Gear Surface DurabilityDevelopment to Enhance

T .. PD·ransmlSSl,on --ower - .IensltyJ'oe, Chen. ,John Flynn ,and Ge·off Semr,BUI

New and upgraded transmissions have re-quired a. higher input power capability to meet thecustomers' expectations, Increasing the powerdensity In the transmission gear system wasurgently demanded, To. shorten the developmentlime in thi area, GM Powertrain's AdvancedGear Sy tems 'Group initialed a systematicapproach to evaluate altemarives that had thepotential to meet the needs. In addition to a single-tooth bending project for a gear bending fatiguestudy and joint projects with bearing suppliers toimprove the durability of planetary bearings (Ref.I), a comprehensive gear surface durabilityimprovement project was established.

This study is at preliminary assessment of theeffects of applying different design methodolo-gies, material selections and manufacturingproce ses on gear surface durability. Ba ed on thisanaly is, an in-depth study willfellow to furtherunderstand the benefits of using differem alterna-tives for future applications.

Objective of tbe Stud.y• To evaluate and verify the effects of design,material selection and manufacturing process onpower density of planetary transmis ion gear ets,• To examine and interpret the effects of factors,uch as surface roughness, residual tres .coating,

material properties and geometry variations ongear urface durability.• To understand benefit and risk factors of eachvariant for future application .

.MetbodologyAll re Is were conducted atthe GM Pewenrain

Gear Laboratory using aback-to-back planetarygear component test machine with a speciallydesigned test fixture (Fig, O. An existing planetarygear set was chosen as the baseline because itssurface durability characteristics were wellknown, and its design permitted easy in pectien

I and maintenance,20 JULY/AUGUST 2002 • GEAR TECHNOLOGY" ......w.gnrltlchl1olog)l.com • w'w....po ..... rlrl'l1lmission.com

IntreductkmGear pitting is one of the primary failure

modes of automotive transmission gear sets. Overthe past years, many alternatives have beenintended to improve their gear surface durability.However, due to the nature of new process devel-opment, it takes a length of time and jciot effortsbetween the development team and suppliers toinvestigate and verify each new approach,

FiglJre 1 Back-lo-back planetary gear .component test ItWcMne and flXl!lr~.

Table 1- Test MaIm lor Surface lJurabilily Comparison

1. B asel.ine vs. IRedesign2. Materiall Selectio~Slln l[jear 11011'1'.Use Standard, Planets II Ring Gear (orJestingl

Sun 'Gear MaterialsAISI!SA:E[U.SAI Steels European Steels

4620M !Ba sslin II I 8620 I 9310 I '5120 18COII I 25C04 I 3OC04X X. X. X X I X I X

3. Ml!nufactllring P~Dcess,........lJslllStandard IRing ,Gear for TestingSun GlIBr Process

Planets Baseline Honed Coating n ual Shot Peening Isotropic 1 Isotropic 2Prod. planet X X X XWithout ShotDeburring XHoned X XPlanetsIisotropic 1 Xlsotropic 2 X ,

The test matrix for the surface durability studyis shown in Table 1. II consists of two differentdesign approaches. seven material selections, andnine manufacturing processes.

Design specifications and stress calculationsare shown in Table 2. The stresses are based on atorque of 1,500 N-m at the ring gear. Lead taperis not considered in the stress calculations.

The test samples were thoroughly inspectedbefore the test Relevant features of each test pariwere identified and recorded for future reference.

The gears were originally tested at. a constanttorque of ].500 N-m at the reaction arm and aconstant speed of 1,500 rpm at the input spindle.Under these conditions, however, the redesignedgear set did not pit ill a reasonable period of timeand displayed inconsistent failure modes.Increasing the torque to 2,'000 N-m produced thedesired results.

The test matrix was run in a random pattern tosatisfy the statistical requirement for a small sam-ple size .. The random pattern is also helpful indetecting faults in the machine and fixture set-tings during testing.

