MECHANiCS OF MMED MODE FATIGUE BEHAVIOUR

OF COLD WORKED ADJACENT HOLES

by

Paraskevas Papanikos, M.A.Sc.

A thesis submitted in conforrnity with the requirements

for the Degree of Doctor of Philosophy,
Graduate Department of Mechanical and Industrial Engineering
University of Toronto

@Copyright by Paraskevas Papanikos 1997

I*m National Library
of Canada
Bibliothèque nationale
du Canada
Acquisitions and Acquisitions et
BibIiographicSenrices servicesbibliographiques
395 Welington Street 395. nie Wellington
Ottawa ON K1A O N 4 Ottawa ON K1A ON4
Canada Canada

The author has granted a non-
exclusive licence allowing the
National Library of Canada to
reproduce, loan, distribute or sell
copies of this thesis in microfom,
paper or electronic formats.
L'auteur a accordé une licence non
exclusive permettant à la
Bibliothèque nationale du Canada de
reproduire, prêter, distribuer ou
vendre des copies de cette thèse sous
la forme de microfiche/nlm, de
reproduction sur papier ou sur format
électronique.

The author retains ownership of the L'auteur conserve la propriété du

copyright in this thesis. Neither the
thesis nor substantial extracts fiom it
may be printed or otherwise
reproduced without the author's
permission.
droit d'auteur qui protège cette thèse.
Ni la thèse ni des extraits substantiels
de celle-ci ne doivent être imprimés
ou autrement reproduits sans son
autorisation.
 MECHANICS OF MIXED MODE FATIGUE BEHAVIOUR
OF COLD WORKED ADJACENT HOLES

Doetor of Phiiosophy

Department of Mechanical and Industnal Engineering

University of Toronto

Fatigue failure of aircraft structures almost invariably begins at the root of geomeuicai
discontinuities. The designer is frequently required to predict the fatigue behaviour of
cracks emanating from these discontinuities so as to assess the residual Iife and to avoid
catastrophic failure. However, the problem is complicated by the fact that these critical Ioad
bearing structures contain residual stresses and interacting stress concentration features such
as holes, inhomogeneities and inclusions. It is with this in mind that the current research
program was undertaken.

Four aspects of the work were accordingly exarnined. The first was concemed with the
development of an anaiytical mode1 capable of predicting the stress state in a solid containing
a number of interacting (adjacent) holes. In this case, a new method for treating arbitrarily
shaped holes is developed. The theoretical formulation is based upon the superposition of
single dislocation solutions, which reduce the original interaction problem into the solution
of singular integral equations in terms of dislocation density functions.

The second was concerned with the evaluation of residual stresses due to cold expansion
of adjacent holes. In this case, a comprehensive elasto-plastic three-dimensional finite
element analysis was conducted and the residuai stress field was obtained through the
thickness of the workpiece.
 The third was concerned with the prediction of mixed mode fatigue crack growth in
the presence of residual stresses. In this case, effective stress intensity factors solutions
for generalized cracks emanating from holes, using a newly developed method, were used
in conjunction with the appropriate crack tracking criterion to determine the potential
incremental crack growth experienced by the component under fatigue loading conditions.

The fourth was concerned with the venfication of the above modelling techniques as
well as the determination of fatigue crack growth data for &Zn 7075 in the presence of
residual stresses.

The results reveal the importance of the location and the magnitude of the stress concen-
tration upon the fatigue Iife of adjacent holes. In addition, they indicate that mixed mode
fatigue crack growth in the presence of residual stresses cm be reliably predicted using a
newly developed model.
Acknowledgements

1offer my sincere gratitude to Dr. S.A. Meguid for his cornmitment to his students, technicd
and financial assistance, and his patience throughout the course of my research. 1also wish
to thank Dr. Xiaodong Wang for his help during the undertaking of the current study.
Notation

surface area

coefficients of the Chebyshev polynornial expansion

Young's modulus

mode 1strain energy release rate

mode II strain energy release rate

mode 1stress intensity factor

mode II stress intensity factor

maximum mode 1 stress intensity factor (no residual stress)

maximum mode II stress intensity factor (no residual stress)

minimum mode 1stress intensity factor (no residual stress)

minimum mode II stress intensity factor (no residual stress)

effective maximum mode 1 stress intensity factor

effective maximum mode II stress intensity factor

effective minimum mode 1stress intensity factor

effective minimum mode II stress intensity factor

residuai mode 1stress intensity factor

residual mode II stress intensitv factor

equivalent stress intensity factor (no residual stress)

effective stress intensity factor

equivalent stress intensity range (no residuai stress)

effective stress intensity range

number of Chebyshev polynomial expansion tems

number of collocation points for segment j

number of fatigue cycles

nominal stress ratio

effective stress ratio

hole radius

transformation matrix

Chebyshev polynomiai of the first kind

Chebyshev polynomial of the second kind

strain energy

crack length

half-length of segment j

crack growth rate

haif centre distance between two holes

Burger's vector associated with normal deformation

Burger's vector associated with tangentid deformation

dislocation density functions for segment j

plate height

plate thickness
 normal deformation of segment j

tangential deformation of segment j

plate width

global coordinates

local coordinates for segment j

inclination angle

radial strain

tangential strain

critical direction of crack growth

Poisson's ratio

remote applied stress

global stress components

I stress components associated with normai deformations
"ij

stress components associated with tangentid deformations

stress components of segment j

prescribed nomai stress of segment j

maximum applied stress

minimum applied stress

yield stress

tangential stress

tangential residual stress

equivalent stress

prescribed shear stress of segment j

orientation angle for segment j

Contents

Acknowledgements iii

Notation

List of Figures

List of Tables xv

Introduction and Justification 1

1.1 Justification for the Study . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 ResearchObjectives ............................ 4
1.3 Method of Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3.1 Stress analysis of interacting holes . . . . . . . . . . . . . . . . . 6
1.3.2 Modelling of mixed mode behaviour . . . . . . . . . . . . . . . 6
1.3.3 Residual stresses due to cold hole expansion . . . . . . . . . . . 6
1.3.4 Mixed mode fatigue behaviour of airfrarne dloys ......... 6
1.4 Layout of Thesis .............................. 7

2 Literature Review 8
2.1 Stress Analysis of Interacting Holes .................... 8
2.1.1 Anaiytical investigations . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2 Finite element and photoelastic investigations . . . . . . . . . . . II
2.2 Residual Stresses Due to Hole Expansion . . . . . . . . . . . . . . . . . 13

vii
 2.2.1 Cold hole expansion parameters . . . . . . . . . . . . . . . . . . 17
2.3 Mixed Mode Crack Growth . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.1 Stress intensity factors evaiuation . . . . . . . . . . . . . . . . . 19
2.3.2 Criteria for rnixed mode crack growth . . . . . . . . . . . . . . . 21

3 Analysis of Multiply Connected Regions 26

3.1 Elastic Field Analysis of an Arbitrarily Shaped Hole . . . . . . . . . . . 26
3.1.1 Basic equations . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.2 Integral equations . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2 Solution of Singular Integral Equations . . . . . . . . . . . . . . . . . . 32
3.2.1 Chebyshev polynomial expansions . . . . . . . . . . . . . . . . 32
3.2.2 Determination of Chebyshev polynomial coefficients . . . . . . . 33
3.3 Tangentid Stress Field . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4 Multiple Interacting Holes . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.5 Verification Exarnples . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5.1 ...
CircuIar hole . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5.2 Elliptical Hole . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5.3 Two circular holes . . . . . . . . . . . . . . . . . . . . . . . . . 37

4 Analysis of Generalized Curved Cracks 40

4.1 Formulation of the Curved Crack Problem . . . . . . . . . . . . . . . . . 40
4.1.1 Basic equations .......................... 40
4.1.2 Boundary conditions and integral equations . . . . . . . . . . . . 41
4.2 Solution of Singular Integral Equations . . . . . . . . . . . . . . . . . . 42
4.2.1 Chebyshev polynomial expansions . . . . . . . . . . . . . . . . 42
4.2.2 Determination of Chebyshev polynomial coefficients . . . . . . . 44
4.2.3 Stress intensity factors . . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Verification of the Mode1 . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3.1 Crack emanating from a circular hole . . . . . . . . . . . . . . . 46
4.3.2 The kinked crack problem . . . . . . . . . . . . . . . . . . . . . 47
4.3.3 Interacting cracks . . . . . . . . . . . . . . . . . . . . . . . . . 47

viii
5 3-D Finite Element Analysis of Cold Worked Adjacent Holes SI
5.1 Two-dimensional Finite Element Analysis . . . . . . . . . . . . . . . . . 52
5.1.1 Details of two-dimensional geometry . . . . . . . . . . . . . . . 52
5.1.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . 52
5.1.3 Limitations of the two-dimensional mode1 .
............ 58
5.2 Three-dimensional Finite Element Analysis . . . . . . . . . . . . . . . . 58
5.2.1 DetaiIs of three-dimensional mode1 . . . . . . . . . . . . . . . . 58
5.2.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . 60

6 Experimental Investigations 70
6.1 Materiais and Test Specimen . . . . . . . . . . . . . . . . . . . . . . . . 70
6.1.1 Selection of material . . . . . . . . . . . . . . . . . . . . . . . . 70
6.1.2 Specimen design and preparation ................. 71
6.2 Hole Expansion Method . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.2.1 Selection of expansion parameters . . . . . . . . . . . . . . . . . 73
6.2.2 Hole alignment and mandrelizing . . . . . . . . . . . . . . . . . 73
6.3 Fatigue Fracture Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.3.1 Fatigue testing equipment . . . . . . . . . . . . . . . . . . . . . 74
6.3.2 Fatigue test program . . . . . . . . . . . . . . . . . . . . . . . . 77
6.3.3 Crack size measurement . . . . . . . . . . . . . . . . . . . . . . 77

7 Analysis of Results and Discussion 79

7.1 Crack Initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.1.1 . . . . . . . . . . . . . . . . . . . . . . . . . 79
Unlreated sarnples
7.1.2 Cold worked sarnples . . . . . . . . . . . . . . . . . . . . . . . 83
7.2 Mechanics and Mechanisms of Fatigue Crack Initiation . . . . . . . . . . 84
7.3 Crack Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.3.1 Crack tracking criterion . . . . . . . . . . . . . . . . . . . . . . 87
7.3.2 Superposition of stress intensity factors . . . . . . . . . . . . . . 88
7.4 Mechanics and Mechanisms of Fatigue Crack Growth . . . . . . . . . . . 90
7.4.1 Fatigue crack growth behaviour of AI-Zn ailoy .......... 90
 7.4.2 Incremental crack tracking .
.................... 94
7.4.3 Fatigue crack growth rate . . . . . . . . . . . . . . . . . . . . . 97

8 Conclusions and Future Work 101

8.1 Statement of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
8.3 ............................
General Conclusions 102
8.3.1 Description of stress field of adjacent holes . . . . . . . . . . . . 102
8.3.2 The curved crack problem . . . . . . . . . . . . . . . . . . . . . 103
8.3.3 Finite element andysis of cold hole expansion . . . . . . . . . . 103
8.3.4 Fatigue crack growth in the presence of residuai stresses . . . . . 104
8-4 Thesis Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.5 Recomrnendations for Future Work . . . . . . . . . . . . . . . . . . . . 105

References 107

A Listing of a Sarnple Program 113

List of Figures

1.1 Modes of failure in aircraft (after [II). ...................

1.2 Exarnples of failures involving interacting holes: (a) aeronautical (after
[l]). and (b) automotive industries (after [2]).
...............
1.3 Corrosion fatigue failure of a steam boiler (after [3]) . . . . . . . . . . . .
1.4 Schematic of method of approach. . . . . . . . . . . . . . . . . . . . . .

Stress concentration factor at different locations for a hole close to a bound-

ary (after [7]) . ...............................
Decomposition of an original problem into a homogeneous problem and
two subproblems (after [9]). ........................
Circumferential and radial stresses prior to unloading (after [22]) . . . . . .
Residual circumferential and radial stresses after unloading (after [22]). .
Schematic of the split sleeve method. . . . . . . . . . . . . . . . . . . .
.
Fatigue crack growth of (a) untreated hole and (b) cold expanded hole
(after [22]). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Stress intensity factors for a cold worked hole (after [37]) . . . . . . . . .
An angled-crack under uniaxial loading. ..................
Prediction of crack growth direction. . . . . . . . . . . . . . . . . . . . .

3.1 Arbitrady shaped hole: (a) exact geometry. and (b) approximate geometry
using line segments. ............................
3.2 Dislocations associated with (a) normal. and (b) tangentid deformations. .
3.3 A schematic of multiple interacting holes. . . . . . . . . . . . . . . . . .
3.4 Normalized tangentid stress distribution around a circular hole. . . . . . .
 Normalized tangentid stress distribution around an elliptical hole. ....
Two equal holes under uniaxial loading...................
Nomdized tangentid stress distribution for two holes. . . . . . . . . . .

Segmented arbitrarily shaped cumed crack. . . . . . . . . . . . . . . . .

Normalized stress intensity factor h
;' of an edge crack emanating from a
circular hole.................................
Details of geometry of a kinked crack. . . . . . . . . . . . . . . . . . . .
Normalized stress intensity factors h" of a kinked crack: (a) kink angle
= lSO, and (b) kink angle q5 = 75".
...................
Schematic of two colinear cracks under uniaxial loading. . . . . . . . . .
Vanation of normalized stress intensity factor with normalized distance
between crack tips. .............................

Cold expansion of two adjacent holes: (a) schematic of geometry used. and
(b) finite element mesh of the shaded region . ...............
Details of equivalent stress contours for different expansion levels: (a)
2%. (b) 4%. (c) 6%. and (d) unloading from 6% expansion. ........
Variation of normalized tangentid residual stress o;/oydong lengths (a)
AB. and (b) CD for 2& R and H'/E= 1%. ................
Variation of nonnalized tangential residual stress oé/oyalong length AB
for a 6% expansion and H f /E=O.1 %. . . . . . . . . . . . . . . . . . . .
Vanation of nomalized tangential residual stress o;/a, along length AB
for a 6% expansion with 2&R .......................
Finite element mesh used for the sequential expansion of the two holes. . .
Variation of normalized tangential residuai stress oijcr, for sequential ex-
pansion with 2c=4 R. 4% expansion and H f /E=û .1 %. ...........
Cold expansion of two adjacent holes: (a) schematic of geometry used. and
(b) finite element mesh .
..........................
Schematic of three-dimensional contact elernent used . . . . . . . . . . . .
5.10 Cornparison with residual strain distributions using interferometry. . . . .

xii
5.1 1 Details of equivalent stress contours at entq and exit for different mandrel
strokes: (a) 10%. (b) 25%. (c) 60%. and (d) 100%. ............
5.12 Variation of normalized equivalent stress 8/uydong length AB for different
mandrel strokes: (a) 10%.(b) 25%. ( c ) 60%. and (d) 100%. . . . . . . . .
5.13 Viriation of normalized equivalent stress 5/09through thickness at point
B (Figure 5.8). ...............................
5.14 Viriation of normalized tangential residual stress ai/c, through thickness
at point B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.15 Tangential residud stress a,'/cgcontours for 2% expansion. . . . . . . .
5.16 Variation of normalized tangential residud stress oi/oy dong lengths (a)
AB. and (b) CD for the entry and exit faces of the workpiece. . . . . . . .

6.1 Detailed geometry of a typical sample....................

6.2 Schematic of mandrelizing apparatus. . . . . . . . . . . . . . . . . . . .
6.3 Mandrel-hole misdignrnent . . . . . . . . . . . . . . . . . . . . . . . . .
6.4 Uniaxial fatigue testing equipment. . . . . . . . . . . . . . . . . . . . .
6.5 Replication of the fatigue specimen. . . . . . . . . . . . . . . . . . . . .

7.1 Geometrical notations for two adjacent holes. . . . . . . . . . . . . . . .

7.2 Variation of normalized tangentid stress for different centre distances:
(a) 2c=3 R. and (b) 2r-4 R . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3 Expenmental results for two untreated sarnples: adjacent holes inclined at
(a) 30". and (b) 60' for a separation distance 2c=3 R . . . . . . . . . . . .
7.4 Variation of nomaiized effective tangentid stress for centre distances (a)
2c=3 R. and (b) 2&R ...........................
7.5 Mechanisms for crack initiation in a cold worked component. . . . . . . .
7.6 Experimental results showing two dominant fatigue cracks in a cold worked
sarnple: P=30°. 2c=3R. ..........................
7.7 Algorithm for prediction of mixed mode crack growth rate in the presence
of residual stresses. ............................
7.8 Experimentally determined crack length versus number of cycles for sam-
ples with 2c=3 R and (a) P=O", and (b) ,8=30°.. ..............
7.9 Calculated equivalent stress intensity factors: (a) details of crack geomeîries
examined, and (b) variation of stress intensity factor versus crack length for
a unit applied load using anaiytical model. .................
7.10 A typical penny-shaped crack in a cold worked sample with P=30° and
2c=3R. ..................................
7.1 1 knalytical prediction of incremental fatigue cracks for untreated and cold
worked geometry for ,û=30° and 2c=3 R. . . . . . . . . . . . . . . . . . .
7.12 Finite element predictions of fatigue crack growth at different lengths:
(a) a= 1 mm,(b)a=2 mm, (c) a=3 mm, and (d) a=4 mm for ,B=30° and 2c=3 R.
7.13 Experirnental results of fatigue crack growth for (a) untreated, and (b) cold
worked sarnples with ,B=30° and 2-3 R. ..................
7.14 Analytical prediction of incremental fatigue cracks for untreated and cold
worked geornetry for p=45* and 2c=3 R. ..................
7.15 Experirnental results of fatigue crack growth for (a) untreated, and (b) cold
worked samples with p45" and 2c=3 R. ..................
7.16 Crack path for a cold worked sample with p=30° and 2-3 R. . . . . . . .
7.17 Fatigue crack growth rate versus equivalent stress intensity range. . 1
7.18 Fatigue crack growth rate versus effective stress intensity range. . 1

xiv
List of Tables

2.1 Life irnprovement due to cold expansion (after [22]) . . . . . . . . . . . . 16

5.1 Materid properties of workpiece and mandrel . . . . . . . . . . . . . . . 61

6.1 Chernical composition of Al-Zn 7075 alloy (wt%) (after [65]). . . . . . . 71

6.2 Mechanical properties of Al-Zn 7075-T65 1 dloy. . . . . . . . . . . . . . 71
6.3 Geometrical details of fatigue specimens (in mm). . . . . . . . . . . . . . 72

7.1 Maximum stress concentration for the geometries exarnined. . . . . . . . 80

7.2 Initiation life in cycles for untreated samples: am.,=87.5 MPa. . . . . . . 82
7.3 Initiation life for untreated and cold worked sarnples. . . . . . . . . . . . 83
7.4 Reduction in effective tangentid stress due to hole expansion . . . . . . . . 83
Chapter 1

Introduction and Justification

In this chapter, we define the problem, justify the undertaking of the study and outline the
method of approach adopted in achieving the set objectives. Furthemore, we provide a
summary of the layout of the thesis.

