You are on page 1of 14

Wear, 151 (1991) 49-62 49

Relationship between normal and tangential contact


stiffness of nominally flat surfaces

(Received November 12, 1990; revised February 13, 1991; accepted March 13, 1991)

Abstract

A study of the static frictiontess contact between two elastic bodies having nominally flat
surfaces is made. The normal and tangential contact stiffnesses between these two elastic
bodies are determined by measuring the system’s natural frequencies under the effect of
different values of normal load. The theoretical analysis has been carried out to compute
the contact stiffness between flat surfaces from protile measurements using the theories
of Greenwood and Williamson and Onions and Archard. Comparison of the calculated
values, based on these theories, with the experimental data shows good agreement. It is
found that the ratio between the normal and tangentiaf contact stiffness is constant and
is merely a function of the Poisson’s ratio of the contact surfaces.

The contact stiffness generated between contact surfaces, caused by the elastic
and plastic deformations of the encountered asperities, plays a significant role in the
static and dynamic analysis of complex structures having joints and efements in permanent
sliding contact. The presence of a joint in machine tools reduces the overali stiffness
of the structure which in turn affects the perfo~ance of the machine tools [I-3].
consequently, to manufacture the different parts of machine tools it is important to
determine the contact stiffness of the machine tool joints at the design stage. Moreover,
the determination of the contact stiffness between any two contact surfaces is important
to elucidate the intricate phenomenon of self-excited vibration 14-61. The decrease in
the natural frequencies with an increase in the speed of gas turbine blades that use
integral shrouding is affected by the contact of shroud surfaces under certain pre-
tightening forces [7, S].
The concept of most experimental techniques used to determine the normal and
tangential contact stiffnesses are usually based on the measurement of deformation
between two flat specimens under different normal and tangential loads [9-121. In
other words, the two Aat specimens are pressed together by normal force and the
deformation of asperities is recorded by means of an electric micrometer or by the
stylus of a profi~ometer. The results obtained from this usual setup must be refined
by eliminating the effect of deformation of parts of the test machine itself. This could
be done by means of a finite element software.
Expressions for computing the normal and tangential contact stiffness using two
different contact models are derived in the present paper. These two models are

~3-1~8~11~3.5~ 0 19% - Elsevier Sequoia, tausanne


described by the elementary theory of Greenwood and Williamson [13], and the theory
of Onions and Archard [14]. According to the first theory the contact stiffness depends
on the values of the standard deviation of the height of distribution of the asperities,
the mean asperity curvature and the density of asperities per unit area. Regarding
the second theory, the contact stiffness depends on the autocorrelation distance, the
apparent area of contact and the standard deviation of the distribution height of the
asperities. The determination of the tangential and normal stiffness of contact is made
by measuring the natural frequencies of the two structures in both the normal and
tangential directions without and with contact under different normal loads. An
experimental setup is constructed to allow contact between the flat surfaces of these
two structures under uniform distributed pressure. This type of contact between the
two elastic structures modifies the natural frequencies of the resulting compound
system. The measured contact stiffnesses as well as the ratio between them are compared
with those computed by the derived mathematical expressions.

2. Test apparatus

The schematic diagram of the test apparatus is shown in Fig. 1 and a photograph
of the whole test rig is shown in Fig. 2. A mild steel cantilever (l), 18 mm X 30 mm
in section and of length L= 175 mm is rigidly attached to a heavy steel block (2) at
one end. A steel block of dimensions 50 mm x 50 mm x 40 mm and a mass of 0.66
kg is hinged to the other end of the cantilever. This block is brought in contact from
its flat side of 50 mm ~50 mm with another steel seismic mass (2.5 kg) rested on
three steel balls (5) of diameter 4.8 mm. The normal load is applied between the two
flat surfaces of the steel block (3) and the seismic mass (4) by means of a wedge
action produced by tightening the screw (6). The block (3) is hinged with the cantilever
(1) by means of a self-aligning single row ball bearing to ensure the plane contact
between the surfaces of the block (3) and the seismic mass (4). The applied normal
load between the two steel blocks is measured by means of two strain gauges (7)
fixed to the upper and lower sides of cantilever (1). The block (3) is excited in both
the normal and tangential directions of contact by slight shocks. The vibrational response
of the block is measured by two accelerometers (8) in two directions. The whole test
rig, including the wedge pieces, is isolated from ground vibrations by means of rubber
pad sheets.

