Professional Documents
Culture Documents
Elastic Bodies in Rolling Contact Three-Dimensional
Elastic Bodies in Rolling Contact Three-Dimensional
by
J.J.KALKER
Department of Mathematics and Informatics.
TH Deljt. The Netherlands
Ka 1ker. J. J.
Three-dlmenslonal elastlc bodles ln roll1ng contact I by J.J.
Kalker.
p. cm. -- (Sol id mechanics and its appl icatlons ; v. 2)
Includes bibliographical references (p. ) and index.
Preface XIII
Introduction XV
Notation XXI
VII
1. 7.3 F rictional boundary condi tions f or bodies of
revolution with the axes almost in one plane 41
1. 7.3.1 Concentrated, e.g. Hertzian, geometry 44
1.7.3.2 A ball rolling in a conforming groove 45
CHAPTER 2 REVIEW 47
VIII
CHAPTER 3 THE SIMPLIFIED THEORY OF CONTACT 99
4.1 The principle of virtual work and its dual for contact problems 138
4.1.I Virtual work 138
4.1.2 Complementary virtual work 144
4.2 Application to elasticity 148
4.2.1 Minimality of the potential energy, maximality of the
complementary energy, and uniqueness of the solution 150
4.2.2 The case Sg 4< 0 154
4.2.3 Existence-uniqueness theory 156
4.2.4 Surface mechanical principles 157
4.2.5 Complementary energy or potential energy in numerical
work? 158
4.3 Implementation 159
4.3.1 The basic algorithm 160
4.3.2 Discretisation of the contact problem 168
4.3.3 The algorithm of 4.3.1 applied to half -space contact
problems 172
4.3.4 Steady state rolling, elastic and viscoelastic 181
4.3.5 Prescription of total force components 181
IX
4.3.6 Sensitivities 182
4.3.7 Calculation of the infIuence numbers in a half -space 183
4.3.8 The subsurface elastic field in a half -space 184
4.3.9 Note on the generalisation to non-concentrated contacts 184
x
Appendix A The basic equations of the linear theory of elasticity 239
Appendix B Some notions of mathematical programming 245
Appendix C Numerical caiculation of the elastic field in a half -space 255
Appendix D Three-dimensional viscoelastic bodies in steady state
frictional rolling contact with generalisation to contact
perturbations 265
Appendix E Tables 285
Bibliography 295
Index 307
XI
PREFACE
This book is intended for mechanicians, engineering mathematicians, and, generally for
theoretically inclined mechanical engineers. It has its origin in my Master's Thesis (J 957),
which I wrote under the supervision of Professor Dr. R. Timman of the Delft TH and
Dr.Ir. A.D. de Pater of Netherlands Railways. I did not think that the surface of the
problem had even been scratched, so I joined de Pater, who had by then become Professor
in the Engineering Mechanics Lab. of the Delft TH, to write my Ph.D. Thesis on it. This
thesis (1967) was weil received in railway circles, which is due more to de Pater's
untiring promotion than to its merits. Still not satisfied, I feit that I needed more mathe-
matics, and I joined Professor Timman's group as an Associate Professor. This led to the
present work.
Many thanks are due to G.M.L. Gladwell, who thoroughly polished style and contents of
the manuscript. Thanks are also due to my wife, herself an engineering mathematician,
who read the manuscript through critically, and made many helpful comments, to G.F.M.
Braat, who also read an criticised, and, in addition, drew the figures together with
J. Schonewille, to Ms. A.V.M. de Wit, Ms. M. den Boef, and Ms. P.c. Wilting, who typed
the manuscript, and to the Publishers, who waited patiently.
XIII
INTRODUCTION
xv
Although there is no difference in principle between concentrated and diversified
contacts, there is a difference in practice. The effect of a concentrated load on a three-
dimensional half-space is known analytically (Boussinesq, 1885; Cerruti, 1882). This
opens the door to three-dimensional contact calculations of concentrated contact problems,
and some diversified and semi-concentrated contact problems. For most diversified
problems the half -space approximation is neither feasible nor desired, and a finite
element method is required. Whereas such a calculation is well-established practice for
two-dimensional problems, it is still in its infancy for three-dimensional FEM analysis.
This justifies the practical distinction between concentrated and diversified contacts.
Consequently, when, in Ch. 4, we consider the general, variational theory of contact, and
the theoretical side of the algorithms, our analysis is valid for three-dimensional diver-
sified contacts with Coulomb friction. But all our results concern half -spaces, i.e. concen-
trated and some semi-concentrated problems.
Galin published two books (1953, 1980), on elastic and viscoelastic contact prob-
lems. The 1953 book is based mostly on Muskhelishvili's works (1946, 1949). This
theory is two-dimensional. The 1980 book contains a large chapter on three-
dimensional, frictionless contact problems.
Fichera (1964) and Duvaut and Lions (1972) established the variational theory of
contact, in which they concentrate on existence and uniqueness proofs of the solu-
tions. Duvaut and Lions consider also viscoelastic and dynamic contact problems.
Gladwell (1980) published a treatise on contact mechanics. For the most part he
considers only frictionless or bonded contacts, concentrated and diversified.
Kikuchi and Oden (1988) wrote a book on elastic, frictionless and frictional
contact mechanics in which they concentrate on the theoretical background of finite
element methods and variational theory.
XVI
In the present work we confine our attention to three-dimensional elastic contact
problems with Coulomb friction. The main theme is rolling contact, with a spin-off to
other types of contact problems such as frictionless contact and the frictional shift. Choice
and treatment of the subject matter are such that they are complementary to the above-
mentioned books. Many results described have been published before as articles, but there
is also much new, as yet unpublished, material. We survey the contents of the book.
