Wear 314 (2014) 305–313

Contents lists available at ScienceDirect

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

A combined simulation procedure for wear assessment

of the HXN5 locomotive
Jianwu Zhang n, Bin Xu, Xiqiang Guan
State Key Laboratory of Mechanical System and Vibration, School of Mechanical Engineering, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang
District, Shanghai 200240, China

art ic l e i nf o a b s t r a c t

Article history: Wear phenomena like flange wear and hollow-worn tread of HXN5 locomotive wheels are studied by a
Received 18 November 2013 nonlinear iterative procedure. For better understanding of these specific wheel wear mechanism, a
Accepted 24 November 2013 numerical solution is developed by a combination of a SIMPACK multibody model and Archard's wear
Available online 4 December 2013
theory. The wheel wears can be predicted by the nonlinear wear formulation, in which real time
Keywords: parameters of state conditions and contact forces are updated by the multibody simulation in a parallel
Co-simulation manner. Five numerical examples of the locomotive simulation are presented and numerical results are
Locomotive in good agreements with the real observations. The present simulation procedure is validated to be
Hollow-worn tread effective in dynamic steady wheel wear and may be used to assess the optimization of locomotive bogie
Flange wear
systems.
Wear prediction
& 2013 Elsevier B.V. All rights reserved.

1. Introduction assumed to be proportional to the normal force and the sliding

distance divided by material hardness [5–8]. The second method is
The wear phenomena due to rolling contact and impact basically attributed to Archard's wear theory [9].
between wheel and rail have attracted much interest and attention Comparative studies of four simulation approaches for wheel
of engineers and researchers. The HXN5 locomotive with a diesel wears, respectively, by Jendel, Pearce and Sherratt, Zobory, and
engine is used to haul heavy duty freight trains with speed Enblom, are undertaken by Arizon [10], introducing the wear
limitation of 120 km/h in Northeast China. This type of locomo- depth per unit time in differential form to wear assessment of
tives is equipped with two pedestal bogies, in each of which three the four different models for comparability. In both severe and
wheelsets with 3
25,000 kg axle load capacity are suspended mild wear conditions, predictions by Zobory's model are closer to
and driven. Since wheel-rail contact conditions are affected by results from Jendel's and Enblom's models in comparison with
dynamic parameters of the locomotive body, bogie suspensions, as these by Pearce and Sherratt's model. Meanwhile, wear in differ-
well as clearances between wheelsets and bogie frame, specific ential form, the wear velocity, is proven to be effective in the wear
wheel wears in the HXN5 locomotive take place such as severe prediction. In the present research, wear depth per unit time is
wheel flange wear on wheels of the first and the fourth wheelsets, derived by application of the Archard equation to certain exten-
and hollow-worn tread on wheels of the third and the sixth sion in the formulation of wheel wears.
wheelsets. In Fig. 1 are shown the hollow-worn tread of one wheel A simplified and effective co-simulation procedure by a combi-
on the 3rd wheelset. These worn wheels must be taken out of nation of the SIMPACK multibody model and the nonlinear wear
service for repair and replacement. model is proposed. The wear depth is determined by the material
Since mid-eighties of the last century, the wheel-rail wear loss of the wheel in circumferential coordinate and mapped in the
phenomena have been studied by both real tests and theoretical profile coordinate. For comparison, four numerical examples are
predictions. The theoretical solutions of wheel wears may be made for examination of wheel wears of HXN5 locomotive on
classified into two types of methods [1]: (1) the solution with curved tracks with small and large radii by the combined simula-
single parameter, in which the removed material is supposed to be tion procedure. In addition, a track with separate portions of
proportional to dissipated energy or friction work [2–4]; (2) the curved and straight tracks is defined to exam the wheel wear
solution with two parameters, in which the material loss is distributions after 104 km running distances.
This combined numerical procedure is applied to assess the
wheel wears of heavy duty locomotives. The wheel wear informa-
n
Corresponding author. Tel.: þ 86 21 34206541; fax: þ 86 21 34206069. tion can be provided by this procedure for locomotives and bogie
E-mail address: jwuzhang@sjtu.edu.cn (J. Zhang). systems with original and optimized parameters. This numerical

0043-1648/$ - see front matter & 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.wear.2013.11.042
306 J. Zhang et al. / Wear 314 (2014) 305–313

Numerical predictions for the wheel-rail contact responses

obtained by the dynamic simulation model can be further applied
to act as input data in a time step manner for the wear calculations.

