Wear 249 (2001) 361371

The tribological behaviour of paper friction plates for wet clutch application investigated on SAE#II and pin-on-disk test rigs
W. Ost , P. De Baets, J. Degrieck
Department of Mechanical Construction and Production, Ghent University, St. Pietersnieuwstraat 41, B-9000 Ghent, Belgium Received 25 September 1998; received in revised form 12 February 2001; accepted 22 February 2001

Abstract The friction behaviour of wet clutches for automatic transmission applications strongly inuences the dynamic behaviour of the entire machine or vehicle including the transmission. The wear, but also the friction curve, determines the lifetime of the clutch. The role of wear is obvious. The friction coefcient of the material couple friction plate/separator plate decreases with number of engagement cycles. As a result, the possible torque of the transmission decreases with time. Under a certain threshold, the clutch has to be revised. But because manufacturers tend to oversize their clutches the decrease in friction coefcient does not yield a limitation to the lifetime of the clutch. In this paper, both SAE#II and pin-on-disk tests are used to investigate the wear and friction characteristics of the friction material. Although the pin-on-disk tests fail in correctly reproducing the wear rate of the SAE#II tests, they can be used for a qualitative analysis of the inuence of material parameters and operating conditions on both friction coefcients and wear rate. 2001 Elsevier Science B.V. All rights reserved.
Keywords: Wet clutch; Friction; Wear; Simulation test

1. Introduction Wet clutches are often used for transmission application in machine and vehicle construction. A typical application is the automatic powershift transmission used in earth moving off-road vehicles [1]. In this case the clutch has to transfer energy continuously while synchronising. In its slipping state, the clutch absorbs both inertial energy and prime mover energy. The automatic transmission connects the output of the motor and torque converter to the drive train of the vehicle. Several clutches are necessary to make different gear changes and directions. This paper mainly focuses on wear of the friction material, as this will affect the lifetime of the clutch, and the consequent friction changes of the clutches. For this reason the transmission clutch subjected to the severest working conditions is investigated (clutch used for a transition from full forward to reverse motion). A research program is started to determine the clutch life according to Czichos [2] scheme (Fig. 1), and to investigate the inuence of important parameters (such as contact pressure, sliding velocity, cooling, etc.) on wear and friction of a specic friction material.

Earth moving machinery equipped with sensors is tested in the eld (1) or on a test bench (2). As it concerns lifetime tests long test periods are necessary, the more that these machines do not work 24 on 24 h. The behaviour of one single clutch is tested on an SAE#II test rig (3) where a ywheel is driven by an electric motor and the kinetic energy of this wheel is dissipated in the clutch (used as a brake) and a scaled down clutch is tested on a smaller SAE#II test rig (4). Finally, the clutch is scaled down to the tribological system of a separator plate, rotating against a friction plate (5) or part of a friction plate (6). The research program consists of tests of types (1), (3) and (6). Main interest of this research is the investigation of the evolution of the friction characteristics and wear of the friction plates. The results of the eld tests are not available up till now so in the present paper the tests of type (3) and (6) will be discussed.

2. Background Fig. 2 shows a schematic view of the clutch. The clutch essentially consists of nine friction plates (1), radially xed by means of a spline on a central shaft (6), and eight separators (3) radially xed in a drum (9). The plates are hydraulically pressed against the end plate (2) by the axial piston (8), in order to transmit a certain torque. The piston force is

Corresponding author. Tel.: +32-9-264-33-05; fax: +32-9-264-32-95. E-mail address: wouter.ost@rug.ac.be (W. Ost).

