EFFECT OF WEAR PARTICLES ON THE WEAR RATE OF UNLUBRICATED SLIDING METALS RIITSU
Laboratory, Tokyo fJafxm,i January 2, 1962f
SUMMARY By using the Okoshi wear machine, in which a hard peripheral surface of a rotating disc is pressed against the specimen plane with a load increasing in proportion to the square root of the sliding distance, the effect of wear particles in raising the wear rate K of unlubricated sliding metals is examined. After some preliminary experiments to ascertain the constancy of K during a wear test, velocity dependence, in the range of 0.05 - 3.6 m/set, and loading dependence, in the range of the end load 1.1 N 9.9 kg, of h: are determined. In the f( vs. sliding speed curve there is a maximum in the low speed range, which shifts to a lower speed when the loading is raised. If the wear particles adhering to the sliding surface of a disc are continuously removed by a wipei, the rise of If in the lower speed range disappears. It is concluded that the abrasive action of wear particles gives rise to the higher K in the lower speed range, where centrifugai force is ineffective in removing the wear particies from the rotating dii periphery. ZUSAMMENFRSSUNG Der Einfluss von Verschleissteilchen auf die Erhohung der Schleissgeschwindigkeitskonstante ice fur den ungeschmierten Gleitverschleiss von Metallen wurde untersucht. Die Untersuchung wurde hierbei wird die harte Seitenflache einer mit der Okoshi Verschleissmaschine ausgefiihrt; zylindrischen Scbeibe gegen eine flache Probe gepresst wahrend die Belastung mit der Quadratwurzel der Gleitstracke zunimmt. In Vorversuchen wurde die Unver&nderlichkeit der Konstante K wshrend eines Verschleissversuches bestltigt. Weiter wurcle der Einfluss der Geschwindigkeit zwischeno.05 m 3.6 m/sot und der Belastung zwischen den Endlasten I. r N 9.9 kg untersucht. Wird die Konstante K gegen die Gleitgeschwindigkeit aufgetragen so ergibt sich ein Maximum filr ziemlich niedrige Gescbwindigkeiten, das mit steigender Belastung nach noch kleineren Geschwindigkeiten ver&iebt. Wenn-jedoch der Schlei&aub mit einem Wischer fottdauernd von der Probefl%che beseitigt wird, verschwindet such diese Steigerung des K-Wertes. Diese Abriebwirkung des Verscldeiss-staubes ist bei niedrigen ~s~hwin~~keiten zu beobachten, weil dann die Zentrifugalkraft der rotierenden Scheibe noch nicht ausreicht urn die Teilchen van der Seitenff%he de; Scheibe weg~usc~leudem.
first proposed a quantitative wear formula for a sliding metal w=
where W is the wear volume, P is the load applied, s is the sliding distance,, and K is a proportionality constant. As he originally conceived the process, wear takes place atom by atom. Although it is now known that wear fragments are far larger than a single atom, this formula is believed to hold so long as all the surrounding conditions remain unchanged%. It is formally analogous to Amontons’ law of friction and may be explained by assuming that the wear volume per unit sliding length is proportional We&v,5 (~962) 435-445
K. TAKAGl, Y. TSlIYA
to the area of real contact, regardless of the magnitude of the apparent contact area. Results of wear experiments carried out by a large number of workers since the pioneering work of DIES3 show that a. wear rate is not really a simple ~~benomenon~ but that K, defined by eqn. (r], depends upon the load, contaci area (or pressure), sliding speed, sliding distance, actual temperature incidental to these sliding c(JIlditions, and other less obvious factors. In the sliding of metals, wear particles play an important role in preventing the adhesion of sliding materials, thereby reducing the wear rate, while at the same time, wear particles tend to oxidize to an oxide harder than the original metal, thereht increasing the wear rate by an abrasive action. It is desirable to solve quantitatively these little understood, hypothetical, effects of wear particles. The object of this paper is to clarify the situation by wear tests of copper alioys pressed against a cast iron or steel rotating disc, varying the load and speed in particular, but keeping the sliding conditions constant througl~o~lt each wear test (shown by the fact that the constant K is independent of the sliding distance). OKOSHI WEAK MACHINE
There are various ways of measuring the wear resistance of sliding materiaIs, but the method initiated by SAWINS is of particular value in indicating readily the wear resistance of the superficial layer of the specimen material. He loads a definite weight radially to the rotating disc, whose peripheral surface is in close contact with the stationary specimen plane surface (Fig. I). .4fter ro,ooo rotations of the disc the
