SlideShare a Scribd company logo
1 of 206
Download to read offline
INITIATION, GROWTH, AND COALESCENCE OF
SMALL FATIGUE CRACKS AT NOTCHES
A Thesis
Submitted to the Faculty
of
Purdue University
by
Eric Nielsen Forsyth
In Partial Fulfillment of the
Requirements for the Degree
of
Master of Science in Aeronautics and Astronautics
May 1993
ii
Dedicated to my parents,
George and Ardith,
and my grandparents,
Arthur and LaVerne Nielsen,
for their endless love and support.
iii
ACKNOWLEDGMENTS
This work was sponsored by the Aluminum Company of America under Project
Number TC919597TC. Special thanks are extended to Dr. A. J. Hinkle and Dr. B. J.
Shaw of Alcoa for their supervision and assistance throughout this study. In addition, the
author would like to express appreciation to the School of Materials Science at Purdue
University for the use of their specimen preparation and optical microscopy facilities.
The author would especially like to thank his major Professor, A. F. Grandt, Jr.,
for his guidance throughout this work. Professor Grandt's experience and insight were
invaluable in shaping the author's perceptions and approach to research in addition to the
course of the research itself. Thanks are also extended to Professor B. M Hillberry and
Professor H. D. Espinosa for providing their unique perspectives as members of the
author's thesis committee.
There are many other people whose support and assistance were instrumental in
the completion of this work. Thanks are due to Mark Yost, Bob Sanders, and the late
Gene Harston for technical assistance ranging from specimen fabrication to testing
equipment maintenance. Special thanks is extended to Chad Zezula for his significant
assistance with specimen testing and replica measurement. Finally, thanks are due to
Mark Doerfler and Marcus Heinimann for their advice and encouragement, as well as
Michelle Wade for her support.
iv
TABLE OF CONTENTS
Page
LIST OF TABLES............................................................................................................. vi
LIST OF FIGURES ......................................................................................................... viii
ABSTRACT..................................................................................................................... xix
CHAPTER 1 - INTRODUCTION.......................................................................................1
CHAPTER 2 - BACKGROUND.........................................................................................2
2.1 LEFM Concepts................................................................................................2
2.2 The Small Crack Problem.................................................................................4
2.3 The 7050-T7451 Aluminum Alloy...................................................................9
CHAPTER 3 - EXPERIMENTAL PROCEDURES .........................................................20
3.1 Small Crack Specimen Design and Testing Procedures.................................20
3.1.1 Specimen Testing.............................................................................22
3.1.2 The Replication Method ..................................................................23
3.2 Large Crack Testing Procedures.....................................................................25
CHAPTER 4 - EXPERIMENTAL RESULTS..................................................................35
4.1 Large Crack Growth Rate Data ......................................................................35
4.2 Small Crack Test Results................................................................................36
4.3 Small Crack Growth Rate Data.......................................................................39
CHAPTER 5 - ANALYTICAL MODELING...................................................................69
5.1 Background.....................................................................................................69
5.2 Description of Algorithm................................................................................71
v
Page
CHAPTER 6 - NUMERICAL RESULTS.........................................................................76
6.1 Back-Prediction in Specimens Used to Calculate Small Crack
da/dN-ΔK curve ...............................................................................................76
6.2 Prediction Results in Specimens Initiating Multiple Cracks ...........................79
CHAPTER 7 - CONCLUSIONS AND RECOMMENDATIONS..................................115
LIST OF REFERENCES.................................................................................................117
APPENDICES
Appendix A - Stress Intensity Factor Solutions...................................................122
Appendix B - Specimen Dimensions and Test Parameters .................................128
Appendix C - Crack Measurements for Double-Edge Notch Specimens............130
vi
LIST OF TABLES
Table Page
3.1 Parameters for fatigue test specimens........................................................27
4.1 Test matrix for the double-edge notch specimens. ....................................43
Appendix
Table
B1 Dimensions and test parameters for the double-edge notch
specimens. All tests were conducted at a stress ratio R =
0.1 and in laboratory air...........................................................................129
C1 Crack measurements for specimen 6611-a12, back notch.......................133
C2 Crack measurements for specimen 6611-a12, front notch.......................135
C3 Crack measurements for specimen 6612-b21, back notch.......................140
C4 Crack measurements for specimen 6714-a11, back notch.......................141
C5 Crack measurements for specimen 6714-a12, front notch.......................142
C6 Crack measurements for specimen 7012-a22, back notch.......................144
C7 Crack measurements for specimen 7012-a22, front notch.......................147
C8 Crack measurements for specimen 7111-b11, back notch.......................150
C9 Crack measurements for specimen 7111-b12, back notch.......................151
C10 Crack measurements for specimen 8B2, back notch ...............................152
vii
Appendix
Table Page
C11 Crack measurements for specimen 8B3, back notch ...............................153
C12 Crack measurements for specimen 8B3, front notch...............................154
C13 Crack measurements for specimen 8T3, front notch ...............................158
viii
LIST OF FIGURES
Figure Page
2.1 The two stages of fatigue...........................................................................14
2.2 The semicircular edge notch geometry and variable
definitions ..................................................................................................15
2.3 Typical fatigue crack growth rate data for large and small
cracks [11]..................................................................................................16
2.4 Stress intensity factor vs. time for constant ampltide
loading. Lower crack opening stress for small cracks
results in a larger effective stress intensity factor than a
large crack under identical loading, translating into faster
growth rates................................................................................................17
2.5 Cumulative fatigue failure distributions from 1984-1987
for the 7050-T7451 thick plate (5.5-5.9 inches = 140-150
mm)............................................................................................................18
2.6 Large crack da/dN-ΔK data for the Al 7050-T7451 alloy,
R=0.1 [29]..................................................................................................19
3.1 Double-edge notch dogbone specimen geometry and
dimensions. ................................................................................................28
3.2 Double-edge notch specimen geometry and dimensions...........................29
3.3 Hoop Stress/Remote Stress, σhoop/σrem, vs. Theta, Θ, for
the 1.11 and 2.00 inch wide specimen geometries.
σhoop/σrem was caculated with a 2-dimensional finite
element analysis of each of the specimen geometries. Θ
is defined as the angle (in radians) from the tip of the
notch to the upper/lower location where the notch meets
the specimen edge......................................................................................30
ix
Figure Page
3.4 Illustration of the replication process ........................................................31
3.5 Definition of replica coordinate system.....................................................32
3.6 CT specimens fabricated from fractured double-edge
notch specimen...........................................................................................33
3.7 CT specimen geometry and dimensions. ...................................................34
4.1 Large crack growth rate vs. stress intensity factor range
data for Al 7050-T7451 (obtained from CT specimens)
The large crack Paris Law was obtained from Reference
[29] (see Figure 2.5)...................................................................................44
4.2 Surface crack length vs. number of elapsed cycles for
specimen 6611-a12 (old material), "back" notch.......................................45
4.3 Surface crack length vs. number of elapsed cycles for
specimen 6611-a12 (old material), "front" notch. .....................................46
4.4 Surface crack length vs. number of elapsed cycles for
specimen 6612-b21 (old material), "back" notch. .....................................47
4.5 Surface crack length vs. number of elapsed cycles for
specimen 6714-a11 (old material), "back" notch.......................................48
4.6 Surface crack length vs. number of elapsed cycles for
specimen 6714-a12 (old material), "front" notch. .....................................49
4.7 Surface crack length vs. number of elapsed cycles for
specimen 7012-a22 (new material), "back" notch.....................................50
4.8 Surface crack length vs. number of elapsed cycles for
specimen 7012-a22 (new material), "front" notch.....................................51
4.9 Surface crack length vs. number of elapsed cycles for
specimen 7111-b11 (new material), "back" notch.....................................52
4.10 Surface crack length vs. number of elapsed cycles for
specimen 7111-b12 (new material), "back" notch.....................................53
x
Figure Page
4.11 Surface crack length vs. number of elapsed cycles for
specimen 8B2 (3-inch plate material), "back" notch. ................................54
4.12 Surface crack length vs. number of elapsed cycles for
specimen 8B3 (3-inch plate material), "back" notch. ................................55
4.13 Surface crack length vs. number of elapsed cycles for
specimen 8B3 (3-inch plate material), "front" notch.................................56
4.14 Surface crack length vs. number of elapsed cycles for
specimen 8T3 (3-inch plate material), "front" notch. ................................57
4.15 Replica photograph of specimen 6612-b21, "back" notch,
0 cycles (after specimen alignment loading). Crack ID # 1:
2a = 0.0016 in. ...........................................................................................58
4.16 Replica photograph of specimen 6612-b21, "back" notch,
10001 cycles. Crack ID # 1: 2a = 0.0038 in. .............................................58
4.17 Replica photograph of specimen 6612-b21, "back" notch,
10001 cycles. Crack ID # 1: 2a = 0.0038 in. .............................................59
4.18 Replica photograph of specimen 6612-b21, "back" notch,
23001 cycles. Crack ID # 1: 2a = 0.0092 in. ............................................59
4.19 Replica photograph of specimen 6612-b21, "back" notch,
33506 cycles. Crack ID # 1: 2a = 0.0179 in. ............................................60
4.20 Replica photograph of specimen 6612-b21, "back" notch,
42509 cycles. Crack ID # 1: 2a = 0.0337 in. ............................................60
4.21 Replica photograph of specimen 6611-a12, "front" notch,
42507 cycles. Crack ID # 4.1: 2a = 0.0208 in. Crack ID #
4.2: 2a = 0.0025 in. Crack ID # 8: 2a = 0.0009 in. ...................................61
4.22 Replica photograph of specimen 6611-a12, "front" notch,
50007 cycles. Crack ID # 4.1: 2a = 0.0275 in. Crack ID #
4.2: 2a = 0.0036 in. Crack ID # 8: 2a = 0.0032 in. ...................................61
xi
Figure Page
4.23 Replica photograph of specimen 6611-a12, "front" notch,
60008 cycles. Crack ID # 4.1: 2a = 0.0379 in. Crack ID #
4.2: 2a = 0.0047 in. Crack ID # 8: 2a = 0.0032 in. ...................................62
4.24 Replica photograph of specimen 6611-a12, "front" notch,
65008 cycles. Crack ID # 4: 2a = 0.0441 in. Crack ID #
8: 2a = 0.0032 in. .......................................................................................62
4.25 SEM fractograph of specimen 6611-a12, "front" notch
fracture surface. The larger crack on the left was
identified as Crack ID # 6 during the replica measurement
process. The smaller crack on the right was identified as
Crack ID # 5 during the replica measurement process.
The reference line on the fractograph is 1000 μm in
length. ........................................................................................................63
4.26 Initiation site of Crack ID # 6. The reference line on the
fractograph is 100 μm in length.................................................................64
4.27 Close-up of initiation site of Crack ID # 6. The reference
line on the fractograph is 10 μm in length.................................................64
4.28 Initiation site of Crack ID # 5. The reference line on the
fractograph is 100 μm in length.................................................................65
4.29 Close-up of initiation site of Crack ID # 5. The reference
line on the fractograph is 10 μm in length.................................................65
4.30 Empirical expression for crack shape vs. nondimensional
length [28]. Measured values are from Crack ID #'s 5 and
6 from specimen 6611-a12, "front" notch..................................................66
4.31 Illustration of corner crack shape based on empirical
expression for c/a vs. a/t (to scale).............................................................67
4.32 Small crack growth rate vs. stress intensity factor range
data for Al 7050-T7451 (obtained from double-edge notch
specimens)..................................................................................................68
5.1 Geometry variable definitions used in the prediction
program for a surface crack and a corner crack.........................................74
xii
Figure Page
5.2 Geometry variable definitions used in the prediction
program for a typical multiple crack configuration...................................75
6.1 Actual and predicted crack growth for specimen 6612-b21,
back notch: surface crack length vs. number of cycles.............................84
6.2 Actual and predicted crack growth for specimen 6714-a11,
back notch: surface crack length vs. number of cycles.............................85
6.3 Actual and predicted crack growth for specimen 7111-b11,
back notch: surface crack length vs. number of cycles.............................86
6.4 Actual and predicted crack growth for specimen 7111-b12,
back notch: surface crack length vs. number of cycles.............................87
6.5 Actual and predicted crack growth for specimen 8B2, back
notch: surface crack length vs. number of cycles.. ...................................88
6.6 Actual and predicted crack growth for specimen 8B3, back
notch: surface crack length vs. number of cycles. ....................................89
6.7 Actual and predicted crack growth for specimen 8T3, front
notch: surface crack length vs. number of cycles. ....................................90
6.8 Predicted crack growth for specimen 6612-b21, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.......................91
6.9 Predicted crack growth for specimen 6714-a11, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.......................92
6.10 Predicted crack growth for specimen 7111-b11, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.......................93
6.11 Predicted crack growth for specimen 7111-b12, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.......................94
xiii
Figure Page
6.12 Predicted crack growth for specimen 8B2, back notch: c/a
vs. a/t for both the predicted values and the empirical
expression assumed in calculating ΔK solutions.......................................95
6.13 Predicted crack growth for specimen 8B3, back notch: c/a
vs. a/t for both the predicted values and the empirical
expression assumed in calculating ΔK solutions.......................................96
6.14 Predicted crack growth for specimen 8T3, front notch: c/a
vs. a/t for both the predicted values and the empirical
expression assumed in calculating ΔK solutions.......................................97
6.15 Stress intensity factor geometry for two offset parallel
cracks in a sheet under uniform uniaxial tensile stress [42]. .....................98
6.16 Actual and predicted crack growth for specimen 7012-a22,
back notch: surface crack length vs. number of cycles.
Note: no crack interaction is considered between the
cracks. ........................................................................................................99
6.17 Predicted crack growth for specimen 7012-a22, front
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.
Note: no crack interaction is considered between the
cracks. ......................................................................................................100
6.18 Actual and predicted crack growth for specimen 7012-a22,
back notch: surface crack length vs. number of cycles.
Note: no crack interaction is considered between the
cracks. The stress concentration factors were adjusted to
account for the crack initiating off the midplane of the
notch at an angle Θ (see Figure 3.3)........................................................101
6.19 Predicted crack growth for specimen 7012-a22, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.
Note: no crack interaction is considered between the
cracks. The stress concentration factors were adjusted to
account for the crack initiating off the midplane of the
notch at an angle Θ (see Figure 3.3)........................................................102
xiv
Figure Page
6.20 Actual and predicted crack growth for specimen 7012-a22,
front notch: surface crack length vs. number of cycles.
Note: the presence of Crack ID #'s 2 and 3 are ignored.. ........................103
6.21 Predicted crack growth for specimen 7012-a22, front
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.
Note: the presence of Crack ID #'s 2 and 3 are ignored.. ........................104
6.22 Actual and predicted crack growth for specimen 6611-a12,
back notch: surface crack length vs. number of cycles.
Note: the presence of Crack ID #'s 1.1 and 2 are ignored. ......................105
6.23 Predicted crack growth for specimen 6611-a12, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.
Note: the presence of Crack ID #'s 1.1 and 2 are ignored. ......................106
6.24 Actual and predicted crack growth for specimen 6611-a12,
back notch: surface crack length vs. number of cycles.
Note: the presence of Crack ID # 2 is ignored.........................................107
6.25 Predicted crack growth for specimen 6611-a12, back
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.
Note: the presence of Crack ID # 2 is ignored.........................................108
6.26 Actual and predicted crack growth for specimen 6714-a12,
front notch: surface crack length vs. number of cycles. ........................109
6.27 Predicted crack growth for specimen 6714-a12, front
notch: c/a vs. a/t for both the predicted values and the
empirical expression assumed in calculating ΔK solutions.....................110
6.28 Actual and predicted crack growth for specimen 8B3,
front notch, Crack ID # 1.22: surface crack length vs.
number of cycles. Note: the presence of other cracks are
ignored. ....................................................................................................111
xv
Figure Page
6.29 Predicted crack growth for specimen 8B3, front notch,
Crack ID # 1.22: c/a vs. a/t for both the predicted values
and the empirical expression assumed in calculating ΔK
solutions. Note: the presence of other cracks are ignored. .....................112
6.30 Actual and predicted crack growth for specimen 8B3,
front notch Crack ID #'s 1.21 and 1.22: surface crack
length vs. number of cycles. Note: the presence of other
cracks are ignored. ...................................................................................113
6.31 Predicted crack growth for specimen 8B3, front notch
Crack ID #'s 1.21 and 1.22: c/a vs. a/t for both the
predicted values and the empirical expression assumed in
calculating ΔK solutions. Note: the presence of other
cracks are ignored. ...................................................................................114
Appendix
Figure
C1 Crack tip locations for Specimen 6611-a12, back notch
(N=30,003 cycles)....................................................................................159
C2 Crack tip locations for Specimen 6611-a12, back notch
(N=40,007 cycles)....................................................................................159
C3 Crack tip locations for Specimen 6611-a12, back notch
(N=50,007 cycles)....................................................................................160
C4 Crack tip locations for Specimen 6611-a12, front notch
(N=45,007 cycles)....................................................................................161
C5 Crack tip locations for Specimen 6611-a12, front notch
(N=65,008 cycles)....................................................................................161
C6 Crack tip locations for Specimen 6612-b21, front notch
(N=29,005 cycles)....................................................................................162
C7 Crack tip locations for Specimen 6612-b21, front notch
(N=37,507 cycles)....................................................................................162
C8 Crack tip locations for Specimen 6612-b21, front notch
(N=47,511 cycles)....................................................................................163
xvi
Appendix
Figure Page
C9 Crack tip locations for Specimen 6612-b21, front notch
(N=62,516 cycles)....................................................................................163
C10 Crack tip locations for Specimen 6714-a11, back notch
(N=60,028 cycles)....................................................................................164
C11 Crack tip locations for Specimen 6714-a11, back notch
(N=75,033 cycles)....................................................................................164
C12 Crack tip locations for Specimen 6714-a11, back notch
(N=90,037 cycles)....................................................................................165
C13 Crack tip locations for Specimen 6714-a11, back notch
(N=110,045 cycles)..................................................................................165
C14 Crack tip locations for Specimen 6714-a12, front notch
(N=113,440 cycles)..................................................................................166
C15 Crack tip locations for Specimen 6714-a12, front notch
(N=130,630 cycles)..................................................................................166
C16 Crack tip locations for Specimen 6714-a12, front notch
(N=142,130 cycles)..................................................................................167
C17 Crack tip locations for Specimen 6714-a12, front notch
(N=163,630 cycles)..................................................................................167
C18 Crack tip locations for Specimen 7012-a22, back notch
(N=53,002 cycles)....................................................................................168
C19 Crack tip locations for Specimen 7012-a22, back notch
(N=65,002 cycles)....................................................................................168
C20 Crack tip locations for Specimen 7012-a22, back notch
(N=80,004 cycles)....................................................................................169
C21 Crack tip locations for Specimen 7012-a22, back notch
(N=94,006 cycles)....................................................................................169
C22 Crack tip locations for Specimen 7012-a22, back notch
(N=106,509 cycles)..................................................................................170
xvii
Appendix
Figure Page
C23 Crack tip locations for Specimen 7012-a22, front notch
(N=41,001 cycles)....................................................................................171
C24 Crack tip locations for Specimen 7012-a22, front notch
(N=56,002 cycles)....................................................................................171
C25 Crack tip locations for Specimen 7012-a22, front notch
(N=71,003 cycles)....................................................................................172
C26 Crack tip locations for Specimen 7012-a22, front notch
(N=89,005 cycles)....................................................................................172
C27 Crack tip locations for Specimen 7012-a22, front notch
(N=106,509 cycles)..................................................................................173
C28 Crack tip locations for Specimen 7111-b11, back notch
(N=231,068 cycles)..................................................................................174
C29 Crack tip locations for Specimen 7111-b11, back notch
(N=239,072 cycles)..................................................................................174
C30 Crack tip locations for Specimen 7111-b11, back notch
(N=247,075 cycles)..................................................................................175
C31 Crack tip locations for Specimen 7111-b12, back notch
(N=211,506 cycles)..................................................................................176
C32 Crack tip locations for Specimen 7111-b12, back notch
(N=221,002 cycles)..................................................................................176
C33 Crack tip locations for Specimen 7111-b12, back notch
(N=229,003 cycles)..................................................................................177
C34 Crack tip locations for Specimen 7111-b12, back notch
(N=233,003 cycles). 177
C35 Crack tip locations for Specimen 8B2, back notch
(N=45,008 cycles)....................................................................................178
C36 Crack tip locations for Specimen 8B2, back notch
(N=57,501 cycles)....................................................................................178
xviii
Appendix
Figure Page
C37 Crack tip locations for Specimen 8B2, back notch
(N=72,507 cycles)....................................................................................179
C38 Crack tip locations for Specimen 8B3, back notch
(N=42,506 cycles)....................................................................................180
C39 Crack tip locations for Specimen 8B3, back notch
(N=57,509 cycles)....................................................................................180
C40 Crack tip locations for Specimen 8B3, front notch
(N=27,503 cycles)....................................................................................181
C41 Crack tip locations for Specimen 8B3, front notch
(N=40,006 cycles)....................................................................................181
C42 Crack tip locations for Specimen 8B3, front notch
(N=50,007 cycles)....................................................................................182
C43 Crack tip locations for Specimen 8B3, front notch
(N=58,509 cycles)....................................................................................182
C44 Crack tip locations for Specimen 8T3, front notch
(N=145,002 cycles)..................................................................................183
C45 Crack tip locations for Specimen 8T3, front notch
(N=162,005 cycles)..................................................................................183
C46 Crack tip locations for Specimen 8T3, front notch
(N=182,008 cycles)..................................................................................184
C47 Crack tip locations for Specimen 8T3, front notch
(N=214,005 cycles)..................................................................................184
xix
ABSTRACT
Forsyth, Eric Nielsen. M.S.A.A., Purdue University, May 1993. Initiation, Growth, and
Coalescence of Small Fatigue Cracks at Notches. Major Professor: Dr. A. F. Grandt, Jr.
This research concerns the initiation, growth and coalescence of small fatigue
cracks at semicircular edge notches in the aluminum 7050-T7451 plate alloy. Three
versions of the alloy were provided by ALCOA, each with a varying degree of
microporosity. The objective of this study was to determine if a reduction in the amount
of microporosity resulted in improved small fatigue crack growth properties. Ten
double-edge notch specimens were tested at varying stress levels with a stress ratio of R
= 0.1. Fatigue crack growth was monitored with the replication method, providing
surface crack measurements as small as 0.0006 inches (15 microns). CT specimens for
all three versions of the alloy were fatigue tested to determine the large fatigue crack
growth properties.
Results from the CT specimen tests compared favorably with fatigue crack
growth rate vs. applied stress intensity factor range data generated previously by
ALCOA, and indicated that all three versions of the alloy had identical large fatigue
crack growth properties. Results from the double-edge notch specimen tests indicate that
after initiation, small fatigue cracks grow at faster rates than large fatigue cracks under
identical ΔK loading. All three versions of the alloy demonstrated similar small fatigue
xx
crack growth rate properties after initiation. However, the versions of the alloy with
reducedmicroporosity demonstrated longer fatigue lives to initiation than the version with
the most microporosity.
The small fatigue crack da/dN-ΔK curve was incorporated in a program to back-
predict the fatigue crack growth after initiation in double-edge notch specimens that
initiated a single crack. The predicted crack growth results to breakthrough showed
reasonable agreement with the data obtained from the specimen tests. However, future
tests should be conducted at different stress levels to generalize the results obtained in
this study.
1
CHAPTER 1 - INTRODUCTION
Fatigue cracks in engineering structures often originate at stress concentrations
such as fastener holes and notched components. These cracks can initiate at initial
defects such as voids and bonded inclusions within the engineering material. Research
conducted by ALCOA on the aluminum 7050-T7451 alloy [1] has demonstrated that the
fatigue life of edge-notch specimens can be improved by reducing the amount of
microporosity within the alloy. Since the majority of a fatigue crack's life in an
engineering structure can be spent in this "small" crack stage, it is of critical importance
to understand how all of these factors interact with one another to effect the crack's
subsequent growth.
The primary objective of this study is to determine how initial microporosity
effects the initiation and growth of fatigue cracks in the aluminum 7050-T7451 alloy. To
accomplish this, fatigue testing was performed on semicircular edge-notch specimens
