Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

263

A GRAPHICAL APPROA<lI TO SHAKEDOWN IN ROllING roNTACf

K L JOHNSON
Cambridge University Engineering Laboratory,
Trumpington Street,
Cambridge CB2 IPZ, UK

ABSTRACf

In repeated rolling contact, even though plastic deformation may occur


during the first passage of the load, through the action of residual
stresses and strain hardening of material, the steady cyclic state may lie
within the elastic limit. Using the Tresca yield criterion it is shown in
this paper how the maximum stress for shakedown - the shakedown limit - in
line contact may be found by a simple graphical procedure.
Cases of tractive rolling, with both partial and complete slip, are
examined for (a) a perfectly-plastic and (b) a kinematically hardening
material.

INTRODUCTION

In 1957 Fessler and Ollerton published a milestone paper in Contact


Mechanics: Contact Stresses in Toroids under Radial Loads [1]. Using the
frozen stress technique of 3-dimensional photoelasticity they made the most
extensive experimental check of the internal stresses in Hertz contact
either before or since*. At that time, before finite elements, stresses
had been evaluated only at the surface and along the axis of symmetry
(z-axis) beneath the surface. Arising from a concern with rolling contact
fatigue, they obtained a closed form expression for the orthogonal shear
stresses T and T throughout the field and checked their values by
yz zx
experiment.
In rolling contact a material element at a particular depth below
the surface experiences a cycle of stress. It was well known that plastic

* They showed that the Hertz predictions of contact area were good up
to the maximum value of contact size to radius of curvature (0.3)
used in their experiments, which was surprising in view of the
idealisations of the Hertz theory.

T. H. Hyde et al. (eds.), Applied Stress Analysis


© Elsevier Science Publishers Ltd 1990
264

yield is initiated at a point beneath the centre of contact by the maximum


shear stress Tmax' which acts on planes at 45° to the surface and whose
value ~ 0.3 Po' where Po is the maximum Hertz pressure. However there was
a view at the time that the orthogonal shear stress TZX ' which acts
parallel to the surface and changes sign from entry to exit, is more
damaging since its range (~ + O.25po to -0.25 po) exceeds that of Tmax
(range ~ 0 to 0.3 p). For this reason the Lundberg-Plamgren theory of
o
ball bearing life [2] was based on the orthogonal shear stress. Its
significance will become apparent from the contents of this paper.
A further seminal paper was published in 1957 by A.W. Crook:
Simulated gear-tooth contacts: Some experiments on their lubrication and
subsurface deformation [3]. Crook showed that repeated rolling contacts at
loads at which exceed the elastic limit of the material can lead to the
accumulation in the near-surface layers of large plastic strains which are
likely to be associated with the initiation of rolling contact fatigue.
This publication stimulated theoretical work at Cambridge into plastic
deformation and shakedown in repeated rolling and sliding contacts which
has continued, on and off, until the present time.

Shakedown
Shakedown in repeated loading is the process whereby plastic deformation in
the first few cycles of load leads to a steady cyclic state which lies
within the elastic limit. The maximum load for which shakedown occurs is
known as the shakedown limit. Three separate processes can contribute to
shakedown:
(i) residual stresses introduced by initial plastic flow inhibit plastic
deformation in the steady state;
(ii) strain hardening may raise the elastic limit; and
(iii) geometry changes brought about by plastic deformation may increase
the conformity of the contact and reduce the contact stress.
In this paper we shall confine our attention to the·plane
deformation of an elastic-plastic half-space by a long cylinder in 'line
contact'. (see Fi~re 1). In this case the surface of the half-space
remains flat after deformation, so that geometry changes referred to in
(iii) above are eliminated. The influence of strain hardening on shakedown
will be investigated using the concept of 'kinematic hardening' as defined
by Ziegler [4], in which the yield surface can be displaced in stress
265

space, without change in size or shape. It is the simplest hardening law


which models the generally observed cyclic behaviour of metals,
particularly steels.
To determine shakedown limits, use is made of the shakedown theorems
of the theory of plasticity:
(a) For an elastic-perfectly plastic material Melan's theorem [5] states
that: 'If system of self-equilibrating residual stresses Pij
~

can be found which, in combination with the stresses due to the


repeated load u . . , do not exceed yield at any time, then elastic
IJ
shakedown will take place. '
(b) For a kinematically hardening material Melan's theorem has been
extended by Ponter [6] to state: 'If any system of fictitious

