Journal of Solid Mechanics

and Materials

Engineering

1080
Vol. 2, No. 8, 2008
Fatigue Crack Growth Behavior of Subsurface
Crack in a Semi-Infinite Body Due to
Rolling-Sliding Contact
*

Asami MIZOGUCHI
**
, Takahito GOSHIMA
**
, Masayoshi SHIMIZU
**

and Sotomi ISHIHARA
**

**Faculty of Engineering, University of Toyama,
3190 Gofuku, Toyama, 930-8555 Japan
E-mail: [email protected]


Abstract
At first, we analyze the stress intensity factors at a crack tip of multiple kinked
crack from a horizontal initial crack in a half-space due to rolling/sliding contact
with frictional heat. On the basis of the results of stress intensity factors and
applying the maximum energy release rate criterion to each kinks in order, the
two-dimensional fatigue crack growth path are simulated, and making use of the
fatigue crack growth law obtained from fatigue experimental results for a high
carbon chromium bearing steel (AISI-52100), the propagation fatigue lives until the
surface tribological failures are predicted due to the repetition of rolling-sliding
contact. And we show that the increase of frictional coefficient, sliding/rolling ratio
reduce the propagation fatigue lives, and the shortening of the depth of a horizontal
initial subsurface crack reduce the propagation life and is apt to be easily induced
the fatigue surface tribological failures.
Key words: Elasticity, Thermal Stresses, Contact Problem, Subsurface Crack, Crack
Growth, Fatigue Life

1. Introduction
The rolling contact fatigue failure, such as shelling in railroads, spalling in rollers ,
may be manifested by originating and growing subsurface crack due to periodic
rolling-sliding contact. These subsurface cracks may be initiated by preexisting defects such
as inclusions, gas pores, or local soft spots, or may be generated during the cyclic straining
process itself. Since the analysis for delamination theory of Suh
(1)
, a considerable amount
of research
(2) - (6)
on fracture mechanics has been performed in order to understand the
mechanism of subsurface crack initiation and propagation in rolling-sliding contacts.
Moreover, Goshima et al.
(7)(8)
have studied the thermal stress intensity factors for a single or
two subsurface cracks due to rolling-sliding contact with frictional heat. The all of these
researches have considered only straight growth of cracks. However, the actual crack is apt
to branch and is propagated into a pitting or shelling failure. As to the analysis for branched
crack due to the rolling contact, Farris et al.
(9)
, Yu et al.
(10)
and Mukai et al.
(11)
have studied
the kinked direction from subsurface crack. However, they do not consider the crack growth
path until the failure and their fatigue life due to the rolling-sliding contact. Although,
Goshima et al.
(12)
have analyzed the surface crack growth path until the pitting failure and
predicted their fatigue lives due to rolling-sliding contact with frictional heat for a high
carbon chromium bearing steel (AISI-52100), their results are not apply to the subsurface
crack.
*Received 27 Feb., 2008 (No. 08-0115)
[DOI: 10.1299/jmmp.2.1080]


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1081
In this paper, we analyzed the stress intensity factors for a multiple kinked crack from
an initial subsurface crack in a half-space due to rolling-sliding contact with frictional heat.
In the crack analysis, we assume that the crack face friction is neglected. In the temperature
analysis, it is assumed that the speed of the moving contact region is much greater than the
ratio of the thermal diffusivity to the contact length (large Peclet number), and the
temperature distribution is not disturbed by the crack. On the basis of the results of stress
intensity factors for the multiply kinked crack model, and applying the maximum energy
release rate criterion
(13)
to each kinks in order, the two-dimensional fatigue crack growth
path are simulated, and making use of the fatigue crack growth law obtained from fatigue
experimental results for a high carbon chromium bearing steel (AISI-52100), the
propagation fatigue lives are predicted. And thermomechanical effects such as the frictional
coefficient, sliding/rolling ratio and the depth of a horizontal initial subsurface crack on the
fatigue crack propagation life and induced fatigue tribological failures are considered.

