Join Us on the Internet

WWW: http://'WINW.thomson.com
EMAIL: [email protected]

thomson. com is the on-line portal for the products, services and resources available
from International Thomson Publishing (ITP).

This Internet kiosk gives users immediate access to more than 34 ITP publishers and over
20,000 products. Through thomson.com Internet users can search catalogs, examine
subject-specific resource centers and subscribe to electronic discussion lists. You can
purchase ITP products from your local bookseller, or directly through thomson. com.

Visit Chapman & Hall's Intemet Resource Center for information on our new publications,
links to useful sites on the World Wide Web and an opportunity to join our e-mail mailing list.
Point your browser to: http://www.chaphall.com or
http://www.thomson.com/chaphail/mecheng.htmifor Mechanical Engineering

A service of I®p®
 .&pplicatioD of
LEI'. "'P.D. Theory

Senior Specialist and MDC Fellow

McDonnel1 Douglas Corporation
Huntington Beach, CA

with
Geor=e BocJaoath
Jam.es Glassco

m
SPRINGER-SCIENCE+BUSINESS MEDIA, B.V.
Cover design: Curtis Tow Graphics
Copyright ~ 1997 Springcr Scicnce+l3usiness Media Dordrccht
Originally published by Chapman & Hali in 1997
Soflcovcr rcprinl of Ihc hardcovcr ISI edilion 1997

AII rights resetved. No pari of this book covered by the copyright hereon may be reproduced or
used in any form or by any means-graphi<;, electronic, or mechanical. induding photocopying.
recording. taping. or information storage and retrieval systems-without the writlen permission of
the publisher.

1 2 3 4 5 6 7 810 9XXX 01 00 99 98 97

Library of Congress Calaloging-in-Publkation Oala

Farahmand, Bahram.
Faligue and fracture mechanies of high risk parts: applkation of
LEFM &: FMDM theory / Bahram Farahmand, George Bockrath, James
Glassco.
p. CJn.
J"dudes bibliograph ical ref."...nces and index.
ISBN 978-1-4613-7753-5 ISBN 978-1-4615-6009-8 (eBook)
DOI 10.1007/978-1-4615-6009-8
1. Metals - Fatigue. 2. Fracture mechanics. 3. Metals - Ductility.
4. Service life (Engineering). 1. Bockrath, George. n. Glassco,
James. JlJ.lltle.

"..,..
TA460.F337 1997
620.I'66--dc21
CIP

British Library Dtiloguing in Publiution Dati iViibble

"Fatigue and Fraclure Mechanics of High Risk Parts" is inlended to present technically accurate and
authoritative information from highly regarded sources. The publisher, editors, authors, advisors,
and contributors have made every reasonable effort 10 ensure the accuracy of the information, but
cannOI assume responsibility for the accuracy of all information, or for the consequences of its use.
This book is dedicated to the memory of my loving father,
Hadi, who instilled in me his enormous respect for higher
education, and the values of patience and discipline.

v
Contents

Contents vii

Preface xiii

Chapter 1 A Brief Introduction to Fatigue

and Fracture Mechanics 1

1.1 Failures in Riveted Joints

l.2 Failures in Welded Joints 2
1.3 Fatigue and Fracture Mechanics 3
References 11

Chapter 2 Conventional Fatigue

(High- and Low-Cycle Fatigue) 13

2.1 Background 13
2.2 Cyclic or Fluctuating Load 15
2.3 Fatigue Spectrum 18
2.3.1 Load Spectrum for Space Structures Using
the Space Transportation System (STS)
or Shuttle 28
2.3.1.1 Flight Cycles 31
2.4 The S-N Diagram 32
2.4.1 Factors Influencing the Endurance Limit 34
2.4.1.1 Brief Description of Constant Amplitude
Axial Fatigue Tests 34
2.4.2 Empirical Representation of the S-N Curve 38
2.4.3 Parameters Affecting the S-N Curve 40
2.5 Constant-Life Diagrams 41
2.5.1 Development of a Constant-Life Diagram 42

vii
viii Contents

2.5.2 Development of a Constant-Life Diagram

by Experiment 43
2.6 Linear Cumulative Damage 47
2.6.1 Comments Regarding the Palmgren-Miner
Rule 50
2.7 Crack Initiation (Stage I) and Stable Crack Growth
(Stage II) 51
2.7.1 Fracture Surface Examination 51
2.7.2 Introduction to Crack Initiation 57
2.7.2.1 Fracture Due to Monotonic Applied
Load 57
2.7.2.2 Crystal Structures of Metals and Plastic
Deformation 59
2.7.3 Crack Initiation Concept (Intrusion and
Extrusion) 63
2.8 Low-Cycle Fatigue and the Strain-Controlled
Approach 67
2.8.1 Crack Initiation Life 68
2.9 The Conventional or Engineering Stress-Strain
Curve 69
2.10 The Natural or True Stress-Strain Curve and Hysteresis
Loop 70
2.1O.l The Natural or True Stress-Strain Curve 70
2.10.2 Hysteresis Loop 71
2.l1 Cyclic Stress-Strain Curve 74
2.11.1 The Power Law Representation of the Cyclic
Stress-Strain Curve 78
2.12 Strain-Life Prediction Models 82
2.12.l Transition Point, NT 86
2.12.2 The Mean Stress Effect 88
2.l2.3 Fatigue Notch Factor, Kf 88
2.12.4 Stress and Strain at Notch (Neuber
Relationship) 93
2.l3 Universal Slope Method 98
References 99

Chapter 3 Linear Elastic Fracture Mechanics 103

3.1 Energy Balance Approach (the Griffith Theory

of Fracture) 103
3.2 The Stress Intensity Factor Approach 105
3.2.1 General 105
 Contents ix

3.2.2 Crack Tip Modes of Deformation 106

3.2.3 Derivation of Mode I Stress Intensity
Factor 107
3.2.3.1 Airy Stress Function, 1> 108
3.2.4 Combined Loading 114
3.2.5 Stress Intensity Factor Equations for Several
Through Cracks 117
3.2.6 Critical Stress Intensity Factor 122
3.3 Fracture Toughness 124
3.3.1 Fracture Toughness and Material
Anisotropy 128
3.3.2 Other Factors Affecting Fracture
Toughness 130
3.4 Residual Strength Capability of a Cracked Structure 134
3.4.1 Residual Strength Diagram for Material with
Abrupt Failure 136
3.4.2 The Apparent Fracture Toughness 138
3.4.3 Development of the Resistance curve (R-curve)
andKR 139
3.5 Plasticity at the Crack Tip 146
3.6 Plastic Zone Shape Based on the Von Mises Yield
Criterion 148
3.7 Surface or Part Through Cracks 152
3.7.1 Stress Intensity Factor Solution for Surface Cracks
(Part Through Cracks) 153
3.7.2 Longitudinal Surface Crack in a Pressurized
Pipe 157
3.7.3 Part Through Comer Crack Emanating
from a Hole 158
3.7.4 Part Through Fracture Toughness, KIe 159
3.7.5 The Leak-Before-Burst (LBB) Concept 160
3.8 A Brief Discription of ASTM Fracture Toughness
Determination 164
3.8.1 Plane Strain Fracture Toughness (KIJ Test 164
3.8.1.1 Standard KIc Test Specimen and Fatigue
Cracking 165
3.8.2 Plane Stress Fracture Toughness (KJ Test 169
3.8.2.1 M(T) Specimen for Testing Kc 171
3.8.2.2 Grip Fixture Apparatus, Buckling
Restraint, and Fatigue Cracking 172
References 174
x Contents

Chapter 4 Fatigue Crack Growth 177

4.1 Introduction 177

4.1.1 Stress Intensity Factor and Crack Growth
Rate 179
4.2 Crack Growth Rate Empirical Descriptions 180
4.2.1 Brief Review of Fatigue Crack Growth
Testing 188
4.3 Stress Ratio and Crack Closure Effect 194
4.3.1 Elber Crack Closure Phenomenon 196
4.3.1.1 Threshold Stress Intensity Factor,
b.Kth 199
4.3.2 Newman Crack Closure Approach 203
4.4 Variable Amplitude Stress and the Retardation
Phenomenon 216
4.4.1 Wheeler Retardation Model 218
4.4.2 Willenborg Retardation Model 219
4.5 Cycle by Cycle Fatigue Crack Growth Analysis 227
4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic
Loading 230
4.6.1 Preloaded Bolt Subjected to Cyclic
Loading 230
4.6.2 Fatigue Crack Growth Analysis of Pads in a Bolted
Joint 237
4.7 Material Anisotropy and Its Applications in Beam
Analysis 248
References 249

Chapter 5 Fracture Control Program

and Nondestructive Inspection 253

5.1 Introduction 253

5.1.1 Design Philosophy 253
5.1.1.1 Slow Crack Growth Design 254
5.1.1.2 Fail-Safe Design and Low Released
Mass 256
5.1.1.3 Material Selection, Testing, Manufacture,
and Inspection 257
5.2. Fracture Control Plan 258
5.2.1 Purpose 258
5.2.2 Causes of Unstable Crack Growth 259
 Contents xi

5.2.3 Fracture Control in the Stages of Hardware

Development 259
5.2.4 Typical Fracture Control Activities 260
5.2.5 Data Required for Fracture Control 261
5.2.6 Contents of a Fracture Control Plan 261
5.2.7 Fracture Control Classification of
Components 261
5.2.8 Analysis and/or Testing to Determine Fracture
Control Acceptability of Hardware 263
5.2.9 Traceability and Design Annotation for
Drawings 265
5.2.10 Overall Review and Assessment of Fracture
Control Activities and Results 267
5.2.11 Fracture Control Board 267
5.3 Nondestructive Inspection Techniques 269
5.3.1 Introduction 269
5.3.2 Liquid Penetrant Inspection 269
5.3.3 Magnetic Particle Inspection 270
5.3.4 Eddy Current Inspection 270
5.3.5 Ultrasonic Inspection 271
5.3.6 Radiographic Inspection 272
5.3.7 Flaw Size Verification 274
5.3.8 Probability of Detection Statistics 276
5.3.9 Variables Affecting NDI Flaw Detectability
Testing 278
5.3.9.1 Variables Affecting the Liquid Penetrant
Inspection 279
5.3.9.2 Variables Affecting the Magnetic Particle
Inspection 283
5.3.9.3 Variables Affecting the Eddy Current
Inspection 284
5.3.9.4 Variables Affecting the Ultrasonic
Inspection 284
5.3.9.5 Variables Affecting the Radiographic
Inspection 286
5.3.10 Critical Initial Flaw Sizes for Standard
NDI 286
5.3.11 Special Level Inspection 288
References 288
xii Contents

Chapter 6 The Fracture Mechanics

of Ductile Metals Theory 289

6.1 Introduction 289

6.2 Fracture Mechanics of Ductile Metals 290
6.3 Determination of the aUF
ac
Term 292

6.4 Determination of the au

ac
u Term 294
6.4.1 Octahedral Shear Stress Theory (Plane Stress
Conditions) 295
6.5 Octahedral Shear Stress Theory (Plane Strain
Conditions) 303
6.6 Applied Stress, (J', and Half Crack Length, c,
Relationship 305
6.7 Mixed Mode Fracture and Thickness Parameters 307
6.8 The Stress-Strain Curve 309
6.9 Verification of FMDM Results with the Experimental
Data 309
6.10 Fracture Toughness Computation by the FMDM
Theory 312
6.10.1 Introduction 312
6.10.2 Fracture Toughness, Kc ' Evaluation for 2219-T87
Aluminum Alloy 313
6.10.3 Fracture Toughness, K c ' Evaluation for 7075-T73
Aluminum Alloy 316
References 322

Appendix A NASAIFLAGRO 2.0 Materials

Constants 323

Appendix B Equations for Uniaxial True Stress

and True Strain 358

Index 367
Preface

In the preliminary stage of designing new structural hardware that must perform a
given mission in a fluctuating load environment, there are several factors the
designers should consider. Trade studies for different design configurations
should be performed and, based on strength and weight considerations, among
others, an optimum configuration selected. The selected design must be able to
withstand the environment in question without failure. Therefore, a comprehen-
sive structural analysis that consists of static, dynamic, fatigue, and fracture is
necessary to ensure the integrity of the structure. During the past few decades,
fracture mechanics has become a necessary discipline for the solution of many
structural problems. These problems include the prevention of failures resulting
from preexisting cracks in the parent material, welds or that develop under cyclic
loading environment during the life of the structure.
The importance of fatigue and fracture in nuclear, pressure vessel, aircraft, and
aerospace structural hardware cannot be overemphasized where safety is of
utmost concern. This book is written for the designer and strength analyst, as well
as for the material and process engineer who is concerned with the integrity of the
structural hardware under load-varying environments in which fatigue and frac-
ture must be given special attention. The book is a result of years of both acade-
mic and industrial experiences that the principal author and co-authors have
accumulated through their work with aircraft and aerospace structures. However,
the material contained in this book can also be applied to other industries, such as
nuclear, pressure vessel, and shipbuilding, where fracture and fatigue are equally
important. Moreover, the scope and contents of the book are adequate for use as a
textbook for both graduate and undergraduate level courses in the mechanical,
material, and aerospace engineering departments. Each chapter has several exam-
ple problems that have been handpicked from industrial experiences that the
authors have accumulated throughout the years in the field of fracture mechanics.
This book addresses the conventional fatigue approach to life evaluation of a
structural part where it is assumed that the structure is initially free from cracks
and, after N number of load cycles, the crack will initiate in some highly localized
stressed areas. In contrast to the conventional fatigue approach, Linear Elastic
Fracture Mechanics (LEFM) assumes the existence of a crack in the structural part
in the most unfavorable location perpendicular to the applied load. This book cov-

xiii
xiv Preface

ers in detail both the conventional fatigue (the stress to life, S-N, and the strain to
life, e-N) and the LEFM approaches to determining the life of structural hardware
in a load-varying environment. In using the LEFM approach to evaluate the life of
a part, an initial crack size provided by the standard nondestructive inspection
method and the fracture toughness data for the material under study must be avail-
able to the analyst. Obtaining fracture toughness data through laboratory testing is
costly and time consuming. A new approach, called Fracture Mechanics of Duc-
tile Metals (FMDM), can satisfactorily correlate the fracture toughness data
obtained through the analysis with the test data and was first developed in the late
1960s by the authors of this book.
In the late 1960s, G. E. Bockrath and 1. B. Glassco, while working at McDon-
nell Douglas Aerospace Company (formerly known as Douglas Aircraft Com-
pany) as principal engineers, were researching high fracture stress under
low-cycle fatigue of high-strength material. It was noticed that some Ti-6AI-4V
titanium alloy (ST A condition) test specimens containing small part through
cracks fractured at the ultimate stress. This led to the question of how large can a
crack be in a specimen that will fracture at the ultimate stress. For metals that have
a large amount of necking, under normal rate of loading, the critical crack size
was large, and for a metal with no necking (such as beryllium) it was very small.
The maximum flaw size that has a failure stress equal to the ultimate stress can be
obtained through the FMDM computer program.
Another important observation as a result of this study was made. While study-
ing the fracture surface of test specimens, the authors noticed that metals whose
full-range uniaxial tensile engineering stress-strain curve showed local plastic
deformation beyond the ultimate stress (necking) had a slope of 45 degrees over
part, or all, of the fracture surface. This indicated that ductile metals with
observed necking absorb more energy at the crack tip than metals whose
stress-strain curves do not show plastic deformation beyond the ultimate stress.
The crack tip plastic deformation defined by the FMDM theory is composed of
two distinct regions: (1) the local strainability at the crack tip (the region of highly
plastic deformation) and (2) the uniform strainability near the crack tip. The
energy absorption rate for these two regions was calculated (see Chapter 6) and
used to extend the Griffith theory of fracture that originally was developed for
brittle materials. In contrast to LEFM, the FMDM theory was shown to accurately
correlate with test data for commonly used structural metals over a wide range of
crack sizes at stresses above as well as below the yield stress. The FMDM com-
puter program is capable of generating the variation of fracture toughness as a
function of the material thickness for ductile metals and requires only the
stress-strain curve as an input.
The contents of this book represent a complete overview of the field of fatigue
and fracture mechanics, a field that is continuously being advanced by many
investigators.
 Preface xv

The author is very thankful to the following individuals: Mr. David Ollodort
and Mr. Kerry Michaels of McDonnell Douglas Aerospace for contributing a por-
tion of the material in Chapter 5; to Mr. Robert Muller for his constructive sug-
gestions pertaining to sections of the fracture control plan. Also my great
appreciation goes to Dr. Bijan Irani Nejad for his contribution to the sections of
load spectrum material. Special thanks also goes to Mr. Vector Kerlins and Mr.
Ray Toosky of the metallurgy department of McDonnell Douglas Aerospace. The
author wishes to thank Professor Ares Rosakis of the California Institute of Tech-
nology for his constructive comments regarding Chapter 6. As a final note, the
author would like to acknowledge the co-authors, Mr. George Bockrath and Mr.
Jim Glassco, for their contribution to the development of the FMDM theory and to
Mr. David Ollodort for his editorial assistance with the manuscript.
I would also like to acknowledge the support of my loving and devoted wife,
children, and especially my dear mother whose sacrifices made it possible for me
to complete this book.
Chapter 1

A Brief Introduction to Fatigue

and Fracture Mechanics

During the past few decades fracture mechanics has become a necessary disci-
pline for the prevention of many structural failures. These are problems resulting
primarily from structures containing part through or through cracks. Such cracks
can originate in many ways. For example, they may be introduced as cracks or as
incipient cracks during manufacture of structural parts; they may grow from
defects in the parent metal, from incomplete welds, from shrink cracks or other
imperfections in weldments; or they may nucleate and grow in structure under
fatigue loading. It is the role of fracture mechanics to determine when these cracks
become critical, that is, when they will reach a size at which the crack will grow
catastrophically at an operational stress well below the yield strength.
The following are descriptions of a few documented fatigue-related failures
that had occurred during the service life of the structure and are directly extracted
from Reference [1] by Parker. Extensive scientific investigation on the nature of
these structural failures indicated that in almost all cases, flat (cleavage) or brittle
fracture occurred in a catastrophic manner, with high velocities and little or no
shear lips. Defects in the welded parts, induced residual stresses in the weld and
during the structural assembly, steel composition, cyclic and corrosion environ-
ment, transition temperature effect (ductile-brittle failure) poor design, and lack of
inspection can all be responsible for these failures. Of the sources of failures
reported in Reference [1], seven started in riveted structures as the source of stress
concentration and 14 in welded parts.

1.1 Failures in Riveted Joints

• Water Standpipe, Gravesend, Long Island, New York, October 7, 1886

• Gasholder, Brooklyn, New York, December 23, 1898
• Water Standpipe, Sanford, Maine, November 17,1904

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts
© Chapman & Hall 1997
2 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

• Molasses Tank, Boston, Massachusetts, January 15, 1919. The tank con-
tained 2,300,000 gallons of molasses that caused 12 deaths and 40 injuries.
• Crude Oil Storage Tank, Ponca City, Oklahoma, December 19
• Eight Riveted Crude Oil Tanks, south and midwest United States, 1930-1940
• Oil Storage Tank, midwest United States, December 14,1943

1.2 Failures in Welded Joints

• Three Vierendeel Truss Bridges, Albert Canal (in Hasselt, Herenthals-

Oolen, and Kaulille), Belgium, 1938 through 1940
• Duplessis Bridge, Three Rivers, Quebec, Canada, January 31,1951
• Spherical hydrogen storage Tank, Schenectady, New York, February 1943.
• Spherical ammonia tank, Pennsylvaia, March 1943
• Spherical pressure vessel, Morgantown, West Virginia, January 1944
• Cylindrical gas pressure vessel, Cleveland, Ohio, October 1944
• Five oil storage tanks in Russia, December 1947
• Crude oil storage tank, midwest area, United States, February 1947
• Two oil storage tanks, Fawley, England, February and March 1952
• Failures of three empty storage tanks in Europe during 1952
• Penstank at the Anderson Ranch Dam, Boise, Idaho, January 1950.

The total cost associated with material fracture and failure in the United States
is estimated to be $88 billion dollars per year (based on 1978 dollars) [2].
Research directed toward fracture-related problems could reduce the costs of
structural failure in the United States by almost 30%; that is, an estimated $21 bil-
lion dollars per year (based on 1978 dollars). Extensive study and research on
these failures indicated that potential savings can be obtained by focusing atten-
tion on two major areas: material and structural. The summary of research in these
areas concluded that (1) a reduction in material variability (tensile and yield stress,
as well as fracture properties) which can contribute to prevention of structural fail-
ures; (2) better use of fracture mechanics in evaluating the life of the component;
(3) increased use of nondestructive inspection and improvement in NDE tech-
niques; and (4) implementation of a fracture control plan can all reduce the costs
of fracture related failure in the United States [2].

1.3 Fatigue and Fracture Mechanics

Fracture mechanics theory began with A. A. Griffith (1893-1963) at the Royal

Aircraft Establishment (RAE) in the United Kingdom with the help of the mathe-
matical work of Professor C. E. Inglis [3, 4], but the major development took
place in the United States at the Naval Research Laboratory (NRL) in 1950 by
George R. Irwin [5]. Griffith's theory applies to brittle materials, that is, materials
 1.3 Fatigue and Fracture Mechanics 3

that, under nonnal rates of loading (monotonically increasing tensile loads as

compared to fluctuating or cyclic loads) and at room temperature, do not undergo
plastic defonnation before fracture in a tensile test.
The Griffith theory as described in "The Phenomenon of Rupture and Flow in
Solids" assumes material contains crack-like defects and that work must be per-
fonned on the material to supply the energy needed to propagate the crack by cre-
ating two new crack surfaces. For brittle materials (such as high-strength steels),
the Griffith theory gives the correct relationship between the fracture strength, O"F'
and critical flaw size, c. However, this theory is not satisfactory as a means for
detennining the fracture strength of ductile material, because cracked structural
metals undergo considerable plastic defonnations in the region at the crack tip
prior to final fracture. Almost 30 years after Griffith's theory, Irwin and Orowan
[6, 7] pointed out that for ductile material to fracture, the stored strain energy is
consumed for both the fonnation of two new cracked surfaces and the work done
in plastic defonnation. This energy balance approach can be stated as:
Energy Input ~ Surface Energy + Energy of Plastic Defonnation
Irwin realized that for ductile materials, the energy required to fonn the new crack
surfaces is negligible compared to the work done in plastic defonnation at the
crack tip. Generally in metallic materials, the energy required for plastic defor-
mation is approximately 10 3 times larger than the surface energy. This is the rea-
son why metallic materials have much greater resistance to fracture than brittle
material. Irwin further proposed that the elastic energy release rate can be
regarded as a force called crack driving force that is denoted by G (after Griffith),
where its critical value, Gc ' corresponds to crack instability. The quantity G is
equal to the derivitive of elastic energy and its magnitude is the same for cracked
plate subjected to constant load or fixed ends conditions.
Fracture mechanics became an engineering discipline with the development of
Linear Elastic Fracture Mechanics (LEFM). Griffith's energy balance approach is
applicable only to the crack analysis cases where the condition of instability under
static load is required. However, this approach is not applicable to other important
cases, such as stable crack growth that occurs when a crack is subjected to cyclic
loading. The stress intensity parameter, K, can address important cracking problems
(e.g., stable crack growth under fluctuating load, as well as the condition of instabil-
ity) that are of interest in the aircraft, space vehicle, pressure vessel, nuclear, and
shipbuilding industries where safety of people and structures is of great concern.
LEFM is an extension of the Griffith theory, modified to account to a limited
extent for the presence of the plastic zone at the crack tip. To be able to correlate
the LEFM results with the test data, the energy consumed at the crack tip for the
plastic defonnation must be considered. Irwin and Dugdale [8, 9] estimated the
size of the plastic zone fonned at the crack tip by assuming that local stresses are
equal to the yield stress of the material. In Dugdale's model, the shape of the plas-
tic zone fonned at the crack tip was considered to be in the fonn of a strip in front
4 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

of the physical crack tip (a region that carries yield stress tending to close the
crack) called the strip yield model, whereas Irwin assumed that the plastic zone
ahead of the crack tip is a circle. In both models, the effective crack length, rather
than the physical crack length, was used to calculate the stress intensity factor.
The critical value of the crack tip stress intensity factor is associated with the
fracture toughness of the material. Fracture toughness can be defined as the mate-
rial's ability to resist unstable cracking in a non-corrosive environment. It can be
described in terms of the critical stress intensity factor under the condition of
plane strain, K Ic ' or plane stress, K c ' under slow rate of loading and normal tem-
perature. Ductile materials under normal rate of loading have the ability to absorb
energy and deform plastically prior to rupture, hence possessing high fracture
toughness. Brittle material on the other hand fails at low stresses in a sudden man-
ner with high velocities and over great distances, hence possessing low fracture
toughness and tending to have cleavage or low energy fracture.
To be able to resolve any cracking problem, the fracture toughness value must
be available to the analyst. This is analogous to the strength analysis requirement
that the limit stress, 1 (T, must always be compared with both the yield, (TYield' and
the ultimate, (TUll' strength of the material. That is, the analysis would be incom-
plete without both material allowables. With the material fracture toughness value
available from ASTM Standard specimen testing, the residual strength capability
of a structural part containing a crack can be determined through a relationship
that describes the critical stress intensity factor, fracture stress, and the crack
geometry. The load-carrying capability or residual strength of a structure is sig-
nificantly affected by the presence of cracks and it is substantially lower than the
strength of the undamaged structure. The residual strength diagram (the fracture
stress as a function of crack length or time) can be used for the purpose of plan-
ning the inspection interval prior to the total loss of the structural part during its
service life. For single load path structure, the residual strength capability of the
structure or the ability of the structure to tolerate damage can be determined with
ease. For fail-safe or mUltiple load path structures or structures with a crack arrest
design, such as stiffened structure, the construction of a residual strength diagram
requires calculation of a failure stress at which partial failure occurs when a load
path fails and determination of whether the redistributed load can be tolerated by
the remaining structure.
Almost alliow-to-medium strength structural steels that are used in construct-
ing bridges, ships, pressure vessels, etc. and many aluminum alloys used in the
design of aircraft and space structure have sufficient ductility and under normal
rate of loading form a large plastic zone at the crack tip. The R-curve method [10],

I The terms "limit stress", "applied stress", "gross stress", and "operating stress" all have the same

meaning and are used interchangeably throughout the literature. No factor of safety is associated with
these values.
 1.3 Fatigue and Fracture Mechanics 5

an extension of LEFM into elastic-plastic fracture mechanics, can be utilized to

describe the crack instability when the plastic zone size is large and substantial
stable crack growth is expected prior to catastrophic failure (tearing type fracture).
The construction of the R-curve for material with considerable plasticity at the
crack tip is well documented in ASTM E-561. It is shown that the R-curve can be
described as a plot of material crack growth resistance (expressed in units corre-
sponding to Kc associated with the plane stress fracture toughness) as a function of
physical crack movement, LlC. Just as in plane strain, K Ic , fracture toughness
testing, the procedure for conducting KR tests are complex and tedious. The exper-
imental values of the fracture toughness for all thickness ranges are essential in
evaluating the life expectancy of the cracked structure. The crack tip stress inten-
sity concept and its derivation, an in-depth discussion of fracture toughness for
both surface and through cracks, crack tip plasticity, and plastic zone shape based
on different yield criteria are covered in Chapter 3.
The presence of surface or embedded cracks in a pressure vessel containing
hazardous gases or fluids can produce catastrophic failure when the induced stress
due to applied pressure overcomes the residual strength capability of a vessel dur-
ing its service life, or while it is subjected to proof test. The failure can occur when
the critical stress intensity factor, K, of the surface crack exceeds the surface frac-
ture toughness of the material, K Ie • When the depth of the growing surface crack
becomes equal to the thickness of the vessel prior to catastrophic burst failure, the
surface crack becomes a through crack and the container will leak. The leakage of
a vessel prior to its final failure (known as leak-before-burst failure mode) is
desirable, since it can be a warning signal for needed inspection and consequently
may prevent catastrophic failure by repairing the container. Chapter 3 provides a
complete description of the leak-before-burst failure mode (when K < KIe and
K ;:: Kc) and the failure criteria defining catastrophic fracture when K ;:: KIe •
A new approach, called the Theory of Fracture Mechanics of Ductile Metals
(FMDM), that could satisfactorily correlate the fracture test data with the analysis
was first developed in the 1960s by the authors of this book [11] . The theory
assumes the existence of two distinct plastic zones (Fig. 1.1) at the crack tip that
absorb energy as the crack advances. That is, the localized plastic deformation is
composed of: (1) the local strainability at the crack tip (the region of highly plas-
tic deformation) and (2) the uniform strainability near the crack tip. In other
words, the fracture characteristics of a metal are directly related to the strainabil-
ity of the metal.
In the FMDM theory, the total energy per unit thickness absorbed in plastically
straining the material around the crack tip, Up, was described in terms of plasti-
cally straining the material beyond the ultimate at the crack tip, UF , and below
the ultimate of the material near the crack tip, Uu' where Up = UF + Uu. Thus,
fracture can be characterized by two parameters shown to be determinable from
the uniaxial stress-strain curve.
6 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

+
Applied Load

Uniform Straining,

~'th.:~ad~
Highly Strained Region,
UF' at the Crack Tip

Figure 1.1 The Crack tip plastic zones as viewed by the FMDM theory.

The FMDM theory was later used to develop analytically the plane strain and
plane stress fracture toughness for different isotropic materials [12]. The conven-
tional approach for obtaining the Kc variation with respect to thickness is to con-
duct several tests with cracked plates of different thicknesses and crack lengths
according to ASTM procedures. These tests are costly and time consuming. The
fracture toughness value provided by the FMDM theory depends on a stress-
strain curve that is readily available in MIL-HDBK-5 and other reliable sources or
can be generated in the laboratory for the material under consideration. The
results of the computed fracture toughness generated by the FMDM approach for
several aerospace alloys were shown to be in excellent agreement with the test
data. A detail description pertaining to the FMDM approach is given in Chapter 6.
Failure of structures under working environments seldom occurs by static load-
ing. The majority of unexpected structural failures are due to fluctuating or cyclic
loads where the preexisting flaw will grow undetected in a stable manner and
finally reach its critical length. Generally speaking, fatigue can be defined as an
accumulative failure of a part under repeated or fluctuating loads. While fatigue
failures start from microscopic cracks, the presence of tool or grinding marks left
on the surface of the part make the formation of fatigue cracks easier. Fatigue fail-
ure occurs at calculated applied nominal stresses considerably below the tensile
strength of the materials involved, and although the materials have sufficient duc-
tility, the failure generally shows little or no ductility. The sequence in which
fatigue failure occurs consists of three parts: microscopic crack nucleation/initia-
tion along the slip surfaces (stage I), stable crack growth perpendicular to the
applied load (stage II), and final rupture, where the crack will propagate in an
 1.3 Fatigue and Fracture Mechanics 7

r
Load

A A A Cycle
\TVV, •

Edge of the
Specimen
~-. ~---I.~~
Stage I Stage II Instability

Figure 1.2 Crack initiation (stage I) and Stable propagation (stage II),

unstable manner. Figure 1.2 illustrates the crack initiation and crack growth stages
where the crack is nucleated from a notch (a source of stress concentration).
August Wohler (1819-1914), the railway engineer who became director of the
German Imperial Railway (1847-1889), was the first investigator to address
fatigue failure by conducting cyclic loading tests on full-scale railway axles, as
well as on small-scale specimens [13]. He plotted his test data in terms of applied
stress versus the number of cycles to failure. This type of plot became known as
the S-N diagram or Wohler line. Later, the Wohler S-N diagram was used in other
applications, such as bridges, ships, aircraft, and machinery equipment, that are
also subjected to fluctuating loads.
The S-N curve is a useful tool for assessing the total life of a structural part
when failure occurs under a relatively large number of cycles, and stresses and
strains in the bulk of the material are within the elastic range of the material. The
results of fatigue tests conducted in the laboratory in accordance with the ASTM
E-466 procedure can be presented in tabular form or in the form of S-N curves.
From the collected S-N data, the total life expectancy of the structural part can be
determined and compared with the total number of cycles that the structure is
exposed to during its service life. When the structural part is subjected to several
load environments having cycles of different stress magnitude, the partial damage
due to each loading environment must be calculated for the total life evaluation.
The ratio of the sum of all cycles, In i , to the total cycles to failure, Nfi , are then
compared (In)Nfi = n1lNfI + n21Nf2 ... ) using the classical linear damage the-
ory proposed by Palmgren and Miner [14, 15]. The total life assessment of a
machinery part, using the S-N diagram is an acceptable approach, provided that
8 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

the service conditions of the part under study are parallel to the test conditions
conducted in the laboratory. This is known as the similitude law. That is, the life
of a structural part is the same as the life of a test specimen if both have undergone
the same loading environment (Chapter 2).
If the magnitude of the fluctuating stress is no longer in the elastic range of the
material, significant plastic straining occurs throughout the body, especially in the
highly localized areas at stress concentration regions and the number of cycles to
failure is expected to be relatively low. Low cycle fatigue failure, sometimes
referred to as the strain-life approach (e-N), can no longer be characterized by an
S-N curve. The number of cycles to crack failure in the region of plastic defor-
mation immediately adjacent to the notch can be estimated by a strain-life predic-
tion model using the Neuber relationship [16] and the cyclic stress-strain curve
conducted under strain-controlled conditions. Chapter 2 contains a comprehen-
sive review of conventional fatigue, including high and low cycle fatigue, together
with several example problems.
To prevent fatigue failure of a structural part in a load-varying environment, it
is important to have a good understanding of (1) all loading events that a compo-
nent will experience and the number of times that each event will occur (the
fatigue spectrum); (2) an empirical equation that can relate the fatigue crack
growth rate, da/ dN, with the crack tip stress intensity, flK; (3) the material frac-
ture toughness; and (4) some estimate of the initial flaw size. The functionf(relat-
ing da/ dN with flK) can be obtained as the result of laboratory test data and can
then be utilized to solve crack growth problems where the structural part has
undergone the same loading conditions. Having the above information available
to the analyst, the remaining life cycles (number of cycles for a crack to grow
from its initial length, ai' to its final length, at) can be calculated.
The earliest and most well known relation between the crack growth rate
(da/ dN) and the stress intensity factor range (M) was given by Paris, Gomez,
and Anderson [17, 18]. The Paris "law" (an empirical equation based on the
experimental data obtained through laboratory testing) was later modified to
account for other parameters, such as stress ratio, threshold and critical stress
intensity factors, retardation effect, etc. Probably one of the most useful crack
growth rate equations is given in the NASA/FLAGRO computer program [19],
the current state-of-the-art computer code that is being used in the aerospace
industry for the safe-life analysis of space structural parts. The constant amplitude
crack growth rate relationship (da/dN, flK) for all regions of the crack growth
rate curve (including the threshold stress intensity factor, flK th , and the critical
regions, Kc) are presented in the NASA/FLAGRO computer code. The Newman's
crack closure phenomenon approach [20] is implemented in the computer code
to account for different stress ratio, R, under constant amplitude loading. Chap-
ter 4 provides a detailed description of the crack growth rate concept, including
a discussion on the NASA/FLAGRO computer code, together with several exam-
 1.3 Fatigue and Fracture Mechanics 9

pIe problems pertaining to crack growth rate prediction for aerospace struc-
tural parts.
In constant amplitude loading conditions where there is no interaction among
cycles (baseline fatigue crack growth condition when dR/ dt = 0, and R =
(Tmin/ (Tmax)' the crack growth rate analysis is relatively simple to assess. However,
the actual cyclic loading that most structures experience during their service life is
by no means of constant amplitude and the crack growth rate delay or acceleration
due to load interactions between low-high or high-low cycles must be taken into
consideration (Fig. 1.3). For example, when aircraft wings are subjected to gust
and maneuver loads, the tensile overload or peak load forms a tensile plastic zone
at the crack tip larger than the subsequent constant amplitude cycles (from high
gust load to low load). Upon load release, the overload-induced plastic zone
causes a crack tip compressive stress and delay in the crack growth rate, da/dN.
Use of the constant amplitude crack growth rate equations to express variable
amplitude loading will give conservative results when the number of cycles to
failure of the structural part is of interest. Two well-known mathematical closure
models, called the Wheeler and Willenborg models, based on yield zone [21, 22],
are presently available and are both discussed in Chapter 4.
Many structural parts used in assembling aircraft and space vehicles are either
welded together or mechanically fastened as bolted or riveted joints. A joint can
be viewed as a source of stress concentration that can shorten the life of the
jointed parts unless preventative measures are taken to minimize structural failure.
Fatigue failure of bolts usually occurs at the threaded location where the bolt and
nut are engaged. In some cases (less probable), bolt failure is observed in the
shank to bolt head area. Experimental investigations have shown that thread pro-
cessing done by rolling the threads improves the fatigue life of the part. Rolling

Load

Constant Constant Constant

Amp~tude Load Amplitude Load Amplitude Load

Figure 1.3 Low-high-Iow tensile and compressive overloads.

10 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

threads will create compressive residual stresses at thread roots that help to pre-
vent fatigue failure of bolts in a joint. A tensile preloaded bolt is also considered
to be a desirable practice in reducing the fatigue failure and increasing the life of
a bolt when it is exposed to cyclic loading. Another area of concern when dealing
with a bolted joint is fatigue cracking that may initiate from the hole where the
two plates are bolted together (a source of stress concentration, Fig. 1.4). The ini-
tiated crack from one of the holes in the joint may grow to its critical size due to
the fluctuating load environment and cause the complete separation of the two
structural parts (Chapter 4).
To maintain trouble-free damage tolerant structural hardware, a multidiscipli-
nary process is required that starts in the early stage of design and continues
through manufacturing and into the operational phase. The implementation of this
process requires the execution of a fracture control plan that can control and pre-
vent damage due to the preexisting flaws in the structure. The required fracture
control procedures are based heavily upon the good engineering and manufactur-
ing practices already embedded in the hardware development process. Fracture
control imposes additional engineering and product assurance requirements
needed to ensure the structural integrity of high-risk or fracture critical structures
throughout all phases of the component's lifetime. The design philosophy, mate-
rial selection, analysis, testing, inspection, and manufacturing are all elements of
the fracture control plan that will contribute to building and maintaining a trouble-
free space structure. Section I of Chapter 5 briefly describes the implementation
approach to a fracture control plan for a man-rated space structure. The fracture
control plan methodology used in this chapter is also applicable to other industries
where safety is the primary concern.
For manned aircraft or space vehicles, it is commonly required to assume
cracks preexist in all structural parts. These cracks shall not grow to their critical

p
o = Hole Diameter
t = Plate Thickness
<J = Bearing Stress

Figure 1.4 Crack emanating from a hole where the two structures (1 and 2) are
jointed together.
 References 11

size when subjected to crack growth analysis at a specified load during their usage
period. In general, the material as received from the vendor will contain defects of
small size, such as porosity, scratches, inclusions, microcracks, and machine
marks. These inherent flaws are considerably smaller than the Nondestructive
Inspection (NDI) capability to detect them and will not grow appreciably in ser-
vice. There are numerous NDI methods utilized for flaw detection in structural
components. The most prevalent of these NDI techniques, commonly used in the
detection of flaws in aerospace components, are liquid penetrant, magnetic parti-
cle, eddy current, ultrasonic, and radiography. Section 2 of Chapter 5 provides a
detail description of different NDI techniques that are currently used in the aero-
space and aircraft industries.

References

1. E. R. Parker, "Brittle Behavior of Engineering Structures," John Wiley & Sons, 1957,
pp.253-271.
2. R. P. Reed, J. H. Smith, and B. W. Christ, "The Economic Effects of Fracture in the
United States," SP 647-1, NBS, March 1983, p. 3.
3. A. A. Griffith, "The Phenomena of Rupture and Flow in Solids," Philos. Trans., R.
Soc. Lond., Ser. A., Vol. 221,1920, p. 163.
4. C. E. Inglis, "Stresses in a Plate due to the Presence of Cracks and Sharp Comers,"
Proc. Inst. Naval Architects, March 14, 1913.
5. G. R. Irwin, "Analysis of Stresses and Strains Near the End of a Crack Traversing a
Plate," Trans. ASME, J. Appl. Mech., Vol. 24, 1957, p. 361.
6. G. R. Irwin, "Fracture Dynamics," Fracture of Metals, ASM, 1948, pp. 147-166.
7. E. Orowan, "Fracture and Strength of Solids," Rep. Prog. Physics, Vol. 12, 1949,
pp. 185-232.
8. G. R. Irwin, Fracture Handbuch der Physik, Springer-Verlag, Heidelberg, VI, 1958,
pp.551-590.
9. D. S. Dugdale, "Yielding of Steel Sheets Containing Slits," J. Mech. Phys. Solids, Vol.
8,1960, p. 100.
10. Fracture Toughness Evaluation by R-Curve Method, ASTM STP 527, edited by D. E.
McCabe, American Society for Testing and Materials, 1973.
11. G. E. Bockrath and J. B. Glassco, Fracture Mechanics of Ductile Metals. California
State University, Long Beach, revised 1985.
12. B. Farahmand and G. E. Bockrath, "A Theoretical Approach for Evaluating the Plane
Strain Fracture Toughness of Ductile Metals," Engin. Fract. Mech., Vol. 53, No.6,
March 1996.
13. A. Wohler, "Wohler's Experiments on the Strength of Metals," Engineering, August
23, 1967, p. 160.
14. M. A. Miner, "Cumulative Damage in Fatigue," Trans. ASME, J. Appl. Mech., Vol.
67, September 1945, p. A159.
12 Chap. 1 A Brief Introduction to Fatigue and Fracture Mechanics

15. H. Neuber, Kerbspannungslehre, Springer, (Berlin), 1958: Translation Theory of

Notch Stresses, U.S. Office of Technical Services, 1961.
16. A. Palmgren, "Ball and Roller Bearing Engineering," translated by G. Palmgren and
B. Ruley, SKF Industries, Inc., Philodelphia, 1945, pp. 82-83.
17. P. C. Paris, M. P. Gomez, and W. E. Anderson, "A Rational Analytic Theory of
Fatigue," the Trend Engin., Univ. Wash., 13 (1),1961, pp. 9-14.
18. P. C. Paris, "The Fracture Mechanics Approach to Fatigue" in Syracuse University
Press, 1964, pp. 107-132.
19. Fatigue Crack Growth Computer Program "NASA/FLAGRO," Version 2.0, Docu-
ment Number JSC-22267 A, January 1993.
20. J. C. Newman, Jr. "A Crack Opening Stress Equation for Fatigue Crack Growth," Int.
J. Fract., Vol. 24, No.3, March 1984, pp. R131-R135.
21. O. E. Wheeler, "Spectrum Loading and Crack Growth," J. Basic Engin., 94D, 1972,
pp. 181-183.
22 J. D. Willenborg, R. M. Engle, and H. A. Wood, "A Crack-Growth-Retardation Model
Using an Effective Stress Concept," AFFDL-TM-71-1 FBR, 1971.
Chapter 2

Conventional Fatigue
(High- and Low-Cycle Fatigue)

2.1 Background

It is known that when metals are subjected to fluctuating load, the failure occurs at
a stress level much lower than the fracture stress corresponding to a monotonic
tension load. 1 With the development of the railway in the nineteenth century, the
fatigue failure of railway axles became a problem and much attention was given
to the understanding of the fatigue failure phenomenon. To understand fatigue
failure mechanism induced by repeated loading, full-scale as well as small-scale
fatigue tests were conducted in the laboratory. In 1852, the German railway engi-
neer August Wohler (Director of Imperial Railways in Germany from 1847 to
1889), conducted several constant amplitude fatigue tests on full and small-scale
railway axles. The results of this work [1] were presented in the form of plots of
the failure stress as a function of the number of cycles to failure. This plot is a
useful tool for the total life prediction of a part subjected to constant amplitude
cyclic loading and is known as the Wohler S-N diagram. The Wohler approach
was extended to other areas of concern, such as bridges, ships, and machinery
equipment that undergo repeated loading. The S-N approach is still a useful tool
to assess fatigue failure of many modern structures that are subjected to repeated
loading, where the applied stress is under the elastic limit of the material and the
number of cycles to failure is large. When material failure occurs under a rela-
tively large number of cycles, and stresses and strains are within the elastic range
of the material, the failure mechanism is called high-cycle fatigue. If the magni-

I Monotonic tension load is defined as the application of a single load to failure during the lifetime

of the member.

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts 13
© Chapman & Hall 1997
14 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

tude of the fluctuating stress is no longer in the elastic range of the material, sig-
nificant plastic straining occurs throughout the body, especially in the highly
localized areas at stress concentration sites, and the number of cycles to failure is
expected to be relatively short. This failure mechanism is referred to as low-cycle
fatigue. Low-cycle fatigue failure, sometimes referred to as the strain-controlled
or strain-life (e-N) approach, can no longer be characterized by an S-N curve.
Low-cycle fatigue life is usually associated with a number of cycles to failure
between 100 and 10,000 cycles (depending on material strength and ductility) and
for high-cycle fatigue the number is above 10,000 cycles. The results of low cycle
fatigue tests can be important in the design and failure analysis of industrial hard-
ware when they are subjected to mechanically or thermally induced repeated
strain where failure occurs in relatively short cycles (less than 10,000 cycles). For
example, aircraft components can be subjected to high mechanically induced
cyclic strain during severe gust and maneuvering load environment. Jet engines,
nuclear reactor parts, and pressure vessels are example of thermally induced
cyclic strain.
In the preliminary stage of design, when a new vehicle or structure is under
consideration to perform a given mission, there are several factors that the design-
ers should consider. Trade studies for different design configurations should be
performed and, based on strength and weight considerations among others, an
optimum configuration is selected. The selected design must withstand the envi-
ronment in question without failure. Therefore, a comprehensive failure analysis
(static, dynamic, and fatigue) is necessary to ensure the integrity ofthe structure
under study. There are two primary groups of information that are necessary as an
input for a comprehensive fatigue analysis. One group of information is the data
related to the material behavior when subjected to cyclic loading, such as the lab-
oratory tests for constructing the S-N diagram, modified Goodman or Gerber dia-
grams, and other factors that would help to evaluate the life of the structure. The
laboratory tests must simulate the stress environment that the structural compo-
nent experience. The stress environment could be inertia, thermal, pressure, sonic,
or other environmentally induced stresses. The second group of information is the
determination of the total number of cycles that the structure will undergo
throughout its life (the life cycle or service history of the structure). This is usually
presented in terms of stress environment versus time. Determination of life cycle
is discussed in Section 2.3. With these two groups of information provided to the
engineer, a complete fatigue analysis is possible. If the analysis reveals that the
structure does not have sufficient life, a redesign might be considered. Alterna-
tively, it is often helpful to reexamine the analysis as a whole to see if it is possi-
ble to reduce the degree of conservatism in the assumptions made.
The S-N curve, in addition to estimation of the structural life, is also useful for
evaluating the following cases:
 2.2 Cyclic or Fluctuating Load 15

1. To study the differences in fatigue behavior between two or more materials when
subjected to a given fluctuating load environment
2. For a given alloy, to select the heat treatment conditions that would give the best
fatigue results when subjected to a given fluctuating load environment
3. To study the fatigue behavior of a given material when it is subjected to mechani-
cal working
4. To study the fatigue behavior of a given material with respect to its material orientation
5. To study the effect of stress concentration on the fatigue behavior.

Having the S-N curve and load spectrum available, a third concept known as
cumulative damage theory must be introduced that relates cycles of different
magnitudes to the S-N curve in order to predict the total life of the structural com-
ponent. Section 2.6 describes the classical linear damage theory proposed by
Palmgren and Miner [2, 3] to estimate failure when summation of the cycles' ratio
for each event becomes equal to 1.
The number of cycles to failure in the region of high plastic deformation,
immediately adjacent to the notch, can be estimated by the strain-life prediction
model via the Neuber relationship [4] and cyclic stress-strain curve conducted
under strain-controlled conditions. Sections 2.8 through 2.11 discuss the develop-
ment of the cyclic stress-strain curve by connecting the tips of stable hysteresis
loops obtained through testing. Each hysteresis loop represents one complete
load-unload cycle of a constant strain amplitude. The strain-life (s-N) prediction
model and relating local stress and strains to the far field applied stress and strains
by the Neuber relationship, are discussed in Section 2.12.
In Section 2.2, different types of cyclic loading occurring in real structures are
defined. These fluctuating loads may have constant amplitude or may vary ran-
domly throughout the service life of the structure (as shown in Fig. 2.1).

2.2 Cyclic or Fluctuating Load

Typical forms of cyclic loading that occur in real structures are almost random in
nature and vary in magnitude during their service life (Fig. 2.la). In constant
amplitude cyclic loading, where the amplitude and mean stress stay constant,
there are five stress parameters that can define the loading characteristics (Fig.
2.1b). These characteristics are: cyclic stress amplitude Sa; mean stress SM; max-
imum stress Smax; minimum stress, Smin; and the stress ratio, R. Any two of the
above quantities are sufficient to completely define the cyclic loading.
The mean or steady-state stress (SM) is the average algebraic sum of the max-
imum (Smax) and minimum (Smin) cyclic stresses. The alternating or variable
stress amplitude, Sa' is defined as:
16 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Random Loading

(a)

Constant Amplitude Loading

Sa

1--_....1._....1.._--......- - Time
(b)

Figure 2.1 Illustrations of random and constant amplitude loading

(2.1)

The stress ratio, R, is also an important parameter and is defined as the alge-
braic ratio of the minimum to maximum cyclic stresses. R is widely used to dis-
tinguish different constant amplitude cyclic loading conditions in fatigue analysis.
Another two less commonly used parameters in fatigue loading are the algebraic
ratio of the stress amplitude to the mean stress (A = Sa/SM) and the strain ratio
Re = Bmin / Bmax during a complete cycle. The strain ratio, Re ,is used when deal-
ing with strain-controlled low cycle fatigue where total strain LlB is controlled
throughout the cycle.
The load case presented in Fig. 2.2a is a fully reversed sinusoidal shape stress
cycle and has a stress ratio R = - 1. In this case, the maximum and minimum
cyclic stresses are equal. For the same range, the case of R = - 1 is considered to
be a less damaging cyclic load case when evaluating the fatigue life of a structure.
An example of the R = - 1 load case is a rotating-bending test that uses
four-point loading to apply a constant moment to a rotating cylindrical specimen,
where the maximum and minimum stresses are equal but of opposite sign.
Figure 2.2b presents the cyclic stress case for the stress ratio R = 0, where the
minimum stress is zero. An example of this case is the pressurization and depres-
surization cycle of a pressurized tank, where the maximum and minimum induced
stresses are related to the maximum and minimum pressure. Figure 2.2c is for the
loading condition where both the maximum and minimum cyclic stresses are pos-
itive (0 < R < O. A preloaded bolt subjected to cyclic tensile stresses such that
the maximum and minimum fatigue stresses are positive represents this case. For
 2.2 Cyclic or Fluctuating Load 17

R =-1

III (a) III (b)

III III
~ ~
oJ oJ
(/) (/) Time

o< R < 1

VV (c)
R> 1
Time

V
III
III
...
41 (d)
oJ
(/)
Time

III

...
III
41 Time
Vi
R=~

Figure 2.2 Typical constant amplitude loading cycles

the same stress range value, the case of R > 0 is considered to be the most dam-
aging cyclic load case when evaluating the fatigue life of a structure.
Figure 2.2d shows the loading condition for the case of R > 1. A plate with a
hole that has undergone a sleeve cold expansion or mandrelizing process [5] and
subjected to fluctuating load is an example of this case. The mandrelizing process
creates a massive zone of compressive residual stress field (SM < 0) around the
hole up to the compressive yield strength of the material (see Fig. 2.3). The com-
pressive stresses are a result of applied tensile load (when it is above the tensile

(Jy
Residual Compressive Region

Adaz.~/
Figure 2.3 Residual compressive stress caused by the cold sleeve process
 .
18 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

j~u~ e v~
A
__ .
... Satellite
Cold Side

s-

R> ,
1--------- Time

::l Compressive
~ Hf---jl--I---+--- Sustaining
Stress

Figure 2.4 lIIustration of the case of R > 1 for a satellite orbiting the earth

yield of the material) that causes the surrounding elastic material upon unloading
to induce compressive residual stress in the plastic region local to the hole. The
residual compressive stresses extend fatigue and crack growth lives of the part.
Note that the static load case represents the condition of stress ratio R = 1.
The case of a satellite orbiting around the earth is another example illustrating
the case of R > 1. For a satellite orbiting around the earth, at a certain position in
its orbit (Fig. 2.4) not exposed to the sun, the induced thermal stress is compres-
sive. The induced compressive stress can be thought of as a compressive sustained
load. The mechanical cycling that occurs at the same time (having much smaller
wavelength) together with the sustaining compressive load (SM < 0) will create a
load cycling that would be similar to the case shown in Fig. 2.2d. The case of
R = 00 is a variation of R > 1, where Smax = 0, as shown in Fig. 2.2e.
The random loading case shown in Fig. 2.1 a is complex and it may contain any
combination of the above mentioned cyclic stresses. The random loadings that
involve more than one amplitude and mean stress are more representative of what
a structure may be subjected to. An example of this case is the loading environ-
ment that an aircraft or space structure will experience during its lifetime. Table
2.1 shows the stress ratio and the corresponding maximum and minimum stresses,
together with an example that represents the pertinent loading condition.

2.3 Fatigue Spectrum

Load, stress, or fatigue spectrum is the engineering definition of the fatigue envi-
ronment that a component experiences throughout its design life and is defined
by the load (or stress) amplitude versus the number of cycles. The fatigue spec-
 2.3 Fatigue Spectrum 19

Table 2.1 Stress ratio and representative load description

Stress ratio R Loading case Load description

R =-1 Smax is positive, Smin is negative Rotating shaft without the overload
with SM = 0 (fully reversed)
R=l Static loading Static loading
R=O Smax is positive and Smin = 0 Pressurization and depressurization
ofa tank
O<R<1 Smax and Smin are both positive Preloaded bolt subjected to fully
(with positive SM) reversed load
R>l Smax and Smin are both negative Mandrelized hole subjected to
(with negative SM) fluctuating load
R=oo Smax is equal to zero and Smin Mandrelized hole under
is negative fluctuating load with Smax = 0

trum should be available early in the engineering phase to permit initiation of

design and analysis activities. This section provides an overview of how a fatigue
spectrum is developed.
The most important step in developing a fatigue spectrum is the definition of
all loading events that a component will experience and the number of times that
each event will occur. Completion of this step requires a comprehensive under-
standing of the component under consideration, its role in the system being
designed, and the design requirements. For example, an aircraft may be required
to perform 1000 missions during its service life. For an aircraft, each mission pro-
file may be divided into distinct events, such as taxi to takeoff, ascent, cruise,
descent, landing, and taxi after landing (Fig. 2.5a). The fatigue spectrum for each
event may be a function of several variables. For example, the wing load magni-
tude and its cyclic behavior during ascent depends on speed, weight, gust factor,
etc. Figure 2.5b shows a simplified version of a complex load spectrum that the
wings of a transport aircraft are exposed to during each mission.
Contractual or top level system requirements usually specify operations and/or
conditions to which the system is exposed and the frequency of this exposure. It is
usually the analyst's responsibility to identify the relevant system level require-
ments for each component and to establish the variation of load or stress versus
time (time history). If the component responds statically to a loading event, the
load/stress time history can be established with relative ease. For example, con-
sider a spur gear that is designed to actuate the wing flaps of an aircraft during
ascent and would undergo a fixed number of complete revolutions under a con-
stant torque. In this case, the stress time history for a typical gear tooth during
each ascent event can be calculated by using the equations of static equilibrium.
In many cases where flexible structures and rapidly changing loads are
involved, however, the component/system responds dynamically to the loading
20 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Altitude

"II)
II)
Co
en
cD
"'p....
::I

;(

Time ----.

(b)

Figure 2.5 (a) Mission profile and (b) simplified load spectrum for an aircraft

environment. In these cases, establishing the time history would require dynamic
analysis and the cyclic behavior of the time history would be a function of both
the load variation and the dynamics of the structure. For example, in the absence
of a shock absorber, the landing gear of an aircraft would experience large load
vibrations during touchdown (Fig. 2.6).
Once the time history for each event is established using a cycle counting pro-
cedure, each time history is converted to a fatigue spectrum for that event con-
sisting of load range/mean range versus number of cycles (range is defined as the
algebraic difference between successive valley and peak loads; see Fig. 2.7). A
variety of cycle counting procedures are available. One of the most commonly
recognized and widely used cycle counting approaches is the rain flow method.
Several variations on the rain flow method for specialized applications have been
discussed in the literature [6, 7]. Other cycle counting methods are peak counting,
level crossing, and range-pair counting. In all of the aforementioned cases, the
irregular load sequence can be converted to a sum of cycles, N;, with different
stress amplitudes, Si' to assess the total damage induced on a given part. It is
important to note that the prediction of fatigue life by one technique (for example,
 2.3 Fatigue Spectrum 21

Touch
/ Time

Down

Figure 2.6 An undamped landing gear force time history during touch down

Load

Range

t-+-lr--+-+-+-~~-+--+-'""""!""'+--- Time

Figure 2.7 A typical fatigue cycle with fatigue loading parameters (range, peak, valley)

the level crossing method) may produce results that differ by an order of magni-
tude from the others.
The following 6 steps are the step-by-step procedures for using a standard rain
flow method technique that are directly extracted from the ASTM E-1049. Figure
2.8 is used here to clarify the procedures defined in steps I through 6. Note that
this procedure should be preceded by identifying the local peaks and valleys in the
time history. Let X denote the range under consideration; Y, the previous range
(see Fig. 2.7 for the definition of range) adjacent to X; and S, the starting point in
the history. The starting point, S, in the load history shown in Fig. 2.8a is A and
the ranges X = 1 B - eland Y = 1 A - B I·
22 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

LOAD
5
4
3
2
1
O~~~~~~-+~~~~--
-1
-2
-3
-4
(a)
-5

LOAD
5
4
3
2
1
O ......~H-f-+-~J,..,.f-t-......,H--t--
-1
-2
-3
-4
-5 (b)

LOAD
5
4
3
2 B
1
O~~~~~~~~~~~~---
-1
-2
-3 A
-4
-5 (e)

Figure 2.8 Rain flow cycle counting technique

1. Read the next peak or valley. If out of data, go to Step 6.

2. If there are less than three points, go to Step 1. Fonn ranges X and Y using the three
most recent peaks and valleys that have not been discarded.
3. Compare the absolute values of ranges X and Y.
(a) If X < Y, go to Step 1.
(b) If X :2:: Y, go to Step 4.
4. If range Y contains the starting point S, go to Step 5; otherwise, count range Y as
one cycle; discard the peak and valley of Y; and go to Step 2.
 2.3 Fatigue Spectrum 23

LOAD
5
4
3
2 B
1
0
-1
-2
-3 A
-4
-5 (d)

LOAD
5
4
3
2 B
1
0
-1
-2
-3 -:
A ::. C
-4 (e)
-5

LOAD
5 - D
4 ~H
3 F
2 B
1
0
=- TIME
-1
-2 E-:'" - I
-3 A
-4 C -=-G
(I)
-5

Figure 2.8 (cont.)

5. Count range Yas one-half cycle; discard the first point (peak or valley) in range Y;
move the starting point to the second point in range Y; and go to Step 2.
6. Count each range that has not been previously counted as one-half cycle.

Example 2.1
By following rain flow cycle count process described by Steps 1 through 6, cal-
culate the fatigue spectrum for the load time history shown in Fig. 2.8.
24 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Solution
Step 1. The starting point is S = A as shown in Fig. 2.8a and the ranges are X =
IB - cl and Y = IA - BI, where X> Y (Y contains the starting point S = A).
Count Y = IA - B I as one-half cycle. The new starting point is S = B. Note that
in Fig. 2.8b the discarded quantity Y = IA - B I is shown by a dashed line which
indicates that it was counted as one-half cycle.
Step 2. The new starting point is S = B, and the ranges X = Ie - I
D and
Y= IB - I,
c where X> Y, (see Fig. 2.8b). Count Y = IB - I c as one-half cycle.
In Fig. 2.8c, the discarded quantity Y = IB - c I is shown by a dashed line.
Step 3. The starting point is S = C. The new ranges are X = ID - E I and
Y= Ie - ID where X < Y, go to the next range.
Step 4. The next ranges are X = IE - FI I
and Y = D - EI'
where X < Y, go
to the next range.
Step 5. The next ranges are X = IF - GI and Y =IE - FI, where X > Y. In
this case, count Y = IE - F E
I as one full cycle, as shown in Fig. 2.8d. Points and
F are discarded and they are shown by the dashed area illustrated in Fig. 2.8d.
Step 6. The new starting point is the same as in step 3, S = C, and X = ID -
I
G and Y = Ie - I' D where X > Y. Count Y = Ie - I D as one-half cycle. The
I
new starting point is S = D and the next ranges are X = G - H and Y = D - I I
I I I
G where X < Y. Count Y = D - G as one-half cycle (Fig. 2.8e).
Step 7. The new starting point is S = G and X = IH - 11
and Y = IG - HI,
I I
where X < Y. Count G - H as one half cycle and H - I 11as one-half cycle (end
of counting), Fig. 2.8f. The resulting fatigue load spectrum is shown in Table 2.2.
In many engineering applications, the spectrum for a typical (or in some cases the
worst) loading event is calculated and the off nominal events are developed by
applying statistical distributions to the range and mean portion of the calculated
nominal spectrum. Such an approach results in the redistribution of cycles among
load ranges and means while maintaining the total number of cycles the same.
The cycle count for each individual event is multiplied by the number of
expected occurrences of that event in the life cycle of the component and is added
to all other expected events to form the fatigue spectrum for the component.

Example 2.2
(a) Use the rain flow technique to calculate the fatigue spectrum for the load time
history shown in Fig. 2.9 and the associated Table 2.3. (b) Calculate the fatigue
spectrum for a component that would experience 100 events characterized by Fig.
2.8 and 200 events characterized by Fig. 2.9 during its life cycle. (c) Calculate the
fatigue spectrum for a component that would experience 200 events that have a
uniform distribution between 2 and 5 for its positive peak load and is character-
ized by a typical time history shown in Fig. 2.9.
 2.3 Fatigue Spectrum 25

Table 2.2 Fatigue spectrum for time history of Example 2.1

Mean load
Load
range -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0

10
9 0.5
(DG)
8 0.5 0.5
(GH) (CD)
7
6 0.5
(HI)
5
4 0.5 1.0
(BC) (EF)
3 0.5
(AB)
2

Solution
Part a. Prior to applying the rain flow procedure the local peaks and valleys are
identified and labeled A through S in Table 2.3. The rain flow procedure is then
applied to the peaks and valleys. A summary of each step is given here:
1. S = A, X = IB - C I, Y = IA - B I, X> Y, Y includes S :. eliminate A,
AB = .5 cycle.
2. S = B, add D, X = Ie - I, I,
D Y = IB - C X < Y.
3. S = B, addE, X = ID - EI, Y = Ic - DI,x > Y, Yexcludes S:. eliminate CD,
CD = 1 cycle.
4. S = B, addF, X = IE - FI, Y = IB - EI, X> Y, Y includes S :. eliminate B,
BE = .5 cycle.
5. S=E,addG,X=IF-GI,Y=IE-FI,x<y.
6. S = E, add H, X = IG - HI, Y = IF - G I ' X < Y.
7. S = E, add!, X = IH - II, Y = IG - HI, X> Y, Yexcludes S:. eliminate GH,
GH = 1 cycle.
8. S=E, X= IF-/I,Y= IE-FI,x<y.
9. S=E,addJ, x=I/-JI,y=IF-/I,x<y.
10. S=E,addK.X= IJ-KI,Y= I/-JI,x<y.
11. S = E, addL, X = IK - LI, Y = IJ - KI,x < Y.
12. S = E, addM,X = IL - MI, Y = IK - L I, X> Y, Yexcludes S:. eliminate KL,
KL = 1 cycle.
26 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Load
F H
/",;"
i G \ J
i '\ ',. L
i .,.' ... '.'.,
ilK \
1 l \R
." \ N '0
A •B I"~" " " , . , I~.. S
t'
o "'!'':'~ III 1111111 11jl11l111 111111111111,1111 111111 :t+tt;~ i<! I 11111111 !1I11 j'f
; " .•. - ... p •••.•'
\
j' Time
\. ;\ I \ .'
.1 • D M \ / Q

·2 C \i 0

.3 E

Figure 2.9 Load time history for the Example Problem 2.2

Table 2.3 Component load time history for Example 2.2

Time Force Time Force Time Force

(A) 0.04 0.00 1.04 0.85 2.04 0.03

0.08 0.00 1.08 1.20 2.08 -0.17
0.12 0.00 1.12 1.64 2.12 -0.29
0.16 0.00 1.16 2.04 (M) 2.16 -0.41
0.20 0.01 1.20 2.41 2.20 -0.30
0.24 0.11 1.24 2.86 2.24 0.02
(B) 0.28 0.17 1.28 3.26 (N) 2.28 0.22
0.32 -0.19 (F) 1.32 3.38 2.32 0.09
0.36 -0.94 1.36 3.19 2.36 -0.15
(C) 0.40 -1.44 (G) 1.40 3.14 2.40 -0.34
0.44 -1.32 (H) 1.44 3.34 2.44 -0.66
(D) 0.48 -1.08 1.48 3.18 2.48 -1.10
0.52 -1.40 1.52 2.40 (0) 2.52 -1.41
0.56 -2.17 (I) 1.56 1.75 2.56 -1.28
(E) 0.60 -2.45 1.60 1.84 2.60 -0.82
0.64 -1.83 (1) 1.64 2.18 2.64 -0.45
0.68 -0.98 1.68 2.17 (P) 2.68 -0.41
0.72 -0.63 1.72 1.91 (Q) 2.72 -0.42
7.76 -0.58 (K) 1.76 1.82 2.76 -0.10
0.80 -0.29 (L) 1.80 1.91 2.80 0.39
0.84 0.12 1.84 1.87 (R) 2.84 0.62
0.88 0.32 1.88 1.61 2.88 0.53
0.92 0.33 1.92 1.20 2.92 0.33
0.96 0.45 1.96 0.77 2.96 0.14
1.00 0.63 2.00 0.35 (S) 3.00 0.01

I local peaks and valleys are Labeled by A through S

 2.3 Fatigue Spectrum 27

13. S = E, x = IJ - MI, Y = II - JI, X> Y, Y excludes S .". eliminate /J,

/J = 1 cycle.
14. S=E, X=IF-MI,Y=IE-FI,x<y.
15. S=E,addN,X=IM-NI,Y=IF-MI,x<y.
16. S = E, add 0, X = IN - 0 I' Y = 1M - N I,X > Y, Yexcludes S .". eliminate MN,
MN = 1 cycle.
17. S=E, X=IF-OI,Y=IE-FI,x<y.
18. S = E, addP, X = 10 - pI, Y = IF - ol,x < Y.
I
19. S = E, add Q, X = P - Q I' Y = I0 - pI, X < Y.
20. S = E, add R, X = IQ - R I, Y = IP - Q I, X> Y, Y excludes S .". eliminate PQ,
PQ = 1 cycle.
21. S=E, X=IO-RI,Y=IF-ol,x<y.
22. S = E, add S, X = R - S I I' Y = 10 - R I' X < Y.
23. Count the remaining peaks and valleys (EF = .5 cycles, FO = .5 cycles, OR = .5
cycles, RS = .5 cycles).

Results are summarized in Table 2.4.

Part h. The solution to part b is simply obtained by multiplying the cycle
counts from the spectrum in Table 2.2 by 100 and adding it to the spectrum of
Table 2.4 multiplied by 200; the result is shown in Table 2.5.

Part c. The detailed solution is discussed for segment BE of the time history.
The BE segment represents [200 events X 0.5 cycles/event] = 100 cycles with a

Table 2.4 Fatigue spectrum for part a of Example 2.2

Mean load
Load
range -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0

10
9
8
7
6 0.5 [EF]
5 0.5
[FO]
4
3 0.5 0.5 [OR]
[BE]
2
1 [CD] 3 2 1 [OR]
[AB, MN, PQ, RS] [KL, IJ]
28 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Table 2.5 Fatigue spectrum for part b of Example 2.2

Mean load
Load
range -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0

10
9 50
8 50 50
7
6 100 50
5 100
4 50 100
3 100 150
2
200 600 400 200

range value of [0.17 + 2.45] = 2.62 and mean value of [017 ; 2.45] - 1.14. Since
the maximum positive peak (point F at 3.38) has unifonn distribution between 2
and 5, the range for segment BE would have a unifonn distribution between
P·6i3~ 2] 1.55 and [26i3~ 5] 3.87. The mean value would have a unifonn distribution
between [-l.j~: 2] - 0.67 and [-l.j~8X 5] - 1.68. Therefore, [::~~ = ~~n = 82%
of the cycles fall in the -1.0 mean category and the remaining 18% are in the - 2.0
mean category. For the range, however, [;:~ =:;;] = 19.4% are in the 2 load
range which translates to [100 X 0.194] = 19 cycles in the "2 range and -1 mean"

in "3 range and -1 mean" category. The remaining m;

category. [;~~ - i~~] = 43.1 % are in the 3 load range which translates to 43 cycles
=~:~] = 37.5% are in the
4 load range. Since only the lower 82% of the cycles fall in the -1 mean load cat-
egory, however, there will be [100 X (82% - 19.4% - 43.1 %)] = 19 cycles in the
"4 load range and -2 mean load" category. The remaining [100 X 18%] = 18
cycles will be in the "5 load range and -2 mean load" category. A similar proce-
dure is followed in analyzing all cycles identified in part a, the complete solution is
shown in Table 2.6.

2.3.1 Load Spectrum for Space Structures Using the

Space Transportation System (STS) or Shuttle
To evaluate the fatigue life of a space structure it is important to consider the total
number of cycles associated with each event that the part is exposed to during its
service life. The initial step is to break the load history associated with each event
into its constituent events, and then the damage effect of each of those events are
added together. For example, the cyclic load environments that a space structure
will experience during its mission can be broken down as follows (see Fig. 2.10):
 Table 2.6 Fatigue spectrum for part c of Example 2.2

Mean load
Load
range -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0

10
9 12 [EF]
8 2 [FO] 19 [EF]
7 24 [FO] 5 [EF] 14 [EF]
6 24 [FO] 19 [EF]
5 24 [FO] 19 [EF]
4 18 [BE] 19 [BE] 11 [EF]
+
24 [FO]
3 43 [BE] 32 [OR]
+
4 [FO] +
24 [OR]
2 19 [BE] 44 [OR]
65 [CD] 135 [CD 100 [AB 39 [II] + 40 [GH] 70 [GH] 70 [GH] 20 [G
]+ ]+ 48 [KL] + + H]
62 [PQ] 200 [MN 115 [II] 46 [II] +
] + 138 [P + 31 [KL]
Q]+ 121 [KL]
100 [RS
]
I\)
(0
30 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Tr3nsportation

Launch
Space Components
in STS

Figure 2.10 The load environment that space componenets experience during their
service life

• Prelaunch cycles (acceptance, proof testing, etc.)

• Transportation (including ferry flight) cycles prior to flight
• Flight (liftoff/ascent) cycles, including abort landing
• On-orbit cycles due to unloading and any other on-orbit activities
• Thennal cycles
• Descentllanding cycles
• Postlanding event cycles

The Ferry flight is referred to the transportation of the shuttle structure (by the
Boeing 747) from one location to its final distination prior to or after its mission.
When the structural hardware is assembled in the shuttle, while it is being trans-
ported to another location for launch, the flight load cycles (as the result of trans-
portation) must be included in the total life analysis of parts.
A brief description of flight environment and the corresponding number of
cycles are provided in Section 2.3.1.1.
 2.3 Fatigue Spectrum 31

2.3.1.1 Flight Cycles

Components of space structures experience cyclic loadings of different stress
levels as a result of launch and landing of the space shuttle. The payload ran-
dom loading spectrum for these events is shown in Table 2.7 and is called the
Goddard load spectrum because it was developed at Goddard Space Flight Cen-
ter (GSFC) [8]. There are a total of 13 load steps associated with the Goddard
load spectrum and the percent of limit load for each stress level, Si' associated
with the number of cycles, N i , is given in Table 2.7. The Goddard spectrum has
been developed for the main load carrying payload structure in the shuttle
orbiter payload bay. Moreover, the payload structural parts major natural fre-
quency must be below 50 Hz. If the payload major frequency is above 50 Hz,
the number of cycles in the Goddard spectrum (see Table 2.7) should be multi-
plied by the factors shown in Table 2.8 [9].
When the load environment is completely evaluated in terms of the number of
cycles, N i , and the associated load amplitude, Si' the total fatigue life of the struc-
tural part can be predicted. However, the S-N curve representing the load envi-
ronment for the material must be well established. Section 2.4 provides a
comprehensive overview of the constant load amplitude S-N diagram which can
be established by plotting test data from the load controlled method.

