Marklund_Wet-clutch-friction-characteristics-obtained-from-simplified-pin-on-disc-test_2008.pdf

Tribology International 41 (2008), Roland1431. Introductionsurface pressures. This is the case in limited slip differ-entials, which normally have a rather low surface slidingspeed, and seldom reach the state of lock up. For thedrivetrain to work smoothly without any unnecessary noiseautomatic transmissions that start the engagement at ahigh difference in rotational speed and then reach a state oflock up. The high speed in these cases implies that theARTICLE IN PRESSclutch will work in full film lubrication, the largest part ofthe engagement. Such an engagement process is possibleto simulate with good results. There will be a torque0301-679X/$-see front matter r 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.triboint.2007.11.014C3Corresponding author.E-mail address: par.marklundltu.se (P. Marklund).Wet clutches are often used in vehicle drivetrains. Theworking conditions of different clutches in the transmissiongreatly vary depending on the application. Wet clutches inautomatic transmissions are often used as lock-up clutchesbetween different rotating parts in the gearbox, where theinitial sliding velocity of the clutch interface can be quitehigh. Other parts of the drivetrain can have wet clutchesthat work with much lower sliding speeds and higherand vibrations, the friction characteristics of the wetclutches have to be thoroughly investigated. Dependingon the working conditions some clutches will work in fullfilm, mixed and boundary lubrication, whereas others willwork mainly in boundary lubrication regime. To get abetter understanding of the frictional behavior of wetclutches, several simulation models have been developed asa complement to traditional measurement methods 16.Most investigations include simulations of clutches inThe frictional behavior of wet clutches in vehicle drivetrains is critical for their overall behavior. During the development of new wetclutch systems there is a need to know this friction behavior. The transferred torque is normally investigated in test rigs where the frictionin a sliding interface between a friction disc and separator disc is investigated. These test rigs can be designed differently, depending onthe working conditions of the investigated clutch. However, it is possible today to simulate the clutch behavior and not limit ourselves toonly using measurements from test rigs for the design of the wet clutch. The torque transferred by the clutch during engagement can beroughly divided into full film torque and boundary lubrication torque. The full film regime is possible to simulate quite well, whereas thefriction in the boundary regime is much more difficult to simulate due to its strong additive dependency. To obtain a good prediction ofthe total engagement, friction measurements in the boundary lubrication regime are still needed. These measurements should be easy toperform and fast tests are preferable. Friction coefficients for the whole range of sliding speed, interface temperature and nominal surfacepressure should be measured. To use these measurements in simulations and get a better understanding of the friction behavior, it is alsopreferable to conduct these measurements on a small test sample, for which the temperature and sliding speed can be regarded asconstant.Here, the friction of a small sample of a wet clutch friction disc is investigated in a pin on disc test and the temperature is measured inthe sample during the tests. Measurements are compared with measurements from a test rig for whole friction discs. A goodcorrespondence between the frictional behaviors of the different measurement methods is achieved.r 2007 Elsevier Ltd. All rights reserved.Keywords: Wet clutch; Friction measurement; Pin on discAbstractWet clutch friction characteristicsimplified pinPa r MarklundC3Division of Machine Elements, Department of Applied Physics and MechanicalReceived 2 January 2007; received in revised formAvailable online824830s obtained fromon disc testLarssonEngineering, Lulea University of Technology, Lulea, SE-97187, SwedenNovember 2007; accepted 16 November 2007January 2008/locate/tribointdesign, but if only the output torque is measured, thefriction coefficient that can be computed for the clutch issimply the mean friction coefficient. Temperature andvelocity are not constant in the interface, meaning that thefriction is also not constant, see 2,6.To get a better understanding of how the friction can bedescribed in terms of temperature, velocity and surfacepressure, a testing method that measures more local effectshas been developed.2.1. Pin on discARTICLE IN PRESSInternational 41 (2008) 824830 825contribution from the boundary friction at the end of theengagement. This friction is much more difficult tosimulate, since it is very additive dependent. For this partof the engagement there is still a need to do frictionalmeasurements that can be used in simulations. For clutchesworking mainly in boundary lubrication, during longerperiods, the simulations will be very dependent onmeasurements. Examples of simulation models for thiskind of application are temperature simulations used topredict changes in torque transfer during a long engage-ment with high surface pressure and a small limited sliptorque as in Marklund et al. 6. For these kinds of semi-empirical simulation models, we encounter a need for amore local friction measurement than what is possible tomeasure in a test rig that measures torque transfer fromone whole friction disc 3,7, or larger parts of a frictiondisc, including grooves 8. A measurement method to dothese local measurements should measure on a quite smallsample of the friction disc, to get the local effects. It shouldalso be possible to measure the temperature inside thissmall sample close to the sliding interface, since it is shownin 6,7, that the temperature will affect the friction. If