2 1 7 8 3

Fig. 1. Schematic of the test rig.


Fig. 2. Photograph of test rig.

3. Experimental procedure

For a certain operating beam length L, the natural frequencies of the block (3)
alone, without contact with the seismic block (4), are measured by applying slight
shocks in the normal and tangential directions. The seismic mass (4) is then brought
into contact with block (3) under the action of normal load varying from 0 N to 380
N by tightening the screw (6). The normal load level is determined from the readings
of the digital strain indicator connected to strain gauges (7) and their calibration chart.
At every loading level the block (3) is excited by slight shocks in the normal and
tangential directions. This loading condition is called test no. 1. After reaching the
maximum normal load in this test, the system is unloaded by untightening the screw
(6) and pushing on the upper moving wedge. When the system is completely unloaded
to zero normal load, test no. 2 starts by reloading the system incrementally without
changing the contact area between the two blocks. Test no. 2 is repeated at the same
normal load levels of test no. 1. Tests are repeated to show the sensitivity of the
variation of the contact stiffness with the repeated loading conditions. The use of the
shock test is preferred in this experiment rather than the frequency sweep method
since it guarantees that the contact configuration of the asperities will not change
during the test, thus ensuring accurate evaluation of the measured contact stiffness.
The shocks are always applied at fixed points lying on the symmetrical axis of the
block (3).

4. Experimental results

An example of the development of the system’s frequencies in the normal and


tangential directions with increasing the normal load is shown in Figs. 3(a), 3(b) and
’ I-

Y(f) iv=220 N fin= 362 Hz , i,,- 3762 Hz


(mv)

Fig. 3.
53

r
TOO

X(f) Vii=
256 N fln= 362 Hs , fzn= 3850 Hz
Cm)

*al
f
75

Fig. 3. Development of frequencies in normal and tangentid directions with increasing normal
load.

3(c) and in Figs. 4(a), 4(b) and 4(c) respectively. It is important to notice that the
first natural frequency of the system, either in the normal or the tangential direction,
remains almost constant with increasing normal load. This fact occurs when two masses
are coupled by a tuned linear spring with a monotonically increasing stiffness [15].

5. Theoretical model

The schematic of the mathematical model is depicted in Fig. 5. Assuming that


all springs have linear stiffness, the equations of motion for the slider block (3) can
be written as
rn%l+(k2+kt)Xl-ktX,+k,btI=0 (1)
miil + (k, + k,)Y, -k, Y2 = 0 (2)
Ili+(k&+k,b2)e+~(Xl-Xz)b=O (3)
and the equations of motion for the seismic mass (4) are
M&+k,(X,-Xl)-k,be=O (4)
ti2+(k,,+kn)Y,-k,Y,=O (5)
Introducing the following notations
54

3;

YIP 164 N rlt=i25 112 , iZt= 2437 Hz


X(f)
fmvl

16

32

X(f)
Y4. 220N flt= 125 Hz , 12t~ 2525 Hz
fm)

24

16

fbf b
!
Fig. 4.
55

j2

‘x(f) Wn 256 N fit' 125 Hz , f2p 2588 He


(mv)

” 1 2 3 4 f wi5) 5
(cl
Fig. 4. Frequency development.

