You are on page 1of 19

COMPUTATIONAL MECHANICS

New Trends and Applications


S. Idelsohn, E. Oate and E. Dvorkin (Eds.)
CIMNE, Barcelona,Spain, 1998
1
AN ORTHOGONAL UNIFORM/PARABOLIC SURFACE PRESSURE
HALF-SPACE ELEMENT FOR THE ANALYSIS OF NON-HERTZIAN
CONTACT
Jan M. de Mul
*
, Joost J. Kalker
**
and Joop M. Vree
*
*
SKF Engineering & Research Centre B.V.
Bearing Theory & Testing Function
Postbus 2350
3430DT Nieuwegein
The Netherlands
e-mail: Jan.De.Mul@skf.com, web page: http://www.skf.com
e-mail: Joop.Vree@skf.com, web page: http://www.skf.com
**
Delft University of Technology
Faculty of Information Technology & Systems
Department of Technical Mathematics & Computer Science
Mekelweg 4
2628CD Delft
The Netherlands
e-mail: J.J.Kalker@twi.tudelft.nl
Key words: parabolic surface pressure, half-space analysis, non-Hertzian contact, finite
bodies, experimental comparison
Abstract. An orthogonal uniform/parabolic surface pressure half-space element is derived
for the analysis of non-Hertzian contact. Restriction is made to parabolic variation of the
surface gap in the contact in one orthogonal direction. The surface pressure is integrated
analytically over the surface element to obtain influence co-efficeints, e.g. for the normal
surface displacement. The co-efficients are assembled into matrices leading to a linear system
of equations. After solving the parabolic pressure variation is transformed to an equivalent
semi-elliptic one co-inciding with Hertzian line contact. Adjustments are introduced to
approximate the finite size of the bodies in contact. The theoretical predictions are compared
to experiments involving the non-Hertzian contact between cylinders of finite lengths.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
2
1 INTRODUCTION
In the first meaningful analysis of the contact problem for parabolic surface separation by
Hertz
1
, the contacting bodies are essentially approximated by elastic half spaces. As a
consequence, the contact compression becomes infinite in the case of line contact. It can be
made finite by involving the dimensions of the bodies, but this is outside the scope of Hertz'
analysis.
Taking the contact surface pressure as uniform over a finite length in one direction and
semi-elliptical in the other, perpendicular direction, Lundberg
2
achieves a finite surface
displacement for a half space in line contact. He also derives the initial surface separation
required to obtain the assumed pressure distribution under load.
Plane strain analysis of a cylinder with circular cross-section under diametrically opposed
loads also yields a finite contact surface displacement, as is shown by Lundberg
3
, Schwartz
and Harper
4
, and Sternberg and Turteltaub
5
.
For general non-Hertzian contact analysis Boussinesq's results for a half space loaded by a
concentrated force on its surface are most popular. By taking some surface element with a
suitable pressure variation, the contact area and pressure mountain can be modelled. Many
researchers like Fridman and Chernina
6
, Conry and Seireg
7
, Kalker and Van Randen
8
, Singh
and Paul
9
, Oh and Trachman
10
, Paul and Hashemi
11
, Nayak and Johnson
12
, Hartnett
13
and
Ahmadi et al.
14
proceed in this way. A special class in this category are constituted by the
multi-grid methods e.g. by Venner
15
. These are particularly computation efficient for large
problems and even allow the inclusion of a lubricant fluid in the contact.
Basically using the half space approach, Reusner
16
introduced adjustments to approximate
the influence of the finite dimensions of the contacting bodies. A complicated but exact
quarter-space method has been developed by Keer et al.
17
Chiu and Hartnett
18
have published
a technique for (almost) straight circular cylindrical bodies that is based on known solutions
for an infinitely long straight circular cylinder. Reusner's relatively simple adjustments for the
finite dimensions of the contacting bodies have been adopted by De Mul et al.
19
in essentially
a half space method based on a rectangular surface element with uniform pressure. In the
present paper the pressure on the rectangular surface element is uniform in only one direction
while in the other perpendicular direction it is assumed parabolic. The analysis is intended for
bodies with parabolic surface separation in at least one direction and is compared to
experiments. For the nomenclature used throughout this paper see section 12.
2 THE HALF SPACE CONTACT MODEL
Two elastic bodies, a and b, are pressed together in the absence of friction. In the region of
the contact of the bodies a plane (z = 0) can be defined that is so close to the surface of the
bodies that the surface displacement of both bodies may be referred to it in the linear theory of
elasticity. A Cartesian co-ordinate system is defined with axes x and y in the mentioned plane
and z-axis pointing inward to body a, see Fig. 1. In the special case that the bodies are bodies
of revolution about parallel axes, the y-axis lies in the plane defined by the axes.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
3
b
z
b
a
w
(
a
y
a
w
y )
b
x
Fig. 1 Contact co-ordinate system, geometry and displacements
In planes of constant y each body has a constant radius of curvature, convex is counted
positive, variation with y is allowed. The centres of the radii of curvature lie in the plane
(x = 0). In this plane the bodies have a surface separation h(0,y) = h(y) in the undisplaced state
which needs not be symmetric about y = 0. With a normal contact displacement along z and
a contact tilt angle about x (both with respect to body a), the resulting interference at
(x = 0, y, z = 0) is given by :

