1、07FTM12The Effect of Start-Up Load Conditionson Gearbox Performance and Life -Failure Analysis and Case Studyby: R.J. Drago, Drive Systems Technology, Inc.TECHNICAL PAPERAmerican Gear Manufacturers AssociationThe Effect of Start-Up Load Conditions on GearboxPerformance and Life - Failure Analysis an
2、d CaseStudyRaymond J. Drago, Drive Systems Technology, 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.AbstractMost gearboxes are rated based on one or more of several
3、criteria such as peak applied load, nominaloperating load, prime mover rated power, driven machine load spectrum, maximum expected overload, etc.These load conditions are used to predict the expected operating life of the gear (and bearing) system. Inmanycases,especiallywheretheloadand/orspeedvarysi
4、gnificantlyduringthenormaloperationalcycleofthe gear system, a great deal of attention is paid to the spectrum of load conditions, often resulting in thedefinitionofextensiveload/timetableswhichareusedinaMinersrule(cumulativedamagetheory)orWeibullapproachtothecalculationoftheexpectedlifefor gear sys
5、temsunder thesewidely varyingload/speed/timeconditions. While all of these are very valid methods of determining the expected life of a gear system, oneimportant factor is very often neglected the starting load! If a gear system is started up and then runcontinuously for long periods of time or if t
6、he starting loads are very low and within the normal operatingspectrum,the effectofthestartupconditions may,and oftenis,insignificantin thedetermination ofthe lifeofthe gear system. Conversely, if the starting load is significantly higher than any of the normal operatingconditions and the gear syste
7、m is started and stopped frequently, the start up load may, depending on itsmagnitude and frequency, actually be the overriding, limiting design condition. In these cases, failure toaccountforthestartupload1conditionsinboththebasicdesignofthegearsystemandintheproperattentionto the details of tooth m
8、odification can lead to premature, seemingly unfathomable, failures.Thispaperaddressestheissueofstartuploadinganditseffectontheperformanceofaseriesofgearboxesinanindustrialapplicationbywayofaspecificcasestudy.Adescriptionofthefailuresthatoccurredandthetestprogram and results that led to a definition
9、 of the root cause of the failures and a path of correction are alsopresented.Whilenotaddressedinthispaper,itshouldbenotedthatshutdownloadcan,inrarecases,alsobeextremelysignificantinthedesignofthegearssystemduetotheveryhigh,shorttermloadsthatcanbeappliedincertainapplications. In one case in the auth
10、ors experience, for example, a gearbox that was used in a high speedturbo compressor application failed due to very high loads that were caused by an improper shut downprocedureonthedrivenmachinethatgeneratedhighbackpressuresandinertialforcessufficienttodamagethe gears.Copyright 2007American Gear Ma
11、nufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314October, 2007ISBN: 978-1-55589-916-51The Effect of Start-Up Load Conditions on Gearbox Performance and Life- Failure Analysis and Case StudyRaymond J. Drago, Drive Systems Technology, Inc.IntroductionWhen gearboxes are
12、 used in applications in whichthe connected load has high inertia, the startingtorquetransmittedbythegearboxcanbeverymuchhigherthattheratedloadoftheprimemover.Powerplants often require several evaporative coolingtowers or large banks of air cooled condensers(ACC) to discharge waste heat, Figure 1. B
13、ecauseof the very large size of the fans used in these ap-plications, they fall into this category of high inertiastarting load devices.Figure 1. Air cooled condenser bankThe typical evaporative cooling tower or air cooledcondenser unit is composed of an electric motorwhichdrivesalargefanthroughagea
14、rbox. Thisas-semblagegeneratesairtoflowforthepurposeofre-moving the waste heat from the power generationprocess. In a typical air cooled condenser applica-tion, many identical units (typically 20 to 40) areinstalledinanarraytoprovidethetotalcoolingre-quired for a large plant. Depending on the type o
15、fplant, these geared fan units may operate almostconstantly(“baseload” plants)or veryintermittently(“peaking” plants).The fans in these units are generally very large andthushavesignificantinertia,Figure3. Whenstartedfrom zero speed, a very high torque is required toaccelerate the fan to normal oper
16、ating speed. If thefan is started infrequently and run continuously forlong periods of time, this high starting torque is ofminimal significance. However, when the fan isstarted and stopped frequently, the number ofcycles at thehighstartingtorquecanaccumulatetoa point where they can cause extensive
17、fatiguedamage even if the gear system is adequatelyrated. Where the gear unit is marginally rated, veryearly, catastrophic gear failure is often the result.Figure 2. Air cooled condenser schematicFigure 3. Installed air cooled fan2Aspartoftheoverallinvestigationofseveralfailuresin such gearboxes, we