Fig. 5. Theoretical model.

and solving eqns. (2) and (5) simultaneously yields the following eigenvalues

(6)
The sum of the two roots gives
r,* + r** = WI2+ %* + ym* (7)
Hence the normal contact stiffness can be calculated from the equation
k,, =mu,,* =m(TIZ + r2* - w2 - q,*)/y (8)
56

r, and r, represent the measured angular frequencies of the system in the normal
direction, while w1 and wb are the angular frequencies in the normal direction for the
slider block and the seismic mass without being in contact.
Equations (l), (3) and (4) could be rewritten as

where

Substituting the measured value of wz and the computed value of o, in eqn. (9) it
is possible to determine the unknown value of w, and consequently the tangential
stiffness k, by a trial-and-error procedure.

6. Theoretical estimation of contact stiffness

6.1. Greenwood and UGlliarnson model


According to the theory of Greenwood and WiIliamson [13] the normal load
between two random surfaces in contact is given by
W= +DAE’/31’2dRF3n(h) (10)
where
m

F,,(h) = (s - h)3nc#x*
(s) ds (11)
s
h

where 4*(s) is the height distribution scaled to make its standard deviation equal to
unity (taken as gaussian).
The normal contact stiffness is given by
dW
k= ds,

where
&=(s-h)o
Differentiating eqn. (10) gives

k,, = 2DAE’@&j(s -h)‘“+*(s) ds (12)


h

According to the theory of Hisakado and Tsukioze [ 161 the elastic tangential displacement
of an asperity in contact with an infinitely rigid smooth surface under a tangential
force T is given as

&= (2-v) Ti
-- (13)
2VG ai
57

The total tangentiat force that can be supported by n asperities will be

(14)

According to hertzian theory [17], the radius of the contact area of an asperity curvature
/3 with this infinite surface is
iZi=J<T--;ija (15)
Therefore the radius of the contact area of all asperity curvatures at height .a will be

where f(z) is the probability that (z>d) or, in nondimensional form, the sum of the
radii of the contact area is

Since the definition of the tangential contact stiffness k, is

k=dT
’ d6r
it is possible to use eqns. (13)-(17) to derive the following expression for the contact
stiffness:

where

Dividing eqn. (12) by (18) the ratio of the normal to the tangential contact stiffness
for two bodies having the same elastic constant is
k2 _ 2(2--v)
--
k, (19)
?r(l-4
This ratio is only a function of the Poisson ratio. For surfaces made of steel the
stiffness ratio according to eqn. (19) is 1.55, while it is only 1.214 according to Mindlin
WI.

6.2. Onions and Archard model


The total load according to this theory is given by
W= $HE ‘(2.3r)crE’h) G33)
58

where

and

f% C)= $& elp(-$exp[-


(s-q]erf(;) (22)

=$7exp(-X*)
erf(f) dr (23)

The normal contact stiffness is

k,= g;

(24)

Applying the same previously mentioned procedure it is possible to show that the
ratio of the normal to the tangential contact stiffness is also given by eqn. (19).
To calculate the theoretical values of the normal and tangential contact stiffnesses,
the surface topography of the polished surface of the block (3) as well as the ground
surface of the seismic block (4) are measured by the stylus of a profilometer (Talyserf
6). Figure 6 shows the surface profile of the two blocks and the bearing ratio curves.
Table 1 gives the topographic data required for the calculation of the contact stiffness.
The measured surface parameters R,, Ii,, S, and P, are used in eqns. (IO), (12),
(19), (20) and (24) to compute the theoretical values of the normal and tangential
contact stiffnesses [19].

5 w-=

t- -
100 um
~
0 50 100 %
(4

Fig. 6. Surface profile and bearing ratio cunre of (a) the slider block and (b) the seismic mass.
59

TABLE 1
Measured topographic data: assessment length=0.8 mm; cut-off=O.lti mm

Parameters Slider block (3) Seismic mass (4)

R, (pm) 0.84 1.59


R, (pm) 2.40 2.70
S, (pm) 49 94
PC (Peaks cm-‘) 25 8

I
0 50 100 150 200 250 300 350 400 4 0

‘a (N)
Fig. 7. Variation of normal contact stiffness with the applied normal load.

n=Kn/Kt

1.5 \
\ ..._
\*-.. .._-..It_r_----_---_.*5-;_..
-..
1.0

- Theory
--- Test No.1
----- Test No.2

-100 200 300 400

Fig. 8. Variation of contact stiffness ratio with applied normal load.