0043-1648/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 3 - 1 6 4 8 ( 0 1 ) 0 0 5 4 0 - 3

362

W. Ost et al. / Wear 249 (2001) 361371

Fig. 1. Tribotesting according to Czichos [2].

determined by the oil control pressure (7). For smooth coupling characteristic this control pressure is not increased suddenly while closing the clutch, but is applied linearly with time (electronically controlled). When the modulation is softer (smaller inclination of the pressure versus time curve), the slipping time of the clutch and the heat dissipated in the clutch becomes larger, but as an advantage the jerk (the derivative of the acceleration) and torque peak remains acceptable. The friction surfaces are cooled by means of oil (4) brought into the clutch through the hollow, central shaft (6) and ejected between the friction surfaces. The surface of the friction plate is grooved in order to increase the oil ow (and cooling capacity) between the plates, even when the clutch is closed. Generally, grey cast iron or carbon steel is used for the separators (counter surfaces) to provide mechanical strength and to dissipate frictional heat to the cooled shaft. The separators normally are non-wearing surfaces, although slight changes in surface topography can be noted. An important

problem of the separator is the occurrence of so-called hot spots, locations where the temperature is appreciably higher than for the surrounding surface [3]. Hot spots occur at a location of greater contact pressure (greater thermal input), often caused by original thickness variations of the separator plates and thermal expansion [4]. Prior to the test program of interest, preliminary tests were performed [5] on the clutches in order to determine the limits of the operating parameters for which hot spots do not occur. In the subsequent research, test conditions were always chosen in a proper way to exclude hot spots. The friction material can be of several types: paper, cork, sintered metal, resincarbon and resinasbestos, depending on required mechanical strength and thermal resistance. Because of the low strength (and toughness) of the friction material the friction plates frequently are fabricated as a steel substrate of a stamped metal covered with a relative thin layer of the friction material. The spline teeth for both friction and separator plates are hardened. Wet friction clutches operate in the boundary lubrication regime, and consequently, the observed wear rates are low. It has been found that wear life can be improved by increased cooling obtained by grooving of the friction plate and increasing the oil ow [6]. The grooves also play an important role in wiping the uid from the friction interface thereby increasing the dynamic coefcient of friction. Paper is a modern friction material, sometimes called organic material, and consists of blends of bres and llers, which are formulated using papermaking processes. The compounds are mixed with water, then rolled and dried to form a continuous sheet of material. Paper material is used because of its high dynamic coefcient of friction and its low ratio static to dynamic friction coefcient [7]. Consequently, the tendency for stickslip in a paper-based clutch is low, resulting in smooth and quiet operation. The tested paper material has a high power absorption, an ability to withstand high compressive loads and a good thermal stability. The porous structure of paper friction material permits a kind of internal cooling oil ow resulting in a good durability of the material. For good operation of paper clutches one has to pay attention to the surface prole of the separators, their production method and hardness. Surface roughness, particularly scratches and burrs presenting jagged edges (even microscopically) can rapidly skive the friction material resulting in high apparent wear [7].

3. Experimental 3.1. Clutch test rig 3.1.1. Characteristics of the test rig The test rig schematically shown in Fig. 3 (SAE#II machine) is used to investigate the friction characteristics of the clutch.

Fig. 2. Schematic view on wet clutch.

W. Ost et al. / Wear 249 (2001) 361371

363

Fig. 3. SAE#II machine.

A rotating inertia (1) is driven by an electric ac motor (2). At a certain moment (pre-set rotational speed) the motor is powered off and the clutch (3) is closed, connecting the inertia to the xed frame (4). The torsion torque between the frame and the inertia is measured by means of a torque transducer (5). The clutch is lubricated and cooled by means of oil pumped with a hydraulic pump (6) into the clutch via a lter (7) and a cooling heat exchanger (8). The rotational speed of the inertia is measured with a magneto-resistive speed sensor on a 60 teeth gear (9), the control pressure of the clutch is measured (10) as well as the oil temperature and the temperature of the several separators in the clutch. All measuring signals are amplied, digitised and stored on PC. The same PC controls the motor speed and the opening and closing of the clutch. Table 1 mentions the main characteristics of the test rig. 3.1.2. Clutch characteristics 3.1.2.1. Friction plate characteristics. A friction plate (Fig. 4) is made of a substrate steel plate (Euronorm 1C60)
Table 1 Main characteristics of SAE#II machinea Numberb 1 2 3 6, 7, 8 Device Inertia Electric ac motor Clutch Cooling circuit