1.1 Justification for the Study

The safety of an aircraft has always been the main concem of aircraft certification authorities.
Economic pressure resulting from the reduced availability of strategic materials, the high
cost of aircraft components and the continued demand, by al1 aircraft manufacturers/users,
for longer life and higher thmst to weight ratio continue to provide a stimulating challenge
for aircraft designers.
Fatigue is undoubtedly the most prevalent mode of failure in aircraft structures (Fig-
ure 1.1). The consequence of failure of a primary load-carrying component in an airfrarne
or engine is usually catastrophic, often resulting in loss of life and hardware. The designers
of such systems are constantly faced with the challenge of establishing stress levels in
these parts that will allow the use of suitable high strength materials to provide lightweight,
high-performance structures. These structures should be sufficiently resistant to fatigue
failure.
Fatigue failure almost invariably begins at the root of geometrical discontinuities in
highly stressed components, such as those depicted in Figures 1.2 and 1.3. For example,
the presence of interacting fastener holes in the upper skin assembly of a F-105 fuselage,
depicted in Figure 1.2(a), resulted in the fatigue failure and the ultimate fracture of the
assernbly. Figure 1.2(b) shows the fatigue failure of a flywheel plate of a Station Wagon
which was caused by excessive vibrations. The effect of the environment can also influence
the fatigue behaviour of interacting holes, as evidenced by the corrosion fatigue failure of
the steam boiler depicted in Figure 1.3.

FATIGUE CORROSION

Figure 1.1: Modes of failure in aircraft (after [l]).

At this stage, it is important to identify the pertinent parameters which influence the
fatigue strength of criticai load bearing componenWassemblies. These include: (i) the
presence of and the interaction between stress concentration features, (ii) the extemally
applied load, (iii) the induced residuai stress field, and (iv) the toughness of the material
selected. The importance of these parameters is evident in providing guidelines for: (i)
the selection of the different geometrical feanires in newly designed/developed aircraft, (ii)
the life assessrnent of aging aircraft, (iii) the introduction of beneficial residual stresses in
airframe and engine components, and (iv) the selection of materials, such as Al-Zn alloys.
The current literature indicates that these issues have not been given their due attention.
Figure 1.2: Examples of failures involving interacting holes: (a) aeronautical
(after [II), and (b) automotive industries (after [SI).
 Figure 1.3: Corrosion fatigue M u r e of a steam boiler (after [3]).

1.2 Research Objectives

It is therefore the objective of this study to examine the fatigue fracture behaviour of a
plate containing two circular holes in close proximity in the presence of residual stresses.
Specifically, it is desired to:
develop an analytical model to predict the stress field and the mixed mode stress
intensity factors for a generdized crack in a solid containing a number of interacting
holes,

conduct three-dimensional finite element analysis to obtain the residual stress field
due to cold hole expansion of adjacent holes,

obtain fatigue crack growth data for Al-Zn 7065-T651 aluminium alloy in the presence
of residud stresses, and

develop a model to predict the mixed mode fatigue crack growth rate of the current
airfrarne dloy in the presence of residual stresses.

Method of Approach
Figure 1.4 shows a schematic of the method of approach adopted to achieve the above stated
objectives. These are summarized below.
1.3.1 Stress anaiysis of interacting holes

The elastic field resulting from two (or more) interacting circular holes in an elastic solid is
examined. The solution is obtained using a new analytical model which is based upon the
superposition of single dislocation solutions which reduce the original interaction problem
into the solution of singular integral equations in terms of dislocation density functions. The
resulting integral equations are then solved by using Chebyshev polynomial expansions.
The accuracy of the model was verified against existing solutions and by using the finite
element method. The model was used to predict the crack initiation site(s) and was verified
with experimental results.

1.3.2 Modelling of mixed mode behaviour

A new model capable of deterrnining the effective stress intensity factor in the presence of
residuai stresses of a crack emanating from one hole was developed. The theoretical model
descnbed in section 1.3.1 was modified to account for the singularity at the crack tip. The
stress intensity factor solutions obtained were used in conjunction with appropriate crack
tracking cnteria to determine incremental crack growth under fatigue loading conditions.

1.3.3 Residual stresses due to cold hole expansion

A comprehensive elasto-plastic three-dimensional finite element analysis was conducted to

evaluate the residual stress field due to hole expansion of two adjacent holes. The contact
between the mandrel and the holes was modelled using special contact elements which
employ the combined penalty-Lagrange multipliers formulations. The effect of expansion
level, separation distance and mechanical properties of the work-piece upon the resulting
residual stress field was examined.

1.3.4 Mixed mode fatigue behaviour of airframe aiIoys

Fatigue tests of aluminum samples containing two circular holes were conducted using a
digitally controlled electro-hydraulic fatigue test apparatus. The two holes were expanded
at room temperature using a mandrelizing apparatus. High cycle fatigue tests were carried
out under load control at room temperature at a stress ratio R = 0.1. Crack initiation and
growth were monitored by exarnining cellulose acetate replicas of lightly polished surfaces
and an optical microscope.

1.4 Layout of Thesis

This thesis is divided into eight chapters. Chapter one defines the problem, justifies the
undertaking of the study and outlines the method of approach adopted. Chapter two
provides a critical review of the relevant work in four main areas: eadier analytical,
numerical and experimental investigations of interacting holes and residual stresses due to
cold hole expansion. Chapter three provides a detailed account of the theoretical approach
adopted in the determination of the stress field associated with interacting holes. In chapter
four, the approach developed in chapter three is extended to obtain expressions for the
stress intensity factors, which are used for the prediction of mixed mode fatigue crack
growth in the presence of residual stresses. Chapterfive outlines the technique adopted
in determining the residual stresses resulting frorn the cold expansion of adjacent holes.
Chapter six describes the different experimental techniques adopted in the study. These
include: hole expansion rnethod, fatigue testing and crack detection. In chapter seven, both
experimental results and analytical predictions of the mixed mode fatigue crack growth are
provided. Finally, chapter eight concludes the thesis and summarizes the findings of this
work.
Chapter 2

Literature Review

This literature review covers three main topics. The first deds with the analytical, numerical
and experimental investigations of the stress field of interacting holes. The second examines
the effect of mechanically induced residual stresses on the fatigue fracture behaviour of
airfrarne alloys. The third deds with the modelling and prediction of fatigue crack growth
in the presence of residual stresses, including stress intensity factor solutions for cracks
emanating from holes.

2.1 Stress Analysis of Interacting Holes

Since the pioneering work of Kirsch [4] of the single hole problem, much attention has
been devoted to the development of solutions for problems involving adjacent holes. These
solutions will enable the designer to assess the mechanical integrity of engineering corn-
ponents by minimizing the stress concentration associated with interacting holes and by
predicting the initiation site(s) of potential fatigue cracks. This section will focus on the
relevant analytical, numerical and experimental research work concerning the behaviour of
adjacent holes under load.
2.1.1 Analytical investigations
The first attempt to solve the problem of two interacting holes in an infinite medium was
cmied out by Jeffery [5] in 1920. This problem was an application of his general solution
of the two-dimensional equations of elasticity in bipolar coordinates. Unfominately, he did
not provide any numerical results. Jeffery also solved the problem of a circular disc with
an eccentric circular hole and the problem of a semi-infinite plate bounded by a straight
edge and a circular hole. Later, Mindlin [6,7]corrected the solution of the latter problem.
Mindlin found that as the hole approaches the free edge, the tangential stress at the hole
boundary close to the edge (point n, Figure 2.1) tends to infinity while the tangential stress
in the diametrically opposite point (p) becomes 4 for a uniaxial applied unit stress. On the
other hand, the stress at the corresponding point on the free edge (point rn) varies from a
value equal to the applied stress when the hole is far away to zero when the hole approaches
the boundary. Mindlin verified the theoretical results using photoelasticity.

Figure 2.1: Stress concentration factor at different locations for a hole close
to a boundary (after [7]).
 In 1974, Yu and Sendeclqj [8] provided a solution to the problem of multiple circular
inclusions in an infinite matrix using the Schwartz aitemating method. The solution of
the single inclusion was obtained using the complex potential method. In 1985, Horii and
Nemat-Nasser [9] introduced the method of pseudo-tractions to calculate the stress and strain
fields in a linearly elastic homogeneous soiid containing a number of defects of arbitrary
shape. However, the method was applied to cases involving interacting circular holes and
cracks. In the case of interacting holes, the solution was obtained by superimposing the
homogeneous subproblem to two subproblems (Figure 2.2) each consisting of an infinitely
extended body with only one hole and zero stresses at infinity. The radial and shear stresses
applied to the hole boundary in each subproblern are called pseudo-tractions and have to be
determined from the boundary conditions of the problem.

Figure 2.2: Decomposition of an original problem into a homogeneous prob-

lem and two subproblems (after [9]).
 Meguid and Shen 1101 developed a general solution descnbing the interaction between
a main hole and an arbitrarily located defense hole under uniaxiai and biaxiai loading
conditions. Their solution was based upon the complex potentials of Muskhelishvili,
an appropriate superposition procedure and the Laurent series expansion method. They
provided approximate closed f o m expressions for the stress field. They concluded that
both the location and the magnitude of the maximum tangential stress at the main hole
change for a different inclination of defense hole. In addition, Meguid and Shen found that
the largest stress concentration at the main hole occurs when the the central line of the two
holes is inclined at 30" with respect to the applied load.
The problem of multiple void-crack interaction was treated by Hu et al. [Il]. They used a
superposition technique to reduce the problem into a number of single-hole and single-crack
subproblerns. Each hole subproblem was modelled using the method of pseudo-tractions
and each crack subproblem was modelled by a distribution of dislocations. They developed
an integral equation approach based on two types of fundamental solutions, one due to
point loads in a solid containing a hole and the other due to point dislocations in an infinite
elastic body. The fundamental solution for a point force acting on the hole boundary was
developed earlier by Dundurs and Hetenyi [ 12,131.
Finaily, Kienzler and Zhuping [14] and Greenwood [15] developed new techniques for
treating a single circula hole under plane loading. However, they did not consider the
interaction between multiple holes andor inhomogeneities.
The above review indicates that existing models which are concemed with interacting
holes, defects and inhomogeneities, require the use of fundamentai solutions. This makes
them inflexible and incapable of treating arbitrarily shaped holes andor defects. It is,
therefore, necessary to develop a mode1 capable of treating generalized holes and cracks.

2.1.2 Finite element and photoelastic investigations

Most of the analytical work in this field is related to the stress distribution in the vicinity
of interacting holes in infinitely extended elastic bodies. Furtherrnore, in view of the
complex nature of the developed equations, the finite element method and photoelasticity
were adopted.
In 1978, Erickson and Riley [16] undertook a study using two-dimensional photoelas-
ticity to determine the optimum size and location of defense holes for a number of plates
with different central hole diarneter to plate width ratios. The defense hole method was in-
troduced by Heywood [ 171 in 1952. According to this method, smailer holes are introduced
on either side of the original hole to help smooth the flow of the tensile principal-stress
trajectories past the original hole. This leads to reduced stress concentration around the
original hole. Erickson and Riley reported maximum stress reductions from 13 to 2 1% for
plates with hole diameter to plate width ratios between O. 1 to 0.6. They concluded that with
such reductions in the maximum stress level, the improvement in fatigue life of a part c m
be very significant.
Jindal [18] examined the effect of the hole geometry and the effect of defense holes
on the stress concentration around the main hole using both the finite element method
and two-dimensional photoelasticity. It was found that by making the main hole oblong
in the direction of loading, the stress concentration factor at the hole decreased by up
to 22%. Jindal concluded that this approach was more effective than the defense holes
approach, since the same reduction in stress concentration was achieved, whilst avoiding
the introduction of two more regions of higher stress concentration. Meguid [ 191considered
the case of a rectangular plate with two main holes and three ways of introducing defense
holes. He pointed out the significance of the position and inclination of the defense holes
with respect to the main hole.
In 1993, Meguid and Gong [20] undertook comprehensive analytical, finite element and
photoelastic studies of the interaction between existing main holes and arbitrarily located
defense hole systems under uniaxial tension. Their analytical solution, which was based on
the complex potentials of Muskhelishvili, an appropriate superposition procedure and the
Laurent series expansion, was used to predict the variation of the stress concentration for
different defense hole configurations. They concluded that the introduction of defense holes
on either side of the original hole helps to smooth the Row of maximum stress trajectories
p s t the original hole.
2.2 Residual Stresses Due to Hole Expansion
Cold expansion of fastener holes has been used for over forty years in the aerospace industry.
It is a very efficient approacli which results in extending the fatigue life of the treated part
without any weight penalty. Uniil recently, particularly for civil aircraft, the technique was
only applied to critical holes in highly loaded zones of the structure, such as landing gears
and engine mounting regions. In modem aircraft, however, one may find over a thousand
cold expanded holes in the wing done. The requirement of increasing structurai efficiency
combined with a reduction in manufacturing costs has demanded a cioser study of the cold
expansion process.
The cold hole expansion is usually conducted using an oversized bal1 or a mandrel. The
radial interference between the "rigid" bal1 or mandrel and the hole results in inhornoge-
neous plastic deformation (Figure 2.3). Upon unloading, the elastically stressed regions.
away from the hole, would tend to fully recover, and as a result impose a compressive
residual stress field on the plasticdly deformed regions (Figure 2.4). This highly localized
compressive field at or near the hole boundary is equilibrated by the development of a ten-
sile residual stress field in the surrounding regions. These compressive residual stresses are
highly effective in preventing premature fatigue failure under conditions of cyclic loading
[2 1,221.
A major impediment to the use of bal1 or mandrel in the cold hole expansion of fasteners
is the surface darnage introduced at the interface during the cold expansion process. To
overcome this difficulty, the split sleeve expansion, split mandrel expansion and roller
expansion methods have been developed. In the split sleeve method (Figure 2.5) a thin, dry
lubricated sleeve is placed over the stem of the mandrel and is pushed into the hole creating
an interference fit. The mandrel is then drawn back through the split sleeve. Aircraft
holes cornmonly treated by this method range from 5 mm to over 40 mm in diameter, with
expansions between 2% and 6% depending on the material and application. In the split
mandrel method, the mandrel is pushed easily through the hole as a result of the presence
of a machined groove dong its length. This method was developed to replicate al1 the
functions of the split sleeve technique and avoid the use of the sleeve. Finally, the roller
 Circumferential

k
1 2 3
Dirnensionless radial position

Figure 2.3: Circumferential and radial stresses prior to unloading (after [22]).

'"1 Circurnferential

\
Radial

2 3
Dimensionless radial position
Figure 2.4: Residual circumferential and radiai stresses after unloading
(after [22]).
expansion technique consists of a set of specially shaped rollers mounted in a rotary tool
and inserted into the hole.

Mandrel

' Sleeve

Figure 2.5: Schematic of the split sleeve method.

Many analytical solutions have been developed to predict the residual stress distribution
surrounding cold worked holes. The complexity of these models, however, depends upon
the number of variables considered. The exact elasto-plastic solution of an infinite plate
containing a circula hole subject to uniform pressure for an ideaily plastic material was
provided in 1950 by Nadai [23]. Later, Hsu and Forman 1241 obtained the residuai stress
field of an expanded hole, accounting for material hardening. Rich and Impellizzeri [25]
improved the mode1 presented in 1241 by assuming isotropie hardening behaviour to account
for compressive yielding on unloading. The solution for plane stress and plane strain
conditions of linear hardening materials was also given by Jongebreur and Koning [26].
Recently, Wanlin [27] extended the solution of Hsu and Forman [24] to account for the
finite width of the plate.
Existing expenmental and axisymmetric finite element investigations reveal that highly
non-uniform residual stresses are introduced by the cold expansion method 128-321. Link
and Sanford [28] obtained the residual strain (stress) field surrounding a single split-sleeve
cold expanded hole in 6.35 mm thick 7075-T651 aluminum plates using moiré interferom-
etry. They concluded that two-dimensional analysis is incapable of predicting the residual
strain (stress) field surrounding cold expanded holes. This may be due to the geometrical
and material non-linearîties associated with the elasto-plastic contact experienced by the
fastener holes, as well as the complexity of the models needed to describe the problem.
Cook and Holdway [29] reached the same concIusion as Link and Sanford (281 using the
X-ray diffraction technique to measure the radial and hoop residual stresses around cold
expanded fastener holes in 7050-T76 alloy. Poussard et al. [30] dso used X-ray diffrac-
tion and two-dimensional and axisymmetric finite elernent analyses to evaluate the residual
stress field around a cold expanded hole in 2024T351 alloy. Recently, an axisymmetric
analysis of the cold expansion process, accounting for contact between the mandrel and the
hole surface was conducted by Forgues et ai. [3 11 and Bernard et al. [32].

MATERIAL LOADING CONDITION EXPANSION LIF'

7075-7745 1 Constant amplitude 5% 5
(cm,,= 207 MPa , R = 0.1)
7075-T65 1 Constant amplitude 6% 20
(a,,, = 262 M'Pa, R = 0.05)
7075-T753 1 Constant amplitude 2.8% 2
(orna,= 80 MPa , R = 0.1)
7075-T735 1 Fighter aircraft spectnim 4.7% 4
(amaz= 207 MPa)
7075-T6 Constant amplitude 3 .O% 5

7075-773 Commuter aircraft 3.5% 10

Ioading spectmm
7075-T6 Transport aircraft spectmm 4% 12
(amaz = 127 MPa)
1 7075-T651 1 Constant amplitude
1

5%
I

1 14
(O,, = 262 MPa , R = 0.05)
7079-T6 Commuter aircraft 3.5% 10
loading spectmm 1
* LIF = Fatigue life of treated hole / Fatigue life of untreated hole.

Table 2.1 : Life improvement due to cold expansion (after [22]).

Table 2.1 shows the reported life improvernent due to cold hole expansion in airframe
aluminum alloys. For commercial applications, the hole expansion is typically between
2% and 695, depending on the material and the desired residual stress field. It can be seen
in Table 2.1 that depending on the treated materials, hole expansion levels and loading
conditions, the fatigue life improvement factor (Lm,
defined as the ratio of the fatigue life
of the treated hole to that of the untreated hole, ranges from 2 to 20. It has been shown in
1331 that larger amounts of coldworking reduces the initiation life of fatigue cracks due to
damage at the hole edge. Therefore, an optimum level of hole expansion has to be found
for each matenal, loading conditions and hole geometry.
Extensive search of the literature indicates that no attempt has been made to mode1 the
cold expansion of adjacent holes. The importance of exarnining the effect of separation
distance, size and orientation of adjacent fastener holes is apparent in the damage tolerant
design of airframes. It is with this in rnind that the three-dimensional finite elernent
investigation of cold worked holes was conducted in this thesis.

2.2.1 Cold hole expansion parameters

This section reviews the various parameters that influence the residual stress field due to the
expansion of an open hole and, therefore, the effectiveness of the process. These include
process, geometncal, material and loading parameters [21,221.

Process parameters

One of the major process parameters is the hole expansion level. As mentioned earlier, the
optimum expansion level should be established for each material, loading and geometry.
Another important parameter is the surface upsetting or part thickening which is inherent
in the hole expansion process. This out-of-plane displacement occurs at both the entrance
and exit sides of the hole and is aiways greater at the exit side (lip formation). This results
in a non-uniform residual stress field throughout the thickness of the component, with
greater values of compressive residual stresses at the exit side. Figure 2.6 shows the fatigue
fracture surfaces of two identical specimens which contain (a) an untreated hole and (b)
a cold expanded hole. The figure indicates that for the untreated hole, the fatigue crack
growth is the sarne on both sides of the specimen, resulting in a unifom crack profile. On
the other hand, hole expansion leads to the introduction of a non-uniform residual stress
field which results in a non-uniform crack profile. Subsequent reaming of the cold worked
hole will also influence the effectiveness of the expansion process.

