Anda di halaman 1dari 8

Tribology

International

Vol. 29, No. 8, pp. 651-658, 1996 CopyrIght 0 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0301-679X/96/$15.00 +O.OO

ELSEVIER SCIENCE:

SO301-679X(96)00011-4

ite element methods contact problems with


P. Wriggers

for friction

The numerical treatment of contact problems involves the formulation of the geometry, the statement of interface laws, the variational formulation and the development of algorithms. In this paper an overview with regard to the numerical simulation of frictional problems is presented when general constitutive equations are formulated in the contact interface. To be most general we will apply a geometrical model and its discretization for contact which is valid for large deformations. Furthermore the algorithms to integrate the interface laws will be discussed for the tangential stress components. Special emphasis is laid on the deveiopment of algorithms which allow an efficient treatment of frictional contact problems. Copyright 0 1996 Elsevier Science Ltd
Keywords:
dry friction, large deformations, finite element method

Introduction Boundary value problems involving contact are of great importance in industrial applications in mechanical and civil engineering. The range of application includes metal forming processes, drilling problems, bearings and crash analysis of cars or car tires. Other applications are related to biomechanics where human joints, impla-Its or teeth are of consideration. Due to this variety contact problems are today often combined either with large elastic or inelastic deformations including time dependent responses. In this article we will restrict ourselves mainly to finite element techniques for the treatment of contact problems despite many other numerical schemes and analytical approaches that could be discussed as well. Furthermore we like to note that the description of the mechanical behaviour of the bodies coming into contact will not be investigated in detail, although this is of great importance. This article thus concentrates on the behaviour in the contact interface for which a contir uum formulation, discretization within the finite element method and the associated algorithms are discussed.

Due to the precision which is needed to resolve the mechanical behaviour in the contact interface, different approaches have been used in the literature to model the mechanical behaviour in contact area. Two main lines can be followed to impose contact conditions in normal direction: these are the non-penetration condition as geometrical constraints and constitutive laws for the micromechanical approach within the contact area. Here we restrict our discussionsto the micromechanical view. Constitutive equations for the normal contact can be developed by investigating the micromechanical behaviour within the contact surface, seee.g. Kragelsky
et al..

lnstitut fiir Darmsladt,

Mechanik, Germany

TH

Darmstadt,

Hochschulstr.

1, D-64289

The interfacial behaviour in the tangential direction (frictional response) is even more complicated. The most frequently used constitutive equation is the classical law of Coulomb. However, other frictional laws are available which take into account local, micromechanical phenomena within the contact interface, see e.g. Woo and Thomas2. An extensive overview may be found in Oden and Martins3; for the physical background see e.g. Tabor4. During the few last years frictional phenomena have also been considered within the framework of the theory of plasticity. This leads to non-associative slip rules. Different relations have been proposed for frictional problems by e.g. Michalowski and Mroz or Curnier6.
International Volume 29 Number 8 1996 651

Tribology

Finite element

methods

for contact

problems:

P. Wriggers

The application of constitutive equations for friction within finite element calculation can be found in e.g. Fredriksson and Wriggers et al.*. The major advantage of this formulation is related to the so called return mapping algorithms, known from plasticity9 and its applications to frictionlO. The weak formulation of contact problems leads to variational inequalities. Different possibilities exist for the numerical solution of these problems. The most frequently used algorithms are based on active set strategies which are applied in combination with Lagrangian multiplier or penalty techniques. Each of the methods has its own advantages and disadvantages. Due to its simplicity the penalty method is favoured by most finite element software developers. A combination of the penalty and the Lagrangian multiplier techniques leads to the so called augmented Lagrangian methods which try to combine the merits of both approaches. These techniques are needed when constitutive interface laws are employed, since often ill-conditioning of the problem occurs, see Wriggers and Zavarise. For the general case of contact including large deformations, arbitrary sliding of a node over the entire contact area has to be allowed. This behaviour is captured by a discretization often called the node-tosegment approach. The associated matrix formulation for large deformations can be found in Wriggers and Simoi for two-dimensional problems. The formulation for frictional contact is discussed in Wriggers et aZ.s. A continuum based approach to contact problems has been derived in Laursen and Sirno. Fig 1 Contact geometry and geometrical approach are ( ),,

tangent vectors of the contact surface a, : = a:,([, t2) and Ai : = XTa(<l, 5) where denotes differentiation with respect to .$a.