Fig. 4. Friction plate.

with hardened external spline teeth, coated with a friction material. Organic bre matrix friction material (paper, Raybestos S-7901-2) with groove pattern (wafe) is used. Before and after the life tests, the thickness of the friction plate is measured. 3.1.2.2. Separator plates. The separator plates are made of stamped steel sheet (Euronorm 1C60, out = 133.25 mm, in = 92.66 mm). The surface is ground and tumbled to a mean surface roughness of Ra 0.32 m. The hardness of the plates equals 215 HB. Before the tests the thickness of the plates is veried and the surface roughness is measured (apparatus Perthen S5P), in accordance with DIN 4768 procedure. 3.1.2.3. Lubricant. For lubrication and cooling of the clutch a commercially available ATF-oil is used (Texamatic

Characteristics Moment of inertia (kg m2 ): 5.16 + i0.78 (i = 0, . . . , 6) Power (kW): 30 Nominal speed (rpm): 1500 Friction material: Raybestos paper Number of friction plates: 9 Type of oil: Texamatic 7045 Flow (l/min): 015 Mean oil temperature ( C): 80 Pressure (bar): 18

10
a b

Hydraulic control circuit Shown in Fig. 3. As in Fig. 3.

364

W. Ost et al. / Wear 249 (2001) 361371

7045, viscosity 81.1 cSt at 20 C). This oil has been developed as a compromise for good gear, bearing and friction plate lubrication in the transmission. 3.1.3. Test procedure Before the test the friction plates and separators are numbered and their thickness is measured. By means of surface prolometry the surface roughness of the separator plates is measured. The plates are correctly mounted in the clutch and in test 3, thermocouples are glued into the eight separators. Preliminary tests [5] have shown that the very thin (insulating) glue lm around the thermocouple affects the temperature measurement. To determine its inuence, a thermocouple with its glue lm is submersed in boiling water. Using a graphical stepresponse method described in [8], a model of form, K esd /(1 + s) (Kpfmller), is t and its accuracy is checked. Time delay is ignored, as it will not affect the maximum temperatures. K is found to be 1.0 and the time constant = 1.39 s. The clutch is placed in the test rig and is cooled with an oil ow of 10 l/min at 80 C. The test is started. The clutch is in open condition and the electric motor accelerates the inertia to a rotational speed of 1300 rpm in a period of time of 5 s. Then the clutch is closed; this means that the axial force in the clutch is gradually increased from 0 to 18,921 N (apparent contact pressure 2.86 MPa) in a period of time of 4 s (values derived from a real application). The clutch absorbs the kinetic energy of the rotating mass, which decelerates to standstill in a period of time determined by the friction in the clutch. The axial force is held for a further period of 4 s. Then the clutch is opened for 2 s, the motor accelerates and a new cycle is started. After a pre-set number of cycles the test is stopped and the friction and separator plates again are inspected (thickness, surface roughness). 3.2. Pin-on-disk test rig Because the full clutch tests are very time consuming (lot of preparation work, long testing time) and expensive it was decided to scale down the clutch to the contact of a coupon from a friction plate on a separator plate. The surface area of the friction coupon is chosen in such a way that although the normal force is limited the apparent contact pressure on the coupon lies in the same range as in the clutch test (03 MPa). The contact pressure and the rotational speed can be varied separately. 3.2.1. Characteristics of test rig Fig. 5 shows a schematic view on the laboratory tribotester. A separator plate (1) is placed on a turning table (2) driven by a frequency controlled ac motor (3). A coupon

Fig. 5. Pin-on-disk test rig.

from a friction plate (4) is pressed against the separator plate by means of a pneumatic piston (5). A ball-and-socket joint is used to avoid misalignment. The friction surface is lubricated with abundant oil ow (6) at 80 C. The normal force and the friction force on the coupon are measured using resistive loadcells (11 resp. 7). Rotational speed of the separator plate is measured with the tachometer (8), the temperature just above the contact surface with a thermocouple (9). The combined wear of the friction plate and the mating surface is measured with the contactless distance probe (10). All signals are amplied and stored on a PC hard disk. The PC also controls the rotational speed of the separator plate. 3.2.2. Specimens characteristics A coupon with a surface of 310 mm2 of the same friction plates (cfr. supra) as for the clutch tests is used. The external spline teeth are removed, and part of the friction material is removed by milling (Fig. 6). The separator plates and the lubricant are the same as used in the clutch tests.

