Вы находитесь на странице: 1из 11

Tribology International 44 (2011) 834–844

Contents lists available at ScienceDirect

Tribology International
journal homepage: www.elsevier.com/locate/triboint

Numerical tribology of a dry contact

Mathieu Renouf , Francesco Massi, Nicolas Fillot, Aurélien Saulot
Université de Lyon, CNRS, INSA-Lyon, LaMCoS UMR5259, F-69621 Villeurbanne, France

a r t i c l e i n f o a b s t r a c t

Article history: Tribologists are confronted on a daily basis by the need to understand the causes and consequences of
Received 22 June 2010 friction on the behaviour of bodies in contact. Understanding contact behaviour is not only a scientific
Received in revised form curiosity but the key to solving numerous industrial issues. Numerical tools have been developed to
24 December 2010
overcome the problems encountered in experiments due to limitations in the local dynamic analysis of
Accepted 15 February 2011
Available online 2 March 2011
multi-scale systems (mechanisms, bodies in contact, interfaces). More than an exhibition of numerical
results, the present paper proposes reviewing the literature on the numerical tribology of dry contacts
Keywords: by analysing the different scales involved.
Tribology & 2011 Elsevier Ltd. All rights reserved.
Dry friction

1. Tribological context One of the main reasons for the difficulties encountered so far is
that problems involving friction and wear are multi-scale and multi-
Friction is part of our everyday lives, in common actions such physical by nature, as well as being strongly coupled. From the
as holding a pencil, walking, etc. From a pragmatic point of view, mechanical viewpoint, constraints at contact level depend on the
it is also the key issue of many industrial problems (mechanism contact itself as well as on the whole mechanism. From the material
lifetime [1–3], different types of wear [4,5], brake squeal noise [6], standpoint, the materials forming the bodies in contact (first bodies)
vibration and noise reduction [7], etc.) and economic challenges and the interface (third body [14]) respond to both temporal and
(energetic expense, warming, etc.). These are just a few of the local spatial distributions of stresses, by exhibiting complex beha-
reasons, why, from the beginning of history (e.g. tribology on viour. In addition, friction and wear models are confronted by the
Egyptian frescoes [8]), human beings have tried to minimise or great complexity of mechanical, thermal and physicochemical
maximise friction according to their needs. phenomena occurring within contacts and affecting the different
Since the Renaissance, with the research of Leonardo Da Vinci parameters used in the models. This complexity results from the
(1453–1519), followed by that of Guillaume Amontons (1663–1705) large number of strongly correlated parameters.
and Charles Augustin de Coulomb (1736–1806), researchers have This complexity, involving different scales and physics, has led to
attempted to interpret friction by formulating basic equations. The the appearance of several approaches to studying tribological pro-
first contact model was proposed in 1881 by Hertz [9] while the first blems that range from experimental validations to numerical simula-
wear laws were proposed by Archard, who developed pin on disk tions. The innovative concept of ‘‘tribological triplet’’, proposed in the
experiments in 1953 [10–12]. These involved total wear volume, 1980s by Godet and Berthier [4,14], takes a different path from that
normal load, running distance and an additional physics—less of conventional schemes, as it underlines the importance of solving a
parameter k. Since Archard’s pioneering works and other directly tribological problem by simultaneously considering the bodies in
related studies, more than 100 wear models have emerged [13]. contact, the mechanism (mechanical system) that contains them, and
These models are based on the fact that the more parameters they the interface that separates them; the latter is called the third body
contain, the easier it is to match them to experimental results. To which refers to the so-called ‘‘first bodies’’ that it separates.
overcome this generality, Lim et al. [13] defined wear maps From an experimental viewpoint, at least four main difficulties
incorporating velocity and contact pressure, making it possible to can be identified when analysing a contact locally:
characterize the nature of wear (soft, hard, delamination, etc.).
Nevertheless, it was soon apparent that this kind of approach could  the confinement of the contact makes it difficult (if not
not be transferred from one contact condition to another. impossible) to perform measurements ‘‘in situ’’ without dis-
turbing the contact itself;
 the three components of the tribological system (mechanism,
 Corresponding author. first bodies and third body) act simultaneously, making it
E-mail address: mathieu.renouf@insa-lyon.fr (M. Renouf). difficult to determine the pilot component and its degree of

0301-679X/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
M. Renouf et al. / Tribology International 44 (2011) 834–844 835

interaction, bringing to mind the image of the snake swallow- area divided into sliding/sticking sub-regions depending on the
ing its own tail; type of stress involved. These models have been used since the
 the local response of the material (work-hardening, superficial first simulations of railway dynamics and can be found in several
tribological transformation (STT) [5,15,16], etc.) to tribological multi-body software applications dedicated to wheel–rail con-
stresses (quasi-hydrostatic pressure, shearing gradient, etc.); tacts. Rolling contacts are still the subject of many works and the
 tribometric measurements are difficult to transpose to real reader can consult specific works for more details [33–36].
mechanisms because they always correspond to specific Due to the huge number and variety of contact and friction
mechanical loads and physicochemical contact conditions. models available in the literature, it is not possible here to provide
a complete review. Therefore, those readers who are interested
To overcome such experimental difficulties, tribologists were can refer to the texts of Bowden and Tabor [37], Rabinowicz [38],
quick to recognise the potential of numerical tools for overcoming Suh [39], Kragelskii [40], and Dowson [8] for further details and
them; nowadays, simulations can be carried out at different references.
scales, taking advantage of the fact that it is possible to vary
one parameter at a time, something that is almost impossible
experimentally. Following a guideline based on the concept of the
tribological triplet, approaches at the scales of the mechanism, 3. Modelling mechanism and first bodies
the bodies in contact and the interface are dealing with in the
following sections. After presenting the first contact models 3.1. Modelling reality
in Section 2, Sections 3 and 4 present numerical tools used at
the scale of the mechanism and the first bodies and at the scale of Although previous analytical models were meaningful tools that
the interface, while Section 6 presents more recent advances in led to understanding of the phenomena involved in contact
dry numerical tribology. For each section, the variety of numerical mechanics and tribology, their limits quickly became apparent when
methods used in tribology at each scale is presented, the hypoth- dealing with more realistic issues in which the role of the mechan-
esis retained to consider the other scales (above or below) is ism, together with the contact, has to be accounted for. For example,
described, and a relevant example is given. predicting the lifetime of a mechanical component like a bearing,
blade disk or brake system must take into account both the dynamic
analysis of the mechanism up to its end-of-life and knowledge of the
2. First contact models local transformations of the bodies in contact (local deforma-
tions [15,41], surface degradation [42,43], etc.).
For many decades, researchers in the field of contact mechanics Thus additional modelling of both the dynamics of the whole
have tried to use simple models to describe the behaviour of the system (mechanism) and the local behaviour of the bodies in
relative motion between two bodies. Hertz was undoubtedly the contact (first bodies) is necessary. Nevertheless, the difference
first to propose a contact model [9] that initially comprised contact between the spatial and temporal scales of the two aspects of the
between a cylinder and a flat surface and then between two spheres problem implies that developing a single model that includes
of different radii. The model offered a simplified representation of both the global behaviour of the mechanism and the detailed
the local deformation of the body, depending on the elasticity of the description of the bodies in contact results in cumbersome
materials in contact, and provided the stress field in the contact area operations that are often impossible to achieve.
relating to the normal force, the radii of curvature of the two bodies Defining objectives as a function of the problem to be solved.
and their elastic modulus in a static configuration. This model is In the case where the model is designed to simulate the dynamic
valid for frictionless contacts, assuming that the contact area is small response of a mechanical system, the contact (either sliding or static
in comparison to the size of bodies in contact and that the stresses contact joints) is accounted for by reduced parameters that are
localised at the contact zone are less than the limit of elasticity. It assumed to represent the effect of the contact on the system’s
also assumes a continuous and smooth distribution of the pressure dynamics. These parameters generally consist of global contact
in the contact area. stiffness factors [44,45], contact damping factors [46] and energy
This theory formed the basis of many models, some of which loss factors if the model includes energetics [47]. Moreover, to
used the hypotheses of the original model. Extensions were overcome the complexity and the non-deterministic effects of the
proposed for other regular geometries [17,18], more complex nonlinearities of contact joints on the dynamics of a mechanism,
geometries and for accounting for the presence of friction and statistical techniques dealing with uncertainties have been devel-
inelastic behaviour [19]. oped, such as statistical energy analysis (SEA) [48]. In such models
In his model, Hertz considered perfectly smooth contact surfaces. the contact between the subsystems (components) of a complex
In the 1960s, Greenwood proposed a static contact model for rough mechanical system are represented by energy loss factors account-
surfaces [20]. This generalisation was to be the basis of many other ing for the loss of energy at the joint, and energy coupling factors
models proposed to describe contact surfaces by more or less accounting for the energy transfer between the subsystems. An
complex functions developed by the combination of trigonometric extensive review of these approaches and the huge amount of
functions [21], by the use of fractal model [22] or by numerical related literature is beyond the scope of this work, although readers
models involving finite element (FE) approaches [23–28]. will find the articles mentioned in the references.
Hertz’s theory was extended to models that took into account On the other hand, when dealing with the contact itself, two
dynamic aspects. Carter [29] proposed a model of a line contact to different approaches have been proposed over recent decades:
describe the contact between wheel and rail, accounting for analytical or semi-analytical models, and numerical models
longitudinal slip only. He highlighted the coexistence of areas of (finite element). These approaches may appear contradictory
sticking and sliding within the contact zone. Carter’s simplified although they sometimes complement each other.
model was then extended to circular contacts [18] and elliptic Table 1 shows a comparison between the different approaches,
contacts [30], allowing the consideration of transverse sliding. emphasizing their respective advantages, drawbacks and several
Based on the work of Johnson, Kalker proposed a set of algorithms typical applications. Readers new to the field can take advantage
using more or less complex models [31,32]. The contact area was of this initial comparison for choosing the approach to be used for
no longer seen as a uniform sliding or sticking region but as an each specific problem (Table 2).
836 M. Renouf et al. / Tribology International 44 (2011) 834–844