fabricated from three versions of the alloy with varying levels of microporosity. Fatigue
crack growth was monitored from the point of initiation, enabling crack growth rate
information to be obtained for physically small cracks. Finally, an existing program was
modified to predict the growth of these cracks from the point of initiation in the
semicircular edge notch geometry.
2
CHAPTER 2 - BACKGROUND
Fatigue, the failure mode associated with cyclic loading, is often separated into
two stages: crack initiation and crack growth (Figure 2.1). Different methodologies have
been developed to treat the life of a crack through these two stages. Stress-life and
strain-life approaches are often used to quantify crack initiation life, while linear elastic
fracture mechanics (LEFM) can be used to quantify the growth of a crack with an initial
size, ao. The damage tolerance design philosophy lends itself particularly well to the
crack growth portion of fatigue life. Using this approach, engineers assume the pre-
existence of flaws in their design. Thus, it is desirable from an analysis point of view to
treat fatigue as primarily crack growth, i. e., have a single analysis method applicable to
all crack sizes. Unfortunately, there is no strict "boundary" where the LEFM assumption
breaks down. The purpose of this research is to monitor crack growth in an aluminum
alloy from the point of initiation, and apply LEFM principles into the "gray" area
between initiation and growth in an effort to predict crack behavior.
2.1. LEFM Concepts
LEFM assumes that crack growth is controlled by the stress intensity factor, K
[2]. This term, introduced by Irwin, relates loading, crack size, and specimen geometry,
and is often given in the form
3
K a a= σ π β( ) (2.1)
where σ is the remotely applied stress, a is the crack length, and β(a) is a dimensionless
function of the crack geometry. Paris, Gomez, and Anderson [3] first demonstrated that
the rate of fatigue crack growth (da/dN) is a function of the applied stress intensity factor
range (ΔK), independent of the particular loading, crack size, and specimen geometry, i.
e.,
da dN f K/ ( )= Δ (2.2)
This expression can be integrated to obtain the cyclic fatigue life
N dN
da
f K
N
a
a
o
f
= =z z0 ( )Δ
(2.3)
If the da/dN-ΔK expression is known for a particular material, these equations can be
incorporated into an algorithm to predict the cyclic fatigue life of a crack under different
loading conditions and geometries.
Although there are many K solutions available for two dimensional geometries,
there are few closed form K solutions for three dimensional geometries. Since fatigue
cracks initially start out having two dimensions (a surface length "a" and a crack depth
"c"), three dimensional K solutions are necessary to study the growth of small cracks.
Specimen geometries that incorporate a semicircular edge notch have been found to be
useful to monitor the initiation and growth of small cracks. Newman [4] has presented
approximate K solutions for corner and surface cracks in a semicircular edge notch.
4
These solutions were developed from finite element [5, 6] and weight function [7, 8]
methods for surface and corner cracks; from boundary force analyses of through cracks
at a semicircular notch [9]; and from previously developed equations for similar crack
configurations at an open hole [10]. The solutions are given in the form
K S a QF a c a t c r c w r t r wjn= π φ( / , / , / , / , / , / , ) (2.4)
The semicircular edge notch geometry and variable definitions are illustrated in Figure
2.2. It is important to note that for a corner crack in Figure 2.2, t is defined in the present
work as the specimen thickness B, whereas for a surface crack, t is defined as B/2. The
full stress intensity factor solutions used here are given in Appendix A. More
information on the actual test specimen design used in this research is given in Chapter 3.
2.2. The Small Crack Problem
Research conducted over the past two decades has shown that for certain
materials, physically small cracks (a ≤ 0.02 inches = 0.51 mm) grow at faster rates than
large cracks under the same ΔK loading. In addition, small cracks have been observed to
grow beneath the large crack threshold, ΔKth. These phenomena are known as the "small
crack effect". Schematic differences between the growth rates of small and large cracks
is illustrated in Figure 2.3 [11]. Since a significant portion of a crack's life in an
engineering structure may be spent as a small crack, any life predictions for that
component based on large crack data would be non-conservative. Thus, it is important to
determine if an engineering material exhibits this difference between the growth of small
and large cracks, and explain why it exists [11].
5
There are several factors that are believed to be involved in the small crack effect.
As mentioned earlier, LEFM assumptions are invalidated as the crack size approaches
zero due to the fact that the plastic zone size in front of the crack is on the same order of
magnitude as the crack size itself. Nonlinear and elastic-plastic fracture mechanics
concepts, such as the J-integral [12] and strain energy densities [12], have been used to
explain the short crack effect. In addition, the continuum assumption of LEFM [13] is
invalidated because grain boundaries as well as voids and inclusion particles affect the
local stresses near the small crack front. For a large crack, these metallurgical effects are
averaged out over the larger crack's long front.
However, there are LEFM concepts which, in part, help explain the small crack
effect. In particular, crack closure has been shown to play an important role in the
accelerated growth rates of small cracks. First proposed by Elber [14], crack closure is
the concept that a crack is not fully open until a "crack opening stress" is reached. This
phenomenon can be attributed to several factors, including plastically deformed material
in the wake of a crack, crack surface roughness, and oxide debris on the crack surface.
All of these factors hinder the opening of a crack, resulting in a stress level that must be
reached before the crack can be fully open and thus propagate. It is believed that small
cracks have smaller crack opening stresses than large cracks do. Therefore, small cracks
would experience a larger effective stress intensity factor range than large cracks, even
though they are experiencing identical ΔK loading. This phenomenon is illustrated in
Figure 2.4 in a K vs. time graph.
In order to study the small crack effect, researchers have developed several
methods for measuring small fatigue cracks. Perhaps the most accurate method for
measuring small cracks is with the scanning electron microscope (SEM). When used in
6
conjunction with stereo imaging, SEM photographs provide useful information in the
closure behavior of small cracks [15]. Although the SEM has both the spatial and strain
resolution for the scale involved, cost makes its use prohibitive for routine laboratory
measurements.
Sharpe [16] has developed the interferometric strain-displacement gage (ISDG)
which acts as a non-contacting extensometer for the specimen. Two indentations are
made with a Vickers hardness tester above and below a surface crack. The diffraction
patterns created by a laser impinging on the indentations can be used to determine crack
opening displacement and thus crack size. Although the ISDG can be used for computer
control and real-time measurement of small fatigue crack tests, the location of the
initiated crack must be known before measurements can be taken.
Another method which allows for computer data acquisition is the direct current
electrical potential measurement (dcEPM) of small cracks [17]. If a current is passed
through a specimen containing a crack, the voltage difference across the crack can be
correlated to the crack length. Drawbacks to the dcEPM method include cost, the
necessity for the specimen to conduct electricity, and the fact that it has only been used
on cracks artificially created with electric discharge machining.
Resch and Nelson [18] have developed an ultrasonic method for the measurement
of small cracks. The method uses surface acoustic waves on the specimen to determine
surface crack depth; in this sense, the method is similar to the SONAR employed by
naval craft to determine underwater features.
7
A relatively simple, but more time consuming method for the measurement of
small cracks is the replication method [19]. It uses an acetate tape which makes an exact
replica of the specimen surface when acetone is applied to the surface. The method can
be used for a variety of specimen geometries and crack length measurements as small as
0.0002 inches (5.1 μm) have been obtained. Unfortunately, only the surface crack length
can be measured with this method - not the crack depth. The research presented in this
thesis utilized the replication method for the measurement of small cracks. A more
thorough discussion on the specifics of the replication method and how it was used in
conjunction with this research is presented in Chapter 3.
In 1984, an AGARD Cooperative Test Program was initiated to investigate the
small crack growth behavior under various loading conditions for the aluminum alloy
2024-T3, a common material used in airframe components [20]. Twelve participants
from nine different countries monitored the growth and coalescence of nearly 950 cracks
in over 250 single edge notch specimens. The tests were conducted at three different
stress levels for both constant amplitude loading (stress ratios, R ≡ minimum/maximum
stress = -2, -1, 0, and 0.5) and spectrum loading (FALSTAFF and GAUSSIAN)
conditions. Surface crack lengths were measured with the replication technique. The
participants involved in the test program showed good agreement on the small crack
growth rates, cyclic fatigue life to crack breakthrough (surface and/or corner cracks
became a through crack), and on crack shapes. The small cracks initiated in the tests
demonstrated the small crack effect mentioned previously by growing below the large
crack ΔK threshold and growing at faster growth rates than large cracks above the
threshold.
8
A fatigue crack growth model accounting for crack closure was developed by
Newman [21] to predict the growth of small cracks from small voids and inclusion
particles on the notch surface. The initial defect size was chosen to approximate the
initiation sites of the cracks monitored in the tests. The model was based on the Dugdale
strip-yield plastic zone [22], but modified for closure by leaving plastically deformed
material in the wake of the crack. Lee and Sharpe's experimentally measured values for
the crack opening stresses (obtained from the ISDG method) [23] showed good
correlation with Newman's analytical model, increasing the confidence in the model.
There was reasonable agreement between the experimental and predicted values for the
small crack growth rates, although the model predicted slightly slower growth rates for R
= -2 loading, and slightly faster growth rates for R = 0.5 loading. However, the model
did indicate that the small crack effect was most predominant in the tests involving
significant compressive loads. This behavior was observed in the tests themselves.
In order to allow participants to test various materials and loading conditions that
were of particular interest to their laboratories, an AGARD Supplemental Test Program
on the growth of small cracks was initiated [24]. The materials tested in the
supplemental program were: 2024-T3 and 7075-T6 aluminum alloys, 2090-T8E41
aluminum-lithium alloy, Ti-6Al-4V titanium alloy, and 4340 steel. The results from the
supplemental program were similar to the first program in that all the materials exhibited
the small crack effect to some extent. However, the effect was less pronounced in some
materials (e. g., 4340 steel) than in others. Once again, the crack growth model predicted
small crack growth rates in reasonable agreement with the experimental measurements
for most loading conditions.
9
2.3. The 7050-T7451 Aluminum Alloy
In an effort to reduce both the size and frequency of potential microporosity in
their aluminum 7050-T7451 plate alloy, the Aluminum Company of America (ALCOA)
has improved their processing techniques for the material over the past decade. Smooth
axial fatigue tests of material produced in 1985 following the process improvements have
resulted in longer fatigue lifetimes than material produced prior to the improvements [1].
Post-test fractography of the specimens fabricated from both materials revealed the size
of the micropores that resulted in crack initiation and subsequent fracture. This
microporosity size distribution was subsequently used in a probabilistic crack growth
analysis, which demonstrated analytically that the reduced microporosity material should
perform better in service than the older material with larger micropores. However, since
smooth axial fatigue tests do not take into account cracks originating from machining
defects, a test program was initiated to examine whether these type of flaws obscure the
process improvements resulting in microporosity reduction.
The objective of this program was to demonstrate the effect of microporosity on
an engineering detail, specifically, a notched specimen subjected to constant amplitude
loading [1]. The material was obtained from a single lot of 5.6 inch (14.2 cm) thick
7050-T7451 plate. Specimens fabricated from the mid-plane of the plate had a higher
degree of microporosity than the specimens fabricated from the quarter-plane of the plate.
The test specimens were 0.126 in. (3.2 mm) thick, 1.00 in. (25.4 mm) wide, and 9.00 in.
(229 mm) long, with two holes of 0.187 in. (4.75 mm) diameter located 1.00 in. (25.4
mm) apart. The goals of this specimen were to provide a symmetric stress field and to
increase the chances that a micropore would be located near a stress concentration. The
10
specimens were cycled to failure at a stress ratio R = 0.1 at maximum stress levels of 10,
12, and 20 ksi (69, 83, and 138 MPa).
Results from these tests show a significant improvement in the fatigue properties
of the low microporosity (quarter-plane) material; this can be seen in the test specimens'
log-life versus log-maximum stress plot of Figure 2.5 [1]. For example, a component
designed for a lifetime of 100,000 cycles could see a maximum stress of 110 MPa in the
low microporosity material as opposed a maximum stress of 98 MPa in the high
microporosity material; this represents an improvement of 12 percent in stress level [1].
All specimen failures in this study initiated at micropores as opposed to
machining defects. The largest micropore initiating a crack in the test program was 0.030
in. (0.75 mm); the average size of a crack initiating micropore, however, was 0.012 in.
(0.31 mm). Both of these sizes fall below current nondestructive inspection (NDI)
capabilities, which can reliably detect flaw sizes of 0.04-0.08 in. (1-2 mm) [1]. Thus,
ALCOA employed destructive techniques such as SEM examination of the fracture
surfaces to quantify the microporosity distribution. This examination revealed that the
frequency of micropores that initiated cracks in the specimens to be the major difference
between the fracture surfaces of the two versions of the material. The high porosity
(mid-plane) version of the material initiated on average a greater number of cracks (2.25
per specimen) than the low porosity (quarter-plane) version (1.33 per specimen) [1].
One of the main conclusions from the test program was that initial material
quality should be considered in the design process. To accomplish this, ALCOA utilized
the United States Air Force (USAF) Advanced Durability Analysis. This method is
based on the concept of an equivalent initial flaw size (EIFS) which represents the initial
11
microporosity distribution in the material. Since all the cracks in the specimens initiated
at micropores, an EIFS distribution (calculated via LEFM principles) based on these tests
could theoretically be equated with the actual initial microporosity distribution of the
material (determined from the earlier smooth axial fatigue tests). Subsequent analysis
demonstrated this hypothesis; the two distributions were very similar, and predicted
specimen lifetimes when used as input in a probabilistic fracture mechanics analysis [1].
Out of this test program arose two objectives for further research. First, it was
desired to further develop and assess the benefits of the probabilistic approach to
durability. ALCOA, in collaboration with Wright Laboratory's Flight Dynamics
Directorate (USAF) [25, 26], has demonstrated through further testing and analysis that a
reduction in the microporosity of Al 7050-T7451 can result in the increased performance
and reduced cost of airframe components where durability is a major design factor. In
addition, they confirmed that the USAF probabilistic failure model captured this
advantage in improved material quality, whereas more conventional fatigue design
practices did not.
ALCOA's second objective was to further quantify crack growth from micropores
by studying the effect of microporosity on the growth of physically small cracks. To
accomplish this, ALCOA initiated a test program to monitor the initiation and growth of
small cracks in the low and high microporosity versions of the 7050-T7451 plate [27].
The specimen design incorporated four semicircular edge notches, two on each side, of a
0.125 in × 2.00 in ×12.00 in (3.2 mm × 51 mm × 305 mm) rectangular specimen. Crack
initiation and growth at the notches was monitored with the replication method. After
fatigue testing of the specimens, the replicas were covered with approximately 100 - 200
Angstroms of gold so that crack measurements could be made with the SEM. Both
12
actual lengths and projected lengths of the cracks were obtained from the replicas with an
automatic image analysis system (IBAS) [27, 28]. In addition, fractography was
performed on the fractured specimen surfaces to examine the crack initiation sites.
From these small crack tests, ALCOA researchers have obtained small crack
length, L, versus number of cycles N, as well as dL/dN-ΔK plots. Although data analysis
is still being performed, some qualitative observations could be made from the
preliminary results. First, the material with the low microporosity initiated cracks later
than the material with the higher microporosity. In addition, large pores in the materials
appear to have the greatest influence on crack initiation and propagation. Finally, the
dL/dN-ΔK plots show little difference between the low and high microporosity versions
of the alloy [27].
The research presented in this thesis is an extension of ALCOA's effort to
determine the effect of microporosity on the initiation and growth of small cracks in the
Al 7050-T7451 alloy. Three versions of the material were supplied for this effort. The
versions of the material shall be referred to in this thesis as "old", "new", and "three-inch
plate" material. Both the "old" and "new" materials were obtained from a six-inch plate,
and contain more microporosity than the "three-inch plate" version of the aluminum
alloy. This is due to the fact that the three-inch version of the material was rolled for a
longer period of time than the six-inch version, effectively "squeezing" out any remaining
microporosity.
Large crack da/dN-ΔK data for the Al 7050-T7451 alloy is shown in Figure 2.6
[29]. Although the large crack growth rates for all three versions of the material exhibit
the same da/dN-ΔK relationship [30], it is believed that the initial microporosity in each
13
of the three versions will effect the small crack growth rates in different ways. It was
hoped that the reduced microporosity versions of the material would delay crack
initiation, and exhibit better overall fatigue properties, thus justifying its increased cost.
The purpose of this research is determine whether this assumption is true by performing
small fatigue crack tests on all three versions of the material. In addition, an existing
crack growth prediction program was modified to analyze the growth of small fatigue
cracks from semicircular edge notches based on the experimental results.
Figure 2.1 The two stages of fatigue.
Figure 2.2 The semicircular edge notch geometry and variable definitions.
Figure 2.3 Typical fatigue crack growth rate data for large and small cracks [11].
Figure 2.4 Stress intensity factor vs. time for constant ampltide loading. Lower crack opening stress for small cracks results in
a larger effective stress intensity factor than a large crack under identical loading, translating into faster growth
rates.
Figure 2.5 Cumulative fatigue failure distributions from 1984-1987 for the 7050-T7451 thick plate (5.5-5.9 inches = 140-150
mm).
Figure 2.6 Large crack da/dN-ΔK data for the Al 7050-T7451 alloy, R=0.1 [29].
20
CHAPTER 3 - EXPERIMENTAL PROCEDURES
In this chapter, the experimental procedures for the test program are presented.
The first section covers the procedures involved with the acquisition of small crack
growth rate data, while the second section covers the procedures involved with the
acquisition of baseline data through large crack testing.
3.1. Small Crack Specimen Design and Testing Procedures
As mentioned previously, specimens which incorporate a semicircular edge notch
are useful in the procurement of small crack growth rate data. The original specimen
design used in this study was a double-edge notch dogbone specimen, and is illustrated in
Figure 3.1. The test specimen was secured to the load frame through pin-hole grips.
Two semicircular edge notches were placed on opposite sides of the specimen in order to
increase the amount of obtainable data in a single test. However, two constraints are
placed on this type of specimen design. First, the width of the specimen must be large
enough so that the growth of a small crack in one notch is not affected by the presence of
the opposite notch and/or other small cracks growing in the opposite notch. At the same
time, the width is limited by the diameter of the pins used to grip the specimen. Trial
tests would often fail in the pinhole grip area if the dog boned width was greater than the
0.75 inch (1.9 cm) diameter of the pins. Two tests were successfully performed with this
specimen design, but considering the constraints involved, a better design was required.
21
Discussions with ALCOA personnel on this problem [30] centered on the method
used to grip the specimen. To circumvent specimen failure in the grips, ALCOA
supplied this study with a grip design that "clamped" the specimen to the grips. The
normal force applied to the specimen faces generates enough friction to prevent the
specimen from slipping out of the grips. These "friction" grips allowed the specimen to
be simplified to a double-edge notch specimen with no dogbone. This rectangular-
shaped specimen is illustrated in Figure 3.2, and was the design used in this rest of the
test program. By increasing the specimen width to 2.00 inches (5.08 cm), this ensured
that the two notches of radii = 0.188 inch (4.78 mm) interacted little with each other.
A two-dimensional finite element analysis was performed on both the dogbone
and rectangular geometries to determine if the specimen width had any effect on the
stress distribution at the notch. The stress concentration factor, Kt, is defined as the ratio
of the hoop stress divided by the remote stress. In Figure 3.3, Kt's obtained from the
finite element analysis are plotted versus the angle off of the mid plane of the notch, Θ,
for both geometries. The figure reveals that both geometries exhibit essentially the same
stress distribution at the notch. Therefore, the crack initiation data obtained from both
specimen geometries were treated as equivalent in this study.
To ensure that crack initiation occurs at material inhomogeneities and not
machining marks, the notch surfaces were polished down to a 600 grit, followed by a
diamond paste. The surface is then etched with a Keller's etch for 10-30 seconds. This
removes any residual stresses generated by the machining and polishing process, and
provides a "map" of the notch surface by extracting grain boundaries.
22
3.1.1. Specimen Testing
Specimens fabricated from all three materials were tested in a servo-hydraulic test
machine with analog-based electronic controls in laboratory air under a constant stress
ratio (R = 0.1). Strain gages were placed on both sides of the specimen to measure the
difference in strain experienced during axial loading, giving an indication of the bending
present in the specimen. All tests involved applying two or three cycles to the maximum
load to ensure the specimen was aligned correctly within the grips, and that potential
bending strains were less than 5 % of the strains induced by the axial loading at the
commencement of testing. Most specimens were loaded at a maximum nominal stress
σnom =16 ksi (110 MPa), although two specimens were loaded at σnom = 15 ksi (103
MPa) and one specimen at σnom = 18 ksi (124 MPa). Table 3.1 lists the test parameters
for the ten semicircular edge notch test specimens.
"Old" material specimens are identified by the number 6•••-••, and were
obtained from blanks of the aluminum 7050-T7451 alloy with the most microporosity.
"New" material specimens are identified by the number 7•••-••, and were obtained from
blanks of the material with less microporosity. The blanks themselves were formed from
a six-inch thick plate at the ALCOA Technical Center. In addition, the numbering
system for the "old" and "new" material specimens is an abbreviation of the numbering
system ALCOA provided with the blanks. "Three-inch plate" specimens are identified
by the number 8••, and were obtained from a three inch thick plate of aluminum 7050-
T7451. This version of the material has the least amount of microporosity of all three
materials [30].
23
3.1.2. The Replication Method
As mentioned previously, surface crack initiation and growth were monitored
with the replication method [19]. Cycling was suspended periodically throughout the
test, and the specimen held under a constant tensile load while the notch surface was
replicated. The tensile load was equal to eighty percent of the mean load, ensuring that
all crack faces were open in the notch and thus making detection of the crack easier. The
notch surface was bathed with 1-2 drops of acetone from a hypodermic needle. Finally, a
0.003 inch (76 μm) thick acetate tape was placed within the notch; this is shown
schematically in Figure 3.4. The acetone softens the tape, allowing it to conform to the
notch surface and flow into the mouths of open cracks. Great care must be taken during
the replication process so that no air bubbles are trapped between the notch surface and
the tape. No information of the notch surface is transferred to the tape where a bubble is
located. After approximately 25 seconds, the tape is dry, leaving an exact replica of the
notch surface. At this point, the tape can be removed from the notch surface and testing
can recommence. Approximately 25-50 replicas were taken throughout each test to
sufficient enough data points are available for analysis.
Once the fatigue test was completed, analysis of the replicas begins. In several of
the tests, individual cracks coalesced into a single crack. To keep track of the crack
coalescence process, the following crack identification system was developed. When
measuring the cracks from the replicas, the last replica taken was examined first. This
replica would usually include a through-the-thickness crack, and sometimes smaller
surface and corner cracks that did not become the dominant crack. Each of these cracks
would be given an integer identification number 1, 2, 3, etc. As these cracks were traced
back in time through earlier replicas, an initial crack, say Crack ID # 2, would "divide"
into two smaller cracks (i. e. crack coalescence). These two cracks would then be given
24
the identification numbers 2.1 and 2.2, indicating that they coalesced into Crack ID # 2 at
a later time in the test. Similarly, Crack ID # 2.1 could "divide" into Crack ID's # 2.11
and # 2.12 as they were traced through earlier replicas. This crack identification system
provides a simple means to keep track of crack coalescence history, and hopefully aids in
following this coalescence process in a single plot of crack length versus cycles for a
particular specimen notch.
Cracks were measured from the replicas via two different methods. Larger
cracks, defined as a ≥ 0.003 inches (76 μm) were measured with a low powered
(magnification ≈ 7 ×) optical microscope. Replicas were mounted on a slide viewing
stage and the crack tip coordinates were measured using two micrometers attached to the
stage. The micrometers provided resolutions of 0.0001 inches (2.5 μm). The crack tip
coordinates were then converted to the x-s coordinate system and subsequently into the
x-Θ coordinate system. The x-coordinate is the distance along the bore of the notch.
The s-coordinate is defined as the notch radius × Θ, where Θ is the angle in radians
above/below the mid plane of the notch; see Figure 3.5. This determined the spatial
location of the crack within the notch, and subsequently its length. For cracks smaller
than 0.003 inches (76 μm), a higher powered optical microscope was used. This
microscope provided magnifications up to 1120 ×, and crack lengths were measured from
a video screen connected to the microscope. Due to the limited viewing field, only crack
lengths could be obtained from this method - not crack tip coordinates. However, spatial
location of the cracks along the notch bore could be obtained from other measurements
with the other microscope once crack lengths become larger. Cracks lengths in the range
a ≈ 0.003 inches (76 μm) were measured with both methods; these lengths showed good
agreement with each other.
25
An important concern with measurement is that the replicas would shrink 5-10%
as they dried on the specimen surface. Therefore, the measurements were normalized
with a shrinking factor. This factor was simply the ratio of the known notch thickness to
the measured replica width, providing a scale for all measurements made on that replica.
Small crack growth experiment results from the semicircular edge notch
specimens are presented in Chapter 4.
3.2. Large Crack Testing Procedures
ALCOA researchers have established the large fatigue crack growth properties for
the aluminum 7050-T7451 alloy through numerous fatigue tests under various loading
conditions and specimen geometries [31]. However, for completeness it was decided to
quantify large fatigue crack growth rate properties for the alloys. In addition,
supplementary large crack testing would further substantiate ALCOA's belief that all
three versions of the alloy exhibited the same large crack growth properties [30].
To accomplish this, compact tension (CT) specimens were fabricated from
fractured semicircular edge notch specimens, as is illustrated in Figure 3.6. The CT
specimens were designed in accordance with ASTM Standard E647 [32], and its
geometry is shown in Figure 3.7. The CT specimens were fabricated from fractured
semicircular edge notch specimens to conserve the material used in this study. Although
the CT specimens consist of material that has been previously cycled, it is believed that
once a pre crack has started in the specimen, the large crack growth properties are
relatively unaffected by the previous loading.
26
Table 3.1 lists the test parameters for the four CT test specimens. CT specimen
pre cracking was conducted according to ASTM Standard E647 [32]. Traveling
microscopes were mounted on both sides of the specimen in order to obtain front and
