Professional Documents
Culture Documents
86 / Vol. 122, JANUARY 2000 Copyright © 2000 by ASME Transactions of the ASME
fully plastic flow, which suggests that the elastoplastic transitional
regime is very long and thus significant. Finally, the results of the
CEB model show that the mean separation is larger and the real
area of contact smaller for the elastic-plastic contact than for the
elastic contact at the same plasticity index and contact load. This
result is contradictory to the experimental results of Powierza et al.
(1992) and physical intuition that the elastic contact should exhibit
a higher stiffness than the elastoplastic contact. Horng (1998)
extended the CEB model to the more general case of elliptic
asperities, but the aforementioned shortcomings were not ad-
dressed. On the other hand, Kucharski et al. (1994) proposed a
model for the contact of rough surfaces based on a finite-element
Fig. 1 Contact of an asperity with a rigid smooth flat
analysis of contact between an elastoplastic sphere and a rigid
plane. Their numerical results show that the CEB model predicts
larger approach and real contact area ratio at the same load. Why
the opposite results are predicted by the CEB model in the two The results obtained by McCool (1986) for anisotropic rough
papers has not been explained as yet and should be clarified. surfaces with random distribution of asperity radii differ negligibly
This paper presents a new elastic-plastic asperity contact model from those of the GW model. The interaction among contacting
for rough surfaces. Its main feature is the modeling of the long asperities and bulk deformation may be important. They are com-
transitional regime from elastic deformation to fully plastic flow. plex problems and, to the authors’ knowledge, have not been
The results of subsequent asperity-contact analyses show that the adequately addressed in previous publications. Efforts are cur-
elastoplastic contact of asperities plays an important part in the rently being made to develop a model which considers asperity
macrocontact behavior of rough surfaces. interactions.
Since the contact of two rough surfaces can be represented by
Modeling the model of an equivalent single rough surface in contact with a
smooth surface, only the contact between a rough surface and a
The following assumptions are made in the asperity contact rigid smooth flat is considered. The formulation of the model is
model to be developed: first presented for the contact of a single asperity and then ex-
1. The asperity distribution is isotropic. tended to the contact of surfaces with many asperities.
2. Asperities are, at least near their summits, spherical. 1 Contact of a Single Asperity With a Rigid Smooth Sur-
3. Asperity summits have a uniform radius R, but their heights face. Consider the contact between one asperity and a rigid
vary randomly. smooth flat, as shown in Fig. 1. Let z and d stand for the height of
4. The interactions among contacting asperities are neglected. the asperity and the mean separation between the two surfaces,
5. Only the asperities deform during contact and no bulk respectively. Then, the contact interference is given by
deformation occurs.
⫽z⫺d (1)
These assumptions are essentially the same as those made in the
GW and CEB models. Successive research on asperity-based mod- The interference is an important variable that measures the
els has shown that assumptions (1), (2) and (3), despite their extent of the asperity deformation. The contact pressure and con-
extreme simplicity, do not greatly limit the generality of the model. tact area of the asperity are fully determined by this interference.
Nomenclature
A e ⫽ elastic contact area of an asperity K ⫽ maximum contact pressure factor y *s ⫽ y s /
A ep ⫽ elastoplastic contact area of an N ⫽ total number of asperities z ⫽ height of asperity measured from
asperity Pa ⫽ mean contact pressure the mean of asperity heights
A p ⫽ plastic contact area of an asperity Pm ⫽ maximum contact pressure z* ⫽ dimensionless height of asperity,
A et ⫽ elastic contact area of two rough R ⫽ radius of curvature of an asperity z/
surfaces we ⫽ elastic contact load of an asperity  ⫽ R
A ept ⫽ elastoplastic contact area of two w ep ⫽ elastoplastic contact load of an ⫽ area density of asperities
rough surfaces asperity 1,2 ⫽ Poisson’s ratios
A pt ⫽ plastic contact area of two rough wp ⫽ plastic contact load of an asperity ⫽ standard deviation of surface
surfaces wy ⫽ critical contact load at the point heights
A n ⫽ nominal contact area of two rough of initial yield s ⫽ standard deviation of asperity
surfaces wf ⫽ critical contact load at the point heights
A t ⫽ total real area of contact, A et ⫹ of fully plastic flow ⫽ distribution function of asperity
A ept ⫹ A pt W et ⫽ elastic contact load of two rough heights
d ⫽ separation based on asperity surfaces * ⫽ dimensionless distribution function
heights W ept ⫽ elastoplastic contact load of two ⫽ plasticity index
d* ⫽ dimensionless mean separation, rough surfaces ⫽ interference
h/ W pt ⫽ plastic contact load of two rough * ⫽ dimensionless interference, /
E⬘ ⫽ equivalent Young’s modulus surfaces 1 ⫽ critical interference at the point of
H ⫽ hardness of softer material Wt ⫽ total contact load, W et ⫹ W ept ⫹ initial yield
h ⫽ separation based on surface W pt *1 ⫽ 1/
heights ys ⫽ distance between the mean of 2 ⫽ critical interference at the point of
h* ⫽ dimensionless mean separation, asperity heights and that of sur- fully plastic flow
h/ face heights *2 ⫽ 2/
k ⫽ mean contact pressure factor
冉 冊 冉 冊
increases: elastic, elastic-plastic and fully plastic. The following
sections develop formulations for the mean contact pressure and 3H 2
4 3 kH 2
2 ⬎ R⫽ R (13)
real area of contact of the asperity as functions of during the 2E k2 4E
three deformation stages.