*
residual stresses Pij can be found which, in combination with the
stresses due to the repeated load, u ij do not exceed yield at any
time, then elastic shakedown will take place. '

Note (i) that the fictitious residual stresses P ij are made up of an*
indeterminate combination of the real residual stresses P ij and the

displacements a ij in stress space; (ii) that the stresses P ij for each .*


material element are independent and are not required (as in Melan's
theorem) to satisfy conditions of self-equilibrium.
If at a given load, no system of residual stresses can be found
which satisfies Melan's or Ponter's conditions, i.e. the load exceeds the
shakedown limit, then plastic deformation will take place with every
loading cycle. It can take two forms: either a closed cycle of reversing
plastic strain or an open cycle in which increments of uni-directional
strain accumulate with repeated cycling as demonstrated in Crook's
experiment [3]. Both types of behaviour are likely to lead to failure; the
first by fatigue and the second by ductile fracture. The shakedown limit,
therefore, provides a rational design criterion for repeated rolling
contacts.

Yield criterion
In this paper we shall follow the Tresca (maximum shear stress) yield
criterion. For plane deformation in the x-z plane, it may be expressed:

(U 1 - U2)2 = (uxx - U ZZ )2 + 4T~x = 4k2 (1)


provided that U3 (= U ) is the intermediate principal stress.
yy
266

Residual Stresses
Consider the cylinder rolling on a half-space shown in Figure 1. The
restriction to plane deformation ensures that any residual shear stresses
are independent of y and that the shear stresses pxy and pyz are absent.
Steady-state rolling ensures that any residual stresses are independent of
the coordinate x in the rolling direction. Finally self-equilibrium with a
traction-free surface requires that: Pzz = Pzx = O.
We are left therefore with only two possible components of residual
stress pxx(z) and Pyy(z) which are functions of depth z only.
It should be noted, however, that the fictitious residual stresses

*
Pij associated with Ponter's theorem for a kinematically hardening
material, since they include an unrestricted displacement a .. of the yield
IJ
locus, are not required to satisfy conditions of equilibrium. The

* *
fictitious stresses Pzz and Pzx are therefore not required to be zero.

ROllING AND SLIDING CDNTACf

The stresses in a plane-strain rolling contact, such as that shown in


Figure I, have been analysed by several authors. Expressions for the
stress components under the action of both a normal load P and a tangential
(tractive) force Q are given by Johnson [7], along with a brief table of
values.
The normal pressure is given by Hertz:

p(x) = p {I - x2/a2)~ (2)


o
In sliding contact, with or without rolling, the tangential traction

Figure 1. Rolling/Sliding contact of a cylinder with an


elastic perfectly-plastic half-space.
267

is taken to be

q(x) = ~po {1 - x2/a2}~ (3)


where ~ is the coefficient of sliding friction.
We now plot, from the above data, stress trajectories of TZX/PO
against ~(u -u )/p due to the combined action of pressure p and
xx zz 0
frictional traction q for several different depths, Z = constant, as shown
in Figure 2 for ~ = 0.2. Such trajectories trace the variations in stress
experienced by material elements at different depths, as the rolling load
passes over. The Tresca yield criterion, given by equation (1), maps in
this figure as a circle of radius k/po' with its centre at the origin. The

maximum Hertz stress for purely elastic behaviour Py (the elastic limit) is
o
given by the radius of the smallest circle which circumscribes the stress
trajectories at all depths. For the case of ~ = 0.2, shown in Figure 2,

thts circle has a radius k/pY = 0.308, giving an elastic limit pY = 3.25 k.
o 0
To apply Melan's theorem to find the shakedown limit we add freely
chosen residual stresses p (z) and p (z) to the contact stresses
xx yy
trajectories shown in Figure 2. The stress p (z) is chosen to ensure that
yy
u (= u + p ) is always the intermediate principal stress. Addition of a
3 yy yy