back through-crack lengths. One of the microscopes was attached to a digital measuring
system accurate to 0.0005 inches (13 μm). However, due to equipment problems with
the second digital measuring system, a microscale accurate to 0.005 inches (130 μm) was
used to obtain crack lengths with the other microscope. Through crack lengths were
taken as the average of the front and back crack lengths, and were recorded to the nearest
0.005 inch (130 μm).
The large crack growth rate data obtained from the CT specimen tests are
presented in Chapter 4.
Table 3.1 Parameters for fatigue test specimens.
Specimen ID Material Type Specimen
Type
Max. Nominal
Stress
(ksi / MPa)
Stress Ratio Frequency
6714-a11 "old" dogbone DEN 15 / 103 0.1 5 Hz
6714-a12 "old" dogbone DEN 15 / 103 0.1 10 Hz
6612-b21 "old" friction DEN 16 / 110 0.1 10 Hz
6611-a12 "old" friction DEN 16 / 110 0.1 10 Hz
7111-b11 "new" friction DEN 15, 18 / 103, 124 0.1 8 Hz
7111-b12 "new" friction DEN 16 / 110 0.1 10 Hz
7012-a22 "new" friction DEN 16 / 110 0.1 10 Hz
8T3 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz
8B3 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz
8B2 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz
Max. Load for
CT Specimens
(lbs / N)
6611-a12-CT2 "old" CT 350 / 1560 0.1 10 Hz
7012-a21-CT4 "new" CT 350 / 1560 0.1 10 Hz
8T3-CT3 "3-inch plate" CT 450 / 2000 0.1 10 Hz
8T3-CT4 "3-inch plate" CT 450 / 2000 0.1 10 Hz
Figure 3.1 Double-edge notch dogbone specimen geometry and dimensions.
Figure 3.2 Double-edge notch specimen geometry and dimensions.
Figure 3.4 Illustration of the replication process.
Figure 3.5 Definition of replica coordinate system.
Figure 3.6 CT specimens fabricated from fractured double-edge notch specimen.
Figure 3.7 CT specimen geometry and dimensions.
Theta vs. (Remote Stress/Hoop Stress)
Semicircular Edge Notch Geometries
Theta (radians)
Srem/Shoop
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Width = 1.11 in.
Width = 2 in.
Figure 3.3 Hoop Stress/Remote Stress, σhoop/σrem, vs. Theta, Θ, for the 1.11 and 2.00 inch wide specimen geometries.
σhoop/σrem was caculated with a 2-dimensional finite element analysis of each of the specimen geometries. Θ is
defined as the angle (in radians) from the tip of the notch to the upper/lower location where the notch meets the
specimen edge.
35
CHAPTER 4 - EXPERIMENTAL RESULTS
4.1. Large Crack Growth Rate Data
Four CT specimens were tested to obtain the large crack growth properties for all
three versions of the aluminum 7050-T7451 alloy. All CT specimens were tested at a
frequency of 10 Hz and a stress ratio of 0.1. The "old" material CT specimen was loaded
at a maximum load of 300 lbs (1330 N). The "new" material CT specimen was loaded at
a maximum load of 350 lbs (1560 N). Finally, two "3-inch plate" material CT specimens
were loaded at maximum loads of 300 lbs (1330 N) and 450 lbs (2000 N). All CT
specimen testing and data analysis was performed in accordance with ASTM Standard
E647 [32]. The stress intensity factor range solution for the CT specimen geometry is
given by
Δ
Δ
K
P
B W
=
+
−
+ − + −
( )
( )
( . . . . . )/
2
1
0 886 4 64 13 32 14 72 5 63 2
2 3 4α
α
α α α α (4.1)
where a ≡ crack length, W ≡ width, B ≡ thickness, ΔP ≡ applied load range, and α = a/W.
Fatigue crack growth rates were calculated with a seven-point polynomial technique [32].
The fatigue crack growth rates for the CT specimens are plotted against the
applied stress intensity factor range in Figure 4.1. Two important things can be discerned
from this
36
"large" crack growth rate curve. First, all three versions of the alloy exhibit essentially
the same large fatigue crack growth rate properties. Second, the CT crack growth rate
data correlates well with numerous fatigue tests performed by ALCOA under various
loading conditions and specimen geometries. The Paris law expression shown in Figure
4.1 for Al 7050-T7451 (R = 0.1) was obtained from Reference [29] for the data
reproduced here in Figure 2.5, and is given by
da
dN
K= × −
3 9 10 10 4 175
. ( ) .
Δ (4.2)
The units for ΔK in Equation 4.2 are ksi√in, while da/dN is measured in inches/cycle.
Although CT test data obtained here do not extend into either the threshold ΔK or the
fracture toughness regions of the da/dN-ΔK curve, it does correlate well with the
ALCOA generated data shown in Figure 2.5 [29]. Thus, the ALCOA Paris Law
expression (Equation 4.2) is used here for subsequent analysis of the materials' large
crack growth properties.
4.2. Small Crack Test Results
Table 4.1 summarizes the results of the double-edge notch specimen tests. Before
studying the results of individual tests, some general information should be noted first.
The terms "front" and back" identify the notch location relative to the servo-hydraulic
test machine. The table indicates that cracks initiated at an equal rate in both the front
and back notches for all the specimens tested. This provided added assurance that
potential bending was kept to a minimum in the tests, i. e., there was no bias as to which
notches caused crack initiation.
37
In addition, two different crack lengths are used here to define fatigue crack
"initiation." Although crack lengths of 2a<0.001 inch (25 μm) were obtained, several
tests had cracks of that length traceable back to "zero" cycles. In actuality, however, the
term "zero" cycles does not include specimen loading which occurred during gripping
and alignment procedures. All tests involved applying two or three cycles to the
maximum load to ensure the specimen was aligned correctly within the grips, and that
potential bending strains were less than five percent of the strains induced by the axial
loading. Therefore, a more generous initiation length of 2a=0.005 inch (127 μm) was
also included in the table. It is important to note that both definitions of "initiation" place
the crack length well within the small crack region of 2a < 0.02 inch (500 μm).
Surface crack length vs. number of elapsed cycles for the double-edge notch tests
are plotted in Figures 4.2 - 4.14. Surface crack lengths are plotted until "breakthrough",
i. e., until the surface crack has become a through crack at the notch.
In several of the tests, a single crack initiated at approximately the center of the
notch and grew into the through crack that eventually caused specimen failure. A series
of replica photographs in Figures 4.15 - 4.20 illustrates the growth of a lone crack in the
"back" notch of specimen 6612-b21. In Figure 4.15, a crack appears to be emanating
from a micropore at "0" cycles. By 10,001 cycles (Figure 4.16), the crack has grown and
established itself. Figures 4.17 - 4.20 follow the growth of the crack at a lower
magnification from 10,001 cycles to 42,509 cycles.
Some tests, however, were characterized by multiple cracks initiating at several
points along the bore of the notch. These cracks in turn coalesced into larger cracks, with
a dominant crack eventually leading to specimen failure. Specimen 6611-a12 (Figure
38
4.3) is the most prolific example of multiple crack initiation, with ten different cracks
initiating in the front notch. The most likely reason for the large number of cracks is the
high degree of microporosity in the "old" material, resulting in a greater number of
initiation sites in this specimen. An interesting phenomenon associated with multiple
crack initiation / interaction is illustrated in specimen 7012-a22 (Figures 4.7 and 4.8).
For some of the cracks, the final length measurements are smaller than measurements
taken at previous cycles. It may be possible that extension of the large dominant crack
prevents complete opening of adjacent smaller cracks, and thus makes them appear to be
smaller as life progresses. For example, in Figure 4.8, Crack ID #1 in the final
measurement is a through crack; it is fully open. However, Crack ID #'s 2 and 3 are only
small surface cracks compared to #1, and are only partially open in the final
measurements.
A series of replica photographs in Figures 4.21 - 4.24 illustrate multiple cracks
interacting with one another in the front notch of specimen 6611-a12. At 42,507 cycles
(Figure 4.21), Crack ID #'s 4.1 and 4.2 are shown in the center and upper-right hand
corner, respectively, while Crack ID # 8 is essentially a micropore to the left of # 4.1. By
50,007 cycles (Figure 4.22), # 8 has established itself, while # 4.1 is growing towards
both # 4.2 and # 8. At 60,008 cycles (Figure 4.23), however, # 4.1 has bypassed # 8, and
has almost coalesced with # 4.2. By 65,008 cycles (Figure 4.24), cracks 4.1 and 4.2
have coalesced into Crack ID # 4. Crack ID # 8 is starting to close due to its close
proximity to the larger # 4.
Examination of the fracture surfaces provides another method in determining
fatigue crack initiation. ALCOA researchers have examined the fracture surfaces of Al
7050-T7451 open hole fatigue specimens with the scanning electron microscope (SEM)
39
in order to locate crack initiation sites [1]. In that study, they determined that the fatigue
cracks initiated from micropores in the material rather than machining flaws. SEM
examination of this study's double-edge notch specimen fracture surfaces is currently
being performed by Jon Elsner on a JEOL JSM-T300 SEM [33]. The accelerating
voltage is 25 kV, and utilizes background scatter electrons as the imaging technique. An
example of Elsner's current work is presented here to illustrate the technique and initial
results.
Figure 4.25 is a fractograph of the "front" notch fracture surface for specimen
6611-a12, and shows two cracks which developed at this notch. Although catastrophic
failure initiated at the "back" notch of this specimen, the elliptical shapes of the dominant
cracks in the "front" notch were preserved. The larger crack on the left was identified as
Crack ID # 6 during the replica measurement process, while the smaller crack on the
right was identified as Crack ID # 5. Figures 4.26 - 4.29 show larger magnifications of
the initiation sites for Crack ID #'s 5 and 6. In both cases, the initiation sites appear to be
micropores in the material just beneath the notch surface.
4.3. Small Crack Growth Rate Data
Specimens where a single crack initiated in one of the notches were used here to
characterize the small fatigue crack growth rate data. As shown in Equation 2.4,
Newman has developed approximate K solutions for corner and surface cracks in a
semicircular edge notch [4]. Variable definitions are illustrated in Figure 2.3, whereas
noted previously t is defined as B for a corner crack, whereas for a surface crack, t is
defined as B/2. Newman's full stress intensity factor solutions are given here in
Appendix A.
40
It is important to note that the specimen design met all restrictions placed on the
edge notch geometry for the K solution to be valid except for the requirement that r/w =
0.0625 (see Appendix A for complete geometry restrictions). For the early dog bone
double-edge notch specimens, r/w ≈ 0.0845; for the friction grip double edge-notch
specimens, r/w ≈ 0.0469. As mentioned previously in Chapter 3, a finite element
analysis was performed to determine the stress distribution at the notch for both specimen
geometries. The stress concentration factor Kt at the mid plane of the notch (Θ = 0) was
calculated to be Kt = 3.03 for the dogbone geometry and Kt = 3.05 for the friction grip
geometry. These results are 3.5 % less than the stress concentration factor used in the
Newman ΔK solutions of Kt = 3.15 for uniform displacement [4]. Because of the close
correlation between the finite element analysis Kt's and the Kt used in Newman's stress
intensity factor solutions, the r/w restriction was considered insignificant in this study.
The replication method can only obtain the surface lengths of cracks, or "2a", and
not the crack depths, "c", defined in Figure 2.2. Since the stress intensity factor solution
for cracks at a semicircular edge notch depends on the crack aspect ratio, a/c, an
expression for a/c is required to calculate ΔK's for the double-edge notch specimen.
Swain and Newman measured crack lengths in both the a and c direction with the use of
marker loads in the 2024-T3 aluminum alloy [34]. Based on the experimental data, they
developed an empirical relationship between the crack shape, c/a, and the non
dimensional length, a/t, given by
c a = −0.9 0.25(a t)2
(4.3)
41
This expression is plotted in Figure 4.30. In addition, a representation of the crack
shapes predicted by the expression is shown to scale in Figure 4.31 for a corner crack.
As mentioned earlier, the SEM examination of the fracture surface of specimen 6611-
a12 allows for actual crack shape measurements to be obtained for Crack ID #'s 5 and 6
in the "front" notch. These measurements are also plotted in Figure 4.30, and correlate
well with Swain and Newman's empirical prediction. Although an exhaustive
examination of all specimen fracture surfaces has not been performed at this time, the
empirical expression for c/a vs. a/t should be adequate for calculating the stress intensity
factor ranges.
The small crack growth rates for double-edge notch specimens where a single
crack initiated along the bore of a notch are plotted against the applied stress intensity
factor ranges in Figure 4.32. The Paris Law expression obtained from ALCOA fatigue
tests for Al 7050-T7451 (Equation 4.2) is also plotted for comparison. The crack growth
rate data obtained from these tests demonstrates that small cracks do, in fact, grow faster
than large cracks at equivalent ΔK loading near the threshold region. However, the small
crack growth rate data merges with the large crack Paris Law at higher ΔK's. A linear
regression was performed on the data to obtain the Paris Law constants for the small
crack growth rate data. The "small" crack Paris Law expression is given by
da
dN
K= × −
8 22 10 9 2 807
. ( ) .
Δ (4.4)
The units for ΔK in Equation 4.4 are ksi√in, while da/dN is measured in inches/cycle. It
is important to note that there is greater variability in the small crack growth rate data
compared to large crack growth rate data. This is not surprising due to the fact that
LEFM principles are being pushed to the limit as well as uncertainties in the small crack
42
measurements. The Paris Law expressions for both small and large crack growth rate in
Al 7050-T7451 are incorporated in a computer program that predicts the crack growth of
surface and corner cracks in a variety of geometries. In addition, the single crack growth
rate Paris Law expressions will be used to predict the growth of multiple cracks by taking
into account interaction between the crack tips. The program's background and
implementation in this study is presented in Chapter 5.
Table 4.1 Test matrix for the double-edge notch specimens.
Specimen ID Nominal
Stress (ksi /
MPa)
Number of
Cracks (front)
Number of
Cracks (back)
Cycles to
2a ≥ 0.001 "
(25 μm)
Cycles to
2a ≥ 0.005 "
(127 μm)
Cycles to
Specimen
Failure
"Old"
Material:
6714-a11 15 / 103 0 2 0 25,002 120,523
6612-b21 16 / 110 0 1 0 20,002 76,952
6714-a12 15 / 103 2 0 30,150 65,485 174,958
6611-a12 16 / 110 10 3 2,501 12,502 75,066
"New"
Material:
7111-b11 15, 18/103,
124†
0 1 183,071 224,564 254,835
7111-b12 16 / 110 0 2 162,511 200,004 243,912
7012-a22 16 / 110 4 4 0 27,501 125,262
3 " Plate
Material:
8T3 16 / 110 1 0 120,708 128,209 245,771
8B3 16 / 110 6 1 15,001 22,502 77,798
8B2 16 / 110 0 1 15,002 22,503 86,962
† Specimen 7111-b11 was loaded at a maximum nominal stress of 15 ksi for 194,403 cycles. At that point, the maximum nominal stress
was increased to 18 ksi.
Fatigue Crack Growth Rate
Data for Al 7050-T7451 Alloy
Delta K, ksi sqrt(in.)
da/dN,in./cycle
1.00E-10
1.00E-09
1.00E-08
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1 10 100
8T3-CT4 (3 in. plate material)
8T3-CT3 (3 in. plate material)
6611-a12-CT2 (old material - more
microporosity)
7012-a21-CT4 (new material - less
microporosity)
Large Crack Paris Law
da/dN=3.9e(-10)*dK^(4.175)
Figure 4.1 Large crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from CT specimens)
The large crack Paris Law was obtained from Reference [29] (see Figure 2.5).
Crack Length vs. Number of Cycles:
Specimen 6611-a12, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 10000 20000 30000 40000 50000 60000 70000
Crack ID # 1
Crack ID # 1.1
Crack ID # 1.2
Crack ID # 2
Figure 4.2 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 6611-a12, Front Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0 10000 20000 30000 40000 50000 60000 70000
Crack ID # 1
Crack ID # 2
Crack ID # 3
Crack ID # 4.1
Crack ID # 4.2
Crack ID # 4
Crack ID # 5
Crack ID 6.1
Crack ID 6.2
Crack ID # 6
Crack Id # 7
Crack ID # 8
Figure 4.3 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "front" notch.
Crack Length vs. Number of Cycles:
Specimen 6612-b21, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 10000 20000 30000 40000 50000 60000 70000
Crack ID # 1
Figure 4.4 Surface crack length vs. number of elapsed cycles for specimen 6612-b21 (old material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 6714-a11, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 20000 40000 60000 80000 100000 120000
Crack ID #2
Crack ID # 1
Figure 4.5 Surface crack length vs. number of elapsed cycles for specimen 6714-a11 (old material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 6714-a12, Front Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
Crack ID # 1.1
Crack ID # 1.2
Crack ID # 1
Figure 4.6 Surface crack length vs. number of elapsed cycles for specimen 6714-a12 (old material), "front" notch.
Crack Length vs. Number of Cycles:
Specimen 7012-a22, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0 20000 40000 60000 80000 100000 120000
Crack ID # 1
Crack ID # 2
Crack ID # 3
Crack ID # 4
Figure 4.7 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 7012-a22, Front Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 20000 40000 60000 80000 100000 120000
Crack ID # 1
Crack ID # 2.1
Crack ID # 2.2
Crack ID # 2
Crack ID # 3
Figure 4.8 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "front" notch.
Crack Length vs. Number of Cycles:
Specimen 7111-b11, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 50000 100000 150000 200000 250000
Crack ID # 1
Figure 4.9 Surface crack length vs. number of elapsed cycles for specimen 7111-b11 (new material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 7111-b12, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 50000 100000 150000 200000 250000
Crack ID # 1.1
Crack ID # 1.2
Crack ID # 1
Figure 4.10 Surface crack length vs. number of elapsed cycles for specimen 7111-b12 (new material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 8B2, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 10000 20000 30000 40000 50000 60000 70000 80000
Crack ID # 1
Figure 4.11 Surface crack length vs. number of elapsed cycles for specimen 8B2 (3-inch plate material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 8B3, Back Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0 10000 20000 30000 40000 50000 60000
Crack ID # 1
Figure 4.12 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "back" notch.
Crack Length vs. Number of Cycles:
Specimen 8B3, Front Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 10000 20000 30000 40000 50000 60000
Crack ID # 1
Crack ID # 1.1
Crack ID # 1.2
Crack ID # 1.21
Crack ID # 1.22
Crack ID # 2
Crack ID # 3
Crack ID # 4
Crack ID # 5
Figure 4.13 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "front" notch.
Crack Length vs. Number of Cycles:
Specimen 8T3, Front Notch
Number of Cycles, N
CrackLength,a(inches)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 50000 100000 150000 200000 250000
Crack ID # 1
Figure 4.14 Surface crack length vs. number of elapsed cycles for specimen 8T3 (3-inch plate material), "front" notch.
Empirical Expression for
Crack Shape vs. Nondimensional Length
a/t
c/a
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Empirical Expression for a/t
vs. c/a
Measured Values of a/t vs. c/a
Figure 4.30 Empirical expression for crack shape vs. nondimensional length [28]. Measured values are from Crack ID #'s 5 and
6 from specimen 6611-a12, "front" notch. Definitions for c, a, and t are illustrated in Figure 4.31.
Figure 4.31 Illustration of corner crack shape based on empirical expression for c/a vs. a/t (to scale). Note: for the corner
crack, t≡B.
Small Fatigue Crack Growth Rate
Data for Al 7050-T7451 Alloy
Delta K, ksi sqrt(in.)
da/dN,in./cycle
1E-10
1E-09
1E-08
1E-07
1E-06
1E-05
0.0001
0.001
1 10 100
Large Crack Paris Law
Small Crack Paris Law
6714-a11, back notch
6612-b21, back notch
7111-b11, back notch
7111-b12, back notch
8B2, back notch
8B3, back notch
8T3, front notch
da/dN=3.9e(-10)*dK^(4.175)
da/dN=8.22e(-9)*(dK)^2.807
Figure 4.32 Small crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from double-edge notch
specimens).
Figure 4.25 SEM fractograph of specimen 6611-a12, "front" notch fracture surface.
The larger crack on the left was identified as Crack ID # 6 during the
replica measurement process. The smaller crack on the right was
identified as Crack ID # 5 during the replica measurement process. The
reference line on the fractograph is 1000 μm in length.
Figure 4.26 Initiation site of Crack ID # 6. The reference line on the fractograph is
100 μm in length.
Figure 4.27 Close-up of initiation site of Crack ID # 6. The reference line on the
fractograph is 10 μm in length.
Figure 4.28 Initiation site of Crack ID # 5. The reference line on the fractograph is
100 μm in length.
Figure 4.29 Close-up of initiation site of Crack ID # 5. The reference line on the
fractograph is 10 μm in length.
Figure 4.15 Replica photograph of specimen 6612-b21, "back" notch, 0 cycles (after
specimen alignment loading). Crack ID # 1: 2a = 0.0016 in.
Figure 4.16 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles.
Crack ID # 1: 2a = 0.0038 in.
Figure 4.17 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles.
Crack ID # 1: 2a = 0.0038 in.
Figure 4.18 Replica photograph of specimen 6612-b21, "back" notch, 23001 cycles.
Crack ID # 1: 2a = 0.0092 in.
Figure 4.19 Replica photograph of specimen 6612-b21, "back" notch, 33506 cycles.
Crack ID # 1: 2a = 0.0179 in.
Figure 4.20 Replica photograph of specimen 6612-b21, "back" notch, 42509 cycles.
Crack ID # 1: 2a = 0.0337 in.
Figure 4.21 Replica photograph of specimen 6611-a12, "front" notch, 42507 cycles.
Crack ID # 4.1: 2a = 0.0208 in. Crack ID # 4.2: 2a = 0.0025 in.
Crack ID # 8: 2a = 0.0009 in.
Figure 4.22 Replica photograph of specimen 6611-a12, "front" notch, 50007 cycles.
Crack ID # 4.1: 2a = 0.0275 in. Crack ID # 4.2: 2a = 0.0036 in.
Crack ID # 8: 2a = 0.0032 in.
Figure 4.23 Replica photograph of specimen 6611-a12, "front" notch, 60008 cycles.
Crack ID # 4.1: 2a = 0.0379 in. Crack ID # 4.2: 2a = 0.0047 in.
Crack ID # 8: 2a = 0.0032 in.
Figure 4.24 Replica photograph of specimen 6611-a12, "front" notch, 65008 cycles.
Crack ID # 4: 2a = 0.0441 in. Crack ID # 8: 2a = 0.0032 in.
69
CHAPTER 5 - ANALYTICAL MODELING
A computer program was employed to predict both crack shape and fatigue crack
growth in the double-edge notch specimens tested in this study. A brief history of the
program and its various implementations is presented in this chapter. This is followed by
modifications made to the program to fit this study as well as a description of how the
algorithm works.
5.1. Background
The multi-degree of freedom algorithm used in this study was originally coded by
Tritsch [34] to predict the fatigue life and crack growth shapes for both single and double
cracks located along the bore of a hole loaded under remote tension; see Figures 5.1 and
5.2. The program utilized the Newman-Raju stress intensity factor solutions [36] for a
single surface or corner crack in a hole. These K solutions could subsequently be
modified with correction factors to handle various geometries as well as crack interaction
effects.
Tritsch's original algorithm utilized Bowie's two-dimensional stress intensity
factor solution for a through-cracked hole [37] to develop correction factors so that the
computer program could be used for various specimen geometries. For this study,
however, Newman's three-dimensional K solutions for corner and surface cracks in a
70
semicircular edge notch [4] were available, thus eliminating the need to use correction
factors based on two-dimensional geomtries. For the cases of two cracks along the bore
of a hole (Figure 5.2), Tritsch employed a crack interaction factor developed by Heath
and Grandt [38]. They used the Finite Element-Alternating Method (FEAM) to obtain
stress intensity factor solutions for both a single corner crack along the bore of a hole and
symmetric corner cracks on the same side of the hole. These solutions were calculated
for crack shapes of a/c = 1.11, 1.5, 2.0, and 3.0. The interaction factor, γ, is then given by
γ =
K
K
symmetric c c
gle c c
L
L
. .
sin . .
(5.1)
Ksymmetric cc is the stress intensity factor for two corner cracks along the bore of a hole
symmetric with respect to the plane at the half-thickness of the specimen, while Ksingle cc
is the stress intensity factor for the single corner crack along the bore of a hole. The
interaction factor is a function of the crack shape, a/c, and the non dimensional separation
distance between the two symmetric corner cracks, ts/a. Polynomial expressions were
subsequently fit to the FEAM results, and incorporated into the program. It is important
to note that the interaction factor was employed only on the tips of the two cracks
adjacent to one another; the crack tips next to the free surfaces were not modified. In
addition, although γ was based on the interaction between two symmetric corner cracks,
it was employed to handle various unsymmetric combinations of corner and surface
cracks.
Scheumann [39] updated the original code to take into account interaction effects
between unsymmetric cracks on opposite sides of a hole in a plate. In addition, Grandt,
Hinkle, Scheumann, and Todd [29] developed an interaction factor based on Trantina and
71
Barishpolsky's [40] effective stress intensity factor for an ellipsoidal void in a large body
with an equatorial crack. This interaction factor was employed to predict the growth of
cracks initiating from particles and micropores in Al 7050-T7451 open hole fatigue
specimens [29]. Although predictions based on the Trantina-Barishpolsky interaction
factor compared favorably with actual specimen lives in [29], the interaction factor was
not used in this study. Instead, the small crack growth rate equation given in Equation
4.4 will be utilized in a modified version of Tritsch's original program [35] to back-
predict fatigue crack growth to breakthrough and crack shapes in the double-edge notch
fatigue specimens tested in this study.
5.2. Description of Algorithm
The goal of a multi-degree of freedom algorithm is to predict both crack size and
shape as a function of the applied load. As mentioned previously, Tritsch's original
algorithm [35] was modified to utilize Newman's three-dimensional K solutions for
corner and surface cracks in a semicircular edge notch [4]. Figure 5.1 illustrates
geometry variable definitions utilized in the prediction code for the case of a single
surface crack or corner crack. The initial dimensions of the crack are defined by the
input coordinates, i. e.,
a a for surface cracks x x
c y
1 1 3 1
1 2
2( ) = −
=
(5.2), (5.3)
The crack depth, c1, was selected to grow an increment Δc1 = 0.0001×c1. The number
of cycles required for c1 to grow an increment Δc1 is
Δ ΔN c dy dN= 1 2( ) (5.4).
72
where dy2/dN, the crack growth rate in the c direction, is given by Equations 4.4 and 4.2
for small and large cracks, respectively. The stress intensity factor solutions for a surface
and corner crack in an edge notch [4] are used to calculate the crack growth rates. The
subsequent growth of the surface crack tips, xi, are then
Δ Δx dx dN Ni i= ×( ) (5.5).
Similarly, dxi/dN is given by Equations 4.4 and 4.2 for small and large cracks,
respectively. The new crack coordinates defining the surface crack length and depth are
then recalculated, and the process is repeated. This algorithm is easily extended to
handle two or more cracks; a typical multi-crack configuration is shown in Figure 5.2.
After each iteration, the program checks for crack tip free surface contact, or in
the case of the double crack configuration, crack coalescence. If a change in the crack
type occurs, e. g., a surface crack becomes a corner crack, the change is noted and all
subsequent calculations are based on the new crack type. Once the surface or corner
crack has broken through, iteration proceeds as a through crack with an initial crack
length equal to the last crack depth calculation, c1. After breakthrough, the program
utilizes the stress intensity factor solution presented by Newman [4] for a through crack
located at a semicircular edge notch; it is given in the following form
K S cF c w c r r wn= π ( / , / , / ) (5.6).
73
The full stress intensity factor solution is given in Appendix A. Iteration continues until
either the fracture toughness of the material is reached, or a specified maximum crack
growth rate is exceeded.
The prediction code was adjusted to include a correction factor for finite notch
width. The finite width correction factor was incorporated because the crack did not
always initiate at the half-thickness of the specimen, i. e., one of the crack tips was closer
to its respective specimen side than the other crack tip. However, the ΔK solution [4]
assumes that a surface crack is located at the center of the notch. The correction factor is
the ratio of stress intensity factors for a centrally located through crack, K2D center,and an
eccentrically located through crack in a sheet of finite width, K2D eccentric, [41], i. e.,
γ fnw
D eccentric
D center
K
K
= 2
2
L
L
..
.
(5.7)
The results of the back-prediction of fatigue crack shape and growth in the double edge-
notch specimen are discussed in Chapter 6.
74
Figure 5.1 Geometry variable definitions used in the prediction program for a surface
crack and a corner crack.
75
Figure 5.2 Geometry variable definitions used in the prediction program for a typical
multiple crack configuration.
76
CHAPTER 6 - NUMERICAL RESULTS
6.1 Back-Prediction in Specimens Used to Calculate Small Crack da/dN-ΔK curve
To verify the validity of the life prediction code, crack growth predictions were
first performed for the specimens that were used to generate the small-crack da/dN-ΔK
data in Chapter 4. Although these tests provided the fatigue crack growth data for the
predictive model (Equation 4.4), these calculations are independent of the input data file
in the sense that crack shape is predicted by the two-degree-of-freedom analysis
described in Chapter 5. Recall that only the crack surface dimension "2a" was measured
from the replicas, so that when computing the cyclic stress intensity factor ΔK to
generate Equation 4.4, it was necessary to assume a corresponding crack depth "c" (i.e.,
Equation 4.3, which assumes that crack shape c/a is a function of crack size a/t). Thus,
applying the crack growth analysis program to these tests provides an independent
prediction for the crack shape c/a, which can then be compared with the assumed shape
(Equation 4.3) employed to establish the da/dN relationship.
Plots are given for both the surface crack length, a, vs. number of elapsed cycles,
N, and crack aspect ratio, c/a, vs. the non dimensional surface crack length, a/t.
Definitions of these crack parameters are illustrated in Figure 2.2.
In the a vs. N plots of Figures 6.1-6.7, there is little difference between the
predicted results of the original algorithm and the predicted results with the finite width
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis
MS_Aero_Thesis