Substituting Eq. (8) into the above expression yields,
(1) Elastic Contact. The asperity deforms elastically when
is sufficiently small. According to the Hertz theory for the elastic 2 ⬎ 4 1 /k 2 (14)
contact of a flat with a sphere of radius R (Timoshenko and With k ⫽ 0.4, one obtains
Goodier, 1951), the contact area A e , contact load w e , maximum
contact pressure P m and mean contact pressure P a of the asperity 2 ⬎ 25 1 (15)
can be expressed in terms of as
Inequality (15) suggests that the interference required to produce
Ae ⫽ R (2) fully plastic flow of the asperity is at least 25 times that at initial
yielding.
w e ⫽ 共 34 兲ER 1/2 3/2 (3) The minimum value of 2 may also be further estimated using
experimental results. Based on the work of Johnson (1985), fully
P m ⫽ 共2E/ 兲共 /R兲 1/2 (4) plastic deformation occurs when the contact force at fully plastic
deformation, w f , is about equal to 400 times that at initial yielding,
Pa ⫽
2
3
Pm ⫽
4E
3 冉冊
R
1/2
(5)
w y . Or,
w f /w y ⫽ 400 (16)
It was shown by Tabor (1951) that, when the maximum Hertz Using Eq. (3), which is valid for elastic contact, the following
contact pressure reaches P m ⫽ 0.6H, or, the average contact expressions are obtained for w f and w y :
pressure P a ⫽ 0.4H, the initial yielding occurs. For a more
general form we can correlate the mean contact pressure P a at the w y ⫽ 共 34 兲ER 1/2 13/2 (17)
point of initial yielding with the hardness by
and
p a ⫽ kH (6)
w f ⱕ 共 34 兲ER 1/2 23/2 (18)
The critical interference, 1, at the point of initial yielding can be
expressed in terms of the parameters of the contacts. Equation (5) Dividing inequality (18) by Eq. (17) yields
is rewritten to give
共 2 / 1 兲 3/2 ⱖ w f /w y ⫽ 400 (19)
⫽
3pa
4E 冉 冊 2
R (7) Or
2 ⱖ 54 1 (20)
Then, substituting Eq. (6) into Eq. (7), yields
This expression suggests that the contact interference at the onset
1 ⫽ 冉
3 kH
4E 冊 2
R (8)
of fully plastic deformation would be at least 54 times that at initial
yielding.
When ⬍ 1, the contact is elastic. When ⱖ 1, the contact is (3) Elastoplastic Contact. When the interference is between
either elastoplastic or fully plastic. 1 and 2, i.e., 1 ⬍ ⬍ 2, the asperity deforms elastoplasti-
cally. Since the total deformation is composed of a mixture of the
(2) Fully Plastic Contact. When is increased to another elastic and plastic deformations in this stage, the relations for the
critical value 2 at which the mean contact pressure P a of the contact area A ep and mean contact pressure P a as functions of the
asperity reaches the value of H, fully plastic deformation occurs. interference become complex. The functional relations are mod-
During the stage of fully plastic deformation ( ⱖ 2), the mean eled below based on results from other researchers in conjunction
contact pressure P a remains constant at H. Or with insights into the physical nature of the asperity contact.
