Engineering Failure Analysis 18 (2011) 1711–1722

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Experimental studies into short crack growth

R. Jones a,b,d,⇑, S. Barter c, F. Chen a
a
DSTO Centre of Expertise in Structural Mechanics, Department of Mechanical and Aerospace Engineering, Monash University, P.O. Box 31, Victoria 3800, Australia
b
CRC for Infrastructure and Engineering Asset Management, Department of Mechanical Engineering, Monash University, P.O. Box 31, Victoria 3800, Australia
c
Air Vehicles Division, Defence Science and Technology Organisation, 506 Lorimer St., Fishermans Bend, Victoria, Australia
d
CRC for Rail Innovation, Department of Mechanical and Aerospace Engineering, Monash University, PO Box 31,Victoria 3800, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper examines short crack growth in two quite different materials, viz: 7050-T7451
Available online 7 April 2011 aluminium alloy and a head hardened rail steel. The experimental data reveals that the so
called short crack effect associated with 7050-T7451 aluminium alloy arises as a conse-
Keywords: quence of attempting to relate da/dN to the range of the stress intensity factor (DK). We
Short cracks also find that, in both cases, cracking crack growth conforms to the Generalised Frost–
Similitude Dugdale model.
Frost–Dugdale
Ó 2011 Elsevier Ltd. All rights reserved.
Fatigue

1. Introduction

This paper arose from the study of short crack growth in two very different materials. The first case involved the growth of
short cracks in 7050-T7451 aluminium alloy. This study arose from an investigation into the crack length versus cycles data
presented in the compendium of F/A-18 fatigue crack growth data by Molent et al. [1]. This compendium examined more
than 350 different cracks mainly in 7050-T7451, but also in other 7000 series aluminium alloys, Mil Annealed Ti–6Al–4V
titanium, and AF1410 steel that arose in a variety of full scale fatigue tests and associated coupon tests. Cracking in Mil An-
nealed Ti–6Al–4V specimens tested under a representative F/A-18 flight spectrum was subsequently studied in [2]. On
examining the crack length versus cycles data presented in [1,2] it was found that the majority of the fatigue life was gen-
erally consumed in the short crack regime, i.e. in growing to a size of approximately 1 mm. As such understanding the
growth of short cracks was particularly important. It was also found [1] that in almost all cases there was a near linear rela-
tionship between the log of the crack length/depth and the number of load blocks/flight hours and that this relationship held
from a starting length of less than 100 lm to lengths in excess of 5 mm’s.
The second problem area studied was associated with the formation and the subsequent growth of small sub mm rail
squats [3]. Squats were first observed in Australia over 19 years ago and in February 1999 the problem was identified as
being among the top 6 high priority items [3] in railway engineering. As such characterising the growth of sub mm cracks
in head hardened rail steel is vital if we are to fully understanding this problem.
As part of the F/A-18 program undertaken by the Australian Defence Science and Technology Organisation [3,4] it was
found that, for initial defects that had a size of approximately 3 lm, the crack growth programs FASTRAN1 [6] and AFGROW
(footnote 1) [7] were unable to model this (near) linear relationship between the log of the crack depth and the number of load
blocks/flight hours. The need to develop a fracture mechanics based methodology that could accurately predict the growth of

⇑ Corresponding author at: DSTO Centre of Expertise in Structural Mechanics, Department of Mechanical and Aerospace Engineering, Monash University,
P.O. Box 31, Victoria 3800, Australia. Tel.: +61 398786265; fax: +61 399051825.
E-mail address: rhys.jones@eng.monash.edu.au (R. Jones).
1
Unmodified and uncalibrated.

1350-6307/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2011.03.012
1712 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

Nomenclature

a crack length
b fatigue strength exponent
c fatigue ductility exponent
C constant in the Paris fatigue crack growth law
C fatigue crack growth constant in the Generalised Frost–Dugdale model
da/dN increment in crack length per cycle
(da/dN)0 reflects both the fatigue threshold and the nature of the notch
EBV equivalent block variant
K stress intensity factor
Kmax maximum applied stress intensity factor in a block
DK stress intensity range
DKth threshold stress intensity range
Kc apparent cyclic fracture toughness
LEFM linear elastic fracture mechanics
N number of cycles to fail the first element
n0 cyclic strain hardening exponent
p exponent of the Kmax dependency in the Walker law
R stress ratio
q critical distance
a constant in the stress strain law
c generalised Frost–Dugdale fatigue crack growth equation exponent
r remote applied stress
r0f fatigue strength coefficient
e0f fatigue strain coefficient
ry material yield stress
r0 distance in front of the crack at which the stresses are evaluated

short cracks in 7050-T7451 aluminium alloy under representative flight load spectra, and yet still be consistent with constant
amplitude crack growth data, then led to the development of the Generalised Frost–Dugdale model [5,8–20] by the Monash–
DSTO Centre of Expertise in Structural Mechanics (CoE-SM) and DSTO. One form of this model is:

da=dN ¼ C   að1c=2Þ ðDK ð1pÞ K Pmax Þc  da=dN0 ¼ rcy C  að1c=2Þ ðDK ð1pÞ K Pmax =ry Þc  da=dN 0 ð1Þ