As the first relevant function for the contact geometry we define a penetration function on the current slave surface qi(T:.) by setting:
gN+

= 1$2)11for [x1 - kF(r,, otherwise of the distance (2) items of r?)] . n < 0 (I)

I/x1 - a$(?, 0 i

Here s = (?I, I) is the minimizer function for a given slave point xi: &(& .$) = 11x1- a:([, function c)l]+MIN (1) contains The penetration information: (l) c2)

Contact

geometry

This section summarizes relations which are necessary to formulate the geometrical contact conditions. In detail the penetration and the relative slip in the contact area are discussed. The first condition also includes the non-penetration condition which is used classically in contact mechanics. The derivation presented here can be used for frictional or frictionless problems. It follows closely the approach discussed in Wriggers and Mieher4. We assume that two bodies which undergo large deformations can come into contact. Let 987, y = 1, 2, denote the two bodies on interest and & maps points Xy E 37 of the reference configuration onto points xy = cp:(X) of the current configuration. Motivated by micromechanical investigations of contact problems we view the mechanical approach of the two contact surfaces as a microscopical penetration of the current mathematical boundaries @(ICY). In this formulation IY,Y C cBJ~ are possible contact surfaces of the bodies By; see Fig 1 for an illustration of this concept. In what follows we denote VP:(Ii) as the current slave surface which penetrates in the case of contact into the current master surface &(I:) (moving reference surface). We parametrize the master surface If in its reference and current configuration by the natural parameters p, p; i.e. we consider material curves X2 = X2 (<I, p) and x7 = %p(t, p). Then the 652 Tribology International Volume 29 Number

two

gN+ serves as a local contact check, i.e. we set: contact g, > 0 enters for gN+ > 0 as a local kinematical gN+ variable the constitutive function for the contact pressure. @N+ which

For later use we compute the variation yields . n2 &if N+= [ q 1 - C(3]

(3) with r) being the virtual displacement or test function. We note in passing that gk+ = [x - %i:(z)] . n 2 0 represents the classical non-penetration condition for finite deformation.

The tangential relative slip between two bodies is related to the change of the solution point (?I, 2) of the minimal distance problem. Thus we can compute the time derivative of 5 at the minimum point in (2). This yields the following result: H,,(P with: = R, (4)

H ap= k+ + gN+%ql R, = [vj - +;($)I . az,+ gN+ii . e:,(J)


8 1996

Finite element methods for contact problems: and 5 are the first and second fundamental f::rn of tl$ deformed surface, well known from differential geometry. With this we can define the tangential relative velocity function on the current slave surface pi (ry ): a

P. Wriggers

(5) Equation (5) determines per definition the evolution of the tangential slip g, which enters as a local kinematical variable the constitutive function for the contact tangential stress, see next section. By exchanging th- velocities with the variations of the tangential slip it can be stated as: 6gT = &$a<I Constitutive equations for contact (6) interfaces

.Yv g, := ta2,

to friction. Here we use a quite general frictional law which incorporates elastic micro-mechanical displacements. The key idea of this model, which is related to an elasto-plastic formulation, is a split of the tangential slip g, into an elastic (stick) part g$ and a plastic (slip) part g+, see Equation (10). The elastic part describes the micro displacement which can be regarded as stick behaviour since the associated deformations vanish once the loading is removed from the system. The constitutive behaviour for the tangential elastic micro-displacements can be deduced from experiments and is related to the elastic deformation of the asperities due to tangential loading. Here we assume, as the simplest possible model, an isotropic linear elastic constitutive equation for the tangential contact stress:
tT = cT i&t%

with

gF:=g,-g+

(10)