-0.4~ __ ~ __ ~ __-L__ ~ __~____~__~__•

-0.4 -0.2 0.2 0.4

Figure 2. Stress trajectories in rolling and sliding, ~ = 0.2.


Yield loci: - - - - - - First yield; -- - -shakedown, perf. plastic;
- - • - - Shakedown, kin. hardening.
268

compressive pxx(z) has the effect of translating the trajectories in Figure


2 to the left by the magnitude of Pxx' The same effect would be achieved
by translating the yield locus to the right by the same distance. In this

way the shakedown limit pS is determined by finding the smallest circle,


o
having its centre on the horizontal axis, which circumscribes all the
stress trajectories. Such a circle (drawn in Figure 2) has radius

k/p~ 0.281, giving the shakedown limit for a perfectly plastic material
s
Po 3.56 k.
With a kinematically hardening material, Ponter's theorem permits

the centre of the circle to be displaced by arbitrary values of Pxx and *


P:X. This enables an even smaller circle of radius k/p~ = 0.25 to enclose

each of the trajectories, which gives the shakedown limit pS = 4.0 k.


o
It is well known that, at high friction (~ > 0.3), yield first
occurs at the surface (z = O) rather than beneath it. The stress
trajectory for z = 0, which takes the form of a semi-circle centred at the

Surra" • S",bsurfau lIow


Sutuurtou flow

4 " ' - - - - -__ \~__Eepeated plaslle flow

'"
Eloslle " .
Shakedown'- .........
--------------------:::~::-, Surtou fI~

Elastic "
""""'"

."
o
o
...J ...........

O~---7~--~----~----~----~--~
o. a 0-1 0·2 0-3 0·4 o-S
Traction coefficient Il.

Figure 3. Shakedown map for complete slip


- - - - - - Elastic limit: - - - Shakedown limit, el. perf. pI
- - - - Shakedown limi t, kin. hardening.
269

origin and radius ~ encloses all the others. The smallest circle which
circumscribes this trajectory is centred at 0 and has the same radius ~.

It follows, therefore, that the elastic limit and the shakedown limits are
all equal whether or not the material is capable of hardening.

i.e. py
o
= pS0 = k/~ (4)
These results were obtained by Johnson and Jefferis [8]. The
elastic limitb and shakedown limits are plotted against traction
coefficient ia Figure 3, which clearly shows the reduction· in cyclic load
capacity with increasing traction coefficient above the critical value of
0.25. The shakedown limit is then governed by the range of 0xx. The
non-proportional cycle of stress experienced by a surface element: tension
(+oxx) followed by orthogonal shear (TZX) followed by compression (-oxx)
and represented by the semi-circular trajectory in Figure 2, is thought to
be particularly damaging (see Bower and Johnson [10J).

TRACfIVE ROll.ING WIlH PARfIAL SLIP

When rolling takes place with a tangential force Q less than limiting
friction ~P, as in the driving wheel of a vehicle for example, 'microslip'
occurs over part of the contact area, while there is no slip over the
remainder. This problem was first analysed by Carter [11], who showed that
microslip took place towards the trailing edge of the contact region, and
that the stresses in the contacting bodies could be found by the
superposition of a tangential traction

q' = ~po {I - x2/a}~ (5a)


acting over the complete contact (-a ~ x ~ +a) together with a traction

q" = - ~po {(c 2/a2 ) - {x + d)2/a2}~ (5b)


acting over the strip (- a ~ x ~ 2c - a). This distribution of traction is
shown in Figure 4, together with the variation of surface stress 0xx Note
that c =a - d and

Q/~P = 1 - c 2 /a2 = {d/a)(2 - d/a} (6)


Tractive rolling was investigated at Nottingham by Haines and
Ollerton [12], again using the technique of frozen stress photoelasticity,
which gave good support for stress distribution shown in Figure 5.