where a is the crack length, N is the number of cycles, da/dN is the increment in crack length per cycle, ry is the yield stress,
C⁄ and p are constants, DK (=Kmax  Kmin) is the stress intensity factor range, Kmax and Kmin are the maximum and minimum
values of the stress intensity factors in the cycle, the term (da/dN)0 reflects both the fatigue threshold (DKth) and the nature
of the defect/discontinuity from which cracking initiates and c is a constant, which for local stresses below the material’s
yield stress, is often taken as c = 3 [21,22]. In this approach the time to crack nucleation is considered to be insignificant.
The link between this formulation, which is referred to as the Generalised Frost–Dugdale law, and other non-similitude
based crack growth laws is discussed in [5,19,23–25] and in the Appendix where it is shown that the Generalised Frost–
Dugdale law can be derived from the local crack tip stress and strain fields regardless of whether the crack is assumed to
be blunt, as in [5], or sharp.
This formulation has been used to accurately represent cracking in 7050-T7451 aluminium alloy in a range of laboratory
tests [14], a DSTO F/A-18 Hornet centre barrel test [5,12,16], cracking under several representative Joint Strike Fighter load
spectra [17], a representative helicopter spectrum [20], cracking in a number of laboratory coupon and sub-component tests
involving a range of steels (including D6ac steel, a large cross-section of rail wheel steels, a 350 MPa Grade locomotive mild
steel, and a propriety steel that is widely used in rail freight rolling stock), Grade 1 Austempered Ductile Iron, a range of 2000
and 7000 series aluminium alloys, and Mil Annealed Ti–6AL–4V ([5,8,10–13,15–20]).
At this stage it should be noted that whilst the ability of the Paris-type crack growth laws to model the Region II growth of
large/long cracks is well documented it has long been known [26–30] that in Region I, crack growth can be a function of the
test geometry. This Region I dependency of the da/dN versus DK data on the specimen geometry and the test methodology
means that the similitude hypothesis, which forms the basis of algorithms that are based on Paris-like crack growth laws, can
not be assumed to be valid in Region I. In this context it will be shown that tests performed at DSTO, NASA and at the Mon-
ash–DSTO CoE-SM on the growth of both short and long cracks in 7050-T7451 confirm the breakdown of similitude in Region
I. Furthermore, these tests reveal that the so called short crack effect appears to arise as a consequence of attempting to re-
late da/dN to the range of the stress intensity factor (DK). Furthermore, the short crack tests presented in this paper show
that in each case crack growth conformed to the Generalised Frost–Dugdale model.
 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722 1713

2. A brief review of similitude based growth laws

It is commonly thought that da/dN can be related to DK, and/or the maximum stress intensity factor Kmax. This approach
was first suggested in 1961 by Paris, Gomez and Anderson [31], who related da/dN to the maximum stress intensity factor
Kmax. The work of Liu [32] subsequently implied that the crack growth was a function of the stress intensity factor range DK
(=Kmax  Kmin). A similar relationship was also proposed by Paris and Erdogan [33]. This led to the well known Paris equation:
da=dN ¼ C DK m ð2Þ
where C and m are experimentally obtained, and are considered to be a constant for a particular material and environment.
Over the years this relationship has continued to be modified to account for a variety of observations [34], including R ratio
(R = Kmin/Kmax), Kmax effects [35–37], crack tip plasticity e.g. Willenborg retardation models [38] and plastic wake-induced
crack closure [6,39,40]. The resulting crack growth laws are all based on the similitude hypothesis, viz:

Two different cracks growing in identical materials and thicknesses with the same stress intensity factor range DK and the same
Kmax, will grow at the same rate.

It is believed that for constant amplitude loading the relationship between da/dN and DK has three distinct regions, see
Fig. 1, with Region III being associated with rapid crack growth, tearing or static fracture modes. Region II, the ‘‘mid growth’’
range, is the region where the Paris equation, and its variants, is thought to hold. In Region I crack growth is slow and several
authors have introduced the concept of a fatigue threshold stress intensity factor range, DKth, beneath which cracks seem not
to grow [41].
It was originally thought that, for any given material and thickness, the da/dN versus DK relationship was unique. How-
ever, Pearson [42], at the Royal Aircraft Establishment Farnborough, revealed that this belief was false. In this work Pearson
[42] revealed that fatigue crack growth laws determined for macroscopic crack growth data could not be used to predict the
growth of small sub-millimetre cracks, and that the constants in the crack growth law were a function of the size of the
crack. He also stated that this inconsistency was not due to crack-tip plasticity effects. This finding essentially reflected
Frost’s earlier work [43] that revealed that the fatigue threshold was crack length dependent, a finding that subsequently
led to the development of the ‘‘Kitagawa diagram’’ [44].
Subsequent studies into cracking in 7075-T6 aluminium alloys [45] confirmed that short fatigue cracks can propagate at
rates faster than that of long cracks subjected to the same nominal DK, i.e. there is an apparent crack length dependency.
Moreover, it is now known that short fatigue cracks can grow at stress intensities well below the long crack DK threshold.
As a result when analysing the growth of short cracks it is common to use data that differs from the long crack da/dN versus
DK data [46].
More extensive reviews of fatigue crack growth are presented in [34,47–51] and for short crack growth by Suresh and
Ritchie [52] and by Miller [30], who suggested a comprehensive classification of short cracks viz: micro-structurally,
mechanically, physically and chemically short cracks. However, the NASA findings [26] that DKth can be a function of the
test geometry, which implies that similitude is not valid in Region I, and Miller’s conclusion [52] that:
‘‘perhaps more progress would be made if LEFM characterisation parameters were ignored, and only the basic da/dN ver-
sus a type data be analysed.’’

raises the questions of: how to determine which da/dN versus DK relationship to use; and how to determine the true fa-
tigue threshold DKth; or whether alternative non-similitude based approaches should be adopted.
These questions take on added importance for short cracks since we now know [30,34,42,52–57] that for short cracks K
dominance is lost and hence the similitude hypothesis, which underpins all Paris based approaches, is highly questionable
for short cracks.