(y)

+

y h(y)

(1)
A combined radius of curvature is defined by:
1

(y)

1
a
(y)
+
1
b
(y)

(2)
The undeformed surface gap between the bodies g(x,y) is approximated by parabolic
variation with x as follows :
g(x, y)
x
2
2

(y)


(y)

(3)
Since interference is physically not admissible a contact pressure p(x, y) is generated in the
contact area A that causes displacements w
a
and w
b
along z, into each body. The sum
displacement w is given by:
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
4
w(x, y) w
a
(x, y) + w
b
(x, y)

(4)
The deformed surface distance then becomes
D(x, y) g(x, y) + w(x, y)

(5)
The contact/no contact conditions are given by,
contact if (x,y) inside A :
p(x, y)

0 and D(x, y) 0

(6)
no contact if (x,y) outside A :
p(x, y) 0 and D(x, y) > 0

(7)
The contact load and tilt moment (on body a) follow from :
Q

p(x, y) dxdy

(8)
T

p(x, y) y dxdy

(9)
Reusner
16
uses a surface element with uniform pressure along y and semi-elliptical
variation over x, with semi-axes p(0,y) and b(0,y) :
p(x, y) p(0, y) 1
x
2
b(0, y)
2

(10)
To obtain the two unknowns p(0,y) and b(0,y) he introduces two conditions. One is
involving the deformed surface distance D(0,y) at the contact centre line, the other the sum of
curvatures 1/(y). In this paper a similar approach is taken. Since the semi-elliptical pressure
distribution cannot be integrated analytically in the expression for the influence co-efficient of
the element, the semi-ellipse is approximated by a parabola, see Fig. 2. With reduced summit
pressure s(y) and contact semi-width c(y) it reads :
p(x, y) s(y) c(y)
2
x
2

(11)
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
5
i
2
i
c
i
s c
c
i
i
y
Fig. 2 Uniform/parabolic surface pressure mountain
The deformed surface distance at the contact centre line becomes :
D(0, y) g(0, y) + w(0, y)

(12)
D(x, y) is developed into a Taylor series around x = 0 :
D(x, y) D(0, y) + x

D

x
(0, y) +
x
2
2

2
D

x
2
(0, y) +
x
3
6

3
D

x
3
(0, y) + ....