Low friction
Provided that the coefficient of friction is below a critical value ~c'
270

first yield and shakedown are governed by subsurface stresses. Subsurface


stress trajectories similar to those shown in Figure 2 have been computed
and then elastic and shakedown limits found as before. The results are
plotted in Figure 6 (perfectly plastic) and Figure 6 (kinematically
hardening) against the traction coefficient QlP for different values of the
coefficient of friction ~.

The results are not very different from those presented in Figure 2
for complete slip. This is not surprising since the subsurface stresses
are not likely to be much influenced by the distribution of surface
traction.
High friction
With high friction, when plastic deformation initiates at the surface, the
behaviour is influenced significantly by partial slip.
The normal pressure p(x) gives rise to equal biaxial compression at
the contact surface: Uxx u zz = - p(x). To examine yielding in the z-x
plane, therefore, this stress state can be ignored and only stresses due to
the shear traction q(x), shown in Figure 4, need to be examined. They may
be expressed:
In the no-slip zone, -a ~ x ~ a - 2d
u
2d
~=
a
o
~Po
(7)
T
zx
--= - q
~Po

O.c

Or-------~----~----r_~------~------~
p.,
o
"'-
o
-O.~

-0.4

-1.5 -1.0 -0.5 o 0.5 1.0 1.5


x/a

Figure 4. Partial Slip: surface stresses.


271

In the slip zone, a - 2d ~ x ~ a

a
--=
xx 2d 2 {(x + d)2 - (a - d)2}~ (8)
a
a 2

T
zx
- - = - {I - x2/a2}~
IlPo
Referring to Figure 4,
a a T
~= 2 d/a;
zz zx
At A (x = -a): - - = - - = 0;
IlPo IlP o /l-P o

a a T
zx
At B (x =a - 2d): ~= 2~
a
~=O;
/l-P
--=

a xx d a
zz
AtC(x=a): -=2--4 --=
/l-Po a

where d/a = 1 - {I - Q//l-P}~.


A typical stress trajectory for a surface element is shown in Figure
5, where the letters A,B, and C correspond to the stress states at A, B and
C in Figure 4. It is then a simple matter to draw in the circumscribing
circles which represent the shakedown states.

-- ..........

" '\ \
\
\

Figure 5. Surface trajectory ABC for partial slip,


_ - - - Yield locus for perfect plastic solid.
--- ----- Yield locus for kin. hardening solid.
272

For a perfectly plastic material the centre of the circle S1 lies on


the horizontal axis, displaced from 0 by the residual compressive stress
pxx . The radius of the circle is thus given by

s BC 2 AC
k/JlPo = CS1 = (2) /(2)
so that the shakedown limit is given by
s los -1
Po/k = {Jl(d/a) (2 - d/a)} (9)

Equation (9) gives pS/k as a function of Q/JlP. The shakedown limits


o
are plotted against Q/P in Figure 6a for various values of Jl.
Similarly, for a kinematically hardening material the centre of the

circumscribing circle S2 is free to displace by arbitrary values of * and


Pxi
*
pzx . Its radius is thus given by

k/JlP s = BC/2
o

so that
s d d -~ -1
Po/k = {(a)(2 - a)} Jl (10)
In the same way equation (10) enables the shakedown limits to be
plotted as a function of Q/P and Jl in Figure 6b.

4.0r-=~~.

