BIOMECHANICS: BASIC AND APPLIED RESEARCH

DEVELOPMENTS IN BIOMECHANICS

Huiskes R, Van Campen DH, De Wijn JR (eds.): Biomechanics: Principles

and Applications. 1982. ISBN 90-247-3047-3.

Perren SM, Schneider E (eds.): Biomechanics: Current Interdisciplinary Re-

search. 1985. ISBN 0-89838-755-8.

Bergmann G, Kolbel R, Rohlmann A (eds.): Biomechanics: Basic and Ap-

plied Research. 1987. ISBN 0-89838-961-5.
Biomechanics:
Basic and Applied Research
Selected Proceedings of the Fifth Meeting of the European Society of
Biomechanics, September 8-10, 1986, Berlin, F.R.G.

edited by

G. Bergmann
Department of Orthopedics, Free University of Berlin,
Berlin, F.R.G.

R. Kolbel
Department of Sports Science, University of Hamburg,
Hamburg, F.R.G.

A. Rohlmann
Department of Orthopedics, Free University of Berlin,
Berlin, F.R.G.

1987 MARTINUS NIJHOFF PUBLISHERS ....

a member of the KLUWER ACADEMIC PUBLISHERS GROUP .11
DORDRECHT / BOSTON / LANCASTER .~
Distributors

jor the United States and Canada: Kluwer Academic Publishers, P.O. Box 358,
Accord Station, Hingham, MA 02018-0358, USA
jor the UK and Ireland: Kluwer Academic Publishers, MTP Press Limited,
Falcon House, Queen Square, Lancaster LA1 1RN, UK
jor all other countries: Kluwer Academic Publishers Group, Distribution Center,
P.O. Box 322, 3300 AH Dordrecht, The Netherlands

Library of Congress Cataloging in Publication Data

European Society of Biomechanics. Meeting (5th: 1986 :
Berlin, Germany)
Biomechanics : basic and applied research.

(Developments in biomechanics)
Includes index.
1. Rehabilitation technology--Congresses.
2. Human mechanics--Congresses. 3. Biomechanics--
Congresses. I. Bergmann, G. (Georg) II. Kolbel, R.
(Reinhard), 1937- . III. Rohlmann, A. IV. Title.
V. Series. [DNLM: 1. Biomechanics--congresses.
WI DE997VKH / WE 103 E889 1986b]
RM950.E97 1986 617'.3 87-21991

ISBN-13: 978-94-010-8007-1 e-ISBN-13: 978-94-009-3355-2

DOl: 10.1007/978-94-009-3355-2

Copyright

1987 by Martinus Nijhoff Publishers, Dordrecht.

Softcover reprint of the hardcover 1st edition 1987

All rights reserved. No part of this publication may be reproduced, ston~d in a

retrieval system, or transmitted in any form or by any means, mechanical,
photocopying, recording, or otherwise, without the prior written permission of
the publishers,
Martinus Nijhoff Publishers, P.O. Box 163, 3300 AD Dordrecht,
The Netherlands.
 v

PREFACE

By definition Biomechanics is the application of engineering

methods to study the mechanical aspects of living beings.
Mostly the life scientists have the questions but lack of the
specialized methods. The engineers on the other hand can
handle very specialized equipment and methods, but lack in
the biological thinking. If both sides are able to adapt to
each other, Biomechanics is a classical field of interdis-
ciplinary cooperation. In the beginning, most biomechanical
research was done in the field of orthopaedics. But other
areas like cardiovascular research, dentistry, sports and
many others gain increasing importance.
This situation is clearly reflected in this book, which
contains a selected number of papers which were presented at
the Fifth Meeting of the European Society of Biomechanics,
held in September 1986 in Berlin. Meanwhile these meetings
have become a well accepted forum and a place of interdis-
ciplinary discussion for scientists in Biomechanics on the
one side and surgeons and other peoples interested in biome-
chanical solutions on the other. It is the third time that
the proceedings are published as a book and the editors are
sure that this volume will help to establish this series
"Developement in Biomechanics" as a valuable tool for all
people involved in Biomechanics.
The Fifth Meeting of the ESB also marks the tenth anniversary
in the short history of the European Society of Biomechanics.
From the first conference 1976 in Bruxelles over the follo-
wing meetings in Strasbourg, Nijmegen and Davos to the event
in Berlin, an increasing number of society members and parti-
cipants at the conferences shows the increasing importance of
biomechanics research in human health.
Regarding the contributors to the meeting and also to this
volume, as well as the membership of the ESB, certain streams
have become apparent: There is a hard core of full time
researchers who consistently work in this field and present
their work. Then there are those temporarily involved in this
interdisciplinary work as part of their carreer in either of
the life- or engineering sciences. There is fluctuation among
the latter. They seek the ESB only as a temporary forum but
are a valuable enrichment because they bring with them a lot
of new ideas and idealism.
During the last years, economical constraints allowed the
establishment of only some new research groups. Although the
financing of new teams became increasingly difficult, the
scientific level of the existing ones became even higher,
which is also reflected in the content of this book.
VI

As in all growing disciplines, certain developments can be

observed also in our society: In the beginning many problems
wait for solution. Those are tackled first which can be
solved with a limited expenditure of time and money and whose
solutions can be applied most easily and on a broad basis.
Success is quick and obvious. At this stage the scientific
language of the researchers is still understood by people who
are eager to apply the results. A common language is an
important factor for the common progress. Examples for this
starting phase are found in the development of fracture
treatment and joint replacement. With time, however, methods
become more sophisticated and expenditure greater. The danger
arises that the scientific language becomes hard to under-
stand for people expecting solutions to their problems. They
are nonplussed and interdisciplinary exchange suffers.
Currently biomechanics seems to be in such a critical phase.
Our society, however, will live and grow only if both sides
understand each other. The language at our meetings must
remain understandable for both sides, even at the risk that
not all the details of the methods are presented orally. The
scientists have the onus of presenting the results and their
importance in a clear fashion rather than to describe theory
and method in unintelligible detail. The Conference Committee
of the Fifth ESB- Meeting has tried to keep this aspects in
mind during the selection of papers, as have the editors of
this book. It is hoped that this will add to the relevance of
these proceedings.
Many people were involved in the preparation of the ESB-
Meeting and this book. First of all the editors want to thank
all authors for the work of preparing excellent talks, pos-
ters and manuscripts. The Conference Committee (P. Brinck-
mann, L. Claes, U. Rehder, U. Soltesz and A. Walter) had a
lot of work in organizing the selection process for the
Meeting, as had the reviewers of the abstracts for the
Meeting. The chairmen of the sessions were not only involved
in the excellent discussions during the Meeting, but also
reviewed the manuscripts for this book.
Great financial support was obtained from the Deutsche For-
schungsgemeinschaft and the City of Berlin. Many companies
also contributed to the financial basis of the Meeting. The
Free University was a generous host for this event and its
Department of Orthopaedics (Oskar- Helene- Heim) supported
the Conference in each possible way. Last but not least
Ms. K. Meyer-Wendecker and the team of the Biomechanics
Laboratory of the Oskar- Helene- Heim organized the event in
the most enthusiastic and effective way.
It is a pleasure for the editors to thank all of them!
Febr. 1987
G. Bergmann, R. Kolbel, A. Rohlmann
 VII

TABLE OF CONTENTS
PREFACE v
I NV ITED LECTURES
BIOMECHANICS IN ORTHOPAEDICS
G. Friedebold and R. Wolff 3
BIOMECHANICS OF TOTAL KNEE REPLACEMENT
P.S. Walker 19
GAIT ANALYSIS: A SURVEY
E.Y.S. Chao 33
BIOMECHANICS OF SPORTS - CURRENT PROBLEMS
W. Baumann 51
BIOMECHANICS OF TENDONS AND OTHER SOFT CONNECTIVE
TISSUES. TESTING METHODS AND STRUCTURE-FUNCTION
INTERDEPENDENCE
A. Viidik 59
CARDIOVASCULAR BIOMECHANICS
T. Arts and R. S. Reneman 73
FLUID MECHANICS IN CARDIOVASCULAR RESEARCH.
CARDIAC VALVE FLOW DYNAMICS
Z.H. Kang 85
THE DEVELOPMENT OF TECHNOLOGY IN MEDICINE AND BIOLOGY
W.J. Perkins 99

METHODOLOGY IN BIOMECHANICAL RESEARCH

THE ACCURACY OF CRITERIA FOR AUTOMATIC 3-D GRAPHICS
RECONSTRUCTION OF BONE FROM COMPUTER TOMOGRAPHY
P. Rothuizen, L. van Erning and R. Huiskes 109
COMPARISON OF OPTOELECTRONIC AND FILM BASED KINEMATIC
DATA IN DYNAMIC BIOMECHANICAL EVALUATION OF BACK MUSCLE
TENS I ON
T. Leskinen, R. Hedberg, B. Jonsson and I. Kuorinka 115
INSTANTANEOUS HELICAL AXIS ESTIMATION VIA NATURAL,
CROSS-VALIDATED SPLINES
H.J. Woltring, A. de Lange, J.M.G. Kauer and R. Huiskes 121
ESTIMATION OF GAIT CYCLE DURATION AND STRIDE LENGTH
FROM ONE MARKER KINEMATIC DATA
H. Lanshammar 129
SPECKLE INTERFEROMETRY: A TECHNIQUE TO INVESTIGATE
DEFORMATIONS IN THE FEMORAL BONE
H.-E. Hoyer and J. Dorheide 135
VIII

SPECKLE INTERFEROMETRY FOR INVESTIGATION OF THE RIGIDI-

TY OF EXTERNAL FRACTURE FIXATION - A PRELIMINARY STUDY
J.R. Rether, H.-E. Hoyer and G. Giebel 141
AN AUTOMATIC METHOD TO EVALUATE GOODNESS OF MUSCULAR
WORK DURING HUMAN LOCOMOTION
G. Ferrigno, G.C. Santambrogio and K. Jaworek 147
QUANTITATIVE EVALUATION OF THE WALKING PERFORMANCE OF
THE HORSE USING FORCE PLATE DATA ANALYSIS
H.W. Merkens and H.C. Schamhardt 153
QUANTIFICATION OF SKIN DISPLACEMENT IN THE WALKING HORSE
P.R. van Weeren and A. Barneveld 159

BIOMECHANICS OF SOFT TISSUE

MEDIAL COLLATERAL LIGAMENT HEALING: A BIOMECHANICAL
ASSESSMENT
M.A. Gomez, M. Inoue, E. McGurk-Burleson and
S.L-Y. Woo 167
THE DETERMINATION OF LIGAMENT STRAIN BY A COMPUTER
CONTROLLED KNEE LOADING APPARATUS
L. Claes, L. Durselen and H. Kiefer 173
REDESIGN AND DEVELOPMENT OF A FORCE TRANSDUCER OF THE
BUCKLE TYPE
G.W.M. Peters, A.A.H.J. Sauren and H. v. Mameren 179
THE WRINKLING OF THIN MEMBRANE-LIKE CONNECTIVE TISSUE
STRUCTURES
D.G. Roddeman, C.W.J. Oomens, J.D. Janssen and
J. Drukker 187
STRESS RELAXATION, USED AS A TOOL FOR DIAGNOSIS OF
INCOMPETENCE OF HUMAN CERVIX IN TERMS OF A MIXTURE
MODEL OF TI SSUE
W.A. van Duyl, A.T.M. van der Zon, C.W.J. Oomens
and A.C. Drogendijk 193

ORTHOPAEDIC BIOMECHANICS: BONE

A STUDY OF HAVERSIAN SYSTEMS
J. J. Sun and J. Ge n9 201
THE ELASTIC MODULUS OF CANCELLOUS BONE: DEPENDENCE ON
TRABECULAR ORIENTATION
N. Inoue, K. Sakakida, F. Yamashita, T. Hirai and
T. Katayama 207
 IX

MECHANICAL PROPERTIES OF SUBChONDRAL BONE FROM NORMAL

AND OSTEOARTHROTIC HUMAN FEMORAL HEADS
W.BonfieId, R.R.H. Coombs, C. Doyle, D.J. Sharp and
K.E. Tanner 213
CONTROL OF CHONDRO-OSSEOUS SKELETAL BIOLOGY BY MECHANI-
CAL ENERGY
D.R. Carter, D.P. Fyhrie, R. Whalen, 1.E. Orr, D.J.
Schurman and D.J. Rapperport 219
MECHANISMS OF CRACK PROPAGATION IN CORTICAL BONE
J.C. Behiri, W. Bonfield and B. Charalambides 225
A BIOMECHANICAL ANALYSIS AT LAMELLAR LEVEL OF FEMORAL
SHAFTS DEFORMED IN BENDING
M. Portigliatti Barbos, S. Carando, A. Ascenzi,
A. Boyde and S. Improta 231
BENDING STRENGTH OF OPERATIVELY STABILIZED FRACTURES
UNDER THE INFLUENCE OF BISPHOSPHONATES
P.C. Kramers, B.A. Rahn, J. Cordey and S.M. Perren 237
HOLOGRAPHIC INVESTIGATION OF MECHANICAL PROPERTIES OF
TIBIA FIXED WITH INTERNAL FIXATION PLATE
A. Kojima, R. Ogawa, N. Izuchi, 1. Matsumoto, K. Iwata
and R. Nagata 243

ORTHOPAEDIC BIOMECHANICS: JOINTS

BIOMECHANICS OF THE SHOULDER AND ELBOW
N. Berme 251
LIGAMENT STRAINS IN WRIST-JOINT MOTIONS
A. de Lange, J.M.G. Kauer and R. Huiskes 255
EXPERIMENTAL STUDY OF THE HIP BIOMECHANICS USING PRE-
SCALE FILMS
J.A. Petit, S. Angevin, R. Darmana, J.P. Morucci and
D. Blanquaert 261
THE SIGNIFICANCE OF THE TORQUE LOADING OF THE TOTAL HIP
PROSTHESIS
E. Savvidls, F. Loer and O. Werner 267
THE LUBRICATION OF THE NATURAL JOINT: VISCOSITY OF
HYALURONIC ACID AND FRICTION IN THE HUMAN HIP
H.E. Langer, S. Altmann, W. Luhrs and H. Zeidler 273
MECHANICAL PROPERTIES OF THE ARTICULAR CARTILAGE
M. Oka and Y.Kotoura 279
MATHEMATICAL SIMULATIONS OF PASSIVE KNEE JOINT MOTIONS
L. Blankevoort and R. Huiskes 285
x
DIRECT MEASUREMENT OF DYNAMIC CONTACT PRESSURE ON THE
TIBIAL PLATEAU OF THE KNEE
H. Inaba and M. Arai 291

A DYNAMICAL MODEL OF THE KNEE JOINT

A. Zabel and U. Rehder 297

GEAR PROPERTIES OF THE HUMAN KNEE JOINT

U. Rehder 303

MOIRE FRINGE CONTOUROGRAPHIC DETERMINATION OF THE

THREE-DIMENSIONAL GEOMETRY OF THE POSTERIOR ARTICULAR
SURFACE OF THE CALCANEUM
P. Klein and F. de Schryver 309

ORTHOPAEDIC BIOMECHANICS: JOINT PROSTHESES

LONG TERM ANIMAL TESTS WITH INSTRUMENTED HIP IMPLANTS
G. Bergmann, F. Graichen, H. Jendrzynski and
A. Rohlmann 317

A HYPOTHESIS CONCERNING THE EFFECT OF IMPLANT RIGIDITY

ON ADAPTIVE CORTICAL BONE REMODELLING IN THE FEMUR
R. Huiskes, M. Dalstra, R. v.d. Venne, H. Grootenboer
and T.J. Siooff 325

STRESS TRANSFER ACROSS THE HIP JOINT IN RECONSTRUCTED

ACETABULI
R. Huiskes and T.J. Siooff 333

STRESS ANALYSIS OF THE UNCEMENTED DSL HIP PROSTHESIS

B. Gasser, S. Tepic and S.M. Perren 341

INFLUENCE OF POROUS COATING THICKNESS AND ELASTIC

MODULUS ON STRESS DISTRIBUTION IN HIP PROSTHESES
A. R0 tl, 1man n, E. J. Che a I and W. C. Hayes 347

A METHOD FOR THE COMPARATIVE EVALUATION OF THE INITIAL

STABILITY OF CEMENTLESS HIP PROSTHESES
J. Eulenberger, E. Schneider, W. Steiner, D. Wyder
and R. Friedman 353

CAD/CAM PRODUCTION OF CUSTOMER MADE BONE TITANIUM ALLOY

IMPLANTS BY MEANS OF CT IMAGES
C. Kranz and U. Ahrens 359

ON LABORATORY EXPERIENCES IN ENDURANCE TESTING OF

ANCHORAGE STEMS OF HIP JOINT ENDOPROSTHESES
A. Walter, M. Bergmann, E. Reithmeier, K. Weinmann
and W. PIitz 365

STRUCTURAL SAFETY AND STRESS DISTRIBUTIONS IN CERAMIC

HIP-JOINT-HEADS FOR DIFFERENT NECK LENGTHS
G. Kleer, U. Soltesz, U. Benzing and D. Siegele 373
 XI

A MATHEMATICAL MODEL OF HUMAN KNEE-MOTION AND EVALUA-

TION OF KNEE ENDOPROSTHESES
J. Hoschek, J. Halt, G. Selvik and U. Weber 379

ORTHOPAEDIC BIOMECHANICS: IMPLANTS AND FRACTURE FIXATION

PARAMETRIC ANALYSIS OF THE STRESS PROTECTION IN BONE
AFTER PLATING
J.R. Cordey, S.M. Perren and S. Steinema.nn 387

LIMITS OF PLATE ON BONE FRICTION IN INTERNAL FIXATION

OF FRACTURES
J. R. Cordey and S. M. Perren 393

THE BENDING OF THE SCREW IN THE OSTEOSYNTHESIS WITH

AUTOCOMPRESSION PLATES
Th. Hopf and W. Krapf 399

MICROCOMPUTER AIDED FINITE ELEMENT ANALYSIS IN FRAC-

TURED LONG BONES FIXED BY EXTERNAL PLATE
T. Zagrajek, K. Kedzior and P. Tulin 405

THE INCORPORATION OF FRICTION INTERFACES IN A NON-

LINEAR, FINITE ELEMENT MODEL OF A PLATED LONG BONE
G.S. Beaupre, D.R. Carter, T.E. Orr and J. Csongradi 411

THE SURGICAL EXPANSION DOWEL: A PROVEN PRINCIPLE OF

MECHANICAL FASTENING AND FIXING TECHNOLOGY TOOSTEO-
SYNTHETIC USES
W. Kramer, A. Fischer, R. Arlt and W. Neugebauer 417

ANATOMICAL CURVATURE OF THE FEMORAL MEDULLARY CANAL FOR

INTRAMEDULLARY RODDING
K. Zuber, J. Eulenberger, E. Schneider and S.M. Perren 423

COMPARATIVE INVESTIGATION ON THE BIOMECHANICAL PROPER-

TIES OF THE CIRCULAR AND OTHER THREE-DIMENSIONAL EXTER-
NAL FIXATORS
B. Gasser, D. Wyder and E. Schneider 429

EVALUATION OF INCREASING STIFFNESS OF HEALING BONE

USING EXTERNAL FIXATION
W. Siebels, R. Ascherl, H. Brehme, H. Albersdbrfer
and G. BlUmel 435

THE EFFECT OF DIFFERENT REGIMES OF AXIAL MICROMOVEMENT

ON THE HEALING OF EXPERIMENTAL TIBIAL FRACTURES
A.E. Goodship, D.J. Kelly, H.S. Rigby, P.E. Watkins
and J. Kenwright 441

THE INFLUENCE OF INTERMITTENT CHANGE OF MECHANICAL

ENVIRONMENT UPON THE HEALING OF TIBIAL FRACTURES
J. Kenwright, J.R. Richardson, A.J. Spriggins,
D.J. Kelly and A.E. Goodship 447
XII

EXPERIMENTAL INVESTIGATION ABOUT THE STABILITY OF THE

GREATER TROCHANTER
F. Leier, E. Savvidis and U. Herrboldt 453
BENDING FATIGUE CHARACTERISTICS OF ARTIFICIAL KNEE
LI GAMENTS
A.H. Soni, M. Kutubuddin, W.A. Grana and M.R. Gudavalli 459
BIOMECHANICS OF THE ACROMIOCLAVICULAR STABILISATION
H. Kiefer, L. Claes, J. Holzwarth 465

ORTHOPAEDIC BIOMECHANICS: SPINE

DETERMINATION OF SPINAL STRESS BY BIOMECHANICAL MODEL
CALCULATIONS AND COMPARISON WITH SPINAL MECHANICAL
STRENGTH
M. Jager, A. Luttmann 473
EFFECT OF TENSION PRELOAD ON LUMBAR SPINE KINEMATICS
AND ITS CLINICAL IMPLICATION
A.H. Soni, J.A. Sullivan, M.R. Gudavalli, W.A. Herndon 479
REGIONAL LUMBAR SAGITTAL MOBILITY AND BACK TROUBLE
K. Burton 485
ELECTROMYOGRAPHIC ACTIVITY AS AN INDICATOR OF BACK LOAD
AT WORK
T. Leskinen, H. Nieminen, S. Kalli, M. Gautreau and
I. Kuorinka 491
AN ATTEMPT AT DETERMINING THE EFFORT OF THE LUMBAR PART
OF THE SPINE ON THE BASIS OF MODEL INVESTIGATIONS
R. Bedzinski and A. Wall 497
EXPERIMENTAL STUDY ON THE TORSIONAL RIGIDITY OF VARIOUS
DORSAL STABILIZATION PROCEDURES ON LUMBAR SPINES
Ch. Ulrich, O. Worsdorfer and L. Claes 505
POSSIBILITIES AND RESULTS OF TOTAL LUMBAR DISK REPLACE-
MENT
K. Schellnack and K. BUttner-Janz 511
THE ROLE OF THE ALAR AND TRANSVERSE LIGAMENTS FOR THE
STABILITY OF THE CRANIO-CERVICAL JOINT
E. Schneider, J. Dvorak, D. Wyder and S.M. Perren 517
3-D AUTOMATIC POSTURE ANALYSIS FOR EVALUATION OF BRACE
EFFECTS ON SPINAL DEFORMITY
R. Assente, G.C. Santambrogio and R. Vigano 523
 XIII

ORTHOPAEDIC BIOMECHANICS: CLINICAL ASPECTS AND REHABILITATION

STRIPPING OF THREAD IN BONE BY COMMERCIAL CORTICAL
SELF-TAPPING AND PRETAPPED SCREWS
X.X. Xu, J. Cordey, B.A. Rahn, W.J. Ziegler and
S.M. Perren 531
COULD KOEHLER-FREIBERG'S DISEASE BE CAUSED BY BIOMECHA-
NICAL FACTORS?
H.A.C. Jacob, H. Zollinger and S. Georgiev 537
MEASUREMENT OF PRESSURE DISTRIBUTION ON CURVED AND SOFT
SURFACES
K. Nicol 543
CRUTCH, STAIRCASE RAILING AND FOOT-FLOOR REACTION
FORCES DURING PARAPLEGIC'S STAIR CLIMBING
D. Rudel, T. Bajd, A. Kralj and H. Benko 551
SLOW WALKING IN NORMAL SUBJECTS ON A TRACK AND A MOTOR
DRIVEN TREADMILL
R.H. Rozendal and H.H.C.M. Savelberg 557
COMPARISON OF SUBJECTIVE GAIT OBSERVATION WITH MEASURED
GAIT ASYMMETRY
D. Attinger, S. Luethi and E. Stuessi 563
IMPULSE CHARACTERISTICS IN AIDED GAIT AND WALKING AID
DESIGN
K.A. Opila, A.C. Nicol and J.P. Paul 569
UPON THE MECHANICAL LOAD OF ELBOW CRUTCHES
Chr. Boosfeld and U. Boenick 575
SIGNIFICANCE OF THE MECHANICAL POWER AT THE LOWER LIMB
JOINTS DURING WALKING BY HEMIPLEGIC SUBJECTS
C. Frigo and V. Licari 581
THE ROLE OF MUSCLE FORCES IN THE INSTABLE HUMAN KNEE
JOINT IN VITRO
H. Kiefer, L. DUrselen and L. Claes 587

DYNAMICS OF THE MUSCULOSKELETAL SYSTEM

RELATION BETWEEN MECHANICAL LOAD AND EMG-ACTIVITY OF
SELECTED MUSCLES OF THE TRUNK UNDER ISOMETRIC CONDI-
TIONS
A. J. Spaepen, W. Baumann and H. Maes 595
MINIMIZATION STRATEGIES FOR ARM MOVEMENT
M. Epstein, J.A. Hoffer and E.C. Mikulcik 601
CHARACTERISTICS OF THE GAIT INDUCED BY A FALL
M.C. Do and Y. Breniere 607
XIV

MUSCLE FIBRE COMPOSITION, ISOMETRIC LEG STRENGTH, AND

ENDURANCE IN MAN
J.H. Challis, D.G. Kerwin, R. Ramsbottom and L.H.Boobis 613
IN-VIVO VISCOELASTICITY OF THE HUMAN SOLEUS MUSCLE
M. R. Shorten and D. G. Kerwi n 619

SPORTS BIOMECHANICS
A NEW METHOD TO MEASURE LATERAL BOW ACCELERATIONS
DURING SHOOTING IN ARCHERY
C. Gallozzi, L.M. Leonardi, A. Pace and G. Caselli 627
REDUCTION OF LATERAL BOW DISPLACEMENT USING DIFFERENT
TORQUE FLIGHT COMPENSATORS AND STABILIZERS IN ARCHERY
L.M. Leonardi, C. Gallozzi, A. Pace and A. Dal Monte 633
AN IMPROVED ARCHERY SIMULATOR FOR OBJECTIVE DYNAMIC
TESTS OF BOWS AND ARROWS
R. Pekalski 639
OPTIMIZATION OF ROWING MOTION TECHNIQUE BY USE OF
INTERACTIVE COMPUTER SIMULATION
A. Dal Monte, A. Komor and L. Leonardi 645
AN IMPROVED MODEL AND COMPUTER SIMULATION OF CYCLING
MOTION TECHNIQUE
A. Komor, W. Kuban and L. Parfianowicz 653
PRESSURE MEASUREMENTS INSIDE SHOES AND APPLICATION IN
ALPINE SKIING
P. Schaff, W. Hauser, R. Schattner and M. Kulot 659
AN ANALYSIS OF HUMAN MAXIMAL ISOMETRIC VOLUNTARY PLAN-
TAR FLEXION AS A FUNCTION OF ANKLE AND KNEE JOINT ANGLE
P.A. Huijing, A.E. Greuell, M.H. Wajon and
R.D. Woittiez 667
AN ATTEMPT AT EVALUATION OF STRENGTH/SPEED ABILITIES OF
LOWER EXTREMITIES FOR ATHLETES
Z. Trzaskoma, G. Bartosiewicz, J. Eliasz, A. Dabrowska,
J. Gajewski, L. Iskra and W. Wit 673
COMPLEX INVESTIGATION OF TRACK START
M. Brand, H. Greiff, H. Theysohn, W. Oberste and
K. Nicol 679
4 x 100 m RELAY: MODELLING BY PERMUTATION TECHNIQUES
AND ULTRASOUND VELOCITY MEASUREMENT ON CURVED TRACKS
F. Liebscher, U. Mielkau, W. Oberste and K. Nicol 685
MODEL OF THE LATERAL YACHT SHELL DISPLACEMENTS USED IN
YACHTSMEN BODY MOTION COORDINATION STUDIES
R. Uklanski, J. Wolf and A. Komor 691
 xv
CARDIOVASCULAR BIOMECHANICS
HYDRODYNAMIC COMPARISON OF DUROMEDICS PROSTHESES AND
ST. JUDE MEDICAL PROSTHESES OF VARIOUS SIZES
R. Heiliger, H. Lambertz and Ch. Mittermayer 699
IN VITRO KINEMATIC STUDY OF THE OCCLUDER MOTION IN AN
ARTIFICIAL TILTING DISC HEART VALVE
P.M. Calderale, G. Bona and F. Vallana 707
NUMERICAL STUDY OF PERISTALTIC PUMPING IN CIRCULAR
CYLINDRICAL TUBES
S. Takabatake, K. Ayukawa and A. Mori 713
MECHANICAL PROPERTIES OF INTACT ARTERIAL SEGMENTS
H.W. Weizsacker and J.G. Pinto 719
ASSESSMENT OF MECHANICAL AND THERMAL PROPERTIES OF
BOVINE PERICARDIUM
P.M. Calderale, F. De Bona, F. Vallana and P. Arru 725

MISCELLANEOUS SUBJECTS
ELASTIC PROPERTIES OF THE HYDROXYAPATITE-BONE COMPOUND
IN RABB ITS
N.M. Meenen, M. Dallek, K.H. Jungbluth and L. Weh 733
FATIGUE BEHAVIOR OF COMPOSITE BONE CEMENT
A. Castaldini, A. Cavallini and R. Pirini 739
BIOMECHANICAL ANALYSIS OF COUPLING BETWEEN MANDIBLE AND
REMOVABLE PARTIAL DENTURES
P.M. Calderale, M. Rossetto and M. Pezzoli 745
ON THE EXISTENCE OF THE ORTHODONTIC CENTER OF
RESISTANCE
P.G. van der Varst, M.C.R.B. Peters and A. de Boer 751
CYCLIC GROUPS OF NON-LINEAR BIOLOGICAL AUTOMORPHISMS
AND THEORY OF CYCLOGENESIS IN BIOMECHANICS
S. V. Petukhov 757

AUTHORS INDEX 763

INVITED LECTURES
 3

BIOMECHANICS IN ORTHOPAEDICS
G. Friedebold and R. Wolff
Free University of Berlin, Oskar-Helene-Heim, Berlin, Germany

Today, there are biomechanic laboratories in the most impor-

tant orthopedic hospitals allover the world. This development
shows the growing importance and influence of biomechanics for
research in orthopedics and clinical everyday life. Biomecha-
nics is an important basic science for the orthopedic surgeon.
In the German speaking area, the cooperation began with the
fundamental idea of Friedrich PAUWELS (39,40) who, being both
physician and engineer, included the results of his theoreti-
cal research in his clinical reflections. PAUWELS was mainly
engaged with the hip joint. He showed the principle of tension
banding by the tractus iliotibialis for this joint. The mus-
cles acting across the joint counterbalance the force caused
by the body weight. Furthermore, he succeeded in showing the
influence of the CCD-angle for the symmetrical stress distri-
bution of all the forces acting on the joint. He layed the
foundation of the term "prearthrotic deformity" created by
HACKENBROCH (16,17). It points out the mechanical factor in
the development of joint degeneration if there is any devia-
tion from the normal axis of the extremities.
So the influence of forces upon the biological structures of
bone and cartilage was shown. The initial conception of ROUX
(43) as well as the basic ideas of WOLFF (51) found a scien-
tific explanation: bone adapts itself to external forces. Com-
pression produces axial orientated cortical bone, tension pro-
duces only a loose tissue structure. Today, these findings of
KROMPECHER (23) and PAUWELS have to be modified, but there is
no doubt, that external forces will influence the formation
and remodelling of bone.
Independent of this, G. KDNTSCHER (24) developed his intra-
medullary stabilization. This method of osteosynthesis allows
early weight-bearing - even during the early period of bone
formation. The mechanical contact between metal and endost
doesn't hinder the new formation and remodelling of bone - on
the contrary, these procedures are promoted. The axial load
causes stress only, if there is a transverse fracture.
Today, we are able to answer the question, whether stress
causes the formation of callus - a matter of controversy in
former days. If the stress produces sufficient stability in
the area of the fracture, so that the biological procedures of
new bone formation are not disturbed by mechanical irritation,
the result will be normal consolidation - dependent on age and
local blood supply only. There is no danger of delayed callus
4

formation or pseudarthrosis. The fundamental point is stabili-

ty. L. BOHLER (4) as well as WATSON-JONES (50) and later on
J. CHARNLEY (11), who standardized a mainly conservative
treatment of fractures, and G. KUNTSCHER, who favoured the
surgical treatment, advocated this principle. KUNTSCHER suc-
ceeded in proving, that even bigger defects of bone will be
bridged and consolidated if only there is optimal immobili-
zation with an adequate shaped nail.
The biomechanical importance of stability for the new for-
mation of bone being recognized, further research was done to
treat fractures which seemed to be unsuitable for nailing,
according to the principles mentioned above. HERZOG (19) de-
veloped special wires (so called "Ausklinkdr&hte") in order to
get stability against rotation. A decisive and important prog-
ress was reached with the introduction of the interlocking
nail. So it is possible to maintain the correct length even in
comminuted and compound fractures under stable condition. The
dynamization some weeks later allows the concentrical approach
of the fragments during weight bearing.
Another important problem for every orthopedic surgeon is
the question how to stimulate the formation of new bone. Here
is not the place to discuss the current concepts of induction
and stimulation. At present the superiority of autogenous bone
grafting is proved. Up to this day, only cancellous bone of
the same individual guarantees an active participation in the
formation of new bone. Under the influence of E. LEXER (25),
it was held for some decades, t i l l the thirties, that a corti-
cal graft has to stabilize the fracture, too - because there
were no proper allogenous implants.
The autogenous cancellous bone graft, advocated by MATTI
(31) in Germany, which had only biological and no mechanical
function, was neglected for years - with no regard to its
osteogenous qualities.
Not until today, when some suitable implants for stabili-
zation are at our disposal, the real value of cancellous bone
is recognized. It is very valuable for the treatment of de-
layed union or pseudarthrosis.
Cancellous bone is incorporated in a bony bed, and dependent
on the earlier possible weight bearing there will be an ade-
quate remodelling. These findings are another proof for the
great practical importance of biomechanical thinking.
An important step in this development yielded the realiza-
tion of a fundamental system for surgical treatment of frac-
tures by the AO-International in Switzerland (38). Compression
guarantees the necessary absence of motion in the area of the
fracture or osteotomy. Therefore, the osteosynthesis under
compression (with a special device, or - if possible - with a
dynamic compression-plate) was largely completed, so that dif-
ferent kinds of fractures could be stabilized - making possi-
ble the beginning of exercises for the injured limb at once.
Subsequently, the so called "fracture-disease" was nearly
eliminated.
Furthermore, the meaning of "primary bone healing" - accord-
ing to the primary wound healing - was recognized. In the mid-
dle of this century, DANIS (13) demonstrated that there is a
 5

bony consolidation without any callus formation if any mechan-

ical disturbance of the fracture healing is eliminated (Fig. i).

FIGURE 1.
Primary bone
healing in a
rabbit, no
callus forma-
tion.

The real importance of this phenomenon becomes evident after

the experimental studies of the AO. If it is possible to es-
tablish an osteosynthesis with sufficient compression at the
surface of the fracture or osteotomy, so that there will be a
high stability and no mechanical disturbance by motion, a pri-
mary consolidation of the bone will be the result. The devel-
opment of the new bone is created by the immediate contact be-
tween the blood vessels along the HAVERSIAN canals. Each visi-
ble callus formation represents movement, increasing with mi-
nor stability. In former days, the callus formation proved by
X-rays showed the progress in healing. Today, there may be
some difficulties to assess the state of bony consolidation
and the moment when weight bearing is possible - because of
the absence of visible increasing formation of new bone.
The type of internal fixative device used in fracture repair
affects the underlying bone: plate fixation can decrease the
mechanical strength of the underlying bone. The biomechanical
effect of compression plating is mainly the increase of rigid-
ity. The original high compression across the fracture surface
may decrease as bone remodelling continues. The so-called
"stress-protection" - defined by the remodelling of bone
structures during the period of high rigidity - causes a minor
resistance against bending forces after the~removal of the
plate, a phenomenon which is not to be found in the use of the
KuNTSCHER nail (Fig. 2).
Recently, the use of less rigid plates made of low modulus
material (Ti-6Al-4V or carbon fiber-reinforced methyl meth-
acrylate polymer) has been advocated. The potential benefits
of such plates are to induce callus formation, to prevent
6

FIGURE 2.
Primary bone
healing in a
rabbit - 4, 8
and 12 weeks
after osteo-
tomy. Consid-
erable atrophy
of bone after
12 weeks
("stress pro-
tection" )

long-term stress-shielding osteoporosis, to facilitate raa~o

graphic observation of bone healing progress and to avoid
plate removal. However, maintenance of initial rigidity and
plate strength in difficult fracture cases are potential prob-
lems (22).
Modern procedures for the stabilization of the vertebral
column are connected with the names of ROY-CAMILLE, LOUIS,
MAGERL and DICK (14,26,28,44,52). As yet, there is no possi-
bility to reconstruct or suture the spinal cord, but the sta-
bilization is the basic requirement for an early rehabilita-
tion. HARRINGTON (18), LUQUE (27), ZIELKE (52) and other or-
thopedic surgeons developed a special device for the correc-
tion of severe deformities of the vertebral column (scoliosis,
M. Bechterew). Profound biomechanic knowledge is necessary for
both the conservative and surgical treatment of these dis-
eases.
The better knowledge of the importance and connec~~on of the
stability on form and function of the spine - an important or-
thopedic problem - has shifted the treatment of spinal lesions
from conservative to surgical therapy. In former times, the
surgical therapy had been restricted to laminectomies of one
or more segments, mainly in cases of compression syndroms in
the spinal canal. As an important disadvantage of this proce-
dure, an additional loss of stability was to be seen.
Stabilization alternatively performed on the vertebral bod-
ies or on the vertebral arches makes a great difference. The
stabilization by an interbody fusion must be of superior suc-
cess, because the weight - bearing axis corresponds to the
line of vertebral bodies. But in many cases of fractures with
or without compression signs of the nerve structures, the dor-
sal approach will be preferred. A combination of mechanical
fixation with transplantation of living bone, and correction
 7

~!C~:!~~:~t~~:d~:sw~~lk~a~o~~:n(;~~w:e:~n:~~=~:G~~:i!~~r~!~~e
of the three columns establishing the human spine. On the base
of this knowledge he developed - according to the ideas of
ROY-CAMILLE (44) - the method of stabilizing mne segment only
by transpeduncular screws and a specially formed plate. This
device can be used in other instable situations lik~ spondylo-
listhesis, too.
Especially in cases of compression fracture of one vertebral
body, which causes a later deformity of the whole vertebral
column, the erection of the concerned vertebral body may be
performed by the device of DICK. He modified and realized the
well-known principle of the Fixateur externe - used for the
extremities - for the situation mentioned above. He changed
the device after implantation into a Fixateur interne by cut-
ting off the long handles of the screws. This modern method
may demonstrate the realization of biomechanic principles for
the use in clinical practice, especially in reconstructive
surgery.
The loss of the stability of the posterior wall of the frac-
tured vertebral body is a sign of danger. The erection of the
deformed vertebral body without supporting this posterior wall
provokes the dislocation of bony fragments in the spinal canal
with neurological complications. With the device of DICK it is
possible to maintain the necessary distance by supporting the.
posterior wall before erecting the anterior part of this body
(Fig. 3) .

FIGURE 3. Stabilization of a vertebral fracture with the

Fixateur interne.

The outstanding importance of stability is proven in the

surgery of cartilage as well. The superficial layers of the
hyaline cartilage are nourished exclusively by the synovia.
The cartilage will be damaged if the joint is immobilized in
8

a cast no matter, if there is mechanical damage or not. The

cessation of mechanical lubrication by inactivity causes an
obstruction of the metabolism of the cartilage with the subse-
quent loss of its typical properties (depending on the age of
the patient and the duration of immobilization). The damaged
cartilage is not able to diminish mechanical forces as before;
the first defects are the beginning of degenerative changes.
Therefore, every joint should be mobilized as soon as possi-
ble - in order to guarantee a physiological lubrication. Con-
sequently, the orthopedic surgeon has to stabilize lesions of
the cartilage, making possible the mobilization mentioned
above.
For some intraarticular fractures, such as chisel fractures,
an anatomical reconstruction of the joint and its surface will
be possible, because the necessary stabilization by an osteo-
synthesis in the bony parts of the fracture doesn't require a
difficult procedure. In cases with a lot of small fragments,
even wires and adhesive don't allow a reconstruction with suf-
ficient stability. The immobilization for a short period may
be unavoidable. Very often, the critical period without lubri-
cation may be bridged by changing the position of the joint in
the bed. The beginning of active exercises, which requires
contraction of muscles with pressure in the joint, will be
allowed after another period of careful passive motion. The
use of a special device for the performance of passive motion
is possible.
The development of artificial joint replacement required
profound biomechanical knowledge. Early designs considered the
acting forces and the transfer of these forces to the bone on
an empirical basis only.
BRAUNE and FISCHER (5,6) tried to calculate the forces act-
ing on limbs and joints by determining the velocities and ac-
celerations of limb segments by using data of mass and centre
of gravity from cadaver dissection.
Realistic estimations of the forces acting across the hip
jOint were published not before the middle of this century
(20). They gave values of around 3 times body weight for one
legged stance phase in walking. The largest component of the
joint force results from the muscles acting across the joint
which counterbalance the external force.
As early as 1891, GLUCK (15) reported on his hemi-replace-
ments for joints. He used natural materials such as ivory and
wood. In 1923, SMITH-PETERSEN (47) introduced his interposi-
tion cup arthroplasty (different types of glass, later cobalt-
chrome alloy). JUDET (21) used a femoral head replacement made
of acrylic which could be keyed in the lateral femoral shaft.
The unfavourable lever arm of the junction of the trunk to the
lower extremity requires a special construction with special
qualities of the involved structures. But these conditions are
not given in the described methods of joint replacement. Stem
breakage, loosening and bone resorption prevented good re-
sults. The cause of the pitfalls was to be seen in biomechani-
cal errors. By the early 1950's, the intramedullary stem was
introduced for a better anchorage (THOMPSON, 48; A. MOORE, 35;
M. LANGE and RETTIG (42) in Germany; MERLE D'AUBIGNE (33) in
 9

France) .
This new technique means an important step forward. But in-
dependant of the better anchorage a new problem became evi-
dent: The friction between the original acetabulum and the
artifical prosthesis. The big difference in the hardness of
the two materials led to a high rate of wear of the cartilage
and of the bone. The biological reaction of the surrounding
tissues caused finally the failure. Pain and a progressive
protrusion of the-metal or acrylic head penetrating in the
acetabulum required another surgical procedure.
In 1960, CHARNLEY (10) introduced his total hip - after many
biomechanical experimental studies in his new established lab-
oratory together with bioengineers. He replaced both parts of
the joint, the socket and the femoral head.
Just before this McKEE and WATSON-FARRAR (32) from Norwich
introduced another model of a total hip joint. They could
show, that the articulation of metal against metal is possi-
ble, if no different metals are used, avoiding a high degree
of electrolytical dissociation (corrosion). J. CHARNLEY pre-
ferred the combination of metal and plastic for the compo-
nents. He succeeded in using methylmethacrylate for tight
fixation. With this kind of cement he makes possible an uni-
form force transfer from device to bone with limited micro-
motion. That means a high rate of initial stability. CHARNLEY
also recognized the importance of tribology - friction, lubri-
cation, wear - and he introduced the principle of low-fric-
tion.
Tribology deals with the interfacial conditions of two
contacting bodies in relative.motion to one another. The wear
and surface damage due to stress effects on the materials can
be most important. The wear may activate foreign body giant
cells and cause focal body destruction. In metal-to-plastic
artifical joints, abrasive wear only occurs to a minimal
degree.
The importance of the low-friction is not only to be seen in
the better actual mobility of the joint permitting really free
motion for all requirements of the daily life, but mainly in
the time of durability. Wear means restriction of motion be-
tween the articular surfaces and, therefore, an increase of
shear forces in the interface between artificial implants and
the bone - that means in the region of the cement. This knowl-
edge has influenced further investigations, because the later
experience that after seven years nearly one third of these
prosthesis became loose (38), required a detailed considera-
tion of the behaviour of the cement in the physiological envi-
ronment and under continuing function during long periods. A
certain aging of the cement was to be found. The low modules
of PMMA allows a gradual transfer of stress to bone, but
mixing and insertion technique are able to influence its homo-
gen ity. Blood and gas inclusions increase the porosity which
significantly reduce the fatigue strength. The adding of anti-
biotics has some clinical advantage but will change the prop-
erties of the cement. In cases of a high rate of wear, these
small particles will restrict the motion of the joint and in-
crease the friction. Consequently, there will be a higher load
10

on the anchorage, and the loosening of the artificial joint

becomes a question of time.
During the last ten years, new ways of anchorage without ce-
ment have been searched for. New materials and new shaping of
the surface of the prosthesis are the current suggestions.
Today, there are three classes of material which are used in
joint prosthesis designs:
- metallic alloys,
- bioceramics,
- polymers.
The biocompatibility of each material is quite well estab-
lished, but the functional characteristics differ. Here is not
the place to discuss the properties of all the metals and me-
tallic alloys used today (Ti, Co-Cr-Mo, Co-Ni-Cr-Mo-Ti,
Ti-Al-V). Studies with titanium and steel showed that there
are po signs of corrosion in the surrounding of titanium but
only in the surrounding of steel.Nearly 20 years ago,the Rus-
sian orthopedic surgeon SIWASH (45,46) introduced an Al-Ti-
alloy for the construction of prosthesis. Titanium wear guar-
antees a high stability,other advantages are the weight and
the elastic properties of this material.
Ceramic contains metallic and nonmetallic elements bonded
ionically. This class of material has a high modulus compared
to bone and low ductility. It is extremely inert, has high
compressive strength but inferior crack resistance characte-
ristics. One favourable property of ceramic material is its
conductiveness to tissue bonding. Moreover, the in vitro wear
rate of ceramic has been the lowest among all artificial ar-
ticulating joint surface designs (9). The ceramic against ce-
ramic combination is of higher quality than the articulation
of a ceramic head against a polymer socket. The wear of the
polymer is of the same degree as in the combination with a
metal head. Therefore, the latter combination should be re-
served for old people only.
The use of ceramics has called special attention to a bio-
mechanical problem of great importance: There has to be an
accurate positioning of both parts of the artificial joint,
especially of the socket.
As shown in the valgus position of the femoral neck, the
functional overloading of the superior part of the acetabulum
provokes an increasing degree of wear at this point. In this
way the typical valgus osteoarthritis is developing. The wear
of ceramic is nearly zero, if the intraarticular forces are
symmetrically distributed. In cases of valgus positioning the
rate of wear surpasses that of the polymers considerably.
The most popular polymer used for implants is ultra-high
molecular weight polyethylene. The weakness of this material
are cold flow and surface wear - small polyethylene particles
may stimulate biological reactions of the surrounding soft
tissues damaging the cement bone interface (the rate of wear
amounts 1 mm during 5 years). Metal backing can minimize the
plastic deformation of the material.
Three basic forms of implant fixations are used today:
1. mechanical fixation through interference fit,
2. interlocking with grouting material (PMMA bone cement),
 11

3. biological fixation through tissue ingrowth.

Each method has some short-term and long-term problems.
Short-term problems include infection, dislocation (both are
problems of surgery), fracture (implant or bone) and loose-
ning. The long-term problems involve articulating surface
wear, material degradation, systematic reaction to implant ma-
terial, bone remodelling or resorption due to physiological
load alteration, and loosening and implant fracture, too (9).
Some problems in Press-Fit-Fixation are:
1. No permanent bond,
2. fibrous tissue interface,
3. stress-related bone remodelling,
4. bone and stem strength.
Mechanical fixation of prosthetic components through inter-
ference of press fit does not provide intimate interface lock-
ing or permanent bonding with bone. Generally, a layer of fi-
brous tissue will develop at the interface as a result of ini-
tial micro-movement. This fibrous tissue is thought by many
authors to act as a shock absorber.
There are some additional problems in cement fixation, e. g.
loosening, loose particle wear (PMMA) and monomer release.
Acrylic cement will compensate in a certain degree for poor
bone quality. However, the surgeon has to be aware of long-
term loosening and unknown biological reactions, especially,
when these implants are used in younger patients. So some
inert materials such as ceramics or carbon are used in order
to enhance direct tissue bonding.
Biological fixation through bone ingrowth into a porous sur-
face of an implant seems to be logical, but there are still
some problems with the initial stability, and as yet, there
are no "real" long-term results (more than 20 - 30 years) .
The bone ingrowth is affected by immobilization, age of the
patient, bone substitute augmentation.
Our experience with more than 550 ceramic hips in ten years
permits the cautious conclusion, that the danger of loosening
is diminishing during the first years after surgery (we are
using the Al 2 0 3 -ceramic prosthesis developed by MITTELMEIER .
Indeed, after the second or third year after surgery the cases
of loosening become more and more rare. For the prosthesis of
the conventional type anchored by bone cement the rate of
loosening is increasing from year to year.
There are a lot of studies trying to assess and analyse
forces and distribution of forces in the hip joint. At lower
loads, the contact areas are mainly at the anterior and poste-
rior horn of the acetabulum and the entire acetabular surface
only makes contact at higher loads. The average area of this
total contact is about 27 cm' giving an average contact pres-
sure at 3 x BW (2250 N) of 0.83 N/mm'.
Experimental and theoretical studies have been carried out
to determine the effect of the prosthesis on bone stresses. Ex-
perimental studies are limited in that only surface strains can
be measured on the bone or stem (using strain gauges). The
number of locations is limited (Fig. 4). Interfacial stresses
cannot be measured using present technology. The data obtained
are only "semi-real", because the muscle actions are not taken
into consideration. The stresses on the lateral side of a
12

FIGURE 4.
Femur armed with strain
gauges (measurement of sur-
face strains).

prosthesis are small at the

top, reach a maximum tensile
value in the middle third of
the stem and then reduce to
zero at the tip. The maximum
tensile stress on the later-
al surface of a CHARNLEY
prosthesis for a femoral
head load of 2000 N (3 x BW)
is about 80 N/mm z , well with-
in the tolerance of prosthe-
tic metals (2,30) i the fa-
tigue limit of modern
"superalloys" is 700 -
800 N/mm 2 , for cast cobalt-
chrome and stainless steel
it is about 300 N/mm' .
More data about load and
load distribution in vivo
will be gained by prosthetic
devices armed with special
measuring instruments. In our laboratory, we have implanted
such prosthetic devices in sheep, and we are intending to use
it in patients.
With the method of "Finite Element Analysis, the engineer
tries to calculate stress and strain even in non-isotropic ma-
terials with a certain irregularity of geometry. The interest-
ing bone area or prosthetic device is divided into numerous
small three-dimensional elements. Some properties (modulus of
elasticity) are assigned to each element. Each element is then
assigned stress-strain equation. The external loads are desig-
nated and the equations are solved using a digital computr.
The output is the stresses and strains at each element, and
generally they are plotted as the stresses along an outer sur-
face or an interface (49).
The failure or the breakage of a prosthesis very often can
be explained by the means of biomechanics. Metallic implants
generally fail as the result of cyclic bending loads. A prop-
erly implanted prosthesis is subjected to an idealized three
point bending. The bending moment applied to ,the prpsthesis
depends on the force across the hip joint and the perpendicu-
lar distance of that force from the collar of the prosthesis.
The moment arm can be reduced by inserting the prosthesis in a
valgus position or by using a prosthesis with a short neck. If
 13

the cement interface fractures along the upper end of the

stem, the point of application of the medial force moves
distally. Here, the cross-sectional area of the stem is less,
a fatigue fracture may be the consequence.

Kinematic, gait analysis:

The measurement of body motions is important to evaluate
functional performance of limbs under normal and abnormal
conditions. With a gait analysis, the joint of the lower ex-
tremity can be examined. Different types of prosthesis for the
same joint can be objectively compared. The patients are eval-
uated on an instrumented walk way that has attached motion-
measuring devices. Moreover, foot-floor contact force and
contact patterns can be measured by a force platform. During
gait analysis, stride length, step length, cadence (strides/
min) and walking speed are measured. The ground reaction force
between the foot and floor during the stance phase of gait can
be measured, as well as the foot to ground pressure distribu-
tion during gait. Dynamic EMG measurement of muscular activity
during walking can provide useful information on muscular
force and coordination (8). There are some efforts to assess
the quantitative muscle force during gait based on electro-
myographic data, too. (As yet, there is only some limited
success in estimating muscle forces based on rectified and
integrated muscle EMG.)
The techniques used for gait analysis should be simple, re-
liable and reproduceable and the data obtained should be cor-
related with the routine clinical evaluation to allow extra-
polation.
The currently available measurement technique of human move-
ment can be classified into three categories (1):
1. Electrical linkage method (electrogoniometer),
2. stereometric methods and biplanar roentgenographic methods
(Monitoring of reference points with high-speed movie
cameras. The reference points on the body can be miniature
light bulbs, reflecting dots, light emitting devices.),
3. accelerometric devices.
For describing the motion of joints, simplified models are
usually adapted. In reality, all anatomic joints have six de-
grees of freedom, in which six independent parameters must be
measured and described (1). Very often, i t will do to describe
the finger joint as a hinge joint with only one degree of
freedom or the hip joint as a ball and socket joint with three
degrees of freedom.
The kinematic analysis can be categorized into two main
areas (7,8):
1. Gross movement of the limb segments interconnected by joints
(described by adopting the Eulerian angle system),
2. detailed analysis of joint articulating surface motion.
Knowing the surface geometric and soft-tissue constraints,
the movements of an articulating joint can be analysed to
provide basic information for lubrication and wear studies.
There are a lot of other topics in orthopedics the bioengi-
neer is engaged in. He has to help in designing special
prosthetic devices for implantation and to modify external
14

braces. prosthetics and orthetics. The development of sophis-

ticated myoelectrical prosthetics (Fig. 5) that can imitate
the function of a hand up to a certain degree would be un-
thinkable without the help of engineers. Moreover, orthopedic
surgery requires a profound knowledge of biomechanics. It is
the basic requirement for the exact position of prosthetic
devices and the planning and performance of surgery in order

conditions,
discharged.
.0
to correct deformities (29) and produce better biomechanical
that the cartilage of damaged joints will be

FIGURE 5.
Myoelectrical
prosthesis.

Here, the principles of orthopedic biomechanics can only be

outlined. The ultimate goal of medical scientists is being
able to select the device and surgical procedure that is best
for their patient.
To get reliable and reproduceable results, the bioengineers
have to pay attention to the laws of physics and experimenta-
tion (MURPHY's law, 3).

Laws of physics: Laws of experimentation:

1. fact is solidified opinion 1. If reproducibility may be a
2. facts may weaken under extreme problem, conduct the test
heat and pressure only once
3. truth is elastic 2. if a straight line fit is re-
quired, obtain only two data
points.

Summary:
Orthopedic biomechanics is a study of forces and motions in
biological systems and the effects that these forces and mo-
tions produce. The bioengineer has to design special prosthe-
tic devices for implantation, modify external braces,
prosthetics and orthetics, and to study the functional perfor-
 15

mance of patients before and after treatment. The understan-

ding of biomechanics has been the basic requirement for the
present possibilities of treatment, the development of frac-
ture fixation devices and artificial joint prosthesis.
While fracture fixation devices are only intended to endure
short periods of in vivo service, until bone fractures unite,
the performance criteria are less stringent - compared to the
performance criteria of a prosthetic device. Even today, not
all the problems of cement fixation (mechanical and biologi-
cal loosening) and of bone ingrowth materials (initial fixa-
tion and stability, stress related, bone remodelling) are
solved.
The ultimate goal of medical scientists is being able to
select the device and surgical procedure that is best for the
patient.

References

1. An KN, Chao EY: Kinematic Analysis of Human Movement.

Annals of Biomedical Engineering, Vol. 12, pp. 585-597,
1984.

2. Andriacchi TP, Galante 10, Rostoker W, Hampton S: An ana-

lytical study of stresses in a femoral stem of a total hip
prosthesis. 21st Annual Meeting of Orthopaedic Research
Society, 1975.

3. Block A: Murphy's Law. Book Two. Price/Stern/Sloan.

Publishers Inc. Los Angeles 1980.

4. Bohler L: Die Technik der Knochenbruchbehandlung.

Maudrich, Wien, 1951.

5. Braune W, Fischer 0: tiber den Schwerpunkt des menschlichen

Korp~rs. Aus: Der Gang des Menschen. Abhdlg. math.-phys.
U. Konigl.-Sachs. Ges. Wiss. 15 (1889).

6. Braune W, Fischer 0: Der Gang des Menschen. Abhdlg. math.-

phys. U. Konigl.-Sachs. Ges. Wiss. Bd. - 6 (1895-1904)

7. Chao EYS: The biomechanics of total joint replacement

surgery. Geriatric, March 1976 (48 - 57).

8. Chao EYS: Prostheses: Implant Materials and Methods of

Fixation. Orthopaedic Knowledge Update. Ed. Biomechanic
Laboratory, Department of Orthopedics, Mayo Clinic/Mayo
Foundation, Rochester, Nov. 1985.

9. Chao EYS: Gait Analysis: A survey. In: Gait Analysis in

Theory and Practice. Proceedings of the 1985 Uppsala Gait
Analysis Meeting. Ed.: Hakan Lanshammar.
16

10. Charnley J: Anchorage of the femoral head prostheses to

the shaft of the femur. J. Bone Jt. Surg. 42-B,28-30(1960)

11. Charnley J: The closed treatment of common fractures.

3 ed. Livingstone, 1961.

12. Charnley J: The bonding of prosthesis to bone by cement.

J. Bone Jt. Surg. 46-B, 518 - 529 (1964).

13. Danis R: Theorie et pratique de l'osteosynthese. Masson,

Paris 1947.

14. Dick W: Innere Fixation von Brust- und Lendenwirbelfraktu-

reno In: Aktuelle Probleme in Chirurgie und Orthopadie,
Bd. 28. Hrsg.: C. Burri, F. Harder, M. Jager. Verlag Hans
Huber, Bern-Stuttgart-Toronto, 1984.

15. Gluck M: Referat uber die durch das moderne chirurgische

Experiment gewonnenen positiven Resultate, betreffend die
Naht und den Ersatz von Defekten hoherer Gewebe, sowie
uber die Verwertung resorbierbarer und lebendiger Tampons
in der Chirurgie. Arch. klin. Chir. 41, 187 (1890).

16. Hackenbroch M: Die degenerativen Erkrankungen des Huftge-

lerikes. In: Handbuch der Orthopadie, Bd. I. Hrsg.: G.
Hohmann, M. Hackenbroch, K. Lindemann. Thieme, Stuttgart,
1961.

17. Hackenbroch M: Faktoren, die die Entstehung und Entwick-

lung der Arthrose beeinflussen. ~rztl. Prax. 23, 3723
( 1971) .

18. Harrington PRo Spine instrumentation. Amer. J. Orthop. 8,

228 - 231 (1964).

19. Herzog K: Nagelung der Tibiaschaftbruche mit einem starren

Nagel. Dtsch. Z. Chir. 276, 227 (1953).

20. Inman VT: Functional Aspects of the Abductor Muscles of

the hip. J. Bone Jt. Surg. 39, 214 (1947).

21. Judet J, Judet R: The use of an artificial femoral head

for arthroplasty in the hip joint. J. Bone Jt. Surg. 32-B,
166 (1950).

22. Kelly PI, An KN, Chao EYS, Rand A: Fracture Healing: Bio-
mechanical, Fluid Dynamic and Electrical Considerations.
Bone and Mineral Research. William A. Peck (ed.). Elsevier
Science Publishers, B.V. 1985.

23. Krompecher S: Die Knochenbildung.Gustav Fischer,Jena,1937.

24. Kuntscher G: Die Marknagelung. Springer, Berlin-Gottingen-

Heidelberg, 1962.

25. Lexer E: Die Verwendung der freien Knochenplastik nebst

 17

Versuchen uber Gelenkversteifung und Gelenktransplantaten.

Arch. klin. Chir. 86, 939 (1908).

26. Louis R: Surgery of the Spine. Springer, Berlin-Heidel-

berg-New York, 1983.

27. Luque E: Segmental spinal instrumentation for correction

of scoliosis. Clin. Orthop. 163, 192 (1982).

28. Magerl F: Der Wirbel-Fixateur externe. In: Fixateur ex-

terne. Hrsg.: B. G. Weber, F. Magerl. Springer, Berlin,
1985.

29. Maquet P: Biomechanics of the Knee Joint. Springer,

Berlin, 1977.

30. Markolf KL, Amstutz HC: A comparative experimental study

of stresses in femoral total hip replacement components:
The effect of prosthesis orientation and acrylic fixation.
J. Biomechanics, 1976.

31. Matti H: tiber freie Transplantationen von Knochenspongio-

sa. Arch. klin. Chir. 168, 236 (1931).

32. McKee GK, Watson-Farrar J: Replacement of arthritic hips

by the McKee-Farrar prosthesis. J. Bone Jt. Surg. 48-B,
245 (1966).

33. Merle D'Aubigne R: Reposition with arthroplasty for con-

genital dislocation of the hip in adults. J. Bone Jt.
Surg. 34-B, 22 - 29 (1952).

34. Mittelmeier H: Zementlose Verankerung von Endoprothesen

nach dem Tragrippenprinzip. Z. Orthop. 112, 27 (1974).

35. Moore A: Metal hip joint. A new self-locking metallic

prosthesis. Sth. med. J. (Bgham, Ala) 45, 1015 (1952).

36. Muller ME: Total hip replacement: planning, technique and

complications. In: Surgical management of degenerative
arthritis of the lower limb, Chapter 10, 91 - 113. Edt.:
R. L. Cruess, N. S. Mitchell. Lea u. Febiger, Philadel-
phia, 1975.

37. Muller ME, Elmiger B: Coxarthrose. 10-Jahres Ergebnisse

der sog. Setzholz-Totalprothese. Orthop. 8, 73-134(1979)

38. Muller ME, Allg6wer,M, Willenegger H: Manual der Os teo-

synthese. AO-Technik. Springer, Berlin, 1969.

39. Pauwels F: Gesammelte Abhandlungen zur funktionellen Ana-

tomie des Bewegungsapparates. Springer, Berlin-Heidelberg-
New York, 1965.

40. Pauwels F: Atlas zur Biomechanik der gesunden und kranken

18

Hufte. Springer, Berlin, 1973.

41. Perren SM: Physical and biological aspects of fracture

healing with special reference to internal fixation. Clin.
Orthop. and Rel. Res. 138, 175 - 196 (1979).

42. Rettig H: Die Huftarthroplastik mit Spezialendoprothesen.

Z. Orthop. 290 (1952).

43. Roux W: Entwicklungsmechanik der Organismen. Gesammelte

Abhandlungen. Engelmann, Leipzig, 1895.

44. Roy-Camille R, Berteaux D: Technique et resultats des

osteosyntheses du rachis lombaire par plaques posterieures
vitsees dans les pedicules vertebraux. Montpellier Chir.
22, 307 (1976).

45. Siwash KM: Development of alloplasty of the hip joint in

the last 15 years. Orthop. Traum. Protez 11, 28 (1969).

46. Siwash KM: The development of a total prosthesis for the

hip joint from a partial joint replacement. Wiederherstel-
lungschirurgie und Traumatologie 11, 53 (1969).

47. Smith-Petersen, MN: Arthroplasty of the hip - a new

method. J. Bone Jt. Surg. 21, 269 (1939).

48. Thompson FR: Two and a half years experience with

vitallium intramedullary hip prosthesis. OJ. Bone Jt. Surg.
36-A, 489 (1954).

49. Walker PS: Biomechanics of Natural and Artificial Joints.

In: Postgraduate Textbook of Clinical Orthopaedics. Ed.:.
N.H. Harris. C.V. Mosby Company.Bristol-London-Boston 1983

50. Watson-Jones R: Fractures and Joint Injuries (ed. 3).

London, Livingstone, 1944.

51. Wolff J: Das Gesetz der Transformation der Knochen.

Hirschwald, Berlin, 1892.

52. Zielke K: Ventrale Derotationsspondylodese. Behandlungs-

ergebnisse bei idiopathischen Lumbalskoliosen. Z.
Orthop. 120, 320 (1982).
 19

BIOMECHANICS OF TOTAL KNEE REPLACEMENT

P.S. Walker
Orthopaedic Biomechanics Laboratory. Brigham and Women's
Hospital. Boston. MA. U.S.A.

INTRODUCTION
Metal-plastic Total Knee Placement (TKR) was first
introduced in the late 1960's. Among the first designs were
the Polycentric, the Freeman-Swanson, the Geomedic and the
DuoCondylar (1). During the 1970's, many other designs were
introduced, some being adaptions to address the problems with
the early designs, and others being new concepts. During the
decade, frequent clinical problems were due to inadequate
instrumentation, inaccurate component alignment, poor cement
technique, excessive device constraint, absence of
patello-femoral replacement, and lack of ligament balancing.
However, institutional reviews published in the early 1980's,
suggested that some designs at least, problems occurred in
only a few percent of cases at up to 10 years follow-up. To
optimise the design for performance and durability, the
designer, manufacturer, and surgeon are faced with a variety
of requirements. Some of these requirements will be discussed
in this article. The first stage is to identify and quantify
the present status, particularly the strong and weak points.
This is done by a review from our own institutional data on
retrieved total knees and clinical follow-ups (2,3,4,5).

THE CLINICAL EVIDENCE

1. ANALYSIS OF 90 RETRIEVED METAL-PLASTIC TKR'S

Materials
The series consisted of Total Knee Replacement's removed
from patients at the Brigham and Women's Hospital. The
plastic was ultra-high molecular weight polyethylene. Full
clinical records were available for 82 cases. Implant time
ranged from 3 months to 10 years with a mean of 5.6 years.
Twenty total knee replacements were in place for more than 8
years, and only 5 for less than 2 years. The cases were all
condylar replacements, with a variety of femoral-tibial
geometry.
Unicondylar 17 Geomedic 6
Duocondylar 12 Kinematic PC 5
Duopatella 23 Robert Brigham 3
Marmor 7 Others 17
20

The reasons for removal were:

Loosening, component shifting 35
Pain in unresurfacec area of joint 15
Sepsis 9
Pain for other reasons 6
Instability/deformity 8
Inadequate data 17

Methods of Stu~y
The plastic components were examined using low power
microscopy. Different wear morphologies were characterised
and assessed on a scale of 0-6 for severity. Thin sections of
plastic from each component were cut uith a scclpel and
examined under transmittec lirht microscopy. Five mm(3)
blocks were cut from 20 components, cnd stucie~ with scanning
electron microscopy. Similar blocks cf plastic from 21
components were used for molecular weight analysis, using high
temperature gel permeation chromatography.
RESULTS
The observed wear and surface damage were as follows:
Embedded acrylic cement particles (43% of components).
These were easily identified. Cement particle craters (74%).
These were identifiable often by jagged sides, dimples at the
base, and high abrasion surrounding. Deformation (cold
flow)(90%). This occurred in the main bearing area, due to
excess contact stresses, near the edges due to subluxation,
and on raised parts such as the tibial eminence due to digging
in of the metal femoral component in internal-external
rotation.
Burnishing (85%) This was glossing of the surface, which
did not appear to result in much material removal.
Abrasion (95%) Scratches, grooves, and stretching of
mUltiple fibers of plastic (- 10-100um) on the surface, fell
into this category. The causes were mainly three-body
abrasion from worn plastic and acrylic.
Delamination (38%) This was seen as sub-surface cracking, a
raised flake or flakes of material, multiple cracks and
fragmentation, and surface cavities. (Figure 1)
The subsurface cracks were often parallel to the surface
and were least 100 um in width.
Section of unused plastic often showed many small cavities
or discontinuities. In some cases, fine lines could be seen
in a network. This pattern became accentuated in used
implants, and in some of the longer-term cases, almost
complete breakdown of the material could be seen. Since the
material is compression molded by fusion of small granules,
it is believed that the fine lines represented granule
boundaries and that height stresses and inadequate fusion led
to the breakdown.
 21

A section through the plastic surface showing delamination

wear.

It has been found in lab tests that wear increases with

decreasing molecular weight (6). In our group. there was a
wide variety.
Low M.W. High M.W.
Minimum 46,000 723,000
Maximum 654,000 4,670,000
Mean 148,000 2,300,000
S.D. 142,000 1,594,000

The overall wear/year correlated with best ambulatory status

at p<O.Ol. For 39 components from one manufacturer (J&J)
there was linear relation between 'overall wear' and years of
implant. (c.c.o.a). There was an inverse correlation of
wear/year with molecular weight (p<.02). Particularly
striking were compressive molded flat Marmor components (which
would show high contact stresses) of high m.w. which showed
minimal wear at 8-10 years follow-up.
The most striking finding from this wear study was the
severity in many cases, far greater than has been reported for
total hips. Where there was delamination and intergranular
cracking, it appeared that severe material breakdown of the
knee components would have occurred in ensuing years.
22

CLINICAL FOLLOW-UP OF 1069 KINEMATIC TKR

The first 1069 consecutive cases of primary Kinematic
Condylar Total Knee Replacement (Figure 2) carried out by nine
staff orthopedists at the Brigham and Women's Hospital were
reviewed. The clinical material was as follows:

1069 cases in 798 patients

667 complete clinical history, complications determined for
all cases.
12-90 years age range,
67 average
77% female. 81% cases over 60 years of age.
47.7% OA 46.8% RA 2.4% JRA
822 cases radiographic study.

Follow-up times (years):

0-1, 386 cases. 1-2, 191 cases. 2-3, 113 cases. 3-4, 67
cases. 4-5, 43 cases. 5-6, 20 cases, 6-7, 2 cases.

Kinematic II Condylar Total Knee.

All clinical data was recorded in the computerized Joint

Registry. Patients were seen at 6 months, 1 year and then
every year. If the patient could not be seen, information was
obtained by telephone. The most recent radiographs were
measured for component and leg alignments. The cement-bone
interface was studied for radiolucent lines (RLL) in 7 tibial
zones (a-p) and 5 femoral zones (m-l). (Figure 3)
 23

RADIOGRAPHIC STUDY OF
KINEMATIC TKR
RADIOLUCENCY SCORING SCHEME

ZONES:

1 t
MED LAT
-1-- 2 3--4 .....

5 7

1 !
-6-
SCORE: 0-1 MM = I
1-2 MM = 2
>2 MM = :3

RLL zones scoring scheme.

Total RLL score was calculated as the sum of the RLL

thickness (to nearest mm) x each zone. Thus a uniform 1mm RLL
all round the tibial would have a score of 7.
Clinical Results
Postoperatively 93% of patients had none or minimal pain on
all activities. 82% could negotiate stairs normally, and the
others had to lead with other leg. Range of motion averaged
2.5-107 degrees.
The Aseptic Revisions were:

Loose femoral component 0

Loose tibial component
Skin necrosis (patellectomy to close) 1
Patella button loose 6
Patella subluxing 5
Patella transverse fracture 2
Patella painful 1
Patella dislocation

17 (1.6%)

Other complications were:

Deep sepsis (9 retained, 1 exch . ) 11 (l %)
Transient problems (skin, hematoma, etc) 29 (2.7%)
(no additional surgery)
24

RadicGraphic Results
A RLL can be taken as an indicator of interface breakdown, but
localise~ RLL may not be significant. The RLL's on the tibial
component occurred primarily at the extreme medial or lateral
zones. (Figure 4)

MED LU
PERCENT
OF CASES -1~~2~ -3.4--
t t
5 7

l~~ -6-
20- ~

10-
~
~

5 2 3

1 2 5 6 7 3 4 ZONE

Tibial RLL

The inner zones had only half the incidence. RLL around the
peg occurred in only a few cases. In the femoral component,
the posterior and anterior flanges were most frequently
affected. The scores on the components were mainly 1,
indicating Imm RLL in only 1 zone. The RLL's occurred in
regions where it is difficult to achieve good cement
pressurization and where high stresses and micromotion occur.
The incidence of a score of 1 or more was 12.9% femoral, 26.5%
tibial, and 6.3% both. The incidence of tibial cases with RLL
increased from 21.3% at 0-1 years, to 45% at 5-6 years, but
the numbers of RLL's with score> 1 did not similarly
increase.
 25

The data on cases with RLL scores> 4 was:

Total number of cases 822
Cases with RLL score> 4 21 (2.6%)
There were only 21 tibial cases (2.6%) with RLL > 4. The
data on these was:

Cement filling wedge-shaped

defect, or thick cement layer 12
Both medial and lateral RLL,
cause not evident 7
Due to varus or valgus deformity 2

CONCLUSIONS REGARDING TKR PERFORMANCE AND PROBLEMS

Aseptic loosening was not a significant problem in the

Kinematic series. No femoral components and only one tibial
component needed revision, the latter occurring due to
deficient bone stock and component misalignment at surgery.
The frequency of cases with RLL's increased over time, but
most of ~he increases were of small RLL's at the sides of the
component, consistent with a.crack propagation mechanism
initiated at the sides (7). The situation did not suggest a
serious failure rate in the future. This result is consistent
with other reports (2,3,4,5). Most of the thick RLL's
occurred beneatq thick cement layers, such as in a wedge
defect. We recommend bone graft, or a metal wedge spacer, in
combination with a long stem, for such cases.
Most mechanical problems were with the patella. The causes
are not fully understood but include both design and surgical
factors (8). Similar problems have been reported in other
series.
Short of a rigorous quality control of the plastic for
internal porosity, intergranular bonding, and molecular
weight, wear at 10 years and more is likely to be a serious
problem in many cases. The effects are unknown but may
include fibrosis, inflammation, and component loosening.
All of the above applies to uncemented TKR as well, except
that the interface is different with unknown long term results
at this time.

OPTIMISING PERFORMANCE
The performance of TKR can potentially be improved by
attention to the patella problems, to reducing wear, and also
in other ways such as increasing motion, minimising shear and
torque on the components, and improving alignment. Many of
these factors can be addressed in the design, but
biomechanical data is needed as design parameters.

1. THE PATELLO-FEMORAL JOINT

There is little information on the effect of total knee
component design and placement, and of the patello-femoral
joint. To test these aspects, we measured 3-D knee and patella
26

motion in a dynamic test rig for different total knee

replacements and a patella button designs.
The tibia was rigidly fixed in the test rig with a defined
axis system using the line through the centers of the
posterior femoral condyles as a transverse axis. A V-shaped
channel was fixed to the femur, and the quadricep tendon
clamped via a transducer to a flexible bicycle cable which was
wound around the spool of a stepper motor. Dynamic
flexion-extension was achieved by winding the cable in and
out. An external weight on the frame provided a joint force
and moment. An arm attached to the channel had three
displacement transducers attached on lateral and medial sides
to measure x,y, and z displacements. The signals from these,
and from a potentiometer to measure flexion, were fed to a
computer, which calculated the three rotations and three
displacements of knee motion. To measure patella motion
relative to the femur, a 3-Space Tracker (Polymus Navigation,
McAuto) was used. The source was fixed to the femoral frame
and moved with the femur; the sensor was fixed on the outside
of the patella. A transformation applied to the 3-D motion
output, gave 3 rotations and 3 displacements referred to the
center of the outer surface of the patella. Knee and patella
motion were determined for the intact knee, the Kinematic
Condylar, PCA, and Kinematic Total Condylar. Data was
obtained for the plastic tibial components in neutral,
internal and external rotation.
Our results for normal patella motion agreed qualitatively
with those of a previous study (9). The ration of patella
flexion to femur flexion averaged 0.69 (0.63 0 0.74). This
ratio was unchanged after prosthetic replacement. In the
natural knee the medial-lateral displacements of the patella
varied from medial displacement of 5mm (mainly in 0-30 degrees
flexion), to a linear lateral displacement of 10mm (Figure 5).

KINEllATIC COHDYlAR P CA
t'JI<I\AL KNEE DOI SllAPED PATELlA AHAT0I11CAL PATELLA BunON

.,.
--k-1I1iH""1Il1l1

.
- - III"I*'&. 101ooliOll
- - _\,I' "'''flO.
01''''''

Patella Motion. A dome patella tracks similarly to normal: a conforming

patella restricts motion.
 27

Most of the m-l displacement occurred in the initial stages of

flexion. This implies that if the patella was initially
resting at the lateral side on subsequent flexion, if the
patella groove did not capture the patella, it could easily
dislocate off the side. The internal rotation paralleled the
medial displacement, varying from 2 to 11 degrees variation.
Patella motions after prosthetic replacement were generally
different from normal. Medial-lateral displacements ranged
from 2-7mms; the direction usually changed to mid-flexion.
Internal-external rotations had ranges from 0 to 10 degrees.
Varus rotations were only about 2-3 degrees and were similar
to normal. When internal-external torques were applied to the
natural knee significant changes were noted in medial-lateral
displacements and internal rotation of the patella; there was
little change in varus rotation. Similar findings were noted
for the Kinematic and the dome button. For the PCA, the
changes were substantially smaller.
Constraint of a prosthetic patella button, especially in
varus-valgus would lead to torsion and possible loosening. A
totally unconstrained button, however, would have excessive
wear. Thus, it is apparent that an appropriate design is a
deep patella groove, especially a the superior flange, with a
partially constrained patella replacement.

2. SURFACE DESIGN
The surface design will affect the factors mentioned above
such as wear, motion, and shear and torque on the components.
In the design of condylar surfaces for TKR, there are
several conflicting requirements. Normal knee motion involves
15 degrees of tibial rotation and 8mm of rollback, from 0-120
degrees flexion(10); about any position, there is 20-30
degrees and 5-10mm of placement at surgery might require
additional laxity. These factors suggest that low constraint
tibial surfaces are needed. One advantage is the low shear
and torque forces transmitted to the interface. However, the
surfaces will be more unstable than normal, placing undue
reliance on the remaining soft tissues. Low conformity gives
high contact stresses on the plastic which could lead to
catastrophic material breakdown in the long term (13,14). In
this study, we computer-generated tibial surfaces with
different motion-laxity-stability criteria, and calculated the
contact stresses, in an attempt to determine the most
acceptable geometry from all aspects.
The starting point was to determine average femoral surface
geometry by slicing 23 knees into 25 sections, dibitizing, and
averaging. The mathematical analog of spherical tcroidal, and
conical surfaces. (Figure 6)
28

A Femoral Surface analog of the anatomic knee for

application to Total Knee Replacement.

Average knee motion was determined by flexing and extending 23

knees dynamically under quadriceps action. Internal-external
rotation and a-p displacement were expressed as a cubic
function of flexion. Equations for laxity curves of intact
knees, (12), were similarly expressed. To reproduce laxity
with the TKR, the criterion was that the surfaces alone would
provide the same laxity curves as the intact knee. The theory
was based on energy considerations (15). Whereby, the work
done by the shear force or torque equals the increase in
potential energy by 'screwing out' the joint surfaces. An
equation defined the vertical height of the femoral surface as
a function of a-p displacement. Similar equations were
derived for internal-external rotation and m-l displacement.
The expressions were obtained for all flexion angles. To
generate a tibial surface, the femoral surface was 'moved' in
the computer through a prescribed path of motion and/or laxity
in mUltiple discrete steps. A horizontal gridwork was defined
on the tibial surface and the lowest y-values at each node
were collected to define the tibial surface. The femur was
then placed through its original motion path. The contact
points were determined and the local radii of curvature along
a-p and m-l axes calculated. Using elasticity theory, the
contact areas and maximum compressive stresses were calculated
(Eplastic = 600 MPA).
 29

The computer generated tibial surfaces gave a range of

stresses depending on the input motion and laxity (Figure 7)

EXTENSION
CYlINDRICAL
RFlEXION

MED~T LAXITY

ANT-POST LAXITY

INT-XT lilT LAXITY

COMBINED LAXITY

o 10 20 30 40
CALCUlATED MAX COMP. STRESS (MPA)

Contact stresses for different types of tibial surface.

The least stresses occurred when the femoral component was

moved uniaxially, a maximally constrained condylar knee. A
cy1inder-on-f1at gave moderately high stresses on lateral and
medial sides. The highest stresses were with the biconvex
femoral condyles on a flat surface. For a 'laxity only'
surface, stresses in the full range were moderate. For
average knee motion, the lateral stresses were high due to the
flat 'rollback' area. With average knee motion and laxity
combined, the stresses were similar to average knee motion.
However, all geometries produced stresses which were higher
than the compressive strength of the plastic. Geometries
which allow free a-p motion by flat tibial surface have both
high str~sses and excessive laxity. For uniaxial motion, the
stresses were much lower, but at the expense of slightly
higher stresses. A reasonable compromise was to reduce
freedom of motion in a-p and rotation as prescribed by normal
knee motion in the form of laxity, which resulted in more
curved surfaces with reduced stresses. A typical surface
which may be suitable for prosthetic application is shown in
figure 8.
30

A Tibial Surface for application to Total Knee

Replacement.

ACKNOWLEDGEMENTS
The author acknowledges the contributions to this work of the
past and present staff of the Orthopedic Biomechanics
Laboratories. Funding was provided by National Institute of
Health, Veterans Administration, and Brigham Orthopedic
Associates.
 31

REFERENCES

1. P.S. Ualker, ch. 6, Human Joints an~ Artificial Replcccments,

publ., Thomas, 1977.
2. J.N. lnsall et al., Clin. Orthc., 192:13, 1985.
3. C.O. Townley, Clinical Orthopedics., 192:82, 1983.
4. N.S. Eftekhar, J. Bone Joint Surgery, 65A:293, 1983.
5. F.C. Ewald et aI., J. Bone Joint Surgery, 66A:I032, 1924; and
Proceedings AAOS, 1986.
6. N.!. Nuebaum, R.II. Rose et ai, J. App Poly Sci, 23:777, 1979.
7. JJ.L. Lewis et al., Proc ORS, p. 144, 1985.
8. R.L. Markow et ai, J. Bone Joint Surgery, 67A:1321, 1985.
9. P. Kaltwasser et ai, Orthopedic Research Society, January,
1987. San Francisco.
10. Van Campen, Huiskes, Blankevoort, Van Rens, Orthopedic
Research Society Proceedings, P. 386, 1986.
11. J.S. Rovick et al., Orthopedic Research Society Proceedings,
January 1987, San Francisco.
12. Harkolf, J. Bone Joint Surgery. 63A:570, 1981.
13. Landy. Proceedings Orthopedic Research Society, p. 96, 1985.
14. Uright, Clinical Orthopedics. P. 205:67, 1986.
15. Ualker, lnst. Hech. Eng., Total Knee Replacement, 1974.
 33

GAIT ANALYSIS: A SURVEY

LLS. Chao
Biomechanics Laboratory, Mayo Clinic and Mayo Foundation,
Rochester, Mi nnesota, U. S. A.

1. INTRODUCTION
Human walking has been studied extensively by many investigators from
various disciplines, including those of anatomy, kinesiology,
neurophysiology, and biomechanics. The historical development of this
branch of biological science since the pre-Renaissance era has been
thoroughly reviewed [1,2,3]. Although gait analysis, or assessment of
gait, has been pursued for a long period of time, our knowledge concerning
human locomotion is far from complete.
Normal walking is an extremely complex biomechanical process. It
involves three-dimensional motion in a multiple linkage system which
requires coordinated control by the central nervous system that far exceeds
the sophistication of any man-made machine. Despite attempts to model
human gait, comprehensive dara are still insufficient to classify
locomotion in quantitative terms. The inability to objectively describe
characteristics of gait and lack of effective data analysis methods to
study large populations are some of the factors which have contributed to
difficulties in analyzing gait.
Although attempts to describe human gait based on neurophysiologic
control principles are still in the developmental stage, phenomenological
aspects of locomotion have shown significant promise in associating normal
and abnormal gait patterns with underlying musculoskeletal pathology.
Since the major aim of treating patients with arthritic disease is to
correct joint deformity and restore functional normality, the ability to
quantitate any visible structural or functional change in limb segments
afflicted by pathology based on well-defined and clinically-meaningful
parameters is important. In addition, a carefully established normative
data base from a broad spectrum of the population is essential to define
anthropomorphic and physiologic variability. Establishment of such a data
base depends on the availability of efficient measuring techniques and data
analysis methods.

2. BASIC GAIT PARAMETERS

Gait parameters can be categorized into four basic groups. Although
some of these are well known, their definitions may vary among different
investigators. It is therefore necessary to briefly define them all for
future reference.
2.1. Temporal distance factors
As shown in Figure 1, this typical footprint diagram illustrates the
basic temporal distance factors studied. To compensate for anthropometric
variation, stride length and step length are normalized against lower
extremity length (LEL). Cadence is defined as the number of strides/min- l
rather than the customary unit of steps/min-I. Walking speed is expressed
in m/min-l.
34

-----I-----.~.~?' length _ _ _- l
right)

--_._--

. . - -- - -walki;'~ direction "

Time

Figure 1. Typical temporal distance factors used in gait analysis. Note

that the single stance period of one leg is the same as the swing phase of
the contralateral leg.

Although left/right symmetry is generally assumed among normals, this

hypothesis is not true for patients. Consequently, the basic temporal
distance factors are defined separately for left and right legs (Fig. 1).
In certain gait patterns, the events of heel strike and toe off cannot be
clearly identified. When this occurs, 'foot strike' and 'foot off' can be
used as temporal markers for these stride parameters [4,51.
2.2 Ground reaction forces and foot contact pressure
The three basic components of ground reaction between the foot and floor
during the stance phase of gait are defined as vertical, fore-aft, and
mediolateral forces (Fig. 2). The entire ground reaction force pattern has
been used as a descriptive criterion for the study of human gait [6,71. In
addition, values can be identified from these curves for parametric
analysis. A total of nine force parameters, in percentage of body weight
(FI-F9), and their chronological incidence of occurrence, expressed in
percent of ~tance phase period (TI-T9), can be defined.
Additional variables such as heel strike impulse [81, foot-ground
vertical impulse [9,10,111, and foot center of pressure distribution
[11,12,13,14,151 can be determined from the data curves. These specialized
definitions are more suitable for investigation of foot mechanics.
Foot to ground contact pressure distribution is a potentially useful
addition to human gait analysis. Static pressure distribution data
obtained using a Harris mat or pedoscope only provide qualitative
assessment of foot pressure during standing [16,171. The measurement of
dynamic foot to ground pressure during gait has been studied recently
[18,19,20,211. Quantitative data can be calibrated using ground reaction
force measurement data displayed as contour plots for the entire stance
phase of gait (Fig. 3). Different subregions of the foot can be divided to
evaluate corresponding pressure intensity and foot anatomy (Fig. 4). These
pressure values and their distributions can be used to determine the
resultant forces applied to different foot regions for characteristic
pattern analysis [211.
 35

Figure 2. Basic parameters to F3

describe foot-floor reaction forces. 100

The force directions are defined as ~
the forces acting from the foot to
the floor. Therefore, the negative
..
iii

ai
80

60

.:e
magnitudes represent the aft and ~
.E
medial force components. 40
u
..
::.
20

0 0 20 40 60
% of stance
15
~

....-
iii 10
5
~ 0
.E
:: -5
\"
~ -10
-15

Figure 3. Graphic display of foot to -20

ground pressure distribution during ~ 15
gait. Pressure contour plots are
shown as a function of percent of ..
iii

ai
10
stance phase. A composite of the 5
~
pressure distribution history .E 0
throughout the stance phase is also -5
illustrated as the "peaks" plot. ~.. -10
::E F7

i,

PEAKS
~
12.77% 17.02%
~ 0

21.28%
~
~29.79%
~lJ ~ ~U '\~
l,j
46.81% 51.06% 55.32%

~'IJ ~
\~
~
\~
U
59.57% 65.96% 72.34% 78.72% 85. 11 % 91.49%
36

Figure 4. Graphic display of foot to ground pressure distribution during

gait. A color-coded foot pressure distribution plot for the entire stance
phase. The foot planar surface is divided into six anatomic areas:
Medial Hindfoot (MH), Lateral Hindfoot (U!:), Medial Forefoot (MF),
Lateral Forefoot (LF), Big Toe (BT), and Small Toes (ST).

2.3 Joint rotational angles

Based on a convenient coordinate system established for joint relative
motion analysis [22,23], three-dimensional rotation of the hip, knee, and
ankle can be described simultaneously. If relative joint rotation only
occurs in one of the traditional planes (sagittal, frontal, transverse)
from the neutral position, the current definition would match traditional
anatomical definitions offlexion-extensio.n, abduction-adduction, and
internal-external rotation (Fig. 5). The key parameters from knee joint
motion curves are defined (Fig. 6). Similar definitions of these
parameters for the ankle and hip joints have been established. In general,
total stance flexion ($3) has a magnitude similar to terminal stance
flexion ($4). Total swing phase flexion (<1>5) and total sagittal motion
(<P6) are also very close. For this reason, <1>3 and $5 were not measured.
In addition, double stance and single stance joint motion data, which can
be determined from temporal distance factors, are useful parameters for
clinical assessment of abnormal subjects.
 37

Figure 5. Knee joint axes and planes of rotation as illustrated in a

three-dimensional perspective view. 'the flexion-extension axis (1-1) is
fixed to the femur, the inte-rnal-external rotation axis ('3-3) is fixed to
the tibia, and the abduction-adduction axis is perpendicular to both of the
otber axes.

2.4 Dynamic electromyography (EMG)

Dynamic EMG measurement of~scular activity during walking can provide
useful inormation on muscular function and coordination. Basically, EMG
data are used to assess on and off actions of certain muscles or muscle
groups during gait. Different instruments and data analysis tecbniques can
also provide the magnitude of muscle contraction on a relative basis. It
bas been a problem to correlate muscle contractile forces with EMG
measurements.. Limited success was accomplisbed only under isometric
functions to estimate muscle forces based on rectified and integrated
muscle EMG [24,25,26]. Although dynamic measurements of EMG have been used
to justify surgical and therapeutic procedures in lower extremity
pathological conditions, no successful attempt has been reported to
quantitate. muscle force during gait. based on electromyograpbic data alone.

3. GAIT MEASUREMENT TECHNIQUES

3.1 Temporal distance factors
A long walkway of 10 m or more in length is necessary for gait analysis
in order to provide ample space at bot.h ends to allow the subject to walk
at a nearly constant velocity. Optical switcbes on other equivalent
38

60
a,
CD
'tI
r::~
40
.2III
...r::
CD
IoC
CD 20
.....
r::
.2
IoC
CD
Li: 0

10

J
-5L-~--~--~~~~~~~1~00~
o 20 40 80

L % of gait/CYCle
Stance : Swing

Figure 6. Basic parameters defined in the study of knee joint rotation in

three dimensions.

instruments placed at either end of the effective walkway length are

required to initiate and terminate the period of gait data transmission and
analysis.
Accurate measurements of step length and step width are difficult to
acquire, especially with the limitations of available optical methods for
motion assessment. In our laboratory, we employ an instrumented walkway
combined with special foot switches to record these measurements. The
middle portion of the walking path has two sections of instrumented
stainless steel tape mats to allow accurate measurement of step length and
step width (with a resolution of 2 em or % 1 cm) for either foot (Fig. 7).
Walking speed is calculated from the signals produced by the optical
switches (averaged speed) and verified by tape mat measurements.
Custom-made foot switches are placed on the soles of the shoes at
locations of the big toe, first and fifth metatarsal heads, and beel. The
heel switches have a metal tag to allow direct contact with the tape mat
for step length and step width calculations. A self-contained electronic
control unit has been de-signed to provide foo-t-ground contact patterns as
well as step length and step width calculations.
 39

Figure 7. Gait laboratory walkways (level and varied ground conditions):

a = piezoelectric force plates, b = instrumented stainless steel mats for
the measurement of step length and step width, and c = infrared light
switches used in the measurement of walking velocity and as triggers for
computer sampling.

3.2 Ground reaction force and foot contact pressure patterns

Two force plates are located in the center of the walkway and are used to
obtain single and double support ground reactions. These plates have a
minimum resonant frequency of 100 to 200 Hz, which is adequate for
recording the main responses between the foot and ground during contact.
They are placed in sequence along the walkway with a 20-em centerline
offset to facilitate clean contact upon loading. This arrangement also
provides for simultaneous observation of double support.
Foot contact center of pressure analysis can be performed using a
transparent force plate with a video camera underneath the plate to record
the images of foot or shoe contact with the plate (Fig. 8). The center of
pressure (point of application of the resultant ground reaction force) as
well as vertical impulse between the foot and ground can be quantitated for
different regions of the foot (Fig. 9). The instrumented force plate
provides the magnitude and location of the center of pressure for each
instance during stance phase in reference to the geometry of the plate.
The video image of the foot or shoe on the plate can then be used to
overlay these data on the contour of the foot for data analysis.
Dynamic foot pressure has been measured using various instruments and
devices containing strain-gauged parallel beams, a matrix of load cells,
photoelastic sheets, and piezoelectric crystals [27,28,29,30]. However,
the method of Critical Light Reflection provides the most effective means
to calibrate pressure intensity as well as to determine localized contact
force [19,21,31]. Fluorescent light is transmitted through the transparent
force plate to obtain images in gradients of gray when the foot is in
contact with the rough rubber mat that covers the surface of the force
plate (Fig. 10). A video analog to digital converter and minicomputer are
used to express the intensity of images in different levels of gray. Using
force plate reaction data from different instances of the stance phase, a
40

Figure 8. Lecation of the video camera und.erneatb the transparent f.orce

plate to record dynamic foot con:tact pressure images f.or quantitative
analysis. The ambient light is used for picture taking purposes only. The
force plate is situated in the middle of an instrumented walkW'ay, and "the
pit is covered with an identical floor covering material during gait
evaluation. The surface of the transparent: plate is cQvered with a robber
mat to enhance the video image for data collection.

pressure density scale factor is determined to quantitate the distribution

of foot to ground contact pressure. Integration of the contact area and
pressure intensity can be used to study the contact forces i11. .different
regions of the foot during gait (F'ig. 3, 4). These data represent a
continuous re.cord of foot contact pressure in normal and abno'rmal
conditions.
3.3 J.oiot a:ngular motion measurement
Several techniques have been described in the .past for the measurement of
human joint motion. T:bese included (1) interrupted light or high-speed
stereophotographic method [32,33,34], (2) television-computer method
[35,36,37,38}, (3) accelerometric method [39,40], (4) stereosonic method
[41], (5) electro-optical method [42,34,44], and(6) exoskeletal linkage
method [45,46,47}. Eac!:! of these methods has advaatages and disadvantages
[.48]. Effecti ve application of any method depends largely on the
refinement of the instruments, technique involved, as well as experience of
the investigator.
Comparison of various methods based on grpes of parameters, illstrument
attachment convenience, equipment cost, on-line computer analytic
capability, and accuracy of data was performed [49}. This comparison
concentrated more on clinical application aspects of gait rather than basic
research on human joint kinematics. One of the unique features of
utilizing photographic, video, and electro-optical methods is the ability
to simulate gait with stick diagrams on a television monitor. Although
suc'h a visual aid appears to be attractive, its value for objective data
analysis has not been demonstrated. Visual analysis of stick diagrams
tends to be subjective, and its usefulness must be carefully evaluated.
The exoskeletal linkage method utilizes devices attached to the
limb-joint system to measure rotation and translation while the human body
performs daily activities. Rotary and translational potentiometers are
 41

120

100

"~'"
~
,. 80

"0
.Q

i:
~
60
"ClQ;
.~
40
~

~"
20

Vertical impulse in percent of total

47
Figure 9. Vertical impulse of the hind-, mid-, and forefoot areas expressed
in percent of the total vertical impulse: the example shown is a normal
subject with shoes.

usually used for analog measurement of joint motion, with no further

reduction of data. However, measurements obtained with certain spatial
linkage designs must undergo computer analysis before the actual joint
motion data can be realized.
The triaxial electrogoniometer has been used to measure three-dimensional
joint rotation. Theoretical and experimental justification of this
instrument has been presented previously [47]. The advantages and
disadvantages of this method compared to other techniques have also been
discussed. The unique ability of this apparatus to measure instantaneous
three-dimensional joint relative motion is the main reason this method was
adopted to manage our relatively large sample population.
Although improper goniometer placement can result in error, a trained
evaluator with the aid of an instrument mounting guide can limit such
error. The reproducibility of data has been demonstrated by test-retest
experiments, and variations of gait patterns in test subjects have been
found acceptable. Normal cross talk due to external placement of the
device can be corrected mathematically. Physical interference with the
instrument does not seem to cause any significant change in normal gait
function.
42

Foot pressure
Plastics
111111111

Total internal reflection Internal light scattering

f(jJ at interface at interface
87507X-7

Figure 10. A schematic diagram illustrating the principles involved in the

Critical Light Reflection method to obtain optical images of foot to ground
pressure distribution during gait for quantitative analysis.

4. DATA REDUCTION AND ANALYSIS

4.1 Data sampling and reduction
All gait analysis systems, regardl'ess of joint motion and ground reaction
force recording method, require a computer for data collection, reduction,
and analysis. However, a backup data recording device is important in case
the computer temporarily fails to function. Under such circumstances, hand
reduction of the data can be performed without losing vital information
from the experiment or examination. Any system that requires a computer to
retrieve raw data is subject to the danger of missing test results.
Precautions to prevent this occurrence must be carefully considered when
designing a gait analysis system.
Only the exoskeletal linkage method using the electrogoniometer system is
capable of providing joint angle measurements in analog form. This offers
a significant advantage for efficient data reduction as well as permits the
capability for manual data analysis in case of a computer system breakdown.
In our laboratory, bilateral joint evaluation is available which requires
multiple data recording channels in order to obtain complete
synchronization of gait information under the same time reference.
For automated gait analysis, 26 channels of analog data are digitized
through the AID converter of a FDP 11/34 digital computer. A 12-channel
Beckman strip chart recorder is used to provide visual inspection of the
analog data recorded from the goniometers, force plates, foot switches, and
tape mats. An instrument panel capable of monitoring all test devices is
designed to check test conditions before and during each gait run. A
computer terminal (VT-55 monitoring unit) is also used to provide immediate
feedback of data from the main computer to ensure correct data entry. This
sampling rate of the AID converter is set at 100 Hz. This rate is by no
means optimal, but judging from our data acquisition system and the amount
of data processed, the current rate is considered acceptable. All analytic
results are available for reporting one hour following completion of the
 43

evaluation. Efficient data analysis and documentation are essential in any

gait system. Flow chart diagrams that illustrate basic instrument
organization and data collection, reduction, and analytic schemes for our
gait laboratory are shown in Figure 11.
4.2 Parametric analysis
All parameters identified for the normal population should be subjected
to routine statistical analysis to obtain means and standard deviations.
Linear correlation matrices are utilized to identify independent gait
variables. The student t-test is used to examine the effects of age and
sex on gait characteristic measurements in varioQs subject groups.
A preselected set of parameters is used to calculate a gait symmetry
index (Is) to determine the presence of any significant side dominance
during normal level walking [50,51]. Different weighting schemes are
applied among all parameters to identify the proper combinations of gait
data for generation of this index.
Due to the comprehensive nature of gait analysis, voluminous data are
generated on each subject which make comparative analysis difficult. It is
helpful to identify the parameters most predictive of gait function so they
can be combined algebraically to form a performance index (I p ). This is
attractive for two reasons: (1) it facilitates data analysis and
comparison and (2) it provides a better means to evaluate a subject's
overall joint function based on the combined effects of selected gait
parameters.
Preoperative and normal data are identified a priori, divided by sex, and
then subjected to the R-squared method of analysis [50,52]. The R-squared
procedure is similar in nature to the stepwise regression procedure;
however, it examines all possible combinations of variables and selects the
best set that can best sep'arate normal from abnormal gait data.
Discriminative analysis will be applied so that the parameters can be
properly rated to generate the performance index I . This value is
normalized so that on a scale of 1 to 100, 50 is the demarcation line
between the normal and abnormal results.
4.3 Nonparametric analysis
There are two main purposes for performing such analyses. The first is
to derive repeatable and reliable variables which can be used as
substitutes for the parameters defined previously. Additional important
parameters may then be identified. Nonparametric analysis includes not
only the magnitude but also the wave form of gait signals and allows
objective evaluation. The second purpose is to present the data in
closed-form expression, so that all individual curves can be conveniently
averaged. In addition, storage of raw data in the computer becomes more
efficient.
The Fourier method of analysis is used, since gait data are periodic, and
this method has been successful in past studies of human gait [53,54].
Ground reaction forces or joint motion patterns can be expressed as
periodic functions of time. Coefficients of the harmonic series (Fourier
Series) are used as additional gait parameters. Given this set of
coefficients for a gait pattern, the original curve of joint motion or
ground reaction force can be easily reproduced. These coefficients for
each gait pattern are used for averaging purposes. The method of selecting
the essential number of harmonics depends on the sampling rate and accuracy
required to reconstruct the original data curve. These essential Fourier
coefficients are used to develop the joint performance index in order to
separate two subject populations in gait analysis.
44

AID
Converter
(64 channel)

Synchronize
Trigger

737I3X-'

PDP 11/34a (256kb)

Computer System

Figure 11. a) Flow diagram for gsit analysis instrument organization and
data reduction scheme. b) Flow chart describing the computer system
structure for gait data reduction, analysis display, and storage.
 45

5. APPLICATION OF GAIT ANALYSIS

Gait analysis is generally used for two main purposes. First, it can be
utilized to characterize the functional behavior of a specific population
group in terms of the gait parameters previously defined. Such study will
-help us further understand the scientific basis of human gait. A similar
approach can also be performed on two different population groups to
establish the differences between them. These population groups can be
either normals or patients with known pathological conditions. The data
obtained is expected to provide criteria to differentiate abnormal patients
from normal subjects, as well as to numerically grade the degree of
abnormality. Basic science and clinical information can be generated to
aid understanding of the anatomical, neurological, and biomechanical
prerequisites of gait. Such analysis can also identify the key parameters
or indices to study individual patients.
The secOnd purpose of gait analysis is to study each patient's
functional performance so that proper treatment can be recommended or
evaluated. In this case, a patient's pretreatment gait data can be nsed as
a ~ontrol for comparison with subsequent evaluation results. This aspect
of gait analysis application has the greatest potential for making it an
acceptable clinical tool in the treatment of patients with musculoskeletal
and neurological control problems. However, without an adequate data base
or established key parameters, it is nearly impossible to interpret the
results obtained from individual patients. Consequently, a proper balance
should be maintained in a gait laboratory whereby both objectives are
pursued in a well-organized manner. Population-based studies can establish
the methodology, key gait parameters, and necessary data base, while
individual patient evaluations can provide a critical test of the
experimental devices and currently developed data reduction and analysis
schemes.
The basic areas of gait analysis application can be summarized as
follows. For brevity, references are not cited because of the large number
involved.

5.1 Establishment of normal data base

This class of application is intended to accumulate normal data on key
gait parameters from the normal population, including variations due to
sex, age, and other physical and biological factors. Basic kinematic data
and ground reaction forces provide the information that is essential to
determine joint loads. Efforts were made to identify the key determinants
of gait performance which could be used to correlate with all other
parameters. Evaluation of walking speed, energy consumption, etc., have
been tried but only with limited success due to the complexity of gait as a
whole. Currently, tbe use of a combination of gait parameters to form a
"performance index" appears to be most attractive for this purpose.
Accumulation of normal data is important for the development of a
standardized reference base that can be used for comparison with abnormal
results. Due to the difficulty involved in data retrieval and analysis,
the temporal and distance parameters are most frequently analyzed based on
a population of sufficient number to overcome anthropometric variation.
Ground reaction forces and joint motion data are relatively less with
smaller sample sizes. Information on pressure distribution under the foot
or within the shoe during gait has been rather scarce, with limited reports
on selected populations and walking conditions. Much more effort is
required in this area of research, particularly with the use of modern
instruments and data analysis t~chniques.
46

S.2 Comparison of norm<rl and abnormal gait performance

This represents the most common application and has demonstrated the
clinical relevancy of gait analysis. Different diseases and functional
limitations have been studied, but the sizes of patient groups were rather
small and could not provide enough detailed information. As disease
processes are complex and have wide variations in the degree of
involvement, only the most general conclusions can be drawn from gait data
which is considered insufficient to demonstrate the specific value of gait
analysis as a clinical evaluation procedure. With the new instruments
available today, it is very possible that gait analysis can assist in
differentiating disease type and severity in patients with walking
abnormalities. However, more effective data analysis methods are needed to
accomplish this task.
S.3 Evaluation of treatment methods, drug therapies, and assistive devices
The most reliable and effective utilization of gait analysis lies in this
class of application. Since patients' pre- and posttreatment results are
more compact, statistical methods of data analysis can be used to thus
provide definitive conclusions. Many surgical procedures, drug treatments,
therapeutic regimens, and assistive devices have been studied and compared
to provide physicians and surgeons with clear knowledge to aid them in
selecting proper care for patients with specific musculoskeletal joint
problems. This application alone can justify the use of gait analysis, as
it is the only method capable of determining and quantitating minute
differences in joint functional performance duI:"ing walking that cannot be
observed visually.
S.4 Assessment of individual patients in clinical follow-up
Patients' lower extremity performance can be effectively monitored using
objective gait analysis. Such information can assist physicians and
surgeons in determining the proper timing and treatment strategy for their
patients. Certain walking abnormalities cannot be diagnosed precisely
using routine medical examination. Gait analysis has been demonstrated to
be effective and useful. However, such cases are sporadic and cannot be
generalized to reflect the reliability of this procedure. More effort must
be devoted to this aspect in the future in order to fully justify the value
of gait analysis as a clinical tool. The true challenge for those working
in the field of gait analysis may very well be in this class of
application.

6. GAIT ANALYSIS TODAY AND TOMORROW

The current gait literature contains primarily the methodology and
statistical results of population-based studies. Patient related
investigations are more demanding, but effective instruments and data
reduction techniques are available to fullfil this purpose well. However,
gait characteristics, even among normals, tend to vary widely. Hence,
sufficient sample sizes must be maintained in order to draw viable
conclusions. Due to the time and cost limitations in each laboratory,
coordinated studies must be performed so that a common data pool can be
established. To achieve this aim, a standardized data collection scheme
and parametric definitions are essential.
The major application of gait analysis today still remains to be the
accumulation of data to form bases for normals and patients with
well-defined clinical symptoms. However, preliminary results on individual
cases have demonstrated the potential value of using this evaluation
technique as a clinical tool. This may be the ultimate goal for gait
analysis in order to justify the cost of such an expensive setup within
medical institutions. The future of such scientific endeavors will greatly
 47

depend upon this development. Otherwise, gait analysis will only be

regarded as a scientific and clinical research tool.
To be clinically acceptable, gait analysis must fullfil the following
requirements:
1. Measured gait parameters must correlate with the patient's functional
abnormalities.
2. The data produced cannot be observed by routine clinical examination.
3. Gait evaluation techniques must be sensitive enough to distinguish
minute differences in functional changes.
4. The instruments and laboratory environment should not alter the
patient's functional behavior.
5. Data must be reliable, accurate, and reproducible.
6. Results should be condensed to facilitate easy communication wixh
physicians.
In most gait evaluation methods available today, the above basic
requirements can be achieved with additional effort. Time and expense
should not be invested to develop new instruments, computer systems, etc.
More effort should be directed toward the accumulation of population-based
results and the development of convenient methods for gait data analysis.
Individual case reports which demonstrate the effectiveness of gait
analysis to provide useful clinical information must be shared among
investigators to broaden the scope of gait application.
Any objective functional evaluation of human musculoskeletal systems,
including gait analysis, contains unique advantages. Besides being able to
provide measurements of joint function in quantitative values, gait
analysis offers result documentation for future comparison and reference.
Studying patients in a laboratory environment also tends to eliminate human
factors related to the patient as well as the treating physician, which
should thus make results more reliable. Finally, the gait laboratory can
be used as a service unit to provide the lengthy functional assessments of
patients as part of preexamination workups for physicians in the more
complex cases. It is also hoped that objective evaluation can help to
establish better subjective assessment schemes which may provide reliable
functional data to clinical institutions that do not have gait analysis
facilities.
The future of gait analysis is certainly very promising in many aspects
of clinical application, not including its established value as a basic
science and clinical research tool. If one wishes to expand the use of
gait analysis to more practical applications in the clinical environment,
the following questions may provide guidelines for achieving this goal.
1. Are the data generated reliable and reproducible?
2. Can the test data be duplicated elsewhere using a different method?
3. Do gait analysis results reflect clinical findings?
4. Can gait evaluation provide more information than physical examination?
5. Are the data easy to understand and can they be made available in a
short period of time?
6. Can the information generated help in making therapeutic decisions?

REFERENCES

1. Contini Rand Drillis R (1954) Biomechanics. Appl Mech Rev 7:49-52

2. Steindler A (1953) Historical view of the studies and investigations
made in relation to human gait. J Bone Joint Surg 35[A):540-542
3. Pauwels F (1948) The significance of the architectural principle of the
supporting and locomotor systems in regard to stresses of long bones.
Z Anat Entwgesch 114:1-2
48

4. Perry J (1974) Clinical gait analyzer. Bull Prosthet Res 10:12:188-192

5. Perry J, Hoffer M M, Giovan P, Antonelli D and Greenberg R (1974) Gait
analysis of the triceps surae in cerebral palsy. J Bone Joint Surg
56[A]:511-520
6. Hargraves P and Scales J T (1975) Clinical assessment of gait using
load measuring footwear. Acta Orthop Scand 46:877-895
7. Jacobs N A, Skorecki J and Charnley J (1972) Analysis of the vertical
component of force in normal and pathological gait. J Biomechanics
5:11-34
8. Simon S R, Paul I L, Mansour J, Munro M, Abernath P J and Radin E L
(1982) Peak dynamic force in human gait. J Biomechanics 14: 817-822
9. Stott J R R, Hutton W C and Stokes I A F (1973) Forces under the foot.
J Bone Joint Surg 55[B]:335-344
10. Cavanagh P R and Lafortune M A (1980) Ground reaction forces in
distance running. J Biomechanics 13:397-406
11. Katoh Y, Chao E Y, Morrey B F, Laughman R K and Schneider E (1981)
Objective evaluation of painful heel syndrome by gait analysis. IN:
Proceedings of the VIIltb International Congress of Biomechanics.
20-24 July, Nagoya, Japan. Human Kinetics Publishers, Champaign, IL
12. Arcan M and Brull M A (1976) A fundamental characteristic of the human
body and foot, the foot-ground pressure pattern. J Biomechanics
9:453-457
13. Grundy M, Tosh P A, McLeish R D and Smidt L (1975) An investigation of
the centers of pressure under the foot while walking. J Bone Joint
Surg 57[B]:98-103
14. Elftman H A (1934) A cinematic study of the distribution of pressure in
the human foot. Allat Res 59:481-491
15. Scranton P E and McMaster J H (1976) Momentary distribution of forces
under the foot. J Biomechanics'9:45-48
16. Chodera J (1957) Examination methods of standing in men. F U CSAV
Praha 1: 3
17. Sharmu M, Dhanendran M, Hutton W C, and Corbett M (1979) Changes in
load bearing in the rheumatoid foot. Ann Rheum Dis 38:549-552
18. Johnson F and Kong D M C (1981) Comparative display of dynamic
foot-ground pressure data (pelmatobarography). Eng in Ned 10:159-161
19. Duckworth T, Betts R P, Franks C I and Burke J (1982) Measurement of
pressure under the foot. Foot & Ankle 3:130-141
20. Cavanaugh P Rand Michiyoshi A E (1980) A technique for display of
pressure distribution beneath the foot. J Biomechanics 13:69-75
21. Spiegl P, Cass J R, Bleimeyer R, Cahill P and Chao E Y (1985) Dynamic
foot pressure in normal subjects. In Transactions of the 31st Annual
Orthopaedic Research Society, January 1985, Las Vegas, NV. Vol. 10,
p. 360
22. Grood E S, Suntay W J, Noyes F R, Butler D L, Miller E H and Malek M
(1979) Total motion measurement during knee laxity tests. IN:
Transactions of the 25th Annual Orthopaedic Research Society, Vol. 4,
p. 80
23. Chao E Y (1980) Justification of triaxial goniometer for the
measurement of joint rotation. J Biomechanics 13: 989-1006
24. DeLuca C J and Forrest W J (1973) Force analyses of individual muscles
acting simultaneously on the shoulder joint during isometric abduction.
J Biomechanics 6:385-393
25. Cnockaert J C, Lensel G and Pertuzon E (1975) Relative contribution of
individual muscles to the isometric contraction of a muscle group.
J Biomechanics 8:191-197
 49

26. An K N, Cooney W P, Chao E Y, Askew L J and Daube J R (1983)

Determination of forces in extensor pollicis longus and flexor pollicis
longus of the thumb. J Appl Phys 54{3):714-7l9
27. Hutton W C and Drabble G E (1972) An apparatus to give the
distributions of the vertical load under the foot. Rheum and Phys Med
2:313-317
28. Dhanendran M, Hutton W C and Parker Y (1979) A quantitative assessment
of normal and pathological foot function. Engrg in Med 8:69-74
29. Stokes I A, Stott J R and Hutton W C (1974) Force distributions under
the foot--A dynamic measuring system. Biomed Engrg 9:140-143
30. Scranton P E Jr, Bootman B H ~nd McMaster J H (1980) Forces under the
foot: A study of walking, jogging, and sprinting force distribution
under normal and abnormal feet. In Foot & Ankle, New York, B C Decker
31. Betts R P, Austin I G, Duckworth T, Crocker S and Moore S (1980)
Critical light reflection at a plastic/glass interface and its
application to foot pressure measurement. J Med Eng Tech 4:136-142
32. Murray M P (1967) Gait as a total pattern of movement. Am J Phys Med
48:290-333
33. Milner M, Brennan P K and Wilberforce C B A (1973) Stroboscopic
polaroid photography in clinical studies of human locomotion. S Afr
Med J 47:948-950
34. Crowninshield R D and Brand R A (1981) A physiological based criterion
of muscle force prediction in locomotion. J Biomechanics 14: 793
35. Winter D A, Greenlow R R and Hobson D A (1972) Television-computer
analysis of kinematics of human gait. Comp Biomed Res 5:498-504
36. Cheng I, Koozekanani S H, Olson K Wand Burnett C N (1974) Computer
television analysis of human locomotion. In Proc 27th Annual Conf
Engineering in Medicine and Biology Vol. 16, p. 454
37. Simon S R, Trieshmann H W, Burdett R G et al. (1983) Quantitative gait
analysis after total knee arthroplasty for monarticular degenerative
arthritis. J Bone Joint Surg 65[Aj:605-613
38. Taylor K D, Mottier F M, Simmons D W et al. (1982) An automated motion
measurement system for clinical gait analysis. J Biomechanics
15:505-516
39. Morris J R W (1973) Accelerometry--A technique for the measurement of
human body movement. J Biomechanics 6:729-736
40. Hayes W C, Feldman J M, Oatis C and Nixon J E (1978) Gait analysis of
multiaxial accelerometry. In Transactions of the 24th Annual
Orthopaedic Research Society 3:104
41. Progress Report No.2 (1973) Krusen Center for Research and
Engineering, Moss Rehabilitation Hospital, Philadelphia, PA,
Dec. 1972-Nov. 1973
42. Tandon Sand Marsalais E (1975) Three-dimensional stroboscope gait
analysis (the use of the Selspot System). In Transactions of the 21st
Annual Q'rthopaedic Resean:h Society
43. Conati F C (1977) Real-time measurement of three-dimensional multiple
rigid body motion. M. S. Thesis, Massachusetts Institute of Technology,
Cambridge
44. Andriacchi T P, Galante J 0 and Rermier R W (1982) The influence of
total knee-replacement design on walking and stair climbing. J Bone
Joint Surg 64[Aj: 1328-1335
45. Lamoreux L A (1971) Kinematic measurements in the study of human
walking. Bull Prosth Res 3:10-15
46. Cbao E Y, Rim K, Smidt GLand Johnston R C (1970) The application of
4 x 4 matrix metbod to the correction of the measurement of hip joint
rotation. J Biomechanics 3:459-471
50

47. Chao E Y and Hoffman R R (1977) Instrumented measurement of human joint

motion. ISA Transactions 17:13-19
48. Chao E Y (1978) Experimental methods for biomechanical measurement of
joint kinematics. In CRC Handbook of Bioengineering in Medicine and
Biology 1:385-411
49. Stanic U et al. (1976) Standardization of kinematic gait measurements
and automatic pathological gait pattern diagnostics. Rehabil Serv
Admin Workshop Report, Locomotion and the Clinical Analysis of Gait,
Philadelphia, PA, pp. 102-119
50. Chao E Y, Laughman R K and Stauffer R N (1980) Biomechanical gait
evaluation of pre and postoperative total knee replacement patients.
Arch Orthop Traum Surg 97:309-317
51. Soudan K (1982) Standardization of gait kinematic data using a gait
symmetry index and Fourier analysis. In Biomechanics: Principles 'and
Applications (Huiskes R, van Campen D H and de Wijn J R, Editors)
Martinus Nijhoff, The Hague, The Netherlands. pp 135-140
52. Laughman R K, Stauffer R N, Ilstrup D M and Chao E Y (1984) Functional
evaluation of total knee replacement. J Orth Res 2:307-313
53. Alexander R M and Jayes A S (1980) Fourier analysis of forces exerted
in walking and running. J Biomechanics 13:383-390
54. Schneider E and Chao E Y (1983) Fourier analysis of ground reaction
forces in normals and patients with knee joint disease. J Biomechanics
8:591-601
 51

BIOMECHANICS OF SPORTS - CURRENT PROBLEMS

W. Baumann
Deutsche Sporthochschule Koln, Institute for Biomechanics,
Koln, Germany

Biomechanics of sports deals with an important social pheno-

menon of the modern world. It has come into prominence since
the late sixties and was focused on two main objectives:
- to analyze the human body and its sports movements in
order to gain a better understanding of the mechanisms
and principles underlying these movements.Violent sports
movements allow to study the human body under extreme
conditions and therefore are interesting also in ortho-
paedics and rehabilitation,
- to improve the performance of the athletes utilizing
scientific methods, taking into account the load toler-
ances of the human body as supplementary condition.
As not unusual, the demand from the practical side preceeded
the scientific interest: great expectations of coaches and
officials in national and international sports organisations
were and still are a very strong motive to carry out biome-
chanical analyses. Another important demand came from the
manufacturers of sports goods and equipment. So everything
moving in sports has been investigated: human bodies, parts
of the human body, real and fictitious points of bodies,
rackets, balls, shoes etc. A real host of data has been
gathered up to now. Parallel to this a certain development of
measuring and data processing techniques took place, mainly
characterized by faster and more comfortable procedures, to
a limited extent by an improved quality of the measurements.
The progress in the development of biomechanical theories
however seems not to be very satisfactory. One may suppose
that the masses of data have not served too much as a basis
of more fundamental scientific research, but mainly for the
technical solution of a detailed problem, which not necessar-
ily had any inner coherence with a more comprehensive issue.
I want to illustrate these statements by some considerations
concerning 1. the application of biomechanics to sport and
2. the determination of muscle forces.

1. APPLICATION OF BIOMECHANICS TO SPORT

In this area the main objectives may be defined as bio-
mechanical characterizations of
- the mechanical structure of the human body
- the most effective sports techniques
- the force/power capacities of the muscle.
I will make some comments related to the last two items.
52

The analysis of sports techniques starts with a description

of the movement by meas of kinematic and dynamic parameters
related to human body characteristics. The majority of the
efforts of biomechanists has been devoted to this field.
Athletes of different levels of performance have been filmed,
the movements digitized and kinematically described. These
data have served as input for many cross sectional studies.
With various analysis of variance techniques it was tried to
establish biomechanical profiles of different levels of per-
formance. Thus an enormous amount of valuable information has
been collected. The affinity of objective kinematic data with
the coaches' subjective and qualitative analysis of the move-
ment facilitated and supported the acceptance of biomechanical
informations by coaches and physical education teachers. To-
gether with the relatively simple methods to obtain planar
kinematic descriptions of movements - even during competitions
- however, this entailed that too much preference was given to
the kinematics and - on the other hand - that the kinetic
structure of the movement was considered less attractive.
From hence the advantages of complex analyses have not been
used too oftenly. There are lots of examples demonstrating
that just the analysis of ground reaction forces is more
powerful in detecting differences between two performances
than any kinematic description. Sometimes the kinematics of
different movements does not show any detectable difference,
but the reaction forces give a very clear separation - and
information more appropriate for interpretation. This of course
is not the result of a higher accuracy of force measurements
(many kinematic parameters concerning certain points or seg-
ments of the body can be measured more exactly than reaction
forces), it is according to mechanical laws, which define the
kinematics always as the smoothed output of a much more differ-
ent and distinctive kinetic control. Therefore forces are the
most sensitive indicators, reacting even to slight changes in
movement technique. There is no doubt, that the limited reach
of kinematic analyses will be further confined, when the 3-
dimensional movement is analyzed in a 2-dimensional projection.
On the other hand the efficiency of the kinetic analysis could
be increased substantially, if it would take into account not
only the external but also the internal kinetic quantities:
the forces in muscles, in joints, in ligaments (the problems
connected herewith will be discussed in the 2nd chapter).
In the analysis of sports techniques it seems that the work
is too much concentrated to the kinematics. However, in the
biomechanical study of the human movement as a whole we have
to consider not only the technique but also the force capab-
ility of the active movement apparatus. Compared to the number
of investigations of sports techniques the area of muscular
mechanics has gained much less attention. Moreover, in most
studies devoted to muscle training exercises the accompanying
kinematic and kinetic analysis is relatively poor. It is con-
fined to the movement of the c.g., the vertical component of
the ground reaction force, the knee joint angle. From these
data, it is impossible to derive the essential quantities like
 53

muscular moments in the different joints of the lower extrem-

ity. But in some of those studies even muscular forces have
been calculated. This is not correct and leads to misunder-
standings and misinterpretations of the investigated training
exercises. Sometimes we are working as if sports techniques
would be completely independent of the force and power capab-
ilities of the muscles and of their coordinated control, both
being only two aspects of the same process and very closely
related to each other.
A simple example may demonstrate the problem. Certain exer-
cises such as hurdle jumps or drop jumps are in use in differ-
ent disciplines in order to improve the strength and velocity
properties of the leg extensor muscles. To characterize the
effects of such exercises we can measure the vertical compon-
ent of ground reaction force. Using the relation: force equals
mass times acceleration we can calculate takeoff velocity
(if initial conditions are given). We can vary the jumping
conditions (input) and then correlate this input with teh out-
put, namely the calculated raise of c.g. (or the takeoff velo~
ity). We can calculate further quantities like total external
power and apply the same statistical procedure. In any case
the result will describe the input-output relation of a black
box. This procedure is by far too simple and too faulty to
yield results of any scientific importance. However this
procedure or similar ones are widely used.
A very simple model of a three joint
F,v kinematic chain representing the
lower extremity in the figure may
illustrate the situation. The joints
No I to 3 correspond to the foot,
knee and hip joint resp., the mo-
ments MI-M3 be the muscular moments
acting in these joints. The power
output of the system is calculated
from velocity and force of the pi-
ston, representing the c.g. of the
body (in this setup the force com-
ponents normal to the piston motion
are not considered). It is evident
that a lot of different combinations
of the three moments can produce the
same power output of the system. ~is
means that from the above cited mea-
2 surements at the periphery of the
system one cannot discriminate the
contributions of the different musde
groups. Possible changes in the
overall performance cannot be as-
cribed to certain components of the
kinematic chain. Moreover compensat-
ing positive and negative effects
within the system cannot be detect-
ed at all. In fact the situation is
still more complicated, when ant-
54

agonistic muscles are active. This method is not well suited

to understand the mechanisms and principles underlying train-
ing exercises.
The methods needed must allow a differential diagnosis of
the system especially for the purposes of
- identifying weak/strong links in the kinematic chain,
- comparision of structural affinities between competitive
and training movements.
The study of the affinity of those movements must take into
account different categories of quantities like kinematic,
kinetic and neuro-muscular control parameters. Then it's
presumably one of the most promising fields of direkt appli-
cation of biomechanical research into sport. It overcomes
the non-existing dichotomy between technique and physical
condition.
The differential diagnosis of training exercises will cer-
tainly increase the effectivity of the training process, but
also can reduce the risks of unnecessary high loading of the
movement apparatus during training.
In the application of biomechanics to sport we have solved
a lot of detailed problems. We have not produced too little,
but sometimes the wrong information. When we have not been
very successful in developing systematic procedures to im-
prove performance, we should start thinking again about the
fundamental issues we can make. The efficiency of the technic-
al services given to athletes and coaches depends to a large
extent on a sound scientific basis - at least on a long-term
perspective.
One of the fundamental scientific problems in biomechanics
connected with most of the above mentioned questions is the
determination of the parameters related to muscle forces,
which are treated in the next paragraph.
2. THE DETERMINATION OF MUSCLE FORCES
Muscle forces are the mechanical output of the neuro-muscu-
lar control system and the most important input variables of
the mechanical system of the human body. There are only a few
problems which can be solved without the basic knowledge of
the spatial and time pattern of the muscular forces. Most of
the fundamental problems connected with mechanical work,ener-
gy and power, with efficiency and economy of muscula: co~
tractions can hardly be solved without a clear quantltatlve
picture of the muscle forces involved. The same is true in
the field of application, where the knowledge of the muscular
and joint forces as such is of importance. In orthopeadics
for instance to judge the effects of conservative or operative
treatment of patients or in sports training for the construc-
tion of specific and effective training methods concerning
both technical and conditional factors. In the important area
of injuries and trouble caused by mechanical factors we are
wandering around nearly blind.
 55

If a mechanical engineer would be confronted with one of

the above mentioned problems and the knowledge presently
available about the mechanical system "body", he certainly
would be very frustrated. Well known fact is, that muscular
forces as most of the other inner forces of the body cannot
be measured directly. It is also well known that very oftenly
muscular moments, muscular forces and joint forces are given
in the literature. All these quantities are calculated by
means of a more or less simple mechanical model, for instance
of the leg, which contain the following quantities:
1. mass and mass distribution of all segments up to the
cutting plane,
2. spatial positions and accelerations of the joint and
segmental axes resp. and of the segmental c.g.s,
3. vector of ground reaction force,
4. free moment in the vertical direction
and - if forces be derived from the joint moments - in
addition:
5. lever arms of the muscles as functions of joint angels.
If we forget for a moment the obvious shortcomings of a
correct determination of the above listed input variables
- which pose a methodological problem -, we still have a
system that is mechanically undetermined, since the number of
unknowns exceeds the number of equations decribing the system
- thus posing rather a fundamental problem. To overcome this
deficiency, we make assumptions concerning the muscular for-
ces acting around a joint e.g. the absence of antagonistic
muscular action, which in some special cases can be checked
roughly by EMG-measurements. Mostly, and always when antag-
onistic activity occurs, the moments be qualified as net
muscular or joint moments, and the forces as minimum estimat-
es. Despite the fact that these data are only of limited value
for practical purposes, one can have the impression that bio-
mechanists got accustomed to that situation and arranged
themselves with it. This is astonishing in view of the fact
that the trivial question for the force of the gastrocnemius
muscle cannot be answered. The simplifying assumptions cited
above can only be made with respect to the ankle joint: For
sake of simplicity we consider the sagittal plane, where the
triceps surae muscles and the tibialis anterior muscle be the
only possible force generators. In many cases we can observe
that there is no antagonistic activity. If not, the ratio of
the cross sections of the two muscle groups suggests that the
net force calculated for the achilles tendon should not be
very much affected by the tibialis' activity. In this case,
we are probably very close to true values of muscular moments
and forces in the ankle joint. However, if we continue to
calculate forces in the knee joint, we are confronted with
the problem, how the force transmitted by the achilles tendon
is distributed to the soleus and gastrocnemius muscle resp.
The gastrocnemius acts as a two-joint muscle in ankle and
knee joint , therefore the gastrocnemius force must be known
if forces and moments in the knee joint should be calculated.
This is not the case, so moments and forces in the knee joint
are faulty. From hence further calculations up to the hip
joint are even more questionable.
56

In order to overcome this calamity, mathematical optimiza-

tion procedures have been used by different authors. These
methods are useful tools for the analysis of complex systems,
which are functioning according to certain principles. The
human body represents such a complex system. The central
question hereby is "which principles can be selected as ob-
jective functions, as biologically meaningful optimization
criteria"? In gait research the application of optimizatiop
methods started. NUBAR and CONTINI (1961) minimized mechanical
work, BECKET and CHANG (1968) minimized the total energy ex-
pended and CHOW and JACOBSON (1971) selected the total work
done during a complete cycle of walking as objective function.
Also for level walking SEIREG and ARVIKAR (1975) minimized
linear sums of forces and moments, PEDOTTI et al. (1978)
defined different sums and ratios of forces as optimization
criteria. In an application to a specific sports movement,
the giant swing in gymnastics, BAUER (1983) minimized mechan-
ical energy.
Summing up these approaches we can discriminate three
different categories of objective functions to be minimized:
- total energy expenditure
- net muscular power output
- net muscular forces
There is evidence that the criterion "total energy expended"
holds for cyclic movements of a certain duration (HATZE 1976).
However there is no direct relation to the mechanical forces
produced by the muscles. In addition this criterion is prob-
ably not relevant in maximum effort movements. All the other
criteria neglect completely antagonistic muscular activities
that play an important role in many everyday and sports move-
ments. Why do we minimize linear sums of forces, of squared
forces? Because of biological evidence? Not at all! We select
them to meet the requirements of the optimization calculus.
Although one might find supporting arguments these criteria
must be regarded as to some extent arbitrarily chosen abstract
mathematical functions. There is no evidence, that they are
biologically meaningful optimization criteria. The EMG measu-
rements made in a part of these studies can only be used for
the falsification of an approach, not for a verification.
What we need are objective functions which have a biological
evidence. ROUX' principle (1895) of the functional adaptation
of biological systems defines qualitatively the relations
between the mechanical stimulus on the biological material
and the biological response: the adaptation of the structure
to the function. Then it must be possible to deduce from the
structure the function, that is the mechanical stimulus.
In this connexion - following some earlier hints - STUCKE
(1986) again drew our attention to the work of PAUWELS (1965),
who studied with greatest carefulness the construction prin-
ciples of the human body from an engineering standpoint. His
work - without any doubt - represents a major contribution
to biomechanics and it deserves much more attention than it
received up to now. The collection of papers appeared in
1965, but the first publication dates back as early as 1935,
that means more than 50 years ago.
 57

The main results of his qualitative and quantitative studies

concerning the mechanical properties of bones and the effects
of muscular forces under static conditions can be summarized
as follows:
- the human skeleton is an ideal lightweight construction.
This is a result of a mechanical analysis of the bending
stiffness of the bones. The biological advantages connect-
ed herewith are evident: accelerations of light weight
constructions require less energy and once more the
maintaining of less biological material safes energy too.
- the most dangerous stress for bones is the bending ~ress.
Shape, density and the microstructure of the bones are
such that bending stress is reduced
- the geometrical configuration of the muscles is such
that bending stresses in the bones be reduced.
These findings allow an assessment of the important
working angles and the lines of action of the muscle
forces.
- the effect of several antagonistic muscles acting at a
joint increases the pressure area and thus reduces the
stresses in the joint
- the broad area of insertion of a muscle into the bone
reduces peak stresses and yields a more evenly stress
distribution in the joint.
The last two principles are important when more muscles
are active than minimally required in a certain mechanic-
al situation.
For all these results PAUWELS produced conclusive evidence.
Furthermore there is no reason to doubt that they hold also
for dynamic cases. Until now no contradictory results have
been reported. Even if that would be the case, it would be
very useful to see how high dynamic loads can possibly modify
the established findings. Obviously these principles can be
regarded as a sound basis to deduce "biologically meaningful"
optimization criteria.
In addition the mechanical properties of the muscle, like
length - tension and velocity - tension functions resp. are
of great importance in this connexion. The problem hereby lies
in the exact knowledge and proper mathematical description of
these functions. Finally the neuro-muscular control functions
even if not very clear but in any case represented in the
electromyographic signals are further indispensable parameters
in the validation of the optimization procedure. In this field
too efforts directed to a more reliable quantification of the
EMG signals are urgently needed.
If Biomechanics is not capable of incorporating more of the
esseritial anatomical and neuro-physiological characteristics
of the human body, it does not deserve its prefix "bio" and
it will not arrive at its real goals. In my opinion it would
be worth while to reflect upon these ideas. Maybe that they
can contribute a bit to the improvement of our theoretical
basis in biomechanics of sport as well as in related areas.
58

REFERENCES

1. Bauer, W.L.: Mathematical Modelling and Optimization and

their Influence on Sports Movements - Possibilities and
Limitations. In: Baumann, W.(Ed.): Biomechanik und sport-
liche Leistung - Biomechanics and Performance in Sport,
p. 129-144, Verlag Karl Hofmann, Schorndorf, 1983.
2. Beckett, R. and K. Chang: An Evaluation of the Kinematics
of Gait by Minimum Energy. J. Biomechanics, 1(1968),
p. 147-159.
3. Hatze, H.: The complete Optimization of a Human Motion.
Math. Biosci. 28(1976), p. 99-135.
4. Nubar, Y. and R. Contini: A Minimal Principle in Biomechan-
ics. Bull. Math. Biophys. 23(1961), p. 377-391.
5. Pauwels, F.: Gesammelte Abhandlungen zur funktionellen
Anatomie des Bewegungsapparats. Springer Verlag, Berlin,
1965.
6. Pedotti, A., Krishnan, V.V. and L. Stark: Optimization
of Muscle-force Sequencing in Human Locomotion. Math.
Biosci. 38(1978), p. 57-76.
7. Roux, W.: Gesammelte Abhandlungen tiber Entwicklungsmechanik
der Organismen. Vol. I and II, Wilhelm Engelmann, Leip-
zig (1895).
8. Seireg, A. and R.J. Arvikar: The Prediction of Muscular
Load Sharing and Joint Forces in the Lower Extremities
during Walking. J. Biomechanics 8(1975), p. 89-102.
9. Stucke, H.: Ermittlung von Muskel-und Gelenkkraften mit
Hilfe der mathematischen Optimierungsmethoden. Paper pre-
sented on the Meeting of the Deutsche Forschungsgemein-
schaft (DFG) in Koln on 18th July 1986.
10. Chow, C.K. and D.H. Jacobson: Studies in Human Locomotion
via Optimal Programming. Math. Biosci. 10(1971), p.239-306.
 59

BIOMECHANICS OF TENDONS AND OTHER SOFT CONNECTIVE TISSUES.

TESTING METHODS AND STRUCTURE-FUNCTION INTERDEPENDENCE
A. Viidik
Institute of Anatomy, University of Aarhus, Aarhus, Denmark

1. INTRODUCTION
One major function of soft connective tissue is to provide
the stability necessary for proper joint function, primarily
by constituting their ligaments, and enable the transmission
of forces from muscles to bones, by forming the tendons. Main-
ly uniaxial tensile forces are created in these mostly paral-
1el-fibred structures, when they perform their physiological
functions. Another important function of dense, soft connecti-
ve tissue is to form the three-dimensional meshwork of fibers
in the dermis, the mechanically protective sheath around the
body. Here the force pattern is more complex; biaxial tensile
forces are combined with some compressive force in the third
dimension under physiological conditions. Protection against
violence from the environment presupposes also some ability to
withstand shear forces. The force pattern is also complex in
the connective tissue, which constitutes the soft "skeletons"
of internal organs and forms the scar tissue, restoring the
continuity of a tissues after injury.
A review of the composition of the soft connective tissues
with these various functions reveals that profound differences
exist, besides in fiber arrangements as indicated above, also
in biochemical composition. The most common tissue element is
collagen, a family of macromolecules, which forms fibers, cha-
racterized by pronounced tensile strength. More than ten dif-
ferent collagenous proteins of vertebrate origin have been re-
cognized so far (Miller, 1985), but only three of them (types
I, II and III) have been shown to form regular fibers. Type I
collagen fibers form tendons and most ligaments, while a mix-
ture of type I and III fibers is found in e.g. skin, arteries
and scar tissue. Type II is the collagen of cartilage. Elastin
is the other fibrous protein in connective tissues, found to-
gether with collagen in e.g. elastic ligaments, skin and blood
vessels (Bailey and Etherington, 1980). The ground substance
is present in varying amounts (e.g. minute amounts 1n tendons
and about 20 percent in skin) and contains macromolecules, es-
pecially the water-binding proteoglycans.
The purpose of this chapter is to give a brief overview of
or introduction to soft connective tissue biomechanics. Most
of the space available will be devoted to parallel-fibred col-
lagenous tissues (tendons and ligaments). Points important for
design of research projects as well as critical evaluation of
research reports in current literature are emphasized: Impor-
tant methodological problems will be discussed in some detail.
60

2. BIOMECHANICS OF TENDONS AND LIGAMENTS

The mechanical properties of connective tissues have been
investigated most extensively for parallel-fibred tissues con-
stituted of type I collagen, e.g. muscle tendons and joint li-
gaments. A large number of reports are available on "straight-
forward" analyses of results from tensile testing until failu-
re; for review see e.g. Butler et al. (1978), Elliott (1965),
Harkness (1961, 1968), and Viidik (1973, 1980a, 1987). Deriva-
tion of empirical as well as constitutive equations and model
fitting has been used for analysis of the viscoelastic proper-
ties of these tissues (for discussion see Fung, 1981; Viidik,
1973, 1980a, 1987).

2.1. Methodo1ogica1 considerations

Some 1nvest1gators have devoted considerable attention to
technical aspects, e.g. acquisition of specimens, storage of
them, testing environment, mounting of specimens for testing,
recording of parameters (especially cross-sectional area and
original length of specimens) and subsequent analysis (see
e.g. Butler, 1978; Viidik, 1973, 1979a; Woo, 1982; Woo et al.,
1986a). A number of such procedural "details" in the experi-
mental protocol can influence the results of testing profound-
ly and be the cause of conflicting data as well as render com-
parison between results from different laboratories impossible
or at best difficult. They will be discussed briefly and the
necessity of paying attention to such "details" illustrated by
data on the not so obvious importance of specimen length (see
2.1.2) .
2.1.1. Hand1ing of specimens prior to testing
There 1S no eV1dence for any slgn1f1cant changes occurring
in the mechanical properties of soft connective tissues, when
the blood circulation in the organism is terminated. When re-
moving a tissue from its in situ location by dissection, care
should, however, be taken to avoid subjecting it to mechanical
strain to the extent possible; the previous strain history of
a specimen (whether from random manipulation or systematic pre-
conditioning) is important for its mechanical properties (e.g.
Viidik, 1979a). The water content of a specimen should be kept
constant, since both overhydration and drying will distort the
results. Overhydration will cause an enhancement of the initi-
al phase phenomena (which are overcome by preconditioning, see
e.g. Viidik, 1979a, 1980a) and of the viscous component in the
stationary phase of tissue behaviour; in single cycle tensile
strength testing this is seen as lower "elastic stiffness" and
larger maximum strain values. Drying has the opposite effect.
The ideal milieu for both short term storage and mechanical
testing would be air with 100 percent relative humidity at 32-
37 0 C (depending on the in vivo temperature of the tissue). In
most instances it is, however, sufficent to keep the humidity
over 65 percent at 21 c (Galante~ 1967) or keep the specimen
wrapped in saline-moistened gauze (Viidik, 1968a). Immersion
of a specimen in any of the solutions tried so far increases
the water content of the tissue; e.g. "physiological" saline
(buffered to pH 7.4) (Viidik and Lewin, 1966) and dextran so-
lution and blood plasma (Galante, 1967). For certain types of
specimens, especially those having very small cross-sectional
 61

areas as rat tail tendons, when introduction of overhydration

is unavoidable for technical reasons, all specimens in a seri-
es should be treated in exactly the same way. This bias beco-
mes important, when groups having biochemical differences in
ground substance (i.e. water-binding capacity) are to be com-
pared.
Storage for less than 24 hours between sacrifice and prepa-
ration for mechanical testning should be by leaving the speci-
men in situ. Storage for longer periods of time requires, be-
sides prevention of dehydration, deepfreezing at -20 C; lower
temperatures do not seem to give any advantage (Butler et al.,
1978). The stress-strain parameters do not change significant-
ly by freezing and thawing (Viidik and Lewin, 1966, Butler et
al. 1978, Woo et al. 1986a) nor are there any changes in the
viscoelastic properties after preconditionrng; the hysteresis
is, however, reduced in the first preconditioning cycles (Woo
et al. 1986a).
2.1.2. Specimen geometry and clamping
The lmportance of speClmen length. It is obvious that the
ratio between length and width of a specimen, which is cut out
as a strip from a meshwork, influences the results of mechani-
cal testing: This ratio determines how many of the obliquely
running fibers are cut in the space between the clamps and how
many of them are clamped at both ends (Galante, 1966; Viidik,
1973, 1979a; Nilsson, 1982; Viidik and Gottrup, 1986). That
the length of tendon specimens, in which the fibers run paral-
lel to the axis for indeterminable lengths (supposedly whole
length of the tendon), influences the results of testing has
been realized only recently (Sanjeeviet al., 1982; Haut, 1986;
Viidik, 1986a). This phenomenon will be illustrated here by da-
ta from Viidik (1986a and in preparation).
"Stress"-strain data were calculated for tail tendon fiber
bundles (parallel-fibred type I collagen) from 4-34 months old
male Wistar rats. The specimen length was 2, 5, 10, 20 and 30
mm. It is seen from Fig. 1 that the shorter specimens are more
extensible, less stiff, and require more energy to break than
the longer ones. The influence of specimen length is most pro-
nounced in the interval between 2 and 10 mm. The pattern for
strain at maximum stress is basically the same pattern as that
reported by (Haut 1986), although he tested 10 to 100 mm long
fibers ana found a pronounced fall in extensibility in the in-
terval between 10 and 20 mm. These two reports contradict San-
jeevi et al. (1982), who found the reverse pattern.
This phenomenon is age-independent for the largest part of
the lifespan of the rat, since the same pattern is seen in im-
mature (from 4 months old rats) as well as aged (from animals
34 months old) tissues. The stress-strain curves for the three
different age groups (including the 12 month group of Fig. 1)
are shown in Fig. 2. An age-dependence for very immature ani-
mals can, however, not be excluded; Haut (1983a) showed a very
high and with maturation decreasing glycosaminoclycan content
in tendons from one to three month old animals. This decrease
was concomitant with a decrease in strain rate sensitivity.
These data clearly illustrate that one parameter (in this
case original length) still may have influence on the results,
also when it is seemingly eliminated (here by transforming de-
62

Pmax (N/UC)

'20
0.4

0.2 10

i 0 (!TU1l) (!TU1l)

10 20 30 10 20 30

Wf (N/UC)
tan, (N/UC)

200
400

100
200

'0 (!TU1l) 10 (!TU1l)

10 20 3C 10 20 30

Fig. 1: Parameters from tensile testing of tail tendons (12

month old rats) versus specimen original length: strain at ma-
ximum "stress", maximum "stress", "elastic stiffness" (slope
of the linear part of the curve) and failure energy (ar.ea bet-
ween the strain axis and the curve up to the point of maximum
"stress"). Strain expressed in percent has been used for fai-
lure energy calculations.

"STRESS" (N/UC) 2 !TU1l

30 !TU1l 2 !TU1l
20
30 !TU1l 2 !TU1l

30 !TU1l
,,
,
,,+
15
, ....t 15
)
10 10

10

STRAIN STRAIN STRAIN

r r r
0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4
4 months 12 months 34 months

Fig. 2: Stress-strain curves for tail tendons from 4, 12

and 34 month old rats, specimens with original lengths of 2
and 30 mm. "Stress" calculation, see legend to Fig. 1.
 63

formation to strain). Such "subtle" variations may thus well

explain some the conflicting results in the literature. The
technical procedures must, therefore, be bescribed in detail.
Problems with specimen clamping. A bone-ligament-bone or a
bone-tendon-muscle-(tendon)-60ne specimen is easily clamped by
fixing the bones into contour-shaped clamps without risks for
slippage and/or premature break at a jaw edge (see e.g. Viidik
et al., 1965 and Viidik, 1969). Another frequently used method
for clamping bones is to drill pins through the bones; this is
safe for specimens from larger animals, while small specimens
(as from rats and rabbits) tend to fracture through the drill
hole, when the force is applied to the pins only. This problem
is overcome by combining pins with some other mechanical'devi-
ce (Woo et al., 1986b). These composite specimens yield infor-
mation only to the point, where the weakest component (usually
the bone of the insertion zone of tendon into bone) fails.
Complete stress-strain curves (including data on ultimate
tensile strength) for tendons and ligaments can, therefore, be
derived only from experiments on these structures clamped iso-
latedly into a materials testing machine. Numerous methods for
clamping have been described, from serrated jaws (that produce
breaks between the sharp edges) to fixed or mobile wedge sys-
tems (that crush or squeaze the ends of the specimen) and cast-
ing the ends of the specimen into a resin (that creates a zone
with a transition from dry to fully hydrated tissue). The cri-
terion for satisfactory clamping is random scatter of jaw and
free space breaks with no gliding of specimen and no differen-
ce in ultimate tensile strength between these two break types
(Viidik, 1967, 1979a). There seems to be no universal clamp
design; an appropriate device must be designed for each type
and size of tissue.
2.1.3. Transformation of load/deformation to stress/strain
Modern materlals testlng machlnes record load and deforma-
tion values continuously with accuracy through a testing cyc-
le. The difficulties and sources of error in this transforma-
tion in performed, reside in the methods avaliable for measu-
ring cross-sectional area and original length of the specimen.
The cross-sectional area can be measured either directly as
a physical area or calculated indirectly from determination of
wet or dry weight or collagen content, expressed per unit spe-
cimen original length.
Most methods for direct measurement have limitations and/or
inherent sources of error: (i) The method to measure the areas
of thin sections cut from the specimen is precise. The varia-
tion of the cross-sectional area of the specimen along its ax-
is would, however, require multiple measurements. This preclu-
des the use of destructive methods. (ii) Most non-destructive
methods rely on instruments with two calipers mounted at right
angle to each other. An error is introduced by the measurement
of a rectangular cross-section, while that of a tendon is ir-
regularly ellipsoid. The slight compressibility of a tendon or
ligament introduces another error. The precision of these me-
thods further decreases with decreasing specimen size. For re-
view of these methods, see Ellis (1969). The development of
new methods, based on laser technology, however, is promising.
The indirect methods are based on the assumption that wet
64

or dry weight or collagen content per unit specimen length re-

present its "functional cross-sectional area" (Viidik, 1973,
1979a, 1980a). These methods have the advantage that they give
an average "area" for the whole length of the specimen. Measu-
rement of wet weight is sensitive to changes in the degree of
hydration (those occurring in vivo, e.g. ageing and change of
hormonal balance, as well as those inadvertedly induced by la-
boratory procedures). Dry weight and collagen content measure-
ments differ from each other only for tissues containing sig-
nificant amounts of non-collagenous structural elements (e.g.
skin and arterial tissue). Dry weight should be used, if it
can be asssumed that all fractions contribute equally to the
strength of the tissue or when calculating true cross-sectio-
nal area indirectly (the density of the tissue must then be
known). Collagen content should be used, when collagen is the
only mechanically important component.
The main difficulty with defining the original length of a
specimen is due to the very insiduous start of the load-defor-
mation curve. This is true also for the stiffest soft connec-
tive tissue structures, tendons and ligaments: Some of the wa-
viness seen by incidental light microscopy on the surface of a
relaxed tendon is straightened out before the recording equip-
ment can measure tension in the specimen (Viidik 1972, 1980b).
Several methods have been used to define this length, calcula-
ted as inter-clamp length of the completely relaxed specimen
and the "deformation" necessary to reach the starting point of
the curve. This point can be defined as that:
(i) when the minimum recordable load is reached (must be in-
dependent of variations in amplification levels within a se-
ries of tests),
(ii) of the intercept between the tangent to the beginning
of the "toe" part and the deformation axis on the raw recor-
der curve,
(iii) when the tangent modulus (load increment per unit de-
formation increment) is significantly different from zero,
(iv) when the first morphological changes can be seen (e.g.
by incidental light microscopy, difficult to determine in a
reproducible way), or,
(v) of the in vivo resting state stress level (theoretically
ideal for stud~es of physiological function ranges but dif-
ficult to determine unambiguously).
2.1.4. Parameters calculated from the stress-strain curve
The whole stress-strain curve ~s needed for the characteri-
zation of the tensile properties of a tissue. It begins with a
"toe" part, convex towards the strain axis, continues with a
fairly linear region and bends off somewhat towards the strain
axis before the point of ultimate stress is reached (Cf. Fig.
2). The stress then declines most often somewhat further, be-
fore complete failure has occurred. The mean curve for a group
of specimens is plotted from a set containing about 20 stress-
strain data points, where mean stress is calculated for each
strain increment of e.g. 0.01 (the size of increment depends
on the extensibility of the tissue). This curve never reaches
the point of mean maximum stress, since strain increment is
its independent variable and the coordinates for the point of
maximum stress are calculated as two independent variables.
 65

The part of the curve above the end of the linear region, must
thus be extrapolated (Cf. the dashed lines in Fig. 2).
The "toe" part can be analysed by plotting Fung's tangent
modulus (stress increment per unit strain increment, in this
case 0.01) versus either stress (Fung, 1967, 1981) or strain
(Viidik et ala 1982). These two plots result in straight lines
bending off towards the stress and strain axis respectively,
when the linear region of the stress-strain curve is reached.
In some instances two linear equations must, however, be used
to describe this exponential stress response (Fung, 1972).
The important parameter describing the linear region of the
stress-strain curve is the "elastic stiffness", i.e. the tan-
gent between the curve and the strain axis. The term modulus
of elasticity or Young's modulus should be avoided, since it
implies that the tissue is elastic, while all soft connective
tissues have viscous components in their mechanical behaviour
(see 2.3). The sets of stress-strain data points for the be-
ginning and end of this region are also calculated at times.
The failure energy or the capacity of the tissue to absorb
energy is calculated from the stress-strain curve as the area
between the curve and the strain axis. This area can be measu-
red from the origin to the point of maximum stress or to that
of complete separation of tissue at th~ site of failure. The
first-mentioned point can easily be determined with accuracy.
The other one is more difficult to define, since the failure
occurs gradually with some fiber strands failing before others
and a few seem at times still to bridge the gap, when stress
has declined to zero. The point of failure can operationally
be defined as the point, where the stress has dropped to e.g.
two thirds of the maximum value.

2.2. In vivo changes in the stress-strain characteristics

The stress-strain curve, ltS shape and the pertlnent para-
meters derived from it have been described in the previous pa-
ragraph (2.1.4). This section will briefly discuss some physi-
ological and pathophysiological conditions that influence the
mechanical properties of soft connective tissues.
Profound changes occur in soft connective tissues during
the maturation and ageing of the organism. There are signifi-
cant species differences, which are important for the evalua-
tion of the literature. While some species, e.g. man, cow and
rabbit, show a rather distinct cessation of maturation at pu-
berty, ot~ers, most notably rodents like rats and mice, have a
prolongec( maturation process, which might continue during the
ageing process of the organism (for review see Viidik 1979b,
1982).
During the maturation of the organism process the collagen
fibers are gradually stabilized, initially by increased cross-
linking and later by changes in cross-linking pattern (on mic-
rofibrillar level). Tensile strength and failure energy incre-
ase rapidly, for the rat up to an age of 4 months, whereafter
levelling off occurs; a slight decrease is seen after the age
of 12 to 24 months (Viidik, 1982; Vogel, 1983). The viscoelas-
tic properties, e.g. strain rate sensitivity (Haut, 1983) and
hystere~is characteristics (Vogel, 1983) change during early
maturation but do not exhibit any decline later. These changes
66

are from a functional point of view not detrimental but render

the performance of tendons and ligaments more precise.
Physical training improves the performance of tendons and
ligaments (e.g. Viidik, 1968a; Woo et al., 1981; for review
see Viidik, 1986b). Immobilization, on the other hand, causes
a rapid deterioration of these connective tissues (Akeson et
al., 1980; Woo et al., 1982).

2.3. Viscoelasticity
The vlscoelastlc properties of a tendon (or a ligament) are
not obvious during a single tensile loading until failure. It
can, however, be seen from experiments performed at different
strain rates, that "elastic stiffness" (Fris~n et al., 1968b)
and ultimate tensile strength (Haut, 1984) become higher with
increasing strain rate. When the loading cycle is halted with-
in the linear part of the stress-strain curve that these tis-
sues exhibit stress relaxation and creep phenomena, when the
strain and stress respectively is kept constant (e.g. Fris~n
et al., 1968a).
Experiments with repeated loading-unloading cycles on the
same specimen show that the hysteresis becomes smaller and the
"elastic stiffness" of the loading part of the stress-strain
curve increases with cycle number, provided the stress level
is low enough not to induce structural damage (Viidik, 1968a;
Fris~n et al., 1968a). With repeated cycles of stress relaxa-
tion or creep at the same initial stress and constant stress
level respectively relaxation and creep gradually decreases.
After six to ten loading-unloading cycles no further changes
can be seen in these parameters; the rheological properties of
the tissue have thus reached their stationary phase (i.e. the
initial phase phenomena have been exhausted by precondition-
ing) (for discussion see Fris~n et al., 1968a; Viidik, 1973,
1979a, 1980a).
There is, however, in this stationary phase an element of
delayed viscoelastic recovery: The hysteresis and the stress
relaxation and creep phenomena are larger, when the cycle is
repeated after a resting period of 5-10 minutes compared to an
immediately repeated cycle.
It can be concluded from these observations (together with
those discussed above in 2.1.4) that parallel-fibred collage-
nous tissue shows:
(i) certain initial phase phenomena, which include a compo-
nent of plasticity and strain-hardening and one of, at least
in vitro, irreverible viscoelasticity (see Fris~n et al.,
1968a,b); these must be eliminated by a preconditioning pro-
cedure before e.g. strain rate dependence is studied in the
stationary phase,
(ii) in its stationary phase non-linear elasticity, i.e. the
shape of the non-linear "toe" part of the ~ress:!train cur-
ve have_lthe same shape at very low (510 sec ) and high
(7.2 sec ) strain rates (Fris~n et al., 1968b; Haut, 1983b,
Viidik and Melvin, unpublished data), and,
(iii) in its stationary phase non-linear viscoelasticity
(Viidik et al., 1978, Viidik, unpublished data), i.e. the
"shape parameters" for stress relaxation and creep respecti-
vely are different when calculated for short (10 min) and
 67

long (1000 min) term experiments; this seems to imply that

there is a rapid viscous component acting in the beginning
and a slow one later on (whether this is influenced by mor-
phological rearrangement of fibrils and ground substance is
not clear).
These observations have led to the formulation of Fung's
law, a set of equations for the "quasi-linear" visco-elastic
behaviour of soft biological tissues, integrating the elastic
and history-dependent parts of the stress-strain relationship
into this law (Fung, 1967, 1968, 1972). While Fung's work fo-
cussed on other tissues than tendons and ligaments, Haut and
Little (1969) showed that this law is valid for ligaments af-
ter preconditioning and extrapolating infinitely low strain
rates.
The other approach has been to describe this set of elas-
tic, viscous and plastic components in the rheological beha-
viour of these tissues by composing models of discrete ele-
ments and constitutive equations based on these models. This
method of analyzing the rheological behavior of macromolecules
was suggested by Alfrey and Gurnee (1957) and implemented for
parallel-fibred connective tissues by Viidik and Magi (1967)
and Viidik (1968b); it was subsequently analysed in detail by
Frisen et al. (1968a,b). (For review see also viidik, 1987).

3. INTERDEPENDENCE BETWEEN STRUCTURE AND FUNCTION

Collagen is the strongest macromolecule in the mammalian
body. Its strength is derived from the triple-helical molecu-
le, each helix of which is a polypeptide chain, about 1,000
amino acid residues long. These alpha chains are left-handed
helices, which are wound together into a right-handed superhe-
lix (length 300 nm, diameter 1.5 nm).
The molecules are assembled into microfibrils, consisting
of five rows of molecules. These rows are wound together into
a left-handed "super-superhelix" (diameter 4 nm). The molecu-
les are aggregated, "quarter-staggered" in adjacent rows, into
microfibrils guided by hydrophobic interactions and, having in
a row no end-to-end couplings but gaps (length 0.6 period) in
between, stabilized by lateral cross-links bridging adjacent
molecules in the same or neighbouring microfibrils. The cross-
links change with maturation and ageing; while two molecules
are linked together in the immature tissue, later on three or
more molecules are probably joined by the same cross-link.
A large number of microfibrils are aggregated into fibrils
forming right-handed "super-super-superhelices" (diameter 30-
150 nm). Whether elements other than cross-links bridging ad-
jacent microfibrils contribute to mechanical stability on this
level of structural hierarchy is not clear (the proteoglycans
presnet might). The fibrils are seen in the electron microsco-
pe with their characteristic striation with a 68 nm periodici-
ty (that of the "quarterstagger"), while that of the molecule
thus is 4.4 periods.
The alternating handedness at the consecutive levels of or-
ganization (from polypeptide chains to fibrils) stabilizes the
collagen fibril mechanically by converting axial stress to la-
teral compression (Piez and Trus, 1977). Man has used the same
principle for rope-making.
68

The fibrils are packed into fibers, the smallest unit seen
in the light microscope (diameter 1-10 um), and assembled into
primary and secondary fiber bundles. On these levels of struc-
tural hierarchy a planar, almost sinusoidal waviness is seen
but no helicality, except that of fiber bundles in certain li-
gaments, is found. The fiber does not attain mechanical stabi-
lity from cross-links (the distance-Detween fibrils being to
long) but probably from interactions between the type I colla-
gen constituting the fibrils and proteoclycans as well as type
IV and V collagens. (For review seeViidik, 1980b).
The waviness seen in the light microscope is straightened
out during the "toe" part of the stress-strain curve (Rigby et
ale 1959; Viidik and Ekholm (1968); Viidik, 1972). No further
changes are seen during the linear region. An irregularwavi-
ness appears in some fiber bundles in the failure region.
The structure on the molecular and fibrillar levels is, on
the other hand, affected by the tensile forces of the linear
region of the stress-strain curve. The pitch of the triple-he-
lix is extended; the period length of 68 nm increases to over
72 nm as measured on electron micrographs (Viidik and Ekholm,
1968) or from low angle X-ray diffraction patterns (from 67 to
70 nm) (Nemetschek et al., 1978). The pitch of the alpha chain
helices (calculated from wide angle X-ray diffraction reflexi-
ons) is also lengthened (Cowan et al., 1955).
These observations, together with those on the initial pha-
se phenomena of plasticity and irreversible viscoelasticity as
well as on the delayed viscoelastic recovery in the stationary
phase (see 2.3 above), suggest that (i) by preconditioning the
paralleling of the fibril and fiber structures is improved and
some ground substance is dislodged, (ii) the "toe" part of the
curve is derived from stretching out the waviness of the fiber
bundles, thereby redistributing some of the fluid and ground
substance, and (iii) the stress-strain behaviour and the vis-
coelasticity of the linear part of the curve are derived from
the fibrillar structure. This hypothesis is supported by data
from measurements of the temperature dependence of activation
energy for viscoelastic deformation (small abrupt temperature
changes during creep): The activation energy for creep during
the "toe" part is of the same magnitude as that for hyaluronic
acid (a major ground substance component) and that during the
linear part of the stress-strain curve is comparable to that
for tape of purified collagen (Hooley and Cohen, 1979; see al-
so Viidik, 1980, 1987 for comparative discussion).

4. BIOMECHANICS OF MESHWORKS AND MIXED TISSUES

Tissue geometry is more complex for e.g. skin (mainly fiber
arrangement and interaction with a larger amount of ground sub-
stance) and arteries (both arrangement and interaction between
collagen and elastin). This is seen in the mechanical proper-
ties (in tensile strength testing) as higher extensibility and
as lower "elastic stiffness" and tensile strength compared to
tissues, which consist mainly of parallel-fibred collagen (see
Viidik 1979a, 1980b).
The influence of tissue geometry and especially composition
is even more pronounced in rheological analysis, i.e. when at-
tempting preconditioning and performing stress relaxation and
 69

creep experiments. It is only possible to reach the stationary

phase at considerably lower stress levels (relative to ultima-
te tensile stress) than when testing tendons or ligaments. The
stress relaxation and creep phenomena are considerably more
pronounced and probably also influenced by geometrical changes
during creep at constant stress. These complications necessi-
tate modifcation of the method to calculate the shape parame-
ter and asymptote according to the constitutive equation for
parallel-fibred collagenous tissue of Viidik (e.g. 1980a; see
also Frisen et al., 1969b). These difficulties are not created
by in vitro testing of strip-shaped specimens. The cutting of
obllquely running fibers (see 2.1.2) and the absence of blood
circulation (for restoring fluid balance in the larger amount
of ground substance) can be ruled out as causes: The same rhe-
ological behaviour is seen in in vivo testing (Larsen and Vii-
dik, to be published).
The influence of the various tissue components can be ana-
lysed by testing tissues, from which one element has been re-
moved at a time by "biochemical dissection", using specific en-
zymes (Hoffman and Daily, 1980; Viidik et al., 1982).
Elastin is generally considered to be more extensible and
have less mechanical strength. Its main functions are to store
energy and restore tissue geometry after removal of mechanical
stress (Serafini-Fracassini et al., 1977). Removal of collagen
by collagenase does not increase the extensibility of liga-
mentum nuchae and aorta (Hoffman and Daily, 1980; Viidik et
al., 1982), although the tensile strength of both tissues de-
creases pronouncedly. An interaction between elastin and col-
lagen is important for these mechanical properties: Removal of
the elastin from aorta (by elastase, with a protease inhibitor
added to avoid damage to collagen) not only decreases the ten-
sile strength but also increases the maximum strain (Viidik et
ale 1982).
Proteoglycans are considered to "glue" the collagen fibrils
together. No changes in mechanical properties can, howver, be
seen, when specific enzymes (hyaluronidase and chondroitinase
ABC) are used. Since the removal of proteoglycans is not comp-
lete after such enzymatic treatment, it is not possible to ex-
clude that the ground substance may have an influence on the
mechanical properties of collagen fibers. The role of the pro-
tenglycans for the viscoelastic properties of skin remains to
be elucidated.
How other collagens (especially types IV and V) and fibro-
nectin contribute to stability of various soft connective tis-
sues is not known; that they should have a role is suggested
by immunohistochemical studies.

REFERENCES

Akeson WH, Amiel D & Woo SL-Y (1980) Immobility effects on

synovial joints- The pathomechanism of joint contracture. Bio-
rheology 17: 95-110. ----
Alfrey T & Gurnee EF (1957) Molecular structure and mecha-
nical behaviour of macromolecules. Pp. 12-32 in: Tissue elas-
ticity (Remington JW ed), Americal PhysiologICal Society, Wa-
70

shington.
Bailey, AJ & Etherington, DJ (1980) Metabolism of collagen
and elastin. Pp. 299-460 in: Comprehensive biochemistry (Flor-
kin M, Neuberger, A & van Deenen LLM eds), Vol 19B: Part I,
Elsevier Scientific Publ, Amsterdam.
Butler DL, Noyes FR & Grood ES (1978) Measurement of the
mechanical properties of ligaments. Pp. 279-314 in: Handbook
of engineering in medicine and biology (Feinberg-SN & Fleming
DG eds), Section B, Vol I, CRC Press, west Palm Beach.
Cowan PM, North ACT & Randall JT (1955) X-ray diffraction
studies of collagen fibres. Symp Soc Exp BioI 9: 115-126.
Hooley, CJ & Cohen, RE (1979) A model for creep behaviour
of tendon. Int J BioI Macromol 1:123-132.
Elliott DH (1965) Structure and function of mammalian ten-
don. BioI Rev 40: 392-421
ElllS DG (1969) Cross-sectional area measurements for ten-
don specimens: A comparison of several methods. J Biomech 2:
175-186.
Frisen M, Magi M, Sonnerup L & Viidik A (1968a) Rheological
analysis of soft collagenous tissues. Part I: Theoretical con-
siderations. J Biomech 2: 13-20.
Frisen M, Magl M, Sonnerup L & Viidik A (1968b) Rheological
analysis of soft collagenous tissues. Part II: Experimental e-
valuations and verifications. J Biomech 2: 21-28.
Fung YCB (1967) Elasticity of soft tissues in simple elong-
ation. Amer J Physiol 213: 1532-1544.
Fung YCB (1968) Bl0mechanics, its scope, history and some
problems of continuum mechanics in physiology. Appl Mech Rev
21: 175-186.
Fung, YCB (1972) Stress-strain history relations of soft
tissues in simple elongation. Pp. 181-208 in: Biomechanics:
Its foundation and objectives (Fung YCB, Perrone N & Anliker M
eds), Prentice-Hall, Englewood Cliffs.
Fung YCB (1981) Biomechanics. Mechanical properties of li-
ving tissues, Springer-Verlag, New York.
Galante JO (1967) Tensile properties of the human lumbar
intervertebral disk. Acta Orthop Scand, Suppl 100.
Harkness RD (1961) Biological functions of collagen. BioI
Rev 36: 399-563. ----
Harkness RD (1968) Mechanical properties of collagenous
tissue. Pp. 247-310 in: Treatise on collagen, Vol 2: Biology
of collagen (Gould BS-ed) , Academic Press, New York.
Haut RC (1983a) Correlation between strain-rate-sensitivity
in rat-tail tendon and tissue glycosaminoglycans. Res Publ
GMR-4243, General Motor Research Laboratories, Warren, MI.
Haut RC (1983b) Age dependent influence of strain rate on
the tensile failure of rat-tail tendon. J biomech Engng 105:
296-299.
Haut RC (1984) Sensitivity of lathyritic tendon to strain.
Res Publ GRM-4662, General Motors Research Laboratories, War-
ren, MI.
Haut RC (1986) The influence of specimen length on the ten-
sile failure properties of tendon collagen. J Biomech 19 951-
955.
Haut RC & Little RW (1969) Rheological properties of canine
anterior cruciate ligaments. J Biomech 2: 289-298.
 71

Hoffman AS & Daily CH (1980) A commentary on the use of en-

zyme probes to elucidate the contributions of individaul com-
ponents to soft-tissue biomechanics in vitro. Pp. 297-312 in:
Biology of collagen (Viidik A & Vuust J eds), Academic Press,
London.
Miller EJ (1985) The structure of firbil-forming collagens.
Pp. 1-13 in: Biology, chemistrx, and pathology of collagen
(Fleischmajer R, Olsen BR & Kuhn K eds), Annals New York Aca-
de my of Sciences, Vol 460.
Nemetschek T, Jonak R, Nemetschek-Gansler H & Riedl H (19-
78) tiber die Bestimmung von Langperioden-Anderungen am Kolla-
gen Z Naturforsch 33C: 928-936.
Nilsson T (1982) Biomechanical studies of rabbit abdominal
wall. Part II. The mechanical properties of specimens in rela-
tion to length, width and fibre orientation. J Biomech 15:131-
135.
Piez KA & Trus BL (1977) Microfibrillar structure and pack-
ing of collagen: Hydrophobic interactions. J mol Biol 110:701-
704.
Rigby BJ, Hirai N, Spikes JD & Eyring H (1959) The mechani-
cal properties of rat tail tendon. J gen Physiol 43: 265-283.
Serafini-Fracassini A, Field JM, Smlth JW & Stephens WGS
(1977) The ultrastructure and mechanics of elastic ligaments.
J Ultrastr Res 58: 244-252.
SanJeevl R, Somanathan N & Ramaswamy D (1982) A viscoelas-
tic model for collagen fibres. J Biomech 15: 181-183.
Viidik A (1967) Experimental evaluation of the tensile
strength of isolated rabbit tendons Biomed Engng. 2: 64-67
Viidik A (1968a) The elasticity and tenslle strength of the
anterior cruciate ligament in rabbits as influenced by train-
ing. Acta Physiol Scand 74: 372-380.
Vildlk A (1968b) A rheological model for uncalcified paral-
lel-fibred collagenous tissue. J Biomech 1: 3-11.
Viidik A (1969) Tensile strength properties of achilles
tendon systems in trained and untrained rabbits. Acta Orthop
Scand 40: 261-272.
---viidik A (1972) Simultaneous mechanical and light microsco-
pic studies of collagen fibers. Zschr Anat Etw-gesch 136: 204-
212.
Viidik A (1973) Functional properties of collagenous tissu-
es. Pp. 127-215 in: "International review of connective tissue
research" (Hall--DA & Jackson DS eds), Vol 6, Academic Press,
London.
Viidik A (1979a) Biomechanical behavior of soft connective
tissues. Pp. 75-113 in: Progress in biomechanics (Akkas Ned),
Sijthoff & Nordhoff,-Xlpen aan den Rijn.
Viidik A (1979b) Connective tissues - possible implications
of the temporal changes for the aging process. Mech Age Dev 9:
267-285.
Viidik A (1980a) Mechanical properties of parallel-fibred
collagenous tissues. Pp. 237-255 in: Biology of collagen (Vii-
dik A & Vuust J eds), Academic Press, London.
Viidik A (1980b) Interdependence between structure and fun-
ction in collagenous tissues. Pp. 257-280 in: Biology of col-
lagen (Viidik A & Vuust J eds), Academic Press, London.
Viidik A (1982) Age-related changes in connective tissues.
72

Pp. 173-211 in: Lectures on Gerontology (Viidik A ed), Vol 1,

Part A, Academic Press, London.
Viidik A (1986a) Soft connective tissue biomechanics. Pp.
276-279 in: Transactions of the 5th meeting of the European
Society for Biomechanics, Freie Universitat, Berlin.
Viidik A (1986b) Adaptability of connective tissues. Pp.
545-562 in: Biochemistry of exercise: Metabolic regulation and
its significance (Saltin Bed), Human Kinetics Publishers,
Champagne, IL.
Viidik A (1987) Properties of tendons and ligaments. in:
Handbook of bioengineering (Skalak R & Chien S eds), McGraw-
Hill, New York (in press).
Viidik A & Ekholm R (1969) Light and electron microscopic
studies of collagen fibers under strain. Zschr Anat Entwgesch
127: 154-164.
Viidik A & Gottrup F (1986) Mechanics of healing soft tis-
sue wounds. Pp. 263-279 in: Frontiers in biomechanics (Schmid-
Schonbein GW, Woo SL-Y-& Zweifach BW eds), Springer-Verlag,
New York.
Viidik A & Lewin T (1966) Changes in tensile strength cha-
racteristics and histology induced by different modes of post-
mortal storage. Acta Orthop Scand 37: 141-155.
Viidik A & Mag~ M (1967) Visco-elastic properties of liga-
ments. P. 507 in: Digest 7th Int Conf Medical and Biological
Engineering (JacObsen Bed), Almqvist & Wiksell, Stockholm.
Viidik, A, Oxlund, H & Andreassen T (1978) Rheological ana-
lysis of collagen, with reference to morphology and experimen-
tal techniques. Pp. 187-188 in: Proc. 3rd Int Congr Biorheolo-
gy, La Jolla. --
Viidik A, Danielsen CC & Oxlund H (1982) On fundamental and
phenomenological models, structure and mechanical properties
of collagen, elastin and glycosaminoglycan complexes. Biorheol
19: 437-451.
Viidik A, Sandqvist L & Magi M (1965) Influence of postmor-
tal storage on tensile strength characteristics and histology
induced by different modes of postmortal storage. Acta Orthop
Scand, Suppl 79.
---VOgel HG (1983) Age dependence of mechanical properties of
rat tail tendon (hysteresis experiments). Akt Geron 13: 22-27.
Woo, SL-Y (1982) Mechanical properties of tendons and liga-
ments. I; Quasi-static and nonlinear viscoelastic properties.
Biorheol 19: 385-396.
Woo SL-Y, Gomez, MA, Woo, Y-K & Akeson WH (1982) Mechanical
properties of ligaments and tendons. II. The relationship bet-
ween immobilization and exercise on tissue remodeling. Biorhe-
01 19: 397-408.
-- Woo SL-Y, Gomez MA, Arniel D, Ritter MA, Gelberman, RH & Ak-
eson WH (1981) The effect of exercise on the biomechanical and
biochemical properties of swine digital flexor tendons. J Bio-
mech Eng 103: 51-55.
Woo SL-Y, Orlando CA, Camp JF & Akeson WH (1986a) Effects
of postmortem storage by freezing on ligament tensile behavior
J Biomech 19: 399-404.
Woo SL-Y, Orlando CA, Gomez MA Frank CB & Akeson WH (1986b)
Tensile properties of the medial collateral ligament as a fun-
ction of age. J Orthop Res 4: 133-141
 73

CARDIOVASCULAR BIOMECHANICS
T. Arts* and R.S. Reneman**
Department of Biophysics* and Physiology**. University of
Limburg. Maastricht. The Netherlands

INTRODUCTION
The circulatory system is a closed loop passing the heart
twice. The blood, returned from the body, enters the right
ventricle via the right atrium. At a relatively low pressure
(approx. 30 mmHg) blood is pumped into the lungs at a rate in
the order of 5 liters per minute. The blood returning from the
lungs is collected in the left atrium, from where it flows into
the left ventricle during each diastolic phase. The left ven-
tricle is a powerful pump having a wall significantly thicker
than the right ventricle. The latter property enables the left
ventricle to generate a relatively high pressure (approx. 130
mmHg) at the same flow rate as the right ventricle. After pas-
sing the aortic valve, the blood enters the arterial system.
The compliance of this system enables uptake of the volume,
ejected by the left ventricle during the relatively short pe-
riod of systole. Because in arteries inertial forces overrule
viscous forces, after each heartbeat a pressure-flow wave is
guided down to the smallest arteries. At the level of arteri-
oles (diameters 15-100 ~m) viscous forces exceed inertial for-
ces, causing attenuation of the pressure wave. At this level
flow regulation takes pace by controlling vessel diameter. In
the narrowest bloodvessels, the capillaries (diameter 4-8 ~m),
for most substrates dimensional scaling factors are favorable
for considerable diffusional exchange between tissue and blood.
Distal to the capillaries, small vessels join to veins. The
veins contain approximately 70% of total blood volume. Because
their volume is large, they have an important role in the regu-
lation of circulating blood volume.
The muscular tissue in the wall of the heart, the myocardium,
delivers the pump energy. Knowledge of the stress strain rela-
tion of myocardial tissue is of great importance, under passive
as well as under activated conditions. Stress and strain oc-
curring in the myocardial fibers are associated with pressure
and volume changes in the cavity. In this relation geometry of
wall and cavity is a major determinant. The left ventricle is
loaded with the aortic input impedance, which is closely asso-
ciated with pressure-flow wave guiding in the arteries.
Within the heart, large arteries, and veins the Reynolds num-
74

ber is high, implying a complicated flow behavior within rela-

tively large spaces. Velocity profiles are important to know in
relation to implant ion of artificial valves in the heart as
well as atherosclerotic plaque formation in large arteries.
This article will not deal with the flow profile aspect of
cardiovascular biomechanics.
Mechanically, the coronary system is a circulatory subsystem
of special interest, because perfusion pressure as well as
pressure in the tissue surrounding the coronary vessels are
generated by the same source, being contraction of the heart.
On the one hand, appropriate coronary circulation needs a suf-
ficiently high perfusion pressure. On the other hand, a high
pressure in the tissue of the wall counteracts perfusion be-
cause of external compression of the coronary vessels.

MECHANICAL CHARACTERISTICS OF CARDIAC TISSUE

The cardiac tissue consists mainly of muscle cells, 15 \l m
thick and 125 \lm long. The ends of the cells are tightly inter-
connected by intercalated disks, thus forming a strong fiber
structure. ~'i'ithin cells, the contractile proteins actine and
myosine form a well organized filament structure, having a
distict periodic structure. The length of one spatial period is
called sarcomere length and is generally found to be in the
range of 1. 6 \lm to 2.3 \lm depend ing on the state of muscle
strech. Under the microscope, within a sarcomere length across
the width of a cell, two dark cross-striations can be distin-
guished. Besides the contractile apparatus, the myocardium also
contains collagen fibers in complicated arrangements. The role
of these structures is not clear yet, but they are likely to
contribute to the passive stress-strain characteristics of the
tissue.
In physiological experiments on isolated cardiac muscle often
papillary muscle is used. The passive characteristics of such
muscle (Ter Keurs 1963) is comparable to other soft connective
tissues (fig. 1). At low values for sarcomere length 1.9 \lm)
the tissue is very compl iant. At large sarcomere length, the
tissue stiffens exponentially (Fung 1970) reaching high, non-
physiological values of stress at a sarcomere length of 2.4 \lm.
In this respect the behavior of cardiac muscle is somewhat
different from skeletal muscle, which is much more compliant at
the latter values for sarcomere length. Hhen the muscle is
activated, an active stress component is added. Keeping muscle
length constant during contraction, this component increases
from zero at a sarcomere length of 1.5 \lm to a value in the
order of 100 kPa at a sarcomere length of 2.1 \l m (Pollack
1976). The shape of the stress sarcomere length relation
depends on the amount of free calcium in the muscular cell. At
higher calcium concentrations, the slope of the curve is steep-
 75

er, and reaches the same value of stress at lower values of

sarcomere length (Ter Keurs 1983).

stress velocity
of
kPa shortening
100

50

o 1.5 IJm 2.5

sarcomere length

Fig. 1: Myocardial stress versus Fig. 2: Velocity of sarcomere

sarcomere length at zero shortening versus
strain rate under pas- stress at different
sive (P) and active con- values of sarcomere
ditions at high (H) and length.
low (L) calcium concen-
tration.

In a dynamically contracting heart, sarcomere length shortens

during muscular activation. Then stress is lower as compared to
isometric contraction. In fig. 2. stress is plotted as a func-
tion of velocity of sarcomere shortening (Edman 1976, Henderson
1973) at different values for sarcomere length. According to
Hill's theory (1939), the curves should be hyperbolic. Skeletal
muscle obeys Hill's equation quite accurately, but at larger
values of stress and sarcomere length the curve may show a
point of inflection.

THE MECHANICS OF THE CARDIAC WALL

The cardiac wall is designed in such a way, that stress in
the myocardial fibers causes a positive pressure in the cavity.
Investigation of cardiac wall mechanics is hampered by the
1 imi ta tions of various techniques to measure local mechanical
parameters. Force transducers attached to the cardiac wall
(Feigl 1967) cause damage of the wall at the site of measure-
ment. Moreover, in determining muscle stress, it is not easy to
associate an accurate value of cross-sectional area with the
76

measured force. As a consequence of this difficulty, investiga-

tors are forced to develop mathematical models of the mechanics
of the cardiac wall (Yin 1981). Like the majority of these
models, we limit ourselves to the left ventricle.
Originally models of left ventricular mechanics were simple.
Woods (1892) introduced Laplace's law to calculate wall tension
from left ventricular pressure and the principal radii of cur-
vature, assuming the wall to be thin. In the model, introduced
by Sandler (1963) and improved by Fasetti (1970), the heart is
considered to be thick walled and ellipsoidal with a ratio of
major to minor axis close to 2. The wall material is taken to
be isotropic. From their calculations at the equator, stress in
the inner layers appeared to be approximately twice as high as
in the outer layers. Calculation of stress at all sites of the
ellipsoid required more complicated modelling (Kim 1985) such
as application of the finite element calculation technique
(Janz 1972). All models based on isotropic myocardium result on
simular transmural stress distributions, quite independent of
geometry (fig. 3; Huismans, 1980).
Introduction of the finite element calculation technique
enabled introduction of the realistic geometry of the heart
(Pao 1974). Moreover, at different sites in the wall, mechani-

normalized
stress

% Chadwick pressure __
- ....r-- 30 IS / Fr.nk -Sl;;rting-...
kPa I
I
20

.... - I
10 I normal t
E

I cycle
J I
oO!:--~-""""=-v-;ol'-u-m-e--
o ~ inner WALL outer~

Fig. 3: Stress normalized to Fig. 4: Left ventricular pres-

the average value ver- sure as a function of
sus depth in the myo- cavity volume during
cardial wall for vari- diastole (D) isovolum-
ous models based on ic systole (IS) and
isotropy (-) as well ejection (E).
as anisotropy (---).
 77

cal properties of the material could be chosen to be different

in order to simulate an ischemic area (Bogen 1980). Neverthe-
less, in the heal thy wall, stress as well as strain in the
inner layers were still twice the value as calculated for the
outer layers, implying an endo- to epicardial difference in
power generation by a factor of four. This finding is difficult
to reconcile with the finding that coronary bloodflow in the
inner layers is only between 5 and 20% higher than in the outer
layers (Bache 1983, Prinzen 1986).
In a next step of model development, deformation of the heart
during the ejection phase is taken to be large. After all, an
ejection fraction of 60% is normal. Models based on isotropy of
the cardiac wall were not able to reach such a high ejection
fraction without introducing unphysiolog ical values of muscle
material properties. A major step forward to a more physiolo-
gical description of cardiac wall mechanics was made by intro-
ducing anisotropic properties of the myocardial tissue (Arts
1979, Chadwick 1982, Beyar 1985). From anatomical studies (Hort
1960, Streeter 1969) it is well known that myocardial tissue is
strongly anisotropic, but introduction of this property in left
ventricular modelling is relatively new. A further essential
improvement is consideration of torsion around the base to apex
axis of the left ventricle, being a result of equilibrium of
shear stresses within the wall (Arts 1979, Chadwick 1982). The
models based on anisotropy are able to simulate a regular left
ventricular ejection phase with normal values for ejection
fraction (50%-70%) and physiological stress-strain characteris-
tics of the myocardial tissue. Transmural differences in muscle
fiber stress and fiber shortening appeared to be significantly
less than when considering the tissue being isotropic. Arts et
al. (1982) calculated that by proper design of the distribution
of muscle fiber orientation across the wall these differences
can be made less than + 5%. Low transmural gradients in mecha-
nical loading are in good agreement with the finding of rela-
tively low transmural differences in coronary perfusion. In
figure 3 the transmural course of stress is indicated for va-
rious models of cardiac wall mechanics.

MODEL CALCULATIONS OF \~ALL DEFORMATION IN RELATION TO EXPERI-

MENTAL RESULTS
In developing a model of a physiological system, some kind of
validation experiments are necessary. Generally, a model de-
scription is based on physiological findings combined with a
hypothesis, expressed in the mathematical structure of the
model. Evaluation of such a model takes place by performing an
experiment on the physiological system, and comparing the re-
sul ts with a pred iction of the model. Most models of card iac
wall mechanics are designed to calculate stress in the wall. As
78

mentioned above, stress cannot be measured easily. So, in eva-

luating the models, other parameters have to be assessed. The
simpler models of cardiac mechanics may be evaluated by mea-
surement of pump function under various loading conditions, but
in this respect inaccuracy of the description of the mechanical
properties of the myocardial tissue greatly affects the related
parameters. Unfortunately, the mechanical properties of myocar-
dial tissue in the intact heart are not accurately known. So,
many simple models of cardiac wall function lack proper assess-
ment. The more complicated models consider a larger number of
measurable entities. For the model, developed in our laboratory
(Arts 1979, 1982), deformation is an important entity of this
kind.
The model of left ventricular mechanics we developed de-
scribes the left ventricle as a section of a cylinder, because
in reality most myocardial mass is concentrated in part of the
left ventricle which resembles a cylinder closely. Fiber orien-
tation changes across the wall. Deformation of the wall is
determined by three parameters, being length, diameter and
torsion angle of the cylinder. In the model, the ratio of cy-
linder length to cylinder radius is determined by an equili-
brium of axial and circumferential stress components in the
wall. The amount of torsional deformation of the cylinder
around the axis is determined by an equilibrium of shear stress
components in the wall. So, the relation between the three
components of deformation of the wall closely reflects the
result of equilibria of stresses in the wall.
Left ventricular deformation can be measured with various
techniques, such as strain gauges (Lab 1978, Feigl 1964), ul-
trasonic dimension gauges (Gallagher 1982), inductive dimension
gauges (Rushmer 1956) or radio-opaque markers (Fenton 1978,
Garrison 1982). The 3 tensorial components of epicardial sur-
face deformation( rate) can be measured with a triplet of in-
ductive coils attached to the epicardium (Arts 1980, 1982) or a
video imaging technique (Prinzen T. 1986). Because, according
to the concept of the model, deformation reflects an equil i-
brium of stresses in the wall, measurement of deformation of
the left ventricle appeared to be appropriate to assess our
model of cardiac wall mechanics. A dimensionless deformation
parameter was found, which appeared to be practically independ-
ent of the stress-strain relation of the myocardial tissue, and
strongly dependent on the gradient of mechanical loading across
the wall. This parameter (STR) is the ratio of inner circumfe-
rential wall shortening to torsion, being quantified as the
gradient of torsion along the axis of the left ventricle multi-
plied by the outer radius. This parameter appeared also to be
nearly independent of hemodynamic loading of the left ventri-
cle, and was calculated to be in the range of 2.4 to 2.6. In an
 79

experimental set-up in the intact, closed chest dog, inner

circumferential shortening as well as torsion of the left ven-
tricle were determined from two-dimensional echocardiographic
images of the heart (Arts 1984). Under control conditions at
rest, in 9 different animals, the parameter STR appeared to be
2.31 + 0.23 (mean + SD), which is close to the model prediction
with a low standard deviation. So, it is concluded that the
description of the wall of the left ventricle by acyl inder
consisting of anisotropic material, is sufficiently accurate to
describe important mechanical features. A precise description
of cardiac geometry appeared to be far less important than
consideration of anisotropy of the material.

