Wear 400–401 (2018) 177–183

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Finite element simulation of the influence of fretting wear on fretting crack T

initiation in press-fitted shaft under rotating bending
⁎
Y.B. Zhang, L.T. Lu , L. Zou, D.F. Zeng, J.W. Zhang
State Key Laboratory of Traction Power, Southwest Jiaotong University, Chengdu 610031, China

A R T I C L E I N F O A B S T R A C T

Keywords: A Finite element based methodology for fretting wear and fretting fatigue prediction in press-fitted shaft is
Fretting wear developed. This methodology integrates wear modelling with fretting fatigue analysis to quantitatively predict of
Fretting fatigue the effects of stress redistribution and material removal induced by the fretting wear on fretting crack initiation
Press-fitted shaft properties. The fretting wear-induced evolutions of mean normal stress, a mean correction multiaxial damage
Finite element
parameter and accumulated fatigue damage are predicted by this methodology. The results show that the
Crack initiation
fretting crack initiation at the contact edge is significantly suppressed by the severe fretting wear, since the
server stress concentration is relieved and the surface material with high accumulated fatigue damage is grinded
off. While, the fretting crack initiation at the inner surface of the contact area is greatly promoted by the slight
wear by introducing the large stress concentration near the edge of fretted wear scar.

1. Introduction
When the press-fitted assembly, for instance, the railway wheelsets
[1–3] and gear hubs/shafts [4], is subjecting the rotating bending loads,
micro-slip occurs near the contact edge and the fretting fatigue happens
inevitably. Consequently, the durability of the assembly is reduced
significantly [5].
The press-fitted part, such as the wheel seat, is the critical part of the
railway axle, since the cracks are easily initiated because of the fretting
fatigue [6]. Thus great emphasis has been placed on the fatigue cracks
of railway axle around the world. In Japan [7], if there is a crack larger
than 0.15 mm in depth, the railway axle will be scrapped. Therefore,
expensive regular maintenance and inspection are indispensable to re-
gard the fretting crack initiation. It has been reported that the crack
initiation life is often short, which only reaches approximately 10–30%
of the total fretting fatigue life for the press-fitted shaft [8,9]. Never-
theless, it is still necessary to investigate the fretting crack initiation
properties in press-fitted shaft, thereby scientific guidance for the reg-
ular maintenance and inspection of the railway axles can be provided.
In order to study the fretting damage of press-fitted shaft, both Lee
et al. [8] and Zhang et al. [9] have carried out the interrupted fretting
fatigue tests, and the results show that the contact surface profile
evolves with the fretting cycles gradually due to the fretting wear.
Meanwhile, the experimental studies [1,5,8–11] conducted with the
press-fitted shaft show that the fretting cracks are found at the inner
surface of the contact area instead of the contact edge where there is a
very high stress concentration. It can be seen that the fretting wear has
a significantly effect on the fretting crack initiation properties. There-
fore, Kubota et al. [10] pointed that to clarify the mechanism of the
crack initiation in press-fitted axle, a comprehensive investigation re-
lated to the balance between the crack initiation and fretting wear and
the stress condition surrounding the incipient cracks is necessary.
However, the corresponding researches are very limit for the press-
fitted shaft. Zeise et al. [12] has presented a finite element (FE) based
methodology to predict the fretting wear in press-fitted shaft, and the
evolutions of the contact surface profile and the contact parameters are
obtained, however the influence of fretting wear on the fretting crack
initiation properties is not clarified in detail. Lee et al. [13] also pre-
sented a numerical analysis of the fretting fatigue of press-fitted shaft
considering the effect of fretting wear, however the investigation is
mainly focused on the effect of contact pressure. In previous study
presented by Zhang et al. [9], in order to discuss the formation me-
chanism of each fretting wear zone, a FE model with the fretted wear
scar for interrupted specimens was developed, while the FE model did
not contain wear model and the fretted wear scars were artificially
added to the FE model based on the measured surface profiles. The
effect of fretting wear on fretting fatigue of press-fitted shaft is dis-
cussed qualitatively and simply. Therefore, the influence of fretting
wear on the fretting crack initiation properties in press-fitted shaft still
far from thorough research.
Comparing with the press-fitted shaft, the fretting wear and fretting
fatigue problems of other applications [14–20] have been well

⁎
Corresponding author.
E-mail address: luliantao@swjtu.cn (L.T. Lu).