Fig. 1. Schematic diagram of the relationship between long crack growth and AK (where KIC is the critical stress intensity factor) for constant amplitude
loading.
1714 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

In this context it should be noted that Schijve [34] remarked that:

‘‘Actually, it should be recognized that the K-concept for such small cracks in a crystalline material becomes questionable.
The plastic zone is a slip band and its size is not small compared to the crack length of the microcrack.’’

2.1. On the concept of K dominance and similitude

The Paris growth law and its subsequent variants are founded on the belief that the crack tip stress field is uniquely char-
acterised by the stress intensity factor. This belief arose as a result of the early papers by Griffiths [58] and Irwin [59]. How-
ever, in 1966 Sih [60] revealed that the Westergard solution [61] for a centre crack in a large panel contained an error. In [60]
it was shown that a constant term, commonly referred to as the T stress, was missing in the series of expansion for the local
stress field in the vicinity of the crack tip. Eftis [62] then used Sih’s corrected solution to reveal that both Griffiths’ and Irwin’s
analyses were in error. Eftis, Jones and Liebowitz [63] subsequently stated that:
‘‘The notion that only the leading term of the series expansion for stress, containing as it does the stress intensity factor
and the square root singular term, can adequately describe the state of stress about the ends of the crack is erroneous in
general.’’
Eftis et al. [63] also suggested that to evaluate crack growth and failure required the crack tip stress field to be evaluated
at a length scale ro in front of the crack. The concept of a characteristic length scale is moderately widely used in the assess-
ment of fatigue crack growth and fracture mechanics [63–67]. For long cracks growing in Mode I under uniaxial loading the
ratio r0/a tends to be quite small and, as a result, the error in using DK and Kmax to characterise the crack tip stress field is also
(generally) quite small. However, for short cracks with length scales of the order of 10 lm the ratio r0/a will not be small.
(Refs. [66,67] give values for r0 that ranged from 0.1 to 3.3 mm.) In this context it should be noted that [57] subsequently
confirmed the conclusions reached in [63] i.e. that for short cracks the crack tip stress and strain fields were a function of
p
both K and the T stress, which is proportional to K/ a and is defined as the non-singular component of the near tip stress
acting parallel to the crack (see [63]). As a result [57] concluded that K dominance is lost for small (short) cracks. Conse-
quently the similitude hypothesis is also invalid for small (short) cracks.
With this in mind it is shown that the crack growth data given in [11,26] reveals that, in Region I, not only is similitude
invalid for Mil Annealed Ti–6AL–4V titanium, but that the crack closure hypothesis is also at odds with the weak R ratio
dependency seen in Mil Annealed Ti–6Al–4V crack growth rate data. (At this point it should be noted that Forth, James, John-
ston and Newman [54] have explained that crack closure approaches are inappropriate for modelling crack growth in mate-
rials that have a weak R ratio dependency. Furthermore, a recent NASA-Sikorsky investigation [68] into cracking in the ST
direction in 7050-T7451 plate revealed that the da/dN versus DK relationship was essentially R ratio independent in both
Regions I and II.)
The conclusion that similitude cannot be assumed in Region I or for the growth of short cracks significantly degrades the
confidence that can be placed in the ability of standard FASTRAN and AFGROW and other similitude based computer pro-
grams, to address potential F/A-18 fleet lifting issues or the growth of small flaws in rail steels and highlights the need to
better understand and quantify the (Region I) growth of short cracks.

3. Short crack growth in 7050-T7451 aluminium alloy

Whilst it has previously been shown [8] that under constant amplitude loading the growth of short (sub mm) cracks in a
350 MPa grade locomotive steel conformed to the Generalised Frost–Dugdale crack growth law the question of whether the
short crack effect was due to attempting to force a relationship between da/dN and DK has not been adequately addressed. To
this end [19,69] presented crack growth data for a series of 7050-T7451 hourglass specimens for R = 0.1 and 0.7. Details of
the tests and the loading are given in Table 1, for more details see [69]. The resultant ‘‘short crack’’ da/dN versus DK data is
presented in Fig. 2 along with data obtained in a joint NASA-Sikorsky [68] study into cracking in 7050-T7451 that was gen-
erated from long cracks. This work used compact tension (CT) specimens tested under ASTM constant R load reducing tests,
p
with R ratio’s of 0.1 and 0.7, constant Kmax (=15 MPa m) tests, and a constant amplitude load test with R = 0.1. Fig. 2 also
includes data obtained by researchers at the Defence Science and Technology Organisation (DSTO) [70,71] using CT and mid-
dle crack tension (MT) specimens respectively, and the details of the specimens used and the associated R ratio’s associated
with these two test programs are given in Table 1.
From Fig. 2 we see that for da/dN < 108 m/cycle (approximately) the R = 0.1 short crack growth rates obtained in the
short crack test program given in [19] are significantly greater than those reported in [68], i.e. there is an apparent short
crack effect when da/dN is presented as a function of DK.
Armed with this experimental data set we were then able to investigate the apparent short crack effect. To this end Fig. 3
presents a plot of da/dN against a(1c/2)(K pmax DK ð1pÞ =ry Þc =ð1  K max =K c Þ, with c = 3 and p = 0.2 and ry = 460 MPa, and
p
Kc = 35 MPa m for the growth of cracks from the small surface discontinuities present in the short crack tests [19,69].
The values c and p were chosen to best collapse the data whilst the value of Kc used in this study for this particular plate
material was the value reported by Barter [72] for thick section 7050-T7451 plate. For the through the thickness cracks
p
tested by Sharp et al. [70] and Finney [71] a value of Kc = 47 MPa m was used. This value was taken from Sharp et al. [70].
 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722 1715

Table 1
Test descriptors, from [69].