Geometrical parameters

These include the sequence of cold working, hole depth, hole surface finish prior to expan-
sion, hoie straightness, bellmouthing, barrelling, ovality and expansion Ievels for multi-hole
patterns. The influence of the hole geometry before and after expansion on the residual
stress field are of particular interest to this thesis and are discussed in details in chapter 5.
Details of the above mentioned parameters and their influence on the performance of the
cold worked part can be found in references 1211 and [22].

Material and loading parameters

The applied stress level and the material of the cornponent constitute two other important
parameters influencing the residual stress field before and during fatigue crack growth.
The results of Table 2.1 clearly demonstrate the importance of these parameters on the
effectiveness of the hole expansion process. It was shown that a high level of applied
stress c m lead to the relaxation of the compressive residual stresses and, hence, reduce
drarnatically the life of the component [21,221. The effect of these parameters upon the
residual stress field resulting from the cold expansion of adjacent holes is described in
chapters 5 and 7.

2.3 Mixed Mode Crack Growth

In order to predict the fatigue propagation life of cracks emanating from cold worked
holes, knowledge of (i) the residuai stress field, (ii) the stress intensity factors, and (iii) an
appropriate crack tracking criterion is necessary. In this section, we will review the work
related to the evaiuation of stress intensity factors and the modelling of fatigue crack growth
in the presence of residual stresses. Furthemore, we will examine the different criteria
developed for tracking mixed mode crack growth.
 Entry Face

Figure 2.6: Fatigue crack growth of (a) untreated hole, and (b) cold expanded
hole (after [22]).

2.3.1 Stress intensity factors evaluation

The first solution of the problem of an infinite plate containing radial cracks emanating from
a circular hole was given by Bowie [34] in 1956. He used the complex variable method
to evaiuate the reduction of the potential energy and, hence, the critical load according to
Griffith's crack theory. Bowie showed that for a very small crack length (a << R) the
critical load is govemed by the local stress field of the hole. On the other hand, for large
crack Iengths, the effect of the stress field caused by the hole is negligible.
In 1972, Cartwright and Ratcliffe [35] used an experimental cornpliance approach to
determine the strain energy release rate for radial cracks emanating from a pip-loaded hole.
Grandt and Gallagher [36,37] proposed a technique based on linear superposition to obtain
the stress intensity factor directly from the hoop stress surrounding an unflawed hole. The
unfiawed stress distribution may result from remote loading or be introduced either by cold
working or by interference fit fasteners. They found that cold hole expansion resulted in a
dramatic reduction in the effective stress intensity factor by up to 80% (Figure 2.7).
 no cold work

Dirnensionless crack length a/R

Figure 2.7: Stress intensity factors for a cold worked hole (after [37]).

A method to evaluate approximate stress intensity factors for both part-through and
through-the-thickness cracks onginating at holes was proposed by Shah [38]. Both open
holes and holes containing either loaded or unloaded fasteners were considered. The stress
intensity factor solutions for semi-elliptical cracks at a fastener hole in a thick plate were
then used to estimate stress intensity factors for quarter-elliptical cracks originating at a
fastener hole in a finite thickness plate. It was shown that for a quater-elliptical crack
originating at a fastener hole, the stress intensity factor reaches its maximum at the edge of
the hole.
Rooke and Hutchins [39] evaluated stress intensity factors for a radial crack at the
edge of a circular hole subjected to a locaiized radial or tangentid force acting at the
hole boundary. Rooke [40] evduated stress intensity factors for cracks at a double row
of holes and the corresponding fatigue crack growth rates. Rooke's solution, however,
neglected mode-II stress intensity factor and the interaction between holes by assuming that
the direction of crack growth in each row of holes was dong a line through the centers of
the holes. Nevertheless, he concluded that arranging fastener holes in two rows, instead of
one, reduces the stress intensity factor.
In 1983, Grandt and Kullgren [4 11 using a superposition technique developed earlier in
[36], presented tabulated values of stress intensity factors for flawed fastener holes. These
solutions included the cases of through-thickness radial cracks, corner and surface cracks.
In two papers [42,43], Kliman et al. exarnined the effect of cold hole expansion on the
fatigue performance of metallic alloys. First, a theoretical model was proposed which can
predict the residual strain induced by a non-uniforrn plastic deformation, such as the cold
hole expansion method. Using the material stress-strain curve, they concluded that the hole
expansion has an optimum residual strain. Klirnan et al. proposed a method for estimating
the value of the residual strain which would yield the maximum fatigue life enhancement.
They verified their theoretical model with experimental results. No attempt was made to
extend their work to predict the effect of residual stresses upon fatigue crack growth.
A more rigorous analysis of problems containing circular holes with edge cracks was
given by Isida and Nemat-Nasser [44]. Their method was based on the analysis of a
unit rectangular region containing a circular hole with edge cracks, where the boundary
conditions of the outer edges are suitably adjusted in order to treat various problems,
including periodic arrays of holes with edge cracks. They also presented polynomiai fits
of the analytical values of the stress intensity factors which are convenient for practical
applicatioiis.
Even though some of the previous work dealt with cracks emanating from interacting
holes [40,44], the mixed mode fatigue growth of these cracks has not been properly in-
vestigated. Furthemore, no solution has been given for a curved crack originating from
interacting holes.

2.3.2 Criteria for mixed mode crack growth

The three most important criteria currently adopted in predicting both the direction of crack
growth and the cntical applied load under monotonic loading conditions for mixed mode
problems are:
 (i) the maximum tangential stress (MTS) critenon.

(ii) the maximum strain energy release rate (Q)cnterion, and

(iii) the minimum strain energy density (SED) cntenon.

These cnteria are aiso used to predict the direction of fatigue crack growth.

Maximum tangentid stress (MTS) criterion

In this criterion, Erdogan and Sih [45] postulated that the crack will propagate dong a plane
normal to the maximum tangential stress c e in the vicinity of the crack tip, where

and h> and KIr are mode 1and mode II stress intensity factors.
An alternative to this cnterion is the maximum-minimumprincipal stress ( M m )
criterion
according to which the crack is assumed to grow in a direction normal to the maximum
principal stress trajectories.

Maximum strain energy release rate (Ç) criterion

Under normal mode 1 loading conditions, the strain energy release rate Ç is equivalent to
the stress intensity factor, such that

1 1 - v2
G1 = -K:
E
(plane stress) ÇI = 7 if (plane strain)

where v is the Poisson's ratio and E the Young's modulus.

The onset of unstable mode 1 fracture has been successfully predicted by the criticai
value of either the strain energy release rate PIC or the toughness of the material h'lc. In
the derivation of equation (2.2), the crack is assumed to grow under mode 1 loading in a
self-similar manner. This can generally be verified expenmentally. Extending the above
expressions to account for mixed mode 1-mode II loading, the strain energy release rate can
be written as:

1 1-9
Ç = - (K:
E
+ K:,) (plane stress) = -(K:
E
+ K&) (plane strain)
(2-3)
Equation (2.3) again assumes that the crack will grow in the same plane. Since ihis
is not the case for mixed mode Ioading, several attempts have been made to provide more
realistic formulations; see, e.g., [46-5 11. In these attempts, it is assurned that an incremental
extension from the tip of the crack takes place in a radial direction. It also stipulates that
crack growth takes place in the direction which maxirnizes the strain energy release rate Ç
and that crack growth occurs when reaches a critical value corresponding to pure mode 1
loading (grc).

Minimum strain energy density (SED) criterion

This cnterion considers the strain energy concentrated in the crack tip region as being the
driving fracture parameter [52]. Assurning unit thickness, the strain energy density in the
vicinity of the crack tip can be expressed as

where W is the strain energy, A the area, and S the strain energy density factor which is
given by
S = aiih'; + 2 a i z h ' I l ~ r+ a 2 2 ~ & (2-5)

with a;, being a function of 0 and the matenal properties [53]. Under these considerations,
the criterion stipulates that crack growth takes place in a direction dong which the strain
energy density factor S reaches a minimum and that crack growth occurs when S reaches
a critical value.
To illustrate the crack growth direction that these critena predict, consider the problem
of an angled-crack under uniaxial loading, as shown in Figure 2.8. The predicted direction
is shown in Figure 2.9 as a function of the inclination B. The MTS and Ç criteria predict
exactly the same angle, while the SED criterion gives slightly different predictions.
 A fundamentai difference amongst the aforementioned critena is the fact that the Ç-
criterion is based on therrnodynarnic energy balance considerations, whereas the MTS and
SED criteria are based on axiomatic considerations established using theoretical and ex-
perimental observations. Specificdly,the Ç-criterion is based on the fundamentai principle
of minimum potential energy and thus satisfies the instantaneous equilibrium of a cracked
geometry. In this case, the crack growth direction is that which rninimizes the potential
energy or rnaximizes the strain energy release rate. This fact gives more credibility to
the Ç-cnterion over the other criteria. Furthemore, the Ç-criterion is the only one that
can be used in the case of generai loading conditions e.g., in-plane and anti-plane loading.
However, the MTS and SED criteria are easier to apply, since they are based directly on
the stress and strain fields existing at the crack tip prior to crack extension. The predictions
based on the above criteria show signifiant variations for monotonie Ioading conditions
and their applicability to fatigue crack growth has been investigated only for simplified
geornetries and for situations where mode I prevails 154,551.
Figure 2.8: An angled-crack under uniaxial loading.

Figure 2.9: Prediction of crack growth direction.

Chapter 3

Analysis of Multiply Connected Regions

The mechanical integrity of interacting hole systems is governed, amongst other parameters,
by the stress field at which fatigue cracks initiate and propagate to ultimate failure. This
chapter is devoted to the treatment of this complex problem using a new and unified
method. In addition to being relatively simple and flexible, the newly developed method is
capable of treating simply and multiply connected regions under complex Ioading conditions
accurately. This generalized approach, as will be seen in the following chapter, can be
extended to treat interacting cracks and microdefects.

3.1 Elastic Field Analysis of an Arbitrarily Shaped Hole

3.1.1 Basic equations

The situation envisaged is that of an arbitrarily shaped hole in a homogeneous isotropic
elastic medium under quasi-static plane loading, as shown in Figure 3.1. The hole is sub-
jected to prescribed normal and shear stresses almg its boundary. The hole is approximated
by N line segments. The respective half-length and orientation angle of the jth segment are
denoted by aj and bjj,with j = 1,2, - - , N. A global coordinate (x, y) and local coordinate
(xj,yj) system are used to describe the hole with the otigin being located at the centre of
segment j. The stress field of each segment is generally characterized by the normal and
tangentid deformations of its surface. These can be formulated as the superposition of a
series of distributed dislocations.
There are two types of dislocations which correspond to two kinds of deformations:
normal and tangential, as depicted in Figure 3.2. The first kind of dislocations is associated
with a displacernent Au = br while the second kind with a displacement Au = bIl dong a
semi-infinite Iine, y = 0,x > O. The plane strain stress fields resulting from the first kind
of dislocations c m be expressed as

- x')
y ) / b r = F i ( x ,Y ) =
&x,
+
( x 2 y2)2

while the second can be expressed as

with cr = E/[4rr(1 - d)]

and F i ( x ,y) and G i ( x ,y ) being the fundamental soIutions of a
- - - - - - - - - -
- - - - -
p p - p p p p - - - - - -

single dislocation [56].

Let us now describe the deformation of the j-th segment in ternis of the following
dislocation density functions

with Avj and Auj being the normal and tangential deformations. The stress field due to
this segment can then be obtained by superimposing al1 distributed dislocations in the local
Figure 3.1: Arbitrarily shaped hole: (a) exact geometry, and (b) approximate
geometry using line segments.

Figure 3.2: Dislocations associated with (a) normal, and (b) tangential defor-
mations.
coordinate system (xj, yj), such that:

The local stress components in (xj,y j ) given by (3.4) can be transferred into the global
coordinate system (x, y), Figure 3.1, by using the following transformation

where

1$ sin 2$ - $ sin 24 cor 29

1
is the transformation matrix. Accordingly, the stress field due to the entire hole can be
expressed in the global coordinate system (x,y) in terms of fi and gj as being

where xj and yj are the local coordinates of segment j and should be replaced by the global
coordinate system using the following transformation
 1 I
Yj = [Y - 5(Yj + cos #j - [X - -(zj
2
+ sin $j
zj+i)]

where (zjv
y,), j = 1,2, - - , N are the edge points of the different segments, as depicted in
Figure 3.1.

3.1.2 Integral equations

The arbitrarily shaped hole is assumed to be subjected to prescribed normal and shear
stresses dong its surfaces, such that

Consider now the stress boundary conditions given by (3.9). By using the general expression
for the global stress field (3.7) and the inverse transformation of ( 3 . 3 , the stress distribution
dong segment k (yk = 0) cm be obtained in rems of the unknown dislocation density
functions f, and gj using (3.9), such that

where z3 and y; are the coordinates of the boundary point (xk,0) in the (xj,yj) coordinate
system and are given by
 { ~ ~ ] = { ~ 0 ~ ( 4 k - # j ) ][COSA
~ ~ + sin41
sin(4k -#j) - sin 4j COS^,

with

sin2 #
? sin
2
cos2 4 sin24
24 - f sin 24 cos 24 1
is a transformation matrix. When j = k, x; = x k and y; = yk = O. Therefore,

By making use of (3.10), (3.13) and (3.14), the stress boundary conditions (3.9) at the hole
surface cm be further expressed as

The singular integral equations (3.15) are the governing equations of the arbitrarily
shaped hole problem in ierms of the dislocation density functions f, and g j .
3.2 Solution of Singular Integral Equations

3.2.1 Chebyshev polynomial expansions

Equation (3.15) can be expressed in terms of the following expansions of Chebyshev
polynomials [57].

with LI, being the second kind of Chebyshev polynornials. Substituting (3.16) and (3.17)
into (3.15) and making use of the relations [58.59]:

and

sin 0 sin(rn6)dO
G,(X;- a j COS 9,y;)
equation (3.15) can be rewritten as

where Tm is Chebyshev polynomiai of the first kind, with

and

where x; and y; are given by (3.1 1).

3.2.2 Determination of Chebyshev polynomial coefficients

The unknown coefficients A$ and B i = 1,2, . - . , N ; rn = 0,1, - - .,oo) are determined

by truncating the Chebyshev polynomial expansions to the &th term, and hence reducing
the number of unknown parameters to 2 N ( M + 1 ) . The solution for these pararneters can
be obtained by considering the boundary conditions at appropriate collocation points d o n g
the hole boundary. If we select .Mkcollocation points for segment k, which are given by
then the stress boundary conditions (3.20) become

+ 1 equation
By selecting the number of collocation points for each segment to be Mk = M
(3.24) provides 2N( M + 1 ) equations which c m be used to determine 2 N ( M + 1 ) un-
known coefficients of Chebyshev polynomial expansions A:, Bm (k = 1,2, .: -.;N ; rn =
O, 1, .. ,M ) .

3.3 Tangential Stress Field

The coefficients of Chebyshev polynomial c m now be used to evaluate the stress field at any
point of the hole boundary. %y using the generai expression for the global stress field (3.7)
.
and the inverse transformation of (3.5) the tangentid stress distribution dong segment
C ( y k = 0) can be obtained in terms of the dislocation density functions fj and gj, such that
Substituting equations (3.16) and (3.17) into (3.25) and making use of relations (3.18) and
(3.19) the tangentid stress of segment k can be expressed as

w here

FI(x; - a j COS 6, Y;)

il" F*(Z;- aj COS 6. y;)

F3(xj - aj cos 8, y;)
sin 0 sin(md)d6

Ln sin 6 sin(m6)dB

with xi and y; given by (3.1 1).

3.4 Multiple Interacting Holes

The developed mode1 can be easily extended to account for multiple interacting arbitrarily
shaped holes. Consider the problem of K arbitrarily shaped and located holes in a homo-
geneous isotropic elastic medium under quasi-static plane loading, as shown in Figure 3.3.
Each hole is subjected to prescribed normal and shear stresses dong its boundary. Let us
approximate the j-th hole with N j line segments with j = 2 K . In this case, the
solution of the problem can be obtained by superimposing expressions (3.1) to (3.28) and
substituting N = N I + Nt + -.-'+N K .
It is worth mentioning that the above developed method is a new and unilfed approach
to treating two dimensional multiple interacting hole problems with complex geometric
configurations and generd loading conditions. Unlike earlier efforts [8,9.11], the current
approach cm be used to treat multiple interacting holes of any shape quite readily without
the need for superimposing the fundamental solutions corresponding to these holes.

Figure 3.3: A schematic of multiple interacting holes.

The newly developed mode1 is verified against known solutions for the problerns of an
infinite plate containing a circular hole, an elliptical hole, and two equal circular holes
under uniaxial loading.

3.5.1 Circular hole

Consider the problem of a circular hole in an infinite medium subjected to uniaxial loading.
The exact distribution of the nomaiized tangential stress on the boundary of the circular
hole is cornpared in Figure 3.4 with the values obtained from the current method using
forty terms in Chebyshev polynomiai expansions and sixteen line segments representing
the hole. An excellent agreement is observed.

Figure 3.5 shows the cornparison between the exact distribution of the normalized tangential
stress at the boundary of an ellipticai hole and the current rnethod for a loading at infinity
parallel to the minor axis of the ellipse. Again, excellent agreement is observed when
forty terms in Chebyshev polynomial expansions and sixteen line segments representing
the elliptical hole are used.

3.5.3 ' h o circular holes

In this example, we consider the problem of two equal circular holes in an infinite medium
under remote uniaxial loading, Figure 3.6, examined earlier by Meguid [IO]. The stress
concentration evaluated using twenty line segments, are compared with the results of Meguid
[ 1O] in Figure 3.7. The agreement between the two solutions is remarkable.
Figure 3.4: Normaiized tangential stress distribution around a circular hole.

Figure 3.5: Normalized tangential stress distribution around an elliptical hole.

 00

Figure 3.6: Two equai holes under uniaxial loading.

Figure 3.7: Normalized tangentid stress distribution for two holes.

Chapter 4

Analysis of Generalized Curved Cracks

To accurately descnbe the fracture behaviour of a solid containing a generdized crack, it

is necessary to account for the complex geometry of the crack in the calculation of the
stress intensity factors. It is therefore the purpose of this chapter to describe the approach
adopted in developing a new technique for treating interacting generaiized cracks under
plane loading. This technique is based on the mode1 outlined earlier in Chapter Three.
However, it is modified to account for the singularity of the stress field at the crack tips.
This singularity has been accounted for using Chebyshev polynomials of the first kind. The
newly developed stress intensity factor solutions were then used with the residual stress
intensity factors to account for the effect of applied and residuai stresses upon the effective
singularity of the resultant field.

4.1 Formulation of the Curved Crack Problem

4.1.1 Basic equations

The situation envisaged is that of an arbitrarily shaped curved crack in a homogeneous
isotropie elastic medium under quasi-static plane loading, Figure 4.1. The crack is subjected
to prescribed normal and shear stresses dong its surfaces. The curved crack consists of N
subcracks. The respective haif-length and orientation angle of subcrack j are denoted by
aj and + j j , with j = 1,2, - ,N. A global coordinate (x, y) and local coordinate (xj7y j )
systems are used to describe the cuwed crack with the origin king located at the centre of
subcrack j. The stress field of the crack is generally characterized by the opening and sliding
modes of deformations of the crack surfaces. These c m be formulated as the superposition
of a series of distributed dislocations. The plane strain stress fields resulting from these
dislocations are given by equations (3.1) and (3.2). The stress field due to the entire curved
crack is given by equation (3.7).

