FRICTION NOISE CAUSED BY FRETTING UNDER GREASE LUBRICATION

FRICTION NOISE CAUSED BY FRETTING UNDER GREASE LUBRICATION T. JIBIKI, M. SHIMA Department of Mechanical Engineering, Tokyo University of Mercantile Marine, 2-1-6, Etchujima, Koto-ku, Tokyo 135-8533, JAPAN; e-mail: jibiki@ipc.tosho-u.ac.jp H. AKITA, K. HATANO Technical Research Centre, Hitachi Construction Machinery Co. Ltd. 650, Kandatsu-machi, Tsuchiura City, Ibaragi Pref. 300-0013, JAPAN SUMMARY Fretting may be accompanied by friction noise and preventing or reducing it can be important for designers and operators. In this paper, such friction noise caused by a grease lubricated fretting contact is investigated, compared with the result of a non-lubricated fretting contact. A 0.45 % carbon steel (Hv730) and a mild steel (Hv240) are used as specimens with the crossed cylinder configuration. The lubrication grease used is a commercially available lithium soap grease in which MoS 2 powder is mixed. Fretting stroke is varied in the range 45 200 µm, and a normal load and frequency are 19.6 N and 7.3 Hz respectively. In addition direct observation of the fretted area is made using an optical microscope, in order to examine the mechanism of fretting noise caused by the grease lubricated contact. For the experiment, fretting contact of a sapphire glass plate against a bearing steel ball is employed. The main results are as follows: the amount of grease supplied onto the fretting contact greatly affects the generation of friction noise, if enough amount of grease is supplied onto or around the fretted area, friction noise is never generated, however, a small amount of grease cannot prevent the generation of friction noise, and eventually leads to "Squeal noise" where sound level is high enough to bother operators. The fretting cycles at which friction noise starts to occur is affected by fretting stroke, the larger the stroke the lower the cycles. From the direct observation of the fretted area it is found that friction noise starts to occur at the cycles at which lubricating grease is almost expelled from the contacting surfaces and oxidized wear debris accumulates between the contacting surfaces and the behaviour after that is almost the same as the nonlubricated case. Keywords: Friction noise, Fretting wear, Direct observation, Grease lubrication, Wear debris 1 INTRODUCTION The aim of this study is to investigate friction noise caused by fretting under grease lubricated condition. Fretting is often accompanied by friction noise, and preventing or reducing such noise is considered to be important for designers or operators. We have developed a system which is able to measure and analyse friction noise under fretting contact, and investigated the influence of various parameters such as fretting stroke, frequency, relative humidity and wear amount on friction noise under non-lubricated condition [1]. From the results it has been shown that the generation of friction noise is strictly related to wear debris accumulated between the contacting surfaces, and that the generation of friction noise may be due to the change in the types of real contact points bearing a fraction of normal load. In addition, there are common features in relation to the generation of friction noise, that is, drastic reduction in coefficient of friction, selfexcited-vibration state, and negative gradient of coefficient of friction are seen without exception. It seems that no study (except for the paper) has been made on the generation of friction noise caused by fretting although a lot of studies have been reported on friction noise of disk and dram brakes, tire for automobile and wheel for railway [2-14]. In this paper, friction noise caused by fretting under grease lubrication is measured and analysed, which is much more important in industry. In general, lubricating fretted surfaces successfully with grease as well as oil is difficult because those lubricants are easily expelled from the fretted surfaces and re-infiltration of the lubricants is also hard, depending on fretting stroke and amount of lubricant existing around the fretted surfaces. [15]. Therefore it is valuable to examine the relationships between the generation of friction noise and both the fretting stroke and the amount of lubricant. In lubricated fretting, contact surface roughness may play an important role. Grease consists of thickening agent and base oil, sometimes including powder such as MoS 2 or additives, so it is interesting to know which substance plays a role in lubricating fretted surfaces. Taking these points into consideration, friction noise caused by grease lubricated fretting contact is measured and analysed, compared with the result of non-lubricated fretting contact. Based on the results and direct observation of lubricated fretting contact by an optical microscope, possible mechanism of the generation of friction noise is discussed.