Nomenclature
c : half contact length, m
h
~
: depth of a horizontal initial subsurface crack, m
V : rolling velocity, m/s
V
s
: sliding velocity during rolling, m/s

1
~
l : initial crack length, m

k
l
~
: each kinked crack length (k=2.3.......n), m

k
: each kinked angle of crack-k from a horizontal initial crack

*
k
: = 180
*
k k

)
~
,
~
( y x : cartesian coordinates fixed to the moving roller, m
)
~
,
~
(
1 1
: cartesian coordinates fixed to an initial crack, m
)
~
,
~
(
1 1
y x : the origin of )
~
,
~
(
1 1
for an initial crack, m
1
~
x : moving distance of roller from an initial crack, m
)
~
,
~
(
k k
: cartesian coordinates fixed to the each kinked cracks (k=2,3........n), m
f : frictional coefficient on the surface of a semi-infinite body
K
t
: thermal conductivity, W/(m.K)

t
: thermal diffusivity, m
2
/s

0
: coefficient of thermal expansion, /K
E
0
: Young's modulus, Pa
G
0
: shear modulus, Pa
: Poisson's ratio
P
0
: maximum Hertzian contact pressure, Pa
) , , (
xy yy xx
: stresses in )
~
,
~
( y x system, Pa
) , , (
k k k k k k

: stresses in )
~
,
~
(
k k
system, Pa
) , (
yy xx
U U : displacements in )
~
,
~
( y x system, m
T : temperature change from the room temperature, K
)
~
(x Q : frictional heat generation
)
~
(
1
x P : contact pressure, Pa
II I
K K , : stress intensity factors, Pa.m
0.5

G : energy release rate, Pa.m
Dimensionless parameters
) /
~
, /
~
( ) , ( c y c x y x = : dimensionless coordinates fixed to the moving roller
) /
~
, /
~
( ) , ( c c
k k k k
= : dimensionless coordinates fixed to the each kinked cracks
z = x + i y : dimensionless complex valuable
c l l
k k
/
~
= : dimensionless crack length


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1082
c h h /
~
= : dimensionless depth of a horizontal initial subsurface crack
) /
~
, /
~
( ) , (
1 1 1 1
c y c x y x = : dimensionless origin of ) , (
1 1
for an initial crack
V V S
s r
/ = : sliding/rolling ratio
t e
cV p / = : Peclet number
0 1
/ )
~
( ) ( P x P x P = : dimensionless distribution of contact pressure
)} 1 ( /{ ) 1 ( 2
0 0 0
+ =
t t
K G H : dimensionless parameter for thermal stresses

2. Problem Formulation

An elastic half-space containing a multiple kinked subsurface crack is subjected to
rolling-sliding contact accompanied by frictional heat with constant velocity V, as shown in
Fig.1. The coordinates )
~
,
~
( y x (which moves with constant rolling velocity V) are fixed to
the roller. The coordinates )
~ ~
(
, k k
; (k=1,2,3..n) are fixed to each kinked cracks.














Fig.1 Geometry and coordinate system

Assuming that all the work done by the friction load is transformed into heat energy, the
frictional heat generation is given as follows.

) ( ) ( )
~
(
0 0 1
x P P S V f x P P V f x Q
r s
= = (1)

The region outside the area of contact is assumed to be thermally insulated.
Furthermore, it is assumed that the temperature distribution is not affected by the presence
of cracks. Thus, the thermal boundary conditions can be given as follows.

>

=
|
|
.
|

\
|

=
1 , 0
1 , / ) (
0
0
x
x K x P P S V f
y
T
t r
y
(2)

0 ) ( =
y
T (3)

The mechanical boundary conditions on the surface and at infinity of the half-space are
given as follows.

>

=
=
1 , 0
1 , ) (
) (
0
0
x
x x P P
y yy
(4)

V
1
~
y
x
~
)
~
(x Q
y
~
1
~
x
)
~
(
1
x P
2

)
~
(
1
x fP
1
~
l
2
~
l
n
l
~
2
~

~
1
~

2
~

~
n

c 2
crack initial Horizontal
1
~

Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1083

>

=
=
1 , 0
1 , ) (
) (
0
0
x
x x P P f
y xy
(5)

0 ) , , ( =
y xy yy xx
(6)

Assuming that the crack face friction is neglected, the boundary condition along the k
th

kinked cracks (k=1,2,3........,n) may be expressed as follows.


k k
l
k k k
< < =
=


0 , 0 ) (
0
(7)


op
k
k
k k k


=
=
, 0 ) (
0
(8)

where
op
k
is the crack face opening region of the k
th
kinked crack.
We require the continuity of displacements throughout the body except at the crack,
and hence the condition for single-valued of displacements is shown as follows.

= + 0 ) ( z d U i U
yy xx
(9)

where

denotes the integration along a contour around the multiple kinked subsurface
crack.

