Vous êtes sur la page 1sur 6

Acta metoll.Vol. 37, No. 3, pp. 161-112,1989 OOOI-6160/89$3.00+O.

OO
Printed in Great Britain Pergamon Press plc

THE EFFECTS OF SLIDING CONDITIONS ON THE


DRY FRICTION OF METALS

S. C. LIM’, M. F. ASHBY’ and J. H. BRUNTON’


‘Department of Mechanical and Production Engineering, National University of Singapore, 10 Kent
Ridge Crescent, Singapore 0511, Republic of Singapore and ‘Department of Engineering, University of
Cambridge, Trumpington Street, Cambridge CB2 IPZ, England

(Received 19 August 1988)

Abstract-A survey of the variations of p in the dry sliding of steel on steel over a wide range of sliding
conditions is presented in the form of a friction-regime map. At slow sliding speeds (u < 1 m/s), the
coefficient of friction p depends strongly on the state of the surface: friction between rough surfaces is
greater than between those which are smooth. At higher velocities (v > 1 m/s), the initial surface condition
is quickly replaced by one characteristic of the wear process, and the coefficient of friction becomes,
increasingly, dependent on the pressure at the surface and the sliding velocity, not the surface state. At
very high loads and velocities, a layer of molten metal forms between the sliding surfaces, reducing p to
very low values.

R&urn&-Un examen d’ensemble des variations de p au tours du glissement B set de I’acier sur l’acier
pour un large spectre de conditions de glissement est prtsentt sous la forme d’une carte de rkgimes de
frottement. Aux vitesses de glissement faibles (v < 1 m/s), le coefficient de frottement p dCpend fortement
de l’ttat de surface: le frottement entre surfaces rugueuses est plus grand qu’entre surfaces deuces. Aux
vitesses plus grandes (v > 1 m/s), la condition initiale de la surface est rapidement rempla&e par un ttat
caracttristique du pro&d6 de glissement, et le coefficient de frottement devient, de plus en plus, d&pendant
de la pression sur la surface et de la vitesse de glissement et non de l’6tat de surface. Aux tr6s fortes charges
et vitesses, une couche de m&al fondu se forme entre les surfaces de glissement ce qui reduit p $ des valeurs
tres faibles.

Zusammeafassung-Es wird ein tiberblick fiber den Reibungskoeffizienten p bei trockener Gleitung von
Stahl iiber Stahl fiir einen grol3en Bereich von Gleitungsbedingungen in Form von einem Diagramm der
Reibungsbereiche vorgelegt. Bei niedrigen Gleitungsgeschwindigkeiten (v i 1 m/s) h5ngt der Koeffizient
stark vom Oberflgchenzustand ab: Reibung zwischen rauhen Oberflbhen ist hiiher als zwischen glatten.
Bei hijheren Geschwindigkeiten (u r 1 m/s) wird die urspriingliche OberflIchenbeschaffenheit durch eine
ersetzt, die fiir den AbtragprozeB charakteristisch ist. Der Reibungskoeffizient wird zunehmend abhingig
vom Druck an der Oberfllche und von der Gleitungsgeschwindigkeit, nicht aber von dem Oberfllchen-
zustand. Bei sehr hohen Lasten und Geschwindigkeiten bildet sich zwischen den gleitenden Oberfl&hen
eine Schicht geschmolzenen Metalles, wodurch auf sehr kleine Werte verringert wird.