More Related Content

Viewers also liked

tata tertib dan pint siswa
tata tertib dan pint siswatata tertib dan pint siswa
tata tertib dan pint siswaHendra Wahyu
 
Pasos para lavarte las manos
Pasos para lavarte las manosPasos para lavarte las manos
Pasos para lavarte las manosSecundaria135
 
Bases sobre a teoria da cor aplicada aos sistemas digitais
Bases sobre a teoria da cor aplicada aos sistemas digitaisBases sobre a teoria da cor aplicada aos sistemas digitais
Bases sobre a teoria da cor aplicada aos sistemas digitaisgonsaloejoao
 
Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...
 Blog 13 and 14 detailed analysis of radio news bulletins and what makes some... Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...
Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...mattsheard1
 
Task 3 ownership of movie indusrty.
Task 3   ownership of movie indusrty.Task 3   ownership of movie indusrty.
Task 3 ownership of movie indusrty.mattsheard1
 
Origenes de las redes [Wanda Juarez, Matias Speratti]
Origenes de las redes [Wanda Juarez, Matias Speratti]Origenes de las redes [Wanda Juarez, Matias Speratti]
Origenes de las redes [Wanda Juarez, Matias Speratti]Matias Speratti
 

Viewers also liked (12)

tata tertib dan pint siswa
tata tertib dan pint siswatata tertib dan pint siswa
tata tertib dan pint siswa
 
Bullying
BullyingBullying
Bullying
 
Evaulating the data. (5)
Evaulating the data. (5)Evaulating the data. (5)
Evaulating the data. (5)
 
Pasos para lavarte las manos
Pasos para lavarte las manosPasos para lavarte las manos
Pasos para lavarte las manos
 
La vida steve jobs
La vida steve jobsLa vida steve jobs
La vida steve jobs
 
Silva telles
Silva tellesSilva telles
Silva telles
 
1 f
1 f1 f
1 f
 
Bases sobre a teoria da cor aplicada aos sistemas digitais
Bases sobre a teoria da cor aplicada aos sistemas digitaisBases sobre a teoria da cor aplicada aos sistemas digitais
Bases sobre a teoria da cor aplicada aos sistemas digitais
 
Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...
 Blog 13 and 14 detailed analysis of radio news bulletins and what makes some... Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...
Blog 13 and 14 detailed analysis of radio news bulletins and what makes some...
 
Task 3 ownership of movie indusrty.
Task 3   ownership of movie indusrty.Task 3   ownership of movie indusrty.
Task 3 ownership of movie indusrty.
 
