The Journal of Engineering

7th International Symposium on Test Automation and Instrumentation

(ISTAI 2018)

Simulation for rubber friction on self-affine eISSN 2051-3305

Received on 15th October 2018
Accepted on 9th January 2019
surface based on energy method E-First on 3rd December 2019
doi: 10.1049/joe.2018.9163
www.ietdl.org

Fuqiang Zhou1 , Shaohong Wang1, Yintao Wei2

1Key Laboratory of Modern Measurement & Control Technology, Beijing Information Science & Technology University, Beijing, People's Republic
of China
2State Key Lab Automotive Safety and Energy, Tsinghua University, Beijing, People's Republic of China

E-mail: fuqiangzhou@126.com

Abstract: Here, the authors propose an energy method to obtain the friction coefficient of a rubber block sliding on self-affine
substrates via simulation. Due to the multi-scale property, the rough surfaces are modelled with four harmonic rigid frictionless
roads. As the rubber block is a viscoelastic solid, energy dissipation in the contact area results in friction. The authors propose a
new simulation method to obtain friction coefficient through energy dissipation. With different loads and sliding velocities, the
change law of friction coefficient is calculated. The authors design a test instrument to verify the friction coefficients of rubber on
some certain surface. The findings of this research constitute a theoretical basis for safe tire design and tire grip improvements.

1 Introduction typical harmonic road surface pattern, and investigating the effect
of conditions on the friction test results. In other words, we obtain
The rubber friction on various rough surfaces has a significant the friction coefficient of the rubber block sliding on the self-affine
impact on tire grip, conveyor belts, and other rubber products. surface with the energy method.
Some models, e.g. the Coulomb friction law, Dahl model, and This paper is structured as follows: Section 2 describes the
Lugre model, provide a relation between the friction force and energy dissipation and introduces a multi-scale method of a rubber
operating parameter. Some models [1, 2] reveal the relationship block on a rough surface. Section 3 simulates the rubber friction
between the microscopic property of the road surface and the test, reports the rheological model of rubber and the self-affine
viscoelastic parameters of rubber and the friction force. In contrast surface approximation with four sine waves, and discusses the test
to metallic friction, which is dependent on the interaction between parameters of the numerical result. Section 4 describes the
contact surfaces, rubber friction is principally affected by the proposed experimental apparatus for measuring friction. Section 5
internal energy dissipation. This conclusion was demonstrated by presents the summary and conclusions.
numerous research studies and is now generally accepted [3]. The
rubber friction coefficient is affected by the properties of the rubber
and friction conditions [4]. Regarding rubber material, the rubber 2 Energy method
friction is the sum of the deformation energy dissipation and the 2.1 Rubber/road surface contact
adhesion of the interface layer. The adhesive force of the interface
layer principally affects dry, clean, and relatively smooth surfaces, At the micro-scale, an object surface is not flat; the contact
e.g. a clean smooth glass surface. The multi-scale analysis of between two planes at the macro-level is actually composed of
rubber friction here ignores adhesion. many small contact areas. The actual contact area is associated
Many roads have the feature of self-affinity [5]. The multi-scale with the external load and various parameters (e.g. the material,
method reported by Wriggers and Reinelt [6] is used to derive the stiffness, and modulus of elasticity) of the contact surface. When
macro-friction coefficient of rubber sliding on a self-affine surface. the contact between two surfaces is observed at coarse scales (e.g.
According to this method, a set of discrete different length scales dimension magnification ξ = 1), the surfaces are smooth planes.
and harmonic functions are superimposed to approximately The two surfaces are in full contact with each other, and the contact
estimate the self-affine surface. The friction is derived from the area is A0. When ξ = 10, peaks and valleys can be observed at the
lower scale surface and used for the next larger scale surface. contact surface. Only the peak is in contact with rubber, and the
Repeated iterations are performed until the maximum scale is load on the surface becomes uneven.
reached for the calculation of the macro-friction coefficient. In view of the microscopic roughness of the road surface, when
In most cases, friction coefficient is obtained by using friction rubber comes in contact with the road surface and slides at a
test rig. For rubber materials, it is difficult to implement. At certain relative speed, the road surface has a road surface
present, there are many phenomenological friction mechanisms, morphology-related load F(ω1, ω2, …ωn) input for rubber, thus
but the friction between the multi-scale pavement and the rubber leading to the energy dissipation of the rubber [7].
block is not much, especially the method of describing the friction A number of energy dissipation patterns exist near the contact
mechanism of rubber through simulation. area between the rubber and substrate asperity during the sliding
Through an analysis of the rubber deformation and energy process of a rubber block on a road surface: (1) the random loading
dissipation, this paper considers contact homogenisation as a of the road substrate asperity leads to rubber deformation; (2) at
function of the sliding velocity and impressed pressure to infer the certain points of the rubber block, significant viscoelastic
macroscopic effective friction coefficient of rubber. We build the deformation induces the stretching and closure of internal micro-
rubber block and the multi-scale road model, perform a sliding test cracks [8]; (3) work is performed by the membrane of impurities
on the numeric self-affine surface at the appointed load and between the rubber and road surface; and (4) the connection and
velocity, and then acquire the energy of each mesh of the rubber. fracture of force between the rubber and road surface molecules
Subsequently, the energy dissipation at each scale is calculated to contribute to the energy dissipation. This paper focuses on the
obtain the effective friction coefficient. We consider the law of energy dissipation and friction issues of rubber during the rolling
friction when choosing the appropriate test parameters, selecting a process of a tire on a road, ignores the cracks in rubber, and does

