Dowling Material Properties
Dowling Material Properties
Dowling Material Properties
Science and Engineering Department, and Engineering Science and Mechanics Department (Jointly Appointed), 2 Engineering Science and
Mechanics Department, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA
Received in final form 12 October 2008
A B S T R A C T Mean stress effects in finite-life fatigue are studied for a number of sets of experimental
data for steels, aluminium alloys and one titanium alloy. Specifically, the agreement with
these data is examined for the Goodman, Morrow, SmithWatsonTopper and Walker
equations. The Goodman relationship is found to be highly inaccurate. Reasonable accuracy is provided by the Morrow and by the SmithWatsonTopper equations. But
the Morrow method should not be used for aluminium alloys unless the true fracture
strength is employed, instead of the more usual use of the stress-life intercept constant.
The Walker equation with its adjustable fitting parameter gives superior results. For
steels, is found to correlate with the ultimate tensile strength, and a linear relationship
permits to be estimated for cases where non-zero mean stress data are not available.
Relatively high-strength aluminium alloys have 0.5, which corresponds with the
SWT method, but higher values of apply for relatively low-strength aluminium alloys.
For both steels and aluminium alloys, there is a trend of decreasing with increasing
strength, indicating an increasing sensitivity to mean stress.
Keywords fatigue of materials; Goodman equation; mean stress effect; Morrow equation; SmithWatsonTopper equation; stress-life curve; Walker equation.
NOMENCLATURE
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
163
164
N . E . D O W L I N G et al.
(a)
(b)
(c)
(d).
(1)
max
0
t
a
min
(2)
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
2, 3
4
4
5, 6
4
4
7
7
4
4
8
5, 6
7
9
9
10
11
10
9
4, 12
13, 14
4
15
5, 6
15
15
5, 6
SAE 1015 St
GSMnNi63 Stc,d
Ck45 St
SAE 4130 St, Normd
49MnVS3 St
17MnCrMo33 St
CC 450 SS, H1150d
CC 450 SS, Soln And
50CrMo4 St, El Pol
42CrMo4 St
AISI 4340 Stc
SAE 4130 St, Hardd
CC 450 SS, H900d
PH138Mo SS, H1000d
300M Std
SAE 1045 St, 705 HB
SAE 1045 St, 55 HRC
SAE 1045 St, 595 HB
6061-T6 Ale
Al Mg4.5Mn, Cld Rlc,e,f
2024-T4 Alc,f
2024-T3 Alc,e,f
2014-T6 Ale
2024-T3 Ale
2024-T4 Ale
7075-T6 Ale
7075-T6 Ale
Ti-6Al-4V c,e
1006
276
298
303
378
438
359
335
489
521
228
312
531
647
566
833
678
1041
970
998
1103
1200
1354
1358
1634
1827
1731
1841
1034
310
363
476
486
494
497
502
567
572
415
501
790
799
840
929
1015
1065
1086
1111
1172
1241
1405
1413
1958
2082
2165
2248
Ultimate
u (MPa)
1271
420
476
631
578
581
591
666
729
736
726
846
1271
1144
1152
1446
1360
1410
1609
1525
1634
1586
1750
1758
2303
2131
2690
2717
Fracture
fB (MPa)
(14.5)
(13)
35
(17.3)
(13.6)
(20.3)
(21.4)
(16.5)
(12.3)
67.9
(26)
60
(15.2)
44
58
(23.2)
(12.2)
48.6
60
56
(8.3)
(15.3)
2
38
40.5
Red. Areag
(Elong.) (%)
990
739
1217
1584
1263
1275
1453
2046
1588
1505
1963
2378
2839
2521
4214
2370
3762
3034
651
1045
2499
1611
949
1772
897
3378
4546
2452
0.1379
0.0811
0.0984
0.0845
0.0948
0.0666
0.0591
0.0600
0.0841
0.0758
0.0958
0.1182
0.0840
