53

Engineering Sciences Division

Explicit equations for the stresses beneath a sliding

spherical contact
G M Hamilton, BSc, PhD
Department of Engineering, University of Reading

Explicit formulae haoe been developed ,for the stresses beneath (I slidiny, normally loaded Hertzian contact. As friction rises a region of‘
impendinyfuilure develops in the back edge of contact.

1 INTRODUCTION 2 DERIVATION OF THE STRESSES

Johnson (1) has recently reviewed the whole field of The method of obtaining the stress functions was given
mechanical contact between non-conforming surfaces. If in the earlier paper and will not he repeated here. It is
the contact is circular and there is no friction then the closely related to the technique devised by Green (6).
problem reduces to one of normal contact first analysed, Other aspects of the same method have been given by
one hundred years ago, by Hertz (2). He set up and Goodman and Keer (7). However the stress functions
solved the integral equation governing the boundary themselves are repeated but in a slightly different form,
conditions and showed that the contact pressure, sub- it being a relatively simple matter, once they have been
ject to certain simplifying assumptions would be hemi- found, to show that they do satisfy the cquations of
spherical. However he did not obtain the stress field elasticity.
beneath the surfaces. This was first done by Huber (3).
Once sliding is introduced it is necessary to re-
establish the boundary conditions and this was done by 2.1 The boundary conditions
Mindlin (4). He made the assumption that for bulk Accepting the simplifications introduced by Hertz and
sliding the shear stress in the surface would be every- Minlin we require the solution to the problem of a semi-
where proportional to the normal stress, the constant of infinite plane, with the contact pressure limited to a
proportionality being the same as the coefficient of fric- circle of radius a, the boundary stresses being given by:
tion between the bounding solids. He was mainly
interested in the surface compliance and the problem of 3P (a’ - x2 - J q 1 / 2
g z = --
partial slip and did not investigate the stresses beneath 271a3
the surface. This was done by Hamilton and Goodman
(5), again with the restriction of a circular region of 3Q (aZ
Tzx = - - - x2 - ).y xz + J’Z < a2
271a3
contact. The paper included a fairly extensive discussion
of the physical implications of the equations for contact zyz = 0
mechanisms.
The equations for the stresses in this earlier paper where P is the total normal load in the z-direction and
were left in an implicit form which involved taking the Q is the total frictional force in the x-direction; all the
imaginary parts of a series of complicated algebraic ex- boundary stresses outside the circle of contact being
pressions. While this can be done easily enough on a zero. There is of course no loss of generality caused by
computer it does mean that explicit expressions are not setting the frictional force in the y-direction to zero. As
available, which complicates subsequent work with the shown by Mindlin the above boundary conditions only
equations in fracture mechanics and plastic shakedown. hold when there is macroscopic sliding, that is:
Recently the author re-examined the expressions and
was surprised to find that it was a relatively simple Q=Pp
matter to extract the imaginary parts explicitly without The radius of the circle of contact a is given by the
making the resulting expressions too unwieldy. well known Hertzian expression :
This leaves the stresses as a series of algebraic formu-
lae which are more convenient for subsequent manipu-
lation. With the equations in this form it is also possible
to show that for zero friction they reduce to those orig-
inally given by Huber (3), although extensive manipu- where E and v are Young’s modulus and Poisson’s ratio
lation is required as Huber’s equations were obtained and R is the radius of curvature. The sufices 1, 2 refer
by a rather different mathematical procedure. to the two bodies in contact.
It is interesting that neither the radius of the circle of
The M S WIF rrceiacd uii 2Y .luly I982 and was accepted j b r publication o n 6
contact nor the contact pressure are affected by the
Septrmber I 9 8 2 presence of the shear stress on the boundary.
26/83 0 IMechE 1983 0263.7 1 54/83/19 7C-0053 X02.0 Proc Instn Mech Engrs Vol 1Y7C

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

54 G M HAMILTON

i
where R‘ = (x’ + y 2 + z ‘ ’ ) ’ : ~ and Z’ = z + it; 5 is a
c. dummy variable inside the contact and 9 implies taking
Direction of sliding the imaginary part of the subsequent function.