2.2. Wear model

According to Archard's wear model [9], Eq. (1), the wear volume
due to wheel-rail rolling contact can be expressed in terms of
normal contact force, sliding distance and the softer material as
follows.
ND
Vw ¼ K ð1Þ
H
Fig. 1. Hollow-worn tread of one wheel on the 3rd wheelset. where Vw is the wear volume, D the sliding distance, N the normal
force, H the Vickers hardness of the soft material and K the
dimensionless wear coefficient.
solution is approved to be effective in prediction of wheel wear
The wear coefficient in Archard's model in association with
forms of heavy duty locomotives and bogies. In the near future,
parametrical nonlinearities is examined by Olofsson and Telliskivi,
optimizations of geometric and dynamic parameters and suspen-
performing real wheel-rail contact experiments and material tests
sion for the pedestal bogie and HXN5 locomotive will be suggested
[12]. The wear coefficient classification is investigated by Jendel
for the purpose of reducing the specific wears mentioned previously.
and Berg, according to the contact pressure and the sliding
velocity [6], as shown in Fig. 2. The most severe wear occurs in
2. Modeling and simulation the top region when the contact pressure exceeds over the value
equivalent to 0.8 times material hardness. Wear coefficient for the
The multibody system dynamics model of the entire locomo- tread contact usually drops in the bottom left region and that for
tive is established in SIMPACK environment, in which Kalker's the flange contact may be allocated in all the four regions. In the
algorithm FASTSIM is integrated and adopted to predict the wheel- present research, median values of wear coefficients in the four
rail contact forces. By simulation of the locomotive dynamic regions, for example k1 ¼350, are taken for computations.
performance, dynamics responses and parameters, creep veloci- Derivation of Eq. (1) with respect to time [10] leads to the
ties, contact area and normal force can be produced and input to following equations as
the Archard wear model built in Simulink. Numerical results of the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N_ N
wheel wears can be obtained directly during each simulation by V_ w ¼ K D ¼K V 2ξ þ V 2η ð2Þ
H H
using the combined simulation procedure. For convenient treat-
where Vξ and Vη are components of the rigid body creep velocity
ment of the two domain numerical solutions, the SIMPACK
to be determined by the dynamic simulations.
Co-simulation Interface is employed in combination of the wear
The conversion from the removed volume ΔVw to the removed
model in MATLAB/Simulink and the multibody model in SIMPACK.
wheel profile depth ΔZ in time step is as follow

2.1. Multibody model ΔV w

ΔZ ¼ ð3Þ
4π Rb
The computational multibody system dynamics model for the Application of Eq. (2) result, the wear depth ΔZ can be calculated
HXN5 locomotive is developed in the SIMPACK environment, in time increment according to
which is composed of 33 rigid bodies including locomotive body, qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bogie frames, wheelsets, axle boxes, electric motors and motor N V 2ξ þ V 2η
hangers. Springs, dampers, baffles and friction units have been ΔZ ¼ K Δt ð4Þ
4π HRb
modeled by linear or nonlinear force elements to connect the rigid
bodies. where R is rolling radius at the contact point and b is the contact
In the simulation, real railway conditions are attributed to three ellipse lateral semi-axis.
kinds of geometrical constraints, curved track with small radius,
curved track with large radius and straight track. The AAR 5th class
spectra of track irregularities containing random vertical, lateral,
rolling and gauge deviations for railway vehicles are treated as
inputs of excitation for the multibody system dynamics model.
Based on the simplified theory of Kalker [11], the fast and
efficient algorithm FASTSIM is used to approximate the contact
between wheel and rail, by which the normal forces and tangential
creep forces can be obtained from solutions of Hertzian normal
contact problem. The FASTSIM algorithm has been proved to be
one of the most powerful and useful computation tools in rail
vehicle dynamics. In Kalker's theory an assumption is made for
contact stresses calculation that the elastic foundation is replaced
by a number of mutually independent springs. The shear stresses
are assumed to keep the stress gradient constant in the contact
patch until the traction boundary is reached. After that, the shear
stresses are assumed the value of traction boundary that is
the normal pressure multiplied by the coefficient of friction. Fig. 2. Wear coefficients in Archard's model.
 J. Zhang et al. / Wear 314 (2014) 305–313 307