2 EXPERIMENTAL 2.1 Apparatus Figure 1 shows a schematic diagram of the fretting apparatus. A driven specimen is attached to a cantilever, which is horizontally oscillated by a motor through a crank mechanism. A fixed specimen is attached to an upper holder. The test apparatus is covered with a chamber in order to control the relative humidity. Friction force is measured by four strain gages attached on two leaf springs of the upper holder. Relative stroke re between both specimens, which is peak to peak amplitude of tangential movement between the upper and lower specimens, is measured by an eddy current pick up. Friction noise is measured by a microphone and a sound level meter, which is able to measure up to a sound frequency of 8 khz. Both AC and DC output signals from the sound level meter are used for the analysis. Friction force, relative stroke between the specimens and friction noise is measured simultaneously during fretting. All the data of these signals are fed into a personal computer through A/D converters that are able to divide 0 to 10 V into 4096 numbers then are analysed. Fig.1: Schematic diagram of the fretting rig 2.2 Specimens Table 1 shows the details of specimens used for the fretting test. Both the fixed and driven specimens are cylindrical, 10 mm in diameter and 30 mm in length. The fixed specimen is a 0.45 % carbon steel quenched by high frequency heating (Hv730), and the driven one is a mild steel (240 Hv). Both specimens were finished within the range of the surface roughness R y 1.0 µm by grinding process. Table 2 shows the details of specimens used for the direct observation of one fretted area. The bearing steel ball is a commercially available one (SUJ2) with mirror finish. Fixed specimen (Upper specimen) Driven specimen (Lower specimen) 0.45% carbon steel quenched Hv730, R y 1.0μm Size : Φ10 30mm Mild steel Hv240, R y 1.0μm Size : Φ10 30mm Table1: Specimens (for crossed cylinders type test) Fixed specimen Sapphire glass plate (Upper specimen) Size : Φ30 5mm Driven specimen Bearing steel ball (Lower specimen) Size : Φ9.525mm Table2: Specimens (for direct observation test) 2.3 Lubricants Table 3 shows the lubricants used for the test. The grease is commercially available lithium soap grease in which MoS 2 powder is mixed. In addition to the grease, the same kind of lithium grease in which MoS 2 powder is not mixed is also used to examine the effect of the MoS 2 powder. To discuss the role of thickening agent in the lubricating fretted area the base oil only, which is the same oil as one contained in the grease, is also used. Type Base oil Viscosity of base oil Lithium soap grease Mineral oil 145.4 mm 2 /s @40 C 14.67 mm 2 /s @100 C Consistency 285(25, 60W) Additive MoS 2 powder Table3: Lubricants Fig. 2: Schematic illustration of direct observation of fretted area Figure 2 shows a schematic illustration of the direct observation of fretted area by an optical microscope. For the observation a sapphire glass plate on a bearing steel ball configuration is employed, and the fretted area is directly observed through the plate by an optical microscope during fretting. 2.4 Fretting test Table 4 shows the experimental conditions. The fretting tests were carried out with normal load 19.6 N, and up to 10 6 cycles in air. Fretting stroke, relative humidity and frequency are varied in the range 45-200 µm, 23-80 % RH and 7.3 Hz respectively. The grease is supplied to the fretted surfaces by the following methods. First, a small amount of grease is put on an optical flat plate, and one of the cylindrical specimens, around whose ends thin films are rolled, is rotated on the plate, then the grease film of about 3 µm in thickness is adhered to the specimen. If the grease

film of 6µm is needed for the experiment, such two specimens are contacted. For the case of the experiment under sufficient amount of grease lying on or around fretted surfaces, grease of the volume of 4 mm 3 is directly put on one of the specimens, and then contacted to the other specimen. The wave signals of the friction noise, together with coefficient of friction and relative stroke, are analysed by FFT analyser. The data sampling frequency of the A/D converter is 5 khz, and 4096 data per second can be stored into a memory. Fretting stroke 45 200 µm Normal load 19.6 N Atmosphere Laboratory air, Grease lubrication Test duration Up to 10 6 cycles Frequency 7.3 Hz Temperature 294±2 K Relative humidity 23 80 %RH Data sampling frequency 5 khz (4096data) Configuration Crossed cylinders, Ball on plate Table4: Experimental conditions 3 RESULTS 3.1 Typical results of measurements Coefficient of friction µ, and sound level of friction noise Lp were simultaneously measured during fretting. The typical examples are shown in Fig. 3(a) for the grease lubricated fretting and (b) for non-lubricated fretting. Under the lubricated condition, µ gradually increases and reaches the same level of non-lubricated case with some fluctuation. Friction noise gradually starts to occur at around 2000 fretting cycles, and finally reaches the same sound level of non-lubricated one. Some fretting cycles are needed to generate friction noise, which behaviour is similar to that of nonlubricated fretting while the fretting cycles to generate friction noise is rather larger than that of non-lubricated case. However, it is apparent that there is no direct relation between the value