Biomechanics

Motion, Flow, Stress, and Growth

Other titles by the same author:

Biomechanics: Mechanical Properties of Living Tissues (1981)

Biodynamics: Circulation (1984)

Springer Science+Business Media, LLC

 Y.C. Fung

Biomechanics
Motion, Flow, Stress,
and Growth

With 254 Illustrations

, Springer
Y.c. Fung
Department of Applied Mechanics
and Engineering SciencelBioengineering
University of California, San Diego
La Jolla, CA 92093
USA

Library of Congress Cataloging-in-Publication Data

Fung, Y.c. (Yuan-cheng)
Biomechanics : motion, flow, stress, and growth / Y.C. Fung.
p. cm.
Includes bibliographical references.
ISBN 978-1-4757-5913-6 ISBN 978-1-4419-6856-2 (eBook)
DOI 10.1007/978-1-4419-6856-2
1. Biomechanics. 2. Human mechanics. 3. Hemodynamics.
4. Biophysics. I. Title.
QP303.F86 1990
591.19'I-dc20 89-22017

Printed on acid-free paper.

© 1990 Springer Science+Business Media New York

Originally published by Springer-Verlag New York, Inc. in 1990
Softcover reprint ofthe hardcover Ist edition 1990
AII rights reserved. This work may not be translated or copied in whole or in part without the writ-
ten permission of the publisher Springer Science+Business Media, LLC.
except for brief excerpts in connection with reviews or scholarly analysis.
Use in connection with any form of information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed is
forbidden.
The use of general descriptive names, trade names, trademarks, etc., in this publication, even if
the former are not especially identified, is not to be taken as a sign that such names, as understood
by the Trade Marks and Merchandise Marks Act, may accordingly be used freely by anyone.

9876543

ISBN 978-1-4757-5913-6 SPIN 10833887

 Dedicated
to the
Memory of
My Father
~1Ii *!B't
Fung
Chung-
Kwang
(1896-
1952),
Educator,
Painter.

He painted
this scroll
in 1939.
He loved
monkeys
because
"they
are like
children."
Preface

Biomechanics aims to explain the mechanics oflife and living. From molecules
to organisms, everything must obey the laws of mechanics. Clarification of
mechanics clarifies many things. Biomechanics helps us to appreciate life. It
sensitizes us to observe nature. It is a tool for design and invention of devices
to improve the quality of life. It is a useful tool, a simple tool, a valuable tool,
an unavoidable tool. It is a necessary part of biology and engineering.
The method of biomechanics is the method of engineering, which consists
of observation, experimentation, theorization, validation, and application. To
understand any object, we must know its geometry and materials of construc-
tion, the mechanical properties of the materials involved, the governing
natural laws, the mathematical formulation of specific problems and their
solutions, and the results of validation. Once understood, one goes on to
develop applications. In my plan to present an outline of biomechanics, I
followed the engineering approach and used three volumes. In the first volume,
Biomechanics: Mechanical Properties of Living Tissues, the geometrical struc-
ture and the rheological properties of various materials, tissues, and organs
are presented. In the second volume, Biodynamics: Circulation, the physiology
of blood circulation is analyzed by the engineering method. In the third
volume, the present one, the methods of problem formulation, solution, and
validation are further illustrated by studying the motion of man and animals,
the internal and external fluid flow, the stress distribution in the bodies,
strength of tissues and organs, and the relationship between stress and growth.
Thus the three volumes form a unit, although each retains a degree of
independence.
The plan of this book is as follows. In Chapter 1, Newton's laws of motion
and the basic equations of solid and fluid mechanics are presented and
illustrated by biological examples. In Chapter 2, the motion of a system of

vii
viii Preface

connected elastic or rigid bodies is considered. Generalized coordinates, gen-

eralized forces, and Lagrange's equations are used. This method is especially
useful in the analysis of the musculoskeletal system because it offers a sys-
tematic way of dealing with the many muscles involved in locomotion.
Then we consider external flow in Chapters 3 and 4, and internal flow in
Chapters 5 through 9. External flow is the flow around bodies moving in wind
and water, in locomotion, flying, and swimming. Internal flow is the flow of
blood in blood vessels, gas in airways, water and other body fluid in interstitial
space, urine in kidney, ureter and urethra. Although blood flow is treated in
Biodynamics (Fung, 1984), a sketch of its salient features is included in Chap-
ters 5 and 6 to make this volume somewhat self-contained, while some new
topics are added. Chapter 7 treats the flow of gas in the lung. Chapter 8 derives
the basic equations of fluid movement in the interstitial space between blood
vessels and cells, as well as the transport mechanisms in cell membranes.
Multiphasic mixtures are also discussed. Chapter 9 presents the analysis of
fluid movement from capillaries to the interstitial space and lymphatics, the
indicator dilution method of measurement, and transport by peristalsis.
The rest of the book is devoted to the biological effects of stress and strain.
In Chapter 10 we set down the basic concepts of stress and strain in bodies
subjected to large deformation. A simple introduction to the finite deforma-
tion theory is offered. I hope to acquaint the reader with the physical meaning
of the most important formulas of the theory, and with the distinctions
between Green's and Almansi's strains, and Cauchy's, Kirchhoff's, and
Lagrange's stresses. With this easy introduction, I trust that the reader will be
confident to use these quantities and formulas when dealing with soft tissues.
In Chapter 11, stress and strain in organs is studied. In Chapter 12, the
strength of tissues and organs is considered. Trauma due to blunt impacts in
airplane or automobile crashes is discussed.
In the last chapter, we consider the phenomena of growth and resorption
of cells, tissues, and organs in relation to stress: phenomena such as hyper-
trophy due to high blood pressure, resorption due to too high or too Iowa
stress, and normal growth with exercise in proper range. These phenomena
are controlled by biological, chemical, and physical stimuli, including physical
stress and strain. Furthermore, growth or resorption of tissues changes the
zero-stress configurations of organs. Conversely, changes of zero-stress con-
figurations are convenient indications of tissue remodeling.
The rapid progress of bioengineering makes a serious attempt at tissue
engineering possible. A person's own cells may be grown rapidly in a polymer
scaffold for use as tissue substitute in surgical repair of serious disease or
trauma. Close attention to the study of growth in the future can be predicted.
Completeness is not claimed. Our subject is young, and progress is rapid.
Seeking greater permanency, I have limited the scope of this book to the more
fundamental aspects of biomechanics. For other aspects and handbook mate-
rial, the reader must turn to the library. Liberal but selected lists of references
are given in each chapter. Problems are proposed for solution, some of them
Preface IX

with references to articles in the literature. These problems will serve to

broaden the scope of our discussion or point to further applications of the
principles, or new directions of research.
When I began writing these books, I made a to-year plan. I did not
anticipate that it would take much longer. The first draft of the manuscript
of this volume was completed in 1980. Successive revisions, simplifications,
and amplifications were made to reflect the advancing frontiers of our subject,
and to gain a greater clarity.
There are many people whose help I must acknowledge. First, on classical
subjects I followed classical books, especially Sir James Lighthill's book on
Biofluiddynamics, and my friend Chia Shun Yih's book on Fluid Mechanics.
Next, on biomechanics research I am indebted to many friends and colleagues
who have influenced me greatly: Sidney Sobin, Michael Yen, Benjamin
Zweifach, Geert Schmid-Schonbein, Hans Krumhaar, Dick Skalak, Savio
Woo, Ted Yao-Tsu Wu, Maw Chang Lee, Zulai Tao, Charles c.J. Chuong,
Evan Evans, Frank Yin, Paul Zupkas, and others. To Sid Sobin and Mike
Yen, I must record my pleasure of having collaborated with them for more
than 20 years. My students Jack Debes, Julius Guccione, Ghassan Kassab,
Shu Qian Liu, Bruce Bedford, Carmela Rider, and Nina Tang read the final
manuscript and gave me errata and other valuable suggestions. I also wish to
express my thanks to the many authors and publishers who permitted me to
quote from their publications and reproduce their figures and data in this
book, especially to Professors Arnold Kuethe on flying, and James Hay on
sports techniques. To my students at Caltech, UCSD, and several universities
in China, I want to say thank you. You shaped this book by your questions,
your enthusiasm, and your dismay. I am grateful to Perne Whaley for typing
the manuscript.

La Jolla, California Yuan-Cheng Fung

Caligraphy by
Il~*fifit
Cheng Sih. alias
Banchiao (1693-1765)
"On Literary Style."
Inscription says:
"Simplified as trees
in late Fall.
New as flowers
before Spring."
Contents

Preface Vll

Chapter 1
Motion
1.1 Introduction 1
1.2 Equilibrium 2
1.3 Dynamics 4
1.4 Modeling 9
1.5 Sports Techniques 12
1.6 Prosthesis 17
1.7 Continuum Approach 18
References 28

Chapter 2
Segmental Movement and Vibrations 29
2.1 Introduction 29
2.2 Examples of Simple Vibration System~ 30
2.3 Strain Energy and the Properties of the Influence Coefficients 32
2.4 Generalized Coordinates 35
2.5 Lagrange's Equations 38
2.6 Normal Modes of Vibration 44
2.7 Decoupling of Equations of Motion 47
2.8 Muscle Forces 49
2.9 Segmental Movement and Vibrations 52
2.10 Systems with Damping and Fluid Dynamic Loads 52
2.11 Sufficient Conditions for Decoupling Equations of System with
Damping 55
References 59

XI
XII Contents

Chapter 3
External Flow: Fluid Dynamic Forces Acting on Moving Bodies 62
3.1 Introduction 62
3.2 Flow Around an Airfoil 62
3.3 Flow Around Bluff Bodies 68
3.4 Steady-State Aeroelastic Problems 71
3.5 Transient Fluid Dynamic Forces Due to Unsteady Motion 76
3.6 Flutter 80
3.7 Kutta-Joukowski Theorem 83
3.8 The Creation of Circulation Around a Wing 87
3.9 Circulation and Vorticity in the Wake 88
3.10 Vortex System Associated with a Finite Wing in Nonstationary Motion 91
3.11 Thin Wing in Steady Flow 96
3.12 Lift Distribution on a Finite Wing 99
3.13 Drag 100
References 104

Chapter 4
Flying and Swimming 106
4.1 Introduction 106
4.2 The Conquest of the Air 106
4.3 Comparing Birds and Insects with Aircraft 109
4.4 Forward Flight of Birds and Insects 113
4.5 Hovering and Other Modes of Motion 121
4.6 Aquatic Animal Propulsion 124
4.7 Stokeslet and Dipole in a Viscous Fluid 128
4.8 Motion of Sphere, Cylinder, and Flagella in Viscous Fluid 132
4.9 Resistive-Force Theory of Flagellar Propulsion 135
4.10 Theories ofFish Swimming 140
4.11 Energy Cost of Locomotion 143
4.12 Cell Movement 148
References 150

Chapter 5
Blood Flow in Heart, Lung, Arteries, and Veins 155
5.1 Introduction 155
5.2 The Geometry of the Circulation System 156
5.3 The Materials ofthe Circulation System 158
5.4 Field Equations and Boundary Conditions 159
5.5 Blood Flow in Heart and Through Heart Valves 161
5.6 Coupling of Left Ventricle to Aorta and Right Ventricle to
Pulmonary Artery 162
5.7 Pulsatile Flow in Arteries 165
5.8 Progressive Waves Superposed on a Steady Flow 167
5.9 Reflection and Transmission of Waves at Junctions 168
5.10 Velocity Profile of a Steady Flow in a Tube 170
5.11 Steady Laminar Flow in an Elastic Tube 172
5.12 Velocity Profile of Pulsatile Flow 176
5.13 The Reynolds Number, Stokes Number, and Womersley Number 179
5.14 Equation of Balance of Energy and Work 181
Contents xiii

5.15 Systemic Blood Pressure 183

5.16 The Veins and Their Collapsibility 185
5.17 Flow in Collapsible Tubes 190
5.18 Pulse Wave as Message Carrier for Noninvasive Diagnosis 192
References 194

Chapter 6
Micro- and Macrocirculation 196
6.1 Introduction 196
6.2 Anatomy of Microvascular Beds 197
6.3 Major Features of Microcirculation 199
6.4 The Rheological Properties of Blood 202
6.5 General Equations Describing Flow in Microvesse1s 203
6.6 An Example of Analysis: Pulmonary Blood Flow 203
6.7 Pulmonary Capillary Blood Flow 211
6.8 Waterfall Phenomenon in Zone 2 215
6.9 Open and Closed Capillary Sheets in Zone 2 216
6.10 Synthesis of Micro- and Macrocirculation in the Lung 220
6.11 Validation of the Flow Model 221
References 224

Chapter 7
Respiratory Gas Flow 226
7.1 Introduction 226
7.2 Gas Flow in the Airway 229
7.3 Interaction Between Convection and Diffusion 237
7.4 Exchange Between Alveolar Gas and Erythrocytes 244
7.5 Ventilation-Perfusion Ratio 249
7.6 Pulmonary Function Tests 253
7.7 Dynamics of the Ventilation System 258
7.8 High-Frequency Low-Tidal-Volume Ventilation 265
References 272

Chapter 8
Basic Transport Equations According to Thermodynamics,
Molecular Diffusion, Mechanisms in Membranes, and
Multiphasic Structure 275
8.1 Introduction 275
8.2 The Laws of Thermodynamics 277
8.3 The Gibbs and Gibbs-Duhem Equations 280
8.4 Chemical Potential 281
8.5 Entropy in a System with Heat and Mass Transfer 283
8.6 Diffusion, Filtration, and Fluid Movement in the Interstitial Space
from the Point of View of Thermodynamics 287
8.7 Diffusion from the Molecular Point of View 292
8.8 Transport Across Cell Membranes 294
8.9 Solid Immobile Matrix 300
References 306
XIV Contents

Chapter 9
Mass Transport in Capillaries, Tissues, Interstitial Space,
Lymphatics, Indicator Dilution Method, and Peristalsis 309
9.1 Introduction 309
9.2 Fluid Movement Across Capillary Blood Vessel Wall 309
9.3 Experimental Determination of the Permeability Characteristics of
Capillary Blood Vessel Wall 313
9.4 The Krogh Cylinder as a Model of Oxygen Diffusion from Capillary
Blood Vessel to Tissue 316
9.5 Fluid Movement in the Interstitial Space 320
9.6 Measurement of Interstitial Pressure 323
9.7 Pressure in an Incompressible Material 325
9.8 Lymph Flow 330
9.9 Measurement of Extravascular Space by Indicator Dilution Method 333
9.10 Tracer Motion in a Model of Pulmonary Microcirculation 337
9.11 Peristalsis 342
References 348

Chapter 10
Description of Internal Deformation and Forces 353
10.1 Introduction 353
10.2 Description of Internal Deformation 354
10.3 Use of Curvilinear Coordinates 360
10.4 Description ofInternal Forces 361
10.5 Equation of Motion in Lagrangian Description 369
10.6 Work and Strain Energy 371
10.7 Calculation of Stresses from the Strain Energy Function 374
10.8 Complementary Energy Function 376
10.9 Rotation and Strain 377
References 381

Chapter 11
Stress, Strain, and Stability of Organs 382
11.1 Introduction 382
11.2 The Zero-Stress State 384
11.3 Stress and Strain in Blood Vessels 388
11.4 Stress and Strain in the Heart 394
11.5 The Musculoskeletal System 400
11.6 The Lung 400
11. 7 Surface Tension at the Interface of the Alveolar Gas and
Interalveolar Septa 418
11.8 Small Perturbations Superposed on Large Deformation 425
11.9 Derivation of Constitutive Equation on the Basis of Microstructure:
Connection Between Micro and Macro Mechanics 431
11.10 Interdependence of Mechanical Properties of Neighboring Organs 438
11.11 Instability of Structures 439
11.12 Collapsed Structure. Example of Atelectatic Lung 443
References 447
Contents XV

Chapter 12
Strength, Trauma, and Tolerance 452
12.1 Introduction 452
12.2 Failure Modes of Materials 453
12.3 Injury and Repair of Organs 460
12.4 Shock Loading and Structural Response 460
12.5 Vibration and the Amplification Spectrum of Dynamic Structural
Response 463
12.6 Impact and Elastic Waves 468
12.7 Wave Focusing and Stress Concentration 473
12.8 Trauma of the Lung Due to Impact Load 475
12.9 Cause of Pulmonary Edema in Trauma 478
12.10 Tolerance of Organs to Impact Loads 482
12.11 Biomechanical Modeling 490
12.12 Engineering for Trauma Prevention 491
References 493

Chapter 13
Biomechanical Aspects of Growth and Tissue Engineering 499
13.1 Introduction 499
13.2 Wolff's Law and Roux's Functional Adaptation Concept 500
13.3 Healing of Bone Fracture 503
13.4 Mathematical Formulations of Wolff's Law 508
13.5 Remodeling of Soft Tissues in Response to Stress Changes 511
13.6 Stress Field Created by Fibroblast Cells and Collagen Synthesis 521
13.7 Growth Factors 524
13.8 Significance of Zero-Stress State: Changes Reveal Nonuniform
Remodeling 527
13.9 A Hypothesis on Growth 529
13.10 Engineering of Blood Vessels 533
13.11 Tissue Engineering of Skin 534
References 539

Author Index 547

Subject Index 559
CHAPTER 1

Motion

1.1 Introduction

To live is to move. In this book we consider locomotion, motion of organs,

motion of fluids around the body as in swimming and flying, and motion of
fluids inside the body, such as gas in respiration, blood in circulation, body
fluids in tissues. We then go on to consider stresses in the body, the effect of
stresses on the physiological function of the organs, the pathological develop-
ment when stresses are either too large or too small, and biological reaction
to stresses in the phenomenon of growth and change.
To analyze motion we accept the following axioms. (1) The space is Eucli-
dean, and time is an independent variable. (2) Each material particle has a
mass m, which is a constant, invariable with respect to time. (3) The location
of a particle can be described relative to an inertial rectangular Cartesian frame
of reference, with respect to which the Newtonian equations of motion listed
below are valid. Let the position of the particle be denoted by a vector* x. Let
t be the time. Let the velocity and acceleration vectors be defined by the
equations:
dx dv
v =dt' a = dt' (1)

The concept of force is introduced through Newton's law. Let F be a force

acting on the particle. Then Newton's first law states that if F = 0, then

v = constant. (2)

* Vectors are represented by boldface characters in this book.

2 1 Motion

If F -# 0, then Newton's second law states that

dv
m- = F or F = ma. (3)
dt
When Eq. (3) is rewritten as
F + (-ma) = 0, (4)
it appears as an equation of equilibrium of two forces. The term - ma is called
the inertial force. Equation (4) states that the inertial force balances the
external force. Stated in this way, Newton's law is called D'Alembert's principle.
Now consider a body that is composed of a system of particles that interact
with each other. Every particle is influenced by all the other particles in the
system. Let an index I denote the Ith particle. Let F IJ denote the force of
interaction exerted by the J th particle on the I th particle, and F J[, that of the
Ith particle on the Jth particle, I -# J. Then Newton's third law states that
(5)
If I = J we set Fll = 0. Let K be the total number of particles in the system.
The force F[ that acts on the Ith particle consists of an external force F~e),
such as gravity, and an internal force that is the resultant of the mutual
interactions between the particles of the system. Thus
K
F[ = F~e) +L FIJ. (6)
J=l

The equation of motion of the I th particle is, therefore,

dv[ _ F(e) ~ F
m[-d - [
t
+ J=l
L.. [b (I = 1,2, ... ,K). (7)

Each particle is governed by such an equation. The totality of the K

equations describes the motion of the system.
To make further progress we must specify how the forces of interaction FIJ
can be computed. Such a specification is a statement of the property of the
system of particles, and is referred to as a constitutive equation of the system.
If the number of particles is very large, then it is awkward to deal with the
large number of equations of type (7). One then asks whether it is possible to
consider the system as a continuum, whether there exists a simplified way to
write down the constitutive equation, and whether one can replace the system
of ordinary differential equations (7) with a partial differential equation. The
answer was provided by Euler, see Sec. 1.7.

1.2 Equilibrium

A special type of motion is equilibrium, i.e., one in which there is no accelera-

tion for any particle in the body. At equilibrium, Eq. (7) of Sec. 1.1 (henceforth
1.2 Equilibrium 3

designated as Eq. 1.1:7) becomes

K
F~e) +L F IJ = 0, (1 = 1,2, ... ,K). (1)
J=1

Summing over 1 from 1 to K yields

(2)

In the last sum, whenever FIJ appears, FJI also appears; according to Eq.
(1.1:5) they add up to zero. Therefore, Eq. (2) is reduced to
K
L F~e) = 0. (3)
]=1

Thus, for a body in equilibrium, the sum of all external forces acting on the body
is zero.
Next let us consider the tendency of a body to rotate. A body rotates if
there is a couple acting on it. A couple is a pair of forces that are equal in
magnitude, opposite in direction, and separated by a certain distance. In Fig.
1.2:1(a), the couple is formed by the forces F and - F, at a distance D apart.
The product FD is the moment of the couple, where F is the magnitude ofF.
If the couple in Fig. 1.2:1(a) is applied to a free body as in Fig. 1.2:1(b) then
the body will rotate. Evidently we would have to distinguish the direction of
rotation. We therefore define a moment as a vector, as in Fig. 1.2:1(c), whose
magnitude is FD, whose direction is perpendicular to the plane containing the
forces F and - F, and whose sense is determined by the right-handed screw
rule. Finally, consider a body pivoted at a point P as in Fig. 1.2:1(d). When a
force F acts on this body, the body tends to rotate about the pivot P. If we
add a pair of equal and opposite forces at P, as in Fig. 1.2:1(c), we see that the

(0)
r
(b) (c)

p----::--~
o

(d) (e) (0 (g)

FIGURE 1.2:1 Couples and moments.

4 1 Motion

action of F is equivalent to a force F and a moment FD (Fig. 1.2:1(f)). Thus,

the moment FD is the cause for rotation about P.
The moment of a force F about a point P is easily expressed in vector
notation. Let F be the force vector and I be a point on the force vector (see,
e.g., Fig. 1.2:1(g)). Let r be the position vector from P to 1. Then the moment
of F about P is
M=r x F, (4)
where the symbol x denotes a vector product.
Now consider a body that is acted on by a system of external and internal
forces. The moment of all the forces about a point P is the sum of the vector
products of r1 and the individual forces, F(e) and F IJ for I and J varying from
1 to K. But the internal forces F IJ and F Jl occur in equal and opposite pairs,
so that the sum of their moments about P vanishes. Hence the total moment
is
K
M= L r1 x F~e). (5)
1=1

If the body is in equilibrium then there is no tendency to rotate, and M = O.

Hence we obtain the second necessary condition of equilibrium: The sum of
the moments of all the external forces acting on the body about an arbitrary
point is zero.
We may rewrite the conditions of equilibrium in terms of the components
of the forces and moments as follows: Let the components of F in the x, y, z
directions of an inertial frame of reference be written as F"" Fy , Fz and those
of the moment M as M"" My, M z ; then the conditions of equilibrium are
(6)
(7)
The moments are obtained by a vector multiplication of the radius vector r
and the force F, Eq. (4). If the components of the radius vector r are written
as x, y, Z, then
(8)

1.3 Dynamics
The motion of a particle in a body is governed by Eq. 1.1:7. By summing these
equations for all particles of the body, and noting that the sum of the last
terms vanishes as in Eqs. (1.2:2) and (1.2:3), one obtains

(1)

where a 1 is the acceleration of the particle I, i.e., dVl/dt, and k is the total
number of the particles.
1.3 Dynamics 5

This equation is simplified by introducing the concept of center of mass (i.e.,

center of gravity). If we have a set of masses mI(I = 1,2, ... , k) located at places
r I with coordinates (x/> y/> ZI) in reference to a rectangular Cartesian frame of
reference, then the radius vector of the center of mass re.G. with components
Xe .G., Yc.G., Ze.G., is defined by the equations:
(2)

where
(3)

is the total mass of the body. If we differentiate both sides of Eqs. (2) with
respect to time, we obtain equations which say that the sum of the momentum
of a set of particles is equal to the product of the total mass of the particles
and the velocity of the center of mass. If we differentiate the resulting equations
again with respect to time, we obtain the result:
(4)
Here m is the total mass of the body, and ae.G . is the acceleration of the center
of mass of the body. Then Eq. (1) becomes
rnae.G. -- L...
~F(e)
I' (5)
I

This states that the total mass multiplied by the acceleration of the center of
mass is equal to the sum of all the external forces acting on the body.
To obtain an equation that describes the tendency of the body to rotate
about the center of mass, we use the same method that was used in deriving
Eq. (1.2:5). Let x, y, Z be an inertial frame of reference in which Newton's laws
are valid, Fig. 1.3:1. Let 0 be a fixed origin. Let re.G. and r I be the radius

~~---------......_ x

FIGURE 1.3:1 Position vectors re-

ferred to a fixed frame of reference
x, y, z, and the center of mass
(C.G.).
6 1 Motion

vectors from 0 to the center of mass (e.G.) and the Ith particle, respectively.
Let PI be the position vector from the e.G. to the Ith particle. Then
(6)
By differentiating with respect to time and denoting the derivative by a dot,
we obtain
(7)
Equation (1.1:7) then reads

ddt [mI(i"C.G. + PI)] = F}e) + f

J=l
Fu · (8)

Forming a vector product with PI and then summing over I, we obtain

~ PI x :t [mI(i"c.G. + PI)] = ~ PI x F}e), (9)

because the summation over the internal forces F u again vanishes in pairs.
The term on the left-hand side is
d . d .
~ mIPI x dt rc.G . + ~ mIPI x dt Pl·
The first term vanishes by the definition of the e.G. The second term is equal
to
d
dt ~ mIPI x PI - ~ mIPI x PI (10)

as can be seen by carrying out the differentiation in Eq. (10). But the last term
in Eq. (10) is identically zero. Hence Eq. (9) becomes

(11)

The quantity:
(12)

is called the angular momentum (or moment of momentum) of the body about
the center of mass. Thus Eq. (11) states that the moment of the external forces
acting on the body about the center of mass is equal to the time rate of change
of the angular momentum of the body about the center of mass.
As a special case, we have the conservation theorem of angular momentum:
With the absence of external moments about the center of mass the total
angular momentum about the e.G. is conserved.
Equations (5) and (11) are useful in biomechanical studies of crawling,
walking, running, jumping, swimming, gait, and sports. They can be applied
to the whole animal, but can be used equally well for an arm, a leg, a bone.
Some examples will be discussed in the following sections.
1.3 Dynamics 7

An entirely similar derivation can be made to prove an angular momentum

theorem in which the words "the center of mass" are replaced by "an arbitrary
fixed point." Thus, with r denoting a radius vector from a fixed point 0 (Fig.
1.3:1), we have

:t(fmIrI XiI) = frI xF~e). (13)

If the system of particles forms a rigid body, then the angular momentum
can be expressed as the product of the moment of inertia and the angular
velocity. In this case, we use a system of coordinates whose origin coincides
with the center of mass and whose axes are parallel to a set of fixed inertial
coordinates x, y, z. Let the angular velocity ofthe body be denoted by roo Then
the velocity of a point of the body relative to the e.G. is
(14)
Now, by vector analysis,
(p x p) = P x (ro x p) = (p' p)ro - (p' ro)p. (15)
Substituting (15) into (12), we obtain the angular momentum:

Lc = LI mIPI x PI = LI mI(PI' PI)ro - LI mI(PI' ro)PI' (16)

This vector equation can be rewritten in terms of components relative to a set

of rectangular Cartesian coordinates. With unit vectors i, j, k in the direction
of the x, y, z axes, respectively, we have
P = xi + yj + zk, (17)
Then, from Eq. (15),
(p x p) = (x 2 + y2 + Z2)ro - (xw x + YWy + zWz)p
= [(y2 + Z2)w x - xywy - xzwz]i
+ [ - yxwx + (x 2 + Z2)Wy - yzwz]j
+ [ -zxWx - zYWy + (x 2 + y2)wz]k, (18)
and it can be shown that Eq. (16) may be written as a matrix equation:

(19)

where
Ixx = L (y; + zf)mIo
are called the moments of inertia, and
(21)
8 1 Motion

are called the products of inertia. Together they are the moment of inertia
tensor. If the mass in the rigid body is continuously distributed, then the
summation in the equations above can be replaced by integrals. In particular,
we may write

(22)

etc., the integration being taken over the entire volume of the body, V, and x,
y, z being the location of the mass element dm.
Many of the equations above can be written in a simpler form if we
introduce the index notation and summation convention. A set of variables Xl'
X2' X3 is denoted as Xi' i = 1, 2, 3. When written singly, the symbol Xi stands
for anyone of the variables Xi' X 2 , X3. The symbol i is an index. In a product
such as aixi, the summation convention means that the repetition of an index
in a term denotes a summation with respect to that index over its range. Thus
3
aix i = a 1 x 1 + a 2 x 2 + a 3 x 3 = L aixi· (23)
i=l

Introducing the index notation by writing III = lxx, 112 = - Ixy , etc., and
using the summation convention, we may write Eq. (19) as
(24)
and Eq. (11) becomes

:t (IijWj) = (~PI x F}e)\ = eijkPIjF}k)' (25)

where eijk is the permutation symbol, which is defined by equations such as

e ll1 = e 222 = e 333 = e112 = e 1l 3 = e 221 = e 2 23 = e3 31 = e 332 =0
e 123 = e 231 = e312 = 1 (26)
e 213 = e 321 = e 132 = -1
i.e., eijk vanishes when the values of any two indices coincide; eijk = 1 when
the subscripts permute like 1, 2, 3; and eijk = - 1 otherwise.
If the moment of inertia tensor Iij does not change with time, then we have

Iijwj = (L PI X F~e»)i = eijkPIjF}kl. (27)

Equations (5) and (27) describe the mechanics of rigid bodies completely.
For rigid bodies there is no need to know the forces of interaction, FIJ , between
the particles of the body. If the body is not rigid, and if one wishes to analyze
the motion of a particle in the body relative to others, then we must return to
Eq. (1.1:7). In that case, we must know the forces of interaction, FIJ.
Many machines (e.g., an automobile engine, an electric motor, a bicycle)
can be considered an assemblage of rigid bodies connected by joints, springs,
1.4 Modeling 9

and dash pots. The forces of interaction between the individual rigid parts are
often the unknown variables in the analysis of the motion of the machine. In
analyzing the locomotion of an animal, the body of the animal may be treated
in a similar manner.

1.4 Modeling
Sometimes it is easy to make a mathematical model of a natural system, e.g.,
an arm holding a load shown in Fig. 1.4:1. The flexor muscle force that
balances the load and the weight of the forearm are computed easily. However,
modeling the chest for the analysis of breathing is not so easy, because the
large number of muscles that control the chest motion is difficult to idealize.
Therefore creating a mathematical model that simplifies the real thing while
retaining all the essential factors is considered an art that must be based on
experience and careful thought.
An example of difficult modeling is that of the human lumbar spine to
which many muscles and ligaments are attached. Nachemson and Elfstrom
(1970) measured the pressure in the nucleus pu/posus, the gelatinous center of
the lumbar disk (L3), with a miniature pressure transducer, and from the
data calculated (with the help of in vitro disk experiments) the loads on the
lumbar spine. Their results are shown in Table 1.4:1. Note how large the load
in the lumbar spine is when one lifts weight the 'wrong way'. Data of this kind

FIGURE 1.4:1 A leaf from Giovanni Alfonso Borelli's book, On the Movement of
Animals, published in 1680/ 1681, illustrating his research on biomechanics. Borelli
(1608-1679) was professor of mathematics at Pisa. This book was translated recently
by Paul Maquet, and published in 1989 by Springer-Verlag.
10 1 Motion

TABLE 1.4:1 Approximate load on the lumbar disk L3 in a 70 kg individual in different

positions, movements and maneuvers. Data from Nachemeson and Elfstrom (1970)
Position/movement/maneuver Load, Newton (kgf)
Supine 490 (50)
Supine in traction (30 kg) 343 (35)
Standing 980 (100)
Upright sitting, no support 1373 (140)
Walking 1128 (115)
Twisting 1177 (120)
Bending sideways 1225 (125)
Coughing 1373 (140)
Jumping 1373 (140)
Isometric abdominal muscle exercise 1373 (140)
Laughing 1471 (150)
Bending forward 20° 1471 (150)
Bending forward 20° with 10 kg in each hand 2108 (215)
Lifting 20 kg, back straight, knees bent 1814 (185)
Lifting 20 kg, back bent, knees straight 3825 (390)
Sit-up exercise, knees bent 2059 (210)
Sit-up exercise, knees extended 2010 (205)
Active back hyperextension, prone 1765 (180)

are significant when one tries to understand the etiology of the low back pain
which affiicts many people.
Schultz (1986) approached the same problem by measuring the myoelectric
activity, the electric signals sent through nerves that cause muscles to contract.
There are evidences that under the right circumstances the myoelectric activity
is linearly related to muscle contraction forces. His results on the forces in
the spine and muscles when a man is attempting maximum backward bend
are shown in Table 1.4:2. It is impressive how large these forces can be! Merely
bending the trunk forward 30 degrees triples the muscle force that prevails at
relaxed-standing condition. Attempting maximum backward bending caused
compression in the spine ten times the relaxed standing value; and very large
tension in the major trunk muscles.
As a further illustration of the complexity of modeling, consider referred
pains which are aches and pains originating in a distant organ. Familiar
examples are pain in the shoulder accompanying diaphragmatic disorders, pain
in the knee and arthritis ofthe hip, the sacral (lower back,just above buttocks)
pains of childbirth, pain in the left shoulder and left arm in angina pectoris,
and the testicular pain referred from lower back. One of the first crucial
experiments on this subject was done by Thomas Lewis in 1936. Wishing to
investigate muscular pain, he injected an irritant deep into the lower lumbar
region. He found a diffuse pain running down the lower limb but little
1.4 Modeling 11

TABLE 1.4:2 Typical magnitude of model-estimated lum-

bar trunk internal forces imposed by physical exertion
when an adult man stands and attempts maximum back-
ward bend. (From Schultz, 1986)
Forces in the spine Newton
Compression 5,050
F orward/back shear 600
Side/side shear 370
Contraction forces per side in major muscles
Rectus abdominus 340
Oblique abdominals (sum of 4) 1,700
Erector spinae
multifidus 320
longissimus 630
iliocostalis 630
Lumbar latissimus dorsi 150
Quadratus lumborum 230
Psoas 440

discomfort at the site ofthe injection (Lewis, 1942). In 1938, Kellgren published
the results of a systematic examination of the phenomenon of referred pain,
showing them to radiate segmentally and not to cross the midline (Kellgren,
1938). Cyriax (1978) presents considerable details of this subject.
Referred pain is a subject for biomechanics because pressure on nerves is
often the cause. When pressure is applied close to a nerve's distal extremity,
the sensation of pins and needles is felt, but the main symptom is numbness.
When pressure is applied on the trunk of a nerve and then released, the sensa-
tion is pins and needles rather than numbness. But if pressure is applied on a
nerve root in the spine, the sensation of pins and needles is felt only as long
as the pressure is sustained, but disappears as soon as the pressure is relieved.
Stretching a nerve root is painful, and a common cause is a disk protrusion.
Figure 1.4:2 illustrates the concept of a cartilaginous disk lesion. An annu-
lar crack has led to a posterior displacement by hinging. The posterior
longitudinal ligament is bulged out backward and pressure is exerted on the
dura mater, causing a backache. No nuclear material has extruded, hence
reduction by manipulation is simple.
Figure 1.4:3 illustrates the concept that a disk lesion may push the posterior
ligament until it presses on the dura mater, causing backache. If the protrusion
proceeds posterolaterally and impinges on the nerve root, then pain will be
felt in the limb. In these cases the treatment of choice is a sustained traction.
Spontaneous recovery is possible if the posteroinferior aspect of the vertebral
bone is resorbed and the protrusion is accommodated so that the dura mater
and nerve roots are no longer subjected to pressure.
12 I Motion

Annulus fibrosus

Nucleus pulposus

Posterior
longitudinal
ligament

FIGURE 1.4:2 A cartilaginous disk lesion. From Cyriax (1978). Reproduced by

permission.

protrusion

duro mater
vertebral

~----ItIHlr",* nerve root

FIGURE 1.4:3 Illustration of a pulp and disk lesion. From Cyriax (1978). Reproduced
by permission.

To model referred pain, we must consider the whole system: the organ in
pain and the nervous lesion. The analysis of a good model may lead to designs
to relieve such pains.
Further discussion of modeling is presented in Sees. 2.9 and 12.11.

1.5 Sports Techniques

In jumping, there are phases when the body is in free flight and the trajectory
of the body's center of gravity is a parabola which is determined solely by the
initial conditions. The trajectory cannot be changed by the athlete's action
1.5 Sports Techniques 13

while in flight, but the body position can. Controlling the body position is of
course important. Let us illustrate:

Long Jump
In the long jump, the distance covered consists of three parts (Fig. 1.5:1): L l ,
the horizontal distance between the front edge of the takeoff board and the
athlete's center of gravity at the instant of takeoff; L 2 , the flight distance; and
L 3 , the distance between his e.G. at the instant his heels hit the sand and the
marks in the sand. The records of good athletes show that Ll is usually about
3.5% of the total, L2 is about 88.5%, and L3 is about 8% (see Hay, 1978, p. 409).
The takeoff distance Ll is a function of the accuracy with which the athlete
places his foot on the takeoff board, his physique, and his body position. The
flight distance L2 depends on his speed, angle of takeoff, height of takeoff, and
the air resistance. The landing distance L3 depends on his body position at
touchdown and the actions he takes to avoid falling backward.
The most important of these variables is the athlete's speed at the instant
of takeoff. This speed depends on his running and on the losses associated
with the adjustments made when preparing for takeoff. The ratio of the vertical
speed (or lift) acquired at takeoff to the horizontal speed is the tangent of
the angle of takeoff. It is well known that, for a given initial speed, the
largest distance is achieved by any projectile when the angle of takeoff is 45°.
However, the lift that the athlete develops at takeoff is influenced by the speed
of his run-up. The faster his run, the less time his foot spends on the ground
at takeoff and the less vertical speed he can develop. Thus, the angle of takeoff
used by top-class jumpers lies in the range of 19-22°.
Once the athlete is in the air, he should try to obtain the optimum body
position for landing. Usually the jumper acquires a forward rotation of the
body at takeoff, which tends to bring his feet beneath his center of gravity at
the very instant of landing, while he actually wants them to be well forward.
Thus, the athlete's principal problem is to overcome this forward rotation.

I,

~1)
-------------------L 2 - - - - - - - - - - - - - - - - - - - - - - - __ L _
3

FIGURE 1.5:1 Contributions to the length of a hang-style long jump. From 1. G. Hay
(1978), p. 409. Reproduced by permission.
14 1 Motion

I "

FIGURE 1.5:2 The in-the-air position adopted in the sail technique. From J. G. Hay
(1978), p. 415. Reproduced by permission.

Compare the two techniques used in the long jump as illustrated in Figs.
1.5:1 and 1.5:2. The sail technique (Fig. 1.5:2) is used by most jumpers. Its
weakness is that it places the athlete's mass close to his transverse axis and
thus obtains a small moment of inertia, facilitating the forward rotation
instead of retarding it. In the hang technique (Fig. 1.5:1) the athlete reaches
forward with his leading leg and then sweeps it downward and backward until
he has both legs together and somewhat behind the line of his body. He swings
his arms backward, too. This results in an increase in the moment of inertia
and a slowing of forward rotation. He continues to swing his arms upward,
then bends his knees and begins the forward movement of his legs in prepara-
tion for landing. At the time oflanding, the extended position, (trunk inclined
slightly backward and hands beside hips), as shown in Fig. 1.5:1, is preferable
to the jackknife position (trunk inclined forward and arms extended toward
the feet), as shown in Fig. 1.5:2. Studies show that the extended position gives
the advantage of approximately 12 more inches of jumping distance than the
jackknife position.
There is a third in-the-air technique called the hitch-kick, or the run-in-
the-air. Its first part is similar to that of the hang technique. Coordinated with
this movement is a pulling through of the takeoff leg. Because of the difference
between the moments of inertia of the extended and flexed legs, the angular
momentum of the leading leg as it swings downward and backward far exceeds
that of the takeoff leg that is moving in the opposite direction. As a result of the
preservation of the total angular momentum, the trunk rotates backward,
which is, of course, the objective to be achieved. Now, the athlete's legs are in
a position that is essentially the reverse of that at takeoff. At this point, the
leg that is to the rear is brought forward, with the knee fully-flexed, to join
the other in preparation for landing. For a longer jump an additional full
stride may be attempted. The hitch-kick method is used by high performers.

High Jump
In the high jump the height that an athlete clears is the sum of three heights:
the height of the athlete's center of gravity (e.G.) at the instant of takeoff, the
1.5 Sports Techniques 15

TABLE 1.5:1 Relative contributions to height in the high jump

The three parts of performance Height (em) Percentage
Height of e.G. at takeoff 144.0 67.5%
Height e.G. is lifted 78.2 36.6%
Height cleared - max C.G. height -8.6 -4.1%

height of his e.G. raised during the flight, and the difference between the
maximum height reached by his e.G. and the height of the crossbar. Hay
(1978) gives the data for a jump of 7 feet by then-world-record-holder Pat
Matzdorf in Table 1.5:1. According to the table the height of e.G. at takeoff
is very important. This explains why high jumpers are tall men with long legs.
The height of his flight is governed by his vertical velocity at takeoff, which
depends on his actions during the last one or two strides of his run-up. If at
the end of his penultimate stride he has sunk low over his supporting leg and
then takes a low, fast step with his takeoff foot, his e.G. is likely to have no
downward vertical velocity at the instant this foot touches down. Then all his
effort will be used to raise his C.G. This is far better than if he sinks down
with the penultimate stride.
The vertical force involved at takeoff results from the swing of his arms and
leading leg and from the extension of the hip, knee, and ankle joints of his
jump leg. The period of time during which the jump foot is in contact with
the ground depends on the style of jumping. Athletes who use the straddle

FIGURE 1.5:3 A frontal approach and a front-piked position over the bar suggested by
J. G. Hay (1978), pg. 433. Reproduced by permission.
16 1 Motion

style have takeoff times in the 0.17-0.23 sec range, whereas those who use the
flop style, popularized by Fosbury, tend to have takeoff times in the 0.12-0.17
sec range. It also depends on the free limbs. Athletes who use a straight leading
leg action generally have a longer time of takeoff than do those who use a
bent leading leg. But in spite of the fact that impulse = force x time, it has
been found that, within limits that are specific to the individual athlete, the
shorter the time of takeoff the greater is the vertical lift (impulse) that the
athlete obtains. This fact is hard to explain unless it is a feature of the muscle.
With the common styles: scissors, straddle, cutoff, and roll, the maximum
e.G. height is higher than the height cleared, (Table 1.5:1). In 1970 Dick
Fosbury originated the Fosbury flop, in which the athlete arches backward
over the bar in such a way that his C.G.lies outside his body. Hay (1978) then
proposed a similar method shown in Fig. 1.5:3, bending the upper body
forward toward the feet. Note that the trajectory of the e.G. does not clear
the crossbar while the athlete's body does.

Use of Friction
If two bodies are in contact along dry surfaces, the limiting friction is equal
to the normal reaction (force of interaction between the two bodies in the
direction normal to the surface of contact) multiplied by a constant whose
value depends only on the nature of the surfaces. Thus,
F = liN, (1)
where F = the limiting friction, N = the normal reaction, and fl = a constant
called the coefficient of limiting friction. If the tangential force between the
surfaces is less than the limiting friction, F, the bodies will remain stationary
relative to each other. If F is exceeded, relative motion ensues. When a body
is actually sliding on the surface of another body, the magnitude of the fric-
tion is given by the equation
(2)
where F. = sliding friction, N = the normal reaction, and fl. = a constant
called the coefficient of sliding friction. fl. is smaller than fl. Our shoes on solid
ground have a coefficient of limiting friction of the order of one. The surfaces
with the smallest coefficient of sliding friction are probably those of the
articular cartilage surfaces of our joints. In Biomechanics (Fung, 1981, p. 410)
we have shown that in cyclic loading a whole bovine synovial joint has a
coefficient of sliding friction of 0.0026 at a normal stress of 500 kPa (~5
kgf/cm 2 ).
All athletes pay attention to ground friction by choosing suitable shoes
according to the nature of sports and the condition of ground surface. Shoes
affect the coefficient of friction. The other factor that can be controlled to a
certain extent is the normal force N. N is affected by the athlete's posture and
body dynamics. For example, a rock climber can increase the limiting friction
1.6 Prosthesis 17

between his feet and the rock by leaning well away from the rock face (see
Prob. 1.11), because N depends on the location of the center of gravity of his
body and the direction of the rope used for the climb.
The dynamics of the athlete also affects the limiting friction because in
arresting a downward movement toward the ground, an upward inertial force
must be created by the ground reaction. Thus the ground reaction acting on
the feet of a man running exceeds that when he is standing, and the technique
of running does have an influence on ground friction.

Use of Hands in Swimming

The forces exerted on the swimmer in reaction to the movements of his arms
is a prime source of propulsion. Normally, this propulsive force is equal to
the drag force acting on the swimmer's arms. However, Counsilman (1971)
and Brown and Counsilman (1971) have suggested that lift may be used. With
a small angle of attack, the hand can function as a hydrofoil. As a hydrofoil,
the lift is larger than the drag, and can be used advantageously with a proper
arm cycle. In a breaststroke cycle, the hands move forward, outward, inward,
and forward again. Now, with an appropriate angle of attack, the lift force
acting on the hands can be directed forward and thus become propulsive.
Similar circumstances can be shown to exist in other competitive strokes: front
crawl, butterfly stroke, and back crawl. The same principle can be applied to
the propulsion derived from the legs in applicable circumstances.
Birds, insects, and fish use lift force for their locomotion. More details about
the fluid mechanical principles oflift generation are given in Chapters 3 and 4.

1.6 Prosthesis

A prosthesis, by definition, replaces a missing body part for structural, func-

tional, or cosmetic reasons. Thus, dentures, artificial heart valves, artificial
limbs are prostheses. Most of them require extensive biomechanical research
before an optimal design can be obtained.
Take an artificial leg or a joint replacement as an example. To make such
a prosthesis, we must know the force that acts in it as the person moves. This
force depends on the adjacent musculature. Records of the variation with time
of the forces in hip and knee joints during level walking show two and three
maximum values respectively in the stance phase of each cycle. An indication
of the variation in joint forces with body weight Wand the ratio of stride
length L to height H is given in Figs. 1.6:1 and 1.6:2. Here the average of the
maxima is plotted to a base of WL/H (Paul, 1970). Note how large the joint
forces are. They are much larger than the body weight. The increase is due to
the action of the muscles, which must keep the basically unstable structure in
balance.
18 1 Motion

5000
z
ai
u
....
0
4000
•
....c 3000
'+-

·0
c.
..c: 2000 •
c
ro
0)
~ 1000

0
40 50 60 70
WL
H,(kg)

FIGURE 1.6:1 Variation of mean hip joint force in level walking with body weight W,
stride length L, and height H. From J. P. Paul (1970). Reproduced by permission.

2000

,• •
z
0)'

• •
u
....

"
0 1500 I
....c::
'+-

e.
• •
0
0)
0)
1000
c
~
0)
Cl
ro
.... 500
0)
>
«
0
40 50 60 70
WL
H,(kg)

FIGURE 1.6:2 Variation of mean knee force in level walking with body weight W, stride
length L, and height H. From J. P. Paul (1970). Reproduced with permission.

1.7 Continuum Approach

In the preceding sections, we presented Newton's equations of motion for a

system of particles in general, and rigid bodies in particular. In biomechanics,
there are many occasions (e.g., in the study of locomotion) in which we can
regard an animal as an assemblage of rigid bodies connected by joints and
1.7 Continuum Approach 19

muscles. If the mechanical properties of the connectors are known, then a

system of equations of the type (1.3:5) and (1}:25) will suffice to describe the
motion. There are other occasions, however, in which this is not efficient. For
example, we may wish to know the stresses in the bones, muscles, spine, heart,
or brain in a man when walking. Then it is no longer convenient to write the
equations of motion in the form of (1.1:7), because the number of particles
to be considered is excessively large, and the way to specify the forces of
interaction between the particles, FlJ of Eq. (1.1:7), becomes very complex. In
these occasions it is simpler to consider the body as a continuum, and search
for a simplified way to specify the forces of interaction between particles by
means of constitutive equations.
The equation of motion of a continuum was derived by Euler, by applying
Newton's laws and several additional axioms. These axioms are: (1) The
material particles form a one-to-one isomorphism with real numbers in a
three-dimensional Euclidean space. (2) The mass distribution is characterized
by the density, p, the mass per unit volume, defined as a piecewise continuous
function over the volume of the continuum. (3) A particle can interact only
with its immediate neighbors. "Interaction at a distance" is ruled out. The
force of interaction of particles on the two sides of an arbitrary infinitesimal
surface in a continuum can be expressed as a surface traction (force per unit
area). The surface traction acting on an arbitrarily oriented infinitesimal
surface can be computed from a well defined stress tensor.
Let Xl' X2' X3 be an inertial rectangular Cartesian frame of reference. Con-
sider an infinitesimal volume dx1dx2dx3. Let Uij' eij be the stress and strain
tensors, respectively. Let Vi be the velocity vector, DvdDt be the acceleration
vector, and Xi be the body force per unit volume. All Latin indicies range
from 1-3. Then Newton's law takes on the form of Euler's equation:
DVi _ oUij X
(i = 1,2,3). (1)
P Dt - ox.
J
+ i'

See a text on continuum mechanics for details, e.g., Fung (1977). In Eq. (1) the
summation convention of tensor analysis is used. A repetition of an index in
any given term means a summation over the range of that index. Thus,
0(1·· 0(1·1 0(1·2 0(1·3
.-....!L = _"_ + _"_ + _"_. (2)
OXj oX l OX2 oX 3

at
DVi.IS t h e matena . 0 f Vi:
. I d erwatwe
.

DVi OV i OVi OV i OVi OVi OVi

-
Dt
= -;-
ut
+ Vj-;-
uXj
= -;-
ut
+ v l -;-
uXl
+ V2 -uX2
; - + V3 -;-.
uX3
(3)

The conservation of mass is expressed by the ~quation of continuity:

op + OPVj = o. (4)
ot OXj
20 1 Motion

If the material is incompressible, then Eq. (4) is reduced to the form

OVj = o. (5)
oXj
Further development requires specification of the properties ofthe material
in the form of a constitutive equation which relates stress with strain or strain
rate, or strain history. For the convenience of the rest of the book let us collect
a few important examples. A material is called a Newtonian fluid if it obeys
the following stress-strain-rate relationship:

OVk) (OV. OV.)

(Jij = -pc5ij + A ( oX k c5ij + Il ax: + O~ , (6)

where p is the pressure, and A and Il are two constants called the coefJicents
of viscosity. If the fluid is incompressible, then according to Eq. (5) the constitu-
tive equation of a Newtonian fluid becomes
OV. OV.)
(Jij = - pc5ij + Il ( ax: + ax'; . (7)

Substituting Eq. (7) into Eq. (1), one obtains the Navier-Stokes equation

au; + u. au; = _~ op +!!.- 02U; + X. (8)

at J OXj pox; p OXjOXj ,
in which the constant III p is the kinematic viscosity of the fluid. To solve Eq.
(8), appropriate boundary conditions must be used. If a Newtonian fluid comes
in contact with a solid body, then the appropriate boundary condition is that
there be no relative motion between the solid and fluid. This is the no-slip
condition. The fluid adheres to the solid whether the surface is wettable or not.
This condition is not so intuitively obvious, but has been found to be valid in
all cases examined in the past, except for rarefied gases in which the mean free
path of the molecules between collisions is comparable to the size of the body.
Fluids whose constitutive equation does not obey Eq. (6) are said to be non-
Newtonian. Blood is non-Newtonian. Most body fluids are non-Newtonian.
At shear strain rate above 100 sec- 1 blood is almost Newtonian. Air and water
are Newtonian.
A material is called an isotropic Hookean solid if it obeys the following
stress-strain relationship:
(9)
or its inverse:

(10)

Here eij is the strain tensor, A, G, E, v are constants. A and G are the Lame
1.7 Continuum Approach 21

constants, G is the shear modulus, E is the Young's modulus, v is the Poisson's

ratio. bij is the Kronecker delta, which is equal to 1 when i = j, and zero
when i =f. j. Equations (9) and (10) are called Hooke's law.
On substituting Eq. (9) into Eq. (1) we obtain
Dv· a
p-' = A-;-eaa
oe··
+ 2G~ + Xi. (11)
Dt uX i uXj

Here DvJDt is given by Eq. (3). The strain must be referred to a configuration
of the body in which the stress is zero, because according to Eqs. (9) and (10),
zero stress implies zero strain and vice versa. Let the displacement vector of
a point in the body be measured with respect to an inertial rectangular
Cartesian frame of reference and be denoted by u i • If U i is finite, the strain-
displacement relationship is nonlinear, see Chap. 9. However, if Ui(X 1 , X 2 , X 3 , t)
is infinitesimal, then

(12)

DVi 02U i
(13)
Dt ot 2 ·

To the same order of approximation, the material density is a constant:

p = const. (14)
On substituting Eqs. (12)-(14) into (11), one obtains the well-known
N avier's equation:

(15)

If we introduce the Poisson's ratio as in Eq. (10),

A
v = ----,-- (16)
2(A + G)'
then Navier's equation can be written as

G -02Ui 1 a OUj 02Ui

-+----+x=p- (17)
OXjOXj 1 - 2v oX i oXj , ot 2 •

Navier's equation is the basic field equation of the linearized theory of

elasticity. It must be solved with appropriate initial and boundary conditions.
Living bodies, however, often take on finite deformations and obey more
complex constitutive equations than the above-mentioned. The field equa-
tions are therefore more complex. Examples of formulation and solutions of
organs subjected to finite deformations and nonlinear constitutive equations
are discussed in Chapters 10 and 11.
22 1 Motion

Problems
1.1 Hold your arm in a horizontal position while lifting a weight in the hand. Name
the major muscles that must provide the tension. How can you determine the
tension in these muscles? Invent a theoretical and/or an experimental way to
determine the tension in individual muscles.
1.2 What factors determine the pressure in the abdomen when one (a) lifts a heavy
weight, (b) swims, (c) relaxes in bed? What muscles are involved?
1.3 When a giraffe moves its head up and down 6 m, how much does the hydrostatic
pressure in its cerebral blood vessels change? In which way would this change in
pressure affect the blood circulation in the giraffe's head? If such large change in
blood pressure occurs in human brain, what ill effect may be expected?
1.4 Locusts (Schistocerca) can jump a distance up to about 80 cm on level ground. If
a locust takes off at an angle of 45°, show that the initial velocity should be
280 cm/sec in order to reach 80 cm.
Figure P1.4(a) shows the hind leg of a locust. The solid outline shows the leg
before it starts to jump. The dotted outline shows its configuration at takeoff.
Figure P1.4(b) shows the skeleton at some intermediate stage of jumping (AB =
tarsus, BC = tibia, CD = femur, G = center of gravity of locust, C' = the point of
attachment of muscle). R. H. 1. Brown has taken cinephotographs oflocustjumps
(Times Sci. Rev. 6-7, 1963) and observed that locusts actually take off at at least
55°, because they jump to get airborne, and then start flying. He reports that the
takeoff speed is about 340 cm/sec for an 80 cm jump. He shows that the muscle is
roughly parallel to the femur, and that the distance CC' in Fig. P1.4(b) is about

",':.,,-,
/" ?,J /
I I
I

I
/
I
"
" "
/
/

'b'
~- T 1.la
I
1,/ /'

I ""
\ rFemur + " r'
, I trochanter ~I
I I ': I-Coxa
\ I 1/

A
--------~-L------x

(a) (b)

FIGURE P1.4 The hind leg of a locust: (a) Before it starts to jump (solid outline), and
at the moment oflifting off (dotted lines); (b) The diagram offorces and angles.
Problems 23

BC/3S. The locust accelerates its body by extending the hind leg rapidly. This
moves the locust's coxa through a distance of about 4 em before the tarsus leaves
the ground.
Using Brown's data, compute the force exerted by the muscle during the jump
by first deriving the formulas for the velocity v and the distance s at time t for the
free fall of a particle subjected to a constant gravitational acceleration g. If v = u,
s = 0 when t = 0, we have
v = u + gt,
v2 = u2 + 2as, s = (u + v)t/2.
Show that a projectile fired from level ground with initial velocity u and an
elevation angle of IX. will reach a height u2 sin 2 1X./2g and a range u2 sin 21X./g if air
resistance is ignored. Then show the following results by Brown successively:
(a) With IX. = S5" and a jump of distance s = 80 em, the initial velocity u must be
290 em/sec. The actual take-off speed of 340 em/sec is greater than 290 em/sec
in order to compensate for air resistance.
(b) Ifu = 0, v = 340cm/sec, and s = 4 em, the acceleration is a = l4,SOOcm/sec 2
if a can be assumed to be a constant.
(c) Find the resultant of the inertial force and the weight of the locust. This force
(shown as FG in Fig. Pl.4b must pass through the center of gravity of the locust
if it does not send the animal spinning.
(d) Assuming an intermediate configuration with the angles L CBX = 100°,
L GFX = sr, LBCD = 90°, and FB = BC/20, calculate the force in the mus-
cle. Show that for a 3g locust this is about S Newton. Quite a force!

Discussion of Problems 1.1-1.4

Questions like those in Problems 1.1 and 1.2 would provide motivation to
study anatomy. Problem 1.2 suggests that we can control the pressure in our
abdomen by voluntary muscle action. In Biomechanics (Fung, 1981, p. 14), it
is shown that in order to relieve excessive loading on the spine it is necessary
to tense up the abdominal muscles and increase the internal pressure. Weight
lifters should have strong abdominal muscles.
In a relaxed state the human abdomen is soft. An effective way to make it
more rigid is to increase the internal pressure. This is analogous to a soft
garden hose which can be stiffened by turning on the water pressure. An
erectile organ becomes erect by squeezing blood with a sphincter muscle. For
example, Vathon reports (Annual Conference of Engineering in Medicine and
Biology, Nov. 19, 1968, Houston, Texas) that the blood pressure in the corpus
cavernosum artery of a bull's penis in erection is as high as 1,750 mm Hg.
Problem 1.3 takes on a physiological significance if you know that the
diastolic pressure in the left ventricle is normally not much higher than the
atmospheric pressure; nor is the pressure in the vena cava at the level of the
heart. (What are the basic reasons for this?) Hence as the giraffe lifts its head,
24 1 Motion

the pressure in its cerebral veins will be lower than atmospheric. Hence there
is a tendency for the veins to collapse. If the vessels were collapsed the
resistance to blood flow would increase. Then, how does the giraffe's heart
supply the needed blood flow to the brain?
On the other hand, when its head is lowered, wouldn't the blood vessels
expand and be gorged with blood? Would the brain tissue be squeezed? With
what consequences?
Similarly in a standing man the hydrostatic pressure in the leg veins will
cause the veins to bulge and become a reservoir of blood. If he lies down, the
hydrostatic pressure in the leg decreases and the volume of the circulating
blood will be increased. Thus, consider the following question: If a person who
is walking suddenly felt a chest pain and thought a heart attack imminent,
should he lie down immediately in order to minimize the work of the heart?
The large tension revealed in Problem 1.4 is significant when you realize
that our own joints are similarly constructed. The muscles that move our legs
and arms are almost parallel to the long bones, and the distances between the
muscles and the fulcrum are small compared with the length of the limb. Thus
the forces in our leg muscles can be large compared with the weight of our
bodies. If a standing man sways a little his leg muscle must provide tension to
control his posture. Hence, even when standing still the tensions in the muscle
are significant.
Compare your elbow with your knee joint. Your have the patella at the
knee but not at the elbow. What is the function of the patella? What is the
consequence of this difference?

Problems
1.5 Design a human-powered hydrofoil for sports.
1.6 What polymer is pressure sensitive? Design a thin-film pressure transducer which
can be used to measure pressure in narrow spaces such as intrapleural space,
space between vertebra and disk, and in joints. Describe possible applications to
sports, sports injury and repair research, gait, diagnosis, mechanical cardio-
pulmonary resuscitation, etc.
1.7 Design recreational sports for the handicapped. Design helpful devices for the
handicapped in their daily life.
1.8 Certain problems in industry are biomechanically oriented. For example, an
airline stewardess often has a sore neck and spine strain due to leaning over
to passengers. Coal miners often have loss of hearing or black lung disease.
On the other hand, assembly line workers may reduce fatigue and improve
efficiency by scheduling frequent short digression and exercise in the course of
the day. Elaborate on these "work physiology" or "industrial biomechanics" by
developing a program for an industry. Discuss the more fundamental, scientific
part of the program.
1.9 Figure P.1.9 shows the type of exercise used effectively by patients with back pain
at the Tientzin Hospital in China. Explain the rationale.
Problems 25

FIGURE P1.9 Hyper-extension exercises to strengthen the back of lumbago patients.

Keep your back hollow.

FIGURE Pl.10 A Russell traction for the leg.

1.10 A Russell traction for immobilizing femoral fractures is shown in Fig. P.l.10.
Determine the resultant force applied to the femur.
1.11 Consider a rock climber. Show that the normal reaction acting on the climber's
feet is equal to the product of the weight of the climber and the perpendicular
distance from the anchor of the rope to the vertical vector passing through the
26 1 Motion

center of gravity of the climber, divided by the distance between the anchor of
the rope and the climber's foot. How can the climber increase the limiting friction
between his feet and the rock?
1.12 A patient suffering from low back pain often needs to stretch his spine. Design a
stretcher that can be used when the patient lies in a hospital bed (e.g., a pulley
system).
1.13 Design a stretcher that can be used by the patient at home. An example is a tilt
table into which the patient can strap his feet and then tilt to an angle so that he
is hung upside down.
1.14 Design an artificial aortic valve. Discuss the pros and cons of your design.
1.15 Since veins have valves and can serve as a one-way tunnel, design a mechanism
which can apply external pressure on the legs in a certain way so that the
mechanism can serve as a heart-assist device. Discuss possible uses and the pros
and cons of your design.
1.16 Acceleration and deceleration can create pressure gradient which may help blood
circulation. Design a machine which can shake a bed on which a patient lies in
such a way as to serve as a heart-assist device. Discuss the pros and cons of your
design.
1.17 Consider a very simple model of a scoliotic spine and three methods of correcting
the deformities, as shown in Fig. P.Ll7. Calculate the corrective bending moment
obtained at the apex of the curve, C, due to the three types of loading. Show that
for severely deformed spines (8 > 53°) method (a) leads to a larger corrective
moment than (b); for milder curves (8 < 53°) method (b) is preferred, whereas (c)
is better for all degrees of deformity. Cf. Panjabi and White (1980).

F
2

(a) (c)

FIGURE PLl7 Several mechanical ways to help straighten a curved spine.

Problems 27

1.18 Model a subsystem of two vertebrae and a disk by a mass-spring-damper

system, and write down the governing differential equation. For simplicity,
consider only one degree of freedom, the vertical compression. Let the lower
vertebrae be fixed, and denote the displacement of of the upper vertebrae by x.
1.19 Soldiers with chest wounds (perforated parietal pleura) often have collapsed lungs
because of pleural space exposure to atmospheric pressure. The local increase of
pleural pressure to atmospheric value reduces the effective pressure that inflates
the lung.
Automatic collapse oflung (pneumothorax) sometimes occurs to healthy athletes
while playing in the field. Explain how this could happen?
1.10 An accident broke a person's neck. Design a mechanism that will hold the head
in place. (Such a device is sometimes called a "halo" device by its shape.)
1.21 A person has one of his facial bones knocked slightly inwarq during an auto-
mobile crash. A surgeon wants to pull that bone out to its original position.
Design a mechanical device to do it.
1.21 A patient has a broken femur due to a fall. The most common treatment is to put
the leg in a plaster of Paris cast while it heals slowly. The Chinese school (Chinese
Academy of Traditional Medicine, Institute of Sone anq Trauma, Beijing and
Tientsin) has evidence that quicker healing can be achieved by a less rigid
approach. They advocate the use of small boards (slats) of soft wood (e.g. balsa,
with a width of about an inch or so) ban4aged reasonably tightly (but not too
tightly) around the broken bone. The patient is then taught a proper course of
exercises which puts some compressive stress in the bone but minimizes shear
stress or torsion. Design such a slat-bandage. Analyze its effect on the broken
bone. Compare the biological process of healing that occurs in the Chinese way
with the conventional way of plaster cast by taking tht} muscle forces into
consideration.
1.23 Design a saw that can cut a plaster of Paris cast over a broken bone without
hurting the soft tissues, (Note: How about using high frequency small amplitude
oscillations?)
1.14 Scotch adhesive tape is used for bandage. The mechanics of tearing off a adhesive
tape is very important to the surgeon and the patient. Develop a theory of peeling
off adhesive tape.
1.15 A proper strategy of surgery and placement of bandages afterwards can be of
great importance to healing. Consider a patient with skin cancer. A piece of skin,
the size of a dollar, ha!! to be removed. Design a strategy to excise the diseased
skin and suture the healthy skin to cover the wound. Design a scheme to put on
the bandage after surgery.
1.26 If Il peam is simply supported and is loaded by a weight W at its center, the
rea!:!tipn at each of the supports is W/2. Consider a man standing. His two legs
support his upper body with ball and socket joints (i.e., simply supported). Yet
the reactions at the joints are far greater than W/2, and are very sensitive to his
posture, Explain why with a good sketch of the anatomy anq free-body diagrams
for the forces, and analytical calculations.
28 1 Motion

References

This bibliography lists books and papers referred to in the text and problems, and a few selected
entries that are not specifically discussed in the text, but are of importance to the topics. Since
this volume is a sequel to the author's books on mechanics, the following references are quoted
throughout this book:
Fung, Y. C. (1955; Revised 1969). Theory of Aeroelasticity. Wiley, New York. Revised, Dover,
New York.
Fung, Y. C. (1965). Foundations of Solid Mechanics. Prentice-Hall, New Jersey.
Fung, Y. C. (1969; 2nd ed. 1977). A First Course in Continuum Mechanics. Prentice-Hall, New
Jersey.
Fung, Y. C. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
Fung, Y. C. (1984). Biodynamics: Circulation. Springer-Verlag, New York.
For the classical mechanics used in Sees. 1.2, 1.3, and 1.7, see Fung (1965,1977), Greenwood
(1965), and Yih (1969, 1989). For anatomy, see Gray (1973). For modeling, see Sec. 12.11 and
References of Ch. 12.
Brown, R. M. and Counsilman, J. E. (1971). The role oflift in propelling the swimmer. In Selected
Topics on Biomechanics: Proc. CIC Symp. on Biomechanics, (J. M. Cooper, ed.), The Athletic
Institute, Chicago, pp. 179-188.
Burns, B. H. and Young, R. H. (1951). Results of surgery in sciatica and low back pain. Lancet
260(1):245-249, (Correction) 358.
Counsilman, J. E. (1971). The application of Bernoulli's principle to human propulsion in water.
In Proc. First Intern. Symp. on Biomechanics in Swimming, Waterpolo, and Diving, (L. Lewille
and J. P. Clarys, eds.), Univ. Libre de Bruxelles Lab. de L'elTort, Brussels.
Cyriax, J. (1978). Orthopedic Medicine, Vol. 1, Diagnosis of Soft Tissue Lesions, 7th Ed. Bailliere
Tindall, London.
Gray, H. (1973). Anatomy, A classic which has been revised and expanded by many authors. 35th
British ed. (R. Warwick and P. L. Williams, eds.), W. B. Saunders, Philadelphia, PA.
Greenwood, D. T. (1965). Principles of Dynamics, Prentice-Hall, Englewood ClilTs, N. J.
Hay, J. (1978). The Biomechanics of Sports Techniques, Prentice-Hall, Englewood ClilTs, N. J.
Kellgren, J. H. (1938). Observations on referred pain arising from muscle. Clin. Sci. 3:175-190.
See also Clin. Sci. 4:35-46.
Lewis, T. (1942). Pain, MacMillan, N. Y.
Nachemson, A. and Elfstrom, G. (1970). Intravital dynamic pressure measurements in lumbar
discs. A study of common movements, maneuvers, and exercises, Scand. J. Rehab. M ed. Suppl.
1, 1-40. See also author's article in Perspectives in Biomedical Engineering, (R. M. Kenedi,
ed.), Proc. of a Symposium. University Park Press, Baltimore, 1973, pp. 111-119.
Panjabi, M. M. and White, A. A. III. (1980). Spinal mechanics. In Perspectives in Biomechanics,
Vol. 1, Part B, (H. Reul, D. N. Ghista and G. Rau, eds.), Harwood Academic Pub., New
York, pp. 617-682.
Paul, J. P. (1970). The elTect of walking speed on the force actions transmitted at the hip and knee
joints. Proc. Roy. Soc. Med. 63(2):200-202.
Schultz, A. B. (1986). Loads on the human lumbar spine. Mech. Eng. 108:36-41.
Williams, M. and Lissner, H. R. (1977). Biomechanics of Human Motion. 2nd Ed. (B. LeVeau, ed.),
Saunders, Philadelphia, PA.
Yih, C. S. (1969,1989). Fluid Mechanics, McGraw-Hill, New York (1969). West River Press, Ann
Arbor, Michigan (1989).
CHAPTER 2

Segmental Movement and Vibrations

2.1 Introduction

In the preceding chapter we described motion in the Newtonian form. An

animal is considered to be a collection of particles, and particle movement is
expressed in terms of displacement, velocity, acceleration, external forces, and
forces of interaction between particles. In application to biomechanics, one
finds that the terms FIJ in Eq. (1.1:7), that describe the mutual interaction
between particles, are the most troublesome. For example, in the analysis of
human locomotion, we must know the forces in all the muscles of the legs.
But the human musculoskeletal system is highly redundant and the deter-
mination of the forces in the muscles is one of the most difficult problems in
biomechanics (see Sec. 2.8 infra). Hence there is a need for a method that does
not require such detailed information. The method of Joseph Louis Lagrange
(1736-1813) offers such an alternative in terms of work and energy. If the
kinetic and potential energies are known as functions of the generalized
coordinates and their derivatives with respect to time, and if the work done
by the external forces can be computed when a generalized coordinate changes,
then the equations of motion can be written down.
This beautiful method is very useful in biomechanics, especially in dealing
with problems of locomotion and vibrations of internal and external organs.
We will introduce this method as an alternative to the Newtonian approach
(Secs. 2.2-2.5). Then we will discuss the normal modes of vibration. The
normal modes of vibration are the most useful generalized coordinates, with
which the equations of motion can be decoupled in general (Secs. 2.6-2.7).
Equations describing systems with damping and fluid dynamic forces and
the possibility of decoupling these equations are discussed in Secs. 2.10 and
2.11.

29
30 2 Segmental Movement and Vibrations

2.2 Examples of Simple Vibration Systems

Example 1. Figure 2.2:1. A mass connected to the ground by a spring and a

damper. The mass has one degree of freedom: vertical displacement x from a
position of equilibrium x = O. The tension in the spring is linearly proportional
to x. The resistance of the damper is linearly proportional to the velocity X.
Thus the spring and damper offer a restoring force Kx + {3x, where K is the
spring constant, {3 is the damping constant, and dot over x signifies a differentia-
tion of x with respect to time. Following Eq. (1.1:7) the equation of motion of
the particle, with an external force F(e) acting on it, is
mx + Kx + {3x = F(e). (1)

Example 2. Figure 2.2:2. Several masses connected by linear springs and

dampers, all confined to move in one direction. An inertial frame of reference is
erected on the ground which is stationary. The mass mi is located at Xi. The
force acting on the mass ml due to the motion of the masses ml , m2 is
Fl2 = -KI(X I - x 2 ) - {31(:X I - x2 )·
That acting on m2 due to the motion ofml' m2, m3 is
F21 + F23 = -K I (X 2 - Xl) - {31(X 2 - Xl) - K 2(X 2 - x 3) - {32(X2 - X3)·
That acting on m3 due to the motion of m2, m3 and the stationary ground is
F32 + F30 = -K 2(X 3 - x 2) - {32(X 3 - X2) - K3 X 3 - {3X3·
These forces can be represented in the form

(i = 1,2,3)

with the appropriate choice of the constants kii' {3ij. Hence, on substituting

FIGURE 2.2:1 A mass-spring-damper system.

The mass is supported on ground by a spring on
the left, and by a damper on the right.
2.2 Examples of Simple Vibration Systems 31

FIGURE 2.2:2 Several masses con-

nected by springs and dampers.

FIGURE 2.2:3 A freely moving body.

into Eq. (1.1:7), we obtain the equations of motion

3 3
miXi + L kijXj + L PijXj = Fl
j=l j=l
e ), (i = 1,2,3). (2)

Example 3. Figure 2.2:3, representing a free-flying astronaut. In this case,

the inertial force depends on the acceleration of the body, whereas the elastic
and damping forces depend only on the deformation of the body. Hence it is
32 2 Segmental Movement and Vibrations

useful to introduce a local frame of reference attached to the body, with origin
located at the center of mass, to describe the elastic deformation. Let us
consider again a unidirectional motion without rotation. Let Xi be the coordi-
nate of particle mi referred to an inertial frame of reference, Xi that referred to
the center of mass, and Xo be the location of the origin of the moving frame
of reference (center of mass). Then
Xi=X o +x i (3)
and the acceleration in the inertial frame of reference is
Xi =Xo + Xi· (4)
On the other hand, the spring and damper forces depend only on Xi. Hence
the equations of motion are

C~ mi) Xo = i~ F?) (5a)

mi(XO + Xi) + L kijXj + L [3ijXj = F?), (i=1,2,3). (5b)

j j

The solution of these equations will be discussed in the following sections.

An application is suggested in Problem 2.1. Literature on modeling human
bodies is cited in Chapter 11.

2.3 Strain Energy and the Properties of the

Influence Coefficients

Consider an elastic body obeying Hooke's law. Let it be rigidly supported in

a fixed space and subjected to a set of forces Ql' Qz, ... , Qn acting at points
1,2, ... , n, respectively. According to Hooke's law, the deflection qi at point i
in the direction of the force Qi is linearly proportional to the forces Q 1, ... , Qn,
and vice versa:
qi = Cil Ql + CizQz + ... + CinQn, (i = 1,2, ... ,n) (1)
Qi = kil ql + kiZqz + ... + kinqn, (i = 1,2, ... ,n). (2)
The constants of proportionality cij , k ij are independent of the forces and
displacements. cij are the flexibility influence coefficients. kij are the stiffness
influence coefficients. The physical meaning of kij is the force that is required
to act at the point i due to a unit deflection at the point j while all points 1,
... , n other thanj are held fixed. In the case of a single degree of freedom, the
stiffness influence coefficient is the familiar spring constant. The physical
meaning of the flexibility constant cij is the deflection at i due to a unit force
acting atj.
In the following, we shall show that the constants kij and cij have very
special properties.
2.3 Strain Energy and the Properties of the Influence Coefficients 33

The linear relationships (1) or (2) imply that there exists a unique unstressed
state of the body to which the body returns whenever all the external forces
are removed. They imply that the principle of superposition applies, and that
the total work done by a set of forces is independent of the order in which the
forces are applied. In particular, if we slowly apply all the forces Ql' ... , Qn
together, beginning with zero and ending with full values, always keeping their
ratios constant, then the displacements q 1, ... , qn will also increase slowly in
constant ratios. The work done by each force is !Qiqi' and the total work done
by the system of forces is

This work is stored as strain energy in the elastic body. Now, there is a
well-known thermodynamic argument (see Fung, 1965, pp. 351-352) that
states that if the material is stable (i.e., if the body will return to its natural,
unstressed state when all the loads are removed), then the strain energy must
be positive definite, i.e., W must be positive, and can be zero only if all the q's
and Q's vanish. The conditions of positive definiteness of the quadratic forms
in Eq. (3) are well known (see Fung, 1965, pp. 29, 30) and are
k ii > 0, (i not summed), (4a)

Ik ii kijl
kjj > 0, (i,j = 1,2, ... , n), (4b)
kji
k ii kij kim
kji kjj kjm >0, (i,j,m = 1,2, ... ,n), (4c)
k mi k mj kmm

That is, all the principal minors, including the determinant of the full kij matrix,
must be greater than zero. Similar conditions hold for cij's.
One of the most important properties of the influence coefficients is that
they are symmetric:
(5)
In other words, the displacement at point i due to a unit force acting at another
pointj is equal to the displacement atj due to a unit force acting at i, provided
that the displacement and force correspond, i.e., they are positive in the same
direction at each point. The proof is simple: Consider two forces Ql and Q2'
When the forces are applied in the order Ql' Q2 the work done by the forces is
W = !(c ll Qi + C22 QD + C12 Ql Q2'
When the order of application of the forces is reversed, the work done is
W' = !(C22Q~ + C ll QD + C21 Ql Q2'
But W = W'forarbitrary Ql' Q2' Hencec12 = C21' and the theorem is proved.
34 2 Segmental Movement and Vibrations

FIGURE 2.3:1 Generalized forces.

The concept of corresponding forces and displacements can be generalized

to include torques and angles (Fig. 2.3:1). If M j is a couple acting at a pointj
and OJ is the rotation of the material at point j in the direction of the couple,
then M/Ji is the work done by the couple M j , and M j and OJ are said to be
corresponding to each other. The couple M j may be called a generalized force
at point j, and OJ is the corresponding generalized displacement. We can then
denote M j , OJ by Qj, qj and show that Eqs. (1)-(5) hold for generalized forces
and displacements.
With these concepts and notations, we can write the equations of motion
of a set of masses embedded in an elastic body. The simplest way is to apply
D' Alembert's principle, which states that the particles can be considered to be
in a state of equilibrium if the inertial forces (mass x acceleration in an inertial
frame of reference) are applied in reversed direction on the particles. Thus, if
the body is supported by stationary supports, then the force Qi applied on the
mass mi is - m;iii + Fie). Here U i is the displacement of the particle and Fie) is
the external force acting on it. Both U i and Fie) are vectors. We may treat each
component of Ui and F?) as a separate variable. This set of forces Qi will keep
the body in equilibrium with displacements Ui. Thus, using a dot over a
variable to indicate the differention of that variable with respect to time, Eq.
(2) becomes
n
..
miui + "k pte)
L.. ijUj = i , (i = 1,2, ... ,n). (6)
j=l

This is the generalization of Eq. (2.2:2) to a three-dimensional elastic body

which is supported at fixed points in an inertial frame of reference.
If the body is free to translate and rotate in space while it deforms elastically,
then we can use a moving frame of reference x, y, z which is attached to the
2.4 Generalized Coordinates 35

body with the origin located at the center of mass. The force-deformation
relationships remain linear as in Eqs. (1) and (2). To apply D'Alembert's
principle, we compute the acceleration that is referred to an inertial frame of
reference. As we have discussed in Sec. 1.3, let r be the position vector of a
point in an inertial frame of reference, Ro be the position vector of the origin
of the moving coordinates, and p be the position vector of the point relative
to the moving coordinates. See Fig. 1.3:1. Then
r = Ro + p = Ro + xix + yiy + ziz (7)
r = Ro + P = Ro + xix + .viy + iiz + x(ixf + y(i,)" + z(iS, (8)
where ix, iy, iz are unit vectors, and (if = 0) x i, 0) being the angular velocity
of the moving frame. Equation (8) may be written as
r=Ro+Pr+O) x p. (9)
This equation defines Pr = xix + .viy + iiz as the relative velocity of the point.
A further differentiation yields, after simplification,
i' = ito + 0) x 0) x P + cO x P + iir + 20) X Pro (10)
Treating the three components as the inertial force - mi' resolved along the
moving coordinates x, y, z as three forces, using D' Alembert's principle, and
denoting the three components of the vector ( ) by ( )i' we obtain the equations
of motion
n
mi (R0 + 0) x 0) x P + 0) x p + Pr + 20)
oO 0 00

X Pr i +
0 ) ~
L..J
kijU = pIe)
i , (11)
j
j=l

(i = 1, ... , n). Equation (11) describes the deformation ofthe body. The transla-
tion and rotation of the body, Ro(t) and w(t), respectively, are described by
Eqs. (1.3:5) and (1.3:25).
In Sec. 2.2 we considered dampers. If the body is linearly viscoelastic we
can add damping terms so that Eq. (6) is generalized to
n n
miui +L kijuj +L PiA = Fie). (12)
j=l j=l

Damping, however, can be much more complex than this. It caq be aero-
dynamic in origin (cf. Chs. 3,4), or nonlinearly viscoelastic, and not necessarily
stabilizing. See Sec. 2.10.

2.4 Generalized Coordinates

So far we have written the equations of motion in terms of the Cartesian

coordinates of particles. We note, however, that there are occasions in which
the movement of the particles in a body can be described by quantities other
than the Cartesian coordinates. For example, the movement of all the points
36 2 Segmental Movement and Vibrations

on a rigid rod can be described by the displacement of one point on the rod
and the rotation of the rod about that point. In general, a set of quantities q 1,
q2' q3"'" qn are called the generalized coordinates of a system if they have the
following properties: (1) The displacement of the system is described com-
pletely by the q's. (2) They are independent, so that one qi can be varied while
the remaining q/s are held constant. Thus each qi describes a degree offreedom
of the system.
Let Xi' Yi' Zi' the Cartesian coordinates of a particle of number i in an inertial
frame of reference, be related to the q's by relations of the form:
(i = 1, ... ,N) (1)
which do not contain time t explicitly. By differentation, we have the velocities

(2)

Now, let us express the total kinetic energy of the system

K = ~ ktl mk(xf + yf + #) (3)

in terms of the generalized coordinates. Here mk is the mass of the kth particle.
On substituting Eq. (2) into Eq. (3), we obtain

K = ttl m{ Ctl ~:: 4jY+ Ct ~~: 4jY+ Ctl ~:: 4jYl (4)

This can be written as

1 n n
L L mij4;4j,
K = -
2 i=l j=l (5)

in which the coefficients mij are functions ofthe q's:

m .. = m .. =
lJ J' k=lf mk (OXk oXk + OYk OYk
Oqi oqj Oqi oqj
+ OZk OZk)
Oqi oqj .
(6)

As the system moves through small displacements dqi' an increment of

work will be done by the forces acting on the system. This work can be written
in the form

(7)

The Q/s are called the generalized forces corresponding to the generalized
coordinates qi'
Example 1. A lower leg is hinged at the knee (x = 0) and acted on by forces
perpendicular to its longitudinal axis at an intensity of p(x) per unit length (as
in swimming), see Fig. 2.4:1. With the angle () chosen as a generalized coordi-
2.4 Generalized Coordinates 37

FIGURE 2.4:1 Forces acting on the arms and legs in swimming.

.. x

FIGURE 2.4:2 A simply-supported beam.

nate, the work done on the leg due to a small change of angle () is

dW = (LL p(x)x dX) d() = Q d(). (8)

The quantity enclosed in the parenthesis is the generalized force corresponding

to ().
Example 2. A simply supported beam vibrates in its first mode. The dis-
placement is a half sine wave as shown in Fig. 2.4:2,

y = a (sin 7)sinwt. (9)

A generalized coordinate for this mode is a quantity that specifies the displace-
ment, e.g., the displacement at x = L/2. Thus
ql = a sin wt y = ql sin(nx/L). (10)
If the beam is acted on by a distributed lateral load of magnitude p(x) per unit
38 2 Segmental Movement and Vibrations

length, then the work done during displacement dql is

dW = IL p(x) dx dy = (IL p(x) sin ~ dX) dql = Ql dql' (11)

Hence the generalized force corresponding to the generalized coordinate ql

is the quantity enclosed in the parentheses:

Ql = Jo p(x)sin-dx.
L nx
L
(12)

2.5 Lagrange's Equations

The kinetic energy K is, according to Eq. (2.4:5), a homogeneous quadratic

function ofthe q's. According to Euler's theorem for homogeneous functions,
K must be of the form:
oK
n
L ~4j=2K.
j=l uqj
(1)

On the other hand, the coefficients mij in Eq. (2.4:6) are functions of the q's.
Hence the rate of change of K, as a function of qj and qj' can be obtained by
following the general rules of differentiation:

. = L (OK
K
OK)
~ijj + ~qj
j uqj uqj

(2)

On substituting Eq. (1) into the first term on the right-hand side of (2), we
obtain

i.e.,

K = L (!!.- oK _ OK)q. (3)

j dt oqj oqj J'

Now, the kinetic energy is one form of energy. For a biological system, other
forms of energy are involved, such as the gravitational potential G, the internal
energy U, and the chemical energy C. According to the first law of thermo-
dynamics, the energy of a system can be changed by absorption of heat, H,
and by doing work on the system. The rate of change of total energy must be
equal to the sum of the rates of heat input, fl, and work done, W:
(4)
2.5 Lagrange's Equations 39

A part of the internal energy U is the strain energy discussed in Sec. 2.3. It
is a quadratic function of the generalized coordinates q1, q2"'" qn, and is, in
general, a function of temperature. If the temperature remains constant then
• n au
u= L -a qj.
j=1 qj
(5)

The rate at which work is done by the forces acting on the system is,
according to Eq. (2.4:7),
(6)

Substituting Eqs. (3), (5), and (6) into Eq. (4), we obtain

~ (d aK aK
L... - - . - - + -au - ) .
Qj qj = H - G - C.
. . (7)
j=1 dt aqj aqj aqj
In the special case when
(8)
then

t (~aK _
j=1 dt aqj
aK
aqj
+ au -Qj)qj=O'
aqj
(9)

But the q/s are independent variables that can assume arbitrary values. In
particular, we may set q1 =F 0 while all other q/s vanish. Then the coefficient
of q1 must vanish. Similarly we can show that the quantity in the parentheses
of every term of Eq. (9) must vanish. Hence
d aK aK au
----+-=Q. (j = 1,2, ... ,n). (10)
dt aqj aqj aqj J'

These are Lagrange's equations.

If the generalized forces are partial derivatives of a function V of q1, q2,

av
Qj= --a' (11)
qj
then the forces are said to be conservative and V is called the potential energy.
Since U and V are independent of qj, au /aqj = av/aqj = O. Then Lagrange's
equations may be written

(12)

in which
L=K+U-V (13)
is called the Lagrangian function of the system.
40 2 Segmental Movement and Vibrations

FIGURE 2.5:1 A simple example.

The aerodynamic or hydrodynamic forces acting on an animal moving in

a fluid are nonconservative. They cannot be derived from a potential by
differentiation. Hence, for problems involving these forces, it is better to use
Eq. (10). (For simple problems, there is no particular advantage in using
Lagrange's equations; advantages often show up in complex problems.)
Bioengineers are constantly faced with the question of how to approximate
real systems. Lagrange's equations tell us that in mathematical modeling one
should concentrate on approximating the energies. The merit of a scheme may
be judged by how well the energies are approximated.

Example 1. Consider a weightless beam with a point mass m at its midspan

(Fig. 2.5:1). A force P(t) acts on the mass. The stiffness influence coefficient for
the load at midspan due to unit deflection is k. Then, if y is the deflection at
midspan, we have
Q = P(t),
my + ky = P(t).
Example 2. Consider a simply-supported beam of uniform material and
uniform cross section (Fig. 2.5:1). Since the deflection curve w(x, t) of the beam
is continuous and satisfies the boundary conditions w = iJ2w/ox 2 = 0 at x = 0,
L, it can be developed into a Fourier sine series:
nnx
I
00
w= ansin-.
n=1 L
The coefficients an may be considered the generalized coordinates qi = ai'
(i = 1,2,3, ... ). When the beam vibrates, wand qi are functions of time. Let
m be the mass per unit length of the beam. Show that

K
1
= -2
fL mw 2 dx
mL
= -4 I a; (t),
00

o n=1
2.5 Lagrange's Equations 41

1 (L (02W)2 (E1 = rigidity = const)

U="2JoE1ox2 dx

If a lateral load of magnitude P(x, t) per unit length acts on the beam, the
generalized forces can be calculated as follows. Let the generalized coordinate
an be given a virtual displacement (jan' The beam configuration undergoes a
virtual displacement (jw(x) = (jan sin(nnx/L). The work done by the external
load is

Qn(ja n = f LP(x,t)(jw(x)dx = fL P(x,t)(jansin-dx.

nnx
o 0 L

But (jan is arbitrary, hence,

Qn = f
L

o
nnx
P(x,t)sin-dx.
L
The equations of motion are

(n = 1,2,3, ... ).

Example 3. Consider a cantilever beam of uniform cross section (Fig. 2.5:2).

The deflection of the beam can be expanded into a series

L
00

w(x, t) = qn(t)f,,(x)
n=l

where fn(x) is the nth mode of undamped free vibration of the beam. The
functions fn(x) are orthogonal and can be normalized so that

(L fm(x)f,,(x)dx = {1, when m = n,

J
0 0, when m #- n.

(a) (b) (c) (d)

FIGURE 2.5:2 A cantilever beam of uniform cross section and its first three modes of
natural vibration.
42 2 Segmental Movement and Vibrations

Show that
mL
L q;,
00

K=-2
n=O

Qn = SoL P(X, t)f,,(X) dx,

in which Kn are the solutions of the equation cos Kn cosh Kn + 1 = o. Therefore,
under a lateral load P(x, t) the equations of motion are

mLiin + EILK:qn =
Jor P(x, t)f,,(x) dx (
L (n = 1,2, ... ).

Example 4. A simple model of posture control. A man is represented by a

mass m, located at the center of gravity at height L, and a centroidal moment
of inertia I (Fig. 2.5:3). The moment produced at the ankles is M, the ground
reaction forces are Fx , Fy • Then, the motion of the center of mass (located at
x, y) and the rotation about it are governed by the equations:
my = Fy - mg, (14)
Since, according to Fig. 2.5:3,
x = Lcos(), y = Lsin(), (15)

----~~~~~------------~__ x

FIGURE 2.5:3 A simple inverted-

pendulum model of posture
control.
2.5 Lagrange's Equations 43

we obtain by differentiation
x= -OL sin lJ - (PLcoslJ,
(16)
ji = OL cos lJ - 92 L sin lJ.
These are five equations for the five unknowns x, ji, 0, F", Fy •
Generalized Coordinate, lJ. The kinetic energy is
K = t(mx 2 + my2 + 192) = t(mL 2 + 1)92. (17)
The potential energy is
v= mgLsin lJ. (18)
The virtual work relation is
<5W = Q<5lJ = M<5lJ, Q=M. (19)
Hence the Lagrangian equation is a single equation
0(mL2 + I) + mgLcoslJ = M. (20)
Note that in Lagrangian approach, the ground reactions F", Fy do not appear
in the final equation.
Feedback Control. Camana et al. (1977) considered four sensing modalities
for maintaining an erect posture:
(21)
where lJ and 9 are ankle angle and ankle angle rates, as shown in Fig. 2.5:3,
lJoe is an approximation to the angular rate information sensed by the semicir-
cular canals, and lJo is a similar approximation to angular position as sensed
by the otolith system. For a normal person, one may also wish to add visual
sensing.

Example 5. A simple model of gait (Fig. 2.5:4). The entire mass of a simulated
biped is concentrated in a single rigid body. Let r, ifJl' ifJ2 be the three generalized
coordinates, and let J, F, M be the moment of inertia, tangential leg force, and
hip moment, respectively, all normalized to the system mass, m. For simplicity,
take the ankle moment as zero. Then the Lagrangian equations of motion are
.. '2 '2
r + LifJ2 sin(ifJl - ifJ2) - rifJl - LifJ], cos(ifJ 1 - ifJ2) + g cos ifJl = F
2···· "2
+ rLifJ], COS(ifJl - ifJ2) + 2rrifJl + rLifJ2 sin(<pl - ifJ2) - gr sin ifJl =
r ifJl - M

Lfsin(ifJ - ifJ2) + Lr~l COS(ifJl - ifJ2) + (J + [2)~2 + 2Lf~1 cos(ifJ - ifJ2)

- Lr~fsin(ifJl - ifJ2) - Lg sin ifJ2 = M.
McGhee (1980) discussed gait and posture on the basis of these equations,
and a comparison of these equations with those obtained by free-body ap-
proach. The latter requires more equations and has to bring the constraint
44 2 Segmental Movement and Vibrations

/
z

,
\
,\ \
1\ \
\1 \
'\ \,
\\ ~
~\
/ \
\ ,,
'
/

/ \
/ \ I
I \ '
/ ,,' .. y
FIGURE 2.5:4 A simplified model of biped locomotion. From McGhee (1980), repro-
duced by permission.

forces such as the ground reactions into account; but each individual equation
appears simpler.

2.6 Normal Modes of Vibration

The most commonly used generalized coordinates for an elastic body are the
vibration modes of the body. By using the vibration modes, the equations of
motion can often be decoupled into a set of independent equations, (see Sec.
2.7). Then the motion in each degree of freedom can be solved separately. This
efficient procedure is the principal reason why people are so interested in
normal modes. The basic theory is presented below.
The equations of free vibration of an elastic body (without damping) is
given by Eq. (2.3:6) with Fl e ) = 0 (without external forces). For simplicity of
notation and manipulation, we shall consider it as a matrix equation. Define
2.6 Normal Modes of Vibration 45

the column and square matrices

o

~J
X=
{ .
I
uU 2 } , K = ~Kll
K21

Un Knl o
(1)
Then the equation of free vibration is
MX + KX = O. (2)
The kinetic and potential energies, given by Eqs. (2.2:1) and (2.3:3), respec-
tively, may be expressed in the following form:
K = !XTMX, U = !XTKX. (3)
The superscript T means the transpose of the matrix. The kinetic energy K is
positive definite (Sec. 2.3). The strain energy U is also positive definite if the
body cannot move as a rigid body; whereas U = 0 in the rigid-body mode.
Hence M and K are subjected to conditions listed in Eq. (2.3:4). In particular,
the determinant of M does not vanish.
Since Eq. (2) is a set of linear differential equations with constant coeffi-
cients, it is expected that the solution will be an exponential function of time,
in the form of
(4)
where Xo is a column of real or complex constants. The real and imaginary
parts of (4) both represent simple harmonic oscillations. The problem is to
determine the circular frequency 0) and the corresponding amplitude function
Xo·
On substituting Eq. (4) into Eq. (2), dropping the subscript 0, and writing
A for 0)2, we obtain
KX - AMX = 0, (5)
Now, the set of linear simultaneous equations (5) can have a nontrivial
solution Xv "# 0 if and only if the determinant of the coefficients vanishes:
det IK - AMI = O. (6)
Since the degree of the polynomial
det IK - AM I = (- 1)n An det IM I + ... + det IK I
is exactly n, M being nonsingular, there exist exactly n roots. The roots are
called eigenvalues, and the corresponding solutions X are called eigenvectors.
Let Al be an eigenvalue and let Xl "# 0 be a corresponding eigenvector, so that
(7)
46 2 Segmental Movement and Vibrations

Xl might be complex valued. Let Xl be the complex conjugate of Xl. Pre-

multiply Eq. (7) with XI. we obtain

(8)

But M and K are real, symmetric matrices. Hence the products XI(MXd and
XI(KX l ) are real numbers; the first is positive, the second is non~ative. It
follows that Al is a real number and is nonnegative, and Oh = y' Al is a real
number, called an eigenfrequency. W l = 0 if the body moves as a rigid body,
W l "# 0 if the body deforms during vibration without rigid-body motion.
Hence we obtain the important result: all the eigenfrequencies of free
vibration of a linear elastic solid are real numbers.
When Al is real valued, it is clear that the solution Xl of Eq. (7) can be
normalized to be a real-valued column matrix. Hence we conclude that all the
eigenvectors can be normalized to be real valued. In other words, all free
vibration modes of a linear elastic body are real, i.e., all masses move in phase
in each mode.
Now we can prove the following:
If the equation of free vibration of a linear elastic solid, Eq. (5), has two
unequal eigenvalues Al and A2 , then any two eigenvectors Xl and X 2, corre-
sponding to Al and A2' respectively, are orthogonal with respect to M and K, i.e.,

(X l ,MX 2) = 0, (9)
(X l ,KX2) = O. (10)

Here the notation (A, B) denotes the scalar product of two vectors A, B; i.e.,
ATB when A, B are represented by matrices.
Proof By assumption, KX l = AlMX l , KX 2 = A2MX 2. According to what
was said earlier, Al and A2 are real numbers. Hence,

(KX l ,X 2) = (A l MX l ,X 2), (lla)

(Xl' KX 2) = (Xl,A2MX2) = A2(X l ,MX 2)· (llb)

But (KX l ,X 2) = (Xl' KX 2), (MX l ,X 2) = (Xl, MX 2) because K and Mare

symmetric. Therefore, using these equations and subtracting (llb) from (lla),
one obtains (Al - A2)(X l , MX 2) = 0, whence Eq. (9) follows, because, by
assumption, Al - A2 "# O. A glance at Eq. (11) then proves Eq. (10). Q.E.D.

Combining the results stated above, one sees that if the determinantal
equation (6) has single roots only, then we can assert that:
The free vibration equation of a linear elastic solid, Eq. (5), has n real
eigenvalues Al , A2' ... ' An and n corresponding eigenvectors Xl> X 2,···, Xn called
the normal modes, satisfying the relations

(v=I, ... ,n),

 2.7 Decoupling of Equations of Motion 47

which are orthogonal by pairs with respect to M and can be normalized so that
(/1, v = 1, ... , n), (12)
where bl'v = 1 if /1 = v; bl'v = 0 if /1 "# v.
Extending this result, it can be shown that if Av is a k-fold multiple root of
the characteristic Eq. (6), then to Av there belongs exactly k of the eigenvectors
Xl' ... , Xn. The proof can be found in Bellman (1970).
The orthonormal character of the normal modes of vibration exhibited by
Eq. (12) is the basis on which the importance of normal modes rests. With Eq.
(12), we see that an arbitrary motion of the body, represented by a vector X
in an n-dimensional space, can be represented in the form
n
X = I qvXV'
v=l
(13)

where qv are constants:

(X, MX v)
qv = (X MX) = (X, MXv)· (14)
V' v

This is easily proven by premultiplying Eq. (13) by X!M and using Eq. (12).
Because ofEq. (13), we see that Xv can serve as the basis and qv the generalized
coordinates. Since Xv are normal modes, qv are called normal coordinates.
Physically, we say that any displaced configuration of a body can be repre-
sented by a linear combination of the normal modes of free vibration of a
linear elastic body of the same geometry and dimensions.

2.7 Decoupling of Equations of Motion

Now we shall show that the equations of motion become particularly simple
when expressed in terms of normal coordinates. Let Xl' ... , Xn be the normal
modes and AI' ... , An be the corresponding eigenvalues of the free vibration
equation:
Mi + KX = 0 or KX = AMX. (1)
We have
KX v = AvMXV' (v = 1, ... ,n), (2)
(XV' MXI') = bl'V' (/1, v = 1, ... ,n), (3)
(XV' KXI') = ).l' bl'V' (/1, v = 1, ... , n). (4)
With these normal modes we can build a square matrix <1>:
<I> = (X 1 ,X 2 , ..• ,Xn ). (5)
The first column of <I> is the column matrix Xl, the second column of <I> is
the column matrix X 2 , etc. It can be easily verified, on account of Eqs. (3) and
(4), that if all the eigenvalues AI, ... , An are different, i.e., none of them is a
48 2 Segmental Movement and Vibrations

multiple root of the characteristic equation (2.6:6), then

C(lTMC(l = I = unit matrix, (6)

Cl>TKCI>
[l,
~ !
0
A2
0
0
...
...
~]d (7)

0 0 An
C(lT is the transpose of C(l. Since the vectors X. are linearly independent, the
matrix C(l is nonsingular. Hence C(l-l exists, and according to Eq. (6) this is
equal to
(8)
The normal coordinates for an arbitrary vector X, given by Eqs. (2.6:13)
and (2.6:14), can thus be written in matrix form
X=~~ 00
(10)
where q is a column matrix whose components are the normal coordinates
ql' q2' ... , qn'
With these ingredients we can now simplify the equations of motion of a
linear elastic system subjected to a set of forces F = (Fl' F2 , ••• ,Fn):
Mt + KX = F. (11)
Premultiplying both sides with C(lT and inserting C(lC(l-1 = I we obtain
C(lTMC(lC(l-l t + C(lTKC(lC(l-l X = C(lTF. (12)
Using Eqs. (6), (7) and (10), and defining the column matrix ofthe generalized
forces Q as
Q = C(lTF (13)
we obtain from Eq. (12)
ii + (C(lTKC(l)q = Q (14)
or, using (7),
ii + Aq = Q. (15)
This is the equation of motion in its simplest form.
To emphasize the difference between Eqs. (11) and (15), we write them in
extenso. Equations (11) are
m l ii l + kll Ul + k12 U 2 + ... + klnun = Fl (t),

m2 ii 2 + k21 Ul + k 22 U 2 + ... + k 2n un = F2 (t),

(16)
2.8 Muscle Forces 49

They are simplified into

iil + WiQl = Ql'
ii2 + W~Q2 = Q2'
(17)

iin + w;qn = Qn·

In these equations, we have written w;
for Av to emphasize the physical
meaning of the constant w., which is the circular frequency of free vibration
of the vth normal mode.
It can be shown that each generalized force Q., defined formally in Eq. (13),
has a simple physical interpretation. It is numerically equal to the work done
by the external forces acting on the body when the body deforms in the vth
mode, which is so normalized in amplitude that
(18)
where u 1 , u 2 , ••• , Un are the displacements of the particles m 1 , m2 , ••• , mn in
the vth normal mode.
We see that if the Q:s are functions oftime only (i.e., if they are independent
of the elastic displacements), Eq. (15) is completely decoupled. Each normal
coordinate varies with time as a single mass-spring oscillator of one degree
of freedom, independent of all other modes. Such is the simplicity introduced
by the use of normal coordinates.
Use of normal modes also relieves the need to account for any constraints
imposed on the body. For example, if we want to study the response of the
spine to shock loading, the end conditions at the neck and the sacrum must
be satisfied. If we study the stresses in the leg in certain maneuver, the
constraint of the hip joint must be imposed. If normal modes are used as
generalized coordinates, the boundary conditions are satisfied by every mode
and no further attention to the constraints is needed.
Fortunately, in most practical problems the deformation of a continuous
body can usually be described to a sufficient degree of accuracy by using only
a few normal modes associated with the lower frequencies of free vibration.
In such cases the continuous body is approximated by one possessing only a
few degrees of freedom.
Unfortunately, if the system were damped, complete decoupling is in general
not possible. See Secs. 2.10 and 2.11.

2.8 Muscle Forces

The high degree of redundancy of animal body structure is the main difficulty
for a detailed analysis of the musculoskeletal system. By degree of redundancy
is meant the excess of the number of unknowns over the total number of
50 2 Segmental Movement and Vibrations

FIGURE 2.8:1 Simple trusses illu-

strating the concept of degree of
redundancy.

available independent equations of equilibrium (or motion) and constraints.

The idea can be illustrated by a simple example. Fig. 2.8:1(a) shows a simple
truss made of steel members that can transmit tension and compression. If we
wish to know the force in each member when a weight is hung on the truss,
we have to determine eight unknown forces. If we make various free-body
diagrams and write down the equations of equilibrium, we find that there are
eight linearly independent equations available. This is exactly the right num-
ber of equations to determine the eight unknowns, so the structure is said to
be statically determinate.
In the truss shown in Fig. 2.8:1 (b), the engineer has added two more
members. Under an external load, there are now ten unknown forces while
the total number of linearly independent equations of equilibrium remains at
eight. Thus the forces cannot be determined by the equations of equilibrium
alone, and the structure is said to be statically indeterminate; and the degree
of redundancy is 10 - 8 = 2.
Now, practically everything in our body is statically indeterminate. You
may compare our finger with the truss shown in Fig. 2.8:1(b); and you will
find that our finger has a much higher degree of redundancy.
A statically determinate structure is usually very efficient in transmitting
2.8 Muscle Forces 51

load from one place to another, and economical in the use of materials, (e.g.,
of low weight, or low cost of construction). But every part has to function.
Breaking one member of the truss shown in Fig. 2.8:1(a) and the whole
structure will collapse.
A statically indeterminate structure may achieve some degree of safety
against failure. For example, two members of the truss shown in Fig. 2.8:1(b)
may be broken without causing the truss to collapse. Engineering structures
which must avoid catastrophic failures are often built with high degrees of
redundancy. An example is the airplane. In this sense, animal bodies are
beautiful fail-safe designs.
How can the forces in an statically indeterminate structure be determined?
In the case of the truss shown in Fig. 2.8: 1(b), the solution calls for an
application of the theory of elasticity. The principle is that the structure must
remain an integral one, so the deformation of every member of the structure
can be consistent with each other as a whole. The deformation of each member
is related to the force in the member according to the law of elasticity. One
analytical procedure introduces imagined cuts in a number of members to
make the structure statically determinate, (e.g., cut the members AF, BE, in
Fig. 2.8: 1(b)), then applies forces on the cut surfaces as in Fig. 2.8:1 (c), computes
the displacements of the joints BCD E F as a functon of these forces, and
finally determines these forces by the fact that the lengths of the members AF
and BE must be changed by exactly the same amount as those caused by the
displacements of the joints B, E, F, and A that were computed in the preceding
step. This method is called the method of consistent deformation.
Similarly, in the musculoskeletal system of an animal the forces in the
muscles, tendons, ligaments, capsules, and joint contact surfaces can be deter-
mined according to the principle of consistent deformation. Each component
obeys its constitutive equation. The analysis is obviously more complex than
that for an inanimate structure because muscles can contract actively, and the
feedback control of the neuromuscular system must be considered.
Few rigorous analysis of muscle forces in the musculoskeletal system exists.
Most publications adopt some gross simplifying assumptions. Examples are:
(a) Functional grouping. Muscles with similar functions or common ana-
tomical insertions and orientations are grouped together and regarded as one.
(b) Optimization. Based on heuristic reasoning, it is hypothesized that
nature works in such a way that some quantities are minimized. Famous
successful examples are the Hamilton principle in dynamics, the minimum
strain energy principle in elasticity, Fermat's minimum time principle in
optics, Maupertius's principle of least action. So people have proposed the
minimum principle for muscle forces at joints. The mathematical problem
takes the form of minimizing
(1)
subjected to the conditions
gj(Xl>X 2 ,,, .,xn ) = 0, (j = 1,2, ... ,m) (2)
52 2 Segmental Movement and Vibrations

and
(i = 1,2, ... , n). (3)
Here Xi stands for unknown muscle and joint forces, 9i stands for equations
of motion or constraints. J is debated. The following have been proposed for
J, but none has been validated.
~ (muscle force) ~ (muscle forcet
~ Goint force) ~ Goint force)2
~ Goint moment) ~ (muscle stress)n
~ (muscle stress) ~ (muscle energy)
Experimental approach using myoelectricity is discussed by Schultz et al.
(1982) and Bean et al. (1988).

2.9 Segmental Movement and Vibrations

In the above, we presented several ways by which the equations of motion of
the musculoskeletal system can be written down, and some of the basic
properties of these equations. We showed the existence of vibration modes in
certain instances and the opportunity of greatly simplifying the equations of
motion by using the normal modes as generalized coordinates.
In the musculoskeletal system, segmental movement and vibrations, of
course, occur together. The major bones move essentially like rigid bodies,
but they respond to dynamic loads by propagating elastic waves and vibra-
tions. When one is interested in stresses in the bones, in their strength and
failure, in their repair and interface with prosthesis, the vibrational features
will come to the foreground.
For internal organs, the segmental movement is less evident. Yet the
response of every organ to dynamic loads can be examined by first looking
at the movement of the center of mass of the organ, and then analyzing the
deformation of the organ relative to the center of mass; e.g. Ex. 3 of Sec. 2.2.
For example, in considering possible injuries to the heart in a car crash, one
may first regard the heart as a rigid body in the chest, then analyze its stresses
and strains under the dynamic loading.

2.10 Systems with Damping and Fluid Dynamic Loads

If the material is viscoelastic or some of the forces are hydrodynamic or
aerodynamic in origin, the system of equations are more complex than those
discussed above. We question whether eigenvalues and normal modes exist,
whether decoupling of the equations of motion is possible or not. For sim-
plicity, we limit our discussion to linear systems.
2.10 Systems with Damping and Fluid Dynamic Loads 53

The most general constitutive equation of a linear viscoelastic solid is due

to Ludwig Boltzmann (1844-1906). Applying Boltzmann's formulation to the
system considered in Sec. 1.3, we may write the internal force acting on a
particle i located at point i by a displacement Uj located at pointj as (see, e.g.
Fung, 1965, 1981):

Flit) = It
-00
du/r)
Gij(t - -r)----;Jtd-r (1)

where Gij(t) are the relaxation functions. For solids with fading memory, which
include Maxwell, Voigt, and Kelvin models,
(2)
where V l , V 2 ••• Vn are the relaxation frequencies, K ij , /3ij ... Yij are spectral
constants.
If the system is subjected to fluid dynamic forces, (e.g. in swimming, in wind,
or with internal flow), the external force at the point i due to motion at point
j may be written as (Chapter 3, and Fung, 1969)

(e) It dui-r)d (3)

Flj (t) = -00 Aij(t - -r)---ar -r

where Aij(t) are aerodynamic influence functions. It is a special property of

aerodynamics that Aij is, in general, unequal to Aji; i.e. the aerodynamic
influence functions are unsymmetric.
When these expressions are inserted into Eq. (1.1:7), we obtain the equation
of motion of a mass number i in an n-mass system embedded in a viscoelastic
medium:

where Fie) is the external force other than the fluid dynamic forces. To avoid
writing integrals, let us use Laplace transformation. The Laplace transforma-
tion of a function u(t) is defined by multiplying u(t) by e-·t, integrating the
product from t = 0 to t = 00, and denoting the product by u(s). Thus,

u(s) = Loo e-·tu(t) dt. (5)

Taking Laplace transformation of Eq. (4), with the initial values of the dis-
placement ui(-r) and velocity ui(-r) equal to U iO and uiO respectively at t = 0, we
obtain

mis2 ui(s) + ±[-

j=l
+ /3ij_1_ + ... + Yij_1_J [suj(s) -
A;j(s)
s + Vl S + Vn
UjO ]

+ K ijUj
-(S) = F-(e)
i + misuiO +miu
.iO , (.,= 1, ... ,n.) (6)
54 2 Segmental Movement and Vibrations

These equations can be written as a matrix equation

s2MX + KX - sAX + _s_BX + ... + _s_rx = F (7)

S + Vi S + Vn

in which M is a diagonal matrix (m;), A, K, B ... rare (n x n) square matrices

(Aij), (KiJ, ({3ij), ... , (Yij) respectively, X is a column matrix (uJ, F is a column

matrix (F;<e) + misu iO + miu iO + f. [- Aij + {3ij _1_ + ...Jujo ). Compared

j=l s+v l
with the equation treated in Sees. 2.6 and 2.7, our new equation involves
additional matrices A, B, ... , r. .
Our question is whether there exists a transformation from (Ul> U2 , ••• , un)
to generalized coordinates (Z[l,"" Z[n) so that Eq. (7) is decoupled into a set of
independent equations of the form
(i = 1, ... , n) (8)
where hi(s), Qi(S) are functions of s, independent of the lfs. If this reduction is
possible, then each individual coordinate qi can be solved separately, and the
features of normal coordinates are obtained.
Suppose that this reduction of Eq. (7) to Eq. (8) is possible and that the
transformation from (ul , ... , ~) to (Z[l' ... , Z[n) is accomplished by a nonsingular
real-valued matrix «J) so that

(::) (::: ::: ::: :::) (~:)

.- .
. ,
. .
..
.,
.,
== «J)q. (9)

Un <fJnl <fJn2 <fJnn Z[n

Then, each column of «J) represents a deformation mode. These modes can be
orthonormalized by the Schmidt process (Sec. 2.6) so that
«J)TM«J) = I = identity matrix. (10)
When such a normalization is done, it is justified to call each column of «J) a
normal mode.
We shall now discuss the conditions under which normal modes exist for
a system described by Eq. (7).
On assuming that the reduction of Eq. (7) to Eq. (8) is possible and writing
Eq. (9) and its inverse as
X = «J)q, q = «J)-l X (11)

we premultiply Eq. (7) by «J)T, and using (10) and (11) to obtain
s2Iq + («J)TK<<J))q - S((«J)T A«J))Z[ = ... = «J)TG. (12)
In order that Eq. (12) be of the form of Eq. (8), it is necessary that
«J)TK«J) = diagonal, «J)T A«J) = diagonal, «J)TB«J) = diagonal. (13)
2.11 Sufficient Conditions for Decoupling Equations of System with Damping 55

Conversely, if such a <I> is available so that Eqs. (13) are satisfied, then (7) is
diagonalized into (8). Hence, the necessary and sufficient condition for the
decoupling of the equations of motion by a transformation of the type (11) is the
existence of a transformation matrix <I> which diagonalizes all the matrices M,
A, B, ... , K simultaneously.
Now, in Sec. 2.6, we have shown that the columns of<l> that satisfies Eq. (10)
and diagonalizes the matrix K are the eigenvectors of the equation
KX=2MX. (14)
Extending the reasoning to other matrices A, B, ... , we can state the result as
follows: The necessary and sufficient condition for Eq. (7) to be decoupled into
Eq. (8) is that the eigenvalue problems
KX = AMX, AX = AMX, BX = 2MX, ... , rx = 2MX (15)
have a common set of linearly independent eigenvectors Xl' ... , Xn •

2.11 Sufficient Conditions for Decoupling Equations

of System with Damping
For arbitrary matrices K, A, B, ... , r, decoupling of Eq. (2.10:7) is, in general,
not possible. But several sufficient conditions for the possibility are known.
We shall quote them without proof.
1) The case in which A, B, ... , K are real and symmetric.
If the matrices A, B, ... , K are real and symmetric, and can be diagonalized
simultaneously, then among these matrices there is one, say K, which is com-
mutable with all the rest; i.e.,
AK = KA, BK = KB, ... , etc. (1)
If one of the matrices, say K, is commutable with all other matrices A, B, ... ,
and if all the eigenvalues of K are distinct (no multiple eigenvalues), then A, B,
... , K can be simultaneously diagonalized.
This result is due to Bellman (1970). The condition that K should have no
multiple eigenvalues weakens the theorem, but it cannot be removed. How-
ever, if only two matrices A, K are involved, then multiple eigenvalue can be
permitted. In this case one must show that if Av is an r-fold multiple root, then
among the linear combinations of the r mutually orthogonal eigenvectors of
K belonging to Av, there are r eigenvectors of A.
2) The case in which A, B, ... , K are normal matrices.
If we do not wish to restrict ourselves to real and symmetric matrices, we
should demand at least the existence of n linearly independent eigenvectors
for the eigenvalue problems (Eq. (2.10:15)). Now, there is a well-known theorem
56 2 Segmental Movement and Vibrations

which states that if AX = AX possesses a system of n linearly independent

eigenvectors, then A is a normal matrix; i.e., it is commutable with its adjoint:
A·(adjA) = (adjA)·A. (2)
[The adjoint of a square matrix A = (aij) is the matrix adj A = (aji ), i.e., the
transpose of a matrix whose elements are the complex conjugate of those of
(a i). If A is real, then adj A is just the transpose of A.]
We quote without proof Bellman's (1970) results:
If M is real, symmetric, and positive definite, and A, B, ... , K are all normal
matrices commutable with M,
AM = MA, BM = MB, ... , KM = MK, (3)
then the matrices M, A, B, ... , K can be simultaneously diagonalized if one of
the matrices, A, ... , K, say K, is commutable with all the rest:
AK = KA, BK = KB, etc., (4)
and if all the eigenvalues of K are single.
Examples. Rayleigh (1894) considered the vibration of a Voigt solid
described by the matrix equation
MX: + BX + AX = F, (5)
where M, B, A are real, symmetric, and positive definite. This equation can
be decoupled; if B is a linear combination of M and A:
B = O(M + f3A, (6)
where 0(, f3 are numerical constants.
Caughey (1960) considered Eq. (5) under the assumption that A is real,
symmetric, and positive definite, and showed that B and A can be simul-
taneously diagonalized if B can be represented as a power series of A,
n=1
B= L O(mAm. (7)
m=O

Caughey (1960) showed further that if A = B", K = an integer, then A and

B are commutable. Define A 1/1< as a matrix B which satisfies the equation
A = B". Then A and B can be simultaneously diagonalized if B can be
expressed in a series
n=1
B= L f3m(A 1/")m. (8)
m=O

Problems

2.1 Solve Eq. (2.2:2) when F(e) is a sinusoidal excitation:

F(e) = Fo sin rot.
Problems 57

Sketch the relationship between the amplitude of motion Ixl and the frequency
of excitation w. Discuss the three cases:
1]2 < 4Km, /3 2 > 4Km, /3 2 = 4Km.
Determine also the response of this system to an impulsive impact load which is
a delta function oftime. Assume that the initial conditions are x = x = 0 at t = O.
Propose a possible application of the solution to a clinical problem.
2.2 Write down the Lagrangian equation of motion for a weightless cantilever beam
with a mass attached to the free end. The mass has a finite moment of inertia, I,
and is rigidly attached to the beam.
2.3 Derive the Lagrangian equation for the vertebrae-disks system (Sec. 1.4, Figs.
1.4:3, 1.4:4) by allowing six degrees offreedom: three relative translations of two
adjacent vertebrae, two bending and one twisting. Consider the lower vertebrae
as fixed while the upper vertebra and the disk in between can be displaced.
2.4 Formulate a model of the head and spine of a man to be used in the investigation
of head injury to drivers in automobile crashes, or neck injury of an aircraft pilot
in an ejection seat. Derive the equations of motion.
2.5 Show that in normal coordinates, the kinetic and strain energies are

K = t(q,q) = t(4r + ... + 4;),

u = t(q,(J)TK(J)q) = t(wrqr + ... + w;q;).
2.6 In Fig. 2.5:3, a man standing on ground is modeled as an inverted pendulum.
Consider this as a dynamic model of posture control. Assume a feedback control
system of the body which yields the following moment at the ankle: M =
kl e + kil, where kl and k2 are constants. Considering only very small angle e,
what are the restrictions on the values of kl' k2 so that the body remains stable,
so that he will not fall down if slightly disturbed. What are the restrictions on the
values of k 1 , k2 so that he is dynamically stable (so that a small disturbance will
not result in an oscillation whose amplitude will increase with increasing time)?
(Cf. Nashner (1971) and McGhee (1980).)
2.7 By increasing the number of joints and degrees of freedom in the inverted
pendulum model, present a better model for posture control. (Cf. McGhee (1980)
and Pedotti (1980). Data on muscles of the leg, their mean length, fiber length,
cross-sectional area, maximum velocity, maximum isometric force, and constants
in Hill's equation are presented in Pedotti's paper (p. 113).)
2.8 Discuss the aerodynamic forces acting on an athlete doing one of the following
sports: parachuting, free fall from an airplane, ski jumping or ski flight, jumping
from a diving board into water, pole vaulting, motor cycling. (Cf. J. Maryniak
(1977). Static and dynamic investigations of human motion. In Euromech
Colloquium, Varna, 1975: Mechanics of Biological Solids, (edited by G. Brankov).
Bulgarian Academy of Science, Sofia, pp. 151-174.)
2.9 To investigate possible injuries to the knee in playing soccer football, make a
model of the leg as a double pendulum, and derive the Lagrangian equation of
motion using the angles of the thigh and the lower leg from the verticle as
58 2 Segmental Movement and Vibrations

fl.')
I

~
./l \

FIGURE P2.1O A fouette turn. The performers body rotates very little in views 2-4
while her extended leg retains most of the angular momentum. In views 5-6 she turns
rapidly while her leg is held close to her body, where it has a smaller moment of inertia.

generalized coordinates. (Cf. L. Lindbeck (1977). Theoretical analysis of reactions

in the knee caused by impact. In Euromech Call., Varna, 1975, loc cit. pp. 142-
150.)

2.10 Analyze the dance steps of pirouettes (a rotation of the body around a vertical
axis over one supporting foot on the floor) in terms of the principle of conservation
of angular momentum. There are many types and styles of pirouettes, from a low
turn on bent legs in modern dance to attitude or arabesque turns, and to the
spectacular multiple fouette turns often seen in classical ballet. See sketches in
Fig. P2.1O. To commence a pirouette, the dancer must exert opposite horizontal
forces with the two feet in order to apply the necessary torque. How is this done?
Can the dancer apply some additional torque in the course of the turn to maintain
the angular momentum against necessary losses? (See Laws, K. (1984). The
Physics of Dance. Schirmer Books.)
2.11 Two dancers, one 5 ft. tall, the other 6 ft tall, are similarly proportioned, and their
muscles are similar. Compare their body masses, cross sectional areas of their
legs, the forces their muscles can exert, the heights they can jump, and the lengths
of time they can remain in the air. Show that in order to allow the larger dancer
to jump to the same height as the smaller dancer a 73% force larger is necessary.
If they want to jump to the heights in same proportion to their body length, the
References 59

larger dancer would have to exert twice the muscular force. If they perform
"beats" with legs at the same rate and angular amplitude, the taller dancer must
exert 2.5 times the torque at the hip. (A beating motion is executed during a
vertical jump, in which the legs, while straight are kicked out to each side, then
crossed, then kicked out, often several times in one leap.) (See Laws, K. (1984).
loc cit.)
2.12 T. Kenner designed a very sensitive instrument to measure the density of blood,
plasma, and other body fluid (Paar Inc., Graz, Austria). The principle is to set
the fluid flowing through a small cylindrical tube which is bent into a U shape
and excite the tube to measure its frequency response. The resonance frequencies
can be measured very accurately with an electronic counter. These frequencies
are related to the density of the fluid. Knowing this much about the principle,
reinvent such an instrument with your own design. (Cf. Applications to pul-
monary problem by J.S. Lee (1988). Microvas. Res. 35: 48-62.)
2.13 Formulate a model of the arm to study the tennis elbow, pains that often amict
tennis players.
2.14 Formulate a model of the human leg with the objective to measure the load on
the articular cartilage of the knee. Discuss the possible relevance of such studies
to the problem of cartilage breakdown and joint degeneration (cf. Mizrahi and
Susak (1982)).
2.15 Review the meaning of Eulerian angles which describe the rotation of a rigid body
in space; cf. e.g. Greenwood (1965, 1977), Kane (1968), Huston and Perrone (1980).
In the modeling of arms, legs in locomotion, the choice of parameters to measure
angular position is very important. There are special choices of the rotation axes
with which the final angular position of the rigid body is specified by three angles
whose values do not depend on the order of the rotations. A general theory for
such choices has been developed by Roth (1967). A gyroscopic system has been
used by Chao (1980), Chao and Morrey (1978), Grood and Suntay (1983), in which
the angles are the clinical measures used by orthopedic surgeons: the flexion-
extension, the internal-external rotation, and the adduction-abduction.

References

For a thorough discussion of Lagrangian equations, see Greenwood (1977). For influence coeffi-
cients (Sec. 2.3), see Fung (1955), and Fung (1965). For aerodynamic loading, see Fung (1955).
For muscle mechanics, see Fung (1981).
Bean, J.e., Chaffin, D.B., and Schultz, A.B. (1988). Biomechanical model calculation of muscle
contraction forces: a double linear programming method. J. Biomech. 21: 59-66.
Bellman, R.E. (1970). Introduction to Matrix Analysis, 2nd ed. McGraw-Hill, New York.
Camana, P.C., Hemami, H., and Stockwell, e.w. (1977). Determination of feedback for human
posture control without physical intervention. J. Cybernet. 7: 199.
Caughey, T.K. (1960). Classical normal modes in damped linear dynamic systems. J. Appl. Meeh.
27: 269-271.
Chao, E.Y.S. (1980). Justification of triaxial goniometer for the measurement of joint motion. J.
Biomeeh. 13: 989-1006.
60 2 Segmental Movement and Vibrations

Chao, E.Y.S. and Morrey, B.F. (1978). Three-dimensional rotation of the elbow. J. Biomech. 11:
57-74.
Fung, Y.c. (1955, 1969). The Theory of Aeroelasticity. Wiley, New York. Revised, Dover
Publications, New York.
Fung, Y.c. (1965). Foundations of Solid Mechanics, Prentice-Hall, Englewood Cliffs, N.J.
Ghista, C.N. (1982). Osteoarthromechanics. McGraw-Hill, New York.
Greenwood, D.T. (1965). Principles of Dynamics. Prentice-Hall, Englewood Cliffs, N.J.
Greenwood, D.T. (1977). Classical Dynamics. Prentice-Hall, Englewood Cliffs, N.J.
Grood, E.S. and Suntay, W.J. (1983). A joint coordinate system for the clinical description of
three-dimensional motions: Application to the knee. J. Biomech. Eng. 105: 136-144.
Huston, R.L. and Perrone, N. (1980). Dynamic response and protection of the human body and
skull in impact simulation. In Perspective in Biomechanics. (H. Reul, D.N. Ghista and G.
Rau, eds.), Vol. 1, Part B, pp. 531-571, Harwood Academic Publishers, New York.
Kane, T.R. (1968). Dynamics. Holt, Rinehart, and Winston, New York.
McGhee, R.B. (1980). Computer simulation of human movements. In Biomechanics of Motion
(A. Morecki, ed.), Springer-Verlag, Wien and New York, pp. 41-78.
Mizrahi, J. and Susak, Z. (1982). In vivo elastic damping response of the human leg to impact
forces. J. Biomech. Eng. 104: 63-66.
Nashner, L.M. (1971). A model describing vestibular detection of body sway motion. Acta
Otolaryng. 72: 429-436.
Pauwels, F. (1980). Biomechanics of the Locomotor Apparatus. (Trans. from the German by P.
Maquet and R. Furlong) Springer-Verlag, New York.
Pedotti, A. (1980). Motor coordination and neuromuscular activities in human locomotion. In
Biomechanics of Motion (A. Morecki, ed., Springer-Verlag, New York, pp. 79-129.
Rayleigh, Baron, John William Strutt (1894). The Theory of Sound. Republished by Dover, New
York, Vol. 1, p. 131.
Roth, B. (1967). Finite position theory applied to mechanism synthesis. J. Appl. Mech. 34:
599-606.
Schultz, A.B., Anderson, G.B.J., Haderspeck, K., Ortengren, R., Nordin, M., and Bjork, R. (1982).
Analysis and measurement of lumbar trunk loads in tasks involving bends and twists. J.
Biomech. 15: 669-675.
HAWK IN THE AUTUMN by Lin Liang (;1* In Light color and ink on silk, 136.8 x
74.8 em, wall scroll, in National Palace Museum, Taipei. Lin was a native of Kwan-
tung. He lived in the latter half of the 15th century, in Ming Dynasty.
CHAPTER 3

External Flow: Fluid Dynamic Forces

Acting on Moving Bodies

3.1 Introduction

When humans exercise, birds fly, fish swim, animals run, and trees sway, we
want to know the forces they experience. The calculation of the forces they
experience from the surrounding fluid is the realm of fluid dynamics. In this
chapter we present the classical theory. In the next chapter we discuss flying
and swimming in nature.

3.2 Flow Around an Airfoil

A real fluid is viscous and compressible. But, if the speed of flow is much less
than the speed of propagation of sound, then the variation of density caused
by the motion of a body in the fluid is so small that the fluid may be regarded
as incompressible. Furthermore, for birds and fish moving in air and water at
Reynolds number much greater than one, the effect of the viscosity of the fluid
is felt only in a thin layer (the boundary layer) next to solid wall of the body.
Outside the boundary layer the fluid may be regarded as nonviscous. A
non viscous and incompressible fluid is a perfect fluid. In many problems of
locomotion, it is sufficient to consider the fluid as a perfect fluid. Yet the
viscosity, however small, has profound effects, for it controls the boundary
layer which may become detached from part of the solid body, and thus affects
the macroscopic picture of the flow.
The force exerted by the fluid on a solid body depends on the relative
velocity between them. The fluid dynamic force consists of two components:
the pressure force normal to the surface of the body, and the skin friction, or
shearing force, tangential to the surface of the body.

62
3.2 Flow Around an Airfoil 63

The parameters affecting the force acting on a body in a flow can be

determined by a dimensional analysis. Obviously the force depends on the
geometry of the body and its attitude relative to the flow; these can be
characterized by a typical length L and a typical angle ct. The force depends
also on the density of the fluid p, the viscosity of the fluid fl, the speed of flow
V, the compressibility of the fluid, and the nonstationary characteristics of the
flow, e.g., the frequency w if the motion is periodic. The compressibility of the
fluid may be expressed in various ways. A simple index of the compressibility
is the speed of propagation of sound in the fluid, because sound is propagated
as longitudinal elastic waves.
Let the speed of sound propagation be denoted by c. Then, in an oscillating
flow of a compressible fluid, the force experienced by a solid body will depend
on the following variables:
L, ct, p, V, fl, w, c. (la)
A dimensional analysis shows that, for geometrically similar bodies, the force
F acting on the body can be expressed as

F = f(ct VLp wL V)~pV2L2 (1b)

'fl'V'c2

where f is a function of ct, V Lp/ fl, wL/V, and V /c. It is easy to verify that the
parameters VLp/fl, wL/V, and V/c are dimensionless. They are known as:
VLp VL
R = -- = - = Reynolds number
fl v

wL
k = U = reduced frequency or Strouhal number (2)

V
M = - = Mach number.
c
Rand k are also written as NR and Ns , resp., elsewhere in this book. The
quantity q = tpv 2 is known as the dynamic pressure. The factor v = fl/P is
the kinematic viscosity.
The speed of sound propagation in a gas is given by the equation

c = ft;, (3)

where p is the static pressure, p is the density, and y is the ratio of the specific
heat at constant pressure to the specific heat at constant volume. For dry air
at 15°C and 1 atm, v = 0.145 cm 2 /sec, y = 1.401, p = 1.225 X 10- 3 gm/cm\
c = 340.6 m/sec.
For water under the same conditions, v = 1.138 X 10- 2 cm 2 /sec, p =
0.9991 gm/cm 3 , c = 1,445 m/sec.
In flying and swimming, we are concerned primarily with bodies like an
64 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

airfoil. The wing geometry and the conventional terminology are illustrated
in Fig. 3.2:1. A chord is a line defined in a cross section, passing through the
trailing edge, pointing in the direction of relative wind of no-lift. The angle
between the chord and the direction of flight is the angle of attack. Two
components of force and one component of moment that act on the body are

Relative wind
---

~Chord
Angle of attack

~<
(a)

Angle of
attack Relative
wind

Sweep back angle

Span
(b)

FIGURE 3.2:1 Definitions of terms. (a) An idealized wing. (b) Upper: A black vulture
with wings outstretched in soaring flight. Lower: Wings flexed in fast gliding flight.
3.2 Flow Around an Airfoil 65

Lift, L

Pitching moment, M
Direction of
flight~

FIGURE 3.2:2 Wing geometry and conventional terminology.

of interest (Fig. 3.2:2):

Lift = L = force perpendicular to the direction of motion,
Drag = D = force in the direction of motion, positive when the force acts
in the downstream direction.
Pitching moment = M = moment about an axis perpendicular to both the
direction of motion and the lift vector, positive
when it tends to raise the leading edge of the
body.
For the wings of birds or airplane, the chords of all cross sections lie
approximately in a plane defined as the mean chord plane. The mean chord
length of the wing is usually taken as the characteristic length of the wing. The
area of the wing projected on the mean chord plane is defined as the wing area.
Let c be the mean chord length, S be the wing area, q be the dynamic pressure.
Then the three primary dimensionless coefficients of interest are:
CL = lift/(qS) = lift coefficient
CD = drag/(qS) = drag coefficient (4)
CM = pitching moment/qSc = pitching-moment coefficient.
CL , CD, and CM are functions of the Reynolds number, Mach number, Strouhal
number, and the body's shape and attitude with respect to the flow.
In a steady flow of an incompressible fluid, the Strouhal number and the
Mach number both vanish, and CL , CD' CM depend on the Reynolds number
66 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

and angle of attack. The Reynolds number characterizes the effect of fluid
viscosity. For large birds, fish and animals, the Reynolds number is much
larger than one. In the remainder of this section, unless mentioned otherwise,
we shall consider only the steady flow of an incompressible fluid at large
Reynolds number. Flow at small Reynolds number, relevant to microbes and
cells, is discussed in Chapter 4.
The lift increases linearly with increasing angle of attack until a certain
value around 10-20° is exceeded. Beyond that value, the lift levels off, eventu-
ally reaches a maximum, and then drops off rapidly. The maximum lift is
reached at a critical angle of attack, O(Lmax, which is the stalling angle. For
angle of attack 0( greater than O(Lmax the wing is stalled. The stalling angle and
the maximum lift coefficient are characteristic numbers for each wing design.
The lift of a wing is zero at some angle of attack. It is convenient to define
that angle of attack as zero, and measure the angle of attack from the zero-lift
line. Such an angle of attack is denoted by 0(. The chord of an airfoil is defined
along the zero-lift line.
For an unstalled wing, the lift coefficient can be expressed as
(5)
in which 0( is the angle of attack and a is the lift curve slope. If 0( is measured
in radians, hydrodynamic theory (see Sec. 3.11) gives the lift-curve slope for
thin airfoils of infinite span in a two-dimensional flow as
ao = 2n (theory, incompressible fluid). (6)
Experimental values of the lift-curve slope are somewhat smaller than this for
most wings, but a/ao is greater than 1 for the so-called NACA low-drag
sections designed and tested by NACA (US National Advisory Committee of
Aeronautics, predecessor of NASA).
According to the theory of thin airfoil (Sec. 3.11), the center of pressure of
the lift is located at i-chord aft of the leading edge. This point is the aero-
dynamic center. If the moment coefficient is computed about the aerodynamic
center, it does not vary with CL • The symbol CMc /4 is used to denote the
moment coefficient that refers to an axis located at the -.t-chord point. The
aerodynamic center remains close to the i-chord point in a compressible fluid
as long as the flow is subsonic; but it moves close to the midchord point if the
flow becomes supersonic.
The subscript "0" of the lift-curve slope ao in Eq. (6) signifies that ao is the
value pertaining to an airfoil of infinitely long span. (The span is the distance
from wing tip to wing tip.) For wings of finite span, the lift-curve slope is
smaller. In Prandtl's finite-wing theory (see Sec. 3.10), a wing is replaced by a
vortex line. Since a vortex line cannot end at the wing tip, it must continue
laterally out of the wing and become a free trailing vortex in the fluid. This
will be discussed later in Secs. 3.9 and 3.10. The vertical velocity induced by
the vortex line and trailing vortices is called the induced velocity, or downwash,
3.2 Flow Around an Airfoil 67

Chord line

----------ra
u

FIGURE 3.2:3 Downwash and induced angle.

and is denoted by w in Fig. 3.2:3. Because ofthe induced velocity, the direction
of flow at the airfoil is changed by an amount e indicated in the figure. If U is
the flight speed, then
tane = wjU. (7)
From Fig. 3.2:3 it is seen that the effective angle of attack 0(0 is smaller than
the geometric angle of attack 0( according to the relation
W
01: 0 = 0( - e == 0( - - (8)
U
when e is so small that tan e == e. The force induced by the effective angle of
attack is proportional to the velocity resultant Ures ' and acts in a direction
normal to the velocity vector Ures • See Sec. 3.7 infra and vector pUre. r in Fig.
3.2:3. It can be resolved into a lift component L perpendicular to the velocity
of flow U and a drag component (induced drag) D in the direction of U. The
resultant pUre. r is proportional to the circulation r (Sec. 3.7) and angle of
attack 0(0. By using Eqs. (5), (6), and (8), we obtain

CL = aO( = aoO(o = ao(O( - ~). (9)

The downwash w is uniform over the entire wing if the wing planform is an
elongated ellipse and is untwisted (having a constant angle of attack across
the span) (see Sec. 3.10). In this case
w CL
(10)
U nAR'
where

AR . (span)2 wing span

= aspect ratIO = = ---::------,:-:------,- (11)
wing area chord length
68 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

Substituting (10) into (9) yields a, the lift-curve slope of a wing of finite span:

a= 1
+
(a j
ao nAR
) (elliptic wing). (12)

Nonelliptic wings would have nonuniform downwash across the span and
a somewhat smaller lift curve slope than that given by Eq. (12). See Glauert
(1947).
These results are applicable to birds. To improve the stalling characteristics
of the wing and to obtain a high maximum lift coefficient, birds spread their
feathers to produce a more curved wing cross-section, and use many other
features that are copied by aeronautical enginers. See Sec. 4.3.

3.3 Flow Around Bluff Bodies

For a blunt body (such as a man) moving in a fluid, usually the drag force
predominates but an oscillatory lift may exist. We shall discuss drag in Sec.
3.15 and lift in this section.
Whoever has rowed a boat must have observed the trail of vortices leaving
the oar. Figure 3.3:1 shows the wake behind a circular cylinder. Vortices are
"shed" alternately from the sides of the cylinder. This shedding of vortices
induces a periodic force in the direction perpendicular to the line of motion,
i.e., an oscillatory lift.
The vortex shedding phenomenon is relevant to the swaying of trees,
rustling of leaves, and bending of the blades of grass. We feel it on our legs

8.---,----,----.----r---.----.----r---.----.---~

y
d

-4

-8~--~----~--~----~--~-- __ ____ __-L____

~ ~ ~__~

o 4 8 12 16 20 24 28 32 36 40
xld

FIGURE 3.3:1 The wake behind a circular cylinder. Reynolds number 56. Measure-
ments by Kovasznay (1949). Figure shows the streamline pattern viewed relative to
the undisturbed flow at infinity. The development and decay of the vortices can be
seen. The lines correspond to differences in the stream function fit/! = 0.1 Ud; the dotted
lines are half-values between two full lines.
3.3 Flow Around Bluff Bodies 69

when we wade in running water in a creek. Cranes and ducks must know it
well. In the manmade world, telephone wires "sing," and smokestacks, sub-
marine periscopes, oil pipe lines, and television antennas vibrate for the same
reason. These vibrations can be controlled either by stiffening the structures
so that the natural frequency is higher than the frequency of the vortex
shedding in wind, or by introducing vibration dampers into the system to
absorb the energy.
The flow around a long circular cylinder will be explained in greater detail.
The flow changes with the Reynolds number, R, defined as vd/v, d being the
diameter of the cylinder. The variations of the drag coefficient (Eq. 3.2:4) and
the Strouhal number (Eq. 3.2:2) of the flow with Reynolds number are shown
in Fig. 3.3:2. At low Reynolds number, the flow is smooth and unseparated,

2.5r-----,------,---------,--------,---~

2.0 t--t---t----+-----+----+------l-i

1.5 t----~+------+----+-----i-----I-----l
k

1.0 t-----fr--7-----"!oO;;;::--=7""""--+-----4-lI---\-----l

0.5 t----r--t----+-----+----+-+--+----j
\
\
,---

Q~--~--~------~------~------~------~
1 2 3 4 5 6

FIGURE 3.3:2 Variation of the Strouhal number and drag coefficient against Reynolds
number for a circular cylinder. CD and R are based on the diameter of the cylinder.
Sources of data are: NPL; Relf and Simmons, Aeronaut. Research Com. R. and M.
917 (1924). Cambridge; Kovasznay, Proc. Roy. Soc. A.198 (1949). CIT; Roshko, NACA
Tech. Note 2913 (1953). Gottingen; Ergebnisse AVA Gottingen, 2 (1923).
70 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

but the fluid at the back of the cylinder is appreciably retarded. At higher
values of R, two symmetrical standing vortices are formed at the back. When
R reaches about 40, the vortices become asymmetrical, detach from the
obstacle, and move downstream as if they were discharged alternately from
the two sides of the cylinder. An eddying motion in the wake is set up. As
the flow moves downstream the eddying motion is gradually diffused and
"decays" into a general turbulence. For R in the range of 40 to 150, the
"shedding" of vortices is regular. The range of R between 150 and 200 is a
transition range, in which the vortex shedding is less regular and its frequency
appears to be somewhat erratic. For R '" 300, the vortex shedding is irregular,
for although a predominant frequency exists, the amplitude appears to be
random. Finally, at R of the order 2 x 10 5 , the separation point of the
boundary layer moves rearward on the cylinder. Consequently, the drag
coefficient of the cylinder decreases appreciably, as shown in Fig. 3.3:2.
The geometry of the wake, when the Reynolds number is in a range in
which vortices are regarded as shedding, is as follows: The frequency at which
the vortices are shed, expressed nondimensionally as the Strouhal number k,
is a function of the Reynolds number, as shown in Fig. 3.3:2. Here the Strouhal
number k is defined as wd/V, where w is the frequency in radians per second
and d is the diameter of the cylinder. The number of vortices shed from each
side of the cylinder every second is n = w/2n:
kV
n = 2nd per second. (1)

The distance A between two consecutive vortices in a row is

(~ '" 0.25 to °),

v- v
A = - -, (2)
n
where v is the relative velocity of the vortices with respect to the free stream.
The flow pattern is approximated by a regular array of vortices shown in Fig.
3.3:3. This flow pattern was analyzed by theodore von Karman and is known
as the von Karman vortex street. The theoretical ratio of the distance H
between the rows of vortices to the distance A is given by von Karman to be

~ '3 + ~
H
~ * ~
I. A J
~ = 0.28
FIGURE 3.3:3 A Karman vortex street.
3.4 Steady-State Aeroelastic Problems 71

H = 0.281A at small distance from the cylinder, but H/A increases as the
distance from the cylinder increases. At large distance, H/A is ofthe order 0.9.
The maximum intensity of the velocity fluctuations occurs in the vicinity of 7
diameters downstream. Thus it appears that the vortices are not really shed
from the cylinder, but are developed gradually.
The shedding of vortices creates an oscillatory lift. In the Reynolds number
range 100-1,000, the periodic lift coefficient has an amplitude of 0.45, while
the drag coefficient is about 1.09. Some experiments indicate that the lift
coefficient can be as large as 1.

3.4 Steady-State Aeroelastic Problems

Let us illustrate the application of the information presented in Secs. 3.2 and
3.3, by considering some aeroelastic instabilities.

Divergence
If a wing in steady flight is accidentally deformed, an aerodynamic moment
will be induced which tends to twist the wing. This twisting is resisted by elastic
moment. However, since the elastic stiffness is independent of the speed of
flight, whereas the aerodynamic moment is proportional to the square of the
flight speed, there may exist a critical speed of flight, at which the elastic
stiffness is barely sufficient to hold the wing in a stable position. Above the
critical speed, an infinitesimal accidental deformation of the wing will lead to
a large angle of twist. This critical speed is called the divergence speed, and the
wing is then said to be torsionally divergent.
As a two-dimensional example, let us consider a strip of unit span of an
infinitely long wing of uniform cross-section. As shown in Fig. 3.4:1 let the
elastic restraint imposed on this strip be regarded as a torsional spring with
an axis at a point G. If the spring is linear then the torque is directly propor-
tional to the angle of twist.
The action of the aerodynamic force on the airfoil can be represented by a
lift force acting through the aerodynamic center, and a moment about the
same point. Let us write the distance from the aerodynamic center to the axis

FIGURE 3.4:1 A two-dimensional airfoil.

72 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

of the torsional spring as ec, c being the chord length and e being a ratio
expressing the eccentricity of the aerodynamic center (positive if the spring lies
behind the aerodynamic center). The lift coefficient Cl (we use lower case I to
indicate lift per unit span) is proportional to the angle of attack, whereas the
coefficient of moment about the aerodynamic center, CmO , is practically inde-
pendent of the angle of attack. Hence, the lift and moment per unit span acting
on the airfoil are
L' = qcCI = qca(O + a),
(1)
M~ = qCmo c2 (about aerodynamic center),
respectively, where a is the slope of the lift-curve (Cl vs. a), q is the dynamic
pressure (!pU 2 ), a is the initial angle of attack, and 0 is the angle of twist. The
prime on L' denotes force per unit length in the span wise direction.
Using Eqs. (1), the aerodynamic moment per unit span about the torsional
spring is
M~ = M~ + L'ec = qc 2 Cmo + qec2 a(O + a)
= qec 2 aO + qec 2 a(a + CmO/ea). (2)
By redefining the angle of attack by absorbing CmO/ea into a we may be rewrite
Eq. (2):
(3)
When equilibrium prevails, the aerodynamic moment is balanced by the
elastic restoring moment. Let Ka be the spring constant, then the elastic
restoring moment per unit span is KaO. On equating this with the aerodynamic
moment given by Eq. (3), and solving for 0, we obtain

0= qec 2 aa (4)
Ka - qec 2 a'
For a given nonvanishing a, the angle 0 will increase when the dynamic
pressure q increases. When q is so large that the denominator vanishes, the
angle 0 tends to infinity and the airfoil becomes divergent. Hence, the condi-
tion of divergence is
(5)
The dynamic pressure at divergence, qdiv' and the divergence speed of flight,
Udiv are given by the equations

and (6)

Thus the critical divergence speed increases with increasing rigidity of the wing
and decreasing chord length and eccentricity. The ratio of the actual angle of
twist of an elastic wing to that of a rigid wing varies with the ratio of the
dynamic pressure to the critical divergence pressure, as shown in Fig. 3.4:2.
3.4 Steady-State Aeroelastic Problems 73

10

/
8

6
/
4 /
,,- . /
/
-----
2

o
o 0.2 0.4 0.6 0.8 1.0
q/qdiv

FIGURE 3.4:2 Ratio of angles of twist of an elastic wing to that of a rigid wing.

It is seen that the twist becomes very large when the divergence speed is
approached.
The analysis presented above can be extended to three-dimensional wings.
For details see Fung (1955, 1969), Bisplinghoff and Ashley (1962), Dowell, et
al. (1978). Among other things it is shown that sweeping back the wing (wing
sheared backward with wing tip pointing towards the tail) increases the critical
divergence speed; whereas sweeping forward (wing so sheared that the tip
points to the head) decreases it. Most bird's wings are sweptback. In the
example shown in Fig. 3.2: 1(b), the inner span is swept forward, the outer span
is swept back.

Loss of Control
Airplanes pitch, roll, and yaw by deflecting their ailerons, flaps, elevators and
rudders. Birds do these maneuvers by feathering their wings and deflecting
their tails. The effectiveness of the deflection of control surface, however,
depends on the speed of flight. The loss of effective control as flight speed
increases puts an upper limit on the speed of flight.
The phenomenon can be explained by considering a simplified model as
shown in Fig. 3.4:3. Consider a wing of chord length c that is held by a torsional
spring located at a distance ec from the aerodynamic center. The wing has an
aileron of width Ec. e and E are constant fractions. The relative wind comes
 n
74 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

L,
Aerodynamic Torsional
cen:-te_r_ _--""'_---'i:++--s:-p"""rin~ ____ 1__
u~ C& ~
FIGURE 3.4:3 A two-dimensional wing with aileron.

at an angle of attack rx against the zero-lift line. The angle of deflection of the
aileron (positive downward) with respect to the main airfoil is denoted by p.
The lift coefficient and the coefficient of moment about the aerodynamic
center can be written in the form
aCI
Cl = arx + p ap' (7)

Cm =
Pacm
ap + CmO , (8)

where a is the lift-curve slope and CmO is the coefficient of moment about the
aerodynamic center of the airfoil with undeflected aileron. According to
Glauert (1947), the coefficients aCdap and acm/ap for a two-dimensional
airfoil in an incompressible fluid are
1 aCI 1 r----

~ ap = -;[arccos(1 - 2E) + 2JE(1 - E)] (9)

aa~m = ~(1 - E)JE(1 - E) (P in radians), (10)

where E is the ratio of the flap chord to the total chord (Fig. 3.4:3). According
to Eq. (7), the lift per unit length of this airfoil is

L, = qc ( arx + aclp)
ap . (11)

Ifno lift can be produced when the aileron is deflected, then the control is lost.
The critical reversal condition is given by the vanishing of the derivative
dL'/dp. This happens, according to Eq. (11) and noting that aCdap is a
constant, Eq. (9), when

dL' (arx acl )

dP = qc a ap + 8jf = O. (12)

Solving this equation for arx/ap, we have

arx 1 aCI
(13)
ap -~ ap'
3.4 Steady-State Aeroelastic Problems 75

To evaluate orx/oP we must consider the elastic constraint. The aerodynamic

pitching moment per unit length of span about the axis of rotation is

, 2(eC, + p oCop + C
M = qc m
mO
)
• (14)

In this equation, the first term qc 2eC, is the product of the lift qcC, acting at
the aerodynamic center times the distance from the aerodynamic center to the
axis of rotation. The last two terms are the moment about the aerodynamic
center according to Eq. (8). This pitching moment is balanced by the elastic
restoring moment rxK, K being the stiffness of the torsional restraint per unit
span of the airfoil. Hence

rxK =
2(
qc eC, + p oC
op + CmO )
m
= qc 2(earx + ep oC,
op + p oC
op
m
+ CmO ) •

(15)
Differentiating with respect to p, we obtain (CmO being a constant),
orx 2 (orx oC, OCm) (16)
K op = qc ea op + e op + op .