3. Stress Analysis

The stress field
pq
is represented by superposition as:

) , , , ( ,
1 0
k k pq pq pq
or y x q p = + = (10)

Here,
0
pq
denotes the stress in an uncracked half-space subjected to the rolling-sliding
contact loading and frictional heat generation in the contact region. The stress
1
pq

denotes the stress disturbance induced by the crack.
The solution of the stress
0
pq
which satisfies the boundary conditions Eqs.(2)-(6) is
represented as follows
(14)
.

> +
< +

=

1
1
1
1
1
1
1
1
1
0
0
1 , ) ( ) (
1 1 , ) ( ) (
1 , ) (
x t d F t P t d G t P
x t d F t P t d G t P
x t d F t P
P
pq pq
x
pq pq
pq
pq

(11)

where,

/ } ) {( } ) ( ) 1 2 ( {
} ) {( ) 2 (
2 2 2
0
75 . 0 2 2 5 . 0
0

+ + +
+ =
y t x B f t x S H y
y t x P A S f H F
pq r
e pq r pq
(12)

( ) ( )
) ( 4 / 5 . 2 5 . 0
2
5 . 0
t x y P
e pq r pq
e
e t x P D S f G

= (13)

Here,
pq
A ,
pq
B and
pq
D are given by Goshima, et al.
(14)
.


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1084
In the present study, the contact pressure is assumed to be Hertzian contact as:


2
1 ) ( t t P = (14)
Then, the integrations in Eq.(11) can be calculated numerically
(15)
.
To account for the stresses caused by the crack, we consider the problem of a discrete
edge dislocation
j
present at the point
j
z z
0
= ) (
j j
= on the jth kinked crack in an
infinite space. Then the dislocation density is defined as:

| | | | { }
j
j j j j
i
j
e U i U
c i 4
G



+

=
) 1 (
0
, (j=1,2,......,n) (15)

where ]} [ ] {[
j j j j
U i U

+ represent the displacement jumps, and
j
z
0
is represented as
follows.

= +
= + +
=

=
1
1
1 1 1
0
) ,......., 3 , 2 ( ,
) 1 ( ,
1
j
m
i
m
i
j
i
j
n j e l e
j y i x e
z
m
j

(16)

The solution to this problem is solved using the following complex potential
functions
(16)
.


j
j
z z
z
0
2
) (

=

(17)


2
0
0
0
2
) (
) (
j
j j
j
j
z z
z
z z
z

=

(18)

Then the stress representation is given by Muskhelishvili
(17)
as:

j
j j j j
i
d d
e z z z z z i




2
2 2 2 2
,
} ) ( ) ( { ) ( ) (
2 2
+ + + =
|
.
|

\
|
(19)

where prime denotes the differentiation with respect to z. An additional potential
3
, which
is required to remove the surface tractions, is conveniently written in terms of
2 2
,
(17)
.

>
<
=
0 ) Im( , ) (
0 ) Im( , ) ( ) ( ) (
) (
2
2 2 2
3
z z
z z z z z
z

(20)

Then the stress representation is written as

:

j
j j j j
i
d d
e z z z z z z z i

2
3 3 3 3 3
} ) ( ) ( ) ( ) ( { ) ( ) (
3
+ + =
|
.
|

\
|
(21)

Superposing Eqs.(19) and (21), the stress field due to a discrete edge dislocation
j
in a
half space is represented in coordinates ) , (
j j
as:

3 2 2
,




|
.
|

\
|
+
|
.
|

\
|
=
d d d d d d
j j j j j j j j j j j j
i i i (22)


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1085

Replacing
j
by distributed dislocation density defined along the each kinked cracks
(j=1,2,.....,n), the stress
1
pq
can be obtained by using Eq.(22) as a Greens function.

=
)
`

=
n
j
l
j
d d
j
j j j j k k k k
d i i
1
0
1 1
) (

(23)

Superposing Eqs.(23) and (11) as Eq.(10), the stress solution
pq
which satisfy the
boundary conditions Eqs.(2)-(6) is obtained. After the coordinate transformation of
pq

along the each kinkeded coordinates ) , (
k k
, substituting these results the boundary
conditions Eqs.(7)-(9), the following singular integral equations for
k
(k=1,2,........,n)
are given.