LEFT VENTRICULAR PUMP FUNCTION

In physical terms pump function is characterized by pressure
and flow. Pressure can be measured accurately by water-filled
catheters or catheter-tip micromanometers. Aortic volume flow
can be measured by electromagnetic flow meters or by thermo-
dilution. In the beginning of this century, in the isolated
heart, changes in ventricular volume were determined by posi-
tioning the heart in a container filled with fluid (Henderson
1906). By measuring fluid displacement to and from the con-
tainer during the cardiac cycle changes in volume were record-
ed. In early studies by Starling (Patterson 1914), and partial-
ly based on data obtained by Frank, during diastole left ven-
tricular pressure was found to increase exponentially with left
ventricular volume. During systole with occlusion of the as-
cending aorta, the increase of left ventricular pressure shows
a maximum (fig. 4). The increase of left ventricular pressure
during systole depends on contractility of the myocardium.
In fig. 4 for a normal cardiac cycle left ventricular cavity
pressure is plotted as a function of cavity volume. The curve
appears to be a loop. The area of the loop reflects the amount
of work, generated during a cycle. In this loop, left ventricu-
lar pressure is lower than found in the Frank-Starling curves
due to muscle fiber shortening, which results in a decrease of
muscle stress (fig. 2). Van den Horn et al. (1985) found mean
aortic flow to be a simple function of mean left ventricular
pressure. At zero pressure, flow is high. At zero flow, or
aortic occlusion, pressure is high. Between these extremes the
relation between pressure and flow is nearly linear. So, power
is maximum at approximately half maximum pressure and flow
level.
THE CIRCULATORY SYSTEM
The left ventricle is loaded with the aortic input impedance.
The aorta branches into a large number of arteries, which are
compliant tubes filled with blood having a substantial density.
80

So, in arterial dynamics, wave-guiding properties of arteries

are essent ial. Therefore, the behav ior of the input impedance
of the aorta is expected to be extremely complicated. However,
Fourier analysis of pressure and flow at the entrance of the
aorta revealed a quite simple behavior of the input impedance
(Westerhof 1973).
The aortic input impedance appeared to be equivalent to a
proximal resistance in series with a distal resistance, which
is in parallel with a compliance. The proximal resistance de-
termines the high frequency behavior (>6 Hz), and is close to
the characteristic impedance of the aorta (approx. 10 MPa s
m-3). The high frequency components of pressure and flow are
associated with pressure flow waves in the large arteries (wave
speed 3.5 - 12 m/s), having a relatively short wave length,
high attenuation, and minor reflections. So for high frequen-
cies aortic input impedance is mainly determined by the phy-
sical properties of the proximal arteries. At frequencies below
2-3 Hz the aortic input impedance is close to the peripheral
resistance which depends on the state of vasodilation in the
various organs. During excercise, peripheral resistance may de-
crease from 100 MPa s m-3 down to one quarter of this value.
Arterioles (10-100 11 m), capillaries (4-8 11m), and venules
(15-150 Ilffi) form the microvasculature. In this reg ion,
Reynold I s number is low, implying a dominant role of viscous
effects in relation to pressure gradients in the flow channels.
In larger vessels, compliance is mainly determined by the ves-
sel wall, but in small vessels, this is not clear. Fung (1966)
proposed a model in which a small vessel behaves like a tunnel
in a gel. Pressure measurement in small vessels is difficult,
and requires skills. Therefore, not many data are available
with respect compliance of the microvasculature.
Generally, not much attention is paid to the mechanics of the
veins. However, beside returning the blood to the heart, they
have another important function. Approximately 70% of the blood
volume is located in the veins. Because little energy is needed
to change the diameter of the thin-walled veins, a relatively
little amount of smooth muscular tissue in the walls of the
veins is able to displace a large amount of blood volume. Con-
traction of these smooth muscles elevates venous pressure. As a
result, right atrial filling as well as pulmonary and aortic
flow increase. Thus, by controlling circulating blood volume,
the veins are extremely important in control of blood pressure.
CORONARY CIRCULATION
The coronary circulation differs from regular vasculatures in
that perfusion is partially obstructed by the influence of the
pressure in the tissue surround ing the coronary vasculature.
During a normal cardiac cycle, volume flow in the coronary
 81

arteries changes periodically, being high in diastole, and low

in systole.
The mechanism of these changes in coronary artery flow is
still matter of a controversy (Spaan 1985, Arts 1985). Downey
and Kirk (1975) postulated that during systole, pressure in the
myocardial tissue would exceed intravascular pressure, causing
collapse of the coronary vasculature. As a result, during this
phase, coronary artery flow would stop predominantly in the
inner layers of the cardiac wall where tissue pressure is maxi-
mum. However, by this theory, several findings could not be ex-
plained. For instance, coronary venous flow does not stop dur-
ing systole, on the contrary, it is maximum during this phase.
Furthermore, under extreme conditions of low arterial pressure,
coronary artery flow is negative during systole. Both findings
imply occurrence ,of changes in volume of the coronary vascula-
ture within a cardiac cycle.
The latter finding raised another possibility to explain
changes in coronary flow during the cardiac cycle (Spaan 1981).
During systole, coronary vessels are squeezed, resulting in a
backward arterial flow component and a forward venous compo-
nent. As a result of the regular arterio-venous pressure drop,
a continuous forward flow component is added. Generally, this
continuous forward flow component is sl ightly larger than the
squeezing component, thus causing systolic coronary artery flow
being close to zero. The amount of squeezed volume per cardiac
beat is in the order of 10% or less of total blood volume pres-
ent in the coronary vasculature.
Uptil now, both hypothesis do not fully explain the pressure
flow relation of the coronary vasculature. However, it is clear
that the compliant properties of coronary vessels playa major
role. Further investigations have to be done.
REFERENCES

1. Arts T, Meerbaum S, Reneman RS, Corday E: Tors ion of the

canine left ventricle during the ejection phase in the
intact dog. Cardiovasc Res 18: 183-193, 1984.
2. Arts T, Reneman RS: Measurement of deformation of canine
epicardium in vivo during cardiac cycle. Am J Physiol 239:
H432-H437, 1980.
3. Arts T, Reneman RS: Interaction between intramyocardial
pressure and myocardial contraction. J Biomech Eng 107:
51-56, 1985.
4. Arts T, Veenstra PC, Reneman RS: A model of the mechanics
of the left ventricle. Ann Biomed Engng 7: 299-318, 1979.
5. Arts T, Veenstra PC, Reneman RS: Epicardial deformation and
left ventricular wall mechanics during ejection in the dog.
Am J Physiol 243: H379-H390, 1982.
82

6. Bache RJ, Schwartz JS: Myocardial bloodflow during excer-

cise after gradual coronary occlusion in the dog. Am J
Physiol 245: H131-H138, 1983.
7. Beyar R, Sideman S: A computer study of the left ventricu-
lar performance based on fiber structure, sarcomere dyna-
mics, and transmural electrical propagation velocity. Circ
Res 55: 358-375, 1984.
8. Bogen DK, Rabinowitz SA, Needleman A, McMahon TA, Abelman
WH: An analysis of the mechanical disadvantage of myocar-
dial infarction in the canine left ventricle. Circ Res 47:
728-741, 1980.
9. Chadwick RS: mechanics of the left ventricle. Biophys J 39:
279-288, 1982.
10. Downey JB, Kirk ES: Inhibition of coronary blood flow by a
vascular waterfall mechanism. Circ Res 36: 753-760, 1975.
11. Edman KAP, Mulieri LA, Scubon Mulieri B: Non-hyperbolic
force velocity relationship in single muscle fibers. Acta
Physiol Scand 98: 143-156, 1976.
12. Falsetti HL, Mates RE, Grant C, Greene DG, Bunnell IL: Left
ventricular wall stress calculated from one plane cinean-
giography - An approach to force-velocity analysis in man.
Circ Res 26: 71-83, 1970.
13. Feigl EO, Fry DL: Intramural myocardial shear during the
cardiac cycle. Circ Res 14: 536-540, 1964.
14. Feigl EL, Simon GA, Fry DL: Auxotonic and isometric cardiac
force transducers. J Appl Physiol 23: 597-600, 1967.
15. Fenton TR, Cherry JM, Klassen GA: Transmural myocardial
deformation in the canine left ventricular wall. Am J Phy-
siol 235: H523-H530, 1978.
16. Fung YC: Mathematical representation of the mechanical pro-
perties of the heart muscle. J Biomech 3: 381-404, 1970.
17. Fung YC, Zweifach BW, Intaglietta M: Elastic environment of
the capillary bed. Circ Res 19: 441-461, 1966.
18. Gallagher KP, Osakada G, Hess OM, Koziol JA, Kemper VlS,
Ross J: Subepicardial segmental function during coronary
stenosis and the role of myocardial fiber orientation. Circ
Res 50: 352-359, 1982.
19. Garrison JB, Ebert Wl, Jenkins RE, Yionoulis SM, Malcom H,
Heyler GA, Shoukas AA, Maugham WL, Sagawa K: Measurement of
three-dimensional positions and motions of large number of
spherical radioopaque markers from biplane cine-angiograms.
Comp Biomed Res 15: 76-96, 1982.
20. Henderson AR, Ocken E, Brutsaert DL: A reapraisal of force-
velocity measurements in isolated heart muscle prepara-
tions. Eur J Cardiol. 1: 105-118, 1973.
 83

21. Henderson Y: The volume curve of the ventricles of the

mammalian heart and the significance of this curve in res-
pect to the mechanics of the heart beat and the filling of
the ventricles. Am J Physiol 16: 325-367, 1906.
22. Hill AV: The transformation of energy and mechanical work
of muscle. Proc Physiol Soc 51: 1-18, 1939.
23. Hort W: Makroskopische und mikrometrische Untersuchungen am
Myocard verschieden stark gefuellter linker Kammern. Vir-
chows Arch (Pathol Anat) 33: 523-564, 1960.
24. Huismans RM, Sipkema P, Westerhof N, Elzinga G: Comparison
of models used to calculate left ventricular wall force.
Med & BioI Eng & Comput 18: 133-144, 1980.
25. Janz RF, Grimm AF: Finite element model for the mechanical
behavior of the left ventricle. Circ Res 30: 224-252, 1972.
26. Kim HC, Min BG, Lee MM, Seo JD, Lee YW, Han MC: Estimation
of local cardiac wall deformation and regional wall stress
from biplane coronary cineangiograms. IEEE Trans BME 32:
503-511, 1985.
27. Lab M, Woollard KV: Monophasic action potentials, electro-
cardiograms and mechanical performance in normal and ische-
mic epicardial segments of the pig ventricle in situ. Car-
diovasc Res 12: 555-565, 1978.
28. Pao YC, Ritman EL, Wood EH: Finite-element analysis of left
ventricular myocardial stresses. J Biomech 7: 469-477,
1974.
29. Patterson SW, Pieper H, Starling H: The regulation of the
heart beat. J Physiol 48: 465, 1914.
30. Pollack GH, Krueger JW: Sarcomere dynamics in intact car-
diac muscle. Eur J Cardiol 4 (suppl): 53-65, 1976.
31. Prinzen FW, Arts T, van der Vusse GJ, Coumans \'VA, Reneman
RS: Gradients in fiber shortening and metabolism across the
ischemic left ventricular wall. Am J Physiol 250: H255-
H264, 1986.
32. Prinzen FW, Arts T, van der Vusse GJ, Reneman RS: Fiber
shortening in the inner layers of the left ventricular wall
as assessed from epicardial deformation during normoxia and
ischemia. J Biomech 17: 801-812, 1984.
33. Prinzen TT, Arts T, Prinzen F\'J, Reneman RS: Mapping of
epicardial deformation using a videoprocessing technique. J
Biomech 19: 263-274, 1986.
34. Rushmer RF, Franklin DL, Ellis RM: Left ventricular dimen-
sions recorded by sonocardiometry. Circ Res 4: 684-688,
1956.
35. Sandler H, Dodge HT: Left ventricular tension and stress in
man. Circ Res 13: 91-104, 1963.
36. Sonnenblick EH, Spiro D, Cottrell TS: Fine structural
changes in heart muscle in relation to the length tension
curve. Physiology 49: 193-200, 1963.
84

37. Spaan JAE: Coronary diastolic pressure-flow relation and

zero flow pressure explained and on the basis of intramyo-
cardial compliance. Circ Res 56: 293-309, 1985.
38. Spaan JAE, Breuls NPW, Laird JD: Diastolic-systolic corona-
ry flow differences are caused by intramyocardial pump
action in the anesthetized dog. Circ Res 49: 584-593, 1981.
39. Streeter DD, Hanna WT: Engineering mechanics for successive
states in canine left ventricular myocardium: II. Fiber
angle and sarcomere length. Circ Res 33: 656-664, 1973.
40. Ter Keurs HEDJ: Calcium and contractility. In: Cardiac
Metabolism, ed. Drake AJ, Noble MIM, John Wiley, 73-99,
1983.
41. Van den Horn GJ, Westerhof N, Elzinga G: Optimal power
generation by the left ventricle. Circ Res 56: 252-261,
1985.
42. Westerhof N, Elzinga G, Van den Bos GC: Influence of cen-
tral and peripheral changes on the hydraul ic input impe-
dance of the systemic arterial tree. MEd Biol Eng 11: 710-
722, 1973.
43. Wood RH: A few applications of a physical theorem to mem-
branes in the human body in a state of tension. J Anat
Physiol 26: 362-370, 1892.
44. Yin FCP: Ventricular Wall Stress. Circ Res 49: 829-842,
1981.
 85

FLUID MECHANICS IN CARDIOVASCULAR RESEARCH

CARDIAC VALVE FLOW DYNAMICS
l.H. Kang
Biomechanics Laboratory. Chengdu University of Science and
Technology. Chengdu. Sichuan. China

It seems that up to now, through developments of generations and genera-

tions, fluid mechanics may be considered as studies of flow of fluids and
their responses. The responses may be of the fluid itself, or of any other
bodies related with the flow of fluid. Those bodies may be within the bulk
of the fluid, beside it, enclosing it or in any other manner. The responses
may be mechanical, thermal, chemical, electro-magnetical and even physiolo-
gical. So we have the branches of study such as: Basic fluid mechanics,
Fluid flow and Heat transfer, Physical and chemical fluid mechanics, Elec-
tro-magnetic fluid mechanics and Bio-fluid mechanics or Physiological fluid
mechanics. Each of these studies shows the consideration of coupling ef-
fects of different flow and responses.
It is true that fluid mechanics has played an important role in Bio-medi-
cal engineering, especially in Cardiovascular research, with emphasis on
blood circulations for quite a long time. The historical period of contri-
butions in this field of study covers early from that of Leonardo da Vinci
in 1513, who found that the vortices in the sinuses of val salva of the aor-
tic valve were actually responsible for the closing mechanism of the valve.
Later on, we had many great contributors to cardiovascular research such as
Harvey, Euler, Young, poiseuille, Helmholtz, Fick, Lamb and Fung of our age
and many others, with all their contributions being closely related with
fluid mechanics. The fields of interest have spread widely to such as blood
flow and wave propagation in arteries, blood flow in veins as collapsible
tubes, blood flow in the lung, hemorheology and micro-circulations etc. At
the same time, it is noticeable that since the years of 1960s, there arises
another trend of study of fluid mechanics in carda,ovascular research, speci-
fically related with cardiac valves, which forms the Cardiac Valve Flow Dy-
namics, showing the evidence of ever growing integration of fluid mechanics
with more aspects of cardiovascular surgical treatments. It has grown up so
rapidly and fruitfully that a short review could not cover even the main
points of it on time. This paper is just an introduction to such a branch
of study - - cardiac valve flow dynamics.
What are the characteristics of the Cardiac Valve Flow Dynamics?
Firstly, so far as the natural heart valves are concerned, on one hand,
they are so fancy for their material, structure and configuration that their
perfection makes us feel they are almost untouchable. Yet on the other
hand, they have been proved physiologically allowable of being taken as a
kind of "passive element" in the analysis of their motion in blood flow.
By passive elements, we mean that they open and close under the forces on
them by the surrounding stream of fluid. There is no source of motion of
the valves themselves, so that fluid mechanics could play an important role
in studying them.
Secondly, the flow field of natural cardiac valves, seemingly simple, is
actually rather complicated. Effects of various causes are usually coupled
86

in action. For example, one of the features of flow which deserves exten-
sive study is that the valve starts to close even when the blood fow is
still in its accelerated ejection phase, a phenomenon called "earlier par-
tial closure". This makes the closure of normal natural valve so smooth
and gradual that causes very little disturbance and very low pressure rise.
Dealing with such mechanisms still remains as study of an unsolved fluid
dynamical problem.
Thirdly, for seriously diseased natural valves, instead of using other me-
dical treatments as in the past days, now surgeons consider the replacement
with artificial ones as the best choice. It has been over twenty years sin-
ce different types of prosthetic cardiac valve become available for clinical
use, yet no one could complete with the normal natural ones in performance.
It is a serious task of study to find out and to fill up such a kind of gap
as much as possible. Unless this is accomplished, we could not actually
expect the development of any artificial valve which could be proved to be
truly satisfactory.
Fourthly, patients with valvular disease, are mostly due to rheumatism,
rheumatic endocarditis or congenital abnormalities. The symptoms are usual-
ly of two kinds: (1) stenosis -- the calcification of valve from antecedent
abnormalities or other causes thickens the valve leaflets and leaves their
surface rough and irregular. These calcified deposits build up and spread,
causing severe obstruction to the blood flow; (2) Deficiency -- incomplete
closure of valve, thus causing excessive amount of blood regurgitation. All
of these abnormalities have their consequences in the flow-field characteri-
stics, such as turbulance, jet flow, back flow, etc. Thus fluid mechanics
can effectively be used in the diagnosis of valve disease. From the consi-
deration of the characteristics mentioned above, we may summarize the purpo-
ses of such branch of study as follows:
a) To understand the function and behaviour of natural heart valves;
b) To help diagnose valvular diseases and its complications;
c) To guide the development of artificial heart valves.
Some of the state-of-the-art of studies of cardiac valve flow dynamics
may be mentioned as follows:
The first main topic of cardiac valve flow dynamics is the field of blood
flow through natural or prosthetic heart valves, which is perhaps one of
the more complex problems in cardiovascular dynamics. Human valve was at
first considered, from the point of view of fluid dynamics, as an area re-
duction, so the orifice flow equation for the relationship of pressure and
flow was used for study. Thus we have such as:
Gorlin equation (1951): LIp = kQ 2
which was derived from Bernoulli's equation for steady flow.
dQ
Yellin equation (1972): LIp = klQ{Q] + k2dt (1)

with local acceleration; being considered,

where / ' d ance coe ff"~c~ent =2"
k 1 = ~mpe A3
PfdA

A
' t '~a coe ff"~c~en t = -P fdX
k 2 = ~ner

x means flow direction; A means the cross-sectional area; A=A(t), P means

the fluid density. The expression of kl and k2 were derived on the assump-
tions of one-dimensional flow and the cross-sectional area of flow being
not strongly dependent on time.
In recent years, the investigation directly related with or indirectly
leading to flow patterns around heart valves have far-reaching advancements.
For example:
Steady flow~ Pulsatile flow (Womersley, 1955)
 87

One dimensional flow + Two dimensional flow (Peskin, 1972)

Laminar flow + Turbulent flow (Yoganathan, 1985)
Coupling effects of fluid motion and valve motion (Peskin, 1972)
Analytical to computational method (Greenfield, 1972)
In vivo observations with Cinematographic high-speed recordings
(Steenhoven, 1982)
Modelling of flow around heart valve (Phillips, 1983)
All of these show the study of flow around heart valves proceeds in acco-
rdance with the developments of analytical, computational and experimental
fluid mechanics.
The second main topic of cardiac valve flow dynamics is the mechanism of
motion of natural heart valve dealing with the process of opening and clo-
sure of valve leaflets.
For the opening process, say for the aortic valve, it was shown that the
valve was pushed forward without opening for the first 53 ms of the systole,
and then reached the fully open position in 47 ms, so the period of opening
is very short, flow is highly accelerated, thus the viscous term in the Na-
vier-stokes equation is negligible compared to the inertial term. Then the
flow pattern during the opening phase can be approximately represented by a
potential flow. The purposes of study are to measure pressure gradients,
position of the leaflet and describe the opening procedure, etc. For these
studies, experimental descriptions were made by Bellhouse and Talbot (1969);
Analysis of numerical methods were used by Huang and Schuessler (1977).
The thing which is very interesting is that the aortic pressure was found
to be very sensitive to the shape and the speed of the moving leaflet, in-
dicating that the change of the geometrical and material characteristics of
the valve leaflet is critical to the fluid dynamics of the blood flow in
the aorta.
Closing mechanism of valve is by far the more important problem to be
studied and is still remained as an unsolved problem. It is true that the
dynamic effect of cardiac contraction is surprisingly effective in moving
blood through the heart even in the absence of competent valves. Neverthe-
less, these thin and delicate cardiac valves, which guard the entrance and
exit of each ventricle, greatly enhanced the efficiency of the cardiac pump
by preventing backflow. In early ages, on the contrary, there prevailed a
kind of consideration that cardiac valve closes due to blood backflow. La-
ter, the misunderstanding was cleared up, yet some kind of artificial car-
diac valve during earlier days such as the caged ball valve was designed on
the basis of such closure mechanism. Even now, it is really surprising to
find out that most mechanical cardiac valves close due to backflow actually,
though they are not designed so intentionally. Experiments of in vitro
closing behavior of BJork-Shiley, St.Jude and Hancock heart valve protheses
in relation to the in vivo aortic valve closure by Steenhoven (1982) show
that the natural valve closure is very gradual, its closure has already
started even during the blood is still in the acceleration phase, and it
continues gradually during the deceleration phase of systolic ejection.
Consequently, only minor backflow in the valve is required to complete va-
lve closure. But for most of the mechanical prostheses now available, they
close only for 5% of their cross-sectional area during the systolic ejec-
tion phase. The Hancock bioprostheses seems better, it closes 45% during
the flow deceleration phase of, systole, which is however significantly less
than the corresponding closure of the aortic valve as recorded in vivo{74%).
Due to the correspondingly higher back-flow value of mechanical valve, the
shear stresses during the backflow phase of diastole will increase as well
as the pressure rise just after valve closure. Other observations of pro-
thetic heart valve flow field show the obvious appearance of turbulent flow,
88

eddies and wakes. From all these effects, it is predictable though undesi-
rable that the incidence of hemolysis will increase. Moreover, no earlier
partial closure during flow acceleration phase can be observed for any type
of artificial cardiac valve so far developed. This is really something re-
markable.
Leonardo da Vinci(1573) is the earliest researcher on the closing mechan-
ism of natural heart valve. He first realized that there is a sinus behind
each cusp of the aortic valve and these vortices must play an important ro-
le in the closing motion of the valve. However, there after no significant
progress about this work was made until 60s of this century because few
people devoted to it. Up to 1960s, following the development of artificial
heart valves, this study unfolded quickly. In 1968, B.J.Bellhouse et al.
advanced their famous "vortex theory" first about closing mechanism of the
aortic valve with the help of series of theoretical analysis and experimen-
tal studies, later, he extended this theory to mitral valve. He explained
that (1) the vortex behind the cusp comes out as soon as the cusp tip goes
into the sinus and last till the valve reaches its closed position; (2) the
vortex in sinus can be modelled by half of a Hill spherical vortex, and is
just the decisive factor of the valve closing; (3) the adverse pressure
gradient produced by flow deceleration can set up a pressure difference
across the cusp only with the help of vortex pushing the valve to move to-
wards closure.
Van. Steenhoven et al.(1979,1982) and H.Reul et al.(1981) found by expe-
riments that the valve can also close without vortices or only with a fain-
ter vortex and in the same time its closing manner is unchanged. They put
forward "adverse pressure gradient theory", in which the adverse pressure
gradient by flow deceleration was considered as the main factor of valve
closure other than the vortex during the whole process of valve closure.
This theory may be simply expressed as follows:
dU dU 1 dP
at + u dX = - P dX (2)

where u is axial velocity, p is pressure, p is density.

Peskin et al. (1982) put forward the "hidden vortex theory". According to
this theory of valve closure, vortex is hidden and not explicit. The circu-
lation around this contour is clearly not zero, so there is vorticity con-
centrated along the valve leaflet themselves.
According to the above-mentioned vortex theory, z.J.Li et al.(1982) of
the Biomechanics Laboratory of CUST(Chengdu University of Science and Tech-
nology), in order to define the closure area of aortic valve, derived a dif-
ferential equation:
(3)
where V is velocity of blood ejection, Vo is peak velocity of blood ejec-
tion, Ro is radius of aortic root, R is radius of valve. From this equa-
tion, we got at the end of decelerated ejection period, the ratio of aortic
valve closure area to the aortic cross-sectional area would be about 61%
which is not far from the normal experimental result.
According to vortex theory, if there isn't a powerful vortex, the valve
should show an insignificant earlier closure, then a great amount of regur-
gitation would follow naturally. This is obviously in contradiction with
the fact that the adverse pressure gradient theory relies. But the adverse
pressure gradient theory is only good for the deceleration phase of blood
ejection, and it cannot explain the phenomenon of earlier partial closure
of the valve during the blood ejection is in the acceleration phase. For
hidden vortex theory, it lacks definite result, in explaining the closure
mechanism of valve. In a word, it is obvious up to now that the moving
 89

mechanisms of the valve during both the opening period and the closing pe-
riod in the deceleration phase of blood ejection have been controlled by the
pressure gradient produced by the varing motion of blood ejection. However,
it is the earlier partial closing mechanism during the acceleration phase
remains to be an unsolved puzzle. Aiming at this, Z.H.Kang and M.Lei of
CUST put forward their study. The consideration is as follows: (1) consi-
der all the forces and moments contributing to the valve motion instead of
only some of them being considered; (2) consider the whole process of clo-
sure motion instead of taking only a certain portion of it into account.
For experimental study, we used a quasi-two-dimensional model. The aorta
was supposed to be with a square section; the radius of the semi-cylindri-
cal sinus was one half of the side length of the aorta section; and two pi-
eces of yak pericardium of rectangular shape with thickness of about 0.5rnrn
were taken as cusps of the valve. The water used was physiological saline.
The flow field was shown by using neutral resin particles with diameters of
some ten to hundreds~. The movement of valve leaflet and the flow pat-
tern near the valve were filmed simultaneously with a 16 rnrn cinecamera, the
filming speed used being 36 frames per second. with the help of GP-200film
motion analyzer, the motion curves 8(t) of the valve were obtained. Mean-
while, the flow and pressure curves at each point near the valve were reco-
rded. Through this experiment, some results were obtained as follows:
1. Figure 1 shows the clo-
sing manner curves of the two
cusps as they close asynchro-
nously. From this figure it 8
is clear that the earlier par-
tial closure of the valve du-
ring acceleration phase is
due to the vortex.
2. The sinus vortex appears
as the cusp goes into the
sinus and attenuates quickly
as the cusp goes out. Near
the downstream edge of sinus,
there is a small reverse se-
condary vortex occuring as
the sinus vortex diffuses.
3. The valve cusp will os-
cillate at 8=0 0 and stay the- t(s)
re for a c.ertain time.
4. Except when approaching 0.2 0.4 0.6 0.8 1.0
A
the close position, the cusp
essentially remains as a flat
plate. FIGURE 1. Experimental Results of
For theoretical study, a the Valve Leaflet Motion; Curve I
2-D analysing model was used has a Stronger Vortex than Curve IT
(Fig.2)
The problem was to find out all the factors making contributions to the
closing moment of valve leaflet based on considering the whole dynamic pro-
cess of motion of valve leaflet. Using the conformal mapping method for
analyzing the assumed flow field. The assumptions were: two-dimensional
flow around a thin leaflet with the aortic sinus; incompressible fluid;
with certain circulation around the leaflet and otherwise irrotational;
sinus pressure being considered as constant and equal to the pressure at
the cusp tip, etc. As a result of such analysis, from the theoretical ex-
pressions of velocity distribution, and pressure distribution, we got the
90

FIGURE 2.
--, Two-Dimensional
y
/
" , \L
Analysing Model
____ _ of the Valve

--
- o
,P s I
I
... x

,-,,'"

moment equation for the valve

leaflet as follows: M

X Me
3000 0 Me 2

inertia moment; 0 Mev

2 ' 2
Me = F2C 8 , called
2000
Mv 2

steady closure moment;

Mev = (F3sin8+F4cos8)Cev2+
+Q,o (F9C~+Fj QVjsin8),
called steady drag moment;
M 2 = (Fscos8+F6sin8)vj2sin8,
v
called retrieval moment;
M = (F7sin8+Facos8)Cdvj ,
Dv dt
called velocity gradient
moment;
Q,2 1 2 dVj 1
Mv = T[ '2 Vj cos8- (2a-Q,) dt >(

Xcos8re-S8 , called
vortex closure moment.
From both the experimental
and theoretical investigation,
we arrived at the following
main conclusions:
1. The motion of the valve is
due to the interactions bet-
ween the valve and fluid, in-
cluding contributions of many
factors;
2. During the opening phase,
the main factor of motion of
the valve is the acceleration FIGURE 3. Variations of Terms
flow, and the deceleration in Eq.4 with Vortex (MvO)
flow plays the main role
 91

during most time of the decele-

ration phase. But during the
quasi-steady phase of blood eje-
M ction, it is the vortex in the
sinus that controls the valve
X M~ closure;
3000
o M~2 3. For the earlier partial clo-
sure of the valve during the ac-
a M~v celeration phase, the vortex is
a decisive factor, but its fun-
ction is limited. It was found
2000 in some time that without vortex,
valve can still close in a simi-
lar manner, but there would be
no earlier closure during the
acceleration phase as compared
1000 with the situation of with vor-
tex;
4. The existence of the sinus
cavity is a prerequisite of the
t(s) effective close of the vlave.
~-t-1~~~~~~~~ucs..~L--------- Its main function is to let the
valve get a certain amount of
closure before the reversing of
the blood flow.
The third main topic of cardi-
-1000
ac valve flow dynamics is the
development of artificial heart
valves. Of course, this depends
to a large extent on fluid me-
chanics. What aerodynamics is
-2000
to airplane design, just shows
what bio-fluid mechanics should

j
be to the design of artificial
earlier closure during heart valve, whether tissue val-
-3000 accel. phase ve or mechanical valve. Up to
now, there still exist some pro-
opening decel. phase blems of current artificial heart
phase valves. They can be mentioned
as: High pressure gradient ac-
ross the opening of valve; Ste-
FIGURE 4. variation of Terms in nosis due to small effectual
Eq.4 without Vortex (Mv=O) opening area; Dysfunctions of the
valve components due to abnormal
stress condition; High percentage of back flow; Throwbosis and hemolysis
formation; Calcification of the leaflet, etc. Cardiac valve flow dynamics
is making substantial contributions towards solving these problems.
For tissue valve, among a lot of fluiddynamical investigations made in
recent years, M.Hou in CUST, are making design and analysis of valves with
trileaflets by trying to solve the problems of insufficient opening of val-
ve and high pressure gradients thus produced during systole. Differentsha-
pes of cusp including spherical, parabolical, elliptical and cylindrical
ones were examined, characteristic geometrical parameter were defined,
their effects on the opening orifice area were analyzed, and then better
shapes were suggested. It seems that the parabolic cusp may be the optimum
92

cusp configuration since it mainta-

,\ ins its shape from the closed to
fully open position. The opening
1.2 Cil A process of the leaflet in a two-
Q(cm 3 /s) dimensional model was described dy-
1.0 namically and solved using the boun-
f---t-~=~~=~500
I 473 dary element method.
The governing equation for poten-
0.8 400 tial flow is the Laplace equation;
denotes the potential function:
(l2 (l2
0.6 P+(ly2 =0 in S1 (5)

0.4 S1 is a J domain circled by boundary

f, the Laplace equation belongs to a
boundary value problem: Once the
0.2 in the boundary is given, the equa-
tion will have an unique solution.
0 t(s) Now in our problem, the mixed boun-
0.5 dary conditions were given, and the

FIGURE 5. Motion Curves of y

the Valve Leaflet with a
Sinusoidal Flow, where the
Closing Parameter,\= 1-!:.sin8,
a & L shown in Fig.l. a
15

boundary was divided into

~~ ______________~~~__________~fl~7___X
four parts f], f2' f3 and
f, as shown in Fig.6 32
18
19 31
in f2,f,
20 21 22
~~ =-voe-rtsinwt in f] (6)
=0 in f3 25
where n designates the nor- a) Boundary Element Model
mal variable.
The leaflet position _________________________________ y
equation is:
Xi (t+6t) =xi_l(t+6t)+
+icos[ai(t)+68il
Yi(t+6t) =Yi_I(t+6t)+ (7)

+isinlai(t)+68il
vi(t) -Ni(t) t
8i = 6t
i
x
The coordinates and seg- b) Leaflet Element Model
ment points refer to Fig.6:
Li -- Length vector of
ith segment; FIGURE 6. Boundary Element
vi -- Velocity vector of Model of the Valve Motion
 93

the end point of ith segment;

tJ.t - Small step of time.
It was found that the larger the flow rate, the faster the leaflet opened.
It was shown that the span of the leaflet in the meridian plane significant-
ly affected the opening orifice, and that the optimum length is about three-
fourths of the sinus height, which agrees with the average value of the na-
tural aortic heart values(about 0.7-0.75 times the aortic diameter). A se-
ries of in vitro hydrodynamic experiments were carried out using cinemato-
graphic high-speed recordings. The experimentally determined ideal cusp
length of valve is just the same as the results of hydrodynamical analysis.
This implies that the natural heart valves are really the optimum design.
The experimental results also indicated that the small geometrical opening
orifice area and the oscillation behavior during the steady opening phase
were the major reasons for high pressure gradients across the valves during
the opening period. Use of the parameters and formulas established in this
study may enable the prosthetic valve to simulate the natural heart valve
more closely. We have developed a kind of tissue valves with yak pericar-
dium with the configuration of'lower stent and larger opening area for prac-
tical use(Fig.7). They have been implan-
ted to more than 40 patients for a year,
also no information of trouble has been
fedback from the hospital so far:
Mechanical valves have been developed
in the recent two decades. Different
shapes of occhider are provided, such as
ball type, disc type, single leaflet type,
bileaflet type, flat plate type, convex-
concave type, etc. ~t is on the basis of

FIG~ 7. The Tissue Valve

airfoil theory of aerodynamics that

a type of hinged bileaflet mechani-
cal valve with cambered profile has
been designed in our laboratory by
Z.H.Kang et al.(1983) for model
test. Here we found that the cam-
ber combined with other parameters
can be designed to fulfill the he-
modynamical requirements advantage-
FIGURE 8. The Bileaflet Valve ously. That is to say, we have mo-
re means to adopt for optimum de-
sign. Take the valve leaflet as an airfeii, the lift.,drag and moment have
close relationship with the camber, chord and location of hinge point. By
properly choosing these parameters, we have more freedom to provide better
valve performances, including better closure mechanism, minimum resistance
in blood flow, etc. The numerical solution of the flow field of cambered
bileaflet mechanical valve by Z.S.Guo of CUST using Galerkin finite element
method is shown as follows:
a) Model: two dimensional and steady viscous flow.
b) Method:
94

.j.J
(])
......
.j.J
~
o

Symmetrical Axis

FIGURE 9. The Cambered Valve (Testing Model)

1. dimensional equations:
t,.VV+Vp =~v (8)
Re
V.t, = 0 (9)
pvmaxr
where Re = - - -
jJ
2. finite element:
c) Some results:
1. Re was taken up to 600;
2. ~p increases, and the
flow stagnation and seperation
become serious as f(camber/
chord) increases;
3. increasing of Re results
in serious stagnation and se-
peration; FIGURE 10. Finite element model of
4. wall shear stress on the the cambered valve
outer surface of the leaflet
is much higher than that in the inner surface.
The next important topic of cardiac valve flow dynamics is model testing.
In in vitro flow testing system, both steady and pulsatile flow tests are
conducted, but for different purposes. For steady flow test, the aim is
mainly for comparative study of characteristics of different types of valve.
It's quite different from the actual physiological flow which is pulsatile,
so the results of steady flow test could not be put to use directly. The
similarity parameters are Reynolds number, Euler number and the form factoL
For pulsatile flow test, more contents and similarity parameters are r~
ired. For the valve, opening and closing time, movements of leaflet or oc-
cluder, functioning manner are required; For the flow, the pattern of flow
field, laminar and turbulent region, wake and stagnation area, velocity and
shear stress distribution, forward and backward flow measurements, variation
of pressure difference across the valve and energy loss across the valve are
required. To fulfill all the similarity requirements is actually impossib-
le. Up to now, the main requirements are: the similarity of non-steady flow
process, geometrical similarity and fluid similarity. Present trend of ef-
forts of study in pulsatile flow test is toward the goal of standardization
 95

so that test results of different sources may be compared and checked with
each other.
In the field of experimental study of cardiac valve flow dynamics, fati-
gue test is also an important one. with the heart pumping 70-80 times per
minute, the heart valve opens and closes about 40 million times a year. The
systole and diastole pressure of the heart act on it, the flowing blood wa-
shes it day and night without stopping. What's their effect on the valve,
especially of fatigue, is a serious problem to be considered. For simula-
tion test, frequency of motion should be made high in order to obtain use-
ful test results within a certain permissible range of time. The absolute
pressure should not be lower than the physiological value. Full valve open-
ing and closing should be insured, this actually becomes the limiting fac-
tor of frequency. It is required for the test apparatus to test several
valves simultaneously, to keep the phase difference between the flow and
pressure curve to be minimum, to have temperature easily controlled and val-
ve easily mounted and observed.
Some of our experimental works in the development of cardiac valve flow
dynamics are as follows:
(1) Hydraulic analog of the systematic circulation: From the Navier-Sto-
kes equation of incompressible Newtonian blood flow in linear artery system,
using the analog with electric transmission system, we can derive out apair
of equations as follows(J.K.Cheng et al.of CUST,1984):
dP dQ
az = -RoQ-L at
(10)
dQ dP
az= -GoP-Coat
where P ,Q are fluid pressure and flow rate respectively; Ro=~'IT4' Lo= ~ ,
o . d 'IT~o
dV) /
Go= ( at dV
P, Co = 3P are ca 11 e d d'lstrl. b ute d f 1 Ul. d '
reslstance, f'ITlr
ow In uctlon
(inertia), flow conduction (wall leakage) and fluid capacity (elasticity of
artery wall) respectively; r o- the radius of the vessel; p -the density of
blood; v - the volume of the vessel with a unit length.
By the electrical-hydraulic analogue theory, the flow parameters can be
simulated correspondingly by the electric parameters. The corresponding
relationships between them are as follows:
flow rate Q - current I, pressure P - voltage V,
elasticity Co - capacitor C, inertia Lo - inductor L,
resistance Ro - resistor R.
Figure 11 is the sketch of our test system as well as its corresponding si-
mulated circuit. In the design of the system, a "Distributed-segmental lu-
mped parameter model" is used. The test section, including the left ven-
tricle, left atrium, ascending aorta and the arch of aorta is considered as
a distributed parameter system, other parts are divided into several lumped
parameter system. Then electrical circuit analog technique is used for sQL-
ving the system equations to obtain each parameter. The values thus obtai-
ned have been confirmed by experL~ents.
(2) Electrochemical method was used to determine the shear stress distri-
bution on the vessel wall in front of and in the rear of the valve by X.Y.
Deng of CUST (1983). Comparative performances can be obtained with this
method easily for different valves. This technique is based on diffusion-
controlled reaction of ferricyanide ion to ferrocyanide ion at a cathode.
3 cathod -4
Fe(CN)6' +e ~_ _ _ ) Fe(CN)6 (11)
anode
cathod and anode were both made of nickel wire. There exists a relation-
ship between wall shear stress and measured limiting current:
96

Aorta
Resorvoir

a) the simulated system with 3

lumped-parameter sections

Q(t) _ _ L3 Rl
~~.<=r~~~~~~~MC~

p (t) Cl

b) the simulated circuit

FIGURE 11. The Sketch of the Simulated Circulation

System as well as Simulated Circuit

3.893
T w =)J -(:-F-'C'--b'--)'-'3:C-d-::-C-
5 D----;C2- i 3 (12)

where )J is fluid viscosity, F - - Faraday's constant, D - - diffusion coeffi-

cient, d - - the cathode diameter, Cb-- the ion species concentrations in
the bulk flow and i - - the measured limiting current.
(3) Hydrogen-bubble technique has been used for visualization of valve
flow study by Y.N.Mao of CUST (1986), especially the flow field pattern.
(4) Fatigue testing system using fluidics has been tried with promising
results by S.Q.Zou et al.(1985) of CUST.
(5) Routine functional testing system of valve for clinical use has been
made.
(6) Videotape recording has been used for showing the motion of test val-
ve on the analog test stand.
(7) Computer-aided-technique are applied mainly.
At last, I would like to mention some aspects of possible developments of
topics of cardiac valve flow dynamics and problems to be taken into consi-
deration in the near future in this field.
1. Establishment of correlation between in vivo and in vitro testings of
valve flow field;
2. Determination of the evidence and related factors making the differen-
ces between the function and morphology of the normal heart valve and
 97

~ 120
~
!

~-.. '. -I

15~~;:fj;
FIGURE 12. The Simulated Circulatory FIGURE 13. CUrves produced by
System the Simulated System

diseased heart valve through fluid dynamical studies;

3. Computer aided design and analysis of artificial heart valves;
4. Turbulence flow, stagnation and regurgitation in the flow field, and
the formation of the thrombus due to improper design of artificial valves;
5. Prediction of the nature of failure from accelerated fatigue testing
of unimplanted valve;
6. Studies of impact of the flow ejection and the characteristics of the
jet flow and its responses in the valve flow field.

REFERENCES
1. Bellhouse,B.J. & Talbot: The Fluid Mechanics of the Aortic Valve.
J.Fluid Mech., vol.35,part 4, pp.72l-735, 1969.
2. Bellhollse,B.J.: Velocity and Pressure Distribution in the Aortic Valve.
J.Fluid Mech., vol.37, part 3, pp.587-600, 1969.
3. Cheng,J.K.et al.: Design Theory and its Application of Pulsatile Flow
Circulation Analog System. Collections of Biomech.Research.CUST.,P.R.C.
1984.
4. Deng,X. Y.: An Investigation on Electro-chemical Measurement of Wall Shell'
Stresses in the Flow Field of Cardiac Valve Prosthesis. Proceedings of
the Second Asian Congress of Fluid Mec~nics. Beijing, 1983.
5. Fung,y.C. Biodynamics, Circulation. Springer Verlag.,New York Inc.,1984.
6. Gorlin,R.& Gorlin,S.C. Hydraulic formula for calculation of area of ste-
notic mitral valve. Am.Heart J. 41, 1951.
7. Guo,Z.S. Numerical investigation of planar steady flow around cardiac
valve prosthesis with bileaflet of cambered profile. MS thesis,CUST.1984.
8. Greenfield,H.& W.Kolff: The prosthetic heart valve and computer graphics.
98

J.Am.Med.Assn., 219(1),1972.
9. Hou,M. Design and analysis of an artificial heart valve with trileaf-
lets. MS thesis. CUST. P.R.C. 1986.
10. Hung,T.K.& Schuessler,G.B. An analysis of the hemodynamics of the open-
ing of the aortic valves. J.Biomech.,vol.lO, pp.597-606, 1977.
11. Hung,T.K. Hydrodynamic analysis of the aortic valve mechanisms.
Adv. Cardiovasc.Phys., vol.5,part 1,pp.l06-118, 1983.
12. Kang,Z.H.et al. On the design of cardiac valve prostheses with bileaf-
let of cambered profile. Proc.of Intl.Sym.on Hemoperfusion and Artifi-
cial Organs. Tianjin, P.R.C., 1983.
13. Lei,M.& Kang,Z.H. Study of the closing mechanism of natural heart val-
ves. Applied Mathematics and Mechanics. vol.7. no.lO.,P.R.C.,1986.
14. Li,Z.J.et al. Mathematical model of aortic valve during the initial
closure process. Journal of CUST. no.2, 1982.
15. Mao,Y.N. Hydrogen bubble technique for study of thrombus formation with
bileaflet cardiac valve prosthesis. MS thesis, CUST.,P.R.C.,1986.
16. Peskin,C.S. Flow patterns around heart valves: A numerical method.
J.Comp.Physiol.lO.262, 1972.
17. Peskin,C.S. Numerical analysis of blood flow in the heart. ibid.25,1977.
18. Peskin,C.S. The fluid dynamics of heart valves:experimental,theoretical
and computational methods. Ann.Rev.Fluid Mech.14.235-259,1982.
19. Phillips,W.M. Modelling of flows in the circulatory system.
Adv.Cardiovasc.Phys.,vol.5,part 1, 1983.
20. Reul,H.et al. Fluid mechanics of the natural mitral valve.
J.Biomech., vol.14, no.5, pp.361-372, 1981.
21. Steehoven,A.A.Van & Dongen,M.E.H.Van: Model studies of the closing be-
havior of the aortic valve. J.Fluid Mech.,vol.90,part 1,pp.21-32,1979.
22. Steenhoven,A.A.Van & Veenstra,P.C. The effect of some hemodynamic fac-
tors on the behavior of the aortic valve. J.Biomech.,vol.15,no.42,
pp.441-450, 1982.
23. Steenhoven,A.A.Van et al. In vitro closing behavior of Bj~rk-Shiley,St
Jude and Hancock heart valve prostheses in relation to the in vivo re-
corded aortic valve closure. J.Biomech.,vol.15,no.ll,pp.841-848,1982.
24. Stevenson,D.M.& Yoganathen,A.P. Numerical simulation of steady turbu-
lent flow through trileaflet aortic heart valves.
J.Biomech. ,vol.18,no.12, 1985.
25. Womersley,J.R. Oscillatory flow in arteries, effect of radial variation
in viscosity of rate of flows. J.Physiol.,107.,38-39,1955a.
26. Womersley,J.R. Oscillatory motion of a viscous liquid in a thin walled
elastic tube. I. The linear approximation for long wave phil.Mog.46:
199-221, 1955b.
27. Womersley,J.R. The mathematical analysis of the arterial circulation in
a state of oscillatory motion, Wright Air Development Center, Technical
Report WADC-JR56-614, 1957a.
28. Womersley,J.R. Oscillatory flow in arteries: the constrained elastic
tube as a model of arterial flow and pulse transmission. Phys.in Med.
Biol.2:118-187, 1957b.
29. Yellin,E.L.& Peskin,C.S.,Frater RWM. Pulsatile flow across the mitral
valve: Hydraulic,electronic and digital computer simulation.
ASME paper 72-WABHF-IO,1972.
30. Zou,S.Q.et al. Fluidics used in fatigue test of cardiac valve prosthe-
ses. Proceedings of international congress on Fluid Control and Measure-
ment. Tokyo, Japan. 1985.
 99

THE DEVELOPMENT OF TECHNOLOGY IN MEDICINE AND BIOLOGY

W.J. Perkins
National Inst. f. Medical Research Mill Hill. London. England

1. INTRODUC7ION
The objective of the lecture was to look at the development of technology
in medicine and biology, and to stimulate thought about the present situat-
ion and the future direction [1-6].

2. EVOLUTION
At the time of the early philosophers they were able to study all sub-
jects. Even in the 15th century it was not uncommon for one person to
embrace a number of subjects, the supreme example being Leonardo da Vinci,
who was considered to be expert in art, astronomy, engineering and the
design of military weapons.
As man's knowledge increased it became necessary for each person to
concentrate their efforts upon a more limited and related range of subjects.
This led to specialisation so that medicine and engineering diverged into
separate disciplines with little evidence of any relationship between them
until the 18th century. This came with the development of electricity
which initially was used as a source of entertainment in drawing sparks
from young boys who were suspended by hempen ropes and charged to high
potentials by friction generators. This interested members of the medical
profession 'ino used bow driven generators to draw sparks from patients
in the hope of providing relief. Similar bizarre applications followed
with the use of low energy, high voltage electrodes being applied to
patients in attempts to cure deafness and to induce childbirth. In another
system a patient 'NoaS placed upon a raised platform which was charged to
a high potential. This device was claimed to provide the patient with
a sense of well-being. As electricity developed further with the intro-
duction of the Leclanche cell, more power became available and the applicat-
ions more hazardous. Thus for the treatment of severe respiratory
disorders, relatively high power was connected to the ailing patient, whose
hands were immersed in saline baths to which the input was connected by
the clinician who kept well back. In this case some success might have
been achieved through stimulation of the phrenic nerve but the technique
could also be lethal. Fortunately, common sense provided better results,
so that with the development by Faraday of eletromagnetism, Duchenne began
to apply the low voltage electrodes from the coils directly to the affected
muscles. However, by this time it was clear that it had been a mistake
to use electricity on patients without a proper understanding of its funct-
ion and with no capability of measuring the output energy of the generator
or the biological effect of the signal being applied. Quite reasonably,
the medical profession decided to stay with their own subject and to keep
away from technology which they did not understand. The possibility of
collaboration does not appear to have been considered at that time.
100

The first major applications of technology in medicine came from outside,

with the discovery of X-rays by Roentgen in 1895. This opened up a whole
new area for non-invasive diagnosis. Then in 1897 D'Arsonval discovered
that electro-magnetic waves above 100 K.Hertz did not stimulate the neuro-
muscular system and that the heat produced by this type of radiation in
bone and tissue could be used therapeutically.
Towards the late 19th, early 20th century, biomedical scientists,
particularly physiologists, began to adapt the principles of existing
technology for the development of their own instruments. The effect of
drugs upon the neuro-muscular system was studied by observing changes in
muscle reaction following an input stimulus. This transient movement was
recorded on to a rotating smoked drum by a marker attached to the particular
limb being stimulated.
To record inspired or expired air, the principle of the gasometer was
used in a miniature form. As air was breathed in or out it passed via
mechanical valves through an empty closed chamber floating on a liquid.
As the chamber rose and fell, a pointer attached to the chamber indicated
on a scale the level of air passing. Greater sensitivity was obtained
by attaching a potentiometer to the chamber instead of the scale and measur-
ing the current passed on a galvanometer. Greater sensivity came with
the amplification of this signal by a thermionic valve and an electrical
pen recorder.
These changes were introduced as the technology developed but the
principle of the system remained unchanged. It was not until about 1950
that greater technological expertise began to be applied to medicine and
biology, in both supportive and innovative roles. On becoming aware of
the biological purpose of the gasometer type spirometer, the electronic
engineer devised a transducer to produce a pressure difference across a
diaphragm and used an electronic differential amplifier to measure and
then record the air changes. The initial stage of measurement was of
interest to both electronic and mechanical engineers, as well as physicists,
in the development of suitable transducers and their associated measurement
systems. Whilst electronic engineers concentrated upon the measurement
of biomedical signals and the control of biological processes, mechanical
engineers were concerned with the development of transducers, prosthetic
and orthotic devices, and medical physicists became involved with the
further development of X-ray techniques, and nucleonics for diagnosis and
therapy.

3. RELATIONSHIP OF THE DIVERSE SUBJECTS

At the early stage of measurement and control there was already an
overlap of common interest between Medical Electronics, Medical Physics
and Biomechanics in respect of automated instruments, implants, tranducers
and stimulation of the neuro-muscular system. This could introduce
difficulties, in that users, in choosing a particular discipline to help
their problems, could also have chosen the technological method to be used.
In fact the three methods, physical, mechanical, electronic, needed to
be assessed for each problem which meant that the region where these three
disciplines overlapped needed to be combined. In organisational terms
this was achieved and the subject became known as Biomedical Engineering
(BME).

4. INFORMATION TECHNOLOGY
Measurement requires the acquisition of data but this in itself is
of little value until it provides information. To obtain this, additional
 101

techniques of processing and analysis are required, for which computing

is a current technology. Biomedical Computing (BMC) is therefore needed
in Medical Electronics, Medical Physics and Biomechanics and embraces the
overlap region between them of BME to become Biomedical Technology (BMT).
Thus the common link between these three groups now is not Engineering,
that is where they differ, but Electronics and Computing. These two disci-
plines form the basis of a relatively new subject of Information Technology
(IT), which is concerned with technological methods for the acquisition,
analysis and effective use of information.

5. OPTIMISATION OF EFFORT
The reward for effort in understanding a new subject usually follows
an exponential type curve, rising steeply at the outset to a general appre-
ciation, then gradually to a more detailed understanding. A level is then
reached where any further advance requires a much greater effort - the
expert region, or the area of professionalism.
In Biomedical Technology, the first stage is usually measurement and
the necessary instrumentation (Fig. ia). An individual may remain in this
subject and become expert but greater effectiveness overall might be more
readily achieved by embracing data processing, which is the rearrangement
of data into a more meaningful form to provide information (Fig. ib).
This subject follows a similar curve and also has its expert region
where one might reside. However, in a mUltidisciplinary subject such as
BMT, it is advisable to become involved with analysis as this will provide
a much better understanding of the biomedical objective (Fig. ic). The
object of analysis is to provide more detailed information about the system
being studied, leading to decision making and consideration of the next
stage of the study.
There is yet another technique, that of modelling (Fig. id). It is often
assumed that modelling cannot be effective until one has a reasonable under-
standing of a system's behaviour. This may be so for producing scaled
models of known situations, but in research, modelling can be valuable
at the outset as a guide to which experiments might provide more useful
information. Thus, an hypothesis, which may well be wrong, can be modelled
exactly and its behaviour compared with that of the biological system being
studied to establish the validity, or otherwise, of that hypothesis. In-
dividuals involved with Biomedical Technology may position themselves
anywhere along these related knowledge curves depending upon the effort
they wish to put into each technological subject.
One's understanding of a system is obtained from facts and the interpre-
tation of those facts in relation to ideas about the system's behaviour
(Fig. 2). These features, both real and imaginary, can be combined into
models of particular features of the system in order to obtain a better
understanding of its behaviour [7].

6. MODELLING
There are various modelling methods that may be used [8,9].
Mathematical models - It is possible to compare a graph of data with
that of a mathematical equation in order to obtain a reasonable match
between them and so be able to describe the system mathematically. With
biological data there are invariably deviations that preclude an exact
match so that the main value of a mathematical representation is to suggest
a possible mechanism of behaviour of the system; for example, an oscil-
latory system or a series of exponentials.
 102

'0
....
CIS
~
Q)
a:

Expert Region
(Professional)

Fig.l :
, Effort
Reward for effort in applying technology to the study of biomedical systems
System

Understanding
Routine

Ideas About I Behaviour

Interpretation
Procedure

Applied Research
t
Basic Research

Facts Obtained from Data

Fig.2: Ratio of ideas: facts in the understanding of a system

 103

As the purpose of modelling is to obtain a better understanding of

behaviour, it is not a good idea to model in a medium that is not understood
by the person who is to use the model. Thus for biologists a biological
derivation of the equations is preferable and more informative to the
modeller who should also be innovative in suggesting mechanisms that could
account for mismatch between model and system behaviour [10].
Descriptive Models - Image or graphic data can be modelled directly
and manipulated to obtain a better appreciation of a system. This type
of modelling was used for building up three-dimensional structures from
data to obtain a better understanding of their function. This was used
to build up molecular models from X-ray diffraction data of atomic position.
A model could then be rotated into any orientation as could components
of the model to determine the minimum energy configuration [11]. Microscope
images of thin serial sections could also be entered into the computer
and reconstructed into a three-dimensional model of the complete structure.
These models had the virtue that defined categories of sub-structures could
be omitted from the main structure or displayed separately so that calculat-
ions could be performed on any selected component [12,13,14].
Deductive models - Alternatively ideas can be modelled and related to
available data. For example, electron-micrograph images of the adeno-virus
hexon provided two-dimensional projections of the unknown 3-D structure
but the orientation of each was also unknown so that reconstruction
techniques could not be used. Instead, a conceptual three-dimensional
model of the hexon was built up in the computer from sections and man-
ipulated into known orientations for comparison of the model projections
with the electron-micrograph data. The model was then successively changed
to obtain a closer match with defined categories of the image orientations.
The final model obtained satisfied all the available criteria but it is
a theoretical model and as yet cannot be supported by experimental proof.
Nevertheless, the model may be used as representative of the true shape
until it is superceded [15,16].

7. EXPERT SYSTEMS
The development of these models all required user intervention with
the computation [17]. At the deductive level and for application, the
models were used to enhance the thought process of the users by incorporat-
ing their data and ideas of a system's behaviour for comparison with further
experimental data and ideas [18]. This meant building the researcher into
the computing, figuratively speaking of course, for ideas of behaviour,
decision making in the selection of data, questionning the response to
intervention and making appropriate changes to the computation. This form
of interactive computing had in effect a human expert system. This worked
very well at the research level but once a particular system was understood
and could be used routinely by others, the human expert wished to be
released. This meant replacing the human expertise by a detailed algorithm
for the particular situation but this unfortunately was no simple task;
the recognition of a biological feature was apparently not definitive but
subjective. Thus it would seem that general purpose expert systems should
allow for user interaction to supplement the expert system. For example,
automated chromosome analysis could identify doubtfuls and allow the user
to arbitrate.

8. EDUCATION FOR BIOMEDICAL TECHNOLOGY

Biomedical Technology is concerned with the application of technology
to medicine and biology, and between these disparate disciplines there
104

is a knowledge gap. There is also a knowledge gap within the technological

disciplines and the biomedical disciplines. Obviously, a high knowledge
of one's subject is desirable but in this multidisciplinary subject of
BMT, this knowledge has to be supported by communication, both in transmiss-
ion and reception. Thus both specialisation and collaboration are expected.
From the technological standpoint, the effectiveness of the technology
in BMT has to be assessed in terms of its biomedical value. Thus some
understanding of the area of application is also necessary but time spent
on acquiring biomedical knowledge could be at the expense of not advancing
technical knowledge.
The assessment of technologists is at present biased towards expertise
in their chosen profession which results in specialisation, i.e. increased
height of the knowledge/subject curve at the expense of bandwidth. An
alternative method of assessment for collaborators, as distinct from
specialists, would be to measure the area under this curve. We need to
encourage education in Biomedical Technology for biomedical effectiveness
and should look carefully at the relative merits of specialists/
collaborators for the tasks to be undertaken.

REFERENCES

I.Perkins WJ: Biomedical Engineering. I.E.E. Reviews, Vol.118, Sept. 1971.

2.Perkins WJ: Problems of Applying Technology in a Non-technological
Environment. J. Aust. Inst. Mech. Engs., No.44, 3-7, May 1972.
3.Perkins WJ: Biomedical Engineering. Proc. 2nd Nordic Meeting on Med.
and BioI. Engineering, Publ. Norweigian Soc. MBE, Oslow, June 1972.
4.Perkins WJ: Medical Electronics. J. of I.E.R.E., Vol.45, No.10, 617-622,
Oct. 1975.
5.Perkins WJ: Medical Electronics - BME - ? Proc. 3rd Mediterranean Conf.
on BME, Portoroz, Aug. 1983.
6.Perkins WJ: Biomedical Engineering - A reappraisal. JMBE, Vol.6, No.7,
157-159, 1984.
7.Perkins WJ: Transforming Data and Ideas Into Information. In: Computing
in Medicine. Edited by J. Paul et al., Publ. MacMillan, 1983.
8.Perkins WJ: Modelling in Biomedicar-Research. Japanese J. Med.
Electronics and Biological Engineering, Vol.15, No.4, 1-5, 1977.
9.Perkins WJ: Introduction to Modelling. In: Biomedical Computing.
Edited by W.J. Perkins, Publ. Pitmans Medical, U.K., pp.238-245, 1977.
10.Perkins WJ, Hammond BJ: The Mathematical Study of Biological Systems.
Bio-Medical Engineering, Vol.2, No.3, Mar. 1968.
11.Perkins WJ, Piper EA, Thornton J: Computer Techniques for Conformational
Studies of Biological Molecules. Computers in Biology and Medicine,
Vol.6, pp.23-31, 1976.
12.Perkins WJ, Green RJ: Three Dimensional Reconstruction of Biological
Sections. J. Biomed. Eng., Vol.4, No.1, 37-43, 1982.
13.Shepherd AM, Perkins WJ, Green RJ, Clark SA: Cell Complement of the
Oesophagus of Aphelenchoides Blastophthorus, as Reconstructed from
Serial Thin Sections Using Computer Graphics. J. Zoology London, 204,
271-288, 1984.
14.Jordan MM, Perkins WJ, Young, MR, D'Arcy Hart, P: Computer Graphics
and Numerical Techniques in the Measurement of Movements Within Cells.
J. MBEC, Vol.23, No.1, 48-4, 1985.
15.Perkins WJ, Barrett AN, Green RJ, Reynolds D: A System for the Three
Dimensional Construction, Manipulation and Display of Microbiological
Models. J. Biomed. Eng., Vol.l, No.1, 22-33, 1979.
 105

16.Nermut MV, Perkins WJ: Assessment of the Three Dimensional Structure

of the Adeno Virus Hexon. Micron, Vol.19, No.3, 247-266, 1978.
17.Perkins WJ: Interactive Computer Models of Biological Systems. In:
Computing in Biological Science. Edited by M.E. Geisow and A.N. Barrett,
Publ. Elsevier, pp.3-21, 1983.
18.Perkins WJ, Hammond BJ: Computer Aided Thought. Nature, Vol.256,
No.5514, 171-175, 1975.
METHODOLOGY IN BIOMECHANICAL RESEARCH
 109

THE ACCURACY OF CRITERIA FOR AUTOMATIC 3-D GRAPHICS

RECONSTRUCTION OF BONE FROM COMPUTER TOMOGRAPHY
P. Rothuizen*. l. van Erning** and R. Huiskes*
lab. Experi.ental Orthopaedics* and Department of Diagnostic
Radiology**. University of Nijmegen. Nijmegen. The Netherlands
1. IftBODOC'fiOti

The purpose of this study was to develop an automatic contour detection

procedure to reconstruct the three-dimensional geometry of bones with Com-
puter Tomography for automatic mesh generation in finite element analysis
and for purposes of artificial joint design.
The application of Computer Tomography, for the quantification of bone
geometry, bone density and morphology of bony structures is meeting in-
creasing popularity in Orthopaedic Biomechanics (1-5). The method is
attractive not only as a result of its nondestructive nature, but also be-
cause the information obtained is directly available in numerical form as
a cross-sectional matrix of relative X-ray attenuation coefficients or
CT-values. Hence, this data is directly applicable in principle for auto-
matic 3-D graphics computer reconstructions.
The first objective is to determine the inner and outer contours of the
cortex per cross-section. In a CT-image these contours are visualized
rather than quantifiedJ due to limitations in resolution they do not
appear as distinct discontinuities in the CT-values. Hence, to estimate
their actual positions, some sort of contour criterion is needed.
Usually a specific scanner dependent window in the CT-values is set and
the visible bone contours are digitized (1,5). In other cases a scanner
dependent CT-value is combined with an automatic contour detection algo-
rithm (3).
A common problem in contour detection in a proximal femoral scan is
caused by non homogeneous density distributions over the cortical wall
(Fig.l.). In this case a specific CT-value as contour criterion can lead
to the construction of a non-closed contour. It is obvious that the
actual cortical contours do not coincide with any specific CT-value in
this case.

FIGURE 1 Typical example of density distributions in a proximal femoral

cross-section in a Computer Tomography image matrix.
110

As a result of these effects it is difficult to accurately reconstruct

the bone geometry without any kind of visual interference of the
operator. To overcome this problem and to make accurate automatic contour
detection possible, an algorithm based on a relative contour detection
procedure was developed, using individual X-ray attenuation profiles.
The subject of this paper is the contour detection procedure and its
accuracy. To investigate the accuracy two comparative studies were done,
one using bone phantoms with known dimensions and one using a proximal
femur.
All objects were scanned in a Siemens Somatom DR3 CAT-scanner u'sing a
slice thickness of 2 mm, an image matrix of 512 * 512 pixels and a 520 mAs
product at a tube voltage of 125 kV.

2. ':I.'HB AO'l"OIIATIC CON'l"OUR DETECTION PROCJmURE

The automatic contour detection procedure developed can be subdivided

in three successive stages:
1. Construction of a base contour.
2. Construction of X-ray attenuation profiles.
3. Evaluation of the individual profiles.

In the first step a base contour is constructed, in such a way that the
bone is with absolute certainty within the area it encloses. As a crite-
rion for this base contour a specific CT-value is chosen. This CT-value
must be higher than the CT-values of the surrounding material but lower
than the CT-values of cortical bone. Detection of this base contour is
performed by a search and contour following algorithm based on the eva-
luation of the eight neighboring pixels of each sequential matrix pixel.
The contour found in this way is tested for appropriateness by calculating
its length L and enclosed area A. If L < 30 It , L > 70 It or L2/A > 101t ,
the contour is rejected and the base CT-value is adapted until a satis-
factory contour is obtained.
In the second step, lines are constructed which intersect the actual
outer and inner cortical contours. In diaphyseal cross-sections, points
of the base contour are connected with its geometrical center. In meta-'
physeal contours, proximal of the trochanter minor, lines are constructed
perpendicular to this base contour. In this way profile lines are

FIGURE 2.a: Profile line over a cortical wall.

 III

constructed at intervals of approximately 1 mm, with a maximum of 1.5 mm

(Fig.2.a). On each line, the attenuation profile is constructed from the
attenuation matrix (Fig.2.b).
In the third step the attenuation profiles on each individual line are
evaluated to estimate the unknown contour points. For this purpose two
criteria were used (Fig.2.b,c). One is an absolute CT-value, C, and the
other a relative CT-value, c, respectively, to estimate the actual abso-
lute CT-value, Cb, and the actual relative CT-value, cb, of the unknown
actual contour points.

CT -profilej (j=1.100 J

H.U. H.U.

(o,max -------

Absolute (= ((bi,j )

Relative: (= ~(bij - (i min

C-I,max- C
I,mln
)( 100%~
(base
(o,min O!;----'-7!;----~----~....lL~

FIGURE 2.b: Typical individual relative X-ray attenuation profile in a

diaphyseal cross-section of the proximal femur.
FIGURE 2.c: Criteria used to estimate actual cortical contour point posi-
tions.

The relative criteria are percentages of the total attenuation distan-

ces between the maximum and the minimum values on the periosteal and endo-
steal sides, respectively.
In the case that an absolute value is used, the corresponding coor-
dinate is the estimate of the contour point. If the absolute value is
higher than the maximum value on the profile, the estimated contour point
defaults to the coordinate of this maximum.
When the estimates of the contour points on each profile are detected,
outer and inner cortical contours are reconstructed by linear interpola-
tion.

3. PHANTOM STUDY

The purpose of this study was twofold. First to investigate the influ-
ences of varying cortical wall thicknesses, second to evaluate the influ-
ence of trabecular bone density variations, on the accuracy and consisten-
cy of the criteria values.
Mater ials used were three PVC tubes with wall thicknesses 2, 5 and
9 mm. The 2 mm tube was subsequently filled with four different concen-
trations of calciumchloride solutions, 0,5,10 and 15%. All phantoms
were scanned in an air environment. In each case the average tube dia-
meter was calculated from the contour estimates, for variable absolute and
relative contour criteria values C and c. Figure 3.a shows the reSUlting
tube diameters, as functions of wall thickness, when the profiles were
evaluated with the criteria C = 600 H.U. and c = 45% (which is approxi-
mately optimal for the 9 mm tube), and with c = 60%.
The graphs show that the relative criterion is more sensitive than the
112

5 10 ...
---7:::=..:--:::----- ~WQUthickness
/"
-s
JI -51 -- - - - absolute (.T.value =600H.U.

- - - relative percentage =45%

~

~~c
l:L ~ -10
---absolute CTvalue = 900HU

"'------'-_ _---relatiYe~= 40%

FIGURE 3.a: Accuracy of criteria chosen with varying wall thicknesses.

3.b: Accuracy of criteria chosen with varying concentrations
CaC12-H20 variations.

absolute criterion for small wall thicknesses, underestimating the actual

contour diameter with 0.4 mm in the case of the 2 mm phantom. This in-
accuracy, however, can be corrected for by using a relative criterion with
c = 60%.
In Figure 3.b the effect of varying CaCl2 concentrations on the accura-
cy is illustrated by showing the resulting diameter deviations from the
actual diameter, using C = 900 H.U. and c = 40% as contour criteria. The
absolute criterion can lead to errors in the contour diameter of up to
0.1 mm; the errors resulting from the use of the relative criterion are
much smaller.

4. BORE STUDY

A proximal femur was fixed in a reference cage and scanned in poly-

esther, water and air environments. After scanning, the femur was sec-
tioned at the corresponding locations. The sections were digitized and
the inner and outer cortical contours compared to those obtained by the
automatic contour detection procedure.
For every section separately, optimal values for the absolute and rela-
tive criteria were determined, both for the inner and the outer contours,
whereby the digitized sections were considered as the true representa-
tions.
Two questions are now addressed; first, what are the optimal absolute
(C, H.U.) and relative (c, %) criteria values, depending on the location
in and the environment of the bone. And second, what would be the errors
in contour detections if specific, consistent absolute or relative
criteria values would be used in all cases.
In Figure 4 the results of the femur scanned in polyesther are summa-
rized, in Figure 4.a relative to diaphyseal cross-sections and in Figure
4.b relative to metaphyseal cross-sections.
In the upper pair of graphs the means and variations of the optimal
criteria values found are shown, for inner and outer contours, respective-
ly; on the left for the absolute criterion (C) and on the right for the
relative criterion (c).
For the diaphyseal contours it is evident that the absolute criterion
 113

plAPHYSEAL CONTOURS
METApHYSEAL (ONTOURS

"~'EC"'" t,,~RECAn"
T~ ----D----11111-",.
.." ::::1--------"'_':'_-
" tlllII
" "

MOt:
OPTIMAL CRITERIA

M~._!
,
,
-.~ -.-,
.Ii:l
DIAMETER DEVIATIONS

MER CONTOURS ~ ~ INNER CONTOURS

FIGURE 4.a: Optimal criteria and aiameter deviations in diaphyseal cross-

sections.
4.b: Optimal criteria and diameter deviations in metaphyseal cross-
sections.

has different optimal values for the inner and the outer contours. Con-
versely, the relative optimal cr iter ion is much more consistent and has
the same value for the inner and outer contours. For the metaphyseal con-
tours both criteria are less consistent in their optimal values, and dif-
fer slightly for the outer and inner contours.
The lower pair of graphs show the means and variations in the devia-
tions from the true bone diameter, in the case that specific average
criteria values are used. In the diaphysis C = 750 H.U. and c = 45%, and
in the metaphysis C = 600 H.U. and c = 50%. Using absolute criteria, the
outer diameter deviates up to 3.9 mm in the diaphysis and up to 1.6 mm in
the metaphysis, the inner diameter deviates up to 3.6 mm in the diaphysis
and up to 5.8 mm in the metaphysis.
Using the relative criterion, the outer diameter deviates up to 1.4 mm
in the diaphysis and up to 2.0 mm in the metaphysis. The inner diameter
deviates up to 1.4 mm in the diaphysis and up to 7.5 mm in the meta-
physis. It must be appreciated that these are not standard deviations,
but represent the maximal deviations obtained, which are sometimes except-
ionally high due to the automatic character of the contour-detection
procedure. In visual inspection, these peak deviations would easily be
recognized and corrected.
The bone environment affects the accuracy markedly if an absolute
criterion is used. Using the absolute criterion value for the femur in
water to detect cortical contours of the femur in air leads to an under-
estimate of the mean outer cortical diameter of 0.7 mm. using the optimal
relative criterion value for the femur in water to detect cortical con-
tours of the femur in air leads to an underestimate of the mean outer cor-
tical diameter of 0.2 mm.

5. DISCUSSION AND CONCLUSIONS

The automatic contour detection procedure worked adequately in all

cases, without any interference of the operator. Closed estimates of
actual cortical contours could be processed automatically to reconstruct
3-D geometry of the proximal femur in a finite element mesh and graphical-
ly displayed as a solid model (Fig.5).
114

Comparing the relative with the

absolute criteria, it can be con-
cluded that the relative cr iter ion
gives adequate accuracy of dia-
physeal contour detection and outer
cortical contour detection in the
metaphysis. The absolute criterion
leads to unacceptable errors in the
diaphyseal region, when consistent
absolute CT-values were used for
both the inner and the outer corti-
cal contours.
In the metaphyseal region of the
proximal femur the results for the
inner cortical contour are somewhat
disappointing for both criteria,
although a closed estimate of the
contour was obtained in all cases.
The accuracy of the absolute
cr iter ion was markedly affected by
bone env ironment, reg ion of scan-
ning and the type of contour to be
detected (inner- or outer cortex).
FIGURE 5 Reconstructed 3-D geometry
of the proximal femur.

Hence, the absolute criterion can not be set universally at a specific

value. Conversily, the relative criterion can be set universally at 45%,
in the diaphyseal region of the femur and at 50% to detect outer meta-
physeal cortical contours, at least when using the High Resolution recon-
struction algorithm on our Siemens scanner.

AC~~:
This study was partly sponsored by Orthopaedic Technology BV, The
Netherlands.

1. Garg A, Deland J, Walker PSg Design of intramedullary femoral stems

using computer graphics. Eng. in Med. 2(1985)89-93.
2. McBroom RJ, Hayes we, Edwards WT, Goldberg RP, and White AA: Prediction
of vertebral body compressive fracture using quantitative computer
tomography. J. Bone Jt. Burg. 67-A(1985)1206-1214.
3. Nelson PC, Robertson DO, Walker PS, Granholm JW: A computerized femoral
intramedullary implant design package utilizing computer tomography
data. Proceedings CA.~ '85, Berlin, Springer Verlag. pp.419-420, 1985.
4. Rhodes ML, Azzawi YM, Chu ES, Pang AT, Glenn WV, and Rothmann SLG: A
network solution for structure models and custom prostheses manufac-
turing from CT data. Proceedings CAR '85, Berlin, Springer Verlag,
pp.403-407, 1985.
5. Schneider E, Weber J?, Gasser B, Cordey J, Robotti G, and Perren 8M:
Determination of geometrical and mechanical properties of the distal
femur using Computer Tomography. Biomechanics X, Proceedings ISB,
Sweden (in press), 1985.
 115

COMPARISON OF OPTOELECTRONIC AND FILM BASED KINEMATIC DATA IN

DYNAMIC BIOMECHANICAL EVALUATION OF BACK MUSCLE TENSION
T. Leskinen*. R. Hedberg**. B. Jonsson** and I. Kuorinka*
Institute of Occupational Health*. Helsinki. Finland and
National Board of Occupational Safety and Health**.Umea.Sweden

I.INTRODUCTION
Spinal stress and its major contributor, back muscle tension, can be
evaluated with biomechanical models based on the analysis of forces and
torques acting on the musculoskeletal system of the human body. Postural
effects due to gravity can be described by static biomechanical models
(e.g. 2, 8). However, static models do not take account of the inertial
forces and torques induced by acceleration, which have been found to
increase the peak load on the spine by 30-60 % as compared to statically
calculated loads in moderate speed lifting tasks (6, 3) and by 19 % when
weights were handled on table level (7). These findings suggest that static
models are of limited value when studying dynamic activities.
To analyse spinal stress biomechanically kinematic data of body postures
and movements is needed. This data can be recorded on film (photography and
cinematography), with video technique, and with optoelectronic methods. In
principle all these methods are applicable for both two- and three-
dimensional analysis, the latter facilitated by the use of two simultaneous
recordings from different directions.
The dynamic analysis of spinal stress has been based on kinematic data
obtained photographically using strobed light with 20 Hz (3),
cinematographically with a high speed (80 frames/s) film camera (7), or
optoelectronically with 100 samples/s (5). Attempts to estimate the dynamic
components mathematically based on the total time of movement and
photographs of the initial and final postures have also been made (1).
The film methods are time consuming especially for dynamic biomechanical
purposes because they involve tedious plotting of posture from several
landmarks in numerous pictures. Moreover, the manual digitizing phase
causes measurement errors, which may be very serious when differentiating
the raw data to obtain acceleration. However, the basic apparatus needed is
cheap as compared to the other methods.
Standard video equipment have low vertical resolution and low sampling
speed for dynamic analysis, but high speed high resolution equipment are
available, which also facilitate automatic analysis of the data by a
computer.
Optoelectronic systems allow computerized movement analysis. They all
use landmarks on the body surface, which causes some sources of error: the
optical connection between the landmarks and the recording apparatus
(camera) must always be preserved, the cables of active (irradiating)
markers may impede normal movement, passive (reflective) markers may be
difficult to identify or they are rather heavy thus increasing problems
arising from the movement of the landmarks on the skin.
In the dynamic biomechanical analysis the precision of the accelerations
calculated is essential for the accuracy of the results. Noise in a direct
movement record decre~ses the precision of accelerations. An increasing
116

signal bandwidth and a decreasing sampling frequency are factors that

decrease the precision, i.e. to obtain reliable results the sampling
frequency must be high enough for a specified signal bandwidth and noise
level. Optimization methods have been developed for smoothing the kinematic
data to minimize the noise of the calculated acceleration data (4, 9), but
their use is limited to cases in which the original sampling rate is high
enough.
In this study both static and dynamic biomechanical sagittal plane
models were used to assess the tension of back muscles during lifts. The
postures and movements of the body were recorded opto-electronically and
with an 8 mm film camera. Our aim was to study if the cheap and portable
but laborious film system is accurate enough for dynamic analysis of
lifting.

2. MATERIALS AND METHODS

Five subjects lifted a box with handles weighing 12 kg from the initial
position, the handles 25 cm from the floor, to knuckle height. The lifts
were done first without instructions concerning the lifting technique (free
style), and then with straight legs (back lift) and with flexed knees (leg
lift). Each lift followed by lowering was repeated three times in
succession.
The handles of the box were equipped with strain gauge transducers to
measure the vertical force.
The movements of the body were recorded simultaneously with an opto-
electronic system (Selspot) sampled 158 times/s and a Super-8 film camera
with 24 frames/so The infra-red light emitting landmarks of Selspot were
attached on the knuckle of middle finger, on the elbow, on the shoulder, on
the hip, on the knee, on the ankle, and on the head just in front of the
ear. The elbow, ankle and head landmarks were not utilized in the
biomechanical analysis. The landmarks of Selspot served as landmarks for
the film analysis, too. The optoelectronic data were directly transferred
to a computer disk, while the positions of the landmarks on the film had
first to be digitized.
with the optoelectronic system two cameras and three-dimensional trans-
formation were used. This system served in inhibiting parallax errors due
to different distances of markers from the cameras, since only the sagittal
plane coordinates were used in the analysis. A Super-8 film camera was used
for the cinematographic recording. The camera was placed 4 m from the
subject. A ruler with I m scale was placed at the same distance with the
subject for length reference. The camera was equipped with electric pulse
unit which gave a pulse for each frame which was exposed during the
collection of the optoelectronic data. These pulses were recorded on
magnetic tape together with the force signal to serve as time reference.
The same anthropometric data and biomechanical algorithms, slightly
modified from those of Leskinen et al (5), were used with both types of
movement data for evaluating static and dynamic components of back muscle
tension. However, the optoelectronic data was first smoothed by averaging
eight successive samples. The averaging window was moved so that the final
sampling rate was 24 samples/so
The muscle force estimates obtained from the two movement data were
compared to each other statistically using regression analysis. Moreover,
the differences of the peak forces obtained with the two methods were
tested with the t-test for paired observations.
 117

kN Muscle force

up down

Figure 1. Static and dynamic muscle force calculated from the opto-
electronic kinematic data, leg lift and lowering of subject 1.

3. RESULTS
A typical time course of dynamic and static muscle forces during lifts
is presented in Fig. 1. showing that the static force is highest in the
beginning of the lift, while the dynamic force reaches its peak about
100-200 ms after the start of the movement.
The linear regression of the peak forces obtained by the two methods was
calculated separately for each lift and lowering. The median coefficients
of correlation of static forces were 0.99 in lifting and 0.97 in lm'lcring
and those of the dynamic forces 0.83 and 0.95, respectively.
To compare the peak forces statistically, the second lift and lowering
of each subject and each technique was taken for the statistical analysis.
The results are presented in table 1. The static forces were systematically
lower with the film method than with the optoelectronic method. A similar
systematic difference was found with the dynamic peak forces in lowering,
though the differences were higher than with the static forces. In lifting
the differences between the dynamic forces of the two methods were often
very high, and also the direction of the differences varied so that no
systematic trends could be detected.
The peak vertical acceleration of the hand calculated from the opto-
electronic data was in the average 2.3 ms- 2 and significantly lower
(p<O.Ol) in lowering: 1.4 ms- 2 There were no statistically significant
differences between the lifting techniques.
118

Table l. Static and dynamic peak forces of back muscles of five subjects
computed from optoelectronic (Selspot) and cinematographic (S-8) kinematic
data in lifting and lowering with different techniques (the second of three
successive lifts) . Significance of differences: NS no significance,
* p<0.05, ** p<O.Ol, *** p<O.OOl.
Static force (N) Dynamic force (N)
Selspot S-8 Diff. Se1spot S-8 Diff

Free style
Lifting
mean 3552 3249 302 4706 4477 228
sd 182 297 220 417 658 444
min 3376 2860 121 4070 3793 -214
max 3850 3659 671 5243 5230 959
signif. * NS

Lowering
mean 3565 3293 272 4587 3555 1032
sd 174 254 149 395 304 175
min 3373 2953 141 4212 3167 852
max 3790 3643 446 5166 3889 1277
signif. * ***

Back lift
Lifting
mean 3893 3637 255 4997 5422 -425
sd 143 125 162 341 1285 1194
min 3706 3445 89 4677 3705 -1948
max 4102 3762 473 5110 6635 972
signif. * NS

Lowering
mean 4020 3707 313 5091 3961 1130
sd 225 178 135 373 437 158
min 3823 3474 122 4690 3554 888
max 4409 3916 493 5549 4529 1308
signif. ** ***

Leg lift
Lifting
mean 3286 2887 398 4179 3698 480
sd 245 360 165 688 587 1032
min 3034 2379 205 3381 3232 -1282
max 3592 3293 655 5214 4663 1387
signif. *** NS

Lowering
mean 3284 2843 441 3989 3027 963
sd 295 540 252 578 604 151
min 2904 2111 174 3253 2212 699
max 3640 3405 793 4649 3578 1071
signif. ** ***
 119

4. DISCUSSION
The static forces obtained from the film based kinematic data were
systematically lower than those obtained from the optoelectronic data.
However, the standard deviation of the difference was quite low suggesting
that the film method gives as reliable results as the optoelectronic method
in static calculations. The systematic differences mean that the scaling of
distances has been different in the two methods, the film method giving
smaller distances than the optoelectronic. This may arise from several
factors including small errors in the calibration process of both methods,
linearity errors of the cameras, and systematic errors in digitizing the
films. Random errors in digitizing have no large impact in static;nalysis.
The digitizing errors were affected by the fact that the visibility of the
landmarks of Selspot was probably too poor for film digitizing purposes.
In the dynamic analysis of lowering the box the systematic difference
between the results of the two methods was larger than in static
calculations, because the difference in scaling of distances also makes the
accelerations of the film method smaller than the accelerations of the
optoelectronic method. The variation of the differences still remained in a
low level. But in the dynamic analysis of lifting the differences between
the methods varied very strongly to both directions suggesting that the two
methods work differently from each other.
In lifting the acceleration peaks were sharper and higher and thus the
frequency range of the signal was higher than in lowering. An increase in
the frequency range of the signal increases the amplitude of the noise in
calculated accelerations considerably, according to Lanshammar (4) with the
power 2.5. Based on his work (4), a noise with the standard deviation of 1
mm in position measurements would lead to the standard deviation in
acceleration of at least 0.28 ms- 2 if the frequency range of the signal is
up to 5 Hz and the sampling rate is 24 Hz, figures quite reasonable for the
film analysis of lowering. If the frequency range is increased to 10 Hz
(reasonable for lifting) the standard deviation of acceleration would be at
least 1.6 ms- 2 The correspondlng standard devlatlons for 158 Hz sampllng
rate would be 0.11 ms- 2 and 0.63 ms- 2 , respectlvely. These examples seem to
indicate that the random errors in digitizing the films are too large for
reliable calculation of accelerations in lifts with the frame rate used. A
higher film speed would allow smoothing of the input data which would
improve the accuracy, but it would increase the digitizing time
considerably, too.
In conclusion, the film method can be used reliably for static force
evaluations and even for dynamic analysis of slow movements. Although
laborious, the film analysis is in many cases the method of choice being
cheap and easy to use in field conditions. However, dynamic analysis of
fast movements must be based on data that has been collected with a
sufficiently high sampling rate. If films with normal frame rate are the
only method available, they can be used to evaluate the static components
of forces even in fast movements, and results of dynamic laboratory
experiments (e.g. 6) may then be used for evaluating the effects of the
dynamic components.
120

ACKNOWLEDGMENTS

This study was supported by the Finnish and Swedish Work Environment Funds.

REFERENCES

1. Ayoub, M. M. & EI-Bassoussi, M. M., 1978: Dynamic biomechanical model

for sagittal plane lifting activities. In Safety in manual materials
handling (ed. by C. G. Drury), DHEW (NIOSH) Publication No. 78-185. U.S.
Government Printing Office, Washington D.C., 88-95.
2. Chaffin, D. B., 1969: A computerized biomechanical model - development
of and use in studying gross body actions. J. Biomechanics, 2, 429-441.
3. Freivalds, A., Chaffin, D. B., Garg, A. & Lee, K. S., 1984: A dynamic
biomechanical evaluation of lifting maximum acceptable loads. J.
Biomechanics, 17: 4, 251-262.
4. Lanshammar, H., 1982: On Precision limits for derivatives numerically
calculated from noisy data. J. Biomechanics, 15: 6, 459-470.
5. Leskinen, T. P. J., Stalhammar, H. R., Kuorinka, I. A. A. & Troup,
J. D. G., 1983a: A dynamic biomechanical model for the evaluation of
spinal stress when lifting. Ergonomics, 26: 6, 595-604.
6. Leskinen, T. P. J., Stalhammar, H. R., Kuorinka, I. A. A. & Troup,
J. D. G., 1983b: The effect of inertial factors on spinal stress when
lifting. Eng Med, 12: 2, 87-89.
7. McGill, S. M. & Norman, R. W., 1985: Dynamically and statically
determined low back moments during lifting. J. Biomechanics, 18: 12,
877-885.
8. Schultz, A. B. & Andersson, G. B. J., 1981: Analysis of loads on the
lumbar spine. Spine, 6: 1, 76-82.
9. Woltring, H. J., 1985: On optimal smoothing and derivative estimation
from noisy displacement data in Biomechanics. Human Movement Science, 4,
229-245.
 121

INSTANTANEOUS HELICAL AXIS ESTIMATION VIA NATURAL, CROSS-

VALIDATED SPLINES+
H.J. Woltring*, A. de Lange**, J.M.G. Kauer** and R. Huiskes**
Pbilips Medical Systems Division*, Best, The Netherlands and
Lab. Exper. Orthop.**, University of Nijmegen, The Netherlands

1. INTRODUCTION
In studies of biological joint motion, quantification of translations
and rotations by means of a reference point and attitude angles does not
provide a clear insight in the relation between kinematics and joint geo-
metry. Because of its geometric simplicity, a better picture can be
obtained by means of the Instantaneous Helical Axis (IHA) , also known as
the instantaneous screw axis, twist axis, or axis of rotation. At each
moment in time, joint motion is seen as the movement of one body segment
with respect to an adjacent segment (usually distal with respect to proxi-
mal), with a translation component along, and a rotation component about a
directed line in space which is uniquely determined as long as the rota-
tory component does not vanish: see Figure 1. The total amounts of trans-
lation and rotation along the path of motion can be defined as the time
integrals of the instantaneous translation and rotation velocities at the
IHA from a given reference time.
The position and direction of the IHA, and the translation and rotation
velocities at the IHA can be estimated by smoothing and interpolating raw
measurements obtained through stereophotogrammetry or electrogoniometry,
followed by conventional rigid-body calculus. In this way, the high errors
incurred by direct estimation of the Finite Helical Axis (FHA) as an
approximation for the IHA can be avoided: see Woltring et al. (1985).
smoothing and interpolation are possible with optimally regularized,
natural splines of sufficiently high order: this is shown to be equivalent
to classical Butterworth filtering, with data-driven determination of the
smallest bandwidth for which signal loss is negligible, under the assump-
tion of a low-pass signal with additive, white measurement noise.

2. INSTANTANEOUS RIGID-BODY KINEMATICS

Given a body-fixed co-ordinate system EX, and a global co-ordinate sys-
tem EY, the position vectors x in EX and X(t) in EY of some point X on the
moving body can be related as-

x(t) = R(t) .~ + .e(t) (1)

where pIt) is the instantaneous position of the orlgln of EX in EY, and

R(t) the instantaneous attitude matrix of EX in EY. The matrix R(t) is
ortbonoraal, i. e

R(t) 'R(t) = R(t) .R(t)' =I (2)

+ From the Laboratory of Experimental orthopaedics, University of Nijmegen
The Netherlands
122

The IHA can be defined as the locus of all points {X} with minimal
velocity li(t) I ' since the rotation component in i(t) vanishes for all
points on the IHA (the translation and rotation components are perpendi-
cular to each other). Differentiation of (1) and elimination of ~ via re-
substitution of (1) results in

i(t) = R(t). R(t)'. {.l(t) - .e(t)} + ,E(t) (3)

The product R(t) .R(t)' is skev-s~tric (cf. woltring et al., 1985) as

can be shown by differentiating the orthonormality constraint (2). The
axial vector w (t) of this skewed matr ix is known as the instantaneous
rotation velocity vector, with amplitude w (t) and unit direction vector
~(t) ,

w (t) = {!;!!,(t) '~(t)}~, ~(t) = ~(t)/w(t) (4)

and relation (3) can be expressed in vector-product form as

y(t) = .!!!.(t) * {.l(t) - .e(t)} + g(t) (5)

The locus of the IHA follows by minimizing li(t) 12 as a function of

.l(t). Taking partial derivatives results in

Q = w2 (t).I - !;!!,(t) .!;!!,(t) , (6)

which yields a class of solutions since Q is singular. For w(t) f 0, Q has

rank 2, with w (t) as its null vector. Thus, for some point y (t) = s (t)
meeting (6), the IHA can be parametrically represented by the line s(t) +
A.n(t). A useful choice for s(t) is the projection of p(t) onto the IHA,
and this yields the additional condition -

.!!!.(t) '{~(t) - .e(t)} = 0 (7)

Combination of (6) and (7) results in the following explicit relation for
~(t) ,

~(t) = .e(t) + .!!!.(t) * g(t)/w 2 (t) (8)

while the instantaneous translation or shift speed v(t) along the IHA fol-
lows by p(ojecting i(t) onto ~(t) ,

v(t) = ~(t) 'g(t) (9)

Given the position p(t), its velocity P(t), and the rotation velocity
w (t), the IRA is completely determined by its position s(t) and direction
n(t), while the amount of instantaneous motion follows from the trans-
lation velocity v(t) and rotation velocity w(t). However, the IHA becomes
undefined for vanishing w(t). (NB: it is often said that the IHA moves to
infinity for vanishing w(t), but this is strictly incorrect). For reasons
of continuity, problems should be expected if the rotation velocity be-
comes "small", especially if the IHA is to be estimated from noisy posi-
tion or angle data.
The occurrence of derivatives, especially in the denominators of most
IHA parameters renders the IHA estimation problem ill-posed or incorrectly
posed (Morozov, 1984). unless special precautions are taken, small
measurement errors have an inordinately large influence on the unknowns in
such problems.
 123

3. DERIVATIVE ESTIMATION VIA OPTIMAL REGULARIZATION

Many practical problems including derivative estimation from noisy data
are ill-posed in the above sense, and additional constraints must be im-
posed in order to arrive at manageable solutions. A common approach is
based on regularization theory (MorOzov, 1984). Here, a balance is sought
between two conflicting goals: goodness of fit to the noisy data, and
smoothness of the fitted signal in terms of a functional of the signal and
its derivatives. An early example of this approach is the procedure of
Anderssen & Bloomfield (1974) for equidistantly sampled, uniformly weight-
ed data which are processed via an FFT-algorithm (cf. Hatze, 1981).

Spline functions
More recently, the use of optimally regularized spline functions (see
Silverman, 1985, for a review) has been found to offer additional advan-
tages, since they can also accomodate non-equidistantly sampled, non-uni-
formly weighted data, while exhibiting less boundary artefacts in the
required derivatives if the order of the spline is choosen sufficiently
high (Woltring 1985, 1986a).
Spline functions can be defined in terms of regularization and variation
calculus. Given a set of N noisy measurements {Yn} at known times {t n },
positive weight factors {wn} which should be inversely proportional to the
local noise variance of the measurements, and an as yet unspecified, posi-
tive regularization factor a, the function salt) is sought which minimizes
the quadratic criterion function

N tN (m)
Ca L wn{Yn - Sa(tn )}> + aJ ISa (t) 12 dt (10)
n=l t1

from the class of functions {s (t)} which are continuous up to the (m-1) st
derivative, and squared integrable in the m-th derivative on the interval
[t1 ,tN]. One can prove that these functions are completely defined by
the class of piecewise polynomials of order 2m which are continuous up to
and including the (2m-2) nd derivative at selected knot positions in the
measurement interval [t1, tN 1. For the purposes of the present paper,
these knots can be taken equal to the measurement times {t n }; furthermore,
certain boundary constraints must be imposed at the record ends. Depending
on these boundary conditions, one can distinguish periodic splines with
sa (k) (t1) sa (k) (tN) for k=0,2m-2, co.plete splines where the boun-
dary derivatives for k=1,m-1 are included as measured constraints with
prior given weights (Hu & Schumaker, 1986), and natural splines where the
derivatives for k=m,2m-2 are set to zero. If the true data underlying the
noisy measurements deviate strongly from the assumed boundary conditions,
problems may occur throughout the aeasureaent interval, especially from
the lowest derivative with conflicting boundary conditions, and higher.
However, the lower derivatives and the smoothing spline proper may be
well-behaved.
Since the splines are completely defined in terms of piecewise poly-
nomials of order 2m, they are called splines of the order 2m. It appears
that they are continuous at the chosen knots at higher derivatives than
m-1; however, these derivatives may fluctuate rather wildly. It is, in
fact, advisable to select m higher than the highest derivative required.
It appears that the solution of (10) can be expressed in terms of a
linear equation system. Denoting the raw measurements as a vector y, and
their spline-predicted values as a vector Xa,
they are related as

Ya = HaY, ( 11 )
124

where Ha is the so-called influence aatrix (in statistics also known as

the "hat" matrix), Band E are certain design matrices defined by the
measurement times {tn} and the spline order 2m, and W = DIAG{wn} is the
matrix of weight factors; see Woltring (1986a). Furthermore, the spline
model allows to evaluate sJk)(t) for k=O,2m-1 and te:[t1,tN]'
For complete and natural splines, Band E can be constructed as band-
limited matrices, and this renders the estimation problem numerically
efficient. For periodic splines, at least for the equidistantly sampled,
uniformly weighted case, the spline model allows an (approximate) treat-
ment in the frequency domain using the Fast Fourier Transform which
renders the matrices Band E purely diagonal as discussed below.
The question now arises how to select a suitable value for the regulari-
zation parameter a. For the uniformly weighted case, Craven & Wahba (1979)
introduced the Generalized Cross-validation Criterion (GCV) which can be
defined as

N 1
GCV a = - 1: {Yn - Sa(tn )}2 / TRACE 2 -(I - Ha)} (12 )
N n=1 N

In essence, cross-validation is based on fitting splines Sa,n(t) to all

datapoints except for the n-th, calculating the predicted residuals at
each omitted point, and finding the value a = aOCV minimizing the rms
value of these residuals. It appears that this ordinary Cross-validation
Criterion (OCV) is ill-behaved if the influence matrix is (nearly)
diagonal; GCV corresponds to making Ha maximally non-diagonal via a prior
orthonormal transformation in y, Ya' and Ha' Generalization to the weight-
ed case is possible (Silverman~ 1985).
The denominator of (12) can be interpreted as the square of the relative
number of degrees of freedom Q a in the smoothing process, and the
numerator as the average value of the squared residuals. Thus, the GCV-
criterion is seen to strike a balance between the residual variance a~ and
Qa'

N 1
L {Yn -Trace(I - H ) ( 13)
n-1 N a

by m1n1m1z1ng the ratio of these quantities.

The residuals {Yn - sa (t n )} at the optimal value a =aGCV should be
largely caused by the stochastic errors in the measurements {Yn}, rather
than by systematic biases in the smoothing process. Thus, the GCV-function
should be largely stochastic in its minimun, and the number of residual
degrees of freedom at a = aGCV must be sufficiently large to ensure a
stable, meaningful (i.e., smooth!) minimum. For multiple datasets as are
typically encountered in the present rigid-body context, this stability
can be enhanced by simultaneous processing of multiple datasets with iden-
tical influence matrices, possibly with additional weight factors per
trajectory to accomodate different noise levels for identical signal
characteristics (Woltring, 1986a). For example, the depth co-ordinate in
photogrammetric movement reconstruction can be much noisier than the two
other co-ordinates (woltring et al., 1985). The mean value of (12) for all
these datasets will exhibit a larger number of degrees of freedom.
In Woltring (1986a), a general package for natural, smoothing splines
has been presented. With Qa the smoothing redundancy, Trace (Ha) can be
seen as the number Na of effectively estimated spline parameters, by ana-
 125

logy to conventional, parametric least-squares. For natural splines, Na

ranges between N (interpoliation, a =0, with Ha=I) and m (a="', Le., zero
derivative constraint in (10), when salt) is an m-th order polynomial with
m characterizing parameters). Na is a useful quantity in the frequency-
domain context discussed in the next section.

4. BUTTERWORTH-FILTER EQUIVALENCE OF THE SMOOTHING SPLINE

If the data are per iodic and sufficiently over sampled , the class of
functions {s (t)} can be reasonable approximated by means of sinusoidal
basis signals with frequencies well below the (average) Nyquist frequency,
rather than by piecewise polynomials. For equidistantly sampled, uniformly
weighted data, Band E in (11) become cyclical, and W becomes the identity
matrix. Thus, (11) reduced to a convolution, with Discrete Fourier
Transform (OFT),

k = O,N-l ( 14)

Y
with Y(k) and Ya(k) the OFT's of X and a , respectively, and Ha(k) the OFT
of the first row of Ha. Using Parceval's theorem, the criterion function
(10) can be transformed to the discrete frequency domain as

1 N-l 2m
Ca = -
N k=O
L 1{1 - Ha(k)}Y(k) 12 + a't.wk IHa(k) .Y(k) I
2
(15)

where 't is the sampling interval, and wk = (211/N't) (k-~N) the discrete
radial frequency. Bij means of a variational argument, the optimal Ha(k)
minimizing (15) can be derived as

(16 )

This relation can be recognized as the discrete transfer function of two

cascaded, m-th order Butterworth filters with cut-off frequency wa ' with-
out phase and frequency distortion (cf. Oppenheim & Schafer, 1975). The
effective number of spline parameters Na can be evaluated as

N-l
Na = L Ha(k) :::: kmwa'tN if W 't II ( 17a)
a
k=O
with

if m - '" (17b)

For example, kl =1/2, k2 = l/Ja, and k3 = 1/3. A similar relation has been
provided by Craven & Wahba (1979). Empirically, the following relation has
been found to hold through experiments with the spline package of Woltring
(1986a) ,

(18 )

both for equidistantly and for slightly non-equidistantly sampled data.

This suggests that m/2 datapoints are used by the natural spline to acco-
modate boundary effects. For the remaining datapoints, the natural spline
126

behaves approximately as a periodic spline.

The m-th order, double Butterworth filter is often applied in a non-
periodic context. First, the data are processed with two cascaded,
recursive time-domain implementations with the second filter in reverse
time in order to cancel phase distortion. Subsequently, the derivatives
are estimated from finite differences. The equivalence with non-periodic
splines explains why m-th derivatives estimated in this way are very sen-
sitive to the selected initialization conditions of the recursive filters,
and it is advisable to select m higher than the highest derivative being
sought (cf. Woltring, 1985).
For IHA estimation, no higher derivatives than the first are required,
so cubic, natural splines and 2nd-order Butterworth filters are appro-
priate. however, the situation is different if also the 2nd-order IRA (Suh
& Radcliffe, 1978) is required, for assessing the pivoting behaviour of
the IHA: in a finite displacement context, Fischer (1907) and Chao & An
(1982) have advocated the utility of the 3-D instantaneous pivot or
central point about which the IHA changes its direction, and of the
trajectory formed by this point.
Under a frequency domain interpretation, the residual variance 0& can be
viewed as the mean residual power per frequency component stopped by the
filter, and Q~ as the relative stopping bandwidth. If the measurements
exist of a low-pass signal with additive, white noise, the effect is that
GCV tries to find the lowest frequency for which the residual noise is
white; this criterion is slightly more conservative than the optimal cut-
off frequencies defined by some authors in Biomechanics (Jackson, 1979;
Wells & Winter, 1980). Thus, the GCv-spline can be seen as a familiar tool
in a new guise, with the additional advantage of also accomodating inter-
polation and non-equidistantly sampled data. However, large datagaps
should be avoided since they are interpolated by local polynomials of
order 2m or less, and their amplitudes may become inordinately large.

5. IHA ESTIMATION RESULTS FROM A KNOWN MOVEMENT

The spline smoothing procedure discussed above has been applied to a
set of marker co-ordinates collected via R6ntgenstereophotogrammetry from
a known movement (de Lange et al 1986). An isotropic distribution of 8
landmarks (p:::: 7 mm, cf. Woltring et aI, 1985) was moved around a fixed
rotation axis, in 180 steps of about 1 each. The rotation axis was
oriented into the x-direction, and the photogrammetric reconstruction
errors of the landmarks had standard deviations of about 100 ~m in the x-
direction, and of about 40 ~m in the y- and z-directions.
For the sllloothing procedure (Woltring, 1986a), the independent Rtime R
variable was defined as the cumulated, finite helical rotations estimated
from the unsmoothed data by means of the procedures outlined in Woltring
et ale (1985). Subsequently, the raw landmark co-ordinates were processsed
with the spline package, as a function of the effective number of spline
parameters N~. A quintic spline (m=3) was choosen in view of future use of
the 2nd-order IHA, and the data were processed in the "simultaneous" mode,
with uniform weighting over time, and weight factors per co-ordinate in-
versely proportional to the position variances. In a pilot study, the
signal-bearing co-ordinates (y and z) were found to possess very similar
QGCV values, while the x-coordinates contained almost pure noise plus a
constant offset.
The 1 rotation increments were sufficiently small to render FHA's as
calculated from the smoothed data suitable estimators for a discrete set
of IRA'S; therefore, the smoothed data were directly passed through the
FHA procedure, and sample statistics were assessed for both sets of
 127

FHA's. In other situations, the smoothed data should be densified between

the measurement times by interpolation. Note that the small-angle noise
efffect of the FHA (Woltring et al., 1985) presupposes uncorrelated noise
between times: this assumption is no longer met after smoothing and inter-
polating the raw data.
Since the movement was about a fixed but numerically unknown axis, dis-
persion measures for the positions and directions of the IHA's with res-
pect to a Mean Be1ica1 Axis (MBA, cf. Woltring, 1986b) provide an indica-
tion of the total precision of the IRA estimation procedure. In essence,
the MBA follows by finding a mean rotation npivot n of all IHA'S, defined
as the point with the smallest rms distance SOs to the IRA's. Subsequent-
ly, an optimal direction vector is defined through this pivot wich has
minimal rms distance of an arbitrary point on the direction vector to the
IHA's. This is equivalent to minimizing the rms value of the sinuses of
the angles between the MBA and the IRA'S, and the angular dispersion SOn
is defined as the arcsine of this rms value. The difference with a truely
angular rms value is negligible, and the solution follows from an eigen-
value problem which plays also a role in the calculation of the mean
pivot.
In Figure 2, the results of this study are shown. The equivalent Butter-
worth cut-off frequency BWfre exhibits the behaviour of relation (18)
except for the case of maximal smoothing (a-co, i.e ,N a - m=3) where the
data are fitted by parabolas. The GCV-function exhibits at its minimum the
smooth behaviour to be expected for a sufficiently overdetermined
problem. The optimal value is located at Na = 8.5, and the strong increase
in the GCV-function for smaller values of Na follows from the inappro-
priateness of least-squares fitting parabolas to the 180 sinusoidal arcs
subtended by the y- and z-coordinates as a function of time. The white-
noise nature of the measurement errors is borne out by the slowly increas-
ing character of the GCV-function for Na > 8.5, and this corresponds with a
horizontal path of the residual variance function (not plotted).
The position and direction dispersions were found to be 34 ~m and 0.078
at the GCv-minimum. In contrast, their values were 2.75 mm and 12 in the
unsmoothed case, i.e., 81 and 154 times as large, respectively. Thus, the
small-angle noise effect discussed by Woltring et al., (1985) could be
effectively abolished.
It appears that some oversmoothing would be acceptable, since SOs and
SOn attain minimal values at Na =4.5. Apparently, reducing the cut-oif
frequency has a beneficial effect in the derivative domain which largely
determines the error sensitivity of the IRA. This is understandable since
high-frequency noise has a strong effect, while low frequency signal com-
ponents are strongly attenuated during differentiation. Thus, the GCV-
criterion can be said to smooth as little as possible, at least if the
data are sufficiently redundant and i f they meet the spline's model
properties of a stationary, low-pass signal with additive, white noise. In
biological joint motion studies, this may be a reasonable assumption since
the opposite might entail large, potentially dangerous inertial forces.

6. ACKNOWLEDGEMENTS
The authors are indebted to mrs. I. de Bruin and mr. J. de vocht for
digitizing 180 R~ntgen stereophotographs.

7. REFERENCES

1. Anderssen RS, Bloomfield P: Numerische Mathematik 22, 157-182, 1974.

2. Chao EYS and An KN: Biomechanics: principles and Applications (R.
Huiskes et al., Eds). Nijhoff, The Rague (The Netherlands), pp. 1-18,
1982.
128

3. Craven P & Wahba G: Numerische Mathematik 31, 377-403, 1979.

4. Fischer 0: Kinematik Organischer Gelenke. vieweg, Braunschweig
(Germany), 1907.
5. Hatze H: Journal of Biomechanics 14(1), 13-18, 1981.
6. Hu CL & SChumaker LL: Numerische Mathematik 49(1), 1-10, 1981.
7. Jackson KM: IEEE Transactions on Biomedical Engineering BME-26 (2) ,
122-124, 1979.
8. Lange A de, Huiskes R, Woltring HJ & Kauer JMG: Effects of data
smoothing on the reconstruction of kinematic parameters (in prepara-
tion)
9. Morozov VA: Methods for solving incorrectly posed problems. Springer,
New York, 1984.
10. Oppenheim AV & Schafer RW: Digital Signal processing. prentice-Hall,
Englewood Cliffs (NJ, USA) 1975.
11. Silverman BW: Journal of the Royal Statistical Society B 47(1), 1-52,
1985.
12 Suh CH & Radcliffe CW: Kinematics and Mechanisms Design. Wiley, New
York, 1978.
13. Wells RP & Winter DA: proceedings Human Locomotion I. Can. Soc. of
Biomechanics (Ottawa), pp. 92-93, 1980.
14. Woltring HJ: Human Movement Science 4(3), 229-245, 1985.
15. Wbltring HJ: Advances in Engineering Software 8(2), 104-113, 1986a.
16. Woltring HJ: Model and measurement error influences in kinematics. In:
Proc. Study Institute and Conference on Biomechanics of Human Move-
ment. Formia, (Italy) 1 1986b (in print).
17. Wbltring HJ, Lange A de, Huiskes R & veldpaus FE: Journal of Biome-
chanics 18(5), 379-389, 1985.

v
V
V/
V
./
/'
k::: ~
BI.
./ re
nIt)
I?'"
.....-: -;:::-
./
/'
wIt) >
<.) /'
V SD
:.-- V V
.l-?
-------~
,. I-- I-'"
1"'-
C!)

. / SD
,/ /
I--'
" ,, GC
" 0.0 . 0 8.0 12.0 16.0 2D.D
Number of 88'lLmoled parameters Na

FIGURE 1: FIGURE 2:
Instantaneous Helical IRA smoothing results for a quintic, natural
Axis. spline, with position dispersion SDs ' direction
dispersion SDn' GCV-function GCV, and equivalent
Butterworth cut-off frequency BWfre' as a func-
tion of Na SD s ' SDn' and BWfre were normalized
w.r.t. arbitrary scaling values.
 129

ESTIMATION OF GAIT CYCLE DURATION AND STRIDE LENGTH FROM ONE

MARKER KINEMATIC DATA
H. Lanshammar
Institute of Technology. Uppsala University. Uppsala. Sweden

ABSTRACT
Several modern gait analysis techniques, such as Selspot, Vicon etc.,
provide coordinate data for markers affixed to the body. These data are
used for different kinds of kinematic and kinetic locomotion analysis.
However, since most variables of interest are depending on stride length
and stride duration, it is important to include these basic gait
parameters in the analysis. For assessment of the validity of collected
data and of the patients ability to carry out stationary gait, it is also
useful to have a measure of the degree of stationarity in the data.

An often used procedure for measuring these parameters is to use foot

switches and photo cells combined with a timer. The objective of the
present study was to develop an algorithm for estimation of the parameters
directly from the measured coordinate data. As input, the algorithm
requires the horizontal and vertical coordinates in the sagittal plane of
one point affixed to the body. For good results the marker should be
located somewhere on the leg, preferrably on the knee or more distal.
1. INTRODUCTION
It is generally agreed that many important gait variables are depending on
walking speed and the closely connected stride length and stride duration
(Lamoreux, 1971, Andriacchi et aI, 1977). Therefore it is essential to
know these parameters also when the analysis primarely concerns other
types of data. Most often these parameters are determined from measument
of the time interval between two events of the same type, such as for
instance left heel strike. Often foot switches are used to determine the
occurence of these events.
Another important issue for gait analysis is to be able to determine if
the measured gait pattern represents stationary gait. Probably the data
are not representative for the subject if the gait is too nonstationary.
In other cases the degree of stationarity might be a usable measure of the
patients ability to control his locomotor apparatus.

Modern gait laboratories are often equipped with instruments that provides
coordinate data for selected landmarks on the body. Examples of such
instruments are Selspot and Vicon. In the Selspot instrument, markers in
the form of light emitting diodes are affixed to the body. The location of
the markers are measured with a special camera and the horizontal and
vertical coordinates are fed to a computer for storing and analysis.
Rather than combining such coordinate measurements with special
instruments for determination of the basic gait parameters it would be
130

more economic to use the kinematic data for determination of these

parameters.
In Gustafsson, Lanshammar (1977) a method is described for the
determination of the point of time for heel contact and toe off for both
left and right leg. These calculations were based on kinematic data in the
sagittal plane for the heel and 5'th metatarsal joint. From these data the
desired gait parameters could be determined.
In the present study a simpler and more robust algorithm is presented
which is not sensitive to the exact location of the marker, and which
determines cycle duration, stride length and velocity. The algorithm,
which is formulated as a Fortran IV subroutine, also provides an estimate
of the gait stationarity.
2. METHOD
The first problem to be addressed is how to estimate stride duration from
the kinematic data.
Let us assume that we have collected sagittal coordinate data for a marker
located on the leg, for instance at the ankle.
A standard procedure for the assessment of cycle duration for data
containing a cyclic component is to use the autocorrelation function.
However, this is suitable only if the data cover many cycles and unless a
treadmill is used for the experiment, a measurement most often includes
only a few steps.
Another possibility is to use the given data for localization of the
occurence of some critical event, such as peak values of the vertical
coordinate. Such a principle can be used and is analogue to the use of a
foot switch in the sence that the time interval between two succesive
occurrences of the same type of event is determined. A major drawback with
such a principle is that the resulting estimate will be sensitive to
errors in the data, since only a few data items are used in the
estimation.
An alternative method, which is used here, is based on the use of all
coordinate data, providing an estimate that is less sensitive to errors in
the data.

In figure 1 an example of sagittal vertical coordinate data for an ankle

marker is given. The data in figure 1, as well as all other data in this
article, were collected with a Selspot I system at 157 Hz. As can be seen
the data are close to but not completely periodic.

The main idea behind the algorithm to be presented is to compare the time
series, corresponding to vertical coordinates, with a time shifted version
of the same data. For periodic data there will be large differences
between the curves for small time shifts, while a shift equal to one
period will make the two curves coincide exactly. Real data, however, are
always more or less nonstationary, and when the time shift approaches
stride duration the curves will be close to but not exactly overlapping.
In figure 2 the difference between the original and time shifted curves
are illustrated for a small time shift (figure 2 a) and when the time
shift is close to the gait cycle duration (figure 2 b).
 131

y (veri leal eoordlnaiel

FIGURE 1. Example of vertical coordinate data in the sagittal plane versus

time for a marker located at the ankle joint.
y (vertical coordinate) y (vertical coordinate)

i.1ID 1.511

FIGURE 2. Comparison of the vertical coordinate data from Figure 1 with a

time shifted version of the same data. a) A small time shift gives large
differences between the curves. b) The curves are close to identical when
the time shift is equal to the stride duration.

To implement this idea in an automatic algorithm the following criterion

was used.
V(j) = ~ ~(y(i+j) - Y(i2
1
where
l32

y(i), i 1,2, ... ,M are the vertical coordinate data.

N is the number of elements in the sum. Note
that N will decrease when j increases, since
only a limited number of data items are
available.
V(j), 0,1, ... , (M-1) is thl! criteria function used for estimation
of thl! qai t cycle duration.
Expressed in words the above formula means that the mean square value of
the difference between the time shifted and the oriqinal data is computed
for different values of the time shift j. This function is then minimized
with respect to the time shif1:. The function V(j) has a local minimum
when the time shift equals one period in the data, as can be seen in
fiqure 3. V(j) starts out at zero and has its first local minimum when
the time shift equals one peri(~. Note that this minimum is rather narrow,
which makes i t possible to detl!rmine the stride duration with a hiqh time
resolution. The resolution ill a certain case will be dependinq on the
samlinq rate and on the resolution of the measurement device.
To avoid the minimization to (:hoose j = 0 as the optimum, the first few
values of j are not included in the analysis. In the present version of
the alqoritha it is assumed that the stride duration is at least 0.5 s,
and therefore a minimum correllpondinq to a stride duration of less than
0.5 s is automatically excluded.

V(jl
.511Gf-8_-----------------------,

37SE-8

.2S8E-0

.125-0

8.8
8.0 .588 1.00 1.50

FIGURE 3. Criteria function, V, defined as the mean squares difference in

vertical coordinate values between oriqinal and time shifted data, as a
function of the time shift. The stride duration is determined by the
minimum of V as marked in the fiqure.
 133

If j is the obtained number of data in one period then the stride

durat~6~, T, is given by the sampling interval, DT, multiplied by the
number of samples

T = j per tOT
We will now turn to the problem of determining the stride length. This can
be estimated from the horizontal data for the same marker. Typical
horizontal data for an ankle marker is given in figure 4. It is
straightforward to use the already obtained estimate of the stride
duration to obtain the stride length, L, as the average horizontal
displacement during one gait cycle:

L = ~ r(x(i+j ) - xCi~
i per
where
xCi), i=1,2, ... ,M are the horizontal coordinate data.

is the number of data in one period as obtained

by minimization of V.

N is the number of elements in the sum, which is

depending on M and j (N = M - J ).
per per

x (hori2ont~l coordi~te)

1.50

.750

0.0

-.750

-1.50

FIGURE 4. Example of horizontal coordinate data in the sagittal plane

versus time for a marker located at the ankle joint.
134

This means that we do not use the distance between two succesive heel
strikes or any other event, which implies that it is not necessary for the
data to cover two heel strikes. The algorithm is still working if for
instance the measurement starts at early swing phase and ends at late
swing phase during the next stride, just as long as the data cover more
than one gait cycle. By this principle we also obtain an algorithm that is
not depending on a few data items and that is therefore not very sensitive
to measurement errors.

3. DISCUSSION
The given algorithm have been used with good results at the Uppsala
Biomechanics Research Laboratory for about two years, for analysis of
normal as well as pathological gait.

A sensitivity analysis have shown that for typical gait data the error in
the estimated gait cycle duration is less than 0.01 s, and the error in
the stride length is less than 0.01 m.

Some advantages with the proposed algorithm are:

* No extra data are needed.
* Not sensitive to stochastic errors in the data.
* Good resolution in estimated gait cycle duration (Narrow minimum in V).
* A measure of the gait stationarity is provided (Given by the minimum
value of V).

Since the algorithm is only involved with the analysis of the cyclic
behaviour of data from one marker, it does not provide any information
about specific events. Therefore, the point of time for heel strike or toe
off cannot be assessed. To achieve this it would be necessary to use more
than one marker, and their exact position on the foot would have to be
prescribed. Further, since only one marker is used, it is not possible to
assess such parameters as left or right step length, since they are
depending on data from both legs. If such data are desired it is necessary
to use markers on both legs or foot switches under both feet.

REFERENCES
Andriacchi T P, Ogle J A and Galante J 0
Walking speed as a basis for normal and abnormal gait measurements.
J. Biomechanics, Vol 10, pp 261-268, 1977.
Lamoreux L W
Kinematic measurements in the study of human walking.
Bullentin of Prostethics Research, 10-15, Spring 1971.
Gustafsson Land Lanshammar H
ENOCH - An integrated system for measurement and analysis of human gait.
Doctoral dissertation, Department of Technology, Upppsala University,
Uppsala, Sweden, UPTEC 77 23 R, 1977.
 135

SPECKLE INTERFEROMETRY: A TECHNIQUE TO INVESTIGATE DEFORMA-

TIONS IN THE FEMORAL BONE
H.-E. Hoyer and J. Dorheide
Institute of Anatomy. Medizinische Hochschule Hannover.
Hannover. Germany

A new technique for measuring the surface displacement in

bones is described. Speckle interferometry was used to study
the deformation behaviour of embalmed human femoral bones.
The bones were measured under varying loads and other condi-
tions. It was shown that the magnitude and direction of dis-
placement could be ascertained at selected points and the
strain between any two points could be calculated. The
present findings differ from those derived from mathematical
models of the femoral bone published so far. The practicabi-
lities and the limitations of this technique are outlined.

INTRODUCTION
A complete knowledge of the deformation behaviour of a
loaded femoral bone is not only of theoretical interest, but
also of great interest when applying osteosyntheses or endo-
prostheses. Numerous investigations have been published and
various techniques have been used during the last 40 years
with respect to the stress and strain distribution of the
loaded human femur. Strain gauge measurements, a conventional
method, give an accurate analysis of surface deformation but
do not allow the displacement of the whole bone or part of it
to be determined. Rohlmann et al. (1980; 1981) calculated the
displacement of a human femur under various loading conditions
using the finite element method. If the bone is assumed
homogeneous, transversely isotropic and the bone material
itself behaves in a linear, elastic manner, then the finite
element method produces results comparable with experimental
results (Huiskes et al., 1981).
Holographic methods have been used in the investigation
of the femoral bone by Hanser (1978, 1979) and Manley et al.
(1983), in particular the mechanical reaction of a femur with
an endoprosthesis. Hanser (1979) determined the bending
behaviour and calculated the displacement of a femoral bone
with and without an osteosynthesis. However, measurements of
deformation by holographic interferometry are restricted to
displacements over a range of a few micrometres. A method
with a wider range of sensitivity (up to 1 millimetre) has
been developed for measuring engineering structures. This
technique is known as speckle interferometry. Archbold et al.
(1970) developed this method further, making use of the effect
that a surface appears speckled when illuminated with coherent
light (laser light). If the object is moved or deformed, the
136

speckle pattern will move too. The change or movement in the

speckle pattern can be recorded on a photographic plate by
double-exposure, whereby the object is photographed before and
after deformation. The processed plate is then illuminated by
an undiverged laser beam and linear fringes can be observed
within the diffraction halo. Thus, the in-plane displacements
can be measured point by pOint.
The objectives of the present study were a) to explore
the practicabilities of speckle interferometry in a biological
context and b) to determine the real deformation of a long
bone under various loading conditions.

MATERIAL AND METHODS

Human femoral bones were obtained from preserved cadavers
which were 60 - 80 years old. The tendons and muscles were
removed completely. The bone samples showed no signs of bone
disease. The distal ends of some bones were embedded in a
polyester resin and mounted on a metal plate so that the angle
between the bone shaft and the vertical axis was 0 0 or 120.
Other bones were not embedded, but standing free on both con-
dyles. We constructed an apparatus which allowed the head of
the bone to be loaded with a vertical force only or with an
additional abductor muscle force, which was simulated by a
metal bar inserted within the trabecular bone at the greater
trochanter. The loading force was produced by a pneumatic
piston and its magnitude was determined and adjusted by a load
cell.
Photographic set-up and procedure. The surface of the
embalmed bone was coated with reflective paint (Scotchlite, 3M
Deutschland GmbH). A 10 x 13 plate-camera (Linhof) with a
3.5/135 lens (Zeiss Planar) and film plates (Agfaortho 25)
were used for photography. The bone was lit with a normal
flash light. (Preliminary studies had revealed that the object
surface could be illuminated by a normal flash light instead
of a laser, resulting in extremely reduced recording times).
The photographic set-up is shown in Fig. 1.
In order to measure in-plane displacement a double
exposure was used, before and after loading the bone. Between
the first and the second exposure a plate mounted on a micro-
metre was displaced for the purpose of calibration. A 5mW He-
Ne-laser (Spectra Physics) was used to interpret the double-
exposed plate (specklegram). These interference lines run
perpendicular to a direction given by a vector s which repre-
sents the surface movement in its own plane. The distance s,
through which the surface of the bbject is displaced, shows an
inverse proportion with speckle width d and obeys the equation
where ~ is the wavelength of the helium laser
s = L A ,light (632.8 nm), L represents the distance
d M
between the specklegram (plate holder) and the
recorded image on a plate and M stands for the enlargement of
the image (Ek, 1977) (see Fig. 2).
The interference fringes were measured with a digitizer
(HP9874A) and the vector calculated by a computer (HP9845B).
The computer was programmed to give a graphical illustration,
i.e. the displacement of points on the bone surface was plot-
ted into its contour. The strain between two points was
 137

calculated from the difference in the length before and after

deformation per unit length.

RESULTS
In Figures 3- 5 the results of three different types of
loading are illustrated. Figures 3a and b: The angle between
the displacement's direction and the direction of the loading
force increases from 235 0 at the head to 265 0 at the femoral
shaft. The magnitude of displacement decreases from 330 ~m at
the femoral head to about 100 ~m at the shaft. The bone's
surface only shows compressive strains except for one very low
tensile strain at the shaft (see Fig. 3b). However, it should
be noted that no strain was calculated at the femoral head
itself. Figures 4a and b: The whole bone shifts laterally.
The angle between the direction of the resultant force and the
displacement's direction is about 1300 , more distally it
decreases to about 110 0 There is no significant difference
in the magnitude of deformation from proximal to distal. The
strain calculations reveal compressive strain at the head,
tensile strain beneath the greater trochanter and more distal-
ly compressive strain again. Figures 5a and b: A loaded bone
was not embedded or clamped, but stood on both condyles. The
whole bone shifts to the lateral side with a sllght increase
in the displacement's magnitude from the proximal to the
distal parts. Tensile strain predominates at the head/neck
region and at the intertrochanteric region, while high com-
pressive strain occurs at the upper shaft.

DISCUSSION
Speckle interferometry has been developed for accurately
measuring surface deformations. This technique offers a
number of advantages. Since measurements are taken on the
object itself, simplifications, necessary for example in a
model, are avoided. It is a non-contact technique, e.g. no
gauges are mounted on the surface of the bone, thus the
results cannot be falsified due to fixation of the gauges.
Speckle interferometry allows displacements over a range of
0.03 millimetre to approximately 1 millimetre to be measured.
Furthermore, every pOint on the bone surface can be investi-
gated, theoretically. Thus, the displacement is measured at a
selected point only.
It should be noted that there are some disadvantages with
this technique. Motions of the object itself, i.e. twisting
around the shaft axis, might influence the speckle pattern.
The accuracy of this measurement technique is limited by the
interpretation of the speckles. Although the surface of the
bones was painted with reflective colour, some speckles could
not be digitized due to poor identification of interference
line borders. In a test speckles were digitized ten times each,
to give an estimate of the possible error. The mean error in
measuring the distance between the lines was about 2.3% for a
narrow speckle width and about 4.3% for a wide speckle width.
In measuring the angle of the interference lines (direction of
displacement) an error of 5.7% was calculated. Measurinq the
distances between the speckle lines on the micrometre sphere
(calibration sphere) reveals a good agreement with the real
138

displacement which was known (error about 1%).

Although numerous investigations have been carried out to
analyse strain on a femoral bone surface only little informa-
tion has emerged about the real deformation of these bones.
The technique of measuring the surface stress with gauges does
not reveal the deformation of the whole specimen. However,
the finite element method only allows an approach to deforma-
tion behaviour. Rohlmann et al. (1980, 1981) showed a pos-
sible shift (shift of nodal points) of the whole femoral bone
under various load conditions. Their results, however, differ
from those found in the present investigation. The displace-
ment's direction contrasts with the contour shift of the bone
calculated by the finite element method. Since there are no
distinct differences between the embedded and the free loaded
bone, our results suggest that clamping conditions are not of
great significance in deformation behaviour.
Based on the results of the present paper the conclusion
is justified that speckle interferometry is an excellent
method of evaluating the surface deformation of a long bone.
Finite element calculation is an approximate method, which
involves an assumption of geometrical and mechanical data and
simplifications are unavoidable. with the technique of spec-
kle interferometry no assumptions are necessary, because the
object measured is the bone itself. It also allows strain
distribution to be calculated in all directions at the whole
bone's surface, which is impossible with strain gauge measure-
ments. Moieover, the distance between the points may vary
between a few millimeters and more than 20 millimeters.
However, it should be noted that strain cannot always be cal-
culated exactly, due to the small difference in the displace-
ment vector and the possible error in measuring, mentioned
above. A minimum difference of displacement, a greater dis-
tance between two points and measuring the speckle width twice
should all help to reduce errors in measurement.
The present results indicate that the technique of spec-
kle interferometry can also be applied to femoral bones with
prostheses or to other bones (e.g. tibia and pelvic bone). It
is a good method for evaluating deformations in a statically
loaded bone, and fills a gap caused by the limits of other
techniques.

REFERENCES
Archbold E, Burch JM, Ennos AE: Recording of in-plane
surface displacement by double-exposure speckle photography.
Optica acta 17, 883-898, 1970.
Ek L: Focused speckle photography with laser and
white light. Herbstschule '77. "Optische Moire- und Speckle-
methoden in der Mefltechnik". Hannover, September 1977.
Hanser U: Holographische Bestimmung von Verformungen in
der experimentellen Biomechanik. Biomed Techn 23, 186-187,
1978.
Hanser U: Quantitative evaluation of holographic deforma-
tion investigations in experimental orthopedics. Holography
in Medicine and Biology (Bally G, ed)., Springer Verlag,
Berlin-Heidelberg-New York, 27-33, 1979.
 139

Huiskes R, Jansen JD, Slooff TJ:A detailed comparison of

experimental and theoretical stress-analyses of a human femur.
Mechanical Properties of Bone (Cowin SC, ed). AMD-Vol 45,
21 0- 2 3 4, 1 981 .
Manley MT, Stern L, Bowins T, DelliSanti G: A biomechani-
cal study of the proximal femur using full-field holographic
interferometry. Orthop Trans 7, 283, 1983.
Rohlmann A, Bergmann G, K6lbel R: Die Beanspruchung des
Femur. I. Intaktes Femur ohne Berlicksichtigung des Tractus
iliotibiales. Z. Orthop 118, 897-904, 1980.
Rohlmann A, Bergmann G, K6lbel R: Die Beanspruchung des
Femur. II. EinfluB des Tractus ileotibialis. Z Orthop 119,
163-166, 1981.

Fig. 1
Arrangement of
the camera and
flash light for
speckle inter-
ferometry.
1: flash light,
2: plate camera,
3: femur,
s: surface
displacement.

Fig. 2
s=~
d'M
I L 1 Method for inter-
preting speckle
photographs.
1: laser,
2: specklegram,
3: digitizer
plate.
For explanation
of equation see
text.
ex: angular
orientation of
interference
3 lines.
140

Fig. 3
a) Magnitude and
direction of the dis-
placements at the
surface of a femoral
bone with a force F
acting parallel to
the shaft axis. The
FEMUR bone was embedded at
its distal end.
BILD b) Calculated strains
at the surface in
the frontal plane.
Data represent com-
pressive (- -) or
tensile strain (- -)
in ~m/mm.

Fig. 4
a) Magnitude and
direction of the dis-
placements of a bone
loaded with a resul-
tant force R of
about 350 N at an
angle of 220. The
FD1UR bone was embedded at
the condyles.
BILD 32 b) Calculated com-
pressive (- - ) and
tensile strain (- -)
in ~m/mm at the
bone's surface in
the frontal plane.

Fig. 5
..-"-.,
a) Magnitude and
direction of displa-
cements at the sur~
face of an unembeddec
femoral bone loaded
'I", " ", /
(. . I with a resultant
force R of about 200~
FEMUR
I :'j at an angle of 17 0 .
b) Calculated strain
w~m/mm) at the sur-

/ '..O\.oi. face in the frontal

I~. ~I
I "/
I ", ')
"'/

-
plane (compressive
_ _ , tensile strain
-).

t~
 141

SPECKLE INTERFEROMETRY FOR INVESTIGATION OF THE RIGIDITY OF

EXTERNAL FRACTURE FIXATION - A PRELIMINARY STUDY
J.R. Rether*. H.-E. Hoyer** and G. Giebel*
Unfallchirurgische Klinik*. Inst. for Anatomy**. Medizinische
Hochschule Hannover. Hannover. Germany

ABSTRACT
Speck1 e interferometry, a coherent-opti c method for the measurement of
the magnitude and direction of in-plane surface displacement, was used
for analysis of a newly designed unilateral fixator externe. The method
offers several advantages in the biomechanica1 analysis of fracture
fixation devices.

INTRODUCTION
The method of 1 aser speck1 e interferometry, a coherent-opti c method for
the non-contacting measurement of surf ace di sp 1 acement and tilt, has
been proven in technical sciences over several years (Archbold et a1.
1970, Chiang et a1. 1976, Brdicko et a1. 1978). When an optically rough
surface is illuminated by coherent light such as that from a laser there
is diffuse reflection by the different parts of the surface. The
reflected light appears to be speckled, which is the result of multiple
interference. If the object's surface is displaced, the reflected speck-
le will displace too because of the constancy of the relative phases of
the waves at any given point. The only condition is that the displace-
ment must be small compared with the illuminated area. By taking
photographs of the speck 1es contai ned ina para 11 e 1 plane in front of
the object -before and after its deformati on- vi a double exposure, a
series of linear fringes are observed on the plate after processing. The
spacing of these fringes indicate the magnitude of the object's displa-
cement in a direction normal to the camera axis whereas the norma1-to-
the-fringe direction indicates the displacement direction. It is there-
fore possible to obtain a complete measurement of the object's in plane
displacement by scanning in steps across the speck1egram. The range of
di sp 1 acemenl that can be measured 1i es between 30 ym and 1 mm, dependi ng
on the di stance between the object and the recordi ng fil mp 1 ate. The
theoretical basis of speckle interferometry is discussed elsewhere
(Archbold et a1. 1970, Brdicko et a1. 1978). In the field of dental bio-
mechani cs, th is method has a 1ready been used for the measurement of
surface displacement (K1eutghen et al. 1983). In order to evaluate the
benefit of this method for biomechanica1 analysis of external fixation
devices, we investigated the rigidity of external fracture fixation
using cadaver tibiae stabilized by a unilateral ventral fixator frame.

MATERIAL AND METHODS

Materials
Ei ght freshly autopsi ed t i bi a 1 bones from human cadavers were coated
with a refl ect i ng pai nt (Scotch 1ite, 3 M). The di sta 1 end was embedded
in a polyester matrix and fixed on a metal plate at a 30 varus angle.
142

Axial loading was performed by a pneumatic piston and adjusted by a load

cell. For external fracture fixation a diaphyseal fracture or osteotomy
resp. was performed and stabilized by a unilateral frame analogous to
previous investigations (Warmbo1d et a1., 1983) (Fig.1). Each bone was
i nvesti g~ted without and afterwards with the frame ina neutral (5 N),
compresslon (500 N) and bridging (gap: 10 mm) position.

FIGURE 1.
Recording of speck1egrams
A 10 x 13 plate camera with a 3.5/135 lens (Zeiss Planar) was used
together with Agf aortho 25-fil mp 1ates. I nstead of a 1aser beam we used a
normal flash for illuminating in order to minimize exposure time. Doub1e-
exposure photographs of the bone were taken before and after 1oadi ng; we
used a sphere adjustable by a micrometer-screw for calibration.
Analysis of speck1egrams
The doub I e-exposed processed plate was placed ina holder. A 5 mW He-Ne-
laser beam was then directed normal to the plate generating a bending-halo
with multi p1 e interference fri nges on a groundg1 ass screen. The spaci ng
and direction of these fringes was determined by means of a digitizer and
calculated by an HP 9845 B computer.
RESULTS
Intact Tibia
In the anteri or-posteri or (a. -p.) vi ew there is very 1itt 1e di sp 1acement
to the lateral outer side and decreases continous1y from the upper to the
lower part of the shaft (Fig. 2a, Tab.1a).
The lateral view shows a deflection to the posterior side (Fig.2b, Tab.1b).

POINT DISTANCE ANGLE

...
200 N ~~ ______ i_~~ _1 __ l~=~!:== l
124
134
142
-7
-6
-8
MEDIAL LATERAL 196 -13
4
183 -11
180 -6
7 207 -1
8 252 -9
9 248 -5
10 262 -7
11 284 -5
12 292 -5
13 284 -5
14 253 -5
15 181 -6
16 219 -1
17 208 -0
18 199 -6

FIGURE 2a. TABLE 1a.

 143

200 N POINT DISTANCE ANGLE

ANTERIOR
" POSTERIOR ~~
1
2
______ i_~~_l __ i~=~~==l
191
194
-45
-38
3 235 -44
4 228 -42
5 245 -39
6 260 -33
7 247 -35

FIGURE 2b. TABLE lb.

Compression-Osteosynthesis
In the a. -p. Vl ew very I itt 1e change can be seen compared to the intact
tibia (Fig.3a, Tab.2a).
In the 1atera 1 vi ew a very s 1i ght gap in the anteri or part of the fracture
can be observed, indicated by the magnitude and direction of the displace-
ment of the upper fragment compared with the lower one (Fig.3b, Tab.2b).
POINT DISTANCE ANGLE
200 N ~~ ______ l_~~_l __ l~=~~==l
MEDIAL
" LATERAL
1
2
3
4
168
167
142
129
-25
-26
-30
-30
5 154 -46
6 143 -45
7 128 -44
8 133 -51
9 122 -53
10 135 -56
11 157 -51
12 191 -57
13 167 -39
14 166 -35

FIGURE 3a. TABLE 2a.

POINT DISTANCE ANGLE

200 N
No ( .m ) (degree)
1--------383------:12---

"
334 -15
274 -17
ANTERIOR POSTERIOR 238 -21
197 -28
6 136 -44
7 129 -71
8 99 -72
100 -70
10 153 -71
11 169 -47
12 213 -32
13 167 -17
14 229 -9
15 279 -7
16 355 -7
17 423 -8
18 247 +4
19 149 -2

FIGURE 3b. TABLE 2b.

144

Neutralization-Osteosynthesis
Due to the removal of the compression to the fracture gap an inversion of
the displacement occurs in the a.-p. view with the upper fragment moving
to the inner side although there is no appreciable movement in the
fracture gap (Fig.4a, Tab.3a).
The lateral view shows a increased bending to the posterior (Fig.4b,
Tab.3b) .
200 N
POINT DISTANCE ANGLE
MEDIAL '" LATERAL No ( 11m)
1--------142------+59---
2 130
(degree).

+57
3 136 +67
4 156 +49
5 191 +38
6 219 +33
7 245 +26
8 268 +28

FIGURE 4a. TABLE 3a.

POINT DISTANCE ANGLE

ANTERIOR
...
200 N
POSTERIOR
~~ ______ l_~~_l __ l~~g~~~l
611
525
-13
-10
388 -20
4
5
314
221
-27
-50
6 164 -43
" 7 130 -46
8 142 -49
9 162 -41
10 273 -59
11 147 -39
12 255 -17
13 396 -15
14 481 -6
15 448 -8
16 596 -8
17 397 -9

FIGURE 4b.
TABLE 3b.
200 N

MEDIAL
'" LATERAL
POINT
~~

1
DISTANCE
______ i_~~_l __ i~~g~~~l
129
ANGLE

-68
2 132 -53
3 462 -77
4 440 -83
5 432 +87
6 430 +80
7 448 +75
8 454 +69

FIGURE 5a. TABLE 4a.

 145

Diastasis
Onder the given load (200 N) compression of the diastasis occurs within
0.4 mm (Tab. 4a). There is bending displacement to the varus (Fig.5a),
therefore inversion of the displacement angle is observed.
Corresponding displacement occurs at the lateral view (Fig.5b, Tab.4b).

POINT DISTANCE ANGLE

~~ ______ l_~~_l __ l~~~~~~l
ANTERIOR POSTERIOR 1 429 -62
2 449 -69
3 405 -74
4 410 -88
5 417 +79
6 128 +15
" 7 84 +22
8 89 +16
9 109 +18
10 451 +78
11 335 +76
12 318 -88
13 357 -72
14 365 -61
15 4~1 -58
16 304 -39
17 153 -59
FIGURE 5b.
DISCUSSION TABLE 4b.
Severa 1 techni ques have been developed for the measurement of surface
displacements in orthopedic biomechanics.
The disadvantages of the conventional methods, e.g. strain gauges or preci-
sion engineering instruments, are obvious: the need for mechanical influen-
ce to the bone may falsi fy the resu lts. Moreover the magnitude of these
instruments restricts the applicability in many cases.
In recent years the method of finite element analysis has obtained excel-
lent results (Huiskes et al., 1983) in the biomechanical analysis of exter-
nal fixators (Chao et al., 1982).
However, the need for considerable experimental work and computing time in
order to obtain satisfactory results restricts this method to a handful of
research centres.
Speck 1e interferometry was introduced in bi omechani ca 1 research for the
measurement of small in-plane displacements on the human shull (Kleutghen
et al., 1983). This method enables the non-contacting measurement of
displacements of any given surface in a range from 0.03 mm to 1.0 mm.
In order to evaluate its benefit in the field of orthopedic biomechanics
we analyzed a newly des i gned unil atera 1 externa 1 fi xator frame (Warmbo 1d
et al., 1983) stabilizing a tibial shaft fracture.
With a double-exposed photograph -the so-called specklegram- we could show
the displacement patterns of the bone-frame specimens in several clamping
conditions according to the conditions most common in external fracture
fixation.
Wi th compress ion (500 N) to the fracture gap, very 1itt 1e differences can
be observed as compared to the intact tibia. In both conditions we found a
small displacement to the outer side in the a.-p. view compatible with the
given load under a physiologic varus angle. In the lateral view a deflec-
tion to the posterior occurs due to the curvature of the native bone with
the maximum tensile strength at the anterior.
146

Without compression a di spl acement of the upper fragment to the side

opposite to the frame occurs due to the lack of support at this side, but
without greater movement in the fracture gap. Even at a fragment-diastasis
of 10 mm compression occurs within 0.4 mm under 200 N axial load, which is
usually performed by the patient before healing occurs.
These findings are corresponding to previous investigations of this frame
(Warmbold et al. 1983).
Special care was taken to the applicability of speckle interferometry for
biomechanical research:
One should mention, that only in-plane displacements normal to the camera
axi s can be detected under the conditi ons reported here, so a sing 1e
speck1egram of both planes would be necessary if displacement in the
z-axi s occurs. Furthermore, di sp 1acement must be small in re 1at i on to the
ill umi nated surf ace. Compared to the methods ment i oned above, there are
several advantages: Because of the non-contacting measurement there ist no
need for any experi menta 1 mode 1s more ore 1ess simi 1ar to real bone; one
can use a real-live bone-frame-specimen. By recording a single speck1e-
gram, the magnitude and di rect i on of the d i sp 1acement of any gi ven poi nt
lying in-plane can be measured. Overall expense ist moderate, recording is
very time-saving and analysis of speck1egrams is easily achieved by means
of a digitizer.
To summari ze we found that speck 1e photography is a usefu 1 tool for the
measurement of displacement in orthopedic biomechanics.
REFERENCES
Archbold, E.; Burch, I.M.; Ennos, A.E. (1970)
Recording of in-plane surface displacement by double-exposure speckle
photography
Optica Acta ll' 883 - 898
Brdicko, I.; Olson, M.D.; Hazell, C.R. (1978)
Theory for surface displacement and strain measurements by laser specklE
interferometry
Optica Acta 25, 963 - 989
Chao, E.K.S.; An, K.-N. (1982)
Biomechanica1 analysis of external fixation devices for the treatment of
open bone fractures.
Finite elements in biomechanics (Edited by Gallagher, R.H.; Simon, B.R.;
Johnson, P.G.; Gross, J.F.) pp. 195-222; John Wiley &Sons, Chichester
Chiang, F.P.; Juang, R.M. (1976)
Laser speckle interferometry for plate bending problems
Applied Optics ~ 2199-2204
Huiskes, R.; Chao, E.K.S. (1983)
A survey of finite element analysis in orthopedic biomechanics: the first
decade.
J. Biomech. ~ 385-409
K1eutghen, I.; Dermaut, L.R.; Boone, P.M.; De Ca1uwe, M. (1983) Measurement
of displacements introduced by extraoral orthodontic forces on a macerated
skull by means of speckle interferometry: a preliminary report.
Proceedings of the Society of Photo Optical Instrumentation Engineers 369,
538-550
Warmbold, M.; Gotzen, L.; Schlenzka, R. (1983)
Stabilitatsuntersuchungen an einem ventralen K1ammerfixateur der Tibia
Unfa11hei1kunde ~ 182 - 186
 147

AN AUTOMATIC METHOD TO EVALUATE GOODNESS OF MUSCULAR WORK

DURING HUMAN LOCOMOTION
G. Ferrigno*. G.C. Santambrogio** and K. Jaworek*
Centro di Bioingegneria. Politecnico di Milano*. Italy and
Warsaw Technical University**. Warsaw. Poland

1. INTROOOcrION
Human locomotion represents the final result of a very complex
coordination of muscular actions producing work around the joints.
Particularly when such coordination plays a basic role in order to reduce
the amount of energy wasted during movement, the objective evaluation of
muscular work behaviour provides a useful contribution to plan training
procedures or physiokinetic treatments devoted to improve the efficiency
of a given performance in sport and in rehabilitation therapy. The aim of
thi s paper is to propose an automati c method whi ch allows the computati on
of a suitable couple of indexes for the estimation of locomotor efficiency.

2. muIPMENT
Figure 1 shows a general scheme of the instrumentation adopted to carry
out this work. It is essentially formed by a force plate for measurement
of the ground reaction forces and by an automatic movement analyser,
named ELITE (1), for kinematic data acquisition. Both these devices are
connected to a host computer for data storage and processing and for the
representation of the results.
The force plate, a KISTLER ~261A, was mounted flush with the floor along a
pathway where the subjects, which carried small reflective hemispheres
marking suitable anatomical points of the leg, were asked to walk or
run straight. A scheme of the movement analyser is illustrated in Fig.
2. The subJect's images, taken by a low persistence TV camera
equipped with an ULTRICON tube, are converted into digital form (coded
over four bits) and processed in real-time by a dedicated hardware
designed for marker recognition. This latter is based on a frame by frame
two dimensional cross-correlation between the whole TV image and a
reference function associated to the marker shape and size. When suitably
lit up by a source placed as close as possible to the optical axis of the
TV camera, the markers described above, which cast back light within a
narrow angle (15 degrees), appear as white circular spots which are
easy to detect by the recognition algorithm. Shape distortions, which
could cause marker mi ssi ng , are prevented from by freezi ng each frame by
means of a rotating shutter. This latter is placed on the optical path of
the TV camera and is essentially an opaque disk, presenting two narrow
transparent windows, which is driven by a synchronous motor powered by
a supply phased with the TV signal. Once the landmarks have been detected,
the movement analyser computes their coordinates achieving an accuracy of
1/2500 of the field of view. This accuracy, which is about ten times
hi gher than the number of pi xel on the TV camera target, is obtai ned by
computing the centroid of the over threshold cross-correlation values
related to each marker. In fact the single marker matches the reference
function in several pixels, giving a different correlation value depending
 148

TV CA"ERA

LA""

~
SHUTTER

FORCE PlATFOR"

/
FIGURE 1. General scheme of the equj pnent adopted for data measurement.

THRESHOLD
COARELATOR
DETECTOR

AID CONVERTER

TV SIGNAL TO HOST
COMPUTER

FIGURE 2. Scheme of the movement analyser.

 149

on the distance from the real marker center. It has been theoretically and
experirrentally proved that the marker center is well approxhated by the
centroid of the cluster of pixels (1).
The whole process is accompli shed wi thi n 20 ms thus leadi ng the system to
a sampling rate of 50 Hz.

3. METHOD
In order to analyse the moverrent of the leg, six markers were uSed;
they were placed on the anterior superior iliac spine, great trochanter,
lateral joint space of the knee, lateral malleolus, heel and fifth
rretatarsal head. The subJects ran along the 15 meters long pathway, at a
cadence imposed by the sound of a rretronome, and contacted the force plate
at about 2/3rd down the starting line. The TV carrera was placed at about 7
m aside the pathway with the optical axjs orthogonal to the running
direction and covering a field of view 1.8 m wide by 1.4 m high centered
with respect to the force plate. When the subject contacts the force plate
both the analog sjgnals proportional to the ground reaction components
(2), sensed by piezo- electric transducers, and the nurrerical coordinates
of the markers are sent to the host computer which performs the following
functions:
- correction of optjcal and electronic distortions
- cali brat jon of coordjnates
- marker classjfjcation and restoration (1)
- ground reaction vector computation (3)
evaluation of muscular moments at the lower limb joints through the
hypothesis of reactive component predomjnance (4)
- estimatjon of jndexes K and J which will be diSCUSSed below.

Correct jon of distortjons and calibration are performed

contemporaneously on the basis of a reference grid of markers placed,
before the experirrental session, on the progression plane over the force
platform. The grid covers the whole fjeld of view and is used for
computing a set of correct jon and calibration coefficients. The algorithm
used js local and applies djfferent coeffjcjent for each different square
belongjng on the grid, supposing that distortion in a given square might
be consjdered linear jn x and y coordinates.
Once the coordinates have been corrected and calibrated, suitable routines
for the classificatjon allow the assocjation of each couple of coordinates
with its related anatomical landmark; this task is achieVed by modelling
the leg and by tracki ng markers frarre by frarre. The model allows an auto-
start of the tracking procedure without requiring manual intervention of
the operator. Restoratjon of those markers hidden during arm oscillation
is perfOrmed by i nterpolati ng procedures.
From kinematic data and ground react jon vector computatjon, the goodness
of muscular work was estimated by the followjng two indexes:

2
K J (M., .)
1
dt

where t2-tl represents the stance phase duration, Mi is the reactjve rate
of the muscular moment around the j-th lower limb joint and <pi j s the
angular velocjty around the sarre Jojnt.
150

Then the index K represents the reactive work wasted for the IOClvement
by the leg, whi Ie J is the Root Mean Square reacti ve power developed by
the leg joints.

4. RESULTS
In order to show an application of the method described above, the
results concerning two long distance runners are reported.
The anatomical parameters of the two athletes were similar as wel~ as
their mean progression speed computed over the six trials they were asked
to perform (see Table 1). All the data were acqui red in steady state
condition which was verified through the integral of the horizontal
component of the ground reaction force along the direction of progression.
The athlete A is an Olympic participant, while the athlete B is an Olympic
gold medal winner.

TABLE 1. Athletes' anatomical parameters and their mean speed.

ATHLETE WEIGHT THIGH SHANK FOOT SPEED

(kg) (m) (m) (m) (m/s)

A 64 .506 .520 .188 3.90.33

B 63 .489 .507 .151 3.75.2l

Figure 3 refers to the results obtained from one of the trials performed
by the athlete B. It illustrates:

- the stick diagram (Fig. 3 A) by which the angles at the hip, knee and
ankle joints were measured (Fig. 3 C)
- the ground reaction evolution (Fig. 3 B) represented both by functions
of time (graphs Z,y,X,PS,PF) and by vectorial form
- the muscular moments (Fig. 3 D) at the three main lower limb joints.

In Tab. 2 the mean values and related standard deviations of theindexes

J and K are reported together wi th each addend J of the index Although
J.
the values of K do not di ffer si gnifi cantly, thus poi nti ng out that
the two runners waste similar work to support IOClvement during stance, the
index J and its addends computed for the hip, knee and ankle Joints of the
athlete A are about 1.5 times greater than the ones for the athlete B.
Since the reactive R.M.S. power, compressed in the index J, is related
with both concentric and eccentric muscular effort provided by the leg
during the stance, the results obtained confirm the excellence of IOClvement
and coordination in the athlete B according with his international
ranking.

5. CONCLUSIONS
Although in preliminary phase, the results obtained have checked the
validity of the proposed method. It is interesting to note that the
simplification adopted in computing the joint moments would lead to an
approximation larger than the one reported in literature (4) because of
the faster advancing speed. Nevertheless the fraction j hip, where such
 200 Kg z SAGITTAL FRONTAL

o
o 0;5
__~IGHT~Ll
I i
15

40 Kg ~o 0 I 0 0 0 0 ~
40eLS FLOOR
A B -40Kg 0,5 1 5. II
'0"1 .... ,... :"'~I rF I5 ~
-40Kg 0,5 15 ~
0,5
o 15.
130" HIP

400 Nm HIP
_ 10
190" KNEE
I"
-200 Nm
200 Nm KNEE
50 ~
C 200 ANKLE
o
-400Nm
400Nm ANKLE
60 ~
0 0,25 0,5 0,75 5. -200Nm 0,25 0,5 O,i 5 5.
STANCE STANCE
PHASE PHASE
FIGURE 3. Representation of kinematic and dynamic results. Graphs Z, Y and X refer to the vertical, for-aft and
mid-lat components of the ground reaction; PS and PF are the displacements of the ground reaction application
vo
point in the for-aft and mid-lat direction. -
152

TABLE 2. Mean values and standard devjatjons of the jndexes J and K.

ATHLETE A B

j hjp 227.32 9.420 141.19 22.20

j knee 577.37 8.360 338.78 16.00
j ankle 584.35 73.900 468.14 26.30

J 1389.04 73.060 947.86 12.82

K 56.89 14.350 60.64 3.220

approxjrnatjon wejghs more sjgnjfjcantly (durjng walkjng the djfference jn

hj p rocment was about 15% whether or not the gravj ty and j nertj a rocments
were jncluded), plays an only mjnor role jn the camputatjon of the jndex J
thus makjng tolerable the hypothesjs of reactjve predamjnance.
Further refjnements of thjs procedure wHl be devoted to search jnto the
problem of optjrnal power generatjon related wjth sport and rehabjljtatjon.

REFERENCES
1. Ferrjgno G. and Pedottj A., ELITE: a djgjtal dedjcated hardware
system for movement analysjs vja real-tjme TV sjgnal processjng. IEEE
Trans. on Bjamed. Engn., vol. BME - 32, n. 11, pp 943 - 950, 1985.
2. Djvjetj L. and santambrogjo G.C., Changes jn muscle characterjstjcs:
basjc prj ncj pIes and appljcatjons. In: Sensoryrnotor plastjcjty:
theoretjcal, experjmental and cljnjcal aspects. Ed. by S. Ron, R. Schmjd,
M. Jeannerod. INSERM, vol. 140, Parjs, pp. 367 - 384, 1986.
3. Pedottj A., Sjmple equjprnent used jn cljnjcal practjce for evaluatjon
of locamotjon. IEEE Trans. on Bjamed. Engn., vol BME - 24, n. 5, pp 456
461, 1977.
4. Boccardj S., Pedottj A., Rodano R. and Santambrogjo G.C., Evaluatjon
of muscular rocments at the lower Ijrnb jojnts by an on-ljne processjng
of kjnematjc data and ground reactjon . J. of Bjamechanjcs, vol. 14, n. 1,
pp 35 - 45, 1981.

Thjs work has been partjally supported by CNR P.F.T.B.S.

Thanks are also due to Centro dj Teorja dej Sjstemj.

 153

QUANTITATIVE EVALUATION OF THE WALKING PERFORMANCE OF THE

HORSE USING FORCE PLATE DATA ANALYSIS
H.W. Merkens and H.C. Schamhardt
Dept. of General and Large Animal Surgery, Fac. of Veterinary
Medicine, University of Utrecht, Utrecht, The Netherlands

SUMMARY
A method was developed to quantify the ground reaction force pattern of
the horse. A number of selected force amplitudes and peak-time positions in
the normalized stance phase of left and right fore and hind limbs were used
to calculate symmetry indices. Data from each limb were compared with those
of a "standard horse" resulting in limb indices. The combination of
amplitude and peak-time symmetry and limb indices yielded one H(orse)INDEX.
The index values of sound horses exceeded 0.920, while index values lower
than 0.880 usually indicated the presence of some kind of lameness.

1. INTRODUCTION
Veterinarians employ a variety of diagnostic tools to clinically detect
disorders in the locomotor system of the horse. There is presently no
equipment available that exceeds human abilities in the diagnosis of
lameness. However, human judgement of locomotor performance is essentially
subjective, and this prohibits quantitative evaluation of, for example, a
certain treatment. A force plate can be used to quantify the loading pattern
of each of the four limbs objectively. Calculation of the 1eft-to-right
symmetry, or comparison of the loading pattern of each limb with those of a
"standard horse" (Merkens et al., 1986) may provide an objective follow-up.
In the present paper a method is described of quantifying the ground
reaction force pattern of the horse. The procedure for quantification of the
symmetry in limb loading is comparable with that used by Jarrett et a1.
(1980). However, a number of selected parameters of each limb ar;- also
compared with previously determined "standard horse" data, resulting in four
limb index values. These values, combined with the symmetry indices, yields
one H(orse)INDEX.

2. MATERIALS AND METHODS

2.1. Procedure. The procedure and equipment used to determine
representative ground reaction force (GRF) patterns of the horse have been
described (Merkens et a1., 1986). At least five GRF patterns of each of the
four limbs, recorded during one session, were computer averaged, after
normalization to the animal's body mass and duration of the stance phase. A
number of peak amplitudes, times of occurrence of the peaks in the stance
phase (peak-time positions) and force impulses were derived from the
representative tracing.
The amplitude indices (Ia) of each of the four limbs were calculated using
the following procedure. Amplitude factors (fa) were obtained by dividing
the amplitude of the selected peaks and force impulses of the horse under
study (Ha) by the corresponding values of the "standard horse" (Sa):
154

fa=lmin(Sa/Ha, Ha/Sa)1 (1)

The accuracy with which the various parameter values were found in the
"standard horse" data was different. This implies that a deviation from the
"standard" is more likely to occur in the parameters having the larger
relative standard deviations (SDa/Sa). Therefore, the amplitude factors were
"weighted" using weighting factors (wa):
wa=abs(Sa/SDa) (2)
which were computed by dividing the mean values Sa of the "standard horse"
data by the corresponding standard deviations SDa. The amplitude index (Ia)
was then calculated from the weighted amplitude factors using:
n Wi]l/m n
I = [ .11 Ui) , where m= ~ wi (3)
1=1 i=l
In this formula, 11 denotes multiplication of the amplitude factors fi after
exponentiation of each factor to the power wi, n is the number of amplitude
factors used and wi are the weighting factors. Exponentiation to l/m was
carried out to normalize the index to the number of factors and their
weights.
The times in the normalized stance phase at which the selected peaks
occurred (peak-time positions) were used to define the peak-time index (It).
However, the procedure was slightly different, because the absolute value of
the standard deviations in the positions of the selected peaks in the
"standard horse" data were (at least in our material) of similar magnitude.
Therefore, the absolute differences t between the peak-time positions of
the horse under study (Ht) and those of the "standard horse" (St) were used
to compute:
t=IHt-Stl (4)
which was subsequently normalized to the duration of the (normalized) stance
phase, resulting in the peak-time factors (ft):
ft=(100- k*t) / 100 (5)
As appears in the discussion below, the factor k was introduced to reduce
the peak-time factors to magnitudes usually encountered in the amplitude
factors, preventing a possible masking of low amplitude indices by
relatively high peak-time indices when combining them in the limb index. In
our material k=3 was suitable. The peak-time weighting factors (wt) were
defined as:
wt=100/SDt (6)
where 100 is the normalized duration of the stance phase and SDt is the
standard deviation of the peak-time positions in the "standard horse" data.
The peak-time index (It) was then calculated using Eq(3).
The limb index included the amplitude and peak-time indices and was
computed as the square root of their product.
The amplitude symmetry index (symIa) was introduced to quantify the
symmetry between the GRF patterns of the both forelimbs and both hindlimbs.
It was calculated by applying a similar procedure and using the same
parameters. However, the amplitude symmetry factors (symfa) were defined as:
symfa=lmin(L/R, R/L)I (7)
where Land R are the amplitudes of left and right limb peaks of the horse
under study, and the weighting factors (symwa) as:
symwa=(symSa/symSDa) (8)
where symSa is the mean amplitude symmetry quotient of left and right limbs
in the "standard horse" data and symSDa is the corresponding standard
deviation. The amplitude symmetry index (symIa) was then calculated using
Eq(3) .
 155

The peak-time symmetry index (symIt) was calculated from the differences
in the peak-time positions between left and right limbs of the peak-times as
used in the calculation of the peak-time indices of the individual limbs.
The peak-time symmetry factors (symft) were defined as:
symft=(100-k*ltL-tRI)/100 (9)
where tL and tR are the peak-time positions of left and right limbs and 100
is the (normalized) duration of the stance phase. The k-value used was again
3. The weighting factors (symwt) were computed as:
symwt=symSt/symSDt (10)
where symSt and symSDt are the mean and the standard deviations of the peak-
time symmetry quotients in the "standard horse", respectively. The peak-time
symmetry index (symIt) was then calculated using Eq.3.
The limb indices and the amplitude symmetry and peak-time symmetry
indic'es of fore and hindlimbs were combined with equal weight in fore and
hind indices. These two finally resulted in one H(orse)INDEX.
2.2. Horses. The applicability of the "factors" and "indices" described
above was tested using the group of 20 selected horses used to create the
"standard horse" data. Detailed clinical examination had confirmed that
these well-trained riding and driving horses were sound.
Index values of the 16 years old Dutch Warmblood riding horse H311164 are
presented as a typical example of the use of the quantification procedure.
At the first measurements, the walk was easy and extended, although some
limited joint movements could be observed. Nine months later the horse was
admitted to the clinic beacause of lameness of the lower left fore limb. At
clinical examination the horse appeared to be moderately lame at the walk. A
third set of' force plate recordings was made after diagnostic blocking of
the digital palmar nerves. The lameness at the walk was largely reduced, but
remained visible at the trot.

Fz N/kg
8l Fy N/kg
15

5
4
0

-5
2

-15
20 1,0 60 80 100% 0 20 1,0 60 80 100%

FIGURE 1. Force-time curves of the longitudinal horizontal (Fy) and vertical

(Fz) ground reaction forces of the left fore limb of horse H311164 prior to
visible signs of lameness, expressed in N per kilogram body mass and
normalized to the duration of the stance phase. The belt indicate the
"standard" range, found by plotting the mean pattern of the "standard horse"
plus and minus one standard deviation.
156

3. RESULTS
The mean value of the amplitude index, the peak-time index and the limb
index of the 80 limbs of the 20 horses used to create the "standard horse"
data, were 0.9220.023, 0.9390.024 and 0.9250.017, respectively, and
ranged from 0.869 - 0.984. The amplitude symmetry index and peak-time
symmetry index were 0.9510.016 and 0.9590.016 and ranged from 0.921
0.994. The H(orse)INDEX value averaged over this group of clinically sound
horses was 0.9420.006. In general, index .values exceeding 0.920 were found
only in sound horses. Values between 0.920 and 0.880 were considered to be
moderate, while lower index values usually indicated the presence of some
kind of lameness.
The vertical and longitudinal horizontal ground reaction force-time curves
of H311164 are shown in Fig.1. Prior to visible signs of lameness, the
force-time curves of both fore limbs were almost identical, but deviated
markedly from the "standard horse" pattern. This is demonstrated by the the
high amplitude symmetry index (Table 1) and the low amplitude indices of the
individual limbs. The index values of the hind limbs ranged from 0.941
0.974. The symmetry in the limb loading pattern is visualized by plotting
the longitudinal horizontal versus vertical ground reaction forces
(vectordynamogram) of all four limbs in one graph (Fig 2). During left fore
limb lameness a marked asymmetry between the patterns of both fore limbs and
both hind limbs was found. The amplitude index of the affected left fore
limb decreased to 0.830, while the sound right fore limb and the ipsilateral
left hind limb showed compensatory responses, reducing the amplitude
symmetry index to 0.837. After diagnostic blocking of the digital palmar
nerves a slight asymmetry between the patterns of both fore limbs and both
hind limbs persisted, but the limb indices returned almost to their pre-
lameness values.

TABLE 1. Fore limb indices of a Dutch Warmblood horse prior to, during left
fore limb lameness and after diagnostic nerve blocking.

fore limb indices

LFI RFI symIa symIt FI HINDEX
prior to lameness .873 .829 .943 .959 .899 .925
during lameness .830 .893 .837 .909 .866 .901
after diagnostic nerve blocking .888 .883 .912 .928 .903 .922

Abbreviations: L, left; R, right; F, fore; I, index; sym, symmetry; a,

amplitude; t, peak-time; HINDEX, H(orse)INDEX.

4. DISCUSSION
A normally walking horse loads both fore and hind limbs almost
symetrically. Therefore, lameness in one limb will immediately become
manifest as a disturbance in the symmetry of the limb loading patterns.
However, a high symmetry index could be calculated despite the presence of
stiffness in both fore limbs (Table 1). This kind of abnormality in the
locomotor system was detected by the comparison of the peak amplitudes of
each individual limb with those of the "standard horse".
The various peak amplitudes and peak-time positions in the "standard
horse" data were found with different reproducibility. Therefore, weighting
factors were introduced inversely proportional to the relative standard
deviations, so that parameters having larger standard deviations have less
influence on the calculated index values. Clearly, the weighting factors are
 157

Fz N/kg _RF _LF

___ RH . _ . LH

8.0

limb index left forelimb .8734

6.0 limb index right forelimb .8285
amplitude symmetry index forelimbs .9429
peak time symmetry index forelimbs .9590

FORE INDEX .8994

4.0

limb index left hindlimb .9405

limb index right hindlimb .9425
2.0 amplitude symmetry index hindlimbs .9480
peak time symmetry index hindlimbs .9739

HIND INDEX .9511

0 H (orse) INDEX horse 11a 311164 .9249

-1.0 0 1.0
Fy N/kg

Fz N/kg _RF _LF

___ RH . _ . LH
8.0

limb index left forelimb .8295

6.0 limb index right forelimb .8933
amplitude symmetry index forelimbs .8367
peak time symmetry index forelimbs .9090

FORE INDEX .8664

4.0

limb index left hindlimb .9473

limb index right hindlimb .9272
ampli tude symmetry index hindlimbs .9197
2.0 peak time symmetry index hindlimbs .9568

HIND INDEX .9376

H (orse) INDEX horse H 311164 .9013

0 11b
-1.0 0 1.0 during left [ore limb lameness
Fy N/kg

Fz N/kg _RF _LF,

__ .RH . _ . LH

8.0

limb index left forelimb .8877

6.0 limb index right forelimb .8828
amplitude symmetry index forel imbs .91,22
peak time symmetry index forelimbs .9282

FORE INDEX .9025

4.0
limb index left hindlimb .9609
limb index right hindlimb .9288
ampli tude symmetry index hindlimbs .9123
.0 peak time symmetry index hindlimbs .9694

HIND INDEX .9425

H (orse) Ii~DEX horse H 31 t 164 .9223

0 during lameness and diagnostic
-1.0 0 1.0 digital palmar nerve blocking
Fy N/kg

FIGURE 2. Fy-Fz vectordynamograms and corresponding indices of horse H311164

prior to (Fig. 2a), during moderate left fore limb lameness (Fig. 2b) and
after diagnostic digital palmar nerve blocking (Fig. 2c).
158

different for the various parameters and they depend on the actual "standard
horse" data.
The procedure used in the calculation of the peak-time indices differred
from that for the amplitude indices. This was a consequence of the "standard
horse" peak-time data. Where the relative accuracies (SDa/Sa) were of
similar magnitude in the selected peak-amplitudes, the same was true for the
absolute accuracies (SDt) in the peak-time positions (Merkens et al., 1986).
The computed peak-time differences were normalized to the duration of the
stance phase (Eq.5), being already normalized to 100 per cent during the
calculation of the representative tracing, to allow a further analysis of
peak-time factors similar to that used for peak amplitudes.
In the early phase of the development of the HINDEX-procedure, Eq.5 was
used while the factor k was equal to 1. Both in normal and in lame horses
the peak-time factors and indices so calculated were rarely below 0.9, while
amplitude indices as low as 0.6 were encountered during lameness. The
inevitable result of combining peak amplitude and peak-time indices in one
limb index was that the high peak-time indices partially masked the lower
amplitude indices. It was therefore necessary to reduce the peak-time
indices to values comparable with the amplitude indices. An arbitrary though
effective method was to multiply the peak-time differences by a factor k
(Eq.5). An approach by trial and error was necessary to select k so that it
reduced the peak-time factors acceptably. In our material k=3 was a suitable
choice.
The combination of amplitude and peak-time indices of four limbs in one
H(orse)INDEX may result in index values which partially mask the seriousness
of affections in one or more limbs. Although the H(orse)INDEX value of
H311164, calculated during severe left fore limb lameness (0.901; Table I),
is lower than that of sound horses, the difference is not in proportion with
the degree of lameness. This may illustrate that it is absolutely essential
to check all amplitude, peak-time and symmetry indices individually, before
an index-based judgement of the locomotor performance of the horse under
study can be made.
The H(orse)INDEX procedure of quantifying the ground reaction force
pattern of the horse at normal walk proved to be a useful technique for an
initial classification of locomotor performance. The real advantage of this
method is that the calculated factors and indices are direct numerical
measures of the resemblance between the loading patterns of each limb and
the "standard", or between both fore or both hind limbs. The interpretation
is clear: a low index means that big differences are present between the
patterns. An increased index value after a certain treatment implies that
the patterns became more alike, which means "improved" when compared with
the "standard horse. It is applicable for sequential characterization after
interventions, training or treatment. Although the method was developed for
analysis of force plate tracings of the Dutch Warmblood horse, the method
can easily be adapted for quantification of human force plate, or other
kinetic or kinematic, signals, provided that appropriate "standard" data are
available.

References
Jarrett MO, Moore PR and Swanson AJG: Assessment of gait using components of
the ground reaction force vector. Med.Biol.Eng.Comput. 18: 685-688, 1980.
Merkens HW, Schamhardt HC, Hartman W. and Kersjes AW: The ground reaction
force pattern of the Dutch Warmblood horse at normal walk. Equine vet. J.
18: 207-214, 1986.
 159

QUANTIFICATION OF SKIN DISPLACEMENT IN THE WALKING HORSE

P.R. van Weeren and A. Barneveld
Dept. of General and Large Animal Surgery, Fac. of Veterinary
Medicine, University of Utrecht, Utrecht, The Netherlands

Abstract.

Kinematic gait analysis techniques normally rely on skin markers for

determination of the position of certain palpable skeletal structures. As
it is well known that during movement the skin shifts over the underlying
bone, a biological error is inevitably introduced by this procedure. A
method is presented for quantitative determination of skin movement over
the underlying skeletal structures during normal locomotion of the walking
horse. The principle of the method is based upon simultaneous visualization
of the positions of the skin and the underlying bony structures, by
marking the bones with implanted Light-Emitting Diodes (LEDs) and the skin
with self adhesive spot labels. Recordings are made using photography. The
major direction of movement is along the bone axis and a gradual increase
is found from the foot in a proximal direction. It is concluded that skin
displacement in the walking horse is substantial around carpus and tarsus
and is very probably even larger around the more proximally located joints.
Thus for correct interpretation of kinemtic data, correction factors
should be established.

INTRODUCTION

Kinematic gait analysis in the horse has a long-standing history of more

than a hundred years (Marey, 1885, cited by Woltring, 1985; Muybridge,1899).
During a long period stroboscopic photography and cinematography were the
methods of choice but rather recently more advanced opto-electronic
techniques have been developed (Woltring,1985).
All these methods use skin markers (Guth et al,1973; Heinrichs,1974) with
one remarkable exception of direct bone-marking as described by Lafortune
et al (1983).
A consequence of the use of skin markers in kinematic gait analysis is the
introduction of an uncontrollable source of biological error caused by the
shifting of the skin over the underlying bony structures which are in fact
the real subjects of study. Although the importance of skin movement has
been recognized by some authors (Fredricson,1972; Woltring,1985), thus far
no thorough quantitative investigation has been done in the horse. In
human gait analysis some research has been started shortly in which it was
shown that the relative movement, due to the movement of the soft tissue,
was by far the greatest source of noise in the measurements. Errors up to
20 mm were found (Macleod, personal communication). Recently a method has
been developed which allows the determination of the direction and
magnitude of skin displacements in the lower parts of the limbs of the
walking horse (van Weeren and Barneveld, in press). In this paper
160

preliminary results obtained with this method are presented.

MATERIALS AND METHODS

Principle.

The method is based on simultaneous photographic recording of both skin and

underlying bone. The skin is marked in a conventional way with self
adhesive labels, whereas the underlying bone is visualized by implanted
Light-Emitting Diodes (LEDs), the light of which penetrates the skin.

Materials.

Six Dutch Warmblood horses (body weight

approximately 600 kg) were used.
LED
High intensity LEDs (50 mcd) were
chosen as a light source (MK 9350-2,
General Instrument Corp., Palo Alto,
Cal.). The LEDs were wired with multi-
Insulation filament copper wire and mounted in
Stainless
stainless steel tubes (diameter 6mm)
Steel of variable length. The space left
Tube
inside was filled with epoxy glue
(Kombi-Snel(R), Perfecta Chemie BV,
Goes, The Netherlands). (Fig.l).

Surgery.
Wires
The horse was positioned in lateral
recumbency with the medial side of the
limb up and kept under general
fig.1 anaesthesia. LEDs were placed just
proximal and distal of the carpal,
tarsal and metacarpophalangeal or
fetlock joints. The exact location of
the canal to be drilled was determined
Fig.l Schematic drawing of the using small needles and fluoroscopic
LED mounted in a stainless control. A 1.5 cm incision was made
steel tube. through the skin and underlying tissues
onto bone. A canal was drilled through
the bone, including the lateral cortex
using a 6 mm handdrill.
An appropriate length of LED implant was chosen and punched into the canal
until the head of the LED just began to elevate the skin on the lateral
side. The skin incision was sutured in a single interrupted pattern, the
emerging wires being attached to the skin by simple sutures.

Recordings.

Recordings were made on the day following surgery. With the horse in normal
standing position, four circular skin markers (diameter 5 mm) were placed in
a square of 3 x 3 cm, the diagonals intersecting over the head of the LED.
 161

The bony landmarks described by Schmaltz (1906) to determine the bone axes
were marked by self adhesive labels (circular, diameter 25 rom). A metal
strip of exactly 15 em (reference length) was attached to the limb. In two
recording sessions a piezo-electric accelerometer (Bruel & Kjaer 4383) was
attached to the hoof and connected to a charge amplifier (Kistler 5001)
that was attached to the horse's back by means of a belt. (Figure 2).
The LEDs were
connected in series
to a power supply
the current of which
was adjusted to
approximatley 120 rnA
DC and attached to
the belt on the
horse's back. To this
belt a cable ran
containing 220 V AC
power supply and the
connection wires of
the amplifier and a
pen recorder
situated on the
photographer's cart.
The animal was
photographed during
normal walk from
this cart moving at
the same speed as
the horse, keeping a
constant distance of
5 m between film and
object.
I'hotographs were
taken using a Nikon
F3 camera with 105
rom objective and a
6 fps motordrive.
The flash contact of
the camer was, when
nn accelerometer was
used, connected to
the recorder. Three
Fig.2 Fully equipped limb. In this case a LED is to four films of 30
placed only in the distal metacarpus. Note frames were used per
the accelerometer taped to the hoof. recording session.
As average stride
time in the walking
horse is 1.2 sec., with a 6 fps motordrive about 4 strides are recorded per
film, thus 12 to 16 (90 to 120 frames) per horse. Kodak Tri-X film (35 rom
format), upgraded to an I.S.O.value of 25,600 was used. The films were
developed in equal amounts of Kodak HC-IIO developer (concentrate) and
water for 32 min. at 20C.

Data processing.

The films were projected frame by frame on an XY-tablet (Summagraphics

162

Bitpad One) and the coordinates of the relevant landmarks (horizontal,

LEDs, self adhesive labels around the LEDs and those indicating the land-
marks pointed out by Schmaltz) were fed into an Apple lIe microcomputer.
From these data the computer calculated the angles of the relevant joints
and the displacement of the centroid of the four self adhesive labels
surrounding the LED relative to the head of the LED itself. The displace-
ment was calculated in X- and Y-direction (X-direction along the bone axis
and Y-direction perpendicular to the bone axis in the sagittal plane).
The accelerograms written simultaneously with action pulses of the camera,
were used to determine the position in time per frame.

RESULTS

Maximal skin displacement during normal walk is shown in Table I.

FRONT LIMB X Y X Y HIND LIMB

Distal radius 22 11 27 15 Distal tibia

Proximal metacarpus 9 6 18 8 Proximal metatarsus
Distal metacarpus 10 8 16 9 Distal metatarsus
Phalanx I, proximal 5 10 12 13 Phalanx I, proximal

Table I. Maximal measured skin displacements in X- and Y-

directions. Values in mm.

The relationship of skin displacement in one horse at the distal tibia

to the angle of the adjacent joint is shown in Fig.3. For the same site
and horse the relationship of displacement and time, obtained using the
accelerometer technique, is shown in Fig.4.

20 EE 20
E ;::
E zw
;:: 15
z ., ::;; 15 '.
w
::;;
w
.. w
0

0 -"
10 D- 10
'-
.... .. .
o
(/)
-"
D-
(/)
0
. ,. .
:
. .. '.
.~~;,.;r.. .. . (','
:'
o - '.:.:.~.~ " '

:.:.. ir,. . ,:.. :..

,. I , ~"" ,:.
.' : ."
10 "

.~
-5L-____L -_ _ _ _L -____ ____~__~~
-5LO------"2-0----~40------6~0------8~0----~100
~

160 150 140 130 120 110

ANGLECdegrees) TIMEC%)

Fig.3 Skin displacement in the Fig.4 Skin displacement in the

distal tibia along the X- distal tibia along the X-
axis plotted against the axis plotted against stride
tarsal joint angle. time (100% = 1.2 sec.).
 163

DISCUSSION

The technique described permits quantification of skin displacement in the

walking horse. During the recording sessions all animals showed normal
locomotion at walk. Following removal of the LEDs the animals recovered
quickly.

Surgery was performed on the medial side to avoid interference with skin
movement on the lateral side. The method is, of course, limited to those
parts of the limbs that can be approached from the medial side. Care should
be taken that the tubes .fit tightly in the canal drilled in the bone. A
canal with a diameter of 6.1 mm resulted in a backward sliding of the
implant, thus causing a substantial loss of light output. Implanting a 6mm
LED in a slightly narrower canal is not possible without damaging the LED
but also appeared not necessary. When a 6 mm canal was drilled no back-
ward or forward sliding nor rotation of the implant was noted.

LEDs were selected as light sources because they generate minimal heat. It
was necessary to use high intensity LEDs to obtain sufficient light
penetrating the skin. The colour of the LED appeared to be important: the
developing procedure of the film caused loss of red light sensitivity, so
LEDs emitting red light could not be used.
The requirements of an exposure time as short as 1/500 sec. (demanded by
the rather rapid movement of the lower parts of the legs of the horse);
reduction of environmental light for not drowning the the light emitted by
the LEDs and a resolution in the order of 1 mm could not possibly be met by
high speed film or video techniques, so photography was the method of
choice. It was necessary, however, to use film of a very high sensitivity
and a specially adapted developing procedure, using an extremely strong
solution.

As every joint angle, except those of maximal flexion and extension,

occurs twice during a complete stride, a complete picture can only be
obtained when the relation in time between photograph and stride is known.
The accelerometer technique offers an opportunity to couple the photographs
made to the fraction of stride time they represent.

Results show a gradual decline in maximal displacement in the X-direction

from carpus and tarsus distally. The difference in displacement proximal
and distal to carpal and tarsal joints is especially marked. Displacement
in Y-direction is 1~ to 2 times smaller than in X-direction, as far as
carpus and tarsus are concerned, so major movement is directed along the
bone axes. Around the metacarpophalangeal joints, displacements are limited
and are more or less the same in both directions. These displacements seem
to be too small to be of any importance in kinematic gait analysis.
Figures 3 and 4 show the largest part of skin displacement occurring during
the swing phase of the stride. As joint angles change most rapidly during
this phase of the stride, these results may have been expected.

Reproducability in the measured skin displacement and joint angles were 2

mm and 1 respectively and were determined from an analysis of one
photograph repeated 20 times.

Out of plane movement of limb segments may interfere with the

164

twodimensional approach of skin movement. Main causes of out of plane

movement are: cinematography oblique to the direction of movement; out of
plane movement due to the kinematics of the limb segment; and finally
craniocaudal sliding of the skin over the circumference of the bone. The
first-two factors may be neglected because the walking direction of the
horse was perpendicular to the (moving) camera and because the anatomy, of
the equine joints allows anly minimal out of plane movement of limb
segments. Craniocaudal sliding of the skin over the circumference of the
bone may be important in the distal segments of the limb. However,
processing of data obtained by manual manipulation and experimental data
resulted in an error of less than half the standard error of the actual
measurements.

The technique described allows quantification of skin movement over the

underlying skeletal structures during normal locomotion of the horse.
Results presented in this paper show a considerable skin displacement
around the carpal and tarsal joints. The anatomy of the limbs of the horse
suggests an even larger skin displacement around the more proximally
located joints such as elbow, shoulder, knee and hip joints. Analysis and
mathematical processing of the data obtained by the technique described will
result in a model providing correction factors for data obtained by modern
gait analysis techniques.

REFERENCES

1. Fredricson,I.(1972) Equine joint kinematics and co-ordination. Acta Vet.

Scand.Suppl.37,1-136.
2. Guth,V.,Abbink,F. and Heinrichs,W.(1973) Eine Methode zur chronozyklo-
graphischen Bewegungsaufzeichnung mit einem Prozessrechner.lnt.Z.Angew.
Physiol.31.1S1-162.
3. Heinrichs,W.(1974) Eine digitale Zeitmesseinrichtung hoher Auflosung
zur Weiterentwicklung der chronocyclographischen Bewegungsaufnahme mit
Prozessrechnern.Eur.J.Appl.Physiol.32,227-238.
4. Lafortune,M.A.,Cavanagh,P.R.,Kalenak,A.,Skinner,S.M. and Sommer III,H.J.
(1983) The measurement of normal knee joint motion during walking using
intracortical pins (abstract).Biomechanical measurement in Orthopeadic
Practice. The Biological Engineering Society,London.
S. Macleod,A. and Morris,J.R.W.(1986) Investigation of Inherent Experimental
Noise in Kinematic Experiments using Superficial Markers. (in preparation)
6. Muybridge,E(1899) Animals in motion.Republished(19S7),(Edited by Brown,
L.S.) Chapman and Hall, London.
7. Schmaltz(1906) Konstuktion und Grosse der Standwinkel an den Beinen des
Pferdes.Berl.Tierarztl.Wochenschr.14,2S7-261.
8. Weeren,P.R.van and Barneveld,A.(1986) A Technique to Quantify Skin
Displacement in the Walking Horse.J.Biomech.(in press).
9. Woltring,H.J.(1985) Contemporary methods for kinematic motion analysis
in sports biomechanics.Movement Analysis by Computer for the Explosive
Events in Track and Field. (Edited by Swinkels,J.M.M. and Langenbach,D.J.
W.), p17-31. De Vrieseborch, Haarlem.
BIOMECHANICS OF SOFT TISSUE
 167

MEDIAL COLLATERAL LIGAMENT HEALING: A BIOMECHANICAL ASSESSMENT

M.A. Gomez. M. Inoue. E. McGurk-Burleson and S.L-Y. Woo
University of California. San Diego. Division of Orthopaedic
Surgery. La Jolla. California. U.S.A.

INTRODUCTION - Many investigations have shown the medial collateral

ligament (MCL) to playa major role in restricting valgus rotation of the
knee (1-4). The healing of collateral ligament injury is therefore of
clinical importance but the best treatment regimen remains controversia 1.
Different treatments of the MCL include primary surgical repair with sub-
sequent immobilization or early mobilization and nonoperative treatment
with varying degrees of. postoperative motion or physical activity. How-
ever, a major problem that remains with the evaluation of the results of
a particular clinical treatment lies in the qualitative nature of the
orthopaedic examination. Manually conducted tests of ligament function
are subject to both personal technique and interpretation. Therefore,
the objective of this study is to establish an experimental model of the
healing canine MCL following common clinical treatment regimens. The
healing MCL is evaluated as a static stabilizer of the knee in the varus-
valgus direction and both the structural properties of the femur-MCL-
tibia complex and the mechanical properties of the healing MCL substance
are measured.

