Академический Документы
Профессиональный Документы
Культура Документы
S. ODI-OWEI
College of Science and Technology, Port Harcourf (Nigeria
Summary
Wear debris extracted from oil samples taken from a twin-disc machine
has been analysed using Quantimet to obtain data in terms of the number,
size and distribution of wear particles in ferrograms. A correlation between
wear and performance in the disc machine is achieved. From optical and
stereoscan examination of the wear particles, the nature of the initial
running-in wear behaviour of the discs is established.
Introduction
i Troct~onMeasurmg
Both Discs Electrlcolly System
Insulated for Measurement
of Film Thickness
ICapacltancel and
Degree of Metallic
Stationary DISC
- 011 Supply by Grawty
011 Sample
Coilectlon Point
Area I which
Temperature 15
Monitored by
Infra-red Camera
\ I / System I Thermovis~on I
To establish the extent and nature of this wear, and also its possible
effect on performance, oil samples were collected at the outlet side of the
contact and subsequently examined for wear debris by means of ferrogra-
phy. The ferrograph method was originally devised and developed by Seifert
and Westcott [ 21 and has quickly gained prominence as a complementary
technique to that of more established methods such as spectrographic oil
analysis. The usual method associated with the analysis of the wear particles
is optical densitometry, coupled with optical and stereoscan microscope
examination. A recent contribution to the study of wear in sliding steel sur-
faces has attempted to clarify and classify the range of wear modes which
can occur in this type of behaviour [ 31.
In the current investigation the optical and stereoscan microscope have
both been utilised to identify the nature of the wear particles. The optical
density of the deposit on the ferrogram can be measured at discrete positions
to provide information about the amount and size distribution of magnetic
particles in the oil. It has been shown that the distribution of particle sizes is
a sensitive indication of wear mode in a number of situations. It was con-
sidered that an alternative, though similar, approach might be investigated in
which an image analysing system such as Quantimet would be utilised to
provide information about the number, size and distribution of the particles.
This would also enable the feasibility of correlating wear with performance
data to be assessed.
239
2LO r
200 1 Load
. 445N (p.=O 15 GPO)
+ 890N (0 21 GPal
A I 33 kN 10 26 GPO)
+ I 78 kN IO 30 GPa)
160 I 0
.
2 22 kN 10 33
2 67 kN(0 36
GPal
GPal
80
0 I_-
0 I 2 3
SItdIng Speed .m/s
Test results
The main purpose of the test programme was to establish the traction
performance between a stationary disc loaded against a disc rotating at cons-
tant speed. Provision was made for testing with different lubricants
under the action of a steadily applied load. Traction performance was deter-
mined in terms of the displacement of the upper stationary disc as measured
241
Sliding Speed,m/s
500 Load
0 890N
A 133kN
+ 178kN
0 222kN
I 2 3 1
Shdlng Speed, m/s
Fig. 5. Degree of inter-asperity contact for various operating conditions (HVI 650 oil).
General details
The discs were EN24 steel, hardened and oil-tempered to 780 HV, 300
mm in diameter and 9.4 mm wide. The initial surface roughness was 0.06 pm,
a super-ground finish, and the initial out of roundness was 0.01 mm.
The oil was HVI 650 (as supplied by Shell Research Centre, Thornton),
with viscosities of 568 X 10M6 m2 s-l at 38 C and 32.8 X 10v6 m2 s--l at
100 C. The tests were performed in the speed range 0.25 - 4 m s-l with
loads from 0.22 kN to 2.6 kN.
243
Fig. 6. Comparison of experimental and theoretical values of film thicknesses (HVI 650
oil).
Microscope examination
Load of 445 N
Two distinct particle shapes were viewed at the lowest speed of 0.6
m s-l. The predominant particle shape was the platelet type which was
244
Load of 890 N
The particle distribution at the lowest speed of 0.6 m s-l was similar
to that of the 445 N load, i.e. platelet stringers with some associated spheres
(Fig. 10). Increasing the speed to 1.27 m s-l produced a small increase in the
particle density. The distribution, however, remained essentially the same
although the platelets were somewhat larger and more well-defined.
A further increase in particle density occurred at 1.9 m s-l. The
spherical particles were no longer present but some cutting wear particles
were found, mostly associated with the platelet stringers (Fig. 11) although a
few appeared as discrete particles. The presence of spongy oxide particles
within the stringers was also observed.
The oxide particle content increased with increase in speed to 3.2 m s-l,
the oxides being a mixture of FesOs and FesOd. The form of the debris other-
wise remained much the same, namely, rubbing wear platelets in stringer
form with some cutting wear particles. The overall particle density, however,
was far less than that for the previous running speed.
246
TABLE 1
Summary of observations from microscope examination
Load of 1.78 kN
Large cutting wear particles were present at the lowest speed of 0.6 m
s-l. The cutting wear particles were found within and separate from the
platelet stringers, the stringers also acting as sites for spongy oxide particles.
The cutting wear particles were absent at the speed of 1.27 m s-l, the
field of view consisting entirely of large rubbing wear platelets and oxide par-
ticles. The platelets were mainly in the form of stringers but some were
distributed as discrete particles. The oxide particles, which were found to be
mainly of FesO, but with some Fes04, were again associated with the plate-
let stringers.