of µ and the generation of friction noise. Figures 4(a),(b) show µ, Lp and relative stroke re curves at around 5000 fretting cycles at which steadystate friction noise is generated, for lubricated and nonlubricated fretting, respectively. It is found that the friction noise starts to occur at the sliding position where µ drastically drops, and then continues until the end of the half fretting cycle, and that the self-excited vibration is occurring to the fretting device during the period, as shown in the displacement curve. Such behaviour is very similar to non-lubricated fretting. Friction noise taking place under the steady-state condition was examined by a FFT analysis. The results are shown in Figs. 5(a),(b). (a) S = 130 µm, G = 6 µm (b) S = 130 µm, G = 0 µm Fig. 3: µ and Lp plotted against fretting cycles N (a) S = 130 µm, G = 6 µm, N = 5000 cycles (b) S = 130 µm, G = 0 µm, N = 5000 cycles Fig.4: Curves of Lp, µ,and re

(a) S = 130 µm, G = 6 µm, N = 5000 cycles (b) S = 130 µm, G = 0 µm, N = 5000cycles Fig. 5: FFT analysis of Lp (a) S = 130 µm, G = 6 µm, N = 5000 cycles (b) S = 130 µm, G = 0 µm, N = 5000 cycles Fig. 6: FFT analysis of dl/dt (a) S = 65 µm, G = 6 µm (b) S = 200 µm, G = 6 µm Fig. 7: Influence of relative humidity on Ng and Lp The power of Lp is maximum at 1.3 khz, which is the same as that of non-lubricated fretting. This fact suggests that the mechanism of the generation of friction noise during grease lubricated fretting is basically the same as that of the non-lubricated fretting. The same result as the friction noise is also obtained for the power of sliding speed dl/dt, which is the differential calculus of displacement curve, as shown in Figs. 6(a), (b). This means that self-excited vibration is taking place in the fretting device during the generation of friction noise. 3.2 Influence of amount of grease supplied, fretting stroke and relative humidity Influences of amount of grease supplied to the fretted area, fretting stroke and relative humidity on the generation of friction noise were investigated. Figures 7(a), (b) show the influence of relative humidity on the fretting cycles N at which friction noise starts to occur, and sound level of friction noise Lp in a steady-state. From the results, it is found that the influence of the relative humidity on N is more marked for larger fretting stroke than for smaller, and that friction noise tends to be generated earlier at around 40 % RH. However, the sound level of friction noise Lp is hardly affected by relative humidity while Lp values themselves are larger for larger fretting stroke than for smaller. Figure 8 shows the influence of amount of grease supplied to the fretted area on the noise generation cycles Ng. It is apparent that the amount of grease greatly affects Ng, the more the amounts of grease the longer the Ng. The tendency was also seen in Figs. 7(a), (b). Especially friction noise is never generated at least until one million cycles in this experiment in the case of enough grease being present at or around the fretted surfaces. The fretting stroke also affects Ng, that is, friction noise is earlier generated with increasing the fretting stroke. 3.3 Direct observation of fretted area by optical microscope The fretted surfaces lubricated by the grease were observed by an optical microscope to examine the morphology of the surfaces and the behaviour of the grease being present at or around the fretted surfaces,

using a sapphire glass plate contacting with a bearing steel ball. The coefficient of friction µ during fretting is shown in Figs. 9(a), (b). Those values are relatively low during all the fretting cycles when enough grease is supplied at or around the fretted surfaces (symbolised by G = 4 mm 3 ), especially consistently low, about 0.1 in the case of the fretting stroke 200 µm. Those values are also low at the early stage of fretting in the case of the grease amount of 3 µm in thickness, however they eventually go up and reach the values of non-lubricated fretting. (a) S = 65 µm (b) S = 200 µm Fig. 9 Relationship between µ and N The series of photographs of the same fretted surfaces are shown in Figs. 10(a),(b),(c) for the fretting tests made with fretting stroke 200 µm. In the non-lubricated fretting oxide wear debris start to occur at only several hundred fretting cycles and the fretted area gradually increases (Fig.10(a)). In the grease lubricated fretting with 3 µm in thickness about 4000 fretting cycles are necessary for such debris to be generated, and after around 10000 cycles the fretted area starts to grow gradually (Fig.10(b)). The appearance of the fretted surfaces at this stage is very similar to that of non-lubricated fretting. Loose wear debris are also present between the contacting surfaces, which are often agglomerated. On the other hand in the case of enough grease being supplied to the surfaces (Fig.10(c)), the appearance is quite different from the non-lubricated one, wear debris hardly take place and the fretted area never grows during the whole cycles of fretting. Almost the same behaviour was also seen in the case of small fretting stroke (65 µm). 