Table 1
Comparison between methods for modelling mechanism and first bodies.

Model Typical application Advantages Drawbacks Scales

Semi-analytical Transmission systems Low computational cost Lost of local contact information Mechanical system scale
Full bearing systems Full system modelling Simplified first body modelling Large time and space scale
Vehicle dynamics Long time simulation Application depending modelling

Energetical formulation Vibration analysis of jointed Low computational cost No local contact information First body scale or simple
structures Interaction between different Simplified first body modelling system scale
Fluid–structure coupling Behaviour of first bodies need of pre-modelling Only stationary states

Standard F.E. Transient model analysis of simple Local contact information Large computational cost result First body scale or simple
systems Full first body modelling precision depending on the contact system scale
Contact instabilities Availability of robust general modelling no third-body effects Contact space scale
First body contact analysis formulations Short time scale

Substructured F.E. Transient model analysis of simple Local contact information Reduced first body modelling result First body scale or simple
systems Reasonable computational cost precision depending on the contact system scale
Contact instabilities Robust formulation availability modelling Contact space scale
First body contact analysis No third-body effects Relative short time scale

Table 2
Comparison between continuous, discrete element (D.E.), coarse grain (CG), united atom (UA), molecular dynamics (MD), Monte Carlo (MC) and ab initio approaches for
modelling the third body and smaller system.

Model Typical application Advantages Drawbacks Scales

Continuous Powder lubrication Low computational cost Unknown a priori Interface scale
Rheological model interface behaviour
No wear modelling
No localisation phenomena

D.E. Powder lubrication No a priori behaviour Large computational cost Part of the interface
Wheel–rail contact Wear modelling Unknown a priori
Machining Localisation phenomena Interaction laws

C.G. Polymer modelling Large number of atoms represented by Difficult to find equivalent interaction Systems with very
an equivalent entity potential long molecules

U.A. Alkane modelling Few atoms grouped in equivalent energy Simplified molecule shape and Nanometer scale
sites interaction potentials
Biological molecules

M.D. Nano-indentation Deterministic simulation of atom motion Very high computational cost Nanometer scale

M.C. Solids at rest Fast to find equilibrium positions No explicit time evolution Nanometer scale

Ab Initio Atomic interactions Includes electron modelling Long computation for only a few atoms Atomic/ subatomic
determination scales

3.2. Semi-analytical models wear are often implemented; these models are then assembled
together and dialog with each other through their respective
Semi-analytical (SA) models were the first tools to be used inputs and outputs.
either for simplifying contact problems by trying to account for SA models have been used for understanding wear phenomena
mechanism behaviour or for understanding the behaviour of like fretting [1,50], in simplified elasto-plastic and wear models [49].
solids subjected to contact stresses over long periods of time. Other researchers have developed semi-analytical methods for
The latter approach is that most commonly adopted for modelling modelling ball bearing behaviour [2,51,3], by simplifying the inter-
the behaviour of mechanical systems over several hours opera- action between the bearing elements with visco-elastic contact
tion. In such cases computing-time becomes one of the main laws. The literature on semi-analytical models is extremely rich
constraints, forcing researchers to develop simplified models of and is growing rapidly every year, with specific branches that each
both the dynamic behaviour of system components and contact deal with particular issues, thus it is not possible to give a complete
stresses. These approaches have become efficient tools for solving overview of existing works and those readers interested should refer
well-defined problems more than general tools for understanding to more detailed reviews that deal with specific applications.
contact problems [1,49], because the phenomena and the scales of Nevertheless, it is worth mentioning the works of Nélias [52,1,49]
the mechanism and the contact are completely uncoupled. Con- concerning dry contact problems and the works of Wensing [7,3]
sequently one of the main drawbacks of these models, as pointed dealing with lubricated contact problems. A detailed review of
out in Table 1, is their specificity with respect to the problem to friction models for friction compensation methods in mechanical
be solved. There is no general formulation capable of being used systems with contact is provided by Armstrong-Héloury [53]. A
for several different problems. bridge between semi-analytical and continuous models is given by
The SA feature of the models described above can be used for the review presented by Oden and Martins [54].
time discretization in analytical models to simulate the evolution An interesting improvement and analysis of SA models can be
of a solution to a problem through time. Different simplified found in the works of Gallego et al. [49]. The authors propose an
analytical models accounting for friction, plastic deformation or alternative method to the FE method to perform wear computations
M. Renouf et al. / Tribology International 44 (2011) 834–844 837