Ref. Initial discontinuity Geometry and thickness Kc (MPa Vm) Test method
depth/length
Forth and Urban [68] 19.1 mm CT, 12.7 mm thick Kmax, R = 0.7 and load reduction (R = 0.1) tests as per
ASTM E647
Sharp et al. [70] 28 mm CT, 12.5 mm thick 47 Fixed load tests with R = 0.1, 0.5, 0.8
Finney [71] 6.6 mm MT, 9.95 mm thick 47 Fixed load test with R = 0.2
Short crack tests [69,70] 0.0076 mm (Test 2) 11 mm thick 35 Repeated blocks were applied
0.021 mm (Test 3) hourglass specimen
Tests 2 block: 5000 cycles at R = 0.1 & 1500 cycles at
R = 0.7 loads. Peak stress = 250 MPa
Test 3 block: 5000 cycles at R = 0.1 & 3000 cycles at
R = 0.7 loads. Peak stress = 225 MPa
For more details see [19]

Fig. 2. Comparison of the various da/dN versus DK test data, from [73].

Here we see that beneath a growth rate da/dN of approximately 106 m/cycle the R = 0.1 and 0.7 short crack data now
appear to (essentially) fall on the same straight line with a slope of approximately 5.1  104, see Fig. 3. We also see that
when presented in this fashion the relationship obtained for the short crack data, the data obtained by Sharp et al. [70],
for R = 0.1, 0.5 and 0.8, using ASTM CT specimens, and the data given by Finney [71], which was obtained for R = 0.2 using
a MT specimen, are very similar, i.e. in this case the ‘‘short crack’’ effect vanishes and both the short and long crack data fol-
lows Eq. (1) with the noted constants.
It thus follows that, in this region, short and long crack growth in 7050-T7451 associated with [19,69–71] conforms to the
Generalised Frost–Dugdale law, given above. Furthermore, this relationship holds over more than 3 orders of magnitude
7  1010 < da/dN < 1.0  106 m/cycle, see Fig. 3. Thus for 7050-T7451 aluminium it would appear that the so called
short crack effect arises as a consequence of attempting to force a relationship between da/dN and the range of the stress
intensity factor (DK).
This counter example to the similitude hypothesis disproves the similitude hypothesis for the Region I growth of both
long and short cracks. (It should be stressed this does not mean that similitude will never apply, only that it can’t be arbi-
trarily assumed to be valid.) In cases when similitude does not hold the experimental data [18,26] suggests that Kmax tests
may well give lower more conservative values for DKth than the ASTM recommended load reducing tests.

4. Material characterisation of short crack growth in a head hardened rail steel

Let next examine the growth of short fatigue cracks in a head hardened rail steel, noting that, as mentioned above, it has
previously been shown [8] that under constant amplitude loading the growth of short (sub mm) cracks in a 350 MPa grade
locomotive steel conformed to the Generalised Frost–Dugdale crack growth law. The rail steel used in this study, which was
1716 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

Fig. 3. Comparison of experimental test results, from [69].

provided by Rail Corp, was taken from the field in NSW. The rail was cut to create a number of 320 mm long by 39 mm wide
and 7.5 mm thick single edge notch tension (SENT) specimens with a 0.5 mm radius semi-circular edge notch on one side,
see Fig. 4. Specimen numbers 1–3 were subjected to a peak remote stress of 475 MPa with R = 0.15 and the resultant failure
surface associated with Specimen number 1 is shown in Fig. 5. Specimens 4–6 were subjected to peak remote stress of
550 MPa with R = 0.5. In each case the failure surfaces were similar, see Figs. 5 and 6.
The crack length history in the various specimens was then predicted by integrating Eq. (1) with da/dN0 set to 0.0 with the
value of the stress intensity factor, for this through the thickness crack, as given in [46]. Jones, Chen and Pitt [13] and Jones,
Peng and Pitt [8] have revealed that many rail steels have a value of c = 3. Consequently in this study it was assumed that
c = 3. As a result in Eq. (1) there were only three unknown parameters, i.e. C, p and Kc. Their value was chosen so as to fit the
p
experimental data associated with Specimen 1. This yielded a value of p = 0.85, C = 1.5  1014 and Kc = 120 MPa m. The
resultant predictions for the entire crack length histories associated with each of these tests are shown in Fig. 7 where, for
each of the R ratio’s, we see good agreement with the experimental measurements.
Having determined and validated the crack growth law for this particular head hardened steel we then predicted crack
growth in Specimens 2 and 3. In these cases crack growth was three dimensional. For Specimen number 2 the final crack
shape, i.e. just prior to failure, was a quarter elliptical edge flaw with an aspect ratio of approximately 2.8–1 and an initial
aspect ratio of approximately 1.5:1, see Fig. 8, that emanated from the 0.5 mm edge notch. The test on Specimen 3 was
stopped after approximately 98,000 cycles and the crack broken open. This revealed a small quarter elliptical flaw (at the
notch) with an aspect ratio of 1.5, see Fig. 9. Eq. (1) was used to predict the crack growth histories for both specimens,
see Fig. 10. (In this analysis we used the expressions for K, for a quarter elliptical edge crack emanating from a notch, given
in [46].) In both cases the predicted crack length histories are in good agreement with experimental measurements, see
Fig. 10.
As a result of this study it would appear that the growth of small sub mm cracks in this head hardened rail steel follows
the Generalised Frost–Dugdale law, i.e. Eq. (1), and that for this particular steel the growth of small sub mm cracks can be
represented by:
3
da=dN ¼ 1:5  1014 ðaÞ1=2 ðDK 0:15 K 0:85
max Þ =ð1  K max =K c Þ ð3Þ