First, the relation between P a and is derived. Francis (1976)
Pa ⫽ H (9) presented the following function for the contact in the elastoplastic
transitional regime based on a statistical analysis of spherical
The contact area, according to Abbott and Firestone (1933), is indentations:
equal to the geometrical intersection of the flat with the original
undeformed profile of the asperity. Or Pa h/b
⫽ C 1 ⫹ C 2 ln (21)
YR Y R /E
Ap ⫽ 2R (10)
where P a is the mean contact pressure, Y R the unaxial flow stress
The contact load w p of the asperity is equal to the contact area of the material, h the displacement of the contact center, b the
multiplied by the mean contact pressure. Or radius of the contact area, and C 1 and C 2 are the regression
constants. Equation (21) suggests a logarithmic relationship be-
wp ⫽ 2RH (11)
tween the contact pressure P a and the geometrical parameter h/b.
While no solid expression for 2 is known, the minimum value Therefore, the dependence of P a on in the regime of the
of this critical contact interference may be estimated based on a elastoplastic deformation of the asperity may analogously be char-
simple analysis. At ⫽ 2, the load carried by the contact is equal acterized by the following logarithmic function:
to 2 R 2 H by Eq. (11). On the other hand, the load carried by the
P a ⫽ a 1 ⫹ a 2 ln 共 /a兲 (22)
contact at ⫽ 2, had it been elastic, would be equal to ( 34) ER 1/ 2 ⫻
23/ 2 by Eq. (3). Therefore, the following inequality is established: where a 1 and a 2 are two constants to be determined and a is the
contact radius of the asperity. The ratio /a corresponds to the
2 R 2 H ⬍ 共 34 兲E 1/2 3/2 (12) geometrical parameter h/b for the case of spherical indentation.
a ⫽ 共2 R兲 1/2
共 ⱖ 2兲 (24)
Therefore, when the asperity deforms elasto-plastically, the fol-
lowing relation is expected
a ⫽ 共C R兲 1/2 (25)
where C is a variable coefficient having a value between one and
two. Substituting Eq. (25) into Eq. (22), the mean contact pressure
P a may be expressed as:
Fig. 2 Relation between the real area of contact and interference in the
P a ⫽ a 3 ⫹ a 4 ln (26) elastoplastic transitional regime
冋 冉 冊 冉 冊册
contact to give
⫺ 1 3
⫺ 1 2
⫻ ⫺2 ⫹3 (34)
ln 2 2 ⫺ 1 2 ⫺ 1
a 3 ⫽ H ⫺ H共1 ⫺ k兲 (29)
ln 2 ⫺ ln 1
This expression is then algebraically simplified to yield the desired
and A ep ⫺ function through the elastoplastic transitional regime:
H共1 ⫺ k兲
冋 冉 ⫺ 1
冊 冉 ⫺ 1
冊册
3 2
a4 ⫽ (30) A ep ⫽ R 1 ⫺ 2 ⫹3 (35)
ln 2 ⫺ ln 1 2 ⫺ 1 2 ⫺ 1
Therefore, the relation between the mean contact pressure and the Equation (35) is a fourth-order polynomial in . It satisfies the four
contact interference in the regime of the elastoplastic deformation boundary conditions and is a monotonically increasing function
is given by within 1 ⱕ ⱕ 2.
