1、08FTM03AGMA Technical PaperEffects of Gear SurfaceParameters on FlankWearBy J.C. Wang, J. Chakraborty,and H. Xu, Dana HoldingCorporationEffects of Gear Surface Parameters on Flank WearJyh-Chiang Wang, Jay Chakraborty, and Hai Xu, Dana Holding CorporationThe statements and opinions contained herein a
2、re those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.AbstractNon-uniformgear wear changes gear topology andaffects thenoiseperformanceofahypoidgearset. Theaggregate results under certain vehicle driving conditions could pot
3、entially result in unacceptable vehiclenoise performance in a short period of time. This paper presents the effects of gear surface parameters ongear wear and the measurement/testing methods used to quantify the flank wear in laboratory tests. Geartoothprofile,transmissionerror,geartoothsurfacefinis
4、hdeterminedbycutting,andgeartoothsurfacefinishdetermined by other processes are the factors considered in this paper. The measurements includetransmissionerror,coordinatemeasurementmachine(CMM),patternrating,andsurfaceroughnesspreandposttest. ASTML-37baseddynamometertestprocedureisadoptedforthewears
5、tudywithgoodcorrelationtofieldsamples. Thelaboratorytestsamplesareestablishedbasedondesignofexperimentconsideringthecontrolledfactors. Theeffectsandinteractionbetweencontrolledfactorsprovidedtheinformationforproductimprovement. Theactionresultedfromthisstudyisanticipatedtosignificantlyimproveproduct
6、reliabilityandcustomer satisfaction.Copyright 2008American Gear Manufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314October, 2008ISBN: 978-1-55589-933-23Effects of Gear Surface Parameters on Flank WearJyh-Chiang Wang, Jay Chakraborty and Hai Xu, Dana Holding Corporati
7、onIntroductionNon-uniformgearwearchangesgear topologyandaffects the noise performance of a hypoid gear set.The aggregate results under certain vehicle drivingconditions could potentially result in unacceptablevehicle noise performance in a short period of time.This paper presents the effects of gear
8、 surfaceparameters on gear wear and the measurement/testing methods used to quantify the flank wear inlaboratory tests. Gear tooth profile, transmissionerror, gear tooth surface finish determined by cut-tingandlapping,andgeartoothsurfacefinishdeter-mined by post hard test processes are the factorsco
9、nsidered in this paper. The measurements in-clude transmission error, coordinate measurementmachine (CMM), pattern rating, and surface rough-ness pre and post test. ASTM L-37 1 baseddynamometer test procedure is adopted for thewear study with good correlation to field samples.The laboratory test sam
10、ples are established basedon design of experiment (DOE) 2 considering thecontrolled factors. The effects and interaction be-tween controlled factors provided the informationfor product improvement. The action resulted fromthis study is anticipated to significantly improveproduct reliability and cust
11、omer satisfaction.Gear wear can be separated into two differenttypes: one is the uniform wear of tooth surface thatdoesnt affect the noise performance, and the otheristhenon-uniformwearoftoothsurfacethataffectsthe noise performance. The differences are asshown schematically in Figure 1. Curve AA rep
12、re-sents the original tooth profile. A uniform wear willremove material from the tooth and create a newprofileBB. Due tothe non-uniformwear, theactualprofileisasshownbycurveCC. The discussionsinthis paper are focused on the non-uniform gearwear.Non-uniformwearOriginaltooth profileFigure 1. Gear wear
13、Gear surface parametersThe parameters considered for the DOE include:S Gear tooth profileS Gear set TE (transmission error)S Gear tooth surface finish determined by cuttingand lappingS Gear tooth surface finish determined byprocesses post hard testOtherfactorsthathaveeffectsonuniformgearwearare iden
14、tified in the test procedure developmentstage and excluded from the DOE for reasonsexplained above.Measurement methodsBoth objective and subjective ratings are used inpre-test and post-test stages for comparison pur-pose.S TE measurements (single flank tester): Thefirstorder drive side TEs aremeasur
15、ed pre-testandpost-test. The difference in TE measurementspre and post test (Delta TE) is an important indi-catoroftheunevenwear. ThisTEmeasurementprocess is defined as hard test. It is performedafter lapping and prior to final processing.S CMM measurements: Sum of squared errorscompared to a master