Table 2.7 Launch and landing load spectrum (called Goddard fatigue spectra),
extracted from [9]

Cycles (NJ/flight Cyclic stress (% limit stress)

Load step
number Launch Landing Total Minimum Maximum

I I 2 -100 100
2 3 1 4 -90 90
3 5 3 8 -80 80
4 12 3 15 -70 70
5 46 3 49 -60 60
6 78 3 81 -50 50
7 165 13 178 -40 40
8 493 148 641 -30 30
9 2229 891 3120 -20 20
10 2132 1273 3405 -10 10
11 2920 2099 5019 -7 7
12 22272 6581 28853 -5 5
13 82954 8701 91655 -3 3
32 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Table 2.8 Major frequencies and the

multiplication factor for the shuttle payload
structural parts

Major Multiplication
frequency factor

0-50 Hz 1
50-100 Hz 2
100-200 Hz 4
200-300 Hz 6

2.4 The S'rN Diagram

In evaluating the number of cycles to failure for a given structure subjected to in-
service fluctuating loads, fatigue test data representing the load environment must
be available. The concept of similitude states that the life of a structural part is
the same as the life of a test specimen if both have undergone the same nominal
stress. Figure 2.11 shows that the service life of a bridge part exposed to a fluctu-
ating load environment can be evaluated by conducting a laboratory fatigue test
that simulates the same environment.
Fatigue test data can be provided to the analyst in tabular form or in the form of
an S-N diagram. The S-N (stress-life) diagram is a plot of stress amplitude, stress
range, or the maximum cyclic stresses, S (selected as the controlled or indepen-
dent variable), versus the number of cycles to failure, N (the dependent variable).
There are two methods of plotting S-N curves: (1) The S-N diagram is plotted as
the actual stress, S, versus the logarithmic scale of cycles, N (semilogarithmic

~~~t~
O(nominal)

Figure 2.11 The concept of similitude: the life of a bridge part can be determined by
obtaining fatigue test data on a Laboratory specimen subjected to the same nominal
stress.
 2.4 The S-N Diagram 33

plotting). (2) Both S and N are plotted in the form of a log-log plot of S versus N
(logarithmic plot). For most materials, the logarithmic plot of the S-N curve is
approximated by a straight line. For some alloys, including the ferrous metals, the
logarithmic plotting method will generate an additional straight line (horizontal)
to account for the endurance limit (also called the fatigue limit) (Fig. 2.12). The
semilogarithmic method is the most widely used in engineering applications.
Several types of machines and specimens are available to develop fatigue data.
Test data are generated under axial loading, plate bending, rotating bending, and
torsion tests. For aerospace use, none of these methods, except for the axial load-
ing tests according to ASTM E-466 (Conducting Constant Amplitude Axial
Fatigue Tests of Metallic Materials) and ASTM E-606 (Standard Practice for
Strain-Controlled Fatigue Testing), are now conducted for developing fatigue
data. A brief description of the constant amplitude axial fatigue tests is provided
in Section 2.4.1.1. The usual laboratory procedure for determining the S-N curve
for a given material is to use about 18 specimens [10]; see Fig. 2.14 for the spec-
imen geometry. The results of testing are expected to sometimes have wide scat-
ter, so that statistical analysis (the best fit curve method by regression analysis
discussed in ASTM E-739 ) is needed to establish a meaningful S-N diagram [11].
The first few specimens are used where the applied cyclic stress is equal to about
70% of the static tensile strength of the material (the induced cyclic stress magni-
tude in the test specimen must be below elastic limit). For this region of the S-N
curve, the number of cycles to failure is expected to be about 10 3 to 10 4 cycles.
The remainder of the test specimens are utilized for other regions of the S-N
curve, where the test stress decreases and the number of cycles to failure is
expected to increase from 10 5 to 10 7 cycles. Usually six to eight stress levels are
selected and for each stress level two to three test specimens are used to develop
an S-N curve. When the number of cycles to failure exceeds 10 7 without failure
(runout condition), the fatigue limit has been reached and the S-N curve becomes

Ferrous Alloys

......
II)

vi
en
....~
......
II)

C)
o
...J
Endurance Limit ~

Log (Number of Cycles to Failure, N)

Figure 2.12 S-N curve for ferrous alloys

34 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

asymptotic to a horizontal line. The stress associated with this limit is called the
endurance limit (or fatigue limit), see Fig. 2.l2. The endurance limit is an impor-
tant parameter when designing a part to have infinite life.
When prior information regarding the shape of the S-N curve is available, it is
unnecessary to select several stress levels to trace out the S-N curve point by
point. Not more than four or five test specimens are needed to establish the shape
of the S-N curve.

2.4.1 Factors Influencing the Endurance Limit

For most materials, the endurance limit or fatigue limit is not a constant and varies
with the stress ratio, R. For a given material, the endurance limit can be influenced
by the type of cyclic loading. Experimental data obtained in the laboratory show
that the endurance limit of a material tested in uniaxial loading is lower than the
endurance limit tested in reverse bending, provided that the two loading cases are
subjected to the same stress ratio, R. Under axial loading, the stresses are uniform
throughout the part, as compared to a nonuniform stress distribution where bend-
ing load is applied.
Other factors affecting the endurance limit are: degree of surface finish, heat
treatment, stress concentration, and corrosive environment. Therefore, it is
expected that the endurance limit will have a wide range of values depending on
the conditions described above. Consider the case of high strength steel with little
ductility exposed to a corrosive environment and containing stress concentration.
In this case, it is possible to have an endurance limit value as low as 15% of its
ultimate tensile strength. On the other hand, consider the case of stainless steel in
the annealed condition and subjected to a noncorrosive environment. The
endurance limit can be as high as 70% of its ultimate tensile strength.
Experimental data have shown that certain alloys, such as ferrous material, ex-
hibit a clear fatigue limit. For the maximum applied stress below this limit, failure
will not occur and, therefore, the material has infinite life. Typical fatigue data
(maximum stress, S, versus N) for 4130 alloy steel is plotted in the semilog plot
shown in Fig. 2.13a [12]. The S-N diagram for this alloy exhibits an endurance lim-
it of 43.3 ksi with stress ratio of R = -1. In comparison with ferrous alloys, alu-
minum alloys do not show a well-defined endurance limit. Fig. 2.13b shows the S-N
curve for 2024-T4 aluminum alloy where the fatigue limit is not well defined [12].
The S-N curves shown in Figs. 2.l3a and 2.l3b can be represented by an empir-
ical relationship that describes the variation of stress amplitude, S, and fatigue
cycles, N. See Section 2.4.2 for a more detailed description related to this topic.

2.4.1.1 Brief Description of Constant Amplitude Axial Fatigue Tests

In determining the fatigue life of a metallic part subjected to constant amplitude
axial cyclic load (where the strains are predominantly elastic and the number of
 2.4 The $-N Diagram 35

rr
II:
Iii 110
!! lilT! if.-I J I fPI :'r-L-I
!! lIill l'H. .1'
j 0
.1 I I
- 111\
I.i. ljl,· ir1i,i, ,I lj ,! l ro-~ . ! Ii! 1 III - 111~!l!
III - , ,lilil
I I
II i

I i IrI.llIii
40

'', ;"I
I II',
!!!i
I
!' !" upon nel section
I NO~~~lre.ses b~sJ II
are
20 :::: !. :: 11ni I ,
I "' ._

l~. II r! III I I 11'1 ",I i II I '!III'" I! ! III i :, 1.1

o , !:! I! I, i ! I ! !d I i i ! [iii I! II LII ·
i
i j i

106 108
FATIGUE LIFE. CYCLES

Figure 2.13a The S-N diagram for 4340 alloy steel (from MIL-HDBK-5E)

SO
NOle: Stresses are based AUJ,I ZllZ40T4 1CT,'oO
SIIESS RAno
10 -1.00
;;;
....0 -G. 50
+ -G.4'
'" 0.00

••
)(
70 CI.06
<Ii

• 0.43
fJ)
w
a:: SO
.... 0.50
'"
z=>
z SO
;:c
0«
Z
40

30

20

,04 105 '0' 107 108

FATIGUE LIFE, CYCLES

Figure 2.13b The S-N diagram for 2024-T4 aluminum alloy (from MIL-HDBK-5E)
36 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

cycles to failure is expected to be large), laboratory fatigue tests in accordance

with the ASTM E-466 Standard must be performed. Small-scale laboratory
fatigue tests, closely resembling or simulating the load environment, are often nec-
essary in order to replace the costly and time consuming full-scale fatigue tests.
The number of cycles to failure is defined as either complete separation of the
test specimen or when a crack of some specified dimension has been reached. The
results of fatigue tests can be presented in tabular form or in the form of S-N
curves. These data are useful for design of machinery parts where the service con-
ditions are parallel to the test conditions. Four types of specimens are recom-
mended by the ASTM E-466 Standard and they are shown in Fig. 2.14. Attention
should be given to ensure that the size of the test specimen is compatible with the
capability of the loading machine employed to generate the S-N data. These spec-
imens are designed such that failure as a result of fluctuating load is expected to
occur in the middle region, where the stresses are uniform and cross-sectional area
is reduced. The end threads of the test specimen which are used to fasten the spec-
imen to the testing machine must be designed such that failure occurs in the
reduced region and not in the threads area. Specimens with circular cross-sections
are of two types: (1) a type in which the test specimen has tangentially blending
fillets between the reduced section and the ends (Fig. 2.14a) and (2) a type where
the test specimen has a continuous radius between ends (Figs. 2.l4b). Circular
specimens should have a minimum of 0.2 in. and a maximum of 1.0 in. diameter

Circular Cross Section

t IR
£I-+--~--r-~---l c
t ~R
{}i---~---r~---l d
Rectangular Cross-Section

Figure 2.14 Four types of fatigue specimens recommended by the ASTM E-466
 2.4 The S-N Diagram 37

in the reduced section. A minimum and maximum grip cross sectional area of 1.5
to 4 times the reduced cross-sectional area is recommended by the ASTM-E466.
Furthermore, the ratio of the length of the reduced section, L, to its diameter, D,
should be greater than 3. To minimize the stress concentration between the grip
and the reduced section, the fillet radius, R, of the blended section should be at
least eight times the test section. The fillet radius stress concentration criterion
mentioned above is also applicable to test specimens with rectangular cross-sec-
tions as shown in Figs. 2.14c and d.
Test specimens with rectangular cross-section must have a minimum and max-
imum reduced area of 0.03 and 1.0 in. 2, respectively. The width-to-thickness ratio,
Wit, in the reduced section should be between 2 and 6. For example, a test speci-
men with rectangular cross-section, in which the width and thickness dimensions
are 0.5 in. and 0.15 in., respectively, should have a cross-sectional length L = 1.5
in., fillet radius R = 4.0 in., and Wgrip = 0.75 in. For a circular cross-sectional test
specimen, with D = 0.25 in., the cross-sectional length L = 1.0 in., R = 2.0 in.,
and Dgrip = 0.75 in.
When calculating the applied load, the width and the thickness dimensions
from which the area is calculated must be measured to the nearest 0.001 in. For
test specimens with dimensions smaller than 0.2 in., the dimensional measure-
ments should be to the nearest 0.0005 in. (ASTM-E466). The same dimensional
requirements are also applicable to the circular test specimens. Consider a rectan-
gular test specimen that is prepared from 2219-T851 aluminum alloy plate (ulti-
mate and yield stress of 62,000 and 45,000 psi, respectively) and is axially loaded.
It is necessary to obtain the number of cycles to failure for a stress equal to 60%
of the material ultimate allowable (37,200 psi) with a stress ratio of R = O. The
calculated applied load, based on a measured thickness, t, of 0.1 in. and width, W,
of 0.5 in. is 1860 lb. If the dimensional measurements of thickness and width dif-
fer by 0.004 and 0.002 in. (not to the nearest 0.001 and 0.0005 in. as mentioned
above) the calculated applied load will be 1778 lb. This would result in a pre-
dicted fatigue life considerably higher or lower than the expected value.
Axially loaded test specimens must be free from any induced rotation or bend-
ing stress introduced as a result of misalignment or rotation of the grips when
mounting the specimen. To eliminate misalignment and to ensure axiality of the
applied load, strain gages can be used to measure the bending strains. The value of
bending strain should be compared with the axially measured strain due to applied
load. The calculated percent bending strain should not exceed 5% of total strain
measured by the strain gages (ASTM-E-466). Care must be taken to ensure that
the installation of strain gages does not cause damage to the surface (resulting in
a stress riser) of the test specimen.
Upon completion of constant amplitude axial fatigue testing, a description of
the parameters that can significantly influence the test results must be provided to
the user of the data. The most important parameters are:
38 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Material
Grade designation, heat number, melting practice, last mechanical and heat treatment,
chemical composition, tensile and yield strength, elongation and reduction of area
Fatigue specimen
Shape size, stress concentration factor, preparation, forming, heat treatment
Fatigue tests
Fatigue testing machine, type of test, frequency, dynamic load verification, dynamic
load mounting procedures, failure criterion, number of specimens tested, laboratory
temperature and relative humidity
Fatigue data
Fatigue test data should include a table that contains dynamic stresses, fatigue life, test
sequence, specimen mark, and remarks related to the nature of failure.

2.4.2 Empirical Representation of the S-N Curve

In some cases, fatigue data are presented in the form of an empirical equation that
describes the variation between the stress amplitude, S, and fatigue cycles, N, for
different stress ratio, R. For example, in Fig. 2.15 the equation that describes the
variation of S versus N for 4130 alloy steel with stress ratios of -1 to 0.02 is
given [12] as:

10gN = 9.65 - 2.85 log [Smax(1 - R)0.41 - 61.3] (2.2)

For stress ratios of R = -1 and R = 0, equation (2.2) becomes:

10gN = 9.27 - 3.57 log (Smax - 43.3) (2.3)

10gN = 9.27 - 2.85 log (Smax - 61.3) (2.4)

The relationship described by Eq. (2.2) is very useful for the cases where the
test data for the intermediate values of R are not available. However, it is impor-

......
(/)

ui
UI
Q)
L..
+-'
(/)
.......
Cl
o
...J

Log (Number of Cycles to Failure, N)

Figure 2.15 Idealized S-N curve shown in a log-log plot

 2.4 The S-N Diagram 39

tant to note that the use of Eq. (2.2) is not recommended in making life predictions
for the conditions beyond the range given by the test data. Finally, Eq. (2.2) is a
convenient way of evaluating the number of cycles to failure when it is pro-
grammed in the computer for cases where the load environment contains stress
amplitudes of different magnitude (see example 2.5).
The fatigue life equation described above is formulated by plotting the stress
data, S, versus the number of cycles to failure, N, in log-log form. For an idealized
case, shown in Fig. 2.15, the slope of a straight line can be written by establishing
the coordinates of two points along the line:

(2.5)

where b is the slope of the line (after Basquin [13], who proposed the idea) and
(Nm' Sm) and (Nn' Sn) are two points taken along the line. By knowing the slope
of the line, b, and any other given point (N, S) along the line, the fatigue cycle, N,
associated with any other given stress amplitude, S, can be calculated. For exam-
ple, let us assume that the endurance limit, Sen' is taken to be 10 7 for alternating
stress equal to one third of the ultimate of the material (0.33Sul )' By knowing the
slope of the line, b, any other number of cycles to failure associated with a given
stress amplitude, S, can be written as:

Or, in general, one can write:

(2.6)

where (NA'SA) is the available point and Sa is the given stress amplitude associ-
ated with the number of cycles, N, to be determined.
The number of cycles to failure, N, obtained by the S-N curve (under load or
stress control condition) is related to the total life of the part up to failure. In real-
ity, fatigue cycles throughout the life of the structural part consist of crack initia-
tion and propagation. Here, the term "propagation" refers to stable crack growth
up to the crack instability. The S-N curve approach does not separate the crack
initiation phase from the propagation. However, in some industries, it is required
to assume that the crack has already initiated in the hardware and only the total
number of cycles associated with the propagation are of interest to the analyst.
Existing or initiated cracks assumed in the structural part are the result of load
cycles induced from manufacturing, machining, or improper handling prior to its
actual usage. The size of the preexisting crack can be assumed based on the capa-
bility of the inspection methods used to detect it. The analyst may assume an ini-
tial surface crack based on the inspection method applied to the part. Using the
40 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

available initial flaw size, the total life of the part can be evaluated by using a
fracture mechanics approach.
Inspection methods, such as penetrant inspection, magnetic particle, eddy cur-
rent, ultrasonic, or x-ray (mainly for embedded cracks due to welds) are currently
in practice throughout the aerospace industry. For example, for the case of a stan-
dard dye penetrant inspection, NASA has recommended that the analyst assume
an initial circular surface crack of 0.075 in. in depth (with depth-to-Iength ratio
ale = 1) to use for evaluating the life ofthe part that are considered as the main
load carrying structural component. The size of the final crack at the instability
depends on the fracture toughness of the material, the crack geometry, and the
applied stress. Section 5.3 of Chapter 5 will cover the inspection methods and the
corresponding flaw size for different crack geometries and Chapter 4 will discuss
the application of the fracture mechanics on fatigue crack growth rate concept that
is currently used in aerospace and other industries.

2.4.3 Parameters Affecting the 5-N Curve

For a given alloy, the shape of the S-N curve will vary according to the conditions
represented. Some factors influencing the shape of the S-N curve are:

• Material conditions
• Heat treatment
• Cold working
• Types of load
• Tension
• Compression
• Torsion
• Combined load
• Stress ratio
• Rate of load application
• Environment
• Corrosive
• Inert
• Temparature

Example 2.3
A component of a space structure is made of 4130 alloy steel. The space structure
is orbiting around the earth every 90 minutes (see also Fig. 2.4). The maximum
thermal stresses induced in the part due to exposure to the sun (once every
90 minutes) is 14 ksi. Determine if the part can survive the environment during its
30 years in space (R = -1).
The equation describing the S-N diagram for 4130 alloy steel is extracted from
Reference [12] and is expressed as:
 2.5 Constant-Life Diagrams 41

logN = 9.65 - 2.85 log [Smax(l - R)OAI - 6.13]

The total number of cycles during the 30 years in space is:

N tota! = (30 X 365 X 24 X 60)/90

= 175,200 cycles

The number of cycles to failure based on 14ksi stress (R = -1) can be calculated as:

Smax (1 - R)OA! = 14 X (2)OA!

= 18.6 ksi

10gN = 9.65 - 2.8510g(18.60 - 6.13)

NF = 10.0 6.52 cycles

It can be seen that the number of cycles to failure (N! = 10.0 6 .52 cycles) is larger
than the total number of cycles (Ntota! = 1.75 X 105 cycles) required for the struc-
ture to orbit around the sun during its 30-year mission in space. Therefore, the
structure can survive 30 years in the space environment.

2.5 Constant-Life Diagrams

The purpose of constant -lifediagrams is twofold: (1) Most of the available fatigue
data are obtained through testing by applying the fully reversed loading case
(R = -1), where mean stress, SM' is zero. The cases of completely reversed
cycles (R = -1) and zero to maximum stress (R = 0) are most commonly used in
developing the S-N curve. However, in the majority of cases, the analyst would
like to design the structural components for the loading case where the mean
stress, SM' is different from zero. It is, therefore, important to be able to utilize
the test data pertaining to the case of R = - 1, collected in the laboratory, for the
life assessment of other loading cases. Constant-life fatigue diagrams are a family
of curves generated for different fatigue life, N, each representing the variation of
either Smax versus Smin' Smax versus SM' or Sa versus SM" The most common type of
constant life fatigue diagram (that can be interpreted easily) is the plot of Sa ver-
sus SM' This diagram was presented by Goodman as a straight line relationship
shown by equation 2.8. The Goodman equation relating Sa and SM can easily con-
vert the available fatigue data of a given material from the case of R = - 1 to
another format (where the mean SM OP 0) useful to the analyst. (2) Constant-life
diagrams are also useful when designing structural components for an infinite life
where N > 10 7 cycles or when designing for finite life, N.
42 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

All methods of plotting fatigue diagrams provide the same type of information.
Because the maximum and minimum stress values of each load cycle are recorded
directly by the testing machine, it would be easier to use these stresses for plot-
ting, for example, the S-N diagram, rather than to convert them to alternating
stress, (Smax - Smin)/2, versus mean stress (Smax + Smin)/2 or maximum versus
mean stress. Since the generation of a constant-life diagram is required to have
several experimental data, which are difficult and expensive to generate, it is
much easier to develop an empirical relationship that can relate the alternating
stress, Sa' to the mean stress, SM. In Sections 2.5.1 and 2.5.2, the empirical and
experimental methods of developing the constant-life diagram are discussed.

2.5.1 Development of a Constant-Life Diagram

Constant-life diagrams can be generated either by empirical relationship or by the
test data. From a series of fatigue test data developed for S-N curves, the variation
of Sa versus SM with various combination of stress ratio, R, was plotted [14] and
the general trend is illustrated in Fig. 2.16. The test data illustrated in Fig. 2.16
were collected for a fixed number of cycles to failure, N. In Fig. 2.16, the com-
pressive mean stress (the region associated with -SM) is not shown because the
data is more sensitive to positive mean stress, SM. Note that the alternating stress,
SaO' represents the fully reversible case (R = -1) where SM = O. Other stress
amplitudes along the dotted line have nonzero mean stress and they are referred to
by Sa. The experimental data indicates that as the alternating stress, Sa' decreases,
the mean stress increases. When stress amplitude, Sa' approaches zero, the data
points tend to approach the ultimate tensile strength of the material, as shown in
Fig. 2.16. Based on this observation, Gerber from Germany (1874) and Goodman

A~..JM=O
Sao- .,. (R=-1)
• I.. Constant N for all the points

J
• ..... obtained experimentaly.
• ·.1:..
••
••••••• •
Sa ....,.. SM = SUi
••• (R =1)
•••
•=: I
Sui -
Figure 2.16 lIIustration of the variation of alternating stress, Sa as a function of mean
stress, Sm
 2.5 Constant-Life Diagrams 43

(England, 1899) [15, 16] proposed an empirical relationship that would approxi-
mate the experimental data describing the constant-life stress relationship illus-
trated in Fig. 2.16.

(Sa/SaO) + (SM/SUl)2 - 1= 0 Gerber [10] (2.7)

(Sa/SaO) + (SM/Sul) - 1= 0 Goodman [11] (2.8)

Where experimental data are not available, the proposed Gerber and Goodman
constant-life stress relationship can be used. Goodman's original law pertaining to
infinite life design assumed the alternating endurance limit (where SM = 0) to be
one third of the ultimate tensile strength of the material. This assumption has since
been modified to the relationship shown by Eq. (2.8). Note that the above con-
stant-life approximation equations described by Eq. (2.7) and (2.8) do not apply to
brittle material.
For an infinite life design, the quantity SaO is set equal to Sen (endurance limit
when SM = 0). Figure 2.17 shows the Gerber and Goodman diagrams describing
the constant-life stress relationship.
It is customary to plot the Goodman and Gerber constant-life diagram in terms
of non dimensional quantities, for example Sa/SaO versus SM /Sul' as shown in Fig.
2.18. Experimental data for aluminum alloys and steels pertaining to infinite life
design had shown that [17] for constant-life analysis, the Goodman line may be
considered as the lower bound and the Gerber line as an upper bound (see Fig.
2.18 where experimental data are falling between the two lines).

2.5.2 Development of a Constant-Life Diagram

by Experiment
The procedure for experimentally obtaining a constant-life diagram for ductile
material is to extract data from several S-N curves having different stress ratios.
Two extreme values of R = -1 (where SM = 0) for fully reversed cyclic loading

¥
R =-, Constant Life, N
SaO~---

R -1
Goodman Diagram
I
Figure 2.17 Illustration of the Goodman and Gerber diagrams
44 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Constant Life, N

Sa/SaO

R =1
• Experimental data falling
between the two diagrams
I

Figure 2.18 Illustration of the Goodman and Gerber diagrams in terms of nondi-
mensional quantities Sa/SaO vs. SM /Su

Data points extracted from several 8-N curves

N3 N2 N,
Number of Cycles to Failure, N

Figure 2.19 Illustration of typical S-N curves for three stress ratios of R l • R 2 • and R3

and R = 1 (where Sa = 0) for steady load with no cyclic loading are essential to
construct the constant-life diagram. Figures 2.19 and 2.20 illustrate the method of
plotting constant-life lines from a family of S-N curves. In Fig. 2.19, three S-N
curves, corresponding to three stress ratios R J ' R2 , and R3 , were converted to
constant-life diagrams with lives of N J , N2 , and N 3 , respectively, as illustrated in
Fig. 2.20.

Example 2.4
A fracture critical rod (i.e., the failure of the rod will cause a catastrophic failure
to the main structure and therefore, it is considered a high risk part) is made of
 2.5 Constant-Life Diagrams 45

R1 (R =-1)

RZ

Figure 2.20 Illustration of the method of obtaining the constant-life diagram through
experimental data

26.0 ksi Preloaded

Stress
13.75 ksi
UI
~
UI
CIl
I-
+.I
I/)

Cycle

Figure 2.21 Load environment for the High Risk Preloaded Bolt in Example 2.4

2219-T851 aluminum alloy and is subjected to maximum cyclic stress, Smax = 26

ksi, as shown in Fig. 2.21. The rod is preloaded to 25% of its ultimate (Sui = 55
ksi). The endurance limit for this material under fully reversed loading (R = -1)
is 8 ksi (where N = 10 7 cycles). Fully reversed stress at 3000 cycles is provided
by the S-N curve as 30 ksi. Use the Goodman constant-life relationship to calcu-
late the life of this component.

Solution
The following quantities can be calculated:
46 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

SM = 0.25 X Sui
= 13.75 ksi

= 1.5 ksi

And the stress amplitude, Sa' is given by:

Sa = (Smax - Smin)/2
= 12.25 ksi

From the Goodman relationship [Eq. (2.8)] the value of SaO' when SM = 0, can be
calculated by using a second point on the Goodman line where SM = 13.75 and
Sa = 12.25 as:

(SaiSaO) + (SM/Sul) - 1 = °
(12.25/S aO) + (13.75/55) - 1 = °
SaO = 16.33 ksi

The constant-life diagram based on the Goodman relationship is shown in Fig.

2.22 for SaO = 30.0, 16.33, and 8.0 ksi.
Using Eq. (2.6), the number of cycles to failure for the preloaded rod can be
calculated. By using any two points (m and n) along the straight line, the slope
associated with the S-N curve can be evaluated. Utilizing Eq. (2.5):

N = 3000 Cycles
30.0 / N Cycles (to be determine)
, 6.33 N = 107 Cycles

, 2.25 t----:::!~~

8.0 [=:=t:::::::~~~~"--t~
, 3.75 SUi = 55

Figure 2.22 Constant Life and Goodman Diagram

 2.6 Linear Cumulative Damage 47

Substituting Sm = 30 ksi and Sn = 8 ksi, with Nm = 3000 cycles and Nn = 10 7

cycles, respectively;

log30 - log8 = -b (IoglO 7 - log3000)

b = -0.16

For SaO = 16.33 ksi, the number of cycles to failure can be calculated from
Eq. (2.6) as:

N = (16.33/8f 1/ o.16 X 10 7
N = 115,650 cycles

All the points on the curve corresponding to N = 115,650 cycles have the same
fatigue life, including the preloaded bolt with mean stress SM = 13.75 ksi.

2.6 Linear Cumulative Damage

The S-N diagram is useful to determine the number of cycles to failure associated
with a given constant amplitude applied cyclic stress. When the total damage
induced on a given part is a result of several load environments, with different
fluctuating stresses, the contributing damage caused by each environment should
be evaluated. Consider the case of a satellite that will experience fluctuating
cycles of different stress level, for example, launch, on-orbit, and thermal cycles,
during its total life. Similarly, the frames of aircraft are subjected to many differ-
ent cyclic loads, depending on the altitude, speed, takeoff and landing, and air tur-
bulence. The ratio of the partial damage, di , due to each individual environment
having stress level, Si' over the total damage, D i , must be additive. The sum of
these partial damage fractions should be unity:

(2.9)

Based on this assumption, the Palmgren-Miner rule [2, 3] (first proposed by

Palmgren and later developed by Miner) states that the fatigue damage contribu-
tion by each individual load spectrum at a given stress level is proportional to the
number of cycles applied at that stress level, ni , divided by the total number of
cycles required to fail the part at the same stress level (Nji ). It is obvious that
each ratio can be equal to unity if the fatigue cycles at the same stress level would
48 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

continue until failure occurs. The total failure, in tenns of partial cycle ratios, can
be written

(2.10)

Equation (2.10) is a useful tool to detennine the life of a given structure sUbjected
to several cyclic load cases of different stress magnitude.

Example 2.5
A component of a space structure made of 2219-T851 aluminum alloy is sub-
jected to fluctuating loads with different stress magnitudes as shown in Table 2.9.

Table 2.9 Launch and on-orbit load spectrum for Example 2.5

Launch On-orbit

# cycles Limit stress # cycles Limit stress

N (ksi) N (ksi)

3 20.5 100 18.29

18 18.5 300 15.97
13 16.4 1300 14.50
27 14.4 2700 13.29
95 12.3 9500 12.00
159 10.3 1.5E + 4 11.68
343 8.20 3.4E + 4 9.36
1134 6.15 11.3E+4 8.04
5349 4.10 53.4E + 4 6.72
5537 2.05 55.3E + 4 5.40
7939 1.40 79.3E + 4 4.70
on-orbit Thermal spectrum

175000 cycles 2.0 ksi

Including a life factor of 4 (4 X number of cycles occurring during the service

life of the structure), detennine if the part will survive the load environment dur-
ing 15 years of its mission on orbit (one mission is equivalent to one service life).
The equation describing the S-N curve described by Eq. (2.2) is:
 2.6 Linear Cumulative Damage 49

10gN = 9.65 - 2.85 log (SmaxO - R)O.4! - 61.3) (2.2)

The load spectrum for launch and on-orbit (including on-orbit thermal cycles
when the structure is exposed to cold and hot temperature; see Fig. 2.4) in terms
of the limit load (maximum operating stress when no factor of safety is included)
are given in Table 2.9:

Solution

The number of cycles to failure, Nf , associated with each step corresponding to

the launch and on-orbit environment can be computed by using Eq. (2.2) (see
Table 2.10). The ratio of N)Nfi is related to partial damage described by Eq.
(2.10). The total failure, in terms of the partial cycle ratios, can be calculated
by employing the Palmgren-Miner rule, see Eq. (2.10). With a life factor of 4
(the safety factor of 4 is included to account for the uncertainties in the mater-
ial properties, load environment and analysis approach), the total summation is
less than 1 (see Table 2.11). Therefore, the structure can survive the environ-
ment.

Table 2.10 Number of cycles to failure Nfi , for each stress

magnitude corresponding to Example 2.5

Launch On-orbit

# cycles Limit stress # cycles Limit stress

Nfi (ksi) Nfi (ksi)

3.6E +5 20.5 5E +5 18.29

4.8E +5 18.5 7.3E +5 15.97
6.8E +5 16.4 9.7E +5 14.50
lOE +5 14.4 12E +5 13.29
15E +5 12.3 16E +5 12.00
26E +5 10.3 18E +5 11.68
49E +5 8.2 34E +5 9.36
112E +5 6.15 52E +5 8.04
356E +5 4.1 87E +5 6.72
257E +6 2.05 16E +6 5.40
710E +6 1.40 24E +6 4.70

Thermal

276E +6 2.0
50 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Table 2.11 Number of cycles to failure Nfi , and the sum of ni INfi used in minor's rule
for Example 2.5

Launch On-orbit

Nfi ni n)Nfi Nfi ni n)Nfi

3.6E + 5 3 8.3E - 6 5E + 5 100 2.0E - 4

4.8E + 5 13 2.7E - 5 7.3E + 5 300 4.1E - 5
6.8E + 5 18 2.6E - 5 9.7E + 5 1300 1.3E-4
lOE + 5 27 2.7E - 5 12E + 5 2700 2.3E - 3
15E + 5 95 6.3E - 5 16E + 5 9500 5.9E - 3
26E + 5 159 6.1E - 5 18E + 5 1.5E + 4 8.3E - 3
49E + 5 343 7.0E - 5 34E + 5 3.4E + 4 .01
112E + 5 1134 1.0E - 4 52E + 5 11.3E+4 .02
356E + 5 5349 1.5E - 4 87E + 5 53.4E + 4 .06
257E + 6 5537 2.1E - 5 16E + 6 55.3E + 4 .03
7lOE + 6 7939 l.1E-5 24E + 6 79.3E = 4 .03
TOTAL 5.3E - 4 TOTAL 0.166
Thermal

276E + 6 175200 6.3E - 4

Sum n)Nfi = 0.166 + 5.3E - 4 + 6.3E - 4
= 0.167 < 1
with a safe-life factor of 4, the summation is:
4 X 0.167 = 0.668 < 1

2.6.1 Comments Regarding the Palmgren-Miner Rule

The Palmgren-Miner rule, described by Eq. (2.10), has been used by engineers in
the past and is still considered a simplified and versatile tool for determining the
total life of a given structure under study. However, the oversimplified assump-
tions that the damage summation described by Eq. (2.10) is linear and that no
account is made of the sequence in which the crack tip is experiencing cyclic
stresses have drawn criticism by investigators in this field. For example, the total
life analysis should yield shorter life if the crack tip experiences high stress cycles
during its early stage of life followed by the smaller amplitude cycles rather than
the other way around. In this case, the smaller stress cycles, following the larger
amplitude cycles, are more effective in damaging the structure (damage becomes
more sensitive to load cycles as it advances and grows in size). It is possible that
using the high-amplitude cycles first and low cycles next may result in a Palm-
gren-Miner sum less than 1, while using the low-amplitude cycles first and high-
amplitude stress next may result in a damage summation greater than 1. Moreover,
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 51

load interactions between high and low amplitude load cycles causing a retarda-
tion effect (i.e. a delay in damage growth) is not considered in Miner's rule.

2.7 Crack Initiation (Stage I) and Stable Crack

Growth (Stage II)

2.7.1 Fracture Surface Examination

The fatigue failure mechanism of metals occurs in three steps: (1) a microscopic
crack is initiated at some localized spot after N number of cycles, (2) the initiated
crack will grow in a stable manner (stable crack growth), and (3) the growing crack
will reach its critical length and create instability. In brittle materials, the crack in-
stability means fast fracture where separation occurs by the cleavage mechanism
along low-index crystallographic planes, known as cleavage planes, by direct
breakage of atomic bonds. Brittle failure also occurs in relatively ductile metals un-
der the plane strain condition where a very small amount of energy is consumed for
the separation mechanism with little or practically no plastic deformation. In con-
trast, in ductile materials under plane stress condition, shear type rupture occurs on
shear planes inclined at approximately 45° to the applied tensile load. In studying
the two fractured surfaces, two distinct zones, related to stable crack growth and in-
stantaneous fracture, are evident and they are called the fatigue and rupture zones.
The fractured surfaces contain information related to the nature of the applied
load (load history), severity of the environment (e.g., corrosion), and the material
quality (type of heat treatment). The two fractured surfaces associated with the
fatigue zone tend to be macroscopically flat, shiny, and smooth and show a
"beach marking" or "clam shell" appearance at uniform intervals due to the vari-
ation in the load amplitude during each cycle. The beach markings (also called
striation marks) can be absent on the cracked surfaces when uniform cyclic load-
ing with little variation in the load amplitude is applied in the laboratory. Striation
marks represent the position of the advancing crack when it is subjected to fluc-
tuating load amplitude and, in most cases, each striation mark represents the
amount of crack growth per load cycle. These markings are perpendicular to the
direction of the advancing crack and are helpful in locating the exact origin of
crack initiation on the structural part by backward tracking of the crack growth. A
schematic representation of fatigue and rupture zone and the surface marks pro-
duced on smooth specimens (circular cross sections) under tension-tension and
reverse bending subjected to high and low loading conditions are shown in Figs.
2.23 and 2.24a, respectively. An electron microscope picture of striation marks
for 6061-T6 aluminum alloy is also shown in Fig. 2.24b which clearly illustrates
the position of the advancing crack when it is subjected to fluctuating load cycle.
Comprehensive representations of beach markings for other loading conditions
are available in Reference [18].
52 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

High Applied Stress

No Stress
Tension Concentration Bending

Crack Initiation

Unstable crack propagation

Note: The rupture zone is larger than the fatigue zone

Figure 2.23 Generation of beach marks due to tension-tension and reversed bending
high load for unnotched specimen

Low Applied Stress

No Stress
Tension Concentration Bending

Crack Initiation Crack Initiation

Stable crack growth

.---../

Unstable crack propagation

Note: The rupture zone is smaller than the fatigue zone

Figure 2.24a Generation of beach marks due to tension-tension and reversed bend-
ing low load for unnotched specimen
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 53

Figure 2.24b The striations marks provided by the electon microscopic technique for
6061-T6 aluminum alloy which represents the position of the advancing crack

Figs. 2.25a and 2.25b show the fracture surface appearance of a bicycle spoke
made of7075-T6 aluminum alloy that is fractured in two pieces due to in-service
fluctuating load, using a light microscope with 25 X and 100 X magnification,
respectively (see also Fig. 2.25c which was taken from the same specimen surface
by an electron microscope with 2000 X magnification). Fatigue crack initiation
usually starts at the free surface or subsurface and macroscopically it can be traced
to either surface irregularity (such as a notch, a scratch, or an abrupt change in
section) or to corrosion pitting. Note that the presence of residual compressive
stresses can cause the crack initiation not to occur on the surface of the material.
Investigation of the fatigue surface shown in Fig. 2.25 indicated that the crack ini-
tiation was due to surface scratches together with a corrosive environment.
Careful examination of the fatigue and rupture zones on the fracture surface
can give information about the kind of load, its magnitude, and the material's frac-
ture toughness. A short fatigue zone with a large final fracture surface indicates
that the applied stress intensity factor exceeded the fracture toughness (material
resistance to fracture) of the material at a relatively short crack length (see Chap-
ter 3 for an in-depth study of the stress intensity factor and fracture toughness con-
cepts) or the material fracture toughness was low. Looking at the fracture surface
of the bicycle spoke shown in Fig. 2.25 indicates that the area of the final fracture
(rupture zone) is small as compared with the fatigue zone. It can be deduced that,
54 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Figure 2.25a Fracture surface appearance of a bicycle spoke made of 7075-T6 alu-
minum alloy (light microscope with 25 x magnification)

Figure 2.25b Fracture surface appearance of a bicycle spoke made of 7075-T6 alu-
minum alloy (light microscope with 100 x magnification)
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 55

Figure 2.25c The Fracture surface of the bicycle spoke failure viewed by the electron
microscope technique with a magnification factor of 2000 X (7075-T6 aluminum
alloy)

in the case of the bicycle spoke failure shown in Fig. 2.25, the applied load was
relatively low and the material fracture toughness was high.
Microscopic examination of fatigue fracture surfaces began with a study by
Zapffe and Wooden using a light microscope [19] with magnification capability
of up to 500 X. The most distinguishing microscopic feature of a fracture surface
is its beach mark or striated surface appearance. Under low-cycle fatigue, where
the applied cyclic load is high, the striations are coarse enough to be observed by
the light microscope [18]. At lower stress levels, however, the striation marks are
difficult to observe even with the help of the light microscope. Higher magnifica-
tions are required to resolve these fine marks. The restricted depth of field of the
light microscope limits such surface examination, and thus electron fractography
is the principal technique used to analyze the striated surfaces. Figure 2.25c shows
the fracture surface of the bicycle spoke failure viewed by the electron microscope
technique with a magnification factor of 2000 X.
Another example of a fractured surface that has beach marking appearance is
shown in Fig. 2.26 using a light microscope with 16 X magnification. The frac-
tured part is a 4130 steel bolt (with 125 ksi ultimate) that failed during the truck
transportation of a missile to Cape Canaveral, in Florida. The bolt attaches the
56 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Figure 2.26 Fracture surface of a 4130 steel bolt viewed by a light microscope with a
magnification factor of 16 X (striation marks on the fatigue zone can be seen on the
surface of the part.)

rocket engine exhaust nozzle to a struts support that was under cyclic loading dur-
ing the transportation. Looking at the surface appearance of the 4130 steel bolt
(see Fig. 2.26), it can be said that the fatigue zone (stable crack growth region) is
larger than the rupture zone from which one may conclude that the material had
good fracture toughness and, moreover, the magnitude of the cyclic load was rel-
atively low. A careful examination of Fig. 2.26 reveals that the bolt was exposed
to a few tensile overload cycles creating a few coarse striation marks.
From a macroscopic point of view, the fracture surfaces due to fluctuating
load, exhibit a flat, shiny surface appearance, as if the failure was associated with
a brittle material, that is, the lack of necking or shear lip formation on the two
fatigue surfaces. This phenomenon occurs even in metals that would be consid-
ered quite ductile when tested under monotonically increasing tensile load. The
flat and shiny appearance of the fracture surface led metallurgist to believe that
the metal had undergone recrystallization and had thereby become brittle.
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 57

2.7.2 Introduction to Crack Initiation

Although there are large amounts of laboratory test data and publications on this
subject, there has not yet been established any fundamental theory that can fully
explain the cause of the crack initiation phenomenon. Most of the development in
this area is based on theoretical equations containing parameters that can be
obtainable through laboratory experiments. In all these cases, the results of the
theoretical work were checked against the experimental results to verify the valid-
ity of the model. In general, the formation of stage I (crack initiation) is influ-
enced by the magnitude of the applied cyclic load, the severity of the corrosive
environment, and temperature [20]. There are several proposed models that can
describe the initiation of fatigue cracks in the load varying environment, all of
which rely on the localization of plastic deformation by a slip movement mecha-
nism and the formation of intrusion and extrusion on the surface of the material
[20-22]. Stage I crack initiation by the slip band mechanism is presented in Sec-
tion 2.7.3. In Section 2.7.2.1 the fracture mechanism when applied monotonic
load is the main cause of failure is briefly discussed.

2.7.2.1 Fracture Due to Monotonic Applied Load

When the applied monotonic tensile load is the principal cause of fracture (a sin-
gle load application to failure versus cyclic load), the structural failure can be
described by a process known as microvoid coalescence. Second phase particles,
inclusions, and dislocation pileups are the source of strain discontinuities that are
suitable regions for initiation of microvoids [23, 24]. As the applied load
increases, the microvoid density increases; they grow and coalesce to a critical
dimension to cause fracture. The two fracture surfaces contain numerous dimples
having a cuplike conical shape and for that reason, this type of fracture is referred
to as dimple rupture (illustrated in Fig. 2.27a). The size of the dimples observed
on the two fracture surfaces is governed by the density of nucleated microvoids in
the volume of the material. A large number of microvoids result in the formation
of small dimples and, conversely, when the nucleation sites for microvoids are
few, the individuals grow to a larger size before the coalescing mechanism occurs
and results in the formation of large dimples. Figures 2.27b, c, and d show the
fracture surfaces of 4340 and A286 steels and 2014-T6 aluminum alloy subjected
to monotonic applied load which clearly show the density of nucleated
microvoids and the corresponding size of the formed dimples after failure.
Figures 2.27b, c, and d were taken with electron microscope techniques using
2000 X magnification.
The fracture surfaces due to shear overload exhibit dimples of different shape
as compared with tension overload. For mode II fracture, the elongated dimples of
58 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Microvoids
coalesence

Dimples (conical shape) as the result of

microvoids coalesence formed on both
fractured surfaces.

Figure 2.27a Microvoids coalescence and the formation of equiaxed dimples for ten-
sion overload.

Figure 2.27b Formation of dimples for tension overload provided by the electron
microscope (4340 steel with 180/200 ksi)
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 59

Figure 2.27c Formation of dimples for tension overload provided by the electron
microscope (A286 steel with 160/180 ksi)

each mating surface are formed in such a way that their heads are oriented in
opposite directions (see Fig. 2.28). In contrast, in mode I fracture (tension over-
loading), the dimples on the two surfaces are almost of equal size and are referred
to as having equiaxed dimples. Equiaxed dimples that have equal rim area in gen-
eral are not equal in the depth direction.
Prior to explaining the crack initiation in the load varying environment, where
the initiated crack length is smaller than or on the order of the material grain size,
it is important for the reader to have some basic knowledge of material crystal
structure and the concept of slip mechanism.

2.7.2.2 Crystal Structures of Metals and Plastic Deformation

Metals generally consist of regions with a regular geometrical arrangement of
atoms, called crystals or grains. The grain is merely a crystal that does not have
smooth faces (Fig. 2.29). However, within a grain, the arrangement of atoms are
as perfect as a crystal. The average diameter of a grain is in the range of 0.01 to
0.001 in. and varies greatly with the mechanical and heat treatment received. Just
60 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Figure 2.27d Formation of dimples for tension overload provided by the electron
microscope (2014-T6 aluminum alloy)

Formation of Elongated Dimples for the

Mating Surfaces (Shear load Application)

Figure 2.28 Microvoid coalescence and the formation of elongated dimples during
shear overload
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 61

A Cubic Cell

Grains Within
a Solid

Figure 2.29 Illustration of slip plane and slip direction of a body centered cubic cell

like a crystal, a grain is made of repeated groups of atoms, called unit cells [25,
26]. A body-centered cubic unit cell of pure alpha iron is shown in Fig. 2.29.
Plastic deformation of metals can occur by a process called slip, where the
adjacent planes of atoms move within the crystal. In simple terms, a slip mecha-
nism can be visualized as a shearing forces that slides one card in a deck of cards
over another. This slip takes place only in certain crystallographic planes, usually
those with the most dense atomic packing. A slip plane and slip direction for the
body-centered cubic unit cell structure are shown in Fig. 2.29. Slip planes for
face-centered cubic structures (such as Cu, Ag, Au, AI, and Ni) are {Ill} and
{11O} for body-centered cubic metals (Li , Cr, K, and Fe). To illustrate plastic
deformation, let us assume we have a number of single crystals of copper in the
form of a rod. The rod can be utilized as a tensile specimen by fixing it at one end
and pulling at the opposite end. The force transmitted across a particular plane of
the rod can be resolved into its components normal and tangent to the plane, as
illustrated in Fig. 2.30. Experiments have shown that when the tangential compo-
nents of the force exceed a certain value on the area associated with the slip plane,
sliding will occur. This critical force value, when acting on the area of the slip
plane, is called the yield stress. Its value is independent of the normal components
of the force on the slip plane. In perfect metals (where imperfection is minimized),
the magnitude of the maximum shear stress in the plane (yield stress), using laws
describing the interatomic forces, is about 10 7 psi. However, the actual yield value
associated with pure iron, obtained by a standard tensile test, is several orders of
magnitude lower than the theoretical value. The difference between the theoretical
and laboratory values mentioned above is based on the imperfections in the crys-
tal, such as voids, inclusions, and dislocations. The concept of crack initiation by
62 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Tangential
Normal

Figure 2.30 Resultant of an axial force into normal and tangential components

slip movement within a crystalline lattice structure of a grain is viewed as the dis-
placement of dislocations in the lattice under the action of shear. Prior to fatigue
failure, dislocations or mismatched atoms tend to pile up along the grain bound-
aries and/or slip bands and cause crack initiation and growth in the structural part.
Microscopic cracks of neighboring grains (initiated along the slip band with the
similar mechanism) will eventually combine to form a sizable crack able to prop-
agate perpendicular to the applied load. Figure 2.31 shows a single dislocation
that created a regional lattice imperfection that is surrounded by a perfect cubic
lattice. In this figure, the upper portion of the crystal is displaced by one atomic
spacing, b, creating a mismatch and internal stress between atoms. In a metal,
there may be 10 7 to 10 9 dislocations in a square inch of area of the crystal
throughout the metal.

A Dislocation A Perfect Crystal

Figure 2.31 Single dislocation in otherwise perfect crystal: introducing mismatch and
internal stresses among atoms
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 63

The sliding mechanism under low-amplitude long life cyclic loading is some-
what different from the monotonic tensile loading case shown in Fig. 2.32. Sec-
tion 2.7.3 describes the slip plane cracking mechanism and creation of intrusion
and extrusion on the surface of the part, which can cause crack initiation under
cyclic loading.

2.7.3 Crack Initiation Concept (Intrusion

and Extrusion)
When the applied tensile load is monotonic, the slip steps that appear on the sur-
face of the specimen have simple geometry, as shown in Fig. 2.32. Coarse slip is
normally associated with monotonic tensile loading; however, under low-ampli-
tude fatigue loading, the surface of the material tends to deform by cyclic slips
which are finer and more localized in nature. Slip lines (illustrated in Fig. 2.32)
tend to concentrate and group into a bundle called slip bands. The surface topol-
ogy corresponding to a cyclic fine slip looks like irregular notches at the specimen
edge consisting of extrusions and intrusions, as shown in Fig. 2.33 [23, 24,
27-29]. An extrusion is a small piece of material that is squeezed out from the
surface of a slip band due to the presence of voids and dislocations pileup. The
inverse of extrusion is intrusion and is also illustrated in Fig. 2.33. Each of these
effects can lead to the localization of plastic strain by the formation of stress ris-
ers on the surface of the material. Continuation of the cyclic load leads to deep-
ening of the intrusion and, eventually, the formation of a microcrack along the
slip band (stage I).
In ideal case of a single crystal of copper, the crack initiation based on slip bands
formation can be described as follow [30-33]: Upon application of low-amplitude
cyclic loading, in those crystals that are oriented favorably with respect to the ap-

Figure 2.32 Illustration of slip by a monotonic tensile load

64 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

P~CYCle
p

Stage I Crack Initiation

Each grain has
P slightly different

~ orientaion so that
Cycle the slip planes are
in different direction.

Figure 2.33 Illustration of intrusion and extrusion due to slip as a result of cyclic
loading

plied fluctuating load, after N number of cycles fine localized slip lines will appear
on the surface of material (see Fig. 2.34, case A). At the early stages of cycling the
slip lines appear on the surface for the tensile portion of the cycle and disappear
when the unloading portion of the cycle takes over. This reversibility continues for
a few cycles until permanent slip bands are formed (see case B). The concentration
of slip lines that are grouped in bundles and tend to increase in width (slip bands 10
to 50 /-Lm wide) are referred to as persistent slip bands (PSBs) which later become
the soft spot for the formation of microcracks. The appearance of slip bands on the
surface occurs if we assume the material is free from any residual compressive
stresses, otherwise the crack initiation may not occur on the surface of the material.
When cycling continues, anyone of the PSB can result in the formation of crevices
(intrusion and extrusion; see case C) which are localized stress concentration sites
for the nucleation of a fatal crack. R. E. Peterson under "Fatigue Cracks and
Fracture Surfaces" indicated that when penetrating oil was placed on the surface of
a fatigue specimen with heavy slip bands (case C of Fig. 2.34) bubbles are formed
under repeated fluctuating load, thus indicating that separation between sliding
plates had occurred. Additional subsequent cycling created crack initiation along
 2.7 Crack Initiation (Stage I) and Stable Crack Growth (Stage II) 65

Extrusion

Intrusion
4·
I
Slip line
Case A Case B Case C

Figure 2.34 Slip lines and slip bands leading to crack initiation

the slip band and crack growth perpendicular to the applied load. The crack initia-
tion and growth mechanism at the early stage of fatigue growth is strongly influ-
enced by microstructural parameters such as grain size and slip band orientations.
Further continuation of the cyclic load will result in the formation of stage II (stable
crack growth), until finally, when the critical length is reached, the crack will be-
come unstable. In the early stages of stable crack growth, the direction of propaga-
tion is not perpendicular to the applied load, but depends on the orientation of the
primary slip band within the crystals of a grain (see Figs. 2.35 and 2.36). When the

Slip Band .....I---i.~ ....

__------tI.~~
Stage I Stage II Instability
Initiation Stable Crack Growth

Figure 2.35 lIIustration of stage I and stage II crack initiation and propagation
66 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Stage I (Initiation) p p~V

".. Grains

r~
V '-J r-
,
S"ge'

"..
Stage II

p
Intrusion

Figure 2.36 Detail description of stage I and stage II

crack proceeds to intersect the grain boundary, the growth direction of propagation
alters (and may be arrested by the grain boundary) and the crack will tend to orient
itself with the next grain having specific crystallographic planes suitable for crack
growth. As the length of the crack increases, the crack will eventually tend to orient
itself normal to the direction of the applied load, as illustrated in Figs. 2.35 and
2.36. Fatigue crack growth of stage II occurs by transgranular fracture (where crack
propagates along the grain boundary) and is influenced by the magnitude of the al-
ternating stress, the mean stress, and the severity of the environment. The extent of
both stage I and stage II is governed by the material type and the magnitude of the
tensile cyclic stress; the lower the magnitude of this stress, the longer the time as-
sociated with stage I will be. On the other hand, if the magnitude of the applied ten-
sile cyclic stress is high, the number of cycles in stage I is short and the remaining
life is consumed for stable crack growth (stage II). Note that there is no simple or
clear outline of the boundary between stage I and stage II.
Both intrusion and extrusion can be described as the localized region of high
stress concentration, where the damage initiation is most likely to occur. These
critical locations can eventually cause localized plastic deformation (the material
surrounding the plastic zone remains elastic) that ultimately under cyclic loading
lead to the complete failure of the part. When the applied cyclic load is above the
tensile yield of the material and the bulk of the structure is plastically deformed,
the number of cycles to failure is small. The failure prediction must be assessed by
 2.8 Low-Cycle Fatigue and the Strain-Controlled Approach 67

the strain-controlled approach. The next section discusses low cycle fatigue and
the strain-controlled approach.

2.8 Low-Cycle Fatigue and the Strain-Controlled

Approach

A complete fatigue failure curve can be divided into both low-cycle and high-
cycle regimes. In the low-cycle fatigue region, the plastic deformation size is
macroscopic and the number of cycles to failure is below 10 4 cycles. High-cycle
fatigue is characterized by microscopic localized plastic deformation 2 and for
many materials the number of cycles to failure is above 10 4 and up to 10 7 or
higher. As mentioned in Section 2.7, a fatigue crack initiates at some surface dis-
continuity (sharp corners, hole, or notch) where material is locally overstressed
and plastically deformed. The number of cycles associated with the crack initia-
tion and stable crack growth (stages I and II) is a function of nominal stress mag-
nitude. For cyclic induced stress amplitude above the yield stress, the bulk of the
material is plastically deformed. In this situation, the extent of stage I is short and
most of the cycles are spent for stable crack growth.
In high-cycle fatigue, the cyclic induced stress is in the elastic range, and plas-
tic deformation is highly localized. The extent of stage I is large and the bulk of
the structure is elastic, with plastic deformation in the locality of stress concen-
trations. In both cases of high-cycle and low-cycle fatigue, the material response
to cyclic loading in the critical locations (such as sharp corners, holes, or notches)
is strain-controlled [34-36]. In high-cycle fatigue, where the strains are predomi-
nately elastic, it is convenient to measure the strain near the highly strained region
directly from linear elasticity by knowing the applied load, since the plastic strain
is small and its magnitude is difficult to calculate or measure adequately. The
important point to remember is that the presence of plastic deformation in the crit-
ical region is necessary for the fatigue crack to propagate and failure to occur.
When a new hardware or a machinery part is subjected to load varying envi-
ronment and is initially free from cracks, it is important for the analyst to deter-
mine the total number of cycles associated with the two stages of growth, crack
initiation and crack growth (stages I and II). Crack initiation usually is defined as
the number of cycles consumed to initiate a crack of a given length. In the aircraft
industry, the crack initiation under low cycle fatigue is selected as a crack size of
approximately 0.01 in. This corresponds to the growth of a crack by the amount of
1/10 of the notch radius (a typical hole size of 0.188 to 0.25 in. hole diameter is

2 Microscopic strain or simply microstrain is defined as the strain over a gage length comparable

to interatomic distances. Microstrain is not measurable by existing techniques (see ASTM E-6). On the
other, hand macrostrain can be measured by several methods, including strain gages and mechanical or
optical extensiometers.
68 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

used in aircraft). The remaining number of cycles associated with crack growth to
failure (for the initial crack oflength 0.01 in. to grow to failure) can then be deter-
mined by using fracture mechanics methodology. Thus, the total life would be the
sum of crack initiation life and crack growth life. The reader should note that the
establishment of crack initiation size is an arbitrary value when defining failure in
the low cycle fatigue environment. Therefore, it is expected that the number of
cycles to initiation will vary considerably.

2.8.1 Crack Initiation Life

As discussed previously, the testing carried out to establish an S-N curve for cal-
culating the total life of a given part was conducted under a load-controlled envi-
ronment. Under load-controlled laboratory tests, the two stages of crack initiation
and stable crack growth are not distinguishable because the induced cyclic stress
magnitude in the bulk of the structure is a constant under constant amplitude load-
ing; the specimen will continue to cycle until it breaks into two pieces. However,
with the strain-controlled approach, where the specimen is fluctuating under a con-
stant amplitude strain, e, the number of cycles required to initiate the crack, at the
root of a notch, can be determined through Coffin-Manson relationship (see Sec-
tion 2.12). It is recommended that the use of the S-N curve for total life evaluation
be avoided when the plastic deformation at the discontinuity is considerable or the
magnitude of the cyclic stress is above the yield stress and the bulk of the material
is plastic. The reason is that under strain-controlled conditions, in the region of
highly plastic deformation, the stress and strain are not linearly proportional.
To compute the number of cycles for a crack to initiate, the material response
to cyclic loading at the geometrical discontinuity must be assessed. It is assumed
that the material's cyclic stress-strain behavior for a smooth specimen, tested
under strain-controlled conditions, can provide useful information to assess the
local material response at the geometrical discontinuity [35, 37]. In other words,
one can relate the fatigue life of a notched part to the life of a smooth specimen
that is cycled to the same strain level at the geometrical discontinuity contained in
the part (the similarity or similitude concept) (Fig. 2.37). The local stress and
strain at the geometrical discontinuity can be related through cyclic stress-strain
curve exhibited as a series of hysteresis loops generated under strain-controlled
conditions. The local stresses and strains can then be related to remotely measured
loads, strains, and stresses by the Neuber relationship.
In Sections 2.9 and 2.10, the engineering and true stress-strain curves are
briefly reviewed for the reader to understand the concept and derivation of the
cyclic stress-strain curve from the hysteresis loops method. In Section 2.11, we
discuss how the cyclic stress-strain curve can be utilized by the analyst in dealing
with the strain-life evaluation of notched parts subjected to fluctuating load
through the Neuber relationship.
 2.9 The Conventional or Engineering Stress-Strain Curve 69

Applied Strain, £'

Smooth
Specimen

Applied Strain, E

Applied Strain, £'

(The crack initiation life for the two above cases is the same
if both experience the same strain, £')

Figure 2.37 Material cyclic response for two specimens, with and without disconti-
nuity (illustrating the concept of simulitude)

2.9 The Conventional or Engineering Stress-Strain Curve

The conventional or engineering stress-strain curve can be obtained by conduct-

ing a simple uniaxial test on standard specimens (ASTM E8, E8M). The tension
test provides useful information on the strength of materials. In a uniaxial tension
test, the specimen is subjected to monotonic increasing load, while its elongation
and corresponding load are simultaneously plotted. All the parameters extracted
from a typical engineering stress-strain curve are based on the original unloaded
gage length and the original cross-sectional area of the test specimen. The engi-
neering stress, u, is defined as the applied load, P, divided by the original cross
sectional area, Ao. The engineering strain, e, is defined as the amount of elonga-
tion, ill = Cf - Co' divided by the original length, Co. A typical engineering
stress-train curve is shown in Fig. 2.38.
The yield stress is defined as the stress associated with the 0.2% offset strain
level and is the limit of elastic behavior. The maximum stress on the curve is
called the ultimate stress and is designated by u ul ' The ultimate stress can be cal-
culated by dividing the maximum load (recorded in the chart) by the original
cross-sectional area. The region to the left of the yield stress is called the elastic
region, where stress and strain are linearly related.
The region to the right of the yield stress, up to the ultimate stress, has uniform
straining, but the stresses and strains are not linearly related. The region from the
ultimate stress, where the necking starts, up to the fracture stress, where the mate-
rial breaks into two parts, displays nonuniform straining. Another important point
on the engineering stress-strain curve is the fracture stress, uf . It can be obtained
70 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Ultimate Tensile Strength

¥
III
III
!
+oJ
en
en
c
''::: Fracture Stress
CI)
CI)
c
'0,
c
w

Engineering Strain

Figure 2.38 Typical engineering stress-strain curve

by dividing the final load recorded on the chart by the original cross sectional
area. The strain associated with the fracture stress, (J'f' is related to the reduction of
the area (RA) and it can be obtained by measuring the total elongation (by putting
the two fractured pieces together) divided by the original crack length, see Appen-
dix B for more information on this topic.

2.10 The Natural or True Stress-Strain Curve

and Hysteresis Loop

2.10.1 The Natural or True Stress-5train Curve

The natural or true stress, (J'p is found by dividing the load, P, by the actual cross-
sectional area, A, at a particular point in time (instead of the original cross-sec-
tional area, Ao). Generally speaking, the analyst likes to compute the stresses
using the original cross-sectional area, which is a measurable quantity. The use of
the engineering stress is acceptable because, in engineering applications, the com-
puted stress is required to be in the elastic range to avoid permanent deformation
and to maintain structural integrity. In the elastic range, little difference can be
seen between the changes in the cross-sectional area.
In the region of uniform straining, in which the specimen has undergone a con-
siderable amount of elongation and reduction in area, the computed true stress is
always larger than the engineering stress. In the region of necking, the cross-sec-
tional area decreases rapidly, while the computed true stress increases as shown in
Fig. 2.39.
The true fracture stress corresponds to the highest point on the true
stress-strain curve and it can be obtained by dividing the final load by the final
cross-sectional area. The true strain is based on the instantaneous gage length, e,
 2.10 The Natural or True Stress-Strain Curve and Hysteresis Loop 71

True Strain

Figure 2.39 Typical true stress-strain curve

and is defined as et = In (fifo) or et = In (1 + e) for strain up to necking (where

e is the engineering strain) . For strain above the necking up to fracture, where the
gage length reading is not accurate, et = In (Ao I A).
In the elastic region, the strain and stress are related by the proportionality con-
stant called the modulus of elasticity, E:

(et)e = atlE (2.11)

and the true plastic straining (et)p can be written in terms of the power law as:

(et)p = (atIK)l/n (2.12)

where the quantities K and n are called the strength and work hardening coeffi-
cients, respectively. The total true strain, (et)r, is the sum of the elastic and plas-
tic strain:
(2.13)
To maintain simplicity and to avoid confusion, the subscript t shown in Eq. (2.13)
that represents the true stress, at' and true strain, ef' will be removed and replaced
by a and e throughout the remaining part of this chapter when dealing with the
cyclic stress-strain curve and the strain-life prediction method. True stresses and
strains in terms of engineering stress-strain curve are discussed in Appendix B.

2.10.2 Hysteresis Loop

If we decide to load and unload a standard tensile specimen (R = 0) below its
elastic limit stress where the bulk of the specimen is purely elastic, the generated
stress-strain loop becomes a straight line with slope equal to the modules of elas-
ticity of the material, E. By loading and unloading beyond its elastic limit for a
72 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Loading

o~~~----~----------~~
True Strain

Figure 2.40 Loading and unloading a standard specimen along the true stress-strain
path

few times, the true stress-strain curve will follow the path shown in Fig. 2.40.
From this figure, it can be seen that, upon loading and unloading to zero stress
level, the path OABCDEF will always follow the original curve that is associated
with the monotonic true stress-strain curve. Repeated loading beyond the elastic
limit will cause elevation of the yield strength of the material, as illustrated in Fig.
2.40. The work hardening exponent, n, used in Eq. 2.12 to represent the original
stress-strain curve (shown by points OAF) is no longer valid to describe the
stress-strain curve shown by BCEF. A new hardening exponent is, therefore,
required to represent the new stress-strain curve BCEF. The elevation of the yield
strength, when unloading along the line Be, can be described as the result of
residual compressive stresses left behind upon loading beyond the elastic limit
and unloading due to the previous loading step (as shown in Fig. 2.40 by OAB).
Bauschinger [38] observed that, upon unloading to zero stress level and on into
compression to a stress equal to -(Tmin = -(Tmax corresponding to point C, the
yielding will occur before -(Tyield is reached, as shown in Fig. 2.41. The same
behavior was observed by reversing the direction of applied cyclic load. That is,
by loading first into compression -(Tmin and then reloading from -(Tmin to stress
level (Tmax' the yield would take place before (Tyield is reached. This behavior was
first noticed by Bauschinger and is known as the Bauschinger effect. This behav-
ior is expected because the residual compressive stresses left behind causes yield-
ing in compression to occur earlier than it would otherwise. The Bauschinger
effect does not indicate that the yield strength in compression or tension occurs at
different stress levels when cyclic loading is applied. It was later realized that the
reverse yielding due to the Bauschinger effect takes place at 2(Tyield from (Tmax or
-(Tmin' as shown in Figure 2.42 (G. Masing, 1923 [39]).
 2.10 The Natural or True Stress-Strain Curve and Hysteresis Loop 73

Stress 0" max

0" Yield -...

Strain

-0" max
C

Figure 2.41 Loading and unloading into compression

Stress

1
0" max

Strain
Aa

1
LElastic

-0" max
I~ AE ·1
ti £ = (ti £) Elastic + (ti £) Plastic

Figure 2.42 Hysteresis loop generated as a result of a complete cycle

If we now decide to apply one load cycle by unloading the specimen from
point A, corresponding to stress u m• x ' to the zero stress level and on into com-
pression (umin ) and load back up to the point A (beyond the elastic limit), a com-
plete hysteresis loop is created (see Fig. 2.42). The total width and height of the
hysteresis loop are described by the total strain range, de, and the total stress
range, du, respectively. The total strain range, in terms of its components, can be
written as:

(2.14)
74 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

where (Lle)e = /).a/E. The shape of the hysteresis loop depends on the thermo-
mechanical process given to the material. For ductile material (for example, steel
in the fully annealed condition), the width of hysteresis loop, representing the total
strain, /).e, is much wider than for high strength steel with little ductility. More-
over, the area within a hysteresis loop is related to the energy dissipated per unit
volume in one cycle.
The procedure for predicting the fatigue life consists of two parts: First, we
need to have the cyclic stress-strain curve that is obtainable from a set of stable
hysteresis loops, and second, a mathematical relationship that represents the total
strain to number of cycles to failure must be developed [40]. The cyclic
stress-strain curve and the power law representation of the cyclic stress-strain
curve is fully presented in Section 2.11 and in Section 2.12 the strain-life predic-
tion models are discussed.

2.11 Cyclic Stress-Strain Curve

To obtain the cyclic stress-strain curve for a given material, several fatigue tests
on a series of smooth highly polished specimens are conducted. These tests are
performed under strain-controlled conditions, with stress ratio of R = -1 and are
subjected to different strain rate amplitude. The cyclic stress-strain curve is the
line drawn through tensile tips of these stabilized hysteresis loops of different
strain amplitudes, plotted in the stress-strain coordinates shown in Fig. 2.43 [34].
For illustration purposes only, three points (each represents the tip of a stable hys-
teresis loop) are shown in Fig. 2.43 to demonstrate the method of obtaining the
cyclic stress-strain curve. Such a curve can be obtained from several specimens or
from one by using the incremental step test procedure [35].
Under strain-controlled conditions, a material's response to cyclic loading
varies depending on the nature of the heat treatment and its initial mechanical con-
dition (called thermomechanical history). In general, annealed materials will
harden when subjected to strain-controlled cyclic conditions. The hardening effect
can be observed in the hysteresis loop as an increase in the stress range, until the
constant amplitude range is reached (Fig. 2.44). On the other hand, a material with
prior cold-working will have a tendency to become soft when undergoing cyclic
straining. This is indicated by the decrease in the stress range (Fig. 2.45), until it
reaches its minimum value. Changes in stress response occur rapidly, as shown in
Figs. 2.44 and 2.45 and become reasonably stable (steady-state or stabilized con-
dition) after cycling about 10% to 20% of the total life [41-44]. In some material,
the hardening or softening effects are not observed, and the steady-state (stable
loop) condition is maintained when subjected to cyclic loading (Fig. 2.46). It
should be noted that the cyclic stress-strain curve is generated when the hystere-
sis loop achieves its steady state condition. By comparing the cyclic stress-strain
 2.11 Cyclic Stress-Strain Curve 75

0"3
0"2
(3)
(2)
CII
"C
::l O"J
.~ (1)
C.
E
<{
1/1
1/1

...
I!!
(I) £, £2 £
L -_ _~______~~______~3~~~

Strain Amplitude

Figure 2.43 Illustration of the cyclic stress-strain curve and the method for obtaining it

Strain Control Condition Response Variable

5
3 5
+ 1-"'"""--,..-----:..--

2 4

Cyclic Hardening

Figure 2.44 Schematic illustration of cyclic hardening

76 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Strain Control Condition Response Variable

+
3 5 3 5
+1-..".---..----..--
TIme
0"

2 4

1
3

2,
Cyclic Softening

Figure 2.45 Schematic illustration of cyclic softening

Strain Control Condition Response Variable

3 5 3 5
+1-2-----..------:.__

2 4 2 4
',3,5

Stable Loop

Figure 2.46 Schematic illustration of stable cyclic loop

curve with the monotonic stress-strain curve it can be concluded that material will
soften or harden upon the application of cyclic load if they fall below or above the
monotonic curve respectively (Fig. 2.47). For example, the cyclic stress-strain
curve behavior for the SAE 4340 steel exhibits the softening effect when com-
pared with the monotonic curve [45] as shown in Fig. 2.48.
 2.11 Cyclic Stress-Strain Curve 77

o Cyclic Stress- Strain

9 Cyclic Stress- Stra in

• Monotonic Stress- Strain

Strain

Figure 2.47 Hardening and softening behavior due to cyclic loading compared with
the monotonic stress-strain curve

For materials that exhibit cyclic softening, it is important to realize that upon
softening, the strength properties of the material are less than what was antici-
pated by the analyst. Therefore, it may be necessary for the designer to consider
the softening effect of the material if a low-cycle fatigue environment is expected.
In general, soft materials, such as aluminum alloys, have a tendency to undergo
cyclic hardening. On the other hand, hard and strong materials, such as high
strength steel, become soft. This general rule can be described in terms of the
Sul/Syield ratio, where Sui and Syield are the tensile and yield strength of the mater-
ial, respectively [45].

(material will harden) (2.15)

and

For Sui / Syield < 1.2 (material will soften) (2.16)

For ratios 1.2 < Sui / Syield < 1.4, hardening or softening may occur.
The softening and hardening phenomena due to cyclic loading may be
explained in terms of dislocation density. In soft materials usually after annealing,
the dislocation density is low and it increases due to cycling. The interaction of
dislocations as the result of density increase will cause the hardening effect. On
the other hand, for an initially hardened material (usually after cold working), the
dislocation density is high and it decreases due to cycling.
78 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

CYCUC STRESS-STRAIN CURVE

200.0

v
~ -- - I- ......

V
175.0 l/
)

I
150.0
/

....
IJ")
125.0
IT

.
~

IJ")
IJ")
w 100.0
Q::
I-<
IJ")

75.0

MenaIonic

Vc,*
50.0

25.0
~ w_
(UOIHNI

0.0
0.000.01 0.02 0.03 0.01 0.05 0.06 0.07 0.08 0.09 0.10
STRAIN, IN.lIN.
Figure 2.48 Cyclic stress-strain curve (courtesy of Douglas Aircraft Company) and
monotonic stress-strain curve for 4340 steel (softening effect), taken from [45].

2.11.1 The Power Law Representation of the Cyclic

Stress-5train Curve
Just like the monotonic stress-strain curve that was described by the Ram-
berg-Osgood relationship, the cyclic stress-strain curve can be expressed mathe-
matically by a power law [43]:
 2.11 Cyclic Stress-Strain Curve 79

BT = alE + (a/K')i/n' (2.17)

where the total true strain, BT , is given in terms of its elastic and plastic compo-
nents, BT = Be + Bp' The quantities K' and n' are the cyclic strength coefficient
and strain hardening exponent, respectively. For most materials, the hardening
exponent, n', varies between 0.1 and 0.2 [41]. If the metal is initially hard with
low n value, the cyclic behavior will lead the metal to soften and cause n' to
increase. On the other hand, for initially soft material, where n is high, the cyclic
behavior cause n' to decrease.
From Eq. (2.17), the plastic component of the cyclic stress-strain, in terms of
the power law, can be described as [43]:

(2.18)

In terms of stress:

(2.19)

The total width aB and length auof a hysteresis loop always corresponds to a point
on the cyclic stress-strain curve [Eq. (2.17)] but differ by a factor of 2. That is:

(2.20)
(2.21)

Therefore, an expression for the tips of hysteresis loops, in terms of the power
law, can be written by utilizing Eqs. (2.20) and (2.21) as:

aB = au/E + 2(au/2K')i/n' (2.22)

Figure 2.49 is the experimentally obtained cyclic stress-strain curve for 4130
steel alloy.

Example 2.6
The cyclic stress-strain curve for a given aluminum alloy, in terms of the power
law, is given as:

Generate the hysteresis loop for the fully reversed case and under strain-con-
trolled conditions, with strain magnitude of 0.04. Assume that stabilization
already has occurred and the loop is stable.
80 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

......
10-

180.0
,.
,.
160.0
v

140.0
l.t'

120.0
I

in
i
lIOi:
100.0

~
80.0

60.0

40.0

20.0

0.0
0.0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
Strain, IN'/IN.

Figure 2.49 Cyclic stress-strain curve for 4130 steel (courtesy of Douglas Aircraft
Company)

Solution
From Eqs. 2.20 and 2.21, any point (e!, O'J) on the cyclic stress-strain curve cor-
responds to (ae!, aO'J) on the hysteresis loop. The portion of the strain-time cycles
under strain-controlled conditions, where ae = 0.04, is illustrated in Fig. 2.50.
The value of the stress belonging to the first point (1) can be calculated from
the cyclic stress-strain equation as:

where the value of 0' can be solved algebraically yielding:

 2.11 Cyclic Stress-Strain Curve 81

Strain Control Condition

3 5

2 4

Figure 2.50 Strain-time cycles for Example 2.6

0'1 = 4600 psi

For point (2) shown in the figure, the stress can be calculated using the equation
that describes the tips of the hysteresis loop [Eq. (2.22)], by considering the total
strain range de = 0.04, as:

0.04 = dO'/10 7 + 2(dO'/2 X 10 4 )1/0.2

Solving the above equation, the value of the stress can be calculated as:

dO' = 9200 psi

and the coordinate of the second point is:

e2 = e 1 - de = (0.02 - 0.04) = -0.02

while the corresponding stress becomes:

0'2 = 0'1 - dO' = (4600 - 9200) = -4600 psi

The coordinates of other points (3, 4, and 5) can be obtained through the same
procedure. The resulting stress-strain hysteresis loop, corresponding to the above
strain-controlled cycling, is shown in Fig. 2.51.

(J 1,3

Stable Loop

Figure 2.51 The resulting stress-strain hysteresis loop for Example 2.6
82 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

2.12 Strain-Life Prediction Models

For many materials, the log-log plot of alternating stress versus the number of
cycles to failure, N, can provide the analyst with a straight line relationship. It was
shown in Section 2.4 that the alternating stress, in terms of the number of cycles to
failure, can be expressed in power law form. In dealing with high-cycle fatigue,
the true fatigue strength, ua ' in terms of half cycle reversal to failure (2Nf ), can be
expressed as:

Ua = U'(2N)b
f f (2.23)

where U j is the fatigue strength coefficient. Its value is equal to the monotonic true
fracture stress, OJ, corresponding to 2Nf = 1. The value of the true fracture strength,
OJ, when N = 1 is larger than the engineering fracture strength of the material (fmal
load divided by the final cross-sectional area). When the fatigue strength coeffi-
cient value is not available to the analyst, an estimate of uf = UUI + 50 ksi can be
used (applicable to ferrous material only) [34], see also Appendix B.
The use of 2Nf reversals to failure, instead of Nf cycles to failure, is common
practice throughout the literature. Imagine a test specimen that is subjected to a
monotonic load up to failure. The test specimen can be regarded as having the
shortest possible life. The failure process as a result of a single load application to
failure is considered as a half cycle, N. Thus, based on this notation a complete
cycle or full reversal can be designated as 2N.
Equation (2.23) was derived by Basquin [13], who in 1910 first proposed the
straight line relationship between the true cyclic stress amplitude and the number
of cycles to failure. The quantity b is called the fatigue strength exponent
(Basquin's exponent) and, for most materials, it varies between -0.05 and -0.12,
with typical values between -0.085 and -0.1 [46,47]. In the elastic range, the
quantity u a ' described by Eq. (2.23), can be replaced by the elastic strain ampli-
tude, Be' as:

(2.24)

The plastic strain amplitude, Bp ' is also related to the number of half-cycle reversals
to failure, 2Nf (or number of cycles to failure Nf ) by a simple power law equation:

B
p
= B'f (2N)C
f (2.25)

Equation (2.25) is called the Coffm and Manson law [42,43]. The quantity Bj is
called the fatigue ductility coefficient and its value is equal to the true monotonic
fracture strain, Bf . The fatigue ductility coefficient, in terms of reduction in area
(RA), can be expressed as:
 2.12 Strain-Life Prediction Models 83

eI
f
= In [I 1
I - RA
(2.26)

The exponent C in Eq. (2.25) varies between -0.5 and -0.7, with a typical value
of -0.6 [42,43].
In Eqs. (2.24) and (2.25), the constants b, a;, e; ,
and c are all considered to be