Fig. 6. Friction plate coupon.

W. Ost et al. / Wear 249 (2001) 361371

365

3.2.3. Test procedure The thickness of the friction material and the surface roughness of the separator plate are measured, each in four different points, and both are mounted in the test rig. The oil is preheated to a temperature of 80 C (kinematic viscosity = 1.715 105 m2 /s) and is applied to the contact zone. Normal force is applied and the test, consisting of several cycles in which the rotational speed of the separator plate is varied, is started. After a pre-set time, the test is stopped and the specimens are inspected. Both the thickness of the friction plate coupon and the surface roughness of the separator plate are measured in four different points.

trol pressure reaches its maximum. As could be expected [9], prelock-up (rotation speed = 0) and postlock-up torsional vibrations occur. The prelock-up frequency (55 Hz) is much higher than the postlock-up frequency (8 Hz), due to the difference between prelock-up and postlock-up inertia (ywheel). The pressure ramp (pressure modulation) is derived from a real earth moving vehicle where this slope had been determined in order to obtain a smooth operation and absence of a heavy torque spike at each engagement. During closing of the clutch the rotational speed of the inertia wheel decreases and the torque between the clutch and the xed frame increases. The torque is the result of the friction between the plates, generated by the axial force on the clutch plates. On each time the instantaneous friction coefcient can be calculated [10] as =
2 r 2) 3M(ro i 3 r 3) 2zFa (ro i

4. Results and discussion 4.1. SAE#II test rig Three different tests were conducted with 18,590, 11,140 and 30,455 number of cycles, respectively. Test 1 was stopped at the moment that a change of the friction coefcient (occurrence of vibrations) was detected. Test 2 stopped accidentally and test 3 was stopped because of heavy vibrations. Fig. 7 shows a typical example of the measurements during one closing cycle of the clutch (control pressure, rotational speed, torque). It can be seen that there is a slight time delay before the oil is admitted behind the axial piston (time td ). The clutch plates are then pressed against each other and the volume behind the axial piston is lled. Once this volume is lled (time tf ), the pressure is linearly increased until a maximum control pressure of 18 bars is reached. Due to the small inclination of pressure ramp the clutch is already closed (lock-up) before the con-

(1)

where is the friction coefcient, M the measured torque, ro the half outer diameter of separator, ri the half inner diameter of friction plate, z the number of friction faces, and Fa the axial force on clutch plates. In deriving formula (1), the friction between the friction plates and the outer drum and between the separator and central shaft have been neglected. A uniformly distributed contact pressure and friction shear force have been assumed. In Fig. 8, the friction coefcient during one SAE#II cycle, calculated with (1) is plotted, together with the rotational speed. The friction coefcient at beginning of the cycle (high rotational speed) is lower than the friction coefcient just before lock-up (low rotational speed), but as the contact pressure also varies (0.35 MPa at the beginning and 1.69 MPa at lockup), no conclusions can be drawn regarding the inuence of rotational speed on friction.

Fig. 7. Clutch characteristics during slipping cycle.

Fig. 8. Friction coefcient during one SAE#II cycle.

366

W. Ost et al. / Wear 249 (2001) 361371

Fig. 9. Friction coefcient during test 1.

Fig. 11. Friction coefcient during test 3.