0.0995
0.1150
0.0750
0.1075
0.1014
0.1151
0.1372
0.1755
0.1324
0.1221
0.1544
0.1051
0.2624
0.2388
0.1285
2262
722
1045
1944
1328
1120
1602
970
4402
4324
1020
672
1267
1168
1325
1232
1365
1326
1547
1452
1722
2437
2317
2889
3612
2418
3542
3149
0.1379
0.1080
0.1394
0.1969
0.1529
0.1080
0.1626
0.09771
0.2412
0.2419
0.1336
0.0925
0.0936
0.1040
0.0899
0.0752
0.0668
0.1078
0.0899
0.0810
0.1077
0.1160
0.1083
0.0879
0.1318
0.0728
0.1147
0.0985
0.5431
0.6330
0.6681
0.5031
0.5655
0.4803
0.4596
0.4090
0.4150
0.4774
0.7352
0.8113
0.6949
0.6903
0.8492
0.6575
0.6253
0.7051
0.7776
0.7781
0.6497
0.5457
0.4758
0.5769
0.4157
0.4839
0.4286
0.5245
fw (MPa)
f (MPa)
bw
Fit m = 0
214
5353
2800
5100
4249
1200
663
1000
1000
5,182
52
15
217
211
246
23
157
231
52
248
222
74
223
2797
684
2.5
26
20
Min (cycles)
1,100,000
83,870,000
2,000,000
1,836,000
2,700,000
7,260,000
1,404,000
29,400,000
929,000
1,992,000
1,241,000
1,900,000
620,000
1,984,000
1,710,000
2,840,000
72,950
94,613
380,000
520,000
901,430
1,023,000
90,941
1,266,000
652,600
161,250
750,791
572,000
Max (cycles)
a Abbreviations: St = steel, SS = stainless steel, Al = aluminium alloy, HB = Brinell hardness, HRC = Rockwell hardness, El Pol = electropolished, Soln An = solution annealed, H900, etc.
= heat treatment conditions, Norm = normalized, Hard = hardened, Cld Rl = cold rolled.
b Numbers refer to the list of References at the end of the paper.
c Samples initially overstrained before testing.
d True fracture strength estimated from + 345 MPa.
u
e True fracture strengths estimated by scaling data on similar material.
f Short life data omitted in stress-life fits; for the = 0 fits, the shortest life data points employed were N = 4750 cycles for Al Mg4.5Mn, 4200 cycles for 2024-T4 Al, and 2050 cycles for
m
f
2024-T3 Al.
g Reduction in area is listed, or if not available, elongation is listed.
Sourceb
Materiala
Yield
o (MPa)
166
N . E . D O W L I N G et al.
max a
1 R
ar = max
2
2
ar = a
1 R
Fig. 2 Stress-life curves for a steel and for an aluminium alloy for
zero mean stress, showing intercepts and true fracture strengths.
a
.
1 m /u
(3)
(a)
(b)
(c).
(5)
ar = max a
1 R
ar = max
2
1
2
ar = a
1 R
(a)
(b)
(c).
(6)
a
.
1 (m /u )2
(7)
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
1/b
1 R 1
N f = max
2
A
1.6
AISI 4340 Steel
u = 1172 MPa
1.4
a/
ar
1.2
0.8
Gerber
0.0
-400
Goodman
u
400
fB
800
1200
m , MPa
y = m1 x1 + m2 x2 + d ,
'f
1600
2000
where
m1 =
1
,
b
x1 = log max ,
m2 =
,
b
A = f 2b .
(8)
Combine this with Eq. (6b) and solve for the dependent
variable N f .
1 R
ar = ANbf = max
(a)
2
x2 = log
1
d = log A.
b
1 R
2
(12)
1
,
m1
= bm2 =
(11)
where
y = log N f ,
ar = ANbf ,
(9)
Morrow
0.4
0.2
(b).
1 R
log A. (10)
log N f = log max + log
b
b
2
b
1.0
0.6
167
m2
m1
f =
A
.
2b
(13)
To distinguish values of f and b obtained by this procedure from those obtained by mere fitting of zero mean
stress data, subscripts w for Walker are added, so that
the resulting three constants are denoted fw
, b w and .
Hence, the fitted stress-life curve is
ar = fw
(2N f )b w
(14)
Consequence
for life estimates
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
N . E . D O W L I N G et al.