https://doi.org/10.1016/j.wear.2018.01.008
Received 10 July 2017; Received in revised form 16 December 2017; Accepted 17 January 2018
Available online 02 February 2018
0043-1648/ © 2018 Elsevier B.V. All rights reserved.
Y.B. Zhang et al.
investigated. More recently McColl et al. [18] developed an incremental
wear approach using Archard equation [21] for the simulation of fret-
ting wear in cylinder on flat configuration. Follow this work, Madge
et al. [14] combines the Archard-based wear model with the critical
plane SWT [22] damage accumulation fatigue model to predict the
effect of fretting wear on fretting fatigue. It turned out that the non-
linear relationship between fatigue life and the slip range obtained from
the experiment [23] can be reasonably interpreted, when the impact of
fretting wear is considered. Then the FE based methodology proposed
by Madge is widely used for complex configurations, such as the spline
couplings [15], thin steel wires [16] and modular hip implants [17]. It
turned out that the predicted fretting fatigue life is more accuracy after
considering the fretting wear effect.
In this paper, a FE based methodology similar with the one proposed
by Madge [14] is developed for the press-fitted shaft. In this metho-
dology, the wear model used for the fretting wear simulation is the
energy-based wear approach proposed by Fouvry et al. [24]. This ap-
proach relates the wear volume to the accumulated dissipated energy
which considers the impact of tangential force. It has been validated by
the experiments [25,26] that this approach is superior to the Archard-
based wear model; Since the SWT parameter might be non-conservative
[27,28], for the cyclic loads that involve relatively large compressive
mean stress, the SWT parameter is replaced by a mean correction model
proposed by Ince et al. [28]. The fretting wear-induced evolutions of
the parameters related to the fretting fatigue have been investigated.
The influence mechanisms of fretting wear on the fretting crack in-
itiation site and initiation life in press-fitted shaft have been discussed
in detail, with the combination of the experimental results presented in
previous work [9] and the simulation results.
2. Experimental details
An extended description of the experimental details and results used
to study the fretting wear-induced surface damage in press-fitted shaft
is reported in [9]. In this section, the specimens, materials, test rig and
load conditions for the interrupted fretting fatigue tests are briefly
outlined.
2.1. Specimens and materials
Fig. 1 shows the shape and dimension of the press-fitted specimen.
The diametral interference 2δ is approximately 21 µm and the corre-
sponding nominal contact pressure P is approximately 141 MPa. Alloy
axle steel (35CrMo) and ER8 wheel steel are used for the shaft and hub,
respectively. And the corresponding mechanical properties are shown
in Table 1.
Fig. 1. Specimen shape and dimension (mm).
178
Wear 400–401 (2018) 177–183
2.2. Fretting tests
The interrupted fretting fatigue tests were conducted on a rotating
bending fatigue machine at four different predefined cycles of 10, 30
50%, and 70% of the total fatigue life. The rotary speed is approxi-
mately 3000 rpm. As it has been mentioned in previous study [9], the
nominal bending stress on the press-fitted part is set to 220 MPa, and
the corresponding total fatigue life Nf is approximately 1.31×106 cy-
cles. After the tests, as it is shown in Fig. 2, all samples were sectioned
by wire electrical discharge machine and then cleaned ultrasonically.
The fretted surface and the fretting cracks near contact edge are ob-
served by a scanning electron microscope (SEM). The contact surface of
the shaft at 10% of the total fatigue life, observed by using SEM,
showed no micro-cracks. Meanwhile, the shaft at 30% of the total fa-
tigue life (N = 393000 cycles), as shown in Fig. 3, possesses micro-
cracks of 30–50 µm in length located approximately 200 µm away from
the contact edge. The surface profiles near contact edge are measured
by a confocal laser scanning microscope (CLSM). The detail processes of
surface profile measurement were presented in [9].
3. FE based methodology
3.1. FE model
The FE model used in this study is the same that has been validated
in a previous work [29], for the simulation fretting wear in press-fitted
shaft. The model consists of shaft and hub as shown in Fig. 4, due to the
symmetry structure and loading, only half of the press-fitted specimen
is created. The axial direction parallel to the X-axis, and the origin of X-
axis locates at the contact edge. Three-dimensional eight-node linear
elements (C3D8) are employed. Based on the optimization conducted in
previous study [29], the highly refined meshes with the size of 15 µm
are adequate to capture the complicated variation of stresses, relative
slip and to give a well representation of the wear scar profile. A master-
slave contact algorithm is employed, and according to the optimal re-
sult of the coefficient of friction (COF) in previous paper [9], a value of
0.6 was used in this study. The penalty approach is used in this paper,
due to the fact that the Lagrange multiplier contact algorithm induces
the convergence problem under cyclic rotating bending applied loads.
To ensure that the penalty approach gives as good results as the La-
grange multiplier contact algorithm, the optimization of the maximum
elastic slip tolerance has been conducted in previous paper [9] and
according to the optimal result, the maximum elastic slip tolerance is
set to 0.0001 in this study.
The elastic–plastic material behaviour is modelled using linear ki-
nematic hardening condition with the material properties shown in
Table 1
Fig. 4(a) shows the loads and constraints imposed on the FE model.
The symmetric constraint along x-axis is applied to the centre of FE
model and the end of the shaft (point A) is fixed in y and z directions.
The concentrated forces varying with the computation time t, Fy =
F0cos(2πt) and Fz = F0sin(2πt), are applied at point B along y-axis and z-
axis respectively, and as a result, the constant bending load F0 rotating
around the axis of shaft with computation time t is applied to the shaft.
Therefore, one fretting fatigue cycle can be simulated within one load
step. More details about the loads and constraints are presented in
previous work [29].
3.2. Wear modelling
The specific method used in this study for the wear simulation of
press-fitted shaft, which was validated by the experimental results, is
described in detail in the previous work reported by Zhang et al. [29].
The wear model used for fretting wear simulation is the energy wear
approach proposed by Fouvry et al. [24]. This approach assumes that
fretting wear can be evaluated by applying local energy wear approach
Y.B. Zhang et al. Wear 400–401 (2018) 177–183