During one test cycle, the mean friction coefcient is dened as m = 1 (tl tc )
tl tc

( ) d

(2)

where m is the mean friction coefcient, tl the lock-up time (rotation speed = 0), tc the time at which the clutch plates are brought into contact, and (t) the instantaneous friction coefcient. Figs. 911 show the friction curve during tests 13, respectively. In Fig. 9, a sudden rise of the coefcient of friction at 5700 cycles can be seen. At that moment, the test rig stopped due to a failure in the hydraulic circuit. When the test was re-started (after 4 days of stand still), a 15% higher coefcient of friction was measured. The friction values measured before the failure were only reached again after another 10,000 cycles. After approximately 17,000 cycles, a somewhat larger scatter in the mean coefcient of friction is observed (vibrations). This effect also occurred in test 3, where it is more explicit. The friction coefcient is slightly decreasing with time. Fig. 10 shows the friction curve during test 2. Compared to test 1, a clear running in stage with rising friction is observed. The reason is that the measurement started from the rst cycle, while in test 1 the rst cycles were not measured

(failure in data-acquisition system). One can see that the running stage is restricted to about 1000 slipping cycles. Just as in test 1, the friction is decreasing with number of cycles in regime conditions. Test 3 (Fig. 11) holds for 30,000 cycles, which is comparable to one fth of the real machine life. The rise in the friction coefcient after 12,000 cycles again is due to an accidental stop (some days of stand still) of the test rig. The friction coefcient is decreasing with number of cycles and above 12,000 cycles the torsional vibrations are increasing. The vibrations are analysed more in detail in Fig. 12. In this gure, the amplitude of the torsional vibrations (expressed in N m as measured by the torque transducer) and the fraction of the cycle during which vibrations are observed, are drawn. The amplitude and the duration of the vibrations increase steadily with the number of cycles up to 15,000 cycles, after which the vibrations were too heavy to measure. The vibration data measured during tests 2 and 3 are analogue (increase in amplitude and time duration). Torsional vibrations are induced either by dynamic friction instability (clutch shudder) caused by a negatively sloped friction versus slip curve [11] or by stickslip, which originates from the difference between static and dynamic coefcient of friction and can also occur when the friction

Fig. 10. Friction coefcient during test 2.

Fig. 12. Vibrations during test 1.

W. Ost et al. / Wear 249 (2001) 361371

367

Fig. 13. Changes in surface roughness of separators.

versus slip curve is positively sloped [12]. As demonstrated above, the slope of the friction versus slip curve cannot be determined from Fig. 8, but pin-on-disk tests will show a negative slope of this curve. Considering the negative slope of the friction versus slip curve, and the fact that the speed measurements are not detailed enough, so that the existence of stick periods cannot be detected, the origin of the vibrations remains unknown. Figs. 1315 give an indication of the wear occurring during the clutch tests. Fig. 13 shows the change of the surface roughness of the separators due to wear, for all friction surfaces. For each friction surface, four measuring data are shown: the Ra surface roughness in two different points before and after the test. It can be seen that the surface roughness decreases during the test (mean surface roughness Ra before test = 0.32 m, after the test = 0.05 m). It is also important to note that the nal roughness after the test merely is inuenced by the original roughness before the test.

Fig. 15. Thickness change rates of different friction plates during SAE#II tests (plate 2 near the end plate, plate 8 near to axial piston).

Fig. 14 shows the thickness changes of the friction plates after the clutch tests. Fig. 15 shows the wear rate. Fig. 16 shows the maximal temperatures of the intermediate separator plates, reconstructed on base of the measured temperatures, using the rst-order model described above. From Fig. 16 it is clear that the maximum temperatures vary along the clutch pack. The temperatures near the end plate are larger than those near the axial piston. This is a result of a non-uniform cooling because the oil, which serves as a coolant is admitted near the axial piston (see Fig. 2). Because of the friction in the splines of the clutch, however,

Fig. 14. Thickness changes of different friction plates during SAE#II tests (plate 2 near the end plate, plate 8 near to axial piston).

Fig. 16. Maximum separator plate temperatures during wear test (separator plate 1 is situated at the end plate, plate 8 is situated near the axial piston).