18、 measuredstarting torqueona typical installation, examined many failed gears,and calculated the load capacity ratings for thegearboxes under actual operating conditions. Thispaper describes the failures observed, the testingconducted, the data analyses and the effect of thehighmeasuredstartingtorque
19、onthelifeandperfor-mance of the gear systems. The test results re-vealed surprising results, especially during startswherethefanwasalreadywind-millingduetonatu-ralairflowintheACC bank.In thefinal analysis,theimportance of appropriate profile modifications isalso clearly demonstrated, Figure 4.Figure
20、 4. Spalled single toothThe initial incidentA failure wasreported inthe lowspeed gearset ofatriple reduction, single helical gearbox, Figure 5.The gearboxes are used in anAir CooledCondens-er (ACC) facility at a power plant and had been inservice for about four years at the time thisinvestigation wa
21、s initiated.Theinitialfailedunitisoneof thirtyidentical, unitsatthe same facility, Site A. An identical bank of units,SiteB,isalsoin servicefor asimilar timeperiod ata“sister”powerplantinthesamestate.ThefirstfailedSite A unit that we examined had accumulated justover 13,000 hours ofloaded serviceand
22、 duringthattime had been subjected to more than 7,000 start/stop cycles. All of the gears in these units arecarburized, hardened, and profile ground.Subsequent to this first failure, several additional,very similar, failures were also discoveredon SiteAgearboxes. Additional, ongoing periodic visual
23、in-spections of the remaining gearboxes at Site Acontinue toreveal additionalfailures. Carefulevalu-ation of the additional damaged low speed gearsshowed them to have very similar characteristics. Itis clear at this point that all of the units at Site A arevery likely to eventually suffer similar fa
24、ilures overtime.Figure 5. Triple reduction gearbox(shown in normally installed attitude)It was initially reported that, though the systems(gearboxes, fans, motors, controllers, etc.) wereidentical at the two sites, the gearboxes at Site Bhad not experienced any failures. As the investiga-tion progre
25、ssed, however, we did find someindications of very earlystage failureson lowspeedgears at Site B which were very similar in nature tothoseobservedonSiteA.Theextentofthedamageobserved on the Site B low speed gears was muchless than that observed on the low speed gears onthe Site A gearboxes and to da
26、te none of the Site Bgearboxes have actually been removed fromservice due to low speed gear damage.Obviously, since the units are truly identical at bothsites,thisconcentrationoffailuresatSiteAandlackof any apparent failures at Site B was something ofapuzzle.Evenwiththediscoveryofsomeveryearlystage
27、damage on the Site B low speed gears, it isclear, that something is different between the twosites.Afterindepthevaluationofthecharacteristicsofthesites, the only significant difference identified wasthe fact that Site A was operated as a “peak load”plant while Site B was operated as a “base load”pla
28、nt. In the operation of a base load plant, the fanswould run almost continuously with very limitedstart/stop cycles. In contrast, the operation of a3peaking plant involves multiple, frequent start stopcycles.Whilethenumberofstart/stopcyclesisdifferentbe-tween the two sites, the nominal power draw on
29、 themotors at both sites A and B during normal, fullspeed, steady state operation was found to be verymuch the same.The failureInitial evaluation of the gearbox indicated that theprimary damage was on the low sped gear set. Thehigh speed and intermediate gear sets did showsomedamage butit waslargely
30、 consequential.Thelow speed gear, Figure 6, exhibited a very hard lineof contact near the tip of the tooth. This hard, local-izedregionofcontactindicatesthatthetoothdidnothave sufficient profile modification for the appliedloading. The shape of the damaged area suggeststhat the gear set had some cro
31、wning (probably ap-plied to the pinion) to accommodate misalignment.The damage pattern observed on the gear toothloadedflanks,however,isslightlyheavierontheleftend of the face (Figure 6). The relatively short areaof damage indicates that the crown applied to thelowspeedgearmeshislikelytolargeforthea
32、ppliedloading. This tends to concentrate the load in thecenter portion of the face while unloading the endsof the face. This increases the localized stresses inthe heavy contact region. This is very apparent onthe loaded tooth flanks shown in Figure 6.Figure 6. Low speed gearThe location of the dama
33、ge observed on the lowspeed pinion, Figure 7, corresponds to the damageobserved on its mating gear; however, the level ofdamage on the low speed pinion is much more se-vere than that on the mating low speed gear. Sincethe pinion sees many more cycles than the pinion,thedisparityinthelevelofdamageise
