Beruflich Dokumente
Kultur Dokumente
LIMITS OF APPLICABILITY
ROUGH SURFACES
105
OF ELASTIC
CONTACT
MODELS
OF
JOHN I. McCOOL
SKF Industries
(U.S.A.)
Inc.,
P.O. Box
515,
1100
First Avenue,
Summary
Recent stochastic models for analyzing the contact of rough surfaces
assume that the asperities are microhertzian,
i.e. that they can be approximated as second-order surfaces in the vicinity of contact points, and that the
asperities deform elastically. Using a plane strain solution from the literature
for a sinusoidally corrugated
half-space, the range of validity of these assumptions is shown to be related to the mean square surface slope and the
macrocontact
pressure. By extension to random surfaces characterized
by a
one-dimensional
spectral density function an interval on the surface spatial
frequency is found over which the asperities deform elastically but without
completely flattening. A numerical example is given.
1. Introduction
Numerous
stochastic
models have been developed
to describe the
relationship between the applied load and the mean plane separation and the
real area of contact which result when two microscopically
rough surfaces
are pressed together. In one of the earliest of such models, that of Greenwood and Williamson [ 11, the asperities on an isotropic surface are modelled
as cones with gaussiandistributed
heights and constant-radius
tips. Gupta
and Cook [2] relaxed this assumption permitting the tip radii to be lognormally distributed but independent
of the tip height. Tallian [3] developed
a model for strongly anisotropic surfaces in which the asperities are modelled
as prisms of triangular cross section, gaussiandistributed
height and constant
tip radius. In equating the base angle of the prisms to the measured r.m.s.
surface profile slope in his model, he recognized that surface frequency and
not just roughness height determined the contact behavior.
Nayak [4] showed the applicability
to engineering surfaces of the rich
two-dimensional
random process characterization
developed
by LonguetHiggins [ 51 for describing ocean waves. With this description the dependence
on frequency
content is explicit, the relation between profile and surface
0043-1648/83/0000-0000/$03.00
(sinusoidal)
= i
COS-(1 - 2Q)
(1)
107
area fraction
(hertzian
approximation)
=-
2
77
Q l/2
(2)
2Q-12
(3)
Both expressions (1) and (2) correctly predict an area fraction of zero
at Q = 0. As Q increases, the predicted area fractions diverge. For Q = 1 the
sinusoidal expression predicts full contact while the hertzian approximation
predicts 63% contact. Table 1 shows the two area fractions and their perTABLE1
True uersussecond-orderapproximation
areafraction
Q
Area fraction
(sinusoidal)
Area fraction
(hertzian)
Percentage
difference
Stress concentration
factor Pmax IP
0.01
0.03
0.05
0.07
0.09
0.11
0.13
0.15
0.17
0.19
0.21
0.23
0.25
0.27
0.29
0.31
0.33
0.35
0.37
0.39
0.41
0.43
0.45
0.47
0.49
0.0637685
0.110825
0.143566
0.170463
0.193973
0.215219
0.234825
0.253183
0.270556
0.287133
0.303053
0.318424
0.333333
0.347849
0.36203
0.375924
0.389573
0.403013
0.416277
0.429394
0.44239
0.45529
0.468116
0.48089
0.493633
0.063662
0.110266
0.142353
0.168434
0.190986
0.211143
0.229537
0.246562
0.262485
0.277496
0.291736
0.305312
0.31831
0.330797
0.34283
0.354455
0.36571
0.376629
0.387241
0.397569
0.407636
0.41746
0.427058
0.436445
0.445634
0.167102
0.504321
0.845466
1.19068
1.54012
1.89395
2.25229
2.61531
2.98318
3.35609
3.73424
4.11782
4.50703
4.90215
5.30338
5.71103
6.12535
6.54667
6.97534
7.41169
7.85612
8.30904
8.77095
9.24232
9.72373
20
11.547
8.94427
7.55929
6.66667
6.03023
5.547
5.16398
4.85071
4.58832
4.36436
4.17029
4
3.849
3.71391
3.59211
3.48155
3.38062
3.28798
3.20256
3.12348
3.04997
2.98142
2.9173
2.85714
< 0.27
(4)
A and frequency
m2 is
= 2(rrfA)
(5)
(6)
which must obtain for an error of less than 5% in the contact area. For a
different allowable error it is necessary only to replace the number 0.27 in
eqn. (6) by the Q value given in Table 1 corresponding
to the desired error
percentage.
Although strictly applicable to harmonic surfaces it is suggested that
this criterion
on the mean square surface slope be applied to judge the
probable error in second-order asperity models for random surfaces as well.
