Large Scores and Radial Cracks onCase-Hardened Worms

Wolfgang Predki Friedrich Jarchow Ralf iDinter and AIexander Rhode

AbstractIn the last couple of years many

research projects dealt with the determi-nation of load limits of cylindrical wormgears These projects primarily focusedon the load capacity of the worm wheelwhereas the worm was neglected Thiscontribution present investigationsregarding damages such as large scoresand cracks on the flanks of case-hard-ened worms

IntroductlnnThe load-carrying capacity of cylin-

drical worm gears is generally limited by

Fig I-Test stand for scorilgdamagecenter distance a = 1(0 mill

Fig 2-Test stand Jar crackiJlgda11lagecenter dislaflle a 250 111111

Fig 3-Large score marks on a case-hardened Klorm

the worm wheel made of bronze Thismaterial has a lower strength than theworm made of steel Critical loads arereached if pitting sliding wear toothbreakage scuffing or high oil tempera-tures occur However in the last coupleof years defects such as radial cracks orlarge scores have increased on the case-hardened worm surface These detectscan cause an increased wear on the wonnwheel which can lead to a premature lossof drive This report shows a way todetermine the maximum permissible loadthai leads to these defects as well as te t-ing new methods which increase the loadcarrying capacity

Test ConditionsThe test runs to examine score marks

and cracks on hardened worm flankswere executed separately on two electri-cally strutted test stands The matingmaterials are case-hardened worms madeof 16MnCr5 steel and worm wheelsmade of GZ-CuSn12Ni bronze tests forscoring were executed with gear setswith a oenter distance 01 a = 100 rnm agear ratio of i= 41 2 a worm shaftarranged below the worm gear am splashlubrication Figure 1 shows the test standthat was employed for scoring tests

After assembly and adjustment of thecontact pattern the worm gear sets wereloaded immediately with the entiretorque and run for 336 hours without anyrunning-in process The rotational speedsof the tested WOlTIlS were

nl = 400 650 1030 and 1470 lIminThe torque varied from l0 to 175

times the nominal torque that is given bythe gear manufacturer (Ref 7) Cracks 011

the worm flanks were tested on largergear sets with a center distance of a ==250 rnrn a gear ratio of i 392 a WOIDl

shaft arranged above tbe worm gear anda combined splashcirculating lubricationwith an external oil cooling system That

test stand is shown in Figure 2The testing period is limited to a max-

imum of 1650 hours The rotationalspeeds are 1500 and 2200 IlminDuring the tests the applied torqueexceeds the nominal torque 175 and 200times The employed lubricant for eachcase is a poly glycol called FVA-refer-ence-oil PG4 (Ref 4) with additive com-bination LPI655

Test Results ScoringFigure 3 illustrates a worm gear dam-

aged by score marks Compared to newnever-run worms the flanks show arough surface within the area of contactThese score marks are transferred duringoperation to the flanks of the wormwheel and cause excessively increasedwear that often leads to a premature fail-ure of the entire gear set

In order to register the amount ofscoring damage measurements of thesurface are carried out in the completemating area of the wonn shaft The meas-urements are taken in the radial directionperpendicular 10 tile expected scoremarks (Fig 4)

In the radial direction parts of theteeth are separated intQo fOQoI center andtip Tests with a low rotational speed ofnl 4001 lImill and J25 times the nomi-nal torque show the largest scoring dam-age The distribution of the measuredarithmetic mean roughness Ra after test-ing is given with Figure 5 The measuredvalues for each position from the begin-ningto the end of the contact area areplotted The roughness of the flanks illthe initial state is the standard of compar-ison New worms have an average arith-metic mean roughness of R(J = 04 -lm

The distribution shown is typical forworms that are damaged by score marksThe greatest roughnes alway occurs atthe end of the contact areaclose to the tipof the tooth Towards the beginning of