material properties. Examples of elastic and plastic strain-life curves for 4340
steel, described by Eqs. (2.24) and (2.25), are shown in Figs. 2.52 and 2.53,

104 ,..---+---+---+--+---+----.,f-o-.,
w
Cl d f = crf = 174 (FATIGUE STRENGTH COEFFICIENT)
=>
j 10 3
a.
~
~ 10 2
w
....--
1
a:
t;;
«stl 10 1 FATIGUE STRENGTH
EXPONENT ~ SLOPE = b = ·0.09

~ 100 1...--.....- .....- -......- .....- ---+-----'

......

10° 101 1()2 1()l 1()4 lOS 1(J5 HY

REVERSALS TO FAILURE, 2Nf

Figure 2.52 Fatigue strength-life plot for annealed SAE 4340 steel (reproduced from
[45])

~ 10 0
=>
I-
::i
a. 10- 1
~
z
~ 10-2
t;;
u
~ 10-3 FATIGUE DUCTILITY
a. EXPONENT - SLOPE = c =-0.57

~
W
10-41...--.....-~--+_-~_~--~~~
<l 100 101 10" 1()l 1()4 lOS 1(J5 HY
REVERSALS TO FAILURE, 2Nf

Figure 2.53 Fatigue ductility-life plot for annealed SAE 4340 steel (reproduced from
[45])
84 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

respectively [45]. Combining the elastic and plastic components of the total
strain-life amplitude described by the Basqin and Coffin-Manson relationship
[Eqs. (2.24) and (2.25)], the total strain-life curve can be expressed as [45]:

(2.27)

Equations (2.24) and (2.25), together with Eq. (2.27), are each plotted in log-log
form and are shown schematically in Fig. (2.54). In this figure the reversal to fail-
ure cycles, 2NI , is the dependent variable and the strain amplitude, B, is selected as
the independent or controlled variable. The term controlled variable is used here
to indicate that the total strain is controlled throughout the test duration between
its maximum and minimum assigned control limits. Figure 2.55 is the strain-life
curve developed for 4340 steel by plotting the elastic and plastic components of
Eq. (2.27), as shown in Fig. (2.52) and (2.53), respectively.
For most metals, the strain-life curve described by Eq. (2.27) can be developed
by testing at different strain amplitudes between ±2.0% and ±O.2% percent [48,
49]. Usually between 9 and 13 specimens are required to obtain the low-cycle
fatigue strain-life curve of a metal [49]. It is preferable to have low cycle fatigue
tests under strain-controlled condition and under axial reversed loading in order to

High Cycle Fatigue

cr

C)
o
.....,
....J

cu
"C
:::l
."!: (f , IE
E f
« (2Nf =1)
c::
.jij
l:i Transition Point I
en (103 to 104 I
Cycles) "

Reverals to Failure, 2N f (Log Scale)

Figure 2.54 Illustration of the total strain-life curve

 2.12 Strain-Life Prediction Models 85

w 100
c !lFJ2 = !l£p/2 + craIE
::J
!:: = 0.58 (2NffO.57 + 0.0062 (2Nf )"0.09
~ 10-1
~
z
JI
< 10-2 From Fig 2.57
~
...J
I----------------~~~~:a==a.~_.r
~ 10-3
~ From Fig 2.56
~ 10-4 ..._ ...._ .....___...._ ......_ _......_ ....._;;a".....

100 101 102 10'3 1()4 10S 16 a

1
REVERSALS TO FAILURE, 2N f

Figure 2.55 Total strain-life plot for annealed SAE 4340 steel (reproduced from [45])

avoid complexities due to the presence of non-uniform stress distribution through

the thickness (low cycle fatigue test by reverse bending, see "Engineering Mate-
rials Evaluation by Reverse Bending" by M. R. Gross). Moreover, the results of
axial reversed loading can easily be compared with the static properties of the
metal generated under monotonic tension load. The number of cycles to failure for
specimens under strain-controlled tests correspond to initiation of a crack usually
between 0.01 to 0.05 in. in length. The crack initiation is detected from the test
machine when the load drops after stabilization cycles by 5% to 10%.
Based on the ASTM E-606 Standard, the definition of failure under strain-con-
trolled conditions (depending on the use of the fatigue life test data) can be
defined as follows:

1. Total separation or fracture of the test specimen into two pieces.

2. Failure occurs when the ratio of modulus for unloading, E UN ' to loading, E L , reaches
half of the value corresponding to the first cycle in the beginning of the test. The
number of cycles when I /2( EUN / E L ) is reached is designated as the number of
cycles to failure, Nf .
3. Failure is defined as occurring when the maximum applied load is decreased by 10%
because of the presence of a crack or several cracks.
4. The number of cycles associated with the presence of a surface crack detected visu-
ally which is equal to or larger than some predetermined size specified by the user
of the data.

Differences in reversal to failure, Ni , in low-cycle fatigue test are expected when

different failure conditions are employed (see definitions above provided by the
ASTM Standards). For this reason the supplier of the eoN data must indicate the
failure definition used when the final results are submitted to the user.
86 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

In real life when damage is the result of different strain amplitudes, the sum of
the all the damage accumulated by the environment on the part based on Miner's
rule can be written as:

D =I nJNfi = 1.0 (2.l0a)

whereNfi is defined as the number of cycles to failure (the initiation of a detectable

crack or total failure) at a given strain amplitude, e j • In aircraft applications, when
the structural part is subjected to a high fatigue load environment, the failure is
associated with the accumulation of all the cycles for the initiation of a detectable
crack around 0.01 to 0.05 in. in size. If the structure under consideratin is now
subjected to additional load cycles with load amplitude much smaller than previ-
ous cycles, the remaining fatigue life may be determined through linear elastic
fracture mechanics.

2.12.1 Transition Point, NT

The intercept of the two straight lines at 2Nf = 1 are ej and uj IE for the plastic
and elastic components of Eq. (2.27), shown in Fig. 2.54. The slopes of the two
lines, b and C, are shown separately in the same figure. It can be seen that the two
curves, corresponding to the elastic and plastic components of Eq. (2.27), are
intersecting at a point called the "transition point." For a number of cycles less
than the transition point, the plastic component dominates; this represents the low-
cycle regime (short lives). For a number of cycles longer than the transition point,
the elastic component dominates the plastic one; this represents the high-cycle
regime (long lives). At the intercept of the two lines, the plastic and elastic strain
amplitudes are the same. The plastic component is governed by the ductility,
defined in terms of the reduction of the area, described by Eq. (2.25), and the elas-
tic component described by the term uj IE.
The number of cycles to failure associated with the transition point, NTo can be
obtained by equating Eqs. (2.24) and (2.25). At the transition point, the quantity
Nf is replaced by NT:

Solving for 2NT (reversals):

2NT = [(ejE)/ujrl/(c-b) (2.28)

For most materials, the number of cycles to failure associated with the transition
point, 2NT , falls between 10 3 and 10 4 cycles. The experimental data shows that
most metals have almost the same reversal to failure at strain amplitude of ±0.01
(de = 0.2) [17].
 2.12 Strain-Life Prediction Models 87

The selection of material to assess fatigue problems is strongly related to the

magnitude of the fluctuating load. In high-cycle fatigue problems, where the nom-
inal stress is in the elastic range, the selection of materials with high tensile
strength and lower ductility, or ductile material with surface treatment (such as
shot peening, carburizing, or nitriding) is appropriate. Thus, for a high-cycle
fatigue design life requirement, where reversal to failure is above 10 3 to 10 4 cycles
(Nf > NT)' it would be appropriate to select a high strength material. On the other
hand, for fatigue life requirements below the transition point, where reversal to
failure is lower than 10 3 to 10 4 , one would pick a ductile material [50, 51].
The following example problem will assess the type of surface treatment
needed when low or high cyclic loads are applied to a given material selected for
use in a structural component.

Example 2.7
A part is made of2014-T6 aluminum alloy with the following fatigue properties:
Modulus of elasticity: lOE + 7 psi
Tensile strength: 55,000 psi
Fatigue ductility coefficient: 0.40
Fatigue strength exponent: -0.10
Fatigue ductility exponent: -0.65
It is recommended by the customer to surface treat the part for better fatigue prop-
erties. Determine whether this treatment is necessary for a load environment where
the part is cycling under a fully reversed constant strain amplitude of 0.018 in./in.

Solution
We would like to know if the applied fully reversed constant strain amplitude of
0.018 is associated with low- or high-cycle fatigue. By employing Eq. (2.28), the
total number of cycles at the transition point can be computed for the 2014-T6
aluminum alloy:

2NT = [(eJE)/O"j]-I/(C-b)
[(10 7 X 0.4)/55,000r 1/(-0.65+0.1)
2NT = 2358 reversals
The total strain at the transition (where 2NT = 2358 reversals) can be computed by
using Eq. (2.25) or (2.24), since, at the transition point the two equations are equal:

ep = eJ (2Nf ),
= 0.4(2358)-065
ep = 0.00256
88 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

The total strain at the transition point is smaller than the fully reversed applied
constant strain amplitude of 0.018; thus it falls under the low-cycle fatigue classi-
fication. For this reason, the surface treatment is not recommended. It should be
noted that increasing the strength of materials through cold working or heat treat-
ment may improve fatigue properties in the high-cycle fatigue regime (make them
worse in the low-cycle fatigue environment).

2.12.2 The Mean Stress Effect

Most of the available fatigue data are collected in the laboratory through testing by
applying the fully reversed loading case (R = -1), where the mean stress, SM' is
zero. In real situations, where SM is not zero, the influence of the mean stress on the
fatigue life of the part cannot be addressed through Eq. (2.27), where R = -1.
However, Eq. (2.27) and the strain-life curve shown in Fig. 2.55 can be corrected
to account for the effect of the mean stress, SM' A correction for the mean stress can
be accomplished by simply adding the mean stress to (if SM > 0) or subtracting it
from (if SM < 0) the strength coefficient, a;
,ofthe elastic part ofEq. (2.27) [48]:

Be = (a; - SM)/E(2Nf )b
(2.29)

Another equation that is available to correct the effect of the mean stress on the
strain-life prediction method was suggested by Smith-Watson-Topper [52]. This
approach is based on the strain life test data obtained for various mean stress val-
ues. They considered the effect of the mean stress on the predicted life through the
maximum stress, where Smax = S M + Sa' The Smax of a cycle for the case of
R = -1 (equation 2.33) was multiplied by the strain-life equation, e, described by
equation 2.27.
Smax e = a,2/E(2N
f f
)2b + a'f e'f (2Nf )c+b (2.30)

2.12.3 Fatigue Notch Factor, K,

Fatigue failure of structural parts during their service life always has its origin at
some surface discontinuity, sharp comer, hole, or the root of a notch, where mate-
rial has deformed locally. For a given nominal stress, the presence of a notch will
create stress concentration at the discontinuity and reduce the service life of the
structure. The ratio of the fatigue strength of an unnotched part (specimen with no
stress concentration) to the fatigue strength of a notched part at N number of
cycles (under the same environment) is designated by Kf and is called the
"Fatigue Notch Factor or Fatigue Stress Concentration Factor." It is also called
the Fatigue Strength Reduction Factor [53]:

Kf = (Fatigue Strength)unnotched/(Fatigue Strength)notched (2.31)

 2.12 Strain-Life Prediction Models 89

~ $, Unnotched S-N Curve with K t= 1

III

I
III
~
+"'
II)

~ 52
·eo --- - +- Corrected Curve
z ~ Notched S-N Curve
with Theoretical
L..._ _ _ _ _ _ _ _ _ _ _ _ _ _• Stress Concentration
Nf Factor, K t

Number of Cycles To Failure, N

Figure 2.56 S-N curves for two cases of notched and un notched conditions

The quantity Kf varies with material, type of notch, and loading. Figure 2.56 illus-
trates the S-N curves for a given material developed for two cases of notched and
unnotched conditions where Kf = S11S 2 corresponding to Nf number of cycles
(note that in figure 2.56 the nominal stress, Sa' is plotted against fatigue life cycle,
N, on a Logarithmic scale). For example, the value of fatigue notch factor, Kf , at
5 X 10 6 cycles (for specimen tested in rotating bending test machine where
Kt = 1.6) are given as 1.0 and 1.6 for Gray Iron and hard steel respectivelly [54].
In terms of the theoretical stress concentration factor of a monotonic tension load
in the elastic range, Kp the fatigue notch factor, Kf , can be approximated [53] as:

K - 1
K=l+-----'t-- (2.32)
f 1+ air
where the quantity r represents the notch root radius and a is a material constant
given in units oflength and is dependent on the material's strength and ductility. In
Figure 2.56, the notched curve is calculated from the unnotched S-N curve by di-
viding each stress point on the curve by the theoretical stress concentration factor,
Kt • A less conservative curve can be obtained by using the value of Kf , from Eq.
(2.32), in place of Kt , for life estimation when long life is expected on the structur-
al part, see the corrected dotted curve shown in Figure 2.56. In Eq. (2.32), the in-
verse of the quantity in the denominator is called the notch sensitivity factor, q:

1
q=---
1 + air

The estimated value of a for heat-treated steel is given by Peterson [33] as:
90 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Table 2.12 Material constant for the Quantity a

(average values)

Material a, inches

Aluminum alloy
2024-T4 Sheet 0.05
bar 0.008
7075-T6 Sheet 0.02
bar 0.003
Steel
Annealed and Normalized 0.01
Quenched and Tempered 0.0025

300000)1.8
a = ( ' X 10- 3 (in inches) (2.33)
SUI (psi)

where SUI is the ultimate of the material. Some average values of a for 2024-T4
and 7075-T6 aluminum alloys and steel are shown in Table 2.12 [53].
From Eq. (2.32) it can be seen that the value of Kf varies between Kf = 1 (no
notch effect when r = 0) and the theoretical stress concentration factor K t (for large
notch with large r). When sufficient data are not available, the value of Kf = K t can
be used where a large number of cycles are expected. On the other hand, the Kf val-
ue is lower, compared to the theoretical stress concentration (Kf < K t ), when short
fatigue lives are expected. In general, designing for fatigue using theoretical stress
concentration factor, Kt , is safe but conservative when the experimental value of Kf
is not available to the analyst.

Example 2.8
A plate that is made of steel (in quenched and tempered condition) is subjected to
cyclic load, as shown in Fig. 2.57. The plate is subjected to fully reversed cyclic
loading with amplitude of 18.0 kips for 2 X 10 5 cycles. Determine the factor-of-
safety with respect to applied stress (Sa) and life (2Nf ). The stress-life relation-
ship in high cycle fatigue (Nf > 10 3), is given as:

Solution
The S-N curve, from the stress-life (S-N) equation described above, is shown in
Fig. 2.57. The theoretical stress concentration factor, Kt = 3, for a hole in a plate
 2.12 Strain-Life Prediction Models 91

W = 4.0 in. D = 0.5 in. P(t)

t = 0.25 in.

tV
10 3
b = -0.088

300
,- (J 'f = 295 ksi

Sa= 295 (2N f) -0.88

111
V)
200 W
as
'C
::J
.t!
C.
E
< 10 2 Sa= 256 (2Nf) -0.88
III
III "
....!!! "
V)

,-...
....
~ Time
Q..

10 102 103 104 105 106 107

Reversals to Failure, 2Nf

Figure 2.57 Steel plate with a hole subjected to cyclic loading for Example 2.8

and the fatigue notch factor, Kf , in tenus of the theoretical stress concentration
factor (Kt ) from Eq. (2.32), is:

K - 1
Kf =1+ 1 ~ air

where a = 0.0025 (for quenched and tempered steel) and r = 0.25 in. Using
K t = 3, the value of Kf = 2.980. The stress amplitude, (Ja' at the stress concen-
tration can be calculated as:

where

A = (W - D)t = (4 - 0.5) X 0.25 = 0.875 in. 2

(Ja = 2.980 X (18.0/0.875) = 61.30 ksi

For 2Nf = 2 X 10 5 cycles, the fatigue strength from the S-N diagram shown in the
Fig. 2.57 is:
92 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

Sa = 100.77 ksi
The computed factor of safety, FS = 100.77/61.30 = 1.64 (with respect to
applied stress). The total number of cycles to failure associated with U a = 61.30
ksi can be computed as:

Ua = 61.30 ksi = 295 (2Nf rO.88

(2Nf ) = 6 X 10 7 cycles
Nf = 3 X 10 7

The computed factor-of-safety, FS = 6 X 107/2 X 10 5 = 300 (with respect to life)

Example 2.9
Repeat Example 2.8 for the case of cyclic loading with amplitude of 18.0 kips and
a mean stress value of 11.5 kips.

Solution
The same procedure should be applied. The mean stress, uM ' can be computed
the same way as u a ' and is given by:

UM = KAP(t)/A]
2.980 X (11.5/0.875)

UM = 39.16 ksi

The S-N curve, corrected for the mean stress, is described by Eq. (2.29) to include
the effect of the mean stress:

Sa = (u; /E - uM)(2Nf )b
(295 - 39.16)(2Nfro.088

255.84(2Nf ) -0.088

The S-N curve for the mean correction is shown in Fig. 2.57 as the dotted line.
The fatigue strength for 2Nf = 2 X 10 5 cycles is found from the above equation:

Sa = 255.84(2 X 10 5)-0.088

Sa = 87.39 ksi
 2.12 Strain-Life Prediction Models 93

The computed factor of safety, FS = 87.39/61.3 = 1.42 (with respect to applied

stress). The total number of cycles to failure associated with (Ta = 61.30 ksi can
be computed by:

61.3 = 255.84(2Nf r 0088

2Nf = 1.2 X 10 7 cycles

The computed factor of safety, FS = 7 X 10 6 /2 X 10 5 35 cycles (with

respect to life)

2.12.4 Stress and Strain at Notch (Neuber

Relationship)
In evaluating the fatigue life of a part that contains stress concentration (as shown
in Fig. 2.58), the actual stress and strain at the notch must be computed. Finite
element methods (FEM) can enable the analyst to transfer the applied load to the
structure and distribute it in the form of stress across the body (the values for the
stress concentrations can also be obtained experimentally through the use of pho-
toelasticity). As a result, the stresses at every point in the body, specifically in
localized areas, such as sharp corners, holes, or the roots of a notch, can be
obtained by the FEM. In the elastic range, when the nominal, S, and local stresses,
(T, are below the yield, the theoretical stress concentration factor (K t ) is:

Nominal Stress, S

p,
(Nominal Region)
~

(Local Region)
Stress Concentration
at Localized Region, a

Figure 2.58 Nominal and local stress in a plate with a hole

94 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

0'
K=-
t S (2.34)

Under this condition, where the plasticity at the locality of the notch is absent,
there is no fatigue failure accumulated in the part. In the presence of local yield-
ing at the notch, the stress concentration can no longer be defined by Eq. (2.34).
Both the stress and strain concentrations in that locality must be addressed. The
strain and stress concentration factors (Ke' K (T) for the case of local yielding
(when applied stress is in the elastic range) can be expressed as:

0' e
K(T = S' K =--
e (SIE)
(2.35)

Based on Nueber's rule [4], the product ofthe two quantities K(T and Ke during the
plastic deformation is related to the theoretical stress concentration (in the elastic
range), Kp by:

(2.36)

Rearranging Eq. (2.36) in terms of the two quantities (0', e) described by Eqs.
(2.34) and (2.35), it follows that:

(2.37)

For a given nominal stress, S, the localized stress, 0', and strain, e, induced at the
notch can be calculated, provided that the stress-strain equation defined by Eq.
(2.13) is available for solving the two unknowns quantities (local stress and strain)
shown by Eq. (2.37). When the applied load is monotonic, the stress-strain rela-
tionship defined by Eq. (2.13) can be expressed by:

e = O'IE + (a/K)l/n

For a cyclic stress-strain curve:

eT = O'IE + (a/K,)l/n' (2.17)

Note that, in the case of cyclic loading, the cyclic stress-strain curve defined by Eq.
(2.17) can be replaced by hysteresis loops or by mUltiplying the cyclic stress-strain
curve by 2 (cyclic 0' - e X 2). Moreover, the nominal and local stresses and
strains, shown in Eq. (2.37), can be replaced by stress and strain ranges:
 2.12 Strain-Life Prediction Models 95

Intersection

I/)
I/)

...I!!
en
Stress-Strain Curve
Strain

Figure 2.59 Illustration of the graphical method of obtaining the stress and strain at
the notch

(2.38)

The graphical solution for obtaining the local stress and strain «(7', e) at the notch
is also illustrated in Fig. 2.59. Here, the intersection of two curves, described by
Eqs. (2.13) and (2.37) for the case of a monotonic loading and (2.17) and (2.37)
for a cyclic loading, respectively, will define the desired values of and (7' and e.

Example 2.10 (Monotonic Tension Load Case)

Determine the stress and strain at the stress concentration point for the applied
nominal stress of 50 ksi, shown in Fig. 2.60. The values of nand K describing the
monotonic stress-strain curve for SAE 4340 steel (annealed condition) are 0.1 and
110.0 ksi, respectively. The value of E for SAE 4340 steel is 3 X 10 4 ksi.

Solution
Using Eqs. (2.13) and (2.37), the two unknowns, (7' and e, can be computed as:

(Kt S)2
e = alE + (a/K) lin, (7'e =
E

Figure 2.60 Determination of stress and strain at the stress concentration site for
Example 2.10
96 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

From (2.37)

And from (2.13)

e = a/3 X 10 4 + (a/11O.0)1/0.1
Solving for a and e:

a = 70 ksi and e = 0.011

Employing Eq. (2.35), the stress and strain concentration can be obtained by:

Ka = 70/50

Ka = 1.4

e
K, = (S/E) = 0.011(50/3 X 10 4 r 1

K, = 6.43

where

Example 2.11 (Cyclic Load Case)

For the same problem, the cyclic load, P( t), is available and is shown in Fig. 2.61.
Properties related to the material are given in Table 2.13.
Determine the number of cycles to failure.

-10 kips

Figure 2.61 The applied cyclic load, P (t), for Example 2.11
 2.12 Strain-Life Prediction Models 97

Table 2.13 Strength and fatigue properties

Modulus of elasticity 3E + 4 ksi

Tensile strength 180 ksi
Fatigue ductility coefficient 0.626
Fatigue strength coefficient -0.080
Fatigue ductility coefficient -0.73
n' 0.16
K' 238 ksi
Cross-sectional area: 0.75 in. 2

Solution
Using Eqs. (2.17) and (2.38), the two unknowns e and (Tcan be computed as follows:

e = (T/E + ((T/K')I/n' and

6.S can be computed by:

6.S = P( t )max/A
= 20 kips/0.75 in. 2
6.S = 26.66 ksi

The local stress and strain associated with applied 6.S = 26.66 ksi can be found by
solving the two Eqs. (2.17) and (2.38):

6.O'6.e = (K t 6.S)2/E
6.mle = (3 X 26.22)2/3 X 10 4
6.O'6.e = 0.213

Using Eq. (2.17) and applying the trial and error approach to obtain 6.0' and:

6.e = 90/3 X 10 4 + (90/238) 1/0.16

6.e = 0.0023 in.jin.

6.0' = 90.0 ksi

The number of cycles to failure can be obtained by using Eq. (2.27):

At the transition point the number of cycles (reversals to failure) is:

98 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

2NT = [(e; E)/a"j]-l/(c-b)

2NT = [(3 X 10 4 x 0.625)/180/r 1/(-O.73+0.08)

2NT = 1222 cycles

The strain associated with 2NT = 1222 cycles is:

e = 180/3 X 10 4 (1222fo.o8 + 0.625(1222fo.73

e = 0.007 in./in.

This value is larger than Lle = 0.0023 and therefore falls into the high cycle
fatigue regime:

e = fJ;/E(2Nf )b
0.0023 = 180/3 X 104(2Nffo.o8

2Nf = 10,000 cycles

2.13 Universal Slope Method

The universal slope method is a useful tool for estimating fatigue life when access
to fatigue data (mainly the exponents band c of Equation 2.27) is difficult. For
this reason, a great deal of effort has been put into finding ways to utilize the
monotonic properties that are available for use in cyclic study. Given the ultimate
strength, fatigue ductility coefficient, and modulus of elasticity (fJuP e; and E)
that are obtainable from monotonic tensile testing, the total number of cycles to
failure can be calculated. Figure 2.62 illustrates the method of universal slopes.
This approach is also called Manson's method of universal slope [55] and it is
the simplified version of Eq. (2.27). The equation that describes the universal
slope method is:

(2.39)

where the exponents band c from Eqs. (2.24) and (2.25) were replaced by an
average slope value of -0.12 and -0.6, respectively. The only variables that are
effective in computing the fatigue cycles for different materials are the static prop-
e;,
erties, fJu1 ' and E that are obtainable through uniaxial tension test. Note that the
quantity e; is replaced by its equivalent quantity, the reduction of the area (RA),
described by Eq. (2.26).
 References 99

£'= {In[_I_]}0.6
~ f 1-RA
,.... 1 0.6
CII 3.S( <TullE )(2Nf)-0.12 + {In [ - - ] }
i;j 1- RA
U
VI
Cl
o
--'
'--'
CII
'0
::s
.~
Q.
E
«
en Transition

,I
en 1 0.6 06
CII Point 1 {In [ - - ]} (2Ntr'
.....
L.
1- RA
VI

Reversals to Failure (Log Scale)

Figure 2.62 Illustration of the universal slope method

It can be seen from Eq. (2.39) that the universal slope method assumes the
exponents band c to be the same for all materials. It should be emphasized that
the values assigned to exponents band c are not always -0.12 and -0.6 and that
somewhat different values have been reported for other materials. Therefore, cau-
tion must be taken in employing Eq. (2.39) for fatigue life estimation.

References

1. "Wohler's Experiments on the Strength of Metals," Engineering, August 23, 1967, p. 160.
2. A. Palmgren, "Ball and Roller Bearing Engineering," translated by G. Palmgren and
B. Ruley, SKF Industries, Inc., Philadelphia, 1945, pp. 82-83.
3. M. A. Miner, "Cumulative Damage in Fatigue," Appl. Mech., Trans. ASTM, Vol. 12,
September 1945, pp. A-159-164.
4. H. Neuber, "Theory of Stress Concentration for Shear-Strained Prismatical Bodies
with Arbitrary Nonlinear Stress-Strain Law," Trans., ASME, Appl. Mech., December
1961, pp 544.
5. Fatigue Technology Inc. (FTI), Extending the Fatigue Life of Metal Structures, Mate-
rial Testing, Seattle, Washington.
6. N. E. Dowling, W. R. Brose, and W. K. Wilson, "Fatigue Failure Prediction for Com-
plicated Stress-Strain Histories," T. A&M. Report No. 337, University of Illinois,
Urbana, January 1971.
7. T. Indo and M. Matsuishi, "Fatigue of Metals Under Random Strain," Pre print of
Japan Society of Mechanical Engineering.
100 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

8. S. J. Brodeur, and M. I. Basci, "Fracture Mechanics Loading Spectra for STS Pay-
loads," AIAA-83-2655-CP, 1983.
9. Fatigue Crack Growth Computer Program "NASA/FLAGRO" Version 2.2, JSC-
22267A, May 1994.
10. ASTM, Manual on Fatigue Testing. ASTM Special Technical Publication NO. 91.
Philadelphia, ASTM Committee E-9 on Fatigue, 1949.
11. ASTM, A Tentative Guide for Fatigue Testing and the Statistical Analysis of
Fatigue Data. ASTM Special Technical Publication No. 91-A. Philadelphia, ASTM,
1963, pp. 19-22.
12. Mil-Handbook 5, Department of Defense, Washington, D.C.
13. O. H. Basquin, "The Exponential Law of Endurance Tests," Proc. ASTM, Vol. 10,
Part II, 1910, p. 625.
14. B. P. Haigh, "The Relative Safety of Mild and High-Tensile Alloy Steels Under Alter-
nating and Pulsating Stresses," Proc. Inst. Automob. Engin, Vol. 24, 1929/1930, p. 320.
15. W. Gerber, "Besdtimmung der Zulossigen Spannungen in Eisen Constructionen," Z.
Bayer Arch. Ing. Ver., Vol. 6,1874, p. 101.
16. J. Goodman, Mechanics Applied to Engineering. Longmans, Green & Co., 1899.
17. H. O. Fuchs, and R. I. Stephens, "Metal Fatigue in Engineering," 1980, pp 69-76.
18. G. Jacoby, "Fractographic Methods in Fatigue Research", Exp. Mech., March
1965, pp. 65-82.
19. C. A. Zapffe and M. Clogg, Jr., "Fractography-A New Tool for Metallurgical
Research", Preprint 36, American Society for Metals, 1944, later published in Trans.
ASM, Vol. 34, 1945, pp. 71-107.
20. Short Fatigue Crack, edited by, K. J. Miller and E.R. de los Rios, ESIS, Publica-
tion 13, p. 31.
21. Small Fatigue Cracks, edited by R. O. Ritchie and J. Lankford, "Proceedings of the
Second Engineering Foundation International Conference/Workshop, Santa Barbara,
California, January 5-10, 1986." A Publication of the Metallurgical Society, Inc.
22. The Behavior of Short Fatigue Cracks, edited by K. J. Miller and E. R. de los Rios,
EGF, Publication 1 (Collection of Papers and References in Crack Initiation).
23. G. F. Vander Voort, "Conducting the Failure Examination," Met. Eng. Q., Vol. 15,
May 1975, pp. 31-36.
24. C. A. Zapffe et aI., "Fractography: The Study of Fracture at High Magnification," Iron
Age, Vol. 161, April 1948, pp. 76--82.
25. The Making, Shaping and Treating of Steel, Ninth Edition, edited by H. E. McGannon,
United State Steel, Pittsburgh, December, 1970.
26. R. A. Flinn, and P. K. Trojan, Engineering Materials and Their Applications, Third
Edition, Houghton Mifflin, 1986.
27. H. Mughrabi, F. Ackermann, and K. Herz, (1979) "Persistent Slip Bands in Fatigued
fcc and bee Metals," Fatigue Mechanism, ASTM STP 675, pp. 69-105.
28. W. A. Wood, "Recent Observation on Fatigue Fracture in Metals," ASTM STP 237,
(1958) pp. 110--121.
 References 101

29. P. J. E. Forsyth, The Physical Basis of Metal Fatigue, American Elsevier, 1969.
30. Achievment of High Fatigue Resistance in Metals and Alloys, edited by J. C.
Grosskreutz and C. E. Feltner, ASTM, STP-467, June 1969.
31. T. Broom, and R. K. Ham, "Hardening of Copper Single Crystals by Fatigue" Proc. R.
Soc., Vol. A251, 1959, pp. 186-199.
32. D.S. Kemsley, and M. S. Paterson, "Influence of Strain Amplitude on Work Harden-
ing of Copper Crystal in Alternating Tension any Compression" Acta Metall., Vol. 8,
1960, pp. 453-467.
33. C. E. Feltner, Philos. Magazine, Vol. 12, 1965, pp. 1229-1248.
34. R. C. Juvinall, "Supplement to Engineering Considerations of Stress, Strain, and
Strength," McGraw-Hill, 1967.
35. B. M. Wundt, "Effect of Notches on Low-Cycle Fatigue," ASTM, STP-490, 1972.
36. Fatigue Design Handbook, edited by J. A. Graham, prepared under the auspices of the
Fatigue Design Subcommittee of Division 4 of SAE Iron and Steel Technical Com-
mittee, Vol. 4.
37. S. S. Manson and M. H. Hirschberg, "Low-Cycle Fatigue of Notched Specimens by
Consideration of Crack Initiation and Propagation," NASA TN D-3146, June 1967.
38. J. Bauschinger, "On the Change of the Position of Elastic Limit ofIron and Steel Under
Cyclic Variations of Stress," Mitt. Mech. Tech. Lab., Munich, Vol. 13, No.1, 1886.
39. S. Timoshenko, "Strength of Materials", Part II, Advanced Theory and Problems,
Third Edition, Robert E. Krieger Publishing Company, 1983, pp. 393-430.
40. R. W. Landgraf, JoDean Morrow, and T. Endo, "Determination of Cyclic
Stress-Strain Curve," J. Mater., Vol. 4, No.1, March 1969, pp. 176-188.
41. R. W. Landgraf, "The Resistance of Metals to Cyclic Deformation," Achievement of
High Fatigue Resistance in Metals and Alloys, " ASTM STP 467, American Society
for Testing and Materials, 1970, pp. 3-36.
42. T. Endo and JoDean Morrow, "Cyclic Stress-Strain and Fatigue Behavior of Repre-
sentative Aircraft," J. Mater., Vol. 4, No.1, March, 1969, pp. 159-175.
43. J. Morrow, "Cyclic Plastic Strain Energy and Fatigue of Metals," Internal Friction,
Damping and Cyclic Plasticity, ASTM, STP- 378,1965, pp. 45-87.
44. L. F. Coffin, Jr. and J. F. Tavernelli, "The Cyclic Straining and Fatigue of Metals,"
Trans. Metall. Soc. of AIME, Vol. 215, October 1959, pp. 794-807.
45. R. W. Smith, M. H. Hirschberg, and S. S. Manson, "Fatigue Behavior of Materials
Under Strain Cycling in Low and Intermediate Life Range," NASA, TN D-1574,
April 1963.
46. J. F. Tavernelli and L. F. Coffin, Jr., "Experimental Support for Generalized Equation
Predicting Low Cycle Fatigue," Trans. ASME, J. Basic Eng., Vol. 84, No.4, Decem-
ber 1962, p. 533.
47. S. S. Manson, discussion of Reference 23, Trans. ASME J. Basic Eng., Vol. 84, No.4,
December 1962, p. 537.
48. J. A. Graham (Ed.), Fatigue Design Handbook, SAE, 1968.
49. ASTM E-606 Standards
102 Chap. 2 Conventional Fatigue (High- and Low-Cycle Fatigue)

50. JoDean Morrow and T. A. Johnson, Material Research and Standards, MTRSA,
Vol. 5, No.1, January 1965, pp. 30-32.
51. R. E. Peterson, Materials Research & Standards, MTRSA, Vol. 3, No.2, February
1963, pp. 122-139.
52. K. N. Smith, R. Watson, and T. H. Topper, "A Stress-Strain Function of the Fatigue of
the Metals," J. Mater., Vol. 5, No.4, December 1970, pp. 767-778.
53. R. E. Peterson, "Analytical Approach to Stress Concentration Effect in Fatigue of Air-
craft Materials," Proceeding of Symposium on Fatigue of Aircraft Structures, pp.
273-299, Wadc Technical Report 59-507, August 1959.
54. H. J. Grover, S. A. Gordon, and L. R. Jackson, "Fatigue of Metals and Structures"
Thames and Hudson, London, 1956 pp. 66-85.
55 S. S, Manson, "Fatigue: a Complex Subject-Some Simple Approximation," Exp.
Mech., Vol. 5, No.7 July 1965, p. 193.
Chapter 3

Linear Elastic Fracture Mechanics

3.1 Energy Balance Approach

(the Griffith Theory of Fracture)

The quantitative statements of the Griffith theory [1] are based on elasticity theory
and are applicable only to those materials that obey Hooke's law up to the instant
of fracture. In confirming his theory, Griffith used glass as the test material
because at room temperature glass does follow Hooke's law to the stress at frac-
ture. Griffith's work must be considered unique, because the surface tension that
appears in his theory as the energy required for fracture was determined indepen-
dently of fracture tests.
Griffith's theory [1] evolved from a consideration of the potential energy of
the system, that is, the equilibrium position of an elastic body deformed by spec-
ified surface forces is such that the potential energy of the whole system is mini-
mum. Griffith took into consideration the increase of potential energy that occurs
due to the formation of two new crack surfaces. Catastrophic crack propagation
under fixed grip conditions begins because the decrease in internal energy of the
system as the crack extends an incremental length becomes available to extend
the crack and is equal to the energy consumed in the dissipative mechanisms of
the growing crack. For brittle materials that do not exhibit plastic deformation at
the crack tip, the critical condition for through cracks can therefore be stated as:

a
oc [UE - Usl =0 (3.1)

UE is the energy per unit thickness available to create the new crack surfaces
resulting from a through crack of length 2c in a plate of infinite size under a ten-
sile stress, (T, normal to the plane of the crack. The term Us is the elastic surface

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts 103
© Chapman & Hall 1997
104 Chap. 3 Linear Elastic Fracture Mechanics

energy and is defined as the energy consumed per unit thickness in creating the
new crack surfaces. The quantity UE can be written [1] as:

(3.2)

and the elastic surface energy (Us) is given by:

Us = 4cT (3.3)

where T is the surface tension of the material, that is, the work done in breaking
the atomic bonds. At the atomic level, one can assume that the residual strength
capability of a material depends on the strength of its atomic bonds. That is, frac-
ture takes place when atomic bonds break between atoms in which two new crack
surfaces are created. Fracture at the atomic level occurs either by cleavage where
the breaking bonds are perpendicular to the fracture plane, with the theoretical
cohesive strength (J'c = E/1O, or by a process called shear across the fracture
plane, where Tc = G/1O (E and G are the elastic and shear moduli, respectively).
In terms of critical stress, (J'c' the surface energy, T, can be written as [2]:

where:

Ea
T = (0.394 X 10- 8)_0 (3.4)
20

and a o is the atomic spacing in Angstroms (lA = 0.394 X 10- 8 in.)

The equilibrium condition described by Eq. (3.1) can further be simplified by
setting the first derivitive of UE and Us [using Eqs. (3.2) and (3.3)] with respect to
crack length, c, equal to zero:

(3.5a)

and for one half of a center crack in an infinite plate:

(3.5b)
 3.2. The Stress Intensity Factor Approach 105

The expression on the left of Eq. (3.5b) is the elastic energy release rate (or the
crack extension force, G [3]) and the expression on the right side is the energy
absorption rate for creation of new crack surfaces. The critical energy release rate,
Gc ' is a constant for truly brittle materials because T is constant.
The energy balance approach described by Eq. (3.5) simply states that, for
some applied stress CT, an ideally sharp crack of length 2c in an infinite body will
become unstable. The energy balance approach is applicable only to the crack
analysis cases where the condition of crack instability is a result of monotonic
increasing tensile load to failure. However, this approach is not applicable to other
important cases, such as stable crack growth that occurs when a crack is subjected
to cyclic loading. The stress intensity parameter, K, is a useful tool that can
address the cracking problems of interest in aircraft, pressure vessels, ships, and
space vehicle structures, such as stable crack growth under fluctuating load, as
well as the condition of instability. The two quantities, energy release rate, G, and
the stress intensity factor, K, are linked together in the energy balance equation by
Irwin [3] as:

(3.5c)

The stress intensity factor approach using linear elastic fracture mechanics is
discussed extensively in the remaining part of this chapter. Later in Chapter 6 un-
der "Fracture Mechanics of Ductile Metals," the modified Griffith fracture criteri-
on is discussed, which considers the dissipated energy absorption rate that results
in the formation of plastic deformation at the crack tip of a metallic material.

3.2. The Stress Intensity Factor Approach

3.2.1 General
As mentioned in Section 3.1, the energy balance approach (expressed as the
energy release rate, G) to crack tip problems leads to the same results as the stress
intensity factor approach. The stress intensity factor, K, characterizes the crack tip
stress field and its applications to fatigue crack growth and life prediction prob-
lems make this a very important parameter in the field of fracture mechanics. For
this reason, its derivation is discussed in Section 3.2.2. The critical value of the
stress intensity factor, called "fracture toughness," is discussed in Section 3.3. In
Section 3.4, methods of constructing the residual strength diagram of cracked
structures are described. The resistance curve approach, known as the R -curve, is
introduced (Section 3.4) for obtaining the fracture toughness value of ductile met-
als with tearing fracture behavior. The failure of structural metals when subjected
to a fluctuating load environment, is in most cases, the result of having surfaces
exposed to scratches during machining, corrosion pitting, or surface damage
106 Chap. 3 Linear Elastic Fracture Mechanics

through improper handling. The behavior of these cracks and their growth can be
assessed through the stress intensity factor parameter. The derivation of crack tip
stress intensity factor for surface cracks (also called part through cracks) is pre-
sented in Section 3.7. Finally, a brief discussion of the plane strain and plane
stress fracture toughness testing is included in Section 3.8.

3.2.2 Crack Tip Modes of Deformation

The stress fields and displacement modes at the crack tip can be classified into
three types (see Fig. 3.1 for crack surface displacement in the x, y, and z directions):
I. The opening mode (mode I) is characterized by displacement of the two crack sur-
faces moving directly apart from each other. In this case, the applied load and dis-
placement, v, are perpendicular to the crack surfaces. The stress intensity factor
corresponding to this mode is denoted by K" [see Fig. 3.1 case (a)].
II. The shear or sliding mode (mode II) occurs when the two crack surfaces are dis-
placed by sliding over each other. The direction of the applied load and displace-
ment, u, are parallel to the crack surfaces [see Fig. 3.1 case (b)].
III. The tearing mode (mode III) occurs when the crack surfaces slide over each other
with displacement w in a direction parallel to the leading edge of the crack [see
Fig. 3.1 case (c)].

Note that the symbols I, II, and III are Roman numerals which refer to the
modes of fracture. The opening mode is shown by mode I. The shear and tearing
modes are represented by II and III, respectively.

y
~ ___ ~w

Mode m

Figure 3.1 Illustration of the three basic modes of crack surface displacement
 3.2. The Stress Intensity Factor Approach 107

The tensile opening mode, mode I, type of failure represents the most frequent
type of separation that engineers design against and that must be prevented. Fail-
ure of the structural parts by mode II and mode III, where fracture is induced by
shear stresses, can also occur. However, these types of fracture seldom occur dur-
ing the service life of the part. In addition, most of the available test data generated
for the critical value of the stress intensity factor, K, (fracture toughness) have
been established for mode Mode I type failure, see appendix A. Therefore,
throughout this book, attention is given to the opening mode, mode I, and its appli-
cation to fatigue crack growth when evaluating the service life of a structural part.

3.2.3 Derivation of Mode I Stress IntenSity Factor

Linear Elastic Fracture Mechanics (LEFM) is based on the application of classical
linear elasticity to cracked bodies. The elasticity assumptions made in deriving
the mode I stress intensity factor are that displacements are small and that linear-
ity between stresses and strains must be obeyed. Moreover, the assumption that
the material is both homogeneous and isotropic is also included in the derivation
of the stress intensity factor, K. The use of the crack tip stress intensity factor, K,
for assessing small crack behavior is not recommended. For more information
pertaining to the anomalous behavior of a small crack, where crack dimensions
are on the order of material microstructure, the reader may refer to Section 4.3.
The following steps are used in deriving the mode I stress intensity factor. No
attempt is made to include all the derivations. The reader may refer to References
[4, 5] for a more in-depth treatment of this topic if he or she so chooses.

1. The equilibrium equations of stresses

2. Hooke's law and the strain-displacement relationship
3. The compatibility equations of strains
4. The Airy stress function
5. The Westergaard function
6. The use of the complex variable method
7. The use of boundary conditions to solve the constants
8. The stress intensity factor equation for different crack geometries.

In solving any two-dimensional stress problem, the equilibrium equations

that describe the relationship between forces in the x and y directions can be
written as [6, 7];

(3.6)
108 Chap. 3 Linear Elastic Fracture Mechanics

From the two above equilibrium equations, there are three unknowns-lTx ' lTy,
7xy - and two equations. To obtain the quantities lTx' lTy' and 7xy' additional equa-
tions are needed.
The stresses at each point in the body are related to the strains by Hooke's law as:

Es x = lTx - VlTy

Esy = lTy - VlTx (3.7)

Eyxy
=7
2(1 + v) xy

where E and v are the modulus of elasticity and poisson's ratio respectively.
From Eqs. (3.6) and (3.7), there are total offive equations and six unknowns (lTx '
lTy, 7xy' sx' Sy, and yxy)' The strain-displacementrelationship can be written as:

au
s=-
x ax

s=-
aV
(3.8)
y ay

au av
y
xy
=-+-
ay ax

where u and v are displacement functions. From Eq. (3.8) the strain--compatibility
equation is:

(3.9a)

In terms of strains Eq. (3.9a) becomes:

a2 yxy a2 a2
axay = ay2 (sx) + ax2 (Sy) (3.9b)

From Eqs. (3.6), (3.7), and (3.9b), there are six equations and six unknowns so
that, together with the boundary conditions, the state of stress at each point in a
body subjected to the plane stress condition can be solved.

3.2.3 .1. Airy Stress Function, cf>

Airy showed that, for any elastic eqUilibrium problem, there is a function that sat-
isfies the equation of equilibrium [6]. Consider a function cf>(x, y) such that:
 3.2. The Stress Intensity Factor Approach 109

(3.10)

Using Eq. (3.10) to replace uX ' uY ' and 'Txy' it is easy to demonstrate that the equi-
librium condition described by Eq. (3.6) is satisfied. Writing Eq. (3.7) in terms of
the Airy stress function and substituting the results into the compatibility equation
described by Eq. (3.9b), we obtain:

(3.11a)

In more compact form:

(3.11b)

where the quantities V\c/J) and V2 (c/J) are called the biharmonic and harmonic
function, respectively. Any function that satisfies Eq. (3.11b) is called an Airy
stress function.
Let Z(z) represent the Westergaard function [5] sue]). that when the Airy stress
function, c/J, is written in terms of Westergaard real, 2(z), and imaginary, 2(z),
components, it can satisfy the biharmonic equation:

c/J = Re 2(z) + y 1m 2(z) (3.12)

In Eq. (3.12), the Westergaard function, Z(z), is a f!Jnction of the complex vari-
able z(x + iy) and, furthermore, the two quantities 2(z) an<l2(z) are the second-
and first-order integrals (Z(z) = dZ(z)/dz andZ(z) = d2(z)/dz). Using this
function can enable us to solve a through center crack in an infinite plate that is
subjected to a remotely applied biaxial load, u(oo), shown in Fig. 3.2. Differenti-
ating Eq. (3.12) to obtain the stresses defined by Eq. (3.10), the stresses U X ' uY '
and'Txy in terms ofthe Westergaard function, Z(z), can be expressed as:

Ux = Re Z(z) - y 1m Z' (z)

u y = Re Z(z) + y 1m Z' (z) (3. lOa)
'Txy = -y Re Z' (z)

where Z' (z) is the first-order derivative of the Westergaard function, Z(z). The
stresses defined by Eq. (3. lOa) are in terms of Z (z) and Z' (z); however, their exact
form can be determined by selecting a suitable Westergaard function that can sat-
isfy the boundary conditions. The selected function Z must be an analytical func-
tion and its derivitive, dZ/ dz, must be continuous (Cauchy-Riemann condition).
110 Chap. 3 Linear Elastic Fracture Mechanics

(),,(co)

0'(00) 0'(00)

0'(00)

Figure 3.2 Illustration of a biaxially loaded plate containing a central crack

For a center through crack in an infinite plate subjected to a biaxial load (illus-
trated in Fig. 3.2), the boundary conditions are: at x = ±oo, the applied stress,
tT(oo) = tTx and accordingly aty = ±oo, the applied stress tT(oo) = tTy; moreover,
the value of tTy = 0 for -a < x < a where y = O. At the crack tip where x = ±a,
the stress tTy = 00 (crack tip singularity).
A function that can satisfy all of these boundary conditions can be written as:

(3.13a)

where z = re i8. This function is called the Westergaard stress function. Moving
the coordinate system from the center to the tip of the crack by simply replacing
the quantity z with (z + a), Eq. (3.l3a) can be expressed as:

(3.13b)

where the derivative of this function is:

(3.13c)

Obtaining the real and imaginary components of Z(r, 0) and Z' (r, 0) [from Eqs.
(3.13b) and (3.l3c)], the stress components described by Eq. (3.1Oa) can be fur-
ther simplified. From (3.13b) and (3.13c), the real and imaginary components of
Z(r, 0) and Z' (r, 0) in terms of (r, 0) can be written as:
 3.2. The Stress Intensity Factor Approach 111

aa cos 0/2
ReZ(z) = (zar )1/2 (3.13d)

'() _ aa(-2arsinO/2 + sin30/2)

1m Z z - ()3/2 (3.13e)
2ar

where re i8 = cos 0 + i sin 0, x = r cos 0, and y = r sin O. Inserting the quantities

shown in Eq. (3.13e) into Eq. (3.10) and replacing y = r sin 0 with its equivalent
2r sin(0/2) cos (0/2), the crack tip stresses can be written as:

aViia
ax = .;:::- cos 0/2[1 - sin 0/2 sin 30/2] (3.14a)
V 27Tr
aViia
ay = • ;:::- cos 0/2[1 + sin 0/2 sin30/2] (3.14b)
V 27Tr

aViia
'Txy = .;:::- sin 0/2 cos 0/2 cos 30/2 (3.14c)
V 27Tr

Note that the out of plane stress az = 0 for the plane stress condition and
a z = v (ax + a y) for the plane strain condition.
The derivation ofEq. (3.14) is based on the assumption that the stress region at
the crack tip is highly localized and the quantity r is limited to a small region at
the crack tip where the term r/ a = O. From Eq. (3.14), it can be seen that there is
a stress singularity at r = 0 and this implies that, as the distance, r, from the crack
tip decreases, the magnitude of the stress increases and approaches infinity. We
show in Section 3.6 that most aerospace material at the crack tip will deform plas-
tically and yield before the stresses approach infinity.
The stresses described by Eq. (3.14) can be written in terms of their radial and
angular positions with respect to the crack tip parameters (1 /v'2:;;; )f( 0) and the
quantity aViia which defines the magnitude of the applied stress and the crack
length. The second quantity, aViia, is called the mode I (or opening mode) stress
intensity factor and it is designated by the symbol Kj having units of (ksi ~).
Thus, the stress intensity factor, Kj , can relate the crack tip stresses (ax' a y' and
'Txy) to parameters that are measurable quantities (a and Viia). The stress inten-
sity factor should not be confused with the stress or strain concentration factor
(Kt or Kg) or with any other similar symbols, such as the strain hardening expo-
nents (k or k').
The influence of external variables, such as the magnitude and the method of
loading, and the crack geometry, on the crack tip stresses can also be described by
the stress intensity factor. The general form of the stress intensity factor, including
the effect of the crack geometry and loading condition, can be written as:
112 Chap. 3 Linear Elastic Fracture Mechanics

(3.15)

where /3J and /32 are the corrections to the loading and crack geometry, respec-
tively. A more comprehensive solution to different crack geometries and loading
conditions is available in [8, 9].
The critical value of K J = (Tcr V7Ta is called the fracture toughness (Kc) and it
is a measure of the material's resistance to unstable cracking. Irwin [10] failure
criteria simply state that if the level of the crack tip stress intensity factor exceeds
a critical value, K cr ' unstable cracking will occur. This is analogous to the case of
a stress at a point and the criteria of yield or ultimate strength. The critical stress
intensity factor and the material's resistance to fracture (fracture toughness) are
discussed in Sections 3.2.6 and 3.3, respectively. A brief description of the ASTM
practice for the determination of plane strain and stress fracture toughness is pre-
sented in Section 3.8.
The stresses defined by Eq. (3.14) are applicable to the case of an infinite plate
containing a through center crack that is subjected to a remotely applied biaxial
load. For the case of a uniaxial tensile load, where remote stress (Tx (00) = 0, this
quantity must be subtracted from the crack tip stresses. However, because the
crack tip stresses are much higher than the farfield or applied stresses the effect is
negligible. The remote stress (Tx (00) is parallel to the crack surfaces and has little
or no influence on mode I crack opening or its stability.
When the applied load is perpendicular to the crack surfaces ((Tx (00) = 0 and
lTy(oo) ¥- 0), the variation of crack tip stresses (local stresses, lTy and lTJ along the
x-axis where () = 0 is illustrated in Fig. 3.3. The reader should note that there is a
local stress (Tx even though the applied load is in the y direction. The induced
stress, (Tx' is the result of straining in the x direction due to the Poisson effect.
The stress intensity factor solutions for a few crack geometry cases important
for use in aircraft and space vehicle applications, subjected to combined applied
load, are discussed in Sections 3.2.4 and 3.2.5.

Load Thin Plate

y Plane Stress Carl ti 00 Ccrz=O)

VI

<1~
VI
Q)
....
'-
Vl
"iii
Crack np
Stress Region
(Local Stress) S~ Nomn.1 Stress

t
x
Load Distance from Crack Tip

Figure 3.3 Stress variation in the region of crack tip

 3.2. The Stress Intensity Factor Approach 113

Kc = 30 ksi (in.112)
0"

Figure 3.4 A Center through crack plate made of 2219-T87 aluminum for example 3.1

Example 3.1

An aluminum plate made of 2219-T87 has a center through crack, as shown in Fig.
3.4. The plate is subjected to an applied stress (design stress) of (T. The value ofthe
stress intensity factor, K, at the instability, where material fracture is 30 ksi~.
Plot the variation of the design stress, (T, as a function of half crack length, c.

Solution:
From Eq. (3.15) for a center cracked specimen subjected to axial load (shown in Fig.
3.4), the two quantities f31 = f32 = 1. The relationship between the applied stress and
the critical half crack length for Kc = 30 ksi ~, can be written as:

30
(Tcr = V7iC
The plot of applied stress and the critical half crack length is shown in Fig. 3.5 for
Example 3.1.

60
III 50
~ 40
en 30
"C
:!
e. 20-
: 10-

0.2 0.4 0.6 0.8 1.0 1.2 1.4

Half Crack Length

Figure 3.5 Variation of stress versus half crack length for Kc = 30 ksi (in.) 1/2
114 Chap. 3 Linear Elastic Fracture Mechanics

3.2.4 Combined Loading

When a cracked plate is subjected to more than one type of mode I induced load
(combined loading), the resultant stress intensity factor is additive. For example,
for the case of a center through crack in an infinite plate that is subjected to an axial
tensile load, P, and bending moment, M, as shown in Fig. 3.6, the total or com-
bined stress intensity factor characterizing the crack tip stresses can be written as:

KTotal = KAxial + KBending

In general, the combined or superposition approach to the stress intensity factor of
a crack plate subjected to different loading conditions that all produce mode I of
fracture can be expressed as [8,9, 11]:

(3.16)

where 0'0' 0'1 ' 0'2' and 0'3 are the applied stresses due to tension, bending through
the thickness, bending through the width, and pin bearing, respectively. Figure
3.7 illustrates different kinds of loading conditions that can all induce the mode I
crack tip stress intensity factor. In Eq. (3.16) the quantities f3o' f31' f32' and f33 are
the corrections to the crack geometries and loading conditions. An example of
combined loading is a bolted joint where the bolted plates experience most of the
load types described by Eq. (3.16) (local plate bending due to bolt tension, bear-
ing stress due to shear, and tension due to far field stress).
The principle of superposition is helpful, not only to evaluate the stress inten-
sity factor solution of a crack plate subjected to different loading conditions (as

Axial &I Moment

Axial

+
Axial Moment
Axial 81 Moment

KTotal + KSending

Figure 3.6 Illustration of the superposition concept for an edge crack with axial and
bending loading
 3.2. The Stress Intensity Factor Approach 115

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

(a): Pure Tension

(b): Bending Through the Thickness
(e): Bending Through the Width
(d): Pin Bearing

Figure 3.7 Illustration of crack plates subjected to different loading conditions

long as the crack geometry remains the same for all loading cases), but also to
obtain the stress intensity factor solutions to the crack cases where the solution
does not exist or is difficult to obtain. For example, consider the case of a finite
width plate containing a center through crack that is wedge loaded, as shown in
Fig. 3.8, case (a). The stress intensity factor solution to wedge loaded center
cracked geometries alone is available [11] and it can be expressed as:

(PIB) (~ 2~a ) (3.17)

KJ = Viia W sin(2~aIW)

2n- a )
W sin(21r a/W)
(a)

Figure 3.8 Stress intensity factor solutions for wedge loaded and uniform tension cen-
ter crack
116 Chap. 3 Linear Elastic Fracture Mechanics

Consider the case of an identical plate with the same crack geometry that is
now subjected to a remote uniform tension load, as shown in Fig. 3.8, case (b).
The stress intensity factor solution for case (b) is also available and expressed by
Eq. (3.15) where f31 = f32 = 1. The additive properties of the stress intensity fac-
tor can enable us to find the stress intensity factor solution for the case of a plate
that is loaded by uniform tension and concentrated force, P, shown in Fig. 3.9.
The equation that describes the combined case can be written as:

K
Total
= (PIB)
v1Ta
(1 2~a
W sin (2~aIW)
) + crv1Ta (3.18)

In Section 3.2.5, the stress intensity factor solution for a few crack geometries and
loading cases that are frequently applied to the analysis of a structural part are dis-
cussed. The reader should pay attention to the fact that when dealing with com-
bined loading, it is understood that each load contributes to the opening mode I of
fracture. The effective stress intensity factor is the superposition of all the indi-
vidual stress intensity factor as described by Eq. 3.16. The K solution for the case
of a circular shaft subjected to combined axial and torsional loadings can no
longer be described by the superposition criteria (shown by equation 3.16)
because torsional applied load does not produce mode I fracture for the same
crack geometry.

~
~
+ ~
~

cr cr

KTOTAL =
2n a
W sin (2Jr at W) )+ (J .JjCa

Figure 3.9 Stress intensity factor solution for combined wedge loaded and uniform
tension center crack
 3.2. The Stress Intensity Factor Approach 117

3.2.5 Stress Intensity Factor Equations for Several

Through Cracks
The linear elastic fracture mechanics approach to the analysis of a cracked struc-
ture and its application to evaluating the service life of high-risk parts (also called
fracture critical parts) created a widespread need for a stress intensity factor solu-
tion to other complicated crack geometries. Figures 3.10 through 3.16 are the
descriptions of some of the most widely used stress intensity factor solutions for
through-the-thickness crack geometry cases that are sUbjected to a combined load-
ing consisting of either tension, through-the-thickness bending, through-the-width

6M 1
°1=-
Wt 2

W : total plate width

t : Plate thickness
M1: plate bending

Figure 3.10 Through crack in a center of plate subjected to axial stress, 0"0' and bend-
ing through the thickness, Ml

Figure 3.11 Through crack at Figure 3.12 Through crack from

edge of a plate a hole in a plate
118 Chap. 3 Linear Elastic Fracture Mechanics

P
0"0 = -
Wt

P
0"3- -
Dt

Figure 3.13 Through crack from a hole in a lu

C=aR i12Rf!
o
R = (D - t)/2

Figure 3.14 A circumferential through crack in a c linder

PR
0"0 =--
t

2R
C/(Rt) 1/2 < , 0 1....t-----1~
R = Mean Radius D
(0 - t)/2
(P is internal pressure)

Figure 3.15 An axial through crack in a pressurized cylinder

bending, or pin bearing pressure [12]. The case of surface and comer cracks is
discussed later in Section 3.7 of this chapter.
To obtain the stress intensity factor equation for the crack geometries shown in
Figs. 3.1 0 through 3.16, one must be able to determine the geometric correction
factors (f3o' f31' f32' (33) applicable to each type of applied stress defined by Eq.
(3.16). It does not matter how complicated the crack geometry and loading condi-
 3.2. The Stress Intensity Factor Approach 119

W» D

Wt 2
W
..;71 0"3
p
=-
Dt

Figure 3.16 Through crack in a plate emanating from a hole subjected to combined
loading

tions are, the stress intensity factor equation for mode I fracture always takes the
form of {3(J"( 7Ta) 1/2. Before attempting to write the stress intensity factor equa-
tions, it is helpful to introduce the following common symbols that are used in all
crack geometries to simplify the mathematics:

u = c/D, W = c/W, y = D/W, and z = (1 + 2ut 1

lo(z) = 0.7071 + 0.7548z + 0.3415z 2 + 0.642z 3 + 0.9196z 4
11 (z) = 0.078z + 0.7588z 2 - 0.4293z 3 + 0.0644z4 + 0.6516z 5

where c is the half crack length, W is the plate width, and D is the hole diameter.
Stress intensity factor equation for a through center crack in a plate (Fig. 3.10) [12]:

K = «(J"0f30 + (J"1f31)v7iC
(3.19)
f30 = (sec7Tw)I/ 2and f31 = f3o/ 2

Stress intensity factor equation for a through crack at edge of a plate (Fig. 3.11) [12]:

K = «(J"0f30 + (J"1f31 + 0"2f32)v7iC

f30 = Y[0.752 + 2.02w + 0.37(1 - sinf3n, and f3 1 = f30/2
f32 = Y[0.923 + 0.199(1 - sinf3)4] (3.20)

where

Y = sec {3[(tan {3)/ {3] 1/2, and {3 = mu/2

120 Chap. 3 Linear Elastic Fracture Mechanics

Stress intensity factor equation for a through crack from an offset hole (Fig.
3.12) [12]:

K = (uo/3o + U I/3I)v7iC
/30 = GoGw and /31 = (Goy/2 + GI)/Gw
Go = lo(z) and G I = II (z) (3.21)

where

Gw = [secA(sin/3)//3p/2

/3 = D/B - 2y

and

A = (7T/2) (1 + u)/(2B/D - u)

Stress intensity factor equation for a through crack from a hole in a lug (Fig.
3.13) [12]:

where

and

where

A = (7T/2) (1 + u)/(2B/D - u), b = (W - D)/2

and

CI = 0.688 + 0.772y + 0.613y 3

C 2 = 4.948 - 17.318y + 16.785y 3
 3.2. The Stress Intensity Factor Approach 121

C3 = -14.297 + 62.994y - 69.818y 3

C4 = 12.35 - 58.644y + 66.387y 3

Stress intensity factor equation for a through crack in a cylinder in the circumfer-
ential direction (Fig. 3.14) [12]:

K = (0"0f30 + O"If3I)ViiC
f30 = (IO/27Ta) 1/2

and

f31 = (II /27Ta) 1/2

where 10 = [-V8 (/2 - 1) + 7Tf32/b] a 2/k and II = [-V8 (g2 - 1) + 7Tf32/b)] a 2/k
The quantitiesf, g, h, b, and f3 are:

f = 1 + h(1 - cota)/2a
g = [1 + h (a + a cot 2 a - cot a) /4]( sin a) / a
(3.23)
h = -V2/(cot[(7T - a)/-V2 + -V2 cot a])
b = a/2k
where k = (t/R) 1/2 [12(1 - v2)rl/4 (see figure 3.14)

f3 = 1 + (7T/16)b 2 - 0.0293 b 3 for b:s 1

or

f3 = (-V8 b/rr)O.5 + (0.179 /b )0.885 for b > 1

Stress intensity factor equation for a through crack in a cylinder in the longitudi-
nal direction (Fig. 3.15) [12]:

K = (O"of3 o)ViiC
f30 = (1 + O.52A + 1.29A 2 - O.074A 3) 1/2 (3.24)

where

A = C/(Rt)I/2
122 Chap. 3 Linear Elastic Fracture Mechanics

Table 3.1 Values of Coefficients An' Bn' en' Dn

n An Bn en Dn

0 -0.00074 0.70920 0.7968 0

1 0.06391 0.68902 0.5326 0.0780
2 -0.10113 0.52270 0.2767 0.7588
3 -0.29411 0.65768 0.0630 -0.4293
4 -0.79179 1.91920 -O.oI7 0.0644
5 1.7197 0.6510

Stress intensity factor equation for a through crack from a hole in a plate under
combined loading (Fig. 3.16) [12]:

4
f30 = 2: [An(1 + A) + Bn]b n, f31 = [(1 + U)b nP/2(1 + v)/(3 + v) (3.25)
n~O

and

5 5
f32 = D/3 2:
n~O
Cnb n, f33 = 2: Dn bn
n~O

where the quantities shown in Eq. (3.25) are given in Table 3.1.

3.2.6 Critical Stress Intensity Factor

The critical value of the stress intensity factor is an important parameter in the
field of fracture mechanics when dealing with structural failure resulting from
unstable cracking. It is analogous to other critical failure criteria that are based on
yield or ultimate strength of the material. The critical value of the stress intensity
factor at which unstable crack propagation occurs is called the fracture toughness.
The fracture toughness data for many aerospace alloys are available in the "Dam-
age Tolerant Design Handbook" [13].
For a thick section of a given material in which the plastic deformation at the
crack tip is constrained and negligible (plane strain condition), the critical stress
intensity factor for mode I at instability is designated by K j c' The failure criteria
for unstable cracking when plane strain condition prevails can be written:

(3.26)

The plane strain fracture toughness, Kj c' is dependent on the type of material,
loading rate, and temperature. The failure criterion described by Eq. (3.26) simply
 3.2. The Stress Intensity Factor Approach 123

states that abrupt failure occurs when the crack tip stress intensity factor, k,
reaches or exceeds the material's fracture toughness, KIc ' A detailed description
of KIc is given in Section 3.3.
Equation (3.26) is valid when the plastic deformation at the crack tip is
assumed localized and small, since the assumptions stated in the formulation of
the stress intensity factor are based on linear elastic analysis (see Section 3.2.2). It
is important to note that in performing a fracture mechanics analysis, both KI and
K Ic are needed and must be available to the analyst. This is analogous to the
strength analysis requirement that the limit stress, (T, must always be compared
with both the yield, (TYieId' and the ultimate, (TVl' strength of the material. That is,
the analysis would be incomplete without both material allowables.
As the material's thickness decreases, the constraint to plastic flow decreases,
and the state of plane stress is reached. The fracture toughness associated with the
minimum thickness is called the plane stress fracture toughness and is designated
by Kc. The plane stress fracture toughness is dependent on thickness as well as on
initial crack size. A brief presentation of the ASTM testing method to obtain the
plane strain and plane stress fracture toughness values of isotropic materials is
available in Section 3.8. The reader may refer to Reference [14] for a detailed
description of the ASTM testing practice.
An analytical approach called the Theory of Fracture Mechanics of Ductile
Metals (FMDM) is available to use for calculating both the plane strain, KIc ' and
plane stress, Kc, fracture toughness values of isotropic materials. The FMDM
computer program is currently being used in some aerospace companies and it is
extremely valuable in obtaining both KIc and Kc fracture toughness values with-
out going through complicated and costly ASTM test procedures. The FMDM
theory is discussed in Chapter 6.
When fracture toughness values of a given material are available, it is easy
to determine the critical flaw size for a given stress level where the structural
instability occurs. Figure 3.17 illustrates the relationship between design stress

K - K 1C - a (ltC)1/Z

Failure Region

Crack length, 2c

Figure 3.17 Schematic plot of critical stress as a function of crack length

124 Chap. 3 Linear Elastic Fracture Mechanics

and critical flaw size for a through centered crack in a plate. From Fig. 3.l7, it
can be seen that for a wide plate with crack length 2e/, unstable cracking will
occur when the stress level becomes equal to or greater than O"c. For any other
stress level, 0", smaller than O"C' unstable cracking does not occur (Fig. 3.l7).
When the load environment is fluctuating in nature, the fracture toughness data
(critical value of stress intensity factor) are essential for a reliable crack growth
analysis and should include plane strain and plane stress as well as part through
fracture toughness. In Section 3.3, the concept of fracture toughness is dis-
cussed and later, in Section 3.7, surface cracks and their critical values (part
through fracture toughness) are briefly reviewed. Extensive fracture toughness
data for many aerospace alloys are available in Appendix A for use in crack
growth study (also see Reference [13]).

3.3 Fracture Toughness

Fracture toughness is defined as the resistance of the material to unstable crack

growth in a non-corrosive environment. This parameter characterizes the inten-
sity of the stress field at the locality of the crack tip when unstable cracking takes
place. The plane strain fracture toughness, KIc ' is considered thickness indepen-
dent; however, it can vary as a function of temperature and strain rate. The letter
I in the subscript is used throughout the literature to indicate the instability as the
result of mode I crack propagation with cleavage type fracture (flat surfaces) with-
out shear lip.
For typical metals the plane strain fracture toughness, K Ic ' is associated with
thick sections. Under the plane strain condition, the state of stress at the crack tip
approaches a triaxial tensile stress state. In the three dimensional state of stress,
the crack tip plasticity formation is constrained and is small compared with the
crack size and other specimen dimensions. Experimental data have shown that for
homogeneous material under plane strain condition the yielding (plastic deforma-
tion) occurs at a much higher stress level compared with uniaxial tension. Specif-
ically, when the applied load is in such a way that principal stresses are almost
equal in magnitude, the material will stay elastic with little plastic deformation up
to failure. However, at the two extreme free edges, the state of stress is biaxial
and plastic deformation occurs by a shear mechanism at a 45° plane with respect
to the flat surfaces. It is shown later in this section that the material resistance to
cracking is always higher at the two extreme free edges than the middle region of
the crack tip due to the degree of plastic deformation resulting from biaxial or tri-
axiality state of stress. The effect of the triaxial constraint at the crack tip is lim-
ited to the center of the cracked section through the thickness of the material and
gradually becomes biaxial at the two outer surfaces, as shown in Figure 3.18.
The cracked surfaces associated with the plane strain fracture toughness are
almost flat as viewed by the naked eye. This type of failure when viewed micro-
 3.3 Fracture Toughness 125

scopically occurs by direct separation of one atomic plane over another and is
called cleavage. Cleavage is defined as low-energy fracture that propagates along
well-defined low-index crystallographic planes known as cleavage planes. Theo-
retically, the two fractured cleavage surfaces should be completely flat. At the
center of a thick plate near the crack tip where the plane strain condition exists,
the state of stress is triaxial and principal stresses are almost equal. That is, the
hydrostatic stress state near the crack tip, where O'x = O'y = O'z is expected and the
maximum shear stress, Tmax ' is negligible which does not cause plastic deforma-
tion [15, 16]. This is illustrated in Fig. 3.18 for the center element selected at the
crack tip in the middle of the specimen. On the other hand, at the two free surfaces
of the same cracked plate, where the triaxial state of stress does not exist (O'z = 0),
ductile failure (sliding of one atomic plane over another) is produced, see the edge
element shown in Fig. 3.18. The material behavior in the locality of the crack tip
through the part thickness (at the middle and the two ends) can be analogous to
several tensile bar specimens that are situated next to each other. Those tensile
bars at the two edges have the flexibility to deform and exhibit reduction of area
(contraction) due to the Poisson effect. In the plane stress condition where O'z = 0
and Sz ¥- 0 the material behavior due to the Poisson effect is illustrated in Fig. 3.18
where contraction in the z direction and consequently the formation of a plastic
zone are permissible. In the plane strain condition where S z = 0 and O'z ¥- 0, those

Plane Strain
Plane
~ (Center and Edge Elements) Stress

......-----~

14 t ---i.~1 ~
t
Plane Strain Plane Stress (Slanted Failure)
(Triaxial state of stress)

Figure 3.18 Illustration of the state of stress at the crack tip for plane strain and
stress condition
126 Chap. 3 Linear Elastic Fracture Mechanics

bars in the middle of specimen are prevented from contraction, which results in
developing stress in the z direction and therefore blocking the formation of plas-
ticity as illustrated in Fig. 3.l8.
Figure 3.19 shows the variation in the amount of flat fracture (failure by cleav-
age) as a function of thickness. For plane strain failure (the region of K,c)' the
portion of the flat surfaces is much larger than the slanted sections. However, for
thin sections where the state of stress at the crack tip is not triaxial, the constraint
to plastic deformation lessens and the failure is associated with plane stress. For
sections with adequate thickness, in which plane strain and plane stress are com-
bined, the state of stress is termed mixed mode.
The portion of the flat region associated with the plane stress state is much
smaller than the slanted region, as indicated in Fig. 3.l9. Thus, the plane strain
failure is associated with flat or cleavage failure and the plane stress with slant
failure (ductile failure). The fact that energy absorption at the crack tip is small
for the thick section where the failure is abrupt explains why the fracture
strength is generally lower for the plane strain state of stress than for the plane
stress.
Figure 3.20 shows the variation in fracture toughness for three regions, called
plane strain, mixed mode, and plane stress, as a function of the material's thick-
ness, with the amount of flat and slanted surfaces corresponding to each region.
The asymptotic portion of the fracture toughness curve is associated with plane
strain fracture toughness and is thickness independent (Fig. 3.20). For thicknesses
less than the plane strain value (t < tid, mixed mode fracture toughness is
obtained and the maximum fracture toughness value, Kc ' corresponds to mini-
mum thickness on the curve, as shown in Fig. 3.20.

......
~
"'" Shear I
...::::I
t;j
CII

t
t!
...
u.. ~:
<: > I <:
1\1
u: tc '>
tIC
.... Almost I
...c
0
Totally
I
Almost
Totally
::::I
0
Slanted < to
> Flat
E
c(
Plane Stress Thickness, t ~ Plane Strain

Figure 3.19 Illustration of the amount of flat fracture surface versus thickness
 3.3 Fracture Toughness 127

Plane K c (Max)
Stress I V" Plane
I Strain

:t;j
I

Thickness, t

Figure 3.20 Variation of Kc as a function of thickness, t

Example 3.2
A structural part shown in Fig. 3.21 has a through crack that is emanating from a
hole. The operating stress, (Top' is 30 ksi. It is requested by the customer to proof
test the part at room temperature, as well as at - 30°F, to a stress level of
1.5 X (Top' Nondestructive inspection indicates that the maximum flaw size in the
part is c = 0.075 in. prior to the proof testing. Determine whether an existing
crack will stay stable upon proof testing.

Solution
From Eq. (3.21), the critical crack length for a through crack emanating from a
hole in terms of the critical stress can be written as:

Klc = 44 ksi (in. 1/2)

(At Room Temp.)

O'ys = 68 ksi

2r = D = 0.45 in.

c = 0.075 in.

W = 2.5 in. and

B = 1.5 in.

O'op Klc = 33 ksi (in. 1/2)

(-30 OF)

Figure 3.21 A through crack emanating from a hole in a palte (for example 3.2)
128 Chap. 3 Linear Elastic Fracture Mechanics

where K[ c = 44 ksi (in. 1/2) (at room temperature) and the proof test stress level is
given as G'p = 1.5 X G'op' The correction factor f30 = f3o(c/r).

c = {[(44.0)/(f3o X 1.5 X 30)]2}/1T

where, from Eq. (3.21), the quantity f3o(c/r) = GoG w • Solving for crack length, c:
c = 0.2 in. > 0.075 in. (by NDE inspection)
Therefore, the part will not fail upon proof test at room temperature. The second
part of the problem is for the proof test conducted at - 30°F. The related analysis
is similar to the room temperature case, except that the fracture toughness value is
replaced by K[c = 33 ksi (in. 1/2). The crack length cis:

c = {[(33.0)/(f3o X 1.5 X 30)]2}/1T

The correction factor f30 = f3o(c/r) = GoGw = 1.45 is the same as the previous case:
c = 0.08 in. > 0.075 in. (by NDE inspection)
The part will not fail upon proof test at - 30°F.

3.3.1 Fracture Toughness and Material Anisotropy

In most materials, the fracture toughness varies with crack orientation and loading
direction. This is due to the anisotropic nature of the material that evolved during
its manufacturing, cold rolling, or heat treatment (see Fig. 3.22). The ASTM
E-616 coding system for manufactured material with rectangular cross-section is

\~
Elongated Grains
Before Cold
After Cold
Rolling a Bar
Rolling
!!o.

TLL ('1
--~

L- Longitudinal Direction
T - Transverse Direction

Figure 3.22 lIIustration of the anisotropic nature of the material that evolved during
the manufacturing
 3.3 Fracture Toughness 129

such that the first letter denotes the loading direction and the second letter repre-
sents the direction of expected crack propagation. The same system would be use-
ful for bar and hollow cylinders. The standard nomenclature relative to directions
of mechanical working (elongated grain direction) for rectangular sections is:

L = Direction of maximum deformation (maximum grain elongation). This is the lon-

gitudinal direction of the rolling, extrusion, or forging process (Fig. 3.23).
T = Direction of minimum deformation. This is the direction oflong transverse (Fig. 3.23).
S = Direction perpendicular to the plane of Land T. This is the direction of short trans-
verse (through the thickness) (Fig. 3.23).

For example, the crack orientation code L-T indicates that the loading direction is
in the longitudinal direction and the direction of propagation is in the direction of
long transverse. It is noteworthy that the designated orientation code for non-
cracked parts contains only one letter. For example, the letter L designates mater-
ial properties (not fracture properties) in the direction of maximum grain
elongation.
For cylindrical sections where the direction of maximum deformation is paral-
lel to the direction of principal deformation (for example, drawn bar, extrusions or
forged parts with circular cross-section), a similar system of nomenclature for the
three directions is:

L = Direction of maximum deformation (longitudinal direction)

R = Radial direction
C = Circumferential or tangential direction.

Figure 3.23 ASTM crack plane orientation code designation for rectangular cross-
section
130 Chap. 3 Linear Elastic Fracture Mechanics

Figure 3.24 ASTM crack plane orientation code designation for cylindrical cross-sec-
tion

Figure 3.24 schematically represents the loading and crack propagation directions
for cylindrical sections. For example, the two-letter code L-R indicates that the
loading is in the longitudinal direction (L) and the expected crack propagation is
in the radial direction (R). It is important for the analyst to specify the crack and
loading directions corresponding to a given fracture toughness for the part under
consideration.

3.3.2 Factors Affecting Fracture Toughness

The change of fracture toughness with respect to material thickness was discussed
in the previous section and its variation depicted in Fig. 3.20. From Fig. 3.20 it
can be seen that several specimens with various thicknesses are needed to gener-
ate a complete curve that can fully describe the variation of Kc versus thickness, t.
When data points are not available for analysis purposes, the following relation-
ship may give somewhat reasonable results, as long as the plane strain fracture
toughness, Kr c' is available for the material under study. Irwin's equation
describes plane stress fracture toughness, Kc' written in terms of Kr c; plate thick-
ness, t; and yield strength of the material, O'Yield' as [17]:

Kc - ( -K,c - )2]
= K,c [ 1 + 1.4 (3.27)
t O'Yield

Another useful relationship describing Kc in terms of Krc is available in

NASAIFLAGRO material library and it is shown by equation 3.58 of Section 3.8.2.
 3.3 Fracture Toughness 131

III
III
cu
c:
.c:
C)
=
~
l!!
=
~
u
Ferrous Alloys

!!
~ L-______________________

Temperature ~

Figure 3.25 Illustration of fracture toughness variation with temperature

¥ Aluminum Alloys

Ferrous Alloys

Loading Rate ~

Figure 3.26 Illustration of fracture toughness variation with loading rate

1/1
III
cu
c:
.c:
C)
=
~
l!!
~=
u

~ ~--~~----------------
Yielding Strength ~

Figure 3.27 Illustration of fracture toughness variation

Other factors affecting the fracture toughness and fatigue crack growth values
are loading rate, temperature, temperature rate, and yield strength (see Figs. 3.25,
3.26, and 3.27, respectively). From Fig. 3.25, it can be seen that for aluminum
alloys the response of fracture toughness to temperature change increases as tem-
perature decreases, whereas an opposite trend was observed to be true for most
ferrous alloys [18, 19]. Experimental test data obtained in the laboratory indicate
132 Chap. 3 Linear Elastic Fracture Mechanics

that most aluminum alloys possess a higher fracture toughness value at the liquid
nitrogen temperature (-320°F) than at room temperature (see Table 3.2) [12].
This positive trend in fracture toughness value is desirable in aerospace, aircraft,
and pressure vessel structures, when proof tests at both room and liquid nitrogen
temperature are required. A higher fracture toughness at liquid nitrogen tempera-
ture indicates that if a proof test at room temperature is successfully completed,
there is no need to conduct an additional proof test at liquid nitrogen. In the case
of steel, where the material possesses sufficient ductility and good fracture tough-
ness at room temperature, cleavage or brittle fracture can occur at a service tem-
perature below the transition temperature. This change in material properties is
known as a ductile-brittle transition (see Fig. 3.28) and was the cause of many
cleavage failures occurring in ships, pressure vessels, bridges, and tanks [20-22].
Figure 3.28 shows the variation of Charpy Impact absorbed energy (representing
the material's notch ductility) versus the temperature for most ferritic steels. Orig-
inally the Charpy Impact Test was developed to relate the amount of energy
absorbed by a material when loaded dynamically in the presence of a notch to the
notch ductility at room temperature. Later it was realized that the absorbed energy
is a function of temperature for several ferritic steels. If several notched bar
impact test specimens (shown in Fig. 3.28) made of ferritic low carbon steel are
impacted by the pendulum of a Charpy Impact Test machine, the energy absorp-
tion value recorded on the machine shows a decrease in value when the tempera-
ture falls below room temperature (see Fig. 3.28). This can be an indication that

Table 3.2 Fracture properties at room temperature and

-320°F

2219-T87 (L-T)
YS UTS K[e K[c
(ksi) (ksi) ksi (in. 1/2) ksi (in. 1/2)

Room temperature
57 68 42 30

Liq. nitrogen temperature

68 83 57 41

2219-T62 (T-L)
Room temperature
43 61 41 29

Liq. nitrogen temperature

51 76 42 30
 3.3 Fracture Toughness 133

Ductile Behavior
Transition
Temperature
>. ~
e'
CLl
c:
....U
UJ

IV
C.
§
~
IV
.c
U

Temperature

Figure 3.28 Illustration of charpy impact energy versus temperature and the pendu-
lum impact testing machine

the material has gone brittle (plane strain mode of failure). The low-energy
absorption value recorded by the Charpy Impact Test machine, when compared
with room temperature, indicates brittle behavior of the material due to a low tem-
perature environment. As a final remark, for the Charpy V-notched test to be
meaningful, other parameters, such as thickness, the rate of loading, and speci-
men geometry must be kept constant. The reader may refer to ASTM E-23, "Stan-
dard Methods for Notched Bar Impact Testing of Metallic Materials," for more
information related to this topic.
The fracture toughness for aluminum alloys seems not to vary with the loading
rate, whereas ferrous alloys (such as ferritic steel) are shown to be sensitive to this
parameter, as indicated in Fig. 3.26. In general, a material with body-centered
cubic (BCC) crystal structure (ferritic steel) shows a reduction in the fracture
toughness with an increase in rate of loading. The reverse trend is true for a face-
centered cubic (FCC) crystal structures (see Fig. 3.29). The variation of the frac-
ture toughness with respect to the temperature for material with FCC and BCC
crystal structures is also shown in Fig. 3.29. A comprehensive review related to
the effects of temperature, loading rate, and plate thickness on fracture toughness
is available in reference [23].
The yield strength varies due to the heat treatment or cold working given to
the material and its variation as illustrated in Fig. 3.27. In selecting high strength
material with little ductility, to reduce the size of the structural part and save
weight, the engineer must be aware that the fracture toughness value has been
reduced considerably. That is, as the ability of the material to absorb energy and
deform plastically decreases, the size of the flaws that could initiate instability
becomes very small.
134 Chap. 3 Linear Elastic Fracture Mechanics

Temperature ~ Rate of Loading ~

Figure 3.29 Fracture toughness variation as a function of temperature and rate of

loading for material with Bee and FCC crystal structures

Additional parameters that may affect fracture toughness value and can be
important when dealing with the fracture analysis of a structural part include:

• The coarse grain size may result in lowering the fracture toughness value.
• Embrittlement [segregation of phosphor (P), nitrogen (N), and possibly sulfur (S), to
the grain boundary which causes an intergranular mode of fracture) due to
microstructure or environmental contamination can result in lowering of the fracture
toughness value.
• Work hardening lowers the fracture toughness value by lowering the hardening coef-
ficient, n.

3.4 Residual Strength Capability of a Cracked Structure

At the atomic level, the fracture phenomenon occurs when the bonds between
atoms break. For materials with perfect crystalline structure, called "whiskers,"
the measured fracture strength is much higher than the value obtained in the lab-
oratory by testing a typical standard tensile specimen. For a long time, scientists
believed that, for a given alloy, the actual tensile strength obtained through labo-
ratory test should result in a lower value than the theoretical value corresponding
to the situation where the alloy is free from defects, such as missing atoms, dislo-
cations, grain boundaries, and cavities. Upon conducting tensile tests with glass
fibers of fine diameter in which the probability of defects per volume is low,
Griffith [24] showed a tensile strength value of 5 X 10 5 psi. He postulated that
this result could be much lower if the diameter of the test specimen was signifi-
cantly larger, due to the presence of more defects and they are known as Griffith
flaws or cracks. Indeed, Weibull, in 1939, was the first to apply statistical methods
to brittle material to explain the greater probability of fracture as a result of find-
 3.4 Residual Strength Capability of a Cracked Structure 135

ing more cracks in a larger specimen than in a smaller one, leading to the possi-
bility of the existence of a size effect on the fracture stress.
From a macroscopic point of view, the presence of a crack introduced in a
structure due to manufacturing, machining, or improper handling will signifi-
cantly lower the strength of the structure as compared with the uncracked condi-
tion. The amount of strength that is left in a structure after crack initiation, which
is supposed to withstand the service load throughout its design life, is called the
residual strength. For uncracked structures, the load carrying capacity or residual
strength capability is simply the ultimate strength of the material. When the
applied load in an uncracked structure exceeds the ultimate strength of the mater-
ial, failure will start to occur.