Substituting orx/oP from Eq. (13), we obtain the critical dynamic pressure for
control reversal:

qrev
1 OC (
= ~ oP' - opm
oC )-1 K c2 . (17)

Hence, the critical reversal speed is given by

= (~ OC,)1/2 (_ OCm)-1/2 (2K)1/2 (18)
Urev a op op pc 2 '
which shows that the critical reversal speed increases with increasing stiffness
K of the airfoil. Note that the quantity e is absent from this formula. Hence
the position of the torsional axis does not affect the reversal speed. This is
because the net change oflift due to aileron displacement is zero at the reversal
speed.
If we define the elastic efficiency of the control surface, or control surface
efficiency, as the ratio of the lift force produced by a unit deflection of the
aileron on an elastic wing to that produced by the same aileron deflection on
a fictitious rigid wing of the same chord length, then it can be shown that
(Fung, 1969)
. ffi'
EIastIc e lClency = 11 - q/qrev
/ . (19)
- q qdiv

This result is shown in Fig. 3.4:4. It is seen that the control surface efficiency
drops to zero very rapidly when the flight speed approaches the critical
reversal speed.
76 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

R=1
1.0
~
~ F=: ~ F=== ~~sl
~
s:s
~
I----- .......
~ ~ ~ i'-
~ ~ r<s ~ ~ \
~~~ ~ 1\ "\ \
~ "",'"1\ !\ 1\ \
\\
0.5

~1\\ \
>.
u
c:
CI>
'0

~ \\ 1\\
~ R= qdiv
q,ev
c:
e
~
C( '[\\ 1\ \ \
\
~\
~
o
~
\\~I\
-0.3
o 0.5 1.0
\ \\
FIGURE 3.4:4 Aileron efficiency versus dynamic pressure when qdiv > qrev'

3.5 Transient Fluid Dynamic Forces Due to Unsteady Motion

When bodies make unsteady motion, the surrounding fluid will of course
respond in a transient manner. To understand flying and swimming we must
study the fluid dynamics of unsteady motion.
The effect of unsteadiness is to delay lift generation and stall. For example,
in steady state a two-dimensional wing flying at a velocity U at an angle of
attack IX will generate a lift force per unit span equal to
(1)
according to Eqs. (3.2:4)-(3.2:6), c being the chord length, and p being the
density of the fluid. But if the same wing starts motion impulsively from rest
to a uniform velocity U at the same angle of attack, the lift force per unit span
 3.5 Transient Fluid Dynamic Forces Due to Unsteady Motion 77

1.0

,.,- L---
~
0.8

L
/"
0.6
~ (T)
/
0.4

0.2

o
o 2 4 6 8 10 12 14 16 18 20
T, distance traveled, in semichords

FIGURE 3.5:1 Impulsive motion of an airfoil.

would be a function of time:

L' = 2nctpU 2 oc<l>(,), <1>(,) = 0 ifr < 0, (2)
where
2U
,=-t. (3)
c
Thus, is time measured by the distance traveled in units of semichord length.
The function <1>(,) was first determined by H. Wagner (1925) and is known as
Wagner's function. It is illustrated in Fig. 3.5:1. An approximate expression
which agrees within 2 percent of the exact value in the entire range 0 < , < 00
is given by Garrick (1938):
2
<1>(,) ='= 1- 4 +" (, > 0). (4)

Another approximate expression is given by R.T. Jones (1940):

<1>(,) = 1 - 0.165e-o.o 4ss < - 0.335e- O. 300<. (5)
From these formulas, and also from Fig. 3.5:2, we see that only half of the lift
is generated immediately following the start of motion. The lift then grows
with time. When 20 semichord lengths are traveled, the lift reaches about 94%
of the steady-state value.
By the same token, if the wing oscillates, then the unsteady effect will cause
the lift to have a phase lag behind the angle of attack. The theories of
Theodorsen (1935), Kiissner (1936), and Biot (1942) refer to oscillating wings
of infinite span in a two-dimensional flow at speeds much lower than the speed
of sound. For simplicity, the fluid is assumed incompressible and inviscid. The
78 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

wing is assumed to be planar and thin and the amplitude of oscillation is so

small that the velocity of the wing due to oscillation is small compared with
a constant flight velocity of the wing relative to the fluid at infinity. The
Reynolds number of flow is assumed to be much larger than 1, and the effect
of fluid viscosity is taken into account by the Kutta condition (Sec. 3.8).

Vertical Translational Oscillations

Using complex representation of harmonic motion, the wing surface is
described by
(6)
where Yo is a dimensionless amplitude representing the ratio of the amplitude
of vertical motion to the semichord b. U is the speed of flight, ro is the circular
frequency of oscillation in rad./sec. The reduced frequency k is equal to rob/U.
The solution involves a function C(k) introduced by Theodorsen (1935):
Kl(ik) .
C(k) = K 1 (ik) + Ko(ik) = F(k) + zG(k). (7)

where K o, Kl are modified Bessel functions of the second kind of orders zero
and one, respectively, with arguments ik. The function C(k) is often referred
to as Theodorsen's function. Its numerical value is given in Fig. 3.5:2. The
complex amplitude of the total lift per unit span, Le iwt, is given by

L = npU 2 Yok 2 b [1 - ~ C(k) J. (8)

The moment about the midchord point is (positive in the nose-up sense)
Ml/2 = -npU 2 iY okb 2 C(k). (9)
A comparison between Eqs. (8) and (9) shows that part of the lift that is
proportional to C(k) has a resultant acting at the i-chord point. This part of
the lift can be identified as that caused by the bound vorticity over the airfoil.
The other part of the lift has a resultant that acts through the midchord point.
This latter term arises from a noncirculatory origin, and is equal to the product
of the apparent mass and the vertical acceleration. For a flat plate the mass
of the fluid enclosed in a circumscribing cylinder having the airfoil chord
as a diameter is the theoretical apparent mass associated with the vertical
motion.

Rotational· Oscillations
The skeleton airfoil, which executes rotational oscillation with a small ampli-
tude about the origin (the midchord point), is represented by the equation
(10)
3.5 Transient Fluid Dynamic Forces Due to Unsteady Motion 79

\
1.0

0.9
\ :\..
"
0.8
F(k)
~

'" --
0.7 ........

0.6 r----
0.5
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Reduced frequency, k

0.20
r - ~ .......
J
--~
I
--
0.16
.............. ........
0.12

/
...............
G(k)
0.08
I
0.04

o
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Reduced frequency, k

FIGURE 3.5:2 The real and imaginary parts of Theodorsen's function F(k) and G(k).
Note the difference in vertical scale in these two figures. F(k) tends to t and G(k) tends
to zero as k tends to infinity.

Writing
• drx
rx = dt' etc., (11)

then the theory yields

L= npUbri. + 2npU 2 b (1 + i~) C(k)rx, (12)

(13)

Comparing the expressions Land Ml/2' we see that the term npUri represents
80 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

a lift that acts at the i-chord point, the term proportional to C(k) represents
a lift that acts at the i-chord point, and the term (np/8)61 is a pure couple. It
can be shown that the term proportional to C(k) represents the lift due to
circulation. The other two terms are of noncirculatory origin.
In both translation and rotation cases, the lift due to circulation can be
written as
L1 = -2npUC(k)w 3 /4 (14)
where W 3/4 stands for upwash at the i-chord point. Thus the upwash at the
i-chord point has a unique significance. For this reason, the i-chord point is
called the rear aerodynamic center.
The theory of oscillating wings has been developed along several directions.
Theodorsen (1935) gave the first rigorous solution to the problem of oscillating
wings of infinite span in a two-dimensional flow of an incompressible inviscid
fluid by conformal mapping. Theodorsen and Garrick (1942) extended the
theory to a wing-aileron-tab combination. Kiissner (1936) and Kiissner and
Schwarz (1940) solved the same problem by the method of superposition of
singularities. Biot (1942) presented a simplified solution using acceleration
potential.
Then the theories were extended to compressible fluid, and to arbitrary
motion of the wing, while experimental results were accumulated.
The indicial response to gust and sudden motion was first solved by
Wagner (1925) and developed by Kiissner (1940), Sears (1941), and many
others.
For literature review, we again refer to Fung (1955, 1969), Bisplinghoff and
Ashley (1962), and Dowell et al. (1978). Experimental results are reviewed in
these references also. New experimental results are surprisingly lacking. It is
hoped that advances in laser velocimeter, ultrasound, and data handling
techniques will soon yield important results.
Further theoretical analyses of more complex situations, especially in the
nonlinear world, are needed. For example, to study fish propulsion, it is
necessary to consider the interaction of the unsteady motion of neighboring
fins. Since the vortex sheets shed by fins in the front interact with the motion
of fins in the back, the effects could be quite complex. See Chapter 4. The
nonlinear interaction of the vortex sheets has not been explored in detail.
Great advancements in computational fluid mechanics are sure to bear fruits.

3.6 Flutter
One possible effect of the phase shift between motion and force is to cause an
important phenomenon of flutter, which is defined as a self-excited oscillation
of the body. Flutter ofleaves of trees, flags, and tents are familiar to all of us.
For birds, insects, fish, and aircraft, the flutter phenomenon imposes another
limitation to their possible speed of motion, in addition to divergence and loss
3.6 Flutter 81

of control studied in the preceding sections. Power availability is not the only
factor that determines speed.
To understand flutter, it may be useful to consider a wind tunnel experi-
ment. Let a wing be mounted in a wind tunnel. When the wind speed is zero
and the model is disturbed by a poke with a rod, oscillation may set in, which
is gradually damped. When the speed of flow in the wind tunnel is increased
continuously, the rate of damping of the oscillation will first increase, then
decrease. Eventually, at a critical speed of flow the damping becomes zero. At
the critical flutter speed, a disturbed wing will oscillate at a steady amplitude.
At speeds above the critical, an accidental disturbance can trigger a violent
oscillation which is flutter.
Thus it is important to understand flutter. As a first step, let us use dimen-
sional analysis to identify the relevant parameters, as was done in Sec. 3.2.
In addition to the variables listed in Eq. (3.2:1a) let us also consider a, a
characteristic material density of the wing structure of dimensions [ML -3],
and a characteristic torsional stiffness constant of the wing, K, of dimension
[ML 2 T- 2 ]. Out of the five variables L, U, p, a, and K, two independent
nondimensional parameters can be formed, e.g.,
p K
, aL 3 U 2
(1)
a •

Any nondimensional quantity relating to the motion can be expressed as a

function of these parameters. Thus, if, in a free oscillation, the deflection at a
point on the wing is described by an expression e-· t cos rot, the damping factor
Il, of dimension [T- 1 ], can be combined with U and L to form a non-
dimensional parameter IlL/U, which then must be a function of the parameters
listed in Eq. (1):
(2)
At the critical flutter condition, Il = 0, the right-hand side of Eq. (2) vanishes.
Hence a relation exist between the parameters p/a and K/(aL 3 U 2 ). This
relation may be written as

2
Uflgtter
K
= aL 3 f
(p)
-;; (3)

which says that the square ofthe critical speed of flight is directly proportional
to the torsional stiffness of the wing, inversely proportional to the cube of the
wing dimension, and inversely proportional to the wing material density. The
constant of proportionality is a function of the ratio of the densities of the
wing material and the fluid.
The frequency of flutter oscillation ro (radians per second at flutter), with
dimension [T- 1 ], can be expressed nondimensionally in the parameter

k = roL (4)
U
82 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

which is the reduced frequency or Strouhal number (Sec. 3.2). Hence the
Strouhal number of flutter is a function of pl(J, and KI((JL 3 Uju).
A physical interpretation of the Strouhal number is as follows. If a periodic
deflection occurs at a point on a body while the fluid moves downstream with
a velocity U, then the spacing, or the "wave length" of the disturbance in the
fluid, is 2nUIw. The ratio of the characteristic length L of the body to this
wave length is the Strouhal number. Thus the Strouhal number characterizes
the way a disturbance at one point is felt at other points in the flow field.
We can show that the phase shift between force and motion is the cause of
energy exchange between a wing and the surrounding fluid. Consider a wing
in horizontal flight performing a vertical translational oscillation, whose
downward velocity h(t) is
(5)
The lift force created by this motion is oscillatory and out of phase with h. We
may write the lift as
(6)
where t/J is the phase angle by which the lift leads the deflection.
When the airfoil moves through a distance dh, the work done by the lift is,
in real variables,
dW = -Ldh = -Lhdt. (7)
It must be recognized that, when Land h are expressed in the complex form
as in Eqs. (5) and (6), the physical quantities are represented by the real parts
of the complex representations. Thus, in complex representation,
dW= -Rl[LJ·Rl[hJdt. (8)

Integrating through a cycle of oscillation, we obtain the work done by the air
on the airfoil per cycle:

W = - f 21t/W
0 Lo cos(wt + t/J)h o cos wt dt
(9)
n .
= -- Loh o cos t/J.
w
Hence, the gain of energy W by the airfoil from the airstream is proportional
to ( - cos t/J). If - nl2 < t/J < n12, W is negative; i.e., the oscillating airfoil will
lose energy to the airstream. t/J can be evaluated by comparing Eqs. (5), (6)
with Eqs. (3.5:7), (3.5:8). It is seen that free vertical translational oscillation will
be damped.
It can be shown (Fung, 1955) that free pitching oscillation of the wing will
also be damped. However, a wing moving with a combination of translation
and rotation can, under certain conditions of frequency and amplitudes, gain
energy from the surrounding fluid stream which sustains flutter. In aircraft,
3.7 Kutta-Joukowski Theorem 83

flutter is dreaded, and it is the designer's duty to know the critical flutter speed
accurately, and the pilot's duty never to exceed it. In nature, it is not known
whether it is used to advantage by some animals.

3.7 Kutta-Joukowski Theorem

The rest of this chapter is devoted to the wing theory. In the idealized case,
we assume that the wing has an infinite span and a cylindrical body; the fluid
is incompressible; and the Reynolds number is so large that the boundary
layer is very thin and the fluid can be considered as nonviscous outside the
boundary layer. The flow field is two-dimensional; the velocity component in
the spanwise direction is zero. A rectangular Cartesian system of coordinates
xyz, with origin fixed in the wing, and the z-axis in the direction of the span,
perpendicular to the direction of flight, will be used.
Under these assumptions, the Eulerian equations of motion (see Sec. 1.7)
are simplified to
au au au 1 op
- + u- + v- + - - = 0, (la)
at ox oy pox
ov ov ov 1 op
-+u-+ v-+--=O, (lb)
at ox oy pay
in which u, v denote the components of the velocity vector of the fluid in the
x, y direction, respectively; p is the density of the fluid, and p is the pressure.
The equation of continuity for an incompressible fluid (see Sec. 1.7), describing
the law of conservation of mass, is
au ov _ 0 (2)
ox + oy - .
The curl of the velocity field v is the vorticity of the field. If the vorticity
vanishes in the whole field, then the flow is said to be irrotational. If the flow
field is irrotational we have

au _ ov = O. (3)
oy ox
Equations (2) and (3) can be solved for specified boundary conditions. Then
Eq. (1) can be used to compute the pressure distribution. It can be shown that
the solution for a flow field satisfying a suitably specified boundary condition
is unique. Hence any method that yields a solution provides the right solution.
By direct substitution, we see that Eq. (2) can be satisfied by an arbitrary
function J/!(x, y) if the velocity components are calculated according to
oJ/! oJ/! (4)
U=-
oy and v = - ox'
 84 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

Equation (3) can be satisfied by an arbitrary function rjJ(x, y) if

orjJ orjJ
u = ox and v = oy' (5)

tf; is the stream function. rjJ is the velocity potential. On substituting (5) into (2),
and (4) into (3), we see that
02rjJ 02rjJ o2tf; o2tf;
ox2 + oy2 = 0 and ox2 + oy2 = O. (6)

Equations (6) are Laplace equations which are also known as harmonic equa-
tions; their solutions are harmonic functions. Thus the stream and potential
functions are both harmonic functions. One method of solving the flow
problem is to look for a harmonic function that satisfies the boundary con-
ditions. It so happens that the real and imaginary parts of any analytic
function of a complex variable z = x + iy are harmonic. Thus, a famous
method of solving the flow problem is to look for an analytic function of
a complex variable, w(z) = rjJ(x, y) + itf;(x, y) that satisfies the boundary
conditions.
The following examples are given in every textbook:
w(z) = (U - iV)z, rjJ = Ux + Vy, tf; = - Vx + Uy, (7)
w(z) = mlnz, rjJ = mlnr, tf; = me, (8)
ir r r
w(z) = -lnz, rjJ = - 2n e, tf; = 2n lnr , (9)
2n

11 I1 X I1Y
w(z) = --, rjJ= + y2" (10)
Z x2 + y2' tf; = x2
Equation (7) represents a uniform flow with velocity components U and V.
Equation (8) represents a source of strength 2nm per unit length. Equation (9)
represents a vortex with circulation r. Equation (10) represents a doublet of
strength 1111. The streamlines of these flows are illustrated in Fig. 3.7:1.

..

(a) (b) (e) (d)

FIGURE 3.7:1 Streamlines of (a) a uniform flow, (b) a source, (c) a vortex, (d) a source-
sink doublet.
3.7 Kutta-Joukowski Theorem 85

Since the governing equations (2), (3), and (6) are linear, a solution can be
superposed to obtain new solutions. Thus the flow past a circular cylinder of
radius a without circulation can be obtained by superposition of a uniform
flow and a doublet:
a2
w(z) = Uz + U-. (11)
z
The flow about a noncircular cylinder can be obtained by a superposition of
sources and sinks and a uniform flow. The flow past a circular cylinder of
radius a with a clockwise circulation of r can be obtained by adding a vortex
to Eq. (11)

w(z) =U (z + az 2
) + ir
2n
lnz. (12)

The streamlines in the last case are illustrated in Fig. 3.7:2.

We now integrate the equations of motion to obtain the pressure p and
then calculate the total force acting on a solid body moving in a fluid. For an
irrotational flow of an incompressible fluid, this is quite easy to do. For the
derivatives au/at, av/at we use the velocity potential defined in Eq. (5). We
then multiply Eq. (la) by dx and Eq. (lb) by dy, then add and integrate the
result along an arbitrary curve C in the fluid. The result is zero because the
right-hand sides of Eqs. (la), (1 b) are zero. Thus

r [(~ax arjJat + u axau + v ayau + ~paxap ) dx

Jc
a arjJ av av 1 ap ) ]
+ ( ay at + u ax + v ay + p ay dy = O.
This can be reduced to

r d (arjJ + u
Jc at
2
+ v2 +
2
f) = o.
p

-u

FIGURE 3.7:2 Streamlines of a flow around a circular cylinder with circulation. This
flow can be obtained by a proper superposition of a vortex, a doublet, and a uniform
flow of velocity u. If the circulation is r, a lift equal to pur is created.
86 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

Hence
(13)

Since the integration is taken with respect to x and y and not to t, the constant
could be a function of time, that can be absorbed in ar/J/at. In the case of steady
flow, the time-dependent term vanishes, and the last equation, after multi-
plying through by p, and writing V 2 for u 2 + v2, becomes the well-known
Bernoulli equation
V2
PT + p = const. (14)

Thus the pressure at any point in the field of flow is given by

p = const -1PV 2 • (15)
Apply this result to the flow about a circular cylinder specified by Eq. (12).
Consider the forces acting on the wall of the cylinder. On an element oflength
ds located at polar coordinates (a, B), as shown in Fig. 3.7:2, the pressure force
pds is directed radially; its components in the x and y directions are
- p cos Bds and - p sin Bds,
respectively. By an integration over the whole circle, we obtain the x and y
components of the resulting force acting on the cylinder per unit length in the
spanwise direction:

Fx = - f:" pa cos BdB, Fy = - L2" pa sin BdB. (16)

To calculate Fx , Fy , we substitute p from Eq. (15), and compute V 2 from Eq.

(12). Now, since
w(z) = r/J + it/!, (17)
dw ar/J .at/! .
dz = ax + 1ax = u - lV, (18)
we have
(19)

From Eq. (12) we obtain

dw = U
dz
(1 _ a2 )
z2
+ ir ~.
2n z
(20)

On the circle z = ae i8 = a(cos B + i sin B), we have

/~; /2 = [U(1 - cos2B) + r;!:BJ + [usin2B + 2:a cosB J

. ( r)2 +
= 4U 2 sm 2 B + -
2na
2Ur.
--smB.
na
(21)
3.8 The Creation of Circulation Around a Wing 87

Clearly Idw/dzl 2 or V 2 at (a, 0) is equal to V 2 at (a, n - 0), hence p is symmetric

with respect to the y-axis. But cos 0 is antisymmetric with respect to the y-axis.
Hence Fx = O. On the other hand, sin 0 is symmetric with respect to the y-axis;
hence the contributions ofthe right and left half of the cylinder to Fy are equal,

f
and
1<12
Fy = -2 pasinOdO. (22)
-1<12

On substituting (15), (19), and (21) into (22), and noting that the integrals of
sin 3 0, sin 0 from -n/2 to n/2 are zero, whereas that of sin 2 0 is n/2, we obtain
Fy = pur. (23)
This shows that the cylinder experiences a lift force equal to the product
of the velocity of flight U, the circulation r, and the density of the fluid p.
There is no drag. The doublet makes no contribution to the lift force. Similarly,
by integration around the wall of a cylinder it can be shown that any enclosed
sources and sinks make no contribution to the lift and drag. Although these
sources and sinks define the shape ofthe wing, they do not affect the lift. Thus
we obtain the famous Kutta-Joukowski theorem:
The force per unit length acting on a cylindrical wing of any cross-section
whatever is equal to pur and acts in a direction perpendicular to U.
Thus, the lift is proportional to the strength of circulation of the vortex line
r, and to the relative velocity U with which the vortex line moves with respect
to the free stream. If the vortex moves with the free stream, then U = 0 and
there will be no lift.

3.8 The Creation of Circulation Around a Wing

The Kutta-Joukowski theorem tells us that the lift per unit span acting on
the wing is equal to pur. But what determines r, the circulation around the
wing, and how is it created?
Insects, birds, and fish solved the problem of creating circulation around
the wing by providing the wing with a sharp trailing edge. The inventors of
the airplane copied the insects and birds and made aeronautics a success.
The importance of a sharp trailing edge can be seen in the illustrations of
Fig. 3.8:1. The airfoil is stationary and the flow comes from the left. Figure
3.8:1(a) shows streamlines of a potential flow around the airfoil. Among all
the streamlines there is one that ends on the body, and defines a front
stagnation point and a rear stagnation point on the body. If the rear stagnation
point does not coincide with the trailing edge, then the flow must turn around
that very sharp corner. This will create a very high velocity gradient and hence
a large shear stress, which will tend to retard the flow around the corner,
causing the rear stagnation point to move toward the trailing edge. The
process will continue until the rear stagnation point coincides with the trailing
edge, as shown in Fig. 3.8: 1(b). Such a flow requires a certain amount of
88 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

-------------
~----

(a) (b)

FIGURE 3.8:1 Flow around an airfoil with a sharp trailing edge. (a) Streamlines about
an airfoil starting to move. The circulation has not been established yet. (b) Streamlines
about an airfoil in steady motion.

circulation r. Thus r is fixed by the condition that the rear stagnation point
coincides with the trailing edge. A mathematical statement of this fact is
known as the Kutta condition, which says that a body with a sharp trailing edge
moving through a fluid will create about itself a circulation of sufficient strength
to hold the rear stagnation point at the trailing edge.

3.9 Circulation and Vorticity in the Wake

So far we have considered two-dimensional airfoils in steady flow. To under-
stand finite wings in oscillatory motion, we must go deeper into fluid
dynamics. The key to this matter is Kelvin's theorem concerning circulation
in the fluid. The circulation r( C) associated with any closed circuit C in the
fluid is defined by the line integral:

r(C) = L Lv·dl = v;dx;, (1)

where the integrand is the scalar product of the velocity vector v (with
components VI' V 2 , v 3 ) and the vector dl (with components dx l , dx 2 , dx 3 ),
which is tangent to the curve C and oflength dl (Fig. 3.9:1). The rate of change
ofr(C) with respect to time, when C is a fluid line (i.e., a curve formed by the
same set of fluid particles at all times), and the fluid is nonviscous and
barotropic (fluid density is a unique function of pressure) is given by the Kelvin
theorem, which states that

Dr =0 (2)
Dt
if the external force field is conservative.
To prove this theorem, we note that since C is a fluid line composed always
of the same particles, the order of differentiation and integration may be
3.9 Circulation and Vorticity in the Wake 89

FIGURE 3.9:1 Quantities defining circulation.

d.

DdS

i i i
interchanged in the following integral:

-D v·dx· =
Dt c I I C
D
-(v.dx.)
Dt I I
=
C
(DV.
-'dx.
Dt I
+ v.-_
DdX.)
Dt
I
I
• (3)

But D(dx;)/Dt is the rate at which dx; is changing as a consequence of the

motion of the fluid; hence it is equal to the difference of the velocities parallel
to X; at the ends of the element, i.e., dv;. Hence the last term in Eq. (3) is V; dv;.
Further, DvdDt is the fluid particle acceleration, which, according to the
Eulerian equation of motion, is equal to the sum of the negative pressure
gradient and the body force per unit volume, pX;, divided by the density, p.
Thus Eq. (3) becomes

Dr =
Dt Jcr [(_2..p ox;
op + X.)dX. + V.dV.]
I I I I (4)

= _ r dp + r X; dx; + Jcr dv
Jc p Jc
2•

Of the terms on the right-hand side, the first vanishes because the fluid is
barotropic (p is a unique function of p) and C is a closed curve; the second
term vanishes because the external force field X; is assumed to be conservative;
the third term vanishes because the value of the line integral is equal to v2 at
a point on C minus v2 at the same point after one goes around C in a closed
circuit. Thus we get Eq. (2). Q.E.D.
Applying Eq. (2) to the flight of birds and insects, we note that since the
Reynolds number is much larger than 1, the boundary layer is thin, and
outside the boundary layer the fluid may be considered to be nonviscous. The
Mach number of bird's flight is small compared with 1; hence the air may be
considered incompressible (thus barotropic). Gravity is the only external force
90 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

and is conservative. Hence all the assumptions for the Kelvin theorem are
valid, and Eq. (2) applies. Now, consider a wing starting to move from a
stationary position (Fig. 3.9:2). The circulation r about any fluid line outside
the boundary layer is zero because it is zero before the motion takes place,
however, the volume occupied by the airfoil and the boundary layer must be
excluded. A fluid line C enclosing the boundary of the wing becomes elongated
when the wing moves forward as shown in Fig. 3.9:2. According to Kelvin's
theorem, the circulation about C is zero. But one cannot conclude that the
vorticity actually vanishes everywhere inside C. In the region occupied by the
wing, and in the wake behind the wing, vorticity does exist if there is a lift force
acting on the wing. Although Fig. 3.9:2 shows I(C) = 0, Fig. 3.9:3 shows that
I(C) may be regarded as the sum of the circulations I(C') and I(C"); where
C' + C = C. If nC') = r, then nC") = -r, so the circulation about the
If

entire wake is equal and opposite to that around the wing.

A curve enclosing
the airfoil, ~

.. .of ~
Direction
OIrfoll motion
~--~
-=.. - --:- ---
(

Curve formed by fluid particles that

constitute the original curve ~~
~~-~~.-------------------~-)
FIGURE 3.9:2 Fluid line enclosing an airfoil and its wake. Top: A stationary airfoil
starting to move. Bottom: The airfo'il has moved forward for a distance, and the fluid
line is elongated.

FIGURE 3.9:3 The circulation about the curve c in Fig. 3.9:2 is the same as that about
the curve c' + elf in this figure. The circulation about the wing (c') is equal and opposite
to that about the wake (elf).
3.10 Vortex System Associated with a Finite Wing in Nonstationary Motion 91

3.10 Vortex System Associated with a Finite Wing in

Nonstationary Motion

An important result deducible from Kelvin's theorem is the Helmholtz theorem

which states that the vortex lines in an inviscid fluid move with the fluid,
meaning that the fluid particles constituting a vortex line will continue to
constitute a vortex line as the fluid flows. We present a proof by Yih (1969).
The proof is facilitated by noting that according to Gauss' theorem, the
circulation defined by Eq. (3.9:1) can be expressed as a surface integral:

nC) = L curlv·vdS, (1)

where S is a surface bounded by C, and v is the unit normal vector of the

surface. Curl v is the vorticity of the flow field. Thus, circulation is the sum of
the normal component of vorticity over a surface.
Now let us consider a continuous field of vorticity, Fig. 3.10:1(a). A vortex
line is a line tangent to the vorticity vectors. On a point A on a vortex line,
draw two lines BAC and DAE intersecting the vortex line, and draw vortex
vectors passing through BAC and DAE. These vectors form two surfaces, Sl
and S2' that intersect in the vortex line under consideration. According to Eq.
(1), the circulation along any closed circuit on Sl or S2 is zero. According to
Kelvin's theorem, it will continue to be zero as the fluid moves. Thus the
circulation along any closed circuit situated on the surface Sl or S2 remains

(a) (b)

FIGURE 3.10:1 (a) A vector field of vorticity. The arrows represent the vorticity at
the points where the arrows start. (b) A circuit on a vortex tube. When D and A
and D' and A' coalesce, there results two circuits on a vortex tube.
92 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

zero. Hence S1 and S2 remain vorticity surfaces consisting of vortex vectors.

Their intersection will continue to be a vortex line, for otherwise at some point
on it a vortex vector would intersect at least one of the surfaces S1 or S2'
contradicting the fact that the circulation along any circuit in S1 or S2 is zero
for all times. Q.E.D.
A vortex tube is formed by vortex lines, Fig. 3.1O:1(b). If a circuit ABCD-
D'CB' A' is drawn on the vortex tube, the circulation along this circuit is zero.
If the points D and A as well as D' and A' coalesce, we obtain two circuits
ABCD and D'CB'A', of which the sum of the circulation is zero. Thus r of
ABCD and r of D' C B' A' are equal and opposite; r of ABCD and r of A' B' CD'
are exactly equal. Therefore the circulation around any vortex tube does not
change.
It follows that a vortex tube cannot end in a fluid, because the circulation
is zero at the end, which, according to the result just derived, requires the
circulation to be zero along the entire vortex tube; a trivial case. Thus, a vortex
line can end only on a solid surface, or on the boundary of the fluid. It can
end on itself, forming a ring, or extend to infinity if the fluid field is infinite.
With the Helmotz theorem, we can now understand the vortex system
associated with a real wing. Let us approach the real wing step by step.

Horseshoe Vortex
Consider a rectangular wing in a steady flow with free stream velocity U.
Attach a frame of reference that moves with the wing so that the wing appears
stationary while the free stream comes from the left. In Fig. 3.1O:2(a) the wing
is represented by a bound vortex. Since the vortex line cannot end at the wing
tips, it has to turn around and move with the free stream, forming the trailing
vortices. Together, the bound and traling vortices look like a horseshoe. The
circulation is constant along the entire vortex line.
If the wing planform is elliptical and untwisted (having a constant angle of
attack across the span), it can be shown that the lift distribution is also
elliptical across the span (see Fig. 3.1O:2(b)). Then the wing can be represented
by a series of bound vortices, the total strength of which varies elliptically
across the span. Since each bound vortex has two trailing vortices, the wing
and its wake can be represented by a series of horseshoe vortices. The trailing
vortices interact with each other, and have a tendency to roll up, as indicated
in Fig. 3.10:2(b).

Downwash
Each vortex line induces a velocity field (see Fig. 3.10:3). For simplicity we
show only the uniform free-stream velocity U, and the velocity on the vortex
lines induced by themselves. Since the trailing vortices move with the free
stream in the direction of U, the induced velocity is perpendicular to U, and
3.10 Vortex System Associated with a Finite Wing in Nonstationary Motion 93

L Trailing vortex

(a)

Trailing vortices Rolling up

(b)

FIGURE 3.10:2 (a) A horse-shoe vortex representing a wing with a uniform lift distribu-
tion. (b) Lift distribution on an elliptic wing.

is called the down wash. In Fig. 3.1O:3(a), the downwash over a rectangular
wing is seen to be nonuniform across the span. In Fig. 3.10:3(b), the downwash
over an untwisted elliptical wing is seen to be uniform across the span.

Induced drag: The Price you Pay for the Lift

In Sec. 3.2 we discussed the effect of the downwash on the wing due to trailing
vortices on the slope of the lift coefficient versus the angle of attack. Here we
shall consider its effect on the drag. As shown in Fig. 3.2:3, the down wash
velocity w deflects the free-stream velocity vector U by an angle e, with
94 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

(a) Horse-shoe vortex

(b) Elliptic wing, uniform downwash

FIGURE 3.10:3 (a) Nonuniform downwash on a horse-shoe vortex. (b) Uniform down-
wash on an elliptic wing.

tan B = wjU. Relative to the wing, the free stream comes in the direction of
the velocity U res . When the circulation is r, the resulting force pUresr acts in
a direction normal to the velocity vector U res . It can be resolved into a lift
component L perpendicular to the velocity of flow U, and a drag component
D in the direction of U. From Fig. 3.2:3, we see that
w
D=LtanB=-L U . (2)

This drag is induced by the down wash as a consequence of creating a lift.

Known as the induced drag, it accompanies the lift, and is a price a flying
animal must pay for the lift.
In Fig. 3.10:3 it is shown that the downwash w is uniform across the span
of an untwisted elliptical wing, whereas it is nonuniform if the wing planform
is not elliptical. This is a result derived by method described in Sec. 3.12.
Glauert showed that for a given total lift the wing with a uniform downwash
yields the minimum induced drag.
For an elliptic lift distribution, it has been shown theoretically that the
down wash is given by
W CL
(3)
U nAR'
where CL is the lift coefficient and AR is the aspect ratio (see Sec. 3.2). Hence
3.10 Vortex System Associated with a Finite Wing in Nonstationary Motion 95

the induced drag given by Eq. (1) is equal to

D. = Cl !pU 2 S or Di = CL (weight). (3)

'n 2' n
S is the wing area and W is the weight ofthe flying object. For a lift distribution
other than elliptical, the induced drag is greater than this. Equation (3) shows
that the induced drag is reduced if the aspect ratio is increased. This is
undoubtedly why a soaring bird spreads out the wing to make its aspect ratio
as large as possible (see Fig. 3.2:1(b)).

Oscillating Wings
If the wing oscillates, either by flapping up and down, or by changing its angle
of attack periodically, then the strength of the bound vortex varies with time,
and there will be vortices shed from the trailing edge into the wake. A crude
sketch is shown in Fig. 3.10:4. Every time an increment of bound vortex is
created, an equal and opposite vortex is left to the wake. Thus the vortex
structure in the wake becomes quite complex. The complexity is increased if
the amplitude ofthe oscillation is large. We can then appreciate how difficult
it is to analyze the flight of birds and insects (see Fig. 3.10:5). Fortunately, the

Lora~ ~
Time, or space ~

FIGURE 3.10:4 The vortices in the wake of an oscillating wing, idealized under the
assumption that the lift fluctuation is very small so that the distortion of the wake due
to the vortices in it is also very small. L is the lift, IX is the angle of attack.

FIGURE 3.10:5 The vortex wake behind a stork in level flight.

96 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

reduced frequency (or Strouhal number based on radian frequency and chord
length) for birds and insects is relatively small (in the order of 0.2 to 0.6), and
it is often permissible to use the quasi-steady approximation to obtain rough
estimates. The same quasi-steady approximation would be less valid for the
lunate tails of fish such as a shark or whale. These tails function as wings, but
operate at Strouhal numbers of the order of 1.

3.11 Thin Wing in Steady Flow

Birds can change their wing planform and airfoil cross-section. The concepts
developed above can be used to estimate the effects of these changes.
Consider first the effect of airfoil cross-section. Let the wing be very thin
as shown in Fig. 3.11:1; it has an infinite span, and the flow is two-dimensional.
Let us use a rectangular cartesian frame of reference with the origin located
at the leading edge, x-axis along the chord line, and y-axis perpendicular to
it. The airfoil camber line is described by the equation
y = Y(x), (0 ~ x ~ c). (1)
If a steady two-dimensional flow with undisturbed speed U and angle of attack
(J. streams past the airfoil, disturbances are introduced into the flow by the

airfoil in such a manner that the resulting flow is tangent to the airfoil. The
thin airfoil can be replaced by a continuous distribution of vortices. Let the
strength of the vorticity over an element of unit length in the span wise
direction and of length dx in the chordwise direction be y(x) dx. According to
the Kutta-loukowski theorem, the lift force contributed by the element
dx is
dL = pUy(x) dx. (2)
The total lift per unit span is therefore

L = pU J: y(x)dx, (3)

where c is the chord length of the airfoil.

~~r--------~~~~~----~~~~------x

FIGURE 3.11:1 Steady flow over a two-dimensional airfoil.

3.11 Thin Wing in Steady Flow 97

For a thin airfoil ofsmall camber, Y(x)« c, the surface of the airfoil differs
only infinitesimally from a flat plate. The induced velocity over the airfoil
surface, to the first order of approximation, can be calculated by assuming
that the vortices are situated on the x-axis. Since the velocity at x induced by
e
a vortex of strength y(e)de located at is y(e)de/[2n(e - x)], the ycomponent
of the induced velocity at a point x on the x-axis is
r c y(e)de
Vi(X) = Jo 2n(e - x)'
(4)

which, to the first order of approximation, is the same as the component of

velocity normal to the airfoil surface at the chordwise location x. The slope
of the fluid stream on the airfoil is then IX + vdU. This must be equal to the
slope of the airfoil surface dY/dx. Hence, the boundary condition on the airfoil
is

IX+-=-. Vi
U
dY
dx
(5a)

The vorticity distribution y(x) must be determined from Eqs. (4) and (5a). Thus
we obtain the integral equation for the vorticity distribution y(x):

rCy(e)de = 2nu(dY
Jo e- x dx
-IX). (5b)

In addition, the Kutta condition y(c) = 0 must be satisfied, i.e., the fluid must
leave the trailing edge smoothly.
Equation (5a) is a singular integral equation of the Cauchy type which has
been treated most thoroughly by Muskhelishvili (1953a,b). The methods of
Glauert, Lotz, Hilderbrand, Multhopp, and Sears are also well known. In
Glauert's method, we introduce a new independent variable t/I so that
c
x = "2(1 - cost/l). (6)

When x varies from 0 to c along the chord, t/I varies from 0 to n. The vorticity
distribution can be written as

y = 2U(Aocott + ~ An sin nt/l) (7)

with unknown coefficients, A o, Al .... Substituting (7) into (4), we obtain (see
Glauert, 1947)

Vi = U( -Ao + ~ An cos nt/l). (8)

Equation (5) then implies

dY
Ao + L Ancos nt/l
00
IX - = -d • (9)
I x
98 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

The left-hand side is a Fourier series. The coefficients can therefore be deter-

°
mined by the usual method. Multiplying Eq. (9) by cos nt/l (n = 0,1,2, ... ), and
integrating from to n, we obtain

IX - Ao = -1 f"dY
- dt/l, An = -2f"dY
-d cosnt/ldt/l (10)
n 0 dx n 0 X

and the problem is solved.

From Eqs. (3) and (7), the total lift can be obtained. The result, expressed
as the lift coefficient, is
(11)
Similarly, the moment about the leading edge, expressed as the moment
coefficient, is
(CM)"e. = -~(Ao + Al - tA 2 ) = -*(A 1 - A2 ) - tCL. (12)
= 0, Ao = IX, the lift curve slope is seen to be 2n.
If the airfoil is a flat plate, Y
From Eq. (12), the lift force is seen to act through the i-chord point.

3.12 Lift Distribution on a Finite Wing

To analyze the stress and strain in a swimming fish or a flying bird, one of the
most important problems is to know the span wise distribution of lift and
moment, which must be resisted by their bones, muscles, and tendons. We shall
present a classical approximate theory credited to Prandtl. The solution also
yields information on down wash and induced drag that is needed to evaluate
the energy cost of locomotion.
Theoretically, a wing can be replaced by a system of vortices as discussed
in Sec. 3.11. For example, if a wing is straight, without significant sweepback
or sweepforward, then in first approximation it can be replaced by a straight
line vortex. Consider such a wing in a symmetric flight. Let a rectangular
cartesian frame of reference be used, with origin located on the vortex line at
midspan, x-axis in the direction of undisturbed flow, y-axis in the direction of
span. The bound vortex lies on the y-axis from - s to s, s being the length of
semispan. The circulation around the vortex line, r(y), is a function of y. In
a steady flight of velocity U, the lift force acting on a segment of the wing of
width dy at a station y is pUr(y) dy (see Sec. 3.7), where p is the density of
the fluid. If the chord length is c and the slope of the curve of lift coefficient
versus angle attack is a o (see Sec. 3.2), then the angle of attack needed to
generate this lift is
IXo(y) = p~r(~) dy = 2r(y) (1)
aozpV cdy ao Vc
The actual angle of attack is equal to IXo(y) plus the induced angle of attack
3.12 Lift Distribution on a Finite Wing 99

due to downwash, w/U (see Fig. 3.2:2). The downwash caused by the bound
vortex on itself is zero when the vortex line is straight. Downwash caused by
the trailing vortices must be calculated. Now on the segment of the wing of
width dy at station y, the strength of the bound vortex is changed by an
amount (drjdy) dy in the spanwise direction. This change must become the
trailing vortex moving with the fluid stream in the direction ofthe x-axis. Thus
the trailing vortex attached to this segment of the wing at y is of strength
°
(dr/dy)dy, and is a semi-infinite straight line extending from x = to x = 00
if the rolling up of the trailing vortex sheet (due to interaction of the vortices)
is neglected. The downwash at the point (x = 0, y = ~) caused by this trailing
vortex is equal to
1 dr
dw(~) = 4n(~ _ y) dy dy, (2)

which is half of what what would have been induced by a vortex line extending
from -00 to 00 (see Sec. 3.7). Hence the downwash at a point (x = 0, y = y)
due to the whole system of trailing vortices can be obtained by an integration
of dw(y) over the entire span from -s to s:

w(y) = ~ f'
_1_ dr(l1) dl1.
4n -. y - 11 dl1
(3)

Note the change of symbols from ~ to y and y to 11 in using Eq. (2) to arrive
at Eq. (3). The absolute angle of attack required to produce the lift distribution
is, therefore, the sum of cxo(Y) given by Eq. (1) plus w(y)/U, i.e.,

cxa(y) = 2r(y)
aoUc
+ _1_ f'
_1_ dr(l1) dl1.
4nU -.y-l1 dl1
(4)

This is the integral equation used to solve for r(y), subject to the boundary
condition that r(y) must vanish at the wing tips, y = -s and s.
Equation (4) is again a singular integral equation of the Cauchy type, and
can be solved by the same methods mentioned in Sec. 3.11. A solution in the
form of a Fourier sine series is most convenient:
aocoU ~ .
r = -2-- n~l An sm nO, (5)

with y = scos O. Details can be found in Kuethe and Chow (1986).

Applying this analysis to birds and fish, we must make at least two more
extensions: First, real wings have sweepback and sweepforward (see Fig. 3.2:1)
so the lifting line is not a straight line. Sometimes the aspect ratio is not much
larger than 1, making it necessary to use a lifting surface theory in order to
achieve the desired accuracy. For fish, the lifting surface is usually not sym-
metric. Second, the elastic deflection of the structure under load is usually
significant, and the reaction and control by muscle must be taken into account.
100 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

Hence the angle of attack, lia(Y} of Eq. (4), (and similarly, the terms dyjdx and
liin Eq. (3.11:5a) is a function of the muscle function, as well as of bones,
tendons, tissues, feathers, or fins. The method of approach used in the theory
of aero elasticity can be used to account for the elastic deflection and active
control (see Fung, 1969, Bisplinghoff and Ashley, 1967, Dowell et aI., 1978),
however, a thorough work on animals remains to be done.

3.13 Drag

So far we have considered lift, moment, and control which makes flight and
swimming possible. Now let us turn to drag which determines the price an
animal must pay to move in a fluid. Drag means force acting on a body in the
direction of motion relative to the fluid at infinity (if the domain is unbounded).
For an airfoil, a dimensionless coefficient, CD' called the drag coefficient, is
defined by Eq. (3.2:4).
F or an airfoil in an unbounded fluid, drag can arise from skin friction, from
wake due to boundary layer separation, and from downwash due to vortices
associated with lift force generation. Skin friction is shear stress associated

0.020 r - - - - - r - - - - r - - - - r - - - " " " T " " - - - - . - - - - - , . - - - . . ,

.
0.016 ··
0.012 ...'.
'.'.'. .....
'.
'
........
0.008
..'.
0.004

o~ __ __ ___ ____
~ ~ ~ ~~ ____ ____ ____
~ ~ ~

-1.2 -0.8 -0.4 o 0.4 0.8 1.2 1.6

FIGURE 3.l3:1 The drag coefficient plotted against the lift coefficient CL of two NACA
airfoils. Experimental data measured at Reynolds number 6 x 10 6 for smooth airfoil.
Data from Abbott and von Doenhoff(1949}. Courtesy of NASA.
3.13 Drag 101

with fluid viscosity and shear strain rate in the boundary layer attached to
the solid surface of the airfoil. Wake is a front-back asymmetric flow pattern
caused by detachment of the boundary layer from the solid wall. It occurs on
blunt bodies (Sec. 3.3). It occurs also on airfoils when the angle of attack is
too large. When a wake exists, the resulting force due to normal stress acting
on the surface of the body will not be zero in the direction of motion. This
drag is the form drag.
The third source of drag is associated with downwash which is velocity
induced by the vortex system around the airfoil and in the wake. It affects the
effective angle of attack of the airfoil. Lift associated with the change of effective
angle of attack due to downwash is the induced drag (Sec. 3.2). Skin friction,

1.8

1.6 ~
Airfoil Ll004
Turbulent rooftop
Rex =3.'0 X 106

1.2

LIo Clean Plus trans

1.0 <> Trans. strip at 5% } strip at
o Trans. strip at 12% 0.0 5c on
o Trans. strip at 20% lower
c[ .8 surface
'i'Theoretical cd for transition
~ at peak of rooftop .
.6 CD Theoretical cd for transition
at the leading edge .
.4

.2

-2
FIGURE 3.13:2 CL vs CD plots of insect wings and airfoils. Reprinted with permission
from A.M. Kuethe and c.y. Chow, Foundations of Aerodynamics, 4th Ed., copyright
© 1986, John Wiley & Sons, Inc. Data adapted from Nachtigall (1974) and Thorn and
Swart (1940).
102 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

form drag, and induced drag are the three principal components of the total
drag.
It is convenient to consider skin friction and form drag separately from
induced drag. If a wing is cylindrical (does not vary in the spanwise direction)
and is tested in a wind tunnel, the flow around the wing can approach a
two-dimensional condition if large flat plates are installed at the tips of the
wing. Then the aspect ratio of the wing is effectively infinity and the induced
drag tends to zero (see Eq. (3.2:10)). The form drag and skin friction can then
be measured by measuring the velocity distribution around a control volume.
The drag coefficient so measured varies with the angle of attack, hence with
the lift coefficient. Figure 3.13:1 shows the measured drag coefficient-lift coeffi-
cient relationship of two smooth NACA airfoils at a Reynolds number of
6 x 106 . It is seen that CD rises drastically when certain CL value is exceeded,
but it is possible to design the airfoil so as to obtain the "low-drag bucket" at
a designed range. Both of the airfoils shown in this figure are so-called
"laminar flow" airfoils. Figure 3.13:2 shows the CL vs CD relationship of a
number of insect wings and airfoils.
Methods of wind tunnel measurement of drag, and means of achieving high
maximum lift coefficient and low minimum drag coefficient are discussed in
Kuethe and Chow (1986), Liebeck (1978), Smith (1975), Stratford (1959),
Walsh (1980).
The airfoil theory is applicable to the fins and tails of fish. But fish swim-
ming near the surface of the water must bear additional drag arising from
making waves on the free surface of the water (Sec. 4.11).

Problems
3.1 Consider the aerodynamics of racing bicycles and cyclists. In order to strive for
the best, what can you do to reduce the drag of the man-machine combination?
(Cf. Kyle, C. and Burke, E.: Mechanical Engineering, 106: 35-45, Sept. 1984. For
an interesting history of human-powered vehicles, see Wilson, D.G.: American
Scientist, 74: 350-357, 1986.)
3.2 Human-powered flight excites people's imagination. How much has man suc-
ceeded? Looking over the horizon, can you suggest some new improvements?
(Cf. "Human-Powered Flight." Mechanical Engineering, 106: 46-55, 1984.)
3.3 There have been attempts to fly nonstop around the world on a single tank offuel.
What kind of features must such an airplane have? What kind of wing span to
chord length ratio? What kind of engine? What kind of control system? What
kind of safety devices must the plane and flyers have? For a two-seater, how large
and how heavy is the plane likely to be? (Cf. "Another World Aviation Record for
Voyager?" Mechanical Engineering, 108: 41-44,1986.)
3.4 Mechanical and geometric properties of the upper and central airways between
the mouth and carina can be inferred noninvasively in individual subjects from
high frequency acoustic reflection data measured at the month. Write down
Problems 103

the equations of motion, continuity, and constitutive equations, and formulate

the inverse problem of determining mechanical and geometric properties from
acoustic waves. (Cf. Fredberg, 1.1.: Acoustic Determinants of Respiratory System
Properties. Ann. Biomed. Eng., 9: 463-473, 1981. The inverse problem is similar
to one of geophysics, see Ware, I.A. and Aki, K.: Continuous and Discrete Inverse
Scattering Problem in a Stratified Elastic Medium. I. Plane Waves at Normal
Incidence. J. Acoust. Soc. Am., 54: 911-921,1969.)
3.5 To prevent aeroe1astic instabilities such as flutter, stall, and divergence, modern
airplane designers use electronic feedback control to actively control the struc-
tural parameters such as wing stiffness and wing shape (aspect ratio, sweepback or
sweepforward angle) as well as the deflections of the control surfaces. Survey the
parameters and discuss the possibilities of active control for a high performance
airplane.
3.6 Following the thin-wing theory outlined in Sec. 3.11, find the lift distribution on
an airfoil whose cross-section is (1) a straight line, (2) a shallow sine curve,
Y = a sin(nx/c), and (3) an S-shaped curve given by Y = a sin(2nx/c). Is there
any difference in the lift coefficient and the moment coefficient about the leading
edge in these three cases?
3.7 Following the lifting-line theory of a finite wing outlined in Sec. 3.12, determine
wing planforms that will produce (1) an elliptic distribution and (2) a uniform
distribution.
3.8 If the lifting line of Sec. 3.12 is not a straightline, but consists of two straightline
segments meeting at an angle at the middle, representing a wing swept back or
forward, what major change in the theory has to be introduced?
3.9 The vortex system of fish's fins has important effect on the dynamics of force
generation in fish swimming. Make a sketch of the vortices of a fish of your choice,
and discuss the dynamics of the system during fish swimming. What is the
difference in the pattern between normal swimming and a condition for hunting
or fighting-for-life situation? Formulate a rigorous mathematical theory to study
the fluid dynamics of this problem. As a biologist, think of the biological side of
the problem also: the muscle system, the nervous control, the neuronetwork.
Formulate an interesting problem to study.
3.10 Describe the vortex system (bound and free) acting on a flapping wing of finite
span.
3.11 Describe the changes occurring in the vortex system when a bird maneuvers to
pitch, yaw, and roll. Discuss the time lag between a movement of a control surface
and the forces and moments generated on the wing.
3.12 Based on the free and bound vortex system, formulate a mathematical approach
to analyze the two-dimensional lift distribution on a bird's wing (not as a single
straight line) under the assumption that the wing is rigid.
3.13 Under the assumption of a rigid wing (as in Prob. 3.12), determine the forces and
moments acting on the bird.
3.14 Bird's wings are generally swept forward near the body and swept back in outer
span. With the anatomical structure and muscle system taken into consideration,
104 3 External Flow: Fluid Dynamic Forces Acting on Moving Bodies

discuss qualitatively the effect of these sweep angles on the stability and control
of the bird.
3.15 An eagle dives from a high altitude to catch a squirrel. What determines the diving
speed? If it dives too fast, would the speed interfere with the ability to catch the
object? How does the eagle pull out of the dive?
3.16 At high speed of diving, what effect does the Corio Ii's acceleration due to earth's
rotation have on the flight of the bird?

References

Material of this chapter is taken mostly from the author's book An Introduction to the Theory of
Aeroelasticity. Additional references can be obtained from the following:

Abbott, I.H. and von DoenhofT, A.E. (1949). Theory of Wing Sections, Including a Summary of
Airfoil Data, McGraw-Hill, New York. (Paperback edition, Dover, New York, 1959.)
Batchelor, G.K. (1967). An Introduction to Fluid Dynamics. Cambridge University Press,
Cambridge, United Kingdom.
Biot, M.A. (1942). Some simplified methods in airfoil theory. J. Aeronaut. Sci. 9: 186-190.
BisplinghofT, R.L. and Ashley, H. (1962). Principles of Aeroelasticity. Wiley, New York.
Bradshaw, P. (1964). Experimental Fluid Mechanics, MacMillan, New York. Pergamon Press,
(p. 235). Introduction to Turbulence and Its Measurement (1975), Pergamon Press, New York.
Cebeci, T. and Bradshaw, P. (1978). Momentum Transfer in Boundary Layers, McGraw-Hill, New
York.
Chow, c.-Y., Huang, M.-K., and Yan, c.-Z. (1985). Unsteady Flow About a Joukowski Airfoil
in the Presence of Moving Vortices. AI AA J. 23(5): 657-658.
Dowell, E.H., Curtiss, H.C., Scanlan, R.H., and Sisto, F. (1978). A Modern Course in Aeroelasticity.
SijthofT and NoordhofT, Alphen aan den Rijn, The Netherlands.
Fung, Y.c. (1955,1969). An Introduction to the Theory of Aeroelasticity. John Wiley, New York
(1955). Paperback, expanded, Dover Publications, New York (1969).
Glauert, H. (1947). The Elements of Aerofoil and Airscrew Theory. 2nd Ed. Cambridge Univ.
Press, London.
Kovasznay, L.S.G. (1949). Hot wire investigation of the wake behind cylinders at low Reynolds
numbers. Proc. Roy. Soc. London, A, 198, 174.
Kuethe, A.M. and Chow, c.Y. (1986). Foundations of Aerodynamics, 4th Edn. Wiley, N.Y.
Kiissner, H.G. (1936). Zusammen fassender Bericht der instationaren Auftrieb von Fliigeln.
Luftfahrt-Forsch. 13: 410-424.
Kiissner, H.G. (1940). Das zweidimensionale Problem der beliebig bewegten Tragflache unter
Beriicksichtigung von Partialbewegungen der Fliissegkeit. Luftfahrt-Forsch. 17: 355-361.
Kiissner, H.G. and Schwarz, L. (1940). The oscillating wing with aerodynamically balanced
elevator. Luftfahrt-Forsch.17: 337-354. English translation: NACA Tech. Memo 991 (1941).
Liebeck, R.H. (1978). Design of Subsonic Airfoils for High Lift. J. Aircraft 15(9): 547-561. Design
of Airfoils on High Lift, Proc. AIAA Symposium on Aircraft Design, 1980.
Nachtigall, W. (1974). Insects in Flight. H. Oldroyd et aI., trans. McGraw-Hill, New York.
Pai, S.I. (1956). Viscous Flow Theory. Van Nostrand, New York.
Reynolds, O. (1883). An Experimental Investigation of the Circumstances Which Determine
Whether the Motion of Water Shall be Direct or Sinuous, and of the Law of Resistance in
Parallel Channels, Phil. Trans. R. Soc. London, 174: 935-982.
Sears, W.R. (1941). Some aspects of non-stationary airfoil theory and its practical application. J.
Aeronaut. Sci. 8: 104-108.
Smith, A.M.O. (1975). High-Lift Aerodynamics, J. Aircraft 12(6): 501-531.
References 105

Stratford, B.S. (1959). The Prediction of Separation of the Turbulent Boundary Layer, J. Fluid
Mech.5: 1-16.
Theodorsen, T. (1935). General theory of aerodynamic instability and the mechanism of flutter.
N ACA Rept. 496.
Theodorsen, T. and Garrick, I.E. (1942). Flutter calculations in three degrees of freedom. N ACA
Rept.74l.
Thorn, A. and Swart, P. (1940). Forces on an airfoil at very low speeds. J. Roy. Aero. Soc.
44: 761-769.
Van Dyke, M. (1982). An Album of Fluid Motion, Parabolic Press, Stanford, California.
Wagner, H. (1925). Dber die Entstehung des dynamischer Auftriebes von Tragfliigeln. Z. angew.
Math. u. Mech. 5: 17-35.
Walsh, M.J. (1980). Viscous Drag Reduction. Progress in Astronautics and Aeronautics, Vol. 72,
(Gary R. Hough, ed.).
Yih, c.-S. (1969,1989). Fluid Mechanics. McGraw-Hill, New York (1969). West River Press, 3530
West Huron River Dr., Ann Arbor, MI 48103 (1989).
CHAPTER 4

Flying and Swimming

4.1 Introduction

Locomotion is, of course, an extremely interesting subject. People are forever

fascinated by sports. We cheer gold medal winners. How athletes are trained
is certainly a legitimate question for biomechanics. There are people who
suffer impairments in locomotion and others who try to help them recover or
overcome their handicaps. These people, the sports lovers, educators, patients,
orthopedic surgeons, engineers, physical therapists, nurses, prosthesis manu-
facturers, and hospital managers, will benefit from a good understanding of
the biomechanics of locomotion. Then there is the world of animals around
us. We see animals walking and crawling on land, flying in air, and swimming
in fluid. From man and mice to birds, fishes, and sperms, there is a tremendous
variety of questions one may wish to ask about locomotion.
The objective of the present chapter is to present a brief discussion of the
mechanics of flying and swimming. We select a few topics from the point of
view offluid and solid mechanics in order to gain some insight to the problems
of locomotion. We shall pass over walking, running, jumping, and crawling
because they are familiar to us, and because their mathematical analysis is
formidably complex.

4.2 The Conquest of the Air

About 400 million years ago, the earth was populated by plants and tall trees;
the air was moist, and the ground was covered with decaying leaves; and the
insects began to appear (Smart and Hughes, 1972). The insects achieved

106
4.2 The Conquest of the Air 107

powered flight in the middle of the Carboniferous period, about 300 million
years ago. Birds came on the scene later, about 150 million years ago. Reptiles,
which flourished 100 to 200 million years ago, included an extensive group of
flying animals, the pterosaurs. Certain flying mammals, including bats, ap-
peared about 50 million years ago.
All the animals we can see are winners of evolution. They represent success-
ful designs that have met all the constraints imposed by their environment. It
is not useful to cover too large a territory in this chapter. To be brief, let us
limit ourselves to two aspects: (1) identifying those features in nature that can
be understood in terms of the fluid mechanics discussed in the previous
chapter, and (2) recording observations of flying and swimming in nature that
involve principles beyond the scope of the preceding chapter. In the latter
category are items such as the hovering of hummingbirds, and the flying of
tiny insects.
The wings of birds and insects not only differ in size and structure, but also
in basic articulation mechanism. The bone structure of a bird does not differ
too much from that of our arms (see Fig. 4.2:1). The bird moves its wings by

FIGURE 4.2:1 Wing articulation mechanism of the bird. The elbow joint, and the
humerus, ulna, radius, and carpals bones are seen. The arrows x and y show tht}
moments created by the muscles about the bone. The aerodynamic force is transmitted
through the bases of the feathers to the wing bones. The resulting tendency to twist
the manus and ulna in the nose-down direction is resisted by the pitch~up moments
(x) supplied by the muscles. Owing to the sharp angle at which the elbow joint is held
in flight, the center of the lift lies ahead of the axis of the humerus and the resulting
nose-up moment is counteracted by the moment y supplied by the pectoralis muscle,
which pulls downward on the deltoid crest, ahead of the axis of the bone. From
Pennycuik (1972, p. 16), by permission.
108 4 Flying and Swimming

(8)
(b)
-~-
(c)
Direct flight muscles
Indirect flight muscles Indirect flight
muscles (side view)
FIGURE 4.2:2 Wing articulation mechanisms of insects. (a) In insects with direct flight
muscles, the muscles are connected to the wings. (b) In insects with indirect flight
muscles, the downstroke isproduced by the raising of the roof of the thorax which is
brought about by the contraction of the dorsal longitudinal muscles as shown in (c).
Upstroke is obtained by the contraction of the vertical muscles. In (b), a double hinge
system that connects the wings to the roof of the thorax is seen. The operation of the
hinges involves a skeletal click mechanism. From G. Goldspink (1977, p. 13) which is
modified from Pringle (1975). By permission.

muscle, so its flying muscles are big and strong. Insects use different arrange-
ments (see Fig. 4.2:2). The insect thorax is shielded by cuticles of thin-walled
chitinous shells with good elasticity and rigidity, joined with an elastic mate-
rial, resilin. The wings and the thorax shell form a vibration system. In some
insects the wing movements are produced by wing muscles directly inserted
into the base of the wing. In others, such as Diptera, the movements are
produced by muscles that pull on the thorax shell, while the shell deformation
moves the wings. In the latter cases [Fig. 4.2:2(b)] the hinges are so arranged
that the wing position is stable either fully raised or fully lowered, whereas in
moving from one stable position to the other it goes through an unstable
position very quickly. A twisting movement can be superposed on the up and
down movement by a proper arrangement of the hinges. Resilin is an almost
perfect rubber. It is capable of large deformation with a nearly linear force -
deflection relationship with very little internal friction. Its mechanical proper-
ties are similar to those of elastin which is discussed in Biomechanics (Fung,
1981, Chap. 7, pp. 197-201), except that the hysteresis loop is even smaller.
Elastin is found in the mammalian blood vessel wall. In the neck of the cow
or horse the large ligament nuchae is almost all elastin, and it holds the head
up without need of muscle action. In an analogous way the resilin can store
elastic energy and can release it very quickly when unloaded. The wing-thorax
resilin-and-muscle system is a mass-spring system which has a characteristic
frequency of vibration. Thus, some insect wings can beat automatically if
4.3 Comparing Birds and Insects with Aircraft 109

excited, and the muscle only needs to supply enough energy to overcome the
aerodynamic drag, and to provide its own excitation.
Resilin exists also in the leg joints of those insects that can jump, e.g., fleas
and click-beetles. To jump, these insects use the muscle to bend to joints and
store energy in the resilin by elastic deformation. Upon release of the muscle
tension the stored energy is released very quickly, causing the insect to be
catapulted.
The insect's flight muscles are rather different from mammalian skeletal
muscles in that they have a very short I-band in the sarcomere (Goldspink,
1977). As a consequence they operate in a very short range of length. This is
consistent with the mechanism exhibited in Fig. 4.2:2.
Be it a bird or an insect, a flying animal must be able to generate thrust
and lift, which is accomplished by superimposing an adequate positive angle
of attack onto a thrust-generating oscillatory motion. For birds and larger
insects in fast forward flight, the lift principle is similar to that of the airplane.
See Sec. 4.3. However, the majority of insects are small, and make use of
hovering and slow flight. To do this without a rotor, unlike a helicopter, most
hovering animals use a mode in which the body stays head-up (almost erect)
and the wings beat back and forth in a horizontal plane, preceding each beat
with a wing rotation that always allows the same leading edge to move forward
at an appropriate angle of attack. Some insects, such as the chalcid wasp
(Encarsia formosa), clap their wings together dorsally once per beat and
achieve a remarkably effective hovering (Weis-Fogh, 1975). The butterflies
(Papilionoidea) can clap their wings ventrally as well as dorsally. The dragonfly
can hover with an almost horizontal body axis by making a complicated
coordinated motion of its four wings. These are discussed in Sec. 4.5.
The story of evolution tells us that out of all known animal species, fossil
and living, those using hovering or slow flight form of locomotion represent
three quarters, or 750,000, of them (Weis-Fogh, 1975). The great success of
using active flapping flight as the principal mode of locomotion on earth is
evident.
But many birds can also utilize natural air currents to soar and to glide
over long distances. The albatross can even take advantage of the wind shear
over the ocean (i.e., the increase in wind speed with the altitude) to spiral and
soar with little expenditure of energy (Sec. 4.4). Nature is full of wonders.

4.3 Comparing Birds and Insects with Aircraft

To understand the mechanism offlight, it is useful to compare birds and insects

with airplanes and helicopters. Man copied nature and developed the aircraft
industry; but the constraints imposed on man's flight are not entirely the same
as those in nature, and the final products are different.
Birds and insects use flapping motion for propulsion, hence their aero-
dynamics is predominantly unsteady. To understand their aerodynamic lift,
110 4 Flying and Swimming

thrust, moment, and induced drag, we must think of the vortices shed into the
airstream at the trailing edge of the wing, as well as the oscillatory horseshoe
vortices which are bounded to the wing in the span wise direction and shed
into the wake in the streamwise direction. These are discussed in the preceding
chapter, see Secs. 3.9 and 3.10, especially Fig. 3.10:5. The most important
consequence of the vortex shedding mechanism is the delay in lift force
generation that causes the aerodynamic lift and moment to get out of phase
with the angle of attack. The system is, therefore, nonconservative, and there
is an exchange of energy between the wing and the airstream in every cycle of
flapping (see Sec. 3.6). This is very different from the fixed-wing aircraft for
which, unless the wing vibrates, only drag consumes energy at a steady flight.
The unsteady condition affects the stalling characteristics also (see Fig.
4.3:1). If a sudden change of angle of attack is so large that in normal steady
flight the wing would have stalled (a drop in lift coefficient occurs in the
CL max region shown in Sec. 3.13, Fig. 3.13:2), in unsteady motion the inertia of
the air may be able to prevent flow separation from occurring. Thus, a higher
CL max may be obtained. This is important for birds when hovering, landing,
and taking off.
A jetliner may extend its flaps to increase wing area when taking off and
landing, deflect the flaps to increase the curvature of the wing cross section to
increase the maximum lift coefficient (CL max), deflect the aileron to roll, move
the rudder to yaw, and lift the elevators to pitch. It may have a leading edge
slot to delay stall and increase CL max' as well as to increase the wing area. It
may have vortex generators (a row of protruding small plates arranged on the
upper surface of the wing in the tip region) to help prevent tip stall. Some high
performance planes are designed to increase their sweep angle (sweep back or
sweep forward) as the flight speed increases (to avoid shock waves). All these
features have prototypes in nature.
Figure 4.3:2 shows six major devices that enable birds to achieve high lift.

-a>O k=~
----a < 0 2V a = 15° + 10° sin wt
--a = 0 With end plates
3

/Z\ ;D
k = 0.004

~
CL

'\--- '\-.::--- -- ... \

p -~.,
-

/ / /
0
0 5 101520250 5 10 15 20 250 5
n' o· o·

FIGURE 4.3:1 Unsteady stalling characteristics of wings. The maximum lift coefficient
as a function. of the rate of change of the angle of attack. Airfoil NACA 0012,
1.22 m chord, 1.98 m span with end plates. Oscillation about quarter-chord point at
various reduced frequencies (Strouhal numbers k) in 29.5 m/sec airstream. From
McAlister et al. (1978). Courtesy of U.S. Army.
 ... ------.-.,. ......

~
~ -- --...;~ ....................

~-----------
...... _--

•
"
fowt erflap "

------.~
--_
Airflow
. " ~
' "~

-",,~

FIGURE 4.3:2 Arnold Kuethe's summary of bird's devices to achieve high lift. From
top down. (1) The "thumb pinion" (alula) of a pheasant as a leading edge slot. The
thumb pinion is particularly highly developed in woodland birds, presumably to better
enable them to avoid obstacles. (2) The drooping leading edge clump offeathers at an
owl as a Krueger flap. (3) The swept forward position of the tail of a split-tail falcon
as a Fowler flap. (4) The "layered" wing feathers as a multi-surface air-foil. (5) The
upward deflected flight feathers of the wing tips of a hawk as "winglets." Reproduced
by courtesy of Professor Arnold Kuethe. From Kuethe (1975a, b). Kuethe and Chow
(1986).
112 4 Flying and Swimming

These pictures were drawn by Cyril H. Barnes to illustrate a paper by Arnold

M. Kuethe (1975b). Some of them are redrawn from versions in von Holst and
Kiichemann (1974). In Fig. 4.3:2(a), the feathers are shown to twist in order
to direct a component of the lift in the forward direction to provide thrust.
The twisting is achieved automatically by a proper aero elastic structural
design of the feather by placing the elastic axis (a line of the centers of twist)
and the line ofthe centers of aerodynamic pressure at a proper relative position
(see Fung, 1969, p. 17).
The equivalent of airplane leading edge slot, split flap, Krueger flap, Fowler
flap, and blown flap are shown in Fig. 4.3:2(b)-(f). The blown flap of advanced
aircraft, using compressed air bypassed from the engine to blowout the
stagnant air in the thickened boundary layer near the trailing edge of the wing,
can raise the maximum lift coefficient by 400-500%. The birds, using the
devices named above, probably can increase the lift coefficient by 50-100%.
Figure 4.3:3 shows the method used by a hummingbird to hover (Weis-
Fogh, 1975). With its body nearly erect, a hummingbird swings its wings about
a near-horizontal plane; each wing describes a "figure 8" through an arc of
about 120°. During the interval when the wings are moving forward, the

Airflow

FIGURE 4.3:3 Hummingbird's hovering method as explained by Weis-Fogh (1973).

Reproduced by permission.
4.4 Forward Flight of Birds and Insects 113

"palms" are down; for the return stroke, they flip to palms-up. Both motions
produce lift in roughly equal proportions.
The Chalcid wasp hovers by a clap-and-fling motion which is described in
Sec. 4.5, Figs. 4.5:1 and 4.5:2. There is no man-made flying machine of this
nature.
There are other similarities and differences. Both animals and aircraft are
elastic, but the animals are more so. The aeroelastic phenomena, discussed in
Sec. 3.4, especially divergence, loss of control, and flutter, will impose restric-
tions on animal flight as they do aircraft. In addition, animals have nerves
and muscles, and exquisite sensory organs; and they achieve remarkable feed-
back and control which are not yet fully understood. Only the most advanced
aircraft designs seek active electronic control of their basic structures (in
addition to their control surfaces), whereas all birds do.

4.4 Forward Flight of Birds and Insects

Only by detailed observation can one gain a true understanding of animal

locomotion. For insects, a monumental piece of work was done by Weis-F ogh
and Jensen (1956), who analyzed the flight of the desert locust Schistocerca
gregaria by observing them in a wind tunnel. They suspended the locust from
a force balance, and used the force measured by the balance to control the
speed of the wind-tunnel fan in order to simulate free sustained forward flight
conditions. The wind speed, once the insect was beating its wings regularly,
was automatically adjusted by feedback control so that the net horizontal
force measured by the balance was zero. Thrust was then balancing drag as
in free flight.
Typical values of the locust's weight, wing length, the speed of flight, U,
and the wing-beat frequency, w/2n, are, respectively, 2 g, 4 cm, 4 m/sec, and
20 Hz. The Reynolds number U c/v is approximately 200 based on the forewing
chord c. The Strouhal number k = wc/U (with w in radians per sec) is about
0.25, low compared with k = 2.2 for most airplane wings in flutter condition,
or roughly 1 for fish with lunate tails, but about the same as that for birds
(Lighthill, 1975). Hence the effect of nonstationarity is not large.
Figure 4.4:1 shows the effective motion through the air of a forewing chord
in the forward flight of a locust. In the downstroke the forewings have a
positive angle of attack so that the lift on them is much larger than the drag
and the resulting aerodynamic force supports both the animal's weight and
the thrust needed for propulsion. This stroke is long and there is plenty of
time for the airfoil to build up its steady-state force, nearly at a right angle
to its instantaneous path through the air.
A rapid "supination" (in aeronautical terms, "pitch-up") marks the onset
of an upstroke of the locust wing with the airfoil at a large positive angle of
pitch relative to the direction of mean motion. At the same time, however, its
angle of attack relative to its instantaneous direction of motion in the air is
114 4 Flying and Swimming

~-------------------------------------------------f-
-y
18

FIGURE 4.4: 1 Tracings from a film of a tethered locust in a wind tunnel, showing the
movement relative to the air of a chordwise section through the mid-point of a locust
forewing in forward flight. From Weis-Fogh and Jensen (1956), by permission.

negative. At this stage of the stroke the wing is bent in the form of a Z-shaped
cross-section, which has some rather peculiar aerodynamic properties. Wind
tunnel tests show that in steady-state the lift is small and acts opposite to the
normally expected direction, probably due to flow separation at the leading
edge due to the bent head of the Z section. But the supination is rapid and
the vortices that shed into the wake of the wing are important to the transient
aerodynamic response. The actual non stationary aerodynamic force acting
during the upstroke is still unknown.
The motion ofthe hind wings is coordinated with that ofthe forewings. The
forewing motion exhibits a phase lag behind that of the hindwings. This phase
lag is characteristic of four-winged insects in general. Detailed estimations of
the lift, drag, thrust, and the interference between the wings are given by
Weis-Fogh and Jensen (1956).
For birds, corresponding pieces of beautiful work were done by Brown,
Tucker, Pennycuick and others. Brown (1953) trained birds to fly along a 60
m long passage to a cage as he photographed them with a high-speed camera
halfway along the passage. Tucker (1968) and Pennycuick (1968) trained birds
to fly freely in wind tunnels. From these studies it was found, for example, that
pigeons can fly at a speed of about 10 m/sec, wing beat frequency of about 5
Hz, corresponding to a Strouhal number of around 0.5, and Reynolds number
of about 104 based on the wing chord. The downstroke, with the wing chord
practically horizontal, is quite similar to the downstroke of the locust shown
in Fig. 4.4: 1. The rapid supination at the end of the downstroke is also just as
marked. The upstroke is highly 'feathered': the bird's wrist is flexed in the
upward movement, and then the primary feathers near the wing tips are swung
violently backward to obtain a significant thrust in a manner depicted in Fig.
4.4:2.
The bird's lift is probably acquired mainly on the downstroke when the
primary feathers are spread out, and the wing area is increased. In the
meantime, the maximum lift coefficient is increased as we have discussed in
4.4 Forward Flight of Birds and Insects 115

Lift

Airflow
Drag -.J
----------
~ ~own (forward) stroke
.............
.............
~ .............

--
............. .............
.............
.............
Lift

LThrust

""~ -- ----
- - __ Up (rearward) stroke

-- ----2:"" .............

............. ---- Airflow

FIGURE 4.4:2 Hovering flight of a gull. The locations of the quills of the feathers are
indicated. The force vectors are shown in the right. Courtesy of Dr. Arnold Kuethe.

Sec. 4.3, and stalling is avoided even when a large angle of attack is used at
the end of the downstroke to initiate the supination and upstroke.
Compared with insects, birds have a higher power requirement per unit
mass. Pennycuick (1968) estimated the specific power requirement for sus-
tained pigeon flight to be around 20 W per kg, amounting to over 96% of the
total metabolic rate. An oxygen consumption of over 130 mljmin is needed
to sustain flight, in contrast to a resting metabolic requirement of only 5
mljmin. On the other hand, the sustained locust flight was estimated by
Weis-Fogh and Jensen (1956) to require only about one-third as much muscu-
lar power output per unit mass of the animal.
Thus the bird needs a lung that can handle a large variation in ventilation
to meet the enormous variation in its energy requirement in life. The bird's
lung is encased in a rigid chest cage, and its ventilation is powered by a number
of air sacs (Dunker, 1972). The lung maintains a constant volume, while the
air sacs expand and contract to generate a flow. Figure 4.4:3 shows a schematic
drawing of the avian respiratory system. The gas exchange apparatus consists
of parallel bundles of thin-walled "air capillaries" of diameter 3 to 10 ~m
(Dunker, 1972) in close proximity to blood capillaries of about the same size.
The construction is similar to some boilers in industrial power plants. Since
the air and blood capillaries are about the same size, the birds are able to pack
the blood- gas exchange surfaces into their lungs more compactly than mam-
mals can in their reciprocating machinery. The exchange surface area per unit
116 4 Flying and Swimming

~ntrobronchi
========-t- Parabronchi
.. - ..........
Cranial
ArSoc Oorsobronchi

.'........
(cut)
....... -

~.--------~-===~)~
\ --------~
M.sobronchus

FIGURE 4.4:3 Schematic drawing of the avian respiratory system by Scheid, Slama, and
Piiper (1972).

volume is about 250 mm- 1 in the bird, as compared with 10-16 mm- 1 for
man. But the most remarkable fact about the bird's lung is that the flow in
the air capillaries is unidirectional while the flow in the air sacs is reciprocat-
ing. This unidirectional flow is achieved without a valve anywhere! It is a
valveless pump. One can understand such a pump only if the inertia of the
fluid and the resistances to flow in the various bronchi and bronchial bifurca-
tion points are considered.
Example. Take a surgeon's glove. Tie up the wrist and cut holes at the tips
of the thumb and ring finger. Into the holes insert glass tubes of 3 to 5 mm
diameter and tie up. Hold the tubes vertical and fill the glove with water to a
level halfway up the tubes. Now you have a valveless pump. Tap on the palm.
With a little practice you can find the right frequency, amplitude and location
of tapping that will create a substantial difference of the heights of the water
columns in the tubes. Explain the mechanics!
4.4 Forward Flight of Birds and Insects 117

Power Requirement for Flight

The thrust required for flight must balance the drag and the inertial force. The
inertial force is equal to the product of the acceleration and the sum of the
mass of the animal and virtual mass of the fluid associated with the mode of
motion. The drag force consists of three parts: One part is due to skin friction
in the boundary layer over the wing, body, and tail. Another part is the induced
drag discussed in Sees. 3.2 and 3.10. The third part is due to the flapping
motion and the transient vortex shedding into the wake.
The sum of the skin friction and induced drag is the drag at steady flight:

(1)

Here D is the drag force, p is air density, V is the speed of flight, A is the wing
area, CD, is the coefficient of drag due to skin friction on the wings, body, and
tail, K is a constant, L is the lift, b is the wing span. The first term on the right-
,
hand side of(1) is Eq. (3.2:4), except that we now let CD include the frictional
drag of the body and tail. The second term in (1) is Eq. (3.10:4).
In steady flight the lift must be equal to the weight of the animal, W:

L = W = 1PV 2 ACL . (2)

Hence the lift coefficients CL , and the corresponding angle of attack (see Fig.
3.2:2) depend on the weight and the flight speed. The drag coefficient CD, varies
with the angle of attack, and hence with CL • When CL is small, CD! is roughly
constant, but CD! increases rapidly if the angle of attack is increased toward
the stalling angle or beyond.
The power required for flight is equal to the product of the thrust T and
the speed of flight V. If we consider steady flight, and very roughly assume
the power required for the flapping motion of the wings to be a constant, Po,
then
Power required = TV = const + DV
= Po + 1PV 3 A CD! + KW 2 /(1PVb 2 ). (3)
If the power required is plotted against the forward speed V we obtain the
results shown in Fig. 4.4:4 Here the power required is seen to be a V-shaped
curve. The minimum power point Pmin corresponds with a velocity Vmim p.
The point of tangency of the power curve with a straight line drawn from the
origin defines the velocity of flight and the power required for the maximum
range. This is because to maximize the range we must maximize distance
travelled per unit work, and this is equivalent to finding the smallest ratio of
power to speed. The maximum-range speed is at least 1.3 times the minimum-
power speed, and usually more-in the pigeon it is around 1.8 times the
minimum-power speed.
The power required to hover (at V = 0) is usually much larger than Pmin •
Eq. (3) does not apply to the hovering mode.
118 4 Flying and Swimming

"t:l
....
Q)

P mr
:::J
0-
.... P min
Q)

Qi
~
0
Q..
Po

U max range
Speed of flight

FIGURE 4.4:4 A plot ofEq. (3). Po represents the constant in Eq. (3). Curve (a) represents
the second term on the right-hand side. Curve (b) represents the last term. Curve TU
is the sum of these two, or the power required. Eq. (3) does not apply when U -+ 0;
that part of the curve is estimated separately.

Range
The work done by flying over a distance dX in still air is equal to the product
of thrust times dX. The thrust is equal to the power P divided by the velocity
V, see Eq. (3). Hence the work done dE is PdX/V. If we regard dE as the
amount of fuel used in order to fly the distance dX, then we can write

dX = VdE. (4)
P
We would like to express this in terms of the lift/drag ratio. For this purpose
note that P = TV, and that at level flight, T is equal to the drag D, and the
lift L is equal to the weight W. Hence
dX _ V dE _ L dE _ L dE
- TV -WV-Jj W· (5)

The lift/drag ratio, L/D, depends on the angle of attack and the wing and body
geometry. The energy dE is obtained by burning a small amount of fat dW.
Writing dE = JdW where J is the mechanical equivalent per unit weight of
fuel (about 8 x 105 joules per kg weight), we obtain, on assuming L/D to be
constant and integrating Eq. (5), the result

L
X=J-In
D
-
Wz '
(WI) (6)
4.4 Forward Flight of Birds and Insects 119

where Wi' W2 are the bird's take off and landing weights, respectively. Thus,
to fly a long distance, as some migrating birds do, the lift/drag ratio should
be large and the bird should be fattened as much as possible before taking off.
A ratio of Wt/W2 equal to 2 is not inconceivable.
Figure 4.4:4 tells us how to find the speed to fly for maximum range. In
this figure the horizontal axis is the bird's airspeed. If there is a tail wind
blowing, the ground speed is equal to the wind speed plus the bird's airspeed.
If the bird's aim is to maximize distance travelled over the ground per unit
work done, then the graph must be redrawn with groundspeed instead of
airspeed as the abscissa. The effect of a tailwind is to shift the origin to the
left, and the tangent now has to be drawn from the new origin, giving a lower
air speed than before for maximum range. Similarly, airspeed must be in-
creased to obtain maximum range against a headwind. Radar tracking of
migrating birds flying high at night shows that they do adjust their speed of
flight this way, although it is not clear how they find out the wind speed!

Stability and Control

Whereas an airplane needs ailerons to control roll, elevators to control
pitch, rudder to control yaw, and fin and stabilizers to give stability, most
insects and birds can achieve stability and control by their wings alone. The
mechanisms used include: (1) variable sweepback of the wings and variable
angle of attack across the span, (2) controlled variable camber, i.e., curvature
of the wing in chordwise direction, and (3) variable wing tip control with
feathers deflected downward. Some of the prehistoric birds (Archaeopteryx)
had long tails like an airplane, but later members dispensed with the tail,
suggesting that the tailless method of control is more efficient aerodynamically,
but more difficult to operate.
Pennycuick (1972) suggests that the function of the bird's tail is like that
of the flap of an airplane: to increase the maximum lift coefficient at low speed.
The tail is typically spread and depressed at landing and take off. This adds
lift surface area and sucks air downwards over the central portion of the wing,
so stalling can be delayed. Birds with long forked tails, such as swifts (Apodidae)
and swallows (Hirundinidae), are able to fly very slowly and hover, and use
the spread tail for steep downward deflection of the airflow leaving the wings.

Gliding and Soaring

If a bird or insect of mass m flies without beating its wings, it can get thrust
only from gravitational acceleration, i.e., by gliding downward. In a still air
let the flight path be inclined at an angle eto the horizon, then the lift L, drag
e
D, and gliding angle are related by the following equations:
L = mgcose, D = mgsine, (7)
Since L is essentially the bird's weight, the last equation shows that the gliding
angle ewill be a minimum if the drag is minimized. With drag given by Eq.(1),
120 4 Flying and Swimming

this occurs at a flight speed U which makes the two terms on the right-hand
side of Eq. (1) equal:
4K L2 )1/4
Umin drag = ( ~ Ab 2 •
(8)
P Df

If U is smaller than that given in Eq. (8), an accidental rise in speed will reduce
the drag and the speed will be increased. Hence gliding at a speed less than
the minimum-drag speed is unstable. Gliding is stable only if U is greater than
the minimum-drag speed, and, of course, also above the stalling speed.
If the bird catches a thermal, then while Eq. (7) remains true relative to the
wind, the path relative to the earth may become inclined upward, and the bird
soars. As this is well known, let us not elaborate any further but turn to the
albatross, which "soars" according to a different principle.

"Soaring" of the Albatross

The famous "soaring" of the albatross Diomedea, by which it varies its height
above the ocean in a periodic cycle with minimum expenditure of energy was
analyzed by Rayleigh (1883) and Jameson (1958). It utilizes the wind shear
over the sea, i.e., the variation of wind speed with height. Stanley Corrsin
explained it as follows (see Lighthill, 1975, p. 169): Let the coordinates x, y, z
be fixed on earth and let the mean natural wind velocity be in the x-direction,
with magnitude u(z) varying with altitude z above the ocean. Let the three
components of the velocity vector of the albatross relative to the earth be
(u(z) + u', v', w'). (9)
Then its velocity relative to the local wind is (u', v', w') with a resultant V. The
dynamic pressure is !pV 2 , p being the air density. At steady flight the lift
must balance the bird's weight W, and the life coefficient is CL = W/(!p V 2 S),
where S is the wing area (Eq. 3.2:4). The angle of attack is IX = Cda if a denotes
the lift curve slope (Eq. 3.2:5). To avoid stalling IX must be smaller than the
stalling angle IXmax , and CL must be smaller than the maximum possible, CL max
(see Fig. 4.3:1). Since CL varies inversely with V 2 , to avoid stalling the square
of the relative velocity must be so large that the lift coefficient remains below
the maximum value CLmax ' In other words the kinetic energy:
(10)
must be steadily maintained above a minimum value against the dissipative
action of aerodynamic resistance. Now, in the theory of turbulence we learn
that any fluctuating motion with velocity (9) (whether of an eddy or of an
albatross) extracts energy from the mean motion at a mean rate
-du
-mu'w'- (11)
dz'
where the bar denotes a mean value. But u'w' can be large and negative,
4.5 Hovering and Other Modes of Motion 121

making (11) large and positive and thus maintaining the kinetic energy (10)
above a minimum level. To do this u' and w' should have opposite signs: the
albatross should move upward when upwind and downward when downwind.
Over the ocean the atmospheric boundary layer thickness can be of the
order of 50 m. The wind shear is significant near the ocean surface. The
albatross turns into the wind in order to gain height, and then turns downwind
to gain speed. At sea level it soars along the slope of the windward face of a
wave until it meets a suitable upward gust, when it turns into the wind and
initiates another upwind climb. It zigzags in this way over the Antarctic
Ocean, progressing on average downwind, and circulating in the prevailing
westerlies 'round and 'round Antarctica.

An Impressive Record
The speed of most insect flight is not impressive, but the deer botfly has been
clocked at 64 miles per hour (28 m/sec) or 300 body lengths per second
(Nachtigall, 1974). By comparison, the Cheetah (fastest of the land animals)
achieves 18 lengths per second, while man achieves 5, an automobile 5 to 15,
a swift 60, a Starling 80, ajet fighter at Mach 3 reaches about 100.

4.5 Hovering and Other Modes of Motion

Most birds and insects hover like a hummingbird, with the body vertical and
the wings flapping in a horizontal plane, see Fig. 4.3:3. A dragon fly hovers
with body horizontal and two pairs of wings executing a coordinated motion.
Weis-Fogh (1975) and Norberg (1975) analyzed this motion with quasi-steady
strip theory (treating every cross section of the wing as if it were a two-
dimensional airfoil reacting to a steady flow whose velocity is the same as that
of the instantaneous field) and concluded that hovering is energy intensive.
Indeed, large birds hover only a short time at take off or landing, and usually
have to use anaerobic power to do so.
Weis-Fogh (1973) described another type of hovering. He studied a small
wasp (Hymenoptera) Encarsiajormosa, that is used in the biological control
of greenhouse aphids. It has a wing span of 0.6 mm, an airborne weight of
25 x 10- 8 N. Its fore- and hind-wings are hooked together and beat as one
at a frequency of about 400 Hz. Like many insects Encarsia can jump but the
wing movements seen in Fig. 4.5:1 relate to free hovering with a slow climb.
Its wings move essentially in a horizontal plane both during the downstroke
and upstroke. The sequence of its wing movements contains three phases: (a)
the clap phase in which the two pairs of wings are brought together as a single
vertical plate (b) a fling phase in which a very rapid pronation of the wings
takes place and the wings are flung open in a manner reminiscent of the
"flinging open" of a book, followed by a horizontal downstroke; (c) a flip phase
during which a very rapid supination takes place before the upstroke. These
122 4 Flying and Swimming

Clap
(a)

Fling (open)
(b)

Fling (forward)
(c)

FIGURE 4.5:1 The hovering motion of Encarsia formosa as sketched by Weis-Fogh

(1973, Fig. 21). Reproduced by permission.

1 - - - - - - 1mm - - - - 0 4

FIGURE 4.5:2 The bristles (setae) of the wings of Terebrantia and haplothrips, tiny
insects commonly known as "thrips." Courtesy of Dr. Kuethe.

three phases are present in Encarsia during all kinds of flight. Moreover, the
lift equalled the body weight long before the wings reached maximum angular
velocity.
Weis-Fogh's explanation (1973) is supported by Lighthill's (1975) analysis.
Lighthill explains that toward the end of the clap, the wings are essentially at
4.5 Hovering and Other Modes of Motion 123

FIGURE 4.5:3 Scanning electron microscopic photographs of a Mulleinthrips Haploth-

rips verbasci. The setae (shown here in a stiffened condition) are 1 x 2 J.1m in cross
section, and 200-300 J.1m long. x 130. Courtesy of Dr. Arnold Kuethe.

rest relative to the air. During the fling the wings open up, and a potential
flow is induced so as to fill the triangular void created between the two upper
wing surfaces, giving rise to two bound vortices of equal strength and opposite
sign. When the wings break apart along the "hinge" and start the horizontal
downstroke, they already have circulation and can produce lift in accordance
with the Kutta-Joukowski Theorem, Eq. (3.7:23). Since the bound vortices
are already there, the generation of this part of the lift has no delay and no
Wagner effect (Sec. 3.5 and Sec. 3.14). In other words, the initial lift is generated
in the fling stage.
124 4 Flying and Swimming

The Reynolds number of this wasp's hovering motion based on the wing
chord and wing velocity in downstroke is about 20. The Reynolds number
based on the speed of the leading edge during its "fling" is about 30. These
Reynolds numbers are sufficiently large so that the boundary layer theory and
vortex wake concept discussed in Chapter 3 are presumably applicable.
For even smaller insects that operate at Reynolds number close to 1 or
smaller than 1, boundary layer concept does not apply, and any analysis
would have to be based on Navier-Stokes equations. In these cases definitive
theory and experiments are both lacking.
Kuethe (1975) points out that many tiny insects have one striking feature
in common. From the tips of their wings, a fringe of bristles (setae) projects
forward, rearward, and outward. Figure 4.5:2 shows a Terebrantia, commonly
known as "thrips", and a haplothrips. Figure 4.5:3 shows a magnified view of
these bristles. Kuethe notes that the flow of air around each bristle has
Reynolds number much smaller than 1, hence must be governed by Stokes
equations. The analysis of flow around such moving bristles is similar to those
around cilia and flagella (Secs. 4.8 and 4.9). It seems natural to suggest that
by a proper wave motion of these bristles, lift can be generated in analogy
with cilia propulsion. Kuethe's idea is that the wave motion of the sheet of
bristles resembles that of the pectoral fin of a skate or ray. He estimated that
the lift force generated by wave motion of the sheet of bristles is more than
100 times greater than that would be if the sheet were impervious to the flow.

4.6 Aquatic Animal Propulsion

From single cells to whales, most organisms live in water. The most elementary
form of motion is chemotaxis. The protoplasm in cells is not stationary, it
moves in response to chemical stimuli. Some cells, such as leucocytes and
amoeba, move this way. Chemotaxis is one of the intensely studied subjects
today; but the mechanism is more closely related to the molecular transforma-
tion of actin molecules and the sol-gel transformation of the protoplasm than
hydrodynamics. See Sec. 4.12.
To survey the field of aquatic animal propulsion, we may begin with
Protozoa. Among these single-celled creatures are those belonging to the class
Mastigophora, which propel themselves by undulatory motions of a whiplike
flagellum. The basic mode of motion is to pass a wave backward along the
flagellum, either by a flexural motion, or by rotation in the manner of a screw.
See Fig. 4.6:1. Fluid resistance to the motion provides the propulsive force to
the organism. This method of propulsion has been most successful in the
evolution process, and has been adopted and developed by practically all the
aquatic animals that are successfully mobile.
Another class of Protozoa, the Ciliophora, includes organisms which propel
themselves by movements of a large number of attached cilia. Each cilium
behaves like a flagellum; but with many cilia there are certain organized
4.6 Aquatic Animal Propulsion 125

5WIM'1IN; ATTACHED

b
FIGURE 4.6:1 Flagella movement computed by Brokaw and Gibbons (1975) under the
hypothesis that active moment is proportional to curvature, and viscous and elastic
resistances are considered.

patterns. Ciliary propulsion is used by animals in the class Ctenophora, and

most members of the phylum Rotifera. But this method has been less successful
in the evolutionary process.
In man and animals, cilia also grow on the surfaces of respiratory airways,
mesenteric membranes, uterian tubes, etc. They help move dust particles from
the lung, mucus in the airway, ovum from the ovary, etc. Their function
depends very much on the viscoelasticity of the fluid. Figure 4.6:2 shows the
ciliary motion of the respiratory tract. Successive stages of the forward stroke
are shown in (a). Successive stages of the retraction stroke are shown in (b).
See Sec. 4.9.

FIGURE 4.6:2 Successive stages of the movement of a cilium on the surface of a

respiratory airway. (a) Forward stroke. (b) Return stroke. The length of arrows indi-
cates the relative speeds of ciliary motion. From Blake and Sleigh (1975), by permission.
126 4 Flying and Swimming

An alternative to undulatory propulsion isjet propulsion, which is used by

animals in the classes Hydrozoa and Scyphozoa in the phylum Coelenterata.
These animals have a characteristic jellyfish shape, and by contraction of the
circular subumbral muscles can expel water from the subumbral cavity and
thereby slowly propel themselves. In the phylum M oliusca, some animals in
the class Cephalopoda, the squids, use a jet concentrated in a narrow funnel
and can achieve quite high speeds of 2-4 m/sec, according to size. In the
phylum Chordata, members of the class Urochordata (headless and accord-
ingly invertebrate) characteristically draw in water and expel it for propulsion.
But jet propulsion is rare among the vertebrates.
Thus undulation is the method of propulsion preferred by animals higher
in the evolutionary ladder. It is remarkable that the same method can be used
by both large and small animals. The forward velocity of the animal U is
related to the speed ofthe transverse wave V passing down the animal's body.
For a flagellate at a typical Reynolds number of 10- 3 , U is of the order of 0.2 v.
(Gray and Hancock, 1955). For a nematode with a typical Reynolds number
of about 1 the forward speed U is of the order of 0.4 V (Gray and Lissmann,
1964). For a leech with a Reynolds number of order 10 3 the forward speed is
about 0.3 V(Gray, 1939). For an eel U is typically of the order of 0.6 V. For
goldfish U is about 0.7 V (Bainbridge, 1963).
The most important modification of the undulatory method of propulsion
introduced by fish is the transverse 'compression' of the body, i.e., flattening
at the posterior end of the animal. The flattening makes the rear end more
flexible transversely, and permits motions of larger amplitude at the tail
region. The next important modification is the enlargement of the caudal fin
to increase the reactive force of propulsion. It can be shown (Lighthill, 1975;
see Sec. 4.10) that the motion at the tail region is really what counts in the
propulsion of the fish, and the width of the caudal fin at the posterior end is
the most important dimension of the fish as far as propulsion is concerned.
Hence the evolution points in the direction of increasing the caudal fin
dimension.
The most advanced development is the lunate tail on such fast swimmers
as the shark, tuna, dolphin, and whale. The caudal fin is transformed into
"wings" of advanced hydrodynamic design. See Figure 4.6:3.
Other fins of the fish, dorsal and anal, pectoral and pelvic, are important
in stability and control, turning and stopping, as well as for auxiliary
propulsion.
Breder (1926) classifies fish swimming into "anguilliform" and "carangi-
form" modes. The former is named after the common eel, of genus Anguilla.
The latter is named after jacks and horse mackerels, of family Carangidae. In
the former the whole body participates in an undulatory mode. In the latter
the tail waves vigorously while the front body remains almost rigid.
This broad classification is widely used but is only indicative. There are
many variations and fine details about each fish that make it unique. And
often the situation is not so clear cut. For example, eels use strict anguilliform
 ~
~
~
~
~
~
.- .. ~.-.-,.
f/
,., _ . , ., ' Wr

FIGURE 4.6:3 Evolution of swimmers. The first row shows some early dwellers of the
sea: protozoa, trilobite, sea scorpion. Rows 2-10 show the evolution offish, beginning
with Devonian period (405 - 345 x 106 yrs ago) at left, to the present time at right.
Row 2 shows Agnatha (Jawless fishes), Pteraspis and Lampetra (lamprey). Row 3
shows a Placodermi (Armored fish) Pterichthyodes. Row 4 shows Acanthodii (Spiny
fishes) Climatius and Acanthodes. Rows 5 and 6 show Osteichthyes (Higher bony
fishes) Cheirolepis, Scomber (mackerel), Oncorhynchus (salmon), Dorypterus, and
Acipenser (sturgeon). Row 7 shows an Osteolepis. Row 8 shows a Dipnoi (Lungfish)
Dipterus. Rows 9 and 10 show Chondrichthyes (Cartilaginous fishes) Cladoselache,
Dogfish shark, Manta (devil ray), Helodus, and Chimaera (ratfish). The last row shows
a mammal, a killer whale.
128 4 Flying and Swimming

propulsion only in fast swimming; for slow motion they use dorsal and anal
fins. They can swim backwards by putting the fin undulations to reverse. Cod,
haddock, and whiting, having numerous separate dorsal fins and a substantial
anal fin and a caudal fin, use these fins in something close to anguilliform. The
vortex shed by the fins is essentially continuous in spite of the gaps between
the fins, which function almost like a continuous one.
We present a mathematical framework for the analysis of flagella and cilia
propulsion in Secs. 4.7-4.9. The theory discussed in Sec. 4.9 could probably
be extended to the anguilliform swimming of some animals with 'poor' hydro-
dynamic form such as water snakes.
Lighthill's theory for carangiform propulsion is presented in Sec. 4.10. This
theory reveals the predominant importance of the tail. In realizing that each
stroke of the tail motion produces a large sidewise force, one must wonder
how the fish can stay on an even course without hopelessly yawing from side
to side. Lighthill (1969) suggests two answers. One is that by narrowing the
depth of the fish cross-section before the caudal fin, the side force is reduced.
The other is that by having a large depth of body and fins in the front part of
the fish, the reaction to side force is reduced.
Bainbridge (1958, 1960,1963) has observed the details of carangiform pro-
pulsion of several species of fish and found that the maximum amplitude of
the caudal fin movement (i.e., the lateral displacement from the unstretched
position) to be about 0.2 times the length of the fish, L. The reduced frequency
or Strouhal number OJLjU (where the radian frequency OJ is 2n times the
frequency in Hz) takes values clustering around 10. This is in sharp contrast
to the reduced frequencies of insects' and birds' flight, which is less than 1, and
usually of the order of 0.2 to 0.6. The influence of vortices in the transient
wake (Sec. 3.10) is thus much more significant for the fish.

4.7 Stokeslet and Dipole in a Viscous Fluid

Microbes swimming with flagella (such as human sperm) or cilia (such as

paramecium) are so small that the Reynolds number of their motion is much
less than 1. In that case the inertial force is negligible compared with the
viscous forces; the Navier-Stokes equation is reduced to the Stokes equation,
and the flow is said to be in the Stokes flow regime. In constructing a
hydrodynamic theory for the flagellar motion, use is made of singularities such
as the Stokeslets and dipoles. Let us first explain what they are.

The Stokeslet
Consider an incompressible Newtonian fluid. Let p denote the pressure, U
denote the velocity vector, p denote the fluid density, J1 denote the fluid
viscosity coefficient. Then, on neglecting the acceleration terms (because the
Reynolds number ---+ 0), and assuming a body force field X, the Navier-Stokes
4.7 Stokeslet and Dipole in a Viscous Fluid 129

equation (see Sec. 1.7) becomes the Stokes equation

-Vp + J.LV 2U + X = O. (1)

The equation of continuity is

V'U = O. (2)
Applying the V operator on Eq. (1) and noting (2), we obtain
_V2p + V'X = O. (3)
Applying the Laplace operator V2 on Eq. (1) and noting (3), we obtain
J.LV4 U = V X V x X. (4)
If the body force vanishes, then X = 0, and
V4 u = O. (5)
Hence in a region in which body force vanishes the pressure is a harmonic
function (because p satisfies the harmonic equation V 2 p = 0), and the velocity
components are biharmonic functions (because they satisfy the biharmonic
equation (5)) in an inertialess flow.
Oberbeck (1876) obtained a general solution of Stokes' equation in terms
of harmonic functions:
u = 2cj) - V(x' cj») + V<Dc (6)
p = -2J.LV·(f). (7)
Here cj) is a vector with three components which are harmonic functions,
V2cj) = 0, and <Dc is a scalar harmonic function, V2<Dc = O. By direct substitu-
tion, it is easy to verify that x' cj) - <Dc is a biharmonic function. Hence u given
by Eq. (6) is biharmonic, satisfying Eq. (5). The solution is general in the sense
that any biharmonic function in a region whose boundary intersects any
straight line in at most two points can be represented by a function of the form
given in Eq. (6) (for a proof, see Fung (1965), p. 208).
On substituting cj) = rxlr and <Dc = 0 into Eq. (6), one obtains a solution
which is called the Stokes let by Hancock (1953). The flow field of a Stokeslet is
rx (rx'x)x
u=-+--- (8a)
r r3
_ 2 (rx·x)
p- J.L-3- · (8b)
r
If rx is a unit vector in the direction in the x-axis, then rx = (0(,0,0), X = (x, y, z),
r2 = x 2 + y2 + Z2; we have
_ (X2 + r2 xy xz)
U-O( 3 '3'3' (9a)
r r r

(9b)
130 4 Flying and Swimming

The solution is singular at the origin. To obtain a physical meaning of 0(,

consider a flow generated by a concentrated force applied at the origin in the
x-direction. Represent this force by the Dirac function Fx<>(r), where r is the
radius vector and Fx can be a function of time. Equations (1) and (3) then
become
- Vp + j1"\Pu + Fx<>(r) = 0 (10)
"\pp = V· Fxc5(r) = Fx{i3c5(r)jox). (11)
At this point it is convenient to recall a familiar solution representing the
source flow of an incompressible fluid (Sec. 3.7), which is described by the
equation
(12)
for the velocity potential t/J. The source is located at the origin and its strength
is a unit volume per second. The solution is
1
t/J=--. (13)
4nr
Comparing (13), (12) with (9b) and (11), we see that the solution

p= -Fx
-- 0 -
4n ox r
(1) (14)

represents the pressure field generated by a concentrated force Fx at the origin.

Comparing (14) with (9b), we see that
(15)
Thus 0( is equal to the force at the origin divided by 8nj1.
The total force acting on a control volume V enclosed in a surface S is given
by integrating the surface traction over the entire surface. If (Jij denote the
stress tensor and Vj denote the unit normal vector of the surface, then the force
acting on S is, by Cauchy's formula (Sec. 1.7),

Pi = - L (-pc5ij + (J;)vjdS. (16)

By Gauss theorem, this can be transformed into a volume integral

F; = _ r (_~
Jv ox;
+ O(Jij)dV.
oX j
(17)

By Euler's equation (Sec. 1.7), the integrand is the body force. Hence

Fl = Iv Fxc5(xddV
= Iv 8nj10(c5(x )dV = 8nj10( = Fx,
1 (18)

F2 = F3 = o.
4.7 Stokeslet and Dipole in a Viscous Fluid 131

The divergence of (Jii' namely, O(JijjOXi, is the term Jl"\PU in Eq. (5). The force
so computed by Eq. (18) is independent of the size and shape of the control
volume V. Thus the force on any control surface is equal to the concentrated
force acting at the origin.
At each point, the radial component of the velocity field of a Stokeslet is,
according to Eq. (9a):
Ur = (r· u)jr = [F j(8nJl)] (2 cos Ojr), (19)
where 0 is the spherical polar coordinate defined so that cos 0 = xjr. The
transverse component is
(20)
Figure 4.7:1 shows the velocity components. Note that the velocity decreases
rather slowly as r- 1 as r increases, and that only the factor 2 in Eq. (19)
prevents the Stokeslet from representing a unidirectional velocity field in the
x-direction with a magnitude equal at all points on a spherical surface of
radius r.

..- --- .......

2 cos eF
- - - 8np

F
I' " 8np
I
I F
I
(a) Stokeslet field

(b) Doublet field

FIGURE 4.7:1 The velocity components in spherical polar coordinates for (a) a stokeslet
field of strength F, and (b) a dipole field of strength G.
132 4 Flying and Swimming

Dipoles
The fow field V¢J derived from a velocity potential
¢J = Gxj(4nr 3) (21)
is said to represent a dipole with strength (G, 0, 0). A dipole is a source and a
sink next to each other, and its potential is obtained by a differentiation of
the potential of a source (Eq. (13). The gradient of the velocity potential is the
velocity vector. The velocity and pressure fields for a dipole are:
_4nG(1r3
u - - - -3x
- -3xy
- -3XZ)
r5
-
2

' r5 ' r5 '

p = const. (22)

The last equation follows from Eq. (1) since u = V¢J satisfied V2 u = 0 when
X = O. The velocity field has radial and transverse components
G 2cos e G sine
= ----, (23)
Ur = - 4n -r-
3 ' Uo
4n r3
which decays like r- 3 • This is illustrated in Fig. 4.7:1.

4.8 Motion of Sphere, Cylinder, and Flagella in Viscous Fluid

From Fig. 4.7:1 or Eqs. (19), (20), and (23) of Sec. 4.7, it can be seen that if one
wants to represent the motion of a sphere of radius a, one should superpose
a dipole of strength
(1)
to a Stokeslet of strength F. Then the velocity on a sphere r = a is a uniform
velocity U = Fj(6nJia) parallel to the direction of the applied force. This yields
the Stokes formula
(2)
relating the external force required to move a sphere of radius a through a
fluid of viscosity Ji at a speed U when the Reynolds number approaches zero.
For an alternative method of deriving this result, see Fung (1984), pp. 250-254.

Slender Cylinder
Since a Stokes flow around a sphere can be described by a superposition of a
stokeslet and a dipole, it may be expected that a Stokes flow around a cylinder
can be obtained by superposing an infinite number of stokeslets and dipoles
on a straight line segment. Indeed, let a line segment stretch from z = - b to
z = c on the z-axis, Fig. 4.8:1. Let the strength ofthe stokeslet on a line element
of length dz located at z be fdz, whereas that of the dipole be gdz, f and 9
being constants. Then on moving the origin of the solutions given by Eqs. (9),
4.8 Motion of Sphere, Cylinder, and Flagella in Viscous Fluid 133

z=c
z

y y

--4~-------~ X
-b c

(a) (b)

FIGURE 4.8:1 Motion of slender cylinders: (a) in a direction perpendicular to its axis,
(b) in the direction of the axis.

(15), and (22) of Sec. 4.7 to Z (replacing z by z - Z), leaving the stokeslet and
dipole pointing in the x-direction, changing F to fdZ, G to gdZ, and integrat-
ing the result from Z = - b to Z = c, one obtains a solution which represents
the results of the superposition. Lighthill (1975) shows that if one choses
fa 2
9 = - (3)
4/1
(not, in fact, the same choice given by Eq. (1) that was needed for the sphere
problem), then the velocity UN on the circle
x2 + y2 = a2, z=o (4)
is uniform, with the components

UN = L
81t/1
(1 + log 4c:a ,0,0) (5)

provided that a « b, a « c. This shows that the result represents approximate

solution of the velocity field generated by a uniform translation of a cylinder
of radius a in a direction normal to its axis. The force applied by the cylinder
on the fluid is equal to f per unit length of the cylinder, and it acts in the
direction of motion perpendicular to the axis ofthe cylinder. The force is equal
to the strength ofthe Stokeslets because the dipoles contribute nothing to the
forces.
This solution is approximate because it does not hold in the neighborhood
of the ends of the cylinder where z - b or z - c is not large compared with a.
Furthermore, a constant f produces a slowly varying velocity field along the
length of the cylinder. The translational velocity is the maximum at the
134 4 Flying and Swimming

midpoint, but the percentage variation over most of the length of the cylinder
is not large.
For the solution named above to be valid, the length of the cylinder b + c
must be larger compared with a, but it must remain finite, because the integrals
do not converge if b, c tends to 00. This is an illustration of the well-known
paradox that the Stokes equations, Eqs. (1) and (2) of Sec. 4.7, possess no solu-
tion satisfying the boundary conditions of a uniform motion of an infinitely
long circular cylinder.
The corresponding solution for a cylinder moving in the direction of its
axis is obtained by superposing stokeslets of strength (fdx, 0, 0) in each element

°
of length dx in the segment -b < x < c of the x-axis. See Fig. 4.8:2(b). The
velocity on the cylinder y2 + Z2 = a2 at x = is U T :

UT = 8~1l ( - 2 + 210g ~: ' 0, °). (6)

In this case, no dipole is needed.

These results are often expressed in terms of the coefficients of resistances
to motions of the cylinder normal and tangential to itself, KN and K t , respec-
tively, defined as the ratios of the force per unit length f to the velocities at
the midsection of the cylinder:
(7)
UN and U T being the magnitude of the vectors given by Eqs. (5) and (6),
respectively. Hence
8Tell 4Tell
KN = 1 + log(4cbja2) , KT = . (8)
-1 + log(4cbja 2)
Note that the resistance decreases with increasing length. Note also that if
log(4cbja 2) is large compared with 1, the value of KN is almost twice as large
as K T • Thus the resistance to normal motion is almost twice that to tangential
motion.
This difference in resistance coefficients KN and KT offers the principal key
to the understanding of the propulsion by undulatory motion of the flagellum
or cilium. The transverse undulation in a flagellum exert a greater force on
the fluid by its normal component than by its tangential component. The
normal component can be made to propel while the tangential component
drags.
Gray and Hancock (1955), Hancock (1953), Gray (1968), Lighthill (1969,
1975), Chwang and Wu (1971, 1974-1976), made extensive use of this kind of
approach to analyze the propulsion of small organisms by flagella or cilia. An
outline of their basic analysis is presented in the next section.
Parallel with the hydrodynamic study, there is another body of research
looking into the biology of the flagella and cilia. The flagella of some organ-
isms are spiral shaped and propel by rotation similar to a propeller. Other
organisms have flagella that undulate in a plane. What are the intrinsic
4.9 Resistive-Force Theory of Flagellar Propulsion 135

oute r

/;nne,
nexin link arm

tubule A
---- arm

tubule B spoke

FIGURE 4.8:2 Interpretive diagram of the structures seen in electron micrographs of

cross-sections of a flagellar or ciliary axoneme. (Modified from an original by K.E.
Summers.) From c.J. Brokaw and I.R. Gibbons (1975), by permission.

structure and mechanism of these flagella that imparts the wave motion?
Almost all flagella have a unique 9 + 2 structure illustrated in Fig. 4.8:2. The
current concept of flagellar motion is built around the possibility of sliding of
these internal filaments in analogy with the actin and myosin fibers in the
muscle. The flagellar motion is modulated by ATP and Ca + + flux. Studies of
these biochemical events yield valuable insights. A convenient reference to
many investigations of this nature is the two-volume book edited by Wu et
al. (1975).

4.9 Resistive-Force Theory of Flagellar Propulsion

Small animals which propel with flagella or cilia usually have lengths less than
1 mm. Some flagella perform undulatory motion in a plane. Some have a spiral
form and rotate like a screw. Some combine undulation and rotation. The
motion appears in general as a wave travelling along the body of the animal
from head to tail. Let us consider a flagellum undulating in a plane. Figures
4.9: 1-4.9:3 show several views of the flagellar motion. Figure 4.9: 1 shows what
appears to an observer who moves with the waveform. Figure 4.9:2 shows
what appears to an observer who moves with the animal. Figure 4.9:3 shows
the motion of the animal as it appears to an observer fixed in the laboratory.
The animal moves with an average velocity U relative to the undisturbed fluid
 136 4 Flying and Swimming

Mean

--
fluid
velocity
c
.---.

---
V-U

Animal vel

FIGURE 4.9:1 The appearance of a flagellum as seen by an observer who moves with
the waveform. Animal appears to move with a velocity c along the wave. In projection
on the x-axis the animal moves forward with an average speed V = IXC. The free stream
appears to move at velocity V-U .

..

u
.. V

FIGURE4.9:2 Observer moves with animal's head. The body wave appears to move
backward with velocity V. The undisturbed free stream appears to flow at velocity U.

FIGURE 4.9:3 Animal swims in a pool in laboratory. Observer in laboratory frame of

reference. The undisturbed fluid is stationary. Animal forward velocity is U.

(called 'free stream') The undulation wave of the flagellum moves backward
(toward the tail) with a velocity V relative to the animal's body.
Lighthill (1975), following Gray and Hancock (1955), analyzed the motion
shown in Fig. 4.9:1 as follows. A flagellum is assumed to have a length L when
stretched straight. To describe the wave form he expresses the coordinate x
of a point on the waveform by a function X(s), s being the distance measured
4.9 Resistive-Force Theory of Flagellar Propulsion 137

along the curve. Similarly, the coordinates y and z are expressed. Thus, the
waveform of the flagellum is described by
(x,y,z) = (X(s), Y(s),Z(s)). (1)
Since s is the arc length the derivatives of X(s), Y(s), Z(s) satisfy the equation
(2)
Hence X'(s), Y'(s), Z'(s) are the direction cosines of the tangent to the wave-
form curve. For an inextensible flagellum moving along the waveform curve
with a constant speed c, the points on the flagellum can be described by
(x, y, z) = (X(s - ct), Y(s - ct), Z(s - ct)). (3)

The point s = 0 is identified as the flagellum's head at time t = O. The speed

c differs from the horizontal velocity V only in that c is measured along the
curved waveform whereas V is its horizontal projection. c is larger than V in
the same proportion as the curved length of the flagellum L is larger than the
horizontal projection of the flagellum.
If the forward velocity of the animal is U while the wave moves backward
on the animal's body with a velocity V, then the free stream would appear to
move forward with a velocity V - U relative to the waveform (see Fig. 4.9:1).
The velocity relative to the fluid of any section of the flagellum is then the
difference of the velocity c directed along the forward tangent ofthe waveform
and the vector V - U. Figure 4.9:1 shows that its component along the
backward tangent is
VT = (V - U)cosO - c, (4)
where 0 is the angle between the local tangent to the flagellum and the x-axis.
Its component along the backward normal, VN , is
VN = (V - U)sinO. (5)
As we have noted earlier, following Eq. (2),
cosO = X'(s - ct), sinO = [1 - X,2(S - ct)]1/2. (6)
Lighthill assumes the resistive force acting on any section of the flagellum
to be linearly proportional to the local velocity of the section relative to the
undisturbed fluid. Thus the tangential and normal forces per unit length of
the flagellum are
(7)
respectively, where K T , KN are resistance coefficients as defined in Eq. (4.8:7).
The sum of the horizontal components of these forces is the thrust per unit
length of the flagellum, and an integration of this thrust over the length of the
flagellum yields the total thrust T:

T = total thrust = IL (fT cos 0 + iN sin 0) ds. (8)

138 4 Flying and Swimming

With Eqs. (4)~(7), this becomes

T= LL {KT[(V - U)X'(s - ct) - c]X'(s - ct)

+ KN(V - U) [1 - X'2(S - ct)]} ds. (9)

Let us write

f: X'(s - ct)ds = f: cos(}ds = aL,

(10)
LL X,2(S - ct) ds = LL cos 2 (} ds = f3L.

Then a, 13 are the average values of cos (} and cos 2 (} over the length of the
flagellum. a, 13 are functions of t. Hence, on noting that V = ac, Lighthill
obtains the total thrust T:
T = KTL[(V - U)f3 - V] + KNL(V - U)(1 - 13). (11)
For a free organism there is no net force acting on the body (inertial force
being neglected since the Reynolds number is «1). If the entire animal is like
a flagellum (e.g., a nematode), then T = 0 and Eq. (11) can be used to solve
for the speed of swimming U. If the flagellum propels an inert head (e.g. a
spermatozoon), then T must balance the drag of the head. Lighthill (1975)
writes the drag of the head as KNLU D, where Dis a coefficient defined as
D= resistance to forward motion of the head (12)
resistance to uniform normal motion of whole flagellum
Then on equating (11) with K NL U D, the ratio of the swimming speed U to the
wave speed V is obtained:
U (1 - 13)(1 - rK )
(13)
V 1 - 13 + rK f3 + D'
where
(14)
If 13 = 1 (the flagellum is flat and horizontal), then U = 0 (there is no forward
velocity). As 13 decreases from 1 (amplitude of undulation increases) the ratio
U IV increases but cannot exceed the limiting value (as 13 -+ 0):

(15)

Thus the maximum achievable speed of swimming depends on the ratio rK of

the tangential and normal force coefficients.
Example 1. Consider a small organism whose length L is very short com-
pared with the wave length of a planar undulation waveform, as illustrated in
4.9 Resistive-Force Theory of Flagellar Propulsion 139

Fig. 4.9:1. Then to an observer moving with the animal the organizm will be
seen as shown in Fig. 4.9:2. Let the waveform in Fig. 4.9:1 be given by the
equation
y = A sin(2nslA) == A sin(2nxI2) (16)
where A is the wavelength along the curve and 2 is the wavelength in the
x-direction. Then the configuration of the animal in Fig. 4.9:2 can be repre-
sented by
y = A sin [2n(x - Vt)!2] (17)
in which x is measured from the head toward the tail. The slope of the animal
is given by
ay 2n
tane = ax = A Tcos[2n(x - Vt)/2]. (18)

Since the length L is assumed to be much shorter than 2, xlA is very small,
and we have the following approximate expression for the entire length of the
organism
2n 2nVt
tane == A-cos-- (0 ~ x ~ L). (19)
A A'
Hence from (10),
4n2 2nVtJ-l
f3 = cos 2 e = (1 + tan 2 er 1 = [ 1 + A2~COS2_A- (20)

A substitution of Eq. (20) into Eq. (13) gives the ratio of the instantaneous
speed of the organism's swimming, U, to the speed of undulation wave V.
Example 2. Ciliary Motion. The method developed above can be used to
analyze the function of cilia. Cilia are cylindrical cellular projections having
t
a uniform diameter of about J..lm and composed of a characteristic array of
longitudinal fibrils. Oscillatory bending movements of cilia are responsible for
the propulsion of fluids over the cell surfaces, resulting in the locomotion of
small, unattached organisms such as Opalina and Paramecium, or in the
maintenance of currents of water or mucus around the cells of larger or
attached organisms. A Frog catches an insect by a flip of the tongue, which is
coated with mucus that glues the insect to it, and then sends the insect down
the throat by ciliary motion. Man clears the respiratory tract with cilia.
Female mammals transport ova in the ovary duct with cilia.
Figure 4.6:2 shows a characteristic pattern of the ciliary beat cycle. Each
cycle consists of two phases: an effective stroke during which the cilium
remains fairly straight and moves through an arc around its basal attachment,
and a recovery stroke in which a region of bending is propagated from base
to top. Details can be found in Sleigh (1974). Typically, a cilium 12 J..lm long
may beat at a frequency of 30 Hz, the effective stroke occupying about 2/5 of
140 4 Flying and Swimming

the cycle, the maximum tip speed is about 2.5 mm/sec. The cilia in the trachea
are about 5 f.lm long. The typical Reynolds number of their motion is «1.
Further Development. The importance of the normal and tangential resis-
tance coefficients KN and KN and their ratio r k is shown in Eqs. (13) and (15).
Equation 4.8:8 gives theoretical values of these constants; but it is based on
an analysis which ignores the curvature of the flagellum, and the proximity of
any other flagella or solid walls. In a flow that has a very low Reynolds number
the mutual interference of solid bodies in a flow field is often surprisingly great.
There is also a difficult question of how to assign values to the constants a, b,
c in Eqs. (3) to (8) of Sec. 4.8: to a because sometimes the surface of the animal
body is so soft (perhaps almost like water) that the effective radius is not what
is seen in the microscope, to band c, because they are somewhat arbitrary;
especially for a curved flagellum. These questions, as well as the rotation of
the animal, are discussed in Lighthill (1975), Chwang and Wu (1971, 1974-
1976), and Wu (1976). The questions of swimming efficiency and scaling (size
effect) are discussed by the same authors, especially Wu (1976).

4.10 Theories of Fish Swimming

There are three approximate theories of aquatic animal locomotion at higher

Reynolds numbers (» 1). They are all developed for animals with elongated
bodies, along the lines of the so-called "slender-body" theory:
(a) For aquatic animals such as the nematode and water snake, a "resistive
force" theory similar to that presented in the preceding section is used. In this
theory the main force of interaction between the animal and the surrounding
fluid is considered to be the resistance caused by the viscosity of the fluid. The
resistance law may be linear (proportional to the local tangential and normal
velocity) or nonlinear (e.g., proportional to the square of velocity), depending
on individual cases.
(b) For fish such as the carp and mackerel, a "reactive force" theory was
developed by Lighthill (1975). In this theory, the force of interaction between
the animal and the surrounding fluid is separated into two parts: the force
tangential to the body of the fish is considered to be resistive. The force normal
to the body of the fish is considered to be reactive, generated by the inertia of
the fluid. The inertial force is equal to the rate of change of the momentum of
the fluid. At large Reynolds numbers, the momentum of the fluid is essentially
concentrated in a cylinder enveloping the animal with a radius not much
greater than the lateral dimension of the animal, and can be calculated from
the 'virtual mass' of the fluid associated with the moving body.
Lighthill's (1975) theory is based on the following three observations:
i) Water momentum near a section offish is in a direction perpendicular to
the backbone and has a magnitude equal to the virtual mass per unit
length m times the component of the fish's velocity in that direction w.
4.10 Theories of Fish Swimming 141

'--------_x

FIGURE 4.10:1 Coordinates and symbols used in the analysis of fish swimming. The
view of a fast swimming fish as seen by one looking down the fish's back. The fish
swims in the x, y-plane. Its caudal fin is parallel to the z-axis which is perpendicular to
the x, y-plane. Hence the caudal fin appears only as a line. The heavy dash-dot line
represents the fishes spine, which bends in the xy-plane as the fish swims.

ii) Thrust can be obtained by considering the rate of change of momentum

within a control volume V which encloses the fish. For convenience, one
surface of the control volume is selected to be a flat surface S perpendicu-
lar to the caudal fin through its posterior end at each instant of time, as
shown in Fig. 4.10:1.
iii) In the momentum balance it is necessary to take into account transfer of
momentum across the surface S named above, not only by convection but
also by the action of the pressure generated by the motion within the
plane S.
The fish's motion is then described by the motion of its spine, which is
assumed to be inextensible, of total length L, and can bend only in the (x, y)
plane. The analysis is quite sophisticated. The final result is as follows. The
resulting force acting on the fish, F, is a vector in the (x, y) plane:

F=[mW(aY,_ax)_!mw2(ax,ay)] _~fL mwnds. (1)

at at 2 as as s=L dt 0
In this formula, m, w is explained in i) above, s = L is where caudal fin is
located. The coordinates (x, y, z) of a point on the spine at time t are functions
of sand t:
x(s, t),
x= y = y(s, t), z=O (2)
from which the derivatives ay/at, ax/at, ay/as, ax/as are computed. n is a unit
vector perpendicular to the spine (Fig. 4.10:1) in the horizontal plane z = o.
142 4 Flying and Swimming

The unit vectors t and n tangential and perpendicular to the spine are,
respectively:
n = (_ oyas' ax)
as . (3)

The horizontal velocity vector v of a point on the spine is

v= (~;, ~~). (4)

Its components tangential and normal to the spinal column are u and w,
respectively:
ax ax oyoy
u=--+--
oyox ax oy
w=-----. (5)
at as at as' at as at as
The resulting force F varies with time. If the swimming movement is
periodic, then the mean value of the last term in Eq. (1), which is a time
derivative, must be zero. Hence the mean force acting on the fish is:

~
F = [ -muwn
1 2
+ zmw t]s=L = mw
{(Oyat 'at
ax) - 21 mw (axas' Oy)}
2
as S=L' (6)

where the bar denotes the mean. If the x-axis coincides with the mean direction
of motion, then the mean thrust is the x component of F:

Fx = [ mw ~~ - ~ mw ~; I=L .
2
(7)

The same method gives an expression for the mean rate of energy
dissipation
15 = nmw 2 u]s=L (8)
by shedding of kinetic energy tmw
2 per unit length at velocity u across the

plane S into the vortex wake. The mean speed of swimming is

~ [ax]
u= u- (9)
as S=L·
The efficiency of propulsion is

(10)
1J = UFx + D
Thus it is seen that the mean terms Fx , D, U, and 1J are all determined at the
posterior end of the caudal fin. To obtain large thrust and high efficiency, the
fish should have a deep caudal fin and execute a large amplitude movement.
The shape and motion of the front portion of the fish really do not matter
very much. In the evolution of the fish the tail end became "compressed," i.e.,
narrowed down so that the tail became more flexible and more able to execute
4.11 Energy Cost of Locomotion 143

large amplitude oscillations; whereas the caudal fin increased in depth (the
virtual mass m is increased in proportion to the square of the depth of the
caudal fin at the posterior end). These fish can achieve an efficiency in the 80%
range, in sharp contrast to the values around 50% expected at a high Reynolds
number from a resistive interaction with perpendicular motions.
Example. A fish swims in a pool. With X-, y- axes fixed in the pool and s
measured along the fish's spine (Fig. 4.10:1), let the undulatory motion of the
fish be described by functions specifying the coordinates of points on the
spine as in Eq. (1). Assume that the fish swims in the way sketched in Figs.
4.9: 1-4.9:3, so that
x = s - Ut, y = h(s - Vt), (11)
where U is the forward speed of the fish, V is the speed of the undulatory wave
passing backward along its spine, h is a function of the variable ¢ = s - Vt.
At t = 0, Y = h(s) describes the waveform of the fish. Then, according to Eqs.
(5) and(ll),
ah
w= -(V - U) as. (12)

Writing Was the lateral velocity of the fish relative to the pool
ayah ah
W=-=-= -V- (13)
at at as
and substituting into (12), we obtain an important relation between the lateral
velocity relative to water and that relative to the pool:

(14)

(c) For fish such as the dolphin and whale, with well developed lunate tails
which look like airplane wings, a "two-dimensional section" theory of the tail
was developed by Wu (1971). These tails have high propulsive efficiency. The
theory pays attention to the vortex wake due to the oscillatory motion of the
tail, much like the wing theory for birds and airplanes described in Chapter 3.
In recent years, extensive computational aerodynamics was developed in
the aeronautical field. Aeroelastic design of airplanes is based on computation
of three-dimensional unsteady flow using concepts outlined in Chapter 3. A
corresponding development for the analysis of flying of birds and swimming
of fish has not taken place.

4.11 Energy Cost of Locomotion

To evaluate the energy cost of locomotion, we must evaluate drag; because

the product of drag and velocity of motion is power. The integral of power
over time is the energy or work done by the drag force. Thus, the energy cost
144 4 Flying and Swimming

in a time interval (t l , t 2 ) in which the body moves from Xl to X2 is

(1)

Here D is drag, V is velocity = dx/dt.

There are several sources of drag. For birds and insects, we have
a} the skin friction acting on the body surface;
b} the pressure drag produced by eddy or wake formation (see Sec. 3.15);
c) the induced drag, (see Secs. 3.2, 3.12, and 3.13)
For fish, we should add
d) the wave resistance,
e) the air resistance on the part above the water.
Of these, the skin friction is the hardest one to evaluate. Even today theoretical
analysis based on Navier-Stokes equations has not been able to adequately
predict the drag coefficient as a function of the Reynolds number for a flat
plate in a uniform flow when turbulence is involved. Hence one must rely on
experimental results.
Direct measurement of drag of animals is very difficult (cf. Sec. 3.15). The
wide scatter of data in the literature is evidence of this difficulty. Hence animal
physiologists often take an alternative approach: by measuring the animal's
oxygen consumption, applying the standard energy conversion factors to
obtain the metabolic power, and then using the energy conservation equation
and flight efficiency to evaluate the drag. By the principle of conservation of
energy, and writing D for drag, V for velocity, P for metabolic power, and '1
for efficiency, we have
DV= '1P, (2)
The power P is appropriately identified ifthe basal metabolic rate is subtracted
from the measured metabolic rate during active swimming. The overall effi-
ciency '1 is equal to '1m'1h' where '1m is the "muscle efficiency" with which
biochemical energy is converted to mechanical power of the muscle, and '1h is
the fluid dynamic efficiency of propulsion.
Figure 4.11:1 shows the results presented by Wu (1984) on the basis of
Brett's (1963) measurement of sockeye salmon (Oncorhynchus nerka) and the
analysis of Wu and Yates (1978). Shown are the data points of the power
coefficient Cp , and the dead-drag coefficient DDd versus the Reynolds number
Re, where
UL
Re=-. (3)
v

P is metabolic power of a swimming fish, p is the water density, v is the

kinematic viscosity of water, L is the fish length (of a single-sized group with
mean L = 0.178 m) and S is the wetted surface area of the fish. A least-square
4.11 Energy Cost of Locomotion 145

\
--
[Slope= -0.40

10'
\
A ~USTAINEO t~
"'1 Cp
\

, Vcrlt
("'1=0.25) '"

',,---
TurbUlent Co - Flat Plate

--- --- --
- _ _ . /(Slope = - 0.20)

--
laminar CD - Flat
(Slope = -0.50)

FIGURE 4.11: 1 Data on the power and drag coefficients of sockeye salmon and com-
parison with the theoretical results of Wu and Yates (1978).

error fit to the drag data yields

CDd = 15.4R;0.4, (4)
On the other hand, the metabolic power coefficient Cp has two branches
meeting near the minimum value of 0.07: one branch for sustained swimming
at lower speeds, another branch for a burst of energy. The efficiency 11m is
estimated by Webb (1975) to be in the range 0.2-0.3. The efficiency 11h is about
0.9 (Wu, 1971), hence 11 is about 0.25. With 11 = 0.25, the values of 11Cp are
calculated and plotted in Fig. 4.10:1. This curve lies considerably below the
dead-drag coefficient CDd' It is suggested that the dead fish data are not useful;
146 4 Flying and Swimming

2 -

!zw 10-'
.~igeOn
U Falcon
u:

':~
u.
w
o()
Cl
«([
o _Astra-Mite........... • Parrott's Vulture

--
w 10-'
f- • --.::::::::--___ SHKbsailPlane
;u:;;---
--
Ci5
« - - __
a: TURBULENT
«

----
11.
L4MIN;:- _ _
RPL4TE --

---:-
10-3 L--_ _...l-_ _-L._...l--L...-L..._ _-..:..._ _--!.._-1...---l---lL--_ _.l....-_--'
10' • 10' B 10 6

REYNOLDS NUMBER

FIGURE 4.11:2 Data on drag coefficients of birds and sail planes. From Kuethe and
Chow (1986), and from Tucker and Parrott (1970). Reproduced by permission.

and live fish has drag coefficients considerably lower than those of the dead
fish.
A corresponding plot for the birds and sail planes is shown in Fig. 4.11:2.
A comparison of Figs. 4.11:1 and 4.11:2 shows that birds and fish are in the
same class as far as drag coefficients are concerned. They are not as efficient
as contemporary sailplanes (Tucker and Parrott, 1970).
Knowing the drag as a function of speed of flight, one can compute the
energy cost according to Eq. (1). It is interesting to examine some results in
the following.

Cost of Distance
If distance covered is the major consideration, one should consider the energy
cost of transport for moving a unit weight for unit distance. To maximize
distance and minimize the cost, the cruising speed should be so chosen
that dE/dx is a minimum (E being the energy, x being the distance). Since
dE/dt = Power P, and dx/dt = speed V, we see that dE/dx = P/v. Hence we
must determine Vat which P/Vis a minimum. For animals, this can often be
done by measuring the metabolic energy (oxygen consumption, CO 2 release
lactate) at various speeds of locomotion. With the speed at which PlY is a
minimum so determined, a dimensionless parameter

P
vw i lIC
= spec lresIstance
· = cost 0 f transport (5)

can be plotted as a function of the weight of the animal as shown in Fig. 4.11 :3.
4.11 Energy Cost of Locomotion 147

IS

:::::
::."
0 •
1·0 INseCTS

~... 00
0 RUNNERS
~ 00
"-
0·5
~
~ ,.
~ FLYERS A o HCLlCOPTCIf
~
'-' 0
~
"-
'"
C)
~ -0·5

'...."
CAT LE HORSES AvrO$
C) aPEDAL AIRPLANE

-/-O
D8/crCLE

-5 -4 -3 -2 -I o 2 3 4 5 6 7
LOG WEIGHT (NEWTONS)

FIGURE 4.11:3 The minimum energetic cost of transport to move a unit of body weight
for a unit of distance in swimming, flying, walking and running animals of various
weights. Data also are included for various machines. The dashed line labeled "theory"
is that predicted for flying birds by Eq. (26) in the text. Other curves are least squares
regression lines. Data for swimmers from Schmidt-Nielsen (1972). Other data from
Tucker (1970), (1973b). Reproduced by permission from Tucker (1975).

Here one sees that the fish are the best performers, followed by the flyers,
runners, walkers, and manmade vehicles. The "theoretical" curve shown in
Fig. 4.11:3 is Tucker's (1975) semiempirical formula for birds and bats in level
flight:
Pi = (0.00723h + 105.9)m1. 3B2 b-1.236 (6)
in which Pi is the power, h is the altitude, b is the wing span, in Sf units of
Watts and meters.
The specific resistance is a kind of global friction coefficient for an animal
or a machine; it is a ratio of the total resistance to the gross weight of the body.

What Price Speed?

In the competition of the fittest, a great role is played by the speed of
locomotion. A possible figure of merit is the "minimum" value of the "power
per unit weight" as a function of the "maximum speed." Theodore von Karm{m
and Gabrielli (1950) chose to use the same dimensionless parameter, PIWV,
defined in Eq. (5) but evaluated at the maximum speed Vmax , to represent the
cost of speed. Here P is the power, W is the weight, V is the speed, in consistent
units such as kg-mis, kg, m/s; or lbs-ft/sec, lbs, and ft/sec, respectively. They
plotted available data of specific resistance versus Vmax for various classes of
148 4 Flying and Swimming

?
I II
'fydrogfitferfrecorrl)
'·0
o I I

0·.1
! I
O¥
;::ceC<7!' ! SlrjilrNIr ~
t'--<L)
lIelicople!'

(1-,r
:; OJ
'->
.~ o·?
I ,I
?Tonf/~/i'P/<7ne-",~c-+-
~IIglile!'
kl
:". 1l 1 liolo!'l)'de 0 :::~d
~
;P?~
I""~
~ ~ 1
Ir~~er _- +-! - -/ StOp/Me
"" :!o:.. 0-10 de;trian~
\leo ~
1-- ~~
C00~1'--
.~~ 0;'%~c' f-l/
31
.::;
"'~" :9 0·05
0'07' ~P-
cllOI lL ~I\
II \~(lO 'o""'",,,,/W,/
..-Aif'Sliip
!ruds wdlioul
/
....
.~

r
~ I!'oile~

-I
O'OJ
~ 3ubmo!'ine 01-~
'S;
~
~
:i!: O'O? su!'f<7ct
v
Rar;t : j
Se
I ,/
'.

::;
"
.2
.;) 0·01
se
-~
J
- '"

v
'
AulorolZ

~
oS:;;
'$
";->
;§
3uomo!'ine
suomu!'j'ed V / ~/
'80 file ship

0'005
.!i;
0·009 //
~ ~ lie!'clion/ sllip./J '
~ O·OOJ
:;"-
"- O'OO?

0·001
1 I J 9 5 10 10 JO 9050 100 100 JOO 500 1000 ?OO()
Y=mox.speedlilmpli.

FIGURE 4,11:4 Specific resistance of single vehicles according to von Karman and
Gabrielli (1950),

vehicles and sketched a continuous curve to represent the minimum specific

resistance as a function of the maximum speed for each vehicle. When these
empirical curves are plotted together, they obtained the result as shown in
Fig. 4.11 :4. The limiting line shown in the figure is

(~)
WV max speed
= 1.75 X 1O- 4 Vmax (mph) = 3.92 X 1O- 4 Vmax (m/sec) (7)

which represents a kind of limit that human technology has achieved so far.
Horse and man are worse by an order of magnitude.

4.12 Cell Movement

The movements of ameba in fluid, leukocytes in blood and blood vessel

walls, and neuron and smooth muscle cells in tissue have been observed. To
understand these movements attention has to be given to the internal structure
Problems 149

of the cells, and adhesion between cells. This subject is extremely important
to the understanding of clinical problems of atherosclerosis, myocardial in-
farction, hemorrhagic death, etc., as well as basic biology. It is beyond the
scope of this book to discuss this rapidly developing field. Interested readers
may consult the following references. On cell motility and Chemotaxis: Berg
and Brown (1972), Dembo et al. (1984), Dembo and Harlow (1986), Gallin and
Quie (1978), Goldman et al. (1976), Huxley et al. (1982), Oster and Perelson
(1987), Zhu et al. (1988), Zigmoid (1978). On cytoskeletal networks: Nossal
(1988), Stossel (1982), Taylor and Condeelis (1979). On actin filaments: Pollard
and Cooper (1986), Sato et al. (1985), Smith (1988), Wang (1985). On clinical
applications: Schmid-Schonbein and Engler (1986), Wilkinson (1974).

Problems

4.1 The shark uses cartilage to form its skeleton instead of bone. Its outer shell is a
cartilaginous bag. Muscles are attached to the skin. When muscles contract the
cartilaginous layer is subjected to strain and stress and its elastic strain energy is
varied. Imagine and describe how a shark can utilize the cartilaginous material for
its locomotion; then check a book on sharks to see if your imagination is right.
The idea that animals are properly designed with respect to materials and
structures is beautifully presented in the book Mechanical Design in Organisms
by S.A. Wainwright, W.D. Biggs, J.D. Currey and J.M. Gosline, published by
John Wiley and Sons, New York, 1976. It would be interesting to speculate on
why the shark was successful in its adaptation through evolution.
4.2 Trees must withstand forces imposed by wind. Their swaying and bending must
be related to an aerodynamic mechanism of reducing the drag force in a storm.
Consider some of your favorite trees in this regard and see how well they are
designed with respect to the specific environment each lives in.
4.3 At "Sea World" one can watch the whale jump out of the water to touch a ball
hoisted 15 feet up in the air. You see the whale make one swoop down and jump
right out. For a tank with 30 feet of water, estimate the achievable lift coefficient
of the whale's tail.
4.4 Man and whale have a diving reflex. As they dive down in the ocean, their heart
rates slow down. Discuss the physiological significance of this reflex.
4.5 The skin of the dolphin is compliant, and is coated by a polymer. Discuss in which
way the compliance of the skin and the viscoelasticity of the polymeric fluid can
affect the stability of the boundary layer on the dolphin's body, the laminar and
turbulent characteristics of the boundary layer flow, and the influence of these
factors on the drag force acting on a dolphin as it swims.
Immitating the dolphin, people have injected polymer into a ship's boundary
layer in the hope of improving the speed of the ship. They have also used
compliant paint on the hull. Fire fighters have introduced polymers into fire
engines so that water can be shot out to a greater height. What success have they
achieved? What possible disadvantage can there be?
150 4 Flying and Swimming

4.6 When a sperm swims toward an ovum, how is its performance affected by the
mechanical properties of the fluid? What characteristics of the rheological prop-
erties are important if the success of a sperm is judged by its ability to be the first
to arrive at the target.
4.7 There is evidence suggesting that the swimming abilities of male and female
sperms are different. Based on this observation, devise a method to separate male
and female sperms in the semen for artifical insemination for sex control.
4.8 Migrating geese are always seen to line up in the form of the letter of A. One
might suspect that there may be an energy saving in doing this as compared with
a disorganized crowd. Can you offer some theoretical evidence? If an experi-
mental approach is preferred, design an experiment. (Ref: Higdon, JJL, and
Corsin, S. Induced drag of a bird flock. Am. Naturalist 112: 727-744, 1978.)
4.9 A chicken cannot fly very well. What is wrong with the chicken's flying machinery?
4.10 The movement of an ovum in oviduct is most likely helped by cilia lining the
oviduct. Describe these cilia and the way they can help the ovum.
4.11 What are the basic factors limiting the speed of flight of a bird? For example, by
diving from a very high altitude, a bird could achieve high speed. What determines
the maximum speed achievable? What determines the maximum speed desirable?
What ill effects will result if the speed exceeds certain critical values? What critical
values are there?
4.12 Active control is undoubtedly used by birds in their flight, but not much is known
about the subject. As a theoretical preliminary to an investigation, list all the
important problems of flight of concern to a bird (it would be simpler if you
choose a specific species) and, with a proper block diagram, describe desirable
neuromuscular control systems that will enable the bird to solve each of these
problems.
4.13 An airplane has to rev up the engine to produce the maximum power to take off.
How does an eagle take off? One would have to consider the aerodynamic,
structural, muscle dynamics, kinematics, metabolic, and energy factors.
4.14 The take-off problem for a duck on water is quite different from that of an eagle
on a rock. How does the duck do it?
4.15 The locomotion methods of sea horse, clams, star fish, snails, and slugs are equally
worthy of study. If you cannot find a satisfactory analysis in the literature, devise
a plan of study of your own.

References
Ayman, G. (1936). Bird Flight, Bodley Head, London.
