Вы находитесь на странице: 1из 9

Inuence of temperature-dependent yield strength on

thermomechanical asperity contacts


H. Yu* S. Liu, Q.J. Wang and Y.W. Chung
Center for Surface Engineering and Tribology, Northwestern University, Evanston, IL 60208, USA
Received 2 February 2003; accepted 30 November 2003
This paper reports the eects of thermal softening on asperity contacts. We modied the thermomechanical asperity model
previously developed by Liu and Wang based on discrete convolution, fast Fourier transform and the conjugate gradient
method and applied it to the current study. The yield strength was assumed to be a function of local asperity temperature.
Asperity contact pressure and contact area with and without such temperature eects were analyzed and compared.
KEY WORDS: thermal eect, temperature-dependent, yield strength, rough surfaces in contact
Nomenclature
A
c
total contact area
A
h
contact area predicted by Hertz theory
A
p
area of the region undergoing plastic deformation

p

poyt,coyt
poyc,coyc
d distortivity, 01 i,1, m/W.
E Youngs modulus
E
*
equivalent Youngs modulus,
1
1


1i
2
1
11

1i
2
2
12
K thermal conductivity, W/m K
p compressive stress or pressure
q
f
l
f
Vp
H hardness
R the equivalent ball radius
R
d
coecient of determinant
DT surface temperature rise, DT=T)T
0
DT
h
, P
ave
maximum temperature rise, average pressure obtained
when the yield strength is temperature-dependent
DT
h
, P
ave
maximum temperature rise, average pressure obtained
when the yield strength is temperature-independent
u normal displacement
V velocity
W applied load
W
*
non-dimensional load, W
*
= 2W/(A
h
E
*
)
x
i
, x
j
coordinates of element (i, j)
Z
ij
initial surface separation of element (i, j)
Greek letters
a the rigid-body approach
l
f
friction coecient
i poisson ratio
o
y
yield strength
h thermal expansion coecient, lm/m K.
Superscripts
e elastic
t thermal
1. Introduction
In order to understand and predict failure transition
of elements in contact and relative motion, one needs
to investigate the contact performance of rough sur-
faces under mechanical and thermal loading. Depend-
ing on load, sliding speed, friction coecient and
contact geometry, one may obtain suciently high
asperity ash temperatures to cause thermal softening
of materials in contact. This reduction of surface
mechanical strength makes interacting surfaces easier
to deform and may lead to signicant plastic ow.
Such surface plastic ow, or smearing, has been
observed on surfaces of scued elements. Understand-
ing such thermally induced behavior requires an inves-
tigation of the mechanical response of the surfaces to
mechanical and thermal loading. Flash temperature
and thermomechanical deformation under the inu-
ence of thermally induced softening are the basic phys-
ical quantities to be determined in this work.
Two approaches, decoupled thermalelastic app-
roach and coupled thermalelastic approach, have
been used for ash temperature analysis. With the for-
mer, an isothermal contact problem is analyzed rst to
obtain the asperity-contact pressure. A heat-transfer
problem is then solved separately, where the heat input
is determined based on the contact-pressure infor-
mation [13]. With the latter, coupled contact, heat
transfer, and thermoelastic problems are solved inter-
actively where the thermoelastic distortion of asperities
is taken into account [410]. One can then readily
determine surface thermomechanical deformation with
this approach.
Contact pressure is limited by the yield strength of
materials in contact. The elasticperfectly plastic yield
model is widely used to limit the contact pressure
* To whom correspondence should be address. E-mail: yhlzpg@
northwestern.edu
1023-8883/04/08000155/0 2004 Plenum Publishing Corporation
Tribology Letters, Vol. 17, No. 2, August 2004 ( 2004) 155
peaks to the hardness of the softer material in contact
in order to take into account the perfectly plastic
deformation. This treatment is more reasonable than
the purely elastic contact model and is widely used in
isothermal [1113] and thermomechanical contact anal-
yses of rough surfaces [10,14]. The work done by Liu
et al. [15] indicates that the elasticperfectly plastic
yield model can predict the upper limit of contact pres-
sure in elasto-plastic contacts and is applicable if the
contact pressure and average gap are of major
concern.
Although the yield strength of many engineering
materials is temperature-dependent, the eect of ther-
mally induced softening has not been investigated in
previous thermomechanical simulations of rough sur-
faces in contact. Temperature aects the yield strength
and hence the distribution of contact pressure, which
in turn inuences deformation and asperity-contact
temperature and material yield strength. This paper
extends the thermomechanical contact model reported
by Liu and Wang [10] by including the interplay
between asperity heating and temperature-dependent
yield property of materials. This provides a more real-
istic analysis of how rough surfaces respond to thermal
and mechanical loading. For simplicity, strain harden-
ing is not considered in the current model. Two com-
mon engineering materials, 52100 steel and an
aluminum alloy 2011 T3, were studied. Results with
and without considering the thermal softening eect
are compared and discussed.
2. Problem description
The contact of two elastic bodies subjected to a
normal load and a tangential relative velocity may be
simplied into the contact between an ideally smooth
ball and a rough but nominally at surface, as illus-
trated in gure 1(a). The ball radius R is the equiva-
lent radius of the curvature found at the contact of
real engineering elements. The roughness of the half-
space surface is the composite roughness obtained
from surfaces of the two elements. Figure 1(b) shows
the deformation of one asperity of the rough surface,
which is decoupled into downward elastic deforma-
tion and upward thermal growth. Thermal softening
will enhance plastic ow due to reduced yield
strength.
Some of the basic equations used in Liu and
Wangs thermomechanical asperity-contact model [10]
are listed below for completeness. The elastic normal
surface displacement caused by the contact pressure,
px
0
1
. x
0
2
, is given by the Boussinesq formula [4]:
u
p
x
1
. x
2