( )
k
l
0
k k
k k i
d e 2
k
k

( ) ( ) ( ) ( ) { }
j
n
j
l
0
j k k j j k k j j
d , ,
j

=
+ +
1

( )
0 =
=
k
k k k k
0 0
i


, (k=1,2,.......,n) (24)

0 ) (
1
0
=

=
n
j
l
j j j
j
d (25)

where, kernel functions ( ) ( )
j k k j k k
, , , in Eq.(24) are given as:

) , ( ) 1 ( } ) , ( ) (
) , (

{ ) , ( ) 1 ( ) , (

) , (
1 0
*
3
0 3
2
0
*
3
2
0 3
j k k kj j k k k
j k
i
j k
i
j k j k k
L z z z z
z z e z z e z z
k k



+

+ =
(26)
) , ( ) 1 ( } ) , (

) (
) , ( ) , (

{ ) , (

) , ( ) , (
2 0
*
3
0
*
3 0 3
2
0 3 0
*
3
j k k kj j k k k
j k j k
i
j k j k j k k
L z z z z
z z z z e z z z z
k



+

+ + =
(27)
k
i
j k j k j k k
e z z z z L


2
0
*
2 0
*
2 1
) , ( ) , ( ) , ( + =

k
i
j k j k k j k j k k
e z z z z z z z L


2
0
*
2 0
*
2 0
*
2 2
} ) , ( ) , ( { ) , ( ) , ( +

+ =

=
) ( 1
) ( 1
) , (

0
0
0 3
z z
z z
z z
,
,

0 ) Im(
0 ) Im(
>
<
z
z

=
0
) ( ) (
) , (
2
0 0 0
0
*
3
z z z z
z z
,
,

0 ) Im(
0 ) Im(
>
<
z
z


) ( 1 ) , (
0 0
*
2
z z z z = ,
2
0 0 0
*
2
) ( ) , ( z z z z z =

= + =
= + + =

1 - k
2 = m
i -
m
i -
k k
i
1 1
n ,....., , k , e l e z
k , y i x e z
m k
k
3 2
1
1 1

=
=
j k
j k
j k
, 0
, 1

j k
j k

=



Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1086
4. Stress Intensity Factors and the Numerical Calculations

4.1 Derivation of Stress Intensity Factors
Equations (24),(25) were solved numerically using the piecewise quadratic method of
Gerasoulis
(18)
. The each dislocation densities ) (
k k
are written with separating the
singularities as

1
2
,
) 1 (
) (
) (
2 / 1
0
=

=
k
k
k
i
k
k k
k k
l
P
k
e

(28)

Let us divide the interval 1 1
k
into 2N
k
equal parts. We define the nodal points
as
s k,
) 1 2 ~ 1 ( + =
k
N s , and use the Lagrange interpolation formula for three nodal
points in the approximation. Setting the collocation points as
k r k r k
N 2 1

, ,
+ =
) 2 ~ 1 (
k
N r = , Eqs.(24), (25) reduce to the simultaneous linear algebraic equation for
) (
,s k k
. Here,
op
k
in Eq.(8) are determined iteratively the degree of crack opening for
a given set of parameters. Iteration was performed under the condition of the absence of
overlap of the material as 0 <
k k
U

. In the present numerical examples, this method
generally converged within three iterations. For the number of collocation points, a good
accuracy was obtained for N
k
=10. At that time, mode I and mode II stress intensity factors
at the n-kinked crack tip are given as follows.

) 1 ( 2
0 n n II I
l c P K i K = (29)

Although the maximum mode I stress intensity factor
max I
K , the maximum and
minimum mode II stress intensity factors
max II
K and
min II
K occur at different locations,
stress intensity factor range
max I I
K K = ,
min max II II II
K K K = over a complete loading
cycle are used for fatigue consideration. Especially, in the present study, as a mixed mode
fatigue propagation should be considered, energy release rate G (Pa.m)over a complete
loading cycle is defined as follows.

{ }
2 2
0
) ( ) (
2
1
II I
K K
G
G

= (30)

4.2 Numerical Examples of Stress Intensity Factors for a Single Kinked Crack
As the most simple example, the stress intensity factors for a single kinked crack from
a horizontal initial crack ) 0 , 180 (
*
1 1
= = as shown in Fig.2. Numerical calculations
are carried out for the case of c = 0.4mm,
P
e
=100 and P
0
=2.5GPa being Hertzian contact
pressure distribution. The material used
here is a high carbon-chromium bearing
steel (AISI-52100). The Poissoin's ratio
is 3 . 0 = and Young's modulus is
E
0
=222.17GPa
(19)
, hence shear modulus is
G
0
=85.45GPa. As the dimensionless
parameter H
0
in Eq.(12) is almost equal to
unity for most metallic materials
(20)
, we
assume H
0
=1. From the result of fatigue
experimental test as shown in Fig.5, we
assumed the threshold value of the stress
intensity factor is K
Ith
=4MPa.m
0.5
, hence
the threshold value of energy release late is
*
2