J. Eng., 2019, Vol. 2019 Iss. 23, pp. 8986-8990 8986

This is an open access article published by the IET under the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0/)
not account for the membrane of impurities; the surface of a typical Regarding the morphological properties of the road surface, a
road is unlikely to be sufficiently smooth and clean for adhesion to statistical contact model, numerical contact model, and multi-scale
occur. contact model can be used to describe the contact with the rough
surface [9]. The road surface can be classified into the following
2.2 Rubber energy dissipation patterns: identical asperities, random non-overlapping substrate,
and self-affine surface [10]. Identical asperities (R(x) = asin(bx))
The elastomer slides on an ideal frictionless road surface, i.e. load are independent of the location, i.e. the time history; in the case of
is only applied from the road surface, and all external force work is movement on a random non-overlapping substrate, the road surface
converted into potential energy; upon the removal of the load, the changes with position and is a function of time that can be
elastomer restores its status observed before deformation, and all of expressed based on the road surface power spectral density or
the energy is released without energy loss. For a viscoelastic elevation difference function Cz(λ) =< (z(x + λ) − z(x))2 >.
material, such as rubber, the stress and strain are in different phases The self-affine surface is characterised by the fact that the micro
along with the alternative variation of load applied, in which case topography of the rough surface is similar at all scales; when the
the material becomes associated with the frequency through the contour of the rough surface is partially enlarged, the contour
hysteresis effect, thereby resulting in energy dissipation. structure is extremely similar at the amplified scale. This feature of
The following conditions are established without regard to the the rough surface can be characterised by self-affine geometric
heat exchange between the rubber and the environment: theory. With respect to the multi-scale road surface expressed by
N
(i) No heat work is absorbed/released by the rubber from/to the the use of several sine functions, R(x) = ∑i = 1 aisin(ωi x)
environment, whereas the work performed by the external force (ω1 > ω2 > ⋯ > ωN ), the establishment of the above-noted road
(surface force and body force) is equal to the work performed by surface and the rubber block of the corresponding size will
the stress against the corresponding strain. consume considerable storage space. In addition, the rubber block
is divided into grids that are sufficiently fine to achieve accurate
∫f
V
(s) (k )
i ui dV + ∫p S
(s) (k )
i ui dS = ∫σV
(s) (k )
ij εij dV (1)
deformation and stress; in the simulation, the computing load is
extremely high and achieving convergence is difficult due to the
(ii) Part of the external force work on the rubber block serves as the non-linearity of the material and contact. Here, the multi-scale
energy for its movement, with part being potential energy and part method is used to determine the friction coefficients μi of the road
being consumed by the hysteresis effect. The heat converted from surface aisin(ωi x) at different scales, which are summed up to
the energy consumed has no effect on the local material of the obtain the total friction coefficient.
rubber, the characteristic of the material remains unchanged.
3 Modelling and simulation
Assuming that strain εij occurs when the rubber mass point changes
3.1 Rheological model of rubber
from t0 to ti, the strain energy density is
The viscoelastic constitutive relation of rubber is a
εij
∫
phenomenological mathematical description of the rubber material
W(εij) = σijdεij (2) and its mechanical properties. Normally, rubber exhibits
0
viscoelastic features above room temperature and dissipates energy
during deformation. Due to its large strain and viscoelastic nature
Rubber deformation is the result of multi-scale road surface
[11], rubber deformation and energy dissipation do not have
ξ1, ξ2, …, ξn loads. The road surface scale ξt causes a certain stress
ξ ξ
explicit solutions. Here, the constitutive equation is composed of
and strain σijt, εijt on the rubber block, and the strain energy six Maxwell elements and a spring. As the harmonic input of
density could be affected by various road surfaces conditions: contact between the road surface and rubber during the sliding
process of the rubber block contains no damping, it is possible to
W(εij) = W(εij)ξ1 + W(εij)ξ2 + ⋯W(εij)ξt (3) describe the load input through the superposition of a series of sine
functions.
The friction coefficient is maximised when the rubber block
The total strain energy density is determined by superimposing the
moves at such a speed that the road surface loading harmonic
strain energy density of the road surface at each scale.
frequency is equal to the inherent harmonic ω̄ = 1/τ. However,
The effective friction coefficient is obtained through a
with regard to sophisticated material models, it is more difficult to
discussion of the stress response and energy dissipation of rubber
find the solution for the equivalent friction coefficient based on the
under the effect of the cyclic strain of the road surface based on the
constitutive relation. The computing load that obtaining the friction
constitutive relation and dynamic mechanical property of the
coefficient can be reduced by using the FEM.
viscoelastic material. The energy dissipation rate of the viscoelastic
For the rubber material described here, the Young's modulus is
material can be expressed as
set to 4.17 MPa and the Poisson's ratio is defined as 0.49; the
viscoelastic property is described using the Prony series of the time
dE
dt
= ∫ d xdtσ ε̇
3
ij ij = Fv (4) domain, as shown in Table 1.