1
pE

Z
1
1
Z
1
1
px
0
1
. x
0
2
dx
0
1
dx
0
2

x
1
x
0
1

2
x
2
x
0
2

2
q 1
If mechanical properties of the two contacting bodies
are identical, the normal surface displacements
caused by the frictional shear are mathematically
canceled out. Friction may be simply analyzed with
the Coulombs law. For simplicity, the observation is
concentrated on the thermal eect on the perfor-
mance of the materials, the frictional energy is
assumed to be completely converted into heat and
then to be evenly transferred between the two bodies
in contact. One half of the heat ows to upper sur-
face, and the other half ows to the lower surface.
The heat ux of the rough surface may be expressed
as a product of the frictional coecient, relative
velocity, and contact pressure between the contacting
surfaces,
q x
0
1
. x
0
2

l
f
Vp,2 2
The corresponding temperature rise in the entire half-
space, Tx
1
. x
2
. x
3
, results in a surface normal dis-
placement u
t
x
1
. x
2
. The steady-state temperature rise
on the surface of the half-space has been obtained by
Carslaw and Jaeger [16]:














(a)
(b)
Figure 1. A ball in contact with a rough surface. (a) A smooth ball
with radius R and a rough half-space. (b) An asperity subjected to
thermal and mechanical loading.
156 H. Yu et al./Inuence of temperature-dependent yield strength
Tx
1
. x
2

Z
1
1
Z
1
1
qx
0
1
. x
0
2
dx
0
1
dx
0
2
2pK

x
1
x
0
1

2
x
2
x
0
2

2
q 3
The normal surface thermoelastic displacement is
found to be [4]:
u
t
x
1
. x
2