368

W. Ost et al. / Wear 249 (2001) 361371

the normal pressure is higher near the axial piston than near the end plate. Nevertheless, the wear rates of the different friction plates in the clutch indicates that this effect of decreasing contact pressure along the clutch-pack is negligible compared to the effect of the variation of temperatures along the clutch-pack. The mean thickness changes are 230 m for test 1, 150 m for test 2 and 170 m for test 3. From the wear rates in Fig. 15, it can be concluded that rst a running-in wear with high wear rate or setting of material occurs, followed by a steady-state wear with lower wear rate. 4.2. Pin-on-disk test rig Using the pin-on-disk test rig the inuence of the contact pressure and the hardness of the separator plate on the friction coefcient and the wear rate of the friction material was investigated. Each test consisted of several cycles in which the rotational speed was varied between 0 and 1300 rpm (See Fig. 17). The clutch tests on the SAE#II machine have shown that the running in period (with rising friction) lasted for about 1000 cycles or, with each cycle equalling a sliding distance of 7 m, a sliding distance of 7 km. Each pin-on-disk test lasted around 100 h, yielding a total sliding distance of 1600 km, long enough to determine the steady state wear rate. To investigate the inuence of the contact pressure on the tribological behaviour of the friction material, tests were run with contact pressure 1.0, 2.0, 2.3, 2.6 and 2.9 MPa. Before and after each test the thickness of the friction material is measured in four different points. Averaging the thickness changes of those four points yields the values presented in Fig. 18. Fig. 19 contains the thickness change rates, i.e. the thickness changes of Fig. 18 divided by the total sliding distance. To check the reproducibility of the wear rates of the previous tests, another batch of tests was performed with contact
Fig. 18. Thickness changes of friction plate coupons for pin-on-disk tests with different contact pressures.

Fig. 19. Thickness change rates of friction plate coupons for pin-on-disk tests with different contact pressures.

pressures 2.0 and 2.3 MPa. The thickness change rates obtained in those tests are presented in Table 2, accompanied by the values obtained in the previous tests. All the values of Table 2 lie within close range of each other. Fig. 19 and Table 2 show that the wear rate is the lowest for 1.0 MPa, and highest for 2.9 MPa. The wear rates for 2.02.6 MPa lie between those two values, but wear rates for 2.3 and 2.6 are lower than could be expected. Nevertheless, the wear rates for the clutch tests (SAE#II), in which the maximum contact pressure is 2.86 MPa, are an order of magnitude larger than those for the pin-on-disk tests. If the thickness change rates found with the SAE#II tests are divided by a factor 2, because for the clutch tests
Table 2 Reproducibility of measured thickness change rates in pin-on-disk tests Thickness change rate of friction plates coupons (105 m/m) Contact pressure (MPa) Test 1 Reproducibility test Reproducibility test 2.0 6.50 7.52 6.43 2.3 3.54 2.38

Fig. 17. Friction force and rotational speed during one cycle of a pin on disk test.

W. Ost et al. / Wear 249 (2001) 361371

369

Fig. 20. Friction coefcient during pin-on-disk wear tests with different contact pressures.

the friction plates wear on both sides, than the wear rates for the SAE#II tests are situated in the interval 4.9 106 and 17.4 106 mm per cycle or 35 105 and 124 105 m/m, while those of the pin-on-disk tests are situated between 2.83 105 and 10.7 105 m/m. Fig. 20 shows the dynamic friction coefcient as a function of time for the different contact pressures, wherefore the dynamic friction coefcient is dened as the mean of the values at the maximum rotational speed. Friction coefcients are highest at the beginning of each test, and decreasing with time. The friction coefcient is highest for a contact pressure of 1.0 MPa and lowest for 2.9 MPa. Friction coefcients for contact pressures 2.02.6 MPa lie within close range from each other.

The friction coefcient during one test cycle, after 10 h of testing, is plotted versus the rotational speed of the separator plate in Fig. 21. The values recorded after 100 h of testing are shown in Fig. 22. From these gures it can be noted that the dynamic friction coefcient drops with increasing sliding speed, coinciding with previous reports by Berger et al. [9] and Ito et al. [13]. The latter reported a dynamic friction coefcient of 0.15, slightly decreasing with sliding speed, for a contact pressure of 1.04 MPa and an oil viscosity of 1.1 105 m2 /s. As Figs. 21 and 22 show, the slope of the friction versus velocity curve is independent of contact pressure and worn volume. Figs. 21 and 22 also show that the dynamic friction coefcient decreases with increasing contact pressure,

Fig. 21. Friction coefcients at the start of the pin-on-disk wear tests.