The existence of a crack in a structure will result in a lowering of the residual
strength of the structure below the ultimate of the material. That is, when the load
on the structure exceeds a certain value called the critical stress, (Tcr' unstable
cracking will occur. For brittle materials or materials with low fracture toughness,
unstable cracking is associated with fast fracture, causing complete fracture of the
structural part. Figure 3.30a illustrates the abrupt failure (brittle failure or fast
fracture) of three specimens with crack lengths a I > a 2 > a 3 , in which the failure
stress for each individual crack length is (T] < (T2 < (T3' respectively. Figure 3.30b
also illustrates the abrupt failure points for the same crack size and crack geome-
try but corresponding to a different material possessing a higher fracture tough-
ness value, (KIch > (KIc )]. For comparison, the fracture points associated with
Fig. 3.30a are also plotted in the same figure. It can be seen that the critical stress
values (residual strength values) for the same crack lengths are higher for the sec-
ond material ((T2 > (T]), since it has a higher fracture toughness value.
For material with adequate fracture toughness, the existing crack may grow in a
stable manner until it becomes unstable (tearing failure). Figure 3.31 illustrates the
tearing failure for three specimens with different initial crack lengths, a] > a 2 > a 3.

I - Onset of Crack
Propagation

a1
...
Crack Length, a

Figure 3.30a Illustration of abrupt failure for three crack sizes

136 Chap. 3 Linear Elastic Fracture Mechanics

a, > a2 > a3
I- Onset of Crack
- 13
..
Propagation
tl 0"2 MatZ
0"\ - 13 Mat1 0"2 > 0"\
0
III 0"2
-""- --12 MatZ
...I!! -- -- I, ..
0"\ Mat1
III
_!.z -,
".!!!a. ~
f-- MatZ

0"\ f-- -- -- I,
Mat1
Cl.
c(

a3 az a,

Crack Length, a

Figure 3.30b Illustration of abrupt failure for two materials

a, > a2 > a3
F3 I - Onset of Crack
~ --. Extension
tl
13
0
Fz
.. F - Crack Propagation

...... " , I'

III
GI Iz
III 1 F,
".!!!
~

a.
Cl.
c(

a3 a2 a1
Crack Length, a

Figure 3.31 Illustration of tearing failure for three crack sizes

From Figs. 3.30 and 3.31, it can be concluded that the load carrying capacity of a
cracked structure is a function of the crack size, a, and fracture toughness, K Ic or
Kc ' and also crack geometry, {3.

3.4.1 Residual Strength Diagram for Material

with Abrupt Failure
The residual strength capability diagram for material with abrupt failure can be
plotted by simply employing the equation of the stress intensity factor, K, that
relates the critical applied stress, O"ef' to the critical crack length, aer , for a given
crack geometry and by replacing the critical stress intensity factor, Kef' with the
material's fracture toughness, KIc ' obtained through testing. That is:
 3.4 Residual Strength Capability of a Cracked Structure 137

• - Experimental data for

<J3 onset of crack propagation
III
I
...f!!
III

I
I/) <JZ -r
"iii
u
.z; I I
.;: <J, -1--1--
U I I
a3 I I a2
I

Critical Crack Length

Figure 3.32 Residual strength diagram and critical stress data (11' 12 , 13 ) extracted
from Fig. 3.30

where

(3.28a)

Figure 3.32 shows the plot of the residual strength diagram from Eq. (3.28a), in
terms of the critical stress, (Tcr' versus the critical crack length, acr ' for a center
through crack structure where the correction factor f3 = 1. It is obvious from Fig.
3.32 that, as the crack length increases, the load carrying capacity of the cracked
structure is reduced, and, if it falls below the maximum design stress level, failure
can be expected.
The same diagram can also be obtained experimentally if the applied stresses
from the test data shown in Fig. 3.30 are plotted as a function of crack length (see
Fig. 3.33). Note that for thick sections or brittle material, the onset of crack

CII

...~
I/)

"iii
.~
...
.t:!
U

Onset of Stable
Crack Extension

Critical Crack Length

Figure 3.33 Residual strength diagram for ductile metals

138 Chap. 3 Linear Elastic Fracture Mechanics

growth (as indicated by the letter I in Fig. 3.30) means fast fracture and failure of
the structural part. In other words, in brittle material or material with small plastic
deformation at the crack tip prior to fracture (state of plane strain), the crack does
not extend in a stable manner prior to abrupt fracture.
The residual strength diagram for a ductile material or metals of thin sections
cannot be described in the same way as the case of a brittle material with abrupt
fracture behavior. For ductile materials, the stable crack extension first occurs at
some stress level below the critical stress. The residual strength diagram for mater-
ial with stable crack growth prior to final failure can be constructed either by ap-
parent fracture toughness [25] or the R-curve approach [26, 27] which will be in-
troduced in Section 3.4.2 and 3.4.3, respectively. The apparent fracture toughness
approach will give conservative results when the residual strength of the material is
evaluated. The apparent residual strength curve will fall between two distinct
curves, shown in Fig. 3.33. The data points with the letter I represent the onset of
stable crack extension and the data points with the letter F describe the final failure.

3.4.2 The Apparent Fracture Toughness

One of the methods of constructing the residual strength diagram for a ductile ma-
terial is based on the apparent fracture toughness approach [25]. In this approach,
the apparent fracture toughness, K APP' can be calculated by using the initial crack
length (represented by the letter I) and final critical stress (as shown in Fig. 3.31 by
the letter F). The apparent residual strength diagram falls between the two curves
(defining the onset of stable crack extension and final crack length) represented by
the dotted line shown in Fig. 3.34. The apparent residual strength diagram shown in
Fig. 3.34 gives the lower bound estimate of fracture toughness for materials with
tearing failure behavior since the failure criterion, K ~ Kef = KAPP ' is based on the

III

...e
III

VI

K = KAPP
Onset of Stable K .. K ONSET
Crack Extension

Crack Length

Figure 3.34 The residual strength diagram for ductile metals for K = KONSET>
K = K APP' and K = K er •
 3.4 Residual Strength Capability of a Cracked Structure 139

original crack length rather than the final crack length. If Kef is evaluated based on
the final crack length, then the R-curve approach described in the next section must
be employed.

3.4.3 Development of the Resistance Curve

(R-curve) and KR
Another available method for constructing the residual strength diagram when
using tough material is the crack growth Resistance curve (R-curve) [26, 27]. This
method can be applied to material with a tearing fracture behavior that occurs in
thin sheet metal or to ductile metals where the crack extension is slow and stable
prior to final failure. The failure criterion described by the apparent fracture
toughness value was considered as a single parameter failure criterion where
K ;?: Kef = K APP' The crack growth resistance approach for tearing type fracture is
based on two fracture parameter criteria that simply state that fracture will occur
when the stress intensity factor, K, becomes equal to or greater than the mater-
ial's fracture resistance, K R • Furthermore, fracture will also occur when the rate of
change of K with respect to crack length becomes equal to or greater than the rate
of change of KR with respect to crack length, that is:

and

aK aKR
-;?: -~
(3.28b)
aa aa

In other words, the two failure criteria say that, at failure, when abrupt fracture
occurs (K ;?: K R ), the energy available to extend the crack becomes equal to or
greater than the material resistance to crack growth (¥f; ;?: a!R).
To obtain the plane stress fracture toughness, Kc ' for material with tearing frac-
ture behavior, a resistance curve, known as the R-curve, must be constructed.
Consider the variation of the stress intensity factor (up to the point of failure) with
respect to total crack lengths for a given material's thickness, as plotted in Fig.
3.35. In this figure, the calculated stress intensity factors for each original crack
length, a 1 > a2 > a3 (shown in Fig. 3.31) correspond to the onset of stable crack
growth, where KONSET = /3uONSET (7Ta O)I/2 is shown as a dotted line. Note that ao
is the original crack length and is equal to the crack length at the onset of stable
crack extension, as shown in Fig. 3.35 by the letter I. In addition, the variation of
the calculated stress intensity factor corresponding to final failure (where the
crack becomes unstable) as a function of the amount of stable crack extension,
da, is plotted in Fig. 3.36. Figure 3.36 presents the crack growth resistance curve
140 Chap. 3 Linear Elastic Fracture Mechanics

!l.a, >!l.a2 >!l.a3

..
~
..: F,
0
u F2
1\1 F3
u.
~ r-- I
-I KONSET

..
'in
c::
CD

-=
13 I IZ
I ~ I+-
I
I
I~ !l.a,I+-

.
!l.a3 I !l.aZ
l+-
C/I
C/I ~ I
....
CD
I I
II) a3 a2 a, I
Crack Length, a

Figure 3.35 Variation of stress intensity factor versus crack length

..
..:
o
U
1\1 KONSET
U. I - - - - ..... ~------
~ '-1,,12,13
'in
c:
...
CD

-=
C/I
C/I
I!!
~
Crack Extension, !l.a

Figure 3.36 Variation of stress intensity factor versus Ila

or R-curve which covers all crack growth resistance behaviors that have been con-
structed for different original crack lengths (shown in Fig. 3.35). It can be con-
cluded that the R-curve developed in Fig. 3.36 is independent of the initial crack
length, but is dependent on the amount of crack extension, !J.a.
In developing the R-curve for a given thickness, the KR value is evaluated by
using the measured effective crack length and the critical load obtained through
testing. The equation describing KR is [28]:

(3.29)

where P is the applied load corresponding to the fracture at instability, W is the

width of the specimen,f(a/W) is the correction to the width, and aeff is the effec-
 3.4 Residual Strength Capability of a Cracked Structure 141

tive crack length. The effective crack length is the total crack length and is
expressed as:

(3.30)

where rp is the correction for the plastic zone (the estimation of the size and the
shape of the plastic zone at the crack tip by using different yield criteria is dis-
cussed in Section 3.5). Note that the R-curve is supposed to be independent of the
original crack length. However, when it is developed for a given crack length and
thickness based on testing, it can be matched with the stress intensity factor curve
to estimate the fracture toughness, Ke, and the load necessary to cause unstable
crack propagation (Fig. 3.37). The tangent point between the developed R-curve
and the stress intensity factor at (T = (Te' where K = Ke, determines the fracture
toughness, as shown in Fig. 3.37. At the tangency point, shown in Fig. 3.37, the
two failure criteria described by Eq. (3.28b) are met.
In general, the construction of a residual strength diagram involves the fol-
lowing steps:

Step 1. The relationship between the crack length, applied stress, and the
stress intensity factor for the crack geometry under consideration [Equation
(3.27)] must be known or developed.
Step 2. The appropriate fracture toughness values must be available for the
material under consideration. Apply the failure criteria described in Eq. (3.27) by
equating the critical stress intensity factor with the fracture toughness value
(K = Kef ~ K1c or KJ
Step 3. Construct the residual strength diagram by plotting the variation of
the fracture stress, (Tc, versus the critical crack size, a Cf ' for the crack geometry
under consideration.

Crack Length, a

Figure 3.37 lIIustration of the R-curve where KR and K curves are tangent at the
instability point
142 Chap. 3 Linear Elastic Fracture Mechanics

KC = 35 ksi (in. 112)

Figure 3.38 A cracked panel made of 2219-T8 aluminum shown for example 3.3

Example 3.3
Construct the residual strength diagram for the cracked panel shown in Figure
3.38. The panel is made of 2219-T8 aluminum with a fracture toughness value of
35 ksi (in. 1/2).

Solution
Applying step 1 of Section 3.4.3 by using the equation for the stress intensity factor
of a wide panel (W» 2a) with a through centered crack [expressed by Eq. 3.19].

K = f30'( 1Ta) 1/2

where

f3=1

The failure criteria described by Eq. (3.27) simply show the relationship that holds
between the critical stress intensity factor and the applied stress. Note that for the
instability to occur, the critical stress intensity factor must be equal to or greater
than the fracture toughness of the material under consideration, K ;=: Ker = Kc'
The residual strength diagram can now be constructed by plotting O'er versus a cr '
as in Fig. 3.39. For example, the residual strength or fracture stress for the cracked
panel shown in the figure with crack length 2a = 1.2 in. is 22.0 ksi. That is, the
induced stress on the structural part due to the load environment must be below
the residual strength of the cracked structure (O'F = 22.0 ksi > (O')serviee ) to pre-
clude the failure of the part.

Example 3.4
Establish the residual strength capability of an eccentric through crack in a finite
plate as shown in Fig. 3.40. The crack geometry and loading conditions are also
 3.4 Residual Strength Capability of a Cracked Structure 143

60
"-
\:) 50
...CI 40
.s:
c
...I/)I!! 30
iii 20
:I
·iii 10
"tI

~
0.1 0.2 0.4 0.6 0.8 1.0 1.2
Half Crack Length, a

Figure 3.39 Residual strength capability diagram constructed for example 3.3

("j -20 ksi t - 0.3 Inch

w-1Slnch
wl- 2 inch
w2-11inch
e - 5.5 inch
Material Propenles
O"Yleid = 70 ksi
0" UI - 80 ksi
kic - 38 ksi (In.) l/Z

Eccentrically Cracked Plate

Subjected to Tension

Figure 3.40 The crack geometry and loading conditions shown for example 3.4.

shown in the figure. Assume the flaw sizes obtained by two different inspection
methods are a through crack of length (1) 2c = 1.5 in. and (2) 2c = 2.4 in.

Solution
From Fig 3.40, it is obvious that the stress intensity factor for the tip-A is more
critical than for tip-B. This is due to the width correction factor, f3" for side A
versus f32 for side B, where w, < w 2 and therefore, f3, > f32. The failure criterion
based on fracture toughness, described in Section 3.4 by Eq. (3.28a), can be
applied here to find the critical crack length for an applied stress of 20 ksi.

Case 1 (2c = 1.5 in.)

The critical flaw size based on tip-A can be obtained through equation 3.19 as:
144 Chap. 3 Linear Elastic Fracture Mechanics

where

f3w = (sec 1TC/W) 112

The width correction factor for this type of crack geometry is a function of dis-
tance, WI' and the amount of eccentricity, e, as shown in Fig. 3.40. A numerical
solution to the width correction factor for the above crack geometry is available in
Reference [29]. Another approach, more conservative but simpler to apply, was
suggested by Kaplan and Reiman [30] in which the width for the crack tip-A is
taken as twice the distance from the crack center to the edge of the plate, 2W I • In
this case, the width for tip-A is W = 4 in. The width correction factor based on
this assumption can be obtained as:

f3w = (sec 1TC/W) 112 = 1.09

K, = 20(1.09)(3.14 X 0.75) 1/2

K, = 33.454 ksi (in. 1/2)

The calculated stress intensity factor based on the initial crack length reported by
inspection (2c = 1.5 in.) is smaller than the critical stress intensity factor K, = 38
ksi (in. 1/ 2). The calculated critical crack length based on K, = 38 ksi (in.1/2),
(J" = 20 ksi, and W = 4 in. is:

38 ksi (in. 1/2) = 20 ksi (sec 1Tc/4) 1/2 (3.14 X c) 1/2

Solving for c:

c = 1.149 in. or 2c = 2.299 in. > 1.5 in. (by inspection)

The calculated critical crack length (2c = 2.299 in.) is larger than the crack length
found by inspection (2c = 1.5 in.). Therefore, the cracked plate will survive the
load environment.

Case 2 (2c = 2.4 in.):

As shown by the analysis performed for the previous case, the calculated critical
crack length 2c = 2.299 in. is smaller than the preexisting crack reported by the
second inspection (2c = 2.4 in.). Based on this assumption the crack at tip-A is
critical and will propagate toward the edge of the plate. The new crack geometry
is now a single edge crack with length c = 2.0 + 2.4/2 = 3.2 in., as shown in Fig.
3.41. The stress intensity factor and width correction factor for an edge crack in a
finite plate are given by Eg. (3.20) that is:
 3.4 Residual Strength Capability of a Cracked Structure 145

0" = 20 ksi

0" = 20 ksi
C= 3.2 in.

Figure 3.41 An Edge Crack in a Finite Plate (when tip-A propagate toward the edge
and become an edge crack)

KI = (J"f3o Viii:
130 = Y[0.752 + 2.02w + 0.37(1 - sinf3)3]

where

Y = sec 13 [(tan 13)/13] 1/2, and 13 = 1Tc/2W

Calculating the quantities Y, 13 and the width correction 130:

13 = (3.14 X 3.2)/(2 X 15) = 0.334
Y = sec (0.334) [(tanO.334)/0.334P/2 = 1.081
130 = 1.081 [0.752 + 2.02(3.2/15) + 0.37(1 - sinO.334)p = 1.398
K] = 20 X 1.398 X (3.14 X 3.2)1/2 = 88.63 > 38 ksi(in. 1/ 2 )

From the above analysis, it is clear that the calculated stress intensity factor for
the new crack geometry (a single edge crack) is much higher than the fracture
toughness of the material. Therefore, as soon as tip-A becomes unstable and is
arrested by the left free edge, the newly formed crack geometry (a single edge
crack) becomes unstable also.
In general, when analysis indicates that the residual strength capability of a
given structural part is not adequate, it is recommended that either (1) the inspec-
tion method be revised to obtain smaller initial crack length or (2) the magnitude
of the applied stress, (J", be reduced at the expense of increasing the part thickness.
However, by doing that the fracture toughness value for the new thickness is now
reduced and the material will tend to approach the plane strain condition.
146 Chap. 3 Linear Elastic Fracture Mechanics

3.5 Plasticity at the Crack Tip

Linear elastic fracture mechanics is based upon the assumption that the size of the
plastic zone formed at the crack tip is negligible as compared to the crack length
and plate thickness. That is, the crack tip plastic deformation is confined to a small
region around the crack tip and the bulk of the structural components is elastic.
For metals that generally go through extensive plastic deformation at the crack tip
prior to failure, the use of linear elastic fracture mechanics yields conservative
results when solving a given crack problem. This is true because the applied load
does work on the cracked body which is stored in the form of strain energy. For
brittle materials, all of the available energy will be consumed in creating two new
crack surfaces. In ductile material, a large portion of the available energy will be
consumed in plastically deforming the material at the crack tip (in metallic mate-
rials the energy required for plastic deformation is approximately 10 3 times larger
than the surface energy). Chapter 6 presents the Fracture Mechanics of Ductile
Metals (FMDM) theory which will incorporate the dissipated energy terms con-
sumed in plastically straining the material in the region at the crack tip.
A problem that arises in applying linear elastic fracture mechanics is that the
calculated crack tip stress approaches the very high value predicted by the quan-
tity 1/~ [shown by equation (3.14)] whenever the term r ~ O. In real situa-
tions, there will be a finite plastic zone (rp) ahead of a loaded crack where the
material will yield prior to final failure. Irwin's first approach [10, 31] to obtain
the plastic zone shape and size was based on the assumption that the plastic zone
formed at the crack tip is a circle (Fig. 3.42) and that the diameter of the circle, rp '
for the plane stress condition (where U z = 0) can be calculated by simply equat-
ing the crack tip stress, u y [from Eq. (3.14)] to the tensile yield of the material.
For a through center crack plate subjected to a far field stress u:

uV7TG
U yield = u y = ~ ~ cos 012 [1 + sin 012 sin3012] (3.31)
V2'TTT

~------------~ x

Figure 3.42 Plastic deformation at the crack tip and stress variation before and after
yielding
 3.5 Plasticity at the Crack Tip 147

At the crack tip where () = 0 and r = rp:

(3.32)

Rewriting Eq. (3.32), the plastic zone size, r p ' becomes:

(3.33)

Irwin further argued that the stress distribution ahead of the crack tip (shown in
Fig. 3.42 as a shaded area) cannot simply be ignored when stresses are above the
tensile yield of the material. Irwin's second estimate of the plastic zone size
resulted in calculating a larger plastic size than that obtained by Eq. (3.33). Fur-
thermore, he stated that the formation of a plastic zone at the crack tip can be
viewed from linear elastic fracture mechanics as if the crack length is longer than
its original size, ao ' by some crack increment, Aa. This is true because the forma-
tion of plasticity at the crack tip can result in having a larger crack tip opening dis-
placement, as if it contained a crack of larger size. That is, the new crack length
(called notional crack) is now equal to (a o + r p )' with its tip a distance, r p ' ahead
of the original crack tip (Fig. 3.43).
To account for the stress distribution above the yield region (as shown in Fig.
3.44), the following equilibrium equation is valid:

(3.34)

Oy

(2xr) 112

Corrected
Crack TIp

....- - - -. . x
--:;1 2rp ~
Ia = a o + rp~
Figure 3.43 Irwin's correction for the crack tip
148 Chap. 3 Linear Elastic Fracture Mechanics

cry
A, cry

crYield A2 crYield

A, + A2 _ A3

Figure 3.44 First and second estimate of plastic zone size

Equation (3.34) simply states that the total forces produced by the stresses in
the shaded and unshaded regions ahead of the crack tip (designated by the let-
ters Al and A 2 , respectively) must be equal to the force produced by the stresses
in the region A3 (see Fig. 3.44). Replacing K] at r = ry with (O"yield) \h7T"ry and
solving for Dp :
(3.35)
Thus, the plastic zone correction based on a notional crack is to extend the
original crack length by the amount of the plastic zone, Dp/2 = rp, derived by
Eq. (3.35) (Fig. 3.43). The stress intensity factor associated with the notional
crack is KNEW = /3O"Y 7T"(a + rp).

3.6 Plastic Zone Shape Based on the Von Mises

Yield Criterion

In the previous discussion related to the plasticity formation at the crack tip (given
in Section 3.5), a simplified assumption that the shape of the plastic zone is a cir-
cle [31] was employed and the radius of the plastic zone was derived based on
crack tip stresses taken to be equal to the uniaxial yield of the material [Eq.
(3.31)]. That is, a given material will fail plastically when the maximum principal
stress becomes equal to the uniaxial yield strength. Moreover, the dependency of
the plastic zone with respect to the angle, 0, was not accounted for in Irwin's
approach, described in Section 3.5.
A more appropriate yield criterion can be implemented to provide the size and
shape of the plastic region for all values of 0 [32, 33]. The most common accept-
able yield criterion is based on the Von Mises criterion (proposed in 1913) [34],
which simply states that, for yielding to occur, the maximum value of the distor-
tion energy per unit volume in that material must reach the distortion energy per
unit volume needed to yield the material in a tensile test specimen of the same
material. In terms of principal stresses, the Von Mises criterion can be written as:
(3.36)
 3.6 Plastic Zone Shape Based on the Von Mises Yield Criterion 149

The equivalent stress, (Te' shown in equation 3.36, is calculated from multiaxial
stress state where (Tl' (T2' and (T3 are principal stresses at a given point in the
body. Yielding occurs when the quantity (Te exceeds the monotonic yield value of
the material.
The principal stresses (Tl' (T2' and (T3 are related to the crack tip stresses (Tx,
(Ty' and (Tz by the following relationships:

_ (Tx - (T y+ [((Tx - (T y)2 + JI / 2

(3.37)
(Tl - 2 2 Txy

_ (Tx + (Ty _ [((Tx - y

(T)2 + JI / 2
(3.38)
(T2- 2 2 Txy

where (T3 = 0 for the plane stress and (T3 = V ((Tl + (T2) for the plane strain condi-
tion and v is the Poisson ratio. Substituting for the quantities (Tx, (Ty' and Txy from
Eq. (3.14) of Section 3.2.3, the principal stresses, 0"1 and 0"2' in terms of stress
intensity factor become:

KJ
0"1 = ~;:::-cos812[1 + sin812] (3.39)
v 21Tr
K,
0"2 = ~ ;:::-cos812[1 - sin812] (3.40)
v 21Tr

Inserting the principal stresses from Eqs. (3.39) and (3.40) into the Von Mises
yield criterion [shown by Eq. 3.36 where O"e = O"yield)' an expression for the plas-
tic zone radius, rp ' as a function of 8 can be obtained. For the case of the plane
stress condition, where 0"3 = 0:

(3.41)

and for plane strain where 0"3 = v (0"1 + 0"2):

(3.42)

The plastic zone size, rp ' for the case of 8 = 0 can be obtained based on the Von
Mises yield criterion via Eqs. (3.41) and (3.42) for the plane stress and plane
150 Chap. 3 Linear Elastic Fracture Mechanics

Shape of the Plastic Zone

Based on the Von Mises Criterion

Plane Strain Mode

----f-+e=O

Figure 3.45 Plastic zone shape for the plane stress and plane strain condition
strain conditions, respectively. The plot of the nondimensional quantity
rp(())/rp(O) versus the angle () is plotted in Fig. 3.45. Note that the plastic zone
size, rp' for () = 0 is equal to the value of rp that was obtained from Eq. 3.33.
The nondimensional quantity rp (())/rp(0) for the plane stress condition can be
written as:

rp(())/rp(O) = 1/2 + 3/4 sin 2 () + 1/2 cos () (3.43a)

and for the plane strain:

(3.43b)

As discussed in Section 3.3, the fracture toughness is related to the amount of

plastic deformation and varies with the thickness. For a structural part with a
given thickness, the shape and the size of the plastic zone vary throughout the sec-
tion. The variation of the plastic zone size and its shape through the thickness
based on the Von Mises yield criterion is illustrated in Fig. 3.46. At the free edges,
where 0'3 = 0, the plastic zone totally resembles the plane stress case (see Fig.
3.46). In the interior or the midsection region, the plastic zone shape and size cor-
respond to the plane strain condition.

Example 3.4
Plot the plastic zone shape and size for the plane stress and plane strain cases by
applying the Tresca yield criterion.
 3.6 Plastic Zone Shape Based on the Von Mises Yield Criterion 151

at the midsection

Figure 3.46 Illustration of plastic zone shape through the thickness

Solution
The Tresca yield criterion [35] is based on the maximum shear stress criterion,
which simply states that a given structural component is safe under multiaxial
state of stress when the maximum value of shear stress, Tmax ' in that component is
smaller than the critical value. The critical shear stress value corresponds to the
value of the shearing stress in a tensile test specimen of the same material as the
specimen starts to yield. In terms of principal stresses, when IT\ > lT2 > lT3 and the
Tresca yield criterion can be written as:

(3.44)

For the plane stress condition, where lT3 = 0, Eq. (3.44) in terms of the crack tip
stress intensity factor can be written [see Eq. (3.39)] as:

lTl = ,~ cose/2[1 + sine/2] = lTYield

V 27Tr

Solving for the plane stress plastic zone size, r = rp:

and for the plane strain condition, where lT3 = v (lTl + lT2 ), the size of the plastic
zone in terms of the crack tip stress intensity factor can be expressed as:
152 Chap. 3 Linear Elastic Fracture Mechanics

Plane Stress Zone

6=0

Figure 3.47 Plastic zone shape based on the Tresca yield criterion

The shape of the plastic zone, based on the Tresca yield criterion (described by the
above two equations), is plotted in Fig. 3.47.

3.7 Surface or Part Through Cracks

In most structures, preexisting cracks are found in the form of surface cracks (also
called part through cracks) that initiate at surface discontinuities or emanate from
a hole in the form of comer cracks (see Fig. 3.48). Surface scratches are intro-
duced into the part as the result of surface machining, grinding, forming, or may
be due to improper handling during manufacturing. These surface cracks may
become through cracks during the service life of the structural part before reach-
ing their critical size. In other cases, embedded cracks found in welded parts will
grow gradually to the surface and become surface cracks.
Surface cracks will grow in both length and depth directions. Therefore, in
analyzing these crack geometries by linear elastic fracture mechanics, it is impor-
tant to have an expression for the mode I crack tip stress intensity factor, Kr•
When the load environment is fluctuating, the expression pertaining to stress
 3.7 Surface or Part Through Cracks 153

An Elliptical Part
Through Crack A Corner Crack Eminating
from a Hole

Figure 3.48 Illustration of a surface part through crack and a corner crack eminating
from a hole

intensity factor, KJ , must be provided to the analyst if fatigue crack growth analy-
sis for both the depth and length directions are needed. In Section 3.7.1, the stress
intensity factor equation for a surface crack in an infinite plate is formulated. The
solutions corresponding to two of the most commonly used surface crack geome-
tries are discussed in Sections 3.7.2 and 3.7.3. Part through fracture toughness for
both the depth and length directions are covered in Section 3.7.4. Finally, the con-
cept of leak-before-burst (LBB) is introduced in Section 3.7.5, together with
example problems.

3.7.1 Stress Intensity Factor Solution for Surface

Cracks (Part Through Cracks)
The stress intensity factor, KJ , for a surface crack in a plate subjected to uni-
form tensile load (as plotted in Fig. 3.48) can be formulated by using the stress
intensity factor equation corresponding to an embedded elliptical crack in an
infinite body subjected to uniform load (see Fig. 3.49 for the geometry and
loading [36]).
At any point, m, along the boundary of the elliptical crack, the mode I stress
intensity factor can be written as:

(3.45)
154 Chap. 3 Linear Elastic Fracture Mechanics

x= C cose

y = a sine

Figure 3.49 Embeded elliptical crack geometry under mode 1 loading

where () is the angle that defines any point around the perimeter of the elliptical
crack (Fig. 3.49). The quantity <P is the complete elliptical integral of the second
kind and is given by:

(3.46)

Empirical expressions that can describe the quantity <P of Eq. (3.46) for different
crack depth to crack length aspect ratios, a/ c, are [28]:

<P = 1.0 + (a)

1.464 -;:
1.65

(3.47)
<P =
c)
1.0 + 1.464 (-;;
1.65

for a/ c ~ and a/ c > 1, respectively.

Using the KJ solution for an embedded crack described by Eq. (3.45), the stress
intensity factor equation at any point, m, around the periphery of a surface flaw in
an infinite plate subjected to tensile load can be written as [36]:

KJ =
1.1 u-Viia
<P -;: cos 2 (J + sin2 ())!4
((a)2 (3.48)

The maximum and minimum stress intensity factors around the crack front are
associated with the angles () = 90 0 and () = 0°, respectively.
When the part through crack is a circular surface crack, as shown in Fig. 3.50,
where the aspect ratio a/ c = 1, the value of the stress intensity factor around the
 3.7 Surface or Part Through Cracks 155

Circular Surface Crack

~ ale = 1

t
t~ 2~--1
r------l-t. t-r--'tll

Shallow Surface Crack

ale = 0.2

Figure 3.50 Illustration of circular and shallow cracks with alc = 1 and alc = 0.2,
respectively

crack front is a constant. From Eq. (3.48), the stress intensity factor corresponding
to a circular crack (ale = 1) can be written as:

(3.49)

where the quantity <P = 2.464 [see Eq. (3.47) for a/ e = 1]. Accordingly, for a shal-
low crack, where a/ e = 0.2 and () = 90°, as shown in Fig. 3.50, the stress intensity
factor can be represented by Eq. (3.49), in which <P takes the value of 1.102.
The two previously mentioned aspect ratios of a/e = 1 (a circular crack) and
a/ e = 0.2 (a shallow crack) as shown in Fig. 3.50 are the limiting aspect ratio
cases that are widely used in fracture mechanics analysis. For example, NASA
requires all fracture critical flight hardware (such as thin-wall pressure vessel
structures) to be examined for safe-life using these two limiting aspect ratios to
ensure that the structural part can survive the expected environment.
The correction factor, {3, employed for the equation of stress intensity factor to
account for the plate width, back surface correction, and loading conditions of a
part through surface crack [see Eq. (3.49)] is provided in References [37, 38] and
can be written as:
156 Chap. 3 Linear Elastic Fracture Mechanics

a-V7Ta
KJ = f3(a/c, aft, c/w, 8) -<1>- (3.50)

where the correction factor f3 is:

The back surface correctionfw = [sec 7Tc/w(a/t) 112] 112. Other quantities in Eq.
(3.51) are given as:

Ml = 1.13 - 0.09(a/c)
M2 = -0.54 + 0.89/(0.2 + a/c)
M3 = 0.5 - [1/(0.65 + a/c)] + 14(1 - a/c)24

g = 1 + [0.1 + 0.35(a/t)2](1 - sin 8)2

f", = [(a/c?cos 2 8 + sin2 8]1/4

for the case of a/c < 1, and for the case of alc > 1, the correction factor para-
meters are:

Ml = ~ [1 + 0.04(c/a)]
M2 = 0.2(c/a)4
M3 = -0.11(c/a)4
g = 1 + [0.1 + 0.35(a/t)2(c/a)](1 - sin8)2

f", = [(c/a)2sin2 8 + cos 2 8p/4

It should be noted that Eq. (3.51) is valid within the limits of:

c I
o : :;:; -a : :;:; 2, - < -, 0:::;:; 8 :::;:; 7T
c w 4

and

~t :::;:; 1.25 (~
c
+ 0.6) for 0 :::;:; ~c :::;:; 0.2
a a
- < 1 for 0.2 :::;:; - :::;:; 00
t c
 3.7 Surface or Part Through Cracks 157

The stress intensity factor solutions for other surface crack geometries are also
available in Reference [39]. In this section, only the two most widely used cases
in aircraft and aerospace industries for life evaluation of high-risk or fracture crit-
ical parts will be discussed. These cases are: (l) a surface crack in a pressurized
pipe and (2) a corner crack emanating from the edge of a hole.

3.7.2 Longitudinal Surface Crack in a Pressurized Pipe

The stress intensity factor for a surface crack oriented in the longitudinal direction
in a pressurized pipe has the form of Eq. (3.50). The u.s. Air Force and NASA
require all pressurized containers to be considered as a fracture critical or high-
risk part and they must demonstrate an analytical life of not less than four times
the service life of the structure (see Chapter 5 for the definition of fracture critical
part). A pressurized vessel is defined as a container that stores fluid with stored
energy of 14,240 ft-Ibs or contains gas with pressure above 14.7 psi. In a pressur-
ized cylinder the circumferential induced stresses (hoop stress) are larger than the
longitudinal stresses. It is therefore necessary to develop the stress intensity factor
solution for a pressurized pipe or vessel with a longitudinal surface crack. The
correction factor, {3, for the stress intensity factor of a part through thickness crack
in a pressurized pipe (shown in Fig. 3.51) with diameter Dis [39]:

where the quantities in Eq. 3.52 are:

Ml = 1.13 - 0.09(a/c)
M2 = -0.54 + 0.89/(0.2 + a/c)
M3 = 0.5 - [1/(0.65 + a/c)] + 14(1 - a/c)24
g = 1 + [0.1 + 0.35(a/t)2] (1 - sin lJ)2
f", = [(a/c)2cos2lJ + sin2 lJP/4

Hoop Stress
(J = P (0 - 2t)12t
p - Internal Pressure

Figure 3.51 Part through crack in a pipe

158 Chap. 3 Linear Elastic Fracture Mechanics

The stress intensity factors for the depth and length directions are associated with
the angles () = 90 0 and () = 0 0 , respectively.
The quantities!c = [(1 + e)/(1 - k 2) + 1 - 0.5(a/t) 1/2] [t/(D/2 - t)] where
k = 1 - 2t/D and the value oft; = 1 for internal crack and 1.1 for external crack.

3.7.3 Part Through Corner Crack Emanating

from a Hole
In fracture mechanics analysis of high risk parts of a given flight hardware, NASA
and the military require the analyst to assume the existence of a comer crack from
the edge of a hole (when this assumption is considered as the worst crack location
and orientation). Understanding comer cracks from a hole is extremly important
when assessing the integrity of a bolted joint. Many structural parts, especially in
aircrafts, are jointed together by fasteners which their failure can lead to loss of
life and the structure. To be able to implement the above requirement, it is neces-
sary to develop the stress intensity factor solution for comer cracks at the edge of
a hole (Fig. 3.52). The solution to the general case of a quarter elliptical crack
geometry is given by Eq. (3.50), with the correction factor, j3, provided by New-
man and Raju as [38, 39]:

where

Mj = 1.08 - 0.03(a/c)
M2 = -0.44 + 1.06/(0.3 + a/c)
M3 = -0.5 - 0.25(a/c) + 14.8(1 - a/c)15
gl = 1 + [0.08 + 0.4(a/t)2] (1 - sin (})3
g2 = 1 + [0.08 + 0.15(a/t)2](1 - cos (})3
!q, = [(a/c)2cos 2(} + sin2(}]l /4

[[]tJ ~Io~

Figure 3.52 Part through corner crack emanating from a hole

 3.7 Surface or Part Through Cracks 159

for the case of a/ c < 1 and

M, = VcTa [1.08 - O.03(e/a)]

M2 = 0.375 (c/a)2
M3 = -0.25(c/af
g, +
= 1 [0.08 + 0.4(e/t)2] (1 - sin 0)3
g2 = 1 + [0.08 + 0.15(e/t)2](1 - cos 0)3
I</> = [( c / a)2 sin 2 0 + cos 20]' /4
for the case of a/e > 1.

3.7.4 Part Through Fracture Toughness, "'e

The critical value of the stress intensity factor for a part through crack is called the
part through fracture toughness and is designated by the symbol K, e' Just like the
plane strain and plane stress fracture toughness data for through cracks, the part
through fracture toughness data are also an important parameter to have when the
number of cycles to failure associated with a given load environment is evaluated
(see Chapter 4 for an in-depth discussion of the fatigue crack growth mechanism).
The part through fracture toughness can be obtained through testing (see
ASTM-E740). The ASTM test specimen for determining the part through frac-
ture toughness value, KI e , is shown in Fig. 3.53. A dog bone shape specimen hav-

ae
2c
a i)< >1

1-
L U2

Figure 3.53 Illustration of ASTM part through crack test specimen

160 Chap. 3 Linear Elastic Fracture Mechanics

ing part through crack (possessing crack depth, a, and crack length, 2c) is sub-
jected to axial load as shown in the figure. The critical value of load at fracture, as
well as final crack length and crack depth after fracture, are measured. Using Eqs.
3.50 and 3.51 [38] the critical value of stress intensity factor at fracture where K =
K'e can be calculated. A typical test specimen configuration can be:

Use 2c = 0.4 in.

Use a = 0.1 in.
Use w = 3.5 in.
Use L = 7.0 in.
Total length 14.0 in.
Note: The above dimensions are for t = 0.3 in.

In the case where the K'e value is not available for the fracture analysis, it can be
approximated in terms of plane strain fracture toughness, K, c' for through cracks
by the following equation [39]:

(3.54)

where Ck is an empirical constant, which for most aerospace material is equal to

1.0 (in. -1/2). Equation (3.54) gives reasonably good results for a variety of
isotropic materials. The value of the part through fracture toughness, K Ie , for
material with high KIci O"yield ratio will be limited to 1.4KIc [33].
As mentioned previously, there are two stress intensity factors that are associ-
ated with a part through crack, corresponding to the angles () = 90° and () = 0°.
The critical values of K, for the depth direction, where () = 90° (also called a-tip),
is given by Eq. (3.54) and is designated by K Ie . The critical value of K, for the
length direction, where () = 0° (c-tip), is approximated by UK, e.
One of the most important applications of K'e is in pressurized tanks used in
aerospace and nuclear structures where the requirement of leak-before-burst
(LBB) must be met. In the case of a pressurized tank that contains a hazardous
fluid or gas, LBB is not desirable and the tank must demonstrate it has safe-life
during its usage. Section 3.7.5 briefly discusses the LBB concept.

3.7.5 The Leak-Before-Burst (LBB) Concept

In most materials, it is desirable to have crack stability even though the part
through crack grows and becomes a through crack [40, 41]. For this to happen, the
following conditions at the onset of part through crack to through crack transition
should be valid:

(K, )Part through crack = (K, hhrough crack (3.55)

 3.7 Surface or Part Through Cracks 161

Moreover:

(K, )Part through crack < K, e (3.56a)

and

(3.56b)

Under these conditions, the part through crack (now it is a through crack) will
grow until the stress intensity factor, K" becomes equal to or greater than Kcor
K,c. This condition is called leak-before-burst (LBB).
The leak-before-burst criteria described by Eqs. (3.55) and (3.56) can be
applied to a pressurized tank. The assumed surface crack (based on the NDE
inspection technique) will grow due to fatigue and sustain load through the thick-
ness. From the safety point of view, for a hazardous fluid or gas, neither leak-
before-burst nor the condition of instability, where K, (at a-tip) > K'e are
desirable. For a nonhazardous fluid or gas, however, leak before burst is required,
as illustrated in Fig. 3.54.
If the criteria for leak-before-burst described by Eqs. (3.55) and (3.56) do not
hold and a leak-before-burst design is desired, the following modifications might
be considered:

1. Thicken the structure to reduce the design stress.

2. Change the material.

Examples of fatigue crack growth problems related to LBB for pressurized vessels
are demonstrated in Section 4 of Chapter 4.

(K I ) Surface Crack • (K I) Through Crack

":;: 2c

..
Stability: K( < Kle
w

Instability: (K I ) > K c, K Ie
..
2c
~t
Through Crack
rn
Figure 3.54 Illustration of Surface Cracks Before and After Transition to a Through
Crack
162 Chap. 3 Linear Elastic Fracture Mechanics

Penetrant Inspection
2e - 0.15 With a/e-l Hoop Stress
(J = P (0 - 2t)l2t
p = Internal Pressure
t = 0.10 in.

o
h;JS 1~
D = 25.0 in. K[ e = 30 ksi (in. lIZ)
p = 100 psi Ke = 50 ksi (in. lIZ)

Figure 3.55 A pressurized tank subjected to internal pressure with a surface flaw
(shown for example 3.5)

Example 3.5

A pressurized cylindrical tank is subjected to 200 psi pressure, as shown in Fig.

3.55. As part of the safety requirement, the tank must undergo a proof test of 1.5
X operating pressure (300 psi). The flaw size obtained by performing penetrant
inspection prior to the proof test indicated that the maximum preexisting circular
surface flaw that can escape the inspection has 2c = 0.15 in. with a/c = 1. The
pressurized tank is made of 2219-T87l aluminum alloy with KIc = 30 ksi (in. 1/2).
Determine (1) if the tank will be leak-before-burst and (2) whether or not the
through crack will be stable (assume K Ie = l.lKIc )'

Solution
The stress intensity factor solution to the crack geometry shown in Fig. 3.55 is
given by Eq. (3.52). The correction factor to the stress intensity factor of the
above crack geometry is:

f3(a/c, a/t, (J) = 0.97 * [M I + M 2(a/t)2 + M J (a/t)4] * g * I¢ * Ie *1;

where a / t and the quantities M I ' M 2' M J for the case of a / c = 1 can be calculated
as follows:

M J = 1.13 - 0.09(a/c) = 1.04

M2 = -0.54 + 0.89/(0.2 + a/c) = 0.201
M J = 0.5 - [1/(0.65 + a/c)] + 14(1 - a/c?4 = -0.1
g = 1 + + 0.35(a/t)2](1 -
[0.1 sin 0)2 = 1
I¢ = [(a/c)2cos 2 0 + sin2(J]1/4 = 1

The stress intensity factor for the depth direction is associated with an angle of (J = 90°.
 3.7 Surface or Part Through Cracks 163

The quantities!c = [(1 + k 2)/(l - k 2) + 1 - 0.5 (a/t)1/2] [t/(D/2 - t)] =

1.015 where k = 1 - 2t/D = 0.992. The value of/; = 1 for an internal crack and
1.1 for an external crack. Substituting these values into the stress intensity correc-
tion factor equation:

Moreover, the induced hoop stress due to 200 psi proof pressure is:
a = p(D - 2t)/2t = 37,200 psi = 37.2 ksi
aViia
KI = f3(a/c, aft, c/w, 0) ~
= 1.21 X 37.2 X (3.14 X 0.75)1 /2/2.464 = 8.865 ksi (in.1/2)

where

<I> = 2.464
Note that the calculated stress intensity factor is much smaller than the part
through crack fracture toughness value, KIe = UK,c = 33 ksi (in. 1/ 2). There-
fore, no catastrophic failure is expected, that is:

K'e = 33 ksi (in. 1/2) > 8.865 ksi (in. 1/2)

Now let us assume the existing crack grows and becomes a through crack hav-
ing length 2c = 2t = 0.2 in. (the part through crack is now leaking). To check for
LBB, it is necessary to calculate the stress intensity factor for a through crack of
length 2c = 0.2 in. From equation (3.56), the condition for stability can be written
as:

The equation for the stress intensity factor from Eq. (3.24) is:

K = (aof3o)ViiC
f30 = (1 + 0.52A + 1.29A2 - 0.074A3)1/2

where

A = C/(Rt)1/2
A = 0.1/(12.5 X 0.1)1/2 = 0.09
164 Chap. 3 Linear Elastic Fracture Mechanics

f30 = (1 + 0.52 x 0.09 + 1.29 x 0.092 - 0.074 X 0.09 3)1 /2 = 1.04

K = (aof3 o)V7TC = 37.2 x 1.04 x (3.14 x 0.1)1/2 = 21.68 ksi (in.1/2)

Because the calculated value of the stress intensity factor is smaller than the frac-
ture toughness of the material, the tank would be leak-before-burst:

(K,hhroughcrack = 21.68 ksi (in. 1/2) < Kc = 50 ksi (in. 1/2)

indicating that the through crack will be stable.

3.8 A Brief Description of ASTM Fracture Toughness

Determination

The present method of determining the residual strength capability of a given

structure by LEFM requires the critical value of the stress intensity factor (called
fracture toughness). The fracture toughness data can be produced by conducting
several tests, performed in accordace with the ASTM practice. Section 3.3 has an
in-depth discussion on plane strain and plane stress fracture toughness and their
importance in generating the residual strength capability curve. Since fracture
toughness is a function of thickness and crack length, it is therefore necessary to
conduct several laboratory tests with various thicknesses that require specimen
preparation, surface finish, and data collection (such as stable crack growth and
load increase measurements). These preparations and measurements are costly
and time consuming.
Standard test methods for plane strain fracture toughness of metallic materials
are briefly discussed in Section 3.8.1 and the determination of plane stress and
mixed mode fracture toughness, K c ' by the resistance curve (R-curve) method is
presented in Section 3.8.2.

3.8.1 Plane Strain Fracture Toughness (~c) Test

As mentioned in Section 3.3, the plane strain fracture toughness, K[c' is indepen-
dent of thickness and crack length. To obtain a valid K,c value, the state of stress
near the crack front must approach the triaxial tensile plane strain where the for-
mation of plasticity at the crack tip is reduced. Moreover, the plastic zone size
formed at the crack tip must be small compared to other dimensions (such as
crack size and specimen thickness). The relationships between the crack tip plas-
tic zone size, ry' and specimen size requirement to ensure elastic plane strain con-
dition throughout the body are:

(3.57)
 3.8 A Brief Description of ASTM Fracture Toughness Determination 165

a ~ 2.5 (Klc)2B ~ 2.5 (KIC )2 W~ 5.0(KIc )2 (3.58)

O"YS O"YS O"YS

where a, B, and Ware the crack length, and specimen thickness and width, respec-
tively. The following calculation shows that the specimen thickness should be
approximately 47 times the radius of the plane-strain plastic zone, ry' in order to
meet the plane strain test specimen requirements. Dividing the crack length, a, or
specimen thickness, B, from Eq. (3.57) by Eq. (3.58):

Specimen Thickness/plastic Zone Size = (a or B)/ry = 2.5(67T) = 47

The plane strain plastic zone shape and size were derived in Sections 3.5 and 3.6.
The Irwin and Von Mises yield criteria described by Eq. (3.33) of Section 3.5
were utilized in determining plastic zone size and shape.

3.B.l.l Standard K Ic Test Specimen and Fatigue Cracking

Two common types of standard specimen are available for K, c testing, namely
the slow-bend test specimen and the compact-tension specimen (CTS or simply
CT). The CT specimen is more commonly used for K,c testing than the slow-bend
test specimen (see Fig. 3.56). Other specimen configurations are also available for

A .. 0.5W
B = WI2 ± O.OlW
D < W/l0
G = 0.275W ± 0.005W
Hl2 = 0.6W ± 0.005W
L = 1.25 W ± 0.01 OW
2R = 0.25W ± 0.005W
W= W± 0.005W

Figure 3.56 Plane strain fracture toughness compact tension specimen (CT)
166 Chap. 3 Linear Elastic Fracture Mechanics

K,c testing and the reader may refer to the ASTM-E399 Standards. The corre-
sponding dimensions for the CT specimen are shown in Fig. 3.56.
Specimen dimensional measurements must be such that the thickness, B, is
measured to 0.1 %, and the crack length measurement after fracture is to the near-
est 0.5%. The crack length measurement should be the average of three measure-
ments at three positions along the crack front.
The crack is introduced in the specimen by a starter notch that extends by
fatigue cracking the notch. The purpose of fatigue cracking is to simulate a natural
crack that can provide a satisfactory plane strain fracture toughness test result.
The stress ratio associated with cyclically loading the notch is 0.1 > R > - 1 and
the Kmax (for fatigue cracking) < 60% K,c' It is common to fatigue-crack the CT
test specimen by the amount 0.05 W, where W is the width of the specimen. Annex
A2 of ASTM-399 outlines the procedure for fatigue cracking.
Note that before the plane strain fracture toughness specimen can be machined,
some estimate of the sizing of the CT test specimen must be known. Table 3.3
provides the ASTM recommended crack length or thickness for the suggested
ratio of yield strength to modulus of elasticity. For example, 2219-T851 alu-
minum alloys with yield strength of 53 ksi give a (J"ys / E ratio of 0.0053. The rec-
ommended crack length or test specimen thickness, B, is approximately 3.0 in.
(see Table 3.3).
To establish that a valid K,c has been obtained, it is first necessary to calculate
the conditional stress intensity factor, KQ , which can be determined through the
construction of test records consisting of an autographic plot of the applied load,
P, versus displacement, Ll, (P-Ll curve). The conditional stress intensity factor,
KQ , is given [42] as:

Table 3.3 ASTM Recommended Crack Length or

Thickness for K 1c Testing

Recommended crack
uys/E length or thickness

0.0050--0.0057 3.00
0.0057-0.0062 2.50
0.0062-0.0065 2.00
0.0065-0.0068 1.75
0.0068-0.0071 1.50
0.0071-0.0075 1.25
0.0075-0.0080 1.00
0.0080--0.0085 0.75
0.0085-0.0100 0.50
0.0100 and Greater 0.25
 3.8 A Brief Description of ASTM Fracture Toughness Determination 167

Displacement, ~

Figure 3.57 Load versus displacement for obtaining the apparent KQ

PQ
KQ = BW 1/ 2 29.6 W
l
(a)I/2 - 185.5 (a)3/2
W + 655.7
(a)S/2
W (3.59)

-1017.0 W (a)7/2 + 638.9 (aW )9/2J

where the applied load, PQ' is the point on the P-t:.. curve shown in Fig. 3.57. The
point P Q is obtained by drawing a secant line with slope 5% from the initial slope
tangent to the linear part of the record. The obtained value of PQ on the P-t:.. curve
is compared with the P max value situated on the curve (see Fig. 3.58 for different
types of load displacement records) to calculate the apparent plane strain fracture
toughness, KQ , to verify the conditions established by Eqs. (3.57) and (3.58). In
Fig. 3.58, the value of Ps is considered to be equal to PQ if there is no other load

(1) (2) (3)

D..

'ci'
III
o
...J

Displacement, t. ~ Displacement, t. ~ Displacement, t. ~

Figure 3.58 Load versus displacement for different possible types of behavior and the
interpretation of KQ
168 Chap. 3 Linear Elastic Fracture Mechanics

in the P-b. record that precedes it and is higher than Ps (see case 1 of Fig. 3.58).
Otherwise, the PQ load is the higher value on the load displacement record as
shown in cases 1 and 2 of Fig. 3.58. If the ratio of Pmax / PQ (the load P max is a load
point situated above the Ps value) is less than 1.10, then the calculated KQ value
from Eq. (3.59) should be used in Eq. (3.57) to compute the plastic zone size, ry.
This value should be compared with specimen thickness, B, and crack length, G. If
both Band G are larger than ry' then KQ = K Ic . In addition to the above mentioned
conditions for obtaining a valid KI c value, the following two items are also impor-
tant to know when conducting the plane strain fracture toughness testing:

• It is necessary to perform a minimum of three tests for each material heat treatment.
• The loading rate should be between 30 ksi (in./min.) 1/2 to 140 ksi (in./min.) 1/2

The plane strain fracture toughness, KI c' values for several selected aerospace
alloys are available in Appendix A. The KI c values for aluminum alloys may
range as high as 46 and as low as 16 ksi (in. 1/2) for 7075-T63 and 2020-T651 alu-
minums, respectively. Plane strain fracture toughness as high as 200 ksi (in. 1/2)
can be found in the NASAIFLAGRO material library among the stainless steel
alloys (see Appendix A). In general, a high K Ic is associated with ferrous alloys
that have undergone appropriate heat treatment which can introduce sufficient
ductility and high resistance to fracture in the material.
Plane strain test specimen thickness as large as 12 in. is reported in the litera-
ture for alloys that possess high ductility and good resistance to fracture [34]. The
specimen preparation for determining the plane strain fracture toughness value
when the thickness requirement is as high as 12 in. is costly and not feasible to
implement. On the other hand, the flexibility of using the FMDM theory in eval-
uating the fracture toughness value will eliminate the difficulties associated with
specimen preparation and time-consuming laboratory testing (see Chapter 6).
Another important point that must be emphasized in conjunction with the plane
strain fracture toughness value is its variation with respect to the grain orientation
that evolves during the manufacturing process. The plane strain fracture tough-
ness values for material that has undergone a forging process are usually expected
to be smaller in the T-L direction than the ~T. For example, the K lc values for
7050-T7452 aluminum alloy in the forged condition are 31 and 21 ksi (in. 1/2) in
the L-T and T-L directions, respectively. As another example, the KIc values for
7050-T76511 in the extrusion condition are 30 and 24 ksi (in. 1/2) in the L-T and
T-L directions, respectively (see Appendix A). From these examples, the analyst
must recognize that the proper K Ic value must be used in assessing the service life
of the structural part. Not recognizing the differences in material properties with
respect to their orientation (and these may differ considerably) can lead to erro-
neous analysis results.
 3.8 A Brief Description of ASTM Fracture Toughness Determination 169

3.8.2 Plane Stress Fracture Toughness (KcJ Test

In contrast to plane strain fracture toughness, K, c' the critical stress intensity fac-
tors for plane stress and mixed mode conditions, Kc ' are thickness and crack
length dependent. Many aircraft and space vehicle structures are made of thin
sheets of aluminum alloy. The life assessment of these structures by a fracture
mechanics approach requires knowledge of the plane stress fracture data that are
obtainable through laboratory testing. The replacement of the Kc by K,c value
yields conservative results (K, c < Kc) when evaluating the life of the part which in
many cases leads into thickening the part to reduce the design stress. The variation
of material fracture toughness, Kc ' versus specimen thickness was discussed in
Section 3.3 and is shown in Fig. 3.59 for 2219-T87 aluminum alloy [33]. From
Fig. 3.59, it can be seen that the fracture toughness value increases as the thick-
ness decreases. The increase in fracture toughness value is strongly dependent on
the state of stress at the crack tip and the extent of plasticity formation in that
region. For a thick section, the extent of the plastic zone through the thickness is
at a minimum, and as the thickness decreases, the constraint to plastic flow
decreases, until the state of plane stress is reached. The fracture toughness associ-
ated with minimum thickness is called the plane stress fracture toughness, Kc. The
maximum Kc value for 2219-T87 aluminum alloy is 70 ksi (in. 1/2) and is associ-
ated with a minimum thickness t = 0.15 in. and half crack length a = 1.5 to 1.6 in.

100 .-----------------------~--------------------~

M(T) Specimen Dala, a b 112

Crack Length
80 o ~.24toO.38

• a-o.45 to 0.68

-
....1.00 to 1.20
...
4
....1.50 to 1.60
....
C 60
j?

.
II:
CJ
U)

"i 40
:!. o

NASAlFlagro Values
20
ICc
YIeld
to
. Z7
58
0.54
Ale 1
Bk 1
0
0 0.5 1.0
Material Thickness, t (inch)
Figure 3.59 Variation of Kc as a function of thickness, t, for 2219· T87 aluminum alloy
170 Chap. 3 Linear Elastic Fracture Mechanics

(as shown in Fig. 3.59). The smallest Kc value (with t = 0.15 in.) in this figure is
44 ksi (in. 1/2) and it applies to a half crack length of 0.24 to 0.38 in.
To develop a curve similar to the one shown in Fig. 3.59 for another alloy or
the same alloy with another heat treatment, it is necessary to conduct several tests
with cracked specimens of various thicknesses. Each specimen will provide a dis-
tinct R-curve different from the others and represents material resistance to frac-
ture for the thickness under consideration. The R-curve represents the material's
resistance to fracture during incremental stable slow crack growth, Lla, under
monotonic increasing load. As indicated in Section 3.4.3, the R-curve is con-
structed by plotting the material resistance to fracture, Kc = KR , for different
crack lengths, (see Fig. 3.60). Because the R-curve is proven to be independent of
original crack length, ao ' it can be plotted as a single curve (KR versus Lla) for a
given thickness using a standard ASTM cracked specimen as recommended by
ASTM E-561 (Standard Practice for R-curve Determination).
In the R-curve concept, unstable crack extension occurs when the applied K
becomes equal to or exceeds the crack growth resistance, K R , of the material. At
the instability point, the crack growth resistance, KR , is equal to Kc as shown in
Fig. 3.60. That is, the plane stress fracture toughness, Kc, is determined from the
tangency between the R -curve and the applied K curve (K versus physical or
effective crack length, ao + Lla) of the specimen. The specimen configuration and
dimensions specified by the ASTM testing procedure for conducting a valid R-
curve test are briefly discussed in Section 3.8.2.1.
The variation of fracture toughness, Kc, with material thickness for many aero-
space alloys can be found in the NASA/FLAGRO material library. This variation
is presented in the form of an empirical equation described in terms of plane strain
fracture toughness, K 1e , [39] as:

Ke = K Ie [1 + Bk e-(A,*tIB)2] (3.60)

...
o
II:
~

Crack Length, a

Figure 3.60 Construction of the R·curve and determination of the instability point
where the applied K and KR are tangent to each other
 3.8 A Brief Description of ASTM Fracture Toughness Determination 171

where B is the thickness that meets the plane strain condition and t is the thickness
associated with the part that could be in the plane stress or mixed mode condi-
tions with fracture toughness, Kc ' to be determined. The constants Ak and Bk are
the curve fit parameters and are given in Appendix A for several aerospace alloys.
Note that the curve fit relationship described by Eq. (3.60) is applicable to mini-
mum crack length, as was shown in Fig. 3.36 for the case of 2219-T87 aluminum
alloy with half crack length a = 0.24 to 0.38 in. The plane stress fracture tough-
ness values associated with minimum crack length will yield a minimum fracture
toughness value that is conservative to use when evaluating the service life of a
fracture critical component.

3.8.2.1 M(T) Specimen/or Testing Kc

For determination of material resistance to fracture by using the R -curve
approach, three types of specimen are recommended by the ASTM committee.
The middle center cracked wide tension panel M(T) specimen, the compact spec-
imen C(T), or the crack-line-wedge-Ioaded specimen C(W) are used to conduct
the plane stress fracture toughness test. Only the center cracked wide panel is dis-
cussed in this section (see Fig. 3.61). The reader may refer to E561 for the rec-
ommended specimens that can also deliver applied stress intensity factor, K, to
the material. Section 3.10.2.2 briefly describes (1) the apparatus (grips and fix-
ture) employed for uniformly transferring the applied load into the M(T) standard
specimen, (2) the method of preventing buckling for the M(T) specimen, (3) spec-
imen fatigue cracking, and (4) load and specimen dimensional measurements.
As discussed in the previous section for the plane strain fracture toughness case
(per the ASTM E-399 test method), the planar dimensions of the specimens are
sized to ensure that elastic conditions are met at all values of load in the net sec-
tion of the specimen. In other words, the width, W, and half crack size, a, must be
chosen so that the remaining ligament is below net section yielding at fracture.
The ASTM guide for designing an M(T) panel is such that a specimen with
W = 27 ry width and crack length 2a = wl3 is expected to fail at a net section

-~

2 Holes (W/3 Dia.)

~cp $~:ra CD W

I I

iIii
...;'-

..t
1;'
wl2
1.SW
III
1.SW I wl2 I
I:I
,
Figure 3.61 Standard center-cracked-tension M(T) specimen
172 Chap. 3 Linear Elastic Fracture Mechanics

Table 3.4 Specimen Width Size and Corresponding Crack Length for an
M(T) Specimen.

Kmax/O"ys Width (W) Crack Length (2a o) Spec. Length

(in. 1/2) (in.) (in.) (in.)

0.5 3.0 1.0 9.0

1.0 6.0 2.0 12.0
1.5 12.0 4.0 24.0
2.0 20.0 6.7 30.0
3.0 48.0 16.0 72.0

stress equal to the yield strength of the material. The selected width dimension
must be larger than 27 ry to avoid net section yielding. Table 3.4 is a list of mini-
mum recommended M(T) specimen dimensions for an assumed Kmax /O"yield ratio,
where Kmax is the maximum K level obtained in the test and O"yield is the 2% offset
yield strength of the material. Note that for the plane stress condition, the value of
ry is given in terms of Km• x and O"yield by Eq. (3.33) as:

1 (K )2max
(3.61)
rp = 27T O"yield

For an M(T) specimen, the Kmax can be calculated by:

(3.62)

3.8.2.2 Grip Fixture Apparatus, Buckling Restraint, and Fatigue Cracking

The grip fixture apparatus must distribute the applied load to the M(T) test speci-
men uniformly to enable one to obtain accurate readings for constructing a valid
R-curve. To ensure that uniform tensile stress is transferred from the pin grip fix-
ture (when a single pin grip is used) into the specimen, the length between the two
single pins on each end shall be 3W (as shown in Fig. 3.62). For example, for a
test specimen with width W = 3.0 in. and crack length 2a = 1.0 in., the length
requirement between the application of loads at the two ends is 9.0 in. In the case
when the specimen width, W, is wider than 12 in. (as indicated in Table 3.4) the
ASTM requires use of multiple pin grips to uniformly distribute the load, with a
length requirement of only 1.5W, as illustrated in Fig. 3.62. For example, for
W = ':0 in., the length requirement is 1.5W = 30 in.
When the test specimen is thin, buckling can develop during loading that can
affect the accuracy of the test data. To prevent the test plate from buckling, rigid
face plates are required to be attached to both sides of the specimen. In the case of
the M(T) specimen, the rigid plates shall be affixed to the test specimen as illus-
trated in Fig. 3.63. The lubricated buckling restraint plates must be attached to
 3.8 A Brief Description of ASTM Fracture Toughness Determination 173

Crack Length
~
I 2a

3W

~ 2a
W>12
inch ~

• 1.5W •
Figure 3.62 The pin grip fixture apparatus for W = 3 and W > 12 in. M(T) specimens

M(T) Specimen

Buckling restraints
Plate

c::::::-
2a
Buckling restraints
Plate

Figure 3.63 Buckling restraint plates affixed to the cracked specimen

both sides of the test specimen in such a way that the crack measurement data
readings are accessible and are not blocked.
To simulate a natural crack in the M(T) specimen, the starting notch is first
introduced in the specimen either by saw cutting or electrical discharge machining
(EDM). The starting notch length plus fatigue precrack should be between 30%
and 40% of the specimen width, W, and situated in the center of the test specimen.
The extended length associated with the fatigue pre crack must not be less than
0.05 in. Fatigue cracking may be eliminated if the cutting saw thickness can sim-
ulate the sharpness of a fatigue starter crack. The ASTM procedures for obtaining
174 Chap. 3 Linear Elastic Fracture Mechanics

a valid fracture toughness value clearly indicate that extensive preparation both
before and after testing is necessary.

References

I. A A Griffith, "The Phenomena of Rupture and Flow in Solids," Philos. Trans. R. Soc.
Lond., Ser. A, Vol. 221,1920.
2. AS. Tetelman and A J. McEvily, Jr., Fracture of Structural Materials, John Wiley &
Sons, 1967 pp. 39-48.
3. G. R. Irwin, "Analysis of Stresses and Strains Near the End of a Crack Traversing a
Plate," Trans. AS ME, J. Appl. Mech., Vol. 24,1957, p. 361.
4. C. P. Paris and G. C. Sih, "Stress Analysis of Cracks," in "Fracture Toughness Testing
and Its Applications," ASTM STP No. 381, ASTM, Philadelphia, 1965.
5. H. M. Westergaard, "Bearing Pressures and Cracks," Trans., ASME, J. Appl.
Mech., 1939.
6. S. P. Timoshenko and 1. N. Goodier, Theory of Elasticity, 3rd Edition, McGraw-
Hill, 1970.
7. N. 1. Muskhelishvili, "Some Basic Problems of the Mathematical Theory of Elastic-
ity," (1933), English Translation, Noordhoff, 1953.
8. H. Tada, P. C. Paris, and G. R. Irwin (Eds.), Stress Analysis of Cracks Handbook, Del
Research Corporation, Hellertown, PA, 1973.
9. G. C. Sih, Handbook of Stress Intensity Factors for Researchers and Engineers, Insti-
tute of Fracture and Solid Mechanics, 31, Series E, No.2, June 1964.
10. G. R. Irwin, Fracture Handbuch der Physik, Springer-Verlag, Heidelberg, VI, 1958,
pp.551-590.
11. D. P. Rooke and D. J. Cartwright, "Compendium of Stress Intensity Factors," Her
Majesty's Stationary Office, London, 1976.
12. Fatigue Crack Growth Computer Program "NASA/FLAGRO," developed by R. G.
Forman, V. Shivakumar, and 1. C. Newman. Appendix C, Reference CI-C27, JSC-
22267A, January 1993.
13. Damage Tolerant Design (Data) Handbook, Revision to MCIC-HB-Ol, (1975), cov-
ered under USAF Contract No. F33615-80-C-3229.
14. "1996 Annual Book of ASTM Standard," Metals Test Methods and Analytical Proce-
dures, Vol. 03.01.
15. J. M. Craft, A M. Sullivan, and R. W. Boyle, "Effects of Dimensions on Fast Fracture
Instability of Notched Sheet," Crack Propagation Symposium, Cranfield 1961, Paper 1.
16. A P. Parker, "The Mechanics of Fracture and Fatigue," E. & F. N. Spon Ltd.,
pp.101-117.
17. M. F. Kanninen, and A T. Hopper, "Advanced Fracture Mechanics," Oxford Engi-
neering Science Series, 1985, pp. 150--200.
18. 1. M. Barsom, "The Development of AASHTO Fracture Toughness Requirement for
Bridge Steel," Engin. Fract. Mech., Vol. 7, No.3, September 1975.
 References 175

19. 1. M. Barsom, "The Development of AASHTO Fracture Toughness Requirement for

Bridge Steel," American Iron and Steel Institute, Washington, D.C., February 1975.
20. The Making, Shaping and Treating of Steel, Ninth Edition, edited by H. E. McGannon,
United State Steel, Pittsburgh, December 1970.
21. American Association of State Highway and Transportation Officials (AASHTO) Mate-
rial Toughness Requirement, Association General Offices, Washington, D.C. 1973.
22. "PVRC Recommendations on Toughness Requirements for Ferritic Materials," WRC
Bulletin 175, New York, August 1972.
23. S. T. Rolfe and 1. M. Barsom, "Fracture and Fatigue Control in Structures, Applica-
tions of Fracture Mechanics," Prentice-Hall, 1977.
24. R. A. Flinn and P. K. Trojan, "Engineering Materials and Their Applications," Third
Edition, Houghton Mifflin, 1986, pp. 321-323.
25. W. S. Margolis and F. C. Nordquist, "Plane Stress Fracture Toughness of Aluminum
Alloy 7475-1/2 in. Plate, Temperes-T7651 and T7351 and of Aluminum Alloy 2024-
1/8 in. Sheet-T81 and T62 Temper," General Dynamics, Forth Worth Div., F-16 Air
Combat Fighter Technical Report TIS GA2300, CDRL A031, USAF Contract
F33657-75-C-0310.
26. J. E. Srawley and W. F. Brown, "Fracture Toughness Testing Method," ASTM STP
381, 1965,pp. 133-195.
27. 1. M. Krafft, A. M. Sullivan, and R. W. Boyle, "Effect of Dimensions on Fast Fracture
Instability of Notched Sheets," Cranfield Crack Propagation Symposium, Vol. 1,
1961, pp. 8-28.
28. 1. C. Newman , Jr. "Fracture Analysis of Surface and Through Cracked Sheet and
Plates," Engineering Fracture Mechanics, Vol. 5, No.3, Sep. 1973 pp. 667-684.
29. D. P. Wilhem, "Fracture Mechanics Guidelines for Aircraft Structural Applications,"
AFFDL-TR-69-111, Air Force Flight Dynamics Laboratory, February 1970.
30. M. P. Kaplan and 1. A. Reiman, "Use of Fracture Mechanics in Estimating Structural
Life and Inspection Intervals," J. Aircraft, Vol. 13, No.2, February 1976, pp. 99-103.
31. F. A. McClintock and G. R. Irwin, "Plasticity Aspects of Fracture Mechanics," ASTM
STP 381,1965, pp. 84-113.
32. A. R. Duffy et aI., "Fracture Design Practice for Pressure Piping, Fracture I,"
pp. 159-232, edited by H. Liebowitz, Academic Press, 1962.
33. D. P. Rooke, "Elastic Yield Zone Around a Crack Tip," Royal Aircr. Est., Farnbor-
ough, Tech. Note CPM 29,1963.
34. W. Johnson and P. B. Mellor, Plasticity for Mechanical Engineers, Van Nostrand, 1962.
35. G. S. Spencer, An Introduction to Plasticity, Chapman and Hall, 1968.
36. G. R. Irwin, "Crack Extension Force for a Part Through Crack in a Plate," J. of Appl.
Mech., December 1962, pp. 651-654.
37. R. M. Engle, Jr., "Aspect Ratio Variability in Part-Through Crack Life Prediction,"
ASTM STP 687, Am. Soc. Test. and Mater., 1979, pp. 74-88.
38. J. C. Newman, Jr. and I. S. Raju, "Stress Intensity Factor Equations for Cracks in
Three Dimensional Finite Bodies," NASA TM 83200, NASA Langley Research Cen-
ter, August 1981.
176 Chap. 3 Linear Elastic Fracture Mechanics

39. Fatigue Crack Growth Computer Program "NASA/FLAGRO," Developed by R. G.

Forman, V. Shivakumar, and J. C. Newman, JSC-22267 A, January 1993.
40. ASTM Committee, "The Slow Growth and Rapid Propagation of Cracks," Mater. Res.
Stand., 1, 1961, pp. 389-394.
41. G. R. Irwin, Fracture of Pressure Vessels: Materials for Missiles and Spacecraft,
pp. 204-229, McGraw-Hill (1963).
42. J. E. Srawley, "Wide Range Stress Intensity Factor Expressions for ASTM E 399 Stan-
dard Fracture Toughness Specimens," Int. Fract. Mech., Vol. 12, June 1976, p. 475.
Chapter 4

Fatigue Crack Growth

4.1 Introduction

As discussed in Chapter 2, the traditional approach to estimate the total life of a

structural part is to use the S-N diagram (alternating stress, S, versus the total
number of cycles to failure, N). This diagram can be constructed by conducting
several laboratory tests on standard specimens subjected to low amplitude fluctu-
ating loads simulating the actual environment. With the S-N approach, it is
assumed that the structure is initially free from defects and that the total life of the
structure is defined as the sum of the crack initiation and propagation cycles. Fur-
thermore, it was mentioned that in the localized region adjacent to the notch
where material is plastically deformed, small fatigue cracks can initiate along the
persistent slip band which eventually coalesce to create larger cracks (stage I
growth). The number of cycles required to generate a small crack, along the active
slip plane, is dependent on the load amplitude, material grain size, and tempera-
ture. Under low load amplitude, the extent of stage I is large and most of the struc-
turallife is consumed for crack initiation. On the other hand when load amplitude
is raised to the low cycle regime where the bulk of the structure is plastic, the
number of cycles used for stage I is small and the entire structural life is expended
for stable crack growth (stage II). In the high cycle fatigue regime, the plastic
deformation is limited to a small region at the tip of the crack, and estimation of
the number of cycles to failure from the end of initiation stage (i.e., the start of
stage II) can be determined from the linear elastic fracture mechanics approach.
The application of linear elastic fracture mechanics in estimating the life of
structural hardware assumes that the crack has already initiated in the material. In
reality, this is often the case, because the initial cracks may have their origin in
many ways. For example, they may be introduced as cracks or as incipient cracks
during the manufacture of structural parts, they may grow from defects in the par-

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts 177
© Chapman & Hall 1997
178 Chap. 4 Fatigue Crack Growth

ent metal, from incomplete welds, or from shrink cracks or other imperfections in
we1dments. The initial crack length associated with these cracks can be estimated
conservatively based on the Nondestructive Inspection (NDI) method. That is, the
largest crack that could escape detection by nondestructive inspection can be
assumed to exist in the material and it can be used as the initial crack size for life
evaluation of the structural components.
The material as received from the vendor will contain defects of a small size,
such as porosities (fine holes and pores in a metal, most commonly in welds and
casting); inclusions (such as oxides, sulfides, and silicates); or microcracks that
can eventually lead to fracture. These inherent flaws are considerably smaller than
the NDI capability to detect them and will not grow appreciably in service. From
a safety viewpoint, the use of a longer initial crack length (obtained through NDI
methods) is conservative, however, from a practical consideration, the longer
crack size assumption should be realistic enough not to impact the weight or cause
rejection of the part. It should be noted that the initial crack size to be used to
evaluate the crack growth behavior must not be so small that it would violate the
concepts of fracture mechanics. That is, the concept of isotropic continuum must
be obeyed, such that the flaw size cannot be smaller than the grain size.
For aircraft and space vehicles, it is required to assume that cracks exist in all
structures that are classified as high-risk or fracture critical parts. These cracks
shall not grow to their critical size at a specified load during their usage period.
Therefore, it is necessary to predict the rate of growth of the assumed flaw and the
number of cycles to failure compared with the number of cycles used during their
service life.
Different crack growth empirical equations capable of calculating the number
of cycles to failure are discussed in Section 4.2. Section 4.2 includes the crack
growth equation employed in the NASA/FLAGRO (NASGRO) computer code
[1]. This program was initially developed for fracture analysis of space hardware
and recently has been extended to crack growth analysis of aircraft structural
problems. The NASA/FLAGRO computer program is currently the standard com-
puter code for the NASA, the European Space Agency, the U.S. Air Force, and
many aerospace companies. A brief review of the ASTM fatigue crack growth
rate testing is also included in this section. Later in Section 4.3, the Elber and
Newman crack closure concepts (used in crack growth equations) are presented
and are followed by a few example problems. Variable amplitude loading and the
retardation concept (including example problems) are discussed in Section 4.4.
Structural integrity analysis of bolted joints under cyclic loading is covered in
Section 4.6. This section covers an in-depth discussion on preloaded bolts sub-
jected to cyclic loading and includes several example problems. Structural
integrity analysis of a crack emanating from a hole in a bolted joint SUbjected to a
fluctuating load environment is also included in this section. Finally, material
 4.1 Introduction 179

anisotropy and its effect on bolted joint analysis when evaluating the life of the
structure is reviewed in Section 4.7.

4.1.1 Stress Intensity Factor Range

and Crack Growth Rate
It is the goal of fracture mechanics to estimate the total number of cycles for the
assumed or the initial crack, ai' to reach its final length, at. The crack tip stress
intensity factor was extensively discussed in Chapter 3 and is an extremely useful
parameter to address crack growth behavior as long as the bulk of the material is
elastic and plastic deformation is limited to a small region at the crack tip. The
stress intensity factor range associated with maximum and minimum stresses of
each cycle (Kmax - Kmin ) is used to describe the amount of crack advancement,
I:l.a = at - ai' during !:IN number of cycles. In using fracture mechanics to
describe fatigue crack growth the minimum value of the stress intensity factor,
Kmin , in a cycle is taken to be zero when R ~ O. The rate of crack growth, l:l.a/!:lN,
in terms of the crack tip stress intensity factor range, I:l.K, can be written as:

I:l.a
I:l.N = /(I:l.K) (4.1)

Equation (4.1) simply states that the material's rate of crack growth, l:l.a/!:lN, is a
function of the stress intensity factor range. The function/(I:l.K) can be obtained
as the result of laboratory test data and can then be utilized to solve crack growth
problems in which the structural part has undergone the same loading conditions
(R = u min / U max)' That is, the crack growth behavior of structural parts can be esti-
mated by testing small specimens in laboratory that simulates the same cyclic
loading conditions. It should be emphasized that when crack tip stress states at a
given time are the same for two separate crack geometry and loading conditions,
I:l.K, the crack growth rate, l:l.a/!:lN, would almost be the same for the two crack
cases. For example, consider the case of a wide center crack plate with initial
crack length of 1.5 in. subjected to a remote constant fluctuating load (R = 0) of
15 ksi. The rate of crack growth for this crack geometry will be the same as if the
crack length is 0.5 in. and subjected to a remote loading of 26 ksi (R = 0):

(note that u min = 0 and therefore, K min = 0)

I:l.K = Km • x = f3umax (7Ta) 1/2
for u max = 15 ksi, Kmax = 15 X (3.14 X 1.5)112 = 32.6 ksi (in. 1/2 )

forumax = 26ksi, Kmax = 26 X (3.14 X 0.5)1/ 2 = 32.6ksi (in.1/2)

180 Chap. 4 Fatigue Crack Growth

In the next section, several empirical relationships will be developed that can
define the crack growth behavior shown by Eq. 4.1.

4.2 Crack Growth Rate Empirical Descriptions

All of the equations that presently describe the functionj(dK) are based on the
trends developed by experimental data. In general, a centered crack specimen with
original crack length, ao ' is subjected to constant cyclic loading of a given stress
ratio, R (see Fig. 4.8 for other crack geometries used in generating crack growth
data). The incremental crack length growth, da, is periodically measured and
recorded together with the number of cycles (!IN). From this information, the
variation of crack advancement versus associated number of cycles is plotted as
shown in Fig. 4.1 which represents the experimental fatigue crack growth data.
Figure 4.2 shows the variation of crack growth, ao + da, versus the number of
cycles, N, for an aluminum lithium alloy tested in accordance with ASTM E-647
practice. The slope of the curve (dal dN = dal!lN) at a point shown in Fig. 4.1 is
computed for any crack length, a = ao + da, along the curve and is called the
fatigue crack growth rate or crack extension per cycles of loading. Currently there
are two methods recommended by the ASTM for determination of the slope,
dal!lN. In the secant method of determining the slope, two data points are needed
per slope, whereas in the incremental polynomial method a minimum of five data
points are required to compute the slope at a point, dal dN [2] (see Fig. 4.3). The
computed slope and the corresponding stress intensity factor range at a point are to
be used in generating the daldN, dK curve. In the secant method of determining

a(t) • Experimental Test Data :

I a (t) Max

/~
A,N;) M;"

......'
G(t) ••••• da/dN

Number of Cycles, N

Figure 4.1 Schematic representation of fatigue crack growth curve data and slope at
a given point
 2.00 1
Spec men ID: FCGR-LT-l/12
Test Temperature(F): Room
---1----'----1- I • -.-.

1 1 1 1
II-----I--r--r--I--I~~~J---I-t t-=1--=---r=-~I=-=--[==~~
I-
1. 681-1--f-+--+---f-+--+---f-+--+--II--+--+-I---+--+-I---t---+-t--f ---1--.---.----.--- -.-.-.-- .... --- .--
_·--+--I--t--
I I t=I I -I-l-I--+--t--I-I--I-----'-~--t---l--·-··t---;-··-I-----
_· __1..__ ·.·_ ---I--.~- - - .---. ____L - l - J. __ ---- --1--- --.-.......
;r;
... - - + - - t - f - - l - - - - \ - · · - I - - - - I - - + - - l - - - l - I - - - l - - --1--1--1---/-----·1·--+-·-1----
. 1.36111-1---r--t--j-+-i-+-+-1~+-+-~+-+-~-l---+-J-J
1=1 I ~-.-I_I=j-.j I t=1--t--j-+--t-t·-r-I·---r-----C-·--
--1-·--- .---~---_t-
_--+-----+ ___L_*_ --
t---+--I--+-----t---I--I-I-I--I--I---~--_I__-I- ·1-1-+- -- .......-1----- .._----
I '---1---1-- - - --I·
1. 04~ 1 --+---. ---- -- - _._ ~=l7 ~____=.~
~ I ---t--I---l--+--t- .--
-- -- . P-- ·f=1
-.J---f--t---~... -- .
+-± +++t . _ . _ .
. 1-
721_~--~--t-_i --- - -
-. __ 1-.____ -+ L...f--++1'ft-+ I"-'-
t _1_~_+._1- ... _________ _ 1----j-3=±
. - - . - - . - 1 - ...-+-;- -t----- --
J--~----,I--;-...-t-+-+--J±_!.. _+ ____ .___ ____ - - - -
_ __
- -.___
1.____
- -_ _1_-_
- -_- - . -__ .

.40 1 I 1 1 1
o 500,000 1,000,000

CYCLES (N)

.... Figure 4.2 Variation of crack advancement (lia) versus number of cycles for alu-
<Xl
.... minum lithium alloy (courtesy of Westmorland Company)
182 Chap. 4 Fatigue Crack Growth

Secant Method (slope evaluated by connecting Incremental Polynomial Method (slope

two adjacent data points by a straight line) evaluated by polynomial fit to five, six or
seven successive data points)
aj +~a

t.K= I!(cmax-"min)(""ave) 1/2

•
•
•

Number of Cycles

Figure 4.3 Determination of slope at a point by secant and incremental polynomial

methods

slope at the midpoint of two data points, one data point has to be sacrificed. That
is, from the two data points, a i and ai + 1, as shown in Fig. 4.3, only one point is
generated to describe the da/dN, D..K curve ((a i + ai + 1)/2). In the incremental
polynomial method, a smooth curve (a parabola) is fit through five, seven, or nine
data points for the determination of the slope, da / dN. With this method, more than
two data points are lost when generating the da/ dN, D..K curve. The logarithmic
plot of the calculated stress intensity factor range, D..K, versus the corresponding
crack growth rate, da/dN, is shown in Fig. 4.4. The crack growth rate curve pro-
duced in Fig. 4.4 is independent of load magnitude, <ret), and crack geometry .

........
Z
'C
.....
III
'C
'-'
Ol
a
....I

• Experimental Test Data

Best Curve Fit

Figure 4.4 Illustration of crack growth rate versus stress intensity range
 4.2 Crack Growth Rate Empirical Descriptions 183

Numerous fatigue crack growth rate empirical and analytical relationships

have been developed and are available in [3]. The earliest relationship describing
the da/dN behavior (shown in Fig. 4.4) was formulated by Paris and his col-
leagues at Lehigh University in 1960 [4] and expressed as:

da
- = c(dK)m (4.2)
dN

where c and m are constants that can be determined from the test data. Figure 4.5
shows typical crack growth data and the curve fit for 2024-T861 aluminum alloy
(plate and sheet, T-L, room temperature) that was generated under stress ratio
R = 0.1 (taken from the NASA/FLAGRO material library [1]). The constants c
and m describing the Paris relation are 3.4 and 4.8E-8, respectively. For most met-
als, the Paris constants m and care: m = 3 to 5 and c = 10- 10 to 10- 6, respec-
tively. Fatigue crack growth data presented in Fig. 4.5 were utilized to compute
the constants used in the crack growth rate curve fit relation described by Eq. (4.5)
for 2024-T861 aluminum alloy [1].
The Paris relation described by Eq. 4.2 is applicable only to the middle region
ofthe crack growth curve, where the variation oflog (da/ dN) with respect to log
(dK) is linear, as indicated in region II of Fig. 4.6. In general, there are three
regions associated with the crack growth curve. In region I, the crack growth rate
(da/ dN) is small and the corresponding stress intensity range, dK, approaches a
minimum value called the threshold stress intensity factor, dKth , below which the
crack does not grow. The value of dKth is not associated with da/dN = 0; rather,
it is associated with a cutoff growth rate of 4 X 10 -10 m/cycle assigned by
ASTM- E647. In region III, the crack growth is rapid and accelerates until the
crack tip stress intensity factor reaches its critical value. The critical value of
stress intensity factor, Kc ' is shown in Fig. 4.6 as an asymptotic line to the crack
growth curve. The value of Kc obtained through cyclic loading (the critical value
of dK obtained from the da/ dN curve) is usually smaller than the monotonic load-
ing case. It should be remembered that an invalid fracture toughness value (as a
result of improper fracture toughness testing) incorporated into the crack growth
equations can result in significant error when determining the life of the part in
consideration.
The number of cycles that a cracked structural part spends in each region of the
da/dN, dK curve (shown in Fig. 4.6) is different. More than 80% of the life of the
structure is usually spent in region I, where the existing crack is short and the
crack growth rate is small. Figure 4.6 illustrates the approximate percent life of a
typical aerospace alloy in regions I, II, and III.
Utilizing the Paris crack growth relation described by Eq. (4.2) beyond its limit
can result in life estimation error. Note that the Paris equation does not apply to
184 Chap. 4 Fatigue Crack Growth

Alloy: 2024 AL
Condition: T861
Environment: LA
Specimen: C(T)
Orientation: T-L
Frequency: 0.10

.....
Q)
u
?J
....... 10- 5
c:
:::. Curve Parameters
z
"C
Smax/cro = 0.3
....... a 1.5
I'G
"C Kc 37.751
at thk=O.02
C 3.09E-8
10- 6
n 2.895
p 0.5
q
Yield 72
Klc 19
AK
BK 1
DKO 2.2
Rei 0.7

t.K [ksi*SQRT (in.)]

10 100

Figure 4.5 Typical crack growth data and the curve fit for 2024-T861 (from
NASA/FLAGRO Material Library [1])
 4.2 Crack Growth Rate Empirical Descriptions 185

5-896

'-296 I
I Kc
Iv

Figure 4.6 Three regions of the fatigue crack growth curve and approximate per-
centage of life spent

regions I or III, where crack growth is slow or rapid. In addition, the effect of the
stress ratio, R, on fatigue crack growth was not considered in Eq. (4.2).
Another attempt to provide an empirical relationship for describing the crack