Bainbridge, R.A. (1958). The speed of swimming of fish as related to size and to the frequency
and amplitude of the tail beat. J. Exp. BioI. 37: 109-133.
Bainbridge, R. (1960). Speed and stamina in three fish. J. Exp. BioI. 37: 129-153.
Bainbridge, R. (1963). Caudal fin and body movement in the propulsion of some fish. J. Exp. BioI.
40: 23-56.
Berg, H.C. and Brown, D.A. (1972). Chemotaxis in Escherichia coli analyzed by three-dimensional
tracking. Nature London 239: 500.
References 151

Blake, J.R. and Sleigh, M.A. (1975). Hydromechanical aspects of ciliary propulsion. In Swimming
and Flying in Nature (T.Y. Wu, CJ. Brokaw and C. Brennen, eds.), Vol. 1, Plenum Press,
New York, pp. 185-209.
Bone, Q. (1975). Muscular and energetic aspects of fish swimming. In Swimming and Flying in
Nature (T.Y. Wu, c.J. Brokow and C. Brennen, eds.), Vol. 2, Plenum Press, New York,
pp.493-528.
Breder, C.M. (1926). The locomotion of fishes. Zoologica 4: 159-297.
Brett, J.R. (1963). The energy required for swimming by young sockeye salmon with a comparison
of drag force on a dead fish. Trans. Roy. Soc. Can. 1: Sec. IV, 441-457.
Brokaw, CJ. and Gibbons, I.R. (1975). Mechanisms of movement in flagella and cillia. In
Swimming and Flying in Nature (T.Y. Wu, CJ. Brokaw and C. Brennen, eds.), Vol. 1,
Plenum Press, New York, pp. 89-126.
Brown, R.HJ. (1953). The flight of birds. II. Wing function in relation to flight speed. J. Exp.
Bioi. 30: 90-103.
Childress, S. (1981). Mechanics of Swimming and Flying. Cambridge Univ. Press, Cambridge.
Chwang, A.T. and Wu, T.Y. (1971). A note on the helical movement of micro-organisms. Proc.
Roy. Soc. B., 178: 327-346.
Chwang, A.T. and Wu, T.Y. (1974, 1975, 1976). "Hydromechanics ofiow-Reynolds-number flow.
Part I: Rotation of axisymmetric prolate bodies. J. Fluid Mechanics 63: 607-622,1974. Part
II: Singularity method for Stokes flows. ibid. 67: 787-815, 1975. Part III: (Chwang alone)
Motion of spheroidal particle in quadratic flows. ibid. 72: 17-34, 1975. Part IV: Translation
of spheroids. ibid 75: 677-689, 1976.
Dalton, S. (1975). Borne on the Wind, The World of Insects in Flight. E.P. Dutton and Co., New
York.
Dembo, M. and Harlow, F. (1986). Cell motion, contractile network and the physics of interpene-
trating reactive flow. Biophys. J. 50: 109-121.
Dembo, M., Harlow, F. and Alt, W. (1984). The biophysics of cell motility. In Cell Surface
Dync.zmics: Concepts and Models. (A.S. Perelson, C. DeLisi and F.W. Wiegel, eds.), Marcel
Dekker, New York, pp. 495-541.
Duncker, H.R. (1972). The structure and function ofbird's lung. Respiration Physiol. 14: 44-63.
Ellington, c.P. (1975). Non-steady-state aerodynamics of the flight of Encarsia formosa. In
Swimming and Flying in Nature (T.Y. Wu, C. Brokow and C. Brennen, eds.), Plenum Press,
pp. 783-796.
Fung, Y.c. (1965). Foundations of Solid Mechanics, Prentice-Hall, Englewood Cliffs, NJ.
Fung, Y.c. (1977). A First Course in Continuum Mechanics, 2nd edn., Prentice-Hall, Englewood
Cliffs, N.J.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living Tissues, Springer-Verlag, New
York.
Gallin, J.I. and Quie, P.G. (eds.) (1978). Leukocyte Chemotaxis: Methods, Physiology and Clinical
Applications, Raven Press, New York.
Goldman, R., Pollard, T., and Rosenbaum, J. (eds.) (1976). Cell Motility. Cold Pring Harbor
Conferences on Cell Proliferation. Cold Spring Harbor Press, New York.
Goldspink, G. (1977). Design of muscle in relation to locomotion. In Mechanics and Energetics
of Animal Locomotion (R.McN. Alexander and G. Goldspink, eds.), Chapman and Hall,
London.
Gray, J. (1939). Aspects of animal locomotion. Proc. Roy. Soc. London, B 128: 28-62.
Gray, J. (1953). How Animals Move. Cambridge Univ. Press, Cambridge.
Gray, J. (1958). The movement of the spermatozoa of the bull. J. Exp. Bioi. 35: 96-108.
Gray, J. (1968). Animal Locomotion, W.W. Norton, New York; Weidenfeld and Nicolson, London.
Gray, J. and Hancock, G.J. (1955). The propulsion of sea-urchin spermatozoa. J. Exp. Bioi.
32: 802-814.
Gray, J. and Lissmann, H.W. (1964). The locomotion of nematodes. J. Exp. Bioi. 41: 135-154.
Hancock, G.J. (1953). The self-propulsion of microscopic organisms through liquids. Proc. Roy.
Soc. A 217: 96-121.
Huxley, H.E., Bray, B. and Weeds, A.G. (eds.) (1982). Molecular Biology of Cell Locomotion.
152 4 Flying and Swimming

Phil. Trans., Roy. Soc. London, B299: 145-327.

Jameson, W. (1958). The Wandering Albatross. Hart-Davis, London.
Kuethe, A.M. (1975a). On the mechanics of flight of small insects. In Swimming and Flying in
Nature (T.Y. Wu, c.J. Brokaw, and C. Brennen, eds.), Plenum Press, New York. pp. 803-813.
Kuethe, A.M. (1975b). Prototypes in nature. The carry-over into Technology. Technium, Engi-
neering Review. 1975, Univ. of Michigan.
Kuethe, A.M. and Chow, c.-Y. (1986). Foundations of Aerodynamics. 4th ed. John Wiley,
New York.
Lighthill, l (1969). Hydromechanics of aquatic animal propulsion-a survey. Ann. Rev. Fluid
Mech. 1: 413-446.
Lighthill, J. (1975). Mathematical Biofluiddynamics, Soc. Indus. Appl. Math. Philadelphia.
Lillienthal, O. (1889). Der Vogelflug als Grundlage der Fliegekunst. R. Oldenbourg, Berlin.
Maxworthy, T. (1981). The fluid dynamics of bird flight. Ann. Rev. of Fluid Mechanics, (M. Van
Dyke and J.V. Wehausen, eds.), Annual Reviews, Palo Alto, California.
McAlister, K.W., Carr, L.W., and McCroskey, W.J. (1978). Dynamic stall experiments on the
NACA 0012 airfoil. NASA Tech. Paper 1100.
Nachtigall, W. (1974). Insects in Flight (Trans. by H. Oldroyd et al). McGraw-Hill, New York.
Newman, IN. (1973). The force on a slender fish-like body. J. Fluid Mech. 58: 689-702.
Newman, J.N. and Wu, T.Y. (1973). A generalized slender-body theory for fish-like forms. J.
Fluid Mech. 57: 673-693.
Norberg, R.A. (1975). Hovering flight of the dragoufly Aeschna Juncea L., kinematics and
aerodynamics. In Swimming and Flying in Nature (T.Y. Wu, C.J. Brokaw, and C. Brennen,
eds.), Plenum Press, New York, pp. 763-781.
Nossal, R. (1988). On the elasticity of cytoskeletal networks. Biophys. J. 53: 349-359.
Oberbeck, A. (1876). Ueber stationiire Fliissigkeitsbewegungen mit Beriicksichtigung der inneren
Reibung. Crelle 81, 62-80.
Oster, G.F. and Perelson, A.S. (1987). The physics of cell motility. J. Cell Sci. 8: 35-54.
Pennycuick, c.J. (1968). A wind-tunnel study of gliding flight in the pigeon Columba livia. J. Exp.
Bio!. 49: 509-526.
Pennycuick, C.J. (1972). Animal Flight. Edward Arnold, London.
Peterson, R.T. (1963). The Birds. Time Inc., New York.
Pollard, T.D. and Cooper, lA. (1986). Quantitative analysis of the effect of acanthamoeba
profilin on actin filament nucleation and elongation. Biochem.. 23: 6631-6641.
Prandtl, L., and Tietjens, O.G. (1934). Applied Hydro-and-Aeromechanics. McGraw-Hill, New
York. (Translated from the German edition, Springer, Berlin/Heidelberg, 1931).
Pringle, lW.S. (1975). Insect Flight. Oxford University Press, London and New York.
Rayleigh, Lord., (lW. Strutt). (1883). The soaring of birds. Nature 27: 534-535.
Sato, M., Leimbach, G., Schwartz, W.H., and Pollard, T.D. (1985). Mechanical properties of
actin. J. Bioi. Chem. 260: 8585-8592.
Schmid-Schiinbein, G.W. and Engler, R.L. (1986). Granulocytes as active participants in acute
myocardial ischemia and infarction. Am. J. Cardiovasc. Patho. 1: 15-30.
Scheid, P., Slama, H., and Piiper, J. (1972). Mechanisms of unidirectional flow in parabronchi of
avian lungs. Measurements in duck lung preparations. Respiration Physiol. 14: 83-95.
Sleigh, M.A. (ed.) (1974). Cilia and Flagella. Academic Press, London and New York.
Smart, J., and Hughes, N.F. (1972). In Insect/Plant Relationships: Sympos. R. Entomol. Soc.
London No.6, pp. 143-155.
Smith, X. (1988). Neuronal cytomechanics: the actin-based motility of growth cones. Science
242: 708-715.
Stossel, T.P. (1982). The structure of the cortical cytoplasm. Phil. Trans. Roy. Soc. London B
299: 275-289.
Taylor, D.L. and Condeelis, J.S. (1979). Cytoplasmic structure and contractility in amoeboid
cells. Int. Rev. Cytology 56: 57-144.
Taylor, G.!. (1951). Analysis of swimming microscopic organisms. Proc. Roy. Soc. London Ser.
A, 209: 447-461.
References 153

Taylor, G.I. (1952). Analysis of the swimming oflong and narrow animals. Proc. Roy. Soc. London,
A, 214: 158-183.
Tucker, V.A. (1968). Respiratory exchange and evaporative water loss in the flying budgerigar.
J. Exp. Bioi. 48: 67-87, Company of Biologists Ltd.
Tucker, V. and Parrott, G.c. (1970). Aerodynamics of gliding flight of falcons and other birds.
J. Exp. Bioi. 52: 345-368, Company of Biologists Ltd.
Tucker, V.A. (1975). Aerodynamics and energetics of vertebrate fliers. In Swimming and Flying
in Nature (T. Wu, C. Brokaw and C. Brennen, eds). Plenum Press, New York, pp. 845-865.
Von Holst, E. and Kiichemann, D. (1974). Motion of animals in fluids. In J. Royal Aeronautical
Soc. 43: 39-56.
Von Karman, T. and Gabrielli, G. (1950). What price speed? Specific power required for propul-
sion of vehicles. Mech. Eng. 72: 775-781.
Wang, Y.-L. (1985). Exchange of actin subunits at the leading edge ofliving fibroblasts: possible
role oftreadmilling. J. Cell BioI. 101: 597-602.
Webb, P.W. (1975). Hydrodynamics and energetics of fish propulsion. Bull. Fish Res. Bd. Can.
190: 1-158.
Weis-Fogh, T. and Jensen, M. (1956). Biology and physics of locust flight. I. Basic principles in
insect flight. A critical review. Phil. Trans., Roy. Soc. London, B 239: 415-458.
Weis-Fogh, T. (1956). Biology and physics oflocust flight. II. Flight performance of desert locust
(Schistocera gregaria). Phil. Trans. Roy. Soc. London B, 239: 459-510.
Weis-Fogh, T. (1960). J. Exp. Bioi. 37: 889-907. J. Exp. Bioi. 59: 169-230.
Weis-Fogh, T. (1964). VIII. Lift and metabolic rate of flying locusts. J. Exp. BioI. 41: 257-271.
Weis-Fogh, T. (1967). Energetics of hovering flight in hummingbirds and in Drosophila. J. Exp.
BioI. 56: 79-104.
Weis-Fogh, T. (1973). Quick estimates of flight fitness in hovering animals, including novel
mechanisms for lift production. J. Exp. Bioi. 59: 169-230, Company of Biologists Ltd.
Weis-Fogh, T. (1975). Flapping flight and power in birds and insects, conventional and novel
mechanisms. In Swimming and Flying in Nature (T.Y. Wu, C.J. Brokaw, and C. Brennen,
eds.), Plenum Press, New York, pp. 729-762.
Wilkinson, P.C. (1974). Chemotaxis and Inflammation, Churchill Livingstone, Edinburgh and
London.
Wu, T.Y. (1971). Hydromechanics of swimming fishes and cetaceans. In Advances in Applied
Mechanics (C.S. Yih, ed.), 11: Academic Press, New York. pp. 1-63.
Wu, T.Y. (1971). Hydromechanics of swimming propulsion. Part 3. Swimming and optimum
movements of slender fish with side fins. J. Fluid Mech. 46: 545-568.
Wu, T.Y. and Newman, J.N. (1972). Unsteady flow around a slender flish-like body. Proc.
International Symp. on Directional Stability and Control of Bodies Moving in Water. Institu-
tion of Mechanical Engineers, London.
Wu, T.Y., Brokaw, c.J. and Brennen, C. (eds.) (1975). Swimming and Flying in Nature. Vols. 1
and 2, Plenum Press, New York.
Wu, T.Y. (1976). Introduction to the scaling of aquatic animal locomotion. In Scale Effects of
Animal Locomotion. (M.J. Lighthill and T.J. Pedley, eds.), Academic Press, London,
pp. 753-766.
Wu, T.Y. and Yates, G.T. (1978). A comparative mechanophysiological study offish locomotion
with implications for tuna-like swimming mode. In Physiological Ecology of Tuna (G.D.
Sharp and A.E. Dizon, eds.), Academic Press, New York.
Yates, G.T. (1983). Hydromechanics of body and caudal fin propulsion. Chapter 6 in Fish
Biomechanics. (P.W. Webb and D. Weihs, eds.), Praeger Scientific, New York.
Zhu, c., Skalak, R., and Schmid-Schiinbein, G.W. (1988). One-dimensional steady continuum
model of retraction of pseudopod in leukocytes. J. Biomech. Eng. 111: 69-77.
Zigmoid, S.H. (1978). Chemotaxis by polymorphonuclear leukocytes. (review) J. Cell Bioi.
77: 269-287.
 I

I
1

CRABS by Fung Chung-Kwang (1896-1952). Ink on rice paper.

CHAPTER 5

Blood Flow in Heart, Lung,

Arteries, and Veins

5.1 Introduction

The next five chapters are concerned with flow inside the bodies of man and
animals. By internal flow of blood, water, and gases, the cells of the body
obtain water, oxygen, and nutrients. To understand the health and disease of
these organisms, it is necessary to know the mechanics of internal flow.
A great variety of things happen in the circulatory and respiratory systems.
To observe them we use a variety of tools. Most important are our eyes. To
help our eyes we use x-rays, cinematography, CAT scan, NMR, ultrasound
imaging, etc. For smaller things, we use optical microscopes, x-rays, electron
microscopes, scanning tunneling microscopes. At different levels of scale,
different objects come into view. Together they reveal the phenomenon of
living. It is the function of mechanics to analyze and integrate the phenomena
at different scales. If the mechanics of a phenomenon at one level of scale is
called macroscopic, and that at a smaller level of scale is called microscopic,
then in biomechanics one often attempts to connect the microscopic to
macroscopic mechanics. For example, the dimension of an endothelial cell
is much smaller than the diameter of the aorta. The mechanics of the endo-
thelial cells is microscopic relative to the mechanics of the aorta. But they are
connected when atherosclerosis is concerned. Again, the diameters of the
collagen and elastin fibers are much smaller than the diameter of the arteries,
but fiber mechanics and arterial mechanics are connected.
In the following, we shall first study blood flow at the scale of the heart and
large arteries. Then (in Chapter 6) we focus on small blood vessels and attempt
to clarify the connection between the micro- and macro hemodynamics. In
Chapter 7 we do the same for the phenomena of respiration. In Chapters 8
and 9, the flow of water and other constituents from blood vessels to extra-

155
156 5 Blood Flow in Heart, Lung, Arteries, and Veins

vascular space, the fluid movement in the tissue space, and the fluid exchange
between interstitium and cells are discussed. The presentations in Chapters 5
and 6 are rather condensed. Interested readers are referred to the author's
book Biodynamics: Circulation (Springer-Verlag, 1984) for a more thorough
treatment. The material in Chapters 7,8 and 9 is presented at a more leisurely
pace.

5.2 The Geometry of the Circulation System

Every animal's circulation system is special, but we shall consider typical

features of man, dog, and cat. In these animals, blood flows from the superior
and inferior vena cava into the right atrium, then through the tricuspid valve
into the right ventricle, then through the semilunar valves into the pulmonary
artery, the lung, the pulmonary veins, the left atrium, the mitral valve, the left
ventricle, and finally through the aortic valve into the aorta. The peripheral
circulation begins with the aorta, perfuses various organs, and returns to the
right atrium. In each organ, flow begins in the arteries, perfuses the micro-
circulatory bed, then drains into veins. The vena cava collect blood from
various organs, and send it to the heart.
Figure 5.2:1 shows the heart in greater detail. The two thin-walled atria are
separated from each other by an interatrial septum. The two ventricles are
separated by an interventricular septum. The left ventricle is thick-walled. In
systolic condition, the pressure of blood in the left ventricle is higher than that
in the right ventricle; hence the interventricular septum bulges out toward the
right ventricle. The left ventricle can be represented as an ellipsoid; the right
ventricle can be represented as a bellow. The four valves are seated in a plane
and their bases are connected into an integrated structure, so that the enlarge-
ment of two opening valves is accompanied with the reduction in size of the
other two closing valves. The mitral and tricuspid valves are attached to
papillary muscles, which contract in systole, pulling down the valves to
generate systolic pressure rapidly, and prevent the valves from inversion into
the atrium.
The lung consists of three trees; see Fig. 5.2:2. The airway tree is for
ventilation. The trachea is divided into bronchi which enter the lung, subdivided
repeatedly into smaller and smaller branches called bronchioles, respiratory
bronchioles, alveolar ducts, and alveoli. The alveoli are the smallest units of
the airway. The walls of the alveoli are capillary blood vessels. Every wall of
an alveolus is exposed to gas on both sides, so each wall is called an interal-
veolar septum. The entire lung is wrapped in a pleural membrane like a
balloon. The chest wall also has a pleura. The pulmonary pleura and the
visceral pleura are apposed to each other with a very small gap (a few 11m
thick) between them. The sealed compartment of space between the pleura
is called the intrapleural space. The pressure in the intrapleural space is
ordinarily lower than atmospheric. The chest wall and the transpulmonary
5.2 The Geometry of the Circulation System 157

sve
Pu lmonary
veins

Ive
wall

FIGURE 5.2:1 Blood flow through the heart. The arrows show the direction of blood
flow. The symbols are: SVC, superior vena cava; IVC, inferior vena cava; RA, right
atrium; RV, right ventricles; PA, pulmonary artery; LV, left ventricle. The valves are
T, tricuspid, P, pulmonary, AO, aortic, M, mitral. From Folkow and Neil (1971)
Circulation, Oxford Univ. Press, New York, p. 153, by permission.

FIGURE 5.2:2 The three "trees" of the lung.

158 5 Blood Flow in Heart, Lung, Arteries, and Veins

pressure (= the difference of the alveolar gas pressure and the pleural pressure)
distends the lung.
The second tree is the pulmonary arterial tree. Beginning with the pul-
monary artery, the tree bifurcates again and again until it form's capillary
blood vessels which separate the alveoli.
The third tree is the venous tree. Beginning with the capillaries, the blood
vessels converge repeatedly until they form pulmonary veins which enter the
left atrium.
Such a sketch of the circulation system cannot give the needed details.
Greater details can be found in Fung (1984) and other references listed therein.

5.3 The Materials of the Circulation System

The heart is a muscle. The lung is blood vessels and airways. All organs are
perfused by blood via blood vessels. Blood vessels consist of smooth muscles,
endothelial cells, and connective tissues. External to blood vessels are body
fluids, cells, and interstitium. The circulation system also includes the lym-
phatic and nervous systems. The blood is a multiphase fluid composed of cells
and plasma. Thus, the variety and complexity of the materials of the circula-
tion system is truly monumental.
A detailed discussion of the chemical composition, molecular and higher
structures (biochemistry, histochemistry), quantitative determination of the
geometrical features of the internal structure of the tissue (morphometry,
stereology, histology, anatomy), and the mechanical properties of the tissue
(biomechanics) and its components (micromechanics) of any ofthe tissues and
organs of the circulatory system would require much space, and is beyond
the scope of this book. The reader is referred to the literature listed in the
Bibliography at the end of this chapter. The author's book Biomechanics:
Mechanical Properties of Living Tissues is a convenient reference. In the
following sections, only the essential data required for immediate discussion
are presented.
It is important to realize that the mechanical properties of many biological
materials are very different from those of familiar engineering materials. For
example, the incremental Young's modulus of the blood vessel wall or the
relaxed muscles vary with the stress acting in the tissue; they do not remain
constant as engineering materials do. For the heart, it is important to know
that the maximum active tensile stress which can be generated in an isometric
contraction of a cardiac muscle varies with the length of the sarcomere. See
the length-tension curve in Fig. 5.3:1. If a heart normally operates at a
sarcomere length marked by the point A in the figure, then when the sarcomere
is lengthened, the maximum muscle tension will increase, and consequently,
the systolic pressure, Pi' will increase. Since the number of sarcomeres in a
heart muscle is fixed, the sarcomere length is proportional to the muscle
length, and by implication, to the radius of the heart. Thus, if the radius of the
5.4 Field Equations and Boundary Conditions 159

B C
toO
E
E 80
.';;
'"
E 60
'-
o
=~ 40
Ii
o
·Vi
s::
Q)
20 slack length
f-
0~__~0~-J______~~__~~____~____~~-= __~
1.0 1.5 2.0 3.0
Sarcomere length (11m)

0.6 0.8 1.0 1.2 1.4 1.6 1.8

Extension ratio, i.

A B C
F=I F=-J F==-J
FIGURE 5.3: 1 The "length-tension" curve of a skeletal muscle. The sarcomere length is
plotted on the abscissa. The maximum tension achieved in isometric contraction at
the length specified is plotted on the ordinate.

heart is increased, the muscle tension will increase, and so will be the systolic
blood pressure. This is known as Starling's law of the heart. This law ceases
to be valid when A moves off the upward-sloping leg of the curve shown in
Fig. 5.3:1.
A similar length-tension curve exists for the vascular smooth muscle. Since
the length of a muscle cell in a tissue depends on the strain at the place where
the muscle cell is located, it becomes clear that the mechanical properties of
the tissues of the circulatory system depend on the strain in the organ.
Another remarkable property of the blood vessels and the heart is the
existence of large residual strains in these organs. Residual strains remain in
an organ when all the extemalloads are removed; e.g., when the transmural
pressure in a blood vessel is reduced to zero. This is discussed in Chapters 11
and 13. See Secs. 11.2 and 13.8.

5.4 Field Equations and Boundary Conditions

The basic equations of biomechanics are the equation of conservation of mass,
the equation of motion, the constitutive equations specifying the mechanical
properties of the materials, and, if heat and transfer and chemical reactions
are involved, the energy equation and reaction rate equation. These and the
160 5 Blood Flow in Heart, Lung, Arteries, and Veins

equations describing the boundary conditions are all that are allowed in a
theoretical analysis of circulation.
The conservation of mass is expressed by the equation of continuity (Sec.
1.7). For a segment of a blood vessel, it says that
The difference of inflow and outflow + the rate of change
of the volume of the segment = O. (1)

The equation of motion is a statement of Newton's law, which takes the

following form when applied to a fluid or solid:
Density x (transient acceleration + convective acceleration)
= - pressure gradient + force due to stress tensor
+ body force per unit volume. (2)

Here density refers to the density of the fluid or solid, the transient acceleration
refers to the rate of change of the local velocity with respect to time, and the
convective acceleration refers to the rate of change of velocity of a material
particle caused by the motion of the particle from one place to another in a
nonuniform velocity field. Pressure gradient is the rate of change of pressure
versus distance. The force due to stress tensor refers to the force per unit
volume due to the rate of change of stresses, taken their directions and areas
into account in a proper way. See Sec. 1.7.
For the blood, the body force consists of inertial forces due to gravitational
acceleration, Coriolis acceleration, and the acceleration of the body due
to walking, jumping, swimming, flying, or other motion; the stress tensor
consists of shear stresses caused by the viscosity of the fluid. For the blood
vessel, the body forces are similar, the stress tensor is mainly caused by
distension of the vessel due to blood pressure.
The constitutive equation of the blood describes the law of viscosity of the
blood, which is, in fact, quite complex (see Fung, 1981). Whole blood is
non-Newtonian, whose viscosity changes with the strain rate.
The constitutive equation of the blood vessel is the stress-strain relation-
ship of the vessel wall material. It is also quite complex because it does not
obey Hooke's law (see Fung, 1981, and Chapters 10 and 11 of this book).
In special situations, it is permissible to use approximate constitutive
equations to simplify the analysis. For example, in large animals such as cat
and man, the shear strain rate of the blood at the walls of the heart and the
pulmonary and systemic arteries and veins exceeds 100 sec- 1 so that in that
region the coefficient of viscosity of blood may be regarded as a constant. The
non-Newtonian feature is important only in a region near the centerline of
the blood vessel. The integrated effect of the non-Newtonian feature is quite
negligible so that flowing blood in these vessels may be treated as Newtonian.
For the blood vessel wall, the stress-strain relationship can be linearized
(into an incremental Hooke's law) if the amplitude of deformation is very
5.5 Blood Flow in Heart and Through Heart Valves 161

small. For pulmonary arteries and veins, the pressure-diameter relationship

has been found to be linear because these vessels are embedded in an elastic
medium-the lung parenchyma (see Fung, 1981, 1984, and Sec. 5.15, Figs.
5.15:4 and 5).
The boundary condition between a viscous fluid and a solid is the no-slip
condition: there is no relative movement of material particles of the fluid at
the boundary and the material particles of the solid at the same interface.
For an ideal fluid whose viscosity is zero, slip must be permitted, the
boundary condition is then reduced to the condition that the materials on the
two sides of an interface must remain contiguous: they must have the same
velocity normal to the interface.
Across any surface on the boundary, the stress vectors on the two sides
of the boundary surface must be equal and opposite by the condition of
equilibrium.
In the analysis of blood flow in any particular blood vessel, one must not
forget the two ends of the vessel. The entry and exit conditions with regard
to pressure and velocity distributions at the ends must be specified.
These are the basic equations and principles. Some special problems are
formulated, solved, and their physiological meaning discussed in the following
sections.

5.5 Blood Flow in Heart and Through Heart Valves

The direction of blood flow in the heart is shown schematically in Fig. 5.2: 1,
the venous blood flows into the right atrium, through the tricuspid valve into
the right ventricle, and then is pumped into the pulmonary artery and the
lung, where the blood is oxygenated. The oxygenated blood then flows from
the pulmonary veins into the left atrium, and through the mitral valve into
the left ventricle, whose contraction pumps the blood into the aorta, and then
to the arteries, aterioles, capillaries, venules, veins, and back to the right
atrium.
An aortic valve with the sinus of Valsalva behind it is sketched in Fig. 5.5:1.
According to model experiments by Bellhouse and Bellhouse (1969,1972) and
Lee and Talbot (1979), the flow issuing from the ventricle, immediately upon
opening of the valve during systole, is split into two streams at each valve
cusp, as shown in the figure. One part of the flow is directed into the sinus
behind the valve cusp, where it forms a vortical flow before reemerging out of
the plane of the figure, to rejoin the main stream in the ascending aorta.
When the aortic pressure rises sufficiently so that deceleration of the flow
occurs, an adverse pressure gradient is produced, P2 at the valve tip exceeds
the pressure Pl at a station upstream. The higher pressure P2 causes a greater
flow into the sinus which carries the cusp toward apposition. The peak
deceleration occurs just before the valve closure. The vortical motion estab-
lished earlier upon the opening of the valve has the merit of preventing the
162 5 Blood Flow in Heart, Lung, Arteries, and Veins

f ---
{~ ",
Top view: \ ~" -+- \
-- I
(~ ,I
......._ _ _ _.....1.....,... . , '

Valve cusp Aortic sinus

I
I

Left
ventricle

-- -
-
--
FIGURE
~

-- ~

5.5:1 Flow pattern within the sinus of Valsalva.

valve cusp from bulging outward to contact the walls of the sinuses. The open
sinus chamber thus can be supplied with fluid to fill the increasing volume
behind the valve cusps as they move toward closure.
Other heart valves and the valves of the veins and lymphatics are operated
by hydrodynamic forces in a similar way, although they do not have sinuses.
In closing these valves, deceleration of the fluid is the essence, not backward
flow.

5.6 Coupling of Left Ventricle to Aorta and Right Ventricle

to Pulmonary Artery

As the heart muscle contracts periodically, blood is pumped from the left
ventricle into the aorta through the aortic valve, and simultaneously from the
right ventricle into the pulmonary artery through the pulmonary valve. The
aorta and the pulmonary artery, being elastic, expand when they receive blood
at a rate faster than the rate at which they send blood away into the peripheral
organs and the lung, respectively. Expanding an elastic vessel causes an
increase of the circumferential strain and stress in the vessel wall. A blood
5.6 Coupling of Left Ventricle to Aorta and Right Ventricle to Pulmonary Artery 163

vessel with an increased circumferential stress in its wall will press harder on
the blood it contains. As a result the blood pressure is increased. The increased
blood pressure in the aorta acts on the aortic side of the aortic valve, tending
to close it. An additional tendency to close the valves comes from the decelera-
tion of the blood in the aorta. The deceleration occurs when the inflow exceeds
the outflow. A consequence of the deceleration is the creation of a longitudinal
pressure gradient through the aortic valve, again tending to close the valve.
Eventually the valve is closed, blood continues to flow from the aorta into the
periphery. By this mechanism the blood flow in the aorta does not have large
swing of pressure as it has in the left ventricle. Similar events occur in the lung.
The process described above can be presented mathematically in various
levels of generality. To be rigorous, it seems evident that the heart, aorta,
arteries, and veins should be represented by three-dimensional network, and
the special geometry and materials of construction of various organs must be
described and incorporated in the mathematical model. In practice it is useful
to consider simplified, crude models first, learn the general features, identify
the important parameters, and then add details when needed. Accordingly,
we shall consider the Windkessel model in this section, and the long wave,
small amplitude pulse waves in the next section. Other features are added in
following sections.
The Windkessel theory is Otto Frank's (1899) interpretation of Stephan
Hale's (1733) explanation of why the pressure fluctuation in the aorta has a
much smaller amplitude than that in the left ventricle. In this theory, the aorta
is represented by an elastic chamber and the peripheral blood vessels are
replaced by a rigid tube of constant resistance. See Fig. 5.6:1. Let Q be the
inflow (cm 3 /sec) into this system from the left ventricle. Part of this inflow is
sent to the peripheral vessels and part of it is used to distend the elastic
chamber. If p is the blood pressure in the elastic chamber (aorta), then the flow
in the peripheral vessel is assumed to be equal to p/R, where R is a constant
called peripheral resistance. For the elastic chamber, its change of volume is
assumed to be proportional to the pressure. The rate of change of the volume
of the elastic chamber with respect to time, t, is therefore proportional to dp/dt.
Let the constant of proportionality be written as C and called compliance.
Then, on equating the inflow to the sum of the rate of change of volume of
the elastic chamber and the outflow p/R, the differential equation governing
the pressure p is
. dp
Q = C dt + p/R. (1)

__ Inflow --.J Elastic \ ' - - - - - ---

---------
__
~ chamber
\------1 Peripheral vessel

FIGURE 5.6:1 The "windkessel" model of the aorta and peripheral circulation.
164 5 Blood Flow in Heart, Lung, Arteries, and Veins

To solve this equation, we can use the method of integration factor. Divid-
ing Eq. (1) by C and multiplying it by e t / RC , we obtain
1 . dp 1 d
_Q(t)e t/RC = _e t/RC + _pe t/RC = _(pe t/RC ). (2)
C dt RC dt
Integrating both sides from t =
we have
° to t and writing the dummy variable as "

p(t)e t/RC = I°c

t 1
Q(,)e</RC d, + Po, (3)

where Po is the value of pat t = 0. If we take the instant of time when the valve
opens as t = 0, then Po is the systolic pressure in the ventricle at the instant
when the valve opens. Multiplying both sides with e- t / RC , we get

p(t) = e-t/(RC) I t1 .
_Q(,)e</(RC) d,
oC
+ poe-t/(RC) (4)

which gives the pressure in the aorta as a function of the left ventricular
ejection history Q(t).
An analog electric circuit can be formulated to represent the differential
equation (Eq. 1). When this electric model is driven by a current I = Q(t) of
the shape of an experimentally determined flow through the aortic value at
the ascending aorta, the voltage V obtained is the analog of the blood pressure
in the aortic arch. On comparing the analog results with an experimentally
determined blood pressure curve, it is found that the actual pressure pulse
deviates from the calculated results in several details: the experimental curve
has a superimposed 3-6 cps oscillation apparent from midsystole throughout
diastole, and a more prominent "incisura" marking aortic valve closure
and a more abrupt rise, often with an "anacrotic" wave. In addition, the
Windkessel model fails to explain the changes of the form of pressure wave
occurring along the arterial network. These limitations of the Windkessel
theory can be alleviated by an improved model such as the one presented in
the next section.
The analysis also applies to the coupling of the right ventricle and pul-
monary artery. The pulmonary circulation, however, is a lower pressure
system. The wall of the right ventricle is thinner than that of the left ventricle;
its systolic pressure is lower. The systolic and diastolic pressures in the
pulmonary artery are much lower than those in the aorta. Since the flows in
the aorta and pulmonary artery are about the same, the shape of p(t) given
by the first term on the right-hand side of Eq. (4) can be similar (except for
the amplitude) only if the values of RC are approximately the same in both
circuits. Hence the low pressure in pulmonary circulation must be achieved
by a lower right ventricular pressure Po, a lower resistance R, and a higher
compliance C of the pulmonary circuit.
The right ventricle and the left ventricle are two pumps working in series.
The flow in them must be matched perfectly, otherwise all the blood would
5.7 Pulsatile Flow in Arteries 165

eventually be accumulated either in the lung or in the periphery. The matching

is stabilized by Starling's law of the heart (Sec. 5.3), namely, if the diastolic
volume is increased, the contracting force of the muscle will increase to pump
harder.

5.7 Pulsatile Flow in Arteries

The weakness of the Windkessel theory is that it allows only one degree of
freedom. The pressures in the aorta and arteries are represented by a single
number. It ignores the change of pressure along the vascular tree. To improve
the understanding of events occurring in the arteries, we go to the next simplest
model: considering each artery as a long, isolated, circular cylindrical elastic
tube, allowing an infinite number of degrees of freedom, approximating the flow
to be one dimensional, and blood as a homogeneous, nonviscous, incompressible
fluid. The flow in each tube is excited at one end by the heart. The excitation
is propagated in the form of elastic waves, much as an earthquake generates
seismic waves. At the distal end each tube bifurcates, and the waves are
partly transmitted to the daughter branches and partly reflected. This theory
was originated by Euler (1775) and Young (1808, 1809), and developed by
many others. It explains many things, but must be supplemented by three-
dimensional theories when one wants to know the velocity profile, flow
separation, stenosis, microcirculation, etc., which are important to the under-
standing of atherosclerosis, hypertension, etc.
To present this theory in the simplest form, it is further assumed that the
wave amplitude is small and the wave length is long compared with the tube
radius, so that the radial and circumferential velocity components are neg-
ligibly small compared with the longitudinal velocity component u(x, t), which
is a function of the axial coordinate x and time t only. Then the basic field
equations (Sec. 3.2) are: the equation of motion,

OU + u OU + ! °Pi = 0 (1)
ot ox p ox
and the equation of continuity,
oA 0
at + ox (uA) = O. (2)

Here A(x, t) is the cross-sectional area of the tube and Pi(X, t) is the pressure
in the tube. The relationship between Pi and A may be quite complex. For
simplicity we introduce another hypothesis, that A depends on the transmural
pressure, Pi - Pe' alone:
Pi - Pe = P(A), (3)
where Pe is the pressure acting on the outside of the tube. Equation (3) is a
gross simplification. In the theory of elastic shells we know that the tube
166 5 Blood Flow in Heart, Lung, Arteries, and Veins

deformation is related to the applied load by a set of partial differential

equations and that the extemalload includes the inertial force of the tube wall.
Hence Eq. (3) implies that the mass of the tube is ignored, and that the
partial differential equations are replaced by an algebraic equation. The
viscoelasticity of tube wall is ignored also.
Equation (1) is the one-dimensional case ofthe Eulerian equation of motion
(Eq. (1.7:1)). Equation (2) can be obtained by integrating Eq. (1.7:5) over a
tube. A special example of Eq. (3) is the pressure-diameter relationship of the
pulmonary artery or vein (Yen et ai., 1980, 1981):
(4)

Here 2a i is the vessel diameter, Pi is the blood pressure, aiO and a are constants
which depend on the pleural pressure PPL and airway pressure PA' but are
independent of blood pressure Pi. a is the compliance constant of the vessel,
and aiQ is the radius when Pi = O.
Let us solve a linearized version of these equations. Consider small distur-
bances in an initially stationary liquid-filled circular cylindrical tube. In this
case u is small and the second term in Eq. (1) can be neglected. Hence

au +! 0Pi = o. (5)
at pox
The area A is equal to nat.Substituting nar
for A in Eq. (2), remembering the
hypothesis that the wave amplitude is much smaller than the wave length, so
that oadox « 1, then, on neglecting small quantities of the second order, we
can reduce Eq. (2) to the form

au + ~ oai = O. (6)
ox ai at
Combining Eqs. (4) and (6), we obtain

au +~ 0Pi = o. (7)
ox ai at

Differentiating Eq. (5) with respect to x and Eq. (7) with respect to t, sub-
tracting the resulting equations, and neglecting the second order term (a/ar)
(oadot) (opdot), we obtain

(8)

where
2 ai
c=- (9)
pa

Equation (8) is the famous wave equation. The quantity c is the wave speed.
5.8 Progressive Waves Superposed on a Steady Flow 167

By direct substitution, one can verify that Eq. (8) is satisfied by the solution
Pi = f(x - ct) + g(x + ct), (10)
where f and g are arbitrary functions of the variables x - ct and x + ct. The
function f(x - ct) represents a wave propagating to the right (increasing x)
whereas g(x + ct) represents a wave propagating to the left.

Velocity, Pressure, and Wall Displacement Waves

The velocity u is linearly related to p through Eqs. (5) and (7), and small change
of the radius a is linearly related to changes in p through Eq. (4). Hence by
eliminating p, it is seen that u and a are governed by the same wave equation
with the same wave speed. If we write
p = Pof(x - ct) + p~g(x + ct),
(11)
u = uof(x - ct) + u~g(x + ct),
then on substituting Eqs. (11) into Eqs. (5) and (7), one sees that the amplitude
Po and U o are related by the simple relationship
Po = pCU o (12)
for a wave that is moving in the positive x direction, and
p~ = -pcu~ (13)
for a wave which moves in the negative x direction.
Equations (12) and (13) show that the amplitude of pressure wave is propor-
tional to the product of wave speed and velocity disturbance and the fluid density.
The pressure and velocity are in phase in an advancing progressive wave; they
are out of phase in the reflected wave.

5.8 Progressive Waves Superposed on a Steady Flow

It can be shown that the equations of Sec. 5.7 are applicable to tubes carrying
a steady flow, provided that we adopt a coordinate system that moves with
the undisturbed flow, and interpret u as the perturbation velocity superposed
on the steady flow and c as the speed of perturbation wave relative to the
undisturbed flow. The proof is as follows.
Let U be the velocity of the undisturbed flow, and u the small perturbation
superposed on it. Treating u as an infinitesimal quantity of the first order, we
see that the equation of motion, Eq. (5.7:1), can be linearized into
au au 1 0Pi
-+U-= ---. (1)
at ax p ax
This can be reduced to Eq. (5.7:6) by introducing a transformation of variables
168 5 Blood Flow in Heart, Lung, Arteries, and Veins

from x, t to x', t':

x' = x - Ut, t' = t. (2)
From Eq. (2) we have
a a at' a ax' a
- = - - + - - = - - U-
a
at at' at ax' at at' ax' ,
(3)
a a at' a ax' a
ax = at' ax + ax' ~ = ax"
Hence, a substitution into Eq. (1) reduces it to
au 1 op
(4)
at' pax' ,
which is exactly Eq. (5.7:5) in the new coordinates.
The equation of continuity, Eq. (5.7:2), now becomes
oa i oa i ai au
-+u-+--=o (5)
at ax 2 ax
when nar is substituted for A, U + u is substituted for u, and the equation is
linearized for small perturbations. Under the transformation Eq. (2), and using
Eq. (3), Eq. (5) becomes

oa '!!:~ _ 0
i
(6)
at' + 2 ax' - ,
which is exactly Eq. (5.7:7).
The pressure-radius relationship, Eq. (5.7:4), is independent of reference
coordinates. Thus all the basic equations are unchanged. But x' and t' are the
distance and time measured in the moving coordinates which translate with
the undisturbed flow. Hence what we set out to prove is done.

5.9 Reflection and Transmission of Waves at Junctions

An arterial tree is composed of segments of cylindrical tubes. Consider a single

junction as shown in Fig. 5.9:1 in which a tube branches into two daughters.
A wave traveling down the parent artery will be partially reflected at the
junction and partially transmitted down the daughters. At the junction, the
conditions are: the pressure is a single-valued function and the flow must be
continuous. Expressing this mathematically: PI denote the oscillatory pressure
associated with the incident wave, PR that associated with the reflected wave,
and PT, and PT2 those associated with the transmitted waves in the two
daughter tubes; then the single-valuedness of pressure means

PI + PR = PT, = PT 2'
(1)
Similarly, let Q denote the volume-flow rate, and let the subscripts, I, R, T1 ,
5.9 Reflection and Transmission of Waves at Junctions 169

FIGURE 5.9:1 A bifurcating artery.

...-.. Incident
~ Reflected

Parent

Tz refer to the various waves as before; then the continuity condition means
(2)
But Q is the product of the cross-sectional area A and the mean velocity u,
which is related to P by Eqs. (12) and (13) of Sec. 5.7. Thus, the flow-pressure
relationship is:
. A
Q = Au = ±-p. (3)
pc
Here p is the density of the blood and c is the wave speed. The + sign applies
if the wave goes in the direction of flow; the - sign applies if the wave
goes the other way.
The quantity pclA is an important characteristic of the artery, and is called
the characteristic impedance of the tube. It is denoted by the symbol Z:

(4)

Equation (3) shows that Z is the ratio of oscillatory pressure to oscillatory

flow when the wave goes in the direction of flow:
P
Z = --;-, ZQ=p, (5)
Q
Z has the physical dimensions [ML -4T- I J, and can be measured in units of
kg m-4 sec-I. With the Z notation, Eq. (2) can be written as

PI - PR = PT 1 + PT 2 •
(6)
Zo ZI Zz
Solving Eqs. (1) and (6) for the p's, we obtain
PR = ZOI - (ZlI + Zil) = qJ
(7)
PI Zo I + (ZI I + Zil)
170 5 Blood Flow in Heart, Lung, Arteries, and Veins

and
PT1 PT2 2Z0l
PI PI Zo 1 + (Zl 1 + Z2 l) = f. (8)
The right-hand sides of Eqs. (7) and (8) shall be denoted by £Jl and ,I,
respectively. Hence the amplitude ofthe reflected pressure wave at the junction
is £Jl times that of the incident wave, the amplitude of the transmitted pressure
waves at the junction is ,I times the incident wave. The amplitude of the
reflected velocity wave is, however, equal to -£Jl times that of the incident
velocity wave, because the wave now moves in the negative x-axis direction,
and according to Eqs. (12) and (13) of Sec. 5.7, there is a sign change in the
relation between u and P depending on whether the waves move in the + or
- x-axis direction.
If the incident wave is
PI = Pof(t - x/co) (9)
and the junction is located at x = 0, so that x is negative in the parent tube
and positive in the daughter tubes, then at the junction x = 0, the pressure is
PI = pof(t). (10)
The reflected and transmitted waves are, therefore,
PR = £JlPof(t + x/co),
PT1 = ,fPof(t - x/c l ), (11)
PT2 = ,IPof(t - x/c 2 )·
Here, co, c l , C2 are the wave speeds in the respective tubes. The wave in the
parent tube is
P = PI + PR = Pof(t - x/co) + £Jlpof(t + x/co)· (12)
. Apo Apo
Q = - f ( t - x/co) - £Jl-f(t + x/co). (13)
pCo pCo
Equations (12) and (13) show that with a reflection, the pressure and flow wave
forms are no longer equal.

5.10 Velocity Profile of a Steady Flow in a Tube

Having analyzed the aortic blood flow by lumped parameter method (Sec.
5.6), and pulse wave in arteries as one-dimensional nonstationary flow of a
nonviscous fluid in an elastic tube (Secs. 5.7-5.9), we shall now consider the
effect of viscosity of blood on the flow. We shall first consider blood as a
Newtonian fluid.
Consider first a steady flow of an incompressible Newtonian fluid in a rigid,
horizontal channel of width 2h between two parallel planes as shown in Fig.
5.10 Velocity Profile of a Steady Flow in a Tube 171
J'

FIGURE 5.10: 1 Laminar flow in y

L
a channel.
u (y)
2h
x

5.10:1. The channel is assumed horizontal so that the gravitational effect (a

body force) may be ignored.
We search for a flow,
u = u(y), v=o, w=o, (1)
which satisfies the no-slip conditions on the boundaries y = ± h:
u(h) = 0, u(-h) = o. (2)
The function u must satisfy the Navier-Stokes equation and the equation of
continuity (Sec. 1.7). It is seen that the equation of continuity is satisfied
exactly. The Navier-Stokes equation is simplified to:
ap d2 u
0= --+11-- (3a)
ax dy2'

0= ap (3b)
ay'

0= ap (3c)
az·
Equations (3b) and (3c) show that p is a function of x only. If we differentiate
Eq. (3a) with respect to x and use Eq. (1), we obtain a2p/ax 2 = O. Hence ap/ax
must be a constant. Equation (3a) then becomes
d 2u 1 dp
(4)
dy2 J1. dx'
which has a solution
1 y2 dp
u= A +By +---. (5)
11- 2 dx
The two constants A and B can be determined by the boundary conditions,
Eq. (2), to yield the final solution
1 2 2 dp
u = --(h - y )-. (6)
211- dx
Thus, the velocity profile is a parabola.
172 5 Blood Flow in Heart, Lung, Arteries, and Veins

Next consider a flow through a horizontal circular cylindrical tube of radius

a. Using polar coordinates, it is easy to show that the solution is [see Fung
(1984), p. 83]
1 2 2 dp
u = --(a - r )-. (7)
411 dx
This is the famous parabolic velocity profile of the H agen-Poiseuille flow.
From the solution (7) we obtain the rate of flow through the tube by
integration:
.
Q = 2n
fa urdr = - -na- d
dp
4
. (8)
o 811 X

This classical solution has been subjected to innumerable experimental

validation. It was found to be invalid near the entrance to a tube. It is
satisfactory at a sufficiently large distance from the entrance but is again
invalid if the tube is too large or too long if the velocity is too high. The
difficulty at the entry region is due to the transitional nature of the flow in
that region so that our assumption v = 0, w = 0, is not valid. The difficulty
with too large a Reynolds number, however, is of a different kind: the flow
becomes turbulent.
Osborne Reynolds demonstrated the transition to turbulent flow in a
classical experiment in which he examined an outlet from a large water tank
through a small tube. At the end of the tube there was a stopcock used to vary
the speed of water through the tube. The junction of the tube with the tank
was nicely rounded, and a filament of colored fluid was introduced at the
mouth. When the speed of water was slow, the filament remained distinct
through the entire length of the tube. When the speed was increased, the
filament broke up at a given point and diffused throughout the cross-section.
Reynolds identified the governing parameter urnd/v-the Reynolds number-
where Urn is the mean velocity, d is the diameter, and v is the kinematic viscosity.
The point at which the color diffuses throughout the tube is the transition
point from laminar to turbulent flow in the tube. Reynolds found that transi-
tion occurred at Reynolds numbers between 2,000 and 13,000, depending on
the smoothness of the entry conditions. When extreme care is taken, the
transition can be delayed to Reynolds numbers as high as 40,000. On the other
hand, a value of 2,000 appears to be about the lowest value obtainable on a
rough entrance. Turbulence is one of the most important and difficult prob-
lems in fluid mechanics.

5.11 Steady Laminar Flow in an Elastic Tube

If the tube is elastic (Fig. 5.11:1), then the high-pressure end would distend
more than the low-pressure end. The diameter of the tube is, therefore,
nonuniform (if it were uniform originally) and the degree of nonuniformity
depends on the flow rate.
5.11 Steady Laminar Flow in an Elastic Tube 173

II nitial External pressure = 0

I
P = 0 no flow

L
I
With flow
External pressure = 0

L
Po

FIGURE 5.11:1 Flow in an elastic tube oflength L.

Vessel as an
elastic body

Elastic
deformation

Flow Vessel as Pressure

fluid conduit

FIGURE 5.11:2 A hemoelastic system analyzed as a feedback system of two functional

units: an elastic body, and a fluid mechanism.

If we wish to determine the pressure-flow relationship for such a system,

we may break down the problem into two familiar components. This is
illustrated in Fig. 5.11:2. In the lower block, we regard the vessel as a rigid
conduit with a specified wall shape. For a given flow, we compute the pressure
distribution. This pressure distribution is then applied as loading on the elastic
tube, represented by the upper block. We then analyze the deformation of the
elastic tube in the usual manner of the theory of elasticity. The result of the
calculation is then used to determine the boundary shape of the hydrodynamic
174 5 Blood Flow in Heart, Lung, Arteries, and Veins

problem of the lower block. When a consistent solution is obtained, the

pressure distribution corresponding to a given flow is determined.
In Sec. S.10 we derived Poiseuille's formula under the assumption of a
laminar steady flow in a circular cylindrical tube of constant radius. If the
radius a(x), as a function of the axial coordinate x, is not a constant, but the
slope da/dx is small, then the fluid dynamic problem can be solved by pertur-
bation method as a power series of the small parameter da/dx. Under an
additional assumption that the Reynolds number is so small that the inertial
force term pu au/ax is negligible in the zeroth order equations, the solution is
the Poiseuille's formula

(1)

in which Q is the volume-flow rate and is a constant for the whole tube, a(x)
is the local radius, dp/dx is the local pressure gradient. For an elastic tube, the
radius a is a function of pressure. Hence we can rewrite Eq. (1) as
dp 81l .
a4 (p)-d = --Q = const. (2)
x n

This is very easy to integrate if the function a(p) is known. For the pulmonary
arteries and veins it is known that the pressure-radius relationship is given
by Eq. (S.7:4)

(3)

where ao is the radius when p is zero, and a is the compliance constant.

Substituting Eq. (3) into Eq. (2), and integrating, we obtain
4dpda 24da 81l .
a --=-a - = - - Q. (4)
da dx a dx n

Since the right-hand side term is a constant independent of x, we obtain the

integrated result
5 20lla .
[a (x)] = ---Qx + const. (S)
n
The integration constant can be determined by the boundary condition
that when x = 0, a(x) = a(O). Hence the constant = [a(O)J5. Then by putting
x = L, we obtain from Eqs. (S) and (3) the elegant result (Fung, 1984)

Q= 20 n {[a(0)]5 - [a(L)]5}
llaL
(6)

= 20;aL [ ( ao + ~ Po Y- ~
(a o + PL YJ
5.11 Steady Laminar Flow in an Elastic Tube 175

Thus the flow varies with the difference of the fifth power of the tube radius
at the entry section (x = 0) minus that at the exit section (x = L). If the ratio
a(L)/a(O) is t, then [a(L)]S is only about 3% of [a(O)]S and is negligible by
comparison. Hence when a(L) is one-half of a(O) or smaller, the flow varies
directly with the fifth power of the tube radius at the entry, whereas the radius
(and the pressure) at the exit section has little effect on the flow.
This analysis applies very well to the lung, in which the phenomenon just
described has an important effect. See Sec. 6.8 infra.
For flow in large blood vessels with Reynolds number much greater than
1, the inertial force terms must be added. Let us consider the case of a steady
flow of Newtonian fluid in an elastic tube whi which is initially a circular
cylinder. Assume that the flow is predominantly one-dimensional. Let q
represent the average velocity in the tube. The convective inertial force is
pq(oq/ox). The pressure drop due to blood viscosity is given by Eq. (1) even
if the flow is turbulent, provided that the coefficient of viscosity J1. is reinter-
preted as the "apparent" coefficient of viscosity which is a function of the
Reynolds number, see Sec. 5.13 infra. Then the equation of motion is
dq dp 8J1..
pq dx = - dx - na4 Q. (7)

Here p is the density of the blood, x is the axial coordinate, p is the pressure,
Q is the volume flow rate, and J1. is the apparent coefficient of viscosity of the
blood corrected for turbulence, secondary flow, or entrance effect, i.e. it is a
function of Reynolds number. Finally, a is the radius of the tube, which is a
linear function of pressure as given by Eq. (3). When the transmural pressure
is zero the tube is assumed to be cylindrical, ao = const. The equation of
continuity is
na 2 q = const = Q. (8)
By differentiation, one obtains
a 2dq + 2qada = O. (9)
On solving Eq. (8) for q, and substituting into Eq. (9) multiplied by q/a 2 , we
have

(10)

Substituting Eq. (10) into Eq. (7) and reducing, one obtains

( a4 _ p(X.Q2) da = _ 4J1.(X. Q. (11)

n 2 a dx n

Integration yields
5p(X.Q2 20J1.(X. .
as - - - 2 - lna = ---Qx + const. (12)
n n
176 5 Blood Flow in Heart, Lung, Arteries, and Veins

The boundary condition a = a(O) when x = 0 yields the integration constant.

On putting this constant into Eq. (12), and then letting x = L where a = a(L),
we obtain

. [p
Q-
a(L)] '2 {5
11:
4Jl1l:L In a(O) Q = 20Jla.L [a(O)] - [a(L)]
5}
. (13)

This is a modification of Eq. (6). The effect of inertial force is embodied in the
second term. If the elastic deformation is small, a(L) == a(O), then the second
term tends to zero, and the solution is the same as Eq. (6) except that the
apparent viscosity Jl must now be considered as a function of the Reynolds
number. If the elastic deformation is significant, in that a(L) differs consider-
ably from a(O), then the second term must be considered. For a given set of
values a(L), a(O), we now have two solutions of Q. Conversely, for a given Q
we now have multiple solutions of a(L), a(O). This is possible at high Reynolds
number, because the inertial force and the viscous dissipation influence the
pressure gradient in opposite ways.

5.12 Velocity Profile of Pulsatile Flow

To obtain the velocity profile of nonstationary flow in a blood vessel, one

must solve the equations of motion and continuity of both the blood and the
blood vessel wall, and boundary conditions that match the displacements,
velocities, and stresses. The calculation is usually lengthy. References to the
literature can be found in Fung (1984), McDonald (1974), Patel and Vaishnav
(1980), Pedley (1980). In the following, a simple example is given.
Assume that the fluid is homogeneous, incompressible, and Newtonian; the
vessel wall is rigid, circular, and cylindrical; the motion is laminar, axi-
symmetric, and parallel to the longitudinal axis of the tube. A pressure
gradient drives the flow, the vessel is horizontal, and gravitation has no effect
on the flow. Then the field equations are the Navier-Stokes equations, and
the equation of continuity. They are simplified to the following under the
conditions named above:

0= _ap (1)
ar'
ap
0= -ae' (2)

pau = _ ap + Jl (a 2u + ! au), (3)

at ax ar2 r ar
au =0.
ax (4)
5.12 Velocity Profile of Pulsatile Flow 177

The boundary conditions are the axisymmetry condition at the center and
no-slip on the wall, at radius a:

au = 0 when r = 0, (5)
ar
u = 0 when r = a. (6)
Here p stands for pressure; (x, r, 0) are cylindrical polar coordinates with x in
the axial direction; u is the velocity component in the direction of x; t is time.
According to Eqs. (1) and (2), p is a function of x and t only. According to (4),
u is a function ofr and t. On differentiating Eq. (3) with respect to'x, one obtains

:x(::) = O. (7)

This shows that the pressure gradient must not vary with x. It can be a function
of t. For a general periodic motion, one can write
ap
- = IN Cneinrot. (8)
ax n=O
On substituting into Eq. (3), one obtains

au
= - IN Cne lnrot + Jl (a- 2u + -1 -au) .
2
.
p- (9)
at n=O ar r ar
The term n = 0 corresponds to a steady pressure gradient investigated in Sec.
5.10; the solution is given by Eq. (5.10:7). To the other terms in (7), we can try
u(r, t) in the form
N
U = I Vn(r)einrot (10)
n=O
which is periodic. Substituting Eq. (10) into Eq. (9) we see that the resulting
equation is satisfied if we set

. = - Cn + Jl (d- Vn + - -n)
1 dV 2
lpnwv
n dr-
2 r dr .
(11)

The boundary condition, Eq. (6), is satisfied if

Vn = 0 when r = a,
(12)
aVn =0 when r = O.
ar
The general solution of Eq. (11) is

(13)
178 5 Blood Flow in Heart, Lung, Arteries, and Veins

Here Jo(kr) is the Bessel function of the first kind of order zero of kr, Yo(kr) is
the Bessel function of the second kind of order zero of kr, k being a constant.
An, Bn are arbitrary constants, and IX is a dimensionless quantity known as the
Womersley number (Sec. 5.13):

IX = aft. (14)

To determine An, Bn, the boundary conditions, Eq. (12), are used. As r
approaches zero, the derivative Jb approaches zero and Y~ approaches infinity.
Hence Bn must vanish, and the first of Eq. (12) requires

AnJo(lXnl/2i3/2) + iCn = o. (15)

pnw
Solving this equation for An and substituting into Eq. (13) together with
Bn = 0, we obtain
_ iCn [ _ JO(IXf,nl/2i3/2)]
vn(r) - pnw 1 Jo(lXnl/2i3/2)· (16)

The problem is solved by substituting Eq. (16) into Eq. (10). The velocity
profile depends on Womersley number IX. An illustration is given in Fig. 5.12:1.

0: = 3.34 0: = 4.72 0: = 5.78 0: = 6.67

180'
180'
165'
165'
150' 150'
135' 135'
120' 120'
105' 105'
90' 90'
75' 75'
60' 60'
45'
30'
15'
0'

Fractional Radial Position

FIGURE 5.12:1 Theoretical velocity profiles of a sinusoidally oscillating flow in a pipe,

with pressure gradient varying like cos cot. IX is the Womersley number. Profiles are
plotted for phase angle steps of I1cot = 15°. For cot> 180°, the velocity profiles are of
the same form but opposite in sign. Reproduced with permission from D.A. McDonald,
Blood Flow in Arteries, copyright © 1974, the Williams & Wilkins Co., Baltimore.
5.13 The Reynolds Number, Stokes Number, and Womersley Number 179

5.13 The Reynolds Number, Stokes Number, and

Womersley Number
The general equations of hemodynamics appear formidable. Some essential
features can be identified when different terms are compared. The Navier-
Stokes equation

aUi + P (aui
P- u l -a + u2-aaUi + u3-aaUi)
at Xl X2 X3

(1)

represents the balance of four kinds of forces. Term by term, they are
transient convective body pressure viscous
inertial + inertial = force + force + force .
force force
Not all the forces are important all the time. In a steady flow the transient
inertial force vanishes. In an ideal fluid the viscous force vanishes. In hydro-
static equilibrium all but the body and pressure forces vanish. Simplifications
are recognized for these cases.
Compare the transient inertial force term with the viscous force term.
To make an estimate, let U be a characteristic velocity, w a characteristic
frequency, and L a characteristic length. Then the first term in Eq. (1) is ofthe
order of magnitude pwU, and the last term is ofthe order of magnitude j1UL -2.
The ratio is
transient inertial force pwU pwL 2 WL2
(2)
viscous force j1UL -2 j1 V

This is a dimensionless number. If it is large, the transient inertial force

dominates. If it is small, the viscous force dominates.
The dimensionless number wL 2 Iv is a frequency parameter, and is called
the Stokes' number because its significance was pointed out by George Stokes
in 1840. It is better known by its square root,

N LJ(!fj,
w = (3)

which is called Womersley number in honor of J.R. Womersley, who made

extensive calculations on pulsatile blood flow in the 1950's. If L is taken to be
the radius of the blood vessel, then Womersley's number is often written as IX:

(4)

D being the blood vessel diameter. In large arteries of all but the smallest
mammals, the value of IX, calculated from the circular frequency of the heart-
180 5 Blood Flow in Heart, Lung, Arteries, and Veins

beat in rad/sec, is considerably larger than 1. For example, a typical value of

IXin the aorta of man is 20, in a dog it is 14, in a cat 8, and in a rat 3. Hence
in these aortas the inertial force dominates in pulsatile flow.
If IX is large, the effect of the viscosity of the fluid does not propagate very
far from the wall. In the central portion of the tube the transient flow is
determined by the balance of the inertial forces and pressure forces (first and
fourth terms in Eq. 1), and the elastic forces in the wall (through the boundary
conditions), as if the fluid were nonviscous. We, therefore, expect that when
the W omersley number is large, the velocity profile in a pulsatile flow will be
relatively blunt, in contrast to the parabolic profile of the Poiseuillean flow,
which is determined by the balance of viscous and pressure forces. This is
shown in Fig. 5.12:1.
Now compare the convective inertial force term with the viscous force term.
With characteristic velocity U and characteristic length L, the order of magni-
tude of the inertial force is p U 2 , that of the viscous force is J1U / L. The ratio is
inertial force pU 2 pUL ld b
. = -- = -- = Reyno s num er. (5)
VISCOUS force J1U jL J1
A large Reynolds number signals a preponderant inertial effect. A small
Reynolds number signals a predominant viscous force effect. In aorta of man
the Reynolds number based on vessel diameter can be 2,000-3,000, large
enough to cause possible turbulence (Sec. 5.10). In the capillary blood vessels,
the Reynolds number is in the order of 0.001 to 0.01, so small that it suggests
complete insignificance of the inertial force.
The occurrence of turbulence in a pulsatile flow in the aorta could be
transient. Even when the condition of flow favors the transition of a laminar
flow into turbulent, the actual transition into turbulence would require a
certain amount of time to develop. Hence if the flow velocity fluctuates too
fast, the turbulence may not develop. Similarly, if a flow is turbulent but the
condition has changed to favor a transition into laminar flow, the actual
transition may lag behind for a while.
Quantitative studies of the laminar-turbulent transition may seek to
express the critical Reynolds number as a function of the Womersley number.
Experimental results can be plotted as shown in Fig. 5.13:1. The ordinate is
the peak Reynolds number. The stippled area indicates the conditions under
which the flow is stable and laminar.
In the experiments whose results are shown in Fig. 5.13:1, the wide varia-
tions of velocity and heart rate were obtained with drugs and nervous stimuli
in anesthetized dogs. In normal, conscious, free-ranging dogs the peak
Reynolds number usually lies in an area high above the stippled area of Fig.
5.13:1. This suggests that some turbulence is generally tolerated in decelera-
tion of systolic flow in the dog.
Turbulence in blood flow implies fluctuating pressure acting on the arterial
wall, and fluctuating, increased shear stress. These stresses are implicated in
murmurs, post-stenotic dilation, and atherogenesis.
5.14 Equation of Balance of Energy and Work 181

5000

()
.... 4000
Q)
..c
E
:::l
; 3000
"'0
0
"0
c: 0
>
& 2000
x
C\)
Q)
0 0
a.
00
1000 0

o I'-------'-,_---L-------L......J
5 10
,
15
,
20
Womersl ey number N w

FIGURE 5.13: 1 The stability of blood flow in the descending aorta of anesthetized dogs
is influenced by the peak Reynolds number and the Womersley number. Points joined
by the lines refer to the same animal. Open circles: laminar flow; filled circles: turbulent
flow; half-filled circles: transiently turbulent flow. From Nerem, R.M., and Seed, W.A.
(1972), by permission.

5.14 Equation of Balance of Energy and Work

According to the principle of conservation of energy, the rate of gain of energy

of a material system (the sum of the kinetic, potential and internal energies)
must be equal to the sum of the rate at which work is done on the system and
heat transported in. Apply this principle to a body of blood contained in a
blood vessel between two arbitrary cross sections, 1 and 2, perpendicular to
the vessel axis, as illustrated in Fig. 5.14:1. Let p denote the pressure, u denote
the axial component of the velocity of flow, q denote the magnitude of the
velocity vector, Q denote the volume rate of flow . Let dA denote a small
element of area in a cross section. At the left end, section 1, the outward normal
vector of the cross section points to the left, the force acting on the area dA
due to the pressure, pdA, points to the right. The positive direction ofthe axial
velocity u point to the right. The rate of work done by the force due to pressure
is pudA. The total work done by the force over the entire cross section is,
therefore

fA,
pudA (1)

where the integral is taken over the area Al of the cross section No. 1.
182 5 Blood Flow in Heart, Lung, Arteries, and Veins

1 ·. ........ ................................... ....... 2I

~..........
~..
~
.......
Ut /:) t,
~
\~
....................·..·..·.... ·....·..·.... """""S$"hi.,,~ I~) U 2 At
..~--..
'I- ., ..•.•....
\ :
t I
. 2·
1 .
~'-- ""...................
.ifi'e............
..
~
. . £.....•..........
::""""""" .........
' ..: ;j$
.. ........... ::::.::.::::;..........................

........... ~:~'\
!

. 'r--~

FIGURE 5.14:1 Two arbitrary cross sections, 1 and 2, of a blood vessel. At an instant
of time t, a control volume of blood is bounded by the plane sections 1 and 2 and the
wall of the blood vessel shown by solid lines. An infinitesimal time /!it later, the
boundary of the control volume becomes that shown by the dotted line: consisting of
two paraboloidal surfaces at sections 1 and 2, and a distended blood vessel wall. The
equations of motion, continuity, and energy are written for the fluid in the control
volume. The symbols U 1 and U2 represent axial component of the velocity at stations
1 and 2, respectively.

The kinetic energy per unit volume of blood is tpq2 where p is the density
of the blood. The total kinetic energy of the blood contained in the volume
between the sections 1 and 2 is, at time t,

(2)

A short time At later, the same body of fluid would occupy a slightly different
volume which is bounded by the dotted lines shown in Fig. 5.14.1. The side
wall distends a little because of vessel wall elasticity. The fluid particles
composing the cross section 1 are displaced by a distance uAt to the right.
The plane cross section No. 1 becomes curved and bulges to the right. The
particles at Section 2 are also displaced to the right by the amount udt. During
the time interval At, therefore, the total kinetic energy of the blood is changed
by the amount

Where V' is the volume bounded by the dotted lines. The rate of change of
kinetic energy is obtained by dividing the quantity above with dt. The - and
+ signs in the expression (3) should be carefully noted.
A similar consideration should be given to the work done by force imposed
on the blood by the blood vessel wall, the potential energy change due to
5.15 Systemic Blood Pressure 183

gravitation, the internal energy change due to temperature change, the heat
transported through the boundary, and the rate of heat generation due to
internal friction equal to the sum of all the products of stress components and
the corresponding strain rates. The last term, the heat dissipation, is denoted
by~:

(4)

Where (iij is the stress tensor and V;j is the strain rate tensor.
Now, on equating the change of energy with the work done, and dividing
through by the volume flow rate Q:

Q= L udA (5)

we obtain, if gravitational effect and heat transfer were negligible, the following
equation:

/'..
Pl - P2
/'.. 1 '1'
= "2Pq2 -
1 '1'
"2PQ1 + pgh2 - pghl
~ 1 a (1r
+ Q + Q Jv at "2 pQ
2) dv. (6)

Here the velocity-weighted pressure p and the square of velocity Q2 are

.0-

defined by dividing Eqs. (1) and (2) by Q:

/'P.. = --;-If pudA, Q2

.0- If
= --;- Q2 u dA. (7)
Q A Q A

Note thatpand tp;f' have the dimensions of pressure.

The energy equation (6) was given by Pedley et al. (1977) and derived in
full detail in Fung (1984), pp. 15-20. It is used frequently in this book.

5.15 Systemic Blood Pressure

If we apply the results derived in the preceding sections to a circuit of blood

vessels beginning at the aortic valve and ending in the right atrium, take the
average of the pressure-flow relationship of every segment over a period of
time which is long compared with a single heart beat, and synthesize the
segments into a whole circuit, then we obtain the result:
Average pressure at aortic valve - average pressure
at right atrium = integrated frictional loss. (1)
This is often written as:
Systemic arterial pressure = flow x resistance. (2)
Here the systemic arterial pressure is the difference between the pressure at
the aortic valve and that at the vena cava at the right atrium, the flow is the
184 5 Blood Flow in Heart, Lung, Arteries, and Veins

cardiac output, and the resistance is the total peripheral vascular resistance.
Hence, writing in greater detail, we have
Pressure at aortic valve - pressure at right atrium
= (cardiac output) x (total peripheral vascular resistance), (3)
where
. integrated frictional loss
Total peripheral vascular resIstance = d' (4)
car lac output
The last term in Eq. (3) represents the sum of the pressure drops due to the
friction loss along all segments of blood vessels. Since there are millions of
capillary blood vessels in the body, there are millions of pathways along which
one can integrate the equation of motion to obtain Eq. (3), so the final result
Eq. (3) is useful only if the pressures at the aortic valve and right atrium are
uniform no matter which path of integration is used. Fortunately, this is the
case.
The integrated frictional loss is the sum of frictional losses in all segments
of vessels of the circuit. To compute the frictional loss of a segment, let us first
consider a steady laminar flow (i.e., one that is not turbulent) in a long, rigid,
circular, cylindrical vessel. To such a flow, Poiseuille's formula, Eq. (5.10:8)
applies. Let the vessel length be L and the vessel diameter be d, then
Q= _ nd 4 Il.p (5)
128 pL'
Here Jl is the coefficient of viscosity of the fluid, and Il.p is the pressure drop.
Equation (5) can be written as
Il.p = (laminar resistance in a tube) x (flow in the tube), (6)
from which we obtain the resistance of a steady laminar flow in a circular
cylindrical vessel:
·
Iammar . . b 128JlL (7)
resIstance m a tu e = ~.

If the nth generation of a vascular tree consists of N identical vessels in

parallel, then the
Pressure drop in the nth generation of vessels
= (resistance in N parallel tubes) x (total flow in N tubes)
(resistance in one tube) ( d ' )
= N x car lac output. (8)

Note that according to Eq. (7) the laminar flow resistance is proportional to
the coefficient of viscosity Jl and the length of the vessel L, and inversely
proportional to the fourth power of the diameter d. Obviously the vessel
5.16 The Veins and Their Collapsibility 185

diameter d is the most effective parameter to control the resistance. A reduc-

tion of diameter by a factor of 2 raises the resistance 16-fold, and hence leads
to a 16-fold pressure loss. In peripheral circulation the arterioles are muscular
and they control the blood flow distribution by changing the vessel diameters
through contraction or relaxation of the vascular smooth muscles.
Equation (7) gives the resistance to a Poiseuillean flow in a pipe and for a
given flow rate Qis the minimum of resistance of all possible flows in the pipe.
If the flow becomes turbulent, the resistance increases. If the blood vessel
bifurcates, the local disturbance at the bifurcation region raises resistance. In
these deviations from the Poiseuillean flow the governing parameter is the
Reynolds number. If a flow is turbulent, then
Resistance of a turbulent flow in a vessel
= (laminar resistance)' (0.005 Nj/4). (9)
Thus, if the Reynolds number is 3,000, the resistance of a turbulent flow is
over two times that of the laminar resistance. In the ascending and descending
aorta of man and dog the peak Reynolds number does exceed 3,000. The
energy loss that occurs at points of bifurcation, entry flow, flow separation,
etc., are also functions of Reynolds number. In these cases one writes the
pressure-flow relationship as
. 1 nd 4 Ap
(10)
Q = Z(NR ) 128 flL'
where Z(NR ) is a dimensionless function of the Reynolds number. Equation
(9) shows that for a turbulent flow Z(NR ) is equal to 0.005 Njf4. Other examples
are given in Sec. 7.2, especially Eqs. (7.2:4)-(7.2:6). All the results of fluid
mechanics research on flow resistance in pipes can be packed into the function
Z(NR)'
Equation (3) or (8) shows the basic factors that control the systemic blood
pressure. The resistance is proportional to the blood viscosity. Hence lowering
the coefficient of viscosity will promote the flow. Hemodilution is thus a
practical clinical tool. The resistance is sensitive to the diameter of the blood
vessel. The diameter is controlled by the vascular smooth muscle. Hence the
control of smooth muscle behavior is the key to the treatment of hypertension.

5.16 The Veins and Their Collapsibility

Veins are similar to arteries in size and construction, but veins have valves
and smaller wall thickness to diameter ratio. In fact the wall thickness of veins
is often quite nonuniform around the circumference. Because veins have
thinner walls, they are more compliant than the arteries. Because the blood
pressure is low in the veins, they are more sensitive to external pressure. If the
external pressure exceeds the internal pressure by an amount known as the
critical buckling pressure, then a vein will collapse. Normally, 80% of a man's
186 5 Blood Flow in Heart, Lung, Arteries, and Veins

blood is in the veins. For this reason veins are said to be capacitance vessels.
The capacitance is sensitive to internal and external pressures. Thus raising
one's leg or moving the leg muscles will reduce the blood volume in the legs,
pushing blood to the heart and circulating to other parts of the body.
The collapsibility of the vein gives the venous blood flow some unique
features. In dynamic condition the transmural pressure (Ap = internal -
external pressures) acting on a vein may be a) positive throughout, b) negative
and exceeding the critical buckling pressure throughout, or c) positive at the
entry section, but negative and exceeding the critical buckling pressure at the
exit section. Then in condition a) the vein is patent, in b) it is collapsed, whereas
in c) something special will happen. If in condition c) the exit end is collapsed,
then the flow would stop, the pressure drop would become zero, the whole
tube would have a Ap equal to that of the entry section, the condition of a)
would then prevail, and flow would start again. But if flow starts, the pressure
will drop along the tube, and the exit section may be choked again. This may
lead to a dynamic phenomenon of "flutter,", or to a limiting steady flow
controlled by a narrowed section. In the last case, the actual value of Ap at
the exit section is quite immaterial as long as the cross sectional area at the
exit section is much reduced. An analogy may be drawn between this and the
waterfall in our landscape, or sluicing in industry or flood control: The volume
flow rate in a waterfall depends on the conditions at the top of the fall, and is
independent of how high the drop is. Thus, the phenomenon of flow in case c)
is described as the "waterfall" phenomenon, or as sluicing.
The waterfall phenomenon occurs in a number of important organs: the
lung, the vena cava, etc. It occurs in thoracic arteries during resuscitation
maneuvers, and in brachial arteries while measuring blood pressure by cuff
and Korotkov sound. The same phenomenon also occurs in male and female
urethra in micturition, and in manmade instruments such as the blood pump
and the heart-lung machine.
Since so much depends on the collapsibility, let us consider the mechanical
property of blood vessels at negative transmural pressure in greater detail.
Moreno et al. (1970) measured the change of the cross-sectional area of
dog's vena cava when the transmural pressure was varied. Shapiro (1977)
measured the same in latex tubing. The characteristics of the vessel and tube
deformations are similar. Shapiro's results are shown by the solid curve in
Fig. 5.16:1. If the tube were circular cylindrical and of Hooke an elastic material
when the transmural pressure is zero, then the elastic stability of the tube is
amenable to mathematical analysis. The theoretical results of Flaherty et al.
(1972) are shown in Fig. 5.16:1 by the curve with long dashes. Theoretically,
the pressure-area curve has a sudden change of slope at each critical
transmural pressure. The deformation pattern changes when the transmural
pressure exceeds the critical value. If one defines the dimensionless variables

(1)
5.16 The Veins and Their Collapsibility 187

~
• a

il 0.5
a" A/Ao
(a ~ I)
a

- 5

-10
~I a.
!:,.lI:

I~
II .
- 15

(a:s a)
- 20

-25 EXPERIMENTAL
- cPa a -3/2
- (P = a- 3/2 _1

-30

-co
FIGURE 5.16:1 Behavior of a collapsible tube. Dimensionless transmural pressure
difference, p, versus dimensionless area ratio, ex. Solid curve shows a typical experi-
mental curve for thin-walled latex tube, and adjacent to it, typical cross-sectional
shapes for the different ranges of ex. Dot-dash curve represents Eq. (8), coincides with
solid curve for ex < a. Dashed curve represents Eq. (10). Curve with long dashes
represents the theoretical result given by Flaherty et al. for cylinders whose cross
sections are perfectly circular when p = O. Point contact occurs at ex = &, and line
contact occurs at ex = a. From Shapiro (1977), by permission.

in which p represents internal pressure, Pe is the external pressure, E is the

Young's modulus of the tube wall material, v is its Poisson's ratio, R is the
tube radius at midwall, h is the tube wall thickness, A is the cross-sectional
area of the lumen, then Flaherty et al. showed that the buckling occurs when
p < - 3. When p = 5.247, the opposite walls touch each other at the midpoint.
Upon further increase in external pressure, the contact area increases and the
open portion of the cross section is reduced in size but remains similar in
shape. For this "self-similar" type of deformation Flaherty et al. obtained the
relationship
(2)
188 5 Blood Flow in Heart, Lung, Arteries, and Veins

Noticing the difference between the experimental curve and the theoretical
curve, Shapiro (1977) proposed an empirical formula
-jj = rx- 3/ 2 - 1. (3)
Now let us see something different. Figure 6.6:5 on p. 209 shows the
thickness of the pulmonary capillary sheet in the interalveolar septa as a
function of the transmural pressure. These pulmonary capillary blood vessels
form a dense network whose "thickness" varies with the blood pressure,
whereas the dimension in the plane of the interalveolar septa is unaffected by

FIGURE 5.16:2 The connection between a pulmonary vein and interalveolar septa in
eat's lung. Courtesy of Dr. Sidney Sobin.
5.16 The Veins and Their Collapsibility 189

the blood pressure. The difference between the blood pressure and the airway
pressure is defined as the transmural pressure. Fung et al. (1972) have shown
that the thickness drops very rapidly to zero when the transmural pressure
changes from positive to negative values.
On the other hand, Fig. 6.6:4 on p. 208 shows the diameter versus trans-
mural pressure relationship of pulmonary veins (Yen and Fappiano, 1981).
It is seen that the relationship can be represented by straight lines. The slope
of the straight line does not change when Ap changes from the positive to
negative value. These veins would not collapse under negative transmural
pressure in the physiological range (Fung et aI., 1983).
Thus the elastic stability characteristics of the pulmonary capillaries is
similar to that of the vena cava, but that of the pulmonary vein is entirely
different from that of the vena cava. Not all veins are alike! The difference is
actually easily explained. The vena cava was tested as an isolated tube.
The pulmonary veins were, however, tested intact, embedded in the lung
parenchyma which was in tension. The lung parenchyma provides an elastic
support to the pulmonary veins.

FIGURE 5.16:3 A photo micrograph of cat lung showing a venule tethered by three
interalveolar septa. Vasculature perfused with a catalyzed silicone elastomer and
hardened; gelatin-embedded; cresyl violet stained. From Fung et al. (1983). Reproduced
by permission.
190 5 Blood Flow in Heart, Lung, Arteries, and Veins

This elastic support can be seen from the photomicrographs of the lung
parenchyma. In Figure 5.16:2 is shown a histological section of a cat lung
parenchyma containing a large blood vessel. The lacy tissue tethering the
outer wall of the blood vessel is the alveolar structure, which is in tension in
an inflated lung. Thus the blood vessel is embedded in a foam-rubber-like
material. In Figure 5.16:3, at a larger magnification, is shown a histological
section of cat lung parenchyma containing a pulmonary arteriole. The diameter
of the arteriole is about 25 /lm, which is small compared with the dimension
of the alveoli of the cat, about 10 /lm. See that the arteriole is pulled by three
interalveolar septa. These septa are in tension in an inflated lung, they tend
to distend the arteriole. When the alveolar gas pressure outside the arteriole
exceeds the blood pressure in the arteriole, the tendency for the vessel to
collapse is resisted by the tension in the interalveolar septa attached to the
outer wall of the blood vessel.

5.17 Flow in Collapsible Tubes

In circulatory physiology, flow in blood vessels in collapsible condition may
occur either in microcirculation, or in large vessels. In microcirculation, a
common example is the capillary blood flow in the lung. The pulmonary
capillaries are readily collapsible (see Fig. 6.6:5) and waterfall phenomenon
occurs in them. This will be discussed in Sec. 6.8.
Waterfall phenomenon occurs in large veins for a different reason. Shapiro
(1977) explained it by an analogy (at infinite Reynolds number). Consider a
one-dimensional, unsteady, frictionless flow in a collapsible tube, a gas flow
in a wind tunnel, and a liquid flow in a uniform, horizontal open channel. The
equation of motion is identical for each of the three cases:
au au 1 ap
-+u-= - - - (1)
at ax pax'
where p is the mass density of the fluid, p is the pressure in the flowing fluid,
u is the velocity, t is time, and x is longitudinal distance. The equations of
continuity are
aA a
For a collapsible tube: at + ax (Au) = 0; (2)

ap a
For the gas flow: at + ax (pu) = 0; (3)

ah a
For the channel flow: at + ax (hu) = o. (4)

Here A is the cross-sectional area in the first case, p is the mass density in the
second case, and h is the height of the free surface above the bottom in the
5.17 Flow in Collapsible Tubes 191

third case. The phase velocity of propagation of small perturbations, c, is, in

the three cases:
2 A d(p - Pe)
c =- . (5)
p dA '

c2 = (~~) at constant entropy; (6)

h dp
c2 = Pdh = gh; (7)

where g is the gravitational acceleration. Thus the analog is seen. Those

readers who are familiar with gas dynamics may recall the shock waves, the
supersonic wind tunnel, the Laval nozzle for steam turbine, the convergent
section to accelerate the fluid in subsonic regime, the sonic throat, and the
divergent section to accelerate the fluid in supersonic regime. Those familiar
with the open channel flow may recall the flow over a dam, and the hydraulic
jump. One could anticipate the existence of analogous phenomena in blood
flow in collapsible vessels. One anticipates also, of course, that similar
phenomena occur in air flow in the airways, Korotkov sound in arteries, urine
flow in urethra, etc.
For a flow from a reservoir with a fixed total pressure head Po into a
collapsible tube, the flow rate depends on the suction pressure p downstream
and the pressure outside the tube Pe' The rate of change of flow with respect
to the suction pressure P is given by the equation (to be derived below):

dQ A (u 2 ) (8)
d(p - Pe) = pu c2 - 1 .

Here Q is the flow rate, P is the internal pressure, u is the mean speed of flow,
and c is the speed of the flexural wave. Note that dQ/dp depends on the
ratio u/c. If the flow speed u is smaller than c, then decreasing internal
pressure increases the flow. If u is larger than c, then the reverse is true. Thus
the condition u = c signifies the maximum flow obtainable with decreasing
internal pressure. This maximum is Qrnax = Ac. At this condition, the maxi-
mum flow depends neither on the upstream pressure, nor on the downstream
pressure. It is an exact analog of the sonic throat ofthe supersonic wind tunnel.
The ratio

(9)

is called the speed index of Shapiro (1977). It plays a central role in liquid flow
through a collapsible tube as the Mach number does in gas dynamics.
Thus the condition u = c signifies a flow limitation. The upstream and
downstream pressures matter only to the extent of getting this condition
established, just as a supersonic wind tunnel has to have suitable conditions
to get it started.
192 5 Blood Flow in Heart, Lung, Arteries, and Veins

The derivation of the intriguing Eq. (8) is as follows. Consider laminar flow
in an elastic tube at a large Reynolds number so that Bernoulli's equation
holds:
(10)
where Po is the stagnation pressure, P is the static pressure, p is fluid mass
density, and u is velocity. The volume flow rate, Q is
Q = Au = AJ~(Po - p) = AJ~[(Po - Pe) - (p - Pe)]. (11)
A is the cross-sectional area which is a function of P - Pe' Although P varies
with distance down the tube, Q remains constant, of course. Differentiating Q
with respect to P - Peo we obtain, after some reduction and using Eq. (10),

dQ du dA A [p dA ] A (12)
d(p - Pe) = -A dp + d(p - Pe) u = - pu + A d(p _ Pe) -pu.
The factor in the bracket of the last term is 1/e 2 • Thus Eq. (8) is obtained.
Shapiro (1977) analyzed a number of cases in which these equations apply.
Experience shows that this one-dimensional analysis is adequate to deal with
the flow leading to the sonic throat. But beyond the sonic throat, the recovery
of flow to the subsonic condition seems to be a three-dimensional pheno-
menon beyond the reach of the simplifed approach.

5.18 Pulse Wave as Message Carrier for Noninvasive Diagnosis

A subsonic flow is influenced by conditions on all of its boundary. The flow

field, governed by the equations of motion and continuity and the constitu-
tive equations, is determined by the conditions on the boundary. Anything
happening anywhere on the boundary will be felt everywhere in the flow field.
In fluid mechanics and the theory of partial differential equations, this is a
feature of potential flow or elliptic differential equations, as distinguished from
supersonic flow or hyperbolic differential equations. Now, blood flow is
subsonic. Therefore, if we have the full, detailed mathematical solution of the
flow field, then, in principle, by examining the flow at a given region, one
should be able to tell any disturbances occurring anywhere on the boundary.
Extending this concept to the diseases of the blood vessels and organs, we can
anticipate that the pulse waves in a given region of an artery should carry
information about stenosis, aneurysm, or atherosclerosis at distant places.
The object of studying the messages carried in arterial pulse waves is similar
to the use of seismic waves to detect oil reserves underground. The mathe-
matical problem has not been solved yet, but anecdotal, empirical information
exists.
In the traditional Chinese medicine, physicians use fingers to feel the pulse
waves of the radial artery on the forearm at the wrist. Through empirical
information accumulated over the years, they have developed an art of
Problems 193

diagnosis which is often marvelous but not well understood. The use of pulse
waves for diagnosis was discussed extensively in one of the most ancient
classics of medicine: the Nei Jing, i.e., the Internal Classic of Huangti (for
Chinese references see Fung, 1984, pp. 14 and 157; and Xue and Fung, 1989)
which is believed to have been written in the Warring Period (475-221 BC).
In essence, the idea is that all disturbances in the function of any organ can
be detected by changes in the pulse waves in the radial artery. The sensation
felt by the fingers when they press on the radial artery at specified points with
varying degrees of pressure is used as diagnostic criterion.
The pulse wave diagnosis method is studied intensively in China clinically,
experimentally, and theoretically. In older literature the waves are treated as
axisymmetric motion in circular cylindrical elastic tubes. Recent literature has
included articles treating non-axisymmetric motion, including lateral oscilla-
tion of the centerline of the blood vessel. Dai et al. (1985) tested the hypothesis
that a disturbance of blood flow at one place can be detected in the arterial
pulse wave at a distant site. They transiently occluded blood flow in a leg and
recorded the pulse waves in both radial arteries. They asked whether the right
and left radial arterial waves can differentiate a disturbance in the right leg
from that in the left leg. The results show that the right and left radial arterial
waves do respond to the disturbances in the right and left legs differently, but
the discrimination is not very strong. Xue and Fung (1989) tried to explain it
on the basis of fluid mechanics. They created an unsymmetrical entry condi-
tion by blocking off one-half of the entry section of a circular cylindrical tube.
As the distance from the entry section increases, the flow tends to become
axisymmetric, but there is an asymmetric component which persists in pro-
pagating downstream with slowly damped amplitude. This suggests that the
asymmetric flow condition from the legs may reach the arms, but whether the
suggestion is quantitatively meaningful or not is entirely unknown. This
remains a fascinating problem.

Problems
5.1 An energy balance equation for blood flow is desired. Consider all the arteries
between two planes, for example, one plane cutting a renal artery, the other plane
cut through the kidney supplied by that renal artery. Identify the rate of gain of
energy of the blood in these arteries (the sum of the kinetic, potential, and internal
energies) and the rate at which work is done on the blood in this system. The
energy balance requires that the rate of gain of energy must be equal to the sum
of the rate of work done on the system and the heat transported in. Express this
energy equation in terms of pressure and velocities in the system. Cf. Fung (1984)
pp. 15-20.
5.2 One of the great achievements of man in the twentieth century is the mechanical
heart. What is the present status of the art in this field? What do you think needs
to be done in order to make this device really available to more people at an
affordable price?
194 5 Blood Flow in Heart, Lung, Arteries, and Veins

5.3 What effect does a stenosis in a large artery have? To study the effect, laser-doppler
velocimeter may be used. Describe the principle of this instrument. Can it be used
for an unsteady flow?
5.4 Describe the theoretical criterion for the velocity distribution in the boundary
layer when boundary separation from a solid wall occurs. (Cf. Yih (1977) pp. 352,
360.)
5.5 Consider the pulsatile flow in the aortic arch, part of which is highly curved as a
torus. Because of the curvature of the vessel, secondary flow exists and the
boundary layer thickness is a function of time and space. For the consideration
of atherogenesis, we need to know the shear stress on the arterial wall. Give a
qualitative discussion on the nature of variation of the flow and shear stress in the
aortic arch. (Cf. Jayaraman, G., Singh, M.P., Padmanabhan, N. and Kumar, A.
(1984). "Reversing flow in the aorta: a theoretical model", J. Biomechanics, 17:
479-490.)
5.6 An aorta has an aneurysm, which is a sac formed by the dilatation of the wall of
the artery. From the point of view of fluid mechanics, discuss the pulsatile velocity
distribution and pressure in the aneurysm and its contiguous parts, and the
possible sound emission (bruit, aneurysmal bruit). From the point of view of solid
mechanics, discuss the stress distribution in the vessel wall. From the general
biological relation between stress and growth or resorption, discuss possible
reasons for the creation of the aneurysm and possible direction of its development.
(Cf. Chapters 10-13 infra.)
5.7 Discuss the stress distribution in the endothelium, the intima, and the adventitia
in the region of arterial bifurcation. Delve into further detail, considering the
stresses acting in the endothelial cells, smooth muscle cells, collagen fibers of
various kinds, elastin fibers, fibronectin, and ground substances in the vessel
wall. Again, precise data are lacking. Develop a research proposal to clarify this
problem. Again, cf. Chapters 10-13, and biological points of view as mentioned
in Prob. 5.6.
5.8 Looking at the stenosis problem of 5.3 from the point of view of solid mechanics
and biology as mentioned in Prob. 5.6, discuss the possible remodeling of the blood
vessel wall when a stenosis develops.
5.9 The place where an artery branches off from the aorta is often the site of athero-
sclerosis. Discuss qualitatively the velocity distribution, fluid pressure, and wall
shear stress on the endothelium in this region. Develop a plan of research to gain
a better understanding of these features.

References

Bellhouse, B.J. and Bellhouse, F.H. (1969). Fluid mechanics of model normal and stenosed aortic
valves. Cire. Res. 25: 693-704.
Bellhouse, BJ. and Bellhouse, F.H. (1972). Fluid mechanics of a model mitral valve and left
ventricle. Cardiovase. Res. 6: 199-210.
Dai, K., Xue, H., Dou, R., and Fung, Y.c. (1985). On the detection of messages carried in arterial
pulse waves. ASME J. Biomed. Eng. Vol. 107, pp. 268-273.
References 195

Euler, L. (1775). Principia pro motu sanguins per arterias determinado. Opera posthuma mathe-
matica et physica. Petropoli, Vol 2, pp. 814-823.
Flaherty, J.E., Keller, J.B., and Rubinow, S.I. (1972). Post buckling behavior of elastic tubes and
rings with opposite sides in contact. SIAM J. Appl. Math. 23(4): 446-455.
Folkow, B. and Neil, E. (1971). Circulation, Oxford Univ. Press, New York.
Frank, O. (1899). Die grundform des arteriellen pulses. Erste Abhandlung, Mathematische
Analyse. Z. BioI. 37: 483-526.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
Fung, Y.C. (1984). Biodynamics: Circulation. Springer-Verlag. New York.
Fung, Y.c. and Sobin, S.S. (1972). Pulmonary alveolar blood flow. Circ. Res. 30: 470-490.
Fung, Y.c., Sobin, S.S., Tremer, H., Yen, M.R.T., and Ho, H.H. (1983). Patency and compliance
of pulmonary veins when airway pressure exceeds blood pressure. J. Appl. Physiol. 54: 1538-
1549.
Hales, S. (1733). Statistical Essays. II. Haemostaticks. Innays and Manby, London, Reprinted by
Haffner, New York. p. 23.
Lee, C.S.F. and Talbot, L. (1979). A fluid mechanical study on the closure of heart valves. J. Fluid
Mech. 91: 41-63.
McDonald, D.A. (1974). Blood Flow in Arteries. Williams and Wilkins, Baltimore, MD.
Moreno, A.H., Katz, A.I., Gold, L.D., and Reddy, R.V. (1970). Mechanics of distension of dog
veins and other very thin-walled tubular structures. tirc. Res. 27: 1069-1079.
Nerem, R.M., Seed, W.A., and Wood, N.B. (1972). An experimental study of the velocity distri-
bution and transition to turbulence in the aorta. J. Fluid Mech. 52: 137-160.
Patel, D.J. and Vaishnav, R.N. (eds.) (1980). Basic Hemodynamics and Its Role in Disease Process,
University Park Press, Baltimore, MD.
Pedley, T.J. (1980). The Fluid Mechanics of Large Blood Vessels. Cambridge University Press,
London.
Shapiro, A.H. (1977). Steady flow in collapsible tubes, J. Biomech. Eng. 99: 126-147. The American
Society of Mechanical Engineers, New York.
Sobin, S.S., Fung, Y.c., Tremer, H., and Rosenquist, T.H. (1972). Elasticity of the pulmonary
interalveolar microvascular sheet in the cat. Circ. Res. 30: 440-450.
Winter, D.C., and Nerem, R.M. (1984). Turbulence in pulsatile flows. Ann. Biomed. Eng. 12:
357-369.
Womersley, J.R. (1957). The mathematical analysis of the arterial circulation in a state
of oscillatory motion. Wright Air Development Center, Technical Report WADC-TR-56-
614.1-123.
Xue, H. and Fung, Y.c. (1989). Persistence of asymmetry in nonaxisymmetric entry flow in a
circular cylindrical tube and its relevance to arterial pulse wave diagnosis. J. Biomech. Eng.
Vol. 111, pp. 37-41.
Yen, R.T. and Foppiano, L. (1981). Elasticity of small pulmonary veins in the cat. J. Biomech.
Eng. 103: 38-42.
Yen, R.T., Fung, Y.c., and Bingham, N. (1980). Elasticity of small pulmonary arteries in the cat.
J. Biomech. Eng. 102: 170-177.
Yih, C.S. (1977). Fluid Mechanics. West River Press, Ann Arbor, MI.
Young, T. (1808). Hydraulic investigations, subservient to an intended Croonian lecture on the
motion of the blood. Phil. Trans. Roy. Soc. London, 98: 164-186.
Young, T. (1809). On the function of the heart and arteries. Phil. Trans. Roy. Sec. London, 99: 1-31.
CHAPTER 6

Micro- and Macrocirculation

6.1 Introduction

In physiology, capillary blood flow is identified with microcirculation. Flow

in small blood vessels supplying and draining the capillaries, the arterioles
and venules, respectively, are also included in microcirculation, but the ques-
tion of how many orders are to be included in microcirculation is sometimes
debated, because different organs seem to demand different answers. From
fluid mechanical point of view, the distinction between micro and macro
circulation can be based on the Reynolds number, VD/v, and Womersley
number, (D/2)JW!V (Sec. 5.16), where V represents the mean velocity of flow
in the vessel, D is the vessel diameter, v is the kinematic viscosity of the blood,
OJ is the circular frequency of oscillation of the blood velocity fluctuations. If
the Reynolds number and Womersley numbers are both much smaller than
1, then the inertial force can be ignored, and the flow is said to be microcircula-
tion. If both numbers are much greater than 1, then the fluid viscosity can be
ignored, and the flow is said to be macro circulation. In between these limits
the fluid mechanical equations are harder to solve, and it is immaterial
whether you classify them as micro or macro circulation.
Some of the general features of microcirculation are the following: The
blood has a lower hematocrit (i.e., the volume fraction of blood cells in whole
blood) than that in the heart. The apparent viscosity of blood decreases
because of the decreasing concentration of red cells. The chance for activated
white blood cells to stick to the blood vessel wall increases because of closer
contact with the endothelium. The proportion of smooth muscle in vessel wall
increases in small vessels. In systemic arterioles 70-80% of the vessel wall is
smooth muscle cells. The contraction of the smooth muscle controls the vessel
diameter, flow resistance, pressure gradient, and thus, eventually, the distribu-
tion of blood to various organs, as well as the systemic blood pressure.

196
6.2 Anatomy of Microvascular Beds 197

One can get an intuitive feeling about what is going on in circulation by

imagining oneself as a red blood cell. In the left ventricle, the cell is so small
that the whole blood may be considered as a homogeneous fluid. A complex,
unsteady flow takes place in the ventricle and aorta. The hydrodynamic force
opens and closes the aortic valve periodically. The aortic diameter is about
2000 times larger than the red cell. Soon the cell gets to a bifurcation point of
the channel. It is swept into one of the branches; then another bifurcation,
then another. Finally, the cell gets into an arteriole, whose diameter is about
10 times larger than the red cell. After several generations of arterioles, the
red cell enters a capillary blood vessel which is so narrow that its wall scrapes
the red cell membrane. The cells are now flowing in single file. The plasma
fluid sticks to the vessel wall and the cell membrane. There is no slip, but
there is some leak back at the wall. Some fluid seeps through the endothelium.
Some dissolved gas leaves the fluid and goes into the vessel wall and the
tissue cells beyond. Other gases come back in reverse direction. Some bigger
molecules move across the wall via tiny channels between the membranes of
the endothelial cells.
After a winding trip in the capillaries, the red cell enters the venules and
slows down. From venules it goes to the veins. The veins have valves which
prevents backflow in unfavorable conditions. In vena cava, there may be a
"sluicing gate" where the flow velocity is equal to the velocity of pulse wave.
This is the choking point of the so-called "waterfall" phenomenon.
Then the red cell enters the right atrium, right ventricle, pulmonary arteries,
pulmonary capillaries, pulmonary veins, left atrium, and returns to the left
ventricle.
Mechanics can help clarify the understanding of these phenomena. As
usual, to understand mechanics we must begin with anatomy and rheology.

6.2 Anatomy of Microvascular Beds

The microvascular beds of all organs are not the same. There are common
features; there are also special features of each organ. To study an organ
one must know its vasculature. Hence morphometry has been a recognized
activity of bioengineers. However, it is beyond the scope of this book to
describe every organ. The reader is referred to the bibliography at the end of
the chapter for general references. For special organs, a search in the library
is necessary; or a research project has to be undertaken.
Let us describe some common features. The smallest blood vessels are the
capillaries. The wall of a capillary blood vessel consists of a single layer of
endothelial cells lying on a basement membrane which occasionally splits to
enclose the pericytes which are thought to have the potential to become
smooth muscle cells. There is a large number of vesicles in the endothelial
cells. These vesicles are believed to be transporters of materials.
The endothelial cells of the capillary wall are apposed to each other. In
electron microscopy, there appears to be a gap of 10 to 20 nm between
198 6 Micro- and Macrocirculation

neighoring endothelial cell membranes. At certain points, these membranes

and the adjacent cytoplasm appear darker; the intercellular clefts are sealed
by tight junctions or maculae occludens, which are formed by close apposition
or fusion of the external leaflets of plasmalemma. In certain areas (e.g., in the
brain) these junctions form an uninterrupted seal, i.e., zonulae occludens,
preventing the passage of molecules with radius of 2.5 nm or larger. These
tight junctions are like spot welding (maculae) and seam welding (zonula) in
industrial metal construction. They connect the endothelial cells together to
form a continuous barrier and play an important role in determining the
permeability of the endothelium to water and other molecules.
The appearance of the endothelial cell lining of blood vessels might be
different in different organs. Generally there are three major types. In the
continuous type, the endothelial cells are joined tightly together. In vessels of
striated muscle, the cells may be quite flat and thin. In postcapillary venules,
they may be cuboidal and form a thick layer. In the fenestrated type, the
endothelial cells are so thin that at some spots the opposite surfaces of their
membranes become so close together as to form small circular areas known
as diaphragms of fenestration approximately 25 nm thick and 100 nm across.
Adjacent endothelial cells are still tightly joined. This type of vessel has been
described in three groups or organs: (1) endocrine gland, (2) structures engaged
in the production or absorption offluids (e.g., renal glomerulus, choroid plexus
of the brain, ciliary body of the eye, intestinal villus), and (3) retia mirabilia
(e.g., renal medulla, fish swim bladder).
The third type of endothelium is the discontinuous type in which there are
distinct intercellular gaps and discontinuous basement membrane. These
occur in those vessels commonly called sinusoids. They are common in organs
whose primary functions are to add or extract whole cells as well as large
molecules and estraneous particles, e.g., liver, spleen, and bone marrow, from
the blood.
The topology of the arterioles that supply the capillaries and venules that
drain them is usually special for each organ. Some are organized as trees;
others are organized into arcades. Each organ is unique.
Lying next to arterioles are lymphatic capillaries which have a single layer
of endothelium surrounded by a basement membrane, but lack smooth muscle
in its walls. The walls of lymphatic capillaries are more porous than those of
capillaries so that larger molecules can pass through them.
The lymphatic capillaries are blind sacs. They merge to form collecting
lymphatics which transport fluid to the vein. Collecting lymphatics have valves
to assure unidirectional movement of lymph.
The complement to the space occupied by the blood vessels, lymphatics,
and cells of a tissue is called the interstitial space of that tissue. It is mainly
connective tissue, containing collagen, elastin, hyaluronic acid, and other
substances, either bathed in some kind of fluid, or embedded in a gel.
Sympathetic nerve fibers invest the aorta, large and small arteries and veins,
and to a variable degree the networks of the arteriolar vessels and muscular
6.3 Major Features of Microcirculation 199

venules in each organ. There appears to be no direct innervation of the

capillary blood vessels and collecting venules, although fibers may be found
in the capillary region. Sympathetic fibers are usually superimposed on the
smooth muscle layer of the blood vessel wall, but do not make direct synaptic
contact with the vascular smooth muscle cells.

6.3 Major Features of Microcirculation

In a typical microvascular bed of the cat mesentery, the arterial to venous

distribution of intravascular pressure and velocity is shown in Fig. 6.3: 1. The
pressure decreases rapidly in arterioles of diameters in the range of 10-35 f.lm.
The decline in pressure within the true capillaries and postcapillaries is much
more gradual.
When the pressure profiles of normal, hypertensive, and hypotensive cats

PRESSURE - VE LOCITY DISTRIBUTION

VELOCITY - mm/sec PRESSURE - mmHg

40 100
ARTERIAL VENOUS
36 90
~PRESSIIRE
32 80 MESENTERY
CAT
28 70 ' PRESSIIRE
o VELOCITY
o .. 24 60

, 20 50
o. .. •

,0
o
, 0

,
00

56 48 40 32 24 16 8 16 24 32 40 48 56

VESSEL DIAMETER- micro

FIGURE 6.3:1 Arterial to venous distribution of intravascular pressure and velocity in

the mesentery of the cat. Vessel diameter (abscissa) is taken to be representative of the
functional position of each vessel in the microvascular network. Each data point
represents the average value of three to five individual measurements at the abscissa
(diameter) value. The solid curves are piece-wise cubic spline fits of the data and
are statistically representative of the arterial to venous trends. From Zweifach and
Lipowsky (1977, p. 386). Reproduced by permission of the American Heart Associa-
tion, Inc.
200 6 Micro- and Macrocirculation

are compared, it is found that in the hypertensive cats the pressure drop in
arterioles is larger than in the normals, whereas that in the hypotensive cats
is smaller. Pressures in the capillaries and postcapillary vessels are similar
in the three groups, the difference between the mean pressures being only
3-5 mmHg. Thus it appears that the arterioles control the blood pressure in
such a way that the capillary pressure is maintained in the normal range while
the central arterial pressure can fluctuate.
In the skin, it is observed that pressures in about 10% of the venules are
sometimes higher than expected (approximately 70 mmHg compared with the
usual value of 30 mmHg). This is taken as evidence of the existence of shunts
(such as thoroughfare channels) in the microvascular bed. These shunts are
more direct low-resistance pathways connecting arterioles to venules.
Pressure and velocity pulsation are detectable in micro vessels, in which the
pressure oscillates with an amplitude of 1-2 mmHg normally, and 2-4 mmHg
if the precapillary sphincter were dilated. There is also a random-fluctuation
of a period in the order of 15-20 sec, with an amplitude of 3-5 mmHg. A third
type of pressure variation is more substantial and lasts longer, in the order of
10 mmHg and 5-8 min, followed by a return to the steady-state condition in
about 2-3 min. See Zweifach (1974).
It is not surprising that cardiac pulses must cause ripples in the capillaries.
These waves are attenuated in the direction of propagation, reflecting the fact
that in the capillaries the viscous stress dissipates the pressure fluctuations.
The wave speed in the capillary is estimated to be approximately 7.2 cm/sec.
So far we have considered the pressure in the blood vessels. Of equal
importance of course, is the pressure outside the blood vessels, because trans-
mural pressure (i.e., internal - external pressures) is relevant in most phe-
nomena involving blood vessels. For example, the collapse of the vein (Sec.
5.19) occurs when the external pressure exceeds internal pressure by a certain
amount. The pulse wave speed depends on the Young's modulus of the blood
vessel wall, the Young's modulus depends on the stress level, the stress level
depends on the transmural pressure. Coming to microcirculation, the pressure
outside the capillary blood vessels plays an important role in mass transport.
This is because capillary walls are semipermeable: it can let water move across
it, but is an impediment to ions and molecules. As a semipermeable membrane,
the rate of movement of water across the capillary wall is commonly assumed
to obey Starling's hypothesis (see Sec. 8.5):
rh = K[Pb - Pt - O"(7tb - 7t t )] (1)
where P stands for hydrostatic pressure, 7t stands for osmotic pressure, the
subscript b stands for blood, t stands for tissue, rh represents the volume flow
rate per unit area of the membrane, K is the permeability constant, and 0" is
the reflection coefficient. In order to calculate the fluid transfer rate rh, we must
know all six quantities, Pb' Pt, 7t b, 7tt , 0", and K; they are equally important.
Since fluid movement in the tissue space is very important to health (too much
or too little fluid in the tissue means edema or dehydration), the measurement
of Pt and 7tt has engaged attention of physiologists for a long time.
6.3 Major Features of Microcirculation 201

We can decrease the tissue pressure PI in our skin by rising in a balloon.

We can increase it by diving into the ocean. We can increase tissue pressure
on an arm by applying a tourniquet, the pressure in abdomen by tightening
abdominal muscle as in weight lifting. We can increase blood colloidal osmotic
pressure by adding dextran into plasma, decrease blood colloidal osmotic
pressure by adding saline. Hence Pb' PI' nb are variable to some extent. The
regulation of Pb' PI' nb' n l , and K in life is one of the central problems in
physiology.
The velocity and hematocrit distributions in any capillary bed are very
nonuniform and change from one instant of time to another. If you watch a
microcirculatory bed in vivo, you see red cells flow and ebb.
The most unique feature of microcirculation is, of course, the prominence
of red blood cells. In small blood vessels, individual red cells are so big that
they may fill the vessel from wall to wall. Flowing in the capillaries, the shape
ofthe red cells has been described as slipper, parachute, bullet, or crepe suzette.
Figure 6.3:2 shows the shape of red blood cells in a capillary, computed by
Zarda et al. (1977) on the basis of a constitutive equation of the red cell
membrane proposed by Skalak et al. (1973) and Evans and Skalak (1979). The
cell and the capillary are both assumed axisymmetric. In the unstressed state
the red cell diameter is assumed to be larger than that of the tube. The values

FIGURE 6.3:2 Shapes computed for red blood cells flowing axisymmetrically in capil-
lary. The unstressed radius of the red cell is 3.91 Jlm, and is 5% larger than radius of
the tube. From Zarda et al. (1977). Reproduced by permission.
202 6 Micro- and Macrocirculation

of Ap on top of the figure is proportional to the total force applied to the red
cell; it is proportional to the velocity of the cell relative to the vessel wall.
The natural cross-sectional shape of a stationary red cell is shown in Ap = O.
The deformed shapes of the cell are shown for successive values of Ap. The
faster the cell moves, the larger the gap between the cell and vessel wall
becomes. The resistance of each cell to the flow of the blood is proportional
to Ap.

6.4 The Rheological Properties of Blood

Blood is a mixture of plasma and blood cells. Plasma is a mixture of water,

proteins, enzymes, and other substances. When plasma is tested in a visco-
meter, it is found that its coefficient of viscosity is a constant independent of
the rate of strain. Hence it is a Newtonian fluid by definition, with a coefficient
of viscosity about 1.2 centipoise.
Inside of the red cell is a hemoglobin solution which has a coefficient of
viscosity of about 6 centipoise. Hence it is also a Newtonian fluid. Thus blood
consists of one Newtonian fluid wrapped in small packages that float in
another Newtonian fluid.
In large blood vessels or in viscometers whose dimensions are much larger
than the diameter of the red blood cell, the mixture appears as a homogeneous
fluid. The shear stress-shear strain rate relationship of the mixture can be
determined by viscometry. The ratio of the shear stress to the shear strain rate
is the coefficient of viscosity, see Eq. (1.7:6). Experimental results show that
blood viscosity is non-Newtonian (see Fung, 1981, Ch. 3). However, as we
have discussed in Sec. 5.4, in large blood vessels the non-Newtonian feature
of blood is unimportant, and blood can be treated as a Newtonian fluid with
a constant coefficient of viscosity without causing a significant error in the
pressure-flow relationship.
In the arterioles and venules, there is a well-known Fahreus-Lindquist
effect: the effective coefficient of viscosity decreases as the diameter of the
blood vessel becomes smaller. This is explained by decreasing hematocrit in
the smaller vessels. If many small blood vessels were attached to a reservoir,
then the hematocrit in smaller vessels is lower. Lower the hematocrit, smaller
is the viscosity.
In capillary blood vessels, the blood has to be treated as a two-phase fluid
consisting of the plasma and the blood cells. The pressure-flow relationship
of such a two-phase fluid in a capillary bed depends very much on the
geometric characteristics of the bed. In long cylindrical capillaries such as
those in muscles, retina, and mesentery, the relationship is nonlinear, espe-
cially in tightly fitting capillaries (Zarda et aI., 1977). On the other hand, in the
pulmonary capillaries, which are organized into sheets (see Figs. 6.6:2, 6.6:3,
and 6.6:6), it has been shown that the pressure-flow relationship is linear (see
Sec. 6.6). Hence an "effective" coefficient of viscosity can be determined by
fitting experimental data with a theoretical formula, and this effective coeffi-
6.6 An Example of Analysis: Pulmonary Blood Flow 203

cient of viscosity is a constant (independent of velocity of flow) in the pul-

monary capillaries.

6.S General Equations Describing Flow in Microvessels

Having identified the constitutive equations of the materials, we can now write
down the field equations and boundary conditions that govern a problem of
interest. For example, if one is interested in the pressure-flow relationship in
a capillary blood vessel, we have the Stokes equation for the plasma in the
vessel:
op
o = -1 -;- 2
+ vV' u i . (1)
P UXi
The acceleration term on the left-hand side of the equation can be set to zero
because the Reynolds number and Womersley number (Sec. 5.13) are both
much smaller than 1. Here U i is the velocity vector that refers to a set of
rectangular Cartesian coordinates fixed in the blood vessel. A similar equation
applies to the hemoglobin solution inside the red cell. The boundary condi-
tions for the fluids on the red cell membrane and the endothelium of the blood
vessel are the no-slip condition for the velocity and continuity of normal
displacements and stresses at the interfaces. The field equations for the blood
vessel wall are the same as those of Sec. 1.7. The field equations for the cell
membrane are the thin shell equations described in books on thin shells.
Now, if the pressure distribution and other stresses at the entry and exit
sections are prescribed, then one can compute the flow in the vessel.

6.6 An Example of Analysis: Pulmonary Blood Flow

The principle of mechanics can be applied to any organ. An infinite variety of
questions may be asked. To pick one for illustration, let us discuss the blood
flow in the lung.
The lung is coupled to the heart. Its vascular system is illustrated in Fig.
5.2:2. Blood flows from the right ventricle to the pulmonary artery, then to
pulmonary capillary blood vessels, to pulmonary veins, and finally to the left
atrium of the heart. Figure 6.6:1 shows a schematic diagram of a lobule of the
lung drawn by Miller (1947), p. 75. It illustrates the relation of the blood vessels
to the air spaces. Note that the arteries are adjacent to bronchi, whereas the
veins stand alone. Bronchial vessels and lymphatics are also shown in the
figure.
The geometry of the pulmonary arterial and venous trees can be described
by the Strahler system. In this system the capillaries are counted as vessels of
order O. The smallest arterioles are called vessels of order 1. Two order 1 vessels
meet to form a larger vessel of order 2, and so on. But if an order 2 vessel
meets a vessel of order 1, the order number of the combined vessel remains
as 2. A similar counting is done for the venous tree. The ratio of the number
204 6 Micro- and Macrocirculation

FIGURE 6.6:1 A schematic diagram of the pulmonary blood vessels, bronchial blood
vessels, and lymphatic vessels in the lung and in the pleural surface. From William S.
Miller, The Lung, 1947. Courtesy of Charles C. Thomas, Publisher, Springfield, Illinois.
B is bronchiole, leading to two alveolar ducts one of which is shown here. A = atrium,
AL V, AL V' = alveoli, S.AL = alveolar sac. P = pulmonary pleura. 1 = pulmonary
artery, dividing into capillaries. 2 = branches of pulmonary arteries distributed
to bronchioles and ducts and then broken up into capillaries which unite with capil-
laries derived from the bronchial arteries. 3 = pulmonary vein. 4 = lymphatics,
5 = bronchial artery and capillaries. 5' = bronchial arterial supply in the pleura (in
animals with a thick pleura). 6 = pulmonary venule. 7, 8, 9, 10 are situations in which
lymphoid tissue is found.

of vessels of order n to that of order n + 1 is called the branching ratio. The

ratio of the diameter of the vessels of order n to that of n + 1 is called the
diameter ratio. Similarly, a length ratio is defined. Yen et al. (1981,1983) have
measured the branching pattern of the pulmonary arteries and veins of the
cat. The results for the arteries are presented in Table 6.6:1. Similar results for
pulmonary veins are presented in Fung (1984), p. 338.
 0-

""~
~
P>
TABLE 6.6:1 Morphometric data of pulmonary arteries of the cat measured at transpulmonary pressure PA - PPL = 10 cm H 20 3
'E..
Compliance 2 "8.,
Number of branches Diameter! Length Apparent
in right lung viscosity coefficient IX
~
Do. L. f3 a
'<
Order (cm) (cm) (10- 4 cm p;/) (l0-4p;! ) en
N. 1l.(CP) ~.

"'C
1 300,358 0.0024 0.0116 2.5 0.00463 1.928
2 97,519 0.0044 0.0262 3.0 0.00848 1.928
"9"o
i:l
3 31,662 0.0073 0.0433 3.5 0.01407 1.928 P>
....
'<
4 9,736 0.0122 0.0810 4.0 0.02352 1.928 1:1:1
5 2,925 0.0192 0.151 4.0 0.02154 1.122 0"
o
0-
6 774 0.0352 0.272 4.0 0.02802 0.796 ."
7 202 0.0533 0.460 4.0 0.03807 0.714 0"
::;:;
8 49 0.0875 0.819 4.0 0.09818 1.122
9 12 0.1519 1.426 4.0 0.4045 2.663
10 4 0.2486 1.187 4.0 0.6620 2.663
11 0.5080 2.500 4.0 1.353 2.663
1 Yen et al. (1983) gives the diameter data of vessels of orders 1-4 measured at P - PA = - 7 cm H 2 0, whereas those of orders 5-11 were measured at
P - PPL = 3 cm H 20. In this Table, Don are the diameters at zero "transmural" pressure, defined as P - PA = 0 for orders 1-4, and P - PPL = 0 for orders 5-11,
and are computed from the data of Yen et al. (1980) according to the equations D = DoW + P(p - PAl] for orders 1-4 and D = Do[1 + P(p - PPL)] for
orders 5-11.
2 Compliance was computed by fitting Eq. (1) to experimental data. Yen et al. (1980) listed data according to vessel diameters, which were interpolated to obtain
data on compliance vs. vessel order. In Eq. (1), r:x = f3D on . Yen et al. (1980), however, fitted data with a parabola. Their r:x is our p. Their f3 is zero in Eq. (1).

tv
oVI
206 6 Micro- and Macrocirculation

In the lung, the smallest unit of air space is the alveolus, which is bounded
by networks of capillary blood vessels. The walls of each alveolus are shared
by neighboring alveoli, and are called interalveolar septa. The overriding fact
that determines the topology of the capillary blood vessels is that all pul-
monary alveolar septa in adult mammalian lungs are similar. Each septum
contains one single sheet of capillary blood vessels and is exposed to air on
both sides.
The function of the lung is to oxygenate the blood and to remove CO2 •
Nature chooses to do this by the principle of diffusion; and for this purpose,
blood is spread out into very thin layers or sheets so that the blood-gas
interfacial area becomes very large. In an adult human lung with a pulmonary
capillary blood volume on the order of 150 ml, the pulmonary capillary
blood- gas exchange area is of the order of 70 m 2 , so that the average com-
puted thickness of the sheets of blood in the pulmonary capillaries is only
about 4 11m. The thin membrane that separates the blood from the air is less
than 1 11m thick; it consists of a layer of endothelial cells, an interstitium, and

FIGURE 6.6:2 Cat lung. Flat view of interalveolar wall with the microvasculature filled
with a silicone elastomer. This photomicrograph illustrates the tight mesh or network
of the extensively filled capillary bed. The circular or elliptical enclosures are basement
membrane stained with cresyl violet and are the nonvascular posts. Frozen section
from gelatin-embedded tissue; glycerol-gelatin mount. The insert shows a detail from
the region indicated by the arrow. From Fung and Sobin (1969), by permission.
6.6 An Example of Analysis: Pulmonary Blood Flow 207

a layer of epithelial cells. Each interalveolar septum is a sheet of blood. Several

billion septa form a space structure which may be compared to the honeycomb
(Malpighi, 1661), or a bowl of soap bubbles. Each bubble is an alveolus that
is ventilated to the atmosphere through a branching airway system.
The dense network of the capillary blood vessels in an alveolar wall of the
cat's lung is shown in Fig. 6.6:2. To characterize the geometry of such a
network, we idealize the vascular space as a sheet of fluid flowing between two
membranes held apart by a number of fairly equally spaced "posts"; see Fig.
6.6:3. In the plane view (a) this is a sheet with regularly arranged obstructions.
The plane may be divided into hexagonal regions, with a circular post at the
center of each hexagon. The "sheet-flow" model is therefore characterized by
three parameters: L, the length of each side of the hexagon; h, the average
height of thickness of the sheet; B, the diameter of the posts.
We shall define a "vascular-space-tissue ratio" (VSTR) as the ratio of the
vascular lumen volume to a circumscribing volume between surfaces T and
B. VSTR represents the fraction of the blood volume over a sum ofthe volumes
ofthe vascular space and the posts. Morphometric data of cat, dog, and human
lungs show an average VSTR of 0.91 ± 0.02. Hence in the interalveolar septa,
91 % of the space is occupied by blood.

I\~I =3L=v'3a 7/~

® }----_J
II " ,,
I
PLAN I \.-l
-----0(
,
@
I-I I
r----~
\ ~ I 0
, I
\ I I
@ I ~-L- =1, -®---L
I \
I ,
I ,
(a)

SIDE
VIEW

(b)

FIGURE 6.6:3 Sheet-flow model. (a) Plane view. (b) Cross section through x-x of (a).
The cylindrical elements (circular in (a) and rectangular in (b)) are the "post." The space
between the top and bottom walls is the flow channel. For bounding surfaces only, T,
top, and B, bottom. Sheet thickness, h. C and D, contact of posts with endothelial
surface at T and B. From Fung and Sobin (1969). Reprinted by permission of the
American Heart Association, Inc.
 208 6 Micro- and Macrocirculation

• loo·;oo,...a
,. lOO· ~I)(JJ.' .a
• UJ(}·800", 0«1
.. 800 I}IJI),..",

+lrItAN~SD

Ei
~
<>
€
Q
100

I.IP . P, - p. em H,O (98 P.) I.Ip. P, - p. em H,O (98 P.)

\0
- 10 10 -II - 10 10

200
''t = - 10(", H;O ,", 98P~)
''t : · 'Orm H,O f'A 9al.) • IOO'}()()",m
~ • 100·}oo "m ... }()()·4/JOIl'"
Q .. 1OO-4IJO "'Iff ~ o 4O(}·800"m

.~
,. lOO·11O()",m
~
Cl4Of}-M}/JPM Q
1\0 A 80Q O{)(J ~e
~

~
tMEAN! SO
tMEAN!SO ~
~
:!!
Q 11\0
o
!ii~
<>
€IOO
c

- II -10 10 - 10 10
I.IP - Py - p. em H,O !91 P.) I.IP , P, - p. em H, O (98 PI)

FIGURE 6.6:4 Percentage change in diameter of pulmonary veins of the cat as a function
of the blood pressure, PV' The airway pressure, PA, is zero. The values of the pleural
pressure, PPL, is listed in the figure. The vessel diameter is normalized against the
nominal diameter when Pv - PA is 10 em H 20 at which the vessel cross section is
circular. From Yen and Foppiano (1981). Reproduced by permission.

The distensibility of pulmonary blood vessels has been measured. A typical

example is shown in Fig. 6.6:4. The empirical formula
D = Do + rx(p - PA)
(1)
applies to pulmonary arteries and veins quite well. Here D is the lumen (inner)
diameter of the vessel, p is the blood pressure, PAis the pressure of alveolar
gas, Do is the lumen diameter when P = PA, and rx is the compliance constant.
In the cat's lung, this formula applies for /1p in the range of - 10 to + 20 cm
H 2 0 (~ -1,000 to 2,000 N/m 2 ).1t is valid even for pulmonary veins subjected
to negative transmural pressure. See Fig. 6.6:4. In Sec. 5.15 we have explained
why the pulmonary veins will not collapse when they are subject to negative
transmural pressure because of the support they receive from the tension in
 6.6 An Example of Analysis: Pulmonary Blood Flow 209

alveolar septa which are externally attached to them. The same tethering by
interalveolar septa explains the linearity of the pressure-diameter curve:
because the vessels are embedded in a linearly elastic cushioning material.
Table 6.6:1 gives the values of a and f3 for the cat's pulmonary arteries. Similar
data for the veins and the variation of the compliance constant with the
transpulmonary pressure (= airway pressure minus pleural pressure) are given
in Fung (1984), p. 339.
The elasticity of the pulmonary capillary sheet is exhibited in Fig. 6.6:5.
The thickness h varies with the pressure difference Ap (equal to the static
pressure of blood minus the pressure of the alveolar air) as follows,
h = 0 if Ap is negative, and smaller than - 8, where 8 is a small
number about 1 cm H 2 0. (2)
h = ho + aAp, if Ap is positive and smaller than a certain
limiting value, Pu' (3)
h tends to a limiting value hco if Ap > Pu, and increases beyond
the limiting value. (4)

--- --
_ 12
§. II
(f)
U)
10
w
z 9
"'"
~
:r: 8
>--
7
>--
w h = 428 ±0219 tlp (CAT)
w 6 "MEAN±oSD
:r:
5
(f)

n::
« 4
-"
0
w
:>
-"
3
«
2
z
«
w
:2'
- 0
.c
-5 0 5 10 15 20 25 30 35 40
t::.p =CAPILLARY - ALVEOLAR PRESSURE (em H20)

FIGURE 6.6:5 A. Sheet thickness-pressure relationship of the cat. Experimental data

can be approximated by a discontinuous curve that is composed offour line segments:
a horizontal line h = 0 for /:lp negative, which jumps to h = ho at /:lp slightly greater
than 0, then continues as a straight line for positive IIp until some upper limit is reached,
beyond which it bends down and tends to a constant thickness. B, The elastic deforma-
tion of the alveolar sheet is sketched for three conditions: IIp < 0, IIp > O. The relaxed
thickness ho of the sheet is equal to the relaxed length of the posts. Under a positive
internal (transmural) pressure, the thickness of the sheet increases; the posts are
lengthened; the membranes deflect from the planes connecting the ends of the posts;
and the mean thickness becomes h. From Fung and Sobin (1972). Reproduced by
permission of the American Heart Association, Inc.
210 6 Micro- and Macrocirculation

TABLE 6.6:2 The compliance constant, DC, and the thickness at zero transmural pressure,
ho, of the pulmonary alveolar vascular sheet at specified values of transpulmonary
pressure, PA - PPL (p A = alveolar gas pressure, PPL = pleural pressure)
ho DC PA - PPL
Animal (11m) (I1m/cm H 2 O) Reference (cm H 2 O)
Cat 4.28 0.219 Fig. 6.5:1 10
Dog 2.5 0.122 Fung, Sobin (1972) 10
2.5 0.079 Fung, Sobin (1972) 25
Man 3.5 0.127 unpublished 10

In the small range -e < Ap < 0, h increases from 0 to ho. A rough

approximation is h = ho + (ho/e)Ap. (5)
Here IX, ho, hoo and e are constants independent of Ap. The parameter ho is the
sheet thickness at zero pressure difference when the pressure decreases from
positive values. The parameter IX is called the compliance coefficient of the
pulmonary capillary bed. The thickness h is understood to be the mean value
averaged over an area that is large compared with the posts, but small relative
to the alveoli. The known values of ho and IX are given in Table 6.6:2.
Furthermore, if Ao represents the area of a certain region of an alveolar
septa when the static pressure of the blood is some physiologic value Po, then
the area of the same region A when blood pressure is changed to p is
A:::::Ao· (6)
In other words, the area is unaffected by the blood pressure.
Now, turn attention to the blood. In pulmonary capillaries, red blood cells
flow in single file just as they flow in systemic capillaries. The cells deform in
response to fluid pressure and shear stress as a consequence of confinement
of the vessel wall and crowding of the cells. In the investigation of overall
pressure-flow relationship of the lung, however, the details of cell deformation
are of little interest. Only the relationship between the average pressure
gradient and average velocity is of concern. These quantities can be averaged
over the cross-sectional area of the blood vessels, and over a small area of the
interalveolar septum. If a local frame of reference with coordinates x, y is
attached to an interalveolar septum, Fig. 6.6:6, then we are interested in the
gradients of the blood pressure, grad p, with components op/ox, op/oy; and
the velocity offlow U, with components Ux, Uy • A dimensional analysis would
enable us to write

pressure gradient = - ~ Jlkf, (7)

where U is the mean velocity of flow, h is the thickness of blood sheet (lumen
ofthe pulmonary alveolar septum), Jl is the apparent viscosity, k is a dimension-
less function of the sheet width to sheet thickness ratio, around 12 by value,
6.7 Pulmonary Capillary Blood Flow 211

Alv ofu~

CApiU.,ry blooci
,. " rl

Alv olu

FIGURE 6.6:6 Left: A plan view of an interalveolar septum of cat lung. Right: A
cross-sectional view of three interalveolar septa. In the left panel, a domain of averaging
around a point (x, y) in an alveolar sheet is shown. From Fung and Sobin (1969).
Reprinted by permission.

f is a dimensionless function of the sheet geometry, the ratio of the sheet

thickness and the post diameter, the ratio of the interpostal distance to post
diameter, and the VSTR (vascular space to tissue space ratio) defined earlier.
The value of the geometricfactor flies between 1.5 to 5.0, see Lee (1969), Yen
and Fung (1973). The apparent viscosity J.l is a function of the coefficient of
viscosity of the blood plasma J.lo, the hematocrit H (ratio of blood cell volume
to whole blood volume), the cell diameter Dc, the Young's modulus of the cell
membrane Ee> and the sheet thickness h. An expression for the apparent
viscosity derived from a dimensional analysis is

J.l = J.loF'(~c , ~c~' H). (8)

Details of F' are given by Yen and Fung (1973), and Fung (1981).

6.7 Pulmonary Capillary Blood Flow

Consider the local pressure gradient and local velocity defined by averaging
over a small volume of an interalveolar septum, as discussed in the preceding
section. The constitutive equation of the blood vessel is given by Eq. 6.6:2 in
the range 0 < p - PA < an upper limit:
h = ho + a(p - pA). (1)
212 6 Micro- and Macrocirculation

The constitutive equation of the blood is given by Eq. (6.6:6). Combining it

with Eq. (1), we obtain
1 20h
u= -kfJ1.a. h OX' (2)

The law of conservation of mass for a steady flow in a sheet with impermeable
walls is

O(~~u) + O(~~ V) = o. (3)

Equations (2) and (3) together yield

~(h3
Ox
Oh) + ~(h3 Oh) = 0,
Ox oy oy
(4)

or

(::2 + :;2 )h4 = o. (5)

Thus the fourth power of h is governed by a Laplace equation. Expressed in

terms of pressure, we have, on defining
(6)
the result

(7)

Solution of the Simplest Case: h Independent of y

If h is a function of x only, then Eq. (5) becomes
d 2 h4
dx 2 = o. (8)

The general solution is

(9)
where C1 , C2 are arbitrary constants. To determine Cl' C2' we notice the
boundary conditions: (a) at the "arteriole," x = 0, the thickness of the "sheet"
is ha; (b) at the "venule," x = L, the thickness of the "sheet" is hv. Hence,
(10)
Solving for C1 , C2 and substituting into Eq. (9), we obtain
h4 = h: - (h: - h~)x/L, (11)
6.7 Pulmonary Capillary Blood Flow 213

< L--[--,-,--x -------,l hlll.-----.cx~

~I xl
:r ~
FIGURE 6.7:1 Plot of Eq. (12) showing the variation of h4 and h with x. With
an appropriate choice of units, we assume that the thickness h. is 1. h~ at x = L is
assumed to be 0.75, 0.25, and 0 for the three cases shown in the figure. The correspond-
ing values of hv are 0.931, 0.707, and 0, respectively.

or
(12)
For various combinations of ha and hv, the distribution of the thickness h
is shown in Fig. 6.7:1. It is seen that the exponent! makes h rather flat near
the arteriole, and constricts rather rapidly toward the venule if hv is small.
From Eqs. (1) and (2), we obtain the flow per unit width:
. 1 ah 4
Q = hU = - 4kfJl.1X fu· (13)

But Q is a constant on account of conservation of mass. Hence ah 4 jax is a

constant, and by Eq. (11) we can write (13) as
. 1 4 4
Q = 4kfJl.IXL (h a - hv)· (14)

This is the flow per unit width in a rectangular sheet. For a rectangular
sheet of length L in the streamwise direction, the total width is equal to the
projected area divided by L. Hence the total flow in the sheet is equal to the
product of Q and sheet area divided by L.

Flow in an Interalveolar Septum

The velocity field in an interalveolar septum can be described by streamlines.
The flow in a streamtube (a tube whose wall is composed of streamlines) is
one-dimensional to which the analysis above applies with x identified as the
coordinate along the axis of the streamtube. Thus, the flow in each streamtube
is equal to the product of Q given by Eq. (14) and the ratio (projected
areaflength L).
214 6 Micro- and Macrocirculation

The flow in a whole septum can be obtained by summing up the flow in all
the stream tubes. Let A be'the area of the septum and S be the vascular-space-
tissue ratio; then the total area of the vascular space is SA, and the total flow
in the septum can be written as
SA 4 4
Flow = 2 [(h o + IXdPart) - (h o + IXdPven) ], (15)
4J1kfL IX
where L is an average length of the stream tubes defined by the relation
SA area of streamtube
(16)
p = L (length of streamtube )2
in which the summation covers all individual streamtubes of the sheet whose
area is SA. Expressed in terms of sheet thickness, Eq. (15) is
1 4
Flow =C [hart -
4
hven], (17)

where
(18)

hart = ho + IX(Part - Palv) (19)

h ven = ho + IX(Pven - PaIJ· (20)

Effect of Gravitation and Anatomy. Regional Differences in the Lung

Equation (15) shows that the total flow in a group of alveoli supplied by an
arteriole and drained by one or more venule is controlled by the blood
pressure at the arteriole and venule, Part and Pven' respectively. Now, Part is
equal to the blood pressure at the pulmonary valve in the pulmonary artery,
minus the integrated product of resistance times flow from the pulmonary
valve to the arteriole, and minus the product of the specific weight of blood
times the height of that arteriole above the pulmonary valve. Similarly, Pven
is the difference of three terms: the left atrium pressure, minus the pressure
loss due to flow, minus the product of specific weight and the height of that
venule above the left atrium. Specific weight is the product of gravitational
acceleration and the density of the fluid. Hence, Part> Pven' and the flow are
influenced by the height which determines hydrostatic pressure, and the
anatomy which determines the pressure loss due to resistance to flow.
Because of the height and anatomical differences, regional differences exist
in the lung. For alveolar blood flow, it is convenient to define three zones as
follows:
Zone 1: Pven - Palv < Part - Palv <0 (21)
Zone 2: Pven - Palv <0< Part - Palv (22)
Zone 3: 0< Pven - Palv < Part - Palv· (23)
6.8 Waterfall Phenomenon in Zone 2 215

In Zone 1, the capillaries are collapsed; there will be very little flow, if any.
In Zone 3, Eq. (15) applies when the blood pressure is modestly high so
that the pressure-thickness relationship of the interalveolar septa is linear.
If the blood pressure in Zone 3 is so high that the thickness tends to a
constant, [Eq. (6.6:3)], then the compliance oc decreases as shown in Fig. 6.6:5.
Then, since the compliance is very low, the thicknesses at the arteriole and
venule are almost equal. Hence Eq. (17), which can be rewritten in the
following form,

Flow = ~(hart - hven)(h:rt + h;rthven + harth;en + h:en ), (24)

can be simplified. First, the four terms in the last parenthesis are almost equal.
Then, using Eqs. (18)-(20), we can reduce Eq. (24) into
. SA 3
Flow = ~(Part - Pven)hart· (25)
JlkfL
This formula gives the flow in alveoli located low in zone 3, where the
hydrostatic pressure is so high that interalveolar septa lose their compliance.
The flow in zone 2 is discussed in the following sections.

6.8 Waterfall Phenomenon in Zone 2

The inequalities
Pven < Palv < Part (1)
defines the zone 2 condition. In a standing man, a region sufficiently high
above the heart is zone 2. In this zone, the capillary sheets tend to collapse at
the venous end. Some sheets will be collapsed, others will remain open. In the
open sheets the thickness at the venule will be small. If hven is smaller than
hart> then the last term in Eq. (6.7:17) is negligible and it becomes

. 1 4 1
Flow = chart =
.
C [h o + OC(Part -
4
Palv)] . (2)

The flow is then independent of the downstream condition, in analogy with

a waterfall.
The waterfall phenomenon in the lung was discovered by Banister and
Torrance (1960) and Permutt et al. (1962). In earlier literature there was a
debate as to whereabout on the vascular tree the "waterfalls" or "sluicing
gates" are located. At that time the prevailing concepts were that capillaries
were rigid, veins were collapsible, and arterioles were muscular and vasoactive.
Hence the sluicing gates were searched for in veins and arterioles. Fung and
Sobin (1972), however, showed that pulmonary capillaries are collapsible,
whereas Fung et al. (1983) showed that pulmonary arteries and veins (and
216 6 Micro- and Macrocirculation

venules) will not collapse under negative transmural pressure, at least to

- 23 cm H 2 0. Hence the sluicing gates must be located at the venular ends
of the capillaries, and Eq. (2) holds.
Fung and Yen (1984) pointed out that one ofthe solutions of Eq. (6.7:5) is
h=O (3)
which can occur if an area of the alveolar sheet is collapsed. Furthermore, the
multifaceted alveolar structure of the lung is such that areas of collapsed
alveolar sheets can be embedded in other alveolar sheets which are open and
perfused. The boundary conditions for the solution (Eq. (3)) are satisfied. In
the opened area there is flow, and Eq. (6.7:17) applies. The total perfused area
S in the constant C of Eq. (6.7:17) must be reduced by the collapsed area.
Hence to determine the flow in zone 2 condition, we must estimate the alveolar
sheet area that is closed. This is done in the next section.

6.9 Open and Closed Capillary Sheets in Zone 2

The estimation of the area of the open capillary sheets is helped by several
pieces of new information:
(1) The details of flow through the sluicing gate, taking into account the
effect of tension in the alveolar wall and the local curvature of the membrane,
as well as the Stokes flow equation have been investigated by Fung (1984) and
Fung and Zhuang (1986). It is shown that as Pven is decreased below Palv, the
flow in the gate is increased while the gate narrows further.
(2) The stability of a partially collapsed interalveolar septum illustrated in
Fig. 6.9:1 (b) has been investigated by Fung and Yen (1986). The sheet shown
in Fig. 6.9:1 (b) is open at the left end, closed on the right end, and partially
closed in part of the sheet. Let the strain energy of the bent wall and the
compressed posts be denoted by WD , the work done by the alveolar gas and
blood in creating a partially collapsed sheet be denoted by ~, whereas the
free energy of the surface be denoted by WCB' (the subscript CB stands for
"chemical bond"). Then if a small increment of the area of contact (adhesion),
bA, occurs, small changes in WCB, WD , ~ would occur. If the sum bWCB +
bWD + b~ were negative, there will be a tendency for the contact area to
increase. At equilibrium, we have

(1)

by which the area of contact Ac can be determined. Furthermore, the collapsed

area is stable if

(2)
6.9 Open and Closed Capillary Sheets in Zone 2 217

J
Eq.A-8
Collapsed h

(a)
~...=--~a-....-it~=~~
L -,

(b)

Cc)
top _ P
A
Y
y
Cd)

Loading

(h : heX h~ Y)
FIGURE 6.9.1 Schematic drawings for analysis of deflection pattern of collapsed inter-
alveolar septa. (a) dotted lines, walls of sheet before collapsing, solid lines, walls of
collapsed sheet; h is sheet thickness; E is collapsed region; L is length of sheet; t1L is
length of transition zone. Posts are drawn as springs, which balance the transmural
pressure, P - PA" (c) elastic characteristics of sheet described by Eq. (6.7:1). a is com-
pliance constant. (d) out-of-balance spring forces causing deflection of wall when
right -hand side of sheet is collapsed. Wall deflection measured from middle line of sheet
is denoted by y. From Fung and Yen (1986), reproduced by permission.

It is unstable if

(3)

When the details were worked out, Fung and Yen (1986) found that Eq. (3)
prevails. Hence if an interalveolar septum in zone 2 starts to collapse, it will
continue to collapse until the whole septum is collapsed. The result is illu-
strated in Fig. 6.9:2. In this figure, the left end of the collapsed septum is
adjoined to two open septa. If one of the open septum started to collapse, it
would continue to do so until the whole septum is collapsed, and so on.
This theoretical result is corroborated by histological evidence obtained
earlier by Warrell et al. (1972), reproduced in Fig. 6.9:3. Dog's lung at zone 2
218 6 Micro- and Macrocirculation

Alveolar gas ~

Terminal
venule

~en < f!

,......
FIGURE 6.9:2 A pulmonary capillary sheet is collapsed and the collapse is arrested by
two open septa at the left end. From Fung and Yen (1986), reproduced by permission.

- "
... ~
~ .•
•..'.t
;
.",.

Ie .
f

.. • ,,

H
lOp
FIGURE 6.9:3 Histological micrographs of dog's lung in zone 2 condition. Specimen
obtained by quick freezing, then fixed and dried at critical temperature. From Warrell
et al. (1972). Reproduced by permission.

flow condition was quickly frozen by pouring liquid freon cooled to liquid
nitrogen temperature. Specimens were taken within several mm below the
pleura, freeze dried, and processed to preserve the geometry. In Fig. 6.9:3, it
is seen that several interalveolar septa are open, several are collapsed, trapping
a few red blood cells in them. Note that those septa that are collapsed did so
as a whole.
Summarizing, it is found that in zone 2 condition, a number of interalveolar
septa connected to the venules may be collapsed, while the remaining septa
6.9 Open and Closed Capillary Sheets in Zone 2 219

are open. Sluicing occurs in the open septa, in which the gates are located at
the junctions with the venules.
A given interalveolar septum may be open or may be closed. To reopen a
closed sheet requires additional energy to recreate the free surface. To make
sure that every sheet is open in a lung, one should impose a large flow while
the lung is in zone 3 condition. If the venous pressure is continuously reduced
from zone 3 condition to zone 2 condition, more sheets will be collapsed as
the venous pressure is lowered. Since the collapsing is stochastic, Fung and
Yen (1986) assumed a normal probability

Ac = F(1 _ e-Ap2/2a2)
(4)
A
to relate the collapsed area, Ae> with the pressure Ap = Pven - PA when Ap < O.
See Figure 6.9:4. Here A is the total anatomical alveolar sheet area, F and (J
are constants. This formula was validated by experimental results which will
be illustrated later in Fig. 6.11:2. The physical meaning of F is the largest
fraction of the alveolar sheet that will be collapsed when the pulmonary
venous pressure is lowered indefinitely. This number is theoretically pre-
dictable when the relationship between the arteries, veins, alveolar ducts and
alveoli is known. A theoretical model of the pulmonary alveolar duct has been
proposed by Fung (1988) and validated. Fung and Yen (1986) predicted an F
derivation of F based on a pentagonal dodecahedral model of the lung

B Pa, ~, PPL
are constant

Zone 2 Zone 3
A

FIGURE 6.9:4 Reduction of area of perfused alveolar sheets due to collapse of capillary
blood vessels and adhesion of endothelial cells in zone 2 condition is directly related
to reduction of flow through lung. A mathematical expression for reduced flow and
reduced alveolar area in region Be (zone 2) is given in Eq. (5). From Fung and Yen
(1986), reproduced by permission.
220 6 Micro- and Macrocirculation

parenchyma equal to 0.156. Experimental data yield F = 0.104 ± 0.016 (SE),

(f = 4.45 ± 0.45 (SE).

Since in zone 2 blood flow exists in those capillary sheets that are open, we
see that Eqs. (15) and (17) of Sec. 6.7 remain valid if the sheet area A in those
equations is replaced by the open area A - Ac. It follows that in zone 2

Ac) SA
Flow = ( 1 - A 4llkf£2rx (hart - hven),
4 4
(5)

with Ac/A given in Eq. (4).

6.10 Synthesis of Micro- and Macrocirculation in the Lung

The circulation of the whole lung is the sum of flow through all of its parts.
But an analysis of a circuit can be done only if one knows the circuit. Anatomic
data in the form of Strahler system (Sec. 6.6) do not specify the circuit
completely. It is, however, consistent with anatomical observation to assume
that the vascular tree can be treated with the successive generations in series,
but within each generation several possible topological arrangements can be
specified. Using this simplifying assumption, Zhuang, Fung, and Yen (1983)
presented a detailed analysis of flow in zone 3 condition; and Fung and Yen
(1986) presented a detailed analysis of flow in zone 2 condition. Experimental
validation of these results were published by Yen et al. (1984, 1986).
The theoretical procedure of synthesis begins with the capillaries, in which