Table 1
Mechanical properties of materials.

Young's modulus (GPa) Cyclic yield strength (MPa) Plastic modulus (GPa) Poisson's ratio Micro-hardness HV0.1

Shaft (alloy axle steel) 212 705 4.5 0.28 310

Hub(ER8 steel) 212 600 8 0.28 286

to local contact conditions along the entire contact interface. The local
wear model is described by:
Δhinc (j, m) = αq (j, m) Δs (j, m) (1)

Where Δhinc(j,m), q(j,m) and Δs(j,m) is the incremental wear depth, the
shear traction and the relative slip for a contact node j at the mth in-
crement time, respectively. α is the energy wear coefficient.
As it has been mentioned in FE model, one fretting cycle can be
simulated within one load step. Supposing there are n wear simulation
increments per load step, and the total wear depth during one step is:
n
Δhstep (j ) = ∑ αq (j, m) Δs (j, m)
t=1 (2)
Fig. 2. Final cut sample. In order to efficiently simulate the wear depth, the cycle-jumping
technique, which has been successfully used in the energy-based
[17,20] wear approach, is adopted in this study. The nodal wear depth
Contact edge Axial direction Δh(j) corresponding to ΔN fretting cycles can be simulated in one load
step by:
n
Δh (j ) = ΔN ∑ αq (j, m) Δs (j, m)
t=1 (3)

Based on the optimization conducted in previous work [29], the

cycle jump ΔN and the increments in one FE cycle are set to 3000 and
20, respectively, which can reduce the computational time without
sacrificing the stability and accuracy of analysis.
The UMESHMOTION user subroutine within ABAQUS [30] is
adopted to implement the wear simulation algorithm in this paper.
Fig. 3. Micrographs of fretting crack on the contact surface at 30% of the total fatigue life
More details about the process of energy-based wear simulation are
(N = 393000 cycles).
presented in previous work [29].
The energy wear coefficient of the shaft was set to αshaft = 4.17 ×
10−8 MPa−1, based on the fretting wear test carried out with ball on
Contact cen
nter
flat configuration [31]. The energy wear coefficient of the hub has not
Fixed in Y an
nd
Load F0 C
Contact edge
e Hub been found in the current literature. However, based on the fretted
wear scar profiles of shaft and hub obtained in previous study [9], the
wear depth of hub is about 1.3 times of that of shaft. Therefore, the
Sha
A B experimental results in previous study were used to relate the hub wear
coefficient to the shaft wear coefficient (the 1.3 factor), and the energy
wear coefficient of hub was set to αhub = 5.42 × 10−8 MPa−1.

Symmetry
3.3. Fatigue modelling
along X-axis
(a) Due to the fact that the fretted zone is subjected to complex loading
conditions, the crack initiation characteristics of fretting fatigue are
1mm usually predicted by the multi-axial critical plane approaches (SWT
Contact edge [22] and F-S [32] criteria). A. Cruzado [16] has indicated that the
Hub
H
bending induced tensile stress leads to fretting cracks in mode I.
Therefore, comparing to the F-S parameter which is suitable to shear
cracking failure mode [32], the SWT parameter and the modifications
Refined meesh of the SWT parameter [28,33] might be more suitable to the press-fitted
Sha
S shaft under rotating bending.
The SWT parameter can be expressed as follows:

(b)
Fig. 4. FE model (a) overall geometry of press-fitted specimen, (b) refined mesh in the
region of contact edge indicated by rectangle in (a).
σ ′f 2
SWT = σmax Δεa = (2Ni )2b + σ ′f ε′f (2Ni )b + c
E (4)
where σmax and Δεa are the maximum normal stress and maximum
normal strain amplitude on the critical plane respectively, E is the
Young modulus, σ’f is the fatigue strength coefficient, ε’f is the fatigue

179
Y.B. Zhang et al.
Table 2
Strain-life constants for 35CrMo (from Ren et al. [34]).
σ’f (MPa) ε’f b c
1146.78 0.95 −0.075 −0.81
ductility coefficient, b is the fatigue strength exponent, c is the fatigue
ductility exponent, and Ni represents the crack initiation life. The strain-
life constants for 35CrMo are given in Table 2, and they are given by
Ren et al. [34].
It is known that the mean stress has a significant influence on the
fatigue life [27]. Both the Morrow model, Eq. (5) [35], and the SWT
parameter, Eq. (4), can give good estimation of mean stress effect, and
as a result, they are all widely used.
σ ′f − σmean
Δεa = (2Ni )b + ε′f (2Ni )c
E (5)
It has been concluded that [35–37] when the compressive mean
stresses is very large, both of the Morrow and SWT models have their
shortcomings. Therefore, Ince and Glinka [28] combined the good
features of the Morrow and SWT models, and proposed a mean cor-
rection model, as follow:
σmax e σ ′f
Δεeq, a = Δεae, eq + Δεap = Δεa + Δεap = (2Ni )2b + ε′f (2Ni )c
′
σf E (6)
σmax
where Δεeq, a is an total equivalent strain amplitude, Δεae, eq = Δεae
σ ′f
an equivalent elastic strain amplitude corrected for the mean stress
effect, σmax is the maximum normal stress, Δεae and Δεap are the elastic
and plastic normal strain amplitudes, respectively. The accuracy of the
predicted results [28] is improved significantly compared to those of
the SWT parameter Eq. (4) and the Morrow's model Eq. (5).
For the press-fitted shaft investigated in current study, large com-
pressive mean stresses can develop near contact edge under rotating
bending. Therefore, with the purpose to study the fretting crack in-
itiation properties in press-fitted shaft under rotating bending, the mean
correction model, Eq. (6), has been implemented.
The implementation of the mean correction model, Eq. (6), here is
approximately follows the detailed procedure previously described by
Sum et al. [38] for the SWT parameter, Eq. (4). In this paper, the angle-
interval of the candidate plane is set to 1°, and as a result, there are
32400 candidate planes for each element. The σmax, Δεae , Δεap and Δεeq, a
are obtained for each candidate plane and each element in one cycle.
The critical plane with the maximum, Δεeq, a , is obtained for each ele-
ment, and the corresponding crack initiation life can be calculated by
solving Eq. (6). Meanwhile, the orientation of the critical plane can be
obtained.
The stresses and strains near contact surface change from cycle to
cycle because of the changing contact surface. The fatigue model
combining the mean correction model, Eq. (6), and the Miner–-
Palmgren (M-P) linear fatigue accumulation damage rule, Eq. (7), is
adopted to capture this phenomenon. The crack initiation life Ni,l is
predicted by Eq. (6), and the corresponding damage Dl is calculated in
each element for each FE fretting cycle l, then the corresponding total
damage D in each element through all the FE fretting cycles is accu-
mulated by M-P rule. When the D reaches 1, the crack initiation hap-
pens. Since a cycle jump technique described above has been used in
this paper, the modified M-P rule as follows is used:
N dually with the increasing fretting cycles.
ΔN
ΔN
D= ∑ Ni, l
l=1 (7)
where ΔN is the cycle jump value, Ni,l is the crack initiation life pre-
dicted by the mean correction model, Eq. (6), at fretting FE cycle l, N is
the total number of different load cycles experienced, N/ΔN is the total
Wear 400–401 (2018) 177–183
number of FE fretting cycles.
On the other hand, due to the use of adaptive meshing to update the
mesh on removal of material, specific elements and nodes are no longer
linked uniquely to actual material points throughout the analysis. So
the effect of material removal in the damage eradication has to be
considered. To solve this problem Madge et al. [14] proposed to create
a material point mesh (MPM) as the global reference model for damage
accumulation. The nodes of the MPM have fixed coordinates
throughout the analysis. Cyclic damage is calculated at the centroid of
each element, and linearly interpolated back to the MPM for accumu-
lation, and the nodes corresponding to the removed material do not
accumulated any fatigue damage. In this study, the damage accumu-
lation model created by Madge et al. [14] is employed to consider the
effect of material removal in the damage accumulation.
4. Results and discussions
The validation of the FE wear model has been conducted by pre-
vious work [29]. It shows that the FE wear model can give relatively
accurate prediction of the peak value of fretted wear scar for both shaft
and hub, and the scar width predicted by FE wear model is approxi-
mately in agreement with that of the experimental result. However, we
need to acknowledge that the prediction of wear depth inside of the
fretted wear scar is slightly larger than that of the experimental result.
More details about the validation of the FE wear model are presented in
previous work [29].
Even though the FE wear model cannot simulate the process of
is
fretting wear very accurately, it can satisfy the investigation of the
fretting wear-induced evolution of the stress and strain near contact
surface, hereafter, the crack initiation properties of press-fitted shaft
can be investigated by adopting the fatigue model proposed above.
4.1. Fretting wear-induced evolutions of the mean normal stress σmean and
the total equivalent strain amplitude εeq, a
It is known that the fretting wear contributes to the material re-
moval, which will induce the evolution of contact surface profile and
the stress redistribution near contact edge. Therefore, in order to un-
derstand the influence mechanism of fretting wear on the fretting crack
initiation properties, it is necessary to investigate the fretting wear-in-
duced evolutions of the parameters related to the fretting fatigue, for
instance, the mean normal stress σmean and the total equivalent strain
amplitude εeq, a. Firstly, it is necessary to explain in advance that due to
the material removal induced by fretting wear, the surface profiles of
shaft and hub are updated all the time during the whole simulation
process, therefore, all the parameters discussed below are the ones on
the new outermost surface of shaft after the material removal.
Fig. 5 shows the predicted evolution of the mean normal stress σmean
distribution on the new outermost surface of shaft with increasing
numbers of fretting cycles. A large compressive mean stress σmean is
predicted near the contact edge during the entire fretting process. This
phenomenon indicates that it is necessary to use the mean correction
model given by Eq. (6) to predict the crack initiation life of press-fitted
shaft. For the unworn case, the minimum value of σmean is predicted at
the contact edge; with the increase of fretting cycles, the depth and
width of fretted wear scar increases gradually, the predicted σmean at
the contact edge increases rapidly and the location of the minimum
value is predicted to move towards inside of contact area with the edge
of wear scar. By the way, the minimum value of σmean decreases gra-
The predicted evolution of the total equivalent strain amplitude
Δεeq, a distribution on the new outermost surface of shaft with increasing
numbers of fretting cycles is shown in Fig. 6. Because of the severe
stress concentration induced by the interference fit, the maximum value
of the total equivalent strain amplitude ΔεeqA, a is predicted to locate at
the contact edge for the unworn shaft. With increased fatigue cycling,
180
Y.B. Zhang et al. Wear 400–401 (2018) 177–183