MATERIALS AND METHODS - Thirty adult mongrel dogs with an average

body weight of 22.1 + 0.7 kg were separated into three groups of ten ani-
mals each. In Group 1, (the non-repaired group), the MCL of the left
knee was exposed through longitudinal incisions in the skin and fascia.
The MCL was completely transected with a scalpel blade at the level of
the tibial plateau and not repaired. Animals were allowed 6 or 12 weeks
of cage activity (cage dimensions: 1 x 1.5 x 2.5 m) and sacrificed. For
Group 2, (the short term immobilization group). the cut MCL of the left
knee was repaired using a modified Kessler technique and the wound was
closed routinely (6). The knee joint was immobilized at approximately
90 0 flexion by passing a 4 mm diameter Steinmann pin sagittally through
the mid-shafts of both the tibia and the femur. After three weeks of
immobilization. the pin was surgically removed and the knee was remobi-
lized for either 3 or 9 weeks of cage activity. In Group 3 (the long
term immobilization group), the MCL was lacerated and repaired similar to
Group 2, but the left knee was subjected to either 6 weeks of immobiliza-
tion or 6 weeks of immobilization plus 6 weeks of remobilization with
cage activity. The contralateral control right knee of all animals was
sham operated to expose the MCL followed by routine closure of the wound.

At sacrifice, the disarticulated hindlimbs of each animal were wrapped in

saline soaked gauze and stored in double plastic bags at -20 o C until bio-
mechanical tests could be performed. To prepare for testing. legs were
first thawed overnight at 4 0 C and then at room temperature until the time
168

of testing. Knees were cleaned of all soft tissues other than the liga-
ments and the menisci. The femur was then shortened to a length of 7 cm.
Distally, the limbs were disarticulated at the ankle joint.

Va~us-Va~_La~ity T~~ - In order to simulate clinical varus-valgus

laxity tests of the knee, a special device was designed based on a
4-point bending test (7) (Fig. 1). To utilize this device, three thread-
ed stainless steel pins 6.5 mm in diameter were inserted sagittally

~ ~ - Schematic of laxity
device which uses principle
~ A A
of 4-point bending to test
knee jOint in varus-valgus
directions. Arrows indi-
cate how rotary and linear
bearings permit pins "A" to
rotate and slide together
as load is applied to pins
"B". Bearing "c" allows
for a constant moment arm
(distance "AC" = 6 cm). A
load cell and two RVDT's
record the load and angu-
lar deformation, respec-
Reversible load cell tively.

through the femur and tibia. A fourth pin was placed coronally through
the proximal end of the femur. This prepared sample was attached to four
rotary bearings encased in support blocks. These rotary bearings allowed
for bending of the joint while linear bearings in the support bloc~s per-
mitted motion in the proximal-distal direction on four longitudinal sup-
port bars. Using a swivel system, equal loads were applied to the proxi-
mal femur and distal tibia with an Instron testing machine to produce a
constant bending moment across the knee joint. The varus-valgus rotation
occurring in the rotary bearing was measured utilizing two rotary vari-
able differential transformers (RVDTs). The rotation vs the applied
momen t wa s p lot ted on an X-Y recorder. For each test, a cyclic bending
moment of + 0.6 N-m was applied. Using the generated bending moment-
angular d;forma tion diagram, the varus-valgus laxity was defined as the
change in angular deformation between: 0.2 N-m,

Tensile Testing - After varus-valgus testing, all soft tissues were dis-
sected from the knee with the exception of the MCL. This resulted in a
femur-MCL-tibia (FMT) complex. The FMT complex was mounted on special
clamps designed to apply the load along the longitudinal axis of the liga-
ment substance during uniaxial testing. The mid-substance of the MCL was
then stained with two, thin parallel black lines 1.5 cm apart using a
Verhoff histological stain. These lines were centered about the level of
the tibial plateau and defined a gauge length by which the surface strain
of the MCL could be measured using a video dimension analyzer system
(5,8). A rectangular cross-section was assumed for the MCL considering
its flattened morphology. The ligament cross-sectional area was then
calculated from width and thickness measurements obtained using the tech-
 169

nique of Woo. et al (S). Tensile tests were performed by first sub-

merging the clamped FMT complex into a 37 0 c saline bath attached to an
Instron testing machine. After IS minutes of temperature equilibration.
a tare load of O.S N was applied to the MCL. The FMT complex was cycled
10 times to an extension of 2 mm at an extension rate of 2 cm/min. The
FMT complex was then stretched to failure at the same extension rate.
The resulting load-deformation curves defined the structural properties
of the FMT complex. Note that the deformation at the pin-bone interface
was considered negligible compared to that experienced by the ligament
and the ligament-bone insertion sites. Finally, utilizing the ligament
cross-sectional area and the mid-substance strain obtained from the VDA
system, the mechanical properties of the MCL substance could be deter-
mined.

RESULTS

Varus-Valgus Laxi~1. - For the six week time period studied, the experimen-
tal varus-valgus laxities for Groups 1. 2 and 3 were on the average 1.2S,
2 and 2.S times, respectively. the contralateral controls (Fig. 2). At
12 weeks. the experimental laxity values of Group 1 approached normal,
while Groups 2 and 3 were still approximately twice as much.

7 Group ~ - Histogram

.3
01 demonstrating varus-
~ 6 valgus (V-V) laxity
...
";;e
~2
for each group. The
...
~z
"'N
... 0
" +1
5 V-V laxity was defined
as the angular defor-
-0
u..
4 mation at a bending
z>- moment of 0.2 N-m.
- ....
oz
>-:e
""0 3 Improvements were
~:e seen in all groups at
0'"
.... -
.... z
2 12 weeks compared to
c~
0:: ....
6 weeks with Group 3
""CD
-'
achieving the best
=> results.
'"z
"" 0
6 WEEKS 12 WEEKS

Structural Properties of the FMT Complex - The cross-sectional areas of

the MCL healing site of both 6 and 12 week animals were found to be simi-
lar for all experimentals and approximately twice that of the contralater-
al controls (Table). The one exception was experimental MCLs of Group 3
at 12 weeks where the cross-sectional areas were only SO% higher than
that of the control.

For the load-deformation (i.e. structural) properties of the FMT complex.

at six weeks. the "linear" stiffness (defined as the slope between 3 and
S mm deformation) for both Groups 1 and 2 was approximately 60% of the
control (Fig. 3) (p<O.OS for both groups). while that for Group 3 was
only 36% of the control (P<O.OS). At 12 weeks. these values approached
that of the control. particularly in Groups 1 and 2 (Fig. 3).
170

Group I Group 2 Group 3

Cont. Exp. Cont. Exp. Cont. Exp.

Cross-sect. 6 wk: 13.9+1.0 26.9+2.7* 14.0+ 1.4 30.1+5.2* 14.0+1.3 31.7+6.0*

area (mm 2 ) 12 wk: 9.2+1.8 17.4+3.0* II .8+ 1.9 19.8:<:4.4* 10.1+1.3 14.1+1.5*

Tensile 6 wk: 53.5+5.1 10.7+2.7* 47.2+ 3.2 9.0+1.8* 60.3+6.5 3.6+1.6*

strength 12 wk: 76.4+18.1 36.4+6.2* 69.6+11.6 32.6:<:5.9* 74.9+7.2 27.5+2.3*
(MFa)

Ultimate 6 wk: 14.0+2.5 11.0+1.0 15.0+ 1.0 8.0+1.0* 14.5+0.5 8.0+0.5*

strain(%) 12 wk: 12.5:<:0.5 14.0+0.5 15.1+ 1.5 13.0:<:1.5 15.0+1.0 13.5+1.5

* p<O .05

TABLE - Cross-sectional area and failure properties of the MCL. For

description of Groups. please see Methods section of text.

(6W) (12 W)
500

~ - The struc-
400 tural properties of
the FMT complex for
both the 6 and 12-
Z 300 week groups. Positive
C>
changes were seen for
::
<l

200 the 12-week groups,

but no group returned
to normal.
100

0
0

DEFORMATION ( mm )

At failure. the ultimate load for each group at six weeks paralleled the
varus-valgus laxity results. That is. values for Groups 1 and 2 showed
positive changes relative to Group 3, but no experimental FMT complex
achieved normal values. The ultimate load values for Groups 1, 2 and 3
were 40, 40 and 20%, respectively of the control. At 12 weeks, the ulti-
mate load for experimental FMT complexes of Groups 1 and 2 achieved val-
ues very close to the control, but Group 3 values were only half. For
the six week study. al1 FMT complexes failed in the mid-substance with
the exception of three sham-operated controls which all failed by tibial
avulsion. At 12 weeks. all FMT complexes failed in the ligament sub-
stance except three FMT complexes on the experimental side (two from
Group 1 and one from Group 2) and two FMT complexes on the sham-operated
side (two from Group 1).

Mechanical properti!:.~~~2'!~~ub:'..taI!~~ - For the stress-strain curves,

the healing MCL substance at six weeks from Group 2 was found to be the
 171

most improved. but was still substantially different from the control
ligament substance (Fig. 4). Improvements were seen for all experimental
MCLs at 12 weeks with Groups 1 and 2 being similar. The tensile strength
of the experimental MCL from Groups 1 and 2 at six weeks were 20% of the
controls while that for Group 3 was only 5% (Table). At 12 weeks. the
tensile strength values for Groups 1 and 2 became approximately half that
of the control. whi Ie those for Group 3 remained only at the 40% level.
At six weeks, the ultimate strain of experimental MCLs from Group 1
approached that of the contralateral side while Group 2 and 3 values were
about 50% less. At 12 weeks, the ultimate strain values between
experimental and controls were equivalent (Table).

6 WEEK 12 WEEK
30 30

~ - The mech-
- Shorn operated - Shom operated and normal anical (stress-
- - - Group 1
- - - Group 1 strain) properties
2 2
20 20 of the ligament
3
,. substance for both

~.
~
it.
'"
J
~
control and experi-
~

mental groups at
~
~
w w

t1"+
GO
GO
6 and 12 weeks.
~
~ ~
~

10 10
Note that these
properties were not
.~t:~~~ .,'"
J
1---- ~- as improved as
/." ___,t;:;;.;::f.;:==::.:.r:::::::r those seen struc-
--i --::i;::;~.~'" , , 0
tura11y.
!

0 0
STRAIN (%) STRAIN ('1,)

DISCUSSION - This study emphasizes the use of experimental models and

techniques to evaluate the effectiveness of different clinical treatment
regimens for MCL injury. A key point to the approach utilized is that
the exact parameters necessary to achieve a "maximum" healing state may
be defined. In other words. these models and techniques not only show
how varus-valgus laxity of the knee correlates to the healing ligament's
biomechanical properties, but help one to follow the healing progress of
the isolated ligament as well.

The varus-valgus laxity data for the present study brings into question
the use of repair and immobilization for the treatment of MCL injury even
though improvements are seen over time. Again, for the twelve week
groups, experimental Group 1 laxity values were approximately normal,
while those for Groups 2 and 3 were two times higher.

Structural properties of the FMT complex (i.e. the load-deformation behav-

ior) for all groups are also seen to improve over a l2-week time period.
Positive changes in these properties reflect the decreased joint laxity
mentioned above and may indicate that the structural properties of the
FMT complex playa significant role with regard to the ligament's ability
to resist valgus loading.

The results show the mechanical properties of the healing MCL (i.e. the
stress-strain behavior) to be comparatively less improved over the same
172

period of healing time. The effect of this reduction in the ligament

mechanical quality on overall joint function is unknown, but it may
affect long term joint stability. If the cross-sectional area of the
healing ligament reduces over time like that seen for Group 3 at 12
weeks, then the mechanical properties of the MCL may override the impor-
tance of the structural properties of the FMT complex in preventing joint
laxity. Further long term studies are therefore indicated.

ACKNOWLEDGMENT - The authors gratefully acknowledge financial support

from the M&D Coutts Institute for Joint Reconstruction and Research and
the RR&D of the Veterans Administration.

REFERENCES

1. Grood, E.S .. Noyes. F.R., Butler, D.L., Suntay. W.J.: Ligamentous

and capsular restraints preventing straight medial and lateral lax-
ity in intact human cadaver knees. J. Bone Joint Surg., 63A:1257-
1269.1981.
2. Kennedy. J.C., Hawkins, R.J .. Willis. R.B.: Strain gauge analysis of
knee ligaments. Clin. Orthop. 129:225'-229 1977.
3. Mains. D.B., Andrews. J.G., Stonecipher, T.: Medial and anterior-
posterior ligament stability of the human knee measured with a
stress apparatus. Am. J. Sports. Med., 5:144-153, 1977.
4. Warren. R.F., Marshall, J.L., Girgis. F.: The prime static stabi-
lizer of the medial side of the knee. J. Bone Joint Surg.,
56A:665-674 , 1974.
5. Woo, S. L-Y., Akeson, W.H., Jemmott, G.F.: Measurement of nonhomogen-
eous directional mechanical properties of articular cartilage on
tension. J. Biomec. 9: 785-791, 1976.
6. Woo. S. L-Y., Ge1berman. R.H., Cobb. N.G., Amie1, D., Lothringer
K., Akeson. W.H.: The importance of controlled passive mobilization
on flexor tendon healing - A biomechanica1 study. Acta Orthop.
Scand 52:615-622, 1981.
7. Woo, S. L-Y . Gomez, M.A., Inoue, M. and Akeson. W.H.: New Experimen-
tal Procedure to Evaluate the Biomechanica1 Properties of Healing
Canine Medial Collateral Ligament (MCL). To appear in J. of Orthop.
Res., 1986.
8. Woo, S. L-Y .. Gomez. M.A .. Seguchi. Y., Endo, C.M .. Akeson, W.H.:
Measurement of mechanical properties of ligament substance from a
bone'-ligament-bone preparation. J. Orthop. Res. 1:22-29, 1983.
 173

THE DETER~IIAnON Of LHjA~EIT STRAIN BY A COMPUTER CONTROllED

KNEE LOADING APPARATUS
L. Ches. l. DQrselen and H. Kiefer
lab. for Exp. TnulIIBatology. Universi t1 of Ula. UIIli. Ger.any

1. I NTRODUCTI ON
Due to the incongruity of the human knee joint surfaces its
ligamentous 'structures are the' most important factor
concerning the joint kinematics and stability. So
insufficient repair after ligament injuries can cause
chronical joint instabilities followed by posttraumatic
arthrosis. Such instabilities can be treated by implantation
of artificial ligaments as there are several types available
on the market. They all differ in material and structure and
thus in their mechanical properties. To figure out the
required properties of such a ligament prosthesis thorough
knowledge of biomechanics of the knee ligaments is needed.
Additionally this knowledge is necessary for an optimised
rehabilitation of knee joints after surgery .
. Some researchers already carried out investigations on the
biomechanical behaviour of knee ligaments but in many cases
their results contrad~ct each~ther. For this re~son we
designed a new apparatus, which allows in vitro measurements
of knee ligament strain as a func ion or the lexi
the main acting muscle forces
and different load cases. By
the special design described
below the physiological
kinematics of the knee joint
were guaranteed at any time.
Object of this investigation
was to d.etermine th. knee
ligament str.ains under various
load andination conditions
both the. n.ormal as well as the'
injured and reconstructed knee.
2. DESCRIPlION OF LOAD SIMULA
Knee specimens were prepared
by resection of all ~uscles and
by seperating the distal part
of the tibia and the proximal
part of the remur in a distance
of approximate.ly 25 em from the.
joint space.. Afterwards the
bony ends were embedded in two
pots by PMMA. These pots were
fixed in the corresponding
174

mounting support in the simulator (fig. 1). A motor driven

lever enabled the femoral part of the knee to flect the
joint in a range from full extension to about 115 degrees.
Two light barriers prevented the knee from hyperextension or
-flexion. In order
not to expose
the specimen to any
compulsive force and
due to the 3-dimen-
sional moving bending
axis of the human
knee joint it was
necessary to design a
multi-joint device,
which allowed the
specimen to move
according to its own
anatomical kinematics.
The scheme of the set-
up (fig. 2) indicates
the different degrees
Fig. 2: Scheme of the knee of freedom as there
load simulator are rotation and
varus-valgus movement
of the tibia, which is mounted in a cardanic suspension.
Additionally the drive suspension provided a bearing that
enabled a horizontal shift of the motor connected to the
femur. To approximate physiological test conditions the set-
up provided the simulation of muscle forces.
The following muscle groups could be applied:
1. musculus quadriceps femoris
2. musculus gastrocnemius
3. musculus biceps femoris + musculus semimembranosus
+ musculus semitendinosus
These muscle forces were generated by pneumatic cylinders,
which transferred the tensile force by ropes and cortical
screws to the insertion points of the muscles. The magnitude
o f t he sin g I e m u scI e muscle forces by knee bending (RlShrle 1880)
forces (fig. 3) were taken~
from calculations carried~
out by R6hrle (1). I n fact
we only applied 10 % of
these values due to the
limited dimensions of the
pressure cylinders and to
the difficulty to intro-
duce high tensile forces
to the insertion points.
The effect of external biceps
+semimembr.+semitend.
loads could be
investigated as well. As 110 110 100 1211
shown in figure 2 one a"
pneumatic cylinder in-
duced a tibial rotational
moment, another one a Fig. 3: Simulated muscle forces
varus-valgus stress. Thus depending on the flexion angle
 175

the application of both isolated and combined external loads

was posai bl.e.

3. MEASURING TECHNIQUE
To determine the strain of the ligaments special measuring
cella (3x12x8 mm in size, fig. 4) were developed (2). Thin Q-
shaped. copper aheets were equipped with strain gaugea
orientated 90 0 to each other working as a half bridge
circuit. The one at the top
delivered the strain aignal, the
other one provided a
temperature compensation. Two
small plastic feet with
transverae drill holes were
fixed under the Q-cell with
epoxy reain and the whole cell
was covered by a polyurethane
covering agent to protect it
against humidity. A calibration
Fig. 4: Strain gauge cell for test has ahown that a linear
the measurement of electricaignal proportional to
ligament e10ngation the change of the foot distance
could be received. The Q-cella
were sutured to the ligaments with
Prolene 4.0 using the transverse
drill holes <fig. 5). The balf bridge
circuit was connected to a multi-
channel digi tal amplifier (HBM/UPM
60), which provided a supply
voltage of 1 Volt, 5 kHz. This
amplifier was controlled by a
computer. The strain signals were
reproducable in a scattering range
of 3J.

Pig. 5 Two measuring cells

sewed to the anterior
cruciate ligament

The following ligaments were investigated:

1. anterior cruciate ligament, anterior medial band (ACLa)

2. anterior cruciate ligament, intermediate band (ACLi)
3. posterior cruciate ligament (peL)
4. medial collateral ligament, ventral part (MeLv)
5. medial collateral ligament, dorsal part (MeLd)
6. lateral collateral ligament (LeL)
176

4. COMPUTERIZED CONTROL
The great amount of data to be processed and the
simultaniously working control of muscle forces in dependence
on the flexion angle required the use of a computer. Our
measuring program was carried out by an OKI personal
computer. The whole control system is shown in figure 6. Each
measurement was started with a zero balance of the strain
cells in 60 0 flexion position. The operating sequence was
performed mainly by two subroutines. The first one was
responsible for the ligament strain measurement, the second
one controlled the muscle forces. The angle of flexion was
measured continuously. Starting with 0 0 (extension of the
knee joint) each 10 0 the program branched into the first
subroutine triggering the
amplifier to deliver the
strain data of all
Ii gaments. As well each 10
degrees but starting with
5 flexion the program
performed the second
subroutine to generate
analogous voltages
according to special stored
data to control the
simulated muscle forces.
These voltages were sent
out in a short
pressure transducer
chronological sequence and
distributed by an
electronic multiplexer to 4
electromagnetic pressure Fig. 6: Scheme of the control
transducers, which output a and measuring procedure
proportional air pressure.
Three of the transducers controlled the muscle forces as
described above, the fourth provided the facility to apply
external loads.

5. EVALUATION OF DATA
The evaluation branch of the program enabled us to look at
the results right after the measuring cycle. The strain of
the different ligaments could be recorded as a function of
the flexion angle (E=E(OCf either on screen or on a x-y
plotter. The relative strain was calculated by deviding the
measured elongation by the distance of the 2 fixation
sutures. The gauges were zero balanced in 60 0 flexion
position. The calculation and plotting of arbitrary mean
values was possible as well.

6. MEASURED LOAD CASES

The experiments included for each specimen the following
meas ureme nts:

flexion cycle

1. without muscle forces

2. wi th si ngle muscle forces
 177

3. with all muscle forces in combination

4. wi th external load (wi th and wi thout muscle forces)
4.1 internal/external rotation(1 Nm)
4.2 varus/valgus stress (5 Nm)
4.3 combination of valgus-ext. rotation or varus-int.rotation

The performance of these cases was repeated after setting of

defined injuries, e. g. resection of the ACL and after
replacement by ligament prostheses.

7. RESULTS
From the great amount of the resulting data one
characteristic plot is presented. Figure 7 reveals the
results of ligament strain measurement after pure fleXion,
i. e. wi thout muscle forces and wi thout acti ng external loads.
The distinct curves represent the different ligaments and
ligament parts. The anterior-medial band of the ACL (ACLa),
the dorsal part of the MCL (MCLd) and the LCL showed its
highest strain between 0 0 and 30 0 flexion. The ventral part
of the MCL (MCLv) manifested an opposite behaviour. Its
strain increased with higher flexion angle. In contrary to
the ACLa the PCL was released most in full extension. Further
on the PCL showed increasing strain till about 60 0 and then a
slight decrease again. Under the measuring conditions of pure
flexion the strain of all ligaments were in the range between
+4% and -4%. Application of external loads in any case led to
ligament strain up to 10%, whereas activated muscle forces
could diminish the occuring stress.
6
III 5
1"\

~ 4
~
....
3
2

I~
ACLa--
PCL - - - -
MCLu---
-1 LCL _ . -
-2
-3 MCLd---

-4
19 29 39 49 59 69 79 89 99 1 119
FLEXION
Fig. 7 Plot of ligament strains without muscles forces
and external loads
ACLa anterior cruciate ligament, anterior medial part
PCL posterior cruciate ligament
MCLv medial colateral ligament, ventral part
LCL lateral colateral ligament
MCLd medial colateral ligament, dorsal part
178

8. DISCUSSION
The apparatus described above allowed to measure ligament
strain under approximated physiological conditions. Further
on this method provides the possibility of knee joint
investigation under standardized conditions. The measurements
performed to date show that the relative high standard
deviation reported in the literature is caused by the
individual anatomical differences and the location of
measurement. Due to this fact a great number of specimens
should be investigated.

9. SUMMARY
The presentation reveals a possibility to evaluate the
strain of ligaments in cadaver knee specimens. The mechanical
set-up and the computerized control system makes it a good
device to get detailed information about a lot of different
load cases of the human knee joint. Knee specimens were
mounted in the apparatus. A motor drive flected the joint and
a cardanic multi-joint suspension provided all degrees of
freedom needed in knee joint motion. Muscle forces could be
simulated with pneumatic pressure cylinders. Special designed
strain gauge measuring cells, which were sutured to the
ligaments, transfered strain data during a flexion cycle to a
computer via a digital amplifier. Additionally external loads
could be applied as there are rotational moments or varus-
valgus stress. The resulting strains were plotted as a
function of the flexion angle.

10. REFERENCES

1. Rohrle H., Scholten R., Sollbach W.

(1977) Kraftflussberechnungen in Knochenstrukturen und
Prothesen, Phase II, BMFT Bericht 01VG 106-ZK14MT267

2. Dtirselen L., Claes L.

Simulation und Messung der Beanspruchung des Knieband-
apparates, rechnerkontrollierte Steuerung eines Simulators
und On-Line MeBwertverarbeitung, Biomed. Technik, Band 30,
Erganzungsband,1985, S. 46,47

3. Claes L., Dtirselen L., Ki efer H.

Influence of load, flexion and muscle forces on the stress
and strain of knee ligaments, Transactions of the 32nd
Annual Meeti ng of the ORS, New Orleans, 1986, p.238

4. Claes L.
Biomechanische Eigenschaften humaner Bander. In: Burri C.
Claes L. (Hrsg.) Alloplastischer Bandersatz. Huber, Bern
Stuttgart Wien, S.12-19

Author's adress:
Pri v. -Doz. Dr. L. Claes
Abt. Experimentelle Traumatologie
der Uni versi tat Ulm, Oberer Eselsberg
7900 Ulm West Germany
 179

REDESIGN AND DEVELOPMENT OF A FORCE TRANSDUCER OF THE BUCKLE

TYPE
G.W.M. Peters**, A.A.H.J. Sauren* and H. v. Mameren**
Eindhoven University of Technology*, Eindhoven, The Nether-
lands, Rijks-University Limburg**, Maastricht, The Netherlands

ABSTRACT A redesign of the so-called buckle transducer is presented

that is used for the local measurement of static tensile forces in membrane-
like connective tissue structures around the elbow joint. This new design
involves also a special tool for installation of the transducer.
To investigate the importance of various design parameters a mechanical
model of the buckle transducer has been developed. This model has proven to .
be an important tool for optimizing the design. Test measurements show the
experimentally determined characteristics to be in good agreement with model
predictions and the transducer to come up to expectations.

INTRODUCTION The so-called buckle transducer is a suitable and, to the

best of our knowledge, only means for direct measurement of static tensile
forces in rope or membrane-like structures. Direct measurement of forces is
an important advantage because the mechanical properties of the test
material often are not known and do not allow calculation of force from
strain measurement. The buckle transducer basically consists of a frame and
a removable crossbar (Fig. 1), the crossbar being much stiffer than the
frame. A strip of tissue is led through the frame and over the crossbar.

FRAME

TISSUE

Fig. 1. An installed Fig. 2. Alignment

buckle transducer errors

Exertion of a tensile force on the strip will cause the frame to deform
which results in a straingage voltage output that is directly related to the
tensile force, thus providing a means for direct force measurement. Existing
designs (1,2), apart from their relatively large size, are prevented from a
succesful application towards our objective by three major disadvantages
affecting accuracy and reproducibility. First, as the frame is considerably
wider than the tissue strip led through (Fig. 1), deformation of the frame
180

will result not only from interaction with the strip but also from an
unknown interaction with the surrounding material. Secondly, the actual
mutual alignment of crossbar, frame, and tissue strip may deviate from the
desired configuration because their relative positions are not uniquely
defined (Fig.2). Finally, from some pilot studies we concluded that the
transducer's characteristics appeared to show significant hysteresis.
Starting from these experiences and with tQe aid of a mechanical model we
arrived at a new transducer design. Because of both the small size ~f the
new transducer and some stringent prescriptions for its installation a
special installation tool was designed as well. The transducer design and
the installation tool are presented in the next section. In the subsequent
sections attention is paid to the formulation of the mechanical model of the
buckle transducer and its application for design optimization, especially
with regard to the reduction of hysteresis phenomena. Finally some test
results are discussed.

REDESIGN OF THE BUCKLE TRANSDUCER The redesigned buckle transducer is

shown in Fig. 3 (dimensions: O.75xl.Sx5 [mm]). The width of the frame is
equal to the width of the tissue strip, so that possible interaction with
the surrounding tissue is minimized. Guides provide proper alignment of the

,----LEADWIRE
,----COATING LAYER 1
2

Ii STRAINGAGE
3

FRAME
GUIDE
L----TISSUE
CROSSBAR

Fig. 3. The redesigned buckle transducer (installed)

transducer on the strip. The applied straingage (Micro Measurements SA-09-

030CG-120) is cut to its minimum dimensions (1.4 x 2.1 [mm]). For electrical
insulation a triple layered coating-system is used (CETA BEVER Two-part
Epoxy resin and Micro Measurement M-Coat D and M-Coat G, where the resin
gives also a firm connection of the wires to the frame). Frame and guides
are made of stainless steel while the crossbar is made of Molybdenum.
Because of the relatively high Young's modulus of Molybdenum the crossbar
combines a high stiffness with minimum dimensions. The transducer is
designed for a maximum load of lOrN]. In Fig. 4 the specially developed
installation tool is depicted and in Fig. 5 (a detail of the tool) the
situation is illustrated in which frame and crossbar are brought together in
the installation tool. When installing the transducer a bending stamp is
moved with the aid of a micrometer spindle so that the ends of the
(disposable) U-formed crossbar are bent over the frame. This spindle is also
 181

used to adjust the distance 60 (Fig. 3), the importance of which will become
clear from the next sections.

FRAME

CROSSBAR

(B') BENDING STAMP

Fig. 4. Installation tool Fig. 5. Installation tool, transducer,

and crossbar (detail)

A MECHANICAL MODEL OF THE BUCKLE TRANSDUCER To derive a relation

between a tensile force F(O) acting on the tissue strip and the strain
induced by this load in the fra'ilc of the transducer, we consider the quasi-

Fig. 6. Forces acting on an infinitesimal part of the strip

182

static equilibrium of a part of the strip that is in contact with the frame
(Fig. 6). It should be noticed that the tissue strip slips over the frame
when it is loaded (and slips back in case of unloading). Friction between
strip and frame will be taken into account supposing the Coulomb friction
model is valid here (which is a simplification). From equilibrium of the
forces acting on an infinitesimal part of the strip and on the assumption
that full slip occurs the following expression may be derived for the force
...
F f acting on the frame:

where F(0)=-F(0)~1' ll\iS the coefficient of friction and the plus and minus
signs in front of the powers of the exponential functions apply for the
cases with increasing and decreasing load F(O), respectively. Neglecting
shear forces, it is easily derived that the maximum principal strain in a
cross-section at a position x (Fig. 7) is given by:

EX = - i~01[(6~ sin(a 1-p) + cos(a 1-p exp(tlla) + 6~sin(p) - cos(p)] (2)

where E is Young's modulus and A the cross-sectional area. In equation (2)

the angle a 1 is a function of the load F(O) and the momentary thickness d of
the tissu~ strip on which F(O) is acting. This means that the load-strain
relationship is non-linear. From Fig. 7 it follows that:

/)0 + d - U 1COS(P) - u 2
tan(a 1)
L 2 + u 1sin(p)
(3)

where:

(4)

~- --
F<o)
L

Fig. 7. Deflections of frame and crossbar

 183

where I represents the moment of inertia of the cross-sectional area of the

1
frame. It can be proved that the deflection u 2 in (3) is negligibly small
compared with u 1 The thickness of the loaded tissue strip is in general a
function of the load F(O) and must be specified for each type of material.
I t is not possible to derive, from equations (3), (4) and (1), an explicit
expression for the angle ~1 as a function of the load FeO). Therefore, for a
known load F(O), the load Ff on the frame and the angle ~1 have to be
determined by means of a numerical method. Subsequently, the strain. EX can
be calculated from equation (2).

HYSTERESIS AND OPTIMIZATION As can be seen from relation (2) a part of

the strain in the frame, which is a measure for the load F(O) on the tissue
strip, depends on whether this load is increasing or decreasing. This means
that, for the determination of the load, not only the value of the measured
strain should be known but also whether the strain is in- or decreasing.
Besides, relation (3) is based on the assumption that full slip occurs in
the contact-area of frame and tissue strip. Reversal of the load introduces
a certain trajectory in the load-strain relationship (2) during which there
is no or incomplete slip (Fig. 8). In that case equation (2) no longer holds

I increasing load bo, full slip

II decreasing load ~<O , incomplete slip
III: decreasing load ~<O , full slip
IV : increasing load bo, incomplete slip
with ~ =dF-
dt

F(O) -

Fig. 8. Load-strain relation with hysteresis.

and the load-strain relation is unknown. It is desirable that the term in

equation (3), accounting for load increase or decrease, equals zero or, at
least, is very small compared with the other term (we suppose ~=O is not
possible; in that case the problem would not occur). In the optimal
situation we have:

o (5)

It should be noticed that in this case equation (2) represents a linear

relation between load and strain.From equation (5) it follows that, for
184

given x, hand p, the angle U 1 has to satisfy:

U1 =P- arctan (6x)

h
(6)

As p, h and x in equation (6) have fixed values for a given transducer and
the angle u 1 is a function of the load F(O), there exists only one specific
load value for which equation (6) is satisfied. Thus some hysteresis will
always be present. To minimize the hysteresis a quantity H is defined
according to:

H (7)
with:

F(O)
r.lax
I (e (~~o,f>o)
x
- e x (~~o,f<O)dF(O) (8)
o
F(O)max
I (9)
o
In words, H is the quotient of the area of the hysteresis loop and the area
under the load-strain curve in case there is no friction. A transducer is
considered to be optimal when H is minimal. For a given thickness of the
tissue strip and a given buckle transducer one of the dimensions bO L2 or P
can be varied. For reasons of simplicity 60 is chosen as the adjustable
quantity. H is minimal for

o (10)

It is not possible to derive an explicit expression for the optimal bO but

again it is easy to find an approximation for bO where H is minimal by means
of a numerical method. An important advantage of, what is called an optimal
transducer, is the linear relation between load and strain as can be seen
from equation (2) if equation (5) is satisfied

TEST RESULTS AND DISCUSSION The new designed buckle transducer was
subjected to three types of tests. All the measurements were made on a linen
rope (d = 0.25 [mm]), so test conditions could easily be controlled and
specimens were unlimited available. First, hysteresis phenomena were
investigated. Fig. 9 shows the measured load-strain curves for two different
values of boo Hysteresis is demonstrated (curve 1), and, indeed, it is
possible to minimize it (curve 2). As stated before there exists only one
specific load value for which the hysteresis term in equation (2) vanishes.
A little amount of hysteresis will always be left as can be seen in Fig. 9
(curve 2). The almost linear behaviour of the transducer in the optimal
situation is also refle.cted by curve 2 of Fig. 9. Theoretical results have
been left out here as the mechanical model is used for a qualitative
analysis of the transducer's behaviour. Secondly, the interfering input
 185

&x10~~----r----r----r---~
t6~__~__+-__~~~

4,~--4----b~~--~

5 FIN)- 10
Fig. 9. Measured load-strain curves.

caused by temperature changes, one of the most occurring error sources, was
tested. Compensation of temperature effects is, as usual, done with a so-
called "dummy transducer" placed so as to assume the same temperature as the
active transducer. Long duration measurements (2000 samples over 8 hours)
demonstrated an average zero drift, in terms of measured force, of less than
0.03 [N) with a standard deviation better than 0.04 [N). Finally accuracy
and sensitivity were determined. Five-hundred random load values, which are
assumed to be errorless, were chosen within the range of operation (0-10
[N). The maximum deviation of a straight line (least squares fit) of the
corresponding output samples was less than 0.25 [N). As can be seen from
equations (1) and (2) the sensitivity of the transducer depends among other
things on the distance 60 between frame and crossbar. In the test situation
installation-accuracy was better than 0,03 [mm). From the mechanical model,
the deviation of the sensitivity corresponding to the installation accuracy
of 0,03 [mm) is calculated to be less than 2.6 [~5train/N] Measured
sensitivities of repeated transducer installations for two different
transducers are given in Table 1. The maximum difference between the
measured sensitivities agrees well with the predicted maximum range.

Transducer 1. Transducer 2.

Installation 1 64.5 60
,, 2 63.5 63.5
3 67 58
"
"
4 - 58

Table 1: Measured sensitivities [~5train/N]

186

CONCLUDING REMARKS The buckle transducer is a suitable and valuable

tool for direct static tensile force measurements in rope- or membrane-like
structures, specially when the mechanical properties of the test material
are not known and do not allow calculation of force from strain measurement.
With the reduced dimensions, the improvements concerning the proper mutual
alignment of frame, crossbar and tissue strip and the controlled way of
installation, the new buckle transducer has shown to be fairly accurate
taking into account the rather complex situations of its application: local
force measurement in soft, biological tissue.
It may be clear that for rational and optimal design a mechanical model is
indispensable. Even though it is based on some coarse simplifications the
model gives a good insight into the behaviour of the transducer and leads to
solutions for established problems.

This paper is based on the article "Redesign and Development of a Force

Transducer of the Buckle Type" published in "Experimental Stress Analysis".
ISBN 90-247-3346-4. Wieringa H. (ed).

REFERENCES
[1] Barnes, R.G., Pindler, D.N., J. Biomech, vol. 7, 1974, pp. 35-42.
[2] Lewis, J.L., Lew, W.D., Schmidt, J., J. Biomech. Eng., vol. 104, 1982,
pp. 125-128.
[3] Salmons, S., Bio-Med. Eng., vol. 4 1969, pp. 467-474.
 187

THE WRINKLING OF THIN MEMBRANE-LIKE CONNECTIVE TISSUE

STRUCTURES
D.G. Roddeman*. C.W.J. Oomens**. J.D. Janssen** and
J. Drukker*
University Limburg*. Maastricht. The Netherlands. and
Eindhoven University of Technology**.Eindhoven.The Netherlands

1. ABSTRACT
A general model describing the wrinkling of thin membranes is given. The
model is able to describe the wrinkling of anisotropic membranes in
geometrically non-linear analysis. Wrinkling is accounted for by replacing a
given deformation tensor, which by using the constitutive equation would
give negative Cauchy stresses, by a modified deformation tensor.
Based on this model a membrane element which is able to wrinkle has been
developed.
2. INTRODUCTION
The objective of our research is to study the force transmission from muscle
to bone near the elbow joint. The connective tissue structures which connect
contractile elements from the muscle to the bones often consist of thin
membrane-like structures. In an experimental set-up the passive structures
are isolated from the contractile elements. While a well defined load is
applied to the tissues, tensile force transmission in these tissues is
determined. Because of their geometry the membrane-like tissues will wrinkle
easily. This wrinkling has much influence on the stress state and the force
transmission. In order to be able to compare theoretical results with the
experiments, wrinkling has to be accounted for.
Wrinkling presents an ubiquitous problem. In biomechanics wrinkling of
skin or cardiac valves can be mentioned. Other examples can be found in
aircraft and spacecraft applications. Furthermore building industry uses
membranes ( fabric constructions ).
The modelling is based on the assumption that membranes have zero flexural
stiffness, so they cannot support any negative stress. When a negative
stress is about to occur the membrane will wrinkle. A model for the stress
field after wrinkling is a tension field. By definition a tension field is
uniaxial in the sence that it has only a single principal component no
matter whether it is two- or three-dimensional, flat or curved. In the
direction perpendicular to the lines of tension the membrane is wrinkled.
Some of the geometrical linear analysis on isotropic as well as anisotropic
membranes has been reviewed by Mansfield (1977). When the deformations are
finite the analysis is more complex. Wu (1981) is the only author we know
of, who modelled wrinkling in finite plane-stress theory, however only
dealing with isotropic materials.
In numerical analysis use has been made of the Finite Element Method.
A membrane element which is able to wrinkle has been presented by
Mansfield (1985). The element is based on geometrically linear analysis,
linear elasticity and isotropy.
Since the connective tissues we study may show large deformations, non-
linear stress-strain curves and anisotropy, a new model has been constructed
capable of dealing with these phenomena. Based on this model a new ,
generally applicable membrane element has been developed.
188

3. THEORY
We restrict our analysis to 'Cauchy-elastic' materials. If it is taken into
account that constitutive equations have to be objective we are allowed to
write:
a(F) J F . H(E) . F C (1)

with: a the Cauchy stress tensor

F the deformation tensor
E the Green-Lagrange strain tensor
J = det(F) the Jacobian
H(E) a tensor function of E
When a theoretical model does not account for wrinkling , deformations F may
occur which result in negative Cauchy stresses. In reality these
deformations will not occur because the membrane will wrinkle. Therefore, a
modified deformation tensor is needed which can be used in the constitutive
equation to generate the real Cauchy stresses. This tensor should take into
account that material lines in wrinkling direction are longer than they
would have been if the membrane would not have wrinkled. Thus, the modified
deformation tensor we use is of the form:

F' = (I + P ~1 ~1 ) . F (2)

where unity-vector ~1 denotes the wrinkling direction in which the principal

Cauchy stress is zero. The tensor (I + P ~1 ~1) lengthens a fictive
non-wrinkled membrane part in ~1 direction to become just as long as the
real wrinkled membrane part. Denoting the perpendicular direction by ~2
( ~1' ~2 is the principal Cauchy frame) , the direction of this frame and
the value of the parameter p have to be determined by using the restrictions
(3) and (4) :

a(F' ) o (3 )

a(F' ) o (4)

Equation (3) states tha~ in ~ri~kling di~ec~ion the principal Cauchy stress
is zero. The fact that n 1 ' n2 1S the pr1nc1pal frame after wrinkling, is
accomplished by equation (4). The parameter p ( never negative) is a
measure of the wrinkliness of the membrane.
At this point there is only one problem left. If a deformation tensor F is
given which, using constitutive equation (1), generates negative stresses,
how can one determine if the membrane is wrinkled or slack ?
Consider the Green-Lagrange strain tensor corresponding with F . If both the
principal values of this strain tensor are negative the membrane part is
completely slack. If however, in this situation, one of the principal values
of the strain tensor is positive, there are directions in which the material
is stretched, thus the membrane is wrinkled.
No assumptions have been made like geometrical linearity or isotropy.
The formulation given above is generally applicable.
4. NUMERICAL ANALYSIS
A Galerkin weighted residual method is used to obtain an integral form for
the field equations suitable for the Finite Element Method. After the
integral equation is put into an incremental form, the unknown nodal
displacements can be solved by an iterative procedure.
 189

Since there may be completely slack regions, the stiffness-matrix of the

entire structure may be singular. If however the iterative procedure is
chosen to be a globally convergent restricted step method, a solution for
the nodal displacements will be found although it might be not the only
solution.
Depending on the nodal displacements the three-noded element decides
whether it is taut, wrinkled or slack. 50, during an iterative proces new
estimates of the nodal displacements may cause the element to change for
example from slack to wrinkled. In such situations the new value for the
parameters p and the direction of the principal frame in the membrane are
determined iteratively using equations (3) and (4).

5. ANALYSIS OF A MEMBRANE SUBJECTED TO SHEAR

An example of a mechanical problem with a wrinkling membrane which can be
solved analytically is the 'simple shear' test (fig. 1) . The dotted lines
in fig. 1 show the undeformed configuration. F1 and F2 are the forces
necessary to shear the membrane over a distance u. We a priori choose the
constitutive equation (1) to be a plane-stress equation thus the analysis
can be made two-dimensional. In a 'simple shear' test a displacement field
is prescribed to the membrane which has the following form:
. ..
u = ~1 u e 2 (5)

resulting in the deformation tensor:

.. .
F = I + ~ e2 e1 (6 )
LO
The material of the membrane is chosen to be transversally isotropic with e 1
.
denoting the axis of rotational symmetry, so the matrix representation of
the tensor function H(E) with respect to the frame ~1 ' ~2 is chosen to be:
e E e v E e
H11 = f -----2 E11 + -----2- E22 (7 )
1 - v 1 - v
e E e v E e
H22 2 E22 + -----2- E11 (8)
- v 1 - v
e e E e (9 )
H12 H21 1 + v E12
where E~. are the components of the Green-Lagrange strain tensor with
respect1to' the base ~j , ~2 . The multiplier f determines the measure of
anisotropy ( if f=1 tne material is isotropic, if f>1 then ~1 is the
stiffest direction ) .
As an example we choose: u =5 [mm] '2LO=100 [mm] , BO=100 [mm] , the initial
thickness DO= 1 [mm] , E = 100 [N/mm ] and v=0.3 .
Using equatlons (6) to (9) would result in negative Cauchy stresses. Thus,
following the preceeding theoretical part, equation (6) is replaced with the
modified deformation tensor:
n 1 n... 1 ) . (I+ uL ....
F ' = (I + p .. e 2 e 1) (10 )
o
and in equations (7) to (9) the Green-Lagrange strains have to be based on
this deformation tensor. Indicating the direction of the principal frame by
the angle a (fig. 2 ) , the parameters p and a have to be determined by
making use of equations (3) and (4) .
Results for geometrically non-linear analysis are given in table 1.
190

For isotropic materials ( f=1 ) the formulation of Wu (1981) would generate

the same results because, in this situation, equation (2) degenerates to
the same modification of the deformation tensor as Wu proposes in order to
describe the wrinkling of isotropic membranes. The results for anisotropic
materials can only be determined by using the preceeding theoretical model.
In analysis without wrinkling the results for the forces Fj and F2 ar~
completely different ( table 2 ). Thus, it may be concluded that wrlnkling
may have much influence on the force transmission in a membrane. Notice that
if the membrane is isotropic ( f=1 ) the angle of the principal frame a does
not change because of wrinkling.
Results for geometrical linear analysis are presented in table 3. Please
notice that if the membrane stiffens in ~1 direction the lines of tension
also tend to this direction. The results of table 3 can also be found by
using the geometrical linear formulation of Mansfield ( 1977 ).
The shear analysis can also be made by making use of the element
discussed before. These calculations have been done with an element mesh
containing 96 elements ( fig. 3 ). When the sides of the membrane ( side
a,b,c and d ) have prescribed displacements according to simple shear, also
the nodal displacements in the inner-area turn out to correspond to simple
shear and the results of the preceeding tables are regained.
However when sides a and d are set free, the stress situation of simple
shear is not possible any more and the stresses and strains become
inhomogeneous. Results for this situation are given in table 4. Because of
the influence of the free ends the results differ significantly from the
results being found with simple shear approximation.
6. CONCLUSIONS
The shear analysis shows that wrinkling may have much influence on the force
transmission in membranes. A model has been constructed capable of
describing the wrinkling of anisotropic membranes in geometrically non-
linear analysis. This model can be used in the analysis of thin structures
, for example connective tissue structures near the elbow joint, cardiac
valves, description of skin, etc. In analysing the force transmission in the
connective tissue structures the membrane element will be used. This element
has shown promosing results in the shear analysis of a membrane.
ACKNOWLEDGMENTS
This study was supported by a grant from the Netherlands Foundation for
Biological Research ( BION ).
REFERENCES
Danielson D.A. , Natarajan S. , Tension Field Theory and the Stress in
Stretched Skin, J. Biomechanics, Vol. 8 , pp. 135-142 (1975)
Kondo K. , Iai T. , Moriguri S. , Murasaki T. , Memoirs of the Unifying
Study of the Basic Problems in Engineering by Means of Geometry, Vol. 1 ,
pp. 417-441
Mansfield E.H. , Analysis of Wrinkled Membranes with Anisotropic and
Nonlinear Elastic Properties, Proc. Roy. Soc. Lond. , Series A. , Vol. 353 ,
pp. 477-498 (1977)
Miller R.K. , Hedgepeth J.M. , Weingarten V.I. , Das P. , Kahyai S. ,
Finite Element Analysis of Partly Wrinkled Membranes, Computers Structures
, Vol. 20 , No. 1-3 , pp. 631-639 (1985)
Wagner H. , Flat Sheet Metal Girders With Very Thin Metal Web ,
z. Flugtechn. Motorluftschiffahrt, Vol. 20 , pp. 200-314 (1929)
Wu C.H. , Canfield T.R. , Wrinkling in Finite Plane-Stress Theory,
Q. Appl. Math. , pp. 179-199 ( July 1981)
 191

f a [-] f3 [-] F1[N] F2 [N]

0.773 1.74 10- 2 131.4 134.7
10 1.072 8.58 10- 3 337.9 183.9
20 1.166 6.46 10- 3 465.4 199.6

Table Simple shear results in geom. non-lin. analysis

with wrinkling

f a [-] f3 [-] F1[N] F2 [N]

0.773 0 13.7 193.0
10 0.924 0 137.4 199.2
20 1.059 0 274.7 206.0

Table 2 Simple shear results in geom. non-lin. analysis

without wrinkling

f a [-] f3 [-] F1[N] F2 [N]

0.785 1. 75 10- 2 125.0 125.0

10 1.059 9.84 10- 3 284.5 159.9

20 1.129 8.27 10- 3 354.7 167.7

Table 3 Simple shear results in geom. lin. analysis

with wrinkling

f F 1[N]

111.7 71.8

10 397.7 156.0

20 559.1 181.8

Table 4 Finite Element Method results in geom. non-lin. analysis

with wrinkling
192

Fig 1 Simple shear of a membrane

"

Fig. 2 The direction of the principal frame is indicated

by the angle"

side c side b

Fi9.3 An element mesh with 96 elements

 193

STRESS RELAXATION. USED AS A TOOL FOR DIAGNOSIS OF INCOMPE-

TENCE OF HUMAN CERVIX IN TERMS OF A MIXTURE MODEL OF TISSUE
W.A. van Duyl*. A.T.M. van der Zon*. C.W.J. Oomens** and
A.C. Drogendijk*
Erasmus University Rotterdam*. Rotterdam. The Netherlands. and
Eindhoven University of Technology**.Eindhoven.The Netherlands

1. INTRODUCTION
Cervical incompetence, that is failure of the cervix uteri to remain
closed under the load of the developing embryonic sac, is one of the main
causes of second-trimester abortion and premature delivery. A cervix known
to be incompetent may be treated by suturing it all the way around with a
synthetic fibre. However, there is a lack of a reliable technique to assess
the diagnosis directly, i.e. based on the determination of the mechanical
properties of the cervix. It is known that the mechanical properties change
during pregnancy under the influence of hormones. Therefore one may doubt
whether measurements done outside pregnancy, which clinically is
preferable, is likely to be of value in predicting incompetence. Neverthe-
less Van Duyl et al. (1) published a retrospective study on non-pregnant
women with cervices known to be incompetent and volunteers with normal
cervices, that relaxation measurements yield parameters which permit a high
level of discrimination between competent and incompetent. The technique is
based on observation of stress relaxation after stepwise dilatation of the
cervical canal by means of a cylindrical shaped nylon, i.e. almost
inelastic, balloon. After insertion into the canal the balloon is inflated
by filling it fastly with saline by means of an infusion pump until a
reference pressure of 33 kPa is reached. Under these isovolumetric
conditions the pressure in the balloon will gradually fall because of
stress relaxation in the cervical wall. For clinical convenience we
restricted the recording of the pressure to 20 minutes. Within this
recording period a static pressure level is almost reached. The pressure
decay curves can be fitted by a model consisting of two exponentials and a
constant. It has turned out that the values of the time constant of the
slower exponential and the value of the constant are negatively correlated
and that these parameters have a high discriminatory power for diagnosis of
incompetence. A prognostic study is going on to test the predictive value
of this new diagnostic technique. Measurements were done also on a group of
pregnant women (6-16 weeks) just before they had an artificial abortion for
medical or social reasons. These measurements reveal that the parameters
are correlated with the duration of pregnancy and tending to values found
in cases of incompetence outside pregnancy (2). We postulate that in cases
'of incompetence the condition is such that an equally directed additive
hormone-induced change during pregnancy brings the cervix in a weak
condition, which is threatening for gestation.
The clinically observed mechanical characteristics need to be related to
basic tissue properties. Histological studies have shown that the cervix
owes its competence to the strength of the elastic and collagen tissue of
the internal os (3). Stress-strain studies in vitro on strips of tissue
taken out of the internal os of fresh human cervix preparations have shown
that cervical tissue has viscoelastic properties which phenomenologically
194

can be presented by a lumped spring-dashpot model (4). Based on known

viscoelasticity coefficients and the assumption of tissue-incompressibility
one could derive the load-deformation relation of the intact cervix.
However, the assumption of tissue incompressibility is not beyond doubt.
Like all soft tissues cervical tissue contains interstitial fluid which
more or less can be expressed and drained by the lymphatic system. For skin
it is known that its water content has a considerable effect on its
plasticity (edematous tissue). Uldbjerg et al. (5) reported that during
pregnancy the water content of cervical tissue increases. These
considerations motivated us to perform compression tests on cervical tissue
in a way similar to what has been published for skin by Oomens et al. (6).

2. MATERIALS AND METHODS

A suitable experiment which has been used with success on articular
cartilage is a confined compression test (7). Figure 1 shows the apparatus
used to perform one-dimensional, confined compression tests x .
The load is applied to a tissue sample by means of a cylindrical indentor
2 with a diameter of 7.25 mm. The indentor is attached to a shaft 3 that is
guided by means of an air-bearing 4 and attached to a cantilever beam 5.
A balance weight 7 keeps the cantilever beam horizontal when unloaded. It
can rotate around an air bearing 9. A load of less than 0.01 N is
sufficient to overcome the remaining friction and to set it in motion. The
load is applied by manually placing a known weight 8 on the cantilever
beam. Disk-shaped tissue samples, 7.4 mm diameter and approximately 2 mm
thick, were taken out of the wall of the internal os of fresh human cervix
preparations by means of a hollow gimlet. A sample is placed in a stainless
steel ring of 7.4 mm diameter. This ring is placed on a sintered stainless
steel filter, which has a very high permeability for water. The accessory
is shown in Fig. 2. A displacement transducer (Sangamo DF/5.0/S) 10 is used
to measure the displacement of the shaft as a function of time.

r-~~I1---- .."ne solution

lucite ring
tISSUE' sampte1~~~~~~mt====(onfinlngring
.- IDlder

.....o..:+---suworling ring

FIGURE 1 FIGURE 2
Apparatus for finite compression tests Detail of Figure

%Placed at our disposal by Twente University of Technology, The

Netherlands.
 195

In order to start from standard initial conditions a small pre-load of

0.1 N was applied during about 10 seconds. Then a load of 4.1 N was applied
during 20 minutes. This load corresponds to a pressure of approximately
100 kPa, which is three times the pressure initially applied in the in vivo
measurements via the balloon. The in vitro measurements have been done at
room temperature.

3. RESULTS
Figure 3 is a representative result of 10 confined compression
measurements. The initial small decay is caused by the pre-load. The
application of the main load is followed almost instant~sly by a
relatively large compression. Then compression progresses slo~ly, which has
been recorded during 20 minutes. Semi-logarithmic plots reveal that about
one minute after the application of the load the continuing compression
follows a mono-exponential function approaching a constant value. To
characterize the compression recordings we determined graphically the
following parameters (see Fig. 3):
Ct - total compression expressed as a percentage of the thickness of the
sample: Ct = (22 2)%
ci - instantaneous compression at application of the main load expressed as
percentage of the thickness: ci = (9.7 5)%
T time constant of the mono-exponential compression:
T = 11.5 1.5 minutes

c(mm)

2.29

2.11

1.60::-mm- - - - - -
--------------------------- - - --- -
- - -----

10 15 t (min)

FIGURE 3
Recording of a compression test
196

4. THEORETICAL CONFINED COMPRESSION MODEL

Results of similar compression measurements on skin have been
interpreted in terms of a bi-phasic mixture model consisting of a porous
elastic matrix filled with water like a sponge. In this model the
compression is governed by the permeability of the matrix for water and the
elasticity of the matrix structure via the equations:

00 op
oy- oy 0 (1 )

0 (ov) 0
(~)
-oy 0 (2)
oy ot oy

where: y position variable in the direction of confined compression

o normal Cauchy stress on a plane with its normal in the y-
direction
p the hydrodynamic pressure
v displacement in y-direction
K permeability in y-direction

For small deformations holds 0 = HE, where E = ov/oy is the strain in

y-direction and H is the confined compression modulus. For the final
equilibrium state of compression, when the hydrodynamic pressure is zero,
holds:

pL
(3)
H

where L is the thickness of the sample.

It has been derived that according to this mixture model the early time
response is proportional to It (8). Oomens (6) applied this result of
analysis to experimental compression data of skin and obtained an
estimation of the permeability coefficient K. The instantaneous compression
ci' seen at the application of the load, however, cannot be explained with
this model. For the measurements on skin Oomens ascribed this phenomenon
to an artefact related to consolidation of the sample in the ring. This
consolidation probably also affects the initial faster compression part of
the curve and consequently reduces the reliability of the values of the
parameter obtained by the applied analysis. Now it can be shown
theoretically, assuming a constant compression modulus and permeability
coefficient, that with increasing time the displacement deviates less from
a mono-exponential decay, i.e. less than 8% for t ~ 0.05 T (9), where
T is the time constant:

4 L2
(4)
2
TI HK

Because this theoretical conclusion fits to our experimental results we

used this analysis to derive Hand K. We found by applying (3) and (4)
H = 0.13 E + 06 N/m 2 20% and K = 1.89 E - 14 m4 /Ns 30%. We note
that in our approach the likely dependence of Hand K on the amount of
compression has been ignored.
 197

5. DISCUSSION
The in vitro tests show that samples of cervical tissue are compressible
because fluid can be expressed. The compression rate is limited by the
permeability of the tissue for the fluid. The time constant of the
(isotonic) in vitro compression process appears to be in the order of 11
minutes, which is about twice the value of the larger time constant found
in the in vivo (isovolumetric) measurements. The in vitro measurements
have been performed on tissue, which has been damaged by excision, and
under unphysiological conditions. Consequently the results are not
representative for tissue properties in the intact cervix. Even the
hypothesis of tissue compressibility needs further verification in in vivo
experiments. Because in connection to this hypothesis incompetence is
related to fluid content of cervical tissue, such experiments are valuable
from a clinical point of view. At this stage of investigation we can refer
to observations which are indicative but not decisive for the validity of
the mixture model for the cervix. In particular the results of the in vivo
relaxation measurements can be explained in terms of the mixture model as
follows.
It has been observed by means of marks fixed at a cervix preparation, that
at the application of a pressure of 33 kPa by means of the balloon, the
radial displacement in the cervical wall predominantly takes place within
a layer of 4-5 mm (2). The initial deformation of the cervix at the
application of the step in pressure is governed by the incompressibility
condition. Subsequently the deformation will change because of
viscoelasticity but also because of gradual tissue compression leading to
a new state of equilibrium. According to the mixture model the relaxation
rate is determined by the rate of drainage of the expressed water instead
of viscous slippage of tissue fibers according to the original
viscoelastic model. Larger values of the relaxation time constant found in
the in vivo measurements in cases of incompetence can be caused by a
reduced rate of drainage. In a series of patients we determined the extra
volume of saline which need to be infused into the balloon after the 20
minutes relaxation period in order to bring the pressure at the level of
33 kPa again. We take this extra volume as an estimation of the amount of
water that has been expressed out of tissue during the relaxation. In the
controls we found that 1.1 0.08 mm 3 water is expressed per mm 2 area of
the canal exposed by pressure. This amount is comparable to the amount
that is expressed out of samples of 4 mm thick in the in vitro compression
tests. For incompetent cervices this amount turned out to be about six
times larger. This indicates the patho-physiological significance of this
parameter.
Aukland (10) describes how local mechanisms are involved for osmotic
control of interstitial fluid. We conclude that a further evaluation of the
validity of the mixture model, as an alternative for the classical
viscoelastic model, to describe the mechanical properties of cervical
tissue is both physiologically and clinically relevant because it puts the
observations in a new framework of interpretation.
198

REFERENCES

1. Duyl WA van, Zon ATM van der, Drogendijk AC: Stress relaxation of the
human cervix: a new tool for diagnosis of cervical incompetence.
Clin. Phys. Physiol. Meas. 5, 207-218, 1984.
2. Zon ATM van der: Cervix insufficientie etiologie en diagnostiek.
Thesis Rotterdam, 1982 (in Dutch).
3. Danforth DN: The fibrous nature of the human cervix and its relation to
the isthmic segment in gravid and non-gravid uteri.
Am. J. Obstet. Gynec. 53, 541-569, 1947.
4. Duyl WA van, Veen BCM in 't, Zon ATM van der, Drogendijk AC: Mechanical
properties of the tissue around the internal os of human cervix uter
determined in vitro. Submitted, 1986.
5. Uldbjerg N, Ekman G, Malmstrom A, Olson KO, Ulmsten U: Ripening of the
human uterine cervix related to changes in collagen, glycosaminoglycans
and collagenoolytic activity. Obstet. Gynecol. 147, 662-662, 1983.
6. Oomens CWJ: A mixture approach to the mechanics of skin and subcutis.
Thesis Twente, 1985.
7. Mow VC, Holmes MH, Lai WM: Fluid transport and mechanical properties of
articular cartilage: a review. J. Biomech. 17, 377-394, 1984.
8. Holmes MH: Comparison theorems and similarity solution approximations
for nonlinear diffusion equation arising in the study of soft tissue.
SIAM J. Appl. Mat. 44, 545-556, 1984.
9. Duyl WA van: Analysis of confined compression test of cervical tissue.
In preparation, 1986.
10.Aukland K: Distribution of body fluid: local mechanisms guarding
interstitial fluid volume. J. Physiol., Paris, 79, 395-400, 1984.
ORTHOPAEDIC BIOMECHANICS: BONE
 201

A STUDY OF HAVERSIAN SYSTEMS

J.J. Sun* and J. Geng**

Dept.Sc.Res.*. Northw.Inst.Archit.Engng . Xian. Shaanxi. and

Dept.Orthop.Surg.**. Xian Med.Univ . Xian. Shaanxi. P.R.China

Abstract
In this paper, a simple method has been developed which can combine the
mechanical tests with histological techniques used in physiology.

By using the method, the deformation of Haversian systems could be

easily observed, and corresponding pictures were obtained. The fracture
initiation is discussed and the mechanical model of a single osteon is
described, based on the observed modes of deformation.

1. Introduction

Investigators are interested in the mechanical properties of human bone

(1, 2, 3, 4). REILLY and KATZ reviewed the mechanical properties of cor-
tical bone (5, 6). ASCENZI and BONUCCI isolated a single secondary osteon
and measured the Young's modulus in tension and in compression, they
suggested the mechanical behaviour of osteons (7,8). In 1971, KATZ pro-
posed a composite material model of bone (4). Using Katz's basic theory,
GOTTESMAN and HASH IN developed another model, with the composite cylinder
assemblage considered (9).

Fracture initiation is a particularly interesting subject.DEMPSTER and

LIDDICOAT pointed out that compression specimens failed in a microbucking
manner (10). FRASCA, HARPER and KATZ fractured the cross-section of the
osteon by bending by hand, and surveyed the mechanism and propagation of
microfissures (11).

The objects of this paper are to present the deformation of Haversian

systems under stress, to describe the fracture initiation of human femur
and to discuss mechanical model of Haversian systems.

2. Material and Methods

The specimens of right femur were taken from the legs of healthy male,
aged 23 to 25, who died of cardiac arrest within one hour. The short
cylinder column specimens and the cubic column specimens were made for
compression tests and standard plates were used for tension tests.

The motion velocity of the Instron cross-head was 0.5 mm/sec. or very
slow.

The specimens were loaded to different extents and then unloaded. As

soon as the specimens were removed from the testing machine, they were
202
decalcified with Plank-Rycho's solution. Slides were then cut along the
longitudinal direction of bone by routine methods, and Hand E and
Mallory's stains were used. By studying all of slides, one can see the
process of deformation of the Haversian systems.

3. The Deformation Procedure

It is clear that there is a need for the development of failure theo-

ries for human bone. The main difficulty is the fact that no data are at
present available on the mechanical behaviour of bone in the body. There-
fore, a simple method by which the specimens of isolated bones can be~b
served and measured has been developed. It can combine the mechanical
tests with section tecrillique used in histology in detailed in the
section II.

The deformation Haversian systems which were subjected to compressive

loading can be clearly observed. (1) At the initiation of loading, the
Haversian systems were hardly deformed, i.e., Haversian canal and Haver-
sian lamellae are not deformed(Fig. 1). (2) With increasing loading, ~he
Haversian canals were slightly bent to the direction of weakness of lamel-
lae and the Haversian lamellae were increased in thickness (Fig. 2). At
this time, each osteon, as a whole, will tend to become wider. This is
caused by microfissures that takes place in the weak interfaces between
osteons, i.e. in the cement line. This is process absorbs energy, (3) At
the ultimate load the Haversian systems were collapsed. From all of the
sections it is seen that the Haversian canals which are originally reason-
ably straight were bent and the Haversian lamellae were cracked along
planes of weakness (Fig. 3).

FIGURE 1. The diameter of Haversian canals

were hardly altered.
 203

FIGURE 2. a) The left represents that the bending occurs in

the directions of forward-backward; the right
shows the bending in the directions of left-right.

FIGURE 2. b) The Haversian lamellae are widened.

FIGURE 2. Bending description of Haversian canals and lamellae.

204

FIGURE 3. a) The Haversian canals FIGURE 3. b} The Haversian lamellae

are clearly bent. have been broken.

FIGURE 3. Th.e specimen is broken.

4. Mechanical Model of a Single Osteon

At present time, the ultrastructural pictures of Haversian systems can

be easily obtained by the scanning electron microscope (SEM) (12). KATZ
produced a composite material model of Haversian bone (4). Other investi-
gators observed the ultrastructure of Haversian bone and produced a simi-
lar model (7,8). These models, which only used anatomical technique, do
not consider the actual mode of deformation under load.

We have studied the deformation of a single osteon under load. It is

well kno~~ that knowledge of the deformations of material is vital to the
understanding of the constitutive relationships, which. depends on the
microstructures. Especially, the mode of fracture of Haversian bone is
very important for th.e stress-strain relationship.

Figure 4 shows that the microfissures appear to arise in the cement line
between osteons. We observed that all Haversian systems did not collaps at
the same time. Because of this, as well as considering the dimensions of
the Haversian canal and lamellae, a single osteon can be considered as a
viscoelastic thick-walled tube which is reinforced by th.e fibres of
cross-ply.

5. Fracture Initiation

Fracture is an important problem in clinical medicine. The point at

which a fracture begins or a crack initiates and then propagates is of
interest for several reasons. Investigators have performed a series of
 205

FIGURE 4. The microfissures appear in the cement lines

between osteons.

tests to salve the unknown factors about fracture initiation. DEMPSTER and
COLEMAN did the unaxial tensile testing with tibial specimens and found
that a "simple cleavage" with less fibrous tearing and lower ultimate
strength was noted. Based on their findings, they postulated that inter-
lam.ellas cement lines are tl1.e weakest structural feature of secondary
Haversian bone. AOJI and co-workers also found many cracks along the
boundaries between Haversian systems and interstitial lamellae, but there
are some differences between the results of DEMPSTER and of AOJI (13,14).

FRASCA, HARPER and KATZ reported that fractures can propagate through
cement lines, according to a thin transverse cross-section subjected to
bending by hand (11).

According to our observations of the deformation of Haversian systems,

it appears that several Haversian systems are not fractured at the same
time, but the deformation is first produced in the weakest portion of the
Haversian lamellae, consequently the microfissures appear to rise in the
cement line between osteons (Fig. 4).

As we know, the osteons have a diameter of about 200 ~ and a length of

1 to 2 cm (15). Cement lines are generally believed to be the regions of
high mineral and low collagen density. Hence, these cement lines are much
weaker than the osteons or the interstitial bone.

It is interesting to note that the same phenomena are observed in terms

of three kinds of different testings, DEMPSTER and COLEMAN did the tensile
test with tibial specimens, FRASCA et a1. performed the bending testing
with a thin cross-section, and we took the compressive testing with femur
specimens, i.e. the microfissures always occur in the weak interfaces
between osteons, and the fracture propagation appears to proceed along the
cement line of bone.
206

References

1. Katz, J .L., "The Structure and Biomechanics of Bone", The Mechanical

Properties of Biological Materials (edited by J.F.V. Vincent and J.D.
Currey), 1980, pp. 137 - 168.
2. Sun Jia-ju, Geng Jie and He Fu-zhao, "The Compressive Properties of
Long Bones", Applied Mathematics and Mechanics, Vol.2, No.6, 1981, pp.
733 - 742.
3. Saha, S., "The Dynamic Strength of Bone and its Relevance", Osteoar-
thromechanics, edited by Dhanjoo N. Ghista, McGraw-Hill Book Company,
1974.
4. Katz, J.L., "Hard Tissue as a Composite Material - I. Bounds on the
Elastic Behavious", J. Biomechanics, Vol.4, 1971, pp. 455 - 473.
5. Reilly, D.T., and Burstein, A.H., "The Mechanical Properties of Corti-
cal Bone", The J. of Bone and Joint Surgery, Vol. 56-A, No.5, 1974,
pp. 1001 - 1022.
6. Katz, J.L., "Composite Material Models for Cortical Bone", Mechanical
Properties of Bone-AMD- Vol.45, edited by Cowin, S.C ..
7. Ascenci, A., and Bonucci, E., "The Tensile Properties of Single Osteons"
Anat. Rec. 158:pp. 375 - 385, 1967.
8. Ascenci, A., and Bonucci, E., "The Compressive Properties of Single
Osteons", Anat. Rec. 161: pp. 377 - 391, 1968.
9. Gottsman, T., and Hashin, Z., "Analysis Viscoelastic Behavious of Bone
on the Basis of Microstructure", J. Biomechanics, Vol.13, pp. 89 - 96,
1979.
10.Dempster, W.T., and Liddicoat, R.T., "Compact Bone as a Non-Isotropic
Material", Am. J. Anat., Vol.91, pp. 331-362, 1952.
l1.Frasca, P., Harper, R.A., and Katz, J.L., "Collagen Fiber Orientations
in Human Secondary Osteons", Acta Anat., Vol.98, pp.l-13, 1977.
12.Frasca, P., Harper, R.A., and Katz, J.L., "Collagen Fiber Orientations
in Human Secondary Osteons", Acta Anat., Vol.98, pp.1167-1172, 1977.
13.Dempster, W.T., and Coleman, R.F., "Tensile Strength of Bone Along and
Across the Grain", J. Appl. Physiol., Vol. 16,pp. 355 - 360, 1960.
14.Aoji, 0., Katsura, M., and Bando, T., "Experimental Investigation on
the Relation Between the Area Percentage of the Interstitial Lamellae
and Compressive Strength of Compact Bone", J. Kyoto Pref. Med. Univ.,
Vol.68, pp. 888 - 890, 1959.
15.Sun Jia-ju, Geng Jie and Zhang Xiao-ru, "Mechanical Failure in
Haversian Systems on the Basis of Microstructure", Applied Mathematics
and Mechanics, Vol.5, No.5, pp. 1639 - 1649, 1984.
 207

THE ELASTIC MODULUS OF CANCELLOUS BONE: DEPENDENCE ON

TRABECULAR ORIENTATION
N. Inoue*. K. Sakakida*. F. Yamashita*. T. Hirai** and
T. Katayama**
Dept.Orthop.Surg.*. Kyoto Prefect.University of Med . Japan and
Dept.Mech.Engineering**. Doshisha University. Japan

ABSTRACT
We have developed a method of quantitative analysis of
trabecular orientation on the cancellous bone by a Fourier
analysis of the soft X-ray image. The relationship between
the trabecular orientation and tensile modulus E was investi-
gated on the cancellous bone specimens removed from the fresh
bovine tibia. The proportion of trabeculae in a group at an
angle e to the loading direction ,f(e), and the volume frac-
tion of the trabeculae,Vf,were obtained by applying the soft X-
ray image of the test specimen. The equivalent volume fraction
in the loading axis, Vfeq, was determined from a netting
analysis, that is, Vfeq = L cos 4 e f(e)Vf. A linear regression
was performed, giving the equation, E=-119+15234 Vfeq (MPa)
( r=O.96, P<O.01 ).

INTRODUCTION
Cancellous bone in long bones or vertebral bodies shows
heterogeneity and anisotropy based on the orientation of the
trabeculae and its three-dimensional structure. However, from
the statistic view point, it might be assumed that cancellous
bone is characteristic of homogeneity and anisotropy caused by
the orientation of the trabeculae.
Many investigators have examined the relationship
between elastic modulus of cancellous bone and apparent
densit y 2)3)4)8)11)12). On the other hand, the significance of
the trabecular orientation for the elastic modulus has been
also pointed out. In this paper the quantitative analysis of
the trabecular orientation is developed and the relationship
between the trabecular orientation and the elastic modulus of
the cancellous bone is discusse~

MATERIALS AND METHODS

!~~!. ~.E~.i!!!~!!
Cylindrical cancellous bone
specimen of 4 mm diameter , were
cut in the principal trabecular
direction from the proximal end
of fresh bovine tibiae. The both
ends of the specimen to a depth
of approximately 5mm were washed
out by water jet to remove the
bone marrow and defatted After Fig.1 Test specimen.
208

the ends were dried, the specimen was inserted into the mould
which was made of Tefron and defatted parts were embedded by
methy 1 me thacryla te (MMA) (Fig.1).

Measurement of trabecular volume fraction

---S;ft-X-':-r;:i image-o-r--t he-fe'S t -spec-rme n we r e t a ken in e a c h
orthogonal direction (Fig.2). A cylindrical cortical bone
specimen of 4 mm diameter from the diaphysis of a bovine tibia
was exposed on each film as a reference object. The soft X-ray
image was taken at 50 kV and 30 mA by using Softex-CMB-2, and
SAKURA contact film CS-100E was exprosured 60 seconds. The
transillumination image of the film was recorded by a charge
coupled device camera (NEC,VC1-B106) and digitized with a
resolution of 256x256 pixels and 64 gray levels. The digitized
image data were transmitted to a super-minicomputer (FACOM
83]00) to analyze the image.
The volume fraction of the
trabeculae, Vf, was calculated
by the ratio of summation of the
gray level of the cancellous
bone pixels to that of the
cortical bone pixels. That is,
LL Gcan(x,y) Soft Soft X-ray
Vf= image
LL: Gcor(x,y)
where Gcan(x,y) and Gcor(x,y)
was gray level of the cancellous
bone pixels and cortical bone
pixels respectively, and, x and
yare the Cartesian co-ordinates Fig.2 Difini tion of
of a pixel point (Fig.]). equivalent plane of test
specimen.
Measurement of trabecular orientation
---The--t~o -;iT me~s-r;-n;:l- F-o-U-rier-t-,-ra nsf 0 r m 0 f Gcan ( x , y ) was
performed using Fast Fourier Transform (FFT). To detect the
directionari ty information contained in the power spectrum of
F(x,y), it was transformed from Cartesian system (x,y) into a
polar system (r,e) (Fig.4). The function P(r,8) was
decomposed into P8(r) for each e , and Pr(e) for each r
yielding a pair of functions [pe(r),Pr(e)] to be used by

Gco,(X,V) Gcan(X ,Y)

x x
Cortical bone Cancellous bone

Fig.] Measurement of trabecular volume fraction.

 209

texture operator. The orientation for each e was indicated by

F(e)= t Peer) (2).
The distribution function of f(8) indicating the orientation
factor of the trabeculae for each angle 8 was defined as
F(e)
f(B)= (3)
L F(B)
The mean orientation angle, Bm, was decided from fee). The
three-dimensional directional vector of the trabecular
orientation was decided from two mean orientation angles in
each orthogonal plane. The plane which was formed by the
directional vector and the loading axis of the test specimen
was assumed an equivalent plane of the test specimen. Another
soft X-ray image was taken in the parallel plane to the
equivalent plane of the specimen, and the distribution
function, f(S) and mean orientation angle, em, for the
equivalent plane was measured by the same procedures (Fig.2)
1 J7)
The orientation efficiency factor in the loading direction,
a1. was determined from the netting analysis, that is,
Cl.1= L cos"ef(e) (4).
The equivalent volume fraction of the trabeculae in the loading
axis, Vfeq, was defined as,
Vfeq= 0>1 Vf (5).

IF( x.v )' I 11'{,.')i'

I 2-0 Fourier transform

X
Cartesian system Po!;ar system

Fig.4 Measurement of trabecular orientation.

Measurement of ~l~sti ~glg~

---f;;;8"11e - test was carried
out at a cross head speed of
0.1 mm/min on an rnstron
testing machine mode 1 1122,
and load-displacement curve
was recorded on the chart. Two
optical targets were attached
to both ends of the measuring
part of the test specimen.
Displacement of each target
during the tensile loading was
measured by non-contacting
optical displacement follower
(OPTOFOLLOW 500SD). The strain
of the measuring part was Fig.5 System of tensile test.
210

obtained by the displacement of two targets. The elastic

modulus was calculated from the stress-strain relation
(Fig.5)6).

RESULT
The load-displacement curve 00.08
U
in tensile test comprised an ~
initial linear part and a non- c
linear part prior to the ulti- 20.04
ro
mate fracture of the specimen. C
<l>
These were similar to Carter and 8
co-workers' results of the o 90 160
tensile study of the cancellous e (DEG.)
bone 4 ). The volume fraction of
the test specimens examined by
image analysis was between Fig.6 Trabeculer orientation.
0.064-0.409. The trabecular
orientation which was quantified
by Fourier analysis of the soft
X-ray image of the test specimen
showed the distributed configu-
ration (Fig.6). The orientation
efficiency factor, Cl.1, ranged
from 0.14-0.68. The equivalent
volume fraction in the loading
axis given by the equation (5)
was ranged from 0.010-0.158
(Fig.7)
For each test specimen
showing approximately short
range of orientation effeciency
factor, that is, above same
range of trabecular orientation
in the loading direction,
Young's modulus were found to be
a lVt= 0.1 04 0,1Vf=0.045
approximately proportional to Fig.7 Equivalent volume
the trabecular volume fraction. fraction in principal axes.

4000 f=0.9566
E=15234 a,v,-119
P<O.01
3000
I
0: 0
~ 2000
w
1000

o 0.05 0.10 0.15 0.20

Fig.8 Relationship between Young's modulus and equivalent

volume fraction in loading axis.
 211

A linear regression was performed on the data in the tensile

test, giving the equation
E = -119+15234 Vfeq (MPa) (6),
where r=O.96, P<O.Ol, n=43.

DISCUSSION AND CONCLUSION

Cancellous bone can be modeled as two-phase composite
material, that is, one phase is a solid phase which consists of
bone tissue, and the other phase is a fluid phase which
consists of blood, red and yellow marrow, nerve tissues,
miscellous cells, interstitial fluid and blood vessels 9) It
has been suggested that the fluid phase did not influence the
strength, stiffness, or energy absorption in the quasistatic
loading condition 3).
The three-dimensional structure of the trabeculae and their
orientation originates the heterogeneous and anisotropic
characteristics of cancellous bone. However, from the statistic
view point, it might be able to be assumed that cancellous bone
is characteristic of homogeneity and anisotropy caused by
orientation of the trabeculae. On this assumption, elastic
modulus is influenced by trabecular volume fraction and
orientation of the trabeculae.
Many investigators observed positive correlation between the
apparent density and elastic modulus of cancellous bone,and
they also pointed out the influence of the trabecular orienta-
tion on the elastic modulus 3 )4)8)12). Townsend et al.(1975)
suggested the dependence of the compressive elastic modulus on
the distribution of the trabecular orientation and proposed the
empirical constant to account for the distribution of
trabecular orientation 11 ). However, the effective evaluation
method on the trabecular orientation has not been investigated
so far.
In the recent study, two
dimensional Fourier analysis for
the image of the cancellous bone
has been applied to examine the
trabecular orientation. Oxnard
(1982) investigated the 9000
trabecular orientation of lumbar r :0.9686 0
vertebra using this analysis and Eex :0.9591En -91
0
found that there were additional P<O.Ol 0
architectural elements organized 6000
in directions other than ~
vertical and horizontal 10 ). ~
A netting analysis on the x
tensile elastic properties of w GI 3000
fibrous materials showing dis-
tributed fiber orientation has
been developed by Cox et al.
(1952). In this analysis, it was
shown that the effects of orien-
o 3000 6000 9000
tation of the fibers might be En (MPa)
represented completely by the
first few Fourier coefficients Fig.9 Measured Young's
of the distribution function for modulus versis estimated
the fibers in respect of Young's modulus.
212

orientation5).
In this study, trabecular orientation of the equivalent
plane for the test specimen was represented by Fourier analysis
for the soft X-ray image on the assumption of homogeneity and
anisotropy. The orientation efficiency factor was defined based
on the netting analysis and equivalent volume fraction of
trabeculae in the loading direction was calculated. Young's
modulus were found to be approximately proportional to the
equivalent volume fraction. The value of the Young's modulus of
bone substance inferred from the equation (6) agrees reasonably
with previous investigations. Correlation analysis shows good
agreement between Young's modulus measured by tensile test
,Eex, and that inferred from the image analysis of the soft X-
ray image of the test specimen using equation (6), En (Fig.9).
Thus Young's modulus of the cancellous bone in the principal
trabecular direction should be calculated effectively by volume
fraction on the trabecular orientation obtained only by image
analysis of the specimen.

REFERENCES
1. Bajcsy,R.:Texture gradient as a depth cue. Computer Grafhics
and Image Processing,5:52-67,1976.
2. Brown,T.D.,Ferguson,A.B.:Mechanical property distribution in
the cancellous bone of human proximal femur.Acta orthop.
scand.51 :429-437,1980.
3. Carter,D.R.,Hayes,W.C.:The compressive behavior of bone as a
two-phase porous structure.J.B.J.S.,59A:954-962,1977.
4. Carter,D.R.,Schwab,G.H. and Spengler,D.M.:Tensile fracture
of cancellous bone. Acta orthop. scand.,51:733-741,1980.
5. Cox,H.L.:The elasticity and strength of paper and other
fibrous materials.British J.Appl.Phy.,3:72-79,1952.
6. Inoue,N.,Sakakida,K.,Yamashita,F.,Hirai,T. and Katayama,T.:
The mechanical properties of cancellous bone(part 1)-tensile
characteristics,torsional characteristics-. Proceedings of
1984 Annual Meeting of Japanese Society for Orthopeadic
Biomechanics:107-113,1984.(Japanese)
7. Inoue,N.,Sakakida,K.,Yamashita,F.,Hirai,T. and Katayama,T.:
Quantitative analysis of trabecular orientation of
cancellous bone by Fourier analysis. Bone Morphometry,5:112-
120,1985. (Japanese)
8. Martens,M.,R.Van Audekercke,Delport,P.,P.De Meester and
Mulier,J.C.:The mechanical characteristics of cancellous
bone at the upper femoral region.J.Biomech.,16:971-983,1983.
9. McElmancy,J.H.,Alem,N.M. and Roberts,V.L.:A porous blocke
model for cancellous bone. New York,A.S.M.E.publication 70-
WA/BHF-2:1-9,1970.
10.0xnard,C.E.:The association between cancellous architecture
and loading in bone:an optical data analytic view.The
Physiologist,25:s37-s40,1982.
11.Townsend,P.R.,Raux,P.,Rose,R.M.,Miegel,R.E. and Radin,E.L.:
The distribution and anisotropy of the stiffness of
cancellous bone in the human patella. J.Biomech.8:363-367,
1975.
12.Williams,J.L.,Lewis,J.L.:Properties and an anisotropic model
of cancellous bone from the proximal tibia epiphysis.
J.Biomech.Engineering,104:1982.
 213

MECHANICAL PROPERTIES OF SUBCHONDRAL BONE FROM NORMAL AND

OSTEOARTHROTIC HUMAN FEMORAL HEADS
W.Bonfield*. R.R.H. Coombs**. C. Doyle*. D.J. Sharp* and
K. E. Tanner*
Dept. of Materials*. Queen Mary Collegue. London. and Orthop.
Surgery**. Royal Postgraduate Medical School. London. U.K.

1. INTRODUCTION
Subchondral bone Is that bone which lies immediately beneath the jOint
surface. comprising a bony plate and trabecular bone. The complex
anisotropic nature of trabecular bone makes its study difficult and because it
Is structurally an open cell foam. determi.nation of mechanical properties can
be complex. The subchondral plate is a thin bony structure with a curved
surface. making the preparation of precisely machined specimens and
mechanical testing difficult.
The understanding of the mechanical behaviour of the subchondral bone of
a joint is of fundamental Importance for several reasons. As an integral part
of the bony structure. the role of trabecular bone must be understood. The
finite element analysis of normal joint behaviour requires detailed information
of the load bearing properties within a particular joint. Skeletal fixation of
Implants depends upon the mechanical behaviour of trabecular bone. and
Improvements of prosthesis design rely upon accurate understanding of these
properties. Such developments also demand an understanding of alterations
with degenerative jOint disease.
Information on the mechanical alterations of bone in a diseased Joint may
also have important bearing upon the understanding of the aetiology and the
recognition of different categories of idiopathic osteoarthrosls. Most previous
studies of bone fall to consider the clinical conditions of the animal or patient
used. There is only limited and Inconsistent data on the mechanical
properties of the subchondral trabecular bone in humans and reported
compression strengths vary from 7.61<+/-45)kPa 1 to 310kPa in the weight
bearing trabeculae and 120-150kPa In the nonwelght bearing trabeculae 2 ; and
elastic modulus values ranging from 581<+/-17) MPa 3 to 4800-6900 for the
primary trabeculae 2 .
Studies which consider pathological bone are limited; the compressive
strength of five osteoarthrotic femoral heads varied from O. 25-13. 5kPa. with
an average elastic modulus of 344(S. D. 27. 6)" and the only work mapping
the variations of mechanical properties within the normal and pathological
human femoral head has been using a small number of heads 5 . Grynpas 6
found that subchondral bone in OA is both structurally thicker and qualitatively
more mineralised in eburnated weight-bearing than in non-eburnated
non-weight bearing areas. There has been no work which carefully correlates
changes in the subchondral bone in the normal or the osteorthrotic head with
the load-bearing geometry.
It is therefore poorly understood whether subchondral bone changes reflect
the histological heterogeneity. and how the bony changes of osteoarthrosis
correspond to these cartilagenous changes. Microhardness provides a readily
available means of studying the physical properties of the subchondral plate
which allows quantifation of the features of bone tissue in a way that histologic
214

and microradiographic tests cannot. Carlstrom 7 was the first to apply

micro-hardness techniques to bone. finding a close correlation between the
degree of mineralisation and the hardness of different structures of bone.
later confirmed by Amprln0 8 .
There has been much debate as to whether adaptive change of composite
structure contribute to variations in bone strength with load-bearing function
and pathology. Weaver 9 demonstrated that the hardness of fibular trabecula,!
bone was generally somewhat less than that of interstitial bone in the adjacent
cortex. and a study of the mlcrohardness of swine bones suggested that
microhardness varies within bone with the function of that part. correlating
with the morphological differences along the bone 1 0 . However. the mineral
content of osteoporotic bone per unit of true volume is the same as that of
normal bone 11. 12 and hardness is not significantly affected in
osteoporosis 9.
Isotope studies suggest that there is increased bone mass in osteoarthrosis.
but no increase In mineral turnover or content; histological studies. however.
suggest that osteoarthrotic bone undergoes qualitative changes making it more
brittle than normal bone 13 whereas the collagen. calcium. phosphorus and
magnesium content showed no difference in normal and osteoarthrotic
bone l ' . The role of the calcified layer of the articular cartilage is uncertain.
Benninghoff 15 considered that calcified cartilage made the transition between
hard bone and the much softer cartilage more gradual and so reduced the
likelihood of the two tissues shearing from each other. but X-Ray microscopic
studies have disclosed that this zone is more heavily mineralised than
bone 16 . The mechanical properties of the calcified layer in the normal and
the degenerative jOint has not been investigated.
We have therefore studied the subchondral plate and trabecular bone of the
human femoral head. using a new technique whereby up to 19 core
specimens may be taken of the subchondral bone. and using methods of
testing which allow analysis of hardness of the calcified cartilage and the
subchondral plate and the most superficial trabeculae where they join the
subchondral plate. These findings may be related to each other and to the
load-bearing mechanics of the jOint.

2. METHODS
Osteoarthrotic femoral heads. taken at the time of total hip replacement and
cadaveric femoral heads have been collected and osteoarthrosls was diagnosed
using the radiological features described by Murray and Jacobson 17 . The
Murray assessment 18 was used to diagnose secondary osteoarthrosis. and
dlopathic osteoarthrosls was classified into hypertrophic. atrophic or necrotic 19.
The femoral neck angle was measured in each case. Osteoporosis was
assessed in the osteoarthrotic cases using the Singh assessment of the proximal
femur trabecular bone pattern 2 0 and using iliac crest biopsies in cadaveric
cases 21 Male cadaveric. idiopathic atrophiC and hypertrophic. and secondary
osteoarthrotic femoral heads were tested. All specimens were stored by
wrapping each femoral head in a gauze soaked in saline immediately following
removal from the patient or cadaver. and stored at -20C. The articular surface
of each femoral head was classified into non-progressive and progressive
articular cartilage changes and early osteophytic changes 22.
A jig was designed which trephines 9.5mm diameter cylinders in the wet"
condition from each femoral head at set reproducible positions related to the
load-bearing geometry of the head. each core taken perpendicular to the surface
of the head. Although this is not the direction of load bearing for all parts of the
head. it does mean that each core specimen is in line with the main direction of
the subchondral trabeculae 2 3 . Each femoral head was orientated on the jig
 215

using the femoral neck angle and the lateral cortex and fovea as landmarks and
mounted with the trephine directly above the pole position. To prevent thermal
damage the trephine speed was kept low and Irrigated with Ringer's solution. Up
to 19 cylinders of subchondral plate and cancellous bone could be trephined from
each femoral head.
The femoral head was divided into three areas classed as major weight
bearing, partial weight bearing and minor weight bearing based on the work of
Brown and Shaw 24. The major weight bearing area was the superior pole of the
femoral head and the area around It, the next area downwards around the head
with the exception of the fovea was classed as partially weight bearing and the
remainder as minor weight bearing. Each core was X-Rayed and then sectioned
just below the level of the subchondral plate. The plate specimen was divided
saggltally, one half fixed for histology, and the other mounted In resin for
mlcrohardness testing. The ends of the trabecular bone specimen were cut
square. The cartilage thickness was measured.
For mlcrohardness measurement each subchondral bone specimen was set In
'Metset' metallurgical resin (Type FT> and then polished using 6p. diamond paste
on an automatic polisher, with quality checks by reflected light microscopy. Each
specimen was marked with 2mm lines on the resin, so enabling accurate
Identification of the site of testing on the specimen, and was then mounted jon
the Reichert micro-hardness testing machine, using a Vicker's diamond pyramid
Indentor. The specimen was visualised under vertical illumination at 400x with
reflected light and the diagonals of each imprint measured by means of a screw
micrometric eye-piece. The approximation In the readings is less than O. Sp..
Applying 20 ponds for 10 seconds. six to eight serial Indentations were made at
6Sp. intervals so that the superficial 400-S00p.m are tested. Local topography
was drawn onto a proforma. noting calcified cartilage. lamellar and Interstitial
bone. and trabecular morphology. The subchondral plate and trabeculae were
tested only with the Indent greater than twice the diagonal from the edge to
exclude edge effects. These were repeated at three different sites on each
specimen. and the results averaged.
The method used in this study was standardised as much as was possible.
Specimen preparation was the same In each case. they were all Immediately
stored by freezing and mlcrohardness testing performed as soon as each
specimen had thawed.
Compressive stress-strain characteristics of wet unconfined specimens were
measured. at room temperature. on a Schenk servomechanical UTM at a
constant strain rate of 0.017sec- 1 . From these studies. values of Young's
modulus of elasticity. the yield stress and yield strain were calculated.

3. RESULTS
These will be considered in two sections. the microhardness results for the
subchondral plate and the compression test results for the trabecular bone. Both
sets of results presented were obtained from males. Mlcrohardness testing was
performed on femoral heads from five cadavers. and patients with idiopathic
osteoarthrosis (2 atrophic. 3 hypertrophic) 2 necrotic osteoarthrosis. two with
secondary osteoarthrosls. and one with ankyloslng spondylitis. an inflammatory
disease affecting synovial joints. Within each of these disease groups there was
no statistically significant difference between the lamellar bone and the calcified
cartilage in any area of the femoral head. although for some of the osteoarthrotic
specimens there was insufficient calcified cartilage in the major weight bearing
areas for assessment. There was also no statistically significant differences
between the different load bearing areas within each disease group with the
exception of necrotic osteoarthrosls with softer bone in the major load bearing
area. a difference significant at the p<. 03 level.
216

All the results for each disease group were combined to give mean values
for each group which are shown in table 1. Unpaired Students T Tests were
performed and the resulting p values are shown in Table 2. There were
statlsticaly significant differences at the p<. 05 level or better between all groups
with the exceptions of Idiopathic Atrophic and Idiopathic Hypertrophic
osteoarthrosis. Idiopathic Atrophic and Secondary osteoarthrosls, and Necrotic
and Secondary osteoarthrosis. The results split into three general _qroups:
cadaveric with a mlcrohardness level of approximately 40kpmm . the
osteoarthrotlc group with mlcrohardness values ran~lng from 21 to 30kpmm- 2 ,
and ankylosing spondylitis at approximately 15kpmm - .

TABLE 1
Mean Mlcrohardness in kpmm- 2

Cadaveric 40.65 5.30 n=105

OA Idiopathic Atrophic 27.26 6.08 n=15
Idiopathic Hypertrophic 29.67 5.25 n=33
Idiopathic Atrophic & Hypertrophic Combined 28.92 5.57 n=48
Idiopathic Necrotic 21.55 4.85 n=14
Secondary 24.93 5.54 n=15
Ankylosing Spondylitis (1 patient> 14.952.49 n=7

TABLE 2
'p' Values for Unpaired Students T-Test between various groups

Cad Idio A Idlo H Idio A+H Necrotic Secondary Ank. Spon.

X X X X X X X Cad
.0000 X X X X X X Idlo A
.0000 N.S. X X X X X Idlo H
.0000 X X X X Idio A+H
.0000 .0095 .0000 . 0000 X X X Nec .
.0000 N.S. .0099 . 0232 N.S . X X Secondary
.0000 .0000 .0000 .0000 .0006 . 0000 X Ank. Spon .

Compression testing was performed on 10 cadaveric femoral heads. The

results from the three load bearing areas were compared. The Yield Stress,
Yield Strain. Young's modulus and cartilage thickness are shown in table 3. For
all four modalities. the major and partial weight bearing areas showed a
statistically significant difference from the minor weight bearing area (p<O. 02) ,
but there was no significant difference between the major and partial weight
bearing areas.
 217

TABLE 3
COMPRESSION RESULTS

Major Weight Partial Weight Minor Weight

Bearing Area Bearing Area Bearing Area

Yield Stress (MPa) 10.08.4.38 9. 12.4. 77 5. 32.4. 28

Yield Stress (%) 12. 57.6.00 10. 24.3. 16 8. 24.3. 30

Young's Mod (MPa) 116. 82.72. 5 118. 27.70. 8 80. 02.61. 2

Cartilage Thickness( mm) 2. 07.. 90 1.871.00 1. 24. 86

4. CONCLUSIONS
The lamellar bone and calcified cartilage of the subchondral shell are of the
same quality. as reflected by their microhardness. in all areas of the femoral
head. This shows that the mineralisation of the subchondral plate does not alter
with its load-bearing function. which is demonstrated in both normal and
pathological femoral heads. and also that the calcified cartilage and lamellar
bone components of the plate function as a single mechanical unit. All types of
OA are significantly softer than cadaveric femoral heads. and microhardness
varies between the different morphological types of osteoarthritis. Such
mechanical appraisal. therefore. will have an important role in the further
description of osteoarthrosis of the hlp. The trabecular bone supporting this
shell has variable mechanical properties which reflect its load bearing function.
This suggests that although the load bearing affects the structural mechanical
properties of the trabecular bone. the quality of the bone in the subchondral
plate is unaffected.

REFERENCES

1. Martens. M.. Van Auderkercke R.. Delport P.. De Meester P.. Muller
J. C. The Mechanical Characteristics of Cancellous Bone at the Upper
Femoral Region. J. Biomech. 16: 971-983. 1983.
2. Brown T. D. and Ferguson A. B. Mechanical Property Distributions in the
Cancellous Bone of the Human Proximal Femur. Acta Orthop. Scand.
51:429-437. 1980.
3. Evans F. G. and King A. Regional differences in some physical properties
of human spongy bone. In: Biomechanical Studies of the Musculoskeletal
System. Chapter 3. 49-67. Ed. by F. G. Evans. Pub. by Charles T.
Thomas Springfield.
4. Schoenfeld C. M.. Lautenschlager E. P.. Meyer P. R. Mechanical
Properties of Human Cancellous Bone in the Femoral Head. Med. and
BioI. Eng. 12. 313-317. 1974.
5. Reimann I. and Christensen S. B. A Histochemical Study of Alkaline and
Acid Phosphatase Activity in Subchondral Bone from Osteoarthritic Human
Hips. Clin. Orthop. 140: 85-91. 1979.
6. Grynpas M. A study of the Subchondral Bone in Osteoarthrosis. Personal
Communication 1985.
7. Carlstrom D. Mlcrohardness Measurements on Single Haversian Systems
In Bone. Experlentla. X/4: 171-172. 1954.
218

8. Amprlno R. Investigations on some Physical Properties of Bone Tissue.

Acta Anat. 34: 161-186. 1958.
9. Weaver J. K. The Microscopic Hardness of Bone. J. B. J. S.
48A:273-288. 1966.
10. Szilagyi M.. Kovacs A. B.. Palfalvi I. Relationship Between the Ash
Content and Microscopic Hardness of Swine Bones. Acta Veterlnarla
Academlae Sclentlarum Hungarlcae. 28: 455-462. 1980.
11. Kelly M.. little K.. Courts A. Bone Matrix and Osteoporosis. Lancet
2: 1125-1126. 1959.
12. Casucclo. C. An Introduction to the Study of osteoporosis.
Proc. Roy. Soc. Med. 55:663-668. 1962.
13. Havdrup T.. Hulth A.. Telhay H. The Subchondral Bone In
Osteoarthritis and Rheumatoid Arthritis of the Knee. Acta Orthop. Scand.
47:345-350. 1976.
14. Lereim P . Langeland N . Romanus B .. Petterson J. E .. Goldie I. F. A
Biochemical Analysis of Subchondral Bone of the Medial Tibial Condyle In
the Normal State and in Osteoarthritis and Rheumatoid Arthritis. Acta
Orthop.Scand. 46:672-684. 1975.
15. Benninghoff A. Form und Bau der gelenknorpel In thren beziehungen 3 ur
funktron. Z. Zellforsch. 2: 782-862. 1925.
16. Sokoloff L. The Biology of Degenerative Joint Disease. Pub. University
of Chicago Press. 1969.
17. Murray R. 0.. Jacobson H. G. The Radiology of Skeletal Disorders.
Exercises In Diagnosis. Vol. 1. 2nd Edit. Pub. Churchill lIvlngstone.
1977.
18. Murray R. O. The Aetiology of Primary Osteoarthritis of the Hlp. B. J.
Radiology 38: 810-824. 1965.
19. Solomon L. and Schnitzler C. M. Bone Turnover in Osteoarthritis of the
Hip. J.B.J.S. 67B:318.1985.
20. Singh M .. Nagrath A. R .. Mainl P. S. Changes In Trabecular Pattern of
the Upper End of the Femur as an Index of Osteoporosis. J. B. J. S.
52A:457-467. 1970.
21. Coupron P .. Meunier P .. Bressot C .. Giroux J. M. Amount of Bone in
Iliac Crest Biopsy. Significance of the Trabecular Bone Volume. Its
Values In Normal and in Pathological Conditions. In: Bone
Hlstomorphometry. Second International Workshop Lyon. 30 June/2 July
1976. p.39-53. Ed. Meunier P. J.
22. Byers P. D .. Contempomi C. A . Farkas T. A. A. Post-Mortem Studyof
the Hip Joint. Ann. Rheum. Dis. 29: 15-31. 1970.
23. Ogston A. On the Growth and Maintenance of the Articular Ends of
Adult Bones. J. Anat. and Physiol. 12: 503-517. 1878.
24. Brown. T. D. and Shaw D. T. "In Vitro" contact stress distributions in the
natural human hlp. J. Biomech. 16: 373-384. 1983.
 219

CONTROL OF CHONDRO-OSSEOUS SKELETAL BIOLOGY BY MECHANICAL

ENERGY
D.R. Carter*.**. D.P. Fyhrie*.**. R. Whalen*.**. T.E. Orr*.**.
D.J. Schurman*** and D.J. Rapperport*
Veterans Administration Medical Center*. RR&D. Palo Alto.
Depts.of Mech.Eng.** and Surg.***. Stanford Univ . CA. U.S.A.
1. INTRODUCTION
The growth and maturation of the chondro-osseous skeleton is accomplished by the
proliferation, maturation, degeneration, and ossification of cartilage. Bone tissue which replaces
cartilage during development is modeled to efficiently resist the loads to which it is exposed.
Osteoarthritis in the aged involves the degeneration and destruction of the remaining cartilage
on the joint surface and osteophyte formation. All of these biological processes are greatly
influenced by the mechanical loading. The role of mechanical stresses in chondro-osseous
biology is, however, incompletely understood.
In this study we present a consistent, comprehensive theory which is successful in describing
many features of skeletal morphogenesis, maintenance, and degeneration. The theory
incorporates a consideration of the multiple loadings imposed on the skeleton and the energy
which is dissipated during this loading history. Some portion of the dissipated energy is, we
propose, instrumental in triggering or controlling the biochemical processes involved in the
regulation of cartilage and bone biology.
2. THEORY
We have proposed that: 1) the natural sequence of events in all cartilage in the appendicular
skeleton is proliferation, maturation, degeneration and ossification; 2) the application of
intermittent deviatoric (shear) stresses (or strain energy) into cartilage will accelerate these
processes; and 3) intermittent compressive dilatational stresses (hydrostatic pressure) will retard
or prevent cartilage degeneration and ossification (1). The very existence of articular cartilage
in a mature joint can thus be ascribed to the inability of the subchondral ossification front
to proceed into areas of high magnitude hydrostatic pressure. Once ossification occurs in the
developing anlage, the new bone adapts its apparent density and orientation to the local stresses.
We assume that the bone adapts its morphology in response to a physiologic mechanical stimulus
which is necessary to maintain its apparent density and orientation.
The regulation of skeletal biology by mechanical forces is accomplished via the transfer of
mechanical energy. When loads are applied to the skeleton, energy is imparted to the tissues.
Most of this energy is stored in the form of intermolecular forces and is recoverable when the
loads are removed. Some of the energy, however, is dissipated or transferred in the tissues in
the form of heat or a change in internal energy. The concept of energy transfer, represented by
the hysteresis observed in the loading curve, must be a central issue in any effort to understand
functional adaptation.
Most investigators have implicitly assumed that only a portion of the energy transferred (or
dissipated in hysteresis) in vivo is in a form which the cells utilize in metabolic regulation.
Thus there are many theories that have been proposed for the specific control mechanisms
of cartilage and bone. These theories include: stress generated electrical potentials, fatigue
microdamage, direct load effects on cell membranes, and changes in mineral solubility under
load. The thermodynamic consideration of energy dissipation discussed above suggest that heat
energy created during cyclic loading may also affect tissue homeostasis.
We postulate that the natural progression of cartilage degeneration and ossification is
accelerated by energy dissipated during repeated loading. Analogously, we assume that the
apparent density and anisotropic orientation of the bone tissue is related to the density of
energy dissipated by repeated loading. The concentration or density of metabolically usable
energy which is dissipated in a tissue per day (U T ) can be approximated as,

UT = Ln;Ut , (1)
_=1
220

where the daily loading history has been summarized as c discrete loading conditions and each
ithloading condition is associated with n, cycles which dissipate an energy per unit volume of
['t.per cycle.
Cartilage
In cartilage, energy is dissipated principally by losses associated with the: I) cyclic deviatoric
(shear) stress, or 2) fluid flow created by pressure gradients. If we assume that fluid flow is
minimal or does not result in a usable form of energy dissipation then Ut is a function only of
the magnitude of the cyclic shear stress, U. i If we represent cartilage as a one-phase, nearly
incompressible material then the cyclic dilatational energy imposed on the cartilage is negligible
and the stored cyclic strain energy density for a cycle, U" is proportional to the square of the
cyclic shear stress. In many materials the energy dissipated per cycle can be approximated by
a power function of the applied stress magnitude (I). If cartilage behaves in a similar manner,
then we can write,
or (2)
and, from Eq(I),
c
(hoe Ln;UY, (3)
;=1

where p is a constant whose value is yet to be determined.

Bone
In bone tissue of apparent density p, a true bone tissue effective stress, Ub, can be determined
by normalizing the continuum model effective stress, U;, with respect to the continuum model
effective failure stress, Uult. If we assume that a power function can be used to relate the cyclic
energy dissipation, Ut ., to the stress magnitude in bone, we can write,
Ut oe u;:: or Uti oe (u;juuld m , (4)
where m is a constant. Since Uult oe p2 (3), Eq(l) implies that,

UT oe Lni(Ui/p2)m. (5)
i=1

For the purpose of investigation we hypothesized that UT is constant for all bone tissue.
Therefore,
C ] 112m
poe Lniui
[
(6)
,&=1

We have previously demonstrated that the continuum model effective stress measure which
is appropriate for bone maintenance theories must define a closed surface in stress space (4).
One effective stress which appears appropriate is the "energy stress" , defined as,
U;nergy == 2EU (7)
where E == continuum model bone modulus and U == continuum model strain energy density.
To determine the anisotropy or trabecular orientation at any location, we consider the
cumulative normal stress stimulus, Sn, on any plane. In 2-D, we let fJ define the plane
orientation so that,
(8)
where (Tn is the normal stress on the plane. Directions of higher strength are normal to planes
(fJ) which define Sn(fJ) maxima.
The energy dissipation concepts presented in this section were applied (Eq. 3,6,8) using finite
element models of the proximal femur and hip joint at various stages in development and aging.
3. APPLICATIONS
Prenatal and Postnatal Development
A plane strain finite element model consisting of 1400 quadrilateral and triangular elements
was created to represent the proximal half of the femur from 48 days after conception to a
 221

postnatal age of approximately 4 months (Fig. 1). The shape of the model was based on that
found four months after birth, since the basic shape of the femur does not change appreciably
over the development stages being considered. There was no attempt to model the changes in
femoral size during growth. It was assumed that size changes would be accompanied (in the
first approximation) by coincidental scaling of the force magnitudes acting on the structures.
The loading conditions in the embryo, fetus,
and neonate are poorly understood. The first
contractile elements of the muscular system ap-
pear in the embryo at about the same time as
the primary ossification site. In utero, loading
of the anlage is achieved by muscular activity
of the embryo and fetus as well as by external
loads imparted by movements and muscular ac-
tivity of the mother. It was assumed that the
range of loading patterns imposed on the prox-
imal femur could be represented by 3 loading
conditions as shown in Fig. 1. The directions of
the joint reaction force resultant for these load
cases were 18, 39, 58 degrees with respect to
the bone shaft. The joint pressure distributions
used were based on previous stress analyses per- Fig. 1 Prenatal/Postnatal FE Model
formed in our laboratory. The magnitudes were
set to be arbitrary but with equal value for all
loading cases. It was assumed that the abductor forces acting at the greater trochanter were
20% of the magnitude of the joint resultant force. The directions of the muscle resultant forces
in the 3 loading cases were 36, 47, and 58 degrees with respect to the bone shaft.
Initially, the entire model was composed of cartilage. Later
stages of development were modeled by changing the material
properties to simulate the progression of endochondral ossifi-
cation. The material properties used in the models were taken
from the literature. Young's modulus and Poisson's ratio for
cartilage were, respectively, 6.0 MPa and 0.47, and for all os-
sified regions, 5000 MPa and 0.35.
The shape of the ossification front was taken from the
literature. It has been previously noted, however, that
the growth plate and thus the mineralization front in the
metaphyseal region is approximately perpendicular to the
maximum principal stress directions. To examine this point,
we calculated the magnitudes and directions of maximum
normal stress from the 3 superimposed stress fields of the 3
different loading conditions in the all cartilage model. By
rotating these vectors 90 degrees, the progression of the Fig. 2 Growth Front Contours
ossification front patterns was evident (Fig. 2). It appears
that the ossification front acts to minimize the strain energy and strain energy dissipation in
the tissue.
At five stages of development, the stress fields were solved using the 3 loading conditions
described above. The density of energy dissipated over a fixed period in each cartilage element
was represented by adding the strain energy density values, U;, calculated under 3 different
loading conditions. This technique of representing the loading history of biological tissue is
consistent with the formulation of Eq(3) when one assumes that an equal number of loading
cycles are imposed in each loading condition (c = 3 and nl = n2 = ng) over a given time period
and the constant, p, is assigned a value of 1. The cumulative, superimposed strain energy
densities calculated were displayed as contour plots.
For the all-cartilage model (embryo at 48 days), the highest superimposed strain energy den-
sity and, by implication, the highest concentration of energy dissipation, was on the periosteal
surface at the midshaft of the diaphysis. This concentration causes formation of the tubular
diaphysis (Fig. 3A). As ossification moves toward the bone ends, the strain energy density in the
cartilage
immediately ahead of the ossification front is higher than average and is greatest at the pe-
222

riosteal surface (Fig. 3B). In what will be the metaphyseal area, the highest strain energy is still
adjacent to the mineralization front. The strain energy. however. is not as concentrated at the
periosteum but begins to be more distributed in the interior regions where cancellous bone will
eventually form (Fig. 3C). When the mineralization front approaches the epiphyseal region, the
energy distributions undergo a distinct change (Fig. 3D). The chondroepiphysis bulges outward
in the central area where it is unconstrained, creating tensile strains in the central region. This
creates an area of high strain energy in the center of the chondroepiphysis where the secondary
ossific nucleus appears (Fig 3D).

A B c D
Fig. 3 Strain Energy Contours (highest is black) Fig. 4 Maximum Hydrostatic
Pressures (highest is black)
Throughout development, biaxial compressive stresses are created near the joint surface
directly under the applied contact stress. This stress state reflects the imposition of high
magnitude, intermittent hydrostatic pressure and relatively low strain energy density in the
area which becomes the articular cartilage. After the osseous epiphysis forms, the articular
cartilage is still exposed to intermittent hydrostatic pressure thus preventing the degeneration
and ossification of the articular cartilage (Fig. 4). The magnitudes of these hydrostatic pressures
control the articular cartilage thickness in a mature joint.

Bone Remodeling and Morphology

Immediately after endochondral
ossification, the bone begins to
adapt its morphology to the func-
tional demands placed upon it.
This functional adaptation is
demonstrated here in the analy-
sis of a plane strain finite ele-
ment model of the adult proxi-
mal femur (Fig. 5) constructed of
1963 quadrilateral and triangular
elements (Fig. 5). An elastic mod-
ulus of 1.0 Gpa was initially as-
signed to all elements. To repre- i = 1 , = 2 , = 3
sent the loading history, we again
selected three distinct loading con-
ditions (c = 3) and assumed an Fig. 5 FE Mesh and Loading Cases of Adult Femur
equal number of loading cycles in
each condition (nl = n2 = n3). The three loading conditions (Fig 5B,C,D) yielded joint re-
action force of 2317, 1158, and 1548 N directed 24, -15, and 56 degrees from the vertical,
respectively. Since the stress distribution is dependent upon the distribution of apparent den-
sity, an iterative scheme was employed to successively remodel the femur until the calculated
densities converged to stable values. Utilizing the energy stress, art updated apparent density
distribution was calculated, Eq(6). New values for the elastic modulus of each element were
then assigned using the relationship, E = 3790p3. To achieve proper convergence we assigned
an apparent density of 0.64 g/cm 3 (a modulus of 1.0 GPa) to one element in the center of the
 223

femoral head. To prevent prediction of bone denser than cortical bone, a saturation density of
1.8 g/cm 3 was used to represent cortical bone.
The solution for apparent density distributions (Eq(6)) was found to be similar for both m=l
and m=4. Other studies (6) which apply our theories to study the influence of exercise on bone
mass suggest that the value of m is probably between 2 and 6. The magnitudes and directions
of Sn(O) maxima (Eq(8)) were solved to find the directions of strong anisotropy. In addition,
if only one maximum existed, the magnitude and direction of the minimum Sn(O) value was
determined so that states of biaxial compressive stress (especially in the epiphysis) could be
visually represented. The corresponding plots illustrate the distribution of bone mass (Fig. 6)
and trabecular orientation (Fig. 6) for the proximal femur.
Degenerative Joint Disease
The osteoarthritic changes in cartilage and subchondral bone are so closely related that it may
be unrealistic to attempt to identify a unique initial event in the disease process. Recent work
(5) has emphasized the fact that there is a slow, continuous growth and remodeling process at
the articular ends of bones throughout life which involves articular cartilage degeneration and
calcification. The site of cartilage ossification is, in fact, the arrested ossification front which
was instrumental in the formation of the osseous epiphysis during skeletal development. In
osteoarthritis, there are significant changes in both the cartilage and bone and this remodeling
process is altered.
The direct attachment of
subchondral bone to articu-
lar cartilage insures that the
mechanical deformations of
these tissues are intimately
related. When attempting to
assess the influence of me-
chanical loading on articular
cartilage, one must consider
the boundary conditions cre-
ANISOTROPY
ated by stresses and deforma-
tions in the subchondral bone
in addition to contact pres-
sures at the articular surface.
Mechanical deformations in
the subchondral bone of the Fig. 6 Bone Density (black=1.8 g/cm3 ) and Trabecular Orientation
femoral head and acetabulum
were calculated using non-linear 2-D contact and linear 3-D finite element models using a
single loading condition (c = 1 and ni = arbitrary constant in Eq(3)) representing the single-
stance phase of gait. The results were compared to the regional histologic and morphologic
osteoarthritic changes observed by others. We were particularly interested in the influences of
intermittent: 1) dilatational (hydrostatic) stress and 2) tensile strain in the subchondral bone.
Subchondral tensile strains create shear stresses in the adjacent articular cartilage. Therefore,
the approach of this study is consistent with our previous work on skeletal morphogenesis.
The magnitude of the subchondral bone compressive dilatational stress was found to correlate
with cartilage thickness and was highest in the superior femoral head and moderate at the
acetabular roof (Fig. 7). Areas of high surface contact pressures on the femoral head were
also areas of high dilatational compression in subchondral bone and cartilage in both the 2-
D (Figs. 4,7) and 3-D (Fig. 8) models and rarely the initial sites of degenerative change.
The seldom contacting surfaces of the medial-inferior and peripheral areas of the femoral head
and the roof of the acetabulum were areas which combined a low dilatational compressive
component with subchondral bone tensile strains (and stresses) tangential to the joint surface
(Fig. 7). Initial cartilage fibrillation and osteophyte formation are often found in these areas.
The findings suggest that fluctuating hydrostatic pressure in a functional joint inhibits vascular
invasion and the degeneration and ossification of articular cartilage. The generation of tensile
strain may accelerate the osteoarthritic process by direct mechanical damage. Additionally,
since tangentially oriented tensile strains are associated with a reduction in the compressive
dilatational stresses and an increase in shear stresses, they may permit or promote vascular
invasion, and cartilage degeneration and osteophyte formation. These mechanical principles
in osteoarthritis are the same as those which have been previously demonstrated to guide the
224

degeneration and ossification of the cartilage anlage during skeletal morphogenesis.

COMPRESSION TENSION

Fig. 7 Principal Stresses Fig. 8 Hydrostatic Pressure (greatest is white)

4. DISCUSSION AND CONCLUSIONS
A comprehensive theory for the influence of mechanical stresses in chondro~osseous biology
has been presented. Central premises in this theory are that in the appendicular skeleton:
1) cartilage proliferation, maturation, degeneration, and ossification is the normal sequence of
events, 2) cartilage degeneration and ossification are promoted by intermittent shear stresses
and inhibited or prevented by intermittent hydrostatic pressure, 3) osteoblastic and osteoclastic
processes are modulated to create bone tissue distributions such that the tissue mechanical
stimulation is equal at all sites and achieved with a minimum amount of bone. When
these premises are viewed in terms of the rates of strain energy dissipation (or transfer) and
intermittent hydrostatic compression, it is possible to encompass and explain chondro~osseous
biology in a consistent mathematical framework.
The pattern of ossification and the formation of the diaphyseal cortex were predicted by the
theory as was the geometry of the primary ossification front. The growth plate location and
geometry were predicted and appear to minimize the energy transfer rate. The developmental
stage and location for the appearance of secondary ossification sites were predicted. The density
distribution and anisotropic principal material directions of bone in the femur were calculated.
The distribution of articular cartilage thickness and the sites of initial osteoarthritic changes in
the aged were also described.
Differing sensitivity to strain energy between individuals may explain variations in skeletal
physiology and may playa role in some disease states. A person who is "strain energy sensitive",
overweight, or very physically active would, according to our theory, tend to have higher density
bone. Persons who are relatively "strain energy insensitive" or less physically active would have
decreased bone mass but might be at a lower risk of osteoarthritis. The fact that idiopathic
osteoarthritis is uncommon in osteoporotic individuals lends credence to this hypothesis.
The versatility and power of the theory presented here suggests that it represents a
fundamental framework for understanding the role of mechanical stress in all chondro~osseous
biological processes. Application of this theory in the study of fracture healing and non~
union, growth plate biology, congenital skeletal disorders, bone remodeling in the response to
prosthetic implants, the influence of exercise on bone mass, and other areas of skeletal research
are appropriate.
References
1. Carter DR, Rapperport DJ, Fyhrie DP, Schurman DJ, and Orr TE: Trans Orthop Res Soc
11:451, 1986 and Orthop Trans, in press.
2. Chang CS, Pimbley WT, and Conway HD: Exp Mech 8:133-137, 1968.
3. Carter DR and Hayes WC: J Bone Jt Surg 59A:594-962, 1977.
4. Fyhrie DP and Carter DR: Trans Orthop Res Soc 10:377, 1985.
5. Bullough PG and Jagannath A: J Bone Jt Surg 24:311-325, 1983.
6. Whalen RT, Carter DR, and Steele CR: Trans N Amer Congress Biomech, 1986.
Acknowledgement: This work was supported by the Veterans Administration and NIH
Grants AM 32377 and AM 01163
 225

MECHANISMS OF CRACK PROPAGATION IN CORTICAL BONE

J.C. Behiri, W. Bonfield and B. Charalambides
Department of Materials, Queen Mary College, London, U.K.

1. INTRODUCTION
The anisotropic nature exhibited by compact bone has been well documented
In the literature. with differences in mechanical behaviour measured between
orientations parallel to (designated as longitudinal> or normal (designated as
transverse) to the long axis of the bone (1-5). Several investigators have
studied the influence of a variation in the direction of osteons in bone in
relation to the associated tensile. compressive and torsional deformation
properties (6-10). but only limited data has been previously reported on the
fracture characteristics as evaluated by J(c (the critical stress Intensity
factor) ( 11) and Gc (the critical strain energy release rate) as a function of
orientation.
The compact tension test method (12) has proved particularly advantageous
for a precise determination of the fracture mechanics parameters in that it is
possible to propagate a crack at various controlled rates (13-15) and has shown
that both J(c and Gc depend sensitively on bone density (16) and the velocity of
the propagating crack (15). Until recently (11) such experiments have been
restricted to a determination of longitudinal fracture. In the current investigation
the compact tension test method approach has been extended to evaluate not
only the critical stress Intensity factor but also the critical strain energy release
rate of compact bovine tibia bone at orientations of 0 0 . 15 0 .30 0 .450 75 0 and 90 0
to the bone axis. The Kc and Gc values obtained for the various orientations
are presented together with an assessment of the fracture morphology.

2. EXPERIMENTAL PROCEDURE
Details of the compact tension test procedure have been reported previously
( 12-15) . The critical stress Intensity factor (J(c) is defined as follows -

J(
C -
-
,;liyn
P (1)

where P is the load associated with a crack length (a)

B Is the specimen thickness
If is the specimen width
y 2 is a geometrical factor given by
1/2 3/2 5/2 7/2
Y2 = 29.6(i) - 185.5(i) + 655.7(i) -1017(j)
a 9/2
+ 638.9(1l) ( 2)

For the evaluation of the fracture characteristics at orientations other than the
longitudinal direction. it was found necessary to groove both side faces of the
compact tension specimens. so as to direct the crack in the desired direction
(Fig. 1) . Since J(c is largely dependent upon specimen geometry. altering
the geometry in this way would affect the equation describing J(c' An analysis
226

for a similarly grooved specimen has been reported (17) which has been
adopted here as a reasonable first approximation for the Kc equation (eq. S).
As the critical strain energy release rate Gc depends on the elastic energy
stored in the entire specimen (which is determined directly by experimental
measurements of compliance of given crack lengths. then the standard
equation 'or Ge Is not significantly affected by the grooving procedure (eq. 4).

I W

II
I
I
I
I

i
}---------.:.--
-$-
101

D=33mm

Ibl

Fig. 1. Schematic diagram of the compact tension specimen (al i11ustratin~

the v-notch as used for the orientation tests (bl.

p
- - - - - - - - - - - - - - Y2 (S)
(88 ) 1/2 W 1/2
n

p2 lie
Gc =-- (4)
2B lIa

Where Bn Is the grooved thickness (m) and e Is the experimentally determined

specimen compliance for any given crack length 'a'.
Bovine compact tension specimens were wet machined at several angles
(0 0 150 SOo. 45 0 75 0 . 90 0 ) to the bone axis from the mid diaphysis of adult
bovine tibia and were stored In Ringer's solution at -20 0 C until tested. The
test specimens. saturated with Ringer's solution were mounted In an Instron
testing machine and subsequently loaded at room temperature at a cross-head
speed of 1. 7 x 10- 7ms -1. As the crack propagated across the test
specimen. the progressive decrease of load (P) was monitored on the Instron
chart recorder while the extension of the crack length (a) was determined
with a microscope.
 227

I
~5

z
'E
:>:
-;:,4
'"
'"'~" 3
I 1 I I
~
1

o 10 20 40 50 60 80 90
Orientation (degrees)

Fig. 2. The dependence of critical stress intensity factor (Kc) on

orientation between 00 - 90.

3000

-2000
N
I
E
::?
<.:>u
<l>

I
Ol
a
~ 1000

0~0~~10~~2~0---3~0--~4~0--~5~0--~~~-=70~-780~~90~-
Orientation (degrees)

Fig. 3. The dependence of critical strain energy release rate (G c )

on orientation between 00 - 90.
228

Fig. 4. Scanning electron micrographs showing pullout of the fracture

surface for a specimen oriented at 15.

Fig. 5. Scanning electron micrographs showing pullout of the fracture

surface for a specimen oriented at 75.
 229

3. RESULTS AND DISCUSSION

Ti'iEID fElSU!ts of the present study shown in figures 2 and 3 indicate that the
critical stress Intensity factor (Kc) and the critical strain energy release rate
(Gc ) for bovine tll;)la specimens increase prOlgresslvely with a vl:ulation In
OIrient.atiOln from 001 to 90 0 to the Done .axis with limiting ratios of
.E!:c{iJo .. )/Kc{o") :2 and GC(!IIo,,)/Gc{o") 1.6. being coml:lI11rsble to those
wound by other Investigators for fracture toughness with for example
KC<!IIlJo)/Kc(O@)=1.7 (2).
Examination OIf the fracture surfaces revealed similar mlcrOlstructural features
at all orientations with the characteristic osteon pull-out especially evident for
fractures at relatively high angles to the long axis of the bons. figures 4 and
5 shOlw the difference in the fracture surface appearance from a fracture that
has been oriented at 150 to the iong axis to one that has propagated at 150
to the long axis. in figure 4 the crack is propagating mainly aiong cement
lines. with a relative absence of Intersecting obstacles. The relative flatness
of the fracture surface illustrates the ease with which the crack has
propagated. In contrast. figure 5. shows that as the incident ang!e of the
crack path increases relative to the long axis of the Done, tl'len ill
considerable obstruction to crack propagation Is provided by the alignment of
the Haversian systems. Considerable osteon pull-out Is observed Hlg.6)
wl'lici'l is .additicnally reflected in a decrease In the velocity of the propagating
crack.

Fig. 6. Considerable osteon pullout fro~ a specimen oriented at 75 to

the long axis of the bone.
230

ACKNOWLEDGEMENTS

The support of SERC through the Specially Promoted Programme in Medical

Engineering is gratefully acknowledged.

REFERENCES

1. Bonfield. W. and U. C. H .. Deformation and fracture of bone.

J. Appl. Phys .. 37. pp. 869-875 (1966).
2. Meivin. J. W. and Evans. F. G.. ASME Biomateriais Symposium. Detroit.
MI. (1973).
3. Margel-Robertson. D .. Ph. D. Thesis. University of Stanford. CA (1973).
4. Bonfield. A. and Datta. P. K. . Fracture toughness of compact
bone .J. Biomechanics. 9. pp. 131-134. (1976).
5. Bonfield. W. and Datta. P. K.. Impact fracture of compact bone In a
shock tube. J. Mater. Sci. .9. pp. 1609-1614. (1974).
6. Hirsch. C. and Da Sliva. 0.. The effect of orientation on some
mechanical properties of femoral cortical specimens"Acta. Orthop. Scand.
38(1) pp.45-56 (1967).
7. Pope. M. H. and Outwater. J. O .. Mechanical properties of bone as a
function of position and orientation J. Biomech .. 7. pp. 61-66 (1974).
8. Bargren. J. H.. Andres. C.. Bassett.l.. and Gjelsvik. A .. Mechanical
properties of hydrated cortical bone J. Biomech .. 7. pp. 239-245 (1974).
9. Reilly. D. T .. and Burstein. A. H.. The elastic and ultimate properties of
compact bone tissue. J. Biomech. 8. pp. 393-405 (1975).
10. Bonfield. W .. and Grynpas. M. D .. Anisotropy of the Young's modulus of
bone. Nature. 270. no. 5636 pp.453-454 (1977).
11. Bonfield. W.. Behiri. J. C.. and Charalambldes. B.. Orientation and
age-related dependence of the fracture toughness of cortical bone.
Fourth Meeting of the European Society of Biomechanics. Davos.
Switzerland. 196 (1984).
12. Bonfield. W.. Grynpas. M. D. and Young. A. J.. Crack velocity and the
fracture of bone. J. Blomech" 11. pp. 473-479. (1978).
13. Behlrl. J. C. and Bonfield. W.. Crack velocity dependence of longitudinal
fracture In bone. J. Mater. Sci. .15. pp. 1841-1849. (1980).
14. Behlri. J. C. and Bonfield. W.. Fracture mechanics of cortical bone.
Biomechanics: Principles and Applications (Edited by Hulskes. Van
Campen. D. and De Wljn. J. ). pp. 247-251. Martinus Nljhoff. The Hague
(1982) .
15. Behlri. J. C. and Bonfield. W.. Fracture mechanics of bone - the
effects of density. specimen thickness and crack velocity on longitudinal
fracture. J. Blomech .. 17. pp. 25-34. (1984).
16. Wrlght.T.M. and Hayes.W.C .. <l97n Fracture mechanics parameters
for compact bone - effects of density and specimen thickness
J. Biomechanics 10. 419-430 (1977).
17. Williams. D. D. and Evans. A. G .. A simple method for studying slow crack
growth. J. of Testing and Evalutatlon V. 1. p. 264 (1973).
 231

A BIOMECHANICAL ANALYSIS AT LAMELLAR LEVEL OF FEMORAL SHAFTS

DEFORMED IN BENDING
M~ Portigliatti Barbos*. S. Carando*. A. Ascenzi**.
A. Boyde*** and S. Improta****
Orthop.Clin.*. Univ.Turin. Italy; Depts.Human Biopathol.** and
Physics****. La Sapienza Univ . Rome; Univ.Coll.***. London.UK

INTRODUCTION
Recent investigations (Portigliatti Barbos et al. 1983, 1984, 1986; Boy-
de et al. 1984) provide evidence that along the femoral diaphysis the
longitudinal lamellae (whose fiber bundles and crystallites have a longi-
tudinal course and withstand loading by traction) and transverse lamellae
(whose fiber bundles and crystallites have a transverse course and with-
stand loading by compression) taken from osteonic and interstitial bone
both have a characteristic rotational distribution consistent with the
distribution of the bending forces normally operative in bone.
These results suggest that if the shaft of a long bone is deformed but
its osteonic structure is left intact, the distribution of longitudinal
and transverse lamellae in its osteonic and interstitial structures may
change, taking up a new arrangement better able to cope with the new dis-
tribution of forces acting on that bone. Some indirect evidence on this
point may be gained from the rearrengement of osteons in bones deformed by
overloading. This subject was examined by Lanyon et al. (1982) in an ex-
perimental investigation on mechanically adaptive bone remodelling in
adult sheep after removal of the ulna ahaft had induced a functional over-
strain of the radius.
The present investigation accepts these premises and aims to extend the
analysis of the cross sectional distribution of longitudinal and trans-
verse lamellae to the femur of an individual who had suffered during
childhood from severe rickets leading to deformation by bending.
MATERIALS AND METHODS
Preparation of the material
The material for this investigation was kindly donated to us by Dr. B.
Kummer, Professor of Anatomy at the University of Cologne (West Germany).
It consisted of two cylindrical fragments or samples about 3 em long sec-
tioned from the upper portion of the left and right femoral shafts of a
woman who had died at 63. The femurs were greatly deformed as a result of
asymmetrical antero-lateral curvature. The two bone specimens were taken
from the point where bending was at a maximum.
Macroscopic examination revealed marked changes in the bone samples.
They had an almost elliptical cross sectional shape with two elongated and
two very short opposite sides. As a result, the medullary canal also
showed an elliptical deformity. The thickness of the bone in the two elon-
gated sides had risen, while that in the two short sides had obviously
fallen.
From the two bone samples a series of undecalcified and plane parallel
sections, 100 micra thick, were prepared using a Leitz annular blade micro-
232

Fig. 1. Section prepared from the right (R) and the left (L) femurs, re-
spectively; p, posterior; 1, lateral.
tome. One representative section for the right (R) &,d one for the left
(L) ferm.ir have been shown in Fig. 1, and their orientation is indicated by
the letters p (posterior), and 1 (lateral). The sections clearly reveal
the almost elliptical deformation of the circumference of the femoral
shafts and offer a clear image of the changes that had occurred in bone
thickness.
In the right femur thickness is found to reach its maximum in the an-
terior part of the elongated lateral wall and in the posterior part of the
elongated medial wall; it reaches its minimum in the short anterior wall,
in the anterior part of the elongated medial wall and in the posterior
part of the elongated lateral wall. In the left femur the thickness of the
medial wall, although almost uniform, rises progressively in its posterior
part; thickness is maximal in the intermedial part of the elongated lateral
wall and minimal in the short anterior wall.
In both sections the short posterior wall appears to be thick &,d has a
pointed shape corresponding to the linea aspera
To conclude, a comparison between~two bone sections reveals that
they are not symmetrical with respect to each other, suggesting that fe-
moral bending did not occur in planes symmetrically oriented with respect
to the longitudinal middle plane of the body.
Quantimet 720 method
The basic procedure used by us for measuring the proportion of tran-
sversely oriented collagen has been described in detail in a previous pa-
per (Boyde et al., 1984). It uses circularly polarized light (CPL) as il-
luminating source and a quantimet 720 image ru'lalysing computer. An analysis
 233

of the results for a series of sections of one normal femur was reported
by Portigliatti Barbos et al. (1984) and for a controlateral femur by Por-
tigliatti Barbos et al. (1986).
In the present case, we have modified the technical procedure to solve
the special problems we have encountered.
Firstly the cortices of the two bones deformed by rickets were very thin
at some point. This meant that we could not sensibly plot the averages of
the results within 1860 x 1860 fields (nine 620 x 620 micra fields). We
therefore decided to halve the side of the measurement field to 310 micra;
for plotting purposes, we averaged the results for 16 of these fields.
Thus each square plotting symbol in Fig. 2 lies at the centre of an area
of 1240 x 1240 micra.
Due to this reduction in the size of the measuring field, we adopted a
different technique in assessing the ratio between CPL bright areas and
the total real area of bone. Any 310 x 310 micra field was excluded if
less than 4% of its area turned out to consist of CPL. A CPL detected
area was only recorded for fields in which there was a "hit". This over-
came the problem of fields only partially overlying bone tissue.
We found a difference in the average value of the CPL bright area for
the right (18.99%) and the left (12.39%) bones under identical detecting
conditions. In plotting Fig. 2 therefore, we have multiplied the stored
data for the left bone by 1.532 (18.99/12.39). As before, the size of the
plotting symbol increases in proportion to the CPL bright area detected,
i.e. the percentage of transversely oriented collagen present. The size of
the square was obtained from the expression k(%CPLO.8).

RESULTS
Proceeding now to an examination of the distribution of transverse lamel-
lae with respect to longitudinal ones in the femoral cross-sections, the
diagrams in Fig. 2 must be considered. They have been recorded from two of
the most representative sections, one for each side.
In the two diagrams the ratio between the surface area of transverse
(bright) lamellae and that of longitudinal (dark) lamellae has been sym-
bolized geometrically by squares. The larger the side (or surface) of the
square, the greater the proportion of bright lamellae present. By giving
each square a position corresponding to that occupied in the bone section,
a diagrammatic but morphologically accurate representation is given of an
entire section, with visualization of the overall ratio between the two
types of lamellae.
In Fig. 2, the diagram obtained from the section of the right femur (R)
shows that the distribution of transverse lamellae with respect to longi-
tudinal ones is not homogeneous. Transverse lamellae are concentrated
within the lateral and medial walls of the femoral diaphysis. In the me-
dial wall, the highest concentration of transverse lamellae with respect
to longitudinal ones is found in the inner and posterior sector where the
wall reaches its maximal surface area. From this site the high concentra-
tion of transverse lamellae continues in the posterior wall of the sectio~
as well as in the anterior part of the medial wall itself. From here, the
high concentration of transverse lamellae drops so that in the adjacent
anterior wall the percentage of transverse lamellae is lower than that of
longi tudinal ones. When all the points in the medial wall in which trans-
verse lamellae reach tIle highest incidence are considered globally, the
evidence points to a tendency for transverse lamellae to be arranged
along straight bands or alignments distributed in an antero-posterior di-
rection.
234

a a I
/
cOCO / a a a a 0 I
00000 / DccaaD / /
cODOOCD I c /
" DO C DOC a / /
" 0 DODO C C D I /
b'l..CODDDacc/a /
oDDDDcric /
DO c ItJD Pcp
o A:l
/ 1:1 Ca
", I /DDDc
,,~ / '/ cOD a
"" V a DOD a
ODCa
G _DOc a
c rnc
DOocca
C a

oOCDa.

D
,~~' c
oDD
o oCOca
O~aOC
DUJODD 0

Fig. 2. Diagrams showing the distribution of transverse and longitudinal

lamellae in the sections prepared from the right (R) and left (L) femurs,
respectively. The directions Gb and Gc correspond to the maximum values of
compression and tension stresses.
In the sites where the incidence of transverse lamellae is high, there
is a recovery in the concentration of longitudinal lamellae toward both
the inner and the outer circumferential edges, but this increase is not
abrupt at the level of the postero-medial outer circumference of the sec-
tion, where numbers of transverse lamellae appear to fall slowly.
In considering now the lateral wall of the same diagram in Fig. 2, it ap-
pears that the highest concentration of transverse lamellae with respect
to longitudinal ones is found in the anterior sector, where the surface
area of the section reaches its maximum. However, in contrast with the pos-
terior portion of the medial wall, here the highest concentration of trans-
verse lamellae is reduced to few, scattered centers. From the sites of
highest incidence the concentration of transverse lamellae falls away to-
ward both the outer and inner circumferential edges, but this fall seems
sharper and steeper in the inner than in the outer edge. On the other hand,
in the posterior portion of the wall, whose surface area is much lower,
the incidence of transverse lamellae appears to fall progressively.
As regards the longitudinal lamellae, these are most frequent in the an-
terior side and in the posterior portion of the lateral side.
In the right side of Fig. 2 there is a distribution diagram for trans-
verse lamellae with respect to longitudinal ones recorded from the section
 235

prepared from the left femur (L). In this diagram too the relative dis-
tribution of the two types of lamellae is not homogeneous, and transverse
lamellae appear to predominate mainly in the lateral and medial walls. As
regards the medial wall, there is a high concentration of transverse lamel-
lae in the inner side where the evidence points to a tendency for the same
lamellae to be arranged along straight bands or alignments distributed in
an antero-posterior direction. Toward the external edge of the same site
numbers of transverse lamellae fall abruptly as numbers of longitudinal
ones rises. At the boundary between the medial and anterior wall there is
a slight decrease in the incidence of transverse lamellae, which reaches
its minimum in the lateral portion of the anterior wall. In the posterior
part of the medial wall at the boundary with the posterior wall concentra-
tions of transverse lamellae fall progressively, although they still re-
main relatively dominant. Only toward the external edge of this last site
is there a clear fall in numbers of transverse lamellae.
In the lateral wall a high concentration of transverse lamellae is ap-
preciable in the inner zone, with a preference for the anterior portion,
where the surface area of the section is at its maximum. However, in con-
trast with the medial wall, the transverse lamellae are not compactly ar-
ranged but appear as few scattered centers. Numbers of the same lamellae
falloff sharply in the following three sites: the posterior portion of
the wall, where the surface area of the section becomes much smaller; the
boundary between the lateral and anterior wall; and the external side of
the lateral wall. Conversely, in the inner edge of the lateral wall hardly
any fall in numbers of transverse lamellae is appreciable.
DISCUSSION
In critically exarrllnlng the results of the present investigation two
main topics call for consideration: first, the changes in the morphology
and dimensions of the sections prepared from the portion of maximunl bend-
ing of the two femurs, and, second, the ralationship between these macro-
scopic changes and the distribution of longitudinal and transverse lamel-
lae in the bone sections.
As regards the first topic, the first attempt to explain the macroscopic
changes occurring in a bended femur was made by Benninghoff (1927), who
used a descriptive morphologic method. According to this author, the femur
loses its circular shape in acquiring an elliptical section whose main
axis lies in the bending plane in order to provide support for the de-
formed bone on its concave side. In other words, the elongated section
could function as a prop capable of compensating for the bending of the
femur.
A sounder, more logical explanation for the deformed shape of the sec-
tions corresponding to the portion of maximum bending in a femoral diaphy-
sis was subsequently put forward by Pauwels (1954, 1968, 1979). This
author applied the most general procedure that has been developed for any
bent material in which the plane of application of a force, including
forces acting on the bone axis, can be oriented in any direction whatever.
In the case of a bent femoral diaphysis, the bending plane and the plane
of application of the force do not necessarily coincide and the bending is
said to be "distorted", according to the translation of the French term
"gauche" used by Pauwels. By using this procedure Pauwels was able to pre-
dict the deformation of femoral sections at the point of maximum bending
as a function of the orientation of the bending plane.
An approach to the second topic has been attempted by developing a pro-
cedure based on the theoretical guidelines proposed by Pauwels in order to
236

establish if a correlation exists between the directions of maximum re-

sistance to compression and traction of the sections and the distribution
of transverse and longitudinal lamellae. The mathematical developments of
such a procedure being too long to be reported here, a graphic represen-
tation of the results is shown in Fig. 2. The directions Gb and Gc, cor-
responding to the maximum calculated value of compression and tension
stresses, are given together with the limits of variability (shown by
dotted lines). It appears resonably obvious that in two bone sections -
but specially in that obtained from the right femur - the directions cor-
responding to the maximum stress in compression are related to surface
~eas mainly composed by transverse lamellae and conversely the directions
corresponding to the maximum stress in traction are related to surface
areas mainly composed by longitudinal lamellae.
This is a first approach for correlating the mechanical properties of a
bone section at macroscopic level with its microscopic lamellar structure.
However it is beyond doubt that other correlation procedures may exist.
So, we are now attempting to find a new approach to the problem.
In concluding, when the here discussed results are translated into mor-
phological qualitative terms, they imply that both the macroscopic con-
figuration of the section and its microscopic structure at the level of
osteons and lamellae are governed by the distribution of the forces ac-
tive in bone. In other words, even pathologically deformed bone provides
evidence that its microscopic structure appears to be related to the mech-
anical forces acting on bone.
REFERENCES
1. Benninghoff v.: Ueber die Anpassung del' Knochenkompakta an geanderte
Beanspruchungen. Anat. Anz. 63: 289-299, 1927.
2. Boyde A, Bianco P, Portigliatti Barbos M, Ascenzi A.: Collagen orien-
tation in compact bone: I. A new method for the determination of the
proportion of collagen parallel to the plane of compact bone sections.
Metab. Bone Dis. & ReI. Res. 5: 299-308, 1984.
3. Lanyon LE, Goodship AE, Pye CJ, MacFie JH: Mechanically adaptive bone
remodelling. J. Biomechanics 15: 141-154, 1982.
4. Pauwels F: Die statische Bedeutung del' Linea aspera. Viertel' Beitrag
zur funktionellen Anatomie und kausalen Morphologie des StUtzapparates.
z. Anat. Entwickl.-Gesch. 117: 497-503, 1954.
5. Pauwels F: Beitrag zur funktionellen Anpassung del' Corticalis del'
R8hrenknochen. Untersuchung an drei rachitisch deformierten Femora. 12.
Beitrag zur funktionellen Anatomie and kausalen Morphologie des Stiltz-
apparates. Z. Anat. Entwickl.-Gesch. 127: 121-137, 1968.
6. Pauwels F: Biomecanique de l'appareil moteur; Contributions a l'etude
de l'anatomie fonctionnelle. Springer-Verlag, Berlin, 1979.
7. Portigliatti Barbos M, Bianco P, Ascenzi A: Distribution of osteonic
and interstitial components in the human femoral shaft with reference
to structure, calcification and mechanical properties. Acta Anat. 115:
178-186, 1983.
8. Portigliatti Barbos M, Bianco P, Ascenzi A, Boyde A: Collagen orienta-
tion in compact bone: II. Distribution of lamellae in the whole of the
human femoral shaft with reference to its mechanical properties. Metab.
Bone Dis. & ReI. Res. 5: 309-315, 1984.
9. Portigliatti Barbos M, Carando S, Ascenzi A, Boyde A: On the structural
and biomechanical symmetry of human femurs. Bone, 1986 (in press).
 237

BENDING STRENGTH OF OPERATIVELY STABILIZED FRACTURES UNDER THE

INFLUENCE OF BISPHOSPHONATES
P.C. Kramers, B.A. Rahn, J. Cordey and S.M. Perren
Laboratory for Experimental Surgery, Davos, Switzerland

SUMMARY

The bisphosphonate HEBP (EHDP), an inhibitor of mineraliza-

tion and to some extent of bone resorption, may inhibit bone
healing clinically. A double blind study in dogs investigated
HEBP-dependent effects on direct bone healing and recovery
from such influences. Dosages were 5.0, 0.5, 0.1 and 0.0
mg/kg/day s.c., given for 20 weeks, the fracture occurring af-
ter eight weeks. Radiography, mechanical testing and histolo-
gical techniques were employed for assessment of drug effects.
At 0.1 and 0.5 mg/kg/day s.c. HEBP had no significant effect
on fracture healing in dogs after rigid plate fixation. The
latter dosage corresponds approximately to the oral dose of 10
mg/kg/day in humans. The excessive 5mg/kg/day s.c. inhibited
fracture healing in dogs. This effect is reversible. Eight
weeks after drug withdrawal gross mineralization defects still
diminished the mechanical strength at the fracture.

IRTRODUCTIOR

The bisphosphonate HEBP, a structural analogue of Pyrophos-

phate resistant to hydrolysis, is a potent inhibitor of mine-
ralization processes and to a lesser extent of bone resorption
(21,4). (HEBP, previously EHDP = 1,hydroxy-1,1-ethylidene-bis-
phosphonate, Procter & Gamble, Cincinatti, Ohio). It is clini-
cally used (20) for the treatment of Paget's disease of bone,
the prevention of pathologic calcification and, diagnosti-
cally, as a bone seeking carrier in scintigraphy (9). Frac-
tures of diseased or deformed bones are a frequent compli-
cation of Paget's disease, and often rigid fixation techniques
are used for their stabilization (12,11). Fracture healing is
reported to be normal in patients who were on HEBP-therapy at
the time of fracture (11). There is also discussion about the
concept of pretreating patients with Paget's disease before
elective surgical procedures (e.g. corrective osteotomies) to
decrease the hypervascularity associated with the disease
(11). Nevertheless, immediate withdrawal of HEBP-medication in
the event of a fracture is strongly recommended on theoretical
grounds, and the problem is unresolved, whether HEBP treatment
may be continued during the period of fracture healing. In a
double blind study in the dog we investigated the effects of
HEBP on direct (primary) bone healing, a type of healing en-
tirely dependent on internal bone remodelling.
238

MATERIALS A.D METHODS

HEBP was administered subcutaneously to 24 beagles at dose

levels of 5.0, 0.5, 0.1 and 0 mg/kg/day. After 8 weeks of drug
administration a standardized fracture of both radii was crea-
ted with three-point bending, using an incomplete osteotomy of
standardized depth as a stress raiser. The resulting trans-
verse or short oblique fracture was stabilized with a precon-
toured plate (Synthes 244.08) applying compression. Half the
dogs were killed 12 weeks after fracture (Series 1), the other
half (Series 2) were kept alive for an additional drug-free
period of eight weeks to assess the mode and degree of
recovery from drug effects. The present paper presents the
assessment of drug effects by mechanical testing of the radii
at the end of each series. Statistical evaluation employed the
Student-Fisher-t-test (P=O. 05), the Kruskal-Wallis test
(P=0.05) and the Friedman test (P=0.05)(Dr.C.Baldauf, lnst.
Dr.W.Ziegler, Basle). Experimental permit: GR 1.2/1983.
After euthanasia, the plates were transversely cut at mid-
plate and reattached to the bones with six of the original
screws. The most distal and proximal screw holes were incorpo-
rated into methyl methacrylate molds fitting the testing
machine. During the preparation procedure the frozen bones
were swathed in gauze moistened with Ringer's solution. Prior
to testing all the bones were slowly thawed. The bones of one
series were all tested within four hours of one another. The
order of testing was randomized. The prepared bones were
clamped in the bending device (Microtron, Russenberger MUller,
Schaffhausen) with the re-attached halved bone plate always on
the tension side during the test. The mechanics of the
experiment were designed to apply a four-point bending load,
constantly registering the applied force against the deforma-
tion of the bone until fracture occurred. A constant loading
speed of 20 mm per minute, with a bending arm of 50 mm on each
side of the bone, was selected. To record the deformation
(strain or angular displacement) of the bone during the test,
a thin flexible bronze plate instrumented with strain gauges,
was attached to two pegs 20 mm apart (L), which could easily
be clamped to the bone perpendicular to the loading plane
(Fig.1). Thus, the strain gauges underwent the same angular
displacement as the bone, allowing a stress/strain gr"aph to be
recorded (Fig.1). The angular displacement of the bronze plate
showed linear characteristics on calibration in the range of
measurement. (100 microstrains = 1,665 degrees of angulation
or bending of the bronze plate in the range of measurement)
The resulting stress/strain graphs were evaluated for
strength and bending (angular displacement) of the bone at
the maximal load point (Mmax; ~max., point immediately prior
to fracture), as well as at 70% of the ultimate strength
(M70%; <j>M70%), an area where the slope of the curved graph was
s t i l l in steep ascension and not yet levelling off into
plastic deformation. The specimen's bending stiffness (El) was
calculated at 70% of the ultimate strength:
E M70% L 2
I = ~ = bwnding stiffness at M70% (Nm , , in rad)
(M 70 %)
 239

M
(111m) max. load point
op(MmaxJ

Fig.t Four-point bending test (schematically). A strain gauge

transducer was clamped to the bone to record angular
displacement during loading. The halved plate was placed on
the tension side. The bone was loaded at a speed of 20 mm per
minute until it fractured. It always failed at the original
fractW'e site. On a two axis recorder the angular displacement
(') ( or I.'e.) was plotted against the applied bending moment
(M)(Nm). The maximal load point and the point at 10% of the
ultimate load were the basis of our calculations (see text).

After 12 weeks all fractures in the 0, 0.1 and 0.5 mg/kgl

day gr6up were radiographically healed. In.no case was forma-
tion of significant periostal callus at the fracture observed.
In the 5 mg/kg/day group a widening of the fracture gap was
evident, and only 1 of 12 fractures was united. At this stage
the bending test demonstrated a degree of stiffness in the
contr61 group, that was not statistically different in the 0.5
and O. 1mgl kgl day groups. A significantly lower (p<0. 01) and
only minimal stiffness was present in the 5 mg/kg/day group,
where solid union was inhibited (Fig.2). After the "recovery"
period all the radii were united or were healing radiographi-
cally. Mechanical stiffness had significantly increased in all
groups (p<O.05), however, the radii from the 5 mg/kg/day group
tended to be weaker than those from the other groups (p=0.06,
the single deviant ~alue was not considered)(Fi&.2).
Histologically there was no difference between the 0.1 mg/kg
and the control group, whereas with 0.5 mg/kg/day very minute
mineralization defects were seen in the microradiographs. With
HEBP at 5 mg/kg/day bone mineralization was completely inhi-
bited during treatment, and was resumed after drug withdrawal
with a delay of about 2 weeks. Flagrant mineralization defects
were still visible eight weeks after drug withdrawal. Bone re-
sorption was not visibly reduced in any group.
240

Fig.2 Bending
EI Bend ing St iff ness stiffness of hea-
ling fractures at
4 twelve weeks (Se-
ries 1), and after

I
* a drug- free reco-
very period of
eight weeks (Se-
2
t ries 2). With HEBP

Itt at 0.1 and 0.5

mg/kg/day results
did not differ
from the control
o i ---L ---L ---L ---L ---L ~ HEBP group (Student Fi-
5 0.5 0.1 0 5 0.5 0.1 o sher t-test). At 5
(mg/kg) mg/kg, HEBP com-
Series I Series II pletely (1)<0.01)
inhibited solid u-
nion, as is shown at 12 weeks (Series 1). Two bones were too
weak even to be tested. After the drug free period, recovery
in this group was still incomplete (p=0.06, when the single
deviant high value (*) is not considered). At this time the
stiffness in the 5 mg/kg/day group corresponded to that of the
other groups of the first series, eight weeks earlier.

DISCUSSIOW

Canine bone structure is similar to that of humans. The in-

fluence of various bisphosphonates have been extensively stu-
died in the dog at multiple dose levels (2,5,6,7,8, 10,11,19).
Direct bone union is possible in the dog within 12 weeks (19).
As the healing process is not completed at this time, diffe-
rences due to drug influence can be readily visualized. The
drug-free period (Series 2) was added to observe the potential
for recovery, because ectopic mineralization has been seen
after withdrawal of the drug (1,16,23).
Radiographically, there is no doubt that the high HEBP dose
of 5 mg/kg/day s.c. seriously impairs bone healing in the dog,
in contrast to the 0.5 and 0.1 mg/kg doses. This confirms pre-
vious conclusions (5). Whereas direct (primary) bone healing
does not seem affected by the 0.5 mg/kg/day dose, indirect
(secondary) healing is, reportedly, delayed somewhat (16).
This might reflect the influence of motion, however small, on
the formation of mineralized tissues.
Testing the mechanical properties of healing fractures pre-
sents some difficulties. Bone is more resistant to compression
than to tensile forces. Yet, i t is virtually impossible to
perform a purely tensile mechanical test, as the tensile axis
of the bone will rarely coincide with the centre of the cross
sectional area. The superposition of a bending moment on the
tensile forces applied does not allow an accurate assessment
of the mode of testing. Therefore, we chose a four point ben-
ding test with an experimental design where the bending
moment between the two pairs of bending forces is constant
even though we are not testing an ideal beam(18). The angular
displacement measured includes deformation of the fracture
 241

site and of the two adjacent bone-plate composites. As the

contribution of the bone-plate composite to total deformation
is similar in all groups, differences in overall stiffness are
indicative of differences in fracture site stiffness.
The radii were prepared for mechanical testing with the
sectioned plate in situ, to protect the screwholes and the
rest of the bone from stress, while the weakest point of the
tested "beam", the fracture area, was subjected to stress. The
bending direction with the plate on the tension side was' cho-
sen, because, directly under the implant the bone is normally
weakened by remodelling due to plate-related vascular injury.
Assuming bone remodelling to be compromised by the drug, alte-
rations in bony union would be most significantly demonstrated
in this manner. Resorption was not visibly affected.
The stiffness of the bones could be reliably determined at
the precisely defined 70% of the maximal load (ultimate
stress), rather than at the starting point of the curves. At
this point the curves were still in steep ascension (elastic
deformation) giving the values a close approximation to the
stiffness at O. At the HEBP dose of 5.0 mg/kg/day a complete
absence of mineralization caused the complete lack of mechani-
cal strength in the fracture area. This effect was partly re-
versed during the eight week "recovery" period, and the amount
of stability achieved was quite surprising, because histologi-
cally, most of the newly mineralized osteons appeared to be
"suspended" in unmineralized osteoid sheaths. In contrast to
indirect fracture healing (16), direct fracture healing was
not impaired by 0.5 mg/kg/day in the same time period in our
study. This supports the assumption that mechanical protection
is prerequisite for mineralization processes (22).
The chosen dosages proved to be well suited to the aim of
this study. The influence of severe overdosing was clearly de-
monstrated at 5.0 mg/kg/day. At 0.1 mg/kg drug effects could
not be detected, indicating that we were below a minimum ef-
fective dose, as planned. The 0.5 mg/kg/day group is interes-
ting, in that the mechanical stiffness of bony union at 12
weeks was, if anything, slightly higher than that of the con-
trol, whereas first drug effects became visible histologi-
cally, in the shape of minute mineralization defects and the
rare occurrence of slightly widened osteoid seams. These chan-
ges were minuscule, yet it is at this dose level that occasio-
nal spontaneous fractures, which heal with a normal radiologi-
cal appearance, are reported in dogs after 12 months of HEBP
administration(6). The slightly higher stiffness at 0.5 mg/kg
(Series 1) could not be explained morphologically.
The oral dose recommended for extended use in humans corre-
sponds to a daily systemic dose of 0.05 - 0.3 mg/kg (1-3 % ab-
sorption of 5-10 mg/kg/day p.o.) (6,14), and no generalized
adverse effects on the ratio of osteoid to mineralized bone
have been reported with this dose within 3-6 months. Neverthe-
less, widening of osteoid seams has been seen in humans under
prolonged intake of a systemic dose of 1.2 mg/kg/day (6% ab-
sorption of 20 mg/kg/day p.o.)(15,3). Osteoid seam widening
was reported after three months in dogs receiving doses not
lower than 2 mg/kg/day s.c.(6). A slight increase in osteoid
seam width was occasionally observed at 0.5 mg/kg/day in this
242

experiment, possibly due to a local accumulation of the drug

in context with the local increase in bone remodelling after
the fracture. When calculating systemic doses, one must consi-
der that the actual drug concentration will be higher at the
target site when the increase in bone turnover is limited to
only few focal sites, than when a patient has widespread di-
sease( 13). In our fracture model bone turnover is only increa-
sed in the vicinity of the fracture when compared to the remo-
delling rate of the intact tibia. We conclude that the drug
concentration at the fracture site is actually higher than the
systemic dose level in our experiment. Based on the aboNe, we
believe i t is a valid assumption that the dose of 0.5 mg/kg/
day s.c. in the dog corresponds to if not slightly exceeds the
oral dosage of 10 mg/kg/day in humans.

1. HEBP at 0.1 mg/kg/day s.c. had no influence on fracture

healing in dogs. 2. HEBP at 0.5 mg/kg/day s.c. showed no me-
chanical effect in this experiment. This dose corresponds ap-
proximately to the recommended therapeutic dose of 10 mg/kg/
day p.o. for pagetic patients. Therefore, direct fracture hea-
ling under rigid plate fixation may be expected to occur in an
approximately normal manner at that dosage. 3. At the exces-
sive dose of 5 mg/kg/day s.c. fracture healing is inhibited in
the dog. This effect is reversible, although eight weeks after
drug withdrawal the bending stiffness of the "recovering" bone
is still diminished due to severe mineralization defects.

REFERENCES

1. Bijvoet,OLM et al: Acta Orthop Scand, 45: 926 1974

2. Cabanela,ME, Jowsey,J: Clin Orthop 100: 364-369 1975
3. Finerman,GA et al: Clin Orthop 120:115-124 1976
4. Fleisch,H: Symposium CEMO IV, Diphosphonates and Bone 1981
5. Flora,L: Symp.CEMO IV, Diphosphonates and Bone 83-1151981
6. Flora,L et al: Metab Bone Dis 2:389-407 1980
7. Flora,L et al: Calcif Tiss Res 27(Suppl):47 1979
8. Flora,L et al: Calc Tiss Res 24(Suppl):31 1977
9. Fogelman,I, McKillop,JH: Symposium CEMO IV, 230-250 1981
10.Hassing,GS et al: Calc Tiss Res 24(Suppl):36 1977
l1.Hassing,GS et al: Calc Tiss Res 27(Suppl):47 1979
12.Johnston,CC et al: Clin Orthop 172:186-194 1983
13.Kanis,JA: Calc. Tiss Int 36:629-631 1984
14.Khairi,MRA, Johnston,CC: Clin Orthop 127:94-105 1977
15.Khairi,MRA et al: Ann Internal Medicine 87:656-663 1977
16.Lenehan,T et al: J. of Orthopaedic Research 3:499-507 1985
17.Milgram,JW: Clin Orthop 127:63-69 1977
18.Popov,EP: Mechanics of Materials, Prentice Hall Inc., 1976
19.Rahn,B et al:Symp.CEMO,Diphosphonates and Bone, 423-27 1981
20.Russell,RGG et al: Symposium CEMO IV, 251-263 1981
21.Schenk,R et al: Calc Tiss Res 11: 196-2141973
22.Schenk,R et al: Symposium CEMO IV, 117-131 1981
23.Thomas, BJ, Amstutz, HC: J Bone Joint Surg (Am) 67:400 1985

* Supported by Procter & Gamble Co., Cincinatti,Ohio

 243

HOLOGRAPHIC INVESTIGATION OF MECHANICAL PROPERTIES OF TIBIA

FIXED WITH INTERNAL FIXATION PLATE
A. KOjima*, R. Ogawa*, N. Izuchi*, T. Matsumoto**, K. Iwata***
and R. Nagata***
Kansai Medical University*, Osaka Prefectural Technical
College** and University of Osaka Prefecture***, Osaka, Japan

1. INTRODUCTION
Osteosynthesis with metal plate and screws is a common treat-
ment for the fracture of the long bone. It is advised that each
hole of a plate should be used for fixation with screw, for
example, a plate with eight holes is used with corresponding
eight screws. However in a limited case, some holes are not oc-
cupied with screws because of types of the fracture, such as a
comminuted fracture or a open fracture with bone defect. In the
post operative course of such cases, complications of bending,
torsion or breakage of the metal plate are not rarely seen.
By the way, the holographic interferometry can measure the
deformation of the object without contact, without fracture and
wi th high accuracy 1 ) . In this paper, mechanical properties of
fixation between the fractured part of the tibia and the metal
plate are investigated under some simulated loading conditions.
Deformation measurements of the bone and the metal plate were
performed by double exposure holographic interferometry. Fol-
lowing three types of specimens were prepared to measure the
deformations, that is : (1 )the normal human tibia, (2)the tibia
to which osteosynthesis was tried by using AO dynamic compres-
sion plate having eight holes with eight screws, after its
central part had been
cut artificially in- HM
stead of a transverse hiNe Laser r-------I'\."->r"+=~
fracture of the tibia,
(3)the tibia in which o : Tibia
IL : Illuminating werve
two screws near the
R : Reference wave
fractured line of the
tibia(2) were removed. P : Plate o
First, the deforma- M : Mirror
tion of the normal HM : Half Mirror R
human tibia (1) was L : Lens M
measured to compare CL: Cylindrical lens
with the specimens with o : Density filter
(3))
1
frac yred part ( (2) and
. Next, the tibia FIGURE 1. Schematic diagram of
of specimen (2) was optical system.
244

measured in the same way as in the first experiment and the

mechanical properties among the three kind of specimens were
investigated. Lastly, the deformations of the specimens sub-
jected to the torsion were measured and then comparison about
the deformations obtained among the three specimens was
performed.

2. EXPERIMENTAL METHOD
Figure 1 shows schematically the optical system used in the
experiment. The light source is a 50 mW He-Ne laser. The light
wave is divided into the reflected (R) and transparent waves
(IL) by a half mirror (HM). The former directly illuminates the
holographic plate (P) as the reference wave (R). After the lat-
ter was collimated by the lens system, it illuminates the tibia
(0). The light wave reflected on the surface of the object ar-
rives at the plate. The specimen is loaded to a predetermined
place and direction between the first and second exposures.
After the holographic plate was developed, it is returned to
its original position. Virtual images of the tibia before and
after deformations were observed at the same time from the ob-
servation point through the plate. Adjacent interference
fringe obtained in the sys~Tm correspond to the displacement to
the out of plane of 0.3pm.
The specimens were dried (a)
human tibias. The proximal end
of tibia was embedded in super
hard plaster . The specimen was
subjected to bending moments in
the anterior, posterior, later-
al and medial directions with tibia weight
respect to the screw fixed to
the center on the distal end of
the tibia. The load was applied
by using string and weight
shown in figure 2-a. One end of
the string was fixed on the end
of tibia and the another end
was pulled by the weight of FIGURE 2. Method to apply
10gf-100gf. Clockwise and coun- bending moment(a). and
terclockwise torsions around torsion(b) to the dlstal
the long axis of the tibia were end of the tibia.
applied at bilateral screws
fixed to its distal end (Fig.2-b). The tibia was loaded to a
predetermined direction between the first and second exposures.
After the holographic plate was developed, it was returned to
its original position and the reconstruction images were
observed. Holograms were recorded from anterior and lateral
views of the tibia. Deformation measurements of three kind of
specimens were performed for bending and torsion moments. Their
results were quantitatively compared and investigated .

3. RESULTS AND DISCUSSIONS

3.1. The deformation measurement of the normal tibia (specimen
~
Figure 3 shows examples of the reconstruction images observed
from the anterior (a),(b) and lateral (c),(d) views of the
 245

tibia The initial and increased weights were 50gf and 20gf ,
respectively. The density of fringes increased gradually from

FIGURE 3. Reconstruction
images observed from the
anterior(a),(b) and lateral
(c) , (d) views of the normal
tibia subjected to the
various bending moments.
(a),(b),(c) and (d) cor-
respond to the anterior,
posterior, lateral and medial
bending moments,respectively.

FIGURE 4. Displacement dis-

tributions (a)-Id) of the
tibia obtained from Fig.3,
O,L-~~~~~~H~--~------~ respectively.
location (%)

FIGURE 5. Reconstruction
images of the specimens (1)-
(3) subjected to the lateral
bending moment, respectively.
(50gf-->70gf)

specimen (n /
(2)

(3)

fracture

+L
FIGURE 6. Displacement dis-
I---o::::......------;oo_------.....,_~ tributions of the specimens
iocatioo (%) (1 )-(3). (50gf--> 70gf)
246

the proximal ends to the distal ends in all cases. Figure 4

shows displacements calculated from the interference fringes
obtained in figure 3. The total amount of deformations of the
tibia in lateral and medial directions are nearly the same (
Fig.4-c and d). Anterior and posterior deformations are also
nearly equal(Fig.4-a and b). However the amount of deformations
in the lateral direction was about twice in the anterior
direction. This means that the strength of the tibia in the
anterior and posterior directions is stronger than the strength
in the lateral and medial directions with respect to the bend-
iI?-g moment.
3.2. The deformation of the specimens (1 )-(3) subjected to the
bending moment.
Figure 5 shows the reconstruction images that were observed
from the lateral view of the tibia. The measurement results in
figure 5 are shown in figure 6. The direction of force is lat-
eral. The density of fringes increased gradually from the prox-
imal to the distal ends (Figs. 5-a and 6-(1 I). For the specimen
(2), it is found that the concentration of fringes on the dis-
tal fracture part of the tibia was more than that on the prox-
imal part (Figs. 5-b and 6-(2. For the specimen (3), a more
greater concentration of fringes in the distal fractured part
of the tibia was observed (Figs. 5-c and 6-(3. This means
that the slope of displacement changed remarkably at the frac-
tured line. Thus, the ability of the plate system to fix tibial
fragments becomes weaker than the results obtained in the spec-
imen (2). The amount of total deformation of specimen (3) in-
creased especially in comparison with them of the other
specimens.
Figure 7 -a shows enlarged. image'
of the deformations of the bone and ...... Sone
the plate in the neighborhood of -0- Plate

the fracture in figure 5-c. On the

both ends of the plate w~ere the
plate is well fixed to the bone by
the screws, the fringes on the
pIa te coincide with the them on . . ..... f ractLireci liM
the bone. However, on the center of
the plate where the screws were not
used for fixation , the fringes on
the plate and bones do not coincide
with each other. This means that
the large gap in the deformation
between plate and bones occurred at
the center of the plate where the
screws were not used for the
fixation. This results shows the (a) (b) displacement
strength of fixation of the plate FIGURE 7. Enlarged image
system decreased remarkably due to (a) near the fracture in
the lack of screws near the frac- Fig.5-(c) and displace-
tured line. Figure 8 shows the ment distributions of the
reconstruction images that were. pIa te and the bone (b)
observed from the anterior view of (50gf--70gf)
the tibia. The measurement results
 247

of figure 8 are shown in figure 9. The direction of force is

anterior. The tendency of the deformation resembles to the
resul ts that described in the experiment of figures 5 and 6.
The amounts of lateral deformation increased remarkably in com-
parison with the anterior deformation under the same load.
Considering the inclinations of displacement at the fractur~d
line for the specimen (3) in figures 6-(3) and 9-(3), the in-
clination of the former is much more than it of the latter, al-
though the inclination of the former about the specimen (2) is
not so much as the latter (Figs.6-(2) and 9-(2. Therefore, if
the two screws were removed from the plate, it is found that
the fixation force of the plate becomes very weak, especially
in lateral direction.

Q.'
\ss:= specimen (1)

;;"
~ fIA -v-
i
.
:I
@
j

,...3.2

!
lo~~-----=
! location (%)

FIGURE 8. Reconstruction FIGURE 9. Displacement distribu-

images(a)-(c) of the speci- tions of the specimens (1 )-(3).
mens (1)- (3) subjected to
the anterior bending moment,
respectively. (50gf-->70gf)

3.3. The deformation of the tibia subjected to the torsion.

Figure 10 shows the deformations of the specimens subjected
to the torsion. As shown in figures 10-b and c, the fringes on
the distal fragment of tibia are cut off at the fractured part.
Especially, in figure 10-c, there was the displacement of about
1pm under clockwise torsional moment of only 100 gf cm. These
results show that the plate system becomes weaker for torsional
moment if the screws of the central part were removed. Tor-
sional deformation is also shown in figure 11-a, which is the
reconstruction image of the tibia subj ected to the anterior
bending moment. The inclinations of the fringe on the distal
fragment and the proximal one face the opposite direction. A
possible explanation for this phenomenon is shown in figure 11-
b. Because the axis of the tibia does not coincide with the
axis of the plate , anterior force to the both ends of the
tibia produces two types of torsional moment force in the op-
posi te directions on the border of plate It was found that
anterior force acts to the specimen as the torsional moment.
248

~
forceW1 .
I. lockw~se
: torsion
i

.~
I .

I i
counterclockwise
torsion): \

c0,
. . .~..-.)
FIGURE 10. Reconstruction FIGURE 11. Reconstruction image!
images(a)-(c) of the spec- a) of the specimen (3) subjected
imens!1 )-(3) subjected to to the anterior bending moment
the torsion. a),(bj--> and its schematic diagram to
200gf cm, (c)-->100gf cm) explain the deformation(b).

4. CONCLUSION
Following contents are concluded on the basis of above
experimental results.
1. Deformations of the normal tibia and the tibia with
fractured part connected by the internal fixation plate
could be measured by using double exposure holographic
interferometry.
2. The strength of the normal tibia in the anterior and
posterior directions was stronger than that in the lateral
and medial directions.
3. The strength of the plate system in the anterior and
posterior directions was stronger than that in the lateral
and medial directions.
4. The fixation force of the plate system decreased remarkably
by the lack of screws near the fractured line especially in
medial and lateral directions.
5. The deformation of the bone near the fracture did not coin-
side with that of the plate.

This work was supported in part by a Grant in aid for Scien-

tific Research from the Ministry of Education,Science and Cul-
ture (No.61480324).

REFERENCES

1. C.M. Vest: Holographic Interferometry (John Wiley &

Sons, New York 1979).
2. K.Arata,R.Ogawa,M.Matushima,T.Matsumoto,K.Iwata and
R.Nagata: Preprint of the meeting of the Jpn.Orthopedic
Ass.59(1985) p.S224.
3. T.Matsumoto,K.Iwata and R.Nagata: Appl. Opt. 12(1973)
p. 961.
ORTHOPAEDIC BIOMECHANICS: JOINTS
 251

BIOMECHANICS OF THE SHOULDER AND ELBOW

N. Berme
Department of Mechanical Engineering, Ohio State University,
Columbus, Ohio, U.S.A.

I.INTRODUCTION
The upper limb, with its mobile shoulder, its extensible and folding
member, the arm and forearm, and its terminal working tool the hand, is
versatile and has a large range of movement allowing the manipulation of
objects. Loss of function in anyone of the upper limb joints translates
into reduced function of the hand, which could hamper such daily
activities as eating, dressing, and personal hygiene. Furthermore, in
patients who's lower limb joints are affected by disease, or who suffer
from a neuromuscular disorder, the upper limb also assumes a weight
bearing function. A significant portion of the body weight can be carried
by the upper limbs in assisting to rise from the seated position, turning
in bed or in using a walking aid. These two conflicting requirements of
mobility and stability as well as its overall versatility makes the
assessment of the function, and loading of joint replacements for the
upper limb significantly more complicated than those for the lower limb.

2.THE SHOULDER COMPLEX

The shoulder complex is the group of structures which connect the arm
to the thorax. The interaction of these structures make the shoulder the
most mobile and complex joint in the human body. The wide range of motion
that the shoulder complex provides for the arm conflicts with the need for
a stable base of operation for the upper limb. Static and dynamic
stability, regardless of the position of the arm, are obtained by the
muscles and ligaments associated with the shoulder.
The shoulder complex consists of the scapula, clavicle, and humerus,
and four main articulations; the sternoclavicular, the acromioclavicular,
the scapulathoracic, and the glenohumeral. Subdeltoid joint is sometimes
considered as the fifth articulation of the shoulder.
Shoulder movement is obtained by a combination of motions at the
articulations of the shoulder complex. The clavicle, scapula, and humerus
all move during shoulder motion. Ranges of motion of the arm and shoulder
are easily measured, but an understanding of individual component motions
of the shoulder complex is a more difficult task. In elevation of the arm
four phases can be identified. Sometimes the last two phases of this
motion is combined and described as one phase. Most authors, however,
agree that the initial portion of elevation is strongly dependent on the
scapulothoracic and sternoclavicular joints, while the remaining motion
becomes more dependent on the glenohumeral and acromioclavicular joints.
Three types of motion occur at the glenohumeral joint. These are
rotation, rolling, and translation of the humeral head relative to the
scapula. Flexion-extension, abduction-adduction, and medial-lateral
rotation about the long axis of the humerus can all occur with a
252

combination of these three types of motion. A fairly detailed description

of shoulder ranges of motion and movements of the joints and links during
these motions are given by Kapandji (1982).
The three abductors and flexors of the humerus, the deltoid,
pectoralis major, and supraspinatus, and the three depressors of the
humerus, the subscopularis, infraspinatus, and teres minor are
continuously active throughout both abduction and flexion, thus indicating
antagonistic stabilizing muscle actions (Inman et.al., 1944).
Due to the complexity of the shoulder joint complex, predicting the
forces in the joints and muscles associated with the shoulder is a
difficult task. Quantitative analyses of the forces transmitted through
the scapula and clavicle and their associated muscles and ligaments are
virtually nonexistent. Forces at the glenohumeral joint have been
predicted, and intersegmental forces and moments at the shoulder joint, as
well as torques around the humeral axis, determined for various
activities. Glenohumeral joint force during isometric abduction in the
plane of the scapula was calculated by Poppen and Walker (1978). They
measured the EMG activity in the supraspinatus, subscapularis,
infraspinatus, latissimus dorsi, and the anterior, middle, and posterior
deltoid muscles. Muscle lines of action and the center of rotation of the
humeral head were determined from cadaver x-rays. The resultant force
between the humeral head and the glenoid was resolved into a shearing
component and a compressive component. To obtain a solution they assumed
that the force in a particular muscle was proportional to the muscle
cross-sectional area times the integrated EMG signal. They found the
resultant force to reach a maximum of 0.89 times body weight at 90 degrees
of abduction and a maximum shearing force up the face of the scapula to
peak at 0.42 times body weight at 60 degrees of abduction. The maximum
compressive force is also at 90 degrees abduction and is found to exceed
0.8 times body weight. The forces obtained by Poppen and Walker were
nearly twice as large as those obtained in Inman et.al. (1944), but this
is attributed to the fact that different lines of action for some of the
muscles were used. A technique for calculating the force generated by
individual muscles that contribute to isometric abduction of the upper
limb in the coronal plane with the humerus medially rotated was developed
by DeLuca and Forrest (1973). Their two dimensional model included the
supraspinatus muscle and the anterior, middle, and posterior portions of
the deltoid muscle.

3.THE ELBOW COMPLEX

The elbow complex is a compound synovial joint which contains three
articulations within a common fibrous capsule. The trochlea of the
humerus with the semilunar notch of the ulna form the humeroulnar joint,
the capitulum of the humerus and the head of the radius form the
humeroradial joint, and the head of the radius in the radial notch of the
ulna form the superior radioulnar joint. The humeroulnar joint is a hinge
type joint which allows for flexion and extension of the forearm. The
head of the radius is also free to rotate on the capitulum of the humerus
and does so during pronation and supination. The radioulnar articulation
is a pivot type joint in which the head of the radius spins in the radial
notch when the forearm is rotated. The bony interlocking of the ulna with
the humerus provides much stability to the elbow joint. Three ligaments
are also major contributors to this joints stability.
 253

The flexion-extension axis of the elbow joint passes through the

center of the trochlea of the humerus and changes only slightly during
complete flexion-extension. As viewed from the sagittal plane, locus of
the instantaneous centers of rotation of the elbow in flexion-extension is
located in the center of the throchlea and varies within a range of only
2-3 mm. The axis of rotation of the forearm for supination or pronation
is through the head of the radius and distal ulnar head. The radius
rotates around the relatively fixed ulna during pronation-supination. The
upper arm and the forearm form an angle in the valgus-varus direction
known as the carrying angle. In a full extension this angle is generally
10 to 15 degrees valgus for males and 20 to 25 degrees valgus for females.
The carrying angle changes, due to the fact that the trochlea is not
symmetrical, to about 6 to 10 degrees varus when the elbow is fully
flexed.
Electromyographic analyses of the muscles crossing the elbow joint
show that the primary elbow flexors are the brachialis, the long and short
heads of the biceps, and the brachioradialis, whereas the major extensors
of the elbow are the three heads of the triceps and the anconeus. A
detailed electromyographic study of both heads of the biceps, the
brachialis, and the brachioradialis was performed by Basmajian and Latif
(1957).
Since the elbow joint is not nearly as complicated as the shoulder
complex, more realistic estimates of the joint, muscle, and ligament
forces are available (Amis et.al., 1980; Berme et.al., 1977). By
measuring the external loading on the limb, noting the limb's position,
and making simplifying assumptions a model can be developed to predict
structural forces. Some simplifying assumptions commonly used which make
the system statically determinate are grouping certain muscles together,
assuming no antagonistic muscle ac tion, negligible j oint friction, and
assuming that j oint surfaces have geometrically specified contours, as
well as other constrait conditions on the muscles, ligaments, and lines of
force.
After an extensive examination of the musculature of the upper limb
Amis et.al. (1979a) predicted maximum forces up to 3 kN during flexion on
both the humero-radial and humero-ulnar articulations. A biomechanical
analysis of elbow joint function by Berme et.al. (1977) was carried out
investigating eating, dressing, pulling a heavy object (table) and
assisted standing from the sitting position. They found compressive loads
of 300 N acting on both sides of the trochlear notch during the dressing
and eating activities, and that the total joint load reaches peaks up to
2500 N in the seat rise and table pull activities.
Elbow joint forces in patients with rheumoatoid arthritis were also
studies by Amis et.al. (1979b). Forces up to 2.4 kN were predicted to act
on the distal humerus, with similar forces acting on both the radius and
ulna showing that even though the subjects, who were mostly joint
replacement candidates, experienced pain, they could still perform
activities that caused joint forces over 2 kN.

REFERENCES
1. Amis, A.A., Dowson, D., and Wright, V., Elbow Joint Force Predictions
for Some Strenuous Isometric Actions. J. Biomech., 13, 765-775,
1980.
254

2. Amis, A.A., Dowson, D., and Wright, V., Muscle Strengths and
Musclo-Skeletal Geometry of the Upper Limb. Engr. in Med., 8, 41-48,
1979.
3. Amis, A.A., Hughes, S., Meller, J.H., Wright, V. and Dowson, D.,
Elbow Joint Forces in Patients with Rheumatoid Arthritis. Rheum.
Rehab., 18, 230-234, 1979.
4. Basmajian, J.V. and Latif, A., Integrated Actions and Functions of
Chief Flexors of the Elbow. J. Bone Jt. Surg., 39A, 1957.
5. Berme, N., Nicol, A.C., and Paul, J.P., A Biomechanical Analysis of
Elbow Joint Function. I. Mech., E., 46-51, 1977.
6. DeLuca, C.J. and Forrest, W.J., Force Analysis of Individual Muscles
Acting Simultaneously on the Shoulder Joint During Isometric
Abduction. J. Biomech. 6,385-391, 1973.
7. Inman, V.T., Abbott, L., and Saunders, J.B., Observations on the
Function of the Shoulder Joint. J. Bone Joint Surg., 26(1), 1-30,
1944.
8. Kapandji, LA., The Shoulder. Clin. Rheum, Dis., 8(3), 595-616,
1982.
9. Poppen, N.K. and Walker, P.S., Forces at the Glenohumeral Joint in
Abduction. Clin. Orthop., l35, 165-170, 1978.
 255

LIGAMENT STRAINS IN WRIST-JOINT MOTIONS

A. de Lange. J.M.G. Kauer and R. Huiskes
Dept. of Anatomy and Embryology. and Dept. of Orthopaedics.
University of Nijmegen. Nijmegen. The Netherlands

1. IIITBODUC'rIOII

The human carpus consists of seven carpal bones (Fig.1) and enables the
hand to perform relatively large motions, while providing adequate stabi-
lity. The passive stability of the wrist is ensured by capsule, ligaments
and the geometry of the articulating carpal bones. In the clinical lite-
rature, carpal instabilities are usually related to lesions of the liga-
mentous structures (7,9). Based on anatomical dissection and conventional
Roentgenography the possible functions of various wrist ligaments in
relation to specific carpal motions were speculated upon (1,6,7,10). The
purpose of the present investigation is to evaluate the precise 3-dimen-
sional motions of the individual carpal bones in flexion and deviation of
the hand and to measure the strains of various ligaments, using a newly
developed method. Results describing the individual carpal bone motions
were presented earlier (4,5).

2. IIA"nmIAL AND MB'.rIIOD

Using a Roentgen stereophotogrammetric measurement system (4,11), the

motions of the individual carpal bones and the strains of the selected
ligaments of five human upper extremity specimens were measured. For
obtaining motion information, a minimum of three tantalum pellets (0.5 -
1.0 mm diameter) were inserted in the seven carpal bones (excluding the
pisiform) using a special spring-actuated syringe. Using tantalum pellets
as well, three different methods have been used to represent the fibre
lengths of selected ligaments, of which two methods accounted for the
actual courses of the ligament fibers: 1. a) by glueing 0.5 mm tantalum
pellets at intervals of 3 mm along the fibre bundles over the ligament
length, 1.b) by glueing a silicon tube, filled with tantalum pellets at
intervals of 3 mm' over the ligament. In method 2, the distances between
origo and insertion of the ligaments were measured only, using one pellet
on either side. The ligaments selected for the investigation were, at the
palmar side, radio-capitate (RCP), radio-lunate (RLP), lunato-triquetrum
(LTP), and triquetro-capitate (TCP), and at the dorsal side the radio-
triquetrum (RTD). In the radio-lunate ligament and in the lunato-
triquetrum ligament two parts were distinguished each. In one specimen
two more ligaments were marked: the short ligament between the radius and
the lunate (RLPS) and the radio-triquetrum ligament (RTP) both at the pal-
mar side (Fig.1). Cleavage of both the palmar and dorsal retinacular
tendon sheets was needed in order to reach the ligaments. After the
marking procedure the tendon sheets and skin were closed by suturing.
Each of the five human upper-extremity specimens were positioned into a
motion rig in which the radius was fixed. The tendons of the muscles
crossing the wrist were connected to stainless-steel wires, each loaded by
256

, /:--.
'-,

,
"-
"\ -'- , .. Mm;,'

R:~~;:Pd ~)~ (~~LTPP

ra ul
FIGURE 1:
Selected ligaments in a left wrist. At the palmar side:
RCP = the radio-capitate ligament: RLPD = the distal part of the radio-
lunate ligament: RLPP = the proximal part of the radio-lunate ligament:
TCP the triquetro-capitate ligament: LTP the lunato-triquetrum
ligament: RTP = the radio-triquetrum ligament: RLPS = the "short" radio-
lunate ligament.
At the dorsal side: RTD = the radio-triquetrum ligament.
Movement around the x-axis: flexion: around the y-axis: deviation.

a 20 N constant force spring. Using a mechanical motion-constraint device

the hand was moved through dorso-palmar flexion and radio-ulnar deviation,
starting from the neutral position (in which the long axis of the third
metacarpal is parallel to the radius), in motion steps of appro 4 degrees.
After each step, stereo Roentgen exposures were made and measured to
determine the 3-D positions of the landmarks. From the landmark configu-
rations in the carpal bones, their Euler rotations and translations as
well as the helical motion axes were determined, using principles of
rigid-body kinematics(4,5). The elongations in the ligaments after each
motion step relative to the neutral position, were calculated from the
ligament landmark configurations, by determining the elongation in each
interval of the landmark series and were evaluated as functions of
flexion or deviation of the hand using the rigid-body motion of the capi-
tate as a reference.
The accuracy of the experimental method is influenced by several error
sources such as the quality of the digitizer, the experience of the opera-
tor and bending effects of the image planes in the cassettes. By remea-
suring one series of Roentgenograms and comparing the recalculated kinema-
tic parameters and ligament strains with the corresponding initial values
in terms of standard deviations, the error dependency of the first two
error factors were quantified. In addition, an additional test was
performed with special constructed "ideal" flat image planes together with
the conventional image planes. Again, both sets of duplicate values of
the kinematic parameters and ligament strains were used to calculate stan-
dard deviations for each parameter (Table I).
 257

3. RBSOL'rS

For the relatively short ligaments (RLP, LTP, RLPS, TCP, RTD) all the
three strains descriptive methods resulted in equal tendencies, although
in quantitative aspect differences occured. Apparently, non-linear geo-
metry effects appeared in the longer ligaments (ReP, RT,P), by which for
these ligaments the bone-to-bone representation method is unsuitable.
Equal trends were observed in different specimens. For the short RLPS
ligament strains were measured upto 30%, relative to .the neutral position,
whereas in all other cases ligament length changes did not exceed 20%. In
dorsal flexion the palmar ligaments RLP, RCP, TCP and RTP increased in
length (Fig. 2), and in palmar flexion they decreased. The dorsal RTD
showed an inverse behaviour. The palmar LTP remained practically unchang-
ed. The short RLPS showed a specific length behaviour. In dorsal flexion
it increased in length~ in palmar flexion up to 40 degrees, it decreased
rapidly to a particular minimal length, after which, with increasing
palmar flexion, it increased (Fig.3.b). In deviation of the hand, the
palmar LTP remained, generally speaking, unchanged again. The RCP
increased in radial and ulnar deviation (Fig .4), while the TCP and RTD
both decreased. The RLP ligament showed to have two functional parts
(Fig .4). Both the RTP and the RLPS increased in length from radial to
ulnar deviation.

"'\\:
RCP' "'. 10 (1%1
TCP

-60 -ItO Ala ,'0 .... 0 -ZO

-dOrMl
., ~60 liD

p.l.ilr-...... ~
.,
-dor'\II -10 0.1111:111'_ (A)
RLPd
RLPS 1
-60 -1.0 -zo 20 40
(FLEXION)
., 60 110

RYO
- dOrsat -10 PAI.... -
dorsal palmar
-60 -40 -20 10 ..

specimen 5
supination

."
FIGURE 2: ."
Ligament strain as function of
capitate flexion~ tantalum pellets ."
glued into the ligaments. .J)

(B)

FIGURE 3:
A) schematical representation of the attitudes of the lunate and capitate
at various positions of the hand during flexion
B) Ligament strain as function of capitate flexion~ RLPS ligament of
specimen 6 ~ supination.
258

RlPd E(%I
10

E(%J

RCP <:
10

-5
- ulnar radial

RlPp 10
E I"!oJ
_ ulnar -'l radial

specimen 6
pronation ulnar -10 radial-

FIGURE 4:
Ligament strain as a function of capitate
deviation; tantalum pellets glued into
the ligaments. Specimen 6; pronation.