Fig. 1 Sign convention for calculations; the stresses relate to

the upper body 2.3 The stresses
The stresses are obtained by application of the standard
equations of the form:
2.2 The stress functions
The displacements and subsequently the stresses in the ox=-
l+v
( vA
-
1-2v
+g)
upper body, that is z 2 0 in Fig. 1, are derived from a
set of four stress functions s;!, Y , Z and Y such that: and

(a) Normal load

u=(l-2v)--z-
as;! ay
ax ax The integrals can all be evaluated in terms of elemen-
tary functions leaving the following equations for the
an aY stresses. Those for the normal load could be simplified
v=(l-2v)--z-
dY aY by taking advantage of thc axial symmetry but as this
8Y does not apply with the tangential loading, it is better to
w = 2(1 - v)Y - z - have both sets in the same coordinates.
a2

(b) Tangential loud (a) Normal load

u =--
aY
ax
ax
ar
2(1- v ) - - z -
ax ax az
ar
azr
a2r
ox = -[(I
3P
2na3
+ v)z# + F“1 7
y2 x2
i -

o = ---2(l-v)--z-
ax
ar a ~ r
ay ay a2
( 1 -vjNzZL-
1 - 2v
3
( N S + 2AN + a3) -
1 vMza

w = (1 - 2v) - - z 2
az az
where Y, $2,r and C are all harmonic functions, that is:
- N(X2 + 2vy’) -
MxZzall
~

v2y = v20= v2r= v2c= 0

with the additional restriction that:
(T

y
=-[(I
3P
2xa3
1
+ V ) Z 4 +r’
i
xz-yz
~

r2
1 2v
as;!
-=
aZ
-y (1 - v)Nz2 --
-

3
(NS + 2AN + a3) - vMza

It is a simple matter to show that with these displace-

ments, the equations of equilibrium, taken in terms of - N(yZ + ~ v x ’ )- -
the displacements are automatically satisfied. That is S
the three equations of the form:
1 8A
v 2 u + -- - -0
1 - 2v ax
where V2 is the Laplacian operator and A is the dila- { - N r 2 + +N(S + 2A)
tation.
Stress functions satisfying the above restrictions and
leading to stresses which meet the boundary conditions
are :
- +
Z(ZN a M ) + $a3) +4
xyz
r
{ aMrZ‘- ZN
--
S
+aM]l
Proc Instn Mech Engrs Vol 197C 0 IMechE March 1983

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

 EXPLICIT EQUATIONS FOR THE STRESSES BENEATH A SLIDING SPHERICAL CONTACT 55

and it is seen that, where appropriate the stresses are in

agreement with the boundary conditions.
(b) Tangential load

6, = $[ (i
+ 1) 4 +-N
{(: $$) -

x (Sv - 2Av + x2z2 7vr’

z’) + -+ -- 2vx2 + r2
3 4 I
+ z({ -).[
3 2x2
-
s
5 (1 - 2v) - 3 (1
A
- - 2v)

1
and - - (22 + 3a’) + -- - - -
2 I s 4 4
G = M 2 - N 2 + Z M - u N ; H = 2MN + U M + zN
On the axis where r = 0, the above formulae are inde-
terminate and have to re-worked, leading to:
3Q [--+r4 3~x4 axM{(i
2na 4 y2)
x [v(S - 2 A + r’) + z’] + 4.22’ 3
-+ - vr2
s 4 I
+ ={(i
r4
;.)[-
1 - 24’2 s6
- (1 - 219 - -
A (1 - 2.)
3
and of course :
z,, = zyz = z,, = 0 2 2
In the surface, that is where z = 0, the original formu-
lae continue to apply but they can be simplified. Thus
on the surface when r 2 a :

-[
1 ayM x2z2
0
y
3P (1 - 2v)(y’ - x”a3
2na3 3r4 1 2na

1 2x2 1 2x2
TXY =
3~
[ 2a3xy(i - 2 4
3r4
1
Isz = zyz = ,z =0

On the surface when r 6 a:

-z2
-
2
3.’
2
f]+ -r2v
2
+ - - -a2x2
4 s
y2
2 --I 3x2
2

- (x2 + 2vy*)(a2
with an equivalent expression for o, with x and y inter-
changed.
- r2)’:.)]
T , , ~= 3Q -
- xyz QM
2nu3 2r4
[ {’1(e
2
+
s 2
- 3a2 - r2
-
2
-)} 2

- (x2 + 2vy2)(u2- r2)l”

I1 x { 1 (s +
- 2A) + z2 -
3
- a2 -
1
- r*
4 4

+-
zy; = 5,, = 0
0 IMechE March 1983 Proc Instn Mech Engrs Vol 197C

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

56 G M HAMILTON

where the symbols A , S , M , N , r and 4 have the same

meaning as before. On the axis r = 0 all the above I.
stresses are zero except :

72, = 3~ [ --a + -32 z tan-’ (:) - 2(aZ +

-az2
zz)
1
On the surface z = 0, the formulae simplify leaving
when r > a, that is outside the contact:

axM, p = 0.25 ’ \\
2na

3Q - - vx$ + a xr4M ,
~

x iv.2 (1 y )+ a- w.}]