thetest sample has no grooves, the measured friction will notbe geometry dependent. This is not the case whenmeasuring torque transfer from a whole friction disc,where grooves and other surface patterns also can affectthe torque. A measurement method based on a pin on disctest can fulfill all these demands. A special pin is designedwith a holder for a small sample of the friction disc where athermocouple is mounted to monitor the temperatureduring friction measurements. This method can also give abetter understanding about the friction phenomena thanwhat is possible in whole friction disc test rigs. The frictioncoefficient and its variation with temperature, slidingvelocity and surface pressure is measured in this paper ina pin on disc test. The test is relatively fast and the normalrange of operational parameters are covered within 2h.Another advantage of the proposed test is the possibility tomeasure the local friction effect, which is of great interestwhen using measurements in simulations.2. MethodA measurement of boundary lubrication is needed to beable to simulate torque transfer in wet clutches working inboundary lubrication regime. This friction is very additivedependent, and is therefore a function of the additivesadsorption and reaction on the surfaces. Adsorption,desorption and reactions depend critically on the operatingconditions temperature, sliding velocity and contactpressure. To obtain a friction coefficient for the wholeworking range of temperature, velocity and pressure, manymeasurements are required to describe the friction. Thefriction in the sliding interface of a wet clutch is oftenmeasured as output torque for one whole friction disc inP. Marklund, R. Larsson / Tribologycontact with one steel separator disc. This is a good methodto measure the final output from an existing wet clutchA special holder is developed for a pin on disc test toenable these local friction measurements for the materialcombinations used in wet clutch systems. In a pin on disctest, a stationary pin is loaded axially in contact with arotating disc, as in the schematic sketch shown in Fig. 1.The friction force on the pin can be measured, thus makingit easy to compute the friction coefficient. The pin on discmachine used in these tests is a Phoenix Tribology TE67.In these measurements, a special pin, Fig. 2, is madewhich has a holder for a small specimen made of a frictiondisc, Fig. 3.A thermocouple that measures the temperature at about0.3mm from the contact surface is inserted in the specimen.Since the specimen only has a diameter of 3.0mm, aconstant velocity and temperature can be assumed over thewhole test specimen contact area. This makes the measuredfriction suitable to use in wet clutch simulations and gives abetter understanding of the boundary friction. The frictionmaterial on the friction discs used in this investigation ismade of sintered bronze.The disc is designed as a holder for a piece of the steelseparator disc used in the real wet clutch system. Thismeans that the test specimen will have the same propertiesas the separator disc used in the clutch. The lubricant usedin these experiments is a semi-synthetic oil tailor-made forthe Haldex Limited Slip Coupling, which is a limited slipdifferential manufactured by Haldex Traction AB. Thisapplication is further described in 9.The friction measurements are in this case made tocorrespond with the working conditions of a wet clutch in alimited slip differential, meaning that the sliding velocitiesFaxFig. 1. Schematic sketch of a pin on disc apparatus.ARTICLE IN PRESSFig. 2. Pin with holderP. Marklund, R. Larsson / Tribology826will be fairly low while there will be quite high tempera-tures and surface pressures. The ranges for temperature,velocity and surface pressure and the resolution of themeasurements are shown in Table 1.The sampling rateduring the measurements is 10Hz.2.2. Test procedureThe tests start at an ambient temperature, 221C, and theequipment is gradually heated during the measurements.Before the test starts, the surfaces are run in with the testlubricant. The disc rotates at a speed of 100 revolutions perminute for 10min with an applied load, corresponding to asliding speed of 0.15m/s.During the test, the velocity is increased from 0 to 0.5m/s,followed by a decrease in speed to a standstill. The wholemeasurement takes about 30s. When the test is finished, aheater is engaged to warm up the test equipment to the nextFig. 3. Test specimens from bronze friction disc. Specimen to the left withdrilled hole for thermocouple.Table 1Working range and resolutionWorking range ResolutionNominal surface pressure, p (MPa) 4.08.0Temperature, T C14C 22100 0.2Rotational speed (rpm) 0318 1.0) Sliding speed, v (m/s) 00.5 0.0016Friction force, Ffric(N) 049 0.015) Friction coefficient, m ()o5:3C210C04temperature level and a new measurement is conducted.There is a temperature difference of 51C between thetemperature levels. The total temperature range for whichthe friction is measured is 221001C. The whole test seriesis therefore performed for 16 different measurements withdifferent start temperatures.3. Results and discussionDuring each test the velocity is increased from standstillto 0.5m/s and then decreased back to standstill. Thisvariation in speed is not linear, and a typical velocity plotfor the tests is shown in Fig. 4(a).3.1. Temperature variation during testThe temperature in the test specimen increases due tofrictional heating during the velocity increase. When thevelocity is decreased, the temperature will decrease. Thistemperature behavior during the test is visualized in Fig. 4(b)for a start temperature of 251C. It is obvious from this figurethat there is no significant delay in the temperaturemeasurement, indicating that the measured temperature isa good measure of the mean temperature in the slidinginterface. The sliding interface is located about 0.3mm fromfor test specimen.International 41 (2008) 