International Journal of Fatigue 119 (2019) 150–159

Contents lists available at ScienceDirect

International Journal of Fatigue

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

Analytical theory for fatigue crack propagation rates of mixed-mode I–II T

cracks and its application
⁎
S. Qia, L.X. Caia, , C. Baoa, H. Chena, K.K. Shib, H.L. Wuc
a
Applied Mechanics and Structure Safety Key Laboratory of Sichuan Province, School of Mechanics and Engineering, Southwest JIAOTONG University, Chengdu 610031,
China
b
Science and Technology on Reactor System Design Technology Laboratory Nuclear Power Institute of China, Chengdu 610031, China
c
Shanghai Electric Power Plant Equipment Co., Ltd., Shanghai 200240, China

A R T I C LE I N FO A B S T R A C T

Keywords: The Manson–Coffin law for the low-cycle fatigue (LCF) of a representative volume element (RVE) and fatigue
Mixed-mode I–II cracks crack propagation (FCP) rate of cracked components can characterize material fatigue resistance. This paper
Low-cycle fatigue proposes an LCF–FCP model for predicting the FCP rate of plane-stress mixed-mode I–II cracks. A semi-analytical
Fatigue crack propagation rate model is also proposed for the Je integral of mixed-mode I–II compact tension shear (CTS) specimens based on
Energy equivalence
energy equivalence method. The combination of these two models predicted the FCP rates of CTS specimens
30Cr2Ni4MoV
made from 30Cr2Ni4MoV rotor steel with mixed-mode I–II cracks and subjected to four different loading angles.
The theoretical results were consistent with the experimental results.

1. Introduction between the FCP behaviors of mixed-mode I–II cracks and the Man-
son–Coffin law for LCF behaviors.
Damage-tolerant design theories for stable cracks in crack-con- Any study of correlations between the FCP rates and LCF char-
taining components and controlling fatigue crack propagation (FCP) acteristics of mixed-mode I–II cracks should be supported by experi-
have been investigated since the 1970s. Over the last 30 years, strong mental data on the FCP of mixed-mode I–II cracks. With respect to
correlations have been found between low-cycle fatigue (LCF) char- experiments on the FCP of mixed-mode I–II cracks, Richard [9], Lal
acteristics and mode-I FCP behaviors of materials, leading to various [10], Mahanty [11], Buzzard [12], and Otsuka [13] have proposed 13
theoretical methods being proposed for this correlation [1–8]. In par- types of specimens to study this. The CTS specimens proposed by Ri-
ticular, Shi and Cai [7,8] developed a theoretical model for predicting chard [9] are widely used in experimental studies on the FCP behaviors
the FCP rate of mode-I cracks based on the Manson–Coffin law for of mixed-mode I–II cracks because of the simplicity of their experi-
uniaxial LCF that accounts for the damage correlations of a material’s mental loading procedures and the ease of observing crack propagation
representative volume element (RVE) by combining the cyclic stress–- in these specimens. Kim [14] and Peixoto et al. [15] have performed a
strain field ahead of a crack tip with the crack closure (CC) effect and large number of experiments on CTS specimens with various loading
fatigue failure criteria for the effective cyclic plastic zone. The predic- angles. In experiments on the FCP of mode I–II cracks for CTS speci-
tions of the Shi–Cai model [7,8] were compared with those of six other mens, it has been shown that the cracks do not propagate along the
theoretical models [1–6] by using experimental data on the uniaxial orientation of the original crack plane but rather propagate over a wide
LCF and FCP of more than 20 different metallic materials. The results range along straight lines at near-constant angles.
indicated that the Shi–Cai model exhibits high generalization and ex- The determination of crack propagation orientations is an important
cellent predictive accuracy. Hence, a high degree of maturity has been part of studies on the FCP behaviors of mixed-mode I–II cracks. Several
achieved in theoretical predictions for the FCP rate of mode-I cracks criteria have been proposed [16–19] to describe the orientation of crack
based on the Manson–Coffin law for LCF behaviors. However, the ma- propagation based on experimental studies on the FCP of mixed-mode
jority of FCP behavior studies focus on mode-I cracks, whereas the I–II cracks. These include Erdogan and Sih’s maximum tangential stress
cracks and defects in actual engineering components (e.g., pressure criterion [16], Sih’s minimum strain energy density criterion [17], Ri-
vessels, pipelines, and fan blades) tend to be plane-stress mixed-mode chard et al.’s criterion [18], and Dai et al.’s J-criterion [19]. The
I–II cracks. Thus far, there have been very few reports on correlations maximum tangential stress criterion is widely used as its expressions

⁎
Corresponding author.
E-mail address: [email protected] (L.X. Cai).

https://doi.org/10.1016/j.ijfatigue.2018.10.004
Received 31 July 2018; Received in revised form 6 October 2018; Accepted 8 October 2018
Available online 09 October 2018
0142-1123/ © 2018 Published by Elsevier Ltd.
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

are neat and concise, and its accuracy has been proven by a large

0
R1
0.32

5
number of experimental studies [20–23]. Although the minimum strain

10
energy density criterion is useful to an extent, there is no theoretical
basis for this criterion, as the minimum strain energy density factor is a
sum of the shape change energy and volume change energy, and 32 14 32
Theocaris and Andrianopoulos [24] have noted that two different
86
physical quantities should not be summable. For mixed-mode I–II
cracks that are dominated by mode-II cracks, Guo [25] demonstrated Fig. 1. Geometric dimensions of the LCF specimens.
that the minimum strain energy density criterion and the Richard cri-
terion [18] both predict significantly larger crack propagation angles
than the actual crack propagation angle. For center-cracked tensile Į
plate specimens dominated by mode-II loading, Gdoutous [26] noted P
that the J-criterion produces results that deviate significantly from ex-
perimental results. In summary, the maximum tangential stress cri-
terion is an effective method for determining the direction of propa- 15 15
gation of mixed-mode I–II cracks.
The accurate measurement of the crack propagation length is a
crucial aspect of FCP experiments. The most common methods for
measuring the crack propagation length include the direct current po-