Zain Ul Abideen_281116
Zain Ul Abideen_281116Zain Ul Abideen_281116
Zain Ul Abideen_281116
 
Origenes de las redes [Wanda Juarez, Matias Speratti]
Origenes de las redes [Wanda Juarez, Matias Speratti]Origenes de las redes [Wanda Juarez, Matias Speratti]
Origenes de las redes [Wanda Juarez, Matias Speratti]
 

Similar to MS_Aero_Thesis

Pressure Distribution on an Airfoil
Pressure Distribution on an Airfoil Pressure Distribution on an Airfoil
Pressure Distribution on an Airfoil Saif al-din ali
 
Fulltext01
Fulltext01Fulltext01
Fulltext01Ngoc Ngk
 
Gene_Merewether_thesis_small
Gene_Merewether_thesis_smallGene_Merewether_thesis_small
Gene_Merewether_thesis_smallGene Merewether
 
Numerical Study of Strong Free Surface Flow and Breaking Waves
Numerical Study of Strong Free Surface Flow and Breaking WavesNumerical Study of Strong Free Surface Flow and Breaking Waves
Numerical Study of Strong Free Surface Flow and Breaking WavesYi Liu
 
Durlav Mudbhari - MSME Thesis
Durlav Mudbhari - MSME ThesisDurlav Mudbhari - MSME Thesis
Durlav Mudbhari - MSME ThesisDurlav Mudbhari
 
grDirkEkelschotFINAL__2_
grDirkEkelschotFINAL__2_grDirkEkelschotFINAL__2_
grDirkEkelschotFINAL__2_Dirk Ekelschot
 
MS Tomlinson Thesis 2004-s
MS Tomlinson Thesis 2004-sMS Tomlinson Thesis 2004-s
MS Tomlinson Thesis 2004-sMSTomlinson
 
Tesi 2017-07-29 t05-23_29.839z
Tesi 2017-07-29 t05-23_29.839zTesi 2017-07-29 t05-23_29.839z
Tesi 2017-07-29 t05-23_29.839zFahadAnees2
 
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...Gregory Waligorski
 
Innovative Payloads for Small Unmanned Aerial System-Based Person
Innovative Payloads for Small Unmanned Aerial System-Based PersonInnovative Payloads for Small Unmanned Aerial System-Based Person
Innovative Payloads for Small Unmanned Aerial System-Based PersonAustin Jensen
 
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_Ions
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_IonsJinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_Ions
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_IonsJeff Jinhua Xi
 
Dissertation
DissertationDissertation
DissertationAndi Wu
 
Freek d. van der meer, carranza, john multi- and hyperspectral geologic rem...
Freek d. van der meer, carranza, john   multi- and hyperspectral geologic rem...Freek d. van der meer, carranza, john   multi- and hyperspectral geologic rem...
Freek d. van der meer, carranza, john multi- and hyperspectral geologic rem...Martha Condori Quispe
 
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...MianHusnainIqbal2
 
Fast Skeletonization of Blood Vessels
Fast Skeletonization of Blood VesselsFast Skeletonization of Blood Vessels
Fast Skeletonization of Blood VesselsAaron Croasmun
 
How Much Does Power Take-off Affect the Dynamic Response of a WEC Device
How Much Does Power Take-off Affect the Dynamic Response of a WEC DeviceHow Much Does Power Take-off Affect the Dynamic Response of a WEC Device
How Much Does Power Take-off Affect the Dynamic Response of a WEC DeviceSam Brown
 

Similar to MS_Aero_Thesis (20)

Pressure Distribution on an Airfoil
Pressure Distribution on an Airfoil Pressure Distribution on an Airfoil
Pressure Distribution on an Airfoil
 
Fulltext01
Fulltext01Fulltext01
Fulltext01
 
Gene_Merewether_thesis_small
Gene_Merewether_thesis_smallGene_Merewether_thesis_small
Gene_Merewether_thesis_small
 
Numerical Study of Strong Free Surface Flow and Breaking Waves
Numerical Study of Strong Free Surface Flow and Breaking WavesNumerical Study of Strong Free Surface Flow and Breaking Waves
Numerical Study of Strong Free Surface Flow and Breaking Waves
 
Durlav Mudbhari - MSME Thesis
Durlav Mudbhari - MSME ThesisDurlav Mudbhari - MSME Thesis
Durlav Mudbhari - MSME Thesis
 
grDirkEkelschotFINAL__2_
grDirkEkelschotFINAL__2_grDirkEkelschotFINAL__2_
grDirkEkelschotFINAL__2_
 
ramaswamy_v
ramaswamy_vramaswamy_v
ramaswamy_v
 
MS Tomlinson Thesis 2004-s
MS Tomlinson Thesis 2004-sMS Tomlinson Thesis 2004-s
MS Tomlinson Thesis 2004-s
 
Tesi 2017-07-29 t05-23_29.839z
Tesi 2017-07-29 t05-23_29.839zTesi 2017-07-29 t05-23_29.839z
Tesi 2017-07-29 t05-23_29.839z
 
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...
Angular_distributions_of_electrons_ejected_from_autoionizing_states_of_barium...
 
Innovative Payloads for Small Unmanned Aerial System-Based Person
Innovative Payloads for Small Unmanned Aerial System-Based PersonInnovative Payloads for Small Unmanned Aerial System-Based Person
Innovative Payloads for Small Unmanned Aerial System-Based Person
 
Get tr doc
Get tr docGet tr doc
Get tr doc
 
Report_FAT1_Final
Report_FAT1_FinalReport_FAT1_Final
Report_FAT1_Final
 
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_Ions
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_IonsJinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_Ions
Jinhua_Xi_PhD_Thesis_Phototetachment_of_Negative_Atomic_Ions
 
Dissertation
DissertationDissertation
Dissertation
 
Freek d. van der meer, carranza, john multi- and hyperspectral geologic rem...
Freek d. van der meer, carranza, john   multi- and hyperspectral geologic rem...Freek d. van der meer, carranza, john   multi- and hyperspectral geologic rem...
Freek d. van der meer, carranza, john multi- and hyperspectral geologic rem...
 
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...
STRETCHABLE STRAIN SENSORS BASED ON POLYPYRROLE AND THERMOPLASTIC POLYURETHAN...
 
Fast Skeletonization of Blood Vessels
Fast Skeletonization of Blood VesselsFast Skeletonization of Blood Vessels
Fast Skeletonization of Blood Vessels
 
thesis_sillero
thesis_sillerothesis_sillero
thesis_sillero
 
How Much Does Power Take-off Affect the Dynamic Response of a WEC Device
How Much Does Power Take-off Affect the Dynamic Response of a WEC DeviceHow Much Does Power Take-off Affect the Dynamic Response of a WEC Device
How Much Does Power Take-off Affect the Dynamic Response of a WEC Device
 

