Beruflich Dokumente
Kultur Dokumente
ELSEVIER
Komvopoulos
Abstract
A linear elastic fracture mechanics analysis of plane-strain indentation of a homogeneous half-space with a subsurface horizontal crack was
performed using the finite element method. Stress intensity factor results obtained for an infinite plate with a central crack subjected to farfield tension and a half-space with a frictionless subsurface horizontal crack under a moving surface point load are shown to be in good
agreement with corresponding analytical results. Crack mechanism maps illustrating the occurrence of separation, forward and backward slip,
stick, and separation at the crack interface are presented for different indentation load positions and crack face friction coefficients. Results
for the stresses in the vicinity of the crack tips and the mode I and mode II stress intensity factors are given for different indentation positions,
crack face friction coefficients, and both concentrated and distributed surface normal tractions. Although indentation produces a predominantly
shear and compressive stress field, mode I loading conditions are shown to occur for certain indentation positions. However, the magnitude
of the mode I stress intensity factor is significantly smaller than that of mode II, suggesting that in-plane shear mode crack gmwth is most
likely to occur in the absence of microstructural defects. The significance of crack face friction and sharpness of the indenter on the subsurface
shear mode crack propagation rate is interpreted in terms of the mode II stress intensity factor range and material behavior.
Keywords: Linearelastic fracturemechanics;Finiteelement method;Stressintensityfactor;Subsurfacecracks;Crack face mechanisms;Friction;Indentation
loading
1. Introduction
lO
2. Modeling procedures
2.1. Statement of the problem
xp
E=O
J
Fig. I. Schematic diagnun of a subsurface horizontal crack in a half-space
subjected to concentrated and distributed indentation loads.
Crock Mechonisrns
Forwotd Slip
Bockword Slip
Seporolion
,',',;;',',',',',;"
Stick
V'lllllllllLIll
lllllllIlll
I/d POintS
II
Crock TiD
I ] I
87 . . / . /
II
''
(o)
I lilllll
Ill lill
II
Crock Tip
Fig. 3. Crack tip isoparamotric quarter-point elements: ( a ) quadratic element
and (b) triangul~ element formed by collapsing one side of the quadratic
element shown in (a).
(b)
Fig. 4. Finite element mesh for modeling subsurface cracking in an indented
half-space: (a) detail of mesh showing elements around the crack and (b)
12
-c<_x<_c; y = l
( I a)
S(x) = f ~ / ( x ) s g n ( d x / d t ) (slip)
-c<x<c;
y=l
(lb)
-c~x<c;
y=l
(Ic)
nodes of the deformable medium at which the contact pressure is zero, andp* is the pressure at zero clearance. Thus, if
the selected value of p* is of the order of the expected contact
pressure, e.g. the maximum Hertzian pressure, the computational time is reduced significantly. Throughout the present
analysis, the parameters c* and p* were fixed at 50 nm and
5 GN m - 1, respectively.
2.4. Solution method
The finite element code ABAQUS and an updated Lagrangian formulation were used to model subsurface cracking in
an indented elastic half-space. Since the objective was to
study subsurface cracking under indentation loading without
considering the deformation effect from previous indentations, the unloaded finite element mesh configuration was
used in all the simulations. Indentation by a concentrated
force was modeled by applying incrementally a load of
i N m - ~ at different nodal points of the contact surface BC,
while constraining the x and y displacements of the nodes of
boundary AD. The nodes of the left and right boundaries of
the mesh, AB and CD, respectively, were not constraint. A
force convergence tolerance equal to a very small fraction of
the load ( 10 -7 N for f~ = 0, and 10- s - 1 0 - 3 N for fc = 0.3
and 0.5) was used to solve the equilibrium equations at each
increment. The modeling of indentation by a rigid indenter
was based on a displacement-controlled procedure similar to
that used in previous normal contact mechanics analyses
[ 14,15 ]. Thus, indentation was simulated by applying a small
incremental vertical displacement of 50 nm to all the nodes
of boundary AD, while at the same time constraining their
horizontal displacements and fixing the rigid indenter to the
desirable position. The advantage of using this method is that
the contact pressure is determined during the solution process,
thus eliminating the need to assume a certain contact pressure
profile. The force convergence tolerance for modeling indentation by a rigid indenter was significantly smaller than the
anticipated nodal forces at the contact interface, in the range
of 0.01-10 N. The maximum number of load or displacement
increments was 30, and in each increment a maximum of 6
to 15 iterations was allowed.
k+l I_ v ~ J
(2a)
Crock
,ll
[.~
a
it, i
Kn-- 2 G 2 ~ [-4u~ - uc ]
k+ I L----~--~J
(2b)
where ux denotes displacements in the x direction.Alternatively,the values of Km and Kn can be calculated from the
displacements of near crack tip nodes using a linearextrapolation technique [ 17 ]. Thus, from the calculateddisplacements u* and u* at severalnodes (r,0) close to the crack tip,
the corresponding apparent stress intensityfactors~ and
can be obtained from the following expressions [ 18]
G
2"~
LLL--U) v r -
K* = ~.-77-----~.