Repeated Flat/Bended
ReCOnstruction I_age Planes
n=12 n=23

flexion (degrees) 0.36 0.33

deviation 0.20 0.39
pronation/ n
0.22 0.44
supination

Strain (%) 0.1 0.3

'.rable I: Standard deviations of the Euler rotations and

ligament strains.
 259

4. DISCUSSION

The results show that precise analyses of ligament strains is possible

by using Roentgen stereophotogrammetry and glued landmarks, although the
strains measured are relative to the initial strain, the value of which is
unknown. In addition, the ligament strains could be related with the
spatial displacements of the carpals. As an example, in Fig.3.a the
lunate motion during flexion is graphically represented and related with
the ligament length behaviour of the RLPS ligament. Even more than in
flexion, in deviation of the hand the carpal displacements were strongly
3-dimensional of nature. In particular, in deviation of the hand, the
present findings mostly invalidate the qualitative speculations on carpal
ligament functions. This might be explained by the fact that in the
qualitative studies fixed motion axes for the carpal motions were assumed,
which is not warranted according to the present results.
Briefly, the Rep ligament resists the displacements of the capitate in
radio-ulnar deviation and in dorsal flexion of the hand. For these hand-
motions the RLP ligament has the same effects for the lunate. Both the
RTD and the TCP ligament seem to play a stabilizing role in the neutral
position of the hand, while the RTD ligament has also a function in palmar
flexion of the hand helping to ultimately resist this motion. The LTP
ligament seems to control the minor relative displacements between the
lunate and the triquetrum. The relative long RTP ligament assists to
resist dorsal flexion of the lunate dur ing dorsal flexion and/or ulnar
deviation of the hand. Relatively seen the RLPS ligament shows large
strain values, by which it is believed that this ligament functions more
for nutricient supply rather than stabilizing the wrist joint. In addi-
tion, the measured strain patterns were not influenced by the position of
the hand in the forearm. Together with geometric models (3) these find-
ings may be used as a database for developing quantitative models of the
wr ist joint.
260

REFERERENCES

1. Bonjean P, Honton JL, Linarte R, and Vignes J: Anatomical Bases for

the Dynamic Exploration of the wrist Joint. Anatomica Clinica
3(1981)73-85.

2. Goldstein H: Classical Mechanics. Reading, Addison-Wesley, 1959.

3. Kauer JMG: Functional Anatomy of the Wrist. Clin. Orthop. 149(1980)

9-20.

4. Lange A de, Kauer JMG, and Huiskes R: Kinematic Behavior of the Human
Wrist Joint: A Roentgen-Stereophotogrammetric Analysis. J. Orthop.
Res. 3(1985)56-64.

5. Lange A de, Kauer JMG, Huiskes R, and Woltring HJ: Carpal Bone Motion
Axes and Pivots in Flexion and Deviation of the Hand. 32nd Annual
ORS, New Orleans, Louisiana, Febr. 17-20, 1985 (Abstract in Proceed-
ings, p.389).

6. Mayfield JK, Johnson RP, Kilcoyne RG: The Ligaments of the Human Wrist
and their Functional Significance. Anat. Rec. 186(1976)417-428.

7. Mayfield JK: Wrist Ligamentous Anatomy and Pathogenesis of Carpal

Instability. Orthop. Clin. N.A. 15, 2(1984)209-216.

8. Mikhail EM: Observations and Least-Squares. New York, Dun-Donelly,

1976.

9. Taleisnik J: Post-Traumatic Carpal Instability. Clin. Orthop. 149

(1980)73-82.

10. Taleisnik J: The Ligaments of the Wr ist. In: The Wr ist (J. Taleisnik,
Ed.) Churchill Livingstone, New York, 1985.

11. Selvik G: A Roentgen Stereophotogrammetric Method for the Study of the

Kinematics of the Skeletal System. Thesis, University of Lund, Lund,
Sweden, 1974.
 261

EXPERIMENTAL STUDY OF THE HIP BIOMECHANICS USING PRESCALE FILMS

J.A. Petit*. S. Angevin*. R. Darmana**. J.P. Morucci** and
D. Blanquaert***
ENI TARBES*. TARBES Cedex; INSERM SC**. TOULOUSE Cedex;
CERAVER***. Courbevoie. France

1. INTRODUCTION
Generally stress distribution in human acetabulum cannot be suitably defi-
ned from the various available experimental means. Also stress values cannot
be accurately measured. So experimental results obtained by different au-
thors working on hip joint biomechanics are very scattered. The heteroge-
neity of the system takes itself a part in this dispersion. Rushfeldt et
al. (1) used pressure sensors placed on a prosthetic femoral head, preser-
ving humanacetabulum. They measured higher pressures of about 11 MPa at
acetabulum roof, and medium pressure of 3.7 MPa, with irregular distribu-
tions. Brown et al.(2), using a human femoral head, bonded with small pie-
zoresistive sensors, measured a maximum pressure of 8.8 MPa and a medium
pressure of 2.9 MPa. Contact areas are located along an anterior-posterior
axis and distributions also appear irregular. Adams et al. (3) could obtain
a medium pressure of about 1.6 to 3.6 MPa. The maximum value, recorded in
the anterior part of the acetabular zenith, is about 6MPa. On the other
hand Brinckmann et al. (4) give a maximum pressure of 3.3 MPa. Using some
cutting of the cartilage, by investigation of collapse resulting from the
applied load, Day et al. (5) estimate a maximum pressure of 3.5 MPa and a
medium one of 1.5 MPa. By a radiographic method, Armstrong et al. (6) give
a maximum pressure of 5 MPa and a medium of 1 MPa. Works done by Greenwald
et al. (7) must also be quoted. Using a colouring transfer process, they
could observe low loaded anterior-posterior contact areas and for higher
loads all the cartilage is concerned.
Then, new experimental tools, which appeared recently - pressure sensiti-
ve tapes already used to study the patello-femoral joint (8) and stresses
distribution in the foot sole (9) - bring a new contribution to the study
of human, then prothetic hip joint. This work is part of a more extended
research on stress transmission at prosthetic acetabular component (cemented
or screwed)/bone interfaces by finite elements method (10). Present results
mainly concern human jOint, but some complementary results about prosthetic
joint are given for comparative purpose.
2. EXPERIMENTAL
Pelvic bones studied were taken off embalmed cadavers. They were mounted
on an experimental compression device, simulating the two-legged stance.
This configuration was taken, because it allows experimentation with a fair-
ly good approximation of bodily statics, and with lessening the error due
to the absence of muscular system. Pelvis were fixed to the compression de-
vice by the way of the sacrum using acrylic cement. Principle of loading is
shown in figure 1.
262

Pelvis ring strain is followed by strain gauges bonded on the pelvis accor-
ding to figure 2.

.
A

\ A
I C
D

"I
C
~

Pigure 2 : Location of the

U straln gauges on anterior
T
and internal faces of the
pelvic ring.

Figure 1 : Experimental set up

Pressure inside and through the joint is measured with Prescale (Fuji) pres-
sure sensitive tapes. These sensors bring an original answer for pressure
measurements, they allow to establish simultaneously contact areas distribu-
tion and pressure intensities. These films are coloured proportionally to the
applied pressure. They are made from two parts, the A part is supporting mi-
cro balls containing a colouring; the other C is coated with a developer
product. To make measurement, parts A and C must be placed as shown in figu-
re 3.

A
IX:.::! Hi:::xl
I I c
Figure 3 : Principle of the Prescale film

Applying a load causes micro balls rupture which colour part C of the film.
The more micro balls broken, the higher is the colour density which is propo~
tional to the pressure involved. Analysis of colour density gives the pressu-
re intensity and the geometry of contact areas. According to pressure intensi-
ty involved, several scales of films are available. Human joint is first stu-
died. A prosthesis is then placed on the same joint which is analysed in the
same condition to compare the physiological and prosthetic systems. Film is
Isticked at femoral head/acetabulum interface in the form of 5 mm wide strips,
with various length depending on acetabular location.According to the accura-
cy required strips number is varying from 5 to 10. The load was applied gra-
Idually to a maximum 800 N.
 263

A total hip prosthesis was used with cemented polyethylene acetabulum and
titanium alloy femoral part with alumina head. Human acetabulum is reamed
in order to preserve sub-chondral bone on which film is sticked. A with
drawing product located at bone/cement interface is necessary to take off
the prosthetic cup. Films are recovered and analysed.
3. RESULTS
3.1. Pelvis ring strain
From experimental results about physiological and prosthetic pelvis strain,
it is possible to establish the stress transmission mode in the pelvis by
the joint. This mode is confirmed by results obtained with pressure sensi-
tive tapes. For normal anteversion internal line of pelvis ring appears
compressed, elongation of the anterior face of pubic branch is observed and
compression of anterior face of sacrum-acetabulum branch is involved. Pel-
vic ring is going to close simultaneously, pubic symphysis is going to
climb (figure 4-5).
SACRUM

Figure 4 : Frontal view Figure 5 : Lateral view

of the pelvic ring strain.
These observations are based on several assumptions :
- the structure is fixed at the sacrum.
- the pubic symphysis has only one degree of freedom because of the
bearing of the structure.
It is justified by the whole anatomy of pelvis/spine. This coupling can be
considered, for a first approximation, as a stiff embedding.Strains symetry
justifies bearing at pubic symphysis. Results obtained for normal antever-
sion indicate that the joint is transmitting a load which vertical component
causes pubic symphysis climb and horizontal component oriented to the pelvis
center favours shutting of pelvis ring. Three femoral rotations: maximum
anteversion (15 0 external), no anteversion (15 0 internal) and maximum retro-
version (30 0 internal) are involved. Results are showing only slight modifi-
cations for all anteversion configurations. On the other hand, retroversion
leads to a significant decrease of the compressive strain of the anterior
face of sacrum-acetabulum area. The vertical stress component transmitted
by the joint to this area is also decreasing.
 264

3.2. Stress distribution in the acetabulum

After loading, from film analysis, load bearing areas in the joint are ob-
served and quantitative pressure values are obtained. Results are shown on
figure 6 for three anteversion configurations.

A
B

7MPa
o o o
3MPa
o
lMPa
Figure 6 Effect of anteversion on pressure distribution, A normal ante-
version, B without anteversion, C retroversion.
For normal anteversion, load bearing areas are located in superior and poste-
rior parts of the acetabulum. Pressure values are about 5 MPa at acetabulum
zenith and 2 MPa in posterior part. These contact areas are situated mainly c
on acetabular fovea sides. Then it appears that the load transmitted by the
joint is oriented to the internal part of pelvis and its posterior component
in the direction of sacrum. With no anteversion, pressures in posterior area
are growing up to 300 per cent by comparison with normal anteversion.
This lack of balance of the load bearing on behalf of posterior part of the
acetabulum could involve an abnormal wear and induce a degeneration of the
joint cartilage. A lowering of the superior contact is also observed (from
5 to 2-3 MPa) and the contact areas move off the anterior bottom of the
acetabulum. In spite of this posterior overpressure, load transmitted by the
joint remains oriented towards the sacrum. On the other hand,a maximum re-
troversion involves an important reduction of the superior contact a~d new
bearing areas appear in the anterior part of the acetabulum. Pressures mea-
sured are about 3 MPa. Complete hip joint arthroplasty modifies stress
transmission by comparison with physiological joint. Particularly, maximum
pressures are only concentrated in superior part of acetabulum. Pressure
intensities are clear~ reduced by arthroplasty.
4. DISCUSSION
This aspect of pelvis biomechanic was already studied by several authors.
Methods are often various and sometimes far from physiological background
because of the disturbance in the articular congruence due to the sensors'
thickness or due to the use of prosthetic components. Our method, based on
the use of low thickness sensors, is also criticable and certainly affects
the pressure distribution. However, the influence on pressure values is dif-
ficult to quantify, involving a discussion of our work with references to
works with the same aim. From a quantitative point of view, pressure values
measured for normal anteversion are about 5 MPa at the acetabulum zenith.
Adams et al. (3) and Armstrong et al. (6) obtained comparable values.
However the measurement method used by these last authors is
difficult to compare with our. Based on cartilaqe collapse, it doesn't aive
 265

global data on pressure involved in the jOint and especially doesn't allow
to establish a map of pressures distribution. Other authors indicate clear-
ly higher pressure values. Particularly Rushfeldt et al. (1) give maximum
values of 11 MPa at acetabulum roof. However the interface studied is no
longer completely cartilaginous and so no longer has its normal damping
ability. Then over pressure are obtained. With an applied load of 2.7 KN
to a completely cartilaginous interface, Brown et al. (2) measured a maxi-
mum pressure of 8.8 MPa. In our case, to satisfy two-legged stance, a 800 N
load corresponding to the mean body weight is applied to the joint. The
pressure differences are directly related to weight differences. Prescale
films are also useful to observe joint load bearing areas. These areas are
mainly located in superior and posterior parts of human acetabulum and lit-
tle in anterior part. This location which can be considered as a characte-
ristic of two legged stance is little related by other authors. For example,
Day et al.(5) mention a bearing area in anterior and posterior parts of the
acetabulum for loads below three times the body weight. Over this limit,
the posterior part also becomes a bearing area. Then, with the exception
of two-legged stance, the superior part of the acetabulum is a no contact
area. Greenwald et al. (11) have the same conclusion but they think the an-
terior and posterior areas are only concerned for loads below a quater body
weight. All these studies are done, without taking account of muscular sys-
tem, in one-legged stance, so it's impossible to get quantitative values of
pressures. On the other hand, from Bombelli et al. (12) two-legged stance
pelvis equilibrium is only possible if bearing areas are horizontal. This
assumption, associated to the observation of sub-chondral bone structure
leads these authors to locate the load bearing area in superior part of the
acetabulum. Finally, Brown et al. (2), using micro-pressure sensors, have
established a map of pressure distribution in the joint and located the con-
tact area along a strip oriented mainly in the antero-posterior direction.
However, they think that irregular results come from the studied pelvis. We
put forward a difference in interpreting location of bearing areas with
other authors. Actually there is no disagreement but just considering dif-
ferent biomechanics systems. Most of works,indeed, related in this discus-
sion, are done in one-legged stance configuration with applied loads very
higher than body weight. Then a general bearing of the whole joint is quite
normal. In our cas~ with a two-legged stance configuration, it's impossible
to consider the whole surface of the joint involved. Interpreting pressure
distribution in human acetabulum shows that sacrum axis is the direction of
load transmitted by the joint. Then a simple system with three stresses at
equilibrium can simulate the two legged stance. From our work, the normal
orientation of femoral neck appears essential. Indeed, the rotation of fe-
moral diaphysis, which leads to a modification of physiological anteversion
of femoral necks, is showing over pressure areas. Some pathologies like
coxarthrosis could result.
5. CONCLUS ION
Using strain gauges to study pelvis ring strain and pressure sensitive tapes
giving a quantitative measure of pressure involved in the hip jOint, several
results were obtained in a two-legged stance:
- load bearing areas in human acetabulum are superior (5 MPa) and posterior
(2 MPa) ; they are only superior in prosthetic acetabulum.
- stress transmission by the joint is oriented in the direction of the sa-
crum. Then a three stresses system at equilibrium is observed (body weight
and joint reactions).
266

- normal anteversion must be obtained to secure a well working of the joint

to avoid modifications of joint mechanics.
N.B. : This work is part of MR. ANGEVIN' S Doctor Degree dissertation.
REFERENCES
1. Rushfe1dt, Mann and Harris (1981) : Improved techniques for measuring
in vitro the geometry and pressure distribution in the human acetabulum
II - Instrumented endoprosthesis measurement of articular surface pres-
sure distribution. J. Biomech. 14, 315-323.
2. Brown, Shaw (1983) : IN VITRO contact stress distributions in the natu-
ral human hip. J. Biomech. 16, 373-384.
3. Adams, Kempson and Swanson (1978) : Direct measurement of local pressu-
res in the cadaveric human hip joint. Med.Bio1.Eng.Comp. 16, 113-115.
4. Brinckmann, Frobin and Hierholzer (1981) : Stress on the articular sur-
face of the hip joint in healthy adults and persons with idiopathic os-
teoarthrosis fo the hip joint. J. BIOMECH. 14, 149-156.
5. Day, Swanson and Freemann (1975) : Contact pressures in the loaded human
cadaver hip. J. Bone Jt.Surg. 57B,302-313.
6. Armstrong, Bahrani and Gardner (1979) : IN VITRO measurement of articu-
lar cartilage deformations in the intact human hip joint under load.
J. Bone Jt. Surg. 61A,744-755.
7. Greenwald and Haynes (1972) : Weight-bearing areas in the human hip
joint. J.Bone Jt. Surg. 54B,157-163.
8. Huberti, Hayes: (1984). Pate110femora1 contact Pressures J. Bone
Jt. Surg, 66-A, 715-724.
9. Aritomi, Morita, Yanemoto (1983) : A simple method of measuring the
footso1e pressure of normal subjects using presca1e pressure detecting
sheets. J. Biomech. 16, 157-165.
10. Angevin, Petit, B1anquaert : Etude de 1a repartition des contraintes
dans 1e coty1e prothetique par 1a methode des elements finis. A paraitre
11. Greenwald, O'connor (1971) : The transmission of load through
the human hip joint. J. Biomech. 4, 507-528.
12. Bombe11i, Santore and Poss (1984) : Mechanics of the normal and osteoar-
thritic hip. A new perspective. C1in.Orthop.Re1.Res.182,69-78.
 267

THE SIGNIFICANCE OF THE TORQUE LOADING OF THE TOTAL HIP

PROSTHESIS
E. Savvidis. F. Loer and O. Werner
Abt. Orthopadie der Med. Fak. der RWTH Aachen. Aachen. Germany

It is generally known that total hip replacements {THR), an-

choraged by the bone cement, are successfully subjected to
immediate postoperative loading. On the other hand the primary
stability of non-cemented types of hip joint prostheses is
still differently estimated, as demonsoated in various forms
of the after-treatment concepts. Thus some authors allow the
instantaneous unrestricted loading, others a partial loading
and others again a temporary release of the affected leg (for
example Judet, 1979; Bombelli, 1983; Mittelmeier, 1984 and
others). The self and force closed mechanisms of a non-cemen-
ted hip joint prosthesis are but to a different extent less
in comparison to those of a cemented hip. In the case under
discussion, comparative investigations for the primary stabi-
lity of a cemented and non-cemented hip joint prosthesis are
carried out, which are of importance for the carrying capacity
in the final stages of recovery.

Of great importance is the fact that in an endoprosthetically

prepared proximal femur the torque load represents an addi-
tional significant instability factor besides the load rates
determined by Pauwels (1935) classical mechanisms. The rota-
tional shearing stresses are produced by loading the bended
hip joint (going stairs, rising from a sitting position)
whereby a dependance of the apparent torsional shearing stres-
ses as a function of the angle () of reflexion (specular
angle) exists (fig. 1).

MATERIAL AND METHODS:

6 pairs of fresh specimens of both thighbones of a donor with

healthy hip were obtained by autopsy. On the specimen of one
side a cemented type of hip joint prosthesis (Mliller-Grad-
schaft) and on the opposite side a non-cemented hip joint (Me-
cron-prosthesis) was typically inserted with an X-ray veri-
fication of a prosper implantation.

We erected the prosthetically prepared femur specimen in a

pressing device in such a way that the angle of action of the
force acted at an anteflexed angle of 45 against the femur
stem in accordance to the transmission of the force on the
diffracted hip joint. Besides the geometrical dimensions and
the value of the force, the obtained femur parts were hereby
268

measured. These results were recorded optically-photographi-

cally and by means of a wire strain gauge on a comprehensive
gauge and evaluation device (fig. 2, apparatus).

RESULTS:

The relative rotary motion of the inserted femur of the endo-

prosthesis opposite the dissection of the proximal end of the
femur is shown in figure 3. The observed torsional angle a be-
tween the prosthesis stem and proximal end of the investigated
femur against the torsional moment are conveyed.
The torsional moment is defined as:
F . r . sin4So
it means F loading force
r displacement between femoral
head centre and femoral axis
Whereas torsional moments of up to 120 Nm in the case of ce-
mented THR could be produced before any damage on the bone
layer occured, the maximum torsional loading in non-cemented
THR with 80 Nm was evidently lower.
The rotation for non-cemented types in comparative torsional
moments amounts to about 4 times (for example for a torque of
60 Nm,the torsion for non-cemented hip joint prostheses was about
2 as against 6 for the cemented).
The enclosed area between the graph and the abscissa gives the
dimension of the energy of deformation.
This shows that almost the same energy of deformation neces-
sary to damage the bony layer of the hip joint prosthesis is
applied until the commencement of the process of damaging.
The analysis of the proximal femura after the commencement of
the damaging situation showed a characteristic fracture beha-
viour for both different kinds of implantation (fig. 4). For
both investigated kinds of implantation different characteris-
tic graphs are produced.
After the implantation with the bone cement even transverse
fractures were observed exclusively below the tips of endo-
prostheses. On the contrary in the case of non-cemented types
spiral fractures of the femur appeared at the level of the
middle and distal stem of the endoprosthesis, an indication
of the varied transmission of the forces of both kinds of im-
plantation. The assumption mentioned above is supported by the
fact, that the transmission of the forces in the case of non-
cemented types takes place as a result of incomplete self and
force closed mechanisms in the region of layers of endopros-
theses.
Punctuated contact zones in non-cemented prosthesis produce
obviously tips of loading within the layers of the prosthesis.
The occurence of fracture lines over the whole of the layer
of the prosthesis is in accordance with this observation.
In contrasting the uniform spreaded contact of the cemented
type of THR to the bony layer produces an evident increased
amount of stability and loading claim distal of the stiffened
 269

layer of the prosthesis.

CONCLUSION:

Non-cemented implanted endoprosthesis show a lesser load capa-

city than cemented due to the torque. Therefore torsional
loading through climbing steps or standing up from a sitting
or squatting posture should be avoided during the phase of
bony ingrowth.

LITERATURE:

Aalam, M.
in Maaz (ed), Aktueller Stand der zementfreien Hliftendoprothe-
tik
Georg Thieme Verlag Stuttgart New York (1985)

Bombelli, R., Santore, R. F.

in Morscher (ed), Die zementlose Fixation von Hliftendoprothe-
sen
Springer-Verlag Berlin Heidelberg New York (1983)

Bopp, H. M.
Postoperative Behandlung nach Alloarthroplastik des Hliftgelen-
kes
Schwarzeck Verlag Mlinchen (1975)

Cotta, H., Heipertz, W., Teirig-Leube, H.

Lehrbuch der Krankengymnastik
Georg Thieme Verlag Stuttgart (1972)

Judet, R., Derzeitiger Stand der zementfreien Verankerung.

Zementfreie Verankerung von Hliftendoprothesen 117 (1979) 476

Mittelmeier, H.
Hliftgelenkersatz bei jungen Menschen
Z. Orthop. 122 (1984) 20

Pauwels, F.
Der Schenkelhalsbruch
Enke Verlag, Stuttgart (1935)
270

(a)

3 - - - - - - - -

/'

--- -
:/~=3
I =--- -
Vl
Vl I ./'
Wo::
0:: 0
>->- I/R
Vl U CD2=2
0::

-
i1ju.
It? ~---
VlZ
~> 1-
:;)'b
0>-
0::-'
-.B....-1
d12-
0:;)
I-~

(b)
0102030 40 45 50 60 10 00 90 0
HIP-FLEXION-ANGLE III

fig. 1: The load relations of the hip as a function of the

angle of hip flection

a) forces and torsional moments on the bended hip

joint

bl torsional shearing tension (TTl-hip flection

angle ()-graphs at different femoral head cen-
tre-femoral axis-displacements
 271

Force F

ForcCl measurement

strain gauge---7!e
Deflection -
gauge

. ----,

L.. _ _ _ _ _ _ _ _ _ _ _ _

fig. 2: construction scheme of the experiment

272

11 !6'
I
I
I
Deg I
I
I
....::J 9
.
E
LL 8 "
I
I

"0
C
c 7
IS ~ 6
41 ~
o."iii 5
c ..
c 4
.. III
::J 0
~
o
0: 3
I-- C
g: 2
~

1i 1

20 40 60 80 100 120 Nm 160

Torque Mr

fig. 3: torque momentan - torque angle graph

a) cemented THR stem

b) cementless THR stem

(a) (b)

fig. 4: two typical damage patterns of the maximal load

case a) cemented THR stem

case b) cementless THR stem

 273

THE LUBRICATION OF THE NATURAL JOINT: VISCOSITY OF HYALURONIC

ACID AND FRICTION IN THE HUMAN HIP
H.E. Langer, S. Altmann, W. Luhrs and H. Zeidler
Medizinische Hochschule Hannover, Hannover, Germany

1. INTRODUCTION
Synovial fluid is commonly accepted as a"biological lubricant" (19), but
conflicting results have been published concerning the role of its visco-
elasticity and of hyaluronic acid for the lubrication and friction in the
human joint. Experimental investigations come to opposite conclusions,
suggesting that the viscosity of hyaluronic acid does influence the
lubrication (14, 18) as well as it does not (8, 9, 10, 15, 22).

In pathological synovial fluids the viscosity is decreased considerably

(12, 26, 27). By theoretical and clinical means, this observation focuses
on possible relations between synovial fluid rheology, lubrication and the
pathogenesis of mechanical joint damage (17, 25). This question, however,
is judged controversially (4, 20, 22).

The present study investigates the hypothesis that high-viscoelastic

hyaluronic acid with the rheological properties of normal synovial fluid
leads to a lower friction in the human joint than preparations with lower
viscoelasticity, which are comparable with pathological synovial fluids.
This implicates the consideration that the high viscosity of normal synovial
fluid could protect the joint against cartilage wear.
2. MATERIALS AND METHODS
2.1. Joint simulator
The experiments were performed with a hip joint simulator that is based on
a hydropulse machine for material testing (POZ 0365, Schenck Company, W.
Germany). The experimental design has been described in detail elsewhere
(2). The simulator allows the vertical loading and an oscillating motion of
a human hip post mortem with various statical and dynamical loads up to
6000 N and frequencies up to 5 Hz, representing the forces in flexion and
extension at walking. Figure 1 shows the general arrangement of the simulator
with the human hip (1) fixed in the loading frame (2) with holder of the
femoral head (3) and acetabular cup (4). Vertical forces are generated by a
servohydraulically controlled actuator (5), the angular movement of the
carriage (6) is done separately by a servohydraulic drive. The equipment
enables the direct measurement of the frictional torque (T F) by a strain-
gauge load cell. Friction tests were done with statical and dynamical
loading (650 400 N) and an oscillating frequency of 1 Hz with an angular
movement off14. The duration of each experiment was 15 minutes. The co-
efficient of friction w was calculated by the formula
frictional torque
w = vertical load x radius of the femoral head
274

2.2 Hip Joints

4 human hip joints were obtained at autopsy from individuals between 16
and 40 years of age, all without history of joint diseases. The specimens
were stored with intact Joint capsule at -25C until use. After thawing the
joints were dissected to remove the synovial membrane and the surrounding
tissue. To remove the adherent synovial fluid, the cartilage was wiped clean
extensively with tissues soaked in Ringer's solution before testing.

2.3. Lubricants
3 joints wRre lubricated with highly purified hyaluronic acid from rooster
comb (Healon , Pharmacia) of various molecular weights and viscosities
(molecular weight = 40.000, 1 million, 3 millions, 7 millions; concentration
in each case 300 mg/l00 ml; zero-viscosity no = 2,1; 36; 3.000; 20.000 mPas);
one Joint was lubricated with high-viscous hyalu-
ronic acid (n = 1.320 mPas) and pooled normal
synovial flui8, that was syringed from human knee
joints post mortem. Each sample of normal synovial
fluid (0.5 - 2 ml) was centrifuged at 22.000 g
for 10 minutes to separate particulate matter,
and then stored at -25C until required. To ob-
tain the needed volume for the experiments (10 ml)
the according number of samples were pooled
(zero-viscosity of the pooled normal synovial
fluid no = 45.000 mPa.s). Ringer's solution was
used in all experiments as a standard due to its
low viscosity (n = 1 mPa.s) and its Newtonian
fluid properties~

2.4. Rheological measurements

The viscosity of the lubricants was measured in
a WeiBenberg Rheogoniometer (type R 18), cone and
plate measu~ing system_f = 7.5 cm) at shear rates
from 9.13 s to 913 s . A low shear viscosimeter
(Couette type LS 100, Contraves) was used to
Fig. 1: measure the z~30_!hear visc~!ity no at shear rates
Hip joint simulator. from 2.9 x 10 s to 4.6 s . The viscosity num-
For explanation of ber [n] was determined by dilution of the hyalu-
the numbers see text. ronic acid preparations with saline as described
previously (1).

3. RESULTS
3.1. Viscosity of the lubricant and friction
For all lubricants, the frictional values were consistent within each
Joint but varied from joint to joint so that the data must be considered
relative rather than absolute. In each experiment, friction increased in the
first minutes up to 25 % and then reached a plateau that nearly remained on
this level for next 7 hours (2). Therefore comparances between the lubri-
cants were done 15 min. after the start of a friction test. Friction de-
creased with increasing viscosity of the lubricant. The frictional torque
was lowest for normal synovial fluid with the highest viscosity (n =
45.000 mPa.s) of all lubricants that were used (fig. 2). In contra~t to
Ringer's solution, high-molecular hyaluronic acid (m.w. = 3 millions) with
high viscosity (n = 1.320 mPas) lubricated nearly as well as normal
synovial fluid (f~g. 2). Comparing the different preparations of hyaluronic
 275

acid (fig. 3), the frictional torque

. --~- -
8411 Ringer was lowest with nearly identical values
for the both preparations with the
highest molecular weight (m.~. = 3 and
7 millions) and the highest viscosi-
= 3.000 and 20.000 mPas,
Q2
ties (n
resp.).oFriction was slightly in-

.---.- -: creased for hyaluronic acid with me-

h.e. 1.320 mPa.s

0.1 ~~ dium molecular weight (m.w. = 1 mill.)

~ and viscosity (n = 36 mPaos) while it
.........---. normal I.f.. 45.000 mPas
was higher for t~e low-molecular pre-
paration (m.w. = 40.000) with low vis-
cosity (no = 2,1 mPas). After an ex-
o 5 12 periment with synovial fluid of high-
molecular hyaluronic acid had been
finished, and after the cartilage had
Fig. 2: been wiped clean with tissues soaked
Frictional torque in a human in Ringer's solution, and then the
hip joint in relation to the joint was lubricated with Ringer's so-
lubricant and duration of lution, the frictional torque remained
loading. low for a time and did not return to
the initial higher values for Ringer's
solution that had been obtained in the tests before. Therefore the experi-
ments were not performed with a random order of the lubricants but with an
ascending sequence, so that the lubricant with the lowest viscosity was in-
serted first.
Presenting the results in another form that might be more familiar to
tribologists, also the graph ~ vs. log no demonstrates the obvious corre-
lation between viscosity and friction (flg. 4; the relation ~ vs. log no
was used instead of the usual form ~ vs. n U/L (where ~ is the frictional
coefficient, no the zero viscosity, U the gliding speed and L the vertical
load applied to the bearings)
due to a better graphic re- 160 1 Ringer
presentation and because
+- + - + - + - +
----_-.1--------.-----
load and speed were iden-
tical in all experiments).
0.3 +-+.....+ - h.a. 2.1 mPa.s

The most decrease in fric-

tion can be observed in 0.2
h.a.36 mPas
the range of low visco- 4~~~4---A---.---4-----4-----4
sities (no = 1 to about 0.1 X3.000 X
100 mPa,s) while the ~~-~---~---~---~-----.-----.
h.a.20.000 mPas
angle of the further
decline tends towards o
o 1 2 3 9 12 15
zero at viscosities above t (min)
100 - 1000 mPas.
Fig. 3:
Frictional torque in a human hip in
relation to the viscosity of hyaluronic acid.

3.2. Structural rheology of hyaluronic acid and joint lubrication

Depending on the critical polymer concentration c 't a solution of hyalu-
ronic acid presents as a particle-solution, a netwofR:solution or a gel-like
network-solution (27). In the particle-solution, the molecules exist as iso-
lated particles. With increasing concentration, the particles occupy the
whole existing space but do not overlap considerably. Above the critical
276

concentration, the molecules overlap and form a network with more or less
entanglements. With the aid of Einstein's law of viscosity:
mass of the chain substance p 2,5
ccrit coil volume = equ LnJ
the critical concentration of the hyaluronic acid polymer can be calculated
from measurements of [n] (1).
A comparison between the lubricational behaviour of hyaluronic acid and
its macromolecular structure is given in fig. 4 by assigning the critical
concentrations to the frictional coefficient and viscosity. According to
that, friction seems to be lowest, when hyaluronic acid is formed as a
macromolecular network, while friction increases in the range of the par-
ticle-solution or the transitional area between particle-solution and
network solution.
4. DISCUSSION
Our tribological investigations with a joint simulator of the human hip
demonstrate an obvious correlation between the viscosity of the normal syno-
vial fluid or preparations of hyaluronic acid and the friction in the natu-
ral joint. In the global orientation, the experiments come to similar re-
sults as the Leeds group (14, 18, 23), who found an influence of the visco-
sity on the frictional properties of the human hip in vitro at low loads.

Opposite findings were presented by a number of other researchers

(8, 9, 10, 16). Digesting the hyaluronic acid fraction of synovial fluid
they reduced the viscosity of the lubricant in their friction experiments
but did not observe a loss of lubricational advantage of the fluid (8, 9,
10, 16). In contrast, digestion of the protein fraction with trypsin did
not influence the viscosity but considerably increased friction (8, 9, 16).
Subsequently, Radin, Swann and Weisser isolated a glycoprotein which they
thought to be responsible for the boundary lubrication properties of syno-
vial fluid and which was called "lubricin" (16, 19, 21). In more recent
experiments with numerous synovial fluids from diseased human joints it was
failed to show a correlation between the viscosity of the fluid and its
lubricating ability (4, 22).

McCutchen stated in 1983 that commercial hyaluronic acid does not lubri-
cate (11) but conceded that the quality of the hyaluronic acid preparations
used in the previous experiments might be not comparable with the highly
purified, high-molecular preparations that we employed in the present expe-
riments (personal communication). Furthermore, the discrepancies between
the particular results may base in part on the different experimental
models used. The experiments were performed with arthrotripsometers (8, 9,
16) or under unphysiological conditions with test devices in which synovial
or the other fluids were inserted as lubricants for latex on glass (4),
rubber on glass (10) or cartilage on glass (22). These experimental condi-
tions directly intend (4) or at least predispose boundary lubrication which
works independently of the viscosity of the lubricant. In a joint simulator
with dynamical loads and the ability of fluid film lubrication, the di-
gestion experiments (8, 9, 16) could not be confirmed (18). Over that,
Linn and Radin (9) might have misinterpreted (6) their findings with the
arthrotripsometer when their original data demonstrate a clear correlation
between the viscosity of the lubricant and the friction within the range of
low viscosity (9, fig. 5, p. 679) and nevertheless they state in the text
that they did not find this correlation.
 277

It was the most interesting result of our experiments that friction de-
creased with increasing viscosity only in the range of low viscosities
(n = 1 to about 100 - 1.000 mPas) while it was nearly unchanged above
th~s level. The initial decrease in friction might be assigned to mixed
lubrication and was seen in similar manner by O'Kelly et al. and Roberts
et al. (14, 18). In contrast to our findings, they observed an increase in
friction with viscosities above n = 50 mPa s (14) or 25 - 100 mPa. s (18),
suggesting that at this level theOmode of lubrication changed from mixed to
full fluid film conditions.

Considering that it might be problematic to transfer the usual tribolo-

gical interpretations from technical bearings to the biological joint, and
considering furthermore that the modes of lubrication may not be necessarily
deduced from friction forces or other indirect parameters (3, 29), from a
formal point of view the unchanging coefficient of friction with increasing
viscosity suggests boundary lubrication in the higher range of viscosity of
hyaluronic acid (no >
1.000 mPas) while the
very low coefficient of friction
( ,AI)
friction speaks against ge'-'II"
,...'...,,"-.0.....'_
such conditions and for 0.04
full fluid film lubri- solution structure
cation. of hyaluronic aCid

The change from mixed 0.03

to full fluid film lubri-
cation could be ex-
plained with increased
squeeze film times with 0.02
... 0.17 2.1 --------. _c_ ---------- 4.8 ----------9.1
increasing viscosity of '... Ccrit
hyaluronic acid. The
assumption, that high-
viscous hyaluronic -no.17~... _-------... - ...
acid has boundary
0.01
'-
nO.16~
lubricating proper-
ties is supported by -- ---
the observation that o
the frictional coeffi- o 2 3 4
cient remains low for viscosity (log '7 0 )
a time after the car-
tilage surface has wiped Fig. 4:
clean and is only lubri- Friction in the human hip in relation to the
cated with Ringer's so- viscosity and the solution structure of
lution. hyaluronic acid.

The comparison of the lubricational properties of hyaluronic acid with its

structural rheology points out that hyaluronic acid lubricates best, when
a macromolecular network structure is present. In this state of solution the
cartilage surfaces might be separated completely by a stable film or a
boundary layer of hyaluronic acid which might line the irregularities of the
cartilage surface and in part might reduce the surface roughness of the
bearings in the loaded area. Ogston and Stanier (13) noted that the large
molecules of hyaluronic acid present in synovial fluid film may be spherical
in shape with a diameter of 0.5 ~m. Taking the surface roughness of a normal
loadbearing human joint like the knee of a young adult with 0.5 - 2.5 ~m
278

(12; mean average roughness; corresponding maximum peak-to-valley-height

about 2.5 - 12.5 ~m),a macromolecular network of hyaluronic acid would be
expected to influence the sliding behaviour of lubricated articular carti-
lage when the separation of the surface was of the order of a few molecules,
and transition to some form of boundary lubrication would be encountered
(5). With increasing viscosity, hyaluronic acid is characterized by an in-
creasing number of entanglements (27) which might consolidate the macro-
molecular network and protect the fluid film or the boundary layer from a
breakdown.
5. SUMMARY
Summarizing the study, we think we have good evidence that in the human
joint the lubricational properties of the lubricant are not independent
from its rheological characteristics as thought by most previous investi-
gations. With increasing viscosity of hyaluronic acid the friction de-
creases, and the decrease in friction possibly is caused by a change in the
structural rheology of hyaluronic acid which transits from a particle-
solution to a network-solution. Until yet it is not clear whether the
remarkably low frictional coefficient with high viscosities must be attri-
buted to a full fluid film lubrication or to exceptionally favourable
boundary properties of hyaluronic acid.
REFERENCES
1. Altmann S, H Zeidler, et al. (1980) Rheol Acta 19: 642
2. Altmann S, H.Zeidler, HE Langer, F Hartmann (submitted for publication)
3. Clarke IC, R Contini, RM Kenedi (1975) ASME J.Lubr.Technol. 97: 358
4. Davis WH, SL Lee, L Sokoloff (1978) Arthr Rheum 21: 754
5. Oowson 0, V Wright, MD Longfield (1969) Biomed Eng, 160
6. Higginson GR, T Unsworth (1981) Tribology of natural joints,
JH Dumbleton (ed), 47
7. Langer HE, H Behrens (1986) Publ Univ Kuopio Med Orig Rep 6: A 17
8. Linn FC (1968) J Biochechanics 1: 193
9. Linn FC, EL Radin (1968) Arthr Rheum 11: 674
10. Mc Cutchen CW (1966) Fed Proc 25: 1061
11. Mc Cutchen CW (1983) B Hasp Jt Dis 43: 118
12. Nuki G, J Ferguson (1971) Rheol Acta 10: 2
13. Ogston AG, JE Stanier (1951) Biochem J 49: 585
14. O'Kelly J, A Unsworth, et al. (1978) Eng Med 7: 73
15. Radin EL, IL Paul (1970) Arthr Rheum 13: 276
16. Radin EL, DA Swann, PA Weisser (1970) Nature 228: 377
17. Rainer R, V Ribitsch (1985) Z Rheumatol 44: 114
18. Roberts BJ, A Unsworth, W Mian (1982) Ann Rheum Dis 41: 217
19. Swann DA, EL Radin (1972) J BioI Chem 24: 8G39
20. Swann DA, EL Radin, et al. (1974), Ann Rheum Dis 33: 318
21. Swann DA, KJ Block, et al. (1981) Arthr Rheum 24: 588
22. Swann DA, KJ Block, et al. (1984) Arthr Rheum 27: 552
23. Unsworth A, D Dawson, V Wright (1975a) Ann Rheum Dis 34: 277
24. Unsworth A (1975b) Discussion to Clarke et al. (5)
25. Wright V (1981) Tribology of natural and arteficial joints, JH
Dumbleton (ed), 23
26. Zeidler H, S Altmann, et al. (1979) Rheol Acta 18: 151
27. Zeidler H, S Altmann (1980): in: G Astarita, G Marucci, L Nicholais
(eds) Rheology vol. 3: Applications, 511
 279

MECHANICAL PROPERTIES OF THE ARTICULAR CARTILAGE

M. Oka and Y. Kotoura
Research Center for Biomaterials, Kyoto University, Kyoto,
Japan

INTRODUCTION
The articular cartilage has a remarkable ability to
diminish
mechanical stress on bone by softening and extending the
loads onto subchondral bone. Loss of its abili ty to
withstand mechanical stress has been considered a factor in
pathogenesis of osteoarthritis (Kempson(l), Maroudas et. al.
(3) ). An understanding of the biomechanical properties of
articular cartilage is of fundamental importance.
Though there are many reports in which an important
role of the interstitial fluid in the deformation of the
articular cartilage is emphasized (Mow et. al.
(4),(5),Torzilli (6)), there appears to be little
documentation of the fluid being observed visually. In order
to gain a clearer understanding of the role of the fluid, we
performed indentation tests on rat knee cartilage in a
silicone oil environment.
Microscopic observation of the appearance of fluid droplets
on the articular surface were made, with the aid of the
hydrophobic properties of the oil.

MATERIALS AND METHODS

Tissue used in this study was taken from proximal tibia
of adult male Sprague-Dawley (SO) rats. Sampled tibia were
preserved in a freezer (-20 DC). After storage, the
materials were soaked in saline for 30 minutes and then the
indentation tests were carried out in saline at room
temperature (20 DC). Some of the tests, however, were
performed in air and for the purpose of morphological
observation, in silicone oil.
An experimental apparatus was set-up, using a framework
of microscope as shown in Fig.l. A clamp (3) is set on the
stage to hold the diaphysis of the rat tibia and adjusted so
that the surface of the articular cartilage was always
vertical to the indenter (1). A fluid bath (4), immersing
the cartilage, is fixed to the column of the clamp. A
stereoscopic microscope (2) is positioned near the indenter
to observe morphological changes on the articular cartilage
during indentation test. A constant load (0.23N) is applied
to the surface of the fixed test material and time-dependent
indentation deformations of the articular cartilage as well
as recovery of deformation after removal of the load are
determined using a microtransducer (5).
280

RESULTS
Time-dependent indentation deformation obtained under a
constant load of O. 23N, for rat articular cartilage in a
saline solution displayed a typical indentation curve, as
described in Fig. 2. The Y axis represents deformation in
microns and the X axis, time since loading in minutes. On
the application of the load the indenter immediately sank
110 micrometer into the cartilage. This was followed by a
gradual lessening of the rate of deformation of the
cartilage over time. One minute after loading, deformation
reached 192 micrometer; it then levelled off to a maximum
around 202 micrometer (balanced state). The load was
removed 2 minutes after initial loading and the cartilage
recovered 20 micrometer instantaneously and thereafter
continued recovering, time dependently. About 3 minutes
were required for maximum recovery which represented about
95 percent of the initial deformation. The constant loading
test performed in air results in a very different
deformation curve (Fig. 3). The initial deformation was as
little as 31 microns. Creep deformation was also slight,
showing no appreciable recovery of deformation after removal
of the load. As regards morphological changes on the
articular surface. Immediately after placing the load, an
outflow of white fluid particles was observed around the
indenter.
Fig.4 is a scheme showing the time-dependent efflux and
influx of the fluid, in association with the time-dependent
deformation of the cartilage during compression. The
indenter is illustrated as a hatched bar. At A (before
loading), no fluid is present on the surface of the
cartilage. At B (when initial instantaneous deformation
occurs), a large quantity of fluid particles, shown as the
black area, gathered around the indenter. With the time-
dependent creep deformation after the initial deformation,
the outflow of fluid particles does not increase markedly;
even at C (when indentation reaches a peak), the fluid
particles only slightly increase in size and tend to gather
around the indenter. Of interest is that at D (when the
load is removed), the fluid particles which flowed onto the
surface of the cartilage disappear rapidly and at E (when
the indentation deformation has recovered to basal level),
fluid particles are no longer evident outside the cartilage.
The appearance and disappearance of fluid particles in
correlation with the deformation of the articular cartilage
were always obtained, when the indentation tests were
carried out using silicone
oil.

DISCUSSION
The typical indentation curve we obtained is
essentially similar to those which were presented by other
authors(l). These indentation experiments have led to the
general acceptance that the deformational character of
 281

articular cartilage is viscoelastic, i. e., it exhibits an

instantaneous elastic response upon sudden application of
compressive load, followed by a slow creep response which
reaches an equilibrium value in time. The mechanical
properties of the articular cartilage are controlled by both
the intrinsic mechanical properties of the tissue components
and their interaction during deformation (4,5,6). One main
component of the tissue is the solid matrix of collagen and
proteoglycan that is anchored to the subchondral bone. A
second component is the interstitial fluid within the matrix
(Linn et. al. (2),(3),(4)). Most of the fluid is free to
move in and out of the tissue on loading and unloading
(2,3,4,5). Certainly there is no doubt that fluid movement
through the matrix plays fundamental role in cartilage
deformation, but there are very few accounts of studies in
which the exudation of the fluid was visually observed. In
association with the deformation of the cartilage, we
observed time-dependent efflux and influx of the fluid
droplets. Contrary to the view that initial elastic
deformation period, the cartilage matrix changes its shape
and water efflux is very small (Weightman et. al. (7) ) , we
observed approximately 60 percent of the outflowing fluid
was in accordance with the initial deformation. The fact
that fluid movement into and out of the articular cartilage
was visually confirmed might be of great significance. From
these observations we suggest that the deformation due to
movement of the fluid plays fundamental role in the
cartilgae deformation.

SUMMARY
Though there are many reports in which an important
role of the interstitial fluid in the deformation of the
articular cartilage is emphasized, there appears to be
little documentation of the fluid being observed visually.
In order to gain a clearer understanding of the role of the
fluid, we performed indentation tests on rat knee cartilage
in a silicone oil environment. Microscopic observation of
the appearance of fluid droplets on the articular surface
were made, with the aid of the hydrophobic properties of the
oil. In association with the deformation of the cartilage,
we observed time-dependent efflux and influx of the fluid
droplets. Contrary to the view that in the initial elastic
deformation period, the cartilage matrix changes its shape
and water efflux is very small, we observed approximately 60
percent of the outflowing fluid was in accordance with the
initial deformation. The fact that fluid movement into and
out of the articular cartilage was visually confirmed might
be of great significance. From these observations we
suggest that the deformation due to movement of the fluid
plays fundamental role in the cartilage deformation.
REFERENCES
1. Kempson G.E., 1979, "Adult Articular Cartilage",
Pitman Med., London.
282

2. Linn, F. and Sokoloff, L., 1965, Arth. & Rheum. 8,

481
3. Maroudas, A and Venn, M., 1977, Annals of Rheum. Dis.
36, 399.
4. MOw, V., Kuei, S., Lai, W. and Armstrong, C., 1980,
J. of Biomech. Engineer 102, 73.
5. Mow V., Holmes, M.H. and Lai, W.,1984. J. of Biomech.
17, 377.
6. Torzilli, P.A., Dethmers, D.A. and Rose, D.E.,1983,
J. of Biomech. 16, 169.
7. ~leightman, B. and Kempson, G. E., 1979, "Adult
Articular Cartilage", Pitman Med., London.

Fig. 1
 283

IYI ,Load "i39~!!111 1IIIIIIHHlIIIII! mlllll~,*ltmllllllJlli lilliliilnHlWlilll1ii

,
Deformation
, ",
(in microns)
,""
t I'.' "
' "II I
I i ~! 1.
I
..
I,
11 , II'II i: IIi::i I! :ii'I i
III I' I! I 111
i I:!: !-II.!
1-1

2'oo~::.=.I .. I.,.:.~W ,t III ffil rf

ili!1 ,I! Ijll [I' iF
Fig. 2 " " , ' 1 J~~"'I Ii' . I .. 111' [II Illi iUi:'
flr!k 'I ! I I Il1i :1" cC

"'-"'-k"........ ",
,!Til " , i,:111
"II, II I.
' I
i' 1
I'
,'I
,",
I ,',' ~i
,iil"I" !11,:1 III II ',1 ,Ii ' " Iii! n
:",/ I i1ij:',,! Iii 1I',!,111 " ' i' I !lliii !'j !

, fl"',i" :'iII !:! [i :

,,1,1 I' I' "[",' II","! 'I' ,',i,:,I,',',I,
ie, . II :ittttlittNmt+tl+ttHl+tttttfl+l.J+l+'4i+l4I+l1W~l
J

.." :"', "" I il

Jjti i uI : :illj
I ! ' ii l ' I ; I:' ILl

. )It'Jtt~.. ~ . ,.
, ," 14+ ...
I'll Jlll
~. 50
..' ... ;

,I J I jl'il
. I I +
I . - . ,..
-,-".
..l .1' ....

Fig. 3
284

EXUDATION OF FLUID IN INDENTATION OF THE ARTICULAR CARTILAGE

C D
.(. !
INDENTATION

TIME

A
BEFORE INSTANTANEOUS
APPLICATION OF LOAD DEFORMATION PEAK DEFORMATION REMOVAL OF LOAD END Of TEST

I I
Fig. 4

Legends for Figures:

Fig. 1 Diagram of the experimental apparatus.

Indentifications. 1 indenter
2 stereoscopic microscope
3 clamp
4 fluid bath
5 microtransducer
6 load-supporting plate

Fig. 2 Deformation curve under constant loading of 23 gm.

in saline.

Fig. 3 Deformation curve obtained under constant loading

of 23 gm. in air.

Fig. 4 Scheme showing the time-dependent efflux and influx

of fluid particles, associated with the
time-dependent deformation of the cartilage.
 285

MATHEMATICAL SIMULATIONS OF PASSIVE KNEE JOINT MOTIONS

L. Blankevoort and R. Huiskes
Lab. Experimental Orthopaedics. University of Nijmegen.
Nijmegen. The Netherlands

1. II!ITRODUCTION

In view of the complex biomechanical behavior of the knee joint, mathe-

matical simulations of its mechanical behaviour are found to be useful in
the scientific process of untwining the complex relationships between the
functional biomechanical characteristics and the structural properties.
As far as the passive properties are concerned, several three-dimensional
models are known. Of the more sophisticated models (1,2) only few results
are known. The purpose of the present investigation is to measure and des-
cr ibe the passive freedom-of-motion character istics of the human knee
joint, on a subject-to-subject basis. The approach is to obtain experi-
mentally the passive motion characteristics of the joint for various
combinations of external loads and to simulate this behavior in a mathema-
tical model. The geometric properties of the individual joint specimens
are measured and are used as input for the model.
This paper describes the general characteristics of the model and some
results of parameter studies. In addition, the effect of malpositioning
an anter ior cruciate (ACL) substitute is determined, in order to illus-
trate the advantage of using a mathematical model for the evaluation of
surgical procedures. This question of the possible effects of malposition-
ing an ACL substitute is an actual one, since Hefzy and Grood (6) found a
change of the insertion location of an ACL substitute to have a large
influence on the length patterns of the ACL fiber bundles. For one
specific flexion motion they choose different tibial and femoral attach-
ment locations and calculated what the length patterns would be if a fiber
bundle was located between variable attachment locations. In this study
this experiment was repeated with the model, but in this case the effects
of alternative insertions on the motion parameters were taken into account
as well.

2. ME'l'IIODS

The mathematical model used is based on the work of Wismans et ale (1)
and is adapted to accomodate for extensive parameter variations (3). The
model calculates the equilibrium position of two rigid bodies connected by
nonlinear spring elements and contact points, while kinematic constraints
and external loads are imposed.
The attachment locations of the ligaments on the tibia and femur were
measured on the experimental specimen (4) and determine the attachments of
the line elements in the model. The strain-force relation of the spring-
type line elements is described by (1):

fIE) o <E < 2

fIE)
286

in which f ligament force

K, El material parameters
E ligament strain

The material parameters are derived from literature data. Since the ini-
tial strain at the reference extension position Er is unknown, this para-
meter is approximated by comparing the model findings with the experiment-
al results.
The articular surfaces are considered to be rigid. The geometry of the
surfaces was measured from the experimental specimen (5). The coordinates
of points on the surfaces are used to fit 3-D polynomials. Since the
femoral posterior condyles can be approximated by spheres, a polynomial
fit in spherical coordinates is applied. The simplest geometrical approx-
imation is by fitting planes on the tibial points and spheres on the
femoral points. Increasing the polynomial degree will increase the quali-
ty of the fits, but will introduce possible problems in maintaining the
condi tions for point contact. Therefore, a method is introduced to fit
low degree polynomials on those surface points that are closest to the
contact points. This means that after every calculation of the equili-
brium position of the joint, for a given set of loading conditions and
kinematic constraints, a new polynomial surface is obtained and a new
equilibrium position is calculated.
In accordance with the experimental protocol (7), subsequent flexion
positions, from extension to 90 degrees flexion, are imposed for a given
loading configuration. The resulting kinematic parameters and the liga-
ment lengths can be compared to the experimental data. Results which can-
not be retrieved from the experiments are obtained in the model, i.e. the
ligament strains, the ligament forces, the contact point locations and the
contact forces.
The parameter study involves a variation of the ligament properties and
the geometrical descriptions. Since it is not possible in the scope of
this paper to extensively document the model characteristics, this paper
focusses on the envelope of passive joint motion, as it was also obtained
from the experiments: the internal and external tibial rotation as
functions of flexion determined with tibial torques of respectively +3 and
-3Nm (Fig.1) (7).
41f internQl rotQtion

FIGURE 1:
The envelopes of passive knee joint motion of
four specimens: internal rotation with a torque
of 3 Nm and external rotation with a torque of
-3 Nm as functions of flexion.
 287

The influence of a variation of the femoral ACL attachment on the

ligament length patterns is studied for a neutral pathway (no tibial
torques) First non-functional phantom line elements are defined which
are attached on the femur 5 mm more anterior and 5 mm more posterior, and
the resulting length patterns are calculated, in fact repeating the
exper imental procedure of Hefzy and Grood (6). Secondly, the femoral
attachments of the line elements representing the ACL are actually replac-
ed anteriorly and posteriorly, and the motion simulation is repeated. Thus
influencing the length patterns as well as the motion parameters.

FIGURE 2:
The femoral surfaces approximated
by spheres, the tibial surfaces by
planes. The line elements repre-
sent the ligaments.

3. RESULTS AND DISCUSSION

The geometrical description of the joint surfaces in the model used for
the following parameter analysis consists of two spheres for each of the
femoral condyles and two planes for the tibial surfaces (Fig.2). The
medial collateral ligament is represented by three line elements (ant: K=
4000, Er= -.031 inf: K= 4000, Er= 0.01 post: K= 4000, Er= .06), the
posterior oblique by two (ant: K= 500, E r= .031 post: K= 500, Er= .03),
the lateral collateral by three (ant: K= 1000, Er= .041 sup: K= 1000, Er=
.031 post: K= 1000, Er= .04), the anterior cruciate by two (ant: K= 1500,
Er= -.041 post: K= 1500, Er= 0.075) and the posterior cruciate ligament by
two line elements (ant: K= 2250, Er= -.0351 post: K= 2250, Er= .045). De-
creasing the stiffness parameter K and decreasing the initial strain of
all ligaments has the same effects on internal and external rotation, in-
creasing rotatory laxity in the model (Fig. 3) Increasing the stiffness
or the initial strain has a reversed effect. Changing the overall stiff-
ness and the overall initial strains of the ligament does not dramatically
change the motion character istics. Alter ing the coordinates of all the
ligament attachments with respect to the geometry of the articular sur-
faces does have a large effect, particularly for changes in the x- and z-
coordinates (Fig.4).
288

CHANGE OF ALL LIGAMENT STlFNESSES

30' external rotation 30' external rotCltion

FIGURE 3:
The envelope of motion calculated with the spheres-and-plane-model.
Variations of the ligament parameters. Stiffness K and reference strain
Er being changed for all ligaments.

An internal motion pathway, i.e. flexion with an internal torque of

-3 Nm, in case the surfaces (Fig.S) are approximated by more realistic
local fits close to the contact points is shown in Fig. 6, compared with
the experimentally obtained data. The decrease in internal rotation after
40 degrees flexion, which is evident in the spheres-and-plane model, is
not present. A general improvement of the agreement between the experi-
mental results and the model calculations is obtained after introducing
this improved approximation of the articular surfaces.
The effects of a femoral attachment variation of the ACL on the length
patterns of this ligament are shown in Fig. 7. The experiments of Hefzy
and Grood were actually reproduced in case the length patterns are calcu-
lated of ACL line elements that are attached S mm anterior or 5 mm
posterior on the femur as compared to the original insertion. However,
when the attachments are actually altered the length patterns are less
affected (Fig.7.b), because the variation is partially absorbed by alter-
native motion characteristics.
This example shows that a mathematical model is a versatile tool for
analyzing experimental results and surgical procedures, i.e. cruciate
ligament reconstruction. However, the assumption of the articular sur-
faces being rigid is a critical one, since changes in surface representa-
tion do have a considerable influence on the model characteristics. There-
fore, future developments of this mathematical knee joint model have to
take into account the non-rigidity of the articular surfaces.
 289

SHIFT OF TOTAL LIGAMENT CONFIGURATION

r----.. . .
SHIFT OF TOTAL LIGAMENT CONFIGURATION
, "
internal rotatioll~ - __ internal -r~tcrtiun-.. "'5 mm posterior
40' -;::::: __ --'. - - - 5 mm proximal 40' /.... """
/." ..... I" "
~~:.:::.:.... 2.5 mm proximal , .' ". 2.5 mm posterior
'- . ,:"
lO' '-,- "', ,: .'
\ "" reference // reference
\ '-, ,.
\ "

" ,,"'.2.5 mm distal

'. 2.5 mm anterior
'5 mm distal

r-+-,--,---r;;--r---r-r.--,--,-,f I exion r-+--,--,---r;;--r---r-r.--,--,-,f I ex ion

lO' 60' . ;9'5 mm distal lO' 60' ~ }J.eS mm anterior
,
~ .2.5 mm anterior
", .... "
/
" .2.5 mm distal
,,"
. 10' ~Z~:~---- -----
:~;;~.~ .' reference '~'", '. reference
20' -"::::'-':''::::::'~-.:'"'' 2.5 mm proximal 20' '<:~ ... -.----.--.... __ .... 2.5 mm posterior
---5 mm proximal - - - - -- - - - - _. 5 mm posterior

lO' external rotation lO' external rotation

FIGURE 4:
Spheres-and-plane-model. Variations of the coordinates of the entire liga-
ment configuration with respect to the articular surface geometry in the
reference position of the joint (extension)
a. proximal and distal shift
b. anterior and posterior shift

!NTERNAL ROTATION

0-- -0 experiment
D--<l model

FLEXION
o
FIGURE 6: Internal motion
pathway obtained with the
FIGURE 5: Representation of the articular local-fit-model compared
geometry in case a local fit is applied. with the experimental data.
290

FIGURE 7: 1.2 reI. length

Relative length patterns //"ANT. SHIF~T.
5:;!f!mm
of the anterior part of -~":-

the anterior cruciate .ANT.

. . . . . *""
. . . . . ." .'. . . . . . . . :
_ ' C'_-."1
~r
.,:
(MC, N1) and the post- ~~

er ior part of the ante- t1

/l ,.
rior cruciate (PAC, N2). :
a) relative length of ,,
line elements with a
,'~'~~"~"~"""
5 rom more anterior or ",
to -i'--,----,r-.,,--, .' flexion
posterior attachment \90' 90'
on the femur. "POST.
5 mm SHIFT POST.
PAC

0.9

b) relative length in reI. length

.ANT.
case the MC and PAC 1.1 ,,
are actually attached .. -----"
, .
5 mmmore anterior or
posterior on the
femur. -POST. , --
,, ,ANT.
,
1.0
90
flexion

0.9

1. Wismans J. et al., J. Biomech. 13: 677-685, 1980.

2. Andriacchi T.P. et al., J. Biomech. 16: 23-29, 1983.
3. Dortmans L. et al., 4th Meeting Eur. Soc. Biomech., Davos, 1984.
4. van Dijk R., "The behavior of the cruciate ligaments in the human
knee", dissertation, Univ. of Nijmegen, 1983.
5. Huiskes R. et al., J. Biomech. 18: 559-570, 1985.
6. Hefzy M.S. and Grood E.S., J. Biomech. Eng. 108: 73-82, 1986.
7. Blankevoort L. et al., In: "Biomechanics, Current Interdisciplinary
Research" (S.M. Perren et al., eds), 309-314, 1985.

ACKNOIILBDGEMBHT
This investigation is performed in co-operation with the dept. of
Mechanical Engineering, Eindhoven Technical University and is partly spon-
sored by the Netherlands Organization for the Advancement of Pure Research
(ZWO) grant 90-90.
 291

DIRECT MEASUREMENT OF DYNAMIC CONTACT PRESSURE ON THE TIBIAL

PLATEAU OF THE KNEE
H. Inaba and M. Arai
Department of Orthopaedic Surgery, Akita University, School of
Medicine, Akita, Japan

Introduction
Although the management of meniscal ~nJuries by open or closed surgical
procedures is very corrmon, the role of the menisci in nonnal knee function
is still unclear. Fairbank (6) suggested that the menisci play an important
part in the transmission of inter-articular loads. In explaining their
function, he accounted for the high incidence of degenerative changes fol-
lowing meniscectomy have been made by others{2,8,9,19) .
M:lre recently, McGinty et al(12) sl1.cMed that by performing a partial rather
than total meniscectany, improved functional and anatomical results could
be achieved. In an attempt to explain the results of McGinty et al, the
present authors have examined the load-bearing patterns in the tibio
-femoral compartment by measuring directly the dynamic contact pressures on
the tibial surfaces, first with an intact menisci and then following par-
tial and total meniscectany.

Materials and Method

Six unembalmed cadaveric knee specimens were used. Their ages ranged from
15 to 58 years (mean age = 35.8 years), whilst the rudy weights ranged from
40 to 66 kg (mean weight = 55.8 kg). Three were from males and three were
from females. Likewise, three were from right and three were from left
limbs. None of them had any degenerative articular change that was visual-
ly perceptible. Prior to testing the specimen were stored at -20C. When
required they were gradually thawed at roan t.ernrerature. Following de-
frosting, the fEmJral and tibial rune shafts were cut 10 em above and below
the joint line. All the muscular tissue was renoved before being set into
special holders using Gypsun. The capsule, ligamentous structures and pa-
tella were preserved intact until imnediately prior to testing.
To measure directly the dynamic contact pressures, 12 small holes were
drilled at various positions on the tibial plateau. Pipes (diameter 1.5rrm)
were inserted through these holes fran the underlying subchondral rune,
Figure 1. Stress concentrations on the cartilage layer were avoided by
limiting the protrusion of the pipes into the cartilage layer to 1. 0 rrm.
This was confinued by loading joints prepared in this fashion with pressure
sensitive paper inserted between the articular surfaces (Prescale by Fuji
Film Co. Ltd., Tokyo). The pipes were then connected simultaneously to two
monitoring systems. A transistor pressure transducer (Toyoda-PMS-10OH)
allowed direct recordings to be made using a storage oscilloscope of the
time dependant contact pressures developed within the joint during cyclic
loading. Buldon tube pressure gauges, monitored with a video camera, were
also connected to the pipes. The tibial canponent and holder were mounted
to an auto-alignment device. This allowed varus-valgus angulation,
anterior-posterior motion and rotation of the tibia about its long axis
enabling natural alignment of the tw:> joint halves. The femoral component
292

am holder were attached to the cross head of a pneumatic loading frame via
a linkage enabling the joint to be flexed through 00-4So in intervals of
7.So. At every angle of flexion, each joint was allowed to assune its pre-
ferred alignment. Once established, the physiological varus-valgusangu-
lation and the rotational [Dsition of the tibia were maintained throughout
the experiment. The contact pressures were then measured while the joint
was subjected to cyclic loading. The maxirmm compressive load applied of
2700 N (approximately four times body weight) was based upon the calcula-
tions of !>brrison(14). He estimated that for normal walking, the maxirm.Jm
joint reaction forces were between tw::> and four times rody weight. Loads
were applied over periods of four seoonds and rerroved over periods of two
seconds. By loading the specimen in this manner, and long term creep de-
formations due to fluid expression were awided(lS). Further rreasurernents
of the ccntact pressures were made during identical cyclic loading follow-
ing partial (i.e. excision of arout 30 % of the menisci) and total menis-
sectany. furing all testing, physiological saline was used as a lubricant
and also as the pressure medium. Saline was also used to ensure the hydra-
tion of the specimen during the e~riments.
Cyclic loading

pressure
transducer oscilloscope
,li===rr:;-Auto-augnment device
'-------.---,-~ Load ceU

Fig.l. Detailed description of Fig. 2. Schematic description of the

the insertion of pipes. experimental apparatus.

Results
1. Q:mtact pressures before meniscectomy at full extension.
Figure 3 shows the changes in dynamic oontact pressure developed across the
tibial plateau of specimen No.4 with an intact menisci. The numbers in
Figure 3 indicate the [Dsitions of the inserted pipes. The pressures on
the cartilage covered by the menisci (Nos 4,6,10,11,12) generally increase
with load. With the exception of No.S, located close to the inner rim of
the menisci, the pressures did not exceed 2 MPa.

2. Contact pressures after partial meniscectany at full extension.

Approximately 30 % of the menisci were circumferentially removed fran the
inner rim. Figure 4 shows the peak pressures at position No.8 to have
increased from 4 MPa to 6 MPa following partial meniscectany. At positions
4,6,10,12 which, prior to partial meniscectany, were underlying the menisci,
the pressures were zero indicating that no contact between the ferroral
condyles and the tibial plateau was occurring. At position No.ll, which
remained covered by the miniscus following partial meniscectany, the pres-
sure was greater than with an intact meniscus whilst, conversely, the
pressure at [Dsition No.S decreased following partial meniscectany. The
cartilage at [Dsition No. S was uncovered as a result of the partial
 293

meniscectany.

3. Contact pressures after total meniscectcmy at full extension.

Following total meniscectany the contact pressures at position No.8 in-
creased rapidly with load. Even with load of 650 N the pressure was around
3 MPa. At 2700 N the pressure was as much as 7 MPa. The peak pressures
developed folla.ving total meniscect:al!Y were 81 % greater than those devel-
oped with canplete and intact menisci. Similar changes were also noted for
position No.2 where the pressures following total meniscectomy were greatly
increased. Pressures measured at positions 4,6,10 and 12 were zero follow-
ing total meniscectcrny. This result, which is similar to that obtained
following the partial meniscectomy, indicate no contact between the feroral
condyles and the tibial plateau in these areas. This suggests that the
overall area of contact between the femur and tibia is much reduced follow-
ing a total meniscectcrny. The pressures measured at position No.ll were
small canpared to those measured with intact and partially removed menisci,
although contact in this area was still present (Fig. 5) .
6.0
MPa
MPa

4.0

.i
~
,
u "
~0
0
2.0 ~o 11

2.0
;/ //'5

1000 2000 3000

~ ...----,~./'"
Load (N)
oL-~~~~--------------------
o 1000 2000 3000
Fig. 3. Contact pressures on the
tibial plateau at full exten- Load (N)

sion with intact menisci Fig.4. Contact pressures on the

(Specimen No.4, Male, Age 22, tibial plateau at full exten-
Body weight 65 kg, right knee). sion after pertial meniscec-
tony (Specimen No.4).
4. The Influence of meniscectcrny on contact pressures.
Areas on both lateral and medial condyles, which were initially uncovered by
the menisci, were subjected to greater pressure folla-.ring the excision of
the menisci. Table 1 smrmarizes the peak pressures obtained for areas of
the tibial plateau uncovered by the menisci, with intact menisci and fol-
lowing partial and total meniscectcrny.

5. Contact pressure with angulation.

The results fran specimen No.3 show that with intact menisci, the contact
pressures on the tibial plateau change dramatically with increased flexion
(Fig. 6) . Under loads of 2400 N (four times body weight), the contact
pressures at position No.3 and No.8 increased progressively with flexion.
Conversely, the pressure at position No.1 (medial tibial plateau and un-
covered by the meniscus) decreased with increased flexion of the knee.
Beyond angles of 22.5 the pressure became zero, indicating no contact
between the femur and the tibial condyle. At position No.6, under the
posterior segment of the meniscus, the pressure increased sharply with
294

Table 1. Mean values and standard deviation of the peak =ntact pressure.
Applied load: 2700 N.

Medial side (MPa) Lateral side (MPa)

With the intact menisci (IM) 39.65.3 41.56.6
After partial meniscectany (PM) 51.2 6.9 56.4 8.2
(PM/IM) (1.29) (1. 36)
After total meniscectany ('lM) 63.78.4 75.5 flO. 7
('lM/IM) (1. 70) (1. 81)

flexion, reaching a rnaximun at 22. 50 flexion. At larger angles of flexion

the pressure subsided slightly from the maximun. At position No.4 and No.
12, located anteriorly on the tibial plateau, the measured pressures at
7.5 0 flexion were small, suggesting only slight contact between femur,
menisci and tibial condyles. At angles of flexion greater than 7.5 the
=ntact pressures are zero. These facts show that the areas of contact
between fEmur and tibia move IDsteriorly with flexion and that peak =ntact
pressures increase.
A
8.0

"'Po
8

6.0
8

.
,
~
~
4.0
~
.
0

c
0
i:;.
G
9
2
11

()
u

2.0 E
0
10

11

1000 2000 3000

Load (N) Flexion angle<deg.)

Fig. 5. Contact pressures on the Fig. 6. Variations in =ntact pressure

tibial plateau at full ex- with angulation with intact
tension after total menis- menisci (Specimen No.5, Male,
cectany (Specimen No.4). AJe 31, Body weight 60 kg,
left knee). Applied load:
2400 N.
Discussion.
The technique of inserting pipes part way into the cartilage layer f = the
underlying subchondral Ixme and the use of physiological saline as a lubri-
cant and pressure medium has the great advantage that they scarcely disturb
the physiological lubricating conditions and patterns of =ntact pressure.
 295

The method was regarded as a satisfactory system for measuring inter artic-
ular cx:mtact pressures. Several other studies on the cx:>ntact areas in the
tibia-femoral compartment have been made (1,7,10,11,13,17,18,20). Maquet
(13) measured the tibia-femoral contact areas under loads of 2500 N and at
full extension am 60 flexion for knees, with and, without menisci. At
full extension the areas of cx:>ntact decreased from 20 cm2 to 12 cm2 follow-
ing meniscectany. At 60 flexion, the areas of contact decreased fran 12
cm2 to 7 an2 following meniscectany. The smaller areas of contact associa-
ted with the angulation of the knee explain the high peak pressures (Fig. 6)
found by the authors.
Fukubayashi and Kurosawa(7) detennined tibia-femoral cx:>ntact areas using a
casting technique. Under loads of 1000 N, areas of rontact decreased from
U.S an2 to 5.2 an2 following meniscectany. In demonstrating that rontact
pressures in areas underlying the menisci (positions 4,6,10 and 12, speci-
men No.4) became zero following total meniscectany, the authors agree with
Maquet(13) and Fukubayashi am Kurosawa(7) that the rontact areas within
the tibia-femoral joint are much reduced by removing the menisci. Walker
and Erkrnen(20) measured tibia-femoral contact pressures under loads of 1000
N and angles of 30 using miniature pressure transducers. They recorded
peak pressures of 1. 3 MPa on the lateral meniscus and 3.2 MPa on the medial
condyle in areas uncovered by the menisci. Fukubayashi and Kurosawa(7)
also demonstrated pressure distributions over the tibial plateau using
pressure sensitive paper (Prescale). Peak pressures doubled fran 4 MPa to
8 MPa following total meniscectany.
~re recently, Almed and Burke (1) employed a plastic rnicroindentation
transducer to measure the static pressure distribution on the tibial plat-
eau. The net thickness of the transducer was 0.285 nm and the duration of
the loading was 120 secx:mds. Although these tedmiq1.ES are likely to have a
profound effect on joint lubrication and also in creating localized stress
concentrations, the results of them(l, 7, 20) are surprisingly sllnilar to
those presented in this paper, especially when cx:>nsidering how the visco
-elastic defonnations of the cartilage rould effect pressure distribution.
Evidence fran previous studies(1,7,10,11,13,17,18,20) regarding the areas
of contact, coupled with data regarding cx:>ntact pressures (1, 7 and the
present study) within the tibia-femoral canpart:Irent suggest that by per-
forming a total meniscectany, the areas of contact are much reduced and
sharp increases in contact pressure occur. Fran this it may be inferred
that the menisci play an important role in load bearing. Seedhan and
Hargreaves (17) tried to further clarify the role of the menisci. They
demonstrated that intact menisci transmit between 77 % and 99 % of the
total loads applied to knee joints.
The results of the present study, which have shown marked increases in
cx:>ntact pressures following meniscectany, clearly show that a major frac-
tion of crnpressive loads are transmitted by the menisci. With such
dramatic alterations in contact pressures following meniscectcmy, it is not
surprising to find tissue renodelling and sings of wear and degenerative
changes. Noble am Alexander(16) have reported large increases in the
density of subchondral bone of the tibial plateau where the meniscus has
been :renoved. This fact suggests that the menisci, by transmitting loads,
have a protective effect on cartilage. The present study, which shows that
partial meniscectany results in lower cx:>ntact pressures than total rnenis-
cectany, perphaps explains why partial meniscectanies lead to less severe
degenerative changes(4,5) and better anatanical and functional results (12) .
en the basis of our findings, it is sUJgested that where meniscal repair is
irnIx:>ssible, a partial meniscectany shaul be performed if, in doing so, a
sufficient residual meniscus can be retained. Such a meniscus may well be
296

beneficial to load-bearing within the joint. Furthernore, the high contact

pressure which is present on the posterior segment of the medial meniscus,
at angles of alx>ve 20 could well provide a clue to why this section of the
meniscus is so susceptable to damage.

Surrnary
1. Dramatic changes in the contact pressure distribution occur with angu-
lation of tiE Jmee.
2. The peak contact pressures increase markedly following meniscectcmy.
3. The most probable functions of the menisci are load bearing and the
protection of the cartilage fran excessive loads.
4. Fbr meniscal injuries a meniscal repair is reccmnended. In circum-
stances where this is impossible, a partial meniscectany soould be per-
fonned in preference to a total meniscectcmy.

References
1) Ahmed, A.M. and Burke, D.L.: ASME J. Bianechanical Engineering, 105: 216
-225, 1983.
2) Appel, H: Acta Orthop. Scand., Suppl. 133: 1970.
3) Cassidy, R.E. and Shaffer, A.J.: American J. Sports Medicine, 9: 209-213,
1981.
4) Cox, J.D., et al.: Clin. OrtOOp., 109: 178-183, 1975.
5) Distefano, V.J.: Clin. Orthop., 151: 143-146, 1980.
6) Fairbank, T.J.: J. Bone Joint Surg., 30-B: 664-670, 1948.
7) Fukubayashi, T. and Kurosawa, H.: Acta OrtOOp. Scand., 51: 871-879, 1980.
8) Jackson, J.P.: Br. Med. J., 2: 525-527, 1974.
9) Johnson, R.J., et al.: J. Bone Joint Surg., 56-A: 719-729, 1974.
10) Kettelkanp, D.B. and Jaoobs, A.W.: J. Bone Joint Surg., 54-A: 349-356,
1972.
11) Krause, W.R., et al.: J. Bone Joint Surg., 58-A: 599-604, 1976.
12) McGinty, J.B., et al.: J. Bone Joint Surg., 59-A: 763-766, 1977.
13) Maquet, P.G., et al.: J. Bone Joint Surg., 57-A: 766-771, 1975.
14) Morrison, J.B.: J. Biomechanics, 3: 51-61, 1970.
15) Mow, V.C., et al.: J. Biomechanics, 17: 377-394, 1984.
16) Noble, J. and Alexander, K.: J. Bone Joint Surg., 67-A: 295-302, 1985.
17) SeedOOm, B.B. and Hargreaves, D.J.: Engineering in Medicine, 8: 220-228,
1979.
18) Shrive, N.G., et al.: Clin. Orthop., 131: 279-287, 1975.
19) Tapper, E.M. and Hoover, N.W.: J. Bone Joint Surg., 51-A: 517-526, 196~
20) walker, P.S. and Erkman, M.J.: Clin. Orthop., 109: 184-192, 1975.
 297

A DYNAMICAL MODEL OF THE KNEE JOINT

A. Zabel and U. Rehder
Orthopadische Universitatsklinik. Hamburg. Germany

1. INTRODUCTION

In the four bar model the ligaments of the knee joint are
considered as being rigid without having any inner structure
(Menschik,1974). In that model, fibers of ligaments stay isome-
trically during flexion.

The knowledge of the exact length patterns as a function of the

angle of flexion is essential for the determination of the best
place of implantation of ligament grafts and for the construc-
tion of condylar prostheses.

The three dimensional model of the knee joint presented here

emphasizes the great importance of the single fibers for the
kinematics of the joint.

The results obtained from the model were compared with the
results of the geometrical involute-model of the knee joint
(Rehder, 1983).

2. METHODS

The femoral and the tibial components of the three-dimensional

model were 1:1 copies of a dissected human knee joint. The four
ligaments of the joint were realized by six fibers for each
ligament. The fibers reached the electronic measuring device
through bore holes in the tibia and were attached to the ground
plate by springs (Fig.l).

The distance between a femoral attachment point of a fiber and

the corresponding tibial point was measured as a function of
the angle of flexion of the knee and was displayed on LED's.

By removing some fibers from the measuring device, partial or

complete ruptures of ligaments were simulated. The consequence
of a rupture for the kinematics of the knee became evidently .
Changes in the length patterns were quantified.

3. RESULTS

The diagrams (Fig.2a-d) show the distances between two femoral

298

"
-1IIIIDmi
, ~ I \.

Fig.l. Measuring device for the measurement of the 1 ength

patterns of single fibers of the ligaments.

and the corresponding tibial attachment points of each ligament

as a function of the angle of flexion of the knee joint.

The curves are normalized on the maximum of distance for each

fiber.

When the distance reaches its maximum, the fiber under conside-
ration is tightened. Only at this angle of flexion the fiber
acts as a force bearing element.

The envelope of the curves is a straight line parallel to the

abscissa at l/lmax = 1, indicating a tight connection between
femur and tibia in the whole range of flexion. The lateral
collateral ligament is not under tension in medium angles of
flexion.
 299

POSTERIOR CRUCIATE LIGAMEHT lATERAL COLLATERAL LI GAMEHT

\ ) \
~ ~

1.8 1.8

THEORY THEORY

8.8 angle of flexion 158' 8.8 angle of flexion 158'

1.8

EXPERIMEHT EXPERIMENT

1 I 1 I I ! I ! 1 1 I

8.8 angle of flexion 15B' 8.8 angle of flexion 158'

Fig. 2a Fig. 2b
300
"EDIAL COLLATERAL LJG~EKT AKTERIOR CRUCIATE LIG~EKT

1.8 1.8

2
1

THEORY THEOR't

8.8 angle of flexion 158- 8.8 angle of flexion

1.8 1.8

EXPERII1EIfr EXPERII1EltT

8.8 angle of flexion 8.8 angle of flexion 158'

Fig. 2c Fig. 2d
 301

4. COMPARISON WITH THEORY

The experimental results were compared with the results given
by the geometrical involute model of the knee jOint.

The values of the rolling-gliding coefficient Aas functions of

the angle of flexion t were estimated from the 3-D-model as
A = 1 for t = 0 ... 25;
A = 0.33 for t = 25.1 0. 80 0;
A = o for t > 80.

Fibers, which are attached to the femur in the near of the

instant center of rotation at a medium angle of flexion, dis-
play small changes of distances, while others, which are at-
tached far from the center, display large changes.

5. RUPTURE OF A LIGAMENT
By removing the six fibers of a ligament from the measuring
device, a rupture of this ligament can be simulated.

The consequence of a rupture of the anterior cruciate ligament

for the kinematics of the joint illustrate figures 3a-c. The
length patterns of the fibers of the remaining ligaments change
significantly.

The anterior part of the medial collateral ligament elongates

in flexion, instead of shortening in normal case. The distances
of the attachment points of the posterior cruciate and the
lateral collateral ligament decrease during flexion.

POSTERIOR CRUCIATE LIGAMENT

za
path.

Fig. 3a-c. The length pat-

terns of the fibers of the
ligaments change signifi-
cantly, when a complete
rupture of the anterior
cruciate ligament is simula-
ted by removing its six
fibers from the measuring
device.
a.a angle of flexion 1sa'
302

LATERAL COLLATERAL LIGAMEKT MEDIAL COLLATERAL LIGAMEHT

28 28

path .

8.8 angle of flexion 8.8 angle of flexion

Fig. 3b Fig. 3c
6. CONCLUSIONS

All ligaments of the knee joint -except the lateral collateral-

are partly tightened in the whole range of flexion. Single
fibers alternate reaching their maximum of flexion, i.e. the
maximum of tension. For the three ligaments no angle of flexion
is found in which all fibers are loosened.

Some bundles of fibers display an isometric length pattern

during a large range of flexion. Their attachment points seem
to be an optimal insertion area of a ligament graft. Neverthe-
less, this area and the diameter of a graft inserting here is
too small to guarantee for an adequate tensile strength.

In surgical reconstruction of knee-ligaments the great functio-

nal difference between groups of fibers with different areas of
attachment has to be taken into account. Separate replacement
of Bin~le parts of ligament grafts could be a better method for
a physiological reconstruction of ligaments.

7. REFERENCES

1. Menschik A: Mechanik des Kniegelenkes.

Z. Orthop. 112, 1974, 4Bl
2. Rehder U: MOrphometrical studies on the symmetry of the
human knee joint: Femoral condyles.
J. Biomech. 12, (19B3), 351

B. ACKNOWLEDGEMENT

Supported by the Vereinigung Nordwestdeutscher Orthopaden and

the Verein zur Erforschung und Bekampfung rheumatischer Erkran-
kungen e.V.
 303

GEAR PROPERTIES OF THE HUMAN KNEE JOINT

U. Rehder
Orthopadische Universitatsklinik. Hamburg. Germany
1. INTRODUCTION
A twodimensional geometrical model of the knee joint based
on the involute geometry of the femoral condyles was estab-
lished. The calculated ligament length patterns were in good
agreement with experimental data from a three-dimensional model
(Zabel, Rehder, 1986). A twodimensional description is an
adequate tool for explaining qualitatively the kinematics of
the knee.

2. RESULTS
The profiles of the femoral condyles in the sagittal plane
can be described by an involute of a circle within an accuracy
of 0.2 mm (Rehder, 1983) (Fig.l). Consequently, the plane
motion of the knee joint can be explained by investigation of
the geometrical properties and the kinematical behavior of the
involute.

The construction of the involute is performed by rolling a

straight line on the base circle (Fig.2). The circle is the
geometrical locus of the centers of curvature of the spiral.
The length of the normal is proportional to the angle of evolu-
tion t.

)
rtl
/

\///
I x /

Fig. 1 Fig. 2
304

During gliding motion, a fixed point on the tibial plateau

comes into contact with the whole circumference of the femoral
condyle. As derived from the construction of the involute,
during flexion motion of the knee the base circle (the evolute
of the femoral profile, in general) rolls on a straight line
perpendicular to the tibial plateau (Fig.3). The base circle is
the polhode of the body fixed system, the contact point defines
the instant center of rotation.

)
/i \
)
?---.;.----....

1< . ~~)
\.
/ "
---~-----

Fig. 3 /
J

Therefore, a point connected with the femur runs through a

cycloid (Fig.3). The loop of the cycloid implies the existence
of a maximum of distance between a point on the femur and a
point on the tibia. A fiber attached to both of the bones is
tightened, when it reaches its maximum of length at a particu-
lar angle of flexion (Fig.4). At this angle the fiber crosses
the instant center of rotation, i.e. the contact point of the
circle and the perpendicular line.

~ 1.8
~~""
)
) lmax

<
1
\ I
\
I
I
e.e angle of flexion 15e'
Fig. 4
 305

In general, the motion of the knee joint is a rolling - gliding

motion. The femur fixed polhode of this combined motion is
positioned between the base circle, i.e. the polhode of gli-
ding, and the involute, i.e. the polhode of rolling. To find
the actual position of the polhode, a rolling-gliding-coeffi-
cient A (Muller, 1963) was defined as the ratio of the length of
the contact lines on tibia and femur in an interval of the
flexion angle t (Fig.5):

ill (tib)/ ilt ill (tib)

=- ----
ill (fem)/ llt ill (fem)

(glide)
with A : ={ ~ < A < 1
(roll)
(roll/glide)

It can be proved mathematically, that the body fixed polhode

divides the normals net) of the involute by the ratio (t).

) \

rL \
/
~~ 1I1II \

A o
~ I 0.5
Fig. 5
Now, the pathways of points of the femur can be calculated.
Parameters of the calculation were
1. the involute geometry of the medial femoral condyle;
2. the course of A (t) and
3. the coordinates of the femoral points.

Data for A(t) were estimated from a three-dimensional model

(Zabel,Rehder,1986) (Fig.6).
A is assumed to be
f. = 1.0 for t 00 ... 25;
A 0.33 for t = 25.1 ... 80 and
A = 0.0 for t > 80.
306

1. a r---

Fig. 6
) 13.13 angle of flexion 1513'

The computed curves were compared with the experimental data of

the physical model (Zabel,Rehder,l986). It becomes evidently
that different fibers contribute to the connection of the joint
at different angles of flexion, when the fibers are reaching
their maximum of length.

1.8 -----====
) \ ///./--" 2
3

~) lmax

123
)
8.13 angle of flexion 158'

Fig. 7a
 307

~\
8.8 angle of flexion 158'
Fig. 7b

With given pathways a calculation of distances between femoral

and tibial attachment points of fibers of ligaments becomes a
simple procedure. The figures 7a-b show the pathways of femoral
points and the computed distances to corresponding tibial
points (e.g. medial collateral and anterior cruciate ligament)
as function of the angle of knee flexion. The distances were
normalized on their maximum.

3. REFERENCES

1. Muller HR: Kinematik, Berlin, DeGruyter, 1963

2. Rehder U: Morphometrical studies on the symmetry of the
human knee joint: Femoral condyles.
J.Biomech.,16, (1983), 351
3. Zabel A, Rehder U: A dynamical model of the knee joint.
5th ESB Meeting 1986, Berlin

4. ACKNOWLEDGEMENT

Supported by the Vereinigung Nordwestdeutscher Orthopaden and

the Verein zur Erforschung und Bekampfung rheumatischer Erkran-
kungen e.V.
 309

MOIRE FRINGE CONTOUROGRAPHIC DETERMINATION OF THE THREE-

DIMENSIONAL GEOMETRY OF THE POSTERIOR ARTICULAR SURFACE OF THE
CALCANEUM
P. Klein and F. de Schryver
laboratory of Functional Anatomy. ISEPK. U.l. Brussels.
Brussels. Belgium
INTRODUCTION

Knowledge of the precise 3-D geometry of articular surfaces becomes

more and more imperative especially in the application of mathematical
models of joints (Wismans et al.,1980) or in the design of implants (Wyss
et al.,1982). These datas will improve the understanding of the mechanism
of a joint and probably most of the classic descriptions of the geometry
of joints could be revised.
The posterior articular surface of the calcaneum plays an important
role in the subtalar joint. Nevertheless the classic descriptions of its
geometry are rather divergent. Fick (1904), Braus (1929), Spalteholz
(1953), Rauber and Kopsch (1968) and Kapandji (1974) describe it as a part
of a cylinder. On the other hand, Gegenbauer (1889), Lanz and Wachsmuth
(1972) attribute a conical shape to this surface. Schmidt (1981) and Huson
(1985) have studied its geometry but without clearing the question on the
cylindric or conical shape of the surface.
Different methods have been described for the acquisition of three-
dimensional coordinates of a number of surface points. They either make
use of a measuring pin connected to a pantograph (Wyss et a1. ,1982) or
connected to a dial gauge (Wisman et al.,1980), or they make use of a cast
(Hiss et al.,1980, Schmidt, 1981). A third category avoides the mechanical
contact with the surface and uses an optical technique as stereophoto-
grammetry (Huiskes et al., 1985) or interferometric holography (Klein et
al. ,1985). The use of Moire contourography, which is reported here, has
not yet been applied to the study of the three-dimensional articular
surface geometry and this in spite of its simplicity. We think it to yield
a relati vel y good accuracy, but probably less then stereophotogrammetry
using professional photogrammetric equipment.

METHODS AND MATERIALS

The principles of the shadow Moire, producing contour lines, have been
published by different authors, some in close sequence (Meadows et al.,
1970, Takasaki, 1970 and Pirodda, 1982). For reasons of convenience to the
reader, some essential principles together with an estimation of some
errors of the method are briefly outlined here.
The experimental set up is schematicaly shown in figure 1. A light
source S illuminates from a distance ls an equispaced plane grating with
line spacing s. The articular surface is placed as close as possible under
the grating. When examinig from a point 0, at a distance 10, through the
grating itself, the shadow cast by it on the surface, one can readily see
Moire patterns. Under specific conditions these patterns are contour
lines.
310

Figure 1. Schematic drawing

of the Moire topography.

If ls 10 1 and if b is the distance source - observer, the

kth order contour line is:

zk = 1 k s (k = 0,1,2, .. ) (1)
b - k s
and the sensitivity factor is (Pirodda, 1982):

( 1 + Zk)2
s 1 1
(2)
AZk=b s zk 2 (k 0,1,2, . )
1 + btl + 1 )
The sensitivity factor is not constant. It increases with k. But
pratically if ~ 1 and if
1 it takes a nearly constant value of:

1
ll:Z=sb (3)

Wi th regard to the maximum height of the articular surface and to the

experimental set up the error due to the changing sensitivity factor did
not exceed 0.01 millimeter.
The experimental set up was arranged in such a way to give a senstivity
factor of 0.5 millimeter.
The effect of errors in the distances of 1, b and s can be evaluated
by differentiating equation (1) and applying the law of error propagation
(van Wijk, 1981). Let ml, ~ and ms be the RMS errors of 1, b and s,
respectively. The RMS error mZk (for a given k) is:

(k s)2 2 (kls)2 2 (klb)2 2

m + m + m (4)
(b _ ks)2 1 (b _ ks)4 d (b _ ks)4 s

Wi th regard to the experimental set up presented here and i f k = 30,

ml = ~ = 2 mm and m = 0.001 mm, the RMS error mZk = 0.05 mm.
In order to washSout the noisy structure of the grating and the fringes
of the Moire patterns formed by the interferences of higher harmonics of
the grating and of its shadow, we opted for the "moving grating technique".
Furthermore, this techniques averages the error (ms) in the pitch of the
grating which contributes to the improvement of the quality of the fringes.
Special care was taken to ensure a pure translation of the grating with a
minimal deviation in the Z direction. This was made possible by the use of
four- precision linear ball bearings supporting the grating and sliding on
two shafts.
 311

The accuracy of the measurements was checked using two test objects: a
flat inclined surface and a cylinder.
For the surface, inclined to 45 degrees with regard to the grating,
the RMS error, calculated on the z values by

1 n 2 1
RMS [ -n~
L-
(zc'1. - Zk')
1.
]':! (where zCi is the known value of z and
i=l Zki is the value determined by the Moire
contourography)
is less then 0.11 millimeter.
For the cylinder, with a nominal radius R, the checking was performed
by comparingR to the radii calculated from the x and Z coordinates deter-
mined by the Moire contourography. The RMS error did not exceed 0.15 mm.
As the test objects and later on, all the articular surfaces were
centered to the optical axis of the lens, no corrections were performed
for the central perspective projection.
Photographs were taken with a Lei tz R camera mounted with an objectif
of f = 90 mm. The photographs were magnified three times.
A total of 21 human calcaneums were investiguated. The ages of the
subjects were unknown, but generally they had advanced ones. The specimens
were constantly kept wet, excepted for the time of mounting and of expo-
sure. This lapse of time did not exceed a couple of minutes. The calcru,eums
were fixed in an anatomical position with the top of the posterior articu-
lar surface as close as possible to the grating. Although it is possible
to visualize fringes on the articular surface, the surfaces were uniformely
covered with a white lustreless powder, the thickness of which did not
exceed 0.005 mm. In such a way the contrast of the fringes became much
higher.
Figure 2. shows a typical photograph of the posterior articular surface
of a calcaneum with the fringes obtained by Moire contourography.

Figure 2. A typical photograph

of the posterior articular
surface of the calcaneum with
contourlines (~z = 0.5 rom).
312

The cross-sections, an anterior, a middle and a posterior one (A,M and

p) were performed at right angle to the main direction of the crest of
the surface. The middle cross-section was outlined on the maximum width of
the surface. The anterior and posterior sections were situated at 5 mm
anteriorly and posteriorly to the middle one. The points corresponding to
the contour lines were fitted to polynomials by a least square method. The
polynomial were either of the third or of the fourth degree, depending on
the residual error wich was kept under 0.1 mm.
The coordinates of the points were measured with a two-dimensional
electronic digitizer, Numonics R , with a given accuracy of 0.25 mm. The
reading of the coordinates is probably the most important source of error.
It may be decreased by the experience of the operator in using the digiti-
zer and especially by the magnification of the photographs.
By differential geometry the radii of curvature could be calculated
which are local invariants, i . e. not depending on the coordinates, and
thus make the comparison possible of the intra- and intervariabili ty of
all specimens.

RESULTS

The quantitative analysis would be beyond the purpose of this presenta-

tion, therefore only qualitative datas will be related here.
Figure 3 represents a typical recording of the points of the three
cross-sections with polynoms fitted to them. The comparison of such curves
is rather contigent and preference is given to the method of the radii
analysis (figures 4 and 5). This analysis is characterized by the fact
that the radius has never a tendancy to be constant, which denotes that
the surface has not a circular section.
C~.LC ~~JEUH 1'0
I ., ~,J

Figure 3. A typical result

of three cross-sections (A,
M and p) with the polynoms
fitted.

II~
I A 0I 5 , I I
L Qt..
[.".,,:
 313

From all the graphical results two categories could be roughly drawn
out.

RAYON CAL [AI~[UI~ 15." p

1...... 1 ''''rIO]

I,. "1 . ,1.

1 ...... 1
L-~~~~~~~~~~~~~~~.

Figure 4 (left) and figure 5 (right). Radii of curvature for two speci-
mens, representing the two categories of surfaces.

Figure 4 shows an example of the first category. The three curves,

representing the radii of curvature of the respective cross-sections A,M
and P, are minimal around the middle area. Medially and laterally the
radii tend to increase denoting a flattening of the surface. This is in
contrast with most of the classic anatomical descriptions, where the
medial part is generally said to be more curved. Authors reporting those
descriptions were probably led into error by the fact that the surface is
much more inclined medially, which is obviously visualised for the ante-
rior and to a less extend for the middle cross-section in figure 3.
314

The second category shows a more complexe evolution of the radii.

Figure 5 represents a typical example. Analysing the curves from the
medial to the lateral border, the radius tends to decrease firstly, i.e.
the curvature of the surface is more pronounced,then it increases gradual-
ly, passing by a maximum. Laterally the radius decreases once more.
In their middle parts all the surfaces present a smaller radius in
the anterior cross-section. This denotes a certain "conical" shape, but
this conus has no circular section. Comparing the radii of the middle
part of the middle cross-section to the corresponding ones of the poste-
riorcross-section they are sometimes higher, sometimes lower. No tendancy
could be found when comparing the three cross-sections A,M and P.

Braus, H.(1929) Anatomie des Menschen. Band 1. Springer Verlag, Berlin.

Fick, R. (1904) Handbuch der Anatomie und Mechanik der Gelenke. Fischer
Verlag, Jena.
Gegenbauer, C. (1889) Traite d'Anatomie humaine. Reinwald Ed., Paris.
Hiss, E., Schwerbrock, B. (1980) Untersuchungen zur diumlichen Form der
Femurkondylen. Z. Orthop. 118:396-404.
Huiskes, R., Kremers, J., De Lange, A., Woltring, H.J., Selvik, G., van
Rens, Th. J G. (1985) Analytical stereophotogrammetric determination of
three-dimensional knee-joint geometry. J. Biomechanics 18,8:559-570.
Huson, A. (1985) Personal communication.
Kapandji, I.A. (1974) Physiologie articulaire. Maloine Ed., Paris.
Klein, P., Rooze, M. (1985) Morphologie de la surface articulaire poste-
rieure du calcaneum. Chirurgia del piede 9,6:399-406.
Lanz, T., von Wachsmuth, W. (1972) Praktische Anatomie: Bein und Statik.
Springer Verlag, Heidelberg, New York.
Meadows, D.M., Johnson, W.O., Allen, J.B. (1970) Generation of surface
contours by Moire patterns. Appl. Optics 9,4:942-947.
Pirodda, L. (1982) Shadow and projection Moire techniques for absolute or
relative mapping of surface shapes. Opt. Eng. 21,4:640-649.
Rauber, A., Kopsch, F. (1968) Lehrbuch und Atlas der Anatomie des Men-
schen. Teil 1. G. Thieme Verlag, Stuttgart.
Schmidt, H.M. (1981) Die Artikulationsflachen der mensch lichen Sprung-
gelenke. In: Advances in anatomy, embryologie and cell biology. Springer
Verlag, Berlin, Heidelberg, New York.
Spalteholz, W. (1953) Handatlas und Lehrbuch der Anatomie des Menschen.
Teil 1. Hirzel Verlag, ZUrich.
Takasaki, H. (1970) Moire topography. Appl. Optics 9,6:1467-1472.
Wijk, van,M.C. (1981) Accuracy of Moire topography. In: Moreland, M.S.,
Pope, M.M., Amstrong, G.W.D. (eds.) Moire fringe topography and spinal
deformity. Pergamon Press, New York.
Wismans, J., Veldpaus, F., Janssen, J., Huson, A., Struben, P. (1980) A
three-dimensional mathematical model of the knee-joint. J. Biomechanics
13,8:677-685.
Wyss, U.P., Doerig, M., Frey, 0., Gschwend, N. (1982) Dimensions of the
femoral condyles. In: R. Van Campen, D. and De Wijn (eds). Biomechanics:
Principles and Applications. M. Nijhoff Publishers, The Hague, The
Netherlands.
ORTHOPAEDIC BIOMECHANICS: JOINT PROTHESES
 317

LONG TERM ANIMAL TESTS WITH INSTRUMENTED HIP IMPLANTS

G. Bergmann, F. Graichen, H. Jendrzynski, A. Rohlmann
Free Univ.Berlin, Dept.of Orthopaedics (Oskar-Helene-Heim),
Berlin, Germany

ABSTRACT
A new telemetry allows long term in vivo measurements on
orthopaedic implants with three strain gauges. The paper des-
cribes the telemetry, the design of hip endoprostheses for the
measurement of spatial joint forces in sheep and the long-term
tests in three animals.
INTRODUCTION
The optimization of orthopaedic implants requires
knowledge of the acting loads or the deformations. If no
realistic calculations or in vitro tests are possible, in vivo
measurements with strain gauges (SG) can be performed. For
humans, this requires an implanted telemetry, mostly with
multichannel capabilities. Commercial devices use batteries
with a limited capacity. Such telemetries are big and must be
arranged separately from the implant. This causes problems with
the sealing and strength of th~ connecting cables. Such teleme-
try can not be used for permanent applications. The only
solution is to develop a miniaturized telemetry which can be
incorporated in the implant and is powered inductively. An
absolutely safe encapsulation without the use of any plastics
is a necessity.
The described three-channel telemetry meets all these
demands. Before it can be used in humans, however, it had to be
tested in long-term animal experiments.
TELEMETRIC MEASUREMENTS ON ORTHOPAEDIC IMPLANTS
RYDELL (9) was the first who measured three-dimensional
joint forces with a strain gauged, battery-powered hip
prosthesis with a separately implanted telemetry. Only two
patients were investigated and some experimental shortcomings
(2) occured. HODGE et.al.(8) implanted a battery-supplied hip
prosthesis in one patient to measure the pressure distribution
on the prosthetic head and to calculate the approximate joint
forces. ENGLISH and KILVINGTON (7) implanted two hip prostheses
and got preliminary readings, but only on the force components
in direction of the neck. BROWN et.al. (5) report on a tele-
metry, incorporated in bone plates. BROWN et.al. (6) also
describe a battery powered hip implant, used in one patient.
BARLOW et.al. (1) are developing an inductively powered tele-
metry for artificial hip jOints. This telemetry has to be
implanted separately and the 5G's have to be glued at the
implant's surface.
318

All these telemetries have at least some of the following

shortcomings: They are so big that they have to be implanted
separately from the prosthesis. Toxic batteries with restricted
capacity are employed. Unreliable plastic materials are used to
seal the implants.
STRAIN- AND FORCE- MEASUREMENTS
The SG's used for the measurement of deformations or loads
are usually arranged in Wheatstone Bridges. This demands a big
number of SG's, additional resistors and connecting wires,
mainly when measuring spatial forces. The implant becomes big
and has to be manufactured very accurately; the SG's have to be
glued with extreme precision (2,3). A use of OP-amplifiers
requires additional components, a powerful supply, and causes
drift problems.
The MATRIX-method (2,3) reduces the number of
required SG's and connecting wires. Although the transducer can
be fabricated with less accuracy, the measuring accuracy is
much improved. This method is best suited for applications with
small implantable telemetries to measure spatial forces or
de form a t ion s ..
A new electronic circuit for gauging the SG-signals had to
be developped for our telemetry. The principle (4) of the
"CURRENT-DIFFERENCE-METHOD" is as follows (Fig.1): An upper
current source sets the current IR inversely proportional to
the value of the resistor R. The current ISG" through the lower
sources is regulated by the values of the Imultiplexed three
SG's and a temperature reference. Then the remaining current
difference In is a function of the resistance change and here-
with of the strain of the activated SG. This circuit has
several advantages, compared to OP-amplifiers:
1. Minimum power demand, which allows an inductive supply.
2. Small number of electronic components and small size
3. Stable long term measurements
4. Easily expandable to multichannel measurements
TELEMETRY
A detailed description of the telemetry circuit is found
in (4). The device transmits a sequence of pulses, and the time
between these pulses is a nonlinear measure for the SG-values
and the temperature. The sampling rate can be chosen according
to the given requirements. With semiconductor SG's of 350 Ohm,
the circuit consumes only 6 mW. With a linearisation between
the SG-signals and the force components prior to the calibra-
tion, an accuracy of 1% for strains and of 3% for spatial
forces (plus 3% max. for neglected friction) is always
achieved.
The telemetry is fabricated in thick-film hybrid technolo-
gy with bonded chips and resistors (Fig.2). The two sides of a
6.9 by 15 mm large ceramic substrate carry 14 chips and 20
passive components. This circuit is built in the cylinder of
Fig.3, which also contains the inner inductive coil. This
cyrrnder improves the power supply and protects the circuit
mechanically and electrically. Only the the SG's and the
antenna need to be connected to this final telemetry.
 319

IMPLANT DESIGN
When built into a titanium prosthesis for a sheep hip
(Fig.4), the power required for the outer inductive coil is
aoour- lOW at 4 kHz. Two electrical feedthroughs connect the
telemetry with the antenna inside the ceramic head. These
feedthroughs from pacemakers are clinically well proved and
shield safely against moisture. Electron beam- and laser wel-
ding techniques are employed to mount the feedthroughs in the
upper closing plate and to assemble the plate and the prosthe-
tic neck.
EXTERNAL EQUIPMENT
Due to the untuned antenna of the implant, the
transmittance range is small. Therefore the receiving antenna
has to be placed directly on the skin. This requires connecting
cables as they are also necessary for the induction coil around
the hip joint (Fig.5). The receiver transfers the pulses to a
microprocessor system for counting and preprocessing. A real
time calculation of the forces will become possible with a
faster computer than used presently.
ANIMAL TESTS
These instrumented hip prostheses were implanted in three
sheep (Febr. 1985, Febr. and June 1986) and are working since
now (Dec. 1986). Prosthesis and plastic socket were cemented.
Before and after the operation, the animals were trained three
times a week on a treadmill, which is also used for the
measurements. Force determinations start one to three days
after the operation and are performed regularly. Data are taken
at walking speed up to 9 km/h, when the animals stand on four
and three legs and when they are lying on the side.
RESULTS AND DISCUSSION
The measured forces were similar to that found in previous
investigations with wire-connected sheep implants (2). Some of
the major findings were:
A. During the stance phase, the resultant joint force acts
in the long direction of the femur, with small anterior-
posterior and mediolateral components. The load direction is
very Similar to that found in humans (9).
B. The reSUltant force varies widely. It reaches typical
maximum values between 90% and 125% of the body weight (RW)
during the first postoperative weeks.
C. It increases towards 90% BW to 140% BW after two or three
month. (one example in Fig.6).
D. The jOint loading increases more with the p.o. time than
with a higher walking speed.
E. Occasional force peaks can reach much higher values. This
occurs most frequently during the first p.o. time, when the
walking is still somewhat handicapped, but also lateron
(example with 240% BW in Fig.7).
F. Joint forces may vary durrng-one run (Fig.7), for different
runs on one day under the same conditio~and on different
days. The variability between different animals is of similar
magnitude as are the intraindividual differences.
G. When the sheep lies on the side with an unloaded leg, the
320

resultant is about 4% BW to 10% BW, slight passive movements

of the leg,however, result in big forces (Fig.8) of up to
140% BW. Remarkably are the big transverse-rDrCes in these
cases.
H. In a standing position, the force on the prosthetiC leg is
typically 20% to 35%, but may increase towards 150% and more
by muscle contractions (Fig.9) or by shifting the animal
towards the affected rear ~
Not all possible peak forces and uncommon force directions
can be detected by such occasional observations. But two major
findings are typical for all animals: Often peak forces occur
during normal walking and standing, which are much higher than
usual and can only be caused by antagonistic muscle activities.
This is most frequently observed during the first p.o. months.
Force calculations will probably never detect such effects.
Even body positions which were thought to cause small
joint forces may lead to high loads, as the passive motion of
the sheep's leg in a sidewards lying position shows. These
findings are probably also valid for humans and may complicate
the future development of improved implants, designed to cause
stresses which are tolerated better by the bone than with the
present prostheses.
REFERENCES
1. Barlow, J.W .. I.F. Goldie, J.M.K. Horwood, A.J.C. Lee, R.P.
Ramson: "In vivo records of strain in a total hip joint", I.
Mech. Eng., C216/84 (1984) 55 - 61
2. Bergmann, G.: " Ein Dreikomponenten- Kraftaufnehmer und
seine Anwendung fUr die Messung von Gelenkkriiften am Tier"
Ph.D., Tech. Univ. Berlin (1981)
3. Bergmann, G. J. Siraky, A. Rohlmann, R. Kolbel: "A
comparison of hip joint forces in sheep, dog and man", J.
Biomech. 17 (1984) 907 - 921
4. Bergmann, G., F. Graichen, J. Siraky, H. Jendrzynski, A.
Rohlmann: "Multichannel strain gauge telemetry for
orthopaedic implants",submitted to the J.Biomechanics (1986)
5. Brown, R.H., A.H. Burstein, V.H. Frankel: "Telemetring in
vivo loads from nail plate implants", J. Biomech. 15 (1982)
815 - 823
6. Brown, R.H., D.T. Davy, K.G. Heiple Sr., G.M. Kotzar, K.G.
Heiple Jr., J. Berilla, V.M. Goldberg, A.H. Burstein: "In-
vivo load measurements on a total hip prostheSiS", Proc.
31st Ann. ORS (1985)
7. English, T.A., M. Kilvington: "In vivo records of hip loads
using a femoral implant with telemetric output", J. Biomed.
Eng. 1 (1979) 111 - 115
8. Hodge, W.A., R.S. Fijan, K.L. Carlson, R.G. Burgess, W.H.
Harris, R.W. Mannn: "Contact pressures in the human hip
joint measured in vivo", Proc. Natl. Acad. Sci., Vol. 83
(1986) 2879 - 2883
9. Rydell, N.W.:"Forces acting in the femoral head- prosthe-
sis", Acta Orthop. Scand., Suppl. 88 (1966)
ACKNOWLEDGEMENT
This work was supported by the Deutsche Forschungsgemeinschaft
 321

fig.4:
Instrumented hip joint im-
plant for sheep. Compo-
nents: Implant, telemetry,
3 semicond. strain gauges,
closing plate with feed-
throughs and antenna, cera-
mic head.

fig.5:
Sheep with measuring prosthesis, inductive power coil and
antenna on treadmill, left: measuring equipment.
 w
~

current
...:
<II
c-
8
r" otransm.
ISGi' I;=I R- ISGj
Id""I.\RSGj
ml.llti-
plexer

fig.1: Fig.2 : FIG.3:

Current-Difference-Method One side of telemetry; Metallic cylinder, contai-
for measurement of three thick film hybrid technique ning telemetry hybrid (in
strain gauges (5G) and one slot) inductive coil
temperature channel. (bottom) and solder pets.
 323

,..., 131i!Jr-~ _ _ __

N 98
L...J

a...
~
66
a
u
l:)
w
34
z
+
I-
--l

::J
(J)
W -31i!J~ .................. -Fx ___ -Fy ____ -Fz _ _ RES
______________________________________ ~________-!

-
0:::
m ... III
lSI
N ......
lSI
~ TIME [ MSEC ] Cl
Cl
N

Fig.6: Resultant hip joint force and components in percent of

lSI
-
body weight (76 weeks p.o., walking speed 5.5 km/h);
inserted: coordinate system.
241i!J~ ____________________________________ ~

,...,

N 186
L...J

~

a... 132
a
u
l:)
W
78
Z
+

~ 24ri~~~~~-+.r-~~~~~~~~~~~~~~~~~~~~~
Ul
W
0::: -31i!J+-___- -__-~__- -__- -__----__- -__- -__- __- - _ +
......
-
151
...
N
...
N

...
N
N
ffi T I ME [ MSEC ]
.......
10
151

Fig.7: Hip force and components ( 23 weeks p.o., 4.5 km/h);

run with varying forces and peak load.
324

83~=-=- ________ __________________________________

~ ~

NEUTRAL FORWARDS f'ORW F ORW

BACKW BACKW

N 63

0...
~
43

a
u
l:J
W
23
Z
+
~ 3~~~~~____________~~____~~~~~~__~\~!"~''''-_''~''~~'
..J
:J
(f)
W
~ -23~_~ _ ..........- ___________________________________________
-Fit - - - -Fy ____ -h _ _ RES -!

...
CSI
...
'<t
...
OJ ...~ TIME [ MSEC ]
Ol
N

...
CSI CSI CSI
N '<t U) CSI

Fig.8: Hip force and components (unloaded position on side,

leg is passively moved slightly forwards and backwards.

,.., lS0r-----------------------------------------------------~

N 110

'-'

Q.
~
70
a
u
l:J
W
30
Z
+
~ -10 \'J (V
I
5 V
(f)
-rx - - - -Fy
w
____----__----__----__--__----__----__----__--__
.................. ____ -F"z _ _ RES
...
~ -S0~ ----~

...
CSI '<t
TIME [ MSEC ] ;
...
01 1/1
01 01
1/1 Ol

Fig.9: Hip force and components, sheep standing on four legs;

two force peaks due to muscle contractions.
 325

A HYPOTHESIS CONCERNING THE EFFECT OF IMPLANT RIGIDITY ON

ADAPTIVE CORTICAL BONE REMODELLING IN THE FEMUR
R. HUiskes*. M. Dalstra*. R. v.d. Venne*. H. Grootenboer**
and LJ. Slooff*
Lab. Exper. Orthop.*. Div. Applied Mechanics**. University of
Nijmegen. Nijmegen. The Netherlands
1. INTRODUCTION

It is well known that bone is continuously active in a remodelling

process, whereby the status quo is a dynamic equilibr ium between bone
resorption and apposition. Due to several factors, the equilibrium confi-
guration of bone can shift towards a change in density (porosity) or a
change in shape. It is generally accepted that one of these causative
factors is related to the mechanical load. Abnormally high loads cause
increases, reduced loads decreases of bone mass. It is, thereby, impli-
citely assumed that there is, for each individual, such a thing as the
normal load level or interval. The effects of these load-induced remodel-
ling mechanisms can be found in active sportsmen, inactive patients and
astronauts, and can be reproduced in animal experiments (Currey, 1984).
Although the remodelling mechanism is relatively well established, the
feed-back link, the signal, between load and remodelling is not. If the
bone senses, locally, a change in load, then this feed-back mechanism must
be based on local deformations, which can be described by strains and
related to stresses.
It is widely assumed, that these strain-adaptive remodelling phenomena
also play a role in bone-mass reductions which are seen in patients around
fracture fixation plates (Tonino et al., 1976) and intramedullary pros-
theses (Harris et al., 1976). These phenomena (osteoporosis or osteno-
penial are often loosely related to "stress shielding", suggesting a
direct link between a loss of bone and a reduction of stress, caused by
load-transfer through the prosthesis rather than bone, which is thereby
partly "shielded". Evidently, the degree of shielding depends on the
structural characteristics of the implant/bone configuration, and thereby
on prosthetic design.
The purpose of the present investigation is, to develop an empirical
mathematical model which can be applied to predict the extent of bone
adapt ions around intramedullary fixated joint prostheses, based on the
structural characteristics of the prosthesis and the initial bone confi-
guration, and on the external load. This development is based on animal
experimental studies on the one hand, and finite element analysis based on
principles of adaptive elasticity (Cowin, 1981) on the other.
This paper reports on the characteristics of the computer models used
and the results of bone-remodelling simulation analyses of a simplified
intramedullary fixation structure, whereby the stiffness of the stem was
related to the possibilities for strain normalisation in the surrounding
bone.

2. MB'THODS

The principles of adaptive elasticity were developed by Cowin and

Hegedus (1976). In this theory a separation is made between internal
326

remodelling (a change of bone porosity whereby the elastic modulus is

affected) and external or surface remodelling (a displacement of the bone
surface). Central to the theory is the remodelling equation, which
relates the bone (surface) growth rate u (mm/sec) to the local strain
patterns (Cowin, 1981)
o
U = Cij ( Eij - Eij), (1)

o
whereby Eij is the strain tensor in the surface point considered, and Eij
a reference strain field in the same point; Cij are surface rate remodel-
ling coefficients. A comparable relation is assumed for internal remodel-
ling.
In the present analysis, the same principles are adapted, but two
alternative choices were made. In the first place, not the strain tensor
is used as a feed-back control variable, but the strain energy density

(2)

where crij is the local stress tensor, and Eij the strain tensor.
In the second place, following a suggestion by Carter (1984), it is
assumed that bones are nlazyn or, in other words, that a certain threshold
level in over- or underloading must be exceeded before the remodelling
process takes any effect. This is illustrated in Fig.1. The local strain
a;n energy density as a function
of time and location U(t,x)
is referred to a local refer-
ence (natural) value UnIx);
.~
25 when U > (1 + s)U n -or
u
o U < (1 - s)U n , net changes in
~ Un Strain bone stock occur, whereby
C> t--,~;---+---L....l...__ Energy
(1 - s) Un ~ U ~ (1 + s) Un is
<: Dens; ty the inactivity range. Hence
for external remodelling, the
local surface displacement
rate u dx/dt (mm/sec,
dx =ld~l, d~ perpendicular to
the bon~ surface):
loss
Fig. 1: Bone remodelling velocity as a function
of strain energy density.

u = I
C { U - (1 + s) Un }
0
C{U - (1 - s)Un }
U > (1 + s) Un
(1 - s) Un < U < (1 + s) Un
U "< (1- s)Un
(3)

For internal remodelling, the elastic modulus rate of change e = dE/dt

(MPa/sec) is described in the same fashion.
Computer programs to evaluate the changes in modulus and geometry based
on a change in strain energy density were developed and combined with the
finite element code NONSYS, as illustrated in Fig.2 (Dalstra, 1986). To
determine the geometrical or material changes, the remodelling rates are
integrated, using forward Euler integration, with a constant time step
lit. The values of C and 1I t have no physical significance as yet, but
 327

were chosen depen-

Boundary Cond it; ~ns
ding on the initial

Load
*
Geometry - - - - - - - - - - - - - - - , ot
maximal
value of Iu - unl
Material in such ~ way t&at
a reasonable
approximation of
the exponential
time/remodelling
curves could be
obtained. In this
study, 2-D plane-
strain finite
elements were
Fig. 2: Scheme, showing the relation of the re- applied. The
modelling programs with the finite element program was tested
system. on a number of
engineering structures for which the mathematically optimized geometry is
known analytically.
The subject of the present analysis is a simplified, general model of
an intramedullary fixation structure (Huiskes, 1980; Huiskes and Chao,
1983), illustrated in Fig.3. The model is two-dimensional but is con-
structed in such a way, using a side-plate
element layer and non-uniform element
thickness, that a precise simulation of a
3-D, axisymmetric geometry is obtained
(Huiskes, 1980). The only external load
is pure bending of the stem. In the case
of external remodelling, the strain energy prox.
density in all the surface nodal points of
the bone are compared to the "natural"
values, based on which they are diplaced
inwards or outwards in each iterative
step, perpendicular to the surface. In
order to maintain adequate element shapes
the nodal points within the bone are dis-
placed as well (Oalstra, 1986). A minimum
cortical thickness of 0.2 mm is always
maintained. The thickness of the side-
plate was varied too, in order to maintain
the relationship between 2-D and axisym-
metric geometry at all times.
Simulating internal remodelling the dist.
analysis is similar. However, in this
case the average strain energy density in
the element is used to determine the Fig. 3: The simplified intra-
change in elastic modulus. medullary fixation model
In order to investigate the effects of used in the analysis.
implant stiffness, the stem thickness was
varied, simulating diameters of 10, 13.3 mm (cemented), and 20 mm (non-
cemented). The threshold levels s were varied between 0, 5, 15 and 30%.
The "natural" strain energy density values Un were determined for the
"bone" cylinder without the stem. Co-Cr-alloy was chosen for the stem
material (Es = 2*10 5 MPa). The (initial) bone modulus is 2*10 4 MPa, the
cement modulus 2*10 3 MPa.
328

3. RESULTS

For the intermediate stem thickness (d s = 13.3 mm, flexural rigidity

Fs = IsEs = 3.06*10 8 Nmm 2 ) the course of the surface remodelling process
is shown in Fig.4 (s=O). The outer radius of the bone decreases to the
default value of 10.2 mm at the utmost proximal side in 5 iteration steps
(Fig.4.a). However, a stable end configuration is reached over the
greater part of the bone, whereby about half of the cortex has disappear-
ed. From a significant strain energy density reduction (Fig.4.b: curve 0

Fs: 3.06 _108 . 0-/. ~.3.o6.,oI.OI.

GPROB c GPR08

dis!. prox. d dis!. prox.

Fig.4: Left: Development of the outer-bone radius after subsequent

remodelling steps.
Right: Development of the strain energy density at the surface
after subsequent remodelling steps.
illustrative for "stress shielding"), an almost "natural" distribution is
reached after 9 iterations. It must be appreciated in interpreting these
graphs, that some inaccuracies at the utmost proximal side are the result
of inadequate mesh density for the high stress gradients in this region.
The end configurations for the two other stems are shown in Fig.5. The
most flexible stem causes bone
resorption at the proximal side
only. The stiff, medula-filling
cementless stem causes all bone to
disappear along the stem.
The effect of the threshold
level s on the end configurations
was a gradational one only. With a
high threshold level, the process
would develop slower and stop
earlier or, in some spots, not
start at all. Of course, the
threshold model causes the end
configuration to be a non-unique
0.ge solution in principle.
The development of internal
remodelling was different in the
sense that a stable end-configura-
Fig. 5: End configurations of the tion could be reached only for the
remodelled bone, assuming most flexible stem and a high
three stem stiffnesses. threshold level.

Fig.6 illustrates, for a stem flexibility of Fs= 1.32*10 8 Nmm 2 and a zero
threshold, that extensive reductions in elastic modulus produce only small
increases in strain energy density, indicating that strain normalisation
through this process is bound to fail.
 329

:;:~~---~----~---------o

FS :1.l2.108 :Q% FS =1.lZ. 1aB.O%

6 PROB
PROB

disl. prox. disl. pro.

Fig.6: Left: Development of the elastic modulus in the cortex after

subsequent remodelling steps.
Right: Development of the strain energy density in the cortex
after subsequent remodelling steps.
4. ANALrl'ICAL CONSIDERATIONS

It was found previously that the simple axisymmetric model applied in this
case behaves in accordance with the theory of beams-on-elastic-foundation,
whereby the middle part behaves as a composite beam (Huiskes, 1980). Con-
sidering the bone as a beam in bending, the maximal strain energy density
at the surface can be written as

(4)

where M is the bending moment, Ro the outer bone radius, I = ;; (Ro - Ri)
11 4 4

the second moment of inertia, E the elastic modulus, and the Poisson's
ratio. When a stem is inserted then, provided composite beam theory is
valid,

U (5)

where Fs= EsIs is the flexural rigidy of the stem (Huiskes, 1980).
Whereas in the natural case (4) the bone will always be able to reach
any strain energy density level by adapting either the modulus E or the
outer radius Ro, in the prosthetic case (5) this relationship is not so
trivial. Fig.7 shows a 3-D graphical representation of eq. (5), as
functions of the outer radius Ro and the apparent density Pb (g/mm3,
related to elastic modulus as E = 2.9*10 3 Pb (Carter and Hayes, 1977)),
for the dimensions of the present model, for different values of the stem
flexural rigidity Fs, cut-off at the level of the natural strain energy
density Un.
As shown in Fig.7.a (Fs= 1.6*10 B), strain normalisation is easily
obtainable by the bone by reducing its cortex with appr. 1 mm, but only
barely by reducing its modulus (or, as indicated here, its density).
When Fs= 4.25*10 8 (Fig.7.b) a boundary situation occurs, beyond which
strain normalisation becomes impossible with any method. Density reduc-
tions would even reduce the strain energy density further.
330

~t'
S.O

40

3.0
:}

's = 1.60*10'
E.--:-I.~=E"I=--b'-'41
r::1

Fig. 7: Three-dimensional graphics presentation of ego (5), as a function

of outer bone radius and apparent density, for two different stem
flexural rigidities.
It can be shown analytically that the boundary values for internal re-
modelling are reached when Fs= EI/4, and for external remodelling when
Fs= EIRi/Ro, neglecting, in this case, a possible threshold range (s=O).
These results are confirmed, in a large sense, by the results of the
computer simulations, in particular where it concerns external remodel-
ling. They explain why remodelling cannot result in strain normalisation
when the stem is very sti ff, and why internal remodelling is so inef-
fective, in the case of a stiff stem even increasing in stead of decreas-
ing the amount of stress shielding.

5. DISCUSSION AND OONCLUSIONS

There should be no question about the hypothetical nature of the

present model, which is based on a relatively large number of assumptions,
which cannot be suppported by experimental data at the present time.
These include the strict separation of internal and external remodelling,
the choice of strain energy density as a feed-back control variable, the
(partial) lineari ty of the response, the threshold levels. Additional
weak points of the theory applied are the consideration of static loads as
gener ic representations of dynamic loading character istics, and the con-
centration of bone response to strain characteristics in a single point.
The actual FE model applied in the present case is, of course, of a very
rough, generic nature.
Nevertheless, the present approach opens the option of actually pre-
dicting the response of bone to "stress shielding" by implants, based on
empirical quantitative descriptions, verifyable (or falsifyable) with
experiments.
The present model, however simplified, gives good insight into the
phenomenon of "stress shielding", and into the mechanical consequences of
bone response, in a gener ic sense, thereby enabling the development of
more precise definitions in this respect.
The most interesting actual findings of the present simulation are,
first that internal remodelling, or osteoporosis in this case, is an in-
effective route towards strain normalisation. And second, that a boundary
value for the stem stiffness can be defined, beyond which strain normal i-
 331

sat ion becomes impossible for the bone by any means, suggesting that the
bone resorption process would be an instable one, leading to complete bone
removal. Examples of this have actually been found in animal experiments
with cementless hip-joint replacement (Miller, 1984).
The present results suggest that this boundary value can be found from

Fs = EIRi/Ro, (6)

hence from the material and geometrical characteristics of the bone,

independent of the load. Although highly hypothetical at this moment,
these findings are certainly verifyable in animal experiments.
Thirdly, and finally, it appears from the present simulation studies,
that whatever the flexibility of the stem, a certain amount of proximal
bone must disappear, if the concept of "stress shielding" is accepted as a
driving mechanism.

ACKNOWLEDGEMENT: Sponsored in part by Grant Nr. 95-189 of the Netherlands

Organization for the Advancement of Pure Research (ZWO).

REFERENCES

1. Carter DR, and Hayes WC: The compressive behavior of bone as a two-
phase porous structure. J. Bone Jt. Surg. 59-A(1977)954-962.
2. Carter DR: Mechanical loading histories and cortical bone remodel-
ling. Calc. Tissue Int. 36, Suppl.I(1984)S9-S24.
3. Cowin SC: Continuum models of the adapt ions of bone to stress. In:
Mechanical properties of bone (SC Cowin, Ed) AMD-vol.45, The Am. Soc.
of Mech. Engrs., New York, 1981.
4. Currey J: The mechanical adapt ions of bone. Princeton University
Press, Guilford, 1984.
5. Dalstra M: De ontwikkeling van een tweetal computermodellen voor het
simuleren van adaptieve botremodellering. MS-thesis. Univ. of Twente
I Univ. of Nijmegen, The Netherlands, 1986.
6. Harris WH, Schiller AL, Scholler JM, Freiberg RA, and Scott R: Exten-
sive localized bone resorption in the femur following total hip
replacement. J. Bone Jt. Surg. 58-A(1976)612.
7. Huiskes R: Some fundamental aspects of human joint replacement. Acta
Orthop. Scand. Suppl. 185, 1980.
8. Huiskes R, and Chao EY: Optimal stem design in tumor prosthesis. In:
Tumor prosthesis for bone and joint reconstruction (EY Chao and JC
Ivins, Eds) Thieme Stratton Inc., New York, ch.44, pp.367-382, 1983.
9. Miller J, Montreal General Hospital, Private Communication, 1984.
10. Tonino AJ, Davidson CL, Klopper PJ, and Linclau LA: Protection from
stress in bone and its effects. J. Bone Jt. Surg. 58-B(1976)107.
 333

STRESS TRANSFER ACROSS THE HIP JOINT IN RECONSTRUCTED

ACETABULI
R. Huiskes and T.J. Slooff
Lab. Experimental Orthopaedics, Dept. Orthopaedics, University
of Nijmegen, Nijmegen, The Netherlands

1. IR'rRODOC'rION
The results of cemented acetabular reconstruction have been somewhat
disappointing relative to femoral reconstruction, in view of loosenings
reported on the longer term (e.g. Stauffer et al., 1983).
The causes for these late loosening effects are not all clear, although
evidently fibrous tissue interposition at the implant/bone interface,
diagnosed as radiolucency, occurs more often in the acetabulum than in the
femur (DeLee and Charnley, 1976). Acetabular loosening is relatively
frequent in relation with surface replacements (Strens, 1986), hence,
either effects of friction or cup flexibility could play an important
role. The negative effects of high cup flexibility on cement stresses have
been emphasized in finite element stress analyses (Pedersen et al., 1982:
Carter et al., 1982: OOnishi et al., 1986: oonishi et al., 1983). 'Metal
backing' of polyethylene sockets appeared to be a solution to that problem
(Harris and White, 1982).
Others have pointed to the negative effects of reaming the subchondral
bone layer, also in relation with loss of acetabular rigidity (Charnley,
1979). Conversely, the positive effects of subchondral penetration of
acrylic cement for the strength of the cement/bone bond have been emphas-
ized. High curing temperatures of acrylic cement, in particular in aceta-
bular fixation, have been mentioned as well (Huiskes, 1980: Eriksson,
1984). Based on the sometimes disappointing clinical results of cemented
cups, cementless fixation was introduced, for instance by using threaded
sockets to be screwed into the acetabulum (Lord, 1979).
The purpose of the present study was to analyse the load-transfer
mechanism through different types of acetabular reconstructions, including
cemented and cement less polyethylene sockets, with and without subchondral
reaming, metal backed sockets, surface replacement cups and threaded fixa-
tion. In addition, several fundamental questions related to finite element
(FE) modelling of acetabular reconstruction in general were addressed:
i.e. the applicability of 2-D FE models versus solid axisymmetric models,
and the effects of elastic coupling between femoral head and socket on
acetabular stress patterns and hip-joint friction.

2. IIB'ftIODS
The acetabular configuration analysed was based on the FE model of
Pedersen et al. (1982). This model describes a frontal section through the
acetabulum and the adjacent pelvic structures, applying axisymmetric
elements, allowing for non-axisymmetric loading (FOurier expansion).
The models applied in the present study are shown in Fig. 1. They
include the natural acetabulum, reconstruction with the cement less CLW*

* CLW is a trade mark of protek AG, Bern, Switserland.

FIGURE 1: FE meshes used in the analyses, with femoral head included,
contact coupled to the acetabulum (compressive stress only)
a: Natural acetabulum: b: conventional sockets:
c: surface replacement: d: CLW threaded cup:
e: local model (2-D) for-CLW cup.
 335

threaded cup, conventional polyethylene cups, and a surface replacement

cup. As in the model of Pedersen et al. (1982), axisymmetric solid ele-
ments were used, but in this case the femoral head was included in the
models, contact coupled to the acetabular cup (Brown and Dig ioia, 1984 ~
Rapperport et al., 1985). As a penalty for this inclusion, only symmetric
loads could be taken into account. In addition, calculations were carried
out, whereby a non-axisymmetric load was assumed (Fourier expansion),
without the femoral head. The CLW reconstruction (being the most rigid of
the cups analysed) and the surface replacement cup (being the most flex-
ible one) were also analysed with 2-D plane-strain elements.
In the global CLW reconstruction model (Fig.1), the region of the
threads was modelled as a composite material of bone and metal. The
detailed stress patterns in and around the threads were analysed with a
local, refined model, for which the boundary conditions were derived from
the global model (Fig.l). The material of the metal ring in the CLW model
was varied between CoCrMo alloy and c.p. titanium.
The conventional cup model was analysed with several variations, inclu-
ding cemented UHMWPE, with and without metal backing, with and without
subchondral reaming, and noncemented UHMWPE with subchondral reaming.
Elastic moduli used are shown in Table I (see also Fig.l).

Nr. Material/Variation E(MPa*10 4 )

1 Cortical bone 1 .7
2 Cancellous bone 0.3
3 Cancellous bone 0.15
4 Cancellous bone o. 1
5 CoCrMo alloy 20.0
6 UHMWPE 0.07
7 Titanium~ CoCrMo alloy 11 0 ~ 20.0
7a Composite bone/metal 5.5~ 10.0
8 UHMWPE~ CrCoMo alloy ~ PMMA 0.07~ 20.0~ 0.3
9 UHMWPE~ PMMA~ cortical bone o. 07 ~ O. 3~ 1.7
10 Cancellous bone 0.5

'I'ABLB I: Elastic moduli used in the models~ bone moduli according to

Pedersen et ale (1982).

All materials were assumed linear elastic, homogeneous and isotropic.

In the axisymmetric models, quadrilateral 8-node isoparametric elements
were used. plane strain quadrilaterals, also 8-node, were used in the 2-D
models. In the case of non-axisymmetric loading, ten Fourier terms were
applied. All mater ials were assumed rig idly connected, except for the
femoral head/cup bond, where only compressive stress was allowed. Stress
results are presented per unit of applied force.

3. RESULTS

3.1. Effects of the femoral head

The femoral head presents a restraint for the deformation of the
cup. This is illustrated in Fig.2, showing cup deformations for the case
that the head is included, and for the case that a distributed load is
directly applied to the polyethylene socket. It is also evident from this
figure that the metal CLW ring behaves almost like a rigid body, relative
to the socket and the bone.
336

The head/socket interaction

causes the compressive-stress dis-
tribution at the inner socket to
depend on the elastic characteris-
tics of the acetabular reconstruc-
tion (Fig.3). This dependency also
has an effect on the stress pat-
terns in the cup, the bone and at
the cup/bone interface. However,
when a relatively rigid recon-
struction is used, as in the case
of the CLW cup, the bone stress
FIGURB 2: patterns only depend on the direc-
Deformations of the CLK-cup: tion of the external force, and
left: femoral head included, not on the precise pressure dis-
right: load application directly tr ibution at the inner cup boun-
to inner polyethelene boundary. dary.

-3
I------i 10 MPa/N

FIGURE 3: Pressure distributions at the head/cup boundary calculated for

the cemented polyethelene cup (left), the surface-replacement
cup (middle), and the CLW cup (right).

3.2. Solid versus 2-D models

The differences in the stress patterns between the axisymmetric solid
models and the 2-D plane strain models are the more pronounced in the case
that the reconstruction is more flexible: i.e. the actual quantitative
differences are greater in the latter case (Fig .4). In the 2-D surface
replacement model the element thickness was taken as 50 mm, which results
in approximately equal compressive stress distributions at the inner
socket boundary in the frontal plane, in both the 3-D and 2-D cases.
Evidently (Fig.4), the stresses in the 2-D model are grossly overestimat-
ed. This overestimation is more extensive the farther away from the cup.
Although the stress patterns are qualitatively similar, the 2-D model
underestimates the contribution of the cortical bone layer surrounding the
acetabulum, in particular at the lateral rim (hoop stress) and the medial
wall. It overestimates the contribution of the superior cancellous bone.
 337
o < 2.S.10'MPa/N

02.5- 5.0.10" ..
DSO-1S.10' ..
!EEZS-l0 .10" ..
FIGURE 4:
Com par ison of von
Mises equivalent
stress patterns in
the frontal bone
section of a 3-D
(axisymmetric) and
a 2-D model of the
surface replace-
ment cup recon-
struction.

3.3. Effect of symmetric versus asymmetric load

Fig.S shows von Mises stress patterns in the bone of the CLW recon-
structed acetabulum, assuming a one legged stance maximal force orienta-
tion (pedersen et al., 1982), and a symmetric force, in both cases distri-
buted over the cup boundary. The femoral head was included in neither of
these models which has, in the case of a relatively rigid reconstruction,
no concequences for the bone stresses (sect. 3.1.). The general effect of
the medially rotated force is an almost uniform increase of stresses in
the superior, lateral bone part, and a corresponding decrease in the infe-
rior, medial part. This indicates that the symmetric load is adequate to
analyse general trends in the bone stress patterns.
In a more flexible reconstruction, the differences may be more pr~
nounced. However, the present findings for the conventional cup recon-
structions, using a symmetric load, compared to the corresponding results
of pedersen et al. (1982), who used a one legged stance load, indicate
that also in this case the differences are not dramatic in terms of
general trends.

Dc:n ..._".
CJtS-!l.o.'IO~_

~-n.11'_
~_10.t1'_

FIGURE 5: Von Mises equivalent stress patterns in the frontal bone

section, as determined for the CLW-cup model. Left a symmetric
load, right an asymmetric load.

3.4. Effects of cup design and fixation technique

The effects of cup design and fixation technique on the stress pat-
terns in the bone are evaluated relative to the natural case, shown in
Fig.6a. In this comparison, only the models with the femoral head included
(symmetric load) are considered.
338

o < 2.5..tD'MPaIN
C]z.S.S.O .,......
1I!iIIIIso-...1If' .
_75-10.10"",.
_>10.10"' ..

Dc2.5 .... MPaIM

c::::Jt!-5.hll"'.
ClSO-'S.""'.
~_10 ..r_

FIGURE 6:
Von Mises equivalent stress pat-
terns in the frontal bone sec-
tions of various models.
a: natural acetabulum~
b: conventional cup reconstruc-
tions with and without PMMA, sub- b
chondral reaming and metal back-
ing.

As evident from Fig.6.b, there is hardly any difference at all between

the stress patterns in the acetabulum with the cemented and the noncement-
ed UHMWPE cups. Although the cement elastic modulus is more than 4 times
higher than the one of UHMWPE, the cement layer is relatively thin, hence,
these reconstructions are structurally about equally stiff. Relatively to
the natural case, stress concentrations occur at the superior cup/bone
interface. More load is transferred through the central cancellous bone
directly to the medial/superior cortex, than via the cortical shell.
Conversely, when the subchondral bone layer is left intact (Fig.6.b),
the stress patterns are very similar to those in the natural case
(Fig.6.a) ~ the stress concentration at the interface, ocurring in the
reamed model, has disappeared.
When metal backing is applied on a
cemented polyethylene cup (Fig.6.b),
2 the central cancellous bone region is
stress-shielded relative to the natural
acetabulum, to a considerable degree.

In this case, even more load is trans-
2 1 ferred directly to the lateral cortical
<!: shell and less load is transferred
~ H-++--++~ through the superior cup/bone inter-
::E face. As a result, the central inter-
face and cement stresses in relation
PE-MB
with the metal backed cup are much
lower than in the non-metal backed case
(area 1 in Fig.7). This mechanism,
FIGURE 7: described earlier (pedersen et al.,
Compressive and shear stresses 1982~ Carter et al., 1982), has been
at the cup/cement interface in considered as the mechanical advantage
two regions (1 and 2), with of metal backing. However, this advan-
and without metal backing (MB). tage has a penalty in increased inter-
 339

face and cement stresses close to the lateral cup rim (area 2 in Fig.7),
both in shear and in compression. The gain in area 1 (in the superior
region) from metal backing is in these models even less than the loss in
area 2 (near the cup rim)
The bone stress patterns in the surface-replacement model (Fig.4, 3-D)
are not very different from those in the conventionally reconstructed ace-
tabulum (Fig.6.b, PE (reamed. The cement/bone interface stresses are
even better spread in the former case, and more load is directly trans-
ferred to the latter cortical shell. This seems surprising in view of the
higher flexibility of the surface-replacement cup. In fact, this is the
result of the deformation restraining action of the stiff femoral head,
which is relatively large in this case.
In the case of the threaded cup reconstruction (Fig.5, symmetric load),
stress shielding of the central cancellous bone relative to the natural
case (Fig.6.a) is evident. Also in this case more load is transferred
directy to the lateral cortical shell. Stress concentrations occur in
small areas near the infer ior and the super ior parts of the threaded
ring. From the stress patterns in and around the threads, determined in
the local FE model (Fig .1) it is found that most of the load is trans-
ferred through the first and the last threads.

4. DISCUSSION AND CORCLOSIORS

In earlier FE analyses of acetabular reconstruction (pedersen et al.,
19821 Carter et al. 19821 Oonishi et al., 1986), the external load was
directly applied at the inner cup boundary. In the present analysis it was
found that this is an unrealistic assumption, because of elastic coupling
between the femoral head and the socket. When the cup has a relatively
stiff metal backing, the effect of the coupling is limited to the direct
(polyethylene) environment of the head, but in the case of a flexible
reconstruction the bone stress patterns are affected also.
The effect of the femoral head restraint is the most pronounced in the
case of the surface-replacement reconstruction. It could be said, that the
femoral head, in this case, acts like a 'pseudo metal backing', thereby
reducing the negative effects of the high cup flexibility. It is possible,
therefore, that the relatively unfavorable clinical results of surface
replacement cups are the effects of friction, rather than cup flexibility.
Because the head/socket contact pressure distribution depends on the
elastic characteristics of the reconstruction, the head/socket friction
will also be susceptible to the elastic characteristics, and thus vary
with prosthetic design.
In a generic way, the use of symmetric -loads versus asymmetric loads or
2-D models versus solid FE models can be useful and appropriate to eval-
uate trends in the stress patterns comparatively. The effects of these
simplifying assumptions, however, depend on the flexibility of the recon-
struction: the more rigid the cup, the less the bone stress patterns are
susceptible to the modelling characteristics. In all cases, however, the
stress values are grossly overestimated in 2-D models, depending on the
assumed thickness of the 2-D elements.
The acetabulum behaves like a 'sandwich' structure, in the sense that
most of the load is transferred through the cortical Shell, and the inner
cancellous bone basically serves to keep these shells apart (Jacob et al.,
1976). This behavior is also evident in the normal stress patterns
(Fig.6.a). The reconstructions can be divided according to those which
violate this behavior (the flexible ones like all conventional cups with
subchondral bone layer reamed) and those which essentially leave the
structural integrity intact (the stiffer ones like the CLW cup, metal-
340

backed cup, or any cup with an intact subchondral layer). The latter
category seems to be the favorable one, although there is a penalty in all
cases: local bone stress concentrations in the CLW cup, increased cement
stresses with the metal backed cup in the lateral rim region (Fig.7), and
the difficulty of cement/bone interlocking when the subchondral bone layer
is retained.
It is well feasible that optimal designs could be developed, based on
compromises between the various conflicting aspects. Eventually, these
compromises must be based on a better understanding of the three-dimen-
sional mechanical characteristics of the acetabulum, in particular in
cases which are representative for the patient population.

ACKNOIrUmGBMENT:
This research project was partly sponsored by protek AG, Bern, Switzerland

1. Brown TO, and DiGioia AM: A contact-coupled finite element analysis of

the natural adult hip. J. Biomechanics 17(1984)437-448.
2. Carter DR, Vasu R and Harris WH: Stress distribution in the acetabular
region - II Effects of cement thickness and metal backing of the total
hip acetabular component. J. Biomechanics 15(1982)165-170.
3. Charnley J: Arthroplasty of the hip, theory and practice. Berlin:
Springer Verlag, 1979.
4. DeLee JB and Charnley J: Radiological demarcation of cemented sockets
in total hip replacement. Clin. orthop. ReI. Res. 121 (1976)20-23.
5. Eriksson AR: Heat-inducted bone tissue injury. Dissertation, univer-
sity of G~teborg, Sweden, 1984.
6. Harris WH, and White RE: Socket fixation using a metal-backed acetabu-
lar component for total hip replacement. J. Bone Jt. surg. 64-A
(1982)745-748.
7. Huiskes R: Some fundamental aspects of human joint replacement. Acta
Orthop. Scand. Suppl. Nr.185, 1980.
8. Jacob HAC, Huggler AH and Dietschi C: Mechanical function of subchon-
dral bone as experimentally determined on the acetabulum of the human
pelvis. J. Biomechanics 9(1976)625-627.
9. Lord GA, Hardy JR, and Kummer FJ: An uncemented total hip replace-
ment. Clin. Orthop. ReI. Res 141 (1979)2
10. OOnishi H, Isha H, and Hasegawa T: Mechanical analysis of the human
pelvis and its application to the artificial hip joint. J. Biomech-
anics 16(1983)427-444.
11. OOnishi H, Tatsumi M, and Kawaguchi A: Biomechanical studies on fixa-
tions of an artificial hip joint acetabular socket by means of
2D-FEM. In: Biological and biomechanical performance of biomaterials.
(P Christel, A Meunier, and AJC Lee, eds.) Amsterdam: Elsevier Science
Publishers BV, 1986, pp.513-518.
12. Pedersen DR, Crowninshield RO, Brand RA and Johnston RC: An axisym-
metric model of acetabular components in total hip arthroplasty. J.
Biomechanics 15(1982)305-315.
13. Rapperport OJ, Carter DR, and Schurman OJ: Contact finite element
stress analysis of the hip joint. J. Orthop. Res. 3(1985)435-446.
14. Stauffer RN: Ten-year follow-up study of total hip replacement, J.
Bone Jt. surg. 64-A(1983)1983.
15. Strens PHGE: Analysis of implant failure in the Wagner resurfacing
arthroplasty. Dissertation, University of Nijmegen, The Netherlands,
1986.
 341

STRESS ANAlfSIS Of THE UNCEMENTED DSl ~IP PROSTHESIS

B. ~asser*. S. Tepic** and S.M. Perren*
M.E.M.-Institute for ~io~echanics*. University of Sern. Bern.
Svitzerland. lab. for Exper. Surgery**. Davos. Switzerland

1. INTRODUCTION
loosening is one of the major factors affecting the long term results
of total hi p arthropl asty. Considerable research effort (l) is presentl y
devoted to systematical improvements of prosthesis geometry. material
properties and surface structure. The goal is always a better fit between
prosthesis and bone in order to create an optimal situation for
appropriate load transfer through the prosthesis - bone interface and to
reduce micromot ion. Most of thi s research. however. is focused on
variations of the surface design as conducting grooves or ribs without
changing the traditional geometry of the stem. Consequently the load
transfer from prosthesis to bone remains more or less concentrated at the
stem ends. The risk of interface stresses exceeding its limit of stability
remains and relative movements will occur in unstable contact areas
causing bone resorption and unavoidable prosthesis loosening.
A new, uncemented and dynamic seH-locldng stem (DSL) system (2) has
been developed. which controls the load transfer by design rather than by
choice of material
properties. Th; s DSL -
stem increases its
thi ckness under phys i 0-
logical loads. The stem
(fig. 1) consists of a
medial and lateral lobe,
connected by thin struts
which are incl ined with
respect to the long axis
of the prosthesi s. When
loaded, because of
tension in the lateral
and compression in the
medial lobe the struts
Fig. 1: prototype DSL-prosthesis are deformed and rotated
and the two lobes sepa-
rate. Thus compression and friction at the contact area between the lobes
and the endosteal surface is increased. Due to the specific geometry of
the stem, already very small amounts of expansion lead to the high contact
stresses necessary to inhibit motion. The prosthesis, preloaded by
insertion, locks itself further under physiological loads.
The goal of this analysis is to examine the basic concept of self -
locking and to vary the geometry of the lobes and struts for optimal
contact stress distribution in the interface of the prosthesis and the
inner wall of the medullary canal. Optimal interface stresses will ensure
342

local stability according to demand.

2. MATERIAL AND METHODS

The analysis was done using the Finite Element (FE) program ADINA.
Mechanical properties and the stress behaviour of the stem as well as the
prosthesi s and bone interface system were investigated. Taking advantage
of the symmetry a simplified, 3-dimensional FE - model applicable for
2-dimensional load cases was generated using the digiti zing tabl et and
mesh generation program FEMGEN. The mesh of the whol e prosthesi s - bone
system includes in addition to the implant, cortical and cancellous bone.
The entire model (fig.2) consists of 6219 nodes and 905 3-dimensional,
20-node solid elements.
Of these, 228 elements
were used to model the
stem of the prosthesis,
551 and 126 elements
respectively were
generated to describe
the cortical shell and
the cancellous bone in
the trochanter region.
The ventral and dorsal
cortical wall of the
model were assumed to be
of uniform thickness.
Only the lateral and the
med ia 1 surfaces of the
stem contact the bone.
v-J.-x The e 1 emen t s i n the
interface were fixed to
each other, because the
Fig. 2: prosthesi s-bone FE-model FE - code used had no
gap analysis available.
This implies that separation or slippage between implant and bone could
not occur. Normal stresses limit the interface friction. Shear stresses
exceeding friction indicated slippage.
The material behaviour was assumed to be 1 inear el astic. A titanium
based alloy is chosen for the prosthesis stem. Relevant material
properties are as follows:

Young's mod ul us: Poisson's ratio:

prosthesis 100'000 N/mm~ 0.3

cortical bone 20'000 N/mm 2 0.3
cancellous bone 100 N/mm 0.0

Taking advantage of the symmetry was possible because the loads were lying
in the model's plane. The joint force was reduced to a concentrated com-
pression load of 2600 N on the head of the prosthesis incl ined 25 0 to the
stem axis. Muscle forces were simulated by a concentrated tension load of
2000 N applied at the tip of the greater trochanter and inclined 30 0 to
the stem axis.
To experimentally verify the results of the FE - analysi s a prototype
prosthesis (fig. 1) was machined in titanium. The prosthesis surface is
treated by sand blasting to enhance interface strength.
 343

In order to vary geometrical parameters such as the strut distance,

strut thickness and strut incl ination, and to investigate their infl uence
on the contact stress distribution, special FE - model s were generated.
Analysis of these smaller models reduced considerably the calculation time
on the computer. The optimization of the strut distance is reminicent of
the problem of elastic beam foundation, where one lobe is considered as
the beam, the cortical shell acts as an elastic foundation, and the struts
are repl aced by concentrated forces. The deformation 1 ine of the beam -
foundation interface is a harmonic function. The ratio of the strut
distance to the fundamental wavelength should influence the contact stress
distribution. After correction of the interface geometry and refinement
of the mesh, the final
model "elastic" (fig.
3) was wnposed of a
quarter tube simu-
lating the femur and a
beam mod ell ing one of
the stem lobes. The
mesh con s i st s of 276
elements, whereof 226
describe cortical
bone. This simplified
model takes advantage
of three pl anes of
symmetry. With a con-
Fig. 3: FE - model "elastic" centrated force lying
in two of these planes
and being parall el to the strut direction, the wavel ength L of the
deformation 1ine could be determined. To get a better strut force
simulation, the concentrated force was additionally replaced by prescribed
displacements with the effect of a more appropriate displ acement
distribution over the width of the lobe. The following equation (1)
yielded the preferred strut di stance a:
a = Lin with n = 1,2,3,... (1)

The strut thickness and inclination variation was done using a smaller
FE - model "strut" (fig. 4) being
part of the whole prosthesis - bone
system. Thi s model incl uded struts
number 5 to 7, if numbering in
proximal direction. The mesh
consi sts of 856 nodes and 106
3-dimensional, solid elements, 72 of
them were used to generate the bone.
The loading of this model was chosen
to produce the same deformation as
the whol e prosthesi s - bone system.
Thus the resulting displacement
components of the whol e model
,~, calculation were used as loading in
the di stal and proximal sectional
Fig. 4: FE-model "strut" plane. Analyzing this model only the
normal stresses in the interface
were investigated to find the parameter situation with the most optimal
344

compression area in the contact zone. The original model had a strut
thickness of 1.2 mm and a strut incl ination of 14 0 with respect to the
normal of the stem axis. Strut thickness d was varied up to 2.0 mm, while
the incl ination of the strut was checked every 4 0 between 10 0 and 26 0 A
combination with both parameters enlarged (d = 2.0 mm, incl. = 20 0 ) was
also investigated.

3. RESULTS
The deformation of the whole prosthesis and bone system yielded a
maximal displacement of 2.4 mm in medial direction at the head of the
prosthesis. However, this result strongly depends on the free bending
length, which in comparison to real ity is too short in the mathematical
model. It can be seen that the struts, espec ially those in the distal
region are rotated. Thi s results in the widening of the prosthesi s stem.
The struts are deformed in an s-shape. This reduces the widening, but can
be countered by originally, negatively s-shaped struts. The effective
amount of widening was calculated as difference of the displacements of
facing nodes of the FE -
mesh. The curve (fig.
5), which shows the
widening of the stem
along its whole length
confirms the effect of
sel flocking. The
maximal amount of
widening of the original
model is 10 micro
meters. The width of the
stem becomes enl arged in
only the distal
two-thirds. However,
this is exactly where
the forces must be
transmitted from stem to
x bone. The interface
v-t normal stresses are cal-
culated as the average
Fig. 5: Widening along DSL - stem of the adjacent element
stresses. In the distal
part of the lateral interface (fig. 6) only approximately 40% of the
contact zone is under compression.

2
Fig. 6: Distal part of lateral interface: normal contact str sses
(tension area diagonally hatched; E = 0, D = -2 N/mm )

Larger compression zones are only at locations where struts push again~t
the lobe. The compression stress peak is normally not more than 2.0 ~/mm ,
only at the distal prosthesis end does it reach values of 20.0 N/mm The
shear stresses had different directions in the distal and proximal parts
because ~f the influence of the trochanter loading. The amount was about
1. 0 N/mm or 1ess. As a consequenc e of the load ing, the max imal stresses
 345

at the prosthesis stem could be found at the loading point. In the stem
regi~n only some bend ing stresses exceeded 50 N/mm2 but never reached 250
N/mm . T2e behaviour of the von Mises stresses was similar, they exceeded
250 N/mm at the loading point only. The value of 250 N/mm2 is half of the
The FE - analysis of the whole system on the VAX 111780 took almost 24
hours, so it was a useful simplification to realize the parameter
variation with the models "elastic" and "strut". Their calculation time
was only 1 and 2 hours respectively. The model "elastic" was corrected two
times during calculation to get a more accurate stress distribution in the
interface. The first two models had bone elements with box geometry,
whereby the second one had a refined mesh in the contact zone. The third
and final model took the curvature of the lateral and medial stem surfaces
in the transverse plane also into consideration. The refinement of the
mesh had minimal influence ( less than 3% ) on the determination of the
hal f wave length, which was 112 mm. With consideration of the curvature,
the compression area could be increased more than 15%. The loading applied
as prescribed displacements yielded larger compression zones than the
concentrated forces. Thi s was because of a better load distribution. The
strut distance variation yielded an increased compression area with
decreasing strut distance. Figure 7a shows the contact zone in case of a

Fig. 7: Model "elastic":

EIHlilHliElHliflH ~ Normal contact stresses:

a) average of prosthesis
and bone
(F = 0, E = -2 N/mm2)
Fig. 7 (a: above, b: below)
b) average of bone
el ements
(0 = 0, B = -2 N/mm2)
(tension area diagonally
hatched)
strut distance of 10 mm corresponding to the original prosthesis. Loaded
with a concentrated force, the compression area was approximately 82% of
the interface surface. The strut distance of 14 mm and 7 mm (equation (1))
resulted in compression areas of 75% and 88% respectively. A 94%
compression area was the result when loaded with prescribed displacements
every 10 mm. If normal stresses in the contact zone (fig. 7b) are averaged
only with stress values belonging to bone elements, there is full contact
in the interface.
The model "strut" was loaded to be deformed as the corresponding part
of the whole prosthesis - bone system. Thus displacements in both x- and
y- direction were exactly the same and the bending stresses were also very
similar. The distribution of stresses perpendicular to the long axis,
having the same direction as normal stresses in the interface, had some
differences in comparison to the whole model. Therefore only parameter
variation model s were compared to each other. The infl uence of the
variation of the strut angle from 10 0 to 26 0 on the normal stress
distribution in the contact zone was so small that it could be neglected.
Also an increase in strut thickness of 0.2 mm did not result in a modifi-
cation of the compression area, which was 50% on the lateral and 40% on
the medial side of the model. Even the combination of the two gossibi-
1 ities of parameter variation with a strut incl ination of 20 and a
thickness of 2.0 mm instead of the original 14 0 and 1.2 mm did not
346

increase the compression zone in the interface.

4. DISCUSSION
The analysis proves, that the basic concept of the DSL prosthesis stem
functions as expected. The resul ts show that further refinement of the
design will be needed to assure an optimal contact area. The struts in the
proximal part are almost perpendicular to the stem axis or even inclined
in the opposite direction. No widening is achieved, nor aimed at in the
proximal region. The separation of the two lobes of the original model
results in a normal force of 1700 N. Considering a coefficient of friction
of 0.38 a total shear force of 650 N can be transferred. With further
parameter optimization, normal and shear forces should suffice to prevent
shearing displacements. A parameter which still has to be investigated to
get larger compression areas in the contact zone is the thickness of the
stem lobes. It must be kept in mind that the analysis with the adjacent
prosthesis and bone elements fixed to each other is not very accurate.
Instead of a smooth contact stress distribution over the length of the
stem, stress peaks were located at the distal end and accordingly the
amount of the stresses in the remaining regions was lower. In addition it
could be establ ished, that compression areas determined with bone el ement
stresses alone were 1arger. Thus it can be expected, that in real ity
normal and shear forces must be greater than results of the FE - analysis.
The investigation of maximal and von Mises stresses show, that the
strength of the prosthesis stem is sufficient, but additional experimental
testing will be necessary.
The bending stiffness of the DSL stem is certainly lower than that of a
full stem of equal dimensions. This is considered an additional, bene-
ficial feature for a gradual load transfer fran prosthesis to bone.
The strut distance variation yielded no special influence on the
contact stresses using preferred distances corresponding to equation (1).
This may be explained with the fact that strut distances were much smaller
than the wavelength.
The s - shaped deformation of the struts suggests a possibil ity to
improve generation of interface stresses by cutting the struts in negative
s-shape. The thickness variation proved, that a general increase of
thickness does not result in any improvement. The variation analysis of
strut incl ination and thickness had no infl uence on the stress
distribution in the contact zone.
Further investigations of the stem geometry and optimization of
parameters, such as lobe thickness, are required. A more accurate
determination of contact stresses and an increased, maybe only a more
harmonized, widening is necessary for sufficient safety against loosening.
However, the basic principle of the dynamic self - locking stem seems to
offer a good possibil ity to prevent motion in the interface of an
uncemented hip prosthesis.

REFERENCES

1. Rohlmann A., Moessner U. et al.: Die Beanspruchung des Femur nach

Hueftgelenkersatz; Z. Orthop. 121 (1983).
2. Tepi c S., Perren S. M.: Dynamic Se If-Loc ki ng Stem for Hi p Prosthesis;
submitted to 33rd ORS meeting.
 347

INFLUENCE OF POROUS COATING THICKNESS AND ELASTIC MODULUS ON

STRESS DISTRIBUTION IN HIP PROSTHESES
A. Rohlmann**. E.J. Cheal* and W.C. Hayes*
Harvard Me~.Schoo~*. Beth Israel Hasp . Boston. MA. U.S.A.;
Free Unlv.Berlln**. Oskar-Helene-Heim. Berlin. Germany

1. INTRODUCTION
Porous coating of femoral component stems has been widely
used in an attempt to improve fixation of prosthesis to bone
and to allow the use of total hip replacement in younger, more
active patients. Several groups have shown that bone tissue
grows into the pores of inert porous materials, thus allowing
a three-dimensional interlock between implant and bone. Such a
fixation has the ability to transfer tensile stresses across
the interface between implant and bone.
Different materials for porous coating have been investi-
gated including titanium, Co-Cr- alloy, carbon, ceramic, and
several plastics (e.g. Proplast, polyethylene). Hip implants
with coatings of titanium, Co-Cr alloy and Proplast have been
used in clinical trials. The elastic modulus of these
materials varies widely as does the thickness of the layer.
Ducheyne et al. (1978) analyzed a two-dimensional model of
a porous coated stem surrounded by trabecular bone. The thick-
ness of the porous metal fiber coating was 0.5 mm. They varied
the elastic modulus between 2,000 and 50,000 MPa and found
that the elastic modulus of the porous coating did not
significantly affect the interface stresses. The influence of
a structured stem surface on the local stress distribution was
studied by Huiskes and Nunamaker (1984). They reported that
the stiffness ratio of bone and implant had a very strong
influence on local stress concentrations.
The objective of our study was to investigate the influence
of coating thickness and elastic modulus on the stress distri-
butions in porous coating and adjacent bone.
2. MATERIAL AND METHOD
We used a geometrically simplified three-dimensional finite
element model consisting of 1110 8-node hexahedral elements
and 1582 nodal points (Fig.1). The femoral diaphyseal bone was
assumed to be a linear, homogeneous and isotropic tube with an
outer diameter of 30 mm and inner diameter of 20 mm. The
prosthetic stem was also assumed to be axisymmetric with a
constant diameter of the solid core of 17 mm. The baseline
thickness of the porous coating was 1.5 mm. The total length
of the model was 184 mm. The model was clamped at the distal
end and loaded proximally with a transverse force F of 1 kN. A
perfect bond between the material components was assumed. The
elastic moduli were chosen as: 1) 18,000 MPa for cortical
bone; 2) 200,000 MPa for the solid prosthetic stem; and 3)
348

DISTAL PROXIMAL

A B
F

c o
FIGURE 1. Hidden line view of the finite element model and
schematic sketch of the longitudinal section.

5,000 MPa for the porous coating. To determine the effects of

coating thickness and elastic modulus, the thickness of the
porous layer was varied parametrically between 0.5 mm and 4.5
mm and the elastic modulus between 100 MPa and 40,000 MPa.
Calculations were performed on a VAX-11/750 computer using the
finite element program ADINA.
3. RESULTS
3.1. Str.esses at the interface
Figure 2 ShoWS-the prIncIpal and the octahedral shearing
stresses at the bone and implant side along the line A-B of
the bone - prosthetic stem interface. The maximum principal
stresses dominate, whereas the minimum and intermediate
principal stresses are very small. The octahedral shear stress
represents the state of shear stress and is proportional to
the von Mises stress. On the bone side the curves for the
dominant principal stress and the octahedral shear stress are
of the same form but different magnitude. On the prosthetic
side the peak for the octahedral shear stress is much more
pronounced than that for the maximum principal stress.
Along the line C-D, the octahedral shear stresses are the
same, but the principal stresses are different. Here the mini-
mum principal stress dominates and the maximum principal
stress is very small.
 349

STRESSES AT THE INTERFACE

PRINCIPAL STRESSES OCTAHEDRAL STRESSES

BONE SIDE A- B BONE SIDE A- B

PROXIMAL
0'+---------------------------
DISTAL PROXIMAL

PROSTHETIC SIDE A- B PROSTHETIC SIDE A- B

w
~ ~
~ ~ W
~ 0
~
~
0 , PROXIMAL
5
~
i
-~ 0
0 oDISTAL PROXIMAL

FIGURE 2. Principal (left) and octahedral shear (right)

stresses on the bone (top) and prosthetic (bottom) side along
the line A-8 of the bone - prosthetic stem interface.
Quadratic symbols: Pl; circular symbols: P2; triangular
symbols: P3.

3.2. Influence of the layer thickness

Figure 3 shows IKe TnTTUence of the porous layer thickness
on the octahedral shear stress at the interface between bone
and porous coating. On the bone side the octahedral shear
stress along the line A-8 increases with the layer thickness.
The greater the thickness, the less pronounced the peak stress
at point A. On the porous layer side an increase of the porous
layer thickness results in a decrease of the octahedral shear
stress near point A.
3.3 Influence of the elastic modulus of the porous coating
The Influence or-the elastIc modulUS or-the porous coating
on the octahedral shear stress at the interface between bone
and porous coating is shown in Fig. 4. The elastic modulus of
the porous layer has only a small influence on the stresses on
the bone side. On the prosthetic side, an increase in elastic
modulus leads to an increase of the octahedral shear stress.
350

INFLUENCE OF COATING THICKNESS ON

OCTAHEDRAL SHEARING STRESSES AT THE INTERFACE

BONE SIDE A- B PROSTHETIC SIDE A- B

STRESSES STRESSES

MAX. STRESS VALUE ~ 21.46 MPa MAX. STRESS VALUE = 23.35 MPa
MIN. STRESS VALUE ~ 1.2B MPa MIN. STRESS VALUE = 0.65 MPa

FIGURE 3. Influence of the thickness of porous coating on the

octahedral shear stress at the interface between bone and
prosthesis along the line A-B. The X-axis represents the
longitudinal axis of the model and the V-axis corresponds to
the thickness of the porous layer. Results are plotted in the
Z-direction. Point A is in the VZ-plane. The curve in the XZ-
plane represents a layer thickness of 0.5 mm. All results are
related to the difference between maximum and minimum value.

An increase in elastic modulus from 100 MPA to 40 GPa

decreases the peak octahedral stresses by 28 % on the bone
side and increases that value by 700 % on the prosthetic side.
For a high elastic modulus the stresses are proportional to
the bending moment resulting from force F.
4. DISCUSSION
4.1. Thickness of the coating
An lncrease--i-n--the layer thickness reduces the peak
stresses at the level of the tip of the prosthesis in the
porous coating at the bone - prosthetic stem interface. The
maximum stress value in the bone is not influenced by layer
thickness. Cook et al. (1985) found in an animal study no
differences in the interface shear stress for a porous coating
thickness of 0.5 mm in comparison to those of 1.0 and 1.5 mm.
 351

INFLUENCE OF ELASTIC MODULUS OF POROUS COATING ON

OCTAHEDRAL SHEARING STRESSES AT THE INTERFACE

BONE SIDE A- B PROSTHETIC SIDE A- B

STRESSES STRESSES

MAX. STRESS VALUE = 22.B1 MPa MAX. STRESS VALUE = 32.33 MPa
MIN. STRESS VALUE = 1.01 MPa MIN. STRESS VALUE = 0.06 MPa

FIGURE 4. Influence of the elastic modulus of the porous

coating on the octahedral shear stress at the interface along
the line A-B. X-axis: longitudinal axis of the model; V-axis:
elastic modulus of the porous coating; l-axis: octahedral
shear stress. The curve in the Xl-plane represents the results
for an elastic modulus of 100 MPa.

4.2. Elastic modulus of the coating

The elastIc moduTUs---of the porous coating has a strong
influence on the stresses in the coating. The peak stresses
increase with an increase of the elastic modulus. However, the
stresses on the bone side are more crucial for long-term fixa-
tion. Here the trend of the stress values is opposite to the
trend of the elastic modulus of the coating. The femur is
mainly loaded in bending. For this kind of loading the
dominate principal stress is nearly in the direction of the
bone axis. The stress component perpendicular to the interface
between bone and implant is much smaller. The force from the
prosthesis to the bone and vise-versa is mainly transmitted by
shear as shown by Huiskes (1980) and Rohlmann et al.
(submitted for publication). A layer with a low elastic
modulus may reduce small stress concentrations due to the
small force perpendicular to the interface but it will
increase the dominant principal stress and the shear stress in
bone.
352

The deformation of a layer with a low elastic modulus is

much higher than of a layer with a high elastic modulus.
Therefore the relative motion in the porous layer between the
coating material and the bone spicules is high for coatings
with a very low elastic modulus. Additionally, the strength of
materials with a very low elastic modulus is often also very
low. Therefore, our results suggest that the use of a low
elastic modulus material provides no advantage when the
prosthesis is loaded in bending. Poor clinical results for hip
endoprostheses with a porous low modulus stem coating (Tullus
et al., 1984) seem to support this conclusion.
In summary, both thickness and elastic modulus of the
coating have a strong influence on the peak stresses in the
porous coating and only a small influence on the stresses in
the adjacent bone.
REFERENCES
Cook SO, Thomas KA, Luedemann RE, Anderson RC, Haddad Jr RJ:
Short Term Mechanical and Histological Characteristics of
Carbon-coated and Uncoated Porous Titanium Implants. Trans.
ORS, 195, 1985.
Ducheyne P, Aernoudt E, deMeester P, Martens M, Mul ier J , van
Leeuwen 0: Factors Governing the Mechanical Behavior of the
Implant - Porous Coating - Trabecular Bone Interface. J.Bio-
mechanics 11, 297-307, 1978.
Huiskes R: Some Fundamental Aspects of Human Joint
Replacement. Acta orthop.scand. Suppl. 185, 1980.
Huiskes R, Nunamaker 0: Local Stress Patterns and Bone-
Remodelling at Implant - Bone Interface. Trans. ORS, 102,
1984.
Rohlmann A, Cheal EJ, Hayes WC, Bergmann G: A Non-Linear
Finite Element Analysis of Interface Conditions in Porous
Coated Hip Endoprostheses. J.Biomechanics (submitted for
publication).
Tullos HS, McCaskill BL, Dickey R, Davidson J: Total Hip
Arthroplasty with Low-Modulus Porous-Coated Femoral Component.
J.Bone Jt.Surg 66-A, 888-898, 1984
 353

A METHOD FOR THE COMPARATIVE EVALUATION OF THE INITIAL

STABILITY OF CEMENTLESS HIP PROSTHESES
J. Eulenberger*. E. Schneider*. W. Steiner**. D. Wyder* and
R. Friedman*
M.E.M.-Institute for Biomechanics*. Univ. of Bern. Bern.
SWitzerland. Regional Hospital**. Interlaken. Switzerland

1. INTRODUCTION
The major problems and causes for revision in total hip arthroplasty
are mechanical loosening of the acetabular and femoral component,
infection and implant fracture. Stauffer (15), in a ten-year evaluation
of 333 total hip replacements, found a 29.9% incidence of roentgeno-
graphic femoral and an 11.3% incidence of acetabular component loosening.
Al though these were early cases, other authors show similar trends in
their long-term studies (10). With the development of cementless total
hip prostheses, one hopes to alleviate these problems.
Motion of a hip stem relative to the femur can cause bone resorption
at the implant/bone interface (13) and might lead to further loosening of
the implant, compromising its long term success. In the uncemented
prostheses, motion at the interfaces exceeding the amount of strain
tolerated by the repair tissues, might lead to an unstable implant and
inhibit "biological" fixation ( 14). Consequently, it may be suggested,
that excessive motion present immediately postoperative under
physiological loading, plays an important role for the long-term success
of cement less hip prostheses. This motion could therefore serve in the
assessment of such implants.
In the evaluation of total hip arthroplasty it has been traditional to
use hip scores such as the Harris (4), which reflect the clinical and
functional results achieved. More recently, a new score has been
published by Kavanagh and Fitzgerald (5), which also includes radio-
graphic assessments of radiolucent lines and subsidence of the femoral
component. Instability has further been quantified clinically by
measuring the motion between hip prostheses and bone from standard x-rays
for several designs of implants. Green and Aufranc (3) presented a
retrospective study of the long-term follow-up of 78 Austin Moore
prostheses, implanted without cement. They found a mean amount of
settling of 16.3 mm, not specifying, however, the contributions of the
femoral and acetabular components. Subsidence of the femoral prosthesis
was investigated by Loudun and Charnley (6), comparing a conventional and
a flanged cemented Charnley prosthesis. They found subsidence (i . e .
vertical displacement exceeding 1.6 mm) in 43 % of the normal, but only
5.3 %of the flanged types. This technique, unfortunately, is only useful
to detect relatively large displacements and other authors therefore used
the more precise method of roentgen-stereophotogrammetry (1). Mjoberg and
co-workers (8) measured the position of the implant relative to the bone
with loaded and unloaded hips in vivo. The load was applied by standing
or compressing the hip with the patient in a supine position. They
defined significant displacements as 0.4 mm in the transverse plane and
1 .6 mm in the long axis of the bone and found 6 of 28 implants to be
354

instable. They never observed instability in the vertical axis alone.

They further observed migration of the stem of up to 4.8 mm after 9
months. In a second study (9), the hips of the patients were loaded in
compression (200-250 N) and internal/external rotation <30-35 Nm). They
found instability in 10 of 24 implants with displacements of up to 6 mm
(subsidence) and demonstrated the importance of rotational stress for
detection of instability.
In vitro, Charnley and Kettlewell (2) measured the sinkage of
uncemented prostheses in a quasi-static experiment under axial load by a
dial gauge and found them to be 1. 1 mm for each incremental load of
450 N, but reduced 200 times, if cement was used. Markolf et al. (7)
determined the flange/calcar and the stem tip/femoral shaft relative
displacement in 16 cemented prostheses loaded quasistatically for
analysis of the effect of a prosthesis collar. They found subsidence of
less than 0.1 mm at the tip of the stem under loads of 2000 N,
independent of calcar contact, but a considerably increased motion at the
calcar, when support was removed.
In light of the many types of uncemented implants presently available
or under development, the need was felt to design an in vitro dynamic
method for the determination of the initial stability of hip prostheses
under axial and torsional loading.

2. ME'11IODS AND MATERIALS

A given prosthesis is implanted according to its clinical protocol and
by means of the original surgical instruments in an entire cadaveric
femur. On completion, the femur is sectioned at a distance of 18 cm
distal to the lesser trochanter and potted in a special fixture (Fig. 2).
The proximal end of the prostheses is connected through a load cell to
the cross-head of an INSTRON materials testing machine. The entire
fixture is mounted on the actuator of the testing machine. The fixture
consists of an x-y-table and a spherical bearing, allowing the
transmission of axial and torsional loads in the direction of the long

,
axis of the bone and still allowing the distal end of the bone to move
y

,
(mm) 004
4

2 001
,
002
003

'J
_V I
I
t t
1
t TOl t T02 T03 T 04

20 40 60 80 T
(min)

Fig. 1: Example of Y-displacements versus time of one specimen,

illustrating the total displacements 'ID1 to TD4 and the dynamic
displacements DD 1 to DD4 for each of the four loading stages.
Transition zone at the beginning of each loading stage due to
setting of the new load values.
 355

freely in the transverse plane. Sinusoidal loads acting in the long axis
of the bone and synchronous rotatory moments around it are applied in 1+
stages (Fig. 1). The amplitudes of the axial loads are increased in
multiples of the bodyweight of the donors (1-, 2-, 3-, 4-fold) to allow
for individual variation of the specimens, the amplitude of the torque is,
8 Nm in each of the stages. Each stage has a duration of 20 minutes and
the loading frequency is set to 0.5 Hz, resulting in a total number of
600 test cycles per stage. The displacement of the tip of the bone is
measured by resistive extensometers in x- and y-direction (Fig. 2),
allowing to assess the tilting of the prosthesis in the frontal and
sagittal plane respectively. The z-displacement of the prosthesis
(sinkage) is measured as the displacement between actuator table. and
prosthesis by means of the in-built actuator transducer. The rotation "an

Fig. 2:
Photograph of ex-
perimental setup with
the prosthesis
connected through
flanges ( 1 ) to the
testing machine. X-
and y-displacements of
the distal end of the
femur (2) measured by
linear resistive
transducers (3), the
rotation of the
prosthesis relative to
the bone by precision
potentiometer (4).
356

of the prosthesis relative to the proximal femur is measured by a

goniometer clamped to the femoral cortex. The conventions are: positive
x-displacements mean valgus-rotation of the prosthesis for left and right
femora, positi ve y-displacements signify dorsal tilting of the head of
the prosthesis for right and vice-versa for left femora, positi ve z-
displacements describe extraction of the prosthesis out of the femur,
positive rotations result for external rotations of the prosthesis
relative to the bone for right and internal rotations for left femora.
The analog signals from the transducers as well as from axial and
torsional loads are digitized at a rate of 50 Hz with a resolution of 12
bits by means of a PDP 11/34 computer. From each cycle, the minimum,
maximum and average value of each signal is determined and stored. These
data are then transferred to a VAX 11/780 computer for display and
further processing. A sample plot of one displacement is given in Fig. 1,
showing the definitions for the total displacements TDl through TD4 and
the dynamic displacements DDl through DD4 for each of the loading stages.
The values for TDl to TD4 and DDl to DD4 are computed at the end of each
loading stage. For calculation of the group averages, the x- and
z-displacements are averaged with their respecti ve signs, the y-
displacements and the rotations are averaged by converting left to right
femora by change of sign.
The femora used for this investigation were removed within 24 hours
after death of the donors, sealed in double polyethylene bags and frozen
at _20 0 C. Before testing, they were thawed and kept moist throughout
implantation and testing. Three groups were established. Group 1
consisted of 19 Mueller straight stem prostheses implanted without
cement, group 2 comprised 6 straight stem prostheses implanted with
cement. This implant was selected for justification of the method,
because its design is based on a tight geometric fit, because there is
much clinical experience and success with this implant (11) and because
it has been implanted in a slightly modified version in small numbers as
an uncemented prosthesis (12). In group 3, six experimental, cementless
implants developped with collar by Prof. Mueller were used. The data of
the femur specimens used for this investigation are summarized in the
following table (means and ranges):

Group Age (yrs) Weight (kg) Sex (m-f) Side (L-R)

1 : str.st. uncem. 64.1 (42-85) 76 . 1 ( 51-89 ) 12 - 7 11 - 8

2: str.st. cern. 62.8 (50-73) 75.2 (69-82) 4 - 2 3 - 3
3: experimental 51.0 (25-64) 71.5 (61-77) 4 - 2 3 - 3

3. RESULTS
The total displacements and rotations TD4 are summarized for each of
the groups as mean (S.D.) in the following table:

x (mm) y (mm) z (mm) a (deg)

1: str.st. uncem. 0.88 (0.85) 1.28 (2.05) - 5.37 (2.90) - 0.94 (3.93)
2: str. st. cern. 0.32 (0.40) 0.54 (0.43) - 0.26 (0.04) 0.00 (0.01)
3: experimental 0.15 (0.18) 0.42 (0.92) - 0.26 (0.11) 0.04 (0.10)

In group 1 (uncemented straigth stem), 17 of the 19 prostheses tended to

rotate towards varus and 14 of 19 towards dorsal. The sinkage observed
ranged from -0.23 mm to -12.3 mm. The rotations occured in both
 357

directions (max. = 11.3 deg.) with a small tendency towards retroversion.

In group 2 (cemented straight stem), the prostheses also tended to rotate
towards varus (in all but 1 specimen) and towards dorsal for all but 1
(other) specimen, but at an amount of one third of group 1. The sinkage
was reduced by a factor of 18 and on average no rotation was observed
(max. = 0.13 degrees). In group 3 (uncemented experimental prosthesis)
the values found for the cemented straight stem could even be slightly
improved with the exception of a small average rotation and one specimen
showing a y-displacement of 1.94 mm.
The dynamic displacements and rotations DD4 for each of the groups are
also summarized as mean (S.D.) in the next table:
x (mm) y (mm) z (mm) a (deg)
1: str.st. uncem. 0.19 (0.11) 0.87 (0.67) 0.22 (0.07) 0.21 (0.15)
2: str. st. cern. 0.20 (0.12) 0.46 (0.45) 0.19 (0.09) 0.07 (0.05)
3: experimental 0.15 (0.09) 0.63 (0.84) 0.24 (0.08) 0.36 (0.15)
In group 1, the amplitude of the dynamic motion after 80 minutes of
testing ranged from 0.05 to 0.41 mm in x-direction, from 0.03 to 2.55 mm
in y-direction, from o. 15 to 0.4 mm in z-direction and from 0.0 to 0.56
degrees for rotation. All displacements showed a slight tendency to
increase over the loading stages, rotation tended to decrease. In group
2, the amplitudes ranged from 0.01 to 0.35 mm (x), from 0.01 to 1.20 mm
(y), from 0.03 to 0.28 mm (z) and from 0.01 to 0.11 degrees rotation. A
considerable reduction seems only apparent in the motion in the sagittal
plane. When comparing group 3 with group 2, the amplitudes of the dynamic
motion seem to be smaller in x-direction, but larger in the y- and z-
directions and especially in rotation.
4. DISCUSSION MID <DNCLOSIONS
As could be shown by the results of groups 1 and 2 (straight stem
prostheses) , the method seems easily able to detect the differences
between the stability of a prosthesis implanted with or without cement.
The technique of implantation did not seem to influence the results in
group 2 and 3, as identical results were achieved by different orthopae-
dic surgeons (in group 1, the same individual implanted all the prosthe-
ses). The test procedure is severe and probably exagerates the amount of
loading in the immediate postoperative time, on the other hand it may be
suitable to better demonstrate the differences between implants.
The results obtained are in agreement with the values quoted above and
underline, that instability should be avoided by any means. Once
migration has started, it is not very likely for many designs, that
stability can be achieved again. The good results obtained with the
experimental prosthesis (group 3) make this implant a good candidate for
successful clinical use. Although the bones used in group 3 were younger
than those of the other groups, its deSign seems to conform well to the
inner geometry of the bones tested.
A critical issue is the establishment of limits to discriminate
between stable and unstable situations. Groups 2 and 3 show, that implant
motions comparable to those obtained by cemented types may be achieved by
cementless implants, and measurements of types not reported here have
shown that clinically less successful implants also show higher
displacements/rotations. So far, the displacements achieved by the
cemented prostheses have been used to define the limits of stability, but
358

actually, it would appear more appropriate to correlate stability with

strain tolerances of the initial repair tissues (14). Such limits should
not be given by the accuracy of the measuring method.
In conclusion, the method seems sensitive and reflecting the actual
postoperative situation of the implant. Further types of prostheses will
be submitted to comparative testing and verification with clinical
experience attempted.
REFEIBfCES
1. Chafetz, N., Baumrind, S., Murray, W.R., Genant, H.K., and Korn,
E. L.: Subsidence of the femoral prosthesis. A stereophotogramnetric
eValuation. Clin. Orthop. Rel. Res. 201: 60-67, 1985.
2. Charnley, J., and Kettlewell, J.: The elimination of slip between
prosthesis and femur. J. Bone Joint Surg. 47-B(1): 56-60, 1965.
3. Green, D.L., and Aufranc, O.E.: Long-term follow-up of patients with
Austin Moore prostheses. South. Med. J. 72(3): 322-325, 1979.
4. Harris, W.H.: Traumatic arthritis of the hip after dislocation and
acetabular fractures: Treatment by mold arthroplastYi and end-result
stage using a new method of result evaluation. J. Bone Joint Surg.
51-A: 737, 1969.
5. Kavanagh, B.F. , and Fitzgerald Jr., R.H.: Clinical and
roentgenographic assessment of total hip arthroplasty. A new hip
score. Clin. Orthop. Rel. Res. 193: 133-140, 1985.
6. Loudun, J.R., and Charnley, J.: Subsidence of the femoral prosthesis
in total hip replacement in relation to the design of the stem. J.
Bone Joint Surg. 62-B(4): 450-453, 1980.
7. Markolf, K.L., Amstutz, H.C., and Hirschowitz, D.L.: The effect of
calcar contact on femoral component micromovement. J. Bone Joint
Surg. 62-A(8): 1315-1323, 1980.
8. Mjoberg, B., Hansson, L.L, and Selvik, G.: Instability, migration
and laxity of total hip prostheses. A roentgen stereophotogramnetric
study. Acta Orthop. Scand. 55: 141-145, 1984.
9. Mjoberg, B., Hansson, L.I., and Selvik, G.: Instability of total hip
prostheses at rotational stress. A roentgen stereophotogramnetric
study. Acta Orthop. Scand. 55: 504-506, 1984.
10. Morscher, E.: Introduction I. In: Morscher, E. (ed.): The cement less
fixation of hip endoprostheses. Springer, 1-8, 1984.
11. MUller, M. E.: Total hip reconstruction. In: McCollister, E. (ed ) :
Surgery of the musculoskeletal system 6:223-247, 1983.
12. Parpan, D., and Ganz, R.: Cementless implantation of a modified
Mueller straight-stem prosthesis. In: Morscher, E. (ed.): The
cementless fixation of hip endoprostheses. Springer, 156-160,1984.
13. Perren, S.M.: The induction of bone resorption by prosthetic
loosening. In: Morscher, E. (ed ): The cementless fixation of hip
endoprostheses. Springer, 39-41, 1984.
14. Perren, S.M., and Boitzy, A.: La differenciation cellulaire et la
biomecanique de l'os au cours de la consolidation d'une fracture.
Anatomia Clinica 1: 12-28, 1978.
15. Stauffer, R.N.: Ten-year follow-up study of total hip replacement. J.
Bone Joint Surg. 64-A(7): 983-990, 1982.
The contributions of Prof. W. Wegmann are gratefully acknowledged.
 359

CAD/CAM PRODUCTION OF CUSTOMER MADE BONE TITANIUM ALLOY

IMPLANTS BY MEANS OF CT IMAGES
C. Kranz and U. Ahrens
MECRON med. Produkte GmbH, Berlin, Germany

The enthusiasm which occured 25 to 30 years ago. when first arthroplastic

joints were implanted. cooled down. because they failed and made reoperations
necessary. The experience' shows that both cemented and cementless
prostheses loosen over a period of time.
The retrogression of the bone. which is mostly the reason for the failure. is
based on an unphysiological loading of the bearing tissue.