18 MAYJUNE 2003 -GEAR lECIHNI(ILOGY bull wwwgeartechdologycom bull wwwpuwertransmissollcom

tlIe mating area the values of the rough-ne s decrease Similar tendencle can beOD erved in the center of th teeth bulthe maximum values are lower In thearea of the foot the measured valuecatter but they reach comparatively

large value Comparative measurement of the

worm shaft and the worm wheel how agood representation IOf the coring dam-age of he worm flank DYmeasurementat the whee tooth center At this poirn 1the malle 1 lubricant film thickness and 1the smallest relative motions directedtoward the height of the tooth occur dur- iing operation between worm and wheel [

Score marks of the worm are reflected ~011[0 the center of the wheel teeth ~Therefore score marksat the wheel are ~uitable for the qualitative de cription of ~

the coring damage at the worm flanks [Figure 6 shows the measured arithmetic 1mean roughne s at the wheel after sepa- ~role test NOS with different load All ~tests were executed with case-hardened 1worms made with 16Mn r5 steel and ~wheels made of GZmiddotCuSnl2Ni bronze l

The te t result show Eli ignificaJltdependence on rotational speed and ltorque The amount of scoring damage jincreases with an increase in thetorque iapplied and decreases with the mtatiomlli 1peed of the worm hafl (Fig 6) On the j

one hand increased load lead to an lincreased coring damage On the other ~hand an increased rotational peed lead 10 Ie coring Tile tronge t coringdamage can be observed at rotationalpeed of III = 400 ]lmin At rotational

-peeds of more than n = ]030 lmin nocore marks result even in the case of

high overloads The roughness of tlleflanks after testing is almost the same ain the beginning The arithmetic mean ~roughness represents the amount of scor- 1ing damage l

A regre ion analysi quantifie the 1dependence of the expected arithmetic ~mean roughne on the worm wheel on ~differenl load a rotational peed n I

torque T2 and oil Sump lelIlpera~ure ts by lfour coefficients Q14 jR al + (112 - II)middotmiddot 11 + a4middotmiddot ) bull Tz ~ 0 IlJTI (W)

Figure 7 shows lite expected arim- --------------metic fOughllessRoR after application ofEquation I A data map show the ~expeeted coring damage for the exam-ined material-lebricam combinationdepending Oil torque and rotationalpeed

A observed during IhetesLs theroughness increases with increasing loadand decreasing rotational speed Mea- 1 I

urement during the te ting reveal an increased wear for RaR = 1-2 um andexcessively increased wear if greaterroughne s applies In some cases themea ured wear exceed the value pre-calculated by Dm 3996 (Ret 3) IIpto 20 time For this reason an arithmeticmean roughness above R = 2 um has tobe particularly regarded as critical Foroperation with constant torque directlyafter gear set assembly the loads shouldbe cho en in a way that the expectedarithmetic mean roughness lays below Ium in order to avoid damaging scoremarks

The presented approach can only beused to predict the behavior of the testedmalerialJlubricant combination and thegiven geometry of the gearsei A predic-tion for operating points which are notrepresented byjhe examined range oftorque and rotational peed is Illghlyuncertain and not verified by practicalinve tigations A tran fer of the results todjfferent size and another tooth geome-III just by conversien of the dependentfactors II T2 Os to other factors such assliiling velocity Vgm and meat contact Ire 0H Jack of lest re ults for thesecoadinon

Method to Avoid Seere MarksScoring on the worm flanks can be

avoided by the use of PVD-coatedworms such as tho e coated withBAUNIT C (Ref 1) or he use of theadditive GH6 Compared to the basic testrunseven higher torque can be appliedNo score marks appear during these tests

on erted oil change after 48 and 144hourshelp to reduce the scoring damageHowever scoring cannot be avoidedcompletely by this particular means

Compared to the basic gear set no

4

[] Roughness Measuring Length

8

1--__ - ----_-

Fig4--Posijon~ wbere measurement oftIle surface are takm to repr-esentscoringdamage