168
1.2
2014-T6 Al
1.0
0.8
a / ar
0.6
0.4
0.2
0.0
-200
200
~
fB
400
600
'f
800
1000
1200
m , MPa
The relationships of Goodman or Morrow can be compared to test data by plotting normalized stress amplitudes
a /ar versus the mean stress m . For the Goodman relationship, Eq. (3), a straight line is expected between the
points a /ar = 1 and m = u . The two forms of the
Morrow relationship, Eqs (4a) and (4b), give similar lines,
but with intercepts at m = fB and m = f , respectively. Example comparisons with test data on this basis
are shown for steels in Figs 3 and 4 and for an aluminium
alloy in Fig. 5. For the data points shown, each a and m
combination corresponds to a laboratory test, and in each
case the ar value used to normalize a is determined from
the experimental life N f using Eq. (2).
For the steels, note that the data are above the Goodman
line for tensile mean stresses, that is, higher stresses than
predicted by this method are required to cause fatigue failure. Hence, fatigue life predictions made from a Goodman ar value for tensile mean stresses will be shorter than
actual fatigue lives, and the method thus errs on the conservative side. (See Table 2.) However, for compressive
mean stresses, the opposite is true, and the method is nonconservative. Either form of the Morrow relationship is
50CrMo4 Steel
u = 1086 MPa
1. 2
1.2
0.6
= 0.650
0.8
0.6
0.4
0.4
0.2
0.2
u
0.0
-400
1.0
ar
a/
ar
1.0
0.8
1.4
1.4
400
'f
800
1200
P
,
M
a
m
0.0
-1.2
fB
1600
2000
Data
SWT
Walker
-0.6
= 0.500
0.0
0.6
/
1.2
1.8
2.4
ar
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
1.4
1.2
= 0.778
0.8
ar
1.0
= 0.500
0.6
0.4
Data
SWT
Walker
0.2
0.0
-0.6
-0.4
-0.2
50CrMo4 Steel
u = 1086 MPa
0.0
0.2
m/
0.4
0.6
0.8
1.0
ar
1.4
Data
SWT
Walker
1.2
0.8
ar
1.0
0.6
= 0.500
0.4
0.2
0.0
-0.4
= 0.480
2014-T6 Al
0.0
0.4
0.8
1.2
/
1.6
2.0
2.4
2.8
ar
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
169
material there are four plots, one for each of Eqs (3),
(4a), (4b) and (5). On any particular plot, the degree to
which the resulting data points consolidate with the line of
Eq. (2) gives an indication of the success of the mean stress
relationship. (Recall that the Eq. 2 constants are obtained
from fitting the m = 0 data.)
For the Walker relationship, ar values from Eq. (6) are
similarly plotted versus the N f data for comparison with
the special fit of Eq. (14), which is noted to result from
fitting the full set of data at all mean stresses. Comparisons
of this type are shown in Figs 11 and 12 for 50CrMo4 steel
and 2014-T6 aluminium.
The trends seen in the ar versus N f plots correspond
to those already noted for steels and aluminium alloys on
the two types of amplitudemean plot. On any particular
plot, where test data points lie above the line for zero mean
stress, the estimated lives are too short, and the method
errs on the conservative side. Conversely, data points below the line indicate that the method is non-conservative.
(See Table 2.) The ar versus N f comparisons shown for
50CrMo4 steel and 2014-T6 aluminium are typical of the
steels and aluminium alloys studied, respectively. Where
the Walker differs significantly from 0.5, the Walker
correlation is noticeably better than the SWT one, and
the Walker method always gives very good correlation of
the test data.
Comments on the data sets analysed
For 18 of the 28 data sets, the tests were conducted in
strain control, with stress control sometimes being substituted for tests at relatively long lives where the strains
were fully elastic. However, only stress control was employed for the seven data sets referenced to sources in
the 1950s, and also for the three data sets obtained from
MIL-HDBK-5H.9
Tabulated numerical data were utilized, usually from the
source referenced, but in a few cases from data obtained
through contacting the original authors. The only exceptions were the data from MIL-HDBK-5H, which were
digitized from stress-life plots in that publication.
In strain-controlled tests at very short lives, that is,
around N f < 50 cycles, and especially for N f < 10 cycles, the very large strains involved may cause buckling or
other anomalous behaviour that affects the life. Hence,
data points in this range that seemed to depart from the
trend of the data at longer lives were not included, specifically, by omitting all points with N f less than a value
chosen as a matter of judgment. In the data sets for stresscontrolled tests, there were sometimes points with anomalously short lives where the maximum stress was around
or even greater than the materials yield strength. Such
data points were omitted, as they are surely affected by
large amounts cycle-dependent creep (ratcheting).
170
N . E . D O W L I N G et al.