Finally, using Eqs. (31) and (35), the contact load (w ep ⫽
ln 2 ⫺ ln P a A ep ) of the asperity in the regime of elastoplastic deformation
P a ⫽ H ⫺ H共1 ⫺ k兲 (31)
ln 2 ⫺ ln 1 can be expressed as a function of the interference by
Next, the relation between contact area A ep and contact inter-
ference is derived. This relation may be modeled by a polyno- 冋
w ep ⫽ H ⫺ H共1 ⫺ k兲 册
ln 2 ⫺ ln
ln 2 ⫺ ln 1
冋 冉 冊 冉 冊册
mial smoothly joining the expressions for A e and A p as functions
of . The domain of the polynomial is from 1 to 2. It should ⫺ 1 3
⫺ 1 2
⫻ 1⫺2 ⫹3 R (36)
monotonically increase and satisfy four boundary conditions: 2 ⫺ 1 2 ⫺ 1
A ep ⫽ A e , dA ep /d ⫽ dA e /d at ⫽ 1 and A ep ⫽ A p ,
dA ep /d ⫽ dA p /d at ⫽ 2. The function may be easily
2 Contact of Two Rough Surfaces. The above theory for
constructed by mapping an appropriate “template” cubic polyno-
the contact of one asperity with a rigid smooth flat may be used to
mial segment into the quadrilateral bounding the transition region
model the elastic-plastic contact between two rough surfaces. If
on the A ep ⫺ plane, as shown in Fig. 2. This template curve is
there are N asperities on a nominal area A n , the expected number
defined by
of asperity contacts will be
冕 冕
y ⫽ ⫺2x 3 ⫹ 3x 2 (32) ⬁ ⬁
冕 冕
0.0339 0.7
d⫹ 1 ⫹⬁ 3.02 ⫻ 10 ⫺4 0.0414 1.0
⫽N A e 共z兲dz ⫹ N A p 共z兲dz 6.576 ⫻ 10 ⫺4 0.0476 1.5
1.144 ⫻ 10 ⫺3 0.0541 2.0
d⫹ 2
d
1.77 ⫻ 10 ⫺3 0.0601 2.5
⫹N 冕 d⫹ 2
A ep 共z兲dz
冉冊 冕
d⫹ 1
1/2 h*⫺y s*⫹ 1*
冕 冕
Wt 4
d⫹ 1 ⬁ W *t ⫽ ⫽  * 3/2 *共z*兲dz*
A nE 3 R
⫽ A n R 共z兲dz ⫹ 2 A n R 共z兲dz h*⫺y s*
冕
d d⫹ 2
2H ⬁
冕 冋 冉 冊
⫹ * *共z*兲dz*
d⫹ 2
⫺ 1 3
E
⫹ A n R 1⫺2 h*⫺y s*⫹ 2*
2 ⫺ 1
冕 冋 册
d⫹ 1
H h*⫺y s*⫹ 2*
ln *2 ⫺ ln *
⫹3 冉
⫺ 1
2 ⫺ 1 冊册 2
共z兲dz (38)
⫹
E
h⫺y s*⫹ 1*
1 ⫺ 共1 ⫺ k兲
ln *2 ⫺ ln *1
冋 冉 * ⫺ *1
冊 冉 * ⫺ *1
冊册
3 2
and ⫻ 1⫺2 ⫹3
*2 ⫺ *1 *2 ⫺ *1
W t 共d兲 ⫽ W et 共d兲 ⫹ W pt 共d兲 ⫹ W ept 共d兲 ⫻ * *共z*兲dz* (41)
⫽N 冕 d
d⫹ 1
w e 共z兲dz ⫹ N 冕 ⫹⬁
d⫹ 2
w p 共z兲dz
where
 ⫽ R (42)
冕 d⫹ 2
* ⫽ z* ⫺ h* ⫹ y *s (43)
⫹N w ep 共z兲dz The integrals in Eqs. (40) and (41) may be numerically evaluated
d⫹ 1 with sufficient accuracy using five-point Legendre-Gaussian
冕
quadrature.
d⫹ 1
4
⫽ A n ER 1/2 3/2 共z兲dz
3 Results and Discussion
d
The model developed in this paper is now used to study the
⫹ 2 A n HR 冕 ⬁
d⫹ 2
共z兲dz
contact behavior of two nominally flat surfaces over a range of
contact load and plasticity index. The results are compared with
those predicted by the GW model and the CEB model. Their
physical implications are also discussed.
⫹ A n R 冕 冋 d⫹ 2
d⫹ 1
H ⫺ H共1 ⫺ k兲
ln 2 ⫺ ln
ln 2 ⫺ ln 1
Contact of steel-on-steel surfaces is considered with Young’s
modulus E 1 ⫽ E 2 ⫽ 2.07 ⫻ 10 11 Pa, Brinell hardness H ⫽
1.96 ⫻ 10 9 Pa and Poisson’s ratio 1 ⫽ 2 ⫽ 0.29. The distri-
bution of the asperity heights is assumed to be Gaussian and is
冋 冉 ⫺ 1
冊 冉 ⫺ 1
冊册
3 2
given, in dimensionless form, by
⫻ 1⫺2 ⫹3 共z兲dz (39)
2 ⫺ 1 2 ⫺ 1 *共z*兲 ⫽ 共2 兲 ⫺1/2 共 /R兲 exp关⫺0.5共 / s 兲 2 z* 2 兴 (44)
Equations (38) and (39) may be normalized by dividing by A n and The surface roughness is described by two parameters,  and /R.
