Graphical Abstract

FCG behavior

Local stress-strain approach

S-N curves

Cumulative fatigue damage theory

Fatigue degradation function

Phase-field model for fatigue fracture

⚫envelope loading
Classical fatigue theory informed phase-field model ⚫cycle jump
 (a) (b) Δ�C

S/E-N curve Paris’ law

�a /�� (log)

da/dN (log)
m
1

Fatigue limit Ⅰ Ⅱ Ⅲ
C Δ�
th
�� (log) � (log)

Fig. 1. (a) S /E–N curve. (b) Paris’ law for fatigue crack growth behavior.

37 and ultimate failure (III), as shown in Fig. 1(b). Stress intensity factor K [7] was proposed to reflect the crack-tip
38 stress field, and it is another parameter besides Gc (Griffith criterion [8]) to determine whether the crack enters the
39 instability state. As well as, the famous Paris’ law [9] was proposed to characterize the relationship between fatigue
40 crack growth rate (FCGR) and K during the fatigue crack propagation. Afterwards, Francfort and Marigo [10] re-
41 visited the brittle fracture as an energy minimization problem, and then proposed a variational model of quasi-static
42 crack evolution. Following this idea, Bourdin et al. [11] regularised the fracture variational theory and introduced a
43 phase-field variable to characterize crack, thus providing a novel approach for fracture simulation. Phase-field fracture
44 (PFF) methods have been developed, and large numbers of studies for PFF have been performed since the beginning
45 of the 21st century.
46 In contrast to PFF method, some numerical techniques, e.g., cohesive zone models [12] and extended finite ele-
47 ment method [13] are developed, which suffer from the problem of describing the topology of evolving cracks and
48 also require a predefined crack path [14]. PFF method, as a kind of variational fracture model, can describe crack
49 evolution by an order parameter without explicit crack tracking and extra crack criterion. Complicated procedures in-
50 volving mesh disconnection, element deletion and re-meshing of the interfaces are not required in the PFM of fracture.
51 The appealing advantage of PFF method is its unified framework, in which the entire fracture evolution (nucleation,
52 propagation, branching, kinking) is covered [15]. Thermodynamically-consistent PFM of fracture was presented by
53 Miech et al. [16], in which the crack irreversibility condition is given and the spectral decomposition of strain is
54 proposed to distinguish tensile and compressive effects on crack growth. It is thought that compressive strain/stress
55 does not drive crack evolution, and a local history field of maximum positive elastic energy is introduced to drive the
56 crack propagation and avert crack healing. Furthermore, various decompositions of elastic energy [16–18], multi-field
57 coupled cases [19–26] , and various material types [27–30] are discussed. PFM has a wide range of applications in
58 multi-field and multi-medium fracture issues, see details in review [31].
59 However, the real service processes usually involve oscillating or fatigue cyclic loads, which usually do not lead
60 to an immediate failure, but rather a failure caused by fatigue or FCG after a certain number of loading cycles. With
61 this in mind, intensive efforts have been made to extend PFM of fracture to fatigue problems with cyclic loading.
62 Three mainstream perspectives are formed. Firstly, one can modify crack geometric energy by introducing a fatigue
63 degradation function [32] (called Gc -based model in this work and A-type model in review [14]). Secondly, one can
64 also define an additional fatigue energy term [33] caused by cycle fatigue damage (called energy-based model in this
65 work and B-type model in review [14]). Thirdly, one can [34] use the mobility parameter to control the FCG behavior.
66 A clear difference between these models is that for Gc -based model, the energy-based fatigue function has no upper
67 boundary [14]. The choice of a model variant should eventually conform to the desired physical interpretation. Gc -
68 based model approaches and applications are suitable for reducing the material’s crack resistance while energy-based
69 model go with an increase of the crack driving force compared to the static case.

2
70 As for the Gc -based model, Alessi et al. [32, 35, 36] first proposed a PFM for fatigue fracture with a piecewise
71 differentiable fatigue degradation function. The accumulated strain history variable is selected as the fatigue variable
72 in PFM, which accumulates only during loading by a Heaviside function. Notably, fatigue degradation function
73 plays a key role in the calculation of FCG behavior and S –N curves. Based on Alessi’s model, Golahmar et al. [37]
74 modified the increment of accumulated strain energy to consider the slope of S –N curve, the endurance limit, and
75 the stress ratio impacts. Besides, Seiler et al. [38] proposed a novel and efficient PFM by using LSA for fatigue
76 fracture in ductile materials. The damage-dependent fatigue degradation function has been developed, where the
77 cumulative damage is calculated by means of LSA based Miner rule [39]. The phase-field damage variable replaces
78 the accumulated strain history variable. Seiler’s model was further extended to study the fatigue crack initiation and
79 propagation characteristics during tooth flank fracture in spur gears [40] and FCG under laser shock peening induced
80 residual stresses [41].
81 As for the energy-based model, Boldrini et al [42, 43] proposed a non-isothermal thermodynamically consistent
82 PFM for damage, fracture and fatigue that is incorporated as an extra order parameter. Schreiber and Müller et
83 al. [15, 21, 33, 44] developed a new PFM for fatigue fracture by introducing an additional fatigue energy term that
84 originates from cyclically accumulated deformation work. The fatigue crack energy is related to the crack order
85 parameter, stress amplitude, load ratio R, strain rate ε̇, and damage history. Interestingly, half model parameters can be
86 obtained from the S –N curve. Similarly to Seiler’s model [38], this PFM is coupled with the fatigue parameters from
87 experiments allowing for a general and elegant incorporation of temperature effect or ambient environment. Based on
88 Schreiber’s model, some studies involving accelerated high-cycle fatigue (HCF) calculation [15] and complex load
89 situations [44] are carried out. However, eight model parameters for the fatigue energy have to be determined by
90 experiment or experience.
91 Differing with Gc - and energy-based model, Lo et al. [34] developed a PFM for fatigue crack growth by introducing
92 a viscous parameter, which is a piecewise description fitted to Paris’ curves. A modified J-integral was progressed to
93 study the relationship between the FCGR and the applied loading. The load is applied statically rather than cycle by
94 cycle. Recently, reviews [14, 45] gave an overview of the main phase-field fatigue models published to date.
95 More importantly, fatigue damage occurs only after a large number of loading cycles, resulting in a heavy com-
96 putational task in numerical calculation, especially for HCF or very high-cycle fatigue (VHCF). By means of the
97 classical fatigue life analysis, Seiler’s model [38] can achieve the HCF calculation with low computational cost. Re-
98 cently, an efficient computational framework for accelerating phase-field fatigue calculations by modified Newton
99 approach is presented [46]. Constant load accumulation [15, 38, 44] (also called envelope loading), cycle jumps [47],
100 staggered solution strategies [48], and adaptive mesh refinement [47, 49, 50] are also applied to the acceleration of
101 HCF and FCG behavior. The envelope loading and cycle jumps are perfectly suitable for HCF and VHCF, where
102 very small changes to the overall system are observed between individual load increments. Yang et al. [51] divided
103 the fatigue fracture issue into a macrochronological problem and a microchronological problem, and reduced the time
104 cost by the adaptive macrochronological time steps. Compared with conventional PFM, a mapped PFM [52] was
105 developed due to its advantages of less computation and no fussy mesh refinement for crack path. In addition, fatigue
106 behavior in elastic [35, 48, 53, 54], plastic [55, 56], elastic-plastic solids [21, 57], piezoelectric materials [58], mul-
107 tiferroic materials [59] and rubber [60] under finite strains are studied. The fatigue cracking cases for mixed-mode
108 loading [61], complex loading [44], and mode I + II/III failure [62] are also modelled and discussed by PFM for
109 fatigue fracture. Further, a phase-field regularized cohesive zone model [62, 63] and a crystal plasticity-based PFM
110 (CP-PFM) for modelling small fatigue crack initiation and propagation were proposed [64–66]. Micro fatigue crack
111 behavior influenced by the formation of slip bands along crystallographic planes can be modelled by CP-PFM.
112 With a focus on the fatigue life and FCG behavior rather than the direct considerations of the damage mechanisms,
113 the current phenomenological PFMs for fatigue fracture has their own merit. In this sense, predicting important laws
114 in material fatigue (e.g., S /E–N curves and Paris’ law) with highly efficient numerical computation is exactly what
115 we desire. Therefore, the aim of this work is to develop an efficient phase-field framework for calculating HCF life
116 and FCG. The central idea is inspired by the classical fatigue-life models, in which the phase-field fatigue damage

3
117 variable in fatigue degradation function of PFM is efficiently calculated by using LSA and cumulative fatigue damage
118 theory. Based on this idea, we propose six kinds of PFMs including three classical fatigue-life formulae (strain-,
119 stress-, and energy-based approaches) and two equivalent-stress criteria (Neuber’s and Glinka’s rules). By means of
120 these six PFM models, the fatigue life and FCG calculation during the full scope of fatigue domain, ranging from
121 low- to high-cyclic and even very-high-cyclic regimes, could be feasible. The effectiveness of our PFMs is verified by
122 thorough and quantitative comparisons between simulated and experimental results. The phase-field framework here
123 provides a simple yet useful and efficient toolkit for calculating HCF life and FCG, and for predicting FCG behavior
124 from S /E–N data. The framework could be further extended to incorporate other newly developed models or theories
125 on fatigue and FCG to make PFM capable of harnessing complicated fatigue fracture issues.
126 The manuscript is organized as follows. In Sect. 2, we present the details on PFM for fatigue fracture, with an
127 emphasis on the incorporation of classical fatigue theory that combines LSA, three kinds of fatigue-life models, and
128 two kinds of equivalent-stress criteria. We then in Sect. 3 carry out a series of numerical simulations for fatigue and
129 FCG specimens to verify the proposed PFMs, with a focus on different fatigue degradation parameters, model types,
130 load forces, load ratios, initial crack and materials’ fatigue parameters, quantitative comparison between simulation
131 and experimental results, and the prediction of Paris’ law in FCG by using S /E–N data. Finally, Sect. 4 gives the
132 conclusive summary.