(a) Goodman
1000
100
10
102
0
0.01
-0.01
R=0
R = -2
0 Fit
ar
ar
, MPa
0
0.01
-0.01
R=0
R = -2
0 Fit
, MPa
1000
Mean Strain
or R
50CrMo4 Steel
u = 1086 MPa
103
104
Nf , Cycles
105
106
100
10
102
Mean Strain
or R
103
50CrMo4 Steel
u = 1086 MPa
104
105
106
Nf , Cycles
(d) SWT
1000
ar
100
10
102
, MPa
0
0.01
-0.01
R=0
R = -2
0 Fit
0
0.01
-0.01 Mean Strain
R = 0 or R
R = -2
0 Fit
ar
, MPa
1000
Mean Strain
or R
103
50CrMo4 Steel
u = 1086 MPa
104
105
106
Nf , Cycles
100
10
102
103
50CrMo4 Steel
u = 1086 MPa
104
105
106
Nf , Cycles
Fig. 9 Equivalent completely reversed stress amplitude versus life correlations for 50CrMo4 steel for the (a) Goodman, (b) Morrow fracture
strength, (c) Morrow intercept, and (d) SWT methods.
2% strain. This follows the logic of Topper and coworkers,14,26 whose work indicates that such overstrains
are useful in advancing the fatigue damage process, so that
the behaviour is similar to that during service loading with
occasional severe cycles. These cases are identified by a
footnote in Table 1.
Additional data sets were located, as in MIL-HDBK5H, but were not included, as they covered only a narrow range of N f values or included only a small amount
of data for non-zero mean stresses. The data on hotrolled and normalized SAE 1045 steel of Fatemi27 were
analysed, but the results are not reported here. Relaxation during these strain-controlled tests resulted in quite
small mean stress values, which hardly exceeded the scatter in the zero mean stress data, causing difficulty with
the Walker fit. As reported by Fatemi, this set of data is
correlated well by the SWT equation. Also analysed but
not included are the data on Ti-6Al-4V generated in a
U.S. Air Force program.28 In this extensive data set, the
stress-life curve begins to flatten around N f = 3 105
cycles and the data extend to 107 cycles. It was omitted only because a more complex form of stress-life
equation would be needed than for the other data
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
171
1000
1000
-1
Approx. R
0.45
MPa
0.05
0.45
-1 Fit
(a) Goodman
2014-T6 Al
100
103
-0.37
0.05
-1 Fit
ar N
-1
ar ,
MPa
ar P
ar ,
Approx. R
-0.37
104
105
Nf , Cycles
106
107
1000
100
103
104
105
Nf , Cycles
106
107
1000
-1
Approx. R
-1
Approx. R
-0.37
MPa
0.45
MPa
ar ,
-1 Fit
ar ,
0.05
0.05
0.45
-1 Fit
(d) SWT
2014-T6 Al
104
-0.37
105
Nf , Cycles
106
107
100
103
104
105
Nf , Cycles
106
107
Fig. 10 Equivalent completely reversed stress amplitude versus life correlations for 2014-T6 aluminium for the (a) Goodman, (b) Morrow
fracture strength, (c) Morrow intercept and (d) SWT methods.
1000
Walker
Approx. R
ar
ar ,
, MPa
0
0.01
-0.01
R=0
R = -2
0 Fit
MPa
1000
Mean Strain
or R
50CrMo4 Steel
u = 1086 MPa
100
10
102
103
10
04
10
05
106
Nf , Cycles
Fig. 11 Equivalent completely reversed stress amplitude versus life
for the Walker method applied to 50CrMo4 steel, for which =
0.778.
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
-1
-0.37
0.05
0.45
Fit
Walker
2014-T6 Al
100
10 3
10
04
105
Nf , Cycles
10 6
107
172
N . E . D O W L I N G et al.
or in Smith et al.30 The missing values were then estimated by multiplying the known one by the ratio of
the ultimate tensile strengths u for the two batches of
material.
u2
fB2 = fB1
.
(15)
u1
However, for 6061-T6 aluminium, no value could be
found in the literature, and so one was obtained by conducting three tension tests on this alloy, averaging the
resulting properties, and then applying Eq. (15).
For steels where fB was not given with the data, the
following estimate was employed:
fB = u + 345 MPa.