(XI9 71)
Figure 4.1: Segmented arbitrady shaped curved crack.

The curved crack is assumed to be subjected to prescribed normal and shear stresses along
its surfaces, such that

In addition, the normal (Av) and the sliding (Au) modes of elastic deformations of the
crack surfaces must be zero at the crack tips, i.e.
This additional continuity condition can be expressed in terrns of f' and gj by integrating
fj and gj dong the curved crack, as ~ O ~ ~ O W S :

The singular integral equation (3.15) and the continuity condition (4.3) form the gov-
erning equations of the curved crack problem in terms of the dislocation density functions
fj and g j -

4.2 Solution of Singular Integral Equations

4.2.1 Chebyshev polynomial expansions

The behaviour of the singular integrai equations (3.15) and (4.3) is governed by a square-
root singularity 1571. Therefore, the general solution of these equations can be expressed
in terms of the following expansions of Chebyshev polynomials

with Tm being the first kind of Chebyshev polynomials. Substituting (4.4) and (4.5) into
(3.15) and making use of the following relations [58,59]:
the stress boundary conditions (3.1 5) can be rewritten as

00

r a ~m=I
um-I(2)
ak
{ }+
A
'
Bk
N
xaj
jfk
{g{ p j m k ( X k ) ]+A[~Q j m k ( z k ) l B i ) }
m=O

where Um is a Chebyshev polynomial of the second kind, and

with xf and y; given by (3.1 1). By making use of (4.4) and ( 4 3 , the continuity conditions
(4.3) are reduced to

eaj
i=l
[COS#,

sin d j
-sin4j]
COS 4j
{ ] {O]=

which provide two additional equations for the solution of the curved crack problem.

4.2.2 Determination of Chebyshev polynomial coefficients

Let us now consider the approach adopted to determine the unknown coefficients AL and
~ ,N ; m = 0,1, -*,
B ~ G =1 , ,S..... m). By tnincating Chebyshev polynomial expansion

to the Mth term, the number of unknown parameters is reduced to 2 N ( M + 1). The
solution for these parameters can be obtained by considering the boundary conditions at the
appropriate collocation points dong the crack surfaces. If we select M kcollocation points
for subcrack k, which are given by

then the stress boundary conditions (4.9) become

For the present single curved crack problem, we take MI = M and M k = M + 1 with
k > 1. Equation (4.14) and the continuity condition (4.12) provide 2N ( M + 1) equations
which c m be used to determine 2 N ( M + 1) unknown coefficients of Chebyshev polynomial
4.2.3 Stress intensity factors
Similar to the traditional straight crack, the stress intensity factors for the present curved
crack can be expressed as

l e t = iim
Zl-F-Ul
J-d)(~l, 0)

By substituting Chebyshev polynomial expansions (4.4) and (4.5) into the stress field
expression (3.10) dong subcracks 1 and N, the stress intensity factors at the left and the
right tips of the curved crack can be expressed as

4.3 Verification of the Mode1

The newly developed mode1 is verified against known solutions for two main problems:
(i) the problem of a crack emanating from a circular hole under uniaxial loading, (ii) the
problem of a kinked crack, and (iii) the problem of two interacting cracks.
4.3.1 Crack emanating from a circular hole
Consider now the problem of a crack emanating from a circular hole subjected to cy, = 0
at infinity. Instead of dealing with this complex boundary condition directly, we mode1 the
hole using a number of subcracks of equal length. The normalized stress intensity factor
K* = K r / 6 obtained from the current method for different subcrack configurations
using forty terms in Chebyshev polynornial expansions (4.4) and (4.5) are compared with
the corresponding result of Bowie [34] in Figure 4.2. A very good agreement is observed
between the two when ten subcracks are used in the modelling of the hole. This example
clearly indicates the ability of the present method to treat complex crack problems with
ease. Furthermore, the current method is not restricted to a straight crack emanating from a
circular hole as the methods reported in the literature [34-411. It will be shown in chapter 7
that the current method cm be used with no modification to reliably evaluate stress intensity
factors for curved cracks emanating from interacting holes.

Figure 4.2: Normalized stress intensity factor K j of an edge crack emanating

from a circular hole.
4.3.2 The kinked crack problem
Figure 4.3 shows a kinked crack subjected to o, = 00 at infinity. The normalized stress
intensity factors ( K I , K r [ )= ( f i ,Krr) / f iagainst e / a for two different kink angles
are shown in Figure 4.4. The values of the stress intensity factors obtained frorn the current
method using forty terms in Chebyshev polynomiai expansions are in very good agreement
with those of Lo [60].The current method, however, is not restricted to one branch of the
kinked crack and can be easily used to evaiuate stress intensity factors for an arbitraïly
shaped curved crack.

Figure 4.3: Details of geometry of a kinked crack.

4.3.3 Interacting cracks

The model was extended to examine the effect of two adjacent curved cracks upon their
corresponding stress intensity factors. In this case, the crack segments represented by their
dislocation density are assigned appropriate numbers so as to enable the modelling of the
cracks and the separation distance between their tips.
Figure 4.5 shows the problem of two colinear cracks under uniaxial loading for which
an exact solution exists [61]. In this case, the model allows us to describe the separation
distance between the cracks (segments 1 and 3, Figure 4.5) using "very small" values of
 1.2 -
1.0 -
0.8 - K, current
k - - - - - K,, current
0.6 - 0 4 [6ol

.
K,, F O I
0.4 -
--*-----------.-------------------O
0.2 -
0.0 - I

0.0 0.1 0.2 0.3 0.4 0.5

e/a

0.8 1 8 1 1

- _______--------
- - - - ----*---
0.6
-a- -
,
a-
K, current
ox0.4 -
6,current

0.2 -
0 K, 1601
4,[601 1i
-1
Li

0.0 - 1 I 1 ,
0.0 O. 1 0.2 0.3 0.4 0.5
ela

Figure 4.4: Normalized stress intensity factors A"of a kinked crack: (a) kink
angle # = lSO, and (b) kink angle 4 = 75O.
dislocation density in segment 2.
Figure 4.6 shows the effect of the separation distance 21 between the interacting cracks
upon the resulting stress intensity factors Kin and KOut.Excellent agreement is obtained
between the newly proposed and the exact analytical solution.
It is worth mentioning that there exists weak singularities in the f o m of r-" with
O <a < 112 at the corners between neighboring subcracks. These singularities would
ultimately affect the local stress field around these corners. Since Our interest is focused on
the crack tip field, the effect of these singularities have been ignored in the current analysis.
The numericai exarnples examined in the present section have demonstrated that the
above treatment is appropriate for the determination of the stress intensity factors for
complex crack geometries. The proposed rnethod is general and simple to apply. By
selecting different subcrack configurations, the present approach can be used to simulate
complex crack problems under plane loading.
 1 out in

(a) Two colinear cracks (b) Simulation of two cofinear

cracks using dislocations

Figure 4.5: Schematic of two colinear cracks under uniaxial Ioading.

Normalized separation distance, Va

Figure 4.6: Variation of normalized stress intensity factor with normalized

distance between crack tips.
Chapter 5

3-D Finite Element Analysis of Cold

Worked Adjacent Holes

As indicated earlier, residual stresses resulting from the inhomogeneous plastic deformation
of cold worked fastener holes extend the fatigue life of the hole. It is the purpose of this
chapter to employ the finite element method to characterize the mechanically induced
residual stress field. Existing work in this area has concentrated on a single hole model,
where the residual stress field can be obtained from simple anaiytical and numerical two-
dimensional analyses. Extensive search of the literature indicates that no attempt has been
made to model the cold expansion of adjacent holes. The importance of examining the
effect of separation distance, size and orientation of adjacent fastener holes is apparent in
the darnage tolerant design of airframes. This chapter addresses these issues using three-
dimensional elasto-plastic finite element analysis of cold worked holes. However, for the
sake of cornparison with the three-dimensionai solution as well as completeness of the work,
some attention was devoted to two-dimensional analysis. This will allow us to establish the
baseline for the field variables.
5.1 Two-dimensional Finite Element Analysis

5.1.1 Details of two-dimensional geometry

A plane stress finite element mode1 was conducted using a plate containing two holes each
of radius R with a centre distance 2c between the holes, as shown in Figure S.l(a). The
dimensions of the plate were: width w = 40R and height h = 40R. These dimensions
were determined carefully, so as to avoid the effect of the boundary on the results. In view
of symrnetry, only one quarter of the plate was discretized. Six-noded triangular elements
were used. A higher density mesh was used in the neighbourhood of the hole to capture the
large stress variation in that region. A number of convergence runs were conducted using
different discretized geometries. As a compromise, the mesh provided in Figure 5.1 (b) was
selected for the two-dimensional analysis. Throughout this work, the ANSYS finite element
code was utilized. The expansion level was imposed on the hole boundaries by applying a
radial displacernent at these boundaries. This assumes a rigid rnandrel configuration.

5.1.2 Results and discussion

In this section, we sumrnarize the resuits obtained from the two-dimensional finite element
modelling of two adjacent fastener holes. The material properties used were that of alu-
minium alloy 7075-T65 1 with E=72 GPa and v=0.32. The elasto-plastic behaviour was
assumed to be bilinear with isotropie hardening and an initial yield stress of 506 MPa.
Figures 5.2(a) to 5.2(c) show the variation of the normalized equivalent stress trajectories
for three different expansion levels (2%, 4% and 6%), while Figure 5.2(d) shows the
unloading residual stress field corresponding to the 6% expansion level. The figures show
clearly the growth of the plastic zone with the increase of the expansion. They also show
interaction effects in the region contained between the two fastener holes. Figures 5.3(a)
and 5.3(b) show the variation of the normalized tangentid residual stress versus the distance
dong the lengths AB and CD, measured from the central line of the workpiece, respectively.
Figure 5.3(a) reveais the presence of tensile residual stresses at the central distance between
the holes for an expansion level of 2%. Increasing the expansion level results in a reduction
Figure 5.1 : Cold expansion of two adjacent holes: (a) schematic of geometry
used, and (b) finite element mesh of the shaded region.
in the tensile field, leading to a purely compressive residual stress field at an expansion
level of 6%. This is compensated for by the increase in the tensile residual stress dong the
length CD of Figure 5.3(b). Figure 5.3(b) also reveals that maximum compressive residual
stresses occur at the hole boundary.

Figure 5.2: Details of equivalent stress a contours for different expansion lev-
els: (a) 2%, (b) 4%, (c) 6%, and (d) unloading from 6% expansion.

Figure 5.4 shows the normalized tangentid residual stress versus the distance along the
length AB for an expansion of 6%. The figure reveals the relatively high tensile residual
stresses for the case where 2c=6R. No tensile residual stresses were observed for 2c=4R.
Figure 5.5 shows the insignifiant effect of the strain hardening rate upon the residual
stress field along length AB. In al1 the above trials, the two fastener hoies were expanded
simultaneously.
We now tum our attention to the case where the expansion was conducted sequentinlly.
Figure 5.6 shows the discretized mesh for the geometry. In this case. an aiternate expansion
of 4% was applied to each of the holes. Figure 5.7 shows the residual stress results for
a single hole expansion; sequential expansion; and simultaneous expansion of the two
fastener holes. The figure reveals the drarnatic difference resulting from the order of the
expansion. It indicates that large tensile residual stresses are deveioped in the case where
the mandrel is applied sequentidly to the holes.
 0.2 0.4 0.6 0.8
Normaiized distance, x/(c- R)

level

4 5
Norrnalized distance, xl(c- R)
(b)

Figure 5.3: Variation of normalized tangentid residual stress oilcr, dong

lengths (a) AB, and (b) CD for 2&R and H'l E=I %.
Figure 5.4: Variation of norrnalized tangentid residud stress oi/ov dong
length AB for a 6%expansion and Hr/E=û.1%.

Nonnalized distance, rd@- R)

Figure 5.5: Variation of normdized tangential residud stress crilo, along

length AB for a 6%expansion with 2 d R .
Figure 5.6: Finite element mesh used for the sequentiai expansion of the two
holes.

- --
- 1.O -0.5 0.0 0.5 1 .O
Nomalized distance, x/(c-R)

Figure 5.7: Variation of normaiized tangentid residual stress o,'/o, for se-
quential expansion with 2&R, 4%expansion and Hf/E=û. 1%.
5.1.3 Limitations of the two-dimensional mode1

The two-dimensional model described above suffers from the following shortcomings.
First, it ignores the asymmetrïc nature of the current cold expansion process, as evidenced
by the difference in the residual stresses at the entry and exit faces of the workpiece.
Second, it did not account for contact conditions at the interface between the rnandrel and
the hole. It assumes a rigid mandrel, as evidenced by the imposition of the displacement
at the boundary of the workpiece. Findly, it is incapable of accounting for friction as
well as surface roughness variations through the thickness of the plate during the cold
expansion process. These shortcomings will manifest themselves in the deformed geometry,
development and growth of plastic zone and unloading residual stresses. This has motivated
the three-dimensional finite element modelling described below.

5.2 Three-dimensional Finite Element Analysis

5.2.1 Details of three-dimensional model

The situation envisaged is that of a plate containing two holes each of radius R with a centre
distance c between the holes, as shown in Figure 5.8(a). The dimensions of the plate were:
width w = 2c + 12R, height h = 12R and thickness t = 1.2R. These dimensions were
determined carefully so as to avoid the effect of the boundary. The results presented were
obtained for R=2.5 mm. In view of syrnrnetry, only one quarter of the plate was discretized.
Twenty-noded hexahedral elements were used in the neighborhood of the rnandrel to capture
the large stress variation in that region. The remaining region of the plate was conveniently
discretized using ten-noded tetrahedral elements. A number of convergence mns were
conducted using different discretized geometries. As a compromise, the mesh provided in
Figure 5.8@) was selected for the investigation.
The interface between the mandrel and the hole surface was modelled using contact
elements. The three-dimensional contact element used, depicted in Figure 5.9, adopts a
contact node-target segment approach in conjunction with the penalty function method.
Elastic Coulomb's friction with a coefficient of friction p = 0.15 was used.
 Thickness t = 1.2R

(a)

Figure 5.8: Cold expansion of two adjacent holes: (a) schematic of geornetry
used, and (b) finite element rnesh.
 A major problem in the implementation of contact elernents is the assignment of values to
the normal (K,) and tangential ( K t )stiffnesses, which govern the convergence and accuracy
of the solution. The values of K, and h; are required to be very large. However, the use
of excessively high values of K, and Kt results in ill-conditioned global stiffness matrices,
leading to numencal errors and divergence. On the other hand, the use of smaller values of
K, and Kt results in convergence to the wrong solution allowing for large interpenetration
and incorrect estimates of the stick and slip regions. In the current sîudy, the appropriate
values of the normal and tangential stiffnesses were selected from convergence tests, where
only minimal interpenetration was allowed. As a result of numerous test runs, the following
values of K, and Kt were selected: 10" and 108(N/m), respectively.

Contact Surface and Node

"1 Target Surface and Nodes

Figure 5.9: Schematic of three-dimensional contact element used.

5.2.2 Results and discussion

This section outlines the results obtained from three-dimensional finite element rnodelling
of the cold expansion process. The material properties used for the workpiece were that
of alurninum alloy 7075-T651. The mandrel's material was that of high strength hardened
steel. Details of the mechanical properties of these two materials are provided in Table 5.1.
 1 Property 1 Workpiece 1 Mandrel (
E (GPa)

506 600
Table 5.1 : Material properties of workpiece and mandrel.

Validation of the model

The current finite element model was validated by comparing with the experïmentai mea-
surernents conducted by Link and Sanford [28]. Link and Sanford obtained the residual
strain field surrounding a single split-sleeve cold expanded hole in 6.35 mm thick 7075-
T651 aluminum plates using moiré interferometry. Numerous finite element mns were
conducted to examine the effect of the centre distance upon the field variables. The work
concluded that virtually no interaction efTects are present for the case where 2 c l R 2 10.
Accordingly, the geornetry shown in Figure 5.8(a) with R=2.87 mm, 2 6 2 8 . 7 mm and
t=6.35 mm was discreîized for cornparison with the existing results of Link and Sanford
for a 3% expansion. The coefficient of friction used for a lubricated steekteet arrangement
was selected to be 0.15 [62].The validation was aiso conducted using different coefficients
of friction. The results reveal the insignifiant influence of p upon the residual strain. For
example, varying the coefficient of friction from 0.1 to 0.4results in a 4% difference in the
maximum equivalent strain. Figure 5.10 shows the residual radial and tangential strains
at the entry face of the specimen. Good agreement between the experimentally obtained
distribution and the current finite element predictions is observed.

Characteristics of residual stress field

To investigate the effect of the geometry, the centre distance between the two holes was
varied from 3R to 10R.The outer boundary of the plate which was ngidly fixed was selected
such that it does not affect the results in the regions surrounding the expanded holes. The
total mandrel displacement (16 mm) was simulated using a 20-step displacement history.
The von Mises stress contours for a centre distance of 4R at four different mandrel
positions is shown in Figure 5.1 1. Both the entry and the exit views are provided. At
 1 2 3 4
Normalized distance tiom hole centre (rlR)

Figure 5.10: Cornparison with residual strain distributions

using interferometry.

the initial entry stage Figure 5.1 1(a), the mandrel is in contact with a small fraction of the
fastener hole. In this case, the equivalent stress trajectories at the entry surface reved the
limited contact between the mandrel and the plate. When the mandrel is pushed further into
the plate Figure 5.1 1(b), both surfaces experience comparable Ievels of stresses. However,
with the additional travel of the mandrel into the hole Figure 5.1 l(c), the equivalent stress
field at exit surface is dramatically increased. Figure 5.1 1(d) shows the exit stage of the
mandrel. The figure reveals the remarkable difference between the equivalent stress at
exit and entry surfaces. This shows clearly the inability of two-dimensional modelling to
accurately capture the plastic zone development and unloading residual stresses resulting
from the cold expansion of two adjacent holes.
The variation of the normalized equivalent stress a/a, along the length AB (see Fig-
ure 5.8 for details) for the four different mandrel strokes discussed earlier is shown in
Figure 5.12. At mandrel strokes of 10%and 25% (Figures 5.12(a) and 5. L 2(b)) the normal-
ized equivalent stress is much higher at the entry than at the exit faces. However, additional
mandrel travel causes both faces to plastically yield. In this case, the equivalent stress varia-
 Entry View Exit View

Figure 5.1 1: Details of equivaient stress i? contours at entry and exit for differ-
ent mandrel strokes: (a) IO%, (b) 25%, (c) 60%, and (d) 100%.
tion is comparable at a mandrel stroke of 6O%, as depicted in Figure 5.12(c). Figure 5.12(d)
shows the residual distribution of the normalized equivalent stress. The figure indicates
that at the hole edge ( x / R = 1). the residual equivalent stress is much higher at the exit
face. This would lead to higher compressive tangentid residual stresses at the exit face of
the workpiece.