a
tential drop (DCPD) method, the compliance method, and the digital
image correlation (DIC) technique. As a unified deterministic formula 60 84
for different materials and specimens do not exist for the DCPD method,
the equations of the DCPD method must be calibrated for each spe- a0
cimen; this method is therefore unsuitable for widespread usage. The
compliance method is widely used for the autonomous measurement of 10
crack length in experiments investigating the FCP of mode-I cracks.
However, a compliance equation for the measurement of mixed-mode
I–II cracks in real time is yet to be found even after almost 10 years of
research. The DIC technique, which is widely used, is also gaining po-
P
pularity for measuring crack length because of its non-contact nature
and ability to track full-field displacement information [27–32]. W 50
Under cyclic loading, a small region of plastic deformation appears
at the crack tip. Hence, the use of stress intensity factors (SIF) in the Fig. 2. Geometric dimensions of the CTS specimens and loading condition.
study of FCP behaviors is a reasonable choice. In a well-known study by
Paris [33] in 1963 that used a large amount of experimental data, it was
integral (the elastic component of the J-integral) of mixed-mode I–II
shown that SIF is related to the FCP rate of mode-I cracks during steady-
CTS specimens. These models were used to predict the FCP rate of
state crack propagation by a power law relationship. After the 1970s,
30Cr2Ni4MoV steel to validate the accuracy of the LCF–FCP model.
the J-integral [34–38] was used by numerous researchers around the
world to describe the FCP behaviors of mode-I cracks, and the J-integral
has also been used to study the FCP rate of small mixed-mode I–II 2. Research conditions
cracks [39]. In 1985, Richard [9] provided expressions for the SIF of
CTS specimens with different loading angles that are useful for pre- 2.1. Investigation of the FCP of mixed-mode I–II cracks
dicting the initiation of deep cracks with length (a0) to width (W) ratios
between 0.55 and 0.7. However, no theoretical equations are currently 2.1.1. Specimens and loading device
available for describing the SIF and Je (the elastic component of the J- The dimensions of the LCF and CTS specimens used for mixed-mode
integral) of FCP in mixed-mode I–II cracks or for the rate of energy I–II FCP rate measurements are shown in Figs. 1 and 2, respectively.
release during this process. The use of finite element analysis (FEA) is The width (W) and thickness (B) of the specimens were 50 mm and
therefore necessary for the derivation of these quantities. In 2017, Chen 5.26 mm, respectively. CTS specimens can be loaded from multiple
and Cai [40] proposed a semi-analytical energy equivalence model for angles. The values of the loading angle α that were used in this ex-
the energy, loading, displacement, and constitutive relations of dis- periment were 15°, 30°, 45°, and 60°. The crack propagation angle and
continuous solids. This model is based on the fact that the strain energy the initial and propagated lengths of the crack are denoted by β, a0, and
of the deformed region of a structural member is equivalent to the a, respectively. The a0/W ratios of the specimens were 0.36, 0.46, and
product of the strain energy density at the median point between the 0.56, respectively.
RVE and the volume of the deformed region and the von Mises A specialized loading device was used to couple each CTS specimen
equivalence. This implies that the RVE strain energy density at the during the mixed-mode I–II FCP experiment. The CTS specimen loader
energy center of the complex stress state is equivalent to the RVE strain shown in Fig. 3(a) was proposed by Richard [9]. The loading of the
energy density of the equivalent uniaxial stress state. specimen is achieved by connecting pins to the loading slots of the
This paper proposes an LCF–FCP model for predicting the FCP rate loading device. However, as gaps are present between the specimen and
of plane-stress mixed-mode I–II cracks considering the effective cyclic the walls of the loading slot, it is difficult to ensure that the loading is
stress–strain field ahead of a mixed-mode I–II crack tip under small- exactly in plane with the through-thickness symmetry plane of the
scale yielding conditions and consideration of the crack closure effect. specimen. This subsequently causes the crack to propagate asymme-
This model assumes that the elastic critical nominal energy density of trically along the opposing sides of the specimen, which could result in
an RVE under uniaxial LCF is equal to the nominal strain energy density significant errors in the measurement of crack propagation length.
of the crack tip plastic zone (CTPZ) of a mixed-mode I–II crack under To eliminate the gap between the loading device and specimen in
fatigue loading. Based on this semi-analytical energy equivalence the loading slot, we divided the Richard loading device [9] into two
model, we also propose a theoretical model for describing the Je parts along the through-thickness symmetry plane and then fastened

151
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

3 1
1

2
2

a˅ b)
Fig. 3. The CTS specimen and its grips.