(16)
arP
arP arN
=
1.
arN
arN
(17)
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
173
Table 3 Values of stress deviation for each material and mean stress equation
Number of data points
s z , Stress deviation
m 0
Goodman
Morrow
fB
Morrow
f
SWT
Sourcea
Number on
bar chartb
Total
Material
Walker
Fit
Walker
Est.
SAE 1015 St
GSMnNi63 St
Ck45 St
SAE 4130 St, Norm
49MnVS3 St
17MnCrMo33 St
CC 450 SS, H1150
CC 450 SS, Soln An
50CrMo4 St, El Pol
42CrMo4 St
AISI 4340 St
SAE 4130 St, Hard
CC 450 SS, H900
PH138Mo SS, H1000
300M St
SAE 1045 St, 705 HB
SAE 1045 St, 55 HRC
SAE 1045 St, 595 HB
Steel Average
2, 3
4
4
5, 6
4
4
7
7
4
4
8
5, 6
7
9
9
10
11
10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
32
31
33
65
30
70
25
25
25
29
25
33
28
17
71
13
28
16
12
14
21
19
16
23
11
9
11
19
10
18
9
6
27
8
13
9
0.142
0.205
0.098
0.382
0.129
0.269
0.069
0.224
0.248
0.237
0.236
0.170
0.159
0.525
0.178
0.091
0.139
0.090
0.200
0.080
0.129
0.082
0.217
0.108
0.079
0.044
0.124
0.104
0.120
0.077
0.111
0.087
0.251
0.145
0.083
0.096
0.068
0.111
0.075
0.147
0.082
0.211
0.102
0.109
0.044
0.139
0.114
0.133
0.070
0.111
0.100
0.061
0.233
0.055
0.148
0.084
0.112
0.118
0.195
0.094
0.220
0.126
0.077
0.038
0.086
0.143
0.146
0.118
0.096
0.076
0.108
0.123
0.059
0.084
0.105
0.112
0.069
0.116
0.069
0.107
0.061
0.061
0.042
0.111
0.041
0.047
0.048
0.103
0.077
0.046
0.099
0.052
0.086
0.070
0.073
0.070
0.133
0.080
0.134
0.097
0.093
0.040
0.080
0.081
0.082
0.062
0.096
0.105
0.067
0.119
0.060
0.069
0.120
0.088
6061-T6 Al
Al Mg4.5Mn, Cld Rl
2024-T4 Al
2024-T3 Al
2014-T6 Al
2024-T3 Al
2024-T4 Al
7075-T6 Al
7075-T6 Al
Aluminum Average
9
4, 12
13, 14
4
15
5, 6
15
15
5, 6
1
2
3
4
5
6
7
8
9
50
22
45
22
32
68
37
34
112
13
11
7
6
6
14
12
9
49
0.518
0.655
0.205
0.344
0.265
0.281
0.467
0.243
0.439
0.380
0.179
0.218
0.130
0.120
0.101
0.118
0.132
0.203
0.172
0.152
0.147
0.191
0.233
0.266
0.199
0.252
0.249
0.394
0.327
0.251
0.222
0.188
0.061
0.124
0.091
0.083
0.107
0.150
0.165
0.132
0.117
0.078
0.052
0.105
0.066
0.081
0.067
0.102
0.156
0.091
Ti-6Al-4V
10
21
14
0.283
0.049
0.267
0.096
0.064
a Numbers
b Numbers
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
174
N . E . D O W L I N G et al.
Fig. 13 Stress deviation values s z for steels for the (a) Goodman, (b) Morrow fracture strength, (c) Morrow intercept, (d) SWT and (e)
Walker methods. Also, s z values are given in (f) for Walker values estimated from ultimate tensile strengths. The numbers on the
horizontal axes identify the steels listed in Table 3.
eter . The average s z values are 0.07 for steels and 0.09
for aluminium alloys. Noting these values and looking at
the bar charts of Figs 13 and 14, as well as the typical life
correlations of Figs 11 and 12, it is clear that this method
gives the best results of all those considered. This might
be expected, as the ability to vary a fitting parameter is
not present for any of the other methods. On normalized
amplitudemean plots similar to Figs 68, adjusting is
seen to allow varying the resulting curve to fit data in cases
where the SWT method is not very accurate. Moreover,
on such a plot, data tend to correlate rather tightly about
a curve that fits a particular value.