Fig. 4. Geometry of the SENT test specimen.

 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722 1717

Fig. 5. A view of the failure surface associated with Specimen 1.

Fig. 6. A view of the failure surface associated with Specimen 6.

10
Crack Length (mm)

Spec 1 Side B
Spec 1 Side A
Predic Spec 1
Spec 4 Side A
Predict Spec 4
0.1
Spec 4 Side B
Spec 5 Side A
Spec 5 Side B
Spec 6 Side A
Spec 6 Side B
Predicted Spec 6
0.01
0 20,000 40,000 60,000 80,000 100,000
Cycles (Total)

Fig. 7. Measured and predicted crack length history for Specimens 1, 4, 5 and 6.
1718 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

Fig. 8. A view of the failure surface associated with Specimen 2.

Fig. 9. A view of the failure surface associated with Specimen 3.

10.000
Crack Length (mm)

1.000

0.100
Specimen 2
Predicted Spec 2
Specimen 3
Predicted Spec 3

0.010
0 100,000 200,000 300,000
Cycles (Total)

Fig. 10. Measured and predicted crack length histories for Specimens 2 and 3.

5. Conclusion

This report has attempted to review, enunciate progress and define problem areas associated with the use of the Gener-
alised Frost–Dugdale model for predicting (short) crack growth in F/A-18 Hornet metallic structural materials and rail steels.
 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722 1719

It has also attempted to briefly address related problem areas associated with characterising crack growth in Region I. As a
result of this study it has been shown that the science base underpinning similitude based crack growth models, in Region I,
can be at odds with material behaviour. Consequently, given that linear elastic fracture mechanics studies have shown that
for short cracks K dominance is lost, alternative crack growth formulations should be further investigated. Furthermore, the
experimental data suggest that for 7050-T7451 the short crack effect arises as a consequence of attempting to relate da/dN to
DK.

Appendix A

Jones et al. [5] have shown that if the crack tip is not sharp, i.e. it has a finite notch radius, then the Generalised Frost–
Dugdale law can be derived from first principles. In this section we will show that it also follows if the crack tip is sharp. To
this end let us consider the approach used by Glinka [73] to derive the Paris fatigue crack growth law. Glinka assumed that
the local stress/strain fields at the crack tip conformed to the HRR solution [74,75]. He also assumed that the material in front
of the crack followed a Ramberg–Osgood law, viz :
0
e=ey ¼ ðr=ry Þ þ aðr=ry Þn ðA:1Þ
where ey and ry are the yield strain and the yield stress respectively, and a and n0 are material constants. Glinka [73] also
assumed that:
1. Crack growth may be regarded as the increment of successive crack re-initiation over a critical distance q (see Fig. A1).
2. The number of cycles, N, for material failure can be calculated from Manson–Coffin’s strain/life relation.
3. The fatigue crack growth rate can thus be calculated as the ratio of q/N.

With these assumptions the increment in the strain per cycle, De, in the direction normal to the crack at a distance q in
front of the crack tip was given as:
! 1 ! n0
ðn0 þ1Þ ðn0 þ1Þ
 EDJ E DJ
De ¼ C 1 þ C2 ðA:2Þ
r2y pq r2y pq
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl}
Elastic Term Plastic Term

where J is the J integral. The increment in the stress was then expressed as:
! 1
ðn0 þ1Þ
E DJ
Dr ¼ C 3 ðA:3Þ
r2y pq
where C1, C2, and C3 are functions of strain hardening exponent n0 and the parameter a, see [73].
Near the crack tip, the elastic terms can be neglected, so that [73] approximated De as
! n0
ðn0 þ1Þ
 EDJ
De ¼ C 2 ðA:4Þ
r2y pq
From the Coffin–Manson relationship, we see that

De rf
0
¼ ð2N Þb þ e0f ð2N  Þc ðA:5Þ
2 E
where b, c, r0f , and r0f are material constants related to the fatigue performance of the material. Neglecting the elastic terms
we see that

Fig. A1. Idealised crack and the critical distance q.