the pressure-flow relationship are given by Eqs. (15)-(25) of Sec. 6.7, and Eq.
(5) of Sec. 6.9. Now consider the first generation ofthe arterioles and venules,
in which the Reynolds and W omersley numbers are smaller than unity. These
vessels are elastic, so the analysis presented in Sec. 5.11 applies and Eq. (5.11.6)
can be used. In larger vessels with Reynolds number greater than 1, the effect
of kinetic energy change on the pressure drop must be considered. The result
is given in Eq. (5.11:13). In general, the needed correction for kinetic energy is
quite negligible in pulmonary arteries and veins.
With these equations, we can synthesize all the segments into a circuit. At
the junctions of the vessels of successive orders a finite jump of velocity occurs
because of a step change in cross-sectional area. The average velocity qn in a
pulmonary artery of order n is greater than that in the next order qn-l (the
order number being counted from capillary upward). Hence, according to
Bernoulli's equation, there is a sudden jump of pressure equal to
(4)
Similarly, at a junction of pulmonary veins of orders n - 1 and n, there is a
corresponding sudden drop of pressure. In addition, there is the effect of entry
flow and exit flow, which, however, can be taken into account by correcting
the apparent coefficient of viscosity of the blood.
Calculations using detailed anatomic and rheological data following the
6.11 Validation of the Flow Model 221

method just outlined were done by Zhuang et al. (1983). Some results of the
cat's lung are illustrated in the next section.

6.11 Validation of the Flow Model

Yen et al. (1984,1986), Fung and Yen (1986) perfused cat's lung and obtained
the experimental results shown in Figs. 6.11:1 and 6.11:2. Figure 6.11:1 refers
to zone 3 condition. Figure 6.11:2 refers to zone 2 condition. In these experi-
ments the pressures in the airway, PA' and the largest artery, Pa' were fixed,
whereas the pressure in the left atrium, Pv, was varied, first from a higher value
downwards, reaching a minimum, then back upward. The lungs were "pre-
conditioned" by giving it a few cycles oflarge flow in zone 3 condition to open
up all vessels. Note that in zone 2 condition (Fig. 6.11:2), the flow reached a
peak at a certain value of Pven' then decreased with further decrease of Pven' A
hysteresis loop exists under cyclic change of Pven'
These experimental results are compared with theoretical predictions in
Figs. 6.11:1 and 6.11:2. In the theoretical calculation, it is necessary to specify
the total area of the alveolar sheets. This area was measured from histological

THEORY
50 o--{) EXPERIMENTS

40

,-..
30
Eli
'-../

20
~
0
...J
U.

'0 10

o~------~--------~------~--------~------~
o 5 10 15 20 25

PULMONARY VENOUS PRESSURE (cmHf> )

FIGURE 6.11:1 Pulmonary blood flow in zone 3 condition. Cat right lung.
222 6 Micro- and Macrocirculation

6 (ml::~k9) - - THEORY
60 0- - -0 EXPERIMENT
Pa • 20 cmH:P
Q------ PA = 10 cmH20

40

20

-20 -15 -10 -5 o 5 10 15 20 25

It PULMONARY VENOUS PRESSURE (cmH:p)

FIGURE 6.11:2 Comparison oftheoretical and experimental results. Theoretical curves

were based on data given in Fung (1984), and A = 0.84 m 2 for half lung. In zone 2
condition, Eqs. (6.9:4 and 5) were used. Curve fitting yields (J = 5.3 and F = 0.093. In
return stroke, it is assumed that the collapsed alveolar sheets are not reopened, unless
the pulmonary arterial pressure is increased under zone 3 condition. From Fung and
Yen (1986), reproduced by permission.

sections by stereological methods, but the standard deviations of the measured

values were large. The origin of the large standard deviation is partly due to
biological variations from one animal to another and from one location in
the lung to another, and partly due to finite sampling, and distorsion caused
by histological preparations, especially shrinking by fixation agents. Since the
available data of the total area [A in Eqs. (6.7:18) and (6.9.5)] are not very
precise, we have selected a value of A that makes the theoretical prediction
coincide with an experiment result at one point on the p-Q curve as noted in
Figs. 6.11:1 and 6.11:2. This value of A was found to lie in the range of
experimentally determined values. If A is varied, the theoretical curve would
move up and down since the flow Qis directly proportional to A. Thus it is
seen that the experimental trend is well predicted by the mathematical model.
In the above, the pressure-flow relationship is validated in steady flow. Of
course, pulmonary blood flow is pulsatile. The features of wave propagation
discussed in Ch. 5 have been shown by Bergel and Milnor (1965), Wiener et
al. (1966), Milnor (1972). Waves are significant in large pulmonary arteries and
veins. In pulmonary capillaries pulsatile flow is recognizable but can be treated
as quasi-static. If we consider the time-averaged pressure-flow relationship,
then we obtain an equation that relates the static and dynamic pressure,
gravitational potential, and energy dissipation. The same equation is obtained
by considering a steady flow. Hence, although steady flow does not occur in
mammalian lung in life, it is still meaningful to consider it theoretically and
experimentally.
Problems 223

In a recent book edited by Will et al. (1987), other successful approaches to

the analysis of the dynamics of blood flow are reviewed. Among them are the
electric circuit analog and occlusion methods by C.A. Dawson, J.H. Linehan,
T.A. Bronikowski, and D.A. Rickaby, the starling resistor interpretation by
W. Mitzner and I. Huang, gravity nondependent distribution by T.S. Hakim,
R. Lisbona, and G.W. Dean, and models of active regulation by B. and
C. Marshall. This reference serves as a convenient reference to recent
literature.

Problems

6.1 Effective diffusivity of platelets in blood with red blood cells. The concept of mole-
cular diffusion can be extended to fluid containing larger particles. In blood, the
local fluid motion generated by individual red cell rotation will lead to greater
random excursion of platelets and thus enhance its diffusivity. Consider a red cell
of radius a. The angular speed (j) will be proportional to the macroscopic shear
rate y. A small amount of nearby fluid will be carried along with the cell move-
ment (due to the no-slip boundary condition at the red cell surface), leading to
increased mixing, and therefore an increase of effective diffusivity. Write the
effective diffusivity, D., as De = D + Dp , where D is the Brownian molecular
diffusion coefficient and Dp is the rotation induced "diffusion" coefficient. Use
dimensional analysis, show that Dp may be written as Dp = Ca 2 y in which C is a
constant, a is the red cell radius, y is the shear rate. (Cf. Keller, K.H. (1971), Effect
of fluid shear on mass transport in flowing blood. Fed. Proc., 30, 1591.) Keller
estimated that for y = 500 sec-l, Dp is about 10- 5 cm 2 /sec in normal whole blood.
This may be compared with the values of molecular diffusivity D of the following
blood components:
O2 D ~ 10- 5 cm 2 /sec

Protein D ~ 10- 7 cm 2 /sec

Platelet D ~ 10- 9 cm 2 /sec.

Thus the effect of red cells is small for O 2 , large for protein, very large for platelet.
6.2 Assuming a Poisseuille velocity profile and using the equation suggested in Prob.
6.1, determine the distribution of effective diffusivity of blood with platelets over
the cross section of the blood vessel (i.e., as a function of the radial distance).
6.3 In a flowing bloodstream, the red blood cells and platelets are distributed differ-
ently over the cross-section ofthe blood vessel if the vessel diameter is much larger
than the red cell diameter. Explain why? (Cf. Eckstein, E.C. (1982) Rheophoresis-
broader concept of platelet dispersivity. Biorheology 19: 717.)
6.4 Flow of a fluid carrying solid particles (such as coal slurry) may be an economical
way of transporting the solid over a long distance. The economy depends on the
rheology of the slurry. Discuss the factors that are important to the economy of
transporting solids this way.
6.5 For a long time it has been controversial whether the pressure and blood flow in
capillary blood vessels of most organs are significantly pulsatile or not inspite of
224 6 Micro- and Macrocirculation

their being driven by the heart. Modern evidence is that they are. The pulse wave
velocity alters from a value in the order of 1 m/sec in large arteries to a value of the
order of 1 cm/sec in microvessels. Can you ascertain this from a theoretical point
of view? (Cf. Salotto, A.G., Muscarella, L.F., Melbin, 1., Li, I.K.I., Noordergraaf,
A.: Pressure Pulse Transmission into Vascular Beds. M icrovasc. Res., 32: 152-163,
1986.)
6.6 To study coronary blood flow, a logical approach is to take steps to identify the
geometry of the vascular system (branching pattern, branching number ratio,
diameters, lengths, and wall thicknesses of successive generations), the rheological
properties of the blood vessels and blood (constitutive equations, zero-stress
state, hematocrit, blood viscosity,), the governing field equations, the boundary
conditions, the mathematical solutions, and the experimental validation. Make a
literature survey to find out what information is missing and what is the present
status of the science of coronary blood flow. Outline a plan to improve the present
status.
6.7 The study of coronary blood flow discussed in Prob. 6.6 should be extended to
other vital organs such as the brain, eye, ear, kidney, liver, stomach, spleen, and
bone. Make a choice and develop a plan of research.

References

Banister, J and Torrance, R.W. (1960). The effect of the tracheal pressure upon flow: Pressure
relations in the vascular bed of isolated lungs. Quart. J. Exp. Physiol. 45: 352-367.
Bergel, D.H. and Milnor, W.R. (1965). Pulmonary vascular impedance in the dog. Circ. Res. 16:
401-415.
Branemark, P.-1. and Lindstrom, J. (1963). Shape of circulating blood corpuscles. Biorheology
1: 139-142.
Evans, E. and Fung, Y.c. (1972). Improved measurements of the erythrocyte geometry. Microvasc.
Res. 4: 335-347.
Evans, E.A. and Hochmuth, R.M. (1976). Membrane viscoplastic flow. Biophysical J. 16: 13-26.
Evans, E.A. and Skalak, R. (1979). Mechanics and Thermodynamics of Biomembranes. Critical
Reviews in Bioengineering. Vol. 3, issues 3 and 4 (in 2 Vols). CRC Press, Boca Raton, FI.
Fitzgerald, J.M. (1969). Mechanics of red-cell motion through very narrow capillaries. Proc. Roy.
Soc. London B, 174: 193-227.
Fung, Y.c. and Sobin S.S. (1972). Pulmonary alveolar blood flow. Circ. Res. 30: 470-490.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
Fung, Y.c. (1984). Biodynamics: Circulation. Springer-Verlag, New York. In press.
Fung, Y.c. (1988). A model of the alveolar ducts of lung and its validation. J. Applied Physiol.
64: 2132-2141.
Fung, Y.c. and Sobin, S.S. (1969). Theory of sheet flow in lung alveoli. J. Applied Physiol. 26:
472-488.
Fung, Y.c., Sobin, S.S., Tremer, H., Yen, M.R.T., and Ho, H.H. (1983). Patency and compliance
of pulmonary veins when airway pressure exceeds blood pressure. J. Appl. Physiol. Resp.
Envir. Exercise Physiol. 54: 1538-1549.
Fung, Y.c. and Yen, R.T. (1986). A new theory of pulmonary blood flow in zone 2 condition. J.
Appl. Physiol. 60: 1638-1650.
Fung, Y.c. and Zhuang, F.Y. (1986). An analysis of the sluicing gate in pulmonary blood flow.
J. Biomech. Eng. 108: 175-182.
References 225

Horsfield, K. and Gordon, W.I. (1981). Morphometry of pulmonary veins in main. Lung 159:
211-218.
Lee, l.S. (1969). Slow viscous flow in a lung alveoli model. J. Biomech. 2: 187-198.
Malpighi, M. (1661). De Pulmonibus. Letters addressed to A. Borelli. Translated by l. Young
(1930), Proc. Roy Soc. Med. 23: Part 1, 1-14.
Miller, W.S. (1947). The Lung. Thomas, Springfield, IL.
Milnor, W.R. (172). Pulmonary hemodynamics. In: Cardiovascular Fluid Dynamics (D.H. Bergel,
ed.), Vol. 2, Academic Press, New York, Ch. 18, pp. 299-340.
Permutt, S., Bromberger-Barnea, B., and Bane, H.N. (1962). Alveolar pressure, pulmonary venous
pressure, and the vascular waterfall. Med. Thorac. 19: 239-260.
Schmid-Schiinbein, G., Fung, Y.C., and Zweifach, B.W. (1975). Vascular endothelium-leukocyte
interaction: sticking shear force in venules. Circ. Res. 36: 173-184.
Skalak, R., Tozeren, A., Zarda, R.P., and Chien, S. (1973). Strain energy function of red blood cell
membranes. Biophysical J. 13: 245-264.
Sobin, S.S., Tremer, H.M., Fung, Y.c. (1970). Morphometric basis of the sheet-flow concept of
the pulmonary alveolar microcirculation in the cat. Circ. Res. 26: 397-414.
Sobin, S.S., Fung, Y.C., Tremer, H.M., and Rosenquist, T.H. (1972). Elasticity of the pulmonary
alveolar microvascular sheet in the cat. Circ. Res. 30: 440-450.
Warrell, D.A., Evans, l.W., Clarke, R.O., Kingaby, G.P., and West, l.B. (1972). Pattern of filling
in the pulmonary capillary bed. J. Appl. Physiol. 32: 346-356.
Wiener, F., Morkin, E., Skalak, R., and Fishman, A.P. (1966). Wave propagation in the pulmonary
circulation. Circ. Res. 19: 834-850.
Will, lA, Dawson, C.A., Weir, E.K., and Buckner, C.K. (editors). (1987). The Pulmonary Circula-
tion in Health and Disease. Academic Press, New York.
Yen, M.R.T. and Fung, Y.c. (1973). Model experiments on apparent blood viscosity and hemato-
crit in pulmonary alveoli. J. Appl. Physiol. 35: 510-517.
Yen, M.R.T., Fung, Y.c., and Bingham, N. (1980). Elasticity of small pulmonary arteries in the
cat. J. Biomech. Eng., Trans. ASME 102: 170-177.
Yen, M.R.T. and Foppiano, L. (1981). Elasticity of small pulmonary veins in the cat. J. Biomech.
Eng., Trans. ASME 103: 38-42.
Yen, M.R.T., Zhuang, F.Y., Fung, Y.C., Ho, H.H., Tremer, H., and Sobin, S.S. (1983a). Morpho-
metry of the cat's pulmonary venous tree. J. Appl. Physiol. Resp. Envir. Exercise Physiol. 55:
236-242.
Yen, M.R.T., Zhuang, F.Y., Fung, Y.c., Ho, H.H., and Sobin, S.S. (1983b). Morphometry of the
cat's pulmonary arteries. J. Biomech. Eng. In press.
Yen, M.R.T., Fung, Y.C., Zhuang, F.Y., and Zeng, Y.l. (1984). Comparison of theory and
experiments of blood flow in cat's lung. In: Biomechanics in China, Japan, and USA (Y.c.
Fung, E. Fukada, and l.l. Wang, eds.), Science Press, Beijing, China. pp. 240-253.
Yen, M.R.T. and Sobin, S.S. (1986). Pulmonary blood flow in the cat: correlation between theory
and experiment. In: Frontiers in Biomechanics (G.W. Schmid-Schiinbein, S.L.-Y. Woo and
B.W. Zweifach, eds.), Springer-Verlag, New York. pp. 365-376.
Zarda, P.R., Chein, S., and Skalak, R. (1977) Interaction of a viscous incompressible fluid with an
elastic body. In "Computational Methods for FluidcStructure Interaction Problems" (Belytschko,
T. and Geers, T.L., eds.) New York: American Society of Mechanical Engineers, pp. 65-82.
Zhuang, F.Y., Fung, Y.c., and Yen, M.R.T. (1983). Analysis of blood flow in cat's lung with
detailed anatomical and elasticity data. J. Appl. Physiol. Respir. Envir. Exercise Physiol. 55(4):
1341-1348.
Zhuang, F.Y., Yen, M.R.T., Fung, Y.c., and Sobin, S.S. (1985). How many pulmonary alveoli are
supplied by a single arteriole and drained by a single venule. Microvasc. Res. 29: 18-31.
Zweifach, B.W. (1974). Quantitative studies of microcirculatory structure and function. I. eirc.
Res. 34: 843-857, II. Cir. Res. 34: 858-868.
Zweifach, B.W. and Lipowsky, H.H. (1977). Quantitative studies of microcirculatory structure
and function. III. Circ. Res. 41: 380-390.
CHAPTER 7

Respiratory Gas Flow

7.1 Introduction

This chapter is focused on the flow of gas into and out of the mammalian lung.
We study the airway tree shown in Fig. 5.2:2. In the airway, the mixing of
gases is given particular attention. In alveoli, the exchange of O 2 and CO 2
between alveolar gas and red blood cells is discussed. The effectiveness of this
exchange depends on the matching of ventilation and circulation.
In Sec. 7.2 energy loss in various segments of the airway is considered. This
is a major part of the cost man pays to get the gas into and out of the lung.
Although a healthy person does not need to worry about this cost, a sick man
may find the demand excessive.
The subject of gas exchange is treated in detail in Secs. 7.3-7.5. The effect
of convection and diffusion on the gas concentration profile is discussed in
Sec. 7.3. The chemical reactions taking place in the red blood cells are taken
up in Sec. 7.4. The integral equation governing the distribution of ventilation/
perfusion ratio in the lung is given in Sec. 7.5.
In Sec. 7.6 we discuss the pulmonary function tests that are often seen in
clinics. In Sec. 7.7 we consider the analysis of the dynamics of the lung by
computer simulation. Finally, in Sec. 7.8, artificial respiration, especially the
high-frequency-Iow-tidal-volume ventilation, is discussed.
Topics such as the neural and chemical control of breathing and disease of
the respiratory system, however, are omitted. The reader must consult other
works for these topics, e.g. the Handbook of Physiology (Fenn and Rahn, eds.,
1964; Fishman, Macklem, Mead, eds., 1986).
The principal variables and their notations will now be explained. A sketch
of the human chest and lung is shown in Fig. 7.1:1. The lung is enveloped in
a membrane-the pulmonary (or visceral) pleura. The internal surface of the

226
7.1 Introduction 227

FIGURE 7.1:1 A sketch of human chest

and lung.

thoracic cavity is lined with another membrane-the parietal pleura. These

two membranes are joined together at their edges and enclose a space known
as the pleural cavity. Normally, the volume of the pleural cavity is small: the
spacing between the pulmonary and parietal pleura is only a few Ilm. This
space is filled with a fluid whose pressure is the pleural pressure, PPL ' The
pulmonary pleura is acted on one side by PPL' and on the other side by the
alveolar gas pressure,* PALV, and the stress from the lung tissue (J. See the
free-body diagram in Fig. 7.1:2. The difference of the pressures, PAL V - PPL> is
responsible for inflating the pulmonary pleura as a balloon, and is called the
trans pulmonary pressure. The tendency to inflate is resisted by the stresses in
pleura and lung parenchyma. To understand lung inflation, we must know the
stress- strain relationships of the pleura and the parenchyma (Chap. 10). On
the other hand, the movement of the gas from the mouth to the alveoli and
vice versa depends on the difference of the pressure at the mouth PM and the
pressure in the alveoli PALV' The pressure PM is usually atmospheric. This
leaves PPL and PALV as the principal variables for ventilation. In normal
breathing, PPL' PALV and the flow vary with time as illustrated in Fig. 7.1:3.
Although in Fig. 7.1:3 the pleural pressure PPL is represented by a constant
at any given time, it actually varies from place to place. The spatial variation
of the intrapleural pressure can be tolerated because the intrapleural space is
very narrow, hence forming a high resistance channel for intrapleural fluid
motion. PPL is generally negative (relative to the atmospheric pressure, which
is, by convention, considered to be zero) but is more so in inspiration than in
expiration. In forced expiration PPL may take on large positive values.

* We use the standard notations PA for airway pressure, P. for arterial pressure, Pv for venous
pressure, PPL for pleural pressure, PALV for alveolar gas pressure. Following the trend of the
literature, we use em H 2 0 as a measure of pressure. 1 em H 2 0 at 4°C is approximately 98 Nm- 2 .
228 7 Respiratory Gas Flow

FIGURE 7.1:2 The forces acting on an element of the pleura of unit area. PPL is the
intrapleural pressure. PALV is the alveolar gas pressure. S is the shear stress due to
relative motion of the pulmonary and parietal pleurae. T is the tensile force per unit
length of the interalveolar septa. (J is the sum of the septal tension per unit area of the
pleura, i.e., the normal stress in the lung tissue acting on the pleura in the direction
normal to the pleura. (J = (I:Tcos (}dL)/area, wher dL is the length of the alveolar septa
intersecting the pleura, () is the angle between T and the normal vector of the pleura,
and the product T cos () dL is summed over the intercept of the interalveolar septa on
the pleura in the area named.

-5£---------.------T~~
Pleural
Pressure -6
em H 2 0

-~~
Alveolar
pres sure
em H2 0

Air flow
l/s
~L¥=l
-0.5

Tidal
volume
l. FIGURE 7.1:3 The variation of pres-
sures in the pleura (ppd and alveoli
(PALV) and flow in the airway with
Time, See time in normal breathing.
7.2 Gas Flow in the Airway 229

PPL is often measured in the esophagus which is also exposed to the pleural

cavity. In making such a measurement, an elongated balloon is inserted into

the esophagus. The balloon is inflated to such a degree that both the balloon
and the esophagus remain buckled (collapsed) circumferentially so that the
elastic resistance to inflation is negligible, and the internal and external
pressures balance each other. Helium is used to inflate the balloon in order
to minimize the inertial effects of the gas and improve the frequency response
characteristics of the instrumentation system.
The alveolar gas pressure is also nonuniform in the lung, but normally the
degree of nonuniformity is small. The alveolar gas pressure cannot be mea-
sured with transducers in life. Therefore, it is usually inferred by calculation.
In normal breathing the alveolar gas pressure is practically the same as the
atmospheric pressure; the difference PM - PALV does not exceed a few cm H 2 0.
In breath holding, or in sneezing, PM and PALV can be very different from the
atmospheric pressure.
The study of the relationship between the gas pressure p, the volume flow
rate V in different parts of the airway, the distribution of PPL over the pleura,
the distribution of the alveolar gas pressure PALV, the partial pressures of CO 2 ,
O 2 , and N 2, and the parenchymal stress (J in various parts of the lung tissue
is the main objective of respiratory mechanics.

7.2 Gas Flow in the Airway

The airway begins at the mouth and nose, extends through the larynx, trachea,
bronchi, and ends at the alveoli. Detailed measurements of the geometry of
the bronchial system have been made by Weibel, Cumming, and their associ-
ates. Weibel (1963) presented measurements of the length, diameter, and area
of successive segments of the airway of five normal human lungs. He showed
that each branch gives rise to two daughter branches, which may vary con-
siderably in length, but less significantly in diameter. Weibel presented his
data in two forms, a model A, with regularized dichotomy, and a model B,
that takes into account the irregular dichotomy. In model A, Weibel took
certain statistical mean values of length and diameter of each generation and
presented data of the human airway which are reproduced in Table 7.2:1. The
trachea is called generation 0, the main bronchi generation 1, and so on.
Generations 0-16 are called conducting airways. Airways of generation 16 are
called the terminal bronchioles. Those of generatitms 17-23 have alveoli on
their walls, and constitute the so-called respiratory zone. Generation 23 termi-
nates in alveoli.
The total cross-sectional area of each generation (i.e., the total area of all
branches of a given generation) varies with the generation number n. When
n> 3 the area increases with increasing n (see Table 7.2:1), and therefore
increases very rapidly as the distance from the mouth increases. As a con-
sequence, the velocity of flow decreases very rapidly toward the alveoli. The
 tv
v.>
o

TABLE 7.2:1 Archiecture of the human lung according to Weibel's (1963) model A, with regularized dichotomy
At flow of 1 l/sec
Cum. Cum.
Diam. Length Length Area Vol. Vol. Speed Reynolds
Generation Number (mm) (mm) (mm) (cm2) (ml) (ml) (cm/sec) No.
Trachea 0 18 120.0 120 2.6 31 31 393 4,350
Main bronchus 1 2 12.2 47.6 167 2.3 11 42 427 3,210
Lobar bronchus 2 4 8.3 19.0 186 2.2 4 46 462 2,390
3 8 5.6 7.6 194 2.0 2 47 507 1,720
Segmental bronchus 4 16 4.5 12.7 206 2.6 3 51 392 1,110

5 32 3.5 10.7 217 3.1 3 54 325 690

Bronchi with 6 64 2.8 9.0 226 4.0 4 57 254 434
cartilage 7 128 2.3 7.6 234 5.1 4 61 188 277
in wall 8 256 1.86 6.4 240 7.0 4 66 144 164
9 512 1.54 5.4 246 9.6 5 71 105 99 -..J

~
10 1,020 1.30 4.6 250 13 6 77 73.6 60 (1)
en
Terminal bronchus 11 2,050 1.09 3.9 254 19 7 85 52.3 34 '9.., .
!»
..,0-
12 4,100 0.95 3.3 257 29 10 95 34.4 20 '<
...>,
oj Cl
Bronchioles with 13 8,190 0.82 2.7 260 44 12 106 23.1 11 !»
en
-g~
o ::l muscle in wall 14 16,400 0.74 2.3 262 70 16 123 14.1 6.5 21
U.o
<1) 0 0
Vl_ 15 32,800 0.66 2.0 264 113 22 145 8.92 3.6 ~
 Terminal bronchiole 16 65,500 0.60 1.65 266 180 30 175 5.40 2.0 ....,
N
resp. bronchiole 17 131 x 10 3 0.54 1.41 267 300 42 217 3.33 1.1 Cl
IU •
'"3 resp. bronchIOle 18 262 x 10 3 0.50 1.17 269 534 61 278 1.94 0.57 ~
.D en '"r1
..9 @ resp. bronchiole 19 524 x 10 3 0.47 0.99 270 944 93 370 1.10 0.31 0"
» .~ :E
~ g s·
8 ::; alveolar duct 20 1.05 X 106 0.45 0.83 271 1,600 139 510 0.60 0.17 ;.
et alveolar duct 21 2.10 x 106 0.43 0.70 271 3,200 224 734 0.32 0.08 "?:
alveolar duct 22 4.19 x 10 6 0.41 0.59 272 5,900 350 1,085 0.18 0.04 ....
:E
I»
alveolar sac 23 8.39 x 106 0.41 0.50 273 12,000 591 1,675 0.09 '<

alveoli, 21 per duct 300 x 106 0.28 0.23 273 3,200 4,800
Notes: (1) Area = total cross sectional area, (2) cum. = cumulative, (3) Dead space, approx. 175 ml + 40 ml for mouth.

tv
!,,;.)
,....
232 7 Respiratory Gas Flow

estimated order of magnitude of the velocity of flow and the Reynolds number
are shown in Table 7.2:1 for a flow rate of 1,000 mljsec.
With Strahler's method (see Sec. 6.6), Cumming and Semple (1973) obtained
17 orders of branches between the terminal bronchiole and the trachea.
Taking the terminal bronchiole as order 1, they obtained a branching ratio
of 2.74, and a diameter ratio of 1.37 between orders 1-17.
The pressure and velocity in the airway vary from point to point. Often the
secondary flow is very strong. An approximate way to analyze the flow in such
a system is to use energy equation (S;-t4:6). The principal variables are p, the
velocity-weighted average pressure, q2, the velocity-weighted average of the
square of the speed, Q, the flow rate, and ~, the dissipation junction, defined
in Eqs. (S.14:7), (S.14:S), and (S.14:4) respectively. In application to the lung, Q
is the gas flow rate, which will be written as V below. Of the terms in Eq.
(S.14:6), the terms representing the potential energy due to the weight of the
gas,/'>.. pgh2 -~ pghl' can be neglected in the lung. The kinetic energy terms
!pqi - !pq~ are important at the bifurcation points because significant
velocity changes take place from one generation to another; but the last
term, representing the transient change of kinetic energy within a segment, is
of minor significance. Hence, in application to the airway, Eq. (S.14:6) can be
simplified to

(1)

where
/'-. = ---;-If
PI pudA,
V A
(2)
~ = Iv (Jij V;j dv,

the integrals being taken oW the cross sectional area A and volume of
segment V. Note that p,
q2 are averages weighted with axial velocity
component u. They are not equal to the average pressure and velocity p and
fi defined by

fi = ~f
AA
udA. (3)

The latter averages p, fi are used in Sec. 7.7 where attention is given to the
airway elasticity in simulating the whole airway system. ~ is the dissipation
function. In order to understand the pressure drop in the airway, major
attention must be focused on the dissipation function.
The dissipation function ~ is the volume integral of the product of
the coefficient of viscosity J1 and the square of the shear strain rate. For a
Poiseuillean flow (laminar flow) in a long circular cylindrical tube, the rate
of energy dissipation ~ in a segment of length L is (see Sec. S.lO):
(4)
7.2 Gas Flow in the Airway 233

u
where U- is the mean velocity of flow, Eq. (3). (For a laminar flow, 2 = 2u- 2 •
See Problem 7.4 infra.) For other flows, one may represent the dissipation
function as the product of f!)p and a factor Z:
(5)
which defines Z. This Z is the same Z(NR ) ofEq. (5.15:10). Examples are given
below.

Entry Flow Into a Tube

Pedley et al. (1977) computed f!) for a flow entering into a circular cylindrical
pipe from a reservoir on the basis of boundary layer theory, and obtained

Zentry
0(1 (d
= 16 L NR
)1/2 , u-d
whereNR = - . (6)
v
The variable d is the diameter of the pipe, v is the kinematic viscosity, U- is the
mean speed of flow, N R is the Reynolds number, and 0(1 is a numerical constant
which is equal to 1 if the velocity profile in the boundary layer is linear and
4/3 if it is parabolic.

Turbulent Flow in a Tube

When the flow rate V exceeds a certain value, the flow becomes unstable and
fluctuation develops. Further increase in flow rate leads to turbulence. In a
fully developed turbulent flow, a well-known empirical formula is

ZT = f!)tu;ulent = 0.OO5Ni'4. (7)

p

The critical Reynolds number for transition from laminar to turbulent flow
in a pipe depends on the frequency ofthe pulsatile flow. At normal breathing
frequency the transition Reynolds number is about 2,300. Data shown in
Table 7.2:1 suggest the existence of turbulence in upper airway of man in
normal breathing. In inspiration, turbulent flow is expected also in smaller
airways where N R is less than 2,300 because disturbances will be carried down
with the flow, decaying rather slowly.

Branching Tubes
Experiments on branching tubes have shown the existence of secondary flow.
Pedley et al. (1977) observed that for inspirational flow the new boundary layer
that develops on the flow divider remains thin for some distance downstream,
and extends around at least half the circumference of the daughter tube. Most
of the dissipation takes place in that boundary layer. They suggest that the
dissipation mechanism is similar to that of the entry flow; hence
d )1/2
Z=y ( L NR , (8)
234 7 Respiratory Gas Flow

where y is a numerical constant, L is the length of the tube, d is the diameter,

and N R is the Reynolds number. The average value of y was found to be 0.33
for laminar flow at Reynolds numbers between 100 and 700. If the formula
yields a value of Z less than 1, it should be ignored and replaced by Z = 1
because Poiseuillean flow yields the minimum dissipation in any tube.

Upper Airways
The flow in the nose, mouth, larynx, and trachea is very complex. The
narrowest passage in the upper airway is the laryngeal constriction. About
half of the viscous resistance to breathing arises in the upper airway; half of
that resistance is from the laryngeal constriction. Upon inspiration the flow
will separate from the larynx and form a turbulent jet in the trachea. With
expiration, separation occurs again and causes a confused flow in the mouth
and nasal passages. The critical Reynolds number at which the inspiratory
flow becomes turbulent in the trachea is about 500, whereas for an expiratory
flow it is about 1,500.
The laryngeal constriction is elastic and its aperture varies with lung
volume, flow rate, and the frequency of breathing. When the lung volume is
increased, the aperture increases and the resistance to flow decreases. Panting
enlarges the larynx aperture and reduces the flow resistance.
Jaeger and Matthys (1968/69) expressed their experimental results on
pressure drop between mouth and trachea with the formula
(9)
where dp is the pressure drop, V is the volume flow rate, and a and Care
constants given in Table 7.2:2. It is interesting to note that except for the very
dense gas mixture, the exponent a is close to 1.5, which is predicted for the
entry flow by use of Eqs. (6), (5), and (4). Pedley et al. (1977) suggested that the
higher exponent (close to 2) for the dense gas sulfur fluoride SF6 is due to the
higher Reynolds number obtained with such a gas at normal flow rate, so that
most of the energy dissipation stems from the turbulent jet issuing from the
larynx. Then ZT of Eq. (7) applies, and an exponent of 2 is obtained by use of
Eqs. (5), (7), and (4).

TABLE 7.2:2 Densities, viscosities, kinematic viscosities, and best fit values
of a and C, in the experiments of Jaeger and Matthys (1968/69)
Gas
mixtures p (g cm- 3 ) Jl (g cm- 1 S-1) v (cm 2 S-1) a C
He 0.45 x 10- 3 2.05 X 10- 4 0.46 1.42 0.35
O2 1.1 x 10- 3 2.07 X 10- 4 0.18 1.55 0.74
Ne 0.88 x 10- 3 2.92 X 10- 4 0.33 1.36 0.82
SF6 4.2 x 10- 3 1.7 X 10- 4 0.04 1.92 1.77
Reprinted from T.J. Pedley, R.c. Schroter, and M.F. Sudlow, Gas flow and mixing
in the airways, in Bioengineering Aspects of the Lung (J.B. West, ed.), pp. 163-265,
by courtesy of Marcel Dekker, Inc., 1977.
7.2 Gas Flow in the Airway 235

Further along the airway tree, the bronchioles divide into terminal bron-
chioles, then into respiratory bronchioles, alveolar ducts, and alveoli. In
bronchioles of diameters less than 0.05 em, the Reynolds number of flow is
smaller than 1 even at high levels of ventilation. Then the inertial force of the
fluid becomes negligible and the pressure drop is proportional to the flow rate.
The contribution of this region to the overall airway resistance is negligible.

Static Pressure Variation Along the Airway Tree

Pedley et al. (1977) calculated the pressure distribution in the entire airway
during inspiration. They used Eq. (8) with an average value of y = 0.33 for all
generations of the airway from bronchi to alveoli, and Eq. (9) for the trachea,
assuming a fully developed turbulent flow there. They calculated the term f!fi IV
in Eq. (1), called it the viscous pressure drop, and denoted it by dpv' They also
calculated the kinetic energy term involving 42 in order to compute the actual
static pressure at every location on the bronchial tree. They assumed Weibel's

125
• • Real Viscous Pressure
0 Poiseuille Viscous Pressure

•
100

0:;-
•
I
E
~ 75
OJ

~
OJ
•
a':
'"0
:J

~
;; 50
a..' •
•
25 0 0 0
0 •
0
0
0 •
0
•
0
0 5 10 15
Airway Generation

FIGURE 7.2: 1 Predicted variation of viscous pressure down the bronchial tree at an
inspired flow rate of 1.67 Ls- 1 . From Pedley, T.J. et al. (1971), by permission.
236 7 Respiratory Gas Flow

symmetric model (Table 7.2:1), and approximate equality of q and the velocity
component in the axial direction u. Then, from Eq. (2),

~ =~ f u 3 dA. (10)

By examining the empirical data, they took

~
u 2 = 1.7u2 • (11)
Figure 7.2:1 shows their result on the viscous pressure drop Pv at the down-
stream end of each generation along the bronchial tree during inspiration,
plotted against the generation number. The pressure in the alveoli is set at
zero. The figure is plotted for a pulmonary flow rate of 100 liter/min. Note
that the pressure drop L1pv is equal to !?fi/V, and !?fi = Z!?fip; hence when Z is
given by Eq.(8), L1pv is proportional to V3/2 (because u ~ v,!?fip ~ V2 ,!?fi ~ VS/2 ,
and L1pv = !?fi/V). The hypothetical Poiseuillean flow results are also shown.
When the static pressure p is computed from Eq. (1) with the kinetic energy
terms taken into account, the result is shown in Fig. 7.2:2.
A much more complex analysis is required for asymmetric branching. The
flow in each branch then depends on the resistance in every other branch,
both upstream and downstream.

80 • Real Static Pressu re

• o Corrected Poiseu ilie Pressu re

60

40 •
••
N
I
20
•• •
E
z
0...- • 0 ~
1:' 0
;;l 0 5 10 15
in 0
ct Airway Generation
u
0;::; -20 0
co
ci)
0
0
-40
0
-60
0

-80

FIGURE 7.2:2 Predicted variation of static pressure down the bronchial tree at an
inspired flow rate of 1.67 Ls- 1 . From Pedley et al. (1971).
7.3 Interaction Between Convection and Diffusion 237

Recapitulation
Among all possible flows of a given rate in a circular cylindrical tube, the
Poiseuille flow has the least energy dissipation. The energy loss due to any
other flow of the same rate can be expressed as a product of a factor Z times
the Poiseuillean loss. Z is a function of the pipe geometry, the entry condition,
the branching pattern, the Reynolds number, the Stokes number (or its square
root, the Womersley number), and the turbulence level.
The use of the Z factor outlined above is no more than a simple way of
summarizing the results of some detailed fluid mechanical analysis. Obviously,
more definitive detailed theoretical and experimental studies of the flow in the
airway are needed. Secondary flow, separation, converging and diverging
flows of expiration and inspiration in real models of the airway, the effect of
flexible walls, the effect of smooth muscle contraction, mucus layer, etc., must
be studied. What is known is minuscule compared to what needs to be known.
This is an excellent field for computational fluid mechanics.

7.3 Interaction Between Convection and Diffusion

In each breathing cycle, we draw in fresh air and expel lung gas through the
same airway. The interface of the old and new gases is somewhat blurred
because of diffusion and convection. The nature of this interface is considered
below.
First consider a steady laminar flow of a gas A in a long straight tube, as
depicted in Fig. 7.3:1. According to Poiseuille, the velocity profile is parabolic.
Imagine that at a certain time, a material B is added to A to the left of a certain
section Xo and that the interface between A and B is flat at this instant of time.
If the solubility of B in A is zero, then at a later time this interface will be
stretched into a paraboloid, with a vertex moving at twice the average velocity
of the fluid, while the fluid in contact with the wall does not move because of
the no-slip boundary condition (see Sec. 1.7). For this nondiffusible case the
stretched interface at successive instants of time t 1, t 2 , t3 is shown by dashed
lines in Fig. 7.3:1, and the concentration of B, averaged over a cross section,
would vary with the distance x as shown by the dashed line in the graph at
the bottom of Fig. 7.3:1. If, however, the material B is diffusible (soluble) in A,
then molecular diffusion will take place to modify the interface. At time tl (see
figure) the concentration gradient of B is essentially pointing in the longi-
tudinal direction. At t2 the concentration gradient of B in the radial direction
becomes significant on the interface and diffusion in the radial direction takes
place. At t3 the average concentration of B will have a spatial distribution as
shown by the solid line in the graph at the bottom of Fig. 7.3:1. The movement
of the diffusible material B from the fast flowing core fluid to regions of lower
velocity changes the stratification of material B, and the effective front of the
material B is not located at the vertex of the paraboloid defined by convection.
238 7 Respiratory Gas Flow

Velocity Profile

average
100%

c oncentration
-- --
Interface with diffusion

_~ no diffusion
of B at t3 j - -'- - .
Effective front of B : - -~
_0_-'--_____ __ . _ _. __ ~ .. _... -:...--::::-:-
x Distance x tip
o

FIGURE 7.3:1 The progressive dispersion of a diffusible material B injected suddenly

at time t = 0 into a long tube in which a fluid A flows steadily. Successive drawings
show the concentration profile of material B as time increases from t = 0 to t!, t z , t3'
The last figure shows the variation of the concentration of the diffusible substance B
with the distance along the tube. The "tip" indicated on the horizontal axis is the tip
of the convective flow profile. The concentration shown is the average value in each
cross section.

In fact, it is shown by Taylor (1953) that at large t the effective front of the
material B moves at a velocity equal to the mean velocity of flow (i.e., at
one-half of the velocity on the center line). Regarding the concentration
distribution of the material B about this effective interface, Taylor shows that
the resulting longitudinal mixing can be described as a process of diffusion,
with an apparent diffusivity ~app (sometimes called the dispersion coefficient)
which is very much larger than the molecular diffusivity D. He finds that ~app
varies inversely with the molecular diffusivity (D), and directly with the square
7.3 Interaction Between Convection and Diffusion 239

of the product of the average velocity (U) and the tube diameter (d); thus
(Udf .
.@app = 192D' (cIrcular tube). (1)

The effective diffusivity is increased as molecular diffusivity decreases be-

cause the radial mixing is less rapid if molecular diffusivity is small . .@app also
increases rapidly with an increasing flow rate and tube diameter.
This analysis also holds qualitatively for turbulent flow, because turbulent
mixing is somewhat similar to molecular diffusion. Applying this concept to
the gas flow in the airways, we see that the O 2 and CO 2 concentrations at the
interface between the newly inhaled and alveolar gases are gently stratified in
the airway during inspiration. As this stratified gas flows into smaller and
smaller airways (bronchioles), U and d decrease rapidly, then the apparent
diffusivity .@app becomes very small, and the stratification pattern becomes
frozen. Beyond the terminal bronchiole the effect of convection is negligible,
and molecular diffusion takes over as the mechanism to spread O 2 and CO 2
toward the flowing blood (Sec. 7.4).

Mathematical Analysis
The qualitative analysis above is exciting, but it may appear hard to believe
at first sight. Thus, consider the component A in the mixture A + B to the left
of the interface. On the centerline of the tube, the mixture is moving at twice
the velocity of the interface. The component A mixed with B becomes purified
again on passing through the diffusion front. How is this done? To remove
any feeling of mystery, it is best to reproduce Taylor's mathematical analysis
here. Let us consider the case of a sudden injection at one station of a solute
B into a steady flow in a channel-a situation similar to that shown in Fig.

°
7.3:1 with the exception that at time zero the solute B is concentrated at the
section x = (i.e., the concentration of B is a delta function of x), see the third
panel of Fig. 7.3:2. When this case is solved, we can obtain the response to a
step function injection as shown in Fig. 7.3:1 by a convolution integration.
To obtain the solution, we first need the basic equations governing the
movement of the solute B. Let C be the concentration of the solute, D be the
coefficient of molecular diffusion of the solute in the fluid A, V be the gradient
vector, and D be the velocity vector. Then the solute flux vector (mass flow
rate of the solute) which is due to molecular diffusion and convection, is
flux = -DVc + DC. (2)
The equation of balance of solute flow, describing the conservation of mass
of the solute, can be derived by considering the influx and outflux through the
surfaces of a control volume, and the rate of change of the mass of the solute
in the volume (see Fung, 1977, Chap. 10, p. 250). The result is
oc
- ot = V·[ -DVc + DC]. (3)
240 7 Respiratory Gas Flow

permeable
walls
MODEL impermeable

impermeable
L /
impermeable

VELOCITY PROFILE
___ ~ _____ \\ _~. ____I~-~:___ ____ ~'J ___

TRACER PROFILE

CONCENTRATION PROFILE
£\
FIGURE 7.3:2 Impulsive injection of a diffusible material at station x = 0 at time = O.

If the coefficient of diffusion D is a constant, independent of both the con-

centration and the spatial coordinates, and if the fluid is incompressible so that
V'U =0, (4)
then Eq. (3) becomes

(5)

Let us first solve this equation for the two-dimension channel illustrated
in Fig. 7.3:1 or 7.3:2. Assume that the walls are impermeable to both the fluid
A and the solute B. Let the coordinate axis x be the center line of the channel
and the z axis be perpendicular to it. Let the components of velocity in the
directions of x and z be u and w, respectively, whereas the third component
perpendicular to the paper is zero. Then Eq. (5) can be written as
oc oc oC (02C 02C)
at + u ox + oz = D ox2+ OZ2 .
W
(6)

In the case illustrated in Fig. 7.3:2, a diffusible solute B is injected at the

station x = 0 impulsively at time t = 0 and uniformly over the cross section.
Assume the flow to be steady except for the diffusion of the solute. We shall
7.3 Interaction Between Convection and Diffusion 241

look for an asymptotic solution of the solute distribution after the elapse of
a sufficiently long period of time following the injection. As the asymptotic
condition is approached the velocity component u consists of a constant mean
velocity U plus a deviation u'(z) which is a function of z only; whereas the
transverse velocity w vanishes. The solute B will have spread over a con-
siderable length of the tube. Since the tube is long and narrow the rate of
change of the solute concentration in the longitudinal direction will be much
smaller than that in the transverse direction: hence iPejax 2 is negligible
compared with a2ejaz 2. Hence at the asymptotic condition Eq. (6) is approxi-
mated by
ae ae a2 e
at
+ [U + u'(z)] ax = D az2' (7)

To remove U from this equation we introduce a set of moving coordinates ~

and ( with the origin moving at velocity U:
x = ~ + Ut, z = (. (8)
Then Eq. (7) becomes

(9)

Now let the average value of the concentration on the centerline z = 0 be

denoted by c(~, t), which is not a function of z. Let the difference between
e(~, z, t) and c(~, t) be e':

e(~, z, t) = c(~, t) + e'(~, z, t). (10)

Substituting this into Eq. (9) yields
a a a2 e'
at (c + e') + u'(z) a~(c + e') = D az 2 . (11 )

Now Taylor argues that e' is essentially a function only of z whereas c is a

function only of ~, so that the equation above is reduced to
ac a2 e' (12)
u'(z) a~ = D az2 .

The validity of this assumption, i.e.,

ae'
-«-
ac
(13)
a~ a~'
can be verified a posteriori. The boundary conditions for Eq. (12) are that
e' = 0 when z = 0 (by definition of c) and ae'jaz = 0 at an impermeable
boundary (z = ± hj2 in Fig. 7.3:2). Hence the integration of Eq. (12) gives
ae'
-a =
IZ 1
-
ac
ay; u'(z')dz', (14)
z -h/2 D ..
242 7 Respiratory Gas Flow

which satisfies the boundary condition at z = - h/2. A second integration

gives

, f( )oc
c = z oe' (15)

where

f(z) = i (fZU
z

o -h/2
1
-u'(z')dz'
D
) dz". (16)

The rate at which mass is transported through a cross section moving at

the mean velocity is
M=Au'c' (17)
in which A is the cross-sectional area and the over bar indicates a cross
sectional mean. Use of Eq. (15) in Eq. (17) yields
. -oc
M = Au,! oe. (18)

This is of the form

. oc
M= -A.@app oe' (19)

which shows that the dispersion follows Fick's law, with a coefficient of
apparent diffusivity:
.@app = -u'f (20)
Now consider a small segment of the tube of length dx. The inflow of the
solute from the left is M. The outflow from the right is M + (oM/oe) de. The
net outflow per unit time is

ar
oM
de = - A.@app
02C
oe 2 de.
By the principle of conservation of mass this must be equal to the rate of
decrease of the mass of the solute in the segment, (- oc/ot)A de. Hence the
mean concentration is governed by the equation
oc 02C
at = .@app oe 2'
(21)

or, by Eq. (8),

oc oc 02C
at + U ox = .@apPox2· (22)

From this equation we obtain the solution which represents the concentration
7.3 Interaction Between Convection and Diffusion 243

of the solute on the centerline following an impulsive input of unit total mass
at x = 0 and t = 0:
_ 1 1 [ (x - Ut)2]
(23)
c=J(x,t)=2 ~exp 4~ t .
hy n~appt app

The apparent diffusivity, ~app, can be computed from Eqs. (16) and (20).
For a two-dimensional channel of width h with a parabolic flow profile, the
velocity at a distance z from the center line is

U=U o
4Z2) ,
(1- Ji2 (24)

where U o is the maximum velocity at the axis. Hence the mean velocity U and
the deviation u' are

U = 1uo, u' =!U (1 _1~:2). (25)

On substituting into Eq. (16), integrating, and then substituting into Eq. (20),
integrating again, and dividing by h, one obtains the apparent diffusivity
h2 U 2
~app = 210D (channel). (26)

This result refers to a two-dimensional channel. If a corresponding problem

of injecting a solute into a flow in a circular cylindrical tube of radius a is
solved, the velocity u at a distance r from the central line is
U = uo/2, (27)
and one obtains the apparent diffusivity:
a2 U 2
~app = 48D (circular tube). (28)

The spatial distribution of the solute at time t after an impulsive injection at

t = 0, x = 0 is given by the same J(x, t) of Eq. (23).

Arbitrary Injection
Since Eq. (22) is linear, the principle of superposition holds. Hence, if one
injects a solute into a tube at a rate of f(x, t) per unit time per unit length, then
in a small interval of time between rand r + dr and space between and e
e + de the amount of solute f(e, r) de dr may be considered impulsively in-
jected. The concentration distribution due to this injection is
f(e, r) de dTl(x - e, t - r).
By taking all such intervals into consideration, we see that the solute con-
244 7 Respiratory Gas Flow

centration at time t and place x is

c(x, t) = foo dr J:oo d( f((, r)I(x - (, t - r). (29)

Further Extensions
Taylor's asymptotic solution must be supplemented by an investigation of its
range of validity with regard to the conditions assumed in Eqs. (13). Obviously
it is not valid at places too close to the site of injection, nor too soon after
injection. Taylor (1953) has presented an extensive theoretical and experi-
mental investigation of these questions. In Taylor (1954) the analysis is ex-
tended to turbulent flows in pipes which are smooth or rough, straight or
curved. Further research has found the method valid for flow in rivers, and is
used for practical determination of river flow. In physiology, use of solute
injection to determine flow rate and arrival time of blood was initiated by
G.N. Stewart (1893, 1900), who assumed that the tracer flows with the mean
speed offlow of blood (which is only one half of the velocity on the center line
of the blood vessel in a Poiseuillean flow). This is plausible, but the correct
explanation was not known until Taylor presented it in 1953.

Diffusion in Alveolar Duct and Alveoli

In the respiratory bronchioles, alveolar ducts, and alveoli, the Reynolds
number is less than 1, the role of convection becomes insignificant, and
molecular diffusion is the mechanism that spreads O 2, CO 2, N2 (or other
gases) in the alveolar space.
Between the terminal bronchiole and the alveoli of human lung, the total
distance is only about 4 mm. The volume available for the gas, however,
increases very rapidly as an exponential function of the generation number,
from 175 cm 3 to 4,800 cm 3 (see Table 7.2:1). In this space, the O 2 and CO 2
concentrations must change from their values in the terminal bronchiole to
their values on the alveolar wall. On the alveolar wall, rapid exchange of
alveolar gas with blood occurs, (Sec. 7.4). The question of stratification of O 2
and CO 2 concentration in the pulmonary alveolar duct has been investigated
by Chang et al. (1973).

7.4 Exchange Between Alveolar Gas and Erythrocytes

With air brought to the alveoli and blood flowing in the capillary blood
vessels, an exchange between blood and alveolar gas takes place. Now we
must consider the diffusion of alveolar gas to the red blood cells, the chemical
reactions in the red cell, and the spatial and temporal distribution of O 2 and
CO 2 •
7.4 Exchange Between Alveolar Gas and Erythrocytes 245

Diffusion Across a Membrane

Analysis of diffusion across the alveolar wall can be based on the equations
derived in Sec. 7.3. Let C be the concentration (mass per unit volume) of a gas
in the wall, and D be the coefficient of molecular diffusion of that gas in the
wall. The wall is incompressible and the convection in the wall can be ignored.
For a wall (Fig. 7.4:1) of uniform thickness h exposed to a gas of concentration
C 1 on one side and C2 on the other (c 1 and C 2 being constants), Eq. (7.3:5) is
reduced to the following:
(1)

and the boundary conditions are

C = Cb at x = 0; C = C2 , at x = h; (2)
x being the coordinate perpendicular to the membrane. A solution of Eq. (1)
that satisfies the boundary conditions (2) is
C2 - C1
C = C1 + h x. (3)

The flux vector, pointing in the direction of the x-axis, has a magnitude:
flux = -DVc = -Dt(c2 - c 1 ). (4)
The rate of mass flow across a membrane of area A is
liz = A . flux. (5)
In pulmonary physiology, it is customary to express the gas concentration in
a tissue in terms of its partial pressure p and the solubility A of that gas in the
tissue:
C == Ap. (6)
Hence, from Eqs. (4), (5), and (6) we have the rate of diffusion across the
membrane and the rate of volume flow V:
DAA . liz
liz = --h-(P2 - pd, v=-.
p
(7)

C1

O~~~h_______________ x

FIGURE 7.4:1 A diffusion barrier.

246 7 Respiratory Gas Flow

D is inversely proportional to the square root of molecular weight:

1
D '" --;======= (8)
J molecular weight
If we compare the diffusion of O 2 and CO 2 in a membrane, we see that CO 2
will move about 20 times faster because its solubility is much larger whereas
its molecular weight is not very different from that of O 2 •

Diffusion from Alveolar Gas to Red Blood Cells in

a Capillary Blood Vessel (Fig. 7.4:2)
If a foreign gas such as carbon monoxide arrives at an alveolus, the gas will
diffuse through the capillary blood vessel wall, the plasma, and the red cell
membrane, and react with the hemoglobin. A certain amount of CO will be
left in molecular form, while the rest is combined with hemoglobin into a
compound. The gaseous CO will be carried by the red cell to the tissues. Hence
we ask: How rapidly does the partial pressure of CO in the red blood cell rise?
The answer is illustrated in Fig. 7.4:3. The partial pressure rises slowly,
because of the high chemical affinity of CO to hemoglobin.
Now let us assume that the foreign gas that arrived at the alveolus is nitrous
oxide. When this gas diffuses into the red blood cell, no combination with
hemoglobin takes place. As a result the partial pressure of N 2 0 in the red
blood cell can rise rapidly. In the example shown in Fig. 7.4:3 the partial
pressure of N 20 in the red cell has virtually reached the alveolar level after
the cell traversed approximately one-quarter of the path along the capillary.
After this point, practically no additional N 20 is transferred. Thus the amount
ofN 2 0 taken up by the blood depends entirely on the amount of blood flow.
Now, consider oxygen, which reacts with hemoglobin to form
oxyhemoglobin:
(9)
The reaction has a finite speed; in the red cell it takes about! sec to complete.

ALVEOLAR
WALL

O2 -f:~!'f--f----'!- O2 + Hb -+ Hb0 2
-D M- - - - - - (J. V c - ---
'--.-.---
FIGURE 7.4:2 Diffusion of gas
from alveolus to red blood cell in
a capillary blood vessel.
7.4 Exchange Between Alveolar Gas and Erythrocytes 247

FIGURE 7.4:3 CO, N 2 0, and O 2 uptake by Enterng End of capillary

!
red blood cells as they travel in the pul- caplary
monary capillary. From J.B. West, Respira-
tory Physiology- The Essentials, copyright !/AlveOlar level
© 1974, the Williams & Wilkins Co.,
Baltimore.

o 025 0.5 0.75

Tine in capillary. sec

When a red blood cell enters a capillary in the alveolus it normally has an O 2
partial pressure of about 40 mmHg. The partial pressure of O 2 in the alveolus
may be about 100 mmHg. The two compartments are separated by a distance
of only about t Ilm. The diffusion and chemical reaction are therefore initiated
immediately in the red cell. This dynamic process continues for about t sec in
a normal lung until the O 2 pressure in the red cell approaches the alveolar
O 2 pressure. This is depicted in Fig. 7.4:3. In abnormal circumstances, in which
the diffusion properties of the lung are impaired, for example, by thickening
the alveolar wall, the rate of increase of O 2 pressure in the red cell will be
slowed, and it is possible that the blood P0 2 at the end of the capillary will not
reach the alveolar value.
Another situation in which O 2 transfer may become diffusion limited
occurs when the time of transit of blood through the capillary is greatly
shortened (e.g., in vigorous exercise), or when the oxygen pressure in the

FIGURE 7.4:4 CO 2 release by red blood

cells as they travel in the pulmonary capil- o OJ5
lary. After P.D. Wagner and J.B. West
(1972). Tine n capilary. sec
248 7 Respiratory Gas Flow

atmosphere is very low (e.g., on high mountains or at high altitude while flying
in an airplane with an unpressurized cabin.) Then the alveolar P0 2 and the
blood P0 2 are both reduced.
Finally, consider CO 2 , Its partial pressure history in a pulmonary capillary
blood vessel is sketched in Fig. 7.4:4. In a normal lung there is no difficulty in
remvoing blood CO 2 and reducing its partial pressure to that of the alveolar
level, but in diseased states with thickening of blood-gas barrier, this elimina-
tion may have some difficulty.

Measurement of Diffusion Capacity

Equation (7) shows that to analyze diffusion across a membrane, we must
know the constants D, A, A, and h. In a real lung, there is no way to measure
these constants during life. To absorb them in a simplified equation that is
applicable to a whole lung, Krogh (1914) writes
V = DdpI - P2)' (10)
The constant DL is called the diffusion capacity of the lung. This equation can
be applied to each gas. Thus, if O 2 is considered, V is V02 ' the rate of transfer
of oxygen from alveoli to red cells measured in terms of volume at standard
temperature and pressure (O°C, 760 mm Hg, no water vapor) per unit time.
PI is PA 0, ,the partial pressure of O 2 in alveoli. P2 is partial pressure of O 2
in the red blood cell. DL is DL0, , the diffusion capacity of the lung for O 2 ,
Similarly, subscripts CO, CO 2 , etc., are used when Eq. (10) is applied to
these gases.
A further separation of the right-hand side ofEq. (10) into two terms would
clarify the role of chemical reaction in gas transfer in the lung; see Fig. 7.4.2:
V02 = DLo ,E(P02 in alveoli - P0 2 in plasma)
+ (P0 2 in plasma - P0 2 in red cell)]. (11)
In analogy with Eq. (7) we write
V02 = DMo ,(P02 in alveoli - P0 2 in plasma), (12)
where DM0, is called the membrane diffusion capacity of O 2 , On the other
hand, to account for chemical reaction, Roughton and Forster (1957) write:
V02 = e02 v,,(P0 2 in plasma - P0 2 in red cell). (13)
Here v" is the blood volume in the capillary blood vessels of the lung, and e02
is a reaction rate, which is usually expressed in ml of O 2 per min per mm Hg
per cm 3 of blood. Combining Eqs. (11)-(13) and omitting the subscripts O 2 ,
we have
1 1 1
-=-+-. (14)
DL DM ev"
If the inverses of Du DM , ev" are interpreted as resistances to gas transport,
7.S Ventilation/Perfusion Ratio 249

then Eq. (14) expresses the idea that the total resistance is the sum of the
"membrane" resistance and the resistance due to a finite reaction rate.
If carbon monoxide is used to measure the diffusion capacity, Pl is the
partial pressure of CO in alveoli and P2 is that in the red blood cell. But we
have seen in Fig. 7.4:3 that P2 of CO is negligible compared with Pl' Hence
the diffusion capacity of the lung for carbon monoxide is

D _ Veo (IS)
Leo - PACO

The normal value of DLco of human depends on sex, age, and height; for
adult man at rest it is about 25 cm 3 per min per mm Hg, and it increases to
two or three times this value with exercise. Oco is very large, and DMCO is
essentially equal to DLco'
For oxygen, DM02 and 002l';, are approximately equal. Hence a reduction
of capillary blood volume l';, by disease is effective in reducing the diffusion
capacity of the lung, DL . For an adult man DLO is about 1.23DLCO ,and DLCO
o~ 2 2

is about 24.6DLco ,based on molecular weights and solubilities of these gases.

Methods of measurements are discussed in detail in Bates et al. (1971).

7.5 Ventilation/Perfusion Ratio

An ideal condition for the lung to work is to have an exactly right amount of
oxygen delivered to the alveoli in time to oxygenate all the blood that is
delivered to the capillary blood vessels in the alveolar walls. This calls for a
certain ideal ventilation/perfusion ratio. In a normal lung, the ideal ratio is
approximately 1 to 1, i.e., the rate offresh air delivery is roughly equal to that
of blood delivery. The exact value of the ideal ventilation/perfusion ratio
depends on a number of factors, such as the hematocrit, the atmospheric
pressure, the altitude, the CO 2 level, and the humidity. Whatever the ideal
ventilation/perfusion ratio may be, however, it cannot be achieved in the entire
lung because both the ventilation and the perfusion are nonuniform in the
lung. One cause of this nonuniformity is gravity. For a man in an upright
position, the weight of the blood creates a hydrostatic pressure in the blood
vessels, whereas the weight of the air in the airways is so small that it's
negligible. The hydrostatic pressure in the blood vessels distends them. As a
result those blood vessels near the base of the lung are distended more than
those near the apex of the lung. The resistance to flow is thus smaller in the
basal region than in the apex region. Hence the blood flow per unit volum~
of the lung is larger toward the base of the lung. In contrast, the alveolar size
is the largest toward the apex of the lung because of the elastic deformation
of the alveolar structure due to gravity. Since the lung parenchyma is very
compliant, its weight pulls the alveoli in the apex region to a larger size while
compressing the alveoli in the region of the base. The net result is that the
ventilation/perfusion ratio is larger near the apex and smaller toward the base.
250 7 Respiratory Gas Flow

That is for the normal lung. In a diseased lung, some airways can be closed
by mucus or by tumor growth. Then the alveoli downstream of these airways
will not be ventilated, and the ventilation/perfusion ratio will be reduced to
zero in that region. Also, a blood vessel can become obstructed by thrombi
or other causes; in that case perfusion is reduced and the ventilation/perfusion
ratio becomes large. This suggests that if we know how to assess the distribu-
tion of the ventilation/perfusion ratio in the lung, we may obtain an effective
tool for diagnosis. The search for means to do this leads to an interesting
integral equation given by Wagner and West (1972), which we shall derive
and discuss.
A method to measure ventilation/perfusion distribution was proposed by
West and Wagner (1977). They took a mixture of several inert gases* and
equilibrated them with normal saline or isotonic dextrose in water. Then they
injected the solution into a peripheral vein, and, after a steady state was
established in the lungs, simultaneously collected samples of the mixed venous
blood and the mixed expired gas. From the samples they measured the
concentration of each gas and the solubility of the gas in blood by gas
chromatography. The total pulmonary blood flow and the minute ventilation
were also measured.
To analyze the situation, consider first a single alveolus and a single inert
gas; see the schematic in Fig. 7.5:1. The mixed venous blood containing the
inert gas flows into the capillary blood vessel, while the inspired gas does not
contain that inert gas. Assuming that at a steady state the diffusion equilibrium
is complete so that the partial pressure of the gas in the end-capillary blood
(PeJ and that in the alveolus (PA) are equal. Let Pv be the partial pressure of
the gas in the mixed venous blood, d~ be the expiratory ventilation rate, and
dQ be the blood flow rate in this small exchange unit. Then PAdVA is the rate
at which that gas is expired from the alveolus to the atmosphere, whereas
dQ(c v - cee) = dQ A(Pv - PeJ is the rate at which that gas is evolved from the
blood to the alveolus. Here the symbol c denotes the concentration of the inert
gas, and A denotes the solubility of the gas in blood. Equating these two
quantities on the basis of conservation of mass, we have
dQA(Pv - PeJ = PAd~.
Noting that Pee = PA as assumed, and solving for PA, we obtain

PA = Pee = Pv(A + d~/dQ)" (1)

This applies to one alveolus. Now let the lung be considered as a collection
of a large number of alveoli in parallel. Each alveolus has a value of d~/dQ.
Let this value be denoted by x. In the whole lung the value of x is statistically
distributed among the avleoli. On the other hand, the ventilation and per-
fusion of each alveolus are also statistically distributed among the alveoli.

* Inert in the sense that the solubility is constant and the dissociation curve is linear, not in the
sense of pharmacological effects.
7.5 Ventilation/Perfusion Ratio 251

Alveolar
gas
Arterialized
blood

Pulmonary Capi Ilary

FIGURE 7.5:1 Schematic drawing of the distribution of an inert gas in a small control
volume of pulmonary capillaries and alveoli. The tube below represents capillary blood
vessels; the bellshaped unit above represents the alveoli. C = concentration of the inert
gas (mass/vol). p = partial pressure of the inert gas. In liquid and solid the concentra-
tion is equal to the product of the solubility (A) and the partial pressure: c = Ap. The
subscripts atm, A, a, c, v denote atmospheric, alveolar, pulmonary arterial, end capil-
lary, and pulmonary venous. V = volume rate of gas flowing out (expiration) of this
small control volume. Q = volume rate of blood (perfusion) flowing into the same
control volume.

Let A~(x) be the sum of the ventilation of all alveoli whose ~/Q ratio lies in
the range between x and x + Ax. Then the inert gas coming from these alveoli
is expired at a rate PA(l,x)A~(x). The rate of expiration ofthe inert gas from
the entire lung is therefore the sum of PA(l,x)A ~(x) over all values ofx. If the
probability distribution of x is continuous we shall write A ~(x) as ~f.,(x) dx,
where ~ represents the total ventilation of the lung, and fv(x) signifies the
probability frequency function of the ventilation: the probability of finding
the ventilation to come from alveolar units with d~/dQ values lying between
x and x + dx. The rate of expiration of the gas is therefore

L PA(l,x)A~(x) = ~ IX) PA(l, x)f.,(x) dx. (2)

Substituting from Eq. (1) and denoting the result as G." we obtain

.
G.,(l) = Pv ~'foo -,-f.,(x)
l dx. (3)
o 11. +X
By an identical reasoning, we let fQ(x) denote the probability frequency of the
blood flowing into those alveoli whose d~/dQ ratio is x, and find the total
amount of gas leaving the lung in the flowing blood, GQ(l):

.
GQ(l) = PvQ 'foo -,-fQ(x)
l dx. (4)
o 11. +x
252 7 Respiratory Gas Flow

West and Wagner (1977) call the ratio G,,(A.}/Pv ~ the excretion of the inert
gas and GQ(A.}/PvQ the retention of that gas. For mathematical reasons it is
more convenient to divide retention and excretion by A., and denote the results
by r(A.} and e(A.}, respectively. They then obtain the basic integral equations:

r(A.} = 1 00 1
-,-fQ(x} dx,
o 1\.+ x
(5)

e(A.} = 1 00

o
1
-,-f,,(x} dx.
I\. +x
(6)

The contention of Wagner et al. is that r(A.}, e(A.} can be measured at several
values of A. by choosing suitable inert gases. The problem is to solve for the
unknown functions fQ(x} and f"(x}.

Solution of the Integral Equation

The equation

r(A.} = 1
o
00 1
-,-f(x} dx,
I\. +X
(7)

is a regular Fredholm integral equation of the first kind. If f(x} is continuous

and tends to zero as x -+ 00 as fast as x-I', where fl > 0, then the integral exists
in Riemannian sense; and the function r(A.} so defined is a continuous function
of A., except possibly at A. = O. If f(O} is finite, then r(A.} tends to - f(O} log A. as
A. -+ 0.1f f(x} tends to zero as fast as XV as x -+ 0, where v> 0, then r(x} is finite
at A. = 0 and continuous for A. ;;?; O.
The theory of the regular Fredholm integral equation is well known. Since
the homogeneous equation obtained from Eq. (7) by putting r(A.} equal to zero
has no solution other than f(x} = 0, the nonhomogeneous equation (7) has a
unique solution.
This equation was exhaustively investigated by Stieltjes, and several solu-
tions are given by Van der Pol and Bremmer (1950). For example, the inversion

1
may be expressed as an integral
i
f(x} = -22
fC+ioo dpxPsin(np} 00

du
r(u}
p+1' (x > 0; - 1 < c < O), (8)
n c-ioo 0 U

or as in infinite product

f(x} = -x~
dx
fI
n=-oo
I (1 + ~n ddx)r(x), (x> O), (9)

in which the prime at the product sign means that the term n = 0 should be
deleted.
These exact solutions are quite useless in practice, because experimental
determination of the retention r(A.} as a function of the solubility A. is limited.
7.6 Pulmonary Function Tests 253

For example, in West and Wagner's extensive research on this subject, r(.Ie) is
determined only for six values of.le (with six specifically chosen gases). Hence
the function r(.Ie) is known only for six points. The question of solving Eq. (7)
is transformed into the following: What can be said about the function f(x) if
r(.Ie) is known for six values of .Ie? Can some major features of f(x) (such as how
many peaks does it have, and at what values of x are these peaks located), be
determined from such scanty data? West, Wagner and their colleagues' answer
is affirmative, and very elaborate theoretical investigations have been done to
assess this question. These theories are not very easily understood. See West
and Wagner (1977) for details.

7.6 Pulmonary Function Tests

Any measurement of the function of the respiratory system is a pulmonary

function test. Pulmonary function tests help physicians to make diagnostic
and therapeutic decisions. The most common pathophysiologic patterns of
pulmonary diseases are obstructive (70%), restrictive (20-25%), and vascular
(5-10%). Dynamic flow measurements are most useful to evaluate obstructive
diseases whereas static elasticity measurements are most useful to evaluate
restrictive diseases. In the following we consider the principles of some of these
tests.

Volume Measurements
The volume of gas moved by normal breathing is called the tidal volume. When
a person takes a maximal inspiration and follows this by a maximal expiration,
then the exhaled volume is called the vital capacity (VC). Some gas remains
in the lung after a maximal expiration; this is the residual volume (RV). The
volume of gas in the lung after a normal expiration is the functional residual
capacity (FRC). The sum of the residual volume and the vital capacity is the
total lung capacity (TLC).
A spirometer is used to measure the tidal volume and vital capacity. It will
not yield, however, the residual volume and functional residual capacity.
In order to measure absolute gas volume in the lung (TLC, FRC, and RV),
three methods may be employed: inert gas dilution or wash out, radiological
techniques (see Bates et aI., 1971, pp. 13-15), and whole body plethysmography
(Dubois, 1964, p. 454). To measure regional lung volume (i.e., the volume of
identifiable parts of the lung), the 133xenon gas-dilution method can be used.
The radioactive xenon is inhaled and detected externally. The scintilation
count rate of 133xenon in the gas in a closed spirometer circuit is compared
with the count rate after a single breath or during equilibration. Analysis is
based on the principle of indicator dilution (Sec. 9.9).
Since the lung deforms under its own weight, the regional distribution of
lung volume is affected by gravity (Sec. 11.6). If breathing takes place at lung
254 7 Respiratory Gas Flow

volumes lower than FRC, the regional volume decreases more rapidly in the
direction of the gravity. The minimal regional volume (regional residual
volume) is about 20% of the regional TLC. In a young person seated upright,
the regional minimal volume is reached first in the lowest zone when the total
lung volume is at about 40% TLC, and in the midzone when the lung is at
about 30% TLC. With loss of elastic recoil in older persons, these minimal
volumes will be reached at higher fractions of the TLC. It appears liklely
that the minimal volume is reached by closure of airways (Bates et aI., 1973,
p. 46). Airway closure beings at the lowest part of the lung which is exposed
to more positive pleural pressure, and progresses upward in the lung. This
phenomenon of airway closure at low volume has, of course, major physio-
logical significance. In conditions in which the lung volume is reduced, as in
gross abdominal obesity, airway closure may be responsible for the low
arterial oxygen tension. In older people, ventilation is diminished immediately
on the assumption of recumbency, presumably as a result of airway closure.
The airway space consists of two compartments: the anatomical dead space
(VD) in which there is no respiratory alveoli and hence no gas exchange with
blood, and the alveolar space in which there is gas exchange with blood. It is
of interest to measure the anatomical dead space. Such a measurement is quite
easy if one makes the assumption that as gas is moved into and out of the lung
the velocity and concentration profiles are flat (i.e., the velocity of flow and the
concentration of any component of the gas are uniform in any cross section
of the airway), but although this assumption is practically always made
(tacitly) in medical literatures, it is certainly wrong (see Sections 7.2 and 7.3.).
The error is generally negligible in ordinary circumstances, except in high
frequency ventilation. Under this assumption, if a volume of gas VE that is
expired at the mouth contains alveolar gas, then, it must consist of a sum of
the dead space and a volume ~ coming from the alveolar space:
(1)
because all the gas in the dead space must first come out before alveolar gas
can. (If VE is smaller than VD then ~ must be zero.) Equation (1) is not valid
if this assumption is not made. For example, if the alveolar gas fluctuates and
diffuses out like ajet in the center of the bronchi and trachea, then the expired
gas can indeed contain the alveolar gas while the expired volume is smaller
than the dead space. Something like this actually happens in high-frequency
low-tidal-volume ventilation (see Sec. 7.8).
Under the assumption named above, the anatomical dead space can be
measured by simultaneous registration of expired nitrogen concentration and
volume of flow, following a deep inspiration of a nitrogen-free gas. If the N 2
concentration is plotted against the volume of gas expired (Fig. 7.6:1) and the
concentration profile can be approximated by a step function, then the expired
gas volume before the rise of the step is the anatomical dead space.
Another method of measuring the dead space is based on CO 2 exchange.
If one inhales fresh air and then collects a complete expiration in a bag, the
7.6 Pulmonary Function Tests 255

START OF
INSP.

80
~

END OF
EXP.

DEEP
__-------i~ CALVo
BREATH -
40 OF O2

I
START OF I
I
EXP.
'A
~
I
I
0 I
I C1NSP.
I
, - VDS _ ' _ _-VALV ~
I .
I

~ VEXP. ~

FIGURE 7.6:1 Fowler's method of measuring the anatomical dead space with a rapid
N2 analyzer. The dead space is the expired volume up to the vertical broken line which
makes the areas A and B equal. From Bates et al. (1971)

amount of CO 2 in the bag is equal to the fractional concentration of CO 2

times the volume of expiration, i.e. FEC02 x VE • This amount of CO 2 is made
up of two distinct portions, (1) a volume of CO 2 from the nonexchanging dead
space VD in which the concentration is the same as that in the inspired
air (FIC02 )' plus (2) the volume of CO 2 derived from alveolar gas, FAC02 x
EVE - VDJ. Hence
(2a)
or
VD = (FAC02 - FEC02 ) VE • (2b)
FAco2 - FIC02
If an assumption is made that the partial pressure of CO 2 in the arterialized
blood, PaC0 2 ' is equal to that in the alveoli, PAC0 2 (see Sec. 7.4), and that the
inspired air contains no CO 2 , then on replacing the fractional concentrations
by partial pressures, Eq. (2b) becomes

v;DC0 - VE (PaC02
P
- PEC0 2 )
. (3)
2 -
aC0 2
256 7 Respiratory Gas Flow

One can thus measure V DC02 by measuring VE' PaC0 2 , and PEC0 2 . V DC02 is called
the physiological dead space. It is equal to the anatomical dead space if the
assumption PAC0 2 = PaC0 2 is valid.

Ventilation
The minute ventilation (VE ) is defined as the quantity of air expired per minute.
By differentiating Eq. (1) with respect to time, we obtain
(4)
where a dot indicates the rate of change with respect to time. The minute
ventilation (VE ) can be measured with a recording spirometer, a pneumota-
chometer, or a collecting bag. If VD is measured then VD is equal to the product
of VD and the frequency of breathing; and ~ can be computed according to
Eq. (4).
Substituting Eq. (3) into Eq. (1), one obtains an important equation

VA-- V E PEC0 2 .
~=
Vco 2.
__ (5)
PaC0 2 PaC0 2

Equation (5) describes the elimination of CO 2 in health and its retention in

disease. For a given metabolic need (Vco 2 fixed) and a decreased ~ (alveolar

~ effort dependent-l

C
10

V
(lPS)

o~------------ ______________•
TLC RV
v
{ll

FIGURE 7.6:2 Flow (6) vs. volume (V) curves in forced expiration between total lung
capacity (TLC) and residual volume (RV) with graded efforts after a full inspiration.
From Bates et al. (1971)
7.6 Pulmonary Function Tests 257

• • •• •
• • •
V
LPS

+10 o -10 -20 -30 -40 -50

FIGURE 7.6:3 An example ofthe iso-volume pressure-flow curve obtained in a normal

subject at 33 percent of vital capacity. From Bates, Macklem, and Christie (1971), by
permission.

.
hypoventilation), the arterial partial pressure of CO 2 (PaCO ,) will rise (hy-
percapnia); conversely, for an increased Y.t (alveolar hyperventilation) PaCO,
will decrease (hypocapnia).
A popular test is the forced expiration after a full inspiration. If a person is
asked to perform a series of graded expiration efforts, the recorded relationship
between the flow rate (V) and lung volume (V) will be different for each effort.
In Fig. 7.6:2, curve A refers to a small effort, curve B is for intermediate effort,
curve C is for maximal effort. Note that all three curves merge at a point and
thereafter follow a common pathway to the residual volume. This common
portion of the curve is said to be effort independent. Hyatt, Schilder, and Fry
(1958) plotted the forced expiration velocity at a fixed lung volume versus the
transpulmonary pressure and obtained a result similar to that shown in Fig.
7.6:3. It is seen that the expiration velocity is independent of the transpulmo-
nary pressure. Later, this was recognized by Permutt, Mead, and others as the
"waterfall phenomenon." The analysis presented in Sec. 5.17 applies.

Resistance and Compliance

With the idea that the flow in the airway is driven by the difference of pressures
at the alveoli and the mouth, the airway resistance to flow R is conventionally
258 7 Respiratory Gas Flow

defined by the equation

R = Pmouth --:- Palveoli . (6)

V
Similarly, with the thought that the lung volume is determined principally by
the pressure difference across the pulmonary pleura (i.e. the difference between
the alveolar pressure, Palv, and the pleural pressure, PPL), the lung compliance
is defined by the equation
dV
C =---::--,-----:- (7)
d(Palv - PPL)'
where V is the lung volume and V is dV/dt. C is approximately constant over
the tidal volume range, whereas at the end of quiet expiration (FRC), Palv - PPL
is approximately 5 cm H 2 0. Therefore, as an approximation
tidal volume
Palv - PPL = (Palv - ppdFRC + C . (8)

Hence, since the transpulmonary pressure is defined as

Pmouth - PPL = (Pmouth - PalJ + (Palv - PPL), (9)
we have, from Eqs. (6) and (8),
tidal volume .
Pmouth - PPL = (Palv - ppdFRC + C + R V. (10)

Quantities in these equations can be measured or calculated. PPL is often

approximated by measurements using an esophageal balloon. Pmouth and V
can be measured by a pneumotachometer. Tidal volume can be measured by
a spirometer. Dubois (1964, pp. 453-454) has shown how to calculate Palv from
Boyle's law using measurements from a body plethysmograph. From such
measurements, Rand C, which are functions oflung volume, can be computed.

7.7 Dynamics of the Ventilation System

Respiratory flow in the airway is a three-dimensional phenomenon of great

complexity. Its analysis awaits future development in computational fluid
mechanics. In Sec. 7.2 we mentioned a number of basic problems of interest.
Among them is the system behavior ofthe branching tree. Flow in one segment
affects the flow in the entire tree. (The same is true also, of course, of circula-
tion.) Simulation ofthe entire tree has been done only under strong simplifying
assumptions so far. A few words about this simplified approach would be
worth while because it points out the advantage of introducing the adjoint
system and formulating a variational principle.
A method presented by Seguchi et al. (1984, 1986) simulates the ariway as
7.7 Dynamics of the Ventilation System 259

(al Element Vector (b) System Vector

FIGURE 7.7:1 Simulation of an airway tree by representing each segment of the airway
as an element. At a branching point three elements meet. The modal vector of the
elements are assembled into a system vector.

a tree (Fig. 5.2:2). Each branch of the tree is called an element, Fig. 7.7:1.
Variables such as the pressure, p, are calculated at a number of nodal points
and listed as nodal vectors. The value of the variables between neighboring
nodal points are specified by certain interpolation formulas. The nodal
vectors of the elements are assembled into a system vector, as illustrated in
Fig. 7.7:1. The objective of the mathematical simulation is to calculate the
system vector according to the basic equations.
The basic equations for the gas in the airways are the equations of motion,
Eq. (1.7:8), continuity, Eq. (1.7:4), state (gas law), and boundary conditions.
Seguchi et al. (1984) used the one-dimensional approximation of Streeter and
Wylie (1967). We must first make clear what the approximation means, and
generalize it as much as possible without introducing unwelcome mathe-
matical complications.

Approximate Equations
The Navier-Stokes equation (1.7:8), reads
au; au; 1 ap J1. a 2u;
- + Uj - = + - - + - - - + X ; . (1)
at aXj pax; p aXjaXj
Here U; is the velocity vector with components Ul , U2 , U3 (or u, v, w) referred to
a rectangular cartesian frame ofreference with coordinates Xl' X 2 , X3 (or X, y,
z); t is time, p is the gas density, p is the pressure, X; is the gravitational force
per unit volume.
Let the axial distance along an airway be represented by X = X l' Consider
a cross section of the airway perpendicular to the X axis. On multiplying Eq.
(1) by an infinitesimal element of area dA = dy dz and integrating the product
260 7 Respiratory Gas Flow

over the cross section A, the first equation (i = 1) becomes

f ou
-dA + f -1 -dA
ou +
2 f (OU
v- + w-
Ou) dA

f f
A ot A 2 ox A oy OZ

= - f A
1 op
--dA
P ox
J1
+-
P A
02U
-dA
ox 2
J1
+-
P A
(02U
- +02U)
- dA.
oy2 OZ2
(2)

To simplify this equation, we introduce the following specific

approximations:
1) The pressure p is uniform over the cross section. Hence, in isothermal or
adiabatic process, the density P is also uniform.
2) The third and the fifth integrals in Eq. (2) are zero.
3) The effect of the gravitational force is negligible.
4) The other two equations (i = 2 and 3) can be ignored.

With these hypotheses, we introduce the following notations for the average
velocity, and transient and convective accelerations and two "conversion
factors" C l and C 2 :

u == ~
A
f A
UdA, OU == ~
ot A
f A
ou dA ==
ot
C
1
OU
ot
(3)

~ ~: == -~ L~ 00: dA == C2~ 0::.

f
We introduce also a factor Z to replace the last term in Eq. (2) by

J1 -
-8n-uZ=-J1 (02U
- +02U)
- dA. (4)
P P A oy2 OZ2
Equation (4) is motivated by a consideration of energy dissipation presented
in Sec. 7.2, where the same Z is introduced to correct the dissipation function
computed by the Poiseuillean formula. Poiseuille flow is presented in Sec. 5.10
under the assumption that the tube is circular, the axial velocity distribution
is axisymmetric, and the flow is laminar (not turbulent). Poiseuille flow satisfies
all the simplifying assumptions 1)-4) above, and Eq. (4) with Z = 1. This can
be verified by dividing Q by na 2 , a being the radius of the airway, to obtain u
using Eq. (5.10:8), substituting the result into Eq. (5.10:7), and using the
transformation
02U 02U 1 0 (Ou) 1 02U
oy2 + OZ2 = -;: or r or + r2 oe 2
from rectangular coordinates x, y, Z to cylindrical polar coordinates x, r, e.
With these approximations, Eq. (1) becomes
ou 1 ou2 op 8nuZ _
PC l at + 2PC2 [j; = - ox - ~u. (5)
7.7 Dynamics of the Ventilation System 261

Next, apply the same procedure to the equation of continuity (1.7:4)

op + o(puj ) = O. (6)
ot OXj

Under the same approximations named above, we obtain

o(pA) + o(puA) = O. (7)
ot ox
Carrying out the differentiation, regrouping terms, using the notations of
"material" derivative (Sec. 1.7)
. DA oA oA . Dp op _op
A=-=-+u- p=-=-+u- (8)
Dt ot ox' Dt ot ox
and dividing through by pA, we obtain
A p ou
-+-+-=0. (9)
A p ox
This equation can be simplified by taking the gas law and the area-pressure
relationship of the tube into account. If the cross sectional area A is a function
of the transmural pressure P - Pe' where P is the pressure in an airway and Pe
is the pressure external to it, then the speed of pulse wave in the airway, c, is
given in Sec. 5.7 (with a change of notation):

c
= [i
p
d(p - Pe)J1 /2
dA . (10)

Hence,
A 1 dA 1
-A d( ) (p - Pe) = - 2 (p - Pe)· (11)
A P - Pe pc

The other term PiP is related to pressure. By the gas law pip = R T, where
R is the gas constant and T is the absolute temperature, we have pip =
constant if the process is isothermal, or pip Y = constant if the process is
adiabatic, y being another constant. In either case, p is a function of p. Hence

P 1 dp.
-=--p=-p.
l.
(12)
p pdp K

Here K = dpI{dplp) is the bulk modulus of the gas. Note that dpI(dplp) =
-dpl(dvlv), v being the specific volume of the gas, v = lip, and dvlv is the
volumetric strain.
Combining Eqs. (9), (11), and (12), we obtain

(13)
262 7 Respiratory Gas Flow

Equations (5) and (13) are the two basic equations in two variables P and
u for each segment of the airway. If we assume C I = Cz = Z = 1, then Eq. (5)
is reduced to the equation used by Streeter and Wylie (1967) and Seguchi et
al. (1984). The introduction of the constants C I , C z, Z permits us to take
secondary flow and turbulence into account as partially illustrated in Sec. 7.2.
Since the number of branches in the airway system is large, a practical way
to solve this large set of equations is by finite difference or finite element
method. The procedure can be based on a variational principle outlined below.

Variational Principle
Seguchi et al. (1984) extended the Lagrangian multiplier method to formulate
a variational principle for the airway system. For each airway branch oflength
L, we multiply Eq. (5) by Al (x, t), Eq. (13) by Az(X, t), sum, and integrate over
the domain (0 ::::; x ::::; L,O ::::; t ::::; T) to obtain a functional J:
Cz ou Z op
J= f 0TfL0 {AI(X,t) [au
PC I Tt+ P 2--a;,-+ ox
811:JiZ_]
+~u

+ AZ(X,t{ C~z + ~)G~ + u:~)

+au 1 (OPe OPe)]}

ox- -pc-Z -at+ uox- dxdt, (14)

where Al (x, t) and Az(X, t) are Lagrangian multipliers. We seek the necessary
conditions for J to assume a stationary value when the functions p(x, t), u(x, t),
Al (x, t) and Az(X, t) are allowed to vary independently and arbitrarily. Pe is
considered an external forcing function and is not varied for the problem of
gas flow in the airway. The first variation of J, after successive integration by
parts, takes the form:

bJ = LT LL {bA I [PCI ~~ + pczu:: + :~ + 811:;Z u]

+ bAz [C~z + ~)p - p~z fie + ::]
7.7 Dynamics of the Ventilation System 263

For the functional J to assume a stationary value, the necessary condition is

that bJ vanishes for arbitrary bAl, bA 2, bu, and bP. Thus the brackets multi-
plying bAl' bA2, bu, adn bp in the double integral in Eq. (15) must vanish, and
we obtain the Euler equations:

(16)

( _1_2 + ~)p
pc K
__ 1_Pe
pc 2
+ au = 0,
ax
(17)

O(AlPC l ) + O(A l PC2U) _ Al 81tJiZ + OA2 _ A2 (_1_2 +~) op = 0, (18)

at ax A ax pc K ax

OA l + ~(~ + A2) + ~(A2U + A2U) = O. (19)

ax at pc 2 K ax pc 2 K
The initial and boundary conditions must render the line integrals in Eq. (15)
as zero; i.e.,

[AlPClbU+A2C~2 + ~)bPJI: =0, (20)

{[Al +A2C~2 + ~)UJbP+ [A l PC2U+A2]bU}I: =0. (21)

The conditions Eqs. (20) and (21) can be satisfied by proper combinations of
the following three sets of conditions:
F oced boundary conditions at t = 0 or T and x = 0 or L:

bu(x,O) = 0, bU(x, T) = 0, bp(X,O) = 0, bp(x, T) = 0,

(22)
bu(O, t) = 0, bu(L,t) = 0, bp(O,t) = 0, bP(L, t) = 0,
i.e. the values of u and p specified at x = 0 or L for all t, or at t = 0 or T for
all x.
Natural initial conditions at t = 0 or T:

[ AlPClU + A2~
pc
+ ~Jp
K
= O. (23)

Natural boundary conditions at x = 0 or L:

[Al +A2C~2 + ~)uJP+ [A l PC2U+A 2]U=0. (24)

It is clear that Eqs. (16) and (17) are our basic equations (5) and (13),
respectively. The initial and boundary conditions given by Eqs. (22)-(24) are
physically realizable. Thus the variational principle bJ = 0 is consistent with
our system of differential equations and boundary conditions. The solution
of ventilation dynamics consists of finding p, u, Al , and A2 which renders J a
stationary value.
264 7 Respiratory Gas Flow

Physical Meaning of J
If we give A1 a physical dimension of [length 3 ]/[time], and A2 a physical
dimension of [force], then J defined by Eq. (14) has the dimension of [force] .
[length] or energy. We shall call J the adjoint energy integral and A1, A2 the
adjoint state variables. The equations (5) and (13) are nonself-adjoint. The use
of Lagrange multipliers or adjoint variables establishes a functional J which
is stationary when the state variables obey the equations of motion and
continuity. But whether J is a maximum or a minimum is uncertain. Rigorous
mathematical study of this type of variational principle for nonself-adjoint
systems remains to be a challenge. In a study of a non-self-adjoint system of
a column loaded by a force which is always tangential to the axis of the
column, Prasad and Herrmann (1969, 1972) showed that the adjoint method
is more efficient for calculation than the Galerkin's method or its extension
using weighted averages. Whittle (1971) has discussed variational principles
of this nature in a general way.

Discretization and Computer Simulation

Computer simulation of breathing based on the variational equation (jJ =
has been presented by Seguchi et al. (1984, 1986). The airway system is
°
subdivided into a number of subregions. See Fig. 7.7:1. In a subregion i the
values of the variables p, U, A1, and A2 are computed at specific points called
nodes and listed as nodal vectors {p}:', {u}:', Pdf, and P2}:" respectively.
Polynomial interpolation matrices [N] are then used to obtain p, U, A1, A2 as
continuous functions of x:

{P}i = [N p ] {p}:', {U}i = [Nu ] {u}:',

(25)
Pdi = [N"J Pdf, {A2}; = [N",] P2}:'.
The overall airway system is assembled from the subregions. The global
vectors can be written as
(26)
etc., if we denote {P}i by Pi. The functional J of Eq. (14) can then be evaluated
as functions of {p}:', etc. By setting (jJ = 0, a set of linear, simultaneous
equations is obtained for {p}:' and {u}:', which can be solved numerically. The
equations for p and u are decoupled from those for the adjoints At> A2 • A key
observation that will greatly simplify the analysis is that the airway of each
generation has nearly constant diameter, and that the diameter, the com-
pliance, the total cross-sectional area, and the velocity of the flow have
stepwise changes at the points of bifurcation where the generation number
changes. Hence according to Bernoulli's equation, the blood pressure has a
stepwise change at the bifurcation point. These changes should be incorporated
in the boundary conditions of the successive generations.
7.8 High-Frequency Low-Tidal-Volume Ventilation 265

7.8 High-Frequency Low-Tidal-Volume Ventilation

Bohn et al. (1980) have shown that a dog can survive ventilation with a tidal
volume smaller than the dead space of the airway at frequencies ranging from
5-30 Hz. This feat was anticipated by Henderson et al. (1915) who supported
the idea with an admirably simple experiment. They blew tobacco smoke
down a tube and found that it formed a long thin spike, concluding that the
"quicker the puff, the thinner and sharper the spike." When the puff stopped
"the spike breaks instantly, everywhere, and the tube is seen to be filled from
side to side with a mixture of smoke and air." Later, Briscoe et al. (1954)
showed that, in man, inspired volumes less than the dead space can reach the
alveoli; and they speculated that this might explain why "some patients can
live despite the fact that they are breathing very small tidal volumes." Lee
(1984), using laser to measure the density of cigarette smoke in a tube, showed
that in high frequency ventilation the effective diffusivity (see Sec. 7.3) of the
smoke can be hundreds or thousands of times larger than the coefficient of
molecular diffusion. Hence high frequency ventilation can be effective to send
some aerosol drugs down to the lung.

Mechanism of Smoke Transport

The mechanism of the smoke transport in Lee's experiment is as follows: On
forward stroke the smoke moves forward with the flow, with higher concentra-
tion in the core and lower concentration at the wall. Simultaneously the smoke
diffuses (by molecular diffusion or by turbulent mixing) from the core into the
wall region. This radial spread of smoke decreases the concentration of the
smoke in the core region. In the reverse stroke the same process continues,
but the core concentration is lower. Hence in a whole cycle there is a net
movement of the smoke to the right. This net movement is repeated in each
subsequent cycle; and by a reasoning similar to that used in Sec. 7.3, the
concentration of smoke is described by the differential equation
ac a2 c
at = Deff ax 2 (1)

in which c is the concentration ofthe smoke, t is time, and x is the axial distance
in the direction of motion. Deff is a constant called the effective diffusivity.

Mechanism of Gas Transport in a Tube

Similar to the smoke transport in Lee's experiment, the basic reason why
oxygen can get to the alveoli at a tidal volume smaller than the dead space is
because the velocity and concentration profiles in the airway are not flat. The
velocity is faster and the concentration is higher in the core region, so that by
a mechanism similar to the Taylor diffusion (see Sec. 7.3) the effective diffusivity
is augmented. Taylor's solution is, however, an asymptotic solution of a steady
266 7 Respiratory Gas Flow

flow. To generalize Taylor's solution to high frequency ventilation of the lung,

one must take into consideration the periodic oscillations, the branching
pattern of the airway, and the fundamental difference in convergent and
divergent flows of a viscous fluid in bifurcating tubes. The velocity and
concentration profiles are affected by these factors.
Some features of oscillatory flows in a circular tube have been discussed in
Chap. 5, Secs. 5.7, 5.12. In oscillatory flow the velocity profile is no longer
parabolic, but is more blunt, and the maximum velocity occurs at a radial
position which varies with time. Depending on the Womersley number, the
high velocity core spreads much closer to the wall. The Taylor diffusion
mechanism is enhanced by this change of velocity profile.
For oscillatory flow in a circular cylindrical tube of infinite length, Chatwin
(1975) has solved the problem of dispersion. When the Womersley number is
greater than 1 and the Schmidt number (1] == v/Dmo1 ) is of order 1 or greater,
Chat win finds the effective diffusivity to be
VO, S G2
Deff = Dmol + 4 3,sf3,sd 2 [F1(1]) + F2(1])(cos4nft - sin4nft)], (2)

where Dmol is the coefficient of molecular diffusion, v is the kinematic viscosity,

f is frequency in Hz, d is diameter in cm, t is time, F1 (1]), F2(1]) are functions
of the Schmidt number 1], and G is proportional to the amplitude of the
pressure gradient.
1 dp
G= - - - . (3)
p dx

Chang et al. (1984), using fluid transmission line theory for a sinusoidal
excitation (Brown, 1962), write
(4)
where VT is the tidal volume in ml, and :!l' is the impedance. For air, and with
j= (_1)1/2, they found

:!l'(f) = 8fj { [( 2v
d 2 1 + 0 nfd 2
)o.SJ} (5)

in which the symbol 0 means "of the order of the argument in the bracket."
For high frequency ventilation of man the argument 2v/(nfd 2) is much smaller
than 1, hence, on omitting the small term involving 0 in Eq. (5) and substituting
Eqs. (3), (4), and (5) into Eq. (2), Chang et al. obtain
64 Vifo.svo. s
Deff = Dmol + n1. S d 6 g(1]), (6)

where g(17) is the quantity in the bracket in Eq. (2). In high frequency ventilation
the second term on the right-hand side of Eq. (6) is much larger than the first;
and we see that according to Chatwin's theory the effective diffusivity increases
Problems 267

with the square of the tidal volume, the square root of the frequency, and the
inverse sixth power of the diameter:

(7)

Hence for the same amount of ventilation (VT f), it is more effective to increase
the tidal volume VT than to increase the frequency; and the effective diffusivity
is much larger in the small airways than in large ones. Experimental validation
ofthis formula has been done by Chang et al. (1984). The maximum values of
Deff reached 180 cm 2 jsec in Chang's experiments as compared to the molecular
diffusivity of CO 2 in air, 0.16 cm 2 jsec at 20°C.

Gas Transport in Airway Tree

The airway is an assemblage of tubes which end on a common alveolar space.
Since the total cross-sectional area ofthe airway increases with the generation
number beyond the third bronchial generation, the inspiratory flow is diver-
gent at the points of bifurcation whereas the expiratory flow is convergent at
these points. The secondary flow of a divergent flow is very different from that
of a convergent flow, and this will influence the mass transport characteristics.
The boundary layer thickness grows faster in a divergent flow and the velocity
profile peaks in the central region. The boundary layer is thinner in a con-
vergent flow and the velocity profile is blunter in the middle. Scherer et al.
(1982) have paid attention to this feature. Using a flow visualization method
they have measured the transport characteristics of small opaque beads (di-
ameter ~ 4011) in a glycerine-water solution flowing in a physical model of
bronchial tree. With the empirical results they constructed a theory of gas
transport in high frequency ventilation.
While the Taylor mechanism dominates the scene from mouth to bron-
chioles, beyond the respiratory bronchioles the molecular diffusion mechanism
takes over. Some authors (see Chang, 1984), however, emphasize the convec-
tive effect ofthe direction offlow relative to the angles of branches at the points
of bifurcation. This last mechanism may persist at the level of alveolar ducts.
Many authors performed animal experiments (e.g. Bohn et al. (1980), Slutsky
et al. (1981)), whereas others investigated the effects of turbulence on effective
diffusivity (Lee, 1984), and the transition between laminar and turbulent flows
in oscillatory condition (Winter and Nerem, 1984). Chang (1984) explored
high frequency ventilation by vibrating the chest and abdominal wall. Clinical
applications have been vigorously pursued.

Problems
7.1 Draw a free-body diagram of the lung. Analyze the condition of equilibrium of
the lung in the thoracic cavity without neglecting the weight of the lung, and
describe a method to determine how the pleural pressure 'varies with the height.
268 7 Respiratory Gas Flow

7.2 It is common practice to assume the mean stress in the lung parenchyma to be
roughly equal to the transpulmonary pressure PALV - PPL. Assess the accuracy
of this hypothesis analytically on the basis of biomechanical principles.
7.3 Analyze the pressure in the esophagus as measured by a catheter tip balloon
inserted into the esophagus and discuss whether the result is approximately equal
to the pleural pressure or not.
7.4 The Dissipation Function f!fi of Eq. (2) of Sec. 7.2. In an incompressible Newtonian
viscous fluid, consider a control volume of unit volume consisting of a parallelo-
piped with edges parallel to the coordinates axes. The stresses acting on the
surfaces of this element are aij. The rate of change of strain is

where u 1 , u 2 , U 3 are the velocity components. The rate at which the work is done
on this element by the stresses is dW = aijdeij. The stress-strain relationship is
given by Eq. (7) of Sec. 1.7. Hence show that W = - pe kk + j1e iiij. This work is
dissipated as heat. The dissipation f!fi is obtained by integrating W throughout
the volume occupied by the fluid. For a Poiseuillean flow, only one derivative of
the velocity field does not vanish; namely, au/or. Deduce f!fip for Poiseuille flow
and verify Eq. (4) of Sec. 7.2.
7.5 Describe qualitatively the mechanics of gas flow in the airway when one sneezes
or coughs. Write down the governing differential equations and boundary
conditions.
7.6 From the point of view of boundary layer development, what is the difference
between inspiring flow and the expiratory flow in the airway? What effect does
this have on the resistance to flow in inspiration and in expiration? Write down
the boundary layer equations and the boundary conditions that apply.
7.7 Discuss the difference between the tendency for flow separation (the failure of
boundary streamlines to follow the solid wall) to occur in the airway in inspiration
and that in expiration. What effect does flow separation have on the resistance
to flow? Formulate this discussion mathematically on the basis of boundary layer
equations.
7.8 Discuss the difference in the Z factor for inspiration from that for expiration.
Why does such a difference exist? Is Z larger in inspiration than in expiration?
(Cf. Pedley et al. (1977).)
7.9 To analyze the dynamics of breathing, the flow in the airway may be idealized
as a one-dimensional flow. The flow is assumed to depend only on the distance
x along the airways. Consider a control volume of length Llx, confined between
sections x and x + Llx. Use the method of continuum mechanics, express the
balance of forces acting on the control volume. Make suitable approximations
to derive a differential equation governing the flow in inspiration and in expira-
tion. (Cf. Schmid-Schonbein and Fung (1978).)
7.10 A small perturbation in pressure is introduced to the air flow at the mouth
by means of an "interrupter." Derive the linearized differential equation and
Problems 269

boundary conditions for the perturbation of flow in the entire airway system.
Take notice of the elasticity of the airways. In order to completely determine the
boundary conditions, it is necessary to consider the lung parenchyma, pleura,
chest wall, diaphragm, abdomen, and the heart. Describe a mathematical pro-
cedure to complete the formulation of the problem. (Cf. loco cit. Prob. 7.9)
7.11 Estimate the Mach number of the flow of air in human lung breathing at a rate
of 1 liter/sec. Can the air be treated as an incompressible fluid in Problems 7.9-10?
Show that the compressibility of air can be ignored.
7.12 Consider a numerical method of solving the equations obtained in Prob. 7.9.
Outline the necessary steps. (Cf. Sec. 7.8)
7.13 From the solutions c(e, t) == c(x - Ut, t) and e'(e, z, t) given in Eqs. (10), (15), and
(23) of Sec. 7.3, reexamine the hypotheses expressed in Eqs. (13) of Sec. 7.3. Derive
the conditions under which Eqs. (13) are valid.
7.14 Derive Eq. (7.3:21) from Eq. (7.3:19) by considering the balance of mass inflow
and outflow of the solute in a control volume with transient change of the total
mass in the volume. Derive Eq. (7.3:23) from (7.3:22).
7.15 Consider a circular cylindrical tube (Fig. 7.3:2). The right-hand side of Eq. (6) has
to be replaced by (see Fung, 1977, p. 277)

V2e = a2 e + !~(r ae),

ax r ar ar
2

where r is the radial distance. Derive the formula for J(r) corresponding to J(z).
Show that ~app is given by Eq. (1) in this case.
7.16 Consider flow in a semicircular cylindrical open channel. The boundary condition
on the free surface is that the shear stress = 0 and pressure = O. The channel is
tilted to such a degree that the fluid fills the semicircular cross section all the time.
The gravitational force and the slope of the channel drives the flow. Show that
the Poiseuille solution for the velocity field in an axisymmetric flow in a full
cylinder satisfies all the boundary conditions here. Consider diffusion of an
impulsively injected solute at a section in analogy to the cases considered above.
Deduce the apparent diffusivity ~app in this case.
The solution considered in Section 7.3 is the fo~ndation of the indicator
dilution method of measuring flow rate and organ volume. The example of the
open channel suggests the applicability of the analysis to measure flow rate in
sewers and rivers.
7.17 With the solution I(x, t) given in Eq. (7.3:23) for the response to an impulsive
input, derive the concentration distribution following a step input at time t = 0
of the solute B of unit mass per unit axial length distributed uniformly from x = 0
to -00, i.e., with the input J(x,t) = l(t)·l(-x). l(t) is the unit-step function,
which is zero when the argument is negative, and is 1 when the argument is
positive.
7.18 Consider an airway tree. Hypothetically, regarding the first generation as an
infinitely long tube extending from x = 0 to -00. When the solute B is injected
as a unit step function l(t) ·1( -x), derive the concentration distribution of the
solute in the tree following the injection.
270 7 Respiratory Gas Flow

7.19 Extend the solution of Prob. 7.18 to the alveolar wall. Then formulate the
mathematical problem of oxygen transport across the alveolar wall. Do the same
for CO 2 ,
7.20 Using qualitative physical reasoning, explain why the ternary diffusion effect on
O 2 and CO 2 transport in the alveoli will be significant if the third gas is helium.
7.21 Consider a liquid layer between two gases (Fig. 7.4:1). Show that in general the
partial pressure of a diffusible gas is discontinuous at the liquid-gas interface.
7.22 The solubility of oxygen in blood is nonlinear. A curve showing the oxygen
concentration in blood versus the partial pressure of O 2 is called a dissociation
curve. Describe the dissociation curve of O 2 as affected by the CO 2 concentration.
Similarly, describe the dissociation curve of CO 2 in blood as affected by the
oxygen concentration. (Cf. West and Wagner (1977), West (1974).)
7.23 Take a breath ofN 2 0. Calculate the time course of uptake ofN 2 0 by the blood.
Answer. Let Vo be the intake volume ofN 2 0, corresponding to a partial pressure
Po. Let V be the volume ofN 2 0 transfered to blood. Then the partial pressures
ofN 2 0 in the alveoli and blood, PI and P2 respectively, are
V
P2 = AV'
c

where v" is the capillary blood volume and A is the solubility of N 20 in blood.
Hence, the equation (7.4:10),
v= Ddpi - P2)
is reduced to

v + DL(~ + A~) V= DLpo·

The solution of this differential equation is

V = Ce-.' + DLPo,
0(

where
Po
0(=-+-
1
Vo Av"
and C is an integration constant. Since V = 0 when t = 0, we have

C = _ DcPo.
0(

Hence the solution is

7.24 Discuss the changes in the V/Q ratio in the following circumstances: 1) obstruc-
tion in blood vessels, 2) local obstruction in airway, 3) restricted ventilation
in some airways, 4) asthma, 5) edema in the lung, 6) black lung disease or
pneumoconiosis of coal miners, 7) silicosis.
Problems 271

12~-----------------' 12
NORMAL 10 EMPHYSEMA
8 8
Vmax
LPS 6 6
4 4

2
2~
0 07 6 5
6 5 4 3 2 0 4 3 2 0
VL liters VL liters
t t t t t t
TLC FRC RV TLC FRC RV
f-VT-1 t-VT-1
t----VC - - - - - - - i ~VC----I

FIGURE P.7.27 Expiratory flow vs lung volume curves for normal and patients with
emphsema. From Bates et al. (1971)

7.25 Explain why the airway resistance to flow varies inversely with lung volume.
7.26 In forced expiration the pleural pressure can become greater than atmospheric.
Based on the pressure-diameter relationship, Eq. (1) of Sec. 5.7, and the depen-
dence of /Talv on PPL' discuss the consequent change in the airway diameter.
7.27 The left panel of Fig. P7.27 shows a normal maximal expiratory flow vs volume
curve together with a flow rate curve during quite breathing over the tidal volume
range. The right panel shows the maximal expiratory flow vs volume curve
obtained from a patient with emphysema. For the emphysema patient the flow
rates during quite breathing are the maximal ones over the tidal volume range.
Explain the cause of difference of the flow-volume curves of the emphysema
patient in terms of the geometric and structural changes of the lung. (Cf. Bates et
al. (1971), Tisi (1980).)
7.28 Name and describe briefly a few principles by which the velocity of flow of a gas
can be measured.
7.29 Name a few principles by which the pressure in a gas can be measured.
7.30 Consider an airway (A) branching into two daughters, Bl and B2, one of which
(B 1 ) is severely constricted. At a given frequency of ventilation, the velocity of
flow into branch Bl will be slower than that in B2 in inspiration and the volume
of the lung supplied by Bl will increase at a rate slower than that supplied by B2.
In other words, there will be a phase lag. If the compliance C is defined by Eq.
(7.6:7). Show that the compliance of the lung supplied by the airway A will be
frequency dependent. (Cf. Otis et al. (1956).)
7.31 What kind of patient needs oxygen therapy? The physiologist Haldane used to
say that intermittent oxygen therapy is like bringing a drowning man to the
surface-occasionally! Why is it?
272 7 Respiratory Gas Flow

7.32 Design an instrument to measure the volume flow rate of breathing through the
mouth.
7.33 Design an instrument which can measure velocity at a point in the trachea.
7.34 Design an instrument to measure the gas volume of inspiration and expiration.
7.35 Design an instrument by which the volume of inspired or expired gas can be
recorded as a function of time during breathing.
7.36 Errors in the use of respirators is a major cause of accidents in hospitals. If you
were a hospital engineer, what would you do to minimize this problem?
7.37 An effective way to measure the resistance to gas flow in the airway, and the
elastic compliance of the lung is to use the small perturbation method (see Secs.
7.6, 7.7). Define "small" for the small perturbations in pressure and flow. Set a
criterion to test the smallness. Design an instrumentation system which is prac-
tical for patient use.
7.38 Small perturbations of pressure that act on the chest wall but not at the mouth
can be introduced by using a loud speaker in a body plethysmograph box. Small
perturbations of pressure and flow that act on the flow-through the mouth, but
not on the chest can be introduced by using a loud speaker in the respirator
circuit. Through a speaker system one can impose perturbations of sinusoidal or
other wave forms. But if a delta function perturbation is desired (so that the
perturbation is very brief in time), an interrupter may be introduced in the
respirator circuit or a small impact may be used on the chest. Discuss the pros
and cons of these methods from the point of view of clinical use.
7.39 Reliable measurements of the partial pressure of oxygen in resting mammalian
skeletal muscle show that the p02 is about 100 mm Hg in large arteries, 40 mm
Hg in large veins, 15-19 mm Hg in the muscle tissue. Explain why the tissue p02
is so much lower than the venous p02? (Cf. Harris, P.D., (1986). Movement of
oxygen in skeletal muscle. News in Physiol. Sci. 1: 147-149.)
7.40 Pulmonary edema is caused by movement of fluid from blood into the interstitial
and alveolar space. One major problem when dealing with critically ill patients
is the accumulation of edema fluid within the interstitial spaces of the lung. The
excess interstitial fluid reduces the lung's ability to provide oxygen to the body,
thus compromising the recovery of patients already at risk. List the factors that
can influence edema. Discuss the clinical measures that will control edema.
Consider especially the role oflymph flow and its relation to the systemic venous
pressure; and of the effects of infection. (Cf. Laine, G.A., Allen, S.T., Williams,
J.P., Katz, J., Gabel, J.e. and Drake, R.E. (1986). A new look at pulmonary edema.
News in Physiol. Sci. 1: 150-153.)

References

Bates, c.Y., Macklem, P.T., and Christie, R.Y. (1971). Respiratory Function in Disease. Saunders,
Philadelphia.
Bohn, DJ., Miyasaka, K., Marchak, B.E., Thompson, W.K., Froese, A.B., and Bryan, A.C. (1980).
References 273

Ventilation by high frequency oscillation. J. Appl. Physiol.: Respir. Environ. Exer. Physiol.
48: 710-716.
Briscoe, W.A., Forster, R.E., and Comroe, J.H. (1954). Alveolar ventilation at very low tidal
volumes. J. Applied Physiol. 7: 27-30.
Brown, F.T. (1962). The transient response of fluid lines. J. Basic Engineering 84: 547-553.
Chang, H.K. (1984). Mechanics of gas transport during ventilation by high frequency oscillation.
J. Appl. Physiol. 56: 553-563.
Chang, H.K., Cheng, R.T., and Farhi, L.E. (1973). A model study of gas diffusion in alveolar sacs.
Respiration Phys. 18: 386-397.
Chang, H.K., Isabey, D., Shykoff, B.E., and Harf, A. (1984). Gas mixing during high frequency
oscillation. In Biomechanics in China, Japan and USA. (y.c. Fung, J.J. Wang, and E. Fukada,
eds.), Science Press, Peaking, pp. 264-280.
Chang, H.K., Tai, R.c., and Farhi, L.E. (1975). Some simplifications of ternary diffusion in the
lung. Respiration Phys. 23: 109-120.
Chatwin, P.C. (1975). On the longitudinal dispersion of passive contaminant in oscillatory flows
in tubes. J. Fluid Mech. 71: 513-527.
Cumming, G., Henderson, R., Horsfield, K., and Singhal, S.S. (1968). The functional morphology
of the pulmonary circulation. In The Pulmonary Circulation and Interstitial Space (A.
Fishman and H. Hecht, eds.), University Chicago Press, Chicago, pp. 327-338.
Cumming, G. and Semple, S.J. (1973). Disorders of the Respiratory System. Blackwell Sci. Pub.,
London.
Dubois, A.B. (1964). Resistance to breathing. In Handbook of Physiology, Sec. 3 Respiration, Vol.
1 (W.O. Fenn and H. Rahn, eds.). Amer. Physiol. Soc. Washington, D.C. 1964, pp. 451-462.
Fenn, W.O. and Rahn, H. (eds.) (1964). Handbook of Physiology: Sec. 3 Respiration, Vols. 1 & 2.,
1696 pp. American Physiological Society, Washington, D.C.
Fishman, A., Macklem, P., and Mead, J. (eds.) (1986). Handbook of Physiology, Sec. 3. Respiration.
Amer. Physiol. Soc. Washington, D.C.
Fry, D.L. and Hyatt, R.E. (1960). Pulmonary mechanics. A unified analysis of the relationship
between pressure, volume and gasflow in the lungs of normal and diseased human subjects.
Amer. J. Med. 29: 672-689.
Fung, Y.C. (1977). A First Course in Continuum Mechanics, 2nd edn. Prentice-Hall, Englewood
Cliffs, N.J.
Fung, Y.c. (1983). Biodynamics: Circulation. Springer-Verlag, New York.
Henderson, Y., Chillingworth, F.P., and Whitney, J.L. (1915). The respiratory dead space. Amer.
J. Physiol. 38: 1-19.
Hirschfelder, J.O., Curtis, C.F., and Bird, R.B. (1954). Molecular Theory of Gases and Liquids.
Wiley, New York.
Hyatt, R.E., Schilder, D.P., and Fry, D.L. (1958). Relationship between maximum expiratory flow
and degree oflung inflation. J. Appl. Physiol. 13: 331-336.
Jaeger, M.J. and Matthys, H. (1968/69). The patten of flow in the upper human airways. Respiration
Phys.6: 113-127.
Krogh, M. (1914/15). The diffusion of gases through the lungs of man. J. Physiol (London) 49:
271-300.
Lee, J.S. (1984). The mixing and axial transport of smoke in oscillatory tube flows. Annals of
Biomed. Eng. 12: 371-383.
Lee, J.S. (1984). A transient analysis of gas transport in oscillatory tube flows. In Biomechanics in
China, Japan, and USA (Y.c. Fung, J.J. Wang, and E. Fukada, eds.). Science Press, Peking,
pp.254-263.
Otis, A.B., McKerrow, C.B., Bartlett, R.A., Mead, J., McIlroy, M.B., Selverstone, N.J., and
Radford, E.P., Jr. (1956). Mechanical factors in distribution of pulmonary ventilation. J.
Appl. Physiol. 8: 427-443.
Pedley, T.J., Schroter, R.C., and Sudlow, M.F. (1971). Flow and pressure drop in systems of
repeatedly branching tubes. J. Fluid Mech. 46: 365-383.
274 7 Respiratory Gas Flow

Pedley, T.J., Schroter, R.C., and Sudlow, M.F. (1977). Gas flow and mixing in the airways. In
Bioengineering Aspects of the Lung (J.B. West, ed.), Marcel Dekker, New York, pp. 163-265.
Prasad, S.N. and Herrmann, G. (1969). The usefulness of adjoint systems in solving nonconserva-
tive stability problems. Int. J. Solids and Struct. 5: 727-735.
Prasad, S.N. and Herrmann, G. (1972). Adjoint variational methods in nonconservative stability
problems. Int. J. Solids and Struct. 8: 29-40.
Pride, N.B., Permutt, S., Riley, R.L., and Bromberger-Barnea, B. (1967). Determinants of maximal
expiratory flow from the lungs. J. Appl. Physiol. 23: 646-662.
Roughton, F .J. W. and Forster, R.E. (1957). Relative importance of diffusion and chemical reaction
rates in determining rate of exchange of gases in the human lung. J. Appl. Physiol. 11:
291-302.
Scherer, P.W. and Haselton, F.R. (1982). Convective exchange in oscillatory flow through
bronchial-tree models. J. Appl. Physiol.: Respir. Environ. Exer. Physiol. 53(4): 1023-1033.
Schmid-Schoenbein, G. and Fung, Y.c. (1978). Forced perturbation of respiratory system. (A)
The traditional model. Annals of Biomed. Eng. 6: 194-211. (B) A continuum mechanics
analysis. ibid, 6: 367-398.
Seguchi, Y., Fung, Y.c., and Maki, H. (1984). Computer simulation of dynamics offluid-gas-tissue
systems with a discretization procedure and its application to respiration dynamics. In
Biomechanics in China, Japan, and USA (Y.c. Fung, E. Fukada, J.J. Wang, eds.). Chinese
Science Press, Beijing, pp. 224-239.
Seguchi, Y., Fung, Y.c. and Ishida, T. (1986). Respiratory Dynamics-Compter simulation. In
Frontiers in Biomechanics (G.W. Schmid-Schonbein, S.L.Y. Woo, B.M. Zweifach, eds).
Springer Verlag, New York, pp. 377-391.
Slutsky, A.A., Drazen, J.M., Ingram, R.H., Jr., Kamm, R.D., Shapiro, A.H., Fredberg, J.J., Loring,
S.H., and Lehr, J. (1980). Effective pulmonary ventilation with small-volume oscillations at
high frequency. Science 209: 609-611.
Slutsky, A.S., Kamm, R.D., Rossing, T.H., Loring, S.H., Lehr, J., Shapiro, A.H., and Ingram, R.H.,
Jr. (1981). Effects of frequency, tidal volume, and lung volume on CO2 elimination in dogs
by high frequency (2-30 Hz) low tidal volume ventilation. Clin. Invest. 68: 1475-1484.
Stewart, G.N. (1893). Researches on the circulation time in organs and on the influences which
affect it. J. Physiol. (London) 15: 1-30. II. Time of the lesser circulation, 15: 31-72. III. Thyroid
gland, 15: 73-89. The output of the heart, 22: (1900) 159-173.
Streeter, V.L. and Wylie, E.B. (1967). Hydraulic Transients. McGraw-Hill, New York.
Taylor, G.I. (1953). Dispersion of soluble matter in solvent flowing slowly through a tube. Proc.
Roy. Soc. (London), Series A, 219: 186-203.
Taylor, G.I. (1954). The dispersion of matter in turbulent flow through a pipe. Proc. Roy. Soc.
(London), Series A., 223: 446-468.
Tisi, G.M. (1980). Pulmonary Physiology in Clinical Medicine. Williams and Wilkins, Baltimore,
Maryland.
Van der Pol, B. and Bremmer, H. (1950). Operational Calculus, Cambridge University Press,
London/New York, pp. 305-307.
Wagner, P.P. and West, J.B. (1972). Effects of diffusion impairment on O 2 and CO 2 time courses
in pulmonary capillaries. J. Applied Physiol. 33: 62-71.
Weibel, E.R. (1963). Morphometry of the Human Lung. Academic Press, New York.
West, J.B. (1974). Respiratory Physiology- The Essentials. Williams and Wilkins, Baltimore.
West, J.B. (1982). Pulmonary Pathophysiology-the essentials. 2nd ed. Williams and Wilkins,
Baltimore.
West, J.B. and Wagner, P.D. (1977). Pulmonary gas exchange. In Bioengineering Aspects of the
Lung (J.B. West, ed.), Marcel Dekker, New York, pp. 361-457.
Whittle, P. (1971). Otpimization Under Constraints: Theory and Applications of Nonlinear Pro-
gramming. Wiley-Interscience, London, New York.
Winter, D.C. and Nerem, R.M. (1984). Turbulence in pulsatile flows. Annals of Biomed. Eng. 12:
357-369.
CHAPTER 8

Basic Transport Equations According

to Thermodynamics, Molecular
Diffusion, Mechanisms in Membranes,
and Multiphasic Structure

8.1 Introduction

Now we shall consider the movement of water and other fluids in our bodies,
especially the exchange of fluid between blood and the extravascular tissues.
Red blood cells cannot leave the blood vessel; but water, ions, and some white
blood cells can. The fluid in the extravascular space moves and exchanges
matter with the cells in the body. The ionic composition of the fluid in the
cells is quite different from that in the extracellular space. Extracellular fluid
is rich in Na +, Cl-, RCO:;, whereas the intracellular fluid is rich in K +, Mg++,
phosphates, proteins, and organic phosphates. The composition of blood
plasma is fairly similar to that of the extravascular fluid, except that the plasma
has some 14 mEqjL of proteins while extracellular fluid has essentially none.
See Table 8.1:1. To talk about mass transport in the body we must explain
how this difference in composition comes about.
In man, the total body water accounts for 45-50 percent of the body weight
in adult females and 55-60 percent of the body weight in adult males. Approxi-
mately 50 percent of this water is in muscle, 20 percent in the skin, 10 percent
in the blood, and the remainder in the other organs. The total body water is
contained in two major compartments which are divided by the cell membrane:
the cell water and the extracellular water (Table 8.1:2). The extracellular
compartment is subdivided into several sub compartments: the interstitial-
lymphatic compartment, the vascular compartment, the bone and dense con-
nective tissue compartment (cartilage and tendons), and the transcellular water
compartment (epithelial secretions). In this chapter we shall mostly be con-
cerned with the interstitial-lymphatic compartment which communicates
directly with the vascular compartment.
Water and solutes are continually exchanged between these fluid compart-

275
276 8 Basic Transport Equations

TABLE 8.1:1 Ionic composition of body water compartments*

Extravascular Intracellular Normal
Extracellular fluid of skeletal blood
fluid muscle cell plasma
(mEqjL) (mEqjL) (mEqjL)
Na+ 145.1 12.0 142.0
K+ 4.4 150.0 4.3
Ca++ 2.4 4.0 2.5
Mg++ 1.1 34.0 1.1
Cations Total 153.0 200.0 149.9
Cl- 117.4 4.0 104.0
HC03" 27.1 12.0 24.0
HPO~-, H 2 PO; 2.3 40.0 2.0
Proteins 0.0 54.0 14.0
Other 6.2 90.0 tt 5.9
Anions Total 153.0 200.0 149.9
* Data from D.M. Woodbury, in Howell and Fulton's Physiology and Biophysics, 20th Ed., T.e.
Ruch and H.D. Patton (eds.). Saunders, Philadelphia, 1974.
tt This
largely represents organic phosphates such as ATP.

TABLE 8.1:2 Distribution of total body water in an average

70-kg male man, according to Woodbury, loco cit., supra
Expressed in percent body weight and approximate volume
Cell water 36.0% 25 liter
Extracellular water
Interstitial fluid 11.5% 8 liter
& lymph
Blood plasma 4.5% 3 liter
Bone 3.0% 2 liter
Dense connective
tissue 4.5% 3 liter
Transcellular water* 1.5% 11iter
Total body water ~ 60 percent body weight
*The transcellular water consists of epithelial secretions such as the
digestive secretions, sweat, and cerebrospinal, pleural, synovial, and
intraocular fluids.

ments. This exchange occurs by passive and active mechanisms. The move-
ment of particles is passive if it develops spontaneously and does not require
a supply of metabolic energy, or active if it depends upon energy derived from
metabolic processes. In humans, solute movement occurs by both passive and
active mechanisms whereas all water movement is passive. Active transport
8.2 The Laws of Thermodynamics 277

requires a certain machinery, certain biochemical processes occurring in some

specific physical space. The space is usually the cell membrane and membranes
within cells. Hence mass transport in membranes has many unique features.
This extensive subject will be treated in two chapters. In this chapter we
shall consider the basic equations that describe the transport phenomena: the
constitutive equations and the phenomenological laws. In the next chapter,
applications of these basic equations will be considered.
The phenomenological laws are empirical equations formulated with the
guidance of theoretical considerations. The most general theoretical founda-
tion is thermodynamics, with the concepts of entropy production, Onsager's
principle, and molecular motion. Greater details are obtained, however, by
consideration of specific molecular and structural models. Hence we begin
this chapter with thermodynamics (Sec. 8.2), emphasizing chemical potential
because of its central importance on mass transport (Secs. 8.2-8.4), then
discuss the internal entropy production (Sec. 8.5) and phenomenological laws
of diffusion, osmotic pressure, Darcy, Fick, and Starling (Sec. 8.6). We then
proceed to molecular theory (Sec. 8.7), add detailed mechanisms in biological
membranes (Sec. 8.8) and solid immobile matrix (Sec. 8.9).

8.2 The Laws of Thermodynamics

One of the ways to arrive at a suitable formulation of the phenomenological

laws of fluid motion in interstitial space is by thermodynamics. In this section,
a brief review of thermodynamics is given with a view to clarify the concept
of chemical potential.
Classical thermodynamics is based on the following axioms:
1) conservation of matter,
2) an isolated system tends toward equilibrium,
3) the first law: conservation of energy,
4) the second law,
5) the third law.
By 2), thermometers are used to characterize empirical temperature, and
calorimeters are used to quantify heat. For a given system, a small increment
of heat (dQ) is proportional to a small change of temperature (dt), so that
dQ = Cdt, (1)
where C is the heat capacity. The accepted unit of heat is calorie, the heat
which must be imparted to 1 gm of water to raise its temperature by 1DC at
15 DC and atmospheric pressure. Davy, Joule, and others established the me-
chanical equivalent of heat: one calorie equals 4.1868 Joule. They showed that
if a small amount of heat dQ is imparted on a body, the body would change
its state and do work:
dQ =dU -dW (2)
278 8 Basic Transport Equations

The variable U is a function of the state of the body; W is the work done by
the system on the surrounding. Equation (2) is a statement of the first law of
thermodynamics.
Joule and others, in establishing the first law, experimented on many forms
of work and energy: mechanical, electrical, chemical, etc. If the body is a solid
and the stress in the body is (Jjk (free Latin indexes range over 1, 2, 3), and the
body deforms so that its strain is changed by dejk , then the work done by the
stress is (Jjk dejk per unit volume, or (Jjk dejk V in a body of volume V. * If a force
F acts on a material particle which is moved by a displacement dx, then the
work done is F . dx. ** If a muscle shortens by an amount - dL under a force
:JF, it performs work equal to -:JF. dL. If a quantity of electricity - de is given
off by a system at an electric potential t/J, an electric work t/J de is performed.
If dn i mole of the ith species of chemicals is transported into the system from
the surrounding, a chemical energy Jli dn i is added to the system (the constant
of proportionality Jli is called the chemical potential). Thus, on denoting the
change of energy by dU for a solid body containing muscle fibers and subject
to a body force F, an electric potential t/J, and a transfer of N species of
chemicals No.1, 2, ... , N into the body by the amounts dn 1 , dn 2 , ••. , dn N , we
have, by the first law:
dQ = dU + Jla dna - :JF . dL + F· dx - t/J de + (Jjk dejk V. (3)
The Greek index rJ. ranges over 1, 2, ... N. The summation over rJ. includes all
chemical species of the system. The energy dU includes kinetic, potential, and
internal energies.
The second law of thermodynamics may be stated as follows:
There exists two single-valued functions of state T, called the absolute
temperature, and S, called the entropy, with the following properties:
I. T is a positive number which is a function of empirical temperature only.
II. The entropy of the system is equal to the sum of the entropies of its parts.
III. The entropy of a system can change in two distinct ways: by transfer from
the surrounding and by internal changes. Thus
dS deS diS
-
dt
= dt
- +dt- ' (4)

where dS/dt denotes the rate of increase of entropy of the system, deS/dt
denotes the rate of transfer of entropy from the surrounding, diS/dt denotes
the rate of internal entropy production.
IV. The internal entropy production is never negative. If it is zero, the
process is said to be reversible. If it is positive, the process is said to be
irreversible.

* The summation convention is used; see Sec. 1.7. The equations derived below can be used for
a gas if the term Ujk dejk is replaced by - p dV.
** Boldface letters denote vectors.
8.2 The Laws of Thermodynamics 279

V. If dQ is the heat absorbed by a system in a reversible process, then the

entropy of the system is changed by the amount:

dS = dQ. (5)
T
The absolute temperature T and the entropy S are defined completely by their
properties as expressed in the second law. For T, William Thomson (Lord
Kelvin) has shown how to calibrate any thermometer into absolute tempera-
ture scale on the basis of the second law. For S, the law says that it is an
attribute of a material body. Its units are Joules per degree Kelvin. For
example, a kilogram of saturated steam at a pressure of 1 atm (101.3 kN/m2)
and a temperature of 373.16 K has an entropy of7.358 kJ/Kjkg. Water at the
same pressure and temperature has an entropy of 1.307 kJ/K/kg.
The physical meaning of entropy has intrigued everybody since its intro-
duction by Clausius in 1865. A thought that is conducive to an intuitive idea
about entropy comes from statistical mechanics. Boltzmann has shown that
entropy is proportional to the number of configurations 0 in which a system
can be realized, according to the relation
S = kInO. (6)
Here k is the Boltzmann constant. An increase in the number of configurations
that a system can assume increases its entropy, while any restriction in the
modes of realization decreases entropy. Thus, mixing, disorganization, and
randomization increase entropy, while organization and ordering decrease it.
The second law states how entropy of a system is changed, but it does not
say how an absolute value can be assigned to entropy. The assignment of an
absolute value of entropy is done by the third law of thermodynamics: The
entropy in the state T = 0 is equal to zero in every system occurring in nature.
This statement is Planck's formulation of Nernst's postulate.
Example. Entropy of a perfect gas. A perfect gas obeys the equation of state
p V = R T, with R a constant, and has a constant specific heat Cv. Hence for
one mole of gas, we have, on defining U, V, Cv, R for one mole,

dS = dQ = dU + pdV = CvdT + (RT/V)dV.

T T T

An integration and use of the equation of state gives

S = Cv log T + R log V + So
= Cvlogp + Cplog V + Sb (7)
= Cplog T - Rlogp + s~,
where So, Sb and S~ are constants.
280 8 Basic Transport Equations

8.3 The Gibbs and Gibbs-Duhem Equations

Combining Eqs. (3) and (5) of Sec. 8.2, we obtain the Gibbs equation for a tissue
containing muscle fibers with tension [IF, subject to external force F, electric
potential t/I, and transfer of N species of chemicals of mass dn l , dn 2, ... , dnN
into the tissue,
dU = TdS + J1.,.dn,. - [IF'dL + F·dx - t/lde + O'jkdejkV. (1)
This Gibbs equation can be integrated. A method for doing so is based on
Euler's theorem on homogeneous function. A function ct>(x, y, z) is said to be
homogeneous of the mth order if
ct>(AX, AY, AZ) = Amct>(X, y, z). (2)
For example, ax + by is homogeneous of the first order, ax 3 + bx 2 y is homo-
geneous of the third order. For such a function, Euler showed that

mct> = x (oct» + y (oct» + Z(oct» (3)

ox Y,z oy Z,x oz x,y

The proof is very simple. Differentiation of the left-hand side ofEq. (2) yields

oct>(h, AY, AZ) = (oct> ) (oct» (oct» (4)

OA OAX x + OA}' Y + OAZ z.
Differentiation of the right-hand side of Eq. (2) yields
OAmct>(X,y,z) _ 1m- l ct>( )
(5)
OA -mil. x,y,Z.

Equating Eqs. (4) and (5), and setting A = 1, we obtain Eq. (3).
Now apply Euler's theorem to Eq. (1), the Gibbs equation, noting that
energy, volume, entropy, and chemical quantity are all proportional to the
mass of the system. Hence, if the internal energy is expressed as a function
U(S, ejk V, nl, n2'" .), then U is homogeneous of order 1:
(6)
therefore according to Euler's theorem, U(S, ejk V, n l , n2, ... ) can be written as
follows:

U= S(~~) + ejk V(O:j~V ) + ~ n,.(:~) + .... (7)

Hence, for small changes of the variables we have

dU = dS(~~) + dejk V(O:j~V ) + ~ dn,.(:~) + .... (8)

Comparing Eqs. (8) and (1), we see that

8.4 Chemical Potential 281

(au)
as V,n, -
- - T.
, (aeau)
--
jk v S,n,
= (Jjk,
(9)

G~)s,v,nl,n2"" ~a' =

etc. Introducing these into Eq. (7), we obtain the integrated form of the Gibbs
equation:

Now if we differentiate Eq. (10), we obtain

dU = T dS + S dT + L (~a dna + na d~a) + .... (11)

On subtracting Eq. (1) from Eq. (11), we obtain the Gibbs-Duhem equation:
Lnad~a + SdT- L'd@, + x·dF - edljJ + ejk d(Jjk V = O. (12)

If experiments were done in such a manner that T, @', F, 1jJ, and (Jjk remain
constant, then
(13)

These equations are used extensively in mass-transport analysis (Sec. 8.5).

8.4 Chemical Potential

Let us define
F = U - TS (1)
as the Helmholtz free energy, and
G = U - TS - (Jj"ej" V (2)
as the Gibbs free energy (or Gibbs thermodynamic potential). Then, on sub-
stituting Eq. (10) of Sec. 8.3 into Eqs. (1) and (2) above, taking differentials of
the resulting expressions and combining them with the Gibbs equation, Eq.
(8.3:1), we obtain
dU = TdS + L~adna -@"dL + F·dx -ljJde + (Jj"dej"V, (3)

Thus we see that

282 8 Basic Transport Equations

(6)

In other words, the chemical potential I-l" of the rxth species of chemicals is the
rate of change of internal energy U with respect to the mass of the rxth species
of the chemical when entropy, strain, muscle length, electric charge, and concen-
trations of all other chemicals, n p, (p -=f. rx) are held constant. Similarly, it is the
partial derivative of the Gibbs thermodynamic potential, oG/on", when the
temperature, stress, etc. are held constant; or the partial derivative of the Helm-
holtz free energy of/on,,, when the temperature, strain, etc. are held constant.
Now, in chemical and biological laboratory experiments, it is relatively easy
to maintain a constant temperature by a heat bath, and a constant pressure
by exposure to the atmospheric pressure. If the material is a solid or liquid
and is left stressfree in atmosphere at constant temperature, then the measure-
ment of the chemical potential is best done by measuring the Gibbs thermo-
dynamic potential under isothermal and isobaric conditions. If the material
is a gas and the container volume is fixed, then it is best done by measuring
the Helmholtz free energy under isothermal condition.
In the chemical literature, following G.N. Lewis, the chemical potential of
a substance is often expressed in terms of activity defined by the following
equation:
I-l = I-ls + RT log a (7)
in which a is called activity, T is the absolute temperature, R is the gas
constant, and I-ls is the chemical potential of the substance at a standard state
at the same temperature T. The standard state can be chosen for each chemical
component independently and arbitrarily with respect to other components
in the system. For example, the activity ofliquid water is 1 at 1 atm pressure,
but at 100 atm pressure it is 1.0757, 1.0728, and 1.0703 at 25°C, 37°C, 50°C,
respectively.
Example 1. Perfect gas. By using Eq. (7) of Sec. 8.2 for S, and dQ = Cp dT
for isobaric process, show that the Gibbs thermodynamic potential G per unit
mass, which, by Eq. (6), is the chemical potential of the perfect gas, is
I-l = I-ls(T) + RTlogp, (8)
where

I-ls(T) = pV + r CpdT _ T Jor CpdT

Jo
T

T
+ T
Uo - TS~, (9)

Uo being a constant. Comparing Eqs. (7) and (8), we see that the activity of a
perfect gas is equal to the pressure. In a real gas, experimentally determined
activity replaces the role of pressure in the chemical potential.
Example 2. A mixture of perfect gases. It follows from the gas law that the
pressure of a mixture of perfect gases is equal to the sum of the partial pressures
8.5 Entropy in a System with Heat and Mass Transfer 283

ofthe individual gases if there is no chemical reaction between the gases. Hence
the internal energy, entropy, and Gibbs free energy of each gas are the same
as if that gas were alone in a vessel, and the chemical potential of a gas species
a in the mixture is given by Eq. (8), with the addition of a subscript a to 11, Ils'
p, Cp , Uo and S~ to indicate their belonging to the species a. Let the mole
number of the gas species a be na , and denote by
na
= a = 1, ... , k (10)
Ca,
n1 + n2 + ... + nk ,
the concentration of the component a in the mixture. Then
Pa = na RT/ V = CaP, (11 )
and we can write the chemical potential of the gas species a as
Ila = Ilsa( T) + R T log Ca + R Tlog p. (12)
It is seen that the product of the concentration and the pressure is the activity.
In a real gas mixture, experimentally determined activities replace the CaP
terms in the chemical potential.

°
Example 3. Dilute solution. For a dilute solution consisting of chemically
pure substances 0, 1, 2, ... k, of which the component is the most plentiful,
denote the mole numbers by no, n 1 , ... , nk , respectively. A solution is dilute
if the internal energy and volume of the solution are linear functions of na/no,
(a = 1,2, ... , k). It can be shown (Epstein, 1937, Chap. 9) that the Gibbs
thermodynamic potential of a dilute solution is
k
G= L na(ga + R Tlog Ca) (13)
a=O

in which Ca is the concentration, ga = Ua - TSa + pVa is a function of p and T,

independent of na. The role of concentration in the chemical potential is seen
again. In real solution, activity replaces the concentration.

8.5 Entropy in a System with Heat and Mass Transfer

We now consider a system in which the thermo mechanical variables are

functions of time and space. The system is not necessarily in equilibrium, and
heat and mass transfer take place within. To analyze such a system, one makes
the basic assumption that the Gibbs equation for internal energy remains valid
locally at any time and place, that is, the internal energy as a function of the
state variables is the same whether the state variables are arrived at by
reversible process or not. Thus, focusing our attention on heat and mass
transfer, and ignoring for the time being th muscle contraction, body force,
electric discharge, and strain energy, we have, for a small volume Ll V of the
system,
284 8 Basic Transport Equations

(1)
"
according to Eqs. (8.2:2) and (8.3:1). Further analysis is facilitated by defining
the entropy per unit volume sv, concentration of a chemical species c", and
the heat per unit volume qv, by the expressions

(2)

Thus Eq. (1) may be written as

Tds v = dqv - L j1."dc". (3)
"
If the system moves and has a velocity field v' which varies from place to place,
then Eq. (3) can be interpreted, in the notation of continuum mechanics, Sec.
1.7, as

TDsv = Dqv _" Dc" (4)

Dt Dt ': j1." Dt '
where D/Dt denotes a derivative with respect to time of a function (such as sv)
belonging to a fixed set of material particles. D/Dt is called the material
derivative. It can be shown (see any book on continuum mechanics, e.g. Fung,
1977) that the material derivative of a function F is
DF of ,
Dt = at + v . grad F. (5)

If the system has no bulk flow so that v' = 0, then Eq. (4) is reduced to
TOsv _ oqv " oc" (6)
at - at - ': j1."at·

Analysis of Entropy Production

When heat and mass transfers occur we define heat or mass flux as the amount
of heat or mass passing through a surface of unit area. Consider a volume V
enclosed in a surface A fixed in space. The rate of increase of the mass of a

i
chemical species Q( in V is due to the flow of matter through the boundary A:

on"= - J 'vdA (7)

at
-
A" ,

where J" is the mass flux vector of species Q( through a surface of area dA out
of the volume V; v is a unit vector normal to the surface pointing outward
from V. But if c" is the concentration of species Q( we have, by definition,

n" = Iv c"dV. (8)

8.5 Entropy in a System with Heat and Mass Transfer 285

Furthermore, by Gauss's theorem, we have

LJa·VdA= IvdiVJadV. (9)

Hence, by Eqs. (8) and (9), Eq. (7) is reduced to

Iv a;; dV = - Iv divJadV. (10)

Since the control volume V is arbitrary, we must have

aC =
ata J
- d·IVa· (11)

This equation expresses the conservation of the mass of the chemical species ()(.
Similarly,

-aqv
at
= - d.IVJ q (12)

expresses the conservation of heat, J q being the heat flux vector. On the other
hand, according to the second law of thermodynamics, the change of entropy
in a system consists of two parts. One part is transported through the boundary;
the other part is produced locally and internally. The first part is transferred by
the entropy flux vector, J s . The second part is the diSjdt term in the second
law, Eq. (8.2:4). Thus

as v = -div J + dis V •
(13)
at s dt

Substituting these expressions into Eq. (6) and dividing through by T, we

obtain

-dis v - d·IV J = - -
1 d·IV J L
+ -J-la d·IV J a. (14)
dt s T qaT

This equation may be modified by making use of the following relation which
holds for a scalar a and a vector b:
divab = adivb + b· grada.
Thus Eq. (14) can be written as

-d . J q + Jq·grad-
dis v - d.IVJ = -dIV- 1+" ( . J-laJa J-la)
L... dIV-- - Ja·grad- . (14a)
s
t T TaT T

The right-hand side can be divided into two parts: a divergence of a vector
and a scalar product of two vectors. Now we shall interpret Eq. (14a) by
splitting it into two parts: We equate the divergence on the two sides of the
equation to obtain
286 8 Basic Transport Equations

· J = d·IV -J q
dIV - d·IV L-
f.1I1. J
- I1. (15)
s T 11. T'
and we equate the entropy production with the scalar products so that

1 + L J ·grad ( -f.111.)
-dis v = J . grad- - . (16)
dt q T 11. 11. T

Equation (15) shows that the flow of entropy across a surface is

J = Jq _ " f.1I1. J I1. (17)

S T '7 T

as suggested directly by Eq. (3). Equation (16) is of the form

dis v = "L...JkXk'
-d (18)
t k

where X k can be called the generalized force "conjugate" to the flux J k • Thus,
by comparing Eq. (18) with Eq. (16), we obtain the important result that the
generalized force which is conjugate to the flux of a chemical species is the
negative gradient of the chemical potential of that species divided by T. This is
why the chemical potential is so important in mass transfer phenomena.

Phenomenological Laws and On sager Principle

The relationships between the fluxes J 1 , J 2 , ••• , J k and the conjugate general-
ized forces Xl' X 2 , ••• , X K , (K being the number of irreversible processes
involved), are called phenomenological laws. If the phenomenological laws are
linear we can write
K
Jk = L LkmXm,
m=l
k, m = 1, 2, ... , K, (19)

where L km are constants. Experience with known phenomenological laws in

thermoelasticity, thermodiffusion, piezoelectricity, etc. has shown that the
matrix of the coefficients L km is symmetric, i.e.
k, m = 1, 2, ... , K. (20)
Onsager derived Eq. (20) from statistical mechanics, and Eq. (20) is known as
Onsager's principle or the Onsager reciprocal relations.
If we substitute Jk from Eq. (19) into Eq. (18), we obtain the rate of entropy
production:

(21)

According to the second law of thermodynamics, the entropy production is

nonnegative. Hence the second degree polynomial on the right-hand side of
Eq. (21) is positive definite. This, together with Eq. (20), imposes restrictions
8.6 Diffusion, Filtration, and Fluid Movement 287

on the coefficients L km • For example, if L km is collectively considered to be a

square matrix, then the diagonal terms L ll , L 22 , ... , LKK must be positive,
the determinant of the matrix must be positive, and the sum of the principal
minors of any given order must be positive (see Fung, 1965, pp. 8,29, 30).

8.6 Diffusion, Filtration, and Fluid Movement in

Interstitial Space from the Point of View
of Thermodynamics

Let us now consider the movement of water and macromolecules in the

interstitial space at constant temperature. We shall follow Zweifach and
Silberberg (1979) to consider a simplified model of the interstitium consisting
of three groups of components: a "solvent" component, a diffusible "solutes"
component, and a "matrix" component. The solvent is essentially water, the
solutes are principally macromolecules. (Small molecules that can move freely
through the tissue space do not create sufficient driving force to move water
and macromolecules and can be lumped with the solvent.) The matrix consists
of collagen, hyaluronate, etc., see Fig. 8.6:1. Our formulation follows Zweifach
and Silberberg (1979), but greatly simplifies it.

• FERRITIN

SERUM ALBUMIN

COLLAGEN (triple helical strand)

...
• FIBRIL (cross -sec/ion)

•i·..•..e\
. ..•....•....
' . .. '
•

.. . .... ..
FIBER (cross-section)

FIGURE 8.6:1 An idealized interstitium consisting of collagen fibers, hyaluronic acid

molecules, globular proteins, ferritin, and serum albumin. From Zweifach and Silber-
berg (1979), by permission.
288 8 Basic Transport Equations

Let us use the subscripts 0, 1, and 2 for the three components, water, solutes,
and matrix, respectively. Then according to Eqs. (8.5:16-20) and under the
assumptions of constant temperature and linear phenomenological laws, we
can write*
J o = -LoogradJlo - L 01 gradJll - L 02 gradJl2
J 1 = -L 10 gradJlo - LllgradJll - L 12 gradJl2 (1)
J 2 = - L 20 grad Jlo - L21 grad Jll - L22 grad Jl2
in which L oo , L 01 , etc. are constants obeying Onsager's reciprocal relations
and the positive-definiteness of entropy production, Eqs. (8.5:20) and (8.5:21).
Chemical potentials of water, solute, and matrix are functions of tempera-
ture, pressure, and composition. According to Eq. (8.4:12), at constant tem-
perature, we have
Jl" = Jl!O) + V"p + RTloga", oc = 1,2,3, (2)
where R is the gas constant, v" is the molecular volume (see Eq. (13) of Sec.
8.4) of the species oc, and a" is its activity.
The chemical potentials of all the components in a system cannot vary
independently. They must obey the Gibbs-Duhem equation, Eq. (8.3:12).
Thus, if the temperature, pressure, and stress deviations are kept constant,
then
3
L c" d(log a,,) = 0,
,,=1
(3)

c" being the concentration of the species oc. Using Eq. (3), one of the d(log a,,)
terms, say, that of water, can be isolated and expressed as a linear function of
all the others. Thus

(4)

Zweifach and Silberberg (1979) define III and Il2 by the equations

C2' T = const.
(5)
Cl' T = const.
It will be shown later that these II's can be revealed as osmotic pressure with

* Here we treat the "immobile" matrix as a fluid also. This leads to greater mathematical
simplicity. If the immobile matrix must be treated as a solid, be it elastic, viscoelastic or otherwise,
then one must consider stress tensor in place of pressure, and a much more detailed treatment
has to be given to the matrix component. This is done in Sec. 8.9, which has applications to dense
connective tissues such as cartilage, spinal discs, etc.
8.6 Diffusion, Filtration, and Fluid Movement 2S9

suitable semipermeable membrane. From Eqs. (2), (4) and (5) we can write
(6a)

(6b)

A similar equation for dl12 is obtained from (6b) by replacing the subscript 1
by 2. An integration of these equations gives
110 = vo(p - ITl - IT 2 ) + 11&°),
(7)

With these equations, we shall consider some simplified cases below.

Dilute Solutions
For a dilute solution, the activity a", is equal to the concentration of the species
(Sec. S.4). Then, with C l and C2 small compared with Co, Eqs. (2) and (4) yield
IX

RT RT
grad 11o = vogradp - ~gradcl - ~gradc2' (Sa)
Co Co

RT
gradl1l = V 1 gradp + ~gradc1' (Sb)
C1

and a similar equation for 11m by replacing the subscript 1 by 2 in Eq. (Sb). In
a nonuniform solution, the fluxes are given by Eq. (1).

Fick's Law of Diffusion

If there is no matrix, no pressure gradient, and no bulk flow, then the second
equation of Eq. (1) is reduced to
J 1 = -const·grad I11· (9)
For a dilute solution, Eq. (9) then yields
J 1 = - D grad C 1 (10)
which is Fick's law of diffusion. D is a constant called the coefficient of diffusion.

Osmotic Pressure
If a solution of chemical species 1 and 2 is put in equilibrium with pure water
across a membrane which is not permeable to 1 and 2, but is permeable to a
water, then when there is no flow of water across the membrane, the chemical
potential of water on the two sides of the membrane must be balanced. On
290 8 Basic Transport Equations

equating J.lo on the "outside" and "inside" of the membrane, via Eq. (7) we have

so that
Pin - Pout = Il lin + Il 2in · (11)
This represents the osmotic pressure of the solution. The pressure difference
has to be mechanically compensated, which is done in this case by the tension
and curvature of the membrane. It is interesting to note that although III and
Il2 are measures of the concentrations C 1 and C 2 of species 1 and 2, respectively,
they do have the units of pressure.
To reveal chemical potential as osmotic pressure a semipermeable mem-
brane is required. One should remember this in order to avoid confusion. For
example, solute concentration can be measured by freezing point depression.
Solute-free water freezes at O°C. If 1 osmol of any solute is added to 1 kg of
water, the freezing point of this water will be depressed by 1.86°C. (1 osmol is
defined as one gram molecular weight of any nondissociable substance and
contains 6.02 x 10 23 molecules.) But one must realize an important difference
between the osmotic pressure of a solution and its osmolality as measured by
the freezing-point depression. When the freezing-point depression is measured,