(13)
Terms of order higher than 3 are neglected. Inside the contact area A it is required
D(x,y) = 0. In Eq. (13) the uneven terms D/x and
3
D/x
3
vanish due to the symmetry in the
loading, yielding the following requirements from Eqs (3), (5), (6), (12) and (13) :
D(0, y) g(0, y) + w(0, y) 0

(14)

2
D

x
2
(0, y)
1

(y)
+

2
w

x
2
(0, y) 0
(15)
With Eqs. (11), (14) and (15) the contact condition Eq. (6) is approximately satisfied.
The total contact load is found by integration over x and summing over the elements i, see
Fig. 2 :

Q
i
4
3
s
i
c
i
3
y
i

(16)
Analogously for the total contact moment :

T
i
4
3
s
i
c
i
3
y
i

y
i

(17)
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
6
Boussinesq's results for a half space loaded by a concentrated force are used to find the
surface displacement w :

2 2
) ' ( ) ' (
' ' ) ' ' (
2
1
) , (
y y x x
dy dx y x p
G
v
y x w


(18)
With Eq. (11) it is found for the displacement in (x,y) due to the uniform/parabolic
pressure on rectangle j :

b a
b a b a
j
G
v
y x w
,
, ,
2
1
) , (

j
j
j
j
j
j
c
c
y
y
y
y
j
j
y y x x
dy x c
dx s
2
2
2 2
2 2
) ' ( ) ' (
' ) ' (
' (19)
Performing variable transformation and partial integration the following transpires :

b a
b a b a
j
G
v
y x w
,
, ,
2
1
) , (

1
1
1
1
1
1
1
1
]
1

1
1
1
1
1
1
1
1
]
1


+ + +
+

'

2 2
1
2
2 2
1
2
2 2
' ' ' )
6
'
(
'
'
)
6
'
(
'
'
'
3
'
' ) (
y x y
x
x
y
x
sinh
y
x c
x
y
sinh x
x
xx x c
j
j
(20)
with :
x c w
j
t
t
'

(21)
2
'
j
i
y
y y y

t
t

(22)
Defining the compliance co-efficient :
w
j
a,b
(0, y
i
) C
ij
a,b
s
j

(23)
It follows from Eq. (20) that :

+


1
1
1
1
]
1

+
+

,
_

+ +

'
'
2 2
1
2
2 1 3
,
, ,
' '
3
'
'
3
'
2
'
3
4
2
1
y
y
j
j
j
j
j
j
b a
b a b a
ij
y c y
c
y
c
sinh y
y
c
c
y
sinh c
G
v
c

(24)
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
7
The derivative with respect to c
j
is required also :

1
1
]
1


'
' 4
'
4
2
1
1 1 2
,
,
,
,
y
c
sinh y c
c
y
sinh c
G
v
c
C
C
j
j
j
j
b a
b a
j
b a
ij
b a
dij

(25)
Combining bodies a and b :
C
ij
C
ij
a
+ C
ij
b

(26)
C
dij


C
ij

c
j
C
dij
a
+ C
dij
b

(27)
From Eqs. (4), (23) and (26) :
w
j
(0, y
i
) C
ij
s
j

(28)
Adding the influence of all rectangles j :

w(0, y
i
)
j
C
ij
s
j

(29)
Substituting this and Eq. (3) in Eq. (14) and re-arranging, one contact equation is found :

j
C
ij
s
j


(y
i
)

(30)
For the other contact equation the second devirative of the surface displacement is needed.
Defining :


2
w
j
a,b

x
2
(0, y
i
) P
ij
a,b
s
j

(31)
it is found from Eq. (19) :
b a
b a b a
ij
G
v
P
,
, ,
) 1 ( 2

1
1
]
1

'
'
1
'
'
y
y
j
y
c
sinh y
(32)
The derivative with respect to c
j
will again be needed :
b a
b a b a
ij
G
v
P
,
, ,
) 1 ( 2

1
1
]
1

+
'
'
2 2
'
'
y
y
j
y c
y
(33)
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
8
Combining bodies a and b :
P
ij
P
ij
a
+ P
ij
b