1. INTRODUCTION a diagram which summarises, with some physical


explanation, the variation of p with F and v.
The coefficient of friction, p, between a given pair of The physical modelling of unlubricated friction is
sliding surfaces is often treated as a constant. For the not a simple matter [5]. At low sliding velocities
dry sliding of mild steel on itself (for example), (roughly speaking, u < 1 m/s) it is the view of some
Smithells [l], states that p = 0.8; the ASM handbook experts that direct metal-to-metal contact at asper-
[2] also advocates p = 0.8, and the American Institute ities creates adhering patches, like tiny spot welds,
of Physics Handbook [3] gives p = 0.78 (static) and which must be plastically sheared [6] or pulled apart
0.57 (sliding). In reality, p depends on the sliding [7], giving friction. Others, observing the subsurface
conditions, by which we mean the force, F, acting plasticity of worn surfaces [8], link friction with the
normal to the sliding surfaces, the relative velocity, u, plastic work dissipated in this deformation. Still
of the two surfaces, and the state of the surfaces: their others, noting that rough surfaces give a higher value
roughness, the degree to which they are oxidised, and of p than smooth ones, relate p to the forces required
so forth. Over a narrow range of these conditions, p to make the asperities on one surface ride over or
is roughly constant, but over the full range (defined plough through those on the other [9-111 or a
in a moment) it varies more widely than is generally combination of these and adhesion [12]. One must
realised: the coefficient of friction for the dry sliding conclude that, in this slow-sliding regime, the
of steel on steel, for example, can be as small as 0.02 physical mechanism of sliding friction is not yet
and as great as 1.6. In this brief paper we report completely clear.
measurements of the dependence of p on surface At higher velocities (u > 1 m/s), p for unlubricated
roughness, and assemble published data to produce steel decreases with increasing v [13-l 61. The changes
768 LIM et al.: EFFECTS OF SLIDING CONDITIONS ON DRY FRICTION

are associated with frictional heating of the surfaces


which first oxidise and then (as sliding conditions
become more extreme), melt. In these regimes, p can
be related, at least qualitatively, to the mechanisms of
wear.
We now describe some experiments to characterise
further the effect of surface roughness on the friction
between dry steel surfaces, before returning to the
broader question of the dependence of p on sliding
conditions.
L
E 0.6-

%
2. EXPERIMENTAL ki 0.5-
w
A pin-on-disk sliding apparatus operating inside u a4-
the scanning electron microscope (SEM) was used to k AA
IL!
measure ,u while, at the same time, observing the 8 0.3- x

mechanism of wear. Details of the rig are given #


elsewhere [I 71. 8
The case-hardened steel disks had four different 0.1-g 4
0.2-
centre line average (c.1.a.) roughness values: 0.05 pm
(mirror polish), 0.20,0.42 and 0.59 pm. The last three a2 0.3 0.4 0.5 0.6 0.7
roughnesses were obtained by the random pattern SURFACE ROUGHNESS Om’I c.l.03
abrasion of mirror-polished disks with three different Fig. 1. Coefficient of friction p plotted against surface
grades of silicon carbide paper. The frictional force roughness for pure-copper pins sliding on case-hardened
and the normal load on each pin were continuously mild steel disks. Values of p for copper pins sliding on steel
recorded using a datalogging system (sampling rate: disks lubricated with liquid paraffin (a poor boundary
lubricant) obtained by Grunberg and Campbell [9] are also
5 per second). Each test lasted about 15 min, with
shown.
the sliding velocities of between 2 x 10m4 and
4 x 10m4m/s. The nominal contact pressures varied
between 0.01 and 100 MPa.
variations in p are simply caused by variations in
2.1. Copper pins on steel disks surface-roughness in the sliding direction.
Data for p from the copper pin-tests show a 2.2. Iron pins on steel disks
positive correlation with surface roughness (Fig. 1).
When iron pins were used instead of copper, there
At the same time it was observed that the copper
was a distinction between the values of p for mirror-
wear-flakes increased in size, thickness and number
polished disks and for disks which were roughened,
with increasing disk roughness.
but the degree of roughening did not seem to matter
For the roughest disk (0.59pm c.1.a. roughness)
much (Fig. 3). The size and thickness of iron wear-
there was severe accumulation of copper wear-flakes
flakes increased when roughened disks were used [ 191,
on the steel disk, and the average p dropped to 0.5
but they did not depend much on the degree of
(compared with 1.13 for the 0.42 pm-roughness disk).
We think this is because wear flakes clogged the disk,
reducing its roughness and giving copper-on-copper
sliding, not copper-on-steel. Values of p between 0.4
and 0.7 are reported for copper sliding dry on itself
[4, 181.
Figure 1 includes the experimental data of
Grunberg and Campbell [9] who also studied the
sliding of copper on steel. Their data for tests carried
out at 10e4 and 10m3m/s agree reasonably well with
data taken for the mirror-polished and the two
differently roughened disks.
Figure 2 shows the changes in p with time for disks
with the three different roughnesses. The periodic I
variations indicate that the grinding direction affects 0 900
TIME (5)
h (the trace for the mirror-polished disk is essentially
Fig. 2. The variations in coefficient of friction p with time
flat). One would expect the surface roughness to be when pure-copper pins slid on case-hardened mild steel
lower (hence lower p) for sliding parallel to the disks of three different surface roughnesses: (a) mirror-
grinding, and we find that this is so. The periodic finished, (b) 0.20pm c.1.a. and (c) 0.42pm c.1.a.
LIM et al.: EFFECTS OF SLIDING CONDITIONS ON DRY FRICTION 769