Fig. 22. Friction coefcients at the end of the pin-on-disk wear tests.

370

W. Ost et al. / Wear 249 (2001) 361371

Fig. 23. Changes in surface roughness of separator plates in pin-on-disk tests.

although this is less pronounced for the higher contact pressures. Fig. 23 shows the average of the surface roughness, measured in four points, before and after the pin-on-disk test. As can be seen in Fig. 23, the changes in surface roughness during the pin-on-disk are smaller than for the SAE#II test, especially for the test conducted at 2.3 and 2.6 MPa, with even an increase in surface roughness for a contact pressure of 2.3 MPa. Important to note is that although the separator plates used in the pin-on-disk test were manufactured according to the same specications as those used in the SAE#II tests, they were not taken from the same production batch and their initial roughness is lower than the initial roughness of the separator plates used in the SAE#II tests. The surface roughnesses after the pin-on-disk tests are similar to those after the SAE#II tests. An overview of the results concerning dynamic friction coefcient, wear rates and changes in surface roughness is presented in Fig. 24. Wear rates and changes in surface roughness are taken from Figs. 19 and 23. The friction coefcient is dened as the average of the values from Fig. 20. In this overview (Fig. 24), a clear correlation between wear of the friction material and roughness decrease of the steel separator plate can be noted. The lowest wear rate (at 2.3 MPa) corresponds with a small increase of surface roughness. A possible explanation could be that much of the wear of the friction material is associated with surface modication at the beginning of the tests (running in). For cases with small roughness decrease this could indicate less running in of the materials, leading to a higher friction coefcient. Fig. 24 bears this out, as the curve of friction coefcient seems to be the opposite of the trend of wear rate and surface roughness decrease. So it seems that at different contact pressures the surfaces run in at a different rate, so that the differences in observed wear rates is not a function of the different loads directly, but more of the running-in rate of the surfaces.

Fig. 24. An overview of the pin-on-disk test results: thickness change rate, roughness decrease and dynamic friction coefcient as a function of contact pressure.

The inuence of the hardness of the separator plate on the wear rate and friction was checked during a test in which hardened separator plates were used with hardness 370 HB but the same initial surface roughness, i.e. R a = 0.19 m. The test was conducted with an apparent contact pressure of 2 MPa, while the rest of the conditions (oil temperature, rotational speed, etc.) remained the same as in the previous tests. The surface roughness after the test equalled R a = 0.23 m and the wear rate was found to be 9.65105 m/m. The dynamic friction coefcient is plotted in Fig. 25, together with the friction coefcient of a previous experiment conducted with the same apparent contact pressure. From Fig. 25, it

Fig. 25. Friction coefcient during pin-on-disk wear tests with different hardness.

W. Ost et al. / Wear 249 (2001) 361371

371

is clear that the friction coefcient is not inuenced by the hardness of the separator plate. Nonetheless, the observed wear rate is higher. This puts previous experiments performed by Fisch and Lloyd [7] into question. In their experiments Fish and Lloyd used two different hardnesses for the separator plates, i.e. HRC 13, which is more or less equal to the hardness of the standard plates used in this paper (215 HB), and HRC 40 (=374 HB) which is equal to the hardness of the hardened plates used here. They found no correlation between the hardness of the separator plates and the wear rate of the friction material.