16、 is a good way of measur-4ingnon-uniformgearwear. Thesumofsquarederrors is the sum of the square of error at eachgrid point.Sum of squared errors = i, j=1 to N(Deviation atrow i and column j)2The uniform gear wear is covered by the tooththickness change from CMM measurement re-sults. The ratio of th
17、e post-test sum of squarederrors versus pre-test number is an indicator ofthenon-uniformwearbasedonappropriateres-olution of mesh points. Figure 2 shows a typicalpinion measured on CMM prior to gear weartest. Figure 3 shows a pinion with less non-uni-formgearwearafterweartest. Figure4showsapinion to
18、oth surface with more non-uniformwear after wear test.Figure 2. Pre-test pinion CMM resultsFigure 3. Post-test pinion CMM results with less non-uniform wearFigure 4. Post-test pinion CMM results with more non-uniform wear5S Pattern rating: This is a subjective rating helpfulduring the development of
19、 test procedure. Theratings range from 1 to 10 with 10 as the bestpattern and 1 the worst. The patternsof thetestgear set are taken and documented before andafter test. A qualified gear set representing thedevelopment has been used as the semi-masters. Patterns of pinion and gear rolledagainst semi-
20、masters separately after test arealso documented to identify the wear on eachpart.S Surface roughness: The pre-test and post-testsurface finishes are part of the measurements.Atypical hypoidgear toothsurface roughnessisas shown in figure 5. The three-dimensionalsurfacecanbeestablishedbyscanninganare
21、a.The surface roughness numbers used in theDOE is along the tooth profile direction.Test methodThe developed lab test is based on ASTM L-37dynamometer testing. Test load and speed wereselected on the basis of testing several similar gearsets and observing the induced tooth wear. Carewastakentoobtain
22、asetofaccelerated testparam-etersthatinduceonlytoothwear,notany othertypeofsystemfailure. Itisahighloadandlowspeedtestintended to duplicate non-uniform gear wear withgearcontactpatterninFigure8. Atwenty-fourhourtest procedure is effective in differentiating non-uniform gear wear of samples with diff
23、erentparameters. The testequipment isshown inFigure6. A typical pinion after wear test is shown inFigure 7.Figure 5. Typical gear tooth surface roughness6Figure 6. Test equipmentFigure 7. A typical pinion post testAtypicalcontactpatternfromfieldreturnisshowninFigure 8. The worn patterns are typicall
24、y narrow inthe profile direction and toward the root. Figure 9and Figure 10 are the post-test patterns from weartest. The contact pattern in Figure 9 exhibits pat-terns typical of a worn gear set. A less amount ofpatternnarrowingandpositionmovementalongtheprofile direction can be observed in Figure
25、10.7Figure 8. A typical contact pattern from field returnFigure 9. Post-test contact pattern from wear test sample8Figure 10. Post-test contact pattern from wear test sample with less severityDOE samplesThe dimensions of the tested gear set are listed inTable 1.Table 1. Tested gear setItem NumberRat
26、io 4.88Diametral pitch, 1/in. 3.2Offset, in. 1.4Pitch diameter, in. 12.2Afullfactorial ofthree controlledparameters iscon-sidered in the DOE. Each parameter is tested athigh/low levels. A total of sixteen samples weretested with two duplicated samples for each testcondition.1stparameter:Gearsetproce
27、ssposthardtest. Twodifferent processes are considered in the DOE.Level 1 is the current processes post hard test.Level 2 is the proposed new processes.2ndParameter: Gear set assembly backlash athigh/lowlevels. Thisparameteristocontrolthepat-ternpositiontosimulatetoothprofilechangeandTEchange due to
28、cutting and lapping processes.3rdparameter: Gearset processpre hardtest. Twolevels of lapping process are considered. Level 1isthe current production lapping cycles. The lappingcycles in level 2 were modified to reflect differentsurface conditions.Both pre-test andpost-test measurementsincludeTE, su
29、m of squared error, contact pattern, andsurface roughness.Test resultsMinitab was used for the test result analysis 3. APareto Chart is essentially a bar chart in which thebars are ordered from highest to lowest. A factorwith standardized effect more than 2.306 is statisti-cally significant. The mai
30、n effect is the differencebetween the factor level mean and the overallmean. An interaction is presentwhen theresponseat a factor level depends upon the level of otherfactors.Hypoid pinion is the part showing most significantnon-uniform wear. This is shown in the patternsrolled with semi-master. Fig