C
a-
*-**

- Entry (A-B)
-- Exit (A-B)

Exit (A-B)

Normdizcd distance (xlR)

Figure 5.12: Variation of normalized equivalent stress s/u, dong length AB

for different mandrel strokes: (a) 1096, (b) 25%, (c) 60%, and (d)
100%.
 The variation of the normalized equivdent stress b / o , across the thickness of the plate
at point B is shown in Figure 5.13 for four different mandrel positions. The figure clearly
indicates that during the entry stage (10% and 25% strokes) the equivalent stress is much
higher at the entry than at the exit surfaces. At 60% stroke, the equivaient stress distribution
is uniform across the thickness. The residual equivalent stress field (mandrel stroke 100%)
aaains a maximum value at mid-thickness and a minimum value at the entry face.

Distance across thickness (mm)

Figure 5.13: Variation of normalized equivalent stress 6/a, through thickness

at point B (Figure 5.8).

The effect of the centre distance between the two expanded holes upon the resulting
residual stress field is shown in Figure 5.14 for a 2% expansion. The normalized tangential
residual stress at point B is plotted across the thickness of the specimen for the three
different centre distances where the interaction effects are dominant. The resu1ts show that
the centre distance of the two cold expanded holes can lead to a significant change in the
residual stress field. The figure also shows that a decrease in the centre distance between
the expanded holes results in a decrease in the compressive residual stresses at the entry
face and an increase at the exit face of the workpiece. These very low compressive residual
stresses have an insignificant effect on the fatigue performance of the cold worked piece.
Figure 5.15 shows the variation of the tangential residual stress contours around the cold
worked hole for the centre distances examined. The figure reveals the large variation in
tangential residual stress in the region of influence.

Entv Exit 1

1 2
Distance across thickness (mm)

Figure 5.14: Viuiation of norrnalized tangential residual stress o ~ / othrough

Y
thickness at point B.

Figures 5.16(a) and 5.16(b) show the residual stresses at the entry and exit faces of the
workpiece for a centre distancc of 2&R. The figures reveal that the smaller compressive
residud stresses are always at the entry face of the workpiece. In the region between the
two holes (Figure 5.16(a)) the residual stress changes rapidly from compressive close to the
hole edges to highly tensile at the central point A. Figure 5.16(a) also shows the inability of
the two-dimensional model to predict the variation of the residual stress field at the entry
and exit faces of the workpiece.
Figure 5.16(b) aiso shows that interacting effects between the adjacent holes can lead to
high tensile residual stresses, even along the length CD. These high tensile residual stress
along lengths AB and CD combined with any tensile stresses resulting from the applied
service load could lead to premature failure under conditions of cyclic loading. Again,
the resu!ts reveal the inability of the two-dimensional model to predict the residual stress
variation at the entry and exit faces of the plate.
Figure 5.15: Tangentid residual stress o,'/o, contours for 2% expansion.
 --- E ~ Q3-D
-- Exit 3-D

0.2 0.4 0.6 0.8

Nonnahzed distance, xl(c-R )
(a)

-
4

--- Entry 3-D

- - Exit
- 2-D 3-D

-1.5 1
3 4 5 6
Normalized distance, x/(c-R )

Figure 5.16: Variation of normalized tangentiai residuai stress o;/av dong

lengths (a) AB, and (b) CD for the entry and exit faces of the
workpiece.
 In conclusion, the three-dimensional finite element analysis of cold expansion of two
adjacent holes revealed the following:
the compressive residual stress varies across the thickness of the specimen, with the
maximum k i n g around mid-thickness and the minimum at the entry face,

two-dimensional finite element models are incapable of accurately predicting the

residual stresses resulting from the cold expansion process,

excessive cold expansion of adjacent holes c m lead to high tensile residual stresses,
and

a decrease in the centre distance between the expanded holes results in a decrease in
the compressive residual stresses at the entry face of the workpiece.
Chapter 6

Experimental Investigations

This chapter comprises three sections. The first deals with the selection of the materials
and the details of the test specimens. Section two is devoted to the hole expansion rnethod,
while section three outlines the experirnental details relating to fatigue testing and fatigue
crack growth measurements.

6.1 Materials and Test Specimen

The materid considered in this investigation was high strength Al-Zn 7075-T651. In view
of its high specific strength, resistance to corrosion and high fatigue resistance, Al-Zn alloys
are currently being used in various airfrarne applications. Rolled Al-Zn 7075-T65 1 alloy,
aged to its peak strength, was received in 3.2 mm plate f o m from ALCOA. The chernical
composition is given in Table 6.1. The designation T651 refers to a precipitation heat
treatment which includes a solution treatment at 480 O C , a stress relief by a 2% stretching
and a 24 hour aging at 120 O C .
MicrostructuraI analysis [64] revealed anisotropic microstructure in the Al-Zn 7075-
T65 1 alloy, with an average grain size of 20 pn, 80 pm and 150 pm in the short transverse
(ST), the long transverse (Lnand the longitudinal (L) orientations, respectively, with the
L orientation being the rolling direction.
 Tensile tests were carried out on specimens machined in different orientations and taken
at different locations of the plate. The measured mechanical properties of Ai-Zn-T651 alloy
is given in Table 6.2. The elongation was measured using a 25 mm gage length and the
results represent the average of three tests with a discrepancy of less than 2% between tests.

Table 6.1: Chemical composition of Al-Zn 7075 alloy (wt%) (after [65]).

Orientation E (GPa) 00.2 (MPa) u s) Elongation (%) KI, ( M P a . 6 )

L 72 506 562 13 28
LT 71.2 508 590 11.5 28

Table 6.2: Mechanicd properties of Al-Zn 7075-T65 1 alloy.

6.1.2 Specimen design and preparation

Uniaxiai fatigue tests were conducted on rectanguiar samples containing two circular holes
of equal diameter. The detailed geometry of a typical sample is shown in Figures 6.1.
Table 6.3 provides the values of the various quantities considered for ail the sarnples used
in the current investigation. The tolerances for the dimensions of Figure 6.1 were fIO pm
The values of the centre distance were selected to ensure strong interaction effects
between the adjacent holes. Accordingly, two values of 7.5 and 10 m m were selected, i.e.,
3 and 4 times the hole radius, respectively. Furthemore, in order to examine the effect
of the inclination of the two holes with respect to the applied uniaxial load, five different
inclination angles were selected; namely: 0'. 30°,45", 60' and 90". To avoid boundary
effects, the two holes were placed near the centre of the specimen.
To minimize expenmental scatter in the fatigue data resulting from the machining
operations and the surface finish of the specimen, the machining operations were carefully
monitored and documented for future reference. The machining procedure resulted in a
 Thickness 3.1 mm

Figure 6.1: Detailed geometry of a typical sample.

Table 6.3: Geometncd details of fatigue specimens (in mm).

final surface finish Raof 0.4-0.5Pm. The surface roughness measurements were conducted
on five specimens which were randomly selected. To allow the detection and accurate
measurement of small cracks, it was necessary to remove the machining marks by carefully
polishing the appropnate surfaces. The final polishing was conducted using fine emery
paper of grade 800-1200, resulting in a final surface finish Ra of 0.2-0.25 Pm.

6.2 Hole Expansion Method

6.2.1 Selection of expansion parameters

The main parameter which characterizes the mandrelizing of art open hole is the percentage
of increase of the hole diarneter due to its expansion. For commercial applications, the
hole expansion is typically between 2% and 6%, depending on the matenal and the desired
residual stress field (Table 2.1). These values of hole expansion, however, are only valid for
a single hole. Limited work for the case of adjacent fastener holes and single holes close
to an edge suggest that one should reduce the values of hole expansion as the two holes
approach each other (Cloud [63]). In order to avoid the excessive plastic deformation and
damage, the expansion level was limited to 2%. This limit was further motivated by Our
desire to examine the mixed mode behaviour of fatigue cracks emanating from adjacent
holes. It is interesting to note that the general characteristics of the distribution of the
residual stress field are the sarne for different expansion levels. Only the magnitudes wili
Vary with the different expansion levels (see chapter 5).

6.2.2 Hole alignment and mandrelizing

The mandrelizing apparatus used in this study is shown in Figure 6.2. It consists of a
guiding and base plates made to very tight tolerances from hardened steel. These plates
can be tightened together with four screws, positioned at the corners of the plates. An array
of hole-pairs, corresponding to the examined specimen geometries, were accurately drilled
through both plates. The diarneter of al1 the holes was 5.05 mm (k10pm). The procedure
followed for the expansion of the specimens' holes is outlined below.
 The screws of the apparatus were loosened and the specimen was placed between the
two plates,

a pair of aiignment pins (4.90 mm (f10 pm))were placed in the corresponding pair
of holes through the two plates and the specirnen and the four screws were tightened,

the two mandrels are placed in the desired holes and are pushed through the specimen
at a speed of 0.2 mmlsec using a servo-hydraulic load frarne, and

the mandrels are removed by pushing two pins through the holes.

Whilst the use of the alignment pins guaranteed the appropriate positioning of the
specimen with respect to the mandrel, it was still necessary to examine the effect of
misalignment, on the resulting residual stress field. It was estimated that the maximum
misalignment due to machining tolerances is approximately 20 Pm. Figure 6.3 depicts the
effect of misalignment (6) on the hole expansion. To evaluate this effect on the resulting
residual stress field, an elasto-plastic finite element analysis was undertaken. The results
indicate that for a misalignment of 6=20 Pm,the difference in the maximum residual stress
field between aiigned and misaligned configurations is less than 8%.

Fatigue Fracture Testing

6.3.1 Fatigue testing equipment

Fatigue tests were conducted using the fatigue test set-up shown in Figure 6.4. The fatigue
test set-up consisted of a servo-hydraulic closed loop 250 kN load frame, a digital control
unit and a 486 micro-cornputer for test control and data collection.
The load frarne is equipped with a linear acniator mounted in a cross-head which c m
be adjusted to accommodate specimens of various lengths. A linear variable differential
transformer (LVDT) is used to monitor the position of the actuator piston. The generated
load is monitored by a load cell. In addition to load and displacement control modes, tests
under strain control are also possible through the use of an external extensometer.
 ' Base plate
Figure 6.2: Schematic of mandrelizing apparatus.

Mandrel

- Misalignment 6

' ' ~ o expansion

region
k

Figure 6.3: Mandrel-hole misalignment.

 Actuator

Hydraulic Grip Computer

Specimen

Load Ce11

Test Frame Digita1 Controlier

Figure 6.4: Uniaxiai fatigue testing equipment.

Hydraulic grips with adjustable gripping force to prevent specimen damage and slippage
were used. The aiignment of the grips and lozd train was checked pnor to testing according
to ASTM standard E- 1012-89. Four strain gages were bonded to the four sides of a
calibration specimen which was then subjected to a tensile load. The measured strains
indicated that the maximum bending strains were within 3% of the average axial strain at
load level of 20 kN,which met with the ASTM requirement of a maximum value of 5%.
The control unit which consisted of a control tower console and a front panel was used to
monitor and display the output from transducers measuring the load and strain experienced
by the test specimen, or the position of the actuator. Commands to setup the system, to
calibrate the transducers and to carry out the required test were entered either at the front
panel or through an interactive program in a 486 micro-computer environment.
6.3.2 Fatigue test program
Two test programs were conducted. In the first, attention was devoted to evaluating the
effect of constant amplitude fatigue loading upon the fatigue behaviour of adjacent holes in
an alurninom plate. Specifically, the work was devoted to determining the effect of the hole
size, inclination and applied load upon the fatigue crack growth of the different untreated
samples.
In the second, attention was devoted to exarnining the effect of coCd work and the
resulting residual stress field upon the fatigue fracture performance of the treated sarnples.
The entire fatigue test prograrn was conducted under load control with a constant
amplitude sinusoidai waveform. The test frequency was 10 Hz. Al1 tests were carried out
at room temperature at a stress ratio of R = = 0.1. A digital oscilloscope,
displaying the output waveform experienced by the specimen, was used to check the
performance of the machine. Deviation of the response signal from the input parameters
was eliminated by adjusting the proportional, integral and derivative gains ( P D values) of
the Ioop controller.

6.3.3 Crack size measurement

Crack initiation and growth were monitored using the acetate replication technique. This is
an accepted and recognized method for providing accurate monitoring of naturally initiated
short cracks. It also has the advantage of providing a permanent record of the crack path.
The technique, which is described in detail in [66],is a discontinuous process that requires
periodic interruption of the fatigue cycling process and may thus result in load relaxation.
The crack size and profile was obtained by examining cellulose acetate replicas of
previously polished surfaces (Ra=0.2-0.25pm). Both areas around the two holes and the
front ald back surfaces of the specimens were replicated, as depicted in Figure 6.5. The
replicas were taken at load equal to 90% of the maximum load to ensure that the crack is
open to its tip.
Replicas were taken frequently during the early fatigue stage so that initiation sites c m
be detected, and a sufficient number of small crack sizes (0.03-2 mm) were recorded. Crack
extensions as low as 30 pm were measured. A low magnification traveling microscope
(50X),mounted ont0 the ioad frarne, was also used to monitor the growth of long cracks
(>2 mm). The traveling microscope had a resolution of 25 pm.

replicas

*
Figure 6.5: Replication of the fatigue specimen.
Chapter 7

Analysis of Results and Discussion

In this chapter, we examine the effect of cold working upon fatigue crack initiation and
growth of Al-Zn 7075-T651 alloys. Specifically, it is desired to determine the effect of
residual stresses and work hardening upon fatigue crack initiation and safe life determina-
tion, as well as fatigue crack propagation and the darnage tolerance behaviour of the alloy.
Modelling of the mixed mode fatigue behaviour of the alloy in the presence of residual
stresses was conducted using the appropriate stress intensity factors and crack tracking
criterion. The findings of the work are summarized in this chapter.

7.1 Crack Initiation

7.1.1 Untreated sarnples

The mode1 developed in chapter 3 was employed to identify the location of potential

initiation site(s). The initiation prediction is based upon the maximum principal stress
criterion. Details of the geometricai notations used are provided in Figure 7.1.
Figures 7.2(a) and 7.2(b) and Table 7.1 show the marked difference in maximum
principal stress, and thus the initiation site as a function of the inclination of the centre line
of two adjacent holes. Figures 7.3(a) and 7.3(b) show typical experimentai results for the
cases where the inclination angles of the adjacent holes are 30" and 60°, respectively. The
figure reveals the close agreement between the analytical predictions and the experimental
 hole radius = R
separation distance = 2c
inclinarion angle = p
orientation angle = 9
extemal load = a,

Figure 7.1 : Geometrical notations for two adjacent holes.

(deg.) SCF 0 (deg.) SCF 8 Wg.1

O 3.38 O 3-11 O

Table 7.1 : Maximum stress concentration for the geornetries examined.

observations, as evidenced by the growing cracks. Table 7.1 shows that for ,O*, 30". 45'
and 90°, the crack initiation site is Iocated close to 0=0°, while for ,û=60° thecrack initiation
site is at 0=180° for 2c=3 R and 4R, respectively. This implies that the examined geometries
will experience mode 1 loading for the cases where p=ûo, 60" and 90°, and mixed mode
loading for /3=3U0 and 4S0, respectively.
In this study, we select a threshold crack size as the length which separates initiation
from propagation. Whilst the selection seems arbitrary, it is governed by the following:
(i) the non-linear behaviour associated with initiation and rnicroscopic crack growth, (ii) the
presence of multiple initiation sites and the average defect size in the material, and (iii) in-
accuracies associated with short crack measurements. Accordingly, a crack depth of 100
pm was selected. The 100 pm crack size also corresponds to the United States Air Force
darnage tolerance specifications Mil-A-83444 [67].In this approach, which neglects the ini-
Figure 7.2: Variation of normalized tangentid stress for different centre dis-
tances: (a) 2c=3R, and (b) 2 4R.
 Figure 7.3: Experimental results for two untreated samples: adjacent holes
inclined at (a) 30°, and (b) 60" for a separation distance 2c-3 R.

tiation life, a corner crack is assumed to pre-exist at holes and notches in aircraft structures
prior to service.
The initiation lives corresponding to the different untreated cases examined are provided
in Table 7.2. The applied load was selected such that the locaiized stress at the hole boundary
does not exceed 70% of the yield stress of the material. Accordingly, a maximum applied
stress ama,=87.5 MPa was selected. Table 7.2 shows clearly the drarnatic effect of the
inclination angle upon the initiation life. For example, for the case where P=30° and 2c=3R
the initiation life is 188,000cycles, while for j?=90° and 2c=3 R the initiation life is 295,000
cycles, i.e., an increase of 57%.

1 B 1 2c 1 Geometry 1 SCF 1 Initiation life (

Table 7.2: Initiation life in cycles for untreated samples: am.,=87.5 MPa.
7.13 Cold worked samples
Let us now focus Our attention on the cold worked samples. Prelirninary tests on the cold
worked samples indicated that for an applied maximum stress of 87.5 MPa, the initiation
life of the specimen exceeded 106 cycles. Accordingly, an increase in the applied load
was necessary. Tests were conducted using a maximum applied stress for each geometry
such that the localized stress at the hole boundary was equal to 90% of the yield stress
of the matenal (uy=506MPa). The applied maximum stress for each sample is shown in
Table 7.3.

Sample geometry cm.= (MPa) Untreated - Cold worked Increase (%)

,û=O", 2c=3 R 135 35,000 68,000 94
,8=30°, 2c=3 R 117 38,000 62,000 63
,û--4So,2c=3 R 131 40,000 52,000 30

Table 7.3: Initiation life for untreated and cold worked îarnples.

Sample geometry aapp ( m a ) or(MPa) aeff (MPa) Reduction (9%)

P=ûO,2c=3 R 455 -66 389 15

Table 7.4: Reduction in effective tangential stress due to hole expansion.

In order to examine the effect of the hole expansion upon the initiation site(s), the
residual stress field of the entry face of the workpiece resulting from cold expansion was
superimposed on the applied field. The resulting tangential stress field is provided in
Figures 7.4(a) and 7.4(b) for different inclination angles and remotely applied loads. The
figures also show that the position of the maximum tangential stress Is marginally affected
by the residual stress field around the hole boundary for the case where the separation
distance is 2c = 3R. For the case where 2&R, the location of the maximum tangential
stress is influenced by the residual stress for the case where the inclination angle is B=60°.
Table 7.4 shows the corresponding reduction in the effective tangentid stress at the hole
boundary due to the residual stress field for four of the considered geometries.
Consider now the effect of the resultant stress field upon the initiation life. Table 7.3
reveals the enhancernent in the initiation life as a result of the hole expansion. For exarnple,
for the examined expansion level of 2%, the initiation life increased by as much as 1568
in cornparison with the corresponding untreated sarnple.

7.2 Mechanics and Mechanisms of Fatigue Crack Initiation

The mechanisrns that govem fatigue crack initiation in the presence of cold working residual
stresses are surnrnarized in the block diagram shown in Figure 7.5. In general, crack
initiation in cold worked components is influenced by: (i) the induced residual stress, (ii) the
work hardening characteristics of the component, and (iii) the associated change in the
surface topography. The first two factors enhance the fatigue crack initiation resistance by:
(i) reducing the applied stress and stress ratio in the compressive Iayer, and (ii) increasing the
flow stress of the exposed surface layers due to work hardening. In high strength materials,
such as Ai-Zn 7075-T651, work hardening effects are insignificant since a,,,/cr, = 1.1.
Measurements of the surface roughness of the untreated and cold worked samples were
comparable ( R a = 0.25pm).It is therefore assumed that the increase in the initiation life
is mostly governed by the presence of the compressive layer in the different geornetries
exarnined.
The experimental results reveai that, for the two applied load levels, the fatigue process
began with a corner crack at the point of maximum stress concentration. Frequent replication
of the two holes permitted the detection of very small cracks (30 pm).
Let us now focus Our attention to the case presented in Figure 7.6 for a cold worked
sample with ,8=30°and îc=3 R subjected to an applied load of oma,=l15 MPa. In this case,
the stress concentration attains a maximum at two positions A and B. The figure shows two
cracks emanating from these stress concentrations sites. Subsequently, both cracks grew to
become dominant cracks.
Figure 7.4: Vanation of normalized effective tangentid stress for centre dis-
tances (a) 2c=3 R, and (b) 2 r - 4 R.
 N
CRACK INITIATION

w
'7i
RESDUAL STRESS SURFACE PROFiLE

r
Expansion level

WORK HARDENlNG
r-7
COMPRESSNE STRESS
COLD WORK
PARAMETERS
Materiais

Surface condition
prior to cold working 1

Figure 7.5: Mechanisms for crack initiation in a cold worked component.

