1、09FTM18AGMA Technical PaperDoes the Type of GearAction Affect theAppearance ofMicropitting and GearLife?by A. Williston, A and, are basedupon solid material stress/strain theory that hasbeentemperedbypractical, real-worldexperience.Statistical probabilities are used to account for vari-ations in mat
2、erial, manufacturing, and geometry.However, an additional variable may be presentwhen looking at the probability of wear in involutegearing.As part of a testing program for wind turbinegearing,aninnovativetestfixturewasdevelopedforparallel shaft gearing that allows a comparisonbetween the two differ
3、ent types of loading: speedreducer (where the pinion drives the gear), andspeed increaser (where the gear drives the pinion)under identical conditions. Preliminary testingresulted in a number of involutegearsets thatexpe-rienced both macro- and micropitting failures.While micropitting can appear ver
4、y early, macropit-ting is usually a fatigue failure that follows somesemblance of predictability. These macropittingfailures occurred consistently in a shorter period oftime and cycles than currently theory predicts.Further, the failures only occurred in the drivingmembers! Additionally, micropittin
5、g did not appearin a consistent manner in the gearing again beingmore prevalent in the driving members.Test fixture configurationThe original test program focused on loading full-sized wind turbine gearboxes (108kW rated power)to 240% of rated torque. After suffering anunexpected bearing failure, it
6、 became readilyapparent that continued testing in this environmentwouldprovecostprohibitiveleadingtodevelopmentof smaller custom gearboxes with a dedicated testfixture. Whilethetestingintheoriginalconfigurationwas not continued, relevant information wasgleaned in the light of results found during te
7、sting inthe3-holefixture.Description of test methodThe gearboxes were mounted to a “back-to-back”test stand (also known as a four-square test) thatallowed a great amount of torque to be circulatedthrough the gearboxes yet minimizing the truepowerusagetoonlytheinefficienciesofthesystem.This method of
8、 torque application assures bothgearboxes are subject to similar loads. However,duetotheflowoftorquethroughthesystem,thetwogearboxesarenotloadedinthesamemanner. Oneisloadedasaspeed reducerwhile theother actsasa speed increaser, thus changing approach andrecess action, bearing loads, deflection, etc.
9、Figure 1 depicts the original test stand using full-size wind turbine gearboxes. This unwieldy fixturewasreplacedwiththemuch moreflexible andcost-effective 3-hole test configuration (Figure 2).Figure 1. Original back-to-back test stand arrangement for shaft-mountedwind turbine gearboxes4Figure 2. Ph
10、oto of the 3-hole test fixtureA surface fatigue failure mode was desired duringtesting. Proper selection of the applied torque wasnecessary to promote a contact fatigue failure onthe gearing. If the applied load is too high, toothbreakage may result potentially damaging thetested component as well a
11、s the test equipment. Iftheloadistoolow,thetime-to-failureofthe testwillbe prohibitive. It was necessary to find a balancebetween these extremes.Itwasalsoimportanttoconstrainasmanyvariablesnot related to the gearing geometry as possible.These constraints included, in part: gearingmaterial, heat trea
12、tment, applied torque, and lu-brication. Of key importance is the lubrication. Ahightestloaddictatedthatanassuredsupplyofcooloilbeavailabletothebearingsandgears. Also,anycross-contamination must be prevented. There-fore, each gearbox was supported by its own oilreservoir, pump, filter, and heat exch
13、anger.Instrumentation forboth gearboxeswas similarandincluded: bearing temperatures, lubricant tempera-tures at various points, lubricant cleanliness andvibrations in three axes. Additional signals trackedcirculated torque, motor amperage and speed.3-hole gearboxThe concept for the 3-hole gearbox wa
14、s to build amuch simpler and less expensive test gearbox thatwould allow gearing to be optimized for the appliedtest loads. However, by reducing the size of thegearing (specifically center distance) connectingthe test gearboxes proved to be troublesome.Couplings, torque actuators, and torque transdu
15、c-ers sufficient to carry the test load were too large tofit when only one gearset was used. By expandingthe scope to two gearsets with the same ratio, theoriginal 4.0” center distance was increased to 8.0”.As shown in figure 3, the concept has a pinion driv-ing one of two identical gears on an inte
16、rmediateshaft, while the othergear onthe intermediateshaftdrives another pinion. With the pinions having thesame number of teeth, the ultimate ratio is 1:1.Thus exact tooth counts between the two test gear-boxes are not necessary, and various gearingdesigns can be compared and optimized withoutlimit
17、ations.5Figure 3. Design for the 3-hole test gearboxes (in section view).An additional advantage to this means of testing isthatineachgearbox, thetwo gearsetswill beactingin different manners. For example, if the torque isappliedthroughtheleftsidepinion (inFigure 3),thispiniondrivesthefirstintermedi