~/~-u * (r,O)
=~V'~u*(r,O)
(3a )
(3b)
27~
,=~V"~Btu,,(rn,~r)--u,(rB,
--'rr) l
KH-4(--('G ~_v)V"~n[Ux(rn,'tr)-u~(rn,
-~)1
K
(4a)
(4b)
(5a)
r~O
K. = lira2V/2-'~r
~'~y(r,0)
(5b)
r~0
13
P 2
J=.~=T[Kf +Kf,]
(6)
The present finite element code yields estimates of the 3integral for closed paths around the crack tip. Thus, for the
simple case of mode I loading, the finite element results for
the J-integral can be substituted into Eq. (6) to determine the
stress intensity factor, i.e. KI = [JE/( 1 - u 2) ] i/2. While the
above displacement and stress methods may be used for either
constrained or unconstrained crack tip nodes, in the J-integral
method it is required that the crack tip nodes are free to move
independently. From the computational standpoint, the methods based on Eqs. (2a), (2b), (4a), (4b) and (6) have the
advantage that calculation of stress intensity factors does not
depend on an extrapolation technique.
Ki=o'~cg(clw)
(7)
14
Table I
Error between finite element and analytical results of mode I stress intensity
factor
Approach
ErrorlK,/cr~c }
Constrained nodes
Free nodes
Displacementmethod
Eq. (2a)
Eq. (3a)
Eq. (4a)
0.038
0.022
0.016
0.049
0.026
0.017
Stressmethod(Eq. (5a))
Energymethod(Eq. (6))
0.005
-
0.000
0.029
cially for unconstrained crack tip nodes, is in excellent agreement with the exact solution. The largest error of about 4.9%
is obtained with unconstrained nodes using Eq. (2). The
generally small effect of the multipoint constraint on the stress
intensity factor is probably because the differences in the
displacements of the crack tip nodes are very small in the
elastic range when the nodes are free to move independently.
Hence, it may be concluded that the chosen crack tip elements
predict fairly accurately the local stress field and that the
predictions of all methods are reasonably close to the exact
result.
To further examine the suitability of the finite element
mesh and the appropriateness of the invoked modeling
assumptions under loading conditions similar to those of the
present study, the problem of an elastic half-space with a
frictionless horizontal crack under a compressive concentrated load was analyzed using the present finite element
model .--'t the analytical method of Hearle and Johnson [ 5 ].
Finite element and analytical results showing the variation of
the mode II stress intensity factor, Kul (2P/,rrv/~), at the left
crack tip with the load position are compared in Fig. 6, The
finite element data were obtained using Eq. (2b). The figure
shows that the agreement between the solutions of the two
approaches is very good. The stress field at the left crack tip
0.6
Concentrated Load
A
~
0.4
0.2
{~
-0.2
-04-3
2c/l=5
f~ =o
Hearle 8~ ~
Johnson (~g~,'3i-"~a~
I
-2
-1
o
Xp/C
.I.
The significance of friction on the interracial crack mechanisms can be interpreted qualitatively based on the deformed
mesh around the crack. Fig. 7 shows magnified deformed
meshes of the crack domain produced from indentation by a
concentraPed load over the left crack tip (xelc= - 1) for
different crack face friction coefficients. Fig. 7 ( a ) shows that
forward slip of the crack faces increases toward the left crack
tip whenf( = 0. However, Fig. 7(b) and (c) demonstrate that
the tendency for forward slip near the left tip diminishes as
the coefficient of friction increases. As a matter of fact, the
deformed elements above and below the crack interface
shown in Fig. 7(c) indicate that sticking near the left tip
occurs whenf~=0.5. Crack face interaction at the right tip
appears to be less sensitive to frictional variations when the
indentation occurs over the left tip.