1720 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

De
¼ e0f ð2N Þc ðA:6Þ
2
So that rewriting Eq. (A.6) in terms of N [73] obtained the relationship
!1c
1 De
N ¼ ðA:7Þ
2 2e0f

Substituting Eq. (A.4) into Eq. (A.7) then yielded

0   n0 11c
ðn0 þ1Þ
C 2 r2 pq EDJ
  !1c ! n0
  n0
 1B
B y
C
C 1 C2 1 cðn0 þ1Þ
EDJ cðn0 þ1Þ
N ¼ B C ¼ ðA:8Þ
2@ 2e0f A 2 2e0f r2y p q

which [73] expressed in the form:

  n0
 EDJ cðn0 þ1Þ
N ¼ C4 ðA:9Þ
q
where
  !1c ! n0
cðn0 þ1Þ
1 C2 1
C4 ¼ ðA:10Þ
2 2e0f r2y p
The average increment in the crack length per cycle da/dN can be now calculated as per [73], viz:
da q
¼ ðA:11Þ
dN N
If we consider the case when R = 0, we find that DJ = DK2 so that substituting Eq. (A.9) into Eq. (A.11) and collecting terms
[73] obtained the following equation for the crack growth rate, viz:
  n0
da q  q cðn0 þ1Þ 0
ð1þcðnn0 þ1ÞÞ 2n0
¼   n0
¼ C5q ¼ C5q ðDKÞcðn0 þ1Þ ðA:12Þ
dN DK 2 cðn0 þ1Þ DK 2
C4 q

If we define
2n0
m¼ ðA:13Þ
cðn0 þ 1Þ
we see that da/dN can be expressed as
da
¼ C 5 qð1 2 Þ ðDKÞm
m
ðA:14Þ
dN
where
!1c
1 C2 n0
C5 ¼ ¼2 ðr2y pÞcðn0 þ1Þ ðA:15Þ
C4 2e0f

If we assume that cracks do not grow if DK is lower than a threshold value DK th , we see that from Eqs. (A.3) and (A.4)
! n0
ðn0 þ1Þ
DK 2th
Deth ¼ C 2 ðA:16Þ
r2y pq
and
! 1
ðn0 þ1Þ
DK 2th
Drth ¼ C 3 ðA:17Þ
r2y pq
From Eq. (A.17) we find that
 ðn0 þ1Þ !
C3 DK 2th
q ¼ ¼ C 7 DK 2th ðA:18Þ
Drth r2y p
 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722 1721

where
 ðn0 þ1Þ !
C3 1
C7 ¼ ðA:19Þ
Drth r2y p
However Tanaka [76] has shown that threshold stress intensity range (DKth) is a function of the crack length, viz:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DK th ¼ k ða=ða þ ao Þ ðA:20Þ
where ao is an intrinsic crack length, and k is a function of the R ratio, for more details see Section 2.3.1 of the NASGRO user’s
manual [77]. For physically small cracks such that a << ao, this yields
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
DK th ¼ k ða=ao Þ ðA:21Þ
so that Eq. (A.18) becomes

q ¼ C 7 k2 ða=ao Þ ðA:22Þ
Substituting Eq. (A.22) into Eq. (A.14) we obtain:
da  ð1m2 Þ
ðDKÞm ¼ C 8 að1 2 Þ ðDKÞm
m
¼ C 5 C 7 k2 ða=ao Þ ðA:23Þ
dN
where
!ð1m2 Þ !1c " ðn0 þ1Þ ! #ð1m2 Þ
C 7 k2 C2   n0
C3 k2 1
2 cðn0 þ1Þ
C8 ¼ C5 ¼2 rp
y ðA:24Þ
ao 2e0f Drth r2y p ao
It is clear that Eq. (A.23) conforms to the Generalised Frost–Dugdale law. We thus see that for physically small cracks, the
Generalised Frost–Dugdale law can be derived either by assuming that the crack is blunt, as in [7], or if it is sharp and
the stress field is described by the HRR solution.