31、ure 11 shows the pat-tern rolled between a tested gear and semi-masterpinion. It doesnt show obvious pattern narrowingand position change as in Figure 9. Figure 12shows the pattern rolled between a tested pinionand the semi-master gear. It is similar to that inFigure 9 with narrowing pattern and pos
32、ition chan-ge. The patterns rolled with semi-masters suggestthat the test pinions are the primary source of non-uniform wear. CMM measurements also confirmthe trend as shown in the sum of squared errors.9Figure 11. Post-test contact pattern between tested ring gear and semi-master pinionFigure 12. P
33、ost-test contact pattern between tested pinion and semi-master ring gearAll the test data are included in Figure 13. “DeltaTE”isthedifferencebetweenpost-testTEandpre-test TE divided by pre-test TE. “Delta Pin Surf”represents the difference between post-testsurface roughness and pre-test surfacerough
34、nessforthepiniondividedbythepre-testnumber. “DeltaGear Surf” presents the numbers for the gear withthesamedefinitionasthepinion. “PinSumofSqrd”is the ratio of the post-test sum of squared errorsdivided by pre-test sum of squared errors for thepinion. “Gear Sum of Sqrd” is the ratio of the post-test
35、sum of squared errors divided by pre-test sumof squared errors for the gear. The numbers werenormalized to have a common ground forcomparison between different measurements.The most significant factor in this study is the postprocess after hard test. Gear sets from process 210showbetterresultsthange
36、arsetsfromprocess 1inthe pinion sum of squared errors ratio, Delta TE,and subjective pattern ratings (Figure 10 versusFigure 9) in Figure 13. Delta pinion surface finish,Delta gear surface finish and gear sum of squarederrors ratio are not closely related to the post testpattern as shown in Figure 1
37、3.The Pareto charts, Main effects plots, and interac-tion plots for post-test pattern rating, pin sum ofsquared errors, and Delta TE are shown in thefollowing figures. The factors shown in the figuresare abbreviated due to software limitation. “PostProcess” refers to gear set process post hard test.
38、“Assembly” means gear set assembly backlash athigh/low levels. “Gear Process” is the gear setprocess pre hard test.The post process after hard test is clearly the mostsignificantfactorthataffectstheresults. TheParetocharts in Figure 14, Figure 15, and Figure 16 dem-onstrate that only factor A is sta
39、tistically significantin the effects of post-test pattern rating, pinion sumof squared errors, and Delta TE.The main effects plots in Figure 17, Figure 19, andFigure 21 show “Post Process” as the mostsignificant factor for post-test pattern rating, pinionsum of squared errors ratio, and Delta TE. Fi
40、gures18, 20, and 22indicate differentlevel ofinteractionsbetween “Post Process”, “Assembly”, and “GearProcess”. The amount of interaction is not signifi-cant enough to change the importance of thefactors.Figure 13. Scatter plot of all data (delta TE, delta pinion surface roughness, delta gear surfac
41、eroughness, pinion sum of squared errors, and gear sum of squared errors)11Figure 14. Post-test pattern rating Pareto chartFigure 15. Pinion sum of squared errors ratio Pareto chart12Figure 16. Delta TE Pareto chartFigure 17. Post-test pattern rating main effects13Figure 18. Post-test pattern rating
42、 interactionFigure 19. Pinion sum of squared errors ratio main effects14Figure 20. Pinion sum of squared errors ratio interactionFigure 21. Delta TE main effects15Figure 22. Delta TE interactionConclusionThe test procedure and measurement methodsdiscussed in this paper prove to be a feasible ap-proa
43、ch to determine the causeof non-uniformgearwear within acceptable time constraints. Theeffects and interaction between controlled factorsprovided the information for product improvementwithout specifically identifying the root causes ofnon-uniform gear wear. The action resulted fromthis study is ant
44、icipated to significantly improveproduct reliability and customer satisfaction. Fur-ther studies with enhanced control of factors cancontribute to the background of the non-uniformgear wear phenomenon in the future.References1. Standard Test Method for Evaluation of Load-Carrying Capacity of Lubricants Under Condi-tionsofLowSpeedandHighTorqueusedforFi-nal Hypoid Drive Axles, ASTM International,Nov. 1, 2005.2. Keri R. Bhote and Ade K. Bhote, World ClassQualityUsingDesignof ExperimentstoMakeItHappen, American Management Association,2000.3. Minitab, Minitab Inc., 2006.