Failure was achieved when any tooth flankdeveloped a pit with all area of ].0 mm2 at a depthexceeding L5 mm. Pit size was measured by acomputerized optical image device.and pit depthwas assessed with a gear profile inspectionmachine.

A Weibun analysis (Ref. 2) was performed toestablish life ranks,

A metallurgical failure analysis was performedon selected samples for a better understanding ofthe causes of the failure.

Test ResultsDesign Appmach. The baseline planetary gear

set was designed approximately 20 years ago andhas been used in various transmissions since then.Recently, increased engine power and more strin-gent noise specifications caused this design tobecome obsolete. A new gear set. with enhancedcapabilities but fitting in the same space, was thenproduced, The design approaches of these twogear systems are quite different. The originaldesign used a relatively coarse pitch to enhancethe tooth bending strength and adopted lead taperto reduce the gear force fluctuation.

The use of taper moderates vibration levels bypreventing the sudden release of load at the toothtrailing end (Ref. 3). The redesign uses a finerpitch to increase both involute and helical. contactratios and incorporates profile modifications to'Optimize the load distribution. Based on 3-B FEA

DesignBase Design IRedesign

Normal Module 1.630 1.439Gear Ratio 3.33, 3.29Normal Pressure Angle 20.00 19'.00Helix Angle 18.00 2MOTotal Conta ct Ratio 2.69 3.63Lead Modification 0-36 um O+/-tBj.lm

Stress 'CDmparisonBase Design Redesigll'

Bending (MPa! 512 513Compressive IMPa.! 1,510 1,250

'60 BASELINE FINAL DAIVEOUT.PUT SPEeo • APPROX. 600-700 RPM1ST ORDER. FINAL DRIVECARRIER ROT.

20 ,.;.......J'

,ai'Baseline' Design

------

Table 2-Design Comparison

60 4T40-E FIN>\L ORIVE GEAR GeOM!OrRYOUTPUT SPEED. APPROX. 600-700 RPM1ST ORDER. FINAL OArVECARRIER ROT.

40

20~ ~

b) Redesign

Jee Chenis a staff project engineerwitt: GM Powertrain GearCenter ill Wixom. MI.He holds a master of sci-em:e degree ill mechanicalengineering, as well as amaster's degree in businessadministration. He hasmore than 25 years ofexperience in gear systemdesign and devetopmeni forall· and off-highwaytransmissions.

John Flynnis a senior project engineerwith the GM PowertrainGear Center and the OMPowertrain ApplicariollGroup. HI! holds a bache-lor's degree In memllurgi-cal engineering arId a mas-

ter s degree III businessadministration. He liasmore than 11 years ofexperience 111 metallurgicaland failure analysis.

Figure 2-Cear noise alia vibration levelcomparisOlI.modeling, the new design has better load distri-butionand lower overall. tooth compressive stressthan the baseline gear set. Noise and vibrationtests verified that the new design has a substan-tially lower gear mesh noise level (Fig. 2).,

As was mentioned above, the two designswere tested under different loading conditions.Figure. 3 shows the pitting life comparisonbetween the baseline and redesigned gear sets.Figure 4 shows the corresponding WeibuU analy-

Geoff Semrauis a senior projectengineerwitn the GM PowertminBase Transmission Group.He holds a master oj sci-(mce degree i", mechanicalengineering ..He has beenwitl: OM Powertrain for 18years, and he specializes ingear system developmentand analysis.

WWW.pow8rtransm eston.com » Www.gB8rtBchn0/ogy.com' GEA.R TECHNOLOGY, JULy/AUGUST 2002 21

;HiD

18D

16:Ct

,..,"EE0 .."'if '"";.

~g' 00

jj..••..••

IRedJ!ollgl!(CAiatJoIIlidi

",""""Figure 3-Sun gear pitting life comparisonbetween designs.

99.00

90.00

50.00

if~~'"~t: 10.00:::J

5.00

1.0010.00 100.00 1.000.00

Pitting Failure Hours (> 1.0 mm»

• 3.33 Ratio (wi 1.500 N·m) ... 3.29 Ratio (wl.2,000 N·m) !! 3.29 Ratio (wi 1 ,500 N-m)~1=3.a2. ~1=44.15. p=0.98 ~2=6.72. ~2 =48.28 ~3=6.72, n8=127.48

Figure 4-Weibull allalysis for design comparison.