It would not be managable to insert and fix a prosthesis into the bone in a
way that the physiological loading path is not disturbed.
It is imagable to keep the change of the stressdistribution as little as possible
by modification of the stem stiffness.

~teasurements show that the combination of a three dimensional geometrical

modelation of a prosthesis stem with a low stiffness of the stem. provides a
nearly 100 % better stressdistribution in the corticalis of the femur. It is
possible to achieve a low stiffness of the stem by using a titanium alloy.
Because of this combination a much better physiological stressdistribution is
possible.
It is necessary to build up a large stock of ready made prostheses in hospitals
to supply all different types of human femura. This stock offers prostheses of
many different sizes and curves.

For a cementless implantation. US specialists publicize a so called "Press-Fit-

Situation", that means: keeping the contact surface between the implant and
the bone as large as possible.
Two different ways can be taken to obtain the largest area of contact surface.

1. Modulating the bone and using a well fitting prosthesis to keep the
resection of the healthy bone as low as possible

2. Exact design of the implant. which means the producing of an individually

custom-made prosthesis to maximize the bearing space, and to minimize the
resection of bone.
360

Our development-department has worked on hoth ~ay5.

The first one - mot1ulation of the bone - has led to a special drill system. which
allo\\'s a preparatIOn of the bone in the best way.
The thickness of the ready made prostheses increase in 2 mm intervalls.
The second \\'ay. the producton of an individual prosthesis will be described as
follo~s.

Description of the development project "Individual custom fitted,

implantable Prosthesis"

The idea, which will be described. is based on the connection of modern

medical technology and production techniques.
Here we have combined the images of the modern Computertomographie with
the technology known as CAD/CAM (Computer aided Design / Computer aided
Manufacturing)

The multi-layered picture of the unhealthy bone- or joint area- given by the
CT. allows an exact three dimentional view of the medular bone. and gives a
picture of the density and thickness of the surrounding cortical bone, too.
The CAD/CAM technolbgy makes it possible to size and process the data.With
assistance of the CAD Computer, a prosthesis stem can be designed, which can
be inserted into the medulary cannel under best conditions.
So called "over hangs" can occur as a problem which. make the insertment of
the stem very difficult or even impossible.
To avoid this problem, the CAD/CAM Computer can be used to simulate the
implantation as often as needed to guarantee no problems during surgery.
Either through remodulizing the prosthesis or through resection of the bone,
the best fitting situation will be found.
This simulation procedure provides the maximum of contact surface between
bone and prosthesis.
If the stem has a thin diameter, it might be necessary to control the fatique
limit of the stem by the Finite-Element-Method. This analysis also can be done
by the CAD Computer with the already known data.
After the definition of the three dimational prosthesis, a software program
needs to be created, which is able to work on a numeric milling-machine to
mill the stem accuratly out of a forged slug.
The method. briefly tiescibed in single steps. is as follows:
1. Preparation of a true 10 scale multi layered CT image of the sick femur
2. Data transfer of the CT -Data to the CAD Computer
3. Processing of the CT -Data into CAD Data
4. Construction and controlling of the prosthesis stem
5. Simulation of the insertment into the medular bone
 361

6. If necessary. controlling the fatique limit with the Finite-Element-Method

7. Conversion of the constructed stem into a NC-Program
8. Milling of a model on the NC-milling machine
9. If needed, production of instruments like graters. milling tools etc. for
surgery, by using the same data.

Remarks:
The principle of this method was developed in the Development and Research
Department of MECRON med. Produkte, and is ready to be implemented in
prod uction.
The following example will show how this procedure can he used to replace
parts of bone which have been distroyed by a tumor.

Description of a practical example

The following example illustrates the procedure:

Because it was the first clinical application, we have differed from the
planning, and have milled first a plastic model. With this model. the resection
has been determined in cooperation with the surgeon.

Picture I shows the x-ray of the patient who is suffering from a tumor. It is
easy to see that parts of the pubic bone and the acetabulum have been
resorbed by the tumor.
Picture 2 depicts a multi layered CT image of the area. After digitizing the
healthy part of the pelvis, a NC Program was made out of the mirror image.
This program reproduces the single layer outlines of the CT .
The plastic layers in picture 3, which have been all made in the same manner.
together form a complete pelvis (Picture 4).
With the newly created complete pelvis, the resection was discussed with the
surgeon (picture Sl.
Picture 6 shows and compares the part of the plevis which will be replaced
with a healthy pelvis.
The finished, milled titanium implant (picture 7 ), here put into the pelvis
model. should be fixed with plates at the healthy pelvis. This kind of fixation
was required by the surgeon. It is also possible to produce the fixation
elements in one piece with the implant.
Picture 8 shows the surgery (here during a test implantation on a human
corpse) . The surgery was successful and the patient is well.
With this example, we simply wanted to demonstrate the flexibility of the
method.
Right now, we are beginning to combine different modern technology. We will
go on with the development, so that in future, much more hip-prostheses can
be produced, individually.
362

The problem of the individual design of an implant can be solved by a high

quality of CT technique, ami by the connection with the precise production
technique with help of the CAD/CAM methOd.

It will take 1-2 weeks to produce such a prosthesis. assumed that there is a
good woorking relationship between the company and the surgeon.
That means that this method can be lIsed for a high amount of medical
indications, where immediate treatment is not required.

Picture 1

Picture 2
 363

f-o Picture 3

Picture 4

f-- Picture 5
364

f=- Picture 6

Picture 7 --+

Picture 8
f--
 365

ON LABORATORY EXPERIENCES IN ENDURANCE TESTING OF ANCHORAGE

STEMS OF HIP JOINT ENDOPROSTHESES
A. Walter, M. Bergmann, E. Reithmeier, K. Weinmann and
w. PIitz
Labor fur Biomechanik und Experimentelle Orthopadie an der
Orthop. Klinik der Universitat Munchen, Munchen, Germany

High dynamic bending and torsional stresses may give rise to

fatigue and stress corrosion failures of prosthetic stems of to-
tal hip joint devices in case of loosening, which up till now
remains one of the major problems of hip joint replacement. A
safety limit of 3-5 million cycles has been settled upon the
basis of clinical experience with loosened femoral components
/1/ and should contribute to preventing the removal of distal
stem fragments involving aggravated conditions to reoperation.
In 1981 the introduction of a load limited to 2 800 N at a
pre-load of 300 N was proposed /2/. This was reported as an
intern manufacturer specification in 1983 and according to /1/
has lead to largely fracture-safe anchorage stems of hip endo-
prostheses. But for standardization no minimum requirements
have been specified so far.
Experimental
In our test procedure the stem position was obtained by a
standard femur /3/. The lower portion of the stem is embedded
up to a distance of 50 mm from the plateau (collar) of the pro-
stheses or from the intersection of the underside of the neck
and the medial side of the stem for prostheses not having a
plateau (Fig. 1). The stem is moulded into a resin for the
pulsating loading. The protruding
part of the test specimen is par-
tially immersed in a quasi physio-
logical fluid (RINGER solution).
A sinusoidal dynamic load with a
frequency of 10 Hz is applied to
the head until the specimen ex-
hibits failure or until a chosen
number of cycles has been attained
at a certain loading range. The
test is interrupted 8 hours per ~'.
solUtion
day thus increasing the possibil-
ity of chemical attack. Further
procedures follow the so-called
staircase technique which is per-
formed until a fracture occurs. Epoxy resin

Metal
Fig. 1: Test design (scheme) for adapter

dynamical loading of THR stems

366

Theoretical Remark on Torsional Effects

During the process of embedding it
may happen that the embedding angles~
and ~ (Fig. 2a) vary slightly. For me- F
tal stems the fracture is usually be-
low the embedd i ng edge and is defi n i te -
ly influenced by the stress conditions
\ I
of the fracture cross section.
A theoretic mechanical approach 1) to the
--t"
torsional moment MTin the embedding ___ .1---;-_
level (see Figs. 2 a and b) for a model
dependent on the angles ~, ~ and r is

1"
t~~~"",~,E MBEDDI NG
EDGE

Fig. 2: a) Scheme of the angle positions; b) Free cut system

1) According to Fig.2a a stereostatic approach to the cut reac-

tion MT , i.e. the torsional moment in the stem axis leads to:

MT = Xl Y2 - X2 Y1 . F
11
where
x1 := 11 sin(~),. Y1:= It cos(~)sin(~) ,
x 2 := 11s1n(~) +12s1n(~-r), Y2:= (l1cos(~)+12cos(~-r))sin(n.
Since MTis in linear dependence on F, the model can be loaded
with a 1 N standard load. Fig. 3 and Fig. 4 present the de-
pendence of the torsional moment MT upon the angles ~,~ and r
at a load of 1 N. For any load F=c,1N, MTis obtained by simply
multiplying the corresponding ordinate value by the factor c.
The geometrical parameters 11 and 12 (Fig.2a) are constant.
Therefore we have taken 11 =45 mm and 12 =55 mm as average values
out of the available models. In order to obtain a ratio between
the bending moment MB and the torsional moment MT ,the relation

should be build up (Fig.2b), where x1 ' x2 , Y1 and Y2 are de-

fined as above. This relationship is independent of the load F.
 367

Fig. 3 and 4:
Torsional moment vs.
angles of embedding position.

shown in Figs. 3 to 5. The

clinically relevant field is
marked in the three-
dimensional representations.
The torsional moment is not
dependent on the embedding 35

angle ex (Fig. 3). Through

the embedding angle 30
corresponding to the ante-
version (i.e. the inclination
of the femoral neck relative
to the femoral shaft in the
transverse plane) apart from
the bending moment MB a
torsional stress in [he
direction of the stem axis
is generated.
The effect on the torsional
moment MT ' caused by the de-
sign-dependent angle r ,
grows with an increasing
embedding angle ~ . For example
an embedding angle ~=5, a
design-dependent angle r=400
and a maximum loading of
about 3 000 N would result
in a torsional moment M of
25 Nm. In an extremely un-
favourable case (~=10,
r= 50) a loading of about
3 000 N results in a tor-
sional moment up to the range Fig. 5:
of 60 Nm (Fig. 4). In the Bending/torsion vs.
graphic presentation (Fig. 5) "anteversion" angle~
the relationship between ben- and "design" angler
368

ding and torsional moment is demonstrated. The bending moment

rises to fifty times the torsional moment for very small angles
~ if the design angle y is in the range of 40 - 50. For lar-
ger and clinically more common angles the torsional moment
rises at most to 20 % of the bending moment. This is quite re-
levant, especially for composite materials sensitive to tor-
sional stresses. In anisotropic materials whith interfaces be-
tween fibres and the matrix the torsfonal moment has an unfa-
vourable effect upon the fatigue strength of these connections.
The same holds for interfaces of stems and coatings.
Static loading tests
New designed hip endoprotheses, such as the "iso-elastic"
models have been subjected to static tests with loading condi-
tions analogous to the dynamic test concept.
Fig. 6 shows the stresses of the stem depending on the dis-
tance from the embedding level. At a vertical static 3 000 N
loading of a titanium alloy stem the strains rise continuously
up to about 3 000 ~m/m from the collar to the embedding edge.
In case of a plastic compound prosthesis with a form-closed
steel core (POM/55), the strains rise much higher to almost
19 000 ~m/m. Down from the embedding surface, the strains
decrease sharply and even negative strains occur.
The deformation behavior may be described as a stem which
is rotating at the embedding edge just like in a hinge. The
high strains are attributed to the extreme deformation of the
stem. The measurement of the horizontal movement of the pro-
sthetic ball by means of an optical device (Fig. 7) confirms
this assumption. In comparison to the straight stem prosthesis
type MUller, the polyacetal/steel composite prosthesis is de-

STRAINS AT LAT.STEM,
3000N LOAD

POMiSS ./ *
./

Ti AL. v.
....-*

---~
. --- ---
('
"'-
.- . *--
-. '-""

STRAINS
1
[pM/M --.L.2o---'----,~0---8~--~--..
0 -"'10---'_.
10 3

Fig 6: Lateral strains vs. distance to the embedding edge.

 369

formed almost 20-times more at 2 000 N of vertical loading

than are common implants. This deformation behavior avoi?ed a
dynamical testing of this composite stem in the test devIce.
3 aaa LOAD [N]

2 aaa

1 aaa

a
a.a 2.5 s.a 7.5 la.a
HORIZ.DISPLACEMENTh[mm]

Fig. 7: Displacement in "medial" direction, cross sections.

Dynamical failure of stemmed femoral hip jOint components
The different procedures and results of 24 hip implants are
presented in Fig. 8, depending on the applied pulsating load
and the number of cycles. Because most of the implants came up
to the present state of material and design, no specifications
of each sample are reported here.
Some stem geometries are not well adaptable to the standard
femur Therefore, a slight variation of the embedding angles ~
and ~ is unavoidable. This results in bending and torsional
moments and hence in a completely different beginning of
cyclic fatigue failure of so-called Young's modulus adapted
carbon reinforced carbon stems. There occurs an irreversible
deformation at the epoxy resin at the lateral edge of the
embedment, contrary to the performance of conventional metal
stems where resin deformation is always generated at the
medial embedding edge.
Apart from varying neck length, the high deviations of the
laboratory results are caused by the interaction of the mechan-
ical and corrosive damage. Some composite materials failed for
370

example after more than 4 million cycles indicating a chemical

sensitivity. Cast alloy prototypes for example failed at 3 mil-
lion cycles, whereas such effects never occured with wrought
Co-Cr alloys. The representation demonstrates the problem of
using a minimum limit of cycles as a criterium for releasing a
stemmed femoral component for clinical application regardless
both the influence of frequency on time-dependent fatigue
effects and design related varying neck lengths (lever arms).

finax /"'\ , ,
~,

, \ I ,
: \
I

Fdyn rN] .... \

\ I
I
,I ___
~
fractured stems
non fractured stems
\ I 2
I
5100 lim
\
-' -IrT
4200

.1 1
~
~

2550
,!2100
- 2
2

1700 1
!lllOO

2 4 5 6

Fig. 8: Dynamical loading regardless the stem designs

(1: cast alloys; 2: composites; without number:
wrought alloys)
The fractographic investigations of dynamically tested
specimens fracture surfaces demonstrate the origins of fail-
ure for different metals (Co-Cr and Ti alloys at the lateral
side in the proximal third of the stems) and for composite
stem materials (at the lateral side). The detected character-
istics of fatigue fractures and of corrosion related failures
seem comparable to topographies of metal stems fractured in-
vivo. For metals the test electrolyte lubricant seems to simu-
late the real chemical interactions adequately. Corrosion
fatigue of a wrought chromium stainless steel (AISI-316L ISO
5832-1) stem is caused in-vivo by the effect of alternating
loads in the presence of a corrosion process in an aqueous or
 371

fig. 9:
in-vivo fractured
stainless steel stem
right top:
fracture surface schew-
left top:
fatigue stri.ations
right bottom:
plane region of
environmental attack
arround pre- or
intraoperative
impact defect
left bottom:
embrittlement causes
feather-like trans-
crystalline fracture
topography

Fig. 10:
in-vivo fractured
cast Co-Cr-Mo stem
right top:
fatigue striations
right bottom (is an
enlargement of left
bottom) :
shrinkage cavity at
the crack origin
(surface structure
limits the efficiency
of nondestructive
defect detection)

synovial environment. This environmental embrittlement first

generates a typical feather-like brittle transcrystalline
cracking (Fig. 9). Because of the time-dependence of the
chemical attack, the number of cycles to failure is dependent
on frequency. Especially tensile stresses of even low values
lead to the propagation of a fracture. Opposed to these
fractures caused by the combination of chemical attack and
mechanical loading, a cast cobalt-chromium prosthesis exhibits
the characteristics of a pure fatigue fracture with striations
in direction of the crack propagation (Fig. 10). These cracks
are caused by alternating mechanical stresses, i.e. successive
opening and closing of the crack flanks. The crack of Fig. 10
starts from a cavity formed by shrinkage during the solidifi-
cation of the cast cobalt chromium alloy.
372

Clinical relevance of the current state of endurance testing

Hip prostheses must be able to support a minimum of 1-2
million cycles per year and additional peak loads which will
stochastically occur. For young patients the prothesis should
last several decades. In case of a prosthesis which is re-
operated one year after it became loose, further 1-2 million
loading cycles may be applied to the pain-afflicted patient.
The test procedure is therefore justified by real conditions.
Broken stems cause intraoperative complications for the re-
moval of the distal fragments. A stem which endures 5 million
cycles of fatigue loading at a minimum pulsating loading 1s
able to exclude most in-vivo failures, but cannot guarantee
clinical success.
Both German standard (DIN 58 840) and international draft
proposal test criteria are based on the assumption that a
breakage of the proximal cement or bone resorption has 0ccured.
They do not take account of design concepts not having a plane
of symmetry, especially "custom-made" multi-curvatured stems.
A specified loading level used for releasing a prototype
stem for clinical application should be designated in relation
to the anatomic geometry and the patient's weight.
The deformation behavior is evidently dependent on the di-
rection of the transmitted loads and thus on the position of
the stem in the intramedullary chanal of the femur. Hence, for
materials which are sensitive to torsion, an anteversion
loading component should be applied. In case of a so-called
"iso-elastic" design, apart from frequency-related temperature
artefacts, extreme deformations avoid an application of the
test concept. The need of design-related test criteria seems
obvious in cases of new biomaterials, which are used for
endoprostheses with unusual anchorage concepts.
Outlook
No minimum requirements are acceptable which do not take
into account varying geometries (stem size related varying le-
ver arms, stressed cross section areas) and materials (time
dependence of specific corrosion mechanisms). What is required
for the future, is:
1 A classification of the broad variety of stem designs being
in clinical use and being newly developed.
2. The evaluation of the test results of a statistically
relevant number of implants should clear up the influence of
two parameters a) dynamic load levels and b) number of loading
cycles.
REFERENCES
1 Semlitsch, M.: Mechanical Properties of Selected Implant
Metals Used for Artificial Hip Joints. CRC Press Boca Raton,
Vol. II (1984), 1-21
2 PIitz W: PrOfung mechanischer Eigenschaften orthopadischer
Implantate, insbesondere von Gelenkprothesen. BMFT For-
schungsbericht T 81-016 (1981), 13-17, 41-42
3. UngethOm, M.: Technologische und biomechani5che Aspekte der
HOft- und Kniealloarthroplastic; Statisch-biomechanische
Spannungsanalysen und dynamische PrOfungen von HOftprothesen.
Verlag Hans Huber, Bern, Stuttgart, Wien (1978), 90-110
 373

STRUCTURAL SAFETY AND STRESS DISTRIBUTIONS IN CERAMIC HIP-

JOINT-HEADS FOR DIFFERENT NECK LENGTHS
G. Kleer. U. Soltesz. U. Benzing and D. Siegele
Fraunhofer-Institut fUr Werkstoffmechanik. Freiburg. Germany

1. INTRODUCTION
In hip-joint replacement surgery surgeons can choose between various
types of ceramic heads differing in the depth of the conical bore and the
thickness of the upper shell. The present contribution considers the stres-
ses effective in these heads under equal loading conditions. Two conditions
are of special interest: Point-contact between cup and head corresponding
to a worst case condition and a regular loading condition using a well fit-
ted cup. For the evaluation of the stresses acting in the heads both expe-
rimental measurements and Finite-Element-calculations were performed. The
knowledge of the stresses is important for obtaining informations on the
probability of failure and furthermore for defining the conditions of proof
testing the heads. Such a proof test is performed after manufacturing the
heads in order to make the failure risk negligible.
2. HEAD TYPES AND LOAD CONDITIONS
Figure 1 shows the three investigated head types with different lengths
of the conical bore and shell thicknesses d, = 4.5 mm. d2 = 6.3 mm.
d3 = 9.8 mm. The total diameter. 32 mm, as well as the external shape is
equal in all three heads. By use of the adequate type of these heads the
effective length of the prostesis neck (L in Fig. 1) and the total length
of the prothesis can be adapted to the physiological requirements.

type (1) type (2) type (3)

short medium long

Fig. 1: Ceramic hip-joint-heads with different depth of the borehole and

thickness d of the shell
374

3. STRESS DISTRIBUTIONS
In order to evaluate the stresses in these heads calculations using the
Finite-Element-Method, and experimental measurements were performed.
3.1 Experimental determination
Loading experiments were carried out under laboratory conditions using
a servohydraulic testing equipment. Small strain gauges were attached at
different positions of the heads. For performing the measurements the heads
were pressed on stems. The point contact was simulated by using plates of
steel and for the simulation of regular contact conditions original cups
made of alumina were used.
In the experiments it was found that the maximum tensile stresses
occured on the inner surface at the center of the curvature. The stress
measured at this point is plotted as function of the force for each head
in Figure 2. The highest tensile stresses were observed in head type (1).
As shown by the figure the stress increases linearly if there is only
a small contact area between cup and head; the nonlinear increase of the
stress which is observed if a cup is used, should be caused by an increase
of the contact area. According to the figure a point contact between head

300 300
;f
~

l! 200 l! 200
~ @
QI
:E
13 13
o 100 o 100
~ ~
~ ~
VI
VI

OL-----L-----~----~----~
o ~ 5 B m 2.5 5 7.5 10
Force F [kN) Force F [kN)

(' iI' Hip-joint head, type 3

.21 200 Fig. 2: Tensile stresses
@ measured at the center of
h the hip-joint-head for
different head types and
13
loading conditions.
o 100

~
VI

owe~--L-----~----~----~
o ~ 5 B m
Force F [kN)
 375

and cup leads to an immediate increase of the stress if the force is in-
creased, so that at 5 kN a stress level of 300 MPa is reached; a further
increase of the force would rapidly lead to the fracture strength limit
which is about 400 to 500 MPa for alumina. If a ceramic cup is used the
stresses are reduced to about one half of these values; considering here a
load of 12 times the body weight which corresponds to the peak loads during
jumping, also an increased tensile stress of about 200 MPa is observed. In
the head types (2) and (3) with higher thicknesses of the shell the stres-
ses are reduced to less than one half of the values observed in type (1).

Fixed bond, 5 kN Fixed bond, 5kN

160 IF 160 IF

'~ 'f ].
a,

80 80
: type (1) a, ;f
~
type (2)
~ ""''' medium

~ +

:a
.~
.,
<5
c.
.
.~
a.
-80 -80

-160 -160
A A B

16 24 32 14 21 28
path [mmJ path [mmJ

Fixed bond, 5kN

160 IF

;f
~
80
'f C1,
point -contact

8. Fig. 3: Calculated stress

:[ distribution at the inside
- 80
(A -- C) of the ceramic
head for different head
types and load conditions
-160
A B

12 18
path [mmJ

3.2 Stress calculations

To determine the stresses of the whole structure Finite-Element calcula-
tions were performed. Also in the calculations it was found that the high-
est stresses generated by the deformation of the spherical shell occur in
376

the center of the curvature. According to our recent work (ref. 1,2) the
sliding on the stem leads only to 10 % higher maximum stresses compared
with the case of no sliding. For this reason in the Finite-Element-calcu-
lation only this case is considered that the head is fixed bonded with the
stem. As load conditions also a point-contact and corresponding to the
loading by the ceramic cup a cosine-shaped force distribution acting on a
contact area of 11 mm diameter were assumed. In Fig. 3 the principal stres-
ses 01,02 in the plane and Ox in the circumferential direction are plotted
along the path A-C for the load of 5 kN. In accordance with the experi-
ment the maximum tensile stresses are reached at the center of the head and
they increase with the thickness of the shell. For the head-type (2), for
instance, and an applied load of 5 kN the following maximum tensile stres-
ses were calculated: 119 MPa in the case of point-contact and 75 MPa for
the contact area of 11 mm diameter. In the experiments maximum stresses of
114 MPa and 63 MPa, respectively, were measured. The differences between
calculation and experiment occur because the increase of the contact area
with the load level and the sliding of the head on the stem is not taken
into account.
4. STRUCTURAL SAFETY BY PROOF-TESTING
In ceramic materials tensile stresses lead to a growing of preexisting
cracks. For reducing the propability of failure heads with not acceptable
large cracks must be selected out and disguarded.
4.1 Basic principle
The concept for this selection is based on fracture mechanics as fol-
lows: If a proof stress op which is higher than the service stress, is ap-
plied to specimens made of a ceramic material with a critical stress in-
tensity factor Kl c all specimens where the length of the most dangerous
crack af.p fulfills the relation

K2Ic
a >-- (1)
f,p f02
P

will fail in the test. For the remalnlng specimens mlnlmum life times,
tmin. k , can be cal cul ated for the case of an effecti ve constant stress
ok according to the equation

2
(2)

where n and A are the crack propagation parameters and f is a form factor
( ref. 3, 4, 5).
The stress distribution can now be used for calculating the peak and re-
sidual stresses for different movement conditions. The proof stresses for
excluding failure of the heads during a requested life time are calculated
according to equation (1). Results for different loading cases are shown in
Fig. 4 for head type (2).
The peak stresses omax and the residual stresses Omin listed in the fig-
ure for two body weights (750 N, 900 N) for normal walking, fast walking
 377

and jumping were calculated by using the multiplication factors (4.5, 7.3,
12) for the body weights as defined in the references (6) and (7). Accor-
ding to the figure the peak stresses are 60 MPa for normal walking and
normal body weight. Compared to this, the higher body weight results in 71
MPa. For fast walking and jumping peak stresses of 115 MPa and 190 MPa re-
spectively were assumed as conservative estimates. After extreme body move-
ments the residual stresses reach their maximum values which are frozen in
for the future, so that e.g. after jumping the residual stresses also for
normal walking increase to 65 MPa which is nearly three times the original
normal walking value.
The minimum life times as function of the proof-stresses (see equ. 2)
were calculated for an average material with a median strength 0S,m = 450
MPa, a Weibull modulus m = 10, crack propagation parameters n = 71 and log
A = -148 (with KI in Nmm- 3/2 ), assuming a semi-elliptical flaw as initial
damage source and sine shaped loading functions closely approximating the
real case. According to the figure the proof-stress has to be higher the
higher the desired life time is. Whereas, in the case of a normal body
weight for normal walking a proof-stress of about 80 MPa is sufficient for
ensuring a life time of 20 years, for the higher body weight a proof-stress
of about 100 MPa is required. The proof-stress has to be increased if fail-
ure for more realistic movement conditions has to be excluded. For the mo-
del cases of fast walking and jumping the proof-stress must exceed 170 and
260 MPa, respectively, if a life time of 20 years is requested. As a conse-
quence of the higher service stresses in head type (1) the proof stresses
have to be higher for this head whereas lower proof stresses are sufficient
for head type (3).

Aluminia hip - joint -head

10"
C1(t) = C1ml1+CsinWll

1010

tmin 10'
Is]
10'
1 day - - - -I-
I
10'
I
I
10' I

0'_.
body weight IN) 750 900
(MF'UI 60 71
900
115
900
190
0'"", lMPol 20 24 39 65
nonnol fast
JUmps
walk~ walJOrg

60 80 100 120 150 200 250

proof-stress a [MPo]

Fig. 4: Minimum life time tmin as a function of the proof-stress for ave-
rage bio-alumina for a special loading case (axial symmetric loading

on a contact area of 11 mm diameter assuming a sine shaped loading
function)

5. CONCLUSION
The stress distributions analysed for the three different head types
show that the highest tensile stresses mainly occur in the center of the
378

curvature at the inner surface. At equal loading conditions the amount of

stress and consequently the risk of failure is different, for the three
head types. In the head type with the thinnest shell (type 1) the stresses
may be 4 to 8 times higher compared to type 3. The stress level is also in-
creased especially if bad fitted cups or other irregular conditions lead to
small contact areas between cup and head. A proof test which can be applied
to all head types makes the risk of mechanical failure of implanted heads
very small.
Acknowledgement: This work was supported by the Bundesministerium fUr
Forschung una Technologie under contract number ZQ 017 120/3
REFERENCES
1. G. Kleer, D. Siegele, U. Soltesz, in: Developments in Biomechanics,
Vol. 2. Martinus Nijhoff, The Hague, Proceedings of ESB4, Davos,
September 24-26, 1984
2. G. Kleer, U. Soltesz, D. Siegele, in: Biological and Biomechanical Per-
formance of Biomaterials, ed. by P. Christel, A. Meunier, A.J.C. Lee.
Elsevier Science Publ. B.V., Amsterdam 1986, pp. 489-494
3. S.M.Wiederhorn, Proc. of the 2nd Army Materials Technol. Conf.,
Hyannis, Mass., November 13-16, 1973
4. A.G. Evans, S.M. Wiederhorn, Intern. J. of Fracture 10 (1974),
pp. 379-392
5. A.G. Evans, Fuller, E.R., Metallurgical Transactions 5 (1974), pp. 27-33
6. J.P. Paul, Loading on Normal Hip and Knee Joints and on Joint Replace-
ments, in: Advances in Artificial Hip and Knee joint Technology Eds.:
M. Schaldach und D. Hohmann, Springer-Verlag, Berlin, Heidelberg,
New York (1976)
7. H. Richter, U. Seidelmann, U. Soltesz, 1980, BMFT Forschungsbericht
T 82-003
 379

A MATHEMATICAL MODEL OF HUMAN KNEE-MOTION AND EVALUATION OF

KNEE ENDOPROSTHESES
J. Hoschek, J. Halt, G. Selvik and U. Weber

Technische Hochschule, Darmstadt, Germany; University of Lund,

Lund, Sweden; Orthop. Universitatsklinik, GieBen, Germany

Kinematic studies of the knee have been carried out by a number of

authors, each resolving a portion of the knee complex. These researches
can be divided into two groups
- on the one side contributions which treated the knee motion as a planar
motion,
- on the other side contributions which interpret the knee motion as a
spatial motion.
The results on the basis of a planar model can be structured into two
branches again
- results, which approximate the knee motions as a rotation on a fixed
axis.
- results, which approximate the knee motion by the rolling of two polodes
or femur condyles.
These contributions have been used to develop an artifical knee. Although
we can find a lot of research papers about the planar knee motion there
are considerably less contributions concerning three dimensional (spatial)
knee motion.
To o.Jtain accurate data on all aspects of spatial knee motion in vivo
measurements were performed using the roentgen-stereophotogrammetric system
of Selvik (1981) with volunteers at the Universities of Lund, Uppsala and
Umae. By the help of these data we will construct a new mathematical model
of the human knee motion.

1. RADIOGRAPHIC TESTS OF KNEE MOTIONS

The motion between rigid bodies (skeletal parts) can be completely
determined if the position of at least three markers, not on a line, in
each body are assessed. Such markers can be attached by metal pins to the
bones and then external measurements can be made - but a more feasible
procedure in many instances is to mark the skeleton with intraosseous
implants and determine their positions by measurements on radiographs.
Such a method has been developed by Selvik (1974) and is used under several
aspects with great success during the last years (see Dijk (1979), Selvik
(1981), de Lange (1982), Wyss (1982)).
To obtain distinct measurement points in the knee the distal femur, the
proximal tibia and the patella were marked each with (at least) four
tantalum balls, diameter 0,8 mID (see fig. 1), under sterile conditions and
using specially constructed instruments. For measurement two film exchan-
gers were used with 35 x 35 cm 2 films, the frame rate was between 2/s and
4/s. To produce radiographic images of the knee which permit a photogram-
metric reconstruction the radiographic equipment has to be calibrated. For
this calibration a "cage" was used. The cage was radiographed together
with a reference plate containing tantalum balls fixed on the grid posi-
tioned in the front of the film in each film exchanger. From this exposure,
380

the positions of the tube foci in the ccordinate system defined by the cage,
as well as L~e projections of the reference markers were determined. In the
subsequent experimental series the cage was removed leaving a free space
for the moving of 'the leg (see fig. 2).

Fig. 2: Investigation of a volunteer.

The two roentgen tubes are seen in
their housings, as well as the two
film exchangers.

Fig. 1: Tantalum balls implanted in

femur, tibia and patella.

For measurements of the film a Wild Autograph AS drawing table was used.
The measured coordinates of reference and of the knee markers in the test
were used to calculate the knee markers three dimensional coordinates in
the cage system. The mathematical methods used are described in Selvik
(1974). In total 15 tests with 6 volunteers were performed until now.
The radiographic tests yield the three-dimensional coordinates of the mar-
kers from the tibia and from the femur. If the femur, tibia and patella are
concerned to be rigid, and the markers are rigidly fixed, then a rigid body
check is performed to assess the quality of the three-dimensional recon-
struction of the markers. Only data which describe a rigid motion are use-
able for knee joint kinematic analysis. For a check of the rigid body model
we had used the following geometric criteria
1. The volume of the polyhedron with the markers as vertices,
2. the area of the triangles which form the surface of the polyhedron,
3. the angles of the triangles,
4. the distances between the markers.
The accuracy obtained was sufficient for the assumption of a rigid body
model: We received the following relative standard deviations: for volumes
about 1,6%, for the areas of triangles between the markers 1%, for the
angles between the markers 0,8% and for the distances between the markers
below 0,5% (Hoschek 1985).
 381

2. MATHEMATICAL MODEL OF KNEE MOTION

The kinematic analysis describes the motion (euclidean displacement) of
a moved rigid body ~ with respect to a fixed rigid body ~. Therefore we
must transform the measured data obtained from femur and from tibia into
one system, for example into a reference system fixed with the femur. For
this transformation we pick out in each position three (arbitrary choosen)
markers Pl, P2, P3 and evaluate their center of gravit;x: S. S might be the
origin of a Cartesian frame of reference of the femur ~. The basis vectors
e, of this frame can be received from the position vectors of the markers.
If X denotes an arbitrary measurement point in the system of reference
(el,e2, e3) (the cage system), the transformed point X
has the representa-
tion
X=S+AX, (1)

where the orthogonal matrix A describes the rotation from the system {~i}
to {ei}' By inversion of (1) all measurement ~oints obtained from the
tibia can be transformed in the fixed system ~ of the femur.
The arbitrarily choosen markers P l , P 2 , P 3 for evaluation of the femur
reference system contain randomly inaccuracies, this error is transferred
on the transformed points from the tibia. To minimize the influence of
these inaccuracies optimization techniques were used.
After the transformation (1) and optimization we have received for each
position k the three-dimensional coordinates of the markers APi with respect
to the fixed femur-system. We will denote these vectors by Xik, where i=l,
... ,4 (number of the markers) and k=O" .. ,n with n+l as number of the eva-
luated positions.
The motion (euclidean dispkacement) of a rigid body r
(moving space) with
respect to a fixed system ~ can be described by the matrix equation
A A

X(t) = A(t)X + b(t) (2)

in which X and X are the position vectors of a point P in the fixed space I
and the moving space ~ respectively. A is an orthogonal matrix, b a trans-
lation vector, t time parameter. If A and b
are functions of a parameter t
(time), (2) gives us a continuous series of displacements, which we will
call a motion. Any point X of the moving space ~ describes a curve, its
path, in the fixed space ~. A
We choose for example the femur as fixed space ~ and the tibia as moving
space r. Further we assume a smooth mot~on passway and will evaluate the
matrix A(t) and the translation vector b(t) out of the path of the measure-
ment points Xik . The index i (i=1, ... ,4) describes the markers in each
position k (k=O,l, ... ,n) with n+l as number of positions in a test. If we
denote by tk the measurement time for position k, we get from (2) directly
(3)
with ok as error vector. Now we eliminate the (unknown) vectors Xi from two
successive equations in (3) and receive
Xik = Rk Xi,k-l + Vk + 8k (4)
with
Rk : = A(tk ) A-l(tk_l)' Vk : = b(tk) - Rk b(tk _ 1 ) (4a)

6
and k as transformed error vector. An overall measure for these error vec-
tors is given by the function f of V and R defined by
A 1 4 A A A T A A A
fk(v,R) 2 ~ (RkXi,k-l + Vk - Xik) (Rk Xi ,k-l+Vk- Xik)
(5)
i=l
Spoor-Veldpaus (1980) have developed a minimizing procedure for the func-
tion f which provides a representation of R by the help of certain eigen-
382

vectors and ei~envalues while the vectors b(tk ) are the centers of gravity
of the points Xik for position k.
If we f~rther suppose that at time t = to the cartesian frames of the fixed
system E and the moved system E are parallel, we have A(to ) = E and obtain
from (4a) subsequently

We assume that the elements of the matrix A are functions of the rotation
angles ~, ~, e, which represent rotations on the coordinate axes (Selvik
1974) and get as an equivalent representation of the matrix A at time tk
the rotation vector (with rotation angles as components)

After these xalculations we have in each position k (time tk) the transla-
tion vector b(tk) and the rotation vector Dk(tk) (k=O,l, .. ,n). To get an
analytic representation of a continuous motion finally we have to approxi-
mate these two sets of points by continuous functions.
For this purpose we can use spline-functions, which are well known in COM-
PUTER AIDED DESIGN (de Boor 1978, B6hm et al. 1984) as Bezier-curves or
B-Spline-curves.
The B-Spline-curves have the parametric representation
1
X(t) = E Qi NiS(t) (6)
i=O
with NiS(t) as B-Spline-functions of degree 5 and Qi as control points
(de Boor pOints). The parameter t runs through the interval t E [0, 1-4J
with the boundary values 5 ~ 1 ~ n (n+1 number of positions). 1 = 5 trans-
forms (6) into the parametric representation of Bezier-curves.
For approximation ot the curves determined by b(tk) and Dk(tk) these points
receive at first parametric values proportional to the distances between
two neighbouring points. The control points Qj will be evaluated by the help
of least square methods combined with optimization of the parameter values.
The design parameter 1 is determined by the condition
JIKids --- min ,
with K as curvature and s as arc length of the approximating curve. This
assumption might by an effective description of the natural knee motion.
At the end of the approximating process we have a parametric representation
of the translation vector b(t) and the rotation vector D(t). D(t) leads to
the displacement matrix A(t), thus we have received a parametric represen-
tation of the knee motion of a volunteer
A A

X(t) = b(t) + A(t) X .

This equation of the knee motion is the basis for the following discussions.

3. A NEW DESIGN OF KNEE ENDOPROSTHESES

By the help of our radiographic measurements we have received a mathema-
tical model of the knee motion. This mathematical model leads to a new
mathematical method for evaluation of the mating surfaces of a knee endo-
prosthesis. Such an endoprosthesis is kinematically accurate in her work.
The mating surfaces of such a prosthesis necessarily contact along a space
curve.
For evaluation of mating surfaces we can choose one surface arbitrarily. If
we assume that we have choosen the profile X(u,v) (u,v parameter) out of the
 383

moving space ~, the shape of the corresponding surface X(u,v) in the fixed
space f is the envelope of X during the motion ~ with respect to ~. The
choosen surface X(u,v) can be attached to the femur or the tibia. For the
parametric representation of X(u,v) we used the Bezier surface method which
is also wellknown in COMPUTER AIDED DESIGN.
While the mating surfaces make contact along a curve K the two surfaces
have a co~on normal N all along this curve K (see fig. 3). The relative
velocity X of the points of the profile X along the curve of contact K must
lie in a plane normal to the common normal. The normal N of a surface in
parametric representation X(u,v) is determined about
N(u,v) = Xu x Xv ,
where Xu' Xv denotes the partial derivatives with respect to u or v. So we
have as condition of contact
:..
N . X = (Xu' XV'
.
X) = 0 . (7)
(7) provides a transcendental equation, whose zeroes with respect to the
motion parameter t determine the curve of contact on the profile X(u,v) at
time t. If we transform all these curves K(t) for all parameter values t by
use of (2) into the system f, we receive the mating surface X(u,v) of the
fixed system. Fig. 4 contains an example of such a pair of mating surfaces.
The tibia profile is given, the femur profile is evaluated, the lines of
contact are drawn in the femur profile. These mating surfaces are calcula-
ted from a test with a knee flexion from 20 0 to 95 0

Fig. 3: Condition of conta~t

for mating surfaces X and X
and their line of contact K.

Fig. 4: Mating surfaces.

The tibia profile is
choosen, the femur profile
evaluated. The tibia pro-
file has deepenings at
each side, the middle part.
is elevated. The femur pro-
file is distorted by pro-
jection
.....
384

Further we have tried to receive a standardization of the knee motion. For

this purpose we have evaluated invariant functions of the motion like the
pitch and the distribution parameter or invariant functions of the corre-
sponding polodes like striction, natural curvature or natural torsion. For
the pitch (s. Bottema - Roth 1979), (s. fig. 5) a certain uniform tendency
could be observed, but the other invariant functions don't have such a
course. The reason for this effect might be, that the evaluation of the
pitch only uses the first derivatives, while the other invariant functions
use higher (up to fourth) derivatives. To receive a standardization of the
knee motion we certainly require
- more measurements,
- more stiff spline approximation like splines in tension.

PITCH
20 r- ------------- ~
Fig. 5: The course of the
pitch for several tests with

::1 II
respect to the flexion angles.
The zero-line describes the
51
..
0r-~~E;~~~~~~~
- 5i
I

i
plane motion.

-lOj i
-15 I !
o 20 40 60 80 100 120
angle (degre.)

REFERENCES
1. B6hm,W.,Farin,G.,Kahmann,J.(1984): A Survey of curve and surface methods
in CAGD. Computer Aided Geometric Design 1, 1-60.
2. Bottema,O.,Roth,B. (1979): Theoretical kinematics. North Holland.
3. De Boor,C.(1978): A practical Guide to Splines. Springer.
4. Dijk,R.v.,Huiskes,R.,Selvik,G. (1979): Roentgen stereophotogrammetric me-
thods for the evaluation of the three dimensional kinematic behaviour
and cruciate ligament length patterns of the human knee joint. J. Bio-
mechanics 12, 727-731.
5. Hoschek,J.,Schelske,H.-J. (1985): MeBverfahren fur mathematische Modell-
bildung zur Entwicklung und Prufung von Kniegelenkendoprothesen. Bundes-
ministerium fur Forschung und Technologie Forschungsbericht T85-042.
6. Lange,deA.,Dijk,R.v.,Huiskes,R.,Selvik,G.,Rens v.,Th.J.G. (1982): The
Application of roentgenstereophotogrammetry for Evaluation of Knee-Joint
Kinematics in Vitro. Biomechanics: Principles and Applications. Nijhoff
Publishers, The Hague, 177-184.
7. Selvik,G. (1974): A roentgenstereophotogrammetric method for the study of
the kinematics of the skeletal system. Diss. Lund.
8. Selvik,G. (1981): Roentgen Stereophotogrammetry in High Tibial Osteotomy
for Gonarthrosis. Arch. Orthop. Traumat. Surg. 99, 73-81.
9. Spoor,C.W.,Veldpaus,F.E.(1980): Rigid Body Motion calculated from Spa-
tial Co-ordinates of Markers. J. Biomechanics 13, 391-393.
10. Wyss,U.P.u.a.(1982): Dimensions of the Femoral Condyles. Biomechanics:
Principles and Applications. Nijhoff Publishers, The Hague, 291-297.
ORTHOPAEDIC BIOMECHANICS: IMPLANTS AND FRACTURE FIXATION
 387

PARAMETRIC ANALYSIS OF THE STRESS PROTECTION IN BONE AFTER

PLATING
J.R. Cordey*. S.M. Perren* and S. Steinemann**
Lab. Exper. Surg.*. Davos. Switzerland; Department de Physique
Experimentale**. Universite de Lausanne. Lausanne. Switzerland

1. SUMMARY

Bone refracture after plate removal has been attributed to the struc-
tural adaptation of bone (bone loss) to reduced stress (stress protection).
The analysis of the stress pattern and the reduction of deformation work in
plated bones seems to be a prerequisite for the assertion that bone loss is
stress related. The strain field within bones under eccentric axial load is
analyzed using the composite beam theory and verified experimentally using
strain gages, with plates made of different materials. The difference bet-
ween the reduction of stress obtained using the less stiff plate materials
is small. Reduction of plate rigidity does not result in proportional im-
provement of the bone strain.

2. INTRODUCTION

Following internal fixation, bone loss has been observed in animal

bones which were experimentally plated. Such bone loss is commonly attribu-
ted to the structural adaptation of the bone due to the stress protection.
It is commonly accepted that this would result in bone weakening and sub-
sequent danger of refracture. It seemed reasonable therefore to try less
rigid plates (2). Plates made of less rigid material (titanium, carbon fi-
bers, etc.) have been proposed (2). The aim of the present paper is to
present a parametrical analysis of the strain distribution of theoretical
and experimental models of stress protection.
The functional strain field within the bone is related to the bony
adaptation (Wolff's law); the physical value which describes such a
possible adaptation phenomenon is the deformation work W produced within
the bone by the physiological loads (5). Therefore both strain field modi-
fication (which is a local estimator) and deformation work (which is a
global estimator) will be taken into account in the present analysis.

3.1. Composite beam theory. The analysis of plated bone is based upon the
model of a composite beam (1, 2), which allows the calculation of the
stress pattern under axial load and bending. The following assumptions are
made: 1. The fracture is completely adapted or healed. 2. The bone is
purely cortical. 3. The cortical bone behaves as a linear elastic material,
388

homogeneous and isotropic, equal in tension and compression. 4. The distur-

bance of the cross-section by the screws and the screw holes is neglected.
5. The plate is a uniform bar. 6. No movement occures between plate and
bone.
The strain distribution in plated bone (z) is given by the superpo-
sition of the strain due to the axial force component F and the strain due
to pure bending moment M, z is the distance from the centroid of the cross-
section to the point in question, positive for the side in tension and
negative for the side in compression:

= -F
AE
+ M.z
IE
A = cross-sectional area AE = axial stiffness
I = moment of inertia IE = bending stiffness
E = elastic modulus

When the plate is applied to the bone, the neutral axis of the beam in pure
bending is shifted toward the plate. The shift Z of neutral axis is given
by the following equation:

Z = AEp.D AEp = axial stiffness of the plate

AEb + AEp AEb = axial stiffness of the bone
D = distance between the centroids plate - bone

The axial stiffness and bending stiffness of the composite beam are then:

AE = AEb + AEp IEp = bending stiffness of the plate

IE = IEb + AEb Z2 + IEp + AEp (D_Z)2 IEb = bending stiffness of the bone

3.2. Deformation work: The effect of stress protection upon bone remodeling
may be described by the reduction of deformation work W for the plated bone
in relationship to the deformation work Wo for the unplated bone. The
deformation work in cortical bone is given by:

The deformation work is given by the addition of work in axial load WF and
the work in bending WM:

= f dx =
bone

3.3. Application. Physiological load i,s simulated by an eccentric axial

load. The eccentricity is characterized by the ratio dID (d = distance
axial load - centroid of the bone, D = bone diameter).
For the calculation of the strain, the bone is modelled by a
circular tube with inside and outside diameters of 13.3mm and 20mm
respectively, and with an elastic modulus of 20GPa.
 389

The size of the plate is assumed to be similar to that of the AO-ASIF

dynamic compression plate, but with a rectangular cross-section (4x12mm)
and no screw holes. The following elastic moduli are taken into account:
Cobalt alloy 240GPa; stainless steel 190GPa; titanium 110GPa; carbon fiber
65GPa.

3.4. Calculated results. The strain field for unplated (1) and plated (2)
bones under varying eccentricity is reported on Figure 1.a and the corres-
ponding reduction of deformation work in Figure 1.b.
The strain is highly reduced under the plate and less reduced at the
opposite cortex; the difference between the different plates is rather
small. The stress protection is very important when the plate is fixed at
the compression side (inapropriate according to the AO technique) and
minimal when the plate is applied at the boundary of the central core about
the tension side. Under physiological load condition, the strain reduction
ranges between 70 and 90%; the plate material is of minor importance.
The ratio bending stiffness of the composite beam EI I bending
stiffness of the unplated bone EIo (EI/EIo - logarithmic scale) for both
bone diameter (D - logarithmic scale) and plate material (Ep - linear
scale) is reported on Figure 2. EI/EIo is a possible estimator of the
stress protection which shows a biphasic behaviour: It is large for small
bones and small for large ones. The limit between the phases is about 10 mm
diameter. The presence of a plate is important, the effect of the plate
material (elastic modulus Ep) is of minor importance, either for large or
for small bones.

W/Wo
100 (%)
A

--,-
50
(
A
&:::;;;-;;;- ----

0
-1 0
(a) (b)

FIGURE 1. Eccentricity of the axial load and stress protection.

1.a. Strain field for unplated (1) am plated (2) bones: Effect of diffe-
rent eccentricity. The loads A and C correspon:! to the IIBximal an:! minimal
stress protection.
1. b. Correspon:!ing deforDBtion work in percentage to the work for unplated
bone WIWo.
390

4 x 12 mm

10 4 FIGURE 2. Reduction of bending stiffness

EI/EIo in relationship to bone diameter
D am elastic modulus of the plate Ep.

1000

100

10

4. EXPERIMENTAL VERIFICATION

4.1. Material and method. Human cadaveric femora, deep frozen and thawed,
were used. Cross-sections were obtained using computed tomography. Six
strain gages were attached to the midshaft of each unfractured femora. The
strain gages were connected through a bridge amplifier and to a X/Y
plotter.
The dimensions of the plates used were 4.8x16mm for the broad AO-ASIF
dynamic compression plates in stainless steel (SS) and in titanium (Ti),
4.0x16mm for a carbon fiber reinforced polysulfone plate (C) and 5.25x16
for a carbon fiber reinforced epoxy resin plate (CE).
A Rumul Microtron testing machine was used to apply eccentric load.
During testing, the distal part of the femur was resting on a mould of
methylmethacrylate. The load to the proximal part of the femur was applied
through a lever arm which applies a compressive load at the femur head and
a tensile load at the trochanter (Figure 3, (3.

4.2. Experimental results. In unplated bone, the position of zero strain

was near the center of the bone (Figure 4.a). In plated bone, the shift of
this value toward the plate is clearly seen. For both plated and unplated
bones, the measured strain varies linearly with increasing distance from
the zero position. In plated bone, the reduction of the strain is greater
 391

under the plate than at the opposite cortex. The difference between the
effects of the plates made of different materials is small. The measured
results are in good accordance with the calculated ones (Figure 4.b).

Fa B

(
(E
Ti
(ETi
(
S5
SS

(b)

FIGURE 4. Strain field in human femur

under eccentric load.
a. Measured values.
b. Calculated values.
FIGURE 3. Eccentric load applied to B: bone alone, SS: stainless steel
human fellllr. plate, Ti: titanium plate, C: carbon fi-
ber reinforced polysulfone, CE: carbon
fiber reinforced epoxy resin.

5. DISCUSSION AND CONCLUSION

1. The modification of the strain pattern following plate fixation is

reduced in average but not in the same amount within all the section. It
can increase, for example about the center of the bone in bending or at the
cortex opposite to the plate in axial load. 2. The modified strain pattern
is not linearly proportional, neither to the elastic modulus of the bone
nor to the bone diameter. It mainly relates to the axial stiffness of the
plate (7). 3. For large bones and/or plates of adequate dimension, the
amount of stress protection is not strongly affected, neither by using
different plate materials nor by using plates of varying cross-sections.
4. The stress protection (estimated using EI/EIo) shows a biphasic beha-
392

viour in relation to the bone dimension and the elastic modulus (Fig. 2):
Bones of small diameter are strongly affected by the plate modulus, but the
modulus of the plate shows no major influence on bones of large diameter.
5. The limit between small and large amount of stress protection is reached
when the bending stiffness of the plate is equal to that of the bone.
Theoretical prediction of stress protection is in good agreement with
the measured values of strain. The composite beam theory leads to compre-
hensive and adequate results. Limitation occurs however, concerning the
stability of the fixation under high shear stresses between plate and bone
and under torsional load. The correlation between strain field and bone
loss is not established (4). Validity concerning size relations between
implant and bone in small animal experiments comparing to clinical situa-
tions: In animals, comparatively large plates are often applied to small
bones and the amount of stress protection risks being out of proportion.

6. ----------
REFERENCES

1. Carter, D.R. and R. Vasu: Plate and bone stresses for single- and
double-plated femoral fractures. J. Biomechanics 1': 55-62 (1981).
2. Cordey, J., and Perren, S.M.: Stress protection in femora plated by
carbon fiber and metallic plates, mathematical analysis and experimen-
tal verification. In Ducheyne, P.; Van der Perre, G.; and Aubert, A.
(Edts.) Biomaterials and biomechanics 1983. Elsevier, Amsterdam, 1984.
3. Cordey, J., and Perren, S.M.: Charges physiologiques sur les fumurs
humains dans Ie plan frontal: Estimation de la force appliquue par Ie
tractus ilio-tibial. J. Biophys: et biomecan. 9: 386-387, 1985.
4. Gautier, E.; Cordey, J.; Mathys, R.; Rahn, B.A.; and Perren, S.M.: Po-
rosity and remodelling of plated bone after internal fixation: Result
of stress shielding or vascular damage? In Ducheyne, P.; Van der Perre,
G.; and Aubert, A. (Edts.) Biomaterials and biomechanics 1983. Else-
vier, Amsterdam, 1984.
5. Steinemann, S.: Bone remodeling - What are the parameters? In Perren,
S.M., and Schneider, E. (Edts.) Biomechanics: Current interdisciplinary
research. Nijhoff, Dorrecht, 1985.
6. Uhthoff, H.K.: Current concepts of internal fixation of fractures.
Springer-Verlag, Berlin, 1980.
7. Woo, S.L.-Y.; W.H. Akeson; R.B. Simon and R.D. Coutts: The advantage
and important design parameters for les rigid fixation plates. In
Uhthoff, H. (Edt.) Current concepts in internal fixation of fracture.
Springer-Verlag, Berlin, 1980.

Acknowledgements. This work, which is part of a thesis of the University of

Lausanne, was supported by part of the grant number 3.857.083 of the Swiss
National Science Foundation.
 393

LIMITS OF PLATE ON BONE FRICTION IN INTERNAL FIXATION OF

FRACTURES
J.R. Cordey and S.M. Perren
Laboratory for Experimental Surgery, Swiss Research Institute,
Davos, Switzerland

Stability in plate fixation of fractures relates to the friction

between plate and bone. When the plate is applied at the tension side (ac-
cording to the tension band principle), high shear force may appear between
plate and bone, which can lead to motion under weight bearing. This possi-
bility was investigated "in vitro" on human femora, using plates made of
smooth (steel, carbon fiber reinforced epoxy resin) or rough (titanium,
carbon fiber reinforced polysulfone) plates. Motion appeared under smooth
plates under relatively low physiological loads. This can be related to
reduced stability in plate fixation, and consequently to clinical problems.

2. INTRODUCTION

Among the important mechanisms of force transmission in the function

of plates in internal fixation of fractures, is the friction between the
undersurface of the plate and bone. Transmission of forces by friction is
expected to be more effective than that through the bending stiffness of
the screws. Since bone loss is believed to be related to stress relief by
the plate ("stress protection"), it is important to attain a better un-
derstanding of this type of force transmission. The aim of this paper was
to explore the limits of friction in vitro.

3. THEORETICAL ANALYSIS

3.1. Theory.
As long as the shear load between bone and plate is small, the rela-
tionship between the strain field within the bone and the external load ap-
plied by weight bearing is calculated using the composite beam theory (2).
When the shear load between bone and plate overcomes the frictional limit
of motion, the plate is assumed to move freely against the bone. Slippage
between bone and plate does occur when the force which is produced within
the plate by a bending moment applied to the bone is larger than the
frictional force:

EI .]J Fz
(D - Z) AEp
394

where: AEp = axial stiffness of the plate

EI = bending stiffness of the beam.
~ = coefficient of friction plate - bone.
Fz = axial force within the screws.
D = distance between the centroids of the plate and of the
bone.
Z = shift of the centroid of bone toward the plate.

Slippage begins to occur at the last plate screw and, after it does appear,
it propagates itself to the next screws. Therefore the axial force which
must be taken into account is the force of one screw.

3.2. Model of physiological load.

Physiological load is simulated by an eccentric load, similar to the
load supported by human femora at midshaft.
The femur is modelled by a tube with 30mm external diameter, 20mm
internal diameter and 20GPa elastic modulus. The cross-section of the plate
is rectangular (4.8x16 mm); its elastic modulus is 190 GPa (stainless
steel) or 110 GPa (titanium), simulating the conditions encountered in the
experimental investigation. The coefficient of friction between plate and
bone is 0.37 for steel and 0.4 for titanium (5, 6).

3.3. Theoretical results.

The superficial bone strain is reported under the plate and at the
opposite cortex on Figure 1, in relationship to the eccentric load applied
to the femur. Slippage is predicted for a load value slightly larger than
500N either for titanium or for steel plates.

4. EXPERIMENTAL INVESTIGATION

4.1. Materials and Methods.

Human cadaveric femora were used. Six strain gauges were attached to
the midshaft of each intact femur. The plates used were: 8-hole DCP either
stainless steel or pure titanium, carbon fiber reinforced polysulfone plate
with round holes and carbon fiber reinforced epoxy resin round-holes plate.
For the steel and the titanium plates, measurements were performed with
screw tightening of torque values monitored to be 1, 2, 3, 4 and 5 Nm.
The distal part of the femur rested in a mould of methylmetacrylate
upon a ball-bearing. The load to the proximal part of the femur was applied
through a ball-bearing onto a lever arm, which produced a compressive load
at the femur head and a tensile load at the greater trochanter (Figure 2).
The load applied was selected according to Rybicki et al. 2 (7), it ranged
from 0 to 1000N; this load resulted mainly in bending. The strain at each
gauge was recorded for each load level. The load-strain relationship was
plotted several cycles after initial adaptation.
 395

strain Fo
Bone
alone
500
Steel
1....../ ' Tit anium
....
_.......... Loa d
o
1000 (Nl

-500

Bone
-1000 alone

FIGURE 1. Superficial bone strain

under the plate and at the opposite
cortex in relationship to the ec-
c en tr ic load applied to the femur. FIGURE 2. Eccentric load applied to
Theoretical prediction. human femur.

4.2. Results.
The load-strain relationship showed a marked hysteresis for steel and
carbon epoxy plates but not for titanium and carbon polysulfone ones
(Figure 3). This was particularly evident for the gauges placed in the vi-
cinity of the plate. The load value at which a bend in the slope in the
strain-load relation appears was proportional to the torque applied to the
screws in the steel plate (Figure 4); this relation was less reliable in
the titanium plate, which required a higher load for appreciable hys-
teresis. The onset of slipping occured at 500N for steel; at the same load,
titanium did not slip. The hysteresis is related to the shearing displa-
cement between the plate and the bone. The surface of both steel and carbon
epoxy plates is very smooth compared with the two other plates.
396

Strain
(11 t) 3
2
500 500

o-filI!!i~_ _-'-_ _ _--r.....;;;L_oa d o~:-__""-___--r_L~oa.d

1000 (N) (N)

-500 - 500

-1000 -1000

-1500
@6 3 5
5
6
-1500
@6 5
6
(a) 4 (b) 4

FIGURE 3. Superficial bone strain

under the plate and at the opposite
cortex in relationship to the ec-
centric load applied to the femur.
Experimental determination.
a. Steel plate. b. Titanium plate.

Theoretical analysis using simple analytical methods leads to

convenient results for stainless steel plates; however more elaborate
methods are of interest for determining the effect of the ends of the plate
and the number of screws (1). The simple theoretical analysis predicts a
similar behaviour for steel and titanium plates; this was not verified by
the experimental verification. Smoothness of the plates seems to playa
 397

Strain

600 (~E) FIGURE 4. Relationship between the

, torque applied to the plate screws
- ,,
" and the bone strain measured next
400 , , to the plate when the slippage
, ,,
occurred. It was proportional to
-
,
the torque applied to the screws in
,
,
" the steel plate
200 ,
,,
"
,"
,-"
"" Torque
0 2 4 6 (N m)

more important role than that assumed by the measurement of frictional

coefficient (5, 6). The experimental results are similar and explain
results obtained by previous investigators (2) on steel plates.

Clinical relevance:
These measurements seem to indicate that the limit of frictional
transmission may be reached in the femur, while in the tibia, at loads
considered to be realistic "in vivo", no slipping was observed. When the
plate is applied using the tension band principle, high shear stress
between plate and bone may result. The advantages of this technique are
partially balanced by the danger of motion of the plate when the frictional
force is not high enough. It is therefore required to protect the fracture
area by a larger number of screws. Furthermore, rough plates (titanium,
carbon fiber reinforced polysulfone) seem to present advantages to smooth
plates (steel, carbon fiber reinforced epoxy resin).
If functional "in vivo" loads led to displacement between plate and
bone, resorption under the plate and subsequent loosening, cortical thin-
ning and fracture could resut. This would offer a different explanation for
bone loss following plate fixation than the hitherto accepted one of stress
protection.

Beaupre, G.S.; D.R. Carter; T.E. Orr et J. Csongradi: The importance of

friction interface in mathematical models of plated long-bones. 32nd
Annual ORS: 476 (New Orleans 1986).
398

2 Burny, F.; R. Bourgois; J.H. Aubriot et P. Halleux: Osteosynthese des

fractures hautes du femur. Premier symposium de biomecanique osseuse
CIBO. Acta orthop. belg. 38, suppl. I: 85 (1972).
3 Cordey, J. and S.M. Perren: The role of plate-bone friction in internal
fixation. In Moyen, B. (Edt.) Proc. Sec. Int. Symp. on Internal Fixation
of Fractures, Lyon - Sept. 16-18 (1982).
4 Cordey J. and S.M. Perren: Stress protection in femora plated by carbon
fiber and metallic plates. In Ducheyne, P.; Van der Perre, G. and Aubert
A. (Edts) Advance in biomaterials, Vol. 5: 189 (Elsevier, Amsterdam
1984).
5 Enzler, M.A. et S.M. Perren: Coefficient of plate-bone friction in vitro:
The effect of duration and amount of load. In Asmussen and Joergensen
(Edts.) Biomechanics VI-B: 239 (University Park Press, Baltimore 1978).
6 Hayes, W.C. et S.M. Perren: Plate-bone friction in the compression
fixation of fractures. Clin. orthop. 89: 236 (1972).
7 Rybicky et al.: On the mathematical analysis of stress in the human
femur. J. biomec. 5: 203 (1972).

Aknowledgements: This work, which is a chapter of a thesis of the Universi-

ty of Lausanne, was supported by the Swiss National Foundation for Scienti-
fic Research, part of the grant Number 3.857.0.83
 399

THE BENDING OF THE SCREW IN THE OSTEOSYNTHESIS WITH

AUTOCOMPRESSION PLATES
Th. Hopf and W. Krapf
Orthop. Univ. Hospital. Homburg/Saar. Germany

I. INTnOJUCTION
In osteosynthesis with selfcompressing plates the interfrag-
mentary compression is produced by especially shaped holes
with oblique gliding path. In the compression holes the axial
force of the screw is transformed in a transversally directed
force, which closes the fracture gap. The screw itself, especi-
ally its head and neck, is stressed by tension, torsion and
bending simultaneously, which leads to a deformation. The deg~
ree of this bending deformation was examined by a biomechani-
cal experiment.

II. METHODS
We used fresh human cadaver tibiae, which were osteotomised in
the middle of the shaft and small AC-Plates (OSTZO AG) with
cortical screws for the osteosynthesis. The principle of the
examination was the simultaneous measurement of the bending
angle of the screw and the interfragmentary compression of the
osteosynthesis. The bending angle was determined by a laser-
optic system. A small mirror was positioned on the head of the
compreSSion screw, another mirror was fixed on the surface of
the bone some centimetres away from the plate. The change of
the angle between these two mirrors, which was produced by the
bending of the screw, was optically measured by a laser light.
This optical measurement had no mechanical influence on the
examination and showed a preCision of 5 angle minutes. The in-
terfragmentary compression force was measured by a piezo-elec-
tric element, which was located in the osteotomy gap; its pre-
cision was at 0,5 kilopond.(Fig.1)

1. Passive bending of the screw

First the influence of the pure tension force of the plate
on the bending of the screw was determined. The screws were in-
serted in the compreSSion holes at different positions on the
oblique gliding path. Then the interfragmentary compression
was produced by an external MULLZR-tension instrument, which
was fixed in the collateral bone fragment griping into the
plate. So the screw was loaded rectangular to its axis by a
static force without using the dynamic compression effect. As
the AC-plate has two different compression holes with 27 0 and
45 0 gliding angle, we measured the bending of the screw sepa-
rate for each in dependence on the inter fragmentary compreSSion
and the position in the hole. The deepest position of the screw
in the hole at the end of the gliding path we called 0 U (U =
400

U~drehungen = rotations). The other positions were described

as 1/2, 1, 1 1/2, 2, 2 1/2 U, according to the gradually tur-
ning out of the screws. As one turn of the thread has a height
of 1,75 mm, 1 U means an increase of the distance screw head -
bone surface of 1,75 mm.

2. Bend~ng deformation whi~e screwing in

2.1. 27 - hole: in the 27 screw hole the measuring began at
zero kp and ended at 50 kp, to prevent the destruction of the
bone thread, which may cause an incorrect result. In three
screw positions ( 1, 1 1/2 and 2 U ) the tests were performed.
2.2. 45 0 _ hole: The examination of the 45 0 _ holes occured
after precompression of the osteosynthes~s, partly with the
27 0 _ hole screw like in the intraoperative matter, partly by
an external MULLER - tensioner. We used a precompression of
20, 40 and 60 kp. Then the insertion of the 45 0 _ screw at the
upper end of the slope took place. The examination was finis-
hed at 80 kp interfra 6 mentary compression. Naturally, in case
of lower precompression, the screw was - at the end of the
test - deeper in the hole, than in higher one. The precompres-
sion was produced by two different methods:
2.2.1. Once we tightened the 27 0 - screw up to 20, 40 and
60 kp and then we inserted the 45_ screw like in the intra-
operative way. Here first the 45 0 _ screw undertook the load
from the 27 0 - screw and then increased the interfragmentary
compression.
2.2.2. In a second series we set the 45 0 _ screw first loose
and then we produced the precomp~ession by an external MULLER-
tensioner, which was fixed at the contralateral fragment. So
the 45_ screw was just loaded in the period of pre compression
(20, 40 and 60 kp). If it was tightened now, it was stressed
more and more continuously.

3. Resulting plastic deformation of the screw

After all examinations the plastic deformation of the screw
was measured in situ by loosening the fixation side of the
plate to prevent a rebending of the screw by the screwdriver.

III. RESULTS
1. Passive bending of the screw
Here the screw was loaded only passively. Therefore the ben-
ding rates were relati~ely 10w 5 There was no significant dif-
ference between the 27 and 45 - holes. As expected, the ben-
ding angle was lower, i f the screw was placed in a deep hole
position and greater in higher positions. If the screw was in
the highest position ( 2 and 2 1/2 U ), the bending angle de-
creased again. In these two positions there was no more con-
tact between gliding path of the plate and screw head, because
it jutted out to much. The transformation of the load occured
now between plate and thread at the end of the hole. The dis-
tance of bone surface and the contact area between plate and
screw (i.e. the liver arm) was reduced and the deformation was
lower. These high hole positions, however, are not very impor-
tant for the clinical practice. Depending on the hole position
the bending angle was 0-3 degrees. (Fig.2)
 401

2.1. ae~ding defDrmation while screwing in, 27 0 _ hole screw:

This test showed a much greater bending rate. Depending on
hole position the angle was up to 20 degrees. The scattering of
this dynamic test was greater because of the higher systematic
mistake (friction!), but the dependency on the hole position
was clear. (Fig.3)
2.2.1. Bending deformation while screwing in, 45 0 -ho1e screw,
precompression with 27 e _ screw: Here we found essentially hig-
her bending angles than in the passive test up to 21 degrees.
It depended on the precompression. The higher it was, the grea-
ter the bending of the screw took place. (Fig.4)
2.2.2. Bending deformation while screwing in, 45 0 _ hole screw,
precompression with tensioner: The measured bending was up to
13 degrees. It was lower than in the last series, but essenti-
ally higher than in the passive test and depended also on the
precompression force. (Fig.5)

3. Resulting plastic deformation

In passive tension the bending of the screw was at 80% elas-
tic and at 20% plastic deformation.
In the dy-;a;;ic tests the plastic part of the deformation was
considerably higher:
- In the 27 0 _ hole the bending was at 90% plastic.
- In the 45 0 _ hole it depended on theway of precompressing:
i f it occured with the 27 0 _ screw, the plastic deformation
was 85 - 92%, in combination with tensioner it was 75 - 90%.

IV. DISCUSSION
The bending of the screw depends essentially on two factors:
- the interfragmentary compression
- the position of the screw on the slope gliding path.
As the interfragmentary compression should not be reduced be-
cause of the stability of the osteosynthesis, the only way to
get a low screw bending is to set the screw as deep as possible
into the compression hole. If the circumstances allow, the
screw should not generally be inserted at the upper end of the
gliding path, but one should estimate the required shifting
distance and then insert the screw at the right point of the
slope, that after screwing together the screw head lies deep
inside in the hole. This is, first of all in the 45 0 _ holes,
necessary.
In the shape of the AC-p1ate, in contrast to other samples
(i.e. the DCP of the AO)T the gliding path reaches nearly the
underside of the plate, that the screw bead juts up minimally
the bone surface.
The relatively high screw bending in the dynamic tests is ob-
viously caused by a material damage, which occurs by the simul-
taneous stress of tension, torsion and bending. A structural
weakening of the neck of the screw takes place. This is shown
by the comparison of the 45 0 _ hole tests with different precom-
pression. In precompression with the 27 0 _ hole the 45 0 _ screw
is deformed in the phase, i f it takes over the load without in-
creasing the interfragmentary compression. In the following in-
crease of compression it is damaged and bended more. The da-
mage is also visible microscopically by warpings and rips in
the material after use.
 Therefore it is wrong,to tiahten one screw extremely and then
screw in the next one. The 45 - compression screws must be
tighened alternatively so that the interfragmentary compression
force is distributed on two or more screws (MITTELMEIER).
According to PERREN a compression osteosynthesis has an elas-
tic reserve of 1 um in the plate (by elastic lengthening) ~
10 um in the bone (by elastic compression of the cortex). An-
other elastic reserve is produced by the overbending of the
plate. The elastic bending of the screws may also increase the
elastic reserve of the osteosynthesis (MITTELMEIER and DIEHL).
According to this examination the elastic bending has an
amount of 2-3 degrees. This corresponds to a horizontal shift
of the plate of 30-130 um, depending on the distance between
screw head and bone surface. A long distance has a greater
shift than a short one. This advantage, however, is combined
with a greater bending and material damage. A deeply inserted
screw has a small longitudinal shift and elastic reserve, but
also a low deformation and material stress.
This goes for the 27 and the 45_ screw in the same way.
As the maximal interframentary compression is essentially
lower in case of the 27 - screw,also the bending and the danger
of damage are lower than in the 45 0 _ screw boles.
In order to take care of the material we recommend a deep
insertion of the 45 0 _ screws. Against that the 27 0 - screw may
be inserted at the upper part of the slope to use the spring
effect of this screw.
Thus, one can combine the advantage of a long horizontal
shift with its better elasticity reserve and a small defor-
mation and damage of the material.

Y. REFERENCES
Krapf Wand Hopf Th: Experimentelle Untersuchungen ilber den op-
ti~alen Plattenvorbiegewinkel bei der Kompressionsosteosyn-
these unter Berilcksichtigung d~s plastisch-elastischen Ver-
haltens der Osteosyntheseplatte. Z.Orthop. 124 (1986) 592
Mittelmeier H: Prinzipien der Osteosynthese mit selbstspannen-
den Platten. MOT 4 (1974) 90
Milller HE,Allg8wer M and Willenegger H: Manual der Osteosynthe-
se.AO-Technik. Springer Berlin,Heidelberg,NY, 1977
Perren S: Pers8nliche Hitteilung

Fig.l: Test object with bone,plate,screws,piezoelectric ele

element and mirrors for the laser measuring.
 403

3 screw bending screw bending

3 1,75~
[degrees 1 1.75U', [ degrees 1 1,5U\~
1,5U '\
,,' 2U""\ ;,
2 "",', 1 U/\"
2 U ", 2
1 U..... ,I, '" \'.\
2.5U"~ I,:,1 o5 U"" \"\
I \ \ \\
1\\\
0,5U '\,\1,"t
....... \ \
\\',
\ \ ~\
\~ 2, 5lJo. , HI
OU
A
'_'\
--~- ,I,
OU~_:~

r
':
40 60
interfragm.
o 40 60 80
interfragm.
T
compres sion comp re SSJon
[ kp 1 [ kp 1

Fig.2: Screwbending in dependence on passive bending force

(=interfragmentary compression) and screw position.
left: 27 0 _ hole, right: 45 0 _ hole

20 screw bending
[degrees 1

15 a=2U
b =1,5 U
c =1 U

10

inter fragm.
III:;........_ _- -.....-_----_~,~ compression
o 20 40 60 0 [kp 1

?ig.3: Screw bending while screwing in, dependant on the inter-

fragmentary compression and the screw position,27 0 - hole.
(I1.2.1.)
404

20 screw bending
[ degrees 1

~ ... 60 kp precompression
15
~ 40 kp

10
- - - - - - .. 20 kp

in terfragmentary
o ~---2~0----:-4":'0---~6+::0----::8't:0--(I--!'O-'" compression [kp1

Fig.4: Screw bending while screwing in (45 0 -screw), dependant

on the inter fragmentary compression and precompression in
the 27 0 _ hole. (11.2.2.1.)

15 screw bending
[degree sl .. .... .......
60 kp precompressi on
10

kp

5
~
... 20 kp
'1-+-+ inter fragmentary
0 20 40 60 80 0 compression [kp 1

Fig.5: Screw bending while screwing in (45 0 _ screw), dependant

on the interfragmentary compression and pre compression with
the tensioner. (11.2.2.2.)
MICROCOMPUTER AIDED FINITE ELEMENT ANALYSIS IN FRACTURED LONG
BONES FIXED BY EXTERNAL PLATE
T. Zagrajek. K. Kedzior and P. Tulin
Warsaw University of Technology. Institute for Aircraft
Engineering and Applied Mechanics. Warsaw. Poland

1. INTRODUCTION
In surgical treatment of certain long-bone fractures, one of the most
commonly used internal fixation devices is the bone plate. Such plates
cause stresses in bone which are significantly different from physiological
stresses. Therefore, it would be useful to analyse such a fixation empha-
sing the stress and displacement in the bone tissue 121. At present the FEM
method is commonly used for this analysis 131. This method is effective but
since it requires big and fast computers, it cannot be used inside the op-
erating theatre.
The aim of this paper is to present a method which overcomes the men-
tioned above inconvenience. This method allows the use of microcomputers
in spite of the fact that the applied mathematical model is universal and
the calculations are precise. The method will be applied in analysing the
ZESPOL fixation system Ill. In the problem solution we have used experien-
ces from Aircraft Engineering with modyfication to our problem.
2. MECHANICAL MODEL
The mechanical model of the ZESPOL fixation system is shown in Fig. lao
The difference from the former construction is the distance between bone
and self compresion-plate. This change was made in order to eliminate a
plate inducted osteopenia 141. Some simplifications were made in designing
the model, namely: the shaft of the long bone was replaced by a tube, it
was assumed that the region of fixation is symmetrical at the fracture
plane, the fixation plate was replaced by a beam and the screw by bars.
These simplifications in the model are admissible since the dimensions of
the bone and its strength parameters are not precisely known. In order to
get mechanical data of the bone-fixator system we have carried out some
experiments (e.g. screw pUll-out test, plate bending test etc.).
External and internal loads are applied to the model. The external
loading results from the action of muscles, gravitation force, etc., while
internal loading results from fixation assembly parameters (pre-bending
of the plate, self-compression displacement, etc.). Both loads can be
modelled by three components of force and three components of torque.
The loads are introduced into a bone in Fourier series form (see Fig.lb)
and into the fixator in concentrated one. This fact shows reality in better
way.
3. MATHEMATICAL MODEL
The mathematical model was based upon FEM principles. The FEM mesh of
bone was composed of special axial-symmetrical shell elements. It was
assumed that the mechanical properties of bone are ortotropic and, what is
more, they are different in each element. FEM mesh of fixator was composed
of 3-D beam elements. The FEM mesh is shown in Fig. 2. While designing this
406

a)
Screw-bone
contact
r
Pw Assembly
parameters

External
loads

Bone

b)

.. , , Iu.._ _- ' ,
\
\
.-l-- , .~
I
I

'-I~~",,,,,,/
/ /
I

~1111111111"'~"'~"

Shearing Bending Pull-out

FIGURE 1. a) The mechanical model of the ZESPOL fixation system. Internal

loads: forces Pu , Pv , Pw; moments Mu , Mv , Mw. One can observe the stress
concentration near the screw. External loads: forces N, Ty , Tz and moments
Ms ' My, Mz . Assembly parameters: u - self-compression displacement; wI' w2
- pre-bending of the plate; a - plate-bone distance. Model is symmetrical
to fracture site. b) Examples of loads in Fourier series form. Using Fourier
series one can describe all force function forms.
 407

Fixator nodes Beam elements

~______,~~~~~I_I~~~'~~i~I~I~~~~ ~
. .-<
(()

(l.J

if-itflHit --tf.......... ~
+-'
(J
C1l
~

I l..L..

Contact nodes

FIGURE 2. Finite Element mesh for bone and fixator. The node of shell
element is the whole circumference. One half of symmetric model.
m
w
a) b)

L m

FIGURE 3. Coordinates and displacements of shell element. Symbols: y - draft

angle; I - length. Strains are in ortotropic coordinates.
mesh, the expected stress concentration and stiffness distribution were
taken into consideration.
Membrane strains of shell element (see Fig. 3) ES ' E~, y~,s in s,~ con-
figuration are:
au

1 (~ 1 + u + w ctg y) 0)
s a~ sin ~
v
- +
1 au
s s sin y - 3
408

and the change of curvature Xs' X , Xs are:

2w
x =a- -
s as2

1 a 2w 1 aw cos y dV (2)
2
2 ~ + 5 dS - 2 2
s sin y a s sin y a
1 a2w 1 aw cos y
cos y y av
Xs = 2( s Sln y 3S31 - -n2-=-- ~ - s sin y -as + )
s sin y s 2 Slny
.

We can describe u, v, w displacement as function of s, in following ,form:

n s. c,
u(s,) I: [u1(s)sin(i) + u 1 (s)cos(i)]
i=O
n s. c.
v (s , ) I: [v1(s)sin(i) + v1(s)cos(i)] (3)
i=O
n s c
w(s,) I: [w1(s)sin(i) + w1(s)cos(i<P)]
i=O
Next, using (3) and (2) to (1) and standard FEM principles we can determi-
nate the stiffness matrix of the shell element 151.
4. SIMULATION MODEL AND RESULTS
The computer simulation model is composed of two programs: one for the
fixator and other for bone. Both programs are joined by the data file and
cooperate in the way shown in Fig. 4. The calculations are made iteratively
until sufficient accuracy has been achieved. This technique makes use of
microcomputer possible.
A great many testing calculations have been made. For example, an analy-
sis has been conducted for a fixator meshed into 17 elements (18 nodes,
108 degrees of freedom) and bone meshed into 15 elements (16 nodes compri-
sing 16 harmonic components of force each, 2048 degrees of freedom). This
example results in a full solution (stress and displacement in bone) in
less than 45 minutes and only 5 minutes to get the appropriate solution
(all the components of loads on screw and the fracture site). The time of
calculation depends closely on the convergence of the iterative process.
The shorter time of calculations enables one to optiminize both the fixator
construction and assembly parameters. Figure 5 and Fig. 6 show some examples
of obtained results.
In order to validate the model we have used three tests:
- results of simple examples were compared with traditional analytical
method;
- results were compared with photo-elasticity tests;
- results were compared with strain-gauge tests.
5. CONCLUSIONS
The hitherto gained experiences in the use of this program show that:
- all types of fixator are very sensitive to assembly parameters and small
changes cause dramatic changes in local stresses in bone;
the accuracy of geometrical and mechanical parameters of bone determines
the accuracy of the results;
the present model is static, yet it is a good starting point for the
development of dynamic models.
 409

pre-be~iing of the plate

{U = assembl y

Loads in contaca nodes
{P1 = [B] U*}
t
r-
Total bone loads (external and internal)
{P} = {!~!d} + {p*}
t
OisPlace{~nts in bone
[K] U} = {p}
t
New contact nodes displacements
{U*} = {U*} - {U} .
t
~(Pi2)n+l - ~(Pi2)n NO
~(p.*2)
1 n
<
t YES
( STOP J
FIGURE 4. Flow diagram of problem solution. Symbols: {u*l- fixator displace-
ments, {U}- bone displacements, (Pl- fixator loads, {p*} - bone loads, [B1-
stiffness matrix of fixator, [K]- stiffness matrix of bone, E - acceptable
error, n - step number.

[MPa]
618
f 522 8"'r 206 9( 316
-14
.262 . 49.4 . 422

r
-12
-10 . Bending stress
\
-B
. . . . . . .
Q)
-6
-4
-2
\: .....
+'
.~

U1 2
Q)
~
4
::J
+'
6
Ll
co B
~
l1..
10
12
5max
s 46 6max
s 38 6max
s 20
6~in 30 6~in = 14 6~in 3
FIGURE 5. Stress distribution in longitudal cross section of the bone.
Assembly parameters: u - 0 mmj wI - 0.5 mm, w2 - 0.25 mmj a - 5 mm. Average
stress on fracture site 2.54 MPa.
410

a)
F r act u r e s i t e

F r act u r e s i t e

FIGURE 6. Longitudal stress distribution - maps of bone side surface:

a) membrane stressj b) bending stress. 8 maps from catalog of 24 parameters
can be chosen. Assembly parameters: u - 0.5 mmj wI - 0.5 mmj w2 - 0.25 mmj
a - 5 mm. Stresses in MPaj "+" marks screw.
REFERENCES
1. Bielawski et al.: Anwendung der Konstruktion als externer Stabilisator.
Drthop. Praxis, 5, pp.411-416, 1985.
2. Piekarski K.R.: Biomechanics of bone. In: Biomechanics VIlA, PWN and
University Park Press, Warsaw-Baltimore, pp.22-31, 1981.
3. Finite Elements in Biomechanics. Int. Conf. Proc., Tuscon, 1980.
4. Perren S.M.: The reaction of cortical bone to compression. Acta Drthop.
Scand. suppl., 125, pp.17-18, 1969.
5. Zienkiewicz D.C.: The Finite Element Method. 3rd Edition, McGraw-Hill,
London, 1977.
 411

THE INCORPORATION OF FRICTION INTERFACES IN A NON-LINEAR,

FINITE ELEMENT MODEL OF A PLATED LON6 BONE
6.S. Beaupre*, D.R. Carter*, T.E. Orr* and J. Csongradi*
Veterans Administration Medical Center*, RR&D, Palo Alta, and
Mech. Eng. Dept.**, Stanford Univ., Stanford, CA, U.S.A.

1. INTRODUCTION
In recent years much attention has been focused on the problem of refracture after the
removal of internal fixation plates. Two concepts that have been implicated as contributory
causes of cortical bone osteopenia or thinning which may result in refracture are: 1) mechanical
stress shielding; and 2) vascular insufficiency. The first of these two possibilities has received
considerable attention and is the focus of the present study.
A number of previous studies have investigated the degree of mechanical stress shielding
within plated bones using in vivo animal models, in vitro studies, and theoretical techniques.
The most extensive data on the stress and strain fields within a plated bone has been obtained
from finite element studies. However, the majority of previous finite element studies have made
relatively simplistic assumptions concerning the nature of the interfaces between the plate and
the bone and between the screw heads and the plate.
The effect of different interface conditions at the bone/plate interface has been examined
in two previous finite element studies. Simon et al. (8) modeled a small spacer in their three-
dimensional finite element model which alleviated any normal or shear stresses between the
plate and the bone. They did not present the results for this model but stated that the results
were not significantly different from the results for the model without a spacer. Cheal et al. (2)
modeled the bone/plate interface in three different ways in separate three-dimensional models.
Their first model employed a continuous interface which approximated composite beam theory,
another model used linear truss elements which eliminated shear stresses between the plate and
bone and the final model used non-linear truss elements to allow "near" stress-free separation
of the plate 'under tensile forces. They found that eliminating all shearing stresses between the
plate and bone resulted in approximately 10% less bone strain shielding compared with the
predictions from composite beam theory.
All of the previous theoretical studies have been based upon relatively simplistic assumptions
about the characteristics of the interfaces between the bone and the plate and between the screw
heads and the plate. Assuming that either no slippage or no shear exists at these interfaces,
these previous models can, at best, lead to estimates of the upper and lower bounds of the
stiffness and stress within a plated bone. However, these bounding values may not provide
good estimates of the stresses within an actual plated bone.
The purpose of this study was to examine the stress shielding which occurs after plate
fixation using finite element models employing more realistic interface conditions. The study
focused on the remodeling stage of healing, i.e., no fracture gap or osteotomy was simulated.
Particular emphasis was placed on the role of screw tightness and sliding, frictional interfaces
between the plate and the bone and between the screw heads and the plate.
412

2. METHODS AND MATERIALS

We modeled a six-hole forearm plate attached to the diaphysis of a idealized long-bone
(Fig. 1). The idealized bone had an inner diameter of 8 mm and and an outer diameter of
13 mm. The cross sectional area of the idealized bone was 82.5 mm 2 The area moment
of inertia was 1200 mm4. The idealized bone had an an elastic modulus of 18.5 GPa and a
Poisson's ratio of 0.30.
~

Figure 1. Quarter symmetric finite element mesh.

The plate which was modeled was based on a six-hole forearm plate. The plate was
73 mm long, 10.7 mm wide and 2.8 mm thick. The area moments of inertias in a plane
perpendicular to the long axis of the plate were 39.1 mm4 and 281.4 mm4. The cross sectional
area was 29.7 mm2 Models simulating both stainless steel (E=200 GPaj 11=0.30) and titanium
(E=110 GPaj 11=0.30) plates and screws were analyzed.
For ease of model generation, the scr~ws in the model had a square cross section with a cross
sectional area which matched that of standard 3.5 mm diameter cortical screws. The screws
passed through the plate with the top portions of the screw shafts allowed to displace in the
plate screw slots. Direct connection to the bone, simulating fully threaded cortices was imposed
where the screw shafts passed through the bone. In the finite element model, pre-tension in
the plate was not simulated. This corresponds to the clinical situation when the plate is used
as a neutralization plate, i.e., no axial compression in the bone is generated when applying the
plate.
Frictional interface elements were used at all interfaces between the plate and the bone and
between the screw heads and the plate. The coefficient of friction at the bone/plate interface
was based on literature values (5,6,7) and was equal to 0.30. The coefficient of friction at the
screw head/plate interface was determined from experiments and was also equal to 0.30.
The applied loading simulated both a static preload representing the forces exerted by the
screws and a cyclic, physiologic load. Three levels of static, axial preload in the screws were
considered. These were: 0 N, 100 Nand 200 N. The following physiologic loading modes
were considered: 1) bending closedj 2) bending open and 3) axial compression combined with
bending closed. In the combined axial and bending closed loading mode, the ratio of the
bending load to axial load was chosen so that 80% of the maximum strain was due to bending
with the remainder due to axial loading.
Two types of models were analyzed. These included: 1) a linear, continuous strain interface
(CSI) model having tight interfacesj and 2) a series of non-linear models which incorporated
frictional interfaces between the plate and the bone and between the screw heads and the plate.
As a qualitative check of the finite element model, the predicted strains at two locations on the
bone surface were compared with strain gage results from the in vitro study by Cordey and
Perren (4).
 413

3. RESULTS
Bending closed loading mode
The shift in location of the neutral axis for the bending closed loading mode is shown in
a longitudinal section in Fig. 2. The screw pre-tension for the two frictional interface models
was 100 N. The CSI model indicates an abrupt shift in the location of the neutral axis near the
outer screw. At the mid-plate cross section (at the right-hand edge of the figure) the neutral
axis for the CSI model is located within the upper cortex. With the frictional interface models,
the shift in neutral axis location is seen to be dependent upon the load level. At a load level of
1.0 Nm the shift is more gradual than for the CSI model, however the neutral axis location at
the mid-plate cross section is nearly the same. For the 5.0 Nm load, very little shift in neutral
axis location is observed, indicating that slippage has taken place. In this case the neutral axis
is located between the upper and lower cortices at all cross sections.