1.2 p p we fro” pins I 1 PfJRE IRdN PINS 1


on Sl.2d a*(6 ( ~NsTEEC DISKS~
” =2..3-3.BxrO-‘nJ/s
Present work 1
1.1
X 52100shl bills
on 52100 51ee,
cvlinders

0.9 -

4 o-3-

ZJ
g 0.7.
u
E
0.6 -
0” Y 3””
TIME (5)
g 0.5-
w
z
E 0.4-
w
P Fig. 4. The variations in coefficient of friction p with time
when pure-iron pins slid on case-hardened mild steel disks
of two different surface roughnesses: (a) mirror-finished and
8 0.3- (b) 0.42 pm c.1.a.

diffusivity of the steel (4.1 x lOA m/s); an approxi-


xx
mate scale of absolute velocity is shown across the
top. The normalised velocity (which can be viewed as
the ratio of true velocity to heat-flow velocity) relates
SURFACE ROUGHNESS (JJ~ c.1.a.)
to the temperature of the sliding surface: surface
Fig. 3. Coefficient of friction p plotted against surface heating becomes appreciable above d = lo2 [47].
roughness for pure-iron pins slid on case-hardened mild The thing to note is that, for v < 1 m/s (or d < lo*),
steel disks. Values of p for steel-ball-on-steel-cylinder tests,
under boundary lubrication, conducted at a higher sliding p shows no dependence on sliding velocity. The data
speed of 0.56 m/s obtained by Furey [lo] are also shown. are spread between 0.05 a value typical of a mirror-
smooth surface-and 1.0 or more-typical of a very
rough one. A glance at the list on the right-hand side
roughening. Others find the same thing: Fig. 3 in- of the diagram shows the large number of sources
cludes, for comparison, the data of Fury [lo] from from which that data were drawn; some used rough
steel-ball-on-steel-cylinder tests. These tests used a disks, some smooth. It is this, not F or v which
boundary lubricant (with generally more than 70% of determines p. At high velocities (v > 1 m/s, 5 > 102)
metallic contact) and were at a higher speed things are different. The data in this regime derive (as
(0.56 m/s) than ours, but show a similar trend. before) from disks of differing roughness, but this no
Figure 4 shows two traces, one from a mirror- longer matters much: the coefficient of friction now
polished disk, the other from a disk having 0.42 pm depends strongly on v, and, as the next diagram
c.1.a. roughness. The pattern in the upper trace is due shows, on F as well.
to the grinding direction of the silicon carbide, as The diagram in question is Fig. 6. Here the axes are
before. Traces for other roughened disks showed normalised velocity z?, and normalised pressure
similar patterns.
F
Our data add to the accepted view: at slow sliding p=-
speeds, the coefficient of friction depends on surface 441
roughness, but not on sliding conditions-meaning where A, is the nominal contact area of the pin with
normal load F and sliding velocity u. But at higher the disk (A, = m-i) and Ho is the hardness of the
sliding speeds, the picture changes. material of the pin. It is used (rather than F alone)
because it allows data from different sources, using
different equipment, to be compared in a sensible
3. THE BROADER PICTURE: EFFECT OF
SLIDING CONDITIONS ON /I way. It can be thought of as the fraction of area of
the pin-end which is in true contact with the disk.
Measurements of the coefficient of friction for the Experiments at a given F and i? are plotted as a point
dry sliding of steel on steel, for velocities, v, from 10m4 on the diagram; the symbol indentifies the source.
to lo3 m/s, are shown in Figs 5 and 6 plotted in two Where it was possible to do so, contours of constant
ways. The first shows p, measured mostly in pin- p have to be drawn through the data.
on-disk experiments, plotted against normalised The figure shows that p is almost independent of
velocity [47] Fand d at sliding velocities below 1 m/s (z?= lo*). But
a clear dependence, shown by the contours, emerges
fi=-- ora at velocities higher than this. This behaviour can be
a
understood when frictional heating is examined.
where r0 is the radius of the pin and CIis the thermal Below 5 = IO*, heating is negligible [47]. Just above
110 LIM er al.: EFFECTS OF SLIDING CONDITIONS ON DRY FRICTION

