Numerical Simulation of Sliding Wear Based on

Archard Model

Xuejin SHEN Lei CAO Ruyan LI

Department of Mechanical Automation Department of Mechanical Automation Shanghai Bearing Technology
Shanghai University Shanghai University Research Institute
Shanghai, China Shanghai, China Shanghai, China
[email protected]

Abstract — A sliding wear simulation approach based on recently, FEM was used to simulate fretting wear as well as the
Archard’s wear model was proposed, in which ABAQUS evolution of fretting variables with the number of wear cycles
scripting interface was used to simulate the progressive in a cylinder on flat configuration both made of Super CMV, a
accumulation of wear between contact surfaces. Removal of hardened steel alloy [8].
material caused by wear was implemented by moving boundary
nodes. In order to measure wear rates of woven fabric liner, a In this paper, a numerical simulation forecasting method
pin-on-disk experiment was performed. The wear rate of was proposed to solve sliding wear problems based on Archard
9.76×10-7mm3/Nm was used as input of the simulation. Then a model and Vishwanath Hegadekatte, Nam Ho Kim, Priit Põdra
wear problem of spherical plain bearing with self-lubricating and others’ research results. Wear simulation of spherical plain
fabric liner was performed to validate above wear prediction bearing with woven fabric liner was taken as an example.
method. Compared with wear experiments of other bearings, Firstly, the wear prediction method was presented. Secondly,
close agreement between experiments and wear simulation was Pin-on-disk wear tests were performed in order to measure
found. At the same time, proper re-meshing technology and time wear rate of woven fabric liner. Thirdly, finite element model
step would increase calculation efficiency and precision. Results and wear simulation routine were presented. Finally,
show that the complex nonlinear wear process can be simulated comparison between finite element results and experimental
with a series of discrete quasi static models and the method results was made to verify this method.
proposed can be widely used to predict wear problems for
engineering.
II. WEAR SIMULATION METHOD
Keywords-Wear Simulation; Archard Model; Finite element
method (FEM); Wear Rate; Spherical Plain bearing (key words) A. Wear Model
The wear process can be treated as a dynamic process,
I. INTRODUCTION depending on many parameters and the prediction of that
Wear is one of the major reliability issues of tribology. It is process as an initial value problem. The most frequently used
desirable to design engineering components for infinite life. model is based on the Archard’s wear law. It is assumed here
Unfortunately, in systems where parts are in intimate contact that wear can be evaluated by applying Archard’s equation to
and relative motion wear is inevitable. Moreover, the diversity local contact conditions along a differential width of the
and complexity of the wear phenomenon make it difficult to contact interface. Archard’s equation for sliding wear is
accurately predict life span of mechanical parts. The most normally expressed as [9]:
confident knowledge about the friction pair tribological V FN
behaviour can be achieved by taking lots of wear experiments. k˜ . ˄1˅
However, experimental exploration is not only costly, but also s H
not satisfactory when it comes to practical problems of uneven Where V is the wear volume, s is the sliding distance, F N is the
load distribution as well as changeable load, such as wear life normal load, H is the hardness of the worn surface and k is the
span of spherical plain bearing with woven fabric liner. dimensionless wear rate. In order to simulate the evolution of
the contact surface profiles with wear cycles, it is necessary to
With the rapid development of computer technology and determine the wear depth at each contact node of the finite
tribology in the past 20 years, “calculation tribology” has element model. Therefore, for an infinitesimally small
become a branch of tribology [1-2] and demonstrated its strong
apparent contact area, 'A, the increment of wear depth, dhw,
vitality. FEM is widely used to simulate wear progress of
associated with an increment of sliding distance, ds, is
mechanical parts. Dr.Vishwanath Hegadekatte [3-6] calculated
determined. This can be obtained by applying (1) locally to the
wear of the micro-mechanical devices based on finite element
area 'A and for the increment of sliding distance, ds:
simulation. The experimental results were in good accordance
with the simulation results. Nam Ho Kim [7] simulated a dV FN
block-on-ring experiment and achieved good results. More k˜ . ˄2˅
ds H

This project is supported by High and New Engineering Program of Shanghai

(No. D.51-0109-09-001) and the Shanghai University Innovation Fund (No.
A.16-0109-09-724).