f--.L,--,.."i.. ____ r_--~' ____ I..,--=L-, CSI MODEL

I I I I I I I

~~~I----~~I-~==~~~~~
f ...j.::r
.y .. ,.............. I ....L-.-+.+---J -I
FRICTIONAL MODEL
BENDING CLOSED (1.0 Nm)

II II II~
______--" I I . . LI -
I I I I II I
-r--------1I- II I -i----I
J FRICTIONAL MODEL
BENDING CLOSED (5.0 Nm)

Figure 2. Shift in neutral axis location.

Contours of axial normal stress in transverse cross sections at the mid-plate location are
shown in Fig. 3a,b,c. The left-hand figure (3a) shows results for an unplated bone. Figures 3b,c
show the axial stress distributions in the frictional models for the 1.0 Nm and 5.0 Nm load levels
and a screw pre-tension of 100 N. Figure 3b shows that for the 1.0 Nm load level the location
of the neutral axis is shifted greatly toward the plate. At the 5.0 Nm load level (Fig. 3c) the
shift in neutral axis location is much less.

A B
:,.....-.. '. c :.........
.. .-',~
.,. ,
.
'.
-',
,

o
-1
5

Figure 3. Axial normal stress (MPa) at the mid-plate cross section (bending closed).
414

In Fig. 4 the axial normal strain at the mid-plate cross section for two locations on the bone
surface is plotted as a function of applied bending moment for the bending closed loading mode.
The left-hand half of the figure corresponds to a point on the periosteal bone surface opposite
the plate. The right-hand half of the figure corresponds to a point directly under the plate. For
each location on the bone, the strain must lie within the shaded wedge-shaped region defined
by a line representing the CSI model and by a. line representing the response of the unplated
bone. The values 0 N, 100 N, and 200 N represent the axial forces in the screws. The non-linear
nature of the response is caused by slippage between the plate and the bone and between the
screw heads and the plate. At low levels of applied load the system responds according to CSI
model predictions. As the load increases the frictional capacity to resist shear at the interface
is overcome and slippage occurs. For screw forces greater than 200 N the stress distributions
approximate the CSI stresses for loads which would produce less than 1000 microstrain in the
unplated bone.

APPLIED BENDING MOMENT (Nm)

...---.,
SCREW PRE FORCES
(NEWTONS)

2500 2000 1500 1000 500 0 500 1000 1500 2000 2500
BONE MICROSTRAIN

Figure 4. Axial normal strain opposite the plate (left) and under the plate (right).

BONE ON 100 N 200 N CSI CSI

ALONE SCREW SCREW SCREW STAINLESS TITANIUM
FORCE FORCE FORCE

DTENSION II COMPRESSION
Figure 5. Axial stress at the mid-plate cross section for a bending moment of 3.42 Nm.

Axial stress distributions at the mid-plate cross section are shown in Fig. 5 for a load level
of 3.42 Nm. This level of applied loading would produce 1000 microstrain in the unplated
bone. Illustrated in the figure are the regions that are in compression and in tension and the
location of the neutral axis or axes. This figure demonstrates that as the screws are tightened
 415

the neutral axes move closer toward one another. The two right-hand figures on the bottom
show the predictions from the CSI models using either a stainless steel or a titanium plate.
With the CSI models, the shift in location of the neutral axis caused by changing the plate
material is small compared with the shift in neutral axis location caused by screw tightness.

Bending open loading mode

The results for the bending open loading mode are qualitatively similar to the results for the
bending closed loading mode. However, significant quantitative differences do exist. For low
applied load the bending open response is similar to the bending closed load response. However,
at the 5.0 Nm load level the neutral axis is closer to the upper cortex for the bending open
loading and therefore significantly greater stress shielding occurs compared with the bending
closed load case. The magnitude of the axial stress directly under the plate at the mid-plate
cross section for the bending open loading mode is 7.7 MPa, while for the bending closed load
case it is 14.4 MPa.

Combined axial compression and bending closed loading mode

With the eccentric plate location, axial loading produces a bending response in addition to
axial compression. This bending is in the same direction as that caused by the bending closed
loading mode. The combined loading causes the neutral axis within the bone to shift toward
the plate. This loading mode was examined in order to compare the finite element predictions
with the results from the in vitro study by Cordey and Perren (4). This comparison is shown
in Fig. 6a,b. Shown in Fig. 6a are the in vitro results from a plated cadaver femur. Values
of the axial strain (measured by strain gages whose locations are indicated by the small dots)
were plotted versus applied load while the specimen was loaded and unloaded. The hysteresis
is an indication of slippage. Figure 6b shows the finite element predictions for the combined
axial compression and bending closed load case. Axial surface strains from the finite element
model for the two locations shown by the small dots are plotted versus applied load. A similar
hysteresis occurs.
EXPERIMENTAL DATA BENDING CLOSED + AXIAL COMPRESSION
A (CORDEY AND PERREN)
(tOO N SCREW FORCE)
B
~ _."~"'~ LOAD
STEP
LOAD

O.IONm
1.&ONm
+
+
n.2N
nl.4N
2.74Nm + lOS'IN
l:=:+m:~
LOAD STEP

Ullo
BONE MICROSTRAIN /.1< BONE MICROSTRAIN /.1<

Figure 6. Comparison of FE results and experiment for combined loading.

4. DISCUSSION AND CONCLUSIONS

The extent to which slippage occurs in vivo between the plate and the bone or between the
screw heads and the plate depends upon a number of factors. Of primary importance is the
magnitude chosen for the screw forces. Previous in vitro studies (1,3,5,6) have shown that the
416

maximum initial screw forces are dependent upon cortical thickness, among other things. For
the radius and ulna the maximum screw force from in vitro tests ranged from approximately
200 N to more than 1500 N. However, in the remodeling stage of fracture repair the initial
screw forces are no longer relevant. Unfortunately, the values of the screw forces after I! years
implantation are unknown.
Standard clinical practice dictates that plates are applied to the tension side of bones. In
this way in the early stages of healing the fracture fragments opposite the plate are compressed
together during physiological loading (bending closed loading). The qualitative similarity
between the responses for the bending closed and bending open loading in the present study
is a consequence of the lack of an osteotomy or fracture gap. The quantitative differences
that do exist between the two loading modes at high load level are a consequence of different
responses at the frictional interfaces. For the bending closed loading mode, the compressive
normal force at plate/bone interface near the outer screw is reduced by the applied load. The
reduced normal force results in reduced shear resistance and more slippage and therefore less
stress shielding. For the bending open load case, the compressive normal force at the interface
near the end of the plate is increased, leading to less slippage and more stress shielding at the
mid-plate cross section.
The results from the present study suggest that applying the plate on the tension side of
the bone is beneficial in the remodeling stage. At the higher levels of applied bending load,
significantly less stress shielding was predicted when the plated bone was subjected to the
bending closed loading mode compared with the bending open loading mode.
Previous theoretical models which have assumed strain continuity at the bone/plate and
screw head/plate interfaces may have over-predicted the extent of stress shielding in plated
long-bones. This study indicates that screw forces and frictional interfaces have profound
effects on the extent of predicted stress shielding within a plated long-bone. The results show
that the effect of slippage between the plate and the bone and between the screw heads and the
plate can be at least as important as plate material or rigidity and placement in determining
the degree of stress shielding.

References
1. Breuing KH, Gotzen L, Haas N, Hammer A: Biomechanical investigation of the new 3.5 mm
ASIF cortical screw. In: Biomechanics: Current Interdisciplinary Research, 1985, pp 435-
439
2. Cheal EJ, Hayes WC, White AA, Perren SM: Stress analysis of a simplified compression
plate fixation system for fractured bones. Comput Struct 17:845-855, 1983
3. Cordey J, Rahn BA, Perren SM: Human torque control in the use of bone screws. Current
Concepts of Internal Fixation of Fractures, 1980, pp 235-243
4. Cordey J, Perren SM: Limits of plate on bone friction in internal fixation of fractures. Trans
Orthop Res Soc 10:186, 1985
5. DeVito PD, Rivlin M, Price D, Goldstein SA, Matthews LS: Friction as a major factor in
plate and screw fracture fixation stability. Trans Orthop Res Soc 7:198, 1982
6. Gotzen L, Haas N, Hiitter J: Biomechanical studies on torque and force of the 4.5 mm AO
cortex screw as a lag screw. In: Current Concepts of Internal Fixation of Fractures, 1980,
pp 259-267
7. Hayes WC, Perren SM: Plate-bone friction in the compression fixation of fractures. Clin
Orthop 89:236-240, 1972
8. Simon BR, Woo SL-Y, Stanley GM, Olmstead SR, McCarty MP,Jemmott GF, Akeson WH:
Evaluation of one-, two-, and three-dimensional finite element and experimental models of
internal fixation plates. J Biomech 10:79-86, 1977

Acknowledgement: This work was supported by the Veterans Administration.

 417

THE SURGICAL EXPANSION DOWEL: A PROVEN PRINCIPLE OF MECHANICAL

FASTENING AND FIXING TECHNOLOGY TO OSTEOSYNTHETIC USES
W. Kramer, A. Fischer, R. Arlt and W. Neugebauer
Chirurgische Klinik der Universitat Tubingen, Tubingen,
Germany

INTRODUCTORY

Contrary to the cortical parts of a bone, the spongiose parts of meta-

physial sections of the skeleton possess elasticity reserves to a much
lesser extent only. Therefore, if pretensioning forces are to be applied
to such skeletal zones they can be maintained over no more than a very
limited timespan. Various investigations into the compression stress load
tolerances of bones (cf. RITTER and GRUNERT, It73; ~~ALDE et aI., 1975)
have shown a value range of 1,000 to 2,000 kp/mm (1,422,330 to 2,844,660
p. s. i. ), whereas the pressure loads which may come to bear on the
spong~osa is only one hundreth that magnitude, i. e. approx. 10 to 20
kp/mm (14,223.30 to 28,446.60 p.s.i.).

Screw-supported osteosyntheses anchored in the spongiose parts of bone

are at greatest danger when stress loads alternate. A great number of
surgeons have experienced that, particularly in the case of elder ly
patients suffering from osteoporosi t y, firmly seated screw anchorings
cannot be achieved. Here in fact, the transition from the maximum
momentum of screw tightening to that of thread stripping is an extremely
short one.

THE PLASTIC EXPANSION DOWEL

The search of alternative technological solutions to the question of how

to anchor screws in bases with a notoriously low pressure-resisting
capacity, the fastening method exemplified by the FISCHER plastic
expansion dowels is almost self-suggesting. Underlying the FISCHER dowel
principle is an augmentation of the anchoring strength, accomplished by
the use of tenaciously elastic plastic material. This material's
intrinsic quality of great elasticity (restoring force), coupled with a
high degree a f stress load resistance and a very distinct! y speci fic
exterior design, enable ist to absorb the influences exercised by
deforming forces. L~hile these dowels, which are presently manufactured in
a multi tude of di fferent configurations, were originally intended to
support anchorings ~n firm base materials (such as concrete), their
importance today its also increasing in other fields of application,
particularly where softer materials, low in pressure resistance (such as
gypsum wallboarding, for instance) must serve as the anchoring ground.
Fundamental to any safe, secure healing of bone fractures is the
immobilization of the individual fragments. In plate osteosynthesis,
screw fixation is, as a rule, performed in the diaphysial-cortical
sections of the bone. When anchorings must be placed in the metaphysial
418

osteospongiosa, however, such as in the head of tibia for example, the

limitations to safe and stable anchoring more often than not do become
all too apparent.

QUESTION

The objective of the research work reported on in this paper was to

answer the question as to which mechanical effects come to bear on the
spongiose bone if, instead of a screw used as the only means of fixation,
the fixation is accomplished with the assist of plastic expansion dowels.

MATERIALS AND METHODS

For the purpose stated, a number of

different plastic materials were
first tested for their usability, 100-
with particular consideration
given to their physical and N~ 80-
chemical properties. As a next ~
step, various forms of dowel
design underwent thorough 6 60-
examination with extraction, ~
torque and alternating load tests ~ 40
POLYPROPYLEN
performed. The finding that a
combination of great mechanical 20- PE-HA
stiffness, nonetheless coupled
with equally great restoring force 0-
(elasticity), must be demanded o 10 20 20 200 400 600 800 1000
from any such plastic material
came as one of the most essential EXTENSION t IN %
conclusions. The example given by
OBERBACH's diagram (Fig.l) Fig. 1: Diagramm by Oberbach
demonstrates this fact clearly:
In crystalline substance, the course of expansion shows a closed pattern
in the so-called "lower area", progressing in proportion to the force of
traction. The polyethylene examples (PE hard and PE soft) show, however,
that small tractional forces already suffice to produce extensive
expansion. This material's behavior is, in other words, extremely
elastic. Quite in contrast, the material Polyamide 6 (with fiberglass
reinforcement) requires a disproportionately greater tensile factor for a
similar degree of expansion to be obtained. But even so, elasticity in
this material is still not neutralized up to an extension factor of
approx. 20 per cent. We found that, for the purpose of our trials, the
materia~s best suited were Polyamide 6, Polyamide 12 and a polyester
(Gripet ) formulated on the basis of polyethylene-terephtalate.

The cyclic-load experiments were performed on trochanters of the femurs

of 68 adult domesticated swines. The reference experiments were in each
case performed on the contralateral trochanter. The microradiographic
pictures confirmed the impression given by the experimental results.
 419

An externally threaded plastic

expansion dowel, speci fically
developed in our preliminary
research for osteoapplications, is
emplaced into its location in the
bone with the assistance of a
specially developed dowel inserter.
Prior to dowel insertion, drilling
(4,5 mm) is performed, while the
internal thread is cut wi th a
special thread-tapping instrument.
The dowels are available in lengths
from 18 to 80 millimeters; they are
designed to receive regular cortex
screws as approved by the AO*
(Osteosynthetic Work Group). (Cf.
Fig. 2). Fig. 2: Dowel inserter,
Dowel,special thread-tapping
* AO = Arbeitsgemeinschaft fur Instrument
Osteosynthese

RESULTS AND DISCUSSION

In the foillwing, a particularly spectacular experiment is to be presented

from the great number of trial results obtained. A Schanz screw was
emplaced in the trochanter of the femur of an adult domesticated pig. The
screw boring depth was 30 mm. In trial A, this screw was inserted without
the extra support of a threaded dowel, whereas in trial B, the screw was
inserted with the dowel. The used screw was in both trial A and B the
standard Schanz scre~~. L~e are aware o.f the fact that by using screws of the
same diameter on both sides, we in fact introduce a bias in our
experimentation, as the diameters of the implants on both sides are there
by not comparable. But - having always the practical application in mind-
we intentionnably wanted to use only that screw, it being a standard AO-
screw-element. In each case, pretensioning was stipulated to a means of 200
Newtons. At a frequency of four c.p.s., 2,000 load alternations ~Jere
performed for each trial fixation.
load-alternation

Fig. 7: lrochanter region

of swine. Inserted
(?) Schanz screw or
dowel Schanz screw
l r
combination{lJ.
420

Fig. 3 the behavior of the

shm~s
dowel/screL~combination under the
cyclical stress of load alter-
nation. To reach the pretensioning
value stipulated, the distance of
the lever path created is 8 mm.
Upon trial ,termination, the
transversal deflection (i.e. the 250 N
path of thread) at the screw's
point of entrance into the cortical 125 N
layer bordering the bone is 2,5
millimeters. 2000 loa d
( ycle
Fig. 3: Cyclical stress of load 5mm
alternation screw-dowel-
combination

By comparison, the distance of the

lever path created with the
"unprotected" screw (Fig. 4) in
order for the stipulated
pretensioning value of 200 Newtons
to be obtained is 12 mm, with the
path of thread at the cortex border 240N
amounting to a distance of 6 mm.
Both tests were performed, on an all 120 N
electronically time/path-controlled
INSTRoN LOAD ALTERNATING
instrument. s
Fig. 4: Cyclical stress of load 6mm
alternation "unprotected"
screw.

The destructions caused by the axial longitudinal movement of the scret~

only on one side and of the screw-pIus-dowel on the other side. The
figures displayed are microradiographic images. Upon completion of the
trials, the screws were removed and the osteoblocks embedded in me thy 1-
methacrylate. The sections' tickness was approximately 70 ~.
 421

Fig. 5 shows the

condition of the
screw-pIus-dowel com-
bination. Even after
2.000 load alter-
nations, sUbstantial
effects of destruction
in the area of the
trabecular structures
cannot be identified.
The spongiose exhibits
a harmonic configura-
tion, a peripheral
seam compaction at the
dowel base is all that
is noticeable.

Fig. 5: After 2000

load-cycles
no damage peg base peg top

. In marked contrast to
the above, the
comminuted spongiose
structures in Fig. 6
clearly demonstrate
the impact of the
forces of deformation,
generated and caused
by the unprotected
screw, readily dis-
cernable as a
destruction of the
spongiosa's trabecular
system. It is
particularly in the
top area where nothing
but comminuted
structures are
observable.

Fig. 6: After 2000

load-cycles
massive
damage in scre1." base screw top
top area
This picture shows the implant si te on the
contralateral trochanter of the same animal - The
reason Irdhy the spacing beb~een the bone lamellas
looks bigger in this last picture is that this cut
was thinner (70 ~) than in Fig. 5 (120 ~).
422

These trial-characteristic pictures as have recurred in a great number of

experiments strongly hint at the basic principle which underlies the
screw-pIus-dowel combination: The deforming forces liberated by the
reverting stresses of cyclically alternating loads are, for the greatest
part, absorbed by the plastic material and have no chance of being
transferred to the spongiose structures.

CONCLUSION

To our knowledge, the experiments presented and discussed in this paper

have, up to this point in time, not been performed at any other place by
any other team of researchers. The principle of isoelastic anchoring of
implants, as in the hip endoprosthesis (MORSCHER, 1975), or as in the use
of the so-called isoelastic joints (applied on interphalangeal joints; cf.
MATHYS, 1975), has, admittedly, been given comparative evaluation, but the
principle of anchor establishment by the use of plastic expansion dowels
has hitherto not even been the subject of basic research.

The very specific nature of the spongiose bone, which fundamentally

differs from that of the cortical bone in mechanical terms, represents no
less than a true challenge to leave the path of conventional anchoring
with screws only and to explore further-going, advanced methods in order
that a substantially improved anchoring be accomplished. Most recent
research results obtained from animal trials must be viewed as strong
confirmation of this thesis. Publication of these results has already been
planned.

REFERENCES

FISCHER, A.: Handbuch fur Befestigungselemente.

Fischer-Werke (1982)

MORSCHER, E. W., R. MATHYS: Erste Erfahrungen mit einer zementlosen

isoelastischen Totalprothese der Hufte.
Z. Orthrop 113: 745 - 749 (1975).

RITTER, G., A. GRUNERT: Experimentelle Untersuchungen zu den mechanischen

Eigenschaften des Knochens im Hinblick auf die Druckosteo-
synthesen.
Archiv fur Orthopadische- und Unfallchirurgie
75: 302 - 316 (1973).

WALDE, H.J., G. RITTER und A. GRUNERT: Experimentelle Untersuchungen

zur Implantatverankerung im spongiosen Knochen.
Hefte Unfallheilkunde 126: 434 (1975).
 423

ANATOMICAL CURVATURE OF THE FEMORAL MEDUllARY CANAL FOR

INTRAMEDUllARY RODDING
K. Zuber. J. Eulenberger. E. Schneider and S.M. Perren
M.E.M.-Institute for Biomechanics. University of Bern. Bern.
Switzerland

1. IIiTRODUCTIOIi
The shape of most of the intramedullary and interlocking
nails commonly used has been determined empirically as a
segment of a circle. The radius of curvature of the Ktintscher
nail, e.g., has a length of 3500 mm, the Aesculap-Ktintscher
3500 mm, the Grosse-Kempf 3000 mm, the AO 4000 mm, and the
Klemm-Schellmann 3500 mm. Despite the fact that all these
implants do not match the geometry of the femoral cavity, the
application of "physiologically" curved nails is reported by
two groups (11,14,22). In clinical use, the exact location of
the point of insertion, axial malalignement, additional
comminution of the fractured area or perforation of the
distal metaphysis are related to a misfit between the
geometry of the femoral cavity and the shape of the implant
(4,10,11,12,15,16,20,21). Although the anatomy of the femur
has been studied extensively (5,8,9,13,17), limited data is
available with respect to the geometrical properties
pertinent for intramedullary nailing of this bone. Two groups
(11,14,22 and Winquist personal communication) determined the
radius of curvature of the femur in the antero-posterior
plane by sagittal x-rays. Radii of curvature of 1150 mm and
1350 mm respectively were found and the successful
application of prebent nails was reported. Because of the
small number of stUdies reported with respect to the geometry
of a "physiological" nail and because the plane with the
maximum radius of curvature of the femur does not normally
coincide with the sagittal plane, a study was undertaken to
determine the radius of curvature of an intra-medullary nail
fitting the femoral cavity exactly, as well as the
orientation of the plane of this curvature.

2. METHOD
X-rays were taken from 100 cadaver femora of a wide
population, ranging from 17 to 94 years (60 % males, 40 %
females). Excluded from the study were bones showing former
fractures, osteoporosis or any metabolic disorder influencing
the bone structure. The radius of curvature of the femoral
medullary canal, the orientation of the plane of this
curvature with respect to the plane touching the posterior
intercondylar surface, the width of the isthmus of the
intramedullary cavity and the nail length were determined
from these x-rays. In order to do this, the femur was fixed
424

Fig. 1: Mounting device allowing rotation and proper

positioning of the femur above the x-ray cassette.

at the tip of the greater trochanter and at the intercondylar

fossa in a special device (Fig. 1), in which the bone could
be rotated until the plane with the greatest curvature (as
determined by the periosteal surface of the femur) was
parallel to the x-ray film. The distance between x-ray focus,
bone and film was kept constant for all measurements. The
radius was measured from the x-ray using circles with radii
between 600 and 2300 mm in steps of 100 mm, drawn on tracing
paper and taking the x-ray magnification factor into account.
The radii had to optimally fit the cavity as well as the
proximal and distal end of the bones.
Six bones of different radii were used to investigate,
whether there is a difference between the position of the
planes of maximum curvature as determined from the periosteal
and endosteal surface of the bones. In three other bones with
different radii, the influence of rotation up to 60 degrees
around the long axis of the femur on the determination of the
radius of maximum curvature was examined.

3. REsm.TS
The difference between the radii of curvature as
determined from the outer and inner surface in the 6 bones
was found to be less than 100 mm, which lies within the
accuracy achievable with this method. The sensitivity of the
determination of the radius of curvature to rotation of the
femur is shown for three bones (Bl, B2, B3) of different
radii in Fig. 2. Within 30 degrees of rotation to either side
from the plane with the smallest curvature, the radii changed
by no more than 100 mm.
For the population investigated in this study, an average
radius of curvature of 1100 mm, ranging from 600 to 2100 mm,
was found (Fig. 3). The distribution of these values shows a
slight tendency towards the smaller radii. The average
rotation of the plane of maximum curvature with respect to
the posterior intercondylar surface was i 5 degrees antero-
laterally, ranging from 0 to 40 degrees. From the 100 bones,
22 % had angles between 0 and 10 degrees, 46 % were in the
range of 10 to 20 degrees, 17 % ranged from 20 to 30 degrees
 425

8'
!1500 B1
~
p::
B2
~

~
p:: 1000
~
U
Ii.
o
rg 500
Q

p:: -60 -30 0 +30 +60
ANGLE OF ROTATION [deg.]

Fig. 2: Influence of rotation of the femur around its long

axis on the determination of the radius of maximum
curvature in three bones (B1, B2, B3)

and 15 % had more than 30 degrees. The wid th of the isthmus

on the average was 14 mm, rang ing from 10 mm to 19 mm. The
average length of the nails, which would fit the bones best,
was found to be 430 mm, ranging from 370 mm to 500 mm. No
statistical correlation could be found among these parameters
nor with age.
n
10

500 1000 1500 2000

RADIUS OF CURVATURE [mm]
Fig. 3: Histogram of the radii of curvature of 100 femora

These results were verified experimentally as follows: out

of a series of custom made nails in the range of 900 mm to
2000 mm, the best fitting nail was selected for 6 femora with
widely varllng radii of curvature, split in half. The
differences between the radii determined by x-ray and the
fitting process were always smaller than 100 mm. In an
additional test, 8 femora of different radii and sizes were
fi tted with their corresponding nail without excessive
426

reaming. The point of insertion was chosen to be physio-

logically, i.e. in the long axis of the medullary canal,
where a lesion of the medial circumflex artery would be
avoided. The fit was checked by x-rays and a transsection at
the distal third. All nails showed a long tight contact and
the hole in the distal metaphyseal area was well centered.

4. DISCUSSION AND CONCLUSIONS

The measurement of the relevant curvature of the femoral
intramedullary cavity as well as the orientation of this
plane was possible using a simple, reproducible and
reasonably accurate method. The average radii determined
match the results of the two other studies (11,14, Winquist,
p.c.) well. The population of almost 400 femora investigated
in all three studies makes i t obvious, that the radii of
intramedullary nails used today do not match the cavities of
their prospective bones. Production of nails having smaller
radii or the prebending of nails during the operation should
therefore be considered. The determination of the position of
the plane of maximum curvature has shown, that in clinical
use, a sagittal x-ray with the femur in 15 degrees of
internal rotation is sufficient to determine the radius of
the nail to be implanted, should there ever be a family of
nails having different radii to choose from.
The absence of correlations between the radius of
curvature and any other of the parameters determined was to
be expected from literature. Selection of the appropriate
implant therefore has to be based upon the geometry of the
medullary cavity and cannot be predicted from any
anthropometric measurement.
Some of the complications encountered during intramedul-
lary nailing may be related to the geometric misfit between
nail and cavity. A false selection of the point of insertion
may lead to lesions of the capsule and the vessels which have
been reported to be important for the arterial blood supply
of the femoral head (12). Additional comminuti9n of the
cortical bone on the dorsomedial aspect of the femur may be
observed during treatment of mid shaft fractures. The reaming
may be excessive to fit a nail with a large radius of
curvature and cracks, fractures and/or fenestration may
occur. Malrotation, delayed union and pseudarthroses are also
reported (4,10,11,18,22).
The following advantages for clinical use of nails with
appropriate curvature could be confirmed: In distal fractures
with a long proximal main fragment, a straight nail would be
severely bent, whereas an appropriate nail would not be bent.
The consequence of such bending in a slotted nail is twisting
of the nail. The medial cortical bone needs lessexcessi ve
reaming. Less additional comminution should occur at the
fracture site. Thinning and fenestration in the distal
metaphyseal femur may be avoided. The fractures may be
reduced more anatomically, although the impact on mal-
positioned neighbouring joints has not yet been examined
sufficiently (4,11,14,19,22).
When following the curvature of the intramedullary cavity
 427

from distal to proximal, the piriformis fossa and the

insertion of the external obturator muscle in the fossa
trochanterica result as the "physiological" point of
insertion. The piriformis muscle inserts on this small notch
on the trochanteric crest. From a lateral point of view, the
"physiological" point of insertion therefore lies on the
lateral border of the trochanteric fossa. The insertion of
the capsule begins 10 mm medially to this site. The classical
point of insertion lies on the lateral side of the greater
trochanter (6,10). Today the dorsal tip of the greater
trochanter is used as described above. Attention must be paid
to the ascending branch of the medial circumflex artery,
which may be damaged in the trochanteric fossa. The ascending
branch passes dorsally to the attachement of the external
obturator, penetrates the capsule and branches the latero-
dorsal epiphyseal arteries, which represent the blood supply
of 2/3 of the weight bearing part of the femoral head. If a
lesion of this vessel group occurs, an avascular necrosis of
the femoral head may occur (1,2,3,7,12,19).
In conclusion, it must be stated, that none of the intra-
medullary nails hitherto used has a curvature comparable to
our results. It is to be expected, that a nail having a
radius of curvature close to the one of the femoral medullary
canal, has a better contact to the bone without the need of
excessive reaming. Such a nail might lead to smaller
misalignement of the main fragments of the bone, especially
in the sagittal plane. Less additional comminution and
splitting of the femur might occur and the nail will remain
distally within the femoral canal all the way down to the
subchondral bone. Considering these advantages, a curved nail
having a larger wall thickness with a higher bending
stiffness and smaller elasticity may be produced to achieve
the results of the presently available thinner nails with
smaller stiffness. Further clinical trials, however, have to
be performed to further elucidate the role of appropriate fit
between intramedullary nails and bone on fracture healing and
to investigate, whether for extreme cases of femur curvatures
peroperative bending is necessitated.

REFERENCES
1. Batory, I.: J(tiologie der pathologischen Veranderungen
des kindlichen HUftgelenkes. BUcherei des Orthopaden,
Enke Verlag, Stuttgart, 32: 25-27, 1983.
2. Chung, S.M.K.: The arterial supply of the developing
proximal end of the human femur. J. Bone Joint Surg.,
58A: 961-970, 1976.
3. Crock, H. V .: A rev ision of the anatomy of the arteries
supplying the upper end of the human femur. J. Anat., 99
(1): 77-88, 1965
4. Hansen, S.T., Winquist, R.A.: Closed intramedullary
nailing of the femur. Clin. Orthop. 138: 56-61, 1979.
5. Holtby, J.R.D.: Some indices and measurements of the
modern femur. J. Anat. Physiol. LII. (Third ser. Vol
VIII), 363-379, 1918.
6. KUntscher, G.: Die Marknagelung von KnochenbrUchen. Arch.
428

klin. Chir. 200: 443-455, 1940.

7. von Lanz, T., Wachsmuth, W.: Praktische Ana tomie , II 4
Bein und Statik. Springer Verlag, Berlin/Heidelberg 1972.
8. Martin, R.: Lehrbuch der Anthropologie. Osteologie.
G. Fischer, Jena: 995, 1038, 1043-1047, 1141-1143, 1928.
9. Merkel, F., Bonnet, R.: Anatomie und Entwicklungsge-
schichte des Femur. In: Anatomische Hefte. J.F. Bergmann,
Wiesbaden. 610, 623-625, 1901.
10. MUller, M.E., Allgower, M., Schneider, R., Willenegger,
H.: Manual der Osteosynthese. Springer Verlag, Berlin,
104-119, 1969.
11.0noue, Y., Sunami, Y., Fujiwara, T., Sadakane, T.,
Yasuda, S.: Treatment of the femoral shaft fracture with
a curved heat-treated COP clover-leaf nail. International
Orthopaedics (SICOT) 3: 203-210, 1979.
12. Pelzl, H.: HUftnekrose nach Femurschaftnagelung. Zbl.
Chir. 95: 202-207, 1970.
13. Parsons, F.G.: The characters of the english thigh-bone.
J. Anat. Physiol. XLVIII. (Third ser. Vol IX) 238-267,
1914.
14. Sadakane, T., Nagano, K., Onoue, Y., Sunami, Y., Engel-
hardt, P.: Anatomische und biomechanische Untersuchungen
zur Femurmarknagelung. Arch. Orthop. Traum. Surg. 91:
31-37, 1978.
15. Raaymakers, E.L.F.B., Marti, R.K.: Spatkomplikationen
nach Marknagelung am Oberschenkel. Hefte zur Unfallheil-
kunde 158: 192-195, 1982.
16. Reinders, J., Mockwitz, J.:Technical faults and
complications in interlocking nailing of femoral and
tibial fractures. Acta Orthop. Belg. Tome 50, Fasc. 5:
577-590, 1984.
17. Ruff, C.B., Hayes, W.C.:Cross-sectional geometry of Pecos
Pueblo femora and tibiae. A biomechanical investigation
I/II. Amer. J. Phys. Anthrop. 60: 359-381, 383-400, 1983.
18. Ryan, J.R., Rowe, D.E., Salciccioli, G.G.: Prophylactical
internal fixation of the femur for neoplastic lesions. J.
Bone Joint Surg. 58-A(8), 1071-1074, 1976.
19. Tencer A.F., Sherman, M.C., Johnson, K.D.: Biomechanical
factors affecting fracture stability and femoral bursting
in closed intramedullary rod fixation of femur fractures.
J. Biomech. Engineering, 107: 104-111, 1985.
20. Trueta, J.: The normal vascular anatomy of the femoral
head in adult man. J.Bone Joint Surg. 35B: 442-461, 1953.
21. Weller, S., Renne, J.: Grundsatzliche Fehler und Kompli-
kationsmoglichkeiten der Marknagelung. Chirurg, 44:
533-538, 1979.
22. Winquist, R.A., Hansen, S.T., Clawson, D.K.: Closed
intramedullary nailing of femoral fractures. J. Bone
Joint Surg., 66-A(4), 529-539,_1984.
ACKNOWLEDGEMENT
We would like to thank the Department of Radiology of the
Inselspital Bern, especially Ms Matter and Mr Fiechter, for
their comprehension and PD Dr. R. Zehnder of the AO
Documention Center, Bern for the statistical analysis.
 429

COMPARATIVE INVESTIGATION ON THE BIOMECHANICAL PROPERTIES OF

THE CIRCULAR AND OTHER THREE-DIMENSIONAL EXTERNAL FIXATORS
B. Gasser. D. Wyder and E. Schneider
'M.E.M.-Institute for Biomechanics. University of Bern. Bern.
Switzerland

L INTRODUCTION
External Fixation has developed during recent years to one
of the standard methods of fracture treatment. A variety of
fixators exists for a wide range of indications (4.5) as open
fractures. pseudarthroses and limb lengthening. Numerous
investigations (1.2.3) have been done comparing the properties
of different types of external fixators among themselves.
Although relevant and comparable parameters of clinical or
mechanical nature are not trivial to find, stiffness of the
fixator frame under different loading conditions has generally
been accepted to be an important property. Fracture heal ing is
believed to be accelerated when stiffness is reduced
appropriately during the healing process.
Among other circular fixators (5), the compression and
distraction device developed by Prof. Ilizarov from Kurgan is
of particular interest in this respect. His circular device has
shown amazing clfnical results in the treatment of
pseudarthroses with bone defects and in limb lengthening. The
main difference between this circular and a conventional
external fixator is the use of multiple, pretensioned, thin
wires leading to a fundamentally different behaviour of
stiffness with load.
The goal of this project was to determine the stiffness of
the circular fixator as a function of several configuration
parameters and to compare the stiffness of this device with the
standard "Hoffmann-Vidal" as well as the un il ateral and
triangular AO type I and III external fixators.

2. MATERIAL AND METHODS

The Finite Element (FE) method was used to calculate the
stiffness of the circular fixator for various configurational
parameters. The sti ffness of sel ected frames of 11 i zarov, Hoff-
mann-Vidal and the AO was determined by material s testing.
The reason to apply the FE - method was to perform the
variation of parameters as effective as possible. The mesh of
the different, FE - models was generated by the program FEMGEN.
The FE - program ADINA was used for the nonlinear calculations
considering large deformations. Taking advantage of the
symmetry, only the lower half of the model, in the standard
case consisting of 173 nodes and 167 beam elements with
different cross-sections for rings, rods, pins and bone, was
analyzed (Fig. 1). In our model, three rods were connecting
430

each of the rings. One of the three rods between the inner
rings was situated anteriorly, one of each of the three rods
between the outer rings posteriorly for definition of A-P- and
M-L- bending. The bone was modelled as a tube with an outer
diameter of 30.0 mm, a wall thickness of 5.0 mm and a total
length of 33.5 cm. The fracture gap had a size of 10 mm. The
standard configuration was defined as:

inner ring distance: 12.5 cm outer ring distance: 7.5 cm

ring radius 8.25 cm pin diameter 1.8 mm
pin pretension 900 N

Keeping the bone geometry and all other parameters constant,

one of them was varied in both directions. The total variation
was 2.5 cm and 3.5 cm for the outer and inner ring distance
respectively. The ring radius was varied by 2.0 cm, the pin
diameter by 0.2 mm and the pretension in the pins by 300 N. The
material behaviour was accepte~ to be linear elastic with a
Young's modulus of 200'000 N/mm for all fixator materials and
3'300 N/mm2 for the bone model. Poisson's ratio was 0.3 for
bone and fixator. Material properties of the bone corresponded
to those of the synthetic model used for the experimental
tests. Mechanical characteristics were determined in 11 load
steps with a maximum load of 1000 N for axial compression,
15 Nm for torsion and 20 Nm for A-P- and M-L- 4-point-bending.
The first load step was used to pretension the pins. With the
remaining 10 steps, the external loads were applied. Corres-
ponding to the load case situation, the axial gap displacement
(a), the lateral gap displacement (b resp. c) or the total
twisting angle (t) were calculated and the stiffness of the
system determined:

Axial stiffness k AX F/a (N/mm)

A-P - Bending stiffness k AP k1 * Mbl b (Nmm 2 )

M-L - Bending stiffness kML k1 * Mbl c (Nmm 2 )

Torsional stiffness k TO k2 * Mt/t (Nmm 2 )

where k1 is a boundary condition and geometry dependent factor

and k2 only depends on the total length L with:

k1 = q*L 2 and k2 = L (q = 0.125, L = 33.5 cm)

An original circular fixator with the standard configuration
was investigated on an INSTRON testing machine for
investigation of the difference between the actual fixator and
the FE - model, where all the connections between the separate
fixator elements are ideal, i.e. rigid. To estimate the
influence of slippage of the pins through the bone, our
standard frame with slippage occurring (ILI1) was modified by
fixing the pretensioned pins with the synthetic bone to prevent
sl ippage and by strengthening the connection between rings and
rods (ILI2). With this modified frame "ILI2", two of the most
 431

important factors responsibl e for the difference between the

numerically and experimentally determined load deformation
curves could be eliminated. In addition, the comparison between
the conventional and the modified device yielded an evaluation
of the discrepancy between the most extreme possibilities
concerning slippage of the pins through the bone.
For a better assessment of the stiffness values calculated
and measured, the standard "Hoffmann-Vidal" device (1,2), the
triangular type III and the unilateral type I of the AO (4)
were tested. The first two are 3-D arrangements such as the
circular device. The unilateral AO type I with two rods was
examined because of its simple design and widespread use. For
ease of comparison, the geometry of these fixators, especially
the dimensions corresponding to the varied frame parameters,
was chosen to be as close as possible to the standard circular
frame. On the other side, the configurations were conform to
their clinical usage.

3. RESULTS
The results of the stiffness calculations of the circular
device by means of the FE analysis are shown in table 1. The
ring radius was the parameter with the most significant
influence under axial compression. A reduction of 2.0 cm
resulted in a rise of stiffness of 77% for 1000 Nand 86% for
500 N axial compression load. A larger ring radius reduced the
axial stiffness approximately 32%. An increase in the pin
diameter and its pretension yielded a larger stiffness about
10%. Inner and outer ring distance had no influence on axial
stiffness. No difference was found between bending stiffness in
A-P- or M-L - direction. With the bone and total frame length
being constant, a larger outer and above all an increasing
inner ring distance increased bending stiffness. An essential
result was the reduction of bending stiffness by almost 15%,
when increasing the ring radius. Pin diameter and pin pre-
tension variation resulted in a negligible influence (less than
3%) on bending stiffness behaviour. The torsional stiffness was
increased about 10% by an increase of 0.2 mm of the pin dia-
meter or 300 N of pretension. An increase of the ring radius of
2.0 cm to the contrary yielded a similar increase in stiffness.
Differences in torsional stiffness according to variations of
inner and outer ring distance are smaller. In particular, due
to variation of the inner ring distance, the total twisting
angle remains almost constant.
Figure 2 shows the comparison of the numerical and experi-
mental load - deformation curve under axial compression. The
maximum difference in stiffness is 10% for 700 N axial load.
Both curves confirm the load dependence of the stiffness of the
circular device under axial load.
Load - deformation curves of all model s tested are shown in
Figure 3. The axial stiffnesses of the conventional (ILll) and
the modified (ILl2) circular devices were distinctly smaller
than those of the "Hoffmann-Vidal" and the AO types I and III.
The circular fixator was also the only one with a nonlinear
behaviour. Under bending, there was a considerable difference
between the "ILIl" and the "ILl2" circular model exhibiting the
432

influence of slippage between pins and bone. In the "ILl1"-

curve~ pin slippage occurred at a bending moment of 4:5 Nm: For
A-P bending, the unilateral type AO-1 was the stiffest
configuration and the "Hoffmann-Vidal" frame as a consequence
of its construction proved to be the weakest fixator type.
Under M-L bend ing, the un il ateral type approx imated the
conventional circular device and the "Hoffmann-Vidal" was
considerably stiffer in comparison with A-P- bending. There was
almost no difference for A-P and M-L bending of the
circular type. The triangular type AO-3 showed a more rigid
behaviour for M-L bending, because its basic frame lies in the
M-L plane similarly to the "Hoffmann-Vidal" fixator. The
torsional stiffness of the circular model was comparable to the
other types. The weakest model for this load case was the
unilateral frame, which additionally allowed for a lateral dis-
placement of the fragments. The triangular type AO-3 was again
the most rigid configuration.

4. DISCUSSION
The loads applied were chosen to guarantee that the range
tested included the physiological loads. The configurations of
the circular and other types of external fixators analyzed were
considered to be a standard model of clinical application. The
experimental tests and calculations with a synthetic bone model
were assumed to be appropriate for this comparative inves-
tigation.
The most significant finding was the load-dependence of the
axial stiffness, which has not been described before in the
literature. It rises from 90 N/mm for the first load step to
173 N/mm for the last one (Fig. 2). This nonlinearity is based
on the behaviour of the thin pins, which transmit tension
forces strongly depending on the deformation and influencing
the bending stiffness of the pins. Axial compression was the
load case with the largest pin deformation and yielded there-
fore the most significant nonlinearity. For the load ranges
examined, the nonlinearity was insignificant under bending and
torsion. Especially the bending load case showed, that rings
rotate as rigid planes so that no deformation occurred in the
pins. The curve "ILl2" of the modified circular model, being
similar to that one of the mathematical model, confirms the
almost 1 inear bend ing sti ffness. The discrepancy between the
numerical and experimental load deformation curve can be
explained with the slippage of the pins through bone and clamps
and the difference in rigidity of the connection between the
separate frame parts. Additionally, it must be assumed that pin
pretension of the tested model, in contrary to the FE - model,
was reduced before the measurements even started because of
creep.
Except for the axial load case, the circular fixator was
comparable to the other types of fixators. It might well be
that the axial elasticity may be an explanation for the
c 1 i n i cal s u c c e s s 0 f the d e vic e wh i c h, 0 f c 0 u r s e, i s not 1 e s s
influenced by clinical aspects not investigated in this
project. Stiffness of the fixation devices was selected in this
study because its definition is applicable to every kind of
 433

device and allows for their comparison. It must be noted,

however; that in a study like the present one, the question
cannot be answered, whether high or low stiffness values or an
intelligent combination of both leads to optimal fracture
treatment: This can only be investigated clinically. But the
present study may contribute the necessary tool s for proper
management of the fixator with respect to its properties.

REFERENC ES
1. Behrens, F., Johnson, W.O., Koch, T.W. and Kovacevic, N.:
Bending Stiffness of Unilateral and Bilateral Fixator
Frames; 1983; Clin. Orthop. and Rel. Res. 178; p. 103.
2. Briggs, B.T. and Chao, E.Y.S.: The Mechanical Performance of
the Standard Hoffmann- Vidal Fixation Apparatus; 1982; J.
Bone and Joint Surg. 64-A; p. 566.
3. Brooker, A.F., Cooney, W.P. and Chao, E.Y.S.: "Principles of
External Fixation"; 1983; Waverly Press, Inc., Baltimore,
USA.
4. Hierholzer, G., Kleining, R., Hoerster, G. and Zemenides,
P.: External Fixation; 1978; Arch. O:-thop. Traumat. Surg.
92; p. 175.
5. Volkov, M.V. and Oganesyan, O.V.: Treatment of Fractures in
Tubular Bones by the Vol kov-Oganesyan Reposi tioning-
Compression Apparatus; 1984; Cl in. Orthop. and r e 1. Re s
186; p. 195.

TABLE 1. Stiffness values of the configurations of the circular

fixation device from FE calculations

Axial A-P - M-L -

Com pre s s ion Bending Bending Torsion
(1000 N) (20 Nm) (20 Nm) ( 15 Nm)

stand ard mod el 119.6 N/mm 155.0 Nm 2 153.3 Nm 2 43.6 Nm 2

= 100 % = 100 % = 100 % = 100 %

outer ring
distance:
5.0 cm 99.9 % 87.9 % 90.6 % 95.5 %
10.0 cm 100.2 % 101.1 % 101.1 % 104.7 %
inn er ring
distance:
9.0 cm 100.2 % 88.3 % 88.4 % 99.5 %
16.0 cm 99.9 % 111.0 % 111.6 % 100.4 %
ring radius:
6.25 cm 177 5 % 113.9 % 114.4 % 114.1 %
10.25 cm 68.1 % 87.4 % 88.8 % 93.1 %
pin diameter:
1.6 mm 92.6 % 97.9 % 97.8 % 90.0 %
2.0 mm 108.7 % 102.3 % 102.2 % 111. 8 %
pin pr e ten s ion:
600 N 89.1 % 98.4 % 98.4 % 86.9 %
1200 N 112.2 % 101. 7 % 101. 7 % 112.6 %
434

" AMIAL CO"PRESSIOH

.
.-.. .'
-
788

......
"
",,'
..
~ .",.- ,,-- "

i '
.'

MeIM. DIIPUCIftIHT
4.' ...-
Fig. 1: FE - model of the Fig. 2: Num. and expo
circular ext. fix. device. load - deformation curve.

"
788
MeIM. CCIIIREIIICIt
.. A-P IIItDlrtG

see

- MeIM.
4.'
DIIPI~

6.' ......
I.

...,.,.- ---I.' ....- _ -

Ll'TIIW. DIIPI Al2I'EtiT

4.'
~.--.
H-U

6.' _
....-'"
.".

a) b)

.. . ," 15
...
...-I
TGaIClt
ILII

,-
I. i
II fJG-3/ / ILU ...
H-U.. ,Jtt'
II
....
,... ....fJG-l
,: ....fJG-l
, .. ...
~

5
, :.'

I.' 4.' 6.' _

LATEM. DIIftLACEI'IErfT
5.'

c) d)
FiS. 3: External Fixators (ILll: standard Ilizarov, ILl2:
mo ified Ilizarov, H-V: Hoffmann-Vidal, AD-I: AD type I and
AD-3: AD type III) under a) axial compression, b) A-P -
c) M-L - bending and d) torsion.
 435

EVALUATION OF INCREASING STIFFNESS OF HEALING BONE USING

EXTERNAL FIXATION
w. Siebels. R. Ascherl. H. Brehme. H. Albersdorfer and
G. Blumel
Institute for Experimental Surgery. Technical University.
Munchen. Germany

INTRODUCTION
Experimental studies dealing with mechanical properties of
healing bone mostly use destructive materials testing proce-
dures such as tension, compression, bending or torsion tests.
The evaluation of time dependent changes in biomechanical para-
meters require a considerable number of specimens to diminish
interindividual differences. Furthermore these analyses need a
minimum of stability of the explanted bones to perform testing.
Black and his coworkers (1) e. g. were not able to measure
mechanical data prior to the 11th p. OPe day. Other authors
(3, 10, 14, 15) started only after nearly three weeks. Non-
destructive examinations such as radiographies proved to be
unreliable to determine mechanical properties of bone,
especially in the early stage of healing (1, 4, 8, 9).
Clinically, Jernberger (6) tried to improve the manual
testing of stiffness by fixing a bending device to the bone and
measuring the applied load versus deformation. Jorgensen (7)
used a similar loading technique with external fixation. Edholm
and coworkers (2) bent the healing bone and took radiographies
of loaded and unloaded legs to calculate a stiffness parameter
from angular deviations. Our early experimental studies on
rabbits also used angular deviations under bending ~oad to
measure stiffness of healing osteotomies treated with external
fixation. To achieve a deviation of 1 0 it was necessary to
remove some stabilization elements of the external fixator and
mount the Schanz - screws onto a bending device. This required
a sufficient bending stiffness to properly refix the external
fixation. Therefore measurements could be performed not before
14 days after the operation. We therefore designed a strain
gauge technique to evaluate stiffness properties without remo-
val of the fixation frame.
436

MATERIALS AND METHODS

Eight mature male and female white New Zealand rabbits
(weight 29 ~ 5 N at surgery) were used in this study. The
animals were anaesthetized (Ketamin/Xylazin) and the right
tibia and fibula were exposed by a longitudinal lateral inci-
sion. Then the fibula was totally removed and the periosteum of
the tibia was incised and circularly elevated to fix a two
dimensional drill guide. Four holes (2.0 mm) were placed and
Schanz - screws (Kirschner wire 2.5 mm with machined thread
M 2.6) were inserted.
Both dorsolateral screws run through the bone perforating the
contralateral skin for 10 mm. The other screws, placed at 90 0
to the first plane, just transfixed both cortices. After remo-
val of the drill guide a modified AO - Mini external fixation
frame was mounted. Then a transverse osteotomy of the tibia was
performed at the distal tibia-fibular junction (halfway between
the screws) using an oscillating saw. The resulting osteotomy
gap measured lmm (fig. 1). The applied fixation arrangement
gave an~initial axial compression stiffness of 30 N/mm.

Figure 1. Radiographs of osteotomized tibia 0 and 5 weeks p.op_

 437

On the 7th p. op. day the animals were again anaesthetized and
two foil strain gauges measuring 3.2 x 5.0 mm (Hottinger &
Baldwin) were applied at two Schanz - screws near by the clamps
(fig. 2). A pair of pliars, also instrumented with strain
gauges were placed on the medial ends of the two perforating
srews. Tension and compression forces were applied manually.
Measurements were repeated every three or four days. Two plane
radiographs were taken once a week.

Figure 2. Application of a foil strain gauge on a Schanz - srew

With progress of healing of the osteotomies strains on the

screws reduced under comparable loads. More load is trans-
mitted through the bone and therefore the slope of the force -
strain curves increase (fig. 3). Thirty-five days after opera-
tion the rabbits were sacrificed and five of the explanted
tibiae were tested in four-point bending (fig. 4)

RESULTS
There were no postoperative complications. Swelling usually
receded within four to five days. The animals began partial
weight-bearing on the operated limb after two days. After ten
days full weight-bearing was apparently resumed. Callus forma-
tion was radiographically visible not before two weeks after
the operation. A circullary periosteal callus appeared to be
flat on the- lateral side and arched on the medial side. The
osteotomy gap remained visible throughout the observation
438

period of five weeks (fig. 1). Histological investigations

demonstrated endosteal callus formation too.
90

force
[NJ
10

100 strain [~mlm]

6 10 14 18 22 26 30days p.op.
Figure 3. Force - strain curves of one animal as recorded.
Slopes of force - strain curves of 8 animals

Increase of stiffness could be detected with this strain gauge

method from the 10 th day p.op. with rapid progress (fig. 3).
The technique was able to determine valid results up to an
average of twelve days following the commencement of stiffness
increase (approximately 22 days after operation). From that
time on, the callus became so rigid that no detectable deforma-
tion occured when forces up to 10 N were applied to the f ixa-
tion frame. Thus the upper plateau in figure 3 expresses the
end of measureable bending of the Schanz - screws and does not
reflect any end in the stiffness increase of the healing bone.
The Schanz - screws where load was introduced manually were not
as suitable for strain measurements as the unloaded pins.
Directly transfered bending moments through the loaded screws
caused superimpositions to gap movement signals.
In four-point bending of the explanted specimens without fixa-
tion frame stiffness as well as ultimate bending moment were
significantly (p -< 0.001) lower compared to intact tibiae
(fig. 5).

DISCUSSION
Strain gauge measurements:
The described method allows progressive evaluation of callus
stiffness in accelerated or delayed healing with external fixa-
tion. Under normal bone healing conditions its validity is
limited to approximately twelve days after a detectable stiff-
ness increase occurs. Changes of the connection between screws
and bone, especially pin loosening, extremly affect the result.
 439

force
[N]
400 INTACT TIBIA

four-point bending
! force

300

10mm

. 200

Tdeformation 100
5 WEEKS P.OP.

54 mm
o~~.--.--~~~--~
0.5 1.0 1.5 2.0 [mm]
deformation

Figure 4. Four-point bending arrangement. Load-deformation

curves of intact rabbit tibia and 5 weeks after osteotomy.

ultimate bending moment bending stiffness

7.91.5Nm 383 85N/mm

4.1 1.6Nm

~
17421N1mm

n-5 n-31
~n5 n31

5 weeks p.op. control 5 weeks p.op. control

Figure 5. Ultimate bending moment and bending stiffness of

explanted tibiae

Four-point bending:
Terjesen and Johnson (13) using fixators with different stiff-
ness characteristics in rabbits found ultimate bending moments
from 57% to 65% of intact tibiae six weeks after operation.
This corresponds to our findings of 53% after 5 weeks. On the
440

contrary stiffness values of the latter authors reached 100% -

120% after six weeks compared to 45% in our five week study.
Therefore it would be very interesting to detect stiffness
parameters up to 100% in vivo.

REFERENCES
1 Black J, Perdigon P, Brown N, Pollack SR: Stiffness and
strength of fracture callus. Clin Orthop 182 (1984) 278-288
2 Edholm P, Hammer R, Hammerby S, Lindholm B: The stability of
union in tibial shaft fractures: Its measurement by a non-
invasive method. Arch Orthop Trauma Surg 102 (1984) 242-247
3 Friedenberg ZB, Roberts PG, Didizian NH, Brighton CT: The
stimulation of fracture healing by direct current in the
rabbit fibula. J Bone Joint Surg 53-A (1971) 1400-1408
4 Hammer R, Hammerby S, Edholm P: X-ray diagnosis of fracture
healing: A comparative study of 127 human tibial shaft frac-
tures. Acta Orthop Scand 54 (1983) 104
5 Henry AN, Freemann MAR, Swanson SAY: Studies on the mechani-
cal properties of healing experimental fractures.
Proc R Soc Med 61 (1968) 902-906
6 Jernberger A: Measurement of stability of tibial fractures.
Acta Orthop Scand 41 (1970) Suppl 135
7 Jorgensen TE: Measurements of stability of crural fractures
treated with Hoffmann osteotaxis.
Acta Orthop Scand 43 (1972) 264-279
8 Matthews LS, Kaufer H, Sonstegard DA: Manual sensing of
fracture stability: A biomechanical study.
Acta Orthop Scand 45 (1974) 373-381
9 Nicholls PJ, Berg ED, Bliven FE, Kling JM: X-ray diagnosis
of healing fractures in rabbits.
Clin Orthop 142 (1979) 234-236
10 Panjabi MM, White III AA, Southwick WO: Temporal changes in
the physical properties of healing fractures in rabbits.
J Biomech 10 (1977) 689-699
11 Terjesen T, Benum P: In vitro effects of external fixation
on intact and osteotomized tibiae.
Acta Orthop Scand 54 (1983) 212-219
12 Terjesen T: Healing of rabbit tibial fractures using exter-
nal fixation. Acta Orthop Scand 55 (1984) 192-196
13 Terjesen T, Johnson E: Effects of fixation stiffness on
fracture healing. Acta Orthop Scand 57 (1986) 146-148
14 White III AA, Panjabi MM, Southwick WO: The four biomechani-
cal stages of fracture repair.
J Bone Joint Surg 59-A (1977) 182-192
15 Wolf JW, White III AA, Panjabi MM, Southwick WO: Comparison
of cyclic loading versus constant compression in the treat-
ment of long-bone fractures in rabbits.
J Bone Joint Surg 63-A (1981) 805-810
16 Wu J-J, Shyr HS, Chao EYS, Kelly PJ: Comparison of osteotomy
healing under external fixation devices with diferent
stiffness characteristics.
J Bone Joint Surg 66-A (1984) 1258-1264
 441

THE EFFECT OF DIFFERENT REGIMES OF AXIAL MICROMOVEMENT ON THE

HEALING OF EXPERIMENTAL TIBIAL FRACTURES
A.E. Goodship. D.J. Kelly. H~S. Rigby. P.E. Wa~kins and
J. Kenwright
Comparative Orthop. Res. Group. Dept. of Anatomy. Univ. of
Bristol. MA. U.S.A . and Nuffield Orthop. Centre. Oxford. U.K.

In both clinical and experimental situations i t has been well

established that bone as a tissue is highly sensitive to changes in
mechanical demand. The application of strain gauges to intact bones in
vivo has resulted in the characterisation of the mechanical variables that
provide an osteogenic stimulus.
The manner in which bone heals may also be influenced by the
prevailing mechanical environment, which will be determined by both the
technique and device used for fixation. Although healing patterns in bone
can be divided into two major categories, direct (osteonal without callus
formation) and indirect (with supporting callus), the rate of healing and
pattern of callus forniation may also be influenced by minor variations in
certain mechanical parameters during the healing process. In fact the
attainment of union will depend upon these mechanical factors being within
a certain range. Deviations from this range may result in delayed healing
or even non-union, with excessive movement leading to exuberant callus and
hypertrophic non-union and insufficient movement resulting in suppression
of callus and atrophic non-union.
Even within defined treatment methods such as external fixation there
are considerable differences in the imposed mechanical environment at the
fracture site resulting from frame type and configuration. These
mechanical differences may not provide an ideal environment for healing in
that too rigid a system will inhibit callus formation and delay union
whereas a more flexible system may induce a good osteogenic response and
encourage rapid restoration of mechanical integrity. A flexible system,
although providing the required stimulus at the fracture site, may also
increase the possibility of excessive pin/bone stresses and excessive
movement at the pin/skin interface leading to pin loosening and an
increased incidence of pin tract infection respectively.
An earlier study in sheep has indicated that a short period (1% of
eaCh day) of imposed axial micromovement superimposed upon an otherwise
rigid fixator system results in an enhanced rate of healing. Thus it is
Possible to effect improved healing in one of two ways, either by
selecting a device with exactly the right degree of flexibility to provide
a continuous mechanical stimulus at the fracture site, or to select a very
rigid device and excite the cell response by applying appropriate levels of
micromovement for a short period each day. By using the application of
micromovement on a rigid device it is possible to achieve the benefits of
both rigid and flexible systems.
The effects on healing patterns, of differing regimes of
micromovement applied to fixators with two different stiffness
configurations have been compared with the healing patterns of the two
configurations alone. The object of this study was to determine the
optimal levels of particular mechanical variables within the loading regime
442

and compare the influence of these short daily stimuli with healing under
conditions of different fixed rigidities.

MATERIALS AND METHODS

Fracture Model
In six groups of six sheep a transverse mid-diaphyseal tibial
osteotomy was created using a jigli saw: a 3mm gap was achieved at the
osteotomy site and care taken to preserve the periosteum. The osteotomy
was stabilised using an Oxford External Fixator system using three 6mm
Schanz bone screws in each fragment Fig. 1. The animals were allowed free
weight bearing immediately post-operatively.

Fixator Configurations
The mechanics of the external skeletal fixation was varied. In two
control groups without applied micromovement (A and B) the geometry of
fixation was varied so that in group A the fixation frame was set at a
distance of 35mm from the tibia and in group B at 25mm. The frame
Configurations were identical in all other respects.

Application of Controlled Micromovement

In two of the treatment groups the frame configuration was set with
35mm pin exposure and a short period (17 minutes) of micromovement was
applied daily. This was achieved by attaching a Dartec miniature
hydraulic actuator to the proximal turrets of the fixator which were fitted
with backplates capable of sliding along the fixator bar for the duration
of the loading period. After the application of micromovement the sliding
turrets were locked to maintain a rigidity comparable to that in the
appropriate control group. The loading regime was controlled USing a
B.B.C. microcomputer and was set to apply 510 cycles of movement at 0.5 Hz
and a rate of 40mm per second. This regime was applied to the osteotomy
five days a week from one week post-operatively, a healing period of
twelve weeks was allowed. In these two treatment groups a varying degree
of displacement was being investigated such that in one the displacement
was 0.5mm and the other 2mm. In a further two groups the frame was set at
25mm from the tibia as in control group B. The loading frequency, strain
rate and number of cycles were identical to the previous two groups but in
these groups a displacement of lmm was applied and the effect of maximum
force allowed was investigated at two levels, 200N and lOOON.

ASSESSMENT METHODS
Radiography
Attachment of the fixator to a jig enabled standard view radiographs
to be taken at intervals during the healing period. These were assessed
quantitatively in terms of the appearance of defined radiological features
by two independent observers in a double blind test.

Photon Absorption Densitometry

A quantitative assessment of bone mineral content was made in
transverse scans in the middle of the osteo.tomy gap, also within the
proximal and distal pone ends. A dual isotope system with 241 Am and 125 I
was used with the fixator and jig allowing sequential scanning of the same
locations at intervals during the healing period.

Fracture Stiffness
A strain gauged transducer was attached to the fixator bar to measure
 443

the deformation of the bar during walking. Concurrent recording of the

peak vertical ground reaction force allowed the calculation of a fracture
stiffness index at intervals throughout the healing period. At twelve
weeks post-operatively the animals were killed and both tibiae dissected
free of soft tissues. The tibial shafts were placed in a torsional test
machine for measurement of torsional stiffness and ultimate torsional
strength.

RESULTS
Radiography
A qualitative assessment showed an overall pattern of fracture healing
which was similar for all groups of treatment. External callus formed
during the first three weeks with bridging occurring between six and ten
weeks and remodelling changes being seen between ten and twelve weeks.
There were differences between the treatment groups, the most
significant being the time at which bridging occurred. The callus was
distinctly delayed in its bridging component in fractures in which there
was low post-mortem stiffness. This delay in bridging was most prominent
in the group subjected to 2.0mm of pulsed micromovement. Examination of
the radiographic parameters which had been quantified did not show any
significant differences between groups and their controls.

Dual Photon Absorption Densitometry

Fig. 2 shows increase in bone mineral content, as a function of time,
taken at the mid-point of the osteotomy gap. In the first four weeks
little change is observed in all groups. Between weeks four and ten there
is a sharp and linear rise with a fall, probably related to remodelling
changes, from week ten to twelve. This overall pattern was influenced by
the fixator stiffness and micromovement regime. A peak level was not
attained in the rigid control group, 2.0mm displacement group and the
group with lOOON applied force. There were differences in overall
mineralisation between the different groups which were statistically
significant. Control group A (less rigid) shows more gap mineralisation
than control group B, demonstrating that a more rigid frame inhibits
healing. The greatest beneficial effect on mineralisation appeared to be
produced by the application of small amounts of displacement (O.5mm) or
low levels of applied force (200N) , applied for very short periods each
day. The difference between this increase in mineralisation and that in
the control groups was noted as well as a statistically significant
difference between O.5mm and 2.0mm of pulsed micromovement.
Scans through the proximal fragment of bone 2.0mm from the bone end
were made. It can be seen that there is less inhibition of mineral
formation in all groups subjected to micromovement. The greatest response
was again seen in the group with O.5mm of applied micromovement, this
difference being significantly greater than both control groups with fixed
frame stiffnesses.

In Vivo Fracture Stiffness

Fig. 3 shows the overall in vivo stiffness measurements for each
group. The results correlate with the mineralisation levels with higher
levels of stiffness being reached within eight weeks post-operatively for
both the less rigid configuration control group and the group subjected to
applied micromovement with O.5mm maximum displacement. In terms of
applied force levels the group subjected to 200N showed a statistically
significant increase in stiffness compared with the corresponding control.
444

Only in the less rigid control and the group subjected to O.5mm
displacement was an early peak in stiffness attained.

Post-Mortem Torsional Stiffness

These values were expressed as a percentage of experimental to
contralateral limb. There were very marked variations, however, in the
groups subjected to different mechanical environments, (Fig. 4). It can
be seen that consistently higher levels of stiffness were seen in the group
subjected to a micromovement regime with O.5mm displacement. A greater
number of very low readings were seen in the less flexible control group
and the group subjected to a regime with 2.Omm displacement. In the rigid
control group half the animals showed values below 25% and could be
considered as delayed or non-unions at this stage. In contrast the
application of a short period of micromovement, with an applied force of
200N, to this inherent rigidity resulted in all stiffness values being
above 75%. Similar patterns were seen for the ultimate torsional strength.
DISCUSSION
The inherent stiffness of an external fixator can be altered by
change in frame configuration. In this study the inherent stiffness was
changed by an alteration in the offset distance of the fixator bar from
the bone, all other aspects remained constant. A significant difference
was seen using objective assessment techniques in the healing pattern of a
standard experimental osteotomy.
In addition it has been shown that a mechanical stimulus superimposed
upon the inherent frame stiffness for only a short period each day can also
significantly change the pattern of healing. Using objective assessment
methods for both bone formation and restoration of mechanical integrity the
healing pattern followed a non-linear course. There were three well
defined phases, an initial slow response to around three weeks followed by
a rapid increase in both bone mineral formation and mechanical integrity
after which an equilibrium was attained during which the callus underwent
remodelling changes. The rates of these stages were affected by the
various mechanical regimes. Thus it would appear that the natural process
of bone healing with external callus can be influenced by both the
consistent prevailing mechanical environment and also by short episodes of
superimposed cyclical loading. The levels of mechanical parameters within
such episodes can be beneficial or detrimental to the healing pattern.
It is conceivable that the optimal levels of the mechanical variables
may be related to the stage of the healing process, in this way the
equilibrium seen toward the end of the healing period might be further
perturbed by increasing the applied force once bridging and consolidation
of callus has occurred.
In the groups with a rigid frame configuration the imposition of
micromovement for a short period each day resulted in a more consistent
attainment of restoration in mechanical integrity. The relatively short
duration of the applied stimulus may indicate that the advantages of a more
flexible frame can be achieved without the associated complications of such
a device. In addition it is conceivable that fracture repair could be
stimulated, using applied micromovement, prior to normal weight-bearing in
the period that has been suggested as critical to progression of an
optimal healing pattern (McKibbin, 1978).
 445

REFERENCES

~1cKibbin B. (1978) The biology of fracture healing in long bones.

J. Bone Joint Surg., 60B, 150 - 162.

FIGURE 1.

1~t ---.........,O.5mm
1000
200lIl
/ : / f - - -.... Cootrol A
} 800
2mm

! 600

iJ 400

10 12

PIGURE 2.
446

25
DYNAMIC STIFFNESS TEST

0.5nm
20 200N

x
w
c 15
~
rn Control A
rn 211m
w
z
IL
IL 10
i=
rn

0L--r~r-~~--4--4--4--+--+--+--+-~
o 2 4 6 8 10 12
Weeks PoaH)perative

FIGURE 3.

PERCENTAGE T~ Sl1FAESS
200 <RKJiT/LEFT)

160

120

60

40

o
Control A 211m o.5nm Control B 200N

FIGURE 4.
 ~7

THE INFLUENCE OF INTERMITTENT CHANGE OF MECHANICAL ENVIRONMENT

UPON THE HEALING OF TIBIAL FRACTURES
J. Kenwright*. J.R. Richardson**. A.J. Spriggins**.
D.J. Kelly*** and A.E. Goodship***
Nuffield Orthop. Ctr.*. Oxford. U.K.; Oxford Orthop. Eng.
Ctr.**; Dept. Comparative Orthop. Res.***. Bristol. MA . ~.S.A.
Serious tibial fractures can cause prolonged morbidity
and their incidence remains high in road traffic accidents:
fracture healing times in excess of 20 weeks are commonplace
for open tibial fractures. (Rosenthal 1977).
When using external skeletal fixation for treatment,
inhibited healing of these high velocity fractures is seen in
nearly every instance. It is known that mechanical
environment affects fracture healing and perhaps this
environment could be made more favourable when using external
skeletal fixation.
In experimental fractures of the tibia fixed with
relatively rigid external fixation, it has been shown that
daily application of short periods of axial micromovement
applied early after osteotomy and continued during healing
will prevent the inhibited bone healing seen when using rigid
fixation (Goodship & Kenwright 1985).
The preliminary results are presented here from an
investigation into the influence of axially applied
micromovement on the healing of tibial diaphyseal fractures in
patients treated with external fixation.

METHOD
Patient Selection. Adults with serious tibial diaphyseal
fractures were treated using a unilateral external skeletal
fixation frame for well recognized indications for severe
injuries. Patients were allocated to one of two treatment
groups using either rigid fixation (Group I) or fixation with
applied axial micromovement (Group II). Allocation to
treatment group was controlled in the first one and a half
years and random in the next one and a half years.
Eighty-two fresh fractures have been treated and followed
through to healing, the frame having been applied within two
weeks of injury.
Fracture Severity was classified according to the degree
of comminution and soft tissue injury (Table 1).
Statistically there was no significant difference in the
distribution of injury grades between the two treatment groups
(Table 1).
Treatment Method. In Group I 50 patients were treated
using a unilateral frame designed specifically to give a high
degree of rigidity of fracture fixation (Evans et al 1979).
Mobilization was encouraged at an early stage and partial
weight bearing allowed within patient tolerance. Frames were
removed w~en clinical and radiological fracture union was
reached.
In Group II 32. fractures were treated using the same
conditions except that the frame allowed controlled ~xial
displacement of the fracture. Two sliding clamps we.re
attached to a spring and to the fixator column on one side of
the fracture (Figure 1). One week after application of the
frame a loading regime was applied for a short period daily
via a pneumatic pump. The regime was similar to that shown to
stimulate healing in experimental tibial fractures, i.e.
0.5-1.5 mm axial displacement applied at 0.5 Hz for 30 minutes
per day. The loading was supervised by a physiotherapist.
Pumping continued until the patient was able to reproduce this
axial excursion through self weight bearing, the spring then
being adjusted appropriately to allow the same displacement to
be maintained during weight bearing.

Figure 1

Figure 1 shows .the two sliding clamps attached to the spring

assembly.

Assessment of Fracture Healing was made every two weeks

by:
(i) Clinical and radiographic examinations:
(ii) Recording the time to free weight bearing from
injury:' I

(iii) Mechanical sltiffness at the fracture was measured

satisfactorily for 49 fractures treated
consecutively at two of the centres.
A strain gauge transducer was placed upon the fixator
column {Burny 1979: Evans et al 1985) and measurement made of
bending deformations of thecolUllln during standardised loading
 449

conditions. As the fracture heals, relatively more load is

taken through the fracture and frame deformation decreases, so
that indices of fracture stiffness can be calculated. The
frame configuration for individual fractures was simulated and
loaded in the laboratory and absolute levels of fracture
stiffness to bending calculated.
The fracture stiffness increased exponentially with time
the period measured. The logarithmic values of the
stiffnesses were plotted and the gradient expressed as the
time taken to reach a bending stiffness of 15 NM/degree. If
15 NM/degree was not reached, extrapolation was made to this
stiffness level so as to represent the gradient. This level
of stiffness was chosen as the definition of healing in that
it corresponded to clinically evident bone union; also in
three cases where the fixation had been removed at lower
levels of bending stiffness, refracture had occurred.

TABLE 1.
Mean times from injury to independent weight bearing

SEVERITY RIGID MICRO/M

n (weeks) n (weeks)

1 Cm 0/1 ST 0/1 14 (24.9) 5 (20.0)

2 Cm 0/1 ST 2/3 12 (26.9) 7 (25.5)
3 Cm 2/3 ST 0/1 9 (30.9) 4 (18.8)
4 Cm 2/3 ST 2/3 12 (32.2) 13 (25.2)
5 Severe bone loss 3 (105) 3 (30.2)

Overall mean healing times (30.7) (23.8) Student's

t-test
after log
transforma-
tion
Total number of patients 50 32 p = 0.02

Cm Comminution Grades 0-3

ST Soft Tissue Injury Grades 0-3
There was no statistical difference between the
distribution of injury severities for the patients in each
treatment group who had undergone strain gauge monitoring.
It has been shown that more than one loose screw will
lead to falsely high levels of bending stiffness being
recorded (Churches et al 1985): such patients were excluded.

RESULTS
Overall Clinical Results: the time from injury to free
weight bearing without splintage is shown in Table 1; the mean
time in weeks is longer for the overall group of fractures
treated by rigid fixation (Group I), this difference being
statistically significant (p = 0.02) after allowance had been
450

made for the abnormal distribution of the data. If those with

bone loss and bone graft were excluded the difference was
still statistically significant though less so (p =< 0.05).
A greater number of patients in the rigidly fixed group
needed secondary surgery in order to achieve sound union.
There was no difference between the two groups in other
complication rates, for infection of the fracture or screw
tract problems. Three refractures occurred in the rigidly
fixed Group I patients.
Radiographs: Subjective assessment often showed callus
ensheathing the bone adjacent to the fracture in the group
treated by micromovement (Group II), a feature not seen in the
rigidly fixed group. It was not possible to quantify callus.
Fracture Stiffness: The gradient of the increasing
fracture stiffness expressed as the time taken to reach
15 NM/degree is shown in Figure 2 for 49 patients in whom the
monitoring was satisfactory. Median time for Group I patients
to reach this level of stiffness was 24.8 weeks and for Group
II patients 17.6 weeks. This difference was statistically
significant (p = 0.0027).

DISCUSSION
There was no significant difference in fracture severity
between treatment groups in these monitored patients. The use
of modern external skeletal fixation frames allows control of
the mechanical environment at the fracture site and these
mechanics can be adjusted throughout healing. As the fixation
screws are placed well away from the fracture healing process,
this process can be studied without attempting to allow for
interference from other influences which could affect healing,
such as operation with implant insertion at the fracture site.
In the present study the influence of two different
mechanical environments has been investigated in patients with
serious tibial diaphyseal fractures.
The loading conditions were standardised as strictly as
was possible but identical mechanical environments would not
have been present throughout healing for individual fractures
in each treatment group because patients vary in their ability
and willingness to weight bear. Axial movement at the
fracture site would al~o have varied in individuals according
to the shape and pattern of the fracture so that the
percentage movement or strain acting upon the healing tissues
must have varied within fractures in each treatment group.
The time to free weight bearing showed a difference of
the medians between treatment groups which was statistically
significantly longer for those fractures not subjected to
pulsed micromovement. The time from injury to free weight
bearing is an indefinite measure of the end point of fracture
healing, though it is important clinically.
 451

Strain Gauge Measure of Bending Stiffness for

49 Fractures.
Log Stiffness Micro/m Rigid
179wks. 23-2wks. (medians)
1S NM/deg

p = 00027

O~--~--r---~--~--~--~--~--~
10 20 30 40
Time in weeks from injury

Figure 2

Strain gauge monitoring of the column has been shown

previously to give results which represent the healing of
fractures in mechanical terms when using external skeletal
fixation (Burny 1979). The differences in time to reach
levels of bending stiffness equivalent to clinical fracture
union were statistically significantly different for the two
treatment groups, there being longer times in the rigidly
fixed group. The median values of 17.9 weeks and 23.2 weeks
do not represent a major difference in fracture healing times
but the difference appears sufficient to conclude that change
of mechanical environment for even very short periods each day
will influence the fracture healing process.
The early weeks after injury have been described by
McKibbin (1978) as an important phase when the fracture
healing process is very susceptible to mechanical influences.
It is also a period when patients with conventional treatment
methods with casts or traction are very inactive.
The most favourable strain magnitUde and rate for the
different phases of fracture healing have not yet been
determined but those applied in this study did appear to
reduce inhibition of fracture healing without increasing
complication rates due to the applied movement.
452

REFERENCES
1. Burny FL: Strain gauge measurement of fracture healing.
In: Brooker AF, Edwards CC, eds. External Fixation.
Baltimore: Williams and Wilkins, 1979.
2. Churches AE, Tanner KE & Harris JD: The Oxford External
Fixator: Fixator stiffness and the effects of bone pin
loosening. Eng. in Med. 14 (1), 3-11, 1985.
3. Evans M, Gwillim J, HarriS-JD & Tanner KE: The monitoring
of fracture stiffness with external fixation. In:
Whittle M & Harris JD: Biomechanical measurement in
orthopaedic practise, 1st edn, 29-48: Clarendon Press,
Oxford, 1985.
4. Goodship AE & Kenwright J: The influence of induced
micromovement upon the healing of experimental tibial
fractures. J. Bone Jt. Surge 67B, 650-655, 1985.
5. McKibbin B: The biology of fracture healing in long bones.
J. Bone Jt. Surge 60B, 150-162, 1979.
6. Rosenthal RE, Macphail JA & Ortiz JE: Non-union in open
tibial fractures. J. Bone Jt. Surge 59A, 244-248, 1977.
 453

EXPERIMENTAL INVESTIGATION ABOUT THE STABILITY OF THE GREATER

TROCHANTER
F. LHer. E. Savvidis and U. Herrboldt
Abt. Orthop!die der Med. Fak. der RWTH Aachen. Aachen. Germany

In reV1Slon operations of an artificial replacement of the

hip joint, an osteotomy of the major trochanter can be indi-
cated. For the refixation of a trochantery medaillon, a num-
ber of osteosynthetical techniques is recommended:

Simple cerclages having a double current (Charnley, 1964),

8-figure wire cerclages (Charnley, 1970; MUller, 1970; Mittel-
meier, 1979; Amstutz, 1984) _ pulling screw osteosynthesis
(MUller, 1970), hook sole plated osteosynthesis (Schneider,
1982) .

Quantitative comparative data of the values of the arrivable

strength of the mentioned forms of osteosynthesis are still
lacking. Therefore we have undertaken model investigations
posing this question.

PROCEDURE:

Gabled roof-like ostetomies of the major trochanter through

the maintenance of the femoral neck were carried out on isola-
ted human cadaveric femora without pathological changes of
the bones (n = 18). Following this, six different forms of
osteosynthesis on two thighbones were prepared respectively
(fig. 1). After that the femora.were respectively fixed into
an apparatus clamped in a device E with the help of a system
of levers in the physiological direction (in an enclosed
angle between the axis of the femur and the direction of the
tensile force = 21). By this means the femora were drawn into
the refixed major trochanter. The required load was in this
way produced on a screwing device and measured by means of
a dynamometer KM (fig 2).

The changes in the osteotomical fissure as a function of the

fixed force were registered by means of a gauge and successive
photographical exposures.
The experiments were carried out on a straight line as well
as an angle of 45 to an antiflexed position of the hip joint.
The loadings were gradually increased until the commencement
of the case of damage.
454

RESULTS:

The force displacement curve for the tested techniques of

osteosynthesis are shown in f'igure 3 and the maximum values
of each of the forms of osteosynthesis in figure 4.
It appeared that cerclages with a single fold of a shoven wire
(a) through loops tolerate tips of loading up to 32 kp. At
higher loadings a complete separation of this osteosynthesis
was achieved by opening the folded wire of the cerclages on
the folded spot. Cerclages with twistings at the end (b)
reached loading values of 42 kp. The maximum accessible load-
ing here was limited by the position of the strong bending
of the wire. An evident improvement of the juncture between
both cerclages with an increase of the capacity of up to 64
kp is obtained by a special switch system (c). For two joint
screws the application of packing shins turned out to be so
important that without them (dl) only loads of up to 32 kp
are tolerated. By applying packing shins (d2) a loading of
up to 75 kp can be obtained.

The clamping (e) with values of up to 26 kp reached the least

strength of all, and in the case of failure the trochantery
cortical bone is completely untightened.

The combination of cerclages and screws produced a slightly

additional increase of the osteosynthetical capacity (without
figures). On the other hand the application of simple os teo-
synthetical plates gives no better tolerance of the capacity
than the simple screwing with packing shins (without figure).
An evident improvement of the values of strength can on the
other hand be obtained with a hooked sole plate (f). This
leads to a maximum tolerance of 184 kp in case of a normal
loading and 124 kp in case of a 45 anteflexion.
The analysis of the cases of damage shows a thorough tighten-
ing of the trochanter lockets of a continuous distal anchorage
in the case of screwing. In the case of plated osteosynthesis
there was a fracture of the trochantery ridge and in case of
hooked sole plate a longitudinal fracture of the trochanter.

DISCUSSION:

In conformity with earlier investigations and clinical ex-

perience the bending positions and closing mechanisms of the
threads of the cerclage proved also to be a limiting factor
of the strength.
The values of strength show that a burdening of the leg with
the weight of the body before commencement of the recovery
of the bone should not be allowed. The value of each of the
forms of osteosynthesis is however not only to be assessed
from our limited number of experiments on strengths, but also
from the special clinical situation, which also essentially
determines the application of the procedure.
 455

SUMMARY:

It was an experimental investigation carried out on 18 cadave-

ric femora on the question of strength of different, clini-
cally practised trochanteric osteosynthesis.
The maximum tolerable values of strength were determined and
the analysis of the causes of damage were carried out as well.

LI'EERATURE:

Amstutz, H. C. et al Results of Interlocking Wire Trochan-

teric Reattachement and Technique Re-
finements to Prevent Complications
Following Total Hip Arthroplasty.
Clin. Orthop. 183 (1984) 82

Charnley, J. Transplantation of the Greater Tro-

chanter in Arthroplasty of the Hip.
J. Bone Jt Surg. 46-B (1964) 191

Charnley, J. Total Hip Replacement by Low Friction

Arthroplasty.
Clin. Orthop. 72 (1970) 7

MUller, M. E. Total Hip Prosthesis

Clin. Orthop. 72 (1970) 46

Mittelmeier, H. Biomechanik der Drahtcerklage

Hanser, U. Z. Orthop. 117 (1979) 701

Schneider, R. Die Totalprothese der HUfte.

Aktuelle Probleme in Chirurgie und
Orthopadie, 24.
H. Huber Verlag, Bern, Stuttgart,
Wien, 1982
456

(d 1 and d2)

fig. 1: systematic representation of the investigated

osteosynthetical techniques

cerclage with a 'single folding (a), the twist-

ing of the end of the wire (b), the closing grip
(c), screwing without shins (d1), screwing with
shins (d2), clamping (e), hooked sole-plating
(f)

-- ------, r--------
iI I

force F : i
I

: I

~
evaluation

fig. 2: systematic representation of the apparatus

 457
mm

]i
U

o

I
t-

CII

,!;
U!
c
20

15

10
1t ~ ' 1
1 '
' 1
1'
, I
I '
o
1 '
0

,,,,
pp

0
01 I,
I,

~ '" "
:'"
9/ / '
:Vi 5 / '
0 ,0"
II

0
0 25 50 75 kp 100
tensile- force F

mm
20
~0
I
t-
o 15
CII

,!;
U!
10
5 .I'
'fi ,-
/

~ 5 "

0
0 25 50 75 100 kp 125
tensile-force F

mm
20
~U

if
I
t-
o 15
~
,S
en 10
,~
U
2
7Ji 5
r---::::~-=:::----jj5'
l5 "" "', . . - ...... - a !
#r;:"''''''
-,,:...
".
0
0 25 50 75 100 125 kp 150 175
tensile-force F

fig. 3: force-displacement diagram of different tech-

niques of osteosynthesis
458

max - Loadingforce in kp

a, cerclage with a single folding

b, cerclage with twisted end of wire

c, cerclage with closed grip I--_ _ _-----Y-v 64

dL screwing without pocking shins

d2, screwing with packing shins

e, clamping

f hooked sole-plating (normal positionl

f, hooked sole-plating (45 0 anteflexedl

fig. 4: maximum tensile-force in kp for different tech-

niques of osteosynthesis
 459

BENDING FATIGUE CHARACTERISTICS OF ARTIFICIAL KNEE LIGAMENTS

A.H. Soni*, M. Kutubuddin*, W.A. Grana** and M.R. Gudavalli*
Oklahoma State Univ.*, Stillwater, Oklahoma; University of
Oklahoma Hlth. Sci. Center**, Oklahoma City, Oklahoma, U.S.A.

Introduction:
Knee joint motion is governed by the tibia and femur
condyle surfaces, menisci and the ligaments. Instability of
the knee joint is a continuing problem to the Physician
treating knee injuries and is often accompanied by the
damage to the knee liagments. 50ni and Grana (2,9) have
studied the kinematics of the knee joint in relation to the
knee instability. The posterior and anterior cruciate
ligaments and the medial and lateral collateral ligaments
are often susceptible to damage during sports activities
such as football, basketball etc. as well as during athletic
activities. The surgical replacement of the damaged or torn
knee ligaments with artifi~ial prosthetic ligaments has
become standard practice. In order to develop and improve
the performance of the prosthetic ligaments, there exists a
need to determine the mechanical properties of the
ligaments. This paper describes the development of a
fatigue testing machine for testing prosthetic ligaments
under bending fatigue.
During the last twenty five years, mechanical
properties of passive soft tissues have been studied by many
investigators. 5eering (7) has given an extensive survey
of literature. Kennedy et al. (4), tested the cruciate
ligaments and the collateral ligaments of the human knee
joints in-vitro to determine yield point, ultimate failure
stress and disruption caused by failure loading. Noyes et
al. (5) recorded energy absorption in the anterior cruciate
ligament of rhesus monkeys as a function of strain rate.
Noyes and Grood (6) reported elastic modulus, failure
stress, and strain energy to failure for human anterior
cruciate ligaments were all age dependent.
460

Few researchers have reported on the cyclic loading of

ligaments. Grana and Soni (1,8) have examined the influence
of cyclic torsional load on the knee ligaments. Weisman et
al. (10) reported on the effect of cyclic loading on rat
medial collateral ligaments. They observed two phenomenons.
One of these describes the softening of the ligament which
is expressed as the ratio of percentage of change in length
of ligament from the last cycle peak to the first cycle peak
load, presented as a percent. The second phenomenon is a
reducti on in strength of the cycl ed 1 i gament, expressed as
percentage change in ultimate strength of the cycled
1 i ga me n t.
Grood and Noyes (3) studied a polyethylene implant for
tensile strength, residual elongation, creep and bending
fatigue. The objective of this paper is to develop a test
apparatus to test 1 i gaments under bendi ng fati gue. An
artificial man-made ligament which is used for prosthetic
replacement of knee joint cruciate ligaments is studied
under bending fatigue testing.
Fatigue Testing Apparatus:
The machine is designed primarily for the testing of
any flexible material having a small cross-section for
fatigue testing under a certain range of bending stresses.
The specimen could be pre-loaded in tension and torsion.
The machine with its combined tensile, torsional, and
bending stress attachment as shown in Fig. 1 is a compact
bench mounted unit, consisting of a rigid bolted frame,
horizontally adjustable vise, variable and provision for
applying pre-tensile and pre-torsional load to the specimen.
The unit occupies a space of 30 inches long, 24 inches wid~,
and 16, inches high.
The dri vi ng force of the machine is transmitted by a
variable speed motor through a pulley and belt to the main
shaft. Force from the main shaft is then transmitted
through a four-link mechanism to the specimen. The variable
throw crank itself, at the end of the main drive shaft is
acting as the input (driving) link of the four-link
mechanism. The variable throw crank adjusts the amount of
deflection (;. e. the crank or input link length) and
transmits the force through a rigid coupler link to the
output (driven) link. The variable throw crank has a range
of 0 to 1.5 inches and is adjusted by loosening the four
socket head screws above the eccentric. The coupler link
(connecting rod), which rides on a precision bearing
transmits the eccentric action of the crank to the output
link. The output link is extended beyond the pivot joint in
a cantilever fashion. The oscillating motion of this end of
the output link is then used to apply cyclic bending load to
the specimen. The effective length of the output link, from
the coupler link to the pivot joint, can be adjusted by
sliding the pivot points along the longitudinal slots
provided in both arms of the output link.
Tensile preload is applied by putting dead weights on a
weight pan, suspended on a steel rope through a pulley fixed
14 .'... ~

1. VARIABLE THROW CRANK 6. SUPPORTING FRAME 12. CLAMPING BASE

(ECCENTRIC) 7. ADJUSTABLE VISE 13. MACHINE BASE
2. CRANK HEAD ASSEMBLY 8. TENSION PRELOAD ATTACHMENT 14. LEVER ARM
3. COUPLER LI NK 9. SPRING LOADED PLUNGER ASSEMBLY 15. CUT-OFF SWITCH
4. DRIVEN LINK 10. COUPLER FOR TORQUE TRANSDUCER 16. CYCLE COUNTER
5. GRIPPER 11. TORQUE TRANSDUCER 17. SPEC IMEN

FIGURE 1. DETAILS OF FATIGUE TESTING MACHINE

~
462

to a bracket. The rope in turn is attached to the right

hand spindle. The spindle which is fixed with a specimen
holding chuck (gripper) is adjustable to accommodate shorter
or longer specimen sizes. Torsional preload to the specimen
is applied by means of a lever attached at other end of the
specimen holding chuck (gripper). A counter which is driven
by a flexible wire attached to the main shaft of the machine
keeps track of the number of cycles the specimen is
subjected to bending load. A cut off switch is provided to
shut off the entire machine if and when the specimen fails.
LIGAMENT TESTING:
A prosthetic man-made ligament with a length of four
inches was tested under bending fatigue. In order to prevent
p 0 s sib 1 e s 1 i PP age 0 f the s pe c i me n i nth e grip per s due to
tensile load and also to prevent failure at the ends of the
grippers due to twisting effect of the torsional load, the
ends of the specimen were tightly reinforced by wounding
good quality cotton threads. Diameters at either ends of the
specimen is increased with enough number of turns of cotton
threads so that the ends of the specimen will get locked
into the inner ends of grippers.
Before mounting the specimen, the machine was set for
30x bending angle at which the specimen would be tested.
The bending arm had its total 60x angle of oscillation and
during each revolution of the driving shaft, the specimen
would undergo a complete reversal of bending load from
maximum tension to maximum compression. The specimen was
then mounted properly on the machine and pre-loaded in
tension and torsion.
A tensile preload of 100 lbs. was applied to the
specimen by putting equal amounts of dead weights on the
weight pan and then locking the device at this load. A pre-
torsional load of 3 in-lbs. was then applied to the specimen
through the torque transducer. The machine was then set
ready for cyclic testing in bending with the above setting
of tensile and torsional preload.
The cycle counter was set to its zero reading and the
machine was turned on to run at a speed of two complete
revolutions per second. Each complete revolution thus
imposed two load-unload bending cycles of opposite direction
on the specimen. The experiment was carried out in room
temperature. After 200,000 stress reversals, the test was
stopped since no sign of failure was observed in the
specimen. The specimen was then put to a tensile test until
failure in the Instron Testing machine. Tensile test is
carried on a fresh specimen and the results of these
tensile tests are shown in Figure 2. A strain rate of2
cm/min was maintained during the tensile tests for both the
fresh and fatigued specimen.
Results and Discussion:
As already mentioned, no sign of failure could be seen
 463

visually in the fatigue test after 200,000 of load - unload

bending cycles of opposite direction on the specimen.
However, the tensile test on the fatigued implant showed a
marked reduction in its ultimate strength. As shown in Fig.
2, the ultimate tensile load for the fatigued specimen is
635 1 bs and that for the fresh speci men is 731 1 bs. Thi s
amounts to a reduction of 13 % of the original strength of
the implant. As could be seen, the ultimate failure points
are well defined in both the fresh and fatigued implant.
However, the elastic range of the curves do not show a
perfect linear relationship between load and strain. As
evidenced for the load - elongation curves and also from the
physical appearance of the specimens after the completion of
the tensi 1 e tests, the fati gued i mpl ant shows a sudden and
sharp decrease in load compared to the fresh implant after
the final rupture of the specimens. The fatigued implant
showed a clean breaking apart sign at the point where
fatigue stress in bending was applied. As compared to this,
the fresh implant showed a gross disruption of the fibers at
certain point within the effective length of the specimen.
At the point of ultimate failure, the fatigued implant had a
total elongation of 11 % as compared to 13 % for the fresh
implant. Structurally, after the final rupture, the fibre
bundles of the fatigued implant disintegrated into its
consti tuents of a thi n fl uffy and wavy materi al interwoven
together into a fine mesh. However, some fibers in the
fresh implant were still intact after the final rupture of
the specimen. The gross disruption of the fibers showed an
irregular pattern of failure around the neck of the specimen
where failure occurred.

References:
1. Grana, W. A., Soni, A. H., and Streit, D. " The
effect of Torsional Load on Axial Strain of Knee-
Ligaments", Proceedings of the 27th Annual ORS Meeting, Las
Vegas, Nevada, February 24-26, 1981.

2. Grana, W. A., and Soni, A. H., "In-Vitro experimental

Determination of Rotary Instability of Knee", Transactions
of ORS,San Fransisco, 1979.

3. Grood, E. S., and Noyes, F. R. "Cruciate Ligament

Prosthesis: Strength, Creep and Fatigue Properties."
Journal of Bone and Joint Surgery, Vol. 58-A, 1976, pp.
1083-1088.

4 . Ken n e dy, J. C., M. D., Haw kin s, R. J., M. D., Will is, R.
B., M. D., and Danvlchuk, K. D. "Tensile Studies of Human
Knee Ligaments." Journal of Bone and Joint Surgery, Vol. 58-
A, 1976, pp. 350-355.

5. Noyes, F. R., M. D., James, L. D., and Peter J. To

"Biomechanics of Anterior Cruciate Ligament Failure."
Journal of Bone and Joint Surgery, Vol. 58-A, 1976, pp.
464

1074-1088.

6. Noyes, F. R., and Grood, E. "The Strength of the

Anterior Cruciate Ligament in Humans and Rhesus
Monkeys." Journal of Bone and Joint Surgery, Vol. 58-a,
1976, pp. 1074-1088.

7. Seering, W. P. "Determination of Mechanical Properties

of the Ligaments of the Knee." PhD dissertation, Stanford
University, Stanford, Calif., 1978.

8. Soni, A. H., Grana, W. A., Streit, D., Jones, D. and

Maiya, M. S., "Ligamentous Damage Due to Torsional Load",
Proceedings of the 7th OSU Applied Mechanisms Conference,
Kansas City, MO, 1981.
9. Soni, A. H., Grana, W. A. and Maiya, M. S. "Application
of the Mobility Principles in Kinematics to Develop
Engineering Criteria of Knee Instability", Proceedings of
the ASME Winter Annual Meeting, Washington D.C, 1981.

10. Weisman, G., Malcom, H. P., and Robert, J. J. "On the

mechanical Properties of Ligaments under Cyclic
Loading." Proceedings of the Seventh New England
(Northeast) Bioengineering Conference, March 1979, pp. 395-
403.

-
-....
7 11

L
o 588
A
D

I-
L
S
281

.ea
e
a 2 3
ELONGATION CI'I'I)
FIGURE 2. LOAD-ELONGATION CURVES
 465

BIOMECHANICS OF THE ACROMIOCLAVICULAR STABILIZATION

H. Kiefer. L. Claes and J. Holzwarth
Laboratory for Exper. Traumatologi. University of Ulm. Ulm.
Germany

1. SUMMARY
A biomechanical in vitro test was performed to determine
the stabilizing effect of various implants for the surgical
treatment of the torn acromioclavicular joint. In a specially
designed testing device, plastic and cadaver specimens of the
shoulder girdle were stressed in various ways. Different
dislocations between the acromion and the clavicle and
between the clavicle and the coracoid as well as the
stiffness of the implants were determined. A convenient
stabilizing effect combined with less rigid fixation to
secure the newly treated AC joint was provided by K-wire
fixation with cerclage. Interfering shear and bending
stresses can be avoided with this method while the other
implants showed various disadvantages. From a biomechanical
point of view, a carbon fibre ligament replacement provides
sufficient stability in eases of chronic acromio-clavicular
separation.

2. INTRODUCTION
The separation of the acromioclavicular joint, type Tossy
III, is generally treated by surgery. The torn ligaments and
the capsule are sutured and the joint is temporariliy
immobilized by an implant to secure the healing ligaments
(1,2,4,5,7,8). For this purpose, various implants and grafts
have been used. In this investigation, the stabilizing and
therefore suture securing effect of six commercially
available implants and of one alloplastic ligament
replacement, has been studied in vitro.
The acromioclavicular (AC) joint is normally stressed in
two ways. Firstly, in opposing the tensile force of the
trapezius muscle at the lateral clavicle, the weight of the
arm causes shear stress in the AC joint. Rupture of the joint
leads to the appearance of a "step" between the acromion and
the clavicle. Secondly, when abduction and flexion of the arm
are performed, movement of the scapula causes relative
motions between the scapula and the clavicle to occur (6,9).
The angle between the clavicle and the scapular spine
increases with progressive flexion of the arm. Additionally,
posterior rotation of the clavicle up to 40 degrees has been
measured (6,9). These relative motions are of a similar
magnitude in flexion and abduction of the arm (6).
The kinematics of the AC joint are well known, but a
review of the literature failed to find information
concerning the magnitude of the forces acting through the
normal acromioclavicular joint. This study aimed to
quantitatively measure the effect of various implants to the
sutured ligaments from simulated joint forces.

3. MATERIAL AND METHODS

A special device has been constructed which could be
adapted to the various anatomical features of the AC joint.
The scapula-clavicle specimens were clamped in such a manner
to allow the scapula to be fixed and the forces were
introduced via the clavicle. Ii. cardan joint at the sternal
end of the clavicle allowed its free rotation (Fig. 2). Seven
standardized plastic models without ligaments as well as 12
fresh human cadaver specimens from the left shoulder were
used for this study. The implants tested are shown in Figure
1. All of the implants were mounted to each of the plastic
models and onto 6 cadaver specimens.

Fig.

Bosworth
srew
2 StreB
plate
3 Balser
plate
4. Wolter
plate
5 Rahmanzadeh
plate
6 Cerclage +
k-wires
7 Carbon
fibre lig.

The cadaveriv specimens were

kept moist with Ringer solution
during investigation period.
The clamped specimens were
mounted in a materials testing
machine (Fig. 3) and could be
tested in 3 different ways:

Fig. 2

Mounting support
Fe cranial force,
Fe = elevation force
:HI' = rot. moment,
E1, E2 = strain
11 - 12 = AC step
 467

Firstl~. the claviole ~as lifted

from a standard position at the
lateral end of tbe bone ,PC = 100
N) to simulate tbe weight of the
arm. Secondly, elevation of the
sternal clavicle ~as performed (FE
= 50 N) while the vertical guidance
was loosened. Thirdly. rotation of
the clavicle was carried out
(rotational moment Ot .. 100 Ncm).
Rotation and elevation of the
clavicle occur naturally during
abduction of the arm (6,9).

Fig. 3

The following measurements were

made:
1. the cranio-caudal dislocation
of the AC joint was obtained
and is referred to here as the
acromioclavicular step lAC
step). This was calculated
from the difference of the
signals of two inductive
displacement pick ups which
were placed on defined
locations on both sides of the
joint (Fig. 4).
2. the increase of the distance
between coracoid process and
clavicle (referred to as the Fig.
CC step) was measured.
3. the stiffness of the implants against the rotation moment
of the clavicle was determined.
4. the increasing strain in the two parts of the coraco-
clavicular (Cel ligament was measured using specially
developed omega shaped strain gauges which were sutured to
the ligaments.
5. the length of the CC ligaments was determined.
Measurements of the strain were carried out in the intact
coracoclavicular ligaments specimens and then repeated
following the attachement of the various implants. The
ligaments were severed and the Ie step was measured. The
ligaments were sutured with PDS sutures and the
468

measurements repeated. Strain and stress curves were plotted

until failure of the sutures occured. Carbon fibre ligaments
were then used to replace the cut ligaments and the
measurements were repeated. The plastic models were used to
obtain measurements of the AC step, CC step and the stiffness
of the joint following the application of the implants.

4. RESULTS
The mean length of the trapezoid ligament was 16.2 mm, and
the average length of the conoid ligament was 14.7 mm. The
results of the AC step are shown in Fig. 6. The values for
the strain in the conoid ligament due to for elevation of the
medial clavicle are shown in Fig. 7 (mean SEM). The
rotational stiffness values of the implants are summarized in
Fig. 8.
Fig. 5 Fig. 6
AC step CC step

mm mm

20

Fig. 7 Fig. 8
strain in conoid rot. stiffness
lig.

Hem
(~) Grd
3
3

2
2
:::;
'~"'

2,0
+
 %9

The mean ultimate tensile force for the cut and resutured
ligaments and capsules while performing elevation of the
medial clavicle was 48.6N ~ 5.5N (SEM). At the time of
rupture, an increase of the coracoclavicular distance of
8.8 mm ~ 0.8 mm had occured.

5. DISCUSSION
Plastic models were used for the first part of the
experiment because of their independence from soft tissue
influences. The standardized size of the spezimens better
enabled the real effect of the various implants to be
studied. Por comparison the same measurements were repeated
using cadaver specimens. Additionally, the strain in the
coracoclavicular ligaments could be measured to obtain
further information about the stress in the joints.
Lifting cranially at the lateral clavicle for measurements
of the AC step led to a shear stress in the joints and
implants. All of the implants were able to prevent the joints
from a major dislocation in this test. The step for the
Bosworth-screw in the plastic models was a lot higher than in
the cadaver specimens due to a softer nature of the plastic
coracoid process. The difference in the results for the
Balser plate between plastic and natural joints must be
explained by the frequent "slipping off" of the plate hook
below the human acromion.
Pixation of the acromioclavicular joint with unhindered
rotation by the Balser and Rahmanzadeh plates was thought to
be their advantage. Measurements of the rotational stiffness
of the implants revealed almost zero for these two plates
while the 4 other implants (no test for the carbon fiber
ligament was performed) showed a uniform stiffness of about
4 Ncm per grade (Pig. 8).
Low rotational stiffness of the protecting implant might
be dangerous for the healing coracoclavicular ligaments when
abduction for more than 90 is allowed and thus should be
avoided.
Elevation of the clavicle at the medial end led to a major
increase of the coracoclavicular distance using the Balser
plate or the Wolter plate while the Bosworth screw provided a
stable fixation. Elevation of th~ clavicle occurs during arm
abduction and causes bending stress for the implant at the
position of the AC joint. Therefore material properties, form
and dimensions of the implants are most important for this
position at the joint. The resistance to that bending by the
soft Streli plate and the ball jointed Rahmanzadeh plate is
very low, which led to high strains in the conoid ligaments.
Increasing the coracoclavicular distance more than 8 mm
leads to failure of the ligament sutures (Pig. 8).
Consequently, implants which allow dislocations higher than
8 mm cannot protect repaired ligaments from redislocation
during the healing period when the same stiffness for intact
and sutured ligaments is assumed.
Stability of chronically dislocated AC joints can be
safely provided by carbon fibre replacements with respect to
the acromioclavicular and the coracoclavicular steps.
470

Due to the weak material

and the huge dimension of
the Streli plate, the
missing rotational
stability of the Balser and
Rahmanzadeh plates, these
implants cannot be
recommended for temporary
stabilisation. The Bosworth
screw causes a rigid
fixation which leads to
failure of the screw due to
fatigue of the material.
With respect to the
investigated biomechanical
properties, k-wires in
combination with a cerclage Fig. 9: Cerclage with k-wires
(Fig. 9) and maybe the
Wolter plate might be the best methods for temporary
stabilisation of surgically treated acromioclavicular
separation.

REFERENCES
1. Bosworth BM: Acromioclavicular Separation. New Method of
Repai r. Surg. Gyn. Obstet. 73: 866-871 (1941)
2. Burri C: Technik des alloplastischen Bandersatzes mit
Kohlenstoffasern am Schultergiirtel. In: Burri C, Claes L,
Helbing G(ed) Bandersatz mit Kohlenstoffasern. Hefte zur
Unfallheilkunde 172, Springer, Berlin Heidelberg New York,
pp 137-148 (1985)
3. Claes L, Mutschler W: Elektrische Messungen von Dehnungen
und Kraften an den Kollateralbandern des menschlichen
Knies. In: Jager M, Hackenbroch MH, Retior HJ
Kapselbandlasionen des Kniegelenkes. Thieme, Stuttgart/New
York, pp 34-49 (1981)
4. Helwing E, otten G: Behandlung der acromioclavicularen
Luxation. Chir. Praxis 24: 275-277 (1978)
5. Holz U, Weller S: Luxationen im acromioclavicularen
Gelenk. Hefte zur Unfallheilkunde 160, Springer, Berlin
Heidelberg New York, pp 222-229 (1982)
6. Inman VT, Saunders M, Abbott LC: Observations on the
Function of the shoulder joint. J. of Bone and Joint Surg.
26: 1-30 (1944)
7. Jager M, Wirth CJ: Kapselbandlasionen, Thieme,
Stuttgart/New York (1978)
8. Tiedke R, Rahmanzadeh R, Faensen M: Die Entwicklung eines
neuen Implantates zur Behandlung der
Schultergelenkssprengung. Operationsverfahren und
klinische Erfahrungen. Hefte zur Unfallheilkunde, 165,
Springer, Berlin Heidelberg New York, 270-273 (1983)
9. Wallace WA, Johnson F: In: Bayley JI, Kessel L( ed)
Shoulder Surgery, Springer, Berlin Heidelberg New York, pp
179-182 (1982)
ORTHOPAEDIC BIOMECHANICS: SPINE
 473

DETERMINATION OF SPINAL STRESS BY BIOMECHANICAL MODEL CALCU-

LATIONS AND COMPARISON WITH SPINAL MECHANICAL STRENGTH
M. Jager und A. Luttmann
Institut fur Arbeitsphysiologie, Universitat Dortmund,
Dortmund, Germany

1. INTRODUCTION
Epidemiological studies have shown that diseases in the musculo-skeletal
system occur more frequently than other kinds of diseases (HETTINGER,
1985). Furthermore, long-term employment in materials handling jobs causes
an increase in orthopedic diseases, particularly in those of the spine
(LUTTMANN et al., 1985). Heavy work results not only in more discomfort
than light work, but may also lead to more irreversible damage in the
vertebral column. Consequently, it is of interest to occupational health to
determine the spinal stress during materials handling tasks. Most spinal
disorders are located in the lumbar region, the smallest number, by con-
trast, being found in the thoracic segment.
The measurement of mechanical parameters like pressure or force is possible
only under clinical conditions (NACHEMSON, 1963; ANDERSSON et al., 1974).
The method employed cannot be applied in routine work-physiology studies.
The attempt has therefore been made in recent years to calculate spinal
stress considering anthropometry, the body weight's distribution and the
kinematics of the segments during the handling tasks under examination.
Most of the biomechanical models provided in literature are two-dimensional
in the sagittal plane or static (for example: AYOUB and EL-BASSOUSSI, 1978;
CHAFFIN and ANDERSSON, 1984; JAGER et al., 1984; LESKINEN, 1985). In this
paper, a three-dimensional dynamic biomechanical model of the human body
composed of 19 segments is presented. The segments are linked by a total of
18 joints; 5 of them are positioned in the lumbar spine, since it is mainly
this part of the vertebral column which renders the flexion and extension
of the trunk possible (PANJABI and WHITE, 1980; LANZ-WACHSMUTH, 1982). The
relevant anthropometric data, body weight etc. were taken from literature
(LANIER, 1939; DEMPSTER, 1955; DRILLIS and CONTINI, 1966; ORNE and LIU,
1971; for further information see JAGER, 1986). The model includes both
static and dynamic influences of the limbs and manipulated loads. In
addition, the influence of abdominal pressure was calculated with the help
of further measurements and biomechanical
considerations by MORRIS et al. (1961) and
CHAFFIN (1969).

2. BIOMECBANICAL MODEL
In fig.1, the kinematics of the human body
are illustrated as assumed in the biomechani-
- =,'"
0p.
k
k-I

i.'
~
L (Wi X P;1'\.11
k-1
+

cal model. All body segments can be moved in +

,
L - - X P;P;.1
Wi X (Wi ---I
i:l

Fig. 1: Human biomechanical model; accelera- 0p, + (w, x P,Ff1 +

tions determined by superposition of rota-
tions of several body segments + W; x (iil; x ~I
( 0 joint; centre of gravity)
474

the natural way. During manipulation tasks, however, one foot usually
remains constant. (Should this not be the case, the kinematic description
must be modified.) The acceleration of the heel (PI) was set at zero. The
time course for the acceleration of the next upper Joint (ankle) can be
calculated from the rotation of the foot with the heel taken as the axis.
The ankle's acceleration is:
a;2 8p
1
+ ( ~ x 2) + W; x ( urI x 2 )
In this equation, ii3' represents the angular velocity, i3 the angular acceler-
ation, and P I I 2 the vector between the points PI and P. By analogy, the'
time course for the acceleration of the knee results lrom the superposi-

-
tion, firstly, of the acceleration of the ankle and, secondly, of the
rotation of the shank:
-ap
3
ap
2
+
Generalized for any joint Pk ' it is:
+

~l .:. _k~l _ _ _
ap L- (Wi x Pi P i+l) + L- wi x ( wi x PiP i +l
k i~l i-I
Similar to the calculation of a joint's acceleration, the acceleration of a
body segment's centr~ of gravity Sk can also be derived:
a;k a; k
+ (W; x Yk) + w: x ( ;; x k
The time courses for the angular velocity and langUlar'l velocity
acceleration of the body segments result from the
actions under examination. This paper's computa-
tions are based on the condition that all limbs
remain constant when starting and finishing the
manipulation task (fig. 2). At the beginning, the
angular velocity is zero, increasing during the
acceleration period. After an initial increase in
,angular velocity, it remains constant for a given
,period and decreases during retardation until the
'movement is finished. The time course for angular
acceleration is yielded from velocity by deriva-
tion and that for the traversed angle by integra-

-
tion. The angle which is totally traversed is
characterized by the initial and final position
of the segment considered in relation to an iner-
tial co-ordinate system. Since usual movements time
are not jerky, the time course for the angular Fig. 2: Assumed time
velocity was adapted to the subordinate condi- courses of moved limbs
tions mentioned above by using a square sinus
curve. Transition between the periods of in-
creasing, constant and decreasing velocity is
therefore smooth.
Fig. 3 shows a free body diagram of an arbi-
trary body segment "i" between the neighbouring
joints Pi and Pi~l. At these points, t~rques Ai.
and forces from tn~ upper segments of the human I~~__~
body chain are transferred to the next lower
one as a reactive force t or reactive torque it.

Fig. 3: "Free body diagram" of an arbitrary

body segment "i"; equations of equilibrium of
(eSI ' W';).
= it. + (-M;':,) +
forces and torques (S.~X~). SI~.'X(-~,)
 475

Both torques and forces must be in equilibrium, especially at the joints.

-The static properties of a segment are represented by its weight ~, the
dynamic ones characterized by a matrix of the moment of inertia 9. The
shape of the limbs has been approximated to be conical or cylindrica~.
On the upper border of the uppermost segment n that is for manipulation
~ks wh~e the load is held in the hands , reactive force and torque,
Rnt1 and Mn+1 respectively, are zero. That results, after some intermediate
ca culations, in a reactive force at the lower border of a limb k:
n
L
i=k
( m ; ;
~ i
The corresponding reactive torque is:
n n n
i\ L
i=k
(~s. ~ ). +
~
L (
i=k
Q'i x mi';;.
1
L
i=k
i x G7
This torque is generated by the well-timed co-ordination of the muscles ex-
tending over the joint considered. In the lumbar region, the muscular
system was simulated in this paper by the following 8 back and abdominal
muscles: erector spinae, rectus abdominis, obliquus externus and internus
abdominis; each on both the right and left side of the body. The muscles'
positions in the human body are described in anatomical textbooks (for
example: GRAY, 1973). Since these muscles may produce forces at the lumbar
spine (for the muscles' effect in the sagittal, frontal and transversal
plane, see JAGER and LUTTMANN, 1985), they must be added to the reactive
forces calculated above. Furthermore, the abdominal pressure represents a
certain help in stabilizing the trunk. This influence had been quantified
in principle by CHAFFIN (1969), but had to be adapted to the special
conditions obtaining in the model in this paper (for further information
see JAGER, 1986).

3. EXAMPLES OF BIOMECHANICAL MODEL CALCULATIONS

The three-dimensional biomechanical model of the human body derived above
allows the determination of the torques and forces acting at the lower
spine. In the following, the results of calculations of lumbar stress dur-
ing two typical materials handling tasks will be introduced: 1) Bi-manual
lifting in the median sagittal plane is very common whenever heavy loads
are manipulated. 2) The handling of light goods in a sitting posture is
usually performed by a single-arm movement. In
this case, a three-dimensional analysis is compressive force on L5-S1
therefore necessary. In fig. 4, the compressive in kN
force on the lumbosacral disc (L5-S1) is shown
for several lifting techniques. In all cases,
16
it was assumed that the load has a mass of 20
kg and that the movements are finished after -5 load 20 kg
1.5 s with the same time for acceleration and
retardation (0.75 s). During the first period, 4
all curves increase due to acceleration of the

r
body segments and the load to be lifted. The
energy of motion may then be utilized resulting
in a decrease in force values. At the end of 2
the movement lumbar stress depends especially ~
on arm postures which were assumed in the
calculations in fig. 4 to be different: 1) arms
1
--t
O+-----r---~-----r
o
. .
Fig. 4: Calculated time courses for lumbar 0.5 1.0 1.5
stress during lifting in various techniques time In S
476

completely hanging downwards, 2) elbows bent with only the forearms

straightened forward and 3) arms completely straightened forward. The dis-
tance of the centre of gravity of the manipulated load from the spine has a
'strong effect on torque, back muscular force and, consequently, on the com-
pressive force on the lumbo-sacral disc (shown in fig. 4). If, furthermore,
the trunk is initially inclined and is increasingly erected during the
movement (back lift), the force values are always higher than in the case
of an upright trunk during lifting (leg lift). On the whole, the time
courses for the 1umbo-sacral compressive force reveal a large dependence on
the kind of lifting technique. The maximum force on L5-S1 disc may take ap-
proximately a quadruple value in the case of the most disadvantageous pos-
ture with an inclined trunk and the arms straightened forward in comparison
with the most advantageous posture with an upright trunk and hanging arms.
In the case of persons in sitting postures, the) compressive force on LS-Sl
manipulated loads are commonly not as heavy as in kN
in the example studied in the foregoing. Fig. 5
shows the calculated lumbar stress during move-
ments of only one arm. It was supposed that the 1.S
person grasps a load (0 - 5 kg) with the arm
straightened forward and brings it to a position
near the body, where, for instance, it can be 1.0
handled bi-manually. The following assumptions 5 kg
4 kg
were made regarding arm movement: the upper arm 3 kg
rotates in the sagittal plane of the shoulder, 2 kg
whereas the forearm and hand are turned in a ns 1 kg
o kg
transversal plane parallel to the surface of the
desk. The superposition of rotations in differ-
ent planes results in a relatively constant
time course for the L5-S1 compressive force with o+-~--~--~~--~-
only a small maximum in the case of an arm o 0.2 0.4 0.6 0.8 1.0
movement without any load (0 kg). Heavier loads, time in 5-
however, lead to force values at a higher mean Fig. 5: Calculated time
level with the maximum more highly developed. courses for lumbar
This is mainly due to an increasing influence of stress for seated lift-
the dynamic components of the load's movement. ing of various loads

4. COMPARISON OF FORCE VALUES WITH SPINAL MECHANICAL STRENGTH

To estimate the human health risk, stress values should be compared with
objective criteria. However, the mechanical strength of the spine cannot be
determined in vivo, but only in vitro. Some investigators studied the
compressive strength of post-mortem material which differed, for instance,
in preparation (inserted in metal, synthetic resin, gypsum etc.). In other
studies, spinal segments were compared with or without the posterior
components. Furthermore, the dissected material was taken from persons who
varied in age, sex or profession. The results of an analysis of literature
on morphological experiments are shown in fig. 6. This diagram shows the
mean value and the standard deviation of 25 experimental series in which
the mechanical strength was determined by static compression (no impacts)
up to fracture. The segments within each group of fig. 6 were the same
length, but the groups may differ from one another. In some cases, complete
lumbar spines were tested. In other cases, segments consisting of only a
few vertebral bodies and intervertebral discs were compressed. In yet other
series, only single bodies or discs were fractured. A special analysis with
regard to the segment's length has revealed no significant influence.
Several experimental results listed in fig. 6 were originally given in
pressure values. In these cases, conversion into compressive force values
 477
f
became necessary. For this purpose, compressive strength of lumbar segments
the area of force transmission in force in kN
the 2umbar spine was assumed to 2e
15 cm (PEREY, 1957: about 217 cm; 8
FARFAN, 1979: about 13 cm). The
mean value of lumbar compressive
strength for the available 283 6
experiments was calculated as 5.1
2.3 kN (SD). This large range of 4
measured compressive strength data
illustrates the impossibility of 2
providing a single value of spinal
mechanical strength valid for all
persons.
A comparison of the strength values
with the calculated force values
illustrated in fig. 4 shows that in
lifting a load of 20 kg the mechan-
ical strength data for the spine
may be reached or even exceeded,
particularly if the trunk is in-
clined (back lift). Lifting heavier
loads would result in correspond-
ingly higher stress values. This
indicates the importance of advan- Fig. 6: Morphological studies to de-
tageous lifting techniques which termine the fracture load of lumbar
minimize health risks by avoiding segments of various length (see text)
high acceleration and large trunk
inclination. Even during single arm movement tasks with unsymmetrical body
postures, shown in fig. 5, the calculated compressive force may exceed some
of the measured strength data given in fig. 6.

In conclusion, the three-dimensional and dynamic biomechanical model of the

human body presented in this paper allows the simulation of the performance
of materials handling tasks and, in consequence, the determination of the
time course for spinal stress. For application in occupational health it
seems to be very convenient to compare the calculated compressive forces
with the mechanical strength of the spine in order to avoid a high health
risk for working persons.

REFERENCES
Andersson BJG, Ortengren R, Nachemson A, ElfstrOm G: Lumbar disc pressure
and myoelectric back muscle activity during sitting. Scand J Rehabil Med ~,
104-114 (1974)
Ayoub MM, El-Bassoussi MM: Dynamic biomechanical model for sagittal plane
lifting activities. In: Drury (ed): Safety in manual materials handling,
pp 88-95. US Department of Health, Education and Welfare, Cincinnati, Ohio
1978
Brown T, Hansen RJ, Yorra AJ: Some mechanical tests on the lumbosacral
spine with particular reference to the intervertebral discs. J Bone Joint
Surg 39A, 1135-1164 (1957)
Chaffrn-DB: A computerized biomechanical model - development of and use in
studying gross body actions. J Biomech 2, 429-441 (1969)
Chaffin DB, Andersson GBJ: Occupational biomechanics. Wiley, New York 1984
Decoulx P, Rieunau G: Les fractures du rachis dorso-lombaire sans troubles
nerveux. Rev Chir Orthop 44, 254-322 (1958)
478

Dempster WT: Space requirements of the seated operator. Wright Air Develop-
ment Center. Ohio 1955 (WADC technical report 55-159)
Drillis R, Contini R: Body segment parameters. Office of Vocational Reha-
bilitation. Department of Health. Education and Welfare. New York 1966
(Report No 1163-03)
Eie N: Load capacity of the low back. J Oslo City Hosp 16. 73-98 (1966)
Evans FG, Lissner HR: Biomechanical studies on the lumbar spine and pelvis.
J Bone Joint Surg 41A. 278-290 (1959)
Farfan HF: Biomechanik der Lendenwirbelsaule. Hippokrates. Stuttgart 1979
Gray H: Gray's anatomy. eds: R. Warwick and P.L. Williams. 35th edition.
Longman. Edinburgh 1973
Hettinger T: Statistics on diseases in the Federal Republic of Germany with
particular references to diseases of the skeletal system. Ergonomics 28.
17-20 (1985)
JAger M. Luttmann A, Laurig W: The load on the spine during the transport
of dustbins. Appl Ergonomics 15. 91-98 (1984)
JAger M, Luttmann A: Biomechanische Analyse ausgewahlter Tatigkeiten im
Baugewerbe. In: W Laurig. L Gerhard. A Luttmann. M Jager. H-E Nau (eds):
Untersuchungen zum Gesundheitsrisiko beim Heben und Umsetzen schwerer
Lasten im Baugewerbe. Wirtschaftsverlag NW. Bremerhaven 1985. pp 43-122
JAger M: Biomechanisches Modell des Menschen zur Analyse und Beurteilung
der Belastung der Wirbelsaule bei der Handhabung von Lasten. Diss. Fach-
bereich Maschinenbau. Universitiit Dortmund. 1986 (submitted)
Lanier RR: Presacral vertebrae of white and negro males. Am J Phys Anthro-
pol 25. 343-420 (1939)
Lanz~ von, Wachsmuth W: Praktische Anatomie. RUcken (J Rickenbacher. AM
Landolt und K Theiler). Springer. Berlin. New York 1982
Leskinen TPJ: Comparison of static and dynamic biomechanical models.
Ergonomics 28. 285-291 (1985)
Luttmann A,-YAger M, Schoo K-C, Laurig W, Puhlvers E: Wirkung erhBhter Wir-
belsaulenbelastung beim Lastentransport auf die Haufigkeit von RUckenbe-
schwerden. In: W Laurig. L Gerhard. A Luttmann. M Jager und H-E Nau (eds):
Untersuchungen zum Gesundheitsrisiko beim Heben und Umsetzen schwerer
Lasten im Baugewerbe. Wirtschaftsverlag NW. Bremerhaven 1985. pp 123-180
Morris JM, Lucas DB, Bresler B: Role of the trunk in stability of the
spine. J Bone and Joint Surg 43A. 327-351 (1961)
Nachemaon A: The influence or-spinal movements on the lumbar intradiscal
pressure and on the tensile stresses in the annulus fibrosus. Acta Orthop
Scand 33. 183-207 (1963)
Orne D:- Liu YK: A mathematical model of spinal response to impact. J
Biomech 4. 49-71 (1971)
Panjabi MM, I.
White III AA:- Basic biomechanics of the spine. Neurosurgery
76-93 (1980)
Perey 0: Fracture of the vertebral end-plate in the lumbar spine. Acta
Orthop Scand, Suppl 25 (1957)
Plaue R: Das Frakturverhalten von Brust- und LendenwirbelkBrpern: Kompres-
sionsversuche an frischen Leichenwirbeln. Z Orthop 110. 357-362 (1972)
Plaue R, Gesche E: Das Frakturverhalten von Brust- und LendenwirbelkBrpern.
Dynamische Kompressionsversuche an Leichenwirbeln. Z Orthop 112, 427-432
(1974) -
Roaf R: A study of the mechanics of spinal injuries. J Bone Joint Surg 42B.
810-823 (1960)
Wy8S T, Ulrich SP: Festigkeitsuntersuchungen und gezielte Extensionsbehand-
lung der Lendenwirbelsaule unter BerUcksichtigung des Bandscheiben-Vorfal-
les. Vierteljahresschrift der Naturforschenden Gesellschaft in ZUrich. 99.
Beiheft Nr 3/4 (1954)
 479

EFFECT OF TENSION PRELOAD ON LUMBAR SPINE KINEMATICS AND ITS

CLINICAL IMPLICATION
A.H. Soni. J.A. Sullivan. M.R. Gudavalli and V.A. Herndon
Oklahoma State University. Stillwater. Oklahoma; University of
Oklahoma Hlth. Sci. Center. Oklahoma City. Oklahoma. U.S.A.

INTRODUCTION :
Human spine is a complex mechanical structure, capable of
withstanding different types of physiological loads, and
allows sufficient motion between the head, trunck and pelvis.
Knowledge of the mechanical and kinematic behavior of lumbar
spine is important for understanding all aspects of the
clinical analysis and management of spinal problems.
Significant contributions have been made by many researchers
(1-8) for understanding the response of the lumbar spine
motion segments under different physiological loading
conditions namely, flexion, extension, right and left lateral
bending, clockwise and countercl~ckwise twisting, anterior and
posterior shear and lateral shear loads under the influence of
tension preload.
EXPERIMENTAL SET-UP
The experimental set-up consists of a spine fixture for
mounting the spine specimen and for the application of preload
as well as physiological loads. The three dimensional
relative motion is measured by means of a Linkage Transducer
comprised of seven links and six rotary potentiometers. The
data collection precedure is controlled by means of a
microcomputer with user interaction. Figure 1 shows the
photograph of the entire experimental set-up. The following
describes briefly the Spine Fixture and the Linkage
Transducer.
The Spine Fixture
The spine fixture has the following features.
(a) The top and bottom vertebrae of the motion segment
are rigidly attached to two aluminum cups using
steinman pins and bone cement (methyl methacrylate).
480

The bottom cup is clamped to the base of the fixture.

The clamping method is designed to allow easy
mounting and dismounting of the spine segment.
(b ) The spine motion segment along with the top cup is
able to execute its normal mode of motion in the
sagittal plane, frontal plane and any other plane in
between.
(c) A preloading bar fastened to the top cup allows the
application of a known amount of compressive preload
by means of static weights hanging from the
preloading bar. In addition traction preload can be
applied by means of cables passing through the
pulleys on the horizontal bar.
(d) The loading fixture consists of a rotating circular
base carrying four vertical bars and a horizontal
bar. The circular base can be rotated and positioned
to apply loading in the sagittal plane, frontal plane
or any other plane that lies at 15 degrees interval
to the sagittal plane.
(e) Shear and moment loads can be applied to the spine
segment by means of cables, pulleys and static
weights. One end of the cable is attached to the top
cup by means of a hook and the other end is connected
to a loading pan. The cables pass through a system
of pulleys attached to horizontal and vertical bars.
linkage Transducer:
A linkage transducer with seven links and six rota ry
potentiometers is used to measure the three dimensional
intervertebral motion. The linkage transducer has the
following features.
(a) The transducer has negligible weight. It does not
load the specimen. Also it does not interfere with
the actual intervertebral motion to be measured.
(b) The transducer can be installed with relative ease.
(c) The size of the transducer is small. Hence i t can be
mounted within the limited space available.
RESULTS

A total of four motion segments in the lumbar region (two

L2-L3 and two L4-L5) are tested with and without the
application of 130 Newtons of tension preload. The motion
segments are subjected to physiological loads simulating
flexion, extension, lateral bending and axial twisting upto a
maximum moment loads of 10 N-M and upto a maximum shear loads
of 180 N. Figures 2 through 5 show the bar charts of the mean
 481

values of rotations and translations about and along the screw

axis also called helical axis of motion under different
physiological loads as well as under the application of
tension preload. The following observations are noticed from
the results presented in the figures. Application of tension
preload has increased the laxity parameters for loading in the
sagittal plane. The rotation has increased by 25 percent in
flexion and '45 percent in extension. Translations have
increased by 37 percent in flexion and 58 percent in
extension. Contrary to these, the laxity parameters have
decreased under lateral bending and twisting moments. The
decrease in the rotations are 11 and 9 percent in right and
left lateral bending and 8 and 10 percent in counter-clockwise
and clockwise twisting moment loads. The decrease in
translations are 50 percent under counter-clockwise twisting
moment, 45 percent under clockwise twisting moment, 55 percent
under right lateral bending and 34 percent under left lateral
bending. Under the application of tension preload, the
rotations have increased by 48 percent under anterior shear
and 57 percent under posterior shear loads. Rotations have
decreased by 1 percent under right lateral shear and 14
percent under left lateral shear. The translations have
increased by 58 percent in anterior shear and 63 percent in
posterior, shear where as the translations have decreased by 42
percent in right lateral shear and decreased by 30 percent in
left lateral shear. Coupled motions of lateral bending,
flexion/extension and axial tWisting along with t~anslations
are observed for the segments with no preloads as well as with
the application of tension preload. No definite trends in the
changes of the coupled motions are observed when the tension
preload is applied. The decrease in the motion of of lateral
bending and axial twisting motions has a clinical significance
in understanding the management of scoliosis by various
devices which use traction. The progression of scoliosis is
reduced in the lateral direction aswell as the rotation of the
vertabrae in the horizontal plane when the spine is subjected
to traction.
REFERENCES :
1. Berkson, M. H. ,Nachemson, A. l. and Schultz, A. B.
(1979) "Mechanical Properties of the Human lumbar Spine Motion
Segments, Part II Responses in Compression and Shear;
I nfl uence of Gros s Morpho logy" Jou rna 1 of Bi omecha ni ca 1
Engineering, Transactions ASME, Vol. 101, pp. 53-57.
2. Panjabi, M. M. and White, A. A. (1977) "Effects of
Preload on load Displacement curves of the lumbar Spine"
Orthopedic Clinics of North America" Vol. 8, pp. 181-192
3. Schultz, A. B., Warwick, D. N., Berkson, M. H., and
Nachemson, A. l. (1979), "Mechanical Properties of the Human
lumbar Motion Segments Part I : Responses in Flexion,
Extension, lateral Bending and Torsion", Journal of
ABiomechanical Engineering, Transactions ASME, Vol. 101, pp.
482

46-52.
4. Soni, A. Ii., Gudaval1i. M. R., Sullivan, J. A. and
Herndon, W. A. (l985) "Migratory Pattern of Vertebral Motion
~n the Lumbar Spine", Proceedings of the 9th ASB Meeting. Ann
Arbor, Mi chi gan. Sept. 1985.
5. Soni, A. fl., Sullivan, J, A., Patwardhan, A. G.,
Gudavall;, M. R. and Chitwood, J. "Kinematic Analysis and
Simulation of Vertebral Motion Under Static Load - Part I :
Kinematic Analysts", Journal of Biomechanical Engineering,
Transactions ASME, vol. 104, pp. 105-11I.
16. Tencer, A. F. and Ahmad, A. M. (1981) "The Role of
Secondary Variables in the Measurement of the Mechanical
Properties of the Lumbar Intervertebral Joint." Journal of
Biornecl1anical Engineering,Transactions ASME, Vol. 103, pp.
129-137.

7. Tencer, A. F., Ahmad, A. M. and Burke, D. l. (1984) "Some

Static Mechanical Properties of the Lumbar Intervertebral
Joints, Intact and Injured." Journal of Biomechanical
Engineering,Transactions ASME, Vol. 104, pp. 143-201.
8. White, A. A. and Gordon, S. (1982] "Synopsis: Workshop on
Idiopathic Low-Back Pain." Spine, 1101. 7, No.2, pp. 141-149.

figure 1. Photograph of the Overall Experimental Set-up

 483

12
FLEX lOll RI GIlT LEFT
LATERAL LATERAL
10 BENDING BENDING

~ 8
cw
Cl

EXTENS ION CC\~

TOROUE TORQUE
rf
~ 6
l-
cC
:; 4
0: rt
2

1 2 1 2 1 2 1 2 i 2 1 2
0.0
1. NO PRELOAD 2. TENSION PRELOAD

Figure 2. ROTATIONS UNDER THE APPLICATION OF MOMENT LOADS

RIGHT LEFT
LATERAL LATERAL
2.54 BENDING BENDING
FLEXION
z 2.032 EXTENS ION
o
I-
:3 1.01(j
Vl CCW CW
z TORQUE TORQUE
:=
cC
0.508
1 2

10 NO PRELOAD 2. TENSION PRELOAD

Figure 3. TRANSLATIONS UNDER THE APPLICATIONS OF MOMENT LOADS
484

6
ANTERIOR RIGHT
SHEAR LATERAL LEFT
5 SHEAR LATERAL
SHEAR
4 .---1--
POSTERIOR
o SHEAR
3
l-
e:(

:; 2
...--
<>::

2 2 2 1 2
1. NO PRELOAD 2. TENSION PRELOAD
Figure 4. ROTATIONS UNDER THE APPLICATION OF SHEAR-LOADS

POSTERIOR
1.778 SHEAR

~ 1 .524
~
::E
~ 1. 270
o
;:1.016
c:(
-'
~ O. 762
c:(
c:::
ANTERIOR
I- 0.508
SHEAR

0.254
2 2
0.0 NO PRELOAD 2. TENSION PRELOAD
Figure 5. TRANSLATIONS UNDER THE APPLICATION OF SHEAR LOADS
 485

REGIONAL LUMBAR SAGITTAL MOBILITY AND BACK TROUBLE

K. Burton
Dept. Life Sciences. Huddersfield Polytechnic. Huddersfield.
U. K.

INTRODUCTION
It is a commonly held view in the fields of biomechanics and clinical
medicine that a relationship exists between abnormal lumbar sagittal
mobility and low back trouble (LST). There is general belief that sagittal
mobility is reduced in patients suffering from LST, and clinicians
traditionally consider the range of mobility when assessing such patients.
However, the evidence available from research studies is often difficult
to interpret and reference/normal values are not yet agreed. These
problems are in part due to the wide range of measurement techniques
available (reviewed by Pearcyl), and to doubts about their accuracy and
reliability. The most frequently used clinical technique, for instance,
employs a tape measure 2 but this has been shown by Reynolds 3 to be less
reliable than inclinometric measurements. In their study of 237 clinically
and radiologically normal subjects, Moll & Wright 2 concluded that lumbar
sagittal mobility was influenced by age and sex (reduced from age 25 years
and reduced in females). Their method of measurement of flexion includes
in its values the movement required to abolish the lumbar lordosis and may
thus be considered to overestimate true flexion (the converse being the
case for extension), which makes comparison of their reference values with
those obtained by, say, inclinometers particularly difficult.
More recently other workers have used a variety of measurement methods
to explore the relationship between sagittal mobility and LST. Anderson
& Sweetman in a study of 432 male workers could not relate lumbar sagittal
mobility to a past history of back pain4. However, Wickstrom et a1 5
reported a limitation of lumbar flexion in males with a history of sciatica
in both heavy manual and sedentary workers. Troup et a1 6 found a similar
relationship in 802 predominantly male working-age subjects. Mayer et a1 7
have shown a reduction of lumbar flexion in 38 chronic LST patients of
both sexes compared with 13 control subjects, whilst Pearcy et a1 8 , using
biplanar x-ray analysis found reduced sagittal mobility in 11 male low back
and sciatic pain patients. The wide range of values found in all these
studies suggests the need for caution in the interpretation of results,
and especially in the latter two due to the small numbers of subjects
involved.
The present report summarises the results of wh~t is essentially a
pilot study to investigate the influence of LST on lumbar sagittal
mobility. It was hypothesised that sagittal mobility is altered
(compared with non-sufferers) during a current spell of LST and also by a
previous history of trouble, and that these alterations will vary
dependant on age and sex.
METHODS
The method of measurement of sagittal mobility chosen has previously
been described 9 and makes use of a modified draftman's flexicurve to
486

record the mid-line contour of the lumbar spine (together with the
locations of T12, L4 and 52) in postures of maximal flexion and extension.
This technique has the capability of measuring 'true' flexion and extension
(the degree of bending of the lumbar spine in each direction from a
straight neutral line), and the ability to record the ranges of sagittal
mobility in upper (T12-L4) and lower (L4-51) regions of the lumbar spine.
The flexicurve is moulded to the lumbar spine in postures of flexion and
extensiori, the locations of the vertebral levels are marked and traces of
each contour drawnion paper. Tangents to the curves at T12, L4 and 52 are
drawn and the intersections of these lines give an angular measure of
'regional' mobility. For the purposes of this study regional mobility was
considered in terms of; (1) total mobility (ROM), (2) range of flexion
(FLEX), (3) range of extension (EXT), (4) flexion + extension in upper
lumbar region (UPPER), (5) flexion + extension in lower lumbar region
(LOWER), illustrated in Fig.l.

47

UPPER

[II
S2
FLL
20

U. 68
27

IDe

S2
m I".
LOWER
21

FIGURE 1. Method of calculating regional mobility from the four angles

measured by the flexicurve technique.

A self-administered questionnaire was used to obtain anamnestic data

concerning experience of LST, defined here as; (1) no recollection of ever
having had LST (NOSACK), (2) previous history but no current spell
(PREVHI5T), (3) experiencing a current spell (CUR5PELL). The PREVHI5T
subjects were further sub-divided as OCCA5 (occasional LST, 1-3 spells) or
FREQ (frequent LST, 4 or more spells). In addition age and gender were
recorded.
The questionnaire was completed by 333 adults recruited from a variety
of industries and two LST clinics. Their regional sagittal lumbar mobility
was measured using the flexicurve technique. The study population
comprised 208 males and 125 females who were placed in three age
categories; (1) YOUNG: 17-34 years, 133, (2) MIDOLE: 35-54 years, 131,
(3) OLD: 55-78 years, 69. There were 104 NOSACK, 106 PREVHI5T and 123
 487

CURSPELL subjects. Because of the lower number of females in the study

population Qnly the males are considered here in relation to frequency of
previous LBT (OCCAS = 49, FREQ = 24). Statistical methods used were the
unpaired t test and analysis of variance.
RESULTS
An initial analysis of the data was performed where the effect of sex,
age and current spell of LBT on the mean ranges of mobility was compared.
These results are shown in the Table. Whilst males and females displayed
a similar ROM, males displayed a greater range of FLEX and females had a
higher range of EXT and LOWER mobility. Subjects with a current spell of
LBT had reduced ROM which was due to reductions in FLEX and UPPER mobility.
ROM was also influenced by age showing a reduction between each category,
this being accounted for largely by a reduction in FLEX and UPPER
measurements.

TABLE. Mean regional lumbar sagittal mobility in 333 adults analysed by sex,
age and current spell of back trouble (measurements in degrees, s.d. in parentheses)

N ROM FLEX EXT UPPER LOWER

MALE -208 52.2(11.6) 22.4(8.3) 29.8(7.8) 37.0(10.2) 15.2(6.0)

FEMALE 125 53.9(1l.7) 18.3(8.2) 35.5(7.8) 37.3(9.4) 16.6(5.6)
ns ** ** NS +

YOUNG 133 56.6(12.1) 23.7(8.6) 32.9(8.3) 40.8(10.5) 15.7(6.1)

MIDDLE 131 52.4(10.5) 20.7(7.5) 31.7(8.3) 36.4(8.8) 15.9(5.7)
* * ns ** ns
OLD 69 46.6(10.1) 15.8(7.4) 30.8(7.8) 31.3(7.4) 15.4(5.8)
** ** ns ** ns
CURSPELL 123 51.0(11. 7) 18.9(8.2) 32.1(9.0) 35.3(9.5) 15.8(5.8)
OTHERS 210 53.9(11.5) 22.0(8.5) 31.9(7.8) 38.1(9.9) 15.7(5.9)
+ * ns * ns

(CURSPELL = current spell of LBT: YOUNG = 17-34y, MIDDLE = 35-54y, OLD 55-78y:
Significances by unpaired t test; + = <0.05, * = <0.01, ** = <0.001,
ns = not significant. Age tests are YOUNG v. MIDDLE and MIDDLE v. OLD)

In order to test the hypothesis that age, sex and LBT have inter-related
effects on mobility further analysis was required so the subjects were
sub-divided into age and sex categories for NOBACK, CURSPELL and PREVHIST.
There were insufficient subjects in some groups to permit full statistical
analysis but certain significant differences and general trends were found.
In the YOUNG/CURSPELL group, females displayed a relative hypermobility in
EXT and UPPER measures compared with males, and also with female NOBACK,
whilst males had reduced FLEX and UPPER compared with NOBACK males (Fig.2).
These patterns were not apparent in the PREVHIST subjects. There was a
tendency for CURSPELL to be accompanied by decreasing mobility in males
with advancing years; OLD/CURSPELL subjects did not show the reduction
of flexion seen in the young subjects (Fig.3). A similar tendency was
found in females where the EXT and UPPER hypermobility was abolished, the
OLD/CURSPELL subjects showing a trend towards hypomobility in these
measures together with a reduced LOWER range compared with NOBACK
488

subjects (Fig.4).
50
- r:!
---<;j?
L--<.-j
1-
X NOBACK

40

,';f,'-'1 ...........
30 ,, ,...... .. ......
,

---
-4p-f('
>-
....J
cc
o
:::E:
20

10
r
O
LO FL EX UP
FIGURE 2. Mean ranges of regional mobility (and 95% c.i.) for YOUNG/
CURSPELL subjects. Mean values for NOBACK are indicated.

---
50' 50'
r:! - YOUNG ~) x HOBACK t> - YOUNG ~) x HOBACK
--OLD " - - OLD

/,,:----_.
40' 40'

-_.
................J'

:......
3~' ....... ;.--- 30
.... ,,"
........",,"
20' ............ / 20' _------ /1
:........::: ........... -( I

.'
I
-----_J/
10' 10'

0' ~-----r------.-----~------~ 0' ~------~------~~------~------~~

LO FL EX UP LO EX UP

FIGURE 3. Mean ranges of regional FIGURE 4. Mean ranges of regional

mobility for YOUNG and OLD CURSPELL mobility for YOUNG & OLD CURSPELL
male subjects, together with age female subjects, together with age
matched NOBACK. matched NOBACK.
 489

The effect of the 'type' of LBT (OCCAS & FREQ) on total range oflmobility
in males is shown in Fig.5 for each age group. The angular differences
were not sufficient to be statistically significant in such small sub-
groups but there is a trend towards altered patterns of mobility between
groups.

60

50

YOUNG MIDDLE OLD

NOBACK
.oil. OCCAS
FREQ

FIGURE 5. Comparison of total mobility (ROM) for PREVHIST males (sub-

divided as OCCAS & FREQ) together with NOBACK values in each age group.

DISCUSSION
The results of this study support the previously reported progressive
loss of lumbar sagittal mobility with increasing age but not that females
have a relatively lower range. The use of a 'regional' measurement
technique together with the distinction between age/sex matched NOBACK,
CURSPELL and PREVHIST groups, increases the informative content of
mobility measures. The hypothesis proposed at the outset, that
alterations of mobility are dependent on these groupings, is tentatively
supported by the results reported here.
The different mobility profiles between YOUNG/CURSPELL males and
females suggests a different biomechanical functioning of the lumbar spine
and the opposite change in mobility in males and females from NOBACK to
CURSPELL invites the suggestion that there may be different functional
bases to LBT in young males and females. The different regional mobility
profiles in young and old CURSPELL subjects compared with matched NOBACK
groups may be an indicator of different types of back trouble affecting
different age groups. The fact that ROM in PREVHIST males (OCCAS and
FREQ) did not follow the normal decreasing pattern with advancing age is
difficult to interpret; perhaps there are residual effects on mobility
following LBT or possibly some patterns of mobility indicate a
predisposition to recurrent LBT. Further work is clearly required but it
seems obvious that PREVHIST subjects should be omitted when defining
reference ranges for normal sagittal mobility.
This pilot study indicates that the relationship between lumbar
sagittal mobility and LBT is more complex than has pre~iously been
4~

reported. Regional measurement of sagittal mobility offers useful

information over other techniques, but a much larger data set is required
to identify reference ('normal') ranges and to determine any
discriminatory or predictive potential of regional mobility profiles.
These extra data are currently been collected.
CONCLUSIONS
1. Sagittal mobility measured 'regionally' offers more information than
other methods.
2. Regional sagittal mobility depends on: sex, age and current spell of
LBT.
3. Patterns of regional mobility may vary in the presence of LBT at
different ages.
4. Patterns of mobility may vary in respect of the 'type' of previous
LBT.
REFERENCES
1. Pearcy, M. Measurement of back and spinal mobility. Clin Biomech 1986;
1:44-51.
2. Moll, JMH, Wright, V. Normal range of spinal mobility;an objective
clinical study. Ann Rheum Dis 1971;30:381-86.
3. Reynolds, PMG. Measurement of spinal mobility; a comparison of three
methods.~heumatol Rehabil 1975;14:180-85.
4. Anderson, JAO, Sweetman, BJ. A combined flexicurve/hydrogoniometer
for measurement of lumbar spine and its sagittal movement. Rheumatol
Rehabil 1975;14:173-79.
5. Wickstrom, G, Nummi, J, Nurminen, M. Restriction and pain during
forward bending in concrete reinforcement workers. Scand J Work Environ
Health Supplement 1:29-38. 1978.
6. Troup, JOG, Martin, JW, Lloyd, OCEF. Back pain in industry; a
prospective survey. Spine 1981;6:61-69,
7. Mayer, TG, Tencer, AF, Kristoferson, S, Mooney, V. Use of non-invasive
techniques for quantification of spinal range of motion in normal subjects
and chronic low-back dysfunction patients. Spine 1984;9:588-95.
8. Pearcy, M, Portek, I, Shepherd, J. The effect of low-back pain on
lumbar spinal movements measured by three-dimensional x-ray analysis.
Spine 1985;10:150-53.
9. Burton, AK. Regional lumbar'sagittal mobility; measurement by
flexicurves. Clin Biomech 1986;1:20-26.
 491

ELECTROMYOGRAPHIC ACTIVITY AS AN INDICATOR OF BACK LOAD AT

WORK
T. Leskinen*, H. Nieminen**, S. Kalli**, M. Gautreau* and
I. Kuorinka*
Institute of Occupational Health*, Helsinki, Finland;
Technical Research Centre of Finland**, Tampere, Finland

1. INTRODUCTION
Back muscles act with a shorter lever arm to lumbar vertebrae, than
external loads while e.g. handling loads. Thus the tension in back muscles
is the most important factor contributing to the spinal load.
In static conditions the electromyographic (emg) activity and the force
production of limb muscles have been found to correlate well with each
other linearly (2) or nonlinearly (9). Assuming the same to be valid for
back muscles, the load on the back has often been evaluated by the emg-
activity of erector spinae muscles, as described by root mean squares (rms)
or rectified averages of the raw emg-signal (e.g. 1, 3). However, in
dynamic work some additional factors influence this relationship. The
relationship between muscle force and emg-activity is strongly dependent on
the length of the muscle (5); in elongated muscles most of the force is
produced by passive stretching, not by an active contraction. In back
muscles usually no electric activity appears in a stooped down posture (4).
Muscles develop greater forces in eccentric than in concentric contractions
with the same electric activity (5).
Our aim was to develop a portable multichannel system with which
electromyography of back muscles can be recorded at work places during
normal work. The relationship between emg and biomechanically computed
muscle force in non-rotated postures was studied in laboratory. Moreover,
we studied the differences in emg-activities of left and right erector
spinae muscles in rotated postures in static and dynamic conditions to see
if the differences could be used as indicators of trunk rotations at work.

2. MATERIALS AND METHODS

The recording system had to be light and small, so that it could be
carried by a worker without disturbing the work. However, both artefacts
induced by electrode and cable movements, and electric interference had to
be rejected effectively. To fulfil these requirements a system was
constructed (Figure 1) consisting of surface electrodes, preamplifiers
attached close to the electrodes, a main unit and a small tape recorder
(TEAC HR-30E). The system has four EMG-channels. Besides band-pass
amplifiers, the main unit has a timer and remote control circuits for the
recorder to allow recording of fixed length samples with fixed intervals
during a whole work-day. Of the seven channels of the recorder one is used
for noise compensation and one for a time reference sighal for the sampling
method. The whole recording system weighs less than 1 kg, and it can be
carried in two pockets.
The laboratory experiments were done with 10 healthy subjects, 5 men
(age 29-36 years, height 178-189 cm, weight 67-79 kg), and 5 women (age
33-42 years, height 160-171 cm, weight 45-64 kg).
492

remote control
preamplifiers
e~t~
---00
J recorder

---00
Main unit
0 0
- filtering, amplification
- signal sampling

Figure 1. Block diagram of the recording system.

The emg-electrodes were placed at the Ll and L4 levels on both sides of

the spine midline. The electrodes of one electrode pair were placed
longitudinally and the displacement between the electrodes was 3 cm when
the subjects sat with their spine flexed. The experiments began with a
period of rest, followed by an isometric maximal voluntary contraction
(IMVC) of back muscles with the subjects lying prone.
Static non-rotated postures studied included standing straight, and
forward inclinations of the body of 15, 30, and 45 degrees. Deeper flexions
were excluded because the emg-activity was expected to be negligible after
45 degrees (8). The subjects stood either without any external load or with
a box weighing 10 or 20 kg. The angle of the hip-shoulder line was measured
visually with an angle measuring device.
In rotated postures the trunk was rotated to the left or to the right so
that the line between the hands was 45 degrees rotated from the frontal
plane. The tests were only done with the box weighing 10 kg, and the height
of the box was adjusted on the same levels as in the non-rotated tests.
In the dynamic tests the subjects lifted the box (or extended without a
load) from the 45 degrees forward inclined posture to erect standing and
then lowered again back to the initial posture. The same weights were used
as in the static tests. The lifts had to be done smoothly without jerks,
but no instructions were given regarding the speed of lifts.
The emg-signals were sampled by the computer with an A/D converter at
1000 samples/so The signal describing the emg-activity was obtained from
the raw emg-signals by full-wave rectification and sliding averaging with a
selectable averaging window. The peak emg-activity during the IMVC test
obtained with the window of 500 ms was used as the scaling value for the
emg-activities recorded during the other experiments.In static experiments
the mean emg-activities of muscles during a 3 s recording period were
computed. For dynamic experiments the window length used was 200 ms and the
peak activities were detected.
The emg-activity in non-rotated postures was compared to biomechanical
muscle force estimates (BMF) based on optoelectronic kinematic recordings.
The method was slightly modified from the method used by Leskinen et al
(6), to produce the force of back muscles and not the compressive force on
the spine. For static postures the parameters of linear regression between
BMF and emg-activity were defined. For dynamic experiments the cross-
correlation function between BMF and emg-activity was computed.
 493

3. RESULTS
Figure 2 shows the means of the 10 subjects' BMF and emg-activity on the
right erector spinae muscles at the L1 and L4 levels in static postures.
The BMF increased homogeneously with the increasing flexion angle, but the
emg-activity stopped increasing after 30 degrees of inclination at the L4-
level, and even before at the L1-level.
The median coefficient of correlation in linear regression was 0.831
(range 0.08 - 0.986) when all postures were included in the regression, and
0.962 (range 0.548 - 0.986) when the inclination of 45 degrees was excluded
from the regression.
Figure 3 shows the means of the peak BMF and the peak emg-activities on
the right side in lifts. The peak BMF increased quite linearly with the
increasing external load. There was no difference between lifting and

EMG- L1-level EMG- L4-level

kN BMF %IMVC activity %IMVC activity
20kg
3 30 30 20kg
10kg 10kg
2 20 20kg 20
Okg
10kg Okg
10 Okg 10

0 15 30 45 deg 0 15 30 45 deg 0 15 30 45 deg

Figure 2. Mean biomechanical muscle force (BMF) and emg-activity of right

erector spinae muscles at L1 and L4 levels in static holding of 0, 10 and
20 kg weights with different trunk inclinations.
EMG-
kN BMF %IMVC activity

lifting
4 80 L4

L1
3 lifting 60

lowering lowering
2 40

20

o 10 20 kg o 10 20 kg
Figure 3. Average BMF and emg-activity peaks when lifting and lowering
different weights from and to 45 degrees inclination.
494

kNBMF
5
/".

4
/ 1,\
I \
3 I \
). \v
2 V\ ./
\
t
0.5 1.0 1.5 2.0 2.5s
%IMVC EMG-activity cross-correlation

1\1
40 1.0

35
rv-, .A
/
/ \
0.5
30
25
r '~
o -- --./
/
20
h !\ /
15 J ~ -0.5
~
10 0
0.5 1.0 1.5 2.0 2.5tS -1.0 -1.0 -0.5 o 0.5 1.0 s

Figure 4. Biomechanical muscle force, emg-activity of erector spinae

muscles (sum of four channels, averaging window 200 ms) and cross-
correlation of BMF and emg-activity in a lift of 20 kg.

lowering in the BMF. Also the peak activity increased homogenously with the
load. However, there was a big difference between lifting and lowering; the
peaks in lifting were almost twice as high as in lowering.
Figure 4 presents an example of the BMF, the summed emg-activity of all
channels, and the cross-correlation between the BMF and the emg-activity in
a lift. The median peaks of the cross-correlation function were 0.72 (range
0.18 - 0.97) in lifting and 0.80 (range 0.10 - 0.96) in lowering.
When the trunk was rotated to the left more activity appeared on the
right erector spinae muscles than on the left as shown in Figure 5 for
static postures. In Figure 6 the emg-activities are shown for lifts in
rotated and straight postures. The differences between the sides were
bigger in lifting than in lowering.
The standard deviations of the results, which usually were high, were
not included in the figures to improve clarity. However, the trends were
common to all subjects, and the observations reported above were of
statistical significance (at least p<0.05) when tested with the t-test for
paired observations.

-4. DISCUSSION
The use of long-emg recordings for the evaluation of spinal load at work
is probably justified when work is done in an erect standing posture, since
the emg-activity and the spinal load are very well correlated to each other
when the load is changed in the same posture (2). However, when the work
includes extremely stooped postures, the emg-activity gives no information
on spinal load ( 4, 8). Our results of decreasing activity after 30 degrees
of inclination support the earlier findings that when the muscle length is
 495

EMG- EMG-
%IMVC activity %IMVC activity

30 30

L1dx L4sin
20 20
L 1sin

10 10

o 15 30 45 deg o 15 30 45 deg

Figure 5. Mean emg-activity of erector spinae muscles in static holding of

10 kg with the trunk rotated to the left.

EMG- EMG-
%IMVC activity %IMVC activity

80 80 L4dx up
L 1dx up
60 60

L4dx down
L4sin down
40 L 1dx down
:::::::== 40

L 1sin down
20 20

LEFT o RIGHT LEFT o RIGHT

Figure 6. Average peak emg-activities of erector spinae muscles in lifting
(up) and lowering (down) in 45 degrees left and right rotated and straight
(0) postures.

increased, an increasing portion of the force output comes from passive

straining of the muscle itself and of the posterior ligaments (e.g. 4).
The difference that we found in peak emg-activities of lifting and
lowering despite the equal peak BMF's is supported by the finding that
muscles create greater forces with equal activities in eccentric
496

contractions than in concentric contractions (5). So when the force

production of muscles in real dynamic work is considered, both the muscle
length and the rate of change in muscle length should be taken into
account.
Our suggestion is to use postural data, e.g. recorded with an
inclinometer, to correct the emg-activity of the erector spinae muscles in
different postures so that the resulting signal would correspond the force
output of the muscles, which is very closely connected to the compressive
load on the spine. There are interindividual differences in how the flexion
is divided between the hip joint and the spine (7) and in what phase of the
flexion the emg-activity ceases (8). Such differences can be taken into
account with a carefully planned test series in the beginning of recording.
The results of the rotated postures and lifts showed that trunk
rotations can be indicated by comparing the emg-activities of left and
right erector spinae muscles, provided that the isometric test contractions
are strictly controlled to be symmetric. However, the results obtained in
this study were only qualitative, the quantitative aspects of rotations
need further research.

ACKNOWLEDGMENT
This study was supported by the Finnish Work Environment Fund

REFERENCES

1. Andersson, G. B. J., Ortengren, R., 1984, Assessment of back load in

assemblyline work using electromyography. Ergonomics, 27, 1157-1168.
2. Asmussen, E., Poulsen, E., Rasmussen, B., 1965, Quantitative evaluation
of the activity of back muscles in lifting. Communications of the Danish
national association for infantile paralysis Nr. 21.
3. Burton, A. K., 1986, Spinal strain from shopping bags with and without
handles. Applied Ergonomics, 17, 19-23.
4. Floyd, W. F., Silver, P. H. S., 1951, Function of erector spinae in
flexion of the trunk. Lancet, 260, 133-138.
5. Komi, P. V., 1973, Relationship between muscle tension, EMG, and
velocity of contraction under concentric and eccentric work. In New
developments in electromyography and clinical neurophysiology, Ed. by J.
E. Desmedt. S. Karger AG, Basel, Switzerland, 596-606.
6. Leskinen, T. P. J., Stalhammar, H. R., Kuorinka, I. A. A., Troup, J. D.
G., 1983, A dynamic analysis of spinal compression with different
lifting techniques. Ergonomics, 26, 595-604.
7. Leskinen, T. P. J., Taka1a, E. P., Stalhammar, H. R., 1987, Lumbar and
pelvic movements when stooping and lifting. In Biomechanics X-A, Ed. by
B. Jonsson. Human Kinetics Publishers, Champaign, Illinois, In press.
8. Takala, E. P., Leskinen, T. P. J., Stalhammar, H. R., 1987,
Electromyographic activity of hip extensor and trunk muscles during
stooping and lifting. In Biomechanics X-A, Ed. by B. Jonsson. Human
Kinetics Publishers, Champaign, Illinois, In press.
9. Viitasalo, J. T., Komi, P. V., 1978, Interrelationships of EMG signal
characteristics at different levels of muscle tension and during
fatigue. Electromyogr. Clin. Neurophysiol., 18, 167-178.
 497

AN ATTEMPT AT DETERMINING THE EFFORT OF THE LUMBAR PART OF THE

SPINE ON THE BASIS OF MODEL INVESTIGATIONS
R. Bedzinski and A. Wall
Technical University of Wroclaw. Wroclaw. Poland

1. INTRODUCTION
The mechanical factors acting upon vertebrae, interarticu-
lar joints and ligament,muscle elements have a considerable
effect on the generation of deformities and dysfunctions of
the spine. These factors are of essential importance because
the spine assumes certain configurations. e.g. 'movement in
the sagital plane, which may cause its overload.
One of the conditions of acquiring a good knowledge of the
pathomechanism of overload is the estimation of the spinal
effort which in this case is understood as an analysis of the
state of stress and deformation of the bony parts of the spi-
ne. which are caused by external and internal loads.
The analysis of the spinal strain under definite biomechani-
cal conditions makes it possible to estimate its effort and
thus determine the pathogenesis of diseases and the dysfunc-
tion of the spine.
The aim of the present paper is an attempt to account fo~
some elements of the overload of the lumbar spine as a result
of mechanical loads. Investigations were carried out into the
lumbar spine since, on account of its anatomic structure and
functions, it is particularly exposed to the action of all
kinds of loads.
2. EXPERIMENTS
Investigations were carried out into the standing position
of the human body, which is apparently static. Nachemson's[2]
investigations have shown, however, that in this position the
spine is exposed to a c~~siderable load.
Because of the complex structure of the spine a decision
was taken to carry out investigations on three-dimensional
spine models by using the stress frozen photoelasticity ~eth
od [1]. The models we re made in the 1: 1 scale and they re-
tatined the differentiated stiffness of compact and spongy
parts of the vertebral body as well as the appropriate rela-
tions of the stiffness of intervertebral discs and vertebrae.
The investigations were carried out on models of the lumbar
part of the spine as well as on the sacrum. The models were
loaded with vertical forces applied to the vertebra Ll.
Moreover, the interaction of ligaments and muscles on the
spinous processes as well as the work of intervertebral joint
capsule were also considered.
The geometric parameters of the models (Tabel~l was obtai-
ned on the basis radiographic examinations of the ten male
498

patients in the 24 to 42 age span X-ray recording of the lum-

bar spine in the sagital plane was performed for the standing
position. For shaping the model the average geometrical valu-
es of model parameters which corresponded to population of
patients were deformined.
Table 1-
Geometric parameters of the models examined .
Index Of Whitmas Value
Number lumbar Loads ot
of Ferguson of interspi-
lor~osis angle (2) vertical nous pro-
model IUS1 ~ load cesses
1 5,7 28u 50 3
2 8,16 36 0 50 3
3 8,16 39 0 50 3
4 6,5 42 0 50 3
Note (1) - index of lumbar lordosis
ILL = d/l 100 [%1
- where d - maximal distance between the chord
of lordosis and arch,
(~ - Whitman-Ferguson angle
<J W-F - inclination angle of the upper pla-
ne of sacrum in relation to hori-
zontal position.
Tab I e 2.
The mechanical and photoelastic properties of vertebrae
(bones) and models.
Bone eve rteb ra} Model (at Inte rve rte-
freezing tem- bral disc
perature)
f-i0rtical ~pongy r-todel of r-todel Disc MOdel
part part ~ortical of (f reezing
part ~pong'l temp .)
part
tyoung's2modulus 12,48.10:! 1,517
E [MN/m] !1,172
103
. 10 3 ;1:
4,17 3
",20,0 ..e,1

~ss~ed value
Ea N/m 2J
12,5 3
10
10
1,5 !
10
- - ",1,4 3,1

Eco rtica l/E spon _ ... 8,3 ... 9,5

gy
Poisson's ratio 0,28 0,14 0,5
Stress-optical
coefficient K
KPa m/isochr.
- - 0,29
patt]
The values of the lumbar lordosis and the inclination angle
of the sacrum were used as the parameters that varied the
particular models.
 One of the essential problems in these investigations was
to determine the physicomechanical properties of the spine
elements. The review of the related literat~re [a.9,10J. has
shown the values of the moduli of elasticity. of Poisson's
ratio and of the rigidity of the spine elements differ within
quite a wide range. Based of the literature quoted, the au=
thors decided to choose the pnysicomechanicsl parameters
which are presented in Table 2. These parameters have become
a basis for selecting t~e properties of the photoelastic mo=
dels (Table 2)
A composition consisting of epoxy resin was used to make
the compact parts of vertebrae. The models of the spongy
parts of vertebrae also were made from epoxy resins,but with
addition of the silicon powder and accelerant (11]. These mo~
dels were cast in moulds from silicon rubber using anatomi~
cally correct preparations of the human vertebrae as casting
models. The use of this technology made it possible to obtain
a diversification of the elasticity moduli of the compact and
spongy parts of the vertebra models (Table 2). The models of
intervertebral discs as well as intervertebral joint capsules
were made from silicon rubber by using moulds of bees-wax.

Fig.l. A Photoelastic model

or the lumbar spine. -
Fig.3. An example of an iso~
chromatic pattern.

Fig.2. SchematiC diagram of cutting

the model into samples.
500

The photoelastic model of the lumbar spine is shown in Fig.1.

The models of the lumbar part of the spine together with its
sacrum part were assembled in the loading system which made
it possible to carry out vertical loads through the vertebral
bodies as well as the interactions of ligaments and muscles
between the spinous processes. The loaded models were put
into an oven provided with programmed temperature control.
where the stress "frozen" process ~] was carried out. After
the models were "frozen" the three-dimenSional problem was
deduced that way into a series of two-dimensional problems by
cutting the model into thin plane slices (fig.2) which were
then analyzed as plane. two-dimensional models. The samples
~hin slices) were placed in the visual fields of a polaris-
cope and the distributions of white and dark fields of iso-
chromatic patterns were recorded (fig.3).
3.RESULTS
On the basis of the isochromatic patterns the diagrams of
distributions of the quasi-principal stress on the contour
were drawn up (fig.4). The values of contour stresses were
determined by using a well-known dependence referred to as
isochromatic equation
61 - 3 2 = mK
since on the contour of models one of the stresses, for exam-
ple, 6 2 " a and hence
GK = 6 1 = mK
where and 6 2 - quasi-principal stresses,
- contoor stresses,
- number of isochromatic pattern,
- stress-optical coefficient of model at frezing
temperature.
According to the contour stress distributions obtained it was
possible to make a comparative analysis of the values of
stresses in the particular parts of the vertebra such as the
vertebral body, pedicles and isthmuses. The analysis of stress
distributions was carried out only for lower lumbar vertebrae
L4 and L5 since according to clinical practice these verte-
brae are exposed to overload most frequently.
On the basis of the diagrams of contour stresses it was
possible to prepare diagrams of changes of the maximal stress
values expressed in the number of isochromatic patterns de-
pending on the changes of the values of lumbar lordosis (ILL)
ang of the inclination angle of the sacrum (JWF) (Fig. 5,6).
It was assumed that in the case of the vertebra being exposed
to overload the values of maximal stresses are of essential
importance, and hence in the analysis presented in the dia-
grams only the numbers of the maximal values of isochromatic
patterns are compared for the particular parts of the verte-
bra (vertebral body, pedicles, isthm~ses). The comparison of
stresses in the above mentioned parts of the vertebrae makes
it possible to estimate their participation in the carrying
of vertical loads through the lumbar spine.
 501

Fig.4. An exemplified di-

stribution of con-
tour st resses.

-VB
---p
--3

Fig.5. A diagram of the changes of maximum stresses (in the

number of isochromatic patterns) depending on ILL and
""W-F for L4 vertebra, in vertebral bodies (VB),
pedicles (P), and isthmuses (I).
4. CONCLUSIONS
The analysis of experimental results, e.g. the isochromatic
patterns, the distributions of contour stresses and the dia-
grams of the changes of the stress values in the particular
pa rts depending on ILL and i:WF changes make it possible to
draw conclusions with regard to the effort of the spine when
exposed to vertical loads.
The analysis of contour stresses allows a conclusion to be
drawn that their distribution in non-uniform, especially in
the lower lumbar vertebrae. The non-uniformity of the distri-
bution is largely dependent on the level of the vertebra, e.
g. on the geometric parameters of the lumbar spine such as .
the v~lue of the lordosis shape and the inclination of the
502

-VB
--- p
_.- :J

(WE
Fig.6. A diagram of the changes of maximum stresses depending
on ILL and ~WF for L5 vertebra, in vertebral bodies
(VB), pedicles (P) and isthmuses (I).
sacrum. The higher values of the maximal stresses are obser-
ved in the lower lumbar vertebrae. The diversified values of
stresses in the particular p~rts of vertebrae such as the
vertebral body, pedicles and the isthmus can be noticed.
This is eVidence that these parts have a signific~nt share in
carrying loads. The higher values of stresses in the pedicles
and isthmus can be observed for the L5 vertebra with the in-
creased inclinations of the sacrum, whereas the higher value
of the lumbar lordosis has a considerable effect on the in-
crease of stresses in the L4 vertebra.
The non-uniformity of the distributions of stresses in the
vertebral bodies as well as the frequently comparable values
of stresses in the isthmuss and in pedicles in relation to
the vertebral body make it possible to conclude that some
diseases such as spondilolisis, spondilolistesis and some
degenerative changes may be caused by the so-called mechanic-
al i factors.
The position which the spine of the human body assumes in
the course of everyday activities has an effect on the confi-
guration of the spine and tends to form the state of overlo
ad, which may lead to some spine diseases.
REFERENCES
1. Kuske A. and Robertson C., Photoelastic stress analysis.
~ohn Wiley and sons London, 1974.
2. Nachemson A., Towards a better understanding of low-back
pain: a review of the mechanics of the lumbar disc. Rheu-
matology and ~ehabilitation 14-129, 1975.
 503

3. Yoshizawa H., Photoelastic study on the relation between

the lumbar lordosis and the load. ~.~ap.Orthop.Ass., 43,
8, 645-662, 1969.
4. Ranieri L., Le spondilolistesis, Proc. LV Congresso della
Societa Italiana di Ortopedia e Traumatologia Napoli, Aulo
Gaggi Editore Bologna, 49-57, 1970.
5. B~dzinski R., Koz~owska A., Wall A., A study of stress
distribution in three-dimensional models of the lumbar
spine. International series of Biomechanics. Volume lA,
Biomechanik VII-A University Park Press Baltimore, PWN
-Warszawa, 65-72, 1981.
6. B~dzinski R., Koz~owska A., Wall A., Analisi dello stato
di tensione in modelli tridimensionali della colonna ver-
tebralle lombare, Progressi in patologi vertebrale. Vol.
VL, Aulo Gaggi-Editore Bologna 65-73, 1984.
7. B~dzinski R., Elastooptyczne badania mode Ii kr~goslupa z
uwzgl~dnieniem korowej i g~bczastej struktury kosci. XI
Sympozjum Badan w Mechanice Ciala Sta~ego, Warszawa, 32,
35, 1984.
8. Yamada H., Strength of Biological Materials; Robert E.,
Kriger Publishing Company Hurtington New York, 1973.
9. Evans F.G., Mechanical Properties of Bone. Thomas Spring-
field, 1973.
10.Hakim N.S., King A.~., A three dimensional finite element
dynamic response analysis of a vertebra with experimental
verification. ~.Biomechanics, Vol. 12, 277-292, 1979.
11.Modina A., Campunuovo, Alcune soluzioni nella realizzazio-
nne di modeli fotoelastici tridimensionali. VIII Convegno
Nazionale dell'Associazione Italiana Per L'Analisi Delle
Sollecitazioni, 47-51, 1980.
 505

EXPERIMENTAL STUDY ON THE TORSIONAL RIGIDITY OF VARIOUS DORSAL

STABILIZATION PROCEDURES ON LUMBAR SPINES
Ch. Ulrich. O. Worsdorfer and L. Claes
Klinik ffir Unfallchirurgie. University of Ulm. Ulm. Germany

I NTRODUCTI ON
The instability in the sagittal plane as well as a marked
rotational instability are characteristic features of a
classic fracture-dislocation of the lumbar spine (3):
rotational motion in the functional units of the lumbar spine
is considerably limited under physiological conditions
compared with the other functional units of the spine (2,9).
As the articular processes are part of the spinal canal,
torsional injury and subsequent damage of the processes will
lead to neurologic complications in particular (3).
Consequently, a high degree of torsional stability must be
achieved by stabilizing procedures carried out on the lumbar
spine in order to bring about adequate fracture
immobilization and to restore physiological conditions;
particularly in case of postoperative functional therapy
without external immobilization.
The torsional rigidity of various posterior stabilizing
procedures and implant systems of the spine were, therefore,
examined in vitro in experimental studies (1,7,10).

METHOD
In order to establish the degree of primary torsional
rigidity achievable by dorsal spinal fixation procedures, 14
fresh, deep-frozen lumbar spines from corpses from T 12 to S
1 were detached from the muscles, and the proximal and distal
ends were embedded in methacrylate. Then the lumbar I;Ipine was
mounted on a material-testing machine and subjected to
torsional moments (Fig. 1).
With the distal end fixed, torque was applied at the
proximal end of the clamped lumbar spine, reaching a maximum
of 5 Nm. With two rotary variable differential transformers
(RVDT) in the vertebral bodies proximal and distal to the
,lesion it was possible to measure the relative torsion angle
lin degrees of angle.
The signals of the RVDT's were amplified and analogously
recorded on an x-y Recorder.
In each specimen, the relative torsion angle was measured
in the following sequence:
1. Intact lumbar spine
2. Severance of the disk with removal of part of the upper
plate
:3. In addition, severance of the ligamentous structures
506

situated posteriorly thereof

4. In addition, severance of the articular processes
The instability produced thereby was then stabilized in
the following sequence:
1. Plate bridging over two vertebrae proximal and distal of
the lesion after ROY-CAMILLE (8)
2. Plate bridging over each individual v~rtebra proximal and
distal of the lesion
3. Fixateur externe after MAGERL(6)
4. Fixateur interne after DICK (1)
5. HARRINGTON distraction system with bridging-over of 2
functional units proximal and 2 functional units distal of
the lesion (4)
'6. Combination of the HARRINGTON distraction system with
segmental sublaminar wiring according to LUQUE (5)
I 6,7
Fig. 1: Test set-up
3r----n~1~
JLCra
1- lumbar spine 2
specimen
2- proximal lumbar
spine fixation
with torque
load-cell
3,4,5- Rotary vari-
able differential
transformer (RVDT)
6,7,9- Amplifier
8,10- Recorder for 4,5
torque angle of
rotation curves

RESULTS
There was an initial load of 0.8 Nm because of the weight
of the RVDT's and their distance from the torsion axis of the
spine.
The torsion-deformation diagram characteristic of each
specimen was then recorded, with increasing torsion moment up
to 5 Nm.

Fig. 2: 3
rorsion-deformation <D Intact spine
curve of the Anterior lesion

lumbar spine @ Additional disseCtion 01 the

posterior ligaments
without stabilization @) Additional transverse fracture
of the posterior elements

5 10 15
 507

In the intact lumbar spine, the maximal difference in

torsion angle was 1.96 1.77; after producing the anterior
defect it increased to 2. 8 2. 19. A hi.gher degree of
instability of 4.82 4.23 was only brought about if, in
addition, the posterior ligaments had been severed. After
severance of the articular processes, the difference in
torsion angle increased to 10.8 4.63 (Fig. 2).
After stabilization of the unstable lumbar spine, the
highest degree of primary stability was found in the spine
instrumented by the Fixateur externe, with a maximal
di fference on torsi on angle of 1. 03 O. 87; wi th 1. 2 O. 89
the long plate fusion with 2+2 bridging-over showed about the
same degree of stability. With the Fixateur interne, the
angle measured was 2.94 2.53 and with the short plate-
fusion it was 4.07 3.21.
The HARRINGTON distraction system with contoured rods for
3-point-fixation showed a difference in torsion angle of
13.41 4.67; a value comparable with that of an unstable
spine. With 14.02 4.23, the combination of this system
with multisegmental cerclage fixation according to LUQUE was
in the same range (Fig. 3).
Mr(Nm)

5
Fig. 3:
torsion-deformation
curve showing the 4
stabilizing effect of
the different fixation
procedures. The poin- 3
ted lines figure the Fixat81lr externe 1+1
Qil Roy-Camille 2+2
curves of the intact C Fixateur interne 1 +1
2
and unstable spine. @ Roy - Camille 1 +1
Harrington - Distraction 2 +2
Harrington-Luque 2+2

Angle (0)
5 10 15
DISCUSSION
Interpretation of in vitro experiments is problematic,
mainly because the effect of the stabilizing muscles cannot
be simulated. Thus the results do not represent in vivo
conditions. Moreover, the various fixation systems, owing to
their different design, are very different in their
mechanical mode of action.
As a rule, the forces and momentums acting on the spine in
the six degrees of freedom of motion are coupled. Their
absolute values are not known (9). Experimental stability
stUdies of specimen can, therefore, take into account only
the main components or the very highly simplified combinatory
forms of these forces or momentums, and thus they are mostly
restricted to the main loading direction of anterior bending
load and axial torsion.
In vitro stUdies on standardized fracture models, however,
permit profitable comparison of the various fixation systems;
508

also with regard to their different interaction between

specimen and implant. The findings arrived at allow in turn
to draw conclusions as to the clinical applicability and
effectiveness of the implants.
For the first instance our results demonstrate the fact
that in the lumbar spine the intact articular processes of
their mode of articulation are most effective in blocking
rotation, but rotational stability os not only dependent on
osseous but also on discoligamentous integrity of the
functional spinal unit (2).

The HARRINGTON system will bring about a stabilizing

effect only if the rods are in exact alignment with the
laminae, thus producing genuine four-point stabilization (4).
If the rods are bent in order to be in alignment with the
lordosis, under load, however, the rods will "disalign" with
the lumbar spine lordosis (10).
Evidently, it is possible to produce additional
instability in vitro in the unstable lumbar spine: by
distraction with the conventional HARRINGTON system even
with contoured rods -, since hooking of the fracture is
eliminated.
The multisegmental sublaminar wiring according to LUQUE
also cannot prevent this completely, because the rods will
rotate under the wiring. This is confirmed by findings on the
torsional stability of the unstable lumbar spine instrumented
by the original Harrington distraction system as compared
with its combination with multisegmental sublaminar wiring
according to LUQUE: when testing torsional stability only
part of the force required for testing flexion stability can
be applied because of the high primary rotational instability
of this construction (7).
A high stability and rigidity in the load range up to 15
Nm is achieved by the transpedicular anchored plate system
according to ROY-CAMILLE (8). However, if this system has the
advantage of a high degree of stability - it also has the
disadvantage of a very high rigidity, which means that only
little energy can be absorbed and that peak loads occur in
the screwed connections. There is another disadvantage which
lies in the fact that reduction by the implant itself is not
possible; therefore, the spine has to be reduced by suitable
means before the plates can be fi xed. As our results show,
anchoring strength is reduced by 50% if the system is limited
to fixation across three segments only; thus short distance
plating c~nnot be recommended for unstable injuries without
anterior support, unless sufficient transpedicular cancellous
bone-graft is applied or reinforcement of the anchoring
strength of the plate by two screws in each lamina is carried
out (10).
The Fixateur externe according to a MAGERL is a frame
construction with stable angles which is nevertheless
elastic. Depending on the material properties of the Schanz-
srews, loading of the non prestressed Fixateur externe system
results in a linear diflection in their elastic range. With
the prestressed system, deflection only occurs if the initial
stress is exceeded; a return to the initial position occurs
 509

in the elastic range. Experimental tests of the anchoring

strength of the Schan2-screws showed that they did not
loosen, even if the bending loads applied to the specimen
spine exceeded 60 Nm (6).
The Fixateur interne according to DICK is a sinkable
implant; in the mounted state there is no flexibility between
the longitudinal support and the shortened and
transpedicularly fixed Schan2-screws, therefore, like with
the Fixateur externe, no further bone supports are required
for its functioning (1). Consequently, also this system can
be used independently of the condition of presence of the
verebral arches, joints and ligaments. Distraction,
compression as well as neutral fixation are possible with the
threaded rod. Unlike the HARRINGTON rod, distraction does not
involve the simultaneous introduction of a kyphotic force
into the functional unit, because fixation of the Schan2-
screws to the threaded rods can be made lockable.
From a biomechanical point of view, the Fixateur interne
system shows characteristics similar to those of the Fixateur
externe system, but because of the short lever-arms of the
Schan2-screws it has the disadvantage of a more rigid
fixation in the vertebra, which may leads to loosening in the
vertebral body under cyclic load. Elasticity of the
Fixateur interne system is maintained until 40 NM bending
load, with plastic deformation occuring beyond this value.
But compared with the Fixateur externe, loosening of the
spinal anchoring occurs much earlier on account of the
Schan2-screws tilting in the vertebral spongiosa. In some
cases, loosening was demonstrated with loads between 20 and
40 Nm; however, satisfactory resiual stability was maintained
even with lossened Schan2-screws (1).
For our opinion, the superiority of the Fixateur externe
and Fixateur interne system lies in the fact that this high
degree of primary stability is created by the immobili2ation
of only two functional units, which is more favorable from
the functional point of view. Compared with the other
systems, the applicability of the Fixateur externe and
Fixateur interne systems is more comprehensive, because of
their adjustability on all planes and their independence of
the ligaments and the articular processes.
The different result of the short plate as compared with
the Fixateur interne system demonstrates the advantage of a
firm connection between vertical and sagittal supports.

SUMMARY
To determine the degree of torstional stiffness achieved
by several procedures, six different sorsal fixation
procedures (ROY-CAMILLE~plates with 1+1 and 2+2 bridging,
Fixateur externe after MAGERL, Fixateur interne after DICK,
HARRINGTON's distraction-rod system and the combination with
multi segmental sublaminar wiring after LUQUE) were carried
out on cadaver lumbar spines. Before fixation, the original
lumbar spines and the stability-reducing effect of various
standardi2ed lesions of discoidal, ligamentous, and ossoeous
structures were examined, the anterior spinal ligament being
preserved in each case.
510

We drew the following clinically relevant conclusions:

1- Rotation of the lumbar spine is not only dependent on
osseous but also on discoligamentous integrity; total
rotational instability is only produced by osseous
lesions.
2- Fracture stabilization and bony fusion of the lumbar spine
is most effectively achieved with transpedicularly
anchored systems.
3- Of the transpedicularly bonded stabilization systems, the
external fixator and the internal fixation plate with 2+2
bridging had the highest degree of rotational rigidity.
4- Distraction with the Harrington system may cause
additional loss of rigidity in completely unstable lumbar
spine.
5- Additional multisegmental cerclage fixation does not
increase torsional rigidity.
6- If decision is taken in favor of these procedures,
external immobilization is required postoperatively as an
additional measure.

REFERENCES
1. Dick W: Innere Fixation von Brust- und Lendenwirbel-
frakuren. Aktuelle Probleme in Chirurgie und Orthopadie,
Band 28. Hans Huber, Bern Stuttgart Toronto, p. 56 (1984)
2. Farfan HF: Mechanical disorders of the low back.
Philadelphia, Lea & Febiger, pp. 74+200 (1973)
3. Farfan HF: The torsional injury of the lumbar spine. Spine
9, 53 (1984)
4. Jacobs RR, Asher MA, Snider RK: Thoracolumbar spinal
inj uries: A comparati ve study of recumbent and operati ve
treatment in 100 patient. Spine 5, 463 (1980)
5. Luque ER: The anatomic basis and development of sequental
spinal instrumentation. Spine 7, 256 (1982)
6. Magerl F: Stabilization of the lower thoracic and lumbar
spine with external skeletal fixation. Clin. orthop. 189,
125 (1984)
7. Munson G, Satterlee C, Hammond S, Betten R, Gaines RW:
Experimental evaluation of HARRINGTON rod fixation
supplemented with sublaminar wires in stabilizing
thora.columbar fracture-dislocations. Clin orthop. 189, 97
( 1984)
8. Roy-Camille R, Saillant G, Berteau D, Salgade V:
Osteosynthesis of thoracolumbar spine fractures with metal
plates screwed through the vertebral pedicles. Reconst.
Surg. Traumatol. 15, 2 (1976)
9. White III AA, Panjabi MM: Clinical biomechanics of the
spine. Li ppincott, Philadelphia, p.79 (1978)
10. Worsdorfer 0, Magerl F, Schlapfer F, Perren SM:
Vergleichende Untersuchungen zu.r Stabilitat verschiedener
Fixationssysteme der lumbalen Wirbelsaule. In: Biomechanik
der Wirbelsaule, Thieme, Stuttgart - New York, p. 155
( 1978)
 511

POSSIBILITIES AND RESULTS OF TOTAL LUMBAR DISK REPLACEMENT

K. Schellnack and K. Buttner-Janz
Department of Orthopaedics. Humboldt-University. School of
Medicine. Berlin. GDR

1. MATERIALS AND METHODS

Based on findings that could be obtained from patent
specifications and publications (Medlars Computers, Stock-
holm), as well as on ideas of our own, a modular type of
endoprosthesis for the lumbar disks was developed after a
series of experimental studies carried out in 1983 and 1984 at
the Department of Orthopaedics, Charite Hospital.
The material combination used was UHMW-polyethylene (Chirulen)
and a I-millimeter titanium sheet, instead of which a
UR X2 Cr Ni Mo N 18.12-steel sheet of the same strength was
initially applied. From the material sheets two centrally
metal-plate bearing shells to be fixed without cement were
each punched. These force into the basic and upper plates of
the vertebral body with the teeth of their cranked edge, and
articulate with each other via the spherically shaped poly-
ethylene core. The size of the endoprosthesis was determined
from the mean values of the internal diameters of the basic
and upper plates of the lumbar vertebrae. The height of the
plastic cores corresponded to the different thickness of the
original disks and intervertebral spaces and was planned in
two sizes (8 and 11 millimeters). To prevent dislocation, the
polyethylene core was equipped with a stabilizing as well as
a guiding gear rim jutting out over the sides of the bearing
shells. Of the latter, two versions have been produced: a
circular shell with five or eleven teeth and a bearing shell
with additional lateral wings designed to enlarge the suppor-
ting area in the case of bone structures that are not capable
of carrying maximum load. A prototype made im casting technique
from non-ferrous alloys is in preparation.
So far, the disk endoprosthesis described herein has been
implanted in thirty-one patients aged twenty-seven to fifty-
five years, among them thirteen males and eighteen females.
The follow-up period is ranging from two to twenty-two months.
The fourth lumbar segment was replaced by an endoprosthesis
seventeen times, the fifth segment seven times, the third
segment two times and the second segment one time. In four
patients two segments were replaced simultaneously. According
to conventional principles of spinal surgery the approach was
extraperitoneally from ventro-lateral using the typical in-
strumental technique.
512

Fig. 1: Disk endoprosthesis with bearing shells and five

teeth with wings in model-position.
2. RESULTS
During the experimental stage, the endoprosthetic mod-
els for the lumbar disk were subjected in the servo-hydrau-
lic laboratory to static, axial as well as dynamic load tests
based on physiological angular degrees. Besides, their
functioning was tested on cadaver vertebrae.
While the metal components showed no deformation at all within
the load limits up to 20 000 newtons, typical surface and
material changes took place in the polyethylene cores. Up to
load limit of plane (max) (F) - 4 000 newtons the diagrams
showed the hysteresis typical of plastic materials. However,
after unloading no permanent deformations were seen. From the
slopes of the curves, the stiffness of the prosthesis used
could be determined. This stiffness increased with growing
load, since the loadcarrying surfaces extendes due to the
pressing-in of the polyethylene into the bearing shells, thus
being remodelled. First permanent deformities appeared in the
plastic core at loads up to 10 000 newtons. These consisted
in an irreversible reduction of height of about 10 per cent
which was due, above all, to material sintering; there was no
considerable progression to be noted in further cycles.
 513

Probe 2 3 4 5
4

VB: 0,32 kN/!>

0 Q5 1,0 2,0 3;0 AS[mrrj]

Il<~
11
1 1.345. Zyklu!>
10

3 Vs= 0,41 kN Is

2,5 3,0

Figs. 2 through 3: Load tests of the polyethylene core

Fig. 2: Load limit 4 kilonewtons,
Fig. 3: Load limit 11 kilonewtons.
On the load levels up to 20 000 newtons distinct deformities
were seen to take place which, finally, led to the destruction
514

of the plastic core.

Cadaver vertebrae were studied in two experimental procedures:
firstly, totally dissected lumbar vertebrae with osteoporotic
or osteomalacic structures, and, secondly, segments with
intact capsular and liga-
mentous structures whose
vertebral bodies had
sufficiently stable basic
and upper plates, were
tested in load cycles.
In the first experi-
mental series the disk
endoprosthesis immigrated
into the bone at 1 500
newtons. In the second
series it was only at
8 000 newtons, when the
bearing shells immigrated
into the basic and upper
plates, that lesions
appeared which limited the
physical load capacity of
the vertebral system.
Fracture of the vertebral
body that could be seen
visually, was then reaches
at loads exceeding 10 000
newtons.
In the clinicalappli-
cation stage, the follow-
up examinations of the
thirty-one patients carried
out after two twenty-two
months showed regression
of prevailing motor and
largely also sensory nerve
disorders. After three
weeks of hospitalization
the patients were dis-
charged for out-patient
aftercare. Some of them
have been rehabilitated
occupationally. In one
case, rehabilitation into
the work-process could be
achieved following tem-
porary disablement.
The results of the clini-
Figs. 4-A through 4-B: cal and radiographic
4-A: Cadaver vertebrae with disk examinations have con-
endoprosthesis firmed the experimental
4-B: Fracture of the vertebral data both as to the ex-
body reaching the load pected changes in material
limit of 10 kilonewtons and the distance stabili-
zation and functional
 515

capacity of the corresponding vertebral segments.

Our present scheme for the surgical treatment of lumbar disk
lesions and advanced degenerative lesions of the vertebral
segments envisages the following indications and contra in-
dications for the implantation of an endoprosthesis:

Fig. 5: Three months after disk replacement in the fifth

intervertebral space.
Indications:
- Advanced degenerative lesions of the intervertebral disk
accompanied by pain caused by the nor rowing of the foramina
intervertebralia and by a facet syndrome
- therapy-resistant pain in the case of segmental instability
of the lumbar spine except changed positions of the ver-
tebral bodies in spondylolysis
- segment-specific post-nucleotomy syndrome with or without
compression of nerve roots.
Contraindications:
Absolute:
- Pronounced osteoporosis or osteomalacia
- previous or present inflamation of the intervertebral disk
and/or the adjacent lumbar vertebrae.
Relative:
- ~Brrow spinal canal~
- spondylolisthesis
- advanced spondylarthritis
- lack of motivation and willingness to cooperate on the part
of the patient.
Despite the relatively short follow-up period limiting the
prognostic information on the therapeutic value of .disk endo-
516

prostheses, the experience we have gained so far is promising

and provides the basis of future research works. We are of the
opinion, that, as the total replacement of joints has shown,
only ten years of follow-up will finally show the advantages
or disadvantages of this method.
4. CONCLUSIONS
Isolated disk lesions with or without radicular signs
and symptoms, and segmental instabilities of the spine with
their objective secondary disorders, cannot always be
successfully treated to restore normal functioning, with
conservative as well as conventional surgical methods.
Indications for dorsal nucleotomy in acute neural or spinal
compression syndromes caused by disk lesions will continue to
exist without restriction.
In spondylolisthesis, advanced spondylarthritis and the
clinical symptoms of a narrow spinal canal as well as in the
inflammatory processes of the spine there is no alternative
method of treatment to ventral or dorsal fusion. In these
cases, too, osteoporosis or osteomalacia will set a limit
to the therapeutic result being highly responsible for the
failure of this method due to graft sintering and the non-
union of vertebral bodies.
The therapeutic gaps indicated herein can be filled with the
implantation of an endoprosthesis as a new and successful
version of treatment, which combines the short peri odes of
postoperative after-care as in nucleotomy with the distraction
and stabilization advantages of interbody fusion providing
and additional functional aspect. This therapeutic method will
be used especially to treat segmental instability associated
with pseudo-radicular symptoms - except structurally -
conditioned spondylolistheses - as well as segment-specific
post-nucleotomy syndromes. It is well feasible, that the range
of indication be extendes. Experimental and first clinical
experiences indicate that there might be a further improvement
of this method to include spondylodesis as well. Using a
different support at the vertebral body and special implan-
tation techniques the endoprosthesis of the "S8+Charit~"
modular type as produced by us can also be applied in the
case of structurally reduced basic and upper plates as well
as such plates loaded only partially as far as their
functioning is concerned.
+~chellnack/~Uttner

REFERENCES
1. Fassio, S., and Ginestie, I.-F.: Proth~se discale en
silicone. Etude experimentale et premi~res observations
cliniques. Nouv. Presse Med. 21, 207, 1978
2. Fernstrom, U.: Arthroplasty with intercorporal Endopros-
thesis in Herniated Disc and in Painful Disc. Acta Chir.
Scandinavica 357, 154 - 159, 1966
 517

THE ROLE OF THE ALAR AND TRANSVERSE LIGAMENTS FOR THE

STABILITY OF THE CRANIO-CERVICAL JOINT
E. Schneider*, J. Dvorak**, D. Wyder* and S.M. Perren*
M.E.M.-Institute for Biomechanics*, Department of Neurology**,
University of Bern, Bern, Switzerland

1. INTRODUCTION
In the treatment of spine fractures, stabilization of the
vertebral column is one of the major objectives. In the
cranio-cervical joint, the alar and transverse ligaments
provide much of the stability of the healthy spine and their
relative contribution seems important both for the diagnosis
of pathology and subsequent, selective stabilization.
Although the anatomy of these ligaments seems well
described (10), the literature with respect to the alar
ligament is not consistent. Whereas the majority of authors
(2, 5, 16) state that the alar ligaments attach laterally at
the occipital condyles only, Cave (4) and Ludwig (11) found
that the anterior portion of the alar ligaments attach via
their inferior fibres anteriorly at the lateral masses of the
atlas. This situation was confirmed in the present study and
is shown in Fig. 1. The alar ligament consists of two
portions, one connecting the dens with the condyles of the
occipital bone and the other inserting on the lateral mass of
the atlas. Its function is to restrict axial rotation,

Fig. 1: Anatomical sketch of the atlas (1) and dens axis (2)
from above, showing the atlanto-alar (3) and the
occipito-alar ligaments (4) as well as the transverse
(5) and the anterior atlanto-dental ligaments (6).
518

sidebending and flexion/extension. The transverse ligament

holds the dens against the anterior arch of the atlas. It has
an upward prolongation to the occiput and a caudal extension
to the posterior surface of the body of the axis, forming the
cruciform ligament of the atlas. Its function is to restrict
flexion of the head as well as anterior displacement of the
atlas.
The incidence of fractures or fracture-disc locations in
the cervical spine has been shown in a retrospective study of
100 patients to be as high as 19% for the levels C1/C2 (1).
Among these, 17 fractures or non-unions of the odontoid were
found. It has been further shown (6), that up to one third of
the patients from motor vehicle accidents having a principal
diagnosis of cervical strain is receiving medical care due to
continuing pain.
The literature on the biomechanics of the cervical spine
has recently been reviewed (9), revealing that most of the
work done in this area centers around impact tests using
intact cadavers for estimation of human tolerance values. Few
investigators focused their attention on single segments in
the cerv ical spine (3, 8, 13, 14, 15) and in only three of
them, the levels C1/C2 were included. Fielding and co-workers
(8) determined the strength of the transverse ligament by
displacing the atlas anteriorly in a transverse plane
relative to the axis and thereby loading the transverse
ligament physiologically. They found failure loads generally
ranging from 400 N to 1100 N with extreme values as low as
120 N or as high as 1800 N. Within the limits of 12 mm of
excursion of the odontoid relative to the axis selected for
these tests, the alar ligaments did not fail. Spence et al.
(15) tested the transverse ligaments by subjecting them to
pure tensile loads and found breaking forces of 380 N to
1040 N in 10 specimens. Pintar et al. (14) measured strength,
failure deflection and energy to failure of the cervical
ligaments in 30 specimens, including the alar but not the
transverse ligament. They found the vertical cruciate
ligament to be strongest (436 N) and the ligamentum flavum
weakest (113 N) with values for the other ligaments between
these extremes. It has been suggested that a lesion of the
alar ligament may cause rotatory instability of the upper
cervical spine (7). With regard to a possible failure
mechanism it could be speculated, that the alar ligament ~ay
rupture in flexion-extension, if the trauma occurs while the
head is rotated to one side, which has been proposed as a
mechanism of injury by Ommaya (12).
Because of this limited information available on strength
and an apparent lack of data on the kinematics of the upper
cervical spine, the goal of this study was to estimate the
contribution of the alar and transverse ligaments to the
stability of the cranio-cervical region by experimental
determination of the range of spinal motion and the strength
of both ligaments under pure axial loading.

2. METHOD
All specimens used in this study were harvested from fresh
 519

cadavers, sealed in polyethylene bags and stored at -25 0 C.

Prior to testing, they were thawed. For the kinematic study,
11 specimens with a mean age of 58 years were used. The
spines were fixed in a radiologically transparent cube and
loaded manually in axial rotation until motion was
significantly inhibited. Images taken by computed tomography
were used to determine the range of motion present in the
occipito-atlantal and the atlanto-axial joint with intact and
sectioned left alar ligaments.
The spines of another 7 subjects with a mean age of
76 years were used for mechanical testing. They were
carefully prepared by freeing the alar and transverse
ligaments of all surrounding soft tissues. The capsule of the
anterior atlanto-dental jOint, the atlantal portion of the
alar ligament as well as the anterior atlanto-dental ligament
were cut. The occipital bone was cut to just retain the
insertions of the alar ligaments and also the posterior arch
of the atlas was removed. For testing of the alar ligaments,
the spinous process of the axis was clamped onto a support
plate connected to the actuator of an INSTRON testing machine
by means of a steel bar for the vertebral body and a cable
for the anterior tubercle. The remnants of the occipital bone
were held by steel cables, connected through cylindrical
bearings to the frame of the testing machine in order to
ensure pure axial loading. The mounting of the transverse
ligaments consisted of two steel cables wrapped around the
arch on each side of the insertion of the ligaments. Axial
load was applied with a cross-head speed of 1.5 mml sec and
load and displacement were recorded on an X-Y-recorder. Prior
to mechanical testing, the lengths of the ligaments were
measured between center-points of insertions, the diameters
were determined by caliper in the middle of the alar
ligaments and where the transverse ligament touches the dens.
Statistical analyses were based on Student's t-test at a
level of significance of 0.05.
3. RESULTS
In the anatomical measurements, the alar ligaments were
found to have a length of 10.8 mm (range: 8-12 mm), with the
caudal side being slightly longer (11.3 mm) than the cranial
one (10.3 mm). No detectable difference was found between
left and right. The transverse ligaments had a length of
20.1 mm (16-23 mm). The cross-sectional averages of the alar
ligaments were 22 mm 2 (16-28 mm 2 ) and of the transverse
ligaments they were 17.8 mm~ (14-24 mm 2 ). The ligamentous
connection between the dens and the lateral mass of the atlas
could be identified in all but one specimen and was app. 3 mm
long. However, it was not possible to include it in the
strength measurements due to its different orientation and
short length.
The motions found in the CT-measurements are represented
graphically in Fig. 2. The average motion between CO and C1
was 4.4 0 to the right and 5.9 0 to the left. Between C1 and C2
these were 31.4 0 and 33 0 respectively. After sectioning of
the left alar ligament, the rotation CO/C1 increased to 9.4 0
S20

and 7.8 0 and the rotation C1/C2 changed to 37.2 0 and 33 0

respectively. The total increase of rotation between occiput
and axis to the right was 10.5 0 or 30% of the initial range
of motion (36 0 ).

CO/C1 C1/C2
Fig. 2: Graphic representation (mean and range) of the
motion found in 11 cadaver specimens between
occiput/atlas and atlas/axis before (open, inner
segment) and after sectioning (dotted, outer
segment) of the left alar ligament.

The tensile strengths (N) of the alar and transverse

ligaments together with age and sex of the specimens are
summarized in the following table:

Specimen Alar ligament Transverse ligament Age Sex

Nr. L R

1 190 + 330 45 f
2 + + 330 87 m
3 *290 *290 400 82 f
4 260 150 290 84 f
5 270 250 260 59 m
6 70 160 700 86 m
7 190 230 170 90 m

Average 212 198 354 76

* Congenital malformation with two additional odontoid bones
+ Missing values due to damage during preparation

No difference between left and rigth alar ligaments could be

found. Failure always occured at the insertions of the
ligaments in the area of the bone/cartilage transition zone.
 521

4. DISCUSSION AND CONCLUSIONS

The methods selected for this study have been appropriate
to provide information on the alar and transverse ligaments
and their importance for the stability of the cranio-cervical
region. Findings by Cave (4) and Ludwig (11) about the
presence of an atlantal portion of the alar ligament were
confirmed. In addition, a ligamentous connection between the
dens and the anterior arch of the atlas, functionally a
synergist of the transverse ligament, was found. It must be
noted, however, that due to their small size, none of these
ligaments could be similarly tested. The total contribution
of the alar ligaments might therefore be underestimated.
Al though the kinematic measurements by means of CT-images
were performed under non-standardized loads, the results were
found to be reproducible and the results were different
enough to sUbstantiate the relative importance of the alar
ligament for the stability of the upper cervical spine.
Biomechanical testing of the ligaments included the
determination of strength, but not elongation, energy or any
viscoelastic property. Strain properties have been published
by Pintar et al. (14). A relatively small number of specimens
was used for the mechanical tests because availability of
specimens was limited and freeing of the ligaments for
testing requires delicate and time-consuming dissection. The
differences in the cross-sectional area of the alar and
transverse ligaments with the stronger ligament having a
smaller cross-section were unexpected.
The results of the strength measurements of the transverse
ligaments compare well with those reported by Spence et al.
( 15), who used ax ial loading, and even with those found by
Fielding et al. (18), who tested the ligaments under loads
perpendicular to the main direction of the ligament. Based on
our data, the wide variation of strengths (also found in
other studies) cannot be explained by differences in the age
of these specimens, but rather by other parameters as
mobility present in C1/C2 or the earlier state of activity of
the patient.
Generally, the strength of the alar ligaments was less
than the strength of the transverse ligaments, but within the
range of extreme values reported by Pintar et al. (14).
According to our findings it can be stated, that the alar
ligament is an important restraint to excess axial rotation
between CO/C1 and C1/2. The use of absolute numbers to
establish limits of instability remains difficult because of
the wide range of values encountered in individual specimens.
This is due to anatomical variations of the bony and
ligamentous structures as well as to degenerative changes.
Our results would support the notion of these ligaments being
a "second line of defense" (8). The question then is, whether
they are more important with respect to rotational or to
sidebending types of trauma or both types as related to
flexion and extension of the head. The relatively low
strength of the alar ligaments as compared to the weight of
the head emphasizes the important function of the
suboccipital musculature. Injury may be exacerbated in cases
522

of unexpected indirect trauma of the cervical spine, when the

muscles are relaxed.
REFERENCES
1. Aebi M., Mohler, J., Zaech, G.A. and Morscher, E.: Indi-
cation, surgical technique and results of 100 surgically-
treated fractures and fracture-dislocations of the cervi-
cal spine. Clin. Orthop. Rel. Res. 203: 244-257, 1986.
2. Arnold, I.: Handbuch der Anatomie, Bd. 1, 1845.
3. Bauze, R.J., and Ardran, G.M.: Experimental production of
forward dislocation in the human cervical spine. JBJS
60-B( 2): 239-245, 1978.
4. Cave, J.E.: On the occipito-atlanto-axial articulation.
J. Anat 68: 416, 193311934
5. Daniels, D.L., Williams, A.L., and Haughton, V.M. :
Computed tomography of the articulations and ligaments at
the occipito-atlanto-axial region. Radiology 146:
709-716, 1983.
6. Dvorak, J., Schmid, S. and Valach, L.: Verletzungen der
Halswirbelsaule in der Schweiz, eine retrospektive Studie
tiber 7 Jahre (in preparation).
7. Dvorak J., Panjabi M.M., Wichmann M., and Gerber M.: CT-
functional diagnostics of the rotatory instability of
upper cervical spine, accepted for publication in Spine.
8. Fielding J.W., Cochran G.V.B., Lawsing III J.F., and Hohl
M.: Tears of the transverse ligament of the atlas. J.
Bone Joint Surg. 56-A, 8: 1683-1691, 1974.
9. Huelke, D.F., and Nusholtz, G.S.: Cervical spine
biomechanics: a review of the literature. J. Orthop. Res.
4: 232-245, 1986.
10. von Lanz T., and Wachsmuth W.: In: Praktische Anatomie
lB: 330-354, Springer Verlag Berlin-Heidelberg, 1979.
11. Ludwig K.: liber das ligamentum alare dentis. Z. Anat.
Entwickl. Geschich. 116, 442, 1952.
12. Ommaya, A.K.: The head kinematics and brain injury
mechanisms. In: AIdman, B. , and Chap on , A. : The
biomechanics of impact trauma. Elsevier Sciences
Publishers B.V., 117-132, 1984.
13. Panjabi, M.M., White, A.A., and Johnson, R.M.: Cervical
spine mechanics as a function of transection of
components. J. Biomechanics 8: 327-336, 1975.
14. Pintar, F.A., Myklebust, J.B., Yoganandan, N., Maiman,
D.J. and Sances Jr., A.: Biomechanics of the upper
cervical spinal ligaments. Proc. 9th ASB, Ann Arbor MI,
J. Biomech 6:469, 1986.
15. Spence, K.F., Jr., Decker, S., and Sell, K.W.: Bursting
atlantal fracture associated with rupture of the trans-
verse ligament. J. Bone Joint Surg., 52-A: 543-549, 1970.
16. White, A.A., and Panjabi, M.M.: Clinical biomechanics of
the spine. Lippincott, Philadelphia, 1978.
ACKNOWLEDGEMENT
The authors wish to gratefully acknowledge the expertise
and help of Mr. D. Wyder during the performance of the
mechanical testing.
 523

3-D AUTOMATIC POSTURE ANALYSIS FOR EVALUATION OF BRACE EFFECTS

ON SPINAL DEFORMITY
R. Assente. G.C. Santambrogio and R. Vigano
Centro di Bioingegneria. Politecnico di Milano. Milano. Italy

1. INTRlJOOCI'ION
Dud ng the last few years, several studj es (l, 2, 3) have
been devoted to non-jnvasjve and non-jonjzjng analysjs of spjnal
deforrnjty. Such approach becomes partjcularly useful for large scale
screenjng or repeated controls on the same patjents where the need js
to avojd X-ray exposures or to reduce them dudng the course of the
djsease and jts treatment.
The ajm of thjs work was to apply a new experjmental method for the
evaluatjon of brace effects jn scoljotjc patjents. Such technjque cambjnes
basjc features of the most oommonly used non-jnvasjve topograms wjth
jnformatjon on load djstrjbutjon durjng posture.

2. INSTRUMENTATION AND EXPERIMENTAL PROCEOORE

Fjgure 1 shows a scheme of the equjpment adopted.
Patj ent to be tested was fj tted wj th small hemj spheres, coated wj th
reflectjve paper, markjng sujtable anatomjcal landmarks suggested by
cljnjcal semejotjcs: lobes, lateral extremjty of the acromjons, sterno-
clavjcular jojnts, anterjor superjor jljac spjnes (SIAS), posterjor
superjor jljac spjnes (SIPS), lateral jojnt space of the knees, lateral
malleolj and spjnous processes begjnnjng from C7 down to S3 on every
second vertebra. Subsequently to that preparatjon, takjng not more than
ten mjnutes, the subject was asked to stand jn orthostasjs on a force
plate sensed wj th a transducer cell j n each corner. Each cell generates an
analog sjgnal proportjonal to the rate of the subject's wejght actjng on
jt. Whjle he was stjll standjng on the force plate, two pajrs of TV
cameras pj cked up hj s front and back vj ews j n order to perrnj t the
measurement of the 3-D coordjnates of each marker. Both the analog sjgnals
fran the force plate and the TV sj gnals fran the cameras JOl n j nto a
specjal devjce, called AUSCAN (AUtanatjc SColjosjs ANalyser), whose
general scheme js jllustrated jn Fjg. 2 (4, 5). The force plate sjgnals,
after convert jon jnto electrjcal voltages and ampljfjcatjon, are sent to
an AID converter and djgjtjzed at a sampljng rate of 50 Hz each channel.
The TV sjgnals are jnput to a multjplexer unjt whjch feeds the Fast
Processor for Shape Recognj tj on wj th a frame every 20 ros thus scannj ng the
four cameras jn 80 ros. The resultjng sampljng rate js 12.5 Hz whjch js
adequate to assure an accurate measurement of postural kjnematjcs. The
FPSR allows the recognjtjon of markers of pre-declared shape fran the
orjgjnal TV frame and the canputatjon of thejr coordjnates. Marker
recognjtjon js essentjally based on a hardware-jmplemented real-tjme
bjdjmensjonal cross-correlatjon between the subject's TV jmage and a
spedal functjon assocjated wjth marker shape. The orjgjnal TV jmage from
each camera and jts related vjdeo map of the recognjsed markers can be
djsplayed on two monjtors for a preljmjnary control of the experjmental
524

TVI TV2

AUSCAN
FORCE
PLATE

TV4 TVa

HOlT
COMPUTER

FIGURE 1. Scheme of the equj ~nt adopted.

FROM Tv
CAMERAS TO MONITOR I

ADORESS LlftES TO MONITOR 2

FROM '---.---r-..... PROCESSED TV SIGNAL

HOST COMPUTER

FROM CHARGE AID

FORCE _ _ _~ AtlPLIFIERS 1-_ _" CONVERTER 1 - - - - - - - ,
PLATE

TO HOST
COMPUTER

FIGURE 2. Scheme of AUSCAN.

 525

set up before data are collected. The host canputer, where the coordjnates
of the markers and the force plate sjgnals are cambjned, performs the
followjng tasks: acqujsjtjon, caljbratjon, marker classjfjcatjon,
barycentrjc vector canputatjon, absolute 3-D reconstrucbon and graphjc
representatjon. ACqujsjtjon js performed jn real-tjme and js merely the
data collect jon and storage. Caljbratjon js requjred jn order to correct
optjcal djstortjons and non-ljnearjtjes jn the deflect jon cjrcujts of the
cameras. Such phase consjsts of an acqujsjtjon of a fjxed grjd of markers
and the estjmatjon of correctjve coeffjdents to transform the djstorted
coordjnates jnto undjstorted ones. Marker classjfjcatjon js based on a
model of the whole body and a sortjng procedure. The force plate sjgnals
are processed to obtajn the ground react jon force jn vectorjal form (6).
Sjnce jnertj al canponents durhlQ posture are negljgjble wj th respect to
gravjtary canponents, the ground react jon vector matches the barycentre of
the body thus pojntjng out the ljne around whjch the load js sjrnnetdcally
dj std buted.
The 3-D coordj nate space j ntersectj on j s performed by consj ded ng the
appljcatjon pojnt of the barycentrjc vector as the absolute coordjnate
frame odgjn. Representatjon of the results can be obtajned jn numedcal
and graphjc form by usjng a prjnter and a plotter.

3. RESULTS
In order to show an example about the analysjs of brace effects on
scoljotjc patjents, the results fran one case among the 17 examjned are
reported. The patjent js a gjrl, 17 years of age, wjth a djagnosed
progressjve jdjopathjc scoljosjs charactedsed by a double curve; the apex
of the lumbar left convex curve j s L 2 (rotatj on +++) whj le the apex of
the thoracjc rjght convex curve js T9 (rotatjon +). The Cobb angles are 37
and 40 degrees for the lumbar and thoracjc curve respectjvely.
Fjgure 3 shows the results obtajned before wearjng a brace. The patjents,
jf canpared wjth normal subjects (4), present evjdent asjrnmetry of the
whole body wj th respect to the barycentrj c vector whj ch j s much closer to
the rjght than to the left sjde. Thjs js due to a djsplacement of pelvjs
toward the rjght as demonstrated by the x coordjnates of the SIAS.
Thoracjc canpensatjon js pojnted out by the posjtjon of the acranjons
whj ch are dj splaced 3.5 rnm left. The spj nal morphology and the dj stance of
the markers placed on the processes confjrm these comments. Partjcularly
jnterestjng js the top vjew where rotatjons of pelvjs and shoulders wjth
respect to the feet are represented. Begjnnjng fran an angle of 4 degrees
between pelvjs and feet, the canpensabon assocjated wj th the thoradc
curve ljmjts the rotatjon of shoulders wjthjn an jncrement of just one
degree. On the same day the patjent wore for the fjrst tjme a MHwaukee
brace. Durjng the followjng perjod the analysjs gave results wjdely
unsatjsfactory and characterjsed by a hjgh varjabjljty but, about fjfty
days later, the brace effects got to be sjgnjfjcantly canparable. Fjgure 4
shows the results taken fran the same patjent whjle wearjng her brace. The
bme jnterval between them and those jn Fjg. 3 js two months. It js to
be noted that those markers plaCed on the last two processes and on the
SIPS are not represented jn the back vjew because of the presence of the
brace.
Although the spjnal curves appear to be less evjoent subsequent to
correct jon jmposed by the brace, a certajn asjrnmetry of the whole body
js stjll present. The djstances of the SIAS fran the barycentrjc vector,
even jf not perfectly balanced, show a recovery assocjated wjth pelvjc
adjustment whjle the upper canpensatjon stjll keeps the shoulders
djsplaced toward the left. Fran the top vjew the reduced angle of pelvjc
 VI
N
a.
Z I Z

....

.I -..
NAME Z X
-.
~ X Z NAME
NAn: Z X X Z NAME
RL06 1695 75 -84 1693 LLOB PROC 1581 -12 211 1510 RACR
RelV 1518 20 -29 1519 LCLV PROC 1534 -4 4 14<16 PROC
RACR 1510 211 -218 1490 LACR PROC 1493 0 6 1392 PROC
RSIA 1076 123 -110 1077 LSIA LACR 1490 . -216 2 134J PROC
RKNE 539 130 -149 536 LKNE PROC 1291 -II 53 1091 R51P
RMAL 68 103 -126 69 LMAL PROC 1216 -12 5 1062 PROC
PROC 1146 -4 5 1026 PROC
LSIP 1093 -56 130 539 RKNE
LKNE 538 -149 103 68 RMAL
LMAL 69 -126

.. ..

..
x x
FRONT VIEW BACK VIEW

y
~
51AS X,Z FOR I---t 10 em
X FOR I---t I em
x Z FOR I---t 10 em
ACROMIONS
MALLEOLI

~ MALLEOLI-SIAS 4'
FIGURE 3. Basal results concernjng a female
~ MALLEOIl-ACROMIONS 5' scoljotjc patjent.
+ TOP VIEW
 z IZ
I
f
--,- - .. \, .
NN1E Z X
- - X Z NAME NAME Z X X Z
1577
NAME
PROC 1663 -3 201 RACR
Rl06 1770 55 -100 1766 LLOB
PROC 1605 -2 2 1534 PROC
RCLY 1594 2 -46 1599 LCLV
LACR 1571 -234 4 1460 PROC
RACR 1577 201 -234 1571 LACR
PROC 1364 0 3 1419 PROC
RSIA 1076 98 -94 1077 LSIA
PROC 1305 -3 0 1146 PROC
RKNE 538 119 -135 539 LKNE
PROC 1226 -4 119 536 RKNE
RMAL 69 93 -112 69 LMAL
LKNE 539 -135 93 69 RMAL
LMAL 69 -112
- - 1
., @
.. -
., <il
x
x BACK VIEW
fRONT_VIEW

~
SIAS
X,Z FOR J---4 10 em
_ACROMIONS X FOR J---4 I em
Z FOR J---4 10 em
X
MALLEOLI

~ MALLEOll-SIAS 2,6'
FIGURE 4. Results concerning the same patient v.
~ MALLEOlI-ACROMIONS' -2,5'
of Fig. 3 while wearing her brace. !::l
+ TOP VIEW
528

rotation confhms the improvement induced in the lumbar region whHe the
angle between SIAS and acrardons, increased to 5.3 degrees, indicates
thoracic unbalancing.

REFERENCES

1. WHIner S.: Moire' Topography for the Diagnosis and Ibcurnentation

of Scoliosis., Acta orthop. Scand., vol. 50, pp. 296-302, 1979.
2. Dacquino G., Divieti L., Muller and Sibilla P.: Thermography:
a New Aid in Scoliosis Diagnosis and Therapy., SICOT Congress, Kioto,
Japan, 1978.
3. Turner-Smi th A. R. and Hard s J D. : ISIS: An Autanated Shape
Measurement and Analysis System., Surface TOpography and Spinal
Deformi ty, Ed. by Harri s and Turner-Smi th, Gustav Fi scher Verlag
Publisher, pp. 31-38, 1986.
4. Assente R., Pedotti A., santambrogio G.C., Vi gano , R., Ambrosini A.
and Cometti A.: An Autanatic Non-Invasive System for the Analysis of
Spi nal Deformi ty., Surface TOpography and Spinal Deformi ty, Ed. by
Harris and Turner-Smith, Gustav Fischer Verlag publisher, pp. 61-70,
1986.
5. Assente R., Ferdgno G., Pedotti A., santambrogio G.C., Vi gano , R.:
AUSCAN System: a New Procedure to Evaluate Spinal Deformities.,
Biomechanics X, in press.
6. Pedotti A., Cametti A., Ambrosini A. and Santambrogio G.C.: A New
Device for Real-Time Analysis of Posture and Gait., INSERM, Human
Gait Analysis and Application, pp. 77-87, 1983.

This work has been partially supported by CNR P.F.T.B.S.

Thanks are also due to Centro di Teoria dei Sistemi.

ORTHOPAEDIC BIOMECHANICS: CLINICAL ASPECTS AND REHABILITATION
 531

STRIPPING OF THREAD IN BONE BY COMMERCIAL CORTICAL SElF-

TAPPING AND PRETAPPED SCREWS
X.X. Xu*. J. Cordey**. B.A. Rahn**. W.J. Ziegler*** and
S.M. Perren**
Bethune Med. Univ.*. Changchun. China; Swiss Res. Inst.**.
Davos; Ziegler's Inst. Plan. and Res.***. Basel. Switzerland

1. SUMMARY

In internal fixation, screws are used to compress a fracture and/or to

fix a plate to bone. In either case their function depends on strong and
sustained anchorage within the bone. Screws are usually inserted after
drilling a hole and tapping a thread. It would seem to be attractive though
to use self-tapping screws to simplify and shorten the operation. Se1f-
tapping screws, however, have drawbacks; the holding power may be infe-
rior, the thread-cutting flutes may become filled with bone so that more
torque is needed to insert or remove them, and during application with any
loosening and retightening, a new thread may be cut, damaging the first
thread with consequent loss of grip. The aim of this paper is to compare
the forces required to strip the thread and to uetermine the respective
holding power of the different screws.
Pretapped cortical screws and two designs of self-tapping screws of com-
parable dimensions were tested on human cadaveric tibiae. The amount of
torque needed to strip the bone thread and the axial force produced were
monitored during experiments comparing the different self tapping screws,
using the pretapped screw as a control. The results showed that the amount
of force needed to strip the bone thread by the pretapped screw was signi-
ficantly higher than with the self-tapping ones. This was more marked in
the distal cortical part of the tibiae than in the proximal cancellous
part.

2. INTRODUCTION

The cortical screws in common clinical use can be of the self-tapping

and pretapped kind. The self-tapping screw was first used by SHERMANN
(1921), and the pretapped one was first advocated by DANIS (1949). The