- 1,
x

0, = T , , ~= T~~ = 0 and M , = (r2 -

Fig. 2 The distribution of the stress 0, in the surface for a
range of coefficients of friction
When r < a, inside the contact:
stress at the back edge. The maximum thus occurs at
y = z = 0, x = - u and the stress is given by :

f J T X = y
3P
2n-a [
1-2v
-+-
3
where p is the coefficient of friction.
4+v
8 ”1
Some results for the distribution of ox along the
x-axis, for a range of values of p are given in Fig. 2.
Even when I(. = 0 there is a significant tensile stress
which causes the we11 known ring-crack in brittle
materials. A fuller discussion of this tensile stress field
cTz = Tpz = 0 with particular reference to the propagation of cracks
has been given by Frank and Lawn (9,lO). Some experi-
and again it is seen that the surface stresses are in agree- mental results illustrating the effect of these stresses
ment with the boundary conditions. were obtained by Hirst and Gilroy (11).
A similar method of simplifying the equations on the
centre plane y = 0 has been devised by Hills and 3.2 Maximum shear stress
Ashelby (8).
The maximum shear stress occurs in normal contact
(p = 0) below the surface. The position of this maximum
3 DISCUSSION depends slightly on the value of Poisson’s ratio but for
v = 0.3 the maximum is on the axis at z = 0.5-a. As it is
The formulae given in this paper are more explicit than
a three-dimensional stress field it is more convenient to
the earlier ones but the physical implications remain the quote the von Mises yield parameter J J , where:
same. These were fully discussed in the earlier paper (5).
Briefly they reduce to two main themes, the effect of - Lf
J2 - ~ ( ( G X- 0.)’ + (gy - uz)’ + (0;- ox)’]
friction on tensile stress in the surface and the likelihood
of plastic flow beneath the surface.
This can be non-dimensioned by diving through by
3.1 Maximum tensile stress the maximum pressure p,,, .
Figure 3 shows a contour map of JJZ/pmaxon the
This occurs on the circumference of the circle of contact. centre-plane y = 0. This has been drawn for a coefficient
The effect of friction is to add a compressive stress to of friction of p = 0.25 with v = 0.3. Friction has several
the front edge of the contact and to intensify the tensile effects. It increases the maximum value reached by
Proc Instn Mech Engrs Vol 197C @ IMechE March 1983

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

 EXPLICIT EQUATIONS FOR THE STRESSES BENEATH A SLIDING SPHERICAL CONTACT 57

- 0.5 0.0 0.5 1.0 1.5

0-334
x/a
Fig. 3 Lines of equal yield parameter , / J , / p , beneath the surface for a coefficient of friction of O.25

x/a
Pig. 4 Magnified view of the yield contours ,,/J,/p, near the critical point
x = - a (shown as a dotted box in Fig. 3 )
0 IMechE March 19x3 Proc Instn Mech Engrs Vol 197C

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

58 G M HAMILTON

it is unlikely that beyond this value the original Mindlin

idea of the local coefficient of friction being constant,

in surface / over the whole of the contact, would continue to hold.

Thus on both counts, maximum tensile stress and
maximum yield criteria, the region near the back edge
of the contact x = - a , y = z = 0 is the most likely
point of failure. The influence of plastic shakedown on
this stress field has been discussed by Hills and Ashelby
(8).
For the convenience of readers who wish to check the
equations after setting them up on a computer Table 1
has been drawn up for a series of spot values; xla, yfa
and zia have been chosen to be distinct except where
the pole points are involved. The stresses have been
non-dimensioned by dividing through by pmax.
Poisson’s ratio has been set to 0.3.
It will be noticed that T ~ ~ has ( ~ the~ same
~ ~ numeri-
~ , )
cal value as c ~ ~ ( ~ ~ It~ is~ possible
~ ~ ~ , to
~ , manipulate
) . the
0.lt formulae to bring out this identity but as originally de-

I I
0.1
I
0.2 0.3
1 I
0.4 0.5
I I
0.6
I
0-7
vised from the stress functions they are quite dissimilar
in appearance.
A similar problem arises if one wishes to compare the
P normal stress values with those of Huber (3). While it is
Fig. 5 The values of the maximum yield parameter in and a simple matter to show that the two sets of equations
below the surface; for coefficients of friction in excess give the same numerical values it is less easy to make
010.5 there is no longer a distinct maximum below the the comparison algebraically. Hubert’s equations only
surface apply to the plane y = 0. but with this limitation they
were shown after considerable manipulation to be
identical to the above equations.
JJ2fpmax so that yielding will occur at lower loads. The
original maximum moves slightly closer to the surface 4 CONCLUSIONS
and most important of all a new region of likely failure
develops in the surface at x = --a. This is given by: A series of explicit formulae have been derived for the
stresses beneath a sliding, normally loaded, circular
l J 2 1=(1
Y -- 2v)’ (1 - 2v)(2 - v)pcn Hertzian contact. Nunicrically they are equivalent to
Pmax 43r 3 + 4 those given in an earlier paper (5) but rather more con-
venient to use. It is shown that as friction rises a region

64
+ (16 - 4v + 7v2)p2n2
A magnified view of this region is shown in Fig. 4.
I of impending failure develops in the back edge of the
contact.