growth data was made by Walker [5]. Walker's crack growth equation is similar to
the Paris relation, but it accounts for the effect of the stress ratio, R, in the region II:

(4.3)

where the constants C, m, and p can be obtained through experimental data.

Forman [6] formulated an empirical equation describing the crack growth
behavior in regions II and III, including the effect of R:

da/dN = c(dKt (4.4)

(1 - R)Kc - dK

where c and n can be obtained through experimental data and Kc is the fracture
toughness of the material and is thickness dependent (see Section 3.3.1 of Chapter 3).
Equation (4.4) was later modified by Forman-Newman-de Koning (FNK) to
account for all the regions of the crack growth curve, including the stress ratio
and crack closure effects [1,7,8]:
186 Chap. 4 Fatigue Crack Growth

da
C(1 - fYD.K" ( 1 - ¥it r (4.5)
dN
(1 - R)" ( 1 - -(l-~-~-)K-cr
where C, n, p, and q are empirically derived constants; R is the stress ratio; and
D.K and D.Kth are the stress intensity factor range and threshold stress intensity fac-
tor, respectively. The parameter f is called the crack opening function and it will
incorporate the effect of closure behavior on crack growth rate under constant
amplitude loading. The Forman-Newman-de Koning (FNK) crack growth rate
relationship described by Eq. (4.5) is widely used in aerospace structures for life
estimation of high risk fracture critical parts.
A detailed discussion related to constant amplitude crack closure phenomena
and crack opening function, I, is given in Section 4.3.

Example 4.1
The crack growth curve for a given aluminum alloy is plotted in Fig. 4.7: (1) Eval-
uate the empirical constants associated with Eqs. (4.3) and (4.4) for the Paris and
Forman relationships, respectively, where Kc = 55 ksi (in. 1/2). (2) For a centered
crack plate with original crack length of ai = 1.0 in. and maximum applied stress
of 20 ksi, determine the number of cycles corresponding to a growth of D.a = 0.7
in. (R = 0).

Test
1E-2

1E-3

....... 1E-4
Z
""0
"-
III
""0
.......
Dl
0
-J
Paris Region of
Application
1E-6

1E-7

10 55 100
Log <t.K)

Figure 4.7 Crack growth curve fit for the aluminum alloy mentioned in the problem
 4.2 Crack Growth Rate Empirical Descriptions 187

Solution
To use the Paris equation for the middle region of the crack growth curve, the two
constants c and m must first be evaluated. Utilizing the two points on the linear
portion of the curve, the constants c and m can be determined as:

For da/ dN = IE - 4 in./ cycle, dK = 24.5 ksi (in. 1/2 )

For da/ dN = IE - 5 in';cycle, dK = 8.5 ksi (in. 1/2 )

Inserting the two da/ dN values into Eq. (4.3):

0.0001 = c(24.5)m
0.00001 = c(8.5t
10 = (2.88)m
m = 2.18 and c = 9.37 X 10- 8

The Paris equation can be written as:

da/dN = 9.37 X 10- 8 (dK)2l8

The Forman constants c and n [from Eq. (4.4)] can be obtained through the same
procedure, where R = O. The same two points on the crack growth curve (Fig.
4.7) are selected:

0.0001 = [c(24.5)"]/(55 - 24.5)

0.00001 = [c(8.5)"]/(55 - 8.5)
Solving for c and n:

c = 1.03 X 10- 5 and n = 1.78

For the second part of the problem:

where

Smax = 20 ksi, Smin = 4 ksi, f3 = 1, and ai = 1.0 in.

Assuming the quantity f3 remains a constant throughout the growth, the total num-
ber of cycles, dN, for the crack to grow 0.7 in. can be computed. Using the Paris
equation shown by Eq. (4.2):
188 Chap. 4 Fatigue Crack Growth

dN =J af [ 1 ] da
a, 9.37 X 10- 8 X 16 2.18 (7Ta?·18/2

where the total number of cycles M = 3759. Using the Forman equation [Eq. 4.4)]:

dN = f (1 - R)K - t::.K
1.03 X 1O-~(t::.K)1.78 da

= f (1 - R)Kc
1.03 X 1O- 5(t::.K)1.78 da -
f t::.K
1.03 X 1O-5(t::.K)1.78 da

where a j and at are 1.0 and 1.7 in., respectively. Simplifying the above integra-
tion, the total number of cycles M = 3084.

4.2.1 Brief Review of Fatigue Crack Growth Testing

The establishment of a complete da/ dN versus t::.K curve (as shown in Fig. 4.5)
for the determination of the number of cycles to failure of machinery parts
requires conducting steady-state fatigue crack growth rate testing in accordance
with ASTM E-647, "Standard Test Method for Measurement of Fatigue Growth
Rates." Two types of specimen configurations, Compact Tension, C(T), and Mid-
dle-Tension, M(T), are recommended for use in generating the da/dN versus t::.K
plot (see Figs. 4.8 and 4.9). This test method involves cycling notched M(T) or
C(T) specimens that have been precracked, while the incremental crack growth
length, t::.a, is periodically measured and recorded together with the number of
cycles, M. Other specimen configurations may be used, provided that stress
intensity factor solutions for the crack geometries are available and moreover, the
net section yielding is avoided.
In precracking the test specimen to simulate an ideal crack, the final Kmax value
associated with precracking should not exceed the initial Kmax value that the test
specimen will experience upon testing. Otherwise, the crack tip will be SUbjected
to the retardation effect at the onset of testing. The retardation effect will slow the
rate of crack growth due to overload (high to low load) and is discussed in Section
4.4. This effect is particularly important when the threshold stress intensity factor
range, t::.Kth , is recorded. The amount of precracking should not be less than 0.05
of the width of the test specimen. Electrical discharge machining (EDM), milling,
 4.2 Crack Growth Rate Empirical Descriptions 189

Dia = O.2SW ---.....

(:r-"" -~ -
-= 1
O.6W

-
--
- - O.6W
--

t
- --L-
w±.o.oosw ~I~~I B I~an =O.20W
1.2SWLO.01W W/ZO<B<W/4
(Recommended Thickness)

Figure 4.8 Standard compact tension C(T) specimen for crack growth rate testing
(all dimensions are written in terms of W.)

LB
or
2 Holes (W/3 Dia.)

~Q

w/2
~
I

I 1.SW
$9Ia
.... or
I 1.SW
G)
.... or
I
I

I w/2
r
If
I I: I
I
III I :I I I

Figure 4.9 Standard M(T) specimen for crack growth rate testing (all dimensions are
written in terms of W.)

broaching, or saw-cutting techniques can be used for notch preparation prior to

precracking. Precracking dimension requirements for the ASTM fatigue crack
growth test are shown in Fig. 4.10.
The incremental crack growth associated with N number of cycles is measured
either visually or with some other optical technique, and the growth rate, I:::..a/!:::..N,
as a function of the stress intensity factor range, I:::..K = Kmax - Kmin , is calculated.
The fatigue crack length reading technique, irrespective of the method of mea-
surement, must be capable of measuring as good as O.002W. To be capable of
reading crack extension visually to O.002W, a low power traveling microscope
with sox magnification capability is recommended. It would be helpful, although
not necessarily required, to apply reference marks at equal intervals on the test
specimen prior to the start of testing to eliminate the potential for errors while
reading the data. The compliance method determines an analytical relationship
190 Chap. 4 Fatigue Crack Growth

I~ anotch
~ ao
Precracking Length>0.05W --.1
Figure 4.10 EDM notch and minimum precracking (ASTM-E647)

between the inverse of the load-displacement slope, v /p, that has been normal-
ized to elastic modulus, E, and specimen thickness, t(Evt/p) and the normalized
quantity of crack length over the specimen width, a/W, and is one nonvisual tech-
nique that can be used to monitor the crack advancement as it is subjected to
cyclic loading. In the compliance method [9] a clip gage is located at the mouth of
the specimen and the displacement of the gage together with the corresponding
load are measured (see Fig. 4.11). Depending on the location of data measure-
ments on the specimen mouth a theoretical relationship between the two dimen-
sionless quantities Evt/p and a/W can be developed. For example, when the clip
gage is situated at the edge of a C(T) specimen (as shown in Fig. 4.12) the poly-
nomial equation describing a/Was a function of Evt/p can be written [9, 10] as:

a/W = 1.001 - 4.6695 (x) + 18.4(x)2 - 236.82 (x) 3 + 1214.9(x)4 - 2143.6(x)5

Ap

DiSPlacemL

Figure 4.11 The compliance method of measuring load-displacement at the mouth

of the loaded crack
 4.2 Crack Growth Rate Empirical Descriptions 191

Specimen

T = Tension
C = Compression

~ainGage

~-- Specimen Edge

Figure 4.12 Clip gage set up for measuring crack length in terms of displacement
measurement at the mouth of the specimen

where

x = [(Evt/p) 1/2 + 1]-1

When using the compliance method, it is good practice to also check the crack
measurements with an optical method at several intervals during the test. Multiple
reading methods are advisable because it is possible that the clip gage could
become loose at the mouth of the specimen due to specimen and machine vibra-
tion which can introduce errors while data are collected.
The crack length readings recorded in Fig. 4.2 of Section 4.2 must be arranged
in such a way that the da/dN data can be evenly distributed with respect to stress
intensity factor range, !lK. In region I, where the incremental crack growth, !la, is
small, the associated number of cycles, !IN, is large. The opposite is true in the
region III, where the crack growth accelerates and the elapsed cycles, !IN, are
small. For the case of the C(T) specimen type, the even distribution of da/ dN with
respect to !lK can be accomplished by measuring the incremental crack growth
length, !la, to be at most equal to or smaller than 0.04W when 0.25 :;S a/W:;S 0.4
(ASTM-EI47). When 0.40 :;S a/W:;S 0.6, the!la measurement should decrease to
0.02W. In the region III with 0.60 :;S a/W, the crack length measurement should
be more frequent (!la :;S 0.01 W). In the case of the M(T) test specimen, with
192 Chap. 4 Fatigue Crack Growth

Lla :::; O.03W and 2a/W < 0.6, the reading should be Lla :::; 0.02W and
2a/W> 0.6, respectively. Generally, the measured value of Lla = 0.01 in. is an
acceptable value for both specimen types mentioned above. As an example, for a
compact tension specimen having W = 2.0 in. and total crack length a = 0.45 in.
(see Fig. 4.8), the incremental crack growth length measurement, Lla, visually
should be equal to or less than 0.08 in. where a/W = 0.225. As the crack length
increases and becomes equal to a = 0.85 in. (where a/W = 0.425), the crack
length measurement interval will be every 0.04 in.
The results of crack growth tests are mostly thickness independent; however,
materials may show thickness dependency in region III of the da/dN versus LlK
curve when crack growth accelerates and Kmax approaches the material fracture
toughness. In selecting specimen size, adequate thickness is needed to avoid buck-
ling. The presence of buckling can introduce error in the test data measurements.
For the C(T) specimen type, the recommended thickness, B, should be larger than
W/20 and should not exceed W/4, as shown in Fig. 4.8.
If an M(T) specimen is used to generate fatigue crack growth data (this type of
specimen is used when data for the case of R < 0 are of interest), the recom-
mended thickness is W/8. For a thin M(T) specimen, lateral deflection may occur
due to improper loading. Strain gages should be mounted on the specimen for
detection of any bending-induced strain. Bending strain as high as 5% of normal
strain may be acceptable in the specimen (ASTM E-647).
In both types of test specimen, the ratio of original crack length, a, to speci-
men width, W, must be such that net section yielding does not occur at all values
of loading (LEFM limitation). That is, the specimen must be predominantly elas-
tic, except in the localized region at the crack tip. To avoid net section yielding,
an empirical relationship based on test results relating the specimen dimensions
(width and crack length) to the material yielding, (J"Yield' and the calculated maxi-
mum stress intensity factor, K max ' is established [10]. For the C(T) specimen, net
section yielding can be avoided when (W - a) ~ 4/1T (Kmax / (J"Yield) 2 and for the
M(T) specimen (W - 2a) ~ 1.25 (P maxlt (J"Yield)' where Pmax is the maximum
load in the cycle and t is the specimen thickness. The maximum stress intensity
factor, K max ' can be calculated for the C(T) specimen type by the following rela-
tionship [11]:

[ Pmax (2 + a/W) ]
K = f3 (4.6)
max tVw (1 - a/W)

where f3 = [0.886 + 4.64(a/W) - 13.32(a/W)2 + 14.72(a/W)3 - 5.6(a/W)4]

and the stress intensity factor range, LlK is given by:
 4.2 Crack Growth Rate Empirical Descriptions 193

!::.K = r(P max - Pmin) (2 + a/W)] f3 (4.6a)

l tVw (1 - a/W)

Accordingly for the M(T) specimen [12]:

!::.K = P max - P mm
tW
. l 7Tal 1/2
(7Ta) sec-
W
(4.6b)

The measured crack length to be incorporated into Eqs. (4.6a) and (4.6b) must be
an average of the measured value from both sides of the specimen to ensure load
symmetry and that the material in consideration is isotropic.
In using the M(T) specimen type, the crack measurement must be the average
value of two crack tips. Including back surfaces, a total of four measurements are
required [(for the C(T) specimen, only two measurements are needed]. If crack
growth directions at the crack tip are not perpendicular to the applied load to
within ±20°, the test must be discontinued and the data obtained are invalid.
Residual stresses can have significant influence in fatigue crack growth data,
specifically in the region where the !::.Kth value is of interest. The residual stress is
added to the applied stress, which can either lower or raise the calculated value of
the crack tip stress intensity factor. For example, when the magnitude of the com-
pressive residual stress is above or equal to the applied stress, the crack growth
rate data will be close to the threshold value since the crack tip stress intensity is
not effective in causing any crack growth (!::.a = 0). This phenomenon can occur
when the test specimen is machined from the weld region where post stress relief
is impossible. Other parameters influencing the crack growth rate are: (1) temper-
ature, (2) severity of corrosive environment, (3) influence of specimen thickness,
(4) fatigue crack size (it is known that short cracks exhibit greater growth rate than
long fatigue cracks; see Section 4.3 for further discussion related to this topic).
In summary, fatigue crack growth testing is performed to provide the analyst
with the following information:

• Fatigue crack growth data can be used to determine the number of cycles to failure for
an initial crack length in a given material subjected to a given cyclic load environment.
• For two or more materials under the same cyclic loading conditions, it can help to
establish the selection of the material that provides the longest life.
• The plot of fracture stress versus the number of cycles to failure can be used to estab-
lish the inspection interval, an important tool for quality assurance purposes.
• The effects of heat treatment, fabrication (material anisotropic) and environment on
fatigue crack growth can be determined by having laboratory crack growth data gen-
erated for different grain orientations.
194 Chap. 4 Fatigue Crack Growth

4.3 Stress Ratio and Crack Closure Effect

Data from experimental testing conducted in the laboratory to generate the crack
growth curve under constant amplitude loading clearly indicate that the stress
ratio, R, and crack closure behavior have a significant effect on crack growth rate
[1]. Figures 4.13 and 4.14 show the influence of the stress ratio on the crack

(MPaoSQRT(nvn))
100 1000
10-2
DllaID M ..... 1l1li
a IaABICIAII0181 0.100 CI.2!O
0 IaABICIAIIOI82 0.5110 CI.2!O
6 IaABICIAIIOI83 0.750 G.21!O
R-o.l
R-o.5

lr ,.",..
CaodIan:
.'AL
1'1
EmIinIow..lt LA
1pec:iMn: M(T)
on..tIan: lINK
fNquonoy: 0.00 R.O.75

11r

lcr .!
1
CUrft~

.
8_1.0 • 0.3
2
K. 44.331
1It1lk. 0.25
C IE.. lcr1
"P 2.3
0.5
q 0.5
VWcI 41
K,. 21
1
""a.. Q.75
2.5
104

~
RIO 0.7

10 100
4K (kaioSQRT(ln))

Figure 4.13 Influence of the stress ratio, R, for 6061-T6 aluminum alloy [1]
 4.3 Stress Ratio and Crack Closure Effect 195

[MPa*SQRT(mm»)
100 1000
lcr2
a... a JIoV.... 'II1II
a M7HBllWB01Cl o.oao CI.2IIO
0 o.oao CI.2IIO
•
M7HBllWB01El
M7HBllWB01Fl ·1.oaa CI.2IIO
"
0
M7HB11WB01F2
M7HBllWB01F3
o.oaa
0.500
CI.2IIO
CI.2IIO
104 AIcIv: 7II7IAL
CancIIIon: Till
EnviIanment: U%NACL
Specjmlll: M(T)
Ollllltallan: L·T
FNqwncy: 0.00

10-4

'i'
U
.f= lcr11
z
~
" eun. ..........
S_"'a. 0.3
II 1.11

.1l1li.
K. 44.272
0.25
C 3.3I1E-7 10"'
n 2.135
p 0.5
q
YIeld 711
K,. za
" 1 10"
a,. 1
AKa z
A.. 0.7
1~1-~~--~~~~~~__~__~~~~~
10 100
AI< [ksj·SQRT(ln))

Figure 4.14 Influence of the stress ratio, R, for 7075· T65 aluminum alloy [1]

growth curve for 606l-T6 and 7075-T65l aluminum alloys [1], respectively.
From Figs. 4.13 and 4.14, it can be deduced that, for a given crack growth rate,
da/ dN, the stress intensity range, l:!.K, is generally higher as R becomes smaller
and approaches a negative value. In addition to the influence of the stress ratio on
fatigue crack growth rate, the crack closure behavior has significant influence on
the behavior of growing fatigue crack and the rate at which it is advancing. The
196 Chap. 4 Fatigue Crack Growth

stress ratio and closure correction to the crack growth rate equation (da/ dN versus
11K) is usually done in one of the two ways: (1) By replacing the stress intensity
factor range, 11K, with its corrected value (called effective stress intensity range,
I1Keff ), and then the amount of crack growth is calculated simply by the Paris
equation or (2) using the apparent stress intensity factor range, 11K, but incorpo-
rating other effects separately, as shown by Eq. (4.5), where the stress ratio and
closure effect is incorporated by the quantities (1 - RY and (1 - !Y respectively.
In Section 4.3.1, the Elber [13-15] and Newman [8] approaches to the crack
closure phenomenon will be discussed. It should be mentioned that crack closure
can be caused also by other mechanisms, such as oxide, roughness, or corrosion
that are not discussed in this book. In Section 4.4, the retardation effect (crack
growth delay due to variable amplitude loading) is described by the two mathe-
matical models, called the Wheeler and Wellenborg models.

4.3.1 Elber Crack Closure Phenomenon

Wolf Elber [13, 15] was the first to explaine the crack closure effect of constant
amplitude loading for various positive stress ratios, R, on crack growth rate behav-
ior by using the concept of plastic yielding at the crack tip. Prior to Elber's observa-
tion on the crack closure behavior and its effect on crack growth, it was believed by
the experts in the field [16] that the two crack surfaces contact each other and close
the crack at zero or under compressive stresses. This assumption is valid for an ide-
al crack where the plastic deformation is limited to a localized region at the crack
tip (an ideal crack was defined as a saw-cut crack of zero width). Elber argued that,
upon application of constant amplitude cyclic loading during the increasing tensile
portion of the load, a plastic zone is formed. Fatigue crack growth tests on alu-
minum alloys generated in the laboratory by Elber have indicated that for constant
amplitude loading, the formation of plasticity occurring at the crack tip remains be-
hind as the crack grows in a stable manner. As illustrated in Fig. 4.15, the tensile
deformation will create compressive stresses which, when the specimen is un-
loaded, are distributed at the locality of the crack tip, together with the plastically
deformed material left behind, cause the two crack surfaces to contact prior to com-
plete unloading. The compressive residual stresses are a result of permanent defor-
mation that stretches the material, and upon load removal it has to fit the elastic ma-
terial surrounding it. The misfit causes the elastic material to induce compressive
residual stress in the plastic region local to the crack tip.
In Fig. 4.15, when the crack length was at location (1), it was surrounded by a
plastic zone larger than any other previous position, while its corresponding crack
length was smaller. At location (2) after N number of cycles, a new plastic zone is
formed that now is larger than location (l) because of its length and higher value
associated with crack tip stress intensity factor. The crack tip plastic zone at all
stages of advancing fatigue crack forms an envelope that contains all the previous
 4.3 Stress Ratio and Crack Closure Effect 197

"
Plastic Zone

K ~Seff = Smax- Sci Plastically deformed

matenalleft behind
(Compressive
Stresses)
~Keff = Kmax-
,-
KcI

V-!Wit
The presence of plastic zone
causes the crack to close at a higher
load Sci >Smin upon unloading

Smin Sci
Loading and Unloading
Stage of a Cycle
Sma x
s~~ S~n •
Cycle

Figure 4.15 Illustration of crack closure based on plastic zone and effective stress
value

An Ideal Crack
(Saw-Cut Crack With Zero Width)

Plastic Zone -...

A fatigue crack with plastic zone

enveloped at the two crack surfaces
in the wake of advancing crack

Figure 4.16 Illustration of plastic deformation formed at crack surfaces in the wake
of an advancing crack (compared with an ideal crack)

plastic zone sizes and its presence is the cause of crack closure behavior. Figure
4.16 illustrates the crack tip plastic zone and the envelope of all plastic zones for
an advancing crack subjected to cyclic load compared with an ideal crack with
zero width.
The plastic deformation helps to partially close the crack surfaces such that the
crack will close and open at a stress level higher than the smin. This effect results
198 Chap. 4 Fatigue Crack Growth

in having a smaller stress intensity factor range, dK, value and, consequently, a
smaller crack growth rate, as illustrated in Fig. 4.15. The effective stress intensity
factor range, in terms of the effective stress range (.ls)eff' is written as [15]:

.lK = f3 .ls(1Ta)1/2
.lKeff = f3(.lS)eff (1Ta)1/2 (4.7)

where (.lS)eff = smax - Sci· The quantity Sci is the stress level at which the closure
occurs (smin < Sci) and KcI is the corresponding stress intensity factor (Kmin < KcI ).
In addition to the quantity Sci (the stress level at which the two crack surfaces are
in contact upon unloading), equal attention must be given to another parameter,
called the crack opening stress, sop' (the stress at which the two crack surfaces are
not in contact with each other upon reloading). Hence, the fatigue crack will
advance only during the opening portion of the cycle. In this book, it is assumed
that the two quantities Sci and sop are the same (in general, for a given material, the
closure stress, Sci' may have different value than sop).
The effective stress range ratio, U, that corrects the apparent stress intensity
factor range to account for closure behavior, can be written [15] as:

Elber's crack growth rate equation, in terms of the apparent stress intensity
range, .lK, and quantity U, can be expressed in terms of the Paris law as:

da/dN = C(U Mr (4.9)

Based on constant amplitude testing of 2024-T3 aluminum alloy sheet, a linear

equation for U in terms of R was established [15] that can be written:

U = 0.5 + O.4R -0.1 <R < 0.7 (4. lOa)

This relationship was later modified by Schijve [17] as follows:

U = 0.55 + 0.35R + 0.IR2 (4.10b)

The quantity U described by Eqs. (4.l0a) and (4.10b) provides the correction for
the closure effect (described in terms of stress ratio, R) on the crack growth rela-
tionship. From Eqs. (4.9) and (4.10) it can be seen that the two quantities, effec-
tive and apparent stress intensity factor range, become equal when the stress ratio
R increases. That is, the crack closure effect becomes less effective when the R
value increases.
 4.3 Stress Ratio and Crack Closure Effect 199

4.3.1.1 Threshold Stress Intensity Factor, !::.Kth

In the limit when Kmax reaches the KcI value, the effective stress intensity factor
!::.Keff= 0 and, therefore, no crack growth is expected. This does not necessarily
imply that the stress intensity factor range, !::.K, is zero; it simply states that there
is a minimum value of !::.K, called the threshold stress intensity factor (!::.Kth ), in
which the growing crack ceases to advance, see region I of crack growth curve
where !::.K is a constant (asymptotic value of !::.K in region I) and da/ dN
approaches zero. As was mentioned previously (in Section 4.2), zero growth is
approximated by a cutoff growth rate of 4 X 10- 10 m/cycle assigned by ASTM-
E647 that provides a practical !::.Kth value near the threshold region. The!::.K level
at which the growth is zero (or when da/ dN = 4 X 10 -)0 m/cycle) is a function of
the stress ratio, R.
The threshold stress intensity range, !::'Kth' in terms of stress ratio, R, and the
threshold stress intensity factor at R = 0, !::.Ko' in a simplified form was empiri-
cally expressed as [I8]:

(4.11)

Co and d are the fitting constants. Note that, in Eq. (4.11) the quantity !::.Ko must
be available through laboratory testing. This quantity is independent of crack
length. By allowing the constant Co = 1 the Klesnil and Lukas [19] equation is
obtainable.

(4.12)

When the fitting constant d = 1, the Barsom [20] relationship is obtained. In cases
where the two fitting constants are not available, the value of Co = d = 1 will
give conservative results for the threshold stress intensity factor, !::.Kth for R :? O.

(for Co = d = 1) (4.13)

In general, the fatigue threshold value, !::.Kth , decreases with increasing stress ratio,
R and becomes a constant at R = 0 (Figure 4.13 illustrates the influence of R on
!::.Kth value for the 6061-T6 aluminum alloy). Equation (4.11) can be modified to
account for the dependency of threshold stress intensity factor on crack size [21].
Frost [22, 23] and Usami [24,25] studied the effect of crack length on the thresh-
old stress intensity factor and concluded that !::.Kth decreases with decreasing crack
length. On the other hand, when crack length has sufficient dimension that its
behavior can be described by the linear elastic fracture mechanics approach, the
threshold stress intensity factor, !::'Kth' is independent of crack size but dependent
on R. Generally, cracks in material can be classified as being long, short, or small
in size. Long or large cracks are usually considered as being through the thickness
200 Chap. 4 Fatigue Crack Growth

cracks in a thick specimen (such as a crack in a notched centered crack specimen

that is used to generate fracture properties of metals). The crack growth behavior
of large cracks is well established in terms of the closure phenomenon through
the effective stress intensity factor, b.Keff [see Eq. (4.9)]. In the threshold regime
(region I), where crack growth is slow, the test measurements for long cracks can
be sensitive to overloading which introduces the retardation effect, and inconsis-
tency in data recording can occur.
Short cracks can be defined as being physically short in dimensions, such as
surface cracks with short crack length smaller than 0.02 and 0.078 in. [26] in
length. Cracks with short dimensions are less affected by closure phenomenon
and will grow faster than corresponding long cracks subjected to the same stress
intensity factor, b.K. In general, crack closure effects increase with crack length.
In addition, short cracks show a lower threshold value when compared with long
cracks, as shown in Fig. 4.17. The transition from short to long crack behavior is
reported by different sources [26-28] to be between 0.5 and 2 mm.
Small cracks (the authors noticed that the term "small" is sometimes used
interchangeably with "short" as was evident in Reference [27]) are defined as
having dimensions as small as their material characteristic microstructure dimen-
sions, such as the grain size. Their growth is controlled by grain boundaries and
cannot be described by the linear elastic fracture mechanics approach. There are
several classes of small cracks that Ritchie and Lankford defined in their work
"Small Crack Problems" [26]. They used microstructurally, mechanically, and
physically small as useful qualifiers to classify small fatigue cracks. For example,
a crack is mechanically small when the crack length, a, is smaller than the plastic
zone size, ry (a < ry)' When the crack length is smaller than the grain size, dg , it
is called microstructurally small (a < dg ). Other classes that fatigue cracks can be
defined as being small include physically small when a < 1 mm and chemically
small when a < 10 mm.

Z Behavior
'0
......
III
'0
'-'
Cl
.9

Log (to.K)

Figure 4.17 Illustration of da/ dN Versus M curve for short and long cracks
 4.3 Stress Ratio and Crack Closure Effect 201

Tanaka et al. [21] had established a theoretical model to describe the small
crack length effect on the threshold stress intensity factor, I1Kth , based on the
crack-tip slip band blocked by the grain boundary concept (called the BSB
model). The BSB model is based on the assumption that a small crack ceases to
advance (the threshold condition) when the crack tip slip band has been arrested
by the grain boundary and prevented from growing into the next grain. The effects
of crack length and grain size on the threshold stress intensity factor, I1Kth were
studied both theoretically and experimentally and a general relationship between
I1Kth ; crack length, a; and the intrinsic crack length, ao was established as:

(4. 14a)

Equation (4.14a) was later modified to account for the stress ratio, R [1]:

where (I1Kth )R =0 is the threshold stress intensity factor for R = o. The intrinsic
crack length, ao' defines the boundary between small and long cracks and repre-
sents the minimum initial crack size for conducting a meaningful life evaluation
using fracture mechanics analysis. Based on the theoretical results and experi-
mental data, an estimated value of 0.004 in. is assigned to a variety of steels and
aluminums [29]. As mentioned in Section 4.1, the minimum initial crack size used
in the damage tolerance crack growth analysis must be larger than the material's
grain size in order not to violate the isotropic continuum concept.
The minimum accepted crack size that can be used for conducting a meaning-
ful crack growth analysis using LEFM can be estimated by employing the Kita-
gawa and Takahashi diagram [30]. Kitagawa argued that in many structural parts
the crack growth process occurs by the initiation and growth of cracks as small as
0.5 to 1.0 mm in length, below which their growth behavior deviates from the
LEFM regime. Therefore, it is necessary to develop a method to observe the ini-
tiation and growth of such small cracks that may behave differently from the
experimentally measured crack growth data associated with large cracks employ-
ing the ASTM standards. Kitagawa defined the threshold stress range, I1s th , for a
fatigue crack to be associated with a crack growth rate less than 2 X 10- 9
mm/cycle (=8 X 10- 11 in./cycle). To study the effect of crack length on the
threshold stress intensity factor range, I1Kth' the experimental variation of I1s th
versus full crack length, 2a, was plotted (using logarithmic scale) for crack
lengths smaller and larger than 0.5 mm. The measured experimental data showed
that for crack length larger than 0.5 mm, a good correlation with the LEFM results
was obtained (a straight line relationship with slope -1/2; see Fig. 4.18). This
observation (constant slope of -1/2) indicates that the threshold stress intensity
factor, I1Kth' is independent of crack length and is a constant for long cracks
where 2a > 0.5 mm. For crack length smaller than 0.5 mm a deviation from the
202 Chap. 4 Fatigue Crack Growth

straight line (anomalous behavior associated with small cracks) was observed
indicating that the LEFM concept may not be applicable to define the fatigue
crack growth threshold behavior. The anomalous behavior can be associated with
large scale plasticity at the tip of small cracks (2c < 0.5 mm.) where the applied
stress amplitude is high and close to the material yield stress.
In the Kitagawa diagram, the boundary between small and large cracks was
estimated to be at the onset of deviation from the straight line behavior. The
experimental data on HT-80 steel, shown in Reference [30], indicates that the con-
tinuation of the trend departing from the straight line becomes gradually asymp-
totic to the fatigue limit of an uncracked smooth specimen, as shown in Fig. 4.18.
Cracks smaller than a* are nondamaging cracks when subjected to an applied
cyclic load equal to the endurance limit. Note that the Kitagawa diagram is depen-
dent on the stress ratio, R, and will shift to a lower stress level as R increases (also
note that the endurance limit for most material decreases with increasing R value,
as discussed in Chapter 2.)
A minimum initial flaw size employed in the aerospace industry for evaluating
the life of fracture critical components has been developed based on the capabil-
ity of crack detection using standard NDI methods. The reader should realize that
the crack size assumption by the standard NDI methods is based on the concept of
a maximum initial flaw size that may escape detection. If a smaller flaw size is
required by the analyst, a special level NDI could be performed on the part. These
values are much higher than the limiting value of ao = 0.5 mm (0.019 in.)
observed in the Kitagawa diagram.
Recently, the NASA has accepted the option of replacing the conventional
fatigue approach (S-N curve) in estimating the life expectancy of non-fracture
critical or low risk parts with fracture mechanics approach that uses a part through
crack of depth a = 0.005 in. and length 2c = 0.01 in .. This value is much smaller
than the initial flaw size provided by standard NDI methods when performing
safe-life analysis for primary space structural components. This method is pre-
ferred over the conventional fatigue approach (see chapter 2) because of its feasi-
bility and availability of matereial data as shown in appendix A. The flaw size
assumption provided by the NASA (0.005 in.) is larger than the material grain

Endurence Umit for Uncracked

.J
Smooth Specimen (R = 0)

~ .r: Non-Damag~ R~on _ _

....... • AS=AKTh I p(lta)li2

~ Small
-I Crack
30< a
+--(LEFM)

a*

Figure 4.18 Kitagawa diagram for estimating a o

 4.3 Stress Ratio and Crack Closure Effect 203

size and it falls in the boundary between small and large cracks and the concept of
isotropic continum is obeyed.

4.3.2 Newman Crack Closure Approach

The crack closure behavior and its effect on fatigue crack growth rate under
constant amplitude loading were discussed by Newman [1, 8, 31]. He devel-
oped an analytical model to describe the crack closure phenomenon for both
plane stress and plane strain conditions (the thickness influence on the crack
closure behavior). The closure model was based on a strip yield model
(Dugdale model) that was modified by Newman to incorporate the envelope of
all the plastic zones in the wake of an advancing crack. In both plane stress and
plane strain conditions, the closure model considers that the material is plasti-
cally deformed along the crack surfaces as the crack grows. Newman argued
that fatigue cracks remain close during part of the cycle when unloading occurs.
When the two crack surfaces are in contact with each other, during the final
part of the unloading portion of the cycle, the material may undergo compres-
sive yielding. Upon reversing the cycle as the specimen is reloaded, the crack
produces tensile residual stresses at the crack tip that cause the crack to open
(zero contact between the two crack surfaces) lower than or equal to the closure
stress occurring in the previous portion of the cycle, Sci' The stress at which no
surface contact is present upon reloading the specimen is called the crack open-
ing stress, sop' The crack advances during each cycle when the two surfaces are
not in contact with each other (where sop> smin) and the calculated stress inten-
sity factor range is smaller than the apparent 11K value when calculating the
crack growth rate. An equation was established by Newman that could define
the crack opening stress, sop' as a function of stress ratio, R; maximum stress
amplitude, Smax; and specimen thickness. The specimen thickness effects in clo-
sure behavior were defined in terms of a two-dimensional closure model [31]
by applying a constraint factor, a, on yielding.
The quantity a is used to describe material yielding in two and three dimen-
sions through the flow stress, (To (tensile yielding when (To is positive and com-
pressive yielding when (To is negative). The flow stress, (To' is taken as the average
value of the tensile yield and the ultimate of the material. Crack tip yielding
occurs depending on the biaxial or triaxial state of stress on that region. Under
plane strain conditions, yielding occurs at a stress level much higher than the uni-
axial yield stress obtained through laboratory testing. The effective yield stress
value, a(To' could be as high as three times the uniaxial stress when a three-dimen-
tional state of stress (plane stress) prevails. Therefore, the material will yield when
the stress at the crack tip has reached the quantity a(To' For plane stress and plane
strain conditions, the constraint factor, a, takes the values of a = 1 and 3, respec-
tively. High a values (2.5 or higher) have been assigned to materials with low
KI c / (TYield ratio (low fracture toughness value such as high strength steels),
whereas high Klc/ (TYield ratios have 1.5 ,,;; a";; 2.0.
204 Chap. 4 Fatigue Crack Growth

The closure effect on the crack growth rate was described by correlating the
effective stress intensity range, !:J.Keff , to the apparent stress intensity range, !:J.K,
through the crack opening stress, sOP' and stress ratio, R, as:

(4.15)

Equation (4.15) simply states that, given a crack growth rate, (da/ dN)} , obtained
from test data under constant amplitude loading and a stress ratio, R} (for sim-
plicity assume R} = 0.2) with the corresponding stress intensity factor range !:J.K;
for any other stress ratio, R2, where (da/ dN)} = (da/ dN)2' the shift in the stress
intensity factor range (!:J.Keff) can be evaluated through the parameter U (for stress
ratio and crack closure adjustment) described by Eq. (4.15); see Fig. 4.19. Exper-
imental data show that for most aluminum alloys, the stress ratio contribution that
can shift the crack growth rate curves is negligible for a stress ratio, R, greater
than 0.7 [32], see Fig. 4.20. Figure 4.20 shows the experimental data obtained for
2014-T651 aluminum alloy with different stress ratios, R = 0.1,0.4, and 0.7 that
were curve fit by employing Eq. (4.5) [1]. Equation (4.5) can be used to describe
the crack closure and stress ratio effect due to varying the stress ratio, R, and it
provides excellent curve fit.
For a center crack specimen under tension loading, the normalized crack open-
ing stress, sop/ smax' describing !:J.Keff [which was expressed by Eq. (4.15)] in terms
of stress ratio, R, and constraint factor, ex, can be written as [8]:

Curve Fit with

Crack Closure • Experimental data

1 E-3

1 E-4
,..."

Z
"C
.......
nI
"C
....... 1 E-S
C)
0
..J

1E-6

1 E-7 (.1K) 2 +1 1 +- (.1K)2

10 "'>K) 1 lOa
Log (flK)

Figure 4.19 Illustration of curve fit with crack closure to crack growth rate data
 4.3 Stress Ratio and Crack Closure Effect 205

[MPa*SQRT(mm))
100 1000

10.5

CUrveP..........
S,...'''o 0.3
Cl 1.5
K. 29.067
10"
V at thk = 0.5
C 6.46E-9
n 3.918
p 0.5
V q 1
10.7 Yield 24
~. 16
A,. 1

~ B.
.11<0 9.5

R" 0.7
10-8
10 100
~K [ksi*SQRT(in)]
Basic curve fit to data

Figure 4.20 Crack growth curves for 2014· T651 aluminum alloy for stress ratios
R = 0.1,0.4, and 0.7 [1]
206 Chap. 4 Fatigue Crack Growth

for R ;:,: I (4.16)

and

for -1 < R:;::; 0 (4.17)

The coefficients Ao, Al ' A2 , and A3 for sop;:': smin are:

Ao = (0.825 - 0.34a + 0.05( 2)[cos(7TSmax/2<To)]I/n (4.18)

AI = (0.415 - 0.7la) X (smax/2<TJ (4.19)

A2 = 1 - Ao - Al - A3 (4.20)

A3 = 2Ao + Al - 1 (4.21)

The flow stress, <To' is taken to be the average of the uniaxial yield stress and uni-
axial ultimate tensile strength of the material:

(4.22)

The contribution of the crack closure effect on the FNK empirical relationship
described by Eq. (4.5) is described by the crack opening function,! For plasticity
induced crack closure, the parameter,!, can be written as [8]:

for R ~ 0
(4.23)

and,

for -2:;::; R < 0 (4.24)

where Kop is the crack opening stress intensity factor below which the crack is
closed. For a stress ratio of R = 0, the crack closure parameter, f, will have the
maximum value given by Eq. (4.23).
Newman and Raju [31] studied the crack closure effect for surface and comer
cracks under constant amplitude loading and variable stress ratio, R. They have
shown that better crack growth results at the c-tip can be obtained by multiply-
ing IJ..K by a crack closure factor (f3R)' The value of f3R for the stress ratio R > 0
is given by:

f3R = 0.9 + 0.2R2 - 0.lR4 (4.25)

and for R :;::; 0 the factor f3R = 0.9.

 4.3 Stress Ratio and Crack Closure Effect 207

Table 4.1 Fracture and fatigue crack growth properties for 7075-T651 aluminum alloy

(TUI K Ic KIe AK BK C n p q t::.Ko Rei Q' SR

85, 28, 38, 1.0, 1.0, 0.2E-7, 2.885, 0.5, 1.0, 2.0, 0.7, 1.9, OJ

The FNK empirical equation describing the crack growth rate [see Eq. (4.5)] is
used in the NASA/FLAGRO computer code that was developed to provide an
automated procedure for analysis of fracture critical parts of NASA space flight
hardware and launch support facilities. In this computer program, the effect of the
r
stress ratio is incorporated through the quantity (1 - R and crack closure behav-
ior through the function f by assuming a constant value of smax / (To = 0.3. This
value was selected because it is close to the average value obtained through
fatigue crack growth tests using various specimen types. For most aluminum
alloys (2000 through 6000 series), ex is chosen to have a value of 1.5 and for 7000
series a = 1.9. Having the two quantities sma,/ (To and a available, the crack clo-
sure effect for any other stress ratio, R, can be evaluated by utilizing Eqs. (4.23)
and (4.24). Table 4.1 presents a listing of all the constants that must be provided
as input to Eq. (4.5) for 7075-T651 aluminum alloy in order to conduct a compre-
hensive safe-life analysis. Information contained in Table 4.1 was extracted from
the NASA/FLAGRO 2.0 material library. The quantities Ak and Bk are the fit
parameters in defining the plane stress fracture toughness, Kc, in terms of plane
strain fracture toughness, K Ic [1]. The NASA/FLAGRO computer code has the
following capabilities:

1. Safe-life analysis of the structural components subjected to a constant amplitude

cyclic load
2. Critical crack size that causes the structure to become unstable
3. Stress intensity factor solutions to different crack geometries and loading conditions
4. Growth rate constants described by Paris law and Eq. (4.5)
5. da/ dt life analysis

NASA/FLAGRO 2.0 has a rich material library that has fracture properties
data for many aerospace alloys subjected to varying heat treatment conditions.
Appendix A provides fracture properties of most aerospace alloys obtained from
the NASA/FLAGRO material library (see Table 4.1 for the definition of the sym-
bols). In addition, the program has enhanced stress intensity solutions that cover
various crack geometries under different loading conditions. For a more detailed
description of the NASA/FLAGRO computer code, the reader may refer to Ref-
erence [1]. The improvement for this code is in progress and as the technology
advances the program will enhance accordingly. Some areas of improvements that
require future attention are:
208 Chap. 4 Fatigue Crack Growth

• Incorporating the da/ dt time-dependent environmental crack growth

• Retardation effect
• Cycle by cycle approach for calculating the crack growth, da/ dN

Example 4.2
A pressurized cylindrical tank is made of Ti-4AI-4V alloy. Its fracture properties
(based on the NASAjFLAGRO material library) are shown in Fig. 4.21. The tank
is pressurized and depressurized eight times per year during its service life. It is
required that the tank undergo a proof test of 1.5 X Maximum Operating Pressure
(MOP) prior to its usage. For operating pressure, Pop = 350 psi; (1) Use a cycle by
cycle analysis by using Eq. (4.5) and compare the results of the first three cycles
with the NASA/FLAGRO 2.0 computer code output (or any other available code);
(2) Determine the final crack length at the end of the year. The original crack
length based on penetrant inspection is given as a circular surface crack with
ao = 0.075 and 2c = 0.15 in. The purpose of part (1) of this example problem is
to allow the reader to learn the steps necessary for computing the crack growth
values and accumulation of growth corresponding to each cycle.

Solution
Using Eq. (4.5), the amount of crack growth associated with each cycle can be
computed. The stress intensity range, b..K, for each cycle is calculated by employ-
ing Eq. (3.52) of Chapter 3 for a longitudinal surface crack in pressurized pipe:

where:

D = 40 inch R=O
a = 0.075 inch
c = 0.075 inch
p = 350 psi p~~
0' =pR/t Cycles

Material Properties from NASAIFLAGRO

(TI-6AI-4V)

0Yield ault Klc Kle AK BK C n p q 6KO Rei a SR

140, 150, 55, 45, 0.5,1.0, 0.2E-S, 3.0, 0.25, 0.75, 3.5, 0.5, 2.5, 0.3

Figure 4.21 Material and crack geometries for the pressurized cylinder in prob-
lem 4.2
 4.3 Stress Ratio and Crack Closure Effect 209

MI = 1.13 - 0.09 (a/c)

M2 = -0.54 + 0.89/(0.2 + a/c)

M3 = 0.5 - [1/(0.65 + a/c)] + 14 (1 - a/e)24

g =1+ [0.1 + 0.35 (a/t)2](1 - sin 0)2

I", = [(a/c)2cos 20 + sin 2 0]1/4
Ix = [1 + 1.464(a/ctL65rl/2

The quantity fc = [(1 + k 2 )/(1 - k 2) + 1 - 0.5 (a/t) 1/2] [t/(D/2 - t)] where
k = 1 - 2t/D and the value off = 1 for an internal crack. Note that the quantity
Ix = (cP) -1/2, as discussed in Eq. (3.47) of Chapter 3.
The calculated values of M I , M2 , M3,fr' andf. for aft = 0.75 and a/e = 1
using Eq. (3.52) are:

MI = 1.04, M2 = 0.201, M3 = -0.106./r = 0.637 and/e = 1.005

The correction factor is f3 = 0.694, where 0 = 1T/2. The stress intensity factor
range, f:.K, for the case of R = 0 (see Fig. 4.21) can be obtained as:

f:.K = Kmax = f3 O'max(1Ta)I/2

= 0.694 X 1.5 X 70(3.14 X 0.075)1/2

= 35.379 ksi (in.I/2)

Note that the applied stress for the first cycle (M = 1) is equal to the proof stress
(1.5 X O'op) and the retardation effect due to proof cycle is not included in this
analysis. When M = 1 andR = 0, Eq. (4.5) can be written as:

e(l - It f:.K n( 1 - -d-

f:.K )P

f:.a = (f:.K)q
1--
K 1e

The crack opening function, I, can be calculated from Eq. (4.23) by replacing the
quantity smax /0'0 by 0.3 if = 0.2745). Moreover, the threshold stress intensity range
Mth = (Mth)R =0 and is designated by Mo in the NASA/FLAGRO material library.
Simplifying the crack growth equation by supplying the appropriate values of
the constants I, M o' Mth' e, n, p, and q (see the material properties shown in Fig.
4.21), the amount of growth for the first cycle is given by f:.a = 0.000106 in. The
original crack depth is advanced by 0.000106 in. and the new crack length at the
210 Chap. 4 Fatigue Crack Growth

a-tip is now acting as the original crack depth for the next cycle (a o = 0.075 +
0.000106 = 0.075106 in.)
The amount of growth in the length direction (c-tip) can be calculated in the
same way as for the a-tip. Note that the equation of the stress intensity factor for
the c-tip was formulated in Chapter 3 [described by Eq. (3.52)] by simply replac-
ing the angle () = 90° with 0°. The correction factor, f3, is slightly higher for
() = 0° due to the dependent quantities g and!", on the angle () [see Eq. (3.52)].
The calculated values of g and!", are 1.296 and 1, respectively. The stress inten-
sity range for the c-tip can be calculated as:

0.9 X 1.5 X 70(3.14 X 0.075)1/2

45.88 ksi (in. 1/2)

Crack growth for the c-tip can be obtained from Eq. (4.5) in the same manner as
was applied for the a-tip direction. However, the fracture toughness value for the
c-tip, K Ie , is now replaced by UKIe [1] (see also Section 3.7.4). The closure
effect for the c-tip was described by Eq. (4.25) and for the case of R = 0 it takes
the value of f3R = 0.9:

C(1 - fr<0.9 D.KY(l - ~:ili)P

D.e = o. K
( 1 _ 0.9 D.K)Q
1.1 K Ie

The calculated amount of growth at the c-tip is given by D.e = 0.000201 in. The
original crack length at the c-tip is advanced by 0.000201 in. and the new crack
length at the c-tip is now acting as the original crack length for the next cycle
(e = 0.075206 in.).
The amount of crack growth in both the a- and c-tip directions can be calcu-
lated for the next cycle in the same manner; however, the aspect ratio a/ e for the
new analysis must include the actual value, rather than the original value of
a/ e = 1. For example, for the second cycle of pressurization and depressurization,
the new aspect ratio a/e = 0.998 = 0.075106/0.075201 must be employed. The
following table shows the calculated values of the quantities that are used to eval-
uate the crack growth up to three pressure cycles for both the a- and c-tips. In
addition, the values of the amount of crack growth for each cycle are calculated by
NASA/FLAGRO computer code and presented for comparison.
 4.3 Stress Ratio and Crack Closure Effect 211

Results of the hand analysis for example 4.2

alc MJ M2 M3 g(90) Ix Iq, Ie {3 flK fla

1.04 0.202 -0.12 1.0 0.637 1.0 1.005 0.695 3.5 1.06E-4
0.998 1.04 0.203 -0.11 1.0 0.637 1.0 1.0054 0.697 3.5 1.705E-5
0.998 1.04 0.203 -0.11 1.0 0.638 1.0 1.0053 0.697 3.5 1.707E-5

alc MJ M2 M3 g(O) Ix Iq, Ie {3 flK flc

1 1.04 0.202 -0.12 1.296 0.637 1.0 1.005 0.901 3.5 2.05E-4
0.998 1.04 0.203 -0.11 1.297 0.637 0.999 1.0054 0.903 3.5 2.88E-5
0.998 1.04 0.203 -0.11 1.297 0.638 0.999 1.0053 0.903 3.5 2.88E-5

Final crack sizes by hand analysis

Cycle 1
Final crack sizes: a = 0.751060E-01, c = 0.752050E-01, alc = 0.9986

Cycle 2
Final crack sizes: a = 0.75123E-01, c = 0.752338E-01, alc = 0.9985

Cycle 3
Final crack sizes: a = 0.75140E-01, c = 0.752626E-0l, alc = 0.9984

Results of NASA/FLAGRO (cycle by cycle)

Cycle 1
Final crack sizes: a = 0.750690E-Ol, c = 0.751050E-Ol, alc = 0.9995

Cycle 2
Final crack sizes: a = 0.750833E-0l, c = 0.751263E-01, alc = 0.9994

Cycle 3
Final crack sizes: a = 0.750876E-0l, c = 0.751476E-01, alc = 0.9993

In using NASAIFLAGRO with three cycles at once rather than cycle by cycle,
the results are:

Final crack sizes: a = 0.750975E-0l, c = 0.751475E-01, alc = 0.999335

Note: The crack growth analysis conducted by the NASAIFLAGRO computer

code is not based on a cycle by cycle approach; rather, it is based on incremental
212 Chap. 4 Fatigue Crack Growth

growth, da = (l/N)a o ' The advantages associated with this method are cost
effectiveness and speed. The NASA/FLAGRO analysis is performed by assuming
the amount of incremental growth always to be 1/200 of the original crack size.
The number of cycles, N, associated with a o /200 is calculated by using Eq. (4.5).
Therefore, the average amount of growth per cycle is:

(da)!eycle = (a o /200)/N
Discrepancies between the hand analysis and NASA/FLAGRO are therefore
expected for the first few cycles until the calculated da value becomes equal to
ao /200.

Example 4.3
In the previous example problem (1) determine the number of cycles needed to
have leak-before-burst, (2) find the stress value for which leak-before-burst is not
achievable (the leak-before-burst criteria were discussed in Chapter 3).

Solution
It would be difficult, if not impossible, to obtain the number of cycles associated
with leak-before-burst by employing a hand analysis as demonstrated in the pre-
vious example. Using the NASA/FLAGRO computer code, however, the results
can easily be provided to the analyst. The following are the results of the
NASA/FLAGRO analysis for part I of this problem.

Results
Transition to a through crack occurs at cycle no. 948.56.
Where the crack length at the a-tip: a = 0.9382E-Ol (t = 0.10 in.) and:
crack size at c-tip: c = 0.105167, alc = 0.892093
FINAL RESULTS:
Unstable crack growth, max stress intensity exceeds critical value:
Kmax = 88.43, Kef = 88.34
at cycle no. 1756.07, where crack size c = 0.405322 in.

The above analysis indicates that the initial crack can grow through the interior
wall of the tank (a = t = 0.1 in.) without failure when it is subjected to fluctuat-
ing load due to pressurization and depressurization cycles. It requires approxi-
mately 948 cycles for the crack to become a through crack. Moreover, it takes an
additional 808 cycles (1756-948 cycles) for the instability to occur.
For the tank to burst before leaking, the level of pressure must be high enough
to have failure before the surface crack becomes a through crack. Two cases are
illustrated in this example problem to demonstrate failure before tank leakage;
 4.3 Stress Ratio and Crack Closure Effect 213

(1) to include the proof test as the first cycle in the fatigue crack growth analysis
and (2) to exclude the proof test cycle. To obtain the maximum stress value for
which leak-before-burst is prevented, several crack growth runs based on different
applied fluctuating stresses were performed. The maximum operating stress that
can be used in order to have a valid proof test (without yielding the material) is:

(O'max)operating = 0.9 X O'Yield/1.5 = 84 ksi

Note that, in conducting the proof test, no permanent deformation is permitted on

the tank. For this reason, it is reasonable to allow the proof test stress to go as high
as 90% of the tensile yield of the material. Based on the above criteria, the maxi-
mum operating pressure will be 84 ksi.

Results
Transition to a through crack at cycle no. 423.38:
Where the crack length at the a-tip: a = 0.913E-0l (t = 0.1000) and:
Crack size: c = 0.101266, alc = 0.902003
FINAL RESULTS:
Unstable crack growth, max stress intensity exceeds critical value:
Kmax = 88.58, Kef = 88.34
at cycle no. 776.28, where crack size c = 0.302632

From the above results, one can conclude that for the tank with thickness t = 0.1
in. subjected to stress level of 84 ksi, the leak-before-burst condition cannot be
avoided. However, by eliminating the proof cycle (not considering its effect in the
crack growth rate analysis) and allowing the maximum operating stress in the
above analysis to go as high as 135 ksi < O'yield = 140 ksi, failure of the tank is
expected to occur before the surface crack becomes a through crack, K Ie ~ Kmax
(see Eq. 3.55 of Chapter 3 for instability condition). As indicated in the results of
the analysis below, it takes approximately 17 cycles for the tank to burst at a stress
level corresponding to 135 ksi.

Results
Unstable crack growth, max stress intensity exceeds critical value:
Kmax = 60.64, Kef = 60.50
at cycle no. 16.80 (the end of the second year)
Crack sizes: a = 0.801573E-OI < t = 0.1 in., c = 0.839072E-OI, alc = 0.955309

Example 4.4
An embedded crack was detected by x-ray inspection in the weld region of a pres-
surized liquid oxygen tank to be used for a space vehicle, as shown in Fig. 4.22.
214 Chap. 4 Fatigue Crack Growth

1E-2
55

1E-3
da/dN -6K curve for the weld

1E-4

- 5E-5
--

L
1E-6

1E-7

10 55 100
log (ilK)

Figure 4.22 Crack growth curve for the welded region described in Example 4.4

The detected crack was shown to be very close to the surface of the weld and con-
servatively considered as a surface crack having length 2c = 0.32 in. The crack is
oriented in the circumferential direction in the shell region of the tank with (TApplied
= PR/2t (where P is the maximum operating pressure, R is the tank radius, and t
is the thickness associated with the weld region in which the crack was detected).
The da/dN - !J..K data curve for this alloy is shown in Fig. 4.22. Use the Forman
equation and compare the results with the NASA/FLAGRO computer code (or
any other available fatigue crack growth computer code) to determine the number
of pressurization cycles that the tank can withstand before failure. Assume the
weld properties are 50% lower than those of the parent material (stress ratio,
R = 0, P = 35 psi, ((TYield )parent = 40 ksi, ((TVlt )parent = 60 ksi, tank radius,
R = 72 in., and t = 0.1 in.).

Solution
To obtain the total number of pressurization cycles in the weld region of the liquid
oxygen tank, a 50% reduction in properties must be considered (as shown in Fig.
4.22) . Furthermore, region III of the curve is approximated by using the asymp-
totic line associated with 27.5 ksi (in. 1/2) (this is equal to 50% of the parent mate-
rial fracture toughness). The Forman constants c and n for the weld material [from
Eq. 4.4)] can be obtained by selecting two points on the da/dN curve, as shown in
Fig. 4.22:
 4.3 Stress Ratio and Crack Closure Effect 215

0.0001 = [c(1O)"]/(27.5 - 10)

5 X 0.00001 = [c(8.5)"]/(27.5 - 8.5)

Solving for c and n:

c = 3.11 X 10- 7 and n = 3.75

The Forman equation for the weld material can be written as:

3.11 X 1O- 7(LlK?·75

da/dN = Ll
27.5 - K

The number of cycles to failure LlN can be evaluated by:

LlN = ff
o
[8.84 X 10 7(LlKr 3.75 ]dc - fr [0.321
0
X 10 7(LlK)-2.75]dc

where cj = 0.16 in. and final crack length, ct ' is computed by:

The number of cycles to failure in terms of final crack length is:

0.379
LlN = J [8.84 X 10 7(LlKr375]dc
0.32

-f 0.379

0.32
[0.321 X 10 7(LlK)-2.75]dc

LlN = f 0.379

0.32
[8.84 X 10 7[25.2V7TCr 3.75 ]dc

-f 0.32
0.379
[0.321 X 10 7 [25.2 V7TCr 2 .75 ] dc

LlN = J
0.379

0.32
r57.4c- 3.75 ]dc - f
0.32
0.379
[93.1c- 2.75 ]dc = 83 Cycles

To use the NASA/FLAGRO computer code for determination of the number of

cycles to failure, the fatigue growth rate constants used to establish the da / dN,
LlK curve must be computed from the data points shown in Fig. 4.22.
216 Chap. 4 Fatigue Crack Growth

Material properties:
UTS: YS: K1e: K1c: Ale: Bk: Thk: Kc
33.0: 27.5: 30.2: 27.5: 1.0: 0.0: 0.10: 27.5

where KIe = 1.1 KIc (see Chapter 3). The crack geometry was assumed to be a
circular surface crack with total crack length 2c = 0.32 and crack depth a = 0.16
in. From da/ dN, DJ( test data with 50% reduction in the properties:

da/dn 11K
0.1000E-02 : 27.500
0.4000E-03 : 25.000
0.2000E-03 18.000
0.5500E-04 9.000
0.2200E-04 4.500

Final results:
Unstable crack growth, max. stress intensity exceeds critical Kef = 27.5 value:
Kmax = 27.53 Kfef. = 0.0000 Kef = 27.50
at cycle no. 691.9 crack size c = 0.369646

By comparing the number of cycles to failure for the two approaches, it can be
seen that it takes only 83 cycles for the Forman and 692 cycles when the
NASA/FLAGRO crack growth equation is used. It can be concluded that the For-
man approach yields a more conservative result when the life of a structural part
is calculated.

4.4 Variable Amplitude Stress

and the Retardation Phenomenon

Fatigue crack growth rate test data needed for life analysis (da/ dN versus 11K)
are produced in the laboratory under the condition of constant amplitude loading.
All of the fracture properties that are compiled in the NASA/FLAGRO material
library [1] (shown in Appendix A) are generated under a constant amplitude load-
ing condition. However, in real situations, most structural parts are subjected to
variable amplitude loading throughout their service life. Crack growth test data
have shown that, under variable amplitude loading, there is delay (retardation) or
acceleration in the amount of crack growth due to high or low loads. Therefore,
the effect of high and low loads on the amount of crack growth must be addressed.
For example, the loads that the wing of a transport aircraft encounters during its
service usage are complex and the effect of stress interaction due to variable
amplitude loading on crack growth rate is an important problem in aircraft design.
Experimental data have indicated that a high tensile load, followed by a con-
stant amplitude load, will reduce or retard the rate of growth, as illustrated in Fig.
 4.4 Variable Amplitude Stress and the Retardation Phenomenon 217

r.-
Retardation Due to High Load
o
III
+
~ High loading

ao Cycle
~--------------------------~~
Number of Cycles, N

Figure 4.23 Illustration of the decrease in the rate of loading due to the high load fol-
lowed by low constant amplitude load

4.23. This phenomenon is called retardation [33]. Retardation or delay in the rate
of crack growth results from the plastic deformation that occurs at the crack tip.
The tensile overload produces large tensile plastic deformation (called the over-
load affected zone) and, upon removal of this load, the material in the vicinity is
elastically unloaded. The plastic zone surrounding the crack tip, however, experi-
ences compressive stresses. Crack growth by subsequent smaller cycles have a
retarded rate across the affected zone. The higher the magnitude of the tensile
overload, the larger the retardation effect when a subsequent low cycle amplitude
is applied.
Even though the plastic deformation under constant amplitude loading has a
smaller dimension, it does not have any retardation effect at the crack tip. The
constant amplitude plastic deformation contributes to the closure phenomenon
described in Section 4.3 where crack surfaces close at a nonzero load level
(Smin < Sci)·
In general, structural parts undergo random loading during their service life.
The load environment contains not only the tensile overloading, as shown in Fig.
4.23, but also other types of overloading, such as compressive and tensile-com-
pressive overload (see Fig. 4.24), which affect the rate of crack growth. Negative
or compressive overloads may have the opposite effect and tend to accelerate the
crack growth rate [34, 35] (it is worth mentioning that for precracking the brittle
material, when it is difficult to precrack under normal tensile mode I loading pro-
cedures, the high compressive load concept can be used to initiate the crack or to
extend the preexisting crack in the material). When a tensile overload is followed
by a compressive overload, the effect of the retardation on the rate of growth is
reduced or may be totally eliminated.
To adjust the crack growth rate due to tensile overload, the empirical crack
growth equations, described in Section 4.2, must be corrected. Various models
218 Chap. 4 Fatigue Crack Growth

Load
Random Loading

Compressive Under Load

Cycle

Tensile & Compressive Overload

Figure 4.24 Illustration of the types of loading that occur in service life

are available that describe the retardation phenomenon. Two well-known mathe-
matical models, called the Wheeler and Willenborg models, based on yield zone
[36, 37] are presently available and will be discussed in Sections 4.4.1 and 4.4.2,
respectively.

4.4.1 Wheeler Retardation Model

In this model, the crack growth equation is corrected to account for the retardation
effect. The effect of the tensile overload on the rate of growth is through the crack
growth reduction factor parameter, Cp [36]. The retarded crack growth rate,
(da/dN)" in terms of the apparent crack growth rate, da/dN, can be written as:

(4.26)

The retardation factor, Cp ' is related to the size of the plastic zones created by the
tensile overload [overload-affected zone, (rp)od, crack size at the onset of over-
load, aOL ' and the plastic zone size for the current or subsequent cycles, (rp)n, as
illustrated in Fig. 4.25:

(4.27)
 4.4 Variable Amplitude Stress and the Retardation Phenomenon 219

Overload Affected Regular Cycles

Plastic Zone Size Plastic Zone Size

I I
~
I (r pn ) I
1
Regular Cycles
1< >1

Figure 4.25 Illustration of plastic zones and the retardation effect as the result of
overload

where m is an empirical constant that can be tailored from variable amplitude test
data to allow for reasonably accurate life predictions. Test data have indicated that
the constant m depends on the material, crack size, and the level of the applied
overload. The quantity (an - a OL ) is the amount of growth associated with n
cycles consumed in the overload affected zone. The plastic zone sizes, (rp)OL and
(rp)n' shown in Fig. 4.25, can be written in terms of the stress intensity factor and
the tensile yield of the material by:

(rp)OL = (1/27T)[(K)oL/O"Yield]2 (4.28a)

(rp)n = (1/27T)[(K)n/O"Yieldf (4.28b)

From Eq. 4.27 (see also Fig. 4.25), it can be seen that the Wheeler retardation
model described by the parameter Cp is effective as long as the plastic zone size
associated with the nth cycle, (rp)n' is within the larger zone (rp )OL. That is, when
the current crack size reaches the end of the yield zone, an = (rp )OL' and the crack
growth reduction factor parameter Cp = 1.

4.4.2 Willenborg Retardation Model

The Willenborg model considers the effect of retardation on the rate of crack
growth through the extent of the plastic zone formed by the overload [37]. The
correction to the crack growth is handled through the use of an effective stress
intensity factor that can be expressed as:

(4.29)
220 Chap. 4 Fatigue Crack Growth

where the reduction stress intensity factor, K red (also called the residual stress
intensity factor, owing to the compressive residual stress state caused by differ-
ences in load levels) as the result of overload is given as [37]:

(4.30)

in which (KOL)max is the maximum stress intensity factor due to the overload and
Kmax,n is the maximum stress intensity factor for the subsequent smaller cycle (see
Fig. 4.25). When the current crack size has reached the boundary of the overload
yield zone (an - a OL = rpOL )' the retardation effect becomes zero, K red = O. Any
overload greater than the previous overload creates a new retardation effect that is
independent from the preceding one,
Equation (4,29) in terms of maximum (Kmax )eff and minimum (Kmin )eff effective
stress intensity factors can be written as:

(Kmax, n)eff = /3 (o-max, n - o-red)( 1Tan) 1/2 (4.31)

(Kmin,n)eff = /3(o-min,n - o-red)(1Ta n)1/2 (4.32)

and the effective stress intensity range is:

(dKn)eff = (Kmax,n)eff - (Kmin,n)eff (4.33)

= dKn

From Eqs. (4.30), (4.31), and (4,32), one can conclude that the effective stress
intensity factor range, (dKn)eff' for a complete cycle is equal to the apparent stress
intensity factor, dKn. That is, in the Willenborg retardation model, both quantities,
the effective stress range (do-)eff and the apparent stress range, do- are the same:

(4.34)

where the effective stress ratio, Reff :

(4.35)

Because the effective and apparent stress intensity ranges are equal in the Wil-
lenborg retardation model, the retardation effect is affected by the effective stress
ratio described by Eq, (4.35). By calculating (dK)eff or dK and Reff , the amount of
growth, da/ dN, in the retardation zone can be computed by any crack growth
equation in which the stress ratio, R, appears, For example, consider the Forman
crack growth relation described by Eq. (4.4):
 4.4 Variable Amplitude Stress and the Retardation Phenomenon 221

The Willenborg retardation effect can be handled cycle by cycle through Eq.
(4.36). This equation is effective as long as Reff is positive. One limitation of the
Willenborg retardation model [described by Eq. (4.30)] is that, for the case of
(KOL )max = 2 X Kmax. n' at the application of the overload where a i = aOL ' com-
plete crack arrest is obtained and crack growth ceases completely. That is:

To account for an overload greater than twice the previous load, the Willenborg
model was revised by Gallagher and Hughes [38] and is known as the "General-
ized Willenborg Model." Gallagher and Hughes introduced a parameter cp into
Eq. (4.30), such that:

where:

(4.38)

The quantity Kmax. th is the threshold stress intensity factor associated with zero
crack growth and (SOL)so is the shutoff overload ratio that can cause complete
retardation when the crack ceases to grow (da/dN = 0):

Example 4.5
An aircraft component is made of7075-T651, L-T, 75F aluminum alloy (constant
amplitude crack growth data as shown in Fig. 4.26 [1]) and is subjected to a vari-
able load environment. The fluctuating load is repeated every three cycles and has
maximum peak stresses of 18, 14, and 10 ksi with R = 0, as illustrated in Fig.
4.27. Visual inspection shows that the maximum detected flaw size is a center
crack with length 2a = 1.0 in. The thickness and width of the part are 0.25 and
10.0 in., respectively. Use the Forman equation to calculate the amount of crack
growth for the following cases: (1) no retardation effect; (2) apply the retardation
effect using the Wheeler model (assume m = 1.4); (3) use the Willenborg model,
and generalized Willenborg models. Conduct the above mentioned analysis for
222 Chap. 4 Fatigue Crack Growth

Crack Growth Data (707S-T6S1 Aluminum Alloy)

10-1

10-2 Compact Tension

Specimen
Z
"0
.......
IV
"0 10-3
...
oj
IV
a::
.J::
~ 10-4
e
<.!) Area of Concern

~
~
()
~ 10-5
jrcraft Part
u C1 I
CJ

(Through Crack)

10-6 10 20 30 40 50 60 70
Kmax , ksi (inch) 112

Figure 4.26 Crack growth rate curve for Example 4.5

Stress

Time
Variable Amplitude Loading
Repeated Every 18, 14, 10 ksi

Figure 4.27 Load spectrum for the aircraft component described in Example 4.5
 4.4 Variable Amplitude Stress and the Retardation Phenomenon 223

the fIrst six cycles (cycles associated with 18, 14, 10, and 18, 14, 10 ksi). The ten-
sile yield of the material is given as 65 ksi [1].

Solution
The equation for the stress intensity factor of a through center crack was formu-
lated in Section 3.2.5 of Chapter 3 [Eq. (3.19)]:

For the first cycle, with u = 18 ksi:

Kmax = 18 X (1T X 0.5) 1/2 = 22.55 ksi (in. 1/2) (/3 = 1)
The Forman crack growth equation for the case of R = 0 can be written as:

The constants c and n can be obtained by selecting two points in the linear region
of the crack growth data (Fig. 4.26), as demonstrated in the example problem of
Section 4.2.

c = 1.13E-8, and n = 4.0

The Forman equation, in terms of constants c and n, becomes:

1. No retardation effect
The amount of growth for the first cycle (u = 18 ksi) with a o = 0.5 (growth at
one tip) is:

(Aa)1 = da = 1.13E-8 (22.55)4'°/(55 - 22.55) = 0.0000900 in.

For the second cycle:

Kmax = 14 X (1T X 0.50009)1/2 = 17.54 ksi (in. 1/2) for u = 14 ksi

(Aa)2 = da = 1.13E-8 (17.54)4'°/(55 - 17.54) = 0.0000285 in.

For the third cycle:

224 Chap. 4 Fatigue Crack Growth

Kmax = 10 x (7T X 0.500118)1/2 = 12.53 ksi (in. 1/2) for IT = 10 ksi

(da)3 = da = 1.13E-8 (12.53)4·°/(55 - 12.53) = 0.0000065 in.

For the fourth cycle:

Kmax = 18 X (7T X 0.5001245)1/2 = 22.5567 ksi (in. 1/2) for IT = 18 ksi

(da)4 = da = 1.13E- 8 (22.5567)4·°/(55 - 22.5567) = 0.0000902 in.

For the fifth cycle:

Kmax = 14 X (7T X 0.500215)1/2 = 17.5472 ksi (in. 1/2) for IT = 14 ksi

(da)5 = da = 1.13E-8 (17.5472)4·°/(55 - 17.5472) = 0.0000286 in.

For the sixth cycle:

Kmax = 10 X (7T X 0.500244)1/2 = 12.533 ksi (in. 1/2) for IT = 10 ksi

(da)6 = da = 1.13E- 8 (12.533)4·°/(55 - 12.533) = 0.00000656 in.

No retardation effect
Original crack length (in.) Final crack length (in.)
0.5000000 0.5000901
0.5000901 0.5001180
0.5001180 0.5001245
0.5001245 0.5002150
0.5002150 0.5002440
0.5002440 0.5002506

2. Wheeler retardation model

To apply the Wheeler model, the quantities associated with the retardation factor,
CP' should be calculated. In this example problem, the first cycle with 18 ksi is
considered as the overload and will retard the growth of the subsequent cycles:

where

(rp)OL = (1/27T)[(K)od ITYieldf

and rpn = (1/27T)[(Kmax)n/ ITYield]2
 4.4 Variable Amplitude Stress and the Retardation Phenomenon 225

(rp)OL = (1/21T)(22.55/65f = 0.0191 in.,

and rpn = (1/21T)[17.54/65F = 0.01159 in.

The amount of growth associated with the first cycle is the same as in the previous
case (no interaction effect) where Cp = 1:

Kmax = 18 X (1T X 0.5)1/2 = 22.55 ksi (in. 1/2) for 0" = 18 ksi
(da)1 = da = 1.13E- 8 (22.55)4°/(55 - 22.55) = 0.0000900 in.

To obtain (dah and (da)3' the quantity Cp for each case must be calculated.
The values of a OL and a1 are 0.5 and 0.50009 in., respectively. The value of Cp
for (dah is:

Cp = [0.01159/(0.5 + 0.0191 - 0.50009)t 4 = 0.5

(da)2 = C/(dK) = 0.5 X 1.13E-8 (17.54)4°/(55 - 17.54)
= 0.00001425 in.

and for the case of (da)3:

Kmax = 10 X (1T X 0.5001043)1/2 = 12.5312 ksi (in. 1/2)

(rp)OL = (1/21T)(22.55/65)2 = 0.0191 in., and rpn = (1/21T)[12.53/65J2

= 0.005917 in.
Cp = [0.005917/(0.5 + 0.0191 - 0.5001043)]14 = 0.195
(da)3 = 0.195 X 1.13E-8 (12.5312)4°/(55 - 12.5312)

= 0.000001268 in.

and the amount of growth associated with the fourth cycle (no retardation) is:

(da)4 = da = 1.13E-8 (22.5567)4°/(55 - 22.5567) = 0.0000902 in.

where

Kmax = 18 X (1T X 0.50010557)1/2 = 22.5563 ksi (in. 1/2)

Accordingly, the values of (da)s and (da)6 can be calculated as shown in the pre-
vious cases of (da)2 and (da)3 by considering the effect of retardation as a result
of 18 ksi overload:
226 Chap. 4 Fatigue Crack Growth

(rp)OL = (1/21T)(22.5563/65)2 = 0.01917,

and rpn = (1/21T)[17.545/65JZ = 0.01160 in.

where

Kmax = 14 X (1T X 0.50019577)1/2 = 17.545 ksi (in. 1/2)

Cp = [0.01160/(0.5 + 0.01917 - 0.50019577)]1.4 = 0.611
(ila)5 = 0.611 X 1.13E-8 (17.545)4.0/(55 - 17.54) = 0.00001746

and in the same way, the quantity (ila)6 can be calculated:

Kmax = 10 X (1T X 0.500213)1/2 = 12.5326 ksi (in. 1/2)

(rp)OL = (1/21T)(22.5563/65)2 = 0.01917 in,

and rpn = (1/21T) [12.5326/65JZ = 0.0059197 in.

Cp = [0.005919/(0.50010557 + 0.01917 - 0.500213)]1.4 = 0.1945
(ila)6 = 0.1945 X 1.13E- 8 (12.5326)4.0/(55 - 12.5326) = 0.000001276 in.

Wheeler retardation effect

Original crack length (in.) Final crack length (in.)
0.50000000 0.50009010
0.50009010 0.50010430
0.50010430 0.50010557
0.50010557 0.50019577
0.50019577 0.50021300
0.50021300 0.50021430

3. Willenborg retardation model

The Willenborg correction to the crack growth is handled through the use of an
effective stress intensity factor and the effective stress ratio. From Eq. (4.29):

(where Kmin = 0)
and

From the above equations, it can be deduced that:

 4.5 Cycle by Cycle Fatigue Crack Growth Analysis 227

which states that for the case of R = 0, both the Willenborg and the "generalized"
Willenborg model cannot be effective to correct the retardation effect. For any
other stress ratio larger than zero, the Willenborg retardation models can be used
to address the retardation effect. Therefore, case 3 is identical to case 1 of this
example problem. The following summary table summarizes the calculated crack
growth for the first six cycles associated with 18, 14, 10, and 18, 14, 10 ksi).

Summary table for example 4.5

No retardation effect Wheeler retardation effect
Original crack Final crack Original crack Final crack
length (in.) length (in.) length (in.) length (in.)
0.5000000 0.5000901 0.50000000 0.50009010
0.5000901 0.5001180 0.50009010 0.50010430
0.5001180 0.5001245 0.50010430 0.50010557
0.5001245 0.5002150 0.50010557 0.50019577
0.5002150 0.5002440 0.50019577 0.50021300
0.5002440 0.5002506 0.50021300 0.50021430

4.5 Cycle by Cycle Fatigue Crack Growth Analysis

To prevent catastrophic failure due to the presence of undetected flaws that would
result in the loss of life and the structure, it is important for the engineer to design
components that provide structural integrity, even in the presence of undetected
flaws. Therefore, it is essential to know whether the preexisting flaw will grow to
a critical length during its service life.
Owing to the complexities in the crack growth analysis that emerge as a result
of a large number of cycles (containing constant and variable amplitude load)
together with complicated load and crack geometries, automated flaw growth pro-
grams are usually employed. Ideally, a flaw growth computer program should
have libraries for (1) standard NDE flaw sizes, (2) stress intensity factor solutions
for different crack geometries and loading conditions, (3) empirical fatigue crack
growth equations, and (4) material properties. In addition, it should be efficient
and easy to run ("user friendly").
Most crack growth computer programs that are tailored for the aircraft industry
have the capability to perform variable amplitude fatigue life evaluation by
employing one or both of the retardation models discussed in Section 4.4. It
should be noted that, based on NASA requirements stated in NHB 8071-1, "Frac-
ture Control Requirements for Payloads Using the National Space Transportation
System (NSTS)," the retardation effect is not allowed to be considered in evalu-
ating the life of fracture critical or high-risk components that will be used as pay-
loads in the NSTS transporter. The cycle by cycle crack growth procedure
generally consists of the following steps:
228 Chap. 4 Fatigue Crack Growth

Step 1. A standard NDE initial crack size, a l , is assumed. As an example, for

standard penetrant inspection, the initial crack size is a part through crack with
depth a = 0.075 and 2c = 0.15 in., see Table 5.5 of Chapter 5.
Step 2. The stress intensity range for the first cycle can be determined, MI =
{3i:1uI (1Ta) 1/2, where i:1uI is the stress range in the first cycle and {3 is the appro-
priate correction factor for crack geometry and loading conditions. Note that the
crack growth analysis performed for a part through crack should consider the
growth for both the depth (a-tip) and the length direction (c-tip).
Step 3. Determine the amount of growth for the first cycle, i:1a l , by using the
appropriate crack growth equation that considers the effect of the stress ratio, R
[(for example, the Forman relationship described by Eq. (4.4)] or directly from
the da/ dN versus M curve. The amount of growth for the first cycle, i:1a l ' is:

The new crack length will be a2 = a l + i:1a l • If the crack geometry in consid-
eration is a part through crack, it is preferable to have the da/ dN versus M test
data for the crack geometry under study.
Step 4. The size of the plastic zone, rp ' in the first cycle is calculated by Eq.
(3.33) described in Chapter 3. If the first cycle is considered as the overload cycle
then the extent of the plastic zone is:

where for plane stress condition:

and for plane strain condition:

Step 5. The new crack length is now a2 and the stress intensity range for the
second cycle i:1K2 = {3i:1u2(1Ta) 1/2. If the new cycle has a load magnitude larger
than the previous cycle, the new overload plastic zone must be calculated. On the
other hand, if the subsequent cycle has a smaller load magnitude, then the effect of
retardation has to be considered. The extent of the yield zone for the new cycle is:

Step 6. If the extent of the plastic zone Z < ZI' the correction to retardation
effect according to the Willenborg [Eqs. (4.33) and (4.35)] or Wheeler model [cal-
 4.5 Cycle by Cycle Fatigue Crack Growth Analysis 229

culate the quantity Cp from Eq. (4.27)] is needed. For the Willenborg model the
new crack length is:

where

da 2 = (da/dNh X 1

The quantity (da/dNh is obtained by using the dKeff and Reff in the Forman
crack growth equation.
When using the Wheeler model, the stress intensity factor, (dK)2' is calculated
and the retardation factor, Cp , is incorporated to calculate the new crack length:

Step 7. If the extent of the plastic zone Z > Zj (no retardation effect), deter-
mine the quantity da 2 = (da/ dN)2 X 1 and the new crack size, a3 , as:

a3 = a2 + (da/dNh X 1

For the subsequent cycles steps 6 and 7 can be repeated in such away that the
nth cycle with new crack length an replaces a2 and the subsequent cycle an + 1
replaces a 3 •
Table 4.2 shows the current state of the art computer codes used in various
aerospace programs.

Table 4.2 Fatigue crack growth computer codes used in several aerospace industry

Programs Computer code

C 17 Program (Douglas Aircraft) Air Force Cracks Program

High-Speed Transporter (Douglas Aircraft) NASGROW
International Space Station NASA/FLAGRO
(McDonnell Douglas Aerospace) NASA/FLAGRO
MSC/pDA Supported Programs PFATIGUE/nCode
F-18 Aircraft (Northrop) NORCRAK
Douglas Aircraft (MD-95 Project) CRACK95(DAC)
NASA'S Crack Growth Analysis FASTRAN
(Using Crack Closure Concept)
230 Chap. 4 Fatigue Crack Growth

4.6 Structural Integrity Analysis of Bolted Joints

Under Cyclic Loading

There are several types of joints that are used to join structural parts. The most
common types have bolted, welded, or pinned connections. A joint can be viewed
as a source of stress concentration and preventative measures must be taken to
minimize structural failure. In this section, bolted joints (also called mechanically
fastened joints) are discussed briefly and the emphasis is given to the integrity of
the bolts and the pads (pads are referred to as abutments, the two pieces being fas-
tened together) when they are subjected to a fluctuating load environment. Many
structures, such as buildings, bridges, space vehicle, and airplanes, rely heavily
upon bolted joints for their structural integrity. The factors that are important to
consider in designing and analyzing a bolted joint are the bolt material, pad mate-
rial, bolt pattern, and the degree of preload.

4.6.1 Preloaded Bolt Subjected to Cyclic Loading

Fatigue failure of bolts usually occurs at the threaded location where bolt and nut
are engaged [39, 40]; however, in some cases (less probable) bolt failure is
observed in the shank to bolt head area, as shown in Fig. 4.28(a). Experimental
data on threaded bolts subjected to rotating-bending tests suggest that a larger
root radius reduces the stress concentration at the root and increases the fatigue
life of the bolt [41]. A large root radius allows smoother flow of stress in that

Maximum Inwced Stress

A Thr eaded Bol t

Stress concentration
Bolt/Nut Engagement
site

~is
Low Fatigue Life Improve Fatigue Life Rolled Process Greatly
Improve Fatigue Life
(Large Radii)
(Large Radii and Rolled
(b) Threads)

Figure 4.28 Probabale locations where a threaded bolt fails

 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 231

region, thus reducing the degree of stress concentration, Kt • Other experimental

studies have show that thread processing (by rolling the threads) improves the
fatigue life and this method is preferred over machining, Fig. 4.28(b) [42, 43].
Rolling threads will introduce compressive residual stresses at thread roots which
help to prevent fatigue failure of bolts in a joint (thread rolling introduces a larger
root radius and smoother stress flow). Threaded bolts used in fastening two frac-
ture critical pieces of hardware which is considered as a high risk joint are
required to undergo a rolling process in order to avoid fatigue failure at the roots
during their service life, see figure 4.28(b). In addition to the rolling process for
reducing the stress concentration in the threaded region, it is also a good practice
to distribute the load among threads for avoiding crack initiation and improving
the fatigue life of the bolted joint. The designer should be aware that the first few
threads which are engaged carry the major part of the load.
A tensile preloaded bolt is also considered desirable in reducing the fatigue
failure and increasing the life of a bolt (by decreasing the load amplitude at the
expense of increasing the mean stress) when it is exposed to cyclic loading as
illustrated in Fig. 4.29. Preloading a bolt in a joint is the result of an applied
torque for the purpose of tightening the joint and to prevent any gapping
between the two fastened plates. Consider a bolt in a bolted joint with maxi-
mum fluctuating stress of O.3lTUL with a stress ratio R = O. The calculated life
of this bolt is higher in the presence of preload. Higgins [42] in his experimen-
tal work with high strength steel bolts concluded that the pre loading can
increase the fatigue life considerably than otherwise possible with similar joints.
The following bolt analysis introduces some of the factors that are involved
when assessing the fatigue life of a bolt.
When a bolt is subjected to an alternating tensile applied load, (Pt )applied' and a
preload, ppreJoad (see Fig. 4.30), the total bolt load, Ph' can be obtained [44]
through the following relationship:

Ph = Ppreload + [K(P)applied] X Factor of Safety (4.40)

Degree of Preload

Figure 4.29 The effect of preload on fatigue life of the part

232 Chap. 4 Fatigue Crack Growth

Bolt
----........ -"- ..
Sh~~...A.b.u.tm.e.n.t.#.l""""""""BlI)J[tl
Abutment # 2 I t2

Figure 4.30 A bolted joint with applied tensile load

where

T
Ppreload = DJL (4.40a)

T is the mean wrenching applied torque and D is the nominal bolt diameter. The
quantity JL is the friction coefficient (at the nut to bolt assembly) and it's value is
dependent on the degree of lubricant at the nut to bolt assembly. The joint stiff-
ness factor, K:

(4.41)

where:

and Am and Ab are the compressed joint area and the bolt shank area, respectively.
Eb is the modulus of elasticity in tension of the bolt and Em is the modulus of elas-
ticity in compression of the abutment material. The equations describing Am' A b ,
and Em are:

where

B = D head + L(tan 30)/2

 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 233

and D head is approximately 1.6Dbolt ' Furthermore,

and

Let us for simplicity assume that the bolt shown in Fig. 4.30 is subjected to a fluc-
tuating stress (O't)applied = 0.30'UL with stress ratio of R = 0 and O'preload = 0.50'UL'
Other pertinent information related to the bolt, plate geometry, and material prop-
erties needed to calculate the joint stiffness factor, K, are: D bolt = 0.25 in.,
tshim = 0, t1 = t2 = 0.319 in., E1 = E2 = 1O.8E + 6 psi, Eb = 29.1E + 6 psi,
O'UL = l.lE + 5 psi, and D head = 1.6D ho1e ' with factor of safety = 1.4.

L = tl + t2 + tshim 0.638 in.

C = tl /2 + t 2 /2 + tshim 0.319 in.
tstack = L 0.638 in.
B = Dhead + 0.5 tstack (tan 30) 0.542 in.
Am = 1T/4(B 2 -D;ole)= 0.217 in 2•
Ab = 1T/4 (D bolt )2 0.049 in 2.
K = C/[L(l + AmEm/AbEb)] 0.189

Equation (4.40) for maximum alternating bolt stress becomes:

O'b = 0.50'UL + 0.189 X 1.4 X 0.30'UL (4.42a)

In terms of ultimate stress, O'UL:

(4.42b)

The second quantity in Eq. (4.42) is fluctuating between zero and 0.07930'UL' Fig-
ure 4.31 illustrates the cyclic loading that the bolt will experience when it is sub-
jected to mean stress slightly higher than the preload induced stress, O'M = 59.34
ksi > preload = 55 ksi. It can be concluded that, in the presence of preload, the
mean stress on the bolt increases, but the amplitude of the fluctuating stress,
O'a = 4.34 ksi < (0.3/2) O'UL = 16.5 ksi, decreases, which results in increasing the
fatigue lifetime of the bolt.
Example 4.6 iIlustrates the effect of preload on service life of a bolt. Fatigue
crack growth analysis of pads in a joint is discussed in Section 4.6.2. Example 4.7
shows the analysis approach to evaluating the integrity of pads in a bolted joint by
234 Chap. 4 Fatigue Crack Growth

°UL
-2- 1-T"""'ir""'T"""'T"""T"""T""......., . . . . . - -

°UL 0max = 33 ksi

4 --_. ~ oM = 16.5 ksi
~~~~~~~-------Time