all solutes contribute in relation to their concentration, including those that
are ineffective osmoles. In contrast, ineffective osmoles do not contribute to
the osmotic pressure of solution. For example, plasma osmolality of 280
mosmoljkg represents a potential osmotic pressure of 5404 mm Hg. However,
almost all the plasma solutes (primarily Na+ salts) are ineffective since Na+
salts are able to cross the capillary wall separating the plasma from the inter-
stitial fluid. The net effective plasma osmolality is only about 1.3 mosmoljkg
(generating an osmotic pressure of 25 mm Hg) and is due to the plasma
proteins which are restricted to the vascular space.
As another example, consider the cells. The cell membrane is permeable to
Na + but the cell Na + concentration is maintained at 12 mEqjL by active
transport ofNa+ out of the cell. Thus the cell membrane is effectively, though
not actually, impermeable to Na +. In this setting, an osmotic pressure is
generated in the extracellular fluid, and these solutes are called effective
osmoles. If the solute is able to cross the membrane and reach concentration
in both compartments, e.g., urea, then no osmotic pressure is generated across
the membrane; hence urea is an ineffective osmole for cells.

Starling's Law of Membrane Filtration

Consider two solutions separated by a semipermeable membrane. Each solu-
tion consists of a solute in water. Assume first that the membrane is impermeable
to the solute but is permeable to water. The chemical potentials of the water
on the two sides of the membrane (indicated as "inside" and "outside") are,
according to Eq. (27),
8.6 Diffusion, Filtration, and Fluid Movement 291

(12)
Here, II stands for the sum II! + II2 if an immobilized matrix exists, or II!
alone if such a matrix does not exist. At equal temperature .u&O?n and .u&OLt are
equal. For dilute solutions, v o in and Vo out are approximately equal. Hence, the
difference ofthe chemical potentials of water on the two sides ofthe membrane
is
(13)
This drives the water across the membrane. Let the flux of water across the
membrane be 1 o (volume rate of flow per unit area of the membrane). If we
assume that flux is linearly proportional to the driving force, then we obtain
(14)
where K is the permeability constant of the membrane. This is Starling's law of
membrane filtration.
If the membrane is also permeable to the solute, then there will be a flux of
the solute across the membrane. Let us use the symbol Ll for the difference of
a quantity on the two sides of the membrane. We may write, in analogy with
Eqs. (1), and using Eq. (7),

= K(Llp - O'LlII), (15)

where K and 0' are constants, Llp = Pin - Pout, and LlII = II in - II out . 0' is
called the reflection coefficient: 0' = 1 when the membrane is impermeable to
the solute, 0' < 1 when it is permeable to the solute.

Darcy's Law of Flow through Porous Media without

Chemical Gradient
If the osmotic pressures can be ignored and the solutes are freely movable Eq.
(1) can be reduced to the form
pk
1o = --gradp, (16)
.u
where p is the density of the fluid, .u is its coefficient of viscosity, and k is also
called a permeability constant. This is Darcy's law offluid flow through porous
media.
One may wish to emphasize the porosity of the porous medium in which
the fluid moves. Let us define the porosity (or the effective pore space per unit
volume of the medium) as the volume fraction <Do:
(17)
292 8 Basic Transport Equations

Then, we have
(18)
where Ie is a constant.

Recapitulation
According to thermodynamics of irreversible prcesses, phenomenological
laws should be based on a consideration of the local entropy production.
From such a consideration it is shown that the gradient of the chemical
potential of each chemical species in the tissue space is the driving force for
the movement of that species. Phenomenological laws relate the fluxes of the
chemical species to the gradients of the chemical potentials. The tissue space
may be divided into several compartments. Each compartment may be con-
stituted of a solution (a liquid) or a mixture (of gases, liquids, and solids). The
chemical potential of all the components in a compartment cannot vary
independently; they must obey the Gibbs-Duhem equation. Using the Gibbs-
Duhem equation, we show that Fick's law of diffusion, Starling's law for
membrane, and Darcy's law of flow through porous media are special cases
of a general relationship based on the consideration of local entropy produc-
tion and the second law of thermodynamics.
So far the matrix is considered mobile and treated as fluid. Immobile matrix
is treated in Sec. 8.9.

8.7 Diffusion from the Molecular Point of View

In the preceding section we saw diffusion driven by chemical potential. Now,

following Einstein (1908), let us look at it from the point of view of Brownian
motion.
Consider a molecule undergoing random collisions in the x-direction. Each
collision results in a step displacement d;. After N collisions, the displacement
Ax and its square are, respectively:
+ d2 + ... + dN ,
Ax = d 1 (1)
(Ax)2 = dr + di + ... + d~ + 2d 1 d2 + 2d 1 d 3 + ... + 2dN - 1 dN • (2)
If the probability for the positive d's is equal to that for the negative, then
when N is large Ax will tend to zero and (Ax)2 would approach the sum of
the squares of the d's. Thus, if d 2 is the mean value of the square of the
displacement step, then
(Ax)2 = Nd 2 • (3)
If T is the average time between steps, then in a time interval t, the number of
steps N is equal to tiT, and Eq. (3) becomes
8.7 Diffusion from the Molecular Point of View 293

(4)

When applied to a whole population of molecules at the origin initially, Eq.

(4) describes a fundamental characteristic of diffusion.
Einstein (1908) identified the coefficient d 2 /1: with twice the coefficient of
diffusion by considering the movement of molecules across a plane. Consider
a plane A with a box attached to each of its sides. Each box has a length d
and a cross-sectional area of 1 cm 2, d being the mean step length between
collisions, i.e., the square root of d 2 • There are C 1 d molecules in the box to
the left and C2 d molecules in the box to the right, (C being the concentration
of the solute, and equal to the product of the Avogadro's number and the
moles of the solute per unit volume). In a time interval 1:, all solute molecules
will leave these boxes, but on average half of the molecules will diffuse to the
left and half to the right. The number of molecules crossing the plane A from
left to right is tdC 1 , that from right to left is tdC 2 • The solute flux J 1
(moles/sec/cm 2) through the plane A is therefore,
d
J 1 = 21: (C 1 - C2 )· (5)

For a continuous distribution of concentration C as a function of x,

(6)

Hence

(7)

This is identical with Fick's law given in Eq. (8.6: 10) if we identify d 2/21: with
the coefficient of diffusion D:

(8)

Einstein (1908) went on to derive an expression relating the diffusion coeffi-

cient to the resistance that a solute molecule encounters when it moves in a
solvent. He obtained the expression
RT
D = --,,--- (9)
NA 6nl1a'
where D is the diffusion coefficient (cm2/s), R is the gas constant (1.99 cal. oK -1
mol- 1 ), T is temperature eK), NA is Avogadro's number (6.023 x 10 23 mole-
cules/mol), 11 is the coefficient of viscosity (dyn. s. cm- 2), a is the solute radius
(cm). Other authors extended Einstein's method to derive expressions for
diffusion through membranes with pores (Pappenheimer et aI., 1951) or in
294 8 Basic Transport Equations

fiber-matrix, Curry and Michel, 1980), see Curry (1984) for a comprehensive
review.

s.s Transport Across Cell Membranes

Cells can retain, gain, or expel ions against gradients of chemical potential,
or transport matter through their membranes at an augmented rate. The
kidney can concentrate urine. The nerve cells can transmit signals. The muscle
cells can rest or contract. All these are initiated through actions of their
membranes.
Obviously, in these membranes there must be some specific machineries,
specific structures, specific materials, each obeying physical and chemical laws,
and together leading to specific functions of each organ. It is equally obvious
that this is a long story, not possible to be compressed into a single section, a
single chapter, a single book. We shall consider the subject only so far as to
define the main topic of this chapter: the movement of fluid in the interstitial
space. This space is bounded by blood vessels on one side, and cells on the
other. Blood vessels and cell membranes are both semipermeable. The inter-
stitial space is bounded by semipermeable membranes.
In the preceding sections we have discussed the transport of fluid across
capillary blood vessels. For a substance (such as water) which moves accord-
ing to the gradient of chemical potential, cell membranes obey the same laws.
The chemical potential in the cell, however, depends on the activity of all
materials in the cell. Hence we must know the transport laws of all substances
in the cell. Certain substances can enter or leave cells by way of specific protein
carriers of the plasma membrane. If the process consists of "pumping" a
substance against a gradient of chemical potential, then it is called an active
transport. If the process tends to equilibrate the substances across the mem-
brane spontaneously but faster, then it is said to be a facilitated transport.

How to Discover the Transport Mechanisms in the Membrane

Since the cell membranes are so thin and the transport mechanisms cannot
be observed directly, one can approach the problem of determining the
mechanisms only by speculation and experimentation. Generally, one mea-
sures what can be measured and relies on physics and chemistry, especially
mechanics, thermodynamics, and statistical mechanics, to define all the restric-
tions quantitatively as far as possible, and to formulate hypotheses that are
not in conflict with these restrictions. If the predictions deduced from the
hypotheses agree with experimental results, then one gains confidence in the
hypotheses. If new experimental results contradict theoretical predictions,
then some hypotheses must be abandoned. There is no unique procedure to
the formulation of hypotheses. If two sets of hypotheses predict the same
results, scientists usually favor the simpler, weaker, more general, less restrictive,
and the least ad hoc.
8.8 Transport Across Cell Membranes 295

Proposed Mechanisms of Facilitated Diffusion

Figure 8.8: 1 shows several proposed models of channels in which ions can pass
through a membrane. In (a) a single membrane protein, an ionophore is shown
which has a hole down its center and is oriented in the direction of membrane
thickness. The channel is lined with charged or polar amino acid residues, and
filled with water. In (b) a set of rod-like proteins arranged in parallel with
space between them lined with polar group is assumed. In (c) and (d) gates are

hanne'l Ion pholc 'hJnllel

(al (bl

Channel IQnophorc

C hargC'd
..
( ') Cd) sclee,,' 11
O. YRC'1l flher
alOm\
FIGURE 8.8:1 Several ion channel models. (a) Tubular channel formed by two mole-
cules of the ionophore, the antibiotic gramicidin. (b) Rodlike ionophores proposed for
the anion channel in the red blood cell (Solomon et aI., 1983). The ionophore is a dimer.
Each monomer consists of five rodlike helices. Three helices (1, 2, and 4) from each
monomer surround the pore. (c) Potassium-selective channel in nerve (Latorre and
Miller, 1983). The "selectivity filter" does not pass Li or Na readily in their hydrated
forms, but does if Li and Na are unhydrated. (d) Sarcoplasmic reticulum maxi-
potassium channel (Miller, 1982). The tunnel is occupied by no more than one ion at
a time. The diffusion is relatively unrestricted at the mouth of the pore. Figure
reproduced by permission from Morton H. Friedman (1986).
296 8 Basic Transport Equations

--d- Low-affinity

0
i . L binding site
ii Interspace
L H
iii High-affinity
binding site
- -- _......... . . . _...
iv (I

~
v () <l
ii
vi

Membrane
---~
Membrane
Membrane
(a) (c)
(b)

FIGURE 8.8:2 Several carrier models. The states designated by Roman numerals repre-
sent successive steps in the cyclic process. (a) The classical ferry boat. (b) A gated pore
model (Patlak, 1957; Klingenberg, 1981). (c) A tetramer model of a sugar carrier in red
blood cells (Lieb and Stein, 1972). From Friedman (1986), by permission.

added and the shape of the channel is defined to match the energy profile.
Many authors assume that the gating process is statistical, and can be opened
or closed by ions, hormones, neurotransmitters, and changes in membrane
potential. Specific rules apply to each of these control agents.
Another class of assumed protein mediators in the cell membrane that can
facilitate diffusions is called carriers. Figure 8.8:2 shows several proposed
carrier models. In (a) the classical ferry boat is shown. The carrier diffuses
across the membrane. In (b) a gated pore model in which the carrier accom-
plishes the task by a conformational change is shown. The dimer undergoes a
conformational change (ii -+ iii) after binding the substrate in step ii, exposing
it to the side No.2 (trans. side) in step iii. Only one substrate molecule can fit
between the two binding sites, so only one molecule is transferred per cycle.
In (c) a tetrameric model, consisting of two monomers (H) whose binding sites
have high affinity for the substrate, and two monomers (L) of lower affinity is
shown. The monomers undergo concurrent conformational changes, such
that the carrier alternates between states i and ii, exchanging solute between
the ambient phases and the interspace. Solute in the interspace equilibrates
alternatively between the high and low affinity inward-facing sites.
Mathematical models of these channels and carriers are fairly similar; they
differ mainly on the physical meaning of the variables that enter the kinetic
8.8 Transport Across Cell Membranes 297

................................. 0;

AX

AXI ... "" AXil

AI +
1l
Xl >.
1f AI
A"

X" +A"
"" X

1-••••••••••••••••••••••••••••••
Side I Side II
1
Side I
1
Side II
(a) (b)

FIGURE 8.8:3 Schematic representations of the simplest carrier system. From Friedman
(1986), by permission.

equations. The kinetic analysis of the simplest carrier system can be presented
in a page or two. An examination of such an analysis will increase our
understanding of these phenomena. The starting point is a list of hypotheses:
HI) The carrier X, transports a single solute A. Fig. 8.8:3.
H2) Each molecule of X reversibly binds a single molecule of A. A can cross
the membrane only as AX.
H3) The binding/unbinding reaction A + X:;::::: AX has an equilibrium con-
stant K:

i = I, II, (1)

where c is concentration (to be replaced by activity if necessary), i is the

side number (I or II). The reaction is fast, and is in equilibrium on both
sides of membrane. K is the same at both interfaces.
H4) The phases on both sides of the membrane are at steady state or quasi-
steady state.
H5) The rate constant D', for the movement of carriers from one side of the
membrane to the other is the same for X and AX in either direction.
Thus the net flux of j (j = X or AX) from side I to side II is
(2)
Under these hypotheses, we see that the conservation of mass of the carrier X
requires that the sum of the parts be equal to the total X T,
(3)
At steady state, the combined transmembrane flux of X and AX must be zero
298 8 Basic Transport Equations

because the carrier X is confined to the membrane. Hence Jx + JAX = 0, i.e.

D'(cl- c~ + c1x - c'2x) = o. (4)
On the other hand, the flux ofthe substrate A is equl to the flux of AX because
A cannot cross the membrane in any other form:
JA = D'(c~x - c'2x). (5)
Solving Eqs. (2)-(5) for cl, c~, c~x' c'2x, and JA , using Eq. (1) as the boundary
conditions, we obtain the important result

J
A
= __
D'XT
2
[c+ c1
K
I
A _
K
c+ c'2 .
II
A ]
(6)

This equation shows that the flux of the solute A is linearly proportional to
c1- c'2 when the concentration CA is very small compared with K; and that
as CA increases relative to K, the flux tends asymptotically to a maximum and
the system is said to be saturated. If c'2 were zero, the maximum flux attainable
by the system is Vm = D'XT /2.
It is easy to imagine more complex models. See Friedman (1986) for
examples and for channels also.

Proposed Mechanisms of Active Transport

The difference between active transport and facilitated diffusion may be illu-
strated by the flux versus chemical potential difference relationship depicted
in Fig. 8.8:4. In a passive transport mechanism, the flow Ji> is a function of its
conjugate, the chemical potential difference I1lli (see Sec. 8.6), but J i always
vanishes when the conjugate chemical potential difference vanishes, i.e., J i = 0
when I1lli = O. The hallmark of active transport is that there exists a flow even
when the difference of chemical potentials I1lli is zero. The total flux may be
written as
(7)

in which Jf = 0 when I1lli = 0; whereas Jr is a result of the coupling of the ith

flux to a driving force that is not its conjugate driving force. The force driving
Jr is derived from a metabolic reaction. Jr is called the metabolic contribution
to Ji , or the rate of active transport of the ith species.
In concept, a metabolic reaction causes a conformational change in a
carrier protein, which then converts chemical energy into transport work. The
driving force for active transport is the affinity A, of the metabolic reaction:
A = I1llreaction = L
reactants
Villi - L
products
Villi' (8)

Here Vi are the stoichiometric coefficients of the reaction, and Ili are the
chemical potentials. Many active transport systems also convert chemical
energy into electrostatic potential energy by contributing to the electric poten-
8.8 Transport Across Cell Membranes 299

(b)

JaI

(a)
Uphill
transport

-----+-:..-----~~--------- a.l'i

FIGURE 8.8:4 Flow vs. its conjugate driving force in (a) a passive transport, and in (b),
an active transport. The terminology of an industrial pump is used to describe the
properties of an active transport mechanism by analogy. From Friedman (1986), by
permission.

tial difference across the membrane. The membrane potential can influence
cellular and transport events.
The major metabolic step relevant to active transport conceived so far is
the hydrolysis of adenosine triphosphate (ATP) to adenosine diphosphate (ADP)
and inorganic phosphate (P0 4 ):
(9)
The reverse step, the phosphorylation of ADP into ATP, is endothermic and
uses the energy of metabolism derived from food. Because the substrates are
oxidized during the energy extraction process, it is referred to as oxidative
phosphorylation.
ATP is the principal carrier of metabolic energy in the cell. When acted
upon by an adenosine triphosphatase (ATPase), it undergoes hydrolysis into
ADP, Eq. (9). The dephosphorylation reaction is accompained by the release
of the energy stored in the terminal phosphate bond, about 7.8 ± 0.5 kcal/mol.
The specific effect of the hydrolysis of ATP is determined by the specific
300 8 Basic Transport Equations

enzyme (ATPase) that causes the hydrolysis reaction. A specific enzyme

coupled with the flux of a specific transported solute results in a primary active
transport process. In animal cells, only three primary transport systems are
known:
1. The sodium-potassium exchange pump, which exchanges two K + for three
Na+ per ATP hydrolyzed.
2. The calcium pump for which the stoichiometry is uncertain; either one or
two Ca++ are transported across the cell membrane per ATP molecule
hydrolyzed.
3. The hydrogen-potassium exchange pump in the stomach, which pumps one
proton into the stomach per potassium removed.
The movement ofNa+ and K+ ions in the Na-K exchange pump may cause
other ions to move across the membrane, e.g. the use of sodium influx to
accumulate a sugar in the cell. This is called a secondary active transport.
Further, proteins may be transported across the cell membrane, the mechanism
of which is still unknown although vigorously studied.
From this brief outline one can see that active transport is initiated by
specific ATPase on the cell membrane. It depends on the availability of ATP.
The chemical reaction also depends on the membrane potential, and the
presence or absence of inhibitors and stimulators. The dynamic interaction of
these carriers on the cell membrane and the chemical and electric environment
of the cell creates the phenomena of nerve function, muscle contraction, urine
formation, and food absorption, which however, are, outside the scope of this
volume. Friedman (1986) is recommended for further reading.

8.9 Solid Immobile Matrix

So far we have treated the matrix component ofthe interstitial space as a fluid
(Sec. 8.5). In reality, especially in dense connective tissues, its quality as a solid
cannot be ignored. In a solid the state of stress must be described by a tensor.
The analog of pressure gradient in a fluid is the divergence of stress tensor in
a solid. In fluid we focus on flow. In solid we focus on strain rate. In fluid one
usually pays little attention to the displacement of particles, whereas in solid
one must. These are the points to be made in the mathematical analysis
presented below. These considerations are essential to answering questions
such as the bearing capacity, rigidity, and lubrication of the articular cartilage
or the intervertebral disc.
Biological tissues are multi-phasic. To present a simple illustration, how-
ever, let us lump solute and solvent together and call it the "fluid" phase, and
lump immobile matrix as the "solid" phase. The fluid phase has already been
discussed in Secs. 8.6-8.8. We view the system as a mixture; and each of the
two phases is present anywhere in the whole space (sometimes called the
"equipresence" hypothesis). Hence each point in the tissue is occupied simul-
8.9 Solid Immobile Matrix 301

taneously by both phases. This concept can be expressed in terms of the

density and volume fractions defined as follows (with the two phases identified
by superscripts s and f):
d(s), d(fLthe true density, i.e. mass per unit volume, of the solid and fluid.
p(S), p(fLthe phase density (mass/tissue vol) in the mixture.
</P), tP(f)-phase volume fraction, defined by
tP(S) = p(S)/d(S), tP(f) = p(f)/d(f). (1)

According to the hypothesis of equipresence, Ps and Pf are well defined, and,

since there is no void,
(2)
With respect to a set of rectangular cartesian coordinates, the displacement
Ui'velocity Vi' acceleration ai' strain eij' and strain rate eij , (i, j = 1,2,3) of the
solid and fluid can be defined in the usual way in view of the equipresence
hypothesis. The law of conservation of mass is expressed, in the absence of
chemical reaction, by the usual equation of continuity

(3)

These equations can be expressed differently if we can assume that both phases
are incompressible by themselves and the densities ds and df are constant.
Then, by Eqs. (1), Eqs. (3) become
OtP(S) o(tP(S)v}'»)_
Tt+ oXj -0, (4)

Adding, and using Eq. (2), we obtain

o(tP(S)v}')+ tP(f)vJf)) = 0
(5)
oXj
which can be written, by Eq. (2), as

a ( (s) + %) -_ 0,
::;- Vj (6)
uXj

Equation (6) says that the divergence of the velocity of the solid phase plus
the filtration flux q (the volume fraction of the velocity of the fluid phase
relative to the solid) vanishes.

Refinement of the Phenomenological Laws

In Sec. 8.6 we used the entropy production concept in thermodynamics to
identify the driving forces conjugate to the fluxes of the materials in various
phases of a mixture. The driving forces turn out to be the negative gradient
of the chemical potentials divided by the absolute temperature T. A chemical
302 8 Basic Transport Equations

species moves when it is subjected to the driving forces, and in motion it

encounters resistance due to friction and momentum transfer from particle to
particle. The actual mechanism of resistance is very complex. Instead of
delving into this complexity, we just assumed in Sec. 8.6 that the mechanism
is such that the fluxes are linearly related to the driving forces, leaving the
coefficients in this relation to be determined by experiments "phenomeno-
logically." The same idea will be used in the present section to deal with the
solid matrix, except that the forces of interaction will be expressed in terms of
stress tensors and the phenomenological laws will be replaced by the equations
of motion.

Simplifying Assumptions
In applying the theory to living tissues, the following assumptions are some-
times acceptable.
(1) All phases are incompressible, but the volume of the phases may vary
so that the tissue as a whole may be compressible (e.g. by squeezing water out
or by swelling).
(2) The velocities are so small that the convective acceleration is negligible.
(Rigid body motion is of no interest to the topic discussed here.)
(3) For the solid phase, a strain energy function Wexists, which is a function
of the strain, and symmetric with respect to the strain components eif) and eJi),
so that the stress is given by the equation (see Fung, 1981, p. 247):

(s) _
aij
oW
--;-. (7)
ueij

This stress is defined with respect to the tissue at a whole. W is part of the
chemical potential of the solid matrix (see Sec. 8.4).
(4) At every point in the fluid phase there is a hydrostatic pressure p, and
a stress deviation tensor which represents the shear stress in the fluid caused
by its moving through the solid matrix, a product of the viscosity of the fluid
and the strain rate due to flow in every "pore." The resultant of the viscous
shear forces in the pores, per unit volume of tissue, can be expressed as a body
force per unit volume, I/J, called the filtration resistance. We shall assume that
the filtration resistance is proportional to the filtration velocity, so that the
ith component of I/J is
i = 1,2,3, (8)
where % is the filtration flux defined in Eq. (6), and Rij is a matrix of constant
coefficients, called the resistivity tensor. That Rij are constants is suggested by
the smallness of the filtration velocity so that the flow is in the Stokes regime,
in which the resistance is linearly proportional to the velocity of flow (see
Chap. 5 in Fung, 1984). Rij depend on the size ofthe pores, hence are in general
nonlinear functions of the deformation.
8.9 Solid Immobile Matrix 303

Thus, we can write the stress in the fluid as

(1.(/)
lJ
= _p,,(f)/j ..
0/ I}
+ e.(f)
lJ'
(9)

where rjJ(f) is the volume fraction of the fluid phase, and elf> is the stress
deviation named above, which has the property that
o A(f) _
- ; - (1ij -
,I,
'I'i' (10)
uXj

(5) The structural fibers in the solid phase are stressed by strain according
to Eq. (7), subjected to the same pressure p of the fluid in contact, and the
shear stress in the fluid, eif), in the opposite direction. Hence the stress in the
solid phase may be written as

(1.(~) = oW _ prjJ(s)[) .. _ e.(f) (11)

u oeij 'J 'J'

(6) The chemical potential of the fluid consists of a part due to the pressure,
RTlnp, a part due to the concentration, RTlnca , of the chemical species
numbered rx (the concentration Ca to be replaced by "activity" in real solution),
and a part due to electric charges of the molecules. See Sec. 8.4. The chemical
potential of the solid has, in addition to those listed above, the strain energy
function W(e ll , el2 , ••. ) named in Eq. (7). However, since p and W have been
named explicitly in Eqs. (9) and (11), we shall use the notations p,(f), p,(s) for
chemical potentials other than RTlnp and W

Equations of Motion
The familiar Eulerian equation of motion (Sec. 1.7)
0(1··
(p . acceleration)i = ~ (12)
uXj

is derived for a homogeneous continuum. If the chemical composition is

nonhomogeneous, we must add the driving forces due to chemical non-
uniformity, namely, the gradients of the chemical potentials divided by the
absolute temperatures T, to the right-hand side (see Sec. 8.6). Thus, in view of
Eqs. (9), (11), and (12), the equations of motion of the fluid and solid phases
are, respectively,
p(f)alf) = _rjJ(f)~ + t/li + ~ op,(f) , (13)
OXi T oX i

p(S)aIS) = _rjJ(S)~ + ~(OW) _ t/li + ~ op,(S) . (14)

oX i oXj oe ij T OXi
Here ai is the ith component of the accleration vector. p,(f) and p,(s) are the
chemical potentials of the fluid and solid which depend on the concentrations
and electric charges of the chemical species.
304 8 Basic Transport Equations

Variations of the chemical potentials of a system must obey the Gibbs-

Duhem equation, Eq. (12) of Sec. 8.3. Under isothermal and constant stress
conditions, Gibbs-Duhem equation reads, for the biphasic system, i.e.
(15)
i.e.

(16)

This means that the effect of chemical potential on the motion of the liquid
phase is equal and opposite to that on the solid phase, i.e., it is an internal
interaction.
Some authors add a buoyance force
arjJ(f)
b.=p- (17)
, aX i
to the right-hand side of Eq. (13) and - bi to the right-hand side of Eq. (14).
This is based on molecular arguments by Muller (1968) for an intrinsic
tendency of material to become homogeneous in density.
Adding Eqs. (13) and (14), and using Eq. (16), we obtain

pa. = _~ + ~(aw) (18)

, ax; aXj aeij
which is the normal equation of motion of the whole tissue. Of the three
equations (13), (14), and (18), two are independent. Hence the solution of
any problem of biphasic flow must involve chemical potential and filtration
resistance.

Boundary Conditions
If a boundary of the mixture is in contact with an impermeable solid, then the
normal outflow will be zero, the normal traction will be specified, and the
normal velocity of the mixture will be equal to that of the solid. The tangential
component of the velocity and displacement are uncertain and need further
specification.
If the boundary of the mixture is in contact with another biphasic tissue,
then the normal stress, flow, velocity, and displacement of the two media must
be continuous at the boundary. If the boundary is a semipermeable membrane,
then IT can be revealed as an osmotic pressure. The chemical potential will
be continuous across the boundary, i.e., p - IT is continuous; hence p will have
a jump if IT does.
The multiphasic theory has been developed and applied to cartilage, skin,
and other connective tissues by Armstrong et al. (1984), Eisenberg and Grod-
zinsky (1985), Lai et al. (1981), Lanir (1987), Mow et al. (1980,1986), Myers et
Problems 305

al. (1984), Omens et al. (1984), Urban et al. (1979). For fundamental mathe-
matical development, one should consult Bowen (1976), Kenyon (1976), and
Truesdell (1962). There are controversies about acceptable approximate sim-
plifying assumptions for various tissues and organs: This has to be sorted out
carefully. Some authors focus attention on frictional reistance to flow. Other
authors speak of concentration effect or osmotic pressure effect. I believe,
however, that it is best to cast the theory in terms of chemical potential. Then
a unification with the classical theories of diffusion, filtration, osmosis, and
permeation in porous media can be achieved.

Problems
8.1 Name two organs whose functions significantly rely on active transport of matter
across membranes. Describe their anatomy, and how they work.
8.2 Generalize Fick's law, Eqs. (37), (39) of Sec. 8.5 to a solution that is not necessarily
dilute.
8.3 Generalize Fick's law to a solution with n species of solutes.
8.4 Under the assumption leading to Starling's law, Eqs. (14), (15) of Sec. 8.6, write
down an expression for the flux of a solute across the membrane.
8.5 Generalize Darcy's law, Eq. (16) of Sec. 8.6 to the movement of a nonhomogeneous
solution in a porous medium, including the effect of the chemical potentials of
various chemicals in the fluid or matrix.
Note: Further generalization to consider the effects of electric charges on the
solutes and matrix can be done. But consideration of the effect of elastic deforma-
tion of the "pores" in the media, leading to a nonlinear constitutive relation,
requires an analysis of a different nature which is presented by Lew and Fung
(1970).
8.6 G.S. Beavers and D.D. Joseph, in a paper entitled "Boundary Conditons at a
Naturally Permeable Wall" (J. Fluid Mech. 30, p. 197,1967), have proposed that
at a permeable wall the no-slip condition should be replaced by

fkoU =
v" or '
-(Xu

where (X is the "slip parameter", and k is the specific permeability of the porous
medium. Thus they postulated the existence of slip at the boundary. They con-
firmed it experimentally.
A qualitative estimation of the effect of slip is quite easy. Consider Poiseuille
flow in a tube. What is the effect of slip?
In blood circulation, could this effect be important?
8.7 Since the xylem fluid in trees is almost pure water, it will freeze in subzero
temperature. On thawing again, some trees can recover, other trees will drop their
frozen branches because cavitation forms in the xylem and water column cannot
be formed on thawing. Explain the difference. (Ref.: Hammel, H.T. (1967) Freezing
of xylem sap without cavitation. Plant Physiol. 42: 55-66.)
306 8 Basic Transport Equations

8.8 Describe the active and passive mass transport processes taking place in the
different segments of the loop of Henle in the kidney. (Cf. books such as P.C.
Johnson (1978), Peripheral Circulation, Wiley, New York.)
8.9 The leaves of a tree in the desert look wilted in the afternoon, but become turgid
by the next morning. From what you know about the negative pressure in the
xylem water, how can this phenomenon be explained? Can this explain how desert
plants get water from the atmosphere? What critical experiments should be done
to verify your hypothesis? (Refs.: Dixon, H.H. (1914) Transpiration and the Ascent
of Sap in Plants. 216 pp., London, MacMillan and Co. and Dainty, 1. (1963) Water
relations of plant cells. Adv. Bot. Res. 1: 279-326.)
8.10 The beautiful flower of water lilies opens up in the sun and closes up when the
sun sets. This is a mechanical event of which I know no literature. So let's
speculate. Propose a possible explanation, and possible experiments, and quanti-
tative analysis to verify your hypothesis.
8.11 Plants are well adapted to the environmental mechanics. Consider pine, bamboo,
and orchid in wind. How well streamlined they are! To be convinced of the
optimal design of the foliage, assume that the petiole (leaf stalk) and the leaf blade
were rigid, and compute the shear force and bending moment that must act in
the petiole at the stem in a wind of 30 km/hr. At what gale force would the
leaf-stalk break?

References

Andreoli, T.E, HolTman, J.F., and Fanestil, D.D. (eds.) (1980). Membrane Physiology. Plenum
Medical, New York, London (being Parts 1,2 and 3 of Physiology of Membrane Disorders
by the same editors, which contains Part 4, Specialized Cells, Tissues and Organs, and Part
5, Clinical Disorders of Membrane Disorders).
Armstrong, e.G., Lai, W.M., and Mow, V.e. (1984). An analysis of the unconfined compression
of articular cartilage. J. Biomech. Eng. 106: 165-173.
Bird, R.B., Stewart, W.E and Lightfoot, EN. (1960). Transport Phenomena. Wiley, New York.
Bowen, R.M. (1976) Theory of mixtures. In: Continuum Physics, (A.C. Eringen, ed.), Vol. III,
Academic Press, New York, pp. 1-127.
Curry, F.E. and C.e. Michel (1980). A fiber matrix model of capillary permeability. Microvasc.
Res. 20: 96-99.
Curry, F.E. (1984). Mechanics and thermodynamics of transcapillary exchange. In Handbook of
Physiology, Sec. 2, Cardiovascular System, Vol. IV, Part 1, (EM. Renkin and e.e. Michel,
eds.). Amer. Physiol. Soc. Bethesda, MD.
Darcy, H. (1956). Les Fontaines Publiques de la Ville de Dijon. Dalmont.
Eisenberg, S.R. and Grodzinsky, A.J. (1985). Swelling of articular cartilage and other connective
tissues: Electromechanochemical forces. J. Orthop. Res. 3: 148-159.
Einstein, A. (1908). The elementary theory of the Brownian motion. Z. Elektrochem. 14: 235-239.
Reprint, 1956, Investigations on the Theory of the Brownian Movement. Dover, New York.
Epstein, P. (1937). Textbook of Thermodynamics. Wiley, New York.
Friedman, M.H. (1986). Principles and Models of Biological Transport. Springer-Verlag, New
York.
Fung, y.e. (1965). Foundations of Solid Mechanics. Prentice-Hall, Englewood ClilTs, N.J.
Fung, y.e. (1977). A First Course in Continuum Mechanics. Prentice-Hall, Englewood ClilTs, N.J.
Fung, y.e. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
References 307

Hargens, A.R (ed.) (1981). Tissue Fluid Pressure and Composition. Williams and Wilkins, Baltimore,
p.3.
Intaglietta, M. and Johnson, P.c. (1978). Principles of capillary exchange. In: Peripheral Circula-
tion (P.c. Johnson, ed.). Wiley, New York.
Katchalsky, A. and Curran, P.F. (1965). Nonequilibrium Thermodynamics in Biophysics. Harvard
University Press, Cambridge, MA.
Kenyon, D.E. (1976). The theory of an incompressible solid fluid mixture. Archs. Ration. Mech.
Anal. 62: 131-147.
Klingenberg, M. (1981). Membrane protein oligomeric structure and transport function. Nature
290: 449-454.
Krupka, RM. and Deves, R (1983). Kinetics of inhibition of transport systems. Int. Rev. Cytol.
84: 303-352.
Lai, W.M., Mow, V.c., and Roth, V. (1981). Effects of nonlinear strain-dependent permeability
and rat of compression on the stress behavior of articular cartilage. J. Biomech. Eng. 103:
61-66.
Lanir, Y. (1987). Biorheology and flux in swelling tissue. 1. Bicomponent theory for small
deformation including concentration effect. Biorheology 24: 173-187. II. Analysis of uncon-
fined compressive response of transversely isotropic cartilage disc. Biorheology 24: 189-205.
Lew, H.S. and Fung, Y.c. (1970). Formulation of a statistical equation of motion of a viscous
fluid in an anisotropic non-rigid porous solid. Int. J. Solids Struct. 6: 1323-1340.
Lieb, W.R and Stein, W.D. (1969). Biological membranes behave as non-poroud polymeric sheets
with respect to the diffusion of non-electrolytes. Nature 224: 240-243.
Lieb, W.R. and Stein, W.D. (1972). Carrier and non-carrier models for sugar transport in the
human red blood cell. Biochim. Biophy. Acta 265: 187-207.
Lightfoot, E.N. (1974). Transport Phenomena and Living Systems. Wiley, New York.
Lotorre, R and Miller, C. (1983). Condition and selectivity in potassium channels. J. Membrane
Bioi. 71: 11-30.
Middleman, S. (1972). Transport Phenomena in the Cardiovascular Systems. Wiley, New York.
Miller, C, (1982). Feeling around inside a channel in the dark. In: Transport in Biomembranes:
Model Systems and Reconstitution. (R Antolini, G. Allessandra, and A. Gorio, eds.). Raven
Press, New York, p. 99.
Mow,V.C., Kuei, S.c., Lai, W.M., and Armstrong, C.G. (1980). Biphasic creep and stress relaxa-
tion of articular cartilage: Theory and experiments. J. Biomech. Eng. 102: 73-84.
Mow, V.c., Kwan, M.K., Lai, W.M., and Holmes, M.H. (1986). A finite deformation theory for
nonlinearly permeable soft hydrated biological tissues. In Frontiers in Biomechanics, (Schmid-
Schiinbein, G., Woo, S.L.Y., Zweifach, B.W., eds). Springer-Verlag, New York. 153-179.
Muller, I. (1968). A thermodynamic theory of mixtures of fluids. Arch. Rat. Mech. Anal. 28: 1-38.
Myers, E.R, Lai, W.M., and Mow, V.c. (1984). A continuum theory and an experiment for the
ion-induced swelling behavior of articular cartilage. J. Biomech. Eng. 106: 151-158.
Omens, C.W.J., van Campen, D.H., Grootenboer, H.J., and De Boer, L.J. (1984). Experimental
and theoretical compression studies on porcine skin. European Soc. of Biomechanics (ESB)
Conference, Davos, Switzerland.
Onsager, L. (1931). Reciprocal relations in irreversible processes. Phys. Rev. 37: 405-426; 38:
2265-2279.
Pappenheimer, J.R., Renkin, E.M., and Borrero, L.M. (1951). Filtration, diffusion and molecular
sieving through peripheral capillary membranes. A contribution to the pore theory of
capillary permeability. Amer. J. Physiol. 167: 13-46.
Patlak, C.S. (1957). Contributions to the theory of active transport. II. The gate type non-carrier
mechanism and generalizations concerning tracer flow, efficiency, and measurement of
energy expenditure. Bull. Math. Biophys. 19: 209-235.
Prigogine, 1. (1955). Introduction to the Thermodynamics of Irreversible Processes. Thomas,
Springfield, IL.
Renkin, E.M. (1977). Multiple pathways of capillary permeability. Circ. Res. 41: 735-743.
Solomon, A.K., Chasan, B., Dix, J.A., Lukocovic, M.F., Toon, M.R., and Verkman, A.S. (1983).
308 8 Basic Transport Equations

The aqueous pore in the red cell membrane: Band 3 as a channel for anions, cations,
nonelectrolytes, and water. Ann. N .Y. Acad. Sci. 414: 97-124.
Starling, E.H. (1896). On the absorption of fluid from the connective tissue spaces. J. Physiol.
(London) 19: 312-326.
Truesdell, C. (1962). Mechanical basis of diffusion. J. Chern. Phys. 37: 2336-2344.
Urban, JP.G., Maroudas, A., Bayliss, M.T., and Filion, J. (1979). Swelling pressures of proteo-
glycans at the concentration found in cartilageneous tissues. Biorheology 16: 447-464.
Zweifach, B.W. and Silberberg, A. (1979). The interstitial-lymphatic flow system. In: International
Review of Physiology, Cardiovascular Physiology III, Vol. 18, (A.c. Guyton and D.B. Young,
eds.). University Park Press, Baltimore.

!i4/.... ~
.at...""._
.y_A ..... """'M.~
«• .1 r;JH
J.'&"<I
It! 'i!r ~

The strain energy function Po W plays a central role in biomechanics. Caligraphy and
lithography by Aphrodite Sobin for the Symposium on Frontiers of Biomechanics,
1984.
CHAPTER 9

Mass Transport in Capillaries, Tissues,

Interstitial Space, Lymphatics, Indicator
Dilution Method, and Peristalsis

9.1 Introduction

In this chapter some applications of the basic equations derived in Chap. 8

are demonstrated. Flow across the walls of the capillary and lymph vessels is
discussed in Sec. 9.2. Methods for measuring the permeability of vessel walls
are presented in Sec. 9.3. A model of oxygen delivery and consumption in
tissues is given in Sec. 9.4. Fluid movement in interstitial space is discussed in
Sec. 9.5.
The "tissue pressure" in the very small interstitial space is very important
but very difficult to measure, and very controversial. This subject is discussed
in Secs. 9.6 and 9.7. Then in Sec. 9.8 the lymph flow is discussed.
In Secs. 9.9 and 9.10, the indicator dilution method for the measurement of
the extravascular space is discussed. In the last section, peristalsis as a mode
of mass transport is analyzed.

9.2 Fluid Movement Across Capillary Blood Vessel Wall

A simplified description of the flow through capillary walls is Starling's law

(Sec. 8.6):
Jo = K[pc - Pt - O'(nc - nt)],
(1)
J1 = <Xl (Pc - Pt) + <X2(nc - nt),
in which Jo , J1 represent the fluxes of water and solutes (rates of volume flow
per unit area from the capillary side to ~he tissue side), respectively, P is static
pressure, n is osmotic pressure, K, 0', <Xl' <X2 are material constants, K is called
hydraulic conductivity, 0' is the reflection coefficient, the subscript c refers to

309
310 9 Mass Transport

the capillary, whereas t refers to the tissue. (J is related to the degree of

permeability of the capillary wall to the solutes: (J = 1 if the wall is imperme-
able to the solutes; (J < 1 if it is permeable. The smaller the (J, the leakier the
wall is to the solutes.
The coefficients K, (J, !Xl' !X2 vary from organ to organ, and are affected by
stress and strain, and the molecular sizes of the solutes.
To understand the nature of the coefficients K, (J, !Xl' and !X 2 , a number of
mechanical models of the capillary endothelium have been proposed. Renkin
and Curry (1982) have examined the multiple pathways for transcapillary
exchange. The pathways that have been documented in the literature are:
(1) Transport through the endothelial cells by crossing two plasma mem-
branes and cell cytoplasm.
(2) Transport within the cell membranes by lateral diffusion in the lipid phase
through intercellular junctions or vesicular channels.
(3) Transport through junctions between endothelial cells in the aqueous
extracellular phase either a) via small pores or a fibrous meshwork imper-
meable to plasma proteins, or b) via large pores or a less-dense fibrous
meshwork permeable to plasma proteins.
(4) Transport through endothelial cell fenestrae which are either specialized
openings across the cells, or are areas at which two cell membranes are
fused and modified, bypassing the aqueous cytoplasmic phase.
(5) Transport by vesicles in the endothelial cells that either a) move back
and forth between cell surfaces, or b) communicate with one another
transiently to exchange their contents or to establish temporary open
channels.
(6) Transport through the basal laminar which consists of a layer of fine
collagenous fibers.
(7) Transport through the cell surface coat which consists of a layer of gly-
coprotein strands that extends into the intercellular junctions and lines
the vesicles.
Water, lipophilic solutes, hydrophilic solutes, and macromolecules may use
different pathways. See Curry (1984) and Table 9.2:1.
In the following, we shall consider a few highly idealized examples to

TABLE 9.2: 1 Partition of transcapilIary

fluxes between different pathways*
Species Pathways
Water 1,3,4
LipophilIic solutes 1,2,3,4
Hydrophilic solutes 3,4
Macromolecules 3,4,5
* According to Curry (1984).
9.2 Fluid Movement Across Capillary Blood Vessel Wall 311

occ lusion ro d
...!:.... •
2R Zo •
' ..
o .. . .
Z -- .... - -- - ...............
lEo

FIGURE 9.2:1 A model of a capillary blood vessel.

illustrate the nature of the mathematical problem. The capillary geometry is

shown in Fig. 9.2:1. It has a length L and a constant radius R. The blood flow
rate in the capillary is Q(x); the filtration rate across the wall is J(x), x being
the coordinate along the length of the capillary, x = 0 at the arteriolar end,
x = L at the venular end. The subscripts c and t stand for capillary and tissue,
respectively, whereas a and v stand for the arteriolar and venular ends of the
capillary, respectively.
The amount of fluid that is filtered by the capillary per unit time is the
filtration rate,

filtration rate = 2n LL R(x)J(x)dx. (2)

If a dimensionless variable e= x/L is introduced, the radius is constant, and

(J = 1, then

filtration rate = A L K(e){ [pcm - pt(e)] - [nc(e) - nt(e)]} de, (3)

where A = 2nRL is the capillary surface area.

Example 1. Assume that K, nc, Pt, and n t are constant, whereas
b = const. (4)
Then
filtration rate = KA[Pa(l - !) - Pt - nc + ntl (5)
Example 2. (Apelblat et aI., 1974; An and Salathe, 1976). Assume that K, nco
nt, and Pt are constant, whereas Pc obeys the momentum equation

(6)

dQ
de = -2nRK[pc(e) - Pt - nc + ntJ, (7)

Pc(O) = Pa' Pc(l) = PV' (8)

312 9 Mass Transport

Then the filtration rate is

KA [ r: r: - 2 - - Pt + 1I:t) - 1I:c] ,
COSh4Je - 1 (Pa + Pv (9)
2yBsinh4yB
where B is a dimensionless parameter:

B=~(~r, (10)

where J1. is blood viscosity. B is usually <0.01. The blood viscosity J1. is a
function of capillary radius, tubular hematocrit, volume rate of flow, elastic
properties of the red cells, and plasma protein concentration. (See Fung, 1981,
Chap. 5.)
Example 3. (Deen et ai., 1972). Assume that K, Pt, and 1I:t are constant, and
Pc(~) is given by Eq. (4). The plasma oncotic pressure 1I:c is required to satisfy
the equation of conservation of mass of protein in the capillary. Then the
filtration rate is

ca , Cv are the plasma concentrations at the arterial and venous ends of the
capillary, respectively. Qa is the volume rate of flow into the capillary, HF is
the (feed) hematocrit of blood entering the capillary.
Example 4. (Papenfuss and Gross, 1978). Assume that K, Pt, 1I:t are constant,
while Pc, 1I:c satisfy the equations of motion and continuity. Consideration of
mass balance leads to

where Peff is defined as the effective pressure:

Peff(~) = Pc(~) - Pt + 1I:t• (13)

The gradient of Peff is given by the Poiseuille equation

dPeff = -8~Q(~) (14)

d~ 1I:R4 •

The "oncotic" pressure 1I:c of the blood plasma in the capillary is a function
of the plasma protein concentration c, and may be written as
(15)
in which the coefficients oc, p, CJ have been reported by Landis and Pappen-
heimer (1963). Conservation of plasma protein in the capillary blood vessel
leads to the equation
9.3 Experimental Determination of the Permeability Characteristics 313

(1 - HF)Qa
c(~) = Ca Q(~) _ HFQa . (16)

Here c(~) is the plasma protein concentration at ~, HF is the hematocrit as

in Eq. (11), Q is the flow, and the subscript a denotes the arteriolar end.
The boundary conditions for Eqs. (12), (13), and (14) are ~ = 0, Pc = Pa'
Q = Qa, C = Ca' These equations can be solved numerically, or by asymptotic
expansion in terms of the small parameter e defined by Eq. (10), after the
equations are properly nondimensionalized. Papenfuss and Gross (1978) have
applied this theory to the rat glomerulus and intestinal muscle, and showed
that it is important to consider the nonlinear terms in Eq. (15).

Discussion
None of the above examples deal with tissue pressure PI and tissue osmotic
pressure 11:1 ; yet these two variables are very important for the health of tissue.
For example, changes in PI are associated with edema, swelling, pain, and
injury. We shall address this topic in Sees. 9.5-9.7.

9.3 Experimental Determination of the Permeability

Characteristics of Capillary Blood Vessel Wall
Landis (1927) first showed how to get hold of a single capillary blood vessel
and measure the permeability of the vessel wall. In the 1960's and 70's,
Zweifach, Michel, and others made extensive investigations with Landis's
method. Other investigators, however, experimented with whole organs and
then used simplified models of the organs to interpret the results, e.g., the
osmotic transient method of Pappenheimer et al. (1948, 1951), and the indicator
dilution method. These methods are outlined below.

Landis Micro-occlusion Technique

If the flow in a capillary is suddenly stopped with a fine needle (Fig. 9.3:1),
then the pressure on the left-hand side suddenly becomes that ofthe arteriolar
pressure, while the pressure on the right-hand side becomes venular. This
upsets the equilibrium condition across the vessel wall, and fluid will move
out of or into the capillary. To measure the filtration rate, Landis used the
red cells in the blood vessel as markers. Assuming that the vessel radius R
does not change, and that the cells sufficiently plug up the vessel so that there
is no leakage of plasma between the red cells and the capillary wall, then any
loss of fluid in the vessel will be seen by a shortening of the distance between
neighboring red cells. The distance L between a red cell and the occluding
needle is measured as a function of time. In a time interval 11t, the distance
314 9 Mass Transport

micropipette occlusion rod

o 00

FIGURE 9.3:1 Landis's (1927) microocclusion experiment.

L
~~ slope = ( ~~ )0
.. ...
.. .....
~.
(0) \
\
\\
'\
t

FIGURE 9.3:2 Data reduction method to determine transport coefficients of capillary

wall by Landis's method. From Papenfuss and Gross (1979), by permission.

changes by tlL, the volume of the capillary segment is changed by nR2 tlL,
while the amount of fluid that flows out of the vessel wall is JwL 2nR tlt, Jw
being the flux across the wall. Equating the outflow with the change ofvolume
and solving for Jw , one obtains

Jw = - 2L & = -K[(pc - Pr)

R (tlL) + l1(nc - n r)]. (1)

When L is plotted against t as in Fig. 9.3:2(a), it is seen that the slope dL/dt
varies with time. Landis extrapolated the slope and L to t = 0 and computed
9.3 Experimental Determination of the Permeability Characteristics 315

Jw at time O. He measured the capillary pressure Pc in the neighboring arteriole

(Fig. 9.3:1). When Jw is plotted against Pc at t = 0, the slope of the resulting
regression line yields the hydraulic conductivity K (Fig. 9.3:2(b)).
This calculation is valid only if Jw , Pt, 1tc, 1tt, R, (1, and K are constant along
the capillary. Zweifach and Intaglietta (1968), however, have shown that this
last assumption is quite wrong: in fact, K varies along a capillary by a factor
of almost 3. The occluding needle certainly disturbs the radius and the tissue.
The hypothesis that the red cells can serve as dividers of plasma segments
without leaking is also uncertain.

Lee-Smaje-Zweifach Method
Zweifach and Intaglietta (1968) modified Landis's method by observing two
red cells and measuring the distance between them z(t), as a function of time
following occlusion (Fig. 9.2:1). This avoids the assumption that the occluding
needle does not deform the blood vessel and surrounding tissue. The theory
is given by Lee et al. (1971). Assuming (1 = 1, they write:
R dz
Jw(t) = - 2z(t) dt = - K[Peff - (t)],
1tc (2)

where Peff = Pc - Pt + 1tt, and 1tc is the colloidal osmotic pressure. The varia-
tion of 1tc with the loss of water as z changes is analyzed and incorporated into
the calculation (cf. Eq. (16) of Sec. 9.2). Equation (2) is solved as a differential
equation for z(t) to obtain a solution in the form of t = f(z). Then a least-
squares method is used to fit the data and obtain K.

Pappenheimer and Soto-Rivera Isogravimetric Method

Pappenheimer and Soto-Rivera (1948) isolated an organ and weighed it
continuously while it was well supported to preserve its natural shape, envi-
ronment, and state of stress. They controlled the conditions so that the
weight did not change with time (isogravimetric), at which the averge osmotic
pressure and static pressure across the capillary blood vessels balanced each
other according to Starling's law. Now, if either the static pressure or the
osmotic pressure of blood in the capillaries were changed, filtration or absorp-
tion of fluid into or from the tissue would occur, and the weight of the organ
would change. By plotting the rate of increase of weight against the changes
in static or osmotic pressure, they identified the slope of the curve as the
filtration coefficient from which the permeability constants can be calculated.
Arterial pressure, venous pressure, osmotic pressure of the perfusing fluid,
blood flow, and temperature can be varied at will, and their influence onthe
filtration-absorption equilibrium can be measured separately. But interpreta-
tion of whole organ experiments is always somewhat difficult because it is
necessary to use some kind of model.
316 9 Mass Transport

The Indicator Dilution Method

The indicator dilution method (Sec. 9.9) was first used by G. N. Stewart to
measure the circulation time in organs in 1893. The theory was clarified by
G. I. Taylor (see Sec. 7.3). It is now widely used as a clinical tool. In applying
it to measure permeability, however, considerable difficulty exists. Issues and
controversies are well documented in Crone and Lassen (1970), Aschheim
(1977), Pappenheimer (1948, 1951), Renkin (1959), and Johnson and Wilson
(1966).

9.4 The Krogh Cylinder as a Model of Oxygen Diffusion from

Capillary Blood Vessel to Tissue

A mathematical idealization that has served as a model of capillary-tissue

gas (or solutes) exchange was proposed by Krogh (1919). In this model, the
capillary blood vessel is represented by a straight circular cylindrical tube
of radius rc; the tissue is represented by a stationary concentric cylindrical
tube of inner radius reo outer radius rt • A large number of identical tubes are
closely packed to represent the tissue of an organ; therefore, the boundary
condition at the outer radius r t is no-flux.
Since diffusion is the main mechanism, the field equations are the same as
those of Sec. 7.4. Let c be the concentration of the chemical species of interest
(moles/volume). The flux vector (the rate of movement of the species across a
cross section of unit area per unit time) consists of two parts: diffusive and
convective. The former obeys Fick's law (Sec. 8.5). The latter is proportional
to the product of velocity and concentration. Hence the flux
oc
Jy = -Dy oy + VyC,
where D", Dy, Dz are the coefficients of diffusion in the x, y, z directions,
respectively, and v (with components v,,, vy, vz ) is the velocity vector of the
medium. (x, y, z) is a fixed frame ofreference.
The law of conservation of mass can be expressed by considering the mass
transfer into a small control volume dx dx dz, as shown in Fig. 9.4:1. The
inflow into the surface dy dz at the left is J" dy dz per unit time. The outflow
through a parallel surface at a distance dx to the right is [J" + (oJ,,/ox)] dy dz.
Therefore, the net inflow rate through these two surfaces is -(oJ,,/ox) dx dy dz.
Considering all six surfaces, we obtain the net inflow rate:

oJ"+oJy
- (- - +OJ
-z ) dx dY dz. (2)
ox oy oz
In the meantime, the mass of the species in the volume is changing at a rate
OC dx dy dz. Further, there may be chemical reaction in the volume dx dy dz
ot
9.4 The Krogh Cylinder as a Model of Oxygen Diffusion 317

I
Jx I
-- -~........ I
---'7--"
I dz
I
I
r---- --

FIGURE 9.4:1 A control element.

which consumes the species at a rate of g moles per unit volume per unit time.
Adding all three rates together, we obtain

oe = _ (OJx + oJy + OJz ) _ g. (3)

at ox oy OZ
On substituting Eqs. (1) into Eq. (3), we obtain

!: = :x ( Dx ;:) + :y (Dy ;;) + :z (

Dz : : ) - g

_ e (ov x + oV y + OVz ) _ v oe _ v oe _ v oe (4)

ox oy oz x ox Yay z oz

which is the basic equation for diffusion. If the fluid is incompressible, then
oVx oVy oVz _ 0
(5)
ox + oy + OZ - .
If, in addition, the coefficient of diffusion D is isotropic and is a constant, then
Eq. (5) is reduced to
Dc oe oe oe oe
Dt == at + Vx ox + Vy oy + V OZ Z

02e 02e 02e)

= D ( ox2 + oy2 + OZ2 - g. (6)

In the tissue space, let the coefficient of diffusion be Dt , the velocity of

motion of the tissue be zero, and the distribution of oxygen be axisymmetric.
Then, on using cylindrical-polar coordinates, Eq. (6) becomes, in the tissue
region,
318 9 Mass Transport

(7)

It is often observed that oxygen consumption in tissue follows Michaelis-

Menten kinetics (White et aI., 1959), which may be expressed as
Ac
g(c) = (B + c) (8)

with two constants A, B.

In the capillary region, let Db be the overall coefficient of diffusion of
oxygen in the blood, which varies with the hematocrit and capillary size. Let
d(c) represent the rate of generation of oxygen within the blood due to
the dissociation of oxyhemoglobin contained in the red cells. Assume incom-
pressibility of blood and axisymmetry in oxygen distribution. Then Eq. (6)
becomes
oC (02C 1 OC 02C)
ot + v'Vc = Db or2 + -;: or + OZ2 + d(c), (9)

The boundary conditions are (a) symmetry about the axis of the cylinder, (b)
no flux at r = rt , and (c) continuity of flux at the interface r = rc. Thus,
r = 0: oc/or = 0, (9a)

r = rc: Db -
(oc) = (oc)
Dt - , (9b)
or blood or tissue
r = rt : oc/or = O. (9c)
These equations have been solved for a variety of conditions. Some com-
ments follow:
(a) Krogh's Steady-State Model (1919). Assuming that 1) g(c) in Eq. (6) is
a constant go; 2) c is equal to a constant Co in the blood vessel; 3) the condition
is steady-state; and 4) the axial diffusion of oxygen is negligible, then Eq. (6)
is reduced to
d2c 1 dC)
0= Dt ( dr2 + -;: dr - go (10)

whereas Eq. (8) is reduced to c = co.

The solution, subjected to the boundary conditions (9c) and c = Co at r = rc ,
is

c = Co = D: --4-
g (r2 _ r2 r2
c
r)
- ~ log;:-; . (11)

(b) Bloch (1943) and Blum (1960) relaxed the boundary condition c = Co
at r = rc in Krogh's approximation, and considered the axial variation of
oxygen concentration in the capillary blood vessel. Several models of the
oxygen generation function and oxygen consumption are considered. Reneau
et al. (1967, 1969) presented numerical solutions of these equations in both
steady-state and nonstationary cases, with particular reference to the brain.
9.4 The Krogh Cylinder as a Model of Oxygen Diffusion 319

They concluded that axial diffusion makes a significant contribution only near
the entrance to the capillary.
(c) Bassingthwaighte et al. (1970) made an analysis of a Krogh tube which
has a circular capillary blood vessel in the center and a tissue cylinder with
hexagonal cross section on the outside. Hexagonal shape is selected because
it is space-filling. The dimensions and parameters are selected for the model
to represent a myocardium, and the boundary conditions used are pertinent
to the problem of fractional extraction of a diffusible indicator during trans-
capillary passage. It is concluded that the intratissue radial concentration
gradients are probably negligible for small solutes, and that the longitudinal
concentration gradients in tissue and capillary are important even for non-
metabolized indicators.
(d) Niimi and Sugihara (1984), using a finite difference method, have ana-
lyzed a Krogh tube with a variable radius, rt(z). They considered counter
current flow in neighboring tubes. They treated blood as a two-phase fluid,
with red blood cell and plasma as two separate phases.
(e) Other Diffusing Substance. Oxygen is lipid-soluble. It can freely pene-
trate the endothelial cells and other elements of the vascular wall. Hence the
boundary condition specified in Eq. (9) is appropriate. For lipid-insoluble
substances such as water or albumin, the boundary condition (Eq. (9)) is
inappropriate. Starling's law is considered a better approximation for trans-
capillary movement of lipid-insoluble substances.
(f) Larger Blood Vessels. Oxygen diffusion takes place not only in the
capillaries, but also in arteries and veins. Figure 9.4:2 shows a profile of

70 92% Hb0 2

60

50

¥E 40
.§.
S
0.. 30

20

10

0 arteries tissue

FIGURE 9.4:2 Perivascular oxygen tension and estimated oxyhemoglobin saturation

in the arterial network of the hamster cheek pouch. Data from Duling and Berne (1970).
Figure by Intaglietta and Johnson (1978). Reproduced by permission.
320 9 Mass Transport

the perivascular partial pressure of oxygen (which is very nearly equal to

intravascular pOz) plotted against vessel type. The figure refers to a hamster
cheek pouch. The arterialized blood from the lung contains approximately
20 ml Oz per 100 ml blood at a partial pressure of 100 mm Hg. By the time
the blood reaches the beginning of capillaries, the pOz is only 25 mm Hg.

9.5 Fluid Movement in the Interstitial Space

A schematic drawing of the interstitial space is shown in Fig. 9.5:1. Blood

enters capillary blood vessels from the left and exits to the right. Blood
and lymph vessels are embedded in a tissue, which consists of cells and an
interstitial matrix. The walls of the capillaries and lymphatics are permeable
to water, solutes, and some proteins. The blood inflow is controlled by
arterioles and precapillary sphinctors (P.S.). The exchange of water and solutes
between the capillaries, lymphatics, and the tissue is determined by the static
pressures in the capillaries (PJ, lymphatics (PL)' and the tissue (Pt), and the
osmotic pressures in the capillaries (n c ), lymphatics (nd, and the tissue (n t ).
The flux of the fluid from capillaries into the tissue is denoted by Ja at the
arteriolar end, Jv at the venular end. JL is the flux across the lymph vessel. In
the interstitial space, fluid moves slowly as in a porous medium. The flux is
measured by the rate of movement ofthe interstitial fluid across a cross section
of the tissue per unit area of the tissue.
Equations governing the fluid movement in interstitial space have been
derived in Sec. 8.5. The simplest version is Darcy's law, which ignores the effect
of osmotic pressure on water movement.
Darcy's law (Eq. (18) of Sec. 8.5) states
J o = -KellVp, (1)
where J o represents the volume flow rate of the interstitial fluid, K is a constant,
ell is the "porosity" of the tissue, i.e., the fraction of a given portion of tissue
occupied by mobile interstitial fluid, V is the gradient symbol, p is the pressure,
and ell and K are functions of p. By the method detailed in Sec. 9.4, the rate of
change of fluid volume in a unit volume of tissue can be written as - V· J o ,
which must be equal to the rate of change of the porosity, aell/at. Hence, the
principle of conservation of mass is expressed by the equation:
a
at ell(p) - V· ell(p)K(p)Vp = o. (2)

For steady flow this is reduced to V· ellKVp = 0; and for constant ell and K, to
the Laplace equation, VZp = O.
The flux of solutes and proteins is given by J 1 of Eq. (22) of Sec. 8.5. To
obtain a simplified version which is consistent with Darcy's law for water flux,
we assume that the solution is dilute, and neglect the term LZ1 grad Ilo in
the second equation of (8.5:22), because the conjugate term L12 grad III is
9.5 Fluid Movement in the Interstitial Space 321

tI.,
ARTERIAL
FLO~
VENOIS RETURN
~~_-t

c~

20,um

FIGURE 9.5:1 Top: Concept of balance of tissue fluid. Transcapillary pressures (closed
arrows) determine the transcapillary flows (open arrows). Reproduced with permission
from A.R. Hargans (Ed.), Tissue Fluid Pressure and Composition, p. 3, copyright ©
1981, the Williams & Wilkins Co., Baltimore. Bottom: A histological photograph
showing an example that in the rat gracilius muscle the interstitial space is very small.
Fascia intact. Capillary blood pressure was 25 cm H 2 0. M = muscle cell, C = capillary
blood vessel, R = red blood cell in capillary, I = interstitium. Courtesy of Jye Lee,
Ph.D., from his dissertation: "Morphometry and Mechanical Properties of Skeletal
Muscle Capillaries." University of California, San Diego, 1990.
322 9 Mass Transport

neglected in Darcy's law. Neglect also the term L 23 grad 112 because the matrix
is immobilized. Further, using Eq. (8.5:29), we obtain
J 1 = -L22(cD 1Vp + kTVc1), (3)
where cD 1 is the volume fraction of the solute, k is the Boltzmann constant. T
is the absolute temperature, and L22 is a constant. The first term on the
right-hand side of Eq. (3) represents transport by convection, the second term
represents diffusion.
The conservation of mass of the solute requires that
OC1
at + V'J 1 = 0, (4)

because V . J 1 represents the rate of loss of solute in a unit volume, and OC 1 jot
the rate of increase in the same volume.
If the effect of osmotic pressure of the solutes or water movement is not
neglected, then Darcy's law is not valid and we must use the full Eq. (22) of
Sec. 8.5.
These equations are nonlinear and difficult to solve. It is clear that the
interstitial pressure is a dynamic quantity, it changes both in space and time.
For example, the vasomotion of the arterioles and venules will affect p
significantly.
At a steady-state, Eq. (2) is reduced to
V· cD(p)K(p)Vp = O. (5)
To solve this equation, Salathe and An (1976) introduced a new variable, "',
which is a function of p defined by
d",
dp = cD(p)K(p), (6)

Then
d",
V", = dp Vp,
and Eq. (5) is reduced to the Laplace equation
V2 ", = O. (7)
A solution that satisfies the condition", -+ 0 as r2 + Z2 -+ 00 is
",(r,z) = r L
f«()d( , (8)
JoJ«(-z)2+r2

where f(O is an arbitrary function defined over the interval 0 :::::; ( :::::; L. This
solution is useful if the interstitial space is infinitely large compared with the
capillaries. It states that the interstitial pressure can be represented as a
distribution of sources and sinks of the function '" distributed along the
9.6 Measurement of Interstitial Pressure 323

capillary axis, and reduced the determination of p(r, z) to the determination

of a function of only one variable J(z). Salathe and An (1976) show further
that Eq. (4) admits a solution of the form C 1 = F(ljJ), which states that the
interstitial protein concentration is a function only of the interstitial fluid
pressure, see Salathe (1980).

Nonuniformity and Other Idiosyncrasies of Interstitial Space

There is much evidence for non uniformity of porosity in tissue space. Diffusible
dyes (e.g., fluorescent dextran) injected into the blood vessel diffuses out into
the tissue, and great nonuniformity of the concentration of the dye is seen in
the tissue. Often there seems to be invisible channels carrying the dyes off.
This means that either K or <I> or both are functions of location in the
interstitial space. In practical application of the mathematical analysis, the
non uniformity must be recognized, but not much is known at this time.
In skeletal muscle, the extravascular space is filled mostly by muscle cells.
The cell volume of the muscle cells are closely regulated by the active pump
ofNa +, K + ions by the cell membrane. As a result, although the cell membrane
is permeable to water, the cell volume changes little, and the movement of
water in the extravascular space takes place only in the extracellular space.
The extracellular-extravascular space in the skeletal muscle is very small in
vivo; perhaps no more than 2 or 3 percent of the muscle volume. Thus the
domain in which the field equations (2), (3), (4) or (5) apply is long and narrow.
Again, anatomical and rheological data are missing, and both theory and
experiment remain to be developed.
In the lung, the interstitial space in the alveolar wall is normally much
smaller than the vascular space. In normal conditions without interstitial
edema, the thickness of the interstitium in the alveolar wall of man, dog, and
cat is no more than 1 /lm, whereas the thickness of the capillary blood vessels
is 5-10 /lm in zone 3 condition (see Chap. 6). The tissue pressure in this space
depends on the structure of the interstitium, especially on the volume of the
macromolecules such as the hyaluronic acid. In this case the geometrical
configuration is sufficiently simple that a detailed analysis can be done. A
theoretical analysis is presented in Sec. 9.10.
The interstitial spaces in tissues of other organs such as liver, kidney,
arterial wall, etc., are also complex in their own way. Each organ needs a
special treatment. This rich field is waiting to be cultivated.

9.6 Measurement of Interstitial Pressure

The time-averaged interstitial pressure PI has been measured by several au-

thors. Guyton and his associates (1963, 1981) imbedded perforated, hollow,
rigid capsules (made of plastic or metal, with diameters ranging from 3-
15 mm) in tissues. They let the interstitial fluid collect itself in the capsules in
324 9 Mass Transport

the course of days, then inserted needles into the capsules to measure the fluid
pressure. If the pressure of the capsule fluid can be assumed to have exactly
the same pressure as thatof the interstitial fluid in the tissue, then the capsule
pressure yields the latter.
Wiederhielm (1969, 1972, 1979, 1981) employed micropipettes to measure
the tissue pressure. These glass micropipettes with dip diameters less than
1 11m are produced by locally heating glass pipettes until they elongate
and break under tension. By adjusting the tension and heating rate, the tip
diameter can be controlled. Each pipette is then filled with a NaCI solution
whose conductivity is different from that of the tissue fluid. The impedance of
the pipette depends on the location ofthe interface between the saline and the
tissue fluid in the tip section. If the saline pressure in lower than that of the
tissue fluid, the interface will be pushed into the tube and the impedance will
change. By a servo control with a pump, the back pressure in the pipette can
be increased until the impedance reaches an extremum. At this condition the
interface is located at the tip of the pipette. The back pressure in the pipette
is then taken as the pressure of the tissue fluid.
Scholander et al. (1968) inserted cotton wicks with connective tissues of
animals. The wicks are enclosed in polyethylene tubes which are perfused with
saline. The wick and tube are introduced into the tissue with a trocar. When
the trocar is withdrawn, t-1 cm of the wick is freely embedded in the tissue,
the rest is enclosed in the polyethylene tube, which is connected to a horizontal
pipette, then to a saline reservoir. A small air bubble is left in the horizontal
pipette. The fluid in the wick is then allowed to come into equilibrium with
the tissue fluid by adjusting the height of the reservoir in such a way that
the bubble in the horizontal pipette remains stationary. In a few minutes
equilibrium is established and the pressure in the pipette is measured and
taken to be the tissue pressure.
Scholander's wick is considered to be equivalent to a large number of
micropipettes by the capillary action. The principle is similar to Wiederhielm's
except that the wick has soft walls and has many, many points of contact with
the interstitial space. Wiederhielm's micropipette, though small, is still like
"telephone poles" to the interstitial space.
The results from the three approaches do not always agree. For animals in
the atmosphere Guyton and his associates have found tissue pressure in
various organs to be negative (zero being atmospheric), in the order of -6
mm Hg, except in the kidney and the brain. In the kidney the tissue pressure
is lower than the pressure under the tight renal capsule. In the brain, it is
lower than the pressure in the cerebrospinal fluid that surrounds the brain.
Scholander et al. (1968) found negative pressure in subcutaneous tissues. But
most investigators using the Wiederhielm probe have found positive mean
pressure, in the order of 1- 3 mm Hg in subcutaneous tissue. Everybody agrees,
however, that edema develops when the tissue pressure exceeds a certain limit.
These differences have not been completely resolved. Injury, healing, large
9.7 Pressure in an Incompressible Material 325

disturbance of the normal state by the measuring instrument, and the changed
boundary conditions are all parts of the story. The disturbed boundary
conditions are especially important, as will be seen in the next section.

9.7 Pressure in an Incompressible Material

The tissue pressure measurements (Sec. 9.6) led to results that are perplexing.
To gain a deeper understanding we shall consider the pressure in an incom-
pressible fluid in general; namely, the biological tissue and body fluids.
Every student of mechanics knows that pressure in an incompressible
material is not defined by strain and strain rate. As far as the constitutive
equation is concerned, the pressure in an incompressible material is an arbitr-
ary constant. It can assume any value. The pressure of a gas (considered
compressible) is determined by its volume or density. The pressure of water
(considered incompressible) is determined not by the water volume or density,
but by its motion and boundary conditions. The pressure in a piece of
steel (which is compressible) is determined by the strains in the steel (being
proportional to the mean principal strains). The pressure in a piece of rubber
(considered incompressible) is not determined by the deformation of the
rubber, but by the equations of equilibrium (or motion) and the boundary
conditions. Note that we usually consider steel as compressible but rubber as
incompressible, not because rubber is harder to deform, but because its elastic
modulus for volume change (the bulk modulus) is 10,000-100,000 times larger
than its Young's modulus and shear modulus. For steel, the bulk modulus,
Young's modulus, and shear modulus are of the same order of magnitude.
Most biological materials are considered incompressible for the same reason
as rubber is: they can be deformed easily, but their volumes are hard to change.
An interstitium without gas bubbles is incompressible. Therefore, the pres-
sure in it has to be determined by the conditions of equilibrium (or motion)
and the boundary conditions. This most elementary fact must be remembered,
or one becomes confused. To assess the interstitial pressure at a certain point,
we must begin with the external environment whose pressure we do know,
and trace it step by step through each membrane and each compartment,
until the point is reached.
In the following, let us make a digression from the interstitium and follow
Per (Pete) Scholander (see Hammel and Scholander, 1976, p. 41 et seq.) to
consider some instructive examples and provocative thoughts.
In Fig. 9.7:1 several cases considered by famous physicists are shown. To
the left is Lord Kelvin (W. Thomson),s (1871) capillary with a drop of water
in its closed end. The capillary is placed in an isothermal evacuated jar with
a layer of water. The vapor from the water will distill down into the capillary
until the height of the meniscus reaches the same level as if the capillary were
open below. Lord Kelvin derived a formula relating the curvature of the
326 9 Mass Transport

FIGURE 9.7:1 A series of experiments considered by famous physicists. (A) Kelvin's

capillary. (B) Arrhenius's osmometer. (C) Noyes's osmometer. (D) and (E) Hulett's
water and osmotic columns. In all these cases the chamber is isothermal and evacuated.
From Hammel and Scholander (1976), reproduced by permission.

meniscus and the lowering of the water pressure immediately below the
meniscus, and explained the movement of the water vapor into the capillary
tube by the lowering of water pressure in the tube.
Ten years later, J. H. Poynting (1881) suggested that the vapor pressure
change is in fact caused by the difference in water pressure. Based on the
theorem of equipartition of energy, he wrote a simple equation:
vL ' dpL = vg' dp g , (1)
i.e., the molar volume of water times a change of its pressure is equal to the
molar volume of the gas phase times its change in pressure.
In 1889, S. Arrhenius saw an important relation to osmosis in Poynting's
equation. He considered an osmotic column (Fig. 9.7:1B) with its semiperme-
able membrane dipping in the water, and concluded that the solution would
rise to a height at which the vapor pressure would match the rise from the
water surface below; otherwise distillation would take place.
In 1900, A. Noyes moved the membrane to the top of Arrhenius's column,
covering it with an infinitely thin layer of solution (Fig. 9.7: 1C). He could then
claim that the osmotic pressure times cross-sectional area is equal to the
weight of the water beneath the free surface.
In 1902, G. Hulett gave a dynamic interpretation of the negative pressure
of water below the membrane of Noyes's osmometer. He saw that the negative
pressure is derived from the reflection of the solute molecules from the free
surface and is analogous to a capillary lift from a membrane (Fig. 9.7:10 and
E). The free surface of the solution is exposed to the vapor. The solute
9.7 Pressure in an Incompressible Material 327

FIGURE 9.7:2 Scholander's experiment

on sedimented starch column at equilib-
rium. Pure water gradients are measured
with wick catheters. From Hammel and
Scholander (1976), by permission.

molecules tend to leave the surface, and are pulled back by the solvent, thus
imparting tension to the solvent.
The last point was made clear by Scholander in his experiment on sedi-
men ted starch matrix (Fig. 9.7:2). A test tube is filled with water and some
starch (left figure). The starch is thixotropic. A cotton wick in a plastic tubing
is put at the bottom of the starch column, while another is put in the clear
water at the top. Each one is connected with an identical glass capillary,
scrupulously cleanned. The menisci are seen to stand at the same level. Now,
let more starch be added until it hits the water surface, making it pasty by
capillary lift. At this time, both capillaries will show negative pressure (Fig.
9.7:2, right). This shows that the interaction between the starch and water
occurs at the free surface, and at that surface only. The relevance of this
example to the matrix force in the interstitium shown in Fig. 9.7:1 is obvious.
Other examples are given in Hammel and Scholander (1976). A particular
application of the principle to botany is very interesting. This is concerned
with the question: What drives the sap of a tree from the root to the top in
such tall trees as the redwood which may stand as tall as 100 m? First, it was
found that the xylem water is nearly pure water, with a freezing point of 0.1 °C
or less. For a tree like the mangrove which can grow in sea water, the root
cells that act as a semipermeable membrane must separate the xylem water
from the sea water at an osmotic pressure difference of some 24 stm. To draw
pure water into the xylem from the sea is a process of reverse.osmosis that
needs a driving pressure greater than 24 atm. The source of the driving
pressure was found to be located in the parenchyma cells of the leaves on
328 9 Mass Transport

which the xylem terminates. The parenchyma cells contain a high concentra-
tion of solutes and have supporting matrix. They are exposed to atmosphere.
When the leaves lose their turgor, the cells are flat and soft, so the internal
pressure in the cells must be nearly atmospheric. Now, across that part of the
cell membranes which are in contact with the xylem water, an osmotic pressure
difference exists. If we apply Starling's law (Eq. (35) or (36) in Sec. 8.5) to the
cell membrane, we find that the difference in static pressure must balance the
osmotic pressure on the two sides of the membrane at equilibrium (noflow).
Since the sum of osmotic and static pressure on the cell side is atmospheric,
the static pressure in the xylem water must be negative because the osmotic
pressure is essentially zero there. Hence it explains the negative pressure in
the xylem.
Using a pressure chamber as shown in Fig. 9.7:3, Scholander et al. (1965)
compressed the leaves until the cells lose their turgor and the xylem water is
expressed from the cut stem of the twig. This pressure must be equal to the
negative pressure in the xylem before the twig was cut. In this way, they
measured the sap pressure in a variety of vascular plants, and demonstrated
that all plants tested had negative sap pressure, most remarkably so in the
mangroves (Rhizophora and Lagunchularia) and desert plans. See Fig. 9.7:4.
Hammel and Scholander (1976) prefer to think of water on the two sides
of a semipermeable membrane as a single continuum with identical properties;

a
b

FIGURE 9.7:3 Pressure chambers used by Scholander et al. for measurement of sap
tension in a twig. (a) Direct observation. (b) Step-by-step sap extrusion and pressure
measurement to obtain a pressure-volume curve. From Scholander et al. (1965), by
permission.
 ATM
\0
• DAY :...,
-aol o NI GHT • "Q

• ~
'"t:
• a
-60t- • 5'
••• I»
• 1- 1 .: •• :::
.....
I.' ~
-401- o •
I.· •• o
Q I 9
-•
11 • • I "0
9 a
I '"~ .
-20 <J
• (b
• ••
• • I •-• , • I :::
I»
0
• fr·< .-- ••• • 1 • • - • • • •
::l.
eo.
"
." } -!::
g .2 4J \... 0 -
" '"
~ .. E
" ): ¢I -;: ~ ~ ~ ."
o .2...2.g "" .~ ~ ~ 0 ..2 )( ~ '- ': -g ~ _ = .2 ~ § § .~ §~ ~ 8 ~
000 0':1 ....
c- v.- "
.. 0 0,
... 0 0 'c: 0
-a~~ ~~ C' E 'r Col
.~ ~ .~ ~ ~ ~ ~
~
" ~ - "g.
~.:o
i. g ~ E ~ i. ~ ~ .~ ~ -~ ~ _2 ,g ~ ~ ~~E te~
e 0 2:. v - "\: oQ.I
-c: :... 0
.~ .~ ~ ~ '~- ~0
a o >. Col E .... E 'c t Q. 0
&nO"," >-. 0 ""
::!:U:x:<>
ell ~ 0
a..
Q:; ."( ... CQ Q ." U ",Ov> ~~OJ;~~~0~ ~~~~£t5~;:OQ£~~

~fftl
~
SEA SHORE o E R T FOREST FRESHWAT E R
W
N
\0
FIGURE 9.7:4 Sap pressure in a variety of vascular plants. Most measurements were taken during daytime in sunlight. Night values are likely
to be less negative by several atmosphere. From Scholander et al. (1965), by permission.
330 9 Mass Transport

in particular, at equilibrium they have the same pressure. If the pressure of

water on the side of pure water is negative, then that on the solution side must
be negative also and at the same value. This is Scholander's tensile water
theory. He generalized it to all solutions in living organisms. To him, the
explanation of osmotic pressure and surface tension has to be sought at the
interface: the very boundary of the fluid, with a dynamic view of the solutes
as in the molecular theory of gases and liquids. Some people challenge his
view on the molecular theory of solutes. I have no doubt that the basic premies
that in an incompressible fluid one can determine pressure only by tracing the
balance of forces to the boundary where stress is known is correct.

9.8 Lymph Flow

There was a time when people believed that fluid enters the tissue space from
the arterial ends of the capillaries and all of it is returned to the capillaries at
their venular ends. The fluid balance in the tissue is thus maintained by the
capillaries. This idea was known as Starling's hypothesis (Starling, 1896).
More recent measurements, however, have shown that the return of fluid to
the blood vessels and the prevention of edema relies on the lymph vessels
(Zweifach and Prather, 1975). This is true not only for water, but also for
various solutes, especially macromolecules.
The lymphatic vessel system is similar to the venous system except that the
terminals (capillaries) are blind-ended. Like veins, the lymph vessels have
one-way valves, and the lymph flow can be propelled by contraction of the
vessels. Such contractions can be produced either by the smooth muscles in
the lymphatic vessel wall, or by the motion of nearby arterioles, or by the
contraction of the skeletal muscles, or movement of the organs (e.g., motion
ofthe arms or legs, peristalsis of the intestine, or breathing of the lung). Figure
9.8:1 shows several records of pressure waves in the lymphatic vessels of the
mesentery of the cat. The pulsatile nature is clearly seen. The period is long.
Between a pair of valves (not necessarily consecutive), the propulsion of the
lymph fluid is similar to that of the ventricles of the heart, consisting of a
diastolic filling phase, and a systolic expulsion phase.
Thus, beyond the first valve, there is no difficulty in understanding the
nature oflymph flow. Conceptual difficulty lies with the terminal lymphatics.
How is it filled? What force drives the fluid from the tissue space into the
terminal lymphatics?
To answer this question, let us first consider the structure of the terminal
lymphatics. These vessels are extremely thin-walled and are much wider than
the blood capillaries. They possess irregular contours, vary in diameter from
15-20 J.1m, and have flattened segments that are as wide as 300 J.1m (Cliff and
Nicholl, 1970). They have different geometric patterns in different tissues.
In the mesentery they conform to the modular configuration of the blood
capillary network. In the skeletal muscle, lymphatics are found in the immedi-
9.8 Lymph Flow 331

[~ ----------
PC'2

j~i \
1.2 tD! HtO
~~ ~t\ , __________ _
(; .\ov V

..."',
' ..........,
"'VALVES

I
,II
I
I
I

~ ~
o OIII¥
------
. ------ ----
COLLECTING LYMPHATIC
CHANNEL (200p)

FIGURE 9.8:1 Pressure waves in the terminal lymphatic network in the mesentery of
the cat. From Zweifach and Prather (1975). Reprinted by permission.

ate neighborhood of the arterioles (see Fig. 9.8:2). In the lung, the terminal
lymphatics lie at the junctions of interalveolar septa. The walls of the terminal
lymphatics consist of an attenuated endothelium which is similar to, but
usually more attenuated and looser than the endothelium of blood capillaries,
and is frequently without a basement membrane. With such a loose structure,
the fluid in the terminal lymphatics seems to be in free communication
with the rest of the interstitium. Thus, the lymph fluid has much the same
332 9 Mass Transport

FIGURE 9.8:2 Top: Arterioles and venules in a spinotrapezius muscle ofthe rat. Bottom:
the same arterioles and the lumphatic vessels which were visualized by microinjection
of Evans Blue. Note the lumph terminal at the right. From Skalak et al. (1984).
Reproduced by permission.

composition as the interstitial fluid, and the osmotic pressure difference cannot
be a significant driving force.
Some authors (see Leak and Burke, 1968) have described connective tissue
fibers that are attached to the surface of the endothelial cells. These fibers
are believed to hold the endothelium tube open and to "anchor" it to the
interstitial matrix. According to this view, the terminal lymphatic is filled
because of the natural tendency of the vessel to be open.
Skalak, Schmid-SchOnbein, and Zweifach (1984) have shown that, in the
spinotrapezius muscle of the rat, all lymphatic vessels in the 20-200 !lm
diameter range lie in immediate neighborhoods of arterioles or venules. See
Fig. 9.8:2. The majority of the lymphatics are in direct contact with the
arterioles. When the arterioles were dilated, the contiguous lymphatics were
seen to be partially or completely collapsed, whereas lymphatics around
contracted arterioles were seen to be wide open. No significant deformation
of the adjacent skeletal muscle cells was observed. These lymphatics have no
smooth muscle; they have only a thin lining of endothelium. These results
suggest that the contraction and relaxation of the vascular smooth muscle in
the arterioles serves to open and close the lymphatics; thus vasomotion of the
9.9 Measurement of Extravascular Space by Indicator Dilution Method 333

arterioles and venules appears to be the motive force for lymph transport in
terminal lymphatics.

Lymph Flow During Hemorrhage

If enough blood is lost in hemorrhage so that the mean arterial blood pressure
is reduced to 50-60 mm Hg, what happens to lymph flow? In such a person,
extensive vasconstriction would raise peripheral resistance and the hydro-
static pressure at the capillary level may fall to 8-10 mm Hg. Under this
condition, reabsorption of fluid is favored, and the uptake of fluid will reduce
circulating plasma protein concentration by up to 30-50%. The hemodilu-
tion develops rapidly and approaches asymptotically to a maximum level in
approximately 60 mins. During the initial response to hemorrhagic hypoten-
sion, lymph flow is drastically reduced. With the completion of hemodilution,
lymph flow reappears at a reduced rate.
Baez (1960) observed that, in animals suffering severe blood loss, the
frequency of spontaneous contractions of the collecting lymphatics in the
intestinal mesentery increases some two- to threefold. Such heightened activity
leads to a visible collapse of the terminal lymphatics, and after 20-30 mins,
the spontaneous vasomotion slows and finally stops.
At a later stage in hemorrhage reaction, as tissue hypoxia develops, espe-
cially in tissue such as the small intestine, there is a marked increase in
peristalsis. At the same time, an increased vascular permeability develops in
these splanchnic viscera, as shown by loss of plasma protein and numerous
petechial hemorrhages into the tissue. Simultaneously, spontaneous contrac-
tion of lymphatics increases and lymph flow picks up. On the other hand,
lymph flow from the skeletal muscle of the extremities remains low (Arturson,
1971). These findings are suggestive of some feedback mechanism in the
intestinal wall which responds to the changes in intestinal plasma protein
level. A host of possible sensors and feedback loops may be suggested, but
identification is a job for the future.

9.9 Measurement of Extravascular Space by Indicator Dilution

Method

An indicator (or tracer) is a substance injected into a system in order to

measure the distribution of some endogenous substance in the system. It must
have the following properties:
1) It can be introduced into the living body.
2) Its presence does not alter any properties of the system.
3) It is neither metabolized nor sequestered in the system; all of it can be
recovered.
4) It can be distinguished from other substances in the system and can be
measured.
334 9 Mass Transport

10 • Cr 51
E 6 01 125
....... ~THO
c
.2
~
u.. 4

-o
~ 1.0
..Q
'S
2
)(

""~
0.1' - ---LLL ---'--_.l----l'----'-_--'------'
o 10 20 30
\:~~::: :~ 30
0 L....-.L.l.l-.-:l~O----'--:2:X:-
.. L - J
0
TI ME (seconds)
FIGURE 9.9: 1 A typical set of indicator dilution curves. The ordinate shows the outflow
fraction per ml, in logarithmic scale on the left, and in linear scale on the right. The
abscissa shows time. The injection time and duration is shown by the black block on
the abscissa. The sampling time interval is 1.09 sec. in this example. From Goresky et
al. (1969). Reproduced by permission.

5) Its distribution in the system is a measure of the way the endogenous

substance of interest is distributed.
6) The input-output relationship is linear; i.e., the distribution of the indica-
tor in the system is a linear function of the input.
7) The basic system is in a steady-state.
Consider an organ fed by one artery and drained by one vein. Assume that
the organ has a constant volume and that the inflow is steady. Let a tracer of
amount qo be injected into the artery as an impulse at time zero. Let the
concentration of the tracer measured at an outflow section in the vein be
denoted by c(t). (See Fig. 9.9:1 as an example). We normalize c(t) and define
an impulse-response function as follows:

h(t) == c(t) II") c(t) dt. (1)

The function h(t) is nonnegative, and

I''' h(t) dt = 1, (2)

therefore, it has the character of a probability density function. Since t is the

time after injection, and h(t) is the portion of the tracer that arrived at the exit
section in a time interval dt after spending a "transit time" t in the system, h(t)
is spoken of as the probability density function of the transit time t.
9.9 Measurement of Extravascular Space by Indicator Dilution Method 335

The precise definition of c(t) is as follows: Assume that the vein is straight
and let x, y, z be the coordinates in the vein, with the x-axis coinciding with
the axis of the vein. Let the tracer be measured at an "exit" cross section
located at x = L, where the velocity normal to the section is u(L,y,z). Then
the mean concentration of the tracer is

c(t) =~ ff c(L, y, z, t)u(L, y, z) dy dz, (3)

where F is the "flow":

F = ff u(L,y, z) dydz. (4)

The integrals are taken over the entire cross section.

From Eq. (3) we can show that

foro c(t)dt = qo/F, (5)

where qo is the total amount of the tracer injected. This is because cu dt dy dz

is the quantity of tracer that flows across dy dz in time dt, so that the integral

fff cudtdydz = qo· (6)

Measurement of Flow
Equation (5) can be used as a basis to measure flow. For example, to measure
cardiac output, one may inject, impulsively, an amount of dye qo, into the
right heart, and record the concentration of the dye as it passes through the
aorta. The area under the concentration curve is the integral on the left-hand
side of Eq. (5). Then Eq. (5) can be used to compute the cardiac output F.

Response to Arbitrary Injection History

Equations (1) and (5) give the impulse response function

c(t) = ~ h(t). (7)

This formula can be generalized to give the response to an arbitrary injection

which has a history q(t) per unit time. Regarding the input as a superposition
of successive impulses of magnitude q(t) dt, we have, under the assumption of
linearity between input and output, the result.

c(t) = -1 It q(r)h(t - r)dr. (8)

F °
336 9 Mass Transport

Measurement of the Volume of the Traced System

If q(t) is a step function with amplitude A, then the concentration at the exit
section is the indicial-response function

c(t) = ~ (t h(t _ -r:)d-r: = ~ (t h(-r:')d-r:'. (9)

F Jo F Jo
Now, FC(t) is the rate at which the tracer leaves the system. Therefore, the
difference AH(t) - FC(t) is the rate at which the tracer is accumulated in the
system, H(t) being the unit-step function. An integration gives the total amount
of tracer that is accumlated in the system.

Q(t) = I [AH(t) - FC(t)] dt. (10)

As t -+ 00, Q(t) tends to a constant Q( (0). From Eq. (9),

Q(oo) = A Loo [1 - f~ h(t)dt]dt. (11)

Integrating by parts and remembering Eq. (2), we obtain

Q( (0) = A Loo th(t) dt = At, (12)

where tis the mean transit time defined by the integral in Eq. (12). Now, under
constant infusion, Eqs. (9) and (12) show that as t -+ 00, the concentration at
the exit tends to a constant c( (0) = AIF. If we assume that the concentration
tends to c( (0) throughout the entire system, then c( (0) multiplied by the volume
V of the system is the total tracer left in the system, Q( (0). Thus, from Eq. (12),
we obtain the principal result
V= Ft. (13)
Thus the flow multiplied by the mean transit time gives the volume of the
system. This derivation was given by Meier and Zierler (1954) and Zierler
(1963) and can be applied to measure the size of the left ventricle, the total
blood volume in the lung, etc.
Zierler's hypothesis that the concentration be uniform throughout the
entire system, unfortunately, is violated in many practical cases. Hence we add
the remark that Eq. (13) hold under the weaker condition that the averge
concentration of the tracer over the entire system be equal to c( (0). The proof
follows Eq. (12) immediately.

Volume of Extravascular Water in an Organ

Consider a specific system, e.g., the lung. Let us simultaneously inject two
tracers and continuously measure their concentrations at an exit section. Let
the probability density functions of the transit time be h(t) and y(t) for these
9.10 Tracer Motion in a Model of Pulmonary Microcirculation 337

two tracers. The mean transit times f(h) and f(y) may be different. The
difference
L\ V = V(y) - V(h) = Ff(y) - Ff(h) (14)
represents the difference of volumes traced by these two tracers.
If one of the tracers is labeled water which permeates through both vascular
and extravascular spaces, whereas the other tracer is confined to the blood
stream, then L\ V gives the extravascular water volume. This is a practical way
to determine extravascular water in the lung and edema.
These formulas can be generalized in a simple way to cover the situation
in which a tracer occupies only part of the fluid. For example, for a tracer
which tags only red blood cells, or another which tags only the plasma, the
corresponding flow rate will be only a fraction of the flow rate of the whole
blood.

9.10 Tracer Motion in a Model of Pulmonary Microcirculation*

As an example of analysis of the motion of fluid in very narrow tissue space,

the tracer motion (Sec. 9.9) in the extravascular and extracellular interstitial
space of the interalveolar septa of the lung is presented. This space lies between
the basement membrane of the pulmonary capillary sheet and the layer of
epithelial cells. In the human lung, it is a multiply-connected sheet of the order
of 100 m 2 in total area and 1 ~m in thickness, with the total volume in the
order of 100 mI. But the thickness and volume are highly variable. The
total area changes with the transpulmonary pressure (airway minus pleural
pressure). The thickness varies with the blood pressure, tissue pressure, osmotic
pressures of the blood and the tissue, the tightness of the epithelial cell
junctions, and the lymph drainage.
Figure 9.10:1 shows a sketch of an idealized capillary blood vessel sheet
between an arteriole and a venule: a channel bounded by two thin layers of
porous material. The channel (compartment No.2) represents the capillary
blood vessel. The porous layers (compartment No.1) represent the tissue
space. The membrances that divide these two compartments are permeable
to water, but are semipermeable selectively with respect to the solutes. The
external boundaries (the epithelia at z = ±(~ + c5», are impermeable to both
water and solutes. Beyond the epithelia is the alveolar air space. The channel
is supplied by an arteriole and drained by a venule, which are represented
here by uniform channels with impermeable walls. Variables in the two
compartments are distinguished by the subscripts 1 and 2.
The justification of this idealized model is discussed in Fung (1983), Chap.
6. The fluid coming from the left in the channel will permeate through the
membrane into the porous layer under the influence of hydrostatic and

* This analysis was given by Tang and Fung (1975) and Fung and Tang (1975).
338 9 Mass Transport

permeable
wall

L-----------~~----------;_~x

impermeable channel space impermeable

~ -
--.~.--.--.--.--.~.--.--.--.~
'\ \,

FIGURE 9.10: 1 Flow of blood, interstitial fluid, and tracer in a pulmonary interalveolar
septum. The upper figure shows the compartments of the blood (No.2) and the porous
interstitial space (No.1), the boundary conditions, the coordinates, and the dimensions.
The lower figure shows a sketch of the velocity distribution. From Fung and Tang
(1975), by permission.

osmotic pressures. Because of fluid viscosity, the horizontal velocity com-

ponent must vanish on the wall. Therefore, at the interface, the velocity must
be perpendicular to the wall. In the porous layer, the fluid flows to the right
and back into the channel.
The blood is assumed to be Newtonian. The interstitial fluid is assumed to
obey Darcy's law (Eq. 9.5:1). The fluids and the tissue are incompressible. The
Reynolds number of flow is very small (<< 1). Hence with the coordinates
shown in Fig. 9.10:1 and restricting our consideration to the two-dimensional
case, we have the following governing equations:
In the porous layer, t ~ z ~ t + 15, Darcy's law applies:

(1)

On the boundaries x = 0, x = L, and z = ±(t + b), there is no slip:

Ul = Wl = 0. (2a)
On the interface at z = ±t Starling's law applies:
(2b)
Here Ul' wl are velocity components in the x and z directions, respectively; K
is Darcy's constant, K is the membrane permeability constant, p is pressure,
9.10 Tracer Motion in a Model of Pulmonary Microcirculation 339

n is osmotic pressure, (J is reflection coefficient, J is the thickness of the

interstitium. The subscripts 1, 2 refer to the compartments 1, 2 respectively.
In the channel, 0 :::; z :::; ~, Stokes equation applies:

(3)

where J1 is the coefficient of viscosity of the fluid. The boundary conditions at

the ends x = 0 and x = L are assumed to conform to Poiseuillean flow:

U2 = ~U
2
(1 _4hZ2 ) 2 '
w2 = O. (4a)

The boundary condition on the walls z = ± ~ is no-slip:

u2 = 0, (4b)
U is the mean velocity of flow in the channel ahead of the porous walls.
The incompressibility condition is expressed by
OU ow (5)
ox + oy = O.
Further, the distribution of each tracer (a solute or a component of blood)
obeys Eq. (9.4:6). Let C be the concentration of the solute, D be its coefficient
of diffusion, (u, w) be the velocity vector; then,

OC OC OC (02C 02C)
ot + u ox + w OZ = D ox 2 + OZ2 ' (6)

and the boundary conditions are

oC l oC 2
Dl Tz - D2 Tz = w(c l - C2), for a permeable wall, (7a)

oC l
Dl Tz = WC 1 , for an impermeable wall, (7b)

OC l = 0 for a wall impermeable to water. (7c)

oz '
The solutions are as follows:
Flow in the Interstitium. Equation (5) is satisfied by an arbitrary stream
function t/J if u, ware defined by
ot/J ot/J
U= - -
OZ' w=a;' (8)

Using (8), Eq. (1) is reduced to

(9)
340 9 Mass Transport

It can be proven by direct substitution that the following solutions satisfy Eqs.
(2a), (2b), and (9) if An = nn/L, n = 1,2,3 ... , and Z ~ 0:

1/11 (x, z) = n~l an sin An x sinh An [ G+ b) - JI Z sinh AA (10)

P1 (x, z) = n~l ~ COsAnX cosh An [ G+ b) - JI z sinh An b+ b

1, (11)

where b 1 is an integration constant. This leaves the boundary condition Eq.

(2c) at z = ±~ to be matched later with the solution in the channel. For z ::;; 0,
1/11 and P1 are mirror images of Eqs. (10) and (11).
Flow in the Channel. With Eq. (8), Eq. (3) can be reduced to the form

(12)

Let

(13)

Then the perturbations u;, w~ due to permeability of the wall must satisfy the
following equations:
u; = 0, w~ = 0, at x = 0, L; (14)
u; = 0, w~ = w 1, at z = ±h/2; (15)
u; = -ol/l;/oz, (16)
'\141/1; = 0. (17)
According to the methods discussed in Fung (1965, p. 206 et seq.), we find the

°
following solution which satisfies the condition of symmetry with respect to
z and the boundary condition u; = at x = 0, L:

1/1; = n~l ( An sin An x sinh An z + Cn 2: sin Anx cosh AnZ)

+ f [Gn(1 -
n=l
~)
L
sinh YnX sinh YnZ

+ Hn f sinh Yn(L - x) sin Yn Z J, (18)

where An = nn/L, Yn = 2nn/h, and n = 1, 2, 3, .... This can be proven by

substitution. The boundary condition w~ = W 1 at Z = ±~is satisfied by choos-
ing
(19)
The remaining unknown constants An, Gn, and Hn can be determined in terms
9.10 Tracer Motion in a Model of Pulmonary Microcirculation 341

of an by using the yet unsatisfied conditions u;

= 0 at z = ±~ and w~ = 0 at
x = 0, L. The complete solution is obtained by matching the solutions, Eqs.
(10) and (18) at the boundary, Eqs. (2b) and (15). The equations are complex,
but the numerical calculations are straightforward. See Tang and Fung (1975)
for details.
Distribution of the Tracer. If a tracer (such as THO) is permeable to
the membrane at z = ±~, then at a steady-state the concentration of that
substance is uniform in both compartments. This is because a solution of Eqs.
(6) and (7a) is
c l = C2 = const (20)
If a substance (such a hyaluronate) in compartment 1 (interstitium) cannot
penetrate into compartment 2 (blood vessel), then in a steady-state the solu-
tion must satisfy Eq. (6) with u l , W l given by Eq. (10), and the boundary
conditions (7b) at z =~, and (7c) at x = 0 and L. It can be verified by
substitution that the solution is
cl(x,z) = coexp[ -q>(x,z)], (21)
where Co is a constant and
1 00
q>(x,z) = Dl n~ an cos AnX cosh An [(h/2 + (5) - z]/sinhAn <5· (22)

The verification is easy if one notices that V2q> = 0 and that the functions q>
and t/ll satisfy the Cauchy-Riemann differential equations for an analytical
function q> + it/ll of a complex variable z = x + iy.
Numerical evaluation of q>(x,z) is given in Fung and Tang (1975). It was
found that for values of L, h, <5 pertaining to mammalian lungs, the values of
q> are small; hence the concentration c 1 (x, z) is nearly equal to a constant
through the interstitium.
Finally, consider a substance (such as albumin) which exists in the vascular
space (compartment 2) but cannot penetrate into the interstitium. Then its
concentration C2 must satisfy Eq. (6) with u 2 , W 2 given by Eqs. (13), (16), (18),
and the boundary conditions (4a), (4b), and (7b). In this case the exact solution
is unknown. An approximate solution can be obtained by relaxing the bound-
ary conditions W = 0 at x = 0, L, yielding a simplified solution t/I; given by
Eq. (17) with Gn= Hn = O. Then the coefficients An, en are proportional to
an of Eq. (10). Alternatively, Ritz's method can be applied to obtain an
approximate solution. In this method we write down a solution that satisfies
all the boundary conditions and contains a set of undetermined coefficients.
These coefficients are then determined by minimizing an error integral I,
representing the square of the deviation from the differential equation when
the approximate solution is substituted into it.
L J2 dx dz.
I= Ji Jih/2 [oc
0 0 U2 0: +
oC0: -
W2 D2 V2C2 (23)
342 9 Mass Transport

If I = 0, then Eq. (6) is satisfied exactly. If an exact solution is unknown, one

determines the undetermined coefficients so that
I = minimum. (24)
Numerical results are given in Tang and Fung (1975), and the following
conclusions are reached:
1) If a tracer (such as THO) is permeable through the membrane that
separates the blood from the tissue space, which in turn is limited by an
impermeable wall, then at a steady-state the concentration of that tracer is
uniform in both compartments.
2) If a tracer is confined to the vascular space, then, because of water
movement through the porous layers, the concentration of the tracer in the
channel is nonuniform, but the average concentration (averaged over the
entire channel) is the same as that at the entry.
3) The hypothesis for the validity of the tracer-volume theorem, Eq. (9.9:13),
or V = Fi, is therefore fulfilled. We conclude that the tracer-volume formula
can be used for the pulmonary alveoli.
4) If a solute is confined to the porous layers, then its concentration
distribution is nonuniform because of the nonuniform convection of water
through the boundary. The nonuniformity, however, is negligible for practical
ranges of physiological parameters of the lung. Thus solutes in the interstitial
space are nearly uniformly distributed.

9.11 Peristalsis

Besides circulation, respiration, and interstitial fluid flow, there are many
other physiological flows, e.g., micturition, biphastic flow in arthroidal car-
tilage, urine extraction in kidney, peristalsis in ureter, intestine, stomach,
arterioles, venules, and lymphatics. The mathematical problem of micturition
is similar to that ofthe collapsible tube discussed in Sec. 5.16. Effort indepen-
dent limitation of flow in urination occurs because of muscle contraction at
the beginning of urethra. Flow in cartilage is an example of biphasic flow
considered in Sec. 8.9. Urine transport in kidney is an example of active
transport (Sec. 8.8).
Peristalsis is a muscle-controlled flow akin to the flow in the heart. Since
peristalsis occurs in many organs, one example, that of the ureter, is discussed
in greater detail below.
Ureteral peristalsis was described by Aristotle (384-322 B.C.) in one of his
books on animals (Historia animalium) several hundred years B.c. The ureters
collect urine from the kidneys and send it to the bladder. In X-ray the waves
of urine appear as shown in Fig. 9.11:1: discontinuous boluses passing down
slowly one at a time. At the bladder each ureter passes through a valveless
one-way valve, called the ureterovesicular junction, which is a Z- or U-shaped
fold of the ureter in the wall of the bladder. It works by the pressure in the
9.11 Peristalsis 343

FIGURE 9.11:1 Au X-ray photograph of a human ureter.

bladder. When the bladder is full, its pressure is high, its wall is in tension in
the circumferential direction. The bladder pressure presses on the Z or U-
shaped fold in the bladder wall and collapses it, forming a check valve, and
stopping back flow into the ureter. This valve can be opened by each bolus
of urine in the ureter if the pressure in the bolus of urine in the ureter exceeds
the lateral pressure imposed by the urine in the bladder and the muscle in the
bladder wall. If the smooth muscle of the ureter is unable to generate such a
higher pressure in the bolus of urine, or if the legs of the Z or U have
inadequate length, then the ureterovesicular junction would not work properly
and a disease called hydro-ureter results. A hydro-ureter is a swollen ureter
with much increased lumen filled with urine. The reason why hydro-ureter is
a disease becomes clear when one considers the hoop stress in a pressurized
tube. In a tube of radius a, a tention T generated by the ureteral smooth muscle
will create a pressure p = T/a. For a given T, p can be large if a is small. But
344 9 Mass Transport

in a hydro-ureter a becomes so large that the pressure which can be generated

by the ureter is insufficient to send the urine through the ureterovesicular
junction. Then urine is collected in the ureter, backs up to the kidney, and
before long causes kidney disease. A weak ureterovesicular junction would
also permit reflux, causing some bacteria which might exist in the bladder to
reach back to the kidney.
The surgical treatment of hydro-ureter is to cut the ureter open longitudi-
nally, remove the excess material, and suture the vessel back into a tube of
smaller radius. Repair the ureterovesicular junction if it were the cause.
Thus, to understand the physiological process of ureteral peristalsis, one
must study the ureteral smooth muscle, find out how it generates tension, what
maximum tension can be generated, and how the tension moves the urine.
Mathematical studies of peristalsis were initiated by Shapiro et al. (1969),
Fung and Yih (1968), and others. Most of the analyses are based on the
Navier-Stokes equation, considering flow in a circular cylindrical tube or
two-dimensional channel with a sinusoidal displacement wave traveling in its
wall at a constant velocity. The objects ofthe studies are: (1) to determine the
longitudinal pressure gradient that can be generated by the traveling wave;
(2) the flow resulting from peristalsis superposed on pressure differences
at the ends; and (3) conditions of reflux. To simplify the analysis, various
approximations are introduced such as: (1) small amplitude of the wall dis-
placement compared with the undeformed radius of the tube; or (2) long wave
length compared with the tube radius; or (3) very small Reynolds number
so that the nonlinear convective acceleration term in the Navier-Stokes
equation can be neglected, or a combination of these. Mathematical tech-
niques include the use of stream functions, resulting in Bousinesq equation
or Oberbeck-Bousinesq equation, series expansion in small parameters, and
numerical methods of finite differences, finite elements, and boundary inte-
grals.
Experimental fluid mechanical studies were made by Yin and Fung (1971a)
who compared theoretical predictions with experimental results. Formulation
of the ureteral peristalsis problem as a muscle controlled flow was first
presented by Fung (1971). Ureteral muscle mechanics was then studied by Yin
and Fung (1971b), and Zupkas and Fung (1985).
One conclusion reached by these studies is that peristalsis is an effective
method to move fluid only if the fluid is transported in the form of a series of
isolated boluses, as seen in Fig. 9.11:1. If the amplitude of the displacement of
the wall is small compared with the tube radius, very little pressure gradient
can be generated by the traveling wave. Pressure gradient increases signifi-
cantly when the radius of the minium section of the wave approaches zero.
No wonder then, that peristaltic waves of the ureter, intestine, and lymphatics
in normal conditions are in this mode. Artificial peristaltic pumps usually
operate in this mode also, fully occluding the tube between boluses.
Figure 9.11:2 shows an example of the theoretical shape of the ureteral
bolus and pressure distribution when the muscle contraction follows the
9.11 Peristalsis 345

0.25

0.2

0.15

2 4

~ = x-Cf (em)

N
E
u
...c
.....
>-
~
~

..."
II>
II>

Ii:
-2 8 10
~ = x-ct (em)

E
S
'"
....
II>

"
15
0
0::

-2.0 60 80 100
~ =x-ct (em)

20
N
E
u
.....
...c 15
>-
~ Vmcx
~ 10
0.2
"
...
II>
II>

Ii:

-20 20 40 60 80 100
~ =x-ct (em)

FIGURE 9.11:2 Theoretical shapes of the ureteral boluses and associated pressure
distribution. From Fung (1971), by permission.
346 9 Mass Transport

constitutive equation given by Fung (1970). The bolus moves from left to right,
rN is the radius of the bolus, x is the distance from the kidney, c is the wave
speed, t is the time. It is seen that the ureter is closed in front of the bolus. The
parameters of the ureter assumed in this example are given in the original
paper, Fung (1971). This example shows that the pressure in the front part of
the bolus is small, and that significant pressure is generated toward the end
of the bolus. This is because the muscle was relaxed before the wave reached
it; it generates tension gradually after stimulation. When the maximum ten-
sion is reached, the lumen of the ureter is forced closed. The closure at the rear
end generates a pressure that may be sufficiently high to send the urine bolus
through the ureterovesicle junction even when the bladder is full.
F or a full understanding of physiological peristalsis, future research should
focus on the smooth muscles, identify their constitutive equations for active
contraction, and their response to drugs and other physical, chemical, and
biological stimuli, as well as the muscle-controlled flow.

Problems
9.1 Design a method to measure the permeability of the capillary blood vessels of an
organ. The permeability constant in Starling's law is defined for fluid transfer per
unit area. Since the total area of the capillary blood vessel wall is unknown for
any organ, what can you do? Is the product of permeability and the capillary
wall area meaningful?
9.2 Using the indicator dilution method (Sec. 9.7), design an experiment to measure
the permeability of capillary blood vessel wall to water.
9.3 Formulate a mathematical theory to study whether red blood cells can serve as
markers for transcapillary fluid transfer in the method of Landis, Zweifach, and
Michel. Estimate the order of magnitude of possible errors due to this source.
9.4 Many authors believe that all capillaries in an organ are not open to flow;
many are closed at any given time. This leads to the concept of recruitment of
capillaries. It is suggested that with the development of essential hypertension
more capillaries are closed. Design an experiment to study this problem. What
animal and what organ will you choose? Why? What instruments are needed?
What man power? What financial resources are needed?
9.S Formulate an experiment to evaluate the distensibility of the capillary blood
vessels with respect to changes in hydrostatic pressure. How significant is the
result with respect to Landis's method for measuring capillary permeability?
Discuss this with reference to a specific organ of your choice.
9.6 Write a computer program using the finite element method to analyze the Krogh
cylinder (Sec. 9.4). Generalize the cylinder to hexagonal cross section so that a
space-filling system can be obtained. What other cross-sectional shapes should
be (or can be) studied?
9.7 Solve the Krogh cylinder problem of Prob. 9.6 with the finite differences method.
Discuss the pros and cons of the two methods.
Problems 347

9.8 What effect does vasomotion (continuous periodic contraction and relaxation)
of arterioles and venules have on blood flow and on fluid and solutes transport
in the skeletal muscle? Formulate a mathematical theory to investigate this
problem. What are the missing pieces of information that need experimental
determination? How can one get this information? Is the fact that the interstitial
space in the muscle is very small ( < 3% muscle volume) significant?
9.9 If the preceding question is asked with regard to the mesentery, what difference
does it make?
9.10 Where are the lymph terminals located in the pulmonary alveoli? Formulate a
theory for water movement in the interstitial space of the alveolar wall and its
relationship to the lymphatics, breathing, and possible damage to epithelium due
to drug abuse or disease.
9.11 Discuss the conditions that may cause edema in the lung.
9.12 What mechanism keeps the volume of living cells constant? Where is the source
of energy? Swelling usually occurs whan a cell dies. Why?
9.13 What controls water balance in an organ? Is there a feedback control mechanism
working for water balance?
9.14 What happens to a person if the oxygen tension in the blood is higher than
normal? What if the partial pressure of CO 2 is higher than normal?
9.15 Select an organ. Design an experiment to determine the permeability of capillary
blood vessels by the isogravimetric method.
9.16 Name some tracer materials that can be used to determine the volume of the
vascular space of an organ, and the extravascular water volume in that organ.
9.17 Consider all the factors that can influence pulmonary edema. Discuss what you
can do to control the respiratory distress syndrome of a young student involved
in a car accident.
9.18 Fluids in the human stomach and small intestine can be characterized as a
non-Newtonian viscous fluid. What kinds of fluid motion are physiologically
significant in these organs?
9.19 In skeletal muscle the level of oxygen in outflowing venous blood is much higher
than the level of oxygen in tissue. This implies that the blood oxygen does not
have enough time to come into equilibrium with the oxygen in tissue. How can
this be explained? Harris named counter current flow in arteries and veins as a
reason: the diffusion between parallel and adjacent arterioles and venules as a
kind of shunt from the tissue space. Discuss this idea more thoroughly. (Cf.:
Harris, P.D. (1986). Movement of oxygen in skeletal muscle News in Physiol. Sci.
1:147-149. Whalen, W.J., Nair, P. and Ganfield, R.A. (1973). Measurements
of oxygen tension in tissues with a micro-oxygen electrode. Microvasc. Res.
5:254-262.)
9.20 In the lung, lymph flow drains into the vein. If the systemic venous pressure is
temporarily increased, how would the pulmonary lymph flow change? How
would this affect pulmonary edema formation? (Cf.: G.A. Laine et al. (1986): A
New Look at Pulmonary Edema, News in Physiol. Soc. 1:150-153.)
348 9 Mass Transport

9.21 Peripheral nerves need nutrition and hence blood supply. Groups of nerve fibers
are enclosed in a specialized, multilayered membrane called perineurium into
fascicles. Small blood vessels enter each fascicle through the perineurium. If the
nerve is injured one of the findings is that water accumulates in the fascicle
(i.e. edema occurs), resulting in an increase of pressure and distension of the
perineurium. This distension takes place mainly in the circumferential direc-
tion-the length ofthe nerve fascicle changes little. The penetrating blood vessel
is distorted by the deformation of the perineurium: its cross section becomes more
or less elliptical or kidney shaped, or biconcave; its cross-sectional area is reduced;
the resistance to blood flow from outside of the fascicle to the inside through the
perineurium, or vice versa, is increased; and the blood flow is reduced. This is an
important problem that is relevant to nerve trauma. Formulate a mathematical
model to analyze this problem. The following tentative data may be used for such
an analysis (see Myers, R.R., Murakami, H., Powell, H.C.: Reduced Nerve Blood
Flow in Edematous Neuropathies: A Biomechanical Mechanism. Microvasc. Res.
32:145-151, 1986):
Perineurium Blood Vessel
Diameter 2,000 Ilm 50 Ilm
Wall thickness 151lm 51lm
Internal pressure 0.8 kPa 33 kPa
External pressure 0.0 kPa 0.0 (outside perin.)
0.8 (inside perin.)
Elasticity (Young's mod.) 3.3 kPa 50 kPa
Poisson's ratio 0.1 0.3

References

An, K.N. and Salatbe, E.P. (1976). A theory of interstitial fluid motion and its implications for
capillary exchange. Microvasc. Res. 12: 103-119.
Apelblat, A., Katzir-Katchalsky, A., and Silberberg, A. (1974). A mathematical analysis of
capillary-tissue fluid exchange. Biorheology, 11: 1-49.
Arrhenius, S. (1889). Einfache Ableitung der Beziehung zwischen osmotischem Druck und
Erniedrigung der Dampfspannung. Z. Phys. Chern. 1: 115-119.
Arturson, G. (1971). Effect of colloids on transcapillary exchange. In Hemodilution: Theoretical
Basis and Clinical Application, (K. Messmer and H. Schmid-Schoenbein, eds.), pp. 84-104.
S. Karger, Basel.
Aschheim, E. (1977). Passage of substances across the walls of blood vessels. In Microcirculation,
(G. Kaley and B.M. Altura, eds.), Vol. 1, Ch. 10, University Park Press, Baltimore,
pp.213-249.
Baez, S. (1960). Flow properties oflymph: A microcirculatory study. In Flow Properties of Blood
and Other Biological Systems, (A.L. Copley and G. Stainsby, eds), pp. 398-411, Pergamon
Press, New York.
Bassingthwaighte, J.B., Knopp, T.J., and Hazelrig, J.B. (1970). A concurrent flow model for
capillary-tissue exchange. In Capillary Permeability, (c. Crone and N.A. Lassen, eds.),
Academic Press, New York.
References 349

Bloch, I. (1943). Some theoretical considerations concerning the interchange of metabolites

between capillaries and tissue. Bull. Math. Biophysics, 5: 1-14.
Blum. lJ. (1960). Concentration profiles in and around capillaries. Am. J. Physiol. 198: 991-998.
Cliff, W.J. and Nicoll, P.A. (1970). Structure and function oflymphatic vessels of the bat's wing.
Quart J. Exp. Physiol. 55: 112-121.
Crone, C. and Lassen, N.A. (1970). Capillary Permeability, Academic Press, New York.
Curry, F.E. (1984). Mechanics and thermodynamics of transcapillary exchange. In Handbook of
Physiology, Sec. 2., Cardiovascular System, Yol. IV, Part I, (E.M. Renkin and e.e. Michel,
eds.) Amer. Physiol. Soc., Bethesda, MD.
Deen, W.M., Robertson, C.R., and Brenner, B.M. (1972). A model of glomerular ultrafiltration
in the rat. Amer. J. Physiol. 223: 1178-1183.
Duling, B.R. and Berne, R.M. (1970). Longitudinal gradients in periarteriolar oxygen tension.
Circ. Res. 27: 669-678.
Fung, Y.C. (1965). Foundations of Solid Mechanics. Prentice-Hall, Englewood Cliffs, N.J.
Fung, Y.e. (1970). Mathematical representation of the mechanical properties of the heart muscle.
J. Biomech. 3: 381-404.
Fung, Y.e. (1971). Peristaltic pumping: a bioengineering model. In Urodynamics: Hydrodynamics
of the Ureter and Renal Pelvis. (S. Boyarsky, C.W. Gottschalk, E.A. Tanagho, and P.D.
Zimskind, eds). Academic Press, New York, pp. 177-198.
Fung, Y.e. (1972). Theoretical pulmonary microvascular impedance. Annals of Biomed. Eng. 1:
221-245.
Fung, Y.e. (1974). Fluid in the interstitial space of the pulmonary alveolar sheet. Microvasc. Res.
7: 89-113.
Fung, ye. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
Fung, Y.C. and Yih, e.S. (1968). Peristaltic transport. J. Appl. Mech. 35, Ser. E. 669-675.
Fung, ye. and Sobin, S.S. (1972). Elasticity of the pulmonary alveolar sheet. Circ. Res. 30:
451-469.
Fung, Y.e. and Tang, H.T. (1975). Solute distribution in the flow in a channel bounded by porous
layers. J. Appl. Mech. 42: 531-535.
Fung, Y.e., Zweifach, B.W., and Intaglietta, M. (1966). Elastic environment of the capillary bed.
Circ. Res. 14: 441-461.
Goresky, C.A., Cronin, R.F.P., and Wangel, B.E. (1969). Indicator dilution measurements of
extravascular water in the lungs. J. Clin. Invest. 48: 487-501.
Granger, H.J. and Shepherd, A.P. (1979). Dynamics and control of the microcirculation. In
Advances in Biomedical Engineering, Vol. 7, pp. 1-63, (J.H.D. Brown, Ed.), Academic Press,
New York.
Gross, J.F. and Popel, A. (Eds.)(1980). Mathematics ofMicrocirculation Phenomena, Raven Press,
New York.
Guyton, A.e. (1963). A concept of negative interstitial pressure based on pressures in implanted
perforated capsules. Circ. Res. 12: 399-414.
Guyton, A.C., Barber, B.J., and Moffatt, D.S. (1981). Theory of interstitial pressures. In Tissue
Fluid Pressure and Composition (A. Hargens, ed.), pp. 11-19, Williams & Wilkins, Baltimore.
Hammel, H.T. and Scholander, P.F. (1976). Osmosis and Tensile Solvent. Springer-Verlag, Berlin.
Hargens, A.R. (Ed.) (1981). Tissue Fluid Pressure and Composition, p. 3, Williams and Wilkins,
Baltimore.
Hulett, G.A. (1902). Beziehung zwischen negativm Druck und osmotischen Druck. Z. Phys. Chem.
42: 353-368.
Intaglietta, M. and Johnson, P.C. (1978). Principles of capillary exchange. In Peripheral Circula-
tion (P.e. Johnson, ed.), Wiley, New York.
Johnson, J.A. and Wilson, T.A. (1966). A model for capillary exchange. Amer. J. Physiol. 210:
1299-1303.
Kedem, O. and Katchalsky, A. (1958). Thermodynamic analysis of the permeability of biological
membranes to non-electrolytes. Biochim. et Biophys. Acta. 27: 229-246.
350 9 Mass Transport

Krogh, A. (1919). The number and distribution of capillaries in muscle with calculations of the
oxygen pressure head necessary for supplying the tissue. J. Physiol. 52: 409-415.
Landis, E.M. (1927). Micro-injection studies of capillary permeability. II. The relation between
capillary pressure and the rate at which fluid passes through the walls of single capillaries.
Amer. J. Physiol. 82: 217-238.
Landis, E.M. and Pappenheimer, J.R. (1963). Exchange of substances through the capillary walls.
In Handbook of Physiology, Sec. 2: Circulation, Vol. 2, Ch. 29, American Physiological
Society, Washington, D.C.
Leak, L.V. and Burke, J.F. (1968). Ultrastructural studies on the lymphatic anchoring filaments.
J. Cell Bioi. 36: 129-149.
Lee, J.S., Smaje, L.H., and Zweifach, B.W. (1971). Fluid movement in occluded single capillaries
of rabbit omentum. Circ. Res. 28: 353-370.
Meier, P. and Zieler, K.L. (1954). On the theory of the indicator-dilution method for measurement
of blood flow and volume. J. Appl. Physiol. 6: 731-744.
Michel, C.e. (1978). The measurement of permeability in single capillaries. Arch. Int. Physiol.
Biochim. 86: 657-667.
Michel, e.e. (1980). Filtration coefficients and osmotic reflexion coefficients of the walls of single
frog mesenteric capillaries. J. Physiol. (London) 309: 341-355.
Middleman, S. (1972). Transport Phenomena in the Cardiovascular System. Wiley-Interscience,
New York.
Niimi, H. and Sugihara, M. (1984). Hemorrheological approach to oxygen transport between
blood and tissue. Biorheology 21: 1-17.
Noyes, A. (1900). Die genaue Beziehung zwischen osmotischem Druck und Dampfdruck. Z. Phys.
Chern. 35: 707-721.
Ogston, A.G., Preston, B.N., and Wells, J.D. (1973). On the transport of compact particles through
solutions of chain-polymers. Proc. Roy. Soc. (London) 333: 297-316.
Papenfuss, H.D. and Gross, J.F. (1978). Analytical study of the influence of capillary pressure
drop and permeability on glomerular ultrafiltration. Microvasc. Res. 16: 59-72.
Pappenheimer, J.R. (1953). Passage of molecules through capillary walls. Physiol. Rev. 33: 337-
423.
Pappenheimer, J.R. (1970). Osmotic reflection coefficients in capillary membranes. In: Capillary
Permeability, (e. Crone and N-A. Lassen, eds.), Munksgaard, Copenhagen, pp. 278-286.
Pappenheimer, J.R. and Soto-Rivera, A. (1948). Effective osmotic pressure of the plasma proteins
and other quantities associated with the capillary circulation in the hind-limbs of cats and
dogs. Amer. J. Physiol. 152: 471-491.
Pappenheimer, J.R., Renkin, E.M., and Borrero, L.M. (1951). Filtration, diffusion and molecular
sieving through peripheral capillary membranes. A contribution to the pore theory of
capillary permeability. Amer. J. Physiol. 167: 13-46.
Poynting, J.H. (1981). Change of state: solid-liquid. Phil. Mag. 5: 32-48.
Reneau, D.D., Jr., Bruley, D.F., and Knisely, M.H. (1967). A mathematical simulation of oxygen
release, diffusion and consumption in the capillaries and tissue of the human brain. In
Chemical Engineering in Medicine and Biology, (D. Hershey, ed.), Plenum Press, New York,
pp. 135-241.
Reneau, D.D., Jr., Bruley, D.F., and Knisely, M.H. (1969). A digital simulation of transient oxygen
transport in capillary-tissue systems (cerebral gray matter) Amer. Inst. Chern. Eng. J. 15:
916-925.
Renkin, E.M. (1959). Transport of potassium-42 from blood to tissue in isolated mammalian
skeletal muscles. Amer. J. Physiol. 197: 1205-1210.
Renkin, E.M. (1977). Multiple pathways of capillary permeability. Circ. Res. 41: 735-743.
Renkin, E.M. and Zaun, B.D. (1955). Effects of adrenal hormones on capillary permeability in
perfused rat tissues. Amer. J. Physiol. 180: 498-502.
Renkin, E.M. and Curry, F.E. (1982). Endothelial permeability: pathways and modulations. Proc.
N.Y. Acad. Sci. 401: 248-259.
References 351

Salatbe, E.P. (1977). An analysis of interstitial fluid pressure in the web of the bat wing. Amer. J.
Physiol. 232: H297-H304.
Salatbe, E.P. (1980). Convection and diffusion in the extravascular space. In Mathematics of
Microcirculation Phenomena (J.F. Gross, and A. Popel, eds.), Raven Press, New York.
Salatbe, E.P. and Venkataraman, R. (1978). Role of extravascular protein in capillary-tissue fluid
exchange. Amer. J. Physiol. 234: H52-H58.
Salatbe, E.P. and An, K.N. (1976). A mathematical analysis of fluid movement across capillary
walls. Microvasc. Res. 11: 1-23.
Scholander, P.F., Hammel, H.T., Bradstreet, E.D., and Hemmingsen, A.E. (1965). Sap pressure
in vascular plants. Science 148: 339-346.
Scholander, P.F., Hargens, A.R., and Miller, S.L. (1968). Negative pressure in the interstitial fluid
of animals. Science 161: 321-328.
Scholander, P.F., Hammel, H.T., Hemmingsen, E.A., and Bradstreet, E.D. (1964). Hydrostatic
pressure and osmotic potential in leaves of mangroves and some other plants. Proc. Natl.
Acad. Sci. 52: 119-125.
Shapiro, A.H., Jaffrin, M.Y. and Weinberg, S.L. (1969) Peristaltic pumping with long wave lengths
at low Reynolds number. J. Fluid Mech. 37, 799.
Skalak, T.C., Schmid-Schoenbein, G.W., and Zweifach, G.W. (1984). New morphological evidence
for a mechanism oflymph formation in skeletal muscle. Microvasc. Res. 28: 95-112.
Starling, E.H. (1896). On the absorption of fluids from the connective tissue spaces. J. Physiol.
(London) 19: 312-326.
Staub, N.C. (ed.) (1978). Lung Water and Solute Exchange, Marcel Dekker, New York.
Staverman, A.J. (1951). The theory of measurement of osmotic pressure. Recl. 7rav. Chim.
Pays-Bas. Belg. 70: 344-352.
Stewart, G.N. (1893-1897). Researches on the circulation time in organs and on the influences
which affect it. J. Physiol. (London) 15: 1-30,15: 31-72, 15: 78-89 (1893); 22: 159-173 (1897).
Tang, H.T. and Fung, Y.C. (1975). Fluid movement in a channel with permeable walls covered
by porous media. (A moel of lung alveolar sheet). J. Appl. M echo Trans. ASME Vol. 97, Ser.
E 42(1): 45-50.
Taylor, G.I. (1953). Dispersion of soluble matter in solvent flowing slowly through a tube. Proc.
Roy. Soc. Ser. A, 219: 186-203.
Taylor, G.I. (1954). The dispersion of matter in turbulent flow through a pipe. Proc. Roy. Soc.
Ser. A, 223: 446-468.
Thomson, W. (1871). On the equilibrium of vapour at a curved surface ofliquid. Phil. Mag. (A)
42: 448-452.
Van't Hoff, J.H. (1886a). Une propriete general de la matiere diluee. Svenska Vet. Akad. Handl.
21: 17-43.
White, A., Handler, P., Smith, E., and Stetten, D., Jr. (1959). Principles of Biochemistry, Chapt.
12, 2nd ed., McGraw-Hill, New York.
Wiederhielm, C.A. (1968). Dynamics of trans capillary fluid exchange. J. Gen. Physiol. 52: 295-615.
Wiederhielm, C.A. (1969). The interstitial space and lymphatic pressures in the bat wing. In The
Pulmonary Circulation and the Interstitial Space. (A.P. Fishman and H.H. Hecht, eds.),
pp. 29-41. University of Chicago Press, Chicago.
Wiederhielm, C.A. (1972). The interstitial space. In Biomechanics: Its Foundations and Objectives.
(Y.c. Fung, N. Perrone and M. Anliker, eds.) Prentice Hall, Englewood Cliffs, N.J.
Wiederhielm, C.A. (1979). Dynamics of capillary fluid exchange: a nonlinear computer simulation.
Microvas. Res. 18: 48-82.
Wiederhielm, C.A. (1981). The tissue pressure controversy, a semantic dilemma. In Tissue Pressure
and Composition (A. Hargens, ed.), Williams & Wilkins, Baltimore, pp. 21-33.
Yin, F.C.P. and Fung, Y.c. (1971a). Comparison oftheory and experiment in peristaltic transport.
J. Fluid Mech. 47: 93-112.
Yin, F.C.P. and Fung, Y.c. (1971b). Mechanical properties of isolated mammalian ureteral
segments. Am. J. Physiol. 221: 1484-1493.
352 9 Mass Transport

Zierler, K.L. (1963). Theory of use of indicators to measure blood flow and extracellular volume
and calculations of transcapillary movement of tracers. Circ. Res. 12: 464-471.
Zupkas, P.F. and Fung, c.y. (1985). Active contractions of ureteral segments. J. Biomech. Eng.
107: 62-67.
Zweifach, B.W. and Intaglietta, M. (1968). Mechanics of fluid movement across single capillaries
in the rabbit. Microvasc. Res. 1: 83-101.
Zweifach, B.W. and Prather, J.W. (1975). Micromanipulation of pressure in terminal lymphatics
in the mesentery. Amer. J . Physiol. 228: 1326-1335.
Zweifach, B.W. and Silberberg, A. (1979). The interstitial-lymphatic flow system. In International
Review of Physiology, Cardiovascular Physiology III, Vol. 18. (A.C. Guyton and D .B. Young,
eds.) University Park Press, Baltimore.

An expression of a philosophy of education: LIKE A MISTY BREEZE OF SPRING. Painted

by Chen Chia Yu (B\i:1JQ~) in Wuchan, during the author's two-month course on
biomechanics in 1979.
CHAPTER 10

Description of Internal Deformation

and Forces

10.1 Introduction
Since living organs normally go through finite deformation, a bioengineer
should know the subject of finite deformation analysis. This subject is not
difficult, but it usually lies outside the common engineering curriculum. It is
not simple, and considerable patience is needed to master it. In the following,
a presentation of its most important aspects is given in easy to understand
physical terms. There are many books and papers on this subject (see Refer-
ences). Fung (1965) is believed to be one of the easiest to read.
The concept of strain is reviewed in Secs. 10.2 and 10.3. The concept of stress
is discussed in Sec. 10.4. The equation of motion is presented in Sec. 10.5, Work
and energy in Sec. 10.6, the use of the strain energy function in Sec. 10.7, and
complementary energy function in Sec. 10.8, all without linearization. The
separation of local rotational motion and strain in a general deformation of
a continuum is discussed in Sec. 10.9.
Of these topics, the most important are the definitions of Cauchy's,
Lagrange's, and Kirchhoff's stress tensors discussed in Sec. 10.4. They are
denoted by (Ji)' 1';) and Si), respectively. The stresses (Jij' Sij are symmetric tensors,
but 1';) is not. Then it is shown in Sec. 10.7 that if the material is elastic and
the strain energy function per unit initial volume, Po W, is expressed in terms
of the Green's strain tensor, E i), we have

S.. = a(poW) (1)

'J aE'J.. ·
On the other hand, if Po W is expressed in terms of the deformation gradient
tensor ax;!aa), then

353
354 10 Description ofInternal Deformation and Forces

o(Po W) (2)
1ji = o(oxdoaj )·
If the strain energy per unit initial volume is expressed as a function of the
Kirchhoff stress tensor Sij' then it is called the complementary strain energy
denoted by Po lv". We have

£.. = oPo lv" (3)

'1 oSij·

These important formulas are used frequently in biomechanics.

After these mathematical concepts are clarified, we return to biology. The
first question we ask is whether living tissues are elastic. The answer, unfor-
tunately, if no. But the concepts of pseudo-elasticity and the pseudo-strain
energy function provides us with a useful approximation. Applications of these
concepts are discussed in Sec. 10.8 and continued in Chap. 11.

10.2 Description of Internal Deformation

To describe the deformation of a body we need to know the position of any

point in the body with respect to an initial configuration which shall be called
the reference state. Moreover, to describe position we need a frame of refer-
ence. Let us choose a rectangular cartesian frame of reference, Fig. 10.2: 1.
Every material particle in the body at the reference state So has three coordi-
nates ai' a2 , a3 which can be written as a column matrix denoted by anyone
of the following forms:

(::) or (a"a"a,) or (a,) m a" i = 1,2,3.

A particle P has the coordinates ai' and a neighboring particle P' has the
coordinates ai + da;. When the body is deformed the particles P, P' are moved
to Q, Q' whose coordinates are Xi and Xi + dx;, respectively. The deformation
of the body is known completely if we know the relationship
Xi = x i(ai,a 2 ,a3 ) i = 1,2,3 (la)
or its inverse
ai = ai(x i ,x2 ,X3 ) i = 1,2,3 (lb)
for every point in the body. If we write
Xi = ai + Ui i = 1,2,3 (lc)
then Ui is called the displacement of the particle P.
In general, we do not have the luxury of knowing the transformation law
(or mapping function), as expressed in Eq. (la) or (lb), at the beginning of
 ' ... v
Su ,,.,
. . ---- "'"", I ;I T* o
N
, ~/
02. X2 ,~ ?(
,, '/ iii
.g.
g.
I (Xi)1fdV '~ :;
I
" \\ o
....,
1. ......
I __ _' 5 a '\ a
C1>
, - _,' I
:3
\ ---- , e..
",' I ?(
u'''' \ Fpdv 0'
dvO ---/ ~, ~"
(a;)~ ____ ~
g.
~\Sa " :;

~
/, "
,'Su
I , ..... " '
FOPOdvO
r--,..----
Deformed configuration V

Original configuration Vo 0,. x1

FIGURE 10.2:1 A body shown by solid line in the reference state is deformed into one
VOl
bounded by the dotted line. The frame ofreference for the reference state is ai' a 2 , a 3 • VI
VI
That for the deformed state is Xl' X 2 , X 3 •
356 10 Description ofInternal Deformation and Forces

solving any biological or engineering problems. More often the seeking of

such a transformation law is the final objective of research. In the process,
however, there is a general belief (axiom) that the stress at a point in a
continuum depends only on the deformation in the neighborhood of that
point. Therefore, we ask what features of the transformation describe the
deformation in the immediate neighborhood of a particle? The answer is that
the principal features are given by the partial derivatives of Eqs. (la-c).
Usually the first derivatives provide all the information we need. These deriva-
tives can be arranged in the form of matrices

OX 1 oX 1 oX 1 oa 1 oa 1 oa 1
oa 1 oa z oa 3 oX 1 oX z oX 3
oXz oXz oXz oa2 oa2 oa z
or (2a)
oa 1 oa z oa 3 OXl OX2 oX 3
oX 3 OX3 oX 3 oa 3 oa 3 oa 3
oa 1 oa z oa3 oX 1 oX z OX3
which can be written simply as
oxjoaj, oajoxj, (i,j = 1,2,3). (2b)
They are called the deformation gradient tensors.
Physically, what the deformation gradients enable us to do is to describe
how the distance between two neighboring points such as P(a;}, P'(a i + da i)
is changed after they are deformed to Q(Xi), Q'(x i + dXi). To measure distance
between two points Q, Q' we assume that the world we live in is Euclidean
and Pythagoras rule applies:
ds 2 = dxi + dx~ + dx~, (3)
where dx 1 , dx z , dX3 are the projections of QQ' on the three coordinate axes
and ds is the length QQ'.
We may write Eq. (3) in anyone of the following forms:
3
ds Z = L dXi dX i = dX i dX i = bij dX i dxj.
i=l
(3a)

Here bij is the Kronecker delta, defined as oij = 1 when i = j, Oij = 0 when i #- j.
The summation convention is used (Sec. 1.3). In the reference state, the distance
between the same two points at P, P' is ds o: Applying Eq. (3) to the line
elements PP'(da i) and QQ'(dx;}, we have
ds~ = da i da i, (4)
Hence we obtain the difference:
ds z - ds~ = dxidx i - daida i = Oij(dXidxj - daidaj). (5)
Now, from Eqs. (1) we have
10.2 Description ofInternal Deformation 357

(6)

A substitution ofEq. (6) into Eq. (5) and several changes of the dummy indices
yield the result:

(7a)

ih aa.)
= ( (j'k - ax; ax: dx , dXk' (7b)

We rewrite Eq. (7) in the form

ds 2 - dS6 == 2Eij da i daj = 2eij dX i dXj (8)
and call the coefficients Eij the Green's strain tensor and eij the Almansi's strain
tensor. Thus

E .. =
IJ
~(aXk
2 aa i aaj
aX k _ (j. ) = ~(aUiaa + aaaUji + aUkaai aUk)
IJ 2 j aa j
(Green) (9)

e .. =
IJ
~((j
2
. _aaaXki aaaXk)
IJ j
= ~(aUi + aUj
2 aXj aX i
_ aUk aUk)
aX i aXj
(Almansi). (10)

If the strains Eij or eij are known, then according to Eq. (8), we can calculate
the change of the square of the distance between any two neighboring points
in a deformed body. (Only the square of the distance can be computed so
easily.) To obtain the change in distance itself we would have to take the square
roots of the righthand sides of Eqs. (4) and form the difference. The resulting
formula for ds - ds o is inconvenient to use. The vanishing of Eij or eij implies
(according to Eq. (8» no change of distance between any two points. Hence
a rigid body motion has no strain. If the deformation is small in the sense that
the values of the derivatives auJaaj , auJaxj are infinitesimal so that their
squares can be neglected in Eqs. (9) and (10), then both Eij and eij are reduced
to

/3 .. = ~(aUi + auj ) = ~(aUi + auj) (11)

IJ 2 aXj aX i 2 aaj aai'
which are called Cauchy's infinitesimal strain tensor.
Returning to what we said at the beginning of this section, we may assert
that biomechanics problems are usually structured so as to find the stresses
and strains first, then the deformation gradient, and finally, the mapping
functions in Eq. (la-c).
The following examples illustrate the meaning of the finite strain
components:
Example 1. A block is deformed in such a way that
(12)
358 10 Description ofInternal Deformation and Forces

Then
Ell = t(AI - 1), E22 = t(A~ - 1), (13)

If the material is incompressible so that the volume of the block does not
change, then
(14a)
or
(14b)
Example 2. Consider a rubber band. Initially its length is Lo. By pulling on
it, its length is increased to L. We can define its "strain" in several ways:
(L - Lo) , (L - Lo)
6=--- 6=---
Lo L '
(LZ - L~)
Ell = 2L~ ,

(L dL L
Y= J Lo L = In Lo .
If L = 2, Lo = 1, we have 6 = 1,6' = t, Ell = ~, ell = i, Y = 0.693. Thus for
a finite deformation each definition leads to a different number. If L = 1.01,
Lo = 1.00, then 6 = 0.01, 6' == 0.01, Ell == 0.01, e u == 0.01, Y == 0.01; and all
definitions lead approximately to the same number when the deformation is
small. y is called the "true" strain by its enthusiastic supporters and is often
used in metallurgy and thermal stress in discussing creep and plasticity.
Depending on personal preference, one or the other is called "engineering"
strain. Our reasoning presented above suggests that Ell or ellis preferred in
biomechanics when dealing with finite deformation.
Example 3. The following deformation is called a pure shear
1
(15)

It preserves the volume.

Example 4 (Fig. 10.2:2). A deformation
(16)
corresponds to the following strains according to the definition, Eq. (9).
Ell = t(AI - 1), E22 = t(A~ - 1), E3 = 0,
E12 = t(A l k + mAz) = E Z1 ' E Z3 = E31 = O. (17)
The maximum shear strain is equal to the radius of the Mohr's circle (see Fung,
1977, p. 95)
10.2 Description oflnternal Deformation 359

FIGURE 10.2:2 Deformation described by Eqs. (16), (17) when Al = 1.5, A2 = 0.9,
k = 0.3, m = 0.1 is shown by the figures at the top. The corresponding Mohr's circle
is shown below. For the construction of the Mohr's circle, the normal strains are
plotted as abscissa, the shear stress is plotted as ordinate. We plot E 11 ,-E 12 to obtain
a point A, plot E 22 , E12 to obtain point B. Join A, B by a straight line. AB intersects
the horizontal axis (normal strain) at C. With C as center, draw a circle passing through
A, B. This is the Mohr's circle. The center, C, represents the mean strain in Xl' X2 plane.
The intercepts E 1, E2 yield the values of the principal strains. The radius is the
maximum shear strain. The angle (J is the angle between the Xl axis and the principal
axis associated with E l • The angle between CA and CE I is equal to 2(J. Analytically,
tan 2(J = E 12 /[E 11 - (Ell + E 22 )/2].

max shear strain = [( Ell - E 11 +

2 E)2
22 + E~2 J12
/

Example 5 (Strain in Myocardium). To measure the strain in the beating

heart Fenton et al. (1978), Waldman et aI. (1985) inserted small lead beads into
the myocardium and took x-ray cine-photographs from two different direc-
tions simultaneously to obtain the coordinates of the beads as a function of
time. Any four nonplanar beads form the vertices of a tetrahedron which has
6 edges. From the measured coordinates, they compute ds 2 , ds~, dXj, daj of each
edge, substitute into Eq. (8), and obtain 6 linear equations for the 6 unknowns
E ll ,E22 ,E33 ,E 12 = E 21>E 23 = E 32 ,E31 = E 13 . These are the average strains
360 10 Description ofInternal Deformation and Forces

in the tetrahedron. Using a standard computing program one obtains further

the principal strains and principal directions. If N beads are inserted, one can
thus obtain the strains in N - 3 nonoverlapping regions in the myocardium,
because, starting with a basic tetrahedron, every addition of a new bead defines
a new tetrahedron. Waldman et al. (1985) used this method to measure the
strains in the heart. By creating an axisymmetric infarct in the left ventricle,
the stress distribution can be computed by finite element method on the basis
of an assumed form of constitutive equation. The identification of the com-
puted and measured strains can then yield the material constants ofthe infarct
myocardium.

10.3 Use of Curvilinear Coordinates

In the preceding section we used a fixed rectangular cartesian frame of refer-
ence to describe the location of a material particle in both the reference and
deformed states. An alternative is to use curvilinear coordinates, such as
cylindrical polar coordinates for a cylinder, spherical coordinates for a sphere.
When the body is deformed, we can use a different set of curvilinear coordinate
to describe the deformed body.
When curvilinear coordinates are used, the square of the length of a
differential element must be written as
ds; = g!J) dai daj, ds 2 = %dxidxj. (1)
The quantities g!J) and gij are called metric tensors. When gij #- J ij , the geo-
metry is non-Euclidean.
A particularly advantageous choice of coordinates is to consider the frame
of reference as attached to the material particles. Then as the body deforms,
Xi remains the same as ai' but the square of the length between any two
neighboring particles become:
(2)
It follows that
(3)
If we write
(4)
then
E ij -_1.( (0»)
2 gij - gij . (5)
Thus the strain tensor is just one-half of the change of the metric tensor.
The elegance of this approach is incomparable. The coordinates so chosen
have been called convective, co-moving, dragging along, embedded, or body
coordinate system. Its use requires, however, a knowledge about the general
tensor analysis. In the rest of this book we limit ourselves to cartesian tensors.
10.4 Description ofInternal Forces 361

10.4 Description of Internal Forces

Internal forces exist in a deformed body. To describe internal forces we
imagine an arbitrary cut in the body and examine the force acting on the cut
surface (Fig. 10.4:1). The magnitude of the force must vary with the area of
the cut. We assume that a limiting value of the force per unit area of the cut
exists, and call it a stress vector or surface traction. The surface traction varies
with the orientation of the surface. To obtain a full description of the state of
stress at any point in a body we imagine an infinitesimal cube enclosing that
point (Fig. 10.4:2). On each surface of the cube there acts a stress vector. On
the surface whose normal vector coincides with the Xl axis the stress vector
has three components 0"11' 0"12' 0"13 in the directions of the Xl' X2, X3 axes
respectively, (Fig. 10.4:3). We call 0"11 a normal stress and define it to be positive
(tensile) if it points in the direction of the xl-axis. We call 0"12 a shear stress
and define it to be positive if it points in the direction of the x2-axis. Similarly,
0"13 (shear) is positive if it points in the direction of the x3-axis. The other

FIGURE 10.4:1 Stress vector f acting on a surface whose normal vector is v.

FIGURE 10.4:2 Stress vectors acting on the surfaces of a small cube with edges parallel
to coordinate axes Xl' x 2 , X3.
362 10 Description ofIntemal Deformation and Forces

~------------~----~ X,
1
FIGURE 10.4:3 The three components of the stress vector T acting on a surface whose
gormal vector points in the direcl\on ofthe positive Xl axis area O'u, 0'12' 0'13' Similarly,
T has components 0'22, 0'23, 0'21, T has components 0'33, 0'31' 0'32'