The normal contact stressescan be obtained generally in two different ways. On one hand the contact stresses follow from the non-penetration constraint. On the other land an approach of both bodies is observed in the ccntact area which then leads to the formulation of associated constitutive interface equations. Normal stress in the contact area

where c-r is a material parameter. The tangential plastic slip g+ is governed by a constitutive evolution equation which can be derived by using standard concepts of the theory of elastoplasticity. Within this framework we can formulate a plastic slip criterion function fS(fTrpNrO,gv) 5 0 for a given contact pressue pN with material parameter p, a given temperature and a hardening/softening variable like the effective slip g,. This slip criterion function can be specialized as follows:
fs(tT$N~osgv) = iid ~s(PN>&gv) so (11)

In the first case the mathematical conditions for the non-pznetration gh+ - 0 precludes the penetration of > one tody into another. Contact takes place when gfir, i$ equal to zero. In this case the associated normal component pN of the stres vector t = pNn + tPap in the ccntact interface must be non-zero leading to: (7) which is well known as the Kuhn-Tucker condition for frictionless contact problems. These conditions provide the basis to treat contact problem in the context of constraint optimization. When the micromechanical behaviour of the contact area is considered. the contact pressure is related to the approach of the physical surfaces which come into contact, i.e. the penetration of the mathematical surfaces results from the deformation of the microasperities which can be elastic or also plastic. In this case we assume the following general form of the constitutive law: prq =f,(d) or d = ~NC.PN) (8) gl;+ 2 0, Pl.4~0, PNgkt =o

with the special case of classical Coulombs model


f;(tT,pN) = IltTli PpN 5 O.

The constitutive evolution equation for the plastic or frictional slip can be stated in a general form of a slip rule for large deformations in the contact zone as follows: Zv g+ = A$
T

= An,

with nT = ~

l;t:II
(12)

& =A

where Equation (12) describes the evolution of the effective slip which is defined as:

where fN and hN are nonlinear functions of the current mean plane distance d or the contact pressure pN, respectively. For a specification of this function, see e.g. Song and Yovanovich5. The current mean plane distance is related to the geometrical approach gN+ (1) as follows: gr.1, = i- d or d=[-g,, (9) where JJ is the initial mean plane distance in the contact area r,. Tangential contact frictional slip stress and tangential

A is a parameter which describes the magnitude of the plastic slip. Equations (lo), (11) and (12), along with the loading-unloading conditions in Kuhn-Tucker form: AZO, fssOo, hfs=O (13) establish the constitutive framework for the tangential slip-stick behaviour. The algorithmic treatment will be discussed later. Boundary strategies value problem, global solution

Many different constitutive models have been develalped to formulate the interfacial behaviour due Tribology

For a numerical solution of the nonlinear boundary value problem we will use the finite element method. Thus we need the weak form of the local balance equations. Due to the fact that the constraint condition (7) is represented by an inequality we obtain in International Volume 29 Number 8 1996 653

Finite

element

methods

for contact

problems:

P. Wriggers work. We just state the matrix weak form of (14): G(v) = i y=l BT7dVformulation of the

general a variational inequality. Algorithms for solving variational inequalities are: mathematical programor sequential quadratic ming, active set strategies programming methods. Here we will apply the active set strategy which is implemented in many existing finite element codes. Within an active set strategy we can write the weak form as an equality since we know the active set of the contact constraints within an incremental solution step:

NT fy dV

- jr:,TiydA}

(17)

where the matrix N contains the shape functions and the so-called B-matrix contains the derivatives of the shape functions. Any standard finite element book will provide the necessary details, see e.g. Zienkiewicz and Tay1ors. A genera1 discretization of the contact interface which allows for large tangential sliding is given by the setup depicted in Fig 2. This discretization is called nodeto-segment contact element and is widely used in nonlinear finite element simulations of contact problems. Due to its importance we will consider this contact element in more detail. Assume that the discrete slave point (s) comes into contact with the master segment (l)-(2), then the kinematical relations can be directly computed using the equations given earlier. With the interpolation for the mater segment, a(t) = x: + (x s - xj)[, the tangent vector of the segment follows as: a: = a([),, = (x2 - x:). It is connected to an orthonormal base vector af = if/l with I = IIxf - x:/l being th e current length of the master segment. Now the unit normal to the segment (l)-(2) can be defined as n7 = e3 x a:. 2 and gN are given by the solution of the minimal distance problem (2), i.e. by the projection of the slave node x,~ in (s) onto the master segment (l)-(2)

@N&N

+fT.

%TjdA

=O

(14)

Here Py is the first Piola-Kirchhoff stress tensor and FY denotes the deformation gradient of the body City. VY is the virtual displacement which is zero at the boundary I; where the deformations are prescribed. In Equation (14) the constitutive equation (8) and (10) have to applied for the determination of pN and tT. The variation Of the normal gap function gN+ and the tangential slip gT is given by (3) and (6). A major problem associated with the numerical treatment of the penalty method and the contact interface laws is the ill-conditioning which arises when the stiffnesses due to law (8) are combined with stiffnesses of the bodies within the finite element formulation. A method to overcome this problem is the augmented Lagrangian technique, well known in optimization theory. This leads to the following form for the contact contribution in Equation ( 14)r1:

subject to

From these equations and the local formulation (3) we compute directly the variation of the gap function 8gSN+ on the straight master segment (l)-(2):
[i@N)] = o

c+(~y,PN) = gN+

The fulfillment of the nonlinear interface law (8) will be practically accounted for by an update formula for the Lagrangian multiplier pN

with the known state.

quantities

{ ...}o,d from the previous

Augmented Lagrangian techniques have also been applied to frictional problemsr6~r7. The latter paper also investigates the possibility of an algorithmic symmetrization of the frictional part of the tangent matrix.

Discretization area

techniques

within

the contact
of the of this

The discretization of the domain contributions bodies being in contact in (14) is not objective 654 Tribology International Volume

Fig 2 Node-to-segment 8 1996

element

29 Number

Finite element methods for contact problems: Equation (4) yields the expression for 62. With the interpolation for the variation, +j2(Q = v: + e(q$ - $), on the strakht master segment (l)-(2) we specialize H,, and RI in (4) which leads to: Sg, = IS?= [7ji - (1-j)gf + yyq; - q:] . n2 - f+] . a: (20)

P. Wriggers

The tangent matrix for the normal contact is derived from the t_erm 6gN 5 P, ~ in (21). Observe that the change in 5 has been considered as well as the change of the normal n2. We obtain with (8) the tangent matrix:

Equations (18) to (20) characterize the main kinematical relations of the contact element. In what follows, we compute the contribution of the node-to-segment element to the weak form (14). We assume that we know the normal force P N 5 = pN .A b and the tangential force TT 5 = fT s As at the discrete contact point (s) of the contact element under consideration, where A, denotes the area of the contact element. Both forces, PN s and TT d, can be obtairted from the constitutive relations discussed earlie!:. This leads to:

(26) Note that in a geometrically linear case all terms vanish which are multiplied by g,,. This results in the IllatriX Kkc, = (df,.J/d&,Tr) NT. N, For the tangential contributions in the contact area we have to linearize the term 6gT s TT s in (21) which leads for the elastic response (stick condition) to:

K+.=cT +gy

gN T, +--NosF 1

gN T, +--No, I I -T,.T,T

iI.,
.c
s-1

i (PN h +
(PNJ &NT

fT

%r)dr

N,,N:

+N,N;,

-T,T;f, (27)

TT~

%Ts)

(21)

- Zgy(N,.T6.,

+ T,,,N;,)

PNs follows directly from (8) multiplied by the area A, of the contact element. For the tangential force T.r s we have to perform an algorithmic update which
is described in the next section. The contributions of one contact element in (21) can now Ike cast into a matrix formulation. For the normal part (19) and the tangential part (20) we have: Q:,.s = v?N.s with the definitions:

Also in this case all terms containing gN .s disappear in a geometrically linear situation which yields K+ : = c,T,~TT. The case of frictional slip leads to an additional contribution in (27) which will be discussed in the next section.

Ii

Algorithms

for contact

problems

6gTs = qT

T, +g*N,,, I t

1
(22)

The algorithm which is applied in many standard finite element programs is related to the active set strategy in which the problem is solved for a chosen active set of constraints. Before we state this algorithm we combine (17) and (25) to the global set of equations: G?(v) = G(v) + i G,(v) = 0 (28)

3-, = (n 2, -(l-j)n, 3 os = (0, -n2, n)P


and

-i$n2)T,
(23)

s=l

where G(v) denotes the contributions of the bodies due to the weak form (17). Now the algorithm is summarized in Box 1.

TH, = (af, -(l-j)af, P os = (0, -al, aT)T

-&I:)~T,
(24) Initialize algorithm set: v, = 0 LOOP over iterations: i = 1, .. .. convergence Check for contact: g, Si 5 0 7 active node Solve: GJVi) = G(v;) + Uzs, Gz(vi) = 0 Check for convergence: /GJvi)jl 5 TOL + END LOOP END LOOP

Thus the virtual mechanical work (21) of the contact eleme:lt can be written as qTGS with the contact element residual vector:

GF=P,.N,

gNs + TTs T, +pNos 1 t

(25)

Often a Newton-Raphson iteration is used to solve the global set of equations. Then the linearization of (25) is needed to achieve quadratic convergence near the solution point. The associated derivation is a little cumbersome and thus only the final results will be summarized. Details of the frictionless case can be found in Wriggers and Sirno and for contact including friction in Wriggers et a1.8. Tribology

Box I method

Contact

algorithm

using

the

penalty

When ill-conditioning occurs the augmented Lagrangian technique has to be applied see e.g. Wriggers and Zavarisell. International Volume 29 Number 8 1996

655

Finite element

methods

for contact

problems:

P. Wriggers

Within the global algorithm the algorithmic update of the tangential stress tT n+l due to friction is performed by a return mapping algorithm based on an objective (backward Euler) integration of the evolution equation (12) for the plastic slip, see e.g. Wriggers or Giannokopoulos19. The results can be summarized as follows: integration of (5) gives the increment of the total slip within the time step Atntl

Punch Problem Plane Strain Load Indentation Curve

Ag T,z+l

(?:+I

khz+l

(29)
, , I , 4

The total slip has to be decomposed into an elastic and a plastic part, see Equation (10). Then we can compute the elastic trial state from (10) and evaluate the slip criterion (11) at time trZ+,:
ttr

t n+l := CT(gTn-1.1 - gk,,)


-FPN

= h-n + CT A&n+, (30) friction relation perform scheme,


0.4 0.8 1.2 1.6

Ef,l,z-f1 := Il%-l~~

ntl

If this state is elastic Cftsrll+i 5 0) then no takes place and we have to use the elastic (10). In the case that firn+, > 0 we have to the return mapping. Using the implicit Euler (12) yields: g+n+1 = !&-II + An Tnil
g, nil = g II +A

(31)

Maximal

Indentation displacement

With the standard arguments regarding the projection schemes, see e.g. Simo and Taylor9, we obtain: t Tn+l
nTFz+l ttr t w-i T-l+1 ACT n,-,+,