133 2. Classical fatigue theory informed phase-field model

134 In this section, a phase-field fracture formulation [38] informed by the classical fatigue theory [67, 68] is proposed.
135 Details of PFM are introduced briefly, including the rate-independent fracture PFM, fatigue degradation function, and
136 classical fatigue theory. The customary LSA is adopted to evaluate phase-field fatigue damage variable (D) during the
137 cyclic loading, resulting in the decrease of damage-dependent fatigue degradation function. Furthermore, six kinds
138 of PFMs including two equivalent-stress criteria and three fatigue-life models are presented. A specific calculation
139 procedure for fatigue life and FCG by using the proposed PFMs is shown in Fig. 2.

140 2.1. Phase-field model for fracture

141 We consider an arbitrary fractured body Ω with an evolving crack surface Γ containing a discrete crack defined by
142 the internal boundary. According to the variational approach to fracture proposed by Francfort and Marigo [10], the
143 entire fracture process is governed by the minimization of the internal energy Π including the body’s stored energy
144 and the fracture induced dissipated energy, i.e.,
Z Z
Π= ψe (εi j )dV + Gc dA, (1)
Ω Γ
145 where ψe (εi j ) is the elastic strain energy density and Gc is the critical energy release rate or fracture toughness of a
146 material.
147 Without explicit tracking of fracture surface Γ(t), PFM can describe this discrete surface Γ(t) by a crack density
148 function γ(d, ∇d), using an order parameter d (d = 1 for the broken material states, and d = 0 for the intact) and length
149 scale parameter l to control the width of the interface zone. Also, the bulk energy term in Eq. 1 is regularised by
150 introducing a degradation function g(d) to account for the subsequent loss of stiffness due to fracture initiation and
151 propagation. The regularised energy functional Π in the domain Ω can be written as
Z Z
Π= g(d)ψe (εi j )dV + Gc γ(d, ∇d)dV, (2)
Ω Ω

152 where the degradation function g(d) = (1 − d)2 that meets the relationship of g(0) = 1, g(1) = 0, g′ (1) = 0.
153 The crack surface energy density γ(d, ∇d) adopts AT2 model [35], where the crack geometric function is d2 . Thus
154 the crack surface energy density is described as
1 2 l
γ(d, ∇d) = d + |∇d|2 . (3)
2l 2
4
 1 Envelope load 2 Revaluation：Neuber or Glinka rule

Local stress-strain approach

4 Lifetime calculation for �� 3 Calculation of fatigue damage parameter

Fatigue-life models for fatigue damage parameter

5 Calculating phase-field damage 6 Evaluating fatigue degradation function

Linear cumulative damage rule for phase-field damage variable

Fatigue life calculation

7 Phase-field variables update Fatigue crack growth simulation

Phase-field model for fatigue fracture

Fig. 2. Illustration of the classical fatigue theory informed PFMs for calculating HCF life and FCG. ①: schematics for effective loading. ②:
Neuber’s rule sets the triangular area equal to the elastically calculated strain energy density, while Glinka’s rule sets the actual elastoplastic strain
energy density equal to the elastic strain energy density [69]. ③: Damage parameter and energy interpretation of the SWT parameter. ④: Lifetime
calculation from PSWT –N curves. ⑤: Local calculation of phase-field damage variable D by linear accumulation. ⑥: Local calculation of phase-
field fatigue degradation function. ⑦: Phase-field simulation of fatigue facture including HCF life and FCG.

155 2.1.1. Decomposition of elastic energy density

156 In order to distinguish the role of compressive and tensile loading on driving crack, the strain volumetric-deviatoric
157 decomposition, spectral decomposition, and directional split are discussed in Ref [18]. Herein, strain volumetric-
158 deviatoric decomposition [17] for small deformation is adopted, i.e.,
1
εi j = εSi j + εD
ij = (ui, j + u j,i ), (4)
2

5
159 where εi j is mechanical strain, and ui is the displacement, εSi j is spherical strain, and εD
i j is deviatoric strain, expressed
160 as
1 1
εSi j = εkk δi j , εD
i j = εi j − εkk δi j , (5)
n n
161 δi j is the Kronecker delta function and n is the dimension. With this decomposition, the strain energy density of a
162 linear elastic isotropic material can be divided into a positive part (ψ+e ) and negative part (ψ−e ), i.e.,

1 1
ψ+e = K⟨tr(εi j )⟩2+ + µεD
i j · εi j ,
D
ψ−e = K⟨tr(εi j )⟩2− , (6)
2 2
163 where λ and µ are the lame coefficients and K = λ + 2µ/n is the bulk modulus of the material. ⟨·⟩ is the Macaulay
164 bracket selecting the positive (negative) part of the argument. Note that the degradation function affects only the
165 energy density related to the deviatoric and positive volumetric part of the strain tensor. Thus the strain energy density
166 with respect to fracture and the stress can be written as

ψ̃e (εi j , d) = g(d)ψe (εi j ) = g(d)ψ+e + ψ−e , (7)

167
∂ψe
σi j = g(d) = g(d)σ+i j + σ−i j . (8)
∂εi j

168 2.1.2. Extension to fatigue

169 In order to incorporate fatigue into the phase-field framework, the fracture toughness Gc is reduced when the
170 material degrades due to cyclic loading. An additional scalar fatigue degradation function α(D) is introduced to lower
171 the fracture toughness Gc locally. In this way, the total energy function in Eq. 2 can be rewritten as
Z h i
Π= g(d)ψe (εi j ) + α(D)Gc γ(d, ∇d) dV. (9)
Ω

172 The fatigue degradation function α(D) and the relevant phase-field damage variable D or ε̄ in Ref [32] is assumed to
173 be continuous and piecewise differentiable. In general, the main feature of this fatigue degradation function is that
174 introducing a phase-field damage variable D to replace the role of accumulated strain history ε̄ in Ref [32]. Herein,
175 α(D) reads
α(D) = (1 − α0 )(1 − D)ξ + α0 , (10)
176 where the fatigue degradation parameters ξ controls the relation between D and d. The threshold α0 represents the
177 residual fracture toughness in terms of the experiments on residual fracture, which is set larger than zero to ensure
178 a certain resistance against crack propagation at the end of lifetime. The phase-field damage variable D ∈ [0, 1] is a
179 history variable that is accumulated strictly locally. For D = 0, the material has experienced no cyclic loading and
180 therefore offers full fracture toughness, and α(0) = 1 must hold. For D = 1, the fracture toughness is reduced to a
181 threshold value α0 .

182 2.1.3. Evolution equation

183 The governing equation for the crack field d is described by Allen–Cahn equation, i.e.,

∂d δΠ
−η = . (11)
∂t δd
184 Substituting Eq. 9 into Eq. 11 yields
Gc
ηḋ − 2(1 − d)ψ+e (εi j ) + α(D) (d − l2 ∇2 d) = 0. (12)
l
185 Note that η = 0 approximates the quasi-static condition.

6
186 Crack evolution is an irreversible process. As a consequence, the phase-field evolution law must fulfill the irre-
187 versibility condition ḋ ≥ 0. To achieve this, the history variable field H with respect to the positive strain energy is
188 introduced, i.e,
H = max ψ+e (εi j ). (13)
d∈[0;t]

189 Note that the history field variable H only relates to the elastic fracture driving force. The final evolution equation for
190 the fatigue fracture is
Gc
ηḋ − 2(1 − d)H + α(D) (d − l2 ∇2 d) = 0. (14)
l
191 2.2. Incorporating classical fatigue theory
192 The stepwise procedure for simulating fatigue fracture by the classical fatigue theory informed PFM is shown
193 in Fig. 2: ① Stress/strain envolope loading with mean load; ② Identification (extraction) of cycles represented by
194 closed stress-strain hysteresis loops; ③ Calculation of fatigue damage parameter PSWT ; ④ Lifetime calculation for
195 Ni ; ⑤ phase-field damage variable summation (Miner–Palmgren hypothesis); ⑥ Entering phase-field calculation and
196 evaluating fatigue degradation function; ⑦ Phase-field variables update. In short, the phase-field damage variable D
197 with respect to fatigue degradation function α(D) is calculated through LSA and linear cumulative damage rule, and
198 then transferred to PFM to reduce the fracture toughness and make crack initiation and propagation.

199 2.2.1. Local stress-strain approach for phase-field model

200 The basic assumption of the LSA is that the fatigue damage accumulation and the fatigue life to crack initiation at
201 the notch tip are the same as in a smooth material specimen if the stress-strain states in the notch and in the smooth
202 specimen are the same. In this work, the evaluation of the phase-field damage variable D follows LSA by means of
203 the concepts of stress revaluation, S /E–N curve, and damage accumulation from the classical fatigue theory.
204 As the first step, stress and strain are revaluated to elasto-plastic values using the Neuber or Glinka rule as a
205 revaluation hypothesis. Inspired by the classical fatigue theory, we adopt the Ramberg-Osgood model [70] for the
206 material’s elasto-plastic behavior, i.e.,
!1/n′
σ σ
εa = + ′ (15)
E K
207 with the cyclic strength coefficient K ′ and cyclic strain hardening exponent n′ . The overline of σ and ε means the
208 equivalent or one-dimensional stress and strain.
209 As shown in ② of Fig. 2, based on the Neuber’s rule [71] and Glinka’s rule [69], the linear elastic strain and stress
210 are replaced by the revaluated stress σ and strain ε. In this way, one has
   1/n′ 
σ σ
σ + for Neuber rule ,


K′

 E
σel εel = 

 1/n′  (16)
σ σE + n′2+1 Kσ′


 for Glinka rule .

211 The revaluation stress σ can be obtained by solving Eq. 16. Note that LSA commonly adopts only equivalent
212 stress and strain since usually only uniaxial stress states are considered in experiments. Hence, the equivalent elastic
213 stress is regarded as Von Mises equivalent stress to consider the uniaxial or multiaxial stress states. Neglecting the
214 contribution from compressive stress, the Von Mises equivalent stress σel for 2D case is modified as
q
σel = g(d) (σ+11 )2 + (σ+22 )2 − σ+11 σ+22 + 3(σ+12 )2 , (17)

215 where σ+11 and σ+22 are the normal stress of positive-part stress tensor, and σ+12 is the shear stress of positive-part
216 stress tensor. Then, using these results from step ② in Fig. 2, one can perform step ③ to evaluate the fatigue damage
217 parameter.