I. Principle of the \vear machine.
magnitude of the rectangular wear trace produced on the specimen surface is measured, the results being given as (W x 3,00o~~10,000, a wear coefficient, where tl’ is the volume worn away in 0.00x mm3. In SAWIN’s method, a large pressure is exerted on a specimen at the beginning of a wear test and this gradually decreases as the wear trace develops. When a material under test wears by different wear mechanisms as the applied pressure decreases, it is desirable to keep the pressure constant during the test. Consequently, Prof. 0KOsK15 devised an apparatus in which the load P applied on the specimen pressed against the peripheral surface of a rotating disc increases so that P is always proportional to the square root of the sliding distance s. Then,
is a constant. The volume of the specimen material sliding for a distance 2s under the load P, is, from eqn. (If2 dW = K’P’&
437 lost by wear &V, after (3)
the wear rate of the material where K, “specific wear”, is a measure representing under test. It is, in principle, a function of sliding speed and pressure, provided that other conditions remain unchanged during a wear test. Combining eqns. (I) and (z), and integrating with the initial condition, IV =T o when s = o, we obtain FV=*K.p*S w When the sliding distance is s and the load is P, K can be calculated by eqn, (4) from a measurement of the wear volume W at the end of a wear test, or by
W can be ~orn~tri~~ll~ calculated from the radius P and thickness disc and the length of the wear rectangle I: Bl3 w zz -I2Y Pressure p between
and kept nearly In calculating attention should zero to any final
the disc and specimen
23 of the rotating
constant throughout a wear test. the specific wear K from the results of a wear test by this machine, be paid to the basic ~sumption that eqn. (3) holds for loads from value during a test at a given combination of materials. EXPERIMENTAL
Materials Materials from which the rotating discs and the specimen were made are listed in Table I with their hardness values. The chemical camposition of some of the important materials is also given. TABLE
~~~e~~~l ._ D&SC
Cast iron FC35 Tool steel SK? .--.-_-_ Specimfm
Cast silzin bronze
R. TAKAGI, Y. TSUYA
Specimen surfaces were ground flat and polished with 3, a, . . ., 4/o Buehler emery papers lubricated with kerosene. The peripheral surface of the disc was, when necessary, ground by a flat flank of a rotating grinder (WA No. 120), which was itself subjected to a preliminary dressing operation by diamond whenever a new disc was to be set. The load reading on a specimen apparently increases with the sliding distance reading as shown in Fig. a. A series of experimental points is shown in Fig. 2 by small circles, while a parabolic curve is drawn so as to pass through most of the
Sliiirg distance reading x (diMsims)
distance relationship in a reduced form. Readings are proportional values.