Fig 4 Total punch force versus maximal

=n%

(32)

The multiplication of (32) by nT n+l yields the condition from which h can be computed:

where & can be a nonlinear function of h. Thus in general we need an iterative scheme to solve K(A). In case of Coulombs model we can solve directly for h:

A= $ Wn+l II - PPNn+l)
Knowing A the stress update follows from (32) and the frictional slip from (31). Again we state the explicit results for Coulombs model:

Fig 3 Purzch problem, 656 Tribology

jkite

element mesh Volume 29 Number

Fig 5 Deformed 8 1996

meshes for cases (I),

(2) and (3)

International

Finite element

methods

for contact

problems:

P. Wriggers

K;.c,

= K

fp-- dfN -r.> %N s

T,++$,

NT

(36)

whick completes interface law.

the

algorithm

for

the

frictional

The tangent matrix which is needed within a Newton iteration can be derived by linearizing the term which- appears in the weak form with respect to the dispkcement field. The explicit matrix form results from the term 6gr .s,l+ 1 T, arz-ml and can be stated for Coulombs model as follows:

with K$- J from (27). This matrix is unsymmetric which corresponds to the non-associativity of Coulombs frictional law. Note that the non-symmetry of the tangent matrix has to be considered to obtain a robust algorithm.

Numerical

example

To demonstrate the applicability of the derived numerical method we consider a punch problem in which a

STRESS 5 Min = O.OOE+OO Max = 1.90E+OO

9.90E-01 l.O9E+OO l.l9E+OO .._ . 1.3OE+OO 1.40E+OO 150E+OO Current View Min = O.OOE+OO X = 4,99E+OO Y = 3.00E+OO Max = 1.90E+OO x = 6.86E+OO Y =-9.81 E-01 STRESS 5 Min = O.OOE+OO Max = 1.86E+OO

I 9.90E-01

Current View Min = O.OOE+OO X = 5.00E+OO Y = 3.00E+OO Max = 1.86E+OO x = 6.62E+OO Y =-7.36E-01
Fig 6 Plastic zoue for cases (I) and (2)

Tribology

International

Volume 29 Number 8 1996

657

Finite element

methods

for contact

problems:

P. Wriggers providing the reader with the underlying theoretical derivations.

rigid punch, as depicted in Fig 3, is pressed against a body made of steel. Large deformations and elastoplastic behaviour are considered. Three different cases are examined: (1) frictionless contact, (2) contact with Coulomb friction (p = 0.2), and (3) frictionless contact when the surface of the body is coated with a material of much higher yield strength. The finite element mesh and the boundary conditions are shown in Fig 3. The material data for the body are: bulk modulus K = 165, shear modulus G = 80, yield strength Y0 = 0.45 and linear isotropic hardening modulus H = 0.15. Within the calculations the displacement of the punch is prescribed. The maximal displacement of 2 is reached within 40 load steps. Within each load step the above-described Newton algorithm needs between six and 11 iteration steps for convergence. This includes the nonlinearities due to elasto-plasticity, large deformations and friction contact. The difference of the total number of iterations per step stems from the fact that Newtons method converges only quadratically near the solution point. Thus, if the number of nodes being in contact changes during a load step then the additional iterations are needed to establish the new contact area. Figure 4 shows the total force versus the maximal punch displacement. We observe that cases(1) and (2) depict almost the same curve. Only in the range of large indentations do the curves deviate and case(2) presents a more stiff behaviour. This can be explained by the fact that the frictional forces only contribute to the punch force in case of a large indentation, as depicted in Fig 5 (2). For case (3) we obtain immediately a different solution since the coating distributes the loading due to the punch over the body which results in a considerable stiffening. The deformed meshes for a punch displacement of 2 show clearly in Fig 5 the different behaviour of the three cases in the contact area. Due to the plastic incompressibility the material is squeezed out from under the punch in the frictionless case (1). This is prevented by the frictional forces in case (2). In Fig 6 the plastic zone is depicted for cases (1) and (2) when the same punch displacement is applied. Here we observe a different elasto-plastic behaviour of the system under the punch.