4 DISCUSSION 4.1 Possible mechanism of generation of friction noise Based on the results described above, possible mechanism of friction noise caused by fretting under the grease lubrication is discussed. From Fig. 3(a) friction noise starts to occur at the stage where coefficient of friction becomes comparable with nonlubricated. This fact shows that friction noise is generated after the lubricating grease is almost expelled from the contacting surfaces and dry contact takes place. In such situation the appearance of the fretted surfaces is very similar to that of non-lubricated, as shown in Figs. 11(a), (b). Thus the possible mechanism of friction noise under the grease lubricated condition may be basically the same as that under non-lubricated condition, while noise generation cycles Ng is dependent on amount of grease supplied. The analysis of friction noise curves by FFT (Figs. 5 and 6) also suggests this assumption. The authors have already presented a possible mechanism of generating friction noise in dry condition [1]. The mechanism can be summarized as follows: various types of real contact bearing a load, which possess different coefficient of friction with each other, are generated during fretting, and if a drastic reduction in average coefficient of friction over the contacting surfaces occurs, resulting from the changes in dominant types of real contact points, it can act as a trigger to release an elastic strain energy stored in the fretting system, and to generate friction noise. Such a drastic reduction in coefficient of friction resulting in the generation of friction noise, was also seen in the case of grease lubricated fretting as shown in Fig. 4(a).

(1) 103 cycles (2) 104 cycles (3) 105 cycles (a)g = 0 µm (1) 103 cycles (2) 104 cycles (3) 105 cycles (b) G = 3 µm (1) 103 cycles (2) 104 cycles (3) 106 cycles (c) G = 4 mm3 Fig. 10: Direct observation of fretted surfaces (a) S = 130 µm, G = 6 µm, after 20000 cycles (b) S = 130 µm, G = 0 µm, after 20000 cycles Fig. 11: Appearance of wear scar (0.45 % carbon steel)

(a) Sufficient amount of grease (b) Small amount of grease Fig. 12: Possible mechanism 4.2 Fretting mechanism under grease lubricated condition Discussing the mechanism of friction noise caused by grease lubricated fretting may result in discussing the fretting mechanism under grease lubricated condition, that is, the mechanism of grease penetration into and the expulsion from the fretted area. Amount of grease supplied onto the fretted surfaces and fretting stroke greatly affect fretting behaviour. Sufficient amount of grease well lubricates the fretted surfaces as shown in Fig. 8, and the fretting damage hardly occurs over total cycles of fretting (Fig. 10(c)). On the other hand small amount of grease is effective for early stage of fretting, however it is gradually expelled from the fretted surfaces and eventually dry contact results. Figures 12(a), (b) are the illustration showing a fretting mechanism under grease lubricated condition. If enough amount of grease exists around the fretted surfaces, certain amount of grease can be dragged into the fretted surfaces with fretting motion. As the result some amount of grease can be always present on one of the fretted surfaces. If the fretting stroke is large in comparison with the diameter of the contact region some fraction of the grease can penetrate the contacting surfaces to be in boundary lubrication. However, even if some grease is present on the surface, the contacting surfaces fretted with small fretting stroke may be difficult to be in boundary lubrication because the grease cannot infiltrate into the contacting surfaces. The result of Fig. 9(a) supports the assumption. On the other hand, in the case of only a small amount of grease lying around the fretted surfaces supplying grease to the fretted surfaces is difficult. Thus, once grease lying on the initial fretted surfaces is expelled with wear debris from the surfaces, the contacting surfaces result in being under nonlubricated condition. The grease consists of thickening agent, base oil and MoS 2 powder, so it is important to know which substance plays a role in lubricating the fretted surfaces. In order to examine this point additional experiments were conducted with several lubricants such as the base oil which is the same oil as that contained in the grease, MoS 2 mixed base oil, Lithium soap grease without MoS 2 and dry MoS 2 powder only. Two kinds of grease were supplied to the fretted surfaces with 6μm in thickness, and two kinds of oil were applied to make a small meniscus between the contacting surfaces. The dry MoS 2 powder was applied to form the thin films on the fretted surfaces. The results are shown in Figs. 13(a), (b). It is found that two kinds of oil are very effective to lower the coefficient of friction and to prevent friction noise and that MoS 2 powder mixed in lithium soap grease does not play any role in lubricating the fretted surfaces. These facts mean that the base oil contained in grease mainly functions as a lubricant for the fretting and that MoS 2 powder does not play an important role in fretting if the powder cannot penetrate into the fretted surfaces with oil flow while the powder originally lying on the fretted surfaces plays certain role in fretting. Small "oil pools" formed on the fretted surfaces, for example, those formed by shot peening is very effective to prevent friction noise, as shown in Fig. 14. The coefficient of friction fluctuates, but friction noise never occurs during fretting. This may be due to "oil pools" effect from which small amount of grease is gradually dragged out to the fretted surfaces by fretting.