on dovetail joints. Although an FE model is used to solve the effort and are thus still preferred for prediction over longer periods
structural problem, the SA model focuses on each contact area and of time.
its speed allow computing cyclic wear easily. Fig. 1 shows the wear An example of more phenomenological approaches is the effort
depth profiles after one fretting cycle with unworn surfaces (i.e. employed by researchers in modelling brake systems by FE meth-
after the first cycle). The solid line in black corresponds to the FE ods, in order to investigate the brake squeal phenomenon. A review
results used as a reference case while the lines with symbols of recent literature [57] dealing with the numerical analysis of brake
correspond to the results provided by the SA model with different squeal shows how FE transient analysis has been employed for
levels of complexity. reproducing and investigating squeal vibrations (Fig. 2) and local
contact stresses, while FE complex eigenvalue analysis has been
developed for predicting brake dynamic instabilities (Fig. 3). While
3.3. Numerical methods for continuous modelling FE models have been of fundamental importance for understanding
the mechanism at the origin of the self-excited friction vibrations
Finite element modelling is the most commonly used numer- that cause squeal noise, their limits have been highlighted as they
ical method for solving contact problems with friction between cannot be used to develop a predictive FE tool. The lack of
deformable bodies [55]. Initially, researchers in tribology were information at the scales of the third body and of the global
not attracted by this numerical tool. One of the main reasons was mechanical system, together with limitations on model size and
the extremely long computational time required in comparison the huge computational efforts demanded for transient analysis,
to more efficient analytical or semi-analytical approaches (see prohibit the development of a quantitative FE model of such systems
Table 1). Nevertheless, numerical approaches using finite element without updating the parameters through experimental comparison.
methods were given increasing attention by tribology journals by Therefore, despite several efforts, FE models remain more useful as
the end of the 1980s [56]. Nowadays, they have become an tools for understanding than for predicting.
essential method for solving contact problems where mechanism Several methods have been developed over recent decades for
and first body behaviour is the main subject of analysis. Notwith- solving contact problems between two deformable bodies. They are
standing, it is important to pay attention to the different appro- generally classified as a function of the contact algorithm and the
aches and methods that can be adopted. Indeed, both the time time integration scheme used. The contact algorithms most com-
integration method and the modelling of contact elements can monly used employ either Lagrange multiplicators [59–61], the
considerably affect the results of finite element simulations. regularization parameters [62], or LCP (Linear Complementary
Transient analysis by the finite element method is better adapted Problems) [63]. Time integration schemes can be either explicit or
to the analysis of local phenomena over short periods (from a few implicit for describing the motion of a structure and for dealing with
milliseconds to several seconds, according to the application) rather a contact [61,64,65,62].
than the simulation of the transient behaviour of a complex system As asserted for the semi-analytical models, it would be difficult
over long periods (several hours). The computational effort, as and, in any case, beyond the scope of this review to cite all the
opposed to semi-analytical methods, is no longer considered to be works dealing with contact problems and finite element methods.
a constraint, despite the fact that it can be cumbersome. Despite this The combination of these issues covers theoretical works dealing
expense in terms of computational effort, these models have been with convergence and the unicity of numerical solutions [66], works
proved as useful tools for understanding problems. Moreover, this dedicated to developing efficient algorithms [67,63,68,69], and an
limitation is being constantly reduced by numerical optimizations extremely wide range of different applications (e.g. [70,34]). With
that use clusters and parallel calculation. Nevertheless, semi-analy- respect to tribological issues, i.e. understanding the causes and
tical models continue to be more efficient concerning computational effects of friction in bodies in contact, the works of Ham [56] and

Fig. 1. Wear depth profiles after one fretting cycle with unworn surfaces (i.e. after the first cycle). The solid line in black corresponds to the FE results while the lines with
symbols correspond to the results provided by the SA model with various levels of complexity [49].
838 M. Renouf et al. / Tribology International 44 (2011) 834–844

Contact forces between pad ans disc x 104 Y-acceleration of pad edge

Acceleration [m.s2]
Normal force 0.5
Force [N]
100 0

0 friction force
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time [s] Time [s]

Fig. 2. Transient FE analysis of brake squeal: (a) normal and tangential forces between brake pad and disc; (b) tangential acceleration calculated at the brake pad. FE
simulation makes it possible to reproduce the vibrations triggered in the contact [58].

Fig. 3. Complex eigenvalue FE analysis of brake squeal: (a) natural frequencies of the brake system cross each other and can become unstable; (b) locus plot of system
eigenvalues: the real part of one eigenvalue become positive, i.e. the system is unstable.

Baillet [59,60] deserve mention. Others works have aimed at under- case of a fluid lubricant, and the same idea can be generalised to
standing specific contact phenomena, for example the behaviour of solid interfaces. Indeed, Godet [14] and Berthier [4] showed that this
composite materials subjected to tribological stresses [71], the more or less heterogeneous and discontinuous interface called
development of contact instabilities [72], wear laws in biomecha- ‘‘third body’’ can have different thicknesses (ranging from several
nics [73], the occurrence of rail defects [34], dynamic instabilities nanometers [82] to several tens of micrometers [83]) and flow with
leading to squeal noise [58,74], contact rupture [75,76], contact an as yet unknown rheology. This discontinuous interface is present
wave generation and propagation [77–79]. The reader will find a in an artificial way in lubrication powder (e.g. MoS2 for space
more complete bibliography concerning each specific domain in the lubrication, etc.) and, in this case, the properties of the media are
works cited. assumed to be known. Nevertheless, the evolution of such media
As highlighted in Table 1, recent advances in substructuring and their behaviour in response to strong pressures and shear rates
methods make it possible to consider a new approach halfway could change, making the interpretation of results complex. There-
between semi-analytical and FE models [80,81]. Substructured FE fore having a continuous view of this kind of interface does not
modelling allows accounting for contact distribution, the connec- always ensure that the reality of the contact is apparent, especially
tions between substructures, and it drastically reduces the num- when the wear rate is accounted for, or when the third body flows
ber of degrees of freedom of the global FE model. First, single are not in steady state (which is rarely the case).
substructures are modelled and the matrix reduction technique is Two kinds of approaches are commonly used to model the third
used to reduce the system matrices to a smaller set of degrees of body flow: continuous and discrete approaches. Continuous
freedoms. Then the reduced matrices of the single substructures approaches offer extensions of models derived from continuum
are assembled together to account for the general boundary mechanics and are essentially used in applications involving lubri-
conditions and the contact stresses at the contact interfaces cation powder, i.e. using an artificial third body. In the case of a
between single substructures. This approach dramatically reduces natural third body (resulting for the most part from the bodies in
computational effort at the expense of simplifying (reducing) contact), continuity hypotheses cannot be retained and, under such
modelling of the first bodies (substructure) response. conditions, it is necessary to verify whether this approach is still
valid and what its limits are. This is the main reason for using
discrete element approaches to describe the evolution of a discon-
4. Within the tribological interface tinuous medium and to infer rheological behaviour.