holding power after screw insertion is one of the important factors affec-
ting the results of fixing a fracture with a plate or with screws alone. Is
there any significant difference in holding power between self-tapping and
pretapped screws? There is still controversy about the methods to be used
in evaluating the advantages and short~omings of each. BECHTOL (1959)
showed that the flutes of the self-tapping screw could reduce the holding
power by 20-30% after fixation. WAGNER (1963) investigated the stability
that could be obtained by the pretapped and self-tapping screws using his-
tological techniques. He showed that after nine months the pretapped screw
was ensheathed in hard remodelled bone. The self-tapping screw by contrast
532

had produced microfractures and bone necrosis and tended to loosen after 23
days. KORANYI et al. (1970) and SCHATZKER et al. (1975) compared the hol-
ding power of both types of screw,but found no relevent difference. In
their investigations, VANGNESS et al. (1981) found microfractures after the
insertion of all the types of screw, but no significant difference in the
pull-out test between either kinds of screw. In these experiments, it can
be argued that the fluted end of the screws protruded through the bone and
could not contribute to the holding power.
The aim of the present experiment was to compare the holding power of
standard pretapped and self-tapping screws in bones of controlled quality.

3.1. Materials and instruments.

3.1.1. Bones. Twenty human tibiae, removed from 10 cadavers (8 males,

aged 18-74 years and 2 females, aged 51 and 64 years), were used. The bones
were not osteoporotic as shown by computed tomography and radiology (XU and
CORDEY, in preparation).

3.1.2. Screws. (Figure 1) The pretapped screws were the standard

stainless steel~O/ASIF cortical type (3.5mm screws type ISO , with
buttress thread, hexagonal impress in the head and 1.25mm pitch). The self-
tapping cortical screws were the Zimmer model (3.5mm screw with buttress
thread and Wooddruff head of smooth stainless steel) and the Howmedica
model C3. 6mm screw with machined thread and Wooddruff head, in vitallium
with a coarse surface).
The ancilliary instruments comprised the AO-ASIF 2.5mm drill bit and
corresponding 3.5mm tap, a Zimmer 2.8mm drill bit and Howmedica 2.8mm and
3.0mm drill bits. The screwdrivers were those produced by the same firms.

3.1.3. Measuring instruments. A RUMUL dynamometric screwdriver (a modi-

fied AO/ASIF type fitted with strain gages) was connected to a RUMUL
Tensicator bridge amplifier to measure the applied torque. A KISTLER 901A
load cell with adaptation washers was connected to a KISTLER 568 load
amplifier. This assembly was connected to a SCHLUMBERGER Sefram X2YT
recorder. The precision and accuracy of the measurements was better than 1N
for the axial force and 0.01N.m for the torque, limited by the reading of
the graphs.

FIGURE 1. Screws used:

A. Pretapped screw.
Z. Self-tapping screw with
buttress thread, 2-lip flutes.
H. Self-tapping screw with ma-
A z H chined thread, 3-lip flutes.
 533

3.2. Methods.

3.2.1. Preparation of the bones. The bones were stored at -20 0 C. Eight
test levels were chosen at the middle part of the bone (Figure 2). X-rays
and tomographs were taken at each of the 8 levels. Before the test the
bones were thawed in Ringer's solution at room temperature for four hours.

3.2.2. Measurements of torque and axial force. The experiment consisted

of inserting the screws and tightening them progressively until the bone
threads were stripped. The screws were matched on the medial and lateral
sides of each pair of tibiae to ensure that both kinds of screws had the
same amount of test at each position (Figure 2). The torque was applied ma-
nually by one of us (Xu), in a succession of tightening twisting movements
at low speed. Torque and axial force were monitored simultaneously on an x-
y diagram (Figure 3).

8 bOlle, 8 bones 4 bones

FIGURE 2. Testing levels for the

Z A H A H A different screws. Comparison bet-
A Z _ A __ H_ A_ H ween the pretapped screws (A) and
Z __ A _H A _ H __ A the different self-tapping screws
__ A Z _ A __ H A H (Z and H) and the drilling diame-
Z A __ H _ A _ H _ A ters (Group II, 2.8mm, Group III,
A Z __ A H A H 3.0mm) were made in sequential al-
_Z A H _ A_ H_ A ternating tests.
A Z .A_H A H
I J L J l J
I II III
Kg
150

100

--- -- 50

Med. M
0
0 0.5 1.0 1.5 2.0 Nm

FIGURE 3. Measurement of torque and axial force.

a. Experimental set-up. Using appropriate washers, each screw is
inserted to penetrate the cortical thickness. One cortex was used
in order to simulate the cortical thickness used for lag screws.
b. Typical chart showing force / torque relationship. The succes-
sive tightening twists are shown by the horizontal lines.
534

The experimentation was carried out in three groups:

Group I: A and Z group, 8 bones. The holes for A screws were drilled with
the 2.5mm drill bit and tapped with the 3.5mm tap; the holes for the Z
screws were made with the 2.8mm drill bit.
Group II: A and H1 group, 8 bones. The holes for A screws were prepared as
in Group I. A 2.8mm drill bit was used for the H screws.
Group III: A and H2 group, 4 bones. The holes for A screws were prepared as
in Group I. A 3.0mm drill bit was used for the H screws.
Following the test, cross-section specimens of different thickness were
taken through or near each pair of screw holes using a Leitz 1600 saw cut
machine. Radiographs of slices of 1 and 4mm were taken, and micror.adiogra-
phs of 100um slices to examine the bone structure and measure the thickness
and density of the bone around the screw holes.
4. RESULTS

(The term "stripping force" is used to indicate the amount of force that
must be applied to a screw to strip the threads from the surrounding bone).
Stripping force of the different screws. In Group I, the stripping force
of the pretapped A screws was 14.5% greater than that of Z self-tapping
ones (P 0.05). In Group II (Figure 4), the stripping force of A screws was
23% greater (P 0.01) than that of H1 self-tapping screws for which the
2.8mm drill was used. In Group III the stripping force of A screws was
26.5% greater (P 0.01) than that of H2 self-tapping screws for which the
3.0mm drill was used. No significant difference was observed for the
maximal torque of different screws.
A significant correlation has been established between stripping force
and computed tomographical density, (values given in "ISOTOM units", ie.
100 x the linear absorption coefficient for a X-ray energy of 27kV - 1125;
Figure 5.a.), as well as between stripping force and bone thickness (Figure
5.b). These correlations, however, are not high enough to justify a predic-
tion of the stripping force (RAHN et al., 1978)
Stripping Force H . - . D.6.Drill 2.8)
N A 0---0 (3.5. Drill 2.5)
3000 tSE
0 0 0
0
0

2000 .0

0
1000
0
0 0

0
P 2 3 4 4 3 2 0

FIGURE 4. Stripping force of the self-tapping

H screws and pretapped A screws in Group II.
 535

Stripping Force
Stripping Force
n- 64 N n#. 48
N
r. 0.7033 3000 r= 0.5604
3000

, '
, ,
'...e, t, : I
"0
0
,
0"

o 2000
2000
,,
,: II

.. "" , , , , ,
I
" .. . I

I I
.. '
I I I '
I

I
I
,
..

I
t.
I

1000
.
1000 "0

I I I
I
2500 30'00 IU 1.0 3.0 5.0 mm
1500 Thickness
Cross-Sectional Oensity

FIGURE 5. Correlation between stripping force and

a. radiological density measured using computed tomography or
b. cortical thickness. Values measured using the pretapped A
screws.

5. DISCUSSION

We think that the cutting flutes of the self-tapping screw are narrow
and short, as bone dust cannot escape during screw insertion, and that the
friction between the screw and bone tissue is much higher, especially in
those screws in which the manufacture had left a rough surface. These types
produce cracks in the bone, damaging the bone structure and compromising
the stability of the plate or screw fixation. If the fluted screw tip is in
the cortex, this too can reduce the holding power of the screw. The cutting
flutes of the tap used to prepare the thread for the pretapped screw are
wider and longer, with sharp cutting edges and a smooth surface. The bone
dust can be easily accommodated in the flutes and washed out when the tap
is removed with a squirt of irrigating fluid. The accurate cutting edges
also cause less damage to bony tissues.

In clinical practice, although the pretapping procedure occupies a few

minutes of operating time, the tapped tract for the screw insertion is much
better and more accurate. The tap should be cleaned every time before use.
After tapping, the screw can be more easily inserted and a higher holding
power can be expected.
536

6. REFERENCES
----------
Bechtol, C.O.: Internal fixation with plates and screws. In Bechtol,
Ferguson and Laing (Edts.) Metal and engineering in bone and joint sur-
gery. (Williams and Wilkins, Baltimore 1959).
2 Danis, R.: Theorie et pratique de l'osteosynthese. (Masson, Paris 1949).
3 Koranyi, E.; C.E. Bowman and M. Jannsen: Holding power of orthopedic
screws. Clin. orthop. 72: 283 (1970).
4 Rahn, B.A.; Th. Matter; E. Mikuschka-Galgoczy; W. Ziegler; J. Cordeyand
S.M. Perren: Relationship between radiodensity, hardness, holding power
of screws and microscopic structure in human cortical bone. In Asmussen
and Joergensen (Edts.) Biomechanics IV-B. (University Park Press,
Baltimore 1978).
5 Schatzker, J.; R. Sanderson and J.P. Murnaghan: The holding power of
orthopedic screws "in vivo". Clin. orthop. 108: 115 (1975).
6 Shermann, W.o.: Vanadium steel bone plates and screws. Surg. gynec. obst.
14: 629 (1921).
7 Wagner, H.: Die Einbettung von Metallschrauben im Knochen und die
Heilungsvorgaenge des Knochengewebes unter dem Einfluss der stabilen
Osteosynthese. Langenb. Arch. Klin. Chir. 305: 28 (1960).
8 Xu, X.X.; J. Cordey; B.A. Rahn and S.M. Perren: Determination of bone
density in human tibia using conventional radiography and densitometric
computed tomography. In preparation.
 537

COULD KOEHLER-FREIBERG'S DISEASE BE CAUSED BY BIOMECHANICAL

FACTORS?
H.A.C. Jacob, H. Zollinger and S. Georgiev
Orthop. Universitatsklinik Balgrist, Zurich, Switzerland

SUMMARY
Koehler-Freiberg's disease, an idiopathic necrosis of the metatarsal head,
most commonly occurs in the 2nd, sometimes in the 3rd and relatively seldom
in the other rays of the foot. Investigations regarding the possible causes
of the disorder have led to an enquiry into the biomechanics of the meta-
tarsophalangeal (MP) articulations. A search through the available
literature did not reveal any information that could be directly applied
and therefore a series of experiments were performed to determine a) the
ground reaction on the MP joints and on the toes, b) the direction of the
flexors and other tendons that cross the joint which might be responsible
for the force acting between the toes and the ground, c) the geometry of
the joint surfaces and finally, d) the force crossing the joint between the
head of the metatarsal and the corresponding phalangeal bearing surface.
It is shown .that the force per unit area, or specific loading, of the ~1P
joint of the 2nd ray could amount to about twice that which occurs in the
same articulation of the great toe. This might well explain the higher
occurance rate of osteonecrosis observed in the 2nd ray compared to that in
the 1st. Furthermore, it is shown that the resultant joint force in the MP
joint of the 2nd ray is more dorsally directed than that of the 1st, which
coincides very well with clinical observation.
Finally, a finite element stress-analysis has been performed to investigate
the stress situation within the metatarsal head under concentrated
load-bearing conditions. This has shown that possibly the cancellous bone
would collapse under repeated loading, leading to the well-known appearance
as first reported independently by Koehler and Freiberg over 60 years ago.
Thus under certain conditions where high specific loading of the metatarsal
head can occur, for instance through edge loading due to subluxation of the
phalangeal bearing surface, it is quite probable that osteonecrosis in the
immediate vicinity of force transmission would ensue.
I NTRODUCTI ON
The changes in the bone structure and form of the metatarsal heads as they
appear on X-rays that are characteristic of Koehler-Freiberg's disease
raise the question as to whether the action of mechanical forces playa
decisive role in the development of this ailment since, as is long known,
the architecture of bone can well be induced to undergo alterations
dependant on the external forces applied (Wolff, 1892). Investigations
relating to the blood vessels in this area have shown qualitative
differences in the localisation and the extent of coverage amongst normal
feet but no quantitative indication was found to explain why, for instance,
the disease attacks usually the 2nd and 3rd metatarsals and less commonly
the 1st (Zollinger and Kubik, 1984). Although it has been shown that
wide-spread bone necrosis could be caused by deficiency of the blood
538

supply, necrosis of the locaHsed type, such as in osteochondrosis

dissecal'ls. in which case a small portion of the joint surface becomes
detached from the rest and eventually floats as a separate entity within
the articulation,is difficult to explain. This type of lesion, sometimes
encountered in the 1st ray (Fig. 1). is almost exclusively seen in the
remaining metatarsal heads, in which cases it is more dorsal,y located than
the position observed when present in the 1st metatarsal.

Fig. 1: A case of osteonecrosis dissecans

of the first metatarsal head
The appearance of these subchondral bone lesions is indeed very reminiscent
of the surface fatigue phenomenon commonly known in mechanical engineering
as "pitting", which is due to the repeated action of concentrated loading
on the surface of an object at almost a point (Hetenyi, 1963). Therefore,
it was decided to investigate the mechanical conditions to which the
articular surfaces of the metatarsal heads were exposed under physiological
conditions in an attempt to perhaps find an explanation for the bone
lesions already described. in this manner.
On commencing this task. it became quickly evident that many of the
anatomical. topographical. and mechanical details that were required to
determine the forces acting across the joint were not available from the
literature known to us and thererfore we obtained the required information
by carrying out the relevant detail-investigations ourselves. These have
been reported at length elsewhere (Jacob et al. 1985) and it is the purpose
of thi s report to present the methods employed and. based upon a fi nite
element analysis only recently completed, to describe the probable manner
in which the pathological condition evolves.
INVESTIGATIONS PERFORMED
In order to determine the forces that act across the metatarsophalangeal
(MP) joints,' it is necessary to determine the external forces acting on the
toes. This was done as follows:
By observing the geometry of the joint surfaces and by determining the
exact position of the tendons relative to the centre of rotation of the MP
joint, the force in the tendons necessary to produce the ground reaction
force at the ends of the toes can be estimated. The tendon force together
with the ground reaction force then leads to the requried force across the
MP joint.
The plantar forces acting at the distal end of the 1st and 2nd rays were
measured by means of transducers that were specially built. employing
strain gauges. Four transducers were used, one under each MP joint and one
in the contact area between the toes and the ground. The transducers were
sensitive only to vertical loads, i.e. only the vertical component of force
was obtained. They were built into a cork sandal as shown in Fig. 2. The
sandal was so formed as to allow walking in a comfortable manner. The
sandals were not flexible so that practically no dorsal flexion occured in
 539

'1.
50

40

30

20

10

30

Bone cement ~ 20
.a;
Fig. 2: Location of force transducers ~ 10

in the cork sandal ~

o
..0
"EQ) 50
tl
40
Cii
a.
30

20

10
Fig. 4: Moire fringes on the MP-I

ttttt:
0
head and socket - Plantar force under MP-II
I

I IttttJ I
10

0
-
I I
Plantar force under 2nd toe
I I
60'1. 60'!.
! I'

Walking cycle
Time

Fi g. 5: Moire fringes on the MP-II fig. 3 : Walking force under MP-I,

head and socket MP- I I,and the first and second toes

.2"

Fig. 8: Resultant joint force in MP-I

Fig. 6: MP-I joint geometry

Fi g. 7; MP- II
joint geometry Fig. 9: Resultant joint force in MP-II
540

the MP joints when walking. For the purpose of our investigation, however,
we believe this to be irrelevant.
Force measurements were carried out on five test persons (four males and
one female) at walking rates of 30, 40 and 50 mlmin, using DC bridge
circuitry and amplifiers. A UV-recorder was used to register the traces. A
definite dependancy was observed between the walking rate and measured
force, but because of scatter, no attempt was made to express the
relationship. Fig. 3 shows the results obtained, depicted as percentages of
boQy weight. The bold curves give the average values measured and the
shaded areas the total scatter observed.
The topography of the metatarsal heads and the corresponding phalangeal
articulating surfaces was determined by use of the moire technique (Jacob
et al, 1985) on three autopsy specimens. Figs. 4 and 5 show typical moire
patterns of the 1st and 2nd MP joints with the cartilage intact. Figs. 6
and 7 show the joint geometry in a sagittal plane.
RESULTS OF FORCE MEASUREMENTS
Force measurements have shown that the average force between the tip of the
great toe and the ground in level walking amounts to about 17% of the total
body weight (BW). The corresponding force on the tip of the 2nd toe was
observed to be 4% BW. Neglecting friction in the MP joints and considering
the conditions for equilibrium (the resultant joint force must pass normal
to the tangent of the joint surface in the bearing area), the joint force
R-I in Fig. 8 amounts to 85% BW while R-II in Fig. 9 reaches 40% BW.
Assuming that the geometrical forms of MP-I and MP-II are similar, their
difference in diameter, which is in the ratio of about 2:1, would lead to a
difference in bearing area of 4:1. Taking the total individual loads
(85% BW and 40% BW) into consideration and also their ratios of bearing
area (4:1), it immediately follows that the specific loading (N/mm 2 ) of
MP-II is about twice that of MP-I.
Observing that R-II in Fig. 9
(MP-II) is more dorsally inclined
than R-I in Fig. 8 (MP-I) - by about
18 0 - it is possible that, if the
dorsal edge of the shallow
phalangeal socket does not proceed
sufficiently dorsally and backwards,
R-II may fall beyond the rim of the
bearing area. This will lead to
subluxation of the joint until the
plantar edge of the proximal part of
the phalanx bears against the
plantar aponeurosis as shown in
Fig. 10. In this state the resultant
Fig. 10: Resultant joint force force between the metatarsal head
in MP-II when subluxation occurs and the phalanx (R in Fig. 10)
becomes even greater than before
and, what is more, is now borne
solely by the dorsal edge of the
socket!
FINITE ELEMENT INVESTIGATIONS
To determine the stress distribution within the metatarsal head when the
joint force is transmitted over a very small area, as would be the case
with edge loading, the finite element method was used. Of several models
that were investigated, we will briefly describe only two. One is a model
 541

in which the metatarsal head of the 2nd ray has an almost physiological,
hemispherical shape and also includes the layer of covering cartilage. The
second model assumes that at the area of initial load transmission the
subchondral bone has so collapsed, that load bearing is now effected over a
narrow ring on a deformed head. Both models were rotationally symmetrical
and were run on an ANSYS-Programme at Sulzer Bros. Ltd., Winterthur. The
subdivision into finite elements was effected as follows: the cortex of
0.2 mm thickness was divided into 4 layers and again into ~ /48 radians.
The cancellous bone and cartilage were divided into almost square meshes
with a side length of rr/48 radians. In the first model (Fig. 11), a force
of 200 N was uniformly distributed in an axial direction over a circular
area of 4 mm diameter on the head of 12 mm diameter. In the second model
(Fig. 12), the same force was uniformly distributed over a ring of 4 mm
inner diameter and 8 mm outer diameter. Assuming Young's modulus for the
cortex, cancellous bone, and cartilage to be 15000, 1500 and 5 N/mm2
respectively, we arrived at the results shown in Figs. 11 and 12. It must
be mentioned that in the second model (Fig. 12) a hemispherical volume of
material at the joint surface was considered to have become detached from
its surroundings and was therefore given an artificial Young's modulus of
1 N/mm2.

DISCUSSION OF FINITE ELEMENT ANALYSIS

Comparative Stresses Total Force: 200 N Comparative Stresses Force: 200 N .

(max. stress difference) (max. stress difference)
Range in N/mm2
48 rtilage 48

10 10

8 8

6 6

4 4
illill[]
2 2
Moduli of Elasticity: MOduli of Elasticity:
II o
cortex 15000 N/mm2
o
cortex 15000 N/mm2
spongiosa 1500 Nlmm2
cartilage 5 N/mm2 :~~~~~~~ 1 50? ~~~~

Fig. 11: Stress intensity pattern Fig. 12: Stress intensity pattern
with load acting on small circular with load acting over a ring area
area' of normal head (with cartilage) on flattened head
For lack of space, only Figs. 11 and 12 have been included. These show the
stress intensities as determined by the Tresca criterion (maximum stress
542

difference) within the metatarsal head. It can be immediately appreciated

that if very localised loading would occur as in Fig. 11, the underlying
cancellous bone would not sustain the high stresses thus imposed. The
material so involved strikingly resembles, in form, the area of necrosis as
seen on radiographs (Fig. 1). Probably following primary multiple fractures
of the trabeculae and subsequent interruption of the micro-vascu1arisation,
the necrotic bone that is particularly vulnerable to the damaging effects
of repeated loading breaks down further. At the same time, in an attempt to
repair the damaged bone, resorption, that always preceeds bone deposition,
takes place at the still vascularised boundary. Due, however, to persisting
movement between the necrotic area and the surrounding material, new bone
formation is prevented, thereby leading in due course to a detached islet
of necrotic bone separated from its surroundings by a demarkation zone
occupied by soft fibrous tissue.
The question now arises as to what happens to the stress pattern within the
bone after the morphological changes described above have occured. To
determine this, we made use of the second model. As Fig. 12 shows, even
after re-distribution of the bearing load on the surface has taken place,
an annular zone of highly stressed cancellous material continues to exist.
Therefore, it is probable that damage to the living cancellous bone would
proceed further so that the surface of the metatarsal head would continue
to collapse, giving rise to the bizarre forms seen in advanced cases.
CONCLUSIONS
Measurement of ground reaction forces and a study of the anatomy of the
metatarsophalangeal (MP) joints have shown that the MP joint of the great
toe can be exposed to loads of about 85% bodyweight (BW) in normal level
walking while that in the 2nd ray could amount to 40% BW. Since the surface
area of the 2nd MP joint is only 1/4 of that of the 1st, the specific
loading of the former could be about twice that of the latter. This would
explain the higher rate of incidence of Koehler-Freiberg's disease in the
2nd ray compared with that observed in the 1st. Also, since the joint
reaction force in the 2nd ray is more dorsally inclined than that in the
1st, the area in which the disease manifests itself, as observed
clinically, is explained.
Finally, a finite element analysis of the metatarsal head under edge
loading conditions, as could occur when subluxation of the phalanx takes
place, shows a stress distribution pattern within the cancellous bone that
could produce damage to the bone in an area very similar to that clinically
encountered.
Therefore, we believe that Koehler-Freiberg's disease can be caused by
mechanical forces under particularly adverse conditions, that is, when the
joint force is transmitted over a very small bearing area.
REFERENCES
1. Hetenyi, M.: Handbook of Experimental Stress Analysis. John Wiley and
Sons (1963) p. 612.
2. Jacob, H.A.C., Zollinger, H., Georgiev, St.: Untersuchungen zur
Biomechanik der Zehengrundgelenke. Z. Orthop. 123 (1985) 929-938.
3. Wolff, J.: Das Gesetz der Transformation der Knochen. Hi rschwald,
Berlin (1892).
4. Zollinger, H., Kubik, St.: Ein seltener Fall von Osteonekrose der
Metatarsalekopfchen 1 und 2 beidseits. Chir. del Piede 8 (1984)
121-124.
 543

MEASUREMENT OF PRESSURE DISTRIBUTION ON CURVED AND SOFT

SURFACES
K. Nicol
University of MOnster, MOnster, Germany

1. '-ETHODES
For the measurement of pressure distribution between human body and
support mostly optical devices were used in the past. An overview is given
in Hennig and Nicol, 1978. More recently several computer supported
electrical measuring methods were developed which significantly support
quantitive analysis. Cavanagh & Hennig (1982) developed a platform
equipped with 1000 piezo-electric sensors sized 7.6 mm x 7.6 mm, each of
which is connected to one charge amplifier. In addition, they used the
same technology to develop a flexible insole (Cavanagh et. al., 1982).
In order to avoid complex circuitry due to high numbers of sensors a
multiplexing circuit was developed by Nicol and Henning (1976). The
principle is shown in Figure 1; it is in particular used for resistance
type and capacitance type sensors. The first leads of these sensors are
joined in rows, the second leads are joined in columns. A first switch
supplies voltage consecutively to all rows, a second switch scans all
columns, while the first switch is in constant position. In this way a
matrix of ax b sensors is scanned by a + b hardware channels only.

MEASURING
CAMCRORS

Figure 1. Schematic
diagram of the
multiplexer circuit for
capacitance type
COMPU1'ER sensors. Three 16-
channel switches are
used for scanning 512
sensors.

Constructing and wireing capacitance type sensors is simplified

particularly by the following technique. The sensor consists of a foam
rubber mat, each side of which is e.g. covered by 16 conducting strips in
the orthogonal direction. In this way 256 capacitors are formed. This
technique is now in particular used by P. Seitz (1986). He designed
measuring platforms and flexible insoles which are intensively used in
medicine and sport sciences.
544

LI NEARITY HYSTERESIS

50

40

0
30
o Loading
'9 Unloading

Ie
V
0

20

10

I
/, .Ell=.
N
Figure 2. Linearity
and hysteresis of
foil capacitor type
sensors.
I I I I I I
4 6 8 10 12 14 16 18 20 22

Due to the flexibility of the mat and due to the stripe technique the
sensor mat can be bent around any axis; the direction of this axis may be
not in parallel to one of the stripe systems. The limitation of this
technique is that the mat cannot be bent to form a sphere or a saddle,
because it will fold in like a sheet of paper would do in this case.
This limitation can be overcome if the mat is used on an uneven but stiff
support. In this case the mat can be glued on the support layer by layer,
so adapting to any arbitrary shape. Moreover, the flexibility of the mat
can be used to gradually adapt its shape to similar supports. E. G. a mat
which was glued to a specific bicycle saddle could be used with a couple
of other saddles by simply pressing the mat into the shape needed.
Nevertheless, there are strong limitations of that technique; e. g.
measurement on soft seats and on mattresses are not possible this way.
Therefore, a new technique was developed which is based on single
capacitance type sensors which are wired individually so that they can be
arbitrarily placed and that they can move individually when being loaded.
 545

RE LAXATION

70
\!l~~ ___
60

50

40

30 loading
29 N

'10

10

~
0 10 20 30 40 50 60 70

70 ~~
60

Unloading
:l.9N

20

Figure 3. Relaxation
10
of foil capacitor
type sensors,
o ------~-=-~-~-~-~~~~----~--------.------- plotted for loading
and unloading,
respectively.
Care had to be taken to keep the sensors as basic as possible so that the
system remains affordable even if hundreds or thousands of sensors are
used.
These sensors used are foil capacitors which contain small air gaps
between neighbouring l~yers of foil. These gaps are kept open by the
spring characteristic of the foil. When pressure is applied to the sensor,
the layers are pressed on each other so that the capacity changes
according to
C(d a ) = Co + Ca (1 - f * 4 * da/d f )
(Co = the capacity of that part of the capacitor which is not affected by
air gaps, Ca = capacity of the affected part when being pressed,
df thickness of the foil, da = average thickness of the air gap,
Ef = dielectric constant of the foil).
546

As both air gap and affected part are small, change in capacity is just
around 1 % at most. It turns out that linearity and hysteresis are
satisfying for biomechanical purposes (see Figure 2) and that relaxation
is acceptable if some care is taken while performing measurements (Figure
3). Temperature coefficient varies significantly for one sensor to the
other, sensors offering neglectable temperature coefficients can be
selected if necessary.
Care has to be taken in particular for the following measuring error.
If the sensors are not very close to each other, and if the surfaces
pressing on the sensors are not extremly soft, a gap can be observed
between the two surfaces in the neighbourhood of the sensor (see Figure
4).

Figure 4. Avoiding interference of sensor reading caused by pressure to

the surrounding area. This is done by providing a protecting rim (P.R.)
Therefore, a considerable amount of force in this area does not load the
support directly but it influences the sensor. For compensation, the
sensor is surrounded by a protecting rim which carries this additional
force, so that only that portion of the force is measured which directly
acts on the sensor.
If the two types of sensors described are compared to each other the
following is seen: The stripe technique has the advantage to be much less
complicated, so that a 32.000-channel system could be realized (Aisslinger
et. al., 1980). On the other hand the capacity of these sensors is around
1 pF so that ample shielding measures are necessary. This is normally
provided by glueing a shielding layer onto the stripes. In contrast, the
single sensors are very insensitive to being touched by a grounded body,
to movement and to interference of noise. Consequently, 1/10 mm unshielded
wire can be used which provides excellent flexibility of the sensor mat.

2. APPLICATIONS
The stripe technique primarily was applied to curved, but hard
surfaces. A selection of these applications are described to some extend
in Nicol and Koerner (1985) and VOcel, Liebscher and Nicol (1985). Typical
examples are measurements on a hard chair for disabled, on a saddle for
handicapped riding, on a bicycle saddle, on a socket for prosthesis, on an
orthotic device for the treatment of kyphosis, between dummy and safety
belt and inside a ski boot.
 547

The stngle sensor technique was first used in same applications with
low . channel nlllibers. Pressure distribution was measured on the teats of
Cows within a milking machine and on a machine hammer. Second, a 512-
channel system was constructed, consisting of two mats of 256 sensors
each, three interface boards and an APPLE rr-Computer into. which the
boards were plugged. The first mat is sized 40 cm x 40 em (sensors are
spaced 25.Mm x 25 mm); it is used for measurement of pressure distribution
while sitting and for high resolution measurements while lying. The second
mat 1s sized 64 cm x 96 cm, the sensors are spaced 40 Mm X 60 Mm. This mat
just covers one half of an adult lying on a mattress. The small mat can be
rewired within a ~uple of hours in order to cover the second half of the
body. The range of, the system is 300 kpa,resolution is .1 kpa, so that
even low load areas can be measured while lying. Figure 5 shows the whole
system. In the foreground the small mat is placed on a pillow; it is seen
that the mat adapts to every single fold of the pillow surface.

Figure 5. Set
up of a 512-
channel mea-
suring system
for mea-
surement on
soft sur-
faces, using
single foil
capacitor
sensors.
548

Figure Ga. Pressure

distribution while lying
on a soft matress,
displayed by means of a
gray scale. Maximum
pressure of 100 kpa
corresponds to full
scale.
 549

Figure 60. Same as Figure

6a. Maximum pressure
corresponds to four times
full scale. Displayed for
demonstration of
resolution.
550

Up to now this system was used for three studies. First, pressu~e
distribution on twelve different pillows was measured in two sUbjects In
the prone and side position. Parameters of the pillows were size,
structure of filling and density of filling. The measurements were
evaluated in respect of maximum pressure, shape of the contact area and
distribution of pressure within this area.
In the second study the mat was placed on selected surfaces ranging
from wood to very soft pillow. The peak pressure was found to vary from
300 kpa to 10 kpa, the area of inprint was 20 cm2 to 200 cm2
In the third study the body was placed on a mattress and a series of
measurements was carried out shifting the small mat beyond the body. So a
total of about 2000 areas was measured. The result is shown in Figure" 6.
Maximum pressure turned out to be 100 kpa under the heals, 80 kpa under
the head and 60 kpa in the hip area.
An intensive study on construction of mattresses and decubitus will be
conducted shortly.

REFERENCES
1. Aisslinger, U., Nicol, K., & Preiss, R. 1981. Device for high
resolution force distribution measurement. In: A. Morecki, K. Fidelus, K.
Kedzior & A. Witt (Eds.) Biomechanics VII-A, pp. 548 - 552. Polish
Scientific Publishers, Warszawa.
2. Cavanagh, P.R., & Hennig, E.M. 1982. A new device for the measurement
of pressure distribution on a rigid surface. Medicine and Science in Sport
and Exercise, 14(2), 153.
3. Cavanagh, P.R., Hennig, E.M., Bunch, R.P., & Macmillan, N.H. 1982. A
new device for the measurement of pressure distribution inside the shoe.
In: H. Matsui und K. Kobayashi (Eds.) Biomechanics VIII-B, pp. 1089-1096.
University Park Press, Baltimore.
4. Hennig, E.M., & Nicol, K. 1978. Registration methods for time-
dependent pressure distribution measurements with mats working as
capacitors. In E. Asmussen & K. Jorgensen (Eds.), Biomechanics VI-A (pp.
361-367). Baltimore: University Park Press.
5. Nicol, K., & E.M. Hennig. 1976. Time-dependent method for measuring
force distribution using a flexible mat as a capacitor. In: P.V. Komi
(ed.) Biomechanics V-B, pp. 433-440. University Park Press, Baltimore.
6. Nicol, K., & Korner, U. 1985. Pressure distribution on a chair for
disabled SUbjects. In D.A. Winter, R. Norman, R. Wells, K. Hayes, & A.
Patla (Eds.), Biomechanics IX-A, pp. 274-280. Champaign, IL: Human
Kinetics Publishers.
7. Yucel, M., Liebscher, F., & Nicol, K. 1985. Pressure distribution in
orthotic devices for treatment of the spine. In D.A. Winter, R. Norman, R.
Wells, K. Hayes, & A. Patla (Eds.), Biomechanics IX-A, pp. 269-274.
Champaign, IL: Human Kinetic Publishers.
 551

CRUTCH. STAIRCASE RAILING AND FOOT-FLOOR REACTION FORCES

DURING PARAPLEGIC'S STAIR CLIMBING
D. Rudel*. T. Bajd*. A. Kralj* and H. Benko**
Faculty of Electrical Eng.*. E. Kardelj Univ . Ljubljana. and
University Rehabilitation Institute**. Ljubljana. Yugoslavia

1. I N'fRODUCTI ON
Stairs represent a major obstacle to people with locomotor system defi-
ciencies. From a biomechanical point of view stair climbing differs consi-
derably from ground level walking. Differences are reflected mainly in the
larger range of knee and hip motion required during the swing phase and
the greater forces generated by knee and hip extensors during body trans-
fer to a higher step. Gait pattern and control principles used during pa-
rapleyic patient ground level walking enabled by Functional Electrical
Stimulation (FES) have to be changed in stair climbing; Several research
yroups have reported successful restoration of primitive ground level
w~lKing in complete paraplegic patients where different stimulation techni-
4ues have been implemented. Implanted nerve stimulation was used by
Willemon et al. (1), Brindly et al. (2), and Thoma et al.(3), intramuscu-
lar stimulation by Marsolais (4). and transcutaneous surface stimulation
by LJublJana (5) and Chicago groups (6) as well as by Petrofsky and his
yroup (7). Marsolais and Kobetic first demonstrated paraplegic stair clim-
biny usiny bilateral railings in 1984. Bajd et al. (8) reported FES enab-
led stair climbing in paraplegic patient using a unilateral railing and
an ordinary crutch. Kobetic et al. (9) reported paraplegic patient's
ascending and descending stairs using one rail and one quad forearm crutch
for balance.
So far, climbing stairs in paraplegic patient has not been quantitatively
assessed although there were several studies performed on normal subjects
(10,11,12). Hands and arms play an important role during paraplegic body
transfer to a higher step. not only because of insufficient forces gene-
rated by FES in appropriate muscles, but also due to the lack of hip sta-
bility and resulting balance problems. It is the aim of this study to
quantitatively estimate the contribution of the arms when ascending stairs.
For that purpose ground reaction forces and forces exerted on both hand
supports were measured (see Fig.l). The results obtained should provide an
estimation of efficiency of lower extremity stimulation pattern.

2. MATERIALS AND METHODS

The instrumentation shown in Fig. 1 included a staircase with 14 cm
riser height and 21 cm tread depth. A section of the first step was cut
out and was placed on the Kistler force plate (dark part of the first step
552

in Fig.I). That enabled direct measurements of foot-stair reaction forces.

Only the vertical component of the reaction force vector Fz was observed.
A specially designed unilateral railing, 88 cm high, was used enabling
three dimensional measurements of forces exerted by body weithgt. Its mec-
hanical construction and strain-gauge type transducer placement prevented
crosstalk among railing segments. The measuring range of forces was between
5 and 1000 Nm for all three directions when cOnnected to a general purpose
strain gauge amplifier. A regular tip crutch instrumented for crutch axis
force measurement up to IOOON was used, supporting the contralateral side.
All the analog signals were recorded on magnetic tape for further proces-
sing. An ordinary VHS camera was also used to document patient's stair
climbing.

FIGURE 1. Paraplegic patient climbing stairs by the help of 4-ch surface

FES during the vertical reaction force measurements. The dark portion of
the first step is exactly above the force plate.

In Fig. 2 reaction forces exerted by patient under the feet (F z ), in ra-

iling (Fud, Ffb), and the crutch (Fc) are schematically presented. A
(D,B,T) coordinate system is used with stair railing and (X,y,Z) for force
plate. As the progression velocity in paraplegic ascent is rather low,
such stair climbing can be considered as a sequence of static events. When
analyzing them, the sum of reaction forces has to be equal to the body
weight in every moment of walking:
 553

FIGURE 2. Schematic diagram of a paraplegic subject on the first step

when ascending stairs during the reaction force measurements.

W=m*g = Fz+Fc+Fud'+Ffb'= Fz+Fc+Fud*CosA+Ffb*sinA, (Eq. 1)

where:
W..... body weight
m............ body mass
g..... gravitational acceleration
Fz .. foot-stair vertical r.force
Fc ....... vertical r.force in crutch
Fud ......... up/down r.force in railing
Ffb .... forward/backward r.force in railing
A... railing angle to vertical position
The calculated values should match the results gathered from force
transducers.
Three outpatients with complete absence of all motor functions in lower
1imbs were involved in the FES stair cl imbing pr09rarr,:W.B. (T 7-8, 8 years
after injury), S.M. (T 5-6,4.5 years after injury), and A.A. (C 7). The
simple reciprocal gait pattern used in level walking was also found useful
in early trials of ascending stairs. Simultaneous flexion in the hip, knee
and ankle joint during the swing phase was elicited by stimulation of the
peroneal nerve. In the stance phase, knee extensors were stimulated. The
parameters of stimulation were: monopolar rectangular voltage pulses of
20 Hz, 0.3 ms pulse width. Movements were controlled by two switches built
into the crutch handle, one for each leg. Activation of them initiated the
swing phase and elicited flexors for the activation period.
554

In Fig.1 the paraplegic patient is shown during the FES assisted stair
climbing, which consists of the following phases:
- in double stance phase the patient leans forward with one elbow on the
hanurail. The crutch in the other hand is positioned on the next higher
step;
- the patient triggers flexor response in the leg close to the handrail
and simultaneously switches off the stimulation of knee extensors of the
same leg. This initiates the swing phase and brings the leading leg to
the height of the next step;
- the patient switches off the excitation of the flexion response and the
stimulation of the knee extensors occurs automatically (see Fig. 1);
- with the help of both arms the patient lifts his/her body and triggers
flexion inthe contralateral leg to bring it to the same step;
- the patien't brings the crutch closer to the leg, erects the trunk, holds
the handrail at a higher position and transfers the crutch to the next
step.

3. RESULTS
The results of measurements of reaction forces in the supports during
paraplegic FES assisted stair climbing depend significantly on the level
of the patient's rehabilitation, on the success of FES application, the
injury level, as well as patient's physical condition. The progression
velocity is rather low in general (6 seconds/stair'for W.B. and 12 seconds
for A.A.). In Fig.3, the results are presented for the paraplegic W.B.
The recordings are divided into four phases labeled with "I" through "IV",
which correspond to different patient's activities. In phase "I" the first
step was initiated by eliciting the flexion reflex.During this time inter-
val the crutch and the railing were loaded to a larger extent. The force
in the crutch Fc and the vertical force Fud in the railing increased (up
to 350N and BON respectively). The patient also pulled himself forward
which resulted in slight increase in forward/backward force Ffb in the
bottom trace in Fig. 3. In phase II vertical reaction force under the foot
increases, as the leading leg was lifted to the first step. This results
also in a slight decrease of reaction forces Fc (250N) and Fud (40N). In
phase III the body weight was transferred to the leading leg. Increase in
the force Fz was caused by initiating the swing phase of the trailing leg
and resulted in an additional loading of the force plate. The force Fc
decreases to ON during this time interval, as the crutch was totaly rele-
ased when the step was completed. Fud changes its polarity as a consequence
of pulling the railing upwards when balancing the body. Phase IV reflects
force distributiGn during double stance with both legs on the first step.
As was expected, there were rio significant cnanges in Ffb through the
whole process of climbing one step.
The results ,were processed according to (Eq. 1) and compared to the
patient's body weight (74 kg or 730N)at the points t2 through t4, when
the trailing leg is already lifted.
 555

VERTICAL REACTION FORCES

I II III IV

50~r"Z(!I~ - - - - - ~-- - - ~ !lQD~

1
- I I

500 Fc(N) I I I

O~II~I
20~t~~ :-_ ::
5~f Ff~)
or
0_

, tt 1
1 ,
2
I
4ft 3 f t4
Pat.W.B.-Th7 8 time(s)

FIGURE 3. Typical recording of reaction forces in hand supports during FES

enabled ascending in paraplegic patients:
- vertical ground reaction force Fz ......... trace 1
- crutch vertical reaction force Fc ......... trace 2
- railing up/down reaction force Fud ........ trace 3
- railing forward/backward r.force Ffb ...... trace 4

F(N) t2(3.24s) t3(4.07s) t4(4.45s)

Fz 384N 621N 755N
Fc 289N 115N 32N
Fud*cosA 41N -5N -34N
Ffb*sinA 2N -2N ON
Summ(N) 716N 729N 753N
Having in mind the simplification made in crutch force calculation and
railing force transducer tolerance of 5%, the results match accordingly.
From the table above, it is evident, that at the point t2 almost 50% of
the body weight was carried by the patient's arms.

4. CONCLUSIONS
The results present the estimation of the arm contribution during FES
assisted stair climbing in paraplegic patients. The measurements in three
patients in the early days of the stair ascending program clearly indicate
what was predicted, i.e. that the arms ana hands perform most of work du-
ring transfer procedure. The hands vertically supported up to 50% of the
556

patient's body weight during the swing phase of the trailing leg. The di-
stribution of forces in crutch and railing support varied from patient to
patient. The variations clearly indicate the patient's personal technique
chosen for the body transfer to a higher step. This approach proceedes
stair climbing analysis as well as synthesis of FES enabled stair ascending
program. When ascending several steps on a normal stairway the average ti-
me to overcome one step was 4.5 s for the patient W.B. for the first five
steps. Later on the fatigue problems showed down the progression velocity.
It is expected that an introduction of additional stimulation channels as
well as improved muscle activation timing will reduce amount of the hand
support and contribute to a better propulsion.
REFERENCES
1. Wilemon WK, Mooney V, McNeal D, Reswick J: "Surgical Implanted Periphe-
ral Neuronic Stimulation", Downey CA, Rancho Los Amigos Hospital,
Internal Report, 1970.
2. Bridnley GS, Pol key CF and Ruskton DM: "Electrical Splinting of the
Knee in Paraplegia", Paraplegia, 16, 1978, pp. 428-435.
3. Thoma H, Frey M, Gruber H, Holle J, Kern H, Reiner E, Schwanda G,
Stoehr H: "First Implantation of a 16-ch Electric Stimulation Device
in Human", ASAIO, Toronto, 1983.
4. Marsolais EB, Kobetic R: "Functional Walking in Paralyzed Patient by
Means of Electrical Stimulation", Clin. Orthopaedics, 175, 1983,
pp. 30-36.
5. Kralj A, Bajd T, Turk R, Krajnik J, and Benko H: "Gait Restoration in
Paraplegic Patients: A Feasibility Demonstration Using Multichannel
Surface Electrode FES", Journal of Rehabilitation R&D, Vol.20, 1983,
pp. 3-20.
6. Cybulski GR, Penn RD and Jaeger RJ: "Lower Extremity Functional Neuro-
muscular Stimulation in Cases of Spinal Cord Injury", Neurosurgery,
15(1) ,1984, pp. 132-146. -
7. Petrofsky JS, Phillips CA: "Computer Controlled Walking in Paralyzed
Individual", J. Neur. Orthop. Surg. 4, 1983, pp. 153-164.
8. Bajd T, Kralj A: "The Impact of Functional Electrical Stimulation on
the Rehabilitation of Paraplegic Patients", Proc. XIV ICMBE and VII
ICMP, Espoo, 1985, pp. 393-395.
9. Kobetic R, Carroll SG and Marsolais EB: "Paraplegic Stair Climbing
Assisted by Electrical Stimulation", Proc. 39th ACMB, Maryland, 1986.
O. Andriacchi TP et al: "A Study of Lower-Limb Mechanics During Stair -
Climbing", J.Bone Surgery, 62-A, 1980, pp. 749-757.
1. Landjerit B, Thairot M, Michaut E, Pelisse F: "A Dynamic Study of the
Human Knee Joint Motion During Walking on flat Ground, Upstairs and
Downstairs", Proc. IFAC Workshop "Human Gait Analysis and Application",
Montpellier, 1983, pp. 23-38.
2. Graille J, Liegeois A: "Locomotion Model Using CAD Methods", Proc. IFAC
Workshop "Human Gait Analysis and Application", Montpellier,1983,pp.1l-22.
ACKNOWLEDGEMENTS
This study was supported by the Research Communities of Slovenia,YU, and
the National Institute of Handicapped Research, Dept. Education, Washington
D.C., USA. Tbe authors are indebted ProLM.K1Jajic from IJS Institute,and
M.Chat~ao,M.Sc., Iskra-Optoelectronics, Ljubljana.
 557

SLOW WALKING IN NORMAL SUBJECTS ON A TRACK AND A MOTOR DRIVEN

TREADMIll
R.H. Rozendal and H.H.C.M. Savelberg
Dept. of Functi ona I Anatomy, Free Un i vers i ty, Amsterdam, The
Netherlands

It is well known that many gait parameters depend on the walking velocity
(Andriacchi et al., 1977; Larsson et al., 1980). Comparison of pathological
and normal gait is only attainable if the latter is performed at the
Same slow or even very slow speed as the former. Descriptions of gait of
"normal" subjects performing slow and very slow speeds are scarce, but
do exist: Larsson et al. (1980) mentioned means and standard deviations
of velocities, cadences of stride lengths at speeds of 0.46 and 0.87
m.s-l. Larsson elaborated on temporal aspects of subphases of the stride
in these and three other classes of speed. Standard deviations in the
various classes of speed are related to the ranges of speeds,
voluntarily chosen by the instructed subjects. The ratios stride length/
cadence of these normal subjects indicated that they had walked at the
very slow speed with relatively long strides and low cadences compared
with hemiplegic patients. The latter took short strides and relatively
high cadences (Mischner-van Ravensberg et al., 1985).
Patients react to the equilibrium problems evoked by the low velocity
in another manner than normal subjects. So it is questionable if
comparisons between patients and normals are meaningfull. Data on
normal subjects should not be used as norms too easely (Quanbury, 1985).
A priori the question arises if normal subjects react uniformly to the
instruction to walk very slowly. To solve this motor problem, the subject
can adopt two strategies, defined as extremes of a continuum: walking
with a stride length nearing the limit of zero or taking longer but very
slow steps resulting in a cadence near zero. This choice could be
described the quotient of step length and step frequency or the stride
length/cadence ration (R).
The solution of a motor problem could be very well related to
environmental factors. A lot of research on walking is done on subjects
walking on a motor driven treadmill (MDT). If one uses a good treadmill
no mechanical differences between treadmill and track walking do exist
(van Ingen Schenau, 1980). Use of a MDT will manipulate and eventually
expel velocity dependent variations in walking.
In view of the differences in visual perception, performance of subjects
not familiar with walking on a MDT could be different from performance
in overground walking. So subjects were asked to walk on a track with a
freely chosen "normal" speed and were instructed to walk very slowly.
Next both speeds were imposed on them on a MDT. In this experiment the
MDT-runs could be considered as walking under environmental constraint
imposed on the subjects.
558

After a period of 12 to 16 weeks a proportion of the subjects was asked

to walk both runs in the order: MDT and track. Apart from the difference
in sequence these subjects were no longer naive with respect to the
environmental constraint.
Interest was focused on the way subjects solved the motor problem posed
by the instructions and by the MDT. The solution was indicated by speed
and the spatiotemporal strategy of walking or stride length/cadence
ratio (R).

METHODS

Subjects were male (n=21) and female (n=17) students not familiar with
walking on a MDT. They walked on an instrumented track of 15 m length and
1.5 m width in which the middle 5 m were used to measure their velocity
(sonar cells) and to count their steps from videotape. They performed
three free speed runs and were instructed to do a very slow run as many
times as were needed to perform at a speed lower than 0.5 m.s-l.
Next they performed on the motor driven treadmill at the median speed of
the three free speed runs and at the speed of slowest run. Steps were
counted during 30 s. After the run in which the free speed was imposed on
them they were asked to comment.
The belt of the treadmill was 3 m long and 1.25 m wide; driven by an
electric motor of 3 kW power and a continuous speed variator. The speed
of the system was not influenced by a human walking on it. The MDT did
meet the requirements deduced from van Ingen Schenau (1980).
Eleven of the male and four of the female subjects were asked to return
to the lab after 12 to 16 weeks and to perform the same runs but in
inverse order. This was called the second session.
The track and MDT were called environmental conditions in which free and
slow speeds were instructed or induced respectively. Of the instructed
runs in the first session, the median free speed and the first slow run
meeting the above mentioned requirements were elaborated.
Statistics used were analysis of variance and Student's t with a level
<
of significance ac. 0.05.

RESULTS

Mean speed and standard deviations of the male and female subjects
instructed free speed runs are tabulated in table I. For comparison data
of Larsson's subjects free "ordinary" run were computed by us.
In both sessions during instructed track and imposed MDT free and speed
and slow speed runs performance was characterized by R as tabulated in
table II.
In the free speed no significant differences in R between environmental
conditions were found in the first session.
The imposed slow runs in both sessions were performed at the speeds of
the first instructed slow run at mean .23 m.s-l s.d. .10 by the male
and mean .21 m.s-l s.d. .12 by the female subjects. The second
instructed slow run (on the track) was performed at mean .15 m.s-l s.d .
. 09 by the male and mean .10 m.s-l s.d. .03 by the female subjects. This
difference was significant in the male subjects, but not in the female.
All slow runs were performed with a wide variation in R .. Due to this,
differences in R between speeds, environmental conditions and sessions
in the slow runs did not reach significance.
 559

In the second session male and female subjects lowered their speed in
the free run on the track significantly. R was significantly different
between sessions on the track in the female subjects. In the male
subjects the difference between sessions on the track was not significant.
On the MDT R during the imposed free speed runs was not different
between sessions.

DISCUSSIONS

After the imposed free speed run on the MDT in the first session all but
one of the subjects expressed that they felt they were going faster than
on the track.
In the second session subjects walked slower on the track than during the
previous run on the MDT. This is suggestive of a certain preponderance
of the propriocepsis over the apparent false visual information on
progression on the MDT. They might have been naive with respect to the
MDT, but recognized the environment very well. Also after this
experience in the second session the instructed free speed run was
different from the customary. This was only expressed significantly in
the speed and the value of R by the female subjects.
The use of uncommon environmental variations in testing the walking
capacity of patients, naive to these environments, is questionable in
view of these results. Normar-SUOjects accustomed to walk on a MDT or
to another environmental condition tend to perform more consistently.
The variation of combinations of step lengths and f~equency was impressive
in the slow speeds. Not only did any combination between the extremes
mentioned in the introduction occur,-but intra-individual differences
were also frequently seen. This is in contrast to the small variation
in the instructed and imposed free speed runs in the first session, in
spite of subjects' impression of speed.
Slow speed walking of normal subjects apparently cannot be
characterized by parameters of just one test run. Data from groups of
subjects might be quite misleading. Expressed in coefficients of
variation (SD / mean = CV) the variation within the group of 38 subjects
doing the free speed run in the first session was .09, while the CV of
the slow speed run on the track was .61. Of the slow and very slow
speed runs of the subjects of Larsson et al. (1980) CV = .16
respectively .26 (25 respectively 27 subjects, our computation).
Their speeds were .87 s.d. .21 respectively .46 s.d. .15 m.s-1. The
slow speed of our subjects was .22 s.d. .11 m.s-1. The whole is
suggestive of an expanding variation with lower speeds. The importance
of mean values for speed- and step frequency - dependent variables of
gait merits some reconsideration. Thus the custom of analysing and
comparing data of one run might be not be valid at these low speeds.
A posteriori the defination of normal walking at low speeds should be
reconsidered. A more elaborate walking performance test should be
developed providing a better basis to study normal and pathological
walking.
 Table 1. Means and (standard deviations) of Larsson's "ordinary" and
our free speed in the first and second session, in m.s-l
Larsson et al. (1980) first session second session
male 1. 26 (.19) 1. 46 (.18) 1.30(.19)
female 1. 28 (.20) 1.55(.10) 1.42(.12)
Table II. Ratio's for both sexes in all the conditions and both
sessions, with (standard deviations).
TRACK MDT
free speed low speed free speed low speed
first session
n 21, males .405(.03) .700(.54) .400 (.03) .539(.35)
n 17, females .354(.03) .576(.34) .355(.03) .471(.26)
second session
n 11, males .424(.05) 1. 137 (.97) .397(.03) .539(.33)
n 4, females .422(.03) .785(.51) 37 (.04) .389(.25)
~
on
 561

REFERENCES

1. Andriacchi TP, Ogle JA & Galante JO: Walking speeds as a basis for
normal and abnormal gait measurements. Journal of Biomechanics 10,
261-268, 1977.
2. Ingen Schenau GJ van: Some fundamental aspects of the biomechanics of
overground versus treadmill locomotion. Medicine and Science in Sport
and Exercise 12, 257-261, 1980.
3. Larsson LE, Odenrick P, Sand lund B, Weitz P & Oberg PA: The phases of
the stride and their interaction in human gait. Scandinavian Journal
of Rehabilitation Medicine 12, 107-112, 1980.
4. Mischner-van Ravensberg CD, Bergkamp BM, Pieters R & Rozendal RH: Gait
parameters of hemiplegic and normal subjects. In: Perren SM & Schneider
E(Eds.), Biomechanics: Current Interdisciplinary Research,
Developments in Biomechanics Vol 2. Martinus Nijhoff Publishers 543-
543, Dordrecht, 1985.
5. Quahbury AO: The clinical gait lab: form and function. In: Winter DA
& Norman RW)Eds.), Biomechanics IX. Champaign: Human Kinetic
Publishers, 509-512, 1985.
 563

COMPARISON OF SUBJECTIVE GAIT OBSERVATION WITH MEASURED GAIT

ASYMMETRY
D. Attinger, S. luethi and E. Stuessi
Swiss Federal Institute of Technology ETH-Center WEO, Zurich,
Switzerland

1. INTRODUCTION
The description of gait is used in medicine as well as in every-day-language
to characterise a person's gait. Hereby the observation of asymmetries is an
important factor. In order to define the reliability of subjective
descriptions of gait asymmetry, visual rating of temporal- and
load-asymmetries performed by a jury of eight medical people is compared with
the equivalent measurements.
Human gait reacts very sensitive and differenciated on functional problems.
It can therefore be used as a mirror of physical disbalance as long as
negative senso-motorical and mental influences are neglectably small. It has
been shown, that due to asymmetric movement patterns problems of the
locomotor system become more often acute (DEWAR 1980). This can also occur
in symmetric movements as for example in walking or running. Increased
loadbearing on the healthy side of the body due to sparing load on the
involved extremity is as long meaningfull as there exists a functional
problem. The movement pattern should later normalize in parallel with
decreasing insufficiency. Nevertheless it often can be observed, that a
"protection behaviour" has become deeply rooted and is stored as a "normal
pattern". Later, thi s can lead to other problems of the locomotor system,
sometimes even on the non-involved extremity or the spine.
One of the aims of medical- and physiotherapeutical treatment is to create
the prerequisites for a normal motor behaviour and to make the patient aware
of it in order to avoid negative consequences. Wishing to systematically
influence a specific motor pattern requires a precise description. This is
not easy, because movements are often too fast and complex to allow a precise
verbal portray. Although, there are good reasons to argue against the use of
computers in a clinical environement, it is reasonable to make use of
biomechanical measuring techniques in such a situation. A quantitative
description can confirm or disprove subjective expectations completing the
picture of a persons gait.
2. METHOD
2.1. Asymmetry parameters
Outgoing from general criteria of observation, a person might be considered
as "limping" if she or he shows a temporal irregularity which can be seen,
heard and, under certain circumstances, also be felt. For this study, the
parameter "temporal asymmetry" is defined as the percentual difference of
contact time between left and right extremity:
shorter contact time
dT [%] ( 1--------
longer contact time
)* 100
564

Another aspect of asymmetry is "load". A patient sometimes is adviced to

load his leg with say "only 40kp" or should spare an extremity.
For this study "load asymmetry" is defined as the percentual difference of
the maximal vertical ground reaction force during one floor contact between
the two legs. Hereby it is not taken into account, whether this peak occurs
during the phase of acceleration or deceleration nor how long it is.
smaller vertical peak force
dF [%] ( 1 - )* 100
greater vertical peak force
2.2. Measurement of gait asymmetry
In order to analyse gait concerning asymmetries, two Kistler force plates of
40x60cm were installed directly behind each other into a walkway of 10m
length and 102m width. The plates can hardly be seen by a "novice". The
set-up allows an unencumbranced gait even with walking aids. The walk-way
can be used as training space for therapy when it is not used for measuring
purposes.
By means of forceplates it is quite easy to measure temporal- and load
asymmetries as defined in the section before: The maximum an duration of
the vertical ground reaction force have to be registred. These values can be
derived and indicated also without the use of a computer.
An inexpensive, easy to handle biofeedback tool with the described features
was constructed at the Biomechanics Laboratory at the Federal Institute of
Technology. It is installed at a gymnasium of the Department of Rheumatology
at the University Hospital in Zuerich. Physiotherapists regularly use the
device during their work independently to quantify asymmetries before and
during rehabilitation, for documentation of therapeutical results or as a
training tool. If a patient shows a disadvantageous gait pattern as for
example no push-off or an asymmetric loading of the limbs, he can be
instructed to train by himself.
In digital form the following numbers are displayed after every trial:
Maximal vertical ground reaction force:
- range: 0 - 200 kp, resolution: kp

Obligated vertical ground reaction force:

- measures the maximal vertical ground reaction force and
compares it with the preset values.
- If the measured force is greater then the preset force,
an accoustic signal sounds.
- range: 0 - 200 kp, resolution: 1 kp
Contact time: - range: 0 - 0.99 sec, resolution: 0.01 sec
In order to define the "normal" range of asymmetry, symmetry parameters of
ten doublestrides of 80 subjects without any bodily ailments have been
statistically analyzed. In this group females and males, athletes and
non-athletes, normal- and over-weight, age ranging from 20 to 75 years have
been included. Foot characteristics have been registered using podograms.
Results show, that in 98% of the double strides the measured temporal- and
load asymmetry is below 5% for this group.
 565

2.3. Documentation
For thi s study, two separate vi deotapes "temporal asymmetry" and "load
asymmetry" have been put together. Twelve respectively eOleven video records
have been selected according to the measured asymmetry. The range of
asymmetry varied over the spectrum defined in the program picture from no to
maximal asymmetry. The two extrems defined the upper respectively lower
limit of the visual-analog-scale used by the jury to define subjectively the
grade of asymmetry. The selected patients were of both sexes, age ranging
from 18 to 67 years, athletes and non-athletes with normal foot
characteristics. The shots of the walking subjects were taken by a
stationary camera during the measurements from behind, from the front and
afterwards from the side.
In a pilot study twelve subjects with different functional problems were
observed walking by five experienced medical people. The jury had to rate
the asymmetry. For further analysis, the jury gave preference to the
videotapes although one was aware of the disadvantages (details could be
overestimated if the same trial is observed several times, filmspeed is too
small for exact movement analysis).
2.4. Trial
In oraer--to test the reliability ob the subjective rating of temporal- and
load asymmetry during gait, an experienced team of three medical doctors and
five physiotherapists of the Departement of Rheumatology at the University
Hospital in Zuerich was asked to:
- Temporal asymmetry:
a) define the side which was loaded longer
b) rate the grade of temporal asymmetry on a visual-analog-scale (dT)
- Load asymmetry:
c) define the side which was loaded more
d) rate the grade of load asymmetry on a visual-analog-scale (dF)
temporal asymmetry load asymmetry
trial injured longer dt[%] injured more dF[%]
number side loaded meas. side loaded meas.
1 L R 4.0 R R 4.0
2 R L 14.0 L L 1.4
3 R L 6.6 L R 4.3
4 L R 12.6 R R 2.5
5 R L 15.6 L L 6. 7
6 L R 4.1 L R 3.0
7 L R 2.0 R L 13.8
8 R L 2.6 R R 2.5
9 L R 2.0 L L 7.0
10 L R 5.5 R L 7. 1
11 R 0.9 R L 2.8
12 R L 7.3
TABLE 1: Measured asymmetry of the selected trials
566

3. RESULTS AND INTERPRETATION

3.1. Temporal asymmetry
a) The probability, that the jury (n=8) hit the correct side that was loaded
longer is 80% with a large deviation from 55 to 100%. 87% of the average
jury observations with measured asymmetries above 5% were correct with
smaller but still remarkable deviations (60 to 100%).
This means, that the reliability of the observation of the jury concerning
the leg that is loaded longer is sufficient for clinical use and needs no
additional measurements. A trend can be observed, that the probability of a
correct observation of the longer loaded side becomes bigger with increasing
measured asymmetry (Figure 1).
Already with this criterion, the reliability of the observations of the
different jury members varies a lot according to their experience in gait
observation.

FIGURE 1: Probability of correct observation of longer loaded side in

dependence of the measured asymmetry (p = 0.59 + 0.04 MA).
b) There exists almost no correlation between the measured and the rated
temporal asymmetry (r e 0.27, n=8).
This means, that a rating of the degree of temporal asymmetry should only be
performed using a measuring system.
3.2. Load asymmetry
c) The probability of a correct observation of the more loaded side is - with
an average of 40 % - very small. It seems, as if a gait pattern contains
plenty informations, that an asymmetry can be recognized, but that a
reduction of the information in order to answer a specific question is very
hard.
Although some members of the jury show a probability of correct observation
of more than 50% (60 to 75%), others - again the less experienced ones - seem
to have difficulties to define the side which is loaded more. The
distribution of the correct observations of the jury as it is shown in
figure2 appears therefore to be randomly distributed over measured asymmetry.
This means, that the jury was not able to observe the side that was loaded
more with a sufficient reliability for clinical use.
Interestingly, two different types of asymmetry can be distinguished:
 567

Some subjects load the healthy extremity more, others the injured extremety
depending on the grade and kind of disability they have. The realiability of
observations is better in the group that load the noninvolved extremity more,
namely 53% compared with 33% in the other group. This could be, because one
expects automatically, that an injured extremity is spared .

0.8 ...
...
0.6

...
0.'

0.2

10 15

FIGURE 2: Probability of correct observation of more loaded side

in dependence of the measured asymmetry
d) There exist almost no correlation between the measured and the rated load
asymmetry (-0.39 ere 0.21, n=8).
It has been observed, that some of the subjects where rated more serious
because of their extraordinary gait pattern, although they showed less
measured asymmetry than others. Again it has been difficult for the jury, to
separate the observation of a single parameter from the observation of the
overall impression. Therefore the use of a measuring device to define the
degree of load asymmetry is appropriate.
The problem of the interpersonal differences of the different members of the
jury in the evaluation of gait asymmetries has already ben mentioned several
times. The reliability of the observations of each jury member concerning
the correct side is shown on the table below.
TIME LOAD
Jury correct side correct side correct side correct side
Nr. all trials trials <;5% all trials trials <;5%

1 67 83 18 a
2 55 60 30 67
3 100 100 44 33
4 74 66 44 67
5 100 100 60 75
6 89 100 36 25
7 82 100 33 50
8 73 83 54 75
TABLE 2: Re 1i abil ity of the jury members concerning judgement of the correct
side which i~ loaded longer respectively more
568

With the exception of load asymmetries above 5%. the jury can be devided into
two groups: The one consisting of five people which work on patients day-in
day-out shows significantly better results of observation than the one with
three members (Nr.1.2.7) which have also a lot of administrative and teaching
responsabilities.
4. CONCLUSIONS
It has been shown in this study. that out of four chosen asymmetry parameters
only the subjective observation by the selected jury of the longer loaded
side during gait was reliable. It is therefore indicated to perform
measurements of time- and load asymmetry whenever a quantification of limping
or unloading is desired. The synthesis between precise quantitative
measurements and extensive subjective gait observation has shown to be
adequate in order to obtain a most optimal result.
In this study. those jurymembers which do observe more regularly show better
results than the others. For better support of this finding. further studies
with more jurymembers should be conducted.
The utilized registering method which is inexpensive and easy to use has
shown to be an appropriate tool to measure the defined asymmetry parameters.
It has shown to be an appropriate training tool for other movements than
walking too as for example climbing stairs.
The restrained informations on maximal vertical ground reaction force and
contact time make it meaningfull to extend the described bio feedback tool
for further use. as for example the detection of the localisation of the
maximal vertical force. a comparison of the peaks occuring during the phases
of deceleration and acceleration or a detection of the required time until
full load is born on an extremity.
5. REFERENCES
1. DEWAR M.E (1980) Temporal asymmetry as a gait quality indicator.
Med.&Biol.Engineering and Computing. 1980.18.689-93.
2. STUESSI E. ET AL. (1985) Analysis of the doublestride in walking.
Proceedings of the X. Intern. Congrss of the ISB (Umea.Sweden)
 569

IMPULSE CHARACTERISTICS IN AIDED GAIT AND WALKING AID DESIGN

K.A. Opila*. A.C. Nicol** and J.P. Paul**
New York College of Osteopath. Med.*. Old Westbury. New York.
U.S.A.; Univ. of Strathclyde**. Glasgow. U.K.

Walking aids in the form of canes, elbow crutches,

axillary crutches and walking frames assist locomotion which
has been affected by arthritis, stroke, fracture, joint repla-
cement, paraplegia, and other neurological and musculoskeletal
disorders. While aid usage is usually short-term, in longterm
aid users such as arthritic and paraplegic patients, the type
and sizing of walking aids and gait patterns prescribed can
accelerate degenerative joint disease. Because of this need
to protect lower and upper limb joints in aided gait, this
study was undertaken firstly, to quantify the function of
walking aids in gait and secondly, to analyze the interaction
of the aid force vector with upper limb joint centers to
determine moment loadings~
While observing patients with various disabilities
ambulating with a range of walking aids and gait patterns, it
became obvious that aids were used for three separate
functions: stability, support, and restraint/propulsion.
Aids and gait patterns are prescribed to meet one, two, or all
three of these functions as required by the patient. Added
constraints in aid prescription are the demands placed on the
upper body from loads transmitted by the aids. Thus a walking
aid must be designed both to achieve its intended function in
gait and to minimize the loads on the upper limbs and shoulder
girdle. Since these two requirements are interdependent, it
was decided to analyze both the contribution of aids to
locomotion and the upper limb loadings simultaneously.
Two previous papers by the authors presented some of the
early results of this study (Opila et al 1984, 1985). These
will be summarized here for convenience. In the first paper
(Opila et al 1984), upper and lower limb loadings of a post-
operative total hip replacement patient using elbow crutches
were presented. Two important conclusions are that 1) aid and
limb forces, resolved into the fore-aft direction (which has
been done infrequently in the literature) are highly descrip-
tive of the contribution of aids and limbs to the
restraint/propulsion function and 2) upper limb flexion-exten-
sion moments can be in error as much as 100% if the deviation
of the aid load from the longitudinal axis of the aid shaft
(peak value 2.5 degrees) is not measured. The second paper
570

(Opila et al 1985) presented aid and limb forces in the fore-

aft and vertical directions of total hip replacement (THR),
tibial fracture, and paraplegic subjects. The diversity in
the use of canes (THR) and elbow crutches (other two subjects)
for restraint and propulsion is noted, particularly in terms
of magnitude and symmetry of early and late loading phases.
It is commented that the analysis of aid and limb force-time
traces becomes unwieldy with numerous types of aids, gait
patterns, and disabilities. Shoulder moment loadings due to
the aid force were shown to be as high as 60 Nm tending to
extend the humerus of the paraplegic subject.
In the continuation of these two papers, the impulse
method for characterizing aid functions and the moment load at
the shoulder joint in reference to aid design are presented
here.
INSTRUMENTATION: The shafts of a pair of aids were strain-
guaged to measure axial and two shear components of force.
The lower portion of the aid was adjustable for height and
could accommodate elbow crutch and cane handles. Three
markers were placed on the aid to monitor its orientation.
The calibrated aids were "walked" across a Kistler
forceplatform and showed agreement with forceplatform values
to within 5 percent.
METHODS: Three groups of subjects were selected who use
walking aids principally for stability, support, and
restraint/propulsion. Respectively, these were post-operative
total hip replacement patients using canes with a four-point
alternating gait pattern, tibial fracture subjects using elbow
crutches with a swing-to gait pattern, and paraplegic subjects
using elbow crutches with a swing-through gait pattern. Sub-
jects walked at their preferred speed across a Kistler
forceplatform. Three-dimensional aid loads, aid orientations,
shoulder, elbow and wrist joint centers of left and right
sides, and ground reaction force data were collected synchro-
nously at 50 Hz.
ANALYSIS AND RESULTS: Aid forces were resolved into three
components: vertical, anterior-posterior, and medial-lateral.
Aid and limb forces for one trial of one subject from each
group are shown in Figures la, b, and c. These results are
not necessarily representative of individual subjects nor of
groups as a whole. In analyzing these figures, particular
attention should be given to the relative magnitude of aid and
limb forces and symmetry of early and late aid and limb
loadings.
The results of the THR subject (Figure la) show that the
affected limb is used for restraining the body (greater magni-
tude of force and more than 50% of the cycle being spent in
deceleration) and the good limb is used for propulsion. The
cane held in the hand opposite the affected limb, loaded
simultaneously with the limb, is shown to initiate propulsion
 VEAr/CAL
vEAr,CAL
VEAr/CAL FY
FY FY {AI/4w}
lHUTrf) 'POf) {N/4 W }
{N/4 w} (II
(.(1
(.(1

",'., ...1\
I'
/ ..,../

us

U'-lFIlIt~

FX FX
FX {N'/4W}
{Al/4W} {Al/4w}
-11./11
-11./11
.11./11 -lAFT

1.11 -lLATERA L
'.11 -lLATCRA L
1." -l LA URAL

FI FI
FI (N/4",)
{N/4.v} ~
(N/4"')
- D." -l "'CD/AL
- './D -l ",CD/AL
- 1.11101 "'CD/AL
(b) (c)
(3)

contra latera l (- ._)

Figure 1. Forces on (3) limbs (--) , ipsila teral (- --) and(b) limbs (--) and
perativ e,
canes of a TIIR patien t 11 days post-o
in swing- to gait
elbow crutch es ~~~) of a tibial fractu re subjec t crutch es (::.::-=)
10 weeks post-tr auma, and (c) 1 inlbs (--) and elbow V.
years post-tr auma. -I
of a parapl egic subjec t in swing- throug h gait 1~ -
 '"~

IMPULSE o.
10 aD.0 IMPULSE 00 00 00 IMPULSE
---
... .I. .I.
o o0 ( N-I) 0 0
(:~) (~~) -OW ......
1 I VERTICAL 1. VERTICAL

0.63 0.63

~ In II In ~ In
H1 F1
0.04~ PROPULSION 0.08 ~ PROPULSION
'T~'fi~
0.08~PROPULSION
P1

0.026
0.02 0.04
0.026 0.04
0.025

-0.027 -0.027
-0.02 -0.04 -0.04
-0.027
-0.04~nESTRAINT -0.08~RESTRAINT
I _
_--
-0.02 1 _....tTTr
-0.04 MEDIAL -0.08 -o.oJ ~ D ~
MEDIAL lMEDIAL

(a) (b) (c)

Figure 2. Impulses on limbs (broad bands) and canes and elbow crutches (narrow bands) of
(a) a TIIR patient 11 days post-operative. (b) a tibial fracture subject 10 weeks
post-trauma. and (c) a paraplegic subject 11 years post-trauma.
 573

prior to the affected limb and has a force value nearly as

large as the affected limb. In the results of the fracture
subject (Figure Ib), the good limb provides a force of more
than twice that of the aids in restraint and propulsion. This
contrasts with the paraplegic subject's aid usage (Figure lc)
whose aid and limb forces in propulsion are nearly equal. The
medially directed aid force of the paraplegic subject is
highly symmetrical in early and late loading as compared with
the fracture subject. This emphasizes the importance of the
stability function of walking aids in paraplegic gait.
Impulses of aids and limbs in the three directions were
calculated by integrating the force-time traces. This method
of representing force-time information has been used to a
limited extent in aided (Bennett et al (1979)) and unaided
gait (Seliktar et al (1979)). In locomotion at constant
velocity, the net impulse is zero -- a trivial result. Of
interest is the contribution of aids and limbs to the
restraining and propelling phases of gait, which varies with
type of aid, gait pattern, and individual style. This is
quantified in the histograms of Figures 2a, b, and c for the
three subject groups. Three gait cycles for each subject are
presented. The first set of histograms for each of the
subjects correlates with the force traces of Figures la, b,
and c respectively. The magnitude of the impulse histograms
representing restraint/propulsion and medial-lateral stability
particularly should be compared. In the paraplegic subject,
the impulses describing restraint/propulsion are nearly equal
to those describing stability. Indeed, in half of the
paraplegic and fracture subjects, the stability impulses were
greater than the restraint/propulsion impulses. The arrows on
the ordinates represent the impulse values of the unaffected
limb of 15 total hip replacement patients twelve months post-
operative.
The impulse results suggest j
that the function of the walking
aids of the paraplegic subjects was
not singularly to propel but also
to stabilize the body. It was
hypothesized that if the stability
function cQuld be provided for by
the geometry of the aid, then upper
limb musculature could be used for x
restraint/pro-pulsion rather than
stability functions. A design MODIFIED
modification of the elbow crutch STANDARD
ELBOW ELBOW
was conceived and fabricated which CRUTCH CRUTCH
angled the shaft portion of the aid
laterally. With this alteration,
loading the crutch with the arms
near the body would result in a
force directed medially stabilizing Figure J. Elbow crutch design modi-
the body. As Figure 3 shows, this fication to enhance medial-
lateral stability.
modified configuration caused upper
574

limb loadings which were biomechanically unsound. In one case

the aid force passed medial to the shoulder which was
inconsistent with measurements; in 100 out of 108 cases,
subjects loaded their aids to cause the aid force to pass
. lateral to the shoulder. In the second situation where the
load line passes lateral to the shoulder as in aid use, the
force also is shown to pass lateral to the wrist, placing
large moment loads on the wrist joint. Two fracture and one
paraplegic subject who attempted to use the aid reported
significant wrist instability. Thus the interdependent
criteria for aid function and the ability of the upper limbs
to transmit aid loads to the trunk were not met simultaneously
in the modified elbow crutch design.
CONCLUSION: In summary, the resolution of walking aid and
limb forces and impulses into three directions was shown to
be a highly descriptive method of assessing function in terms
of stability, support, and restraint/propulsion. The
stability requirement of walking aids for paraplegic subjects
prompted a modification on the elbow crutch design to increase
the medial-lateral base of support. The modified design had
limited success in use due to wrist instability. This
reinforced the requirement that the aid design must optimize
function with the ability of the upper limbs to transmit
loads.
ACKNOWLEDGEMENTS: This work was carried out at the University
of Strathclyde. Staff and technical support of the
Bioengineering Unit are gratefully acknowledged.
REFERENCES:
Bennett Leon, Murray M.P, Murphy E.F, Sowell 1.1. (1979).
"Locomotion assistance through cane impulse." Bull Pros Res
10-31:39-47.
Opila K.A, Nicol A.C., Paul J.P. (1984). "Biomechanical analy-
sis of limb loads in aided gait using elbow crutches."
Developments in Biomechanics Vol.2 publ. Mortinus Nijhoff, The
Hague. pp567-572.
Opila K.A, Nicol A.C, Paul J.P. (1985). "Quanti fication of
the function of walking aids." loth ISB 15-20 June 1985,
Umea, Sweden.
Seliktar R, Yekutiel M, Bar A. (1979). "Gait consistency
based on the impUlse-momentum theory." Pros & Orthotics
International 3:91-98.
 575

UPON THE MECHANICAL LOAD OF ELBOW CRUTCHES

Chr. Boosfeld and U. Boenick
Department of Biomed. Eng. & Test Center for Orthopaedic Aids,
Technical University, Berlin, Germany

Abstract - The report presents a field-trial investigating

crutch loading under everyday conditions. The trial lasted for a
6-day-period/subject (13 subjects) and the acquired data was
evaluated in regard of obtaining appropriate test parameters for
a test procedure recommended by the DIN-committee.
A mechanical model of a forearm crutch is established, and the
way, the force and moments are measured with a light and porta-
ble, ~P-controlled device is explained.
The measuring and data storing technique, transferable to any
mechanical system subjected to irregularly recurring, stochastic
loads, is based upon the "Peak Counting Method" with threshold
and hysteresis.
The desirable surveillance, control and -if necessary- Off-
Line correction of the non-observed field measurements is made
possible by the electronic realization of a mechanical law: the
"Condition of Equilibrium".
To gain the test parameters, the averaged frequency distribu-
tions of each subject are normalized by the criterion of an
equal number of steps and then superimposed. Using the 95%-
confidence limit of the resulting "Group-Histogram", the test
force and its direction are calculated according to the test
procedure. Results: In average 1,756 steps were taken during a
7.3 hour test day. The vertical force, applied at the hand grip,
did not exceed 500 N in a 95% probability, although, 800 N were
reached in a few cases. A rule of thumb for the dynamic test
load applied at the grip -no "safety factor" taken into account-
is: 65 % of body weight (crutches bilaterally). Recommended test
load cycles: > 1 350,000.
1. INTRODUCTION
In order to prevent accidents caused by mechanical failure of
forearm crutches it is necessary to test these mass-produced
aids for their operational stability, functional efficiency and
ergonomic design prior to introducing them into service.
An initial proposal concerning a standardized method of test-
ing was published in 1979 by the "PrOfstelle fOr orthop~dische
Hilfsmittel der Technischen Universit~t Berlin". The resulting
preliminary DIN-standard 32 972 was based on this proposal,
however, all participants agreed that the data on quantitative
loading and frequency of utilization of these aids were not
sufficient. Therefore, further investigations into normal daily
use had to be carried out to determine exact test-load values.
576

2. ~EASURING APPARATUS
The portable measuring device consists of two components (Fig.
1): 1. A c rut c h wit h in t e g r at ed mea sur i n 9 p YI on s (s t r a in - 9 aug e
sensors) and 2. a measuring and storage system, which can be
fastened either to the trouser belt or to the crutch itself.

@ Measuring poln.ts FUF

eo Crossing (Reference) point K

FIG.l: Measuring gear, FIG.2: Bending(B) moment distribu-

left: pylons in detail tion and definition of model
The model of the crutch is based on a bOdy-related set of
coordinates with positive orientation corresponding to that
shown in Figure 2. Five "strain-gauge" Half-Bridges -two active
resistances of a total of four (Wheatstone)- were installed to
measure the bending (8) moments in the sagittal (5) and frontal
(F) plane of the upper (0) and lower (U) section as well as the
sagittal bending moment of the hand-grip (G). For determination
of the axial force (DRU) in the upright part of the crutch the
lower pylon was additionally equipped with a Full-Bridge con-
sisting of four active strain gauges.
Fig. 2 shows the measuring points marked by means of circles.
Bending moments measured at these points can be extrapolated by
application of the intersect theorem since the bending moments
are zero at the point of application of the force (bottom, grip,
cuff). Fig. 2 shows one possible distribution of bending moments
along the body of the crutch.
 577

The six-channel measuring and storage system was developed at

the Test-Center. The unit is composed of the mobile measuring
and storage device which the patient can carry himself. 6 inter-
mediate memories to which all measured information of each day
is communicated and the stationary evaluation computer (Fig. 3).
3. TESTING
A total of 3 measuring units were fabricated. After completion
of the first unit. laboratory experiments were executed in which
2 A/K amputees and 1 non-disabled person used crutches bilate-
rally on a treadmill. These pre-tests were essential to give
estimates of the magnitude. direction and phase relationship of
_. _.
the occurring loads and allowing calibration of the sensors .
......
---- / :.-
p

~~ "
,,-/\
rv
/\
P "" .
~ /'\ ,
IV 1'-'

,
Meuurlng and
Storage IYllem
""--
,n_ 1/4 ......
..I
114 ....

~-----il
~
Evw.llon Computer

FIG.3: Schematic illustration of FIG.4: Analog-oscillograph of

components of the system the 5 bending momentsandthe
a x i a 1 for c e ; ( tim el a p s e : 1 s t e p )
Fig.4 illustrates typical curves of 6 measured values over the
period of one pace qualitatively correlated to the "sagittal"
3ngular position Lt of the crutch (below).
The "duration of support" can be divided into 3 distinctive
phases. characterized by both the varying magnitude and sign of
the mom e n trs (f 0 r c e ) and the po sit ion Lt..
At the start of the step the aid is brought forward and after
"Touch-down" (tip on the floor) phase I begins whereby the lower
part and the grip carry the weight. The forearm is loose and
unstable in the arm cuff -thus. frontal bending is high- and
there is no support of the arm in the sagittal plane (850 :: 0).
578

Turning the crutch to where a approaches zero consumes approx.

half of the total(lsec.)time lapse. The maximum value of the
axial force (DRU) corresponds to the beginning of phase II and
the load upon the upper part increases gradually, stabilizing
the arm.
Considering the system as being (sagittal)-planar and "qua-
sistatic", the conditions of equilibrium require that the sum of
all acting forces (moments) is zero for each single instant and
given reference point. Thus, due to the increase of BSO and BSG
during phase II/III, BSU must decrease -and here even changes
sign-.
During phase III maximal load is transferred onto the crutch,
presuming "ordinary" use.

From July to September 1984 measurements were taken on a group

of 13 subjects for 6 days each. Table 1 contains specific data
on each subject:
No. of Body Mean Mean time
Subj ect Sex Weight(kg) Age Steps/day in use/day(hr)
1 m 87 60 3,408 5.7
2 m 90 64 1,771 7.3
3 m 81 48 565 8.2
4 m XXX XXX XXX XXX
5 m 85 61 856 5.7
6 f 60 34 1,456 8.9
7 m 79 17 1,213 7.4
8 m 71 71 3,029 8.0
9 f 50 14 721 12.2
10 m 66 66 2,526 10.4
11 m 79 58 599 7.3
12 f 65 29 161 1.6
13 f 53 22 737 5.1
14 m 79 64 5,786 8.2
Table l:Data of Subjects

Seven persons of the group constantly used crutches: six leg

amputees (Subj.No:l,2,5,8,10,11) who did not wear a prosthesis
and 1 paralysed person (3). Three patients (7,9,12) had sus-
tained minor injuries (ligament rupture/tibia fracture) and
needed crutches while walking with a cast.
The remaining three were:
- a hip exarticulated girl (13) who, during the test period,
walked without her prosthesis,
- a woman (6), recently amputeed and
- a man (14) who s e hip j 0 in twa s rem 0 v e d.
For each day, the load exerted on one crutch (either side) was
recorded in form of a set of six load-time histories, each hi-
stogram representing the occurrences of the peaks of the mea-
sured value during a particular step.
 579

4. ILLUSTRATION AND DESCRIPTION OF THE HISTOGRAMS

In tatal apprax. 420 daily f AxIal forea (DRU)(t)
histagrams were recarded. Fig. l'~
5 shaws an example af the hi- ...., DevlaUon
<1. <!>
stagrams far the faur maments
in the sagittal plane. -
The example shaws the re- _
suIts far subject 1, an A/K
amputee, who on that day walk- ~
ed 5 616 steps using the fare- ~
arm crutch. This value exceeds
by far that far average use.
Hawever, these laad-time hi- F""", N

staries were faund to. be use-

ful demanstrating tendencies.
The axial farce histagram
(DRU) gives infarmatian abaut
the effect af bady weight and
the accurring dynamic farces. 000
Often the latter are small and
the made af the distributian 000

carrespands to. appraximately 400

half af the subject's bady
weight when crutches are used
bilaterally.

One measurement cycle carrespands to. each step taken; there-

fare, the sum af absalute frequencies seen in the axial farce
histagram is equal to. the number af steps an that specific test-
ing day.
The bending mament BSG, at the paint where the hand-grip is
fa-stened, reaches its highest frequency at -21 Nm. Maments af
32 Nm accurred mast frequently at the upper part of the crutch
(BSO) .
A'Kpl Ying the candi tians af equ ilibrium, :tife difference (BSOK-
BSG ) results in the sagittal mament BSU ; thus, the sum af
these 3 values must equal zero.. The deviatian histagram delta(s)
-shawn in the upper right af Fig.5- indicates this sum, meaning
it shaws haw aften the peak value af the sum was nat zero..
That implies that delta(s) serves bath as a cantrol criterian
and a "praper use" indicatar: miscalibratian, large affsets ar
wire mal- and discannectian af ane af the 3 sagittal channels
wauld have led to. either a significant dis persian af delta(s) ar
"autliers".
On the ather hand, using the crutch impraperly, wauld also. have
been detected. In that case, the mechanical assumptians regard-
ing the paints af farce applicatian and the leverages are nat
met.
Due to. the fact that the bady weight is nat always evenly
distributed and applied at the hand-grip, a minar deviatian must
appear in the histagram.
580

5. EVALUATION
To obtain absolute
test-load values cn_d_'"
NumberoftlH
I
.0
n
2.
m
.,
lY i

"
iI
11

from the recorded ."

Bodywelght
72 II
70 II 70

histograms, the fre-

quency distributions
-..
8 ..~ ..k ... 1

115.
at

2021
,.
7

371.
'43'
had to be normalized
2031
da,
Time In u I
by the criterion of da, 7,' 7 7,1 7.'

an equal number of ORU

'N' 324,1 2 . 354.3 371,4 275,' 341.2

steps using a spe- il: ItIm)

aa _13.1 -11.1 -14 . -1',' ....7 -1'.S
cial computer pro-
gram. Thus, norma- J. ,ton' -0.2 -0.3 -0,1 -0.2

lized histograms of
each subject were
created and then ac-
g:.o ,.... ,
... 'N'
F ....
15.'

"17.1
13.3

.00
11.0

1512.S
17.1

512.5
14.0

'00 ...
14.4

cording to the u::: INmI -I' -22.& -21.1 -215.1 -21 -27

groups I to VI "'=:: tHIn!

34.1 27
28.5
shown in Tab.2- su- 474.' 317.' "".7 .to.l
perposed.
F,
,.,'N'
"' . 7
'"
The test force F f3 17.' t .... 17.' 17.' 17.' 17

and its directioR~ NLW 141000 1521000 740000 74'000 113000 1.387000

were then calculated

by solving the equa- L Group! AI Nit. .
arc..!
tions of equilibrium
L Patlenla Ighlng .... than 76kg
.. Group! P........ ...,.~ lIKIra ItIan 15"_

for the model of the IV. Group! Va"" (Longt .... unn)
V. Gn:Iwt!' Acddent Y1ctim. (Shortt ..... u,.)
YL Group! Active Patientl (Slepliday ccIq 2000)

Table 2: Group data, mean values and test

parameters (lower 3 lines)
crutch, appointing a "95%-margin" value for each histogram, i.e.
the value which is not exceeded with a probability of 95%. The
range of the specification factors F , ~ and the recommended
load cycles NLW -annual estimation- isPpresented in the lower 3
lines of Table 2.
6. CONCLUSION
Table 2 provides both a basic guidance for the manufacturer
telling him what a crutch has to withstand and specific test-
load values for the DIN testing procedure.
The value of 500 N for dynamic testing, assumed by the stan-
dard to be appropriate can be confirmed by this field tr~al.
The test force angle should be modified from the proposed 21 to
17.5 0 , whereby the vertex of this angle remains unchanged at the
middle of the grip Ef.65 mm).
A minimum of 1.35xl0 test load cycles should be executed.

REFERENCES
III Boosfeld, Chr.: Untersuchung der Beanspruchungsverhalt-
nisse an UnterarmstockstOtzen im Feldversuch zur Ermittlung
von PrOflastwerten. Diplomarbeit, TU MOnchen/TU Berlin 1984
 581

SIGNIFICANCE OF THE MECHANICAL POWER AT THE LOWER LIMB JOINTS

DURING WALKING BY HEMIPLEGIC SUBJECTS
C. Frigo and V. Licari
Centro di Bioingegneria. Politecnico di Milano. Fondazione Pro
Juventute Don Gnocchi. Milano. Italy

1. INI'ROOOCI'ION

A camprehensjve descrjptjon of the relatjve energy requjrements of the

varjous jojnts and segments of the body durjng a cycle of normal level
walkj ng can be found j n a report by Bresler and Berry (1951).
In that publjcatjon the authors state that an jnvestjgatjon of energy
requjrement of the leg must consjder not only the energy necessary to
actuate the leg j tself, but also the energy supplj ed by the muscles j n
the leg and transferred to other segments of the body. Thejr work on
normal subjects was ajmed at provjdjng a basjs for evaluatjng energy
requjrements of prostheses and to provjde same estjrnate of the power
avaHable to an amputee. OUr work, dealjng wjth subjects affected by
neurolog j c dj seases (hemj pleg ks) j s aj med at j nvestj gatj ng same features
of the functjonal jmpajrement resultjng fram muscle control defjcjency and
analysjng the campensatjng mechanjsms adopted by the controlateral sound
ljmb fram the energy requjrement pojnt of vjew.
The rate of energy exchange at each jojnt (jojnt power) seemed to us the
nore representatjve varjable for descdbjng the muscles jnvolvement, both
because j t represents the j nstantaneous muscle j nvolvement durj ng the
strj de and also because j t contaj ns the j nformatj on on wether a muscle
group js actjve jn concentrjc or excentrjc condjtjons.

'2. METHOD
2.1. Experjmental set-up

Kjnematjcal measurements were made by the ELITE system (Ferrjgno and

Pedottj,1985) on two normal subjects and fjve hemjplegjc patjents.
Reflectjve markers were plaCed on the apex of the Hjac crest, the greater
trocanter, the lateral fenoral condIe, the lateral malleoulus and the
fjfth metatarsal head. The subjects were asked to walk at natural cadence
on a pathway 15 m long covered by a conductjve allulldmjum plate. A thjn
sheet of metal was applj ed beneath the shoes j n order to detect the
temporal phases of each strj de. One lV-camera pj cked up the subject j n the
central part (2.5 m fjeld ampljtude) of the walkjng pathway where a force
platform was embedded on the floor. The ELITE system elaborated on Ij ne
the lV-sjgnal at a frequency of 50 Hz and suppljed a mjnj-camputer wjth
the coordjnates of each marker. Sjgnals fram the force platform and from
the contact phases analyser were acqujred jn the meanthne through an AID
converter.
582

2.2. Elaboration prOCedure

Rough kinematical data were calibrated and filtered as described by
Ferrigno and Pedotti (1985). Marker coordinates were then reduced to the
six free coordinates that are sufficient for geometrically describing a
four rigid links system equipollent to the pelvis and lower limb system.
The free COOrdinates arbitrarily chosen were the two hip coordinates and
the four angles of the pelvis, the thigh, the shank and the foot with
respect to the vertical.
First and second derivatives were camputed for each free coordinate
according to the method proposed by Cappozzo et ale (1975).
Anthro);X:llletri c parameters such as the mass and the i nerti a m:::ment of each
lower limb segment were estimated for each subject on the basis of
anthro);X:llletric tables (Drillis and Contini, 1966) and direct measurements
on the subjects (length of the segments, hei ght and total wei ght)
Joint torques were thus obtained by applying the D'Alembert principle
to the system of dgid links defined above. Accordingly, the mechanical
action of the reaction forces measured by the force platform were
conSidered as well as the weight of the segments of the lower lirrb and the
forces and moments of inertia.
The joint power was then camputed in each instant of time by simply
multiplying the joint torque by the relative angular velocity between
two adyacent segments. A sign convention was adopted by considering of
positive value the power prodUCed by muscles (torque and angular velocity
on the same directjon) and negative the power absorbed.
3. RESULTS
On the two normal subjects examined the time course of the joint power
showed the same basic features, even j f different absolute values were
observed depending on the dHferent shes and on the individual patterns
of movement (fig. 1). At the hip and knee joints three typical phases are
observed charactedsed by a) production of power at the hip and
absorption at the knee at the beginning; b) absorption of power at the hip
and production at the knee in the middle stance; c) production of power at
the hip and absorption at the knee in the last stance/early swing. The
remai ni ng swi ng phase is characted sed by a power absorpti on at both the
joints starting earlier at the hip than at the knee, justified by the need
of braking the forward movement of the thigh and the shank. At the ankle a
phase in whi ch power is absorpted lasts approxi matel y 40% of the stri de.
Then a production of power gives rise to a typical lobe peacking at 50-53%
of the stride corresponding to the push-off phase. During the swing phase
and the early stance the power curve is very close to zero. All the above
features are strictly in agreement with data fran Bresler and Berry
(1951). Quite different was the pattern of power requirements in the
hemiplegic subjects depending on the level of impairment and on the degree
of functional recovery. Fig. 2 shows bianechanical data fran a subject 56
years old affected on the right side as a consequence of a spontaneous
cerebra-vascular accident occurred one year before our examination. At our
analysis resulted a marked asymmetric gait. The stance duration was 80% of
the stride duration on the sound limb and 67% on the affected limb. Such
an asymmetry is reflected also in the ti me curves of the joi nt torques and
joint powers.
In particular, concerning the joint torques higher values are achieved
both at the hip and the ankle Joints of the sound limb, while on the
affected side sane oscillations are shown probably connected to clonic
contraction of muscles. The joint power curves show a big production of
energy at the sound hi p duri ng the most of the stance phase and the
 583

begjnnjng of the swjng. On the plegjc sjde developrrent of power js very

Ijttle, except jn the last stance. Surpdsjngly jn thjs pedod the
opposjtlon of phase between hjp and knee power development js roughly
preserved as jn normal subjects. Instead on the sound Ijmb an abnormal co-
absorptj on of power j s seen j n the last stance. Another j nterestl ng
feature js presented by the knee at the begjnnjng of the stance, where
power j s produced j n order to extend the knee after heel strj ke ( see also
the stjck djagram above).
A stdkjng dHference between sound and plegjc Ijmb js shown alsO at the
ankle joj nt, where a quj te normal tl me course of the joj nt power j s
realjSed at the unaffected sjde, whjle a large negatjve lobe followed by a
small pushjng power js shown at the affected sjde. A sjmHar behavjour was
observed j n three of the subjects examj ned, all characted sed by a marked
asyrrrnetry. On the other two subjects, rrore syrrrnetd c, we could observe two
dHferent behavjours: one of them (the less jrnpajred) was able to produce
a power tjme course sjmjlar to the normal also concernjng the ampljtude
(fjg. 4), the other one showed very Ijttle power jnvolvement on both sjdes
(fj g. 3). Anyway a hj gher energy exchange occurred at the ankle of the
sound Ij mb than at the opposj te sj de

4. DISCUSSION
On the fjve subjects examjned dHferent patterns of movement and power
development have been found connected to the djfferent level of functjonal
jrnpaj rment. Nevertheless sane ccmnon features could be observed j naIl
of them. Fjrstly the plegjc Ijmb was characterjSed by power developrrent
lower than jn the sound Ijmb along the whole strjde. Secondly the more
altered curve, wjth respect to the normal subjects was jn general that of
the hjp at the sound Ijmb. Thjs was characterjSed by a hjgher power
productj on for the most of the stance phase and can be j nterpreted as a
mechanj sm to j nject power j nto the system j n order to canpensate for
muscular defjcjt of the plegjc Ijmb. Thjrdly a roughly normal tjme course
of the power at the ankle js preserved,even H dHferent relatlve
ampljtudes between the negatjve and posjtjve phases are present. In the
three subjects characterjZed at the sound Ijmb by a marked knee flexjon at
the heel strjke,the followjng extensjon requjres a product jon of power by
the extensor muscles. Thj s mechanj sm seems to be dj sadvantageous for the
energy requjrement of the sound Ijmb,but probably js requjred to overcame
the defjcjt jn power product jon of the plegjc Ihm.

5. CONCLUSIONS
Thjs preljmjnar data are very pranjshjng for the possjbjljty of
jnterpretatjng sane characterjstjc features of the pathologjcal gajt and
canpensatjng mechanjsms,bejng connected to the problem of muscular
coordjnatjon and optjmjzatjon of the resjdual muscular capabjljtjes.
Further studj es j n thj s dj rectl on and correlatj on wj th other parameters
are necessary jn order to obtajn jnformatlons useful for cljnjcal
appljcatjons.

REFERENCES

1. Bresler B,Berry FR: Energy and Power jn the Leg durjng Normal Level
Walkj ng. Report on the Prosthetj c Devj ces Research Project, Instj tute
of Engjneerjng Research, Unjversjty of Caljfornja, Berkeley, serjes
II,jssue 15, 1951.
584

2. Cappozzo A, Leo T, Pedottj A : A General canputjng Method for the

Analysjs of Human Locanotjon. Journal of Bjanechanjcs, 8, pp 307-320,
1975.
3. Drj 11j s RJ, Cont] nj R: Body Segment Parameters. Technj cal Report
no 1166, 3, School of Engjneerjng and Scjence, New York Unjversjty,
1966.
4. Ferrjgno G, Pedottj A: ELITE: a Djgjtal Dedjcated Hardware System for
Movement Analysjs Vja Real-tjme TV Sjgnal Processjng. IEEE Transactjons,
BME,32-11, pp 943-950, 1985.

NORMAL SUBJECT

NORMAL SUBJECT 2

CWJ 50
o
50
50
CWJ ,.,.- ...
0

50
100
100
50
CWJ ~~~ __________~~____~~____________~ANKLE
I
0 I
I
..........-"
513

ex 50X 100X

FIGURE 1. Jojnt tjme course fran two normal subjects walkjng at free
speed.
 585

tNmJ 100
o
~
jo~~ --JH
t '"'Or-.
I

: <===1
tNmJ
::~
0~
50
~ ! :~
1
IK ~ ............ c J
~I
: :
tNmJ
100~ ~
0_~1 : j
A_~::
1 1
I 1
1
59~
5:~Vrr=--IH}=V~ -4
I 1
t\.lJ

t\.lJ 50
9~
j l : j :
59
'IK
""'J! -=
1
:
I
"V: '
<=>

t:\.lJ ~
50

0 :
( \1 ,

:A J-+-___--------,~c-=-....=-.:_----
~ IJ ~

2
50 ,
I 1

[STANC~ i","GJ FE;=___S_T_A_I-I_C_E--.:==:::l..._5_\.l_I_N_.G-I

50X leex e sex 109

FIGURE 2. From top to bottom: stjck djagram and vector djagram jojnt
torques at the hjp, knee and ankle; jojnt power at the same
artjculatjons; strjde temporal phases of both ljrnbs. Sound and
affected ljrnb of a hemjplegjc subject wjth marked degree of
spastj cj ty.
 586

SOUND LIMB PLEGIC LIMB

[\lJ , I

HJ~
I I
S:f' c;=-
, ~
,
I

, ,,
I
4

,
5:]
50
,,
I I

[\lJ I I

KJ
I
I

V.I
I

50 "=:J. I

5:1 ~i
I

[\lJ
A~
Aj ~!I
I
I
50 I
I
I I
STANCE ,I
I S\lING
I
I STANCE I S\I.ING

E F dE
I I
,
~
I
I

~
0X S0X 100% 0X 50% 100X

_ FIGURE 3. Joint rx>wers of a hemiplegic subject wi th mHd degree

of spasti cj ty. Both lirms analysed.

SOUND LIMB
PLEGIC LIMB

[\lJ
5:J~ I I

50 ]
v~ ~ I
I
I
I I
50

==---~\7.~:--..:::--<====1~:
[1.0
9+oc----r"'==---=::..,.----:.._~-...s
K -+--=]
59

199
,
I

[\lJ 50
,,
I

9 ~~~~=----+--~----~ A
I
I
50
STANCE : SUING S\lING

E ,7? IE
I

~ --Lr-r;='r---_ ---'~
0X say, 100X saX 100%

FIGURE 4. Joint rx>wers of a hemiplegic subject with slight degree of

spasticity. Both limbs analysed.
 587

THE ROLE OF MUSCLE FORCES IN THE INSTABLE HUMAN KNEE JOINT

IN VITRO
H. Kiefer, L. Dtirselen and L. Claes
Lab. for Exper. Traumatology, University of Ulm, Ulm, Germany

SUMMARY
Anterior drawer signs (ADS)[ anterior tibial displacement in
[mmll were measured in human cadaveric knee joints in 90 0
flexion in vitro. Section of the anterior cruciate ligament
(ACL) led to an increase of the ADS up to 180%. Additional
section of the medial collateral ligament (MCL) raised the ADS
to 470%. Carbon fiber ligament replacements reduced the drawer
signs to initial values. Each of the 3 muscle groups of
the knee joint reduced the ADS: the quadriceps femoris by 20%,
the gastrocnemius and the pes anserinus group by around 50%
each. In combination the 3 muscle groups minimized the ADS to
about 20% of the values in the intact knee joint even when ACL
or both ACL and MCL were cut. Thus, postoperati ve
physiotherapy should preferably include training of the
hamstring and calf muscles.

I NTRODUCT! ON
Various experimental investigations under different
conditions have been carried out to determine the instability
of human cadaveric knee joints in vitro (4, 5) and in vivo (6,
11). In recent years the importance of the thigh muscles has
been learnt from clinical experience (7, 12). To strengthen
the quadriceps muscle, particularly after sport injuries,
special training is recommended (1, 10). Not much attention
has been paid to the hamstring and calf muscles so far.
Especially, quantitative data on their effect is not
available.

OBJECT! VE
This study aimed to measure the anterior instability (i. e.
the anterior tibial displacement with respect to the femur) of
fresh cadaveric knee joints with defined ligamentous injuries,
as well as the stabilizing influence of the knee spanning
muscle groups. Finally, the effect of surgical re-
stabilisation, using a ligament prosthesis, was determined.

MATERIALS AND METHODS

Eight fresh human cadaveric knee joints [mean age of 44
years, range 18-72 yearsl were prepared wi th respect to the
capsule and the ligaments. They were stored deep frozen until
the date of investigation.
The specimens were clamped in a mounting support in 90 0
flexion position (fig. 1), allowing free rotational and
varus/valgus position (2, 3). The tibia could be set in
SSg

neutral as ~ell as in hoth 15@ internal and external

rotational position with respect to the axis of the femur. The
tihial head was pulled anteriorl~ by a 40 N force causing an
anterior drawer sign (fig.2). For the knee bend the
stabili~ing muscle forces bave been calculated by Rahrle CS,.
For the load case, simulated here, fractions of the forces of
the 3 major muscle groups [quadriceps femoris, gastrocnemius
and semitendinosus-semimembranosus-biceps femoris muscles,
(referred to bere as pes anserinus group)] were applied by
computer controlled plulumatic cylinders (2,3, fig. 3).

fig. 1: ~ounted knee specimen fig.2: Amount, insertions

in tbe test apparatus and directions of
in !H)@ flexion. the muscles and
Anterior tibial dis- tensile forces.
placement is measured.

Since anchorage of the

ropes and cortical srews
(fig. 2) at tbe normal
insertions was not able to
transduce tbe complete
forces, only 10% of these
physiological tensile forces
of quadriceps (120 N),
gastrocnemius (50 N) and pes
anserinus muscles (25 N)
were applied. ~easurements
of the anterior drawer
signs (ADS) were taken in
neutral, external and inter-
nal rotational position from
the intact knee joint with-
out muscle forces and after fig. 3: Computer controlled
the application of single pneumatic cylinders
muscle group forces as well simUlate the muscle
as their combination. After forces.
 589

cutting the anterior cruciate ligament (ACL) the ADS

determinations were repeated and compared with the data
obtained after restabilizing the knee joint using a
commercially available carbon fibre ligament prosthesis.
Finally ACL and medial collateral ligament (MCL) were
resected and both replaced by carbon fibre ligaments.

RESULTS
The mean ADS and the standard
error of mean (SEM) was found
in NR 2,9 % 0,6 mm in the
t
N ..... t ... 1
E ute,...'
J .lntemlll
rot.tkxl
intact knee joint.
in ER were
The results
2,0 % 0,5 mm, and
2, 2 % 0, 5 mm i n I R,
CUT respectively. Cutting the ACL
increased the ADS to 180% and
with an additionally sacrified
5 MCL up to 470% in NR (fig. 4).
Carbon fibre ligament
replacements of the ACL, and
ACL plus MCL, respectively,
N E J N E J
decreased the ADS to almost
Aa.
+ initial values of the uninjured
MeL-
knee joint (fig. 5). In NR
fig. 4: ADS (anterior tibial isolated quadriceps force
displacement) in 3 reduced the ADS to 76% when
rot. posi ti ons for ligaments were intact.
intact and cut ligg. Gastrocnemius and pes anserinus
muscles decreased the ADS to
45% and 38%, respectively. Their combination led to minor ADS
of 21 % (fig. 6). After resection of the ACL or both ACL and
MCL the simulation of the quadriceps reduced the ADS to values
which still remained on the levels of 120% and 300%,
respectively. However, the gastrocnemius, the pes anserinus,
and the combined muscle forces diminished the ADS to values of
the intact joints (fig. 6). The influence of the muscle forces
in ER and IR was only little less and qualitatively similar.
The detailed results are plotted in fig. 7 and fig. 8.

neut ...1 rot8t1on

fig.5:

Straight ant .
CUT
instability.
Ligaments cut
400
and replaced
by carbon
fibres.
200 ADS ( anteri or
tibial dis-
placement)
in intact
Intact AoCL,.
.a...
AQ..
ACL+MCL
joints = 100%
590

(mm) neutral rotation

ACL+MCL.

fig. 6:
Influence of the
ACL. muscles on the
straight anterior
instability. 90 0
C-FIB
flexion knee joint,
C-FIB
8
intact
ACL ACL+MCL neutral rotational
position of the
ti bi a.

fi g. 7:
Stabilizing effect
of the muscles on
the antero-medial
rotatory instabi-
lity. 90 0 flexion
knee joint, 15 0
external rotation
of the tibia.

(mm Int.rnel roteUon

ACL+IICL.

ACL.
fig. 8:
Effect of muscle
C-FIB
forces on the knee
II
Intact C-FIB
ACL
ACL+MCL stability in 15 0
into rotation. 90 0
flexion position
of the knee joint.
 591

DISCUSSION
Standardized test conditions allowed quantitatively
reproducable measurements of the influence of muscle forces on
the anterior instability in human knee joints. The values
(fig. 2) of the muscle forces have been derived from finite
element analysis for the knee bend (9). These forces up to
1200 N can not be introduced into the ligamentous origins
since the screws would be pulled out of the bone. However,
preliminary tests have shown similar results of the ADS when
5%, 10% or 20% of the calculated muscle forces were applied.
Consequently, the anterior tensile force at the tibial head
could be reduced to 40 N. Thus, usi ng 10% of the calculated
forces general validity of the obtained results may be
assumed. Compulsive forces on the specimens could be avoided
by the bearing of the mounting support (2, 3). A 90 flexion
position of the knee joint was chosen in this test because the
lever arms of the muscles are the longest and their effect
could be shown best in this position. in addition, the
anterior dislocation of the tibia is mostly evident in 90.
According to the literature (4, 5), section of the ACL leads
to positive anterior drawer signs in all of the tested
rotational positions. The ADS was enhanced when the MCL was
cut additionally. In this experiment anterior stability of the
knee joint could be improved significantly (p<O,01) by carbon
fibre ligament replacements for the ACL and/or ACL plus MCL.
The stability obtained can be compared with the values of the
intact knee joints.
Clinically, training of the thigh muscles commonly means
training of quadriceps (1, 10). Simulating isolated quadriceps
muscle force, however, reduces the ADS by about 15-25%
depending on the actual angle of rotation and the extent of
injury. This is valid for both, the intact and the instable
knee joint. The anterior drawer was significantly diminished
(p<O,01) by isolated tension of the gastrocnemius or pes
anserinus groups. The same effect could be observed by a
combination of these muscles with the quadriceps group. Here
the reduction effect was up to 75-80% for the ADS. This was
also true after resection of both the ACL and MCL structures.
The higher stability compared to normal ligaments through
carbon fibres for the ACL stabilisation (fig. 5) can be
explained by 3 knee specimens from older indi viduals. They
were, although macroscopically intact, somewhat more lax and
showed a triplicated ADS with uninjured ligaments, compared to
the remaining 5 joints. By ACL replacement so the anterior
stability could be improved.

CONCLUSIONS
In consequence of the presented results, in rehabilitation
after ligamentous injuries of knee joints attention should be
paid not only to the strengthening of the quadriceps muscles
but in particular, to the gastrocnemius and pes anserinus
muscle groups. This is also recommended by other authors (7,
8, 12). In the simulated knee bend, simUltaneous contraction
of all knee spanning muscles provides maximal dynamic
592

stability, even in gross anteromedial instability. The static

stability can also be improved up to a normal level by a
braided carbon fibre ligament prosthesis for the ACL and MCL.

REFERENCES

1. BrUckner H., BrUckner H. ( 1912) Bandplasti ken im

Kniebereich nach dem "Baukastenprinzip". ZBL. Chir. 91:
65-11
2. Claes L., DUrselen L., Kiefer H. (1986) Influence of load,
flexion and muscle forces on the stress and strain of knee
ligaments. Transactions of the 32 d Annual Meeting of the
ORS, New Orleans : 238
3. DUrselen L., Claes L. ( 1985) Simulation und Messung der
Beanspruchung des Kniebandapparates, rechnerkontrollierte
Steuerung eines Simulators und On-line
MeBwertverarbeitung. Biomed Technik 30: 46-41
4. Fukubayashi T., Torzilli P. et al (1982) An in vitro
Biomechanical Evaluation of Anterior-Posterior Motion of
the Knee. J Bone Joint Surg 64 A: 258-264
5. Furman W., Marshall J., Girgis F. (1916) The anterior
cruciate ligament. J Bone Joint Surg 58 A: 119-185
6. Kennedy JC., Fowler PJ. (1911) Medial and anterior
instability of the knee. J. Bone Joint Surg 53 A: 1251-
1210
1. MUller W. (1982) Das Knie. Springer, Berlin Heidelberg New
York
8. Paulos L., Noyes FR., Grood E., Butler DL. ( 1981> Die
komplexe vordere Knieinstabilitat. Thieme, Stuttgart New
York
9. R6hrle H. , Scholten R. , Sollbach W. (1911)
KraftfluBberechnungen in Knochenstrukturen und Prothesen,
Phase II, BMFT. Bericht 01 VG 106-ZK14MT261
10.Smillie JS. (1985) Kniegelenksverletzungen. Enke,
Stuttgart
11. Torzilli PA., Greenberg RL., Hood RW., Pavlov H., Insall
IN. (1984) Measurement of anterior-posterior motion of the
knee in injured patients using a biomechanical stress
technique. J Bone Joint Surg 66 A: 1438-1442
12. Wirth CJ., Jager M., Kolb M. (1984) Die komplexe vordere
Knieinstabilitat. Thieme, Stuttgart New York
DYNAMICS OF THE MUSCULOSKELETAL SYSTEM
 595

RELATION BETWEEN MECHANICAL LOAD AND EMG-ACTIVITY OF SELECTED

MUSCLES OF THE TRUNK UNDER ISOMETRIC CONDITIONS
A.J. Spaepen*, W. Baumann** and H. Maes*
Inst. of Physical Ed.*, Heverlee, Belgium; Inst. of
Biomechanics, Deutsche Sporthochschule**, Koln, Germany

1. INTRODUCTION
A mechanical analysis of muscle and skeletal forces in the human body
often requires the use of a model. As also described by Andersson,
Schultz and others (1,2), elements of such a model include: the anato-
mical description of the body segments and the making of a number of
assumptions concerning ligamental forces and the simultaneous action of
flexor and extensor muscles. In this paper an alternative technique is
described in which the isometric activity of selected muscles of the
trunk is quantified and compared to the net moment of force in the lum-
bar section around vertebra L5. Results from such an analysis may then
be used to verify the assumptions or to find an additional input for
the model.
2. METHOD
To this end, the following experimental set-up was used. Three male
subjects (between 20 and 30 years of age) participated in a series of
tests in 10 different body positions, ranging from an upright position
of the trunk to a forward inclined position of approximately 70 0 with a
support for the hands placed at three different levels. In order to
enable a larger range of loads on the trunk, the feet were fixed to a
Kistler force platform by means of a pair of ski-boots (figure 1).
Vertical forces on the four load cells and horizontal forces in the
sagittal direction were sampled at 500 Hzwith a 12-bit AID-converter.
Spatial coordinates of LED-markers attached to the wrist, elbow, shoul-
der, thorax, waist, hip, knee and tibia (at a known distance from the
ankle) were calculated from the 12-bit data of a Selcom-II system re-
gistered at the same rate as the force plate data. In addition E.M.G.-
signals of M. rectus abdominis and M. erector spinae from both the left
and right sides were sampled. Anthropometric measurements in conjunc-
tion with a modified Hanavan model were used to estimate the segmental
masses and their centres of gravity.
Special calibration tests were carried out both on the platform and
the Selcom system because of the high degree of accuracy needed in the
calculation of net joint moments of force.

This study is part of a project supported by the Alexander von Humboldt

Stiftung.
596

FIGURE 1. Body position in one of the tests.

The following tests were then completed. For each of the ten body
positions, the subjects were asked to remain in that position for a
period of two seconds without touching the horizontal bar that was in
front of them and then to grasp the bar and exert maximal forces on it
in a downward, upward, forward and backward direction during four con-
secutive 2-second time intervals.
3. DATA PROCESSING
For each set of values sampled simultaneously, the net moment of for-
ce about a lateral axis through vertebra L5 was calculated from the
force measurements and from a free-body-diagram of the lower part of
the body (feet, legs and pelvis). The position of the vertebra was
determined from the position of the LED-markers and the cross-section
depth at the level of L5.
The muscle activity was derived from the EMG-signals of both flexors
and extensors, using a differentiation technique based on elements
which have been described earlier (3,4). The technique has led to the
following differential equation
A(t) ~ -a A(t) + IE{t)l 1/2
where, A(t) the muscle activity (in arbitrary units);
E(t) the EMG-signal measured;
a constant, experimentally determined for different muscle
groups and equal to 6.25;
indicating time derivative;
I indicating absolute value.
A first approximation of the numerical solution of this equation is
given by the following recursive formula for a signal sampled at 500
Hz:
 597

A(tn+1) = p . A(tn) + q. I E(tn+1) - E(tn) I 1/2

where, A(tn+1) : the muscle activity at time tn+1;
A(tn) : the muscle activity at time tn;
p : constant value, equal to 0.9875;
q : constant value, equal to 0.04475;
E(tn+1) : EMG-signal at time tn+1;
E(tn) : EMG-signal at time tn.
In the last step of the calculations, the regression coefficients a,
band c of the following linear equation are determined:
Me(t) = a * Efl(t) + b * Eext(t) + c
where, Efl(t) the activity of both flexors at time t;
Eext(t) the activity of both extensors at time t;
Me(t) the net moment at L5, estimated from the EMG-activity
at time a.
To this end at least squares algorithm has been applied on each se-
ries of 5000 values of the net moment and the corresponding muscle ac-
tivity. In this way, muscle activity, mechanical load and load esti-
mated from the activity may be compared.
4. RESULTS
Coordinates of the markers on the skin indicate that during a 10-se-
cond test, displacements of hip and shoulder are limited to a maximum
of 6 cm. The task was therefore considered to be quasi-static. This
is also illustrated by figure 2, in which ten stick figures of the body
(one per second) are drawn. It can also be seen that the body is more
stable in the upright rather than in the inclined positions.

FIGURE 2. Plot of ten stick diagrams of the body position at one second
time intervals.
Comparing the different tasks for the same body position, the maximal
effort was found not always to be limited by the same joint. This con-
clusion was derived from the values of the net moments around L5,
shoulder and knee (figure 3). Forces of -210 Nand 175 N were register-
ed when subjects were asked to push downward or upward. The corres-
ponding net moments in the shoulder were -120 Nm and 95 Nm and around
L5 were -110 Nm (flexion) and 210 Nm (extension) respectively. During
the action of forward pushing, the magnitude of the force exerted in a
horizontal direction was 350 N. The extension action at the shoulders
598

was at that time even slightly higher than in the previous task. The
extension activity at L5, however, was smaller. Therefore, it is pos-
sible that the shoulder is the limiting factor during forward pushing
in the given position and that a relatively high extension load of the
back is attained during the lifting task. Since the highest values for
flexion at the shoulders and L5 occur during the same exercise, it is
difficult to determine the limiting joint. When the hands are pulled
backwards, a maximal value of 180 Nm was found for the effort of fle-
xion at the knee joint in a body position with an inclined trunk and
with the arms almost horizontal.

200 N -
N
E
T
EXTENSIOi~

ojvj o
1'1

E
\
, /
I
FLEXION
N
T
'~I
V
KNEE
L5
-200 tJrv SHOULDER

FIGURE 3. Mean values of the net moment at knee, L5 and shoulder for
five different tasks.
When different body positions are compared for the same task, mecha-
nical loads on the trunk are shown to decrease with a decreasing in-
clination of the trunk. The height of the support at the hands is of
minor importance.
This summary illustrates the range of static loads, both for the fle-
xors and the extensors of the trunk, provided by the experimental set-
up used in this study. Flexion and extension appeared alternately du-
ring the same test situation. This is of importance in the discussion
of the EMG-activity. An example of the registered EMG-signals of M.
erector spinae and M. rectus abdominis is given in figure 4. The net
moment at L5 is indicated as well. The alternation of flexion and ex-
tension activity is clearly reflected in the net moment. In this dia-
gram the mechanical load varies between 200 Nm (extension) and -100 Nm
(flexion).
Quantification of the EMG-signals and calculation of the regression
coefficients lead to a linear equation with which the net moment at L5
is predicted. The result is also plotted in figure 4 (curve c, conti-
nuous line). The correlation between the net moment and its value as
 599

calculated from the regression equation, is high. Limits of this coef-

ficient for all tests are 0.88 and 0.98 with a mean of 0.951. The mean
value of the coefficient of non-determination is 8.2 %, and corresponds
to a mean deviation between both results of 18.5 Nm.

(
t - -
EI'lG

EflG

200 Nl'l I I I
EXTEfl5ION I

"\. I
"- L
NET l,iOHENT ./
" I
FLEXION
I
i

-200 NM

200 i~M

flET nOl,lENT
(PREDICTED)

-200 tJM +-----+-----+-----+-----+-----1

o 2 4 6 8 10

TIr1E (5)

FIGURE 4. Example of test results, showing muscle activity and mechani-

cal load during five different isometric conditions. EMG-signal of M.
erector spinae and M. rectus abdominis, net moment at L5 and its value,
predicted from the EMG signals.
600

The method applied here, has two advantages. Firstly, it is possible

to take the simultaneous activity of several muscles or muscle groups
(flexors and extensors) into consideration and to assess their influen-
ce on the net moment about the joint. In this way, it is also possible
to analyse exercises in terms of the synergic action of agonists and
antagonists.
Secondly - as is shown by the example - it is not necessary to have
at one's disposal EMG-samples at a constant level. The method may also
be used in static situations, where relationships are studied between
increasing or decreasing muscle-activity and mechanical loads.
It is expected that the application of more EMG-signals might contri-
bute to decrease the discrepancies between the net moment and its pre-
dicted value, also in assymmetrical body positions or loads.
5. CONCLUSIONS
The results of the quantified muscle activity of M. rectus abdominis
and M. erector spinae show a clear linear relation with the mechanical
load of the trunk at the level of vertebra L5 during static exercises.
This relationship also applies to increasing or decreasing muscle acti-
vity and to simultaneous action of flexors and extensors. More than 90
% of the net moment may be derived from muscle activity. Further ana-
lysis will be needed to examine to what extent the remaining 8.2 % is
caused by muscle activity which has not taken into account, by the me-
thod to quantify the EMG-activity or by errors in measurement.
6. AKNOWLEDGMENTS
The authors wish to express their gratitude for the profound support
of the Von Humboldt Stiftung, which was essential for the co-operative
success of this study.

REFERENCES
1. Andersson GBJ, Ortengren R, Nachemson AL &Schultz AB: Biomechanical
Analysis of Load on the Lumbar Spine in Sitting and Standing Pos-
tures, in Matsui H, Kobayashi K(ed): Biomechanics VIII, Illinois:
Human Kinetics Publishers, 1983, 543-552.
2. Schultz A, Andersson GBJ, Ortengren R, Bjork R, Nordin M: Analysis
and Quantitative Myoelectric Measurements of Load on the Lumbar
Spine when Holding Weight in Standing Postures, Spine 7(4): 390-397,
1982.
3. Van Leemputte MF, Spaepen AJ &Willems EJ: Quantification of EMG
Using a Differentiation Technique, in Ducheyne P, Van der Perre G &
Aubert AE(ed): Biomaterials and Biomechanics 1983, Amsterdam:
Elsevier, 1984, 103-107.
4. Zeng V-F, Hemani H &Stokes BT: Muscle Dynamics Size Principle and
Stability, IEEE Trans. Biomed. Eng. 31(7): 489-497, 1984.
 601

MINIMIZATION STRATEGIES FOR ARM MOVEMENT

M. Epstein*. J.A. Hoffer** and E.C. Mikulcik*
Dept. Mech. Eng.* and Depts. Clin. Neurosci. andPhysiol.**.
The University of Calgary. Calgary. Alberta. Canada

1. PREAMBLE
The question of which laws, if any, govern the movement of limbs
in humans and animals, has occupied the attention of researchers in recent
years, both in the field of.Neuroscience and of Biomechanics [1, 2, 3, 4].
That trajectories, control strategies, dynamic parameters, etc., are task
dependent is almost a truism - as can easily be observed experimentally -,
but the question still remains of whether a unique, simple, criterion can
be found which underlies at least some class of movements with specified
tasks. Because of experience in other fields of science and also because
of thier philosophical and aesthetical appeal, the search for a "minimum
principle" appears to be most promising. Indeed, in a recent paper [5],
based on their earlier work, Flash and Hogan have suggested the minimization
of the mean-square jerk of the hand trajectory both for single- and multi-
Jo~nt systems. We shall not discuss whether their reasons for choosing
the jerk (third derivative of the displacemen~are justifiable, namely,
"smoothness maximization" so as to be able to enforce boundary conditions
of zero velocity and acceleration of both ends of the path. Let us only
point out that discontinuous modellin~ of continuous phenomena with large
slope(large variation in a small interval of time or space) is an acceptable
technique in Continuum Physics (eg., wave-front propagation), and also that
models of the "boundary layer" type should not be discarded. It is con-
ceivable, therefore, that insistence on satisfying too many smoothness
conditions may have a detrimental effect on the model. But rather than
studying alternatives to the minimum jerk criterion, we shall exploit it
here to the fullest - in a modified fashion - and explore the final con-
sequences of its application. For simplicity, we confine the analysis to
unconstrained point-to-point arm movements in which the hand is regarded as
rigidly attached to the wrist. A quantity which can be thought of as a
"distributed jerk" for the entire arm will be minimized allowing thus the
hand trajectory to deviate from a straight line. Although a comparison
with experimental results is not attempted, such deviations appear to refle-
ct an awkwardness for extreme locations of the end points of the motion. An
envisaged application of the model is in restoration of motor funciton in
persons with spinal chord injuries by replicating, using functional electri-
cal stimulation of paralyzed muscles, the normal limb trajectories in parti-
cular movements [6].

* Department of Mechanical Engineering.

t Departments of Clinical Neuroscience and Physiology.

602

2. THE ONE-DEGREE-OF-FREEDOM SYSTEM H

For a single-joint movement such
as planar or conical motion of the
forearm around the elbow joint or the
whole straight arm around the shoulder,
~-....I------)(
the configuration of the system can be
described by a single parameter. J
FIGURE 1 - Single degree of
freedom system

In Figure 1 the limb J-H, considered as a rigid body, is shown on the plane
x-yo The joint J is an ideal hinge. Time will be denoted by t.
Our minimum principle reads as follows: "Of all possible motions
9 = 9(t) between given positions and given times the limb will choose the
one that minimizes the total square jerk P of the limb".
Denoting by L, A and p the length, cross-section and density of
the limb, respectively, we may write

P= i T fL. . .,.lL.lL p A dA dt,

o 0
'" '"
where a superimposed dot denotes time differentiation, and where T is the
time interval, lL the position vector and A is a coordinate measuring length
along the limb.'" In general, p and A may be functions of A. The main dif-
ference between our criterion and the criterion presented in [5], is that
here we minimize the total (physical) square jerk of the whole limb rather
than the one associated with the trajectory of the hand alone and therefore
not just the kinematics but also a dynamical effect related to the mass
distribution along the limb is included. This seems to be more in keeping
with the basic physical laws, where only physical quantities are involved
in a manner independent of the coordinates used to describe them. In our
particular case, a number of extra terms are thus included, involving lower
order derivatives, which will have an effect on the predicted velocity pro-
files.
Referring back to Figure 1 we have
~(A) = A cos 9 ~ + A sin 9 ~,
whence
~ ~ = 1. 2 (8 2 _ 288 3 + 98 282 + 86 ),
'"
which
'"
yields

P V

1" (8 2 _ 1'e'8 3 + 982~2 + e 6 )dt,

V .f
where V is the constant

pA
A2dA.

It follows that in the given one-degree-of-freedom system, the mass distribu-

tion and the cross sectional area play no role in the minimization process.
On the other hand, the last three terms within the parentheses are new and
 603

will have some effect on the velocity profiles, as remarked earlier. Spec-
fically, if the boundary conditions be given as

6(0) 6(0) = 8(0) = 6 (T) = e(T) = 0

6 (T) 8,

then the velocity profile, although always bell shaped, will have a tenden-
cy to flatten out more and more as 8 increases. This seems as a logical
behaviour pattern, since for long trajectories, such as a complete revolu-
tion or more, the limb will have a tendency to attain a constant speed
once the initial (and final) period of transition is overcome.
The solution of the variational problem

oP = 0

leads now to a non-linear ordinary differential equation of order 6, whose

exact solution is not available. Instead, a finite difference scheme was
implemented and programmed for a computer using a direct variational ap-
proach [7]. Results are plotted in Figures 2 and 3 for various values of
8.

el

~~ ______ ~ ______ ~t

FIGURE 2 - Angle vs. time FIGURE 3 - Velocity vs. time

3. THE TWO-DEGREE-OF-FREEDOM SYSTEM
For a two-joint planar system, such as shown in Figure 4, the config-
uration can be completely specified by means of 2 parameters, such as the
angles a and S. (In the application to the arm-forearm system the con-
straint S > a has to be added). Although the analysis in this case is
considerably more complicated than for the one-degree-of-freedom system, it
follows quite directly from the same principle, namely, the minimization of
the total square jerk of the system.
Denoting by a and b the lengths S-E and E-H, respectively, and letting
A and p stand for area and density, as before, we obtain for the position
vectors ~ and ~ of E and H, respectively,

~ = a(~os a ~ + sin a ~),

604

y H

f3

o---~--------~x

FIGURE 4 - Two degree-of-freedom system

~ = ~E + b(cos S ~ + sin S i)
Noting that, by the rigidity assumption, the position vector to any inter-
mediate point can be obtained by linear interpolation, we get, after con-
siderable amount of algebra, the following expression for the total square
jerk of the system:

+ sin (S-a) (3i:tii.S

where the quantities V2 ' WO' Wl and W2 are given by

0, 1, 2

For constant density and cross-section they reduce to

1
pAb, i = 0, 1, 2
(Hl)

The functional P is now to be minimized for the unknown functions aCt) and
Set) under the boundary conditions

a(O), S(O), a(T), SeT) prescribed

a(O) = a(O) S(O) = B(O) = ~(T) = ~(T) SeT) S (T) = 0

Again, the use of finite difference techniques into the variational expres-
sion
 605

oP = 0

permits to obtain directly a discretized system of algebraic equations to

solve the minimum problem. Numerical results for a variety of conditions
are shown in Figures 5, 6, 7 and 8.

FIGURE 5 - Hand trajectories for a=b FIGURE 6 - Arm angular coordinates

vs. time

v/

~ ______ t

FIGURE 7 - Forearm angular coordinates FIGURE 8 - Velocity profiles

vs. time

One of the salient features of these results is that the trajectories

of point H are seldom perfectly straight, the deviations being more pro-
nounced the more "awkward" the motion is in terms of the initial and ter-
minal points. The speed profiles keep their bell shape characteristic but
the maximum is biased towards one end or the other depending on the loca-
tion of the extreme points of the path. This is in contradiction with ob-
servations by some authors who seem to assume the straightness of the path
ab-initio.
606

ACKNOWLEDGEMENTS
The authors wish to express their thanks to Mr. J. Franke for his tech-
nical assistance. This work has been supported in part by the National
Science and Engineering Research Council of Canada.

REFERENCES
1. Flash, T. and Hogan, N., "Evidence for an Optimization Strategy in Arm
Trajectory Formation", Soc. Neurosci. Abstr., 8, 282 (1982).
2. Hogan, N., "An Organizing Principle for a Class of Voluntary Movements",
J. Neurosci., 4, 2745-2754 (1984).
3. Soechting, J.F., Lacquaniti, F. and Terzuolo, C.A., "Coordination of Arm
Movements in Three-Dimensional Space", J. Neurosci., 17, 295-311(1986).
4. Hollerbach, J.M. and Flash, T., "Dynamic Interactions Between Limb Seg-
ments During Planar Arm Movement", BioI. Cyber., 44, 67-77 (1982).
5. Flash, T. and Hogan, N., "The Coordination of Arm Movements: an Exper-
imentally Confirmed Mathematical Model", J. Neurosci., 5, 1688-1703
(1985).
6. Hoffer, J.A. and Sinkjaer, T., "A Natural 'Force Sensor' Suitable for
Closed Loop Control of Functional Neuromuscular Stimulation", Proc. 2nd.
Vienna Int. Workshop on Functional Electrostimulation", 47-50 (1986).
7. Epstein, M. and Huttelmaier, H.P., "A Finite Element Formulation for
Multilayered and Thick Plates", Compo & Struct., 16, 645-650 (1983).
 607

CHARACTERISTICS OF THE GAIT INDUCED BY A FALL

M.C. Do and Y. Breniere
Laboratoire de Physiologie du Mouvement, Universite de Paris-
Sud, Orsay, France

1. INTRODUCTION
The performance of a voluntary movement brings into play anticipatory
postural adjustment processes (BELENKII et al., 1967; BOUISSET and
ZATTARA, 1981) which contribute to the general dynamic organization of
balance and serve to reduce postural disturbances due to the forthcoming
movement. A similar anticipatory postural adjustment has been described in
spontaneous initiation of gait (BRENIERE and DO, 1985). Initiation of gait
has been defined as the transient phase elapsed between the initial
postural phase and the stationary gait. It is composed of an anticipatory
adjustment phase and a step execution phase. The anticipatory adjustment
phase aims to free the stepping foot and to create the necessary
conditions to obtain a stationary forward progression, i.e. to create the
propulsive forces. It has been shown that in spontaneous initiation of
gait, time elapsed between onset of movement and onset of oscillatory
phase increased with respect to the speed of progression, whereas the
duration of the oscillatory phase decreased. However, the sum of the two
times was constant whatever the speed of progression (BRENIERE and DO, in
press). It was suggested that the speed of progression was the relevant
parameter to build up the motor program.
The aim of this study was to examine the two phases, anticipatory
adjustment and execution, of the gait process in the particular
experimental situation of a balance recovery process in reaction to a
forward fall.

2. METHODS
Ten adult subjects participated in the experiment. The subject, bare
foot on a force platform, feet flat and in symmetrical position, arms
fully extended along the straight-body, was maintained in equilibrium in
an initi~l forward inclined posture by a horizontal.cable coupling a belt
around the subject's abdomen to an electro-magnet mounted on a
dynamometer. When the subject was motionless, the cable was released
without warning. The instruction was to recover balance by walking, which
is the natural behavior, at least as soon as the inclination goes beyond
several degrees. The value of the dynamometer was used to calculate the
subject's initial inclination. The adjustable cable length allowed the
initial inclination to be changed between trials. For each examination,
between 20 and 50 trials were recorded and the initial inclination ranged
between 5 and 30, or more depending on the capability of the subject to
keep his feet flat.
The force platform and its characteristics were detailed in BRENIERE et
al. (1981). In summary, it is an equilateral triangular structure, 2 m on
each side, which is suspended by means of cables to nine extensiometry
gauge systems, three vertical and six horizontal (three active and three
608

passive mounted in opposition), measuring in the range 0-1000 N with a .2%

linearity error. The rigidity of the gauge system was 10 7 N m- 1 and the
cables had a Young's modulus of E = 8750 kg mm 2 The cable platform
suspension type gives six degrees of freedom to the system. The natural
frequency of the platform when loaded is approximately 50 Hz, which is
higher than physiological frequencies of articulated systems (PAYNE,
1968). The resultant forces, RX, RY, RZ at the center of gravity (C.G)
following the three axes and the coordinates (XI and YI) of the center of
foot pressure (C.p) were calculated from the force platform data.
Data processing and data acquisition were carried out by a mini-computer
DEC PDP 11/34. The sampling period was 4 ms per channel. Data were stored
on disk for subsequent analysis. Data acquisition was triggered once the
cable was released. Software permitted recording 500 ms prior to the
release of the fall and 1.5 s after. Data were displayed off-line on X-Y
plotter (TEKTRONIX 4662).

3. RESULTS
3.1. Biomechanical pattern
Examination of the recordings showed that the step execution was not
instantaneous to the disequilibrium and was preceded by large variations
in biomechanical parameters (Fig.1). Within the first step cycle, the
vertical resultant of forces, ~Rz, showed four large oscillations around
the reference line, i.e. the subject's body weight. It started with a
negative wave, meaning that the movement of C.G started with an
accelerated fall, then was broken when the variations became positive. And

~t---_ __

RF
LF
Fig.1: Typical recording of provoked
gait.
From top to bottom: Fr, dynamometer
trace; Diagram of the different

-=--+-+-+-,...-+..;.--+-- 1 50 N
phases of right (RF) and left (LF)
foot; TO, time of release; HO, T01,
HC, T02, respectively times of
Heel-off, Toe-off, Heel-contact, and

Rv --+-+"c-i:----:::-i:;,-q-- I 50 N
Toe-off of the other foot. The right
foot was in oscillatory phase between
T01 and HC; ~ RZ, vertical resultant
force, the reference line represents
the subject's weight; RY, RX lateral
Rx 1100 N and antero-posterior resultant force,

YI
--+-----i---~++_- I .1 M direction of force to the left (L),
to the right (R), to forward (F); YI
and XI, coordinates of the center of
pressure displacement

XI
...1:..-1-"-:"_ _-'---_ _ 115 M

300 MS
 609

at the time of step execution (T01), the Rz trace was negative again
till heel-contact time (HC). Times of T01 and HC were determined from the
lateral displacement (YI) of the C.P which showed level displacements
between positive and negative values. Onset and end of a levelling of YI
indicated onset and end of a single stance phase, which coincided with the
negative peaks of 1::,. RZ. A positive level indicated a right stepping foot.
Between two successive levellings of YI were a double stance phase. Trace
of the resultant of force following the lateral axis, RY showed an
opposite direction of variation to that of YI. In other terms, during the
ongoing stance phase, for exemple on the right foot, the C.G was
accelerated towards the left foot. For weak initial inclination, a
displacement of C.P towards the stepping foot, can be seen just after the
release, as in the case of initiation of gait. Following the
antero-posterior axis, the C.P showed successive plateau displacements.
Just after the release, XI showed a first plateau of weak value, then at
time of H.C it increased steeply and reached a second plateau. Distance
between two successive levels gave the step length (L). The velocity of
step execution was calculated as V = L / /::; T, where /:;T = (T02 - T01)
corresponded to the duration of half of step cycle. On Rx trace, onset of
variation started at time of HO and showed two positive humps. It became
negative after time of HC. Times of the first peak and of the relative
minimum corresponded to times of the Rz first peak and of TO. Sometimes,
just after the release and prior to heel-off (i.e. prior onset of Rx
increase), Rx trace showed a decrease with respect to the inital value.
This was possibly due to the movement of the trunk which would be, in the
initial posture, more or less tensed by the hip muscles.

3.2. Effect of the initial inclination on the biomechanical parameters

All dynamic time-events (Zl, Z2, TO) of the adjustment phase, i.e. which
precede the step execution phase, were independent of the initial
inclination (Fig.2). Time of first /::; Rz negative peak (Zl) ranged between
80 ms (S.D = 6 ms) and 100 ms (10 ms), depending on the subject. The mean

ms
600

500
..
". :
.. He
Fig.2 Relationships of 400
time-events of provoked
gait.
Zl, Z2, TO, HC, respect i- 300
..
~ . .. .. .. TO
vely times of negative (Zl)
and positive (Z2) peak of
/::;RZ, times of step execu-
200
.~.
.. '.':.' .. .. Zl
tion (TO) and heel-contact
(HC). In abscissa, initial
inclination of subject. In
100
.. - .. .. .' -..... . ... Z2
ordinate, time scale with
respect to time release.
10" IS" 20" 25"
610

value of these averages was 90 ms (6 ms). Time of llRz positive peak (Z2),
corresponding to maximum push occurred between 180 ms (8 ms) and 240 ms
(11 ms). Time of step execution (TO) ranged between 235 ms (7 ms) and
343 ms (15 ms). Examination of these mean-times with the Student-Fischer t
test applied to two subjects at a time, did not show any significant
difference for time of negative peak, but the statistical test was
significant for times of positive peak and step execution.
The amplitude of positive peak of llRz was independent of the initial
inclination, whereas the amplitude of maximum of Rx increased with the
initial inclination (Fig.3).
Whereas the adjustment phase was constant, the step execution phase was
related to the initial inclination. Times of heel-contact (H.C) and of
second swing phase (T02) decreased with respect to the initial
inclination. A linear regression test showed a highly significant
(p < .01) correlation coefficient, which ranged between .72 and .95. Step
length and progression velocity increased with respect to the initial
acceleration (Fig. 4). The step length rarely went beyond 1 m., but the
progression speed could reach 6 km/h. A linear regression test between the
step length and the initial inclination showed highly significant
correlation coefficients (for 9 subjects out of 10) which ranged between
.71 and .91. The correlation coefficient between the progression velocity
and the initial inclination ranged between .88 and .97. It can be noted
that the correlation coefficients of the progression velocity were always
higher than those of step length.

L
em)

500 N .. AX .9
.8
. ' .' .
.7
375 N

,
V
250 N em/s)

1.7

..
1.5
.' ..
125 N ~ .. AZ
1.3
1.1
.9

10' 15' 20' 25' 10 15 20 25

Fig.3 Relationships between Fig.4 Relationships between

the amplitude of positive peak the step length (1) and of the
of t:, RZ, AZ, and maximum peak progression velocity (V) with
of RX, AX, with respect to respect to the initial
the initial inclination inclination.
 611

4. DISCUSSION
Analysis of- the results permitted the representation of gait provoked
by a forward fall. During the first 100 ms after release, the fall of C.G
of the subject was articulated around the ankles with essentially a
vertical acceleration, because the feet had not yet moved. Then the heels
lifted off the ground and the C.G was accelerated upward, i.e. the fall
was decelerated vertically, and at the same time the postero-anterior
acceleration started to increase. Then, some 300 ms after the fall, step
execution occurred. Depending on the initial acceleration, one or several
steps were necessary to recover balance. Hence, the process of provoked
step comprised an adjustment phase and an execution phase. It can
therefore be compared to voluntary movements where the postural basis has
been perturbed, and in particular to the spontaneous initiation of gait.
The characteristics of the two phases which composed the provoked gait
process were that: i) all dynamical time-events of the adjustment phase
were chronologically fixed and their values were not dependent on the
initial inclination, it also meant that they were independent of the
initial acceleration, ii) only the amplitude of Rx was related to the
initial acceleration, iii) the execution phase was dependent on the
initial acceleration.
Given that in spontaneous initiation of gait, the duration of the
adjustment phase increased with the progression velocity, but time to
initiate the gait (time elapsed between onset of movement and
heel-contact) was constant, the processes brought into play in the two
experimental situations would not seem to be identical. This discrepancy
could come from the initial experimental conditions. In provoked gait, at
time of release the C.G had an initial acceleration, whereas in
spontaneous initiation of gait, the propulsive forces were nil and had to
be generated. The fact that in spontaneous initiation of gait the lateral
adjustment, whose purpose is to free tbe moving limb, was always present,
whereas in provoked gait, over a certain limit of the initial inclination,
no lateral adjustment was observed might mean that this lateral adjustment
was not necessary for step execution when the propulsive forces were
sufficient.
The adjustment phase ended at time T01, which was the result of
chronologically fixed dynamical phenomena, and may be considered as a
physiological time whose mechanical expression was the time necessary to
realize the displacement of the stepping foot. On the assumption that one
of the purposes of the postural adju~tment associated with movement was to
enable the displacement of a limb that was previously supporting a part of
the body (MASSION and GAHERY, 1979), then time T01 was the minimum
duration necessary to complete this function. As a matter of fact, if the
subject did not have this minimal time, step execution would be impossible
and balance could not be recovered. In such a case, the subject would
probably adopt a different strategy which would consist in breaking the
fall with his arms. The redundancy of strategies to avoid a complete fall
is an indirect indication ehat the sensory messages can participate either
hierarchically or jointly in triggering the balance recovery program. This
was suggested by the disclosed relationships in the amplitude of the
vertical force resultant and of the postero-anterior force resultant,
which suggested that two independent processes were brought into play
during the adjustment phase. Following the vertical axis, a biomechanical
process reacted to decelerate the fall with roughly the same intensity.
Following the antero-posterior axis, it would seem that the fall was not
decelerated.
612

The kinematics characteristics of the execution phase, contrary to the

adjustment phase, were related to the initial inclination. Times of
heel-contact (HC) and of second swing phase (T02) decreased, while the
step length (L) and the progression speed (V) increased. It was shown (DO
et al., 1982) that the step length increased in the same direction as the
progression speed. The increasing relation between step length and
progression speed is similar to those obtained in stationary gait (CAVAGNA
and MARGARIA, 1966). However, the comparison showed that the step length
is relatively shorter for equivalent speed, in other terms that for
equivalent spontaneous progression velocity, the step length is shorter.
The fact that a non-significant correlation between the step length and
the initial acceleration could be obtained, whereas the progression speed
was always highly significantly correlated, shows that the step length and
the duration of the half cycle (T02 - TOl) were linked in a special way in
order to give a stable relationship to the progression velocity V. It
suggests that the progression speed would be the relevant kinematic
parameter.

REFERENCE
BELENKII, Y.Y., GURFINKEL, V. and PALTSEV, Y.I. Element of voluntary
movements. Biofisika, 12 (1), 135-141, 1967.
BOUISSET, S. and ZATTARA, M. A sequence of postural movements precedes
voluntary movement. Neuroscience Letters, 22, 263-270, 1981.
BRENIERE, Y., DO, M.C., and SANCHEZ, J. A biomechanical study of the gait
initiation process. J. Fran~ais Biophysique Medecine Nucleaire, 5,
197-206, 1981.
BRENIERE, Y. et DO, M.C. Les conditions de la marche stationnaire
sont-elles reunies a l'issue du premier pas J. Biophysique et de
Biomecanique, 9, 4, 1985.
BRENIERE, Y. and DO, M.C. When and how does steady state gait movement
induced from upright posture begin? J. Biomechecanics, (in press).
CAVAGNA, G.A and MARGARIA, R. Mechanics of walking. J. Applied Physiology,
21, 271-278, 1966.
DO, M.C., BRENIERE, Y., BRENGUIER, P. A biomechanical study of balance
recovery during the fall forward. J. Biomechanics, 15, 12, 933-939, 1982.
MASSION, J., and GAHERY, Y. Diagonal stance in quadrupeds: a postural
support for movement. In Reflex Control of Posture and Movement, Ed.
GRANIT, R. and PONPEIANO, 0., 219-226. Elsevier, Amsterdam, 1979.
PAYNE, A.H. The use of platforms for study of physical activity.
Biomechanics 1, 1st Int. Seminar, Zurich, 1967, 83-86. Karger, Basel,
1968.
 613

MUSCLE FIBRE COMPOSITION, ISOMETRIC LEG STRENGTH, AND

ENDURANCE IN MAN
J.H. Challis*, D.G. Kerwin*, R. Ramsbottom* and L.H. Boobis**
Loughborough University of Technology*, Leicestershire, and
University of Leicester Medical School**, Leicestershire, U.K.

INTRODUCTION
Attempts to analyse human movement in more sophisticated ways, has
placed increasing demands on basic knowledge of the musculoskeletal
system. For example the development of mathematical models of human
movement, in which the properties of the different types of muscle fibre
are accounted for (eg. Pierrynowski and Morrison, 1985), requires a more
precise specification. A review from the literature reveals that there
is conflicting evidence about the relationship between isometric leg
strength and muscle fibre composition. Hulten et ale (1975) reported no
significant relationship between muscle fibre type in the vastus
lateral is and a maximum voluntary isometric contraction (M.V.I.C) of the
knee extensors (no correlation coefficient was reported). Clarkson et
ale (1980) in a similar experiment, compared the M.V.I.C of the knee
extensors, of an endurance trained group and a group engaged in power
training. The correlation between M.V.I.C and percentage type I fibres
in the vastus lateralis for the first group was non-significant
(n=7,r=0.63), but was significant for the second group
(n=8,r=0.80,p<0.05). In contrast to these studies Tesch and Karlsson
(1978) showed a significant relationship between percentage type II
fibres in the vastus lateralis and M.V.I.C (n=31,r=0.55,p<0.001).
Certain basic questions about the human musculoskeletal system are
still unanswered or clouded by contradictory evidence. The purpose of
this study was to re-examine the relationships between the force achieved
by a M.V.I.C, local endurance capacity, and skeletal muscle fibre
composition.
METHOD
Before the study was undertaken, agreement was obtained from the
University Ethical Committee.
Twenty physically active volunteers participated in the study. Their
details are presented in table I.
Table I. Mean values for subjects age,height and weight
Mean Age Mean Height Mean Weight
27.4 (7.8) 1.69m (0.10) 65.55Kg. (10.72)
(Figures in brackets represent standard deviations)
A purpose built isometric chair was used for the strength and
endurance tests. The chair ensured that all subjects assumed the same
test position, and that the knee extensors were the only muscles involved
in the activity. The subjects all adopted the same relative position in
614

the chair. A restraining belt was secured across their waist to maintain
the required lower body isolation.
A strain gauged load cell was mounted on an adjustable fixing at the
rear of the test chair. All the subjects were seated with their shanks
in a vertical positon, (knee flexion angle of 90 degrees). Signals from
the strain gauge were amplified and passed via an analogue to digital
converter to a BBC Microcomputer, where the data and subject details were
stored on floppy diskette for subsequent analysis. Subject feedback was
provided via a visual display screen producing a real time graphical
representation of the force being exerted.
All tests were preceeded by a standard warm-up, and each leg was
tested seperately with the dominant leg being tested first. This
provided a period of familiarisation for the subject and maintained a
standard experimental protocol. Both tests showed a high test-retest
repeatability (p<O.OOl).
The first test administered was the M.V.I.C. The subject was asked
to exert their maximum voluntary isometric force during a period of ten
seconds. Effective lever length was measured from the restraining ankle
cuff to the estimated knee joint centre; from these data maximum torque
was determined. Maximum force values were normailsed with respect to
subject weight, and torque values were normalised with respect to subject
weight and height.
The M.V.I.C test was followed within a five minute period by the test
of isometric endurance of the knee extensors. To establish the endurance
capacity of the knee extensors the subject was asked to maintain 50
percent of their M.V.I.C, until the point of volitional fatigue. Similar
visual feedback allowed the subject to monitor the required target level.
Muscle fibre samples from the vastus lateralis were collected using a
needle biopsy technique (Bergstrom,1962). Muscle fibre populations were
identified by standard histochemical staining procedures (Brooke and
Kaiser,1970). A recent report by Blomstrand and Ekblom (1982) showed
that the variance in the identification of muscle fibre types could be
reduced by taking two biopsies from the same site. The present study
followed these recommendations. The mean number of fibres being used to
establish muscle fibre type being 683 (standard deviation 187).

RESULTS
The results of the tests are presented in Table II, together with the
correlation coefficients between the percentage type I fibres in the
vastus lateralis, endurance time, maximum force, and maximum torque
values.

DISCUSSION
In the present study the correlation between the percentage type I
fibres and maximum torque was non-significant, see figure I.
Results from the endurance part of the study showed a small
non-significant correlation between percentage type I fibres and
endurance time, see figure II. In a similar study involving 19 subjects,
Hulten et al. (1975) examined two-legged endurance at 50 percent of
M.V.I.C. and reported a significant relationship between endurance time
and percentage type I fibres in the vastus lateralis
(n=19,r=0.70,p<0.001). Komi et al. (1982), in another two-legged study
found a significant relationship between percentage type I fibres in the
vastus lateralis and endurance time at 50 percent of M.V.I.C.
(n=35,r=0.75,p<0.001).
 615

Table II. Values for percentage type I fibres, endurance time,

maximum torque, and maximum force and correlation coefficients
between percentage type I fibres, and endurance time, max. force,
and torgue values and the normalised values (adjusted).
MEAN S.D MINIMUM MAXIMUM CORRELATION
WITH % TYPE I
ACTUAL ADJUSTED
% TYPE I 56.9 1l.5 30.3 88.0
FIBRES
ENDURANCE 88.3 34.2 43.0 170.0 0.255
(secs)
MAX fORCE 479 125 293 712 0.184 0.317
(Newtons)
MAXIMUM 165.4 53.0 87.7 256.3 0.1l6 0.320
TORQUE( Nm)
Variation within the findings from very similar studies casts some
doubt on the appropriateness of the methods being employed.
Methodological problems within these studies may account for the
variation in results.
It is well documentented that the vastus lateralis is not homogenous
with respect to muscle fibre type (eg. Halkjaer-Kristensen and
Ingemann-Hansen,1981). Lexell et al. (1983) examined a whole
cross-section of the vastus lateralis and reported an increase in
percentage type I fibres with increasing muscle depth. Obviously the
accuracy of muscle fibre composition estimation, could be affected by the
biopsy technique's ability to reflect whole muscle fibre type
composition. It should be stated that the results of Lexell et al.
(1983) are contradicted by other studies (eg. Edgerton et al.,1975, and
Elder et al.,1982).

180

160
Vi
u
cu 140

~
cu 120

E

i=
cu
u
100
c:

10
t-
:::l
"C
c:
LU
80

60

,

40
30 40 50 60 70 80 90
Percentage Type Fibres

fIGURE I. Plot of endurance time (secs) against the percentage of

.type I fibres.
616

X10
28
26
24
E 22
~
QI
=> 20
t:r
'-
18
0
l-

16
E
=>
x 14
E
'" 12
l:
10

8
30 40 50 60 70 80 90
Percentage Type Fibres
FIGURE II. Plot of maximum torque (Nm) against the percentage of type
I fibres.

In the present study two muscle samples were taken from each subject
to reduce the variance in the identification of muscle fibre types
(Blomstrand and Ekblom, 1982). Thus, although, muscle fibre type
composition in the vastus lateral is was identified, in this study no
account was made for the area variation of these fibres. As the force a
muscle fibre can develop is directly proportional to its cross-sectional
area, (Ikai and Fukunaga, 1968), some account should have been taken of
the relative contribution of the fibre types to cross-sectional area of
the muscle under examination. Such an approach has been followed by
Nygaard et al. (1983), investigating the biceps brachii, and by Young
(1984) for the vastus lateralis. In the latter study type II fibres were
found to be 'stronger' than type I fibres. A comparable result was not
found by Nygaard et al. (1983), however the sample size was very small in
that study.
Torque rather than force values have not consistenly been used, nor
have they always been normalised for each individual subject. Although
these measures were taken in this study, direct comparison with other
findings has proved difficult and could explain some of the apparant
variabHty.
Finally it must be remembered that the force measured is a composite
of a number of muscles and is not necessarily representative of the
biopsied muscle force. A superior approach would be to estimate the
force actually developed by the muscle of interest, with knowledge of the
muscle geometry and refined measuring and analysis techniques this would
be possible. .
Perhaps the number of dependent variables acting within this
experimental design is still too extensive to enable a focus to be drawn
upon the relationship between muscle fibre type, M.V.I.C. and isometric
endurance capacity. An alternative approach might be the selection of a
different muscle site where its action could be more readilly isolated.
 617

REFERENCES
Bergstrom,J. (1962)
Muscle Electrolytes in Man. Scand.J.Clin.Lab. Suppl. 68.
Blomstrand,E., and Ekblom,B. (1982)
The Needle Biopsy Technique for Fibre Type Determination in Human
Skeletal Muscle - A Methodological Study. Acta.Physiol.Scand.
116:437-442
Brooke,M.H., and Kaiser,K.K. (1970)
Three "Myosin Adenosine Triphosphatase" Systems: The nature of their pH
lability and Sulphydryl Dependence J.Histochem.Cytochem. 18:670-672
Clarkson,P.M.,Kroll,W., and McBride,T.C. (1980)
Maximal Isometric Strength and Fiber Type Composition in Power and
Endurance Athletes Eur .J~Appl. Physio1.44: 35-42
Edgerton,V.R.,Smith,J.L., and Simpson,D.R. (1975)
Muscle Fibre Type Populations of Human Leg Muscles Histchem.Journ.
7:259-266
Elder,G.C.B.,Bradbury,K., and Roberts,R. (1982)
Variability of Fiber Type Distributions within Human Muscles
J.Appl.Physiol:Resp.Env.Ex.Physiol. 53:6;1473-1480
Halkjaer-Kristensen,J., and Ingemann-Hansen,T. (1981)
Variations in Single Fibre Areas and Fibre Composition in Needle Biopsies
From the Human Quadriceps Muscle Scand.J.Clin.Lab.Invest. 41:391-395
Hulten,B.,Thorstensson,A.,Sjodin,B., and Karlsson,J. (1975)
Relationship between Isometric Endurance and Fibre Types in Human Leg
Muscles Acta.Physiol.Scand. 93:135-138
Ikai,M. and Fukunaga,T. (1968)
Calculation of Muscle Strength per Unit Cross-sectional Area of Human
Muscle by means of Ultrasonic Measurement Int.Z.Angew.Physiol. 26:26-32
Komi,P.V.,Karlsson,J.,Tesch,P.,Suominen,H., and Heikkinen (1982)
Effects of Heavy Resistance and Explosive Training Methods on Mechanical,
Functional, and Metabolic Aspects of Performance In: Exercise and Sport
Biology Ed. Komi,P.V. pp90-102 Human Kinetics, Champaign, Ill.
Lexell,J.,Henriksson-Larsen,K., and Sjostrom,M. (1983)
Distribution of Different Fibre Types in Human Skeletal Muscles.2 A Study
of Cross-sections of Whole M.Vastus Lateralis Acta.Physiol.Scand.
117: 115-122
Nygaard,E.,Houston,M.,Suzuki,Y.,Jorgensen,K., and Saltin,B. (1983)
Morphology of the Brachial Biceps Muscle and Elbow Flexion in Man
Acta.Physiol.Scand. 117:287-292
Pierrynowski,M.R., and Morrison,J.B. (1985)
A Physiological Model for the Evaluation of Muscular Forces in Human
Locomotion: Theoretical Aspects Math.Biosci. 75:69-101
618

Tesch,P., and Karlsson,J. (1978)

Isometric Strength Performance and Muscle Fibre Type Distribution in Man
Acta.Physiol.Scand. 103:47-51
Young,A. (1984)
The Relative Isometric Strength of Type I and Type II Muscle Fibres in
the Human Quadriceps elin.Physiol. 4:23-32
 619

IN-VIVO VISCOELASTICITY OF THE HUMAN SOLEUS MUSCLE

M.R. Shorten* and D.G. Kerwin**
Nike Sport Research Laboratory*. Beaverton. Oregon. U.S.A.;
Dept. of PE & Sports SCience**. Loughborough University. U.K.

1. IN1RODU010N
Muscle elasticity and the storage and recovery of strain energy in stretched muscle and
tendon appear to enhance both the effectiveness of human movement (Bosco and Komi,
1979) and its efficiency (Asmussen and Bonde Petersen, 1974). While the enhanced perfor-
mance of movements involving a pre-stretch is at least partly the result of elastic energy
storage and recovery, other factors such as increases in myoelectric activity also playa role
(Bosco, Viitasalo, Komi and Luhtanen, 1982). In order to reliably determine the contribu-
tion of elasticity to muscular performance, the in-vivo viscoelastic behavior of individual
muscle groups under different loads must be established.
This paper describes the viscoelastic characteristics of the structures around the ankle joint,
determined using a low frequency vibration method that largely isolates the single joint
plantar-flexor muscles.

2. METHODS
Seated subjects supported a loaded frame on ,their knees while their forefeet rested on
a Kistler force plate (Figure la). With the subjects' knees flexed, it was assumed that the
gastrocnemius muscles were too short to make a significant contribution to load bearing
and that the load was supported primarily by the soleus muscles, in series with the Achilles
tendons (Hof and van den Berg, 1977).

a b

P I V 0 TED O~==~======:;::~----J.
SUPPORT

FIGURE 1. (a) Experimental apparatus and (b) system model for determining the viscoelastic
characteristics of the single joint ankle extensor muscles.
620

A brief downward force applied directly above the knees initiated lightly damped oscilla-
tions at frequencies between 3 and 6 Hz, depending on the mass of the system (Figure 2b).
The soleus EMG recorded during this procedure is similar to that recorded during an isometric
contraction at the same load (Figure 2a). This suggests that contractile component activity
was constant and that the oscillations were due to muscle and tendon compliance rather
than to voluntary muscle activity. By comparison, when subjects deliberately maintained
forced oscillations of the load, clear cyclic variations in the electromyogram were observed
(Figure 2c).
In order to determine their viscoelastic characteristics, the loaded ankle extensor muscles
were modelled as a damped mass-spring system (Figure 1b). The damped frequency and
logarithmic decrement of the oscillations in the vertical component of the ground reaction
force (Fz, Figure 2) were used to calculate an elastic constant k, damping ratio, s, and dam-
ping coefficient, c, from the equation of motion of a damped mass-spring model (Thomp-
son, 1981, pp 25-34). Data were adjusted to account for the damping and stiffness of the
measuring apparatus.
Nine male subjects (mean body mass 80.1 kg11.5sd) participated in the experiment. Each
subject completed 5 trials at each of 8 different loads in the range 100 to 1200 N. The distance
between the point of application of Fz and the estimated ankle joint centre was used to
convert stiffness and damping coefficients to angular equivalent units.

3. RESULTS
The results of this study are summarized in Figure 3. The damping ratio, s, of the free
oscillation in Fz had an overall mean value of 0.123 but decreased non-linearly with increas-
ing muscle moment (Figure 3a). The damping constant c is given by c=2Mwns where Wn
is the natural frequency of the damped mass-spring system in Figure 1b. The mean value
of c, averaged over all trials and all subjects was 13141sd Nms.rad-1, in angular equivalent
units.

a b c

Z 600

N
LL

0.0 1.0 0.0 1.0

Time, S
FIGURE 2. Soleus EMG and vertical component of ground reaction force during experimental trials
with (a) isometric contraction (b) free oscillation initiated by experimenter, (c) forced oscillation
maintained voluntarily by subject.
 621

0.4
a

-
0.3
-
0
II
~.
0

a-

0

f)
<>.
o 0

<>.0
0
0.,
<>0

0

o It

0
e
"..
~

~ .
f)

" 0.0
II
.!: 1.0
E
z b
~

ftJ

ftJ

--
CD
C

ftJ
0.5
c-
CD
II
>
~
0"
CD
a-
li
~
D
0.0
c 0 50 100
4(
Ankle extension moment. Nm

FIGURE 3. (a) Damping ratio and (b) angular equivalent stiffness of the system shown in Figure
1. Data are expressed as angular equivalents about the ankle joint and are for one leg. Each sym-
bol represents one of nine subjects, and each point records the mean of 5 trials at a given load.
Solid line in (b) is best fitting curve determined using the model described in the text.
622

The stiffness constant, k, of the supporting muscle-tendon system of one leg varied bet-
ween 28 and 860 Nm.rad-1, increasing curvilinearly with increasing load (Figure 3b). At low
loads, angular equivalent stiffness increased rapidly as the extensor moment about the ankle
increased. At higher muscle moments, the rate of increase of stiffness with increasing load
was lower.

4. MODEL OF MUSCLE TENDON STIFFNESS

The non-linear variation of the elastic stiffness of the whole muscle-tendon system with
joint moment can be described by the model shown in Figure 4. The elasticity of the tendon
and parallel elastic components are represented by stiffnesses kT and kp respectively. Ex-
periments on isolated muscle suggest that both active tension in the contractile component
and the elastic stiffness of the contractile proteins depend upon the number of crossbridge
attachments. There is evidence therefore, that contractile component stiffness increases ap-
proximately linearly with muscle tension (Morgan, 1977; Haugen 1982). Contractile compo-
nent stiffness was therefore modelled as a linear function of muscle load, given by ks.P
where ks is a constant and P is muscle tension, expressed in angular equivalent units. Com-
bining parallel and series stiffness as shown in Figure 4a, the overall stiffness of the model,
A is given by

A=------ (1)
kT + kp + ks.P

-
~ 3.0
...
E
z
.:.=

II)

--
II)

~ 2.0

I I)

-c
G)
as
> ~--------~-------------kT
::I 1.0
C"
G)

...as
::I
CI
C
<
50 100
Ankle extension moment. Nm
FIGURE 4. (a) Configuration of a model describing overall muscle-tendon stiffness in terms of three
components. (b) Component stiffnesses of the model best fitting experimental data in Figure 3b.
 623

Figure 4b shows the component stiffnesses that best fit the experimental data recorded
in Figure 3b. The parallel component, kp, has a relatively small influence. With increasing
muscle tension, the contractile component becomes increasingly stiff, and the overall stiff-
ness, A approaches kr the stiffness of the more compliant tendon. The model best fitting
the experimental data has the angular equivalent constants kr=1091 Nm.rad-1, kp'=14.7
Nm.rad-l and ks=13.5 rad-l for one leg.

5. DISCUSSION
The elastic compliance about the ankle joint is anatomically distributed thoughout the
muscle-tendon system and surrounding tissues. Hunter & Kearney (1982) have described
a method which effectively isolates the viscoelastic properties of the ankle joinf and con-
trols voluntary muscular activity. Their results were restricted to relatively low ankle joint
moments, however. Cavagna (1970) reported ankle extensor stiffness at a number of loads
in excess of bod~eight, determined by applying the damped mass-spring model to oscilla-
tions in Fz recorded when subjects landed on a force plate with the knees locked and the
Triceps surae held in voluntary sustained contraction. The relative contributions of different
muscle groups and the effects of muscle activity variations on these results are unknown.
The experimental method described here allows a wide range of loads, in the form of ad-
ditional mass, to be applied in a controlled manner. We have attempted to isolate the
viscoelastic properties of the single joint ankle extensor muscles and passive elastic struc-
tures acting around the ankle joint but, clearly, elasticity and damping at the knee and hip
joints will have an effect on the results. The recorded stiffness of the parallel component
(kp ,;, 14.7 Nm.rad-1) includes elements of passive stiffness at the hip and knee. Seigler
et al (1984) presented data for the passive moment-angle function of the normal ankle joint
from which an average passive ankle stiffness of 12 Nm.rad-l can be calculated. This value
is only slightly lower than the average value of kp, which suggests that the influence of hip,
and knee components on the stiffness recorded in this study is small. Damping at these
joints may have a greater effect on the results, however, contributing to the relatively high
damping ratios recorded at low loads.
The rate of increase of stiffness with increasing muscle load was lower at high loads than
it was at low loads. The model presented here mimics this effect, since the overall stiffness
of the model system tends to kT, the tendon stiffness, in the limiting case of infinite con-
tractile component stiffness. While the calculated value of kT (1091 Nm.rad.s-1) is compatible
with the known stiffness of the achilles tendon (Benedict et aI, 1968), it must include the
contributions of the proximal and distal attachments of all the extensor muscles of the ankle
which can apply tension with the knee flexed.
While reflex control mechanisms may contribute to the regulation of muscle stiffness under
dynamic movement conditions, the results of this study suggest that overall muscle-tendon
stiffness is dominated by a compliant tendon, especially at higher loads. This property has
a functional advantage, since a compliant system stores more strain energy than a stiffer
one when both are subjected to the same load. Limiting the ultimate stiffness of the muscle-
tendon system therefore increases the possibilities for performance enhancement through
elastic energy storage and recovery.

ACKNOWLEDGEMENIS
The authors wish to acknowledge the technical assistance of Tony Sadler in Loughborough
and Les Cooper, Richard Durost and Bob Mueller in Beaverton. Their contributions to this
project are much appreciated.
624

REFERENCES
1. Asmussen, E. & Bonde Petersen, F. (1974) Apparent efficiency and storage of
elastic energy in human muscles during exercise. Acta Physiol. Scand. 92: 537-545.
2. Benedict, J~., Walker, L.B. & Harris, E.H. (1968) Stress-strain characteristics and
tensile strength of unenbalmed human tendon. J. Biomechanics 1: 53-63.
3. Bosco, C. and Komi, P.V. (1979) Potentiation of the mechanical behavior of the
human skeletal muscle through prestretching. Acta Physiol. Scand. 106: 467-472.
4. Bosco, c., Viitasalo, J.T., Komi, P.V. and Luhtanen, P. (1982) Combined effect of
elastic energy storage and myoelectric potentiation during stretch-shortening cycle
exercise. Acta Physiol. Scand. 114: 557-565.
5. Cavagna, G.A. (1970) Elastic bounce of the body. J. Applied Physiology 29: 276-282.
6. Haugen, P. (1982) Short range elasticity after tetanic stimulation in single muscle
fibres of the frog. Acta Physiol. Scand. 113: 487-495.
7. Hof, A.L. and van den Berg, Jw. (1977) Linearity between the weighted sum of
the EMGs of the human Triceps surae and the total torque. J. Biomechanics 10:
529-539.
8. Hunter, I.W. and Kearney, R.E. (1982) Dynamics of human ankle stiffness: varia-
tion with mean ankle torque. J. Biomechanics 15: 747.!752.
9. Morgan, D.L. (1977) Separation of active and passive components of short-range
stiffness of muscle. Am. J. Physiol. 232: 45-49.
10. Seigler, S., Moskowitz, G.D. and Freedman, W. (1984) Passive and active com-
ponents of the internal moment developed about the ankle joint during human
ambulation. J. Biomechanics 17: 647-652.
11. Thompson, W.T. Theory of vibration with applications. 2nd edition, George Allen and
Unwin, London, 1981.
SPORTS BIOMECHANICS
 627

A NEW METHOD TO MEASURE LATERAL BOW ACCELERATIONS DURING

SHOOTING IN ARCHERY
C. Gallozzi, L.M. Leonardi, A. Pace and G. Caselli
Sport Science Institute, Department of Physiology and
Biomechanics, Rome, Italy

1. INTRODUCTION
During the release of an arrow the bow undergoes vibrations and
torsions caused by the impact of the bowstring with the tips of the limbs
of the bow. These vibrations are transmitted to the entire apparatus
including the arm of the archer who is holding it and influence the flight
direction of the arrow and the precision of the shot, as well and can cause
various physical problems in the archer (tendon troubles, muscolar pains
etc.) .
In 1955 experiments were carried out for the first time to find a
stabilization system to reduce the above described phenomena. From then
until the present day considerable progress has been made. Recently, with
the deregulation of the number of stabilizers the possibilities have
increased even more but the choice of this accessory is still linked to
empirical criteria and convictions which are entirely personal.
The principal objectives which are taken into consideration are:
1. to allow the correct functioning of the bow and the arm of the
archer at the moment the bowstring is released;
2. to reduce the oscillations of the bow-limbs;
3. to subdue the noise of the bowstring in the closing phase;
4. to reduce the movement transmitted to the handle (grip) of the bow
by any trembling of the muscles during the aiming phase;
5. to reduce partly the torsions on the handle (grip) caused by the
bowstring or by an insufficiently linear release.
The aim of this study is to formulate a system of measurement which
is capable to evaluate of the effectiveness of the stabilizers by measuring
the accelerations and so the displacements which the bow is subjected to,
especially during the mos t cri tical phase of the shot i. e. , during the
period of contact with the arrow.

2. MATERIAL AND METHOD

For this research we utilised a good quality bow which is widely used
among archers. The accelerations borne by the apparatus during the shot are
measured by the adoption of a mono-axial sensor mounted integrally to the
"riser" (the rigid part of the bow), near the arrow-rest point (fig. 1).
For the sake of simplicity only the lateral accelerations of the
system were analysed, but the accelerations which develop along the other
spatial directions could also be studied in the same way by using a
628

FIGURE 1.

tri-axial sensor.
One of the main difficulties which had to be tackled in studying a
phenomenon which lasts for a total of a few hundredths of a second only was
to succeed in combining the accelerations measured with the sequence of
mechanisms which occurs during the execution of a shot.
This is necessary if we are able to analyse in particular those
phenomena which occur during the critical period of contact between the bow
and the arrow and which can therefore condition the precision of the shot.
To meetthis need some of the accessories of the bow were transformed
into mechanisms which were capable, once they were activated during the
shot, to emitt a signal which could be recorded along with the
accelerometric signal.
The "clicker" was transformed into a sort of switch which was capable
to indicat the moment in which the archer begins the action of releasing
the bowstring and to automatically start up the entire recording system
(fig. 2).
The "Berger buttom" (or arrow "shock assorber") was modified in such
a way as to be able to indicate the period of contact of the arrow on the
riser (fig. 3).
Finally the bowstring was modified in such a way as to signal the
instant in which it parts company from the arrow (fig. 4).
These devices offered us the possibility of working on a series of
 629

FIGURE 2.

FIGURE 3.

analogical signals including the accelerometl'ic one which were digitised

simultaneously at a frequency of 2500 data/second/signal by an A/D
630

converter (Hewlette Packard) and then processed by the electronic computer

(Hewlette Packard 9845C) (fig. 5).

FIGURE 4.

FIGURE 5.
 631

3. RESULTS AND CONCLUSIONS

An example of a the recording of a shot is shown in the diagram
(graph. 1).

ACCEL.ERATION tgl
BERGER BUTTON [Volt.J
STRING [Volt.l
6

2 ............ A
/CI.ICKER pJ.
A fl
-0.: I
-2 ...........
, ~ V
\.

,\ 1\ "
-4
\ I ,\ : , ' : , '
.. 11'\' """ /' ,"\/ ,~ \l
v I " V,,,.." .. \I

TIME tel
-6
...
II)

lSi
GRAPHIC 1.

The total recording time comes to 15 hundredths of a second.

Along with the signal referring to the accelerations expressed in g,
the signals relating to the devices used in the Berger and at the nocking
point of the bowstring expressed in volts, can also be seen.
In the centre of the diagram (graph. 2), the period going from the
moment the arrow starts to the moment it leaves the bowstring is shown by
the square.
Even if we take into account the limi tations deriving from the
analysis of accelerations which is in a lateral direction only and which
are highly specific to the apparatus being used certain interesting
considerations can be made.
The first element to be revealed comes from the possibility of
calculating the times of the succession of the different stages of the
shot.
We can see for example that the length of the interval between the
release of the clicker (beginning of the recording) and the beginning of
the acceleration of the system towards the right which presumably coincides
with the arrow startig to move can be seen to be about 5 hundredths of a
second with a maximum of 8 hundredths of a second recorded in other trials.
This datum corresponds with the archer I s reaction time and posses
numerous questions of a physiological nature which certainly merit further
study and confirmation.
632

BOW'S ACCEl..RATION DURING l.INBS' CLOSING PHASE

[9]2T---------------------------------------------~

TINE [eJ

GRAPHIC 2.

The times referring to the period of contact between arrow and riser,
which lasts about 3 hundredths of a second and the activation time of the
Berger which is only 1 hundredth of a second seem to be equally
interesting.
As far as the accelerations borne by the system are concerned, we can
see how during the phase of contact between arrow and riser these occur in
a rightwards direction (in right-handed archers) and have a value of about
1 g.
These accelerations are probably caused by the thrusting action
exerted on the bow by the arrow in its combined mevements of torsion and
advance. This movement is abruptly interrupted at the moment the Berger
returns to the inert position giving way to larger-scale phenomena in the
order of about 4 g in intensity which are probably linked to the
oscillatory movement of the bow which occurs after the bow-limbs close and
which represents the sensation of vibration in the apparatus felt by the
archer.
Apart from these considerations of a mainly technical nature we
believe that the system of evaluation which we have tested has been shown
to posses good characteristics of sensitiveness and reliability. We also
feel that it can be used with positive results for the evaluation not only
of different types of stabilization and other accessories present on the
bow (Berger, bow-limbs etc.) but also of the influence which the archer's
technical abilities can have on the behaviour of the competition
equipement.
 633

REDUCTION OF LATERAL BOW DISPLACEMENT USING DIFFERENT TORQUE

FLIGHT COMPENSATORS AND'STABILIZERS IN ARCHERY
L.M. Leonardi. C. Gallozzi. A. Pace and A. Dal Monte
Sport Science Institute. Department of Physiology and
Biomechanics. Rome. Italy

1. INTRODUCTION
A bow is capable of shooting arrows, even without any stabilizers at
all.Naturally, all the vibrations and torsions caused by the impact of the
bowstring on the ends of the flexible bow-limbs are transmitted to the
entire apparatus, and also to the arm of the archer who is holding it.
These continual vibrations can in the long run be the cause of
various problems in the archer (aches and pain, tendinitis, etc .. ).The
first stabilization, with two short bars positioned in the front part of
the bow, was introduced in 1955 by Mr. Hoyt (who nowadays manufactures the
bow of the same name).
From 1.955 up to the present day, considerable progress has been made
with stabilization of the bow.Recently, with the liberalization by the
FITA (the organization which governs the discipline of archery at the
international level) of the number of stabilizers allowed, the
possibilities have increased even further.
Modern compensators and T.F .C.s (Torque Flight Compensators) have
inside various types of rubber pad which, because of the adjustable
pressure given by the screw base, allow a bar which is screwed in at the
front and which has a weight at the end, to oscillate more or less, as
required.
The main purposes of stabilization are:
1) To allow the normal functioning of the bow and of the arm of the archer
at the moment of release.
2) To reduce the oscillations of the bow-limbs.
3) To lessen the noise of the bowstring in the closing phase.
4) To reduce the movement transmitted to the handle of the bow by any
slight trembling of the muscles in the full draw position.
5) To reduce the oscillations acting on the bowsight during the aiming
phase.
6) To partly reduce the torsions on the handle caused by the bowstring or
by an insufficiently smooth release.
Nowadays, many archers either use various T.F.C.s systems, copying
them from the devices used by higher level athletes who obtain the best
resul ts, or, more simply, they take advantage of their own personal
experience.
The aim of this study is to evaluate the effectiveness of different
634

types of T.F. C. s for the purpose of optimizing a choice which is still

dictated by empirical criteria or by entirely personal convictions.

2. METHODS AND PROCEDURE

In order to be able to study the differences in behaviour in the Bow
- T.F.C.s - Arrow system at the moment when the arrow is released, a
mono-axial accelerometer (measuring only the lateral component of the
accelerometric signal) was fitted on the rigid part of the bow itself

BOW

ACCELEROMETER

FIGURE 1.

(fig.1) , and the signal which was sent out by this device during the
bowstring release phase and during the entire period in which the arrow,
dri ven by the bowstring, was still in contact with it, was recorded
(fig.2) .
For determining the time of release of the bowstring by the archer
and the time of the parting of the arrow from the bowstring, both the
clicker and a small portion of the bowstring itself functioned as a
switch.
Because the phenomenon we were studying is very shortlived, the
sampling took place at a high frequency (2500 Hz) using a HP 6940B
MULTIPROGRAMMER.
The following three situations were compared:
1) In the first, the behaviour of the bow (model Hoyt TD4 66" 52) with
no stabilizing support at all was analyzed.
2) In the second, the same bow was fitted with stabilization consisting
 635

of a central bar (24") with standard masses (Hoyt) , two lateral bars
and an upper-front bar (10") with the same masses and T.F.C.s (model
Spigarelli) equipped with very soft elastic elements, and adjusted
fairly loosely.
3) In the third, the same system described in point 2) was modified, with
the adoption of rather hard elastic' elements, and the adjustment
tightened up.

ACCELERATION [gl
BERGER SUTTON [vol~
STRING [Volt:.J

III VCLICKER }Vi' n

-.: , V
-2

-4

TIME Cal
~+-~~~~-+--~~-+--~~~--+-~~~~~

= III In
... ...
iii iii iii

FIGURE 2.

Fifteen trials were carried out for each of the situations being
considered; these tests consisted of shots released by one single top
athlete (from the Italian National Archery team) capable of performing
numerous trials with pratically identical characteristics.
Both the gathering of the data and the biomechanical and statistical
processing (calculation of the mean values and standard deviation) were
entrusted to a HP 9845C computer.All the software used was developed by
the authors of this study.

3. RESULTS AND DISCUSSION

In fig.3 is reported the behaviour of the mean value and the
standard deviation (S.D.) of the lateral acceleration (rightwards for the
righthanded archers) undergone by the bow in the three experimental
situation considered.
636

The time interval analysed stretches to 0.03 seconds, and goes from
the instant the bow-limbs begin their elastic return movement until the
instant in which the arrow leaves the bowstring.

LATERAL BOW ACCELERATION (MEAN VALUE & s. Do)

1) WJTIIDUT STAJlJLJZEIIS
2) WJ1M ID'T STA8JLJZEIIS
3) WJ1M IWIII STAJlJLJZEIIS
~ ~I~----------------------~

1.5

1.1

LS

I iI I I
"' "' "' "'
FIGURE 3.

Within this period, we can identify three successive phases in time:

1) An initial phase,lasting from 0" to 0.022",in which the arrow, bending,
presses on the berger (arrow shock-absorber) and causes it to recede,
with a force which increases over time until it reaches a maximum when
the berger itself has reached its limit.
2) A second phase, lasting approximately from 0.022" to 0.026" in which
the berger and the arrow perform their elastic return movement to
the starting position and the arrow is pushed to the left by a force
which decreases over time and which is equal to zero when the arrow is
no longer touching the berger. The failure to reach the zero line in
fig.3 is caused by the torsions of the bow which, in the final part of
this second phase represent the greatest percentage of the lateral
acceleration from the central portion of the bow itself.
3) A third phase,lasting approximately from 0.026" to 0.030",in which the
arrow is no longer in contact with the riser, but still with the
bowstring.During this phase the lateral acceleration 'of the bow is
caused principally by its internal oscillations which are started on
by the maximum peak of force exerted on the berger by the arrow in the
 637

instant t = 0.022", approximately.

The diagrams of the rightwards displacement of the bow, starting
from the data of fig.3 and continuing for the first 0.026", that is, from
the period in which the arrow is still in contact with the berger, are
reproduced in fig.4.

LATERAL BOW DISPLACEMENT <MEAN VALUE & s. 0. )

1) IIJMIJT STASJUZERI
2) IIIT)! SOFT STASJLJ1EIIS
3) IIJ1M lIARD STASJUZERI
~ ~.~------------------------~

1.5

..
I .
iii
..
I .
I

FIGURE 4.

The greater extent of the displacement of the bow when no type of

stabilization is used is due to the smaller mass of the bow itself. The
displacement is smaller, for exactly the opposite reason, with a
stabilized bow with T.F.C.s (which significantly increase its mass) firmly
connected to it.
But the repeatibility of the characteristics of the shot can be considered
to be of greater importance than the lateral displacement of the bow, and
for the analysis of this it is necessary to consider the S.D. of the mean
of the displacements. These values are reported in fig.4 every 1/2500th of
a second.
Bearing in mind the actual meaning of the term S. D. , we have a
certain probability (34%) that the error caused by the displacement during
a single trial, will be included within the range of the mean value plus
or minus the S.D. itself.
By carrying out this calculation for each of the three experimental
protocols, in relation to the last temporal step analyzed, the results
reported in tab.l were obtained.
638

DISPLACEMENT (mm.) AT THE TIME OF 0.026 sec.

VALUE S.D.

1) 0.94 0.35
2) 0.84 0.20
3) 0.70 0.25

wi th p < 0.05 in all the cases

1) without stabilizers
2) with soft stabilizers
3) with hard stabilizers

TABLE 1.

4. CONCLUSION
It is seen that the values of the displacement reported in point 2)
of tab.1 are higher than the corresponding values reported in point 3) of
the same table. This is due to the greater delay caused by the inertial
contribution of the stabilizing masses, which consist of softer "shock
absorbers" and a slacker grip.
We can consider, therefore, that with regard to the effects of the
lateral displacement, for short time intervals, it is as though the bow
had a slightly lower mass than it actually has, with, as a consequence,
greater displacements.
If, on the other hand, we examine the value of the S.D., to be considered
in this case an indication of repeatibility and of more consistent
behaviour in the apparatus, it is with set-up 2) that we obtain the best
results. Since this is the most important of the properties required of a
system of stabilizers and T F C. s in the sport of archery, we can
reasonably conclude that T.F.C.s of the soft kind are the most effective
ones to use.
 639

AN IMPROVED ARCHERY SIMULATOR FOR OBJECTIVE DYNAMIC TESTS OF

BOWS AND ARROWS
R. Pekalski
Institute of Sport. Dept. of Biocybernetics. Warsaw. Poland

INTRODUCTION
Presented study deals with the archery shooting technique. The
mechanical model of a competitor-bow-arrow system was developed. It can be
used to determine the connections between the parameters of this system and
the accuracy of hitting the target. It is shown that these connections may
be useful in the evaluation and improvement of the equipment as well as the
shooting technique.
The main stress in the presented study was put on formulation of some
directions for:
- elaboration of the methods of control for shooting a bow,
- more conscious and complete utilization of the features of available
archery equipment.
It was assumed that the sport result depends mainly upon:
- parameters and quality of the equipment being used,
- selection of the equipment parameters for each competitor,
- physical and psychological features and the training level of the
competitor.
Thus it follows that the object of biomechanical studies in archery should
be the competitor-equipment system. Said system can be presented as a
structural scheme (Fig. 1).

FIGURE 1. Structure of a competitor-bow-arrow system.

640

This scheme includes connections between individual elements of the system

during the aiming phase. The sequence of the contact loss for said
connections during a shot is as follows (the numbers are the same as these
in Fig. 1.):
1) competitor's right arm - bow strings,
2) arrow - handle,
3) arrow - bow strings,
4) bow - competitor's left arm.
The main effect of the competitor-equipment system action is arrow's
movement. This movement may be divided into two phases:
- first phase - interaction between an arrow and the subsystem
competitor-bow, lasts from the moment of loosing bow strings by
competitor's fingers up to the moment of contact loss between an
arrow and a bow and bow strings,
- second phase - ballistic flight, lasts from the end of first phase
until the moment when an arrow hits a target.
This division is characterised by the fact, that the competitor-bow
subsystem may interact with an arrow only during the first phase.
Competitor's action can be treated as the initial values generation for the
action of the bow-arrow subsystem during the first phase of the arrow
movement. Analysis of the physical phenomena taking place during shooting a
bow shows, that the initial parameters describing movement of an arrow are
as follows:
1. parameters determining the interaction between the competitor-bow
subsytem and an arrow (notation as in Fig. 2):
a) length of an initial drawing 10 dependent on the bow strings length
b) length of drawing 1 = xp
c) protrusion of the arrow's rest yp
d) angles of twist of a bow around the axis Y and Z (denoted a y and a z '
respectively).
The twist of a bow around Z axis equal a z is equivalent, in the system
related with the bow handle, to the displacement b = a z 1 of arrow's
nock and bow strings in the Y direction. Zero values for angles a z and
a y are defined for such a position of bow strings against a bow, that
the bow strength components in Z and Y directions are zero.
2. Position and velocity of two points of the long axis of an arrow, in the
coordinate system connected with the ground.
3. Shape of the long axis of an arrow (initial deflection of an arrow).
The identification of the action of the competitor-bow-arrow system
means identifying the influence of its individual components on the whole
system. Especially important are the relations connecting equipment
parameters and competitor's shooting technique with the accuracy of shots.
One can investigate these relations by observing the results achieved by a
good competitor while changing the parameters of used equipment or the
shooting technique. Unfortunately,the investigations mentioned above are
very often impossible to run. The main difficulties are:
- the changes of shooting technique mean a substantially long preparing
time for a competitor, which very often means a long duration of an
experiment and makes it impossible to find competitors willing to
participate in it,
- a big number of shots needed for the statistical puproses (to enable the
 641

recognition of the influence of the equipment changes among the usual

distribution of shots).
Therefore the models were built and the simulation experiment was performed
for the purposes of this study.
PHYSICAL MODEL OF A MAN-ARROW-BOW SYSTEM
The physical moqel of a man-equipment system was built which
constitutes of a device for the mechanical loosing of an arrow from the bow
strings. This device meets the following requirements:
- it allows shooting with every kind of bows and arrows used in sport
practise, with the initial values the same as for shooting a bow by a
man,
- it allows repeated shooting with the same initial values (set by a user).
The device constitutes of three systems:
1. frame of a high rigidity, settled in concrete, being the connecting
element for two other units: bow grip and release mechanism;
2. bow grip - fastens the middle part of a bow (bow handle) to the frame by
means of two gripping jaws. These jaws are fixed within the system of
three ball bearings. Said bearings enable turns (practically without
friction) of the bow handle around two axises parallel, respectively, to
Z and Y axis of a coordinate system defined in Fig. 2. The bearings may
be locked in arbitrary set angle positions ~z and ~y.
3. release mechanism enables:
a) fixing of the bow strings for every drawing length,
b) smooth, practically friction-free, loosing of the bow strings,
c) fluent displacement of the bow strings fixing point along the axis X,
Y, Z.
There were two experiments performed to verify constructed model of the
competitor-bow-arrow system.
During the first one the fast camera (1000 frames/s) was used to film
the first phase of an arrow flight. The filming was done for an arrow
released by the member of the Polish national team and for an arrow
released by the device for mechanical loosing described above. These films
proved that there are no qualitative differences in the movement of an
arrow released by these two methods.
The purpose of the second experiment was to evaluate the accuracy of
shots released by means of the device. The shooting distance was 30 m. It
was noted that the accuracy variance was 0.9 cm. This implies that almost
all (99.6%) shots reach the circle of 4-centimeter-radius, so-called "10".
It means that "the device secures results better than that obtained by the
best competitors in the world.
SIMULATION EXPERIMENT WITH THE PHYSICAL MODEL OF A MAN-BOW-ARROW SYSTEM
The final effect of the competitor-bow-arrow system action is hitting
a target by an arrow. The measure of this hit is the accuracy e.g. the
distance R between the hit point and some settled before point of a target.
There are many conditions influencing the accuracy. From the practical
point of view it is important to find a set of parameters {xi} affecting
the accuracy, and then to investigate quantitative relations between each
of the parameters xi and the accuracy measure R. The best method is to find
functions ~~.' which will be further called the accuracy sensitiveness to
1
642

X
lOCKING POINT ---.:tt:::::::;::::;;;~;;;;;;;;--rr':::::""""'~
OF THE ARROW .- '- . . . . . . . . . REST
( O,b, 0) (Xp,Yp,O)

10

FIGURE 2. Carthesian coordinate system XYZ associated with the bow handle
and some notations used in the paper.
Angles ay ' az ' called in the paper the angles of twist of a bow
around the axis Y, Z, respectively, are the components of the
directed angle ~z= (D,ay,az ) between the positive semi-axis OX
and the vector directed along the arrow's long axis. The origin
of this vector is the nocking-point of the arrow, and the arrow
head is the end-point of this vector. The X axis is defined by
the elasticity of elongation of the bow. Nocking-point of the
arrow has the coordinates (x,D,D) when the elasticity of the
bowls arms is directed along the X axis.
 643

the parameter xi. These functions may form an objective base for the
evaluation of the quality of a man-equipment system elements and their
mutual matching.
The shooting technique of a competitor may be, approximately, brougth
to generating the initial and boundary values for the bow-arrow subsystem.
These values are described by the set of parameters. The accuracy
sensitiveness to these parameters will differ depending on the equipment
used. Well-selected equipment is this one, for which the compensation of
competitor's faults is the best in comparison with the other equipment. The
fault of a competitor is the lack of recurrence of his movements, which
causes the scatter of the parameters values, generated by him for the
equipment.
In the presented study we used the device for mechanical loosing of an
arrow to determine the accuracy sensitiveness to the angle ~z (twist of a
bow around the vertical axis Z). The shooting distance was 30 m, which is
among one of the FITA round events. The following equipment was used:
- bow: make - Hoyt; bow strength Fn = 150 N; bow arms length 167.6 cm
(66 in)
- arrow: make - Easton; code X7, 17/14, length I = 0.67 m.
All the parameters of the equipment and the device were set in such a way,
as to obtain correct, in point of view of sports practice, arrow flight.
The only changeable parameter was the ~z angle. The accuracy sensitiveness
to angle Z function was found as follows:
N

aR -15.8 [cm/deg]
-=
21. 9 [ cm/ deg ]

The observed assymetry should be connected with the influence of ~z angle

on the transverse vibrations of an arrow in the horizontal plane
(Pekalski,1981). With the sign change of ~z the direction of the first
deflection of an arrow changes, thus changing the interaction between an
arrow and a nocking point.
CONCLUSIONS
On the base of the above results the following conclusions were
formulated:
1. making a mistake in setting ~z angle for ~z < 0 gives the difference in
the accuracy 1/3 smaller than for the same mistake for ~z > 0;
2. the condition necessary to hit "10" with all the arrows from the 30-
meter-distance (circle of 8 cm diameter) is for competitor to assure the
reccurance of angle ~z setting with the accuracy ~~z = 0.250 (for ~z <
0) or ~~z = 0.170 (for ~z > 0).
The above values are related to the equipment used in the experiments.
Nevertheless, they should be very interesting for the sport practise,
especially because this equipment is representative for that used by top-
performance competitors.
The values given in conclusion 2 show that to obtain the results close
to maximal a very high precision of angle ~z control is needed. It is
difficult to achieve because with only one aiming point a competitor does,
not have a visual control over ~z.
Thus an important conclusion for sport practise. The main feature to
be trained is competitor's ability to reccurance the ~z angle setting,
which highly influences the result achieved. For this purpose it should be
possible to control the ~z angle during the training process. The simplest
way to realise this is to connect two aiming points to the bow handle.
The presented mechanical device for loosing arrows from a bow may be
used in testing of an arrow set homogeneity. In this way the equipment
quality can be determined by producers as well as by users. It also can be
useful for selecting arrows for use in competitions.
REFERENCES
Pekalski R.: Mathematical model of the initial phase of arrow flight. In:
Biomechanics VII-A. Proceedings of the VIIth International Congress of
Biomechanics, Warsaw. Ed. by A. Morecki et ale Warszawa, PWN.; Baltimore,
Univ. Park Press 1981, 267-271
 645

OPTIMIZATION OF ROWING MOTION TECHNIQUE BY USE OF INTERACTIVE

COMPUTER SIMULATION
A. Dal Monte*, A. Komor** and L. Leonardi*
Dept. Physiol.*, Institute of Sport Science, Rome, Italy, and
Dept. of Biocybernetics**, Institute of Sport, Warsaw, Poland

1. I NTRODUCT I ON
Continuous growth of sport results level makes it
necessary to implement more precise methods of scientific
analysis and training in all sport disciplines. Nowadays the
application of computer simulation of an athlete's performance
with the use of mathematical models becomes more common and
seems to be a very effective method of gaining better knowledge
of sport motion phenomena, their limitations and ways of
improvement (Dal Monte et al.,1985, Nolte,1985, Pope,1973).
Generally, while the given motion technique under
investigation is considerably complex, the "model experiment"
has many advantages in respect to common sport training
practice like possibility of research in highly selctive manner
by isolating of various factors and avoiding of direct
investigation of an original object (athlete). The model and
simulation procedure may also be an useful mean of
communication between a coach and an athlete.
The optimization of sport motion performance due to the
complexity of human body dynamics and control usually requires
the use of complex, nonlinear algorithms and computational
methods. However, it seems necessary to develop more "friendly"
and user-oriented methods and software of sport motion
optimization in order to increase their direct use by coaches
in sport training practice.
In this paper an attempt is made to simulate and optimize
of rowing motion performance with the use of personal computer
and user-oriented software.
The main goals of presented rowing simulation are:
identification of the most limiting factors of rowing
performance,
- finding the best initial position of an athlete in respect to
the boat (or ergometer) and taking into account his
anthropometric data,
- finding the best sequence of muscles groups coaction in order
to produce the greatest output - force impulse applied to the
oar grip.
2. SYSTEM MODEL
2.1. Main assumptions
The effective performance of rowing requires energizing
muscles groups according to strictly defined sequence. The
following assumptions were found to be relevant in formulation
of the rowing motion model:
- the motion of an athlete can be considered as plannar,
- the pull phase should be mainly analysed as the most
important part of the rowing cycle,
- the motion results from the action of the three main muscles
groups: upper extremities fle:<ors, trunk and lower extremities
extensors; the main static and dynamic characteristics o~ these
groups for a given athlete are known, ,
- in the first step of simulation the "rowing" on the Giessing-
646

Nilsen rowing ergometer was taken into consideration - the boat

motion was neglected in system dynamics description.
2.2. Model formulation
Recent investigations of rowing performance (Dal Monte et
al., 1985. Nolte, 1978, Pope. 1973) and the assumptions stated
above lead to the simple model presented in Fig. 1.

Fo (t)

FIGURE 1. Model of rower's dynamics. m - mass of upper body

part, 10 - momentum of inertia, s gravity center
location. h - veri tical distance between oar grip and
feet support.FI - lower extremities force. Fu - upper
extremi ties force, Mt - trunk muscl es torque, F>:, Fy -
ground reaction force components
2.3. Model equations
The rower's dynamics equations were formulated basing on
Lagrange principle:
aaT aT av
(----) - ---- + = Uj (1)
at a q 3 q 8 q

where T kinetic energy of the system

V potential energy of the system
U external forces (torques); u = [FI, Fu, Mtl
q system variables; q = [x. l
They have the following form:
2 .. ..
( 10 + m * s ) * - m * s * sin * x = Mt- Fu*L*sin - m*g*cos
- m * *
s sin ~ + m ~ *
m s cos * - * *
~2 = FI - Fu (2) *
The limitations of muscular forces for each group were
formulated as follows (Komor. 1981):
(3)
 647

approximation of i-th maximum muscular torque (force)

limit measured under dynamic conditions
qrl'lQJC - maximum linear (aAgular) velocity of i-th body
segment
2.4. Performance criterion
Taking into consideration the system equations (2) the
problem of optimization of rowing motion is two-fold i.e. the
optimum initial conditions should be chosen and optimization
with respect to the control set available to the rower u(t) E U
should be found In the present example the maximization of
normalized force impulse applied to the oar, which is the
essential factor the sport result in rowing depends on, was
defined as a performance index in the form:
tp
Ip = (J Fo(t)*dt/tp * (So/Ss) ----) max (4)
o
where: Fo(t) "oar grip" resistance force
tp pull phase time period
So oar grip displacement during the simualtion of
the pull phase
Ss standard distance to be covered by the oar grip
during the pull phase (~ 1.5 m )
The performance index combines two aspects i.e. evaluation
of normalized in time force impulse and simple penalty
function which performs checking if the final condition
concerning the oar displacement was satisfied.
3. METHODS
3.1. Experimental measurements
The following experimental measurement have been done in
order to obtain necessary data for model verification and
simulation procedure:
- anthropometric - lenghts of athlete's body segments
- dynamography - ground reaction forces, seat acceleration and
"oar grip" force characteristics were recorded and processed by
Kistler- Hewlett&Packard computerized system (Fig.2)
- dynamometric - force/torque characteristics of lower and
upper extremities and trunk for given athletes were measured.

FORCE PLATFORM

FIGURE 2. Measurement system - dynamography

3.2. Simulation and opt mizat on procedure
The problem of opt mizat on can be solved by use of :
- application of an opt mizat on algorithm - in most cases the
648

use of complex, nonlinear algorithms is necessary (Hatze, 1981,

Hubbard, 1980, Komar, 1981),
- interactive action of an experienced operator (coach) with
the simulation software (Bauer, 1980, Morawski & Wiklik, 1978).
Considering the wide use of the presented method in rowing
sport practice the second method was applied for searching the
optimum solution of rowing motion pattern. An improved
interactive simulation software .was developed on HP9845
computer and some additional software aids were implemented
which enable:
- continuous calculation of performance index and its gradients
in respect to selected main system parameters i.e. input
conditions, control law parameters, etc.,
- continuous estimation of temporary solution with regard to
predefined state and control variables transients,
- easy modification of system parameters.
The software gives the full graphics support (real-time
stick figures graphs and transients of main state and control
variables) during the simulation and enables easy simulation
process control via soft keys (Fig 3 and Fig.4)
_ _ .,. ~ ;'-'"
........... .,. ... ~. ~ ~i"~!'.~t"n
t.,. No.
... le,. tryn'"
"'ff .,.,..

...
No.
_ _ _ Par_l.,. No. 4 o .allon

les - - - - - - -

..
,'7.
"....
13.
12.
..
II.
,
h---~

,.

\
,.
5

,.
"
FIGURE 3. Simulation
1 TIHE ["eel
results - kinematics

I1IB

./
I."
, .. ./
/
/
...-----~-- ---~
\
\
g ~
// '-.. . \
7.'
,.. /
J/ "'.\\\
,..
/

\~
~ ,..
,..
,..
,..
FIGURE 4. Simulation
results - dynamics
T[H[ [ucl
 649

4. SIMULATION RESULTS
The following cases were studied during the simulation
procedure:
1) HOdel valida~ion i.e.simulation of real motion with the main
data taken from the experiment. The output results
comparison of "oar grip" output force transients from the model
and experiment are given in Fig.5.

912 FOmclll. 85? [N]

837

767-
.....
z 687
u

&
612

537 I
,I

III
I
,
u 452
It I
0 387
IL I
S1Z I
/
217

162

87
.. 12 eN] ..
- -
J.\rA ~IION EXPERIMENT Fi -5311.11 liS
.ArA F _ SIlIUIATION FjoSSS.4 HS
II.
1.6 1.7 1.8 .... 2.0 2.1 1.2. 2.S 2.4 Z.S 2.6 2.7
TIME [He]

FIGURE 5. Comparison between "oar grip" force transients from

model and experiment (pull phase)
The simulated pull time was 1.10 sec while the real value
was 1.08 sec. The comparison of output force impulses gave the
following values: IFoS = 535.5 Ns, I,.1i 530.9 Ns. It was
concluded that the assumed model was satisfactory for further
simulations.
2) EvaluaLion of ~he influence of ~he iniLial condiLions on
perforJDance index
The study of the influence of the distance between "oar
lock" (ergometer) and feet strechers, Ds (Fig.6), initial angle
of the trunk, initial seat position etc. has been done.
3) EvaluaLion of effecL of chan~in~ of Lhe muscles groups
ac~ion synchroniza~ion
As an example the effect of changing the time instant of
upper extremities activation, Toa, is shown in Fig.?
4) Op~imiza~ion of ~he maLion performance
Considering the results obtained in steps 1, 2 and 3 an
attempt of iterative optimization was performed. The
improvement of about 57. of performance index was found by
changing Ds and the time instant of lower and upper
extremities muscles activation.
5) Searchinc for Lhe opLimum soluLion wiLh Lhe exLended force
Ii IIIi ~aLi ons
In this case it was assumed that the athlete under
investigation is "selectively stronger" of about 107.. It means
that at first the effect of stronger lower extremities only was
assumed and examined, later stronger upper extremities were
assumed, than stronger trunk muscles and eventually the effect
of all together stronger actuators was studied. Finally the
improvement of about ? 57. of Ir. was found wi th necessary changes
650

of force actuators activation - upper extremities should be

"faster" (Fig. 8)

DISTANCE .3

)(1 - .cat. di,..cc1.Ion

DISTANCE .4

XI - .cat. dlrect.lon

I,
500

400

300

200

100

2 0.3 0:4 0.5 0.6 ....,Ift,

FIGURE 6. The effect of changing the distance between "oar

lock" and feet strechers, Ds
 651

I - .
:"..9 12 \6 28 2. III 32:n "s.,

'
IFiRMS .45

........
.,",._ ..
500

400

300

200

100
FIGURE 7. The effect of changing
the time instant of upper
extremities activation, Toa 0.1 02 03 04 0.5 0.6 1oa1_1

FIGURE 8. Results of "selectively

stronger" athlete simulation
652

5. CONCLUSIONS
The presented simplified model follows many major asp~cts
of the rowing motion. In particular the effects of varIOUS
initial conditions and various coaction of force actuators can
be easily studied with acceptable accuracy. The model and
simulation software package was found as an useful aid for
coaches and athletes in studing various, individual patterns of
rowing performance. Basing on the presented model and
simulation procedure a new computerized rowing erometer (CORE)
was developed in Italian National Rowing Center in Piediluco.
REFERENCES
Bauer,W : Mathematical modelling and optimization and their
influence on sport movements - possibilities and limitations.
In: W.Baumann (ed), Biomechanics and Performance in Sport.
Hofmann Verlag,1980.
Dal Monte,A, i<omor,A, Leonardi,L Bi omechani cal analysis
and mathematical modelling of human motion during F"Dwing.
Report of Institute of Sport Science, Rome, 1985.
Hatze,H : Miocybernetic optimization of sport motion model,
methods for parameter identification, practical implementation.
Report of CSIR, Pretoria, 1981.
i<omor,A :Mathematical model and optimization of weight lifting
performa.ncce. Abstracts of "Sport Bi omechani cs" Conference.
Santa Barbara, 1985.
i<omor.A : Identification of force constraint functions of human
joint~ under dynamic conditions. Biomechanics IX-A, Human
Kinetics. 1985.
Morawski~ J and Wiklik, K: Application of analog and hybrid
simulation in sports. In: Simulation of Control Systems, troch
I (ed), North Holland Publishing Company, 1978.
Nolte, V: Die effektivitat des Ruderschlages. Bartel & Wernitz,
1985.
Pope,D : On dynamics of men and boats and oars, Mechanics and
Sport, American Soc. of Mech. Eng., 1973.
 653

AN IMPROVED MODEL AND COMPUTER SIMULATION OF CYCLING MOTION

TECHNIQUE
A. Komor, W. Kuban and L. Parfianowicz
Department of Biocybernetics, Institute of Sport, Warsaw,
Poland
1. INTRODUCTION
The general objective of this study was to understand the
interaction of a bicycle and athlete as a coupled biomechanical
system. Up to now the top level cycling as a long-lasting human
motion gained mainly the attention of the research dedicated to
physiology of human performance. Only limited number of
biomechanical studies were done in order to evaluate the motion
pattern (kinematics and dynamics) of cycling technique ( Davis
& Hull, 1981, Gregor,1981, Lafortune & Cavanagh, 1981, Sargeant
et al., 1981).
However it seems to be very useful to apply advanced
optimization methods in order to assist an athlete in
improvement of his sport performance with respect to individual
physical and energetic limitations.
Considering the problem of searching for optimum solution
of the given sport performance this solution mainly results
from:
- constraints related to the kinematics of athlete's body,
muscular forces limitations, energetic limitations, etc.,
- constraints resulted from interaction of an athlete and a
sport equipment (if exists),
- control action and limits of a neuromuscular system.
It seems that the use of mathematical models and modern
computer simulation software and hardware should enable to
introduce in a wider scale the "model experiment" procedure in
cycling sport practice (bicycles and motion technique
improvement ); (Davis and Hull, 1981, Komor et al., 1986,
Redfield and Hull, 1981).
2. MAIN GOALS
Assuming direct application of the simulation results in
cycling sport practice the following goals were formulated:
- simulation of selected real motion of a given athlete and
analysis of kinematics and dynamics (external and internal
forces and torques) of a given motion pattern,
- optimization of motion from the view point of muscles energy
cost minimization for selected cycling tasks like:
** steady state cycling (constant velocity),
climbing with constant velocity on the slope with constant or
non-constant slope angle,
* starting the motion i.e. gaining the speed as fast as possible
As variables to be optimized the parameters describing the
geometry of bicycle-athlete system (the "best" bicycle
selection) and the kinematic parameters of cycling pattern
(the best motion technique problem) were taken into
consideration.
3. MATHEMATICAL MODEL FORMULATION
3.1. Main assumptions
The effective performance of cycling requires the carefully
studied athlete-bicycle interaction and proper activation of
the muscles groups. The following assumptions were found to be
654

important in formulation the model of the athlete-bicycle system:

- the motion of cyclist is plannar,
- all parameters defining the geometry of a bicycle and
an athlete's position are known,
- the main parameters determining the aerodynamic
characteristics of the system are known,
the anthropometric data and static / dynamic force
characteristics of muscles groups under investigation are known.
3.2 Model
The structure of the athlete-bicycle system as shown
in Fig.l was taken into consideration. The dynamics of the
system is presented in Fig.2.

r-----l
I
I

FIGURE I.Athlete-bicycle system FIGURE 2.System dynamics

As the system has two degrees of freedom its kinematics and
dynamics was defined in following way:
KINEMATICS
Coordinates of joints (X and Y) are calculated as follows:
Poc [0. 0] crank axis
Poh [Xs, Ys] hip axis
Pop [Poc>: + L4*cos(o',c), Pocy + L4*sin(cf~c)] pedal.,..xis (1)
Poa [Popx + L3*cos(t3), Popy + L3*sin(t3)J ankle axis
Pok [Poax + L2*cos(2), Poay + L2*sin(2)] knee axis
Angular positions of segments (thigh, shank and foot) were
evaluated by use of formulas:
THIGH 1 = n - u (2a)
where: p a;:'ctg' Yh - Va) I (Xh - Xa)
~ = 2 * arctg (sqrp -Lx'*(p-LIJ/p*'p-L2)
p = (LI + L2 + Lx)/2
SHANK 2 = arctg Yk - Ya)/(Xk - Xa (2b)
FOOT 3 = A1 * sin(ac) + A2 * cos(dc) + A3 (2c)
where: AI. A2. A3 - Fourier coefficients to be found during
.. optimization procedure
The angular velocities and accelerations were calculated
by double differentiation of the following formulas:
 655

Li * cos (i> Xs
{ (3)
Li * sin (i> Ys
DYNAMICS

The system dynamics equations (of lower extremity) were

formulated basing on Lagrange principle:
e ,5 T e T e V
(---)- + = Q (4)
e t {j q e q a q
where: T kinetic energy of the system
V potential energy of the system
Q external forces
q system variables ; q = [ 1, 2, 3]
and have the form:
.2 .2-
A11 * <iil + A12 2 + A13 if3 + B12 * !~2 + B13 3+Cl Mh -Mk
*+ A22
* .2 * ~~+C2
A21 <iii ~2 + A23 * qi3 + B21 * 1 + B23 Mk -Ma
* * cp2 + A33 * (3 + B31 .2 * ,:P2+C3
.2
A31 * 1 + A32 * 1 + B32 Ma +
Rx
Rx
* L3*si n (q:,3) *- Ry * L3*cos <q:,3)
Rn * cos(ctc) + Rs * sin(ctc)
* (5)

(6)
Ry Rs * cos (ctc) + Rn * sin (ctc)
where: Aij, Bij, Cj coefficients depending on geometry (Li,si)
masses (mi) and momenta of inertia (Ii)
of the system segments
Mh, Mk, Ma muscular torques in hip, knee and ankle
joints, respectively
Rx, Ry pedal reaction forces according to eq. (6)
The limitations of muscular torques for each group were
formulated as follows (Komor, 1981):
Mi{t) < Mimax(q.~) (7)

where: Mima:dq,q) = (ali * q2 + a2j

approximation of
q + a3i>*(1 - q /qimqx) -
i-th muscular torque limits measured
*
under dynamic conditions
qimax = maximum angular velocity of selcted i-th
lower extremity segment
3.3 Performance criterion
Optimized pedal force profiles were determined for the
following joint moment cost function:
2 2 2
I = Mh + Mk + Ma -----> min (8)
It means that for a gi ven e:{ternal load and a gi ven velocity
(or a given velocity function) the muscles should act in such a
way that the minimum of the index I should be maintain at each
position of the crank. Additionaly, in order to maintain the
cyclic motion the following formula of mean crank torque should
should be also satisfied:
1'0
=
L4 * { f Rn(ctc)*dctc)/180
Mc
o
where: Mc - resistance torque on the crank shaft resulting from
the air and ground friction drag.
656

3. METHODS
3.1 Experimental measurements
The following experimental measurements have been done in
order to obtain necessary data for model validation and
optimization procedure:
- system geometry lenghts of athlete's body segments and
bicycle,
- cinematography- film recording (50 f/sec) has been done while
athlete cycled on ergometer,
- dynamography - crank torque (Mc) and pedal forces components
(Rx, Ry) have been simultaneously recorded during filming,
- electromiography - main hip, knee and ankle flexors and
extensors EMG activity have been recorded during the ergometer
test.
3.2 Simulation and optimization procedure
Considering the direct use of the developed software in
cycling training practice the interactive user oriented software
on MACSYM 150 and PDP11/34 minicomputers was developed The
software gives the full graphic support easy input data
mofification, simulation and optimization process control via
soft keys.
The optimization alghoritm assumed that the Rn force
(normal component of the pedal resultant force) has the form:
n
Rn(a,c) =,E Ai * cos (n*c(c) + Bi * sin(n*a,c) ( 10)
.on
where: Ai, Bi Fourier equation coefficients ; till now
unknown
Calculating the values of Mh, Mk and Ma from equations (5)
the function of I can be obtained in the following general form:
I = f(Mh, Mk, Ma, Rs, Rn) ( 11 )
Due to the fact that the Rs pedal force component is
useless during the cycling motion (does not produce the torque)
it should be minimized. The shape of minimum Rs transient can be
found from the equation:
elfeRs = 0 (12)
To solve the given optimization problem the searching of
the permitted solutions space and selected gradient alghoritms
were used.
4. SIMULATION AND OPTIMIZATION RESULTS
As the output the following parameters can be calculated:
transients of angular segments (hip, shank and foot)
positions, their angular velocities and accelerations,
transients of components of force applied to the pedal i . e.
Rs(t). Rn(t), Rx(t}. Ry(t} (Fig.3),
transients of joint muscular torques Mh(t}. Mk(t} and Ma(t}
As an additional information, each simulation protocol
consists of minimum and maximum values of these parameters and
performance index value.
The several cases were studied in order to evaluate the
influence of selected geometry parameters on optimum motion
performance for the individual athletes like the influence of
crank lenght , L4, seat position (Xs. Ys) and angular position
of trunk ,13.
As an example the function I f (L4) for two different
pedalling frequencies (100 and 120 rpm) is shown in Fig.4.
 657

RX
Ry

100
-- ----- RX

a1 - - - - + - - - - + - - - - 9 - + 0 - - - - 1....2-0----i150-~~-:t:::"-~'[dIg]

150
/
200 /
I
250 /
~--""/
-'
300 - - - - actual pattern
optimum pattern

FIGURE 3 Actual and optimum pattern of Rx and Ry components

of pedal force
1
~tOOO 100 rpm
65 - - - - 120 rpm
\
60 \

55 " "-\.

50 '" ....
T 0,16 O,1? 0,175 O,t8
FIGURE 4 Performance inde;{ versus crank lenght
S. CONCLUSIONS
The presented athlete-bicycle model follows many major
aspects of the cycling motion. However the preliminary results
of constant cycling velocity were only presented the model and
software package were found as an useful aid for coaches and
athletes in training process control and in objective adjustment
of bicycle parameters. Basing on the presented model and
optimization procedure the improved computerized cycling
ergometer with on-line graphic feedback of the selected optimum
and actual pattern parameters is under development by Italian
and Polish Cycling Federations (Fig.S).
658

MONITOR
ACTUAL
~ERN

~~
~. ... OPTIMUM

DATA OF
OPTIMUM
FEEDBACK I ATHLETE ON CVCLING YA
DISPLAV ERGOMETER
TECHNIQE

MEASUREMENT
PROCES$ING
YA SYSTEM
$OFTWAR
... At/Dt

MICROCOMPUTER

FIGURE 5. Functional scheme of computerized cycling ergometer

6. REFERENCES
Davis.R and Hull.M : Measurement of pedal loading in bicycling