these parts using bolts, as shown in Fig. 3(b). The improved loading was used to get a digital image of the specimen surface every 500 cy-
device was then connected to an MTS hydraulic chuck. The require- cles, and the number of cycles corresponding to each digital image was
ments of the fatigue tests were thus satisfied by the connections be- recorded as N. The position of the crack tip was defined as the co-
tween the specimen, grips, and MTS chuck. ordinate corresponding to the greatest value of strain in the region
As fatigue cracks have to be pre-made for an CTS specimen prior to where the crack-tip was being tracked. This was used to determine the
mixed-mode I–II FCP testing, sine-wave cyclic loading was performed propagation of the crack tip at 500-cycle intervals.
with a loading angle (α) of 0°, loading ratio (R) of 0.1, and frequency of
20 Hz, until a crack 1–2 mm long was formed. The same loading con- 2.1.3. Experimental materials and their properties
ditions were used in the FCP testing. The maximum load of the CTS The specimens were made from 30Cr2Ni4MoV rotor steel, with
specimens with a0/W fixing at 0.36 at each loading angle is 8500 N. The chemical composition in mass fractions of: C, ≤0.35%; Si, ≤0.10%;
maximum load of the CTS specimens with a0/W fixing at 0.46 at each Mn, 0.20–0.40%; Mo, 0.25–0.60%; Ni, 3.25–3.75%; Cr, 1.50–2.00%; V,
loading angle is 6000 N. The maximum load of the CTS specimens with 0.07–0.15%; P, ≤0.01%; and S, ≤0.005%. This material has a Young’s
a0/W fixing at 0.56 at each loading angle is 4500 N. modulus of elasticity of 202 GPa and a yield strength of 794 MPa. The
room-temperature LCF up-to-failure tests on 30Cr2Ni4MoV steel spe-
2.1.2. Test equipment cimens were conducted according to the ASTM-E602 standard [41].
An MTS 810 25-kN electro-hydraulic servo universal testing ma- The stress amplitude was calculated from the steady stress–strain hys-
chine was used alongside a Test Star II control system to perform the teresis loops of the LCF specimen. These stress amplitudes were sub-
LCF and FCP testing on mixed-mode I–II cracks. The strain ex- sequently used to determine the relationship between the cyclic stress
tensometer used to control LCF strain was an MTS632.53F-14, which amplitude and the strain amplitude for the specimen material, ac-
has a gage length of 12 mm, a range of ± 2%, and an accuracy class of cording to the following expression:
0.5. An MTS632.02F-20 crack opening displacement (COD) ex- Δσ / 2 Δσ / 2 1/ n c
⎧ +( )
tensometer, with a gage length of 5 mm and a range of 4 mm, was used E Kc
Δε /2 = Δσ / 2 σsc Δσ / 2 Nc
for the displacement test in the mixed-mode I–II FCP testing. A com- ⎨ Δεe/2 + Δεp/2 = + αc ( σ )
puter was used to control the experimental processes and to record ⎩ E E sc (1)
data. Here, Δε/2 is the cyclic strain amplitude, Δεe/2 the cyclic elastic
A Correlated Solutions (USA) Video Image Correlation in 3-D (VIC- strain amplitude; Δεp/2 the cyclic plastic strain amplitude; Δσ/2 the
3D) system was used to measure crack propagation length. Fig. 4 il- cyclic stress amplitude; E the Young’s elastic modulus; σsc and Kc are the
lustrates the experimental setup, which consisted of the MTS machine, cyclic yield stress and cyclic strain intensity coefficient, respectively;
the VIC-3D system, the CTS specimen and its grips. The VIC-3D system and Nc and nc are the cyclic stress intensity exponent and cyclic strain
intensity exponent, respectively, where Nc = 1/nc; and αc is the cyclic
MTS system strain intensity coefficient, and αc = E/(KcNc·σsc1-Nc). The stress ampli-
Hydraulic tude, plastic strain amplitude, and fatigue life of a material are de-
grip scribed by the Manson–Coffin law:
VIC-3D testing system
b
COD meter ⎧ Δσ /2 = σ ′f (2Nf )
⎪
Δεp/2 = ε′f (2Nf )c
⎨
CTS specimen ⎪ Δε /2 = ΔσE/ 2 + Δεp/2
⎩ (2)
CTS grip
The Mises strain Here, Δεe/2 is the cyclic elastic strain amplitude; Δεp/2 the cyclic
nephogram of CTS specimen plastic strain amplitude; σf’ and b are the material’s cyclic strength
coefficient and cyclic strength exponents, respectively; εf’ and c are the
material’s cyclic ductility coefficient and cyclic ductility exponent, re-
spectively; and Nf is the fatigue life of a smooth specimen. Δε/2 and Δσ/
2 are given by the steady-state hysteresis cycle of the fatigue specimen.
Major direction LCF experiments were performed on 16 specimens made from
30Cr2Ni4MoV steel, and the experimentally obtained parameters of
Eqs. (1) and (2) of the LCF model are shown in Table 1. Fig. 5 illustrates
Fig. 4. Test system. the experimentally obtained relationship curves between Δε/2 and 2Nf,

152
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

Table 1
The LCF model parameters in Eqs. (1) and (2).
Material Kc/MPa nc σ'f /MPa b ε'f c σsc/MPa

30Cr2Ni4MoV 995.4 0.07404 869.7 −0.03580 2.560 −0.8619 635.0

0.1
Here, σij is the stress tensor; α’ the stress intensity coefficient; N the
stress intensity exponent; IN the dimensionless constant, which is de-
pendent on N; εs the yield strain of the material, where εs = σs/E and σs
is the yield stress of the material; J the J-integral of mixed-mode I–II
cracks; r and θ are the polar radius and angle, respectively, with the
0.01
-1

crack tip being the origin; and ∼

/2 /mm.mm

σij is the dimensionless constant related

to θ, N, and the plastic mixity parameter, Mp, in the plane-stress state,
expressed as follows

2 σθθ |r , θ = 0
1E-3 /2-2Nf
e
Mp = tan−1 lim
π r → 0 σrθ |r , θ = 0 (4)
p
/2-2Nf Test
/2-2Nf In this equation, σrθ and σθθ are stress tensor components in polar co-
ordinates.
Prediction of Manson-Coffin
Shih [42] noted that IN and Mp are derived from the relationship
1E-4
100 1000 10000 100000 curves between IN, Mp, and N and the relationship curve between Mp,
2Nf /Cycle
the elastic mixity parameter Me, and N, with Me expressed as
Fig. 5. LCF curves of 30Cr2Ni4MoV steel. σθθ |r , θ = 0
2 2 KI
Me = tan−1 lim = tan−1
π r →∞ σrθ |r , θ = 0 π KII (5)
Δεe/2 and 2Nf, and Δεp/2 and 2Nf, as well as the relationship between
Δε/2 and 2Nf predicted by the Manson–Coffin model. A high degree of where KI and KII are the tensile and shear SIFs of mixed-mode I–II
consistency was observed between these results. The parameters of Eqs. cracks, respectively. Under small-scale yielding conditions, the von
(1) and (2) of the LCF model for 30Cr2Ni4MoV steel constitute the basic Mises equivalent stress, σe, of mixed-mode I–II cracks may be expressed
data for validating the LCF–FCP model proposed in this paper. as
1
J N+1
⎤ ∼
2.2. Finite element method for the analysis of CTS specimens σe = σs ⎡ σe (θ , N , Mp)
⎢
⎣ α ′ε s σs IN r ⎥
⎦ (6)
FCP experiments were performed on CTS specimens made from where ∼ σe is a dimensionless constant that is dependent on θ, N, Mp, and
30Cr2Ni4MoV steel at various loading angles to determine the re- the planar-stress state, the value of which may be derived from the
lationship between the FCP rate of mixed-mode I–II cracks and the relationship curves between ∼ σe and θ, N, and Mp that were obtained by
value of the J-integral. The results were used to validate the effective- Shih [42] using a large number of finite element analyses. If the crack
ness of the LCF–FCP model proposed in a later section of this paper. An tip zone is in the small-scale yielding regime, J is equal to its elastic
FEA was performed in ANSYS to characterize the J-integral of the CTS component, Je.
specimens semi-analytically. If the components of plastic strain tensor remain constant propor-
Before using the FEA simulation, we compared the calculation re- tion to one another at each point of the plastic strain region as in the
sults under different mesh densities and selected the appropriate ele- case of the total strain theory, this permits a general treatment of the
ment size in this paper. A linear elastic FEA model was constructed for response to cyclic loading through the plastic superposition method
CTS specimens under plane-stress conditions, as shown in Fig. 6. developed by Rice [44]. According to the plastic superposition theory
Plane183 elements were used for the planar model, which contained proposed by Rice [44], a description for the cyclic equivalent stress
88,393 elements. The elastic modulus (E) and Poisson’s ratio (v) of the field of mixed-mode I–II crack tips can be obtained from the expression
material were 202 GPa and 0.3, respectively. A rigid beam constraint for Eq. (6) for the uniaxial loading condition
was used to achieve the equivalent of pin-loading of the specimen. 1
Fig. 7 shows the relationships between the load, P, and the crack E ΔJ ⎤Nc + 1 ∼
Δσe = 2σsc ⎡ σec (θ , Nc, Mp)
mouth opening displacement (CMOD), V, which were obtained via FEA ⎢ 2α c σsc2 IN c r ⎥ (7)
⎣ ⎦
and under experimental conditions. The FEA conditions (grid model,
boundary conditions, and material properties) and experimental con- Here, Δσe is the cyclic equivalent stress range; ΔJ the J-integral range
ditions (specimen, loading device, and testing machine) were consistent for mixed-mode I–II cracks; Nc the cyclic stress intensity exponent in Eq.
with each other. Hence, the characterization of the J-integral by the (1); and INc the dimensionless constant that is dependent on Nc. When
finite element method could be used to analyze the FCP data obtained the crack tip region is in the small-scale yielding regime, ΔJ is
from the CTS specimens. equivalent to its elastic component, ΔJe.
Based on consideration of the crack closure effect in cyclic loading,
3. Analytical LCF–FCP model the effective stress field of the crack tip in a mixed-mode I–II cracks is
described as
In 1988, Symington, Shih, and Ortiz proposed a crack tip stress field 1
ME ΔJ ⎤Nc + 1 ∼
for mixed-mode I–II cracks under plane-stress conditions for homo- Δσe = 2σsc ⎡ σec (θ , Nc, Mp)
⎢ 2α c σsc2 IN c r ⎥ (8)
geneous, isotropic, and ductile power-law materials [42,43] ⎣ ⎦