4.1. From reality to modelling

4.2. Continuous approaches
When focusing on the interface that separates two bodies in
contact, flows of the latter can be observed under certain conditions There is no universal method for describing lubrication powder or,
(high pressures and shear rates). This is easily understandable in the more generally, granular flows. Therefore many approaches attempt
M. Renouf et al. / Tribology International 44 (2011) 834–844 839

to describe such flows that are often reduced to specific geometries. interaction forces between particles must satisfy the mechanical
In the granular tribology community, the most widely used equations equilibrium of the system and use local laws with a physical
are those proposed by Haff [84] or Lun [85]. They are used in many sense involving the fewest possible calibration parameters.
applications involving bearings [51], Couette geometries (annu- Ten years after the first discrete element simulations in fluid
lar) [86], parallel plates [87,88] and converging plates [89]. lubrication [103,104], discrete element methods were used for
The equations of conservation (mass and momentum) are used the first time to simulate lubrication by bearing powder [105].
to describe the continuous behaviour of the medium [90]. These The third body is represented by a collection of rigid spheres
equations are similar to the energetic equations used for fluids in moving between two rough inclined planes. This work paved the
which variations of granular temperature are a balance between way for many succeeding studies. The first were very exploratory
the energy injected into the system and the energy dissipated by and concerned the influence of different numerical [100,106]
friction and collisions. In this set of equations, the system is and geometric parameters (size of samples and particles) [107]
closed by a constitutive equation that connects the stress tensor on the mechanical response of the medium. Then more repre-
to the strain tensor. Regarding the available models satisfying sentative laws involving phenomena within an interface were
such a condition, the reader can refer to the work of Hesh- used (JKR contact model [108]) [109] and the influence of particle
mat [91,92], Tichy [93,94,89] or those of Khonsari [88] involving size and the cohesion between particles on the flow behaviour were
different geometries. studied.
These methods were then extended to the study of wear
4.3. Discrete approaches mechanisms with simple shear [110,111], offering wear laws
based on discrete approaches. Other works concerned third body
Discrete element approaches allow representing the interface flows in fretting-like simulations [112].
as a collection of more or less heterogeneous particles indepen- More recently some authors have accounted for the deformation
dent of each other. This type of method emerged in the early of particles in simple shear simulations [101,113]. The models are
1970s to model the evolution of rockfalls [95,96]. It was quickly not representative of dense granular flows observed experimentally,
adapted to modelling numerous problems involving media with but they propose a new approach in the simulation of interfaces that
divided features (geophysical structures [97], ballast [98], masonry accounts for particle deformation. Fig. 4 presents the sample used in
[99], etc.). the works of Kabir et al. [101] using an explicit dynamic finite
The equations of continuum mechanics are no longer used to element model of parallel shear cells, containing 52 particles and
describe the evolution of systems, instead the equations of motion consisting of round and multishaped granules (diamond, triangle,
and laws managing the interactions between particles are imple- and rectangle). From a different perspective this kind of approach is
mented. They play the role of conservation and constitutive equations used to revisit the theory of Hertz in a medium with inclusions
respectively. Although there are many variations of the discrete [114].
element method (DEM) used in tribology, they all rely on three key Although several improvements have been performed in the
ingredients: last decade, three questions remain unanswered relating to
modelling the third body by using the discrete element method.
 the time integration of the equation of mechanics; The first question is that of how to choose the size of a single
 contact detection between particles; particle, since numerical particles do not represent real particles.
 computation of interaction forces between particles. The shapes used in models remain geometrically simple (disks or
spheres according to the dimension) and exhibit bulk behaviour
Time integration must not modify the mechanical balance of the (essentially rigid even if recent improvements take into account
system. Indeed, if a given energy is injected into the system, it is elastic deformation). Consequently, numerical particles play the
not dissipated numerically by the numerical scheme used role of (representative?) elementary volume when used with
(Euler [100], Newmark [101], theta-method [64], Verlet [102], homogenisation techniques, while the law that governs their
etc.). Contact detection must be accurate to properly describe the interaction represents the integration of local phenomena.
connections between the particles of the system. Although this The determination of such laws is the second main open
appears easy for spherical particles, it is more difficult when question. How can local elasticity, plasticity, viscosity, damage,
bodies with complex geometries are used. Finally, computation of etc. be accounted for? Models use phenomenological laws based

Fig. 4. Two-dimensional 52 particle granular flow shear cell finite element model [101]. (a) Round particles and (b) Multi-shaped particles.
840 M. Renouf et al. / Tribology International 44 (2011) 834–844

on basic assumptions. This is due to the lack of experimental MC approach solves the Boltzman equation where the motions
results concerning real interactions of third body particles. There- of atoms are simulated by statistical laws. This method proves to
fore current models offer only phenomenological results and be very efficient for finding equilibrium positions but is unable to
should not be considered as generic. obtain an explicit evolution for atoms through time. On the
The last open question is related to spatial and temporal contrary, MD simulation sums up inter-atomic forces, and then
scales. Simulating third body flows takes into account only a solves the classical Newton equation of motion over each atom,
small part of the contact interface within a short time. How can providing a deterministic way of simulating their displacement
these local results be extended to the evolution of a whole through time. Nevertheless they are very computationally costly
contact? At present mean local values are used by the upper and simulate only several nanometers. A variety of adaptations of
scale. The Section 6.1 will propose a few alternatives. molecular dynamics can be found in the literature to increase the
size of the system, ranging from nanoscale to higher scales. Atoms
can be grouped together, represented in only one point of space
5. Nanometric characterization of interfaces and their interaction potentials merged together in order to
define an equivalent interaction site (united atoms, coarse grain
One of the major problems for discrete simulations of third methods). However, when going up in scale, as soon as geometry
body flows at the mesoscale (because of the complexity of matter) and potentials are simplified in the models (as matter becomes
is the quantitative definition of the interaction forces between increasingly complex) the latter looses its precision.
particles. Nevertheless, at the scale of the atoms and molecules Concerning ‘‘nano-tribology’’, early models for atomic friction
composing this medium, it is possible to define reliable inter- such as the Tomlinson model [117] or Frenkel–Kontorova model [118]
atomic laws from quantum calculations [115]. Thus two main involve two crystalline surfaces in direct contact, assuming that each
kinds of method are used to describe the motion of atoms: the surface atom is linked only to the center of the mass of the rubbing
Monte Carlo (MC) approach and molecular dynamics (MD) [116]. body [117] or that the surface atoms are linked only to their nearest

Fig. 5. Snapshots showing the evolution of a Cu(1 0 0) tip on a Cu(1 0 0) substrate during sliding to the left [119].
M. Renouf et al. / Tribology International 44 (2011) 834–844 841