Fig. 5. Predicted evolution of the mean normal stress σmean distribution on the new
outermost surface of shaft with increasing numbers of fretting cycles.

Fig. 7. Predicted evolution of the accumulated damage D distribution on the new out-
ermost surface of shaft with increasing numbers of fretting cycles.
Fig. 6. Predicted evolution of the total equivalent strain amplitude Δεeq, a distribution on
the new outermost surface of shaft with increasing numbers of fretting cycles.
Table 3
Comparisons of the predicted results and the experimental data.
the stress concentration at the contact edge is significantly relieved by
fretting wear. Consequently, the maximum value of the total equivalent Initiation life Ni Initiation location
strain amplitude ΔεeqA, a at contact edge is decreased significantly. As it is
shown in Fig. 6, for the inner surface of the contact area, another stress Experimental 131000 < Ni < 393000 50–200 µm away from the
results [9] contact edge
concentration appears and the corresponding maximum value of the Prediction from Ni = 318000 Approximately 250 µm away
total equivalent strain amplitude ΔεeqB , a is predicted to locate approxi- Zone A from the contact edge
mately near the edge of wear scar. With the increase of the width of Prediction from Ni = 23000 The contact edge
fretted wear scar, the location of the maximum value ΔεeqB , a moves to- Zone B
wards inside of contact area with the edge of wear scar. With the in-
crease of the depth of fretted wear scar, the stress concentration at the
contact edge is relieved significantly and part of the stress concentra- area and the maximum values were decreased gradually. While for the
tion induced by the interference fit shifts from the contact edge to the model used in current study, the wear scar is predicted by fretting wear
edge of wear scar, as a result, the maximum value ΔεeqB , a increases model, and both the stress concentrations near contact edge and near
gradually. the edge of wear scar are predicted successfully. By comparison, it can
In previous study [9], the FE model with the fretted wear scar of the be found that the FE wear model is more accurate and reasonable than
interrupted specimen has been created and the evolutions of the SWT the FE model used in previous study, and the FE wear model is neces-
and F-S parameters distributions with increasing numbers of fretting sary for the quantitative investigation of the fretting wear-induced
cycles have been investigated qualitatively. Since the wear scar was evolutions of the stress and strain near contact surface.
artificially added, the FE model in previous study could not predict the
severe stress concentration near contact edge induced by the inter-
ference fit, and only the new stress concentration near the edge of wear 4.2. Influence of fretting wear on the fretting crack initiation properties
scar was simulated successfully. The locations of the maximum values
of the SWT and F-S parameters were moved inward towards the contact As it is discussed in above section, the stress redistribution near
contact edge induced by the fretting wear dramatically impacts the