J
1/(1 + N ) where M is the crack closure factor of mixed-mode I–II cracks, it is a
σij = σs ⎛ ⎞ ∼
σij (θ , N , Mp)
⎜ ⎟
dimensionless quantity.
′
⎝ s s IN r ⎠
α σ ε (3)
According to Ref. [43], the relationship between the size of the

153
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

ELEMENTS
U SEP 18 2018
ROT 19:58:22
F

Y
Y

Z
X
Z X

model

Fig. 6. Finite element net model of CTS specimen.

0.25 E
Loading angle=0deg ρc = M ΔJth
4πσsc 2 (10)
Loading angle=15deg
Ansys
Crack mouth opening displacement V /mm

0.20
Loading angle=45deg Here, ΔJth is the threshold of the J -integral range of mode-I cracks
Loading angle=60deg [45].
On average, the magnitudes of plastic strains in mixed-mode I–II
Loading angle=0deg
0.15 Loading angle=15deg
CTPZs are far greater than those of elastic strains. The contribution of
Test elastic strain to the FCP of mixed-mode I–II cracks is therefore negli-
Loading angle=45deg
Loading angle=60deg gible. The the product of the cyclic effective plastic strain range and
0.10 cyclic effective stress range can thus be obtained from the Eq. (1) as
following expression,
Nc + 1
σsc Δσe /2 Nc 4σ 2 α c Δσ
0.05 Δσe Δεpe = Δσe [2α c ( ) ] = sc ⎛ e ⎞ ⎜ ⎟

E σsc E ⎝ 2σsc ⎠ (11)

In this equation, Δεpe is the cyclic equivalent plastic strain amplitude.

0.00 To obtain the average nominal strain energy density up of the RVE of
1000 2000 3000 4000 5000 6000
Load P /N crack surfaces in the CTPZ of mixed-mode I–II cracks, the integral
average of ΔσeΔεpe may be taken in the effective plastic zone of the
Fig. 7. P versus V curves given by FEA and tests.
crack surface, [ρc, rpc],
2α
Δσe Nc + 1
CTPZ of mixed-mode I–II cracks, rpc, and ΔJ based on the von Mises
up =
r
∫ρcpc Δσe Δεpe dr
=
∫ρcpc
r 4σsc
E
c
( )
2σsc
dr
yield criterion is
rpc−ρc rpc−ρc (12)
E Because the material RVE of the CTPZ is controlled by the magni-
rpc = M ΔJ
4πσsc 2 (9) tude of cyclic strain, the propagation of the CTPZ of mixed-mode I–II
cracks is equivalent to mechanical failure in the material’s LCF beha-
In this equation, rpc is the size of the cyclic plastic zone. viors. Therefore, the Manson–Coffin critical damage model for de-
Crack propagation will not occur if the crack tip driving force of scribing the LCF behaviors of a material may be used to describe the
mixed-mode I–II cracks, ΔJ, is lower than the threshold for the J-in- failure of the RVE in the CTPZ of mixed-mode I–II cracks. Based on Eq.
tegral range, ΔJth. Because of the existence of crack tip passivation and (2), the equivalent critical energy density of the material RVE when
plasticity-induced crack closure, the passivation of the crack tip is re- subjected to uniaxial LCF is
duced when the critical point for crack propagation is reached. To ac-
count for this effect, a correction for crack closure of the same magni- u p = Δσ·Δεp = 4σ ′f ε′f (2Nf )b + c (13)
tude as that of Eq. (8) is applied to the crack tip passivation radius, ρc.
Because FCP is the result of repeated surge-like crack-tip expan-
The critical passivation radius is then
sions, if we suppose that each propagation length is the length of the
effective plastic zone, rpc-ρc, N* then corresponds to the number of

154
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

cycles required for the crack to propagate over the propagation length. Table 2
Combining Eqs. (12) and (13), N* is Parameters of the Je0 model.
1 kij j=1 j=2 j=3 j=4 j=5
1 ⎡ M·ΔJ rpc b + c
N∗ = · σ 1 + Nc (θ , Nc, Mp)·ln ⎤
·∼ −14.1041 −0.232837 −0.000188000
2 ⎢ 2σ ′f ε′f (rpc−ρc ) IN c e ρc ⎥ (14)
i=1 16.8497
⎣ ⎦ i=2 14.0317 0.498318 −119101 −0.429513 0.000155000
i=3 −2.11915 −8.87107 0.156761
The FCP rate of mixed-mode I–II cracks, da/dN, may be expressed as
i=4 −4843.63 1077.16 7.65657 1.99893 0.163779
the ratio between rpc-ρc and N*. The analytical model for predicting the
FCP rate of mixed-mode I–II cracks according to Eqs. (9), (10) and (14)
is therefore: 700