7. Discussion of results

A comparison of the experimentat results of the normal and tangential contact


stiffnesses with the theoretically predicted values by eqns. (lo), (12), (20) and (24) is
shown in Figs. 7 and 8. These figures show the variation of k, and stiffness ratio with
the normal load W for two successive tests on the same area of contact. It is found
that the frequency spectra of the contact surfaces in the tangential direction are not
60

correlated with the theoretical model of contact in the normal load range of B-150
N (Fig. 9). This may be explained by the slip of asperities retative to each other and/
or by the weakness of the elastic junction in the tangentiaf direction in this range of
loading. The frequency spectra in the normal direction are very well correlated with
the proposed model in the whole range of loading (Fig. 10). However, the theoretical
values of the normal contact stiffness k, predicted by the Greenwood and Williamson
model are, as expected, underestimated when compared with the experimental values,
since this model assumes that all asperities have constant curvature. The influence of
the distribution of the asperity curvature has been considered by Onions and Archard
who have shown that the mean pressures between the contact surfaces are significantly
higher than those calculated from the Greenwood and Williamson model. It is useful
to mention that both of the two theoretical models have not taken the plastic deformation
into account, which may explain the discrepancy between the theoretical and experimental
results. The theoretically predicted values of the ratio between the normal and tangential
contact stiffnesses by eqns. (10) and (19) are found to be in good agreement with
the experimentally determined values, as shown in Fig. 8.

0.0 I-
50 100 153 200 250 3Gir 3 0

w (N)
Fig. 9. Variation of tangential contact stiffness with the applied normal load.

0.5

z 0.4

5 0.3

?=
0.2

0.1

0 50 130 150 200 250 300 350 LOO 450

W (N!

Fig. 10. Variation of normal contact stiffness with the applied normal load.
61

8. Conclusions
A new measurement technique for estimating the normal and tangential contact
stiffnesses of two flat surfaces has been developed. The experimental results have been
compared with those computed by the expressions derived from the theories of
Greenwood and Williamson and Onions and Archard. The authors think that the
discrepancy between the theoretical and experimental results is mainly due to the
assumption of considering only the elastic deformation of the asperities. The experimental
results indicate that the contact stiffness in both the normal and tangential directions
increases with an increase of the number of repeated loadings. The important result
obtained here is that the ratio between the normal and tangential contact stiffnesses
of two flat surfaces in contact is constant and agrees with the value obtained by eqn.
(19), which reveals also its dependence on the Poisson ratio of the contacting surfaces.
However it would be expected in practical applications and/or in experimental work
that the ratio k,/k, will not be completely constant or independent of surface roughness,
except for especialty smooth surfaces, since most surfaces have plasticity indices Iarger
than unity, which means that asperities will flow plastically under the action of the
lightest loads.