Figure 4.31 Illustrating the effect of preload on a bolt subjected to fluctuating load

0cf'0.44 in
-.....
ill11

0
SeQ7 Model

~~( '~I 'ffffl

1~
~

SCOB Model

Figure 4.32 A surface crack shown on the shank area of a bolt described in Exam-
pie 4.6

assuming the existence of a comer crack emanating from the holes. The final
results, in terms of the NASA/FLAGRO computer code output, are included.

Example 4.6
A fracture critical bolt (shown in Fig. 4.32) is subjected to a fluctuating load of
(Pt)applied = 5000 lb with stress ratio R = O. The bolt is made of A286 high
strength steel with fracture properties shown in Table 4.3. The threads are rolled
Table 4.3 Material properties from NASA/FLAGRO, A286, 200 ksi, (forged rod, L-R)

lTYield lTUI K 1c K 1e AK BK C n p q l1Ko Rei a SR

190 200 100 140 1.0 0.5 0.3E-8 2.1 0.25 .25 3.5 0.2 3.0 0.3
 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 235

to minimize the stress concentration I. The standard eddy current inspection per-
formed on the bolt indicated that the initial flaw size in the shank area is a part
through crack having half crack length c = 0.075 in. and crack depth a = 0.064
in.; see Fig. 4.32. Evaluate the life of the bolt for the preload of 0.50'UL where the
joint stiffness factor, K = 0.25. Compare the results of the analysis with no pre-
load case. The factor of safety for the load environment is 1.4.

Solution
From Eq. (4.40), the bolt load can be written as:

Pb = Ppreload + [K(P).pphed] X 1.4

where K(Pt).pplied = 1250 Ib and O'prelo.d = 0.50'UL' The load variations that the
bolt is subjected to for the two above cases (with and without preload) are shown
in Fig. 4.33. The crack geometry used for crack growth analysis is a thumbnail
crack as shown in Fig. 4.32 (designated by SC07 in the NASA/FLAGRO com-
puter code, see also the note provided at the end of this example problem) [1].

<JUl
(200ksi)

0.75<JUl
(lS0ksi)

<JM = 105.76 ksi

R=0.67
O.S<JUl
(Preload= 100 ksi)
(100ksi)

0.25<JUl
(SOksi)

<JM = 23.04 ksi

R=O.O
(No Preload Case)
_.....11_...11.._......&......III......JI._......._ _ _ _ _ _ Cycle

Figure 4.33 The Cyclic load environment for two cases: (1) preload and (2) no pre-
load

I The rolling process will introduce compressive residual stresses at the threaded roots. For this
reason, the crack growth analysis in this example problem is performed in the shank area where the
probability of failure is higher (see the note provided at the end of this example problem).
236 Chap. 4 Fatigue Crack Growth

The crack depth length, a, in terms of crack length 2e, was expressed in terms of
the diameter D = 2r = 0.44 in. [1] as:

D
a = - [tan (2e/D) + 1 - (tan 2 {2e/D) + 1)1/2]
2

The fracture mechanics analysis for both the preloaded and no preload (where the
joint stiffness factor K = 1) cases are conducted and the final results in terms of
NASA/FLAGRO output are shown here. The final results clearly indicate that
longer life is expected in the presence of the preload.

FATIGUE CRACK GROWTH ANALYSIS

(computed: NASA/FLAGRO Version 2.03, January 1995)

PROBLEM TITLE
BOLT ANALYSIS WITHOUT PRELOAD

GEOMETRY
MODEL: SC07-Part-circular Surf. crk on cylinder eire. plane
Cylinder Diameter, D = 0.4400

FLAW SIZE
a (init.) = 0.6461E-Ol
c (init.) = 0.75E-Ol

MATERIAL
MATL 1; A286 (200ksi Bolt material)
Forg. rod, L-R
Material Properties;
UTS YS Kle Klc
200.0 190.0 140.0 100.0
Material Crack Growth Eqn Constants
C n p q DKo Rcl Alpha Smax/SIGo
0.3D-08: 2.1 0.25 0.25 3.50 0.20 3.00 0.30

FINAL RESULTS
Net-section stress exceeds the flow stress.
(Flow stress = average of yield and ultimate)
At cycle no. 91536.03
Crack size a = 0.203986
Corresponding semicrack length, c = 0.750000E-Ol

Note: The final results (net-section stress exceeds the flow stress) do not represent
the final failure where the total separation occur (failure ofthe bolt in two parts).
 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 237

When the total length of the advancing crack is in such a way that the net section
(W - a) is yielding, the linear elastic fracture mechanics cannot further be used to
describe the crack tip stress field and the corresponding crack growth rate.
Based on the above result, the number of cycles to failure for the bolt when it
is not preloaded is shown to be 91,536. However, the life of the bolt is expected to
increase by applying (Tpreload = 0.5(TUL to the bolt. This is true because upon the
application of preload, some of the applied load is absorbed by the jointed plates
in compression and only a portion of it will be affected by the bolt, K (P)applied'
where K = 0.25. The following is the final result of the analysis conducted by
NASA/FLAGRO where the magnitude of applied fluctuating stress is reduced by
the joint stiffness factor K = 0.25 «(Tmax = 111.5, and (Tmin = 100 ksi).

FINAL RESULTS (WITH PRELOAD) :

Net -section stress exceeds the flow stress.
(Flow stress = average of yield and ultimate)
At cycle no. 832977.38:
Crack size a = 0.132176
Corresponding semicrack length, c = 0.750000E-0l

The fracture analysis of the bolt with preload clearly indicates that the number
of cycles to failure has increased by nine times as compared to the previous analy-
sis where preload was not incorporated.
Final Note: If a thorough inspection is possible to ensure that the rolling
threads are properly performed and fully effective, then the area of concern would
be in the shank and bolt head-to-shank region. However, because 100% inspec-
tion at the roots are not possible, it is recommended that bolt fracture analysis be
performed in the threaded region only. Note that in analyzing the threaded region
(using Se08 crack model in the NASA/FLAGRO computer code or any other
available code, see Fig. 4.32), the far field stress will be based on the minor diam-
eter which will give higher preload stress compared with the shank region.

4.6.2 Fatigue Crack Growth Analysis of Pads

in a Bolted Joint
Generally, in joints that contain more than one bolt, the failure of one bolt due to
cyclic loading will not necessarily cause the total destruction of the joint. If by
analysis it can be shown that the remaining bolts can withstand the redistributed
load during its service life, the system has a fail-safe design. Even though the bolts
in a given joint might be classified as having a fail-safe design and the failure
might be prevented by redistribution of load to the remaining bolts, the fatigue
cracking may initiate at other locations in the joint and cause a detrimental effect
on the structural parts fastened together. For example, a crack may initiate from
the hole in the fastened plates (as illustrated in Fig. 4.34) where the two plates are
238 Chap. 4 Fatigue Crack Growth

Four Bolted Joint

Crack emanates
from a hole in the
Joint

Structural Part 1 I~ ~I
Figure 4.34 A four bolt joint where the two plates are fastened together

bolted together by a four-bolted joint pattern. The initiated crack from one of the
holes in the joint may grow to its critical size due to the fluctuating load environ-
ment, and cause the complete separation of the two structural parts.
When a bolted joint such as the one shown in Fig. 4.34 is subjected to a fluc-
tuating load environment, the tension and shear forces applied to the bolts will
induce localized bending and bearing stresses on the plate. The calculated induced
stresses are useful information for determining whether the joint can maintain its
integrity during its usage period. Figure 4.35 illustrates the possibility of a crack
emanating from a hole in a joint as a result of localized cyclic bearing and bend-
ing stresses. The following analysis briefly describes the methodology used in
evaluating the structural integrity of a bolted joint when it is subjected to a resul-

Bolted Joint

Plates

pr
tl t2
Crack ~ Load

Figure 4.35 The axial and shear induced loads at a bolt

 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 239

tant force and moments (Fx' Fy ' Fz' Mx ' My, MJ, applied at the centroid of the
bolt pattern, shown in Fig. 4.35. Using the bolt load analysis approach, described
by Bruhn in Reference [45], the maximum resultant bolt tension load, PI' and
shear, V, can be calculated. The bearing stress that is effective for growing the
crack (see Fig. 4.35) is given by:

O"br = V/Dt (4.43)

where D and t are the diameter of the hole and plate thickness, respectively. More-
over, the bending stress induced in the pad, due to the axial load on the bolt, PI'
also contributes to growing the crack from a hole and can be calculated as a bath
tub channel fitting (case a), angle fitting (case b), or from the plate bending
approach (case c), depending on the geometry of the joint [46] (see Fig. 4.36).
For the simple case of plate bending (case c), the induced plate bending stress,
O"bending' to be employed in the NASA/FLAGRO computer code can be estimated
for the region designated as ABeD by the following relationship (see Fig. 4.37):

O"bending = 6M/bt 2 (4.44)

Example 4.7
A bolted joint with a four-bolt pattern (situated in the flange of an I-beam), shown
in Fig. 4.38, is subjected to forces and moments at its centroid location. The bolt

(1) The Bending Plate (Flange)

(2) Plane of The Web
(a) (3) Integerated Stiffnere Plate
Bath Tub (Channel) Fitting (4) Integerated Stiffnere Plate

Figure 4.36 The three types of fitting design used in designing a joint: (a) channel,
(b) angle and (c) plate bending design [46]
240 Chap. 4 Fatigue Crack Growth

Plate Bending

Where b=2d + Dhead

Applicable to the Region of ABeD

Figure 4.37 Evaluating bending stress in a plate subjected to tensile bolt load

Figure 4.38 Surface crack in a preloaded bolt subjected to cyclic loading (part 1 of
Example 4.7)

Table 4.4 Material properties from NASA/FLAGRO, 2219-T851 L-T

lTYield lTVI K1c K1e AK BK C n p q I!..Ko Rei a SR

53 65 100 46 1.0 1.0 O.l2E-7 3.2 0.50 1.0 2.5 0.7 1.5 0.3

is made of A286 high strength steel (Fig. 4.38) with fracture properties for bolts
and plated shown in Tables 4.3 and 4.4 respectively. The analysis indicated that
the maximum fluctuating bolt tension load Pt = 2500lb with resultant shear load
V = 2000 lb (R = 0). To improve fatigue life and avoid gapping, the bolts are
required to be preloaded as high as 0.35 0'U1' The load spectrum for one service
life is given as:
 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 241

Load Spectrum Table for Example 4.7

Cyclic stress (%Limit)

Steps Cycles Minimum Maximum

100 0.0 100

2 300 0.0 90
3 400 0.0 80
4 700 0.0 70
5 1000 0.0 60
6 2000 0.0 50
7 2800 0.0 40
8 5500 0.0 30
9 9000 0.0 20
10 15019 0.0 10
II 28853 0.0 7

(1) Determine if the bolts can survive four service lives by assuming the exis-
tence of a part through crack in the shank location with a = 0.04 and 2c = 0.1 in.
and (2) Assume a comer crack of length c = 0.05 and depth a = 0.05 in. is ema-
nating from the hole in the joint (see Fig. 4.39). Determine if the joint can survive
the above environment (minimum of four service lives) with far field tension

Figure 4.39 A bolted joint and crack geometry for a crack emanating from a hole
(Part 2 of Example 4.7)
242 Chap. 4 Fatigue Crack Growth

stress O'Tension = 10 ksi (R = -1), as shown in Fig. 4.39. The joint stacking thick-
ness and other related information are: tl = t2 = 0.4 in., tshim = 0, D bo1t = 0.5 in.,
D head = 0.8 in. and EI = E2 = 1O.8E + 6 psi, Eb = 29.1E + 6 psi, O'UL = 1.99E
+ 5 psi, d = 2D.

Solution
The stiffness factor, K, for this joint can be calculated by employing Eq. (4.38):

L = t) + t2 + tshim 0.800 in
C = t)/2 + t2 /2 + tshim 0.400 in
tstack = L 0.800 in
B = Dhead + 0.5 tstack (tan 30) 1.014 in
Am = 1T1/4 (B 2 -D;ole) 0.611 in 2
Ab = 1T1/4 (D bo1t )2 0.196in 2
K = C/[LO + AmEm/AbEb)] 0.232

The alternating tensile stress in the bolt can be calculated with the presence of
induced preload stress of O'preload = 0.3 X 200 = 60 ksi. From Eq. (4.40), the total
bolt stress can be written as:

O'bolt = O'preload + K O'applied

where the maximum and minimum alternating bolt stresses (based on PI = 2500
lb) are 62.96 and 60 ksi, respectively. The final results pertaining to fracture
analysis of bolts subjected to the fluctuating load environment (described in the
load spectrum table for Example 4.7) are shown for four service lives as:

FINAL RESULTS:
All stress intensities are below the fatigue threshold.
NO growth in four service lives
Crack size a = 0.455928E-Ol
Corresponding semicrack length, c = 0.500000E-Ol

The final result on the fracture analysis of bolts indicates that all the calculated
stress intensity factors are below the threshold stress intensity factor, LlKth
(described in Section 4.3.1.1) and thus no growth is expected when the bolt is sub-
jected to the fluctuating load environment shown below. The induced tensile
stress in the bolt due to the 2500 lb fluctuating load is not effective in driving the
crack above the threshold value.
 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 243

Load Spectrum

% Limit Load

Steps Cycles Min Max

100.00 60.00 62.96

2 300.00 60.00 62.66
3 400.00 60.00 62.36
4 700.00 60.00 62.07
5 1000.00 60.00 61.78
6 2000.00 60.00 61.48
7 2800.00 60.00 61.18
8 5500.00 60.00 60.88
9 9000.00 60.00 60.59
10 15019.00 60.00 60.29
11 28853.00 60.00 60.20

To examine the integrity of the pad in the joint, the bending stress, (Tbending' and
bearing stresses, (Tbr' should be calculated to check if the comer crack is stable
during its four service lives. From Eq. (4.44):

(Tbending = 6M / bt 2

where M = Ptd = 2500 X 2D = 2500 in-lb. From Fig. 4.37: b

(4D + D head ) = 2.8 in.

(Tbending = 6 X 2500/(2.8 X 0.4 2) = 33.48 ksi

From Eq. (4.43), the bearing stress, (Tbr:

(Tbr = V/Dt
(Tbr = 2000/(0.5 X 0.4) = 10.0 ksi
FINAL RESULTS:
Unstable crack growth, max stress intensity exceeds critical value:
Km.. = 56.44 Kfef = 0.0000 Kef = 50.60:
At the end of two service lives:
Crack sizes: a = 0.381844, c = 0.294645, a/ c = 1.29594

The above results indicate that unstable crack growth occurs when the pads in
the bolted joint are exposed to two service lives (associated with the number of
cycles used in the two load spectrum). In summary, the pads in the bolted joint
244 Chap. 4 Fatigue Crack Growth

could not survive the minimum of four service lives. The designer may be advised
to make one of the following changes in order to have a safe joint:
1. Reduce applied loads by redesigning the joint.
2. Use special NDE inspection to obtain a smaller flaw size, as discussed in Section 5.4
of Chapter 5.
3. Consider changing the pad material.

Example 4.8
A low risk structural component carries a fluid line (a clamp is holding the fluid
line as shown in Fig. 4.40) and is attached to a primary piece of hardware (a high
risk part) by a four-bolt pattern (bolt diameter D = 0.25 in. and plate thickness
t = 0.1 in.). The fluid line and the clamp assembly have a total mass of 5 lb-m and
are subjected to 50 ft./sec 2 acceleration (in the x and y directions) in a fluctuating
load environment. Use NASA/FLAGRO or any other computer code to determine
the integrity of the interface during its service life (one service life is equal to four
times the number of cycles shown in the load spectrum table). Both the primary
and secondary structures are made of 2219-T851 aluminum alloys with fracture
properties as shown in Table 4.4. Assume a comer crack of length 0.05 in. with
aj c = 1 preexisting at the hole.

Bolted Joint (clamp interface with

the primary structure)

Figure 4.40 Loads and crack geometry descibed in Example 4.8

 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 245

Load Spectrum for Example 4.8

%LimitLoad

Steps Cycles Max % Min%

100.00 100.00 -100.00

2 500.00 90.00 - 90.00
3 1000.0 80.00 - 80.00
4 2000.0 70.00 - 70.00
5 3000.0 60.00 - 60.00
6 4000.0 50.00 - 50.00
7 5000.0 40.00 - 40.00
8 10000.0 30.00 - 30.00
9 12000.0 20.00 - 20.00
10 15000.0 10.00 - 10.00

Solution
The forces induced in the x and y directions due to 50 ft./sec 2 acceleration are
Fy = Fx = 5 X 50 = 250 lb. These forces carry moments Mx and My at the cen-
troid of the four-bolt pattern as shown in Fig. 4.41. The reactions at the bolted
joints due to Fx. Fy. Mx. and My can be calculated:

~Y ~
R R
R= Fx /4=250/4=62.5Ib
~
~F
R<e
R= Fy /4=250/4=62.5Ib
Y

Mx=Fy x L
My=Fx xL
X4

-....
~2
j..e2
V 3
3i'R

Y
Y R.~ rMx i'.,
l~Z My ~ ~ 4 R

R= (Fx x L)/a R= (F y x L)/b

62.5x2.2/1.5=91.66 62.5x2.2/2.5=55Ib

Figure 4.41 Bolt load due to forces and moments applied to the centroid of the bolt
pattern
246 Chap. 4 Fatigue Crack Growth

The maximum tension load is taken by bolt number 1 where the reactions from
Mx ' and My contributions are added together (91.66 + 55 = l46.66lb). The bend-
ing stress induced on the plate (t = 0.10 in.) due to tension on the bolt can be
determined by using an equation describing the bending stress for case (c) of Fig.
4.36; see also Fig. 4.37. From Eq. (4.44) (see also Fig. 4.42):

M = 146.66 X (1.5/2) = 110 in.-Ib

b = 2 X 0.75 + 1.6 X 0.25 = 1.9 in., and t 2 = 0.010 in. 2
O'bending = 6M/bt 2 = (6 X 110)/(1.9 X 0.010) = 34736 psi
The bearing stress due to shear forces will also contribute to crack growth. The
resultant of the shear forces for the bolt at location 4 is (see Eq. (4.43) and also
Fig. 4.43):

R = V = (91.66 2 + 55 2)1/2 = 107lb

O'br = V/Dt = 107/(0.25 X 0.1) = 4276 psi

Fatigue crack growth analysis
(computed: NASA/FLAGRO version 2.03, January 1995.)
U.S. customary units [inches, ksi, ksi sqrt(in)]

GEOMETRY
MODEL: CC02-Comer crack from hole in plate (2D)

M=P t x d
a=6M1bt 2
Where b=2d + Dhead

Figure 4.42 Bolt tension load and plate bending as described in Example 4.8
 4.6 Structural Integrity Analysis of Bolted Joints Under Cyclic Loading 247

Figure 4.43 Shear resultant and the induced bearing stress as described in Exam-
ple 4.8

Plate thickness, t = 0.1000

Plate width, W = 2.0000
Hole diameter, D = 0.2500
Hole-center-to-edge dist., B = 0.5000
Poisson ratio = 0.30

FLAW SIZE
a (init.) = 0.5000E-Ol
c (init.) = 0.5000E-OI
a/ c (init.) = 1.000

MATERIAL PROPERTIES
UTS YS Kle Klc Ak Bk Thk Kc
65.0 53.0 46.0 33.0 1.00 1.00 0.100 65.7

Crack Growth Eqn Constants

C n p q DKo Rei ll' Smax/SIGo
0.119E-7 3.l56 0.50 1.00 3.00 0.70 1.50 0.30

ANALYSIS RESULTS
FINAL RESULTS
Critical crack size has NOT been reached.
At cycle no. 15000.00 of load step no. 10
of block no. 4 (at the end of 4 servise lives)
Crack sizes: a = 0.66534IE-0l, e = 0.183923, a/e = 0.361750
248 Chap. 4 Fatigue Crack Growth

The result of the analysis indicated that the part can survive the load environ-
ment with final crack length a = 0.0665 in. (initial crack length was 0.05 in.) and
ale = 0.36.

4.7 Material Anisotropy and Its Applications

in Beam Analysis

Almost all of the beams used in structural hardware are oriented in such a way
that the elongated grain direction (L-direction) is along the length of the beam.
The long transverse, T, and short transverse, S, directions are associated with the
width of the flange and the web, respectively (as illustrated in Fig. 4.44 for a beam
with I cross sectional area). Fracture toughness test data on most alloys have
shown lower values in the T-L and S-L compared to the L-T and L-S directions,
as indicated in Table 4.5 for a few selected materials extracted from the
NASAIFLAGRO material library [1] (see Section 3.3.1 of Chapter 3 for standard
nomenclature relative to directions of mechanical working (grain direction) for
rectangular sections). For 2219-T851 aluminum alloy, the fracture toughness val-
ues for the L-T and T-L directions are almost equal; however, for other materials,
such as 2124-T85I aluminum alloy, the difference is no longer small. The analyst
must use the appropriate value of fracture toughness for the life evaluation of the
part under study.

L-S S-L

o
Flange

- -

m
Ji~~~~~:::::::::::::::::::::::~(~K:C:)S:-L (I<c)L-S
<

ill
(Kc)T-L < (!<c)L-T
Web Web

O-.r
~ L-T T-L

Figure 4.44 The material anisotropy, crack orientations, and fracture toughness for
an I-beam
 References 249

Table 4.5 Fracture toughness properties for several aluminum alloys with respect to their
material grain orientaions

2219-T851 K Ic K Ie YS UTS
(Pit & Sht.)
(L-T) 33 46 53 65
(T-L) 31 43 50 66
2219-T87 K Ic K Ie YS UTS
(Pit & Sht.)
(L-T) 30 42 57 68
(T-L) 27 38 58 69
2219-T851 K Ic K Ie YS UTS
(Pit & Sht.)
(L-T) 30 42 63 71
(T-L) 23 31 64 72
(S-T), (S-L) 21 28 60 69

When a beam is mechanically bolted to another structure (bolted joints at the

flange or web locations), it is important to select the correct material orientation
when evaluating the usage life of the part under study. In most cases, when the
beam is subjected to far-field induced bending and tension stresses (as shown in
Fig. 4.44), the L-S and L-T crack orientations (for the flange and the web, respec-
tively) are considered to be the appropriate orientation to use for fracture analysis
of the joints, even though the (Ke) T-L and (Ke) S-L values are lower than
(KJL-T and (Ke)L-S, respectively. For this reason, in the case of a flange, as
well as the web, it is recommended that the analyst orient the assumed initial flaw
(crack emanating from a hole) perpendicular to the far field stress when conduct-
ing safe-life analysis (Fig. 4.44). It should be noted that the S-L and T-L crack
orientations are not affected by the far field induced stresses. However, when
dealing with localized induced stresses such as bearing stress at the hole of a joint
(no contribution due to far field stress), the material orientation must be consid-
ered. In using the L-T or L-S fracture toughness values, less conservative results
(a larger number of cycles to failure) can be expected.

References

1. Fatigue Crack Growth Computer Program "NASA/FLAGRO," developed by R. G.

Forman, V. Shivakumar, and J. C. Newman. JSC-22267A, January 1993.
2. W. G. Clark Jr. and S. J. Hudak Jr., "Variability in Fatigue Crack Growth Rate Test-
ing," J. Test. Evaluat., Vol. 3, No.6, 1975, pp. 454--476.
250 Chap. 4 Fatigue Crack Growth

3. D. W. Hoeppner and W. E. Krupp, "Prediction of Component Life by Application

of Fatigue Crack Growth Knowledge," Lockheed California Company, Burbank,
CA, pp. 8-9.
4. P. C. Paris, M. P. Gomez, and W. E. Anderson, "A Rational Analytic Theory of
Fatigue", Trend Engin. Univ. Wash., Vol. 13, No.1, 1961, pp. 9-14.
5. E. K. Walker, "The Effect of the Stress Ratio During Crack Propagation and Fatigue
for 2024-T3 and 7075-T6, The Effect of Environment and Complex Load History on
Fatigue Life Life," ASTM-STP 462, ASTM, pp. 1-15, Philadelphia, 1967.
6. R. G. Fonnan, V. E. Kearney, and R. M. Engle, "Numerical Analysis of Crack Propaga-
tion in Cyclic Load Structures," J. Bas. Engin. Trans. ASME, Ser. D, 89,1967, p. 459.
7. R. G. Fonnan and S. R. Mettu, "Behavior of Surface and Comer Cracks Subjected to
Bending and Tensile Loads in Ti-6AI-4V Alloy," Fracture Mechanics: Twenty Sec-
ond Symposium, Vol. 1, ASTM STP 1131, edited by H. A. Earnest, A. Saxena, and
D. L. McDowell, American Society for Testing and Materials, Philadelphia, 1992,
pp.519-546.
8. J. C. Newman Jr., "A Crack Opening Stress Equation for Fatigue Crack Growth," Int.
J. Fract., Vol. 24, No.3, March 1984, pp. R131-R135.
9. J. K. Donald, and D. W. Schmidth "Computer-Controlled Stress Intensity Gradient
Techniques for High Rate Fatigue Crack Growth Testing," J. Test. Evaluat., Vol. 8,
No.1, January 1980, pp. 19-24.
10. A. Saxena and S. J. Hudak Jr., "Review and Extension of Compliance Infonnation for
Common Crack Growth Specimens," Int. J. Fract., Vol. 14, No.5, October 1978.
11. J. C. Newman Jr., "Stress Analysis of Compact Specimens Including the Effects of
Pin Loading," Fracture Analysis (8th Conference), ASTM STP-560, ASTM, 1974,
pp. 105-121.
12. C. E. Fedderson, "Wide Range Stress Intensity Factor Expression for ASTM Method
E 399 Standard Fracture Toughness Specimens," Int. J. Fract., Vol. 12, June 1976,
pp. 475-476.
13. W. Elber, "Fatigue Crack Propagation," Ph.D. thesis, University of New South Wales,
Australia, 1968.
14. W. Elber, "Fatigue Crack Closure Under Cyclic Tension," Engin. Fract. Mech., Vol.
II, No.1, July 1970.
15. W. Elber, "The Significance of Fatigue Crack Closure," Damage Tolerance in Aircraft
Structure, ASTM STP 486,1971, p. 230.
16. J. R. Rice, "Mechanics of Crack Tip Defonnation and Extension by Fatigue," Fatigue
Crack Propagation, ASTM STP 415,1967, p. 247.
17. J. Schijve, "The Stress Ratio Effect on Fatigue Crack Growth in 2024-T3 Aluminum
Clad and the Relation to Crack Closure," Delft University of Technology Department
of Aerospace Engineeing, Memorandum M-336, August 1979.
18. J. Backlund, A. F. Blom, and C. J. Beevers, "Fatigue Threshold, Fundamentals and
Engineering Applications," in Proceedings of an International Conference held in
Stockholm, June 1-3, 1981.
19. M. Klesnil and P. Lukas, "Effect of Stress Cycle Asymmetry on Fatigue Crack
Growth," Mater. Sci. Engin., Vol. 9,1972, pp. 231-240.
 References 251

20. 1. M. Basom, "Fatigue Behavior of Pressure Vessels Steel," WRC Bulletin 194, Weld-
ing Research Council, New York, May 1964.
21. K. Tanaka, Y. Nakai, and M. Yamashita, "Fatigue Growth Threshold of Small
Cracks," Int. 1. Fract., Vol. 17, No.5, October 1981, pp. 519-533.
22. N. E. Frost, Proceedings of Institution of Mechanical Engineers, Vol. 173, 1959,
pp. 811-835.
23. N. E. Frost, Alternating Stress Required to Propagate Cracks in Copper and Nickel-
Chromium Alloy Steel Plates, 1. Mech. Engin. Sci., Vol. 5, 1963, pp. 15-22.
24. H. Ohuchida, S. Usami, and A. Nishioka, Trans. Japan Soc. Mech. Engin., Vol. 41,
1975, pp. 703-712.
25. S. Usami and S. Shida, Fatigue Engin. Mater. Struct., Vol. I, 1979, pp. 471-482.
26. Small Fatigue Cracks, edited by R. O. Ritchie and 1. Lankford, Proceedings of the Sec-
ond Engineering Foundation International ConferencelWorkshop, Santa Barbara, CA,
January 5-10, 1986." A publication of the Metallurgical Society, Inc.
27. Short Fatigue Crack, edited by K. 1. Miller and E. R. de los Rios, ESIS, Publication 13.
28. The Behavior of Short Fatigue Cracks, edited by K. 1. Miller and E.R. de los Rios,
EGF, Publication 1 (Collection of Papers and References in Crack Initiation).
29. T. C. Landley, "Near Threshold Fatigue Crack Growth: Experimental Methods, Mech-
anism, and Applications," in Subcritical Crack Growth Due to Fatigue, Stress Corro-
sion, and Creep, edited by L. H. Larsson, Elsevier Applied Science, 1985, pp. 167-213.
30. H. Kitagawa and S. Takahashi, "Applicability of Fracture Mechanics to Very Small
Cracks or the Cracks in the Early Stage," in Proceedings of the Second International
Conference on Mechanical Behavior of Materials, Boston, MA, 1976, pp. 627-631.
31. 1. C. Newman Jr. and I. S. Raju, "Prediction of Fatigue Crack Growth Patterns and
Lives in Three-Dimensional Cracked Bodies," presented at the Sixth International
Conference on Fracture, New Dehli, India, December 1984.
32. R. A. Schmidt and P. C. Paris, "Threshold for Fatigue Crack Growth Propagation and
Effects of Load Ratio and Frequency," in Progress in Flaw Growth and Fracture
Toughness Testing, ASTM STP-536, American Society for Testing and Materials,
Philadelphia, pp. 79-94.
33. J. Schijve and D. Broek, "Crack Propagation Based on a Gust Spectrum with Variable
Amplitude Loading," Aircraft Engin., Vol. 34,1962, pp. 314--316.
34. J. Schijve, "The Accumulation of Fatigue Damage in Aircraft Materials and Struc-
tures," AGARDograph No. 157, 1972.
35. 1. Schijve, "Cumulative Damage Problems in Aircraft Structures and Materials," The
Aeronaut. 1., Vol. 74,1970, pp. 517-532.
36. O. E. Wheeler, "Spectrum Loading and Crack Growth," J. Basic Engin., Vol. 94D,
1972, pp. 181-184.
37. 1. D. Willenborg, R. M. Eagle, and H. A. Wood, "A Crack Growth Retardation Model
Using an Effective Stress Concept," AFFDL-TM-71-1 FBR, 1971.
38. 1. P. Gallagher and T. F. Hughes, "Influence of the Yield Strength on Overload
Affected Fatigue-Crack-Growth Behavior of 4340 Steel," AFFDL-TR-74-27, 1974.
252 Chap. 4 Fatigue Crack Growth

39. S. M. Arnold, "Effect of Screw Threads on Fatigue," Mech. Engin., Vol. 65, July,
1943, pp. 497-505.
40. A. Schwartz Jr., "New Thread Form Reduces Bolt Breakage," Steel, Vol. 127, Sep-
tember,4, 1950, pp. 86-94.
41. A. M. Smith, "Screw Threads, the Effect of Method of Manufacture on the Fatigue
Strength," Iron Age, Vol. 146, August 22,1940, pp. 23-28.
42. J. O. Almen, "On Strength of Highly Stressed, Dynamically Loaded Bolts and Studs,"
SAE J., Vol. 52, No.4, April, 1944, pp. 151-158.
43. T. R. Higgins, "Bolted Joints Found Better Under Fatigue," Engin. News-Record, Vol.
147, August 2,1951, pp. 35-36.
44. "Preloaded Bolts and Screws," Lockhead Aircraft Corporation, Report No. 2072,
March 1, 1950, pp. 1-7.
45. E. F. Bruhn, "Analysis and Design of Flight Vehicle Structure," Jacobs Publishing,
pp. Dl-Dl4.
46. "Tension Type Fittings," Lockhead Report No. 2072, Lockhead California Company,
November 1, 1968, pp. 1-12.
Chapter 5

Fracture Control Program and

Nondestructive Inspection

5.1 Introduction

Full assurance of maintaining a trouble-free damage tolerant space vehicle during

its operational life requires a continuing, multidisciplinary process which begins
in the preliminary design phase and extends through manufacturing and into the
operational planning of the space structure. Figure 5.1 illustrates the main ele-
ments of this process. The purpose of this section is to provide guidance for
implementing a general approach, defined in a Fracture Control Plan (FCP), to
control and prevent damage that could cause a catastrophic failure due to the pre-
existing cracks in the structure. The general concepts used in this chapter are also
used in other industries, such as nuclear, pressure vessels, and shipbuilding, where
safety is of great concern.
The design philosophy, material selection, analysis approach, testing, quality
control, inspection, and manufacturing are elements of the fracture control pro-
gram that contribute to building and maintaining a trouble-free space structure.
The first part of this chapter (Sections 5.1 and 5.2) discusses the elements that are
essential in implementing an effective fracture control plan. In the latter part of this
chapter (Section 5.3), different nondestructive inspection techniques that are cur-
rently used for inspecting high risk critical space hardware are discussed. Section
5.1.1 briefly describes the two types of design concepts (namely slow crack growth
and fail-safe) presently used in the design of aerospace and aircraft structures.

5.1.1 Design Philosophy

To provide adequate tolerance and safety, a space structure must be designed to
withstand the environmental load throughout its service life, even when the struc-

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts 253
© Chapman & Hall 1997
254 Chap. 5 Fracture Control Program and Nondestructive Inspection

Design Category
* Safe Ute (single load path) Load'" Corrosive In Depth Analysis of:
* Fail Safe/Low Risk! Environment * Static/Dynamic Loads
Contained * Strength
C1l1l:k I!llint~ * Fabrication * Thermal Stress
Stress Concentration Factor * Test * Fatigue
Process Effects * Storage * Crack Growth
Tolerance Build-lJp * Serllice * Residual Strength
Stress Corrosion
Residual Stress

",

,
Quality Control
* Non-Destructive Inspection) " Fracture Process Control
* Traceability ... Control ~ * FactOl}'/Logistics
* Identification Program * Subcontractor
* Special Handling ¥' t
Material'" Process Testing
~ Operation and DaigD DmlllllllDlIIDl
Fracture Toughness Maintenance * Material Properties
Stress Corrosion * Accessibility Development
Fabrication '" Joining *Inspection Interval * Safe Ufe
Process Auids * Corrosion Protection * Qualification
Temperature * Acceptance
Environment

Figure 5.1 Main elements of a fracture control program

ture has preexisting flaws or when part of the structure has already failed. To pro-
vide adequate safety and to ensure that unstable crack growth is not reached, two
types of damage tolerant design concepts are presented in this section: single load
path (slow crack growth) and fail-safe design (also called multiple load path
design). Figures 5.2 and 5.3 illustrate the slow crack growth and fail-safe design
concepts, respectively.

5.1.1.1 Slow Crack Growth Design

The slow crack growth design concept simply states that single load path struc-
tures must be designed such that any preexisting cracks will grow in a slow and
stable manner during their service life, and thus the growing crack will not
achieve a critical size that could cause unstable propagation. In this design phi-
losophy, the damage tolerance is ensured by maintaining a slow crack growth rate
during the life of the structure. Moreover, the residual strength capability of the
structural part must always be kept above the applied stress. The keel structure
shown in Fig. 5.2 is used to attach a segment of a space structure to the shuttle
during the flight. The pin is made of 455 custom stainless steel and was classified
as a single load path structure. The structural integrity requirements of this part
are to show that the pin has sufficient margin of safety based on static analysis, as
well as a minimum of four service lives using a fracture mechanics approach.
 5.1 Introduction 255

Keel Pin ~

~ Keel Pin Structure

Payload Bay (This is a single load path structure,
L.._ _.;..._......::......_-'---.Io.IJI=- Ie. the failure of this part will
Shuttle create catastrophic hazard
during shuttle launch)

Figure 5.2 Illustration of a single load path structure. The payload is attached to the
space shuttle payload bay by a keel pin structure that has a single load path design

Space Primary Structure.

V
Umbilical Support Structure
Integrated with
Primary Structure

Payload Bay
National Space Transporter

Figure 5.3a Illustration of a fail-safe design structure. The umbilical support struc-
ture carries avionics and fluid lines for space components.
256 Chap. 5 Fracture Control Program and Nondestructive Inspection

Each bolted joint has a fail-safe feature (failure of

one bolt causes the redstributlon of the load to the
Fail-Safe Locations rema/nlng bolts), see figure S.3a.

Boned Joint

Boned Joint

Figure S.3b Fail-safe design of the umbilical structure and upper and lower bolted
joints

5.1.1.2 Fail-Safe Design and Low Released Mass

In the fail-safe design philosophy, the structure is designed such that the failure of
a structural component due to crack propagation is localized and safely contained
or arrested. The damage tolerance in the fail-safe design category must be assured
by allowing the partial structural failure, while the remaining undamaged struc-
tural components maintain their integrity and are able to operate safely under the
remaining distributed load. In addition, it must be shown that all released frag-
ments due to partial failure have a total mass of less than 0.25 lb (called low
released mass) during the launch or else they are contained and will not pose a
catastrophic hazard to the surrounding structural parts or to the shuttle. A fail-safe
structure has a multiple load path feature (see Figs. 5.3a and b) in its design. Fig-
ure 5.3a illustrates a fail-safe designed space structural component called the
umbilical support structure (situated in one of the sections of the main structure).
It functions as a support structure to the avionics lines, which carry power, data,
optical signals, instrumental wiring, and the fluid lines that carry ammonia (a
coolant) to different regions of the space structure (avionics and fluid lines are not
shown here). The failure of one beam (for example, one of the vertical beams
shown in Fig. 5.3b) produces load redistribution to other members of the struc-
ture. By analysis, it can be shown that the remaining redistributed load will not
cause damage to the umbilical structure or to the surrounding hardware during the
launch and on-orbit environment.
 5.1 Introduction 257

Another example that can be useful to the reader for understanding the fail-
safe design philosophy is the upper and lower joints that connect the umbilical
support structure to the primary structure (8 bolted joints on the top and two on
the bottom). Failure of anyone of the bolted joints will redistribute the load to the
remaining joints. In addition, each bolted joint has a fail-safe feature by having the
four-bolt pattern shown in Fig. 5.3b. It can be shown by analysis that the failure of
one bolt can cause the redistribution of the load in the joint and will prevent cata-
strophic damage to the main structure.
The residual strength capability of a fail-safe structure must be evaluated to
ensure the structural integrity of the part. In other words, the redistributed load
must fall below the residual strength capability of the structural part. To assess
the residual strength of a fail-safe structure (1) the stress level at which partial
damage occurs must be known and (2) the load-carrying capability of the remain-
ing structure that must withstand the redistributed load must be evaluated.

5.1.1.3 Material Selection, Testing, Manufacture, and Inspection

Material selection is a critical element of the fracture control process. Tradeoff
studies in the early stage of design are conducted between materials and the use of
comparative property data is necessary in the selection process. Ultimate strength,
yield strength, plane strain, and plane stress fracture toughness (including the frac-
ture toughness for surface cracks) and stress corrosion property data must be con-
sidered in the material selection process for the expected environment. The crack
growth rate as a function of stress intensity factor (da/ dN, I:1K data) is also
required. If the environment is corrosive, the crack growth rate data (da/ dt versus
1:1K) should be available to the analyst. Mechanical and fracture properties with
respect to the anisotropic nature of certain materials must also be considered. A
comprehensive list of materials that are currently recommended by NASA for
space application (which may also be utilized for other industries) together with
their static and fracture properties is available in Appendix A. In this appendix,
the ultimate and yield strength as well as plane strain, K 1c ' and plane stress, K c '
part through fracture toughness, K, e' and crack growth properties for the listed
material are available. Table 5.1 shows an example of static and fracture proper-
ties of 7075-T851 aluminum alloy that must be available for conducting a mean-
ingful fracture analysis.
Testing methods for each critical part and assembly (an assembly is a piece of
hardware that is made of several parts) are also developed and incorporated as
part of the fracture control process.
Manufacturing processes must be selected for the high risk or critical parts
such that they do not reduce the damage tolerance level required by the design.
Gouges and surface scratches due to improper handling and machining must be
avoided throughout the manufacturing process since their presence (if undetected)
can reduce the life of the part. Control of processes and selection of inspection
258 Chap. 5 Fracture Control Program and Nondestructive Inspection

Table 5.1 Fracture properties of 7075-T85I aluminum alloy (NASA/FLAGRO

material library)

Yield Vlt.
Stress Stress K,c K'e A B C n p q llKo Rei a SR

..
76 85 26 38 1.0 1.0 .2E-7 2.88 0.5 1.0 2.0 0.7 1.9 0.3

... • ...Fracture Adjustement

... • ... •
Strength Fatigue Crack Growth Properties
Properties Toughness to Plane
Properties stress

procedures to maintain process quality are prime considerations of the fracture

control plan.
This introduction to the fracture control plan illustrates the interconnection
between design, testing, manufacture, and inspection and their roles in maintain-
ing the desired damage tolerant structure and to reduce the incidence of fracture-
related failures and loss.
The structural integrity of a hardware that contains several parts (such as an
aircraft, a space structure, or a pressure vessel) can be maintained by ensuring the
implementation of an effective fracture control plan. The implementation of a
sound fracture control plan for space structures is required by the regulating
agency (i.e., NASA). Throughout this chapter, the emphasis will be given to the
fracture control policies and requirements for space structures. Section 5.2
describes the detailed fracture control policies and requirements that must be
implemented for any space structure that will be used as a payload on the space
shuttle. The fracture control plan described in Section 5.2 can be similarly applied
to other structures where safety is a great concern, such as nuclear, military, or
commercial aircraft. Prior to implementation of the fracture control program, a
fracture control plan must be submitted to the customer for review. Implementa-
tion of the plan and its related activities are monitored by representatives of the
procuring agency after its approval.

5.2 Fracture Control Plan

5.2.1 Purpose
The purpose of implementing a fracture control plan is to prevent unstable crack
growth in structures of all elements and systems of flight vehicle structure during
the design service life. Catastrophic failures as the result of unstable crack growth
can cause the total loss of the structural hardware, and more importantly, the loss
of human life. The required fracture control procedures are intended as an aug-
 5.2 Fracture Control Plan 259

mentation to, and are based heavily upon, the good engineering and manufactur-
ing practices already embedded in the hardware development process. Fracture
control imposes additional engineering and product assurance requirements
needed to ensure the structural integrity of fracture critical structures throughout
all phases of the component's lifetime.

5.2.2 Causes of Unstable Crack Growth

The basic assumptions that underlie the need for fracture control are based on the
knowledge that all structures contain preexisting flaws that could be located in the
most unfavorable location and orientation with respect to the applied stresses.
Furthermore, it is essential that these cracks do not grow undetected to a critical
length during the usage period.
Under sustained tensile stress in the presence of an aggressive environment,
some materials exhibit a tendency to initiate cracks. The circumstances under
which this stress corrosive cracking phenomenon occurs are sufficiently well doc-
umented to provide appropriate avoidance measures and criteria (MSFC-HDBK,
Material Selection Guide for MSFC SpaceLab Payloads criteria or MSFC-SPC-
522B called "Design Criteria for Controlling Stress Corrosion Cracking"). In non-
corrosive environments, cracks will initiate owing to a sufficient number of high
tensile stress cycles. However, since the assumption of preexisting cracks in all
structural elements is more severe than the assumption of crack initiation (see
Chapter 2 on crack initiation concept), the requirements for preventing failure are
completely enveloped and satisfied by the Nondestructive Inspection (ND!)
requirements (see Section 5.3 for a comprehensive review of NDI methods and
their application to aerospace hardware). The reader should understand that the
actual initial crack size in the material could be much smaller than the crack size
assumption based upon the NDI method. Therefore, the analytical life evaluation
of a part using the standard NDI initial flaw size yields conservative results.
The fracture control plan specifies the type of NDI and minimum initial crack
sizes (a,) that may be used for single load path versus fail-safe design. The initial
crack size associated with any given NDI method can be reduced to extend the
analytical life of the component if the contractor can demonstrate that the reduced
initial flaw size is reliably detectable (see Section 5.2.8).

5.2.3 Fracture Control in the Stages of Hardware

Development
Fracture control for flight hardware is implemented starting at conceptual design
and continues through all phases of engineering, manufacture, and in-service
usage. Therefore, the implementation of the program should be organized accord-
ing to the logical sequence of a given structural component's life history. In the
engineering phase, the part is designed, analyzed, and qualification tested. At this
260 Chap. 5 Fracture Control Program and Nondestructive Inspection

stage, the structural integrity and safe-life analysis are fully assessed. The end
products of the engineering phase include the detailed production drawings used
by manufacturing to produce the part.
The production phase occurs when the part is produced according to the pro-
duction drawing, inspected, and/or acceptance tested. Satisfaction of NDI, proof
testing, traceability, and special handling requirements are essential elements of
this phase. During the operational phase, the part may be subjected to interval
inspection, maintenance, and possibly repaired or replaced.

5.2.4 Typical Fracture Control Activities

All activities that influence the control of structural crack initiation and propaga-
tion in flight hardware elements and systems and their test articles are subject to
the requirements and procedures defined in a Fracture Control Plan. These activ-
ities generally fall into the following categories:

1. Structural design
2. Structural analysis
3. Materials selection, procurement, and storage
4. Fabrication process control
5. Quality assurance and nondestructive inspection (ND!)
6. Test
7. In-service operations, maintenance, inspection, and repair

Examples of typical fracture control activities performed by most hardware devel-

opers include:

1. Minimization of flaw sizes through proper inspection, processing, and fabrica-

tion procedures
2. Selection of materials having adequate fracture toughness, resistance to stress cor-
rosion cracking (SCC), and crack growth rate properties (da/ dN, !:l.K and possibly
da/ dt, !:l.K data)
3. Minimization of residual stresses during processing and fabrication
4. Complete identification and documentation of design loads and environments
5. Design for sufficiently benign internal stresses, including minimization of stress
concentrations
6. Use of proper analytical tools and tests to verify that the design fulfills fracture con-
trol requirements
7. If, owing to conflicting requirements, all fracture control requirements cannot be
completely met, additional controls are placed on the design.
 5.2 Fracture Control Plan 261

5.2.5 Data Required for Fracture Control

In support of fracture control activities, considerable data are developed and
maintained by the contractor. Engineering data necessary for adequate assessment
and classification of hardware include the following:

1. Definition of environments and load spectrum history. The environment could be

inertia, thermal, pressure, sonic, corrosive, or a combination of several environments.
2. Detailed design and assembly drawings. All critical dimensions and characteristic
must be denoted on the drawing.
3. Mechanical and fracture properties of materials in the appropriate environments
4. Stress and fracture analysis results
5. Level of NDI performed and the corresponding flaw size for through and
part through cracks.

A large portion of these data is contractually required and generated for purposes
other than fracture control (e.g., objective evidence of design and qualification
documentation).

5.2.6 Contents of a Fracture Control Plan

Each major element of a fracture critical hardware system is governed by a Frac-
ture Control Plan prepared by the contractor or system developer and approved by
the responsible procuring agency. The Fracture Control Plan defines the elements
of the Fracture Control Program and the responsibilities for managing and accom-
plishing them. The Plan must describe in detail the approach to be used to imple-
ment the following:

1. Fracture control classification of components

2. Analysis and/or testing to determine fracture control acceptability of hardware
3. Control of materials, manufacturing processes, nondestructive inspections, testing,
design changes, and part handling transportation
4. Overall review and assessment of fracture control activities and results

Implementation approaches for the above four major elements of the fracture con-
trol program are described in Sections 5.2.7 through 5.2.10.

5.2.7 Fracture Control Classification of Components

Fracture control classification of all hardware components is typically determined
as shown in the logic diagram of Fig. 5.4. Components that are classified as non-
hazardous released parts, contained, fail-safe, or low risk must meet the specific
requirements associated with at least one of these categories. These components
262 Chap. 5 Fracture Control Program and Nondestructive Inspection

Does component meet low rei

mass requirements of Section 5.1.1.2?
, No
Engineered and manufactured using
Does component meet released mass conventional aerospace
containment requirements of Section 5.1.1.2? uirements and standards
, No

Subject to fracture critical or high

risk requirements

Figure 5.4 Evaluation of structural components for fracture control classification

are then considered nonfracture critical and are not required to be further evalu-
ated for fracture control. However, a life expectancy analysis of nonfracture crit-
ical parts, either through a conventional fatigue (stress to life approach using the
S-N curve data) or a damage tolerance approach that assumes a smaller initial
flaw size, must be performed. Components classified as fracture critical or high-
risk parts will have their damage tolerance and/or safe-life (one safe-life is equal
to four service lives) verified by test and/or analysis, in addition to meeting the
standard design, control, and verification requirements for aerospace structures.
Note that for other structures, such as aircraft, one safe-life may be defined as two
service lives. As fracture control classification and analysis of parts are per-
formed, an up-to-date listing is maintained of all fracture critical parts.
 5.2 Fracture Control Plan 263

5.2.8 Analysis and/or Testing to Determine Fracture

Control Acceptability of Hardware
A fracture critical component is acceptable if it can be shown by analysis or test
that the largest undetected flaw that could exist in the component will not grow to
failure when subjected to the cyclic and sustained loads encountered in four ser-
vice lifetimes. One service lifetime includes all significant loading cycles occur-
ring after flaw screening (to establish minimum initial flaw size) and will include
testing, transportation, flight, operations, or any other loading cycles considered
relevant. Figure 5.5 illustrates a simplified version of a complicated load spec-
trum that an aircraft structural part can be exposed to. The number of cycles asso-
ciated with each event must be considered in evaluating the life of the part.

(2) (' )
Fracture Critical (High Risk)

\~;Handling Cycles
Transportation Cycles

O~t

Landing Cycles Launch and Flight Cycles

Figure 5.5 Illustrating a simplified version of an aircraft load environments. Load

cycles for each environment contributes to accumulation of failure on aircraft struc-
tural parts
264 Chap. 5 Fracture Control Program and Nondestructive Inspection

Either of two analysis approaches may be used to show that a part subjected to
NDI meets safe-life requirements. The first or direct approach is to select the
appropriate NDI technique and to use the flaw sizes representative of the capabil-
ity of that technique to show that the part will survive at least four lifetimes. The
second approach is to calculate the critical (i.e., maximum) initial crack size for
which the part can survive the required lifetimes and to verify by inspection that
there are no cracks greater than or equal to this size. In the second approach, the
initial crack length is estimated and by analysis the final crack length, ai' corre-
sponding to four service life is calculated. Figure 5.6 illustrates the two above
approaches for a beam subjected to a fluctuating load environment.
In addition to the safe-life analysis or testing, all fracture critical parts must be
subjected to NDI inspection to screen flaws. The selection of NDI methods and
level of inspection is based primarily on the safe-life acceptance requirements of
the part. The initial crack sizes correspond to a 90% probability with a 95% con-
fidence level of inspection reliability. Minimum detectable initial crack sizes for
specific NDI methods are given in Table 5.1 for the crack geometries shown in
Figure 5.16. These are the minimum sizes that must be used in the safe-life analy-
sis. Use of initial crack sizes for other geometries or NDI techniques requires the
approval of the procuring agency. Where adequate NDI offinished parts cannot

QMoment, M

Part To Be Analyzed .7 t.K=I3t.cr(na)o.s

da/dN ~ f("'K)
(see chapters 3 & 4 for more description of these equations)

First Approach Second Approach

Select a, from appropriate NDI method and Evaluate the critical crack length, ac,
compare the calculated K (after four service for four service life and compare it's
life) with the critical K value (Kc). value with the initial crack length, a.
(a. <ar ).

Figure 5.6 lIIustrating the two approaches used for safe-life analysis
 5.2 Fracture Control Plan 265

be accomplished, NDI may be required on the raw material and/or on the part
itself at the most suitable step of fabrication.
Pressure systems (pressure vessels), rotating machinery, and fasteners that are
classified high risk or as fracture critical have additional analysis and/or test
requirements imposed upon them. For example, all pressure systems (such as a
pressure vessel with stored energy of 14,240 ft-Ib or greater) and high-energy
rotating machinery components (a rotating mechanical assembly that has a kinetic
energy of more than 14,240 ft-Ib based on KEnergy = I (j}12 where the two quanti-
ties I and ware the mass moment of inertia and w is the rotational frequency in
radians per second respectively) require qualification proof testing. The purpose
of this proof test is to verify the integrity of the structural part. The stress associ-
ated with proof testing for pressure vessels and rotating machinery is selected as
1.5 x limit load (see Figure 5.7).

5.2.9 Traceability and Design Annotation for Drawings

Traceability is defined as the ability to relate designated product data to specific
products by retrieving information logged through each level of hardware records
to the configuration item level. For example, the material mill heat number should
be traceable from the part at any stage of manufacture and assembly. In addition,
one should also be able to trace the NDI performed on the hardware and other
critical characteristic records of fracture critical hardware at any stage of its devel-
opment. Traceability is required to be maintained on all fracture critical parts

Proof Test of 1.SxLimit load is Required

Rotating Machlnary
Oassified as
Fracture Critical

K~,,~
_ ..............
Pressurized Tank
Oassified as
Fracture Critical

Stored Energy of
14240 ft-lb or

Rotating Machlnary
with Kinetic Energy
Greater than 14240 It-lb.

Figure 5.7 High energy (high risk) systems. A pressurized tank and A rotating wheel
266 Chap. 5 Fracture Control Program and Nondestructive Inspection

throughout their development, manufacture, testing, and flight. Specific proce-

dures for traceability are established at the discretion of the procuring agency, and
may include serialization or other means of configuration control. For manned-
rated flight hardware, serialization is typically required only for pressure vessels
and components of high-energy rotating machinery. All other fracture critical
hardware must maintain traceability to the raw material heat treat lot as a mini-
mum. For each pressure vessel, a log is maintained to record all pressure cycles
and associated environmental conditions occurring during the time period from
fabrication to the end of the service life of the vessel.
Engineering drawings and equipment specifications for fracture critical parts
contain notes that identify the part as fracture critical and specify the appropriate
inspection Dr flaw screening method to be used on the part. Critical characteristics
are also noted on the drawings to identify mandatory inspection points. Some
examples of critical characteristics are critical dimensions, traceability, proof test,
and NDI results. An example of fracture control program notes on the fracture
critical part drawings (designated by the symbol [FCP]) are shown in Fig. 5.8.
[FCP] note number I indicates that this component is a high risk fracture crit-
ical part and must meet fracture critical requirements per XXX document. [FCP]
note number 2 is related to the material serialization. [FCP] note number 3 is
related to NDI penetrant inspection of the part after machining per XXX docu-
ment. Finally, [FCP] note number 4 describes the identification method using
electrochemical etch masking. Here the XXX refers to special documents that
contain the procedures to be used to satisfy fracture control requirements.
Changes in design or process specifications, manufacturing discrepancies,
repairs, and finished part modifications for all fracture critical parts are reviewed
according to criteria established by the responsible fracture control authority to
make certain that the parts still meet fracture control requirements. Changes to the
Fracture Control Plan must be incorporated into a revised Plan, which is then
resubmitted to the procuring agency for approval.

[FCP) 1: THIS PART HAS BEEN IDENTIFIED AS A FRACTURE CRITICAL PART. SHOULD
MEET FRACTURE CRITICAL HARDWARE REQUIREMENTS PER XXX DOCUMENT.

[FCP) 2: SERIALIZE PER YYY DOCUMENT. EACH ITEM SHALL HAVE A DIFFERENT
SERIAL NUMBER. GAPS IN SERIAL SEQUENCE ARE PERMISSIBLE.

[FCP) 3: PENETRANT INSPECT PER XXX DUCUMENT. ETCH SURFACES PRIOR TO

PENETRANT INSPECTION.

[FCP) 4: IDENTIFY USING ELECTROCHEMICAL ETCH METHOD PER XXX DOCUMENT.

DIRECT MARKING LOCATION APPROXIMATELY IN AREA SHOWN ON THE DRAWING

Figure 5.8 Related fracture control notes on drawings for high. risk or fracture criti·
cal parts
 5.2 Fracture Control Plan 267

Special handling requirements are applied to fracture critical hardware to avoid

material damage that could act as a crack initiator. Fracture critical parts and their
assemblies must be protected at all times during the manufacturing process and
installation operations from damage caused by environmental deterioration or
rough handling. The procedures include requirements to ensure individual han-
dling in protective packaging labeled with "FRACTURE CRITICAL-HANDLE
WITH CARE." The XXX document mentioned in the drawing notes (see Fig.
5.5) contains program derived requirements for traceability, NDE inspection, and
the special handling of fracture critical parts.

5.2.10 Overall Review and Assessment of Fracture

Control Activities and Results
Each contractor or hardware system developer designates a specific individual or
group to be the responsible fracture control authority. This authority is responsible
for direction and implementation of its Fracture Control Program and for assuring
its effectiveness. The designee is responsible for monitoring, reviewing, and
approving fracture control activities performed both internally and by subcon-
tractors. These activities are coordinated with other key organizations, including
engineering, manufacturing, safety, reliability, and quality assurance as required.
The total Fracture Control Program is subject to oversight and review by the
procuring agency.
To certify compliance with fracture control requirements, the prime contractor
or subcontractor responsible for the system development prepares a fracture con-
trol summary report on the total system for review and approval at a Design Cer-
tification Review by the procuring agency. The documents required to support the
acceptability of a fracture critical part include a crack-growth analysis (or safe-life
test) report and a Nondestructive Inspection (NDI) or proof test report. A docu-
mented description of the load spectrum and material crack growth properties
used in the analysis is included in the safe-life analysis report. The NDI report
includes the date of inspection, serial number or identification of the part
inspected, and name of the inspector. If special NDI was used (to obtain a smaller
flaw size), additional data are required in the inspection report to assure accept-
ability and traceability of the process. Documents or analysis supporting the frac-
ture control summary report and the fracture control classification of components
are normally kept for the life of the system and made available for audit.

5.2.11 Fracture Control Board

The purpose of the Fracture Control Board (FCB) is to direct and oversee the
Fracture Control Program. The FCB is responsible for:
268 Chap. 5 Fracture Control Program and Nondestructive Inspection

1. Ensuring preparation, maintenance, review, and approval of the FCP

2. Monitoring, coordinating, reviewing, and approving fracture control activities per-
formed both internally and by subcontractors and vendors
3. Ensuring effective implementation of the fracture control program through activities
in all affected organizations, and for maintaining awareness of fracture control
progress or problems arising within these organizations.
4. Ensuring that each subcontractor engaged in the design and manufacture of fracture
critical hardware establishes a Fracture Control organization that assures confor-
mance with the requirements of the FCP. The typical FCB organization consists of
a permanently assigned member from each of the following disciplines:
Program Management (Chairman)
Strength Analysis
Materials and Processes (M&P)
Quality Assurance (QA)
Safety
Structures Design
Mechanical Systems Design
Flight Crew Systems Design
Fluid Systems Design
Subcontract Technical Manager (STM) for subcontractor hardware

At FCB meetings concerning a particular system or element, the FCB Chairman

and the assigned members representing Strength Analysis, M&P, QA, Safety, and
the cognizant design authority or their delegates will be present. Figure 5.9 shows
the organization of a typical FCB.

Strength Analysis Materials and PTocesses Quality Assurance Safety

Fracture Control Board

Chairman

Structures Design Mechanical Systems Flight Crew Systems Fluids Systems

Figure 5.9 Organization of a typical Fracture Control Board

 5.3 Nondestructive Inspection Techniques 269

5.3 Nondestructive Inspection Techniques

5.3.1 Introduction
Nondestructive Inspection (NDI) can be defined as the use of nonintrusive meth-
ods to ascertain the integrity of a material or structure. Many nonintrusive meth-
ods have been developed to evaluate materials for property determination, verify
quality of workmanship, and evaluate a component for the existence of flaws. A
flaw, in this sense, can be considered as any nonconformity that exceeds an estab-
lished size criteria.
Flaw detection is by far the most important aspect of NDI in regards to safe-life
assessment of fracture critical parts. Fracture mechanics analysis assumes the
existence of a maximum flaw size in the part that grows in a stable manner during
its service life. NDI provides the assurance that a flaw larger than the identified
maximum size does not exist in the part. From a safety perspective, the initial
assumed crack lengths provided by the NDI methods are longer than any preex-
isting flaw that could be present in the structure after inspection. However, the
degree of conservatism as a result of the longer initial crack size assumption (used
to evaluate the life) must be realistic enough not to impact the structural weight or
cause unnecessary rejection of parts.
There are numerous NDI methods utilized for flaw detection in structural com-
ponents. Although many specialized methods are developed for specific materials
and configuration, most techniques are variations on several general methods that
use visual enhancement of defects or measure some form of energy transmission
through materials and its interaction with defects. The most prevalent of these
NDI techniques commonly used in the detection of flaws in aerospace compo-
nents are: liquid penetrant, magnetic particle, eddy current, ultrasonic, and radi-
ography. The purpose of this section is to briefly describe and compare the
common NDI methods and to discuss the variables affecting them.

5.3.2 Liquid Penetrant Inspection

Liquid penetrant inspection is the most widely applied NDI method for the detec-
tion of all types of surface flaws, porosity, shrinkage area, and similar types of
discontinuities in metals. Essentially an enhanced visual evaluation, penetrant
inspection relies on capillary action to penetrate into flaws open to the surface.
Excess dye is removed and a white powder developer is applied to the surface to
act as a blotter to the trapped dye, thus enhancing flaw detection sensitivity. Flu-
orescent dyes are most commonly used in penetrant fluids, providing high sensi-
tivity for critical part inspections. A schematic depicting the mechanism of crack
detection by the penetrant process is shown in Fig. 5.10 [1]. Penetrant inspection
is fast, portable, and easy to interpret, but does require that the flaw be open to the
surface for detection. For metal components subjected to mechanical operations
270 Chap. 5 Fracture Control Program and Nondestructive Inspection

(a)
" Crack
Ceaned Surfaces Penetrant Covered
(b)

Excess Penetrant Developer Applied

Removed (d)
(c)

Figure 5.10 Mechanism of penetrant flaw detection

that may mask existing flaws, such as grinding or machining, etching may be
required to remove the smeared metal prior to inspection. Variables affecting the
penetrant inspection are discussed in Section 5.3.9.1.

5.3.3 Magnetic Particle Inspection

Magnetic particle inspection is another NDI method which uses a fluorescent
media to reveal flaws. This approach is applicable only to ferromagnetic materials,
such as plain carbon or alloy steels and ferritic stainless steels. A magnetic field is
generated on the part surface by flowing electric current through the part or adja-
cent to it. Surface and near surface discontinuities cause variations in the magnetic
field. The fluorescent magnetic oxide particles are then flowed over the part sur-
face and are attracted to the magnetic field variations, where they can be detected
under ultraviolet light. Figure 5.11 illustrates two types of magnetic particle
inspection setups. Figure 5.8a shows an inspection of a tube segment for longitu-
dinal cracks, while Figure 5.8b depicts a weld inspection utilizing contact prods
[2]. The magnetic particle inspection method is quick and simple in principle,
however, several different magnetic field orientations and magnetization strengths
may be required to thoroughly inspect the suspected flaw orientations. Demagne-
tization of the part is also required after the inspection is completed. Variables
affecting the magnetic particle inspection are discussed in Section 5.3.9.2.

5.3.4 Eddy Current Inspection

Another nondestructive technique used to detect surface and near surface flaws is
eddy current inspection. Eddy current is an electromagnetic technique that utilizes
 5.3 Nondestructive Inspection Techniques 271

Ibl

Figure 5.11 Magnetic particle circular magnetization setups

small diameter coils to induce electric currents in conductive materials. The eddy
currents induced in the material, in tum, generate their own magnetic field. The
magnitudes, time lags, phase angles, and flow patterns of the eddy currents within
the test material are detected as changes in the electrical characteristics of the
inducing coil. The presence of defects causes significant disruptions in the eddy
current flow, and results in impedance and voltage variations. Calibration is per-
formed on reference defects of known size.
Eddy current inspection methods have been used extensively in the aircraft
industry for in-service inspections of components subjected to a fatigue environ-
ment. The high sensitivity of eddy current methods for crack detection provides a
notable advantage over other NDI methods for this application. Conventional
eddy current equipment designed for crack detection utilizes a CRT screen that
displays the phase and amplitude of the received signal on an impedance plane
[Inductive Reactance (X) versus Resistance (R)]. A typical signal response from
a crack in aluminum is shown in Figure 5 .12a. Figure 5 .12b illustrates the change
in signal response at various gain settings [3]. Variables affecting the Eddy cur-
rent inspection are discussed in Section 5.3.9.3.

5.3.5 Ultrasonic Inspection

Ultrasonic evaluation methods are most prominently used to detect subsurface
flaws in components. This capability permits ultrasonic evaluation of material in
its raw stock form, thereby screening material containing rejectable flaws before
a large investment in material processing is made. Ultrasonic evaluations require
272 Chap. 5 Fracture Control Program and Nondestructive Inspection

x x (b)
(a) Vertical
Volts/Division
0.2

0.5

1.0

Aluminum 2.0
R Air 5.0
50KHz
Vertical=2.0 Volt/Division SO kHz Aluminum
Horizental=2.0 Volt/Division Horizontal-2.0 Volts/Division

Figure 5.12 Eddy current CRT response from a crack in aluminum (a) Normal
impedance plane response, (b) response with phase rotated to horizontal and at vari-
ous gain settings

generating high-frequency (1 MHz to 50 MHz) acoustic waves with a vibrating

crystal transducer and then introducing them into a component. A liquid or gel
couplant is used to transmit the ultrasonic waves between the transducer and the
material. As the ultrasonic wave encounters an interface, such as at the cou-
plant/test part surface, part of the wave is reflected back to the transducer while
part is transmitted into the part. The amplitude of the signal reflected from and
transmitted by the liquid/solid interface is dependent upon the ratio of the acoustic
impedances of the part and the coupling media. This reflectance and transmittal of
the ultrasonic wave occurs at every subsequent interface (e.g., defects as well as
the front and back surface of the test part). The amplitude and position in time of
the reflected or transmitted signal are monitored producing a marked signal
change when encountering material flaws. The signal can be represented as an
analog trace on an oscilloscope (A-scan), through-thickness depth position
(B-scan), or a plane view (C-scan). Ultrasonic signal displays for each of these
data representations are shown in Figs 5.13 a, b, and c [4]. Variables affecting the
ultrasonic inspection are discussed in Section 5.3.9.5.

5.3.6 Radiographic Inspection

Radiography, or x-ray imaging, is another NDI method employed in the detection
of subsurface flaws. This method relies on the differential absorption of x-ray
photons as they propagate through a material. In conventional film radiography,
x-rays are generated via electrons striking a tungsten target and exit the x-ray tube
 5.3 Nondestructive Inspection Techniques 273

rl M''''C'''.~ OS( lIos ~PE Sc ~en

I Power Supply ~ 1
(Ootionail
Fla\ E~ ~we pT lice

J ~I IClock
"\
.'i
V I
I Pulser Circuit I J Initial Pulse _I-' ~
I I
rl Sweep Circuit I
I

Search r j- i1.J h. h ri (r-,

Unit T
\..r~ ~Receiver.An1>lnie1
> Circuit
J I

I
Range Marker
First Back

~lF ~ - Sound Beam

Echo Depth Rellection

L- Echo intensity
Raw

Figure 5.13a A-Scan display

Test

Oscilloscope Screen

Figure S.13b 8-Sean display

Figure 5.13e C-Sean display

274 Chap. 5 Fracture Control Program and Nondestructive Inspection

in a conical beam. The photons that penetrate the material are imaged on the
underlying photographic film. Areas of greater thickness or density will absorb
more of the penetrating photons. If the difference in density between a flaw and
the surrounding material is significant this difference will appear on the x-ray film
as a difference in film density. A typical film setup is shown in Fig. 5.14 [5].
X-ray systems are usually sensitive to changes that result in an apparent change of
at least 1% to 2% of the material thickness or density. This limits the flaw sensi-
tivity to defects oriented parallel to the x-ray beam such as cracks, voids, and
inclusions. X-ray inspections are of particular value to evaluations of welds and
castings where defect type and orientation is favorable. Variables affecting the
radiographic inspection are discussed in Section 5.3.9.5.