181
Y.B. Zhang et al.
distributions of the mean normal stress σmean, and the total equivalent
strain amplitude εeq, a. Therefore, it can imagine that the stress redis-
tribution near contact edge would greatly impact the fretting crack
initiation properties. Meanwhile, according to the quantitative simu-
lation of cylinder on flat configuration by Madge et al. [14], the ma-
terial removal induced by severe fretting wear would wear away the
incipient fretting fatigue cracks, and therefore suppress the fretting
crack initiation. In this current study, in order to quantitatively consider
the effect of material removal in the damage eradication, the damage
accumulation model using MPM technique created by Madge et al. [14]
was employed.
Fig. 7 shows the predicted evolution of accumulated damage D
distribution on the new outermost surface of shaft with increasing
numbers of fretting cycles. As it is shown, because of the severe stress
concentration induced by the interference fit, one crack initiation zone
(Zone A) is predicted to appear at the contact edge of press-fitted shaft.
The maximum accumulated damage DA is predicted to appear at the
same location of the maximum value of the total equivalent strain
amplitude ΔεeqA, a . In the early stage, as shown in Fig. 7(a), due to the
slight fretting wear and the severe stress concentration, the predicted
accumulated damage D in Zone A increases rapidly. With the increase
of fatigue cycle, as shown in Fig. 7(b), the worsening of fretting wear
induces the great decrease of stress concentration near contact edge,
and by the way, the surface material with high accumulated fatigue
damage is removed by fretting wear and replaced by the less damaged
material. As a result, the predicted accumulated damage D on new
outermost surface in Zone A decreases gradually, and the fretting crack
initiation at the contact edge is significantly suppressed by the severe
fretting wear. As it is shown in Fig. 7, according to the evolution of
accumulated damage in Zone A, the fretting crack is predicted to in-
itiate at the cycle of 23000 at the contact edge.
As it is shown in Fig. 7, for the inner surface of the contact area, due
to the effect of fretting wear, another stress concentration is predicted
near the edge of wear scar, and another crack initiation zone (Zone B) is
predicted to appear. In the early stage, as shown in Fig. 7(a), due to the
slight stress concentration, the accumulated fatigue damage in Zone B
increases slowly. With the increase of fatigue cycle, as shown in
Fig. 7(b), the stress concentration in Zone B increases rapidly, mean-
while the fretting wear in Zone B is slight. As a result, the material with
high accumulated fatigue damage is hard to be removed by the fretting
wear and the accumulated damage in Zone B increase rapidly. As it is
shown in Fig. 7, according to the evolution of accumulated damage in
Zone B, the fretting crack is predicted to initiate at the cycle of 318000.
And the corresponding predicted initiation location is approximately
250 µm away from the contact edge.
The comparisons of the predicted results and the experimental data
of the fretting crack initiation properties are show in Table 3. According
to the experimental results presented in previous study [9] the fretting
fatigue cracks initiates from the contact area 50–200 µm away from the
contact edge, the crack initiation life is longer than 131000 cycles and
shorter than 393000 cycles. As it is shown in Table 3, the predicted
cycle and location to the crack initiated in Zone B is much more con-
sistent with the experimental result than that in Zone A.
In Zone B, the fretting wear is slight while the stress concentration
induced by wear is severe, as a result, the fretting wear greatly pro-
motes the fretting crack initiation. The interactions between fretting
wear and fretting fatigue is similar with that of the cylinder on flat
configuration under partial slip condition as presented by Madge et al.
[14]. Based on the consistent of the predictions from Zone B and ex-