MS_Aero_Thesis

  • 1. INITIATION, GROWTH, AND COALESCENCE OF SMALL FATIGUE CRACKS AT NOTCHES A Thesis Submitted to the Faculty of Purdue University by Eric Nielsen Forsyth In Partial Fulfillment of the Requirements for the Degree of Master of Science in Aeronautics and Astronautics May 1993
  • 2. ii Dedicated to my parents, George and Ardith, and my grandparents, Arthur and LaVerne Nielsen, for their endless love and support.
  • 3. iii ACKNOWLEDGMENTS This work was sponsored by the Aluminum Company of America under Project Number TC919597TC. Special thanks are extended to Dr. A. J. Hinkle and Dr. B. J. Shaw of Alcoa for their supervision and assistance throughout this study. In addition, the author would like to express appreciation to the School of Materials Science at Purdue University for the use of their specimen preparation and optical microscopy facilities. The author would especially like to thank his major Professor, A. F. Grandt, Jr., for his guidance throughout this work. Professor Grandt's experience and insight were invaluable in shaping the author's perceptions and approach to research in addition to the course of the research itself. Thanks are also extended to Professor B. M Hillberry and Professor H. D. Espinosa for providing their unique perspectives as members of the author's thesis committee. There are many other people whose support and assistance were instrumental in the completion of this work. Thanks are due to Mark Yost, Bob Sanders, and the late Gene Harston for technical assistance ranging from specimen fabrication to testing equipment maintenance. Special thanks is extended to Chad Zezula for his significant assistance with specimen testing and replica measurement. Finally, thanks are due to Mark Doerfler and Marcus Heinimann for their advice and encouragement, as well as Michelle Wade for her support.
  • 4. iv TABLE OF CONTENTS Page LIST OF TABLES............................................................................................................. vi LIST OF FIGURES ......................................................................................................... viii ABSTRACT..................................................................................................................... xix CHAPTER 1 - INTRODUCTION.......................................................................................1 CHAPTER 2 - BACKGROUND.........................................................................................2 2.1 LEFM Concepts................................................................................................2 2.2 The Small Crack Problem.................................................................................4 2.3 The 7050-T7451 Aluminum Alloy...................................................................9 CHAPTER 3 - EXPERIMENTAL PROCEDURES .........................................................20 3.1 Small Crack Specimen Design and Testing Procedures.................................20 3.1.1 Specimen Testing.............................................................................22 3.1.2 The Replication Method ..................................................................23 3.2 Large Crack Testing Procedures.....................................................................25 CHAPTER 4 - EXPERIMENTAL RESULTS..................................................................35 4.1 Large Crack Growth Rate Data ......................................................................35 4.2 Small Crack Test Results................................................................................36 4.3 Small Crack Growth Rate Data.......................................................................39 CHAPTER 5 - ANALYTICAL MODELING...................................................................69 5.1 Background.....................................................................................................69 5.2 Description of Algorithm................................................................................71
  • 5. v Page CHAPTER 6 - NUMERICAL RESULTS.........................................................................76 6.1 Back-Prediction in Specimens Used to Calculate Small Crack da/dN-ΔK curve ...............................................................................................76 6.2 Prediction Results in Specimens Initiating Multiple Cracks ...........................79 CHAPTER 7 - CONCLUSIONS AND RECOMMENDATIONS..................................115 LIST OF REFERENCES.................................................................................................117 APPENDICES Appendix A - Stress Intensity Factor Solutions...................................................122 Appendix B - Specimen Dimensions and Test Parameters .................................128 Appendix C - Crack Measurements for Double-Edge Notch Specimens............130
  • 6. vi LIST OF TABLES Table Page 3.1 Parameters for fatigue test specimens........................................................27 4.1 Test matrix for the double-edge notch specimens. ....................................43 Appendix Table B1 Dimensions and test parameters for the double-edge notch specimens. All tests were conducted at a stress ratio R = 0.1 and in laboratory air...........................................................................129 C1 Crack measurements for specimen 6611-a12, back notch.......................133 C2 Crack measurements for specimen 6611-a12, front notch.......................135 C3 Crack measurements for specimen 6612-b21, back notch.......................140 C4 Crack measurements for specimen 6714-a11, back notch.......................141 C5 Crack measurements for specimen 6714-a12, front notch.......................142 C6 Crack measurements for specimen 7012-a22, back notch.......................144 C7 Crack measurements for specimen 7012-a22, front notch.......................147 C8 Crack measurements for specimen 7111-b11, back notch.......................150 C9 Crack measurements for specimen 7111-b12, back notch.......................151 C10 Crack measurements for specimen 8B2, back notch ...............................152
  • 7. vii Appendix Table Page C11 Crack measurements for specimen 8B3, back notch ...............................153 C12 Crack measurements for specimen 8B3, front notch...............................154 C13 Crack measurements for specimen 8T3, front notch ...............................158
  • 8. viii LIST OF FIGURES Figure Page 2.1 The two stages of fatigue...........................................................................14 2.2 The semicircular edge notch geometry and variable definitions ..................................................................................................15 2.3 Typical fatigue crack growth rate data for large and small cracks [11]..................................................................................................16 2.4 Stress intensity factor vs. time for constant ampltide loading. Lower crack opening stress for small cracks results in a larger effective stress intensity factor than a large crack under identical loading, translating into faster growth rates................................................................................................17 2.5 Cumulative fatigue failure distributions from 1984-1987 for the 7050-T7451 thick plate (5.5-5.9 inches = 140-150 mm)............................................................................................................18 2.6 Large crack da/dN-ΔK data for the Al 7050-T7451 alloy, R=0.1 [29]..................................................................................................19 3.1 Double-edge notch dogbone specimen geometry and dimensions. ................................................................................................28 3.2 Double-edge notch specimen geometry and dimensions...........................29 3.3 Hoop Stress/Remote Stress, σhoop/σrem, vs. Theta, Θ, for the 1.11 and 2.00 inch wide specimen geometries. σhoop/σrem was caculated with a 2-dimensional finite element analysis of each of the specimen geometries. Θ is defined as the angle (in radians) from the tip of the notch to the upper/lower location where the notch meets the specimen edge......................................................................................30
  • 9. ix Figure Page 3.4 Illustration of the replication process ........................................................31 3.5 Definition of replica coordinate system.....................................................32 3.6 CT specimens fabricated from fractured double-edge notch specimen...........................................................................................33 3.7 CT specimen geometry and dimensions. ...................................................34 4.1 Large crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from CT specimens) The large crack Paris Law was obtained from Reference [29] (see Figure 2.5)...................................................................................44 4.2 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "back" notch.......................................45 4.3 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "front" notch. .....................................46 4.4 Surface crack length vs. number of elapsed cycles for specimen 6612-b21 (old material), "back" notch. .....................................47 4.5 Surface crack length vs. number of elapsed cycles for specimen 6714-a11 (old material), "back" notch.......................................48 4.6 Surface crack length vs. number of elapsed cycles for specimen 6714-a12 (old material), "front" notch. .....................................49 4.7 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "back" notch.....................................50 4.8 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "front" notch.....................................51 4.9 Surface crack length vs. number of elapsed cycles for specimen 7111-b11 (new material), "back" notch.....................................52 4.10 Surface crack length vs. number of elapsed cycles for specimen 7111-b12 (new material), "back" notch.....................................53
  • 10. x Figure Page 4.11 Surface crack length vs. number of elapsed cycles for specimen 8B2 (3-inch plate material), "back" notch. ................................54 4.12 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "back" notch. ................................55 4.13 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "front" notch.................................56 4.14 Surface crack length vs. number of elapsed cycles for specimen 8T3 (3-inch plate material), "front" notch. ................................57 4.15 Replica photograph of specimen 6612-b21, "back" notch, 0 cycles (after specimen alignment loading). Crack ID # 1: 2a = 0.0016 in. ...........................................................................................58 4.16 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles. Crack ID # 1: 2a = 0.0038 in. .............................................58 4.17 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles. Crack ID # 1: 2a = 0.0038 in. .............................................59 4.18 Replica photograph of specimen 6612-b21, "back" notch, 23001 cycles. Crack ID # 1: 2a = 0.0092 in. ............................................59 4.19 Replica photograph of specimen 6612-b21, "back" notch, 33506 cycles. Crack ID # 1: 2a = 0.0179 in. ............................................60 4.20 Replica photograph of specimen 6612-b21, "back" notch, 42509 cycles. Crack ID # 1: 2a = 0.0337 in. ............................................60 4.21 Replica photograph of specimen 6611-a12, "front" notch, 42507 cycles. Crack ID # 4.1: 2a = 0.0208 in. Crack ID # 4.2: 2a = 0.0025 in. Crack ID # 8: 2a = 0.0009 in. ...................................61 4.22 Replica photograph of specimen 6611-a12, "front" notch, 50007 cycles. Crack ID # 4.1: 2a = 0.0275 in. Crack ID # 4.2: 2a = 0.0036 in. Crack ID # 8: 2a = 0.0032 in. ...................................61
  • 11. xi Figure Page 4.23 Replica photograph of specimen 6611-a12, "front" notch, 60008 cycles. Crack ID # 4.1: 2a = 0.0379 in. Crack ID # 4.2: 2a = 0.0047 in. Crack ID # 8: 2a = 0.0032 in. ...................................62 4.24 Replica photograph of specimen 6611-a12, "front" notch, 65008 cycles. Crack ID # 4: 2a = 0.0441 in. Crack ID # 8: 2a = 0.0032 in. .......................................................................................62 4.25 SEM fractograph of specimen 6611-a12, "front" notch fracture surface. The larger crack on the left was identified as Crack ID # 6 during the replica measurement process. The smaller crack on the right was identified as Crack ID # 5 during the replica measurement process. The reference line on the fractograph is 1000 μm in length. ........................................................................................................63 4.26 Initiation site of Crack ID # 6. The reference line on the fractograph is 100 μm in length.................................................................64 4.27 Close-up of initiation site of Crack ID # 6. The reference line on the fractograph is 10 μm in length.................................................64 4.28 Initiation site of Crack ID # 5. The reference line on the fractograph is 100 μm in length.................................................................65 4.29 Close-up of initiation site of Crack ID # 5. The reference line on the fractograph is 10 μm in length.................................................65 4.30 Empirical expression for crack shape vs. nondimensional length [28]. Measured values are from Crack ID #'s 5 and 6 from specimen 6611-a12, "front" notch..................................................66 4.31 Illustration of corner crack shape based on empirical expression for c/a vs. a/t (to scale).............................................................67 4.32 Small crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from double-edge notch specimens)..................................................................................................68 5.1 Geometry variable definitions used in the prediction program for a surface crack and a corner crack.........................................74
  • 12. xii Figure Page 5.2 Geometry variable definitions used in the prediction program for a typical multiple crack configuration...................................75 6.1 Actual and predicted crack growth for specimen 6612-b21, back notch: surface crack length vs. number of cycles.............................84 6.2 Actual and predicted crack growth for specimen 6714-a11, back notch: surface crack length vs. number of cycles.............................85 6.3 Actual and predicted crack growth for specimen 7111-b11, back notch: surface crack length vs. number of cycles.............................86 6.4 Actual and predicted crack growth for specimen 7111-b12, back notch: surface crack length vs. number of cycles.............................87 6.5 Actual and predicted crack growth for specimen 8B2, back notch: surface crack length vs. number of cycles.. ...................................88 6.6 Actual and predicted crack growth for specimen 8B3, back notch: surface crack length vs. number of cycles. ....................................89 6.7 Actual and predicted crack growth for specimen 8T3, front notch: surface crack length vs. number of cycles. ....................................90 6.8 Predicted crack growth for specimen 6612-b21, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................91 6.9 Predicted crack growth for specimen 6714-a11, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................92 6.10 Predicted crack growth for specimen 7111-b11, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................93 6.11 Predicted crack growth for specimen 7111-b12, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................94
  • 13. xiii Figure Page 6.12 Predicted crack growth for specimen 8B2, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................................95 6.13 Predicted crack growth for specimen 8B3, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................................96 6.14 Predicted crack growth for specimen 8T3, front notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.......................................97 6.15 Stress intensity factor geometry for two offset parallel cracks in a sheet under uniform uniaxial tensile stress [42]. .....................98 6.16 Actual and predicted crack growth for specimen 7012-a22, back notch: surface crack length vs. number of cycles. Note: no crack interaction is considered between the cracks. ........................................................................................................99 6.17 Predicted crack growth for specimen 7012-a22, front notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: no crack interaction is considered between the cracks. ......................................................................................................100 6.18 Actual and predicted crack growth for specimen 7012-a22, back notch: surface crack length vs. number of cycles. Note: no crack interaction is considered between the cracks. The stress concentration factors were adjusted to account for the crack initiating off the midplane of the notch at an angle Θ (see Figure 3.3)........................................................101 6.19 Predicted crack growth for specimen 7012-a22, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: no crack interaction is considered between the cracks. The stress concentration factors were adjusted to account for the crack initiating off the midplane of the notch at an angle Θ (see Figure 3.3)........................................................102
  • 14. xiv Figure Page 6.20 Actual and predicted crack growth for specimen 7012-a22, front notch: surface crack length vs. number of cycles. Note: the presence of Crack ID #'s 2 and 3 are ignored.. ........................103 6.21 Predicted crack growth for specimen 7012-a22, front notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: the presence of Crack ID #'s 2 and 3 are ignored.. ........................104 6.22 Actual and predicted crack growth for specimen 6611-a12, back notch: surface crack length vs. number of cycles. Note: the presence of Crack ID #'s 1.1 and 2 are ignored. ......................105 6.23 Predicted crack growth for specimen 6611-a12, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: the presence of Crack ID #'s 1.1 and 2 are ignored. ......................106 6.24 Actual and predicted crack growth for specimen 6611-a12, back notch: surface crack length vs. number of cycles. Note: the presence of Crack ID # 2 is ignored.........................................107 6.25 Predicted crack growth for specimen 6611-a12, back notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: the presence of Crack ID # 2 is ignored.........................................108 6.26 Actual and predicted crack growth for specimen 6714-a12, front notch: surface crack length vs. number of cycles. ........................109 6.27 Predicted crack growth for specimen 6714-a12, front notch: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions.....................110 6.28 Actual and predicted crack growth for specimen 8B3, front notch, Crack ID # 1.22: surface crack length vs. number of cycles. Note: the presence of other cracks are ignored. ....................................................................................................111
  • 15. xv Figure Page 6.29 Predicted crack growth for specimen 8B3, front notch, Crack ID # 1.22: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: the presence of other cracks are ignored. .....................112 6.30 Actual and predicted crack growth for specimen 8B3, front notch Crack ID #'s 1.21 and 1.22: surface crack length vs. number of cycles. Note: the presence of other cracks are ignored. ...................................................................................113 6.31 Predicted crack growth for specimen 8B3, front notch Crack ID #'s 1.21 and 1.22: c/a vs. a/t for both the predicted values and the empirical expression assumed in calculating ΔK solutions. Note: the presence of other cracks are ignored. ...................................................................................114 Appendix Figure C1 Crack tip locations for Specimen 6611-a12, back notch (N=30,003 cycles)....................................................................................159 C2 Crack tip locations for Specimen 6611-a12, back notch (N=40,007 cycles)....................................................................................159 C3 Crack tip locations for Specimen 6611-a12, back notch (N=50,007 cycles)....................................................................................160 C4 Crack tip locations for Specimen 6611-a12, front notch (N=45,007 cycles)....................................................................................161 C5 Crack tip locations for Specimen 6611-a12, front notch (N=65,008 cycles)....................................................................................161 C6 Crack tip locations for Specimen 6612-b21, front notch (N=29,005 cycles)....................................................................................162 C7 Crack tip locations for Specimen 6612-b21, front notch (N=37,507 cycles)....................................................................................162 C8 Crack tip locations for Specimen 6612-b21, front notch (N=47,511 cycles)....................................................................................163
  • 16. xvi Appendix Figure Page C9 Crack tip locations for Specimen 6612-b21, front notch (N=62,516 cycles)....................................................................................163 C10 Crack tip locations for Specimen 6714-a11, back notch (N=60,028 cycles)....................................................................................164 C11 Crack tip locations for Specimen 6714-a11, back notch (N=75,033 cycles)....................................................................................164 C12 Crack tip locations for Specimen 6714-a11, back notch (N=90,037 cycles)....................................................................................165 C13 Crack tip locations for Specimen 6714-a11, back notch (N=110,045 cycles)..................................................................................165 C14 Crack tip locations for Specimen 6714-a12, front notch (N=113,440 cycles)..................................................................................166 C15 Crack tip locations for Specimen 6714-a12, front notch (N=130,630 cycles)..................................................................................166 C16 Crack tip locations for Specimen 6714-a12, front notch (N=142,130 cycles)..................................................................................167 C17 Crack tip locations for Specimen 6714-a12, front notch (N=163,630 cycles)..................................................................................167 C18 Crack tip locations for Specimen 7012-a22, back notch (N=53,002 cycles)....................................................................................168 C19 Crack tip locations for Specimen 7012-a22, back notch (N=65,002 cycles)....................................................................................168 C20 Crack tip locations for Specimen 7012-a22, back notch (N=80,004 cycles)....................................................................................169 C21 Crack tip locations for Specimen 7012-a22, back notch (N=94,006 cycles)....................................................................................169 C22 Crack tip locations for Specimen 7012-a22, back notch (N=106,509 cycles)..................................................................................170
  • 17. xvii Appendix Figure Page C23 Crack tip locations for Specimen 7012-a22, front notch (N=41,001 cycles)....................................................................................171 C24 Crack tip locations for Specimen 7012-a22, front notch (N=56,002 cycles)....................................................................................171 C25 Crack tip locations for Specimen 7012-a22, front notch (N=71,003 cycles)....................................................................................172 C26 Crack tip locations for Specimen 7012-a22, front notch (N=89,005 cycles)....................................................................................172 C27 Crack tip locations for Specimen 7012-a22, front notch (N=106,509 cycles)..................................................................................173 C28 Crack tip locations for Specimen 7111-b11, back notch (N=231,068 cycles)..................................................................................174 C29 Crack tip locations for Specimen 7111-b11, back notch (N=239,072 cycles)..................................................................................174 C30 Crack tip locations for Specimen 7111-b11, back notch (N=247,075 cycles)..................................................................................175 C31 Crack tip locations for Specimen 7111-b12, back notch (N=211,506 cycles)..................................................................................176 C32 Crack tip locations for Specimen 7111-b12, back notch (N=221,002 cycles)..................................................................................176 C33 Crack tip locations for Specimen 7111-b12, back notch (N=229,003 cycles)..................................................................................177 C34 Crack tip locations for Specimen 7111-b12, back notch (N=233,003 cycles). 177 C35 Crack tip locations for Specimen 8B2, back notch (N=45,008 cycles)....................................................................................178 C36 Crack tip locations for Specimen 8B2, back notch (N=57,501 cycles)....................................................................................178
  • 18. xviii Appendix Figure Page C37 Crack tip locations for Specimen 8B2, back notch (N=72,507 cycles)....................................................................................179 C38 Crack tip locations for Specimen 8B3, back notch (N=42,506 cycles)....................................................................................180 C39 Crack tip locations for Specimen 8B3, back notch (N=57,509 cycles)....................................................................................180 C40 Crack tip locations for Specimen 8B3, front notch (N=27,503 cycles)....................................................................................181 C41 Crack tip locations for Specimen 8B3, front notch (N=40,006 cycles)....................................................................................181 C42 Crack tip locations for Specimen 8B3, front notch (N=50,007 cycles)....................................................................................182 C43 Crack tip locations for Specimen 8B3, front notch (N=58,509 cycles)....................................................................................182 C44 Crack tip locations for Specimen 8T3, front notch (N=145,002 cycles)..................................................................................183 C45 Crack tip locations for Specimen 8T3, front notch (N=162,005 cycles)..................................................................................183 C46 Crack tip locations for Specimen 8T3, front notch (N=182,008 cycles)..................................................................................184 C47 Crack tip locations for Specimen 8T3, front notch (N=214,005 cycles)..................................................................................184
  • 19. xix ABSTRACT Forsyth, Eric Nielsen. M.S.A.A., Purdue University, May 1993. Initiation, Growth, and Coalescence of Small Fatigue Cracks at Notches. Major Professor: Dr. A. F. Grandt, Jr. This research concerns the initiation, growth and coalescence of small fatigue cracks at semicircular edge notches in the aluminum 7050-T7451 plate alloy. Three versions of the alloy were provided by ALCOA, each with a varying degree of microporosity. The objective of this study was to determine if a reduction in the amount of microporosity resulted in improved small fatigue crack growth properties. Ten double-edge notch specimens were tested at varying stress levels with a stress ratio of R = 0.1. Fatigue crack growth was monitored with the replication method, providing surface crack measurements as small as 0.0006 inches (15 microns). CT specimens for all three versions of the alloy were fatigue tested to determine the large fatigue crack growth properties. Results from the CT specimen tests compared favorably with fatigue crack growth rate vs. applied stress intensity factor range data generated previously by ALCOA, and indicated that all three versions of the alloy had identical large fatigue crack growth properties. Results from the double-edge notch specimen tests indicate that after initiation, small fatigue cracks grow at faster rates than large fatigue cracks under identical ΔK loading. All three versions of the alloy demonstrated similar small fatigue
  • 20. xx crack growth rate properties after initiation. However, the versions of the alloy with reducedmicroporosity demonstrated longer fatigue lives to initiation than the version with the most microporosity. The small fatigue crack da/dN-ΔK curve was incorporated in a program to back- predict the fatigue crack growth after initiation in double-edge notch specimens that initiated a single crack. The predicted crack growth results to breakthrough showed reasonable agreement with the data obtained from the specimen tests. However, future tests should be conducted at different stress levels to generalize the results obtained in this study.
  • 21. 1 CHAPTER 1 - INTRODUCTION Fatigue cracks in engineering structures often originate at stress concentrations such as fastener holes and notched components. These cracks can initiate at initial defects such as voids and bonded inclusions within the engineering material. Research conducted by ALCOA on the aluminum 7050-T7451 alloy [1] has demonstrated that the fatigue life of edge-notch specimens can be improved by reducing the amount of microporosity within the alloy. Since the majority of a fatigue crack's life in an engineering structure can be spent in this "small" crack stage, it is of critical importance to understand how all of these factors interact with one another to effect the crack's subsequent growth. The primary objective of this study is to determine how initial microporosity effects the initiation and growth of fatigue cracks in the aluminum 7050-T7451 alloy. To accomplish this, fatigue testing was performed on semicircular edge-notch specimens fabricated from three versions of the alloy with varying levels of microporosity. Fatigue crack growth was monitored from the point of initiation, enabling crack growth rate information to be obtained for physically small cracks. Finally, an existing program was modified to predict the growth of these cracks from the point of initiation in the semicircular edge notch geometry.