5.3.7 Flaw Size Verification

Although great efforts are expended to produce critical components free from
defects, material processing operations have the potential for creating flaws. Cast-
ings are susceptible to voids, cold shuts, differential cooling cracks, and inclu-
sions. Mill operations such as rolling of bar, plate, sheet, and extrusions can create

lACK

~L
fiLM

HIGH DINSI"'" IN IADIOO ••V" LOW DENSlfT IN l/tOIOOIA'M

Figure 5.14 X-Ray film exposure setup

 5.3 Nondestructive Inspection Techniques 275

laminar defects such as seams, laps, stringers, and cracks. Machining with
improper feeds and speeds can cause tears, laps, and cracks. Heat treating to
obtain desired mechanical properties can also cause various types of cracking
upon quenching or cooling. Joining operations such as welding or brazing can
result in casting defects in the weld area as well as cracks in the adjacent heat-
affected zone.
Once a component has been placed in service, flaws can initiate and propagate
from a wide variety of factors. The most prevalent of these include: fatigue, corro-
sion, overload, creep, stress corrosion cracking, and hydrogen embrittlement, to
name a few. Regularly scheduled in-service inspections using non-destructive
evaluation techniques can allow components to be utilized far beyond what could
safely be achieved without inspection. The importance of NDI for in-service in-
spections is illustrated in the crack growth rate curve and corresponding residual
strength curve shown in Figure 5.15. The crack growth rate curve (top) plots crack
size as a function of time based on anticipated service load profile. Underneath this
curve is the corresponding residual strength curve which shows the reduction in
strength with time and consequently crack instability. Point A corresponds to the
crack size at which the defect becomes reliably detectable for the specific NDI
method. Point B is the point at which residual strength has decreased to the failure
load. The time between Point A and Point B is then the period available for the de-
tection of cracks that may propagate in service before attaining a critical size [6].

1
en~ ,.,r CrItical
.w - - - - - - - - - - - -
~ Minimum
U Detectable

AI ~Time I
DesignI I
Ultimate I
.s::
~ r-------~ ~ __
II
~ .r Fail-Safe I I
!,Load I
~ ------...J-----
I Detectable
I Period
Ai<
a ~Time

Figure S.lS Crack growth rate and residual strength curves

276 Chap. 5 Fracture Control Program and Nondestructive Inspection

Computational approaches to fracture mechanics require the assumption of a

maximum initial flaw size. The true value of NDI for stress analysis lies in its
application to screen critical components for flaws, assuring the validity of the
critical maximum flaw size assumption. For NDI flaw screening to be effective,
the detection capability of a particular NDI technique for a given application must
be established. The task of assessing NDI capability is complex and is under con-
stant refinement. The basis for all quantitative NDI capability assessments is to
subject test specimens containing known flaws of varying sizes and locations to
specific NDI inspection processes and measuring capability as a function of the
detected flaw size. This is not a trivial task.
Extensive testing programs have been undertaken by NASA as well as several
aerospace contractors to characterize the flaw detection capability of the predom-
inant NDI methods. In typical NDI reliability testing, specimens containing
fatigue cracks are utilized to provide uniform controlled flaws for meaningful
comparisons. Starting with an electrodischarge machined (EDM) notch in the test
specimen, a crack is initiated and grown until the desired size is reached. The
EDM starter notch is then machined off, leaving only the fatigue crack. Several
flaws of various orientations can be grown in a single specimen. The use of
fatigue cracks allows tight control over flaw length, depth, and aspect ratio. How-
ever, it also adds some conservatism in the flaw size detection capability as tight
fatigue cracks represent one of the most difficult flaws to detect. For the inspec-
tion of new production hardware, fatigue cracks are not an expected type of defect
unless the hardware has already been subjected to a fatigue environment.

5.3.8 Probability of Detection Statistics

Nondestructive evaluation involves the measurement of complex physical para-
meters with inherent variations in the measurement technique as well as in the test
specimens [6]. From a statistical point of view, the inspection outcome is not a
simple accept/reject decision but rather a conditional acceptance with four possi-
ble outcomes: true positive (flaw exists and is identified), false positive (flaw does
not exist but is identified), false negative (flaw exists but is not identified), and
true negative (flaw does not exist and is not identified). Two independent proba-
bilities can be considered to describe the inspection reliability-the probability of
detection (POD) and the probability of false alarms (POFA). These are deter-
mined as follows:

POD = Total True Positive CallslTotal Number of Defects

POFA = Total False Alarms/Opportunities for False Alarms

For a large flaw, the signal and noise distributions will be well separated, and the
POD will be high and the POFA low as shown in Fig.5.13a. If the inspection is
repeated on a smaller flaw, signal and noise probability density distributions will
 5.3 Nondestructive Inspection Techniques 277

k" Noise Alone k" Noise

Threshold
Signal Alone

(a) Response Level

a) Large Raw SignalJNoise Distribution b) Small Raw SignaVNoise Distribution

Figure 5.16 Signal/noise density distributions for large and medium size flaws

be closer together as depicted in Fig. 5.16b. For this flaw size, the signal and noise
distributions overlap to a degree and the POD will decrease while the POFA
increases [6].
NDI capability and reliability, however, are also affected by variations in the
inspection process. Small flaws may be easily detected in the laboratory by a
given technique, while large flaws may be missed by the same technique under
less than ideal conditions. The factor of greatest importance in determining capa-
bility and reliability of an NDI technique is not the smallest flaw detected, but is
defined by the largest flaw that could go undetected. Under this premise, NDI
technique reliability is best determined by statistical methods. A valid statistical
method of defining inspection performance is to specify percent probability
(POD) at a given confidence level (C) for a crack of a given size. The values for
POD and C commonly used for NDI reliability demonstrations have been 90%
POD with 95% confidence level in order to meet MIL Handbook 5B require-
ments. The significance of establishing a confidence level for a given POD lies in
the specification of a sample size to achieve statistical significance. The confi-
dence level using a binomial distribution analysis is defined by the following [6]:
N
C = 1 - ~ (XN)(PODY(1 - POD)N-X
x=s

where
C = confidence level
N = sample size
S = the number of successes (flaws detected)
POD = the required probability of detection.
278 Chap. 5 Fracture Control Program and Nondestructive Inspection

By specifying the confidence level (C) as 95% and the lower bound probabil-
ity (POD) as 90%, a set of values for sample size (N) and number of successes (S)
can be calculated. Each combination of N and S that satisfies the above equation
indicates the number of inspections and the number of detections required to
achieve the specified 90% POD at the 95% confidence level for a given flaw size.
Table 5.2 lists the required number of successes for a given sample size that meets
the 90% POD/95% confidence level criteria.
The binomial distribution analysis is an effective means for qualifying inspec-
tion personnel to a specific flaw size detection requirement. A more descriptive
NDI capability is determined from the relationship between POD and crack size.
This POD/crack size relationship is shown in Fig. 5.17. Various analytical tech-
niques have been applied to generate POD/crack size curves. The two most
widely applied statistical methods are the moving average method and the maxi-
mum likelihood method [6].

5.3.9 Variables Affecting NDI Flaw Detectability Testing

The results of reliability testing for a given NDI method are influenced by many
factors. The tests are a measure not only of an inspector's ability to detect and
characterize flaws but also of the capability of the equipment and processing

Table 5.2a Initial flaw size detectability assumptions for standard penetrant NDI processing

Part Aspect Crack Crack

NDI thickness Crack ratio depth length
method Area (in.) type (aI2e) (in.) (in.)

PT Open
standard surface t < 0.075 TC 0.200
PTC 0.1 0.250
0.2 0.200
0.5 (0.075 max.) 0.175
t;;;' 0.075 TC 0.150
PTC 0.1 0.250
0.2 0.200
0.5 0.l50
Surface
edges t < 0.075 TC t 0.100
t;;;' 0.075 CC 0.075 0.100

PT = Penetrant inspection
PTC = Part through crack
CC = Comer crack
TC = Through crack
 5.3 Nondestructive Inspection Techniques 279

Table S.2b Initial flaw size detectability assumptions for standard eddy current NDI
processing

Part Aspect Crack Crack

NDI thickness Crack ratio depth length
method Area (in.) type (a/2c) (in.) (in.)

ET Open
standard surface t? 0.020 TC 0.100
PTC 0.1 0.100
0.2 0.100
0.5 0.100
Surface
edges t? 0.020 TC t 0.050
CC 0.020 0.050

ET = Eddy current inspection

PTC = Part through crack
CC = Comer crack
TC = Through crack
Table S.2e Initial flaw size detectability assumptions for standard magnetic particle
NDI processing

Part Aspect Crack Crack

NDI thickness Crack ratio depth length
method Area (in.) type (a/2c) (in.) (in.)

MT Open
standard surface t? 0.070 TC 0.250
PTC 0.1 0.375
0.2 0.275
0.5 0.250
Surface
edges t? 0.070 TC t 0.250
CC 0.070 0.250

MT = Magnetic particle inspection

PTC = Part through crack
CC = Comer crack
TC = Through crack

steps. The following section briefly discusses the variables affecting each partic-
ular NDI method.

5.3.9.1 Variables Affecting the Liquid Penetrant Inspection

Liquid penetrant inspection is the simplest NDI method in principle. It is, how-
ever, a multistep process is which improper performance of any single step can
f\)
co
o

Table S.2d Initial flaw size detectability assumptions for standard ultrasonic NDI processing

Part Aspect Crack Crack

NDI thickness Crack ratio depth length
method Area (in.) type (aI2e) (in.) (in.)

UT Standard
(L-wave or S- Raw stock or Embedded
wave machined orPTC, TC, 0.200 dia. (Equivalent Area)
t;;" 0.300 CC Class B
Class A 0.130 dia. (Equivalent Area)

UT = Ultrasonic inspection
PTC = Part through crack
CC = Comer crack
TC = Through crack
 Table S.2e Initial flaw size detectability assumptions for standard radiographic NDI processing

Part Aspect Crack Crack

NDI thickness Crack ratio depth length
method Area (in.) type (a/2e) (in.) (in.)

t
RT standard Raw stock or (0.150 min.)
machined t? 0.050 Embedded Ellipse 0.7t
PTC, TC,CC 1/2 Ellipse 0.7t t
(0.150 min.)
t < 0.050 Embedded Ellipse 0.7t (0.025 0.150
min.)
PTC, TC,CC 1/2 Ellipse 0.7t (0.025 0.150
min.)

RT = Radiographic inspection
PTC = Part through crack
CC = Corner crack
TC = Through crack

f\)
<Xl
......
282 Chap. 5 Fracture Control Program and Nondestructive Inspection

100

80
~
>.
~ 60
:c
III
.c
e 40
c..
c
0
'ij 20
U
....
III

c3
•
0

Flaw Size

Figure 5.17 Typical probability of detection curve

seriously affect flaw detection. Surface preparation is the first area of concern,
since flaws must be open to the surface for detection. If grinding, machining, or
even polishing operations have been performed, flaws may be smeared over,
reducing or precluding detectability. Acid etching a light layer (0.0002 in. typi-
cally) to remove the smeared metal may be required. Surfaces must also be clean
and free from contaminants that may fill flaws preventing the penetrant dye from
entering. Rough surfaces can also reduce detectability by holding penetrant dye in
scratches and grooves, making the detection of defects difficult.
The selection and application of penetrant dyes also plays a serious role in
defect detection. Penetrant dyes are classified as fluorescent (type I) or visible
(type II) as well as by sensitivity level (1, 2, 3, or 4). Type I fluorescent dyes are
easier to see and thus are more sensitive to defect detection. Sensitivity level
refers to the brightness of the dye under ultraviolet light. The brighter the dye, the
easier it is to detect. The brightest dyes are sensitivity level 4, specified mainly for
highly stressed rotating engine component inspections. Penetrant dyes must also
be allowed to dwell on the part long enough to penetrate into the defects. Removal
of the penetrant must also be performed properly. Overzealous removal can wash
out dye from shallow defects while underwashing will leave a significant back-
ground, adversely affecting detectability. Proper application of a light coating of
developer is also important. Proper lighting during evaluation is also a concern.
Excessive white light during fluorescent penetrant evaluations can reduce defect
detection, as can using ultraviolet lights of inadequate intensity.
Figure 5.l8a shows a POD curve developed using the maximum likelihood
analysis method for a type I, sensitivity level 3 nonaqueous developer inspection
of Haynes 188 alloy test specimens. The lower curve represents the 95% confi-
dence level and indicates a flaw size of approximately 0.080 in. in length. In com-
parison, Figure 5.l8b shows a POD curve for a type I sensitivity level 2 penetrant
 5.3 Nondestructive Inspection Techniques 283

taG
10
... _-- .--
10
...
~
"'" •"
GO
0
0 •
3D
II.
20
to

0.01 0.01 O.t 0." '.Sl 0 •• 0 •• 0.:1

Crack Length (Inches)

Figure 5.1Sa POD curve examples for fluorescent penetrant inspection (sensitivity
level 2 penetrant). NASA/Inconel 71S. Inspection #2S. Type I. Method A. Form d.
(Reproduced From Ref. [6])

sao
•
.-
•
70

•
~

!lei
0
0 •
3D
II.
10
10

0.02 0.01 0.1 0.14 O.SI 0.22 0.. 0.3

Crack Length (Inches)

Figure 5.1Sb POD curve examples for fluorescent penetrant inspection (sensitivity
level 3 penetrant). NASA/Inconel 71S. Inspection #36. Type I. Method A. Form d.
(Reproduced From Ref. [6])

with nonaqueous developer inspection of similar specimens. The 95% confidence

level indicates a flaw size of about 0.120 in. in length [6].

5.3.9.2 Variables Affecting the Magnetic Particle Inspection

In reliability testing using magnetic particle methods the results are greatly
affected by the proper development and application of the inspection technique.
The proper selection of magnetizing currents is important to assure attraction of
the ferromagnetic particles to the defect without obscuring its detection by too
much current flow. As with penetrant inspection methods, magnetic particle
284 Chap. 5 Fracture Control Program and Nondestructive Inspection

detectability performance is also influenced by the type of materials used. Fluo-

rescent particles provide better sensitivity; however, too much white light or inad-
equate ultraviolet light intensity will reduce sensitivity to defect detection. The
orientation of the flaws away from 90° to the applied magnetic field will also
decrease detectability, many times requiring multiple orientations and field
strengths for complete inspection coverage. Magnetizing shots should also be per-
formed in increasing field strength, as residual magnetism could obscure subse-
quent evaluations. Proper flow of the ferromagnetic particle medium during or
immediately following magnetization is also important. Flaw depth can also affect
inspection sensitivity. Subsurface flaw indications are usually less distinct than
surface flaws. Complex part geometries and sharp edges also can produce false
indications.

5.3.9.3 Variables Affecting the Eddy Current Inspection

Eddy current inspections are most seriously influenced by the equipment selection
and scanning techniques. The pattern of eddy current flow is greatly affected by
probe size, selected frequency, and the conductivityImagnetic permeability of the
material being inspected. Higher frequencies and smaller probe coils provide a
smaller eddy current field and thus increase the sensitivity to smaller surface
defects. This, however, reduces the current flow into the material depth and
reduces subsurface flaw detection capabilities. Other equipment settings such as
gain and signal/noise ratio influence flaw detection performance.
Inspection of an entire part surface with eddy current requires scanning the
probe over the surface. Defect detection can be affected by the speed of scanning,
indexing increments, and the distance from the probe to the part, called lift-off.
The flow of eddy current drops exponentially with distance from the probe coil.
Rough surfaces and inconsistent probe pressure can adversely affect flaw detec-
tion capabilities of an eddy current inspection. Automated scanning methods with
uniform speed, indexing, and probe pressure have been found to improve flaw
detection capabilities. Figure 5.19 shows eddy current POD curves determined by
the maximum likelihood analysis method on Haynes 188 test specimens using
manual (Fig. 5.19a) and automated (Fig. 5.19b) inspection techniques [6].

5.3.9.4 Variables Affecting the Ultrasonic Inspections

Ultrasonic inspection flaw detectability is affected by many of the same variables
as eddy current methods. Frequency selection and transducer design are signifi-
cant factors. High-frequency scans offer greater sensitivity but are more easily
attenuated, thus increasing signal noise. Smaller probe diameters or focused
probes can also provide increased defect detectability. Complete part inspections
with ultrasonics requires scanning the transducer over the surface using a liquid
 5.3 Nondestructive Inspection Techniques 285

100

•
.... •
...~ •
70

a •...
0
D. •
10
10

a-oJ .0115 .on .01S .oe .au .0115

Crack Depth (Inches)

Figure 5.19a POD curve examples for eddy current inspections (manual inspection,
differential probe). NASA/Haynes 188. Inspection #79. Eddy Current. 1 MHz.
(Reproduced From Ref. [6])

-110

•
10

'"
~
711
10

a ••
0
D. •
•
'0
.-oJ .011 .on .as ... .• ._
Crack Depth (Inches)

Figure 5.19b POD curve examples for eddy current inspections (automated inspec-
tion, absolute probe). Haynes 188. Inspection #82. Eddy Current. 1 MHz. (Repro-
duced From Ref. [6])

couplant such as water or oil. Insufficient couplant can cause signal loss while
excess couplant on shear wave inspections can cause spurious reflections.
Flaw orientation and configuration also have a great effect on detectability per-
formance. For optimum scan sensitivity, the ultrasonic beam must be oriented per-
pendicular to the flaw, thus providing maximum signal reflection back to the
transducer. Also, as the ultrasonic signal propagates through a material it loses
energy. Thus, a flaw located deep within a material produces an ultrasonic signal
of less amplitude than an identical flaw located close to the surface. Figure 5.20
286 Chap. 5 Fracture Control Program and Nondestructive Inspection

...
lOG ----.:--:::;;-:;-:;:--;.:::=!!!!!:I"--

a 4
oD. 30

10

O.OZ 0.01 0.1 0." 0.18 O.D 0.21 0.3

Crack Length (Inches)

Figure 5.20 POD curve example for shear wave ultrasonic inspections. NASA/lconel
718. Inspection #84. Hand Scan Ultrasonic (Reproduced From Ref. [6])

shows a POD curve detennined by the maximum likelihood analysis method for
a contact shear wave ultrasonic inspection of Inconel 718 specimens.
Similar to eddy current inspections, ultrasonic scans are also subject to scan speed
and indexing inconsistencies. Automated scanning can be used to improve consis-
tency, but may not always provide proper orientation for optimum flaw detection.

5.3.9.5 Variables Affecting the Radiographic Inspection

Radiographic defect detectability testing typically produces poor perfonnance and
relatively large maximum initial flaw sizes compared to other NDI methods. This
can be attributed to the nature of the detectability test flaws which are exclusively
fatigue cracks. Radiography is a volumetric inspection where density variations
provide image contrast on photographic film for defect detection. While sensitive
to changes of 1% to 2%, tight fatigue cracks do not provide significant volume
unless the x-ray beam is oriented perfectly parallel with the crack. Other variables
affecting detectability perfonnance are x-ray tube kilovoltage and current, length
of exposure time, type of x-ray film used, distance from the x-ray tube to the film,
and distance from the flaw to the film. The farther a feature is from the film, the
less distinct it appears.
X-ray images are essentially a 2-D density representation of a 3-D object.
Complex or multilayer parts result in x-ray images with superimposed features
that may be difficult to interpret. Interpretation is also adversely affected by
excessive light during evaluation as well as defects in the x-ray film itself such as
scratches and blemishes.

5.3.10 Critical Initial Flaw Sizes for Standard NDI

Conducting NDI perfonnance testing for every flaw screening application would
prove cost prohibitive for most manufacturing budgets. The need for standardized
 5.3 Nondestructive Inspection Techniques 287

critical initial flaw sizes for fracture mechanics computations was evident. During
the space shuttle development program in the 1970s, NASA addressed this issue
and performed comprehensive NDI reliability testing in the five major NDI disci-
plines using a wide variety of inspection labs. The results were then compiled and
minimum critical flaw sizes were established at a 90% POD and a 95% confi-
dence level using standard NDI processing techniques. These are listed in Table
5.2 while the accompanying crack geometries are depicted in Figure 5.21 [7].
These flaw sizes represent the smallest flaws that can be detected using a particu-
lar NDI method at the 90% POD and 95% confidence level employing NDI indus-
try standard processing methods. These results have been utilized on numerous
military and commercial programs throughout the aerospace industry over the
past 20 years. Recently, NASA has undertaken a program to reexamine the NDI
POD flaw sizes in light of advanced processing techniques. Initial results indicate
that although NDI evaluations are now performed in less time and at lower costs,
the critical initial flaw detectability limit using standard processing methods is rel-
atively unchanged from the initial study. The initial flaw sizes listed in Table 5.2
can be used by an analyst to conduct safe-life analysis of fracture critical parts. If
the results of the analysis indicated that the part does not have sufficient life, a
smaller initial flaw size must be assumed. Section 5.3.11 addresses the special
NDI inspection method for obtaining the desire initial flaw size (smaller than the
standard initial flaw sizes listed in Table 5.2) to ensure safe-life.

Partia l Through Surface Cracks

Through Cracks

-
Comer Cracks

ciilfJ~~c
1" .
t a . t

Figure 5.21 Initial crack geometries

288 Chap. 5 Fracture Control Program and Nondestructive Inspection

5.3.11 Special Level Inspection

For some critical design applications, the maximum initial flaw sizes verifiable
using standard NDI processing techniques are still too large to assure safe life.
For these applications, NDI methods may still be used to screen for the desired
flaw sizes by employing special level NDI. To perform special level NDI an
inspector must first undergo NDI reliability testing in the applicable discipline
using test materials and processing techniques similar to those of the components
of interest. Results of this testing must verify that the inspector can detect flaws as
small as the maximum initial design flaw at a 90% POD and 95% confidence
level. Recertification is then required a minimum of every two years.

References

1. R. c., McMaster, (Ed.), Liquid Penetrant Tests, Nondestructive Testing Handbook,

Second Edition, Vol. 2, American Society for Nondestructive Testing, 1982.
2. 1. T., Schmidt and K. Skeie, (Technical Eds.), and P. MacIntire Ed., Magnetic Particle
Testing, Nondestructive Testing Handbook, Second Edition, Vol. 6, American Society
for Nondestructive Testing, 1989.
3. L. E. Bryant, (Technical Ed.) and P. MacIntire (Ed.), Radiography and Radiation Test-
ing, Nondestructive Testing Handbook, Second Edition, Vol. 3, American Society for
Nondestructive Testing, 1985.
5. A. S. Birks and R. E. Green, Jr. (Technical Eds.), and P. MacIntire (Ed.), Ultrasonic
Testing, Nondestructive Testing Handbook, Second Edition, Vol. 7, American Society
for Nondestructive Testing, 1991.
5. M. L. Mester, (Technical Ed.), and P. MacIntire, (Ed.), Electromagnetic Testing, Non-
destructive Testing Handbook, Second Edition, Vol. 4, American Society for Nonde-
structive Testing, 1986.
6. B. K. Christner, D. L. Long, and W. D. Rummel, NDI Detectability of Fatigue-Type
Cracks in High-Strength Alloys, NDI Reliability Assessments, MCR-88-1044, Martin
Marietta Astronautics, 1988.
7. Standard NDE Guidelines and Requirements for Fracture Control Programs, MSFC-
STD-1249, NDI Branch, Materials and Processes Laboratory, NASA-MSFC, 1985.
Chapter 6

The Fracture Mechanics of

Ductile Metals Theory

6.1 Introduction

Almost all metallic materials manifest some plastic deformation in the region at
the crack tip before catastrophic crack propagation. The general principle from
which the Griffith theory [1] (see Section 3.1 of Chapter 3) is derived is not lim-
ited to materials that obey Hooke's law. The principle applies as well when dissi-
pative mechanisms, such as plastic deformation, are present. Irwin and Orowan
[2,3] showed that Griffith's principle can also be applied to materials that mani-
fest ductile behavior, that is:

(6.l)

where UE is the stored energy, and Up is the energy consumed per unit thickness
in plastic straining in the region at the crack tip. For ductile metals, where
Up » Us' the expression for surface energy, Us' was omitted from Eq. (6.1) to
obtain:

a
-
ac [UE - U ]
p
=0 (6.2)

From Eq. (6.2), the following well-known Irwin-Orowan equivalent of the Griffith
energy balance equation is obtained:

(6.3)

B. Farahmand et al. (eds.), Fatigue and Fracture Mechanics of High Risk Parts 289
© Chapman & Hall 1997
290 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

The quantity to the left is the energy release rate or the crack extension force at the
instability, Gc ' and has dimensions of force per unit of extension. The crack exten-
sion force, G, is the same for the cracked plate under fixed grip condition (where
the external load cannot do work) and free ends where the work is done by the
external load. In the former case the available energy is delivered by the elastic
energy and in the latter case by the load. The quantity to the right of Eq. (6.3) is
the critical value of the stress intensity factor, Kc' called fracture toughness which,
for the thick sections when plane strain condition prevails, is designated by K Ic '
Fracture toughness values for thicknesses smaller than the plane strain case are
referred to as mixed mode where the maximum value is associated with the plane
stress condition labeled as Kc' In Section 6.2, the fracture mechanics of ductile
metals (FMDM) are discussed. A relationship between fracture stress and the cor-
responding critical half crack length is developed that enables the analyst to esti-
mate the residual strength capability of the structural part without relying on test
data generated in the laboratory. To verify the results and accuracy of the FMDM
theory, several comparisons were made with the experimental data provided from
different reliable sources. The correlation between the FMDM theory and the
experimental data with various alloys is presented in Section 6.9. In addition, the
fracture toughness values for several alloys were calculated by the FMDM
approach and the results compared with ASTM fracture toughness testing (Sec-
tion 6.10).
To appreciate the simplicity of calculating the fracture toughness with the
FMDM theory, the reader may refer to Chapter 3 pertaining to the complexities
associated with the ASTM specimen preparation procedures, including the
machine starter slot, fatigue cracking the specimens prior to testing, method of
applying load to the specimen, load recording, and crack measurements.

6.2 Fracture Mechanics of Ductile Metals

The theory of ductile fracture assumes the fracture characteristics of a metal, local
to the crack tip, are directly related to strainability of metals which is of two kinds:
(1) local strainability (necking) at the crack tip, the region of highly plastic defor-
mation, and (2) uniform strainability near the crack tip. Thus, fracture is charac-
terized by two parameters that represent the absorbed energy associated with two
plastic deformation regions which can be shown to be determinable from the uni-
axial stress-strain curve. The two aforementioned strained regions are illustrated
in Figs. 6.1 and 6.2 for the crack tip and uniaxial stress-strain curves, respectively.
The total energy per unit thickness absorbed in plastic straining of the material
around the crack tip, Up, from Eq. (6.1), can be written as:

(6.4)
 6.3 Determination of the auF / ac Term 291

Uniform Straining,
UU Near the Crack Tip

Crack
.-:~---

Highly Strained Region,

UF at the Crack Tip

Figure 6.1 The crack tip plastic zones

Engineering
Stress-Strain Curve

Uniform Straining Region

Strain

Figure 6.2 Different regions of the stress-strain curve

where UF and Uu are the energy absorbed per unit thickness in plastic straining of
the material beyond the ultimate at the crack tip and below the ultimate stress near
the crack tip, respectively. Equation (6.1), in terms of U F and U u , described by
Eq. (6.4), can be rewritten as:

J
- [U - U - U - U ] = 0 (6.5)
Jc E S F U

where aUF
ac
and au
ac
u are the rates at which energy is absorbed in plastic straining

beyond the ultimate stress at the crack tip and below the ultimate stress near the
crack tip, respectively.
To obtain the residual strength capability curve of a cracked plate subjected to
tension (the plot of the applied stress, (J'c' versus the critical half crack length, c) it
is therefore necessary to determine the energy absorption rates for the two plastic
regions around the crack tip. Sections 6.3 and 6.4 discuss the theoretical approach
and assumptions used in obtaining the aUF ae
and a_uiJcu terms for the plane stress con-
dition. The plane strain condition is also discussed in Section 6.5.
292 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

6.3 Determination of the aUF Term

ac

To obtain the energy absorption rate for the highly strained region at the crack tip,
a~F , let us assume that the remaining bulk of the material (except at the crack tip)
is uniformly strained up to the ultimate of the material. This can be done by apply-
ing to the cracked plate an applied stress equal to the ultimate strength of the
material, O'Applied = O'uc· In this case, the term Uu ofEq. (6.5) can be eliminated,
since there is no energy absorption contribution at the crack tip from the Uu term
(see Fig. 6.1). For O'Applied = O'uc' Eq. (6.5) becomes:

7TO'ZC
- -= U
-Ua[ U]
E ac +
S F

or

7T(J'ZCu = 2T + aUF (6.6)

E ac
where Cu is the crack length associated with 0' Applied = O'uc. This value is impor-
tant in the field of fracture mechanics when initial flaw size for fatigue crack
growth analysis is needed (the quantity Cu can be viewed as the largest flaw size
preexisting in the material). At any other applied stress below the ultimate,
O'Applied < O'uc' Eq. (6.5) becomes:

(6.7)

where T is the surface tension of the material, that is, the work done in breaking
the atomic bonds and its value was given by Eq. (3.4) of Chapter 3. The quantity
aUF in Eq. (6.6) is set equal to WFh F, where WF is equal to the unrecoverable
ac
energy density (energy per unit volume) represented by the area under the plastic
uniaxial engineering stress-strain curve from the stress at which necking begins to
the stress at fracture for an uncracked tensile specimen (see Fig. 6.3 for different
types of stress-strain curves). It is assumed that essentially all of the energy rep-
resented by WF is absorbed at the crack tip in a single dominant coarse slip band.
The quantity hF is the effective height of the highly strained portion of the small
region at the crack tip, and its minimum value is considered to be equal to the
effective height of a coarse slip band oriented to make an angle of 45° with the
plane of the crack. The movement of the two slip surfaces by the amount of hF
due to movement of dislocations results in release of energy in the locality of the
crack tip. The size of coarse slip bands at high stress and at room temperature is
approximately 10 micrometers [4,5] and therefore:

hF = 0.00057 in. (6.8a)

 6.3 Determination of the aUF/acTerm 293

A B
~ A~ I
I

~ I
1
I I
I I
I
.... I
...

C ~
D
~

I
I
I
I
I
I
I ..
I ....

Figure 6.3 Typical engineering stress-strain curves for: (A) brittle materials, (B)
metals that do not neck, (C) many structural metals, (D) some precipitation-harden-
ing steel

An empirical relationship based on test data [6] has been developed for tough
metals that have large "neck" strains (see for example case D of Fig. 6.3) which
gives a higher hF value and better correlation with the test data than indicated by
Eq. (6.8a):

(6.8b)

where a is equal to 6.31 X 10- 5 and 5.68 X 10- 5 for iron nickel and beryllium
alloys, respectively. The area under the stress-strain curve from necking up to
fracture, WF , is equal to:

(6.9)

where the quantity (j UF is the neck stress and its value is at the centroid of the
plastic energy bounded on the top by the engineering stress-strain curve from the
294 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

beginning of necking to fracture and on the bottom by a straight line from the
beginning of necking to fracture. For material with a negligible amount of necking
the neck stress aUF = ((J"U + (J"F)/2. Moreover, cPN is the plastic straining at the
onset of necking and its derivation is shown in Appendix B. Equation (6.7) in
terms of hF' aUF' and C PN can be expressed as:

(6.7a)

The critical flaw size associated with applied stress equal to the ultimate stress of
the material, cU' described by Eq. (6.6), can be written as:

(6.10)

As discussed in Chapter 4, for structural life evaluation, when the load envi-
ronment is cyclic, the initial crack size assumption (the original flaw that preex-
isted in the material prior to its usage) must be available in order to conduct a
meaningful fracture mechanics analysis. From equation 6.10 it can be seen that for
material that do not exhibit necking (where hF(J"UFCPN = 0) the critical flaw size at
ultimate, cU' becomes small. This is in accordance with the experimental obser-
vation which shows that a brittle material can tolerate a much smaller initial flaw
at failure as compared with ductile materials with an appreciable amount of neck-
ing prior to their final failure.

6.4 Determination of the a~u Term

The quantity a~u can be expressed as W uhu, where W U is the unrecoverable energy
density represented by the area under the plastic uniaxial stress-strain curve from
the elastic limit stress, (J"v to the ultimate stress, (J"Uf' for an uncracked specimen.
Equation (6.7a) becomes:

(6.7b)

The effective height of the volume in which WU is absorbed is hu. The expression
for W uhu is derived in Sections 6.4.1 and 6.5 for the two cases of plane stress and
plane strain conditions, respectively. In Section 6.4.1, the octahedral shear stress
theory is used to develop an expression for the total true strain near the crack tip
from elastic limit load up to the ultimate of the material. This theory is used here
 6.4 Determination of the auu/ac Term 295

to describe the crack tip multiaxial stress state in terms of the uniaxial stress
obtainable through laboratory testing.

6.4.1 Octahedral Shear Stress Theory (Plane Stress

Conditions)
From linear elastic fracture mechanics (LEFM), the crack tip true stresses for
plane stress conditions (when r / c « 1.0) were derived in Section 3.2 of Chapter
3. In polar coordinates the crack tip true stress can be written as:

(TTr(r, e) = V2
(T fC (5 e 1 e)
'Y~ "4 cos "2 - "4 cos 3"2

(r e) = - (T /; - (3- e+ -41cos 3 -2e)

cos -
(T
T8' V2 r 4 2

(TTr8(r, e) = V2
(T fC (1 e 1 e)
'Y~ "4 sin "2 + "4 sin 3 "2 (6.11)

In Eq. (6.11), it is assumed that the true stresses are approximately equal to the
engineering stresses within the elastic range. For simplicity let the quantities
inside the parentheses [shown by Eq. (6.11)] be represented by:

5 e 1 e
'"r ( e) == "4 cos "2 - "4 cos 3 "2
3 e 1 e
'" 8 ( e) == "4 cos 2" + "4 cos 3 2"

(6.12)

Substituting Eq. (6.12) in Eq. (6.11) the crack tip true stress can be rewritten as:

(6.13a)

(6.13b)

(6.13c)
296 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

The equation for octahedral shear stress, TOCT (similar to the strain energy of
distortion) satisfactorily correlates states of biaxial tension stress ((J"Tr and (J"TO)
with uniaxial tension stress, (J"T' and is used here to relate the stresses in the zone
just ahead of the crack tip to the uniaxial tension stress, (J"T [7]. The octahedral
shear stress, TOCT ' measures the intensity of stress that is responsible for bringing
a solid substance into the plastic state. The equation for octahedral shear in terms
of crack tip stresses is:

TOCT -
_ 3
1 [( (J"x - (J"y)
2
+ ((J"y - (J"z
)2
+ ((J"z - (J"x
)2

+ 6(T;y + T;z + Tz~)]1/2

or in terms of rand ():

2 2 2 )]1/2
+ 6 ( TTrO + TTOz + TTzr (6.14)

For plane stress, (J"TZ = 0, TTOz = 0, and TTzr = °and Eq. (6.14) becomes:
(6.15)

(6.16)

For simple uniaxial stress, (J"T' where (J"TO = 0, and (J"TrO = 0:

(6.17)

Equating TOCT [Eq. (6.16)] for uniaxial stress to TOCT [Eq. (6.17)] for plane stress:

(6.18)

or,

(J"T
= (J"
Tr
[1 - !/J o(()
!/Jr(()
+ (!/JO(())2
!/Jr(()
+ 3 12
(!/Jro(())2J /
!/Jr(()
(6.19)
 6.4 Determination of the aUu/ac Term 297

and,

Let the quantity inside the bracket be equal to o/oc (lJ):

Substituting Eq. (6.21) in Eq. (6.20):

(6.22)

The true crack tip uniaxial stress, (FT' in tenns of applied stress, (F, crack length, c,
and the crack tip distance, r, using Eq. (6.13a) becomes:

(6.23)

and in tenns of strain, BT , Eq. (6.23) becomes:

(6.24)

Equations (6.23) and (6.24) define the variation of true stress and strains at the
crack tip up to the elastic limit of the material, BTL' For small plastic defonnation
(when the crack tip plastic defonnation is small and the bulk of the material is
elastic) the quantity, (FT' can still be detennined by Eq. (6.23) with reasonable
accuracy. However, for ductile materials where an appreciable amount of strain-
ing occurs at the crack tip, equation 6.23 must be avoided when assessing the
stresses in that locality. Eq. (6.24) can be extended to describe the variation of
unifonn true straining, BT , from the elastic limit to the ultimate of the material.
Figure 6.4 illustrates the variations of true elastic stress, (Fp and strain, Bp as
described by LEFM, and plastic strains from limit up to ultimate strain with
respect to the crack tip distance, r, for a typical metal. The quantity BT as described
by Eq. (6.24) can be used to detennine the rate at which plastic energy is absorbed
in the larger zone (unifonn plastic region). The sum of all the elements acting on the
area inside the unifonn plastic zone that contribute to energy absorption rate as the
crack extends, can be expressed as (Fig. 6.5):
298 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

Uniform Elastic Range

Plastic
£T

(}"-----
Elastic Strain Uniform Stress, (}"

rU Distance, r rL Distance, r

Figure 6.4 (1) Stress and strain distribution in the elastic region up to the limit and
(2) uniform strain distribution near the crack tip up to the ultimate

~dc l
c

Figure 6.5 Illustration of an element (dA) chosen in the plastic zone for dUu / de
analysis

auu =
ac
I
(AJ
U"r ae rp dA
ac
= 2 Ie,.
0
frL

ru
U"r ae rp rdrdO
ac
(6.25)

where Oe is the angle at which the plastic energy absorption rate, dede vanishes, TP ,

i.e., the angle at which no contribution to plastic deformation is made at the crack
 6.4 Determination of the aUujac Term 299

tip. Replacing the total strain expressed via Eq. (6.24) into Eq. (6.25) (by assum-
ing the total strain is approximately equal to the plastic strain, cr = crp):

au
_~u = 2 J O
' ac rdrde
IrL O"r_r (6.26)
ac 0 r, ac

When the magnitude of the applied stress is in such a way that the net section
yielding is avoided except in the localized region at the crack tip, the limits of the
integral, described by Eq. (6.26), are correct. However, when the applied stress is
above the elastic limit of the material or when the bulk of the structure due to the
applied stress is plastic, Eq. (6.2) can be rewritten as:

au
_~u =2 J I O"r_r
ac rdrde
O
'
r
(6.26a)
ac 0 r, ac

Note that when the applied stress becomes larger than the elastic limit stress of
the material, the plastic deformation contribution at the crack tip due to the energy
absorption rate term associated with the uniform plastic region (shown in Fig. 6.4)
is less and only a portion of it is included in Eq. (6.26). Later in Section 6.5, the
FMDM results will be given in a general form, where both cases of applied stress
(above or below the elastic limit of the material) can be used for obtaining the
corresponding half critical crack length, c. The reader should note that one of the
limitations associated with the LEFM, is that, the formation of the crack tip plas-
ticity must be limited to the small region at the crack tip and the bulk of material
must be elastic. Otherwise, the LEFM cannot be utilized to assess the residual
strength capability determination of a part when the applied stress is above the
yield of the material or the bulk of the structure is plastic.
From Eq. (6.24), the quantity~; can be simplified as follows:

aCr O"'Poc (e) O"'Poc (e) ar

(6.24a)
ac 2 V2Evcr 2 V2Ery, ac

Under constant load conditions, the stress would be constant as the crack begins to
grow catastrophically. In this case, ilI7:
ac
would be zero and the last term in Eq.
(6.24a),"-"-~, would vanish. From simplification ofEq. (6.24a):
ac "

(6.27)
300 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

From Fig. 6.5, the quantities ~

ac
and ~
ac
can be replaced by:

~ a8 ~n(~
- = -cos8, and - =-- (6.28)
ac ac r

Substituting Eq. (6.28) in Eq. (6.27):

aer
- =-
er [-r + cos 8 + 2 sin 8 a8a ]
'Poe(8)
( ) (6.29)
ac 2r c 'Poe 8

But, ~
('
« 1; therefore:

aer e
[ a]
a8 'Poe (8)
- = ~ cos 8 + 2 sin 8 ( ) (6.30)
ac 2r 'Poe 8

Let

[ aea 'Poc(]8)
1 (8) = cos 8 + 2 sin 8 'Poc (fJ) (6.31)

Substituting Eq. (6.31) in Eq. (6.30):

aer
- = -er
18() (6.32)
ac 2r

Substituting Eq. (6.32) in Eq. (6.26a), the energy absorption rate can be written as:

au
-----.ll
ac
=2f
0
B
,. f
r
rL
e 1(8)rdrd8
(Tr~
2r
(6.33)
u

(6.33a)

For metals whose true stress versus true plastic strain curve can be approximated
by the Ramberg-Osgood Equation (when fitted at (Tn and (Tru) the true stress in
terms of true plastic strain can be written as:
 6.4 Determination of the aUu/acTerm 301

(6.34)

The quantity B TPU is the true plastic strain at ultimate and n is the strain hardening
coefficient (for perfectly plastic material, the value of n = (0) and can be
described by:

(6.35)

where BTPL is the true plastic strain at the elastic limit. For material with an appre-
ciable amount of plastic strain where B TP = BT and B TPN = BTN

(6.36)

Substituting Eq. (6.36) in Eq. (6.33):

auu
__ = (J'TUBTU 1°" frL [B--.L]~ f«())drd() n (6.37)
ac 0 ru BTU

From Eq. (6.24), the ratio of the quantity BrJB TU can be replaced by its equiva-
lent:

~ = P;rU (6.38)
BTU V--;
Substituting Eq. (6.38) in Eq. (6.37)

au
ac =
_u (J'
TU
B
TU
1°" frL [r---.!lr ]~ f«())drd()
2n (6.39)
o ru

Integrating and simplifying equation 6.39:

au
_u
ac = (J'
TU
BT
U
f O
,-2n-
n - 1
r [(-r )"-=--.!]rL
U
2n
r
f«())d()
o U ~

au = -n- U B [(r~ )n~

ac n - 1 TU TU r
_U
-I
U
- 1
]
I
0,.

0
2r f«())d()
U
(6.40)
302 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

From test data for plane stress and infinite width, an empirical relationship has
been established that can describe the r U value for ductile metals [8]:

(6.41)

where Q = h(BTFB n )/ BTU [the value of h = 0.00057 in.; see Eq. (6.8)]. For finite
width, w:

(6.41a)

For center cracked panels, the width correction, Y, is expressed as:

The quantity r U represents the radius of uniform plastic zone which extends from
r = r L to r = ru. This quantity is a function of B n , n, O"TU' and BT which in tum
is a function of applied stress and the crack length via Eq. (6.24). The simplified
presentations of plastic zone size through Irwin's and Von Mises' yield criteria
was discussed in Chapter 3 [see Eq. (3.42)].
Substituting Eq. (6.41) in Eq. (6.40) and noting that the ratio of (rdru) 1/2 =
Bru/Bn[see Eq. 6.24]:

(6.42)

{3

By numerical evaluation the thickness parameter, {3, for plane stress is:

(6.43)

The term thickness parameter, {3, is used here to show the distinction between the
energy absorption rate, a~u, defined by Eqs. (6.42) and (6.49) (defined in Section
 6.5 Octahedral Shear Stress Theory (Plane Strain Conditions) 303

6.5) for the two cases of plane strain and plane stress conditions, respectively; see
Eqs. (6.43) and (6.S0a) which describe the f3 values.
For the general case of loadings where the applied stress could be either below
or above the elastic limit stress of the material, Eq. (6.42) can be rewritten as:

(6.43a)

6.5 Octahedral Shear Stress Theory (Plane Strain

Conditions)

Using the crack tip stresses described by Irwin for the plane strain condition [see
Eq. (6.11)] where:

The octahedral shear stress from Eq. (6.14) becomes:

TaCT
_ 1 [(
-"3 O"Tr - O"TO)
2
+ (O"TO - VO"Tr - VO"TO)
2

+ (O"n - vO"Tr - VO"TO)2 + 6T;rO] (6.44)

or:

TaCT -
_V2[
3 (1 - v + V 2 )(O"Tr2 + 2
O"TO)

- (1 + 2v - 2V2)0"Tr 0"TO + 37"2TrO ]1/2 (6.44a)

For plastic straining, v = 0.5

Then:

or:

(6.44b)
304 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

As before, equating Eq. (6.44) to simple uniaxial stress, where (TTIJ = 0 and (TTrlJ = 0:

or:

(6.45a)

and:

(6.46)

Let:

(6.47)

For the plane strain condition, the energy absorption rate in the larger zone can be
simplified the same way as was done for the plane stress condition; see Eq. (6.25).
The quantity ru from test data for plane strain and infinite width:

(6.48a)

For finite width:

(6.48b)

The energy absorption rate in the larger zone, auu , in its final form for infinite
ae
width is given by:

[( e:; n - I
) -n- -
]
1
2
'TT2 L'I/!;cCe)f(e)de
IJ
(6.49)
 6.6 Applied Stress, (T, and Half Crack Length, c, Relationship 305

By numerical evaluation, the thickness parameter, /3, for plane strain is:

2 fOe
f3 = ~2 ljJo~(())f(())d() = 0.127 (6.50a)
7T' 0

Note that the value of /3 for a thickness, t, greater than the maximum thickness (to)
for plane stress and less than the thickness for plane strain (this condition is called
mixed mode where the value of /3 falls between plane stress and plane strain con-
dition) is given by:

/3 = l.30(to/t) + 0.127 [(t-to)/t] (6.50b)

6.6 Applied Stress, (T, and Half Crack Length, c,

Relationship

The quantity W U of Eq. (6.10a) can be evaluated from the area under the
stress-strain curve taken from the stress equal to the applied stress to the ulti-
mate of the material:

(6.51)

From Eq. (6.36):

(6.52)

Taking the derivative:

deTP =
(T
nCTPU ( _ T
)n-1 d(T
_T (6.53)
(TTU (TTU

Substituting Eq. (6.53) in Eq. (6.51):

(6.54)
306 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

Integrating:

(6.55)

Wu = -n- cTpuO'TU
n+1
l1- (0')n+l1
_T
O'TU
(6.56)

Substituting Eqs. (6.56) and (6.50) in Eq. (6.49):

auU = n + 1W h cTFcTL l(CTU)n~ 1 - 11 f3 (6.57)

ac n - 1 U cTUcT cT
Define hu by:

auu
--=W h (6.58)
ac u u

From Eq. (6.58) and Eq. (6.57):

(6.59)

From Eq. (6.10) and Eq. (6.5):

(6.60)

For a thin plate of infinite width (plane stress conditions, f3 = 1.3):

(6.61)

Substituting Eqs. (6.56), (6.59), and (6.61) into Eq. (6.60) gives the critical half
crack length in a plate of infinite width for plane stress fracture at applied stress iT.

* (n ~
n
1)1 h cTFcTL
cTUcT
[(CTU)n~ 1] f3}
cT
1 _ (6.62)
 6.7 Mixed Mode Fracture and Thickness Parameters 307

For mixed mode fracture, Eq. (6.62) becomes:

(6.63)

and for the finite width condition:

(6.63a)

(See Section 6.7 for a detailed description of the thickness parameters (3, k, and f.L).
Equations (6.62) (plane stress) and (6.63) (mixed mode) simply state that,
for a given crack length 2c, the corresponding fracture stress, IJ, can be easily
computed provided that the full stress-strain curve for the material under study
is available. The plot of fracture stress versus half crack length, c, is called the
residual strength capability diagram and is extremely important in the field of
fracture mechanics. From the residual strength data obtained from Eq. (6.63)
the material resistance to fracture for different thicknesses and crack length can
be computed.

6.7 Mixed Mode Fracture and Thickness Parameters

Expressions for k, {3, and f.L for mixed mode fracture [shown in Eq. (6.63)] are
described as follows:
A shear lip is formed in plane stress fracture, whereas plane strain fracture is
characterized by a flat fracture surface. It is assumed that the energies released
and absorbed in plane stress and plane strain are in proportion to these thick-
nesses, as shown in Fig. 6.6.
Let to be the maximum thickness for plane stress fracture. The thickness cor-
rections for mixed mode are applied to two strained regions at the crack tip and
near the crack tip (Figure 6.1). The thickness parameter, k, for the highly strained
region at the crack tip [see Eq. (6.6)] can be written as:
308 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

to
-,--...
2 '" ~ ~I
/
.
-4-_
to
2

-
t

Figure 6.6 Shear lip and flat fracture surfaces

where the two quantities k and j), are defined as:

k = ~ + 2C ~ to) (6.65a)
and

(6.65b)

The minimum values of k and j), are equal to unity when t = to. This is associated
with the maximum plane stress condition where complete shear lip prevails. Note
that, for the case of maximum plane stress condition where = t to'
equation 6.64
can be simplified to equation 6.6. Furthermore, the thickess parameter, {3, for the
mixed mode condition in the uniform strained region near the crack tip [see Eq.
(6.63)] can be described as:

= 2T + WFhFk + Wuhu [ 1.30 tto + 0.127 - t -o (t - t )l (6.66)

In terms of {3 and k Eq. (6.66) can be rewritten as:

(6.66a)
where

{3 = l1.30~ + 0.127C ~ to)l (6.66b)

The maximum and minimum values of {3 are 1.3 and 0.127 [see Eq. (6.66)] for the
plane stress (when t = to) and plane strain (when to = 0) conditions, respectively
[see also Eq. (6.43) and (6.50a)].
The thickness parameters K, j)" and {3, described by equations 6.65a, 6.65b, and
6.66b, each contain two terms which are normalized to the total plate thickness, t.
 6.9 Verification of FMDM Results with the Experimental Data 309

The first term (toft) represents the adjustment to the plane stress condition. The
latter one describes the amount of flat fracture, t~~ot
, that differs for the two

absorbed energy terms (quantities to the right of equation 6.66) as well as for the
energy released term to the left of equation 6.64.

6.8 The Stress-Strain Curve

The short-time uniaxial full-range tension test is used to determine the mechanical
properties of structural metals. Information obtained from this test can be used to
calculate the fracture properties of structural metals. As shown in Fig. 6.3, a typi-
cal stress-strain curve for a ductile metal shows that the metal undergoes initial
elastic straining, then uniform plastic straining and finally local plastic straining
(necking). The stress-strain curve in Fig. 6.3A is typical of brittle material, such
as glass. The curve in Fig. 6.3B is representative of metals that do not "neck,"
such as cross-rolled beryllium. The curve in Fig. 6.3C is typical of many structural
metals, and the curve in Fig. 6.3D, which shows very little uniform plastic strain-
ing, is typical of some precipitation hardening steels.
In a uniaxial tension test, strains are obtained from deflections measured by a
gage that is usually longer than the necked portion of the test specimen. The typ-
ical gage length employed for most uniaxial tension tests has a 2 in. length. This
average strain must be corrected to obtain the actual local strain in that part of the
test specimen. The correction can be made by multiplying the local average strain
by the length of the gage and dividing by the thickness, t" of the tensile test spec-
imen. The correction for the actual true "neck" strain and the procedure to obtain
the true "neck" strain, B pn ' is discussed in Appendix B.
Typical full-range uniaxial tensile stress-strain curves should be obtained from
reputable and reliable sources. If such curves are not available, five or six fuIl-
range tensile stress-strain curves should be obtained by tests. Either flat or round
test specimens are suitable. A typical stress-strain curve should be developed from
the test data by the strain departure method, which is outlined in MIL-HDBK-5,
Metallic Materials and Elements for Aerospace Vehicle structures.

6.9 Verification of FMDM Results

with the Experimental Data

The following are the calculated residual strength capability data provided by the
FMDM theory for several aerospace alloys compared with test data gathered from
several reliable sources. Both cases of finite and infinite cracked plates were used
to check the analysis. In Chapter 3, the ASTM procedures for determination of
fracture toughness, including standard specimen preparation, surface finish, and
310 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

prefatigue cracking of the specimen, as well as data gathering, are briefly dis-
cussed. In Section 6.10 the FMDM approach for determination of fracture tough-
ness for 2219-T87 and 7075-T73 is discussed and the calculated results compared
with the test data obtained through the ASTM procedure.
To employ the FMDM theory for constructing the residual strength diagram
(fracture stress as a function of half critical crack length), it is necessary to have
the full range stress-strain curve for the material under study. The full range
stress-strain curve must be obtained from reliable sources. The results of the
analysis conducted by the FMDM theory indicated exceptionally good agreement
with experimental data. In Fig. 6.7, the stress-strain curve for 2219-T87 alu-
minum alloy (with plate thickness ts = 0.125 in.) is plotted from MIL-HDBK-5.
The pertinent information is extracted as input to the FMDM equation to obtain
the variation of fracture stress as a function of half critical crack length, as shown
in Fig. 6.8. The experimental data plotted in Fig. 6.8 were extracted from Refer-
ence [6]. An excellent correlation between the test data and the FMDM results
was observed. As another example, the HP-9NI-4CO-0.2C Steel [8] was selected
and proper information that is needed for generating a residual strength diagram
was taken from the stress-strain curve shown in Fig. 6.9. Figure 6.10 is the con-
structed residual strength capability curve which describes the variation of the
fracture stress as a function of half critical crack length. Good agreement between

-
70

60

SO /I
]l Stress-Strain Curve, 2219-T87 I

,,
IS
40
Aluminum Alloy (t-O.l 25 Inch) I
.,; I
...
III
:!
en 30
I

,,
I
20 I

,,
10
I

0 I
0.02 0.04 0.06 0.08 0.10 0.12
Strain, E, in.lin.

Figure 6.7 The stress-strain curve for 2219-T87 aluminum alloy

 6.9 Verification of FMDM Results with the Experimental Data 311

200 Fracture Stress, 2219- Tal AI uminum All OJ

:;.......- UI ti mate Stress

iii
.><
Infi ni te wi dth
ti 60

~ ' ..., ...

50

..
vi
u
40
............
' ......
'" , .....
30
5l
.
"\J

'a
20 Width =6.0 Inch ..I '" '" ,
=12.0 inch
Y,-\,'"
~'"
Q.
<I( Width
Wi dth =24.0 inch

0.02 0.03 0.04 0.1 0.2 0.3 0.4 1.0 2 3 4

Critical Hal f Crack Length, c, inch

Figure 6.8 Residual strength diagram for 2219-T87 aluminum alloy

200

180
Forging, Room Temp.
RA·0.667

160

140

120

I
:fll00 ~
o "
"
"
/I"
60 "
'I
"
40 "
'I
I I
Stress-Strain Curve, HP 9Ni-4CO-O.20C I I
20 Steel I I
I I
Strain, E, in/in -> I I
I I
0.0 0.02 0.04 0.06 0.08 0.10 0.12 0.14

Figure 6.9 The stress-strain curve for HP-9NI-4CO-O.2C steel

312 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

Fracture Stress, H.P. 9N-4c()'0.2OC Steel

300
Ultimate Strength
II" Infinite Width
~~--------------~--

c 100
,jj
!
Iii
"II
Finite Width (4.0 inch)
i 400 • Test Data
~

0.02 0.03 0.04 0.05 0.1 0.2 0.3 0.4 0.5 1.0 2.0
Initial Half Crack Length, C, Inch.

Figure 6.10 Residual strength diagram for HP-9NI-4CO-0.2C steel

the test data and the FMDM result can be seen. Additional data that can validate
the results of the FMDM analysis with the test data are available in Reference [6].

6.10 Fracture Toughness Computation

by the FMDM Theory

6.10.1 Introduction
As mentioned previously, plane stress fracture toughness testing is costly and time
consuming because of its dependency on the material thickness and the crack
length.This dependency necessitates having standard specimens with different
thicknesses that must be prepared based on the ASTM procedures. The specimen
preparation for fracture toughness testing includes the machine starter slot, fatigue
cracking of the specimens prior to testing, method of applying load to the speci-
men, load recording, and crack length measurements, all of which are elements
that can contribute to a costly and time-consuming process. The following sce-
nario, which can often happen in the aircraft and aerospace industries, clearly
illustrates the need for knowing the variation in fracture toughness as a function of
 6.10 Fracture Toughness Computation by the FMDM Theory 313

material thickness. For example, consider a situation in which the fracture

mechanics analysis of a component of an aircraft structure shows that a part can-
not survive the load varying environment. Thus, there is a need to increase the
part thickness, tl ,in order to reduce the applied stress, (T. A new fracture mechan-
ics analysis of the part with increased thickness (t2 > t l ) requires a corrected frac-
ture toughness value that corresponds to the new assigned thickness, t 2 • With the
FMDM theory, the Kc value can easily be computed for the material, provided
that the material stress-strain curve from reliable sources, such as MIL-HDBK-5,
is available to the analyst.
Equations (6.62) and (6.63) provide the variation in fracture stress, (T, with
respect to the crack length, c. Incorporating the fracture stress, (T, and half critical
crack length, c, into the crack tip stress intensity factor equation (for a center crack
panel in an infinite plate or plate of finite width, W), the obtained K value will
correspond to the apparent fracture toughness. The stress intensity factor equation
for a cracked plate having thickness t and width W can be written as:

Kc = (T( 1TC sec (1TcIW)) 1/2 where (T < (TYield (6.67)

The analyst must remember that the computed value of the fracture stress from
the FMDM theory should not fall above the material yield value nor is net-sec-
tion yielding allowed when using Eq. (6.67) for the fracture toughness computa-
tion. That is:

«(TJnet-section = (Te [W I(W - 2c)] < (TYield (6.68)

Sections 6.10.2 and 6.10.3 contain example problems pertaining to computa-

tion of the fracture toughness by the FMDM theory [9, 10] for 2219-T87 and
7075-T73 aluminum alloys which have stress-strain curves taken directly from
MIL-HDBK-5.

6.10.2 Fracture Toughness, Kc ' Evaluation

for 2219- T87 Aluminum Alloy
The stress-strain curve for 2219-T87 aluminum alloy (sheet and plate, longitudi-
nal direction, room temperature) is shown in Fig. 6.7. The FMDM residual
strength capability data generated by the FMDM Computer code [see Eqs. (6.62)
and (6.63)] is plotted in Figs. 6.11 through 6.14 for plate thicknesses of t = 0.1,
0.2,0.4, and 0.8 in., respectively. Using Eq. (6.67) the variation of the calculated
fracture toughness, K e , with respect to material thickness (t = 0.1, 0.2, 0.4, 0.8,
1.0,1.4, and 2.0 in.) can be plotted (Fig. 6.15). The FMDM residual strength capa-
bility diagrams for other thicknesses (t = 1.0, 1.4, and 2.0 in.) are not shown here.
Because fracture toughness is a function of crack length, two arbitrary values
of crack length were chosen. One is associated with a fracture stress (Te = 90% to
314 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

100

ZZl9-T87 (t- 0.1 Inch)

w- WIdth of T1Ie Plate

1+-------------~------------~------------_1------------~.
0.01 0.1 1 10 100
Half Crack Length, Inch

Figure 6.11 The residual strength capability diagram for 2219·T87 generated by the
FMDM computer code (thickness t = 0.1 in.)

100

:=
1,0
i
<
Z219-T87 (t- 0.2 Inch)

W - Width of Th. Plet.

1+----..........+-...............~..............~............~
0.01 0.1 10 100
Half Crack Length, inch

Figure 6.12 The residual strength capability diagram for 2219·T87 generated by the
FMDM computer code (thickness t = 0.2 in.)

95% of CTYield' and the other one with half crack length c = 1.5 in. Two other inter·
mediate crack lengths were also used and the corresponding fracture toughnesses
are shown in Fig. 6.l5. The reader should remember that the larger the original
crack length, the higher is the computed fracture toughness value, Kc. Therefore,
the minimum fracture toughness is expected to be associated with a half critical
crack length having fracture stress of CTc = 90% to 95% of CTYield. A minimum
fracture toughness value, corresponding to the minimum crack length, where
 6.10 Fracture Toughness Computation by the FMDM Theory 315

100

Inllnlte
Plate

=-!
r
10 W.ZO
~
~
I
-<
2219-T87 (t. 0.4 Inch)
W-6 W_IZ

W • Width of The Plate

1+--------P------~~------~------~
0.01 0.1 10 100
Half Crack Length, inch

Figure 6.13 The residual strength capability diagram for 2219-T87 generated by the
FMDM computer code (thickness t = 0.4 in.)

100

Infinite
Plate

'iii
..><
iii
III

~
II)
10 W- 20

"C
.S!
C.
c. Wc6 W-12
-< 2219-T87 (t = 0.8 inch)

W = Width of The Plate

1+---------~------~--------_r------~
0.01 0.1 10 100
Half Crack Length, inch

Figure 6.14 The residual strength capability diagram for 2219-T87 generated by the
FMDM computer code (thickness t = 0.8 in.)
316 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

80
70 • • 2219-T87 Aluminum Alloy
! "• •.. • • •
• • • • .• "• •.. •"• •• •"• •
60 A-

•
u
50
'" A- 0

~
• • • • • • • • ••
40 A- &

.g,
::>
30
B 20
~
;l 10
g
ol:: 0
0.5 1.5 2

Material Thickness, t (Inch)

• -+ Minimum Crack Length is associated with applied stress of

90 to 95% of the material yield.

e, .... -+ Intermediate Crack Length • -+ Max Crack Length
(2c =3.0 inches)

Figure 6,15 Variation of material fracture toughness vs. thickness for 2219-T87 alu-
minum alloy

eTc = 0.9 to .95 X eTYield' gives a conservative result when conducting crack
growth rate analysis of a structural component. In most real cases, the initial crack
length assumption in the structure is small, so that the minimum fracture tough-
ness is an acceptable value to use in the analysis. Also note that the fracture tough-
ness values shown in Appendix A [11] are the lower bound Kc values based on
minimum initial crack size. The fracture stress corresponding to the minimum
crack size should not cause plastic deformation in the bulk of the test specimen.
Figure 6.16 shows the variation of the experimentally obtained fracture tough-
ness, K c ' versus material thickness for 2219-T87 aluminum alloy taken from Ref-
erence [11]. Fracture toughness curves shown in Fig. 6.16 correspond to half
crack lengths c = 0.24 to 0.38, 0.45 to 0.68, 1.0 to 1.2, and 1.5 to 1.8 in. Both fig-
ures (6.15 and 6.16) clearly illustrate that the fracture toughness value increases
with decreasing thickness and, moreover, that Kc is a function of crack length, c.

6.10.3 Fracture Toughness, Kc ' Evaluation

for 7075- T73 Aluminum Alloy
The pertinent information extracted from the 7075-T73 aluminum alloy (sheet and
plate, longitudinal direction at room temperature) stress-strain curve and it is
shown in Fig. 6.17. Figure 6.17 is a portion of the FMDM computer output that
contains information which can be used to construct the FMDM residual strength
capability curve. Using the FMDM computer code the residual strength diagrams
were plotted in Figs. 6.18 through 6.21 for different plate thicknesses t = 0.1, 0.2,
 6.10 Fracture Toughness Computation by the FMDM Theory 317

100 ~-----------------------------------------,

80 o ...o.241DQ.38

... • a-().4S III 0.68

~l.oo ID 1.20
....:•....
&
.". • a-1.501DL60
:::: •... i/' • .:.... .
C 60 .1' .~..--..-...............~
i=
II:
.' • •••• Ii'.
_ _.......ca:::::::::................:~.:~:.::
~
!.... 40
o

20
Kc 27
YIIId • 58
ID 0.54
IJ( 1
III 1
o+-________~__--------~__------~--------~
o 0.5 1.0

Material Thickness, t (inch)

Figure 6.16 Variation of Kc as a function of thickness, t, for 2219-T87 aluminum alloy

[10]

0.4, and 0.8 in., respectively. Equation (6.67) was utilized to calculate the mater-
ial fracture toughness and to establish the variation of Kc versus material thick-
nesses (t = 0.1,0.2,0.4,0.8, 1.0, 1.4, and 2.0 in.) for different crack lengths, 2c
(Fig. 6.22). No attempt was made to include all the residual strength capability
diagrams for the thicknesses that were used to generate Fig. 6.22.
A minimum fracture toughness value corresponding to the minimum crack
length is assigned to a fracture stress equal to 90% to 95% of the material yield
strength. Also, a half crack length c = 1.5 in. was selected as the upper limit of the
fracture toughness (two other intermediate crack lengths were also used and the cor-
responding fracture toughnesses are shown in Fig. 6.22). Note that the upper limit of
the fracture toughness was arbitrarily chosen to be associated with a half critical
crack length, c = 1.5 in. A higher fracture toughness value for this material can be
obtained by using a larger crack length. The fracture toughness dependency of 7075-
T73 aluminum alloy on the crack length and the material thickness is also shown in
Fig. 6.23, in which data points were obtained through laboratory tests conducted in
accordance with the ASTM standards. Both figures (6.22 and 6.23) clearly illustrate
that the fracture toughness is dependent on the material thickness (decreases as
thickness increases) and the crack length, c (increases as crack length increases).
318 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

Aluminum Alloy, 7075-T73, Plate, L-T Oir., Room Temp.

The following mechanical properties were obtained from flat tensile test
bars.

Atom. Spacing. un. Str. Yld. Str. RUp. Str. NeckStr. E

2.860 83.000 73.000 75.500 80.900 10300.00

Polsn. R. un. Strn. Rup. Strn. Thk. Tst. Sp. Gagel.

0.3300 0.0880 0.12 0.0625 2.000

Computed Basic Data

to WF EPN STU STF n

0.146 68.82 0.850 90.037 152.18 17.89

Atomic spacing is in Angstroms; stresses, modulus and WF are in ksi; to,

thicknesses, gage length, widths and crack sizes are in inches.

Neck Str. is the average stress from the beginning of necking to rupture of
the tensile test specimen.

Ult. Stm. is the strain at the beginning of necking.

to is the maximum thickness of a through cracked test specimen for plane-

stress fracture.

WF is the density of plastic energy under the stress-strain curve from strain
at the beginning of necking to strain at rupture.

EPN is the corrected neck uniaxial plastic tensile strain. (corrected from
gage length).

STU is the uniaxial true ultimate tensile stress.

STF is the uniaxial true rupture tensile stress.

n is the exponent in the Ramberg-Osgood relation for uniaxial true plastic

tensile strain.

Str. is the computed gross area fracture stress.

Figure 6.17 Portion of the FMDM computer output for 7073-T73

 6.10 Fracture Toughness Computation by the FMDM Theory 319

Thickness t - 0.1 Inch

707S-T63
100

.....
ui
1/1
OJ 10
Vl
"C
.'!!
a.a.
«
0.01 0.1 10 100
Half Crack Length, inch

Figure 6.18 The residual strength capability diagram for 7075-T63 generated by the
FMDM computer code (thickness t = 0.1 in.)

Thickness t -.2 Inch

100

W - 16.0 inch.

1 +-------------~------------~-----------_+-----------~
0.01 0.1 10 100
Ha lf Crack Length, Inch

Figure 6.19 The residual strength capability diagram for 7075-T63 generated by the
FMDM computer code (thickness t = 0.2 in.)
320 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

Thlcknen: t _.4 Inch

100

1 +-------~----~------~----~
0.01 0.1 10 100
Hllf cntk length, Inth

Figure 6.20 The residual strength capability diagram for 7075-T63 generated by the
FMDM computer code (thickness t = 0.4 in.)

Thiem .... t -.8 Ineh

100

1
i
10

a
Q.
«
1
0.01 0.1 10 100
Hllf Cl'lCk length, Inch

Figure 6.21 The residual strength capability diagram for 7075-T63 generated by the
FMDM computer code (thickness t = 0.8 in.)

• •
120

10
• • • •
...
... • • • •
•
u
><

...• •
10

.
= • • ... ... •...
a
c:
• ...
"
{!.
lO
• • • •
~ 40
1j
I!
u..
20

0
0 02 0.4 0.1 0.1 I.Z

Material Thlc kn.... t (I nchel)

• - Minimum Crack length is assocIaled with appUed stress of

90 to 95% of the malerial yield.
• , ... - Intennedlate Crack leng1h • - Max Crack length
(2<>3.0 Inches)

Figure 6.22 Variation of material fracture toughness versus thickness for 7075-T63
aluminum alloy
 12B I Se.al .... Iyp.
• PTSC
o CCP
• CCP l.t."
:r CCP Id."dN
lBB • 4 CT
o 0 V CT Id."dn'
o NB
... 4-NB
0
• CANT
BB-t o SENT
~ 0 0
c 0
0 r:P
0 0 0
t=-
IC 0
e
~ 6B
--
I.
u .. B
~ v v
Klc
~---II
ta 7075-T73 Aluminum Alloy,l-T, 75F

B.6 B.B 1.B 1.2 1. .. 1.6 loB 2.9

Material Thickness, t (Inches)

Figure 6.23 Variation of Kc as a function of thickness, t, for 7075-T73 aluminum alloy

U>
[10]
I\)
......
322 Chap. 6 The Fracture Mechanics of Ductile Metals Theory

References

1. A. A. Griffith, "The Phenomena of Rupture and Flow in Solids," Philos. Trans., R.

Soc. Lond., Ser. A., Vol. 221, 1920, p. 163.
2. G. R.lrwin, "Fracture Dynamics," Fracture of Metals, ASM, 1948, pp. 147-166.
3. E. Orowan, "Fracture and Strength of Solids," Rep. Prog. Physics, Vol. 12, 1949,
pp. 185-232.
4. A. H. Cottrell, Dislocations and Plastic Flow in Crystals, The Claredon Press, 1958, p. 3.
5. H. Conrad, "Experimental Evaluation of Creep and Stress Rupture," Chapter 8, in
Mechanical Behavior of Materials at Elevated Temperature, edited by John E. Dorn,
McGraw-Hill, 1961.
6. G. E. Bockrath and J. B. Glassco, "A Theory of Ductile Fracture," McDonnell Dou-
glas Astronautics Company, Report MDC G2895, August 1972, revised April 1974.
7. A. Nadai, Plasticity. McGraw-Hill, 1960, pp. 103,260.
8. J. B. Glassco, B-1 Fracture Analysis by Ductile Fracture Method, Rockwell Interna-
tional Report TFD-76-1368-L, 1977.
9. B. Farahmand and G. E. Bockrath, "A Theoretical Approach for Evaluating the Plane
Strain Fracture Toughness," Engin. J. Fract. Mech., March 1996.
10. B. Farahmand, "A New Analytical Approach to Obtain the Plane Stress Fracture
Toughness Value by the FMDM Theory," The Fourth Pan American Congress of
Applied Mechanics, Buenos Aires, Argentina, January 5-9,1994.
11. Fatigue Crack Growth Computer Program "NASA/FLAGRO," developed by R. G.
Forman, V. Shivakumar, and J. C. Newman. JSC-22267 A, January 1993.
Appendix A

NASAIFLAGRO 2.0 Materials

Constants U. S. Customary
Units (ksi, ksi Yin)

323
 324 Appendix A NASAIFLAGRO 2.0 Materials Constants

Material; Condition; Environment* Code YS UTS

[A] Iron, alloy or cast

ASTM Specification
A536 Grd 80-55-06
As cast Al AC50AB1 58 80

[B] ASTM spec. grd. steel

AIO Series
A36
Pit (Dyn Klc, < 500Hz); LA, HHA, 3% NaCl BOCBlOABl 44 78
ES Weld & HAZ (Dyn Klc, < 500Hz); LA, HHA, 3% NaCl BOCZKlABl 44 78
A200 Series
A203 Grd E (3.5% N1)
Pit B2CE12ABl 71 80
PIt; -IOOF B2CE12LA7 82 96
A216Grd WCC
Casting B2GC5lABl 48 80
A300 Series
A302 Grd B
Pit B3AB12ABl 55 90
A372 Type IV
Forg B3GD2lABl 65 105
A387 Grd 22, CI 2
Pit B3IQ1OABI 50 75
A400 Series
A469 CI4
Forg B4JD26ABl 85 105
A469 CI5
Forg B4JE20ABl 95 115
ASOO Series
A508 CI2 & CI3
Forg B5AC2lABl 65 100
A514Typ F
Pit B5BFlOABl 105 120
GMA SR Weld B5BFC2ABl 100 115
A517 Grd F (Tl Steel)
PIt B5DF12ABl 100 125
A533-B, CIl & Cl2
Pit B5HDlOABl 70 100
SMA Weld B5HDFlABl 60 90
A553 Typ I
Pit B5QA12ABI 95 110
Pit; -320F B5QA12LA4 150 175

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 325

(continued)

n p

45 32 0.75 0.50 .700E-9 2.900 0.5 0.5 8.0 0.7 2.5 0.3

100 70 0.75 0.5 .100E-8 3.000 0.5 0.5 7.0 0.7 2.0 0.3
100 70 0.75 0.5 .100E-8 3.000 0.5 0.5 7.0 0.7 2.0 0.3

230 170 0.75 0.5 .915E-9 2.688 0.5 0.5 7.0 0.7 2.5 0.3
280 200 0.75 0.5 .915E-9 2.688 0.5 0.7 7.0 0.7 2.5 0.3

200 150 0.75 0.5 .300E-9 3.000 0.5 0.5 7.0 0.7 2.5 0.3

120 100 0.75 0.5 .600E-1O 3.480 0.5 0.5 7.0 0.7 2.5 0.3

120 100 0.75 0.5 .600E-1O 3.480 0.5 0.5 6.0 0.7 2.5 0.3

120 100 0.75 0.5 .400E-9 3.000 0.5 0.5 7.0 0.7 2.5 0.3

230 170 0.75 0.5 .900E-9 2.800 1.0 0.5 5.5 0.7 2.5 0.3

230 170 0.75 0.5 .100E-8 2.800 0.5 0.5 4.3 0.7 2.5 0.3

140 100 0.75 0.5 .100E-8 2.800 0.5 0.5 6.0 0.7 2.5 0.3

115 85 0.75 0.5 .200E-8 2.570 0.25 0.25 4.0 0.7 2.5 0.3
115 85 0.75 0.5 .150E-8 2.570 0.25 0.25 4.0 0.7 2.5 0.3

140 100 0.75 0.5 .100E-9 3.500 0.25 0.25 6.0 0.7 2.5 0.3

200 150 0.75 0.5 .100E-8 2.700 0.5 0.5 6.5 0.7 2.5 0.3
140 100 0.75 0.5 .100E-8 2.700 0.5 0.5 8.0 0.7 2.5 0.3

250 180 0.75 0.5 .l50E-7 2.000 0.25 0.25 4.5 0.7 2.5 0.3
140 100 0.75 0.5 .300E-9 3.200 0.25 0.25 6.0 0.7 2.5 0.3
 326 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[B] ASTM spec. grade steel

A500 Series
A579 Grd 75 (12% Nl)
Forg B5VW20ABI 180 190
A588 Grd A & Grd B
Pit B5XA11AB1 55 80
Pit; 3% NaCI, > 0.2 Hz B5XAllWBI 55 80
A645 (5% Nl)
Pit B6GA12ABI 75 105
Pit; -320F B6GAI2LA4 110 165

[C] AISI - SAE Steel

AISI lOxx -12xx Steel
Low Carbon 1005 - 1012
Hot rolled pit CIABllABl 25 45
Low Carbon 1015 -1025
Hot rolled pit C1BBllABl 30 58
AISI 43xx - 48xx Steel
4330VMOD
180-200 UTS; Pit & Forg C4BSIOABI 175 190
200-220 UTS; Pit & Forg C4BTlOABl 195 210
220-240 UTS; Pit & Forg C4BUlOABl 215 230
4340
160-180 UTS; Pit & Forg C4DC21AB1 155 170
180-200 UTS; Pit & Forg; HHA C4DD11AD1 175 190
180-200 UTS; Pit & Forg C4DD21AB1 175 190
200-220 UTS; Pit & Forg C4DE11AB1 195 210
200-220 UTS; Pit & Forg; -50F C4DEIILB7 200 220
220-240 UTS; Pit & Forg C4DFllABl 215 230
220-240 UTS; Pit & Forg; -50F C4DFI1LB7 225 240
240-280 UTS; Pit & Forg C4DGlIABl 240 260
240-280 UTS; Pit & Forg; -50F C4DG11LB7 250 270

[D] Misc. u. S. Spec. Grade Steel

SAE Spec. Steel
0030 Cast D5AC50ABl 44 72

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, Sol, CoR, Col, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 327

(continued)

125 90 0.75 0.5 .150E-7 2.000 0.25 0.25 3.0 0.7 2.5 0.3

140 100 0.75 0.5 .350E-9 3.300 0.5 0.5 8.5 0.7 2.5 0.3
140 100 0.75 0.5 .7ooE-9 3.300 0.5 0.5 8.5 0.7 2.5 0.3

250 180 0.75 0.5 . 170E-8 2.500 0.25 0.25 5.0 0.7 2.5 0.3
110 80 0.75 0.5 . 170E-9 3.500 0.25 0.25 7.0 0.7 2.5 0.3

100 70 0.75 0.5 .800E-1O 3.600 0.5 0.5 8.0 0.7 2.5 0.5

100 70 0.75 0.5 .800E-1O 3.600 0.5 0.5 8.0 0.7 2.5 0.5

150 110 0.75 0.5 . 130E-8 2.700 0.25 0.25 5.5 0.7 2.5 0.3
110 80 0.75 0.5 . 130E-8 2.700 0.25 0.25 4.5 0.7 2.5 0.3
85 65 0.75 0.75 . 130E-8 2.700 0.25 0.25 4.0 0.7 2.5 0.3

190 135 0.75 0.5 . 170E-8 2.700 0.25 0.25 6.0 0.7 2.5 0.3
155 110 0.75 0.5 .160E-8 2.700 0.25 0.25 5.5 0.7 2.5 0.3
155 110 0.75 0.5 .130E-8 2.700 0.25 0.25 5.5 0.7 2.5 0.3
110 80 0.75 0.5 . 130E-8 2.700 0.25 0.25 4.5 0.7 2.5 0.3
70 55 0.75 0.5 . 170E-8 2.700 0.25 0.25 4.5 0.7 2.5 0.3
85 65 0.75 0.75 . 130E-8 2.700 0.25 0.25 4.0 0.7 2.5 0.3
55 45 0.75 0.75 . 170E-8 2.700 0.25 0.25 4.0 0.7 2.5 0.3
65 55 0.75 1.0 .130E-8 2.700 0.25 0.25 3.5 0.7 2.8 0.3
45 40 0.75 1.0 . 170E-8 2.700 0.25 0.25 3.5 0.7 2.8 0.3

95 70 0.75 0.5 .2ooE-1O 4.000 0.25 0.25 8.0 0.7 2.5 0.3
 328 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[E] Trade/common name steel

Ultra High Strength Steel
18 Nt Maraging
250 Grd; Plt & Forg EIADIOABI 240 260
300 Grd; Pit & Forg EIAEIOABI 280 290
300M
270-300 UTS; Plt & Forg EIBF21ABI 240 285
AF1410
220-240 UTS; Pit & Forg; -65F EICCI2AA7 240 250
220-240 UTS; Pit & Forg; LA, HHA/DW > 1Hz EICCI2ABI 220 230
D6AC
220-240 UTS; Pit & Forg; Nom. KIc (70); -40F ElDDlOAA8 225 240
220-240 UTS; Plt & Forg; Nom. KIc (70) ElDDlOABI 215 230
220-240 UTS; Plt & Forg; Nom. KIc (70); HHA/DW > 0.1 Hz ElDDlOADl 215 230
220-240 UTS; Pit & Forg; High KIc (90) ElDJlOABI 215 230
HP-9-4-20
190-210 UTS; Plt & Forg; L-T, T-L; HHA, SW > I Hz ElEB23ABI 190 200
190-210 UTS; Plt & Forg; L-T, T-L; -65F ElEB23AC7 200 210
190-210 UTS; GTA Weld + SR; LA, HHA, SW > I Hz ElECB2WAI 185 195
190-210 UTS; GTA Weld + SR; -65F EIECB2AC7 195 205
HP-9-4-30
220-240 UTS; Pit & Forg; L-T, T-L; LA, HHA, SW > I Hz EIGC23ABI 205 230
220-240 UTS; Pit & Forg; L-T, T-L; -65F EIGC23AA7 215 240
220-240 UTS; Plt & Forg; L-T, T-L; 600F EIGC23AA14 165 195
HY-180 (10NI)
Plt & Forg ElIB13ABl 180 200
Pit & Forg; DW, ASW > 0.1 Hz ElIB13WAI 180 200
Plt & Forg; SW > 0.1 Hz ElIB13WBI 180 200
HY-TUF
220-240 UTS; VAR Forg ElJB23ABI 200 230
H-ll MOD
240-260 UTS; Plt & Forg EILE23ABI 215 250

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 329

(continued)

c n p q DKo Rei a S.R.