For random surfaces the value of m2 is conveniently
expressed in terms
of the one-sided spectral density function s(f) as
m2 =
CW2
f2s(f) df
(7)
3. Elastic flattening
The index commonly used to assess the severity of the contact of rough
surfaces is the film parameter A defined as the ratio of the lubricant film
thickness h to the composite r.m.s. roughness u of the contacting bodies (cf.
ref. 11 and ref. 12). Leaver et al. [ 131 recognized that in constructing
this
index it is possible to be seriously misled by the relatively large contribution
to the r.m.s. roughness value of long spatial wavelengths. They argued that
wavelengths which were long relative to the size of a hertzian contact would
not cause interactions but would merely be ridden by the moving contact.
They recommended
the removal by filtering of all low frequency waves
having wavelengths
that exceeded twice the contact width prior to the
computation
of the r.m.s. roughness.
109
Sayles et al. [14] proposed, as a refined cut-off criterion, the elimination as harmless of all waves which will elastically flatten under the applied
load. For purely harmonic surfaces this criterion is readily translated into a
constraint
on the mean square surface slope using the analysis of an elastically deformed sine wave presented in Appendix A.
A harmonic surface of amplitude A and frequency f will, when pressed
against a smooth elastic plane, be fully flattened when the contact halfwidth
b is equal to half the period, i.e. when
b = 1/2f
(8)
= 1.0,
i.e.
p = rfAE
(9)
Using the expression for the mean square slope in eqn. (5), the flattening condition can be expressed as
p>
(~YFI#~E
rfAE=
(10)
or
2
(11)
4. Plastic deformation
derived by Sayles et
>
(12)
some degree of plastic flow will occur
y/2
(13)
0.96Y4
(PE,)2
141
the maxi-
= 0.3OP,,X
criterion
al.
= Y4(pE)2
for
(14)
110
5. Frequency
models
Let us define
(15)
and
h2)P
Y4
__(@)2
(16)
1122 <
(17)
(m2)p
p = Y/2/2
m2W
= CW2
/f2s(f)
df
(19)
fL
Mfl)
f is a general frequency
mz(f) is a monotonically
= W2)E
(20)
111
f*
Fig. 1. The function mz(fl.
various frequencies each below fi will flatten, even though the mean square
slope rn2 for the composite surface is less than (~tl~)z. Nevertheless it seems
reasonable to regard frequencies below fi as relatively harmless with respect
to asperity interactions.
In the same sense, single-frequency
sinusoids of
frequency lower than f2 defined implicitly by
m2V2)
(21)
= (mz),
6. Numerical
example
112
E'+ ~ ___-E
2(1 - V)2
so that
i 2 x 135 x 10
I*
w = 7~X 1.26 X 10
and the maximum
is
(23)
P=
7r x 10
This pressure
bearings.
is typical
of the contact
pressure
lbf in-*)
engine
of the form
(24)
3/L
(25)
where L is the total length of profile record used in compiling the spectrum.
Frequencies
below fL are believed to be damped as a consequence
of
the removal of a linear trend from the profile record. The assumption that
L = 1500 pm gives
between
the limits
fL and f, is
fu
$
=
s
s(f) dt
= c(fL-'-fu-I)
(26)
fL
=
- (5 x lo-)-)
113
mz(f) is
df
(27)
fL
m2(f)
(N2c(f
(28)
- fL)
- 2 x lo-s)
macrocontact
for p
pressure
gives
1040
(m2)E
there are
2w
2 X 82.6
-=
= 25.9
111.57 x 1o-2
l/f1
wavelengths within the contact width.
It should be noted that the criterion of eliminating waves longer than
twice the macrocontact
width would have resulted in the removal of frequencies lower than
flZ
1
__A__
2(2w) = 2 X 2 X 82.6 = 3.03 X 10m3 j.fm- (75.7 in-)
AISI-SAE
52100
2ooo4
(m2)p = (1040 X 1.26 X 1O5)2
= 9.32
Equating
x 1o-4
to m2(f2) gives
represented
by the measured
spectrum
will
114
The variance
computed as
02 =
s
of the surface
rLl
s(f) df = 3.14 X 10-5((1.57
excluding
X 10-2)-
frequencies
below fi may be
- (5 X 10-i)-}
f,
=
so that
u = 4.40 X 10e2 E.trn(1.76 pin)
Thus by excluding elastically flattening waves the r.m.s. roughness is reduced
to 35% of its unfiltered value.
If instead fi, corresponding
to the elimination
of waves longer than
twice the macrocontact
width, is used the variance would be
u 2 = 3.14 x lOP((3.03
x 10-s))
- (5 x lo-)-}
range
criterion
value.
the
= 22pjm22E
212 X 1040
= (6.07 X 10-4)12 x 1 ?F .: ;$
= 0.47
area calcula-
5%. To estimate
the
115
error in area is
Acknowledgments
The work described in this paper was performed under Contract DEAC02-80ER10693
with the U.S. Department
of Energy. This support and
the encouragement
of Dr. Oscar Manley, the technical project monitor, are
gratefully acknowledged.
References
1 J. A. Greenwood and J. Williamson, Contact of nominally flat surfaces, Proc. R. Sot.
London, Ser. A, 295 (1966) 300 - 319.