Prof Dr-Ing Wolfgang IPredkiiis Ihf head OIlIf chair of mlcwnical compo-nents indu tria and aUollloilC power Irans-mission 01 Rlltr Unifrsily Iocaled 11 BochumGermany The chairs research focuses all wearoptimiunion of Iorm gears plastic ana sintered~orm f(lars and bl1Ol~eoptimization bull

Prof em Dr-Ing IFriedrich Janhowis lireformer head of rht chair of mechanicalCOflJt()nemIindustrial altd automotive powertransmission elf Ruhraquo University

Dr-Ing Ralf IDinteris employed at Winergy II G a leading pTrJloitltrof drive systems for wind turbines located illVOlrtielFrjldriclLgeldGermany Willergy AG isII subsidiary of A Friedt Flmdu GmbH basedin BCIltJwlr Germany

Dipl middotIng AIexander Rhodeis a research assistant ar the ciJlliroflfJlchimjmiddotcal components indusrial and automotivepo er transmission fIf Ruhr Univer iry Hiwork focuses on the inve jgIJJionof large scoreOIl -orms

wwwpowerIr8nsmfssolJcom www gesrtfulmoio(1y com bull IGEAIRnCHNOLOGY bull MAVJUNIE 2003 19

1120

10 0

e 80Ii

Roughness Measuring LengthNIEW

up the results of everai tests The testing

period is represented by thelengtll ofeach bar Different hatching indicatesseveral defects on the worm flanks The

employed WOIm materials a we1I as the

hardening processes are given with thisfigure

The nominal torque must be exceeded

by far to produce cracks High load leadto heavy pitting and in some casesexcessi ve wear on the worm wheel Thecracks on the worm flanks always occur

at the end of the mating area and how aradial alignment There are two di fferenlareas for the origin of the cracks and theirdirection of growth On the one handcracks start at the inner delimitation of

the contact field with a growth direction

towards the tip of the worm tooth (Fig9) On the other hand crack tart at the

tip and grow toward the tooth (Fig 10)

For both cases the origin ofthe cracks is

located within an area of disadvanta-geous lubrication conditions After the

tests a local transfer of bronze can beobserved in these areas

Theoretical examinations (Ref 2 6)how the smallest lubricant film thick-

ness at the inner delimitation of the wormmating area Compared to other areas onthe flank higher friction factors apply lor

this region The combination of high fric-lion factor and the existing peed leadparticularly towards the end of the mat-ing area to high temperatures on the

flankCracks that start at the tip of a worm

tooth can be observed if large pitied areas

are located on the wheel teeth andincreased wear commence The toothface tip of the worm digs into the feotof the wheel tooth The development of asufficient lubricating film in this area irestrained by the sharp-edged tip of the

worm tooth and the prevailing conditionsof motion

A theoretical analysis of the mating

Fig 5-Arilllmetic mean roughness after testing (336hOllrS) 11=400 11min T2 = 12STIl16MnCrS case-bordenedGZ-CuSn12Ni -

drives For economical bench te ts gear process show a contact or wear betweendrives of the size a 250 mm were the tip of the worm tooth and the foot ofexamined These drives were charged the wheel tooth at the end of the wormwith overloads up to twice the nominal mating area which eorre pond to the

4035

30

25

2015

10 I

05

0 0

400

125

T2l r15 175

test resultsAs metallogrsphic examination

201 MAVJUNE 2003 bull GEAR IECHINOUIGY bull wwwgear bullbull choologycomwwlwpowertrlJosmissioomiddotcom

Fmiddotg6--ATitJlmetic mean ~ol~glmessof the wlee after336 hours

improvement should be expected bytheapplication of polished or crownedworms made of 16MnCr5 case-hardenedsteel Spelling can be observed within thenitrided fringe of gas nitrided worms

