Beruflich Dokumente
Kultur Dokumente
WEAR OF ELASTOMERIC
BRUCE H. BURR
Product Research Group, Hughes Tool Division, Hughes Tool Company, Box 2539,
Houston, TX 77001 (U.S.A.)
KURT M. MARSHEK
Department of Mechanical Engineering,
78712 (U.S.A.)
Summary
An empirical equation was obtained via dimensional analysis, which
relates the abrasive wear volume to the friction force, specimen load, sliding
distance and specimen breaking strength for O-ring materials. Wear experiments on O-rings molded from four nitrile compounds and one polyurethane
material were conducted on a special pin-disc-type
testing machine. Specimens cut from a size 330 O-ring were held against a roughened rotating steel
wear cylinder with a load which varied from 5 to 15 lbf. Both the specimen
and the wear cylinder were immersed in an abrasive mud of the type used for
oil well drilling. The sliding velocity was held constant at 11 in s-l. The wear
resistance of the polyurethane
was two times better than the best nitrile
compound.
1. Introduction
The literature reveals many simple expressions for predicting the wear
rates of various types of materials. For sliding electrical contacts, Archard
and Hirst [l] related the wear volume to the sliding distance, normal load
and yield pressure of the softer material. Ratner et al. [ 21 proposed a relationship for plastic materials which showed that the wear volume depended
on the breaking elongation, breaking strength, hardness and coefficient of
friction. Kar and Bahadur [3] used experimental
data and dimensional
analysis to develop an equation for the volumetric wear rate of filled and unfilled polyoxymethylene.
The literature suggests that material properties such as yield pressure,
hardness, breaking strength, elastic modulus, specific heat and thermal conductivity are important in predicting wear volume. The work presented in
this paper uses the techniques of dimensional analysis to establish relation0043-1648/82/0000-0000/$02.75
348
ships
349
2. Experimental equipment
2. I. O-ring specimens
Physical properties of the nitrile compounds and polyurethane
are
listed in Table 1. Size 330 O-rings were chosen for the experimental
work because they are used to seal standard-size rock bit bearings. This O-ring has a
nominal 2$ in inside diameter and r&in diameter cross section.
TABLE 1
Properties of O-ring materials
~u~erial
Shore A hardness
Breaking strength
(lbf iK2)
Specific gravity
NBR 1
NBR 2
NBR 3
NBR 4
Urethane
80
70
80
90
90
2367
1516
2093
2381
4078
1.28
1.28
1.17
1.34
1.17
,-Loading
Arm
Pulley
Abrasive
Fluid
Holder
WearCylinder
Weight
3. Experimental
results
A series of wear tests, where the specimen volume loss was measured as
a function of time, was run for each O-ring material. The test results were
then plotted as the logarithm of volume loss uersus the logarithm of time
(Figs. 4 - 8). A series of wear tests, where the specimen volume loss was measured as a function of specimen load, was also run for each material. The
results of these tests were then plotted as the logarithm of volume loss uersus
the logarithm of load (Figs. 9 - 13). The experimental
procedure is described
in ref. 4.
1.
[;
3.0
0.4 Y
3
2 0.2 -
0.1 1
2
I
4
III
6 8 IO
20
TIME,
40
I I I
608OKJo
200
10
MINUTES
Fig. 4. Volume loss vs. time for NBR 1: - - -, analytically predicted; -c-,
periment.
from ex-
351
0.1
II,
6 IO
20
40
III
6060100
200
)O
TIME, MINUTES
Fig. 5. Volume loss us. time for NBR 2: - - -, analytically predicted; -c-,
periment.
0.1
66D
20
TIME,
40
II,
6060100
from ex-
200
4 IO
MINUTES
Fig. 6. Volume loss us. time for NBR 3: - - -, analytically predicted; -G-,
periment .
from ex-
The full straight line on each graph is a least-squares fit of the experimental data. The broken straight line is the relationship predicted by a wear
equation derived using dimensional analysis.
As in all wear testing, there was some scatter in the results. The scatter
is caused by small differences in test parameters such as (1) the surface finish
on the cylinder and O-ring, (2) sliding surface material properties and (3)
abrasive fluid consistency.
The graphs of volume loss uersus time (Figs. 4 - 8) indicate that the
initial wear rate during the first 6 min is greater than the subsequent wear
rate. This is a result of the sharp asperities on the sand-blasted wear cylinder,
which provide a high degree of two-body abrasive wear during the first few
0.1 1
2
,,I
680
20
TIME,
40
II,
608000
20
4 10
MINUTES
predicted; -o-,
from ex.
m;i!;
3
9
/J
0.2-
0.12
6I
8/ IO
I
20
/
TIME,
40
608oloO
I n I
200
LK 0
MNUTES
- - - , analytically
minutes. The peaks of these asperities are quickly worn down by the abrasive
particles and the lapping action of the elastomer. The valleys of the sandblasted wear cylinder surface remain and draw abrasive particles in between
the wear surfaces. Three-body abrasion then emerges as the predominant and
steady state mode of abrasive wear.
The graphs of volume loss versus load (Figs. 9 - 13) show that wear is a
power function of load, and analysis predicts an average load exponent of
1.35. This unusual relationship could be caused by the draping effect of the
elastomer surface over asperities. An increased load would cause more draping and thus even deeper penetration of asperities into the elastomer.