(34)
P
dij


P
ij

c
j
P
dij
a
+ P
dij
b

(35)
From Eqs. (4), (31) and (34) :


2
w
j

x
2
(0, y
i
) P
ij
s
j

(36)
Adding the influence of all rectangles j :


2
w
j

x
2
(0, y
i
)
j
P
ij
s
j (37)
Substituting this in Eq. (15) the other contact equation is found :

j
P
ij
s
j

1

(y
i
)

(38)
Using matrix [M]=[M
ij
] and vector {v}={v
j
} notation Eqs. (30) and (38) read :

[
C
]
s


(39)

[
P
]
s {
1
}

(40)
Eqs.(39) and (40) may be solved iteratively by linearising with respect to variations in the
contact semi-width c
j
. Thus :
[ ] [ ] [ ]{ } { } + s c C C
j dij

(41)
[ ] [ ] [ ]{ }

'

1
s c P P
j dij
(42)
From step k to k + 1 in the iteration :
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
9
{ } [ ] [ ] [ ]

'

+

+

1
1
1 j dij k
c P P s
[ ] [ ] [ ] [ ]

'

1
1
1
1
1
P c P P
j dij
[ ] [ ] [ ][ ]

'

1
. 1
1 1
P c P P
j dij

(43)
From Eq. (40) :
[
P
]
1
{
1
} s
k

(44)
Into Eq. (43) :
s
k+1

[1 [
P
]
1
[P
dij

c
j
]] s
k
(45)
Substitution into Eq. (39) :

[[C+ [C
dij

c
j
]] [1 [
P
]
1
[P
dijc
j
]] s
k

[[
C
] + [C
dijc
j
] [
C
][
P
]
1
[P
dijc
j
]] s
k

[[
C
] + [C
dijc
j
] [CP
1
P
dijc
j
]] s
k



(46)
[[C
dijc
j
] [CP
1
P
dijc
j
]] s
k


[
C
]
s
k
(47)
alternative :

[[C
dij
s
j
] [CP
1
P
dij
s
j
]] c

(48)
With this the basic system of linear equations that needs to be solved for the contact has
been established. Once the contact pressures and semi-widths have been found for a particular
value of approach and tilt angle between the bodies, the total contact load Q and tilt
moment T are calculated from Eqs. (16) and (17). If the difference from the desired load and
moment values is too large and are changed and the iteration is continued until the
difference drops below the limit set.
This process is carried out using the Newton-Raphson method, making use of the
derivatives of the load and moment with respect to the approach and tilt angle, which can be
established analytically. This is not further explained since it is not essential for solving the
contact problem.
3 MODELLING THE LENGTH OF THE BODIES
Reusner
16
developed adjustments of the influence co-efficient to approximate the effect of
the finite length of the bodies along the contact axis (y). It is based on mirroring the pressure
mountain about the body side faces to cancel the subsurface shear stresses on these side faces.
This is explained in detail by De Mul et al.
19
and therefore not repeated here. All "influence"
matrices [C], [C
d
] (defined by Eq.(27)), [P] and [P
d
] (defined by Eq.(33)) are affected.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
10
4 MODELLING THE DEPTH OF THE BODIES
Also for this Reusner
16
developed adjustments of the influence co-efficient to approximate
the effect of the finite depth of the bodies. It is based on the displacement difference between
the surface and the location at a certain depth below the surface. This has been adapted and
explained in detail by De Mul et al.
19
and is therefore repeated here only for those parts that
are changed. Defining :

1
1
]
1

+
+


) (
'
) 1 ( 2
) ( '
'
'
,
1
2
,
2
,
y d
y
sinh v
y d y
y
Y
b a
b a
b a
(49)
Eq. (24) of De Mul et al.
19
reads while dropping here the redundant x argument and
defining q
j
= f
xj
:
w
a,b
(y, d
a,b
(y))