Test ResuWfs CrackingRadial cracks on the flanks of case-

hardened worms can be encountered ongear sets with center distances above Ci

200 mm but they predominantly occur

with center distances above a 400 mmHigh sliding speeds and incornplete tart-ing contact patterns lead to high specific

loads on the tooth flanks of the e gear

torque

Figure 8 shows a bar chart that sums

show the flanks of the worm werestressed by very high temperatures with-in the mating area The fringe texture i

thermally influenced Even rehardeningof the Hank urfaces occurred Rehard-ening require temperatures above theau tenite temperature of at lea t aCI

723degClruemal stress analyses reveal that the

loads applied to the wonn flanks lead tohigh tensional internal stresses within thefringe A new worm shows compressiveinternal stre se within the same areas

The change in the stale of iaiemalstresses by high surface temperatures isshown in igllre U In this examplethere is no internal stress at the begin-ning A short application of a temperaturefield to the surface of a part causes com-pre ive tre ses within areas cia e to theurface becau e a free temperature-

caused exten ion i prevented by tile or-rounding colder areas An increase of thetemperature causes higher compressiveinternal stre es If these stresse reachthe hot-strain limit pennanem trainsOCCIIf The value of the hot strain-limitdepends on the temperature Afterremoval of the thermal Ire thedeformed area cool down and the SIXain

are aved as lensional internal tres e Hertzian contact stresses and tangentialloads caused by friction promote thiseffect If the local trength of the materi-al is exceeded by tensional internalstre eSLhe growth of a cruck com-mences

Test ConclusionThe nominal torque ha to be exceed-

ed by far to produce cracks on the wormflank For thi reason Lhese defects

sheuld not be expected under normalworking condition with worm geardri e of the tested size The defects arecau ed by high thermal loads in combi-nation with a local transfer of bronze Inparticular cracks occur in combinationwith heavily pitied worm wheels

That is another reason wily piltingshould be avoided Gas mtridcd wormstend to pan within the nitrided layer

In order 10 reduce the risk of cracks oncase-hardened worms the specific load

1900180017001600150014001300120011001000900BOO700600SOO400300200100

101 s 00 s10 N IN N

os 0 0 0 0 0 0 00 0 0 0 0 0 0- I) II) CD - 11- N N

Rotational worm speedi n [min]

Fig 7-E timatedscoring damage calculated bgtequation 1 GeaHet a = 100 mm =205 161~1nCr5ehIGZmiddotCuSIl12 it operoJion will omillat wad flO running-in nominaltorque TlbullV takenrom mauuJacturer data (Ref n

8OC n T 1500mn81501

10000Nm

5- C 0o IQC) 10

T1

T2

13 1500 mintOOOONm2200 rrun

22tIO I11III

IOOOONm

T7

T8

T91110

16MnCl5 case-hardened + MllkIed

HSMnCr5 case-hardened niIrideltI + polished

34CtNiMc8 nitrideltI=======-n 2200 mIIiI

1 10000 Nm

n 2200 rrIn1 10000 Nrn

o 500

3OCrMoV9 regrinded

1000 1500 20001

lesthlgl period [h]S _1m Pit11l11

Fig 8-Tes results cracking Wheel is GZ-CS112Wi lrtbricant FVAmiddotPG4 LPl655

on the flanks lIould be reduced and me vent local transfer of bronzeloud-capacily again I cuffing should beincreased Ia general this can be reachedby he following meansbull Largest possible starting contact pat-tern

bull Fasr running-m by use of a bronzewith low Iccnglh in combination with amineral oil subsequent oil change anduse of a more efficient polyglycol Use of a lubricant with high load-apacity against scuffing in order to pre-

bull Prevention of pitting on th wonnwheel (for precalculatioa ee Ref 5)

Conh ionsThis research project explores the

criticalloads of hardened worms that arecombined with wheels made of bronzeTwo different types IOf defects largescores and radial cracks are examinedeparately