An increase in particle density was apparent when the speed was
increased to 1.9 m s- . The particles remained predominantly rubbing wear
platelet stringers containing some oxide. However, some small cutting wear
particles had reappeared as had a few spherical particles.
The particle distribution at 3.2 m s-l remained essentially the same as
for 1.9 m s-l , i.e. rubbing wear platelet stringers containing oxide plus the
occasional cutting wear particle or sphere. The particle density, however, was
somewhat less.
A summary of these findings is presented in Table 1.
247
@
A3 L45 N
83 890 N
C3 1.78 kN
Speed - 1.9 m/s
Oil -H VI 650
780 VPN Discs
50
0 5 10 15 20 25 30
1
\
100 Slldlng Speed
Cl 06 ms-'
c2 127 II
c3 190 ,,
90 -
CL 320 ,,
Dl ,, ~Control)
Load - 178 kN
011. HVI 650
30 r
0 5 10 15 20 25 30
PosItion along Sllde.mm
, -
& 1 10 kN
IL
0 1 2 3 4
Sliding Speed m/s
ratio of particles to contacts when plotted against speed (Fig. 16) provides
an indication of the wear in relation to performance. It is evident that, as
before, the behaviour at light load differs from that observed under heavier
load at a later stage in the test.
Discussion
cl 3 PP
0 5 PP
A 8 PP Load = 890 N
V 10 PP
x 15 PP
i 2 3 4
Slfdtng Speed, m/s
tion of particles along the slide (Figs. 12 and 13) an indistinct pattern of
behaviour emerges which suggests a change in wear conditions as operating
conditions vary. A more clearly defined pattern is apparent in Fig. 14 in
which the behaviour at light load differs from conditions at higher load. An
examination of particle size (Fig. 15) fails to reveal any clear indication of
size effects as the film thickness and rate of contact change.
When the number of particles is related to asperity contact (Fig. 16) the
basic difference in behaviour noted in Fig. 14 becomes more clearly defined.
To try and explain this difference requires a knowledge of the nature of the
particles, as determined from examination with the optical and stereoscan
microscopes.
The different particle densities and particle shapes suggest that two
different wear regimes dominate which are linked by a transitional type of
behaviour. A mild wear regime is indicated by the flat platelet particles. This
contrasts with a severe wear regime as indicated by cutting wear chips or
swarf.
At the lowest load (445 N) a mild wear condition predominates in the
initial stages. This gradually changes to severe wear as speed increases. It is
251
890 N
I
0
A 1.78 kN
0 I I -
0 1 2 3 4
considered that in this instance the cutting wear could have been caused by
one of two possible mechanisms. Firstly the discs were hard and the final
surface condition prior to testing was obtained by grinding using an
aluminium oxide wheel. Examination of the ferrograms did not yield any
clear evidence of foreign grit of this type. This would provide a suitable
source for an abrasive cutting medium. An alternative explanation is that
cracking of the surface layers occurred due to a fatigue action resulting from
the grinding operation. If the latter hypothesis were correct it would help to
explain the gradual build-up in severe wear cutting action observed during
the initial period of the test.
During the examination of the ferrograms two types of spherical par-
ticles were observed, one having a completely smooth surface as shown in
Fig. 8, the other having a surface seemingly made up of overlapping layers
(Fig. 17). The presence of spherical particles is a well documented [9, lo]
feature of fatigue cracking in rolling contact, the particles exhibiting smooth
surfaces. In the present tests, sliding conditions in the contact dominate. The
extent to which this affects the mechanism for forming spherical particles is
not known at this stage.
Fig. 17. Load 445 N, speed 1.27 m s-l (4200 x )
The wear behaviour at the two higher loads, though similar, exhibits
different characteristics from the conditions at the initial load. At low speed,
mild rubbing wear is again in evidence but this gives way to severe wear when
the speed is in the region of 2 m s _ . During the tests an audible knocking
sound was noted in this speed range, the noise subsiding as the speed was
increased above 2 m s-l. This coincides with the indication of a return to
mild wear, in which the film thickness and temperature in the contact were
increasing as speed increased. This is substantiated by evidence of an increase
in the amount of metallic oxide present, which is associated with mild wear
behaviour [ 31.
The overall impression which emerges from Fig. 16 is that the increase
in severity of wear which occurs initially is gradually superseded by a milder
wear mechanism as the test proceeds, even though the load during the test
is increasing. This suggests that the surfaces are experiencing some form of
self-healing process which is indicative of the conditions in the contact
progressing from a metastable to a stable condition as the test proceeds.
This manifests itself in the form of a running-in wear process in which the
surfaces in contact are adapting to their operating environment under con-
ditions of gradually increasing severity in operating condition.
Conclusions
Acknowledgments
The authors wish to thank Shell Research (Mr. G. Galvin) for the
supply of the lubricant and details of the appropriate properties and Mr.
Malcolm Williams (Metallurgy Department, University College Swansea) for
assistance in the use of the Quantimet 720 machine and the subsequent
analysis of results.
For the production of ferrograms the authors are indebted to Mr.
Vernon Westcott of Transonics Inc. and Dr. G. Pocock of the Admiralty
Materials Laboratory. Their helpful comments and contribution to the tech-
nical discussion of the results is greatly appreciated.
References