SLIDING VELOCITY v (m/s)


i-2 1 1p2
q Johnson et al 119Lll
STEEL B Bucklev et al I19621
COEFFICIENT OF FRICTION p q Jahonmir et 01119751
PI yonget 01 11965)
+ Dokos (19L61
o Bowden and Persson It9611
B Ouinn 119671
B Mokhtar 119621
e Tokagi 09701
@ Mohdavion and Moi (19641
d GIOU 119601
d Ouinn et 01 119801
0 Suh and Sin 11960
X Morcimok and Otmianowski (19611
W Antler 1196L)
0 Cocks 119581
x Montgomery (19761
+ Eorles and Powell 11966/67)
1( Earles ond Kadhim 11965/661
. Williams and Giffen 11963/6&l
. Suh et al (197&l
I Bhottachoryyo 119.901
\ Arnell et al ll9751
* Gr6tsch and PlokellOLO)
c Cocks l196L)
A Kmsello and Childsi1976)
v Jahanmir et al 119741
X Saks et 01 119771
M Abrohomson et oll197Ll
6 Ramishtili (19611
* Soda and Sosada (19761
1 Present work

NORMAUSED VELOCITY F

Fig. 5. The variations in coefficient of friction p with normalised velocity d for steel. The range of p
obtained in the present work are shown. (The references, in order, are 120-24, 14, 25-30, 12, 31-33, 16,
34, 35, 15, 36461.)

SLIDING VELOCITY v (m/s)

qJ+-pgk
I! Johnson et al 119MI
q Buckley et al 11962)
El Johanmir et al 11975)
II Yang et al I19651
$ Dokos I19461
0 Bowden and Perwn 09611
6 Ouinn 119671
B Mokhtar (19621
@ Tokagi 119701
@f Mohdovion and Mai 1196&I
& Ouinn et 01 1196Ol
v Suh and Sin (19811
w Antler 119611
0 Cocks (19561
x Montgomery (19761
+ Eorles ond Powell ls66/611
% Eorles and Kadhim ll965/661
. Williams and Giffen 11963&I
. Suh et al ll97V.1
‘ Bhattachoryyo 114801
. Arnell et al (19751
l Soda and Sosodo 119761
@ Cocks H96Ll
A Kinsello ond Childs 119761
v Jahanmir et.ol l197Ll
X Sako et 01 119771
n Abrahamson at al 119711
I Present work

1
NORMALISED VELOCITY v

Fig. 6. The variations in coefficient of friction ~1with normalised pressure and normalised velocity for steel.
The numbers given against the points are the values of p. Contours of constant p are superimposed on
the data points at velocities greater than 1 m/s. (The references, in order, are [21X24, 14, 2528, 30, 12,
32, 33, 16, 34, 35, 15, 3638, 46, 4&44].)
LIM ef al.: EFFECTS OF SLIDING CONDITIONS ON DRY FRICTION 771

SLIDING VELOCITY v (m/s)

16* 1 lo*
STEEL
FRICTION-REGIME MAP

SEIZURE

NORMALBED VELOCITY 5

Fig. 7. The friction-regime map for the unlubricated sliding of steel surfaces. Contours of constant /J
(replotted from Fig. 6) are superimposed on the different regimes of different frictional behaviour. These
regimes are separated by field boundaries taken from the wear-mechanism map for steel [47]. The limits
of p for most engineering applications involving steel sliding surfaces are given.