7
 Table 1. Different types of fatigue-fracture PFMs informed by classical fatigue theory.

Strain-life Energy-life Stress-life

Neuber’s rule model 1 model 2 model 3
Glinka’s rule model 4 model 5 model 6

218 2.2.2. Fatigue-life models for fatigue damage parameter

219 Agoda et al [72] compiled various SWT parameters to calculate the fatigue life of metals. Herein, three quantities
220 including strain amplitude εa [5], stress-strain energy σa εa [6] and stress amplitude σa are used to describe the fatigue
221 damage parameters PSWT in fatigue life curves. In addition, Kujawski et al. [73] proposed a deviatoric interpretation
222 of Neuber’s rule and SWT parameter, demonstrating a fairly good correlation for R < 2. Here, we adopt three models
223 to calculate the fatigue lifetime Ni , including strain-life approach, stress-life approach, energy-life approach, i.e.,
 σ′
ε εa = εa,e + εa,p = Ef (2N f )b + ε′f (2N f )c for strain-life approach,
 
a


 


σ′f 2
 
PSWT =  σa εa ⇒ 
 
σ ε = E (2N f ) + σ f ε f (2N f )
′ ′ (18)
  2b b+c

 
 a a for energy-life approach,
σ

 
σa = σ′ (2N f )b

for stress-life approach,

a f

224 where σ′f is the fatigue strength coefficient, ε′f is the fatigue ductility coefficient, b is the fatigue strength exponent, c
225 is the fatigue ductility exponent, and 2N f is the number of reversals to failure. Note that, the calculation of fatigue life
226 should be corrected for different loading modes and different stress ratios. Eventually, six kinds of models combining
227 two rules (Neuber and Glinka) and three fatigue-life approaches are developed, as shown in Table. 1.
228 When taking into account the mean stress and stress ratio, the strain-, energy-, and stress-life approaches should
229 be modified as
σ −σ




 εa = f E m (2N f )b + ε f (2N f )c Morrow mean stress correction,
σ2f


σmax εa = E (2N f )2b + σ f ε f (2N f )b+c SWT mean stress correction, (19)





σmax = σ′ (2N f )b

 Modified Basquin equation
f

230 where σmax = σa + σm = 2σa /(1 − R). The fatigue lifetime Ni can be derived by solving Eq. 19, which describes
231 the number of load cycles the material could bear by assuming continuous stressing with only the damage parameter
232 PSWT . The corresponding step is ③ and ④ in Fig. 2. The phase-field damage variable Di of the single load cycle then
233 is obtained by Eq. 20 using the Miner’s rule, as shown in ⑤ in Fig. 2. Finally, the phase-field fatigue degradation
234 function α(D), the evolution of order parameter d, and stress, strain and energy update are implemented, as shown in
235 Fig. 2 ⑥ to ⑦.

236 2.2.3. Linear cumulative rule for phase-field damage variable

237 The classical fatigue theory has been widely applied to efficiently evaluate the fatigue life of structures or materials
238 in engineering for a long time. Its core elements include LSA and cumulative fatigue damage related fatigue-life
239 models (e.g., stress-, strain-based, energy-based approaches). Inspired by the linear cumulative damage rule (i.e.,
240 Miner’s rule [39]) from the classical fatigue theory, we set the phase-field damage variable D in our PFM in Eq. 9 as
1 X 1
∆Di = , and D = ∆N . (20)
Ni i
Ni

241 For instance, if one adopts stress-based fatigue life model, then Ni is the fatigue life under σa,i during the single
242 loading cycle i and can be calculated by using S /E–N curve. It should be noted that the choice of the cycle number
243 increment ∆N is usually determined by a compromise between simulation cost and accuracy. We desire that phase-
244 field simulation of cyclic fatigue fracture only requires moderate computational cost with acceptable accuracy. Three
8
 Table 2. Material and simulation parameters for 42CrMo4 [38, 75].

parameter name 42CrMo4

E elastic modulus 210 GPa
v poisson ratio 0.3
η mobility parameter 0
Gc fracture toughness 27 kN/m
K′ cyclic strength coefficient 907 MPa
n′ cyclic strain hardening exponent 0.089
σ′f fatigue strength coefficient 1587 MPa
ε′f fatigue ductility coefficient 4.72
b fatigue strength exponent −0.05
c fatigue ductility exponent −0.67

245 stages for FCG are identified and we apply different cycle number increments in each damage stage. For instance, the
246 ∆N = 1 for the uniaxial loading, ∆N =10–1000 for the constant amplitude loading, and finally ∆N = 0.1 for fatigue
247 failure to make solver convergence. Due to the HCF or VHCF, the stress/strain amplitude at the previous step and
248 current step will not change too much. As a result, it is still reasonable to take dD = ∆NdD even if the cycle number
249 increment ∆N is relatively large.
250 Overall, the phase-field damage variable in fatigue degradation function is calculated by using LSA, PSWT , and
251 Miner’s rule, resulting in a classical fatigue theory informed PFMs that could be applicable to both low cycle and
252 HCF. In this following sections, we will carried out a series of numerical simulations to quantitatively demonstrate
253 the ability of the proposed PFMs in calculating HCF life and FCG.

254 3. Numerical examples

255 The above PFMs are numerically implemented by finite element method in the open source Multiphysics Object
256 Oriented Simulation Environment (MOOSE) [74]. The six kinds of PFMs in Table 1 are utilized in the numerical
257 simulations. Benchmarks including one- and two-dimensional cases are carried out to serve the purpose of qualitative
258 verification and to demonstrate fundamental principles. Further, a quantitative comparison between simulation and
259 experimental S –N results is presented. FCG behavior of CT specimen is modelled and its influencing factors are
260 systematically investigated. The model parameters of an example material 42CrMo4 are shown in Table. 2.

261 3.1. Benchmark simulations

262 Benchmark simulations are first carried out for high-strength steel 42CrMo4 with yield strength > 1000 MPa to
263 verify and compare the above six PFMs. Simulation examples include fatigue life in one-dimensional case and a
264 dog-bone bar, as well as FCG in two-dimensional case regarding to type-I, type-II, and three-point bending fracture.

265 3.1.1. One-dimensional examples for fatigue life

266 In order to quantitatively demonstrate the difference in fatigue life, stress-strain state, and fatigue damage param-
267 eter PSWT among the above six PFMs, a one-dimensional uniaxial tension test is presented using six PFMs under the
268 same strain loading (εapp = 0.4%, ∆N = 10, and R = −1). As shown in Fig. 3(a), the fatigue life calculated by model
269 1 is the most conservative and that by model 6 is the most overestimated. For the fatigue-life approach, strain-life
270 analysis is the most conservative, and stress-life analysis is the most overestimated. Comparing Neuber’s rule and
271 Glinka’s rule (Eq. 16), the equivalent stress calculated by the former is theoretically larger than that of the latter, thus
272 resulting in a greater damage parameter PSWT and lower fatigue life for the case of Neuber’s rule (as shown in Fig. 4).
9
 (a) 1 (b) 800
Von Mises equivalent stress
N1 N2 N4 N5 N3 N6
Neuber equivalent stress
0.8 Glinka equivalent stress
600

Stress (MPa)
0.6

400
d

0.4

200
0.2

0 0
10 20 30 40 50 60 70 80 90 100 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
3
N (Cycle) (×10 ) Strain (%)

Fig. 3. One-dimensional solutions for fatigue life at εapp = 0.4%, ∆N = 10, and R = −1. (a) Crack phase field d as a function of loading cycle N.
(b) Stress-strain curve for Von-mises equivalent stress, Neuber equivalent stress, and Glinka equivalent stress during fatigue cyclic loading.

(a) 6 (b) 4 (c) 0.8

Strain-life approach Energy-life approach Stress-life approach
PSWT = σmaxεa (×10-3)
PSWT = εa (×10-3)

5 0.7
0.00415 Neuber rule 3

PSWT = σmax
0.00243 0.6 0.585
4
0.00368 Glinka rule
0.00204 0.554
2 0.5
3

0.4
N1 N4 N2 N5 N3 N6
2
4 5 6 4 5 6 4
10 10 10 10 10 10 10 105 106
Nf Nf Nf

Fig. 4. PSWT –N f curves with the case of Fig. 3 for (a) strain-life approach, (b) energy-life approach, (c) and stress-life approach.

273 The equivalent stress evaluated by the PFM is consistent with theoretical results [76], as shown in Fig. 3(b). The
274 PSWT –N f curves for various fatigue-life approaches are shown in Fig. 4. Obviously, PSWT calculated by Neuber’s rule
275 is larger than the one by Glinka’s rule. Based on Eq. 18, the fatigue life can be obtained, as shown in Fig. 4, which is
276 in line with the results calculated by PFM (Fig. 3(a)).

277 3.1.2. Dog-bone bar under uniaxial cyclic loading

278 We then consider a dog-bone bar and its fatigue life when subjected to a uniaxial cyclic loading. Schematic illus-
279 tration of the tensile-compressive dog-bone bar and the envelope load replacing real cyclic load are shown in Fig. 5(a)
280 and (b), respectively. The application of envelope load and cyclic equivalent method [15, 44] significantly reduces the
281 computational cost, thus enabling the efficient calculation of HCF or even VHCF. The damage of monotonic loading
282 has little contribution to total fatigue damage, and the cycle number increment ∆N is set as 0.1 in the monotonic
283 loading phase. Then ∆N = 100 is set for the amplitude load until it fails to return to 0.1. The bar has a length of 100
284 mm, and its cross section is decreased from A = 100 mm2 at both ends to 0.5A at the center. The characteristic length
285 l = 0.6 mm is five times the minimum mesh size (hmin = 0.12 mm). Similarly to the previous literature [38], the
286 fatigue degradation function parameters are chosen to ξ = 1 and α0 = 10−7 that is closed to zero to evade numerical
287 difficulties. A displacement umax = 0.13 mm with zero mean load is applied, and model 4 is chosen for this simulation.
288 Stress-strain behaviors for displacement-control and stress-control conditions are illustrated in Fig. 5(c) and (d),

10
 envelope load
(a) (b)
real load
U/F

mean load

t
single cycle single simulation step

(c) (d)
�a
Stress loading
Cycle number
1st last cycle
σ

Cycle number
1st last cycle

Strain loading Ua
ε ε
Fig. 5. Loading settings in PFM for fatigue fracture in a dog-bone bar. (a) Set up for phase-field simulation of tensile-compressive specimen. (b)
Schematic illustration of replacing direct-cyclic loading by envelope loading within simulation. (c) Strain-stress curve for strain loading model. (d)
Strain-stress curve for stress loading model.

