1、07FTM14Roughness and Lubricant ChemistryEffects in Micropittingby: A.V. Olver, D. Dini, and E. Lain, Imperial College,and T.A. Beveridge and D.Y. Hua, Caterpillar, Inc.TECHNICAL PAPERAmerican Gear Manufacturers AssociationRoughness and Lubricant Chemistry Effects inMicropittingA.V. Olver, D. Dini an
2、d E. Lain, Imperial Collegeand T.A. Beveridge and D.Y. Hua, Caterpillar, Inc.The statements and opinions contained herein are those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.AbstractMicropitting has been studied using a d
3、isc machine in which a central carburised steel test roller contactsthree, harder, counter-rollers (“rings”) with closely controlled roughness. We varied the roughness usingdifferent finishing techniques and investigated the effects of different oil base-stocks and additives, whilstkeeping the visco
4、sity approximately constant. We also developed a predictive model for the approximateanalysis of rough-surface elastohydrodynamic lubrication based on the FFT approach of Hooke.Damageonthetestrollersincludeddensemicropitting and“micropittingerosion”inwhichtensofmicronsofthe test surface were complet
5、ely removed. This phenomenon is particularly damaging in gear teeth where ithas the potential to destroy profile accuracy. It was found that anti-wear additives led to a high rate ofmicropitting erosion and that the effect correlated more or less inversely with simple sliding wear results.There were
6、 also appreciable effects from base-stock chemistry.The key parameter affecting the severity of damage seemed to be the near-surface shear stress amplitudearising from the evolved roughness; different chemistries led to the evolution of different roughness duringinitial running and thence to differe
7、nt contact stresses and different levels of damage.Copyright 2007American Gear Manufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314October, 2007ISBN: 978-1-55589-918-91Roughness and Lubricant Chemistry Effects in MicropittingA.V. Olver, D. Dini, and E. Lain, Imperial
8、College,and T.A. Beveridge and D.Y. Hua, Caterpillar Inc., Peoria, ILINTRODUCTIONIn gear teeth, the phenomenon of micropitting isnow widely recognized,either asa damagemecha-nism with a particular visual appearance (“greystaining”), or alternatively by its characteristic fea-turesoffinesurfacecracks
9、withaparticulargeome-try and orientation. 1-3. However, micropitting isa phenomenon which extends beyond gearing af-fecting a wide range of mechanical components inwhichrolling-slidingcontacttakesplace,notablyin-cluding rolling bearings and cams. The key diag-nostic feature appears to be surface cra
10、cking on ascale appreciably smaller than the dimensions ofthe nominal area of contact (usually Hertzian). Inturn this is associated with the stresses caused bysurface roughness. It is rolling contact fatiguecaused by roughness 3.Roughness of surfaces which are brought into con-tact creates local hig
11、h pressures and stresses. Inrolling contacts these high stresses will always becyclic in nature since the rolling motion causes thecontact pressures at a particular position to pulsateasthecontactsweepsacrossthesurface. Iftheop-positesurface,the“counterface”,isroughandthereisslidingpresent,eachpassa
12、geofthecontactgivesrise to many cycles of stress as the crests of theroughness slide past each point on the originalsurface, causing a rapidly pulsating pressure.In turn, it is hardly surprising that the high cyclicstresses cause metal fatigue and thereby lead tocracking inthe near-surface. However,
13、this formoffatigue is evidently rather peculiar in nature andshows some features not shared, for example, bytooth root fatigue in gears. Among these are thefollowing:Thestressfieldismostlycompressivebutwithahigh shear stress range, so crack opening can-noteasilybecausedbytheexternalloads;thereisfric
14、tionbetweenthecrackfacesand,nearlyal-ways,thereiscrackfacedamagemakingfrac-tography particularly difficult.The stresses are also non-proportional (stresscomponents do not rise and fall in proportion toone another) and randomly distributed in time,so stress cycles are difficult to count even if thein
15、stantaneous pressure field can be analyzed.The resultant cracking is often very dense andaccompanied by extensive plasticity. Thismakes formal fracture mechanics treatmentcomplex anddifficult. It alsopredisposes theaf-fectedsurfacetoseverewear(“micropittingero-sion”) a feature particularly damaging
16、to gearteethwhereseriousprofilelossmayresult,lead-ingtonoise,vibration,dynamicloadsandsome-times, consequential damage to the transmis-sion system.As a result for these difficulties, the development ofcalculation methods to ensure protection from mi-cropitting has often resulted in procedures whicha
17、re complex, conjectural in nature and difficult tovalidate. The subject has occupied a large amountofeffortinthegearingindustryworld-wide,notleastfrom AGMA 4.However, recent years have seen substantial ad-vancesinthetheoryoflubrication. Untilrecently,al-though designers could calculate the “EHL film