REFERENCES
Beyond p = 0.25 this region dominates the stress field
as is shown in Fig. 5 which is a plot of the local maxi- 1 Johnson, K. L. One hundred years of Herti contact. Proc. Insrn
mum values of J J J P , , , ~ in
~ the sub-surface region and M e t h . Enqrs, 1982, 196, 363-378.
those at the back edge of the contact as a function of 2 Hertz, H. O n the contact of elastic solids. J . Reine und Ange-
wandie Mathematik, 1882, 92, 156-171 (the date is often given
coefficient of friction. The graph is terminated at 0.7 as wrongly as 1881).

Table 1 Spot values of the stress fields inside and outside the immediate contact area. The stresses have been non-dimcnsioned by
dividing through by the maximum normal pressure. Notice that zzLy= ozr
Stressesip,,,,,

Position xio yja z/a Field Ox OY flz Tyr T:x

Axis 0 0 0 Normal - 0.8 - 0.8 - 1.0 0 0 0

Tangential 0 0 0 0 0 - 1.0

0 0 0.5 Normal -0.1803 -0.1803 -0.8 0 0 0

Tangential 0 0 0 0 0 -0.2696
Surface 0.3 0.7 0 Normal - 0.5445 -0.4925 -0.6481 0.0273 0 0
Inside contact Tangential -0.5066 -0.1060 0 -0.4673 0 - 0.648 1

Surface 1.5 1.9 0 Normal - O.OO53 0.0053 0 0.022 1 0 0

Outside contact Tangential - 0.0589 - 0.0379 0 - 0.0499 0 0
Below surface 0.3 0.7 0.5 Normal -0.1092 - 0.1492 - 0.4492 - 0.02 10 -0.1 867 -0.0800
(r < a) Tangential -0.1058 -0.0239 -0.0800 -0.1105 -0.0333 -0.1520
Below surface 1.3 1.7 1.5 Normal -0.0168 -0.0268 -0.0334 -0.0185 -0.0326 -0.0249
(r > 4 Tangential -0.0192 -0.0169 -0.0249 - 0.023 1 - 0.0243 - 0.02 1 1

Proc Instn Mech Engrs Vol 197C 0 IMechE March 1983

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015

 EXPLICIT EQUATlONS FOR THE STRESSES BENEATH A SLIDING SPHERlCAL CONTACT 59

3 Huber, M. T. On the theory of elastic solid contact. Annln. Phys. tangential compliance of bodies in contact. Proc. I U T A M , Delft,
Lpz., 1904 14, 153-163. A minor misprint occurs in equation (10) 1975, 127-151.
of this paper which should read (1 - p)J(a’ + u ) in the fourth 8 Hills, A. A. and Ashelby, D. W. The influence of residual stress on
term. contact load bearing capacity. Wear, 1982,75,221-239.
4 Mindlin, R. D. Compliance of elastic bodies in contact. J. Appl. 9 Frank, F. C. and Lawn, B. R. O n the theory of Hertzian fracture.
Mech., 1949,16,259-268. Proc. R. Sac., 1967,299,291-306.
5 Hamilton, G. M. and Goodman, L. E. The stress field created by a 10 Lawn, B. R. Partial cone crack formation in a brittle material
sliding circular contact. J. Appl. Mech., 1966, 88, 371-376 (a factor loaded with a sliding spherical indenter. Proc. R. SOC.,1967, 299,
of ?I is missing at the end of equation 10). 307-3 16.
6 Green, A. E. On Boussinesq’s problem and penny shaped cracks. 11 Gilroy, D. R. and Hirst, W. Brittle fracture of glass under normal
Proc. Cumb. Phii. Soc., 1949, 45, 251-257. and sliding loads. Br. J . Appl. Phys., 1969, Ser. 2, 2, 1784-1787.
7 Goodman, L. E. and Keer, L. M. Influence of elastic layer on the

0 IMechE March 1983 Proc Tnstn Mech Engrs Vol 197C

Downloaded from pic.sagepub.com at UNIV NEBRASKA LIBRARIES on April 8, 2015