surface that is perpendicular to the xl-axis has a normal pointing to the

negative xl-axis. On that surface the stresses are 0'11' 0'12' 0'13 also, with all the
directions reversed. Similarly, on the surfaces perpendicular to the X 2 and X3
axes we have normal stresses 0'22' 0'33 and shear stresses 0'21> 0'23' 0'31' 0'32' By
conditions of equilibrium we can show that 0'12 = 0'21,0'23 = 0'32,0'31 = 0'13'
Hence 6 stress components define the stress state at a point in the body. This
was first clarified by Cauchy, hence O'ij is known as Cauchy stress. If the body
is subjected to a body force X (with components X l ,X2 ,X3 ) per unit volume
of the material in the deformed state, then it can be shown that the equation
of equilibrium of the internal forces is given by

aO'ij
aXj +
x. = 0 I • (1)

If the internal forces are not in equilibrium, then motion ensues, and according
to Newton's law, the equation of motion is
aO'ij
Prxi = -a
Xj
+ Xi' (2)

Here p is the density of the material in the deformed state, rxi is the acceleration
of the material particle.
All this is discussed in much greater detail in mechanics books, e.g., Fung
(1977). In the course of analysis, however, we must relate stresses to strains.
Hence, if strains were referred to the original position of particles in a con-
tinuum, it would be convenient to define stresses similarly with respect to the
original configuration.
Consider an element of a strained solid as shown on the right-hand side of
Fig. 10.4:4. Assume that in the original state this element has the configuration
as shown on left side of Fig. 10.4.:4. A force vector dT acts on the surface
PQRS. A corresponding force vector dTo acts on the surface PoQoRoSo. If we
10.4 Description ofInternal Forces 363

0, •• ,
FIGURE 10.4:4 The corresponding tractions in the original and deformed states of a
body.

assign a rule of correspondence between dT and dTo for every corresponding

pair of surfaces, and define stress vectors in each case as the limiting ratios
dT/dS, dTo/dSo, where dS and dSo are the areas of PQRS, PoQoRoSo, respec-
tively, then by the usual method (see Equations (15), (16) infra) we can define
stress tensors in both configurations. The assignment of a correspondence rule
between dT and dTo is arbitrary, but must be mathematically consistent.
The following is known as the Lagrangian rule:
dT~~) = dTi , (dT~L) = dT). (3)
According to this rule, the force vector is transported without change of
direction and magnitude from the surface element in the deformed configura-
tion to the same surface element in the undeformed configuration. An alterna-
tive, known as the Kirchhoff's rule, however, requires a change of magnitude
and direction of the force vector. The Kirchhoff's rule states that

dT,(K)
0,
= oa
'"
i dT
l'
(4)
uX j

so that the force vectors dT~K) and dT are related by the same rule as the
transformation of line elements
da·, = -
oa·
oX' dx.J
j
(5)

when the body is subjected to deformation. These correspondence rules are

illustrated in Fig. 10.4:5 for a two-dimensional case, in which the rectangle
represents an element in the deformed body, and the parallelogram represents
364 10 Description ofInternal Deformation and Forces

(0)
0,. x, (b)
opX,

FIGURE 10.4:5 The correspondence offorce vectors in defining (a) Lagrange's and (b)
KirchholTs stresses, illustrated in a two-dimensional case.

Stress = 0"
Force=o,1"2"3

Y' / Force=
o,'''~3
.z /
Force = 0,,"2 113
II,

l/
Stress = T" Stress = s"
(b) (c) /
FIGURE 10.4:6 The relationship of stresses defined according to (a), Cauchy, (b),
Lagrange, and (c) Kirchhoff.

the same element in the undeformed configuration. Let us illustrate these

relations by a few examples.
Example 1. Let us consider the deformation described by Eq. (10.2:12). A
unit cube becomes a parallelopiped with edge lengths ,11, ,12' ,13. A Cauchy
stress 0"11 acts on the surface perpendicular to the xl-axis in the deformed
state. Since the area of that surface is ,12,13, the force is 0"11,12,13 (Fig. 1O.4:6a).
To derive a Lagrangian stress associated with the original cube we transfer
this force to the corresponding surface of the original cube (Fig. 1O.4:6b)
without change. The area of that surface in the cube is 1, and the Lagrangian
stress, written as T11 , is
(6)
To derive a Kirchhoff stress associated with the original cube (Fig. 10.3:6c)
10.4 Description ofInternal Forces 365

Lo
I Area
.-l.-Ao
F

FIGURE 10.4:7 A bar in the original and deformed configurations.

we transfer the force O'u A2A3 to the corresponding face on the cube according
to the rule (4). The inverse deformation gradient oat/ox 1 is l/Al> thus the force
Su = O'UA2A3/A1' Since the area is 1, the Kirchhoff stress is

A2 A3 Tu
Su = O'u T = ~. (7)

If the material is incompressible, so that A1 A2A3 = 1, then

Su = O'u ;i' if A1A2 A3 = 1. (8)

Figure 10.4:7 shows a usual experimental set up in which a cylindrical test

specimen is clamped at the top and loaded by a force F at the bottom. If the
cross-sectional area is A, the Cauchy stress is F / A. If the unloaded specimen
has a length Lo and a cross-sectional area Ao, we have A1 = L/Lo and
A2A3 = A/Ao. Hence
F F F Lo
O'u =-:4' Tu =T' Su = - - . (9)
o Ao L

Example 2. Consider a deformation described by the mapping function

relating the coordinates of a point in the original position (a 1 , a 2 , a 3 ) to those
in the deformed position (X 1 ,X 2 ,X 3 ):
Xl = a1 + ka 2 , (lOa)

a1 = Xl - kx 2 , (lOb)
A unit cube in the deformed state corresponds to a parallelopiped in the initial
state, as it is shown in Fig. 10.4:8. Let the unit cube be subjected to shear
stresses 0'12 = 0'21 while all other stress components vanish. Compute the
Lagrangian and Kirchhoff stresses.
Solution. Consider first the top surfaces ACDA and A'CD'A'. They have
the same orientation and same area, hence under either Lagrangian rule or
366 10 Description ofInternal Deformation and Forces

Kirchhoff rule, the force acting on the surface A' C' D' A' consists of a horizontal
force 0"21. Therefore, we have
T22 = 0, S22 = 0, (11)
Next consider the wedge formed by the inclined surface A' B' and the
surfaces OA', OB' normal to the axes ai' a2 respectively. Figure 10.4:8 (b) shows
the Lagrangian case, with the stress vectors and the areas indicated. In view
of T21 = 0"21 (Eq. (11)), the conditions of equilibrium in the vertical and
horizontal directions respectively yield
(12)
Hence in this case T12 = T21 and Tl1 = -k0"21.

c'~---< CF----:-f

L..-----!L--- x,
Undeformed state Deformed state
a

b c

FIGURE 10.4:8 (a) The corresponding shapes of a body in the undeformed and de-
formed states described by Eq. (10). The Cauchy shear stress 0"21 = 0"12 acts on the
body in the deformed state which is a unit cube. The forces on the surfaces AB and
AC are 0"12 and 0"21' resp. since the area = 1. These forces are transported to the surfaces
A'B', A'e' in the undeformed state according to Lagrange's rule. (b) Forces acting on
a wedge A'B'D in the undeformed state. (c) The force that acts on the surface A'B' in
the undeformed state obtained from the Cauchy stress 0"12 according to Kirchhoff's
rule.
10.4 Description ofInternal Forces 367

Similarly, for the Kirchhoff's case, consider the wedge shown in Fig.
10.4:8(c). According to Kirchhoff's rule Eq. (4), the force acting on the inclined
surface A' B' has the following components:

· I ~0"12
h onzonta: Gal = -
k 0"12'
UX 2

Hence the conditions of equilibrium yield

(13)

But, according to Eq. (11) S21 = 0"21' and 0"21 = 0"12' Hence

Example 3. Consider the same deformation as in Example 2, but now let

°
there be, in the deformed state, a normal stress 0"22 -# in addition to 0"12 =
0"21' Then according to the Lagrangian rule, there acts a force with a vertical
component 0"22 and a horizontal component 0"21 on the top surface A'C'D'A'.
Thus, in this case, Eqs. (11) are replaced by
T22 = 0"22'

and Eqs. (12) are replaced by

T12 + kT22 - 0"12 = 0,
Hence

Thus, in this case T12 -# T 21 , i.e. the Lagrangian stress tensor is asymmetric.
For the Kirchhoff stresses observe that according to the transformation
rule given by Eq. (4), on the top surface A'C'D'A' the vertical force 0"22 in the
deformed state is transformed into a force with the following components:
. I Ga 2 · I ~0"22
Gal
vertIca: ~0"22 = 0"22; honzonta: = - k 0"22'
UX 2 UX 2

The horizontal force 0"21 acting on the surface ACDA in the deformed state is
transformed into a force acting on the surface A' C'D' A' in the undeformed
state with the following components:

vertIca: Ga2
. I ~0"21
UX I
= °h
,
· I ~0"21
onzonta: Gal
UX I
= 0"21'

Hence, on the surface A'C'D'A', we have the vertical and horizontal stresses:

Next, consider the wedge shown in Fig. 1O.4:8(c), which remains applicable.
The condition of equilibrium yields, in the horizontal direction;
Su + kS21 + kO"12 = 0,
368 10 Description ofInternal Deformation and Forces

or
Su = -k(a21 - k(22 ) - ka 12 = -2ka12 + k2a22'
In the vertical direction:
S12 + kS22 - a 12 = 0, i.e. S12 = a12 - ka22'
= S21' i.e. the Kirchhoff stress tensor remains symmetric.
It is seen that S12
These results can be obtained rapidly by substitution into the general
formulas given below, see Eqs. (17), (18).
Example 4. The general case (Fig. 10.4:4). The vector dT denotes a force
acting on a surface element in the deformed body with a unit outer normal v:
whereas dT 0 denotes the corresponding force assigned to the corresponding
surface in the original configuration with area dSo and unit outer normal Vo'
If aij is the stress tensor that refers to the strained state, we have Cauchy's
relation:
(14)
We now define stress components that refer to the undeformed state by a law
similar to Eq. (14). IfEq. (3) is used, we write
dTJ~) = 1]ivOjdSO = dT; (Lagrange). (15)
If Eq. (4) is used, we write
(K) _ _ oa i
dToi - SjivOjdSO - a;:d'Fa (Kirchhoff). (16)
a

These equations define the stresses aij , T;j' and Sij' which are called the Cauchy,
Lagrangian, and Kirchhoff's stress tensors, respectively.
The following relationships between aij' T;j' and Sij are given in standard
textbooks:

(17)

S _ Pooa oa i j
P oXa oXp apa·
ji - (18)

The Eulerian stress tensor rij is symmetric. The Lagrangian stress tensor 1]i is
not necessarily symmetric. The Kirchhoff's stress tensor Sji is symmetric. The
Lagrangian tensor is inconvenient to use in a stress-strain law in which the
strain tensor is always symmetric; the Kirchhoff stress tensor is more suitable.
From Eq. (18), we have

(19)

(20)
10.5 Equation of Motion in Lagrangian Description 369

(21)

Example 5. The calculation can be speeded up if one recognizes that oajoxa

is the element in the ith row and IXth column of the square matrix {oajoxp};
(Jij is the element of the ith row and jth column of the square matrix {(Jij}.
Thus, for the calculation of Example 3, we note that

p/Po is equal to the determinant of the {oa)ox m } matrix which is equal to l.

Hence according to Eq. (17), Tl1 is equal to the scalar product of the 1st row
of {oa)ox m } and the 1st column of {(Jij}:
oa 1 oar oa 1
Tl1 = -;--(J11 + -;--(J21 + -;--(J31 = (J11 - k(J21 + O.
UX 1 UX 2 UX 3

Similarly,

and so on.

to.5 Equation of Motion in Lagrangian Description

Consider a continuum occupying a region V with a boundary surface S in the

deformed state, which corresponds to a region Vo with a boundary surface So
in the original state (Fig. 10.5:1). The body is subjected to external loads. In
Eulerian description the external loads consist of a body force F per unit mass,

FIGURE 10.5:1 Notations.

370 10 Description of Internal Deformation and Forces

and a surface traction t per unit area acting on a surface element dS whose
unit outer normal vector is v. In Lagrangian description, we shall write the
body force per unit mass as F o' The original density Po corresponds to the
density P in the deformed state. We shall specify that
(1)
We shall consider static equilibrium of the body.* The resulting body force
acting on the region V or Vo is

r F;p dv = Jvr Fo,Po dvo.

Jv o
(2)

The result of the traction acting on the surface S or So is

f s
T;ds = f s
(Jjivjds. (3)

According to Eq. (10.4:3), this is equal to Jso

1}i VOj dSo, which can be trans-
formed by Gauss's theorem into Jvo (o1}joaj) dv o. Hence

f s
v
I;dS = 1oT.
Vo
~dvo.
uaj
(4)

The condition of equilibrium is that the sum of the body forces and surface
tractions vanishes. Hence, by Eqs. (2) and (4), we have, at equilibrium,

r (PoFo, + OT.)
Jvo o~; dv o = O. (5)

Since this equation must be valid for an arbitrary region Vo, the integrand
must vanish, Hence, we obtain the equation of equilibrium
01}i
--;-
ua + PoFo. = O. (6)
j I

Further, using Eq. (10.4:2), we obtain the equation of equilibrium expressed

in terms of the Kirchhoff stress tensor:
o(
oaj Sjk oa~
(]x.) + PoFo, = O. (7)

Finally, by an analysis similar to Eqs. (2)-(6) but staying in the deformed

configuration, we obtain the well-known equation of equilibrium expressed
in terms of Cauchy stress:

(8)

* If the body is in motion, we can apply D'Alembert's principle to reduce an equation of motion
to an equation of equilibrium. We apply the inertial force, which is the product of the mass and
the negative of the acceleration, as an external load. Hence F, and Fo, include the inertial force.
10.6 Work and Strain Energy 371

10.6 Work and Strain Energy

Force multiplied by the distance it traveled in the same direction is the work
done by the force. Force multiplied by the velocity of its travel (scalar product)
is the power of the force. Consider a cylindrical test specimen (of muscle, or
blood vessel, or bone) clamped at the top and loaded by a force at the bottom
(Fig. 10.6:1). In the deformed state the length of the specimen is L, the
cross-sectional area is A, the longitudinal stress is u, so the force is uA. If the
length is extended by a small amount bL then the work done is uAbL. Let us
express this quantity in terms of dimensions and stresses in the original
configuration of the specimen, which has a length of L o, cross-sectional area
A o, Lagrangian stress T, Kirchhoff stress S. According to Sec. 10.5 we have

uA
TAo = uA, SAo=T· (1)

Hence the work done by the force per unit original volume of the specimen is

Work done by force uAbL ASAoLobA TAoLobA

(2)
~~ ~~ ~~ ~~
Let us denote the work done by the force per unit mass of the specimen by
the symbol bW, and the mass per unit volume in the original state, i.e. the
initial density, by Po. Then bpo W represents the work done per unit initial
volume. The left-hand side of Eq. (2) can be written as bpo W. Note further
that Green's strain Ell = !(A 2 - 1) implies AbA = bEll. Thus Eq. (2) becomes

(3)

Lo

L
Area Ao

I Cross sectional
Lagrangian Force=TAo=aA

Kirchhoff Force SAo = = a: .JT-

l ...
J....:Area=A

J Force=aA
FIGURE 10.6:1 Illustration of the principle embodied in Eqs. (3) and (4).
372 10 Description ofInternal Deformation and Forces

from which we obtain:*

(4)

but (J does not have a correspondingly simple expression.

The work done (jpo W could be dissipated into heat, or consumed by
chemical reaction, or stored in the system ready to be recalled. The last
eventuality is very important. Most engineering materials and biological
tissues can do that. To describe this eventuality, people invent a name: by
calling the material elastic. Mathematically, we assume the existence of a
quantity Po W which is a function of the strain, and for which equations like
(4) are valid. The leads to the following formal definition:
A material is said to be elastic, if it possesses a strain energy function per
unit mass W which is an analytic function of the strain components measured
with respect to the stress-free state, with the property that the rate of change
of the strain energy per unit mass is equal to the power of the stresses.
Expressing this definition in terms of Eulerian variables, we have
D 1
-W=-(J··V·· (5)
Dt p!J '1'

whereas in terms of Lagrangian variables, we have

DID 1 a
- W= -S··-E ..
Dt Po!J Dt !J
= -S··-£.·
Po'l at 'l'
(6)

In these formulas, (Jij is the Eulerian stress tensor, Eij is the Green strain tensor,
and Sij is the Kirchhoff stress tensor: Po and p are the density in the original
and deformed states, respectively. V;j is the rate of deformation defined by

V;j = 2:
1 (av. av.)
ax: + a~ , (7)

where Vi is the velocity field.

In Eq. (5), the material derivative DWjDt means the rate of change of W
associated with a specific group of material particles with respect to time. On
the right-hand side, (Jij V;j is the sum of the products of stresses and the
corresponding strain rates, which is the power of the stresses per unit volume.
Dividing (Jij V;j by the material density P results in power of the stresses per
unit mass. Thus Eq. (5) describes the definition of elasticity as stated.
We should like to prove that Eqs. (6) and (5) are equivalent, thus they
describe the same definition. The proof is as follows: Begin with Eq. (5) of

* These are important formulas. On the next 4 pages a generalization of these formulas to the
most general deformation of an elastic body is given, resulting in Eqs. (1), (2) of Sec. 10.7. The
remark that in an incompressible material the pressure is an arbitrary quantity (Eq. (12) of Sec.
10.7), is also important.
10.6 Work and Strain Energy 373

Sec. 10.2. Taking material derivatives of terms on both sides of the equation,
we obtain

D
-(ds D
2 - ds 2 )= -(dx.dx.(j
o Dt .. - da.da.(j .. ) = 2 (DEii)
- Dt da·da·. (8)
Dt 1 J IJ I ) IJ I)

Now,
D aV' D
-dx· = dv· = _" dX 1 -da· =0. (9)
Dt 1 1 aXI ' Dt 1

Hence the middle terms in Eq. (8) is

= 2l'il dx i dx l

aXi aXI
= 2l'iI-
a -a dakda m• (10)
ak am
A substitution back into Eq. (8) and changing the dummy indices several times
yields the desired relation:

~E .. = V. axp aX q (11)
Dt IJ pq aa i aaj'

Equation (11) shows that DEij/Dt vanishes when l'ij = 0, as one should expect
for a rigid-body motion. We now have, according to Eq. (10.4:20) and Eq. (11),
1 1 aXi aXj 1 D
-Uijl'ij = - - a -a SPIXl'ij = -SPIX-EplX' (12)
P Po a lX ap Po Dt
Thus, the equivalence of Eqs. (5) and (6) is proved.
The last equality in Eq. (6) follows the definition of material derivative. Eij
is a function of the initial coordinates of the particles, Eij = Eij (ai' a 2, a 3, t).
Since the labeling of each particle does not change with time, we have
D
i a
-D E i al,a2,a3,t) = -a Eij(al,a2,a3,t). (13)
t t·

This is to be contrasted with the material derivative of a function F(x l , X2, x 3 ,

t) of the current coordinates Xl' X2, X3 and time. Since Xi = xi(a l ,a 2,a 3), F is
an implicit function of ai' a 2, a 3. Hence, by definition,

aF aF
= -a + Vi-a . Q.E.D
t Xi
374 10 Description oflnternal Deformation and Forces

10.7 Calculation of Stresses from the Strain Energy Function

In this section we shall describe some salient properties of the strain energy
function Po W that characterize an elastic solid. We assume that Po W, as a
function of the nine components of a symmetric strain tensor, is written in a
form which is symmetric with respect to the symmetric strain components Ei}
and Eji , and that in forming the derivative of Po W with respect to a typical
component E ij , the component Eji will be treated as independent of E ij . With
this convention, we shall prove that the Kirchhoff's stress tensor can be
expressed as

S.. = 8(poW) (1)

'J 8Ei}'
where Po is the uniform mass density at the natural state, and Eij are the
components of Green's strain tensor. On the other hand, if W is considered
as a function of the components of the deformation gradient tensor 8x;/8aj =
bij + 8u;/8aj [see Eqs. (10.2:1) and (10.2:2)],* then we shall show that
Lagrange's stress tensor can be expressed as

T,. = 8( Po W) = 8(po W)
(2)
J' 8(8x;/8aj ) 8(8u;/8aj ) •
To prove Eq. (1), we note that by the convention stated above we have
8W 8W
--=--, (3)
8Eij 8Eji
and
.E.- W = 8W DEi} (4)
Dt 8Eij Dt .
A comparison of Eq. (4) with Eq. (10.6:6) shows that

(~Sij
Po
- 8W) DEi}
8Eij Dt
= O. (5)

Both factors of this equation are symmetric with respect to i,j. Since Eq. (5)
is valid for arbitrary values of the rate of strain DEi}/Dt, we must have
1 8W
-Sij--=O, (6)
Po 8Eij
which is Eq. (1).

* We use Xi to describe the position of a particle in the deformed configuration of the body, and
ai for that in the original, undeformed configuration. Xi is also used as the geometric variable in
Eulerian description of the deformed by body, and ai for that in Lagrangian description of the
undeformed body. Ui is the displacement vector. Xi' ai' u i refer to the same rectangular Cartesian
frame of reference.
10.7 Calculation of Stresses from the Strain Energy Function 375

If all the components of the rate-of-strain tensor DEij/Dt are not allowed to
vary independently, we cannot claim the valididity of Eq. (6) on the basis of Eq.
(5). For example, if we insist that the material is incompressible, then the
divergence of velocity field vanishes: OV;/OXi = l'ii = O. This can be expressed
in terms of DEij/Dt as follows. Multiply (10.6:11) by (oa;/ox,)(oaj/ox s ), note
that

(7)

and we have
v. = DEij oai oaj (8)
's Dt ox, oX s '
Hence the condition of incompressibility implies that

(9)

Writing

(10)

we have

B.. DEij = 0 (11)

IJ Dt .

Bij is called Finger's strain tensor. Since DEij/Dt are restricted by Eq. (11), we
cannot say that the first factor in Eq. (5) must vanish, but only that it must
be proportional to Blj . Therefore, for an incompressible elastic material, we
must have

(12)

where p is an arbitrary scalar which can be identified with pressure.

To derive Eq. (2), we return to the general compressible material and regard
Was a function of the nine components of the tensor ox;/oaj. Let us write Xi,j
for ox;/oaj. Then

(13)

From Eq. (10.2:9) we have

OEkl _ ~(OXi ~ oXi ~ ) (14)

- uk'+ UI'
OXi,j 2 oa l J oak J •
376 10 Description oflnternal Deformation and Forces

Hence

OPo W = ~ (OPo W OXi + OPo W OXi)

OXi,j 2 OEjl oal oEkj oak
0POWOX i OXi
= oEjk oak = Sjk oak = 1ji' (15)

The last equality is obtained in accordance with Eq. (10.4:21). Thus Eq. (2) is
verified.
If one likes to consider W as a function of the derivatives of the displace-
ments ui .j = oujoaj , we note that uij = Xi,j - c5ij , where c5ij is the Kronecker
delta. From Eq. (9) of Sec. 10.2 we obtain the following equation, which is
equivalent to Eq. (14):

OEkl
-:1- = -21 ( c5ki c5lj + c5li c5kj + ~c5kj
OUi OUi)
+ -;-c5lj . (16)
U~j ~ U~

Hence, in analog with Eq. (13) we obtain

OPo W _ oPo W
-",-- - ~E
(15ik + 0OUi) -_ OXi_
Sjko -1ji'
uUi,j U jk ak ak
which is the last part of Eq. (2).
Known strain energy functions of biological tissues are presented in Fung
(1981).

10.8 Complementary Energy Function

If the constitutive equation

su.. =OPoW
--
oEij
(1)

can be inverted; i.e. if Eij can be expressed as functions of Su. S12' ... , then
there exists a function Po l¥" which is a function of the stress components Sij'
and is called the complementary energy function of the material, with the
property that
(2)
and

E .. = oPo l¥" (3)

u oSij'

This theorem can be proved very easily, because if Eij are continuous and
differentiable functions of stresses then Po W is an implicit function of the
stresses Sll' S12'"'' S33.lience by differentiating Eq. (2) with respect to Sij'
10.9 Rotation and Strain 377

we obtain
iJpo w., iJE kl iJpo W iJEkl
- - = E .. +Skl - - - - - - -
iJSij !J iJSij iJEkl iJSij
which reduces to Eq. (3) because the sum of the last two terms vanish on
account of Eq. (1). Q.E.D.
The "If" at the beginning of the first sentence of this section is a big "If".
Most nonlinear stress-strain relationships cannot be inverted. For example,
a simple relation such as
Sij = Eij + EikEkj + EikEkmEmj
cannot be inverted explicitly. Fortunately, however, for many biological soft
tissues this inversion can be done, and their complementary energy functions
are known, see Fung (1979), and Biomechanics (Fung, 1981, pp. 253-256).

10.9 Rotation and Strain

The deformation gradient tensor given in Eq. (10.2:2) in the form of a matrix,
can be resolved into the product of two matrices with special properties. Let
F be the deformation gradient matrix of a reversible transformation (10.2:1),

F = G:;). (1)

The polar decomposition theorem of Cauchy yields two unique expressions for
F in terms of an orthogonal tensor R and positive symmetric tensors U and V:
F = RU = VR. (2)
R is orthogonal but need not be proper-orthogonal, so that the product of R
and its transpose R T has the value 1, and the value of the determinant of R is
either + 1 or - 1 throughout the domain Xi and time t:
RRT = I, det IR I = + 1 or - 1. (3)
Thus
detU = detV = IdetFI. (4)

R is called the rotation tensor. U and V are called the right and left stretch
tensors, respectively. Obviously,
V = RUR T • (5)
The right and left Cauchy-Green tensors C and B are defined as follows:
C == U 2 = FTF
B == V2 = FFT = RCR T. (6)
378 10 Description ofInternal Deformation and Forces

Books on continuum mechanics usually contain illustrations of the meaning

of the rotation tensor R and the stretch tensors C and B. Note that the Green's
strain tensor defined in Eq. (10.2:9) and (10.3:5) is related to C simply by
2E=C-I. (7)
The expression RU means a stretching following rotation. The expression VR
means a rotation following stretching. The processes of stretching and rota-
tion decomposed this way are non-commutative.
Chen (1988) established a new decomposition theorem:
F=(I+Y)+X (8)
in which I + Y is an orthogonal tensor
(I + Y)(I + Y)T = I
detlI + YI = 1 (9)
and X is a symmetric tensor
XT=X. (10)
Chen decomposes the deformation gradient tensor into the sum of a rotation
and a strain. The processes are then commutative. Chen went on to show that
his strain X defines normal strains in terms of change of length of linear
elements, unlike those of Green's tensor which is based on the change of the
square of length of linear elements. Chen's shear strains also have simpler
interpretation than Green's shear strains. This is a significant new develop-
ment which will have important applications to biomechanics especially to
cases in which large local rotation is involved.

Problems

10.1 Consider a rectangular block of compressible material of unit thickness that

does not change. The cross section changes from a cube to a paralleloid, with
the following coordinates for their comers:

(0, 0, 0) -+ (0, 0, 0)
(0,1,0) -+ (1,2,0)

(1,0,0) -+ (1.5, 0, 0)
(1, 1,0) -+ (2.5,2,0).

What are the Lagrangian and Green's strain components. Using matrix algebra,
determine the principal strains, stretch ratios, and the rotation.
10.2 The block named in the problem above is subject to the following stress in the
deformed configuration:
Problems 379

(a,,) { : :)x to' Njm'

What are the components of the Lagrangian and Kirchhoff stress tensors? N.B.:
Consider a unit square in the deformed configuration. Use the mathematical
transformation determined in Prob. 1 to find the shape of that element in the
initial configuration. Then apply the Lagrangian and Kirchhoff method to
determine the stresses.
10.3 Sketch the deformed shape of a cube defined by the following transformation.
(a) Xl = a l + a l a2
x 2 =a 2 +a l a2
X3 =a3
(b) Xl = a l + hin(a l + 2a 2)
X 2 = a2 + tsin(a l + 2a 2)
X3 = a 3

10.4 Design an instrument that will enable you to measure the tension in the skin of
a man in vivo. The instrument must be safe and tolerable to the subject being
measured, and convenient to use by the operator. Discuss the special design
considerations if the instrument is to be used by a beautician for cosma tic
applications or by a surgeon in various skin cancer operations.
10.5 In an operation for hernia in the abdomen, it may be desirable to prestretch the
skin so that at the end of the surgery the tension of the skin will remain normal.
Too large a tension may make the operation difficult and may hinder healing.
The prestretch may be accomplished by pumping air into the abdomen, so that
the skin area may increase by the process of creep. Design an instrument that
will measure the creep and the skin tension so that the operation can be planned
scientifically.
10.6 The tension in a tout string can be measured by anchoring the string at two
points, putting a lateral load perpendicular to the string, and measuring the
deflection of the string as a function of the load. From this premis, design an
instrument which will enable you to measure the tension in the Achilles tendon
when you exercise.
10.7 A student broke one of his fingers when playing basket ball. A tendon in that
finger broke. As a result he cannot straighten the finger. A surgeon can repair
this trouble by taking a spare tendon from his foot and using it to reconnect the
ends of the broken tendon. The problem is to decide how long the segment of
the new tendon must be for him to tie and suture it to the old ends. Or, how
much tension should the tendon in the finger have at the time of suturing. If the
tension is too large, the finger will be bent one way. If this is too small, it will be
bent the other way. How much is just right? To settle this question, a bioengineer
was called in to measure the tension needed for the correct positioning of the
finger, and to enable the surgeon to put in the correct tension at the time of
surgery. You be the engineer. You must invent an instrument and a procedure
to do it. Giving due consideration to the conditions of the operation room,
present a design for this instrument.
380 10 Description of Internal Deformation and Forces

10.8 Blow Out of a Spherical Bailon. Let the strain energy function of a material be

I/J = ~(A~ + A~ + A~ - 3), (1)

where AI' A2, A3 are the principal extensions and Il and k are material constants.
Let the initial (zero-pressure) thickness of the balloon be ho, the initial radius be
Ro. When the circumferential extension is A. Show that the inflation pressure p
is given by
(2)

Demonstrate the possibility of blow out for a rubber balloon for which k = 2.
Show that a ventricle will not blowout, (the value of k for myocardium is about
18).
Solution. Assuming plane stress and incompressibility, the stress strain rela-
tionship of a membrane under equibiaxial tension is
(J = Il(A k - r 2k ), (3)
where (J is Eulerian stress and Ais the surface extension, A= RIRo. For thin-walled
sphere, the equation of equilibrium is p = 2(JhIR, and the conservation of wall
volume requires 4nR;ho = 4nR 2 h. A combination of these relations yields Eq.
(2). (Ref. Ogden, R.W. Large deformation isotropic elasticity-on the correlation
of theory and experiment for incompressible rubberlike solids, Proc. Roy. Soc.
London, A., 326:565-584,1972.)
10.9 Discuss the stability of various states of inflation ofa spherical balloon from the
point of view of stationary total potential energy of the system. (Ref. Alexander,
H. Tensile instability of initially spherical balloons. Int. J. Engineering Science,
9:151-162,1971.)
10.10 Show that the pressure p as a function of A given by Eq. (2) has a relative
maximum if -~ < k < 3. Outside of this range of k a local pressure maximum
does not exist. Hence a balloon made of a material which may be described
approximately by a strain energy function given in Eq. (1) with k » 3 (such as
a ventricle or a urinary bladder, or an artery, or a piece of skin) will not blow
out. (Ref. Ogden, ibid 1972).
10.11 Experimental Strain Analysis. Experimental methods to reveal stress and strain
in a body can be based on a wide variety of physical and chemical phenomena
that are affected by stress and strain. Some examples are listed below. Expand
the list.
1) Piezoelectricity:
Electric resistance of thin wires of metals change
with strain - bone
Semiconductors - bone
Polymer plastics - soft tissue
Mercury enclosed in rubber tubing - plethysmograph
2) Optical properties, birefriengence:
Photoelasticity - bone model
Frozen stress in plastics - bone model
Polymer fluids - blood, body
fluids
References 381

3) Thermoelasticity:
Generation of heat by strain and blood flow - cancer
Revelation of temperature by color - cancer
4) Morphological changes
X ray of markers - heart, lung
Radioopaque filler - blood vessels
5) Variation of electric property with displacement:
Capacitance, inductance - plethysmograph
6) Acoustic phenomena
Ultrasound - blood vessel
Doppler - blood flow
7) Electromagnetic flow meters - blood flow
8) Chemical sensors (02, glucose, etc.)
Immobilized enzyme - diabetes
9) Nuclear magnetic resonance, metabolic imaging - seeing proton
contents, H,
H 2 0, P, etc.
10) CA TT scan
11) Antibody marker - cell membrane
12) Fluorescence - cell division,
nerve

References

Biot, M.A. (1965). Mechanics of Incremental Deformations. (Theory of elasticity and viscoelasticity
of initially stressed solids and fluids, including thermodynamic foundations and applications
to finite strain). Wiley, New York.
Chen, Zhi-da (1986). Rational Mechanics, (in Chinese). China Institute of Mining and Technology
Press, Xuzhou, Jiangsu, China.
Eringen, A.C. (1967). Mechanics of Continua. Wiley, New York.
Fenton, T.R., Cherry, J.M., Klassen, G.A. (1978). Transmural myocardial deformation in the
canine left ventricular wall. Am. J. Physiol. 235: H523-H530.
Fung, Y.C. (1965). Foundations of Solid Mechanics. Prentice-Hall, Englewood Cliffs, N.J.
Fung, Y.c. (1977). A First Course in Continuum Mechanics. 2nd ed., Prentice-Hall, Englewood
Cliffs, New Jersey.
Fung, Y.c. (1979). Inversion of a class of nonlinear stress-strain relationships of biological soft
tissues. J. Biomech. Eng. 101: 23-27.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living Tissues, Springer-Verlag, New
York.
Green, A.E. and Zerna, W. (1954). Theoretical Elasticity. Oxford University Press, London.
Green, A.E. and Adkins, J.E. (1960). Large Elastic Deformations and Nonlinear Continuum
Mechanics. Oxford University Press, London.
Malvern, L.E. (1969). Introduction to the Mechanics of a Continuous Medium. Prentice-Hall, N.J.
Waldman, L.K., Fung, Y.c., and Covell, J.W. (1985). Transmural myocardial deformation in the
canine left ventricle. Circ. Res. 57: 152-163.
CHAPTER 11

Stress, Strain, and Stability of Organs

11.1 Introduction

In this chapter we discuss the stress and strain in organs. The significance of
the subject is established, the methods of approach are summarized, highlights
on a few organs are surveyed, and then we focus on some topics which are
important from the point of view of mechanics, such as the zero-stress state,
the connection between micro- and macromechanics, the effect of surface
tension, the incremental laws, the interaction between neighboring organs, the
stability of some structures, and the behavior of some structures that become
unstable and collapsed.
Why do we want to know the stress and strain in an organ? The primary
reason is to gain an understanding of the function of the organ. Consider the
lung. To analyze pulmonary gas exchange we must know the size and shape
of the alveoli. The size and shape of the alveoli are related to the strain of the
lung parenchyma. The gas exchange depends on blood flow, which in turn
depends on the diameter of the blood vessels. Pulmonary blood vessels are
embedded in an elastic parenchyma; the parenchymal stress affects the caliber
of the pulmonary blood vessels. Stretching the alvelar wall narrows the
capillary blood vessels, increasing the resistance to blood flow, affecting the
ventilation/perfusion ratio, and modifying the interstitial fluid pressure and
filtration of water from pulmonary capillaries to the interstitium, lymphatics,
and aveoli.
Consider the heart, which is a muscle. The contraction of a heart muscle
cell depends on its length at the time of stimulation. If the length is fixed, the
maximum force developed is a function of the muscle length. If the force of
contraction is fixed at any time during the course of contraction, the velocity

382
11.1 Introduction 383

of shortening is a function of the muscle length. What determines the length

of the muscle cells in the heart? It is the stress and strain in the diastolic heart.
Hence, in order to determine the performance of the heart we must know the
stress and strain in the diastolic heart. In systole, the pressure of the blood in
the left ventricular chamber is determined by the tension in the muscle cells.
Hence the stresses in the myocardium determines the performance ofthe heart.
Similar considerations would show the importance of stress and strain in
the smooth muscles in internal organs. This is evident in peristalsis of the
intestines, stomach, ureter, Fallopian tubes, arterioles, venules, and sphincters
of all kinds. Moreover, the systemic blood pressure is controlled by the
vascular smooth muscle in arterioles (Ch. 5), and the smooth muscle action is
controlled by stress and strain.
The autoregulation of the blood flow to vital organs such as the kidney
and brain is based on the mechanical properties of vascular smooth muscles.
The vascular smooth muscle of the brain can react to change of strain and
stress in such a way that a ± 25 to ± 50 mmHg change in the mean arterial
blood pressure would not change the rate of cerebral blood flow, neither would
a ± 25 mmHg change in mean intracranial pressure affect the cerebral blood
flow. The renal blood vessels have similar property. They react to changing
stress and strain.
The importance of stress and strain in the musculoskeletal system is of
course well known. When certain limits of stress and strain are exceeded, a
tendon may be painfully sprained, a ligament may be broken, a bone may be
fractured. In repair and surgery, success comes to the surgeon who under-
stands biomechanics.
We know also that overstress or understress of a tissue may cause resorp-
tion, and change in stress level causes structural remodeling of an organ
(Ch. 13). Hence stress is related to the safety, growth, and change in the organs.

Method of Approach
Analysis of stress and strain in organs requires knowledge of the anatomy of
the structure, the mechanical properties of the materials, the forces that act in
the organ, and the boundary conditions imposed on the organ. The method
of continuum mechanics can be applied to living tissues at various levels of
scale. For example, in considering the heart, one may limit one's attention to
the phenomena at the scale of the whole heart. But if one also wants to know
what is going on inside a single muscle cell, then the length scale of a single
sarcomere is more pertinent. Further, in studying a sarcomere, one can ignore
details that take place in atomic dimensions, and so on. Similarly, in con-
sidering the mechanics of the whole lung, the stresses and strains in lung
parenchyma can be evaluated in regions which are large compared with an
individual alveolus; whereas in considering an interalveolar septum as a
continuum, the collagen fibers do not need to be individually accounted for.
384 11 Stress, Strain, and Stability of Organs

At each level of consideration, the property of the continuum can be described

by a constitutive equation; and each constitutive equation is written for a
specific range of length scale.
At each chosen level, we choose an appropriate set of variables to describe
the deformation and motion of an organ, which must satisfy the laws of
conservation of mass, momentum, and energy, constitutive equations, and
boundary conditions. The solution of the field equations yields stress and
strain at the level of scale chosen for the investigation. On the other hand, a
study of the relationship between different levels of scales (e.g., a derivation of
the constitutive equation on the basis of micromechanics) would provide a
deeper understanding of physiology and mechanics.

11.2 The Zero-Stress State

Any analysis of the stress and strain in an organ must begin with the zero-stress
state of that organ. This state must be determind experimentally. For example,
take a blood vessel. Cut a thin slice out with two plane cross-sections per-
pendicular to the axis of the vessel. We obtain a ring. If we cut the ring radially
at one place, the ring opens up suddenly into a sector, as shown in Fig. 11.2: 1.
If we characterize the sector by its opening angle, the angle extended by two
radii joining the midpoint of the inner wall to the tips of the ends, we find that
the opening angle of the blood vessel can differ significantly from zero. The
section opens up because certain residual strains are released.
If a body is in a zero-stress state, then any cut will not cause any deforma-
tion. One may ask whether the cut arterial ring, which has zero resultant shear

Outside

LU
,I
I

Posterior.
I
I
"

1 \
posterior / 1"---"
e \
Posterior 1 \
Outside 1 \
Anterior
IOPENING \
/ ANGLE \
/
1

FIGURE 11.2: 1 Cross section of a cut vessel at zero-stress, defining the opening angle.
From Fung and Liu (1989).
 N
...,
0-
<1>
Inside Posterior Outside Anterior N
<1>
Before Cut ....
o

p
o
~
til
OJ
I ";: 0
0 () "
d
~ r/J~c 0 0
A a
a b
~.::~!:~.

::'
d'''' Jb al. ),b d('\b a<Jc
~",d"
C
d ~ c ~"'/'
c d

5 mm
FIGURE 11.2:2 Test of the hypothesis that one cut is sufficient to reduce an arterial segment at no load to the zero-stress state. See text. From
Fung and Liu (1989).
w
00
VI
386 11 Stress, Strain, and Stability of Organs

and zero bending moment, has approached the zero-stress state. To test the
hypothesis that it has, one must cut the sector further into many pieces, and
show that the pieces can be fitted together, without strain, into the original
sector. This is obviously a difficult experiment to perform; but a somewhat
reassuring example is shown in Fig. 11.2:2, which shows four consecutive
segments of a thoracic aorta of a rat, each 1 mm thick. These four rings were
cut at four different radial sections named "inside", "anterior", "outside",
"posterior", at approximately 90° apart. The four sectors look different, im-
plying that the aorta is not axisymmetric: its property varying circumferen-
tially. In the lower row, four pieces cut from one of these sectors are shown;
and the four pieces, named a, b, c, d, were reassembled in four different ways:
abcd, bcda, cdab, dabc, beginning with a, or b, or c, or d in turn. The center
line tangents of successive pieces were matched and lined up at the places of
joining. A close resemblance of the reassembled configurations to the original
sectors is seen. This indicates that the cut ring is a good approximation to the
zero-stress state. In Sec. 11.4 we shall show an example of the heart in which
this question is examined in greater detail.

Variation of Zero-Stress State with Location and Sudden Change of

Blood Pressure
The opening angle varies with the location on the vascular tree, as it is
demonstrated in Fig. 11.2:3. The photos in the first column show the zero-
stress configurations of the sections of an aorta of a normal rat. The location
of the sections is expressed by x/L, x being the distance of the section to the
aortic valve, L being the length of the aorta from the aortic valve to the iliac
bifurcation point. Large variation of the opening angle along the aorta is seen.
The figures in the other columns of Fig. 11.2:3 show the changes that occur
in the rat aorta when the blood pressure is changed by aortic banding: the
imposition of a metal clamp of 2 mm width onto the aorta at x/L == 0.60, just
above the celiac trunk, occluding the aortic lumen area locally 97%. The clamp
causes the blood pressure of the upper body to rise gradually to about 25%
above normal, with half of the rise achieved in 7 days. The blood pressure in
the lower body decreases at first, then rises to a value somewhat below that
of the upper body. The photographs in column 2 show the zero-stress con-
figurations 2 days after banding. The third column shows the configurations
4 days after banding, etc. until 40 days after banding. Large changes in
zero-stress state are seen in the aorta in response to the blood pressure change.
Moreover, the peak of change is reached in 2 to 4 days. Hypertrophy and
remodeling of the blood vessel occurs in this period and afterwards.
Other experiments show that the zero-stress state of arteries change
also with other physical, chemical, and biological stimuli, such as hypoxia,
diabetes. Similar changes are seen in small arteries and arterioles; and also in
veins and in pulmonary and coronary vessels.
 Normal 2 4 6 8 10 13 20 40
N
-
~
4%tlllrlllDIlDOtl N
"..,o
o
~
20% RtJIJElBfl~ElEJ '"Vl
~

40%
IIlImllllllllll
~~~~:7gn-~60% IIliBaaliDIlII
~~~~,~c/1~80% IIl1lJIJllflBIlD
A~100% tlllaBI11111D 1 em

FIGURE 11.2:3 Photographs of the cross sections of the rat aorta when they were cut first transversely and then radially along the "inside"
line indicated in Fig. 11.2:1. The first column shows the zero-stress state of the aorta of normal rat. The rest shows the change of zero-stress
state due to vessel remodeling after a sudden onset of hypertension. The photos are arranged according to days after surgery from left to
right, and according to location on the aorta from top to bottom, expressed as distance from the heart in percentage of the total length. The w
00
location of the metal clip used to induce hypertension is shown in the sketch at left. The arcs of the blood vessel wall do not appear smooth -.l
because of some tissue attached to the wall. In reading these photographs, one should mentally delete these tethered tissues. From Fung
and Liu (1989).
388 11 Stress, Strain, and Stability of Organs

These examples lead us to expect that in a living organ the zero-stress state
is in general different from the no-load state, and that the zero-stress state can
change with blood pressure or other stimuli.

11.3 Stress and Strain in Blood Vessels

Stress and strain in arteries have been analyzed by Bergel (1972), Patel and
Vaishnav (1972), Chuong and Fung (1983), Fung, Fronek and Patitucci (1979),
and many others. In these analyses, the assumption is made that the no-load
state of an artery is its zero-stress state, and the conclusion is reached that at
physiological blood pressure the circumferential stress is very nonuniform,
with a high value at the inner wall and a low value at the outer wall. The ratio
of the stress at the inner wall to that at the outer wall is of the order of two
or three if a linear stress-strain relationship is assumed, but reaches the range
of 10-20 when a nonlinear, experimentally determined constitutive equation
is used.
The reason for finding this stress concentration at the inner wall under the
stated hypothesis is quite simple: Arteries are thick-walled tubes at the no-load
condition, with radius to thickness ratios in the range of 2-4. When they are
inflated to the normal blood pressure in the range of 110-130 mmHg, the tube
radius is enlarged, the wall thickness is decreased and the radius to thickness
ratio is increased. Figure 11.3:1 shows that in this situation the circumferential
strain at the inner wall is larger than that at the outer wall. The stress-strain
relationship being exponential, the stress rises much faster than the strain

IF stres

and zero-stress state were 0<. = O.

then at 100 mmHg: stress

FIGURE 11.3:1 An explanation ofthe nonuniformity ofthe stress distribution in blood

vessel wall. The blood vessel is a thick-walled tube at no-load state. The wall material
is incompressible. When the vessel is inflated to 100 mmHg pressure, the circumferen-
tial strain at the inner wall is larger than that at the outer wall, the stress is more
concentrated at the inner wall because of the nonlinear stress-strain relationship.
 w
-
~
>-\

~
po
IF ZERO STRESS \I ::s
p.
\1
STATE WERE I --- ----- ~
>-\
~ po
© ({) s·
then s·
DC. =0 <X = 10 ()I.. = 70 0(.=155 t:O
0"
0
p.
<:
STRESS ~
AT 0 mmHg !l-
'"
IS @ @-
© 0
whereas
AT 100 mmHg
IT IS
aOG
FIGURE 11.3:2 Illustration of the effect of residual strain on the homeostatic stress distribution in blood vessel wall. Successive rows refer
successively to zero-stress state, state at no-load, and state at 100 mmHg transmural blood pressure. Successive columns refer successively
to a thick-walled tube with no residual strain, a thin-walled tube with a small opening angle, a thick-walled tube with a moderate opening
angle, and a thick-walled tube with a large opening angle. The circumferential stress distribution in the wall of each tube is shown at the
right-hand side of each tube. The larger opening angle correlates with more uniformly distributed circumferential stress.

w
00
\0
390 11 Stress, Strain, and Stability of Organs

when strain increases. As a result the stress at the inner wall becomes large,
and it decreases exponentially to a lower value at the outer wall.

Effect of Zero-Stress State on Stress Distribution In Vivo

Figure 11.3:2 shows qualitatively how the circumferential stress would vary
throughout the vessel wall when the opening angle of the zero-stress state is
changed. If the opening angle a: is zero and the vessel wall is thick, the
circumferential stress (10 is zero when blood pressure is zero, but can become
very nonuniform at a blood pressure of 100 mmHg. If a: is small but the vessel
wall is thin, then the residual stress is small and the stress at 100 mmHg blood
pressure is relatively uniform because of the thinness of the vessel wall. On
the other hand, if the vessel wall is thick and the opening angle is large, then
the residual stress would be large, but the stress distribution at 100 mmHg
could be fairly uniform if a: is sufficiently large. For each vessel a certain
opening angle can be found which can make the circumferential stress at
the inner wall equal to that at the outer wall at the normal blood pressure.
This opening angle increases with increasing thickness to radius ratio. These
qualitative statements can be quantitized by method presented below.

An Example of Stress Analysis

Consider a circular cylindrical artery whose zero-stress state is a circular
sector (Fig. 11.3:3). Let the zero-stress state be called state 0, the unloaded

STATE 0 STATE 1 STATE 2,3, ... N

Pure Bending Pressurization

Axi a 1 St retch

(R, e, Z)

FIGURE 11.3:3 The cross-sectional representations of arteries at the stress-free state,

the unloaded state, and the loaded state. From Chuong and Fung (1986).
11.3 Stress and Strain in Blood Vessels 391

tube as state 1, and the subsequent loaded states as states 2, 3, ... , N. With
cylindrical polar coordinates, a material point is denoted by (R, e, Z) in state
0, and (r, (J, z) in states 1, 2, ... , N. The subscripts i and e indicate the inner
and external walls respectively. eo represents half of the polar angle of the
zero-stress state. Note that eo is related to the opening angle ot by the equation
sin eo
tan(ot/2) = (1 - cos e"0 ) (1)

Using the notations of Sec. 10.2, we see that the incompressibility condition
of the vessel wall requires that
(2)
where Az is the stretch ratio in the axial direction. The deformation of the vessel
is described by the mathematical transformation:
r = r(R), z = z(Z). (3)
The principal stretch ratios are:
8r
A, = 8R' (4)

With incompressibility condition A,AoAz = 1, Eq. (10.2:14), the first of Eqs. (3)
can be written as

(5)

The pseudo-strain energy function is assumed to be (Fung, 1981)

c
poW=-expQ (6)
2
where
Q = blE: + b2 E; + b3 E; + 2b4EOEz + 2bsE zE, + 2b6 E,Eo. (7)
Po is mass density of the wall, c, bl , b2 , •.• , b6 are material constants, Eo, Ez,
E, are Green's strains in the circumferential, longitudinal, and radial direc-
tions, respectively.
E; = t(Al- 1), (i = r, (J, z). (8)
A Lagrangian multiplier H is introduced to impose the condition of incom-
pressibility on the strain energy function, hence:
H
pW* = Po W + "2[(1 + 2Eo)(1 + 2E z)(1 + 2E,) - 1]. (9)

It is known that H has the significance of a hydrostatic pressure (Sec. 10.7). If

a; denotes the coordinates of a material particle at the reference state 0, and
392 11 Stress, Strain, and Stability of Organs

Xi denotes that at the deformed state, then the Cauchy stress components can
be obtained from
P OXj OX i 0
(J··=-----PoW
* (i, j, IX, f3 = r, e, z), (10)
'J Po oaex oap oEpex '
where p, Po denote the densities of the material in the deformed and un-
deformed states (p = Po in the present case). See Sec. 10.7. Introduction of
Eqs. (6)-(9) into (10) yields
(J8 = c(l + 2E8)[b1 E8 + b4 Ez + b6ErJe Q + H
(Jz = c(l + 2Ez ) [b4 E8 + b2 E z + b5 ErJe Q + H (11)
(Jr = c(l + 2Er) [b6E8 + b5 Ez + b3 ErJe Q + H.
The problem of an artery subjected to transmural pressure and longitudinal
tethering force can be solved by substituting Eq. (11) into the equation of
equilibrium

(12)

and the boundary conditions:

1) (Jr = 0 on the external surface r = re ,
2) (Jr = Pi on the inner surface r = ri,
3) On the ends of the blood vessel segment, there acts an external force F.
Application of the conditions 1 and 2 yields

Pi = (i c{(l + 2Er) [b6E8 + b5 Ez + b3 ErJ

For static equilibrium the longitudinal force F + nrlpi equals the integral of
(Jz over the vessel wall cross section; hence

(14)

Use of Eqs. (11) and (13) in (14) yields

F = 2nc fe re Q [(l + 2E z)(b4 E8 + b2 E z + b5 Er)

- t(1 + 2Er)(b6E8 + b5 Ez + b3 Er)
- t(l + 2E8)(b1 E8 + b4 E z + b6Er)] dr. (15)
11.3 Stress and Strain in Blood Vessels 393

2.0~--------------------------~

1.5
~

A.
~
'"
...0 1.0

N
e
...... 0.5
c
~
---- - -----
. 0.0
--
t)

..-"- ..--
....
III
III
cu
-0.5 --- c9
----az
II)

>0-
J:;
u -1.0
:J
IV
U

-1.5

-2.0~----~----~~----~------~
1.0 1.1 1.2

Normalized Undeform Radial Coordinate

R/Ri

FIGURE 11.3:4 The residual Cauchy stress distribution through the thickness of a cat
thoracic aorta at the no-load condition (transmural pressure and longitudinal force
=0). Az = 1. For this computation, the zero-stress state ofthe artery is assumed to be
a circular arc with a polar angle eo equal to 71.4°, outer radius Re = 4.52 mm, inner
radius Ri = 3.92 mm, as in a photograph given in Fig. 2.9:4, p. 59, Fung (1984). From
Chuong and Fung (1986).

Equations (13) and (15) are two integral equations from which we can
determine the material constants with the knowledge of Pi' F, re , r, and Ee, Ez ,
E" i.e., intraluminal pressure, longitudinal stretch force, external radius, inter-
nal radius, and distribution of strain components, respectively. Once the
material constants are determined, we can evaluate the residual stress at state
1 and the stress distribution at loaded states with residual stress and residual
strain taken into consideration. Details are given in Chuong and Fung (1986).
Sketches of some of the results are given in Figs. 11.3:4 and 11.3:5. Figure
11.3:4 shows the residual stress distribution. Figure 11.3:5 shows the stress
distribution at a normal blood pressure of 120 mmHg. It is seen that the
residual stress due to a finite opening angle of 143 has the effect of mak-
0

ing the circumferential and longitudinal stresses CTe and CTz fairly uniform
throughout the blood vessel wall.
394 11 Stress, Strain, and Stability of Organs

300
~

IV
~
~

...0
N 200
E
......
c
~

'"
-
...'"
~
Q)

I/)

>-
..c
u
~ o~----------~----------~
IV
U 1.0 1.1 1.2
o ....... ....... .
-10
-20
... ••••••••••••

-30
Normalized Undeform Radial Coordinate
R/R,

FIGURE 11.3:5 The distribution of the Cauchy stresses (J9' (J. . (J, through the thickness
of a cat thoracic aorta at Pi = 120 nun Hg (~16 kPa ), Az = 1.691, under the same
zero-stress state assumed in Fig. 11.3:4. From Chuong and Fung (1986).

11.4 Stress and Strain in the Heart

Stress and strain in the heart have been analyzed extensively by Grimm, Hood,
Janz, Mirsky, Sandler and Dodge, Spaan, Streeter, Wong, Yin, and others
under the assumption that the unloaded diastolic state is the zero-stress state.
A variety of constitutive equations have been examined, as well as a variety
of simplifying assumptions. See extensive reviews by Mirsky (1979) and Yin
(1981).
The zero-stress state of the left ventricle of the rabbit, dog, and rat has been
studied by Fung (1984) and Omens and Fung (1989). If an unloaded left
ventricle is supported in such a way that the effect of its weight is minimized
(e.g., floating in normal saline and resting on a soft nylon net), then a longitu-
dinal cut would cause the left ventricle to open up as shown in Fig. 11.4: 1. If
an isolated, resting, diastolic left ventricle is sectioned by planes parallel to its
base, Fig. 11.4:2, we obtain a ring. If the ring is cut radially, it opens up. An
opening angle can be defined as shown in Fig. 11.4:3. Changes of strain in the
myocardim from the state of unloaded whole heart to that of the ring and the
11.4 Stress and Strain in the Heart 395

FIGURE 11.4: 1 A left ventricle of the rabbit cut longitudinally on the interventricular
septum. The cut line opens up. From Fung (1984).

B
A

FIGURE 11.4:2 Schematic representation of a left ventricle, showing equatorial slice.

This extracted slice, when placed in the holding tank, will be in the no-load condition.
The slice is cut radially opposite to the right ventricle (not shown). The resulting
configuration is referred to as the radially cut state. From Omens and Fung (1990).
396 11 Stress, Strain, and Stability of Organs

FIGURE 11.4:3 Drawing of slice after radial cut,

OPENING showing definition of opening angle. Note the
ANGLE asymmetry of the posterior and anterior free
walls. From Omens and Fung (1990).
~

sector are measured. Further changes of strain by additional cuts are mea-
sured also.
For quantitative studies a well perfused relaxed ventricle was arrested by
clamping the ascending aorta and injecting an iced, heparinized hyperkalemic
Krebs-Henseleit solution directly into the left ventricle through the apex. To
avoid ischemic contraction, we added a calcium channel blocker nifedipine
(2 x 10- 4 g/l), and a calcium chelator EGTA (10 mMjl) to the perfusate. To
obtain markers for strain measurement, we sprinkled stainless steel micro-
spheres (60-100 !lm diameter) onto the ring shaped slice, and pressed them
lightly into the tissue. The slice was then cut radially as indicated above.
The microspheres were photographed immediately before and immediately
after the cutting (Fig. 11.4:4). Computer graphics record the coordinates of
light spots on selected microspheres, compute the distances between the par-
ticles, and evaluate the Green's strain tensor components Eij according to the
definition given in Eq. (8) of Sec. 10.2. In this computation, the distances ds,
ds o, da; (in notations of Sec. 1.2) are measured and the strain components Eij
are computed from a sufficient number of simultaneous equations or by the
least-squares method if the number of equations exceeds the number of
unknowns. With E;i' the principal strains E 1 , E 2 , and the principal directions
are computed, and the principal stretch ratios,
(1)
are determined.
A typical result is shown in Figure 11.4:5. If the cut ring is considered as
stressfree, then what is shown here is the residual strain in the ring. The
principal directions of the residual strain differ from the radial and circum-
ferential directions within 15 ± 10 degrees (mean ± S.D.), hence the stretch
 ~

~
~
",
'"
s.
~
~.
::s
s·
s-
'"
::t:
",
'"
:l

I __ 1cm
I

FIGURE 11.4:4 An equatorial slice sprinkled with small steel balls which served as .....,
'-0
markers of location. Local strains were computed from the displacements of the steel -.J

balls. From Omens and Fung (1990).

398 11 Stress, Strain, and Stability of Organs

o Circumferenlial • Radial
1.20

g 1.10
•
~
""£
f:
1.00

en 0.90 . 0
... "
0 00
tob 11

tf>" 0
JjlCc90
Db ell

..
1.20 0.80
Endo Ep; 1.20
Radius
1.10 1.10

1.00 t----,..,---"-~...-----

/''"
1.00
08°0 0oe' 8 00
cut
/
0.90

0.80 ' - - - - - - - - - 0.80

.... ... . . .0"

1.20 1.20

1.10 1.10

-1.00 ~
0 0
00
0.90 o 0.90

0.80L----! 0.80

1.20

. ....
\1.20
1.10 •
. ... 1.10
110 Co:'
o
1.00 /----,O,,---:cO"'b-'''''-·e-
o "0".~o--'0"---- 1.00
o 0 o
o 0
0.90 0.90
0.80 L -_ _ _ _ _ _ __
0.80

...
1.20

1.10

0.90
o
a. .. Eqe
1.00 I---.:.,o----==~-..J'--
0 0
1.00

0.90 00
0"

c
..
0'"

O.SO ' - - - - - - - - - - 0.80

FIGURE 11.4:5 Principal stretch ratios of an equatorial slice with a radial cut through
the center of the left ventricular free wall. The no·load geometry is shown, along with
the location of the radial cut. The horizontal axis (radius) of each group ranges from
endocardium to epicardium, although the measured stretch radios do not span this
entire range. The right ventricular wall is ignored. The line of zero strain (A.i = 1.0) is
also shown. The stretch ratio values are ±0.02 due to digitization errors. From Omens
and Fung (1990).

radios are simply labeled as radial and circumferential. (The third principal
strain is perpendicular to the slice, and is not shown here.) Data are plotted
for points along nine radii. It is seen that the residual strain in the circumferen-
tial direction is largely compressive in the endocardial side, and tensile in the
epicardial side; the magnitude being smaller in the interventricular septum.
The radial strain is indeed predominant and mostly tensile. The sign and
magnitude of the radial strain can be predicated by the condition of incom-
11.4 Stress and Strain in the Heart 399

pressibility of the material, which requires that the product of the three
principal stretch ratios be exactly equal to one. If the circumferential and
longitudinal stretch ratios are both less than 1, then the radial stretch ratio
must be greater than 1.
One wonders whether one cut relieves all the residual stress in the slice of
myocardium. The result shown in Fig. 11.4:6 partially answers this question.
The figure shows the changes in strain due to a second radial cut in addition
to the first one. The second cut does cause a little additional strain, implying

o Circumferential 6 Radial o Circumferen:ial 6 Radial

1.20 1.20
6
6 0
,g 1.10
6 6
'"
6 6 00 .Q 1.10
e 0 "§
6 66
6

U i.OO .c 1.00
6 U 00 0
00
~ 0 0
€J ~
(J) 0.90 66
6 iii 0.90

0.80 0.80
Endo Epi Endo Epi
Radius Radius

Deformation due Additional deformation

to first cut due to second cut

First cut __

............ Second cut

FIGURE 11.4:6 Principal stretch ratios after one radial cut, and the additional deforma-
tion due to a second radial cut. These distributions are taken from an area near the
first cut (shown below plots). The stretch ratios after a second cut are referred to the
configuration after the first cut. On a global scale, the second cut does not result in
substantial deformations compared to those after the first cut. From Omens and Fung
(1989).
400 11 Stress, Strain, and Stability of Organs

that one cut is not enough; but it also shows that the changes are small.
Rigorously speaking, many cuts are necessary to relieve the stress completely.
Theoretical analysis of myocardium with measured zero-stress state is best
done with the finite element method. A number of good programs have been
written. See an extensive review by Hunter and Smaill (1988), paper by
McCulloch et al. (1987), and an ASME proceeding edited by Spilker and
Simon (1989).
Accurate measurement of strain in a beating heart by Waldman (1985),
Waldman et al. (1985) and others is an important new development. It yields
information about the dynamics of the heart in normal and diseased condi-
tions. The method uses embedded markers (small lead beads or stainless steel
beads, or the best, gold beads) implanted into the myocardium. The three-
dimensional coordinates of these beads are obtained with high-speed cine-
radiography. Any four noncoplanar markers forming a small tetrahedral
volume «0.1 cc) are used to calculate finite strains according to Eq. (10.2:8).
Waldman et al (1985) showed that the directions of the principal axes of
shortening vary substantially less than the variation of fiber direction across
the wall (20°-40° compared with 100°-140° for fiber direction). Hence the
principal axes of the strain do not coincide with the direction of the muscle
fibers. There are substantial interactions between neighboring fibers in the left
ventricular wall.

11.5 The Musculoskeletal System

The spectacular development of the mechanics of the musculoskeletal system

in recent years has made biomechanics an integral part of medicine and
surgery. The bone has a very complex structure. Its strength, failure, growth,
and resorption have been studied in detail. Replacement of damaged bone
and arthritic joints by artificial bone and artificial joints has led to a rapid
development of biomaterial science and technology. The soft tissues in
the joints: ligaments, tendons, cartilage, and the muscles that mobilize the
musculoskeletal system, have been and continue to be subjects of intensive
study. There is huge literature on this topic but its review is beyond the scope
of this book. Bibliographies of Chaps. 1, 2, and 12 contain references to
mathematical modeling of the musculoskeletal system.

11.6 The Lung

The lung is a porous structure. In applying the concept of continuum me-

chanics to such a structure, it is extremely important to have the objective of
the study and the size of the structure clearly stated at the beginning. For
example, if one is interested in comparing the difference of strain in the upper
part of the human lung from that in the lower part, then individual alveoli
11.6 The Lung 401

can be considered infinitesimal and one can speak of deformation averaged

over volumes that are large compared with the volume of a single alveolus,
but small compared with the whole lung. Such an approximation would be
appropriate in studying the interaction oflung and chest wall, the distribution
of pleural pressure, or the distribution of ventilation in the whole lung. On
the other hand, if one is interested in the stress in a single alveolar wall, then
even the individual collagen and elastin fibers in the wall cannot be ignored.
There are interesting problems at every level; appropriate use of averaging is
often the key to simplification.
In the following, several aspects of the mechanics of the lung will be
discussed. Our general objective is to study the connection between the
mechanical properties of the whole lung and that of the aveolar ducts and
alveoli, i.e., the connection between macro- and micromechanics, with micro
scale identified with alveolar dimension.
We shall take the following steps:
1) Identify the geometry of the alveoli and alveolar ducts.
2) Identify the materials of construction and the geometric configurations of
the materials.
3) Determine the rheological properties of the materials of construction and
the interfacial energy or surface tension.
4) Derive the constitutive equations of the alveolar walls, alveolar mouths,
and the lung parenchyina.
5) Validate the derived constitutive equations and solve useful problems of
the lung.
As it is stated in the Introduction (Sec. 11.1), these same steps are necessary
to study any organ. The lung is used here as an example. In carrying out these
steps, we shall recognize that there exist certain gaps of knowledge. These
gaps will be filled in today by ad hoc hypotheses which will be replaced by
fundamental data and rigorous theoretical derivations in the future.
We explain the geometry, the materials, and a very simple and very crude
derivation of a constitutive equation in this section. Then in Sec. 11.7 the
surface tension on alveolar wall is examined in greater detail. In Sec. 11.8 we
consider small perturbations of lung parenchyma, and in Sec. 11.9 derive the
incremental elastic moduli on the basis of the microstructure. The rest of the
chapter deals with some applied stress problems.

Models of the Alveoli and Alveolar Ducts

Analysis of the elasticity of the lung must begin with a description of the
geometric structure of the lung. Hence it is necessary to have a mathematical
model of the alveolar ducts and alveoli. Looking at histological photographs
such as that shown in Fig. 11.6:1, one gets the impression that alveolar ducts
are bounded by curved edges, but it is very difficult to get a three-dimensional
model. Hanson and Ampaya (1975) reconstructed a region of postmortem
402 11 Stress, Strain, and Stability of Organs

FIGURE 11.6: 1 Histological photograph of alveolar ducts in a thick section (150 J.1m)
of human lung fixed by quick-freezing technique using OS04 and strained by silver to
reveal coHagen fibers. Curvature of alveolar mouths is evident. Specimen and histology
are described in Matsuda et al. (1987).

human lung wall by wall, and obtained a great deal of data, but it is still
difficult to derive a mathematical model.
Pulmonary researchers have long engaged in model building. Miller (1947)
reviewed the history of the development of concepts of the structure of the
alveoli and alveolar ducts from Malpighi to Loeschcke. Orsos (1836), Weibel
(1963), Oldmixon et al. (1987,1988), Mercer et al. (1987) have added new results.
Figure 11.6:2 shows a series of alveolar models published by famous authors.
Those in the upper row visualize alveoli as bunches of grapes. Each alveolar
wall has an "inside" and an "outside". On the inside there acts the alveolar
gas pressure. On the outside, the pressure is presumably the pleural pressure.
This model culminates in that of von Neergaard (1929), shown in the first figure
of lower row, which was taken from Clements et al. (1965). This is the model
that has asserted the importance of surface tension on the stability of the lung.
One difficulty this model leads to is shown in the drawing (a) of Fig. 11.6:3. If
two soap bubbles were blown at the ends of two tubes and these two tubes
are connected and open to each other, than at equilibrium the smaller bubble
 ........
1 .1 111 1-"" '- ,5w: I, "" .

c '~"
§
OQ

,
II. ~""';.1"~1 (lrIS - 7 0)
l~ ~ Q
/I. LDe.ScJ,clc~ ( "ll _ -')

W. -,;,,.,,.r ("31.-I1")
r: WiWS ("" -7')

.. ,"WAY ... 'It~C4

(-'0~
_._- ---- -" .

A
------3 t ----- ! \ I
:1. c/ef'lle t1 f.s (I'I'S)
~- -':-;::->'

r-b HOffin
G. c.,,,,,,,i"j urn)
51' Sem,/e. (lf1J) J. HilJebr-a"dt (lf71)
ER. w~;J>~1 ( l f~J)

FIGURE 11.6:2 A series of well-known alveolar models in the literature. ~

w
404 11 Stress, Strain, and Stability of Organs

TOPOLOGICALLY
WRONG MODEL
FOR ALVEOLAR MECHANICS

NEERGAARD MODEL SEPT. a-a: Both faces in DUCT A.

APPLIES
SEPT. b-b: One face in DUCT A.
another in DUCT B.

Difference of pressures on both sides

of a -a is zero. That on the two sides
of b- b is a few flrn H 20.

FIGURE 11.6:3 Topological structure of pulmonary alveoli. Top left: Topologically

wrong model for alveolar mechanics. Top right: Schematic sketch of the typical
topological arrangement of interalveolar septa. Septum a - a: Both sides are inthe
same alveolar duct; the net pressure difference between the two sides is zero. Septum
b - b: One side is in duct A, and the other side is in duct B; the difference in the
pressures acting on the two sides is on the order of a few Ilm H 2 0. Bottom left:
Equilibrium of pulmonary (visceral) pleura, to which von Neergaard's model applies.
From Fung (1975a).

would have collapsed. * Similarly, if a human lung were considered as 300

million bubbles connected in parallel to an airway, then one would conclude
that at equilibrium all but one alveolus will be collapsed, unless the surface
tension is zero, or the elastic force overwhelms the surface force. The lung
obviously does not behave this way. The fault of this model is that it ignores
the fact that an alveolar wall is an interalveolar septum. The inside of one
alveolus is the outside of another alveolus. The pressure difference across each
septum is negligibly small. A septum must be a minimal surface; i.e., one whose
mean curvature is zero. Any calculation of septal deformation based on
pressure difference of airway and pleura would be wrong.

* On account of Laplaca's fonnula, Eq. (11.7:1), which states that the pressure in the smaller bubble
is higher; hence on opening the connecting valve, the gas in the smaller bubble will flow into the
larger bubble.
11.6 The Lung 405

Other drawings in Fig. 11.6:2 show the models of alveolar duct described
by Weibel (1963) Cumming and Semple (1973) and Hoppin and Hildebrandt
(1977) who did not describe the rules how these units are organized into an
airway tree that fills the entire space of the lung. Karakaplan et al. (1980)'s
model is similar to Hoppin and Hildebrandt's. Wilson and Bachofen (1982)'s
model is based on Orsos' (1936) comments, and it seems to me an over-
simplification.
To meet the space-filling requirement, the author (Fung, 1988) proposed a
model and validated it against all available morphometric data. This model
is based on the assumptions that all alveoli are equal and space filling before
they are ventilated, that they are ventilated to ducts as uniformaly as possible,
reinforced at the edges of the ventilation holes (alveolar mouths) for structural
integrity, and distorted by lung weight and inflation according to the theory
of elasticity.
It is well known that there exists only five regular convex polyhedra in
three-dimensional space: the tetrahedron, the cube, the regular octahedron,
the pentagonal dodecahedron bounded by 12 regular pentagons, and the
icosahedron, bounded by 20 regular triangles. But the last two are not space
filling, and the tetrahedron and octahedron are space filling only if they are
used in combination. On the other hand, the cube, the nonregular octahedron
(every face a triangle), the garnet-shaped rhombic dodecahedron (12 sided,
each side a rhombus), and the tetrakaidekahedron (14 sided polyhedron
formed by cutting the 6 corners off a regular octahedron in such a way as to
leave all edges equal in length) are space filling. (See Fig. 11.6:4.) Thus only
the 4- and 8-hedra with triangular faces, 6- and 12-hedra with rectangular
faces, and 14-hedron with square and hexagonal faces meet the equal and
space-filling requirement. Microscopic observations of the microstructure of
the lung show that the alveolar walls are not all triangles or all rectangles.
This leaves the 14-hedron as a more reasonable candidate.
Fung's basic unit of alveolar duct model is the second order 14-hedron
formed by a 14-hedron surrounded by 14 identical polyhedra, Fig. 11.6:5, with
all the walls of the central 14-hedron removed for ventilation. Two Order-2
14-hedra can be joined together to form a longer duct by removing a few
additional walls. Figure 11.6:6 shows two ways in which two Order-2 14-hedra

FIGURE 11.6:4 Several polyhedrons. The

three on the right-hand side are space
filling. From Fung (1988).
406 11 Stress, Strain, and Stability of Organs

FIGURE 11.6:5 The order-2 14-hedra is the

basic unit of the alveolar duct. Fourteen
14-hedra surround one central 14-hedron.
Several 14-hedra in front are removed to re-
veal the central 14-hedron. To form a basic
unit of lung structure, all faces of central
14-hedron are removed, whereas all of its
edges are reinforced to take up the load
and keep the structure in stable equilibrium.
From Fung (1988).

FIGURE 11.6:6 Two order-2 14-hedra are joined together to form an alveolar duct. At
least one additional face must be removed in order to ventilate the alveoli. The two
assemblies differ by length of one unit. The longer one can be identified as ducts of
generations 4 and 5. The shorter one can be identified as ducts of generations 6 and
7. Ducts of generations 1, 2, 3 are formed by joining one more order-2 polyhedron to
the longer one shown here. From Fung (1988).
11.6 The Lung 407

A ----

~ @ ..

Gen 1,2,3 Gen 4,5 Gen 6,7

FIGURE 11.6:7 Schematic drawings of how order-2 polyhedra are assembled into
alveolar ducts of generations 1, 2, ... 7 and a ductal tree. Generation 8 consists of a
single 14-hedron filling in left-over space. From Fung (1988).

are joined together, differing in lengths by one basic unit (diameter of the first
order 14-hedron). These units can be joined to form a ductal tree, Fig. 11.6:7.
Space can be filled with a combination of first and second order 14-hedra.
Ventilation of every alveolus can be assured by removing a suitable number
of walls. Removal of a wall upsets the balance of forces at an edge, but the
equilibrium can be restored by reinforcing the edge into a "cable" and allowing
it to become curved, Fig. 11.6:8, which explains why the alveolar ducts and
mouths always look rounded in Fig 11.6:1.
Comparison of morphometric predictions ofthis model and the experimen-
tal data published by Hanson and Ampaya (1975) and others is given in Fung
(1988). The model predicts that the length of alveolar ducts of generations 1,
2, 3 is approximately 4Ll, Ll being the dimension of a single alveolus (height
of a 14-hedron). The length of generations 4, 5 is 3Ll, that of generations 6, 7,
8 is 2Ll, whereas the length of the alveolar sac is Llj2. This is in good agreement
with the measured data. The predicted dihedral angles between interalveolar
septa, 110° and 125°, are in agreement with morphometric data. However, in
order to explain the great variety of shapes of the alveoli observed by Hanson
and Ampaya in human lung, it is necessary to assume that an additional
408 11 Stress, Strain, and Stability of Organs

--
-
I

-----
-=::::...----

; c

FIGURE 11.6:8 Illustration of balance of forces at an edge where 3 alveolar walls

intersect. A: An edge where all 3 membranes are intact. B. One of 3 membranes is
perforated. C. Two membranes are perforated. The edges with perforated membranes
must carry tension and be curved in order to balance the membrane tension. From
Fung (1988).

number of interalveolar septa were perforated either naturally or by disease

or aging.
This model is simple enough and detailed enough to serve as the foundation
for theoretical analysis of the lung elasticity and lung ventilation.

Materials of Construction
Collagen, elastin, ground substances, and cells are the principal materials of
the lung parenchyma, which is composed of alveoli and alveolar ducts. Muscle
exists in blood vessels and bronchi. Cartilage exists in trachea and large
bronchi. The mechanical properties of these materials are discussed at length
in Fung (1981). The specific information needed for the lung is their structure
and distribution in lung parenchyma.
Sobin, Fung, and Tremer (1988) have measured the alveolar size and the
width and curvature of collagen and elastin fibers in postmortem human
pulmonary alveolar walls. Matsuda, Fung, and Sobin (1987) measured the
width of these fibers in alveolar mouths. Figure 11.6:9 shows a photograph of
the collagen fiber bundles in a human pulmonary alveolar wall. Figure 11.6: 10
shows a photograph of the elastin fibers. Tables 11.6: 1-11.6:3 list some data
obtained by morphometry.
Statistical data show that the histograms of the width and curvature of
collagen and elastin fibers are skewed to the right, but the square root of the
width and cubic root of the curvature of the fibers in the interalveolar septa
11.6 The Lung 409

FIGURE 11.6:9 Collagen fibers in pulmonary alveolar walls of the lung of a 19 years
old male, inflated to a transpulmonary pressure of 10 em H 2 0. The scale is marked
on the border; 800 pixels are equal to 200 J.1m in the tissue. OS04 fixed. Silver stained.
From Sobin, Fung, and Tremer (1988). Reproduced by permission.

are normally distributed, whereas the fourth root of the width of the fibers
in the alveolar mouths are also normally distributed. Thus the probability
frequency function is

f( x ) -- _ 1_ e -(X-I')2/ (2u 2), (1)

afo
where /1 is the mean of x, a is the standard deviation of (x - /1), and
x = (curvature)1/3 or (width)1/2 in wall,
x = (width)1/4 in mouth. (2)
410 11 Stress, Strain, and Stability of Organs

FIGURE 11.6:10 Elastin fibers in pulmonary alveolar waH of the lung of a 24 years old
male inflated to a transpulmonary pressure of 10 cm H 2 0. OS04 fixed. Orcein stained.
From Sobin, Fung, and Tremer (1988). Reproduced by permission.

The data show several important features:

a) Although the fibers may show some predominant orientation in an
individual septum, over all the septa examined, no predominant orientation
can be identified relative to the alveolar mouth or septal edges.
b) Wider fibers are straighter, but the trend is not strong.
c) The mean value of the curvature of collagen fibers in the interalveolar
septa is larger than that of the elastin fibers; but they are of the same order of
magnitude. For example, in young lungs inflated at a transpulmonary pressure
of 4 cm H 2 0, the curvatures of collagen and elastin fibers in the septa are,
respectively, 0.052 ± 0.048 (SD) vs 0.031 ± 0.030 (SD) 11m -1. The large stan-
dard deviations are reflections of the skewness of the distribution. The data
11.6 The Lung 411

TABLE 11.6:1 Collagen and elastin fiber width in pulmonary interalveolar septa
Age (yrs) Young (15-35) Middle (36-45) Old (65+)

Pressure 4 14 4 14 4 14
TPPcmH 2 0
Collagen

Width Dc Mean 0.966 0.973 0.982 1.044 1.166 1.186

11m S.D. 0.481 0.490 0.522 0.501 0.590 0.631
Dcl /2 Mean 0.952 0.955 0.958 0.994 1.045 1.053
11m 1/2 S.D. 0.242 0.246 0.255 0.237 0.270 0.279
Elastin
Width De Mean 0.973 0.969 1.037 1.045 1.270 1.242
11m S.D. 0.472 0.477 0.531 0.522 0.631 0.632
Del /2 Mean 0.957 0.956 0.970 0.988 1.093 1.079
11m 1/2 S.D. 0.239 0.237 0.213 0.263 0.274 0.281
Note: TPP = transpulmonary pressure = airway p - pleural pressure.
Data from Sobin, Fung, and Tremer (1988).

TABLE 11.6:2 Collagen and elastin fiber curvature in human pulmonary interalveolar
septa
Age (yrs) Young (15-35) Middle (36-45) Old (65+)

Pressure 4 14 4 14 4 14
TPP cm H 2 0
Collagen
Curv Mean 0.052 0.035 0.040 0.029 0.034 0.029
I1m- 1 S.D. 0.048 0.031 0.033 0.027 0.033 0.028
(Curv)1/3 Mean 0.349 0.305 0.319 0.286 0.297 0.280
I1m- 1/3 S.D. 0.094 0.087 0.087 0.078 0.092 0.088
Elastin
Curv Mean 0.031 0.029 0.031 0.027 0.026 0.024
I1m- 1 S.D. 0.030 0.028 0.030 0.028 0.025 0.024
(Curv) 1/3 Mean 0.288 0.285 0.288 0.270 0.273 0.267
I1m- 1/3 S.D. 0.088 0.081 0.086 0.075 0.080 0.079
Note: TPP = transpulmonary pressure = airway pressure - pleural pressure.
Data from Sobin, Fung, and Tremer (1988).
412 11 Stress, Strain, and Stability of Organs

TABLE 11.6.3 The width, D, in 11m, and the fourth root of the width, Dl/4, in (I1m)1/4,
of fibers in alveolar mouths·
WidthD Transpul. Thickness of Mean ± S.D.
or Lung pressure of specimen for 15, 11m
Dl/4 Fiber specimen cmH 2 0 11m n for Dl/4, Ilml/4

D Collegen CPWA 4 150 3095 4.98 ± 2.24

Collagen CRQA 14 160 1405 6.37 ± 2.98
Elastin CGHI 10 85 403 5.72 ± 2.74
Elastin CIVI 4 96 466 6.19 ± 2.23
Dl/4 Collagen CPWA 4 150 3095 1.467 ± 0.165
Collagen CRQA 14 160 1405 1.557 ± 0.182
Elastin CEHI 10 85 403 1.512 ± 0.170
*n = number of measurements.
Data from Matsuda, Fung, and Sobin (1987).

of (curvature)l/3, (curvature)1/4 in Table 11.6:1 show a much smaller relative

SD because the distribution of these transformed variables is nearly normal.
d) The mean value of the width of collagen fibers in interalveolar septa is
also of the same order of magnitude as that of the elastin fibers. In young lungs
at 4 cm H 2 0 transpulmonary pressure, the widths of collagen and elastin fibers
are 0.966 ± 0.481 (SD) and 0.973 ± 0.472 (SD) 11m respectively.
e) In alveolar mouths collagen fibers run parallel, but separately, with
elastin fibers. The widths of the collagen and elastin fibers in alveolar mouths
vary considerably from one individual to another. For example, the widths of
collagen fibers in alveolar mouths of two lungs are 4.98 ± 2.24 (SD) and
6.37 ± 2.98 (SD) 11m. The smallest and largest widths of elastin fibers in
nine lungs are 4.69 ± 2.22 (SD) and 7.11 ± 2.93 (SD) 11m.
f) The ratio of the total volume of collagen fibers in alveolar mouths to
the total volume of collagen fibers in interalveolar septa is 0.312 for young
lungs, 0.460 for middle aged, 0.307 for old lungs. The corresponding ratios for
elastin are 0.478, 0.422, 0.299, respectively.
g) Data from other references are discussed thoroughly in Sobin et al.
(1988).
These data provide a set of fundamental information, which has to be
combined with the rheological data of collagen and elastin, and information
on the contributions of ground substances and cells in order to evaluate the
elasticity of the interalveolar septa and alveolar mouths. Then, with the model
of alveolar ducts and alveoli, we can go on to evaluate the elasticity of lung
parenchyma.

Derivation of a Very Simplified Constitutive Equation

Instead of proceeding rigorously from the parenchymal model and fibrous
structure outlined above, we shall make an ad hoc hypothesis and derive a
11.6 The Lung 413

,,' I
~- -,1-- -"'--
I- .J.B
r,': 11. [,': ~ II' ;
i : : I
,
),--
I
I
,;t"d
-J:-- _~c:.
[" ~ J) 11',
i i
.J-_. -J:- /t_. -~I
I

~' I : ,. II II'
~' I

i :
)- .. . .;-- -.J-- -oJ
I

lot' ~' [/
"

FIGURE 11.6:11 Deformation of an idealized cubic model of the alveolar structure.

From Fung (1975b). Reproduced by permission.

simplified constitutive equation of the lung parenchyma.* We shall use a very

crude model, representing the lung parenchyma by a cubic lattice (Fung, 1975),
Fig. 11.6: 11. In this model all alveoli are equal, cubic in zero-stress state,
parallelopiped in deformed state. The idealized structure consists of three
orthogonal sets of parallel membranes.
Let Xl' X2' and X3 be a rectangular Cartesian frame of reference with
coordinate planes parallel to the interalveolar septa in the zero-stress state,
and assume that Xl' X2' and X3 are the directions of principal axes of macro-
scopic strain. Let A,l' A,2, and A,3 be the principal stretch ratios. Let A be the
distance between the interalveolar septa in the reference state, and let 2y,
(N jm), be the surface tension (two surfaces to each membrane), and N(e), (Njm),
be the elastic tension (stress resultant). To be specific, the tension in Xl
direction in a membrane whose normal is parallel to X2 is denoted by
2Y21 + N1el. Now, consider a unit square perpendicular to Xl. This section
cuts 1j(A,2A} membranes whose normals point in the X2 direction, and Ij(A,3A}
membranes whose normals point in the X3 direction. The force resultant in
the direction of Xl is the macroscopic stress 0"11. It is, therefore,
0"11 = [2Y21 + N1elJ(A,2 A} + [2Y31 + N~elJ(A,3A} - PA(1 - hL}. (3)
Here the first term represents the resultant of tension from all membranes
whose normals point in the X2 direction, the second term represents that from
* See Sec. 11.9 for a derivation based on the model presented in Figs. 11.6:5-11.6:7 and the
rheological data of Tables 11.6:1-11.6:3 and Fung (1981).
414 11 Stress, Strain, and Stability of Organs

membranes with normals in the X3 direction, the last term is the alveolar gas
pressure which acts on an area which is somewhat smaller than the cross
section by an amount equal to the product ofthe thickness ofthe interalveolar
septa, h, and the length of the intercepts of the interalveolar septa per unit
cross-sectional area of the lung, L. Other principal stresses, (122' (133' can be
obtained from Eq. (3) by cyclic permutation of the subscripts 1, 2, and 3.
Now if we can express y and N(e) in terms of A1 , A2 , A3 , we would have
obtained the desired stress-strain relationship. The surface tension will be
considered in Sec. 11.7 infra. The elastic tension can be derived from a strain
energy function as discussed in Sec. 10.7. With the membranes being stressed
in a state of plane stress, we can define a two-dimensional pseudo-strain-
energy function for a membrane (Fung, 1981, p. 249)
(4)
in which C, a 1, a2' a4 are constants, and E 1, E2 are principal strains. Then

N(e) _ A1 ~ w(2) (5)

12 - A2 aE 1
(see Fung, 1981, pp. 246, 248, 276). The stress resultant N!~ can be obtained
by cyclic permutation of subscripts. Substitution of (4) and (5) into (3) yields
the stress-strain relationship. Expressed in terms of a pseudo-strain-energy
density function of the lung parenchyma, Po W per unit volume, we have

Po {1 C
Po W = P 2" A. exp(a1E1 + a2E2 + 2a4E1E2)
2 2

+ symmetric terms by permutation of E 1, E 2, E3

+ interfacial energy} (6)

where C, a 1, a2' a4, are material constants, P is the density ofthe parenchyma
in deformed state, Po is that in zero-stress state, and A. is the alveolar dimension
at zero-stress state. This strain energy is called pseudo-strain-energy because
loading and unloading require different sets of material constants to account
for hysteresis.
These equations have been compared with experimental results by Vawter,
Fung and West (1978, 1979) who tested biaxial stretching of saline-fllied lung
parenchyma. The elastic part ofEq. (6) compares very well with experimental
results. Figure 11.6:12 shows that Eq. (6), which is derived for triaxial loading,
can be used to fit biaxial loading experimental data. The three material
constants, C, a 1 = a2' a4 identified for one biaxial test can be used to predict
the outcome of other biaxial tests; but cannot be used to predict uniaxial test
results. Reliable triaxial test data do not exist. But Eq. (6) was also used to
compare with the triaxial data given by Lee and Frankus (1975), and it is
shown by Fung et al. (1978) that the expression with three constants (by setting
a 1 = a 2 ) fits their data as well as their polynomial with nine constants.
11.6 The Lung 415

10
• ['9 I1 i.. ,nlo l dolo .~In Fy • 0 . 2 N

-. 0
SOLIO CURVES' prtd itlld lot Ih. ten
.....
.
oct Fy - 0 . 2 N on I~t bos i, 01 co n,lo nl, ~
..... 8
u.. dtltr.,;n,d Iro" tlPor ;.,•• " In whith '
(I")~

<nO Fy • 0
w- I'
C:Cx
l-
6 2' Fy.O.IN
V> N
-
z E 3' Fy • 0 . 2 N
oct' 4, Fy • 0 .5""
;:;2
z 4
oct
c:c
<.::I
oct
..J

b
. 2

O.B 1.0 1.2 1.4 1.6 I.B

A. (STRETCH RATIO L / LOll

FIGURE 11.6:12 Comparison of the predicted curves for the case of a biaxial loading
with a lateral load of 0.2 N with experimental data of that case (discrete squares).
Curves 1, 2, 3, 4 are plots of results computed from Eq. (6) with the constants C, at,
a z , a4 identified from experiments with lateral load equal to 0, 0.1, 0.2, and 0.4 N
respectively. Note that curves 2, 3, 4 of the biaxial loading cases fit well; but curve 1
of the uniaxial loading case does not. From Vawter et al. (1978). Reproduced by
permission.

Fung et al. (1978) derived a constitutive equation similar to the above under
a different assumption. They introduced the concept of ensemble average of
alveoli. The ensemble average alveolus is called a mean alveolus. They assume
that the mean alveolus is a sphere in the zero-stress state, and that the
mechanical properties of all the alveolar septa are the same and isotropic
in their own planes. This assumption is said to be the initially isotropic
assumption.
When the lung is deformed, the initial unit sphere is deformed into an
ellipsoid with principal axes equal to the stretch ratios ..1 1 , ..1 2 , ..1 3 , which is the
ensemble average of the deformed alveoli. A detailed calculation leads to the
following strain-energy function of the lung tissue:

Po w(e) = :0 ~eXp[IX(Ei 1+ E~2 + E~3) + ({3 + 21X)

X(E11E22 + E22 E33 + E 33 E 11 ) - {3(Ei2 + m3 + E~I)]. (7)
416 11 Stress, Strain, and Stability of Organs

where 0:, /3, C are constants and Po, p, A have the same meaning as in Eq. (6).
The last three terms in this equation show that the parameter /3 should be
negative valued, otherwise the shear stress will not have the same sign as the
shear strain because the shear stress is

S12
o (e)
= oE12 (PoW) =
Po 2C
-/3pT E 12 exP[ ]. (8)

Since a positive E12 should correspond to a positive S12' and since C is always
positive, /3 must be negative.
Equation (7) has been used by Fung et al. (1978) to reduce the experimental
data obtained by Lee and Frankus (1975), Hoppin et al. (1975), Vawter et al.
(1978). It can fit any given set of data very well by proper choice of the
constants.
Zeng et al. (1987) further discussed how to use the theoretical formula to
fit an entire set of experimental data by a "global" method. They published a
much more extensive set of experimental data. There is no doubt that the
formulas (6) and (7), simple as they are, contain sufficient flexibility to accom-
modate the major features of the stress-strain relationship of the lung. The
major weaknesses of these formulas are: (1) The cubic alveoli model is an
oversimplification; hence the physical meaning of the constants C, aI' a 2, a 4
is not clear, and cannot be related to the microstructure and microrheology
of the materials in detail. (2) There is a hidden hypothesis that the micros train
is the same as the macrostrain. The alveolar mouths or ducts are not given
an additional degree of freedom. (3) Experimental data used for checking are
derived from biaxial loading of liquid-filled lung parenchyma. The surface
tension part needs a more refined treatment, see Sec. 11.7. (4) There is a
lack of a good method to perform triaxial loading experiments on lung
parenchyma. The conventional pressure-volume test of the whole isolated
lung does not yield precise information on stress-strain relationship because
one has to evaluate the effect of pleura and to assume a state of uniform
isotropic strain in the parenchyma when it is not. Hajji et al. (1979) have
estimated that the pleura contributes as much as 25% to the bulk modulus of
the lung. The biaxial testing method (Fung, 1981, p. 243; Vawter et aI., 1978)
solves the problem of edge condition, but the alveolar configuration in biaxial
loading is different from that in normal lung. Hence, the elastic constants
determined may not apply to normal lung. For these reasons, further improve-
ments are necessary. In Secs. 11.8 and 11.9, we shall discuss the derivation of
constitutive equation of the lung more thoroughly.

Boundary-Value Problems
Some problems can be solved by the equations of equilibrium (Eq. 1.7:1) alone:
OUil OUi2 OUi3
-0- + -0- + -0- + Xi = 0, i = 1, 2, 3 (9)
Xl X2 X3
11.6 The Lung 417

where (1ij are the stress components in the parenchyma, Xi is the body force
per unit volume, and Xl' X2' X3 are rectangular cartesian coordinates. For
example, when Xi = 0, a solution of (9) is:
(111 = (122 = (133 = a constant, (1.
(10)
ifi ~ j.

This represents a uniformly inflated weightless lung parenchyma. It can be an

exact solution if the boundary condition on the pleura is satisfied. As seen
from the sketch in Fig. 11.6:13, the boundary condition is
(11)
where (1 is the tissue stress, PAis the alveolar pressure, PPL is the pressure acting
on the pleura, N~U, N~i) are the principal membrane tensions in the pleura,

T T
'--------~
~
0'"

0' = normal component of the resultant

force per unit area of the tension
in interalveolar septa

FIGURE 11.6:13 The boundary condition on the pulmonary pleura. On the side of the
pleural cavity, there acts the intrapleural pressure PPL and the shear stress t due to the
motion of the intrapleural fluid. On the side of the alveoli, there acts the alveolar gas
pressure PA , and the surface traction 1; = uij Vj due to stress uij in the lung parenchyma
(see Sec. 1.7), Vi being the normal vector of the pleura. The normal component of the
surface traction is U = 1; Vi' In the plane of the pleura, there acts principal stress re-
sultants N~t> and N~i) in two orthogonal principal directions. The pleura has principal
curvatures kW and ,,~:r. We assume that the directions ofthe principal axes of pleural
curvature coincide with those of pleural membrane stress, and place the local frame
of reference with origin on the pleura and Xl' X 2 axes pointing in the direction of these
principal axes. The X3 axis is normal to the pleura. The equation of equilibrium of
forces in the direction normal to the pleura is Eq. (11) or (13) in the case illustrated.
418 11 Stress, Strain, and Stability of Organs

K~\!, K~~ are the principal curvatures of the pleura, (1 - hL) is the fraction of
the pleura surface that is exposed to alveolar gas as explained in Eq. (3); the
rest being occupied by the interalveolar septa.
Another solution of Eq. (9) when Xl = pg is

(12)
if i ~ j,
where Ch is a constant, and 9 is the gravitational acceleration. The solution is
exact if the shear on the pleura is zero, and the normal stress on the pleura is
Ch - pgXl = (1 - hL)PA - PPL - NM!K~\! - N~i>K~~. (13)
Eq. (13) defines the pleural pressure that must exist for the solution given by
Eq. (12) to be correct.
For a lung in the chest, the boundary conditions are imposed by the chest
wall and diaphragm, and the motion of the lung relative to the chest wall.
Equations (11) and (13) are not realizable. Hence Eqs. (10) and (12) are only
mathematical ideals.
The boundary conditions on the visceral pleura of the lung are 1), the
normal component of pleural displacement is confined by the chest wall, and
2), the pleura is subjected to the shear stress and pressure imposed by the
intrapleural fluid which flows in a very narrow space. The normal amount of
intrapleural fluid between the visceral and parietal pleura is small; the gap
between the pleura is only a few flm. But the intrapleural fluid is viscous. The
flow of a viscous fluid in the very narrow gap caused by gravity and relative
motion of the lung and the chest wall can generate significant shear stress and
pressure gradient. To determine the boundary condition of the lung, the
problem of flow in the pleural space must be solved. Hence, a study of the
mechanics of the lung in the chest must solve three problems simultaneously:
the lung, the intrapleural flow, and the chest wall motion.

11.7 Surface Tension at the Interface of the Alveolar Gas and

Interalveolar Septa

Energy per unit area of an interface between two materials or two phases of
a material can be revealed as surface tension, which has physical units offorce
per unit length (e.g. N/m). Figure 11.7:1 shows a surface S containing an
arbitrary line element oflength L. In a "free-body" diagram of a small element
of surface containing L, T represents the tension per unit length acting on the
line element. In a length L, the total force is TL. If the line is displaced to the
right by a distance dx, the tensile force will do work equal to TL dx. The
surface area is increased by L dx. If W represents the energy per unit area of
the membrane, the total increase in energy is WL dx. This increase in surface
energy is equal to the work done. Hence TL dx = WL dx or T = W, i.e. the
surface tension per unit length is equal to the surface energy per unit area.
11.7 Surface Tension at the Interface of the Alveolar Gas and Interalveolar Septa 419

&7 L-------------------x
FIGURE 11.7: 1 Illustration of the equivalence of surface tension and energy per unit
area of interface.

FIGURE 11.7:2 The condition of equilibrium of a membrane subjected to internal and

external pressures and tensile stress in the membrane, leading to the Laplace's equation
(1).

Surface tension depends, of course, on the media on both sides of the

interface. Examples are interfaces between oil and water, water and air.
If the surface is curved, Fig. 11.7:2, the effect of surface tension on the
balance of pressures acting on the two sides of the surface is given by Laplace's
equation (see any elementary book on continuum mechanics, e.g., Fung, 1977):

T(~r 1 + ~)
r2
= Pi - Po, (1)

where r1 , r2 are the principal radii of curvature of the surface, and Pi and Po
are the internal and external pressures, respectively. If r1 and r 2 are small,
Pi - Po can be large. For example, the surface tension between pure water and
air being 7.8 x 10- 4 N/cm, ifr1 = r 2 = 1 Ilm, the equilibrium pressure differ-
ence is 15.6 N/cm 2 (1.54 atm).
420 11 Stress, Strain, and Stability of Organs

100
/;-
I I
I I
I I
80
I I
I I
I I
U
-J 60
I-
~

W 40
::l!
::l
-J
0
>
20

o 10 20 30
AIRWAY PRESSURE (em H20)

FIGURE 11.7:3 Pressure-volume curves of a cat's lung. (a) Solid curves: lung inflated
by air. p = airway pressure-pleural pressure. (b) Dotted curves: lung inflated by filling
the airway with saline. The liquid-air interface was thus eliminated. From Bachofen et
al. (1979). Reproduced by permission.

Because the interalveolar septa are moist and are exposed to alveolar gas,
surface tension is significant. Its effect can be seen in Fig. 11.7:3 which shows
two pressure-volume relationships of a dog lung, one inflated by air, the other
inflated by filling the airway with saline. Filling the lung with saline eliminates
the gas-liquid interface, decreases the surface tension, and dramatically alters
the pressure-volume relationship.
When a newborn baby takes the first breath to inflate its lungs, it has to
create the needed interface. If the surface tension is too high, inflation can be
difficult (see Sec. 11.12). Hence the fetus must reduce the surface tension of
the amniotic fluid to an acceptable level. The fetus accomplishes this in the
last few weeks of pregnancy, see Fig. 11.7:4. In this period, certain cells on the
alveolar wall begin to secrete surfactants such as lecithin, sphingomyelin, etc.
which reduce surface tension. Birth will be safer if a sufficient concentration
of surfactants exist in the amniotic fluid. Gluck et al. (1971) first based clinical
decisions on whether a baby is ready to come into this world or not by the
lecithin/sphingomyelin ratio (Fig. 11.7:4) of the amniotic fluid, which is easily
measured by chromatography. Since then childbirth has become much safer.
Methods of measuring surface tension of a fluid-gas interface are described
by Shaw (1970) and others. Clements and Tierney (1965) used a Wilhelmy-
11.7 Surface Tension at the Interface of the Alveolar Gas and Interalveolar Septa 421

v. .··.1!\
22 AMNIOTIC FLUID

20

18 o
i
E
E 16
! Lecith in" ~
~

I
14
;;
~ 12
Z
010
I- o I
0(
0::
I-
8 ......)
... '~""'" i

~ 6
o
Z 4
o
o
2

18 20 22 24 26 28 30 32 34 36 38 ~rm
GESTATION (week,,)

FIGURE 11.7:4 Mean concentrations in amniotic fluid of sphingomyelin and decithin

during gestation. The acute rise in lecithin at 35 weeks marks pulmonary maturity.
From Gluck et al. (1971). Reproduced by permission.

Langmuir trough to measure the surface tension of pulmonary surfactants. A

surfactant is spread on the surface of water in the trough in a layer of about
one molecule thick. The interface is then compressed or extended by a sliding
bar while all the surfactant is confined in the surface and the surface tension
is measured. Figure 11.7:5 shows a typical result. It is seen that the surface
tension varies with the surface area. Increasing area decreases surfactant
concentration and increases surface tension. In a cyclic variation of area there
is a large hysteresis loop, because as the area changes the configurations of
the surfactant molecules are changed, stretched, tumbled, overlapped, piled
up, etc. in an irreversible way, which affects their effectiveness to reduce surface
tension.
Reifenrath and Zimmermann (1973) collected fluid directly from the alveolar
walls of a dog with a microneedle, injected the fluid into a small cubicle of
water, then measured the surface tension of a small gas bubble formed at the
tip of a pipette. They showed that the hysteresis loop of the surface tension-
area relationship can be stabilized only after hundreds of cycles. This method
is discussed by Enhorning (1977).
Schiirch, Goerke, and Clements (1976) attempted to measure the surface
tension of an interface on an interalveolar septum directly by depositing a
metal microsphere on a septum in vivo and observing the change of curvature
of the septum under the weight of the sphere. Flicker and Lee (1974) measured
422 11 Stress, Strain, and Stability of Organs

.75

<
[,l

~
[,l
:::
1-<
.50

<
-1
[,l
~

.25

SURFACE TENSION (dynes/em)

FIGURE 11.7:5 Surface area-surface tension relationship oflung extract from an experi-
ment using Wilhelmy trough. From Clements and Tierney (1965). Reproduced by
permission.

the change of dimensions of the pulmonary alveoli just beneath the pleural
surface of a dog's lung, combined the data with the pressure-volume curves
of air-filled lungs, and computed the surface tension-area relationship under
the hypothesis that the alveoli far from the surface behave the same way as
the subpleural alveoli.
To describe the surface tension-area relationship in cyclic changes of area
mathematically, Fung et al. (1978) wrote Ymin, Y, Ymax for the surface tension at
the minimum area (A min ), intermediate area (A), and maximum area (Amax) of
the interface, respectively, after the hysteresis loop is stabilized by repetition
of periodic changes of area. Define a dimensionless variable e:
(2)
Then the curves as shown in Fig. 11.7:5 can be represented by a straight line
plus a Fourier sine series:

Y = Ymin + (Ymax - Ymin) ( e+ n~l en sin nne) . (3)

One set of the coefficients en applies when the area is increasing, another set
applies when the area is decreasing. Zupkas (1977) has evaluated the coeffi-
cient en from published results of surfactants tested in Wilhelmy-Langmuir
trough type of experiment. Some of his results are listed in Table 11.7:1.
11.7 Surface Tension at the Interface of the Alveolar Gas and Interalveolar Septa 423

TABLE 11.7: 1 Fourier coefficients of the surface tension-surface area relationship given
by Eq. (3). Data by Clements (1962). Computation by Zupkas (1977)
Inflation Deflation
Harmonic Cn x 103 Cn/CI Cn x 103 Cn/ICII
1 366 1.000 -185 -1.000
2 -174 -0.475 -133 -0.719
3 94.1 0.257 -72.6 -0.398
4 -61.0 -0.167 -46.4 -0.251
5 50.0 0.139 -40.3 -0.218
6 -32.6 -0.089 -24.4 -0.132
7 27.1 0.074 -21.3 -0.115
8 -22.2 -0.061 -14.3 -0.077
9 18.9 0.052 -15.7 -0.085
10 -15.6 -0.043 -10.6 -0.057
11 15.2 0.042 -9.42 -0.051
12 -12.1 -0.040 -5.90 -0.032
13 13.8 0.038 -6.27 -0.034
14 -9.73 -0.027 -4.31 -0.023
15 11.3 0.031 -4.72 -0.026
16 -10.8 -0.030 -2.16 -0.012

On the other hand, Flicker and Lee (1974) expressed their results in the form
y = c 1 (A/A min )'2 (4)
whereas Vawter and Shields (1982) wrote
= c~ [1 - C2 exp( - c;A/Amin)].
y (5)
Some typical values are: cdDA = 0.1966 cm H 2 0, DA = alveolar diameter =
60.1 ~m for the dog lung, C2 = 2.71; C~/DA = 6.78 cm H 2 0, C2 = 4.83, c; =
2.35.
In 1982, Wilson introduced a method to evaluate the surface tension on
the alveolar walls of a lung. Their method is based on energy consideration.
They assume that the total energy of an air-filled lung, E, is the sum of the
elastic energy stored in the tissue and the surface energy. The tissue energy is
the sum of two parts, the energy of a saline-filled lung (in which the surface
tension can be ignored), U., and the additional strain energy associated with
the distortion of the alveolar wall caused by the imposition of the surface
tension when the lung became air-filled, AU(V, S). S represents the total area
of the alveolar walls. V represents the lung volume. It is assumed that Us is a
function of Valone, and AU is a function oftwo independent variables V and
S. The surface tension y is a function of S. Thus

E = Us(V) + AU(V, S) + S: ydS. (6)

424 11 Stress, Strain, and Stability of Organs

At a fixed lung volume, the equilibrium state ofthe lung structure minimizes
the total energy. Hence the partial derivative of E with respect to S must be
zero at the equilibrium state:
all V
as+y=o. (7)

If the lung volume is increased by dV, the work done by the transpulmonary
pressure PA - PPL is (PA - ppd dV, whereas the energy is increased by dE.
These must be equal. Hence
dE dVs oLlV OLlV dS dS
PA-PPL=dV=dV+ oV +asdV+YdV' (8)

The sum of the last two terms vanishes on account of Eq. (7). The first term
on the right-hand side, dVs/dV, can be identified as the recoil pressure of the
saline-filled lung Ps • Therefore, Eq. (8) is reduced to the following form:
all V
PA - PPL - Ps = oV . (9)

Differentiating Eq. (7) with respect to V and Eq. (9) with respect to Sand
eliminating 02LlV/OV as, one obtains
oY O(PA - PPL - Ps)
(10)
oV as
Then an integration yields Wilson's equation:

Y
= -IV O(PA - PPL - Ps) dV.
as' (11)
Vs

where V, is the volume of the saline-filled lung at which the surface tension
vanishes. The integrand in Eq. (11) is a function of the alveolar area A and
lung volume V; and can be determined experimentally. Hence Eq. (11) can be
used to calculate y(V, S).
Bachofen et al. (1979) and Gil et al. (1979) fixed rabbit lungs by perfusing
fixatives through pulmonary blood vessels, then prepared histological slides
from which the surface area of the alveolar walls was measured by stereo-
logical methods. They obtained the surface area of alveolar walls in air-filled,
saline-filled, and detergent-rinsed and then air-filled rabbit lungs inflated to
various percentages of total lung capacity. From these data, Wilson and
Bachofen (1982) derived PA - PPL - Ps and V, as functions of V and Sand
calculated y as a function of S. Their results are shown in Fig. 11.7:6. It is seen
that the calculated values of surface tension decrease to less than 2 dyn/cm as
surface area decreases along the deflation limb of the pressure-volume curve.
Surface tension increases very steeply with surface area on the inflation limbs,
reaching a limiting value of just under 30 dynes/cm.
11.8 Small Perturbations Superposed on Large Deformation 425

30'.-----------------------------------------,

.100% TLC
E x 90%
o 80%
~ v 70%
en
CI>
c 60%
~20 .. 50%
.:g.
• 40%
Z
o ... 30%
en
z
w
I-
WlO
u
~
0::
::l
en

o~------~~------~~------~~------~~~
1.5 2.0 2.5 3.0 3.5
SURFACE AREA (104 cm2 )

FIGURE 11.7:6 The surface area-surface tension relationship of rabbit's lung as a

function oflung size. TLC = total lung capacity. From Wilson (1982). Reproduced by
permission.

11.8 Small Perturbations Superposed on Large Deformation

Limiting considerations to small perturbations superposed on a known large
deformation of a body is one of the ways to linearize the nonlinear equations
of biomechanics. For example, we may superimpose a small perturbation on
a uniformly inflated lung, so that the transformation law for the location of a
particle a i in the zero-stress state to the location of the same particle in the
deformed state Xi is
(i = 1,2,3). (1)
Here ai and Xi are referred to the same rectangular cartesian frame of reference,
is a constant and Ui is a function of ai' a2 , a3 • If Ui is so small that the
,1.0
derivatives louJoajl are infinitesimals of the first order, then on substituting
Eq. (1) into the definition of the strain Eq. (10.2:9) and neglecting the second
order terms, we obtain the green's strain

1 (OXk
Eij =="2 oai oaj -
oXk {)ij
)

= (A~ - 1) ().. A ~ (ou i oUj ) (2)

2 'J + 0 2 oa.J + oa., .
On defining the Cauchy infinitesimal strain e;j in the usual way,
426 11 Stress, Strain, and Stability of Organs

elj == ! (ou i + oUj ) (3)

2 oXj OXi
and noting that ou;/oaj is equal to (OU;/OXk)(OXk/fJaj) = A.o(OU;/OXk) according
to Eq. (1) and to the infinitesimals ofthe first order, we can write Eq. (2) as
Eij = !(A.~ - 1)<>jj + A.~elj. (4)
If the stress (Tij is an analytic function of Eij' then on substituting Eq. (4) into
the analytic function we can expand it into a power series of elj' and on keeping
only linear terms of elj' (Tij can be written in the form
(5)
where (TjJ) depends on A.o, but is independent of U;, and (T[j is a linear function
of elj. The terms elj' (T[j are called the incremental strains and incremental
stresses, respectively.
If the relationship between (T[j and elj is the same irrespective of how the
frame of reference is oriented, then the relationship is said to be isotropic. It
is important to note that isotropy in stress-strain relationship does not require
that the microstructure have a spherical symmetry. Only the stress-strain
relationship needs to be. If a stress-strain relationship is derived from a
microstructure presented in Secs. 11.6 and 11.9, and it is shown that the
resulting equation is the same in three different orientations of the frame of
reference, then by tensor transformation rules it can be shown that the equa-
tion will be the same under arbitrary rotation of coordinates, and the isotropy
is established.
An isotropic nonlinear constitutive relation will lead to an isotropic in-
cremental stress-strain relationship in the case of small perturbations of a
uniformly inflated lung, because in this case the stress (TjJ) is isotropic. If the
large deformation is anisotropic, then (T17) is anisotropic and the incremental
stress-strain relationship will be anisotropic.
A basic theorem in continuum mechanics (see, e.g., Fung, 1977, p. 168)
states that a linear isotropic relation between two symmetric tensors of the
second order can be defined by two constants. For stress and strain tensors
the linear isotropic relationship is the Hooke's law,
(6)

or

where A. and J.I. are the Lame constants, E is the Young's modulus, v is the
Poisson's ratio. In engineering literature J.I. is usually denoted by G, and is called
the shear modulus. If such a material is subjected to a uniform pressure p under
which the volume v changes by an amount L\v, then the ratio of - p divided
by L\v/v is the bulk modulus, K, which is related to other constants by the
ll.S Small Perturbations Superposed on Large Deformation 427

formula

K= E =2G(I+v)=A.+~G, (8a)
3(1 - 2v) 3(1 - 2v) 3
9KG E 3K -2G
(8b)
E = 3K + G' G = 2(1 + v)' v = 2(3K + G)·
See Fung (1977), pp. 195,217. These equations are applicable to the perturba-
tion of a uniformly inflated lung.. Many solutions of the classical theory of
elasticity can be applied to the lung. A few examples follow:
Example 1. Cylindrical hole in an initially uniformly expanded parenchyma
(Fig. 11.8:1). Assume isotropic incremental stress-strain relationship. In cylin-
drical polar coordinates, a solution that satisfies the equations of equilibrium is
A
~r=-' ~9 = 0, ~z = 0, (9)
r

where ~" ~9' ~z are displacements in the radial, circumferential, and axial
directions respectively; A being a constant, and r the radial coordinate. See
Foundations of Solid Mechanics (Fung, 1965, pp. 243, 244). This solution can
satisfy the boundary conditions of a specified displacement or radial stress at

FIGURE 11.8:1 A circular cylindrical hole in a uniformly expanded lung requires a

uniform tension to act on the surface of the hole in order to maintain the uniform
stress field. If the hole expands the parenchyma will be distorted with no change in
volume. The radial stress at the wall is reduced by 2G times the fractional change in
hole radius, G being the shear modulus. This disturbance in displacement dies away
at a rate proportional to r- 1 • From Wilson (1986), reproduced by permission.
428 11 Stress, Strain, and Stability of Organs

the inner wall of a cylindrical hole; and zero stress and zero displacement at
r -400. This solution can be used to estimate the interaction between the lung
parenchyma and an embedded blood vessel or bronchus, see Sec. 11.10. The
deformation described by Eq. (9) is such that a cross-hatched element of the
parenchyma shown at the one o'clock position in Fig. 11.8:1 will deform into
a shape shown by the dotted line. The element expands in the circumferential
direction and shrinks in the radial direction, with no change of volume. It is
a "pure shear". If the boundary condition is that ~r = b at r = a, then the
constant A is equal to ab. The stresses that accompany the deformation
depend on the shear modulus of the parenchyma alone, not on the bulk
modulus. The decrease in normal stress at the boundary is 2Gb/a. Hence when
a vessel expands the perivascular pressure increases by 2G times the fractional
change in vessel radius. Since the volume of the parenchyma does not change
with this deformation, ventilation per unit mass of parenchyma is not affected
by the expansion of the vessel.
Example 2. Gravitational deformation of a cylinder of parenchyma in a
rigid container (Fig. 11.8:2). Consider a linear elastic solid in a rigid circular
cylindrical container of radius R and height L. Assume that in the initial state
the solid is subjected to a uniform isotropic tensile stress (lo. Now impose the
gravitational force in the direction of the z-axis. The material is then subjected
to a body force in the z-direction, of magnitude pg per unit volume, p being
the density of the material and g the gravitational acceleration. Let the body
be free to slide but remain in contact with the container. Then the equations

I L

1
FIGURE 11.8:2 Distortion of a linearly elastic solid in a rigid circular cylindrical
container due to gravitational force acting in the axial direction. The distortion
produces local changes in volume and shape. From Wilson (1986), reproduced by
permission.
11.8 Small Perturbations Superposed on Large Deformation 429

of equilibrium and boundary conditions are satisfied by a solution in which

the displacements ~r' ~9' ~z referred to a set of cylindrical polar coordinates r,
e, z are:
~r = 0, ~9 = 0, (10)

The displacement is zero at the top and bottom as required by the boundary
conditions and is maximum at midheight. Horizontal planes shown by the
solid circles in Fig. 11.8:2 are displaced to the positions shown by the dashed
lines. The top half of the solid is expanded relative to the initial state of uniform
expansion, whereas the bottom half is compressed. The corresponding stress
perturbation is:

, = KK-+ 1~G(L
(Jrr pg
)
G 2" - z , (J~9 = 0,

Thus the normal stress in the z-direction is tensile at the top and compressive
at the bottom. The radial normal stress is linear in z as shown by the dashed
lines in Fig. 11.8:2. The vertical gradient of the pressure acting on the cylin-
drical wall (analog of the pleural pressure in the lung) is smaller than the
hydrostatic gradient in a fluid with specific weight pg by the factor
[1 - (2G/3K)]/[1 + (4G/3K)].
This simple example cannot model a lung, but it leads one to expect an
intrapleural pressure gradient different from pg.
Example 3. Half space loaded on the surface. A classical solution for a
semi-infinite space bounded by a plane loaded by a concentrated force has
been used by Hajji et al. (1979) and Lai-Fook et al. (1978) to determine the
incremental modulus of a uniformly inflated lung. The classical solution is
associated with the names of Boussinesq, Kelvin, Hertz, Mindlin, and others,
and can be found in books on elasticity, e.g., Fung (1965).
Example 4. Incremental modulus of lung parenchyma. Experimentally deter-
mined values of the bulk modulus (K) and shear modulus (G) of a dog's lung
for small perturbations of a uniformly inflated state are shown in Figs. 11.8:3
and 11.8:4, from Lai Fook et al. (1978), and Lai-Fook and Toperoff(1980). As
functions of the inflation pressure and volume, we have, roughly
K ~ 4(PA - PPL), G ~ 0.7(PA - PPL)' (dog). (12)
If we compute the Young's modulus E and Poisson's ratio v from the formulas
(8b), we obtain, from Eqs. (12),
E ~ 2(PA - PPL), v ~ 0.42. (13)
Thus, the Poisson's ratio is about 0.42, not far from 0.5.
H is well known that in small deformation, a Poisson's ratio of 0.5 means
that the material is incompressible. For finite deformation this conclusion is
generally untrue even for a material that is isotropic and obeys Hooke's law.
430 11 Stress, Strain, and Stability of Organs

100

so
I
I
/1 I
I
~
I

I I
60 f I
f
I

,,1 tfi
, I

,,
I
I
~
20 , ".

"I I I J
Oil 40 60 SO 100
P. em H2 0 VDlume. "
FIGURE 11.8:3 Bulk modulus of lung parenchyma of excised dog lungs obtained by
small volume perturbations is shown as a function of transpulmonary pressure on the
left and lung volume on the right. Inflation (open circles and dashed line) and deflation
(closed circles and solid) histories are shown. Reprinted with permission from J.
Biomechanics, 12: 757-764, "Elastic properties of lung parenchyma: The effect of
pressure-volume hysteresis on the behavior oflarge blood vessels," copyright © 1979,
Pergamon Press pIc.

24

22

~~O.7 P

o 4 8 12 16 20 24 28
Transpulmonary pressure. em H 20

FIGURE 11.8:4 The shear modulus of excised dog lungs determined by punch identa-
tion tests is shown as a function oftranspulmonary pressure. From Lai-Fook (1979a).
Reproduced by permission.
11.9 Derivation of Constitutive Equation on the Basis of Microstructure 431

Vawter (1983) demonstrated this as follows: The condition of incom pressibil-

ity, expressed in terms ofthe principal stresses 0"1,0"2,0"3 for a material obeying
Eq. (7), is

{1 + ~[0"1 - V(0"2 + 0"3)J}{ 1+ ~[0"2 - V(0"3 + O"dJ}

x{1 + ~ [0"3 - V{O"l + 0"2)J} = 1. (14)

This equation is satisfied by v = 0.5 if the strains O"t/E, 0"2/E, 0"3/E are infini-
tesimal or if the stress state is isotropic (0"1 = 0"2 = 0"3). Otherwise Eq. (14)
cannot be satisfied for arbitrary loading whether v = 0.5 or not. Hence v = 0.5
does not guarantee the absence of volume change in finite strain if the stress
is not isotropic.

11.9 Derivation of Constitutive Equation on the Basis of

Microstructure: Connection Between Micro and
Macro Mechanics

Constitutive equations are phenomenological. They are regarded as empirical

by experimenters, and axiomatic by mathematicians. In biomechanics, we
often try to derive them on the basis of microstructure, e.g., Eq. (11.6:6), in
order to gain a better understanding, or to get some guidance to the mathe-
matical form. Our approach is discussed more fully below.
Consider a lung. If we want to derive a constitutive equation of the whole
lung, we have to solve the problem of stress and strain distribution in the
alveoli and alveolar ducts, for which we need the constitutive equations of the
interalveolar septa and the alveolar mouths. In order to derive the constitutive
equations of the interalveolar septa and alveolar mouths, we have to solve the
problem of stress and strain distribution in the collagen and elastin fibers in
the interalveolar septa and mouths, for which we need the constitutive equa-
tions of the collagen fibers, elastin fibers, ground substances, and cells. Con-
versely, if the constitutive equations of the fibers and ground substances are
known, then we can derive, in succession, the constitutive equations of the
interalveolar septa and alveolar mouths, and that of the whole lung.
The hierarchy of the system is illustrated in Fig. 11.9: 1. To derive the
constitutive equation of the whole lung (the first column), we need to know
the zero-stress state of the whole lung (the upper left-hand box). A macrostrain
of the parenchyma can then be defined relative to the zero-stress state. In
response to this macrostrain, the microstructures of the alveoli and alveolar
ducts deform and produce stresses. The problem can be solved according to
steps listed in the second column. In the process, we need to solve the
sub-problem of the interalveolar septa and alveolar mouths by steps listed in
the third column. The solution of the problems in the third column is then
used to complete the solutions of the problems listed in the second column,
432 11 Stress, Strain, and Stability of Organs

Lung
Macro
Zero-Stress
State
Ducts
and
Alveoli
Micro
Macrostrain Zero-Stress
State Collagen and
!
I
Microstrain
Elastin Fibers
in
Septa and Mouth
Intera Iveo la r
Septa Zero-Stress State

I Equilibrium

Stress
Microstrain
Alveolar Constitutive
Mouth Eqs.

I
Equilibrium
Compatability
Constitution

I
Stress
i
Macro
Stress
I
FIGURE 11.9: 1 Steps involved in the derivation of a macroscopic stress-strain relation-
ship of the lung.

which in turn gives answer to the box at the bottom of the first column. The
relation between the top and bottom boxes of the first column yields the
desired constitutive relationship.

Example: Incremental Elastic Moduli of Lung

We take the first order 14-hedron as the basic model of the alveoli, and the
second order 14-hedron as the basic unit of the alveolar ducts. As explained
in Sec. 11.6, an assemblage of order 1 and order 2 14-hedra is the model of
the lung parenchyma.
11.9 Derivation of Constitutive Equation on the Basis of Microstructure 433

Suppose that we wish to study the incremental elasticity of a uniformly

inflated lung. We choose a rectangular cartesian frame of reference x, y, z with
the origin located at the center of a second-order 14-hedron as shown in Fig.
11.9:2. Impose an incremental macroscopic strain described by the tensor

(1)

,
0;2
i

I 1
j
Z
1 II1 1 <r;z

\ .~
...•....

a-'

y x
r:r
.•.•...
t ., ....

1 j

i I
\
i
(J' (J
zz
27..
t t t t t ~ ~ ~ ; y I
FIGURE 11.9:2 Coordinates for strain and stress analysis. The lung parenchyma is
represented by a continuum of 14-hedra. The incremental loads shown here are: a
uniform tensile stress (J, and a uniaxial tension (Jzz. The length of every edge is L at the
homeostatic condition.
434 11 Stress, Strain, and Stability of Organs

on this continuum of 14-hedra. Consider first the case in which Gxx = Gyy =
Gzz = G. We expect the incremental macroscopic stress tensor to be

(o~ ~ ~).
0 a
(2)

If a can be found, then a/(3G) is the bulk modulus of the lung. To find a, we
note that since the 14-hedra fill the entire space, the condition of compatibility
is satisfied if the strains specified by Eq. (1) apply to the outer shell of all order-2
14-hedra as illustrated in Figs. 11.6:5-11.6:7. However, internally, the strain
in the walls of the 14-hedra may be different from Eq. (1) because the duct has
the freedom to deform differently from the macroscopic average. The strain
ofthe central duct (shown by heavy lines in Figs. 11.6:5-11.6:7) can be specified
by a microscopic strain tensor

(3)

By specifying tensors (1) and (3) on the outer and inner borders of the second
order 14-hedra, the interalveolar septum and alveolar mouths are strained
and stressed in specific ways. The mathematical problem is to determine G;j

FIGURE 11.9:3 Schematic drawing of an infinitesimal element of a pulmonary alveolar

wall membrane with collagen and elastin fibers embedded in it. Stresses exist in the
fibers, ground substances, cell, interfaces. The forces that act on an edge perpendicular
to the x-axis on the right-hand side are depicted here: ai' a 2 , •.• , a i acting in the fibers
in directions tangential to the axes of the fibers. The normal force fxx representing
forces derived from the surface tension, ground substances, cells, and bending and
transverse shear of the fibers; and the shear force tangential to the edge are not shown.
The dimensions L1x, L1y, may be considered as unit length.
11.9 Derivation of Constitutive Equation on the Basis of Microstructure 435

and the stresses in the interalveolar septa and mouths according to the theory
of elasticity.
In the process of the solution we need the constitlJ.tive equation of the
interalveolar septa. Figure 11.9:3 shows such a septum with collagen and
elastin fiber bundles in them. When displacements are imposed on the borders
of the septum, the fibers are stressed and strained. Analysis of the detail is a
sub-problem which requires the constitutive equations of the fibers.
The principle of the analysis is thus very clear. Generalization to other types
of macroscopic and microscopic strains can be done.

The Mathematical Problem

The simplest method of solution is the method of minimum potential energy.
First, the strain distribution in every interalveolar septum and alveolar mouth
is written according to the method of finite elements consistent with the strains
given in Eq. (1) on the outer shell, and Eq. (3) on the inner shell. Second, the
stress-strain laws of the interalveolar septa and the alveolar mouths are
derived according to the morphometric and rheologic data of the collagen
and elastin fibers and ground substances. Third, the strain energy of every
interalveolar septum and every edge of the duct is derived. Fourth, the strain
energy of the whole parenchyma is obtained by summing together the con-
tributions from all elements. Adding to the strain energy the potential energy
of the external load, we obtain the total potential energy. By minimizing the
total potential energy with respect to the unknown variables, we obtain as
many linear equations as the number of unknowns. The solution of these
equations as functions of the specified boundary conditions yields the answer.
Lack of information on the zero-stress states of the collagen and elastin
fiber bundles in the lung is the major hurdle at present. Quantitative deter-
mination of the constitutive equations of these fibers must be pursued. Many
types of collagen have been identified in the lung parenchyma. The constitutive
equation of every type needs to be identified. All fibers may not have the same
zero-stress state in the interalveolar septa. Statistics on the zero-stress states
of the collagen and elastin fibers are of fundamental importance.
Experiments on the zero-stress states of the collagen and elastin fibers and
ground substances in the interalveolar septa can be done by puncturing slits
and circular holes in the septa. If there were residual stresses in these fibers or
ground substance, they will be revealed by the changes of the shape and size
of the slits and circles. The holes and slits are stress indicators which can be
interpreted on the basis of known solutions in the theory of elasticity.

The Influence of Tissue Growth and Change on the

Homeostatic State
In Chapter 13, the relationship between stress and growth is studied. We
shall see that the homeostatic state is, among other things, a result of the
436 11 Stress, Strain, and Stability of Organs

stress state. When the stress state changes, for example, by immobilization or
vigorous exertion, the tissue may change its shape, size, composition, and
zero-stress state.
In Sec. 13.10, especially in Fig. 13.10:4, a basic hypothesis is proposed which
states that the rate of growth of a tissue varies with stress acting in the tissue.
There exists several stress levels at which the rate of growth is zero, i.e.,
homeostatic. A deviation from these stresses will cause either growth or
resorption. If this hypothesis is verified, then it follows that at a homeostatic
condition there is a unique stress. If this rule applies to the collagen fibers in
the interalveolar septa, then we may assume that all collagen fibers have the
same stress. This would greatly simplify the analysis. In particular, the deriva-
tion of the incremental stress-strain law, which describes the relationship
between small incremental strains and stresses superposed on a homeostatic
condition (Sec. 11.8), will be simplified greatly.

Incremental Stress-Strain Relationship

Using the conclusion named at the end of the preceeding paragraph, we
assume that all collagen fibers at the homeostatic state have the same (collagen)
tensile stress; all elastin fibers also have a uniform (elastin) stress at the state
of homeostatis. Then it can be shown that the stress-strain relationship of the
interalveolar septa can be put into the following form:
Nll = (3C + B)e~1 + (C + B)e;z + (Nll)h,
(4)
N12 = 2C e~z·
Here e~I' e~z, e;z are the microstrains in each membrane relative to a set of
rectangular Cartesian coordinates (~, 1]) introduced in each membrane, and
N ll , N IZ , N zz are stress resultants. (Nll)h' (Nzzh are the stress resultants at
homeostatic condition. Band C are constants. B, C are functions of the fiber
statistics, collagen and elastin elasticity, and (Nll)h' (Nzzh. For fiber statistics,
we use the data given in Tables 11.6:1-11.6:3 for human lung. For collagen
elasticity we use the formula given in Fung (1981):
T = (T* + p)eC«;'-;'*) - p, (5)

(6)
Here T is the Lagrangian stress, Ais the stretch ratio. a and T* are two material
constants defined by Eq. (5); they are determined by fitting experiment data
with Eq. (5). Physically, a is the ratio of the incremental modulus of elasticity
to the stress, i.e., (dT/dA)/T. For a linear material obeying Hooke's law, dT/dA
is a constant. Collagen is nonlinear: its dT/dA is proportional to T, and the
constant of proportionality is a. T* is the homeostatic tensile stress at a
homeostatic stretch ratio of A*. pis a constant computed from Eq. (6). Accord-
11.9 Derivation of Constitutive Equation on the Basis of Microstructure 437

TABLE 11.9:1 The coefficients C and B defined in Eq. (4), in units of dynes/em
Transpulm. Hexagonal Membrane at Square membrane at
pressure surface tension y (dyn/cm) surface tension y (dyn/cm)
C,B Age em H 2 0 10 20 10 20

C Young 4 91.0 56.1 111.5 75.2

Young 14 517.9 475.2 612.2 567.0
Middle 4 106.2 70.3 129.8 92.7
Middle 14 684.9 641.3 808.3 764.2
Old 4 147.8 112.4 178.7 142.1
Old 14 905.2 861.6 1066.7 1022.5

B Young 4 1.5 0.9 1.8 1.2

Young 14 4.5 4.1 5.3 5.0
Middle 4 0.4 0.3 0.6 0.4
Middle 14 1.8 1.7 2.2 2.1
Old 4 1.0 0.7 1.2 0.9
Old 14 4.7 4.5 5.5 5.3

ing to the assumption named at the beginning of this paragraph, the stress
T* is the same for all collagen fibers (hence the fiber force is proportional
to the fiber cross-sectional area). The values of a have been determined for
tendon and for arteries; their values differ a great deal. We use IX of the
arteries, and T* computed from equilibrium. For elastin, we used a linear
stress-strain curve (Fung, 1981). For ground substances it was found that
their contribution is negligible. The computed values of Band C depend
largely on the values of IX and surface tension. Examples are given in Table
11.9:1.
The elasticity of alveolar mouth is computed similarly. The strains in the
interalveolar septa and alveolar mouth are then computed from the assumed
macroscopic strain, Eq. (1), applied to the outer border of the 2nd order
14-hedra and microscopic strain, Eq. (3), applied to the central ducts, by finite
element method. The stress is calculated by Eq. (4). Two cases shown in Fig.
11.9:2 are considered: incremental expansion and incremental uniaxial tension
superposed on a uniformly inflated lung. The macroscopic incremental stress
is computed for each case. The bulk modulus K, Young's modulus E, and
shear modulus G, and Poisson's ratio v are computed. The results are listed
in Table 11.9:2.
Note that the principal axes of strain and stress coincide. In fact, the x, y,
z axes in Fig. 11.9:2 are axis of material symmetry of the 14-hedra. Hence the
incremental stress-strain relationship of the lung parenchyma is isotropic, i.e.,
it is the Hooke's law (Eq. 11.8:7).
The computed results compare reasonably well with the experimental
results on postmortem human lung obtained by Yen et al. (1987). Further
calculations show that the bulk modulus and the shear modulus are nearly
438 11 Stress, Strain, and Stability of Organs

TABLE 11.9:2 The effect of the material constant ex of collagen on the Young's modulus
E, bulk modulus K, shear modulus G, and Poisson's ratio v of human lung parenchyma
under the assumptions of T* = 6.56 X 106 dyn/cm 2 when ,1* = 1.3 for collagen,
Eh = 4 X 106 dyn/cm 2 for elastin, surface tension = 10 dyn/cm, morphometric data of
middle aged human
Transpulmonary
pressure P ex 7.5 10.0 12.5 15.0
E(cmH 2 O) 11.77 14.97 18.08 21.13
EIP 2.94 3.74 4.52 5.28
4cmH 2 O KIP 2.40 3.05 3.67 4.28
G/P 1.14 1.45 1.75 2.05
v 0.292 0.291 0.289 0.288
E (cm H 2 0) 53.88 70.68 87.48 104.26
EIP 3.85 5.05 6.25 7.45
14cm H 2 O KIP 3.13 4.11 5.09 6.07
G/p 1.50 1.96 2.42 2.89
v 0.287 0.287 0.287 0.287

proportional to the transpulmonary pressure, agreeing with dog lung result

shown in Fig. 11.8:3. But reasonable agreement with experimental results does
not guarantee the correctness of the basic hypotheses underlying the analysis.
This example should encourage us to really carefully study the zero-stress state
of collagen and elastin fibers in the interalveolar septa, and determine the
constitutive equations of these fibers.

11.10 Interdependence of Mechanical Properties of

Neighboring Organs
When different organs are put next to each other, they will of course interact
mechanically. The mechanical property of an organ in isolation is not the
same as that organ in contact with neighbors. We have illustrated this in
earlier chapters. For example, Fung et al. (1966), Fung (1966) have shown that
the capillary blood vessels in the mesentery are well supported by surrounding
gel and are rather rigid. The capillary blood vessels in a bat's wing, which is
a very thin membrane of approximately 30-40 11m thickness, is fairly disten-
sible (Bouskela and Wiederhielm, 1979) because the surrounding tissue is
relatively small. The capillaries in the pulmonary interalveolar septa are
distensible in the direction perpendicular to the septa because they are un-
supported in that direction, but are indistensible in the planes of the septa
with respect to blood pressure because there are no free space to move (Sec.
6.6). For comparison, the distensibility (change of diameter per cm H 2 0
divided by diameter) of the mesenteric capillaries of the frog given by Baldwin
11.11 Instability of Structures 439

and Gore (1989) is of the order of 0.1% per cm H 2 0, that of the pulmonary
capillaries of the cat is about 5% per cm H 2 0.
The pulmonary arteries and veins are embedded in the lung parenchyma.
When the lung is inflated the parenchyma behaves as an elastic body. The
pulmonary arterial wall is internally subjected to blood pressure and shear
stress, externally to alveolar gas pressure and parenchymal stress, and long-
itudinally to stretch due to lung inflation. The dimensions of the vessel are
influenced by these loads, which are usually coupled together. For example,
when the blood pressure is increased the diameter of the blood vessel will
increase. The increase of diameter induces incremental stresses in the paren-
chyma which resists the expansion of the vessel. Conversely, in a reduction of
blood pressure the lung parenchyma participates in resisting the reduction of
vessel diameter. This is undoubtedly the reason why the pressure-diameter
relationship of the pulmonary blood vessels shown in Fig. 6.6:4 is so different
from that of the aorta shown in figures of Ch. 8 of Fung (1981), the former
being linear whereas the latter being exponential.
Lai-Fook (1979a, b) and Lai-Fook et al. (1978, 1980, 1982) analyzed the
situation by considering a circular cylindrical hole in the parenchyma into
which is fitted a blood vessel. By matching the boundary conditions at the
interface, the radial stress at the interface can be determined, and the "inter-
dependence" of the vessel and parenchyma clarified.

11.11 Instability of Structures

In this and the next sections instability and atelectasis of the lung are discussed.
The word instability, like the word disease, has no unique meaning. Here I
define stability as the tendency of a system toward returning to the initial state
after an arbitrary infinitesimal perturbation: A system is stable if it would,
unstable if the returning is not guaranteed.
The term atelectasis may also mean different things to different people.
Here I mean the existence of some groups of completely collapsed alveoli
in which there is no ventilation. In contrast, Wilson (1982) considered the
pressure-volume curves (PV) of the lung and used a positiveness of the bulk
modulus of the parenchyma as a criterion for stability. Stamenovic (1986)
generalized this concept and defined atelectasis as a coexistence of several
phases of expansion, with each phase having a uniform volume expansion
ratio. While their investigations throw light on the phenomenon, their defini-
tions are different from ours.
Thus defined, atelectasis and instability are not the same thing. A lung
which is stable with respect to small perturbations may be changed into an
atelectatic state by a "large" deformation. On the other hand, an atelectatic
lung may be quite stable in the atelectatic state. However, it is most likely that
atelectasis follows instability. Hence we investigate them both: the initial
tendency toward instability, and the persistent atelectatic plaque.
440 11 Stress, Strain, and Stability of Organs

Atelectasis is often seen in surgery, trauma, disease, airway obstruction,

high oxygen breathing, high acceleration, etc. To a patient or a physician, the
most important question is how to reinflate the collpased region. A related
question is the inflation of a new born baby's liquid filled lung. The answers
are discussed below.

Physical Principles of Stability Analysis

Figure 11.11:1 shows the common concept of equilibrium and stability. A ball
in a concave dish will rest at the bottom and is stable there. Turn the dish over,
the ball will be in equilibrium at the top but unstable there. This concept is
expressed mathematically by saying that the potential energy of the ball is a
relative minimum at the stable equilibrium position, and a relative maximum
at the unstable equilibrium position. The potential energy of the ball, derived
from gravitation, is equal to the product of the weight of the ball and the
height. The relative maximum or minimum is examined against variation of
the radial distance from the equilibrium position.
This concept can be generalized to an elastic' system by the method of
calculus of variation (see, for example, Fung, 1965, Chap. 10). One examines
the variation ofthe potential energy of the system against all possible displace-
ments. If the first variation of the potential energy function is zero at a certain
state of strain, then that state is in equilibrium. If the second variation of the
potential energy is positive at the equilibrium state, then that state is stable.
If the second variation of the potential energy were negative, then that state
of equilibrium is unstable. This is illustrated in the lower half of Fig. 11.11:1.
In this generalization, the slope of the dish is replaced by the first variation,

The concept of stability

height VS. radius

can be generalized by using strain energy PoW

as a function of the 6 components of strain.

~Wl=P'W~
Strain Strain

FIGURE 11.11: 1 The common concept of equilibrium and stability and its generaliza-
tion to a complex elastic structure.
11.11 Instability of Structures 441

the curvature of the dish is replaced by the second variation of the potential
energy.
From the general theory of thermodynamics we recognize the existence of
internal energy. For the lung, the internal energy per unit volume is designated
as Po W; and it consists of two parts: 1) the strain energy in the tissue, and 2)
the surface energy of the liquid-gas interfaces of alveolar walls. As to the
external forces acting on the lung, some, like gravity, is conservative and has
a potential function; others, like aerodynamic forces in airway and viscous
shear force in blood vessels, are nonconservative. All external forces multiplied
by the corresponding displacements is equal to the work done, which has the
same dimension as energy. When the lung deforms a little in an arbitrary way,
the displacement of the lung can be described by a continuous vector field ou.
This displacement causes a change of strain OE ii , a change of stress OSii' a
change of internal energy per unit volume o(Po W), a work done by body force
Xi equal to Xi OUi' and work done by surface force 11 equal to 11 (jUi' Then a
rigorous analysis yields the result (see, e.g., Fung, 1965, p. 450)
apow (1)
Sii=~'
'J

(j2(pO W) > 0: stable, (2)

(j2(pO W) < 0: unstable. (3)
The critical condition of instability is
(j2(pO W) = o. (4)
Here (j2(pO W) is the second variation of Po W; i.e. the second order terms of
the change of Po W due to the small variations of displacements OUi'

Inflated Disturbed

FIGURE 11.11:2 Illustration of the idea that a general, arbitrary deformation is con-
sidered in the stability study. The left-hand side is a mathematical model of a basic unit
of pulmonary alveolar duct. The right-hand side shows a perturbation.
442 11 Stress, Strain, and Stability of Organs

Applying this theorem to the lung, we must allow c5Ui to be completely

arbitrary (subjected only to the restrictions at the boundary of pleura). This
is illustrated in Fig. 11.11:2, in which a model of an alveolar duct is shown in
the left, with each alveolus represented by a 14-sided polyhedron, ventilated
to the duct bounded by heavy lines at the center. A disturbed configuration
is shown on the right hand side, in which every wall and every edge is deformed
in an arbitrary manner. In the theoretical examination of the critical condition
of stability, the tendency to collapse is examined against any possible (acci-
dental or otherwise) small perturbations. If the lung is stable, then any small
transient disturbance will die away. If it is unstable, then a small disturbance
may cause the lung to collapse.
The principle of stability analysis is embodied in the four equations given
above. The steps to be taken for a quantitative evaluation oflung stability are
then very clear: One must evaluate the potential energy, i.e. the sum of strain
energy, surface energy and the potential of external load, as a function of
arbitrary deformation of the lung.

The Potential Energy Function

We have discussed the strain energy function in Secs. 10.6, 10.7, 11.6 (Eqs.
11.6:6, 11.6:7), and 11.7 (Eq. 11.7:6). Since the stress tensor (iij can be obtained
by differentiating the strain energy function with respect to the strain com-
ponents Eij' an integration of the stress with respect to strain yields the strain
energy per unit volume, Po W:

Po W = f Sij dEij. (5)

In the lung, the surface energy of the gas-alveolar wall interface is part of the
strain energy Po W.
The potential energy is the sum of Po Wand the potential of the external
loading, which has been enumerated earlier.

The Stability Criterion

Consider arbitrary small perturbations of an inflated lung. Let the stretch
ratios at equilibrium be AlO , A20 , and A30 . Let the stretch ratios of the
perturbed lung be Al = AlO + c5A l , A2 = A20 + c5A 2, and A3 = A30 + c5A 3. The
c5A values are arbitrary, but infinitesimal. The strain energy Po W is changed
to Po W + c5(po W) due to the perturbation. The stability is determined by
the second variation of the strain energy, c5 2 (po W). The system is stable if
c5 2 (po W) :;::: 0; otherwise it is unstable.
Substituting Ai = AiO + c5Ai into Po Wand retaining only the second-order
terms results in
3
c5 2 (po W) = L kij c5A ic5Aj, (6)
i,j=l
11.12 Collapsed Structure. Example of Atelectatic Lung 443

where kij are the values of the second derivatives evaluated at Ai = AiO:
02p O W
kij = OA.-OA.- . (i,j = i, 2, 3). (7)
, J

The right side of Eq. (6) is a quadratic form. If the equilibrium is to be stable,
the quadratic form must be positive definite, i.e., >0 for whatever values of
<5Ai and <5Aj , and 0 only when <5Ai = <5Aj = O. The conditions for the positive
definiteness are (see Fung, 1965, pp. 29-30):
kll + k22 + k33 > 0, (8)

Ikll
k21
k121
k22
+ Ik22
k32
k231
k33
+ Ik 33
k13
k311 > 0
kll '
(9)

kll k12 k13

k21 k22 k23 >0. (10)
k31 k32 k33
The quantities in Eqs. (9) and (10) are determinants. Thus the problem of
stability is reduced to the checking of the three equations, (8), (9), and (10).
The simplest and most important case to check is the uniformly inflated
state. Detailed examination shows (Fung, 1975) that the inflated state is stable
above a critical transpulmonary pressure. On the other hand, an unstable state
is obtained if the surface tension is a constant independent of area, and the
elastic stress is zero (in which case kll = 0).

11.12 Collapsed Structure. Example of Atelectatic Lung

When the critical condition for instability is met, the lung will have a tendency
toward collapse. The following three types of collapsed lung are of special
interest: (1) atelectasis of the focal type, in which the alveoli collapse toward
a central focus, (2) atelectasis of the axial type, in which the alveoli collapse
toward a line, and (3) atelectasis ofthe planar type, in which the alveoli collapse
toward a plane. These are illustrated in Figs. 11.12:1 and 11.12:2.
In each case it is assumed that at the core a number of the interalveolar
septa are coalesced. When two interalveolar septa touch each other, their
liquid coverings will fuse, thus eliminating surface tension. Pressure, tension,
and shear stress can be transmitted through the coalesced alveolar septa.
Immediately next to these coalesced septa are the open alveoli, whose walls
are moist, on which surface tension acts. The dimensions of the alveoli next
to the coalesced region may conceivably be reduced because of the necessary
continuity of the membranes (septa). Farther away from the atelectatic core,
the alveoli are less and less influenced by the localized perturbation; they
become those of the normally inflated lung.
There are, however, considerable differences in the behavior of the three
types of atelectasis. For the planar type, the transition from the coalesced
444 11 Stress, Strain, and Stability of Organs

2L
-K{A.)o6 K("f)O~
.-
:-....

./
:-....
-
;"

i.-:: J....

-
22 -+ +-_L_
r-r-, K-k
;" r--..
/'" .......
\

-,
J

----
......... ./
:-.... ;"

1...-1---

2k alveolar layers
collapsed

FIGURE 11.12:1 Conceptual description of a planar type of atelectasis. 2k layers of

alveoli are collapsed into a plaque. From Fung (1975b).

region to that of the normally inflated alveoli can be immediate. In other

words, once created, a planar atelectasis can exist in an inflated lung. Since
an inflated lung is stable with respect to all infinitesimal disturbances, an
atelectic plaque can be introduced only by a large disturbance, such as an
obstruction of the airway, a compression by a tumor, a pressure by a surgeon's
finger, or a transient local reduction of stress to a level below the critical stress
of instability. Because planar atelectasis is stable, it can be very persistent.
For atelectasis of the axial type, we ask two questions: 1) How large is the
transition zone between the coalesced core and the normal alveoli? 2} Can we
pull the coalesced alveolar walls out by inflating the lung? The answer to 1}
is zero; that to 2} is that we can. These answers are based on exact solutions
of the equations of equilibrium, compatibility, and boundary conditions
(Fung, 1975). In the transition zone it is shown that the displacement must be
proportional to r, resulting in a constant radial strain similar to a pattern
11.12 Collapsed Structure. Example of Atelectatic Lung 445

Spacing >A

••
FIGURE 11.12:2 An axial or a focal atelectasis. The alveoli in the central region are
collapsed. A transition zone between circles of radius a and radius b is assumed. Beyond
radius b is the normally inflated zone. A rigorous mathematical analysis shows that