18、thickness” (using smooth body assumptions) andcompare it to the measured roughness of gearteeth, it was not possible to describe how thestresses in the affected surface would be affected.For many practical gear engineers or would-be fa-tigue analysts, this rather made the whole exercisepointless! St
19、okes 5 is particularly scathing aboutthe contribution of elastohydrodynamic lubrication(EHL) to gear design.In the present paper we examine how a simplifiedmicro-EHL analysis can be used to make predic-tions of near surface stresses and we comparethese to some experiments to show how thestresses are
20、 related to the development ofmicropitting erosion.In a previous FTM paper 6 it was shown that cer-tain oil additives promote micropitting by maintain-ingtheroughness(and theconsequent stresses)at2a relatively high level. Here we made use of thesame additive (a zinc diethyl-dithio-phosphate,ZDDP) so
21、 that we could keep the surface rough-ness approximately constant during the experi-ment, facilitating the analysis. An alternative ap-proach in which the developing topography isexplicitly analyzed has also been attemptedrecently 7.Theory of micro-elastohydrodynamiclubricationThetheoryofelastohydro
22、dynamiclubrication(EHL)ofsmoothsurfaceswasdevelopedinthemid-twen-tiethcenturybyErtel,Petrusevich,Dowsonandoth-ers,largelyinresponsetothequest forunderstand-ing of gear-tooth related problems. (See 8.) Itsapplication to (smooth) gear teeth is fairly straight-forward although it is necessary to addres
23、s the fric-tionaltemperaturerise inthe teeth. However,whenroughness is present, although a continuous fluidfilm persists, there are variations of pressure andfilm thickness associated with the roughness. Ingeneral, the roughness sweeps through the con-tact, so the problem is very much complicated by
24、the time-dependent (as well as spatially depen-dent) nature of the pressure and separation. It wasfound that numerical solution of the governing elas-tic and Reynolds equations was very difficult be-cause the elastic deformations are very largecomparedtothesurfaceroughnessandtothethick-ness of the f
25、ilm, leadingto computationalinstability.Nevertheless, these problems were successfullysolved from the late 1980s onwards, by Snidle,Evans, Venner, Chang, Hooke, Lubrecht, andothers. Experimentalvalidationofthesepredictionslaggedbehindforsomeyears,duetopracticalprob-lems in measuring such thin films
26、but have nowbeen carried out, notably by Choo, Spikes and co-workers. A review of the application of these theo-ries to gear lubrication was given by one of thepresent authors 9.Recent years have seen two further highly signifi-cant developments. Firstly, a simplified version ofmicro-EHLtheoryhasbee
27、n developedin whichthemeasured roughness is decomposed into constitu-ent harmonic components using fast Fourier trans-forms (FFT) 10, 11. Approximate predictions ofpressure are then made by separate considerationof each Fourier component and subsequently su-perposed. This means that relatively rapid
28、 calcula-tions of pressure, film thickness and, moreimportantly, subsurface stress can be carried out.In addition, attempts are now being made to tacklethe corresponding subsurface stress and fatigueanalysis12. Thisismadeeasierbytheadoptionofthe FFT approach since the resultant pressurewaveshaveafor
29、m inwhichcyclecountingispartic-ularly straightforward; they are sinusoids, witheither a constant, or steadily decaying amplitude.Strictly the FFT approach is only valid when theroughness is small compared with the smooth filmthickness, a condition which might appear to pre-vent its use in micropitti
30、ng. In contrast, experienceactually shows that features which are largecompared to the film thickness can indeed be ana-lyzedsatisfactorily11. Nevertheless,oneproblemwith the method is that it presumes that all rough-ness is entirely flattened when the lubricating film isabsent (e.g. at very low spe
31、ed or viscosity). Thisisnot satisfactory for the analysis of steel gear teethwhere some roughness invariably persists understatic conditions.For this reason, we have here used a modified ver-sion of the standard FFT method in which we firstfound the deformed surface shape and pressurescorresponding
32、to elastostatic (dry) conditions. Wethen carried out the analysis following the methodsdescribed in 10, 11 but we used the dry contactcondition as the low-film-thickness-limit, using theFourier transform of the elastostatic profile. Themethod is very straightforward; we did our calcula-tions using t
33、he data analysis function in MicrosoftExcel which also supports the necessary complexnumber manipulation. A full description of this isgiven in recent papers 13, 14.AnalysisIn order to show how, in principle, the contactstresses are affected by the roughness and by thelubrication condition, a plot o