d
2p
Z
1
1
Z
1
1
qx
0
1
. x
0
2

ln

x
1
x
0
1

2
x
2
x
0
2

2
q
lnr
0

dx
0
1
dx
0
2
4
where d is distortivity, r
0

x
2
1
x
2
2
q
is the distance
between the origin and some reference point, x
1
. x
2
.
The relative value between the observed point and the
reference point is used in a way similar to the treat-
ment for the Flamant problem [4]. The corresponding
inuence coecients for equations. (1), (3) and (4)
may be obtained through analytical integration shown
in Ref. [10].
The temperature obtained from the thermoelastic
analysis was then used to determine the yield
strength at each asperity contact position, r
y
T.
Based on the elastic perfectly-plastic model and con-
tact mechanics [4], the maximum contact pressure is
limited by the hardness H of the softer material,
which is about three times the yield strength of a
metallic material:
p H; H 3r
y
T 5
Knowing the normal surface displacements, we can
express the thermomechanical contact conditions as
follows for any surface point, i. j 2 X
S
, where X
S
is
the computational domain and includes the contact
regions, W
C
:
u
i j
z
i j
a 0. p
i j
2 0. H . i. j & X
C
6
u
i j
z
i j
a 0. p
i j
0. i. j 6& X
C
7
and
Z
X
S
pdA W 8
where z
i j
is the initial separation, a, the rigid-body
approach; and W, the total applied normal load.
3. Numerical procedures
The equations for the contact problem presented
in the previous section were solved by the discrete
convolution and fast Fourier transform method (DC-
FFT) [10] and the conjugate gradient method [12].
The path of DC-FFT/inuence coecient/Greens
function was employed. Due to the dependence of
the yield strength on temperature, the code devel-
oped in Ref. [10] was modied to include the yield
strengthtemperature relationship for asperities in
contact. The contact constraints, equations (6)(8),
were solved by two iteration processes with the outer
one corresponding to the temperature-dependent
yield strength and the inner one to the pressure dis-
tribution. Starting from a given load, the geometry
of surfaces, and material properties, the computation
entered the outer loop, initiating T
OLD
as T
0
and
r
y OLD
as r
y
(T
0
). Then pressure was initiated with a
uniform distribution as P
OLD
, and the computation
went into the inner iteration loop to determine the
thermoelastic displacements. Using the conjugate gra-
dient method, we can determine the nal pressure
distribution P
NEW
, which satises:
P
OLD
P
NEW
k k
P
P
OLD
P
NEW
j j
P
P
OLD
j j
2 10
5
9a
Load W k k
P
P
NEW
ds W j j
P
P
NEW
ds j j
2 10
3
9b
where ds is the area of an element region. Otherwise,
the pressure distribution was updated with P
OLD
=
P
NEW
, and one could then proceed to the inner itera-
tion loop again. The temperature distribution was
computed until the following condition was satised:
T
OLD
T
NEW
k k
P
T
OLD
T
NEW
j j
P
T
OLD
j j
2 10
4
9c
Here, the convergence limits are chosen to ensure that
the numerical results are stable and convergent. The
ow chart for the entire analysis is given in gure 2.
4. Results and discussion
Yield strengthtemperature relationships for a
group of steels and aluminum alloys [17] are shown in
gure 3. In both cases, the yield strength decreases
H. Yu et al./Inuence of temperature-dependent yield strength 157
with increasing temperature. Properties of the two
selected materials (52100 steel and 2011 T3 aluminum
alloy) are listed in table 1. The hardnesstemperature
relationship for the 52100 steel is tted with a linear
function given by equation (10a), and that for the alu-
minum alloy with a Gaussian function given by equa-
tion (10b), where DT = T)T
0
and T
0
= 27 C. The
coecients of determinant (R
d
2
) for the curve tting
equations are also given below:
HDT 3r
y
DT 1819.34494
1.77287DT R
2
d
0.98569 10a
HDT 3r
y
DT 10.0337
70. 245.14236
195.19381
exp
DT 3.62606
195.19381

2
" #
R
2
d
0.99655 10b
Two contact pairs were formed with the same materi-
als, namely the 2011 T3 aluminum alloy pair and the
52100 steel pair. The formulation and procedures
mentioned in the previous section were used to
analyze these cases numerically. Here, E
1
= E
2
and
m
1
= m
2
. The surface topography was measured and
digitized by a phase-shift white-light interferometer.
The root-mean-square (RMS) surface roughness value
for the 52100 steel at was measured to be 0.21 lm.
In order to compare the results on the same basis,
Figure 2. Flow chart of the numerical calculation.
Table 1
Properties of 2011 T3 aluminum alloy and 52100 steel.
Properties 2011 T3 52100
Youngs modulus E (GPa) [17] 80 200
Poisson ratio m [17] 0.33 0.3
Thermal conductivity K (W/m K) [17] 152 50.2
Friction coecient
*
relative velocity
(m/s) l
f
V
0.2 0.2
Thermal expansion coecient
h (lm/m K) [17]
25 11.7
(a)
(b)
Figure 3. Yield strength as a function of temperature variation [17].
(a) A few steels. (b) A few aluminum alloys.
158 H. Yu et al./Inuence of temperature-dependent yield strength
the 2011 T3 aluminum at should be analyzed at the
same roughness value. This is also the composite
roughness of two contacting surface pairs because in
the current study, the ball is considered to be ideally
smooth.
4.1. Inuence of the temperature-dependent yield
strength on the maximum ash temperature
Figure 4 plots the maximum ash temperature for
both alloys. A non-dimensional load W
*
is dened as
W
*
= 2W/(A
h
E
*
) for result presentation, where A
h
is
the Hertzian contact area between the ball and a
smooth half-space under load W and E* = 1/[(1)m
1
)/
E
1
+(1)m
2
)/E
2
]. Figures 4(a) and (b) show the maxi-
mum temperature rise as a function of the non-
dimensional load. The yield strength is either constant
(r
y
r
y
T
0
and T
0
= 27
o
C) or temperature-depen-
dent (r
y
r
y
T). The results show that in both
cases, the maximum temperature rise is slightly higher
when the yield strength is treated as constant. The
results from smooth Hertzian contacts show the same
trend .
Figure 4(c) plots the ratio of the maximum tempera-
ture rise when the temperature-dependent yield
strength is considered to that obtained with the con-
stant yield strength (temperature ratio) versus load. At
light loads, the temperature ratio is close to 1 for both
alloys as expected. When the load is increased, the
temperature ratio for the steel decreases and becomes
smaller than that for the 2011 aluminum alloy.
Figure 4. Results of the maximum temperature rise as a function of the non-dimensional load. (a) The 52100 steel case. (b) The 2011 alumi-
num alloy case. (c) The DT*W* curves for the rough surfaces in contact, DT*= DT
h
/DT
h
, which is the temperature ratio, and
W