Figure 7.6: Experimental results showing two dominant fatigue cracks in a

cold worked sarnple: P=30°, 2c=3 R.
7.3 Crack Propagation

7.3.1 Crack tracking cntenon

As described in chapter 2, the most reliable crack tracking criterion is the Ç-criterion, since
it is based on the fundamental principle of minimum potential energy and thus satisfies
the instantaneous equilibrium of a cracked geometry. The application of the Ç-criterion
requires the evaluation of the stress intensity factors KI and KII for each crack increment
and an expression for the elastic strain energy release rate Ç which accounts for the change
in crack growth direction under mixed mode loading. Following Nuismer 1471, the strain
energy release rate (plane stress) is given by

The direction of crack growth O, is then evaluated by maximizing Ç in equation (7.1)

and is given by

which coincides with the expression obtained using the maximum tangentid stress criterion.
To assess the influence of the combined effect of hf and A'[[, it is appropriate to use
an equivalent stress intensity factor IC,, for the development of the fatigue crack growth
curves. The simplest approach is to consider the total strain energy release rate as being:

Equating (7.1) and (7.3) for 8 = O,, the equivalent stress intensity factor h',, is given

1 8
hC, = 4COS* 2
2
[K:( l + COS +9 : 0, ~- 6h>Klr(1
~ sin2 + cos O,) sin O,]
(7.4)
 Equation (7.4) can now be used to evaluate the equivalent maximum and minimum
stress intensity factors, as descnbed in details in the next section.

7.3.2 Superposition of stress intensity factors

The approach adopted in this investigation to account for the effect of residual stresses on
the crack tip stress field is based on the superposition of the respective applied and residual
stress intensity factors giving an effective stress intensity factor, such that

where fi-,,, and &r-mat are the applied maximum mode 1 and mode II stress intensity
factors, respectively. These are obtained from the mode1 descnbed in chapter 4, using the
maximum applied stress .
,
a The residual mode I and mode II stress intensity factors
(Kie",Ki;') are obtained using the residual stress field evaluated from the finite element
analysis, descnbed in chapter 5 and the mode1 of chapter 4.
The effective maximum stress intensity factor K ~ ~ ! , ~can
, now be found by substituting
hfI = ~ e f j and f i r = cf f into equation (7.4). Similady, the effective minimum

mode 1 and mode II stress intensity factors can be expressed as

and the effective minimum stress intensity factor K:{!mi, as

The effective stress intensity range and the effective stress ratio can now be evaluated
from the following expressions:

eff -
AK:;' = Ke9-maz ef1
h yeq-min
ef /
R C / ~ = rGq-min
A""eq-maz

For ~ ' f =
j 0, the predicted crack growth rates are caiculated by substituting the
corresponding AK:if values in a fatigue crack growth relationship which does not account
for the R-ratio, such as Paris's relationship [68]; viz:

In the case where ~~f~ > 0,the crack growth rate is predicted by substituting ~~f in a
Forman's type equation which accounts for the influence of different stress ratios R [69]:

where K, is the fracture toughness of the material under consideration. The empirical
constants D,n,D Rand m* are obtained from curve fitting fatigue crack propagation data
( d a / d N - Ai<)which are generated for R = O, using specimens free from residual stresses
and under pure mode 1 loading.
Figure 7.7 summarizes the steps needed to evaluate the equivalent stress intensity Ni,,
and effective stress intensity range ~ K : iThe
f . figure shows that knowledge of the fatigue
crack growth rate data will enable the determination of Kr and Iirrfor a given crack
increment using (4.17). and Krl can now be used to determine the direction of crack
growth 8, given by expression (7.2). The critical angle of growth together with the
corresponding values of the stress intensity factors Kr and f i r are used to determine IiPe,
using expression (7.4). To evaluate the effective stress intensity range, the applied and
residual stress intensity factors are superimposed. These effective values are then used to
calculate the effective stress intensity range A K:!~. The entire process is repeated for each
incremental crack growth.
 Figure 7.7: Algorithm for prediction of mixed mode crack growth rate in the
presence of residual stresses.

Mechanics and Mechanisms of Fatigue Crack Growth

Fatigue crack growth behaviour of Al-Zn alloy

Figures 7.8(a) and 7.8(b) show the recorded surface crack length a versus the elapsed
number of cycles for the untreated and cold worked configurations, respectively. In both
untreated and cold worked samples, a minimum of two sarnples were tested to generate the
crack growth data. The results were obtained at a constant nominal stress amplitude which
results in a maximum stress oe=û.9a,at the hole boundary (Table 7.3).
Figure 7.8(a) shows the variation of a versus N for 2 ~ R and
3 ,B=Oo, while Figure 7.8(b)
 O Untreated
@Cold worked

Number of cycies

initiation of
second crack

O a

'.
shieMing
begins shielding 0
"O begins

O
i O Untreated
Cold worked

Number of cycles

Figure 7.8: Experimentally determined crack length venus number of cycles

for samples with Zc=3 R and (a) ,û=Oo,and (b) /3=30°.
shows the crack growth curve for the sarne separation distance, but for an inclination angle
0=30°. For the former, the crack is subjected to pure mode 1loading in both untreated and
cold worked conditions. In this case, the presence of the second hole increases the stress
intensity factor at the dominant crack. The effect of cold working residual stresses delays
the fatigue crack growth considerably. For example, for respective crack lengths of 0.5 mm
and 1.O mm, an increase in the propagation life of 1 10%and 60% was observed. This delay
in the fatigue crack growth was aiso observed for the case where ,B=30°. In this case, an
increase in the propagation life of 60% and 120% was observed for crack lengths 1 mm and
2 mm, respectively.
Whilst Figure 7.8(a) shows typical characteristicsof mode 1loading for ail crack length,
Figure 7.8(b) shows that the fatigue crack growth relationship changes at a crack length
of 1.5 mm for both the untreated and cold worked conditions. For the untreated case, the
"inflection" point corresponds to a shielding effect resulting from the interaction between
the adjacent holes. For the cold worked condition, a second crack was initiated at 96,000
cycles and the crack growth rate was reduced dramatically, as depicted in Figure 7.8(b).
The reason behind the change in the slope of the crack growth curves is best explained
using Figure 7.9. In Figure 7.9(a), we provide the details of the different geometries
exarnined, while in Figure 7.9(b) we show the variation of the stress intensity factor against
the crack length for a unit applied load (ao=l MPa). For the case where ,B=30°, the initial
increase in the stress intensity factor is dominated by the stress field of a single hole. As the
crack length increases, the influence of the second hole develops (AB). This influence is
more pronounced in the range a l R d . 4 to 0.8, reaching a maximum value when a/R d . 4 .
As the crack grows further, the stress field at its tip experiences shielding from the second
hole. This is manifested by the reduction in the stress intensity factor in region (BC). This
effectively implies that for crack lengths a d . 2 5 mm and 3 mm, the stress intensity factors
are the sarne.
This behaviour c m aiso be observed for the case where P=4S0, where mixed mode
loading dominates. For the cases where B=ûO and 90°,mode 1loading dominates. However,
because of the nature of the interaction between the holes, the stress intensity factor values
for P=ûOare much greater than for P=90°.
 -Single hole
p=oo
...-. .-.....
8=30°
- - - - - ~=45*

Figure 7.9: Calculated equivalent stress intensity factors: (a) details of crack
geometries examined, and (b) variation of stress intensity factor
versus crack length for a unit applied load using andytical model.
 Whilst the initiation in both the untreated and cold worked sarnples began with a corner
crack at the root of one of the holes, the growth for the untreated sample was through the
thickness. In contrast, the growth in the cold worked sample was predominantly in the
surface as a result of their arrest by the compressive residual stress field present in the
thickness direction. The crack profile was penny-shaped. as depicted in Figure 7.10.

crack final crack

initiation profile

Figure 7.10: A typical penny-shaped crack in a cold worked sample with

,B=3Oo and 2c=3 R.

It was also observed that for the cold worked samples, al1 cracks initiated at the mandrel's
entry side of the specimen. This side is associated with a lower residuai stress than the exit
side, as depicted in Figure 5.16 of chapter 5.

7.4.2 Incremental crack tracking

After the crack initiation site had been determined, the next objective was to predict the
direction of the propagating cracks using an appropriate incremental crack tracking criterion.
The analytical mode1 described in chapter 4 was used to predict the incremental crack growth
direction for different configurations. Attention was, however, devoted to the cases where
mixed mode crack growth took place, Le., 2c=3 R and P=30° and 45".
 Figure 7.1 1 shows the segrnented geometry of the two holes for B=3Oo and 2c=3R as
well as the two propagating cracks for the untreated and cold worked cases, respectively.
Each increment was taken to be 0.5 mm and a cornputer program, specially designed to
obtain the direction O,, was used. Details of this program are provided in Appendix A.
In order to examine the applicability of the analytical solution for a finite width plate, the
analytical predictions were venfied using the finite element method. As already indicated,
cracks are most likely to initiate at the hole boundary, where the maximum tangential
stress is experienced. Accordingly, cracks were initiated at the appropriate initiation site,
as depicted in Figure 7.12(a). These cracks were modelled by creating two appropriate
surfaces with unrestrained nodes. To mode1 the singularity at the crack tip, special crack tip
elements were used. Initially, an incremental crack of length 0.5 mm was allowed to grow
in the appropriate initiation site. The finite element solution for each increment allowed the
determination of the values of K I and f i I . The strain energy release rate criterion was then
used to evaluate the crack growth direction and a new increment was allowed to grow in this
direction. The predicted crack growth is shown in Figures 7.12(b)-(d). Excellent agreement
between the analytical and the finite element predictions is observed; see Figures 7.1 1 and
7.12.
These predictions were compared with experimental results of both untreated and cold
worked sampies, depicted in Figure 7.13. Figures 7.13(a) and 7.13(b) show the rernarkable
agreement with the predictions.
The segmented geometry of the two holes for /3=4S0 and 2c=3R, as well as the propagat-
h g crack for the untreated and cold worked cases are shown in Figure 7.14. No noticeable
difference was observed in the prediction of the crack direction in the untreated and cold
worked conditions. The average inclination of the fatigue crack growth predicted was 22".
The experimental results of fatigue crack growth shown in Figure 7.15 have an average
inclination of the fatigue cracks 23" and 2 5 O for the untreated and cold worked samples,
respectively.
It is worth noting that the direction of crack growth in both the untreated and cold
worked sarnples is similar. It must be remembered, however, that for the untreated case
the crack grows through the thickness, whilst in the cold worked case it is restricted to
Figure 7.1 1: Analytical prediction of incremental fatigue cracks for untreated
and cold worked geometry for P=30° and 2c=3 R.

Figure 7.12: Finite element predictions of fatigue crack growth at different

lengths: (a) a=l mm, (b) a=2 mm, (c) a=3 mm, and (d) a=4 mm
for /3=30°and 2c=3 R.
 Figure 7.13: Experimental results of fatigue crack growth for (a) untreated,
and (b) cold worked samples with P=3Oo and 2c=3 R.

a depth of 2.8 mm for P=30°. For cases where crack growth takes place under mode 1
loading (P=ûOand 90°), the crack grows through the thickness in both the untreated and
cold worked cases. The effect of the compressive layer slows down the growth of the crack
through the thickness in the cold worked sarnples.

7.4.3 Fatigue crack growth rate

Fatigue tests were conducted on the considered samples according to the conditions de-
scribed in chapter 6. The crack length and direction were measured by using the replica
technique and the corresponding number of cycles was recorded. Figure 7.16 shows the
crack profile for a cold worked sarnple with P=30° and 2c=3R.
As discussed earlier in this chapter, the crack growth rate in a residual stress field is
governed by the effective stress intensity range AK$f and the effective stress ratio FteII.
First consider the variation of fatigue crack growth rate ( d a / d N ) against Ah'.,. Figure 7.17
shows the effect of cold working upon the fatigue crack growth rate. The figure also shows
the fatigue crack growth rate for the same material under pure mode 1 loading [70]. The
results reveal a remarkable reduction in the fatigue crack growth rate as a result of cold
working. This figure also shows that the fatigue crack growth behaviour of the untreated
samples is comparable to mode I loading.
 O Unneatcd and
coid workcd

Figure 7.14: Analytical prediction of incremental fatigue cracks for untreated

and cold worked geometry for ,8=4S0 and 2 ~ = R.3

It is now important to modify A K,, to account for residual stresses. This is carried out
according to the procedure descnbed earlier in this chapter and is sumrnarized in Figure 7.7.
Figure 7.18 shows the newly predicted values for the untreated and cold worked sarnples
using the effective stress intensity range aK;if. These prediciions indicate that mode 1
loading represents an upper bound limit for fatigue crack growth rate for the different
geornetries examined both in the untreated and cold worked conditions. This effectively
means that one is now able to use the current theory to predict both fatigue crack growth
direction and rate to ensure darnage tolerant designs.
Whilst the crack initiated at the corner, it propagated through the thickness for al1 tested
samples with the exception of the cold worked samples with @=30° and 2c=3 R. It was
therefore felt unnecessary to complicate the analysis to account for the three-dimensional
nature of that single crack geometry. However, in order to obtain a conservative value for
fatigue crack growth data most of Our analysis was conducted using a lower bound of the
residual stress field, i.e. at the eniry face of the cold worked samples.
Figure 7.15: Experimental results of fatigue crack growth for (a) untreated,
and (b) cold worked sarnples with P=4S0 and 2c=3 R.

Figure 7.16: Crack path for a cold worked sample with 8=30°and îc=3 R.
 OUntreated
&old worked
Mode I PO]

6 10
MW(MPa mlR)

Figure 7.17: Fatigue crack growth rate versus equivdent stress intensity range.

t
0 Untreated
&old worked
Mode 1 [70]

6 10
AlCe: (MPa mlR)

Figure 7.18: Fatigue crack growth rate versus effective stress intensity range.
Chapter 8

Conclusions and Future Work

8.1 Statement of the Problem

Fatigue is undoubtedly the most prevaient mode of failure in aircraft structures. The
consequences of failure of a primary load-carrying component in an airfrarne or an engine
are usually catastrophic, often resulting in loss of life and hardware. It is therefore the duty
of designers to develop high performance fatigue resistant structures.
Residual stresses play an important role in determining the fatigue life of engineering
components. For example, many welded structures experience high tensile residual stresses
at the heat affected zone as a result of large thermal gradients and possible phase change
of the material. These high tensile residual stresses promote fatigue crack initiation and
growth. In contrast, compressive residual stresses, such as those induced by shot-peening
and cold hole expansion, are highly effective in improving the fatigue strength of aerospace
components. A major impedirnent to the wider use of these cold working processes stems
from Our inability to characterize and optimize the pertinent parameters which govem iheir
performance.
8.2 Objectives
It is therefore the objective of this study to examine the fatigue fracture behaviour of a
plate containing two circular holes in close proximity in the presence of residual stresses.
Specificdly, it is desired to:

(i) develop an analyticd model to predict the stress field and mixed mode stress intensity
factors for a generalized crack in a solid containing a number of interacting holes,

(ii) conduct three-dimensionai finite element anaiysis to obtain the residual stress field
due to cold hole expansion of adjacent holes,

(iii) obtain fatigue crack growth data for Ai-Zn 7065-T651 alloy in the presence of residual
stresses, and

(iv) develop a model to predict the mixed mode fatigue crack growth rate of the current
airfrarne alloy in the presence of residual stresses.

8.3 General Conclusions

8.3.1 Description of stress field of adjacent holes

A new analytical model capable of predicting the stress state in a solid containing a num-
ber of interacting holes was developed. The theoretical formulation was based upon the
superposition of single dislocation solutions, which reduce the original interaction problem
into the solution of singular integral equations in terms of dislocation density functions.
The resulting integrai equations were then solved using Chebyshev polynomial expansions.
The significance of this newly proposed method was manifested by its versatility and appli-
cability to complex hole configurations, thus enabling the detenination of potential crack
initiation site(s). The application of this model to the case of two adjacent circular holes
arbitrarily onented in a plate subjected to uniaxiai loading revealed the following:

(i) both the location and the magnitude of the stress concentration and, subsequently, the
fatigue life are dramatically influenced by the centre distance between the two holes
and the inclination of these holes with respect to the applied load, and
(ii) the maximum stress concentration occurs at an inclination angle P=3Oo.

8.3.2 The curved crack problem

A new model capable of determiningthe effective stress intensity factor of a crack emanating
from one hole in the presence of residual stresses was developed. In this case, the theoretical
mode1 described in chapter 3 was modified to account for the singularity at the crack tip.
The stress intensity factor solutions obtained were used in conjunction with the strain energy
release rate crack tracking criterion to determine incrementd crack growth under fatigue
loading conditions. The application of this model to the case of a curved crack emanating
from two adjacent circular holes revealed that the model is capable of accurately predicting
the direction of fatigue crack growth.

8.3.3 Finite element analysis of cold hole expansion

A comprehensive elasto-plastic three-dimensional finite element analysis was conducted to

evaluate the plastic zone developrnent and growth, and unloading residual stresses resulting
from cold expansion of two adjacent holes. The analysis accounted for the friction between
the mandrel and the hole surface using contact elements. The results revealed the following:
the compressive residual stress varies across the thickness of the specimen, with the
maximum being around mid-thickness and the minimum at the entry face,

two-dimensional finite element models are incapable of accurately predicting the

residual stresses resulting from the cold expansion process,

excessive cold expansion of adjacent holes cm lead to high tensile residual stresses,
and

a decrease in the centre distance between the expanded holes results in a decrease in
the compressive residual stresses at the entry face of the workpiece.
8.3.4 Fatigue crack growth in the presence of residual stresses
Fatigue tests of aluminum sarnples containing two circular holes were conducted using a
digitally controlled electro-hydraulic fatigue test apparatus. The two holes were expanded at
room temperature using a mandrelizing apparatus. High cycle fatigue tests were conducted
under load control at room temperature at a stress ratio R = 0.1. Crack initiation and growth
were monitored by examining cellulose acetate replicas of lightly polished surfaces and an
optical microscope.
The mixed mode fatigue crack growth of the resulting cracks was accounted for by using
an equivalent value for the stress intensity factor based on the strain energy release rate
criterion. Furtherrnore, the effect of residuai stresses was accounted for by the superposition
of the applied and residual stress intensity factors. The resulting equivalent and effective
stress intensity factors were then used in a reference mode 1 crack growth curve to predict
the propagation rate in the untreated and cold worked specimens. Based on these studies
the following conclusions are made:
the cold expansion of adjacent holes only rnarginaily affects the location of crack
initiation,

a cold expansion of 2% resulted in an increase in the initiation iife between 30% to

150% for the considered geometries,

crack initiation always takes place at the entry face of the cold worked sample,

cold work of adjacent holes does not affect the direction of the propagating cracks,

propagation life improvement of up to 120% was achieved by the cold expansion.