3.0

2.0

~ = 0.6

LO LO

o 0.2 0.4 0.6 0.2 0.4 0.6


TRACTION COEFFICIENT Qfp
(a) TRACTION COEFFICIENT QfP
(b)

Figure 6. Shakedown maps for partial slip: (al El.Perfect-plastic;


(bl Kin. hardening.
273

It may be shown that, for the range of parameters displayed in

Figures 3 and 6, out of plane residual stress Pyy or Pyy develop to ensure *
that 03 0 +p or 0 + p* is the intermediate principal stress. Yield
yy yy yy yy
is then governed by equation (1) in terms of the in-plane stress plotted in
Figures 2, 4 and 5, and lateral plastic flow is prevented. (The results in
this section are new).

roNa..USION

Use of the Tresca yield criterion permits a simple graphical approach to


shakedown in rolling contact which has been presented in this paper.
Modelling the material as an elastic-perfectly plastic solid in plane
deformation shows the role of residual stresses alone in promoting
shakedown. Modelling the material as a kinematically hardening solid shows
how cyclic hardening properties of the material can, in some but not all
circumstances, increase the shakedown limit. The diagram, in which the
trajectories of stress experienced by material elements and the yield loci
are represented, provides a clear visual indication of the influence of
both residual stresses and hardening upon shakedown.
The method of this paper reproduces a shakedown map (Fig. 3) for
rolling with complete slip, which is already in the literature. The
situation of tractive rolling with partial slip, as found with a locomotive
driving wheel, has been examined and new shakedown maps obtained. These
results demonstrate the damaging effect (reduction in the shakdown limit)
of a high coefficient of limiting friction ~, even if the traction
coefficient QlP at which the contact is operating remains rather modest.

ACKNOWLEOCEMENT

The author is grateful to Dr H R Shercliff for carrying out some of the


computations.

REFERENCES

1. Fessler, H., and Ollerton, E., Contact stresses in toroids under


radial loads, Brit.,J.Appl.Phys .. 1957, L 387-393.
274

2. Lundberg, G., and Palmgren, A., Dynamic capacity of rolling bearings,


Acta Polytechnica, Stockholm, Mech.Eng.Series 1947, I, 1-50.

3. Crook, A.W., Simulated gear-tooth contacts: some experiments on


their lubriction and subsurface deformation, 1957,
Proc.lnst.Mech.Engrs. 171, 187-214.

4. Ziegler, H., A modification of Prager's hardening rule, Quart.of


Appl.Math. 1959, 17, 55-65.

5. Melan, E., Der spannungsgudstand eines Henky-Mises schen Kontinuums


bei Verlandicher Belastung, Sitzungberichte der Ak. Wissenschaften
Wien, Ser. 2A, 1938, 147, 73.

6. Ponter, A.R.S., A general shakedown theorem for elastic-plastic


bodies with work hardinng, 3rd Int.Conf. on Structural Mechanics in
Reactor Tech., London, 1976.

7. Johnson, K.L., Contact Mechnics, C.U.P., Cambridge 1985, pp. 103-205


and 429.

8. Johnson, K.L., A shakedown limit in rolling contact, Proc. 4th


US Nat.Conference of Appl.Mech., Berkeley, 1962, ASME.

9. Johnson, K.L., and Jefferis, J.A., Plastic flow and residual stresses
in rolling and sliding contact. Proc.Inst.of Mech.Engrs. Symposium
on Rolling Contact Fatigue, London, p. 50.

10. Bower, A.F., and Johnson, K.L., The influence of strain hardening on
cumulative plastic deformation in rolling and sliding line contacts',
1989 J.M.P.S., 37, 471-493.

11. Carter, F.W., On the action of a locomotive driving wheel,


Proceedings, Royal Society, A112, 151.

12. Haines D.J. and Ollerton, E., Contact stress distributions on


elliptical contact surfaces subjected to radial and tangential
forces, Proceedings, Institution of Mechanical Engineers, 177, 95.

You might also like