(a) Noise generation cycles Ng 3) In the case of sufficient amount of grease, the friction noise is never generated during whole fretting, because the 4) grease around the fretted surfaces is gradually supplied into the contacting surfaces with fretting action. The coefficient of friction µ is constantly low and fretting wear hardly occurs. 5) Very small holes lying on fretted surfaces such as those by shot-peening form oil pools, and are effective to prevent the friction noise. They gradually supply the lubricants to the fretted surfaces. 6) Under the grease lubricated fretting condition, base oil separated from the grease may play an important role in lubrication. 7) MoS 2 powder is effective, only when it flows into the contacting surfaces, with base oil. 8) Relative humidity greatly affects the generation of friction noise, not only under non-lubricated fretting but also under grease lubricated fretting. 6 ACKNOWLEDGEMENTS (b) Coefficient of friction µ Fig. 13: Effect of some kinds of lubricants Fig. 14: Effect of shot peening (S = 65 µm, G = 6 µm) 5 CONCLUSIONS Friction noise caused by grease lubricated fretting contact was investigated, and compared with the results of non-lubricated fretting contact, using direct observation of fretted surfaces by an optical microscope. The following conclusions were obtained. 1) The generation of friction noise under grease lubricated condition is greatly affected by the amount of grease supplied and fretting stroke. 2) In the case of a small amount of grease, friction noise similar to that of non-lubricated fretting generates, because once the grease is expelled from thecontacting surfaces, it cannot be supplied from around the fretted contact. The worn surfaces are basically the same as those of non-lubricated fretting. We would like to express our appreciation to Prof. R. B. Waterhouse of Nottingham University for many helpful suggestions and discussions. 7 REFERENCES [1] Jibiki T., Shima M., Akita H., and Tamura M., Wear of Materials 2001 (to be published). [2] Inoue M., Journal of Japan Society of Mechanical Engineering, (C), 51, 466 (1985) 1433-1439. [3] Ohta K., Kagawa K., Eto T., and Nishikawa S., Journal of Japan Society of Mechanical Engineering, (C), 50, 457 (1984) 1585-1593. [4] Okamura H. and Nishiwaki M., Journal of Jap.Soc. of Mech. Engineering, (C), 54, 497 (1988) 166-174. [5] Senda T., Nakai M., Yokoi M. and Chiba Y., Journal of Japan Society of Mechanical Engineering, (C), 50, 449 (1984) 125-133. [6] Nakai M., Chiba Y. and Yokoi M., Journ. of Jap. Soc. of Mech. Eng., (C), 47, 423 (1981) 1466-1475. [7] Fosberry R. A. C. and Holubechi, Z., MIRA Report, 1957/3. [8] Lnag A. M. and Smales, H., I Mech E. C 37/83 (1983) 223. [9] Felske A., et al, SAE paper 780333 (1978). [10] Eales S. W. E., SAE paper 770181 (1977). [11] Millner N., SAE paper 780332 (1978). [12] Miller N., Loughborouugh, Paper C 39/76, London (1977). [13] Lamb H., London Mathematical Society, (1888). [14] Stappenbeck H., VDI-Z, 96-6, (1954), 171. [15] Shima M., Suetake H., McColl I. R., Waterhouse R. B., and Takeuchi M., Wear 210 (1997) 304-310.