70

;~ 60iiiif

50~:EC!A

40~:::l 300;;.of:::i 20CI.:'Eii 10

04620M 8620

Baseline9310 51'20 18CD4 25C04 30CD4

Figure S-SZllt gear piuing life camp.arisonamong materials.

Mal.eriall.D.

sis result of these two design approaches, Theanalysis shows that the baseline coarser pitch sungear has a B-50 pitting life of 40 hours. This com-pares to 46 hours at the higher load or 120 hoursat the same load for the finer pitch gear set. The120 hours is determined by extrapolation using anSIN curve with slope of 3.375. The data indicatethat a significant improvement in pitting life hasbeen achieved by the redesign.

Material Selection. Seven common gear steelswere selected for surface durability testing,Figure 5 shows the surface durability test results,and Figure 6 shows the Weibull analysis. Thesample gears were identical in design and manu-facture, Several parameters possibly affectingsurface durability were tabulated, Table 3 showsthe chemical composition. retained austenite,case depth and hardness of each material. Table 4shows a comparison of surface finishes.

Manufacturing Processes. Seven differentmanufacturing processes were evaluated. Bothsun and pinion gears were fabricated using thevarious processes, and a total of nine combina-tions were tested (Table I). Standard ring gearswere used for all tests. To eliminate unwantedvariations, the test gears-except those to behoned-were manufactured and heat treated as asingle lot and later finished by the specificmethod. Figure 7 shows the durability life com-parison of the manufacturing processes, andFigure 8 shows the Weibull analysis that ranks allthe test variants,

The test results reveal that manufacturingprocesses can have a significant effect on the sur-face durability, While further investigation isneeded, some general observations may be made:

The surface finish of the planet pinions has agreat influence on the sun gear durability. This isillustrated as follows:.' Honed sun gears run with standard pinions haveonly one-third the life of standard sun gears runwith honed pinions .. The honed pinions have abetter surface finish than the standard pinions.• Honed sun gears run with standard pinions haveapproximately one-fourth the life of honed sungears run with honed pinions.• The shot deburring process, which is intendedto remove gear edge burrs and heat-treatmentscale, is detrimental to gear life if poorly con-trolled, Standard sun gears run with pinionsthat were not subjected to shot deburring havefour times the life of the standard SUll gears runwith pinions that were shot deburred. The shotdeburring process produces much rougher sur-

22 JULY/AUGUST 200.2 ,. GEAR TECHNOLOGY' www.geartechnology.com' www.powertrenemisstori.com

faces on the gear teeth arid. if not under tightcontrol. can cause severe surface damage.Figure 10 shows a comparison of gear teethbefore and after a poorly controlled shot debur-ring operation.

Both isotropic urface finishing processes(Rd. 4) use a sequence of chemkalandmechanical, operations to smooth the gear sur-face by removing a few micrometers of materi-al. One of these proprietary proces es appearsto have the potential of enhancing surface dura-bility when both mating gears ace treated.

Compressive reoidual stres near the toothfillet has long been recognized as contributingto inc rea ed gear tooth bending life. The sametheory can also be applied to gear surface dura-bility (Ref. 5) with a certain degree of uccess,WI is imperative, however, that the gear surface

not ub tantially damaged by the processused '10 impart that compre sive residual stress.Figure 10 how the comparison of the residualstress produced by the conventional and twosecondary proce se . The dual shot peening andthe honing process. both completed after heattreatment, add compressive re idual stress tothe gear surface. These stresses will reduce ur-face crack progre . ion and thu increase usefulgear life. The duaJhol peening requires twodifferent hot size and intensities, The coarsershot provides deep penetration at the root, andthe finer shot produces high compres ive re id-ual stress close to the surface. Similarly. thehoaing proce s provide high compressiveresidual stre dose to the tooth surface,

All test ariant except the honed paris wereconventionally bobbed and haved and thenbatch heat-treated. However, tooth profile dis-iortion during heal. treatment is not consistentfrom lot to lot or even part 1.0 part. Because thehoning process is completed after heat-treating,better profile consistency can be achieved.