1
4
$
G
a,b

j
q
j
Y
a,b


(50)
With :
q
j

4
3
s
j
c
j
3

(51)
Eq. (50) becomes :
w
a,b
(y, d
a,b
(y))


j
c
j
3
3

G
a,b
Y
a,b

s
j

(52)
The adjusted displacement difference then becomes :
w
a,b
d
(0, y) w
a,b
(0, y, 0) w
a,b
(y, d
a,b
(y))

(53)
The effect is taken into account by adapting the "influence" co-efficients in Eqs. (24) and
(25) :

C
ij
a,b

c
j
3
3

G
a,b
Y
a,b

(54)
C
dij
a,b

c
j
2

G
a,b
Y
a,b

(55)
If | y
j
- y

| << d
a,b
(y) the equations above can be approximated to :
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
11

C
ij
a,b

3 2
a,b
3G
a,b
d
a,b
(y
i
)
c
j
3

y
j
(56)

C
dij
a,b

3 2
a,b
G
a,b
d
a,b
(y
i
)
c
j
2

y
j
(57)
So far mirroring has not been considered. When this is done the additional terms coming
from the mirroring each cause their own adjustment, see De Mul et al.
19
Eq. (32).
Since the elastic displacement at depth d(y) does not depend on x [P] and [P
d
] are not
affected.
5 COMPARISON WITH HERTZIAN ANALYSIS
The uniform/parabolic pressure element is first compared to Hertzian line contact. For
equal elastic constants Eq. (32) becomes :

1
1
]
1


'
'
) 1 ( 4
1
y
c
sinh y
G
v
P
j
ij

(58)
i = j is set and the limit of the element length y
j
is taken. Using l'Hospital's rule it is found
:
i ii
y
c
G
v
P
lim
i

) 1 ( 8

(59)
Using Eqs. (10), (16) and (38), De Mul and Van Engelenburg
20
Eqs. (1) - (5), and :

G
E
2(1 +

)
(60)
it is found that the Hertzian semi-width and maximum pressure are given by :

b
2
3
c 0.81650 c

(61)
p
0

4 6
3
s c
2
1.03960 s c
2

(62)
This compares with p(0,0) = sc
2
from Eq.(11).
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
12
So the parabola underestimates the maximum contact pressure and overestimates the
contact width. The factors shown in Eqs. (61) and (62) can be used to modify the parabolic
values to achieve the Hertzian values. In the case of rolling bearing fatigue life analysis this is
important since the fatigue life depends on the 9.3rd power of the subsurface shear stress
amplitude, which depends linearly on the maximum elliptic pressure, see De Mul at al.
21
. In
the results shown in the present paper the modification of the parabolic values to the semi-
elliptic values is made everywhere.
Modification factors having been established comparison may now be made to an elliptical
Hertzian point contact example from De Mul et al.
21
, see Fig. 3. The Hertzian elliptical
pressure and half-width distribution (symmetric quarter shown only) is compared with the
computed distributions. The correspondence between the present model and the Hertzian
analysis is excellent.
0
0.2
0.4
0.6
0.8
1
1.2
y/a
p
(
y
)
/
p
0
,