1 Calculated results Prediction by Equation (25)

da EM ΔJth (λJ −1) ⎡ σ ′f ε′f EIN c (λJ −1) ∼−(1 + Nc) b+c
600 by EFA
= · σe (θ , Nc, Mp ) ⎤
dN 2πσsc 2 ⎢ 2πσsc 2λJ ln λJ ⎥ (15) 0 0 0 0
⎣ ⎦ =10 =15 =10 =15
0 0 0 0
500 =20 =25 =20 =25
where, λJ is the dimensionless driving force for the propagation of =30
0
=35
0
=30
0
=35
0

0 0 0 0
mode I/II cracks, and λJ = ΔJ/ΔJth and is independent of the crack 400
=40
0
=45
0
=40
0
=45
0
=50 =55

2
=50 =55

2WB EJe0/P
closure effect. =60
0
=65
0
=60
0
=65
0

0 0 0 0
=70 =75 =70 =75

2
300 0 0 0 0
=80 =85 =80 =85
4. A model for predicting J-integral (Je) of CTS specimens =89
0
=89
0

200
For the maximum tangential stress criterion proposed by Erdogan
and Sih [16] in 1963, it was assumed that the crack propagates along 100
the direction of the maximum tangential stress and that crack propa-
gation occurs when the circumferential stress in this direction reaches a 0
critical value. The maximum tangential stress criterion is widely used to 0.45 0.60 0.75
a0/W
predict the propagation direction of mixed-mode I–II stresses under
plane-stress conditions. The expression for the propagation angle β of Fig. 9. 2WB2EJe0/F2 versus a0/W curves that were calculated by Eq. (25) and
the cracks is FEA.

3KII2 ± KI4 + 8KI2 KII2

β = arccos Table 3
KI2 + 9KII2 (16) The parameters of the Je model.
The fatigue crack propagation angle β of a CTS specimen is de- qij j=1 j=2 j=3 j=4
pendent on α and a0/W, and these cracks will propagate in straight lines
i=1 −1.98600 −2.39435 5.51115 −5.51169
at near-constant angles over a wide propagation range. Fig. 8 shows the
i=2 −3.46338 1.29190 0.622069 1.07966
β-versus-α curves corresponding to the FCP of the cracks of the CTS i=3 −7.43093 −0.00127813 5.69644 10.6266
specimens that were obtained using Eq. (16) and experimental methods i=4 132.760 −272.459 177.354
with various a0/W values. Here, it is shown that the maximum tan- i=5 −481.592 −742.014 576.901
gential stress criterion is highly effective in predicting the FCP direction i=6 3.63971 3.03853 −2.52773 0.653689
i=7 −2.72640 −1.28035 0.429740
of the mixed-mode I–II cracks.
i=8 −12.0987 8.79046 0.00458932 0.988408
Chen and Cai [40] proposed a semi-analytical energy equivalence i=9 21.2490 −5.2538 0.88285
model for the energy, loading, displacement, and constitutive relations
of discontinuous solids. This model is based on fact that the strain en-
ergy of the deformed region of a structural member is equivalent to the
product of the strain energy density at the median point between the
RVE and the volume of the deformed region and the von Mises Loading angle a=15° relative error=2%
10
Je / Je0 Prediction by Equation (26)

Loading angle a=30°

75 Loading angle a=45°
Loading angle a=60°
a0/W=0.36 Loading angle a=75°
a0/W=0.46 Prediction of Maximum Tangential Stress Criterion relative error=2%
60
Crack propagation angle ȕ/degree

a0/W=0.56

45 a0/W=0.36
Test
a0/W=0.46
a0/W=0.56
30

1 10
Je / Je0 Calculated results by ANSYS
15
Fig. 10. The Je/Je0 curves given by Eq. (26) and FEA with different loading
angels.
0
15 30 45 60 75
Loading angle Į /degree equivalence, which implies that the RVE strain energy density at the
energy center of the complex stress state is equivalent to the RVE strain
Fig. 8. Comparisons of the crack propagation angles given by tests and those
energy density of the equivalent uniaxial stress state as followings
predicted by Eq. (16) with different α and a0/W values.

155
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

10
a0/W=0.36 3#
a0/W=0.46 3# 0
Loading angle 15
a0/W=0.56 3#
8

a0/W=0.36 5#
a0/W=0.46 5# 0
Loading angle 30
Crack length a /mm

6 a0/W=0.56 5#
a0/W=0.56 6#

4 a0/W=0.36 5#
0
a0/W=0.46 1# Loading angle 45
a0/W=0.56 1#
2
a0/W=0.36 2#
0
a0/W=0.46 2# Loading angle 60
a0/W=0.56 2#
0
80000 160000 240000 320000
Cycle N /Cycle
Fig. 11. Relations between FCP length a and the cyclic number N.

1E-5 elastic component (Je) in units of energy is

30Cr2Ni4MoV steel
∂Ue
R=0.1 Je =
B ∂a0 (20)
1E-6
where B is the thickness of the specimen. Under small-scale yielding
da/dN / mm/Cycle

conditions, Je is equal to the energy release rate, G, which yields

1E-7
Je = G (21)