  • 22. 2 CHAPTER 2 - BACKGROUND Fatigue, the failure mode associated with cyclic loading, is often separated into two stages: crack initiation and crack growth (Figure 2.1). Different methodologies have been developed to treat the life of a crack through these two stages. Stress-life and strain-life approaches are often used to quantify crack initiation life, while linear elastic fracture mechanics (LEFM) can be used to quantify the growth of a crack with an initial size, ao. The damage tolerance design philosophy lends itself particularly well to the crack growth portion of fatigue life. Using this approach, engineers assume the pre- existence of flaws in their design. Thus, it is desirable from an analysis point of view to treat fatigue as primarily crack growth, i. e., have a single analysis method applicable to all crack sizes. Unfortunately, there is no strict "boundary" where the LEFM assumption breaks down. The purpose of this research is to monitor crack growth in an aluminum alloy from the point of initiation, and apply LEFM principles into the "gray" area between initiation and growth in an effort to predict crack behavior. 2.1. LEFM Concepts LEFM assumes that crack growth is controlled by the stress intensity factor, K [2]. This term, introduced by Irwin, relates loading, crack size, and specimen geometry, and is often given in the form
  • 23. 3 K a a= σ π β( ) (2.1) where σ is the remotely applied stress, a is the crack length, and β(a) is a dimensionless function of the crack geometry. Paris, Gomez, and Anderson [3] first demonstrated that the rate of fatigue crack growth (da/dN) is a function of the applied stress intensity factor range (ΔK), independent of the particular loading, crack size, and specimen geometry, i. e., da dN f K/ ( )= Δ (2.2) This expression can be integrated to obtain the cyclic fatigue life N dN da f K N a a o f = =z z0 ( )Δ (2.3) If the da/dN-ΔK expression is known for a particular material, these equations can be incorporated into an algorithm to predict the cyclic fatigue life of a crack under different loading conditions and geometries. Although there are many K solutions available for two dimensional geometries, there are few closed form K solutions for three dimensional geometries. Since fatigue cracks initially start out having two dimensions (a surface length "a" and a crack depth "c"), three dimensional K solutions are necessary to study the growth of small cracks. Specimen geometries that incorporate a semicircular edge notch have been found to be useful to monitor the initiation and growth of small cracks. Newman [4] has presented approximate K solutions for corner and surface cracks in a semicircular edge notch.
  • 24. 4 These solutions were developed from finite element [5, 6] and weight function [7, 8] methods for surface and corner cracks; from boundary force analyses of through cracks at a semicircular notch [9]; and from previously developed equations for similar crack configurations at an open hole [10]. The solutions are given in the form K S a QF a c a t c r c w r t r wjn= π φ( / , / , / , / , / , / , ) (2.4) The semicircular edge notch geometry and variable definitions are illustrated in Figure 2.2. It is important to note that for a corner crack in Figure 2.2, t is defined in the present work as the specimen thickness B, whereas for a surface crack, t is defined as B/2. The full stress intensity factor solutions used here are given in Appendix A. More information on the actual test specimen design used in this research is given in Chapter 3. 2.2. The Small Crack Problem Research conducted over the past two decades has shown that for certain materials, physically small cracks (a ≤ 0.02 inches = 0.51 mm) grow at faster rates than large cracks under the same ΔK loading. In addition, small cracks have been observed to grow beneath the large crack threshold, ΔKth. These phenomena are known as the "small crack effect". Schematic differences between the growth rates of small and large cracks is illustrated in Figure 2.3 [11]. Since a significant portion of a crack's life in an engineering structure may be spent as a small crack, any life predictions for that component based on large crack data would be non-conservative. Thus, it is important to determine if an engineering material exhibits this difference between the growth of small and large cracks, and explain why it exists [11].
  • 25. 5 There are several factors that are believed to be involved in the small crack effect. As mentioned earlier, LEFM assumptions are invalidated as the crack size approaches zero due to the fact that the plastic zone size in front of the crack is on the same order of magnitude as the crack size itself. Nonlinear and elastic-plastic fracture mechanics concepts, such as the J-integral [12] and strain energy densities [12], have been used to explain the short crack effect. In addition, the continuum assumption of LEFM [13] is invalidated because grain boundaries as well as voids and inclusion particles affect the local stresses near the small crack front. For a large crack, these metallurgical effects are averaged out over the larger crack's long front. However, there are LEFM concepts which, in part, help explain the small crack effect. In particular, crack closure has been shown to play an important role in the accelerated growth rates of small cracks. First proposed by Elber [14], crack closure is the concept that a crack is not fully open until a "crack opening stress" is reached. This phenomenon can be attributed to several factors, including plastically deformed material in the wake of a crack, crack surface roughness, and oxide debris on the crack surface. All of these factors hinder the opening of a crack, resulting in a stress level that must be reached before the crack can be fully open and thus propagate. It is believed that small cracks have smaller crack opening stresses than large cracks do. Therefore, small cracks would experience a larger effective stress intensity factor range than large cracks, even though they are experiencing identical ΔK loading. This phenomenon is illustrated in Figure 2.4 in a K vs. time graph. In order to study the small crack effect, researchers have developed several methods for measuring small fatigue cracks. Perhaps the most accurate method for measuring small cracks is with the scanning electron microscope (SEM). When used in
  • 26. 6 conjunction with stereo imaging, SEM photographs provide useful information in the closure behavior of small cracks [15]. Although the SEM has both the spatial and strain resolution for the scale involved, cost makes its use prohibitive for routine laboratory measurements. Sharpe [16] has developed the interferometric strain-displacement gage (ISDG) which acts as a non-contacting extensometer for the specimen. Two indentations are made with a Vickers hardness tester above and below a surface crack. The diffraction patterns created by a laser impinging on the indentations can be used to determine crack opening displacement and thus crack size. Although the ISDG can be used for computer control and real-time measurement of small fatigue crack tests, the location of the initiated crack must be known before measurements can be taken. Another method which allows for computer data acquisition is the direct current electrical potential measurement (dcEPM) of small cracks [17]. If a current is passed through a specimen containing a crack, the voltage difference across the crack can be correlated to the crack length. Drawbacks to the dcEPM method include cost, the necessity for the specimen to conduct electricity, and the fact that it has only been used on cracks artificially created with electric discharge machining. Resch and Nelson [18] have developed an ultrasonic method for the measurement of small cracks. The method uses surface acoustic waves on the specimen to determine surface crack depth; in this sense, the method is similar to the SONAR employed by naval craft to determine underwater features.
  • 27. 7 A relatively simple, but more time consuming method for the measurement of small cracks is the replication method [19]. It uses an acetate tape which makes an exact replica of the specimen surface when acetone is applied to the surface. The method can be used for a variety of specimen geometries and crack length measurements as small as 0.0002 inches (5.1 μm) have been obtained. Unfortunately, only the surface crack length can be measured with this method - not the crack depth. The research presented in this thesis utilized the replication method for the measurement of small cracks. A more thorough discussion on the specifics of the replication method and how it was used in conjunction with this research is presented in Chapter 3. In 1984, an AGARD Cooperative Test Program was initiated to investigate the small crack growth behavior under various loading conditions for the aluminum alloy 2024-T3, a common material used in airframe components [20]. Twelve participants from nine different countries monitored the growth and coalescence of nearly 950 cracks in over 250 single edge notch specimens. The tests were conducted at three different stress levels for both constant amplitude loading (stress ratios, R ≡ minimum/maximum stress = -2, -1, 0, and 0.5) and spectrum loading (FALSTAFF and GAUSSIAN) conditions. Surface crack lengths were measured with the replication technique. The participants involved in the test program showed good agreement on the small crack growth rates, cyclic fatigue life to crack breakthrough (surface and/or corner cracks became a through crack), and on crack shapes. The small cracks initiated in the tests demonstrated the small crack effect mentioned previously by growing below the large crack ΔK threshold and growing at faster growth rates than large cracks above the threshold.
  • 28. 8 A fatigue crack growth model accounting for crack closure was developed by Newman [21] to predict the growth of small cracks from small voids and inclusion particles on the notch surface. The initial defect size was chosen to approximate the initiation sites of the cracks monitored in the tests. The model was based on the Dugdale strip-yield plastic zone [22], but modified for closure by leaving plastically deformed material in the wake of the crack. Lee and Sharpe's experimentally measured values for the crack opening stresses (obtained from the ISDG method) [23] showed good correlation with Newman's analytical model, increasing the confidence in the model. There was reasonable agreement between the experimental and predicted values for the small crack growth rates, although the model predicted slightly slower growth rates for R = -2 loading, and slightly faster growth rates for R = 0.5 loading. However, the model did indicate that the small crack effect was most predominant in the tests involving significant compressive loads. This behavior was observed in the tests themselves. In order to allow participants to test various materials and loading conditions that were of particular interest to their laboratories, an AGARD Supplemental Test Program on the growth of small cracks was initiated [24]. The materials tested in the supplemental program were: 2024-T3 and 7075-T6 aluminum alloys, 2090-T8E41 aluminum-lithium alloy, Ti-6Al-4V titanium alloy, and 4340 steel. The results from the supplemental program were similar to the first program in that all the materials exhibited the small crack effect to some extent. However, the effect was less pronounced in some materials (e. g., 4340 steel) than in others. Once again, the crack growth model predicted small crack growth rates in reasonable agreement with the experimental measurements for most loading conditions.
  • 29. 9 2.3. The 7050-T7451 Aluminum Alloy In an effort to reduce both the size and frequency of potential microporosity in their aluminum 7050-T7451 plate alloy, the Aluminum Company of America (ALCOA) has improved their processing techniques for the material over the past decade. Smooth axial fatigue tests of material produced in 1985 following the process improvements have resulted in longer fatigue lifetimes than material produced prior to the improvements [1]. Post-test fractography of the specimens fabricated from both materials revealed the size of the micropores that resulted in crack initiation and subsequent fracture. This microporosity size distribution was subsequently used in a probabilistic crack growth analysis, which demonstrated analytically that the reduced microporosity material should perform better in service than the older material with larger micropores. However, since smooth axial fatigue tests do not take into account cracks originating from machining defects, a test program was initiated to examine whether these type of flaws obscure the process improvements resulting in microporosity reduction. The objective of this program was to demonstrate the effect of microporosity on an engineering detail, specifically, a notched specimen subjected to constant amplitude loading [1]. The material was obtained from a single lot of 5.6 inch (14.2 cm) thick 7050-T7451 plate. Specimens fabricated from the mid-plane of the plate had a higher degree of microporosity than the specimens fabricated from the quarter-plane of the plate. The test specimens were 0.126 in. (3.2 mm) thick, 1.00 in. (25.4 mm) wide, and 9.00 in. (229 mm) long, with two holes of 0.187 in. (4.75 mm) diameter located 1.00 in. (25.4 mm) apart. The goals of this specimen were to provide a symmetric stress field and to increase the chances that a micropore would be located near a stress concentration. The
  • 30. 10 specimens were cycled to failure at a stress ratio R = 0.1 at maximum stress levels of 10, 12, and 20 ksi (69, 83, and 138 MPa). Results from these tests show a significant improvement in the fatigue properties of the low microporosity (quarter-plane) material; this can be seen in the test specimens' log-life versus log-maximum stress plot of Figure 2.5 [1]. For example, a component designed for a lifetime of 100,000 cycles could see a maximum stress of 110 MPa in the low microporosity material as opposed a maximum stress of 98 MPa in the high microporosity material; this represents an improvement of 12 percent in stress level [1]. All specimen failures in this study initiated at micropores as opposed to machining defects. The largest micropore initiating a crack in the test program was 0.030 in. (0.75 mm); the average size of a crack initiating micropore, however, was 0.012 in. (0.31 mm). Both of these sizes fall below current nondestructive inspection (NDI) capabilities, which can reliably detect flaw sizes of 0.04-0.08 in. (1-2 mm) [1]. Thus, ALCOA employed destructive techniques such as SEM examination of the fracture surfaces to quantify the microporosity distribution. This examination revealed that the frequency of micropores that initiated cracks in the specimens to be the major difference between the fracture surfaces of the two versions of the material. The high porosity (mid-plane) version of the material initiated on average a greater number of cracks (2.25 per specimen) than the low porosity (quarter-plane) version (1.33 per specimen) [1]. One of the main conclusions from the test program was that initial material quality should be considered in the design process. To accomplish this, ALCOA utilized the United States Air Force (USAF) Advanced Durability Analysis. This method is based on the concept of an equivalent initial flaw size (EIFS) which represents the initial
  • 31. 11 microporosity distribution in the material. Since all the cracks in the specimens initiated at micropores, an EIFS distribution (calculated via LEFM principles) based on these tests could theoretically be equated with the actual initial microporosity distribution of the material (determined from the earlier smooth axial fatigue tests). Subsequent analysis demonstrated this hypothesis; the two distributions were very similar, and predicted specimen lifetimes when used as input in a probabilistic fracture mechanics analysis [1]. Out of this test program arose two objectives for further research. First, it was desired to further develop and assess the benefits of the probabilistic approach to durability. ALCOA, in collaboration with Wright Laboratory's Flight Dynamics Directorate (USAF) [25, 26], has demonstrated through further testing and analysis that a reduction in the microporosity of Al 7050-T7451 can result in the increased performance and reduced cost of airframe components where durability is a major design factor. In addition, they confirmed that the USAF probabilistic failure model captured this advantage in improved material quality, whereas more conventional fatigue design practices did not. ALCOA's second objective was to further quantify crack growth from micropores by studying the effect of microporosity on the growth of physically small cracks. To accomplish this, ALCOA initiated a test program to monitor the initiation and growth of small cracks in the low and high microporosity versions of the 7050-T7451 plate [27]. The specimen design incorporated four semicircular edge notches, two on each side, of a 0.125 in × 2.00 in ×12.00 in (3.2 mm × 51 mm × 305 mm) rectangular specimen. Crack initiation and growth at the notches was monitored with the replication method. After fatigue testing of the specimens, the replicas were covered with approximately 100 - 200 Angstroms of gold so that crack measurements could be made with the SEM. Both
  • 32. 12 actual lengths and projected lengths of the cracks were obtained from the replicas with an automatic image analysis system (IBAS) [27, 28]. In addition, fractography was performed on the fractured specimen surfaces to examine the crack initiation sites. From these small crack tests, ALCOA researchers have obtained small crack length, L, versus number of cycles N, as well as dL/dN-ΔK plots. Although data analysis is still being performed, some qualitative observations could be made from the preliminary results. First, the material with the low microporosity initiated cracks later than the material with the higher microporosity. In addition, large pores in the materials appear to have the greatest influence on crack initiation and propagation. Finally, the dL/dN-ΔK plots show little difference between the low and high microporosity versions of the alloy [27]. The research presented in this thesis is an extension of ALCOA's effort to determine the effect of microporosity on the initiation and growth of small cracks in the Al 7050-T7451 alloy. Three versions of the material were supplied for this effort. The versions of the material shall be referred to in this thesis as "old", "new", and "three-inch plate" material. Both the "old" and "new" materials were obtained from a six-inch plate, and contain more microporosity than the "three-inch plate" version of the aluminum alloy. This is due to the fact that the three-inch version of the material was rolled for a longer period of time than the six-inch version, effectively "squeezing" out any remaining microporosity. Large crack da/dN-ΔK data for the Al 7050-T7451 alloy is shown in Figure 2.6 [29]. Although the large crack growth rates for all three versions of the material exhibit the same da/dN-ΔK relationship [30], it is believed that the initial microporosity in each
  • 33. 13 of the three versions will effect the small crack growth rates in different ways. It was hoped that the reduced microporosity versions of the material would delay crack initiation, and exhibit better overall fatigue properties, thus justifying its increased cost. The purpose of this research is determine whether this assumption is true by performing small fatigue crack tests on all three versions of the material. In addition, an existing crack growth prediction program was modified to analyze the growth of small fatigue cracks from semicircular edge notches based on the experimental results.
  • 34. Figure 2.1 The two stages of fatigue.
  • 35. Figure 2.2 The semicircular edge notch geometry and variable definitions.
  • 36. Figure 2.3 Typical fatigue crack growth rate data for large and small cracks [11].
  • 37. Figure 2.4 Stress intensity factor vs. time for constant ampltide loading. Lower crack opening stress for small cracks results in a larger effective stress intensity factor than a large crack under identical loading, translating into faster growth rates.
  • 38. Figure 2.5 Cumulative fatigue failure distributions from 1984-1987 for the 7050-T7451 thick plate (5.5-5.9 inches = 140-150 mm).
  • 39. Figure 2.6 Large crack da/dN-ΔK data for the Al 7050-T7451 alloy, R=0.1 [29].
  • 40. 20 CHAPTER 3 - EXPERIMENTAL PROCEDURES In this chapter, the experimental procedures for the test program are presented. The first section covers the procedures involved with the acquisition of small crack growth rate data, while the second section covers the procedures involved with the acquisition of baseline data through large crack testing. 3.1. Small Crack Specimen Design and Testing Procedures As mentioned previously, specimens which incorporate a semicircular edge notch are useful in the procurement of small crack growth rate data. The original specimen design used in this study was a double-edge notch dogbone specimen, and is illustrated in Figure 3.1. The test specimen was secured to the load frame through pin-hole grips. Two semicircular edge notches were placed on opposite sides of the specimen in order to increase the amount of obtainable data in a single test. However, two constraints are placed on this type of specimen design. First, the width of the specimen must be large enough so that the growth of a small crack in one notch is not affected by the presence of the opposite notch and/or other small cracks growing in the opposite notch. At the same time, the width is limited by the diameter of the pins used to grip the specimen. Trial tests would often fail in the pinhole grip area if the dog boned width was greater than the 0.75 inch (1.9 cm) diameter of the pins. Two tests were successfully performed with this specimen design, but considering the constraints involved, a better design was required.
  • 41. 21 Discussions with ALCOA personnel on this problem [30] centered on the method used to grip the specimen. To circumvent specimen failure in the grips, ALCOA supplied this study with a grip design that "clamped" the specimen to the grips. The normal force applied to the specimen faces generates enough friction to prevent the specimen from slipping out of the grips. These "friction" grips allowed the specimen to be simplified to a double-edge notch specimen with no dogbone. This rectangular- shaped specimen is illustrated in Figure 3.2, and was the design used in this rest of the test program. By increasing the specimen width to 2.00 inches (5.08 cm), this ensured that the two notches of radii = 0.188 inch (4.78 mm) interacted little with each other. A two-dimensional finite element analysis was performed on both the dogbone and rectangular geometries to determine if the specimen width had any effect on the stress distribution at the notch. The stress concentration factor, Kt, is defined as the ratio of the hoop stress divided by the remote stress. In Figure 3.3, Kt's obtained from the finite element analysis are plotted versus the angle off of the mid plane of the notch, Θ, for both geometries. The figure reveals that both geometries exhibit essentially the same stress distribution at the notch. Therefore, the crack initiation data obtained from both specimen geometries were treated as equivalent in this study. To ensure that crack initiation occurs at material inhomogeneities and not machining marks, the notch surfaces were polished down to a 600 grit, followed by a diamond paste. The surface is then etched with a Keller's etch for 10-30 seconds. This removes any residual stresses generated by the machining and polishing process, and provides a "map" of the notch surface by extracting grain boundaries.
  • 42. 22 3.1.1. Specimen Testing Specimens fabricated from all three materials were tested in a servo-hydraulic test machine with analog-based electronic controls in laboratory air under a constant stress ratio (R = 0.1). Strain gages were placed on both sides of the specimen to measure the difference in strain experienced during axial loading, giving an indication of the bending present in the specimen. All tests involved applying two or three cycles to the maximum load to ensure the specimen was aligned correctly within the grips, and that potential bending strains were less than 5 % of the strains induced by the axial loading at the commencement of testing. Most specimens were loaded at a maximum nominal stress σnom =16 ksi (110 MPa), although two specimens were loaded at σnom = 15 ksi (103 MPa) and one specimen at σnom = 18 ksi (124 MPa). Table 3.1 lists the test parameters for the ten semicircular edge notch test specimens. "Old" material specimens are identified by the number 6•••-••, and were obtained from blanks of the aluminum 7050-T7451 alloy with the most microporosity. "New" material specimens are identified by the number 7•••-••, and were obtained from blanks of the material with less microporosity. The blanks themselves were formed from a six-inch thick plate at the ALCOA Technical Center. In addition, the numbering system for the "old" and "new" material specimens is an abbreviation of the numbering system ALCOA provided with the blanks. "Three-inch plate" specimens are identified by the number 8••, and were obtained from a three inch thick plate of aluminum 7050- T7451. This version of the material has the least amount of microporosity of all three materials [30].
  • 43. 23 3.1.2. The Replication Method As mentioned previously, surface crack initiation and growth were monitored with the replication method [19]. Cycling was suspended periodically throughout the test, and the specimen held under a constant tensile load while the notch surface was replicated. The tensile load was equal to eighty percent of the mean load, ensuring that all crack faces were open in the notch and thus making detection of the crack easier. The notch surface was bathed with 1-2 drops of acetone from a hypodermic needle. Finally, a 0.003 inch (76 μm) thick acetate tape was placed within the notch; this is shown schematically in Figure 3.4. The acetone softens the tape, allowing it to conform to the notch surface and flow into the mouths of open cracks. Great care must be taken during the replication process so that no air bubbles are trapped between the notch surface and the tape. No information of the notch surface is transferred to the tape where a bubble is located. After approximately 25 seconds, the tape is dry, leaving an exact replica of the notch surface. At this point, the tape can be removed from the notch surface and testing can recommence. Approximately 25-50 replicas were taken throughout each test to sufficient enough data points are available for analysis. Once the fatigue test was completed, analysis of the replicas begins. In several of the tests, individual cracks coalesced into a single crack. To keep track of the crack coalescence process, the following crack identification system was developed. When measuring the cracks from the replicas, the last replica taken was examined first. This replica would usually include a through-the-thickness crack, and sometimes smaller surface and corner cracks that did not become the dominant crack. Each of these cracks would be given an integer identification number 1, 2, 3, etc. As these cracks were traced back in time through earlier replicas, an initial crack, say Crack ID # 2, would "divide" into two smaller cracks (i. e. crack coalescence). These two cracks would then be given
  • 44. 24 the identification numbers 2.1 and 2.2, indicating that they coalesced into Crack ID # 2 at a later time in the test. Similarly, Crack ID # 2.1 could "divide" into Crack ID's # 2.11 and # 2.12 as they were traced through earlier replicas. This crack identification system provides a simple means to keep track of crack coalescence history, and hopefully aids in following this coalescence process in a single plot of crack length versus cycles for a particular specimen notch. Cracks were measured from the replicas via two different methods. Larger cracks, defined as a ≥ 0.003 inches (76 μm) were measured with a low powered (magnification ≈ 7 ×) optical microscope. Replicas were mounted on a slide viewing stage and the crack tip coordinates were measured using two micrometers attached to the stage. The micrometers provided resolutions of 0.0001 inches (2.5 μm). The crack tip coordinates were then converted to the x-s coordinate system and subsequently into the x-Θ coordinate system. The x-coordinate is the distance along the bore of the notch. The s-coordinate is defined as the notch radius × Θ, where Θ is the angle in radians above/below the mid plane of the notch; see Figure 3.5. This determined the spatial location of the crack within the notch, and subsequently its length. For cracks smaller than 0.003 inches (76 μm), a higher powered optical microscope was used. This microscope provided magnifications up to 1120 ×, and crack lengths were measured from a video screen connected to the microscope. Due to the limited viewing field, only crack lengths could be obtained from this method - not crack tip coordinates. However, spatial location of the cracks along the notch bore could be obtained from other measurements with the other microscope once crack lengths become larger. Cracks lengths in the range a ≈ 0.003 inches (76 μm) were measured with both methods; these lengths showed good agreement with each other.
  • 45. 25 An important concern with measurement is that the replicas would shrink 5-10% as they dried on the specimen surface. Therefore, the measurements were normalized with a shrinking factor. This factor was simply the ratio of the known notch thickness to the measured replica width, providing a scale for all measurements made on that replica. Small crack growth experiment results from the semicircular edge notch specimens are presented in Chapter 4. 3.2. Large Crack Testing Procedures ALCOA researchers have established the large fatigue crack growth properties for the aluminum 7050-T7451 alloy through numerous fatigue tests under various loading conditions and specimen geometries [31]. However, for completeness it was decided to quantify large fatigue crack growth rate properties for the alloys. In addition, supplementary large crack testing would further substantiate ALCOA's belief that all three versions of the alloy exhibited the same large crack growth properties [30]. To accomplish this, compact tension (CT) specimens were fabricated from fractured semicircular edge notch specimens, as is illustrated in Figure 3.6. The CT specimens were designed in accordance with ASTM Standard E647 [32], and its geometry is shown in Figure 3.7. The CT specimens were fabricated from fractured semicircular edge notch specimens to conserve the material used in this study. Although the CT specimens consist of material that has been previously cycled, it is believed that once a pre crack has started in the specimen, the large crack growth properties are relatively unaffected by the previous loading.
  • 46. 26 Table 3.1 lists the test parameters for the four CT test specimens. CT specimen pre cracking was conducted according to ASTM Standard E647 [32]. Traveling microscopes were mounted on both sides of the specimen in order to obtain front and back through-crack lengths. One of the microscopes was attached to a digital measuring system accurate to 0.0005 inches (13 μm). However, due to equipment problems with the second digital measuring system, a microscale accurate to 0.005 inches (130 μm) was used to obtain crack lengths with the other microscope. Through crack lengths were taken as the average of the front and back crack lengths, and were recorded to the nearest 0.005 inch (130 μm). The large crack growth rate data obtained from the CT specimen tests are presented in Chapter 4.