The differential equation in Eq. (4) for the wear depth may be component of the gravity while the locomotive is negotiating
solved by Matlab/Simulink, in which variables, N, R, b, Vξ and Vη, curved tracks.
are obtained at each time step from the dynamic simulations. For Case 3 a series of curved tracks with variant superelevation
are idealized and given in Table 3 for details. The running speed of
the locomotive is assumed to be constant at 120 km/h. For Case
2.3. Combined procedure

Using simulation results of dynamic responses for the wheel

Table 1
contacts in SIMPACK, the modified Archard's wear model in Small radii of curved tracks for Case 1.
Matlab/Simulink is employed to predict the wear depths simulta-
neously. The SIMPACK Co-simulation Interface, a Simulink block, is Radius [m] Superelevation Length of circular Velocity
employed to communicate data between SIMPACK and Simulink. [m] arc/total track [m] [km/h]
Accordingly all parameters required in Eq. (4) are updated by the
300 0.142 500/1000 60
dynamic simulation in SIMPACK and exchanged by the SIMPACK 400 0.106 500/1000 60
Co-simulation Interface block, as shown in Fig. 3. After the wear 600 0.071 500/1000 60
calculation, the wear depths of all wheels are mapped in the wheel 800 0.053 500/1000 60
profile coordinate.
Using difference track settings in the multibody model, a series
of wear distributions of all wheels can be obtained by the combined
procedure. By classification and conclusion of the numerical wear Table 2
Large radii of curved tracks for Case 2.
results, wear features of all wheels can be deduced.
Radius [m] Superelevation Length of circular Velocity
[m] arc/total track [m] [km/h]
2.4. Examples and assumptions
1000 0.170 500/1000 120
Numerical examples are made for the specific co-simulations of 2000 0.085 500/1000 120
wheel wears for the HXN5 locomotive. Five cases of track condi- 3000 0.057 500/1000 120
4000 0.0425 500/1000 120
tions are considered. For Case 1 a series of small radii of curved
tracks are idealized and given in Table 1 for details. The running
speed of the locomotive is assumed to be constant. For Case 2 a
series of large radii of curved tracks are idealized and listed in Table 3
Table 2 for details. The running speed of the locomotive is also Curved tracks for Case 3.
assumed to remain constant. According to Eq. (4) the wheel wear
Radius [m] Superelevation Length of circular Velocity
depth is proportional to the normal contact force. The wheel-rail [m] arc/total track [m] [km/h]
normal forces are directly affected by the centrifugal force of
locomotive, which changes along with the variation of super- 1000 0.090 500/1000 120
elevation of the curved track. Therefore, determination of super- 1000 0.110 500/1000 120
1000 0.130 500/1000 120
elevations in Case 1 and Case 2 is based on that the centrifugal
1000 0.150 500/1000 120
force of the locomotive has been balanced by the horizontal

Fig. 3. Architecture of the combined numerical procedure for wheel wear prediction.
308 J. Zhang et al. / Wear 314 (2014) 305–313