neighbour by springs [118]. With the development of computers and At the scale of the interface, models offer richer descriptions
the use of both MC and MD methods, these friction models have been by mixing discrete and continuous approaches in a single simula-
developed over the years to represent a crystalline layer of adsorbed tion [134,133]. The influence of the deformation of the bodies in
atoms, including rearrangements of atoms due to the equilibrium contact is directly accounted for by the rheology of the interfaces.
condition (i.e. temperature) or sliding. Other models provide a dialogue between different scales by
To investigate atomic scale asperities and finite contact areas, applying an iterative method [134,135]. The friction at the scale of
many authors (such as [119], see Fig. 1) have modelled AFM-like the bodies in contact is the result of a simulation performed at the
tips-on-substrate contacts. Since the shearing velocity in classical scale of the interface (third body), while the boundary conditions
MD simulations is very high compared to real AFM experiments, of the simulations at the interface scale are provided by the local
large scale molecular dynamics simulations are still under devel- dynamics at the scale of the first bodies.
opment, involving massive parallel programming [120] (Fig. 5). Some models couple the finite element method and molecular
Many tip-substrate simulations focusing on indentation were dynamics to account for a richer description of the interface and
performed in the 1990’s [121,122]. They showed that when the surface condition [136,137], while other methods couple
approaching the two surfaces (usually metallic surfaces) to each discrete models at the scale of the material’s structure to reflect
other, the normal force remains almost null, after which it suddenly local degradation in gear dynamics [138].
becomes extremely attractive with a jump to contact. This phenom-
enon was not so clear when approaching two diamond surfaces in 6.2. Towards multi-physics models
the case where at least one was hydrogen-terminated [123].
Regarding sliding, nanoscale simulations often highlighted inter- Multi-physics models are commonly used at the scale of the
mittent motion, with the system alternatively sticking and slipping first bodies. Indeed, thermo-mechanical models based on robust
forward, as described by Rabinowicz [38] and Bowden and formalism have been implemented [139].
Tabor [124]. The stiffness of the contacting bodies and the commen- Concerning solid interfaces, multi-physics models have been
surability of the surface, i.e. the way atoms are arranged regularly (or developed only recently. Models coupling mechanical and electrical
not) depending on the crystalline lattice orientation [125–127], play aspects have been developed to understand phenomena such as
an important role in the dynamic friction process. shunting in the wheel–rail contact [140]; thermo-mechanical mod-
The first tribochemical reactions in molecular dynamics simu- els have been developed to describe energy dissipation within the
lations were observed by Harrison and Brenner in 1994 [128], contact interfaces [141–143]. This work allowed a new approach to
who used reactive potentials that allow for the breaking and the concept of friction by incorporating thermal discontinuous
formation of chemical bounds. This brought to light the origin of aspects and updating the Blok theory [144].
the formation of molecular wear debris. More recent models are capable of coupling mechanical,
An impressive compilation has been proposed in [129], where the thermal and physical chemistry within the interface. This kind
reader will find many complementary aspects related to this section. of approach allows linking every physical feature within a single
simulation [145,146]. The limits of these approaches depend on
the experimental data used feed the models.
6. Numerical tribology today

During the last decade, numerical tribology has evolved con- 7. Discussion and conclusion
siderably. It has taken advantage of the numerous numerical
developments achieved in the domain of contact mechanics and The numerical tools used at the scales of the mechanism, the
adapted them to its own needs. Numerical tribology has now bodies in contact and the interface have been presented following
become an indispensable tool for studying contact problems. a guideline based on the concept of the tribological triplet. The
Nevertheless, the need for in-depth understanding of the semi-analytical model, finite element method, discrete element
phenomena involved is leading to the development of models method and molecular dynamics approach appear to be the most
that must be as accurate as possible. They are neither longer commonly used tools.
purely mechanical, nor based on only one scale, as they incorpo- At the scale of the mechanism and the first bodies, semi-
rate different physics and consider several scales. analytical models and finite element methods appear to be the most
appropriate tools for describing the behaviour of an entire system
and that of each component. SA models can be used to simulate the
6.1. Towards multi-scale models long-term evolution of full systems at low computational cost.
Nevertheless, they entail a loss of local contact information, the use
In the past, most of the numerical work in tribology was of simplified first body modelling, and models dedicated to only
performed at a single scale, accounting for the other scales by one application. The finite element method made full first body
boundary conditions or local conditions. Although the idea of modelling and the measurement of local contact characteristics
coupling the different scales is not a new one [5,130], it is only possible. Moreover, robust general formulations are available. On
recently, in tribology, that different approaches have tried to the other hand, the approach is computationally costly when the
account for the different scales of the tribological triplet within a results depend on contact modelling and it is not possible to simulate
single simulation, either directly or indirectly. the long-term evolution of a contact, even when using substructured
For example, certain works couple the mechanism and bodies approaches.
in contact to analyse bearings [51,131]. In the first case, a nodal At the scale of the interface, rheological models are computa-
reduction is performed on the whole system and coupled with tionally inexpensive, enabling description through wear model-
condensation techniques. In the second case, the mechanism ling and reproducing localisation phenomena. Finally, they can be
results from a multi-body model and injected as boundary used to postulate an a priori unknown interface behaviour. On the
conditions into the finite element simulation of the contact. contrary, the discrete element method can be used to consider
Other works couple smaller scales (the scale of the bodies in wear and localisation processes but suffers from high computa-
contact and the interface) to account for the effect of local damage tional costs and is based on a priori unknown interaction laws (in
on the behaviour of bodies in contact [132,133]. most cases).
842 M. Renouf et al. / Tribology International 44 (2011) 834–844