18、ategear. Astheloadpasses through the gearbox to the right side, thesecond intermediate gear drives the right sidepinion. Much discussionin thegearing industryhascentered on the difference in affects of a reducergearset (pinion driving a gear) versus an increasergearset (gear driving a pinion). Speci
19、fically, thesliding action between the gear teeth determineshow well oil can provide lubrication between thecontacting flanks. Some in gear design adamantlyrecommend different gear geometries for reductiongearing versus increaser gearing for this purpose.Others in the industry are not convinced of t
20、henecessity! This configuration allows these theoriesto be tested.To keep the size of the test manageable, a gearingcenter distance of 4.00” was chosen. As the centerdistance increases, the amount of applied load in-creases (since load capacity is proportional to thecube of the center distance). All
21、 gearing in currenttesting are expected to have the same materialsand heat treatments as production wind turbine(WT) gearboxes (High quality 8620 steel to AGMA2001class2specifications),however,futuretestingwith other materials may be considered.A robust housing design using an industry provendesign
22、incorporating readily available tapered rollerbearings was chosen. A stiff two-piece designwitha split at the bearing bore center lines allowed foreasy part changes and reassembly. Within thegiven space of the 4.0” center distance, as muchbearingaspossiblewaschosento beassured ofnobearing failures.L
23、ubricating oil is supplied to each bearing and toboth sides of the gear meshes. The bottom of thegear housing is open, allowing oil to drop into areservoir. Each gearbox has separate lubricationsystems.Gearing designGearing was designed to achieve expected L1livesof approximately 200 hours with the
24、original testload (pinion torque = 6300 lb-in 1800 rpm).6When calculating the average lives at this load(L50 1E9hours),itwasrealizedthattoomuchtimewouldpossiblypassbetweensubsequenttests. Anincrease in load to 7000 lb-in and speed to 2200rpm reduced the average life to approximately23700 hours (L1=
25、42 hours; L10= 740 hours). Allanalyses were conducted using the AGMA GearRating Suite of software for involute gearing. Addi-tionaltorqueincreaseswereplanned,butearlygearfailures (surface contact, macropitting) suggestedmaintaining the current loading.All gear components were analyzed for deflection
26、susing an LVR analysis to determine the properamountoftoothmodificationsnecessarytoachievea smooth, balanced load across the active toothflank. This analysis required creating an analyticalmodel of the gearing, supporting shafts, bearingstiffness,andhousingstiffness. Modificationswerekept symmetric
27、to allow placement of the partswhere necessary during testing, however it isacknowledged that this resulted in a compromisemodification. Light-load and full-load contact pat-terns appeared to confirm the modifications.A summary of the gearset geometry for these testsis shown in Table 1. A complete l
28、isting from theAGMAGearRatingSuiteisincludedin AppendixA.Table 1. Test gearing geometryInvolute gearsetPinion GearNumber of teeth 16 35Normal diametralpitch (module)7.000 (3.629)Normal pressureangle20.0Helix angle 18.6974Operating center dis-tance4.00”Outside diameter 2.793 5.757Face width 1.50 1.45
29、Gear ratio 2.1875Addendum modifica-tion coefficient0.4328 0.7771Tooth modifications 0.0018 tiprelief0.0018 tip relief0.001” circularcrown0.0006” circularend reliefThe sliding velocities and specific sliding ratio forthe gearset operated at the test speed of 2200 rpmare given in table 2.Table 2. Slid
30、ing velocities and specificsliding for the test gearsetPinion rollangle (deg)Slidingvelocity (in/s)Specific slidingratioang egPinion Gear Pinion Gear14.02 (SAP) -93.30 93.30 -1.4696 0.595119.51 (LPSTC) -57.09 57.09 -0.6463 0.392628.16 (PITCH) 0 0 0 036.52 (HPSTC) 55.18 -55.18 0.3336 -0.500742.01 (EA
31、P) 91.38 -91.38 0.4804 -0.9245Approach action: 49.48%; Recess Action 50.52%,maximum Specific Sliding ratio: -1.4696The original project specification was to duplicatereal-world operating conditions (in the wind indus-try) where possible; therefore a commonly usedISO320 viscosity gear oil was chosen
32、for testing.Sump temperatures were adjusted from an original155F to 140Finanattempttominimizetheappearance of micropitting (yielding a minimumspecific film thickness of 1.31 at a gear toothtemperature of 176F). See Appendix D for asample calculation. Additional testing could look atthe affects of di
33、ffering lubricants at differenttemperatures.Test fixtureThepurposeofthe testfixture isto properlysupportthegearboxes,supplyloadtothegearing,rotatethegearing, and supply clean lubricant of a specifiedtemperature. During operation, data should begathered to monitor the test including torque ap-plicati