90 75 0.75 0.75 .350E-S 2.600 0.25 0.25 3.5 0.7 2.5 0.3
S5 70 0.75 0.75 .300E-S 2.600 0.25 0.25 3.0 0.7 2.S 0.3

65 55 0.75 1.0 .500E-S 2.460 0.25 0.25 3.0 0.7 2.S 0.3

150 110 0.75 .0.50 .600E-S 2.200 0.25 0.25 3.5 0.7 2.5 0.3
ISO 135 0.75 0.50 .100E-7 2.200 0.25 0.25 3.5 0.7 2.5 0.3

70 50 0.75 0.75 . 150E-S 2.570 0.25 0.25 4.0 0.7 2.5 0.3
100 70 0.75 0.75 .320E-S 2.570 0.25 0.25 4.0 0.7 2.5 0.3
100 70 0.75 0.75 .150E-7 2.570 0.25 0.25 4.0 0.7 2.5 0.3
120 90 0.75 0.75 .320E-S 2.570 0.25 0.25 4.0 0.7 2.5 0.3

135 110 0.75 0.50 .S20E-S 2.320 0.25 0.25 4.5 0.7 2.5 0.3
135 110 0.75 0.50 .500E-S 2.320 0.25 0.25 4.5 0.7 2.5 0.3
135 110 0.75 0.50 .S20E-S 2.320 0.25 0.25 4.5 0.7 2.5 0.3
135 110 0.75 0.50 .500E-S 2.320 0.25 0.25 4.5 0.7 2.5 0.3

115 90 0.75 0.50 .S20E-S 2.320 0.25 0.25 4.0 0.7 2.5 0.3
115 90 0.75 0.50 .S20E-S 2.320 0.25 0.25 4.0 0.7 2.5 0.3
110 S5 0.75 0.50 .S20E-S 2.320 0.25 0.50 4.0 0.7 2.5 0.3

200 150 0.75 0.50 .600E-S 2.300 0.25 0.25 5.0 0.7 2.5 0.3
200 150 0.75 0.50 . 120E-7 2.300 0.25 0.25 5.0 0.7 2.5 0.3
200 150 0.75 0.50 .600E-7 2.300 0.25 0.25 5.0 0.7 2.5 0.3

150 110 0.75 0.75 .350E-S 2.500 0.25 0.25 5.0 0.7 2.5 0.3

60 50 0.75 0.75 .150E-S 2.700 0.25 0.25 4.5 0.7 2.5 0.3
 330 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[E] Trade/Common Name Steel

Pressure Vessel/Piping
HY80
PIt E2AA13AB1 90 105
PIt; 3.5% NaCl/SW > 0.1 Hz E2AA13WB1 90 105
HY130
PIt E2CA13AB1 140 150
PIt; 3.5% NaCl/SW > 0.1 Hz E2CA13WB1 140 150
GMA Weld E2CAC1AB1 130 140
SMA Weld E2CAF1AB1 130 140
Construction Grade
HT-80
PIt E3BA13AB1 110 120
SA Weld E3BAHlABl 95 115

* Unless noted. assume Lab Air (LA) environment and any orientation except s-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 331

(continued)

c n p q

250 200 0.75 0.50 .150E-8 2.500 0.25 0.25 5.5 0.6 2.5 0.3
250 200 0.75 0.50 .300E-8 2.500 0.25 0.25 5.5 0.6 2.5 0.3

250 200 0.75 0.50 .300E-8 2.500 0.25 0.25 5.0 0.6 2.5 0.3
250 200 0.75 0.50 .700E-8 2.500 0.25 0.25 5.0 0.6 2.5 0.3
250 200 0.75 0.50 .150E-8 2.500 0.25 0.25 5.0 0.6 2.5 0.3
250 200 0.75 0.50 .150E-8 2.500 0.25 0.25 5.0 0.6 2.5 0.3

200 150 0.75 0.50 .700E-9 3.000 0.25 0.25 8.0 0.7 2.5 0.3
200 150 0.75 0.50 .250E-9 2.700 0.25 0.25 2.5 0.7 2.5 0.3
 332 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[F] AISI type stainless steel

AISI 300 Series
AISI 301/302
AnnPlt & Sht F3AA13ABI 40 90
1/2 Hard sht F3AC13ABI 125 165
Full Hard sht F3AE13ABI 190 205
AISI 304/304L
Ann PIt & Sht, Cast; 550F Air F3DA13AA13 24 64
Ann PIt & Sht, Cast; 800F Air, > 1Hz F3DA13AAI6 20 63
Ann Plt & Sht, Cast F3DA13ABI 40 90
Ann Plt & Sht, Cast; - 320F LN2 F3DA13LA4 100 205
SA weld (308 filler) + SR; 800F Air, > 1Hz F3FAH2AA16 49 69
SA weld (308 filler) + SR F3FAH2ABI 66 107
AISI316/316L
Ann PIt & Sht, Cast; 600F Air F3KA13AA14 32 60
Ann Plt & Sht, Cast; 800F Air F3KA13AAI6 20 60
Ann PIt & Sht, Cast F3KA13ABI 36 90
Ann PIt & Sht, Cast; -452F LHe F3KA13LA2 80 215
Ann Plt & Sht, Cast; - 320F LN2 F3KA13LA4 70 185
SMA weld (316 filler) + SR; 800F Air, > 1Hz F3KAH2AA16 49 69
20% CW Plt & Sht F3KB13ABI 100 110
AISI 400 Series
AISI430VAR
Ann Rnd, C-R F4LAI6ABI 34 59

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 333

(continued)

c n p q

280 200 1.0 0.50 .600E-9 3.0 0.25 0.25 3.5 0.7 2.5 0.3
140 100 1.0 0.50 .130E-8 3.0 0.25 0.25 3.5 0.7 2.5 0.3
110 80 1.0 0.50 .550E-8 2.2 0.25 0.25 3.5 0.7 2.5 0.3

200 150 1.0 0.50 .400E-1O 4.0 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.50 . 120E-9 4.0 0.25 0.25 7.0 0.7 2.5 0.3
280 200 1.0 0.50 .600E-9 3.0 0.25 0.25 3.5 0.7 2.5 0.3
280 200 1.0 0.50 .120E-9 3.2 0.25 0.25 7.0 0.7 2.5 0.3
140 100 1.0 0.5 .100E-9 4.0 0.25 0.25 8.0 0.7 2.5 0.3
200 150 1.0 0.5 .580E-I0 3.68 0.25 0.25 6.0 0.7 2.5 0.3

200 150 1.0 0.5 .750E-1O 4.0 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.5 .100E-9 4.0 0.25 0.25 8.0 0.7 2.5 0.3
280 200 1.0 0.50 .800E-9 3.0 0.25 0.25 3.5 0.7 2.5 0.3
280 200 1.0 0.5 .220E-9 3.2 0.25 0.25 8.0 0.7 2.5 0.3
280 200 1.0 0.5 .220E-9 3.2 0.25 0.25 7.0 0.7 2.5 0.3
110 80 1.0 0.5 .260E-ll 5.0 0.25 0.25 8.0 0.7 2.5 0.3
140 100 1.0 0.50 .400E-9 3.0 0.25 0.25 5.0 0.7 2.5 0.3

114 80 1.0 0.5 .350E-1O 3.8 0.50 0.25 14.0 0.7 5.84 1.0
 334 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[G] Misc. CRES/heat resistant steel

PHxx-x Alloys
PH13-8 Mo
HlOOO; Plt, Forg, Extr GIADl3ABI 200 208
HlOOO; Plt & Forg; OW & SW, > 1Hz GIADl3WDI 200 208
Hl050; Pit & Forg GlAF13ABI 185 190
H1050; Plt & Forg; OW & SW, > O.1Hz GlAF13WDl 185 190
xx-xPH Alloys
15-5PH
H900; Rnd, CoR G2AB16ABI 170 190
H1025; Rnd, CoR G2ADl6ABI 165 175
Hl025; Forg G2A023ABI 155 165
HI100; Rnd, CoR G2AF16ABI 145 155
17-4PH
H900; Pit, L-T G2CBllAB1 170 195
H900; Plt, T-L G2CB12ABI 170 195
Hl050; Pit G2CE13ABI 155 160
H1025; Rnd, CoL G2CE19ABI 160 163
H1025; Cast; HHA G2CE50ADl 160 163
H 1100; Plt; HHA G2CH13ADl 145 150
17-7PH
TH1050; Pit G2EH13ABI 170 190
AMxxx Alloys
AM 350
CRT; Sht, L-T G4AHIIABI 185 205
AM 367
SCT (850); Sht G4FCIIABI 240 243
Custom xxx Alloys
Custom 455
H1000; Pit & Forg G5B023ABI 195 205
Hl025; Forg, CoR GSBE26ABI 185 195
Nitronic xx Alloys
Nitronic 33
Ann; Pit G7AA13ABI 64 115
Ann; Pit; -452F LHe G7AA13LA2 220 260
Ann; Plt; - 320F LN2 G7AAI3LA4 165 220
Nitronic 50
Ann;Plt G7CAI3ABI 77 120
Ann; Plt; -452F LHe G7CA13LA2 210 275
Ann; Pit; - 320F LN2 G7CAI 3LA4 170 230
Nitronic 60
HR, CR; Rnd Rod G70CI8ABI 137 192

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 335

(continued)

c n p q

140 100 0.75 0.75 .250E-S 2.620 0.25 0.25 5.0 0.6 2.5 0.3
140 100 0.75 0.75 .350E-S 2.S00 0.50 0.50 6.0 0.6 2.5 0.3
160 115 0.75 0.75 .2ooE-S 2.620 0.25 0.25 5.0 0.6 2.5 0.3
160 115 0.75 0.75 .250E-9 4.100 0.50 0.50 7.0 0.4 2.5 0.3

65 50 1.0 0.5 .7S0E-1O 4.000 0.25 0.25 5.0 0.7 2.5 0.3
SO 60 1.0 0.5 .2ooE-9 3.600 0.25 0.25 5.0 0.7 2.5 0.3
150 110 1.0 0.5 .S30E-9 2.S70 0.25 0.25 5.0 0.7 2.5 0.3
110 SO 1.0 0.5 .450E-9 3.100 0.25 0.25 5.0 0.7 2.5 0.3

60 50 1.0 0.5 .600E-9 3.110 0.25 0.25 4.0 0.7 2.5 0.3
55 45 1.0 0.5 .600E-9 3.400 0.25 0.25 4.0 0.7 2.5 0.3
SO 60 1.0 0.5 .340E-9 3.420 0.25 0.25 4.0 0.7 2.5 0.3
70 55 1.0 0.5 .150E-9 3.500 0.25 0.25 4.0 0.7 2.5 0.3
70 55 1.0 0.5 .400E-9 3.500 0.25 0.25 4.0 0.7 2.5 0.3
120 90 1.0 0.5 .250E-S 2.470 0.25 0.25 4.0 0.7 2.5 0.3

60 50 1.0 1.0 .350E-9 3.400 0.25 0.25 3.0 0.7 2.5 0.3

120 90 1.0 1.5 .130E-S 2.500 0.25 0.25 5.0 0.7 2.5 0.3

80 65 1.0 1.0 .350E-8 2.150 0.25 0.25 4.0 0.7 0.7 0.3

140 100 1.0 0.5 .300E-S 2.440 0.25 0.25 4.0 0.7 2.5 0.3
150 110 1.0 0.5 .91OE-9 2.760 0.25 0.25 4.2 0.7 2.5 0.3

2S0 200 1.0 0.5 .150E-1O 3.S50 0.25 0.25 9.0 0.7 2.5 0.3
SO 65 1.0 0.5 .700E-ll 5.000 0.25 0.25 9.0 0.7 2.5 0.3
150 110 1.0 0.5 .500E-1O 4.000 0.25 0.25 9.0 0.7 2.5 0.3

250 ISO 1.0 0.5 .200E-10 3.S00 0.25 0.25 9.0 0.7 2.5 0.3
120 90 1.0 0.5 .230E-11 4.300 0.25 0.25 9.0 0.7 2.5 0.3
150 110 1.0 0.5 .170E-1O 3.650 0.25 0.25 9.0 0.7 2.5 0.3

S4 60 0.75 0.5 .475E-S 2.100 0.50 0.50 3.1 0.5 5.S5 1.0
 336 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[H] High temperature steel

Nickel Chromium
A286 (140 ksi)
Pit & Sht; 600-800F HIAB13AAI5 95 138
Pit & Sht HIAB13ABI 100 140
Forg, L-T, T-L, L-R HIAB23ABI 100 140
A286 (160 ksi)
Pit & Sht; 600-800F HIAC13AA15 95 138
Pit & Sht HIAC13ABI 105 160
Forg. rod, L-R HIAC28ABI 120 160
A286 (200 ksi Bolt Material)
Forg. rod, L-R HIAD28ABI 190 200
JBK-7S
ST-CR-A; Pit, T-L HlCB12ABI 150 180

[J] Tool Steel

AISI Tool Steel
M-SO
61-63 Rc; Pit J1IKlOABI 325 375
T1 (18-4-1)
60-63 Rc; Pit J1MAlOABI 325 350

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, Sol, CoR, Col, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 337

(continued)

c n p q DKo ReI a SR.

110 80 1.0 0.5 .150E-8 2.930 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.5 .500E-9 3.000 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.5 .100E-9 3.300 0.25 0.25 6.0 0.7 2.5 0.3

110 80 1.0 0.5 .120E-8 3.000 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.5 .260E-9 3.200 0.25 0.25 6.0 0.7 2.5 0.3
140 100 1.0 0.5 . 177E-8 2.450 0.25 0.25 4.5 0.5 3.0 0.3

140 100 1.0 0.5 .300E-8 2.l00 0.25 0.25 3.5 0.2 3.0 0.3

120 90 1.0 0.5 .200E-8 2.400 0.25 0.25 8.5 0.7 2.5 0.3

13.5 13 1.0 0.5 .l40E-8 3.180 0.1 0.1 2.4 0.7 3.0 0.3

15.5 15 1.0 0.5 .350E-9 3.800 0.1 0.1 2.4 0.7 3.0 0.3
 338 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[M] 1000-9000 Series aluminum

2000 Series
2014-T6
Pit & Sht; L-T M2ADllABl 65 74
Pit & Sht; T-L M2AD12ABl 63 71
2014-T651
PIt & Sht; L-T M2AFllABl 64 71
Pit & Sht; T-L M2AF12ABl 64 71
Pit & Sht; OTA Weld M2AFBlABl 24 47
Pit & Sht; OTA Weld, SR M2AFB2ABl 14 27
2020-T651
Pit & Sht; L-T M2CBllABl 77 82
Pit & Sht; T-L M2CB12ABl 78 82
2024-T3
Clad, PIt & Sht; L-T; LA & HHA M2EAllABl 53 66
Clad, PIt & Sht; T-L; LA & HHA M2EA12ABl 48 65
Clad, PIt & Sht; L-T; DW M2EAllWAl 53 66
Clad, PIt & Sht; T -L; DW M2EA12WAl 48 65
2024-T351
PIt & Sht; L-T; 300F to 400F Air M2EBllAAll 52 66
Pit & Sht; L-T; LA & HHA M2EBllABl 54 68
Pit & Sht; T-L; LA & HHA M2EB12ABl 52 68
2024-T3511
Extr; L-T; LA & HHA M2EC3lABl 55 77
2024-T62
Pit & Sht; L-T; LA, HHA & ASW M2EGllABl 58 66
PIt & Sht; T-L; LA, HHA & ASW M2EGl2ABl 57 66
2024-T81
Pit & Sht; L-T; 350F Air M2EIllAAll 52 61
PIt & Sht; L-T M2EIllABl 63 75
Pit & Sht; L-T; DA M2EIllACl 63 70
Pit & Sht; L-T; HHA M2EIllADl 63 70
PIt & Sht; T-L; 350F Air M2EI12AAll 52 65
Pit & Sht; T-L M2EI12ABl 62 68
2024-T851
Pit & Sht; T -L; 300F to 350F Air M2EJ12AAll 56 61
Pit & Sht; LA, DA, IP-4 M2EJ13ABl 64 74
Pit & Sht; 3.5% NaCl M2EJ13WBl 64 7C

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 339

(continued)

c n p q

38 27 1.0 1.0 .350E-7 2.800 0.5 1.0 2.7 0.7 1.5 0.3
23 18 1.0 1.0 .350E-7 2.800 0.5 1.0 2.7 0.7 1.5 0.3

28 22 1.0 1.0 .150E-7 2.800 0.5 1.0 2.7 0.7 1.5 0.3
27 20 1.0 1.0 .150E-7 2.800 0.5 1.0 2.7 0.7 1.5 0.3
22.4 16 1.0 1.0 .646E-8 3.918 0.5 1.0 9.5 0.7 1.5 0.3
22.4 16 1.0 1.0 . 115E-8 5.005 0.5 0.5 4.0 0.7 1.5 0.3

29 22.5 1.0 1.0 .3lOE-8 3.695 0.5 0.5 2.2 0.7 1.5 0.3
21 17 1.0 1.0 .135E-7 3.074 0.5 0.5 2.2 0.7 1.5 0.3

46 33 1.0 1.0 .829E-8 3.284 0.5 1.0 2.9 0.7 1.5 0.3
41 29 1.0 1.0 .244E-7 2.601 0.5 1.0 2.9 0.7 1.5 0.3
46 33 1.0 1.0 .169E-7 3.090 0.5 1.0 2.9 0.7 1.5 0.3
41 29 1.0 1.0 .892E-8 3.282 0.5 1.0 2.9 0.7 1.5 0.3

46 33 1.0 1.0 .334E-7 2.956 0.5 1.0 2.6 0.7 1.5 0.3
48 34 1.0 1.0 .922E-8 3.353 0.5 1.0 2.6 0.7 1.5 0.3
41 29 1.0 1.0 .922E-8 3.353 0.5 1.0 2.6 0.7 1.5 0.3

35 25 1.0 1.0 .200E-7 2.700 0.5 1.0 2.9 0.7 1.5 0.3

50 36 1.0 1.0 .100E-7 3.200 0.5 1.0 2.9 0.7 1.5 0.3
42 30 1.0 1.0 .100E-7 3.200 0.5 1.0 2.9 0.7 1.5 0.3

35 25 1.0 1.0 .105E-6 2.761 0.5 1.0 2.9 0.7 1.5 0.3
30 22 1.0 1.0 .804E-7 2.763 0.5 1.0 2.8 0.7 1.5 0.3
29.7 22 1.0 1.0 .217E-7 2.890 0.5 1.0 2.8 0.7 1.5 0.3
29.7 22 1.0 1.0 .863E-7 2.566 0.5 1.0 2.8 0.7 1.5 0.3
32 23 1.0 1.0 .340E-7 3.255 0.5 1.0 2.9 0.7 1.5 0.3
28.1 21 1.0 1.0 .219E-7 3.313 0.5 1.0 2.8 0.7 1.5 0.3

34 24 1.0 1.0 .243E-8 4.308 0.5 1.0 2.8 0.7 1.5 0.3
31 23 1.0 1.0 . 150E-7 3.100 0.5 1.0 2.8 0.7 1.5 0.3
31.3 23 1.0 1.0 .566E-7 2.480 0.5 1.0 2.8 0.7 1.5 0.3
 340 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[M] 1000-9000 Series aluminum

2000 Series
2024-T852
Forg;LA&DA M2EK23ABl 57 70
2024-T861
PIt & Sht; L-T; 300F to 400F Air M2ELlIAAll 56 58
PIt & Sht; L-T; LA & HHA M2ELlIABl 73 76
PIt & Sht; T-L M2EL1 2AB1 72 76
2048-T851
PIt & Sht; L-T; LA, DA M2FCIIABl 63 69
PIt & Sht; T-L; LA, DA M2FC1 2AB1 62 69
2124-T851
Pit & Sht; L-T; 120F to 350F Air M2GCllAAlO 53 55
PIt & Sht; L-T; LA, DA, HHA M2GC11 AB1 63 71
PIt & Sht; T-L; 300F - 400F Air M2GC 12AA 11 53 55
Pit & Sht; T-L; LA, HHA M2GC1 2AB1 64 72
PIt & Sht; T-L; -200F to -150F GN2 M2GC12GB6 72 80
PIt & Sht; SoT, Sol; LA, HHA M2GC15ABl 60 69
2219-T62
Pit & Sht; L-T M2IA11 AB1 43 61
Pit & Sht; L-T & T-L; 350F Air M2IA 13AA 11 37 46
Plt & Sht; T-L M2IA12ABI 43 61
PIt & Sht; L-T & T-L; -320F LN2 M2IA13LA4 51 76
2219-T851
PIt & Sht; L-T, LA, DA M2IC11 AB1 53 65
PIt & Sht; T-L; LA, DA M2IC1 2AB1 50 66
2219-T87
PIt & Sht; L-T; 300F to 350F Air M2IFllAAlI 43 49
Pit & Sht; L-T M2IFllABI 57 68
PIt & Sht; L-T; - 320F LN2 M2IFIILA4 68 83
Plt & Sht; T-L; 300F to 350F Air M2IF12AAll 44 49
PIt & Sht; T-L M2IF12ABl 58 69
Pit & Sht; T-L; -320F LN2 M2IF12LA4 66 85
Pit & Sht; GTA weld, PAR M2IFBIABl 20 42
PIt & Sht; GTA weld, PAR; - 320F LN2 M2IFBILA4 27 46
2324-T39
PIt & Sht; L-T M2JA11 AB1 65 72
2090-T8E41
PIt & Sht; L-T M2PAllABl 80 85

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 341

(continued)

C n p q a S.R.

39.2 28 1.0 1.0 .666E-7 2.060 0.5 1.0 2.9 0.7 1.5 0.3

50 36 1.0 1.0 .283E-7 3.181 0.5 1.0 2.8 0.7 1.5 0.3
30 23 1.0 1.0 .280E-7 2.920 0.5 1.0 2.2 0.7 1.5 0.3
24 19 1.0 1.0 .309E-7 2.895 0.5 1.0 2.2 0.7 1.5 0.3

49 35 1.0 1.0 .386E-8 3.460 0.5 1.0 2.7 0.7 1.5 0.3
42 30 1.0 1.0 .203E-8 3.951 0.5 1.0 2.7 0.7 1.5 0.3

43 31 1.0 1.0 .162E-7 3.233 0.5 0.5 2.9 0.7 1.5 0.3
42 30 1.0 1.0 .14IE-7 3.146 0.5 0.5 2.7 0.7 1.5 0.3
36 26 1.0 1.0 .452E-8 3.715 0.5 1.0 2.9 0.7 1.5 0.3
31 23 1.0 1.0 .119E-7 3.061 0.5 1.0 2.7 0.7 1.5 0.3
35 26 1.0 1.0 .200E-8 3.400 0.5 1.0 2.2 0.7 1.5 0.3
28 21 1.0 1.0 .259E-8 3.867 0.5 1.0 2.7 0.7 1.5 0.3

43 31 1.0 1.0 .377E-7 2.657 0.5 1.0 3.6 0.7 1.5 0.3
42 30 1.0 1.0 .458E-7 2.611 0.5 1.0 3.8 0.7 1.5 0.3
41 29 1.0 1.0 .420E-7 2.648 0.5 1.0 3.6 0.7 1.5 0.3
42 30 1.0 1.0 .257E-8 3.178 0.5 1.0 2.9 0.7 1.5 0.3

46 33 1.0 1.0 .119E-7 3.156 0.5 1.0 3.0 0.7 1.5 0.3
43 31 1.0 1.0 .307E-7 2.863 0.5 1.0 2.1 0.7 1.5 0.3

39 28 1.0 1.0 .293E-7 2.533 0.5 1.0 3.5 0.7 1.5 0.3
42 30 1.0 1.0 .572E-7 2.487 0.5 1.0 2.9 0.7 1.5 0.3
57 41 1.0 1.0 .617E-8 3.151 0.5 1.0 2.3 0.7 1.5 0.3
38 27 1.0 1.0 .269E-8 3.903 0.5 1.0 3.5 0.7 1.5 0.3
38 27 1.0 1.0 .700E-8 3.400 0.5 1.0 2.9 0.7 1.5 0.3
46 33 1.0 1.0 .450E-1O 5.279 0.5 1.0 2.3 0.7 1.5 0.3
50 28 1.0 1.0 .640E-8 4.060 0.5 1.0 4.0 0.7 1.5 0.3
38 27 1.0 1.0 .304E-1O 6.140 0.5 1.0 4.0 0.7 1.5 0.3

55 39 1.0 1.0 .831 E-8 3.072 0.5 1.0 2.9 0.7 1.5 0.3

46 33 1.0 1.0 .485E-8 3.267 0.5 1.0 4.0 0.7 1.5 0.3
 342 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[M] 1000-9000 Series aluminum

5000 Series
5083-0
Pit; T-L M5BA12ABI 20 43
6000 Series
6061-T6
Pit; T-L M6AB13AB1 41 45
PIt; GTA weld, PAR M6ABAIABI 23 26
6061-T651
PIt; 300F Air M6AC13AA10 36 37
PIt M6AC13ABI 44 47
6063-T5
Pit & Sht; T-L; LA M6BA12ABI 21 27
7000 Series
7005-T6 & T63
Pit & Sht; L-T M7BAllABI 48 53
PIt & Sht; T-L M7BA12ABI 48 53
7010-T73651
PIt & Sht; L-T & L-S M7DA11 AB1 64 .73
7050-T73511
Extr; L-T; LA, HHA, DA M7GE31ABI 72 80
7050-T736 & T74
Forg; L-T M7GI21ABI 65 72
Forg; T-L M7GI22ABI 62 72
7050-T73651 & T7451
Pit & Sht; L-T; LA & HHA M7GJ11ABI 66 77
Pit & Sht; L-T; DA M7GJ11ACI 66 77
PIt & Sht; T-L; LA & HHA M7GJ12ABI 65 77
PIt & Sht; T-L; DA M7GJ12ACI 65 77
PIt & Sht; S-T M7GJ15ABI 61 75
7050-T74511
Extr; L-T; LA & DW M7GL31ABI 70 79
7050-T73652 & T7452
Forg; L-T M7GM21ABI 70 79
Forg; T-L M7GM22ABI 70 79
7050-T7651
Pit & Sht; L-T; LA & HHA M7GQ11 AB1 75 80
PIt & Sht; T-L M7GQ12ABI 75 80
7050-T76511
Extr; L-T; LA & HHA M7GS31ABI 79 87
Extr; T-L M7GS32ABI 79 87

* Unless noted, assume Lab Air (LA) enviromnent and any orientation except S-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 343

(continued)

q DKo ReI a SR.

63 45 1.0 0.1 .249E-6 1.938 0.5 1.0 4.5 0.7 1.5 0.3

36 26 1.0 0.75 .900E-7 2.300 0.5 0.5 3.5 0.7 2.0 0.3
36 26 1.0 1.0 .113E-7 3.524 0.5 1.0 4.5 0.7 1.5 0.3

38 27 1.0 1.0 .288E-8 3.884 0.5 1.0 4.0 0.7 1.5 0.3
38 27 1.0 1.0 .133E-6 2.248 0.5 1.0 3.5 0.7 1.5 0.3

34 24 1.0 0.75 .946E-8 3.447 0.5 1.0 3.5 0.7 1.9 0.3

64 46 1.0 1.0 .990E-7 2.248 0.5 1.0 3.4 0.7 1.75 0.3
56 40 1.0 1.0 .839E-7 2.313 0.5 1.0 3.4 0.7 1.75 0.3

43 31 1.0 1.0 .929E-8 3.122 0.5 1.0 2.5 0.7 1.75 0.3

49 35 1.0 1.0 . 133E-7 2.908 0.5 1.0 2.1 0.7 1.9 0.3

46 33 1.0 1.0 .615E-8 3.368 0.5 1.0 2.3 0.7 1.9 0.3
33 24 1.0 1.0 .213E-7 2.677 0.5 1.0 2.3 0.7 1.9 0.3

43 31 1.0 1.0 .462E-7 2.379 0.5 1.0 2.4 0.7 1.9 0.3
43 31 1.0 1.0 .363E-7 2.084 0.5 1.0 2.4 0.7 1.9 0.3
35 25 1.0 1.0 .246E-7 2.738 0.5 1.0 2.4 0.7 1.9 0.3
35 25 1.0 1.0 .660E-7 2.865 0.5 1.0 2.4 0.7 1.9 0.3
33 24 1.0 1.0 .132E-7 2.674 0.5 1.0 2.8 0.7 1.9 0.3

50 36 1.0 1.0 .279E-7 2.682 0.5 1.0 2.2 0.7 1.9 0.3

43 31 1.0 1.0 .266E-7 2.654 0.5 1.0 2.4 0.7 1.9 0.3
27 21 1.0 1.0 .266E-7 2.654 0.5 1.0 2.4 0.7 1.9 0.3

43 31 1.0 1.0 .446E-7 2.412 0.5 1.0 2.0 0.7 1.9 0.3
38 28 1.0 1.0 .446E-7 2.412 0.5 1.0 2.0 0.7 1.9 0.3

41 30 1.0 1.0 .502E-7 2.329 0.5 1.0 2.0 0.7 1.9 0.3
32 24 1.0 1.0 .502E-7 2.329 0.5 1.0 2.0 0.7 1.9 0.3
 344 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS un

[M] 1000·9000 Series aluminum

7000 Series
7075-t6
Pit, Sht & Clad; LA M7HA13ABl 75 8L
Pit, Sht & Clad; HHA M7HA1 3ADl 75 8L
7075-T651
PIt & Sht; L-T; LA, DA M7HBllAB1 76 8:
PIt & Sht; L-T; HHA M7HBIIADl 76 8:
Pit & Sht; L-T; 3.5% NaCI M7HBIIWBl 76 8:
Pit & Sht; T-L; DW M7HB12WAl 76 8:
PIt & Sht; S-T M7HB15ABl 66 7:
7075-T6510
Extr; L-T; LA & DA M7HC31ABl 79 8~
7075-T6511
Extr; L-T; LA & HHA M7HD31AB1 80 8'
Extr; T-L M7HD32AB1 80 8'
7075-T73
PIt & Sht; L-T; LA & HHA M7HGllABl 60 7'
Plt & Sht; T-L M7HG12ABl 61 7
7075-T7351
Plt & Sht; L-T M7HHl1AB1 62 7
Pit & Sht; L-T; DA M7HHIIACI 62 7
Pit & Sht; L-T; HHA M7HHllADI 62 7
Pit & Sht; T-L; LA & DA M7HH12AB1 60 7
Pit & Sht; S-T; LA M7HH15ABl 58 6:
7075-T73510
Extr; L-T; LA M7HI31AB1 64 7
7075-T73511
Extr; L-T; LA, DA, HHA M7HJ31AB1 65 7·
7075-T7352
PIt, Sht & Forg; L-T; LA & DA M7HKllABl 59 6
Pit, Sht & Forg; T-L; LA & DA M7HK12ABl 53 6
7075-T7651
Pit & Sht; L-T; LA & DA M7HMIIABl 66 7
Pit & Sht; T-L; LA & DA M7HM12ABl 66 7
7079-T651
Plt & Sht; L-T M7ICllABl 75 8

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 345

(continued)

37 27 1.0 1.0 .209E-7 2.947 0.5 1.0 2.0 0.7 1.9 0.3
37 27 1.0 1.0 .145E-6 2.497 0.5 1.0 2.0 0.7 1.9 0.3

38 28 1.0 1.0 .233E-7 2.885 0.5 1.0 2.0 0.7 1.9 0.3
38 28 1.0 1.0 . 147E-7 3.985 0.5 1.0 2.5 0.7 1.9 0.3
38 28 1.0 1.0 .339E-6 2.135 0.5 1.0 2.0 0.7 1.9 0.3
32 24 1.0 1.0 .191E-6 1.917 0.5 1.0 2.0 0.7 1.9 0.3
23 18 1.0 1.0 .578E-7 2.435 0.5 1.0 2.3 0.7 1.9 0.3

38 28 1.0 1.0 .184E-6 1.869 0.5 1.0 1.8 0.7 1.9 0.3

38 28 1.0 1.0 .609E-7 2.324 0.5 1.0 1.8 0.7 1.9 0.3
32 24 1.0 1.0 .609E-7 2.324 0.5 1.0 1.8 0.7 1.9 0.3

50 36 1.0 1.0 .149E-7 3.321 0.5 1.0 2.7 0.7 1.9 0.3
32 23 1.0 1.0 .252E-7 2.908 0.5 1.0 2.7 0.7 1.9 0.3

41 29 1.0 1.0 .348E-7 2.529 0.5 1.0 2.6 0.7 1.9 0.3
41 29 1.0 1.0 .120E-7 2.994 0.5 1.0 2.6 0.7 1.9 0.3
41 29 1.0 1.0 .655E-8 3.696 0.5 1.0 2.6 0.7 1.9 0.3
35 25 1.0 1.0 .397E-8 3.652 0.5 1.0 2.6 0.7 1.9 0.3
25 19 1.0 1.0 .200E-7 2.300 0.5 1.0 2.6 0.7 1.9 0.3

43 31 1.0 1.0 .138E-6 1.918 0.5 1.0 2.4 0.7 1.9 0.3

46 33 1.0 1.0 .347E-7 2.508 0.5 1.0 2.4 0.7 1.9 0.3

46 33 1.0 1.0 .150E-7 3.100 0.5 1.0 2.7 0.7 1.9 0.3
35 25 1.0 1.0 .150E-7 3.100 0.5 1.0 2.7 0.7 1.9 0.3

45 32 1.0 1.0 .552E-7 2.305 0.5 1.0 2.4 0.7 1.9 0.3
32 23 1.0 1.0 .654E-8 3.338 0.5 1.0 2.4 0.7 1.9 0.3

36 26 1.0 1.0 .212E-6 2.101 0.5 1.0 2.0 0.7 1.9 0.3
 346 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UT!

[M] 1000-9000 Series aluminum

7000 Series
7149-T73511
Extr; L-T; LA M7NA3lAB1 66 71
Extr; T-L; LA M7NA32AB1 63 7·
7178-T6 & T651
PIt & Sht; L-T; LA & HHA M7RA11 AB1 84 8'
Pit & Sht; T-L; LA & HHA M7RA12AB1 79 8
7178-T7651
PIt & Sht; L-T M7RF11AB1 72 8
PIt & Sht; T-L M7RF12AB1 70 7
7475-T61
PIt, Sht, Clad; L-T; LA, DA, HHA M7TBllAB1 74 7
Pit, Sht, Clad; T-L; LA, DA, HHA M7TB12AB1 71 7
7475-T651
PIt & Sht; L-T; LA, DA, HHA M7TD11AB1 75 8
7475-T7351
PIt & Sht; L-T; LA, DA, HHA, DW M7TFllAB1 63 7
PIt & Sht; T-L; LA, DA, HHA M7TF1 2AB1 57 7
7475-T7651
PIt & Sht; L-T; LA, DA, HHA, DW, 3.5% NaCI M7TJ1IABI 69 7

[0] Misc. and cast aluminum

300 Series cast
A356·T60
Cast 03FB50ABI 31 ~

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, SoL, CoR, CoL, and R-L.
 Appendix A NASNFLAGRO 2.0 Materials Constants 347

(continued)

C n p q DKo RcJ a S.R.

43 31 1.0 1.0 .278E-6 1.544 0.5 1.0 2.4 0.7 1.9 0.3
33 24 1.0 1.0 .352E-6 1.435 0.5 1.0 2.4 0.7 1.9 0.3

31 24 1.0 1.0 .156E-6 2.180 0.5 1.0 2.3 0.7 1.9 0.3
27 21 1.0 1.0 .405E-7 2.736 0.5 1.0 1.8 0.7 1.9 0.3

39 28 1.0 1.0 .140E-6 1.800 0.5 1.0 2.1 0.7 1.9 0.3
31 23 1.0 1.0 . 140E-6 1.800 0.5 1.0 2.1 0.7 1.9 0.3

43 31 1.0 1.0 .898E-7 2.211 0.5 1.0 2.1 0.7 1.9 0.3
36 26 1.0 1.0 .360E-6 1.496 0.5 1.0 2.1 0.7 1.9 0.3

52 37 1.0 1.0 .232E-6 2.328 0.5 1.0 3.0 0.7 1.9 0.3

60 43 1.0 1.0 .175E-7 2.877 0.5 1.0 2.4 0.7 1.9 0.3
49 35 1.0 1.0 .696E-7 2.212 0.5 1.0 2.8 0.7 1.9 0.3

56 40 1.0 1.0 .18IE-7 2.808 0.5 1.0 2.2 0.7 1.9 0.3

22 16 1.0 1.0 . 142E-9 4.980 0.5 1.0 6.3 0.7 1.9 0.3
 348 Appendix A NASNFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[P] Titanium alloys

Tl Unalloyed
Tl-SS
Plt & Sht P1AA13AB1 55 65
Pit & Sht; DW & SW P1AA12WA1 55 65
Tl-70
Pit & Sht P1CA13ABI 70 80
Pit & Sht; DW & SW P1CA13WA1 70 80
Binary Alloys
Tl-2.S Cu; STA
Sht; LA, HHA, DW P2AA13AB1 97 110
Ternary Alloys
Tl-S AI-2.SSn; Annealed
Sht; LA, HHA, DW P3CA13ABI 120 130
Tl-S AI-2.SSn (ELI); Annealed
Forg P3CB23ABI 115 120
Forg; -423F LH2 P3CB23LA3 200 210
Tl-3 AI-2.SV; CW, SR (7S0F)
Extr P3DB33ABI 105 125
Tl-6 AI-4V (MA)
Plt & Sht, -lOOF P3EA13AA7 165 170
Pit & Sht P3EA13AB1 138 146
Forg P3EA23AB1 135 145
Extr P3EA33AB1 125 140
Tl-6 AI-4V; BA (1900F/.Sh + 132SF/2h)
Pit & Sht; LA, DA, 3.5% NaCI P3EB12AB1 120 135
Forg; LA, DA, HHA, 3.5% NaCl P3EB23AB1 110 125
Tl-6 AI-4V; RA
Sht; L-T; LA, DA, HHA, DW, 3.5% NaCI P3EC11AB1 140 150
Sht; T-L; LA, DA, HHA, DW, 3.5% NaCI P3EC12AB1 140 150
Pit; -IOOF P3EC13AA7 160 175
Plt; LA, DA, HHA, DW P3EC13ABI 125 135
Forg; LA, DA, HHA, 3.5% NaCI P3EC23AB1 115 130
Tl-6 AI-4V; ST (17S0F) + A (lOOOF/4h)
Pit & Sht; SR (l000F/4h) P3ED13AA1 155 167
Plt & Sht; SR (l000F/8h) P3EDl3ABI 140 150
Plt & Sht; SR (l000F/4h); -320F LN2 P3EDl3LA4 230 240
Forg; SR (1000F/4h) P3ED20ABI 150 163
Forg; SR (1000F/4h); -320F LN2 P3ED20LA4 225 235

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, SoL, CoR, CoL, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 349

(continued)

n p q DKo Rei a SR.

70 50 1.0 0.5 .542E-9 3.670 0.5 0.5 5.0 0.7 2.5 0.3
70 50 1.0 0.5 .100E-S 3.670 0.5 0.5 5.0 0.7 2.5 0.3

70 50 1.0 0.5 .210E-S 3.1S0 0.5 0.5 5.0 0.7 2.5 0.3
70 50 1.0 0.5 .300E-S 3.1S0 0.5 0.5 5.0 0.7 2.5 0.3

70 50 1.0 0.5 .750E-S 2.750 0.5 0.5 5.0 0.7 2.5 0.3

90 65 1.0 1.0 .120E-7 2.500 0.5 0.5 5.0 0.4 2.5 0.3

90 65 1.0 1.0 .900E-S 2.600 0.5 0.5 5.0 0.4 2.5 0.3
SO 60 1.0 0.5 .150E-S 2.S50 0.5 0.5 5.0 0.4 2.5 0.3

70 50 1.0 0.5 .300E-9 4.000 0.25 0.75 4.5 0.7 2.5 0.3

65 50 1.0 0.5 .100E-S 3.300 0.25 0.75 3.5 0.7 2.5 0.3
65 50 1.0 0.5 .252E-S 3.010 0.25 0.75 3.5 0.7 2.5 0.3
65 50 1.0 0.5 .311 E-9 3.667 0.25 0.75 3.5 0.7 2.5 0.3
75 60 1.0 0.5 .147E-9 3.S34 0.25 0.75 5.0 0.7 2.5 0.3

100 80 1.0 0.5 .900E-9 3.250 0.25 0.75 3.5 0.7 2.5 0.3
100 SO 1.0 0.5 .600E-9 3.250 0.25 0.75 6.0 0.7 2.5 0.3

SO 60 1.0 0.5 .150E-S 3.300 0.25 0.75 3.5 0.7 2.5 0.3
SO 60 1.0 0.5 .250E-8 3.300 0.25 0.75 3.5 0.7 2.5 0.3
S5 65 1.0 0.5 .120E-9 4.200 0.50 0.75 5.0 0.7 2.5 0.3
100 75 1.0 0.5 .250E-S 3.000 0.50 0.75 6.0 0.7 2.5 0.3
100 75 1.0 0.5 .500E-9 3.300 0.50 0.75 S.O 0.7 2.5 0.3

55 45 0.5 1.0 .500E-S 2.S50 0.25 0.75 3.5 0.5 2.5 0.3
55 45 0.5 1.0 .200E-S 3.000 0.25 0.75 3.5 0.5 2.5 0.3
44 3S 1.0 0.5 .4OOE-S 2.750 0.25 0.75 3.5 0.5 2.5 0.3
50 42 0.75 0.75 . 150E-S 3.200 0.25 0.75 3.5 0.5 2.5 0.3
47 40 0.5 0.75 .220E-9 3.300 0.25 0.75 3.5 0.5 2.5 0.3
 350 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[P] Titanium alloys

GTA Weld; SR; thk < 0.2" P3EDB2AB1A 125 135
GTA Weld; SR; thk >= 0.2" P3EDB2AB1B 125 135
Tl·6 AI·4V; ELI; BA (l900F/.5h) + (1325F/2h)
Pit & sht; LA, 3.5% NaCI P3ELl2ABI 115 127
Tl·6 AI·4V (ELI) RA
Pit P3EM13AB1 120 130
Forg; -IOOF P3EM23AA7 145 155
Forg P3EM23AB1 120 130
Forg; -452F LHe P3EM23LA2 240 248
Forg; - 320F LN2 P3EM23LM 200 213
Forg; EB welded, SR; weldine P3EMD2AB1 120 130
Forg; EB welded, SR; weldine; - 320F LN2 P3EMD2LM 200 213
Forg; EB welded, SR; HAZ P3EMD8AB1 120 130
Forg; EB welded, SR; HAZ; - 320F LN2 P3EMD8LM 200 213
Quaternary Alloys
Tl·4.5 AI·5 Mo·1.5 Cr
Pit; LA, 3.5% NaCI P4BAlOABl 170 182
Tl·S AI·l Mo·lV
Sht P4CBllABI 138 150
Tl·6 AI·6V·2Sn MA
Pit, Forg, Extr; LA, DA, HHA, DW P4DA33ABI 150 160
Tl·6 AI·6V·2Sn RA
Pit P4DB12ABI 150 160
Tl·6 AI·6V·2Sn BA
Pit P4DC12AB1 140 155
Tl·6 AI·6V·2Sn ST (l600F); A (lOOOF/6h)
Forg; -65F P4DD21AA7 205 212
Forg; 300F P4DD21AAlO 165 172
Forg; LA, DA, HHA P4DD21AB1 180 190
Tl·lOV·2Fe·3AI
STA (140-160 UTS, 70Klc) PIt & Forg P4MD23AB1 140 150
STA (160-180 UTS, 60Klc) PIt & Forg P4MF23AB1 153 170
STA (180-200 UTS, 40Klc) Pit & Forg P4MG13AB1 178 190
STA (180-200 UTS, 30Klc) Pit & Forg P4MG20ABI 178 190
STA (180-200 UTS, 25Klc)Forg P4MG23AB1 178 190
Tl-6A; ·2Zn·2Sn·2Mo·2Cr (ST or STA)
Pit; HHA P5FBllADl 155 165

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 351

(continued)

K1e KIc Ak Bk C n p q DKo Rei a S.R.

50 40 1.0 0.5 .SOOE-S 2.5S0 0.25 0.25 3.5 0.7 2.5 0.3
70 55 1.0 0.5 .SOOE-S 2.5S0 0.25 0.25 3.5 0.7 2.5 0.3

100 SO 1.0 0.5 .900E-9 3.250 0.25 0.75 3.5 0.7 2.5 0.3

95 75 1.0 0.5 .250E-S 3.000 0.50 0.75 6.0 0.7 2.5 0.3
95 75 1.0 0.5 .440E-9 3.300 0.50 0.75 6.0 0.7 2.5 0.3
95 75 1.0 0.5 .l50E-S 3.000 0.50 0.75 S.O 0.7 2.5 0.3
60 50 1.0 0.5 .500E-9 3.S00 0.50 0.50 5.0 0.7 2.5 0.3
65 55 1.0 0.5 .l40E-S 3.000 0.50 0.50 6.0 0.7 2.5 0.3
90 65 1.0 0.5 .lOOE-7 2.000 0.50 0.75 5.0 0.7 2.5 0.3
65 55 1.0 0.5 .SOOE-S 2.000 0.50 0.50 5.0 0.7 2.5 0.3
90 65 1.0 0.5 .l20E-7 2.000 0.50 0.75 5.0 0.7 2.5 0.3
65 55 1.0 0.5 .350E-S 2.500 0.50 0.50 5.0 0.7 2.5 0.3

57 45 1.0 0.5 .650E-S 2.650 0.25 0.75 3.5 0.7 2.5 0.3

77 55 1.0 0.5 .500E-S 2.500 0.25 0.75 3.5 0.7 2.5 0.3

65 50 1.0 0.5 .600E-S 2.700 0.25 0.75 3.5 0.7 2.5 0.3

91 65 1.0 0.5 .l50E-7 2.400 0.25 0.75 3.5 0.7 2.5 0.3

84 60 1.0 0.5 . 161E-7 2.207 0.25 0.75 3.5 0.7 2.5 0.3

32 2S 1.0 0.5 .1 62E-S 3.021 0.25 0.75 3.5 0.7 2.5 0.3
70 53 1.0 0.5 .370E-S 2.635 0.25 0.75 3.5 0.7 2.5 0.3
35 30 1.0 0.5 .736E-S 2.279 0.25 0.75 3.5 0.7 2.5 0.3

95 70 1.0 0.5 . 120E-7 2.500 0.25 0.25 2.0 0.7 2.5 0.3
SO 60 1.0 0.5 .200E-7 2.500 0.25 0.25 2.0 0.7 2.5 0.3
4S 40 1.0 0.5 .200E-7 2.500 0.25 0.25 2.0 0.7 2.5 0.3
35 30 1.0 0.5 .200E-7 2.500 0.25 0.25 2.0 0.7 2.5 0.3
2S 25 1.0 0.5 .300E-7 2.500 0.25 0.25 2.0 0.7 2.5 0.3

75 55 1.0 0.5 .1 25E-7 2.500 0.25 0.75 3.5 0.7 2.5 0.3
 352 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[Q] NI Alloys/superalloys
Hastelloy Alloys
Hastelloy B
RndRnd QIAAI6ABI 60 127
Hastelloy X-280; ST (2150F)
Pit; 600-800F Air QIQA1OAA15 39 91
PIt; 1000-1200F Air; >.67Hz QIQA1OAA19 33 84
Pit QIQAIOABI 53 109
Inconel Alloys
Inconel600
Pit & Sht; WOOF Q3AB1OAA18 28 83
Pit & Sht; 75-800F Q3ABIOAB1 35 94
Inconel625
Pit & Sht; 600F Q3EA1OAA14 50 122
PIt & Sht; 800F Q3EA1OAA16 48 119
Pit & Sht; 1000F Q3EA1OAA18 48 119
Pit & Sht; Q3EA1OABI 66 133
Inconel 706; ST (1800-1950F); A (1375F/8h; 1150F/5-8h)
Forg & Extr Q3JB33ABI 145 177
Forg & Extr; -452F LHe Q3JB33LA2 177 230
ST Pit - GTA weld - STA Q3JBB3ABI 145 164
ST Pit - GTA weld - STA; -452F LHe Q3JBB3LA2 177 213
Inconel 718; ST (1700-1850F) + A (1325F/Sh 11150F/I0h)
Pit; 600F air, >.3Hz Q3LBllAA14 150 190
PIt; 800F air, > .3Hz Q3LBllAA16 142 180
Pit; 1000F air, >.3Hz Q3LB13AAI8 135 178
Sht (t <.25") Q3LB13ABIA 175 210
Pit Q3LB13ABlB 170 200
Forg Q3LB23ABI 165 190
Forg; 300F air, >.3Hz Q3LB26AA1O 160 187
Forg; 600F air, > .3Hz Q3LB26AA14 155 185
Forg; 800F air, > .3Hz Q3LB26AA16 147 173
Forg; 1000F air, > .3Hz Q3LB26AA18 140 167
GTA weld-STA; 600F air, > .6Hz Q3LBB3AA14 140 170
GTA weld-ST A; 800F air, > .6Hz Q3LBB3AA16 135 160
GTA weld-STA; WOOF air, >.6Hz Q3LBB3AA18 132 160
ST plt-GTA weld-aged Q3LBB3ABI 160 192
ST pIt-GTA weld-aged; - 320F LN2 Q3LBB3LA4 178 227
ST pIt-EB weld-aged Q3LBD3ABI 163 205
ST pIt-EB weld-aged; - 320F LN2 Q3LBD3LA4 193 224

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, Sol, CoR, Col, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 353

(continued)

c n p q

267 100 1.0 0.5 .170E-9 3.200 0.50 0.50 20.0 0.6 2.5 0.3

150 110 0.75 0.5 .250E-9 3.600 0.50 0.50 6.0 0.7 2.5 0.3
140 100 0.75 0.5 .400E-9 3.600 0.50 0.50 4.0 0.7 2.5 0.3
165 120 0.75 0.5 .100E-9 3.600 0.50 0.50 8.0 0.7 2.5 0.3

140 100 1.0 0.5 .450E-9 3.300 0.50 0.50 5.0 0.7 2.5 0.3
140 100 1.0 0.5 .658E-ll 4.647 0.50 0.50 8.0 0.7 2.5 0.3

125 90 1.0 0.5 .484E-1O 4.124 0.50 0.50 5.0 0.7 2.5 0.3
125 90 1.0 0.5 .623E-9 3.365 0.50 0.50 4.0 0.7 2.5 0.3
125 90 1.0 0.5 .475E-8 2.792 0.50 0.50 2.5 0.7 2.5 0.3
140 100 1.0 0.5 .954E-1O 3.745 0.50 0.50 5.0 0.7 2.5 0.3

120 85 0.75 0.5 .343E-9 3.069 0.50 0.50 12.0 0.65 2.5 0.3
147 105 0.75 0.5 .150E-8 2.144 0.50 0.50 16.0 0.30 5.84 1.0
55 45 0.75 0.5 .500E-9 3.069 0.50 0.50 6.0 0.65 2.5 0.3
65 50 0.75 0.5 .225E-1O 3.457 0.50 0.50 8.0 0.30 5.84 1.0

125 90 0.75 0.5 .332E-9 3.250 0.25 0.50 7.5 0.7 2.5 0.3
120 85 0.75 0.5 .265E-9 3.395 0.25 0.50 7.5 0.7 2.5 0.3
95 70 0.75 0.5 .21OE-8 2.775 0.25 0.50 7.0 0.7 2.5 0.3
120 85 0.75 0.5 .225E-9 3.250 0.25 0.50 8.0 0.7 2.5 0.3
125 90 0.75 0.5 .225E-9 3.250 0.25 0.50 8.0 0.7 2.5 0.3
125 90 0.75 0.5 .248E-9 3.295 0.25 0.50 6.0 0.7 2.5 0.3
125 90 0.75 0.5 .400E-1O 3.500 0.25 0.50 6.0 0.7 2.5 0.3
125 90 0.75 0.5 .158E-9 3.425 0.25 0.50 5.5 0.7 2.5 0.3
120 85 0.75 0.5 .376E-9 3.216 0.25 0.50 5.5 0.7 2.5 0.3
95 70 0.75 0.5 .329E-8 2.747 0.25 0.50 5.0 0.7 2.5 0.3
85 60 0.75 0.5 .199E-9 3.536 0.25 0.50 6.0 0.7 2.5 0.3
75 55 0.75 0.5 .629E-9 3.108 0.25 0.50 5.5 0.7 2.5 0.3
60 45 0.75 0.5 .916E-9 3.165 0.25 0.50 5.0 0.7 2.5 0.3
70 55 0.75 0.5 .392E-1O 3.954 0.25 0.50 6.5 0.7 2.5 0.3
70 55 0.75 0.5 .392E-1O 3.954 0.25 0.50 6.5 0.7 2.5 0.3
55 45 0.75 0.5 .109E-9 3.846 0.25 0.50 6.5 0.7 2.5 0.3
50 40 0.75 0.5 .109E-9 3.846 0.25 0.50 6.5 0.7 2.5 0.3
 354 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[Q] Ni Alloys/superalloys
Inconel 718; ST (1900F) + A (1400F/10h + 1200F/10h)
Pit; - 320F LN2 Q3LCIOLA4 191 225
Pit; 600F air; > .6Hz Q3LCIIAA14 158 180
Pit Q3LC11ABI 165 205
Pit; l000F air; >.6Hz Q3LC12AA18 145 170
Inconel 718; ST (2000F) + A (1325F/4h + 1150F/16h)
Pit; 800F air; > .2Hz Q3LEllAA16 133 170
Pit; l000F air; >.6Hz Q3LE13AAI8 125 157
Pit Q3LE13ABI 145 187
GTA weld-STA; 600F air, >.6Hz Q3LEB3AA14 140 180
GTA we1d-STA; 800F air, >.6Hz Q3LEB3AA16 132 175
GTA weld-STA; l000F air, >.6Hz Q3LEB3AA18 129 167
GTA weld-STA Q3LEB3ABI 147 193
Inconel 718, Bolt material
185 ksi UTS BoIts Q3LP18ABI 180 205
225 ksi UTS Bolts Q3LQ18ABI 205 225
Inconel X-750; ST (2100F) + A (1550F/24h + 1300F/20h)
Pit; 600F air, >.6Hz Q3SDlOAA14 87 147
PIt; 800F air, > .6Hz Q3SDIOAA16 83 143
Pit; 1000F air, >.6Hz Q3SDlOAA18 81 134
Plt& Forg Q3SD26ABI 100 150
Forg; -452F LHe Q3SD26LA2 125 220
Rene and Udimet Alloys
Rene 41; ST (1950F) + A (1400F/16h)
Pit & Forg QQ7AD13ABI 138 184
Forg; 1100F air Q7AD26AA19 112 157
Forg; 1200F air Q7AD26AA20 110 155

* Unless noted, assume Lab Air (LA) environment and any orientation except S-T, S-L, C-R, C-L, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 355

(continued)

c n p q

125 90 0.75 0.5 .300E-9 2.900 0.25 0.50 8.0 0.7 2.5 0.3
125 90 0.75 0.5 .284E-9 3.234 0.25 0.50 7.5 0.7 2.5 0.3
125 90 0.75 0.5 .400E-9 2.900 0.25 0.50 8.0 0.7 2.5 0.3
100 70 0.75 0.5 .224E-8 2.829 0.25 0.50 7.0 0.7 2.5 0.3

120 85 0.75 0.5 .1 46E-8 2.733 0.25 0.50 7.5 0.7 2.5 0.3
100 70 0.75 0.5 .500E-8 2.251 0.25 0.50 7.0 0.7 2.5 0.3
125 90 0.75 0.5 .250E-9 2.900 0.25 0.50 8.0 0.7 2.5 0.3
85 60 0.75 0.5 .262E-8 2.386 0.25 0.50 6.0 0.7 2.5 0.3
75 55 0.75 0.5 .259E-8 2.452 0.25 0.50 5.5 0.7 2.5 0.3
70 50 0.75 0.5 .1 54E-7 1.984 0.25 0.50 5.0 0.7 2.5 0.3
85 60 0.75 0.5 .469E-9 2.844 0.25 0.50 6.5 0.7 2.5 0.3

100 70 0.75 0.5 .l3IE-9 3.626 0.25 0.50 4.0 0.7 2.5 0.3
70 55 0.75 0.5 .750E-9 2.838 0.25 0.50 3.0 0.7 2.5 0.3

70 50 0.75 0.5 .672E-1O 3.705 0.25 0.50 6.0 0.7 2.5 0.3
70 50 0.75 0.5 .200E-9 3.200 0.25 0.50 5.0 0.7 2.5 0.3
70 50 0.75 0.5 .1 14E-8 2.727 0.25 0.50 4.0 0.7 2.5 0.3
85 60 0.75 0.5 .102E-9 3.421 0.25 0.50 10.0 0.7 2.5 0.3
100 70 0.75 0.5 .207E-15 7.228 0.25 0.50 12.0 0.7 2.5 0.3

105 75 0.75 0.5 .306E-1O 3.847 0.25 0.50 8.0 0.7 2.5 0.3
84 60 0.75 0.5 .581E-9 3.384 0.25 0.50 6.0 0.7 2.5 0.3
77 55 0.75 0.5 .250E-7 2.150 0.25 0.50 6.0 0.7 2.5 0.3
 356 Appendix A NASAIFLAGRO 2.0 Materials Constants

(continued)

Material; Condition; Environment* Code YS UTS

[R] Misc. superalloys

Multiphase alloys
MP35N Rnd Rod R3AB18ABl 254 274

[S] Copper/Bronze alloys

Be-Cu Alloys
CDA 172
RndRod SOBA13ABl 159 179
C17510
Peak-aged Pit SlLBllABl 110 125
Peak-aged PIt; - 320F LN2 SlLBllLA4 120 141
Overaged PIt SlLCllABl 86 99
AI-Bronze Alloys
CDA 630 AI-Bronze Extf S6JB36AB1 82 117

[T] Magnesium alloys

AM 503 Pit TlAAllABl 16 29
AZ-31 B-H24 Pit TlDA12ABl 26 39
ZK-60 A-T5 Pit TlMA12ABl 39 50
ZW1 Pit TlNA11ABl 24 36
QE22A-T6 Pit T2LB13ABl 26 33

[U] Misc. non-ferrous alloys

Beryllium
Cross-rolled sht UlCA90ABl 55 70
Hot-pressed b1k UlCA93ABl 35 60
Columbium Alloys
C- 103 PIt UlCAlOABl 43 50
Zinc Alloys
Zn-4 AI-O.04Mg Die cast alloy No.3
As cast; l35-200F airO U4BA50AA9 25 34
As cast U4BA50AB1 30 41

* Unless noted, assume Lab Air (LA) environment and any orientation except SoT, Sol, CoR, Col, and R-L.
 Appendix A NASAIFLAGRO 2.0 Materials Constants 357

(continued)

n p q DKo Rei a SR.

105 80 1.0 0.5 .264E-8 2.280 0.25 0.25 3.0 0.7 2.5 0.3

27 26 0.35 0.5 .532E-1O 4.595 0.5 1.0 8.0 0.7 2.0 0.3

110 75 1.0 0.5 .315E-7 1.950 0.75 0.50 8.0 0.7 2.0 0.3
150 110 1.0 0.5 .404E-8 2.591 0.75 0.50 8.0 0.7 2.0 0.3
105 70 1.0 0.5 .900E-8 2.500 0.75 0.50 8.0 0.7 2.0 0.3

74 53 1.0 0.5 .123E-8 3.190 0.50 0.50 8.5 0.7 2.5 0.3

17 12 1.0 0.5 .304E-6 3.183 0.25 0.25 0.75 0.7 1.5 0.3
28 20 1.0 0.5 .294E-6 2.717 0.25 0.25 1.4 0.7 1.5 0.3
28 20 1.0 0.5 .292E-6 2.576 0.25 0.25 1.4 0.7 1.5 0.3
18 13 1.0 0.5 .294E-6 3.210 0.25 0.25 0.75 0.7 1.5 0.3
18 13 1.0 0.5 .294E-6 3.210 0.25 0.25 0.75 0.7 1.5 0.3

11 9.5 1.0 1.5 .200E-9 3.750 0.15 0.15 7.0 0.7 1.75 0.3
12.5 10 1.0 1.5 .500E-1O 5.000 0.15 0.15 8.5 0.7 1.75 0.3

42 30 1.0 1.0 .112E-7 2.322 0.5 0.5 10.0 0.7 2.0 0.3

20 14 1.0 0.5 .132E-6 2.812 0.25 0.5 1.5 0.7 1.25 0.3
20 14 1.0 0.5 .500E-7 3.700 0.25 0.5 2.5 0.7 1.25 0.3
Appendix B

Equations for Uniaxial True

Stress and True Strain

Typical engineering and true stress-strain curves are shown in Figure B-1. The
following equations describe the method of obtaining the true stress-strain para-
meters in terms of known quantities extracted from an engineering stress-starin
curve generated in the laboratory in accordance with the ASTM Standards.
The engineering or conventional stress is defined as

p
(J' =- (B.l)
Ao

where P is the tensile load on the specimen, and Ao is the initial cross-sectional
area of the test specimen. The true stress is defined as

P
(J' =- (B.2)
T A

where A is the cross-sectional area of the test specimen under the load P. After the
load is removed the area is Ap .
From elasticity theory:

A = Ap (1 - v i) 2
(B.3)

Substituting Eq. (B.3) in Eq. (B.2) gives

P
(BA)

358
 Appendix B Equations for Uniaxial True Stress and True Strain 359

Stress

Failure

o
<C
"c..
I/)

:3
"c..
<C
....
ti)
I/) C)
I/) C
CI.l ';: Failure
.... CI.l
+" CI.l
II) C Engineering Stress-Strain Curve
~ '0,
.... C
I- W

Strain

Engi neeri ng Strai n, tULo

True Strai n, In AotA

Figure B-1 A Typical Engineering and True Stress-Strain Curves

For plastic straining the volume remains constant. That is, the volume of the uni-
axial test specimen remains constant (area X length) during uniform plastic
deformation and as the specimen length increases (Lo + Mo). Its cross-sectional
area decreases uniformly up to the maximum load corresponding to the ultimate
of the material.

(B.5)

From which

(B.6)

where Lo and Lp are the lengths of the strained region before loading and after
removal of the load, respectively. Substituting Eq. (B .6) in Equation (B.4) gives,
360 Appendix B Equations for Uniaxial True Stress and True Strain

(B.7)

Substituting Eq. (B.l) in Eq. (B.7) gives

(B.8)
O"T = 0" ( O"T)2
1- v-
E

Plastic strain is defined as:

(B.9)

From which

(B.IO)

and

Lp
Lo = I + e p
~ (B.I1)

Substituting Eq. (B. II) in Eq. (B.8) gives

(B.12)
O"T = 0" ( O"T)2
1- v-
E

True strain is defined as:

e
T
= IL dLL (B.I3)
Lo

Integrating and substituting limits gives

eT = InL/Lo (B.14)
 Appendix B Equations for Uniaxial True Stress and True Strain 361

From Eq. (B.ll), (B.l4) becomes

er = In (1 + e) (B.15)

or

(B.16)

The total true strain is the sum of the elastic plus plastic true strain.

(B.l7)

Rearranging terms gives

(B.l8)

From Eq. (B.16)

(B.19)

Substituting Eq. (B.18) in Eq. (B.l9) gives

(B.20)

or

(B.2l)

Substituting Eq. (B.16) in Eq. (B.2l) gives

(B.22)

Substituting Eq. (B.22) in Eq. (B.12) gives

I +e
(B.23)
362 Appendix B Equations for Uniaxial True Stress and True Strain

The reader should note that true stress/true strain parameters can be obtained from
the engineering stress/strain curve using equations B-16 and B-23 prior to onset of
the necking strain I. Equations B-16 and B-23 are valid based on the assumption
of uniform distribution of strain along the gage length, L G , and the material vol-
ume due to the application of monotonic load remains constant as described by
equation B-5. Thus, equations B-16 and B-23 should be used prior to necking up
to the maximum load on the engineering stress-strain curve (points to the right of
ultimate strength of the material). Beyond the maximum load the true stress, uT '
should be calculated from the actual load and cross-sectional area (uT = PIA).
Furthermore, the true strain, eT , beyond the maximum load should be calculated
based on the actual diameter, D, (if the test specimen has a circular cross section)
or area,A (i.e. eT = InA/A o or 2 InDo/D).
To find plastic necking strain, the total deflection of the tensile specimen is
the sum of the deflections from uniform plastic straining, neck strain, and elas-
tic straining.

(B.24)

where eF is the total deflection of the tensile specimen divided by the initial gauge
length, L G , epu is the plastic strain at ultimate stress; ePN is the plastic necking
deflection divided by the effective initial length of the necked region, LNo ; and eEF
is the elastic strain at fracture. It is assumed that for thin tensile specimens with a
thickness, ts:

(B.25a)

and for specimens thicker than the maximum thickness for plane stress, to'

(B.25b)

Substituting Eq. (B.25) into Eq. (B.24) and solving for the plastic necking
strain gives

(B.26)

From Eq. (B.16)

(B.27)

1 Necking strain is defined as the strain required to defonn the test specimen from maximum load
to fracture.
 Appendix B Equations for Uniaxial True Stress and True Strain 363

From elastic theory

(B.28)

Substituting Eq. (B.28) in Eq. (B.27) gives

(B.29)

Substituting Eq. (B.29) in Eq. (B.26) gives

8
PN
= (eF - 8
PU
- e<TTdE + 1) LG
t (B.30)
s

Again, from Eq. (B.l6)

(B.31)

From Eq. (B.l7)

_ a TU
8 TPU - 8 TU - - (B.32)
E

Substituting Eq. (B.32) in Eq. (B.31) gives

(B.33)

or

(B.34)

From Eq. (B.16)

(B.35)

Substituting Eq. (B.35) in Eq. (B.34) gives

1 + 8U
8
PU
=e<TTU/E
---1 (B.36)
364 Appendix 8 Equations for Uniaxial True Stress and True Strain

Substituting Eq. (B.36) in Eq. (B.30) gives

8
PN
= ( _8 F -
1+8 - e(]"T.lE + 2
__u
e(]"ru/E
)Lt
-.Sl (B.37)
s

For the sum of the true plastic strain at ultimate plus that for necking the total true
plastic strain

(B.38)

From Eq. (B.18)

(B.39)

Substituting Eq. (B.39) in Eq. (B.38) gives

(B.40)

From Eq. (B.15)

(B.4t)

From which

(B.42)

or

(B.43)

From Eq. (B.12)

(B.44)
 Appendix B Equations for Uniaxial True Stress and True Strain 365

Substituting Eq. (B.43) in Eq. (B.44) gives

(B.4S)

Substituting Eq. (B.3?) in Eq. (B.4S) gives

_ 1 + eu [ (- _ 1 + eu _ Lc]
(1 _V ";)'e"wlE 1+ 2 t,
O"TF/E )
"" - "F FF e"wlE e + (8.46)

where ts is greater than to' from Eq. (B.2Sb), Eq. (B.3?) becomes

e
PN
= ( _e F -
l+e
__
O"TV/E
u - eO"TF/E + 2 )L
~
t0 (B.4?)
e

or

e = ( _e - _
l+e
_u - e O"TF/E + 2) LC
(B.48)
PN F eO"TV/E (to/e PN )ePN

From which

(B.49)

Again substituting Eq. (B.49) in Eq. (B.4S) gives

_ 1 + eu
)2 {
1+
[( __
eF
~_
e
O"TF/E ) Lc
+ 2 (to/ e pN)
]1/2}
e O"ru/
(TTF - (TF ( E
1-v- e (TTF 0"
TV
/E

(B.SD)

Substituting Eq. (B.4D) in Eq. (B. I?) gives

(B.Sl)
366 Appendix B Equations for Uniaxial True Stress and True Strain

From Eq. (B.lS)

(B.S2)

From which

(B.S3)

or

(B.S4)

or

(B.SS)

For a tensile specimen with a circular cross-section where reduction of area, RA,
is given, reduction of area is defined as

(B.S6)

or

(B.S7)

Substituting Eq. (B.6) in Eq. (B.S7) gives

(B.S8)

Substituting Eq. (B.1l) in Eq. (B.S8) gives

1
RA=l--- (B.S9)
1 + cp

From which for the necked region

RA
c
PF
= 1 - RA (B.60)
Index

Accumulative damage, 12,60--62,97 Compact tension specimen (see testing)

Airy stress function, 108 Compliance method, 189-191
Anisotropy Constant amplitude fatigue test, 34-37
beam analysis, 248-249 Constant life diagram, 41-47
material, 128-134 Conventional fatigue
Annotation for drawing, 265-267 High cycle fatigue, 13
Anti buckling device, 207-208 low cycle fatigue, 13
Apparent fracture toughness, 138-139 Conventional stress-strain curve, 80-81
Correction
Back surface correction, 155 Back surface, 155, 156
Basquin exponent, 82 Plastic zone, 307-309
Bauschinger effect, 72 Thickness, 126, 130, 169,307-309,317,
Beam analysis, 248-249 321
Biaxial stress state, 124-125 Width,302
Bolts integrity assessment Crack closure effect, 194-196
bearing stress, 239 Crack extension force, 105
bolted joint, 9-10, 230-248 Crack initiation concept, 57-67
fatigue crack growth, 231-248 PSB,64
pad analysis, 237-248 Crack growth analysis
pad bending, 239 bolts, 231-236
preloaded bolt, 230-237 pads, 237-248
Brittle failure Crack growth rate, 179
cleavage, 1,4, 124-126, 132, 135 ASTM testing, (see testing)
ductile brittle transition, 132 cycle by cycle, 227-229
equations, 180-188
Center crack, 117 NASA/FLAGRO, 178, 185-186,
Charpy test, 132-133 207-216,
Cleavage, 1,4, 124-126, 132, 135 stress ratio effect, 194-196
Closure Phenomenon, 196-198 Threshold value, 241-247
Elber, 196-198 Crack initiation life, 68
Newman, 203-206 Crack opening stress intensity, 206
Raju, 206 Crack tip
Coffm-Manson law, 82 modes, 122-123
Combined stress intensity, 114-117 plasticity, 146-152
Comments on Critical stress intensity factor, 6, 124-132
Miner's rule, 50-51 Crystal structures, 70-72

367
368 Index

Curve Effect of yielding

Empirical representation of S-N, 38-40 fracture toughness, 130
S-N, 7,13-15,32-34 Effective crack length, 141
s-N, 13-15,32,92-98 Effective stress intensity, 196, 198
cyclic stress-strain, 8, 86-88, 90-92 Effective stress range, 198
Cycle by cycle fatigue crack growth, 227-229 Effective yield stress, 203
Cyclic or fluctuating load, 15-18 Elastic energy release rate, 3
Cyclic stress-strain curve, 15,74-82 Endurance limit, 42-43
Cyclic hardening and softening, 74 Energy balance approach, 5, 6,117,118
Energy release rate, 105, 290
Damage tolerance, 10, 253-268 Engineering stress-strain curve, 69-70
Deformation Equiaxed dimples, 59
plastic (see plastic zone) Examples
Modes, 106 chapter (2), 23, 24, 44, 48, 79,87,
Design philosophy 90,95
fail safe, 4, 256 chapter (3), 113, 127, 142, 150, 192
slow crack growth, 4, 254 chapter (4), 186,208,212,213,221,234,
Determination of 239,244
au~ energy term, 292 Extrusion, 63-66
oC
auu energy term, 294
oc Factors influencing
Development of endurance limit, 34-35
R-curve, 161-164 fracture toughness, 130-134
Constant life diagram, 42-44 threshold stress intensity
Diagram S-N, 38, 40
Gerber, 42-43 Fail safe design, 4, 256
Goodman, 42-43 Failure criterion
S-N, 40-42, 48-51 Octohedral shear stress, 295-296, 303
s-N,92-98 Von-Mises, 148-150
residual strength capability, 134-146 Fatigue ductility coefficient, 82
universal slope, 98-99 Fatigue strength coefficient, 82
Kitagawa, 202 Fatigue
Dimple shape, 57-59 low cycle, 13
Dislocation, 62 high cycle, 13
Ductile behavior spectrum, 18-31
brittle transition, 132 Fatigue notch factor, 88-93
fracture Flat failure, (see cleavage)
Dugdale'S model, 3, 248 Flaw
elliptical, 153
Eddy current, 270-271 embedded, 153-154
Effective crack size, 141 size inspection (see NDI)
Effect of crack size Flight spectrum, 28-31
threshold stress intensity, 199-203 FMDM theory, 5, 289
crack closure, 200 Forman equation, 185
Effect of thickness Fracture
fracture toughness, 130-131 brittle, 64
Effect of temperature critical parts classification, 261-262
fracture toughness, 130-134 monotonic load, 57-59
Effect ofrate of loading plane strain (see plane strain)
fracture toughness, 130 plane stress (see plane stress)
 Index 369

surface examination, 51-57 Monotonic failure, 57-59

toughness, 4, 124-128 Multiple load path design, 253-257
Fracture control
acceptability of hardware, 263-265 NASAIFLAGRO computer code, 8, 178
activities, 260-261 Natural stress-strain curve, 81-83
board, 267-268 Net section yielding, 313, 317
contents, 261 Neuber rule, 8, 15,68,93-98
development, 259 Newman crack closure, 247-252
parts classification, 261-262 Non destructive inspection (NOI)
plan, 10, 258-268 Eddy current, 270-271
Fracture toughness magnetic particle, 270
apparent, 138-139 penetrant, 269
plane strain, 122-123 radiographic, 272
plane stress, 122-123 special, 288
surface crack, 159-160 ultrasonic, 271
testing (see testing) Notional crack, 148
computation, 312-320
Opening mode, 106
Griffith theory, 2, 3, 103-105,289 Opening stress, 198, 203
Grip fixture apparatus, 172-174 Overload,216-218

High cycle fatigue, 13, Paris Law, 8, 183

High energy components, 264, 265 Part through cracks
Hysteresis loop, 15,71-74 emanating from a hole, 158-159
fracture toughness, 188-190
Intrusion, 63-66 stress intensity factor, 179-184
Inspection (see NOI) Penetrant inspection, (see NOI)
Intergranular fracture, 66 Persistent slip bands (PSB), 64
Irwin plastic zone, 170-172 Plane strain fracture toughness, 122, 123
Initiation Plane stress fracture toughness, 122-123
fatigue crack, 57, 177-178 Plastic zone shape
Irwin, 146-148
Kitagawa diagram, 202 Tresca, 150-152
Von-Mises, 148-150
Low released mass, 256 Power law representation
Leak-before -burst, 5, 160-164 cyclic stress-strain curve, 78-79
LEFM,103 stress-strain curve, 71
Linear cumulative damage, 47-51 Plate bending, 239
Low cycle fatigue, 13,67-68 Pre-cracking, 166, 189
Preloaded bolts, 230-237
Magnetic particle inspection, 270 Probability of flaw size detection, 276-278
Mandrelizing, 17
Manson, 98 Radiographic inspection, 272-274
Material selection, 270 (see also appendix A) Rain flow technique, 21-23
Material anisotropy, 128-130 Range, 21
Mean stress effect, 88 Residual strength capability, 4, 7,156-160
Microvid coalescence, 57-59 abrupt failure, 158-160
Miner's rule, 7,15,47-49 Retardation effect, 9
Mixed mode fracture, 106,306-309 Wheeler model, 9
Modes of deformation, 106 Willenborg model, 9
370 Index

Reversal to failure, 82 Tearing failure, 135, 138-141

R-curve, 5, 8, 139-141 Testing
Rolled threads, 9 crack growth test, 188-193
fracture toughness, 164--173
K1c test, 164--168
Safe-life, 8, 155,202,206
Kc test, 169-174
Shuttle, 38-39
S-N test, 34--38
Simulative law, 8
specimen, 165, 171-174
Slip mechanism
Theoretical stress concentration
monotonic load, 61-62
factor, 89, 93-94
cyclic load, 63-66
Threaded bolt, 230-237
Slope method
Threshold stress intensity factor, 199-203
polynomial, 182
Through cracks, 137-142
secant, 180-181
Tractability, 265
Small cracks, 199-202
Transition point, 86-88
Smith-Watson-Topper,88
Tresca yield criterion, 150-152
Special level inspection, (see NDI)
Triaxial stress state, 124, 125
Stage I and stage II, 7
True stress-strain curve, 70-71 (see also
Strain ratio, 26
appendix B)
Strain-controlled, 76-78
Strain-life prediction, 8, 82-86 Ultrasonic inspection, 271
Stress and strain at notch, 104--112 Universal slope method, 98-99
Stress concentration factor, 153-157
Variables affecting NDI, 278-286
Stress intensity factor, 105-122
Variable amplitude stress, 216-218
derivation, 107-111
Verification ofFMDM result, 309-312
equation, 117-122
Von-Mises yield criterion, 148-150
range, 179
surface crack, 153-157 Walker equation, 185
Stress function Welded joints failure, 2-3
Airy, (see Airy) Westergaard stress function, 109
Stress ratio, 18 Wheeler model, 15,218-219
Stress-strain curve, 69-71, 309 Willenborg model, 15,219-221
Striation (beach) marking, 51 Wohler, 7,13
Superposition principle, 114--116
X-Ray inspection, 272-274
Surface cracks, 152-160
from a hole, 158-159 Yield criterion
longitudinal in a pipe, 157-158 Tresca, 150-152
fracture toughness, 159-160 Von-Mises, 148-150
Surface energy, 104 Yield strength effect, 130, 203