b
(
y
)
/
b
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
Hertz
model p-ratio
model b-ratio
Fig. 3 Elliptical contact pressure and semi-width variation - model versus Hertz
6 COMPARISON WITH PLANE STRAIN CIRCULAR CYLINDERS
The approach taken by De Mul and Van Engelenburg
20
to fit the depth adjustment is
followed here. Their Eq. (6) gives the contact centre displacement of a cylinder in plane strain.
Analysing two long cylinders with the present model plane strain can be simulated.
Comparing the results with the equation mentioned above gave a depth adjustment equal to
the diameter of the cylinder considered to yield excellent agreement for different curvature
sums 1/. Apart from this simple proportionality to the diameter the depth was found to be
unaffected by the load level.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
13
Using the same procedure De Mul and Van Engelenburg
20
had found a different depth
adjustment, namely 3/4 diameter of the cylinder considered. This is attributed to the different
pressure element, i.e. uniform instead of uniform/parabolic
In the next section the uniform/parabolic model with the depth adjustment of the diameter
is applied to the contact between rollers or needles and bearing rings.
7 COMPARISON WITH EXPERIMENTS
The B-TAN-2 roller and CON needle experimental cases are taken from De Mul and Van
Engelenburg
20
. The (measured) roller and needle profiles are shown in Fig. 4 and 5. The rings
had straight profiles.
Fig. 4 B-TAN-2 roller profile
Fig. 5 CON needle roller profile trace
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
14
In Fig. 6 the measured and calculated footprints of the contact are shown in overlay, while
the latter is also shown separately. Excellent agreement is observed.
Fig. 6 Footprint of B-TAN-2 profiled roller on (plugged) inner ring at 20 kN load - comp. vs. exp.
In Fig. 7 the contact pressure and semi-width distributions over the contact centre line are
shown. Fig. 6 and 7 can be seen to agree well with the symmetric quarter of the pressure
mountain shown in Fig. 9 of De Mul and Van Engelenburg
20
, calculated with a uniform
pressure rectangular surface element.
-10 -8 -6 -4 -2 0 2 4 6 8 10
-4
-2
0
2
4
y [mm]

-
b
(
y
)

[
m
m
/
1
0
]
















p
(
0
,
y
)

[
G
P
a
]