(a) 1 (b) 1

0.8 0.8

dmax
0.6 Dmax 0.6
d, D, α

α (D)

αmin
0.4 0.4

0.2 0.2

0 0
0.1 1 10 0 0.2 0.4 0.6 0.8 1
5
N (Cycle) ×10 D

Fig. 6. (a) d, D, α-N as a function of loading cycle N under cyclic constant displacement. (b) α-D curve under cyclic constant displacement load.

289 respectively. Material stiffness is gradually decreased with fatigue cyclic loading. The evolutions of crack phase field
290 d, phase-field damage variable D and fatigue degradation value α during cyclic loading are presented in Fig. 6(a).
291 Accordingly, D increases continuously as the loading cycles increases, degrading the fracture toughness through
292 α(D). Meanwhile, d develops much slower, and exhibits an abrupt increase when the bar is close to failure. This
293 suggests that a crack after initiation will rapidly propagate until failure in the smooth bar. The evaluation of α(D) and
294 D during envelope load is shown in Fig. 6(b), which is governed by Eq. 10. Besides, Fig. 7 shows the distribution of

11
 (a) (b) 0.2 Nf = 3e5
7e5
0.15 9e5
10.4e5

0.1

d
0.05

0
0 20 40 60 80 100
x (mm)

Fig. 7. (a) Evolution of the distribution of phase-field damage variable D along the bar length as the load cycle N increases. (b) Evolution of the
distribution of crack phase-field order parameter d.

(a1) Nf = 10 (a2) Nf =35000 (a3) Nf =75000 (a4) Nf =107000

(b1) (b2) (b3) (b4)

Fig. 8. Phase-field simulation of fatigue fracture in a single-edge notched tension test: (a) fatigue crack evolution and (b) fatigue damage evolution.

295 D and d along with the length of the bar subjected to load cycles N. A crack first occurs at middle edge of the bar
296 after a load cycle of 1.04 × 106 load cycles. This is attributed to the higher stress and strain at the middle edge, leading
297 to a lager damage D.
298 In general, the fatigue life of a component can be expressed as the sum of the number of loading cycles required
299 to initiate a crack and the number of cycles it takes that crack to propagate to failure [77]. In S –N curves, the sample
300 is thought perfect and the fatigue crack initiation life dominates total fatigue life. Our PFM results are consistent with
301 observations in S –N experiments for metals, i.e., the specimen is damaged continuously without losing much stiffness
302 until a crack occurs abruptly.

303 3.1.3. Two-dimensional examples for FCG

304 We then simulate FCG in two-dimensional examples including type-I, type-II, and three-point bending fracture.
305 The evolutions of crack phase-field order parameter d and phase-field damage variable D in the two-dimensional
306 notched plate are displayed in Fig. 8 (tension, type-I) and Fig. 9 (shear, type-I), respectively, for nucleation, stable
307 and unstable propagation of the crack. The sample size is 100 × 100 mm2 , minimum mesh size hmin = 0.2 mm, and
308 l0 = 3hmin = 0.6 mm. Model 1 is chosen in for the phase-field simulations. The parameters of fatigue degradation
12
 (a1) Nf = 10 (a22)) Nf = 40000
(a (a33))
(a Nf = 55000 (a44))
(a Nf = 64100

Nf = 55000

(b1) (b2) (b3) (b44))

(b

Fig. 9. Phase-field simulation of fatigue fracture in a single-edge notched shear test: (a) fatigue crack evolution and (b) fatigue damage evolution.

(a1) Nf = 20 (b1)

(a2) Nf = 32500 (b2)

(a3) Nf = 42900 (b3)

Fig. 10. Phase-field simulation of fatigue fracture in a three-point bending test: (a) d-N f evolution and (b) D-N f evolution.

309 function are set to α0 = 0.05 and ξ = 1. A displacement is applied with umax = 0.15 mm for tension and umax = 0.2 mm
310 for shear with zero mean load. For the tension test, fatigue crack path is similar to the monotonic case, as shown in
311 Fig. 8. The zone of D = 1 can be associated with the plastic zone. Crack initiates and grows because of the fracture
312 toughness losses originating from the accumulating damage variable D.
313 During the shear test, we can observe that, although the fatigue history variable is accumulated in the whole
314 domain over large regions, the observed crack maintains its localized nature and its path is similar to the one observed
315 in the monotonic case. Namely, the crack originates at the initial notch tip and propagates downwards toward the
316 lower right corner [35]. Due to the changing load direction, fatigue damage also accumulates at the top and bottom
317 crack front alternately, but the fatigue crack do not initiate here. The difference in crack path between statical fracture
318 and fatigue fracture by PFM is attributed to the loss of fracture toughness Gc caused by phase-field damage variable
319 D. Three-point bending test is also simulated by using model 1, as shown in Fig. 10. The FCG path is similar to the
320 case of statical load, which is in agreement with related literature [16, 46].

13
321 3.2. Comparison with experimental S–N curves
322 Based on the above six models (Table. 1), we proceed now to compare PFMs’ predictions with S –N curves
323 obtained from uniaxial tension-compression fatigue experiments on the fatigue specimen (Fig. 5(a)), with the con-
324 sideration of mean load. The experimental fatigue properties of 42CrMo4 is obtained from Ref [78]. Similar to the
325 uniaxial tension-compression fatigue experiments, R = −1 and stress-control load is applied. As shown in Fig. 11,
326 an excellent prediction ability for HCF of 42CrMo4 is demonstrated when the simulation results by PFM are com-
327 pared with experimental S –N curves [78]. It is important to highlight that S –N data evaluated by model 1 (Neuber
328 + Strain-life) and model 4 (Glinka + Strain-life) is in good agreement with cast and wrought 42CrMo4, and fatigue
329 life calculation by strain-life approach are conservative compared with other fatigue-life approaches. Overall, the fa-
330 tigue life prediction by Neuber’s rule is lower than Glinka’s rule, and the predicted fatigue life decreases in the order:
331 strain-life approach < energy-life approach < stress-life approach. This is owing to that the smaller revaluated stress
332 by Glinka’s rule leads to lower fatigue damage, and the various correlations between PSWT –N f , including strain-life,
333 energy-life, and stress-life approaches present multiplicate Ni and Di (see Sec 3.1.1 for details). On the other hand,
334 Glinka’s rule can be used as an essential supplement to enhance the accuracy of the fatigue-life prediction model.
335 As shown in Fig. 11(a), by using Glinka’s rule, the PFM based on strain-life approach is suitable for low-cycle and
336 high-cycle fatigue cases, energy-life process works well for HCF case, and the PFM based on stress-life approach is
337 suitable for VHCF case. This seems to be consistent with the classical fatigue life analysis found in literature [79].
338 In addition, the mean load plays an important role on the fatigue life and fatigue strength. In this work, the mean
339 stress is built-in the PSWT –N f curves, see Eq. 19. As shown in Fig. 11(b), the predictions of various stress ratios with
340 R = −1, −0.5, 0, 0.1, and 0.3 are conducted by using model 2. Most of the existing studies are mainly focused on the
341 fatigue life in low cycle and HCF regimes [80]. Based on SWT’s equation [6] for fatigue strength in both HCF and
342 VHCF regimes, the fatigue strengths of 107 cycles at various stress ratios are evaluated by S –N curves and SWT’s
343 equation, respectively. The SWT’s equation [6] for fatigue strength can be expressed as
r
1−R
σa,R = σa,−1 (21)
2
344 where σa,R and σa,R are the fatigue strength of 107 cycles at stress ratio R and −1, respectively.
345 From the S –N data, as shown in Fig. 11(b), the fatigue strengths of 107 cycles are as follows: σa,−1 ≈ 476 MPa,
346 σa,−0.5 ≈ 412 MPa, σa,0 ≈ 337 MPa, σa,0.1 ≈ 319 MPa, and σa,0.3 ≈ 284 MPa. Based on the Eq. 21, the fatigue
347 strength is calculated to be 412, 335, 319, and 281 MPa for R = −0.5, 0, 0.1, and 0.3, respectively. Therefore, the
348 fatigue strength calculated by PFM is greatly consistent with the prediction by SWT’s equation. This may be attributed
349 to the unity of SWT’s fatigue life approach (Eq. 19) and fatigue strength equation (Eq. 21). These results suggest that in
350 general the classical fatigue theory informed PFM can reproduce the S –N curves for HCF and the effects of the stress
351 ratio and stress amplitude on S –N curves, and a good agreement between simulation and experimental results [78]
352 can be ensured.

353 3.3. FCG in CT (compact-tension) specimen

354 3.3.1. Model verification
355 To further assess the proposed PFMs’ capability in reproducing Paris’ law in FCG behavior, a CT specimen
356 subjected to cyclic loading is simulated. The Paris’ law [9] is expressed as
da
= C(∆K)m (22)
dN
357 where a is the crack length, and da/dN is the FCGR. C and m are constants that depend on the material, environment,
358 and stress ratio. ∆K is the range of the stress intensity factor during the fatigue cycle.
359 The geometry is shown in Fig. 12 with W = 80 mm and thickness B = 2 mm. To simplify, the free ends behind
360 the drilled hole are not included in the model, instead of loading along the axis of drilled hole, as shown in Fig. 12(b).
14
 (a) (b) 500
R = 0.3
600 0.1
Stress amplitude (MPa)

Stress amplitude (MPa)

0
450 −0.5
−1
500
Wrought 42CrMo4 400

Cast 42CrMo4
400
350

model 2
300 4 300 4
10 105 106 107 108 109 1010 1011 10 10
5
10
6
10
7
10
8
10
9
10
10

Nf (Cycle) Nf (Cycle)

Fig. 11. (a) Comparison of S –N curves from simulation by various PFMs and from experimental data [78] for 42CrMo4. (b) S –N curves from
PFM simulations with various stress ratios.