At the nanoscale, molecular dynamic approaches provide good [25] Pei L, Hyun S, Molinari JF, Robbins MO. Finite element modeling of elasto-
descriptions of matter and are well-suited to nano-applications, plastic contact between rough surfaces. J Mech Phys Solids 2005;53:
but in the case of more microscopic problems, mapping informa- [26] Persson BNJ. Elastoplastic contact between randomly rough surfaces. Phys
tion becomes complex. Rev Lett 2001;87:116101.
Although many studies at different scales have led to clarifying [27] Persson BNJ, Alboht O, Tartaglino U, Volokitin AI, Tosatti E. On the nature of
the concept of friction, there is still some way to go before reliable surface roughness with application to contact mechanics, sealing, rubber
friction and adhesion. J Phys: Condens Matter 2005;17:R1–62.
predictive models of friction and wear can be obtained. Until we [28] Persson BNJ, Sivebaek IM, Samoilov VN, Khao K, Volokitin AI, Zhang Z. On
are able to model a full system, friction will remain a scientific the origin of amonton’s friction law. J Phys: Condens Matter 2008;20:
smokescreen. A great deal of research work is currently taking 395006.
[29] Carter FW. On the action of a locomotive driving wheel. Proc R Soc London,
this direction and integrating different physics, with dialogue
Ser A 1926;112:151–7.
between different scales. In conclusion, it is possible to assert [30] Haines J, Ollerton E. Contact stress distribution on elliptical contact surfaces
that, thanks to numerical tribology and its interactions with subjected to radial and tangential forces. Proc Int Mech Eng 1963;177:
experimental tribology, tribologists are wearing friction down. 95–114.
[31] Kalker JJ. The computation of three-dimensional rolling contact with dry
friction. Int J Numer Methods Eng 1979;14(1293–1307):1293–307.
[32] Kalker JJ. Mathematical models of friction for contact problems in elasticity.
Acknowledgements Wear 1986;113:61–77.
[33] Colette C. Usure ondulatoire des rails dans les transports urbains. Master’s
thesis, Université libre de Bruxelle; 2003.
Part of this work is funded by the Agence National de la [34] Saulot A, Baillet L. Dynamic finite element simulations for understanding
Recherche in the framework of the ANR-JCJC08-0020 DiNEET wheel–rail contact instability states occurring under sliding conditions.
ASME J Tribol 2006;128:761–70.
project (Dialogue Numérique Entre Echelles Tribologiques). [35] Andersson C, Johansson A. Prediction of rail corrugation generated by three-
dimensional wheel–rail interaction. Wear 2004;257(3–4):423–34.
[36] Telliskivi T, Olofsson U. Wheel–rail wear simulation. Wear 2004;257(11):
References 1145–53.
[37] Bowden FP, Tabor D. The friction and lubrication of solids. 2nd corrected ed.
Oxford University Press, Clarendon Press; 2001. [oxford classic text in the
[1] Gallego L, Nélias D, Deyber S. A fast and efficient contact algorithm for fretting
physical sciences edition].
problems applied to fretting modes i, ii and iii. Wear 2010;268(1–2):208–22.
[38] Rabinowicz E. Friction and wear of materials. New York: Wiley; 1965.
[2] Morales-Espejel G. Semi-analytical methods for surface life analysis in
elastohydrodynamically lubricated and dry contacts; 2009. [39] Suh NP. Tribophysics. PrenticeHall; 1986.
[3] Wensing JA. On the dynamics of ball bearings. PhD thesis, University of [40] Kragelskii IV. Friction and wear. London: Butterworth; 1965.
Twente; 1998. [41] Carroll RI, Beynon JH. Rolling contact fatigue of white etching layer: Part 1:
[4] Berthier Y. Experimental evidence for friction and wear modelling. Wear crack morphology. Wear 2007;262(9–10):1253–66.
1990;139:77–92. [42] Cristol-Bulthe A-L, Desplanques Y, Degallaix G. Coupling between friction
[5] Berthier Y. Handbook of materials behavior models. In: Background on physical mechanisms and transient thermal phenomena involved in pad-
friction and wear. Lemaı̂tre Academic Press; 2001. p. 676–99. disc contact during railway braking. Wear 2007;263(7–12):1230–42.
[6] Massi F, Giannini G, Baillet L. Brake squeal as dynamic instability: an [43] Kasem H, Bonnamy S, Berthier Y, Jacquemard P. Fiber-matrix unbonding
experimental investigation. J Acoust Soc Am 2006;120(3):1388–99. and plastic deformation in c/c composites under tribological loading. Wear
[7] Tillema H. Noise reduction of rotating machinery by viscoelastic bearing 2010;269(1–2):104–11.
supports. PhD thesis, University of Twente; 2003. [44] Nobari AS, Robb DA, Ewins DJ. Model updating and joint identification, their
[8] Dowson D. History of tribology. 2nd ed. London and Bury St. Edmunds, ranges of application, restriction and overlap. IJAEMA 1993;8(1):93–105.
England: Professional Engineering Publishing; 1998. [45] Liu W, Ewins DJ. Substructure synthesis via elastic medi part i: joint
[9] Hertz H. Über die berührung fester elastischer körper (sur le contact entre identification. In: Proceedings of the 18th international modal analysis
corps élastiques). J für reine und angewandte Mathematik 1881;92:156–71. conference, San Antonio, USA, vol. II; 2000. p. 1160–6.
[10] Archard JF. Contact and rubbing of flat surfaces. J Appl Phys 1953;24(8): [46] Gaul L, Bohlen S. Identification of nonlinear structural joint models and
981–8. implementation in discretized structure models. In: Proceedings of the 5th
[11] Archard JF. Elastic deformation and the contact of surfaces. Nature international modal analysis conference, vol. I. London; 1987. p. 86–91.
1953;172(4385):918–9. [47] Ji L, Mace BR. Statistical energy analysis modelling of complex structures as
[12] Archard JF. Elastic deformation and the laws of friction. Proc R Soc London, coupled sets of oscillators: ensemble mean and variance of energy. J Sound
Ser A 1957;243:190–205. Vib 2008;317:760–80.
[13] Lim SC, Ashby MF, Brunton JH. Wear-rate transitions and their relationship [48] Culla A, Sestieri A, Carcaterra A. Energy flow uncertainties in vibrating
to wear mechanisms. Acta Metall 1987;35:1343–8. systems: definition of a statistical confidence factor. Mech Syst Signal
[14] Godet M. The third-body approach: a mechanical view of wear. Wear Process 2003;17(3):635–63.
1984;100:437–52. [49] Gallego L, Fulleringer B, Deyber S, Nélias D. Multiscale computation of
[15] Busquet M, Descartes S, Berthier Y. Formation conditions of mechanically fretting wear at the blade/disk interface. Tribol Int 2010;43(4):708–18.
modified superficial structures for two steels. Tribol Int 2009;42(11–12): [50] Fouvry S, Kapsa Ph, Vincent L. An elastic-plastic shakedown analysis of
1730–43. [Special issue: 35th Leeds–Lyon symposium]. fretting wear. Wear 2001;247(1):41–54.
[16] Eleod A, Oucherif F, Devecz J, Berthier Y. Conception of numerical and [51] Weinzapfel N, Sadeghi F. A discrete element approach for modeling cage
experimental tools for study of the tribological transformation of surface
flexibility in ball bearing dynamics simulations. ASME J Tribol 2009;131:
(tts). In: Dowson D, editor. Lubrication at the frontier, proceedings of the
25th Leeds–Lyon symposium on tribology. Tribology series, vol. 36. Else-
[52] Boucly V, Nélias D, Liu SB, Wang QJ, Keer LM. Contact analyses for bodies
vier; 1999. p. 673–82.
with frictional heating and plastic behavior. ASME J Tribol 2005;127(2):
[17] Gladwell GML. Contact problems in the classical theory of elasticity. Alphen
aan den Rijn: Sjthoff and Noordhoff; 1980.
[53] Armstrong-Hlouvry B, DuPont P, Canudas de Wit C. A survey of models,
[18] Johnson KL. The effect of a tangential contact force upon the rolling motion
analysis tools and compensation methods for the control of machines with
of an elastic sphere on a plane. ASME J Appl Mech 1958;25:339–46.
[19] Barber JR, Ciavarella M. Contact mechanics. Int J Solids Struct 2000;37: friction. Automatica 1994;30(7):1093–138.
29–43. [54] Oden JT, Martins JAC. Models and computational methods for dynamic
[20] Greenwood JA, Williamson JBP. Contact of nominally flat surfaces. Proc R friction phenomena. Comput Methods Appl Mech Eng 1985;52:527–634.
Soc London, Ser A 1966;295:300–19. [55] Zienkiewicz OC, Taylor RL. The finite element method in engineering
[21] Ciavarella M, Delfine V, Demelio V. A new 2d model with interaction for science. London: McGraw-Hill; 2000.
studying the contact of multiscale rough random profiles. Wear 2006;261: [56] Ham GT, Rubin CA, Hahn GL, Bharghava GV. Elasto-plastic finite element
556–67. analysis of repeated, two-dimensional rolling sliding contacts. ASME J
[22] Ciavarella M, Demelio G. Elastic multiscale contact of rough surfaces: Tribol 1988;110:44–9.
Archard’s model revisited and comparisons with modern fractal models. [57] Ouyang H, Nack W, Yuan Y, Chen F. Numerical analysis of automotive disc
ASME J Appl Mech 2001;68:496–8. brake squeal: a review. Int J Vehicle Noise 2005;1:207–31.
[23] Karpenko YA, Akay A. A numerical model of friction between rough [58] Massi F, Baillet L, Giannini O, Sestieri A. Brake squeal: linear and nonlinear
surfaces. Tribol Int 2001;34:531–45. numerical approaches. Mech Syst Signal Process 2007;21(6):2374–93.
[24] Pearson JD, Zikry MA, Wahl K. Computational design of thin-film nano- [59] Baillet L, Sassi T. Simulations numériques de différentes méthodes d’élé-
composite coatings for optimized stress and velocity accommodation ments finis pour les proble mes de contact avec frottement. C R Mecanique
response. Wear 2009;267:1137–45. 2003;331(11):789–96.
M. Renouf et al. / Tribology International 44 (2011) 834–844 843