1
~
l
2
~
l
cont act surface
horizont al
init ial crack
h
~

Fig.2 Single kinked crack model



Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1087

0
2 2
/ ) 1 ( E K G
Ith
th
= =65.54 Pa.m (31)

Actually, before the crack propagation, the crack initiation should be considered. For
example, Fan et al.
(21)
estimated the crack initiation life by using the Mura-Tanaka crack
initiation model. However, in the present study, it is assumed that the initial crack length
can be determined by the criterion of
th
G G . Thus , the horizontal initial crack length
1
~
l = 100m, 80m and 60m can be obtained for the initial crack depth h
~
=100m,
30m, and 10m, respectively, for the case of f=0.1, S
r
=0.1. Here, it is assumed that the
kinked crack length
2
~
l =10m. Figures 3 and 4 show the effect of kinked angle
*
2
on the
behavior of stress intensity factors K
II
over a complete loading cycle for f=0.1 and 0.7
respectively, for the case of h
~
=100m and S
r
=0.1. In all these cases, as K
I
were almost
equal to zero, mode II crack growth seems to be predominant. In Fig.3,
II
K
(=
max II
K -
min II
K ) takes a maximum value for the case of =0
*
2
, therefore the
initial crack start to propagate in the horizontal direction.











Fig.3 Stress intensity factor K
II
variations Fig.4 Stress intensity factor K
II
variations
with
*
2
for the case of f=0.1 with
*
2
for the case of f=0.7

5. Prediction of Fatigue Crack Growth Path and Life until Surface Failure

In order to clarify the mechanism of tribological failure such as shelling or pitting etc.,
extending the above calculation for n=2 to for n>2, we consider the behavior of the fatigue
crack growth path and fatigue life due to a repetition of the rolling-sliding contact. As
mentioned above, it is assumed that the initial crack is parallel to the contact surface
) 180 (
1
= and the crack length are
1
~
l = 100m, 80m and 60m for h
~
=100m,
30m and 10m, respectively. In the prediction of crack growth path, it is assumed that the
crack growth increment
k
l l
~ ~
= =10m (k=2,3,......,n) are inclined in the direction of
G
k
k
*
max
*
= at which G in Eq.(30) takes a maximum value. Thus, getting the value of
G
k
*
max
for each steps (k=2,3,.......,n), we can predict the crack growth path starting from
the horizontal initial crack. In order to predict the propagation life, in Fig.5 the experimental
crack growth behavior ('a' is a crack length) is shown by rotating bending fatigue test under
the applied stress amplitude of 1450 and 1500MPa for a high carbon-chromium bearing
steel (AISI-52100). From Fig.5 it is assumed that the threshold value is about
th
K = 4
MPa.m
0.5
. Within the range of stable crack growth in Fig.5, the experimental results are
arranged as Fig.6, by making use of the least squares method. From this experimental result,
in the present study, the following crack growth low is assumed.

<
>
=
th
th th
K K
K K K K C
N d
l d



, 0
, ) (
~
0
(32)

Where, l
~
is the total crack length , N is the cycle number and K is a stress intensity
factor range ( K = K
max
-

K
min
). From Fig.6, = 1.377 , C
0
=1.678810
-10
m
(1- /2)
/

(MPa)

-3 -2 -1 1 2 3
-4
-2
2
4
0
=60
=30
=0
=30
=60
x
K

, MPa.m
0.5
1
*
2

*
2

*
2

*
2

*
2

-3 -2 -1 1 2 3
-4
-2
2
4
0
=60
=30
=0
=30
=60
K

, MPa.m
0.5
x
1
*
2

*
2

*
2

*
2

*
2

Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1088
are obtained. Substituting K = [G E
0
/ (1-
2
) ]
0.5
into Eq.(30), the following crack growth
law is obtained using the energy release rate range G.