34、xpected.Thelocation of the very hard line of contactnear thetipsof the gear teeth corresponds closely to the similarhardlineofcontactatthelowestcontactpointonthepinion teeth. As noted above, this hard contact indi-cates a lack of adequate profile modification on thepinion or gear or both.Figure 7. L
35、ow speed pinionThe specific failure mode observed on the pinion isspallingwhichisatypeofsurfacedurabilitydistress.Specifically,spallingisafatiguemechanismthatoc-curs when very high local stresses initiate cracks ator near the tooth surface. These cracks progressinto hard caseon thetooth surfaceand p
36、rogressupalong the tooth surface from the lowest contactpoint in the direction of sliding on the tooth surface.As the cracks progress, material on the tooth sur-faceis“undermined”andeventuallyrelativelylarge,somewhat fan shaped pieces of tooth surface areliberated. This mechanism is clearly shown in
37、 Fig-ure 7. As the cracking progresses through thecarburized case on the loaded flanks of the teeth, iteventually propagates across the tooth thickness,generally fairly close to the tips of the teeth andfractures the entire case off the tooth, even extend-ing back onto the coast flanks. Essentially,
38、 the car-burized case is “peeled” from the loaded andunloaded tooth flanks. More detail on this failuremode will be presented below.4In addition to the hard line at the lowest contactpoint, the pinion distress also suggests that theteetharesomewhatover crowned.This conditionissimilar to that observe
39、d on the mating low speedgear, as noted above. Careful examination of bothFigures 6 and 7 also shows small areas of theloadedtooth surfacesat bothends ofthe facewidthwhere the original witness marks from the tooth fin-ish grinding operation are plainly visible. Theseareas of relatively light loading
40、 further indicate thepossible over crowned condition. In addition theyalso indicate that misalignment across this gearmesh is not a major factor in the occurrence of thelowspeedgearsetfailure.Whileitmaybemoreofafactoronotherlowspeedsetsitisstillnottheprima-rycausativeagentthoughitiscertainlycontribu
41、tory.Load capacity evaluationIn order to better understand the cause of thefailures observed, we calculated the basic loadcapacity rating of each gear set in the gearbox. Theload capacities are best understood by looking atthe Service Factors for each gear mesh, assummarizedinTable1.Table 1. Service
42、 factorsParameter Strength DurabilityHigh speed Pinion 2.6 1.2pGear 2.9 1.2Intermediate Pinion 2.0 1.0Gear 2.6 1.1Low speed Pinion 1.5 0.8pGear 2.1 0.9NOTE:Ratings based on 175,000 hour required life (24hours/day, 365 days/year, 20 years) using AGMAGrade 1 materials to motor nameplate power, perAGMA
43、 2001-C95.Asclearlyshownbythe datain Table1 althoughthestrength ratings are above the applied power, thedurability ratings of the low speed stage pinion andgear are less than the applied power. These lowpower ratings resultin durabilityservice factorsthatare less than Unity. In an application such a
44、s this,we would normally recommend a minimum ServiceFactor of at least 2.00. The lower Service Factor onthe low speed pinion relative to the higher (but stillbelow the recommended minimum) Service Factoron the mating low speed gear is consistent with therelativelygreaterdamageexperiencedbythepiniona
45、s compared to the gear. Based on these ratings,prematuredurabilityfailuresof thelow speedpinionand gear would be expected. The relatively highstrength Service Factors would allow the units tocontinueoperatingfora longperiod oftime afterthespalling damage had initiated and progressed,though tooth fra
46、cture would ultimately beexpected.In this particular application, the power supplied bythe motor during normal steady state operation isslightly less than the motor nameplate rating thusthe durability service factors for the low speed gearsetareactuallyslightlyhigherthanshowninTable1(very close to u
47、nity).While the low durability service factors are certainlyofsignificantconcern,ofandbythemselvestheydonot fully explain the rather catastrophic failures thatoccuronthelowspeedgearsetsatSiteA,especial-lyinviewofthefactthatthedurabilityservicefactorsfor the intermediate and high speed gear sets areo
48、nly slightly higher (there were no reports of cata-strophicfailureorevensignificantsurfacedamage),andtherewerenocatastrophicfailuresofanyofthegears at Site B.In orderto betterunderstand justwhat ishappeningwe had to first develop the actual failure scenariowhichexplainsthecatastrophicdamagethatoccur
49、son the low speed gears of the Site A gearboxes.Failure scenarioWhile it may seem obvious that the low durabilityratings of the low speed gear set, particularly thepinion, are fully responsible for the failures ob-served,theyarenotthesolecause.Someevidenceof a small amount of misalignment is apparent andthis certainly plays a role in the failure as well bygenerating a load maldistribution across the facethatresultsinhighlocalized stresslevels. Further,itappears that there may be too much crowning onthe low speed set (either by error or design) whichalso results in load concentrat