The occurrence of the various crack mechanisms for different indentation load positions can be elucidated in light of
the stress results obtained from the interface elements at the
crack interface. Fig. 8 shows that for(c=0.5, two or three
different crack mechanisms may arise concomitantly,
depending on the position of the indentation load and the
stress condition satisfied at the crack interface (Eqs. ( l a ) ( l c ) ) . Thus, for indentation position xelc = - 1, stick and
separation occur at the left and right crack tips, respectively,
whereas forward slip commences at the central region of the
crack interface (Fig. 8(a) ). This result reveals the existence
of the mode I (tensile) crack mechanism in a predominantly
shear and compressive stress field. However, only two mechanisms are encountered when the indentation occurs at
xelc = 0; ba~:kward and forward slip at the left and right crack
tips, respectively, and stick at the center of the crack
(Fig. 8(b)). Hence, separation of the crack faces does not
(a)
- -
Deformed
.....
Original
~:lesh
Mesh
l[[iiii :
t: t::]:::t::t _It:t:::t:--I:
.
15
Is,,:~.,.
:i-f(o,
I
ro,..0s:,.
_,~o,,o.
..,::.,.
l~xlc/Pl --o--
7,,
.['~ck.o,a s,~),_
/
o~
........
r--r- ~ - - r
. . . . . . .{---'i
.
I"-
~T
stick
(b)
p'o'q
~. F~.ora slip I
x,,/c=0
---~
V,,,V.,,
-05
0-I
Fig. 8. Dimensionless
G5
=!
(a)
(c)
Fig. 7. Deformedfiniteelementmeshesof the crackdomainforconcentrated
lead positionxp/e= - I: (a) f~=O (mag. 12x 10~), (b) fc=0.3 (mag.
12 10;), and (c) fc=0.5 (mag. 13x 10"~).
occur, and the crack is subjected to pure mode II (shear)
loading.
Dimensionless stress distributions on the crack plane
(y/l= 1) are presented in Figs.9 and 10 for f=0.5 and
different indentation load positions. Schematics of associated
crack mechanisms are also shown to facilitate the interpretation of the results. As shown in Fig. 9(a), the effect of the
crack on the normal stress distribution is small. The stress
profile is fairly similar to that of an indented homogeneous
half-space, except at the right tip where, as a result of crack
face separation, the continuity of the stress distribution is
disturbed, thus giving rise to a stress singularity. The variation
of the tensile normal stress at the right crack tip for indentations at xpIc = - 2 and - I is shown in Fig. 9(b). When the
indentation load is closer to the left tip, larger tensile stresses
arise in the vicinity of the right tip. The profiles shown in
Fig. 9(b) exhibit a stress singularity at the root of the right
tip.
0
X/
'..Z...~,~_2'
i
i
i
i
,
i
xp/ =-2
,,
~3
(b)
001
-2
-i
xp~
-I
---o---
-OO]
"0020
/
02
04
O~~
O~I
LO
16
~"
(O) ~
xp/ =-2
O~
o. O)
.02
-03
-04
@~
oz
)
I
(b)
x,/c:-t
Q;
~o
~:~%
-O4~
o~
ii,
I
(c)
x,/c=O
~m
-z / (o)
'
-2.0
-1.5[-
fc=O \'~
'
-Ul
'
-o.,~
'1
.~
-2
!
-1
I
O
x/
"
L[ (b)
-1.5 V
-0.5 -
fc
-----0.3
t I s-
,x\\
'~'~"~
1.O
1.5
201
-LO
I
-0.5
|~
0
I
0.5
t.O
x/c
Fig. I 1. Regions of stick, slip, and separation at the crack interface vs.
concentratadload position: (a)fc = 0 and (b)f~ = 0.3 and 0.5.
17
1.6
2.4
3.2
4.0
0.8
I .Z
1.6
Crack
<..8
5.6
iio ..o
~'.0
I".
(a
,
1'6
::o
II- -G ~
"
c,nckl
0.3
0.6
0.9
1.2
2
3
4
0.4
0.6
0.8
6
7
8
9
I0
1.2
1.4
1.6
1.8
2.0
Crock
1.5
(b)
o.a
L2
1.6
2.0
Crock
2.4
2.8
3.2
3.6
10 4.0
1.0
(c)
Fig. 12. Contoursof the yon Misesequivalentstressat the left and rightcracktips for concentratedloadpositionxp/c ffi - I: (a)f~ = 0o (b)f~ffi0.3, and (c)
fc = 0.5. ( Stressunitsare MPa.)
friction reduces the contribution of the tensile mode, thereby
promoting the dominance of shear mode crack growth. The
results presented in Fig. 11 demonstrate the significance of
the indentation load position and magnitude of inteffacial
friction coefficient on the crack growth rate during indentation loading.