3.2 Simulation
where σij represents the stress component, ε̇ij denotes the strain
change rate, the friction force on a typical unit of rubber block is F, When coming into contact with road substrate asperities, the rubber
and the mean movement velocity is v. If the force applied to the is normally subjected to considerable deformation, whereas the
unit is L, then the effective friction coefficient μ is deformation of the road surface is less significant; accordingly, the
road surface is supposed to be a rigid body, and the rubber block is
F assumed to be a viscoelastic body. Define the road surface as a
μ= (5)
L discrete rigid grid and set the rubber block to be a CPS4R four-
node grid. The contact and sliding simulation test of the rubber
2.3 Multi-scale method block on the rough road surface involves material non-linearity,
geometric non-linearity and contact non-linearity; such non-
The substrate asperity is vital to rubber friction. The micro linearities are likely to result in convergence difficulties during
topography of the road surface affects the load amplitude, and the iterative computations. By breaking down the road surface level by
load exerted on the rubber block results in different amounts of level until the minimum scale is reached, the action of road surface
energy dissipation with different sliding velocities of the rubber on rubber block approaches zero. Define the tangential friction
block. coefficient between the road surface and rubber block as 0.

J. Eng., 2019, Vol. 2019 Iss. 23, pp. 8986-8990 8987

This is an open access article published by the IET under the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0/)
Table 1 Constitutive parameter for the finite viscoelastic material
Serial number g_i_Prony k_i_Prony tau_i_Prony
1 0.000592717 0 0.01134034
2 0.002536278 0 0.0002628
3 0.024459921 0 1.316E-005
4 0.17432432 0 1.295E-006
5 0.387971288 0 7.708E-008
6 0.408678482 0 4.2E-010

Fig. 1  Friction coefficients obtained through simulation with different height-to-width ratios