2 P. K. Gupta and N. H. Cook, Statistical analysis of mechanical interaction of rough
surfaces, J. Lubr. Technol., 94 (1) (1972) 19 - 26.
3 T. E. Tallian, The theory of partial elastohydrodynamic contacts, Wear, 21 (1972)
49 - 101.
4 R. P. Nayak, Random process model of rough surfaces, J. Lubr. Technol., 93 (3)
(1971) 398 - 407.
5 M. S. Longuet-Higgins, The statistical analysis of a random moving surface, Philos.
Trans. R. Sot. London, 249 (1957) 321- 347.
6 A. W. Bush, R. D. Gibson and T. R. Thomas, The elastic contact of a rough surface,
Wear, 35 (1975) 87 - 111.
7 M. OCallaghan and M. A. Cameron, Static contact under load between nominally flat
surfaces in which deformation is purely elastic, Wear, 36 (1976) 79 - 97.
8 A. W. Bush, R. D. Gibson and G. P. Keogh, Strongly anisotropic rough surfaces,
J. Lubr. Technol., 101 (1) (1979) 15 - 20.
9 J. I. McCool and S. S. Gassel, The contact of two surfaces having anisotropic roughness geometry, ASLE Spec. Publ. SP-7, 1981, pp. 29 - 38 (American Society of
Lubrication Engineers).
10 J. Dundurs, K. Tsai and L. Keer, Contact between elastic bodies with wavy surfaces,
J. Elast., 3 (2) (1973) 109 - 115.
11 T. E. Tallian, Y. P. Chiu, D. Huttenlocher, J. Kamenshine, L. B. Sibley and N.
Sindlinger, Lubricant films in rolling contact of rough surfaces, ASLE Trans., 7 (2)
(1964) 109 - 126.
12 P. Dawson, Effect of metallic contact on the pitting of lubricated rolling surfaces,
J. Mech. Eng. Sci., 4 (1) (1962) 16 - 21.
13 R. Leaver, R. Sayles and T. Thomas, Mixed lubrication and surface topography of
rolling contacts, Proc., Inst. Mech. Eng., London, (1974) 461 - 469.
14 R. Sayles, G. desilva, J. Leather, J. Anderson and P. MacPherson, Elastic conformity
in hertzian contacts, Tribol. Int., 14 (6) (1981) 315 - 322.
15 F. Seely and J. Smith, Advanced Mechanics of Materials, Wiley, New York, 2nd edn.,
1952.
16 E. Church, M. Howells and T. Vorburger, Spectral analysis of the finish of diamondturned mirror surfaces, Proc. Sot. Photo-Opt. Instrum. Eng., 315 (1982) 202 - 218.
116
Appendix
A.1. Elastic contact area fraction for a sinusoidal surface: exact analysis
versus hertzian approximation
A.l.l. Sinusoidal surface
Let us consider a semi-infinite body with a sinusoidal surface in elastic
contact with another perfectly flat infinite body as shown in Fig. Al.
The equation
and sinusoidal
half-spaces.
of the sinusoid is
y = A cos(2nfx)
(Al)
= +EfA{I
-cos(27rfb)}
et al. [Al]
(-1
;;lfcos-(l--&)
(-43)
117
The contact
sinusoidal
length as a fraction
2b
= l/f
area fraction
Full contact,
of the period is
1
= - cos -+-&)
,r
(A4)
2p/1rE fA = 2
or
&-p/fA=rE
A.l.3. Hertzian approximation
For a standard hertzian analysis of the contact of a flat plane and the
tip of a sinusoidal asperity the curvature in the vicinity of the tip may be
calculated as
P=
d2@(xN
dx2
(A51
x=0
where
6(x) = A { 1 - cos(2rfx))
is the separation
position x.
This gives
(A6)
of the undeformed
p = A(2rf)2
(A7)
corresponding,
R,L
to a cylinder
of radius
P
(2rf)2A
The contact length for a cylinder
space under a load P per unit length is
(A8)
of radius R contacting
a flat half-
(A91
and hence the contact length when the sinusoid
der of curvature radius R = 1/(2nf)2A is
is approximated
as a cylin-
l/2
(AlO)
The area fraction,
hertzian
area fraction
= 2b
llf
= 4
of p = P/f,
(All)
118
l/2
hertzian
area fraction
(A=)
solution
is
hertzian
_ 2 (TE\~
*nE
area fraction
-7
predicts
full contact,
2
= - = 0.6366
7T
1 ot
2
P(x)=Pmax
l-
p(x) = 2nEAf
the
l/2
across
(A131
contact
cos(nxf)
(cos(2rfx)
- cos(27rfb)}12
(A14)
stress is
(A15)
(A=)
factor. It represents the factor by
tips exceeds the uniform stress in
J. Dundurs, K. Tsai and L. Keer, Contact between elastic bodies with wavy surfaces,
Elast., 3 (2) (1973) 109 - 115.
J.