978-1-4244-7739-5/10/$26.00 ©2010 IEEE

Then, dividing both sides by 'A, the following equation is
obtained:
dV FN
k˜ . ˄3˅
ds ˜ 'A H ˜ 'A
The F N /'A term is the local contact pressure, p, while dV/'A is
the required increment of local wear depth, dhw. The following
equation is thus obtained for the prediction of the increment of
local wear depth:

dh w
kD ˜ p . ˄4˅
ds
Where the quantity k/H is replaced here by k D , the dimensional
Fig. 3 Schematic of Test Pin.
wear rate. Thus, the total wear depth hw of every element
between the contact surfaces could be formulated as
Considering actual operating conditions of spherical plain
bearing (Fig. 2), and geometry measurement convenience after
³k ˜ p ˜ ds .
w
h D ˄5˅
the tests, loads 14N and 33N were imposed, with speeds
(m/min) of 90 and 15, respectively. The results were shown in
B. Wear Modeling Procedure Fig. 4.
The prerequisite for wear simulation is to discretize a
continuous process into several steps. During each step,
assume that the contact parameters, such us contact pressure,
surface morphology, are constant. The whole process of wear
simulation begins with the calculation of contact pressure
between contact surfaces. Because contact problems usually to
be non-linear, the commercial finite element code ABAQUS
6.8 is implemented to calculate the contact parameters. Then,
equation (4) is used to calculate wear increment and update
contact nodes of every element. At last, the total wear depth on
the contact surfaces could be integrated with (5).

III. EXPERIMENTAL METHOD AND RESULTS FOR

DETERMINATION OF A WEAR RATE
In order to measure wear rates, pin-on-disk tests were (a)
performed on the UNIVERSAL MICRO MATERIALS
TESTER (UMT–2), produced by Center for Tribology, Inc.
(CETR), USA (Fig. 1). Materials of the contact pair were
aluminum alloy plate (aluminum alloy 7075) and self-
lubricating liner (FSZ-Z02) adhered to the lower cylinder part
of the test pin (Fig. 3).

(b)
Fig. 4 (a), (b) Relationship between the wear depth and sliding distance of
liner under different loads and speeds.

Since surface of the disc was anodized, its hardness and

anti-friction performance were significantly increased. Also,
there were almost no wear after experiments. Thus, wear of
aluminum alloy disc was ignored when wear volume of the
Fig. 1 UMT-2 universal Fig. 2 Schematic of spherical
micro materials tester. plain bearing.
liner was calculated. Equation (6) was used to calculate the
wear volume of liner.
 ­° a series of finite element analyses and to estimate the profile of
ª L º ½°
2
ªL º L the worn surface.
V W ˜ ®arcsin« c » ˜ R2  c ˜ R2  « c » ¾. ˄6˅
°̄ ¬ 2R ¼ 2 ¬ 2 ¼ °¿
A. Finite element modeling
Where V is volume of the material loss, W is length of the wear
The spherical plain bearing shown in Fig.2 was taken as an
scar (Fig. 5), R is radius of the pin (Fig. 3), L c is width of the
example of wear simulation. The outer ring and inner ring of
wear scar (Fig. 6).
the bearing were made of aluminum alloy and the self-
lubricating liner was woven fabric liner. In order to simulate
the experimental conditions of the bearing system [10], a test
ring and shaft were also implemented (Fig. 7), both of which
were rigid body. Since analytical method was not available to
analyze spherical plain bearing problems, finite element
analysis code ABAQUS was used to solve the non-linear
contact problem. Reference [11] showed that the maximum
contact pressure between inner ring and outer ring occurred on
the axial cross-section, which was the intersection plane
between XOY plane and the bearing system (Fig. 8). Therefore,
the maximum wear depth would occur on the same section in
terms of Archard’s wear model.
Fig. 5 Magnification of wear scar on the lower cylinder part of the pin.

(a) Before wear (b) After wear

Fig. 6 Comparison on the lower cylinder part of pin before and after wear.

Wear rates of self-lubricating liner under different loads

and speeds could be calculated according to Archard’s [9] Fig. 7 Model of bearing test system.
wear model:
V
kD . ˄7˅
FN ˜ s
TablHĉ
ĉshows experimental results of the wear rate. The
average wear rate is 9.76×10-7 mm3/Nm, which would be a
reasonable input to the finite element simulation of spherical
plain bearing.