4 one track is idealized with 800 m radius. The running speed of 3. Simulation analysis and discussion
the locomotive is from 60 km/h to 120 km/h stepped by 20 km/h.
The typical tracks for the combined simulation in Cases 1 to It can be easily verified by simulation that the wears of left and
4 are treated to be identical. Total length of all tracks in these four right wheels on a single wheelset are identical after locomotive
cases is 1 km, as shown in Fig. 4 and in Tables 1–3. Fig. 4 gives negotiates a sinusoidal track. For simplification, definition is made
variation of the curvature and superelevation in the longitudinal on the track with one right hand curve and only wear results of
direction of the curved track with 300 m radius. The first 150 m, left wheels, as well as the outer wheels, are presented. In Fig. 5 are
the following 100 m and the middle 500 m are straight track, shown simulation results for the wear depths of left wheels of six
transition track and circular arc respectively. After the middle wheelsets obtained for Case 1 and in Fig. 6 are given simulation
circular arc, the following 100 m is also the transition track. The results for the wear depths of left wheels of six wheelsets obtained
final 150 m becomes the straight track. All other tracks have the for Case 2.
same shape but different values of curvature and superelevation. As shown in both Figs. 5 and 6 the maximum wear depths of all
For comparability, irregularities of the track used herein possess wheels are reduced by increasing the curved track radius for the
the same statistic properties as those of the AAR Class 5 spectra. locomotive at the same speed. The locations of maximum wear
For Case 5 a track with separate portions of curved and straight depths on the 1st and 4th wheelsets occur in the flange area for
tracks are used, which possesses of statistical properties of a the curved tracks of radii less than 3000 m. The value of wear
typical railway line in China [13], as given in Table 4. The track is depths of left wheels on the 1st and 4th wheelsets are at least one
divided into eleven intervals, in each of which a curved track with order larger in quantity than those of the other wheelsets. This
the mean radius, Rm and the mean superelevation, hm are used to means that wheel wears of the leading wheelsets of the front and
represent the geometry of all curved tracks. The definition of the rear bogies become severe than other wheels. In this condition,
track is mainly dependent on curved tracks with smaller radii severe flange wear can occur on wheels of leading wheelsets of
which affect more to wheel wear distribution in comparison with both bogies.
larger ones. Curved tracks with radii larger than 3500 m are The wheel flange wear in the 2nd and 5th wheelsets are found
simplified to be straight. If so, the wear distribution can be in Case 1, as shown in Fig. 5(b and e), and also normal wears are
determined for each curved tracks and weighted by the total observed in Case 2, as shown in Fig. 6(b and e). Locations of the
length of intervals, which ensures that each interval contributes to maximum wear depths are moved to the tread zone from the near
the total wear distribution proportionally. flange area by increasing the curved tracks radii.
Both in Figs. 5(c) and 6(c) are plotted, respectively, the left
wheel wears on the 3rd wheelset of the front bogie. Positions of
the wears remain almost in the tread zone. As a result this rolling
contact condition may cause hollow-worn tread. The similar
simulation results produced for the wheels of the 6th wheelset
are presented in Figs. 5(f) and 6(f).
In Fig. 7 are shown simulation results for the wear depths of
left wheels on six wheelsets obtained for Case 3. The normal
contact force between the left wheel and rail is decreased by the
increasing superelevation, because the horizontal component of
gravity of locomotive balances more centrifugal force. Therefore,
the maximum wear depths of left wheels of all wheelsets are
reduced by increasing the superelevation of the curved track
according to Eq. (4).
As shown in Fig. 7(a and d), the locations of the maximum wear
depths of left wheels on the 1st and the 4th wheelsets remain in
the near flange area with the increasing superelevation. Compar-
ing with results of other four wheels, the maximum wear depths
of left wheels on the 1st and 4th wheelsets are at least one order
larger in quantity than those of the other wheels, like results in
Case 1 and Case 2.
Fig. 4. Definition of the curved track with 300 m radius in Case 1. (a) Curvature As the superelevation is larger than 0.11 m, the wear zone
against track length and (b) superelevation against track length. of the left wheels on the 2nd and 5th wheelsets moves to the
tread area instead of remaining in near flange area, as shown in
Table 4 Fig. 7(b and e). The results indicate that the location of maximum
Discretization of track profile. wear depths varies on the wheel profile and wear distributions of
wheels on the middle wheelset of each bogie are better than those
Radius interval Rm hm Total track length Total track sum of other wheels.
[m] [m] [m] [%] [%]
For the left wheels of the 3rd and 6th wheelsets, the situation is
0–350 300 0.142 0.10 0.1 similar to the tread wears of the same wheels in Case 1 and Case 2.
350–450 400 0.106 0.42 0.52 The wear zone is independent of variation of the superelevation
450–550 500 0.085 0.6 1.12 and remains in the tread area. The similar numerical results
550–650 600 0.071 1.11 2.23
produced for the left wheels of the 3rd and 6th wheelsets in the
650–750 700 0.061 0.51 2.74
750–900 800 0.053 1.17 3.91 Case 4 can also be observed in the following Fig. 8.
900–1200 1000 0.170 4.04 7.95 In Fig. 8 are given wear results for left wheels of six wheelsets
1200–1700 1500 0.113 2.25 10.20 obtained for Case 4. The variation of locomotive speeds influences
1700–2500 2000 0.085 3.21 13.41 the wheel wears of six wheelsets differently. Wear depths of left
2500–3500 3000 0.057 4.84 18.25
Straight 81.75 100
wheels on the 1st and 4th wheelsets are proportion to the speed of
locomotive. On the contrary, wear depths of left wheels on the 3rd
 J. Zhang et al. / Wear 314 (2014) 305–313 309