[60] Baillet L, Linck V, D’Errico S, Laulagnet B, Berthier Y. Finite element [95] Cundall PA. A computer model for simulating progressive large scale
simulation of dynamic instabilities in frictional sliding contact. ASME J movements of blocky rock systems. In: Proceedings of the symposium of
Tribol 2005;127(July):652–7. the international society of rock mechanics, vol. 1; 1971. p. 132–50.
[61] Carpenter NJ, Taylor RL, Katona MG. Lagrange constraints for transient finite [96] Cundall PA, Strack ODL. A discrete numerical model for granular assemblies.
element surface contact. Int J Numer Methods Eng 1991;32:103–28. Géotechnique 1979;29(1):47–65.
[62] Wriggers P. Finite element algorithms for contact problems. Arch Comput [97] Renouf M, Dubois F, Alart P. Numerical investigations of fault propagation
Methods Eng 1995;2:1–49. and forced-fold using a non smooth discrete element method. Rev Euro
[63] Chabrand P, Dubois F, Raous M. Various numerical methods for Method Num 2006;15:549–70.
solving unilateral contact problems with friction. Math Comput Modell [98] Saussine G, Cholet C, Dubois F, Bohatier C, Gautier PE, Moreau JJ. Modell-
1999;28(4–8):97–108. ing ballast behaviour under dynamic loading. Part 1: a 2d polygonal
[64] Jean M. The non smooth contact dynamics method. Compt Methods Appl discrete element method approach. Comput Methods Appl Mech Eng
Math Eng 1999;177:235–57. 2006;195(19–22):2841–59.
[65] Klarbring A. A mathematical programming approach to three-dimensional [99] Acary V, Jean M. Numerical modeling of three dimensional divided
contact problems with friction. Comput Methods Appl Mech Eng 1986;58: structures by the non smooth contact dynamics method: application to
175–200. masonry structure. In: Topping BHV, editor. The fifth international con-
[66] Basseville S. Dynamique et stabilité d’un syste me discret en présence de ference on computational structures technology, 6–8 September. Civil-
contact et de frottement. PhD thesis, Université d’Aix-Marseille II; 2004. Comp Press; 2000. p. 211–22.
[67] Alart P. Méthode de newton généralise en mécanique du contact. J Math [100] Lubrecht AA, Berthier Y. Granular lubrication; a simple model and trends.
Pures Appl 1997;76:83–108. Tribol Ser 1995;30:53–62.
[68] Jourdan F, Alart P, Jean M. A Gauss–Seidel like algorithm to solve frictional [101] Kabir A, Jasti VK, Higgs III CF, Lovell M. An evaluation of the explicit finite
contact problem. Comp Meth Appl Mech Eng 1998;155:31–47. element method approach for modelling dense flows of discrete grains in a
[69] Renouf M, Alart P. Conjugate gradient type algorithms for frictional multi- coquette shear cell. I Mech E 2008;222:715–23.
contact problems: applications to granular materials. Comput Methods [102] Iordanoff I, Se ve B, Berthier Y. Solid third body analysis using a discrete
Appl Mech Eng 2004;194(18–20):2019–41. approach: influence of adhesion and particle size on the macroscopic
[70] Busquet M, Baillet L, Bordreuil C, Berthier Y. 3D finite element investigation behavior of the contact. ASME J Tribol 2002;124:530–8.
on the plastic flows of rolling contacts—correlation with railhead micro- [103] Lane JE, Spurling TH, Freasier B, Perram JW, Smith ER. Inconsistency of the
structural observations. Wear 2005;258(7–8):1071–80. density-functional theory of adsorption when using computer simulations.
[71] Peillex G, Baillet L, Berthier Y. Dynamic behavior of a homogenized Phys Rev A 1979;20-5:2147–53.
composite under contact with friction loading. C R Mecanique 2007;335: [104] Magda JJ, Tirrell M, Davis HT. Molecular dynamics of narrow, liquid-filled
394–400. pores. J Chem Phys 1985;63-4:1886–901.
[72] Linck V, Baillet L, Berthier Y. Modeling the consequences of local kinematics [105] Elrod HG, Brewe DE. Numerical experiments with flows of elongated
of the first body on friction and on the third body sources in wear. Wear granules. Tribol Ser 1991;21:219–26.
2003;255:299–308. [106] Lubrecht AA, Chan-Tien C, Berthier Y. A simple model for granular lubrica-
tion influence of boundaries. In: Dowson D, editor. The third body concept.
[73] Jourdan F. Numerical wear modeling in dynamics and large strains:
Tribology series, vol. 31; 1995.
application to knee joint prostheses. Wear 2006;261(3–4):283–92.
[107] Ghaouti A, Chaze M, Dubujet P, Sidoroff F. Particulate and granular
[74] Massi F, Berthier Y, Baillet L. Contact surface topography and system
simulation of the third body behaviour. Tribol Ser 1996;31:355–65.
dynamics of brake squeal. Wear 2008;265(11–12):1784–92.
[108] Johnson KL, Kendall K, Roberts AD. Surface energy and the contact of elastic
[75] Dang Van K, Maitournam MH. On some recent trends in modelling of
solids. Proc R Soc London A 1971;324:301–13.
contact fatigue and wear in rail. Wear 2002;253(1–2):219–27.
[109] Seve B, Iordanoff I, Berthier Y. A discrete solid third body model: influence
[76] Baietto MC, Rannou J, Gravouil A, Pelletier H, Gauthier C, Schirrer R. 3D
of the intergranular forces on the macroscopic behaviour. Tribol Interface
crack network analysis during a scratch test of a polymer: a combined
Eng Ser 2001;39:361–8.
experimental and multigrid X-fem based numerical approach. Tribology
[110] Fillot N, Iordanoff I, Berthier Y. A granular dynamic model for the degrada-
International, doi:10.1016/j.triboint.2010.04.014.
tion of material. ASME J Tribol 2004;126(3):606–14.
[77] Ben-Zion Y. Dynamic ruptures in recent models of earthquake faults. J Mech
[111] Fillot N, Iordanoff I, Berthier Y. Simulation of wear through a mass balance
Phys Solids 2001;49:2209–44.
in a dry contact. ASME J Tribol 2005;127(1):230–7.
[78] Adams GG. An intersonic slip pulse at a frictional interface between
[112] Renouf M, Saulot A, Berthier Y. Third body flow during a wheel–rail
dissimilar materials. J Appl Mech 2001;68:81–6.
interaction. In: Mota Soares CA, editor. III European conference on
[79] Di Bartolomeo M, Meziane A, Massi F, Baillet L, Fregolent A. Dynamic
computational mechanics solids; 2006.
rupture at a frictional interface between dissimilar materials with aspe-
[113] Kabir A, Lovell M, Higgs III CF. Utilizing the explicit finite element method
rities. Tribol Int 2010;43(9):1620–30.
for studying granular flows. Tribol Lett 2009;29(2):85–94.
[80] Zhaojun B, Ren-Cang L. Substructuring, dimension reduction and applica- [114] Raje N, Sadeghi F, Rateick Jr. RG, Hoeprich MR. Evaluation of stresses around
tions: an introduction. In: Dongarra J, Madsen K, Wasniewski J, editors. inclusions in hertzian contacts using the discrete element method. ASME J
Applied parallel computing. Lecture notes in computer science, vol. 3732. Tribol 2007;129:283–91.
Berlin, Heidelberg: Springer; 2006. p. 266–6. [115] Doucet J-P, Weber J. Computer-aided molecular design: theory and applica-
[81] Magoule s F. Substructuring techniques and domain decomposition meth- tions. Elsevier; 1996.
ods. Kippen: Saxe-Coburg Publications; 2010. [116] Allen MP, Tildesley DJ. Computer simulations of liquids. Oxford: Clarendon
[82] Descartes S, Berthier Y. Rheology and flows of solid third bodies: back- Press; 1987.
ground and application to an mos1.6 coating. Wear 2002;252(7–8):546–56. [117] Tomlinson GA. A molecular theory of friction. Phil Mag Ser 1929;7:905–39.
[83] Berthier Y, Descartes S, Busquet M, Niccolini E, Desrayaud C, Baillet L, [118] Frenkel YI, Kontorova T. On the theory of plastic deformation and twinning.
Baietto-Dubourg MC. The role and effects of the third body in the wheel– Zh Eksp Teor Fiz 1938;8:1340.
rail interaction. Fatigue Fract Eng Mater Struct 2004;27:423–36. [119] Sorensen MR, Jacobsen KW, Stoltze P. Simulations of atomic-scale sliding
[84] Haff PK. Grain flow as a fluid-mechanical phenomenon. J Fluid Mech friction. Phys Rev B 1996;53:2101–13.
1983;134:401–30. [120] Martini A, Dong Y, Perez D, Voter AF. Low-speed atomistic simulation of
[85] Lun CKK. Kinetic theory for granular flow of dense slightly inelastic, slightly stick–slip friction using parallel replica dynamics. Tribol Lett 2009;36:
rough spheres. J Fluid Mech 1991;233:539–59. 63–8.
[86] Yu C-M, Craig K, Tichy JA. Granular collision lubrication. J Rheol [121] Landman U, Luedtke WD. Nanomechanics and dynamics of tip-substrate
1994;38(4):921–36. interactions. J Vac Sci Technol B 1991;9:414–23.
[87] Hou K, Kalousek J, Magel E. Rheological model of solid layer in rolling [122] Harrison JA, Stuart SJ, Brenner DW. Atomic scale simulation of tribological
contact. Wear 1997;211:134–40. and related phenomena. In: Handbook of micro/nanotribology. CRC Press;
[88] Zhou L, Khonsari MM. Flow characteristics of a powder lubricant sheared 1999. p. 525.
between parallel plates. ASME J Tribol 2000;122(1):147–55. [123] Harrison JA, White CT, Colton RJ, Brenner DW. Nanoscale investigation of
[89] Sawyer WG, Tichy JA. Lubrication with granular flow: continuum theory, indentation, adhesion and fracture of diamond (111) surfaces. Surf Sci
particle simulations, comparison with experiment. ASME J Tribol 2001;123: 1992;271:57–67.
777–84. [124] Bowden FP, Tabor D. The friction and lubrication of solids. 2nd corrected ed.
[90] Wornyoh EYA, Jasti VK, Higgs III CF. A review of dry particulate lubrication: Oxford University Press, Clarendon Press; 1986. [oxford classic text in the
powder and granular materials. ASME J Tribol 2007;129:438–49. physical sciences edition].
[91] Hesmat H. The rheology and hydrodynamics of dry powder lubrication. [125] He G, Müser MH, Robbins MO. Adsorbed layers and the origin of static
Tribol Trans 1991;34(3):433–9. friction. Science 1999;284:1650–2.
[92] Hesmat H, Godet M, Berthier Y. On the role and mechanism of dry [126] Hirano M, Shinjo K. Atomistic locking and friction. Phys Rev B 1990;41:
triboparticulate lubrication. Lubric Eng 1995;51:557–64. 11837–51.
[93] Higgs III CF, Tichy JA. Granular flow lubrication: continuum modeling of [127] Hirano M, Shinjo K. Superlubricity and frictional anisotropy. Wear
shear behaviour. ASME J Tribol 2004;126:499–509. 1993;168:121–5.
[94] Higgs III CF, Tichy JA. Effect of particle and surface properties on granular [128] Harrison JA, Brenner DW. Simulated tribochemistry: an atomic scale view of
lubrication flow. I Mech E 2008;222(703–713). the wear of diamond. J Am Chem Soc 1994;116:10399–402.
844 M. Renouf et al. / Tribology International 44 (2011) 834–844