perimental results, it can be seen that, for the inner surface of the
contact area, the model is able to quantitatively clarify the influence of
fretting wear on the crack initiation life and location.
In Zone A, the fretting wear is severe and the severe stress con-
centration induced by the interference fit is greatly relieved, as a result,
the fretting wear significantly suppresses the fretting crack initiation.
The interactions between fretting wear and fretting fatigue is similar
182
Wear 400–401 (2018) 177–183
with that of the cylinder on flat configuration under gross slip condition
as presented by Madge et al. [14]. While, based on the difference be-
tween the predictions from Zone A and experimental results, it can be
seen that, only the qualitative interaction and competition between
fretting wear and fretting fatigue initiation in Zone A can be given by
this model.
5. Conclusions
This paper has presented a finite element based methodology which
combines a FE wear model and a mean correction multiaxial damage
accumulation approach for the simulation of the fretting crack initia-
tion properties in press-fitted shaft. The fretting wear-induced evolu-
tions of the parameters related to the fretting fatigue have been in-
vestigated. The influence mechanisms of fretting wear on the fretting
crack initiation site and initiation life in press-fitted shaft have been
discussed in detail. The main conclusions are:
• Due to the stress redistribution induced by fretting wear, the severe
stress concentration at the contact edge induced by the interference
fit is relieved significantly and another stress concentration appears
at the inner surface of contact area approximately near the edge of
wear scar.
• The severe fretting wear significantly suppresses the fretting crack
initiation at the contact edge by greatly reducing the stress con-
centration and grinding off the surface material with high accu-
mulated fatigue damage.
• The fretting wear greatly promotes the fretting crack initiation at
the inner surface of the contact area by introducing the large stress
concentration near the edge of fretted wear scar.
Acknowledgments
This work is supported by the National Natural Science Foundation
of China (51375406)
References
[1] S. Foletti, S. Beretta, G. Gurer, Defect acceptability under full-scale fretting fatigue
tests for railway axles, Int. J. Fatigue 86 (2016) 34–43.
[2] S. Maciej, M. Maciej, A method of fretting wear reduction in an automatic wheel set
gauge change system, Eng. Fail. Anal. 45 (2014) 363–375.
[3] K. Mädler, T. Geburtig, D. Ullrich, An experimental approach to determining the
residual lifetimes of wheelset axles on a full-scale wheel-rail roller test rig, Int. J.
Fatigue 86 (2016) 58–63.
[4] C.E. Truman, J.D. Booker, Analysis of shrink-fit failure on a gear hub/shaft as-
sembly, Eng. Fail. Anal. 14 (2007) 557–572.
[5] B. Alfredsson, Fretting fatigue of a shrink-fit pin subjected to rotating bending:
experiments and simulations, Int. J. Fatigue 31 (2009) 1559–1570.
[6] K. Hirakawa, M. Kubota, On the fatigue design method for high-speed railway axles,
Proc. Inst. Mech. Eng. 215 (2001) 73–82.
[7] H. Ishizuka, Thirty years of axles maintenance for Japan bullet train-Shinkansen, in:
Proceedings of 11th international wheelset congress, Australia; 2, 1995, pp. 43–46.
[8] D.H. Lee, S.J. Kwon, J.B. Choi, Y.J. Kim, Observation of fatigue damage in the press-
fitted shaft under bending loads, Key Eng. Mater. 326–328 (2006) 1071–1074.
[9] Y.B. Zhang, L.T. Lu, Y.B. Gong, J.W. Zhang, D.F. Zeng, Fretting wear-induced
evolution of surface damage in press-fitted shaft, Wear 284–285 (2017) 131–141.
[10] M. Kubota, S. Niho, C. Sakae, Y. Kondo, Effect of understress on fretting fatigue
crack initiation of press-fitted axle, JSME Int. J. 46 (2003) 297–302.
[11] C. Song, M.X. Shen, X.F. Lin, D.W. Liu, M.H. Zhu, An investigation on rotatory
bending fretting fatigue damage of railway axles, Fatigue Fract. Eng. Mater. Struct.
37 (2014) 72–84.
[12] B. Zeise, R. Liebich, M. Prölß, Simulation of fretting wear evolution for fatigue