2W,A
h
,E

.
H. Yu et al./Inuence of temperature-dependent yield strength 159
4.2. Inuence of the temperature-dependent yield
strength on the average contact pressure and the
area undergoing plastic deformation
Figures 5(a) and (b) plots the results of the average
contact pressure as a function of the non-dimensional
load. Figure 5(c) shows the variation of the ratio of
the average pressure on the contact surfaces with and
without taking into account the inuence of tempera-
ture (pressure ratio) versus load. For the 52100 steel
Figure 5. The average pressures as a function of the non-dimen-
sional load. (a) The 52100 steel case. (b) The 2011 aluminum alloy
case. (c) The P*W* curves for the rough surfaces in contact. Here,
P*= P
ave
/P
ave
is the pressure ratio.
Figure 6. Ratios of the plastically deformed region to the total con-
tact area as a function of the non-dimensional load. (a) The 52100
steel case. (b) The 2011 aluminum alloy case. (c) The A

p
W* curves
for the rough surfaces in contact, Where A

p

Apryt,Ac ryt
Apryc,Ac ryc
.
160 H. Yu et al./Inuence of temperature-dependent yield strength
with smooth surfaces in contact (Hertzian contact), the
curve is perfect linear until W
*
reaches 0.004, indicat-
ing that plastic deformation is not signicant. For the
rough surface in contact, the rst few points in the g-
ures show a linear variation. Beyond that, the average
pressure of the temperature-dependent case decreases
faster. Compared to the results for smooth surfaces,
the average pressure for the rough surface in contact is
higher and reaches the yield point at a lower load.
Figure 5(c) shows that the pressure ratios are smaller
than one and that they decrease as the load increases.
For the 52100 steel, the pressure ratio is smaller than
that for the 2011 aluminum alloy. This indicates that
the thermal softening eect is more signicant for the
steel alloy due to the higher temperature rise. The zig-
zag shape of the curve for the 52100 steel is caused by
the non-linear variation of the asperity contact area.
The 2011 aluminum alloy behaves dierently because
of its low yield strength. The contact area of the 2011
aluminum alloy increases so rapidly with load that
most of its asperities are involved in the contact. As a
result, the zigzag phenomenon does not appear to be
as obvious as that with the 52100 steel.
Figures 6(a) and (b) shows the ratio of the area
undergoing plastic deformation to the total contact
area (area ratio, A
p
,A
c
), or the relative plastic defor-
mation area of the two materials with and without
considering the temperature eects on the yield
strength. The data indicate that yield appears at lower
loads when the thermal softening eect is considered.
Figure 6(c) shows the variations of another area ratio,
A

p
. For the rough surface in contact, A

p
is dened as
Figure 7. Contact pressures under dierent loads for the 52100 steel. (a) r
y
= Const, (b) r
y
= r
y
(T) W*=0.00175. (c) r
y
= Const,
(d) r
y
= r
y
(T) W*=0.00225. (e) r
y
= Const, (f) r
y
= r
y
(T) W*=0.00325.
H. Yu et al./Inuence of temperature-dependent yield strength 161
the ratio of the relative plastic deformation area to the
total contact area, with and without considering the
temperature eects. From this gure, one can see that
for the case of 52100 steel, the plastic deformation
area with thermal softening is much larger than that
when the temperature eect is not taken into account.
For the 2011 aluminum alloy case, the dierence is not
signicant though the trend is similar to that for the
52100 steel. Figure 6(c) also shows the ratios, A