Fig. 5. Numerical predictions for wear depths of left wheels on curved tracks in Case 1. (a) The 1st wheelset, (b) the 2nd wheelset, (c) the 3rd wheelset, (d) the 4th wheelset,
(e) the 5th wheelset and (f) the 6th wheelset.

and 6th wheelsets are inverse proportion to the speed of locomotive wheelsets can also be concluded by Case 4. However, the wear
and the maximun value remains in a small interval, in the sense that zone of the 2nd and the 5th wheelsets remains in the near flange
the maximum wear values of these wheels are not very sensitive to area in Case 4 rather than move to the tread zone with changing
the speed in Case 4. As the speed is from 60 km/h to 100 km/h, the simulation conditions in the other cases.
maximun wear depths of left wheels on the 2nd and 5th wheelsets In Fig. 9 are shown the numerical results of wear distributions
reduce and then increase after the speed is larger than 100 km/h. after 104 km running distance on the track defined in Case 5. As
In Case 4 the wear zone of left wheels on the 1st, 3rd, 4th and can be observed, the wears of the 2nd and the 5th wheels are
6th wheelsets, as shown in Fig. 8(a, c, d and f), are similar as Cases normal in comparison with other four wheels. Severe flange wears
1, 2 and 3. The flange wear of left wheels on the 1st and 4th can be observed on the 1st and 4th wheels, especially the 1st
wheelsets and the hollow-worn tread of the 3rd and the 6th wheel, and the maximum wear depths are at least one times larger
310 J. Zhang et al. / Wear 314 (2014) 305–313

Fig. 6. Numerical predictions for wear depths of left wheels on curved tracks in Case 2. (a) The 1st wheelset, (b) the 2nd wheelset, (c) the 3rd wheelset, (d) the 4th wheelset,
(e) the 5th wheelset and (f) the 6th wheelset.

than those of the 2nd and the 5th wheels. The wear zones of the 3rd wheelsets occurs for the curved tracks of radii less than 3000 m
and 6th wheels located in the tread area and the maximum wear and the hollow-worn tread of the 3rd and the 6th wheelsets
depths are about one times larger than those of the 2nd and the 5th appears in all five cases. In the condition that wear depths of left
wheels, which may cause the hollow-worn tread. For further wheels on the 1st and 4th wheelsets are at least one order larger
predicting wheel wear distributions accurately after a long running in quantity than those of the other wheelsets, the flange wear
distance, the wheel profile evolution should be involved in the could be severer than usual. The specific wears, severe flange wear
simulation procedure, which will considered in the future work. on wheels of the 1st and the 4th wheelsets and hollow-worn tread
It is worthy of being noted that the present predictions in five on wheels of the 3rd and the 6th wheelsets, are identical to real
cases are in good agreement with real observations for wheel observations mentioned in the first section. Because the HXN5
wears phenomenon of the HXN5 locomotive. Concluding these locomotive has only one driver cab in the front of the locomotive,
five numerical predictions, the flange wear of the 1st and the 4th wheels will retain the wear features between two maintenances.
 J. Zhang et al. / Wear 314 (2014) 305–313 311

Fig. 7. Numerical predictions for wear depths of left wheels on curved tracks in Case 3. (a) The 1st wheelset, (b) the 2nd wheelset, (c) the 3rd wheelset, (d) the 4th wheelset,
(e) the 5th wheelset and (f) the 6th wheelset.

With assumptions for tracks, causes of these wears are the wheel flange wear on wheels of the 1st and the 4th wheelsets and
dynamic parameters of the locomotive body, bogie suspensions, hollow-worn tread on wheels of the 3rd and the 6th wheelsets are
as well as clearances between wheelsets and bogie frame rather predicted by the effective nonlinear combined simulation procedure.
than conditions of curved tracks. The numerical solution is developed by the multibody model
and the Kalker simplified theory in SIMPACK, and the modi-
fied Archard's wear model in MATLAB/Simulink. The SIMPACK
4. Concluding remarks Co-simulation Interface was employed to integrate both software
and communicate with each other. Five typical numerical exam-
The HXN5 locomotive wheel wears are formulated by a non- ples, locomotive on curved tracks with small radii at the speed of
linear computational procedure based on the multibody system 60 km/h, curved tracks with large radii at the speed of 120 km/h,
dynamics and Archard's wear theory. Numerical results of severe curved track of radius 1000 m with variant superelevations, curved
312 J. Zhang et al. / Wear 314 (2014) 305–313