Considering the single titanium alloy studied, it is of
course not possible to draw general conclusions for titanium alloys from one set of data. Nevertheless, it is notable
that the behaviour was generally similar to that for aluminium alloys, with poor correlation for the Goodman
and Morrow f methods, and good or excellent agreement for the remaining three methods.
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
175
(19)
Hence, the fatigue life deviates from the zero mean stress
(or Walker-fitted) line by a factor of 2.59. Of course,
when the above calculation is applied to a z value, the
result characterizes the error in life for only an individual
data point, and when applied to an s z value, the result
characterizes the overall error in life for the set of data.
For the typical b = 0.10, some additional values of the
life factor are given below.
z or s z
Life factor
0.072
2.00
0.10
2.59
0.20
6.19
0.30
13.8
0.50
57.7
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
N . E . D O W L I N G et al.
176
Walker, Estimated
0.8
1000
0.6
, MPa
Steels
ar
0
0.01
-0.01
R=0
R = -2
0 Fit
0.4
= 0.0002000
0.2
+ 0.8818
100
10
0.0
0
500
1000
1500
2000
u , Ultimate Tensile Strength, MPa
(u in MPa).
Mean Strain
or R
1 02
(20)
50CrMo4 Steel
u = 1086 MPa
104
103
Nf , Cycles
2500
105
106
1.0
Aluminum alloys
0.8
0.651
0.6
6061
Al Mg4.5Mn
0.473
0.4
2014
2024
7075
0.2
0.0
300
350
u,
400
450
500
550
Ultimate Tensile Strength, MPa
600
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
f
E
(2N f )b + f (2N f )c .
(21)
f
E
(2N )b + f (2N )c .
(22)
The quantity N is an equivalent life value corresponding to the zero mean stress case, that is, N is simply
the life calculated from a as if the mean stress were
zero. An expression for N can be developed for any
particular ar = f (a , m ).
Because the Walker mean stress relationship has been
shown to provide better correlation with test data than
other common methods, its incorporation into the strainlife curve is of particular interest. In this case, Eq. (6b)
leads to the particular N value for the Walker method,
here denoted N w .
1 R (1 )/b
Nw = N f
(a)
2
1 R (1 )/b
(23)
(b).
N f = Nw
2
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
0.1
a , Strain Amplitude
177
0.01
= 0.650
0.001
102
103
104
105
106
N*w , Walker Equivalent Cycles
107
108
178
N . E . D O W L I N G et al.
6 For steels, the correlation of with u of Eq. (20) allows to be estimated with sufficient accuracy to still
give excellent results for this class of material. Relatively
high strength aluminium alloys have 0.5, which corresponds with the SWT method, but higher values of
apply for relatively low strength aluminium alloys.
7 For both steels and aluminium alloys, there is a trend
of decreasing with increasing strength, indicating an
increasing sensitivity to mean stress, and also more brittle
behaviour that is increasingly dominated by the maximum
stress max , and less affected by the stress amplitude a .
Acknowledgements
The latter portions of this data analysis study that spanned
several years were supported by the U.S. Naval Air Systems Command, Patuxent River, MD. Technical direction was provided by Nam D. Phan of NAVAIR Structures, with the aid of Trung T. Nguyen. Project administration was handled by Pamela F. Bowen of Integrated
Systems Solutions, Inc., California, MD. Gratitude is expressed to these individuals and organizations for their
able assistance.
Institutional support was provided by the Engineering
Science and Mechanics Department, and by the Materials Science and Engineering Department, Virginia Tech,
Blacksburg, VA. We are especially indebted to staff member Lisa L. Smith of the ESM Department, who was helpful in a variety of ways.
The authors wish to thank several individuals who
aided with useful advice, comments, or data, including:
Jamie T.P. de Castro (Pontifical Catholic University,
Rio de Janeiro, Brazil); Ali Fatemi (The University of
Toledo, Toledo, OH); Nagaraja Iyyer (Technical Data
Analysis, Inc., Falls Church, VA); Chih-Kuang Lin
(National Central University, Chung-Li, Taiwan) and
Surot Thangjitham (Virginia Tech, Blacksburg, VA.).
REFERENCES
1
4
5
10
11
12
13
14
15
16
17
18
19
20
21
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
22
23
24
25
26
27
c 2009 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 32, 163179
28
29
30
31
32
179