Isostress contours of the van Mises equivalent stress in the
vicinity of the crack tips are presented in Fig. 12 for different
crack face friction coefficients and indentations over the left
crack tip. The magnitudes and gradients of the stresses at the
left tip are greater than those at the right tip, especially for
f~=O (Fig. 12(a)). A comparison of the stress data shown
in Fig. 12(a) and (c) shows that increasing the friction coefficient from 0 to 0.5 reduces the maximum stress at the crack
tips by approximately 50%.
The dimensionless mode II stress intensity factor at the left
and right crack tips vs. indentation load position is shown in
Fig. 13 forf~ = 0 and 0.5. The Kn data were calculated based
on Eq. (2b). For f~=O, Ku at the left crack tip increases
rapidly as the load approaches the crack, reaching a maximum
value of 0.51 at xp/c = - 1.33. Alternatively, Ku at the right
tip increases at a lower rate because crack fac,.~separation at
the right tip diminishes gradually (Fig. 11 (a)). For "~nden-
18
0.6
,ro~. t
0.4
Table2
DimensionlessmodeI and mode11stressintensityfactorsat the rightcrack
tip vs. indemationloadposit~onandfrictioncoefficientat thecrackinterface
Coneentrated2c//=3Lo0d
f~
:'~-0.2
Crack Tip
Left Right
--.o--.---~
-o,
""=a'"
.
f
o
o.5
~.~=.o.o,o.-~.~t"
~
. . ~ / "
I
-1
-~
0.0
0.3
0.5
z~"
~ ~'
~,
1
/
V
o
xp~
-I ~..~,,~
Posilive Shear
0,5
0.5
Negative Shear
1.0
~Io
-o~
I
I
0.5
lo
X/C
K11(2PI1ryr~)
KnI(2PI~V~)
xplc= - 2
xj,Ic= - I
xelc = - 2
xelc = - 1
0.0112
0.0093
0.008 !
0.0380
0.0180
0.0085
0.0517
0.2057
a.00,55
0.1262
Indentations by the rigid indenter revealed that the maximum pressure at the center of the contact region is in the
range of 1.28-1.30 GN m -z and the contact width is equal
to about 0.5c, independent of the indenter position relative to
the crack. The relatively small size of the contact width may
be considered to be representative of a semi-infinite halfspace having a relatively long subsurface crack.
Fig. 15 shows deformed mesbes of the crackdomain resulting from indentations approximately above the left crack tip
forf= = 0.5. From the crack face elements shown in Fig. 15(a)
and (b) it can be observed that indentations closer to the
crack promote the occurrence of stick at the left tip over a
larger region compared with forward slip at the fight tip. The
intensification and relaxation of the crack tip stresses as a
function of indentation position can be interpreted in light of
the stress distributions shown in Fig. 16. Similar to the crack
behavior observed in the indentation simulations involving a
point load (Fig. 9(a)), Fig. 16(a) shows that the normal
Deformed
....
Originol
0.1
Mesh
'
'
'
Mesh
-01
O" -02
-Q7
-O8
Q2
(a
~o
T
I
I
m4~
I
I
I
i
I
I
I
n~
. . . .
(c)
. . . . . . . . . . .
(11
~
I
I
~421
02
-0:
I
I
(dl
-2
-1
0
x/
indenters.
The intensification of the crack tip stresses occurring with
indentations closer to the crack neighborhood is demonstrated
by the isostress contours of the yon Mises equivalent stress
shown in Fig. 18. For an indentation at xp/c= -0.833, the
stresses at the right tip are higher than those at the left tip
(Fig. IS(a) ). However, when the indentation occurs closer
to the crack center (Xp/C = - 0.333), higher stresses arise at
theleft tip (Fig. I8(b)). From Fig. 18(a) and (b) it follows
that indentations in the range - 1 <xp/c<O have a signifi-
20
-zo:
-Ls~
rao.5
f,=o.I
,\
._~
-0.5 -
22V
-1.0
",,
-0.5
0
X/C
03
1.0
Fig. 17. Regionsof stick, slip, and separationat the crack interface vs.
indentationpositionforfc=0.5 andf,=0.1.
cant effect on the stresses at the left tip, while the effect on
the stresses at the right tip is relatively secondary. This is
further illustrated by the variation of the mode II stress intensity factor shown below. Evidently, the opposite situation
arises for indentations in the range 0 < x e l c < I.