(a) (b) (c)

� �
�0

� �
10 mm
y
�
x
Fig. 12. (a) Sketch of CT specimen according to the ASTM E647 standard [81]. (b) Simplified model for CT specimen in PFM simulations. (c)
Local mesh refine for CT specimen. Geometry with W = 80 mm and thickness B = 2 mm

361 Refining mesh of the crack growth path by HyperMesh, as shown in Fig. 12(c), and l0 = 0.6 ≈ 3hmin is set for CT
362 specimens. A constant force amplitude of F = 2.08 kN with R = 0.1 is applied The initial crack length is set to
363 a0 = 0.25W, and model 2 is adopted. The expression of stress intensity factor for CT specimen was proposed by
364 Srawley et al. [82], i.e.,

∆F 2 + a/W  a  a 2  a 3  a 4 
∆K = √ 0.866 + 4.64 − 13.32 + 14.72 − 5.6 (23)
B W (1 − a/W)3/2 W W W W

365 where ∆F is the applied force, and a/W ≥ 0.2. Based on the Paris’ law (Eq. 22), the FCGR da/dN is described as a
366 function of ∆K, which basically incorporates effects of external load, actual crack length and geometry.
367 Fig. 13(a), (b), and (c) present the a–N curve, da/dN–∆K curve, and the fatigue crack propagation, respectively.
368 Three stages during FCG, involving crack initiation (c1 in Fig. 13), stable propagation (c2 -c3 ), and rapid propagation
369 (failure, c4 ), can be readily reproduced. The FCG life accounts for the majority of the total lifetime. For the stable
370 crack propagation stage, Paris parameter C and m in Eq. 22 can be fitted. As shown in Fig. 13(b), the result well
√
371 matches Paris’ law with parameter C = 3.01 × 10−10 mm/cycle/(MPa m)m and m = 3.9. Sec. 3.3.6 will offer the
372 comparison of Paris parameter C and m between phase-field simulation and experiment.

15
 (a) 20 c4 (b) 10−3
model 2
a0/W=0.25
16 F R = 0.1
F = 2.08 kN

da/dN (mm/cycle)
12 F
c3 10−4
a (mm)

8
10−5

4
c2
c1 C = 3.01e−10, m = 3.9
0 10−6
0 10 20 30 40 50 60 70 10 20 30 40 50
N (Cycle) (×104) ∆K (MPa√m)
(c) c c4
1 c2 c3
d

Fig. 13. Phase-field simulation of FCG in CT specimens. (a) a–N curve under fatigue loading in CT specimen. (b) FCGR with respect to stress
intensity factor. (c) Fatigue crack evolution. Model 2, and F = 2.08 kN, and R = 0.1 are applied.

373 3.3.2. Model comparison

374 In addition, the impact of fatigue degradation function parameters ξ and α0 is shown in Fig. 14. Based on the
375 fatigue degradation function (Eq. 10), the increase of ξ clearly accelerates the fatigue degradation and the loss of
376 fracture toughness. Therefore, shorter fatigue life and higher FCGR presents for a bigger ξ. The influence of parameter
377 α0 is similar, but not so great. Taking into account Paris’ law, increasing ξ and α0 seems to rise the slope of da/dN–∆K
378 curve (m) in the double logarithmic coordinate. Notably, the results of da/dN for ξ = 1 and α0 = 0.05 is typical for
379 most crack growth experiments, and ξ = 1 and α0 = 0.05 is selected to carry the following FCG simulation. It is
380 important to highlight that the phase-field fatigue parameters are less than the Gc - and energy-based models, only
381 involving the fatigue degradation function parameters α0 and ξ. The fracture toughness Gc and crack length scale l0
382 are necessary for any PFM of fracture. Specially, the experimental cyclic stress-strain curve parameters (K ′ , n′ ) and
383 E–N curve parameters (σ′f , ε′f , b, c) are meaningful and mandatory for the PFM to calculate damage variable. These
384 fatigue parameters can be obtained from the classical fatigue theory in which fatigue database of typical materials
385 have been developed.
386 In a CT specimen, phase-field simulations for FCG behavior under six models are carried out and the associated
387 results are summarized in Fig. 15. a0 /W = 0.25, F = 2.08 kN, and R = 0.1 are utilized. FCG life obtained by six
388 models is similar to fatigue life in Sec. 3.2. Obviously, the effects of PSWT –N f analysis and stress equivalent criteria
389 on FCG behavior are similar to the fatigue life analysis (see Sec. 3.2). The FCGR by Neuber’s rule is bigger than that
390 by Glinka’s rule. The FCGR decreases in the order: strain-life approach > energy-life approach > stress-life approach.
391 It is found that model 6 combining Glinka’s rule with stress-life approach yields the longest FCG life to the critical
392 crack length.

16
 (a) 10 −2 (b) 10 −3
Model 2 Model 2
F = 2.08 kN F = 2.08 kN
10
−3 R = 0.1 10
−4 R = 0.1
�0 = 0.05 �=1

da/dN (mm/cycle)
da/dN (mm/cycle)

−4 −5
10 10

−5 −6
10 10
ξ = 1, C = 3.21e−10, m = 3.89 α0 = 0.01, C = 5.25e−10, m = 3.77
ξ = 2, C = 3.52e−10, m = 3.97 α0 = 0.05, C = 3.21e−10, m = 3.89
ξ = 10, C = 1.93e−10, m = 4.61 −7
α0 = 0.1, C = 2.01e−10, m = 4.02
−6
10 10
10 20 30 40 50 10 20 30 40 50
∆ K (MPa√ m) ∆ K (MPa√ m)

Fig. 14. Phase-field simulation results about the influence of (a) exponent ξ and (b) threshold α0 in the fatigue degradation function (Eq. 10) on
da/dN–∆K curves.

(a) 20 (b) 10 −3

16
−4
10
da/dN (mm/cycle)

12
a (mm)

−5
10
8

−6
10
4

−7
0 10
0 5 10 15 20 25 30 35 40 10 20 30 40 50
N (Cycle) (×105) ∆ K (MPa√ m)

Fig. 15. (a) a–N curves and (b) da/dN–∆K curves under different models with R = 0.1.

393 3.3.3. Influence of load and stress ratio

394 Load conditions including force amplitude and stress ratio play an essential role in FCG behavior, as shown in
395 Fig. 16. It is well known that the Paris’ parameters of a material are insensitive to force amplitude, agreeing with
396 the simulation results in Fig. 16(a). Paris’ stage during the FCG with different force amplitudes is almost the same.
397 The differences of initial ∆K between various force amplitudes is attributed to the ∆F in Eq. 23. Stress ratio effect
398 is tested on R = 0, 0.1, 0.3, 0.5, and 0.7 by using model 2 and F = 2.082 kN. The phase-field simulation results of
399 da/dN–∆K curves with various stress ratios are shown in Fig. 16(b), which indicates that a bigger stress ratio with
400 R > 0 will cause a higher FCGR. For R > 0, the lines fitting Paris’ stage of da/dN–∆K curves under different stress
401 ratios are almost parallel, in accordance with the well-known empirical trends. These results prove the validity of the
402 SWT mean stress correction (Eq. 19) that implicitly includes the effect of mean load on Paris’ law.

403 3.3.4. Influence of initial crack length

404 Fig. 17 presents the crack evolution under cyclic loading and the related FCGR in a CT specimen with various
405 length of initial crack. It is clear that FCG life exhibits a correlation with the initial crack length. A short initial
406 crack length has a longer fatigue life, which is in line with known empirical trends. For the FCG behavior, the initial
407 crack length has little effect on Paris’ law. Actually, the Paris’ law or parameters C and m is insensitive to the force

17
 (a) 10−2 (b) 10−2
model 2 model 2
R = 0.1 F = 2.082 kN
10−3 10−3
da/dN (mm/cycle)

da/dN (mm/cycle)
10−4 10−4

R = 0.0
10−5 F = 2.082 kN 10−5 0.1
2.288 kN 0.3
2.496 kN 0.5
2.704 kN 0.7
10−6 10−6
10 20 30 40 50 10 20 30 40 50
∆K (MPa√m) ∆K (MPa√m)

Fig. 16. (a) da/dN–∆K curves for various force amplitudes under F = 2.082, 2.288, 2.496, and 2.704 kN. (b) da/dN–∆K curves under various
stress ratios R = 0, 0.1, 0.3, 0.5, and 0.7.

(a) 20 (b) 10
−1

16
10
−2 a0
da/dN (mm/cycle)

12
a-a0 (mm)

−3
10
8

a0 = 1 mm 10
−4 a0 = 1 mm
4 5 mm 5 mm
7.5 mm 7.5 mm
10 mm 10 mm
0 −5
10
0 5 10 15 20 25 30 35 16 20 30 40 50
N (Cycle) (×104) ∆K (MPa√m)

Fig. 17. Influence of initial crack length on (a) a–N curves and (b) da/dN–∆K curves with model 2, F = 2.48 KN, and R = 0.01.

408 amplitude, specimen geometry, and initial crack length.

409 3.3.5. Influence of E–N parameters

410 The concept of E–N and S –N curves and Paris’ law plays a key role in the analysis of engineering failures caused
411 by fatigue. Davila et al. [83, 84] calculated the FCGR during Paris’ law in composites from the known S –N curves.
412 This seems to prove the feasibility of the conception in Fig. 1. In this work, we simulate FCG behavior with E–N data
413 by using the classical fatigue theory informed PFMs for fatigue fracture. The effect of E–N parameters (b, c, σ′f , and
414 ε′f ) on FCG behavior is investigated with model 2, F = 2.08 kN and R = 0.1. Note that, parameters b and c govern
415 the slope of E–N curve during elastic and plastic stages in double logarithm coordinate, respectively. σ′f and ε′f are
416 the intercept of E–N curve during elastic and plastic phases in double logarithm coordinate, respectively. Obviously,
417 the Paris’ law is sensitive to E–N parameters, as shown in Fig. 18.
418 Fig. 18 shows that the FCGR or parameter C depends on the parameters of E–N curves. The increases of |b| and
419 |c| or the decreases of σ′f and ε′f will lead to a higher FCGR or C, which is consistent with Ref [85] that deducing C
420 proportional to |b| and |c|. This is attributed to that the phase-field damage variable D increases with the increasing of |b|
421 and |c| or the decreasing of σ′f and ε′f under a constant PSWT , according to Eqs. 20 and 19. Besides, Noroozi et al. [86]
422 proposed the similar crack propagation model and deduced the expression of constant C in Paris’ law using SWT

18
 (a) 10−3 (b) 10−3
model 2
R = 0.1

da/dN (mm/cycle)
da/dN (mm/cycle)

−4 −4
10 10

PSWT

PSWT
10−5 10−5
Nf Nf
b = −0.07, C = 7.44e−9, m = 3.16 c = −0.77, C = 2.38e−9, m = 3.60
b = −0.06, C = 2.89e−9, m = 3.33 c = −0.67, C = 1.20e−9, m = 3.46
b = −0.05, C = 1.17e−9, m = 3.50 c = −0.6, C = 4.23e−10, m = 3.41
10
−6
10−6
10 20 30 40 50 10 20 30 40 50
∆K (MPa√m) ∆K (MPa√m)
(c) 10−3 (d) 10−3

da/dN (mm/cycle)
da/dN (mm/cycle)

10−4 10−4
PSWT

PSWT
10−5 10−5
Nf Nf
σ'f = 1.787 GPa, C = 8.55e−10, m = 3.48 ε'f = 4.72, C = 1.17e−9, m = 3.50
σ'f = 1.587 GPa, C = 1.18e−9, m = 3.47 ε'f = 3.72, C = 1.32e−9, m = 3.52
σ'f = 1.487 GPa, C = 1.45e−9, m = 3.63 ε'f = 3.0, C = 1.45e−9, m = 3.57
10−6 10−6
10 20 30 40 50 10 20 30 40 50
∆K (MPa√m) ∆K (MPa√m)

Fig. 18. The influence of E–N parameters (a) b, (b) σ′f , (c) c and (d) ε′f on Paris’ law.