Fig. 8. Numerical predictions for wear depths of left wheels on curved tracks in Case 4. (a) The 1st wheelset, (b) the 2nd wheelset, (c) the 3rd wheelset, (d) the 4th wheelset,
(e) The 5th wheelset and (f) the 6th wheelset.

track of radius 800 m at different speeds and a track with separate wears, severe flange wear and hollow-worn tread, are mainly
portions of curved and straight tracks, are presented and discussed. affected by dimensions and parameters of bogie and locomotive
The flange wears of wheels on the leading wheelsets of the front rather than conditions of tracks.
and rear bogies are found than those of other wheels for the HXN5 By the proposed procedure for prediction of wheel wear forms,
locomotive on the curved tracks with radii less than 3000 m. Wear synthetic optimization of the locomotive and bogie system may be
depths of wheels on the 1st and 4th wheelsets are at least one order carried out in the earlier stage of design to limit the abnormal
larger in quantity than those of the other wheelsets, which makes wears. This combined simulation procedure should be validated by
the flange wear severer and may cause severe flange wear. The measurements of worn wheel profiles in the future. In addition,
tread wears on the 3rd and the 6th wheelsets occur in all numerical the tractive effort and braking of locomotive in curved and straight
examples and may cause hollow-worn tread. By assumptions of tracks can also affect the wheel wear significantly, which will be
track system and the numerical results, it is also shown that the considered and studied in the future works.
 J. Zhang et al. / Wear 314 (2014) 305–313 313

References

[1] R. Enblom, Deterioration mechanisms in the wheel-rail interface with focus on

wear prediction: a literature review, Veh. Syst. Dyn. 47 (2009) 661–700.
[2] H. Krause, G. Poll, Wear of wheel-rail surfaces, Wear 113 (1986) 103–122.
[3] I.J. McEwen, R.F. Harvey, Full-scale wheel-on-rail wear testing: comparisons
with service wear and a developing theoretical predictive method, ASLE/ASME
Lubrication Conference, Hartford, U.S.A, 1983.
[4] F. Braghin, S. Bruni, F. Resta, Wear of railway wheel profiles: a comparison
between experimental results and a mathematical model, Veh. Syst. Dyn. 37
(2003) 478–489.
[5] R. Enblom, M. Berg, Impact of non-elliptic contact modelling in wheel wear
simulation, Wear 265 (2008) 1532–1541.
[6] T. Jendel, M. Berg, Prediction of wheel profile wear – Methodology and
verification, Veh. Syst. Dyn. 37 (2002) 502–513.
[7] R. Enblom, Simulation of Wheel and Rail Profile Evolution: Wear Modelling
and ValidationRoyal Institute of Technology (KTH), Stockholm, 2004.
[8] T. Jendel, Prediction of wheel profile wear - comparisons with field measure-
ments, Wear 253 (2002) 89–99.
[9] J.F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys. 24 (1953)
981–988.
[10] J. De Arizon, O. Verlinden, P. Dehombreux, Prediction of wheel wear in urban
railway transport: comparison of existing models, Veh. Syst. Dyn. 45 (2007)
849–866.
[11] B. Jacobson, J.J. Kalker, Rolling Contact Phenomena, Springer, Italy, 2000.
[12] U. Olofsson, T. Telliskivi, Wear, plastic deformation and friction of two rail
steels – A full-scale test and a laboratory study, Wear 254 (2003) 80–93.
[13] Z. Wen, Fatigue load spectra study of C70 type railwail freight car, National
Key Laboratory of Traction Power, Southwest Jiaotong University, Chengdu,
China, 2011.

Fig. 9. Numerical results of wear distribution after 104 km running distance on the
track defined in Case 5. (a) Wear distributions of wheels in the leading bogie and
(b) Wear distributions of wheels in the second bogie.

Acknowledgments

The present work financially supported by Scientific Research

and Development Program of China Ministry of Railways under
Project no. 2010J015 is gratefully appreciated.