34、f the distribution ofshear stress range with depth below the surface isshowninfigure 1for threeexample conditionsiden-tified in table 1. Details of the calculation methodare given in another recent paper 13.3Table 1. Contact conditions modelled.All cases were p0= 1.6 GPa, = 5.2%; e=4MPa;u = 3.05 m./
35、s.Roughness LubricationconditionViscosity, cP (Smooth) filmthickness, nmSmooth Dry - -Rough Ra=0.5mm Lubricated 33.0 250Rough Ra=0.5mm Lubricated 5.47 73Rough Ra=0.5mm Lubricated 9.8410- 31.0Rough Ra=0.5mm Dry - 0Figure 1. Variation of shear stress range with depth for three different film thickness
36、es (obtainedby changing the viscosity of the oil).Resultsofthecalculation areshown inthe Figure1.As a guide, the upper bound estimate of the shake-down limit stress for this material is expected to beabout 2.1 GPa. The 1nm film is effectively unlubri-cated and shows a very high shear stress, in ther
37、ange where plastic collapse would be expected.Thelubricatedcasesgive muchlower stresseswiththe thickest film showing stresses little greater thanfor smooth (Hertzian) contact.The two lubricated, rough cases with appreciablefilm thickness (73 and 250 nm) show how the effectof the lubrication is to re
38、duce the stresses in thenear surface region to a depth which, although itincreases with the film thickness, also dependsuponcontactconditionsandslidingspeed. Thisoc-cursbecausetheshorterwavelengthsofroughnessare accommodated within the film whereas the lon-ger wavelengths are compressed. Since,in th
39、eab-senceoflubrication,theshorterwavelengthsarere-sponsible for higher and shallower stresses, thishas the effect of mitigating the highest stresses.Figure 2 shows the effect of varying the roughnessandrunningconditions. Theroughnessdataforthisstudy were derived from measurements of thecounterface d
40、iscs from the experiment describedinthe next section. Again itcan beseen thatalthoughthe rougher surfaces create higher stresses, theseare limited in magnitude near the surface by the ef-fect of the lubricating film.4Figure 2. Effect of changing roughness speed and temperature on the range of sub-su
41、rfaceshear stress. (See Table 2 for the relevant conditions.) Also included are the stresses derived forthe same profiles but under elastostatic (dry) conditionsInthenextsection,somediscmachineexperimentsare described and the results compared to the pre-dictions of shear stress.Experiments on microp
42、ittingRoughness StudyTheexperimentswereconductedonatriplecontactdisc machine, described elsewhere and illustratedin figure 3. The test rollers were smoother andslightly softer than the counterface “rings”. Asummary of the experimental conditions is given intable 2.Theoutcomeofthetestsisshowninthelas
43、tcolumnoftable2whichshowsthediametralwearandinfig-ure 4 which illustrates the surface of the test rollersafter 5 minutes and 5 hours running (7.3104and4.5106contact cycles respectively).Table 2. Test temperature, film thickness and wear. All cases were p0= 1.7 GPa, =5.2%.Lubricant was poly-alpha-ole
44、fin synthetic base stock (3.19 100 C) + 1.3% ZDDP additive(0.1%P). Rings were transverse ground as figure 3c.TestNo.RingroughnessRa,nmEntrainmentspeed, m/sInlet Temp.,CViscosity,cPhc,nm Wear, mm1 185 3.05 76.0 5.47 73.5 22 195 2.00 99.7 3.39 44.3 23 265 2.00 99.7 3.39 44.3 34 400-610 3.05 76.0 5.47
45、73.5 875Figure 3. Disc machine tests, showing triple contact disc machine (a) with central testroller and three counterface “rings”; ring geometry (b) showing detail of transverselyground (c); and longitudinally polished (d) topography. The arrows show the directionof the finishing marks. (e) shows
46、the rollers before and after the test and thewear depth measurement with a stylus profile instrument.Figure 4. Optical micrographs showing appearance of the test rollers during andafter the test. Only in test 4 did micropitting extend to the whole roller surfaceand develop into severe micropitting e
47、rosion.6Base stock and additive studyFigure 5 shows earlier results from tests with longi-tudinal (hand polished) topography compared withthe transverse topography considered here. Testconditions were as given in table 2 for test 4. Thereis a little more micropitting wear with the transversegrinding
48、 of approximately the same roughness.However, the behavior is essentially the same withsevere micropitting erosion for the cases with theZDDP present throughout the test.Figure 6 shows corresponding results (same testconditions, additive, longitudinal roughness) for anAPIgroupImineraloil(2.7cPat100C
49、). Again,forGroup 1 mineral oil and 1.3% ZDDP the wear isslightly worse than the corresponding results withthe PAO but the pattern is still the same: the ZDDPcausesseveremicropittingwear. Forthemineraloilhowever, the specimen tested without the additiveshowed more damage with larger pits on the testsurface than was the case for the PAO.Figure 5. Micropitting wear with PAO base stock, showing the effect of the direction theroughness direction and finishing method. The roughness of all the rings was in the ran