b = a, i.e., the transition zone does not exist. Lower figures: Illustration that (1) the
direction of the idealized alveoli is immaterial in the stability argument and (2) the
edge of a planar atelectatic plaque provides a region in which the interalveolar septa
can be pulled out to reinflate the alveoli. From Fung (1975b).

given by Eq. (11.8:9). The boundary condition with the normal alveoli then
leads to the first conclusion. For the answer to the second question, we
consider a small perturbation of the uniformly inflated region of the lung
surrounding the core. For small perturbations a linearized theory suffices. An
exact linear solution is that the radial displacement is inversely proportional
to r. Consequently, the radial strain decreases with r, the circumferential strain
increases with r, the areal strain is zero, i.e., the cross-sectional area of any
small element in r, () plane is preserved. This means that the tissue can be
pulled out of the core and the alveoli reinflated by an outward radial move-
ment, or coalesced into the core by a radial movement in the opposite
direction.
The focal type of atelectasis behave similarly. A transition zone does not
exist. Small centripetal radial displacement may push the parenchyma into
the core, outward radial displacement may pull the alveoli coalesced in the
core out into the inflated lung.
 446 11 Stress, Strain, and Stability of Organs

Reinflation of Atelectatic Lung

We have explained above that from a mechanical point of view, focal and
axial atelectasis can be pulled out by reinflating the lung to a larger size. But
how can a planar atelectasis be pulled out? It is not effective to pull the planar
coalesced region which can transmit tensile stress. But the edges of a plaque
of planar atelectasis must behave somewhat like half of an axial atelectatic
core. Since there is no axial symmetry, the condition there is quite complex,
but it is clear that to remove a planar atelectasis one should work on the edges
of the plaque. Overinflation will pull out the alveolar septa at the edges. One
needs to overcome, in this process, not only the surface tension of the newly
created interfaces but also the viscou~ friction between the septa because of
their relative motion. Therefore, the duration of overinflation should not be
too brief. But excessive positive airway pressure compresses the alveolar
capillaries and decreases the blood flow; hence, the duration of overinflation
should not be too long. Intermittently applied positive-pressure breathing at
a suitable frequency, or negative pressure applied to the chest or abdomen,
are local procedures. When a surgeon massages a lung after an operation to
remove signs of atelectasis, he is applying this principle. The same principle
tells why it is difficult to open an atelectatic plaque which borders on a pleura
where the peeling action is ineffective.
Fleischner (1936) first described platelike atelectasis in roentgenograms of
the lung. The linear shadows in the lung, variously called Fleischner's lines,
platter atelectasis, or discoid atelectasis, have been discussed in detail by
Fraser and Pare (1970, p. 301). These lines are undoubtedly shadows of planar
atelectasis. Fraser and Pare (pp. 196-239) presented detailed patterns of
lobar and total pulmonary collapse. They showed that when atelectasis is
approached a collapsed lobe or segment tends to look like a curved plate
whose edges tend not to retract from the chest wall and the mediastinum.
Thus, an atelectatic lobe is also planar.

Problems
11.1 Develop a finite element program to analyze the stress distribution in arterial
wall by taking into account the open zero-stress configurations as shown in
Fig. 11.2:3.
11.2 Continuing the development suggested in Problem 11.1, develop a finite element
program to analyze the stress distribution in the vessel wall in the aortic arch
region, where the vessel is toroidal.
11.3 Develop further a program for computing the stress distribution in the left
ventricle, taking into account the zero-stress configurations shown in Fig. 11.4:4.
11.4 In order to measure the mechanical properties of vascular smooth muscles, small
arteries are used because smooth muscle cells occupy a larger portion of the vessel
wall in smaller vessels. In vascular smooth muscle research, arterioles at the
periphery are favored because smooth muscle cells occupy 80-90% of the walls
References 447

of these vessels (the rest is the intima, collagen, elastin, and adventitia). Propose
a constitutive equation for the smooth muscle, containing several unknown mate-
rial constants. In order to identify these unknown constants, it is expedient to
test the vessel in unusual deformation modes in addition to the usual inflation
by internal pressure. Fung used a pair of micro pipettes with rectangular mouths
to deform the vessel wall. The pipettes can push the vessel into elliptical shape.
Using a vacuum pump, the pipette can suck the wall into the pipette mouth
to an extent depending on the transmural pressure. The deformation is measured.
Develop a computing program to identify the material constants.
11.5 Similar to the scheme outlined in Pro b .. 11.4, micropipette aspiration method can
be used to study the mechanical properties of cell membranes. Develop a comput-
ing program to evaluate the material constants of the membrane constitutive
equation from cell aspiration data. (Cf.: Chien et aI., 1978.)
11.6 One of the simplest ways to evaluate residual stress in a membrane is to poke
holes in an unloaded membrane and observe the change in geometry of the holes.
A circular punch creating a circular hole, or a knife creating a slit would provide
very useful data. To understand the results quantitatively, it is necessary to
assume a constitutive equation of the membrane material. Develop a computing
program to relate the deformation of the hole to the residual stress and constitu-
tive equation. Include the identification of material constants in your program.

References

Ardila, R., Horie, T., and Hildebrandt, J. (1974). Macroscopic isotropy of lung expansion. Resp.
Physiol.20: 105-115.
Bachofen, H., Gehr, P., and Weibel, E.R (1979). Alterations of mechanical properties and
morphology in excised rabbit lungs rinsed with a detergent. J. Appl. Physiol.: Resp. Environ.
Exercise Physiol. 47: 1002-1010.
Bachofen, H., Hildebrandt, J., and Bachofen, M. (1970). Pressure-volume curves of air and
saline-filled excised lungs: Surface tension in situ. J. Appl. Physiol. 29: 422-431.
Baldwin, A.L. and Gore, RW. (1989). Simultaneous measurement of capillary distensibility and
hydraulic conductance. Microv. Res. 36: 1-22.
Bergel, D.H. (1972). The properties of blood vessels. In: Biomechanics: Its Foundations and
Objectives, (y.c. Fung and M. Anliker, eds.). Englewood Cliffs, N.J. Prentice-Hall, Ch. 5,
pp. 105-140.
Bergel, D.H. (1961). The static elastic properties of the arterial wall. The dynamic elastic properties
of the arterial wall. J. Physiol. (London) 156: 445-469.
Bouskela, E. and Wiederhielm, C.A. (1979). Microvascular myogenic reaction in the wing of the
intact unanesthetized bat. Amer. J. Physiol. 237(1): H59-H65.
Chien, S., Sung, P., Skalak, R, Usami, S., and Tozeren, A. (1978). Theoretical and experimental
studies on viscoelastic properties of erythrocytes membrane. Biophy. J. 24: 463-487.
Chuong, c.J. and Fung, Y.c. (1983). Three-dimensional stress distribution in arteries. J. Biomech.
Eng. 105: 268-274.
Chuong, C.l. and Fung, Y.c. (1986). Residual stress in arteries. In: Frontiers in Biomechanics
(G.W. Schmid-Schonbein, S.L.-Y Woo, and B.W. Zweifach, eds.), Springer-Verlag, New
York, pp. 117-129.
Clements, J.A. (1962). Surface phenomena in relation to pulmonary function. Physiologist 5:
11-28.
Clements, J.A. and Tierney, D.F. (1965). Alveolar instability associated with altered surface
448 11 Stress, Strain, and Stability of Organs

tension. In Handbook of Physiology, Sec. 3, Respiration, Vol. 2 (W.O. Fenn and H. Rahn,
eds.). Amer. Physiol. Soc., Washington, D.C. pp. 1565-1583.
Cumming, G. and Semple, S.J. (1973). Disorders of the Respiratory System. Blackwell, Oxford.
Dale, P.J., Mathews, F.L., and Schrotter, R.C. (1980). Finite element analysis of lung alveolus.
J. Biomech. 13: 865-873.
Doyle, J.M. and Dobrin, P.B. (1973). Stress gradients in the walls of large arteries. J. Biomech.
6: 631-639.
Enhorning, G. (1977). Pulsating bubble techniques for evaluating pulmonary surfactant. J. Appl.
Physiol.: Resp. Environ. Exercise Physiol. 43: 198-203.
Fleischner, F.G. (1936). Dber das Wessen der baselan horizontalen Schattenstreifen im Lungen-
feld. Wien Arch. Inn. Med. 28: 461-480.
Flicker, E. and Lee, J.S. (1974). Equilibrium of force of subpleural alveoli: implications to lung
mechanics. J. Appl. Physiol. 36: 366-374.
Frankus, A. and Lee, G.c. (1974). A theory of distortion of lung parenchyma based on alveolar
membrane properties. J. Biomech. 7: 101-107.
Fraser, R.G. and Pare, J.A.P. (1970, 1977, 1979). Diagnosis of Diseases of the Chest. 1st ed. 1970,
2nd ed., Vol. 1,2,1977, Vol. 3, 4,1979. Saunders, Philadelphia.
Fung, Y.c. (1965). Foundations of Solid Mechanics, Prentice-Hall, Englewood Cliffs, New Jersey.
Fung, Y.c. (1966). Theoretical considerations of the elasticity ofred cells and small blood vessels.
Fed. Proc. 25: 1761-1772.
Fung, Y.c. (1974). A theory of elasticity of the lung. J. Appl. Mech. 41: 8-14.
Fung, Y.c. (1975a). Does the surface tension make the lung inherently unstable? Cir. Res. 37:
497-502.
Fung, Y.c. (1975b). Stress, deformation, and atelectasis of the lung. Circ. Res. 37: 481-496.
Fung, Y.c. (1977). A First Course in Continuum Mechanics, 2nd Edn, Prentice-Hall, Englewood
Cliffs, N.J.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, N.Y.
Fung, Y.c. (1984). Biodynamics: Circulation. Springer-Verlag, New York, pp. 53-66.
Fung, Y.c. (1985). What principle governs the stress distribution in living organs? In: Bio-
mechanics in China, Japan, and U.s.A. (Y.c. Fung, E. Fukada, and J.J. Wang, eds.). Science
Press, Beijing, China, pp. 1-13.
Fung, Y.c. (1988). A model of the alveolar ducts of lung and its validation. J. Api. Physiol. 64:
2132-2141.
Fung, Y.c., Zweifach, B.W., Intaglietta, M. (1966). Elastic environment of the capillary bed. Circ.
Res. 19: 441-461.
Fung, Y.c., Tong, P., and Patitucci, P. (1978). Stress and strain in the lung. J. Eng. Mech. Div.,
ASCE 104: 201-223.
Fung, Y.c., Fronek, K., and Patitucci, P. (1979). Pseudoelasticity of arteries and the choice of its
mathematical expression. Amer J. Physiol. (Heart, Circ.) 237: H620-H631.
Fung, Y.c. and Sobin, S.S. (1982). On the constitutive equation of the lung tissue. In 1982 Advances
in Bioengineering. Amer. Soc. Mech. Engineers, New York, pp. 84-87.
Fung, Y.c., Sobin, S.S., Tremer, H., Yen, R.T. and Ho, H.H. (1983). Patency and compliance of
pulmonary veins when airway pressure exceeds blood pressure. J. Appl. Physiol.: Respirat.
Environ. Exer. Physiol. 54(6): 1538-1549.
Fung, Y.c., Yen, R.T., Tao, Z.L., and Liu, S.Q. (1988). A hypothesis on the mechanism of trauma
oflung tissue subjected to impact load. J. Biomech. Eng. 110: 50-56.
Fung, Y.c. and Liu, S.Q. (1989). Change of residual stress in arteries due to hypertrophy caused
by aortic constriction. Circ. Res. 65: 1340-1349.
Gil, 1., Bachofen, H., Gehr, P., and Weibel, E.R. (1979). The alveolar volume-to-surface-area
relationship in air-and-saline-filled lungs fixed by vascular perfusion. J. Appl. Physiol.: Resp.
Environ. Exercise Physiol. 47: 990-1001.
Gluck, L., Kulovich, M.V., Borer, R.c., Jr., Brenner, P.H., Anderson, G.G., and Sepllacy, W.N.
(1971). Diagnosis of the respiratory stress syndrome by amniocentesis. Am. J. Obstet. Gynecol.
109: 440-445.
References 449

Haiji, M.A., Wilson, T.A., and Lai-Fook. S.J. (1979). Improved measurements of the shear
modulus and pleural membrane tension of the lung. J. Appl. Physiol.: Resp. Environ. Exercise
Physiol. 47(1): 175-181.
Hansen, J.E. and Ampaya, E.P. (1975). Human air space shapes, sizes, areas, and volumes. J. Appl.
Physiol. 38: 990-995.
Hayashi, K., Handa, H., Mori, K., and Moritake, K. (1971). Mechanical behavior of vascular
walls. J. Soc. Material Science Japan 20: 1001-1011.
Hildebrandt, J. (1969). Dynamic properties of air-filled excised cat lungs determined by fluid
plethysmograph. J. Appl. Physiol. 27: 246-250.
Hoppin, F.G., Jr., Lee, G.c., and Dawson, S.V. (1975). Properties of lung parenchyma in distor-
tion. J. Appl. Physiol. 39: 742-751.
Hoppin, F.G., Jr. and Hildebrandt, J. (1977). Mechanical properties ofthe lung. In Bioengineering
Aspects of the Lung, (J.B. West, ed.). Marcel Dekker, New York, pp. 83-162.
Hunter, PJ. and Smail, B.H. (1988). The analysis of cardiac function: a continuum approach.
Prog. Biophys. Molec. Bioi. 52: 101-164.
Karakaplan, A.D., Bieniek, M.P., and Skalak, R. (1980). A mathematical model of lung paren-
chyma. J. Biomech. Eng. 102: 124-136.
Lai-Fook, S.J. (1979a). A continuum mechanics analysis of pulmonary vascular interdependence
in isolated dog lobes. J. Appl. Physiol.: Resp. Environ. Exercise Physiol. 46: 419-429.
Lai-Fook, S.J. (1979b). Elastic properties of lung parenchyma: the effect of pressure-volume
hysteresis on the behaviour oflarge blood vessels. J. Biomech. 12: 757-764.
Lai-Fook, S.J., Wilson, T.A., Hyatt, R.E., and Rodarte, J.R. (1976). The elastic constants of inflated
lobes of dog lungs. J. Appl. Physiol. 40: 408-513.
Lai-Fook, S.J., Hyatt, R.E., and Rodarte, J.R. (1978). The effect of parenchymal shear modulus
and lung volume on bronchial pressure-diameter behavior. J. Appl. Physiol.: Resp. Environ.
Exercise Physiol. 44: 859-868.
Lai-Fook, S.J. and Toperoff, B. (1980). Pressure-volume behavior of perivascular interstitium
measured directly in isolated dog lung. J. Appl. Physiol.: Resp. Environ. Exercise Physiol. 48:
939-946.
Lai-Fook, S.J. and Kallok, M.J. (1982). Bronchial-arterial interdependence in isolated dog lung.
J. Appl. Physiol.: Resp. Environ. Exercise Physiol. 52: 1000-1007.
Lambert, R.K. and Wilson, T.A. (1973). A model for the elastic properties of the lung and their
effect on expiratory flow. J. Appl. Physiol. 34: 34-48.
Lanir, Y. (1983). Constitutive equation for lung tissue. J. Biomech. Eng. 105: 374-380.
Lee, G.c. (1978). Solid mechanics of lungs. J. Eng. Mech. Trans. ASCE, 104: 177-200.
Lee, G.C. and Frankus, A. (1975). Elasticity properties of lung parenchyma derived from ex-
perimental distortion data. Biophys. J. 15: 481-493.
Lee, G.c., Frankus, A., and Chen, P.D. (1976). Small distortion properties of lung parenchyma
as a compressible continuum. J. Biomech. 9: 641-648.
Liu, J.T. and Lee, G.c. (1978). Static finite deformation analysis of the lung. J. Eng. Mech. Div .•
ASCE 104: 225-238.
Liu, S.Q. and Fung, Y.c. (1988). Zero-stress states ofarteries. J. Biomech. Eng. 110: 82-84.
Liu, S.Q. and Fung, Y.c. (1989). Relationship between hypertension, hypertrophy, and opening
angle of zero-stress state of arteries following aortic constriction. J. Biomech. Eng. 111:
325-335.
Love, A.E.H. The Mathematical Theory of Elasticity. Cambridge University Press, Cambridge,
1st ed., 1892, 1893; 4th ed., 1927. Reprinted Dover Publication, New York, 1944. Note
especially the "Historical Introduction", pp. 1-31.
Matsuda, M., Fung, Y.c., and Sobin, S.S. (1987) Collagen and elastin fibers in human pulmonary
alveolar mouths and ducts. J. Appl. Physiol. 63: 1185-1194.
McCulloch, A.D., Smaill, B.H. and Hunter, P.J. (1987). Left ventricular epicardial deformation
in isolated arrested dog heart. Am. J. Physioi. 252: H233-H241.
Mead, J., Takishima, T., and Leith, D. (1970). Stress distribution in lungs: a model of pulmonary
elasticity. J. Appl. Physiol. 28: 596-608.
450 11 Stress, Strain, and Stability of Organs

Mercer, R.R. and Crapo, J.D. (1987). Three-dimensional reconstruction of the rat acinus. J. Appl.
Physiol. 63: 785-794.
Miller, W.S. (1947) The Lung, 2nd ed. Springfield, IL., c.c. Thomas.
Mirsky, I. (1979). Elastic properties of the myocardium: a quantitative approach with physio-
logical and clinical applications. In: Handbook of Physiology, Sec. 2, The Cardiovascular
System, Vol. 1, The Heart (R.M. Berne and N. Sperelakis, eds.). Amer. Physiol. Soci.,
Bethesda, MD. pp. 497-531.
Oldmixon, E.H. and F.G. Hoppin. (1987) Lengths and topology of septal borders (Abstract)
Federation Proc. 46: 820, 1987.
Oldmixon, E.H., Butler, J.P., and Hoppin, F.G., Jr. (1988) Dihedral angles between alveolar septa.
J. Appl. Physiol. 64: 299-307.
Omens, J. and Fung, Y.C. (1989). Residual strain in the rat left ventricle. Cire. Res. 66: 37-45.
Orsos, F. (1936). The frameworks of the lung and their physiological and pathological significance.
Beitrage zur Klinik der Tuberkulose und speziejischen Thberkuloseforschung. 87: 568-609.
Patel, D.J. and Vaishnav, R.N. (1972). The rheology of large blood vessels. In: Cardiovascular
Fluid Dynamics, (D.H. Bergel, ed.). Academic Press, New York, Vol. 2: Ch. 11, pp. 2-65.
Reifenrath, R. and Zimmermann, I. (1973). Surface tension properties oflung alveolar surfactant
obtained by alveolar micropuncture. Resp. Physiol. 19: 369-393.
Rodarte, J. and Fung, Y.C. (1986). Distribution of stresses within the lung. In: Handbook of
Physiology, Sec. 3, The Respiration System Vol. 3, Part 1. (A.P. Fishman, P.T. Macklem, and
J. Mead, eds.). American Physiological Society, Williams and Wilkins, Baltimore, MD.
pp.233-246.
Rosenquist, T.H., Bernick, S., Sobin, S.s., and Fung, Y.C. (1973). The structure of the pulmonary
interalveolar microvascular sheet. Mierovasc. Res. 5: 199-212.
Schiirch, S., Goerke, J., and Clements, J.A. (1976). Direct determination of surface tension in the
lung. Proc. Nat!. Acad. Sci. USA, 73: 4693-4702.
Seguchi, Y., Fung, Y.C., and Ishida, T. (1986). Respiratory dynamics-computer simulation. In:
Frontiers of Biomechanics. (G. Schmid-Schonbein, S. Woo, and B.W. Zweifach, eds.),
Springer-Verlag, New York, pp. 377-391.
Shaw, D.J. (1970). Introduction to Colloid and Surface Chemistry, 2nd. ed. Butterworths, London.
Sobin, S.S., Fung, Y.c., and Tremer, H.M. (1982). The effect of incomplete fixation of elastin on
the appearance of pulmonary alveoli. J. Biomech. Eng. 104: 68-71.
Sobin, S.S., Fung, Y.c., and Tremer, H.M. (1988). Collagen and elastin fibers in human pulmonary
alveolar walls. J. Appl. Physiol. 64(4): 1659-1675.
Spilker, R.L. and Simon, B.R. (eds.) (1989). Computational Methods in Bioengineering. ASME
BED-Vol. 9 Am. Soc. Mech. Eng., New York.
Stamenovic, D. (1986) The mixture of phases and elastic stability of lungs with constant surface
forces. Mathematical Modeling 7: 1071-1082.
Stamenovic, D. and Wilson, T.A. (1985). A strain energy function for lung parenchyma. J.
Biomech. Eng. 107: 81-86.
Takamizawa, K. and Hayashi, K. (1987). Strain energy density function and uniform strain
hypothesis for arterial mechanics. J. Biomech. 20: 7-17.
Takishima, T. and Mead, J. (1972). Tests oCa model of pulmonary elasticity. J. Appl. Physiol. 28:
596-608.
Tao, Z.L. and Fung, Y.C. (1987). Lungs under cyclic compression and expansion. J. Biomeeh.
Eng. 109: 160-162.
Vaishnav, R.N. and Vossoughi, J. (1987). Residual stress and strain in aortic segments. J. Biomech.
20: 235-239.
Valberg, P.A. and Brain, J.D. (1977). Lung surface tension and air space dimensions from multiple
pressure-volume curves. J. Appl. Physiol.: Resp. Environ. Exercises Physiol. 43: 730-738.
Vawter, D.L. (1983). Poisson's ratio and incompressibility. J. Biomech. Eng. lOS: 194-195.
Vawter, D.L., Fung, Y.c. and West, J.B. (1978). Elasticity of excised dog lung parenchyma. J.
Appl. Physiol. 45: 261-269.
References 451

Vawter, D.L., Fung, Y.c., and West, J.B. (1979). Constitutive equation oflung tissue elasticity. J.
Biomech. Eng., Trans. of Amer. Soc. Mech. Engineers 101(1): 38-45.
Vawter, D.L. and Shields, W.H. (1982). Deformation of the lung: the role of interfacial forces. In
Finite Elements in Biomechanics (R.H. Gallagher, et a!., eds.). Wiley, New York, pp. 83-110.
von Neergaard, K. (1929). Neue AutTassungen iiber einen GrundbegritT der Atemmechanik: Die
Retraktionskraft der Lunge, abhiingig von der Oberfliichenspannung in den Alveolen. Z.
Ges. Exp. Med. 66: 373-394.
Waldman, L.K. (1985). In-vivo measurement of regional strains in myocardium. In: Frontiers in
Biomechanics. (G.W. Schmid-Schonbein, S.L. Woo, and B.W. Zweifach, eds.). Springer-
Verlag, New York, pp. 99-116.
Waldman, L.K., Fung, Y.c., and Covell, J.W. (1985). Transmural myocardial deformation in the
canine left ventricle. Circ. Res. 57: 152-163.
Weibel, E.G. (1963) Morphometry of the Human Lung. Academic Press, New York.
West, J.B. and Matthews, F.L. (1972). Stresses, strains, and surface pressures in the lung caused
by its weight. J. Appl. Physiol. 32: 332-345.
Weyl, H. (1952). Symmetry. Princeton Univ. Press, Princeton, N.J.
Wilson, T.A. (1972). A continuum analysis of a two-dimensional mechanical model of the lung
parenchyma. J. Appl. Physiol. 33: 472-478.
Wilson, T.A. (1981). Relations among recoil pressure, surface area and surface tension in the lung.
J. Appl. Physiol.: Resp. Environ. Exercise Physiol. 50: 921-926.
Wilson, T.A. (1982). Surface-tension-surface-area curves calculated from pressure-volume loops.
J. Appl. Physiol: Resp. Environ. Exercise Physiol. 53: 1512-1520.
Wilson, T.A. (1983). Nonuniform lung deformation. J. Appl. Physiol.: Resp. Environ. Exercise
Physiol. 54(6): 1443-1450.
Wilson, T.A. (1986). Solid Mechanics. In Handbook of Physiology, Sec. 3. The Respiratory System.
III. Mechanics of Breathing, Part I (A.P. Fishman, P.T. Macklem, J. Mead, eds.). American
Physiol. Soc. Bethesda, MD.
Wilson, T.A. and Bachofen, H. (1982). A model for the mechanical structure of the alveolar duct.
J. Appl. Physiol.: Resp. Environ. Exercise. Physiol. 52: 1064-1070.
Yen, M.R.T., Fung, Y.c., and Artaud, C. (1987). The incremental elastic moduli of the lungs of
rabbit, cat and man. The 1987 Adv. in Biomech. Amer. Soc. Mech. Engineers, New York,
pp.39-4O.
Yen, M.R.T., Fung, Y.c., Ho, H.H., and Butterman, G. (1986). Speed of stress wave propagation
in lung. J. Appl. Physiol. 61: 701-705.
Yen, R.T., Fung, Y.C., and Bingham, N. (1980). Elasticity of small pulmonary arteries in the cat.
J. Biomech. Eng. 102: 170-177.
Yin, F.C.P. (1981). Ventricular wall stress. Circ. Res. 49: 829-842.
Zeng, Y.J., Yager, D., and Fung, Y.c. (1987). Measurement of the mechanical properties of the
human lung tissue. J. Biomech. Eng. 109: 169-174.
Zupkas, P. (1977). Mathematical analysis of surface tension diagrams of mammalian lung com-
ponents. M.S. Thesis, Univ. of Calif., San Diego.
CHAPTER 12

Strength, Trauma, and Tolerance

12.1 Introduction

There are many reasons why the study of the strength of biological tissues
and organs is important. In the first place any living organism must be strong
enough to withstand the loads imposed on it by its environment and its
activities. The history of evolution is a history of cells forming more efficient
organizations for competition and survival. The shapes of plants and animals
depend largely on the structural materials these organisms can manufacture
and organize into structures of adequate strength. See Currey (1970) and
Wainwright et al. (1976).
Engineering science is concerned with the description and measurement of
the strength of natural or artificial biological materials, and the determination
of their significance. The determination of the failure characteristics of living
tissues and organs is especially complex, because there are many ways a
material can "fail" in biological sense. Besides yielding, plastic deformation,
creep, rupture, fatigue, corrosion, wear, and impact fracture, one has to
consider other kinds of failure. An impact causing severe edema in the lung
can be fatal. A sprain of the ankle can be very painful. A concussion of the
brain has neurological effects. To study the strength of biological materials
one has to correlate clinical observations and pathological lesions with stress
and strain in the tissues.
In this chapter we shall discuss the strength of organs from the point of
view of trauma research. Trauma is a Greek work for wound, meaning an
injury to a living body caused by the application of external force or violence.
Next to heart disease and cancer, trauma is the third greatest killer in the
United States. For people between 15 and 45 years of age, it is the number 1
killer. Minimizing trauma is on everybody's mind. To the medical community

452
12.2 Failure Modes of Materials 453

the problem is critical care and management. To an automotive engineer the

problem is design for safety and protection. To police and paramedics the
problem is quick transportation. To government the problem is legislation.
To parents and teachers the problem is safety education. Everybody has a
role to play. Everybody has something at stake. From the point of view of
reducing the possibility of getting into a traumatic situation in the first place,
it is a problem of culture, of life style, of war and peace, of law and order.
When it gets to the stage when bioengineers, physicians, surgeons, and nurses
can do something about it, it is already toward the end of the line.
Biomechanics is involved in many ofthese stages. Specifically, it is involved
in process causing traumas, in trauma management, in surgical treatment, in
pathophysiology, in recovery, in physical therapy, in rehabilitation, in the
design of vehicles, and in personal protection. A clear understanding of
biomechanics in all these stages will be helpful to minimize trauma as a
national and personal problem.
Trauma is a vast subject. In this chapter we can only provide a survey of
basic principles in trauma research. We shall discuss the types ofloading that
may cause trauma, the failure modes of materials, the strength and tolerance
of organs, and biomechanics in trauma management, healing, recovery, and
rehabilitation. We shall discuss engineering for trauma prevention, taking
passenger aircraft design as a concrete example; and end the chapter with
suggestions for future research.
A tool for trauma research is modeling. What this chapter offers is a sketch
of the basic features that detailed calculations are expected to reveal.

12.2 Failure Modes of Materials

The mechanical properties of biological tissues have some features similar to

those of the familiar engineering materials and other features very different
from them. To study this very complex subject let us consider the metals first.
If rods of metals are pulled in a testing machine at room temperature, typical
load-elongation relationships shown in Fig. 12.2:1 are obtained. The initial
region, appearing as a straight line, is the region in which the law of linear
elasticity holds. The maximum load M is called the ultimate load. At the point
C the specimen breaks.
For a given material, the stress-strain relationships in tension, shear,
compression, bending, and torsion are somewhat different, and the loading
condition should be mentioned in stating the values of the elastic limit, yield
stress, and ultimate stress.
The stress-strain relationship of bone is somewhat similar to that shown
in Fig. 12.2:1c. Typical examples are shown in Fig. 12.2:2. See Yamada (1970)
for a comprehensive set of data. Saha (1982) for dynamic strength of bone.
Soft tissues have very different stress-strain relationships. See some typical
examples in Fig. 12.2:3. In the physiological range the stress generally in-
454 12 Strength, Trauma, and Tolerance

c
"0
o
o
...J

(Elastic response
exaggerated)
p

.\.:~--- €(P~----+l+(:r- Elongation

(0)

c c
"0 "0
o o
o o
...J ...J

Elongation Elangotion
(b) (c)
FIGURE 12.2:1 The stress-strain relationship of several engineering materials.

HUMAN FEMUR (AFTER EVANS 1969)

10'lb/,n' kp/mm'
.-----,.----,,.....--,.---,

12

10 7

~
UJ
a::
l-
V)

° 0
o 0,002 0,004 ° 0,002 0,004 0,006 0,008 0,010 0,012
INCHES PER INCH
STRAIN
FIGURE 12.2:2 The stress-strain relationship of bone.
12.2 Failure Modes of Materials 455

T,,,~,,. 9'~'" I
60

50 m","OOO
~
C~ A
II
40
II
_ _ XOength of body) I'
E
'" --- ~. (width of body)

':
l<. 30
Fy .r:; ( ) . y ) - I ,"
,I
20 If
/1
I ,
I I
10 I I
,,/ /

10
__
O~ ~~~D~_~_~_~-_~--~~~-~--~_~-=;!B%===~I::~~~~_~_~
1.1 1.2 1.3 1.4 1.5 6

FIGURE 12.2:3 The stress-strain relationship of soft tissues.

creases exponentially with increasing strain. At certain strain higher than

physiological the tissue yields and breaks. Hysteresis, creep, and relaxation
exist in both hard and soft tissues. See Fung (1981) for a detailed discussion
of the mechanical properties of soft tissues.
For most tissues the strength depends on the strain rate. The degree of
dependence varies with the tissue: strain rate effect is significant for the bone,
less so for the ligament.
Experimental data on the strength of soft tissues are difficult to obtain. Test
specimens of suitable sizes are often unavailable, or are difficult to keep in in
vivo condition. Clamping of the specimen for strength testing can be a difficult
problem. Furthermore, it is important to realize that the damage that a force
can do to an organ depends not only on the magnitude of that force, but also
on a number of other factors. Consider the following simple experiments (Fig.
12.2:4).
(1) A twine is to be cut by a pair of dull scissors. I have difficulty cutting it
when the twine is relaxed. But if I pull it tight and then cut it, it breaks
easily. Why?
(2) A stalk of fresh celery breaks very easily in bending. An old, dehydrated
one does not. Practice on carrots also!
(3) A balloon is inflated. Another is not inflated but is stretched to a great
length. Prick them with a needle. One explodes. The other does not. Why?
(4) A thin-walled metal tube is filled with a liquid. Strike it on one side.
Sometimes the shell fails on the other side. This is known as "contre
coup". How can this happen?
456 12 Strength, Trauma, and Tolerance

Cutting string
Bending celery Piercing a rubber balloon

Uniaxial stretching

O·• .~
C it)
cSohr.• cirde ..
;
3Mr
2Ebn 3

(a)

Bouncing
ball
Liquid-filled
shell
o

(b)

FIGURE 12.2:4 Several experiments demonstrating that the meaning of the term
"strength" depends on the condition the specimen is in: whether it is subjected to large
initial strain, internal fluid pressure, uniaxial versus biaxial or triaxial loading condi-
tion, or focused elastic waves. See text.
12.2 Failure Modes of Materials 457

(5) Take a small nylon ball, or a pearl, or a ball bearing, throw it onto a hard
surface. It bounces. Throw it onto a thin metal plate such as that used in
the kitchen for baking, it won't bounce.
Think of biological analogs of these experiments. The twine is similar to a
blood vessel, a tendon, or a muscle. The celery is similar to an erectile organ.
The balloon experiment shows the difference between the behavior of a
material under a uniaxial tension and one subjected to biaxial tension. Many
organs of our body are subjected to biaxial tension: pericardium, pulmonary
pleura, interalveolar septa of the lung, a taut skin, a diaphragm, a filled
bladder, etc. A human head impacting on a windshield of a car is not unlike
a tube filled with liquid in the contre coup experiment, or the ball against a
hard surface in the bouncing experiment. The same head impacting on a thin
sheet of metal may have its kinetic energy transferred to the sheet without
bouncing. The kinetic energy is absorbed and the danger of head injury
lessened. These examples show that it is instructive to understand what is
going on in these experiments.
In the first example, the twine can be understood if we postulate that the
fibers in the twine break when the maximum principal tensile stress exceeds
the ultimate stress. If shear is applied to the twine when it is slack, the principal
tensile stress in the twine is numerically equal to the shear stress imposed. See
the Mohr's circle on the lower left part of Fig. 12.2:4(a) which is a graphical
method for determining the principal stresses. On the other hand, if the twine
is pulled taut and then the shear is applied, the principal tensile stress is
numerically equal to the initial tensile stress plus the shear stress (draw
a Mohr's circle for this case) and is, therefore, larger than that in the slack
case. Therefore it is easier to cut the twine when it is taut.
In the second example, the specimen fails by bending. When the specimen
bends, however, half of the specimen is stretched, whereas the other half is
compressed. The fibers in the stretched side are expected to be taut in the fresh
and plump celery, whereas those in the dehydrated specimen are likely to be
slack. Upon bending, the taut fibers in the fresh celery will be stressed more,
the slack fibers in the dehydrated celery will still be slack. The highly stressed
fibers break when the ultimate stress is reached; the slack fibers will not break.
Thus, the difference in the failure characteristics of these two celeries can be
explained in the same way as in the first example. This is an interesting
example. The contrast between a fresh celery and a dehydrated one is not so
different from that of some tissues in vivo and in vitro, with blood perfusion
and without, edematous or normal. The expected difference in the mechanical
properties of these specimens is worth remembering.
The third example is also an interesting one to remember. It shows that
some materials are ductile under uniaxial loading; but become brittle
under biaxial and triaxial tension. The reason for the difference is illustrated
in the sketches given in the lower right corner of the figure. Rubber is a high
polymer of long chain molecules. These molecules are bent and twisted in a
458 12 Strength, Trauma, and Tolerance

complex and random fashion. When the rubber membrane is stretched uni-
axially, some molecules in the direction of stretching become straightened and
take up the load. If a hole is now made in the membrane (as by the needle),
some of those tauted molecules in the direction of stretching will be broken,
but the molecules in other directions remain bent and twisted, the hole remains
a hole, and nothing dramatic happens. Consider, on the other hand, the
situat,ion of the inflated balloon. In this case the membrane is stretched in
every direction. The long-chain molecules in every direction are stretched
straight and' taut. If a hole is made in the middle, the chains in every taut
molecule intersecting the hole will be broken, and an explosion results!
Biological soft tissues are composed of collagen and elastin fibers and other
long chain molecules embedded in ground substances. The fibers and chains
can be stretched when the tissue is under strain. The relevance of the example
is evident.
The fourth example shows what focusing of stress waves can do. The
compression wave in the fluid initiated by the impact moves to the right. The
flexural wave of the metal shell also moves to the right along the curved surface
of the tube wall. If the flexural wave and the compression wave arrive at the
other side simultaneously, a concentration of stress may occur which may
exceed the ultimate stress of the materials and cause fracture on the far side.
This is a mechanism that may occur in head injury.
The fifth experiment illustrates the possibility of transferring energy be-
tween two bodies in impact. When a ball impacts a hard surface, it bounces;
its kinetic energy is transformed first into elastic strain energy in the ball when
the ball is stopped; then the elastic strain energy in the ball is transformed
back into kinetic energy propelling the ball upward. Against a good hard
surface the ball loses little energy in the collision. On the other hand, when
the ball hits a thin plate which is more flexible than the ball, the plate deforms
and sets off elastic waves which propagate away from the point of impact,
carrying kinetic energy with the motion and storing elastic energy in the plate
as it deforms. By this process only a small part of the original kinetic energy
of the ball is transformed into strain energy in the ball. At the time of rebound
the strain energy in the ball is small, therefore the rebound is small.
The design principle for using the sheet metal (or plastic) as shock absorber
is its flexural compliance against a concentrated load. The plate should be
more flexible than the ball (or head).
In examining the questions of strength and tolerance of man to impact
loading, it may be beneficial to remember these simple examples. The magni-
tude of initial stresses, the difference between uniaxial, biaxial, and triaxial
states of stress, and the dynamic effects are important factors determining the
strength of the organs.
There are other factors that may have great effect. The existence or absence
of stress raisers such as notches or microscopic cracks is important. Cracks,
though small, may induce large stress concentration at their ends. A sharp
notch has the same effect. So notches and cracks should be avoided in load
12.2 Failure Modes of Materials 459

bearing members. This is an important principle to remember in surgery and

in manufacturing or installing prosthesis.
The art of predicting the strength of biological tissues from their structure
and ultrastructure is still in the developing phase. Yamada (1970) has pre-
sented collections of strength data on bones, tendons, cartilage, ligaments, skin,
muscles, and some other tissues of man and other animals of different ages
and sexes under static tension, compression, torsion, and bending, dynamic
impact loads, or fatigue oscillations. Standard engineering testing machines
are used to obtain data on bone specimens or soft tissues in simple elongation.
Special testing methods have been developed to test soft tissues in biaxial
loading condition to take care of finite deformation and large strain (Lanir
and Fung, 1974; Fung, 1981, p. 242). A convenient reference to testing methods
is the CRC Handbook edited by Feinberg and Fleming (1978).
To find the strength of tissues such as the skin, a new tearing test was
developed by Schneider (1982). The method was originally used by Rivlin and
Thomas (1953) for testing rubber membrane and adopted by Sharma (1979)
for the testing of blood vessels. A rectangular test specimen is cut as shown
in Fig. 12.2:5. The middle flap is folded over and pulled. The critical value of
the tension at which the specimen is tom at constant velocity is measured.
This method has the advantage that the configuration of the tear remains

FIGURE 12.2:5 Tearing as a method of testing membraneous tissue.

460 12 Strength, Trauma, and Tolerance

self-similar during failure, yielding meaningful data with little interference

from the method of clamping of the specimen in the testing machine.

12.3 Injury and Repair of Organs

For man, the level of injury that one is willing to tolerate may be much lower
than the breaking failure discussed in Sec. 12.2. Often a minor pain is intolera-
ble to some. The physician, surgeon, physiologist, bioengineer, patient, and
lawyer are interested in evidences of injury and repair at the cellular level.
Injury is deviation from the normal. All the knowledge and tools of science,
technology, and medicine must be brought to bear to develop this subject. A
recent publication edited by Woo and Buckwalter (1988) presents a detailed
discussion of injury and repair of the musculoskeletal soft tissues, including
tendon, ligament, bone-tendon and myotendinous junctions, skeletal muscle,
peripheral nerve, peripheral blood vessel, articular cartilage, and meniscus.
Future directions of research are pointed out. The biomechanics of injury and
repair, however, are largely untouched. Evidently, this is a field for future
cultivation.

12.4 Shock Loading and Structural Response

Penetration by bullets and bomb fragments is a well-known cause of trauma.
Strong electromagnetic and heat radiations from nuclear weapons are, of
course, traumatic to the extreme and we hope that these weapons will never
be used again. Control of these things require intelligence of man beyond
technology.
Blunt impacts occur daily and are more controllable. They are our main
concern in the following discussion. Blunt impacts do not penetrate the skin.
They occur in automobile crashes, survivable airplane crashes, falling, col-
liding, boxing, diving, football, or shock waves of explosives. How much
impact a man can tolerate depends on many factors: the strength ofthe shock,
the magnitude of the peak force, the duration of the pulse, and how rapidly
the load is applied. Is it slow enough for the organs to respond quasistatically?
Or fast enough to induce vibration? Or even faster so that the major feature
of the response consists of stress waves in the organs? It depends also on the
initial stress in the organ. For example, if one tenses up, one can induce large
stresses in muscles, bones and joints. This initial stress is superposed on the
response to the shock loading. The sum is what the body has to bear.
In an automobile crash accident, if a passenger did not fasten his seat belt,
then he will fly off as a projectile, and be decelerated when he hits the car. If he
has his seat belt fastened, then he will acquire certain deceleration. Each part
of his body has a different deceleration history. To analyze the stress in any
of his organs, a free-body diagram ofthat organ can be constructed. According
12.4 Shock Loading and Structural Response 461

to D' Alembert's principle, the body can be considered as in a static equilib-

rium with the inertial force (mass x deceleration) applied as an extemalload.
To find the stress in the organ is to find the response of the organ to the
dynamic load. Although the full analysis usually require elaborate mathemati-
cal modeling, major features can be understood through simple examples.
As a simple example, consider an elastic wire of uniform cross-sectional
area A and length L, made of a uniform linear elastic material with a Young's
modulus E (see Fig. 12.4:1). One end of the wire is attached to a rigid ceiling,
the other end has a stopper and hangs free. Now let a load W be applied,
infinitely slowly, to the free end. The wire extends, infinitely slowly, so that
equilibrium is maintained at all times. When the load is fully applied, the free
end extends a distance (). The strain in the wire must be ()/L. The longitudinal
stress u is, therefore, E()/L. A multiplication with the cross-sectional area A
gives the total tension in the wire, AE()/L. Since this must be equal to Win
an equilibrium condition, we see that
() W
AE()/L = W, hence u=E-=- (1)
L A'
Now let us consider a different situation. Let us lift the weight W slightly
above the stopper. At t = 0 we suddenly drop the weight W on the stopper.
The wire extends, but equilibrium cannot be maintained. The free end ac-
celerates downward at first, then it is pulled back by the tension in the wire,
and finally at a certain instant of time t = to the free end becomes stationary,
with a zero velocity and a finite displacement ()'. At this instant, t = to, the
wire and the weight are not in equilibrium; an upward acceleration exists and
the weight starts an upward motion. Eventually the weight vibrates about an
equilibrium position. If we make the assumptions that the strain in the wire

A
L

FIGURE 12.4:1 A suddenly applied load on a wire.

462 12 Strength, Trauma, and Tolerance

is uniform throughout the length of the wire at all times, and that the weight
Wand the stopper are perfectly rigid, then the maximum strain and stress in
the wire are reached at t = to, and can be calculated as follows. The strain at
t = to is [)'jL, the stress is E[)'jL, and the total strain energy in the wire is
[)')2 AL. This strain energy must be equal to the work done by the external
i E (L
load, W[)'. The balance of energy, therefore, yields the equation

1 -
-E
2
([)'L )2 AL = W[)' .
Hence

[)' = 2 WL (2)
AE'
Comparing Eqs. (1) and (2), we see that
[)' = 2[), (1' = 2(1. (3)
Thus the dynamic stress and strain are twice the corresponding static values.
The difference between the two cases examined above arises from the
difference in the work done by the external load. In the first case the equili-
brium is maintained so that the load experienced by the wire is linearly
proportional to the elongation, as shown in Fig. 12.4:2 and the work done

/
)"
1/'"
/
V
IV FIGURE 12.4:2 The load-deflection relationship

~' of a spring when the load is applied slowly. How

slow the loading must be to qualify for such a
"static" response is discussed in the text.
12.5 Vibration and the Amplification Spectrum of Dynamic Structural Response 463

by the external load is equal to the area of the triangle. In the second case the
full load W acts on the wire, and the work performed is equal to the area of
the shaded rectangle in Fig. 12.4:2.
The reasoning given above is so simple that it cannot fail to impress us.
But is the numerical factor 2 infallible? Under what conditions is it valid?
Under what conditions would it be in gross error?
The condition of validity is contained in the statements "we suddenly drop
the weight Won the stopper", and "we assume that the strain in the wire is
uniform throughout the length ofthe wire". If these assumptions are not valid,
the ratio between the maximum dynamic stress and the static str:ess will be
different from 2. In fact, it may be much smaller or larger than 2. This will
become clear in the following sections.

12.5 Vibration and the Amplification Spectrum of Dynamic

Structural Response

The problem discussed in Sec. 12.4 and sketched in Fig. 12.4:1 can be presented
as follows. A mass M is attached to a spring (wire). Under the assumption of
uniform stress distribution in the spring, let the length of the spring be Land
the deflection be u. The tension in the spring is then EAu/L. At time t = 0 a
gravitational acceleration g is suddenly imposed on the system. The equation
of motion of the mass is then
d 2u EA
M - = --u+Mg. (1)
dt 2 L
The initial condition is
du
u = - = 0 when t = O. (2)
dt
The solution of Eq. (1) is
. LM
u = C1 Slllwt + C2 coswt + EA g, (3)

where w is the circular frequency of free vibration

w= J:~. (4)

The initial conditions Eq. (2) are satisfied if

C1 = 0, (5)

Hence
LM
u = EA g (1 - coswt). (6)
464 12 Strength, Trauma, and Tolerance

Thus the deflection history is sinusoidal. The maximum deflection is reached

when wt = nn (n being odd integers), at which the deflection is 2LMg/(EA),
corresponding to a stress 2Mg/A in the spring, This proves that the maximum
dynamic stress in the spring is twice the static value Mg/A. Q.E.D.
This simple example can be easily extended to give many applications. We
know that shock problems facing an engineer do not always arise from
solid-to-solid impact. In most cases the blow reaches an object after being
softened by some intermediate structures. Such are the cases of a passenger
in a train, an instrument in a package, a machine soft-mounted on the ground,
a building subjected to an earthquake.
The ground shock problem can be illustrated by a simple case shown in
Fig. 12.5:1. A mass M is attached to a massless spring, which in turn is built-in
to a "ground". When the ground moves horizontally with a displacement
history s(t), the mass M moves to x(t). If the spring constant (force per unit
displacement) is K, and the damping constant (force per unit velocity) is c,
then the equation of motion of the mass is
Mi + c(i: - s) + K(x - s) = 0, (7)
where x represents the horizontal displacement of the mass, and a dot over x
indicates a differentiation with respect to time. If we let
y=x-s (8)

A. Mass sustained by a massless B. Moss on stationary ground C. Simplified

beam subjected to horizontal subjected to variable Analog
ground acceleration inertia field Systems

Head-neck
m m

x
00-
-s _ Spine

Eyeball-
Spring Constant k
socket

Damping Constant c Heart

Aorta
s
- s(t) Chestwall

Abdomen
mx +c(x -s)+k(x -s)=o With a gravitational field

of acceleration -i; • the Testes

equation of motion of the
my+cy+ky=-ms
mass is:

my+cy +ky = -ms

FIGURE 12.5:1 Ground shock of a cantilevered structure.

12.5 Vibration and the Amplification Spectrum of Dynamic Structural Response 465

represent the displacement of the mass relative to the ground, then Eq. (7)
may be written as
My+cy+Ky= -Ms. (9)
It is convenient to write this equation in the form
ji + 2ewy + w 2 y = -s, (10)

2 K c
(11)
w = M' e=2JKM'

where w is the natural frequency of the system, and e is the ratio of actual
damping to the critical damping of the system.
The solution of Eq. (10) for an arbitrary ground acceleration s(t) and
arbitrary initial conditions
y(O) = Yo, y(O) = Yo when t = 0 (12)
can be written in a closed form:
1
y(t) = yoe-erot cos w~t + -(Yo + yoew)e-erot sin w~t
w

(13)

and mathematically the problem is solved.

The motion expressed by Eq. (13) represents a forced motion during the
interval in which s(t) =1= 0, and a residual oscillation after s(t) transpires. An
engineer is interested in y(t) because it gives information on the dependence
of the response on the structural parameters w, the natural frequency, e, the
damping factor, and s(t), the pulse history. In particular, the maxima and
minima of y(t) as functions of these parameters are of interest. The maximum
response as a function of the natural frequency of the structure is known as
the response spectrum.
Real problems of structural response to ground acceleration differ from the
idealized example considered above by having many (or an infinite number
of) degrees of freedom. But there exists a classical method of generalized
coordinates which enables us to write down the Lagrangian equations of
motion. See Chap. 2. For certain types of damping and forcing functions these
equations can be decoupled by introducing appropriate normal modes offree
vibration as generalized coordinates (Sec. 2.10). In that case for each normal
mode the mathematical problem is the same as that of the single-degree-of-
freedom case.
Example 1. Let the pulse s(t) be a trapezoid as shown in the first panel of
Fig. 12.5:2 (rise from 0 to peak in time t m , constant for a period tm , then decay
in another t m ). The response y(t), calculated from Eq. (13), is shown in the
figure for several values ofthe parameters 2ft m and the damping factor e. Here
466 12 Strength, Trauma, and Tolerance

- "\
.
---- Pulee

-
• • 0.04
..

Reeponee

2ftm = 3.0
2ftm • 1.0
= 0
.. 0

2ftm =
=
2.0
2ftm . 1.0
0
• . 0.04

2ftm = 1. 75
2ftm . 0.5

• ..
0
0

2ftm = 1.5
2ftm = 0.5
= 0
= 0.04

f = Frequency of oscillator, cyclee per eec.

tm = Rise time to peak of pulse, sec.

f = Percent critical damping

FIGURE 12.5:2 The dynamic response of a single degree of freedom elastic oscillator
to a trapezoidal pulse.
12.5 Vibration and the Amplification Spectrum of Dynamic Structural Response 467

tm is the rise time of the pulse, J is the frequency of natural vibration of the
system. The parameter 2ftm is called the rise-time parameter:
2fi = rise time (14)
tm half period of vibration

Example 2. Amplification spectrum. For an impact force, the maximum

static displacement under the maximum absolute value of the force is
[) = max IF(t)1 = max IF(t)1 (15)
K mOJ 2 '

The ratio of the maximum deflection of the spring in a mass-spring system to

[) is defined as the dynamic amplification Jactor or amplification spectrum:

Amp l1'filcation
. _ IYlmax _ max. dynamic displacement
spectrum - ~ - . d' I (16)
u max static ISP acement
If an acceleration pulse is considered, we define
[) = max Is(t)1
(17)
OJ2

and the amplification spectrum as

IYlmax lji + slmax max acc. of mass
(18)
[) Islmax max ground acc. .
The amplification spectrum for the trapezoidal pulse illustrated in Fig. 12.5:2
shows that the maximum response is reached when 2Jtmis in the neighborhood
of 0.5. When 2Jt mis small, the amplification factor increases linearly with 2Jt m.
When 2Jt m is greater than 2, the amplification factor tends to 1.

The force shock amplification spectrum is the same as the

ground acceleration spectrum for sIt) = F(tl/m. The tm for
the pulses are the time at which the peak impact force is
reached in each impact.

eu
II:

if
z
.... ~-~

........
1.0
52
!:!
!i
~ °0~-----+------~Z~----~3~----~4
TIME, SEC 2ft",

FIGURE 12.5:3 The amplification spectrum of a single-degree-of-freedom oscillator to

a set of impact loads recorded from air-plane landings by a variety of airplanes and
airports. e is the damping factor, e = 1 corresponds to critical damping.
468 12 Strength, Trauma, and Tolerance

Example 3. Amplification spectrum of aircraft landing pulses. Figure 12.5:3

shows a set of records of the landing impact force of an airplane and the
amplification spectrum computed for each impact. It is seen that although the
impact force histories vary considerably from one landing to another, the
amplification spectra look much alike. For the purpose of engineering design
and analysis of injury potential, it often suffices to know the mean amplifica-
tion spectrum.

12.6 Impact and Elastic Waves

John Hopkinson (1872) performed an interesting experiment of which the

explanation would help us understand the nature of elastic wave propagation
and its significance in shock and trauma. He tried to measure the strength of
a steel wire by suddenly stretching the wire. As is shown schematically in Fig.
12.6:1, a ball-shaped weight pierced by a hole was threaded on the wire and
dropped down from a known height so that it struck a clamp attached to the
bottom of the wire. For a given weight we expect the existence of a critical
height beyond which the falling weight would break the wire. Using different
weights dropped from different heights, however, Hopkinson obtained the
remarkable result that the minimum height from which a weight had to be
dropped to break the wire was, within certain ranges, almost independent of
the size of the weight!

FIGURE 12.6:1 John Hopkinson's experiment on a weight

falling on a stopper supported by a metal wire. The ex-
x= 0 periment was designed to test the strength of the metal.
12.6 Impact and Elastic Waves 469

Now, when different weights are dropped from a given height, the velocity
reached at a given level is independent of the magnitude of the weight.
Hopkinson's result suggests that in breaking the wire it is the velocity of the
loading that counts. Following this lead, Hopkinson explained his result on
the basis of elastic wave propagation. He knew that in a plane progressive
wave propagating in a homogeneous isotropic elastic medium (see Sec. 5.7),
the stress (J is proportional to the particle velocity v (Eq. 5.7:12):
(J = pcv, (1)
where C = .JEiP is the speed oflongitudinal waves in the wire, p is the density
of the material, E is Young's modulus of elasticity. The stress in the wire,
however, is not largest at the instant of impact at the lower end. The largest
value is reached sometime later when the elastic wave had propagated up and
down the wire a few times. When this largest stress equals the ultimate stress
of the wire, the wire breaks.
When the weight hits the clamp in Hopkinson's test, the end of the wire
acquires a particle velocity Vo which is equal to that of the weight. A steep-
fronted tension wave is generated and propagated up the wire. In the mean-
time, the weight is slowed down by the tension of the wire. The elastic wave,
on reaching the fixed end at the top, is reflected as a tension wave of twice the
intensity ofthe incident wave. The reflected wave is reflected again at the lower
end, and so on. John Hopkinson used a 27-foot long wire and weights ranging
from 7 lbs to 41 lbs; and the absolute maximum tensile stress was reached
near the top after a number of reflections. Bertram Hopkinson (1914), in
repeating his father's experiment, used a smaller weight (lIb) so that the weight
was slowed down faster. G.!. Taylor (1946) showed theoretically that the
maximum tensile stress in B. Hopkinson's experiment occurred at the third
reflection, i.e. the second reflection at the top of the wire when the tensile stress
reached 2.15 pc Vo.
The impact wave analysis is very similar to the arterial pulse wave analysis
presented in Secs. 5.7-5.9. Consider a wire as shown in Fig. 12.6:1. Choose an
axis x along the length of the wire, with the origin 0 located at the lower end.
When the wire is loaded, each particle in the wire is displaced longitudinally
from its original position by a small amount u, which is positive in the direction
of x. Assuming that plane cross sections remain plane, then u is a function
only of x and time t. The strain in the wire is
au (2)
e= ax'
If the wire material obeys Hooke's law, then the axial stress is
au
(J = Ee = E ax' (3)

The equation of motion, Eq. (1) of Sec. 1.7, becomes, in the present case
470 12 Strength, Trauma, and Tolerance

A substitution of Eq. (3) yields the wave equation

02U 1 02U
----=0 (5)
ox 2 c 2 ot 2
with the wave speed

c=fp. (6)

The general solution of Eq. (5) is

u = f(x - ct) - g(x + ct), (7)
where f and 9 are two arbitrary functions. The function u = f(x - ct) repre-
sents a wave propagating in the positive x direction; whereas u = g(x + ct)
represents a wave propagating in the negative x direction. In either case we
have
AU lou v
-=
ox
+--= -+-c
- cot
(8)

with v = au/at denoting the particle velocity. A substitution ofEq. (8) into Eq.
(3) yields the formula
E
(J = ±-v = ±pcv (9)
c
where the - sign applies to a wave propagating in the positive x-direction,
and the + sign applies to a wave in the other direction. This is the equation
quoted at the beginning of this section.
The Hopkinson problem (Fig. 12.6:1) is specified by the initial and boun-
dary conditions
u = 0 at x = L for all t, and when t ~ 0 for all x, (10)
dV
Mdt = A(J - Mg at x = 0, (11)

V = - Vo when t = O. (12)

Here V IS au (x, t) at x
. t h e vaIue 0 f -o-t- = O. T h e mat h ematIca
. I pro bem
l 'IS reduce d

to finding the arbitrary functions f(x - ct) and g(x + ct) so that u(x, t) given
by Eq. (7) satisfies Eqs. (10)-(12).
G.!. Taylor's calculated results about the elastic stress history at the two
ends of the wire in Hopkinson's experiments are shown in Figs. 12.6:2 and
12.6:3. Figure 12.6:2 refers to one of J. Hopkinson's experiments, with the mass
12.6 Impact and Elastic Waves 471

Q)
>
0- 6r-----------------------------,
Mass of wire Lm
ti5 a = Mass of impact weight =M = 0.1
5
Ja- sin ({ii t)
"0
OJ
OJ
0-
Ul
OJ
> 4
OJ ro Stress in wire
.~ ~
~
.1;;
c -5
Cl
.Q c
Ul OJ
C
OJ .....
....J
Ul
I- Ul
ro
~

II
0
:::I
I-
ro
E
°0&---~--~----~--~4----~5--~6~~7

Dmensioniess time. aVL

FIGURE 12.6:2 Elastic waves in a spring induced by a load suddenly applied to a mass
attached to the end ofthe spring. Solution by G.I. Taylor (1946) when the mass ofthe
wire is equal to 10% ofthe mass ofthe impact weight.

~ 3r_--.---~----r_--._--~----r_--,
>
2~--~----~--~i-Lowerend Upper end
~I
,
J (fixed)

-2 -Stress in wire neglecting its mass-+----+----I

o I 4 5 6
Dimensionless time, at/L
FIGURE 12.6:3 Taylor's solution when the mass of the impact weight is comparable to
the mass of the wire.
472 12 Strength, Trauma, and Tolerance

of the wire equal to 10 percent of the mass of the impact weight. The clamp
at the top of the wire and the impact weight itself were assumed to be rigid.
The ordinate represents (JI pc Yo, i.e., the ratio of the stress at any time to the
maximum stress in the wave before the first reflection. The abscissa represents
ctlL so that reflections occur at the top when (ctIL) = 1, 3, and 5, and at the
bottom when (ctIL) = 2, 4, and 6. Figure 12.6:3 refers to B. Hopkinson's
experiments with pALIM = 1.3.
If the density of the wire tends to zero, then the wave speed tends to infinity,
and the tension in the wire becomes uniform and varies harmonically in the
manner of the simple spring-mass system discussed in Sec. 12.4. As a spring-
mass system the stress in the spring can be expressed by the formula

1m
- (J- = - s . ( yOC-
r:.ct)
pcVo Ja L
(13)

where, in Taylor's notation,

pAL length of wire
(14)
oc = M = characteristic length MI(pA)·
In Fig. 12.6:2 it is seen that when oc = 0.1 the stress at each end varies above
and below the stress represented by Eq. (13) by an amount approximately
equal to pcVo, the overall stress history is a disturbed sine curve. On the other
hand, for oc = 1.3, the stress history has lost all resemblance of a sine wave, as
shown in Fig. 12.6:3.
This simple analysis calls attention to the importance of elastic waves,
sound speed, and particle velocity in the creation of trauma.
The speed of sound in several human organs is listed in Table 12.6:1. It is
seen that the cortical bone has a sound speed of 3500 m/sec. This may be
compared with the sound speed of 4800 m/sec in steel, aluminum, copper, etc.
The speed of elastic waves in the lung is of the order of 30-45 m/sec (Yen et
al., 1986). The lung has a complex structure. There are many pathways along
which forces can be transmitted; hence there are many types of sound waves
and sound speeds. Table 12.6:1 lists several sound speeds in the lung. The
speeds given by Yen et al. (1986) were measured from lungs of man, cat, and
rabbit under impact pressure and wall velocity comparable with those induced
by shock waves in an air blast; and the values depend on the transpulmonary
pressure and animal species. The sound speed given by Rice (1983) was
measured by a microphone listening to a sound made by an electric spark.
The speed given by Dunn and Fry (1961) was measured by ultrasound waves.
The ultrasound waves are apparently very different from the waves measured
by Yen et al. and Rice. Note that the speed of sound in the lung is much lower
than both the speed of sound in the gas and in the tissue, in analogy with the
fact that the sound in water containing gas bubbles is much slower than the
sound speeds in pure water and in gas alone.
 12.7 Wave Focusing and Stress Concentration 473

TABLE 12.6:1 Velocity of sound in various tissues, air, and water

Sound speed
Density TPP* mean ± S.D.
Tissue (g/cm 3 ) (kPa) (m/sec) Reference
Muscle 1580 Ludwig (1950), Frucht (1953),
von Gierke (1964)
Fat 1450 Ludwig (1950), Frucht (1953)
Bone 2.0 3500 Clemedson and Jonsson (1962)
Collapsed lung 0.4 650 Dunn and Fry (1961)
(ultrasound)
Collapsed lung 0.8 320 Dunn and Fry (1961)
pneumonitis (ultrasound)
Lung, horse 0.6 25 Rice (1983)
Lung, horse 0.125 70 Rice (1983)
Lung, calf 24-30 Clemedson and Jonsson (1962)
Lung, goat 0 31.4 ± 0.4 Yen et al. (1986)
0.5 33.9 ± 2.3
1.0 36.1 ± 1.9
1.5 46.8 ± 1.8
2.0 64.7 ± 3.9
Lung, rabbit 0 16.5 ± 2.4 Yen et al. (1986)
0.4 28.9 ± 3.3
0.8 31.3 ± 0.9
1.2 35.3 ± 0.8
1.6 36.9 ± 1.7
Air 340 Dunn and Fry (1961)
Water, distilled, ODC 1407 Kaye and Labby (1960)
Air bubbles (45% by 20 Campbell and Pitcher (1958)
vol.) in glycerol
and H 2 O
* TPP = Transpulmonary pressure = airway pressure - pleural pressure.
1 kPa = 103 N/m2 - 10.2 cm H 20.

12.7 Wave Focusing and Stress Concentration

In the preceding three sections, we have analyzed a mass on an elastic wire

three ways. In Sec. 12.4, we obtain a dynamic amplification factor of 2 under
the assumption of uniform stress in the wire. In Sec. 12.5, we obtain a dynamic
amplification spectrum when the stress in the wire is assumed to be uniform
but the loading is not instant but consists of a pulse of specified shape, and
the system is allowed to vibrate. In Sec. 12.6, the assumption of uniform stress
throughout the wire is removed so that elastic waves are revealed. The third
analysis is of course the most general. If the second analysis is generalized to
include all the normal modes of the wire (Chap. 2), the wave features will be
474 12 Strength, Trauma, and Tolerance

revealed. If the third analysis is applied to a loading pulse of specified shape

and continued long enough, the vibration aspect will be seen. Thus, both the
second and third analysis are valuable, and in special situations simplified
versions reveal important aspects of dynamic response. To appreciate the
special situations, note that the loading pulse and the structure have the
following characteristic times:
Loading pulse: rise time t m .
Vibration mode: period, inverse offrequency, 1-1.
Elastic wave: L/c == body length/wave speed.
When tm « L/ c, the simplified analysis of Sec. 12.5 reveals almost the whole
story. When tm » L/c, the vibration analysis, generalized to include all rele-
vant normal modes, solves the problem of maximum stress faster. In the latter
case, the dimensionless variable 2 ftm is an important parameter with regard
to the prediction of maximum response, as demonstrated in Fig. 12.5:3.
The same comment applies to the three-dimensional elastic bodies of more
complex geometry.
In an infinite domain of a linear, homogeneous, isotropic, elastic medium,
there can be longitudinal waves in which the material particles of the medium
move in the direction of propagation of the waves, and transverse waves in
which the material particles move in a direction perpendicular to the direction
of propagation of the waves. The longitudinal waves are also said to be
irrotational waves, or dilatational waves, or compression waves. In seismology,
it is called a P-wave, P signifying "push". The transverse waves are also said
to be rotational waves, or shear waves. In seismology, it is called an S-wave, S
signifying "shake".
The wave speed for a longitudinal wave is denoted by CL ; that for a
transverse wave is denoted by CT' CL and CT are material constants of the
medium. For a linear isotropic elastic medium, they are related to the Young's
modulus E, shear modulus G, Poisson's ratio v, and the density of the material
p, by the equations
E(1 - v)
(1)
p(1 + v)(1 - 2v)'
See e.g., Fung (1977), p. 310.
In a semi-infinite medium bounded by a plane which is free from surface
traction, a plane S-wave incident upon the free surface will be reflected as a
combination of a P-wave and an S-wave. Similarly, an incident P-wave will
be reflected as both P- and S-waves.
If two elastic media which have different wave speeds are joined together
in a plane, then the diffracted waves also consist of both P- and S-waves.
In a semi-infinite body there exists a type of surface wave called the Rayleigh
wave, which propagates at a slower speed then C T , and for which the elastic
displacement decreases exponentially as the distance from the free surface
increases. For a layered medium there exists a Love wave, which is caused by
12.8 Trauma of the Lung Due to Impact Load 475

the interface between the media. These surface waves are dispersive. More
complex wave patterns arise when the interface is curved.
In a finite elastic body, the wave patterns are more complex. In a sphere
there exists spherical waves which converge at the center with increasing
intensity. In a cylinder there exists similarly cylindrical waves which converge
at the center. These are examples of focusing. Focusing is not limited to spheres
and cylinders. A curved surface or interface in an elastic body can cause
focusing and increased intensity of wave motion.
Extending these concepts to man and animals subjected to impact load,
one can visualize the complexity of stress distribution in internal organs.
The stress-strain relationships of the soft tissues of man and animals are
nonlinear. Their elastic moduli increase with increasing stress. This kind of
strain-stiffening material property can lead to shock waves analagous to the
strong shock of supersonic airplane and the surf breakers in water waves along
the sea shore. The basic cause of these shock waves is that particles that have
higher stress also have faster speeds of propagation. Hence the strong stress
is pushed to the wave front, causing a jump condition.
Although a detailed analysis is beyond the scope of this book, the general
features of impact and trauma are not beyond comprehension.
Focusing by curved surfaces is an effective way to concentrate energy into
small regions in a material. If the energy of an impact on a body can be
concentrated into a small region by a mechanism of focusing, the small region
may be first to become endangered. For example, if a man is subjected to a
pressure wave due to explosion, the gas-containing organs, ear, lung and
intestine, are the most susceptible to damage (Secs. 12.8, 12.9). The conver-
gence of stress waves and the reflection of waves on the heart and spine
undoubtedly playa role in this case.

12.8 Trauma of the Lung Due to Impact Load

In the next two sections, we analyze the trauma of the lung in order to illustrate
the application of the wave theory and to discuss the meaning of injury and
tolerance levels.
It is known that when a man or animal is subjected to a shock wave due
to a bomb or industrial explosion, the ear and lung are most prone to damage
(Bowen et aI., 1965; Clemedson, 1956; Clemedson and Jonsson, 1962). Lung
injury is revealed by edema and hemorrhage. Trauma patients of automobile
crash accidents may also suffer lung injury and pulmonary edema.
The following sequence of facts may be noted: (1) In a plane progressive
wave, the stress is equal to the product of the tissue velocity, sound velocity,
and the mass density ofthe tissue, Eq. (12.6:9). (2) If the impact load is applied
rapidly, the induced velocity of the lung tissue can be high. (3) The velocity of
sound in the lung is singularly low among all organs, Table 12.6:1. (4) The
Mach number of impact of the lung, i.e., the ratio ofthe velocity oflung tissue
476 12 Strength, Trauma, and Tolerance

to the sound velocity oflung, can approach 1 or exceed 1, i.e. can be transonic
or supersonic. A supersonic shock wave concentrates disturbance and energy
in a small region behind the wave front. (5) Trauma occurs when stress or
strain exceed certain critical values. (6) High speed impact injury is therefore
expected to be localized. (7) Focusing by a curved wave front may cause further
concentration of damage. Evidences of hemorrhagic injury in the lung are
usually localized and are usually most severe next to the spine, heart, ribs.
These facts suggest the importance of the wave feature of the phenomena.
Clemedson (1956) was the first to state that the initial surface velocity of
the lung is the key parameter with respect to lung injury. Jonsson et al. (1979)
constructed a drop tower to test lung injury based on this concept. Clemedson
and Jonsson (1962) stated that the lethal injury level for rabbit lung lies at a
velocity of around 15 ms- t .
Yen et al. (1988) experimented on a greatly simplified model: They excised
rabbit lung, supported it on a fine soft net (Fig. 12.8:1), perfused it with saline
at isogravimetric condition (i.e. a condition in which the lung weight does not
change with time), impacted it with a shock wave or a light weight pellet,
measured the initial velocity and the maximum displacement at the point of
impact, and continuously monitored the weight of the lung following impact
for many hours. The pulmonary venous pressure was fixed at 3 cm H 2 0, the
airway pressure was held at 10 cm H 2 0, and the pulmonary arterial pressure
ranged from 12.5-15 cm H 2 0.
Two kinds of impactors were used: A shock tube which sends air shock

\i\O

o
0

J?OOO 0
~VALVE

L--J SEQUENCER
TO PRESSURE REGULATOR
AND COMP.RESSED AIR
J 00
0 0
0

l~
0 001:3:3::1.88 ..100
TO o 0
AMPLIFIER, COMPARATOR, AND

---1
AMPLIFIER AND STRIP BULLET HIGH SPEED COUNTER

'"'" rORDER BARREl ... _.- --

TO
-TO~
TO PERFUSION SYSTEM
FORCE
TRANSDUCER

BALANCE

FIGURE 12.8:1 Test set-up used by Yen et al. (1988) to study lung edema due to impact
load. The lung was perfused. Isogravimetric condition must be established before
testing. Change of lung weight after impact was measured.
12.8 Trauma of the Lung Due to Impact Load 477

FREELY SUPPORTED

1
1000
0

'1
990 0 1 hr.
. 0 .5 hr.

510
•
~ 500

50
Z
<i 40
(!)

f- 30
J: () II
(!)
iii 20

• ••
~
10

0
• ••
6 8 10 '2 ,. '6
VELOCITY OF IMPACTING PELLET ml $

FIGURE 12.8:2 Rate of the gain of lung weight (as a percentage of the weight of the
lung at isogravimetric condition before impact) as a function of impact velocity of the
pellet. Pellet weighed 1.49 g with a cross-sectional area of2 cm 2 . From Yen et al. (1988).

waves of prescribed Mach number and overpressure, and a compressed air

gun which shoots a pellet of 1.49g with a cross section of 2 cm 2 . The surface
deflection was measured by a SP 2000 Motion Analyzer System (Spin Physics,
Inc.) at 2000 frames per sec.
Pellet test results are shown in Fig. 12.8:2. The lungs were "freely" sup-
ported on a nylon net. No edema was found if the impacting pellet velocity
was less than 11 ms- l . But edema reached 20% of initial lung weight 1 h after
impact when the impacting pellet velocity was 11.5 ms- l . At an impact velocity
of 13.5 ms- l and above, the lung weight increased at a rapid rate. Then edema
was massive, and the shock loading has done a great damage.
The advantage of the excised-lung approach is a better defined measure-
ment. The disadvantage is that the effects of many other factors are omitted
from consideration, hence it cannot directly yield a tolerance level of the whole
animal.
Lau and Viano (1981), Viano and Lau (1988) found that impulsive injury
primarily of the alveolar region is prevalent at impact velocities above 15 ms- l .
In studying anesthetized intact rabbits, Yen et al. (1988) found that the critical
impact velocity for the initiation of edema is lower for the whole animal than
that for the isolated lung. This is probably due to the more complex geometric
structure of the whole animal, resulting in more complex wave patterns and
stronger stress concentration.
478 12 Strength, Trauma, and Tolerance

Edema is not the only feature of lung trauma. Hemorrhage is another.

There are evidences that hemorrhage is not necessarily bad; it leads to blood
coagulation and then stoppage of flow, cutting off edema. In post-mortem
examination of victims of impact accidents, it was often noted that there were
red marks on the surface of the lung. They are named "rib markings". Since
the rib markings are of red color, they are often considered a sign of hemor-
rhage of the delicate pulmonary capillary blood vessels (Clemedson and
Jonsson, 1962, Frisoli and Cassen, 1950). Yen and Fung (1985) showed,
however, that rib markings often are not marks of hemorrhage; Often they
are marks of atelectasis, i.e., collapsed alveoli, removable by rebreathing or
reinflation.

12.9 Cause of Pulmonary Edema in Trauma

Since soft tissues have good strength in compression, why does a compression
wave cause edema? The hypothesis of Fung et al. (1988) is that tensile and
shear stresses are induced in the alveolar wall on rebound from compression,
and that the maximum principal stress (tensile) or maximum shear stress in
the lung during the dynamic process may exceed critical values for increased
permeability of endothelium and epithelium to small solutes, or even fracture.
Furthermore, small airways may collapse and trap gas in alveoli at a critical
strain, causing traumatic atelectasis. The collapsed airways reopen at a higher
strain after the wave passes, during which the expansion of the trapped gas
will induce additional tension in the alveolar wall.
Increased permeability of the epithelium of the interalveolar septa due to
traumatic stretching is considered to be a factor of great importance in trauma.
Iffluid movement obeys Starling's law, Eq. (8.6:14), then the rate of movement
of fluid per unit area of a membrane is proportional to the difference of static
pressures on the two sides of the membrane minus the osmotic pressure
difference. Edema can be caused by a change of the distributions of the
concentrations of the solutes or the static pressures in such a way that fluid
will move from the interstitium to the alveolar space. For example, if the
epithelium becomes permeable to a certain small solute, then that solute in
the interstitium will cross the epithelium into the alveolar side, increase the
osmotic pressure there, and pull fluid from the interstitium into the alveolus.
The movement of fluid and solutes can be studied by the is 0 gravimetric
method, indicator dilution method, electron microscopy, lymph measure-
ment, etc. (Chap. 9). See Crone and Lassen (1970), Fishman and Hecht (1969),
Giuntini (1971), and Staub (1978). Brigham (1978), Effros et al. (1982), Egan
et al. (1976), Nicolaysen and Hauge (1982) have shown that the epithelium is
less permeable to small solutes than the endothelium, and that the change of
permeability of the epithelium to small solutes is probably the reason for
alveolar edema.
Our hypothesis stated above assumes that it is the tensile stress that does
12.9 Cause of Pulmonary Edema in Trauma 479

the damage. There are two ways to induce tensile strains in alveolar walls.
One is by macroscopic dynamic response of the lung-chest system to the
impact load. The other is the micromechanical response ofthe alveoli to stress
waves. The dynamics of the chest and lung was analyzed by Bowen et al. (1965)
and White et al. (1971) as a single degree-of-freedom elastic shell enclosing a
gas which has a uniform pressure. Chuong (1985) analyzed the stress waves
with the finite-element method and showed that the pressure distribution is
very nonuniform when the lung is SUbjected to a traveling shock wave, and
that in some locations tensile strain with a magnitude comparable with the
absolute value of the maximum initial compressive strain is induced. Super-
posed on this macroscopic dynamic response are the micromechanical re-
sponse of the airways and alveoli, as well as the normal tensile stress due to
inflation. The bronchioles are compressed when the shock wave arrives. If a
critical level of strain is exceeded, the bronchioles will collapse, trapping gas
in the alveoli. After the shock wave passes, the trapped and compressed gas
in the alveoli rebounds, creating tension in the interalveolar septa. The tension
causes damage. This mechanism needs an experimental verification. Our
validating experiments are described below.

Experimental Test of the Hypothesis

To test this hypothesis, Fung et al. (1988) used isolated rabbit lung perfused
with saline at the isogravimetric condition and measured the rate of lung
weight increase due to a transient increase of stretch of the alveolar membrane.
The stretch lasted for 1 min., then the lung was returned to the normal
condition to measure the rate of weight change. The weighing platform is
similar to that shown in Fig. 12.8:1. At the 5 minutes resting period, the airway
pressure was maintained at 10 cm H 2 0, and the lung weight was monitored.
The results for four rabbits are given in Fig. 12.9:1. The rate of increase of
lung weight in the resting period, dw/dt, is expressed as a percent ofthe initial
lung weight per hour, and is plotted against the 1 minute stretching pressure
preceding that period. It is seen that as the lung was increasingly stretched,
the rate of increase of lung weight rises, i.e. edema increases. Hence stretching
the lung promotes edema.
The second experiment demonstrates gas trapping in compressed lung and
the re-expansion of the gas when the compression wave passes.
Tao and Fung (1987) compressed isolated rabbit lungs with trachea open
to the atmosphere and recorded the relationship between the lung volume and
pleural pressure. The lung was hung in a Lucite box, which can be pressurized
to a variable pressure acting on the pleura, PPL' The airway was open to the
atmosphere so that the alveolar gas pressure, PA , is zero. The transpulmonary
pressure, Pt = PA - PPL' was 6.4 cm H 2 0 at the outset. With increasing pleural
pressure they measured the lung volume. They found that for the rabbit lung
a range of pleural pressure exists in which gas will be trapped in the alveoli,
see Fig. 12.9:2. At a critical closing pleural pressure, Pch a limiting lung volume
480 12 Strength, Trauma, and Tolerance

Isogravimetric Condition

CD ":100
=
When Pt 1 OcmH 2 0
CIl .£:
co ...... Capillary Pressure
CD .£:
... -
o Cl 0'-···-
11.4 em H20

:;:
c:.o:; 80 • - - 13.5cm H20
* --- 14.2 em H2 0
.£: Cl
Cl c:
.- ::J
CD...J
;: 60
Cl .~
c:

--
j c:

o 0
40
_CD CD
CIl
co co
ex: CD
20

20 25 30 35 40 45 50 55 60
1 min. Stretching Pressure, Pt =PA -PpL (cm H 2 0)

FIGURE 12.9:1 The rate of increase of the wet lung weight due to edema following
successive steps of increased stretching of rabbit lung. Each stretch lasted 1 min. Each
resting period was 5 min.; during which the transpulmonary pressure was 10 em H 2 0
and the rate of weight increase was measured. Four rabbits; each symbol represents a
rabbit. The initial isogravimetric condition is stated in the inset at top left. The rate of
increase of lung weight is expressed as percent of the initial lung weight per hour. The
regression lines and the initial weight are, from top down,
(1) Y = 4.58x - 104.30, (r = 0.9151), w = 13.30g.
(2) Y = 0.06X2 - 2.04x + 25.75, (r = 0.9882), w = 26.55g.
(3) Y = 0.05X2 - 2.25x + 29.14, (r = 0.9884), w = 23.40g.
(4) Y = 0.03X2 - 1.33x + 16.04, (r = 0.9961), w = 41.50g.

is reached at which the lung behaves like a closed balloon obeying Boyle's law
on further compression. In the rabbit this limiting volume was roughly one-
quarter to one-half of the initial lung volume which was about 60% of the
total lung capacity. On inflating the lung again, there exists another critical
pressure, Pre •op , at which the lung volume begins to increase again. This critical
pressure for reopening is higher than that for closing, and varies with the initial
lung volume, the rate of strain, and the maximum compression imposed on
the lung. When Boyle's law applies, gas is trapped in the alveoli. In the case
12.9 Cause of Pulmonary Edema in Trauma 481

Pleural Pressure, PPL (cm H2 0 )

12 8 4 0 -4 -8 -12 -16 -20
10
PA - 0 cm H2 0
Pto = 6.4 cm H2 0
o
,..... max(PPL)1 = 10 cm H2 0
E
..... max(PPL)2 = 245cm H 2 0
~ -10 Vo = 58 ml at point S
oj
E
r = 8.9 mil sec
.ao -20
-
>
o
Q)
g' -30

I-=~-~---.-"-,,.-,,-...//
co
J:

PC..
U
_40t--_ _ _ _

-50~J-~~~~-L~~~~~--~~~~~~~~
-12 -8 -4 0 4 8 12 16 20
Transpulmonary Pressure, Pt = PA - PPL (cm H2 0 )

FIGURE 12.9:2 The pressure-volume relationship of rabbit lung with the regime of
small and negative transpulmonary pressure emphasized. The airway pressure was
atmospheric (zero). At the starting point, S, the transpulmonary pressure was 6.4 cm
H 20, and volume was 58 ml. The rate of volume reduction was 8.9 mljsec. In
application, these curves may be read as a stress-strain relationship of the lung because
the tissue stress is equal to the transpulmonary pressure (1 em H 20 = 98 N/m2), and
the change of volume divided by initial volume is the volumetric strain. From Tao and
Fung (1987).

shown in Fig. 12.9:2, the lung was compressed to Pr = -10 em H 2 0 in the

first cycle (solid curve), but to a maximum of Pt = - 245 cm H 2 0 in the second
cycle (dotted curve). At the point P' the slope of the PV curve in expansion is
equal to that in compression at the same volume.
If the strain rate changes, the pressure at the inflection point, P;nf, and the
closing pressure, Pclo remain relatively unchanged, but the reopening pressure,
Pre -op , and the characteristic pressure P', where the PV curve bends upward
sharped on reinflation, will change to the extent of several em H 2 0. If the lung
is compressed to a much higher pressure, the reopening pressure is increased.
If the trans pulmonary pressure remains lower than the reopening pressure,
then the collapsed airways will not be reopened and traumatic atelectasis
results. These results show that gas trapping in the compressed lung is a reality.
One should note that the gas trapping mechanism is related to the lung
482 12 Strength, Trauma, and Tolerance

strain only. PA or PPL alone does not cause the alveoli or small airways to
collapse. It is the transpulmonary pressure PA - PPL that is relevant.
Fung et al. (1988) also presented a theoretical analysis of the response of a
group of alveoli with trapped gas to a tension or compression wave. It was
shown that when a traveling stress wave arrives and passes, the alveoli
respond and moves past the equilibrium condition by the inertial force; and
an oscillation follows. In the oscillation the alveoli contracts and expands.
In expansion tensile stress' is generated in the wall of the alveoli. These facts
support the hypothesis that lung trauma is caused by overstretching the
alveolar membrane.
Clinical and experimental studies of gas trapping in the lung have been
reviewed by Anthonisen (1977), Bates et aI., (1971), and Hoppin and Hildebrandt
(1977) in relation to the maximal expiratory flow phenomena (Sec. 7.6).
Kooyman (1981) has shown that diving animals such as Weddell seal have
cartilage in their bronchioles all the way to the alveoli, presumably to prevent
the closure of the bronchioles before the closure of the alveoli. Therefore in
deep sea diving nitrogen will not be trapped in the seal lung. This may explain
why seal do not get bends or decompression sickness whereas human divers
do in deep sea diving. In man the trapped nitrogen will be dissolved in blood
under high pressure, and evolve as bubbles when the pressure returns to
normal. These observations are not directly related to trauma. But according
to our reasoning, we anticipate that a Weddell seal has a better tolerance to
impact trauma than man.
These theories and experiments, taken together, show the reasonableness of
the expectation that pulmonary edema due to traumatic impact is related to
the velocity of the surface of the lung induced by the impact.

12.10 Tolerance of Organs to Impact Loads

In the 1950's and 60's an audacious group of people in the United States
undertook a remarkable research program to determine the tolerance of man
to body acceleration. They built rocket sleds on long tracks in Western deserts.
Col. John Paul Stapp and others tested themselves to the limits of their
physical tolerance. They recorded their observations and collected data on
some primates and animals. The results were extremely helpful to the develop-
ment of aeronautics and astronautics. In the meantime, people concerned with
highway safety and automobile design began their research on the tolerance
of various organs to impact load. They realized that only through such
quantitative research can progress be made. In the following, we present some
data on human tolerance to whole-body acceleration, head impact, and spinal
injury. Data on other organs are not summarized, but some leading references
are given.
Human tolerance data are difficult to obtain. Data from accident investiga-
tions are often clouded. Use of volunteers is expensive and difficult. Use of
12.10 Tolerance of Organs to Impact Loads 483

cadavers or animal models requires extrapolation and has questions in inter-

pretation. Every piece of information in the literature came from some special
circumstances and cannot be used indiscriminately.

Whole-Body Acceleration Tolerance

Approximate tolerance limits for well-restrained sitting human subjects have
been summarized by Eiband (1959), Roth et al. (1968), and Stapp (1957, 1961,
1970). Unfortunately, evaluated test data are available only in terms of ideal-
ized trapezoidal forcing functions with peak acceleration, duration, and rate
of onset as the only parameters evaluated.
Tolerance to spineward acceleration (eyeballs-out, - Gx ) as a function
of magnitude and duration of impulse is illustrated in Fig. 12.10:1. For
sternumward acceleration (eyeballs-in, + GJ, the tolerance limit is similar, but
can be higher or lower depending on how well the head is supported. In the
longitudinal axis, tolerance to head ward acceleration (+ Gz ) (Fig. 12.10:2)
exceeds tolerance to footward acceleration (eyeballs-up, - Gz ) by a small
margin. But note the great difference between human tolerances to Gx and Gz :
some volunteers have withstood 45 G with duration less than 0.044 sec or

200
I I t"~~l r I 1 I II I I I II I I I I

<!)
~~~~~'~~iA:r:e:a
.
100
Q) Area of -- of severe injury I
moderate injury ~
0 60 ~. ~
1: II

- ~~~~~~~~==~~~~~~~~~
1"..:::~~=-~~~;-rr~-------;
Q)
> 40 IIII1I ....
~.4
, ~4
,.". HH+l~~~~~-*~ ..~,~T,~.~--'
e Area of voluntary human exposure '. , '. ;;}.\".;:f.~~ ,-
e: 20
e (uninjured. undebililaled) '. ~ ""~
+=:
«S
~
Q) 10
Q3
0 I- -
0
«S I-
G -
E
4 I- -
V~'3 --
l-
o Human ~,
e
~

:t:: c Hog ..,.

e: l-
10 t\ 12
::> .. Chimpanzee
1 J J J j llJJ l I
0,001 0.004 0.01 0.04 0.1 0.4 2

<t2-t 1), Duration of uniform acceleration, sec

FIGURE 12.10:1 Duration and magnitude of spineward acceleration endured by vari-

ous subjects. Survivable exposures. Maximal body support used in all cases. From
Eiband (1959), by permission.
 484 12 Strength, Trauma, and Tolerance

200
I .1 11111 I I I I 1111 I
I
). ....
(!' 100
a) , Area of severe injury
o 60 ,

-
:E
Q) I+I----I--+..:.~
> 40
II I I I ....
o Area of moderate injury
c::: 20
o
:;:
I 11111 I I I
I , I lo"
«J I '" I
~
10 . Area of vountary human exposures
j1
(I) 8~ (uninjured , undebilaled) ,
o
o 6 f--
.) " G

•
r-- ..
«J
4 >-

-
o Human ...~~
E -
~
C Hog Acc
o
~

" Chimpanzee
'2 2- 10 I,
:l
1
0.004
I III I 0.1
I I I 1L1
0.001 0.01 0.04 0.4
Duration of uniform acceleration, (t1-t2)' sec
FIGURE 12.10:2 Duration and magnitude of head ward acceleration endured by various
subjects. From Eiband (1959), by permission.

No shock No shock
Definite shock signs signs · (conj, hemorrhage)
40r--------------,~---

c::: 20
~

.2
.-
«J
~

~ 10
Q)
0
0
«J 6

t 1 - to, Time, sec.

FIGURE 12.10:3 Initial rate of change of spineward acceleration endured by various
subjects. The 500-600 Gis curve were obtained by human volunteers fully strapped.
The 1060 Gis curve was obtained from Chimpanzee, 1370 Gis curve was from human,
3400 Gis curve was from Chimpanzee, From Eiband (1959), by permission.
12.10 Tolerance of Organs to Impact Loads 485

25 G at duration 0.2 sec in eyeballs-out accelerations; but in headward,

eyeballs-up accelerations the tolerance limit is only 15 G for a duration of
0.1 sec. All the data quoted in these figures are for well-restrained young male
adult subjects in top physical condition, with a variety of full-torso and head
restraints, not just a lap belt. It is expected that the tolerance limits would be
much lower if the subjects were not so well restrained.
The rate of onset of the applied force also has a definite effect on human
tolerance. Under some impact conditions, the rate of onset appears to be a
determining factor, as indicated in Fig. 12.10:3. The rate effect can be under-
stood from the point of view of the dynamic amplification factor and elastic
waves, as was discussed earlier in Sees. 12.5 and 12.6.
Tolerance to lateral (Gy ) accelerations seems lower (9 G) for a duration of
0.1 sec. Other data are summarized by Stapp (1970).
A variety offactors are involved for human volunteers in determining where
the tolerance limits lie. A variety of trauma are quoted in the area of "severe
injury" in these figures. There exists a wide corridor of uncertainty.

Tolerance Level of Individual Organs

More definitive tolerance information can be obtained from experiments
designed to study individual organs. Then the injury level can be correlated
with the input acceleration pulse.
A certain amount of data exists with respect to the fracture of the cranium,
deformation and fracture of facial bones, injuries to the neck, thorax,
abdomen, and arms and legs. A general reference is the book edited by Nahum
and Melvin (1985). Additional data on tolerance level can be found in the
following, which contain further references. For cranium, Hodgson and
Thomas (1972). For thorax, Lau and Viano (1986), Neathery (1974), Stalnaker
et al. (1973), Viano and Lau (1987). For abdomen, Mertz and Kroell (1970),
Stalnaker et al. (1972). For liver and kidney, Melvin et al. (1973).
In the following we discuss two topics: head and spinal injuries.

Head Injury
Head injury is the most serious frequent trauma of automobile accidents. It
is also the most frequent cause of death of passengers in military aircraft in
survivable crashes. Table 12.10:1 lists some data from the U.S. Army Safety
Center. It is seen that next to the head, chest injuries are the major cause of
death. The frequency of serious vertebral injuries is lower for light fixed-wing
aircraft than it is for helicopters, implying that the vertical component of
impact load experienced by the occupants of helicopters during impact is
larger than that in light, fixed-wing aircraft. Rollover of helicopter is a major
cause of fatalities. The British Royal Air Force experience was similar (Hill,
1978).
The brain can be injured by fracture, impingement, excessive acceleration,
486 12 Strength, Trauma, and Tolerance

TABLE 12.10: 1 Frequency of injuries to each body part as percentages of total injuries
(U.S. Army Aircraft, 1971 through 1976, total injuries in parentheses). Data from
Laananen (1980)
Major and fatal Fatal injuries
injuries combined only
Light Light
Helicopters fixed-wing Helicopters fixed-wing
(1,114) (104) (403) (53)
Head 19.7 19.2 31.5 30.2
Face 9.4 14.4 5.0 5.7
Neck 2.6 0.0 2.7 0.0
Arms, hands 12.1 11.5 7.7 5.7
Thorax 12.5 19.2 21.6 28.3
Abdomen 7.1 5.8 11.7 9.4
Pelvis 3.0 1.9 1.0 0.0
Spine 16.5 12.5 6.5 9.4
Legs, feet 17.1 15.4 12.4 11.3

high localized pressure or tensile stress, high localized shear stress and strain,
and cavitation in high-tension regions. The regions where the maximum
normal stress occurs are usually different from where the maximum shear
stress occurs; and they are affected significantly by the flow through foramen
magnum (opening at the base of the skull) during impact. The brain tissue can
be contused and blood vessels ruptured.
A trauma most widely studied in brain concussion, which is defined as a
clinical syndrome characterized by immediate transient impairment of neural
function, such as loss of consciousness, and disturbances of vision and eq uili-
brium due to mechanical forces. Normally, concussion does not cause per-
manent damage. It is the first functional impairment of the brain to occur as
the severity of head impact increases. It is reproducible in experimental
animals.
Concussion has been studied with respect to rotational acceleration
(Holbourne, 1943, Gurdjian et aI., 1955, Gennarelli et aI., 1971, and Hirsch et
aI., 1970), translational acceleration (Lissner et aI., 1960), and flexion-extension
of the upper cervical cord during motion of the head-neck junction. Hirsch et
ai. (1970) have attempted to establish injury criteria and tolerance levels for
rotational acceleration on the basis of experiments on monkeys. Lissner et ai.
(1960), Gurdjian et ai. (1953, 1955) proposed a tolerance specification for
translational acceleration on the basis of experiments on cadavers and ani-
mals. Their proposal is known as the Wayne State University Concussion
Tolerance Curve. As it is shown in Fig. 12.10:4, the ordinate is the "effective"
acceleration (which is an average front-to-back acceleration of the skull
measured at the occipital bone over a "duration" T) for impacts of the
12.10 Tolerance of Organs to Impact Loads 487

~
600
~I
(9
500 ,,
§ 400 1 - r--
\
'';::;

....co Dangerous to life

Q)
300 --
\
Q)
()
()
co
Q) 200
>

~
'';::;
()
Q)

BJ 100
Not ............ .!

o
dangerous to life
I I :
o 2 4 6 8 10 12
'f
100
Time duration of effective acceleration, msec
FIGURE 12.10:4 The "Wayne State" tolerance curve for the human brain in forehead
impacts against plane, unyielding surfaces.

forehead against a plane, unyielding surface; the abscissa is the duration of

the effective part ofthe pulse. The curve was derived from clinical observations
of skull fracture, cadaver experiments with the duration of pulse in the 2-6
msec range, animal experiments with pulse duration in the 6-10 msec
range, and long-duration human volunteer experiments by Stapp (1957), Figs.
12.10:1 and 12.10:3. Considerable exploration of observed data was involved.
Gadd (1966) showed that the curve shown in Fig. 2.10:4 can be represented
fairly well in the 2.5 to 50 msec range by the equation
{j2.ST = const. (1)
(but badly elsewhere). Here (jis the "effective" or "average" acceleration in the
"duration" T. Determination of the duration T was troublesome, so Gadd
(1966) replaced Eq. (1) with the following:

LX) [a(t)J2.S dt = a constant called the Severity Index. (2)

If the acceleration is given in units ofthe earth's gravitational acceleration, G,

and the time in seconds, then a severity index of 1000 is recommended by the
Society of Automotive Engineers as a criterion of head injury for car design
with respect to frontal impacts (Gadd, 1966).
Versace (1971) pointed out that the upper limit of the integral in Eq. (2)
488 12 Strength, Trauma, and Tolerance

causes difficulty if the acceleration pulse has a long tail which is hard to
measure accurately and is probably unimportant. To remedy the situation,
an arbitrary measure called Head Injury Criterion (HIC) was devised:

HIC = Max {_1_

t z - tl
If2 [a(t)]Z.5 dt} (t z -
f.
ttl < 1000, (3)

t 1 , t z are time in seconds, (t 1 < t z ). The quantity on the left-hand side of the
inequality sign is a function of tl and t z . The criterion calls for varying t 1 , t z
to obtain a maximum of the quantity in the braces. Hodgson and Thomas
(1972) have shown that t z - tl must be less than 15 msec in order to pose a
concussion hazard, even if the HIC value exceeds 1000.
Similar investigation has been done with regard to lateral impact of the
head. Nahum et a!. (1980) measured the pressure in cadaver skulls when they
were subjected to lateral loads. A contracoup phenomenon was observed as
in frontal blows. Stalnaker et al. (1973) reported a side-impact threshhold of
head injury at a peak translational head acceleration of 76 G with a pulse
duration of 20 msec. On the other hand, in a rocket sled test, Stapp sustained
without injury an acceleration pulse of Severity Index 1500 "at the head
(without impact with a solid surface) while 45 G was measured on the seat.
Ommaya (1968), Hirsch et a!. (1970) have shown that head injury occurs
much more readily if the head is allowed to rotate. The brain in the skull can
tolerate considerable frontal or side impact if rotation of the brain relative to
the skull does not occur. On the other hand, if rotation does occur, the same
impact that was tolerable for translational motion may cause severe injury.
Numerous other criteria have been proposed (see King's review, 1975).
Some are approximations of an approximation. As experimental data ac-
cumulate, they appear to be less and less justifiable. In practical applications,
their apparent simplicity is superficial, because an engineer has to predict the
acceleration pulse a(t) in the first place. This is often done by mathematical
modeling. There is a trend toward mathematical modeling of the brain (see
Goldsmith, 1972; Ward et a!., 1980) A future task of trauma research is to
correlate the clinical syndromes and pathological lesions with the stress and
strain in the brain, not merely with the deceleration of the head.

Spinal Injury
As it is shown in Table 12.10:2, spinal injuries are the most prevalent injuries
among the crew and passengers of modern civil air transport accidents. These
injuries may range from fatalities to minor complaints. Often the complaints
are not due to broken bones or burst discs. A disturbance sufficient to pinch
the nerves in the spine may elicit pain in the lower back, limbs, or other parts
of the body.
The load acting at any point of the backbone during an impact does not
simply arise from the inertial force of the body, but also from the reaction of
the muscles and ligaments attached to the spine, and the variable pressure in
12.10 Tolerance of Organs to Impact Loads 489

TABLE 12.10:2 Impact injuries to crew and passengers in civil air transport accidents
in the 1970-78 period in the United States mentioned in Chandler et al.'s (1980) report*
No. of individuals Percentage**

Spine, vertebra, neck 80 48%

Lower extremities 49 30%
Head 11 6.6%
Ribs 11 6.6%
Other bone 5 3%
Shoulder, clavicle, hand 5 3%
Hip 2 1.2%
Kidney 1 0.6%
Spleen 1 0.6%
Intestine 1 0.6%
Unclassified (contusions, head 232
injury, abrasions, lacerations,
ribs,limbs, vertebra, pelvis,
arms legs)
* Data are from 30 major "survivable" accidents, involving Douglas DC8, DC9, DC10, Boeing
707, 727, 737, Martin 404, Convair 340, 440, Fairchild FH227, F27B, Lockheed LI011, 188A.
Some persons who suffered multiple injuries are listed in more than one category.
** Percentages are calculated under the assumption that the injuries included in the "unclassified"
category are distributed in the same ratio as those itemized above.

the abdomen caused by the reaction of the abdominal muscles to the impact.
The loads in the vertebrae and discs are usually much larger than those
required by static equilibrium against external load alone, because there exists
a system of redundant self-equilibrating forces in the muscles which act on the
spine. In Chapter 1, Table 1.4:1, we have shown the results obtained by
Nachemson and Elfstrom (1970) on the pressure in the lumbar disc (the nucleus
pulposus of L3) of the spine. Note how large the load in the lumbar spine is
when one lifts weight the "wrong way". If one lifts a 20 kg weight with back
straight and knees bent, the load acting in the lumbar disc (L3) is 185 kg. If,
however, one lifts the same 20 kg weight with knees straight and back bent,
the load on the lumbar disc becomes as high as 390 kg. A similar difference
can be expected in the crash impact situation. If a vertical impact load is
applied on a spine that is kept straight, the stress in the spine would be much
lower than that acting on a spine that is bent. In other words, a bent spine is
in no position to resist large loads.
The spine is a very complex structure. Spinal injury is a very complex
problem. Tolerance data is woefully lacking. Many people complain of low
back pain in normal life, suggesting that people's tolerance level to spinal load
may be quite low.
In the case of forced landing of an air transport, some general observation
can be made from a biomechanical point of view. For a passenger sitting in
an airplane subjected to verticle deceleration, a straight spine is estimated to
490 12 Strength, Trauma, and Tolerance

be 2 to 5 times stronger than a bent spine (Kazarian and Graves, 1977;

Chandler and Trout, 1979). The reason is as follows: If the spine is straight
and the inertia force is parallel to the spine, then the bending moment which
tends to bend the spine forward is relatively small. This moment is resisted, by
the back muscles, and in the lower thoracic and lumbar region, also by the
abdominal pressure (Fung, 1977, p. 29). In contrast, if the person sits with a
bent spine, then the vertical inertial force will create a larger bending moment
on the spine because of the larger moment arm. If the person bends forward
with arms hugging the knee as some airlines recommend, then the bending
moment in the waist region tends to bend the spine backward. In this position
the abdominal pressure and the back muscle cannot help, and the bending
moment would have to be resisted by compression and shear between the
superior and inferior articular processes of successive lumbar vertebrae (see,
e.g., Jacob et al. (1982, p. 133) and Fig. 1.4:2, 1.4:3). This may cause slip, undue
deformation, or dislocation, which endangers the integrity of the spinal cord.
Hence a straight spine is stronger.
As the proverb says: An ounce of prevention is worth a pound of cure. A
good design should put a person in the strongest configuration. In this respect,
the rearward-facing seating in an airplane looks very attractive (Fung, 1982).
Support of spine can be obtained easily in a forward crash if the seats face
rearward, and are designed to be strong enough to survive crashes without
collapsing. The rearward facing seats should be high enough to support the
head to avoid whiplash in a plane crash. In a forward crash the deceleration
of the plane will keep the passenger's spine straight against a rearward facing
seat, support his head naturally and firmly, keep his arms and legs in position,
not to fly off and injure themselves. Seated rearward, children and pregnant
women can be supported much more easily compared with wearing conven-
tional seat belts in the forward facing arrangement. The only requirement is
that the seats must be designed to take the load.
Since today's airplanes are not so designed, and since spinal injuries incur
in many other human activities, it is necessary to study the problem closely.
For reviews of current state-of-the-art, see Jacobs and Ghista (1982), Liu
(1980), Panjabi and White (1982), Saha (1982), and Sonnerup (1982).

12.11 Biomechanical Modeling

In assessing human tolerance to impact loads, and in the design of vehicles

for crashworthiness, it is necessary to calculate the stress and strain at specific
points in various organs, and this is best done by mathematical modeling.
Through mathematical modeling, one can connect pieces of information on
anatomy, physiology, and clinical observations with people, vehicle, and
accident. A validated model can then become a foundation of engineering.
Biomechanical modeling is in its infancy. Vigorous development is sure to
come. The trend is toward dynamics of three-dimensional bodies capable of
Problems 491

revealing the vibrational amplification and wave propagation features dis-

cussed in Sections 12.4-12.6. Much closer study of the physiological, patho-
logical, and clinical aspects will be done.
A sampling of the huge literature is given below: A general review is given
by King (1984). Head injury is reviewed by Goldsmith (1972), Ward et al.
(1980). Spinal injury is reviewed by Huston and Perrone (1980). The effect of
wearing helmets was studied by Huston and Sears (1979). Intervertebral discs
were modeled by Kulak et al. (1976). Spinal load was analyzed by Schultz and
Anderson (1981), Koogle et al. (1979), Privitzer and Belytschko (1980), and
validated by Kazarian and Graves (1977), Engin (1979). Bones have been
modeled by Brown et al. (1980), Piziali et al. (1976), Hayes et al. (1978), Orne
and Young (1976), Viano et al. (1976).
Chest models were developed by Roberts and Chen (1970), Sundaram and
Geng (1977). Chaffin (1969), Kane and Scher (1970), Passerello and Huston
(1971) modeled the astronauts, swimming and kicking, and human attitude
control. Muskian and Nash (1974), Fleck (1975) modeled a seated man.
And then there are models of heart, lung, blood vessels, muscles (Chapters
6, 11) and single cells (see References in Ch. 4), as well as astronauts in
spaceship (Thomson and Fung, 1965).

12.12 Engineering for Trauma Prevention

A bioengineer can help to make this world a safer place to live. Understanding
trauma is an important step. Improving the design of automobiles, aircraft,
factories, work places, and sports equipment, etc. to make them intrinsically
better for people is the next step. Search, develop, and promote safer and better
work habits, sport techniques, general culture is our responsibility. Manu-
facturing and marketing safer and better products, and teaching the public
about healthier ways of life is a further step.
In trauma research, the application of biomechanics is not limited to the
engineering aspects. It must be extended to all aspects of emergency handling,
patient evaluation, testing, treatment, healing, recovery, and rehabilitation. A
deeper understanding of tissue engineering, of growth and resorption is neces-
sary. A broad horizon is in front of us.
Trauma is a world problem. Trauma research is expensive. Its objective is
to benefit all mankind. It is a fitting topic for international cooperation.

Problems
12.1 In testing a rectangular specimen of cortical bone in bending, it was found that
when the bending moment reached a certain limit, microcracks were formed on
the tensile side. At a somewhat greater bending moment, a few cracks appeared
on the compressive side while many more microcracks formed on the tensile
side. The cracks on the tensile side were perpendicular to the beam axis whereas
492 12 Strength, Trauma, and Tolerance

the cracks on the compressive side were inclined at about 45°. Based on a
consideration of principal stresses and maximum shear (use Mohr's circle if you
wish, see Fung, 1977, Chap. 4), explain what do these experimental results reveal
about the relative magnitude of fracture strength of the bone in tension, com-
pression, and shear?
Given the dimensions of the beam, derive formulas expressing the strength
of the bone in tension, shear and compression.
12.2 In a torsion test of a long bone, the bone broke along a line at approximately
45° to the torsion axis. What does this tell about the strength of the bone with
respect to shear, tension, and compression.
Given the dimensions of the shaft, derive formulas expressing the strength
of the bone.
12.3 A bone subjected to a loading applied very slowly breaks into two pieces. A
load of similar magnitude applied impulsively breaks the bone into many small
pieces. How do you explain this?
12.4 Consider Hopkinson's experiment, Fig. 12.4:1. The mass M has a velocity Vo
when it hits the stopper. As the elastic wave goes up the wire, a force Au = Apcv
acts on the mass. The equation of motion of the mass is, therefore,
dv
M-=Apcv.
dt
Solve this equation with the initial condition v = Vo at t = O. With this result,
discuss the waves in the wire.
12.5 Show that the equations of motion of an elastic oscillator subjected to a force
acting on the mass and one subjected to a ground acceleration are similar.
12.6 Derive wave equation in spherical polar coordinates. Find a solution symmetric
with respect to the origin. Discuss the phenomenon offocusing (Cf.: S. Temkin,
Elements of Acoustics, Wiley, New York, 1981, Ch. 4).
12.7 Design a program to study the tolerance levels ofthe brain in response to impact
load. List possible impairment of functions, clinical symptoms, and the cor-
responding lesions. Discuss how the lesions can be observed and correlated to
the impact load; what should be measured in your proposed experiment; and
how.
12.8 Design a program to study the tolerance levels ofthe spine to various kinds of
injuries that may arise in people of different sex, age, and occupation.
12.9 Verify the statements made in Sec. 12.2 with regard to experiments 1 and 2
through· a calculation of the principal stresses. Set up equations to analyze
experiments 4 and 5.
12.10 Assessment of gait may be helped by the use of electromyograms of major
muscles during walking. Design an equipment for use on patients. (Cf.: R. Shiavi
et aI., J. Rehabilitation Res. and Dev. 24: 13-30, 1987.)
12.11 High pressures in the human hip joint. W.A. Hodge et al. (Proc. National Acad. Sci.
83: 2879-2883, 1986) reported that the local pressure measured on the opposing
References 493

layers of cartilage of the human joint can be as high as 18 MPa (~ 180 atm),
which was obtained when a 73-year old woman (68 kg) was rising from the
sitting to the standing position. In normal level walking, the maximum pressure
was 4 MPa. Records show also that the local joint pressures were only 5%
greater with the use of a cane than with a crutch. Formulate a mathematical
model to analyze these movements.
12.12 In a head-on car crash from a speed of 60 mph, compute the impulse on the
heart, aortic arch, lung, head, neck, and thoracic and lumbar spines. Explain
clearly all the assumptions used in your calculation.

References
Anthonisen, N.R. (1977). Closing volume. In: Regional Differences in the Lung, (J.B. West, ed.),
Academic Press, New York, pp. 451-482.
Bates, D.V., Macklem, P.T., and Christie, R.V. (1971). Respiratory Function in Disease, W.B.
Saunders, Philadelphia.
Belytschko, T., Kulak, R.F., and Schultz, A.B. (1974). Finite element stress analysis of an inter-
vertebral disc. J. Biochem. 7: 277-285.
Bowen, I. G., Holladay, A., Fletcher, E.R., Richmond, D.R., and White, e.S. (1965). A fluid-
mechanical model of the thoraco-abdominaI system with applications to blast biology.
Report No. DASA-1675, Lovelace Foundation for Medical Education and Research,
Albuquerque, New Mexico.
Brigham, K.L. (1978). Lung edema due to increased vascular permeability. In: Lung Water and
Solute Exchange (N. Staub, ed.), Marcel-Dekker, New York, pp. 235-276.
Brown, T.D., Way, M.E., and Ferguson, Jr., A.B. (1980). Stress transmission anomalies in femoral
heads altered by asectic necrosis. J. Biochem. 13: 687-699.
Campbell, I.J. and Pitcher, A.S. (1958). Shock wave in a liquid containing gas bubbles. Proc. Roy.
Soc. (London), A, 243,534-545.
Chaffin, D.B. (1969). A computerized biomechanicaI model-development ofand use in studying
gross body actions. J. Biomech. 2: 429-441.
Chandler, R.F., Neri, L.M., Pollard, D.W., and Caiafa, e.A. (1980). Crash injury protection in
survivable air transport accidents-U.S. Civil Aircraft Experience from 1970 to 1978. FAA
Report No. FAA-CT-80-34.
Chandler, R.F. and Trout, E.M. (1979). Evaluation of seating and restraint systems conducted
during fiscal year 1978. FAA Report No. AM-79-17, ADA074881/4.
Chuong, C.J. (1985). Biomechanical model of thorax response to .blast loading. Report on
Contract No. DAMD 17-82-C-2062. Jaycor, San Diego, CA, 92138.
Clemedson, C-J. (1956). Blast injury. Physiol. Rev. 36: 336-354.
Clemedson, C-J. and Jonsson, A. (1962, 1964) Distribution of extra-and intra-thoracic pressure
variations in rabbits exposed to air shock waves. Acta Physiol. Scand. 54: 18-29, 1962. See
also, ibid, 62: Suppl. 233: 3-31, 1964.
Courant, R. and Friedrichs, K.O. (1948). Supersonic Flow and Shock Waves. Interscience, New
York.
Crone, e. and Lassen, N.A. (eds.) (1970). Capillary Permeability, Proc. of a Sym., Academic Press,
New York.
Currey, J.D. (1970). Animal Skeletons. Edward Arnold, London.
Dunn, F. and Fry, W.J. (1961). Ultrasonic absorption and reflection by lung tissue. Phys. Med.
BioI. 5: 401-410.
ElIros, R.M., Mason, G., Uszler, J.M., and Chang, R.S.Y. (1982). Exchange of small molecules in
the pulmonary microcirculation. In Mechanisms of Lung Microvascular Injury. Annals of
N.Y. Acad. of Sci. 384: 235-245.
494 12 Strength, Trauma, and Tolerance

Egan, E.A., Nelson, R.M., and Oliver, R.E. (1976). Lung inflation and alveolar permeability to
non-electrolytes in the adult sheep in vivo. J. Physiol. (London) 260: 409-424.
Eiband, A.M. (1959). Human tolerance to rapidly applied accelerations: A summary of the
literature. NASA Memorandum No. 5-19-59E.
Engin, A.E. (1979). Measurement of resistive torques in major human joints. Report No. AMRL-
TR-79-4. Air Force Aerospace Medicine Res. Lab., Wright-Patterson Air Force Base, Ohio.
Feinberg, B.N. and Fleming, D. G. (eds) (1978). CRC Handbook of Engineering in Medicine and
Biology. Sec. B. Instruments and Measurements Vol. I (A. Burstein and E. Bahniuk, eds.). CRC
Press, Boca Raton, Florida.
Fishman, A.P. and Hecht, H.H. (eds.) (1969). The Pulmonary Circulation and Interstitial Space.
The Univ. of Chicago Press, Chicago.
Fleck, J.T. (1975). Calspan three-dimensional crash victim simulation program. In "Aircraft
Crashworthiness". (K. Saczalski, G.T. Singley, W.D. Pilkey, and R.L. Huston, eds.), Univ. of
Virginia Press, Charlottesville.
Frisoli, A. and Cassen, B. (1950). A study of hemorrhagic rib markings produced in rats by air
blast. J. Aviation Med. 21: 510-513.
Frucht, A.-H. (1953). Die Schallgeschwindigkeit in menschlichen und tierischen Geweben. Ges.
Exp. Med. 120: 526-557.
Fung, Y.C (1957). Some general properties of the dynamic amplification spectra. J. Aeronautical
Sci. 24: 547-548.
Fung, Y.C (1962). On the response spectrum oflow-frequency mass-spring systems subjected to
ground shock. J. Aeronautical Sci. 29: 100-101.
Fung, Y.C (1965). Foundations of Solid Mechanics. Prentice-Hall, Englewood Cliffs, New Jersey.
Fung, Y.C. (1977). A First Course in Continuum Mechanics. 2nd ed., Prentice-Hall, Englewood
Cliffs, New Jersey.
Fung, Y.C (1981). Biomechanics: Mechanical Properties of Living Tissues. Springer-Verlag, New
York.
Fung, Y.C (1982). On human tolerance to impact loads. Report to U.S. Department of Trans-
portation, Transportation Center, Cambridge, Massachusetts.
Fung, Y.C (1984). The application of biomechanics to the understanding and analysis of trauma.
In: The Biomechanics of Trauma (A.M. Naham, ed.). Appleton-Century-Crofts, Norwalk,
Connecticut, pp. 1-16.
Fung, Y.C (1985). The application of biomechanics to the understanding and analysis of trauma.
In The Mechanics of Trauma, (A.M. Nahum and J. Melvin, eds.), Appleton-Century, Crofts,
Norwalk, Conn., pp. 1-16.
Fung, Y.C. and Barton, M.V. (1958). Some shock spectra characteristics and uses. J. Appl. Mech.
25: 365-372.
Fung, Y.C. and Barton, M.V. (1962). Shock response of a nonlinear system. J. Appl. Mech. 29:
465-476.
Fung, Y.C, Yen, R.T., Tao, Z.L., and Liu, S.Q. (1988): A hypothesis on the mechanism of trauma
oflung tissue subjected to impact load. J. Biomech. Eng. 110: 50-56.
Gadd, CW. (1966). Use of a weighted impulse criterion for estimating injury hazard. 10th Stapp
Car Crash Conf., Soc. of Automotive Engineers. pp. 95-100.
Gadd, CW., Culver, CC, and Nahum, A.M. (1971). A study of responses and tolerances of the
neck. 15th Stapp Car Crash Conf., Soc. of Automotive Engineers.
Gennarelli, T.A., Ommaya, A.K., and Thibault, L.E. (1971). Comparison of translational and
rotational head motions in experimental cerebral concussions. 15th Stapp Car Crash Conf.,
Soc. of Automotive Engineers.
Giuntini, C (ed.) (1971). Central Hemodynamics and Gas Exchange. Minerva Medica, Torino,
Italy.
Goldsmith, W. (1972). Biomechanics of head injury. In "Biomechanics: Its Foundations and
Objectives". (Y.C Fung, ed.), Prentice-Hall, Englewood Cliffs, N.J., pp. 585-634.
Griffith, A.A. (1920). The phenomena of rupture and flow in solids. Phil. Trans. Roy. Soc. London,
A221: 163-198.
References 495

Gurdjian, E.S., Lissner, H.R., Latimer, F.R., Haddad, B.F., and Webster, J.E. (1953). Quantitative
determination of acceleration and intracranial pressure in experimental head injury.
Neurology 3: 417-423.
Gurdjian, E.S., Webster, J.E., and Lissner, H.R. (1955). Observations on the mechanism of brain
concusion, contusion and laceration. Surgery, Gynecology and Obstetrics 101: 688-890.
Hayes, W.e., Swenson, Jr., L.W., and Schurman, D.J. (1978). Axisymmetric finite element analysis
of the lateral tibial plateau. J. Biochem. 11: 21-33.
Hill, I.R. (1978). Injury mechanisms analysis in aircraft accidents (of Royal Air Force, U.K.) In
AGARD Conf. Proc. No. 253: Models and Analogues for the Evaluation of Human Bio-
dynamic Response, Performance and Protection. NATO, AGARD, Paris. Article A12.
Hirsch, A.E., Ommaya, A.K., and Mahone, R.H. (1970). Tolerance of sub-human primate
brain to cerebral concussion. In Impact Injury and Crash Protection, Charles C. Thomas,
Springfield, IL., pp. 352-369.
Hodgson, V.R. and Thomas, L.M. (1972). Effect oflong-duration impact on head. 16th Stapp Car
Crash Conf., Soc. of Automotive Engineers.
Holbourne, A.H.S. (1943). Mechanics of head injury. Lancet 245: 438-441.
Hopkinson, J. (1872). Collected Scientific Papers, Vol. II, p. 316.
Hopkinson, B. (1914). A method of measuring the pressure produced in the detonation of high
explosives or by the impact of bullets. Proc. Royal Soc. (London) 213: 437-456.
Hopping, F.G. and Hildebrandt, J. (1977). Mechanical properties of the lung. In: Bioengineering
Aspects of the Lung, (J.B. West, ed.), Marcel Dekker, New York.
Huston, R.L. and Perrone, N. (1980). Dynamic response and protection of the human body and
skull in impact situations. In "Perspective in Biomechanics (H. Reul, D.N. Ghista, and
G. Rau, eds.), Harwood Academic Publishers, New York, pp. 531-572.
Huston, R.L. and Sears, J. (1979). Effect of protective helmet mass on head/neck dynamics, 1979
Biomechanics Symposium, ASME, (W.e. Van Buskirk, ed.), pp. 227-229.
Jacob, S.W., Francone, e.A., and Lossow, W.J. (1982): Structure and Function in Man. Saunders,
Philadelphia.
Jacobs, R.R. and Ghista, D.N. (1982). A biomechanical basis for treatment of injuries of the
dorsolumbar spine. In Osteoarthromechanics, (D.N. Ghista, ed.), McGraw-Hill, New York,
pp.435-472.
Jonsson, A., Clemedson, C-J., Sundqvist, A-B., and Arvebo, E. (1979): Dynamic factors influencing
the production of lung injury in rabbits subjected to blunt chest wall impact. Aviation, Space
& Envir. Medicine 50: 325-337.
Kane, T.P. and Scher, M.P. (1970). Human self-rotation by means oflimb movements. J. Biomech.
3: 39-49.
Kaye, G.W.C. and Labby, T.H. (1960). Tables of Physical and Chemical Constants, 12th Ed.,
Longmans, Green, New York.
Kazarian, L. and Graves, G.A., Jr. (1977). Compression strength characteristics of the human
vertebracolumn. Spine 2 (No.1).
King, A.I. (1975). Survey of the state of the art of human biodynamic response. In Aircraft
Crashworthiness, (K. Saczalski, G.T. Singley III, W.D. Pilkey and R.L. Huston, eds.), Univ.
Press of Virginia, Charlottesville, VA., pp. 83-120.
King, A. (1984). A review ofbiomechanical models. J. Biomech. Eng. 106: 97-104.
Koogle, T.A., Swenson, Jr., L.W., and Piziali, R.L. (1979). Dynamic three-dimensional modelling
of the human lumbar spine. In "Advances in Bioengineering" (M.K. Wells, ed.) Amer. Soc.
Mech. Engineers, pp. 65-68.
Kooyman, G.L. (1981). Weddell Seal: Consummate Diver. Cambridge Univ. Press, London.
Kraman, S.S. (1983). Speed oflow-frequency sound through lungs of normal man. J. Appl. Physiol.
55: 1862-1867.
Kulak, R.F., Belytschko, T.B., Schultz, A.B. and Galante, J.O. (1976). Nonlinear behavior of
human intevertebral disc under axial load. J. Biomech. 9: 377-386.
Laananen, D.H. (1980). Aircraft crash survival design guide Vol. 2. Aircraft Crash Environment
and Human Tolerance. Report prepared by Simula Inc., Tempe, Arizona for Applied Tech.
496 12 Strength, Trauma, and Tolerance

Lab., U.S. Army Res. and Tech. Labs., Fort Eustis, Virginia. Report No. USARTL-TR-79-
22B.
Lanir, Y. and Fung, Y.C. (1974): Two-dimensional mechanical properties of the rabbit skin. I.
Experimental system. J. Biomech. 7: 29-34. II. Experimental results. J. Biomech. 7: 171-182.
Lau, V.K. and Viano, D.C. (1981). Influence of impact velocity and chest compression on
experimental pulmonary injury severity in rabbits. The Journal of Trauma, 21: No. 12.
Lissner, H.R., Lebow, M., and Evans, F.G. (1960). Experimental studies on the relation between
acceleration and intracranial pressure changes in man. Surgery Gynecology, and Obstetrics
3: 329-338.
Liu, Y.K. (1980). Mechanisms, evaluation, prognosis, and management of head and neck injury.
In Perspectives in Biomechanics, (H. Reul, D.N. Ghista, and G. Rau, eds.), Harwood Academic
Publishers, New York. pp. 537-581.
Ludwig, G.D. (1950). The velocity of sound through tissues and the acoustic impedance oftissues.
J. Acoustic Soc. Am. 22: 862-866.
Melvin, J.W., Stalnaker, R.L., Roberts, V.L., and Trollope, M.L. (1973). Impact injury mechanisms
in abdominal organs. 17th Stapp Car Crash Conf, Soc. of Automotive Engineers.
Mertz, Jr., H.J. and Kroell, C.K. (1970). Tolerance of the thorax and abdomen. In Impact Injury
and Crash Protection, Charles C. Thomas, Springfield, III.
Muskian, R. and Nash, C.D. (1974). A model for the response of seated humans to sinusoidal
displacements of the seat. J. Biomech. 7: 209-215.
Nachemson, A., and Elfstrom, G. (1970). Intravital dynamic pressure measurements in lumbar
discs. A study of common movements, manoeuvres, and exercises. Scand. J. Rehab. Med.,
Suppl1. See also author's paper in Perspectives in Biomedical Engineering, (R.M. Kenedi, ed.),
Univ. Park Press, New York, 1973, pp. 111-119.
Nahum, A.M. and Melvin, J. (eds) (1985). The Biomechanics of Trauma. Appleton-Century-Crofts,
Norwalk, Connecticut.
Nahum, A., Ward, c., Raasch, E., Adams, S., and Schneider, D. (1980). Experimental studies of
side impact to the human head. 24th Stapp Car Crash Corif'., Soc. of Automotive Engineers.
Neathery, R.F. (1974). Analysis of chest impact response data and scaled performance recom-
mendations. 18th Stapp Car Crash Conf, Soc. of Automotive Engineers, pp. 459-493.
Nicolaysen, G. and Hauge, A. (1982). Fluid exchange in the isolated perfused lung. In Mechanisms
of Lung Microvascular Injury, Annals of N.Y. Acad. of Sci. 384: 115-125.
Ommaya, A.K. (1968). Mechanical properties of tissues of the nervous system. J. Biomech. 1:
127-138.
Ommaya, A.K., Fass, F., and Yarnell, P. (1968). Whiplash injury and brain damage. J. Amer.
Med. Assoc. 204: 285-289.
Orne, D. and Young, D.R. (1976). The effects of variable mass and geometry, pretwist, shear
deformation and rotary inertia on the resonant frequencies of intact long bone: A finite
element model analysis. J. Biomech. 9: 763-770.
Panjabi, M.M. and White, A.A. III (1982). Spinal mechanics: Kinematics, kinetics, and mathema-
tical models, procedures and devices for correction of deformaties and fixation of spinal
fractures. In Perspectives in Biomechanics, (H. Reul, D.N. Ghista and R. Raul, Eds.),
pp. 617-682. Harwood Academic Pub. New York.
Park, J.B. (1984). Biomaterial Science and Engineering. Plenum Press, New York.
Passerello, C.E. and Huston, R.L. (1971). Human attitude control. J. Biomech. 4: 95-102.
Perrone, N. (1972). Biomechanical problems related to vehicle impact. In: Biomechanics: Its
Foundations and Objectives. (Y.c. Fung, ed.), Prentice-Hall, Englewood Cliffs, New Jersey,
pp.567-584.
Piziali, R.L., Hight, T.K., and Nagel, D.A. (1976). An extended structural analysis of long
bones-Application to the human tibia. J. Biomech. 9: 695-701.
Privitzer, E. and Belytschko, T. (1980). Impedance of a three dimensional head-spine model.
Intern. J. of Math. Modeling (No.2)
Rice, D.A. (1983). Sound speed in pulmonary parenchyma. J. Appl. Physiol. 54(1): 304-308, 1983.
Richmond, D.R., Damon, E.G., Fletcher, E.R., Bowen, I.G., and White, C.S. (1968). The relation-
References 497

ship between selected blast-wave parameters and the response of mammals exposed to air
blast. Annals of N. Y. Acad. Sciences, Vol. 152, pp. 103-121.
Rivlin, R.S. and Thomas, A.G. (1953). Rupture of rubber. I. Characteristic energy for tearing. J.
Polymer Sci. 10: 291-318.
Roberts, S.B. and Chen, P.H. (1970). Elastostatic analysis of the human thoracic skeleton, J.
Biomech. 3: 527-545.
Roth, E.M., Teichner, W.G., and Craig, R.L. (1968). Compendium of human response to the
aerospace environment, (E.M. Roth Ed.), Vol. II, Sec. 7. Acceleration. NASA Contractor
Report No. CR-1205 (II), Prepared by Lovelace Foundation, Albuquerque, N.M.
Saha, S. (1982). The dynamic strength of bone and its relevance. In Osteoarthromechanics, (D.N.
Ghista, ed.), McGraw-Hill, New York, pp. 1-44.
Schneider, D.C. (1982). Viscoelasticity and tearing strength of the human skin. Ph.D. Thesis,
University of California, San Diego, Department of AMES/Bioengineering.
Schultz, A.B. and Andersson, G.B.J. (1981). Analysis of loads on the lumbar spine. Spine 6: 76-82.
Sharma, M.G. (1979). Development of a tear criterion for rupture of thoracic aortas in vitro. In
1979 Adv. in Biomech. 119-122. Amer. Soc. Mech. Eng. New York.
Sonnerup, L. (1982). Stress and strain in the intervertebral disk in relation to spinal disorders. In
Osteoarthromech. (D.N. Ghista, ed.), McGraw-Hill, New York, pp. 315-352.
Stalnaker, R.L., Roberts, V.L., and McElhaney, J.H. (1973). Side impact tolerance to blunt trauma.
17th Stapp Car Crash Conf., Soc. of Automotive Engineers.
Stapp, J.P. (1957). Human tolerance to deceleration. Am. J. of Surgery 93: 734-740.
Stapp, J.P. (1961). Human tolerance to severe, abrupt deceleration. In Gravitational Stress
in Aerospace Medicine, (O.R. Gauer and G.D. Zuidema, eds.), Little, Brown, Boston,
pp.165-88.
Stapp, J.P. (1970). Voluntary human tolerance levels. In Impact Injury and Crash Protection,
(E.S. Gurdjian, W.A. Lange, L.M. Partrick, and L.M. Thomas, eds.), Charles C. Thomas,
Springfield, Ill.
Staub, N. (1978). Lung fluid and solute exchange. In "Lung Water and Solute Exchange"
(N. Staub, ed.). Marcel Dekker, New York, pp. 3-16.
Sundaram, S.H. and Geng, c.c. (1977). Finite element analysis of the human thorax. J. Biomech.
10: 505-516.
Tao, Z.L. and Fung, Y.c. (1987). Lungs under cyclic compression and expansion. J. Biomech.
Eng. 109: 160-162.
Taylor, G.1. (1946). The testing of materials at high rates of loading. J. Inst. Civil Engineers 26:
486-519.
Thomas, A.G. (1960). Rupture of rubber. VI. Further experiments on the tear criterion. J. Appl.
Polymer Sci. 3: 168-174.
Thomson, W.T. and Fung, Y.c. (1965). Instability of spinning space stations due to crew motion.
AIAA J. 3: 1082-1087.
Versace, J. (1971). A review of the severity index. 15th Stapp Car Crash Conf., Soc. of Automotive
Engineers.
Viano, D.C. Helfenstein, U., Anliker, M., and Ruegsegger, P. (1976). Elastic properties of cortical
bone in female human femurs. J. Biomech. 9: 703-710.
Viano, D.C. and Lau, V. (1988). A viscous tolerance criterion for soft tissue injury assessment.
J. Biomech. 21: 387-399.
von Gierke, H.E. (1964). Biodynamic response of human body. Appl. Mechanics Review 17:
951-958.
von Gierke, H.E. (1971). Biodynamic models and their applications. J. Acoustic Soc. of America
50: 1397-1412.
Wainwright, S.A., Biggs, W.D., Currey, J.D. and Gosline, J.M. (1976). Mechanical Design in
Organisms. Wiley, New York.
Ward, c., Chan, M., and Nahum, A. (1980). Intracranial pressure-A brain injury criterion. 24th
Stapp Car Crash Conf., Soc. of Automotive Engineers, pp. 161-185.
White, C.S., Jones, R.K., Damon, E.G., Fletcher, E.R., and Richmond, D.R. (1971). The bio-
498 12 Strength, Trauma, and Tolerance

dynamics of airblast. Report No. DNA 2738T, Lovelace Foundation for Medical Education
and Research, Albuquerque, New Mexico.
Woo, S.L.Y. and Buckwalter, J.A. (eds.) (1988). Injury and Repair of the Musculoskeletal Soft
Tissues. Park Ridge, IL. American Academy of Orthopedic Surgeons.
Yamada, H. (1970). Strength of Biological Materials. (F.G. Evans, ed.), Williams and Wilkins,
Baltimore.
Yen, R.T. and Fung, Y.c. (1985). Thoracic trauma study: Rib markings on the lung due to impact
are marks of collapsed alveoli; but not hemorrhage. J. Biomech. Eng. 107: 291-292.
Yen, R.T., Fung, Y.c., Ho, H.H. and Butterman, G. (1986). Speed of stress wave propagation in
the lung. J. Appl. Physiol. 61(2): 701-705.
Yen, R.T., Ho, H.H., Tao, Z.L. and Fung, Y.c. (1988). Trauma of the lung due to impact load.
J. Biomech. 21: 745- 753, 1988.

The flower of water lily is loved for its poetic purity. But it revealed to me a certain
aspect of growth. Last summer I returned home after a three-month trip and found
my lily pond almost dry and that the lilies were shriveled and had turned black. Each
plant was reduced to the size of a hand. I drained and refilled the pond. The next day
some tiny leaves sprouted. They grew rapidly; their slender stems lengthened at a rate
of about a foot a day. In three days new round green leaves were floating on the surface
of water, and before long I got the picture shown here. What tenacity! What is their
secret?
CHAPTER 13

Biomechanical Aspects of Growth and

Tissue Engineering

13.1 Introduction
All parents want healthy children. Everyone wants a strong and handsome
body. We all believe that a proper level of exercise, i.e. a proper level of stress
and strain, is necessary for health. This is common sense. If we can turn this
common sense into a precise knowledge about stress and growth, then it will
enlighten biology in general, help surgeons to engineer healing, throw light
on physical education, sports techniques, health care, rehabilitation.
In our bodies, cells either live in steady state (homeostasis), reproduce,
move, or die. Tissues hypertrophy or resorb. In this chapter we are concerned
with tissues and organs, and not with cell culture. The distinction is important
because in cell culture, cells may proliferate at a great rate, and are active
individually. When cells form a confluent layer, they become quiescent and
then change slowly. In a tissue, every cell has neighbors on all sides, and the
rate of turnover is usually very low.
There is no doubt· that cells and tissues change and grow molecule by
molecule and must have a biochemical foundation. But it must have a bio-
physical foundation also. This chapter is concerned about the latter, with a
focus on the effects of stress and strain.
In the following, we shall first present a historical review of the concept of
growth and change in bone, then present information on the growth and
change of soft tissues, especially the heart, lung, and blood vessels. We then
offer the idea that growth and change in an organ can be revealed simply by
examining the changes of its zero-stress state. A theoretical examination of
the meaning of the residual stresses in organs then leads to an experimental
hypothesis on growth presented in Sec. 13.10.
The concepts of growth and change lead to the proposal of tissue engineer-

499
500 13 Biomechanical Aspects of Growth and Tissue Engineering

ing as a new discipline. Tissue engineering is defined as the application of the

principles of engineering and biology toward a fundamental understanding of
the structure-function relationship of tissues, and the development of bio-
logical substitutes to restore, maintain, or improve tissue functions. Examples
are artificial blood vessels covered with the patient's own endothelial cells;
skin substitute made of the patient's own keratinocytes, etc. The present status
and future prospects are discussed in Secs. 13.11-13.13.

13.2 Wolff's Law and Roux's Functional Adaptation Concept

On the subject of the relationship between structure and stress, Wolff's law is
the most famous, although its meaning and foundation are hazy. In 1981,
Roesler presented a thorough account of the history of Wolff's law. The story
is interesting and instructive, hence we present an outline of Roesler's findings
below.
In 1866, G.H. Meyer presented a paper on the structure of cancellous bone
at a meeting of the Ziiricher Naturforschende Gesellschaft. Meyer demon-
strated that "the spongiosa showed a well-motivated architecture which is
closely connected with the statics and mechanics of bone." C. Culm ann, a
mathematician who had published a book on graphic statics in the same year
was in the audience. Culm ann remarked that the lines in Meyer's drawings
resembled the principal stress trajectories in cantilever beams. Meyer, in his
paper published the following year (1867), stated that Culm ann, stimulated
by the drawings of bone structure, asked a student of his to construct the
principal stress trajectories in a crane-like curved bar loaded in a fashion
similar to the human femur (see Fig. 13.2:1). This was later to become the
famous "Culmann's crane."
Meyer (1867) formulated his ideas in three questions:
1) Is it possible that structures like the observed ones are formed by static
(force-equilibrium) conditions?
2) What is the internal metamorphosis that makes these structures so "fit for
service?"
3) Can these structures be understood if one adds to the external loads the
mechanical influence of the traction of muscles and ligaments?
Two years later, Wolff (1869) claimed that he could prove the following:
1) There is a perfect mathematical correspondence between the structure of
cancellous bone in the proximal end of the femur and the trajectories in
Culmann's crane.
2) There is a statical importance and necessity of the trajectorial structure of
the bone.
3) Bone growth can occur only in the interstitial space.
4) The compact bone is nothing but a compressed cancellous bone.
13.2 Wolff's Law and Roux's Functional Adaptation Concept 501
1~r .f1I•

.... I
,j

I
j.
A B

FIGURE 13.2:1 Culmann's crane presented by Wolff in his 1870 paper in Virchow's
Archiv.

Wolff then published a sequence of papers in 1870, 1872, 1874, 1884, 1891,
and a book in 1892. In his 1884 paper he called this the "law of bone
transformation". His "proof", besides relying on the rough similarity between
the trabecular pattern in the cancellous bone of the femur to Culmann's crane,
(Fig. 13.2:1), does contain a valid detail: just as the principal directions of a
stress tensor at any given point are orthogonal, the trabeculae in the bone,
when they intersect, do seem to intersect at right angles (Fig. 13.2:2). Other-
wise there are many questions: why should the bone be molded according to
the stress trajectories of one particular loading? What happens to other loading
conditions? Why is it sufficient to consider a curved beam of homogeneous,
isotropic, elastic material, when the bone is not? Furthermore, Wolff's ideas
about interstitial growth, and medullary cavity, turned out to be wrong.
The "crane" constructed by Culmann's student was also questionable. The
figure obviously does not agree with the known results of the theory of
elasticity of curved beams, which have stress concentration on the concave
side. In Wolff's 1870 paper, a figure of the stress distribution in the crane
502 13 Biomechanical Aspects of Growth and Tissue Engineering

FIGURE 13.2:2 Photograph of bone structure presented in Wolff's paper of 1870 in

Virchow's Archiv.
13.3 Healing of Bone Fracture 503

showed a linear distribution of the normal stress and a parabolic distribution

of the shear stress in the beam cross section, which are true only for straight
beams. Roesler (1981), after a detailed investigation, concluded that Culmann
(or his student) just used the St. Venant theory of a straight cantilever rec-
tangular beam loaded by shear at the free end, plotted the intercepts of the
trajectories on 8 normal sections of the crane, and faired in with smooth
curves. Yet this drawing became a historical landmark!
The three questions asked by Meyer, however, remain meaningful and need
to be answered.

Roux's Functional Adaptation Concept

In 1880 and 1881 Wilhelm Roux introduced the idea of "functional adapta-
tion", which means "adaptation to a function by making use of it". He
called the process of functional adaptation a "quantitative self-regulating
mechanism" controlled by a "functional stimulus". Being a contemporary of
Charles Darwin (1809-1882) and· Alfred Wallace (1823-1913), Roux (1850-
1924) formulated his idea in a general way to deal with all aspects of evolution.
One of his important examples is the trajectory hypothesis of cancellous bone
structure. He said: "There exists a group of very fine appropriate structural
formations which could not result from selection alone .... This is the structure
of cancellous bone corresponding to the static pressure lines, which enables
bone to resist external forces with a minimum material consumption". "Many
cells are influenced by functional stimuli, the nerve, muscle, gland cells by the
related (electric) pulse, bone cells and connective tissue cells by pressure and
tension.... It is plausible that, if this influence is disadvantageous for sOme
cells of a tissue or some parts of it, then these cells will disappear during the
process of physiological regeneration. If, however, variations appear for which
the functional stimulus is favorable, then in the course of time these cells will
replace aU the other cells, which are indifferent to the functional stimulus"
(Roux, 1881).
In this way, Roux put the question of the structure of cancellous bone in
a more fundamental form. By focusing on functional stimulus, the question
can be examined in detail and answered quantitatively. But his own analysis
of the cancellous bone seems to be faulty, as was pointed out later by Pauwels
(1980), see the following section, and Roesler (1981).

13.3 Healing of Bone Fracture

Bone is a solid material which may fail in modes discussed in Sec. 12.2. If a
crack exists in a material, it may lead to fracture because the ends of the crack
are stress raisers. The critical state of stress at which a crack will enlarge itself
is reached when the energy needed for the creation of new free surface of an
advancing crack is balanced or exceeded by the sum of the energies released
504 13 Biomechanical Aspects of Growth and Tissue Engineering

from the strain in the body due to the formation of new surface, the work done
by the loading caused by the advancing crack, and the energy dissipation due
to plastic deformation in regions near the ends of the crack. This basic concept
by Griffith (1920) has lead to important advances in fracture mechanics. See
reviews in Erdogen (1983), and Sih (1981). To the bone, additional considera-
tions are needed to account for the composite structure of the bone (being an
organized mixture of collagen and crystalline hydroxyapatite), the unit struc-
ture of the osteon, and the existence of voids in cancellous bone.
Pauwels (1935, see his book, 1980, pp. 1-105) has shown that fractures of
the femoral neck can be classified into three types on the basis of stress acting
on the surface of fracture imposed by the muscles and the weight of the body.
If the normal stress is denoted by (Tn and the shear stress is denoted by r, then
Pauwel's three types of fractures are:
Type 1: (Tn < 0 (in compression), r < f
Type 2: (Tn < 0 (in compression), r > f
Type 3: (Tn > 0 (in tension).
Here f is a certain critical value of shear stress. Pauwels said that the Type 1
fracture usually heals by itself into a bony union without treatment. Type 3
fractures usually will not form a bony union. Fractures of this type that appear
to have been healed often progressively evolve into a pseudoarthrosis (false
joint) after being loaded. The Type 2 fracture generally has a pessimistic
prognosis. He argues that the newly regenerated bone tissue cannot resist a
shear stress r appreciably higher than f
Figure 13.3:1 shows Pauwels examples of the three types offracture. In this
figure, R is the resultant of the forces in the muscles and the body weight. Ks
is the "effective" shearing force. Z is tension, D is compression. The figure

FIGURE 13.3:1 Pauwels' illustration of three types of fracture. From Pauwels (1980).
Reproduced by permission.
13.3 Healing of Bone Fracture 505

shows that the angle between the normal to the plane offracture of the femoral
neck and the verticle axis is a deciding factor. If the angle is less than 30°, the
fracture is Type 1 and healing can be assured. If it is greater than 60°, the
fracture is Type 3 and pseudoarthrosis will result. In between, much more
carefully analysis is needed, and some ingenious methods of treatment have
been devised.
The healing of a fracture in the femoral neck is very different from that of
a long bone (such as a diaphyseal fracture). In the latter, a collagenous
periosteal callus will develop first to take up any tensile stress if it exists. In
the femoral neck, no such periosteal callus will form. The healing of the
femoral neck relies on the medullary healing tissue.
How does the medullary healing tissue develop? It is known that the new
tissue can develop into either a connective tissue, or a cartilage, or a bone. If
it develops into a connective tissue or cartilage, then the fracture will become
a false joint. If it develops into a bone, then the fracture will be healed.

Roux and Pauwels' Hypotheses

Roux and Pauwels recognized the importance of stress in healing. Roux (1985)
formulated the following hypotheses: tensile and shear stresses provoke the
formation of connective tissue. Friction or shearing movement leads to the
formation of cartilage. 'Functional' compression (defined as a periodically
varying compressive stress) provokes bone formation.
Pauwels (1980, pp. 1-105) at first believed in Roux's hypotheses. Then
he expressed doubts (1980, pp. 106-137), and finally presented (1980,
pp. 375-407) the following hypothesis: elongation and hydrostatic 'pressure'
are the two specific stimuli for the differentiation of the mesenchymal tissue.
Elongation (i.e. unequal principal strains) provokes the formation of collagen
fibrils, and therefore of connective tissue. Hydrostatic "pressure" (here defined
as a state of strain in which the three principal strains are equal) stimulates
the formation of cartilage tissue. There is no specific mechanical stimulus for
the formation of bone tissue.
Pauwels quotes many observations (e.g., Krompecker, 1937) that contradict
Roux's hypotheses and support his own. Among these, the most important is
the following: In bone remodeling of the femoral neck which is cancellous, the
regions in which the largest principal stress is in tension are calcified along
the tensile trajectories, showing that tensile stress stimulates bone growth as
well as the compressive stress.

Carter's Hypothesis
The question was studied recently by Carter et al. (1987), Carter (1988),
Carter and Wong (1988). They examined the growth, maturation, and aging
of the skeleton in relation to the proliferation, maturation, degeneration, and
506 13 Biomechanical Aspects of Growth and Tissue Engineering

ossification of cartilage and the local stress or strain histories of the skeletal
tissue. Using a strain energy function as a probe, and finite element analysis
as a tool, they reached a qualitative conclusion as shown in Fig. 13.3:2. The
figure delineates the tissue development under cyclic loading of an undifferen-
tiated mesenchymal tissue. The left figure is drawn for a tissue with good blood
supply and adequate tissue oxygen tension. The right figure is for a tissue with
poor blood supply and low oxygen tension. The abscissa refers to the am-
plitude of cyclically applied dilatational (mean) stress, the ordinate refers to
the amplitude ofthe cyclic octahedral shear stress. For each stress state defined
by the coordinates, the end product is labeled in the figure. The coordinates

GOOD VASCULARITY POOR VASCULARITY

S
,~~~S Cyclic
Octahedral
Shear Stress

FIBROUS
TISSUE
CARTIL~A~G~E~~~~~u]ITn~~~~__~C~A~R~G11~LA~G~E~~~~~_
o D o D
Cyclic Dilatational Stress Cyclic Dilatational Stress

(-) Compression - - Tension (+) (-) Compression- - Tension (+)

FIGURE 13.3:2 Carter's hypothesis.

FIGURE 13.3:3 "Reorientation" surgery of a femeral head. Drawn from photographs

given by Pauwels (1980).
13.3 Healing of Bone Fracture 507

FIGURE 13.3:4 Resorption process turns a bad type 3 fracture into a healable type 2
fracture. From Pauwels (1980). Reproduced by permission.

are supposed to represent some kind of weighted average of the entire history
ofloading imposed on the bone over a period of time. A few "bad" cycles with
high shear or tensile hydrostatic stress may be sufficient to promote fibrous
tissue formation at the expense of osteogenesis. The details are not entirely
clear.
While the cellular dynamics remain to be clarified, it is clear that consid.era-
tion of mechanical stress acting on the wound should guide any surgical
procedures for the treatment of bone fracture.
Example of Application: Beneficial use of Resorption. Figure 13.3:3 shows
an operation of "reorientation", which turns a pseudoarthrosis into a Type 1
fracture which heals. Figure 13.3:4 shows a Type 3 fracture turned into a Type
2 fracture after resorption and controlled later by nailing. Both of these
examples are from Pauwels (1980, pp. 36,37).

Piezoelectric Effect
Bone is piezoelectric. In fact, all collagenous tissues, soft and hard, are piezo-
electric. Fukada (1974), Bassett (1978), Yasuda (1974), and others have pro-
posed that the piezoelectric property can be used clinically to heal fractures
and non unions in bone. The electromagnetic waves can be imposed on the
organ noninvasively. Bassett stresses the importance ofthe wave form because
he thinks that the electromagnetic waves must provoke interactions between
electrically charged cell membranes and charged hormones, antibodies, and
508 13 Biomechanical Aspects of Growth and Tissue Engineering

ions as normally achieved in nature. Impressive clinical results have been

reported.

Bioelectric Effect on the Growth of Whole Organ

There is no doubt that electromagnetic waves induce certain changes in cells.
Bioelectricity was brought to public attention in the mid 1700's by Luigi
Galvani. In 1903, A.P. Matthews discovered that living hydroids (small rela-
tives of sea anemones) are electrically polarized-that is, the 'head' end of the
animal, with the flowerlike hydranth, has a different electric potential than the
'foot' end, which is fastened to rock. Subsequently, E.J. Lund (1921) observed
the growth of eggs on the seaweed Fucus in sea water in the presence of an
applied current created by placing electrodes at the opposite sides of a petri
dish, and found that the eggs always produced their first outgrowths directed
toward the negative electrode. He then showed that he could control which
end of a piece of hydroid would become the head and which the foot by
applying an electric field. In 1920, Ingvar showed that chick nerve cells
cultured in vitro would survive and send out nerves toward a cathode when
a small DC current was applied to the culture dish. This was confirmed by
Sisken and Smith (1975) who found also that the growth rate in the electric
field was two or three times faster than normal. More recent efforts on electric
stimulation of nerve growth include the use of implantation of a bimetal
(platinum and silver) electrode (insulated except at the tip) across severed ends
of a nerve (with platinum in the proximal segment and silver in the distal
segment) and electromagnetic waves. Faster and better organized growth has
been reported.
How about limbs? Robert Becker (1961) discovered that the electric be-
havior of an amputated frog's limb was different from that of a newt's limb.
Now, frog's limbs are no better than human's so far as regeneration is con-
cerned, but a newt can grow new ones in a month or two when its limbs are
cut off. Smith (1974) experimented on frog's ability to regenerate limbs by
imposing the electric pattern seen in an amputated newt's limb. He implanted
a bimetallic couple of the sort mentioned in the preceding paragraph in the
limb stump of an amputated adult frog's forearm. When the cathode was at