This is due to the slower crack growth resulting from the compressive residual stress
field,

mode 1 loading represents an upper bound limit for mixed mode fatigue crack growth
rate for the different geornetries examined both in the untreated and cold worked
conditions, and

the newly developed mode1 can reliably predict the direction and rate of mixed mode
fatigue crack growth in both the untreated and cold worked samples.
8.4 Thesis Contribution
In this thesis, comprehensive analytical, numerical and experimental work has been con-
ducted to evaluate the fatigue fracture behaviour of cold worked and untreated arbitrarily
oriented adjacent holes in Bat plates. The contribution can be surnrned up as follows:
(i) a new closed form solution has k e n developed to evaluate the elastic stress field in
two-dimensional domains containing adjacent holes,

(ii) the analytical work was further extended to describe the behaviour of a crack ema-
nating from one of the holes under generalized loading conditions. This model can
be used quite readily by designers,

(iii) extensive three-dimensional elasio-plastic non-linear finite element analysis was con-
ducted to evaluate the plastic zone development and growth, and unloading residual
stresses resulting from the cold expansion of two adjacent holes. Three remarkable
features were observed: (a) unlike existing two-dimensional and axisymrnetric work,
the current three-dimensional mode1 reveals the significant differences between the
entry and exit faces of the cold worked plate, (b) the high tensile residual stresses
present at the centre line of the plate, and (c) the effect of the level of expansion upon
the resulting residual stress field. These are important findings ready for use by the
aerospace industry in critical aerospace applications, and

(iv) a reliable model which is capable of predicting the mixed mode fatigue fracture
behaviour in the presence of cold work residuai stresses of two adjacent holes was
developed.

8.5 Recommendations for Future Work

During the course of these studies, the following items were considered worthy of future
work:
(i) extension of the work to account for multiple interacting holes,

(ii) study the effect of strain rate upon the resulting residuai stress field and depth of the
compressive layer,
(iii) examine the relaxation of residual stresses with cyclic loading and removal of material
frorn subsequent reaming operations,

(iv) evaluate the effect of interfacial conditions (friction, surface roughness, mismatch in
properties) upon the performance of the cold work process, and

(v) extend the two-dimensional crack mode1 to account for generai three-dimensional
curved cracks.
References

T.D.Cooper and C.A. Kelto, Fatigue in Machines and Structures - Aircraft, in

Fatigue and Microstructure, American Society for Metds, pp. 29-56, Metals Park,
Ohio, USA, 1979.

F.K.Naumann, Failure Analysis: Case Histories und Methodology, American Society

for Metals, Metals Park, Ohio, USA, 1983.

C.R. Brooks and A. Choudhury, Metallurgical Failure Analysis, McGraw-Hill, New

York, 1993.

G.Kirsch, Die Theorie der Elastizitat und die Bedurfnisse der Festigkeitslehre, Verlag
des Vereins Deutscher Ingenieure, 42. pp. 797-807, 1898.

G.B.Jeffery, Plane Stress and Plane Straïn in Bipolar Co-ordinates, Philosophical

Transactions of the Royal Society of Lundon, A221, pp. 265-293, 1920.

R.D. Mindlin, Stress Distribution Around a Tunnel, Transactions of the American

Society of Civil Engineers, 105, pp. 11 17-1 153, 1940.

R.D. Mindlin, Stress Distribution Around a Hole Near the Edge of a Plate Under
Tension, Proceedings of Society of Erperirnental Stress Analysis, 5, pp. 56-68, 1948.

LW. Yu and G.P.Sendeckyj, Multiple Circular Inclusion Problems in Plane Elasto-

statics, ASME Journal of Applied Mechanics, 41, pp. 2 15-22 1. 1974.

H. Horii and S. Nemat-Nasser, Elastic Fields of Interacting Inhomogeneities, Inter-

national Journal of Solids and Structures, 21(7),pp. 73 1-745, 1985.

S.A. Meguid and C.L. Shen, On the Elastic Fields of Interacting Defense and Main
Hole Systems. International Journal ofMechanica1Sciences, 34( 1), pp. 17-29, 1992.
[ I l ] K.X. Hu, A. Chandra and Y. Huang, Multiple Void-Crack Interaction, International
Journal of Solids and Structures, 30( 1 1), pp. 1473- 1489, 1993.

J. Dundurs and M. Hetenyi, The Elastic Plane With a Circular Insert, Loaded by a
Radial Force, ASME Jounal of Applied Mechanics. 28, pp. 103-1 11, 196 1.

M. Hetenyi and J. Dundurs, The Elastic Plane With a Circular Insert, Loaded by a
Tangentially Directed Force, ASME Jounal of Applied Mechanics, 29, pp. 362-368,
1962.

R. Kienzler and D. Zhuping, On the Distribution of Hoop Stresses Around Circular

Holes in Elastic Sheets, ASME Journal of Applied Mechanics, 54, pp. 1 10-114, 1987.

J.A. Greenwood, Exact Formulae for Stresses Around Circular Holes and Inclusions,
International Journal of Mechanical Sciences, 31(3), pp. 2 19-227, 1989.

P.E.Erickson and W.F. Riley, Minimizing Stress Concentrations Around Circular

Holes in Uniaxially Loaded Plates, Experimental Mechunics, 18, pp. 97-10, 1978.

R.B.Heywood, Designing by Photoelasticity, Chapman & Hall, Ltd., London, 1952.

U.C.Jindal, Reduction of Stress Concentration Around a Hole in a Uniaxially Loaded
Plate, Journal of Strain Analysis, 18(2), pp. 135- 141, 1983.

S.A. Meguid, Finite Element Analysis of Defense Hole Systems for the Reduction
of Stress Concentration in a Uniaxially-loaded Plate with Two Coaxial Holes, Engi-
neering Fracture Mechanics, 25(4), pp. 403-4 13, 1986.

S.A. Meguid and S .X. Gong, Stress Concentration Around Interacting Circular Holes:
A Cornparison Between Theory and Experiments, Engineering Fracture Mechanics,
44(2), pp. 247-256, 1993.

R.L. Champoux, Overview of Hole Cold Expansion Methods, Proceedings of the

International Conrefence on Fatigue Prevention and Design, pp. 35-6 1, Amsterdam,
The Netherlands, 1986.
M. Bernard, T.Bui-Qum, D. Julien and S. Forgues, Feasibility Study of Cold Expan-
sion Processes Modelling, Report 063SS.WS477- 1-AC89, Department of National
Defence, 1993.
[23] A. Nadai, Theory of Flow and Fracture of Solids, McGraw-Hill, New York, 1950.

[24] YC.Hsu and R.G.Forman, Elastic-plastic Analysis of an Infinite Sheet Having a

Circular Hole Under Pressure, Journal of Applied Mechanics, 42, pp. 347-352, 1975.

[25] D.L. Rich and L.F. Impelliueri, Fatigue Analysis of Coldworked and Interference
Fit Fasteners Holes, ASTM STP 637, pp. 153-175, 1977.

[26] A.A. Jongebreur and A.U. de Koning, Results of a Study of Residual Stresses and
Fatigue Crack Growth in Lugs with Expanded Holes, National Research Laboratories,
NRL, Amsterdam, The Netherlands, Report No. NRL-MP-83024-U, 1983.

[27] G. Wanlin, Elastic-plastic Analysis of a Finite Sheet with a Cold-worked Hole,

Engineering Fracture Mechanics, 45, pp. 857-864, 1993.

[28] R.E. Link and R.J.Sanford, Residual Strains SurroundingSplit-Sleeve Cold Expanded
Holes in 7075-T65 1 Aluminum, Journal of Aircrufi, 27, pp. 599-604, 1990.

[29] R. Cook and P. Holdway, Residual Stress Induced by Hole Cold Expansion, In
Computer Methods and Experimental Measurements for Surjke Treamient Efects,
pp. 9 1- 100, Computationai Mechanics Publications, Southampton, U.K., 1993.

[30] C. Poussard, M.J. Pavier and D.J. Smith, Andytical and Finite Element Predictions
of Residual Stresses in Cold Worked Fastener Holes, Journal of Strain Analysis, 30,
pp. 29 1-304, 1995.

[3 11 S.A. Forgues, M. Bernard and T.Bui-Quoc, 3-D Axisymmetric Numerical Analysis

and Experimental Study of the Fastener Hole Coldworking Process, In Computer
Methods and Experimenral Measurements for Surjiace Treatrnent Effects, pp. 6 1-70.
Computational Mechanics Publications, Southampton, U.K., 1993.

[32] M. Bernard, T. Bui-Quoc and M. Burlat, Effect of Re-coldworking on Fatigue Life

Enhancement of a Fastener Hole, Engineering Fracture Mechanics, 18. pp. 765-775,
1995.

[33] N. Chandawanich and W.N. Sharpe, An Expenmental Study of Fatigue Crack Ini-
tiation and Growth from Coldworked Holes, Engineering Fracture Mechanics, 11,
pp. 609-620. 1979.
O.L. Bowie. Analysis of an Infinite Plate Containing Radial Cracks Originating at
the Boundary of an Internai Circular Hole, Journal of Mathematics and Physics, 35,
pp. 60-71, 1956.

D.J. Cartwright and G.A. Ratcliffe, Strain Energy Release Rates for Radial Cracks
Emanating from a Pin Loaded Hole, International Journal of Fracture Mechanics,
8(2), pp. 175- 181, 1972.

A.F. Grandt Jr. and J.P. Gailagher, Proposed Fracture Mechanics Criteria to Select
Mechanicd Fasteners for Long Service Lives, Fracture Toughness and Slow-Stable
Cracking, ASTM STP 559, pp. 283-297, 1974.

A.F. Grandt Jr., Stress Intensity Factors for Some Through-Cracked Fastener Holes,
International Journal of Fracture, 11(2),pp. 283-294, 1975.

R.C.Shah, Stress Intensity Factors for Through and Part-Through Cracks Originating
at Fastener Holes, Mechanics of Crack Growth, ASTM STP 590, pp. 429-459, 1976.

D.P. Rooke and S.M. Hutchins. Stress Intensity Factors for Cracks at Loaded Holes
- Effect of Load Distribution, Joumal of Strain Analysis, 19(2),pp. 8 1-96, 1984.
D.P.Rooke, Stress Intensity Factors for Cracks at a Double Row of Holes, Journal of
Strain Analysis, 19(4), pp. 23 1-236, 1984.

A.F. Grandt Jr. and T.E. Kullgren, Tabulated Stress Intensity Factor Solutions for
Flawed Fastener Holes, Engineering Fracture Mechanics, 18(2), pp. 435-45 1, 1983.

V. Klirnan, M. Bily and J. Prohacka, Improvement of Fatigue Performance by Cold

Hole Expansion. Part 1: Model of Fatigue Limit Improvement. International Joumal
of Fatigue ,2, pp. 93- 100, 1993.

V. Klirnan, M. Bily and J. Prohacka, Improvement of Fatigue Performance by Cold

Hole Expansion. Part 2: Experimental Verification of Proposed Model, International
Journal of Fatigue ,2, pp. 101-107, 1993.

M. Isida and S. Nemat-Nasser, A Unified Analysis of VWious Problems Relating to

Circular Holes with Edge Cracks, Engineering Fracture Mechanics, 27(5),pp. 57 1-
591, 1987.
F. Erdogan and G.C. Sih, On the Crack Extension in Plates under Plane Loading and
Transverse Shear, Journal of Basic Engineering, 85, pp. 5 19-527, 1963.

M.A. Hussain, S.L. hi and J. Underwood, Strain Energy Release Rate for a Crack
under Combined Mode 1and II, ASTM STP 560, pp. 2-28, 1974.

R.I. Nuismer, An Energy Release Rate Criterion for Mixed Mode Fracture, Interna-
tional J o u m l of Fracture, 11(2), pp. 245-250, 1975.

K. Palaniswamy and W.G. Knauss, On the Problem of Crack Extension in Bnttle

Solids under General Loading, Mechanics Toduy, ed. S. Nemat-Nasser, Pergarnon
Press, 4, pp. 87- 148, 1978.

C.H.Wu, Fracture under Combined Loads by Maximum-Energy-Release Rate Cn-

terion, Journal of Applied Mechanics, 45, pp. 553-558,1978.

B. Cotterell and J.R.Rice, Slightly Curved or Kinked Cracks, International Journal

of Fracture, 16, pp. 155- 169, 1980.

K.Hayashi and S. Nemat-Nasser, Energy-Release Rate and Crack Kinking under

Combined Loading, Journal of Applied Mechanics, 48, pp. 520-524, 1981.

G.C.Sih, Some Basic Problems in Fracture Mechanics and New Concepts, Engineer-
ing Fracture Mechanics, 5, pp. 365-377, 1973.

G.C.Sih, Mechanics of Fracture Initiation and Propagation, Kluwer Academic

Publishers, 1991.

G.C. Sih and B.M. Barthelemy, Mixed-Mode Fatigue Crack Growih Predictions,
Engineering Fracture Mechanics, 13, pp. 439-45 1, 1980.

G.C. Sih and E.T. Moyer, Ir., Path Dependent Nature of Fatigue Crack Growth,
Engineering Fracture Mechanics, 17, pp. 269-280, 1983.

1. Kovacs and L. Zcoldos, Dislocations and Plastic Deformation, Pergamon Press,

New York, USA, 1973.

F. Erdogan, G.D. Gupta andf T.S. Cook, Numerical Solution of Singular Integral
Equations, in Methods of Analysis and Solutions of Crack Problems (edited by G.C.
Sih), pp. 368-425, Noordholf, Leyden, The Netherlands, 1973.
G. Szego, Orthogonal Polynomials, American Mathematical Society, Providence,
USA, 1975.

T.J. Rivlin, Chebyshev Polynomials, John Wiley & Sons, New York, USA, 1990.

K.K. Lo,Analysis of Branched Cracks, Journal of Applied Mechanics, 45, pp. 797-
802, 1978.

J. Eftis, Influence of Load Biaxiality on the Fracture Characteristics of Two Collinear

Cracks, International Journal of Fracture, 24, pp. 59-80, 1984.

Machinery's Handbook (editet by R.E. Green), 24th edition. Industrial Press Inc.,
New York, 1992.

G.Cloud and S. Paleebut, Three-dimensional Nature of Strain Field Near Coldworked

Holes, Air Force Wright Aeronautical Laboratories, Report AFWAL-TR-80-4204,
1982.

M. Ferahi, On the Effect of Residual Stresses Upon the Fatigue Behavior of Advanced
Airframe Alloys, Ph.D. Thesis, University of Toronto, 1994.

P.S. Pao, M.A.Imam,L.A. Cooley and G.R. Yoder, Cornparison of Corrosion-fatigue

Cracking of AI-Li Alloy AA 2090-T8E41 and Alloy AA 7075T651 in Salt Water,
Corrosion, 45, pp. 530-535, 1989.

M.H. Swain, Monitoring Smali-crack Growth by the Replication Method, In Smnll

Crack Test Methods, ASTM-STP 1 149, pp. 34-56,1992.

Military Specifications, Airplane Damage Tolerance Requirements, US Air Force:

MIL-A-83444,1974.

P.C. Paris and F. Erdogan, A Critical Analysis of Crack Propagation Laws, Journal
of Basic Engineering, 85, pp. 528-534, 1963.

R.G.Forman, V.E. Keamey and R.M. Engle, Numerical Analysis of Crack Propa-
gation in Cyclic-loaded Structures, Journal of Basic Engineering, 89, pp. 459-464,
1967.
Damage Tolerant Design Handbook, Volume 4, Wright Laboratory, Air Force Mate-
rial Command, US Air Force, Wright Patterson, USA, 1994.
APPENDIX A

Listing of a Sample Program

 PROGRAM GENERALIZED CURVED CRACK
THIS IS THE PROGRAM USED TO CALCUL,ATE THE STRESS
INTENSITY FACTORS OF A CURVED CRACK UNDER PLANE DEFORMATION
THE INPUT OF THIS PROGRAM IS
NC NUMBER OF THE SEGMENTS OF THE CRACK
M ORDER OF THE CHEBYSHEV POLYNOMIALS
NE NUMBER OF POINTS USED IN NUMERICAL INTEGRATION
CL(1) THE HALF-LENGTH OF SEGMENT I
FI(1) THE ORIENTATION ANGLE OF SEGMENT I
CX, CY, TAU APPLIED LOAD
YYR (I), XYR (I) RESIDUAL NORMAL AND TANGENTIAL STRESS OF SEGMENT I

IMPLICIT DOUBLE PRECISION (A-H,0-Z)

COMMON /PARA/NC,NE,M
COMMON /CONS/A,PI
COMMON /ICRACK/NCRACK,KO
DIMENSION TK(3OOO,3OOO),CC(3000),BB(3000),BB1(3000,1),WA(3000)
DIMENSION FI(l8O) ,CL(180),YYR(180),XYR(l8O)

OPEN(1,file='out8)
PX=2 .OXDATAN(1DO)
A=l .OD0
WRITE(*,*) 'NUMBER OF ITERATION/N. OF CURVED-CRACKS: I OR 2/KO8
READ(*,*) NIT
R E A D ( * , * ) NCRACK
READ(*,*) KO
WRITE(*,*) 'NC= I , 'NE= ' , 'M= l

READ(*,*) NC,NE,M
WRITE(*,*) 'CX= ' , ICY= ' 'TAU=
I I

READ(*,*) CX,CY,TAU
NN=2*NC* (M+l)
WRITE(*, * ) 'FI(NC1, CL(NC)'
ITERATION FOR FI
DO 1011 IT=l,NIT
DO 1 I=l,NC
PcEAD(*,*) FI(I),CL(I)
write(*,*) fi(i),cl(i)
FI(I)=FI(I)+PI/l8O.
CONTINUE
RESIDUAL STRESSES
R E A D ( * , * ) NRESI
DO 2 1=1,180
YYR(I)=O.ODO
XYR(I)=O.ODO
CONTINUE
DO 3 I=I,NRESI
READ ( * , * ) NRES
READ(*,*)W ( N R E S ) ,XYR(NRES)
CONTINUE

CALL FTK(TK,FI,CL)
 CALL FBB(BB,FI,CX,CY,TAU,YYR,XYX)
C
Dû 5 I=l,NN
5 BBl(1,l)=BB(I)
CAU LEQTlC (TK,3000,3000tNN,BBl,l,O,WA,IER)
Dû 6 I=l,NN
6 CC(I)=BBl(I,l)
s1=0.
S2=0.
S3=0.
S4=0.
S5=0.
S6=0.
S7=0.
S8=0.
DO 10 I=l,M
Sl=Sl+(-1.ODO)**I*CC(I)
S2=S2+(-1.ODO)**I*CC(M+I)
S3=S3+CC( (KO-2)*S*K+I)
S4=S4+CC( (KO-2)*S*M+K+I)
SS=SS+ (-1.ODO)**I*CC(KO*2*M+I)
S6=S6+(-1.ODO)**I*CC(KO*2*M+M+I)
S7=S7+CC( (NC-1)*SfM+I)
S8=S8+CC( (NC-1)*2*M+M+I)