circles,y2 = 9.16 (x + 0.92). The curve does not pass strictly through the origin and it deviates from the actual loading near the origin of the graph in such a way that y is preset to 5 divisions to ensure physical stability, instead of to 2.90 at x = o. This causes minor errors which can be ignored when the experimental sliding distance is sufficiently long. The Okoshi apparatus permits five different values of actual loading and also of actual sliding distance to be independently selected by exchanging pairs of mating gears, so that there are 25 feasible P-s curves. In practice only five representative combinations were used. In these five combinations P and s have, at the end of a test, when s reaches 200 divisions, the values listed in Table II. For every combination in Table II, the product P‘s, which appears in equation (5), is the same at each point on the curve in Fig. 2, and this allows a curve to be constructed illustrating the dependence of the wear on pressure. In this work a sliding distance of 50 divisions was often adopted in order to obtain a value of 1 suitable for reading in a short test time; TABLE
WEAR RATE OF UNLUBRICATED
P and s then have values one half and one fourth respectively of those given in Table II. In the wear test shown in Fig. I, wear particles of the specimen adhering to the hard surface of the disc take part in the further wearing of the specimen after a further single rotation of the disc. The wear particles adhering to the disc quickly increase to an equilibrium amount, excess powder falling off. To examine the effect of the adhering wear particles a wiper and a grinder were used in the following ways : Method
Remarks Wear particles build up on the disc
Wear particles lightly attached to the disc surface are removed Sliding surface of the disc is continuity ground by flat flank of a grinder
The wiper was a wad of cotton lightly pressed against the disc periphery at the position X or Z in Fig. I. The rotating grinder (WA No. 120) continuously advanced to the disc at a rate approximating to 0.x mm per IOO m of sliding distance in method M3. WEAR TEST RESULTS
Reliabilkty of the wear test Figure 3 shows a series of .l values obtained by repeating wear tests in the same P-s condition successively with the same disc surface, untreated after an initial grinding. The sliding speed was kept constant at 0.21 mjsec, and final values of P and s are 3.3 kg and 50 m, respectively (C3). The figure shows a tendency to gradual increase of wear as a result of repeated use of the disc. This is probably so because the wear debris adhering to the disc surface slides on the specimen. This condition constitutes Ml M3 Peel-FC35Q AIecl*< b 1 SzBC2- II 0 .
Fig. 3. Examination
of the wear test. Wear, 5 (v&zf
R. TAKAGI, Y. TSUYA
a similar metal pair which is known to produce severe wear. The outstanding example is that of curve I in the combination of PBCI - FC35. Here, black fine powder, supposedly connected with less adhesion, falls during the first and the second tests. During the third and following tests, squeaking begins before the end of a test and coarse, bright, metallic powder falls off abundantly. In the case of the combination AlBCr - FC35, curve II, the wear rate increases steadily up to the tenth run, accompanied by the fall of black powder. If the sliding surface of the disc is vigorous13 rubbed with dry cotton, there is no evident resultant change in the appea.rance of the surface but some unknown change, possibly due to transfer of some lubricating impurity, takes place on the disc surface and the wear rate suddenly decreases (the eleventh run). The wear test was then changed from Mr to M3 (the thirteenth run), and an approximately constant value of I was obtained. Specimens of the other metals tested show generally similar behaviour. Since 1/1.is proportional to 13, ant’ can estimate, from eqn. (6), the large fluctuations of the wear rate, which are clearl) beyond the limits of experimental error. In the case SzRC2-FC35, curve III, however, an approximately constant value is obtained throughout by method MI except for the slight rise after the first test. Method M3 shows a lower value at the seventh and tight runs. On returning to the condition Mr, the wear rate measured originally is reproduced. It is quite impossible to determine a definite wear value in the case of wear behaviour such as that shown by curves I or II. Although reasonably stationary wear 6-
Sliding dis+_atxes h)
4. Z-S relationship.
Sliding speed 0.21 mjsec. P-s: C3. The results and arrows indicate the order of measurement.