References

[1] Molent L, Sun Q, Green AJ. The F/A-18 fatigue crack growth data compendium. DSTO-TR-1677; February 2005.
[2] Zhuang W, Barter S, Molent L. Flight-by-flight fatigue crack growth life assessment. Int J Fatigue 2007;29:1647–57.
[3] Kerr M. Rail Services Australia Internal Report on Asset Data Collection and Analysis. Heavy Haul Workshop; March 1999.
[4] Molent L, Singh R, Woolsey J. A method for evaluation of in-service fatigue cracks. Eng Fail Anal 2005;12:13–24.
[5] Jones R, Molent L, Pitt S. Crack growth from small flaws. Int J Fatigue 2007;29:1658–67.
[6] Newman JC. FASTRAN-II – A fatigue crack growth structural analysis program. NASA Technical Memorandum 104159; February 1992.
[7] Harter JA. AFGROW users guide and technical manual, air vehicles directorate. Air Force Research Laboratory OH, AFRL-VA-WP-TR-2004-XXXX; June
2004.
[8] Jones R, Pitt S, Peng D. The Generalised Frost–Dugdale approach to modelling fatigue crack growth. Eng Fail Anal 2008;15:1130–49.
[9] Jones R, Molent L, Pitt S. Similitude and the Paris crack growth law. Int J Fatigue 2008;30:1873–80.
[10] Jones R, Molent L, Pitt S, Siores E. Recent developments in fatigue crack growth modelling. In: Proceedings 16th European conference on fracture.
Greece: Alexandropoulos; July 2006.
[11] Jones R, Farahmand B, Rodopoulos C. Fatigue crack growth discrepancies with stress ratio. Theor Appl Fract Mech 2009;51(1):1–10.
[12] Jones R, Peng D. Tools for assessing the damage tolerance of primary structural components. In: Farahmand B, editor. Virtual testing and predictive
modelling: fatigue and fracture mechanics allowables. Springer; 2009.
[13] Jones R, Chen B, Pitt S. Similitude: cracking in steels. Theor Appl Fract Mech 2007;48(2):161–8.
[14] Jones R, Wallbrink C, Pitt S, Molent L. A multi-scale approach to crack growth. In: Proceedings mesomechanics 2006: multiscale behaviour of materials
and structures: analytical, numerical and experimental simulation, Porto, Portugal, 19–22 July 2006.
[15] Jones R, Molent L, Krishnapillai K. An equivalent block method for computing fatigue crack growth. Int J Fatigue 2008;30:1529–42.
[16] Jones R, Pitt S, Peng D. An equivalent block approach to crack growth. In: Sih GC, editor. Multiscale fatigue crack initiation and propagation of
engineering materials: structural integrity and microstructural worthiness. Springer Press; June 2008. ISBN: 978-1-4020-8519.
[17] Molent L, Barter S, Jones R. Some practical implications of exponential crack growth. In: Sih GC, editor. Multiscale fatigue crack initiation and
propagation of engineering materials: structural integrity and microstructural worthiness. Springer Press; June 2008. ISBN: 978-1-4020-8519-2.
[18] Jones R, Forth SC. Cracking in D6ac steel. Theor Appl Fract Mech 2010;53(1):61–4.
[19] Jones R, Molent L. Critical review of the generalised Frost–Dugdale approach to crack growth in F/A-18 Hornet structural materials. DSTO-Research
Report 0350; March 2010.
[20] Tiong UH, Jones R. Damage tolerance analysis of a helicopter component. Int J Fatigue 2009;31(6):1046–53.
[21] Frost NE, Dugdale DS. The propagation of fatigue cracks in test specimens. J Mech Phys Solids 1958;6:92–110.
[22] Barter S, Molent L, Goldsmith N, Jones R. An experimental evaluation of fatigue crack growth. Eng Fail Anal 2005;12/1:99–128.
[23] Carpinteri A, Spagnoli A, Vantadori S. Size effect in S–N curves: a fractal approach to finite-life fatigue strength. Int J Fatigue 2009;31:927–33.
[24] Carpinteri A, Paggi M. A unified interpretation of the power laws in fatigue and the analytical correlations between cyclic properties of engineering
materials. Int J Fatigue 2009;31:1524–31.
[25] Carpinteri A, Paggi M. A unified interpretation for the interpretation of anomoulos scaling laws in fatigue and comparison with existing models. Int J
Fracture 2010;161:41–52.
[26] Forth SC. The purpose of generating fatigue crack growth threshold data. NASA Johnson Space Center. <http://ntrs.nasa.gov/>.
[27] Vecchio RS, Crompton JS, Hertzberg DRW. The influence of specimen geometry on near threshold fatigue crack growth. Fatigue Fract Eng Mater Struct
1987;10(4):333–42.
[28] Garr KR, Hresko GC. A size effect on the fatigue crack growth rate threshold of alloy 718. In: Newman JC, Piascik RS, editors. Fatigue crack growth
thresholds, endurance limits, and design, ASTM STP, vol. 1372. West Conshohocken, PA: American Society for Testing and Materials; 1999.
[29] Hutař P, Seitl S, Kruml T. Effect of specimen geometry on fatigue crack propagation in threshold region. In: Proceedings international conference on
fracture, Ottawa; 2009.
1722 R. Jones et al. / Engineering Failure Analysis 18 (2011) 1711–1722