endurance limit estimation of assemblies, Wear 316 (2014) 49–57.
[13] D.H. Lee, S.J. Kwon, W.H. You, Numerical analysis of the effect of contact pressure
on the fretting fatigue life in a press-fitted shaft, J. Korean Phys. Soc. 54 (2009)
1115–1118.
[14] J.J. Madge, S.B. Leen, I.R. McColl, P.H. Shipway, Contact evolution based predic-
tion of fretting fatigue life: effect of slip amplitude, Wear 262 (2007) 1159–1170.
[15] J. Ding, S.B. Leen, E.J. Williams, P.H. Shipway, Finite element simulation of fretting
wear-fatigue interaction in spline couplings, Tribol. Mater. Surf. Interfaces 2 (2008)
10–24.
[16] A. Cruzado, S.B. Leen, M.A. Urchegui, X. Gómez, Finite element simulation of
fretting wear and fatigue in thin steel wires, Int. J. Fatigue 55 (2013) 7–21.
Y.B. Zhang et al.
[17] T. Zhang, P.F. Harrison, P.F. McDonnell, P.E. McHugh, S.B. Leen, A finite element
methodology for wear-fatigue analysis for modular hip implants, Tribol. Int. 65
(2013) 113–127.
[18] I.R. McColl, J. Ding, S.B. Leen, Finite element simulation and experimental vali-
dation of fretting wear, Wear 256 (2004) 1114–1127.
[19] A. Cruzado, M.A. Urchegui, X. Gómez, Finite element modeling and experimental
validation of fretting wear scars in thin steel wires, Wear 289 (2012) 26–38.
[20] T. Zhang, P.E. McHugh, S.B. Leen, Computational study on the effect of contact
geometry on fretting behavior, Wear 271 (2011) 1462–1480.
[21] J.F. Archard, Single contacts and multiple encounters, J. Appl. Phys. 32 (1961)
1420–1425.
[22] K.N. Smith, P. Watson, T.H. Topper, A stress–strain function for the fatigue of
metals, J. Mater. 5 (1970) 767–778.
[23] O. Vingsbo, S. Soderberg, On fretting wears, Wear 126 (1988) 131–147.
[24] S. Fouvry, T. Liskiewicz, P. Kapsa, S. Hannel, E. Sauger, An energy description of
wear mechanisms and its applications to oscillating sliding contacts, Wear 255
(2003) 287–298.
[25] S. Fouvry, P. Duo, Ph Perruchaut, A quantitative approach of Ti–6A1–4V fretting
damage: friction, wear and crack nucleation, Wear 257 (2004) 916–929.
[26] K. Elleuch, S. Fouvry, Experimental and modelling aspects of abrasive wear of a
A357 aluminium alloy under Gross Slip Fretting Conditions, Wear 258 (2005)
40–49.
[27] T. Zhao, Y. Jiang, Fatigue of 7075-T651 aluminum alloy, Int. J. Fatigue 30 (2008)
834–849.
183
Wear 400–401 (2018) 177–183
[28] A. Ince, G. Glinka, A modification of Morrow and Smith–Watson–Topper mean
stress correction models, Fatigue Fract. Eng. Mater. Struct. 34 (2011) 854–867.
[29] Y.B. Zhang, L.T. Lu, Y.B. Gong, J.W. Zhang, D.F. Zeng, Finite element modeling and
experimental validation of fretting wear scars in press fitted shaft with open zone,
Tribol. T (2017) (DOI: 10.1080/10402004.2017.1378395).
[30] Abaqus User Manual: Abaqus Theory Guide. Version 6.14. USA: Dassault Systemes
Simulia Corp.; 2014.
[31] P.Z. Wang, Y.L. Liang, H.C. Song, Y.L. Yi, Fretting wear behavior of 35CrMo steel
under gross slip condition, Heat Treat. Met. 39 (2014) 141–144.
[32] A. Fatemi, D. Socie, A critical plane approach to multiaxial fatigue damage in-
cluding out of phase loading, Fatigue Fract. Eng. Mater. Struct. 11 (1988) 149–165.
[33] N.E. Dowling, C.A. Calhoun, A. Arcari, Mean stress effects in stress-life fatigue and
the Walker equation, Fatigue Fract. Eng. Mater. Struct. 32 (2008) 163–179.
[34] W.X. Ren, The growth of short crack under fatigue test for 35CrMo alloy steel, J.
Exp. Mech. 8 (1993) 174–179.
[35] Morrow JD. Fatigue properties of metals. In: Fatigue Design Handbook Sec. 3.2. SAE
Advances in Engineering, Warrendale, PA, 1968, pp. 21–29.
[36] S.K. Koh, R.I. Stephens, Mean stress effects on low cycle fatigue for a high strength
steel, Fatigue Fract. Eng. Mater. Struct. 14 (1991) 413–428.
[37] T. Wehner, A. Fatemi, Effects of mean stress fatigue behavior of a hardened carbon
steel, Eng. Int. J. Fatigue 13 (1991) 241–248.
[38] W.S. Sum, E.J. Williams, S.B. Leen, Finite element, critical plane, fatigue life pre-
diction of simple and complex contact configurations, Int. J. Fatigue 27 (2005)
403–416.