(6 = lo’-103) asperity heating causes mild oxidation surface roughness together with material properties
of the surface [48], reducing the amount of metal-to- such as modulus, yield strength [22] hardness [52],
metal contact. As the velocity increases further, surface energy [53] determine the friction. Many
higher flash temperatures generate more oxide until models have been proposed but no complete
the entire surface is covered with a continuous film of description is yet available.
plastically deformed oxide [ 15,491 reducing p further.
At very high loads and speeds, surface melting gives 4. CONCLUSIONS
“melt lubrication” [39,50, 511 which reduces p to
very low values; as low as 0.02 have been reported The coefficient of sliding friction, p, between dry
u41. metal surfaces is determined, at low sliding speeds
These findings are summarised by a friction-regime (u < 1 m/s for steel) by surface roughness and by the
map (in Fig. 7). There are three main regimes: plastic (and perhaps elastic) properties of the sur-
faces. At higher speeds (u > 1 m/s for steel), the
1. Seizure
surface condition is modified by local heating (which
2. Plasticity (or roughness) controlled friction
can cause oxidation or even melting); then /J depends
3. Sliding-condition controlled friction
in a reproducable way on sliding velocity and bearing
The last regime can be further subdivided into pressure F/A,. The higher-velocity regime can be
further subdivided, according to the extent of ox-
1. Mild-oxidation controlled
idation and degree of melting. These observations are
2. Severe-oxidation controlled
summarised in a “friction-regime map” which shows
3. Melt controlled.
the field of Y’and Pin which each regime is dominant,
The boundaries between the regimes are taken from and the way in which p depends on them.
the wear-mechanism map for steel [47].
The picture is clear in the sliding-condition con-
trolled regime. In these situations, the sliding is Acknowledgemenu-Part of this work was carried out while
one of the authors (SCL) was at Cambridge University. The
influenced by surface heating and the surfaces are award of a scholarship by the National University of
separated by either a layer of oxide or a layer of Singapore for SCL to undertake this piece of work is
molten metal. In the plasticity controlled regime gratefully acknowledged.
112 LIM et al.: EFFECTS OF SLIDING CONDITIONS ON DRY FRICTION

REFERENCES 28. S. M. Mahdavian and Y. W. Mai, Wear 95, 35 (1984).