In Ch. I the problem is stated and modeled. The boundary conditions are derived. The
chapter is introducory, and required reading for the succeeding chapters which can be
read independently.
In Ch. 2 the elastic rolling contact theories that have been set up are reviewed exten-
sively. We discuss the accuracy and the calculating speed of the finite element methods
(FEM) and the boundary element methods (BEM) and explain why the half -space
approximation, where applicable, leads to faster computer programs. The chapter is
intended for the rail vehicle dynamicist whom it will aid in choosing the roIling contact
theory appropriate to his work.
Duvaut and Lions, who founded the variational theory of frictional contact in 1972, gave
no physical explanation of the variational inequality upon which they based their
existence and uniqueness proofs. Such a basis is given in Ch. 4, which also contains a
statement and proof of the active set algorithm that numericaIly solves the contact
problem on the basis of the variational theory. This chapter is primarily intended for
contact mechanicians.
In Ch. 5 numerical results and applications are given. It contains many interesting and
chaIlenging test cases; care has been taken to specify the problems completely. This
chapter is intended for researchers who want to test their own algorithms.
XVII
Appendices A and B contain some elementary not ions on the theory of elasticity and on
mathematical programming, respectively.
In Appendix C an algorithm is given to calculate the elastic field in the interior and on
the boundary of a half -space wh ich is loaded by a uniform load on a rectangle Iying in
the bounding plane.
Appendix D contains the extension of the three-dimensional elastic half -space theory of
steady state rolling to the viscoelastic case. Further , it contains an extension of the theory
of sensitivities (described in Ch. 4, Sec. 4.3.8), to perturbations that are periodic in time,
both in the elastic and in the viscoelastic case. Finally an explanation is given for the fact
that the calculation of the sensitivities, notably the creepage and spin coefficients of
rolling contact, see Kalker (l967a), Ch. 4, are calculated so inaccurately by the program
CaNT ACT that is based on the theory of Ch. 4.
Appendix E contains tables for the Hertz theory and related problems, and for the linear
theory of rolling contact.
My own results that are described in this book were obtained with two programs, viz.
ROLLEN (Hertzian rolling contact: simplified theory) and CaNT ACT (complete elastic
half -space contact theory).
- ROLLEN implements the Hertz theory, the linear theory of Hertzian rolling
contact, and the simplified theory of Hertzian rolling contact. In addition, it
implements Hertzian deep groove rolling.
- CONTACT computes:
Contact formation;
Shift problems, such as Cattaneo's problem, both single step and transient;
Rolling contact problems, both steady state and transient;
The elastic fields inside a half -space.
All these problems can be solved for Hertzian and non-Hertzian contact, and for
equal and different materials of the contacting bodies. All calculations concern
three-dimensional, homogeneous, isotropic half -spaces in concentrated contact.
XVIII
The programs are available as a FORTRAN source code on a floppy disk, ready for use
on a 640 kB pe AT. Prices are available upon request. For information contact:
FAX (31)(15)787022.
XIX
NOTATION
NI. GENERAL
A roman, non- bold faced capital letter signifies a point set or an index set.
A roman bold faced letter signifies a vector or a matrix, except B, C, D, E, K, see below
in S2.
An italic letter indicates a scalar variable.
All symbols may carry an index:
Index a: body number, I or 2.
Indices i, j, h, k run from I to 3, or over x, y, z; they indicate Cartesian components.
Indices 0:, ß, 1, r run from I to 2; they also indicate components.
r has the connotation: "tangential component".
Indices I, J indicate element numbers.
All these indices, except a, are subject to the summation convention of summing over the
entire range of repeated indices in a product, except where otherwise indicated.
ft
All other indices serve to complete the identifier.
(0) = ,material derivative.
( ') may complete the identifier; in addition, it may indicate that the variable is taken at
the time t ' .
. = 8/8x.; analogously for j, h, k.
,I 1
"sub" means "subject to the auxiliary condition(s)".
This list shows variables whose meanings extend beyond the section where they are
defined. The construction of an entry is as follows:
Symboles) Definition, comment Reference
XXI
In the reference, numbers between brackets refer to equations, and without brackets to
sections.
XXII
E Young's modulus App.A
F.
I
i = 1,2,3 or = x,y,z total force transmitted by contact 1.4,( 1.64)
FEM finite element method
XXIll
N index set of inactive constraints ( 4.3Ia)
N(x.) index set of constraints not active in x. ( 4.28g)
I I
n, nj outer normal on V; can carry subscript a (4.3c)
XXIV
displacement of body a, at time t I (1.7),(1.8)
prescribed displacement in A (1.39),(4.4)
ua
x plane X = Xl = 0
x, xl coordinate in rolling direction between (1.23) and (1.24)
x. Cartesian coordinates, origin 0 1.2
1
X. Cartesian coordinates of body a 1.7.1
a/
x a ' x ai particle coordinates in undeformed state at time t 1.2
x'. as X ., at time t I 1.2
a/ a/
y plane Y = x2 = 0
y = x2 coordinate in lateral direction
z 1. plane z = x 3 = 0
2. feasible set (4.3)
coordinate pointing normally into body 1
xxv
'1 lateral creepage (1.82)
v combined Poisson's ratio (1.44)
v Poisson's ratio of body a
a
v. Lagrange multiplier (4.29)
I
v[ Lagrange multiplier of non-penetration conditions 4.3.3,Point 4
e longitudinal creepage (1.82)
p density
r7 .• stress App.A
I)
r7[ ideal stress (5.19)
r/> spin ( 1.82)
w angle between planes of principal curvature 1.7.1
a
XXVI