Table 5 shows a comparison of the profilecontrol obtained by the conventional and hon-ing processes. The much improved urfacedurability of the honed sun and planet re 111£from the combination of benefits from con is-tent geometrical control and higher residualcompressive stre s.

Gear surface hardness i' essential. to sustainextended gear life. on-metallic coatings canprovide very high surface hardnes (HRc 90)and lower friction coefficients. The non-coatedsamples in this studyhad conventional surfaceand core hardness. The mechanism by which

l1li00 -QOOO

'legend~ I 4620'M

i so 00 . 8620

" 9310i' S 52.10!~ o 18C04:J 25C041000 x

+ 30CDA

500

100100 1000

1'1'!LngUft (Hour.l ~l,Omrn·2I'JlSQ:.

Figure ,6-Weibull' allalysi for mat'erial comparison.

Ra(~mlflsk!'V1pm)Rms(~ml

The ~mage surl·ece roughnessSkewnm. a measure 01 symmetry of vellsys: e positive skew indice!es e ·peaky' surlaciMWmwn peak'I~VIljey heigh! OYB/the sampl,The root·mean·square average 01 the measured height deviation taken l'Iithin the mlul~o" area endmeasured from the mean surtace

140

120:'10:'!l. 100"eIEc:t 80...."F::1' 60!":s 40'to,e:

=ii: 20

o..... ' --...:,' ... twit. ....... 1 ...... D.~ ..... MIM....,

a... !!III!i!!iII. liIIII!!

www.pow9rtransmlsslon.com • Www.IlBSrr8chnolog'y.com· GEAR TECHNOLOGY' JULV/AUGUST 2002 ,23Figure 7-Momifaclurillg process ,effectOR'slm gear pitthlg life.

coatings provide improved surface durability isnot fully understood at this time and is the sub-ject of continuing research,

figure II shows typical microstructure pho-tographs of tested samples. Failure analysisconducted by the GM Pow-ertrain GearLaboratory revealed that most distress is sur-face-initiated fine grain pitting. Surface cracksinitiating at inclusions were also found. Theinclusions may have been non-metallic impuri-ties in the steel or could have been caused by amanufacturing operation. There were also flnd-ings of intergranular oxidation ([GO), related

I

·• to the high alloy content of the steel. and

I

I improper environment control during heal .1.- ---11 • treatment

Fig.ure8-Weibull analysis for process comparison. ir-------------------~----------------------------_, iii!

I...,

I- I

Before Debut""g

'Ii II~ . -

~

II

Aller DebyrrlngI

-!lUll

!II.

~

I1l11Q

.~

1D.00 10UM! 1.101.l1liPitU1ll1l Flll".1! HOllr.l> 1.D m.. ~

.- _, 1!24.53, ,.1!1.5!1, ralS7

+ II<WDpIo 1 b7.1.B7.h7oQO.01),_~ DSP &11b),5,64, _.so.....:l$!!( _.&11 b&o«7~-'-.opIo2 bll<Q.74,, __

ow.oPlnlon,~ ,1l8o/L73,hio4Ul._,.,II<4C_ .b1=U2,hl0M.99,_• .__ ·l>b2.6:2;l*l!!1!l6,r.zl.QOO __ ,PInIon ,b5011:52,I'6=IlBS:l,_

Figure 9-Plallet surface roughness before alld aJter sllot deburnng process.

-0.50

Cepth I( mml)

-0.30-0.20 0.00200

-0.40 -0.10

10

-200

-400

-600

...... H'oned -800

-1000

-1200

Figure 10-Residual stresses produced by differelll manufacturing processes.

ConclusionsTwo similar planetary gear sets wi.lh different

design approaches were tested to compare surfaceendurance life. The redesigned finer pitch gear sethas about three times the B·50 life of the baselinecoarser pitch gear set.