pressure
semi width
Fig. 7 Variation of contact presssure and semi-width along contact centre line of B-TAN-2 profiled roller on
(plugged) inner ring at 20 kN load - computation
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
15
In Fig. 8 and 9 the measured and calculated load-compression relationships for B-TAN-2
and CON are shown. With the depth adjustment of the diameter good agreement between
experiments and calculations is obtained for B-TAN-2 on the stiff side while for CON
reasonable correspondence is achieved on the weak side.
Fig. 8 Computed and experimental load vs. compression for B-TAN-2 profiled roller on (plugged) inner ring
Fig. 9 Computed and experimental load vs. compression for CON profiled needle roller on solid cylinder
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
16
8 MODEL PERFORMANCE CHARACTERISTICS
Compared to models using a uniform/uniform pressure element e.g. De Mul et al.
19
, the
uniform/parabolic element has the advantage that the contact semi-width b(y) does not need to
be known a priori as it is a result of the calculation. In practice this reduces the modelling
effort significantly. Also the lower number of elements increases the computation efficiency
with calculation time according to O(n
2
), n being the number of elements. Since the element
itself is more complex the theoretical maximum benefit is not achieved. Compared to De Mul
et al.
19
an improvement by a factor of about 2 in CPU time is obtained. While the contacting
bodies in this model may have arbitrary profile in the y-direction (axial), in the x-direction
(circumferential) they must be parabolic, consistent with their use as an appriximation to
circular bodies. The model is thus particularly applicable to roller/raceway profiling
investigations, where the number of elements required is moderate.
A class of methods using a uniform/uniform pressure element that are particularly
computation efficient for large problems is the multi-grid technique e.g. by Venner
15
. The
basis for this advantage is that the calculation effort increases with O(nlogn) rather than O(n
2
).
The multi-grid models are typically used for analysing 3-D surface features/topography, EHL
or both and require very powerful PCs or work stations. Compared with multi-grid methods as
so far implemented, the present uniform/parabolic element approach enjoys several
advantages. One is its ability to correct for the finite size of the contacting bodies, shown here
to be of some considerable importance. Another is the ease with which local grid refinement
can be introduced, since no restriction is placed on the element aspect ratio. The latter effects
boost calculation efficiency. In addition to this, if the problem is not large the method benefits
from being simpler and therefore having less overhead than multi-grid and shows higher
computation efficiency. Both the present uniform/parabolic element method and the
uinform/uniform element method programs are running on a VAX 4100A computer and
require less than 1 minute CPU time for a typical profile calculation.
9 CONCLUSIONS
An orthogonal uniform/parabolic surface pressure half-space element for the analysis of
non-Hertzian contact has been developed. The major advantage of the element is that the
parabola allows analytic integration of the relevant influence co-efficients and lowers the
number of elements compared to the frequently used orthogonal uniform surface pressure half
space element while the contact semi-width is even an explicit part of the solution. For
subsequent analysis involving the contact pressure, such as the particularly sensitive
subsurface-stress-based rolling bearing fatigue life, the parabolic pressure variation is scaled
to the more realistic semi-elliptic distribution. Comparison to Hertzian contact then shows
excellent agreement.
In addition relatively simple adjustments are introduced in the model to approximate the
finite dimensions of the contacting bodies, making it particularly applicable to the finite
cylinders occurring in roller bearings. Comparison of the model with experiments shows, for
the most part, good agreement.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
17
10 ACKNOWLEDGEMENT
The authors wish to thank Dr. H.H. Wittmeyer, Managing Director of SKF Engineering &
Research Centre B.V., for his kind permission to publish this paper, Robert-Jan Buunk for
assisting with the computer program runs and preparation of the pictures, and Hugo
Willeumier for contributing to the conversion to MS Word.
11 REFERENCES
[1] H. Hertz, ber die Berhrung fester elastischer Krper, 1881, ber die Berhrung
fester elastischer Krper und ber die Hrte, 1882, Gesammelte Werke, pp. 155-196.
[2] G. Lundberg, Elastische Berhrung zweier Halbrume, Forschung auf dem Gebiete
des Ingenieurwesens, Sep./Oct. 1939, pp. 201-211.