  • 47. Table 3.1 Parameters for fatigue test specimens. Specimen ID Material Type Specimen Type Max. Nominal Stress (ksi / MPa) Stress Ratio Frequency 6714-a11 "old" dogbone DEN 15 / 103 0.1 5 Hz 6714-a12 "old" dogbone DEN 15 / 103 0.1 10 Hz 6612-b21 "old" friction DEN 16 / 110 0.1 10 Hz 6611-a12 "old" friction DEN 16 / 110 0.1 10 Hz 7111-b11 "new" friction DEN 15, 18 / 103, 124 0.1 8 Hz 7111-b12 "new" friction DEN 16 / 110 0.1 10 Hz 7012-a22 "new" friction DEN 16 / 110 0.1 10 Hz 8T3 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz 8B3 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz 8B2 "3-inch plate" friction DEN 16 / 110 0.1 10 Hz Max. Load for CT Specimens (lbs / N) 6611-a12-CT2 "old" CT 350 / 1560 0.1 10 Hz 7012-a21-CT4 "new" CT 350 / 1560 0.1 10 Hz 8T3-CT3 "3-inch plate" CT 450 / 2000 0.1 10 Hz 8T3-CT4 "3-inch plate" CT 450 / 2000 0.1 10 Hz
  • 48. Figure 3.1 Double-edge notch dogbone specimen geometry and dimensions.
  • 49. Figure 3.2 Double-edge notch specimen geometry and dimensions.
  • 50. Figure 3.4 Illustration of the replication process.
  • 51. Figure 3.5 Definition of replica coordinate system.
  • 52. Figure 3.6 CT specimens fabricated from fractured double-edge notch specimen.
  • 53. Figure 3.7 CT specimen geometry and dimensions.
  • 54. Theta vs. (Remote Stress/Hoop Stress) Semicircular Edge Notch Geometries Theta (radians) Srem/Shoop -0.5 0 0.5 1 1.5 2 2.5 3 3.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 Width = 1.11 in. Width = 2 in. Figure 3.3 Hoop Stress/Remote Stress, σhoop/σrem, vs. Theta, Θ, for the 1.11 and 2.00 inch wide specimen geometries. σhoop/σrem was caculated with a 2-dimensional finite element analysis of each of the specimen geometries. Θ is defined as the angle (in radians) from the tip of the notch to the upper/lower location where the notch meets the specimen edge.
  • 55. 35 CHAPTER 4 - EXPERIMENTAL RESULTS 4.1. Large Crack Growth Rate Data Four CT specimens were tested to obtain the large crack growth properties for all three versions of the aluminum 7050-T7451 alloy. All CT specimens were tested at a frequency of 10 Hz and a stress ratio of 0.1. The "old" material CT specimen was loaded at a maximum load of 300 lbs (1330 N). The "new" material CT specimen was loaded at a maximum load of 350 lbs (1560 N). Finally, two "3-inch plate" material CT specimens were loaded at maximum loads of 300 lbs (1330 N) and 450 lbs (2000 N). All CT specimen testing and data analysis was performed in accordance with ASTM Standard E647 [32]. The stress intensity factor range solution for the CT specimen geometry is given by Δ Δ K P B W = + − + − + − ( ) ( ) ( . . . . . )/ 2 1 0 886 4 64 13 32 14 72 5 63 2 2 3 4α α α α α α (4.1) where a ≡ crack length, W ≡ width, B ≡ thickness, ΔP ≡ applied load range, and α = a/W. Fatigue crack growth rates were calculated with a seven-point polynomial technique [32]. The fatigue crack growth rates for the CT specimens are plotted against the applied stress intensity factor range in Figure 4.1. Two important things can be discerned from this
  • 56. 36 "large" crack growth rate curve. First, all three versions of the alloy exhibit essentially the same large fatigue crack growth rate properties. Second, the CT crack growth rate data correlates well with numerous fatigue tests performed by ALCOA under various loading conditions and specimen geometries. The Paris law expression shown in Figure 4.1 for Al 7050-T7451 (R = 0.1) was obtained from Reference [29] for the data reproduced here in Figure 2.5, and is given by da dN K= × − 3 9 10 10 4 175 . ( ) . Δ (4.2) The units for ΔK in Equation 4.2 are ksi√in, while da/dN is measured in inches/cycle. Although CT test data obtained here do not extend into either the threshold ΔK or the fracture toughness regions of the da/dN-ΔK curve, it does correlate well with the ALCOA generated data shown in Figure 2.5 [29]. Thus, the ALCOA Paris Law expression (Equation 4.2) is used here for subsequent analysis of the materials' large crack growth properties. 4.2. Small Crack Test Results Table 4.1 summarizes the results of the double-edge notch specimen tests. Before studying the results of individual tests, some general information should be noted first. The terms "front" and back" identify the notch location relative to the servo-hydraulic test machine. The table indicates that cracks initiated at an equal rate in both the front and back notches for all the specimens tested. This provided added assurance that potential bending was kept to a minimum in the tests, i. e., there was no bias as to which notches caused crack initiation.
  • 57. 37 In addition, two different crack lengths are used here to define fatigue crack "initiation." Although crack lengths of 2a<0.001 inch (25 μm) were obtained, several tests had cracks of that length traceable back to "zero" cycles. In actuality, however, the term "zero" cycles does not include specimen loading which occurred during gripping and alignment procedures. All tests involved applying two or three cycles to the maximum load to ensure the specimen was aligned correctly within the grips, and that potential bending strains were less than five percent of the strains induced by the axial loading. Therefore, a more generous initiation length of 2a=0.005 inch (127 μm) was also included in the table. It is important to note that both definitions of "initiation" place the crack length well within the small crack region of 2a < 0.02 inch (500 μm). Surface crack length vs. number of elapsed cycles for the double-edge notch tests are plotted in Figures 4.2 - 4.14. Surface crack lengths are plotted until "breakthrough", i. e., until the surface crack has become a through crack at the notch. In several of the tests, a single crack initiated at approximately the center of the notch and grew into the through crack that eventually caused specimen failure. A series of replica photographs in Figures 4.15 - 4.20 illustrates the growth of a lone crack in the "back" notch of specimen 6612-b21. In Figure 4.15, a crack appears to be emanating from a micropore at "0" cycles. By 10,001 cycles (Figure 4.16), the crack has grown and established itself. Figures 4.17 - 4.20 follow the growth of the crack at a lower magnification from 10,001 cycles to 42,509 cycles. Some tests, however, were characterized by multiple cracks initiating at several points along the bore of the notch. These cracks in turn coalesced into larger cracks, with a dominant crack eventually leading to specimen failure. Specimen 6611-a12 (Figure
  • 58. 38 4.3) is the most prolific example of multiple crack initiation, with ten different cracks initiating in the front notch. The most likely reason for the large number of cracks is the high degree of microporosity in the "old" material, resulting in a greater number of initiation sites in this specimen. An interesting phenomenon associated with multiple crack initiation / interaction is illustrated in specimen 7012-a22 (Figures 4.7 and 4.8). For some of the cracks, the final length measurements are smaller than measurements taken at previous cycles. It may be possible that extension of the large dominant crack prevents complete opening of adjacent smaller cracks, and thus makes them appear to be smaller as life progresses. For example, in Figure 4.8, Crack ID #1 in the final measurement is a through crack; it is fully open. However, Crack ID #'s 2 and 3 are only small surface cracks compared to #1, and are only partially open in the final measurements. A series of replica photographs in Figures 4.21 - 4.24 illustrate multiple cracks interacting with one another in the front notch of specimen 6611-a12. At 42,507 cycles (Figure 4.21), Crack ID #'s 4.1 and 4.2 are shown in the center and upper-right hand corner, respectively, while Crack ID # 8 is essentially a micropore to the left of # 4.1. By 50,007 cycles (Figure 4.22), # 8 has established itself, while # 4.1 is growing towards both # 4.2 and # 8. At 60,008 cycles (Figure 4.23), however, # 4.1 has bypassed # 8, and has almost coalesced with # 4.2. By 65,008 cycles (Figure 4.24), cracks 4.1 and 4.2 have coalesced into Crack ID # 4. Crack ID # 8 is starting to close due to its close proximity to the larger # 4. Examination of the fracture surfaces provides another method in determining fatigue crack initiation. ALCOA researchers have examined the fracture surfaces of Al 7050-T7451 open hole fatigue specimens with the scanning electron microscope (SEM)
  • 59. 39 in order to locate crack initiation sites [1]. In that study, they determined that the fatigue cracks initiated from micropores in the material rather than machining flaws. SEM examination of this study's double-edge notch specimen fracture surfaces is currently being performed by Jon Elsner on a JEOL JSM-T300 SEM [33]. The accelerating voltage is 25 kV, and utilizes background scatter electrons as the imaging technique. An example of Elsner's current work is presented here to illustrate the technique and initial results. Figure 4.25 is a fractograph of the "front" notch fracture surface for specimen 6611-a12, and shows two cracks which developed at this notch. Although catastrophic failure initiated at the "back" notch of this specimen, the elliptical shapes of the dominant cracks in the "front" notch were preserved. The larger crack on the left was identified as Crack ID # 6 during the replica measurement process, while the smaller crack on the right was identified as Crack ID # 5. Figures 4.26 - 4.29 show larger magnifications of the initiation sites for Crack ID #'s 5 and 6. In both cases, the initiation sites appear to be micropores in the material just beneath the notch surface. 4.3. Small Crack Growth Rate Data Specimens where a single crack initiated in one of the notches were used here to characterize the small fatigue crack growth rate data. As shown in Equation 2.4, Newman has developed approximate K solutions for corner and surface cracks in a semicircular edge notch [4]. Variable definitions are illustrated in Figure 2.3, whereas noted previously t is defined as B for a corner crack, whereas for a surface crack, t is defined as B/2. Newman's full stress intensity factor solutions are given here in Appendix A.
  • 60. 40 It is important to note that the specimen design met all restrictions placed on the edge notch geometry for the K solution to be valid except for the requirement that r/w = 0.0625 (see Appendix A for complete geometry restrictions). For the early dog bone double-edge notch specimens, r/w ≈ 0.0845; for the friction grip double edge-notch specimens, r/w ≈ 0.0469. As mentioned previously in Chapter 3, a finite element analysis was performed to determine the stress distribution at the notch for both specimen geometries. The stress concentration factor Kt at the mid plane of the notch (Θ = 0) was calculated to be Kt = 3.03 for the dogbone geometry and Kt = 3.05 for the friction grip geometry. These results are 3.5 % less than the stress concentration factor used in the Newman ΔK solutions of Kt = 3.15 for uniform displacement [4]. Because of the close correlation between the finite element analysis Kt's and the Kt used in Newman's stress intensity factor solutions, the r/w restriction was considered insignificant in this study. The replication method can only obtain the surface lengths of cracks, or "2a", and not the crack depths, "c", defined in Figure 2.2. Since the stress intensity factor solution for cracks at a semicircular edge notch depends on the crack aspect ratio, a/c, an expression for a/c is required to calculate ΔK's for the double-edge notch specimen. Swain and Newman measured crack lengths in both the a and c direction with the use of marker loads in the 2024-T3 aluminum alloy [34]. Based on the experimental data, they developed an empirical relationship between the crack shape, c/a, and the non dimensional length, a/t, given by c a = −0.9 0.25(a t)2 (4.3)
  • 61. 41 This expression is plotted in Figure 4.30. In addition, a representation of the crack shapes predicted by the expression is shown to scale in Figure 4.31 for a corner crack. As mentioned earlier, the SEM examination of the fracture surface of specimen 6611- a12 allows for actual crack shape measurements to be obtained for Crack ID #'s 5 and 6 in the "front" notch. These measurements are also plotted in Figure 4.30, and correlate well with Swain and Newman's empirical prediction. Although an exhaustive examination of all specimen fracture surfaces has not been performed at this time, the empirical expression for c/a vs. a/t should be adequate for calculating the stress intensity factor ranges. The small crack growth rates for double-edge notch specimens where a single crack initiated along the bore of a notch are plotted against the applied stress intensity factor ranges in Figure 4.32. The Paris Law expression obtained from ALCOA fatigue tests for Al 7050-T7451 (Equation 4.2) is also plotted for comparison. The crack growth rate data obtained from these tests demonstrates that small cracks do, in fact, grow faster than large cracks at equivalent ΔK loading near the threshold region. However, the small crack growth rate data merges with the large crack Paris Law at higher ΔK's. A linear regression was performed on the data to obtain the Paris Law constants for the small crack growth rate data. The "small" crack Paris Law expression is given by da dN K= × − 8 22 10 9 2 807 . ( ) . Δ (4.4) The units for ΔK in Equation 4.4 are ksi√in, while da/dN is measured in inches/cycle. It is important to note that there is greater variability in the small crack growth rate data compared to large crack growth rate data. This is not surprising due to the fact that LEFM principles are being pushed to the limit as well as uncertainties in the small crack
  • 62. 42 measurements. The Paris Law expressions for both small and large crack growth rate in Al 7050-T7451 are incorporated in a computer program that predicts the crack growth of surface and corner cracks in a variety of geometries. In addition, the single crack growth rate Paris Law expressions will be used to predict the growth of multiple cracks by taking into account interaction between the crack tips. The program's background and implementation in this study is presented in Chapter 5.
  • 63. Table 4.1 Test matrix for the double-edge notch specimens. Specimen ID Nominal Stress (ksi / MPa) Number of Cracks (front) Number of Cracks (back) Cycles to 2a ≥ 0.001 " (25 μm) Cycles to 2a ≥ 0.005 " (127 μm) Cycles to Specimen Failure "Old" Material: 6714-a11 15 / 103 0 2 0 25,002 120,523 6612-b21 16 / 110 0 1 0 20,002 76,952 6714-a12 15 / 103 2 0 30,150 65,485 174,958 6611-a12 16 / 110 10 3 2,501 12,502 75,066 "New" Material: 7111-b11 15, 18/103, 124† 0 1 183,071 224,564 254,835 7111-b12 16 / 110 0 2 162,511 200,004 243,912 7012-a22 16 / 110 4 4 0 27,501 125,262 3 " Plate Material: 8T3 16 / 110 1 0 120,708 128,209 245,771 8B3 16 / 110 6 1 15,001 22,502 77,798 8B2 16 / 110 0 1 15,002 22,503 86,962 † Specimen 7111-b11 was loaded at a maximum nominal stress of 15 ksi for 194,403 cycles. At that point, the maximum nominal stress was increased to 18 ksi.
  • 64. Fatigue Crack Growth Rate Data for Al 7050-T7451 Alloy Delta K, ksi sqrt(in.) da/dN,in./cycle 1.00E-10 1.00E-09 1.00E-08 1.00E-07 1.00E-06 1.00E-05 1.00E-04 1.00E-03 1 10 100 8T3-CT4 (3 in. plate material) 8T3-CT3 (3 in. plate material) 6611-a12-CT2 (old material - more microporosity) 7012-a21-CT4 (new material - less microporosity) Large Crack Paris Law da/dN=3.9e(-10)*dK^(4.175) Figure 4.1 Large crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from CT specimens) The large crack Paris Law was obtained from Reference [29] (see Figure 2.5).
  • 65. Crack Length vs. Number of Cycles: Specimen 6611-a12, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 10000 20000 30000 40000 50000 60000 70000 Crack ID # 1 Crack ID # 1.1 Crack ID # 1.2 Crack ID # 2 Figure 4.2 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "back" notch.
  • 66. Crack Length vs. Number of Cycles: Specimen 6611-a12, Front Notch Number of Cycles, N CrackLength,a(inches) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 10000 20000 30000 40000 50000 60000 70000 Crack ID # 1 Crack ID # 2 Crack ID # 3 Crack ID # 4.1 Crack ID # 4.2 Crack ID # 4 Crack ID # 5 Crack ID 6.1 Crack ID 6.2 Crack ID # 6 Crack Id # 7 Crack ID # 8 Figure 4.3 Surface crack length vs. number of elapsed cycles for specimen 6611-a12 (old material), "front" notch.
  • 67. Crack Length vs. Number of Cycles: Specimen 6612-b21, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 10000 20000 30000 40000 50000 60000 70000 Crack ID # 1 Figure 4.4 Surface crack length vs. number of elapsed cycles for specimen 6612-b21 (old material), "back" notch.
  • 68. Crack Length vs. Number of Cycles: Specimen 6714-a11, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 20000 40000 60000 80000 100000 120000 Crack ID #2 Crack ID # 1 Figure 4.5 Surface crack length vs. number of elapsed cycles for specimen 6714-a11 (old material), "back" notch.
  • 69. Crack Length vs. Number of Cycles: Specimen 6714-a12, Front Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 20000 40000 60000 80000 100000 120000 140000 160000 180000 Crack ID # 1.1 Crack ID # 1.2 Crack ID # 1 Figure 4.6 Surface crack length vs. number of elapsed cycles for specimen 6714-a12 (old material), "front" notch.
  • 70. Crack Length vs. Number of Cycles: Specimen 7012-a22, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 20000 40000 60000 80000 100000 120000 Crack ID # 1 Crack ID # 2 Crack ID # 3 Crack ID # 4 Figure 4.7 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "back" notch.
  • 71. Crack Length vs. Number of Cycles: Specimen 7012-a22, Front Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 20000 40000 60000 80000 100000 120000 Crack ID # 1 Crack ID # 2.1 Crack ID # 2.2 Crack ID # 2 Crack ID # 3 Figure 4.8 Surface crack length vs. number of elapsed cycles for specimen 7012-a22 (new material), "front" notch.
  • 72. Crack Length vs. Number of Cycles: Specimen 7111-b11, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 50000 100000 150000 200000 250000 Crack ID # 1 Figure 4.9 Surface crack length vs. number of elapsed cycles for specimen 7111-b11 (new material), "back" notch.
  • 73. Crack Length vs. Number of Cycles: Specimen 7111-b12, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 50000 100000 150000 200000 250000 Crack ID # 1.1 Crack ID # 1.2 Crack ID # 1 Figure 4.10 Surface crack length vs. number of elapsed cycles for specimen 7111-b12 (new material), "back" notch.
  • 74. Crack Length vs. Number of Cycles: Specimen 8B2, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 10000 20000 30000 40000 50000 60000 70000 80000 Crack ID # 1 Figure 4.11 Surface crack length vs. number of elapsed cycles for specimen 8B2 (3-inch plate material), "back" notch.
  • 75. Crack Length vs. Number of Cycles: Specimen 8B3, Back Notch Number of Cycles, N CrackLength,a(inches) 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0 10000 20000 30000 40000 50000 60000 Crack ID # 1 Figure 4.12 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "back" notch.
  • 76. Crack Length vs. Number of Cycles: Specimen 8B3, Front Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 10000 20000 30000 40000 50000 60000 Crack ID # 1 Crack ID # 1.1 Crack ID # 1.2 Crack ID # 1.21 Crack ID # 1.22 Crack ID # 2 Crack ID # 3 Crack ID # 4 Crack ID # 5 Figure 4.13 Surface crack length vs. number of elapsed cycles for specimen 8B3 (3-inch plate material), "front" notch.
  • 77. Crack Length vs. Number of Cycles: Specimen 8T3, Front Notch Number of Cycles, N CrackLength,a(inches) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 50000 100000 150000 200000 250000 Crack ID # 1 Figure 4.14 Surface crack length vs. number of elapsed cycles for specimen 8T3 (3-inch plate material), "front" notch.
  • 78. Empirical Expression for Crack Shape vs. Nondimensional Length a/t c/a 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Empirical Expression for a/t vs. c/a Measured Values of a/t vs. c/a Figure 4.30 Empirical expression for crack shape vs. nondimensional length [28]. Measured values are from Crack ID #'s 5 and 6 from specimen 6611-a12, "front" notch. Definitions for c, a, and t are illustrated in Figure 4.31.
  • 79. Figure 4.31 Illustration of corner crack shape based on empirical expression for c/a vs. a/t (to scale). Note: for the corner crack, t≡B.
  • 80. Small Fatigue Crack Growth Rate Data for Al 7050-T7451 Alloy Delta K, ksi sqrt(in.) da/dN,in./cycle 1E-10 1E-09 1E-08 1E-07 1E-06 1E-05 0.0001 0.001 1 10 100 Large Crack Paris Law Small Crack Paris Law 6714-a11, back notch 6612-b21, back notch 7111-b11, back notch 7111-b12, back notch 8B2, back notch 8B3, back notch 8T3, front notch da/dN=3.9e(-10)*dK^(4.175) da/dN=8.22e(-9)*(dK)^2.807 Figure 4.32 Small crack growth rate vs. stress intensity factor range data for Al 7050-T7451 (obtained from double-edge notch specimens).
  • 81. Figure 4.25 SEM fractograph of specimen 6611-a12, "front" notch fracture surface. The larger crack on the left was identified as Crack ID # 6 during the replica measurement process. The smaller crack on the right was identified as Crack ID # 5 during the replica measurement process. The reference line on the fractograph is 1000 μm in length.
  • 82. Figure 4.26 Initiation site of Crack ID # 6. The reference line on the fractograph is 100 μm in length. Figure 4.27 Close-up of initiation site of Crack ID # 6. The reference line on the fractograph is 10 μm in length.
  • 83. Figure 4.28 Initiation site of Crack ID # 5. The reference line on the fractograph is 100 μm in length. Figure 4.29 Close-up of initiation site of Crack ID # 5. The reference line on the fractograph is 10 μm in length.
  • 84. Figure 4.15 Replica photograph of specimen 6612-b21, "back" notch, 0 cycles (after specimen alignment loading). Crack ID # 1: 2a = 0.0016 in. Figure 4.16 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles. Crack ID # 1: 2a = 0.0038 in.
  • 85. Figure 4.17 Replica photograph of specimen 6612-b21, "back" notch, 10001 cycles. Crack ID # 1: 2a = 0.0038 in. Figure 4.18 Replica photograph of specimen 6612-b21, "back" notch, 23001 cycles. Crack ID # 1: 2a = 0.0092 in.
  • 86. Figure 4.19 Replica photograph of specimen 6612-b21, "back" notch, 33506 cycles. Crack ID # 1: 2a = 0.0179 in. Figure 4.20 Replica photograph of specimen 6612-b21, "back" notch, 42509 cycles. Crack ID # 1: 2a = 0.0337 in.
  • 87. Figure 4.21 Replica photograph of specimen 6611-a12, "front" notch, 42507 cycles. Crack ID # 4.1: 2a = 0.0208 in. Crack ID # 4.2: 2a = 0.0025 in. Crack ID # 8: 2a = 0.0009 in. Figure 4.22 Replica photograph of specimen 6611-a12, "front" notch, 50007 cycles. Crack ID # 4.1: 2a = 0.0275 in. Crack ID # 4.2: 2a = 0.0036 in. Crack ID # 8: 2a = 0.0032 in.
  • 88. Figure 4.23 Replica photograph of specimen 6611-a12, "front" notch, 60008 cycles. Crack ID # 4.1: 2a = 0.0379 in. Crack ID # 4.2: 2a = 0.0047 in. Crack ID # 8: 2a = 0.0032 in. Figure 4.24 Replica photograph of specimen 6611-a12, "front" notch, 65008 cycles. Crack ID # 4: 2a = 0.0441 in. Crack ID # 8: 2a = 0.0032 in.
  • 89. 69 CHAPTER 5 - ANALYTICAL MODELING A computer program was employed to predict both crack shape and fatigue crack growth in the double-edge notch specimens tested in this study. A brief history of the program and its various implementations is presented in this chapter. This is followed by modifications made to the program to fit this study as well as a description of how the algorithm works. 5.1. Background The multi-degree of freedom algorithm used in this study was originally coded by Tritsch [34] to predict the fatigue life and crack growth shapes for both single and double cracks located along the bore of a hole loaded under remote tension; see Figures 5.1 and 5.2. The program utilized the Newman-Raju stress intensity factor solutions [36] for a single surface or corner crack in a hole. These K solutions could subsequently be modified with correction factors to handle various geometries as well as crack interaction effects. Tritsch's original algorithm utilized Bowie's two-dimensional stress intensity factor solution for a through-cracked hole [37] to develop correction factors so that the computer program could be used for various specimen geometries. For this study, however, Newman's three-dimensional K solutions for corner and surface cracks in a
  • 90. 70 semicircular edge notch [4] were available, thus eliminating the need to use correction factors based on two-dimensional geomtries. For the cases of two cracks along the bore of a hole (Figure 5.2), Tritsch employed a crack interaction factor developed by Heath and Grandt [38]. They used the Finite Element-Alternating Method (FEAM) to obtain stress intensity factor solutions for both a single corner crack along the bore of a hole and symmetric corner cracks on the same side of the hole. These solutions were calculated for crack shapes of a/c = 1.11, 1.5, 2.0, and 3.0. The interaction factor, γ, is then given by γ = K K symmetric c c gle c c L L . . sin . . (5.1) Ksymmetric cc is the stress intensity factor for two corner cracks along the bore of a hole symmetric with respect to the plane at the half-thickness of the specimen, while Ksingle cc is the stress intensity factor for the single corner crack along the bore of a hole. The interaction factor is a function of the crack shape, a/c, and the non dimensional separation distance between the two symmetric corner cracks, ts/a. Polynomial expressions were subsequently fit to the FEAM results, and incorporated into the program. It is important to note that the interaction factor was employed only on the tips of the two cracks adjacent to one another; the crack tips next to the free surfaces were not modified. In addition, although γ was based on the interaction between two symmetric corner cracks, it was employed to handle various unsymmetric combinations of corner and surface cracks. Scheumann [39] updated the original code to take into account interaction effects between unsymmetric cracks on opposite sides of a hole in a plate. In addition, Grandt, Hinkle, Scheumann, and Todd [29] developed an interaction factor based on Trantina and
  • 91. 71 Barishpolsky's [40] effective stress intensity factor for an ellipsoidal void in a large body with an equatorial crack. This interaction factor was employed to predict the growth of cracks initiating from particles and micropores in Al 7050-T7451 open hole fatigue specimens [29]. Although predictions based on the Trantina-Barishpolsky interaction factor compared favorably with actual specimen lives in [29], the interaction factor was not used in this study. Instead, the small crack growth rate equation given in Equation 4.4 will be utilized in a modified version of Tritsch's original program [35] to back- predict fatigue crack growth to breakthrough and crack shapes in the double-edge notch fatigue specimens tested in this study. 5.2. Description of Algorithm The goal of a multi-degree of freedom algorithm is to predict both crack size and shape as a function of the applied load. As mentioned previously, Tritsch's original algorithm [35] was modified to utilize Newman's three-dimensional K solutions for corner and surface cracks in a semicircular edge notch [4]. Figure 5.1 illustrates geometry variable definitions utilized in the prediction code for the case of a single surface crack or corner crack. The initial dimensions of the crack are defined by the input coordinates, i. e., a a for surface cracks x x c y 1 1 3 1 1 2 2( ) = − = (5.2), (5.3) The crack depth, c1, was selected to grow an increment Δc1 = 0.0001×c1. The number of cycles required for c1 to grow an increment Δc1 is Δ ΔN c dy dN= 1 2( ) (5.4).
  • 92. 72 where dy2/dN, the crack growth rate in the c direction, is given by Equations 4.4 and 4.2 for small and large cracks, respectively. The stress intensity factor solutions for a surface and corner crack in an edge notch [4] are used to calculate the crack growth rates. The subsequent growth of the surface crack tips, xi, are then Δ Δx dx dN Ni i= ×( ) (5.5). Similarly, dxi/dN is given by Equations 4.4 and 4.2 for small and large cracks, respectively. The new crack coordinates defining the surface crack length and depth are then recalculated, and the process is repeated. This algorithm is easily extended to handle two or more cracks; a typical multi-crack configuration is shown in Figure 5.2. After each iteration, the program checks for crack tip free surface contact, or in the case of the double crack configuration, crack coalescence. If a change in the crack type occurs, e. g., a surface crack becomes a corner crack, the change is noted and all subsequent calculations are based on the new crack type. Once the surface or corner crack has broken through, iteration proceeds as a through crack with an initial crack length equal to the last crack depth calculation, c1. After breakthrough, the program utilizes the stress intensity factor solution presented by Newman [4] for a through crack located at a semicircular edge notch; it is given in the following form K S cF c w c r r wn= π ( / , / , / ) (5.6).
  • 93. 73 The full stress intensity factor solution is given in Appendix A. Iteration continues until either the fracture toughness of the material is reached, or a specified maximum crack growth rate is exceeded. The prediction code was adjusted to include a correction factor for finite notch width. The finite width correction factor was incorporated because the crack did not always initiate at the half-thickness of the specimen, i. e., one of the crack tips was closer to its respective specimen side than the other crack tip. However, the ΔK solution [4] assumes that a surface crack is located at the center of the notch. The correction factor is the ratio of stress intensity factors for a centrally located through crack, K2D center,and an eccentrically located through crack in a sheet of finite width, K2D eccentric, [41], i. e., γ fnw D eccentric D center K K = 2 2 L L .. . (5.7) The results of the back-prediction of fatigue crack shape and growth in the double edge- notch specimen are discussed in Chapter 6.
  • 94. 74 Figure 5.1 Geometry variable definitions used in the prediction program for a surface crack and a corner crack.
  • 95. 75 Figure 5.2 Geometry variable definitions used in the prediction program for a typical multiple crack configuration.
  • 96. 76 CHAPTER 6 - NUMERICAL RESULTS 6.1 Back-Prediction in Specimens Used to Calculate Small Crack da/dN-ΔK curve To verify the validity of the life prediction code, crack growth predictions were first performed for the specimens that were used to generate the small-crack da/dN-ΔK data in Chapter 4. Although these tests provided the fatigue crack growth data for the predictive model (Equation 4.4), these calculations are independent of the input data file in the sense that crack shape is predicted by the two-degree-of-freedom analysis described in Chapter 5. Recall that only the crack surface dimension "2a" was measured from the replicas, so that when computing the cyclic stress intensity factor ΔK to generate Equation 4.4, it was necessary to assume a corresponding crack depth "c" (i.e., Equation 4.3, which assumes that crack shape c/a is a function of crack size a/t). Thus, applying the crack growth analysis program to these tests provides an independent prediction for the crack shape c/a, which can then be compared with the assumed shape (Equation 4.3) employed to establish the da/dN relationship. Plots are given for both the surface crack length, a, vs. number of elapsed cycles, N, and crack aspect ratio, c/a, vs. the non dimensional surface crack length, a/t. Definitions of these crack parameters are illustrated in Figure 2.2. In the a vs. N plots of Figures 6.1-6.7, there is little difference between the predicted results of the original algorithm and the predicted results with the finite width