The combination of Eqs. (19) and (20) yields

1# J th 0.2998 MPa mm P 2 ∂ ⎛ h∗ ⎞
1E-8
Je =
2# J th 0.2857 MPa mm 2BE ∂a0 ⎝ A∗ ⎠ (22)
3# J th 0.2803 MPa mm
When deflected propagation has not occurred in a crack, the effec-
Average J th 0.2886 MPa mm
tiveness of expressing A* is the function of b0, b0 the length of the re-
1E-9
1 sidual ligament in the specimen, where b0/W = 1 − a0/W, h*and A*
J / MPa·mm can be assumed by:
Fig. 12. The experimental results of the thresholds for the J-integral range. ∗
⎧ ∗ h =W
⎨
⎩ A = WB / f (b0 / W ) (23)
h
⎧
Ue
⎧ ( h∗ )2 Here, WB is the cross-sectional area at the crack surface of the
⎪
⎪ Ue∗ = ⎨ ( P∗ )2 specimen. As Eq. (22) is equal to Eq. (23), the following formulation
⎨ ⎩ Pe can be obtained
⎪ P h
⎪ Pe∗ = h∗ (17) P 2 ∂ [f (b0 / W )]
⎩ Je =
2B2E ∂a0 (24)
In Eq. (17), P is the external load, Pe* the characteristic load, h the
displacement, h* the characteristic displacement, Ue the elastic strain The FEA was performed in ANSYS to characterize the J-integral of
energy, and Ue* the characteristic energy. The Pe* and Ue* are expressed the CTS specimens semi-analytically. A model for Je0 for mixed-mode
as follows I–II cracks of CTS specimens (with a0/W values between 0.36 and 0.8)
that have not undergone deflected crack propagation can be obtained as
∗ ∗
⎧ Pe = EA shown in Eq. (25). The values of the parameters in this equation are
∗ EV ∗ listed in Table 2.
⎨Ue =
⎩ 2 (18)
P2
Je0 = [k1 + k2·(b0 / W )−1 + k3·(b0 / W )−2 + k 4·(b0 / W )−3],
In this equation, V is the characteristic volume, where V = A*h*,
* *
2WB 2E
*
A is the characteristic area of the deformed region. The characteristic ⎧ k1 = k11 + k12cos1.5 α + k13cos2 α + k14 /ln cos α
volume, characteristic area, and characteristic displacement are used to ⎪ k2 = k21 + k22 cos α + k23cos2 α + k24 e cos α + k25/ln cos α
nondimensionalize the loads, displacements, and energies corre- ⎨ k3 = k31 + k32cos2 α + k33cos2.5 α
sponding to various a0/W conditions. The relationship between Ue and ⎪
⎩ k 4 = k 41cosk 42 α + k 43cosk 44 α + k 45 (25)
P may be obtained by combining Eqs. (17) and (18)
2 2
The 2WB EJe0/P versus a0/W curves that were calculated for CTS
P 2 h∗
Ue =
(19) specimens tested at various α values using the finite element method
2E A∗
and Je0 model or Eq. (25) are shown in Fig. 9. It may be observed that
The J-integral [46] is a fundamental aspect of fracture mechanics. the relative deviation between these curves is less than 1%.
Under small-scale yielding conditions, the expression of the J-integral’s When deflection occurs in the cracks of the CTS specimen (for a0/W

156
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

0.1 0.1

30Cr2Ni4MoV steel 30Cr2Ni4MoV steel

0.01 R=0.1 0.01 R=0.1
0
Loading angle 15 Loading angle 30
0

1E-3 1E-3
da/dN / mm/Cycle

da/dN / mm/Cycle
1E-4 1E-4 a0/W=0.36 5#
a0/W=0.36 3# a0/W=0.46 5#
Test a0/W=0.46 3# Test a0/W=0.56 5#
1E-5 1E-5
a0/W=0.56 3# a0/W=0.56 6#

1E-6 1E-6 a0/W=0.36 5#

a0/W=0.36 3#
Prediction FCP-PSED a0/W=0.46 5#
Prediction FCP-PSED a0/W=0.46 3#
1E-7 with Crack Closure 1E-7 with Crack Closure a0/W=0.56 5#
a0/W=0.56 3#
a0/W=0.56 6#

1E-8 1E-8
0.1 1 10 100 1000 0.1 1 10 100 1000
J / MPa·mm J / MPa·mm
(a) (b)
0.1 0.1

30Cr2Ni4MoV steel 30Cr2Ni4MoV steel

0.01 R=0.1 0.01 R=0.1
0 0
Loading angle 45 Loading angle 60
1E-3 da/dN / mm/Cycle 1E-3
da/dN / mm/Cycle

1E-4 1E-4
a0/W=0.36 4# a0/W=0.36 2#

1E-5 Test a0/W=0.46 1# 1E-5 Test a0/W=0.46 4#

a0/W=0.56 1# a0/W=0.56 2#

1E-6 1E-6
a0/W=0.36 4# a0/W=0.36 2#
Prediction FCP-PSED a0/W=0.46 1# Prediction FCP-PSED a0/W=0.46 4#
1E-7 with Crack Closure a0/W=0.56 1# 1E-7 with Crack Closure a0/W=0.56 2#

1E-8 1E-8
0.1 1 10 100 1000 0.1 1 10 100 1000
J / MPa·mm J / MPa·mm
(c) (d)
Fig. 13. Comparisons of da/dN versus ΔJ curves given by FCP tests of mixed-mode I–II cracks and theoretical results from Eq. (15) for 30Cr2Ni4MoV steel.

values in the range of 0.36–0.64 and a/W values in the range of 0–0.2), effectiveness and accuracy of our Je model.
the elastic component of the J-integral for mixed-mode I–II cracks, Je
model (the elastic component of the J-integral when deflection has not
5. Results and discussion
occurred in the mixed-mode I–II crack), may then be described using
Eq. (26). The values of the parameters in this equation are listed in
Fig. 11 shows the relations between the FCP length a in a CTS
Table 3.
specimen and the cyclic number N, as measured by VIC-3D.
The threshold for the J-integral range was obtained by experiment,
Je q1 + q2·(a0 / W ) + q3·(a / W ) + q4·(a / W )2 + q5·(a / W )3
= the J-integral range (ΔJ) was obtained by Eqs. (25) and (26). The
J0e 1 + q6·(a0 / W ) + q7·(a0 / W )2 + q8·(a / W ) + q9·(a / W )2
3 measurement was based on the force drop test, the fatigue crack growth
⎧ q1 = q11 + q12 cos α + q13cosα /ln(cosα ) + q14 /ln(cosα ) rate was close to zero or the crack tip J-integral range corresponding to
⎪ q = q + q cos3α + q ln(cosα )/cos2α + q /cos2α
⎪ 2 21 22 23 24 the crack stop expansion. Generally, the range of J-integral corre-
⎪ q3 = q31 + q32cosα /ln(cosα ) + q33ln(cosα )/cos2α sponding to the fatigue crack growth rate equal to 10−7 mm/cycle was
⎪
defined as the fatigue crack growth threshold ΔJth. Fig. 12 shows the
⎪ + q34 /cos2α
⎪ relationship between the FCP rates da/dN and the ΔJ by the experiment
⎪ q4 = q41 + q42 α 2 + q43 α 0.5lnα
for the threshold of the J-integral range. The value of ΔJth is
⎨ q5 = q51 + q52 α + q53 e α 0.2886 MPa mm.
⎪
⎪ q6 = q61 + q62 (lnα )2 + q63/ α1.5 + q64 / α 2 The threshold for the J-integral range for mode I cracks in
⎪ q7 = q71 + q72 α 2lnα + q73/ α1.5 + q74 lnα / α 2 30Cr2Ni4MoV rotor steel for a loading ratio of 0.1 was obtained from
⎪
⎪ q8 = q81 + q82 cos2.5α + q83/ln(cosα ) + q84 /cos0.5α experimental measurements as shown in Fig. 12. The values of ΔJth and
⎪ −q (cosα − q93) M are 0.2886 MPa·mm and 0.25, respectively.
⎪
⎩ q9 = q91 e−e 92 (26) Fig. 13 shows that the FCP rates of mixed-mode I–II cracks predicted
by LCF-FCP model, Eq. (15) for 30Cr2Ni4MoV steel agree closely with
Fig. 10 shows the Je/Je0 curves that were calculated by Eq. (26) and the experimental data, the prediction model can safely predict the crack
FEA with different loading angels for CTS specimens whose cracks had growth rate at each loading angle. In addition, it was found that the
undergone deflection, for various values of α. The relative deviation of effects of interactions between crack surfaces are not negligible in the
95% of these data is less than 2%. Figs. 9 and 10 therefore validate the measurement of FCP rates for mixed-mode I–II cracks. For a given