4.0
3.0
2.0
-i
2
1.0
lfj 0.8
8
_I 0.6
!I OA
G
0.2
IO
15 20
1520
LOAD, LB,
LOAD,
from ex-
Fig. 10. Volume loss us. load for NBR 2: - - -, analytically predicted; -G-,
periment.
0
x
7
4.0
4.0 IC
3.0
3.0
2.0
20
LBF
from ex-
%
I..0
-is
x
0.6
1.0
4 0.6
9
%
0.6
g 0.4
!g 0.4
0.8
9
0.2
0.2
0. I
IO
15 20
0.1 2
40
I
_
I
_
I
_
I
_
,
_
IU
19
I
_ ^
zo
LOAD, LBF
LOAD,
LB,
Fig. 11. Volume loss us. load for NBR 3: - - -, analytically predicted; u,
periment.
from ex-
Fig. 12. Volume loss us. load for NBR 4: - - -, analytically predicted; -G-,
periment.
from ex-
?k
-
l.O-
CJ
0.8-
0.6
if
$
0.4
0.2
0.1
i-
01
I1
IO
LOAD,
1520
LBF
4. Dimensional
- - -, analytically
predicted;
+,
from ex-
analysis
v =h(f, w,x, S)
In dimensionless
V
-=
x3
&
(1)
sx2
i WW
The dimensionless
relation:
1
variables
(2)
may be assumed
(3)
355
TABLE 2
Experimentally
Material
NBR 1
NBR 2
NBR 3
NBR 4
Urethane
1.42
1.14
1.51
1.40
1.29
mAVE
1.35
STDb
nAVE
STD
0.14
0.36
0.44
0.29
0.38
0.28
0.35
0.07
aAVE, average.
bSTD, standard deviation.
TABLE 3
Wear constant K and related variables
Material
V (in3)
at 60 min
W (lbf)
x (in)
at 60 min
S (lbf inm2)
NBR 1
NBR 2
NBR 3
NBR 4
Urethane
1.24
1.86
1.33
1.34
0.65
0.366
0.404
0.505
0.492
0.305
10
10
10
10
10
4.17
4.17
4.17
4.17
4.17
2367
1516
2093
2381
4078
0.0041
0.0034
0.0038
0.0045
0.0044
x
x
x
x
x
1O-4
10-4
10-4
10-4
1O-4
x
x
x
x
x
104
104
104
104
104
aThe coefficient of friction values are the averages of all the coefficient of friction values
for each material at 6 min of testing.
or
V=Kf"WmxnSb
(4)
where K is a dimensionless wear constant, m = - a - b and n = 3 + 2b. Values for m and n were obtained from the least-square fits illustrated in Figs.
4 - 13 and are summarized in Table 2.
If the average values for m and n from Table 2 are used and the coefficient of friction is defined as 1_1=f/W,eqn. (4) becomes
V=Kp-
0.025w1.325x0.35~1.325
(5)
The wear constant K for each material was calculated using eqn. (5);
the data and material properties are given in Table 3. The K values are also
listed in Table 3.
The wear constants K and friction coefficients (u were then substituted into eqn. (5) to yield the wear curves predicted by dimensional
analysis. These curves are plotted as broken lines alongside the actual experimental data points (see Figs. 4 - 13).
Inspection of eqn. (5) shows the importance of specimen load, specimen
breaking strength and sliding distance. The friction coefficient has very little
effect on wear, and this variable could be eliminated from the equation. The
356
appropriateness
of the choice of variables, selected for dimensional analysis,
is established by the agreement between the experimental
data obtained
from the laboratory and the results predicted by the dimensional analysis.
To use eqn. (5), the wear constant K must first be established by running one wear test. This wear constant may then be used in the equation
to predict specimen wear for varying conditions of load and sliding distance
or time. In dynamic O-ring service, the specimen load is related to the surface contact stress resulting from the squeeze imposed on the O-ring by its
gland.
5. Conclusions
An equation has been developed which shows that the abrasive wear
of an O-ring is related to the material breaking strength, load, sliding distance, coefficient of friction and a material wear constant. If one relatively
short wear test is made first to determine the material wear constant, the
equation may then be used to make predictions of the wear volume loss for
varying conditions of the elastomer surface contact stress and time.
References
1 J. F. Archard and W. Hirst, The wear of metals under unlubricated conditions, hoc.
R. Sot. London, Ser. A, 236 (1956) 397 - 410.
2 S. B. Ratner, I. I. Farberova, 0. V. Rayukevich and E. G. Lurie, Connection between
wear-resistance of plastics and other mechanical properties, Sou. Plast., 7 (1964) 37.
3 M. K. Kar and S. Bahadur, The wear equation for unfilled and filled polyoxymethylene,
Wear, 30 (1974) 337 - 348.
4 B. H. Burr and K. M. Marshek, O-ring abrasive wear test machine, Wear, 68 (1981)
21 - 32.
5 G. Murphy, Similitude in Engineering, Ronald Press, New York, 1950.
6 M. J. Moran and K. M. Marshek, Some matrix aspects of generalized dimensional
analysis, J. Eng. Math., 6 (1972) 291 - 303.