423 fatigue damage parameter by using LSA. C is found inversely proportional to σ′f ε′f , in agreement with the simulation
424 results. The above E–N parameters seem to have a little influence on the slope of Paris’ curves (m). Note that, for the
425 strain-life approach, transition fatigue life NT = 32285 cycle with PSWT = εa = 0.0076(2N f )−0.05 + 4.72(2N f )−0.67 .
426 Herein, the fatigue life is in the order of 105 cycles, and thus the elastic part of E–N curves contributes to the impact on
427 Paris’ law. In other words, the FCGR is more sensitive to E–N parameters b and σ′f in the case N f > NT . In contrast,
428 b and σ′f play a greater role in LCF. For stress-life approach, only parameters b and σ′f have a notable influence on
429 Paris’ law.

430 3.3.6. Comparison with experimental FCG behavior

431 To compare with the experimental FCG behavior [87, 88], fatigue fracture of a CT specimen with F = 2.038 kN,
432 R =0.1 and a0 /W =0.25 is studied by phase-field simulations based on models 1, 2, 4, and 5. As shown in Fig. 19(a),
433 we present both the the fatigue life from both phase-field simulations and experimental test. It is found that the strain-
434 life approach could achieve the results that agree well with the experimental fatigue life of 42CrMo4. Using the data
435 in Fig. 19(b), the fitting and experimental parameters for Paris’ law in 42CrMo4 are summarized in Table. 3. It can
436 be seen that C and m from simulation and experiment agree well with each other, further verifying the capability our
437 PFMs to predict FCG behavior and its Paris’ law. Furthermore, the good agreement between phase-field simulated
438 and experimental da/dN–∆K curves in Fig. 19(b), as well as the Paris’ parameters in Table. 3, indicates that classical
439 fatigue theory informed PFMs for fracture fatigue is capable of correctly predicting FCG behavior by a solely input
440 of fatigue data (e.g., S /E–N curves).

19
 (a) 35 (b) 10−2 Sim−Model 1
Sim−Model 2
Sim−Model 4
30 Sim−Model 5
10−3 Lesiuk et al. 42CrMo4−HT1

da/dN (mm/cycle)
Lesiuk et al. 42CrMo4−N
Sui et al. 42CrMo−M42
25
a (mm)

Sui et al. 42CrMo−M52

10−4
Sim−Model 1
20 Sim−Model 2
Sim−Model 4
Sim−Model 5 10−5
15 Lesiuk et al, 42CrMo4−HT1
Lesiuk et al, 42CrMo4−HT2
Lesiuk et al, 42CrMo4−N
10 10−6
0 20 40 60 80 100 120 140 160 180 10 20 30 40 50
N (Cycle) (×104) ∆K (MPa√m)

Fig. 19. (a) a–N curves based on model 1, 2, 4 and 5. Green point data from experimental results [87]. (b) Fatigue crack growth rate curve based
on model 1, 2, 4 and 5 with F = 2.038 KN, R = 0.1 and a0 /W =0.25. Point lines represent the experimental results of 42CrMo4 from Lesiuk
et al. [87] and Sui et al. [88], where HT1 and HT2 are heat treatment conditions, and N is N-normalized. 42CrMo specimens in Ref [88] were
machined from the center of M42 and M52, two different specifications of bolts.

Table 3. Paris’ parameters C and m of 42CrMo4 from phase-field simulations and experiments.

Model Model 1 Model 2 Model 4 Model 5 HT1 [87] N [87] M42 [88] M52 [88]
C 3.47e-9 2.74e-9 3.44e-9 2.13e-9 2e-8 2e-9 1.45e-8 3.67e-9
m 3.06 3.15 2.85 2.91 3.14 3.15 2.68 3.08

441 4. Conclusion

442 In summary, we have developed a classical fatigue theory informed PFM of fatigue fracture for efficiently sim-
443 ulating HCF life and FCG, with the aim of extending fracture PFM to HCF or even VHCF regime and meanwhile
444 realizing efficient computation instead of heavy direct-cyclic calculation. The proposed PFM inherits the handiness
445 and conciseness of classical fatigue theory that has been widely adopted in the failure analysis and life design of
446 engineering structures. Specifically, our PFM incorporates LSA and cumulative fatigue damage theory to evaluate
447 the phase-field damage variable as a function of a local lifetime. Fatigue effects are then implemented by reducing
448 the fracture toughness through the fracture energy degradation function that is governed by the phase-field damage
449 variable. Furthermore, envelope loading and cycle jump techniques are incorporated into our PFM to simulate HCF
450 or VHCF fracture that is formidable for traditional PFM relying on direct-cyclic loading.
451 In detail, six kinds of PFMs combining three fatigue-life formulae (strain-, stress-, and energy-life approaches)
452 with two equivalent-stress criteria (Neuber’s and Glinka’s rules) are developed. Taking 42CrMo4 as an example
453 material, benchmark phase-field simulations for fatigue crack initiation and propagation in 1D and 2D specimens are
454 performed to validate the applicability and versatility of the proposed PFMs. The PFM combining Neuber’s rule with
455 strain-life approach is the most conservative. The predicted fatigue life by PFM is revealed to agree well with the
456 experimental one. Phase-field simulations of FCG in CT specimens are conducted to reproduce important laws in
457 materials’ fatigue, e.g., Paris’ law. It is demonstrated that the full scope of fatigue domain ranging from low-cycle,
458 high-cycle, and even to very-high-cycle regimes can be covered by means of these PFMs. Using S /E–N data as a
459 solely input, the PFM is shown to capture the correct FCG behavior and reproduce the da/dN–∆K data and Paris’ law,
460 both of which excellently coincide with experimental results, thus enabling interconversion between fatigue and FCG
461 data.

20
462 As a future work, this PFM framework here provides a great potential to merge other newly developed fatigue-
463 life approaches and fracture mechanics method to make these PFMs suitable for the study of fatigue fracture in new
464 materials such as additively manufactured alloys. It is also necessary to retrospect the elastic-plastic behavior and
465 their mathematical models to broaden the application range of the proposed PFMs in fatigue field. The modifications
466 of these PFMs for multiaxial fatigue case are also important directions for the future research.

467 CRediT authorship contribution statement

468 Wei Tang: Formal analysis, Investigation, Visualization, Data curation, Writing – original draft. Min Yi: Con-
469 ceptualization, Resources, Supervision, Project administration, Funding acquisition, Writing – original draft & review
470 & editing. Long-Qing Chen: Conceptualization, Supervision, Writing – review & editing. Wanlin Guo: Conceptu-
471 alization, Supervision, Writing – review & editing.

472 Declaration of competing interest

473 The authors declare that they have no known competing financial interests or personal relationships that could
474 have appeared to influence the work reported in this paper.

475 Acknowledgements

476 This work is partially supported by High Performance Computing Platform of Nanjing University of Aeronautics
477 and Astronautics. Simulations were also performed on Hefei advanced computing center. We thank Prof. Bai-Xiang
478 Xu form TU Darmstadt for her valuable comments.

479 References
480 [1] A. Wöhler, Ueber die Festigkeits-versuche mit Eisen und Stahl, Vol. XX, Zeitschrift für Bauwesen, 1870.
481 [2] W. Schütz, A history of fatigue, Engineering Fracture Mechanics 54 (2) (1996) 263–300. doi:10.1016/0013-7944(95)00178-6.
482 [3] B. OH, The exponential law of endurance tests, in: Proc Am Soc Test Mater, Vol. 10, 1910, pp. 625–630.
483 [4] L. F. Coffin Jr, A study of the effects of cyclic thermal stresses on a ductile metal, Transactions of the American Society of Mechanical
484 Engineers 76 (6) (1954) 931–949.
485 [5] S. Manson, Fatigue-a complex subject-some simple approximations, Experimental Mechanics 5 (1965) 193–226.
486 [6] K. Smith, A stress-strain function for the fatigue of metals, Journal of Materials 5 (1970) 767–778.
487 [7] G. R. Irwin, Analysis of stresses and strains near the end of a crack traversing a plate, J. Appl. Mech. 24 (1957) 361–364. doi:10.1115/1.
488 4011547.
489 [8] A. A. Griffith, The phenomena of rupture and flow in solids, Philosophical transactions of the royal society of london. 221 (1921) 163–198.
490 [9] P. Paris, F. Erdogan, A critical analysis of crack propagation laws, Journal of Basic Engineering 85 (1963) 528. doi:10.1115/1.3656900.
491 [10] G. A. Francfort, J. J. Marigo, Revisiting brittle fracture as an energy minimization problem, Journal of the Mechanics and Physics of Solids
492 46 (8) (1998) 1319–1342. doi:10.1016/S0022-5096(98)00034-9.
493 [11] B. Bourdin, G. A. Francfort, J.-J. Marigo, The variational approach to fracture, Journal of Elasticity 91 (2008) 5–148.
494 [12] B. Yang, S. Mall, K. Ravi-Chandar, A cohesive zone model for fatigue crack growth in quasibrittle materials, International Journal of solids
495 and Structures 38 (22-23) (2001) 3927–3944.
496 [13] C. Daux, N. Moës, J. Dolbow, N. Sukumar, T. Belytschko, Arbitrary branched and intersecting cracks with the extended finite element
497 method, International Journal for Numerical Methods in Engineering 48 (12) (2000) 1741–1760.
498 [14] M. Kalina, T. Schneider, J. Brummund, M. Kästner, Overview of phase-field models for fatigue fracture in a unified framework, Engineering
499 Fracture Mechanics 288 (2023) 109318. doi:10.1016/j.engfracmech.2023.109318.
500 [15] S. Yan, C. Schreiber, R. Müller, An efficient implementation of a phase field model for fatigue crack growth, International Journal of Fracture
501 237 (1-2) (2022) 47–60. doi:10.1007/s10704-022-00628-0.
502 [16] C. Miehe, M. Hofacker, F. Welschinger, A phase field model for rate-independent crack propagation: Robust algorithmic implementation
503 based on operator splits, Computer Methods in Applied Mechanics and Engineering 199 (45-48) (2010) 2765–2778. doi:10.1016/j.cma.
504 2010.04.011.
505 [17] H. Amor, J. J. Marigo, C. Maurini, Regularized formulation of the variational brittle fracture with unilateral contact: Numerical experiments,
506 Journal of the Mechanics and Physics of Solids 57 (8) (2009) 1209–1229. doi:10.1016/j.jmps.2009.04.011.