References
[1] S. Winkelmann, H. Harmuth, Reibkupplungen Grundlagen, Eigen-schaften, Konstruktionen, Springer, Berlin, 1985. [2] H. Czichos, Reibung und Verschleiss von Werkstoffen, Bauteilen und Konstruktionen, Expert verlag 7031 Grafenau 1/Wrtt., 1982. [3] H.A. Rothbart, Mechanical Design and Systems Handbook, McGraw-Hill, New York. [4] P. Zagrodzki, Analysis of thermomechanical phenomena in multidisc clutches and brakes, Wear 140 (1990) 291308. [5] S. Dupont, P. Verhaert, Theoretical and experimental study of a wet clutch, Dissertation University Ghent, Faculty of Applied Sciences, 1997 (in Dutch). [6] P. Payvar, Laminar heat transfer in the oil groove of a wet clutch, J. Heat Transfer 34 (7) (1991) 1791. [7] R.L. Fisch, F.A. Lloyd, Surface nish requirements of spacer plates for paper friction applications, SAE Reprint 730840, August 1981. [8] H. Rake, Step response and frequency response methods, Automatica 16 (1980) 519526. [9] E.J. Berger, F. Sadeghi, C.M. Krousgrill, Torque transmission characteristics of automatic wet clutches: experimental results and numerical comparison, Tribol. Trans. 40 (4) (1997) 539548. [10] B.J. Hamrock, B. Jacobson, S.R. Schmid, Fundamentals of Machine Elements, McGraw-Hill, New York, 1999. [11] C.G. Slough, M.P. Everson, R.C. Jaklevic, D.J. Melotik, W. Shen, Clutch shudder correlated to ATF degradation through local friction vs. velocity measurements by a scanning force microscope, Tribol. Trans. 39 (3) (1996) 609614. [12] C. Gao, D. Kuhlmann-Wilsdorf, D.D. Makel, The dynamic analysis of stickslip motion, Wear 137 (1994) 112. [13] H. Ito, K. Fujimoto, M. Eguchi, T. Yamamoto, Friction characteristics of a paper-based facing for a wet clutch under a variety of sliding conditions, Tribol. Trans. 36 (1) (1993) 134138. Patrick De Baets graduated in Mechanical Engineering in 1989 at the University of Ghent, Belgium. In the same year, he became an Assistant at the Department of Mechanical Construction and Production at the same university, and in 1995, received his PhD degree. As Senior Lecturer, he is involved in lecturing and research on the dimensioning of machine elements and tribology. His research interests are mainly in the area of dry friction and wear, especially fretting wear, large-scale wear testing and industrial research. He is a member of the Dutch materials science group Bond voor Materialenkennis, tribology section and member of the Mechanical Engineering and Transport Mechanics Group of the Flemish Organisation of Engineers Koninklijke Vlaamse Ingenieursvereniging. Wouter Ost graduated in Mechanical Engineering in 1999 at the University of Ghent, Belgium, and is currently working as an Assistant at the Department of Mechanical Construction and Production. His research interests are in the eld of applied tribology in transmission components. He is a member of the Dutch materials science group Bond voor Materialenkennis, tribology section and member of the Mechanical Engineering and Transport Mechanics Group of the Flemish Organisation of Engineers Koninklijke Vlaamse Ingenieursvereniging.

5. Conclusions Pin-on disk tests can lead to a better understanding of the frictional behaviour and the effect thereupon of the operating conditions of the paper friction material used in clutches. In this study, pin-on-disk tests have shown that Dynamic friction coefcient increases as the apparent contact pressure decreases. Independent of the applied contact pressure, the dynamic friction coefcient decreases with increasing sliding speed. Hardness of the separator plates does not inuence the dynamic friction coefcient. However, the pin-on disk tests fail in correctly reproducing the wear of the friction material, mainly because their inability to reproduce the same thermal conditions as used in the SAE#II tests. Though this limits the validity of the data obtained using the pin-on-disk test concerning the inuence of contact pressure on wear rates, it can be concluded that the wear rate of the friction material increases with increasing hardness of the separator plate.

Acknowledgements The authors are grateful to Dana, Spicer Off-Highway Products Division in Brugge, Belgium, for the nancial support of this research and for providing the SAE#II machine, and Raybestos for providing the friction plates and separators. The authors especially thank Dr. Saskia Pattyn and Mr. Filip Demazire at Dana for their valuable suggestions and continued interest.