[3] G. Lundberg, Cylinder Compressed between Two Plane Bodies, SKF, Gteborg,
Sweden, 1949.
[4] J. Schwartz and E.Y. Harper, On the Relative Approach of Two-Dimensional Bodies
in Contact, International Journal of Solids and Structures, Vol. 7, 1971,
pp. 1613-1626.
[5] E. Sternberg and M.J. Turteltaub, Compression of an Elastic Roller Between Two
Rigid Plates, Continuum Mechanics and Related Problems of Analysis,
Muskhelishvili Anniversary Volume, pp. 495-515, Nauka Publishing House, Moscow,
1972, L.I.
Sedov et al., Eds.
[6] V.M. Fridman and V.W. Chernina, Iteration Methods Applied to the Solution of
Contact Problems in Bodies, (in Russian) Mekh. Tverdogo Tela AN SSSR, Vol. 1
(1967), pp. 116-120.
[7] T.F. Conry and A. Seireg, A Mathematical Programming Method for Design of
Elastic Bodies in Contact, ASME Journal of Applied Mechanics, Vol. 38 (1971),
pp. 387-392.
[8] J.J. Kalker and Y. van Randen, A Minimum Principle for Frictionless Contact with
Application to Non-Hertzian Half Space Contact Problems, Journal of Engineering
Mathematics, Vol. 6, No. 2 (Apr. 1972), pp. 193-206.
[9] K.P. Singh and B. Paul, Numerical Solution of Non-Hertzian Elastic Contact
Problems, ASME Journal of Applied Mechanics, June 1974, pp.484-490.
[10] K.P. Oh and E.G. Trachman, A Numerical Procedure for Designing Profiled Rolling
Elements, ASME Journal of Lubrication Technology, Oct. 1976, pp. 547-552, 574.
[11] B. Paul and J. Hashemi, An Improved Numerical Method for Counterformal Contact
Stress Problems, Report No. FRA-ORD-78/26, US Department of Transportation,
Federal Railroad Administration.
[12] L. Nayak and K.L Johnson, Pressure Between Elastic bodies Having a Slender Area
of Contact and Arbitrary Profiles, International Journal of Mechanical Science, Vol.
21 (1979), pp. 237-247.
[13] M.J. Hartnett, A General Numerical Solution for Elastic Body Contact Problems,
Solid Contact and Lubrication, AMD-Vol. 39, ASME 1980, Eds. H.S. Cheng and
L.M. Keer.
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
18
[14] N. Ahmadi, L.M. Keer and T. Mura, Non-Hertzian Contact Stress Analysis for an
Elastic Half Space - Normal and Sliding Contact, International Journal of Solids and
Structures, Vol. 19, No. 4, pp. 357-373.
[15] C.H. Venner, Multilevel Solution of the EHL Line and Point Contact Problems, Ph.D.
Thesis, Feb. 1991, Twente University of Technology, Enschede, the Netherlands.
[16] H. Reusner, Druckflchenbelastung und Oberflchen-verschiebung im Wlzkontakt
von Rotationskrpern, Dissertation, May 1977, University of Karlsruhe, Germany.
[17] L.M. Keer, J.C. Lee and T. Mura, A Contact Problem for the Elastic Quarter Space,
International Journal of Solids and Structures, Vol. 20 (1984), No. 5, pp. 513-524.
[18] Y.P. Chiu and M.J. Hartnett, A Numerical Solution for the Contact Problem Involving
Bodies with Cylindrical Surface Considering Cylinder Effect, ASME Journal of
Tribology, Vol. 109 (July 1987), pp. 479-486.
[19] J.M. de Mul, J.J. Kalker and B. Fredriksson, The Contact Between Arbitrarily Curved
Bodies of Finite Dimensions, ASME Journal of Tribology, Vol. 108 (Jan. 1986), pp.
140-148.
[20] J.M de Mul and H. van Engelenburg, The Line Contact Between Two Cylinders
Some Experiments and Theoretical Predictions, ASME Journal of Tribology,
Vol. 110 (Apr. 1988), pp. 285-291.
[21] J.M. de Mul, J. M. Vree and J.C. Kuijpers, The Influence of Certain Raceway Dent
Geometries (3-D) on Contact Stresses and Rating Fatigue Life of Rolling Bearings,
ASME Journal of Tribology, Vol. 109 (July 1987), pp. 452-461
12 NOMENCLATURE
12.1 Parameters
a
Hertzian contact semi-length [L] F = force unit, L = length unit
b
Hertzian contact semi-width [L]
contact element semi-width (semi-elliptic pressure) [L]
c
contact element semi-width (semi-parabolic pressure) [L]
C
compliance co-efficient [L
5
/F]
d
depth below surface [L]
D
deformed surface distance [L]
E
elastic modulus [F/L
2
]
g
undeformed surface gap [L]
G
shear modulus [F/L
2
]
Jan M. de Mul, Joost J. Kalker

and Joop M. Vree
19
h
undisplaced surface separation [L]
p
contact pressure [F/L
2
]
p
0
maximum Hertzian contact pressure [F/L
2
]
P
partial derivative (influence) co-efficient [L
3
/F]
q
load intensity [F/L]
Q
contact load [F]
s
parabolic summit pressure [F/L
4
]
T
contact moment (about x) [FL]
w
surface displacement [L]
x, y, z
right hand Cartesian co-ordinate axes [L]
' , ' y x integration variables (along x and y) [L]
t
' x x c x
j
t
t
' [ ] L
t ' y
2
'
j
j
y
y y y

t
t
[L]
contact tilt angle (about x) [radian]
contact normal approach [L]
c increase of c [L]
y element length [L]
Poisson's ratio
radius of curvature [L]
12.2 Subscripts
a,b body
d derivative with respect to c
j
i point considered
j location of pressure element
k iteration step
12.3 Superscripts
a,b body
d depth adjusted

You might also like