21
507 [18] C. Steinke, M. Kaliske, A phase-field crack model based on directional stress decomposition, Computational Mechanics 63 (5) (2019) 1019–
508 1046. doi:10.1007/s00466-018-1635-0.
509 [19] M. J. Borden, T. J. Hughes, C. M. Landis, C. V. Verhoosel, A higher-order phase-field model for brittle fracture: Formulation and analysis
510 within the isogeometric analysis framework, Computer Methods in Applied Mechanics and Engineering 273 (2014) 100–118.
511 [20] M. Ambati, T. Gerasimov, L. De Lorenzis, Phase-field modeling of ductile fracture, Computational Mechanics 55 (5) (2015) 1017–1040.
512 doi:10.1007/S00466-015-1151-4.
513 [21] C. Kuhn, T. Noll, R. Müller, On phase field modeling of ductile fracture, GAMM Mitteilungen 39 (1) (2016) 35–54. doi:10.1002/gamm.
514 201610003.
515 [22] C. Miehe, L.-M. Schaenzel, H. Ulmer, Phase field modeling of fracture in multi-physics problems. Part I. Balance of crack surface and failure
516 criteria for brittle crack propagation in thermo-elastic solids, Computer Methods in Applied Mechanics and Engineering 294 (2015) 449–485.
517 doi:10.1016/j.cma.2014.11.016.
518 [23] C. Miehe, M. Hofacker, L.-M. Schänzel, F. Aldakheel, Phase field modeling of fracture in multi-physics problems. Part II. Coupled brittle-
519 to-ductile failure criteria and crack propagation in thermo-elastic–plastic solids, Computer Methods in Applied Mechanics and Engineering
520 294 (2015) 486–522. doi:10.1016/j.cma.2014.11.017.
521 [24] C. Miehe, S. Mauthe, Phase field modeling of fracture in multi-physics problems. Part III. Crack driving forces in hydro-poro-elasticity and
522 hydraulic fracturing of fluid-saturated porous media, Computer Methods in Applied Mechanics and Engineering 304 (2016) 619–655.
523 [25] Z. A. Wilson, C. M. Landis, Phase-field modeling of hydraulic fracture, Journal of the Mechanics and Physics of Solids 96 (2016) 264–290.
524 [26] Y. Heider, A review on phase-field modeling of hydraulic fracturing, Engineering Fracture Mechanics 253 (2021) 107881.
525 [27] S. Teichtmeister, D. Kienle, F. Aldakheel, M.-A. Keip, Phase field modeling of fracture in anisotropic brittle solids, International Journal of
526 Non-Linear Mechanics 97 (2017) 1–21.
527 [28] F. Dammaß, M. Ambati, M. Kästner, A unified phase-field model of fracture in viscoelastic materials, Continuum Mechanics and Thermody-
528 namics 33 (4) (2021) 1907–1929.
529 [29] C. Xing, T. Yu, Y. Sun, Y. Wang, An adaptive phase-field model with variable-node elements for fracture of hyperelastic materials at large
530 deformations, Engineering Fracture Mechanics 281 (2023) 109115.
531 [30] Z. A. Wilson, M. J. Borden, C. M. Landis, A phase-field model for fracture in piezoelectric ceramics, International Journal of Fracture 183
532 (2013) 135–153.
533 [31] J.-Y. Wu, V. P. Nguyen, C. T. Nguyen, D. Sutula, S. Sinaie, S. P. Bordas, Phase-field modeling of fracture, Advances in Applied Mechanics
534 53 (2020) 1–183.
535 [32] R. Alessi, S. Vidoli, L. De Lorenzis, A phenomenological approach to fatigue with a variational phase-field model: The one-dimensional
536 case, Engineering Fracture Mechanics 190 (2018) 53–73. doi:10.1016/j.engfracmech.2017.11.036.
537 [33] C. Schreiber, C. Kuhn, R. Müller, T. Zohdi, A phase field modeling approach of cyclic fatigue crack growth, International Journal of Fracture
538 225 (1) (2020) 89–100. doi:10.1007/s10704-020-00468-w.
539 [34] Y. S. Lo, M. J. Borden, K. Ravi-Chandar, C. M. Landis, A phase-field model for fatigue crack growth, Journal of the Mechanics and Physics
540 of Solids 132 (2019) 103684. doi:10.1016/j.jmps.2019.103684.
541 [35] P. Carrara, M. Ambati, R. Alessi, L. De Lorenzis, A framework to model the fatigue behavior of brittle materials based on a variational phase-
542 field approach, Computer Methods in Applied Mechanics and Engineering 361 (2020) 112731. doi:10.1016/j.cma.2019.112731.
543 [36] R. Alessi, M. Ambati, T. Gerasimov, S. Vidoli, L. D. Lorenzis, Comparison of Phase-Field Models of Fracture Coupled with Plasticity,
544 Computational Methods in Applied Sciences (2018). doi:10.1007/978-3-319-60885-3.
545 [37] A. Golahmar, C. F. Niordson, E. Martı́nez-Pañeda, A phase field model for high-cycle fatigue: Total-life analysis, International Journal of
546 Fatigue 170 (2023) 107558. doi:10.1016/j.ijfatigue.2023.107558.
547 [38] M. Seiler, T. Linse, P. Hantschke, M. Kästner, An efficient phase-field model for fatigue fracture in ductile materials, Engineering Fracture
548 Mechanics 224 (2020) 106807. doi:10.1016/j.engfracmech.2019.106807.
549 [39] M. A. Miner, Cumulative damage in fatigue, Journal of Applied Mechanics 12 (3) (1945) A159–A164. doi:10.1115/1.4009458.
550 [40] T. Schneider, D. Müller, M. Seiler, T. Tobie, K. Stahl, M. Kästner, Phase-field modeling of fatigue crack growth during tooth flank fracture in
551 case-hardened spur gears, International Journal of Fatigue 163 (2022) 107091. doi:10.1016/j.ijfatigue.2022.107091.
552 [41] M. Seiler, S. Keller, N. Kashaev, B. Klusemann, M. Kästner, Phase-field modelling for fatigue crack growth under laser shock peening-
553 induced residual stresses, Archive of Applied Mechanics 91 (2021) 3709–3723.
554 [42] J. L. Boldrini, E. A. Barros de Moraes, L. R. Chiarelli, F. G. Fumes, M. L. Bittencourt, A non-isothermal thermodynamically consistent
555 phase field framework for structural damage and fatigue, Computer Methods in Applied Mechanics and Engineering 312 (2016) 395–427.
556 doi:10.1016/j.cma.2016.08.030.
557 [43] G. Haveroth, M. Vale, M. Bittencourt, J. Boldrini, A non-isothermal thermodynamically consistent phase field model for damage, fracture
558 and fatigue evolutions in elasto-plastic materials, Computer Methods in Applied Mechanics and Engineering 364 (2020) 112962. doi:
559 10.1016/j.cma.2020.112962.
560 [44] C. Schreiber, R. Müller, C. Kuhn, Phase field simulation of fatigue crack propagation under complex load situations, Archive of Applied
561 Mechanics 91 (2) (2021) 563–577. doi:10.1007/s00419-020-01821-0.
562 [45] P. Li, W. Li, B. Li, S. Yang, Y. Shen, A review on phase field models for fracture and fatigue, Engineering Fracture Mechanics 289 (2023)
563 109419. doi:10.1016/j.engfracmech.2023.109419.
564 [46] P. K. Kristensen, A. Golahmar, E. Martı́nez-Pañeda, C. F. Niordson, Accelerated high-cycle phase field fatigue predictions, European Journal
565 of Mechanics - A/Solids 100 (2023) 104991. doi:10.1016/j.euromechsol.2023.104991.