34、on, temperatures, vibrations, and lubricantcleanliness.Duringtesting,alargeamountoftorquewasappliedto the system. This torque acts to “twist” one testgearbox with respect to the other. Therefore, itwasimperative that the mounting surface of the fixturebe stiff enough to maintain gearbox shaft alignm
35、entunder test loads. Additionally, to create a compacttesting area, the fixture was designed with internalreservoirs for accommodating all the test oil re-quired. Further, to prevent cross-contaminationbetween the two gearboxes, separate reservoirs7were made. Each reservoirhad twobaffles toallowpart
36、icles to settle and to reduce foaming.The intent of the design was to create a platformformountingnotjust thegearboxes butthe drivemotorandlubricatingpumps. Thisplatform,alongwiththerest of the test equipment (heat exchangers, filters,and computer control system), would then be ableto fit into a sta
37、ndard 20 foot shipping containerallowing for unobtrusive placement outside ofnormal manufacturing space. This was necessi-tated by requests to free up the main test platformand reduce noise.Lubrication was straight forward with oil being di-rectly applied to each bearing through the bearingcaps and
38、to each gear mesh via oil jets. A dry-sumpdesignallowsalllubricatingoiltofalloutofthegearboxandintothefixturetank. Eachgearboxhasa separate lubricant system with separate filtersand heat exchangers.All design intents were met with the fabricatedfixtureshowninfigure4. Eachlubesystem (oneforeach gearb
39、ox) uses two oil pumps. One pumpdirectsoilfromthereservoirthrougha3micronfilterinto a distribution manifold that in turn, feeds oil toeach bearing and to both sides of the gear meshes.Oil flow is controlled with small orifices in themanifold, thereby ensuring oil flow to all compo-nents. A second pu
40、mpcirculates oilfrom thereser-voir through a heat exchanger, past a heater, andbackintothereservoirthusmaintainingaconsistentoil temperature.After initial alignments, the test gearboxes werepinnedtothetopofthefixture. Itwasfoundthatpartchangeovers were more easily accomplished if agearboxwasremovedf
41、romthetestfixture. Taperedpins provided assurance that the shafts would bealigned when the gearbox was remounted to thefixture.Load applicationRotation was provided by a directly coupled motor.By using a hydraulic rotary torque actuator with arotating union, torque could be applied in either di-rect
42、ion without affecting shaft alignment - allowinguseofstandardcouplings. Beingabletochangethetorque direction was necessary for testing bothflanks of the gear teeth. Torque was monitoredusing a torque transducer.A standard 3-phase motor with a variablefrequency drive(VFD) was chosentodrive thesys-tem
43、. Using the design power of 244 hp ( 2200rpm) and an assumed total 6% loss in the system,14.7 hp will berequired. With windinglosses duetothe increase in speed and to keep motortemperatures down, a 30 hp motor was used.Figure 4. Top view of the 3-hole test fixture configuration8Operation of this 3-h
44、ole test allows two differentmesh conditions (a reduction gearset and anincreasergearset)ineachgearbox. Therefore,four(4) gear meshes are being tested at a time. Inaddition, subsequent testing may be conducted byreversing the direction that torque is applied (as-suming no tooth breakage). As an exam
45、ple, for agivendirectionoftorque,gearsets5-6and3-4mayact as increasers while gearsets 1-2 and 7-8 willact as reducers. This changes as the torquereverses directions to gearsets 5-6 and 3-4 actingas reducers and gearsets 1-2 and 7-8 acting asincreasers. Determiningwhetheragearsetisactingas an increas
46、er or a reducer is based upon the mo-tordirectionandwhichgeartoothflanksareloaded.InstrumentationAnumberoftestparametersweremonitoredduringthe test. A detailed listing of those instrumentationpoints is given in Appendix B.Controlled valuesMotor speed and applied torque were controlled bynumerical in
47、puts to the computer control system.Adjustments to the motor variable frequency drive(VFD) set the motor speed, and a computercontrolled hydraulic rotary actuator was used toestablish torque with a torque transducer controlfeedback.Monitored valuesOf high importance was the monitoring of bearingtemp
48、eratures and gearbox vibrations so that anyadverse deviations could be captured. If a warningthreshold was surpassed, then visible and audiblealarms were set in additionto automaticshut-downof the test.From previous test experience, temperature couldnot necessarily be considered a dependablevariable
49、. Also, total vibration did not reveal aproblem until major damage occurred. However,one aspect of vibration did directly correlate to thebeginning of damage (Figure 5 - see vertical linethat corresponds to test documentation made bythe operator of a change in sound in the test). Inone direction the levels were generally small.However,oncedamagestartedthisverysmallcom-ponent of the total vibration dramatically changed.Analysis of the starting value of the small vibrationcomponent was used to establish an alarm value.Figure 5. Vi