References
I. Kragelsky I.V., Dobychin M.N. and Kombalov VS. Friction and Wear ~ Calculation Methods, (Translated from the Russian by N. Standen), Pergamon Press, 1982 2. Woo K.L. and Thomas of experimental works. T.R. Contact Wear 1980, of rough surfaces: 58, 331-340 a review

3. Oden J.T. and Martins J.A.C. Models methods for dynamic friction phenomena. Mech. Engng. 1986, 52, 527-634 4. Tabor D. Friction J. Lubr. Technol. 5. Michalowski sliding rules 30, 259-276 6. Curnier 637-647 A.A. - The present state 1981. 103, 169-179 Associated problems.

and computational Comp. Meth. Appl. of our understanding.

R. and Mroz Z. in contact friction Theory of friction.

and non-associated Arch. Mech. 1978, Struck 1984, 20,

Inf. J. Solids

7. Fredriksson B. Finite element in structural mechanics with fracture mechanics problems. 8. Wriggers of large Comput.

solution of surface nonlinearities special emphasis to contact and Comput. Struct. 1979, 6, 281-290

P., Vu Van T. and Stein E. Finite element formulation deformation impact-contact problems with friction. Struct. 1990, 31, 319-331 operators for Appl. Mech. contact (Eds. J.

9. Simo J.C. and Taylor R.L. Consistent tangent rate-independant elastoplasticity. Comp. Meth. Engng. 1985, 48, 101-118 10. Wriggers problems. Middleton

P. On consistent tangent matrices for frictional In Proceedings of NUMETA 87 Conference and G.N. Punde) Nijhoff, Dorbrecht. 1987

11. Wriggers P. and Zavarise G. On the application of augmented Lagrangian techniques for nonlinear constitutive laws in contact interfaces. Comm. Num. Meth. Engng. 1993, 9. 815-824 12. Wriggers P. and Simo J.C. A note on tangent stiffness for fully nonlinear contact problems. Comm. Appl. Num. Meth. 1985, 1, 199-203 13. Laursen T.A. and Simo J.C. A continuum-based formulation for the implicit solution of multibody, mation frictional contact problems. Znt. J. Num. 1993, 36, 3451-3485 14. Wriggers P. and Miehe C. Contact constraints thermomechanical analysis - a finite element Meth. Appl. Mech. Engng. 1994, 113, 301-319 finite element large deforMeth. Engng. within model. coupled Comp.

15. Song S. and Yovanovich M.M. Explicit relative contact pressure expression: Dependence upon surface roughness parameters and Vickers microhardness coefficients. AIAA Paoer 87-0152. 1987 16. Alart P. and Curnier A. A mixed formulation for frictional contact problems prone to Newton like solution methods. Comp. Meth. Appl. Mech. Engng. 1991, 92, 353-375 17. Laursen T.A. and Coulomb frictional Camp. Meth. Appl. Simo J.C. Algorithmic symmetrization of problems using augmented Lagrangians. Mech. Engng. 1993, 108, 133-146 Element method Comput. Method, for the Struct.

Conclusion
This overview summarized some of the current research work in computational contact mechanics for frictional problems. Due to the broadness of contact formulations and algorithms and the limitation of space not all promising new approaches have been discussed in detail. One numerical example, which included large deformation, has been presented. However this could not stand for all mentioned topics. More examples have been omitted, since this overview was aimed at

18. Zienkiewicz O.C. and Taylor R.L. The Finite 4th edn.. McGraw-Hill, London, 1989 19. Giannokopoulos A.E. integration of friction 1989, 32, 157-168 The return constitutive mapping relations.

658

Tribology

International

Volume 29 Number 8 1996

Anda mungkin juga menyukai