<
>
=
th
th th
G G
G G G G C
N d
l d



, 0
, ) (
~
1
(33)

Where,
th
G is given by Eq.(31) and C
1
is given as:

2 /
2
0
0 1
1

|
|
.
|

\
|

=
E
C C = 6.35910
- 11
m/(MPa.m)
/2
(34)

10
0
10
1
10
2
10
-10
10
-9
10
-8
10
-7
K , MPam
1/2
d
a
/
d
N

,

m
/
c
y
c
l
e
:1500MPa
:1450MPa

-1 0 1 2
-10
-9
-8
-7
y=an x
n
a0=-9.77477534e+00
a1=1.37707360e+00
l
o
g
(
d
a
/
d
N
)
log(K -K
th
)

Fig.5 Crack growth behavior by rotating Fig.6 The least squares method in the range
bending fatigue test for AISI-52100. of stable crack growth in Fig.5.

Corresponding to the crack growth path as mentioned above, the cycle number increment
k
N for
k
l l
~ ~
= can be obtained by using Eq.(33). Thus, we can obtain the variation of
crack length l n l l
~
) 1 (
~ ~
1
+ = with the cycle number
n
N N N N + + + = ........
3 2

until n-steps (hereafter, we call this relation N l
~
curve).
As a first example, for the case of the very shallow horizontal initial crack of the depth
h
~
=10mthe frictional effect (frictional coefficient f ) and thermal effect (sliding/rolling
ratio S
r
) on the crack growth path until the contact surface are shown in Figs.7a) and 7b)
respectively. Then, the corresponding N l
~
curves are shown in Figs.8a) and 8b),
respectively. From Figs.7a) and b), the crack growth path until the tribological failure such
as the very thin shelling or delamination, do not change very much by the frictional or
thermal effects. However, from Figs.8a) and b), the frictional or thermal effects reduce the
fatigue lives. Especially, from Fig.8a), an increase of frictional coefficient f extremely
reduces the fatigue life. For example the fatigue life changes from N
f
=1.8310
4
for f=0.1 to
N
f
=4.5610
2
for f=0.7 for the case of S
r
=0.1.











Fig.7 a) Crack growth path showing Fig7 b) Crack growth path showing
the frictional effects for 10
~
= h m the thermal effects for 10
~
= h m

0 20 40 60 80 100 120
0
20
40
60
80
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
Width, m
D
e
p
t
h
,

m
f=0.1
Initial crack-tip
0 20 40 60 80 100 120
0
20
40
60
80
f=0.1
f=0.3
f=0.5
f=0.7
Width, m
D
e
p
t
h
,

m
S
r
=0.1
Initial crack-tip


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1089












Fig.8 a) N l
~
curve showing the Fig.8 b) N l
~
curve showing the
frictional effects for 10
~
= h m thermal effects for 10
~
= h m

Figures 9a) and 9b) display the crack growth path from a horizontal initial crack of the
depth h
~
= 30m. And Figures 10a) and 10b) show the corresponding N l
~
curves,
respectively. The tendency of frictional and thermal effects on the crack growth path and
fatigue life are qualitatively similar to the case of h
~
=10m as mentioned above. However,
in quantitatively, the size of the failure is larger and the lives are longer than the case of
h
~
=10m. From Figs.9a), frictional effect cause a little early rising up the crack growth path
toward to the contact surface. And, from Fig.10a), an increase of frictional coefficient f
reduces the fatigue life. For example fatigue life changes from N
f
=7.4610
4
for f=0.1 to
N
f
=6.7010
3
for f=0.7 for the case of S
r
=0.1.










Fig.9 a) Crack growth path showing Fig9 b) Crack growth path showing
the frictional effects for 30
~
= h m the thermal effects for 30
~
= h m











Fig.10 a) N l
~
curve showing the Fig.10 b) N l
~
curve showing the
frictional effects for 30
~
= h m thermal effects for 30
~
= h m

While, for the case of the relatively deep horizontal initial crack of the depth
h
~
=100mthe frictional effect and thermal effect on the crack growth path until the contact
surface are shown in Figs.11a) and 11b) respectively. Then, the corresponding
N l
~
curves are shown in Figs.12a) and 12b), respectively. From Fig.11a), the crack
growth path until the tribological failure such as the micro-pitting, are affected significantly
by the frictional coefficient, an increase of frictional coefficient cause the early rising up
crack growth path toward to the contact surface. From Fig.11b), an increase of
sliding/rolling ratio cause only a little difference in the crack growth path toward to the
10
0
10
1
10
2
10
3
10
4
10
5
0
50
100
150
N, cycle
C
r
a
c
k