157
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

loading angle, the effects of a0/W on crack propagation rates are neg- ICBMFF2. 1985.
ligible. [10] Lal D. Fatigue crack propagation and fracture in compressive members. Mater. Sci.:
MS Thesis, Syracuse Univ 1970.
[11] Mahanty DK, Maiti SK. Experimental and finite element studies on mode I and
6. Conclusions mixed mode (I and II) stable crack growth—I. Exp Eng Fract Mech
1990;37:1237–50. https://doi.org/10.1016/0013-7944(90)90065-O.
[12] Buzzard RJ, Gross B, Srawley JE. Mode II fatigue crack growth specimen devel-
In this study, we developed an analytical LCF–FCP model for pre- opment. Fracture Mechanics: Seventeenth Volume ASTM International; 1986.
dicting the FCP rate of mixed-mode I–II cracks, and we compared the https://doi.org/10.1520/STP17405S.
da/dN-versus-ΔJ curves that were obtained from experiments on [13] Otsuka A, Tohgo K, Matsuyama H. Fatigue crack initiation and growth under mixed
mode loading in aluminum alloys 2017–T3 and 7075–T6. Eng Fract Mech
30Cr2Ni4MoV steel specimens and produced by theoretical predictions. 1987;28:721–32. https://doi.org/10.1016/0013-7944(87)90065-8.
The following conclusions were drawn from the results of this study. [14] Kim JK, Kim CS. Fatigue crack growth behavior of rail steel under mode I and mixed
1. There are a number of factors that affect the effective cyclic stress mode loadings. Mater Sci Eng A 2002;338:191–201. https://doi.org/10.1016/
S0921-5093(02)00052-7.
field of the crack tip zone in mixed-mode I–II cracks. The effect of crack
[15] Peixoto DFC, de Castro PMST. Mixed mode fatigue crack propagation in a railway
closure during cyclic loading was accounted for through the introduc- wheel steel. Procedia Struct Integrity 2016;1:150–7. https://doi.org/10.1016/j.
tion of an empirical crack closure factor M for mixed-mode I–II cracks. prostr.2016.02.021.
In this way, we obtained an effective cyclic stress field for the crack tip [16] Erdogan F, Sih GC. On the crack extension in plates under plane loading and
transverse shear. J Basic Eng 1963;85:519–25. https://doi.org/10.1115/1.
zone of mixed-mode I–II cracks under plane-stress small-scale yielding 3656897.
conditions. [17] Sih GC, Macdonald B. Fracture mechanics applied to engineering problems-strain
2. The LCF–FCP model proposed in this paper can be used to predict energy density fracture criterion. Eng Fract Mech 1974;6:361–86. https://doi.org/
10.1016/0013-7944(74)90033-2.
the FCP rate of plane-stress mixed-mode I–II cracks. It is based on [18] Richard HA, Fulland M, Sander M. Theoretical crack path prediction. Fatigue Fract
consideration of the effective cyclic stress–strain field ahead of a mixed- Eng M 2005;28:3–12. https://doi.org/10.1111/j.1460-2695.2004.00855.x.
mode I–II crack tip under small-scale yielding conditions and the crack [19] Dai Yi ZG, Qilin H. On fatigue crack growth under mixed mode cyclic loading. Acta
Aeronautica Et Astronautica Sinica 1987;11:004.
closure effect. It also assumes that the elastic critical nominal energy [20] Abdel Mageed AM, Pandey RK. Fatigue crack closure in kinked cracks and path of
density of an RVE under uniaxial LCF is equal to the nominal strain crack propagation. Int J Fracture 1990;44:R39–42.
energy density of the crack tip plastic zone (CTPZ) of a mixed-mode I–II [21] Brown MW. Analysis and Design Methods in Multiaxial Fatigue. Advances In
Fatigue Science and Technology. Dordrecht: Springer; 1989. p. 387–401. https://
crack under fatigue loading. doi.org/10.1007/978-94-009-2277-8_16.
3. This paper proposes a semi-analytical energy equivalence model [22] Mahajan RV, Ravi-Chandar K. An experimental investigation of mixed-mode frac-
is proposed in this paper for the energy, load, displacement, and con- ture. Int J Fracture 1989;41:235–52. https://doi.org/10.1007/BF00018856.
[23] Pook LP. The fatigue crack direction and threshold behaviour of mild steel under
stitutive relations of mixed-mode I–II cracks, based on the fact that the
mixed mode I and III loading. Int J Fatigue 1985;7:21–30. https://doi.org/10.1016/
strain energy of a deformed region is equal to the product of the strain 0142-1123(85)90004-0.
energy density at the median point of an RVE and the volume of the [24] Theocaris PS, Andrianopoulos NP. The Mises elastic-plastic boundary as the core
deformed region and also the von Mises equivalence. This implies that region in fracture criteria. Eng Fract Mech 1982;16:425–32. https://doi.org/10.
1016/0013-7944(82)90120-5.
the RVE strain energy density at the energy center of the complex stress [25] Guo Y, Li Q. Material configurational forces applied to mixed mode crack propa-
state is equivalent to the RVE strain energy density of the equivalent gation. Theor Appl Fract Mec 2017;89:147–57. https://doi.org/10.1016/j.tafmec.
uniaxial stress state. A semi-analytical model is also proposed for the Je 2017.02.006.
[26] Gdoutos EE. Fracture Mechanics Criteria and Applications. Springer Science &
integral of mixed-mode I–II compact tension shear (CTS) specimens. Business Media; 2012.
4. The FCP rates of mixed-mode I–II cracks at various loading angles [27] Yan JH, Sutton MA, Deng X, et al. Mixed-mode crack growth in ductile thin-sheet
in 30Cr2Ni4MoV steam turbine rotor steel that were predicted by our materials under combined in-plane and out-of-plane loading. Int J Fracture