It. Analysis and results, J.Biomechanics, Vol.14,
No. 12. 1981.
Gregor,R : A biomechanical analysis of lower limb action during
cycling at a four different loads, Doctoral Thesis, The
PennState University, 1976.
J<omor.A : Identification of force constraint functions of human
joints under dynamic conditions, Biomechanics IX-A. Human
J<inetics.1985.
J<omor,A,'J<uban,W and Parfianowicz,L :An application of computer
motion technique optimization and athlete-equipment system
improvement on the e::ampl e of bi eycl i ng ( i n Pol i sh ),
Proceedings of VIth School of Biomechanics, Olejnica (to be
published)
Lafortune,M and Cavanagh,P :Force effectivness during cycling,
Medicine and Science in Sports, Vol.12, No.2, 1980.
Redfield,R and Hull,M: Biomechanical modelling and analysis
of bicycle riding, M.Sc. Thesis, The University of California,
Davis, 1984.
Sargeant,A, Hoinville,E and Young.A : Ma:<imum leg force and
power output dur i ng hot -term dynami c e:-: erc i se, J. Amer i can
Physiological Society.1981.
 659

PRESSURE MEASUREMENTS INSIDE SHOES AND APPLICATION IN ALPINE

SKIING
P. Schaff. W. Hauser. R. Schattner and M. Kulot
Institut fur Biomedizin und Ergonomie. TOV Bayern. Munchen.
Germany

Introduction

Pressure distributions inside shoes are of great importance for orthopedic

and biomechanical research. Safety and comfort in sports, in particular,
depend essentially on this pressure quantity, which also determines if a
shoe is well suited for a certain discipline. Therefore, the measurement of
the pressure distribution allows detailed and objective statements about
these factors. The development of a special measuring system was necessary
for inshoe measurements of the human leg.

There are different principles for pressure measurements. Some are based on
an optical principle, others use strain gauges or even inductive and piezo-
electrical systems with a crystal as a transducer for sensoring pressure.
Nearly all of them can only be used for measurements of barefeet on flat,
ground installed platforms. Although some inshoe measurements have been
performed with a piezoelectrical system, we chose the capacitive principle
with a flat measuring mat, which seems to be an optimal basis for the
development of thin and highly flexible mats in different shapes to be
placed in the shoe (Fig. 1).

FIGURE 1: Symbolic diagram of the capacitive pressure measuring system

660

Ski boots for alpine skiing often cause pain, discomfort or even increase
the risk of inboot fractures if they are poorly constructed (FIG. 4). Much
research has been done on bindings but only a small amount of inshoe
measurements have been performed up to now (FIG. 3), whereby an artificial
leg is equipped with strain gauges for pressure measurements. However,
only direct measurements on the human leg can lead to objective state-
ments which may influence further boot construction. A special charac-
teristic of ski boots is a stiff and inflexible sole, which is one reason for
less flexion stress on a thin measuring mat. The first inshoe measurements
were done in the area of alpine skiing, due to the excellent prerequisites
offered for the application of a capacitative measuring mat on a shoe with
a stiff sole, as well as a need for measurements to be performed in ski
boots (FIG. 2).

pevelopment of special ~easuring mats:

Based on various prototypes, we developed a set of four different measur-

ing mats for measuring between the boot shaft and the front of the lower
leg, as well as for determining the pressure distribution underneath the
foot. The measuring mat is connected to special, lightweight electronic
equipment for processing, storing or sending data by telemetry so that it is
possible to make measurements in a laboratory or on the slope. By using
single measuring points, problematic regions of interest, e.g. over the instep
can be analysed.

Positioning of the mats

The position of the mat is shown in figure 2. The measuring mat is fixed
with adhesive strips to the lower leg. Measuring point 8 of the 3 x 24
pointmat (mat A) and point 1 of the 9 pointmat (mat B) lies 4 cm above
the ankle joint.

The mats for measuring the sole (14 point measuring mat =
mat C, 70 point
measuring mat = mat D, measuring area 1 cm per point) are constructed in
size 9 and fit well on top of the insole material underneath the sole,
without influencing its geometry.

FIGURE 2: Positioning of 8 point measuring mat

 661

Testing the system

While testing linearity, we found a r of almost 0,99 for linear regression

of all mats. Reproducability measured at a forward lean of 30 degrees
proved to be very high for a biological measuring system. The variation of
the middle shaft region was about 5 % using mat B. Hysteresis of the whole
system is shown in FIGURE 2.
Reproducability of mat D is about 21 % for the toe-region, 11 % for the
forefoot and 8 % for the heel-region comparing one of several foot to
surfaces contacts during gait of a patient.

Eight different types of ski boots (sizes 5 and 9) worn by 14 subjects were
tested at different forward leans and temperatures, using mat B for the
right and the left foot. Forward lean was determined by a gravity ankle
measurement system, fixed onto the lower leg. Additional measurements on
three different types of boots using mat A for the shaft region and mat C,
as well as mat D for measuring underneath the foot, were performed. A
complementary determination of the force at the heel element of a ski
binding and a registration of muscular activity (EMG) helped in interpreting
the results. Some field research using telemetry and video completed our
study.

The data was stored on a floppy disk so that various analyses by means of
computer programs were possible. For quick orientation, the pressure values
are shown on a RGB-monitor in numeric and in special color codes as in
thermographic pictures.

Results

There are two different ways to make a forward flexion of the lower leg.
You can bend your knees or you can go into a forward position without
bending your knees by just laying straight forward. Although the pressure
distribution on the lower leg is not greatly influenced by the different
positions, it causes great variations on the force at the heel. Our meas-
urements were performed with bended knees and it was proven by EMG-
measurements, that muscular activity of the gastrocnemius was minimal.
Active contraction of the gastrocnemius at a forward lean of 35 degrees
could cause an increase of over 100 % on the force of the heel.

However, all positions with up to 35 degrees forward lean could be done

without gastrocnemius activity.

Measuring the pressure distribution caused by the shaft of different boots

on the lower leg for different forward leans with mat A, we always found
the highest pressure on the medial positioned test series, which later
became the position of mat B. All following results measured with mat B
correspond well with the results of mat A. Even a resolution of only 8
points in one test series is sufficient for exact statements on pressure
distribution. for the amplitude of the pressure curves to be expected within
the solution of mat B. The flexibility and reproducability of mat B is a
little better than that of mat A.
662

Different subjects in the same boot showed little difference, except over
the instep, in the pressure distribution on the lower leg. For interpretation
purposes, it is useful to divide mat B into two regions. Region 1 includes
point 1 and 2, region 2 includes points 3 up to 8. Region 1, which is over
the instep, has a variability of 10 % to 45 %, region 2, mainly above the
tibia, shows a variability of 10 % to 29 %. Higher variability over the
instep can be easily explained by the tibialis anterior tendon. Two models
in particular did not fit very well into that region, causing great variation
by pressing onto the tendon. Over all variability between different subjects
for values over 15 N/cm is about 16 %. Differences of less than 10 % could
be observed when comparing the pressure values of the right boot to the
those of the left boot.
Remarkable differences could be seen between different boot constructions.
Whereas rear entry boots have higher values near the boot top and low
values over the instep, traditional boots show highest value over the
instep and a decrease of pressure at the boot top. Maximum pressure peaks
of up to 28 N/cm over the instep can be seen in traditional boots; max. for
rear entry boots was only 14 N/cm. The reacting force at the heel (FA V) is
influenced by this. Rear entry boots have significantly higher FA V values
for the same flexion angle.

30 30

~
25
u

~
25
REAR ENTRY BOOTS
.: TRAO IT1 ONAL BOOTS
.:
! 20
,
, 20

a
15 ~''''
.. 15

~ .
10 ~
10
...... ~..;;,"':-.
5 5

0 0

"r I "I-PO Int. ...... r lng-point.

FIGURES 3 + 4: Remarkable differences in pressure distribution can be seen

between different boot constructions. Rear entry boots
show maximal pressure near the boot top and low pressure
over the instep, whereas an opposite behaviour is observed
in traditional boots.

Pressure values and distribution were only minimally influenced by the size
of the boot. For model D (rear entry) maximum pressure was located 15 cm
above the ankle joint in both sizes with values of 18 N/square cm, respec-
tively 20 N/square cm. Only one model using a spring for adjusting the
stiffness of the shaft had different values for both sizes (11 N/sq. cm, 18
N/sq. cm). The position of maximum remained the same.

The influence of temperature on pressure distribution was demonstrated at

o degrees C and 21 degrees C. For a flexion angle of 35 degrees, models
using distorsion of the plastic for adjusting the shaft's stiffness led to
over 80 % higher pressure values at the freezing point. Only spring
dampened models are less influenced by temperature.
 663

30 30

25 25

20 20

15 IS

10 10

5 5

oL---------------------~----
2 .5

o~----~'---------
2 4
________

~

urlnupotnt.. urlng-polnt.

FIGURES 5 + 6: The temperature dependence of the shaft stiffness is

strongly influenced by the boot construction. Models
using the distorsion of the plastic (FIG. 6) lead to 80 %
higher pressure .alues at freezing point (--) compared
to room temperature (- -).

Looking at the pressure distribution underneath t.he foot, the influence on

the compression of the forefoot by adjusting the buckles of a conventional
boot could easily be demonstrated. There was an increase of almost. 100 N
for the total force in comparison to the open boot. The forefoot load in
rear entry boots with one back buckle is only minimally influenced by
adjusting the buckle.

The forefoot load during flexion in the ankle joint in different boots is
influenced by whether the boot fits well or not and by the adjusted buckle.
Rear entry boots have lower values than traditional boots.

Although the area of maximal pressure load varies widly from boot to boot
and is influenced by the insole material as well as by the construction of
the shaft, respectively the artificial ankle axis. The adjusting abilities for
the so-called canting position may cause an asymetric load on the foot if
adjusted to a wrong position.

Soft insole material can lower the maximum pressure peaks underneath the
foot. Especially if skiing on an icy slope, a homogenous pressure load,
based on a well suited insole, may provide skiing with less fatigue.

Testing 3 different boot constructions at increasing flexion angles, we

always found a decrease of the total force on the sole when the boot fits
the subject well, especially over the instep. 450 N at 0 degrees flexion
angle decreases to 140 N at 40 degrees flexion angle in boot B. Compara-
tive results can be ovserved for boot A and C.

Measuring a subject with the poorly adapted boot B (high pressure load
over the instep in boot B) led to almost high values (500 N for 0 degrees
and 540 N for 40 degrees) for the total force on increasing forward lean.
664

boot B
,--,
Z
L........J

- Q)

0
VI
Q)
..c
.....
500

fits well

c 300
0
Q)
0
u poorly
- '-
0

-; 100
.....0
adapted

.....
0 40 flexion

FIGURE 7: The pressure load in the forefoot region is considerably reduced

during flexion, if the boot fits well over the instep.

By greatly adjusting the buckles, an additional reduction of 20 % of the

total force during a flexion angle over 10 % could be demonstrated.

Discussion

Pressure course by increased forward lean, considerations in safety and

technology:

The various heights and positions of the maximal values in different ski
boot models allow the biomechanical relations in a frontal fall to be ob-
served in essentially more complex relations than before.
Models in which only the upper part of the shaft acts as the main support
for the thigh in a frontal fall and which connects all leverage forces or
moments exclusively to the height of the shaft, are certainly not suffiGient
for the description of the true relations in shoes with maximal pressure
values close to the ankle joint. Thus, Menke (5) refers exclusively to the
shaft's upper region as the primary support in his studies. The applicabili-
ty of rear entry models (high pressure towards the top of the shaft) may
be justified and sufficient in many cases, but the pressure relations in
conventional ski boots (higher pressure in the lower part of the shaft)
must be incorporated into a mathematical model when describing the bio-
mechanical load in a frontal fall.

The influence of the form fit on the total force on the sole during flexion
must be observed for further boot constructions and may offer an explana-
tion for the occurance of pain underneath the foot when skiing. An in-
creasing forefoot load during flexion may cause pain and discomfort. A well
suited boot has a decrease of total force by increasing flexion.
 665

Shaft construction

In comparison to rear entry boots, the ,influence of the shaft construction

on the pressure course is clearly displayed in conventional boots. The
differences in the position of the pressure maxima (closer to the top in
rear entry models, lower in conventional models) are not only important in
judging a possible flexion load on the bone, but also decidedly influences
the skiing characteristics of a boot. Thus, a higher moment can be rendered
in a rear entry model with punctual, relatively smaller pressure loads than
in the conventionally buckled boot. The same pressure sensation, in com-
parison 1.0 a conventional boot, means higher moment conveyance in a rear
entry boot. The experiences of skiers using rear entry boots for the first
time and who at first had great difficulties steering their skis, could con-
firm these considerations. However, forces leading to the release of the
heel elements FA V is also increased through the already increased moment
conveyance in lesser angles by forward leans. This can lead to increased
faulty release of the heel element in bindings which are correctly adjusted
according to the adjustment charts being used until now. In the fliture,
adjustment of the heel element with consideration given to the influence of
boot construction may be necessary, especially in high shafted rear entry
models.

Conclusion and summary

Great variations in the pressure resulting on the shin at the same forward
lean can be found in the various ski boot models represented on the mar-
ket. The conveyed moment and thus the force FAV also vary greatly from
model to model. In addition, these parameters are highly temperature
dependent. By measuring the shins of any number of test persons with the
described capacitative measuring system under application of an 8 point
measuring mat, highly reproducible pressure values and courses in the shaft
area of ski boots can be attained. Thus, it is possible to make standard
measurements directly on people. In cases of forward leans with angles up
to 35 degrees, this measuring system replaces or surpasses other measuring
methods using artificial measuring prothesis which are equipped with sen-
sors placed in a shoe and which simulate human movements. The great
differences between the boots, which can barely be noticed in its entirely
by the customer or the salesperson, but which become evident in the
above-mentioned results, lead to a demand for summarizing objective and
testable criterion - into the current lAS guideline 150, for example - in
order to differentiate by way of a test mark between constructions which
are anatomically and functionally suitable and models which are dangerous
and ill-suited. Such a guideline should therefore be published in the near
future. The demands placed therein are connected with the conclusions
derived from this work, amongst others. General and medical requirements,
as well as requirements in safety and ski technology can be summarized as
follows.
666

Requirements for Safety Technology:

1. Gritica flex moments should not affect the tibia. Thus, pressure curves
should show no pressure peaks in the lower area of the tibia.
2. The pressure develOpment on the tibialis anterior tendon must remain as
low as possible in all forward lean angles 10 N/sq. cm), in order to
prevent irritation.
3. The protective function of the ski boot for the ankle jOint must be
secured by blocking movements of over 45 degrees in the ankle joint.

fu!..9.!!irem.~nt~J!Lki 'I:.echnology:

1. The pressure curve course should show higher values in the upper area
than in the lower leg.
2. A steady course, free of pressure peaks, is desirable.
3. The pressure increase with increasing forward lean should respond to a
function of first or second rank. Plateau formatiori and pressure jumps
are to be avoided.
4. For good to very good skiers, a regulating resistance in forward leans
or also 4amping a forward lean is desirable.
5. The increase of the shaft's stiffness when the boot cools down from 21
degrees G to 0 degrees C must be lower than 20 %.

General Requirements

1. The boot should be divided into respective groups (L-A-S) which should
be clearly noted in a description of the ski boot.
2. The forward lean of the ski boot shaft should be constructed according
to the respective group.
3. One should be able to totally remove the forward lean adjustment in
just one hand movement.
4. Boots in women's sizes should have a softer shaft resistance than those
in men's sizes.
5. The pressure development at a 35 degree forward lean should lie below
the 95 percentile of the individual pain threshold of the respective
group (women app. 46 N/sq.cm).
6. An optimal adaption possibility must be ensured in the instep region in
order to avoid pressure peaks .

.References

1. Cavanagh, P.R., Lafortune, M.: Ground reaction forces in distant running.

J. Biomechanics Vol. 13 Pergamon Press Ltd. (1980) pp. 397-406
2. Hauser, W., Asang, E., Muller, B.: Injury Risk in Alpine Skiing. ASTM STP
860, Skiing Trauma and Safety. Philadelphia 1985, pp. 338-348
3. Schattner, R.,Hauser, W., Asang, E.: Basic Mechanics of Boot Skier
Interaction. ASTM STP 860 pp 151-158, Philadelphia 1985
4. Shealy, J.E.: The In-Boot Fracture. In Skiing Trauma and Safety, 6.
Internat. Symposium ASTM publication (in' press) 1986
5. Menke, W.: Biomechanische Analyse des Skisturzes. Ferdinand Enke
Verlag, Stuttgart I, 1985
 667

AN ANALYSIS OF HUMAN MAXIMAL ISOMETRIC VOLUNTARY PLANTAR

FLEXION AS A FUNCTION OF ANKLE AND KNEE JOINT ANGLE
P.A. Huijing. A.E. Greuell. M.H. Najon and R.D. Noittiez
Dept. Functional Anatomy. Interfaculty of Physical Education.
Free University. Amsterdam. The Netherlands

1. INIROIXJCl'ION

For obvious reasons direct measurement of muscle length force

relationships are only rarely feasible for human muscles. Using models for
calculating muscle force on the basis of muscle length and morphology
validated for rat muscle, length force characteristics of human calf
muscles were predicted (Huijing, 1985, Huijing et al. 1987).
Canparison of results of these studies to those involving direct
measurement of force or m::rnent exerted voluntarily by subjects would be of
interest.
As m::rnents and forces exerted maximally are likely to be the net result of
excitation of more than one muscle, special techniques are neccesary to
extract information concerning single muscles.
Under certain circumstances, manipulating the length of biarticular
muscles by more than one joint independently yields a possibilty to
extract such information.
The aim of the present study was to perfOll!l such an analysis for the human
m. gastrocnemius, by manipulating its length through variation of ankle
and knee angle during maximal exertion of iscmetric plantar flexion
m::rnent.

2. ME'IHODS

Maximal voluntary plantar flexion was exerted by a male and a female

subj ect on an ergcmeter which allowed measurent of net plantar flexion
m::rnent at a large variety of ankle and knee angles. The ergcmeter used,
which apart fran minor modifications was identical to the one described by
Hof and van den Berg (1977), was designed for measuring exerted -m::rnents
exclusively and allows alignment of estimated knee and ankle axes with
respective axes of the ergcmeter. The stiffness of the measurement segment
of the apparatus was 2100 Nm/rad.
For each of eight ankle angles maximal plantar flexion m::rnent was measured
for twenty different knee angles. All series of measurements were
perfoll!led twice on different days. Each of the experiments perfoll!led on
one day were preceded by a reference contraction to exclude the
possibility of local muscle fatigue.
Data relating knee angles to maximal plantar flexion m::rnent at different
ankle angles were least square fitted with the following polynanial:
668

Y = aOX3 2
+ a 1x + a 2x + a 3

where Y represents maximal plantar flexion mcment, X knee angle and a ,

a 1 , a 2 and a 3 are constants. These fitted curves were used in further da~
treatment.
Using the equations of Grieve et ale (1977) change of origin - insertion
distance (1 .) of the gastrocnemius muscle with respect to a reference
position wa~lcalculated for each knee and ankle angle combination. 1 . at
the reference position was estimated using a simple planimetric @6del
decribing the geanetry of the lower leg (Bobbert et al, 1986)
Using these length data the maximal plantar flexion mcment curves were
shifted with respect to each other in order to line up equal gastrocnemius
origin insertion lengths.
I t should be noted that using this method changes of muscle length as a
consequence of tendon canpliance are disregarded.
As considerable overlap occurred with regard to 1 . between successive
curves the contribution of MG to the total mcment caR1be calculated if the
following assumptions are made:

------ Voluntary excitation of MG during exertion of maximal plantar

flexion rncrnent is independent of the origin insertion length at which this
rncrnent is exerted.
b
------ MG can not exert a moment wi th the knee bent to 60 0 , full
extension being 1800 ) and the ankle plantar flexed (to 140 , full plantar
flexion being 1500 ) and as a consequence the rncrnent measured is the net
rncrnent exerted by the synergists and antagonist of MG at the ankle.
------ The contribution of these synergists and antagonists to net plantar
flexion rncrnent is not influenced by knee angle.

If these assumptions are valid, the change of moment between

overlapping curves at equal MG 1 . are due to length changes as well as
changes of rncrnent arm of plantar ~lexion synergists and antagonists of MG.
By subtracting the sum of contributions to the total rncrnent of these
synergists and antagonists an estimate of MG exerted mcxnent is obtained.
For MG of six human cadavers a regression analysis was performed
relating tibia length to mean number of saroomeres of MG. Using this
relationship the number of sarcameres of fibres of MG was estimated for
the subjects of this study, fran which an estimate of fibre optimum length
was obtained by multiplying with sarcomere optimum length. Methods used in
these cadaver studies were described previously (Huijing, 1985).
MG length rncrnent curves were canpared with the length range estimated on
the basis of fibre optimum length.

3. RESULTS

Knee angle - rncrnent diagrams, as obtained by least square fitting

third power polynanials to the raw knee angle and net plantar flexion
rncrnent data, are represented in Fig. 1 for both subjects.
Note that the shape of the curves is highly dependent on ankle angle,
being concave upwards in the plantar flexion region and concave downwards
in the dorsal flexion region.
In order to facilitate canparison of these results with similar ones
presented in literature, Fig. 2 gives an example of some of the same data
represented as ankle angle - rncrnent diagrams.
 669

Calculated length moment curves for MG are shown in Figure 3. Fran these
curves one could get the impression that for the ranges of both knee and
ankle angles studied only a small part of the length range, at which MG
can exert force actively, was used: With increasing muscle lengths the
moment exerted increased but never reached an optimum.

180 Moment (Nm)

MG

140

100

60

20

36 40 44 48
Length MG (em)

Fig. 3. Calculated length - moment curves for gastocnemius

muscle of both subjects. Bars indicate half length range
estimated on the basis of fibre optimum length (see below).

The results of a regression analysis performed on cadaver material, for

the relation between tibia length and mean number of sarcaneres of fibres
of MG is shown in Figure 4.

:j:I: Sareomeres
+
20000

18000
+
+
16000 + +
y= 90.1828 x -15390.3
14000
r=.92

12000~~~r----r----.----.----.
30.0 35.0 40.0
Length tibia (em)

Fig. 4. Regression between tibia length and mean number of

saroameres of human gastrocnemius muscle.
670

200 Moment (Nm) 86 200 Moment (Nm)

77

160 160
A~ 95

~:
120 120

80

40
80

40
;a:
60 100 140 60 100 140
Flex Knee Ext Flex Knee Ext
Angle Angle

Fig. 1. Knee angle - Plantar flexion moment curves for the

two subjects. Numbers next to curves indicate ankle angle.

200 Moment (Nm)

146.7

160

66
120

80

40

140 100 60
Plant Ankle Angle Oors

Fig. 2. Example of Ankle angle - Plantar flexion moment

curves. Numbers next to curves indicate knee angles.
 671

Using this relationship the number of saroomeres of fibres of MG was

estimated for the subjects of this study, Fran the number of sarcaneres an
estimate of fibre optimum length was obtained. Fibre optimum lengths were
54.4 mm and 70.3 mm respectively. As muscle fibres will start delivering
active force at approximately 60% of their optimum length and, under
conditions of equal activation, will increase force production up to
optimum length, the length range of the left half of the length force
relationship will involve approximately 22 mm and 28 mm respectively for
the subjects of this study.
Comparison of these estimated ranges with the calculated length moment
relationships (Fig. 3) indicate that length moment curves calculated using
the methods of this study can not be considered to be realistic ones.

4. DISClJSSIOO AND mNCLUSIOOS

The results of the present study with respect to joint angle moment
relationships agree quite well with those of Belanger et al. (1983), Fugl
- Meyer et al. (1980,1982) and Sale et al. (1982).
However fran the ccmparison of length moment data to fibre length range it
is clear that the experimental length force relations are not a realistic
ones because its length range is approximately 2-3 times greater than
could be expected on the basis of estimated fibre length.
Reconsideration of the assumptions (see methods) necessary for the
method of calculating the moment exerted by MG is indicated. Essentially
they require the following:
Constant excitation of gastrocnemius at the one hand as well as its
synergists and antagonists with respect to plantar flexion on the other
hand.
The gastrocnemius to be at or shorter than its active slack length
(Le. the shortest muscle length at which isanetric active force
generation decreases to zero).
Changes of voluntary excitation of MG as well as its synergists and
antagonist with changes of joint angle seem to be the most likely
explanation of the results of the present study.
Preliminary scrutiny of electranyographic data on these muscles seems to
confirm this notion, and indicates that changes of excitation may occur
during a experimental session when measurements were performed at constant
ankle angle but with increasing knee angles.
It is concluded that for maximal voluntary plantar flexion it is not
possible to construct usefull length mcment or length force curves for m.
gastrocnemius, as voluntary activation of synergists and antagonists of
MG, none of which cross the knee, is influenced by knee angle.

REFERENCES

Belanger, A. Y., A.J. McCanas and G.B.C. Elder. Physiological properties of

two antagonistic human muscle groups. Eur. J. Appl. Physiol. 51: 381-
393, 1983

Bobbert, M.F., P.A. Huijing and G.J. van Ingen Schenau (1985) A r-bdel of
human triceps surae muscle-tendon ccmplex applied to jlUl1ping. J. Biomech.,
1986, in press.
Fugl - Meyer, A.R., L. Gustavson, Y. Burstedt. Isokinetic and static
plantar flexion characteristics. Eur. J. Appl. Physiol. 45: 221 - 234,
1980.
672

Fugl - Meyer, A.R., K.H. Mild, J. Hoernsten. Output of skeletal muscle

contractions. Acta Physiol. Scand. 115: 193 - 199, 1982.

Grieve, D.W., S. Pheasant and P.R. Cavanagh. Prediction of gastrocnemius

length fran knee and angle joint posture. In: E.Asmussen and K. Jorgensen
(Pds.), Bianechanics VI-A, University Park Press, Baltimore, 1977.

Hof, A.L. and J.W. van den Berg. Linearity between the weighted sum of the
EM;'s of tthe human triceps surae and the total torque. J. Bianech. 10:
529 - 539, 1977.

Huijing, P.A. Architecture of the human gastrocnemius muscle and some

functional consequences. Acta Anat. 123: 101 - 107, 1985.

Huijing, P.A., P.C. Vossen, W.H. Rijnsburger and R.D. Woittiez. Length
range of active force generation and in situ length range of human m.
soleus and its fibers during maximal ankle excursion. In: Bianechanics X,
B. Jonsson (Pd.), Human Kinetics Publishers, Champaign, 1987, (in press).

Sale, D., J. Quinlan, E. Marsh, A.J. McComas and A. Y. Belanger. Influence

of joint position on ankle plantar flexion in humans. J. Appl. Physiol.:
Respirat. Environ. Exerc. Physiol. 52: 1636 - 1642, 1982.
 673

AN ATTEMPT AT EVALUATION OF STRENGTH/SPEED ABILITIES OF LOWER

EXTREMITIES FOR ATHLETES
Z. Trzaskoma, G. Bartosiewicz, J. Eliasz, A. Dabrowska,
J. Gajewski, L. Iskra and W. Wit
Dept. of Biomechanics, Institute of Sport, Warsaw, Poland
INTRODUCTION
In numerous cases of training process, as well as in biomechanical
studies, counter-movement jump (CMJ) is used to evaluate strength/speed
abilities of athletes of different sport disciplines. This exercise was
applied by Alabin and Krivonosov [1J for track and field, by Komi and Bosco
[5J, Viitasalo et ale [8J for volleyball, Bartosiewicz et ale [3J for
fencing, Bartosiewicz and Wit [2J for volleyball and judo. Values of ground
reaction forces as functions of time as well as height of jump are usually
taken as criterions for the evaluation [4,5,7,8J.
Presented study is an attempt at multiparameter evaluation of movement
with CMJ being a testing exercise. The aim of this evaluation is to build
an objective, accurate, and prognostiC estimation of strength/speed
abilities of lower extremities of athletes.
We tried to find the answers to the following questions:
1. What are the maximal mechanical power values of lower extremities and
trunk developed by athletes of different sport disciplines?
2. Can the maximal height of center of body mass uplift serve as a
measurement of maximal power of lower extremities and trunk?
3. Do the highest values of maximal power achieved during jumps coincide
with maximal height of center of body mass uplift?
4. What are the changes of work and maximal power with respect to an
additional external load?
ASSUMPTIONS AND METHODS
On the basis of the mathematical model of a man performing CMJ [2J the
parameters characterizing this performance were chosen which (from the
point of view of mechanics) are the best chracteristics of strength/speed
abilities. The basic parameter, which can be derived directly from the
ground reaction force plot, is maximal force (or maximal acceleration) at
the moment of take-off:

1F1max = J (F z - Q)2 + F~ + F; Imax (1)

lal max = 1F1max /M (2)
where: Q - body weight, M- body mass, Fx ' Fy ' Fz - components of groung
reaction force.
It was assumed that the basic parameter determining movement dynamiCS
(and also strength/speed abilities) is maximal power (Pmax ). In order to
compute Pmax values the velocity of center of body mass was calculated:
VCM(t) = oft F(x) dx (3)
The maximal velocity values carried also some significant informations
674

about the dynamics of a movement performed:

IVCMlmax = J V~ + V~ + V; Imax (4)

where: Vx ' Vy ' Vz - components of the velocity of body mass.
On the basis of force and velocity the mechanical power was calculated:
P = F VCM (5)
which gives:
P = Fx'V x + Fy'V y + (Fz - Q).vz (6)
and then the maximal power (Pmax ):
Pmax = P(t) Imax (7)
The second parameter taken to evaluate strength/speed abilities was the
maximal height of center of body mass uplift (h max ):

hmax = [t /4 Vz(x) dX.] (8)

3 max
where: t3 - moment when Vz = max, t4 - the end of movement.
Strength/speed abilities of athletes were determined on the basis of
three tests:
A - CMJ, individual voluntary performance;
B - CMJ performed up to a given (varying) height, with the assumption of
the shortest time of movement;
C - CMJ performed with an additional external load (5-60 kg).
The above tests were performed on a Kistler force platform. Minicomputer
Neptune 184 with an analog/digital convereter was used to process the data.
The diagram of measuring set is shown in Fig. 1.

A MAX-2.17G
V MAX 2.9iM/S
P MAX-4140W
H-47CM.

MINICOMPUTER NEPTUNE 184

Figure 1. Diagram of the setup used for studies of strength/speed abilities

of lower extremities and trunk
The subjects of this study were: weightlifters (seniors, n=12), volleyball
players (men - national team, n=12; women - national team, n=15), handball
players (men, national team - seniors, n=17; and juniors, n=12), fencers
(men - juniors, n=21; and women - seniors and juniors, n=10), judo
competitors (national team men - juniors, n=26), and the b~st Polish
juniors of the following disciplines: weightlifting, wrestling, fencing,
speed skating (n=20). Results achieved by athletes practising marathon and
 675

long jump as well as untrained persons were also taken into consideration.
RESULTS
Mean values of maximal power (Pmax ) obtained by athletes of different
sport disciplines are shown in Fig. 2. Mean values of maximal height of
center of body mass uplift (h max ) are shown in Fig. 3.
h mQx 1m!
mo. ( kW J

5 T
I
I

--
r-- .. _

ass
4
-~

0.45 1
3 - -~rr i1-

-'""r
0.35

A t " 'I lL- ~ ,.k.

~ i.- ,.
~ ~.S
M.S M.S M.J M.J W.
t A l . ~ V
M.S M.S M.S M.J W.S M.J
~ if
'I
W. ~
""S S.J S.J
n12 n12 n17 n-12 ",,21 in026 ne15 nl0 n12 n-12 n-17 n'21 n15 no12 n'lO noi!E
dlSClph~S

Figure 2. Mean values of maximal Figure 3. Mean values of maximal

power (Pmax) obtained' by height of center of body
athletes of different mass uplift (h max ) for
sport disciplines athletes of different
sport disciplines
Denotation: disciplines - olympic symbols, M- men, W- women, S -seniors,
J - juniors, n - number of tested subjects.
The largest values of maximal power were achieved by weightlifters (4560
1050 W) and volleyball players men (4420 500 W). The highest value of
maximal power was obtained by a weightlifter and was equal 6342 W. The
smallest values of maximal power were recorded for women practising
volleyball (2270 420 W) and fencing (2050 380 W), and for junior judo
competitors (2370 550 W). The greatest values of hmax were achieved by
volleyball players men (0.592 0.07 m) and weightlifters (0.579 0.052
m). The lowest values of hmax were obtained by judo competitors (0.360
0.053 m). Individually, the highest value hmax = 0.77 m was recorded for a
volleyball player.
In order to answer the second of the research Questions the
correlation coefficient between Pmax and hmax was calculated (Fig. 4). It
676

was significant (on 0.05 level) for volleyball players (men and women) and
for fencers (men).
In order to answer the third research question test B was conducted.
Every subject was to execute consecutive jumps with the command "take off
as fast as possible". The height of the jumps was preset starting from 15
cm and increasing in steps of 5 cm up to the maximum for each subject. In
Fig. 5 changes of maximal power as function of height of center of body
mass uplift are shown for selected subjects.

correlqtion
coefficient
Pmax- h max

UIIDD p<O.OS
0.9
Pmax fkWJ

_. _ _ . _wetgtlththng

4.'
0.7
long JUmp
,-
3.' .",' ",

". "/ "

.../ ............ .
-
/~
0.5 2.'
. untrained IST"I

~.
-"-"-- -'marathon

0.3

t 1-11- ,.I- "'" M.

0.1 W.
W.S M.J M.S S.J M.l M.S
ft &1 .tz natO 012 ".12
~I
j. ~26
disciplines

Figure 4. Correlation Figure 5. Variations of maximal power

coefficients (Pmax ) with height of
between Pmax center of body mass uplift
and hmax (h max )
The variations of work and maximal power with an additional
external load were studied using a special (prototype) vest. Consecutive
CMJ tests were performed with an external load ranging from 5 to 60 kg
(with 4 or 8 kg steps). In Fig.6 the changes of maximal power as function
of additional external load are shown. The external load is expressed as
percents of Subject's body mass. Experimental data are shown as regression
lines. The best Polish junior competitors of four disciplines
(weightlifting, wrestling, fencing, speed skating) were subjected to this
part of experiment. Maximal power decreases monotonically with an increase
of external load. The changes of maximal power were similar for athletes of
different sport disciplines. For the same athletes (test C) the changes of
external work (expressed in relation to 1 kg of subject's body mass with *)
external load were determined (Fig. 7). It was noted that, contrary to
maximal power changes (Fig. 6), the external work does not change
significantly.
 677
Pmax 'wI ,6Pmox
C!) WRESTLING 1-35.3'1,1 L>~

1 WRESTLING 1-4.8 1/1"2em I
FENCING (-35.5 1/11 2 FENCING 1-12.1 1/ , .. 6eml

WEIGHTlIFTlNGI-32.7'/o1
3 WEIGHTLIFTING 1-12.0"1, .. 6em I
4 SPEED SKATING (- 6.11/,_2.5eml

(0 SPEED SKATING (-36.9 1/1)

2500

~~'--------~~~---------8Tb~~ L.~--------~~,-,----------~~__
bOdy weight I~I bOdy weq.t1"l.J

Figure 6. Variations of maximal power Figure i. Variations of external

(Pmax ) with additional work (*) with
external load additional external
load
DISCUSSION
At the beginning of this experiment it was assumed that the best
parameters describing strength/speed abilities of athlete's lower
extremities are maximal power (Pmax ) and maximal height of center of body
mass uplift (h max )' measured during CMJ. In tests conducted with volleyball
players (men and women) and judo competitors [2] mean power values were
also calculated. We presume that maximal power values achieved in CMJ are
close to the maximal possible ones [3]. Bartosiewicz and Wit [2] performed
a similar study. They registered values ranging from 2975 to 4320 Wfor the
best Polish volleyball players (men). In our study the highest values of
maximal power were achieved by athletes of those disciplines in which the
high level of strength/speed abilities is necessary for effective
performance of specific movements (volleyball, weight! i fting). Relatively
low maximal power values achieved by judo competitors are probably related
to typical movements of this discipline, which take place mainly in
horizontal plane. Women (volleyball, fencing) achieved significantly lower
Pmax and hmax values than the best men. It is probably caused by lower
strength/speed abilities of women. Komi and Bosco [5] explain this
phenomenon by smaller mass of muscles, different structure of muscles, and
different usage of muscle elastic energy. Values of hmax registered in this
study (volleyball,judo) correspond to those published by Bartosiewicz and
Wit [2], and are little higher than those published in the study of Komi
and Bosco [5]. Relatively high values of height of jump published by
Skladanowska [6] are caused by the fact, that in so-called Sargent test the
measured value is the height of hand uplift and not of the value of the
center of body mass uplift. It has been previously stated by Bartosiewicz
and Wit [2] that there is no correlation between hmax and Pmax
(independently of sport discipline, sex, age, physical state of an athlete
such as warm up, effort,_ fatigue). In our study there is a significant
relation between Pmax and hmax observed for volleyball players and fencers.
Very often the jumping ability is identified with the power {Bartosiewicz
678

and Wit [2J). With the assumption that hmax is the parameter describing
jumping ability, then the simple relation with this parameter is present
only for square of initial value of center of mass velocity. Even though
there is not physical nor statistical basis for the statement that CMJ may
be an objective test of power, we believe that for athletes of some sport
disciplines the power of lower extremities can be indirectly evaluated on
the basis of height of jump. This assumption is often applied in training
practise (Alabin and Krivonosov [1J). On the basis of the results of test B
we can state, that the highest values of Pmax are not always achieved in
jumps for which the highest values of hmax are observed. It implies, that
to evaluate the greatest values of Pmax for some athletes test CMJ
performed to preset height shoud be applied (e.g. 80% of maximal value of
center of body mass uplift). By means of a special vest the variations of
external work and maximal power with an external additional load were
investigated (test C). Wachowski [9J, similarly to us, did not find any
significant correlations between changes of maximal power and practised
sport discipline.We believe that changes of maximal power during CMJ
performed with an additional external load are related to individual
strength/speed abilities of an athlete and not to sport discipline. The
values of external work developed during CMJ with different external load
did not change significantly. In our opinion it was related to the fact,
that even though with an increase of an external load the hmax values
decreased, but the decrease of the external load caused that the product of
these values (W=mgh) did stay almost constant. Observed changes of work and
maximal power in relation to an external load may be used to evaluate
strength/speed abilities at different stages of training process as well as
to objective registering of training loads applied. On the basis of this
study we cannot state anything about their usefulness in comparison of
athletes of different sport disciplines.
REFERENCES
1. Alabin W.G., Krivonosov M.P.:Trienazery i specjalnyje upraznienija w
legkoj atletikie. Fizkultura i Sport, Moskwa, 1982
2. Bartosiewicz G., Wit A.:Skocznosc czy moc? Sport Wyczynowy 1985, 6, 7-
14
3. Bartosiewicz G., Skladanowska K., Trzaskoma Z.: Proba oceny mozliwosci
silowo-szybkosciowych szermierzy. Sport Wyczynowy 1986, 5, 3-14
4. Desipres M.: Polyparametric study of the vertical jump. In: Paavo V.
Komi (ed.), Biomechanics V-B, 73-80, Park Press. Baltimore, 1976
5. Komi P.V., Bosco C.: Utilization of stored elastic energy in leg
extensors muscles by men and women. Medicine and Science in Sports,
1978, 10, 4, 261-265
6. Skladanowska K.: Rozwazania na temat poziomu przygotowania
sprawnosciowego szermierzy. Sport Wyczynowy 1975, 12
7. Tveit P.: Variation in horizontal impulses in vertical jumps. In: Paavo
V. Komi (ed.), Biomechanics V-B, 81~6, Park Press. Baltimore, 1976
8. Viitasalo J.T. et al.: Vertical jumping height in elite male volleyball
players. Volleyball 1982, 5, 18-21
9. Wachowski E. Speed and useful power in strength exercises. In: Adam
Morecki et al. (ed.), Biomechanics VII-B, 379-385, Park Press.
Baltimore, 1981
 679

COMPLEX INVESTIGATION OF TRACK START

M. Brand, H. Greiff, H. Theysohn, W. Oberste and K. Nicol
University of Munster, Munster, Germany
Up to now there are only a very few investigations of track start using
EMG analysis. Oespires (1973) compares standing start and crouch start
technique and encounters significant advantage of the crouch start
technique in terms of acceleration and load applied to the muscular system
during the first second of the starting process. Jung (1984) measured
electrical activity (EA) of four muscle groups of the rear leg using a
telemetry device. Baranow (1973) studied the fact that the heel of the rear
leg swings back during the push off phase because starting blocks normally
support a part of the foot only. His result was that this causes a loss of
time up to .12 s.
On the other hand there are a few points from which the hypothesis may
be derived that this swing back of the heel may be advantageous to the
starting process. According to muscle physiology, elastic energy is stored
during the eccentric movement phase in the cross-bridges of the muscle
fibres which can be used during the following concentric phase (e. g. Noth,
1985, p. 250). Similar, Komi (1985, p. 258) says that restitution of
elastic energy is the main source of the increased power of a stretch- and
shortening-cycle.
Another explanation of the increased power of a prestretched muscle may
be the prestretch reflex. Yamazaki et ale (1985, p. 346) give a
quantitative analysis on this subject. Bosco & Komi (1979) and Bosco, Komi
& Ito (1981) found out that the amount of stored eleastic energy grows
stronger with increasing speed of the prestretching process. This finding
was proved by Funato, Ohmichi & Miyahshita (1985, p. 64).
A further aspect of this phenomenon is given by the dynamographical
approach by Hochmuth (1982, p. 163) who derived quantitative relations for
the optimum amount of prestretch in terms of the impulses excerted during
eccentric phase and concentric phase.
Based on these findings the hypothesis is derived that the swing back of
the heel during sprint start increases performance because the increased
power of the prestretched muscle can be used this way. This will hold true
if the force increase is large enough to overcompensate the time lossj this
is most likely, if the time loss due to swing back is accordingly small.
This study is aimed at (a) giving a biomechanical description of sprint
start in general, (b) comparing trained and untrained sprinters and (c)
investigating the time loss for swing back for various types of starting
blocks which is the first step for verifying or falsifying the hypothesis
derived above.

1. METI-DDES
The study was performed as a joint investigation of the orthopedical
hospital and of the institute for sports sc!ences of Muenster University.
The design was based on the simUltaneous use of electromyography (EMG) ,
cinematography and dynamography, in order to enable complex analysis of
track start.
Two starting blocks were separately set up, the rear block was mounted
on a force-platform. Consequently the vertical force and the horizontal
force in saggital direction could be recorded and analyzed by a computer.
680

The measuring frequency was about 500 Hz. The platform was mounted at the
edge of the long jump-pit of an indoor track at equal height with the
surrounding track surface. In addition, the computer recorded two EMG
channels for temporal synchronisation. The force histories included the
spike of a hammer blow which functioned as acoustic starting signal and
served as a visual synchronisation with the 16 mm - film. The trials were
filmed at a frequency of 55 fps. The marked bodypoints were digitized and
the angles of the bodysegments could be calculated.
For registration of the EMG of several muscles of foot, leg and hip we
used a modified 16 channel telemetry device offering a band-width up to
1000 Hz (433 MHz Pulse Duration Modulation, Glonner Corp.). Usually, this
device has only a passive low-pass cutoff of 6 dB/oct. This allows for a
lower cutoff-frequency of 30 Hz only, if the amplitudes of the EMG-signal
should not be reduced by more than 15 - 20 %. This is a severe limitation
for the following reason.
The EMG of the legs recorded during fast movements is disturbed by great
artefacts. The mean origin of these artefacts are variations in chemical
polarisation evoked by motions and vibrations of the surface electrodes on
the skin the frequency of which are up to 20 Hz.
In order to determinate the beginning and the end of the active phase of
a muscle exactly, it is necessary to reduce these artefacts significantly.
Passive high-pass filters (Re-type) are not capable to solve this problems
resulting in less skirt selectivity. So we used active Bessel high pass
filters offering 36 db/oct and 50 Hz cutoff frequency. The beginning and
the end of the phase of muscle activity could now be determinated with an
accuracy of 10 ms. We feel that these characteristics are the best
compromises of artefact reduction and signal loss. The effect in relation
to commercial passiv filters are shown in Figure 1.

11+++-11110+-1 60 Hz

MW~PII~~ 20 Hz

Figure 1. Comparison of EMG-

signals, smoothed with passiv Figure 2. Location of electrodes
Re-filters (below) and active on the rear leg side (left) and on
Bessel-filters (above). the front leg side (right).
Surface EMG's of the following 15 muscles were recorded: (see Figure 2).
Rear leg: MusculUS glutaeus medius; M. vastus medialis and lateralis; M.
rectus femoris; M. semitendinosus; M. gastrocnemius (caput laterale and
mediale): M. soleus; M. tibialis anterior; M. peroneus longus; M. abductor
hallucis; Front leg: M. glutaeus medius; M. vastus medialis and lateral is;
M. rectus femoris. In addition, a micro switch mounted in the shoe of the
rear leg was used for indicating contact phases. Figure 3 gives an example.
Three different types of starting blocks were used, providing a support
height of 25 cm, 15 cm and 10 em respectively. Ten students served as
SUbjects, seven of whom were active sprinters. The 100 m performance time
 681

of three male sprinters was 10.8 to 11.0 s, four female sprinters performed
11.9 to 13.5 s. One female and two male students of physical education were
members of the control group.

10

11 ! ~ /\ D .A.-.---
12
I c==J 0 CJ
13

14

15

15

Figure 3. Plots of the amplitude of EMG-

recordings. Numbers refer to electrodes, as
indicated in Figure 2. Time axis is related to
starting signal.

2. RESULTS
2.1. Average Rates
After averagIng all results, the following biomechanical description of
sprint starts can be given (see Figure 4 a).
682

During "ready position" preinnervation is registered which is

significantly reduced 100 ms before the main innervation is built up. After
the starting signal the subjects firstly innervate the proximal muscle
groups (M. glutaeus medius and M. quadriceps femoris), then the distal
groups (M. triceps surae, M. peroneus longus and M. abductor hallucis) are
activated. EA of M. quadriceps femoris begins about 52 ms before EA of M~
triceps surae. Consequently, M. triceps surae which is already activated by
that time, is stretched excentrically. Then the active phase of this muscle
starts, at the beginning of which the electrical activity is rised by the
factor of three. After the rear foot left the block, the knee is bent by M.
semitendinosus, while M. rectus femoris explosively bends the hip joint and
the rear leg passes the front leg. Then a phase of inactivity of M. triceps
surae is observed Wlile M. tibialis anterior prepares touch down of" the
rear leg by performing dorsal extension.
The following time sequence was encountered in relation to the starting
signal:
Start of EA of M. glutaeus medius and M. quadriceps femoris: 80 ms.
Start of EA of M. triceps surae: 136 ms
Start of force signal: 132 ms
End of EA of M. glutaeus medius and M. quadriceps femoris: 195 ms
End of EA of M. triceps surae: 276 ms
End of force signal: 284 ms
The standard deviation of all these data is around 30 ms.
2.2. Differences of Grou P t
First, traIned! sprln ers compared to untrained yielded the following
results. The path of the foot in the frontal plane is nearly a straight
line for the trained sprinters, whereas a curved path is seen in the
untrained group (see Figure 5). Consequently, the more economical behaviour
of the trained group leads to an earlier touch down. In the trained group,
phases of electrical activity were shorter (186 ms compared to 232 ms), the
overlap of activity phases was smaller and the degree of EA was smaller in
the trained group (see Figure 4).

o
1 ,
-~TS
I I I I fF
Figure 4. Duration of electrical
0,
f I
--OF
I I I I t~ activity of M. quadriceps femoris (QF)
liZ
1 t
--OF
I I I I t"J
and M. triceps surae (TS), and force
platform signal (F) of the rear leg,
c:, 1 --OF
---IS related to starting signal (left
I I I I "F
arrow) and end of contact (right
zt
C I
~TS
, I I I tF arrow). Data of the following groups
and individuals are shown: (a) all
C3
1 I
~TS
I I I I tF groups; (b1 ) trained sprinters; (b 2)
d
f I
--OF

,~J} untrained sprinters; (c1 ) 25 em block;

.t T', ,
OF
(c 2) 15 cm block; (c3) 10 em block;
I t
(d) trained individual, average
,t OF
d
IS
F performance;
OF (e) trained individual, extraordinary
l o
I
""
I I
2110
I I ),.,
~'J,
"...
performance; (f, g) two trials of an
untrained individual.
 683

15
32 17 F) .
r~(. .
,, , 21. '.9 \ f Figure 5. Path of the rear
./ \J foot.
(above),
Untrained individual
trained individual
(below).

Second, the influence of starting blocks was studied, the results are
given in table 1. It is seen that the variation in the foot angle which is
the parameter of the swing back of the heel during push up phase varies
according to the block height. Statistical analysis proved significant
differences on the .05-level between all blocks. On the other hand it can
be seen that these differences generate no difference at all in terms of
electrical activities of the M. quadriceps and M. triceps surae, in the
duration of the combined activity, and in the duration of the force signal.

Table :I,
Influence of block type on variation of foot angle and on
duration of phases. Combined EA means: duration of electrical
activity from the beginning of the active phase of M. quadriceps
femoris to the end of the active phase of M. triceps surae.

Block- No.of Variation EA of M. EA of M. Combined Duration

height trials of foot quadri- triceps EA of force
angle ceps surae si{nal
(cm) (degrees) (ms) (ms) (ms) ms)

25 21 2,2+1,2 114+35 144+43 196+40 153+31

15 19 4,7"+2,3 114+23 134+24 196+28 153+33
10 20 lO,3~3,5 116"+30 141"+21 205+32 150+38

In contrast to the sequence of innervation described above, one highly

trained female sprinter activates M. triceps surae before M. quadriceps
femoris (see Figure 4 e). Moreover she obviously activates other muscle
groups in that way that in none of her trials any force was recorded. It is
striking to observe such movement errors on a high level of performance.
684

3. DISCUSSION
Baranow's hypothesis of large time losses due to the swing back of the
heel could not be proved. In addition to this, an indication of time loss
could not be detected at all. So we raise the hypothesis that the low
height blocks provide an advantage because (a) there is no time loss and
(b) higher muscle force can be expected as a consequence of the
prestretching. Unfortunately, to the present day our data are not
sUfficient to prove that the force output is really higher in this case.

REFEREI\CES
1. Baranow, K. (1973). Weiter Ober den Tiefstart. Der Leichtathlet, 16,
8-9.

2. Bosco, C., & Komi, P. V. (1979). Potentation of the mechanical behaviour

of the human skeletal muscle through prestretching. Acta physiologica
Scandinavia, 106, 467-472.
3. Bosco, C., Komi, P. V. & Ito, A. (1981). Prestretch potentation of human
skeletal muscle during ballistic movement. Acta physiologica
Scandinavia, 111, 135-140.
4. Desipres, M. (1973). Comparison of the Kneeling and Standing Sprint
Starts. In: S. Cerquiglini, A. Venerando, J. Wartenweiler (Eds.),
Biomechanics III, (pp. 364-369). Basel: Karger.
5. Funato, K., Ohmichi, H. & Miyashita, M. (1985). Electromyographic
Analysis on Utilisation of Elastic Energy in Human Leg Muscles. In: D.A.
Winter, R.A. Norman, R.P. Wells, K.C. Hayes & A.E. Patla (Eds.),
Biomechanics IX-A (pp. 60 - 64). Champaign: Human Kinetics Publishers.
6. Hochmuth, G. (1982). Biomechanik sportlicher Bewegungen. Berlin:
Sportverlag.
7. Hoster, M. (1979). Weg-, Zeit- und Kraft-Parameter als EinfluBgroBen
beim Sprintstart in der Leichtathletik. Braunschweig.
8. Jung, R. (1984). Zur Bewegungsphysiologie beim Menschen: Fortbewegung,
Zielsteuerung und Sportleistungen. In: W. Berger (Ed.), Haltung und
Bewegung beim Menschen (pp. 7-63). New York: Springer.
9. Komi, P.V. (1985). Dehnungs-VerkOrzungs-Zyklus bei Bewegungen mit
sportlicher Leistung. In: M. BOhrle & A. Gollhofer (Eds.), Grundlagen
des Maximal- und Schnellkrafttrainings. (pp. 254-270). Schorndorf:
Hoffmann.
10. Noth , J. (1985). Neurologische Aspekte der Muskelelastizitat. In: M.
BOhrle & A. Gollhofer (Eds.), Grundlagen des Maximal- und
Schnellkrafttrainings (pp. 238-254). Schorndorf: Hoffmann.
11.Yamazaki, Y. Mano, T., Mitarai, G. & Kito, N. (1985). Contribution of
Stretch Reflex to Knee Bouncing Movement. In: D.A. Winter, R.A. Norman,
R.P. Wells, K.C. Hayes & A.E. Patla (Eds.), Biomechanics IX-A (pp. 346-
351). Champaign: Human Kinetics Publishers.
 685

4 X 100 M RELAY: MODELLING BY PERMUTATION TECHNIQUES AND

ULTRASOUND VELOCITY MEASUREMENT ON CURVED TRACKS
F. Liebscher*. U~ Mielkau**. W. Oberste** and K. Nicol**
University of Siegen*. Siegen. Germany; University of
Munster**. Munster. Germany
In order to select a relay team, coaches usually test sprinters over
distance of 120 m and 130 m on straight lane and curve. After selection the
runners are instructed to start their run at the beginning of the approach
zone and to perform the change over in the middle or in the second half of
the take over zone.
This procedure does not take into account that the athlete's ability in
acceleration and endurance may vary significantly so that it might be
advantageous to mind these parameters when it must be decided who is to
run, on which position he should run and how every single change over point
should be fashioned best considering the limits given by the regulations.
With this in mind an investigation was designed to
(a) develop a mathematical model of the total running time
(b) measure velocity profiles of a selected team of sprinters
(c) test variations in all these parameters in order to reach the minimum
running time.
1. MA.TI-EMA.TICAL ~EL
The dependent variable of the model which has to be minimized is the
total running time. Independent variables are
a) the selection of four runners out of a group of k runners which totals
in (~) possibilities.
b) the sequence of four runners out of a group of four which gives 4! = 24
poss ibi! i ties.
c) variation of the starting points for runners # 2 to # 4 in steps of 1 m.
d) variation of take over points for runners # 1 to # 3 in steps of 1 m.
So, the total running time can be indicated as:
t
T = .=1 t (i, j, A.,1
S.,
1
U.)
1
~ Minimum
with t = running time of sprinter j on position i, with starting point A.,
take over point S. and delivery point U.. 1
For this function the following boundary conditions have to be taken into
account
j ' {1,2, ...... k} jn # jl (i.e. every sprinter runs once)
~ = Sl = 0 (starting coordinates of the relay)
U4 = 400 (finishing coordinate of the relay)
Ai' Si' Ui of the respective take over zones. (In order to
guarantee a good take over in our model it should
have taken place 3 m before the end of the take over
zone.)
i = 1,2,3 (gained distance 1 m for every exchange
when the incoming runner passes the baton to the
outgoing runner)
686

The maximum difference in speed of the two exchanging

runners should be within a given limit.
Based on this it can be shown that a total of (k) * 4! * 103 * 133
possibilities have to be compared. This leads to more4 than 790 million
possibilities of selection for a group of 6 runners. In order to reduce
this number an operations research procedure (Muller-Merbach, 1970) was
used to preselect the promising combinations.
2. SnDY 1: SIMJLATION
The model works best if a complete velocity profile of every relay
sprinter can be fed in. Before going through the toil of taking real
profiles, a simulation of 5 velocity profiles with different running times
and racing characteristics was carried out to test possible achievements
of the model compared to the results of a team slected by an experienced
relay coach.
The performance of these 5 imaginary sprinters over 100 m and 120 mare
shown in Table 1: A and B have comparable good results, C shows a bad
endurance, E a poor start. The experienced coach selected the sequence O-A-
B-G and laid down the start at the beginning of the approach zone and the
change-over in the second half of the take over zone. For these conditions,
the model calculated a final running time of 42.72 s.
Sprinter 100 m 120 m

A 10.83 13.16
B 10.95 13.17
C 11.06 13.50
0 12.01 14.58
E 12.20 14.70

Table 1. Running times for five imaginary sprinters (in s).

In order to show some of the possibilities of the model, the
improvements in this group are explained step by step: First, the first and
the last runner eXChange their positions which results in an improvement
of .12 s (see Table 2). Second, runner 0 is replaced by runner E who has
not been selected by the coach because of his bad results. As the results
of runner E basically are influenced by his bad start, which is of no
importance on positions 2 - 4, a big improvement of .54 s is obtained.
Third, the exchange point is varied. It turns out that the optimum is
obtained if the sequence of runners is exchanged once more and if the third
point of take over lies 8 m before the point used before. The improvement
is .22 s. Fourth, the starting position is varied. The model indicates that
runner C should start his run 5 m after the point used before thus runnig a
shorter distance making up his bad endurance. By these procedures a total
improvement of .92 s is obtained.
It turns out that particular advantages can be obtained if the crew
from which the selection is made is very inhomogenious in their velocity
profiles.
3. STUDY 2: HIGH PERFORMANCES SPRINTERS
For this study the following method was used in order to obtain the
velocity profiles needed. The athletes wear a belt with a wave emitter and
 687

Sprinter Starting Reception Delivery Running

point point point time
D a a 106 12.74
A 81 107 206 10.25
B 181 207 306 10.09
C 281 307 400 9.67
Conditions defined by the coach 42.72

C a a 106 11.74
A 81 107 206 10.25
B 181 207 306 10.09
D 281 307 400 10.52
Starting runner and final runner exchanged 42.60

C a a 106 11.74
A 81 107 206 10.25
B 181 207 306 10.09
E 281 307 400 9.98
Final runner replaced 42.06

A a a 106 11.48
E 81 107 203 10.34
C 181 204 298 9.72
B 281 299 400 10.30

Individual exchange points 41.84

A a a 106 11.48
E 81 107 203 10.34
C 186 204 298 9.68
B 281 299 400 10.30

Individual starting points 41.80

Table 2. Change of sequence of runners, of exchange coordinates (in m),

and running times (in s) during optimizing procedure.
an electric power supply unit weighing about 100 g around the waist during
the test. Thus the runner can race unhandicapped. The receivers on fixed
positions detect the frequency f changing according to the Doppler Shift
and produce a signal proportional to the frequency shift which is measured
by an analog to digital converter and stored in a micro computer. This
method (Hennig and Nicol, 1976) was improved in the following way.
688

To guarantee usable signals on the whole distance several receivers

were installed along the track. As the geometry of the four relay distances
differs a special configuration of the receivers was needed for the four
different starting positions. Taking measurements on the straight parts of
the track is simple, because the major direction of the emitted signals
fits the direction of the receivers positioned beside or in extension of
the track. In this case the range of the emitter is long so that only few
receivers are needed. In the curve however the emitter rotates relative to
the fixed receiver thus limiting the range considerably. That is why at
least four receivers were needed for an undisturbed continuous measurement
(see Figure 1). These were positioned on the side outside the curve in such
a way that their optimum reception line cuts the track twice.

1
CENTRE OF
~ ---Ii
TI CUM: '. - - - -IIIIEClDI OF
'. HOVaDT

------~~~-----
~.coa~..

Figure 2. Parameters for

Figure 1. Positions of the calculating racing
receivers in the curve. velocity
The racing velocity on the straight track is directly equal the
measured radial velocity. In case the receiver is standing 1 m beside the
track and the athlete has run 7 m after pgssing the receiver, the
observation angle'(see Figure 2) is less than 8, thus cos 1 ... 1. Therefore
vE ... vr according to
vr = vE * cos,," (1)
Unfortunately it is much more difficult to make velocity profiles of
the runs in the curves, because the angle,r changes constantly and cos\r
can no longer be neglected. For the evaluation of the signals we used a
method developed by Liebscher (1985).
The measured radial velocity is known only in its amount, not in its
direction, but if it is integrated the distance between the emitter and the
receiver can be definded. This distance defines as sphere around the
emitter which must intersect the path of the emitter. The special
geometrical conditions of the curve entail the following test and
evaluation conditions: (a) The receiver is positioned in such a way that it
is on the same level as the emitter carried in an orbit of r=37 m for track
2. (The inaccuracy caused by the inclination of the runner in the curve for
v = 10 mls is about 0.2 m.) (b) Receiver, emitter and path of the emitter
lie on one plane so that for the determination of the position of the
emitter it is only necessary to intersect one circle around the receiver
 689

defined by the distance of the receiver and emitter RE and another circle
defined by the radius r T of the curved track.
To simplify the calculation a Cartesian coordinate system should be
chosen in such away that the centre of the path of the emitter lies in its
origin and the receiver on the axis in CR(o/yc) (see Figure 3). The
equations for determining the coordinates of the point of intersection S
are:
r T2 - r R2 + yc 2
,,-
ys =
2yc
(2) "
2
Xs = (r T - ys 2)1/2 (3)

---arT _
_ ._. -rR-RE

Figure 3. Determination of the position of

the emi tter.
with. r T as radius of the track and r R as distance between emitter and
receIver.
To obtain the distance r R at any given moment the measured radial
velocity must be integrated:
rr = r Ro +, vr dt
f (4)
This calculation is made for all receivers. As the measurements of the
receivers refer to different coordinate systems these coordinates must then
be transformed into one common system.
The initial radius r R must be taken before the start and is
calculated for the folloWIRg receivers from the data of the previous
receiver.
Finally the data are smoothed by cubic-spline-functions (Jordan-
Engelen and Reutter, 1978). The velocity in x- and y-direction can now be
calculated by numerical differentiation of the position coordinates. The
equation for the track velocity in the curve is
2 2 1/2
v = (v
T + Vy )
x (5)
Figure 4 shows an exaqJle. 4J to now this method was used in an
investigation with ten German elite sprinters. As the investigation had to
be carried out in very cold weather, only the starting bend and the home
straight were measured. The profiles of the remaining positions were
calculated from these data. Moreover, as these athletes were afraid of
Injuries, they did not run full speed so that neither acceleration nor
endurance capability showed up to full extent. Consequently, the profiles
were very similar so that the exchange procedures performed by the model
showed only marginal improvements. It is hoped that the investigation can
be repeated under better conditions.
6W

VP 7 KURVE
A 12
1
1
1
1~
,
~

E
B
>

~
~ 2 4 6 B 1~ 12 14 16
t [8] --->

Figure 4. Velocity profile of an athlete running the curve.

It seems probable that the improvements attained in the simulation can

also be transferred to the high performance level so that an improvement of
up to one second will be possible even on national level.

REFERENCES
1. Hennig, E. & Nicol, K. (1976): Velocity measurements without contact on
body surface points by means of the acoustical Doppler-effect. In: P. V.
Komi (Ed.) Biomechanics V-B, pp. 449-455. University Park Press, Baltimore.
2. Jordan-Engelen, G. & Reutter, F. (1978): Numerische Mathematik fur
Ingenieure. Mannheim.
3. Liebscher, F.F. (1985): Der Einsatz von Ultraschall-Geschwindigkeits-
MeBsystemen in der Biomechanik des Sports. Siegener Studien, 37, 77-96.
4. Muller-Merbach. (1970): Operations-Research. p. 312. Franz Vahlen, Berlin-
Frankfurt.
 691

MODEL OF THE LATERAL YACHT SHELL DISPLACEMENTS USED IN

YACHTSMEN BODY MOTION COORDINATION STUDIES
R. Uklanski. J. Wolf and A. Komor
Dept. of Biocybernetics. Institute of Sport. Warsaw. Poland

SUMMARY
Applying Hie special physical model of the lateral sail
yacht shell displacements the mathematical model was
constructed and simUlation of the dynamic behaviour of a yacht
during sailing (heels) was performed. This model was applied in
special measurement system for the purpose of determination of
the body control quality in the movement characteristic for
sailors.
The model is run on the EAI 2000 analog computer. A sailor,
by means of balasting on the motionless dummy of a yacht side
and observing on the oscilloscope screen a yacht mast position
in relation to the water surface, fullfils the task of
maintaining the simulated yacht in the heelless position.He is
balancing the generated stochastic interference which simUlates
changes of the ya'cht heeling aerodynamic force. During the
two-minute test an indicator of quality control for such an
exercise is calculated.
The results of the tests showed that it is a good tool for
the initial selection of candidates for intensive sailing
training. The proposed test was accepted by Polish Yachting
Association and now is in the wide use.

1. INTRODUCTION
The movement 'coordination which we understand here as a
control activity with an imposed program (imposed either by the
central nervous system or externally - as a certain task of
reproduction or compensation of the external signals imposed)
is very interesting and important sportsman's feature, often
deciding about success. A man, from the point of view of
cybernetics, acts as a regulator controlling a determinated
objec,t (Fig.t).

error E(i)az({)-y({)

\()~o__J"'.JI.
_in...!p_ut_z.!.(t-'-)__ A EGULATOR
output y(t)

FIGURE 1.Scheme of the control system

692

~his is a classic control system with the feedback, where the

neuromuscular system of a man is a regulator, with the man's
body (~r an optional external object) as a controlled
object.The task of the regulator is to minimiz~ the error (t).
A standard method for testing of the motion coordination of the
sportsmen was developed in the Department of Biocybernetics,
Institute of Sport in 1980: The purpose of it was to analyze
the stabilization process of a man's body position. This
procedure, providing an objective and measurable evaluation of
coordination ma~ be one of the sport selection factors
establishing a group highly predestined with respect to the
coordinative abilities to practice competitive sport.
Based on this assumptions the measurement system for testing
the quality of body control in 'the balasting movement
characteristic for sailors was built.

2.MATERIAL AND METHOD

The measurement system consists of several elements. First
of them is a special balasting stand which is designed as a
dummy of the side of the Finn class yacht. During the exercise
a competitor is balasting on this stand in the way similar to
reality. However, the whole stand does not make any movement,
since it is fixed fastly to the KISTLER force platform and the
whole dynamics of the simulating exercise is incorporated in a
mathematical model run on the analog computer.

""''''''----1 X !loolit., control indicator x

--I
x I
co: I
I
I
I
I
I
I
I
I
\!plaslinQ I
.... ond (II;

balasting MATHEMATICAL MODEL

I
belt OF YACHT DYNAMICS
shain gouge I
force indicator ..-,................ I
I I
Kdl!~~~:!nfo~~ ~~ ~NAL~ eOM~E~ ___ J

FIGURE 2.Scheme of the measurement system

 693

A sailor, during balasting on the motionless stand generates

two signals to the computer: a moment of force (resulting from
balasting action) which is measured by the KISTLER force
platform and the signal from the sheet indicated the level of
the take in of the sheet (i.e. ,a measure of an aerodynamic
force on the sail). Moreover, the measurement system consists
of: TEKTRONIX .oscilloscope, RACAL 7D tape recorder, and EAI
2000 analog computer with a plotter (Fig.2).
The principal part of that system is a mathematical model of
the dynamics of a yacht in the lateral plane, affected by the
action of the wind and of the balasting sailor, simulated on
the EAI 2000 analog computer.
Two signals are introduced into the above model:
1) the aerodynamic force ~ heel ing the yacht CA is imi tated by
the generated stochastic interference x and modifyied by the
level of the take in of the sheet),
2) the moment of force HZ, resulting from the balasting of a
sailor on the dummy of the yacht side, straightening the yacht.
The d.angle as an output signal constitutes the value of the
yacht heel and results from the dynamic balance between the
straightening and heeling moments of the yacht existing in a
given while.The angle is calculated from the differential
equations presented further. The task of the competitor tested
is such balasting, which causes that the indicator of the mast
position of the simulated yacht in relation to the water
surface, observed on the oscilloscope, is maintained in the
heelless position (vertical), therefore angle =0. Any other
value of ~angle calculated during two-minute standard test is
a basis for the evaluation of the objective quality control
indicator in the form:
1= JcA: S
2 dt/ idt

Please note that our control system is one-dimensional. This is

a conscious simplification of the proper task of the balasting
on the water, where the sense of sight is not an exclusive
channel of the feedback, and the heel angle is not an exclusive
value determined by the competitor.

3.THE MATHEMATICAL MODEL OF THE LATERAL YACHT SHELL

DISPLACEMENTS
The model takes care of four types of external forces acting
on sailing yacht (Fig.3):
1) aerodynamic forces ~generated in the sails,
2) hydrodynamic forces Hi) generated on the hul'l,
3) forces of hydrostatic lift HS according to Archimedes law,
4) forces of gravity (yacht and competitor) GY, ~.

-
These forces generate the moments of forces with respect to the

-
determined yacht axis of rotation in the lateral plane. The
moments of A an~HD....L0rces4auses the heel of t~e yacht,- and
the moments of HS, Gc and Gy forces - the straIghtening. For
the certain value of C!. angle, for the determined moments 6f
forces in a given time, the balance of the yacht takes place
(we are interested exclusively in the one-dimentional motion of
the yacht in the lateral plane). In case of small competitive
yachts the balasting competitor, i.e. the moment of the <k
694

force is crucial from the point of view of the global yacht's

balance. Basing on the scheme from Fig.3, a physical model of
the lateral~acht .l.!ell displacements was developed, . in. whic.!l
the forces HS and HD were replaced by the ground reactIons N
and T in the rolling motion of the model on the plane (Fig.4-) .

FIGURE 3.Scheme of the later~l yacht balance

FIGURE 4.Physical model of the lateral yacht shell

displacements
 695

An influence of the competitor on the yacht was presented only

in the form of straightening moment ~. It is measured directly
during the test on the KISTLER.... force platform on which the
balastingstand is fixed. The A force is generated bymeaItS' of
the tape recorder and has a stochastic form - wh-ich imi tates a
fluctuations of the wind force durrng the real sailing.
Additionally. the A force is modi fyied- by - the sheet tak-e- -Ih
level. The forces T and N were determined from the motion
equations of the assumed model on tne basis of the known.
assumed geometry of thi-s model. That geometry was established
in such a way that the modeled characteristics of the
straightening moment in relation to the heel angle a( is
consi stent wi th the real yacht-' s one. - The curvature of the
model geometry (Fig.4) is constructed of the two arches, which
are the sections of the cil'cles: the first one wi th the centre
in the point 0t and of radius R. and the second one with the
centre in the point ~ and of radius R'. The change of the
radius from R to R' takes place for the angle d=W1, and the
centre 02, is chosen according to the following rule: the
straight line crossing this point <020) and the mass centre of
the yacht C is pe.rpendicular to the tangent to the arch of the
model in the crossing point. By this geometry, the arm r of
the straightening moment varies according to the following
relation:
r.., =r sinal for 0( ( 0/4
ry =r' sin(~-a() for II( ~C{1

The above expresses well the character of the real curve of the
arm ry for the small competitive yachts (for al t the arm r.,
reaches the maximum, for at]. r.., =-0 and for the higher values of
0/2 r., is negati ve ; it means that the moment varies from the
straightening to the heeling one).
Applying the basic principles of the dynamics and presented
above considerations the following equations are obtained:
2 . )/( _~2,
ot=(-lICmrR sin'+A(R+(h-r)cos)-mgr slnct-i'Iz , Ic+m(R+r-2Rr coso(.
for a( (ct.
Z Z
01.=(-0( mr'R' sin(OIa-ol)+A(R'+h cosetTr' cos( ",.-(;( -Mz)/Oc+m(R'
+r,a+ 2R ' r ' cos("1.-Dl.) for a/>J'0i.

where: m-mass of the yacht

Ic-moment of inertia of the yacht in respect to its
longitudal axis
The rest of the indexes as in Fig. 4
The elaborated model was used for the purpose of simulation of
the effect of -maintaining on the lateral balance of the yacht
in the tests of the motion coordination. The equations
presented above are run during the simulation on the EAI 2000
analog computer (Fig.2).

4. RESULTS
The several tests have been performed up to now. Two main
groups have been tested:
1) sailors (Polish national olympic teams) - 23 male and 7
female subjects
696

2) nontraining persons (as a control group) - 8 male subjects

The results achieved by the sailors indicate the similar high
level (low values of the I indicator - mean value I=0.4172).An
interesting fact is that the highest level have achieved
~ingle-class yacht sailors (Finn, OK Dinghy). A statistically
higher scatter is to be noted for the results of the control
group (including an uncompleted tests oversteps of the I
indicator limi t and "the fall down" of simulated yacht).
Additionally, the mean value 1=0.5417 obtained for the" control
group is much higher (i.e. worse execution of the test) than
that for the sailors.

5. CONCLUSIONS
The sailors possessing an exellent balasting technique due
to the sailing training achieved much better results than the
nontraining subjects. It confirmed the assumptions that the
tests of the movement tasks, precisely determined for each
sport event, indicate the diagnostic character, since the
higher the sport championship, the better quality of control
for these movement task is. Moreover, the equal level of the
sailors' results may indicate an approach to a certain level of
perfectness in the performance of this activity. It was also
found that the elaborated theoretical model of the lateral
yacht displacements expressed excellently, in the sailors'
opinion, the real phenomenon of sailing (excluding the fact
that a dynamic behaviour of the yacht was received by the
tested sailor by means of sense of sight only). Very important
is that such a simulation enables the precise and objective
evaluation of the quality of balasting for the competitors in
the laboratory conditibns. It seems that the presented method
can be used as an objective tool for the preliminary selection
of the candidates for competitive sailing and for control of
training process in sailing as well.
It seems resonable to continue this type of tests by adaptation
of the standard method of the motion coordination tests to "the
other selected motion exercises, characteristic for different
sport disciplines.

REFERENCES

I.Komor A.,Morawski J.,Parrianowicz L.: Human Engineering

Approach to Body Motion Coordination Studies. In: Abstracts
of III International Congress of Biomechanics, Nagoya, 1981.
2.Marchaj C.A.: Aero-hydrodynamics of sailing, Granada,
Publishing USA, 1979.
3.Morawski J.,Komor A.,Parfianowicz L.: A Method, Test
Equipment and First Results of Body Motion Coordination
Studies. International Symposium-Biomechanics and Performance
in Sport, Koln, Verlag Karl Hofmann Schorndorf, 1983.
4.Parfianowicz L., Uklanski R. ,Wolf J.: The Measurement Methods
of Body Motion Coordination in the Selected Disciplines of
Sport (in Polish), Report of -Institute of Sport, 1984.
5.Uklanski R.: Aero-hydrodynamic Researches and Mathematical
Modelling in the Sailing (in Polish). II Conference:
Theoretical Basis of the Board-sailing, Olejnica, Zeszyt
Naukowy AWF Wroclaw nr 37, 1983.
CARDIOVASCULAR BIOMECHANICS
 699

HYDRODYNAMIC COMPARISON OF DUROMEDICS PROSTHESES AND ST. JUDE

MEDICAL PROSTHESES OF VARIOUS SIZES
R. Heiliger*. H. Lambertz** and Ch. Mittermayer*
Dept. of Pathol.* and Dept. of Internal Medicine 1**. Univ. of
Technology. Aachen. Germany

INTRODUCTION
The mechanical valves are the biggest single group of pros-
theses for valve replacement in the mitral position. Though
the mechanical valves and bioprosthetic valves show different
hydrodynamic advantages and disadvantages, none of these dif-
ferent valve types is superior in all hydrodynamic criteria
Basically two different types of mechanical heart valves,
tilting disc valves and bileaflet valves are available for
valve replacement. The aim of this study is to determine the
effectiveness of the two different models of mechanical bileaf-
let valves available for valve replacement: St. Jude Medical
and Duromedics. Altogether 10 mechanical bileaflet valves, 5
St. Jude Medical and 5 Duromedics, with different sizes from
No. 21 to No. 29 have been perfused in the mitral position of
a mock circulation under pulsatile flow conditions.

MATERIALS AND METHODS

The investigated Duromedics valves were clinical valves for
mitral valve replacement supplied by the manufacturer. Within
the group of the St. Jude Medical valves only valve No. 21 was
supplied by the manufacturer, the other valves without patho-
logical changes. Time of implantation was varying between 1
month and 34 months: St. Jude Medical No. 23: 19 months; valve
No. 25: 34 months; valve No. 27: 1 month; valve No. 29: 29
months.
The pulse duplicator for heart valve testing consists of a
left atrium, left ventricle, aorta, reflux tubing and a reser-
voir. Atrium and ventricle are made from resilient silicon rub-
ber; theri shape resembles the natural ones. The atrium has
four entrances, corresponding to the four pulmonary veins. The
silicon rubber atrium fits into a glass vessel which is similar
in shape. By means of this vessel the atrium can be admitted
with pressure so that a gentle atrial kick can be simulated in
late ventricular diastole. The atrial systole is regulated by
a constant throttle and regulating membrane throttle. Supple-
mentary pathological atrial pressure of a maximum of 50 mmHg
can be applied. The pulse duplicator is described in details
elsewhere /2/.
The investigated valve fits between atrium and ventricle,
in the mitral position. The 90 0 orientation by SCOTTEN and
WALKER /4/ was considered. Within the glass atrium a fibreglass
optic is mounted so that the valve movement can be observed
700

during the pulsatile flow.

FIGURE 1. Diagrammatic representation of the St. Jude Medical

bileaflet valve (left) and the Duromedics bileaflet valve
(right) .
Cardiac output Q is continuously variable. In this investi-
gation pulse frequency f is raised from 60 beats per minute to
100 beats in steps of 10 beats per minute. The filling volume
of the left ventricle Vlv is raised from 20 ml to 70 ml. Car-
diac output varies between 1.2 litres per minute and 7.0 litres
per minute.
Flow is measured electromagnetically 1 cm above the mitral
valve level and within the ascending aorta. Pressure is measur-
ed within left atrium, left ventricle and the aortic arch
(Figure 2). The movement of the occluder of the valve during
pulsatile flow is registered by means of a fibreglass optic
and a 16 mm camera with 64 frames per second. The orifice area
is determined from the film by means of a planimeter. Flow and
pressure signals are recorded continuously by means ofax/t-
recorder. Flow and pressure are evaluated by means of a digi-
tizer and a computer.
The filling volume of the left ventricle Vlv and the reflux
volume Vr can be determined by the course of the integrated
flow signal of the measurement immediately above the mitral
valve level.
The filling volume of the left ventricle Vlv is the volume
flowing through the open mitral valve from the left atrium into
the left ventricle during ventricular diastole. Reflux volume
Vr is the volume flowing back through the mitral valve into
the left atrium during ventricular systole. Effective filling
volume Veff is defined as the volume available to the ejection
 701

, LeI1 AlliUm
2 _YaM
3..-_
5 __
4 LeI1 IIontricle

6 Arcus Aortae
7 Oeacending Aorta
a Bifurcation
9 Return Piping
10 Reservoir
11 Fibregl Optic
12 Constant Tlvottle
13 Membrane Throttle
14 Membrane Pump

FIGURE 2. Diagrammatic representation of the pulse duplicator.

through the aortic valve into the aorta during one cycle /2/:
Veff = Vlv - Vr
Insufficiency I is defined as the reflux volume in relation
to the filling volume as a percentage:
I = Vr . 100
Vlv
In order to determine Vlv, Vr and Veff for each set of tests
six signals are evaluated and a mean value + 1 standard devia-
tion is calculated. The values of regurgitation are presented
as mean values.
Whilst filling volume of the left ventricle Vlv, reflux vol-
ume Vr, orifice area A and maximal orifice area Amax have been
measured, effective filling volume Veff, insufficiency I, mean
orifice area Amean, discharge coefficient CD, performance index
PI and efficiency index EI have been calculated.
In addition to the maximal orifice area Amax the mean ori-
fice area Amean during the opening phase of the valve T is cal-
culated:
702

Amean
if J A (t) . dt

This discharge coefficient CD is defined as the mean orifice

area Amean related to the internal valve area Ain:
CD = A~ean
A1.n
The performance index PI is the mean orifice area Arnean re-
ferred to the external orifice area Aex, which is the mounting
area:
PI = Arnean
Aex
The efficiency index EI is calculated as described by
GABBAY /1/:
EI = PI Veff
Vlv

RESULTS
Reflux volume Vr of the bileaflet valves decreases or is
constant with increasing pulse frequency f. The St. Jude Medi-
cal valves show always higher values than the Duromedics valves
with corresponding size (Figure 3). Vr of the St. Jude Medical
No. 29 decreases from 14.5 ml to 11ml; the regression equation
between Vr and f is : y = -0.098 x +20.34; the correlation fac-
tor is: r = 0.90. The reflux of the corresponding Duromedics
valve decreases slightly from 8.8 ml to 8.1 ml (y = -0.028 x
+10.89; r = 0.47). Reflux of the St. Jude Medical No. 25 de-
creases from 8.7 ml to 5.8 ml (y = -0.072 x +13.00; r = 0.95)
and Vr of the St. Jude Medical No. 21 decreases from 7 ml to
5.4 ml (y = -0.047 x +9.95; r = 0.85), whilst Vr of the Duro-
medics No. 25 varies between 2.3 ml and 2.8 ml (y = 0.013 x
+1.52; r = 0.36) and that of the Duromedics No. 21 remains
nearly constant: Vr = 2.5 ml (y = -0.003 x +2.74; r = 0.94).
The mean values of insufficiency I for all sets of data
pOint are presented in Figure 4. The insufficiency of the St.
Jude"Medical is always higher than I of the corresponding Duro-
medics.
As the St. Jude Medical valves show higher values of mean
orifice area Arnean and maxi~l orifice area Arnax, discharge
coefficient and performance index of the St. Jude Medical
valves are higher than the values of the Duromedics valves. CD
of the St. Jude Medical varies - depending on the valve size -
between 0.77 and 0.85, CD of the Duromedics varies between 0.59
andO.67. PI of the St. Jude Medical varies between 0.47 and
0.55, whilst PI of the Duromedics varies between 0.36 and 0.42.
Although the St. Jude Medical shows higher values of reflux
and insufficiency, the efficiency index of this valve is higher
than EI of the Duromedics and varies between 0.41 and 0.46,
whilst the corresponding values of the Duromedics are varying
between 0.31 and 0.36.
 703

15r--=------~---.----,-----r---~
~
., ml
CD
E "'v "55.0I t 6.4ml
j

oj 10 1--_ _---f_Du_'O_m_ed_iC'_ +

><
j
0;
&
5~---+----~----+-----~--~==~~

o
<) o

o 70 80 90 100 min-1
Frequency f

FIGURE 3. Reflux volume Vr as a function of pulse frequency f,

Vlv = 55 rnl.

~~----------------------------------------,

I1!'i!!ii!I St. Jude lIedlcal - - - - - - - - - - - - - 1

4 0 1 - - - - - - - - - III!III!I Du,omedici

H 30~-------------------1----~

f H~----------~--I---I-----
10

o
Val No.

FIGURE 4. Mean values of insufficiency I of St. Jude Medical

and Durornedics.
704

,.a ,.0

o.a - 0.8

o.a

... -
- -

1-
--

--
-

-
--

--
rmm
_
I-
It.JUH ".dlul
Duro ...dlc*
-

-
-

--
-

-
r-

-
0..

...
... f- - -- - -- I- -- i- I--- - - 0.2

.... '-- .,'--

----
'--
--- '-- - '--
27 ---
'-- '-- 0.0

FIGURE 5. Mean values of discharge coefficient CD of St. Jude

Medical and Duromedics.

,.a ,.0

... 0 .8

; D.8
I"4m It. Jude
_
'''dle._
Duro.... dlc. ...

"c

I
i ... r- 1- - I- I- - - ..4

0. 'r- 1- -- I- I- I- - - -- - - ...
..0 '-- '--
Z1
- '--
--- L-
'-- L-
.7 '-- - L- 0.0

FIGURE 6. Mean values of performance index PI of St. Jude Medi-

cal and Duromedics.

DISCUSSION
Though both valve types have the same construction principle
of two mechanical leaflets, the behaviour of the St. Jude Medi-
cal and the Duromedics under the same pulsatile flow conditions
is different. Reflux volume and insufficiency of the St. Jude
Medical is always higher than Vr and I of the Duromedics. Re-
flux of the St. Jude Medical No. 21 is even higher than that of
the Duromedics No. 25 and very close to that of the Duromedics
 705

No. 29. The insufficiency I of the St. Jude Medical is also

much higher than that of the Duromedics and the difference in
I between these two valves is higher at smaller valve sizes.
The values of regurgitation are higher for the St. Jude Medi-
cal than for the Duromedics due to the higher closing volume of
the St. Jude Medical. From the course of orifice area during
the opening phase of the valves it was determined that the St.
Jude Medical needs a longer time for opening and closing of the
valve than the Duromedics. The reflux volume decreases slightly
with increasing pulse frequency because the valves are opening
and closing slightly faster at higher pulse frequencies; thus,
the closing volume becomes smaller (Figure 7). This effect is
clearer pronounced at the St. Jude Medical because of the
higher closing volume of that valve in comparison to the Duro-
medics.

3.0 3.0 1}l'P-". . .f!i/~"'fl~. .~. -. \

~5

L,a-
Ouromedies Duromedics
2.5
Veff-= 52.48ml I \ Ve" =68.34ml

1 2.0. ,c60min-1 1 2.0. \ ,_eOmin-1

.
~
1.5 d.=29mm ..
~
1.5

C 1.0.

C 1.0.

~
0
0.5
~
0
0.5

0. 0.1 0.2 0..3 DA 0.5 0.6 0..7 0.8 0.9 to. 0. 0..1 0..2 0..3 0..4 0..5 0..6 0.7 0.8 0.9 1.0.

4.f 4.5

4.C 4.C .i''-'''''~.;\!i:\ St-:Jude Medical

lie" 56.17ml I " Ven = 50.72 ml

1 3.5 1 3.5
j
.
f.60min- 1 f = BOmin-'
~ \
..
~

1 .
3.0 d.=29mm 3.0 d..=29mm

C C 2.5

2.5

~
0
2.0
~
~
2.0.

0. 0.1 0.2 D.3 D.4 0.5 D.6 0.7 0.8 D.9 1.0 0. 0.1 0.2 D.3 D.4 0.5 D.6 0.7 0..8 D.9 1.0

Time I [ oJ Time I [secJ

FIGURE 7. Course of orifice area during the opening phase of

the Duromedics No. 29 (above) and the St. Jude Medical No. 29
(below) at 60 beats per minute (left) and 80 beats per minute
(right)
This orifice area of the St. Jude Medical is higher than the
orifice area of the Duromedics. That means the St. Jude Medical
shows smaller values of pressure drop. As the values of orifice
area are higher for the St. Jude Medical, this valve shows
higher values of discharge coefficient and performance index.
The discharge coefficient of both valves models shows that
7~

minimal 60 % and maximal 85 % of the inernal area is available

during the opening phase of the valve. In comparison to the
mounting area minimal 36 % and maximal 55 % of the natural
ostium is available for the blood flow during the opening phase
of the valve (Figure 6). The described values of discharge co-
efficient and performance index are very high when compared
with the corresponding results of bioprosthetic heart valves
/2/.
The efficiency includes both criteria for determination of
hydrodynamic effectiveness: orifice area and regurgitation. The
St. Jude Medical shows the highest values of EI even for smal-
ler valve sizes which present high insufficiencies and values
of orifice area close to these of the Duromedics valves. The
efficiency index cannot be chosen as the only criteria for the
effectiveness of artificial heart valves. Not only the absolute
values of Vr, I, Arnax and Arnean, but especially the behaviour
of the valve during different circulation conditions have to be
taken into account. Thus, the results of efficiency index are
not presented.

CONCLUSION
The St. Jude Medical shows higher values of orifice area but
also higher yalues of reflux and insufficiency. The Duromedics
as a new bileaflet model confirms that the bileaftlet valves
must not necessarily show higher values of regurgitation than
the tilting disc valves.
For smaller valve sizes the Duromedics shows nearly the same
values of orifice area, but much lower values of insufficiency
than the St. Jude Medical. For larger valve sizes the St. Jude
Medical prostheses show very big orifice areas - that is, small
pressure drops - but also very high values of insufficiency.
The values of insufficiency of the Duromedics valves are much
lower when compared with the St. Jude Medical valves and very
similar to the insufficiency of the Bjork-Shiley tilting disc
valves /3/. But for the larger valve sizes the orifice area of
the Duromedics prostheses is smaller than the orifice area of
the St. Jude Medical valves. Thus, for valve sizes equal or
higher than No. 27 none of the two valve models can be chosen
as superior. While for valve sizes smaller than No. 27 the Duro-
medics seems to be superior to the St. Jude Medical due to an
equal orifice area, but smaller insufficiency when compared
withthe St. Jude Medical.

REFERENCES
/1/ Gabbay S.; McQueen DM.; Yellin EL.; Frishman WH.; Frater
RWM: In vitro hydrodynamic comparison of mitral valve biopros-
theses. Circulation 60 Suppl.l: 62-70,1978.
/2/ Heiliger R.; Richter HA.; Mittermayer ch.: Hydrodynamic
test of bioprostheses in the mitral position. Life Support Sys-
tems 3: 327-334,1985.
/3/ Heiliger R.; Geks J.; Mittermayer Ch.: Preliminary results
of a compative test of mechanical bileaflet valves and tilting
disc valves. Biomedizinische Technik 31 (7-8): 175-177,1986.
/4/ Scotten LN.; Walker DN.; Brownlee RT.: Prosthetic mitral
valve orientation. Life Support System 2 Suppl. 1: 86-88, 1984.
 707

IN VITRO KINEMATIC STUDY OF THE OCCLUDER MOTION IN AN

ARTIFICIAL TILTING DISC HEART VALVE
P.M. Calderale*. G. Bona** and F. Vallana**
Dipartimento di Meccanica*. Polito di Torino. Torino. Italy.
Cardiovascular Prothesis Dept. SORIN Biomed.**. Saluggia.Italy

1. INTRODUCTION
Cardiac valve mechanical prostheses have been in clinical use since 1960.
Tilting disc prostheses, commercially available since 1967 are mostly used.
Several models (Bjork-Shiley, Sorin, Hall-Kaster, Lillehey - Kaster, Omni-
science) based on the tilting disc concept were developed. All these valves
include a disc type occluder retained and supported by an annular cage so
that the disc is free to oscillate around a virtual chordal axis under the
action of the blood pressure. Moreover, the disc can rotate around its own
longitudinal axis. The prostheses of this family show satisfactory hemody-
namic efficiency, good mechanical reliability and limited dimensions. par-
ticularly in terms of protrusion of the open disc in the cardiac cavity,
(1). The obturator motion law strictly depends on the geometry of the
obturator itself and of the support cage. Even small variations in the
geometry of the valve components may involve important variations in the
obturator movement characteristics. This can easily result in important
differences of the prosthesis performances in the interaction with the very
critical biologic environment, where the following characteristics are
requested: minimal resistance to forward flow, minimal pressure drop
across the valve in open position, minimal turbulences induced in the blood
stream (these can cause thrombus formation and thromboembolic problems) and
low hemolysis due to the mechanical disturbance of the blood cells (2).
From these remarks it can be clearly seen that an accurate analytical
kinematic study of the occluder motion is essential to support the research
in improving valve models and the correct prostheses design. This paper
describes the method used by our group for the valve kinematic analysis.

2. MATERIALS AND METHODS

The experimental method is the most suitable method of analyzing the
occluder motion in view of the complexity of the phenomena related to the
environment surrounding the valve in its actual application site (the
heart). Moreover, the geometry of the device is so complex that it would be
very difficult to design a satisfactory mathematical model which would be
capable of taking into account even the slight deviations from design
dimensions existing in individual devices industrially manufactured. To
implement the experimental method it is necessary to operate on a mock
circulation system that accurately simulates the actual fluidodynamic con-
ditions existing in the heart in the vicinity of the valve under test (3).
For our purpose the simulator must perform the features listed in Table
I. The research was carried out on a circulatory system simulator (Fig.l)
(4) according to the project of Swanson and Clark. This apparatus consists
of a transparent test section that simulates the left heart and aorta, with
aortic (Fig.2) and mitral valves, and it is connected to a rubber tube with
variable impedance simulating the systemic circulation. The driving system
70&

T"llLE I: ESSENTIAL FEATlJ"RES <?L 1L CIRCULATORY SYSTEM SIMIl'...AroR

1. Variable stroke volume (20 - 150 ml)
2. Variable pulse rate (40 - 200 c.p.m.)
3. Controllable systolic ejection period, fraction of beat period (0.2-0.5)
4. Flow pulse shsper to closely represent the blood flow across the prosthesis
5. Systolic, diastolic and atrial pressures measured and controllable to closely represent
physiological values
6. Direct accurate measurement of pulsatile (instant and mean) values of:
- atrial, ventricular and aortic pressure
- differential pressure across the valve
- direct and regurgitant flow through the valve
7. Test sections should model actual geometry be easily intercangeable, and suitable for
observation and photography.

FIGURE 1: Circulatory system simulator: 1) Speed control. 2) Gear motor and cross-heed drive.
3) Ventricle cylinder. 4) Valve test section. 5) Atrium. 6) Differential pressure servo
valve. 7) Servo valve switch. 8) Left ventricle pressure transducer. 9) Aortic pressure
transducer. 10) Electromagnetic flow meter. 11) Vascular tubing. 12) Clamps. 13) Reservoir.

consists of a piston having a

spring-loaded connection to the
motor in order to give appro-
priate systolic/diastolic ra-
tio. With this equipment it is
possible to operate with diffe-
rent frequences from normal
physiological rate up to 150
beats/min. Fluidodynamic para-
meters may be directly measured
by means of specific tran-
sducers.
The operative parameters of
the simulator were trimmed to
simulate physiological condi-
tions in the heart at rest: FIGURE Z: Valve test cb.!tI!Ibre
 709

75 beats/minute pulse rate;

55 cm3 stroke volume. The
duration of the ejection phase
is taken to be 33% of the dura-
tion of the cardiac cycle. The
systemic simulated impedance
was controlled in order to
obtain a waveform of the flow
through the valve and of the
aortic and ventricular pres-
sure, similar to the standard
physiological values. Fig. 3
shows a typical set of simulta-
neous measurements of aortic
(AOP) and ventricular (LVP)
pressure (superimposed) and FIGURE 3: Simultaneous recording of aortic and
flow rate (Q) obtained in the ventricular pressure (AOP, LVP) and flow (Q) at
circulatory system simulator at physiological conditions
75 beats/min.
The study was carried out on
standard clinical Sorin tilting
disc prostheses (5) manufac-
tured by Sorin Biomedica
(Fig.4) .
This valve (Fig.5) includes
a pyrolytic carbon disc with a
lenticular cross sectional
shape and a metallic cage con-
sisting of a peripheral ring
with two disc support stems.
One of these stems is placed on
the blood outflow side of the
valve (outflow stem) and it
engages into a circular groove
existing in the outflow face of FIGURE 4: SORIN cardiac valve prosthesis
the disc, preventing disc es-
cape. The opposite convex face
of the obturator (inflow face) Support Cage
is sustained by the second stem
(inflow stem). As a result of
Outfloll Stu
the stems and disc geometry,
the disc is free to tilt around
a chordal axis and to
rotate around its own longi-
tudinal axis. The struts geome-

.
try has been designed to opti-
mize the stress distribution
and the fluidodynamic behaviour Blood nOli
of the valve. The whole cage Ihrectlllll
(peripheral ring and both
stems) of this valve is pro-
duced as a single piece by
means of an original casting
technique. This manufacturing
method that avoids weldings and FIGURE 5: Prosthesis section view
710

results in a perfectly homogeneous material structure, was selected in

order to improve the mechanical durability of the prosthesis. For the
kinematic study, the disc motion of the valve placed in the simulated
aortic position was recorded through high speed cinematography. The speed
of the film (16 mm) was set at 2800 photograms per second. The cinematogra-
phic apparatus automatically superimposes a time syncronization light
spot flashing at 100 pulses per minute on the photograms.
Subsequent projection of the film at low speed and correct use of the
synchronization light spots makes it possible to study the motion characte-
ristics with a time scale which is much slower but precisely proportional
to the actual one.
The method of analysis is based on transferring the traces of the disc
profile from film photograms to diagrams. This is done by sequentially
projecting the photograms through a precision optical comparator and plot-
ting the disc image appearing on the screen. The resulting enlargement rate

TAB. 11- Operative conditions In 3 experiments

Experiment. Cardiac Frequency Peak Ventric.Mean Aortic Closure
Output pressure pressure Time
(II min) (b.p.m.) (KPa) ( KPa ) ( m sec

.-
D- 4,5 75 18,6 13,3 39
e- 4,5 75 18,6 13,3 38
4,5 75 18,6 16,0 30

u u 6
G)

--
G)

--
(II (II

E 5 E
5

-

4 4
::
u (j
o 0
Q; G) 3
> >
...as ~ 2
:::J :::J
Cl Cl
c:: 1
as i 1

65 50 40 30 20 10 0 .10 2030405060708090100
60
disc angular position (0) % of total closure time

FIGURE 6: Diagrams of angular velocity of the occluder versus fraction of total closure time
(a) and versus disc angular position (b) at different operative conditions (see Table II)
 711

FIGURE 7: Couples of sequential positions of the disc profile during the closure cycle

is 2.5. Time reference is given by synchronization light spots on the

photograms. The sequence of the angular position of the main geometrical
plane of the valve obturator, referred to time, allows analysis of the disc
movement kinematic characteristics.

3. RESULTS
The movement of a Sorin valve was filmed in different operating condi-
tions in the physiological range (Table II). The diagrams of the disc
positions during the closure cycle were studied in order to:
a) analyze the modality of variation of the disc velocity and acceleration;
b) detect the pattern of the polar curve of the disc movement (i.e. the
locus of traces of the successive instant rotation axes) comparing it
with a purely rolling motion;
c) evaluate the typology of the disc movement as this results from the
geometrical, fluidodynamic and constraint conditions.
The angular velocity of the occluder, computed from the disc position
diagrams, was plotted against the fraction of the total closure time (Figu-
re 6a) and against the angular position of the disc in respect to the main
plane of the supporting ring (Figure 6b). The results were qualitatively
analogous in the 3 different operating conditions analyzed.
In fact the angular velocity of the disc during the closing phase in-
creases irregularly in such a way that the angular acceleration never
becomes negative. This means that the disc impacts against the supporting
structures at the end of the closure cycle with a finite, non-zero, kinetic
energy. This characteristic is inherent in the concept of the tilting disc
valve, where the pivotal axis, which is in a chordal position with respect
to the base ring, can never pass over the position of a diameter thus
reversing the pressure momentum acting on the disc (this would be the
necessary condition to reduce the impact kinetic energy).
The sequential positions of the instant rotation axes were plotted ana-
lyzing couples of the disc profiles (Figure 7), corresponding to the se-
quential positions during the closure cycle, taken with reference to the
annular support ring.

4. DISCUSSION
It has been observed that the disc movement follows a complex pattern
which cannot be described as a purely rolling motion over the inflow stem.
In fact, if that were true, the polar curve of the movement (i.e. the locus
712

of traces of the successive instant rotation axes) would lay on the trace
of the convex disc surface and would travel continuously and with univocal
motion sense, from an eccentric position at the beginning of the closure
movement (when the disc is fully open) towards a diameter of the disc. On
the contrary, on the observed valves, the motion of the istantaneous rota-
tion axis is not monotone. The motion characteristics are variable on
different cycles of the same sample.
This experimental finding supports the consideration that in a tilting
disc valve of the type studied it is practically impossible to obtain a
purely rolling motion of the disc over the supporting stem because of:
a) The disc-support contact area during the movement is only a very narrow
portion on the stem and moreover it is not possible to machine the con-
tact surfaces to create a geometrical pattern capable of making them
behave as a rolling cam.
b) Very low friction exists between the disc and stem surfaces as a conse-
quence of the high level polishing required by the blood contacting
devices for hemocompatibility and for reasons of low thrombus genera-
tion. For the same reasons it is not possible to add solid elements like
teeth or grooves. This prevents the univocal coupling between corre-
sponding points on the disc and the support surface which is essential
to obtain a purely rolling movement without slipping.
c) The fluidodynamic conditions of the environment where the valve is used
are such that tangential forces act on the disc. These produce disc
translations parallel to disc's main plane.

5. CONCLUSION
The illustrated method permits the kinematic characteristics of a mecha-
nical valve prosthesis to be evaluated with sufficient accuracy to supply
useful data for design and development. In particular it is possible to
analyze the occluder angular velocity and acceleration with respect to time
and/or position, the relevant momentum and the law of motion. Moreover the
impact conditions of the disc against the support can be studied.
On the basis of these results we think that the same basic experimental
method can be used for the kinematic analysis of other types of mechanical
and biological valve prostheses.

6. REFERENCES
1. Bonchek LI: Current status of cardiac valve replacement: Selection of a
prosthesis and indications for operation - Am Heart J. 101 (1): 96
(198!)
2. Bruss KH, Reul H, Van Gilse J and Knott E: Pressure drop and velocity
fields at four mechanical heart valve prostheses: Bjork-Shiley standard,
Bjork-Shiley concave-convex, Hall-Kaster and St. Jude Medical, Life
Support System 1 (1): 3 (1983)
3. Swanson WH, Clark RE: Cardiovascular System simulation requirements, J.
Bioengineering 1 (1): 121 (1977)
4. Swanson WH, Clark RE: A simple cardiovascular system simulator: Design
and Performance, J. Bioengineering 1 (1): 135 (1977)
5. Pellegrini A, Colombo T, Peronace B, Quaini F: Results of heart valve
replacement with Sorin prosthesis, in Heart Surgery 1985, D'Alessandro L
Ed., CESI, Roma (1985).
 713

NUMERICAL STUDY OF PERISTALTIC PUMPING IN CIRCULAR CYLINDRICAL

TUBES
S. Takabatake. K. Ayukawa and A. Mori
Dept. of Mechanical Enginee~ing. Ehime University. Bunkyo-cho.
Matsuyama. Ehime. Japan

SUMMARY
A numerical method employing an upwind finite-difference technique is
adopted for an investigation of peristaltic pumping in circular cylindrical
tubes like as some organs in the living body. The fluid mechanics of peri-
staltic mixing and transport are studied in detail by analyzing the reflux
and the trapping phenomena. The results show that the nature of the reflux
phenomenon in the flow depends upon the Reynolds number as well as the wave-
number (mean tube radius/wavelength). Much greater peristaltic mixing and
transport are realized in a circular tube as compared to a two-dimensional
channel.

INTRODUCTION
The mechanism of the mixing and transporting peristaltic motion has
acquired general interest in the field of hydrodynamics, and a number of
studies have been undertaken with respect to peristaltic pumping in a two-
dimensional plane channel and in a circular cylindrical tube. Some of them
are summarized in the review by Jaffrin & Shapiro [1]. But many of these
studies have been limited by one or more simplifying assumptions, such as
small wave amplitude, long wavelength, small Reynolds number, or zero mean
volume flux, because the mathematical treatment of the full problem is dif-
ficult, and because a qualitative interpretation has been the motivation
for the study in the first place.
It is the purpose of the present article to investigate the peristaltic
pumping in circular cylindrical tubes which would seem to be more realistic
for biological systems. A numerical investigation of the influence of the
magnitudes of wave amplitude,wavelength, Reynolds number and time-mean flow
on axisymmetric peristaltic flow is conducted. In particular, the fluid
mechanics of peristaltic mixing and transport are studied in detail by ana-
lyzing the reflux and trapping phenomena. Also, from a comparison of these
results with those for the two-dimensional plane case, the quantitative dif-
ference between both cases is evaluated.

DESCRIPTION OF THE NUMERICAL ANALYSIS

The numerical method used here is
based on that developed for the two-
dimensional plane case in the earlier
paper[2]. The fluid is assumed to be
Newtonian, viscous, homogeneous and
incompressible. An axisymmetric tube
in which an infinite train of sinuso-
idal waves progresses along the wall
with velocity c is considered (see Figure 1. Geometry ofaxisymmet-
Fig.l). ric tube and traveling wave system.
714

There are four dimensionless parameters for the problem: that is, the
amplitude ratio , the wavenumber a, the Reynolds number Re and the dimen-
sionless time-mean flow ~. They are here defined by
<p = E/h. a = hll... Re = (chlv)a. ~ = Q/7I:ch2. (1)

where E and l.. are the wave amplitude and the wavelength, h is the mean tube
radius, v is the kinematic viscosity and Q is the time-mean volume flow
rate. The first three are the same definitions used in the analysis of two-
dimensional channel flow[2], whereas the fourth was defined as ~=Qlch in
the two-dimensional plane case.
The flow induced by an infinite train of peristaltic waves is expected
to be the same as a periodic flow that appears at the central part of the
finite calculation region, where no influences of the boundary conditions
on the end sections are found. We thus consider the finite region with an
integral number of waves in the wave-fixed relative frame. As boundary con-
di tions, arbitrary conditions on the two end sections, the symmetry condi tion
on the center axis, and the no-slip condition on the wall are considered.
The Navier-Stokes equations are solved numerically in the finite calcu-
lation region by using a finite-difference technique. Simpleandappropriate
oblique lattice coordinates are introduced in obtaining the finite-difference
expressions for the governing equations. The upwind successive-over-relaxa-
tion method is employed to solve these equations.
Although the previous analyses are only applicable for the specific flow
in which their simplifying assumptions are valid, the present numerical
method has no restriction in principle on the magnitudes of the wave ampli-
tude, the wavenumber, the Reynolds number, or the time-mean flow, so it is
possible to obtain solutions over a wider range of these quantities. In the
following sections, representative results of these cOIllputations are pre-
sented, and arguments about the mixing and the transport mechanisms of
peristalsis are developed.

PERISTALTIC REFLUX
In order to ascertain the presence of the reflux, it is necessary to
examine Lagrangian trajectories of individual fluid particles and to deter-
mine whether there are any particles undergoing net negative displacement.
Examples of the particle trajectories for
a typical case, with =O.2,a=O.1 ,Re=1 and ~=O, 1.2
are illustrated in Fig.2. In the figure, the
open circles indicate the initial locations 1.0
and the filled circles show the locations at
the end of one and two wave periods. Although
"[, "\, R./h =0.792
the individual particles in the flow repeat .c:: 0.8
the same trajectories periodically, they do a:: .......
I \)V
not describe exactly closed paths and their
orbital motions possess net longitudinal dis-
0.6
- R./h= 0.297
placements. Moreover, the resultant particle 0.4
period Tp, which is defined for the time in- ~( b~
terval of one orbital motion, does not always
0.2
( )
coincide with the wave period T These inter-
esting conclusions are identical with those
obtained in the two-dimensional flow study [2] o
0.3 0.4 0.5 0.6 0.7 0.8
Figure 3 shows the motions of fluid parti-
Z/l..
cles on the axis in the case of zero time-mean
flow. In this figure, the present results for Figure 2. Examples of
Re =O.01 are compared with the solution of Sha- particle trajectories;
piro et al. [3] obtained in the limit Re-O.a-O, .,o.2.a=O.1 ,Re=1.0.~=O.
 715

and with the numerical results of Tong & vawter [4] employing a finite-element
method for the limit Re-O. Each fluid particle initially(T=O) positioned at
the trough section of the wave undergoes first a backward motion followed
by a forward one, then a backward one again, as time progresses. At the end
of each particle period ( T;Tp=1), the particle experiences a net positive
longitudinal displacement for small 0., while it undergoes a net negative
longitudinal displacement for large 0.. The latter shows the presence of the
reflux.
In our results, shown by solid curves, the a=O. 01 case is in excellent
agreement with the theory of Shapiro et ai. On the other hand, the results
by Tong & vawter show that the displacements during the backward and the
forward motion are both greater than our estimations, and, therefore, their
particle trajectories differ from those obtained in the present study and
in the study of Shapiro et ai.
The meen speedU. of displacement of
the fluid particle, averaged over the
particle period, are plotted in Fig. 4 0.4
as a function of mean lateral position R.
of the particle trajectory. We may as-
certain the presence of the reflux when 0.2
there is a region in a tube where U. is ~
negative. It is seen in the figure that N
U. is negative in the region near the 0
wall of the tube and is positive in the
region near the axis when both Re and 0.
are small, for small values of ~. For -0.2
this situation, Shapiro et ai. conclud-
ed that the reflux occured near the wall o 0.2 0.4 0.6 0.8 1.0
of the tube. But Fig.4 indicates that T/Tp
the location of the reflux phenomenon in
the flow depends upon the magnitudes of Figure 3. Trajectories of fluid
Re and a. That is, the reflux takes place particles on the axis for =O.25,
near the axis for large Reynolds number ~=O. - - - , Present method for
and/or for large wavenumber, whereas it Re =O.01; , Shapiro et ai. [3]
occurs near the wall for small Re and for Re-O.o.-O; -----, Tong & Vaw-
small a as pointed out by Shapiro etai. ter[4] for Re-O.

1.0 ...-""r--~.----'-""""-'

0.75
.-:
"-
~ 0.5

0.1
0.25
0.01

o
-0.2 o 0.1 0.2 0.1 0.2 0.3 -0.2 0.2 0.4 0.6
U./c U./c U./c

(a) (b) (c)

Figure 4. Curves of mean speed of advance for the fluid particles,

plotted against the mean lateral position of the parti-
cle trajectory. (a) =o.2.a=O.Ol.~=o; (b) =O.25.Re=O.Ol.~=O;
(c) =O.2.Re =O.Ol.a=O.01.
- - - , Present method; -----, Shapiro et ai.[3].
716

Although this interesting result wi th

respect to the appearance of reflux is 1.0 '---"""-"""'r----r----,
essentially identical with that of the
two-dimensional case, there is a quanti-
tative difference between the magnitudes 0.75
of the mean speeds of advance in the two
cases. A comparison between them is made
in Fig.S. It is shown in the figure that 0.5
the magnitudes of the mean speeds in the
axisymmetric case are a factor of ten
greater than those in the two-dimensional 0.25
case. Thus it may be concluded that the
reflux phenomenon is much stronger in a
circular tube than in a two-dimensional o
channel. In other words, the peristalsis -0.2 -0.1 o 0.1 0.2
in a tubular configuration results in a U./c
strong mixing flow compared to that in a
two-dimensional plane configuration. Figure 5. Profiles of mean speed
of advance of the particle for
TRAPPING PHENOMENON =o.2.(l=O.Ol.~=o. --,Axisym-
It has been observed for certain values metric tube; ----, Two-dimensional
of and ~ that a part of the fluid in the channel [2]
tube is enclosed by a streamline separated
from the axis in the wave-fixed relative
frame and is transported with the wave speed in the laboratory frame as if
it were trapped by the wave. Examples of this interesting phenomenon called
-'trapping' are shown in Fig.6 (al for an axisymmetric flow and in Fig.6 (bl
for a two-dimensional flow. In these figures, the computational flow pat-
terns observed in the wave frame are illustrated for various flows over a
wide range of Reynolds numbers.
In the case of two-dimensional flow, as the Reynolds number increases,
the trapping becomes gradually smaller and moves backward due to the back-
ward flow from the narrowest region of the channel. Trapping ceases toexist
when the Reynolds number becomes sufficiently large. However, in axisymmetric

Zj). Z/).

(al Axisymmetric flow (bl Two-dimensional flow

Figure 6. Streamlines in wave frame for =o.7.(l=O.01.~=O.6.

 717

flow, as shown in Fig.6(al, although the front part of the trapping is seen
to be broken off, the major part does not change shape and the trapping
does not move. In addition, the maximum magnitude of the stream function in
the trapping is kept nearly constant when the Reynolds number is increased.
In order to discuss the appearance and growth of the trapping in detail,
we define certain quantities as follows: The dimensions of the trapping La
and lb may be measured by the longitudinal and lateral length of the
trapped fluid region enclosed by the split streamline. These quantities are
normalized with respect to the wavelength A and the maximum tube radius ~.,
respectively. Also the relative intensity of the trapping may be character-
ized by the ratio of the maximum magnitude of the stream function wi thin
the trapped region """= to the magnitude of the stream function along the
tube wall 1/1",,11.
The effects of Reynolds number upon the trapping lengths La and Lb and
upon the trapping intensity I"""=/~lll are represented in Fig.7. The results
for axisymmetric flow presented by the solid curves are compared with those
for two-dimensional flow presented by the dashed curves . In these figures,
the value of ~ at the left end of each curve expresses the lower limit of
the time-mean flow at the first appearance of the trapping, and the value
at the right end stands for the maximum time-mean flow transported by the
peristaltic pumping, that is, when no pressure difference is imposed along
the tube. However, because the calculations for the axisymmetric case at Re>2

1.0 1.0

.... "......
I:'
......
..! .:! ~,
,If' Re=O.Ol
,'!f- 0.1
0.5

1fI-+0.01, 0.1, 1
0.5
:It::
/I,~5
21

'H-i--2
"" ,, ,,
" , , 10
" H--S
/' , '--- R.=10
0 0.5 1.0 0 0.5 1.0
~ ~
(al Longitudinal length (bl Lateral length

0.3
, 6

,,
, R.=10 ~
~ !l
l' 0.2 " 4
~
!:! R.=10~"

J 1
5 ----H'
"
1
~

)0.1 0.1 -.....:::

2)
0.01 " '
~

/'"'"/ /
0
0 0.5 1.0
~
(cl Relative intensity

Figure 7. Effects of Reynolds number on the trapping phenomenon;

~=O.7.a=O.Ol. - - - , Axisymmetric flow; -----, Two-
dimensional flow[2].
718

failed to converge in the range of small time-mean flow, our result-s in

this range are limited.
In both the axisymmetric and two-dimensional flow cases, the curves of
la and l". shift toward the right side and their slope become steeper as R.
increases. It is evident from this fact that the time-mean flow for the
trapping limit increases as the Reynolds number increases, and, in addi-
tion, the trapping at large Reynolds number can grow suddenly as the result
of a slight increase in the time-mean flow, in contrast to that at small
Reynolds number.- In Fig~7(c), representing the trapping intensity, it is
seen that the intensities for the axisymmetric case are twenty times as
large as those for the two-dimensional case, and the dependence of the in-
tensity upon Reynolds number is less notable in the former case than.in the
latter case. Thus, it is concluded that the dimension and the intensity of
the trapping are less sensitive to an increase in Reynolds number in axi-
symmetric flow than in two-dimensional flow. The convective transport of
fluid by means of the trapping phenomenon in a circular cylindrical tube
can therefore be performed actively within a wider range of Reynolds num-
bers than in a two-dimensional plane channel.

CONCLUDING REMARKS
The influences of the magnitudes of the wave amplitude, the wavelength,
the Reynolds number and the time-mean flow on the flow are investigated
through numerical calculations over a wide range of these quantities. A
quantitative comparison is also made between the results for the axisym-
metric tube and for the two-dimensional plane channel.
The nature of the reflux phenomenon in the flow depends upon the Reynolds
number as well as the wavenumber. That is, the reflux is found near the axis
of the tube at large Reynolds number and/or at large wavenumber, while it
occurs near the tube wall when both Reynolds number and wavenumber are small,
as pointed out by Shapiro et al. The fluid mixing that results from peri-
staltic reflux is much stronger for the case of circular cylindrical tubes
than for two-dimensional channels. Also, the contribution of the trapping to
fluid transport is much greater in cylindrical geometry.
It is evident from these facts that the mixing and transport mechanisms
in peristaisis are more effective in a cylindrical tubular configuration
than in a plane configuration. This result is supported by the fact that
many organs in the living body are cylindrical in shape. It is supposed
that, in the peristalsis of the living body, such mixing motion by reflux
helps the absorption of fluid by the wall of the organ, and the convective
transport of fluid by the trapping phenomenon also promotes fluid absorption.
ACKNOWLEDGEMENT - This work was partially supported by a Grant-in-Aid for
Scientific Research from the Ministry of Education(Japan).

REFERENCES
1. Jaffrin, M.Y. & Shapiro, A.H. 1971 Peristaltic pumping. Ann. Rev. Fluid
Mech. l .. 15-35.
2. Takabatake, S. & Ayukawa, K. 1982 Numerical study of two-dimensional
peristaltic flows. J. Fluid Mech. 122, 439-465.
3. Shapiro, A.H., Jaffrin, M.Y. & Weinberg, .. S.L. 1969 Peristaltic pumping
with long wavelengths at low Reynolds number. J. Fluid Mech. 37,799-825.
4. Tong, P. & Vawter, D. 1972 An analysis of peristaltic pumping. Trans.
ASME, J. Appl. Mech. 39, 857-862.
 719

MECHANICAL PROPERTIES OF INTACT ARTERIAL SEGMENTS

H.W. Weizsacker* and J.G. Pinto**
Karl Franzens Universitat*. Graz. Austria. and San Diego State
University**. California. U.S.A.

1. INTRODUCTION
Due to their windkessel function, the mechanical properties of arteries
play an essential role in hemodynamics. Detailed knowledge of vascular rheology
is diagnostically relevant for early detection of disease processes and is cli-
nically important in vascular surgery and prosthetics. It is generally recogni-
zed that from a mechanical viewpoint arterial tissue is highly deformable, in-
elastic, incompressible, nonlinear and orthotropic (.1, 9 ). However, considerable
disagreement is evident in literature, concerning both the overall mechanical be-
havior of tubular arterial segments (3, 9, 12, 14 ) as well as the character of
anisotropy displayed by vascular tissue ( 1, 3, 9, 12 ). Conflicting claims
exist over the direction of anisotropy ( 3) and the question still appears open
whether vascular tissue can be considered isotropic under certain circumstances
(1, 4, 6, 13). The purpose of the present study was to examine the passive mech-
anical behavior of a wide variety of arteries and to find out which deformational
aspects are common to arteries from different topographical sites. To develop a
picture of wall mechanics as complete as possible, experiments covering a wide
range of deformations both in the circumferential and longitudinal direction were
considered. The anisotropic elastic response of the arterial tissue was analyzed
using the concept of incremental moduli of elasticity.

2. METHODS
2. 1. Experimental
The experiments reported here were performed on 9 cylindrical segments of
carotid artery, on 6 segments of abdominal. aorta, 6 thoracic aortas and 3 tail
arteries from 24 adult male Wistar rats of average weight 380 grams. The experi-
mental setup and protocols used in the investigation are described in detail el-
sewhere ( 7, 11, 12 ); they are shortly summarized below. The main parts of the
setup are a tensile testing machine, a piston pump and a Video Dimension Analyzer
for automatic noncontact measurement of the vessel diameter ( 5, 7 ). In a typi-
cal experiment the artery is exposed and its retraction upon excision is measured
between ink dots applied axially on the wall. The vessel segment is then closed
at one end, cannulated at the other, connected to the pump and inserted between
the clamps of the tensile machine. After preconditioning the vessel is stretched
to several successive levels of axial strain, including and past the in vivo
prestretch, and pressurized at each level, from 0 to 200 mmHg. During the test
internal pressure and axial force are recorded by transducers, while deformation
at midportion of the specimen is tracked by the video system; the undeformed
wall thickness is measured under microscope.
2.2. Analytical
For analysis we consider the arteries to be thin-walled, homogeneous, per-
fectly elastic tubes; the radial stress is neglected as being small when compar-
ed with stresses in circumferential and axial direction (2, 3, 5, 6 ).
Consider an arterial segment of length 1, midwall radius r and wall thickness
h under the action of a pressure p and an axial force F. Let L, R, H be the corres-
ponding load- free dimensions. Referred to a cylindrical frame of reference, the
extension ratios A and the natural strains E are:
~ = r / R e: z a in A
z
720

The internal pressure and axial force are sources of stress in the vascular
tissue; the average values of the circumferential and axial stress are (9):
F
+ (2 )
e 2Ttrh
Consider now incremental changes dp and dF around an overall state of loading
caused by p and F. The .incremental stress - strain relationship can then be
written as
dOe / Ee - (vez / Ezl dO z
(3)
- (v ze / Eel dOe + dO z / Ez

Ee and Ez.in E~. 3 a~e incremental moduli of elasticity and vaz ' v ze are incre-
mental P01sson s rat10S.
The symmetrical relationship between Poisson's ratios and elastic moduli( 3 ) is
vez / E z = v ze / Ee which inserted in Eq. 3 yields

dEe = dOe / Ee - (vez / Ezl dOz

(4)
dE Z = - (vez / Ezl dOe + dO z / Ez

The two equations in Eq. 4 embody three unknown material parameters, namely Ea ,
E , Va . Hence follows that one incremental experiment alone is not sufficient
t~ det~rmine these parameters. A second incremental experiment around the same
overall state of deformation must be considered. Denoting with stars (*)
quantities involved in this experiment we get

dE: = do: / Ee - (v ez / Ezl dO z *

(5)
dE: = - (V ez / Ezl dO: - dO: / E z

Notice that all four equations of Eq. 4 plus Eq. 5 refer to the same overall de-
formation state, for which the parameters are to be determined. Proper choice of
three equations out of Eq. 4 and Eq. 5 can thus be used to compute Ee' E I ve
from experimental data. The two incremental experiments considered in thfs in~
vestigation are symbolized by arrows in Fig.2. They correspond to experiments in
which A and p are held constant respectively (12 ). The following set of three
equatiofis from Eq. 4 and Eq. 5 form a well conditioned system:

dE z = - (vez / Ezl dO z + dO z / Ez (a)

*
dE: = dOe / Ee - (vez / Ezl dO z.* (b) (6)

a - *
(vez / Ezl d e + dO: / Ez (c)

Here starred quantities refer to the experiment in which Az const . From Eq.6c
and Eq. 6a follows:

E
dO z - (dO: / doe l dOe* and
dO*
z (8)
z
dE z do*
e
which inserted in Eq. 6b yields:
(9)
 721

Equations 7 - 9 were used in this study; values of midwall radius and wall
thickness in Eq. 2 were computed taking into account arterial incompressibility.

3. RESULTS AND DISCUSSION

Fig. I shows. typical pressure- diameter hysteresis loops for t?e.thoracic and
proximal (upper) abdominal aorta under passive smooth muscle cond1t10~s. Th~
diagrams reflect steady state conditions as reached after some 4 - 6 1nflat10n/
deflation cycles at constant rates of pressurization.

THORACIC AORTA ~ UPPER ABDOMINAL AORTA

~

PERIOD,1= 3 min. PERIOD, 1=3 min.

If. If.
~ 4
Ii! PARAMETER: Ii! PARAMETER:
AXIAL EXTENSION RATIO, Az
.;. AXIAL EXTENSION RATIO, Az .;.
lI!
..,
:::>
a:
VI
:::>
VI
t:l
a: t:l
n. a:
n.

:a :a
2.0 DIAMETER,d,IIO"'m] 3.5 1.5 DIAMETER, d, 110"'m] 2.5
Figure 1. Actual online experimental curves recorded with the video system and
the XY-plotter: pressure- diameter response of the aorta around the in-vivo axial
prestretch. Arrows indicate the direction of loading and unloading trajectory
We observe that the loops are very slim at least when compared with analog ones
from other biotissues (10). The hystereses are somewhat larger for the abdominal
than for the thoracic aorta and generally increase with axial prestretch. Quite
sil!lilar in trend is the axial force vs. pressure response. While the tissue be-
haves as nearly perfect elastic body at low deformations, the inelastic component
becomes apparent for axial stretches equal or larger than in vivo,as reflected
by large hystereses observed before steady state is achieved.
Curves in Fig.2 show the dependence of diameter and axial force on pressure
and longitudinal stretch. The sequence of symbols corresponding to increasing
parameter values is:
~~+X+~ZY~*XIQ-IL~~a~r

Only the ascending limbs of the hystereses are considered in the figure. As can
be seen from the upper panels, all diameter vs. pressure curves are sigma-sha-
ped. At low pressures the diameter increases more than linearly with pressure
thus displaying "rubberlike behavior" (13). As the pressure increases the curves
show an inflection reflecting maximum compliance in the physiological pressure
range. At still higher pressure the slope increases rapidly indicating that the
vessel gets very stiff. The sigmoid character of the curves is most pronounced
in the thoracica, less evident in the distal abdominal aorta. We observe further-
more that curves pertaining to different prestretches run close to each other,
above all in the physiological pressure-range,thus revealing that the diameter
is fairly independent of axial prestretch. This observation is particularly true
for the thoracic aorta. A role apart in Fig.2 is played by the tail ar-
tery which, unlike the other vessels of the set ( 8), is a muscular artery, its
wall consisting mainly (75%) of' densely packed, circumferentially oriented smooth
muscle bundels. ( 7 ). The lower panels in Fig.2 show the dependence of force on
pressure for increasing longitudinal stretches. For each vessel, the increment in
stretch from one curve to the next (i.e~ ) is constant. According to this dia-
grams, the force decreases with pressure a~ low prestretches, but increases for
large. For a specific axial extension ratio AC , the axial force remains constant
during pressurization i.e. is independent of Zinternal pressure (11). In each
experiment this very extension ratio turned out to be nearly congruent with the
 -..I
CAROTID ARTERY THORACIC AORTA L. ABDOMINAL AORTA TAl L ARTERY ~

:g PARAMETER, LONG. EXTENSION RATIO ,A2 ~

PARAMETER LONG. EXTENSION RATIO ,7I z :g PARAMETER LONG. EXTENSION RATIO ,7I z
~
PARAMETER. LONG. EXTENSION RATIO, 'liz
:8
~ ~ ~ ~

~ ~
~
~ ~
~
~
~ ~
o 0 ~
o

~
~

'"=> '"=> '"=> '"=>

'"
~
'"
~ '" ~
'"'" '"'"
~
~ ~
'"
'" '" ~
'"
0
0
~ ci g 1 J"",:,
1.73 DIAMETER,d,110-3m] 3,20 1.30 DIAMETER, d, I 10-3 m] 2.02 0.70 DIAMETE R, d, 110-3 m ] 0.99

PARAHETE R, LONG. EXTENSION RATIO, Az PARAMETER, LONG. EXTENSION RATIO ,Az PARAMETER, LONG. EXTENSION RATIO, 'liz ~ PARAMETER, LONG. EXTENSION RATlO,'lIz
OX)
'" ~
~
:= ~

~
0
,
0 o

~. ~.
~. ~.

u u
'" 0
'"
~~E~ .,;
'"z z '"z '"z
0
o

8 g
ci ci ~ i''''''~
PRESSURE,p,1103pa] 26.66 0.00 PRE S SUR E, p , I 10 3 Pa ] 26.66 0.00 PRESSURE, p,l 10 3 Pa] 26.66 0.00 PRESSURE,p,110 3 pa] 26.66

FIGURE 2. Representative examples of deformational behavior of arteries from different topographical sites.
Dependence of the outer diameter (upper panels) and longitudinal force (lower panels) on pressure for increas-
ing axial prestretches. See 'the text for sequences of symbols. The range of longitudinal extension ratios
are: Carotid artery: A.z = 1.,047 to 1.887, t::.iI.z = 4.66 x 10- 2 ; Thoracic aorta: A.z = 1.025 to 1.480, t::.A. z =
2.07 x 10- 2 ; L. abdominal aorta: A.z = 1.0168 to 1.587,t::.A.z = 3.357 x 10- 2 ; Tail artery: A.z = 1.059 to 1.299,
t::.A.z = 1.845 x 10- 2 . Ring on the left indicates average physiological deformation state; arrows in the middle
show the kind of increments used for computation of the moduli.
 CAROTI D ARTERY THORACIC AORTA U. ABDOMINAL AORTA L. ABDOMINAL AORTA
PARAMETER, LONG. EXTENSION RATIO, ~ PARAMETER LONG. EXTENSION RATIO ,'II,
A, .N
PARAMETER, LONG. EXTENSION RATIO.A, PARAMETER, LONG. EXTENSION RATIO. A,
o

{ ~; ~~ ~~
w~ .J. w~ w~
",'
=> => '"=>
~
'"=>
~
~
=>
'"
=> => =>
CJ CJ CJ
g 0 o o
:>: :>: :>: :>:
u: u: u: u:
:>: :>:
~ => => =>
"
u u u u
C< C< C< C<

0 0 "! ~oP'~:r:
.; 0
, ci I
0.0 PRESSURE, p, I 10 3 Pa J 14.00 0.0 PRESSURE; P J 110 3 Pa 1 24.00 0 00 PRESSURE,p,110 3 paJ 1S.33 0.0 PRESSURE,p,110 3 paJ 14.00

~ PARAMETER LONG. EXTENSION RATIO ,71, PARAMETE R, LONG. EXTENSION RATIO. A, PARAMETER, LONG. EXTENSION RATIO, 71, PARAMETER, LONG. EXTENSION RATIO .11,
o o
:;;
'" '"
~ ~ ~
a. a. a. a.
'"o '"o '"o '"o

'"=>
~
'"=> '"=>
~
=> => =>
CJ CJ CJ
o o o
>: >: >:
co co co
z z z
o o o

ci~~ ciP!P!~, ,,!~', o ~~-'--~-r-.--r-.--.--.-~

0.0 PRESSURE J PJ[10 3 pal 24.00 0.0 PRESSURE J PJ[10 3 pa124.00 o 0.0 PRESSURE,p,1101paJ 1S.33 .:;; 0.0 P'RE~SUREJPJ[103p~1 24.00

FIGURE 3. Typical examples of the dependence of incremental elastic moduli on pressure and axial stretch. Upper
panels: circumferential incremental modulus of elasticity; lower panels: longitudinal modulus. Inset rings de-
note states of deformation at which Ee = Ez . The range of axial stretches considered are: Carotid artery: Az =
1.093 to 1.840, 6Az = 4.66 x 10- 2 ; Thoracic aorta: Az = 1.066 to 1.460, 6Az = 2.07 x 10- 2 ; Upper abdominal
aorta: Az = 1.24 to 1.60, 6A z = 6.00 x 10- 2 ; Lower abdominal aorta: Az = 1.118 to 1.554, 6Az =
3.357 x 10- 2 . -.J
tv
...,
724
,c
in-vivo prestretch AV as measured during dissection (12). Table compares "'z
with AV; values are ~eans(+SD) over all experiments:
z -
Carotid Thoracica Abdominal Tail
Table 1 ACz 1. 715 + .05 1.490 + .08 1.574 + .06 1.290+.10
AVz 1. 710 + .06 (J.450!, .10) 1.579 + .07 1.310+ .10
Since results as in Fig.2 are valid at least up to deformation frequences of ap-
proximately 1 Hz ( the upper limit allowed by the setup), the findings have far
reaching consequences as for the role of "tethering" in vascular mechanics (14).
To conclude it should be mentioned that all arteries displayed large biological
variability; in the average/compliance decreased toward periphery, the thoracica
being the most distensible artery.
The results presented hitherto pertain to the overall mechanical behavior of
intact tubular segments. In Fig. 3 attention is turned to the elastic properties
of the wall material in terms of incremental moduli E and E . For easy correla-
tion, results in this figure were obtained with data ~rom Fi~. 1 and Fig.2; they
give a direct idea of the noise encountered in moduli computations, which stems
from the numerical differentiation of experimental curves. Nevertheless common
and general features, valid for all arteries investigated, are easily detectable
in Fig.3. It is apparent from the upper panels that the circumferential modulus,
while being a strong function of pressure, is practically independent of axial
prestretch. Reversely, the spacing of the curves in the lower diagrams reveal
that the longitudinal modulus is strongly dependent on longitudinal stretch but
only a weak function of pressure. The' poor coupling between the longitudinal and
circumferential directions is confirmed by the Poisson's ratios va ' which were
of the order of 0.25 and reached 0.5 around the physiological defo?mation state.
In Fig.3 each pair of panels, pertaining to the same vessel, have equal scales
on the axes. Hence superposition of each pair allows easy identification of the
deformational states for which Ea = E . This states of biaxial incremental iso-
tropy are marked by rings in Fig.3 an~ are linked by a dotted line. Notice that
this line departs from the undeformed state and passes through or in vicinity of
the average physiological state. This finding answers the longstanding question
in literature as to whether arteries can be considered isotropic in the undefor-
med or some other state of deformation (J, 6, 3,.4, 13). Results in Fig.3 are in
good agreement with literature (4). They are furthermore surprisingly similar
to results obtained when more complete analytical formulations are used to inclu-
de the radial modulus in the computation (12). This suggests that neglecting the
radial stress, as often done in literature, does not strongly alter the values
of the circumferential and longitudinal moduli. This aspect is important if re-
sults from different authors are to be compared.

REFERENCES
1. Busse R (ed): Kreislaufphysiologie: Georg Thieme Verlag.Stuttgart,New York,
1982. 2. Demiray H, Weizsacker HW and Pascale K: Biomed.Technik 31,46 - 52,1986.
3. Dobrin PB: Handbook of Physiology.The Cardiovascular System, Vol.III,Chap. 3,
pp. 65 - 102. American Society: Bethesda, Maryland,1983. 4. Dobrin PB: J. Bio-
mechanics 19, 351 - 358,1986. 5. Fung YC, Fronek K and Patitucci P: Am.J.Physiol.
237, H620 - H631,1979. 6. Kenner T: Arch. Kreislaufforsch. 54, 68 - 139,1967. 7.
Pascale K, Weizsacker HW and Gallasch E: Medical & Biological Engineering & Compu-
ting, Vol.23, Supplement Part I,pp. 715 - 716,1985. 8. Pascale K, Weizsacker HW,
Gauster A and Dohr G: Rheolog. Acta 21, 607 - 610,1982. 9. Patel DJ and Vaishav
RN: Basic Hemodynamics and its Role in Disease Processes. University Park Press,
Baltimore, 1980. 10. Weizsacker HW: Ph.D. Thesis. Karl-Franzens-Universitat, Graz,
1974. 11. Weizsacker HW and Pascale K: Basic Res. Cardiol. 72, 619 - 627,1977.
12. Weizsacker HW, Lambert H and Pascale K: J. Biomechanics 16, 703 - 715,1983.
13. Weizsacker Hwand Pinto JG: J. Biomechanics (submitted) ,1986. 14. Weizsacker
HW and Pascale K: 5th ICMMB, Bologna, pp. 83 - 86,1986.
Supported by grants FWF P4682 (Austria) and USPHS 29779/03 (U.S.A.).
 725

ASSESSMENT OF MECHANICAL AND THERMAL PROPERTIES OF BOVINE

PERICARDIUM
P.M. Calderale*. F. De Bona*. F. Vallana** and P. Arru**
Dipartimento di Meccanica*. Polito di Torino. Torino. Italy.
Cardiovascular Prosthesis Dept. SORIN Biomed.**. Saluggia.Italy

1. INTRODUCTION
At present bovine pericardium is widely used as an engineering material
in the manufacturing of biological heart valves prostheses(l); in order to
prevent proteolytic digestion and antigenicity reactions and minimize
microbial contamination. the tissue undergoes a fixation treatment, gene-
rally performed with glutaraldehyde solutions (2).
Since the elastic properties and the fatigue behaviour of the tissue can
be significantly affected by the chemical stabilization treatments, a
careful evaluation of the mechanical properties of the biological material
is of primary importance. Several studies (3,4,5) concerning the influence
on material mechanical behaviour of chemical treatment parameters (concen-
tration, pH and purity of the reagents, exposure time, etc.) are now
available. However none of these parameters, if taken alone, is sufficient
to characterize the real degree of tissue fixation. Recently it was
established that a thermal parameter called shrinkage temperature (Ts)
gives a more comprehensive description of stabilization properties of the
biological material (6,7). In fact, shrinkage temperature seems strictly
related only to the chemical status of the material and is therefore
independent of the morphological and geometrical characteristics of the
biological tissue.
For these reasons tensile and thermal stability tests were performed in
order to highlight the relation between the mechanical parameters and the
degree of fixation of bovine pericardium.
As we previously established (5), bovine pericardium mechanical
characteristics show a great variability. To overcome this problem tests
~ere carried out on a great number of samples, in order to analyze
statistically the results obtained.

2. MATERIALS AND METHODS

Bovine pericardium sample preparation
Bovine pericardium is a membrane structure that surrounds the heart; the
tissue is mainly composed of a thick fibrous sheet with several layers of
collagen bundles embedded in a proteoglycan matrix; because of collagen
fibers different orientation, the material can be considered isotropic (8).
At slaughter, bovine pericardium was drawn, immediately stored in
physiological solution at a temperature near 0 C and transported to our
laboratory; then, the tissue was washed and cleaned in order to obtain
patches quite uniform in thickness (0.2-0.3 mm); pericardium patches were
then chemically treated with glutaraldehyde solutions at different concen-
trations (0.2-0.5 %) and exposure time.
Tensile test specimens were cut following ASTM D 1708 Standard Procedure
(Fig. 1); the cross-section area measurement was performed with a special
apparatus consisting of a travelling microscope with graded measuring grid
726

and a hydraulically-actuated thickness gauge, permitting a constant and

controlled stylus downstroke; this methodology gives measurement re-
peatability within 10%.

FIGURE 1. Tensile test specimen FIGURE 2. Tensile test apparatus

For thermal stability test we used rectangular samples (30X8 rom) cut
near the tensile test samples.
Tensile tests
Tensile tests were performed using an Instron machine with pneumatic
grips to prevent material slippage. To overcome viscous phenomena and
obtain a nearly-static load condition, we used a low deformation rate (2
rom/min); therefore, tensile testing took a long time and it was necessary
to prevent tissue specimen from drying by means of a special apparatus
(Fig.2) that permits testing with samples dipped in physiological saline
solution.
Thermal stability tests
For thermal stability test
we used an apparatus (Fig.3-
4) consisting of a water
bath, heating the samples at
a rate of 0.5 C/min. a tem-
perature transducer, a linear
variable displacement trans-
ducer (LVDT) and a X-Y recor-
der, in order to plot sample
length variation versus tem-
perature. The temperature
corresponding to the shorte-
ning of tissue length is
called shrinkage temperature
(Fig.S); Ts may be evaluated FIGURE 3. Overall view of the thermal
with an accuracy of O.soC. stability test apparatus
 727

Ii'
~
I
20
~
li
.!'".,
~
15

FIGURE 4. Close up view of thE FIGURE 5. Evaluation of shrinkage

specimen ready for the test temperature

3. RESULTS AND DISCUSSION

In samples obtained from fresh tissue a shrinkage temperature of 62C
was measured. The fixation treatment increases Ts' In agreement with some
authors (2.6) bovine pericardium may be considered completely fixed after
a 48 h, 0.5% glutaraldehyde solution treatment, in this case Ts=84C. As
Fig.6 shows for both fresh and completely fixed tissue, Ts seems to be
little variable; in fact, thermal stability test allows to evaluate
denaturation temperature of collagen which is related to the stability of
collagen macromolecule.
100r--------------------.

cOlllpletely fixed
tissue
111111111. of s3111111es, 10 ilumb.!lf sampi!lS: 1B2
meam: 61.115 0(; lIIeaD: 84.11 0&
stand. dew.: 11.35 c staDd. dew.: 11.321.:

FIGURE 6. Histogram of Ts distribution for fresh and completely

fixed tissue
728

Therefore. Ts is influenced only by the chemical treatment that the'

tissue undergoes. and it is completely independent of tissue morphology
and thickness. collagen fiber orientation and concentration. sampling
area. etc.
For these reasons Ts may be considered to be a suitable parameter to
describe tissue degree of fixation. that may be defined as:

F ---100
Tsf-Tso

where: F= Degree of fixation (per cent)

Ts= Shrinkage temperature of the tested sample
Tso= Shrinkage temperature of the fresh tissue (62C)
Tsf= Shrinkage temperature of the completely fixed tissue (84C)

The stress strain curve of

bovine pericardium follows a R1---------71
typical pattern (5). In parti-
cular. it is possible to
evaluate some characteristic
mechanical parameters (Fig.7):
R= Maximum stress that the
sample can support
A= Elongation at the R stress
point measured from the
"zero stress point". becau-
se of the difficulty in
defining the initial sample
length
El=modulus of elasticity
measured in the linear por-
tion of the stress-strain
curve FIGURE 7. Stress-strain curve and
Es=Specific energy at failure mechanical parameters evaluation

Mechanical parameters of tissue specimens equally treated show a great

variability also for samples obtained from the same pericardium.

.-.
c:
50

i.. 40
lumb. of sa_pies: 78
~
.. 30
-r-
mean: 21.70 .,.
stand. dew.: 6.13 "10
r-
10

o
o
I
20
n 40
,......,
A 60 80 r/.) 100

FIGURE 8. Histogram of A distribution for fresh pericardium

 729

The great data scatter must be related to the particular characteristics

of the biological material (high water content. no proper shape. etc.).
which do not permit accurate cross-section area measurements. Moreover.
the concentration of collagen fibers varies randomly from sample to sam-
ple. indipendently of their thickness and sampling area; therefore. the
measured cross-section area is generally different from the actual load
supporting area.
For these reasons it is difficult to find a good correlation between
mechanical parameters and degree of fixation; Rand El seem to be
unaffected by fixation treatments. whereas A and Es values increase with
fixation. Neverthless. the specific energy at failure shows a wide scat-
ter. so that it is possible only to distinguish the fresh status from the
completely fixed status.
50

numb. of samples: 182

mean: 34.86 'I.
stand. dew.: 11.84 'I.

o ~~~~-L~~~~~~==~~~==~~

o 20 40 A 60 80 ('I.) 100
FIGURE 9. Histogram of A distribution for completely fixed pericardium

Elongation A is less variable (Fig.8-9); it increases quite linearly with

F. Fig.10 shows the results obtained from linear regression analysis; as
the fairly large interval of re-
gression enphasizes for a
90% confidence interval of regression
given measured value of A. this
model does not permit to obtain 60 80%
the degree of fixation of the ~k) 70%
tissue. but only describes qua-
litatively a variation law. It '0 .-.-.-._.- ..
- ------~:.~-::.=----:..=.= ==-=--
-.:...-.:...--"::::::::.:.-;:.:::
--
is thereby confirmed that the
elastic properties of bovine A
pericardium are affected by fi-
xation treatment; in particular 20
a changing of A values must be
related to the stiffening effect
that fixation induces at low O~~--~----L- __~L_ _ _ ~_ _ _ _ ~

deformation. resulting in in- o 20 40 60 80 ("!o) 100

creased Es and a shift in the
"zero stress point" with an
apparent increase in A values FIGURE 10. Elongation (A) versus degree
(Fig.7) . of fixation

4. CONCLUSIONS
The shrinkage temperature of bovine pericardium shows low scatter and so
permits to evaluate the degree of fixation of the tissue. On the other
730

hand. mechanical parameters show a wide scatter : Rand El remain con-

stant with fixation. while Es seems to increase. However it was not
possible to find a good correlation between specific energy at failure and
F. because of the large spread of Es values.
Elongation A is statistically less variable; it increases with fixation
showing a virtually linear relation. It is possible to explain this
behaviour by considering the stiffening effect of the fixation treatment
at low deformation.
From the engineering point of view it is surely necessary to evaluate
the relation between mechanical parameters and degree of fixation of the
tissue. in order to obtain a correct design of the bioprosthesis and
optimize the manufacturing process variables.
Neverthless because of the great variability of the mechanical parame-
ters. the thermal stability test must be considered more suitable for
Quality Control tests of fixation degree of bovine pericardium.

This work was partially supported by Grant M.P.I.

REFERENCES

1. Bodnar E. Jacoub M Eds. "Biologic and bioprosthetic valves. Jorke

Medical Books. New York. 885pp. 1986
2. Woodroof EA: The chemistry and biology of aldehyde treated tissue heart
valves xenografts. in Tissue Heart Valves. M.I.Ionescu Ed . Butterworths.
347-362. 1979
3. van Noort R. Yates SP. Martin TRP. Barker AT. Black MM: A
study of the effects of glutaraldehyde and formaldehyde on the mechanical
behaviour of bovine pericardium. Biomaterials. 3. 21-26. 1982
4. Liangjung H. Junijang W. Dnyuan C. Jinyi Y: Pericardium mechanics
Proc. of Int. Conf. of Biomech. in China. Japan and USA. Wuhuan. 328-336.
Science Press. 1984
5. Calderale PM. De Bona F. Arru p. Galloni M: Evaluation of mechanical
behaviour of bovine pericardium in relation to tissue fixation and steri-
lizing treatments. Proc. of Int.Conf.on Experimental Mechanics. Beijing.
1045-1050. 1985
6. Calderale PM. Amoretti p. De Bona F. Vallana F. Arru P: The influence
of fixation treatments on thermal stability of bovine pericardium. Trans.
of the Europ. Conf. on Biomat . Paris. 135. 1985
7. De Bona F. Amoretti p. Arru p. Calderale PM: Determinazione del grado
di fissazione del pericardio bovino mediante prove di stabilita termica. I
Congresso Nazionale SIB. Pisa. 84-85 1986
8. Ishiara T. Ferrans VJ. Jones M. Boyce SW. Roberts WC: Structure of
bovine pericardium and of unimplanted Ionescu-Shiley pericardial valvular
bioprosthesis. J.Thoracic Cardiovasc. Surg . 81. 747-752. 1981
MISCELLANEOUS SUBJECTS
 733

ELASTIC PROPERTIES OF THE HYDROXYAPATITE-BONE COMPOUND IN

RABBITS

N. M. Meenen *1, M. Dallek*, K.H. Jungbluth* and L.Weh**

Dept. Traumatology*, Orthopaedics Dept.**, University Hospital

Eppendorf, Hamburg, Germany
ABSTRACT
Porous hydroxyapatite ceramic (HAC) is known as a substance suitable for
surgical substitution of bone. As there is only little degradation of this material for
intervals up to years, the mechanical behaviour of the ceramic-bone compound is
of great importance especially when implanted close to joint surfaces.
In the rabbit model we drilled defects of 3.1 mm diameter into the cancellous
bone of the medial condylus of the femur leaving an approximately co-planar
layer of cartilage and subchondral bone of 0.5 mm thickness. We implanted
granules of hydroxyapatite ceramic (Osprovit) and studied the joints after 18
months of development and biomechanical adaptation of the compound under
permanent physiological load.
Pressure was applied by indendation with a compression testing machine within
physiological limits (39N/mm2) with and after the removal of the cartilage from the
subchondral bone. We demonstrated the topography of elasticity of the jOint
surfaces.
As no significant difference in the mechanical properties of the tested
hydroxyapatite-bone compound in comparison with normal untreated specimens
was found, hydroxyapatite with its good biocompatibility can be used as bone
substitute under biomechanical aspects even for filling bone defects under jOint
surfaces without negative effects on the articular cartilage.

INTRODUCTION
The anorganic component of the bone consists of hydroxyapatite. This name
describes the crystalline structure of the pentacalcium- mono hydroxy-
triorthophosphate Cas(P04bOH. SynthetiC powder of hydroxyapatite is thermally
converted (sintered) to ceramic with interconnected porosity. In this state the
material is suitable to fill surgical or traumatic bone defects. Chemical and
physical (including mechanical) tests reveal its structural similarity to natural
bone mineral and the shape maintaining properties. Biological experiments
demonstrate histocompatibility and conductivness to the ingrowth of bone during
regeneration. There is no toxidity, antigenicity, carcinogenicity and no marked
inflammatory response 6 . Bone is directly deposited on the implant surface
without any limiting membrane of connective tissue as seen in all other synthetic,
even ceramic implants. This leads to an ingrowth into the ceramic pores by

1 Dedicated to Prof. Dr. T.H. Schiebler, Head of the Department of Anatomy, University of
WOrzburg/Germany, on the occasion of his 65th. birthday.
Osprovit= registered trademark FeldmOhle AG Plochingen, West Germany
734

penetration and significant increase in strength of the originally brittle material9.

With our understanding of the bounds we call this "bonding-osteogenesis"7. The
maximum penetration and integration is reached after one year. Hydroxyapatite
is not significantly replaced by bone, and little degradational activity, is seen.
Therefore the mechanical behaviour of the HAC-bone composite system is of
special importance for the final result of the implantation.
In order to clarify this in a previous work4 we filled reproducible subchondral
bone defects in rabbits with hydroxyapatite ceramic. The sensitive joint surface
functioned as a biological test area for the elastic properties of the implant-bone
compound. Increased stiffness would lead to degenerative changes in articular
cartilage under physiological stress lO . As no joint alteration was ever seen in
histological examinations, we started this present long-term research, that
centres on quantification and comparability of the elastic deformation behaviour
of the entire jOint surface over the subchondral defect filled with
bone-hydroxyapatite-ceramic compound developed with permanent remodelling
under defined load.

MATERIALS AND METHOPS

Adult rabbits were chosen as experimental animal.They were anesthesized with
ketamin and the medial condylus of the femur exposed through a longitudinal
incision. A dental drill directed by a specially designed stereotactic device (see
Fig.1) was inserted from the medial view to gain a reproducible bone defect of 3.1
mm diameter under the central contact area of the condylus leaving a 0.5mm
co-planar layer of subchondral bone and cartilage.The bone defect was filled
with granules of porous hydroxyapatite (Osprovit@). In this material the pore
volume is 50%, the granules diameter 1.2mm. To assure bilateral incapacitation
we operated both femurs. Postoperativly all rabbits were allowed to move freely
in their cage. 18 months after the implantation the rabbits were sacrificed, both
femurs removed, and freed of surrounding tissue. We secured the femur with 3
claws to a solid steel plate which is mounted on a ball and socket joint.

Fig.1.Stereotactic drilling device Fig.2. Femur in indentation position

 735

On a hand driven mechanical testing machine (Reicherter Elasticometer EE

2000) the fixed distal femurs underwent impressive force of up to 50N (39N/mm2)
by a plane ended indenter of 1 mm diameter in perpendicular contact on the
articular surface. In order not to destroy the mechanical unit of the distal femurs
we indented only within physiological load zone. The indentation was performed
at 0.6mm/min rate of cross head movement. In a second series the cartilage was
carefully removed and the measuring process repeated on the subchondral
cortical bone plate.

, 3
2

Fig.3. Diagram showing the distribution and numbering system

of the indented areas on the medial femur condylus.

The 18 reproducible measuring areas on every condylus were fixed as shown in

Fig (3).The depth of indentation was measured by a displacement register in
steps of 1/100 mm. Its tip was placed on the upper end of the indenter assembly.
We plotted continuously by feeding the signal from the impressive force testing
machine into the x-y-plotter.For calculation the plotted curves underwent linear
regression from 10-50N. The displacement values were correlated statistically
and graphed as shown in Fig. 5.

RESULTS
During the postoperative course and the following 18 months no infection in the
operated joints was noticed. The animals used their legs normally.
After removal of the femurs no damage of the articular surface could be verified.
The cartilage was smooth and shiny. The subchondral plate under it was entirely
intact. All indentations were fully reversible after removal of the load, no fracture
occured during the mechanical tests. To establish the reproductiveness, repeated
indentations were performed in the same position and less than 5% of difference
were found. We evaluated 8 operated distal femur articular surfaces and 5 not
operated ones. A total number of 468 indentations were made and correlated.
The absolute indentation values ranged from 0.1 to 0.58mm on cartilage, from
0.06 to 0.28 after removal of the cartilage for 50N of applied force (39N/mm 2).
Specimens showed no significant difference between right and left leg of one
rabbit, however, there were slight differences in the indentation values of
corresponding areas of different animals. The breaking load was found at
approximately 135 to 185N/mm 2 , specimens after removal of the cartilage
fractured sooner. Examples of the original registration curves of cartilage
736

impression (a) and displacement of the subchondral bone (b) are shown in Fig.4.

N N
50 50

10 - - - - - - - I

l.-Q078mm ~ o.04mm
>----<
0 - = - - - - - - -_ _ _ __
o -~------~---

a b
Fig. 4 Original registration curves (load vs. indentation depth)
Note different registration speed (a=5sec/cm and b=2sec!cm)

HAC, carl. mad. HAC, cart. cantr. HAC, carl. lal.

no HAC, carl. mod. no HAC, cart. cantr.

1ml"
~I~
2 :I ,

no HAC, carl. lal.

5 Ii

HAC, no carl. mad. HAC, no cart. cantr. HAC, no cart. lat.

no HAC, no carl. mad. no HAC, no cart. cantre

~lffiBII
2 :I

no HAC, no carl. lal.

5

~11inH ]buII ~lhitiI

Z :I $' '2:S" 5' ':r:l" 5 Ii

Fig. 5 Mean indentation values (x 10-2mm) vs. observation points. Overlay chart
for minima and maxima. HAC=hydroxylapatite ceramic (Osprovit)
 737

The mean indentation values varied from 0.11 to 0.366mm for cartilage
measurement and from 0.066 to 0.19mm for the subchondral plate impression.
We graphed the mean displacement values for the defined areas (1-6) on three
arcs (centre, medial, lateral) of all joint surfaces divided into four groups
(operated and control, with cartilage and without respectively). (see Fig. 5).
With marked consistency from these graphs we find, that all of the 12 curves of
mean indentation values demonstrate similar behaviour:
-All curves harmonize in shape with similar topographical variation
-The displacement has its greatest numerical value at point 1
-Between point 3 and 4 the values of indentation were found to be lowest
-At point 6 the deformation increases again
-there is no discrepancy in curves of mean indentation at the centre over
HAC-ceramic filled defect areas (row 1 and 3 in the graph)
-indentation of the subchondral bone plate results in less elastic deformation

DISCUSSION
In our dynamic experimental model the integration process of hydroxyapatite
ceramic in reproducible subchondral bone defects proceeds under permanent
physiological conditions. Native hydroxyapatite is brittle as physical testing
reveals 2. Its ability to satisfy the mechanical demands of the functional use of the
leg in the early phase after implantation is documented by intact articular surface
3. A few days after the operation the onlay of osseous material and the
penetration of the interconnecting pores starts. By remodelling and substantial
integration into a trabecular arrangement the typical bonding at the interface of
bone and the ceramic material results 8. As evaluation of the force / deformation
curves reveals, the elastic properties of the whole joint loadbearing area
consisting of articular cartilage, subchondral bone plate and epiphyseal bone
with the ceramic implant shows no significant differences in the topographical
survey by indentation within physiological range of load in comparison with
untreated animals.Stiffening of the subchondral area after implantation of HAC
would lead to degenerative joint desease 10,11. In a previous work 5 we
demonstrated such lesions 12 weeks after filling the same subchondral defects
with homologous cancellous bone, that obviously does not become replaced by
autologous bone in an appropriate manner to withstand the physiological stress
during the entire integrational process.
In this present study we could not find such changes of cartilage and subchondral
plate 18 months after ceramic implantation. The documented unchanged
deformation / stress relation is known to be a very sensitive criterion for
discerning normal and osteoarthritic joint surfaces 1. The equal elastic response
of the ceramo-osseous regeneration complex and the overlying structures in the
intact joint never provokes degenerative cartilage changes.
The intent of this study is to compare the mechanical influence of implantation of
HAC rather than to determine the exact elasticity of the articular structures.
We did not deal with cartilage layer thickness and did not calculate the
E-modulus (Young's modulus) of the different components of the joint
loadbearing area. Others 1,9,12 discuss the many problems that arise from testing
standardized material bodies like plugs and cylinders.
Natural (coral) sintered hydroxyapatite has been analysed under these standard
738

conditions and found to get much stronger against compressive force when
integrated by bone. Piecuch 9 pOints out, that this is the result of the bone-
implant combination and does not reflect a change in the properties of the HAC
itself.
CONCLUSIONS
After 18 months of orientation by subchondral bone remodelling and with the
small amount of degradation processes at the granules surface the fully
integrated porous hydroxyapatite ceramic, with its outstanding biological features
in the field of synthetic bone substitutes, shows no discernable difference in its
elastic behaviour, as shown by indentation under physiological perpendicular
compressive load on the articular surface and the subchondral bone plate. This
proves the good functional compatibility of the HAC-bone compound, that makes
it suitable for filling large bone defects even close to articular structures, without
altering their important specific ability to distribute and absorb longitudinal
loading stress.
REFERENCES
1. Hayes, W.C., Keer, G., Herrmann, G., Mockros, I.F.(1972) A mathematical
analYSis for indentation tests of articular cartilage. J. Biomech 5:541-551
2. Katz, J.L. (1971) Hard Tissue as a composite material-I bounds on the
elastic behavior. Biomech 4: 455-473
3. Meenen, N.M., Jungbluth, K.H., Donath, K. (1985) Besonderheiten des
subchondralen Knochendefektes. H.Unfaliheilkunde 185: 145-153
4. Meenen, N.M., Mommsen, U., Osborn, J.F., Flosdorff, W., Jungbluth, K.H.
(1985) Hydroxylapatitkeramik zur UnterfOtterung in subchondral gelegenen
Knochendefekten. H. Unfallheilkunde 174:50-53
5. Mommsen, U., Meenen, N.M., Osterloh, J., Jungbluth, K.H. (1984) Der Wert
homologer Spongiosa bei der AuffOliung subchondral gelegener
Knochendefekte. Unfallchir 10:273-277
6. Osborn, J.F., Weiss, Th. (1978) Hydroxylapatitkeramik - ein
knochenahnlicher Biowerkstoff. Schweiz. MSchr. Zahnheilk. 88: 1166-1172
7. Osborn, J.F., Kovacx, E., Kallenberger, A. (1980) Hydroxylapatitkeramik-
Entwicklung eines neuen Biowerkstoffes und erste tierexperimentelle
Ergebnisse. Dtsche Zahnarztliche Z. 35: 54-56
8. Osborn, J.F. (1985) Implantatwerkstoff Hydroxylapatitkeramik.
Quintessenzverlag Berlin.
9. Piecuch, J.F., Goldberg, A.J., Shastry, C.Y., Chrzanowski, R.B. (1984)
Compressive Strength of implanted porous replamineform hydroxyapatite.
J. Biomed. Mat. Res. 18: 39-45
10. Pugh, J.W., Radin, E.L., Rose, A.M. (1974) Quantitative studies of human
subchondral cancellous bone. Its relationship to the state of its overlying
cartilage. J. Bone Joint Surg.56-A: 313-321
11. Simon, S.A., Radin, E.L., Paul, I.L., Rose, A.M. (1972) The response of
joints to impact loading- II. In vivo behaviour of subchondral bone. J.
Biomech 5: 267-272
12 Vriijhoef, M.M.A., Driessens, F.C.M. (1971) On the interaction between
specimen and testing machine in mechanical testing procedures.
J Biomech 4: 233-238
 739

FATIGUE BEHAVIOR OF COMPOSITE BONE CEMENT

A. Castaldini. A. Cavallini and R. Pirini
Dipartimento di Fisica. Universita di Bologna. Bologna. Italy

INTRODUCTION
Implant loosening, one of the major reasons for failure in total
arthroplasty, is a very complex phenomenon because of the large number of
mechanisms (mechanical and biological) which may cause it (1). Even so, in
a wide range of cases, fracture of polymethylmethacrylate bone cement ~rom
now on referred to as PMMA) appeared to be the source of loosening. This
led to characterization analyses of the cement and, in recent years, many
efforts have been made in an attempt to improve it. Particular attention
was paid to the fatigue properties since PMMA fracture was seen to be
cycle-dependent and not time-dependent on the applied stress (2).
In addition, bone cement porosity turned out to be the chief factor
producing PMMA failure and, therefore, the key feature in its fatigue
strength (3); thus its reduction, for example by centrifugation immediately
after mixing (4), improves the PMMA fatigue life.
This work aims to study the fatigue life behavior and the features of
fatigue-fracture-surfaces by comparing standard cement control samples with
hydroxyapatite (HA) composite ones: previous results (5) showed that
porosity morphology can also be modified by adding crystalline HA to the
bone cement.

MATERIALS AND METHODS

This paper investigates the effects induced on PMMA fatigue strength
by the presence of crystalline hydroxyapatite in its composition: for this
examination a weight fraction XO of the powder from batches of plain
Simplex P bone cement (Howmedica, Rutherford, New Jersey) was substituted
with HA fillers. The dough was prepared in the standard way following the
manufacturer's instructions. An original specimen preparation technique was
developed to make easily and directly samples with design and size suitable
for the fatigue tests, so as to avoid mechanical machining, a dangerous
operation from the point of view of surface flaw introduction.
For this purpose, after powder and liquid mixing, the dough was
injected into a cylindrical PTFE mold (Fig.1a), then a plunger, also made
of Teflon, with five finished-reamed holes was inserted into the die,
forcing the material to reflow, and held there until polymerization was
complete.
Testing was performed on specimens containing a weight fraction XO
equal to 0, 3 and 5%; measurements were also made on specimens obtained
under an applied pressure of 5 MPa to eliminate the air inclusion voids
trapped in the dough during mixing (from now on these specimens will be
referred to as HP (High Pressure) specimens).
The fatigue life behavior was measured "dry" at 37C at different
stress levels, by cycling the specimens up to failure with the following
two methods:
740

~Fnh FIGURE 1

DIElJU
a) C~lindrical mold;
b) rotating bending
fatigue apparatus;
c) sinusoidal alter-
nating three point load

@) apparatus.

(a) a rotating bending cantilever beam (length 55mm, diameter 4mm) machine
operating at 3Hz (the loading frequency bone cement experiences "in vivo"
(6)) (Fig.lb), and (b) an alternating load apparatus at 0.3Hz (bar cast in
PTFE mold dimensions 25x5x2 mm 3 ), with a cycle asymmetry coefficient R,
defined as the ratio between the minimum stress "0 min" and the maximum
stress "0 max " borne by the sample during the test, equal to zero (7)
(Fig.lc).
It should be noticed that these frequency values are sufficiently low
to assure that crack propagation is the sole mechanism responsible for
failure, i.e. it avoids thermal softening.
The number of cycles to failure was automatically recorded and a
statistical comparison, at each stress level, was made between the
different composition cements on the basis of the Student t-test at a 95%
confidence level.
Porosity evaluation was made by means of: (a) hydrostatic scale
method, and (b) weight-to-volume ratio determinations. Its morphology, that
is the statistical density and size distribution of the voids, more than
porosity itself, turned out to be the main factor determining the
mechanical performances of these composite cements (8). Particular
attention was thus paid to determining porosity patterns and their
quantitative distribution.
Therefore, the void features were examined with both optical and
scanning electron microscopes, and the pattern characteristics of the
different composition resins carefully compared (9).
The micrographic examination of the fatigue fracture surfaces was also
carried out with a scanning electron microscope (Philips 515); the
specimens were gold-coated prior to examination to avoid charging the
polymeric components.
From the fracture features, and more precisely from the failure mirror
area of the rotating bending samples (6), toughness data were deduced which
enabled the critical stress intensity factor KIc to be calculated.

RESULTS AND DISCUSSION

The results of both test data are presented in the form of standard
fatigue Wohler "S-N" plots. The introductionl of hydroxyapatite fillers into
the acrylic resin composition induces significant changes in the fatigue
strength of the starting material. This behavior can be seen in the plots
of both the sinusoidally alternating load data (Fig.2) and the rotating
bending ones (Fig.3). For practical reasons the endurance limit, or fatigue
strength, was arbitrarily chosen as the stress at whiCh the specimen will
last for 106 loading cycles.
 741

40

I
~30
~ FIGURE 2. "S-N" plots of the
~ commercial bone cement (~) and
~20 of the 3 wt% HA composite one
II>
.!:: (-0-), alternating load tested
U)
Each point represents the mean of
10 5 experimental observations and
the uncertainties are given by the
a standard deviations.
70' 702 703 70' 7rY 706 707
Number of cycles

40

FIGURE 3. "S-N" plots of standard

130 bone cements (~-), of the 3 wt%
~ HA composite cement (-0-), and of
the 5 wt% HA composite one (;J-) ,
~20 rotating bending tested. Here, as
~ in Fig.2, the graph reports the
.!::
U)
means, and their standard
70
deviations, of the observed
fatigue lives.
a
Id 102 lif 10' las 106 107
Number of cycles

It must be noticed that the presence of HA not only causes an increase

in fatigue strength, but also a shift the "S-N" diagram, as a whole,
towards higher values. This phenomenon was quantitatively described by
best-fitting the experimental data of fatigue life S with the least squares
line to the expression given by:

In(S-Se)=A-K*ln(N) (1 )

where Se is the asymptote of the diagram when the cycle to failure number N
becomes infinite, that is, when the slope of the "S-N" curve approaches
zero, and A and K are constants (10). The considerable evidence gathered
indicates that the hydroxyapatite improves the PMMA fatigue performances;
but this improvement does not monotonically depend on the filler amount:
a 5 wt % HA corresponds to an "S-N" curve lower than the 3 wt% one, as can
be seen in Fig.3, where the 5% curve lies between the standard cement and
the 3 wt% HA curve.
Cement porosity is central to this issue. In fact these results can be
compared with previous analyses (8,11) on creep and elastic properties and
their correlation with the porosity morphology modifications induced by
crystalline fillers. From the above studies, fatigue behavior was shown to
present a significant improvement exactly at hydroxyapatite content XO
equal to 3%, where the porosity patterns show the void maximum density and
minimum diameter.
To check beyond all doubt whether the void distribution itself is the
controlling factor of the cement mechanical properties, and not the HA
742

content, fatigue tests were performed

on the HP specimens: the specimens
are identical to the ones examined so
far, but completely lack voids, as
observed by means of cross section
microscopy. This last series of
experimental data confirms the above
hypothesis, i.e. non-porous standard
and composite cement sample strengths
coincide from a statistical point of 10
view (Fig.4).
Table I reports the results of
least-squares fit of the fatigue
data to the straight line oKf 10' lOS 10 6 107
In(S-Se)=A-K*ln(N), together with the Number ofcyc/es
number n of experimental points of
the data samples. FIGURE 4.Rotating bending fatigue
The correlation probability performances of HP specimens
analysis gave the linear-correlation (~-) standard cement, (-0-)3wt%
coefficients r, reported here. HA composite cement.

TABLE I Least Squares Fitting Data

ALTERNATING LOADi ROTATING BENDING iROTATING BENDING (HP)

%HA 0 3 0 3 5 0 3
Se(MPa) 5.70 6.80 4.75 6.80 5.75 8.50 8.50
A 2.06 1.89 1.87 2.01 1.98 2.89 2.85
KIc 0.18 0.17 0.16 0.18 0.18 0.41 0.40
r 0.98 0.98 0.98 0.99 0.99 0.97 0.97
N 45 45 108 108 108 24 24

For all the populations examined the probability P(r,n) is less than
0.001, indicating a very high
degree of correlation probability.
7.5 D-111Hr'<_-----------, As can be seen, the fatigue
.m
life, which ought to be correctly
defined as "the stress at which
--.. the specimen life becomes
~ infinite", is substantially
I
~7.21 ....................~, ........................................... I modified between the standard
specimens and the composite ones
~ for both failure tests. When
-.J

-
however specimens lack voids, the
filler effect does not appear
(results of high pressure specimen
Q9~------~--~~--~-1~ tests), and the PMMA fatigue life
7.8 27 3.6 4.5 54 63 is the same in the two materials.
Log(N)- The graph in Fig.5 shows
the theoretical function
In(S-Se)=A-K*ln(N) best-fitting
FIGURE 5. Calculated best-fit of the the experimental data points of
3 wt% HA composite resin fatigue the rotating bending test.
lives vs. cycles to failure in The analysis of the fatigue test
rotating bending test. (See function results was extended to the criti-
parameters in Table I). cal stress intensity factor Klc
 743

determination: fractography of
rotating bending sample surfaces
was thus carried out to identify
the failure mechanism.
When the failure occurs at
stresses smaller than 20 MPa, the
fracture surface appears textured
and less flat and smooth than at
higher loads. It is uniformly
grooved by cracks and
imperfections, with no possibility
of identifying a preferential
direction of flaw propagation.
When the samples fail at stresses
higher than 20 MPa, Le with
shorter lives, the typical
fracture surface of fatigue
process is observed, as can be
seen in Fig.6.
Here the cracked beams present a
FIGURE 6. Fatigue fracture more uniform fracture surface and
surfaces of standard (a) and the cracks, nearly all perpendicu-
3 wt% HA composite (b) cement lar to the fracture line, are less
specimens, stressed at 40 MPa. numerous than the cracks in
specimens cycled longer.
Two areas with different appearances corresponding to the change from
fatigue to fast fracture, can be distinguished, as reported in (6).
By measuring the area of the fatigue sector, it was possible to determine a
critical crack size "a" which allowed an estimation of the fracture
toughness by the equation

(2)

where Y is a geometric correction factor including a stress term and the

specimen dimensions.
The fracture toughness values so obtained by the fatigue cracked beam
fractography are equal to (1.510.09) MPa*m' and (1.610.07) . MPa*m' for
the commercial cement and for the 3 wt% HA composite one, respectively.
These data coincide, within the st~ndard deviations, to previous
corresponding results of standard charpy-V bar tests (8); moreover, the
former coincides with the Simplex ~fracture toughness reported in the
literature (12).
Table II summarizes the Student t-test analyses made between the
standard cement and the 3 wt% cpmposite one, for all the tests carried out,
with DF degrees of freedom and at a stress level referred to as $.

TABLE II Student t-test Analysis

%HA DF S(MPa) t .P Test

o 3 14 40 -2.92 0.02 KIc Fractography

o 3 22 12 -6.08 0.001 Rotating Bending Fatigue
o 3 8 12 -3.97 0.005 Alternating Stress Compression
o 3 6 12 -0.5 0.7 Rotating Bending Fatigue
(High-Pressure Specimen)
7M

CONCLUSIONS
The fatigue life behavior and the fracture surface characteristics of
PMMA and hydroxyapatite added PMMA bone cement were compared.
Some of the most important results can be summarized as follows:
1) the presence of the HA fillers increases significantly (P~O.OOl) the
PMMA endurance limit, in both the fatigue tests carried out;
2) the relationship between endurance limit and HA content is not a
monotonic function;
3) observations made on samples without voids proved that the porosity
morphology, and not the hydroxyapatite itselI, is the key factor of the
fatigue behavior;
4) fractography of the cracked beams showed that the critical stress
intensity factor in the composite cement is larger than in the standard
one.
Hence, the results indicate that the hydroxyapatite presence induces
an improvement of the fatigue behavior of the bone cement prepared in the
normal way.

REFERENCES
1. Wide bibliography in: Raab S, Ahmed M and Provan JW: The Quasistatic and
Fatigue Performance of the Implant/Bone-Cement Interface. J.Biomed.
Mater.Res.,15,159-182, 1981.
2. Feltner CE: Cycle-Dependent Fracture of Polymethyl Methacrylate. J.Appl.
Phys.,Vol.38,N.9,3576-3584, 1967.
3. Schurman DJ, Swenson LW,Jr. and Piziali RL:Bone Cement with and without
Antibiotics:a Study of Mechanical Properties.The Hip: Proceedings of the
6th Open Scientific Meeting of the Hip Soc.,Mosby CV, Co(ed) ,87-96, 197a
4. Burke DW,Gates EI and Harris WH: Centrifugation as a Method of Improving
Tensile and Fatigue Properties of Acrylic Bone Cement. J.Bone and Joint
Surgery, Vol.66,N.8,1265-1273, 1984.
5. Castaldini A, Cavallini A, Moroni A and Olmi R: Young's Modulus of
Hydroxyapatite Mixed Bone Cement. Biomaterials and Biomechanics,
Ducheyne P, Van der Perre G and Aubert AE(eds), Elsevier Science
Publishers, 5,427-432, 1983.
6. Freitag T and Cannon SL:Fracture Characteristics of Acrylic Bone Cements
II. Fatigue. J.Biomed.Mater.Res. Vol.11,609-624, 1977
7. Feodosev VI: Soprotivlenie Materialov. Mir, (ed) Moscow, 1977.
8. Castaldini A and Cavallini A: Fracture of Acrylic Composite Bone Cement.
Biomechanics:Current Interdisciplinary Research. Perren SM and Schneider
E (eds), Martinus Nijhoff Publishers, 147-152, 1985.
9. Paper presented at the European Conference on Biomaterials, in Bologna,
Italy, September 14-17th, 1986.
10.Yokobori T: The Strength, Fracture and Fatigue of Materials.Noordhoff
(Groningen) The Netherlands, Crip JDC (ed), 1965.
11.Castaldini A, Cavallini A: Creep Behaviour of Composite Bone Cement.
Biological and Biomechanical Performance of Biomaterials. Christel P,
Meunier A and Lee AJC (eds) Elsevier Science Publishers,6,525-530,1986.
12.Robinson RP, Wright TM and Burstein AH: Mechanical Properties of Poly-
methylmethacrylate Bone Cements, 27th ORS. Las Vegas, Nevada,l14, 1981.
 745

BIOMECHANICAL ANALYSIS OF COUPLING BETWEEN MANDIBLE AND

REMOVABLE PARTIAL DENTURES
P.M. Calderale*. M. Rosseito* and M. Pezzoli**
Dept.of Mech.*. Politecnico di Torino and School of
Dentistry**. University of Torino. Torino. Italy

1. INTRODUCTION
The Department of Mechanics and the School of Dentistry of Turin have
been involved in research concerning dentistry over the last ten years. The
experimental methods used in our work were the reflection photoelasticity
and the strain gage technique.
Some authors studied these problems by means of mandible replicas made of
photoelastic resin material (1). This is an isotropic. homogeneous mate-
rial, quite different from bone, which is notoriously anisotropic, non
linear and non homogeneous. Moreover, the shape of a mandible is not regu-
lar, and it is very difficult to interpret the resulting strain pattern
using transmission photoelasticity. In fact, when the thickness of photo-
elastic resin changes, the value of the fringe constant changes too, and
the same colour indicates different strains in the tested structure, or the
same strain level is indicated by different isochromatic fringes (2).
Reflection photoelastic method allows to study the strain pattern on the
real bone surface.
Our previous research concerned the relationship between structure, shape
and strain patterns experimentally obtained in human mandibles (3)(4)(5).
In particular, the effect of edentulism has been studied by looking at the
strain patterns both in the horizontal branch and in the condyles of
mandibles, referred to the shape of the condylar surface and the remodel-
ling index. This research has shown that a consistent relationship exists
between the principal strain direction experimentally induced, the direc-
tion of the bony trabeculae, radiographically observed, and the orientation
and the shape of the condyles.
Now we are studying the behaviour of removable partial dentures (RPD)
(6), that are employed to restore bilateral distal extension edentulism,
the statistically most widely encountered situation. These dentures are not
totally supported by teeth, but also by the edentulous mucosa. Because of
such a mixed support, this type of denture can torque and harmfully stress
the abutment teeth, or compress the edentulous ridge causing resorption. On
the ground of our clinical experience, biomechanical ~oncepts and litera-
ture (7)(8)(9), we can assert that the ideal prosthesis should present the
following characteristics 1) even load distribution between abutment
teeth and edentulous mucosa, 2) axial loading of abutment teeth, 3) uniform
loading of the edentulous mucosa.
The purpose of this research is to compare the biomechanical behaviour of
six removable partial dentures with distal extension bases.

2. MATERIALS AND METHODS.

Six different partial dentures were constructed on the same dried mandi-
ble, that was bilaterally edentulous to the first premolar. The designs
tested were:
746

- RPD with distal rest, circumferential clasps (Akers)

- RPD with mesial rest, circumferential clasps (Nally-Martinet)
- RPD with continuous clasp and resilient connection (Koller)
- RPD with precision attachments and splinted abutments
- RPD with stress breakers and splinted abutments
RPD with telescopic crowns with distal crest and return wires.
Reflection photoelasticity has been utilized in order to compare the RPD.
The external surface of the mandible horizontal branch was coated with a
3 mm thick layer of resin PL-l (Photolastic inc.) glued to the bone by the
PC-l reflecting glue (Photolastic inc). Because the Young's modulus of the
coating is E=2950 MPa (about 1/4 that of the cortical bone) the reinforce-
ment effect due to the resin is negligible for the qualitative strain
distribution in the mandible surface. The fringe value, considering a wave
length of 575 nm (yellow) is f=958 ~m/m.
In order to simulate mandible links
and apply physiological load a dedi-
cated device was utilized (Fig.l).
This device allows to secure in posi-
tion a brass replica of the upper arch
and glenoid fossae. A rubber bolus was
fixed between the first and second
premolar of the two arches on the
right mandible side. Temporal and
masseter muscles were simulated by
means of rope bundles bonded to the
muscular insertion areas by a cyano-
acrilic glue. The load directions can
be varied by means of guide pulleys. D
The temporo-mandibular joint is simu-
lated by the brass reproduction of the
p
glenoid fossa. The total load applied
to the mandible can be increased con-
tinuously, and it can be gauged by a
dynamometer. FIGURE 1. Loading device
The research was developed in 3
steps. Firstly, the prostheses were classified recording the minimum load
causing the first isochromatic fringe on the mandible surface. As the
tested structure was the same , the difference in minimum loads causing the
same strain level depends upon the mode of loading, and therefore on the
tested RPD. Obviously, the larger the load required to generate the first
isochromatic fringe, the better the biomechanical behaviour of a prosthe-
sis. Then the isoclinic lines were recorded maintainig the load at a con-
stant value, and isostatics were plotted. Lastly, increasing the load, the
isochromatic flow due to two dentures was recorded.

3. RESULTS
For all RPD designs tested the load was increased from zero to the load
causing the first isochromatic fringe. These measurements were made under
white circular polarized light, so the first fringe (1-2 = 958 ~m/m) was
recognizable by the purple colour. The exact fringe order was verified by a
Babinet-Soley compensator.
Five RPDs caused the formation of the first fringe near the "lateral
mental foramen" on the loaded side. Only the precision attachments, splin-
ted abutments RPD caused the formation of the first fringe under the distal
abutment tooth. This fact enables us to affirm that this denture transfers
 747

the load mainly through the attachments, whereas the edentulous ridge is
less loaded. The results of this first classification are shown in table I.

RPD Designs Load causing

first fringe [N]

Distal rest, circumferential clasps (Akers) ................. 200

Mesial rest, circumferential clasps (Nally-Martinet) ......... 230
Continuous clasp and resilient connection (Koller) ........... 240
Precision attachments and splinted abutments .......... 250
Stress breakers and splinted abutments ....................... 250
Telescopic crowns with distal crest and return wires .......... 305

Table I Classification results

After the classification, the isoclinic lines were recorded maintaining

the total load constant at 130 N. In all cases, at this load, isoclinic
lines were seen clearly, and isochromatic fringes did not appear. Then the
isostatic lines were plotted. Our previous work showed that isostatic lines
on the surface of a dentate mandible have a regular, smooth distribution
(3). It follows that a regular isostatic pattern would represent a good
denture behaviour, that can be estimated qualitatively from the isostatic
disturbance due to the RPD.
- With the telescopic denture, on the bolus side, the isoclinic appeared
only under abutment teeth and in the area under the mesial part of the
denture base (Fig. 2). Moreover, we noticed an isotropic zone between the
two abutments which is indicative of equal loading of the abutments by the
denture. Another point of load transmission is in the mesial part of eden-
tulous crest. The isostatic flow is regular, being only slightly disturbed
under abutment teeth.
- The precision attachments denture produced a considerable alteration of
isostatic flow in the area under the bolus (Fig. 3). This denture loads the
mandible through the distal abutment tooth and through the central portion
of the posterior edentulous crest.

FIGURE 2. Telescopic crowns RPD on FIGURE 3. Precision attachments RPD

loaded side: Isostatics on loaded side: Isostatics

- In the case of stress breakers denture the portion of edentulous crest

under the bolus was loaded in a similar way as with prec1s1on attachment
denture (Fig. 4). Moreover, it is possible to recognize a strain concentra-
tion in the distal area of posterior edentulous crest.
- With continuous clasps denture we notice a perturbed area under abutment
748

teeth, especially under mesial abutment, but the isostatic flow is regular
in the other areas (Fig. 5).

FIGURE 4. Stress breakers RPD on FIGURE 5. Continuous clasp RPD on

loaded side: Isostatics loaded side: Isostatics

- The denture with circumferential clasps and mesial rest causes a per-
turbation in the area under the edentulous crest (Fig. 6). Under the abut-
ment teeth the isostatic flow is only slightly perturbed.
- With circumferential clasps and distal rest RPD the most perturbed area
is under mesial abutment, the area under the bolus is slightly perturbed
and the distal area is regular (Fig. 7).

FIGURE 6. Akers RPD on loaded side: FIGURE 7. Nally-Martinet RPD on

Isostatics loaded side: Isostatics

On the opposite side to the bolus only two types of flow were noticed.
Telescopic crowns RPD, continuous clasps RPD and circumferential clasps
with distal rest RPD gave a smooth flow pattern, the other three RPD gave a
perturbed pattern.
As in the prostheses classification step the telescopic crowns RPD shows
the better biomechanical behaviour. In fact it perturbes only slightly the
isostatic flow. The stress breakers RPD, that caused the first isochromatic
fringe at the lower load, causes the largest isostatic flow disturbance, so
it shows the worst biomechanical behaviour. The other prostheses show an
intermediate biomechanical behaviour.
Increasing the load, we recorded the isochromatic fringe flows using the
telescopic crowns RPD and continuous clasps RPD, that showed the best
biomechanical behaviour in the previous steps. The test of different pros-
theses could have caused the debonding of the coating, because we had to
use the PC-l glue, containing aluminium powder, and its adhesion to bone is
 749

lower than to metals. We recorded the strain pattern under loads of 350,
400, 450, 500 N. Increasing the load further the coating had come unstuck,
and we had to stop the test. The higher fringe order obtained was n=1.6
(1-2 = 1561 ~m/m). Comparing the isochromatic fringes due to these pros-
theses we can make the following considerations:
- At the lower load the telescopic denture produces less strain in the
edentulous area. Both the dentures present a black area under the abut-
ments. Also on the unloaded side, a higher strain magnitude was generated
by continuous clasp denture (Fig. 8).
- At the higher load the continuous clasps RPD changes its mode of loading
the mandible, and the magnitude and distribution of strains becomes practi-
cally the same for the two dentures. However, the isotropic zone is confir-
med with the telescopic crowns RPD, while the black area under abutments
disappears with the continuous clasps RPD (Fig. 9).

FIGURE 8. Isochromatics at 350 N: FIGURE 9. Isochromatics at 500 N:

Telescopic crowns RPD above Telescopic crowns RPD above
Continuous clasps RPD below Continuous clasps RPD below

In all phases of our research, the telescopic crowns RPD showed the
better biomechanical behaviour, providing the most equitable transmission
of occlusal loads to the mandible. The stress breakers RPD produced the
largest perturbation of the strain flow lines, and loaded the mandible in
a incorrect way.

4. CONCLUSIONS AND DEVELOPMENTS

This research confirms that it is possible to obtain satisfactory results
for studying the RPD biomechanical behaviour by means of reflection photo-
elasticity. However, this technique is not easy to employ because there are
a lot of difficulties in the preparatory phases: the coating must present a
constant thickness, it must copy perfectly the mandible surface, the glue
layer must be thin and constant. Moreover, the interpretation of isoclinic
pattern can be difficult. The room temperature must be kept constant: a
difference of lOC, between the test and gluing temperatures causes
residual birifrangence.
To obtain results concerning strain pattern at higher load it is necessa-
ry to use a glue having a better adhesion than PC-I, normally used in
photoelastic tests. This is important because the denture, with increasing
total load applied, can change the mode of loading the mandible.
750

Finally, no information about internal strain pattern can be achieved by

this technique. Moreover, it is very difficult to assess the stress, be-
cause we know approximately the Young's modulus of the bone, but not Pois-
son's ratio.
To complete the research and to study the whole internal structure, it is
desiderable to use a mathematical model, employing, for example, the finite
element method. The mesh of a mandible must be very detailed because it has
to reproduce a very complex structure. A mesh of this kind is under
developement. An initial approximate mesh is showed in fig.10. It employs 8
and 6 node elements. In every section the internal elements are employed to
simulate the spongy bone, the external ones to simulate the cortical bone.
The mesh of the whole dentate mandible consists of about 1000 elements. A
parametric analysis will allow us to determine the best Poisson's ratio
comparing the model results with the experimental data.
In this model, teeth are
roughly simulated by 8 node e-
lements.
To study the teeth behaviour we
will employ a detailed mesh si-
mulating the real supporting
structure.

This work was sponsored by the

Italian Research Council.
(CNR Grants n. 83.00150.07 and
n. 85.02152.07)

FIGURE 10: Mandible mesh

5.REFERENCES
1. Thompson WO, Kratochvil FJ, Caputo AA: Evaluation of photoelastic
stress patterns produced by various designs of bilateral distal-exten-
sion removable partial dentures. J. Prosthet. Dent. 46, 21-, 1981
2. Frocthe MM: Photoelasticity. J. Wiley & Sons, New York, 1949
3. Mongini F, Calderale PM, Barberi G: Relationship between structure and
the stress pattern in the human mandible. J. Dent. Res., 58, 12, 2334-,
1979
4. Calderale PM, Fasolio G, Mongini F: Experimental analysis of stress
influencing mandibular remodelling. Ac. Orth. Belgica, 46, 5, 601-,
1980
5. Mongini F, Preti G, Calderale PM, Barberi G: Experimental strain ana-
lysis on the mandibular condyle under various conditions. Med. & Bio.
Eng., 19, 521-, 1981
6. Pezzoli M, Calderale PM, Rossetto M: Evaluation of load transmission by
distal-extension partial dentures by using reflection photoelasticity.
J. Prosthet. Dent., 56, 329-, 1986
7. Kratochvil FJ: Influence of occlusal rest position and clasp design on
movement of abutment teeth. J. Prosthet. Dent., 13, 114-, 1963
8. Cecconi BT, Asgar T., Dotz E: The effect of partial denture clasp
design on abutment tooth movement. J. Prosthet. Dent., 25, 44-, 1971
9. Kratochvil FJ, Caputo AA: Photoelastic analysis of pressure on teeth
and bone supporting removable partial dentures. J. Prosthet. Dent., 32,
52-, 1974
 751

ON THE EXISTENCE OF THE ORTHODONTIC CENTER OF RESISTANCE

P.G. van der Varst*, M.C.R.B. Peters* and A. de Boer**
Dept. Cariology & Endodontology*, Nijmegen, The Netherlands,
Natl. Aerospace laboratory**, The Netherlands

1. INTRODUCTION
Effective design and usage of orthodontic appliances depend strongly on a
thorough knowledge of the biomechanics of tooth movement. One basic concept in
this field of biomechanics is called: the center of resistance (COR). It is
asserted that a single force acting through the center of resistance effects
pure translation of a tooth (Burstone 1985, p. 196). (See also Smith et al. 1984;
Nikolai, 1985, p. 47; Burstone et al., 1980; Nikolai, 1975; Bowden, 1978; Hoce-
var, 1981; Roberts et al., 1982; Nikolai 1982). Invariably it is assumed that
such a center of resistance always exists for single- and multi-rooted teeth and
even for units of teeth. However, some basic biomechanical reflections lead to
the suspicion that the general existence of the COR might not be obvious at all.
Therefore, a study was carried out to investigate this matter in more detail.
After a general part (sect. 2), in which the system to be analysed is des-
cribed, a theoretical analysis of the problem at hand will be presented in sec-
tion 3. Section 3 also contains necessary and sufficient requirements which have
to be met to ensure the existence of the COR. Using experimental data concerning
the shape of the central maxillary incisor of two persons, it is next verified
(section 4) whether aforementioned requirements are met in practice. The results
are discussed in the closing section 5.

2. GENERAL PART
Applying orthodontic loads *) to teeth gives rise to various processes, each
having its own characteristic time scale.
a. Due to the visco-elastic propert~es of the periodontal ligament the tooth
is almost instantly displaced* t~*t new equilibrium position.
b. Afterwards several other processes are set in motion (Yoshikawa,
1981; Burstone, 1985; Reitan, 1985) resulting in a change of shape,
position and orientation of the socket due to bone remodelling and,
probably, in changes of the mechanical properties of the tissues because
of histochemical changes of the periodontal ligament.
In orthodontics it is the bone-remodelling that counts because otherwise
teeth could not be translated over distances much larger than the periodontal
space. Moreover, the instantaneous displacements disappear as soon as the ortho-
dontic appliance is removed. Hence, after some time, the total displacements of
a tooth consist of a (hopefully) permanent part due to bone remodelling and a,
much smaller, temporary part that disappears after ending the treatment.
The instantaneous (temporary) displacements of the tooth result in non-
physiological strain and stress levels within the bone determining the ensuing

*) The term loads denotes both forces and/or couples. Similarly the term
displacements denotes both translations and/or rotations.
**) Processes leading to unwanted reactions such as root resorption are not
considered here.
752

bone remodelling processes. In orthodontics it is generally assumed that the

permanent displacement is of the same type as the instantaneous one. (For exam-
ple: a pure instantaneous translation results also in a permanent translation.)
The question whether matters are always as simple as that is left aside in this
paper; here we take the same point of view. The question whether the COR exists
then boils down to the question if a point exists such that a single force ac-
ting through this point results in a pure instantaneous translation. This quest-
ion will be studied using the system as shown in figure ,. As the tooth and the
socket consist of hard tissues compared to the periodontal ligament it will be
assumed that tooth and sockets are rigid bodies~ Moreover, it will be assumed
that the connective tissue behaves as a linear elastic material.

Figure 1. Left: Section of

tooth
(hard j
crownffj

tissue root-
I

:
jaw with two incisors. Gin-
giva not shown.
Right: Coordinate system con-

'
nected to socket. Part of the
periodontal
ligament (soft
tissues not shown. Open ar-
tissue) rows indicate the loads (==t>
I the force and ~ the coup les)
I and closed arrows denote the
resulting displacements of
the tooth with resvect to the
socket
(bone: hard
socket (--. the tr~nslations
tissue) and ~ the rotations).

3. THEORETICAL ANALYSIS
Consider the system consisting of tooth, periodontal ligament and socket as
shown schematically in Fig. ,. Also shown in this figure is a coordinate system
with x, y and z axes and origin O. The system is fixed with respect to the
socket. As stated in the previous section we assume that tooth and socket are
rigid bodies and that the periodontal ligament is a linear elastic material.
Now, let the tooth be loaded with a force F in point r and a couple m. Both F
and!!!. are vectors with canponents FX,Fy,F Z- and mx,m;'m with respect to the
coordinate system. The resulting displacement of the tJ6th consists of three
translations u x uy u z along the coordinate axes and of three rotations Ifix.IfJy.lfiz
around the coordinate axes. COllecting the translations and rotations in the
vectors ~ and.!!!. respectively and assuming that lfix ' lfiy and lfiz are sufficiently
small it holds that the 6 loads and the 6 displacements are connected by 6
equations. In matrix notation they read as shown in (,). The first 3 equations
express equilibrium between applied and reaction forces and the last 3 express
equilibrium between applied and reaction moments.

(1 )

Here the symbol denotes the usual vector product. The entries in the 3x3
matrices S", S'2 and S22 are the various stiffness coefficients of the
system. The superscript T denotes matrix transposition. Note that S" and
S?2 are symmetric and positive. The' stiffness coefficients depend on the mate-
r1al properties and on the shape of tooth root and socket.
The vector u can be elimimated fran the equations to obtain the following
vector equation.
 753

(2) (D(~)-PA)~ - P8~ + ~ = QJ!..

The matrices Q, P z ' Pa and 0 are defined as:
T -1 -T
PA 1:;(8 12 8 11 - 8 11 8 12 ) (anti-symmetric) ;
-rr ry~
D'' 'TO
r z o -rx P8 1:;(81'2 8 "11 + 8"118 12) (symmetric) ;
8-18
-ry rx 0 Q 8 22 - 8 T 12 11 12
(rx,ry, and r z are the components of r).
Note that Q is a regular matrix because in the nonloaded case, that is F. <= 0 and
m = 0, the tooth does not displace, i.e. u = 0 and Cfi = O. The question whether
the COR exists can now be phrazed as follows. --
Let the tooth be loaded with forces only; that is let in eqs. (2) m = O. If
the COR exists there should also exist some point Ee such that for
arbitrary force~, acting in Ee' the equations

have the solution if! = Q

After analysing eqs. (3) it is found that the following theorem holds:
THEOREM 1. Necessary and sufficient conditions for the existence of the center
of resistance (COR) in the 3-D case, are that the 3x3 matrix P~
vanishes. In that case the position of the COR is found by solv~ng
the equations D(~c) = PA.
The proof of the theorem is straightforward because of the regularity of the
matrix Q.
It follows that apart from some fairly obvious cases, such as axisymmetric
systems, there is no reason to assume in advance that the matrix P s always
vanishes. Hence, the existence of COR cannot be taken for granted.
The analysis given above can be repeated for the 2-D case.
Again taking a coordinate system fixed with respect to the socket we have now
two force components Fx,F z , collected in the vector F. but only one couple m.
This loading system produces two translations ux,u z , collected in the vector il,
and one rotation l{J If F acts at the point r (components: r , r ) the basic
equations read x z
(4a) F,
(4b) det(y,F) + m.

Note that the equation (4a) expressing equilibrium between applied and reac-
tion forces is a vector equation whereas eq. (4b), expressing equilibrium be-
tween ap~li~ an~ reaction moment is a scalar equation. Consequently, S11 is a
2x2 matr~x, s1~ ~s a vector and s22 is a scalar. These parameters are again
the various st~ffness coefficients of the system. Also S11 > 0 and s22 > 0
Again eliminating u from the eqs (4a) and (4b) one obtains the scalar equation
(5) (d(r) - p)Tp + m = qCfi
in which
d(Y)

_T
[-::J
-T -1
(d: vector; r ,r
x z components of r);

p s12 811 (p: vector);

( _T -1_ ) scalar) .
q s22 -s12 8 11 s 12 (q:
754

As in the previous sUbsection the question whether the COR exists can be
phrazed as follows:
Let the tooth be loaded with forces only, that is: m = 0 in eq. (5). If the
COR exists there should also exist some point re such that for arbitrary
forces acting in re the equation
- _ _ T-
(6) (d(r e ) - p) F = qql ,
has the solution qI = 0
Because q > 0 it follows readily from (6) that, compared to the 3-D case, now a
quite different theorem holds.
THEOREM II. The center of resistance in the 2-D case always exists. Its position
r c is determined by equation d (r ) = p .
SOlving for 1:e yields: e

re = [~ -~] p
Summarizing the results obtained so far the conclusion is that in "reality",
which is always the 3'-D case, the COR mayor may not exist in each particular
case. Whether the COR exists can only be determined experimentally.

4. EXPERIMENTAL AND NUMERICAL PART

Using stereophotogrammetric methods the shape of root and socket of the cen-
tral maxillary incisor of two males was determined in vivo (de Boer et al.,
1983). With the aid of these data, assuming that tooth and socket are rigid ma-
terials compared to the periodontal ligament the stiffness matrices S11' S12
and S22 where determined numerically using finite element methods. Linear ele-
ments were used with nodes lying on the root and on the socket surface (Fig. 2).
The determined root and socket shapes and element mesh are shown in Figure 3.
To study the influence of material properties it was assumed that the perio"
dontal ligament is a ,linear elastic homogeneous and isotropic material charac-
terized by two elastic constants, say Lame's constants A andp.

Figure 2. Details, element

mesh.
Above: Typiaal loaation of:
briak-shaped maaro element.:
Below: Deaomposition of
maaro element into 5 tetra-
hedral elements. Equal num-
bered aorners aoinaide in
maaro element.
 755

Figure 3. Experimentally
determined shapes of socket.
periodontal ligament and the root
of the central maxillary incisor
of two males. To enhance the
visibility of the root. is part of root surface _~mttS:
the socket surface and the
periodontal ligament not shown.
Also visible is the mesh of the
'111. socket surface

macro elements. Each brick consists

of 5 tetrahedra. Total number of
elements: 6480.

Provided the rotations are sufficiently small, it can then be shown that the
stiffness matrices Sij take the form
(7) Sij = AAij + ,uB ij , (i,j = 1,2)
The 3x3 matrices Aij and Bij are independent of the material properties. The
finite element analysis was carried out in such a way as to obtain numerical
values for the entries of the matrices Aij and Bij
As shown in section 3 the COR exists if Ps = O. With the aid of the (regu-
lar) matrix S11 it is straightforward to show that the vanishing of Ps is
equivalent to the anti symmetry of S12S11' Consequently the COR also exists
if the matrix S12S11 is antisymmetric. From eq. (7) it is found that
S12S 11 = A2(A12A11 +
,
(x) (B 12A11 + A12Bll ) + (X)2B12Bll)
,
H
Because A is a scalar the matrix H has the same symmetry properties as
S12S11' As
e
= (l-2v)/(2v)
A
(v: Poisson's ratio),
it is found that H = H(v).
Using the obtained matrices Ai' and B" next H(v) was for each tooth de-
termined for various values of v wIthin tt~ range 0.1~v ~ 0.4. Although H was
not exactly antisymmetric it was found however that the antisymmetric part
strongly dominates. Taking the sup norm as matrix norm it turned out that in all
cases considered, that is for both teeth and for O. 1 ~ v ~ 0.4, the symmetric part
of H was less than 1% compared to H itself. Specifically
1Il.:!(HT+H)iI sup }
(8)
{ < 0.01
IIHII sup
In view of experimental and numerical inaccuracy it follows that the two inci-
sors, for which the result (8) applies, indeed do have a COR.

5. DISCUSSION AND CONCLUSIONS

The main result from section 3 that the COR always exists in the 2-D case
whereas its existence in the 3-D is not guaranteed apriori has methodological
and clinical implications.
To start with the methodoiogical aspects it should first be stressed that the
phenomenon of qualitatively different properties of systems in the 3-D case as
compared to the 2-D case is not unusual in physics; the classical example being
756

the wave equation. Once such differences are known to exist, the general theory
on the behavior of a system should be derived from the 3-D case because this is
generally much closer to reality. Reducing reality by considering the 2-D case
instead of the 3-D one is only permissible if the differences are sufficiently
small for the problem at hand. If such a reduction is carried out for ease of
exposition in textbooks or research papers, one should take care to consider
only those aspects of the theory that have the same methodological status in the
exposition as they have in reality.
In clinical orthodontics a moment-to-force ratio theory (Burstone, 1962,
19801 Christiansen et al., 1969), based on the existence of the COR is used.
This theory furnishes a formula through which moment-to-force ratio, root length
and distance of the COR to the center of rotation are quantitatively related.
In view of the results obtained in sections 3 and 4 it must be concluded that,
apart from incisors, Christiansen'S and Burstone's formula probably cannot be
used in the general 3-D situation because the COR might not always exist. To
conclude it should, however, be pointed out that the analysis as given in
section 3 can be utilised to develop a general moment-to-force-ratio theory. To
do so a separate study is necessary. That will be done in future.

LITERATURE
1. Smith RJ, Burstone CJ (1984): Mechanics of tooth movement. Am J Orthod~:
294-307.
2. Nikolai RJ (1985): Bioengineering analysis of orthodontic mechanics. Lea &
Febiger, Philadelphia.
3. Burstone CJ ( 1985) : Application of bioengineering to clinical orthodontics.
In: Graber TM, Swain BF (eds.): Orthodontics. Current principles and
techniques. The VC Mosby Company, St. Louis: 193-227.
4. Burstone CJ, Pryputniewics RJ (1980): Holographic determination of centers
of rotation produced by orthodontic forces. Am J Orthod 77: 396-409.
5. Nikolai RJ (1 975): On optimum orthodontic force theory as applied to canine
retraction. Am J Orthodon 68: 290-302.
6. Bowden DEJ (1978): Theoretical consideration of headgear therapy: A
literature review. Br .J Orthod 5: 145-152.
7. Hocevar RA (1981): Understanding, planning and managing tooth movement:
Orthodontic force system theory. Am J Orthod 80: 457-477.
8. Roberts WW, Chacker FM, Burstone CJ (1982): A-Segmental approach to
mandibular molar uprighting. Am J Orthod 81: 177-184.
9. Nikolai RJ (1982): Analytical mechanics andlanalysis of orthodontic tooth
movements. Am J Orthod 82: 164-166.
10. Yoshikowa DK (1981): Biomechanical principles of tooth movement. Dent Clin
North Am 25: 19-26.
11. Reitan K (1985): Biomechanical principles and reactions. In: Graber TM,
Swain BF (eds.) Orthodonics. Current principles and techniques. The VC Mosby
Company. St Louis: 101-192.
12. Christiansen RL, Burstone CJ (1969): Centers of rotation within the
periodontal space. Am J Orthod 55: 353-369.
13. Burstone CJ (1962): The biomechanics of tooth movement. In: Kraus BS, Riedel
BA (eds.): Vistas in orthodontics. Lea & Febiger, Philadelphia 1962: 197-
213.
14. Boer A de, Varst PGT van der, Bosman F (1983): Determination of the three
dimensional shape of the periodontal ligament. J Int Ass Dento-maxillo-
Facial Radiology, suppl. 5: 49.
 757

CYCLIC GROUPS OF NON-LINEAR BIOLOGICAL AUTOMORPHISMS AND

THEORY OF CYCLOGENESIS IN BIOMECHANICS
S.V. Petukhov
Mechanical Engineering Research Institute, USSR Academy of
Sciences, Moscow, USSR

Many problems in mechanics of living matter, including the

problems of stato-kinetic regulation and structure of ele -
ments of skeletal and muscular apparatus, are connected with
geometrically ordered multiblock biostructures like the lum-
bar spine or the net-like structures of bone tissue. The pre-
sent report is devoted to the exposition of general biologi -
cal algorithms in the formation of such structures. Such
knowledge provides a useful basis for studing the problems of
normal and pathological structures and the functioning of the
sensor, motor, skeletal and muscular apparatus of a man and
animals. In addition the analysiS of such algorithms is esse-
ntial for the determination of the fundamental morphogenetic
properties of living matter and the development of the theor~-
es of biological evolution. '
The ordered multiblock structures can be widely observed in
living organisms: such examples are metameric bodies of myri-
apods and annelidae, fins of fishes, flowers, molluscs shells
etc., many of them are constructed according to known rules
of symmetry and are represented by self-similar cyclomery,
i.e. the multiblock bodies in which the sequence of blocks
X1,X2,.",Xk,'" is formed due to recurrent equation:
Xk+1 = T(Xk) ,
where T is similarity transformation ( 1,2,) ). In other
words, a classical biological cyclomery (if one takes into
account that it consists of a limited number of cyclomers) is
described in terms of an orbit of the cyclomer (or a motive
unity) with respect to the cyclic group with regenerating si-
milarity transformation T. This cyclomery of similarity is
organized into relatively simple cylindrical, conic, spiral
and other forms.
The author has posed and investigated the problem of a pos-
sible biological value of cyclomeries using regenerating
transformations T of a more general type, for example, these
can belong to groups of non-linear transformations which have
the similarity group as their subgroup. Interest is primarily
focussed on the groups of transformations emanating from the
Erlangen program of F.Klein (4) the foundation of non-eucli-
dean geometries, in the same manner as the similarity group
constitutes the foundation of euclidean geometry (4). As a
result of this mathematical investigation, the existence and
growth of idealized multiblock biobodies has been described
in terms of cyclomeries with regenerating transformations
758

from mobius, affine and projective transformation groups.

These transformation groups contain the similarity group and
~the bases of the conformal, affine and projective geomet~
ries. The propositions of these abstract non-euclidean geome-
tries turn out to be realized in living bodies. This supports
the V.r.Veraadsky hypothesis (5) on the possible fundamental
value of Don-euclidean geometry for living matter.
The non-euclidean cyclomeries have configurations which we-
re never derived before in the science of biosymmetries: ly-
re-like, sickle-like, bud-like etc. Figures 1 - 8 represent
some examples of non-euclidean cyclomeries in biostructures.
Ion-euclidean cyclomeries and similarity cyclomeries are usu-
ally observed in similar multiblock organs: se~ented horns
of aniaals, fins of fishes, feathers of birds, blastulae
(uniform and non-uniform) etc.
Returning to the spine as an example of an iaportant biome-
chanics multiblock bio-object, one may follow a similar ap -
proach for human and animal vertebrae: the structure of ver-
tebral bodies of the spine is described as a cyclomery of si-
milarity (see lizards etc.) or non-similarity. In Fig.1, the
configuration of a human spine is reproduced in terms of the
cyclomery using the mobius regenerating transformation of the
loxodrome type: (16,87-i' 2,88). (Xk +iYk >+64,Q,6+i 56,8
+
.a-------------------------------------
Xk 1 + i'Yk 1
+ (0,04-i'0,01)'(Xk+i'Yk )+16,09-i'5,96
Let us keep in mind that by modifying this regenerating
transformation we can obtain models for morphogenetic anoma-
lies of the spine. This fact is in agreement with our other
data on the development in man, animals and plants of eucli-
dean and non-euclidean cyolomeries, not only for normal, but
also for pathological form-building. This corresponds to the
preposition (weJ.[ mown in plant morphology), that abno:rmali-
ty is a variant of the Dorm and that "possibly one cannot na-
me a single type of deviation which has not come to be a norm
for any taxons"(6,p.32). A similar approach is fruitful while
modelling the deformities of the skull by acromegalics.
Euclidean and non-euclidean cyclomeries are observed in the
structure and functioning of vestibular, acoustic and eyemo-
ving apparatae,the organisation of the nervous system, bone
tissue,vessels,muscles,the position of biologically active
points on the surface of a human body,series of colour pat-
ter.ns,time biorythas,the cyclomer constructions instinctively
builted by protozoae and social insects, the trajectories of
aniaals and human motion in normal and extraordinary conditi-
ons,the forming of space illusions, the aesthetics of propor-
tions and forms in architecture and art eto. The totality of
the obtained results in cyclomeries is a further development
of the conformal blocks prinCiple (7,p.213) and supports the
thesis (7,p.84) that many aspects of the biological configu-
rations including the peoularities of psychophysios are stru-
cturally allied with morphogenesis. These results also lead
us to the idea of the importance of special biochemical me -
cbanisms (conneoted with networks of metabolic reactions) for
realization of biological structurization processes.
 759

ty

9 7

6

0
8

~
5
%
4 ~
S ~
E 2 d
[J
I 0
X X
0 I 2
Fig.1 The vertebral column of the Fig.2 Segmentized horn
torso section of human spine (ac- of Pantholops hodgsoni
cording to (8 and its geometri- (according to A.Bannikov
cal model in the form of cyclomery et al.,1971) and its
with the mob ius regenerating configuration in the
transformation. form of the Mobius cy-
clomery.
y

~.. \ 20

10
x
o 20 40 60
Fig.4 Multiblock structure
of the fish-scale (9,table
87,N15) modelled by the cy-
clomery with the affine re-
generating transformation:
Xk+1 = Xk-O,Q99Yk +5,24
{
Yk+1 = Yk

Fig.) The chamber of protozoa (9,table 12) and its configur.a-

tion in the form of cyclomery wi~h the Mobius regenerating
transformation.
 A

Fig.5 The caudal fins of the fishes (according to L.Berg,1961

& P.Grasse,1960) with the euclidean sad non-euclidean cyclome-
ries. A - the fin of Acanturus sp. and the cyclomery with the
affine regenerating transtormation: Xk+1-Xk-O,9Yk; Yk +1-O,OJXk
+Yk~ B - the tin ot Lumpenus lampetraetormis and the cyclo -
mery with the projeotive regenerating transformation:
Xk +1-(O,8JXk+1,27Yk-2,26):(1-0,05Yk )}
Yk+1-(-O,14Xk+O.91Yk -J,26):(1-0,05Yk ) ,
C-the tin of Siniperoa ohua-tsi with the olassioal rotatory
oyolomery; D - the tin of Tungusiohtys acentrophoroides with
the projective cyclomery.
 761

@~
c@)~ci) Fig.7 The development of sa-
lamandr~sembryo (according
Fig.6 The development of phy- to O.Dodson,1960) is accom-
tomonadidae's colony (accor - panied by reconstruction of
ding to O.Ivanova-Casas,1975) sequence of somites from the
at the stages A-F as example classical! similarity oyclo-
of cyclomeries'participation mery of spiral type to the
in kinematics of biostructu- mob ius cyolomery of loxodro-
res. me type.

.' . .
(iD " \ \\~
~
I

Fig.8 The examples of the mobius cyclomeries' configu-,

rations in physiologioally normal poses of the elephant s
trunk, body of the defending itself cobra, tail of the
frightened rat.
762

Let us add here that the phyllotaxical laws of morphogene-

sis and the biological role of the Fibonaoci numbers can be
explained on the basis of special connections of these num -
bers with the oyolio groups of non-euolidean transformations.
The use of cyclio groups of automorphisms in multiblock
biostruotures is functionally admissible in conneotion with
the obvious reduotion of information necessary for genetio
ooding and the morphogenetio realization of suoh structures.
The biologioal cyclomeries can b~ completely discrete but so-
metimes praotioally continuous. This relates to the reoently
discussed contradiction between the discrete and/or continu-
ous characteristics of genetic coding and morphogenesis. A
living organism can be considered as a colony of cyclomeries.
The evolution of the organism can be described in terms of
the evolution of the corresponding colony. The symmetry rela-
tions between the neighboring blocks in the multiblock biobo-
dies of the cyclomeric type can sometimes fluctuate depen -
ding,for example,on the feeding conditions while constructing
a new block. For modelling of such structures the formalismS
of Markov chains and multisteps processes are useful. These
series of questions represents the essence of the author's
theory of biologioal cyclogenesis. It conneotes with problem
of geometrization in biology.
_ The cyclomeric principles of biological organization are
also related to the phenomena of volume growth of living bo-
dies and the kinematics of morphogenetic,growth and motor mo-
tions, many oases of which can be described in terms of a su-
ccessive transposition from one cyclomery to another. This
transpOSition may sometimes propogate along the multiblock
body in the form of one or several variably delayed dispersed
body waves. Exaaples of such waves are the excurvation in the
development of the colony of volvox, the locomotor kinematics
of myriapods,annelidae,snakes etc. The proposed concept of
th.waves of changing cyclomeries can also be applied to mo-
delling of the kinematics of heart pulsation and other wave-
-like motions of organs and tissues in man.
REFERElfCES
1. Beklemishev VB: Bases of comparative anatomy of inverte-
brates. Mosoow: Iauka, 1964, 699 p. (in Russian).
2. leyl H: Symmetry. Bew Jersey: Princeton, 1952, 191 p.
3. Shubnikov AV: Similarity symmetry.-Cristallography, 1960,
v.5, pp.489-496 (in R~ssian2.
4. Klein F: Vorlesungen uber hohere Geometrie. Berlin: Ver -
lag, 1926, 399 S.
5. Vernadsky VI: Chemical structure of the biosphere and its
medium. Moscow: Nauka, 1965, 205 p. (in Russian).
6. Meyen SV: Plant morphology in its nomogenetical aspects.-
Botanic.rev., 1973, v.39, B 3, pp.205-260.
7. Petukhov SV: Biomechanics, bionics and symmetry. Moscow:
Bauka, 1981, 239 p. (in Russian).
8. Haeckel E:Naturlicheschop!ungs Geschichte. Berlin: 1898.
9. Haeckel E: ~unstforaen der Natur. Leipzig-lien: 1904.
 763

AUTHORS INDEX
Ahrens,U. 359 Dabrowska,A. 673
Albersdorfer,H. 435 Dal Monte,A. 633,645
Altmann,S. 273 Da 11 e k , M. 733
Angevin,S. 261 Dalstra,M. 325
Arai,M. 291 Da rmana, R. 261
Arlt,R. 417 De Boer, A. 751
Arru,R. 725 De Bona, F. 725
Arts,T. 73 De Lange,A. 121,255
Ascenzi,A. 231 De Sch ryver, F. 309
Ascherl,R. 435 Do,M.C. 607
Assente,R. 523 Doyle,C. 213
Attinger,D. 563 Drogendijk,A.C. 193
Ayukawa,K. 713 Drukker,J. 187
Bajd,T. 551 Dvorak,J. 517
Barneveld,A. 159 Dorheide,J. 135
Bartosiewicz,G. 673 Durselen,L. 173,587
Baumann,W. 51,595 Eliasz,J. 673
Beaupre,G.S. 411 Epstein,M. 601
Bedzinski,R. 497 Eulenberger,J. 353,423
Behiri,J.C. 225 Ferrigno,G. 147
Benko,H. 551 Fischer,A. 417
Benzing,U. 373 Friedebold,G. 3
Bergmann,G. 317 Friedman,R. 353
Bergmann,M. 365 Frigo,C. 581
Berme,N. 251 Fyhrie,D.P. 219
Blankevoort,L. 285 Ga jew ski, J. 673
Blanquaert,D. 261 Gallozzi,C. 627,633
Blumel,G. 435 Gasser, B. 341,429
Boenick,J.P. 575 Gautreau, M. 491
Bona,G. 707 Geng,J. 201
Bonfield,W. 213,225 Georgiev,S. 537
Boobis,L.H. 613 Giebel,G. 141
Boosfeld,Chr. 575 Gomez,M.A. 167
Boyde,A. 231 Goodship,A.E. 441,447
Brand,M. 679 Graichen,F. 317
Brehme,H. 435 Grana,W.A. 459
Brenier,Y. 607 Greiff,H. 679
Burton,K. 485 Greuell,A.E. 667
Buttner-Janz,K. 511 Grootenboer,H. 325
Calderale,P.M. 707,725,745 Gu d a vall i , M. R. 459,479
Carando,S. 231 Halt,J. 379
Carter,D.R. 219,411 Hauser,W. 659
Caselli,G. 627 Hayes,W.C. 347
Castaldini,A. 739 Hedberg,R. 115
Cavallini,A. 739 Heiliger,R. 699
Challis,J.H. 613 Herndon,W.A. 479
Chao,E.Y.S. 33 Herrboldt,U. 453
Charalambides,B. 225 Hirai,T. 207
Cheal,E.J. 347 Hoffer,J.A. 601
Claes,L. 173,465,505,587 Holzwarth,J. 465
Coombs,R.R.H. 213 Hopf,Th. 399
Cordey,J. 237,387,393,531 Hoschek,J. 379
Csongradi,J. 411 Hoyer,H.-E. 135,141
764

Huijing,P.A. 667 Merkens,H.W. 153

Huiskes,R. Mielkau,U. 685
109,121,255,285,325,333 Mikulcik,E.C. 601
Improta,S. 231 Mittermayer,Ch. 699
Inaba,H. 291 Mor i , A. 713
Inoue,M. 167 Mo r uc c i , J . P. 261
Inoue,N. 207 Nagata,R. 243
Iskra,L. 673 Neugebauer,W. 417
Iwata, K. 243 Nicol,A.C. 569
Izuchi,N. 243 Nicol,K. 543,679,685
Jacob,H.A.C. 537 Nieminen,H. 491
Janssen,J.D. 187 Oberste,W. 679,685
Jaworek,K. 147 Ogawa,R. 243
Jendrzynski,H. 317 Oka,M. 279
Jonsson,B. 115 Oomens,C.W.J. 187,193
Jungbluth,K.H. 733 Opila,K.A. 569
Jager,M. 473 Orr,T.E. 219,411
Kalli,S. 491 Pace,A. 627,633
Kang,Z.H. 85 Parfianowicz,L. 653
Katayama, T. 207 Paul,J.P. 569
Kauer,J.M.G. 121,255 Pekalski, R. 639
Kedzior,K. 405 Perkins,W.J. 99
Kelly,D.J. 441,447 Perren, S. M.
Kenwright,J. 441,447 237,341,387,393,423,517,531
Kerwin,D.G. 613,619 Peters,G.W.M. 179
Kiefer,H. 173,465,587 Peters,M.C.R.B. 751
Kleer,G. 373 Petit,J.S. 261
Klein,P. 309 Petukhov,S.V. 757
Kojima,A. 243 Pezzol i, M. 745
Komor,A. 645,653,691 Pinto,J.G. ~19
Kotoura,Y. 279 Pirini,R. 739
Kralj,A. 551 Portigliatti Barbos,M. 231
Kramer,W. 417 Rahn,B.A. 237,531
Kramers,P.C. 237 Ramsbottom,R. 613
Kranz,C. 359 Rapperport,D.J. 219
Krapf,W. 399 Rehder,U. 297,303
Kuban, W. 653 Reithmeier,E. 365
Kulot,M. 659 Reneman,R.S. 73
Kuorinka,I. 115,491 Rether,J.R. 141
Kutubuddin,M. 459 Richardson,J.R. 447
Lambertz,H. 699 Rigby,H.S. 441
Langer,H.E. 273 Roddeman,D.G. 187
Lanshammar,H. 129 Rohlmann,A. 317,347
Leonardi,L.M. 627,633,645 Rossetto,M. 745
Leskinen,T. 115,491 Rothuizen,P. 109
Licari,V. 581 Rozendal,R.H. 557
Liebscher,F. 685 Rudel,D. 551
Luethi,S. 563 Sakakida,K. 207
Luttmann,A. 473 Santambrogio,G.C. 147,523
Uier,F. 267,453 Sauren,A.A.H.J. 179
LOhrs,W. 273 Savelberg,H.H.C.M. 557
Maes,H. 595 Savvidis,E. 267,453
Matsumoto,T. 243 Schaff,P. 659
McGurk-Burleson,E. 167 Schamhardt,H.C. 153
Me e nen, N. M. 733 Schattner,R. 659
 765

Schellnack,K. 511 Wyder,O. 353,429,517

Schneider, E. Xu, X. X. 531
353,423,429,517 Yamashita,F. 207
Schurman,O.J. 219 Zabel,A. 297
Selvik,G. 379 Zagrajek,T. 405
Sharp,O.J. 213 Zeidler,H. 273
Shorten,M.R. 619 Ziegler,W.J. 531
Siebels,W. 435 Zollinger,H. 537
Siegele,O. 373 Zuber, K. 423
Slooff,T.J. 325,333
Soltesz,U. 373
Son i , A. H. 459,479
Spaepen,A.J. 595
Sp rig gin s , A. J . 447
Steinemann,S. 387
Steiner,W. 353
Stuessi,E. 563
Sullivan,J.A. 479
Sun,J.J. 201
Takabatake,S. 713
Tanner, K. E. 213
Tepic,S. 341
Theysohn,H. 679
Trzaskoma,Z. 673
Tulin,P. 405
Uklanski,R. 691
Ulrich,Ch. 505
Vallana,F. 707,725
Van Ouyl,W.A. 193
Van Erning,L. 109
Van Mameren,H. 179
Van Weeren,P.R. 159
Van der Varst,P.G. 751
Van der Venne,R. 325
Van der Zon,A. T.M. 193
Vigano,R. 523
Vi i di k, A. 59
Wajon,M.H. 667
Walker,f'.S. 19
Wall,A. 497
Walter,A. 365
Watkins,P.E. 441
Weber,U. 379
Weh,L. 733
Weinmann,K. 365
Weizsacker,H.W. 719
Werner,O. 267
Whalen, R. 219
Wit,W. 673
W6rsd6rfer, o. 505
Wo i tt i e z , R. O. 667
Wolf,J. 691
Wo Iff, R. 3
Woltring,H.J. 121
Woo,S.L-Y. 167