L
e
n
g
t
h
,

m
f=0.1
f=0.3
f=0.5
f=0.7
S
r
=0.1
10
0
10
1
10
2
10
3
10
4
10
5
0
50
100
150
N, cycle
C
r
a
c
k

L
e
n
g
t
h
,

m
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
f=0.1
0 50 100 150
0
20
40
60
80
100
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
f=0.1
Width, m
D
e
p
t
h
,

m
Initial crack-tip
0 50 100 150
0
20
40
60
80
100
f=0.1
f=0.3
f=0.5
f=0.7
S
r
=0.1
Width, m
D
e
p
t
h
,

m
Initial crack-tip
10
0
10
1
10
2
10
3
10
4
10
5
0
100
200
f=0.1
f=0.3
f=0.5
f=0.7
S
r
=0.1
N, cycle
C
r
a
c
l

L
e
n
g
t
h
,

m
10
0
10
1
10
2
10
3
10
4
10
5
0
100
200
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
f=0.1
N, cycle
C
r
a
c
k

L
e
n
g
t
h
,

m


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1090
contact surface. From Fig.12a), an increase of frictional coefficient f causes a little decrease
of fatigue life. For example fatigue life changes from N
f
=1.5810
5
for f=0.1 to
N
f
=5.2610
4
for f=0.7 for the case of S
r
=0.1. However, from Fig.12b), the fatigue life is
not affected very much by the change of sliding/rolling ratio (thermal effect). This tendency
by the thermal effect is depend on the fact that the surface temperature caused by the
friction abruptly decrease along the depth. Moreover, from Figs.8, 10 and 12, a decrease of
initial crack depth (shallow crack) makes a greater reduction in life for any cases.













Fig.11 a) Crack growth path showing Fig11 b) Crack growth path showing
the frictional effects for 100
~
= h m the thermal effects for 100
~
= h m













Fig.12 a) N l
~
curve showing the Fig.12 b) N l
~
curve showing the
the frictional effects for 100
~
= h m the thermal effects for 100
~
= h m

6. Summery
Stress intensity factors are calculated for a multiple kinked subsurface crack in a semi
-infinite body due to rolling-sliding contact with frictional heat. On the basis of theses
results, the crack growth from a horizontal initial crack until the contact surface are
simulated by maximum energy release rate criterion, and the fatigue growth lives are
predicted by the crack growth law obtained by the fatigue experimental results, for a high
carbon chromium bearing steel (AISI-52100), for the case of the horizontal initial crack
depth h
~
= 10m, 30m, 100m, frictional coefficient f= 0.1, 0.3, 0.5, 0.7, and
sliding/rolling ratio S
r
= 0.1, 0.3, 0.5, 0.7, respectively. For the case of the very shallow
horizontal initial crack, the frictional effect (frictional coefficient f ) and the thermal effect
(sliding/rolling ratio S
r
) on the crack growth path until the contact surface , do not change
very much by the frictional or thermal effects. However, the frictional or thermal effects
reduce the fatigue lives. Especially, an increase of frictional coefficient extremely reduces
the fatigue life. While, for the case of the relatively deep horizontal initial crack, an increase
of frictional coefficient cause the early rising up crack growth path toward to the contact
surface, but an increase of sliding/rolling ratio cause only a little difference in the crack
growth path toward to the contact surface. And the fatigue life decrease a little by an
increase of frictional coefficient, but the fatigue life is not affected very much by the change
of sliding/rolling ratio. Moreover, a decrease of initial crack depth (shallow crack) makes a
greater reduction in life for any cases.

10
0
10
1
10
2
10
3
10
4
10
5
10
6
0
100
200
300
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
N, cycle
C
r
a
c
k

L
e
n
g
t
h
,

m
f=0.1
10
0
10
1
10
2
10
3
10
4
10
5
10
6
0
100
200
300
N, cycle
C
r
a
c
k

L
e
n
g
t
h
,

m
f=0.1
f=0.3
f=0.5
f=0.7
S
r
=0.1
0 100 200
0
50
100
150
Width, m
D
e
p
t
h
,

m
f=0.1
f=0.3
f=0.5
f=0.7
S
r
=0.1
Initial crack-tip
0 100 200
0
50
100
150
S
r
=0.1
S
r
=0.3
S
r
=0.5
S
r
=0.7
D
e
p
t
h
,