for x arc very irregular
WEAR RATE OF UNLUBXICATED
values may be obtained by repeating wear tests either by method MX with preliminary grinding of the disc surface before each test, or by method M3, it will be of iittle value to discuss the wear rate in terms of K value calculated from eqn. (5), using final P and s values, if the sliding conditions that decide the wear mechanism are not kept constant throughout a wear test. Variation of wear rate with the sliding distance Figure 4 shows the l-s relationship in the wear test under MI, Mz and M3 conditions. Each point in the figure except for A represents an independent test result obtained after sliding uninte~uptedly up to a predetermined s. By combining eqns. (2), (5) and (6), E is found theoretic~ly to be proportions to s*. Curve I rises faster, and curves II and III slower, than the curve I oc sg of X = const. The former observation must be due to transfer of wear debris to the disc and subsequent severe wear; the latter to the gradual decrease of K as the contact area grows larger, as occurs frequently when adhesion is not marked. In the case of specimen SzBC2, wear tests were similarly conducted, as shown by curves IV and V. In both methods MI and Mz, 1 is proportional to s* for both disc materials, FC35 and SK2. In one case in Fig. 4, A, (i.e. method M2 and disc SK2), the values of 1 were measured on a single specimen, halting the rotation of the disc appropriately to follow the growth of the same wear trace. The values show some deviation from a straight line. The interruption of smooth rotation in order to measure the length 1 by using dividers, often leaving slight scars, may have affected the ensuing wear test results. It is shown clearly that the different disc material used here does not affect the wear rate of SzBC2 and that the curve of method M2 lies considerably lower than that of MI. The weight of the specimens before and after the wear tests as shown by o (MI, SzBC2-FC35) in Fig. 4, was measured by an automatic balance. The resultant weight losses, divided by the experimentally obtained density 8.14 g/ems, are plotted in Fig. 5. A straight line drawn in the figure represents eqn. (6), and this is in complete agreement with the weight loss measurements. Therefore, it can be stated
3 Lefgth ofwca-
4 5 trwelbmm)
Fig. 5. W-l relationship. Straight line is calculated from W = BP/IYZ Y.
lccl 200 31iding distance s&t-,)
Fig. 6. P-s relationship. Ww,
R. T_4KAGI, Y. TSUYA
that the amount of plastic deformation of the specimen material near the entrance and exit of the rotating disc (which contributes to errors in reading the length I) is negligibly small. The pressure calculated from the end load divided by the product B x I in each of this series of wear tests is plotted in Fig. 6. The value remains nearly constant throughout as expected. Dependence o_fthe wear rate on the sliding speed Curve MI in Fig. 7 shows the dependence of the wear rate on the sliding speed in the case of method MI (SzBCz-SKz). There is maximum wear at 0.4 m/see, decreasing with increasing speed up to 3.6 mjsec. Observation of the mutual sliding surfaces and their circumferences after each wear test shows that the abundant amount of adherent wear particles present at the slower speed decreases more and
Fig. 7. Velocity dependence of the wear rate of SzBCz-SKz for F-s: C3 (3.3 kg and 50 m at the end). Temperature rise (“C) of the disc after a test measured by a thermistor probe is given in parentheses.
more with higher speed. At the highest speed tested, scarcely any wear particles are visible. In the case of the wear test by method Ma (curve Mz), where wear particles are almost completely removed by a wiper, 2 is somewhat larger at the slower speeds tested, but varies very little over the speed range 0.4 mjsec - 3,6 m/set, with only a slight decrease at higher speeds. Above I m/set, the two methods of wear tests, MI and M2, give practically the same results. Figure 8 shows the dependence of the wear rate on velocity when the applied load P at the end of test is changed from 1.1 kg to 9.9 kg, while the product of P and s remains constant. There is a general tendency for the amount of wear particles at the end of a wear test to decrease either toward higher speeds for the same loading, or toward higher loading for the same sliding speed. In consequence, the speed at which maximum wear occurs shifts to a lower value when the applied loading is increased, but the maximum wear rate seems to be but little affected by the change in loading. In the case of method M3, on the other hand, the wear rate depends on velocity as
shown in Fig. 9, which cannot be explained by a mere absence of the wear particles. Not only is the wear rate value itself larger than before, but also the general directions of the veIocity dependence curves vary remarkably for different ioadings.
Fig. 8. Velocity dependence of the wear rate of SzBCn - SK2 (MI) for P-S: Cr, Cz,. . . , C5, keeping the product P x s, at the end, constant. A Cr, x Cz, 0 C4, o Cs. EZoints for the curve C3 are given in Fig. 7.
Fig. g. Velocity
skiing speed ~rnked
of the wear rate of SzBQ-SKz given in Fig. 8.