Fig. 2  Variation of the node strain energy density of rubber block

Considering that the representative contact element is a road

Fig. 3  Friction coefficients at different sliding velocity
surface harmonic length, which is infinitesimally small considering
the overall rubber sample, the binding of the left and right
boundaries of the rubber block exhibits a synchronous deformation model of the material) is established based on multi-scale finite
trend. element analysis to simulate the friction caused by sliding rubber
According to Persson [1], the roughness below the micrometre on the self-affine road surface. The test procedure is described in
scale can be excluded in multi-scale finite element simulation detail in the following: apply perpendicular pressure L on the upper
analysis because of dust and dirt particles that fill the smaller surface of the rubber block; apply a certain movement velocity v to
cavities of a road surface. On this basis, a sine harmonic road the upper surface of the rubber block; determine the strain energy
surface 0.015sin(35x) is established at the level-1 scale, density at each grid node of the rubber block; derive the strain
0.006sin(80x) is established at the level-2 scale, 0.003sin(200x) is energy density and further determine the strain energy change rate
and equivalent tangential force F; use (5) to obtain the friction
established at the level-3 scale, and the 0.002sin(300x) sine
coefficient μ; Fig. 4 shows the resulting sliding friction of the
harmonic road surface is established at the level-4 scale. A
rubber block. At low perpendicular pressures, the rubber block
harmonic wave length λ is taken for the rubber block, and the
cannot fill the entire profile, whereas at high perpendicular
height of the rubber block must be properly selected.
pressures, the rubber block can deform to completely follow the
Rubber blocks with different height-to-width ratios are slid on
corrugated substrate profile. At low sliding velocities, the rubber is
harmonic road surface under the same load, and the obtained
able to fill the profile. At high sliding velocities, the rubber is not
results are shown in Fig. 1. As shown in the figure, the friction
able to fill the entire profile because of the increase in the stiffness
coefficient increases with the height-to-width ratio of the rubber
of the rubber at high perturbing frequencies. As the steps increase,
block before the sliding velocity reaches the peak; however, as the
the velocity of the rubber block saturates, as do the deformation
sliding velocity continues to increase, the friction coefficient of
and friction coefficient, as shown in Fig. 5.
rubber blocks with different height-to-width ratios tends to be
stable, and the difference is not significant. As shown in Fig. 2, as
the node of the grids becomes increasingly further away from road 3.3 Simulation results
surface, its strain energy density decreases and tends to maintain The results of simulation tests at various scales are shown in Fig. 6.
the steady-state value. As shown in Fig. 3, the friction coefficients As shown in the figure, the friction coefficient is not constant but
at different sliding velocity tends to be stable when the height-to- instead changes with the variation of such parameters as velocity
width ratio is >0.75. Hence, the rubber block with a height of 0.75λ and pressure. There is a certain movement velocity that enables the
is taken as the typical test unit. road surface load input frequency ω to lead to the maximum
The simulation test is performed using the six sliding speeds, as friction-induced energy loss, whereas the equivalent friction
shown in Table 2. coefficient obtained through calculation is the largest. The
A rubber friction analysis and forecasting method involving calculated total friction coefficient of the macroscopic asperity is
operating parameters (sliding velocity, perpendicular pressure) and shown in Fig. 7.
geometric physical parameters (road substrate asperity, rheological

8988 J. Eng., 2019, Vol. 2019 Iss. 23, pp. 8986-8990

This is an open access article published by the IET under the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0/)
Table 2 Sliding velocities (mm/s)
Serial number 1 2 3 4 5 6 7
velocity 0.1 1 10 100 500 1000 10,000

Fig. 4  Simulation result of rubber block at different load and velocity

Fig. 5  Friction coefficient saturates with steps Fig. 6  Friction coefficient of rubber block at each scale of rough surfaces
(a) Friction coefficient on micro-scale, (b) Friction coefficient on scale 3, (c) Friction
4 Result validation and test data analysis coefficient on scale 2, (d) Friction coefficient on scale 1
We design a friction instrument as shown in Fig. 8 and perform
tests on the sliding process of the rubber block under various is fixed at the front of the sliding sledge, which is bound to a steel
conditions to determine the friction coefficient of rubber at rope that moves the sliding sledge slide under the dragging force of
different values of road substrate asperity, sliding velocity, and the steel rope guided by the leading sheave. A force sensor and
perpendicular pressures. The rubber block is pasted onto the displacement sensor are used to record the vertical load, traction
bottom of the flange force sensor and maintains connection with load and sliding velocity. By varying the weight attached to the
the road surface in the sliding sledge. The depth of the sliding rope, we can obtain the trail load and sliding velocity. A
sledge is 10 mm. Different substrate asperities can be placed in the displacement sensor is set at the rear of the test rig to measure the
sliding sledge, such as a steel plate, sand paper, a concrete road position of the sledge. As a result, we can calculate the sliding
sample, a marble slab, or a glass plate; in addition, water, oil, and velocity. Before starting the test, we lubricate the sliding rail to
ice can also be put in it. The sliding sledge and sliding rail are lower the unwanted resistance. The relationship between the
linked by six ball bearings, which can lower the resistance between calculated relative sliding velocity between the rubber block and
them. A hand wheel is used to apply a vertical load. A force sensor road surface and the friction coefficient and the relationship