m
Width, m
f=0.1
Initial crack-tip


Journal of Solid Mechanics
and Materials Engineering

Vol. 2, No. 8, 2008
1091
References
(1) Suh, N.P., The Delamination Theory of Wear, Wear, Vol.44 (1977), pp.1-16.
(2) Fleming, J.R. and Suh, N.P., Mechanics of Crack Propagation in Delamination Wear,
Wear, Vol.44 (1977), pp.39-56.
(3) Rosenfield, A.R., A Fracture Mechanics Approach to Wear, Wear, Vol.61 (1980),
pp.125-132.
(4) Hearle, A.D. and Johnson, K.L., Mode II Stress Intensity Factors for a Crack Parallel to
the Surface of an Elastic Half-Space Subjected to a Moving Point Load, Journal of
Mechanics and Physics in Solids, Vol.33 (1985), pp.61-81.
(5) Sheppard, S., Barber, J.R. and Comninou M., Short Subsurface Cracks under Conditions
of Slip and Stick Caused by a Moving Compressive Load, Transactions of ASME, Journal
of Applied Mechanics, Vol.52 (1985), pp.811-817.
(6) Salehizadeh, H. and Saka, N., Crack Propagation in Rolling Line Contacts, Transactions
of ASME, Journal of Tribology, Vol.114 (1992), pp.690-697.
(7) Goshima, T. and Soda, T., Stress Intensity Factors of a Subsurface Crack in a
Semi-Infinite Body due to Rolling/Sliding Contact and Heat Generation, JSME
International Journal, Ser.A, Vol.40, No.3 (1997), pp.263-270.
(8) Goshima, T., Ishihara, S., Shimizu, M. and Soda, T., Mutual Interference of Two
Subsurface Cracks in a Semi-Infinite Body Due to Rolling Contact with Frictional Heating,
JSME International Journal, Ser.A, Vol.44, No.4 (2001), pp.542-549.
(9) Farris, T.N., Keer, L.M. and Steele, R.K., The Effect of Service Loading on Shell
Growth in Rails, Journal of Mechanics and Physics in Solids, Vol.35,No.6 (1987),
pp.677-700.
(10) Yu, M. M. H. and Keer, L. M., Growth of the Shell/Transverse Defect in Rails,
Transactions of ASME, Journal of Tribology, Vol.111 (1989), pp.684-654.
(11) Mukai, D.J., An Analysis of a Near-Surface Crack Branching Under a Rigid Indenter,
Transactions of ASME, Journal of Tribology,Vol.122 (2000), pp.23-29.
(12) Goshima, T., Ishihara, S., Shimizu, M., Yoshida, H. and Tsuchida, Y., Surface Crack
Growth Path and Fatigue Life Prediction Due to Repeated Rolling/Sliding Contact, JSME
International Journal, Ser.A, Vol.46, No.4 (2003), pp.582-589.
(13) Wu, C.H., Fracture Under Combined Loads by Maximum-Energy-Release-Rate
Criterion, Transactions of ASME Journal of Applied Mechanics, Vol.45 (1978), pp.553-558.
(14) Goshima, T and Keer, L.M., Stress Intensity Factors of a Surface Crack in a Semi -
infinite Body due to Rolling-Sliding Contact and Frictional Heating, Transactions of JSME,
(in Japanese), Ser.A, Vol.56, No.532 (1990), pp.2567-2572.
(15) Abramowitz, M. and Stegun, I. A., Handbook of Mathematical Functions, (1970),
p.889, Dover Publications, Inc., New York.
(16) Dundurs, J., Mathematical Theory of Dislocations, (1975), p.70, ASME Publication.
(17) Muskhelishvili, N.I., Some Basic Problems in the Mathematical Theory of Elasticity,
4
th
Ed, (1954), p.141, p.473, Noordhoff .
(18) Gerasoulis, A., The Use of Piecewise Quadratic Polynomials for the Solution of
Singular Integral Equations of Cauchy Type, Comput. Math. Applics., Vol.8 (1982),
pp.15-22.
(19) JSME, The Modulus of Elasticity of Metals and Alloys (in Japanese), (1980), p.69,
JSME Publication.
(20) Azarkhin, A., Barber, J.R. and Rolf, R.L. Combined Thermal-Mechanical Effects in
Frictional Sliding, Key Engineering Materials, Vol.33 (1989), pp.135-160.
(21) Fan, H., Keer, L.M. and Mura, T., Near Surface Crack Initiation Under Contact Fatigue,
Tribology Transactions, Vol.35, No.1 (1992), pp.121-127.