10 CONTINUE
C
Sl=Sl+CC(NC*2*M+1)
S2=S2+CC(NC*2*M+S)

SKRII=-A*PI*DSQRT(PI*CL(NC)) *S7
SKRI=-A*PI*DSQRT(PI*CL(NC))*S8
WRITE(*,*) 'R-R1,SKRI/DSQRT(P1),SKRII/DSQRT(PI)
WRITE(*,*) 'R-L',SKrlI/DSQRT(PI),SKrlII/DSQRT(PI)
WRITE ( , ) 'L-R',SKlrI/DSQRT(PI),SKlrII/DSQRT(PI)
WRITE ( +, * ) ' L-L ' ,SKLI/DSQRT(PI:),SKLII/DSQRT (PI)
sssl=SKlrI
sss2=SKlrII
IF (sss2.EQ. O ) GOTO 1111
RRRR=sssl/4.O/sss2
CRITANG=2/PI*180.0*ATAN(RRRR-
ABS(SSS~)/SSS~*SQRT(RRRR**~+O.~))
GOTO 1112
CRITANG=O.O
WRITE ( * , * ) ' C r i t i c a l Angle = ' ,CRITANG
CONTINUE
STOP
END
SüBROUTINE F'BB(BB,FI,CX,CY,TAU,WR,XYR)
IMPLICIT DOUBLE PRECISION (A-H,0-2)
COMMON /PARA/NC,NE,M
COMMON /CONS/A,PI
COMMON /ICRACK/NCRACK,KO
D m S I O N BB(3OOO),FI(l8O),YYR(I8O),XYR(l8O)
NN=2*NCf(M+1)
DO 10 I=I,NN
BB (1)=O.
II=O
DO 1 I=l,NC
MM=M
IF(I.GT.1) MM=M+l
IF(I,EQ.NC-and.ncrack.eq.2) MM=M
DO 2 J=l,MPlr
II=II+l
BB(2*II)=(CX+CY)/2--(CX-CY)/ 2 - * D C O S ( 2 . * F ( I )
BB (2*II-1)=- (CX-CY)/2.*DSIN(2- *FI (1)) +TAU*DCOS(2.*FI (1)) +XYR(I)
BB (2'11-1) =-Ba (2*II-1)
BB(2*II)=-BB(2*II)
CONTINUE
CONTINUE
RETURN
END

SUEROUTINE FTK (TK,FI,CL )

IMPLICIT DOUBLE PRECISION (A-H,O-Z)
COMMON /PARA/NC,NE,M
COMMON /CONS/A,PI
COMMON /ICRACK/NCRACX,KO
DIMENSION TK(3OOO,3OOO)
DIMENSION TK(2* (NC*(M-1)+1),2* (NC*(M-1)+1))
DIMENSION EMA(2,SOO),EMB(2,200),FI(i80),CL(l8O)
NN=2*NC*M+2' (NC-1)
Do 1 I=l,NN
DO 1 J=l,NN
TK(I,J]=O.ODO

IF(K.EQ.NC.and.ncrack.eq.2) MM=M
DO 3 Kl=l,MM
TK (II,JJ)
II=II+l
XKCT=FLOAT[Kl.)*PI/FLOAT(MM+1)
XK=CL (K)*DCOS (XKCT)
DO 5 J=l,NC
CALL EMETR(J,K,XK,FI,CL,EMA,EMB)
JJO=(J-1)*2*M+1
JJl=JJO+M
DO 6 JJJ=l,M
JJA=JJO+JJJ-1
JJB=JJl+JJJ-1
TK(S*II-1,JJA)=EMA(l,JJJ)
TK(2*11,JJA)=EMA(2,JJJ)
TK(2*II-1,JJB)=EMB(l,JJJ)
TK(2*II,JJB)=EMB(2,JJJ)
CONTIMJE
TK(2*1I-l,NC*2*M+2*J-l)=EMA(l,M+l)
TK(2*II,NC*2*M+2*J-l)=EMA(Z,M+l)
TK(2*II-18NC*2*M+2*J) =EMB(l,M+I)
TX(S*IItNC*2*M+2*J)
=EMB(S,M+l)
c o m m
CONTINUE
CONTINUE
if(ncrack.eq.1) goto 1001
DO 7 J=l,NC
TK(2*NCe (M+1)-3 ,2*NCfM+2*J-l)=CL(J)*DCOS(FI (J))
TK(2*NC* (M+l)-3,SfNC*M+2*J)=-CL(J) *DSIN(FI(J))
TK(S*NC* (M+l)-St2*NC*M+2*J-1)=CL(J)*DSIN(FI(J))
TK(2*NCf(M+1)-2,2*NC*M+2*J)=CL(J) *DCOS(FI(J))
kkO=kO
if(ncrack-eq.1) kkO=nc
DO 17 J=l,KKO
TK(2*NCf (M+~)-I,S*NC*M+~*J-~)=CL(J)*DCOS(FI(J) )
TK(2*NC* (M+lI-lt2*NC*M+2*J)=-CL(J)*DSIN(FI(3))
TK(2*NC* (M+I),2*NCfM+2*J-1)=CL (J)*DSIN(FI(J))
TK(2*NCe (M+1),2*NC*M+S*J)=CL (J)*DCOS (FI(J))

TK(IIl+I,JJl+I)=TK(II1+ItJ J ~ + I+1
) .OD15
CONTINUE
TK(II2+1,JJ2+1)=TK(II2+1,JJ2+1)+ l m 5
TK(112+2,JJ2+2)=TK (II2+2,~ ~ 2 ++l.0~15
2 )
continue
RETURN
END

SUBROUTINE EMETR (J,K tXK,FI,CL,EMA, EMB)

IMPLICIT DOUBLE PRECISION (A-H,0-2)
COMMON /PARA/NC,NE,M
COMMON /CONS/AO,PI
DIMENSION FI(l8O),CL(l8O)
DIMENSION EMA(2,200),EMB(2,200),TT(2,3),CGM(3),CGM1(2)
DIMENSION CT(200)
DIMENSION AAl(200),AA2 (200),AA3 f200),BB1(200),BB2(200),BB3 (200)
FKJ=FI (K)-FI (J)
CALL STRESST ( FKJ, TT)
DO 1 I=I,NE
CT (1)=FLOAT(I-1)*PI/FLOAT(NE-1)
DO 2 I=l,M+l
IF(K.NE.J) GOTO 100
IF(I.EQ.M+l) THEN
EMA(1,I)=O.ODO
EMA(S,I)=O.ODO
EMB(l,I~=O.ODO
EMB(S,I)=O.ODO
GOTO 2
ENDIF
CTXK=DACOS (XK/CL(K))
IF(DABS ( C m )-GT.l.D-3) GOTO 1001
Sl=AO*PI*FLOAT(1)
GOTO 1000
IF(DABS(CTXK-PI).GT.l.D-3) GOTO 1002
SI=AO*PI*(-1.0~0) * * (I+I)*FLOAT(I)
GOTO 1000

EMA(2,I)=O.ODO
EMB(l,I)=O.ODO
EMB(2,I)=Sl
GOTO 2
IF(K.LT,J) THEN
DXJK=O.ODO
DXlJK=O - ODO
DYJK=O .ODO
DYlJK=O.ODO
DO 5 III=K,J-1
DXJK=DXJK+2.ODO*CL(III)*DCOS(FI(III))
DXlJK=DXlJK+2.ODO*CL(III+l)*DCOS(FI(III+l))
DYJK=DYJK+2.ODO*CL(III)*DSIN(FI(III))
DYIJK=DYIJI(+S.ODO*CL(III+l)* D S I N (FI(III+l))
ENDIF
IF(K,GT.Jl THEN
DXJK=O.ODO
DX1SK=O.ODO
DYJK=O .ODO
DYlJK=O.ODO
DO 6 III=J,K-1
DXJK=DXJK-2.ODOfCL(III)*DCOS(FI(II1)1
DXlJK=DXlJK-2.ODO*CL(III+l)*DCOS(~1(111+1) )
DYJKzDYJK-S.ODO*CL (III)*DSIN(FI (III 1

m = o .SDO* (DXJK+DXlJK)
YJK=O .SDO* (DYJK+DYIJK)
CALL TRANPO (J,FI (JI,K,FI (KI,XJK,Y=,m, YJ,mtO -DO)
DO 3 II=I,NE
X=XJ-CL(J)*DCOS(CT(I1))
Y=YJ
MMM=I
IF(I.EQ.M+l) MMM=O
AA1 (II)=CX1[X,Y)*DCOS(FLOAT(MMM)*CT(II))
A U (II)=CYl(X,Y)*DCOS(FLOAT(MMM)*CT(II))
AA3 (II)=CXYl(X,Y) *DCOS(FLOAT(MMM)*CT(II))
BBl (II)=CX2(X,Y)*DCOS(FLOAT(MMM)*CT(II) )
BB2 (II)=CY2(X,Y)*DCOS(FLOAT(MMM)*CT(II))
B?33(II)=CXY2(X,Y) *DCoS(FLOAT(MMM)*CT(II))
CONTINUE
CALL INT(NE,CT,AAl,RESl)
CALL INT(NErCT,AA2,RES2)
CALL INT(NEnCT,AA3,RES3)
CGM(1)=RESI
CGM (2) =RES2
CGM ( 3 ) =RES3
CALL MATRCON(2,3,TT,CGM,CGMl)
EMA(l,I)=CGMl(l)*CL(J)
EMA(2,I)=CGM1(2)*CL(J)
CALL INT(NE,CT,SBl,RESI)
CALL INT(NEnCT,BB2,RES2)
CALL 1NT(NE,CTnBB3,RES3)
CGM(1)=RESI
CGM (2) =RES2
CGM (3) =RES3
CALL MATRCON(2,3,TT,CGMnCGM1)
EMB(l,I)=CGMl(l)*CL(J)
EMB(2,I)=CGM1(2)*CL(J)
CONTINUE
RETURN
END

SUBROUTINE INT(N,X,Y,RESULT)
IMPLICIT DOUBLE PRECISION (A-H,0-2)
DIMENSION X(200),Y(200)
RESULT=O .
DO 1 I=l,N-1
REmT=RESULT+O.S*(Y(I+1)+Y(I))*(X(I+l)-X(1))
CONTINUE
RETURN
END

SUBROUTINE STRESST (F,TT)

IMPLICIT DOUBLE PRECISION (A-H,0-2 )
DIMENSION TT (2,3)
TT(l,l)=-O.5DOfDSIN(2.0DO*F)
TT(1,2)=-TT(1,l)
TT(1,3)=DCOS(2.0DOfF)
TT ( 2 , 1 )=DSIN(F)*DSIN (F)
TT(2,2)=DCOS(F)*DCOS(F)
TT(2,3)=-DSIN(2.ODO*F)
RETURN
END

S U B R O U T m TRAND (FI,DX, DY,DXI ,DY1 )

IMPLICIT DOUBLE PRECISION (A-H,0-2)
DIMENSION A(2,2) ,B(2),C(2)
A(1, 1)=DCOS (FI)
A(l,S)=DSIN(FI)
A(2,I)=-DSIN(F1)
A(2,2)=DCOS (FI)
B(I)=DX
B(2)=DY
CALL MATRCON(2,2,A,B,C)
DXl=C ( 1 )
DYl=C (2)
RETURN
END

M, A, B,Cl
SUBROUTINE MATRCON (NI
IMPLICIT DOUBLE PRECISION (A-H,0-2 )
DIMENSION A (N,M) ,B (Ml ,C (N)
DO 1 I=l,N
S=O .
DO 2 J=l,M
S=S+A(1,J)*B (J)
C(I)=S
RETURN
END
SUBROUTINE TRANPO(J,FJ,K,FK,XJK,YJK,XJ,YJ,XK,YK)
IMPLICIT DOUBLE PRECISION (A-H,0 - 2 )
DIMENSION A(2,2), B ( 2 ) ,C(2),D(2)
FKJ=FK-FJ
A(1,I)=DCOS (FKJ)
A(I,S)=-DSIN(FKJ)
A(2,1)=DSIN(FKJ)
A(2,2)=üCOS(FKJ)

CALL MATRCON(2,2,A,B,C)
CALL MATRCON(2,2,AtB,D)
XJ=C(l)+D(l)
YJ=C(2)+D(2)
RETURN
END
FUNCTION CXl (X,Y)
IMPLICIT DOUBLE PRECISION (A-H,O-2)
COMMON /CONS/AG,PI
IF(DABS(Y) .LT.l.oD-3) THEN
CXl=O .ODO
GOTO 1
ENDIF
CXl=AG*Y* (3.ODO*X*X+Y*Y)/ (X*X+Y*Y)**2
RETURN
END
FUNCTION CYI (X,Y)
IMPLICIT DOUBLE PRECISION (A-H,O-2)
COMMON /CONS/AG,PI
IF(DABS(Y).LT.l.OD-3) THEN
CYl=O .ODO
GOTO 1
ENDIF
CYl=-AG*Y* (XfX-Y*Y)/ (X*X+Y*y)**2
RETURN
END
mJNCTION CXY1 (X,Y)
IMPLICIT DOUBLE PRECISION (A-H,0-2 )
COMMON /CONS/AG,PI
IF(DABS(Y).LT.l.OD-3) THEN
CXYl=-AG/X
GOTO 1
ENDIF
t+S
CXYI=-AG*X* (X+X-Y*Y)/ (X*X+Y*Y)
RETURN
END
J?üNCTION CXS (X,Y)
IMPLICIT DOUBLE PRECISION (A-H,O-2)
COMMON /CONS/AG,PI
IF(DABS(Y).LT.l.OD-3) THEN
CX2=-AG/X
GOTO 1
ENDIF
CX2=AGtX*(Y*Y-X*X)/ (XeX+Y*Y)
**2
R E m
END
FUNCTION CY2 ( X , Y)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
COMMON /CONS/AG,PI
IF (DABS(Y).LT. 1.OD-3 ) THEN
CY2=-AG/X
GOTO 1
ENDIF
CY2=-AG*X* (3.0DO*Y*Y+X*X)/ (X*X+YtY)
**2
RETURN
END
FUNCTION CXY2 (X,Y)
IMPLICIT DOUBLE PRFCISION (A-H,O-Z)
COMMON /CONS/AG,PI
IF (DAES(Y).LT. 1.0D-3 ) THEN
CXY2=O.ODO
GOTO 1
 ENDIF
CXY2=AG*Yf(Y*Y-X*X) / (X*XfYtY) **2
RETURN
END

THIS IS AN IMSL SCTBROUTINE FOR SOL-G LINEAR COMPLM EQUATIONS

ASSOCIATED WITH A UNSYMMETRIC MATRIX IN FULL STORAGE.
+ INPUT *
m----------MAXIMüM NUMBER OF ROWS OF THE LEFT HAND SIDE MATRIX
m~---------
MAXIMUM NUMBER OF COLUMNS OF THE LEFT HAND SIDE MATRIX
N----------- NUMBER OF COLUMNS OF THE LEFT HAND SIDE MATRIX
M----------- NUMBER OF COLUMNS OF THE RIGHT HAND SIDE MATRIX
IB----------NUMBER OF ROWS OF THE RIGHT HAND SIDE MATRIX
IJOB--------JOB OPTION
=O, SOLVE THE EQUATION
A(NM,NDM)---THE LEFT E3AND SIDE MATRIX
B (NM,M)-----THE RIGHT HAND SIDE MATRIX. EACH COL= AS A VECTOR
REPRESENTS A GIVEN CASE
WA(NM)------WORKING VECTOR
IER---------CONDITION INDEX
=O, NO= TERMINATION
=129, TEIIMINAL ERROR
*OUTPUT*
A(NMINDM)---THE LEFT HAMi SIDE MATRIX, DECOMPOSED-
B (NM,M)-----THE RIGHT HAND SIDE MATRIX. EACH COLUMN AS A VECTOR
IS THE RESULT FOR A GIVEN CASE.

SUBROUTINE LEQTlC (A,NM,NDM,N,B,M,IJOB,WA,IER)

IMPLfCIT DOUBLE PRECISION (A-H,0-2)
INTEGER N,fA,M,IB,IJOB,IER
DIMENSION A (NM,NDM) ,B (NM,M) ,WA (NM)
COMPLEX+16 A(NMINDM),B(NM,M)
DOUBLE PRECISION WA (NM)
DOUBLE PRECISION P IQ ,ZERO,ONE,T ( 2
COMPLEX*16 ÇUM, TEMP
INTEGER 1,J IJM1,IMl,K,IMAX,JP1,IW
EQUIVALENCE (SUM,T(l))
DATA ZERO/O.ODO/,ONE/l.DO/
IA=N
IB=N
IER=O
IF(IJOB.EQ.2) GO TO 7 5
RN=N
DO 10 I=1,N
BIG=ZERO
DO 5 J=I,N
TEMP=A(I,J)
P=DABS (TEMP)
IF (P.GT,BIG)BIG=P
CONTINUE
IF(BIG.EQ.ZER0) GO TO 105
WA (1) =ONE/BIG
CONTINUE
DO 70 J=l,N
JMl=J-1
IE (JMl.LT.1) GO TO 25
DO 20 I=l,JMl
S W = A (1,J)
IMl=I-1
IF(IMl.LT.1) GO TO 20
DO 1s K = 1 , IM1
SUM=SUM-A(1,K) * A ( K , J)
15 CONTINUE
 A(I,J)=SUM
c o m m
P=ZERO
DO 45 I=J,N
SUM=A(I,J)
IF(JMl.LT.1) GO TO 40
DO 35 K=l,JMl
SUM=SUM-A(1,K)* A ( K , J)
CONTINlTE
A(1, J)=SUM
Q=WA(I)*DABS(SUM)
IF(P.GE.Q) GO TO 45
P=Q
IMAX=I
CONTINUE
Q=RN+P
IF(Q.EQ.RN) GO TO 105
IF(J.EQ.IMAX) GO TO 60
DO 50 K=l,N
TEMP=A (ml K)
A(IMAX,K)=A(J,K)
A(J,K) =TEMP
CONTINUE
WAIIMAX)=WA(J)
WA(J)=IMAX
JPl=J+l
IF(JP1.GT.N) GO TO 70
TEMP=A( J IJ)
DO 65 I=JPl,N
A ( f ,J)=A(I,J)/ T m
CONTIrn
CONTINUE
IF(IJOB.EQ.1) GO TO 9005
DO 103 K=1,M
IW=O
DO 90 I=l,N
IMAX=WA( 1)
SUM=B( IMAX,K )
B(IMAX,K)=B(I,K)
IF(IW.EQ.0) GO TO 85
IMl=I-1
DO 80 J=IWIIM1
çUM=SUM-A(1, J)*B(J.K)
CONTINUE
GO TO 88
IF(T(1) .NE.ZERO.OR.T(2).NE.ZERO)IW=I
B(I,K)=SUM
CONTINUE
NI=N+l
DO 100 IW=l,N
I=Nl-IW
JPl=I+l
SUM=B( 1,K)
IF(JP1.GT.N) GO TO 98
DO 95 J=JPl,N
SUM=çUM-A(1,J)*B(J,K)
CONTINUE
B(I,K)=SUM/A(I,I)
CONTINUE
CONTINUE
GO TO 9005
IER=129
CONTINUE
C U L UERTST ( IER,6HLEQTlC)
RETURN
END
IMAGE NALUATION

APPLIED
.-. --- I M G E lnc
= ,
1653 East Main Street
--
-
--
--
--- I
Rochester. NY 14609 USA
Phone: 7161482-0300
F a : 716/288-5989

O 1993. mieci Image. Inc.. All Rights Reserved