J. Eng., 2019, Vol. 2019 Iss. 23, pp. 8986-8990 8989

This is an open access article published by the IET under the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0/)
Fig. 7  Friction coefficient of the rubber block with P = 5 MPa on the rough surface

Fig. 8  Test rig for rubber specimen

Fig. 9  Relationship between the sliding velocity and friction coefficient of
between the perpendicular pressure and friction coefficient are rubber friction test
shown in Fig. 9; similar to the modelling results presented in the
previous section, the friction coefficient increases gradually with 7 References
increases in the relative sliding velocity between the rubber block
and road surface, and the speed begins to decrease after reaching a [1] Persson, B.N.J.: ‘Theory of rubber friction and contact mechanics’, J. Chem.
Phys., 2001, 115, (8), pp. 3840–3861
certain peak. [2] Klüppel, M., Heinrich, G.: ‘Rubber friction on self-affine road tracks’, Rub.
Chem. Technol., 2000, 73, (4), pp. 578–606
[3] Persson, B.N.J., Albohr, O., Tartaglino, U.: ‘A I volokitin and E tosatti, On
5 Conclusion the nature of surface roughness with application to contact mechanics,
Based on the energy method, this paper considered the friction sealing, rubber friction and adhesion’, J. Phy. Condens. Matter Inst. Phy. J.,
2005, 17, (1), pp. R1–R62
model for self-affine road substrate asperities and rubber and [4] Lorenz, B., Persson, B.N.J., Dieluweit, S.: ‘Rubber friction: comparison of
presented a rubber friction analysis and forecasting method that theory with experiment’, Eur. Phys. J. E., 2011, 34, (12), pp. 1–11
involves various operating parameters and condition parameters. A [5] Heinrich, G., Klüppel, M., Vilgis, T.A.: ‘Evaluation of self-affine surfaces and
multi-scale harmonic road substrate asperity and rubber finite their implication for frictional dynamics as illustrated with a rouse material’,
Comput. Theor. Polym. Sci., 2000, 10, (1–2), pp. 53–61
element model was built to perform the friction test to identify the [6] Wriggers, P., Reinelt, J.: ‘Multi-scale approach for frictional contact of
energy method-based law of rubber friction on a self-affine surface. elastomers on rough rigid surfaces’, Comput. Methods Appl. Mech. Eng.,
The multi-scale method and its reliability were demonstrated 2009, 198, (21), pp. 1996–2008
through friction tests of a rubber block on a road surface. The [7] Lorenzis, L.D., Wriggers, P.: ‘Computational homogenization of rubber
friction on rough rigid surfaces’, Comput. Mater. Sci., 2013, 77, pp. 264–280
research findings presented here constitute a theoretical basis for [8] Yamada, S.: ‘General shear-thinning dynamics of confined fluids’, Tribol.
safe tire design and tire grip improvements. Lett., 2002, 13, (3), pp. 167–171
[9] Hu, Z.W., Liu, K., Wang, W., et al.: ‘Rewiew and prospect of contact models
of rough surfaces’, Cryo. Supercond., 2011, 39, (12), pp. 71–74
6 Acknowledgments [10] Persson, B.N.J., Tosatti, E.: ‘Qualitative theory of rubber friction and wear’,
J. Chem. Phys., 2000, 112, (4), pp. 2021–2029
This research was supported by the National Natural Science [11] Kaliske, M., Nasdala, L., Rothert, H.: ‘On damage modelling for elastic and
Foundation of China (grant nos. 11672148, 51275265, and viscoelastic materials at large strain’, Comput. Struct., 2001, 79, pp. 2133–
51175286). The authors would like to thank Professor Wei who 2141
provided expertise that greatly assisted the research.

8990 J. Eng., 2019, Vol. 2019 Iss. 23, pp. 8986-8990

This is an open access article published by the IET under the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/3.0/)