TABLE I. WEAR RATES OF LINER Fig. 8 Axial cross-section model of spherical plain bearing.

Wear rate (mm3/Nm)/ Wear rate (mm3/Nm)/ Average wear rate B. Wear simulation routine
F=14Nˈv=90m/min F=33Nˈv=15m/min (mm3/Nm) The flow chart of the finite element wear simulation
procedure, consisting of a series of structure solution steps
7.23×10-7 1.23×10-6 9.76×10-7 combined with additional calculations, was shown in Fig. 9. In
this paper, ABAQUS scripting interface was introduced to
carry out the wear simulation program.
IV. FINITE ELEMENT MODELING AND SIMULATION
The initial parameters given defined the model geometry,
The strength of finite element analysis in making wear loads, constraints and wear model parameters along with the
predictions is its ability to accurately consider both the element and material data. Special subroutines were developed
variation of the contact pressure and the progressive change of for every configuration to generate the FE model and define
the surface geometry caused by material removal in complex the loads and constraints automatically. Simulation of wear
three-dimensional components. In this section, the wear rate began with the solution of the general contact problem using
that is obtained from the pin-on-disk testing is used to perform
ABAQUS. Contact pressure at each of the surface nodes was Thus, adopting appropriate time step is necessary to improve
computed from the normal vector and the stress tensor the computational efficiency. As for spherical plain bearing,
averaged at each surface node. Then, Archard’s wear model wear depth of the central region increased quickly at the
was implemented to calculate linear wear. According the wear beginning because of the higher contact pressure. With wear
depth, updated the FE model. If the cumulative wear depth did distance went on, contact pressure on the border region of the
not exceed the maximum predefined wear depth, went to the contact surface increased and the middle region decreased. At
next iteration process until 25000 oscillation cycles. last, oscillating wear of the bearing achieved a steady state,
called linear wear state. In order to simulate the rapid wear
depth change at the beginning, a varying time step was
employed for the wear process. As shown in TableĊ, firstly, a
small time step was used. Then, the time step increased
gradually with the increase of oscillating cycles.

TABLE II. ADAPTIVE TIME STEP

Oscillating Cycles r Time Step
r < 500 100
500 d r < 5000 500
5000 ” r < 10000 1000
r t 10000 3000

D. Wear simulation results

Fig. 10 shows wear depth of the inner face of outer ring
until 25000 oscillating cycles. The maximum wear depth is
0.185mm and occurs at the central contact region. Through
comparison with experimental results (Tableċ) of the same
kinds of spherical plain bearing, close agreement between the
FEA predictions and experimental results is found for
maximum wear depth. Actually, the practical wear rate is not a
constant during the oscillating process. However, during the
simulation, the average wear rate of the inner is employed.
This may account for that the FEA result is slightly larger than
the experimental results.

Fig. 9 Wear simulation flow chart of spherical plain bearing.