22
566 [47] A. Jaccon, B. Prabel, G. Molnár, J. Bluthé, A. Gravouil, Adaptive mesh refinement and cycle jumps for phase-field fatigue fracture modeling,
567 Finite Elements in Analysis and Design 224 (2023) 104004. doi:10.1016/j.finel.2023.104004.
568 [48] K. Seleš, T. Lesičar, Z. Tonković, J. Sorić, A residual control staggered solution scheme for the phase-field modeling of brittle fracture,
569 Engineering Fracture Mechanics 205 (2019) 370–386. doi:10.1016/j.engfracmech.2018.09.027.
570 [49] A. Gupta, U. M. Krishnan, T. K. Mandal, R. Chowdhury, V. P. Nguyen, An adaptive mesh refinement algorithm for phase-field fracture
571 models: Application to brittle, cohesive, and dynamic fracture, Computer Methods in Applied Mechanics and Engineering 399 (2022)
572 115347. doi:10.1016/j.cma.2022.115347.
573 [50] C. Xing, T. Yu, Y. Sun, Y. Wang, An adaptive phase-field model with variable-node elements for fracture of hyperelastic materials at large
574 deformations, Engineering Fracture Mechanics 281 (2023) 109115. doi:10.1016/j.engfracmech.2023.109115.
575 [51] S. Yang, Y. Shen, An acceleration scheme for the phase field fatigue fracture simulation with a concurrent temporal homogenization method,
576 Computer Methods in Applied Mechanics and Engineering 416 (2023) 116294. doi:10.1016/j.cma.2023.116294.
577 [52] T. Xue, S. Adriaenssens, S. Mao, Mapped phase field method for brittle fracture, Computer Methods in Applied Mechanics and Engineering
578 385 (2021) 114046.
579 [53] M. M. Hasan, T. Baxevanis, A phase-field model for low-cycle fatigue of brittle materials, International Journal of Fatigue 150 (2021) 106297.
580 doi:10.1016/J.IJFATIGUE.2021.106297.
581 [54] F. P. Duda, A. Ciarbonetti, P. J. Sánchez, A. E. Huespe, A phase-field/gradient damage model for brittle fracture in elastic-plastic solids,
582 International Journal of Plasticity 65 (2014) 269–296. doi:10.1016/j.ijplas.2014.09.005.
583 [55] J. Ulloa, J. Wambacq, R. Alessi, G. Degrande, S. François, Phase-field modeling of fatigue coupled to cyclic plasticity in an energetic
584 formulation, Computer Methods in Applied Mechanics and Engineering 373 (2021) 113473. doi:10.1016/j.cma.2020.113473.
585 [56] K. Seleš, F. Aldakheel, Z. Tonković, J. Sorić, P. Wriggers, A general phase-field model for fatigue failure in brittle and ductile solids,
586 Computational Mechanics 67 (5) (2021) 1431–1452. doi:10.1007/s00466-021-01996-5.
587 [57] Z. Khalil, A. Y. Elghazouli, E. Martı́nez-Pañeda, A generalised phase field model for fatigue crack growth in elastic-plastic solids with an
588 efficient monolithic solver, Computer Methods in Applied Mechanics and Engineering 388 (2022) 114286. doi:10.1016/j.cma.2021.
589 114286.
590 [58] Y. Tan, Y. He, X. Li, G. Kang, A phase field model for fatigue fracture in piezoelectric solids: A residual controlled staggered scheme,
591 Computer Methods in Applied Mechanics and Engineering 399 (2022) 115459. doi:10.1016/j.cma.2022.115459.
592 [59] Y. Tan, C. Liu, J. Zhao, Y. He, P. Li, X. Li, Phase field model for brittle fracture in multiferroic materials, Computer Methods in Applied
593 Mechanics and Engineering 414 (2023) 116193. doi:10.1016/j.cma.2023.116193.
594 [60] P. J. Loew, B. Peters, L. A. Beex, Fatigue phase-field damage modeling of rubber using viscous dissipation: Crack nucleation and propagation,
595 Mechanics of Materials 142 (2020) 103282. doi:10.1016/j.mechmat.2019.103282.
596 [61] Q. Xie, H. Qi, S. Li, X. Yang, D. Shi, A phase-field model for mixed-mode elastoplastic fatigue crack, Engineering Fracture Mechanics 282
597 (2023) 109176. doi:10.1016/j.engfracmech.2023.109176.
598 [62] J.-Y. Wu, Y. Huang, H. Zhou, V. P. Nguyen, Three-dimensional phase-field modeling of mode I+ II/III failure in solids, Computer Methods
599 in Applied Mechanics and Engineering 373 (2021) 113537.
600 [63] J.-Y. Wu, V. P. Nguyen, A length scale insensitive phase-field damage model for brittle fracture, Journal of the Mechanics and Physics of
601 Solids 119 (2018) 20–42.
602 [64] S. Lucarini, F. P. E. Dunne, E. Martı́nez-pañeda, An FFT-based crystal plasticity phase-field model for micromechanical fatigue cracking
603 based on the stored energy density, International Journal of Fatigue 172 (2023) 107670. doi:10.1016/j.ijfatigue.2023.107670.
604 [65] K. Appunhi, S. Ghosh, Crack tip enhanced phase-field model for crack evolution in crystalline Ti6Al from concurrent crystal plasticity FE-
605 molecular dynamics simulations, European Journal of Mechanics / A Solids 100 (2023) 104983. doi:10.1016/j.euromechsol.2023.
606 104983.
607 [66] Z. J. Li, T. Wang, D. Y. Chu, Z. L. Liu, Y. N. Cui, A coupled crystal-plasticity and phase-field model for understanding fracture behaviors of
608 single crystal tungsten, International Journal of Plasticity 157 (July) (2022). doi:10.1016/j.ijplas.2022.103375.
609 [67] M. Kamal, M. M. Rahman, Advances in fatigue life modeling: A review, Renewable and Sustainable Energy Reviews 82 (2018) 940–949.
610 doi:10.1016/j.rser.2017.09.047.
611 [68] W. Tang, Z. Tang, W. Lu, S. Wang, M. Yi, Modeling and Prediction of Fatigue Properties of Additively Manufactured Metals, Acta Mechanica
612 Solida Sinica (2023). doi:10.1007/s10338-023-00380-5.
613 [69] S. Kilambi, S. M. Tipton, Numerical evaluation of the original “Neuber’s rule” for pure out-of-plane shear loading, The Journal of Strain
614 Analysis for Engineering Design 48 (8) (2013) 522–535. doi:10.1177/0309324713496407.
615 [70] W. Ramberg, W. R. Osgood, Description of stress-strain curves by three parameters, Tech. rep., NASA Technical Reports Server (1943).
616 [71] H. Neuber, Theory of Stress Concentration for Shear-Strained Prismatical Bodies With Arbitrary Nonlinear Stress-Strain Law, Journal of
617 Applied Mechanics 28 (4) (1961) 544–550. doi:10.1115/1.3641780.
618 [72] T. Łagoda, S. Vantadori, K. Głowacka, M. Kurek, K. Kluger, Using the Smith-Watson-Topper Parameter and Its Modifications to Calculate
619 the Fatigue Life of Metals: The State-of-the-Art, Materials 15 (10) (2022). doi:10.3390/ma15103481.
620 [73] D. Kujawski, P. C. Sree, On deviatoric interpretation of Neuber’s rule and the SWT parameter, Theoretical and Applied Fracture Mechanics
621 71 (2014) 44–50. doi:10.1016/j.tafmec.2014.03.006.
622 [74] M. R. Tonks, D. Gaston, P. C. Millett, D. Andrs, P. Talbot, An object-oriented finite element framework for multiphysics phase field simula-
623 tions, Computational Materials Science 51 (1) (2012) 20–29. doi:10.1016/j.commatsci.2011.07.028.
624 [75] M. A. Terres, H. Sidhom, Fatigue life evaluation of 42CrMo4 nitrided steel by local approach: Equivalent strain-life-time, Materials and

23
625 Design 33 (1) (2012) 444–450. doi:10.1016/j.matdes.2011.04.047.
626 [76] A. A. Roostaei, Y. Ling, H. Jahed, G. Glinka, Applications of Neuber’s and Glinka’s notch plasticity correction rules to asymmetric magne-
627 sium alloys under cyclic load, Theoretical and Applied Fracture Mechanics 105 (2020) 102431. doi:10.1016/j.tafmec.2019.102431.
628 [77] M. D. Sangid, The physics of fatigue crack initiation, International Journal of Fatigue 57 (2013) 58–72. doi:10.1016/j.ijfatigue.
629 2012.10.009.
630 [78] A. Schmiedel, T. Kirste, R. Morgenstern, A. Weidner, H. Biermann, The fatigue life of 42CrMo4 steel in the range of HCF to VHCF at
631 elevated temperatures up to 773 K, International Journal of Fatigue 152 (2021) 106437. doi:10.1016/j.ijfatigue.2021.106437.
632 [79] D. F. Socie, Fatigue-life prediction using local stress-strain concepts: The primary emphasis of this paper is on the mechanics of a computer
633 algorithm for cumulative fatigue damage, Experimental Mechanics 17 (2) (1977) 50–56.
634 [80] G. Li, L. Ke, X. Ren, C. Sun, High cycle and very high cycle fatigue of TC17 titanium alloy: Stress ratio effect and fatigue strength modeling,
635 International Journal of Fatigue 166 (2023) 107299. doi:10.1016/J.IJFATIGUE.2022.107299.
636 [81] A. S. for Testing, Materials, Standard test method for measurement of fatigue crack growth rates: ASTM E 647-05, ASTM International,
637 2008. doi:10.1520/E0647-05.
638 [82] J. E. Srawley, Wide range stress intensity factor expressions for ASTM E 399 standard fracture toughness specimens, in: Conf. of Am. Soc.
639 for Testing and Mater., Committee E-24, no. E-8654, 1976.
640 [83] C. Dávila, From SN to the Paris law with a new mixed-mode cohesive fatigue model for delamination in composites, Theoretical and Applied
641 Fracture Mechanics 106 (2020) 102499. doi:10.1016/j.tafmec.2020.102499.
642 [84] Z. Man, W. Yang, L. Qiang, Y. Min, Prediction of mode II fatigue delamination propagation in fibre reinforced composites: From G-N curve
643 to Paris’ law, Acta Materiae Compositae Sinica 40 (2022) 1–14.
644 [85] P. Miarka, S. Seitl, V. Bı́lek, H. Cifuentes, Assessment of fatigue resistance of concrete: S-N curves to the Paris’ law curves, Construction
645 and Building Materials 341 (2022) 127811. doi:10.1016/j.conbuildmat.2022.127811.
646 [86] A. Noroozi, G. Glinka, S. Lambert, A two parameter driving force for fatigue crack growth analysis, International Journal of Fatigue 27 (10)
647 (2005) 1277–1296. doi:10.1016/j.ijfatigue.2005.07.002.
648 [87] G. Lesiuk, M. M. Duda, J. Correia, A. M. de Jesus, R. Calçada, Fatigue crack growth of 42CrMo4 and 41Cr4 steels under different heat
649 treatment conditions, International Journal of Structural Integrity 9 (3) (2018) 326–336. doi:10.1108/IJSI-01-2018-0003.
650 [88] G. Sui, Z. Wang, F. Jiang, C. Zhao, Y. Guo, W. Li, Fatigue crack growth behavior of 42CrMo high-strength steel with tempered sorbite/bainite
651 microstructures: Roles of grain and constituent in microstructures, Engineering Fracture Mechanics 279 (2023) 109051. doi:10.1016/j.
652 engfracmech.2023.109051.

24