A n E, respectively. Furthermore, all the length parameters and Table 1 presents the values of  and /R of different surface
variables in the equations are normalized by . The resulting topographies. These values, taken from the experimental observa-
dimensionless equations are given by tion of Nuri and Halling (1975) for typical engineering surfaces,
were also used in the analysis of CEB model (Chang et al., 1987).
A *t ⫽
At
An
⫽  冕 h*⫺y s*⫹ 1*
h*⫺y s*
* *共z*兲dz*
The surface roughness may also be characterized by the plasticity
index, which is related to the two parameters and material con-
stants of the contact members by (Chang et al., 1987):
⫹ 2  冕 ⬁
* 共z*兲dz
⫽
2E
1.5 kH 冉冊 冉
R
1/2
1⫺
3.717 ⫻ 10 ⫺4
2 冊 1/4
(45)
h*⫺y s*⫹ 2*
The value of k is taken to be 0.4 to be consistent with other
冕 冋 冉 冊
researchers. Then the plasticity index of the contact system can be
h*⫺y s*⫹ 2*
* ⫺ *1 3
calculated and is also given in Table 1 along with the correspond-
⫹  1⫺2
*2 ⫺ *1 ing  and /R. The critical contact interference at the point of fully
h*⫹y s*⫹ 1*
plastic flow of the asperity, 2, is taken to be 54 1 based on Eq.
⫹3 冉 * ⫺ *1
*2 ⫺ *1 冊 2
* *共z*兲dz* (40)
(20).
Figure 3 shows the ratio of the contact area contributed by the
Conclusion
An elastic-plastic asperity microcontact model for rough sur-
faces is presented in this paper. The main feature of the model is
incorporation of the transitional regime from elastic deformation to
fully plastic flow of the asperity. The dependence of the mean
contact pressure and contact area of the asperity on its contact
interference in the elastoplastic transitional regime is modeled by
logarithmic and fourth-order polynomial functions, respectively.
Fig. 5 Real area of contact of the surfaces as a function of load and The relation for the ratio of the elastoplastic contact area versus
plasticity index (a) ⴝ 0.7, (b) ⴝ 1.5, (c) ⴝ 2.5 load is derived. It is shown that the elastoplastic contact of asper-
ities plays an important part in the macrocontact behavior of rough
surfaces. A detailed comparison of the present model with the CEB
model and the GW model is made based on the predicted mean
the plasticity index increases. With ⫽ 1.5, the area becomes
surface separation and the real area of contact for a wide range of
larger than that predicted by the GW model and is about 10 to 20%
plasticity index and contact load. Smaller mean separation and
below that by the present model. With ⫽ 2.5, it eventually
larger real area of contact are predicted by the present model than
exceeds results of the present model by about 10% at the load of
the GW model at any given plasticity index and contact load. The
W t /A n E ⫽ 10 ⫺3 .
results are consistent with experimental observation and physical
The physically unreasonable values of the mean surface sepa-
intuition. Based on the results comparison and analysis, the present
ration and real area of contact predicted by the CEB model are
model is shown to be more complete than both the CEB model and
largely caused by the assumption of abrupt transition from elastic
the GW model in describing the elastic-plastic contact phenomena
deformation to fully plastic flow of the asperity. In the CEB model,
between rough surfaces.
the mean contact pressure is assumed to jump from 2/3KH to KH
(from Eqs. (10), (12), and (27) in Chang et al., 1987) when the
mode of the asperity deformation changes. Furthermore, the rate of Acknowledgment
increase of the contact area with respect to the contact interference, This research was financially supported by the Tribology and
dA/d , is also discontinuous at this point, jumping from R to Surface Engineering Program, the National Science Foundation
2 R. As a result of the abrupt increase in mean contact pressure, through Grant CMS-9501877.
the calculated mean separation of the surfaces is higher than those
given by the GW and present models and the difference is most References
pronounced with an intermediate plasticity index such as ⫽ 1.0 Abbott, E. J., and Firestone, F. A., 1933, “Specifying Surface Quality—A Method
to 1.5 as shown in Fig. 4. With the plasticity index in this range, Based on Accurate Measurement and Comparison,” Mech. Engr., Vol. 55, p. 569.