Stress calculations show that the finer pitchgear set has about 21 % lower compressi ve stress-es and better load sharing than the baseline coars-er pitcll design. The redesigned gear set utilizesfiner pitch to increase the involute and helicalcontact ratios and employs tooth profile modifi-cations to prevent uneven load distribution alongthe line of loath action.

Seven different gear steels were tested. Allsamples were manufactured and. heat-treated byidentical processe . WeibuU B-50 lives for sungears between the baseline and the best.perform-ing steel were spread from 18 hours to 37 hoursrespectively, The life ranks from the lowe t to thehighest are: 4620M (baseline), SAE-93W, SAE-5120, SAE-862!l', 3OCD4, [8CD4 and 25CD4 ..The last three steels are European products.

AJthough the chemical composition, surfacefinish, and microstructure of these materials wereclosely examined, no conclusive explanation oftheir varying urface durability can be made.

Iine different manufacturing processes weretested. Except for the honed gears. all test partswere identically manufactured and heat-treated.Secondary processes were then completed toenhance the power density capability. Weibull B-50 lives between the baseline and Ihe best per-forming samples were spread from 9.5 hours toU5 hours. The gear surface lives among the lestvariants are ranked from lowest to highest as fol-lows: baseline (standard) process 0). isotropic 1(2), double shot-peened sun gear (3), honed sun

2.4 JULY/AUG,US,T 2002 • GEAR TECHNOLOGY' www.gesrtschnology.com ., www.powertran.smlssion.com

gear with standard pinions (4). isotropic 2 (5).standard sun gears mated with pillions that werenot shot debarred (6), non-metalliccoated sungear (7), standard sun gears with honed pinions(8),. and sun gears and pinions both hailed (9).Among tile characteristics producing significant.surface durability improvement are the surfacefinish, the residual stress pattern, the surfacehardness. and superior tooth geometry control.

Planet tooth surface roughne s appears to havea significant effect on the SUIl gear surface dura-bili.ty··O

A,cknowledgmentsThe accomplishment of this project occurred

through tbe joint efforts of all the project teammembers, the manufacturing facility of GMPT·St.Catharine and GM Powertrain gear lest and inspec-tion laboratories. The authors would also like to

express their appreciation to those who providedassi stance and information to this project.

R,efer,ences.1. Chen. J. "Planetary eedle BearingDevelopment to Enhance Transmission PowerDensity," To be published.2. PC· WAS WeibuU AJ1aiysis System User Guide,developed by EDS Group, March 1993.3. Hardy, A. "An Action Analysis of Gear TeethWith Modified Lead and Profile," SAE 670050,4. Hashimoto. F. "Modeling and Optimization ofVibratory Finishing Process," Annals of CIPP,Vol. 45, January 1.996.5. Simpson. R . and R. Garibay .."Quantification ofShot Peening Process Variables AffectingWorkpiece Performance-Process Control. andDouble Shot Peening Medi.a," SAE-8507'14.

This artlcle was fiirst Ipublished intbe'P,roceedingsohhe International Conference 'onMechanicall Transmissions. heidi in Apr:il 20011in Chongqing •.thina ..

Tell Us Mat Younink ...VISit www,,,'lfIChnDlogy.comto• Rate this article• Request more information• CoAIIct the author or companies mentioned• Make a suggestionOr call (847)437-&604 to talk to one of our editors!

lal15x Case microstructure to examinelemperedrna rte nsite and retained a ustenite. 500~.Etchant4% Nital.

(b) 500x Core microstructure to examine temper,edmartensite and some transformation products.500x. Etcha lit 4% Nilai.

Intergrannlar oxidation and inclusion examl'nation nea r ha rdened case area.5OOx.Etchan! 2% Nital.

Ie) 1.0Ci0x

SEM photographs showing the pit examined.

Figure l1-Typical microstructure photograph of tested parts.www.pow8rlran$mission.com' www.geartechnoiDgy.com ,. GEAR TECHNOLOGY' JULY/AUGUST 2002 25