Fig. 19 shows the variation of the dimensionless mode II
stress intensity factor at both crack tips with the indentation
position. The Kn values were calculated from the crack tip
stresses based on Eq. (5b). The rapid increase of Kn at the
left crack tip is again observed for indentations at the lefthand side of the crack ( x e l c < - 1), followed by a decrease
to lower negative Kn values for indentations over the crack
domain ( - 1 < x e l c < 1). For the same range of indentation
distances, n at the right tip increases gradually reaching
higher values when the indenter is between the left tip and
1
0
'= ~/2'rrr2
~
cos:[Ktz sin 0 + / ( . ( 3 cos 0 - 1 ) ]
lio.
2.1
2.8
3.5
,-,
5.6
6.3
10 7.0
3 2.O
3.0
4.0
5.0
6.0
7.0
8.0
9.0
I ,o I0.0
(8)
,,.o
ck
L.%.%-
3
4
2.8
8
9
I0
7.3
8.2
9.1
3.7
5 4.6
Crock
-
eI
1.9
2.8
3.7
4.6
5.5
6.4
7.3
82
9:,
Fig. 18. Contoursof the yon Misesequivalentstressat the leftand fightcracktips forfc= 0.5,f, = 0.1, and indentationposition: (a) xp/c = - 0.833 and (b)
xe/c = - 0.333. (Stressunitsare GPa.)
0.4
i
i
Distributed Lood
2c/1=3
fc=0.5
fs=O,I
0.2
~o
ff-o.2
-OA
-0.6
-3
t
-2
I
-1
I
0
Xp/C
I
1
I
2
21
Fig. 19. Variation o f dimensionless mode !! stress intensity factor at the left
(9)
4. Conclusions
Acknowledgements
This research was supported by the Surface Engineering
and Tribology Program of the National Science Foundation
under Grant No. MSS-8996309.
22
A p p e n d i x A, N o m e n c l a t u r e
a
A
c
c*
E
f
g
G
~'
k
Kt, Kn
l
n
N
P
p*
r
R
S
u~,, Uy
V
w
x, y
x~,
AKetf
AK~, A K n
0
v
References
[ 1] LR. Fleming and N.P. Suh, Mechanics of crack propagation in
delamination wear, Wear, 44 (1977) 39-56.
[2] L.M. Keer, M.D. Bryant and G.K. Haritos, Subsurface cracking and
delamination, in H.S. Chang and L.M. Kcer (eds.), Solid Contact and
Lubrication, AMD-Vol. 39, ASME, New York, 1980, pp. 79-95.
[3l A.F. Bower, The influence of crack face friction and trapped fluid on
surface initiated rolling contact fatigue cracks, J. Tribol.. I10 (1988)
704-71 I.
[4[ M.T. Hanson and L.M. Kent, An analytical life prediction model for
the crack propagation occurring in contact fatigue failure. TriboL
Trans., 35 (1992) 451--461.
[5] A.D. Hearle and K.L. Johnson, Mode II stress intensity factors for a
crack parallel to the surface of an elastic half-space subjected to a
moving point load, J. Mech. Phys. Solids. 33 (1985) 61-81.
[6] F.-K. Chang, M. Comninou, S. Sheppard and J.R. Barber, The
subsurface crack under conditions of slip and stick caused by a surface
normal fo~rce,J. Appl. Mech., 51 (1984) 311-316.
Biography
Dr. Komvopoulos is an Associate Professor in the Department o f Mechanical Engineering, University o f California at
Berkeley. He received a P h D degree in Mechanical Engineering in 1986 and MS degrees in Aeronautics a n d Astronautics and Civil Engineering in 1981, all from
Massachussetts Institute o f Technology, and a BS in Civil
Engineering from the National Technical University o f Athens in 1979. His current research activity is in tribology,
contact mechanics, tool wear, finite element modeling o f contact problems, ion beam and laser-aided surface modification
techniques, sputtering o f thin films, tribology o f magnetic
storage media, and surface engineering for reducing stiction
in microeleetromechanical systems ( M E M S ) . Dr. Komvo-
23
sented numerous seminars on tribology and contact mechanics in industry and at various universities and has delivered
key notes at the Gordon Research Conferences on Tribology.
Dr. Kotuvopoulos is an active member of the Research Comtuittee on Tribolngy of the American Society of Mechanical
Engineers (ASME), past Associate Editor of the ASME
Journal ofTribology, and is listed in Who's Who in the World,
Who's Who in Engineering, Who' s Who in Science and Engineering, and Who's Who in American Education. His most
recent experience includes consulting with IBM, HewlettPackard, Kaiser Aluminum & Chemical Co., Lucas NovaSensor, and Advanced Cardiovascular Systems, Inc.