p
, of
the two cases are either greater than or equal to 1, and
the values for the 52100 steel case are greater. This dif-
ference is due to the lower yield strength of the alumi-
num alloy and the smaller temperature rise. The ratio
of the relative plastic deformation area for the 52100
steel shows a strongly non-linear dependence on load.
This may be caused by the uneven change of the real
asperity contact region due to the load increase as
mentioned before the contact area and the plastic
deformation area both increase, but the area of asperi-
ties under elastic deformation may increase with load
in a non-smooth manner. In order to view the defor-
mation more clearly, gure 7 plots the pressure distri-
butions of the 52100 steel when the non-dimensional
load is 0.00175, 0.00225 and 0.00325, respectively, cor-
responding to local maxima and minima of the curve
in gure 7(c), marked as A, B and C. Where plastic
deformation occurs, pressure peaks are attened and
interconnected. The plots in gure 7 indicate that the
percentage of the plastic deformation region when con-
sidering the thermal softening eect is much greater
than that when such eect is omitted at a non-dimen-
sional load of 0.00175 (A
p
(r
y
(t))/A
p
(r
y
=c)=2.6) and
0.00325 (A
p
(r
y
(t))/A
p
(r
y
=c)=3.4) where the curve
reaches a local maximum. At a non-dimensional load
of 0.00225, where the curve in gure 6(c) reaches the
local minimum, such dierence is smaller (A
p
(r
y
(t))/
A
p
(r
y
=c)=2.0).
Figures 8 and 9 compare the pressure distributions
when W
*
= 0.004 for the 52100 steel and 2011 alumi-
num alloy cases. One can see that for the 52100 steel
when thermal softening is included, the pressure peaks
are more interconnected and the area undergoing plas-
tic deformation is larger. A concave shape is clearly
seen, which is created by the dierence of the asperity
yield strength due to the dierence in asperity ash
temperature. The asperities at the center of the contact
experience more plastic deformation and lower pres-
sure as a result of the higher temperatures there. For
the 2011 aluminum alloy case, the trend is the same
though such interconnection is not signicant because
the temperature variation in this contact spot is not as
large as that observed for the 52100 steel. There is no
Figure 8. Contact pressures under W*=0.004 for the 52100 steel case. (a) r
y
= const. (b) r
y
= r
y
(T).
Figure 9. Contact pressures under W*=0.004 for the 2011 aluminum alloy case. (a) r
y
= const. (b) r
y
= r
y
(T).
162 H. Yu et al./Inuence of temperature-dependent yield strength
signicant yield strength dierence among the asperi-
ties of 2011 aluminum.
5. Conclusion
The elasticperfectly plastic contact problem was
studied assuming that the yield strength is tempera-
ture-dependent. Numerical results for two typical engi-
neering materials were obtained and compared with
those assuming constant yield strength. Inclusion of
the thermal softening eect results in lower tempera-
ture rise and lower average pressure, but with higher
percentage of plastic deformation. The thermal soften-
ing eect on asperity plastic deformation is more
prominent in materials with higher yield strength and
lower thermal conductivity.
Acknowledgments
The authors would like to express their sincere grati-
tude to the U.S. Oce of Naval Research and National
Science Foundation for their support of the research.
References
[1] L. Qiu and H.S. Cheng, T ASME J. Tribol. 120 (1998) 310.
[2] J.Q. Gao, S.C. Lee and X.L. Ai, ASME J. Tribol. 122 (2000)
519.
[3] X. Tian and F.E. Kennedy, ASME J. Tribol. 116 (1994) 167.
[4] K.L. Johnson, Contact Mechanics (Cambridge University Press,
Cambridge, 1996).
[5] F.E. Kennedy, Thermomechanical Eects in Sliding Systems
(Elsevier Sequoia S. A., New York, 1985).
[6] F.F. Ling, Surface Mechanics (John Wiley & Sons, New York,
1973).
[7] F.E. Kennedy, Wear 100 (1984) 453.
[8] B.Y. Ting and W.O. Winer, ASME J. Tribol. 111 (1989) 315.
[9] J.R. Barber, Int. J. Mech. Sci. 14 (1972) 377.
[10] S. Liu and Q. Wang, ASME J. Tribol. 123 (2001) 17.
[11] S.C. Lee and H.S. Cheng, Tribol. Trans. 35(3) (1992) 523.
[12] I.A. Polonsky and L.M. Keer, Wear 231 (1999) 206.
[13] S.C. Lee and N. Ren, STLE Tribol. Trans. 39 (1996) 67.
[14] G. Liu, Q. Wang and S. Liu, ASME J. Tribol. 123 (2001)
595.
[15] G. Liu, J. Zhu, L. Yu and Q. Jane Wang, Tribol. Trans. 44
(2001) 437.
[16] H.S. Carslaw and J.C. Jaeger, Conduction of Heat in Solids
(Oxford University Press, London, 1959).
[17] J.R. Davis, in: Metals Handbook, 2nd ed. Desk ed.; ASME
International, Materials Park, OH, 1998.
H. Yu et al./Inuence of temperature-dependent yield strength 163

Вам также может понравиться