29. P. J. Blau, presented at ASME Winter Meet, Chicago,
C. J. Smithells, Metals Reference Book, 4th edn, Ill (1980).
Butterworths, London (1967). 30. T. F. J. Ouinn. D. M. Rowson and J. L. Sullivan, Wear
ASM Handbook. Am. Sot. Metals, Metals Park Ohio. 65, 1 (1980).
American Institute of Physics Handbook, (Edited by 31. J. Marciniak and T. Otmianowski, Proc. 3rd Znt. Tribol.
D. E. Gray), 3rd edn. McGraw Hill, New York (1972). Corm. (Eurotrib 81). Warsaw. Poland, The Polish
F. P. Bowden and D. Tabor, The Friction and Lubri- Tribvology Council, Vol. I/A, p..241 (1981).
cation of SoUs, Part II, Clarendon Press, Oxford 32. M. Antler, Wear 7, 181 (1964).
(1964). 33. M. Cocks, A.S.L.E. Trans. 1, 101 (1958).
5. T. H. C. Childs, Tribol. Int. 13, 285 (1980). 34. S. W. E. Earles and D. G. Powell, Proc. Inst. Mech.
6. D. Tabor, Trans. A.S.M.E., J. Lub. Tech. 103, 169 Engrs 180, 16 (1966/67).
(1981). 35. S. W. E. Earles and M. J. Kadhim, Proc. Inst. Mech.
I. I. V. Kragelsky, M. N. Bodychin and V. S. Kombalov, Engrs 180, 531 (1965/66).
translated by N. Standen, Friction and Wear, Calcu- 36. N. P. Suh, S. Jahanmir, E. P. Abrahamson II and A. P.
lation Methods. Pergamon Press, Oxford (1982). L. Turner, Trans, A.S.M.E., J. Lub. Tech. 96, 631
8. D. A. Rigney and J. P. Hirth, Wear 53, 345 (1979). (1974).
9. L. Grunberg-and R. B. Campbell, Proc. Co@ Lubri- 37. S. Bhattacharyya, Wear 61, 133 (1980).
cation and Wear. London. 1957. Inst. Mech. Engrs, 38. R. D. Amell, A. P. Herod and D. G. Teer, Wear 31,237
v. 291 (1957). (1975).
10. -M. J. Furey; A.S.L.E. Trans. 6, 49 (1963). 39. G. Griitsch and E. Plake, 2. Schiess. Sprengst. 35, 3
11. T. Tsukizoe and T. Hisakado. Trans. A.S.M.E. J. Lub. (1940).
Tech. 92, 264 (1970). 40. M. docks, J. appl. Phys. 35, 1807 (1964).
12. N. P. Suh and H.-C. Sin, Wear 69, 91 (1981). 41. F. H. Kinsella and T. H. C. Childs. Inst. Mech. Engrs
13. D. Tabor, Proc. Int. Conf. Tribology in the SO’s, NASA Co& 1978(6), 55 (1978).
Lewis Research Centre. Cleveland. Ohio, v. 119 (1983). 42. S. Jahanmir, N. P. Suh and E. P. Abrahamson II, Wear
14. F. P. Bowden and P. A. Persson,’ Proc. R. Soc.‘At60, 28, 235 (1974).
433 (1961). 43. N. Saka, A. M. Eleiche and N. P. Suh, Wear 44, 109
15. K. Williams and E. Giffen, Proc. Inst. Mech. Engrs 178, (1977).
24 (1963/64). 44. E. P. Abrahamson, II, S. Jahanmir, N. P. Suh and D.
16. R. S. Montgomery, Wear 36, 275 (1976). A. Coiling, Proc. Int. Conf on Production Engineering,
17. S. C. Lim and J. H. Brunton, Wear 101, 81 (1985). Tokyo, 1974, Japan Sot. Precision Engng CIRP,
18. M. 0. A. Mokhtar, M. Zaki and G. S. A. Shawki, Tokyo, Part 1, p. 408 (1974).
Tribol. Znt. 12, 265 (1979). 45. G. Ramishvili, Proc. 3rdInt. Triboiogy Congr. (Eurotrib
19. S. C. Lim and J. H. Brunton, Wear 113, 383 (1986). 81) Warsaw, Poland. The Polish Tribology Council,
20. R. L. Johnson, M. A. Swikert and E. E. Bisson, NACA Vol. I/A, p. 287 (1981).
Tech. Note, No. 1442, National Advisory Committee 46. N. Soda and T. Sasada, Trans. A.S.M.E. J. Lub. Tech.
for Aeronautics, Washington (1947). 100,492 (1978).
21. D. H. Buckley, M. A. Swikert and R. L. Johnson, 47. S. C. Lim and M. F. Ashby, Actu metall. 35, 1 (1987).
A.S.L.E. Trans. 5, 8 (1962). 48. T. F. J. Quinn, Wear 18, 413 (1971).
22. S. Jahanmir, N. P. Suh and E. P. Abrahamson II, Weur 49. D. G. Powell and S. W. E. Earles, A.S.L.E. Trans. 11,
32, 33 (1975). 101 (1968).
23. Z. Y. Yang, M. G. S. Naylor and D. A. Rigney, Wear 50, E. Saibel, R. Aircraft Es&b. Trans. no. 391 (1951).
105, 73 (1985). 51. B. Sternlicht and H. Avkarian, A.S.L.E. Trans. 2, 248
24. S. J. Dokos, Trans. A.S.M.E. J. appl. Mech. 68, A 146 (1960).
(1946). 52. A. J. W. Moore and W. J. McG. Tegart, Proc. R. Sot.
25. T. F. J. Quinn, A.S.L.E. Trans. 10, 158 (1967). A212, 452 (1952).
26. M. 0. A. Mokhtar, Wear 78, 297 (1982). 53. E. Rabinowicz, Friction and Wear of Materials, Wiley,
27. R. Takagi, A.S.L.E. Trans. 13, 87 (1970). New York (1965).

Vous aimerez peut-être aussi