[129] Robbins MO, Müser MH. Computer simulations of friction, lubrication, and [138] Sfantos GK, Aliabadi MH. Multi-scale boundary element modelling of
wear. In: Modern tribology handbook. CRC Press LLC; 2001. p. 717–65. material degradation and fracture. Comput Methods Appl Mech Eng
[130] Sham TL, Tichy JA. A scheme for hybrid molecular dynamics/finite element 2007;196:1310–29.
analysis of thin film lubrication. Wear 1997;207:100–6. [139] Maugin GA. The thermomechanics of nonlinear irreversible behaviors.
[131] Massi F, Rocchi J, Culla A, Berthier Y. Coupling system dynamics and contact World scientific series on nonlinear science, series A, vol. 27. World
behaviour: modelling bearings subjected to environmental induced vibrations Scientific; 1999.
and ‘false brinelling’ degradation. Mech Syst Signal Process 2010;24(4):1068–80. [140] Renouf M, Fillot N. Coupling electrical and mechanical effects in discrete
[132] Bucher F, Dmitriev AI, Ertz M, Knothe K, Popov VL, Psakhie SG, Shilko EV. element simulations. Int J Numer Methods Eng 2008;74:238–54.
Multiscale simulation of dry friction in wheel/rail contact. Wear [141] Richard D, Iordanoff I, Berthier Y, Renouf M, Fillot N. Friction coefficient
2006;261:874–84. as a macroscopic view of local dissipation. ASME J Tribol 2007;129(4):
[133] Temizer I, Wriggers P. A multiscale contact homogenization technique for 031404.
the modeling of third bodies in the contact interface. Comput Methods Appl [142] Richard D, Iordanoff I, Renouf M, Berthier Y. Thermal study of the dry sliding
Mech Eng 2008;198:377–96. contact with third-body presence. ASME J Tribol 2008;130(3):031404.
[134] Berthier Y, Cao H-P, Nhu V-H, Renouf M, Saulot A. Couplage et dialogue [143] Renouf M, Berthier Y. Modélisation numérique discréte des matériaux
entre méthode discre te et continue pour la tribologie numérique. In: Actes granulaires, Modélisation numérique de la génération et transmission de
du 9e Colloque National en Calcul des Structures, vol. 1; 2009. p. 155–60. la chaleur, Lavoisier; 2010. p. 381–400.
[135] Saulot A, Renouf M, Berthier Y. Fem-dem dialogue for tribological under- [144] Blok HA. Theoretical study of temperature rise at surface at actual contact
standing. In: Proceedings of the ASME/STLE international joint tribology under oilness lubricating conditions. I Mech E 1937;2:222–35.
conference; 2008. [145] Champelovier T, Renouf M. Mode les multiphysiques par éléments discrets -
[136] Anciaux G, Molinari JF. Contact mechanics at the nanoscale, a 3d multiscale application a la modélisation du troisie me corps. In: Actes du 9e Colloque
approach. Int J Numer Meth Eng 2009;79:1041–67. National en Calcul des Structures, vol. 2; 2009. p. 665–70.
[137] Luan BQ, Hyun S, Molinari JF, Bernstein N, Robbins MO. Multiscale modeling [146] Renouf M, Cao H-P, Nhu V-H. Multiphysical modeling of third-body
of two-dimensional contacts. Phys Rev E 2006;74:046710. rheology. Tribol Int 2011;44(4):417–25.