4 (MJ). P x s is the same as that
ROLE OF WEAR PARTICLES
During a WXE test as represented by a point on the curve ailx in Fig, 71 a considerable amount of the wear particles fall off and many of them remain attached near the sliding surface after a test at the slower sliding speeds. At the higher speeds, the wear
TAKAGl I Y ’ TSUYA
particles produced durin,g a test are far less in number and only a small amount of such material is attached around the sliding traces on the tested sliding surface, In the case of curve MZ there are, naturally, scarcely any visible wear particles attached near the sliding surfaces. Though the quantity of wear particles produced is only indicative of the magnitude of the wear rate, the difference that exists between the curves MI and 141~can only be due to the abrasive action of the wear particles present between the sliding surfaces. When method MI is used, increasing speeds cause the wear particles attached to the disc periphery, especially the larger particles, to be scattered about by the increasing centrifugal forces; increasing pressures make it more difficult for the wear particles attached to the sliding surface of the disc, but out of contact with the wear trace on the specimen, to get into the contact zone again beca.use the particles are stemmed at the entrance to the wear trace of the disc rotating about a vertical axis. The consequent decrease in number of abrasive particles hetween two surfaces in contact causes the lower wear rate curve in Fig. 8. The sliding surface of the specimen observed immediately after a wear test displays a variation in roughness and brightness. Surfaces produced by method MI in Fig. 7 exhibit thick striations parallel to the sliding direction, but are full of metallic lustre. Those of curve Mz in the same figure are smooth with thin striations, but not so bright as those of method MI. The thickness of the MI striations decreases slightly as the sliding speed increases, corresponding to the decrease in number of the larger masses of the wear particles functioning as abrasives. In the case of method Mz, all the larger wear particles are removed by the wiper and only very fine particles arc attached to the surface, so that the wear surface is smooth. In the case of method Mg, on the other hand, since there are no wear particles to prevent welding between steel and bronze and to produce a rough surface, the surface is very smooth but the small scale welding, shearing and transfer that occur continually at the real contact points on the sliding surface give rise to a lustreless wear trace along the path of sliding. However, specimens tested at speeds less than 0.4 mjsec are externally quite different from those tested at larger speeds, corresponding to a depression in the wear rate curve in Fig. 9. The former are rougher and brighter than the latter. The velocity dependences of the wear rate are also very different for different loadings. These characteristic features in the case of method M3 need further experiment in order to clarify the results. To obtain estimates of the temperature rise due to sliding, the sliding surfaces of the disc were touched inlmediatel~7 after a wear test by a thernlistor probe. The temperature at the points of real contact in sliding is naturally higher than these temperatures, but some measure of the temperature rise may be obtained by this means. An example of the temperatures thus measured at the room temperature of ~5 N 28°C is given in Fig. 7. The insignificant temperature rise recorded, which increases little on increasing the sliding speed or loading, is much lower than that given by [email protected]
, who used the same machine to test the wear of steels under severe conditions. In the C:~W of method M3, values only a little higher than those given in Fig. 7 were obtained in the lower speed range, but they rise to some extent at high speed under heavy loading. .4&KNVWLEDGEMENTS The authors thank the ~orishima Foundry Works and Sato Metal Engineering Company for furnishing some of the specimen materials.
WEAR RATE OF UNLUBRICATED
IiEFERENCES 1 R. HOLM, Electric Confacts, H. Gebers F&lag, Stockholm, 1946, p. 214. 2 J. T.BURWELL,JR., Wear, 1(rg57/58) IIQ. G. W. ROWE, Appl. Mech. Rev., 13 (rg6o) 787. 3 K. DIES, [email protected]
, 83 (1939) 307. K. MAILANDER AND K. DIES, Arch. Eiser&hiittenw., 16 (1942/43) 385. 4 N. SAWIN, Werkstattstechnik, 33 (1939) 165. 5 M. O~osar, Japan Pat. No. 208,782 and No. zrg,ogg. M. OKOSHI, T. SATA AND M. MIZUNO,?'~~~~.J.S. M.E., 21(1955) 555. 6 5. ITO, Rep. Govt. Mech. Lab., No. 41 (1961). Wear, 5 (1962) 435-445