824830the thermocouple. The temperature measurement is also agood indicator that the temperature is very dependent on thesurface heat flux, since the temperature will immediately startto decrease when the velocity is decreased, see Fig. 4.Themeasured temperature in this point, 0.3mm from the surface,will in this paper be referred to as interface temperature.3.2. Friction measurementsTo use the value of the friction coefficient in wet clutchsimulations, or for a wet clutch control software, the mostimportant is to describe the friction coefficient as afunction of sliding speed, v, interface temperature, T, andnominal surface pressure, p. One test cycle in this pin ondisc test will give this frictional behavior for one combina-tion of friction material, lubricant and load. Results fromthe measurements can be visualized in differently. One wayARTICLE IN PRESS0v m/s0 2040600.40.5t sFig. 4. Temperature variation and sliding speed duringP. Marklund, R. Larsson / Tribologyis to plot the friction coefficient as a function of slidingspeed and interface temperature, as in Fig. 5. Themeasurements are statistically very good with little spreadbetween the measurements.There are basically two ways to describe the relationshipbetween friction coefficient, sliding speed and interfacetemperature. A mathematical expression can be fitted tothe measured friction data, or the data could be stored in alarge matrix from which friction coefficients could beinterpolated from nearby cases. A mix between these twomethods can also be used 6. The advantage with anapproximated function is that it will not need a largestorage space, and that is vital for control softwares withsmall memory capacities. Another advantage with thismethod is that the friction coefficient will be easy tocompute. The disadvantage is the limited flexibility of thechosen expression that can only be applied for a specificfrictional behavior. For this case, the expressionm a1a2C1tanhvC1a3a4C1T a5C1T2a6C1T3a7C1va8C1vT a9C1vT0.450751000.040.060.080.10.12afii9839-T Cv m/sFig. 5. Friction coefficient as function of sliding speed and interfacetemperature. Nominal pressure 8:0MPa.0 20406025262728T Ct sone test cycle: (a) sliding speed; (b) temperature.0.12International 41 (2008) 824830 827gives a good approximation to the measured frictionaldata, see Fig. 6.In this equation m is the friction coefficient, v is thesliding velocity and T is the interface temperature.Another way to visualize frictional behavior is to use thisexpression and plot the friction coefficient as functionof sliding speed for different interface temperatures, seeFig. 7.3.3. Comparison with other test rigTraditional friction measurements of wet clutches areperformed in test rigs with one whole pair of friction discs3,10.In8, larger parts of a friction disc, includinggrooves, have also been tested in a pin on disc test. Withthis test method, where a small sample of the friction disc istested, it is important to investigate the correlation withother performed tests. Fig. 8 shows a comparison betweencurve fits from the measured friction coefficient in the pinon disc and a wet clutch test rig 10 with nominal pressuresof 8.0MPa.0.40.52550751000.020.040.060.080.1afii9839-TCvm/sFig. 6. Friction coefficient versus sliding speed and interface temperature.Measured data and approximative mathematical surface. Nominalpressure 8:0MPa. The mesh is the approximation according to Eq. (1).ARTICLE IN PRESS30C50C70C0.040.060.080.10.12afii9839-P. Marklund, R. Larsson / Tribology828Friction coefficients from the different measurementsshow the same trends in variation of friction coefficientthroughout the whole range of speed and temperature.However, the friction coefficient is consistently slightlysmaller in the wet clutch test rig measurements. Themeasured friction coefficient should not be exactly thesame for the different test rigs, since grooves are notincluded in the pin on disc test.The nominal pressure on the friction discs used in thewet clutch test rig is calculated for the net surfaced area incontact in the interface, i.e. the groove area subtractedfrom the total disc area. A smaller difference in this areafrom the manufacturing process of the discs, could affectthe geometry of the friction material and therefore the net90C0 0.1 0.2 0.3 0.4 0.50.02v m/sFig. 7. Friction coefficient as function of sliding speed at differentinterface temperatures. Based on curve fitted mathematical surface.Nominal pressure 8:0MPa.35C Pin on Disc35C Wet Clutch Test Rig60C Pin on Disc60C Wet Clutch Test Rig00.040.050.060.080.07afii9839-0.20.1v m/sFig. 8. Comparison between friction measurements in pin on disc and wetclutch test rig 10. Nominal pressure 8:0MPa.friction coefficient for 601C is computed. Fig. 10(b) showsthe largest absolute deviation from the mean frictioncoefficient for each sliding velocity. This illustrates that themaximum absolute deviation in friction coefficient forpressure of the three investigated pressures. This makesthe difference in friction coefficient to be dependent onthe load less plausible. It is possible that the differencein friction coefficient is instead dependent of other vari-ables, such as surface structure and friction materialcomposition, indicating that the friction coefficient couldbe described just as a function of sliding velocity andtemperature for the whole temperature range. Fig. 9 alsoshows that for sliding velocities about 0.5m/s, the frictioncoefficient, m, will be about 0.1 for all investigated loadsand temperatures.3.5. Error analysisAs described in Section 3.4 the difference in frictioncoefficient for different surface pressures is not large. Thefriction coefficient could therefore be described as afunction of only interface temperature and sliding velocitywithout loosing much precision. As a measure of thedeviation of the measured data for the maximum andminimum load, the maximum deviation from mean frictioncoefficient computed for m