the wound surface, he found that the frog grew regenerates which contained
recognizable organized limb structures.

13.4 Mathematical Formulations of Wolff's Law

Continued efforts have been made to clarify Wolff's Law. On the side of
morphology, quantitative stereology is used to determine the directions and
areal density of the trabeculae in any cross section. On the side of mechanics,
finite element method has been used to determine the stresses and strain in
the structure. Photoelasticity has served for many years as a means to deter-
13.4 Mathematical Formulations ofWollT's Law 509

mine the principal stresses in two-dimensionallucite models of bone. The use

of "frozen stress" in transparent plastics has extended photoelasticity to
three-dimensional models.
In the meantime, bone cells are studied with electron microscopes and
biochemical means. Menton et al. (1984) noted that all periosteal and endo-
steal bone surfaces are covered with bone lining cells. Among the bone lining
cells are osteoblastic precursor cells (Gr. blastos, germ). As the precursor cell
becomes an osteoblast and matures, it develops numerous pseudopods that
contact the pseudopods of maturing and mature osteocytes. As the mineraliza-
tion front approaches the osteoblast, the osteoblast appears to round up and
begins to function as an osteocyte.
Chambers (1985) describes osteoclastic cells (Gr. Klastos, broken) which
can reach the bone via blood circulation. Whether these are mononuclear
phagocytes or not is not clear. They participate in calcium homeostasis of the
blood and bone. Chambers (1985) presents a hypothesis which assumes that
strain stimulates the osteocytes which in tum induces osteoblasts to produce
osteokinetic agents that attract the osteoclasts. Thus the osteoblasts have the
capacity either to suppress (through prostaglandin production) or to stimulate
(through mineral exposure) osteoclastic resorption.
Lanyon (1984) has shown that bone resorption will occur for strains less
than about 0.001 and bone deposition will occur for strains greater than about
0.003. How osteocytes can sense strains so small is not known. Gross and
Williams (1982) and Salzstein and Pollack (1987) proposed that the mechan-
ism may be the ion streaming of fluid in cavities in bone tissue, including
lacunae, canaliculi, Haversian, and Volkmann canals. EI Haj et al. (1988)
showed that the orientation of the proteoglycan molecules within bone tissue
is related to strain rate. If there were no further loading, the proteoglycan
orientations persist for over a day. Further data are provided by Yeh and
Rodan (1984), Nulend (1987), and Gross et aI. (1988).

Surface Remodeling
Cowin and associates (1979, 81, 84, 85) have studied the mathematical form
of Wolff's law of bone remodeling. Their basic idea is to describe the osteo-
blastic and osteoclastic activities on the surface of bone tissue, including
trabecula, Haversian canal, and the periosteal or endosteal surface of a whole
bone. Bone cell activity is assumed to depend upon genetic, metabolic, and
hormonal factors as well as the strain history experienced by the bone cell.
The osteoblastic and osteoclastic function are written as ab and ac respectively.
The rate of surface deposition or resorption is expressed in terms of U, with
the dimensions of velocity (typically Jlm/day or mm/year). Then, following
the ideas of Martin (1972) and Hart (1983), Cowin and Van Buskirk (1979)
wrote
(1)
510 13 Biomechanical Aspects of Growth and Tissue Engineering

in which nb and nc denote the number of osteoblasts and osteoclasts, respec-

tively, per unit area, and A b , Ac are the surface areas available to the osteo-
blasts and osteoclasts, respectively. Further, for simplicity, they assumed
(2)
(3)
in which e is the strain, H b , H" Gb , G" C, eo are constants. eo represents a strain
at which no remodeling occurs. C is the remodeling rate constant. Cowin et
al. (1981, 1985) have applied these equations to the remodeling of long bones
with the finite element method.

Mathematical Description of the Cancellous Bone Architecture

A cross section of a cancellous bone specimen is shown in Fig. 13.2:2. The
calcified bone tissue, called trabeculae, is surrounded by bone marrow. White-
house (1974) and Whitehouse and Dyson (1974), used stereological methods
to measure the mean intercept length of the trabeculae as a function of the
direction of the test lines. Let 0 be the angle between a test line and an
arbitrarily chosen x-axis. Let L be the average distance between two bone/
marrow interfaces measured along all test lines inclined at an angle O. White-
house showed that the following equation fits the measured results very well

£11(0) = MllCOS 20 + M 22 sin 2 0 + 2M12 sinOcosO, (4)

where M ll , M 22 , M12 are constants. Harrigan and Mann (1984) then showed
that in three dimensions their experimental results can be fitted by an equation
which is a generalization of Eq. (4):
1
L2(0) = M;jn;nj' (5)

in which 0, with components (n;, n2, n3) referred to a rectangular cartesian

frame of reference, denotes a unit vector in the direction of the test line.
Equations (4) and (5) represent an ellipse and an ellipsoid, respectively, in the
space of n1, n2, n3. Mij can be considered the components of a tensor of rank
2, M, with Mij as components. In matrix form,

Mll M12 M13)

(M) = ( M21 M22 M 23 . (6)
M31 M32 M33

The quadratic form defined by the right-hand sides of Eqs. (4) and (5) are
positive definite. Hence an inverse of the matrix M exists. The validity of Eq.
(5) was confirmed by Cassidy and Davy (1985).
Cowin (1986) then introduced a tensor H defined as the inverse of the
square root of M:
13.5 Remodeling of Soft Tissues in Response to Stress Changes 511

(7)
and called it the fabric tensor of cancellous bone. He showed that the Young's
modulus of cancellous bone is larger if the Hij values are larger. Hence it is
preferable to correlate Hij with the mechanical properties of the cancellous
bone.

Mathematical Statement of Wolff's Law

When a cancellous bone is subjected to a loading, resulting in a stress repre-
sented by a tensor 1ii • a remodeling process will go on so that the fabric tensor
Hij will change with time. If 1ii were a constant, then after the elapse of a
certain period of time the remodeling process will cease when a new homeo-
static condition (remodeling equilibrium) is reached. Concerning this homeo-
static condition, Cowin (1986) states Wolff's law in the following form:
"At a homeostatic condition, the principal axes of the stress tensor 1ii and
those of the fabric tensor Hij coincide."
Cowin shows that this coincidence of principal axes is assured if the matrix
multiplication of (1i) and (Hij) is commutative, i.e., if
(8)
The proof of condition (8) is the same as that described in Sec. 2.10.
W oUI's law must be part of a more general law defining the rate of change
ofthe fabric tensor as a functional ofthe stress 1ii , strain Eii , tensor Hii , blood
flow Q, biochemical factors Ck' age, and time t:
(9)
By saying that it is a functional, we mean that it is a function of the entire
histories of these variables. Finding the functional and validating it are objec-
tives of future research.
One aspect of bone remodeling is the change of the density and strength
of the bone, or more restrictively, the solid volume fraction of trabecular
structure, denoted by v and defined as the volume of solid trabecular struts
per unit bulk volume of the tissue. At homeostatic condition of the patella
bone, Hayes and Snyder (1981) found a strong correlation of v with the von
Mises effective stress component of 1i} in two-dimensions. A subsequent
three-dimensional study by Stone et al. (1984) found, however, that v correlates
not with von Mises stress, but with principal stresses. Fyhrie and Carter (1985)
suggested that v correlates with the strain energy density. So again future
research has to focus on validation.

13.5 Remodeling of Soft Tissues in Response to Stress Changes

Heart, lung, blood vessels, and muscle also remodel in response to stress and
strain. Their features are described below.
512 13 Biomechanical Aspects of Growth and Tissue Engineering

FIGURE 13.5:1 A comparison of three dog's hearts: (a) Top left: Volume-overloaded
heart. (b) Top right: Pressure-overloaded heart. (c) Bottom: Normal. The normal heart
was photographed 4 hours post mortem. Photo by courtesy of Dr. Toshio Nakamura,
Tokai University, Isehara, Kanagawa, Japan.

Hypertrophy of the Heart

It is well known that when the heart is overloaded its muscle cells increase in
size. Figure 13.5:1 shows a comparison of the cross sections of (c) a normal
heart, (a) a heart chronically overloaded by an unusually large blood volume
(b) a heart chronically overloaded by an unusually large diastolic and systolic
left ventricular pressure, all from autopsies of man. Figure 13.5:2 shows a
comparison of the histology of a normal and a pressure-overloaded heart. It
is seen that the muscles in hypertrophic heart are much bigger in diameter
than those of the normal heart. Figure 13.5:3 shows the changes in myocytes
in hypertrophy.
Note the difference of these two types of hypertrophy. The volume-
overloaded heart increases its ventricular volume. The pressure-overloaded
heart increases its wall thickness both inward and outward, resulted in a
decrease in the volume of the ventricular chamber. Ultrastructural studies
of the muscle fibers showed (Anversa et aI., 1971) that the structure of the
muscle cells of the pressure-overloaded hypertrophic heart are pretty uniform
throughout the ventricular wall, as expressed in terms of the volumetric
density of the contractile mass, mitochondria, and T system, and the surface-
13.5 Remodeling of Soft Tissues in Response to Stress Changes 513

FIGURE 13.5:2 A comparison of the histology of a normal (left) and a pressure-over-

loaded heart (right). Photographed at the same magnification. Courtesy of Dr. Toshio
Nakamura.

to-volume ratio of these components in the outer, middle, and inner layers of
the ventricular wall. The fiber diameter of the hypertrophic muscle cells of the
left ventricle was said (Lund and Tomanek, 1978) to be larger in the sub-
endocardium than in the subepicardium, consistent with the concept that the
tensile stress is higher in the endocardial region; whereas the muscle fiber
diameter of the right ventricle was the same as the control. The muscle fiber
orientation was found (Tezuka, 1975) to be somewhat rearranged in the inner
myocardial layer: the average angle between the muscle fibers at the endo-
cardial surface and the basal plane was 63° for the normal heart, 56° for
pressure overloaded heart, 50° for the volume overloaded heart.
Nakamura et al. (1982), using data on the pressure-volume relationship of
the left ventricle to deduce the stress-strain relationship of the myocardium,
found that the stress-strain relationship of the hypertrophic myocardium is
not different from that of the normal. Arai et al. (1968) measured the growth
of the coronary artery and the capillary blood vessels in hypeftwphic hearts
and found that the mean blood flow per unit volume of the heart muscle
estimated by the anatomical radii of the three major branches of coronary
arteries was practically the same as that in the normal heart. * The total

* Based on Suwa et a\"s (1968) result that the mean blood flow of an arterial branch, Q, can be
related to the radius of that branch, r, by the equation Q= qrn, where q and n are constants. The
value n = 2.7 was found to hold for all the arterial systems examined.
514 13 Biomechanical Aspects of Growth and Tissue Engineering

FIGURE 13.5:3 Scanning electron micrographs. A: Ventricular cardiocytes isolated

from autopsied human hearts. HeM = hypertrophic cardiomyopathy with asym-
metrical hypertrophy. B: En face views of cellular architecture in situ in ventricular
myocardium. From K. Kawamura (1982). Reproduced by permission.

capillary length was found to increase in hypertrophic hearts in parallel with

the increase in the total surface area of the muscle fibers. The capillary length
per unit volume of the heart muscle, was found to be somewhat decreased in
hypertrophic heart.
The change in the behaviour of the hypertrophic heart seems to be due
more to the changed geometry ofthe ventricular chamber than to the changed
properties ofthe muscle fibers or blood vessels. Nakamura et al. (1982) showed
that at an end-diastolic pressure of 20 mmHg the end-diastolic volume of
hypertrophic heart was smaller than the normal by 47% (p < 0.01), and the
peak systolic pressure in isovolumic beats was higher than the control by 24%
(p < 0.01). When ejection pressure was the same, the difference between the
stroke volumes of the hypertrophic and control hearts was insignificant. The
isovolumic pressure line (see Biodynamics: Circulation, Fung (1984), p. 39)
changed, its slope and volume-axis intercept decreased in proportion to the
change in chamber geometry.
The question arises why is hypertrophic heart prone to failure? (Braunwald,
1980). The results quoted above suggest no simple answer. It is clear, however,
13.5 Remodeling of Soft Tissues in Response to Stress Changes 515

that the process of hypertrophy, its initiation and its equilibrium state, must
be stress related.
Volume and pressure overload are not the only causes for hypertrophy of
the heart. Frolich (1983) and Alpert (1983) reviewed other hemodynamic and
nonhemodynamic factors such as the increased heart rate, the augmented
ventricular contractility, the increased total peripheral resistance, aging, colla-
gen deposition, coexistant diseases (such as atherosclerosis, myocarditis,
diabetes mellitus), as well as race, sex, obesity, hormones (catecholamines,
angiotension II, growth hormone), and therapy. Limas (1983) showed that the
micro tubules are involved in isoproterenol-induced hypertrophy of the heart.

Remodeling of the Lung

Cowan and Crystal (1975) showed that when one lung of a rabbit was excised,
the remaining lung expanded to fill the thoracic cavity, and it grew until it
weighed approximately the weight of both lungs. What are the factors respon-
sible for this growth response? A study of the effects of growth hormone on
lung growth after unilateral pneumonectomy suggests that hormone is not
directly involved (Brody and Buhain, 1973). Increasing the blood flow to one
lung does not result in a growth response of the lung (Romanova et aI., 1970).
The hypoxemia seen immediately after pneumonectomy resolves shortly after
surgery and is probably not a major factor (Nattie et aI., 1974). It remains to
consider the mechanical factor of increased stress and strain (Cohn, 1939 and
Gaensler and Strieder, 1951).
For the purpose of quantitative investigation, collagen synthesis before and
after surgery was studied by Crystal (1974), Cowan and Crystal (1975), using
New Zealand white rabbits. They determined the course of collagen synthesis
in the lungs, performed surgery at 70 days of age, and continued monitoring
until 180 days after birth. At a specified time the lungs were excised and
weighed. Tissue slices (0.5 mm) were incubated from 1 to 4 hours in a medium
e
containing 0.012 mM 4C]proline. After incubation, the slices were rinsed,
homogenized, and measured for dry weight, DNA, protein, hydroxyproline,
e e
4C]proline in noncollagen protein, 4C]hydroxyproline, and the specific
activity of free [14C] proline as described by Bradley et ai. (1974). The rate of
collagen synthesis (nmols [14C] proline incorporated into [ 14 C] hydroxy-
proline per mg of DNA per hour) was calculated. This reflected the synthesis
of collagen chains (proline is hydroxylated only after it is incorporated into
collagen), because it has been verified that the number of hydroxylated pro-
lines per collagen molecule remained constant.
Cowin and Crystal's (1975) results are shown in Fig. 13.5:4. The ordinate
shows the total content of collagen in the right lung. Normal growth, marked
by open circles, shows a rapid increase of collagen in the 1st month of life. A
similar rapid growth is seen after left pneumonectomy. The rate of collagen
synthesis as a percentage of the total protein synthesis of the right lung is high
in the first month after birth and 1st month after left pneumonectomy.
516 13 Biomechanical Aspects of Growth and Tissue Engineering

Total right
lung collagen
I
,""
I
Ci 100 I Post-pneumonectomy
E I
z I
w
(,!)
«
yl
--l
--l 50
o
u

30 90 180
AGE (days)
Pneumonectomy

FIGURE 13.5:4 Cowan and Crystal's experiment showing lung collagen content
in the right lung after left pneumonectomy (e---e) compared with the normal right
lung at several ages (0--0). From Cowan and Crystal (1975). Reproduced by
permission.

Cowan and Crystal saw similar results if the left lung was tied off at the
hilum and left in the chest. On the other hand, if the left lung was excised and
the cavity left by the excised lung was ftlled with liquid paraffin embedding
wax after chest closure, then they found little change in the right lung from
normal.
These experiments show that there is a direct relationship between the
changed strain level and collagen synthesis in the lung.

Remodeling of Pulmonary Blood Vessels in Hypertension

When the oxygen supply to the lung is reduced suddenly, pulmonary blood
pressure increases and changes in blood vessels occur. The thicknesses of the
adventitia and the smooth muscle layer in the pulmonary arteries increase,
the lumens of small arteries deccrease, and new muscle growth occurs on
smaller and more peripheral arteries than normal. To identify cell division
activities in these vessels, Meyrick and Reid (1979) studied the incorporation
of 3H-thymidine into the DNA of the new cells in rats exposed to hypobaric
hypoxia (exposure to 380 torr) after 1,2,5,7,10, and 14 days. Using autoradio-
graphs of l-~m sections, they measured and expressed the 3H-thymidine
incorporation by a "labeling index". Figure 13.5:5 shows one of their results.
It is seen that the remodeling occurs within days. In a separate study, Meyrick
and Reid (1982) showed that pulmonary arterial endothelial cell mitosis
13.5 Remodeling of Soft Tissues in Response to Stress Changes 517

20 .. New muscle cells

o Arterial endothelial
cells

15

x
'"c
"0

en
c 10
a:;
.n
'"

:.. ..•..
....J

9·.
f····.
5 ·:· ....
.:-
f
. .
.....
·b ........ .
: •••••• -0
.. G-......
...... ~
i
O~L---~---L--~------~------~
3 5 7 10 14

Control Hypoxia (days)

FIGURE 13.5:5 The cell division activities reflected by the "labeling index" of inc orpo rat-
ing 3H-thymidine in the pulmonary peripheral intra-acinar arteries (external diameter
between 15 and 100 11m) after exposure to hypoxia. Standard deviations too small to
be shown. From Meyrick and Reid (1979). Reproduced by permission.

occurred also in rats fed with Crotalaria spectabilis seeds which caused pul-
monary artery hypertension; but in this case not only did the changes take
weeks to occur, but also there was a lack of evidence of cell division of small
muscle cells.
Further, Sobin et al. (1983) have shown that when rabbits were put into a
chamber low in partial pressure of oxygen, changes in the structure of their
pulmonary arterioles occurred within a few hours. Bevan (1976) induced
hypertension in dogs by tying off a renal artery, and showed that mitotic
activity began in the otolic artery soon afterwards, reaching a maximum in
about 2 weeks, and then subsiding gradually. Such mitotic activity was asso-
ciated with an increase in blood pressure, and hence is stress related.
Further discussion of the remodeling of blood vessels due to physical,
chemical, and biological stimuli is given in Sec. 13.9.
518 13 BiomechanicaI Aspects of Growth and Tissue Engineering

Changes in Blood Vessel Diameter Caused by Increased Blood Flow

Kamiya and Togawa (1980) presented results of a very interesting experiment
on the adaptive change of canine carotid artery diameter in response to
changes in blood flow. They constructed surgically an arteriovenous shunt
from the common carotid artery to the external jugular vein. As a con-
sequence, blood flow was increased in one segment of the artery while de-
creased in another. They showed that in 6 to 8 months after the operation,
the segment with increased flow dilated while the other segment with de-
creased flow atrophied to a smaller diameter. The diameter of the artery
increased or decreased in such a way that the wall shear rate* y remained
almost constant (within 15%) if the change of flow was within 4 times of the
control. The transendothelial protein permeability, evaluated at the Evans-
blue-dye-T-1824-stained surface by a reflectometric method, also showed a
close correlation with wall shear (r = 0.934). They therefore suggested a local
regulatory mechanism of wall shear stress controlling the vessel growth in-
volving protein turnover in the vascular wall.
According to Liebow (1963), Thoma (1893) first observed in chicken em-
bryos that the pathways of the fastest blood velocity became the main arteries
while those with slower velocity atrophied. Liebow (1963) and others, on
studying arteriovenous fistulas and collateral circulation, have also showed
that increased blood flow induces blood vessel dilatation. Rodbard (1975)
collected clinical evidences of the same. More recent work on the relationship
between shear stress and changes in vascular endothelial cells initiated by Fry
(1968) in search of atherogenesis has also indicated this trend.
Kamiya et al. (1984) collected data from the literature and obtained the
following estimates of the wall shear stresses (dynes/cm 2 ) in the dog:
aorta large arteries small arteries arterioles venule veins vena cava
12 10-16 19 14 3 1.5 6.3
For the capillaries, data from rat and cat show that the wall shear stress lies
in the range of 10-26 dyn/cm 2 • The narrowness ofthe range ofthe wall shear
stress in the entire vascular tree is remarkable.
If the shear stress on the endothelial wall due to the viscosity of the flowing
blood really controls the blood vessel caliber, then the feat is all the more
remarkable in view of the smallness ofthe shear stress. We know that the state
of stress at any point in a continuum can be stated either in terms of three
principal stresses (tensile or compressive) or in terms of two maximum shear
stresses and a mean stress. The two maximum shear stresses are numerically
equal to one half of the differences of the principal stresses, and act on planes
inclined at 45° to the principal planes. The mean stress is equal to one third
of the sum of the principal stresses. When blood flows in a blood vessel, the
blood pressure induces stresses in the vessel wall. At a normal blood pressure

* Yis calculated from y = 4Q/(:n:r3 ) under the assumption of laminar flow. Q is the flow rate.
13.5 Remodeling of Soft Tissues in Response to Stress Changes 519

the circumferential stress is ofthe order of 106 dyn/cm 2 , the longitudinal stress
is smaller but of the same order of magnitude, the radial stress is an order of
magnitude smaller. The maximum shear stresses in the arterial wall due to
blood pressure are, therefore, 5 orders of magnitude larger than the shear
stress imposed by the blood on the vessel endothelium. Yet we know that
pressure overloading induces thickening of the vessel wall (usually with de-
creased vessel lumen), whereas the flow overloading induces enlargement of
the vessel diameter. This suggests that there is somethig very special about
the endothelial cells, and that the biochemical-biomechanical pathway must
be very carefully studied.

Healing of Surgical Wounds

Healing of wounds and development of new strength reveal structural and
mechanical changes in a tissue. The author and his associates (Lee et aI., 1985)
have measured the strength of surgically anastomosed arteries of the rat
sutured with a polyglycolic acid surgical thread (Dexon). The abdominal
aortas and the carotid arteries were severed, sutured, and then the wounds
were closed and the animal healed. After a specific period of time, the vessels
were taken out and tested in uniaxial loading condition. The stress-strain
relationship of the vessels was measured, and then the vessels were pulled to
failure. It was found that the strength of the anastomosis was the lowest in
about 4 months. Figure 13.5:6. In the first day, the force at failure was about
the same as that of the control. Then the strength decreased with time, until
a minimum was reached in 4 to 6 months; at which time the tensile force to
failure was about 25% of the control for the carotid artery and 49% of the
control for the abdominal aorta. The corresponding values of the tensile stress
at failure were 17 and 11%, respectively. The differences between ultimate
forces and stresses were caused by the thickening of the vessel wall in the
neighborhood of the suture line in the healing process. After 4 to 6 months,
the strength gained again. At 13 months, the strength ofthe anastomosis was
about the same as that of the control. The stretch ratio at failure was constant
through all periods.
Thus the healing ofthe artery after surgery appears to be quite slow: return
of the maximum tensile force to normal takes 12 months. In contrast, wound
healing in the gastrointestinal tract is much faster. Gottrup (1980) has
shown that when incisions were made in rat stomach and duodenum and
immediately resutured with polypropylene thread 6-0, the wounded tissue
can regain the breaking strength of the intact tissue in 5 to 10 days.
The development of the strength of an artery after surgery depends on the
process ofresorption of the suture, the regeneration of the smooth muscle and
the connective tissues, and the changes in structure in the process of resorption
and regeneration. It is believed, but not proven, that adequate stress is neces-
sary. Gottrup (1981) has attempted to show a correlation between the collagen
content and the gain in strength in wound healing in gastrointestinal tract.
 520 13 Biomechanical Aspects of Growth and Tissue Engineering

xl0 4 CAROTID ARTERY

8

CAROTID ARTERY
Ma xlmum Force
,.... 7
200 N tvlaxlmum Stress at Failure
E at Failure =
.....0 Cross Section Area
E E 6
Ol
.....
Ol

I.1J I.1J
a: 150 a:
:::> :::> 5
....J ....J
<I: <I:
lJ... lJ... 11

I-
4 11
I-
<I:
: 100 C/)
U C/) 3
a: I.1J
0 a:
lJ... 11 I-
C/)
2 2
:::>
2
50
11
" 2
:::>
X
2
<I: X 1
2 <I: 11
2
0 00
0 4 8 12 4 8 12
TIME (month) TIME (month)

FIGURE 13.5:6 Left: Force at failure for carotid artery as a function of time after
surgery. Failure loads of the controls at day zero are shown in solid black triangles.
The curve is a smooth cubic spline curve. Right: Stress at failure of carotid artery as a
function oftime after operation. Black triangles are for controls at day zero. From Lee
et al. (1985). Reproduced by permission.

Goldin et at (1980) have examined the change of various mechanical and

biochemical factors in the process of wound healing of digital tendon of
the rabbit. Theoretical analysis of tissue strength based on morphological,
rheological, and structural data needs to be developed.

Muscle Atrophy During Space Flight or Immobilization

Animals exposed to the weightless condition of space flight have demonstrated
skeletal muscle atrophy. Leg volumes of astronauts are diminished upon
return to earth. In flight vigorous daily exercise is necessary to keep astronauts
in good physical fitness over a longer period of time. Herbison and Talbot
(1985) reviewed the literature of muscle growth and noted the following: The
rate of muscle protein synthesis decreases within 6 h of immobilization ofrat
limbs (Booth, 1982); it returns to normal 6 hrs after remobilization (Booth et
at, 1982). Insulin appears to be an important hormonal factor for short-term
regulation and maintenance of protein balance in skeletal muscle (Goldberg,
1979). Immobilized muscle atrophies; but there is a marked difference between
13.6 Stress Field Created by Fibroblast Cells and Collagen Synthesis 521

stretched immobilized muscle versus muscles immobilized in the resting or

shortened position (Booth, 1982; Simrad et aI., 1982, Spector et aI., 1982).
When muscle is denervated, strong electric. stimulation may retard atrophy
(Karpovich, 1968). During limb immobilization and space flight, the blood
cortisone level is elevated (Leach and Rambaut, 1977). Passive tension or
repetitive stimulation retards protein degradation in isolated muscle prepara-
tion (Goldberg et aI., 1975). Muscle protein degradation may be induced by
thyroid hormones (Goldberg et aI., 1980). Inactivity of skeletal muscle induced
by joint immobilization or by pharmacologic or surgical denervation is asso-
ciated with an increase in extrajunctional acetylcholine receptors (Fischbach
and Robbins, 1971; Lavoie et al., 1976; Pestronk et aI., 1976).
Fundamentally, growth is a cell biological phenomenon at molecular level.
Stress and strain keep the cells in a certain specific configuration. Since growth
depends on this configuration as well as other factors, it depends on the stress
and strain (Coan and Tomanek, 1979; Faulkner et aI., 1983.)

13.6 Stress Field Created by Fibroblast Cells and

Collagen Synthesis
In a continuum, any local change in stress and strain affects the entire body.
In fact, it can be proven mathematically that if a simply connected body obeys
the linear law of elasticity and is free from body force, then if there exists a
finite domain in the body, no matter how small, in which all strains vanish,
then the strain must vanish in the entire body, which is thus stress free.
Similarly, in a field of flow of a Newtonian fluid in a simply connected domain
free from body force, one can show that if the velocity is identically zero in a
small finite region, then the flow is zero everywhere. Mathematically, this is a
property of the partial differential equations ofthe 'elliptic' type which governs
these continua. Conversely, if strain or flow is changed in one region, no matter
how small, then the whole field will be changed. Applying this mathematical
result to a living body, we infer that if one cell changes its volume or shape,
then every cell in the body is stressed and strained. Thus, in principle, the stress
field of each cell is influenced by other cells in the body no matter how remote
they are. The intensity of the signal, of course, decreases with increasing
distance between the cells.
A biological experiment by Harris et al. (1981) shows this beautifully. They
prepared a thick gel of reprecipitated collagen and planted two groups of
chicken heart fibroblasts in it at a distance of about 1-2 cm apart. When the
growing cells spread and cluster, they create a field of tension between the two
explants. The tensile stress stretches the collagen fibers into alignment. Figure
13.6:1 shows a photograph of the straight bundles of collagen that form
between the adjacent explants.
The phenomenon illustrated by Fig. 13.6:1 suggests that tissues such as
tendons, ligaments, organ capsules and dermal tessellations do not need to be
522 13 Biomechanical Aspects of Growth and Tissue Engineering

FIGURE 13.6:1 Harris et al.'s experiment on stress field created by growing cell colonies.
Two groups of cells separated by a distance of 1.5 em were grown in a collagen gel. In
the process of polymerization, the newly formed collagen fibers become aligned into
long axially oriented tracts interconnecting two centers of traction. Heart explants
from 8-day chick embryos after 96 h in culture. Scale bar, 1 mm. From Harris et al.
(1981), Nature Vol. 290, p. 251. Reproduced by permission.

laid down by direct secretion of collagen fibres into their eventual arrange-
ment. Harris et al. (1981) believe that fibroblast traction can rearrange and
repack collagen into these patterns, even beginning from a totally random
meshwork.
An earlier experiment by Harris et al. (1980) may help visualize what is
going on. They prepared very thin sheets of silicone rubber membrane by
briefly exposing a silicone fluid to a flame. The outer layer of the fluid was
cross-linked into a membrane of about 1 !lm thick. This membrane was then
lifted up and put in a tissue culture fluid and a fragment of embryonic chick
heart was placed on the membrane. Figure 13.6:2 shows what was seen after
48 hours. Under dark-field illumination, the bright radiating lines show the
wrinkles in the membrane, wrinkles similar to those in cloth fabric under
tension. In engineering this is known as 'tension field' in thin-walled structures.
These lines represent the lines of tension (a principal stress trajectory) in the
membrane. The membrane is compressed and wrinkled ('buckled' in engineer-
ing terminology) in the direction perpendicular to the tension line; and since
the membrane is very thin the compressive stress (the so-called critical buck-
ling stress) is very small and can be treated as zero. Figure 13.6:2 is a graphic
demonstration of the tension field created by the fibroblasts. The rays are the
principal tensile stress trajectories.
When the area of contact between the fibroblasts and the membrane
substrate was examined under the microscope, it was seen that the cells adhere
to the membrane along the outer border of the area of contact. Behind the
border (of width 5 to 25 !lm) the membrane was wrinkled, with the direction
of the wrinkles perpendicular to the rays outside, indicating that the mem-
brane was under compression beneath the cells. These facts are interpreted as
follows (Fig. 13.6:3): The outgrowing cells are rather flat (like pancakes). When
they thicken and round up (into balls), the edge pulls the membrane substrate
in, creating compression wrinkles under the cells and tension rays radiating
away from the cells.
Thus the fibroblasts first grow out thin and/or elongated and then round
up and contract into a ball. In so doing, depending on the circumstance, it
FIGURE 13.6:2 Harris et al.'s experiment on cells grown on very thin membranes of
silicone rubber. The contraction of the cells caused wrinkles in the rubber membrane.
Photograph shows a low magnification, dark-field illumination view of a chick heart
explant that had been spreading on the membrane for 48 hours. The bright radiating
lines are the tension wrinkles. The bar is 1 mm long. From Harris et al. (1980), Science,
Vol. 208, p. 178. Reproduced by permission.

FIGURE 13.6:3 Harris et al.'s explanation of how the growing cells interact with the thin
membrane on which they rest. (A) Diagrammatic side view of an individual cultured
fibroblast distorting and wrinkling the elastic silicone substratum upon which it has
spread and is crawling. (8) Diagrammatic side view of the margin of an explant whose
cells are spreading outword on a silicone rubber substratum. The traction forces
exerted by the outgrowing cells compress the rubber sheet beneath the explant and
stretch it into long radial wrinkles in the surrounding area. From Harris et al. (1980).
Science, Vol. 208, p. 177. Reproduced by permission.
524 13 Biomechanical Aspects of Growth and Tissue Engineering

may stretch the surrounding medium and induce tension in a certain direction.
The tension may serve to line up the procollagen and then form collagen fibers.
If there were two colonies of fibroblasts, each anchored in the medium, then
their contraction may induce tension between these colonies and eventually
create collagen fibers linking these colonies. This is believed to be the cause
of the condition shown in Fig. 13.6:1. It resembles Paul Weiss's (1955) "center
effects" theory of nerve growth.
The apparent contraction of collagen networks around deep wounds,
burns, and surgically implanted prostheses may be caused by the same pro-
cess. The idea is very suggestive and may have many applications.
The concept that an internal stress field is created by the growing cells
themselves adds an exciting dimension to the stress modulation of growth. In
engineering language it adds a feedback control channel. It suggests that with
a proper design, tissue growth can be engineered. See Secs. 13.8-13.10.

13.7 Growth Factors

Biological growth is a cellular activity. At the foundation, its study belongs

properly to molecular biology, genetics and biochemistry. From molecules to
organism, mechanics plays a role because there are many steps that are
influenced by flow and stress.
To study growth from the point of view of physiology is to begin at the
macroscopic end. One accepts the existence of a homeostatic state of an
individual organism, and asks how does the homeostatic state change when
the environment and the internal stress are changed.
The trend of research is to bring the microscopic and macroscopic
views together. In cellular biomechanics, biochemistry, cell biology and
biomechanics are brought together. In biomechanics of growth, growth
hormone, and growth factors are essential players. But the subject is vast, and
it is beyond the scope of this book to describe it in any detail. In the following,
a bare outline of the historical highlights is given to serve as an introduction.

Growth Hormone
The term hormone was chosen by Bayliss and Starling in 1902 from a Greek
work hormaein meaning "I arouse to activity". Earlier, in 1825, Leuret and
Lassaigne showed that in the dog, acidification of the upper intestine led to
secretion of pancreatic juice. In 1849, Berthold observed that the atrophy of
the comb and loss of male behavior in cockerels following castration could
be prevented by grafting a testis into the abdominal cavity. These and other
experiments lead to the idea that endocrine glands regularly release substances
into the blood stream that are necessary for the normal development and
function of other parts of the body. In 1902, Bayliss and Starling demonstrated
that an extract of dog intestinal mucosa could, on intrayenous injection into
13.7 Growth Factors 525

the dog, cause intense pancreatic secretion. They gave the name secretin to
the agent in the extract that caused the pancreas to secrete, and the term
hormone to describe similar messengers.
The following are endocrine glands: hypothalamus, hypophysis (pituitary
gland), thyroid, parathyroid, adrenal, islet of Langerhans of the pancreas,
ovary, testis, pineal gland, placenta. Growth hormone (GH) was first found
in pituitary gland, growth hormone-releasing factor (GHRF) and growth
hormone release-inhibiting hormone (GHRIH, also called somatostatin) exist in
hypothalamus, but was first found from pancreatic tumors.
The great stimulus to an understanding of the pituitary was a burst of
interest in acromegaly toward the end of the nineteenth century. The interest
was stimulated by Pierre Marie in 1886 who described and named acromegaly
(enlargement of the extremitas of the skeleton-the nose, jaws, fingers, and
toes) as a disease (see Raben, 1959). In 1921, Evans and Long showed that a
pituitary extract could induce gigantism in rats. Cho-Hao Li and associates
(1948) then produced highly purified crystalline growth hormone in 1940's,
and Li and associates determined the complete structure of the molecule in
1971.
The growth hormone accelerates growth, increasing the size of all organs
and promoting the growth of bone before closure ofthe epiphyses. It increases
protein formation, decreases carbohydrate utilization, and increases the
mobilization offat for energy use (Knobil and Greep, 1959).
In growth hormone experiments, the rat has been favored because it is
exceptional in two ways. The epiphyses of almost all of its long bones remain
open throughout life, and growth continues for as long as good nutrition and
good health prevail. It is also peculiar in responding indiscriminately to the
growth hormone of all other mammals thus far explored.
Growth in human dwarfs in response to human growth hormone is
normally proportioned, but there is no sexual maturation. No features of
gigantism or acromegaly have yet been described, but prolonged treatment
with large doses has not yet been attempted.

Growth Hormone Releasing Factor

The discovery of a peptide with specific and unusually potent stimulatory
effect on growth hormone secretion was achieved in 1982. The source of this
peptide was not the hypothalamus, but from pancreactic tumors obtained
from two patients. In both cases, the patient had symptoms of acromegaly.
In one patient, pituitary enlargement was diagnosed and transsphenoidal
surgery undertaken; the pituitary showed hyperplasia of the GH-secreting
cells, the symptoms of acromegaly persisted postoperatively, and a pancreatic
tumor was subsequently found. The second patient had a normal pituitary
and two pancreatic tumors. In both cases extracts of the pancreatic tumors
revealed a peptide of similar structure. Within a few weeks of each other,
Guillemin et al. (1982) and Vale et al. (Rivier et al., 1982) revealed the structure
of the human pancreas growth hormone releasing factor.
526 13 Biomechanical Aspects of Growth and Tissue Engineering

Guillernin's Language of Polypeptides

With modern techniques of cell biology, many growth factors have been
identified. Peptides can be recognized by radioimmunoassay techniques and
visualized in cells that produce them by immunocytochemical methods.
Neurons can be stained for various hormone-releasing factors. The mor-
phologist can work out a complete map for the distribution of each kind of
peptide-containing neuron in the brain and spinal cord. Palkovits (1984) has
shown that particular peptides are not confined to specific regions but are
instead widely distributed in patterns which have no known or obvious
functional correlations. Guillemin (1985) likened the peptides in the brain as
a language. He concludes that the central nervous system is a gigantic multi-
endocrine organ secreting within itself or to very closely attached structures
(the pituitary gland, the periventricular organs) multiplicity of peptide mole-
cules, most of them are also known to be secreted by nonneuronal endocrine
cells or tissues at the periphery (and thus not specifically neuropeptides). He
said that we do not really know the role(s) of these peptides in the brain. But
we know what many of these same peptides do at the periphery. So he sums
it up by saying that we know a local pituitary "dialect" for some pep tides, a
local pancreatic "dialect", but we do not know the brain "dialect(s)".

NGF,EGF
Then 1986 Nobel Prize in Medicine was awarded to Rita Levi-Montalcini
and Stanley Cohen for their discoveries of nerve growth factor (NGF), and
epidermal growth factor (EGF), respectively.
The story began when Levi-Montalcini was invited to St. Louis by Viktor
Hamburger. In St. Louis they repeated an experiment by Elmer Bueker in
which he had noted that mouse sarcomas transplanted to chick embryos
induced vivid nerve growth from the chick embryo nervous system into the
tumor. Levi-Montalcini noted that sensory and sympathetic ganglia that
remote from the tumor also become enlarged. So she suggested and proved
that a diffusible substan,ce (NGF) exists that stimulates nerve growth. Joined
by Cohen, they began to characterize this material enzymatically, and dis-
covered that the addition of snake venom greatly enhanced the nerve
growth-stimulating activity of the tumor extract. They soon realize that the
snake venom itself was able to stimulate nerve growth. They reasoned that if
venom glands in snakes contained NGF-like activity, perhaps salivary glands
of mammals might also be active. Thus it was discovered that the sub-
mandibular gland of adult male mice is an incredibly rich source of NGF.
Cohen then isolated NGF. Levi-Montalcini developed an NGF bioassay.
Stanley Cohen injected salivary gland extracts into newborn mice. He
found precocious opening of the eyes and precocious tooth eruption. He soon
realized that the salivary gland extract contained a second biologically active
factor different from NGF. He isolated this factor, named it epidermal growth
13.8 Significance of Zero-Stress State 527

factor (EGF), and demonstrated its effect on the epithelial cells of the skin,
mucous membrane, cornea, mammary gland; and several other types of cells
such as connective tissue, cartilage, liver, smooth muscle, and hormone-
producing cells. Typical references are Cohen (1959), Levi-Montalcini and
Angeletti (1966). See Schreiber et a!. (1986).

Angiogenesis: ECGF, FGF

Folkman et a!. started the study of angiogenesis in the 1960's by observing
that transplanted tumors in mouse mayor may not grow depending upon
whether the transplant stimulates the growth of new capillaries or not (Folkman,
1985, Thomas et a!., 1985). Using cell culture, rabbit cornea, biopolymers, and
chick embryo chorioallantoic membranes as media to observe capillary growth,
they saw new capillaries arise as offshoots of established vessels. In the
presence of an angiogenic stimulus, endothelial cells begin to extrude them-
selves from their parent vessels by releasing specific collagenases that cause
local lysis of the vascular basement membrane. The endothelial cells elon-
gate, undergo directional locomotion toward the angiogenic stimulus, and
form a spout in which endothelial cells proliferate. A lumen is then formed.
Next, two sprouts join to form a loop through which blood begins to flow.
New sprouts branch from the loop and a capillary network is gradually
constructed.
Folkman et a!. found that the tissue mast cells are associated with increased
capillary growth. Since mast cells secrete heparin, they were led to the discovery
that heparin can potentiate angiogenesis: it has a high binding affinity for the
endothelial growth factors (ECGF). This discovery led to the isolation and
purification of ECGF by heparin-affinity chromatography in 1986. Now an
entire class of ECGF has been identified. They are peptides, and are related
either to the basic fibroblast growth factor (FGF) or to acidic FGF. Genes
for basic FGF from the brain and the ECGF, a precursor of acidic FGF, were
cloned. Three other angiogenic factors were reported: angiogenin, transform-
ing growth factor alpha, and transforming growth factor beta, all purified from
human tumor cells, but were later found in a variety of normal cells.
Steroids that inhibit angiogenesis in the presence of heparin have been
found. The tetrahydro derivatives of cortisone were found to be the most
potent angiogenesis inhibitors.

13.8 Significance of Zero-Stress State: Changes Reveal

Nonuniform Remodeling of Tissue

Blood vessel remodeling can be revealed in many ways, such as:

(1) Overall changes in vessel lumen diameter, wall thickness/lumen diameter
ratio, connective tissue/smooth muscle ratio.
528 13 Biomechanical Aspects of Growth and Tissue Engineering

(2) Nonuniform growth in different parts of the vessel wall. Change of opening
angle at zero-stress state.
(3) Mechanical property changes in stress-strain relationship, smooth muscle
tone, viscoelasticity, and constitutive equation.
(4) Cellular and molecular changes. Cell morphology. Ultrastructure. Bio-
chemical changes.
Among these ways the change of zero-stress state has a unique significance,
because the zero-stress state is the only state at which every cell and extra-
cellular material is at natural shape. When the tissue is remodeled because of
growth or resorption of the cells and extracellular matter, we would like to
see how do the shape and size of these matters change. For this purpose it is
best to compare everything at zero-stress state. To compare them in any other
state is to bring in the complication of deformation due to internal stress.
In Fig. 11.2:3 we have seen that the zero-stress state of arteries change a
great deal due to a step change of blood pressure. The change in opening angle
is large and easy to observe.
In Fig. 13.8:1 a simple illustration of several types of remodeling is shown.
Uniform changes lead to elongation, widening, shortening, shrinking. But
bending must be caused by a nonuniform growth. The opening angle changes
shown in Fig. 11.2:3 are due to bending, hence they exhibit nonuniform
remodeling of the blood vessel wall caused by hypertension.
Conceptually, tissue growth is expected to be nonuniform because cells die
or divide one by one. Death of a cell leaves a vacancy which will be filled by

GROWTH: Change of cellular and extracellular

mass and configuration
Original zero-stress configuration:

Elongated

Widened I I I I
Resorbed

Proliferated

Bent

o
FIGURE 13.8: 1 Illustration that the remodeling of an organ is best described by change
of its zero-stress configuration.
13.9 A Hypothesis on Growth 529

neighboring cells, in the process these cells deform and create residual stresses
in themselves. To see how much these neighboring cells have changed, we
must make suitable number of cuts to render the tissue stress-free.
In Sec. 11.3 we have explained that a consequence of the residual stress
coming from the opening angle of the blood vessel is to reduce the stress
concentration in the inner wall region at a normal physiological blood pres-
sure. Calculations of the blood vessel wall stress distribution by Fung
(1984, 1985), Chuong and Fung (1986), Vaishnav and Vossoughi (1986),
Takamizawa and Hayashi (1987) (see ref. in Ch. 11) have shown the con-
sistency of the opening angle and the hypothesis of uniform circumferential
stress distribution in arterial wall at homeostatic condition at normal blood
pressure. When the blood pressure is raised above the normal value, the tensile
stress at the inner wall is increased more than that at the outer wall. The inner
wall hypertrophies more than the outer wall, and the opening angle at zero-
stress increases. As remodeling proceeds, the opening angle continues to
change until a new homeostatic condition is established, at which the circum-
ferential stress distribution is again quite uniform at the new steady blood
pressure.
Conversely, if one assumes that tensile stress is uniformly distributed
throughout the blood vessel wall at the homeostatic (physiological) condition,
then one can calculate the needed residual strain, and the opening angle of
the zero-stress state. The results have some resemblance with experimental
data. Fung (1984, 1985) examined the dense body distribution in the vascular
smooth muscles of the arteries of the rabbit mesentery and showed that they
are quite uniform at physiological blood pressure. Similarly, the sarcomere
length of the myocardium in the left ventricle has been found by Hort (1960)
and others (see Fung, 1984b, p. 61) to be fairly uniform across the ventricular
wall. But the scatter of the data is large. The ratio of the standard deviation
to the mean are so large that any attempt to prove uniformity is difficult.
Furthermore, a sarcomere length distribution that is spatially uniform at
diastole may become nonuniform in systolic period. Hence in an organ in
which stress and strain vary cyclically there is a fundamental question of how
to define homeostasis. Could it be better defined in terms of work done per
unit volume of tissue per unit time? Or in terms of oxygen demand? Or
microcirculatory blood flow? These questions are topics for further study.

13.9 A Hypothesis on Growth

To help theoretical and experimental studies, I would like to propose a

hypothesis. The hypothesis is that the remodeling of the blood vessel involv-
ing growth or resorption of cells and extracellular materials is linked to
the stress in the vessel. We recognize that (1) the transport of matter through
cell membranes by active or passive mechanism depends on strain in the cell
membranes, (2) the granular-to-fibrous transformation of actin molecules and
530 13 Biomechanical Aspect~ of Growth and Tissue Engineering

growth rate

stress

FIGURE 13.9:1 The author's proposed stress-growth law.

the function of actin-myosin cross-bridges are strain dependent, (3) chemical

reaction rate depends on pressure, stress, and strain. Hence, specifically, we
propose that growth depends on stress and strain in a manner shown in Fig.
13.9:1. There exist equilibrium states of stress, a, b, c, ... At a, an increase of
stress causes growth, a decrease of stress causes resorption. At b or c an
increase of stress causes resorption, a decrease of stress causes growth. Joining
the trends at a, b, c together by a smooth curve, I shall express the growth
rate by an equation
(1)
Here mis the rate of growth, s is stress, C, kl' k2' k3' a, band c are constants.
If kl' k2' k3 are less than one, the curve is very steep at a, b, c. If kl' k2' k3 are
greater than one, the curve is very flat at a, b, c. Taking this as an experimental
hypothesis, then the problem is to determine the constants as functions of
growth factors and other physical, chemical, and biological stimuli, and
whether s should be dilatational, or deviatoric, or principal stresses or their
invariants.
Some explanations are due. The existence of homeostatic states is axiomatic
in physiology. The normal physiological condition of blood vessel behaves
like point a. Hypertrophy due to increased blood pressure agrees with the
condition to the right of a. Astronauts at zero G seems to be in a condition to
the left of a (Mack et ai., 1967). A metal screw installed in a bone may reach
the state b: an over-tightening leads to resorption. Tissue culture in stress-free
state is possible; hence I give ma positive value near the s = 0 axis. Joining
that to the point a requires a zero-crossing homeostatic state c. A step
reduction of blood pressure so that the stress is reduced to a value below c
will induce growth.
Some consequences of this hypothesis are:
1) At homeostasis, all collagen fibers have the same stress, all elastin fibers
have another uniform stress, all sarcomeres in skeletal or cardiac muscle
cells have the same length, all dense bodies in smooth muscle cells have
the same spacing.
13.9 A Hypothesis on Growth 531

2) If the homeostatic condition is described by the equilibrium state a, then

in hypertension, growth starts at the inner wall, and the opening angle at
zero-stress would be increased.
The first conclusion follows our hypothesis if the hypothesis holds for the
collagen and elastin and muscle fibers. If this conclusion is true, then it is
tremendously important to the derivation of constitutive equations. We have
used it in Sec. 11.9.
The experimental results on the sarcomere length and dense body spacing
mentioned in Sec. 13.8 support this conclusion.
The second conclusion is supported by the features illustrated in Fig. 11.2:3.
A feature deducible from the results of hypertension experiment presented
in Fig. 11.2:3 also lends support to our hypothesis: If we plot the opening angle
of various segments of the rat aorta against the number of days post surgery,
we obtain the result shown in Fig. 13.9:2. If we compute the hoop stress in the
artery by the formula
pr
(j=- (2)
h

200
~ 4%L

~'\Q. • 20%L
-
::l
150
. ~ • 40%L
(J ""1!r-- c

- . •.
0
Q) 100
:5!
CI) k··~.
50
: :l
OQ;' L ....
";-=-.
• _.~ •
o
.!o,
~Q)

10 20 30 40 50
C)Q) 200 0
CO
«- 60%l
III

• 80%L
C) 150
• • 100%L

K.·""'·- .·
c:
c: 100
••
Q)
Q. .~
o 50 .~o
" ~~ ~
o 0

o 10 20 30 40 50

Number of Days Postsurgery

FIGURE 13.9:2 The course of change of the state of zero-stress of rat aorta following
aortic banding described in Sec. 11.2 and shown in Fig. 11.2:2. The opening angle
(inside cut) is plotted against the number of days following surgery. %L marks the
location of a section along the aorta, see Sec. 11.2, Fig. 11.2:3.
532 13 Biomechanical Aspects of Growth and Tissue Engineering

1 <,-------------------------------------,
_ -<;J-- 4% L
- . - 20"1. L

1 2 ~
11!(0
V1\~
I'
I ,
" Sl
1.0 o" . • ~----
_---
• 0
•

-
f/)
f/)

....C1>
( J) 1. 6 r-------- -- ------------------------,
--0-- 40% L
a.
o - . - 60% L
o
::I:
r:
nl--.
00
~.Q.
1.0 10
"C
C1>
N
nl
....E 12 ,--------------------------------------.
o
z - - 0-- 80% L
- . - 100"1. L
1.0 "
,
.
\

08 '"" "-
D

. ..... - : - - - - - JI.. - - --------0

06 ...... ~
•• •----------.
0.4 '----~--"---~--"--~-'----~--"------'
o 10 20 30 40 50

Number of Days Postsurgery

FIGURE 13.9:3 Change of hoop stress in the aorta following the occlusion of aorta
mentioned in Fig. 13.9:2.

where p is the blood pressure, r is the lumen radius, h is the wall thickness, a
is the mean circumferential stress, and plot it against the number of days post
surgery, we obtain the result shown in Fig. 13.9:3. Now the hoop stress is the
mean circumferential stress in the vessel wall. If we assume that the maximum
circumferential stress in the vessel wall is proportional to the mean, then the
stress that drives the hypertrophy and change of opening angle is proportional
13.10 Engineering of Blood Vessels 533

to the hoop stress. If, in addition, we assume that in the neighborhood of the
equilibrium point a, rh is proportional to s - a, then the curve of rh vs days
will look like the hoops stress vs days curve, Fig. 13.9:3. The opening angle
change is due to a cummulative effect of rh, i.e., IY. is proportional to Srh dt.
Thus the curve in Fig. 13.9:2 should be the result of integrating the curve in
Fig. 13.9:3, as it seems to be.
The choice of the quantity s for the abscissa is still debatable. In the
discussion of Wolff's law of the bone, Sec. 13.4, we have seen advocates for
principal stress in tension or compression, von Mises stress, mean stress, strain
energy, and full stress tensor. The corresponding rh could be the rate of increase
of tissue mass, calcified tissue, or fabric tensor.
Validation of the hypothesis is a task for the future.

13.10 Engineering of Blood Vessels

Many vascular reconstruction surgeries involve the use of some type of graft
as a conduit to bypass an occluded arterial segment. The graft may be natural,
such as saphenous vein or internal mammary artery, or synthetic, such as
dacron. Use of saphenous vein has been very successful with 90% long-term
patency rates. Large synthetic vessels have been quite successful, but smaller
synthetic vessels (say, less than 6 mm in diameter) have difficulty remaining
patent over a period of several years. Among many reasons for this difficulty,
two principal ones are (1), cellular hyperplasia at or around the junction be-
tween the vascular graft and the native artery, leading to progressive stenosis
(Clowes et al., 1985), and (2), failure to grow a layer of endothelial cells to cover
the graft surface to protect it against blood clotting. Greater understanding
of these phenomena must be obtained.

Endothelial Cell Seeding

Stanley et al. (1985) and Zamora et al. (1986) have shown that pre-seeding of
synthetic vascular graft improves graft patency. Davies et al. (1986) showed
that turbulence enhances cell proliferation. Jarrell et al. (1986) showed that
microvascular endothelial cells can be isolated from fat tissue, and deposited
on synthetic polymer such as polystyrene and dacron by gravity. Adhesion
develops within a few minutes of incubation. Adhesion is improved by coating
the surface with fibronectin. The rate of spreading varies greatly with the
nature of the surface. With suitable optimization, it is possible to cover the
surface with a confluent monolayer of endothelial cells in less than an hour
following incubation.
Litwak et al. (1988) developed a small diameter, compliant vascular prosth-
esis using polyurethane urea as basic material. A surface active component
polydimethylsiloxane-based polyurethane is added to obtain a composite
material. Grafts are extruded as multilayered tubes. The inner, blood contact-
ing layer presents an open foam face. A solid impervious layer, adherent to
534 13 Biomechanical Aspects of Growth and Tissue Engineering

the inner foam, provides the graft with strength and prevents fluid transfer.
An outer layer of textured foam serves as a tissue anchor. Optimal bulk and
biocompatibility properties for vascular implants have been obtained.

Autologous Blood Vessel Substitute

Bowald et al. (1980), Greisler et al. (1985) have formulated a research hypothesis
that a normal arterial wall can be formed through natural repair processes if
a scaffold of a vascular graft is made of a slowly disappearing, nontoxic mesh.
Galletti et al. (1988), Gogolewski et al. (1987), and Yue et al. (1988) have studied
transient biocompatible scaffolds for regeneration of arterial wall. The chal-
lenge is to design bioresorbable tubular fabrics which at the time of implanta-
tion has suitable mechanical strength and anticlotting property, but lead, after
the disappearance of the synthetic polymers, to a regenerated vessel of adequate
strength and hemocompatibility, without rupturing at any time. Th~ develop-
ment of the new blood vessel substitute involves simultaneous decay of the
polymer and growth of cellular constituents.
Galletti et al.'s experiments have yielded encouraging results. They used
homopolymers such as polydioxanone or copolymers such as composites of
polyurethanes and polylactides or fibrin, or fast resorbing polyglycolide yarns
coated with "retardant" polymers, e.g., a mixture of polylactide and poly 2, 3
butylene succinate. They implanted 8 mm internal diameter, 8-9 cm long
aortic graft in dogs and found that the graft functioned effectively for six
months when the animal was sacrificed. The polymers have largely decayed
at three months. No infection was found with fully bioresorbable prosthesis.
Galletti and other authors state that the compliance of the original prosthesis
is a critical factor in obtaining the development of a circumferentially oriented
smooth muscle layer and the deposition of elastin.
Initial success has to be bolstered by true understanding. On the chemical
side, the use of peptide factors that promote cell recruitment and proliferation
to build tissue of host origin must be studied. Types of cells recruited into the
implant must be identified. Angiogenesis and development of microcirculation
in the implant must be clarified. The reaction of the implant and cells to a
variety of physical, chemical, and biological stimuli such as activated white
blood cells and bacteria must be understood. The design and manufacturing
of the bioresorbable polymer scaffold must be optimized. Mechanical forces
involved must be understood. But the road to success is discernible.

13.11 Tissue Engineering of Skin

Each year more than 100,000 people are hospitalized and about 10,000 die
from burn injuries in the United States. Conventional treatment for severe
burns includes early excision of the eschar, coverage of wounds with cadaver
allograft, or topical anti-microbial agents and dressings. Definitive wound
13.11 Tissue Engineering of Skin 535

closure is achieved by grafting skin from an unburned area on the patient.

These established procedures for victims of severe burns result in long hospitali-
zations and multiple surgery. Greater availability of skin substitutes for
wound closure offers the prospects for reduction or elimination of the donor
site for skin autograft, fewer surgical procedures, shorter hospitalization time,
improved functional and cosmetic results, and substantial economic benefits
in reduced health care costs. Skin substitutes can also be used for testing the
safety of consumer products that contact the body, of foods and drugs, and
of occupational and environmental hazards; including skin irritancy test, skin
corrosion, and percutaneous absorption, and skin inflammation.

Skin Substitute Made of Patient's Own Cells

A permanent skin substitute can be created by growing the patient's own skin
cells in a collagen scaffold. The purpose is to heal a deep skin wound of a
patient with his own cells rather than his own skin, thus minimizing the
trauma of an accident to a much smaller degree. The collagen scaffold is
biodegradable. When the skin substitute is grafted surgically, and becomes
well vascularized so that microcirculation is reestablished, then a new perma-
nent skin is obtained. Such a substitute for skin autograft has been developed
which presently consists of cultured human epidermal keratinocytes attached
to a porous and resorbable collagen and chondroitin sulfate membrane that
is populated with human fibroblasts (HF). Initial application of the substitute
to athymic mouse has shown advantages such as a large saving of donor's
skin, and a reduction of the contraction of the healing wound.
A variety of materials that are obtained from either in vitro or ex vivo
preparations have been proposed. The use of tissue culture techniques for
normal human epidermal keratinocytes can accomplish expansion ratios of
area coverage that can exceed lOOO-fold in a period of 3-4 weeks. Cultures of
these kinds will form multilayered sheets and have been shown to provide
wound closure after application to excised full thickness burns. See Bell et al.
(1983), Boyce and Hansbrough (1988), Burke et al. (1981), Gallico et al. (1984),
Ham (1981), Pittlekow and Scott (1986), Yannas and his associates (1980a, b,
1982), Hefton et al. (1987).

Biodegradable Scaffold
Synthesis of collagen-GAG membranes was first accomplished by Yannas et
al. (1980a, b, 1982). Comminuted bovine collagen is partially solubilized in
O.05M acetic acid and coprecipitated with chondroitin-6-sulfate in a refrige-
rated homogenizer. The coprecipitate is cast into sheets and frozen. The frozen
coprecipitate is lyophilized overnight and cross-linked in a vacuum oven at
105°C and 10- 4 torr for 24 hours. Dry membranes may be stored for extended
periods of time.
Before tissue culture or grafting, dry membranes are rehydrated in O.05M
536 13 Biomechanical Aspects of Growth and Tissue Engineering

acetic acid, cross linked with 0.25% glutaraldehyde, washed exhaustively, and
stored in 70% isopropanol.
Pore size can be controlled by controlling the freezing temperature and the
concentration of starting materials. By controlling the freezing temperature
gradient, a layer of the substitute with smaller pores can be created by quick
freezing to a lower temperature.

Wound Contraction
Wound contraction is a principal problem with skin graft and wound healing.
The contraction creates tension in the neighboring healthy skin and distorts
the skin and the limb. A wound caused by burn and healed without graft may
contract to 15-20% of the original wound size. Skin substitute must be
designed to reduce this wound contraction. A thorough understanding of
growth and stress would have tremendous applications here. Cf. Doillon et al.
(1987), Kennedy and Cliff (1979), McGrath and Simon (1983), Snowden and
Cliff (1985).

Problems to Be Solved
Tissue engineering of the skin needs the solution of the following problems:
1) Creation of an optimal collagen scaffold with optimal pore size in
relation to cellular mechanics, strength, contraction, and adhesion, and opti-
mal compliance with regard to the controlling of contraction. If angiogenic
factors were needed, they can be immobilized in the scaffold.
2) Understand the contraction of healing wound. Pharmaceutical, chemical
(e.g. nerve growth factor) and mechanical means have been sought to control
wound contraction, but the degree of success has been rather limited. The
study of skin substitute provides a unique opportunity to understand and to
control wound contraction. Molecular mechanics of the actin fibers and cell
mechanics will playa central role. On the other hand, contraction is related
to growth and resorption. Since growth is modulated by stress (Sec. 13.9),
control of stress during the healing process is believed to be a key to the
solution of the contraction problem.
3) Vascularization of the skin substitute.
4) Cellular adhesion and the mechanical properties of the skin substitute.
It is known that the adhesive stress between cells is very sensitive to the
environmental parameters s~ch as pH, temperature, and chemicals or phar-
maceutical agents in the bathing fluid (Evans, 1985). Hence it is feasible to
design a fluid so that a desired strength of adhesion is obtained.

Future Prospects
One can think of many other tissues for engineering: cartilage, blood, nerves,
etc. To make living substitutes for them is an exciting objective. Although
Problems 537

nobody can say when such an objective will be achived, we are sure that a
sound engineering knowledge is a prerequisite.

Conclusion
Here we bring our discussion of biomechanics to conclusion. Biomechanics
helps us understand nature. It sensitizes us to observe, and to appreciate. It
is a gentle tool: sharp, precise, natural, and unavoidable. It is the foundation
of engineering for prosthesis, rehabilitation, tissue substitutes, trauma preven-
tion, and health.

Problems
13.1 To search for evidence of residual strains in bone, the place to look is around
the growth plate. To find any changes of the zero-stress state with respect to
time, one should consider the period of growth in young animals, or period of
healing of a fracture. Design a research plan to study this question.
13.2 The stress in living organisms is influenced by earth's gravitation. To study the
effect of stress on growth and change, one of the best ways is to study space
flight. Design experiments to study astronaut's bone demineralization and
cardiovascular deconditioning during space flight. (Cf. Anderson, S.A., Cohn,
S.H.: Bone Demineralization During Space Flight. The Physiologist, 28: 213-
217, 1985; Levy, M.N., Talbot, J.M.: Cardiovascular Deconditioning of Space
Flight. The Physiologist, 26: 297-303, 1983. Consult the !UPS and NASA
publications on Proc. Annual Mtg. of IUPS Commission on Gravitational
Physiology, as Suppl. to Physiologist.)
13.3 Aortic arch and the first couple of generations of pulmonary arteries are highly
curved. They have large opening angles at zero-stress state (often in the range
of 180° to over 360°). Locally, the geometry of these vessels may be approximated
by a torus, a curved shell with a planar circular centerline of radius Po and a
circular cross section of radius a. At the center of curvature of the centerline,
erect an· axis perpendicular to the plane of the centerline, and use it as the polar
axis of a cylindrical-polar frame of reference, with coordinates (p, 8, z) to describe
the location of points on he shell. The nature of the shell depends on the radial
coordinate p. For the outer shell, p > Po, the two principal curvatures of the
midsurface ofthe shell have the same sign. For the inner shell, p < Po, the signs
of the two principal curvatures are different. In differential geometry, the outer
shell is said to be elliptic, the inner shell is said to be hyperbolic. The stress
distribution in the shell in response to the pressure in the tube is different in
these two parts. Set up equations to analyze the stress distribution in the vessel
wall. For simplicity, assume that the vessel wall obeys Hooke's law of elasticity.
13.4 Using the equations derived above, make analysis to answer the following
questions: (1) If we assume that at the equilibrium condition subjected to certain
internal pressure, the circumferential stresses are uniformly distributed, what
would the incremental membrane stresses be when the internal pressure is
changed by an infinitesimal amount? (2) By reducing the internal pressure step
538 13 Biomechanical Aspects of Growth and Tissue Engineering

by step until it is zero, what residual stresses remain in the shell? (3) What are
the residual strains, and the opening angle at zero-stress state if the shell is cut
by the method described in Sec. 11.2?
13.5 A description of cancer growth. Let x(t) be the number of living cancer cells in
an organ at time t. Let q(t) be the fraction of cancer cells eliminated in the growth
period. Then in an infinitesimal period of time I1t the number of cancer cells
increased may be assumed to be proportional to [1 - q(t)]x(t). Hence
I1x = K(t) [1 - q(t)]x(t)l1t

when K(t) is the constant of proportionality. If the number of cancer cells at

t = 0 is x(O), then an integration of the equation above yields

x(t) = x(O) exp {I K(t) [1 - q(t)] dt}.

Experiments suggest that the integrand is ofthe form e-«t". In chemotherapy, IX

is proportional to the concentration of drug above the minimal effective con-
centration. Propose a way to evaluate the effectiveness of drugs through the
evaluation of IX and n.
13.6 The mechanics of cellular growth is beautifully discussed by Ingber et al. (1985,
1989). Follow up the literature and offer a critical discussion. Suggest a theory
of your own if you feel there is a need.
13.7 Consider further the aortic arch studied in Problems 13.3 and l3.4. Based on
the effects of centrifugal forces, curvature, Bernoulli's equation, and boundary
conditions, give a qualitative description ofthe flow field. Estimate the pressure
distribution, and wall shear stress. These variables change rapidly, of course,
with the activity of the left ventricle and the aortic valve. Describe the course
of variation of these variables thorough a cycle of heart beat.
13.8 Problems l3.3 and l3.4 are concerned with stress distribution in the vessel wall.
The endothelial cells seem to be very sensitive to the shear stress. What is known
about the effect of shear stress on the morphology of the endothelial cells, and
on the orientation and shape of the cell nuclei? What is known about the effect
of the shear stress on the mass transport of cholestrol: LDL, HDL?
13.9 The stress in the smooth muscle cells is important to the muscle tone and
contraction force and shortening. Describe the mechanical behavior of the
vascular smooth muscle in the aorta. The connective tissues, especially collagen
and elastin fibers, are stressed. How can these stresses be calculated?
13.10 Stress is a tensor. If one knows the principal stresses, one can easily compute
the shear stress acting on any plane. Consider a blood vessel in physiological
condition. Compare the maximum shear stress in the vessel wall with the surface
shear stress acting on the endothelium due to the flowing blood. How different
are they?
13.11 In response to stresses, how do the various components ofthe blood vessel wall
change instantaneously and in long term? If the stresses become abnormal, how
do the various components in the vessel wall remodel themselves? How does
References 539

the remodeling of one component interfere with that of another neighboring

component? How can the whole system remodel? Perhaps you cannot find
answers to some of your questions in the library. The'Ll it is necessary to
experiment or theorize. Formulate a research proposal to settle some of your
questions.
13.12 Blood rheology has a major effect on the phenomenon of flow separation, i.e.,
the failure of streamlines near a solid wall to follow the boundary of the solid
wall. Where in the circulation system of man are the sites of possible flow
separation? What characteristics of blood rheology affect separation? How can
these characteristics be modified in chemical condition?
13.13 What are the effects of flow separation on the pressure and shear stress in the
blood vessel? What are the effects of the stress distribution in the blood vessel
wall on the tendency of the blood vessel to remodel itself? If the flow separation
condition changes, how would the blood vessel remodel itself in the long run?

References

Alpert, N.R. (ed.) (1983). Myocardial Hypertrophy and Failure. Raven Press, New York.
Anversa, P., Vitali-Mazza, L., Visioli, o. and Marchetti, G. (1971). Experimental cardiac hyper-
trophy: a quantitative ultrastructural study in the compensatory stage. J. Molecular and
Cellular Cardiology 3: 213-227.
Arai, S., Machida, A. and Nakamura, T. (1968). Myocardial structure and vascularization of
hypertrophied hearts. Tohoku J. Exp. Medicine 95: 35-54.
Arai, S., Nakamura, I. and Suwa, N. (1976). Quantitative analysis of cardiac hypertrophy due to
pressure load in reference to the relations of blood pressure, left ventricular weight, and left
ventricular capacity. Tohoku J. Exp. Medicine 118: 299-309.
Bassett, C.A.L. (1978). Pulsing electromagnetic fields: A new approach to surgical problems. In
Metabolic Surgery, (H. Buchwald and R.L. Varco, eds), Grune and Stratton, New York,
pp.255-306.
Bayliss, W.M. and Starling, E.H. (1902). Proc. Roy. Soc. (London) 69: 352. J. Physiology 28: 325.
Becker, R.O. (1961). The bioelectric factors in amphibian limb regeneration. J. Bone Joint Surg.
43A: 6431.
Bell, E., Ivarsson, B. and Merrill, C. (1979). Production of a tissue-like structure by contraction
of collagen lattices by human fibroblasts of different proliferation potential in vitro. Proc.
Natl. Acad. Sci. U.S.A. 76: 1274-1278.
Bell, E., Sher, S., Hull, B., Merrill, c., Rosen, S., Chamson, S., Asselineau, D., Dubertret, L.,
Coulomb, B., Lapiere, C., Nusgens, B. and Neveux, Y. (1983). The reconstitution of living
Skin. J. Invest. Dermatol. 81(1), Supp. 2S-lOS.
Bevan, R.D. (1976). An autoradiographic and pathological study of cellular proliferation in rabbit
arteries correlated with an increase in arterial pressure. Blood Vessels 13: 100-128.
Booth, F.W. (1982). Effect of limb immobilization on skeletal muscle. J. Appl. Physiol. 52:
1113-1118.
Booth, F.W., Nicholson, W.F. and Watson, P.A. (1982). Influence of muscle use on protein
synthesis and degradation. Exercise, Sport Sci. Rev. 10: 27-48.
Bowald, S., Busch, C. and Erikson, I. (1980). Absorbable meterial in vascular prosthesis: A new
device. Acta Chir. Scand. 146: 391.
Boyce, S.T. and Hansbrough, J.F. (1988). Biologic attachment, growth and differentiation
of humen epiderman keratinocytes onto a graftable collagen and chondroitin-6-sulfate
membrane. Surgery 103(4): 421-431.
540 13 Biomechanical Aspects of Growth and Tissue Engineering

Bradley, K.H., McConnell, S.D. and Crystal, R.G. (1974). Lung collagen composition and syn-
thesis. J. Bioi. Chern. 249: 2674-2683.
Braunwald, E. (1980). Pathophysiology of heart failure. In Heart Disease: A Textbook ofCardio-
vascular Medicine. Saunders, Philadelphia, pp. 453-471.
Brody, J.S. and Buhain, W.J. (1973). Hormonal influence on post-pneumonectomy lung growth
in the rat. Respir. Physiol. 19: 344-355.
Burke, J.F., Yannas, I.V., Quinby, W.c., Bondoc, C.C. and Jung, W.K. (1981). Successful use of
a physiologically acceptable artificial skin in the treatment of extensive burn injury. Ann.
Surgery 194: 413-428.
Carter, D.R. (1987). Mechanical loading history and skeletal biology. J. Biomech. 20: 1095-1109.
Carter, D.R. (1988). The regulation of skeletal biology by mechanical stress histories. In Tissue
Eng. (R. Skalak and C.F. Fox, eds.), Alan R. Liss, New York, pp. 173-178.
Carter, D.R., Fyhrie, D.P. and Whalen, R.T. (1987). Trabecular bone density and loading history:
Regulation of connective tissue biology by mechanical energy. J. Biomech. 20: 785-794.
Carter, D.R. and Wong, M. (1988). Mechanical stresses and endochondral ossification in the
chondroepiphysis. J. Orthop. Res. 6: 148-154.
Cassidy, J.J. and Davy, D.T. (1985). Mechanical and Architectural Properties in Bovine Can-
cellous Bone, Trans. Orthop. Res. Soc. 31: 354.
Chambers, T.J. (1985). The pathobiology of the Osteoclast. J. Clin. Pathol. 38: 241.
Chuong, c.J. and Fung, Y.c. (1986). Residual stress in arteries. In Frontiers in Biomechanics (G.W.
Schmid-Schonbein, S.L.-Y. Woo, B.W. Zweifach, eds.), Springer-Verlag, New York,
pp.I17-129.
Cliff, W.J. (1963). Observations on healing tissue: A combined light and electron-microscopic
investigation. Roy. Soc. of London, Phil. Trans. B,233: 305
Clowes, A.W., Gown, A.M., Hanson, S.R. and Reidy, M.A. (1985). Mechanisms of arterial graft
failure. I. Role of cellular proliferation in early healing of PTFE prostheses. Am. J. Path. 118:
43.
Coan, M.R. and Tomanek, R.J. (1979). Regeneration of transplanted muscle with reference to
overload. In A. Mauro, (ed.). Muscle Regeneration, Raven Press, New York, pp. 509-522.
Cohen, S. (1959). Purification and metabolic effects of a nerve growth-promoting protein from
snake venom. J. Bioi. Chern. 234: 1129-1137.
Cohn, R. (1939). Factors affecting the postnatal growth of the lung. Anat. Rec. 75: 195-205.
Cowan, MJ. and Crystal, R.G. (1975). Lung growth after unilateral pneumonectomy: Quantita-
tion of collagen synthesis and content. Am. Rev. Respir. Disease 111: 267-276.
Cowin, S.c. (1984). Modeling of the Stress Adaptation Process in Bone, Cal. Tissue Int. 36, (Supp.
S99-S104.
Cowin, S.C. (1986). Wolff's law of trabecular architecture at remodeling equilibrium. J. Biomech.
Eng. 108: 83-88.
Cowin, S.c. and Van Buskirk, W.c. (1979). Surface bone remodeling induced by a medullary pin.
J. Biomech.12: 269.
Cowin, S.C. and Firoozbakhsh, K. (1981). Bone remodeling of diaphyseal surfaces under constant
load: theoretical predictions. J. Biomech. 14: 471.
Cowin, S.C., Hart, R.T., Balser, J.R. and Kohn, D.H. (1985). Functional adaptation in long bones:
establishing in vivo values for surface remodeling rate coefficients. J. Biomech. 12: 269.
Crystal, R.G. (1974). Lung Collagen: Definition, diversity, and development. Fed. Proc. 33:
2243-2255.
Culmann, C. (1866) Die graphische Statik. Meyer und Zeller, Zurich.
Davies, P.F., Remuzzi, S., Gordon, E.F., Dewey, C.F., Jr., Gimbrone, M.A., Jr., (1986). Turbulent
fluid shear stress induces vascular endothelial cell turnover in vitro. Proc. Natl. Acad. Sci.
83: 2114-2117.
Doillon, c.J., Hembry, R.M., Ehrlich, H.P. and Burke, J.F. (1987). Actin Filaments in Normal
Dermis and During Wound Healing. Am. J. Path. 126(1): 164-170.
El Haj, A.J., Skerry, T.M., Caterson, B. and Lanyon, L.E. (1988). Proteoglycans in bone tissue:
References 541

identification and possible function in strain related bone remodeling. Trans. Orth. Res. Soc.
13: 538.
Erdogen, F. (1983). Stress intensity factors. J. Appl. Mech. SO(4b), 992-1002.
Evans, E.A. (1985). Membrane Mechanics and Cell Adhesion. In: Frontiers in Biomechanics
(G. Schmid-Schonbein, S. Woo and B.W. Zweifach, eds.), Springer-Verlag, New York,
pp.I-17.
Fischbach, G.D. and Robbins, N. (1971). Effect of chronic disuse of rat soleus neuromuscular
junctions on postsynaptic membrane. J. Neurophysiol. 34: 562-569.
Faulkner, J.A., Weiss, S.W., and McGeachie, J.K. (1983). Revascularization of skeletal muscle
transplanted into the hamster cheek pouch: intravital and light microscopy. Microvasc. Res.
26: 49-64.
Folkman, J. (1985). Tumor angiogenesis. Adv. Cancer Res. 43: 175-203.
Frolich, E.D. (1983). Hemodynamics and other determinants in development of left ventricular
hypertrophy. Fed. Proc. 42: 2709-2715.
Fry, D.L. (1968. Acute vascular endothelial changes associated with increased blood velocity
gradients. Circ. Res. 22: 165-197.
Fry, D.L. (1977). Aortic Evans blue dye accumulation: Its measurement and interpretation. Am.
J. Physiol. (Heart, Circ. Physiol. 1) 232: H204-H222.
Fry, D.L., Mahley, R.W., Weisgraber, K.H. and Oh, S.K. (1977). Simultaneous accumulation of
Evans blue dye and albumin in the canine aortic wall. Am. J. Physiol. 233 (Heart Circ.
Physiol.) 2: H66-H79.
Fukada, E. (1974). Piezoelectric properties of biological macromolecules. Adv. Biophys. 6: 121.
Fung, Y.c. (1981). Biomechanics: Mechanical Properties of Living TIssues. Springer-Verlag, New
York.
Fung, Y.c. (1984). Biodynamics: Circulation, Springer-Verlag, New York.
Fung, y.c. (1985). What principle governs the stress distribution in living organs? In Biomechanics
in China, Japan, and USA. (y.c. Fung, E. Fukada, and J.J. Wang, eds.) Science Press, Beijing,
China, pp. 1-13.
Fung, Y.C. (1988). Cellular growth in soft tissues affected by the stress level in service. In: Tissue
Engineering (R. Skalak and C.F. Fox, eds.), Alan Liss, New York, pp. 45-50.
Fyhrie, D.P. and Carter, D.R. (1985). A Unifying Principle Relating Stress State to Trabecular
Bone Morphology, Trans. Orthop. Res. Soc. 31: 337.
Gaensler, E.A. and Strieder, J.W. (1951). Progressive changes in pulmonary function after pneu-
monectomy: The influence of thoracoplasty, pneumothorax, oleothorax, and plastic sponge
plombage on the side of pneumonectomy. J. Thorac. Surgery 22: 1.
Galletti, P.M., Aebischer, P., Sasken, H.F., Goddard, M.B. and Chiu, T.H. (1988). Experience
with fully bioresorbable aortic grafts in the dog. Surgery 103: 231.
Gallico, G.G., O'Conner, N.E., Compton, c.c., Kehinde, O. and Green, H. (1984). Permanent
coverage of large bum wounds with autologous cultured human epithelium. New Eng. J.
Medicine 311(7): 448-451.
Gogolewski, S., Galletti, G. and Ussia, G. (1987). Polyurethane vascular prostheses in pigs. Colloid
Polymer Sci. 265: 774.
Goldberg, A.L. (1979). Influence of insulin and contractile activity on muscle size and protein
balance. Diabetes 28: 18-24.
Goldberg, A.L., Etlinger, J.D., Goldspink, D.F. and Jablecki, C. (1975). Mechanism of work-
induced hypertrophy of skeletal muscle. Med. Sci. Sports 7: 148-261.
Goldberg, A.L., Tischler, L.M., DeMartino, G. and Griffin, A. (1980). Hormonal regulation of
protein degradation and synthesis in skeletal muscle. Federation Proc. 39: 31-36.
Goldin, B., Block, W.D. and Pearson, J.R. (1980). Wound healing oftendon. I. Physical, Mechan-
ical, and metabolic changes. II. A mathematical model. J. Biomech. 13: 241-256; 257-264.
Gospodarowicz, D., Bialecki, H. and Greenburg, G. (1978). Purification of the fibroblast growth
factor activity from bovine brain. J. Bioi. Chem. 253: 3736-3743.
Gottrup, F. (1980,1981). Healing of incisionaI wounds in stomach and duodenum. Am. J. Surg.
542 13 Biomechanical Aspects of Growth and Tissue Engineering

140:296-301;141:222-277.
Greisler, H.P., Kim, D.V., Price, J.B. and Voorhees, A.B. (1985). Arterial regeneration activity
after prosthetic implantation. Arch. Surgery 120: 315.
Griffith, A.A. (1920). The phenomena of rupture and flow in solids. Phil. Trans. Roy. Soc. (London),
Ser A, 221: 163-198.
Gross, S.B., Spindler, KP., Brighton, C.T. and Wassell, R.P. (1988). The proliferative and synthetic
response of isolated bone cells to cyclical biaxial mechanical stress. Trans. Orth. Res. Soc. 13:
262-263.
Gross, D. and Williams, W.S. (1982). Streaming potential and the electro-mechanical response of
physiologically moist bone. J. Biomech. 15: 277-295.
Guillemin, R. (1985). The language of polypeptides and the wisdom of the body. The Physiologist
28: 391-396.
Guillemin, R., Brazeau, P., Bohlen, P., Esch, F., Ling, N., and Wehrenberg, W.B. (1982). Growth
hormone releasing factor from a human pancreatic tumor that caused acromegaly. Science
218: 585-587.
Ham, R.G. (1981). Survival and growth requirements of nontransformed cells. In: The Handbook
of Experimental Pharmacology 57: 13-88.
Harrigan, T. and Mann, R.W. (1984). Characterization of Microstructural Anisotropy in Ortho-
tropic Materials Using a Second Rank Tensor, J. Material Sci. 19: 761-767.
Harris, A.K, Wild, P. and Stopak, D. (1980). Silicone rubber substrata: A new wrinkle in the
study of cell locomotion. Science, 208: 177-179.
Harris, A.K, Stopak, D. and Wild, P. (1981). Fibroblast traction as a mechanism for collagen
morphogenesis. Nature 290: 249-251.
Hart, R.T. (1983). "Quantitative Response of Bone to Mechanical Stress", (Doctoral Dissertation)
Case Western Reserve University, Cleveland, Ohio.
Hayes, W.C. and Snyder, B. (1981). Toward a quantitativeformulation ofWolfT's law of trabecular
bone. In Mechanical Properties of Bone, (S.c. Cowin, ed.), Am. Soc. Mech. Engineers, New
York, Pub. No. AMD-Vol. 45, pp. 43-68.
Hefton, J.M., Madden, M.R., Finkelstein, J.L., Oefelein, M.G., LaBruna, A.N. and Staianao-
Coico, L. (1987). The grafting of cultured human epiderman cells onto full-thickness wounds
on pigs. J. Burn Care 19: 29.
Hempstead, B., and Chao, M.V. (1989). The nerve growth factor receptor: biochemical and
structural analysis. Recent Prog. Hormone Res. 45: 441-466.
Herbison, GJ. and Talbot, J.M. (1985). Muscle atrophy during space flight: Research needs and
opportunities. The Physiologist 28: 520-527.
Hort, W. (1960). Makroskopische und mikrometrische untersuchungen am Myokard verschieden
stark gefullter linker kammern. Virchows Arch. Path. Anat. 333: 523-564.
Hudlicka, O. (1982). Growth of capillaries in skeletal and cardiac muscle. Circ. Res. 50: 451-461.
Ingber, D.E. and Jamieson, J.D. (1985). Cells as tensegrity structures: architectural regulation of
histodifTerentiation by physical forces transduced over basement membrane. In Gene Expres-
sion during Normal and Malignant Differentiation (L.c. Anderson, C.G. Gahmberg,
P. Ekblom, eds.) Academic Press, New York. pp. 13-32.
Ingber, D.E. and Folkman, J. (1989). Mechanochemical switching between growth and difTeren-
tiation during fibroblast growth factor-stimulated angiogenesis in vitro: Role of extracellular
matrix. J. Cell Bioi. In press.
Ingvar, S. (1920). Reaction of cells to the galvanic current in tissue cultures. Proc. Soc. Exper.
Bioi. Medicine 17: 198 fT.
Jarrell, B.E., Williams, S.K., Stokes, G., et aI., (1986). Use of freshly isolated capillary endothelial
cells for the immediate establishment of a monolayer on a vascular graft at surgery. Surgery
100: 392-399.
Jaye, M., Howk, R., Burgess, W., Ricca, GA, Chiu, I.-M., Ravera, M.W., O'Brien, S.J., Modi,
W.S., Maciag, T. and Drohan, W.N. (1986). Human endothelial cell growth factor: cloning,
nucleotide sequence, and chromosome localization. Science 233: 542-545.
Kamiya, A., Bukhari, R. and Togawa, T. (1984). Adaptive regulation of wall shear stress optimiz-
References 543

ing vascular tree function. Bull. Math. Biology. 46: 127-137.

Kamiya, A. and Togawa, T. (1980). Adaptive regulation of wall shear stress to flow change in the
canine carotid artery. Am. J. Physiol. 239 (Heart Circ. Physiol.) 8: HI4-H21.
Karpovich, P.V. (1968). Exercise in medicine: a review. Arch Phys. Rehabil. Medicine 49: 66-76.
Kawamura, K. (1982). Cardiac hypertrophy-scanned architecture, ultrastructure, and cyto-
chemistry of myocardial ceUs. Japanese Circ. J. 46(9): 1012-1030.
Kawamura, K., Imamura, K., Uehara, H., Nakayama, Y., Sawada, K. and Yamamoto, S. (1984).
Architecture of hypertrophied myocardium: Scanning and transmission electron micro-
scopy. In Cardiac Function (H. Abe et al., eds.), Japan Sci. Soc. Press, TokyofVNU Sci. Press,
Utrecht, pp. 81-105.
Kennedy, D.F. and CIiIT, W.J. (1979). A Systematic Study of Wound Contraction in Mammalian
Skin. Pathology 11(2): 207-222.
Knobil, E., and Greep, R.O. (1959). The physiology and growth hormone with particular reference
to its action in the rhesus monkey and the "species specificity" problem. Recent Prog.
Hormone Res. 15: 1-69.
Krompecher, St. (1937). Die Knochenbildung. Verlag Fischer, Jena.
Lanyon, L.E. (1984). Functional strain as a determinant for bone remodeling. Cal. Tiss. Int. 36:
S56.
Lavoie, P.A., Collier, B. and Tennenhouse, A. (1976). Comparison of a a:-bungarotoxin binding
to skeletal muscle after inactivity or denervation. Nature, London, 349-350.
Leach, C.S. and Rambaut, P.C. (1977). Biochemical responses of skylab crew men: an overview.
In Biomedical Results from Skylab. (Johnston, R.S. and Dietlein, L.F., eds.), Washington,
D.C. NASA, pp. 204-216.
Lee, S., Fung, Y.c., Matsuda, M., Xue, H., Schneider, D. and Han, K. (1985). The development
of mechanical strength of surgically anastomosed arteries sutured with dexon. J. Biomech.
18: 81-89.
Leuret, F. and Lassaigne, J.-L. (1825). Recherches Physiologiques et Chimiques pour Servir a
I'Histoire de la Digestion. Madame Huzard, Paris.
Levi-Montalcini, R. and Angeletti, R.U. (1966). Nerve growth factor. Physiol. Rev. 48: 534-569.
Li, C.H., Evans, H.M. (1948). The biochemistry of pituitary growth hormone. Recent Prog.
Hormone Res. 3: 3-44.
Li, C.H., and Chung, D. (1971). Human pituitary growth hormone. Int. J. Prot. Res. 1lI: 73-80.
See also Li et aI., ibid, 81-92, 93-98, 99-108,185-189.
Li, C.H., and Ramasharma, K. (1987). Inhibin. Ann Rev. Pharmacol. Toxicol. 27: 1-21.
Liebow, A.A. (1963). Situations which lead to changes in vascular patterns. In Handbook of
Physiology, Sec. 2: Circulation, Vol. 2, Chap. 37, Washington, D.C., Am. Physiological Soc.
pp. 1251-1276.
Limas, C.J. (1983). Biochemical aspects of cardiac hypertrophy. Fed. Proc. 42: 2716-2721.
Linzbach, A.J. (1960). Heart failure from the point of view of quantitative anatomy. Am. J. Cardiol.
5: 370-382.
Litwak, P., Ward, R.S., Robinson, A.J., Yilgor, I. and Spatz, C.A. (1988). Development of a small
diameter, compliant vascular prostheses. In: Tissue Engineering (R. Skalak and C.F. Fox,
eds.), Alan Liss, New York, pp. 25-30.
Lund, D.D. and Tomanek, R.J. (1978). Myocardial morphology in spontaneously hypertensive
and aortic-constricted rats. Am. J. Anatomy. 152: 141-152.
Lund, E.J. (1921). Experimental control of organic polarity by the electric current I. ElTects of the
electric current of regenerating internodes of obelia commisuralis. J. Exper. Zool. 34: 471.
Lyman, D.J., Fazzio, F.J., Voorhees, H., Robinson, G. and Albo, D. Jr. (1978). Compliance as a
factor elTecting the patency of a copolyurethane vascular graft. J. Biomed. Mat. Res. 12: 337.
Maciag, T., Cerundolo, I.S., Kelley, P.R. and Forand, R. (1979). An endothelial cell growth factor
from bovine hypothelamus: identification and partial characterization. Proc. N atl. Acad. Sci.
U.S.A. 76: 5674-5678.
Mack, P.B., LaChange, P.A., Vose, G.P. and Vogt, F.B. (1967). Bone demineralization of foot and
hand of Gemini-Titan IV, V, and VII astronauts during orbital flight. Am. J. Roentgenol.
544 13 Biomechanical Aspects of Growth and Tissue Engineering

100: 503-511.
Martin, R.B. (1972). The effects of geometric feedback in the development of osteoporosis. J.
Biomech. 5: 447.
Matthews, A.P. (1903). Electrical polarity in the hydroids. Am. J. Physiol. 8: 394.
McGrath, M.H. and Simon, R.H. (1983). Wound geometry and the kinetics of wound contraction.
Plast. Reconstr. Surgery 72: 66-72,1983.
Menton, D.N., Simmons, D.J., Chang, S.L., Orr, B.Y. (1984). From bone lining cell to osteocyte-
An SEM study. Anat. Rec. 209: 29-39.
Meyer, G.H. (1867). Die Architektur der spongiosa. Arehiv. fur Anatomie, Physiologie, und
wissenschaftliehe Medizin (Reichert und wissenschafliche Medizin, Reichert und Du Bois-
Reymonds Archiv) 34: 615-625.
Meyrick, B. and Reid, L. (1979). Hypoxia and incorporation of 3H-thymidine by cells of the rat
pulmonary arteries and alveolar wall. Am. J. Path. 96: 51-69.
Meyrick, B. and Reid, L. (1982). Crotalaria-induced pulmonary hypertension: Uptake of 3H-
thymidine by the cells of the pulmonary circulation and alveolar walls. Am. J. Path. 106:
84-94.
Nakamura, T., Nakajima, T., Takishima, T., Suzuki, N., Arai, S. and Suwa, N. (1975). Analysis
of diastolic pressure-volume relation of the canine left ventricle: Half-inflation pressure as
an index ofleft ventricular compliance. Tohoku J. Exp. M ed. 117: 311-321.
Nakamura, T., Nakajima, T., Suzuki, N., Arai, S. and Suwa, N. (1976). Left ventricular stiffness
and chamber geometry in the pressure-overload hypertrophied heart. Tohoku J. Exp.
Medicine 119: 245-256.
Nakamura, T., Abe, H., Arai, S., Kimura, I., Kushibiki, H., Motomiya, M., Konno, K. and Suzuki,
N. (1982). The stress-strain relationship of the diastolic cardiac muscle and left ventricular
compliance in the pressure-overloaded canine heart. Japanese Cire. J. 46: 76-83.
Nattie, E.E., Wiley, C.W. and Bartlett, D., Jr. (1974). Adaptive growth of the lung following
pneumonectomy in rats. J. Appl. Physiol. 37: 491-495.
Nulend, J.D. (1987). Cellular Responses of Skeletal Tissues to Mechanical Stimuli", (Doctoral
Dissertation) Amsterdam, Free University Press.
Palkovits, M. (1984). Distribution of neuropeptides in the central nervous system: a review of
biochemical mapping studies. Prog. Neurobiol. 23: 151-189.
Pauwels, F. (1980). Biomechanics of the Locomotor Apparatus, German edn. 1965. English trans.
by P. Maqnet and R. Furlong, Springer-Verlag, Berlin, New York.
Pestronk, A., Drachman, D.B., Griffin, J.W. (1976). Effects of muscle disuse on acetylcholine
receptors. Nature, London. 260: 352-353.
Pittelkow, M.R. and Scott, R.E. (1986). New techniques for the in vitro culture of human skin
keratinocytes and perspectives on their use for grafting of patients with extensive burns.
Mayo Clin. Proc. 61: 771-777.
Raben, M.S. (1959). Human growth hormone. Recent Prog. Hormone Res. 15: 71-114.
Rakusan, K. and Turek, Z. (1986). A new look into the microscope: proliferation and regression
of myocardial capillaries. Can. J. Cardiol. 2: 94.
Rivier, J., Spiess, J., Thorner, M., and Vale, W. (1982). Characterization of a growth hormone-
releasing factor from a human pancreatic islet tumour. Nature 300: 276-278.
Rodbard, S. (1975). Vascular caliber. Cardiology, 60: 4-49.
Roesler, H. (1981). Some historical remarks on the theory of cancellous bone structure (Wolff's
law). In Mechanical Properties of Bone, (S.C. Cowin, ed.), Am. Soc. Mech. Engineers, New
York, Pub. No. AMD-Vol. 45. p. 27-42.
Romanova, L.K., Leikina, E.M., Antipova, K.K. and Sokolova, T.N. (1970). The role of function
in the restoration of damaged viscera. Sov. J. Dev. BioI. 1, 384.
Roux, W. (1880-1895). Gesammelte Abhandlungen uber die entwieklungs meehanik der Organismen.
W. Engelmann, Leipzig, 1895. No.3: Dber die Leistungsrlihigkeit der Deszendenzlehre zur
Erkliirung der Zweckmiissigkeiten des fierischen Organismus", Bd. 1, pp. 102-134, (1st pub.
1880. No.4: Der ziichtende Kampf der Teile (Theorie der funktionellen Anpassung), Bd 1,
pp. 137-422, (1st pub. 1881). Nr. 5: "Der zuchtende Kampf der Teile im Organismus"
References 545

(Autoreferat), Bd. 1, pp. 423-437 (1st pub. 1881).

Salzstein, R.A. and Pollack, S.R. (1987). Electromechanical potentials in cortical bone. II. Experi-
mental analysis. J. Biomech. 20: 271.
Schreiber, A.B., Winkler, M.E. and Derynck, R. (1986). Transforming growth factor 2: a more
potent angiogenic mediator than epidermal growth factor. Science 232: 1250-1253.
Sih, G.C.M. and Chen, E.P. (1981). Cracks in Composite Materials. M. NijhotT, Hague/Boston.
Simard, c.P., Spector, S.A. and Edgerton, V.R. (1982). Contractile properties of rat hind limb
muscles immobilized at ditTerent lengths. Exp. Neurol. 77: 467-482.
Sisken, B.F. and Smith, S.D. (1975). The effects of minute direct electrical currents on cultured
chick embryo trigeminal ganglia, J. Embroyl. Exper. Morph. 33: 29.
Skalak, R. and Fox, C.F. (eds). (1988). Tissue Engineering. Alan Liss, New York.
Smith, S.D. (1974). EtTects of electrode placement on stimulation of adult frog limb regeneration.
Ann. N. Y. Acad. Sci. 238: 500.
Snowden, J.M. and ClitT, W.J. (1985). Wound Contraction. Correlations Between the Tension
Generated by Granulation Tissue, Cellular Content and Rate of Contraction. Quart. J. Exp.
Physio. 70(4): 539-548.
Sobin, S.S., Tremer, H.M., Hardy, J.D. and Chiody, H. (1983). Changes in arterioles in acute and
chronic pulmonary hypertension and recovery in the rat. J. Appl. Physiol. 55: 1445-1455.
Spector, S.A., Simrad, c.P., Fournier, M., Sternlicht, E. and Edgerton, V.R. (1982). Architectural
alterations of rat hind-limb skeletal muscles immobilized at ditTerent lengths. Exp. Neurol.
76: 94-110.
Stanley, J.C., Burkel, W.E., Graham, L.M. and Lindbald, B. (1985). Endothelial cell seeding of
synthetic vascular prostheses. Acta Cir. Scand. Suppl. 529: 17.
Stone, J.L., Snyder, B.D., Hayes, W.c. and Strang, G.L. (1984). Three Dimensional Stress
Morphology Analysis of Trabecular Bone, 1rans. Orthop. Res. Soc. 30: 199.
Suwa, N. (1982). Myocardial structure of hypertrophied hearts. Japan. Circ. J. 46(9): 995-1000.
Suwa, N. and Takahashi, T. (1971). Morphological and morphometrical analysis of circulation in
hypertension and ischemic kidney. Urban & Schwarzenberg, Miinchen.
Takamizawa, K. and Hayashi, K. (1987). Strain energy density function and uniform strain
hypothesis for arterial mechanics. J. Biomech. 20: 7-17.
Tezuka, F. (1975). Muscle fiber orientation in normal and hypertrophied hearts. Tohoku J. Exp.
Medicine 117: 289-297.
Thoma, R. (1893). Untersuchungen uber die Histogenese und Histomechanik des Gefiissystems.
Enke, Stuttgart.
Thomas, K.A., Rios-Candelore, M., Gimenez-Gallego, G., DiSalvo, J., Bennett, c., Rodkey, J. and
Fitzpatrick, S. (1985). Pure brain derived acidic growth factor is a potent angiogenic vascular
endothelial cell mitogen with sequence homology to interleukin 1. Proc. N atl. Acad. Sci.
U.S.A. 82: 6409-6413.
Vaishnav, R.N., and Vossoughi, J. (1987). Residual stress and strain in aortic segments. J. Biomech.
20: 235-239.
Weiss, P.A. (1955). Nervous system (neurogenesis). In Analysis of Development, (B.H. WilIier, P.A.
Weiss and V. Hamburger, eds.), Saunders, Philadelphia, pp. 346-401.
Whitehouse, W.J. (1974). The Quantitative Morphology of Anisotropic Trabecular Bone,
J. Microscopy, 101: 153-168.
Whitehouse, W.J. and Dyson, E.D. (1974). "Scanning Electron Microscope Studies of Trabecular
Bone in the Proximal End of the Human Femur," J. Ana.U8: 417-444.
WoltT, J. (1869). Dber die bedeutung der Architektur der spongiosen Substanz. Zentralblatt for
die medizinische Wissenschaft. VI. Jahrgang: 223-234.
WoltT, J. (1870). Dber die innere Architektur der Knochen und ihre Bedeutung fiir die Frage vom
Knochenwachstum. Archiv for pathologische Anatomie und Physiolgie und for klinische
Medizin (Virchovs Archlv) 50: 389-453.
WoltT, J. (1892). Das Gesetz der Transformation der Knochen. Hirschwald, Berlin.
Yannas, I.V. and Burke, J.F. (1980a). Design of an artificial skin. I. Basic design principles. J.
Biomed. Mater. Res. 14: 65-81.
546 13 Biomechanical Aspects of Growth and Tissue Engineering

Yannas, LV., Burke, I. F., Gordon, P.L., Huang, C. and Rubenstein, R.H. (1980b). Design of an
artificial skin. II. Control of chemical composition. J. Biomed. Mater. Res. 14: 107-131.
Yannas, LV., Burke, IF., Orgill, D.P. and Skrabut, E.M. (1982). Wound tissue can utilize a
polymeric template to synthesize a functional extension of skin. Science 215: 174-176.
Yasuda, I. (1974). Mechanical and electrical callus. Ann. N.Y. Acad. Sci. 238, 457-465.
Yeh, C.K. and Rodan, G.A. (1984). Tensile forces enhance prostaglandin E synthesis in osteo-
blastic cells grown on collagen ribbons. Cal. Tiss. Int. 36: S67.
Vue, X., van der Lei, B., Schakenraad, T.M., van Oene, G.H., Kuit, I .H., Feijen, T. and Wildevuur,
C.R.H. (1988). Smooth muscle seeding in biodegradable grafts in rats: A new method to
enhance the process of arterial wall regeneration. Surgery 103: 206.
Zamora, I.L., Navarro, L.T., Ives, C.L., Weilbaecher, D.G., Gao, Z.R. and Noon, G.P. (1986).
Seeding of arteriovenous prostheses with homologous endothelium. J. Vasco Surg. 3: 860.

We discussed animals. But biomechanics

applies equally well to plants. BAMBOO by
Ku An, ~ ~. Ink on silk, 123 x 53 cm
wall scroll, in National Palace Museum,
Taipei. Ku was a native of East Whei,
Kiangsu. He lived in 14th century, Yuan
Dynasty.
Author Index

Abbott,I.H. 101, 104 Ashley, H. 73,80, 100, 104

Abe, H. 544 Ayman, G. 150
Adams, S. 496
Adkins, J.E. 381
Aebischer, P. 541 Bachofen, H. 405,420,424,447,448,
Aki, K. 103 451
Alexander, H. 380 Baez, S. 333, 348
Allen, S.T. 272 Bahniuk, E. 474
Allessandra, G. 307 Baldwin, A.L. 438, 447
Alpert, N.R. 515,539 Balser, J.R. 540
Alt, W. 149, 151 Barber, B.J. 349
Altura, B.M. 348 Bartlett, D., Jf. 544
Ampaya, E.P. 401,407,449 Barton, M.V. 494
An, K.N. 322,323,348,351 Bassett, C.A.L. 507,539
Anderson, G.B.J. 60,448,491,497 Bassingthwaighte, J. 319,348
Anderson, S.A. 537 Batchelor, G.K. 104
Andreoli, T.E. 306 Bates, D.V. 482,493
Angeletti, R. U. 527, 543 Bayliss, W.M. 524, 539
Anliker, M. 447,497 Bean, J.e. 52, 59
Anthonisen, N.R. 482,493 Becker, R.O. 508,539
Antipova, K.K. 544 Bell, E. 535, 539
Antolini, R. 307 Bellman, R.E. 55, 59
Anversa, P. 512, 539 Belytschko 491,493,495,496
Apelblat, A. 311, 348 Bergel, D.H. 388, 447
Ardila, R. 447 Berne, R. 319,349
Aristotle 342 Bernick, S. 450
Armstrong, e.G. 304, 306, 307 Bevan, R.D. 517,539
Arrhenius, S. 326, 348 Bieniek, M.P. 449
Artaud, C. 451 Biggs, W.B. 149,497
Artinson, G. 348 Bingham, N. 451
Arturson, G. 333 Biot, M.A. 77, 104,381
Arvebo, E. 495 Bisplinghoff, R. 73, 80, 100, 104
Ascheim, E. 326, 348 Bjork, R. 60
547
548 Author Index

Block, I. 318,348 Caughey, T.K. 56,59

Block, W.D. 541 Cebeci, T. 104
Blum, 1.1. 318,348 Chaffin, D.B. 59,491,493
Bohlen, P. 542 Chambers, T.l. 509,540
Boltzmann, L. 53 Chan, M. 497
Booth, F.W. 520,539 Chandler, R.F. 489, 490, 493
Borelli, G.A. 9 Chang, H.K. 244,266,267,273
Boren, R.C., lr. 448 Chang, R.S.Y. 493
Borrero, L.M. 350 Chang, S.L. 544
Bouskela, E. 438, 447 Chao, E.Y.S. 59,60
Boussinesq, 1. 429 Chasan, B. 307
Bowald, S. 534,539 Chatwin, P.C. 266, 273
Bowen, I.G. 475, 479, 493, 496 Chen, E.P. 545
Boyce, S.T. 535, 539 Chen, P.D. 449
Bradley, K.H. 515,540 Chen, P.H. 491,497
Bradshaw, P. 104 Chen, Z.D. 378, 381
Bradstreet, E.D. 351 Cheng, R.T. 273
Brain,l.D. 450 Cherry,l.M. 381
Brankov, G. 57 Chien, S. 201,225,447
Braunwald, E. 514,540 Childress, S. 151
Brazeau, P. 542 Chillingworth, F.P. 273
Brenner, B.M. 349 Chiu, T.H. 541
Brenner, P.H. 448 Chow, c.Y. 101,102,104,111,146,
Brigham, K.L. 478, 493 152
Brighton, C.T. 542 Christie, R.V. 272,493
Brody,l.S. 515,540 Chung, D. 543
Brown, R.H.l. 22,23 Chuong, C.l. 388,393,447,479,493,
Brown, R.M. 17,28 529,540
Brown, T.D. 491,493 Chwang, A.T. 134, 140, 151
Bruley, D.F. 350 Clarke, R.O. 225
Buchwald, H. 539 Clausius, R.l.E. 279
Buckwalter,l.A. 460,498 Clemedson, C.-I. 473, 475, 476, 493,
Bueker, E. 526 495
Buhain, W.l. 515, 540 Clements,l.A. 402,403,420-423,447,
Bukhari, R. 542, 543 450
Burke, E. 102 Cliff, W.l. 330,349,536,540,543,
Burke,l.F. 330,350,535,540,545,546 545
Burkel, W.E. 545 Clowes, A.W. 533,540
Burns, B.H. 28 Coan, M.R. 521,540
Burstein, A. 494 Coefficients, D. 316
Busch, C. 539 Cohen, S. 526, 540
Butler,l.P. 450 Cohn, R. 515, 540
Butterman, G. 498 Cohn, S.H. 537
Collier, B. 543
Comroe, J.H. 273
Caiafa, C.A. 493 Condeelis, J.S. 149, 152
Camana, P.C. 43, 59 Cooper, J.A. 149, 152
Campbell, 1.1. 473,493 Copley, A.L. 348
Campen, V. 307 Corsin, S. 120, 150
Carr, L.W. 110, 152 Counsilman, J.E. 17,28
Carter, D.R. 505,506,511,540,541 Courant, R. 493
Cassen, B. 478, 494 Covell, J.W. 381,451
Cassidy, 1.1. 510 Cowan, M.J. 515,516,540
Caterson, B. 540 Cowin, S.C. 509, 510, 540
Author Index 549

Craig, R.L. 497 Eckstein, E.C 223

Crapo, J.D. 450 Edgerton, Y.R. 545
Crone, C. 316, 348, 349, 478, 493 Effros, R.M. 478, 493
Cronin, R.F.P. 349 Egan, E.A. 478, 494
Crystal, R.G. 515, 516, 540 Ehlich, H.P. 540
Culmann, C. 500,540 Eiband, A.M. 483, 484, 494
Culver, c.c. 495 Einstein, A. 292, 293, 306
Cumming, G. 229,232,273,404,405, Eisenberg, S.R. 304, 306
448 Elfstrom, G. 10, 28, 489, 496
Curran, P.F. 307 EI Haj, A.J. 540
Currey, J.D. 149,452,493,497 Ellington, C.P. 151
Curry, F.E. 294,306,310,349, Engin, A.E. 491,494
350 Engler, R.L. 149, 152
Curtis, C.F. 273 Enhorning, G. 421, 448
Curtiss, H.C. 104 Epstein, P. 306
Cyriax, J. 11, 12,28 Erdogen, F. 504, 541
Erikson, I. 539
Eringen, A.C. 381
Dai, K. 193, 194 Esch, F. 542
Dainty, J. 306 Etlinger, J.D. 541
Dale, P.J. 448 Euler, L. 2, 19, 165, 195
Dalton, S. 151 Evans, E.A. 201,224,536,541
Damon, E.G. 496, 497 Evans, F.G. 454,496,498
Darcy, H. 277,291,205,306,320 Evans, H.M. 543
Darwin, C. 503 Evans, J.W. 225
Davies, P.E. 533, 540
Davy, D.T. 510,540
Davy, H. 277 Fanestil, D.D. 306
Dawson, C.A. 223, 225 Farhi, L.E. 273
Dawson, S.Y. 449 Fass, F. 496
Dean, G.W. 223 Faulkner, J.A. 521,541
De Boer, L.J. 307 Feinberg, B.N. 459, 494
Deen, W.M. 212, 349 Fenn, W.O. 226, 273, 448
De Martino, G. 541 Fenton, T.R. 381
Dembo, M. 149, 151 Ferguson, A.B., Jr. 493
Deves, R. 307 Filion, J. 308
Dewey, C.F., Jr. 540 Firoozbakhsh, K. 540
Dix, J.A. 307 Fischbach, G.D. 521,541
Dixon, H.H. 306 Fishman, A.P. 225,273,351,451,478,
Dobrin, P.B. 448 494
Dodge, H.T. 394 Fitzgerald, J .M. 224
Doenhoff, Y. 100, 104 Flaherty, J.E. 186, 195
Doillon, C.J. 536,540 Fleck, J.T. 491,494
Dou, R. 193, 194 Fleischner, F.G. 448
Dowell, E.H. 73, 80, 100, 104 Fleming, D.G. 459, 494
Doyle, J.M. 448 Fletcher, E.R. 493, 496, 497
Drachman, D.B. 544 Flicker, E. 421,423,448
Drake, R.E. 272 Folkman, J. 527,541,542
Drazen, J.M. 274 Folkow, B. 157, 195
Dubois, A.B. 258,273 Foppiano, L. 189, 195, 225
Duling, B.R. 319,349 Forster, R.E. 273, 274
Duncker, H.R. 115, 151 Fosby, R. 16
Dunn, F. 472, 473, 493 Fox, C.F. 540, 545
Dyson, E.D. 510,545 Francone, C.A. 495
550 Author Index

Frank, O. 163, 195 Goerke, J. 421,450

Frankus, A. 414, 416, 448, 449 Gogolewski, S. 534, 541
Fraser, R.G. 446, 448 Gold, D. 186, 195
Fredberg, J.J. 103, 274 Goldberg, A.L. 520, 541
Friedman, M. 295,2%,297,298,300, Goldin, B. 520,541
306 Goldman, R. 149, 151
Friedrichs, K.O. 493 Goldsmith, W. 488, 494
Frisoli, A. 478, 494 Goldspink, D.F. 541
Froese, A.B. 272 Goldspink, G. 108, 151
Frolich, E.D. 515, 541 Gordon, E.F. 540
Fronek, K. 388, 448 Gordon, W.1. 225
Frucht, A.H. 393, 473, 494 Gore. R.W. 439,447
Fry, D.L. 257,273,518,541 Goresky, C.A. 334, 349
Fry, W.J. 472,473,493 Gorio, A. 307
Fukada, E. 273,507, 541 Gosline, J.M. 149,497
Fung, Y.C. 28,53, 59, 60, 73, 80, 82, Gospodarowicz, D. 541
100, 104, 108, 112, 132, 151, 176, Gottrup, F. 519, 541
183, 189, 193, 195, 204, 206, 209, Gown, A.M. 540
210,211,215,216,217,219,221, Graham, L.M. 545
224, 225, 269, 273, 274, 306, 307, Granger, H.J. 349
312, 337, 338, 340-342, 344-346, Graves, G.A., Jr. 490,491,495
349, 351-353, 358, 362, 376, 377, Gray, H. 28
381,384, 387, 388, 391, 393-399, Gray, J. 126, 134, 136, 151
404-416,419, 422, 42~29, 436, Green, A.E. 381
438-445,447-451,455,459, 478- Greenwood, D.T. 28, 59, 60
482,490-492, 494, 497, 498, 514, Greep, R.O. 521, 543
529,541,543 Greisler, H.P. 534, 542
Fyhrie, D.P. 511,541 Griffin, J.W. 544
Giffith, A.A. 494,505,541,542
Grimm, A.F. 394
Gabel, J.C. 272 Grodzinsky, A.J. 304,306
Gabrielli, G. 147, 148, 153 Grood, E.S. 59, 60
Gadd, C.W. 487,494 Grootenboer, H.J. 307
Gaensler, E.A. 515, 541 Gross, D. 509, 542
Galante, J.O. 495 Gross, J.F. 312, 313, 314, 349, 350
Galletti, P.M. 534,541 Gross, S.B. 509, 542
Gallico, G.G. 535, 541 Guillemin, R. 525,526,542
Gallin, J.1. 149, 151 Gurdjian, E.S. 486,487,495,497
Galvani, L. 508 Guyton, A.C. 324, 349, 523
Ganfield, R.A. 347
Gao, Z.R. 546
Garrick, I.E. 77, 80, 105 Haddad, B.F. 495
Gauer, O.H. 497 Haderspeck, K. 60
Gehr, P. 447,448 Hajji, M.A. 416, 449
Geng, C.C. 491,497 Hakim, T.S. 223
Gennarelli. 486, 494 Haldane, J.S. 271
Ghista, D.N. 28,60,490,495,497 Hales, S. 163, 195
Gibbons, I.R. 125, 135, 151 Ham, R.G. 535, 542
Gil, J. 424, 448 Hamburger, V. 526,545
Gimbrone, M.A., Jr. 540 Hammel, H.T. 305, 325-328, 349, 351
Giuntini, C. 478, 494 Han, K. 543
Glauert, H. 104 Hancock, G.J. 126, 134, 136, 151
Gluck, L. 420,421,448 Handa, H. 449
Goddard, M.B. 541 Hansbrough, J.F. 535,539
Author Index 551

Hanson, J.E. 401,407,449 Hull, B. 539

Hanson, S.R. 540 Hunter, P.J. 400, 449
Harf, A. 273 Huston, R.L. 59,60,491,495,4%
Hargens, A.R. 307, 321, 349, 351 Huxley, F.R.S. 149, 151
Harlow, F. 149, 151 Hyatt, R.E. 257, 273, 449
Harrigan, T. 510, 542
Harris, A.K. 521-523, 542
Harris, P.D. 272, 347 Imamura, K. 543
Hart, R.T. 509, 540, 542 Ingber, D.E. 538, 542
Haselton, F.R. 274 Ingram, R.H., Jr. 274
Hauge, A. 478, 496 Ingvar, S. 508, 542
Hay, J. 13, 14, 15, 16, 28 Intaglietta, M. 307,315,319,349,448
Hayashi, K. 449, 450, 529, 545 Isabey, D. 273
Hayes, W.e. 491,495,511,542,545 Ishida, T. 274,450
Hazelrig, J.B. 348 Ivarsson, B. 539
Hecht, H.H. 273, 351,478, 494 Ives, C.L. 546
Hefton, J.M. 535, 542
Helfenstein, U. 497
Hemami, H. 59 Jacobs, S.W. 490,495
Hembry, R.M. 540 Jacobs, R.R. 490,495
Hemmingsen, A.E. 351 Jaeger, M.J. 234, 273
Hempstead, B. 542 Jaffrin, M.Y. 351
Henderson, Y. 265,273 Jameson, W. 120, 152
Herbison, G.J. 520,542 Jamieson, J.D. 542
Herrman, G. 264, 274 Janz, R.F. 394
Hertz, H.R. 429 Jarrell, B.E. 533, 542
Higdon, J.J.L. 150 Jayaraman, G. 194
Hight, T.K. 496 Jaye, M. 540
Hildebrandt, J. 404,405,447,449,482, Jensen, M. 113, 114, 115, 153
495 Johnson, J.A. 316, 349
Hill, I.R. 485, 495 Johnson, P.C. 306, 307, 319, 349
Hirsch, A.E. 486, 488, 495 Jones, R.K. 497
Hirschfelder, J.~. 273 Jones, R.T. 77
Ho, H.H. 215, 224, 225, 448, 498 Jonsson, A. 473, 475, 476, 493, 495
Hochmuth, R.M. 224 Joseph, D.D. 305
Hodge, W.A. 492 Joule, J.P. 277
Hodgson, V.R. 485,488,495
Hoffman, J.F. 306
Holbourne. 486, 495 Kaley, G. 348
Holladay, A. 493 Kallok, M.J. 449
Holmes, M.H. 307 Kamiya, A. 518,543
Hood, W.B. 394 Kamm, R.D. 274
Hopkinson, B. 469,471,495 Kane, T.P. 59,60,491,495
Hopkinson, J. 468, 495 Karakaplan, A.D. 404, 405, 449
Hoppin, F. 404,405,416,449,450, Karpovich, P.V. 521,543
482, 495 Katair-Katchalsky, A. 348
Horie, T. 447 Katchalsky, A. 307, 349
Horsfield, K. 225, 273 Katz, A.1. 195
Hort, W. 529,542 Katz, J. 272
Huang, I. 223 Kawamura, K. 514,543
Huang, M.K. 104 Kaye, G.W.C. 473,495
Hudlicka, O. 542 Kazarian, L. 490,491,495
Hughes, N.F. 106, 152 Keller, J.B. 186, 195
Hulett, G .A. 326, 349 Keller, K.H. 223
552 Author Index

Keligren, J.H. 10, 11,28 Lavoie, P.A. 521,543

Kelvin, L. T. 325, 326, 429 Laws, K. 58, 59
Kenedi, R.M. 496 Leach, C.S. 521,543
Kennedy, D.F. 536,543 Leak, L.V. 332,350
Kenyon, D.E. 305,307 Lebow, M. 496
Kim, D. V. 542 Lee, C.S.F. 161, 195
Kimura, I. 544 Lee, G.C. 414,416,448,449
Kinetics, M.-M. 318 Lee, J.S. 59, 225, 265, 267, 273, 315,
King, A.I. 488, 491, 495 350,421,423
Kingaby, G.P. 225 Lehr, J. 274
Klassen, G.A. 381 Leikina, E.M. 544
Klingenberg, M. 296, 307 Leimbach, G. 149, 152
Knisely, M.H. 350 Leith, D. 449
Knobil, E. 525, 543 Levi-Montalcini, R. 526, 543
Knopp, T.J. 348 Levy, M.N. 537
Kohn, D.H. 540 Lew, H.S. 307
Konno, K. 544 Lewis, G.N. 282
Koogle, T.A. 491,495 Lewis, T. 10,28
Kooyman, G.L. 482,495 Li, C.-H. 525, 543
Kovasznay, L.S.G. 68,69, 104 Li, J.K.L. 224
Kraman, S.S. 495 Lieb, W.R. 296, 307
Kroell, C.K. 485, 496 Liebeck, R.H. 103, 104
Krogh, A. 316, 318, 350 Liebow, A.A. 518,543
Krogh, M. 248, 273 Lightfoot, E.N. 306, 307
Krompecher, S. 543 Lighthill, J. 113, 120, 122, 126, 128,
Krupka, R.M. 307 134, 136, 138, 140, 152
Kuchemann, D. Ill, 153 Lillienthal, O. 152
Kuethe, A.M. 101, 102, 104, Ill, 112, Limas, c.J. 515, 543
115, 122, 123, 124, 146, 152 Lindbald, B. 545
Kulak, R.F. 491,493,495 Lindbeck, L. 58
Kulovich, M.V. 448 Lindstrom, J. 224
Kumar, A. 194 Linehan, J.H. 223
Kushibiki, H. 544 Ling, N. 542
Kiissner, H.G. 77,80, 104 Lipowsky, H.H. 199,225
Kwan, M.K. 307 Lisbona, R. 223
Kyle, C. 102 Lissmann, H.W. 126, 151
Lissner, H.R. 28,486,487,495,496
Litwak, P. 533, 543
Laananen, D.H. 486, 495 Liu, J.T. 449
Labby, T.H. 473,495 Liu, S.Q. 384,385,387,448,449,
Lagrange, J.L. 29, 38 494
Lai, M. 304, 306, 307 Liu, Y.K. 490,496
Lai-Fook, S.J. 429, 430, 439, 449 Loeschcke, H. 402,403
Laine, G.A. 272, 347 Loring, S.H. 274
Lambert, R.K. 449 Lossow, W.J. 495
Landis, E.M. 312-315, 350 Love, A.E.H. 449
Lange, W.A. 497 Ludwig, G.D. 473,496
Lanir, Y. 304,307,449,496 Lukocovic, M.F. 307
Lanyon, L.E. 509, 540, 543 Lund, D.D. 513, 543
Lassaigne, J.-L. 524, 543 Lund, E.J. 508,543
Lassen, N.A. 316,348,349,478,493 Lyman, D.J. 543
Latimer, F.R. 495
Latorre, R. 295, 307
Lau, V.K. 477, 496 Machida, A. 539
Lauret, F. 524, 543 Maciag, T. 543
Author Index 553

Mack, P.B. 530, 543 Mizrahi, J. 59, 60

Macklem, P.T. 226,272,273,451, Moffat, D.S. 349
493 Morecki, A. 60
Mahley, R.W. 541 Moreno, A.H. 186, 195
Mahone, R.H. 495 Mori, K. 449
Maki, H. 274 Moritake, K. 449
Malpighi, M. 207,225,402 Morkin, E. 225
Malvern, L.E. 381 Morrey, B.F. 59,60
Mann, R.W. 510,542 Motomiya, M. 544
Marchak, B.E. 272 Mow, V.C. 304, 306, 307
Marchetti, G. 539 Muller, I. 304, 307
Maroudas, A. 308 Murakami, H. 348
Marshall, B. 223 Muscarella, L.F. 224
Marshall, C. 223 Muskian, R. 491,496
Martin, R.B. 509,544 Myers, E.R. 304, 307
Maryniak, J. 57 Myers, R.R. 348
Mason, G. 493
Matsuda, M. 402, 408, 449, 543
Matthews, A.P. 508 Nachemson, A. 489, 496
Matthews, F.L. 448, 451 Nachtigall, W. 101, 104, 121, 152
Matthys, H. 234, 273 Nagel, D.A. 496
Matzdorf, P. 15 Nahum, A. 485, 488, 495, 496, 497
Maxworthy, T. 152 Nair, P. 347
Mayrick, B. 516,517,544 Nakajima, T. 544
McAlister, K.W. 110,152 Nakamura, T. 512, 513, 539, 544
McConnell, S.D. 540 Nakayama, Y. 543
Mc Croskey, W.J. 110, 152 Nash, C.D. 491, 496
McCulloch, A. 400, 449 Nashner, L.M. 10, 28
Mc Donald, D.A. 176, 178, 195 Nattie, E.E. 515, 544
McElhaney, J.H. 497 Navarro, L.T. 546
McGeachie, J.K. 541 Neathery, R.F. 485,496
Mc Ghee, R.B. 43,44,57,60 Neil, E. 157, 195
McGrath, M.H. 536,544 Nelson, R.M. 494
McIlroy, M.B. 273 Nerem, R.M. 181, 195,267,274
McKerrow, C.B. 273 Neri, L.M. 493
Mead, J. 226,257,273,449,450, Newman, J.N. 151, 153
451 Nicholl, P.A. 330, 349
Meier, P. 336, 350 Nicholson, W.F. 539
Melvin, J. 224, 485, 496 Nicolaysen, G. 478, 496
Menton, D.N. 509, 544 Niimi, H. 319,350
Mercer, R.R. 402,450 Noon, G.P. 546
Merrill, C. 539 Noordergraaf, A. 224
Mertz, H.J., Jr. 485,496 Norberg, R.A. 121, 152
Messmer, K. 348 Nordin, M. 60
Meyer, G.H. 500,503,544 Nossal, R. 149, 152
Michel, C.C. 294, 306, 349, 350 Noyes, A. 326, 350
Middleman, S. 307,350 Nulend, J.D. 509, 544
Miller, C. 295, 307
Miller, S.L. 351
Miller, W.S. 203,204,225,402,450 Oberbeck, A. 129, 152
Milner, W.R. 222, 224, 225 Ogden, R.W. 380
Mindlin, R. 429 Ogston, A.G. 350
Mirsky, I. 394, 450 Oh, S.K. 541
Mitzner, W. 223 Oldmixon, E.H. 402,450
Miyasaka, K. 272 Oldroyd, H. 121, 152
554 Author Index

Oliver, R.E. 494 Prather, J.W. 330, 331, 352

Omens, C.W.J. 305,307 Preston, B.N. 350
Omens, J. 394-399,450 Price, J.B. 542
Ommaya, A.K. 488, 495, 496 Pride, N. 274
Onsager, L. 307 Prigogine, I. 307
Orne, D. 491,496 Pringle, J.W.S. 108, 152
Orr, B.Y. 544 Privitzer, E. 491,496
Orsos, F. 405, 450
Ortengren, R. 60
Oster, G.F. 149, 152 Quie, P.G. 149, 151
Otis, A.B. 271,273

Raasch, E. 4%
Padmanabhan, N. 194 Raben, M.S. 525, 544
Pai, S.L. 104 Radford, E.P., Jr. 273
Palkovits, M. 526, 544 Rahn, H. 226,273,448
Panjabi, M.M. 26, 28, 490, 496 Ramasharma, K. 543
Papenfuss, H.D. 312, 313, 314, 350 Rambaut, P.C. 521,543
Pappenheimer, J.R. 293,307,312,313, Rau, G. 28, 60
315, 316, 350 Rayleigh, J.W.S. 56,60
Pare, J.A.P. 446,448 Rayleigh, L. 120, 152
Park, J.B. 496 Reddy, R.V. 186, 195
Parrott, G.C. 153 Reid, L. 516,517, 544
Passerello, C.E. 491, 496 Reidy, M.A. 540
Patel, D.J. 176, 195, 388, 450 Reifenrath, R. 421, 450
Patitucci, P. 388,448 Remuzzi, S. 540
Patlak, C.S. 296,307 Reneau, D.D. 318,350
Patrick, L.M. 496, 497 Renkin, E.M. 306-307,310,316,349,
Paul, J.P. 17, 18,28 350
Pauwels, F. 60,503-507 Reul, H. 28, 60
Pearson, J.R. 541 Reynolds, O. 104
Pedley, T.J. 176, 195,234, 235, 268, Rice, D.A. 472, 473, 496
273,274 Richmond, D.R. 493, 496, 497
Pedotti, A. 57, 60 Rickaby, D.A. 223
Pennycuick, C.J. 107, 1I4, 1I5, 119, Riley, R.L. 274
152 Rivier, J. 522, 544
Perelson, A.S. 149, 152 Rivlin, R.S. 459, 497
Permutt, S. 215,225,257,274 Robbins, N. 521,541
Perrone, N. 59,60,491, 495, 496 Roberts, S.B. 491, 497
Pestronk, A. 521, 544 Roberts, V.L. 496, 497
Peterson, R.T. 152 Robertson, C.F. 349
Piiper, J. 116, 152 Robinson, A.J. 543
Pilkey, W.D. 495 Rodan, G.A. 509,546
Pitcher, A.S. 473, 493 Rodarte, J.R. 449, 450
Pittlekow, M.R. 535,544 Rodbard, S. 518, 544
Piziali, R.L. 491,495,496 Roesler, H. 503,544
Pollack, S.R. 509, 545 Romanova, L.K. 515, 544
Pollard, D.W. 493 Rosen, S. 539
Pollard, T.D. 149, 151, 152 Rosenbaum, J. 149, 151
Popel, A. 349 Rosenquist, T.H. 188, 189, 195,450
Powell, H.C. 348 Roshko, A. 68
Poynting, J.H. 326, 350 Rossing, T.H. 274
Prandtl, L. 152 Roth, B. 59, 60
Prasad, S.N. 264,274 Roth, E.M. 483, 497
Author Index 555

Roth, V. 307 Simmons,D.J. 544

Roughton, F.J.W. 274 Simon, B.R. 400, 450
Roux, W. 503,505,544 Simon, R.H. 536, 544
R.....inow, S.1. 186, 195 Simrad, C.P. 521,545
Ruegsegger, P. 497 Singh, M.D. 194
Singhal, S.S. 273
Singley, G. T ., III 495
Saczalski, K. 495 Sisken, B.F. 508, 545
Saha, S. 453,490,497, 597 Sisto, F. 104
Salathe, E.P. 311,322,323,348,351 Skalak, R. 149,153,201,202,225,447,
Salotto, A.G. 224 449, 540, 545
Salzstein, R.A. 509, 545 Skalak, T. 332, 351
Sanada, K. 543 Skerry, T.M. 540
Sandaram, S.H. 491,497 Slama, H. 116, 152
Sandler, H. 394 Sleigh, M.A. 125, 139, 151, 152
Sasken, H.F. 541 Slutsky, A.A. 274
Sato, M. 149, 152 Smail, B.H. 400, 449
Scanlan, R.H. 104 Smaje, L.H. 350
Scher, M.P. 491, 495 Smart, J. 106, 152
Scherer, P.W. 267,274 Smith, A.M.O. 102, 104
Schilder, D.P. 257,273 Smith, S.D. 508, 545
Schmid-Schonbein, G. W. 149, 152, Smith, X. 149, 152
153, 225, 267, 274, 332, 348, 351 Snowden, J.M. 536,545
Schmidt-Nielsen, 147 Snyder, B. 511,542, 545
Schneider, D.C. 459,496,497,543 Sobin, S.S. 188, 189, 195, 206, 207,
Scholander, P. 324-330, 349, 351 209,210,211,215,225,349,408-
Schreiber, A.B. 527, 545 412,448,450,517,545
Schroter, R.C. 273,274,448 Sokolova, T.N. 544
Schultz, A.B. 10, 11,28,52,59,60, Solomon, A.K. 295,307
491,493,495,497 Sonnerup, L. 490, 497
Schurch, S. 421,450 Soto-Rivera, A. 315, 350
Schurman,D.J. 495 Spann, J.F. 394
Schwartz, W.H. 149, 152 Spatz, C.A. 543
Schwarz, L. 80, 104 Spector, S.A. 521, 545
Scott, R.E. 535, 545 Spiess, J. 544
Sears, J. 491,495 Spilker, R.L. 400,450
Sears, W.R. 80, 104 Spindler, K.P. 542
Seed, W.A. 181, 195 Stainsby, G. 348
Seguchi, Y. 258, 259, 262, 274, 450 Stalnaker, R.L. 485, 488, 496, 497
Selverstone, N.J. 273 Stamenovic, D. 439,450
Semple, S.J. 273,404,405,448 Stanley, J.e. 533, 545
Sepllacy, W.N. 448 Stapp, J.P. 482,483,485, 497
Shapiro, A.H. 186, 190, 191, 192, 195, Starling, E.H. 308, 351, 524, 539
274, 344, 351 Staub, N.e. 351,478,493,497
Sharma, M.G. 459,497 Staverman, A.J. 351
Shaw,D.J. 420,450 Stein, W.D. 296,307
Sheid, P. 116, 152 Stewart, G.N. 244,274,316,351
Shepherd, A.P. 349 Stewart, W.E. 306
Sher, J. 539 Stockwell, C.W. 59
Shiari, R. 492 Stokes, G. 542
Shields, W.H. 423,451 Stone, J.L. 510, 545
Shykoff, B.E. 273 Stopak, D. 542
Sih, G.C.M. 504,545 Stossel, T.P. 149, 152
Silberberg, A. 287, 288, 308, 348, 352 Strang, G.L. 545
556 Author Index

Stratford, B.X. 102, 105 Trout, E.M. 490, 493

Streeter, D.D. 394 Truesdell, e. 305, 308
Streeter, V.L. 259,262,274 Tucker, V.A. 113,146,147,153
Strieder, J.W. 515,541
Sudlow, M.F. 273,274
Sugihara, M. 319,350 Uehara, H. 543
Sundqvist, A.-B. 495 Urban, J.P.G. 304, 308
Sung, P. 447 Usami, S. 447
Suntay, W.J. 59,60 Uszler, J.M. 493
Susak, Z. 59, 60
Suwa, N. 513,539,544,545
Suzuki, N. 544 Vaishnav, R.N. 176,195,388,450,
Swart, P. 102, 105 529,545
Swenson, L.W., Jr. 495 Valberg, P.A. 450
Vale, W. 525, 544
Van Buskirk, W.e. 509, 540
Tai, R.e. 273 Van Der Pol, B. 252, 274
Takahashi 545 Van Dyke, M. 105, 152
Takamizawa, K. 450, 529, 545 Van't Hoff, J.H. 351
Takishima, T. 449,450,544 Varco, R.L. 539
Talbot, J.M. 520,537,542 Vathon, A. 23
Talbot, L. 161, 195 Vawter, D. 414-416,423,431,450,451
Tang, H.T. 337,338,341,342,351 Venkataraman, R. 351
Tao, Z.L. 448,450,479,481,494,497, Verkman, A.S. 307
498 Versace, J. 487,497
Taylor, D.L. 149, 152 Viano, D.C. 477,491,496,497
Taylor, G.I. 152, 153,238,241,244, Visioli, O. 539
274,316,351,469-471,497 Vitali-Mazza, L. 539
Teichner, W.G. 497 Von Gierke, H.E. 473, 497
Temkin, S. 492 Von Holst, E. Ill, 153
Tennenhouse, A. 543 Von Karman, T. 147, 148, 153
Tezuka, F. 513, 545 Von Neergaard, K. 402,403, 451
Theodorsen, T. 77,80, 105 Voorhees, A.B. 542
Thorn, A. 101, 105 Vossoughi, J. 450,529,545
Thomas, A.G. 459, 485, 488, 497
Thomas, K.A. 527,545
Thomas, L.M. 495,497 Wagner, H. 77,80, 105
Thompson, W.K. 272 Wagner, P.P. 247,250,252,270,274
Thomson, W.T. 351,491,497 Wainwright, S.A. 149, 452, 497
Thorner, M. 544 Waldman, L. 359,360,381,400,
Tierney, D.F. 420,422,447 451
Tietjens, O.G. 152 Wallace, A. 503
Tischler, L.M. 541 Walsh, M.J. 102, 105
Tisi, G.M. 271,274 Wang, J.J. 273
Togawa, T. 518,542 Wang, Y.L. 149, 153
Tomanek, R.J. 513,521,540,543 Wangel, B.E. 349
Tong, P. 448 Ward, e. 488, 496, 497
Toon, M.R. 307 Ward, R.S. 543
Toperoff, B. 429, 449 Ware, J.A. 103
Torrance, R.W. 215,224 Warrell, D.A. 217,218,225
Tozeren, A. 201,225,447 Watson, P.A. 539
Tremer, H. 188, 189, 195, 215, 224, Way, M.E. 493
225, 408-412, 448, 450 Webb, P.W. 145, 153
Trollope, M.L. 496 Webster, J.E. 495
Author Index 557

Weeds, A.G. 149, 151 Wu, T.Y. 134, 135, 140, 143-145, 152,
Wehausen, J.V. 152 153
Wehdrenberg, W.B. 542 Wylie, E.B. 259, 262, 274
Weibel, E.R. 229, 230, 231, 274, 402,
403,405,447,448, 451
Weilbaecher, D.G. 546 Xue, H. 193-195, 543
Weinberg, S.L. 351
Weir, E.K. 225
Weis-Fogh, T. 108, 109, 112-115, 121, Yager, D. 451
122, 153 Yamada, H. 453, 459, 498
Weisgraber, K.H. 541 Yamamoto, S. 543
Weiss, P.A. 524, 545 Yan, C.Z. 104
Weiss, S.W. 541 Yannas, LV. 535,540,545,546
Wells, J.D. 350 Yarnell, P. 496
West,J.B.225,247,250,252,270,274, Yasuda, I. 507, 546
414,450,451,495 Yates, G.T. 144, 145, 153
Weyl, H. 451 Yeh, C.K. 509, 546
Whalen, R.T. 540 Yen, M.R.T. 166, 189, 195, 204, 205,
Whalen, W.J. 347 211, 215-217, 219, 221, 224, 225,
White, A. 318,351 437,448,451,472,473,475-478,
White, A.A., III 26,28,490,496 498
White, C.S. 479, 493, 496, 497 Yih, C.-S. 28,91, 105, 194, 195,344,
Whitehouse, W.J. 510, 545 349
Whitney, J.L. 273 Yilgor, I. 543
Whittle, P. 264, 274 Yin, F.C.P. 344,351,394,451
Wiederhielm, C.A. 324, 438, 447, Young, D.R. 491,496
451 Young, J. 225
Wiener, F. 225 Young, R.H. 28
Wild, P. 542 Young, T. 165, 195
Wiley, C.W. 544 Yue, X. 534, 546
Wilkinson, P.c. 149, 153
Will, J.A. 223, 225
Williams, J.P. 272 Zamora, J.L. 533,546
Williams, M. 28 Zarda, P.R. 201,202,225
Williams, S.K. 542 Zaun, B.D. 350
Williams, W.S. 509, 542 Zeng, Y.J. 225,416,451
Willier, B.H. 545 Zerna, W. 381
Wilson, D.G. 102 Zhu, C. 149, 153
Wilson, T.A. 316, 349, 405, 423-425, Zhuang, F.Y. 216,220,221,224,225
427,428,439,449,450,451 Zierler, K.L. 336, 350, 352
Winter, D.C. 195,267,274 Zigmoid, S.H. 149, 153
Wolff, J. 545 Zimmermann, I. 421,450
Womersley, J.R. 178-180, 195 Zuidema, G.D. 497
Wong, A.Y.K. 394 Zupkas, P. 344, 352, 422, 423, 451
Wong, M. 505, 540 Zweifach, B.W. 199, 200, 225, 287,
Woo, S.L.Y. 460,498 288, 308, 315, 330-332, 349, 350,
Wood, N.B. 181, 195 351,352,448
Subject Index

Acromegaly 525 alveoli 231

Activated white blood cells 196 bronchi 229, 230
Active mass transport 276, 294 bronchioles 229, 230
and ATP, ADP 299 conducting 229
vs. facilitated transport 294 respiratory bronchioles 231
mechanisms 298 respiratory zone 229
metabolic contribution 298, 299 Strahler system 232
primary processes 300 terminal bronchioles 229, 231
secondary 300 trachea 229, 230
Activity 282 Airway resistance 257
of dilute solution 283, 289 Airway tree 157, 226
of perfect gas 282 Albatross 109, 120, 121
related to chemical potential 282 Alveolar gas pressure 227
Aerodynamic center 66 Amplification spectrum 463, 467, 468
rear aerodynamic center 80 Angiogenesis 525
Aerodynamic influence function 53 Angle of attack 64
Aeroelastic problems 71-83 Anguilla 126
divergence 71-83 Anguilliform of swimming 126
flutter 80-83 Angular momentum 6
loss of control 73-75 theorem 6,7
Airfoil theory 62-68, 83-102 Apparent coefficient of viscosity 175
dimensionless parameters 63 Apparent diffusivity 238, 239, 242, 243
Airway, gas flow 299-237 Aquatic animal propulsion 124-148
alveolar gas pressure 227, 228 anguilliform 126
dissipation function 232, 233 carangiform 128
energy balance equation 232 ciliary 125, 134-140
mouth pressure 227 energy cost 143-148
static pressure variation 235, 236 fish 140-148
transpulmonary pressure 227, 228 flagella 125, 134-140
upper airways 234 jet 126
viscous pressure drop 235 lunate tail theory 143
Airways, geometry 229-232 protozoa 124, 148-149
alveolar ducts 231 reactive force theory 140-143
559
560 Subject Index

Aquatic animal propulsion (cont.) Blood rheology 202

resistive force theory 140 apparent viscosity in lung 210, 211
wing theory 143 in pulmonary capillaries 210
Arteriole controls flow resistance 196 Blood vessels
Artery diameter vs. flow 518
resorbable suture 520 engineering of 533
strength after surgery 519 healing 518-519
Articulation mechanism residual stress 159
birds 107 stress analysis of 390-394
insects 108 stress concentration 388
Artificial leg 17, 18 structure 159
Aspect ratio of wing 67 zero-stress state 384-385
Astronaunt, model 31 Bluff bodies, flow around 68
Atelectasis of lung 439-446 circular cylinder 68
axial 443, 445 drag 69
criterion 442-443 influence of Reynolds number 69,70
definition 439 lift 68
focal 443,445 stalled wing 101-102
physical principle 440-442 vortex street 70
planar 443, 444 Boltzmann constant 279
potential energy 442 Boltzmann principle 53
reinflation 446 Bone formation 505
x-ray viewing 446 Roux, Pauwels, Carter
ATP, ADP 299 hypotheses 506
hydrolysis 299 Bone fracture healing 503
phosphorylation 299 Boundary conditions 20, 304
Automobile injuries 485-488 Boundary layer 62
Autoregulation 383 Boundary vortex 92
Axioms of mechanics I Bousinesq equation 344
Brain injury 485-488
Branching ratio 204, 205
Back pain II Bristles 122-124
Bandage 27 Brownian motion 292
Bellman theorem 55-56 Bulk modulus 426, 427
Bernoulli equation 86 Butterfly 109
Blood flow 153-225
basic equations 159, 160
in collapsible tube 190-192 Calcification of bone 506
elastic tube 172-176 of cartilage 506
energy equation 181-183 Calcium pump 300
Poiseuille formula 172 Calorie 277
pulsatile flow 176-178 Capillary blood vessels
pulse wave 165-170 fluid transport across 200, 309
resistance 184-185 permeability 313-316
Reynolds number effect 172 Carangidae 126
steady velocity profile 170-172 Carangiform propUlsion 128
transition to turbulence 172, 180, Cardiac muscle 158, 159
181 Carrier models 296-297
in veins 190-192 analysis of 297
Blood pressure Cauchy-Green tensor 377
control by arteriole 200 relation to Green's strain 378
distribution in vessels 199 Cells
in penis 23 proliferation, die, move 499
systemic 183-185 Cell membrane, transport in 294-300
Subject Index 561

Center of gravity, mass 5 Critical flutter speed 81

Channel models 295, 297, 298 Critical reversal speed of flight
analysis of 297 75
Channels of transport 295 Crotalaria spectabilis 517
Characteristic impedance 169 Ctenophora 125
Cheetah 121 Culmann 's crane 500
Chemical potential 278,281,282,288 Cylinder in Stokes flow 132-134
of dilute solution 283
of mixture of perfect gases 282, 283
as osmotic pressure 290 D'Alembert's principal 2, 34, 35
of perfect gas 282 Damping in vibration system 52
Chemotaxis 124, 148, 149 Dancing 58
Chinese traditional medicine 192-193 Darcy's law 277,291,305,338
Chord, wing 64 Dead space
Ciliophora 124 anatomical 254
Cilium 124 physiological 256
ATP 135 Decoupling of equations of motion 47-
Ca + + flux 135 49,52-56
9 + 2 structure 135 Deer botfly 121
resistive force theory 135-140 Deformation gradient tensor 356
stokes flow 134-139 Cauchy decomposition 377
thrust 137, 138 linear decomposition 378
Circular cylinder, flow field 69 product decomposition 377
Circulation of flow field 87 Diameter ratio 204, 205
in wake of wing 88 Diffusion 287
Clap-fling-flip mechanism of insect in alveolar duct 244
flight 121-123 carrier models 296
CO 2 transport 247, 248 channel models 295
Collagen coefficient 289
in lung 408-412 molecular 275, 292-294
synthesis and stress field 521 to red cells 248, 249
Collapsibility of veins 186-190 Diffusion capacity, Du lung 248
flow in veins 190-192 of CO, O 2 , CO 2 249
Complementary strain energy 376 Diffusion constant, gas 245
Concussion, head 486 re molecular weight 246
Conservation of angular momentum 6 Dilute solution 289
Constitutive equation 2, 20 Dimensional analysis 63
Continuum approach 18-21 Dipole 132
Convection and diffusion 237-244 Diptera 108
arbitrary injection of indicator 243- Dispersion coefficient (see Apparent
244 diffusivity)
coupling 238 Dissipation function, Poiseuillean 232,
interface offresh and dead gases 237 233
Taylor asymptotic solution 238-244 corrected for turbulence, entry,
Coordinate branching 233
body coordinates 360 Distensibility
convective, co-moving, dragging pulmonary blood vessels 208
along, embedded 360 pulmonary capillaries 209
Corresponding forces and displace- Divergence of wing 71-73
ments 33, 34 divergence speed 72
Couple 3 Dodecahedron 405
Coupling of heart, lung, aorta 162-165 Downwash 66, 67
Creation of circulation around a on airfoil 92-94
wing 87 in wake 92-94
562 Subject Index

Drag, drag coefficient 65, 100-102 Equipresence hypothesis 300, 301

of birds, insects 144 Euler's equation 19
circular cylinder 69 Excretion 252
form drag 101 Extracellular fluid 275-277
induced drag 67, 101
stalled wing 100-102
Dragonfly 109 Facilitated transport 294
Dynamics of particles 4-9 analysis 297-298
Dynamic structural response 463, 467, diffusion 295
468 saturated 298
Dynamics of ventilation system 258- Fading memory, system with 53
264 Fahreus-Lindquist effect 202
nonself-adjoint equation 264 Fail-safe designs 51
simulation 259, 264 Failure modes 453
variational principle 262-263 Fibroblast cells, stress field 521
Fick's law 242,277,305,316
coefficient of diffusion 289
Eigenvalues, eigenvectors 45,46 Filtration 287
Elastic eaves 468, 472-474 coefficient 315
focusing 473, 475 flux 301, 302
sound wave 472, 473 rate 311
speeds 472-474 resistance 302
Elastice efficiency of control sur- resistivity tensor 302
face 75, 76 Finger's strain tensor 375
Elastic material Finite deformation 21
definition 372 Finite wing theory 98, 99
pressure in 375 Flagellum 124
strain energy function 372, 374-376 ATP, Ca++ flux 135
Elastin 108 9 + 2 structure 135
in lung 408-412 resistive force theory 135-140
Emphysema 271 Stokes flow 124
Encarsia formosa 109, 121 thrust 137-138
Endocrine glands 525 Fleischner's lines 446
Endothelium 197, 198 Flight muscles 108, 109
Energy cost of locomotion 143-148 Flow limitation 191
of distance 146-147 Flutter in veins 186
of speed 147-148 Flutter of wing 80-83
Energy equation, blood flow 181-183 Flying 106-124
Entropy 278, 279, 283 comparing aircraft 109-113
re heat, mass transfer 283 comparing birds, insects 109-113
of perfect gas 279 evolution 106-124
Entropy production 278, 284-286 gliding, soaring 119, 120
Equation of continuity 19 insects 121-124
Equations of motion power requirement 117
Cauchy description 362, 370 range 118
decoupling 29, 47-49, 52-56 stability, control 119
Eulerian 19 Flying safety 488-490
Kirchhoff description 370 Forced expiration 256, 257
Lagrangian 29, 38-44, 369, 370 effort independent 257
Newtonian 2, 4-9 Fouette tum 58
system with damping 52 Fowler's method 255
vibration system 44, 47 Freezing point depression 290
Equilibrium 2 Frequency of vibration 45
equations of 3, 4, 362 Friction 16
Subject Index 563

Functional adaptation 503 Heart valves, flow through 161, 162

Functional residual capacity 253 Heat capacity 277
Helmholtz theorem on vortex line 91
Hematocrit distribution 201
Gait 43,44 Hemodilution 185
Gas exchange with red cells 244-249 Hemoglobin-oxyhemoglobin 246
in capillary blood vessel 246 High frequency ventilation 265-267
diffusion across membrane 245 High lift devices of birds 11 0-112
diffusion capacity, lung 248 Highway safety 485-488
Gas flow (see Respiration) 226-274 Homeostasis 499
Gas flow in airway (see Airway, gas Hookean solid 20
flow) Hooke's law 21
Gates of transport 295, 296 Horse shoe vortex 92
Generalized coordinates 29, 34, 35-38 Hovering 109, 112
Generalized displacements 33,34, 36 insects 121-124
Generalized forces 29, 34, 36 power requirement 117
Geometry of circulation system 156- Humming bird 112
158 Hydraulic conductivity 309
Gibbs, J.W. 280 Hydrogen-potassium exchange
equation 280, 283 pump 300
free energy 281 Hydrolysis, ATPase 299
thermodynamic potential 281 Hypercapnia, hypocapnia 257
Gibbs-Duhem Equation 280,281 Hyper-extension exercise 25
applications 288 Hypertension, pulmonary 517
Growth 499-537 Hypertrophy
Growth factors 524 of heart 512-515
Growth hormone 524 of lung 515-516
angiogenesis 527 of pulmonary blood vessels
ECGF, FGF 525 516-517
history 524-525 Hypoventilation, hyperventilation 257
NGF, EGF 526
releasing factor 525
Growth of organs 508 Impact 468
Growth-stress law 529 Impedance, characteristic 169
consequences 530-531 Impulsive motion 77
Incompressible material 20, 358
stress in 375
Hagen-Poiseuille flow 172 Incremental stress, strain
Harmonic functions 84 constitutive equation 426
Haversian canal 509 elastic moduli 429
Head injury 485-488 of lung 426-431
criterion 488 Index notation 8
Wayne State curve 487 Indicator dilution 333-342
Heart definition 333
coupling to lung, aorta 162-165 formulas 334-336
flow through heart valves 161, 162 in pulmonary circulation 337-342
geometry of 156, 157 Indicator dilution method 309, 316
hypertrophy 512-515 Induced drag 67, 93-95
mechanics 382, 394 Induced velocity 66, 67
muscle 158, 159 Industrial biomechanics 24
residual strain 396-400 Inertial force 2
residual stress 159 Inertial frame of reference 1
Starling's law 159 Influence coefficients 32, 33
zero-stress state 394-396 symmetry condition 33
564 Subject Index

Influence function 32, 53 Lighthill's theory, flying, swimming

aerodynamic 53 insects 122
flexibility, stiffness 32, 33 flagella, cilia 134, 136
Injury reactive force theory 140-143
head 485-488 Locomotion 106
of organs 460 Locust (Schistocerca) 22, 113-115
spinal 488-490 Longitudinal waves 474
Instability of structures 439-446 Loss of control in flight 73-75
of lung 443-446 Love wave 474
physical principle 440 Low back pain 10, 11
Interalveolar septa of lung 206 Lumbar spine 9, 11
Internal forces (see Stress) Lung
Interstitial pressure 323-325 alveoli, alveolar duct 401-408
Interstitial space 198,287,294,321 boundary value problem 416-418
fluid movement 320-323 collagen, elastin in 408-412
pressure 323-325 connection of micro-
Intracellular fluid 276 macromechanics 421
Ionophore 295 constitutive equation 412-416,431-
Irrotational flow field 83 438
Isogravimetric method 315 geometry of 156-158, 400, 401
incremental moduli 429-43 I
incremental stress, strain 426
Kamiya's shear stress theory 518 interdependence 438-439
Karman vortex street 70, 71 newborn baby 420-421
Kelvin's theorem on circulation 88,89 small perturbations 425-43 I
Kinematic viscosity 20 stress, strain, and physiology 382
Kinetic energy 29, 36, 40 stress analysis hypotheses 401
Korotkov sound 186 surface tension effect 418-425
Krogh cylinder 316-319 Lung, collapsed 27
Kronecker delta 21 Lung, remodeling 515
K T , KN coefficients of resistances Lung, trauma 475-482
134 Lung of bird 115, 116
Kutta condition 78, 88 Lymph flow 330-333
Kutta-J oukowski theorem 83-87, 123 Lymphatics 198

Lagrange's equation 38-44 Mach number 62

Lagragian function 39 Mackerels 126
Lame constants 20, 21 Macromechanics 384
Landis method 313 Macroscopic mechanics 155
Laplace equation 84,419 Mass transport 309
Laplace transformation method 53 in capillaries 309-316
Lift curve slope 66, 68, 74, 98 in interstitial space 320-330
Lift generation 87 in lymphatics 330-333
finite wing theory 98-100 in tissues 316-320
impulsive motion 76, 77 Mastigophora 124
thin wing theory 96-98 Material derivative 19, 284
Lift, lift coefficient 65 Materials
with down wash 67 of circulation system 158, 159
elastic wing 72 elastic 372, 374
with flap 74 incompressible 358, 375
oscillating wing 79-80 pressure in 375
relation to circulation 87 Mechanisms of transport in cell mem-
unsteady motion 76-80 brane 294-300
Subject Index 565

active 298-300 Newtonian fluid 20

carrier models 296--298 Newton's laws 1,2
channel models 295-298 Noninvasive diagnosis, pulse 192
facilitated 295 Non-Newtonian fluid 20
Metabolic contribution to trans- Normal coordinates 47-49
port 298,299 Normal modes of vibration 29,44-49
affinity 298
ATP and ADP 299
Microcirculation 196 Oberbeck-Bousinesq equation 344
basic equations 203 Obstructive lung disease 253
fluid transport 200 Onsager's reciprocal relation 286
geometry 197 Opalina 139
pressure, velocity distribution 199, Opening angle 384, 531
200 after surgery 532
pulmonary 203-223 blood vessels 384-388
Micro-macromechanics of lung 431- heart 394-396
438 relation to growth 531
Micromechanics 384 Optimization principle for muscle force
Microvascular beds determination 51-52
anatomy 197 Orthogonal properties of normal
topology 198 modes 47
Minute ventilation 256 Orthogonal tensor 377
Modeling, biomechanical 9, 490-491 Oscillatory lift
Mohr's circle 358, 457 airfoil 78-80
Molecular diffusion 275, 292-294 bluff body, vortex shedding 68
Moment of force 3, 4 Theodorsen ' s function 78
Moment of inertia tensor 7, 8 Osmol 290
Moment coefficient 64 Osmolarity 290
Momentum, angular, moment of 6 Osmotic pressure 289, 290, 313
Motion, equations of I, 2, 4-9, 19 Oxidative phosphorylation 299
Multiphasic structure 300-305 Oxygen transport 246, 247
boundary conditions 304
equations of motion 303
filtration resistance 302 PA , Po, P ALV , PPL notations 277,228
phenomenological laws 30 I, 302 Pain, referred 10
Muscle forces 49-52 Papilionoidea 109
consistent deformation 51 Paramecium 139
estimation by model II Parenchymal stress, lung 228
myoelectricity, relation to 10 Partial pressure of gas in tissue 245
optimization principle 51 Passive mass transport 276
redundancy 49, 50 Pauwel's hypothesis 505
Muscles Peripheral vascular resistance 184-185
cardiac 158, 159 Peristalsis 309, 342-346
flight 108, 109 Permeability, capillary wall 313-316
Musculoskeletal system 400 Permeability constant 200, 291
Myocardium, strains 359 Permutation symbol 8
Myoelectric activity 10 Phenomenological laws 301-302
Phosphorylation 299
Piezoelectric effect
Navier's equation 21 on bone 507
Navier-Stoke's equation 20 on growth of organs 508
Neonatal lung 420, 421 Pleural cavity, pressure 227
Nerst's postulate 279 Pneumonectomy 515-516
Nerve, stressed II Pneumothorax 27
566 SUbject Index

Poiseuille formula 172 by hypoxia 516

modified by elasticity 174, 176 on pulmonary arterioles 517
Poisson's ratio 21,426 Pulmonary microcirculation, indicator
Polar-decomposition theorem 377 dilution 337-342
Polyhedron 405 Pulmonary pleura 226
lung structure 405-408 Pulsatile flow 176--178
Polypeptides 526 Pulse wave, arteries 165-170
Porous medium 291 characteristic impedance 169
Positive definitiveness of quadratic reflection, transmission 168-170
forms 33 Pulse wave diagnosis 192
Posture control 42 Pure shear 358
Potential energy 29
Potential function 84
Power of flight 117, 118 Range of flight 118, 119
Pressure, osmotic, tissue 313 Rayleigh waves 474
Product of inertia 8 Red blood cell 201, 202
Prosthesis 17 Reducedfrequence 63,69,70,81
Protozoa 124 Redundancy, degree, of muscles 49
Pulmonary blood flow 203-223 Reference state 354
branching pattern 204-205 Referred pain 10,12
capillary blood flow 211-223 Reflection coefficient 200, 291, 309
collapsibility, capillaries 209 Regional difference, pulmonary blood
distensibility, blood vessels 208,209 flow 214-223
interalveolar septa 206 Regional lung volume 253
pulmonary capillaries 207 Relaxation function of visco-
Strahler system 203 elasticity 53
synthesis 220 Remodeling
validation 221-223 blood vessels 518-519
vasculature 203, 204 Fung's hypothesis 529-530
Pulmonary capillaries 207 heart 512-515
compliance 210 lung 515, 516
pressure-thickness relation 209 muscle 520
sheet model 207 nonuniform process 527-529
vascular-space-tissue ratio 207 pulmonary blood vessels 516, 517
Pulmonary capillary blood flow 211- soft tissues 511
223 in space flight 520--521
basic equations 211-212 surgical wounds 519-520
effect of gravity 214, 215 zero-stress state 527
flow in whole septum 213, 214 Reorientation surgery 506
solution, simplest case 212, 213 Repair of organs 460
synthesis with macrocircula- Residual strain and stress 384
tion 220--223 in blood vessel 388-393
waterfall phenomenon 215 in heart 396--400
zone 2 flow 215-220 Residual volume 253
Pulmonary function tests 253-258 Resilin 108, 109
airway resistance 257, 258 Resistance, blood flow 184, 185
dead space 254, 256 Resorbable suture in artery 519
forced expiration 256, 257 Resorption 499
Fowler's method 255 beneficial use of 507
lung compliance 258 Respiration 226--271
minute ventilation 256 anatomy 227, 229
volume measurements 253, 254 convection and diffusion 237-244
Pulmonary hypertension dynamic analysis 258-264
by Crotalaria spectabilis 517 gas exchange with red cells 244-249
Subject Index 567

gas flow in airway 229-237 Spine 9, II

high frequency ventilation 265-267 curved 26
pulmonary function tests 253-258 disk 9, ll, 12
ventilation/perfusion ratio 249-253 muscle forces 11
Response spectrum 465,467,468 nucleus pulposus 9
Restrictive lung disease 253 Sports techniques 12-17
Retention 252 Stagnation point 87
Reversal of flight control 74, 75 Stall, wing, stalling angle 66, 102
Reynolds number 62, 179, 180, 196 Stalling, birds 110
Rheology of blood 202 Starling's law of filtration 200, 277,
Rigid body 7 290,291, 305, 309, 315, 330, 338
Rotation 377, 378 Starling's law, heart 159
Cauchy approach 377 Stokes flow 128-139
Chen Zhi-da approach 378 of cilia 139-140
polar decomposition 377 coefficient of resistance 134
Rotifera 125 of cylinder 132-134
Roux's hypothesis 505 Dipole 132
Russell traction 25 of flagella 134-139
of sphere 132
Stokeslet 128-131
Salmon 144 Stokeslet 128-131
Sarcomere 158, 159 Stokes number 179
Schistocerca gregaria 113 Strahler system 203,205
Schmidt number 266 Strain 350-360
Scoliosis 26 Almansi's strain tensor 357
Seal, Weddell 482 Cauchy's infinitesiinal strain
Segmental movement 29, 52 tensor 357
Semi-permeable membrane 289,290 curvilinear coordinates 360
Setae 122-124 deformation gradient 356
Severity Index, head injury 487 examples 357-360
Shapiro's analogy 190 Finger's strain tensor 375
Shapiro's speed index 191 Green's strain tensor 357
Shear modulus 21, 426, 427 incompressible material 358, 375
Shear waves 474 in myocardium 359
Shock loading 460 reference state 354
Skin substitute 534 Strain energy 29, 32, 33, 41, 371
biodegradable scaffold 535 Strain energy function 374-377
patient's own cells 535 stress relationship 374
wound contraction 536 Stream function 84
Slender-body theory 140 Strength of tissues 452
Small boards (slats) bandage 27 failure modes 453-459
Soaring of birds 119, 120 tearing 459
Sodium-potassium exchange pump 300 Stress 353, 361-377
Solid immobile matrix 277, 300-305 Cauchy's 362
boundary conditions 304 examples 364-369
equations of motion 303 general formulas 368, 369
phenomenological laws 301, 302 in incompressible material 375
Solubility of gas in tissue 245 Kirchhoff 363, 364
Somatostatin 525 Lagrangian 363, 364
Sound speed in gas, water 63 matrix operation 369
Spectrum, response, amplification 465 normal 361
Speed of sound, in organs 472-475 shear 361
Sphere in Stokes flow 132 strain energy 353, 372
Spinal injury 488-490 surface traction 361
568 Subject Index

Stress (cont.) endothelial cell seeding 533

symmetry property 368 skin 534-536
vector 361 Tissue pressure 200, 201, 313
Stress concentration 473 Tissues
Stress field of sound speed 472
collagen synthesis 521-524 strength 452
fibroblast 522-523 tearing 459
Stress-strain relationship Tolerance 452
elastic material 372, 374 head injury 485-488
incompressible material 375 of organs 482-490
via strain energy 353, 354, 372 severity index 487
Stretch tensor 377 shock loading 460
right and left 377 spinal injury 488-490
Strouhal number 63 whole body 483-485
birds 95,96 Total lung capacity 253
fish 96 Trailing vortex 92
flow around cylinder 69, 70 Transient fluid dynamic forces 76-80,
physical interpretation 82 94-96
wing flutter 81 Transport equations 275-305
Structural response 460 cell membrane 294-300
Summation convention 8 molecular diffusion 292-294
Superposition principle 33 solid matrix 300-305
Surface tension thermodynamics 275-292
definition 418 Transpulmonary pressure 227
equation of equilibrium 419 Transverse waves 474
Laplace's equation 419 Trauma 452
in newborn baby's lung 420, 421 aircraft 488-491
surfactants 420, 421 automobile 485-488,491
Surfactants, pulmonary 420, 421 edema 478
Sweep back, sweep forward, wings 73 of lung 475-482
Swimming 106, 124-148 of organs 482-490
Synthesis of micro- and macro- circula- prevention 491
tion, lung 220-223 Turbulence in arteries 180-181, 184-
Systemic blood pressure 183-185, 196 185
Two-phase fluid, blood 202

Taylor theory of convection-diffu-

sion 238-244 Ultimate load 453
Tearing experiment 459 Upper airways, flow in 234
Temperature, absolute 278, 279
Tetrakaidekahedron 405
Theodorsen 's function 78 Vascular lung disease 253
Thermodynamics, first, second, third Veins 185-190
laws 277-279, 285 collapsibility 186-190
Thermodynamics of transport 275-292 flow in 190-192
Thin wing theory 96-98 Ventilation/perfusion ratio 249-253
Thoroughfare channels 200 exact solution 252
Thrips 122 integral equation 251, 252
Thymidine, incorporation 516 method of measurement 250
Tidal volume 253 practical solution 252, 253
Tissue engineering 499, 500 probability distribution 250, 251
autologous substitute 534 Ventilation system 258-264
blood vessels 533 Vibration 463
Subject Index 569

Vibrations 29, 44-47 loss of control 73-75

system with damping 52-56 oscillating 95, 96
system with fluid dynamic loads 53- power requirement 117-118
54 range 118
Visceral pleura 226 thin wing, steady 96-98
Viscoelastic material 53 Wolffs law 500-502, 508
Viscosity, coefficient 20 Cowin's formulation 509-511
Viscosity of blood 202 density and strength 511
Vital capacity 253 math formulation 508
Volkmann canal 509 solid volume fraction 511
Vortex, shedding 68 surface remodeling 509
Vortex street 70, 71 Work done by air on oscillating
Vorticity of flow field 83 wing 82
Vorticity in wake of wing 88 Work done in generalized coordi-
finite wing, unsteady 91 nates 36-38
VSTR 207 Work in finite deformation 371
Work physiology 24
Womersley number 178-180, 196
Wagner effect, function 77, 123 Wound contraction 536
Wake, behind cylinder 68 Wound healing 519
behind a wing 88-96
down wash 92-94
oscillating wings 95-96 Young's modulus 21
Wasp, chalcid 109, 113
Wasp flight 121-124
Waterfall phenomenon 186, 190, 215- Z factor for dissipation 233
220,257 Zero-lift line 66
Wave equation 166, 167 Zero-stress state 384
Wave speed 166 of blood vessels 384-394
Wayne State curve 486, 487 effect on stress concentration 388-
Weis-Fogh theory of hovering 112,122 390
Whole-body acceleration toler- of heart 394-400
ance 482-485 hypertension 386-388
Wilhelmy-Langmuir trough 420-422 Zone 1, 2, 3 of lung 214, 215
Windkessel theory 163-164 Zone 2 blood flow, lung 215-220
Wind shear 109, 120-121 collapsed area 218, 219
Wings flow formula 220
of birds 107, 109, 110-115 histological confirmation 218
divergence 71-73 open and closed capillaries 216
finite wing, steady 98-100 reduced flow due to closed sheet 219
flutter 80-83 sluicing 216, 219
high lift 110-112 stability of open sheet 216,217
of insects 108, 110-115 stable form 218