2009;160:169. https://doi.org/10.1007/s10704-009-9420-x.
LCF–FCP theoretical model (which accounts for the crack closure effect [28] Kamaya M, Kawakubo M. Strain-based modeling of fatigue crack growth–an ex-
in cyclic loading) agreed closely with experimentally measured FCP perimental approach for stainless steel. Int J Fatigue 2012;44:131–40. https://doi.
rates. org/10.1016/j.ijfatigue.2012.05.006.
[29] Zhang R, He L. Measurement of mixed-mode stress intensity factors using digital
image correlation method. Opt Laser Eng 2012;50:1001–7. https://doi.org/10.
Acknowledgments 1016/j.optlaseng.2012.01.009.
[30] Lorenzino P, Beretta G, Navarro A. Application of Digital Image Correlation (DIC) in
resonance machines for measuring fatigue crack growth. Frattura ed Integrità
The authors gratefully acknowledge the financial support for the
Strutturale 2014;30:369. https://doi.org/10.3221/IGF-ESIS.30.44.
present work from National Nature Science Foundation of China (Grant [31] Hos Y, Vormwald M, Freire JLF. Measurement and simulation of strain fields
No.: 11472228). around crack tips under mixed-mode fatigue loading. Frattura ed Integrità
Strutturale 2015;33:42. https://doi.org/10.3221/IGF-ESIS.33.06.
[32] Decreuse PY, Pommier S, Poncelet M, et al. A novel approach to model mixed mode
References plasticity at crack tip and crack growth. Experimental validations using velocity
fields from digital image correlation. Int J Fatigue 2012;42:271–83. https://doi.
[1] Glinka G. A cumulative model of fatigue crack growth. Int J Fatigue 1982;4:59–67. org/10.1016/j.ijfatigue.2011.11.021.
https://doi.org/10.1016/0142-1123(82)90061-5. [33] Paris P, Erdogan F. A critical analysis of crack propagation laws. J Basic Eng
[2] Ellyin F. Crack growth rate under cyclic loading and effect of different singularity 1963;85:528–33. https://doi.org/10.1115/1.3656900.
fields. Eng Fract Mech 1986;25:463–73. https://doi.org/10.1016/0013-7944(86) [34] Dowling NE, Begley JA. Fatigue crack growth during gross plasticity and the J-
90260-2. integral. Mechanics of crack growth 1976. https://doi.org/10.1520/STP33940S.
[3] Kujawski D, Ellyin F. A fatigue crack propagation model. Eng Fract Mech [35] Tanaka K. The cyclic J -integral as a criterion for fatigue crack growth. Int J
1984;20:695–704. https://doi.org/10.1016/0013-7944(84)90079-1. Fracture 1983;22:91–104.
[4] Pandey KN, Chand S. An energy based fatigue crack growth model. Int J Fatigue [36] Wüthrich C. The extension of the J-integral concept to fatigue cracks. Int J Fracture
2003;25:771–8. https://doi.org/10.1016/S0142-1123(03)00049-5. 1982;20:R35–7. https://doi.org/10.1007/BF01141264.
[5] Li DM, Nam WJ, Lee CS. An improvement on prediction of fatigue crack growth [37] Srivastava YP, Garg SBL. Study on modified J-integral range and its correlation with
from low cycle fatigue properties. Eng Fract Mech 1998;60:397–406. https://doi. fatigue crack growth. Eng Fract Mech 1988;30:119–33. https://doi.org/10.1016/
org/10.1016/S0013-7944(98)00029-0. 0013-7944(88)90217-2.
[6] Chen L, Cai L, Yao D. A new method to predict fatigue crack growth rate of materials [38] Chow CL, Lu TJ. Cyclic J-integral in relation to fatigue crack initiation and pro-
based on average cyclic plasticity strain damage accumulation. Chin J Aeronaut pagation. Eng Fract Mech 1991;39:1–20. https://doi.org/10.1016/0013-7944(91)
2013;26:130–5. https://doi.org/10.1016/j.cja.2012.12.013. 90018-V.
[7] Shi KK, Cai LX, Chen L, et al. Prediction of fatigue crack growth based on low cycle [39] Hoshide T, Socie DF. Mechanics of mixed mode small fatigue crack growth. Eng
fatigue properties. Int J Fatigue 2014;61:220–5. https://doi.org/10.1016/j. Fract Mech 1987;26:841–50. https://doi.org/10.1016/0013-7944(87)90033-6.
ijfatigue.2013.11.007. [40] Chen H, Cai L. Unified elastoplastic model based on a strain energy equivalence
[8] Shi KK, Cai LX, Qi S, et al. A prediction model for fatigue crack growth using ef- principle. Appl Math Model 2017;52:664–71. https://doi.org/10.1016/j.apm.2017.
fective cyclic plastic zone and low cycle fatigue properties. Eng Fract Mech 07.042.
2016;158:209–19. https://doi.org/10.1016/j.engfracmech.2016.02.046. [41] Standard A. Standard Practice for Strain-Controlled Fatigue Testing. Annual book of
[9] Richard HA. Specimens for investigating biaxial fracture and fatigue processes. ASTM standards. PA: American Society for Testing and Materials West

158
S. Qi et al. International Journal of Fatigue 119 (2019) 150–159

Conshohocken; 2004. propagation 1967. https://doi.org/10.1520/STP47234S.

[42] Shih CF. Elastic-plastic Analysis of Combined Mode Crack Problems. Harvard [45] Hua G, Brown MW, Miller KJ. Mixed-mode fatigue thresholds. FFEMS 1982;5:1–17.
University; 1973. https://doi.org/10.1111/j.1460-2695.1982.tb01220.x.
[43] Symington MF, Shih CF, Ortiz M. Tables of plane strain mixed-mode plastic crack [46] Rice JR. A path independent integral and the approximate analysis of strain con-
tip fields. Brown University Division of Engineering, 1988. Report No.: MRG/DMR- centration by notches and cracks. J Appl Mech 1968;35:379–86. https://doi.org/
8714665/1. 10.1115/1.3601206.
[44] Rice J. Mechanics of crack tip deformation and extension by fatigue. Fatigue crack

159