C. Re-meshing technology and proper time step

In order to simulate wear of the contact surface, coordinate
position of the nodes must be changed. Whereas, changing
positions of boundary nodes may yield inaccurate sensitivity
results or a distorted finite element mesh, and thus fail in
achieving an optimal solution. Thus, it is necessary to re-mesh
the model after each cycle. Boundary displacement method
(BDM) of the design sensitivity analysis (DSA) is widely Fig. 10 Comparison of the inner surface of outer ring before and after
employed to update the finite element mesh for FEA. Before wear.
using BDM, displacements of each element node must be
calculated. Then, the displacements are used to update finite Since the maximum wear depth occurs on the center node
element model. of the inner face of outer ring, the wear simulation program
The integration time step is another parameter regarding records the relationship between wear depth and oscillating
the reliability of simulation results. Too long steps cause cycles of the node. As shown in Fig. 11, the wear depth and
erratic results and possibly the un-convergence of FEA oscillating cycles show a linear trend overall. For one thing,
procedure. Too short intervals take too much computing time. the wear rate is constant. For another, changes of the contact
pressure during simulation process do little contributions to mechanical components. Through discretizing wear process
wear depth. into wear simulation step, total wear volume could be
calculated using Euler integration scheme. When it comes to
TABLE III. WEAR TESTS RESULTS FOR THE SAME KINDS OF spherical plain bearing, close agreement between experiments
SPHERICAL PLAIN BEARINGS [12, 13] and wear simulation is found. Results show that the complex
Load Oscillating Wear Depth nonlinear wear process can be simulated with a series of
Samples of Bearing
(KN) Cycles (mm) discrete quasi static models, and the method proposed can be
‡ 110-1 510 25000 0. 048 as an available tool to predict wear problems for engineering.
‡ 110-2 510 25100 0.130
2) Pin-on-disk wear test is an efficient way to get wear
‡ 110-3 510 15210 0.213
rates of materials. In this paper, UNIVERSAL MICRO
MATERIALS TESTER (UMT̢2) was implemented to test
As shown in Fig. 12, the finite element model of the outer wear rate 9.76h10-7 mm3/Nm of liner.
ring changes because of volume loss after 25000 oscillating 3) Since element distortion is a key factor that affects
cycles. Every node of the inner surface of outer ring moves accuracy of the results, re-meshing technology must be taken
along their normal direction except for the boundary nodes B
into consideration during the process. In order to improve the
and C, which move vertically, along the positive direction of Y
axis. calculation efficiency, an adaptive method for the time step
between the geometry updates was used. In the early steps, a
smaller step was implemented while a greater one was used in
the subsequent steps.
ACKNOWLEDGMENT
The authors wish to thank Professor Xiaoyang CHEN at
Shanghai University for valuable discussions. The authors also
wish to thank Youguang CHEN and Yi ZHANG at Shanghai
Bearing Technology Research Institute for technical assistance.

REFERENCES
[1] J. F. Archard, Wear theory and mechanisms. In: Peterson MB, Winer
WO, editors. Wear control handbook. New York: ASME, 1980.
[2] R. D. Cook, Concepts and applications of finite element analysis. New
York: John Wiley and Sons, 1981.
Fig. 11 Wear depth changes of the central node on the outer ring with [3] V. Hegadekatte, N. Huber, and O. Kraft, “Finite element based
oscillating cycles. simulation of dry sliding wear,” Modelling and Simulation in Materials
Science and Engineering, Vol. 13, pp. 57-75, 2005.
[4] V. Hegadekatte, N. Huber, and O. Kraft, “Modeling and simulation of
wear in a pin on disc tribometer,” San Antonio, TX, United states, 2006.
[5] V. Hegadekatte, S. Kurzenhauser, N. Huber, and O. Kraft, “A predictive
modeling scheme for wear in tribometers,” Tribology International, Vol.
41, pp. 1020-1031, 2008.
[6] V. Hegadekatte, N. Huber, and O. Kraft, “A finite element based
technique for simulating sliding wear,” Washington, D.C., United states,
2005.
(a) FE model of the outer ring after wear simulation
[7] N. H. Kim, D. Won and D. Burris, B. Holtkamp, G. R. Gessel, P.
Swanson, and W. G. Sawyer, “Finite element analysis and experiments
of metal/metal wear in oscillatory contacts,” Wear, Vol. 258, pp. 1787-
1793, 2005.
[8] I. R. McColl, J. Ding and S. B. Leen, “Finite element simulation and
experimental validation of fretting wear,” Wear, Vol. 256, pp. 1114-
1127, 2004.
[9] J. F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys. 24,
1953, pp.981–988.
[10] L. B. Wei and Y. G. Chen, “Research and application of spherical plain
(b) Worn magnification on the left side of outer ring bearings,” Bearing, Vol. 5, pp. 8-10, 2008 (in chinese).
[11] K. W. Li, X. J. Shen, Y. G. Chen, and R. Y. Li, “Numerical analysis of
Fig. 12 Profile of the outer ring after wear simulation. woven fabric composites lubricated spherical plain bearings,” Orlando,
FL, United states, 2008, pp. 1392-1399.
[12] B. Z. Di, F. Y. Hong and Y. Wang, “Wear tests of spherical plain
V. DISCUSSION AND CONCLUSIONS bearings,” Bearing, Vol. 6, pp. 25-26,40-41, 1999 (in chinese)
[13] R. Y. Li, “The application of polytetrafluoroethylene fabrics to spherical
1) Combination of finite element theory and numerical plain bearings,” Function materials, Vol. 35, pp.2221-2223, 2004 (in
simulation method provides a good way to predict wear life of chinese)