References
1 R. H. Thomley, K. Connoly and F. Koenigsberger, The effect of flatness of joint faces upon
the static stiffness of machine tool joints, Proc. Inst. Me&. Eng., 1812 (18), Part 1 (1967-1968~,
pp. 271-277.
2 R. H. Thomley, K. Connoly, R. Barash and F. Koenigsberger, The effect of surface topography
upon static stiffness of machine tool joints, ht. J. MTDR, 5 (1965) 57.
3 Z. M. Levina and D. N. Reshetov, Machine design of contact stiffness, Mach. Tool., 36 (12)
(1963) 9.
4 M. ToIstoi, Significance of the normai degree of freedom and natural normal vibrations in
contact friction, Wear, 10 (1967) 199-213.
5 B. V. Budanov, V. A. Kudinov and D. M. Tolstoi, fnteraction of friction and vibrations,
Trenie Iznos, 1 (1980) 79-89 (in Russian).
6 A. Soom and C. Kim, Roughness-induced dynamic loading at dry and boundary lubricated
sliding contact, Trans. ASIUE, J. Lubr. Technol., 105 (1983) 514.
7 C. H. Menq, J. H. Griffin and J. Bieiak, The forced response of shrouded fan stages, ASME
J. tibration, Acoustics, Stress and Reliability in Design, I08 (1985) 50.
8 C. H. Menq, Modeling and vibration analysis of friction joints, ASME J. Kbration, Acoustics,
Stress and Reliability in Design, 111 (1989) 71.
9 M. P. Dolbey and R. Bell, The contact stiffness of joints at low apparent interface pressure,
Ann. CZRP (1970).
10 M. Burdkein, N. Back and A. Cowley, Experimental study of normal and shear characteristics
of machined surfaces in contact, J. Mech. Erg. Sci., 20 (3) (1978) 129.
11 P. Chicate and S. K. Basu, Contact stiffness of machine tool joints, Tribal. Znt., 8 (1) (1975) 9.
12 T. Tsukada and Y. Anno, An analysis of the deformation of contacting rough surfaces, Bull.
JSME, 15 (86) (1972) 982.
13 J. A. Greenwood and J. B. P. Williamson, Contact of nominally flat surfaces, Proc. R. Sot.
London, A295 (1966) 300-319.
14 R. A. Onions and J. F. Archard, The contact of surfaces having a random structure, J. Appl.
Z’hys., 6 (1973) 289.
15 W. T. Thomson, Theory of Vibration with Applcation, Prentice-Hall, New York, 1972, pp.
130.
16 T. Hisakado and T. Tsukizoe, Effect of distribution of surface sfopes and flow pressures of
contact asperities on contact between solid surfaces, Wear, 30 (1974) 213.
17 A. E. H. Love, Treattie on the Mathematical Theory of Elasticity, Cambridge University Press,
4th edn., 1927.
18 R. D. Mindiin, Compliance of elastic bodies in contact, I. Appt. Mech., 16 (1949) 259.
62

19 H. A. Sherif, Parameters affecting contact stiffness of nominally flat surfaces, Wear, 145
(1991) 113.

Appendix A: Nomenclature

ai radius of the hertzian contact area of asperity i with an infinitely rigid


surface
apparent area of contact
separation
density of asperities per unit area
Young’s moduli of the contacting material
equivalent Young’s modulus (l/E ’ = (1 - vlz)/E, + (1 - v,~)/E,)
ith frequency of the system in the normal direction
ith frequency of the system in the tangential direction
shear modulus of elasticity
standardized separation (d/v)
mass moment of inertia
linear stiffness of the slider block in the normal direction
linear stiffness of the slider block in the tangential direction
normal and tangential stiffness of contact
stiffness of ball-flat contact
mass of the slider block
mass of the seismic block
number of contact asperities
the number of local peaks which project through a selectable band centred
about the mean line
4 maximum height of the profile above the mean line
% root mean square of the profile ordinates
s normalized coordinate (height/v)
&n mean spacing between profile peaks at the mean line
t time
T total tangential force
W total normal load
x y displacement in the tangential and normal directions
P mean radius of curvature of asperities (l//3= l/p1 + l/h)
constant depending on mass ratio between the contacting bodies
L normal approach
s, tangential approach
e angle of rotation of the slider block
l-1, r2 angular frequencies of the system in the tangential direction
r,, c, rs angular frequencies of the system in the normal direction
Vl, Y Poisson’s ratio of surfaces 1 and 2
V standard deviation of height distribution of asperities
7 correlation distance
64 angular natural frequency of the slider block in normal direction without
contact
e? angular natural frequency of the slider block in tangential direction without
contact
angular natural frequency of the seismic mass in contact with the steel
balls, in normal direction
angular contact frequency in normal and tangential direction

You might also like