[30] Miller KJ. The behaviour of short fatigue cracks and their initiation. Part I – a review of two recent books. Fatigue Fract Eng Mater Struct
1987;10(1):75–91.
[31] Paris PC, Gomez RE, Anderson WE. A rational analytic theory of fatigue. Trend Eng 1961;13/1:9–14.
[32] Liu HW. Fatigue crack propagation and applied stress range. ASME Trans J Basic Eng 1963;85D(1):116–22.
[33] Paris PC, Erdogan F. Critical analysis of crack growth propagation laws. ASME Trans J Basic Eng 1963;85D(4):528–34.
[34] Schijve J. Fatigue of structures and materials in the 20th century and the state of the art. Int J Fatigue 2003;25:679–702.
[35] Noroozi AH, Glinka G, Lambert S. A two parameter driving force for fatigue crack growth analysis. Int J Fatigue 2005;27:1277–96.
[36] Dinda S, Kujawski D. Correlation and prediction of fatigue crack growth for different R-ratios using Kmax and DK+ parameters. Eng Fract Mech
2004;71:1779–90.
[37] Sadananda K, Vasudevan AK. Analysis of fatigue crack closure and threshold. Fract Mech, edited by F. Erdogan, ASTM, STP; 1993. p. 484–501.
[38] Willenborg J, Engle RM, Wood H. A crack growth retardation model using an effective stress concept. American Rockwell Technical Report TFR 71-701,
LA, USA; 1971.
[39] Newman Jr JC. FASTRAN-II – a fatigue crack growth structural analysis program. NASA Technical Memorandum 104159, USA; February 1992.
[40] Elber W. The significance of fatigue crack closure. Damage Tolerance of Aircraft Structures, ASTM STP-486; 1971. p. 230–42.
[41] Suresh S. Fatigue of materials. UK: Cambridge University Press; 2001.
[42] Pearson S. Initiation of fatigue cracks in commercial aluminium alloys and the subsequent propagation of very short cracks. Eng Fract Mech
1975;7:235–47.
[43] Frost NE. The growth of fatigue cracks. In: Yokobori T, editor. Proceedings of the first international conference on fracture. Sendai: The Japanese Society
for Strength and Fracture of Materials; 1966. p. 1433–59.
[44] Kitagawa H, Takahashi S. Fracture mechanical approach to very small fatigue cracks and to the threshold. Trans Jpn Soc Mech Eng 1979;45:1289–303.
[45] Lankford J. The growth of small fatigue cracks in 7075-T6 aluminum. Fatigue Fract Eng Mater Struct 1982;5(3):233–48.
[46] Newman JC, Wu XR, Venneri SL, Li CG. Small-crack effects in high-strength aluminium alloys. NASA Reference Publication; May 1994.
[47] Schijve J. Fatigue crack growth under variable-amplitude loading, engineering fracture mechanics. Eng Fract Mech 1979:20–44.
[48] Schijve J. Fatigue damage in aircraft structures, not wanted, but tolerated? Int J Fatigue 2009;31(6):998–1011.
[49] Skorupa M. Load interaction effects during fatigue crack growth under variable amplitude loading—a literature review. Part II: qualitative
interpretation. Fatigue Fract Eng Mater Struct 1999;22:905–26.
[50] Maddox SJ. The effect of mean stress on fatigue crack propagation: a literature review. Int J Fracture 1975;11(3):389–408.
[51] Davidson DL. How fatigue cracks grow, interact with microstructure, and lose similitude;;. In: Piascik RS, Newman JC, Dowling NE, editors. Fatigue and
fracture mechanics: ASTM STP 1296, vol. 27. American Society for Testing and Materials; 1997. p. 287–300.
[52] Suresh S, Ritchie RO. Propagation of short cracks. Int Met Rev 1984;29(6):445–76.
[53] Miller KJ. The behaviour of short fatigue cracks and their initiation. Part II – a general summary. Fatigue Fract Eng Mater Struct 1987;10(1):75–91.
[54] Forth SC, James MA, Johnston WM, Newman Jr JC. Anomalous fatigue crack growth phenomena in high-strength steel. In: Proceedings int. congress on
fracture, Italy; 2005.
[55] Sih GC. Crack tip mechanics based on progressive damage of arrow: hierarchy of singularities and multiscale segments. Theor Appl Fract Mech
2009;51:11–32.
[56] Sih GC. Segmented multiscale approach by microscoping and telescoping in material science. In: Sih GC, editor. Multiscaling in molecular and
continuum mechanics: interaction of time and size from macro to nano. Springer; 2006. p. 259–89.
[57] Kardomateas GA, Carlson RL, Soediono AH, Schrage DP. Near tip stress and strain fields for short elastic cracks. Int J Fracture 1993;62:219–32.
[58] Griffith AA. The phenomena of rupture and flow in solids. Philos Trans Roy Soc A 1921;221:163–81.
[59] Irwin GR. Analysis of stress and strains near the end of a crack traversing a plate. J Appl Mech 1957;79:361–8.
[60] Sih GC. On the Westergaard method of crack analysis. Int J Fract Mech 1966;2:628.
[61] Westergaard HM. Bearing pressure and cracks. J Appl Mech 1939;6:A49–53.
[62] Eftis J. Load biaxiality and fracture: a two-sided history of complementing errors. Eng Fract Mech 1987;26:567–92.
[63] Eftis J, Jones DL, Liebowitz H. Load biaxiality and fracture: synthesis and summary. Eng Fract Mech 1990;36(4):537–74.
[64] Sih GC. Experimental fracture mechanics: strain energy density criterion. In: Sih GC, editor. Mechanics of fracture, vol. 7. Martinus Nijhoff; 1972.
[65] Taylor D. Geometrical effects in fatigue: a unifying theoretical model. Int J Fatigue 1999;21:413–20.
[66] Bellett D, Taylor D, Marco S, Mazzeo E, Guillois J, Pircher T. The fatigue behaviour of three-dimensional stress concentrations. Int J Fatigue
2005;27:207–21.
[67] Ritchie RO, Knott JF, Rice JR. On the relationship between critical tensile stress and fracture toughness in mild steel. J Mech Phys Solids
1973;21:395–410.
[68] Forth SC, Urban MR. Fatigue crack growth thresholds for 7050-T7451 aluminum. NASA Johnson Space Centre, Houston, Texas, Private communication;
April 2009.
[69] Chen F, Jones R, Barter SA. Crack growth data for 7050-T7451, work in progress, unpublished data.
[70] Sharp PK, Byrnes R, Clark G. Examination of 7050 fatigue crack growth data and its effect on life prediction. DSTO-TN-0729. Australia; 1998.
[71] Finney J. Centre cracked panel tests. Private Communication; 1992.
[72] Barter SA. Fatigue crack growth in 7050-T7451 aluminium alloy thick section plate with a surface condition simulating some regions of the F/A-18
structure. DSTO-TR-1458; July 2003.
[73] Glinka G. A cumulative model of fatigue crack growth. Int J Fatigue 1982:59–67.
[74] Rice JR, Rosengren GF. Plane strain deformation near a crack tip in a power-law hardening material. J Mech Phys Solids 1968;16(1):1–12.
[75] Hutchinson JW. Singular behaviour at the end of a tensile crack in a hardening material. J Mech Phys Solids 1968;16(1):13–31.
[76] Tanaka K, Nakai Y, Yamashita M. Fatigue growth threshold of small cracks. Int J Fracture 1981;17(5):519–33.
[77] Fatigue crack growth computer program ‘‘NASGRO’’ Version 3.0: reference manual, JSC-22267b; March 2002.