A map for score damage can be creat-ed It show that all increased amount of

WWWpowflfflansmissioncom wwwgearteclmology com GEAIR TlECHNOLOGy bull MAYJUNiE 2003 21

Fig 9-Worm t~ken from test T3 Originothe crackallle imlerdelimitaliOll otheeontac field (cracks redrawn for bet1erillustraJion)

Fig 1000Worm takenfllom ret Tl Originotlle crack at Ihe tip othl tooth

JensJon

Compression --

Temperature

~traquoJ 10middotbullCompression

=

Hot-strain limit

Permanent strain

Fig ll-Changesoillternal stresses by tliermalloads

scoring is to be expected with anincreased load An increase in rotationalspeed reduces [he expected damageHowever even high overloads do notproduce score marks if high rotationalspeeds apply An effective way to preventscoring is tile use of PV[)coated wormsor the use of a lubricant with the additiveGH6

To generate cracks on the wormflanks the nominal torque must beexceeded greatly For this reason cracksare not to be expected under normalworking conditions with worm geardrives of the tested sire The defects arecaused by high thermal loads in combi-nation with a local transfer of bronzeThe origin of the cracks is located within

areas of disadvantageou lubricationconditions and consequently higlt ther-mal loads Material tests showed ther-

mally-influenced fringe texture and inorne cases even zones where reharden-

ing occurred Internal stress analy e indicate a

change from initially compressive tohigh tensional internal stresses causedby high nash temperatures These stress-es lead to cracks if the local strength ofthe material is exceeded In particular cracks occur in combination with heavi-ly pitted worm wheel That is anotherreason why pining should be avoided

Further information can be obtainedfrom the final report on research project237 of the FVA (Ref 5)0

This paper was printedwilh the permis-sian of the publiisherrom VOlmiddotGesellschaft EntwicklungIKonstruktiolJVertrieb (iEdI International Conferenceon Gears Vot 1 VDI-Barichts 1665Dilsseldorf VDI-Verlag 2002 Ipp435-449

Rd renees1 Balzers AU Fi rmenschrift bullDel Weg lUumiddoterliissigeren Bauteilen Liechtenstein19932 Bouche B Relbungslahlen lionSchneckengetrieben imMischreibllllgsgebiet Dissertation Ruhr-Universitiil Boch-um 19913 DlNmiddotEntwurf 3996 Tragfiilligkeits-berechullg von Zyiillder-Sclmeckeng-etrieben mit Achsenwinkel E90deg Entwurf1195bull4 fV A-Porschungsheft 316 SynthetischeReerellzole Datensal1lmlung arSyntheseole 19905 FVA-Albeilsblatt zum Forschungs-vorhaben Nr 12f1V Versuche zUI

GrfJbchentragflihigkeil Vorl Schneck-engetrieben Stand Juni 19966 fVAmiddotPbrschungsvorhaben 237 Schneck-enrragiiJligkeitsgrenun ermitteln linderilollen Heft 518 der Forschungsvereinigung Antriebstechnik 19967 Kalalog der Firma Thyssen MUTAXSchneckenrodsatze 1-7508 Kltiber Finnen chrift Kliibersyntli GH6-Ole Produktinformation SUScI Ausgabe03939 Predki W Hertzselle Driicke Sch-mierspalthohen und Wirkulgsgrade vonScbneckengetrieben Dissertation Ruhr-Universitiit Boehum 1982

Tell Us What You Think Viit WWIIfIflrtet tabull RIll dli Irticlebull Req more infonlllltiDilbull ConllCt the 1IIIon or OIJIIizItiaa tionldbull Mike I gestioaOr cln (147)CJ1-6IM to tllk to OIl of our H-tori

22 MAVJUNE 2003 bull GEAR TECHINOLlJGY bull IWlwwgstulecmoogcom wwwpowertuJnsmssoncom