1、08FTM13AGMA Technical PaperHydrogen and InternalResidual Stress GearFailures - SomeFailure Analyses andCase StudiesBy R.J. Drago andR.J. Cunningham, Drive SystemsTechnology, Inc.Hydrogen and Internal Residual Stress Gear Failures - SomeFailure Analyses and Case StudiesRaymond J. Drago and Roy J. Cun
2、ningham, 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.AbstractHydrogen and internal stress failures are relatively rare; however, when they
3、 occur they are alwaysspectacular, often very costly and sometimes quite catastrophic. While hydrogen and internal stress issuesaregenerallyrecognizedassignificantinthedesignandmanufactureoflargergears,theyarealsoimportantfor smaller gears as well. Unfortunately, such failures are difficult to diagn
4、ose and, because of their oftencatastrophic nature, may well go unrecognized.This paper presents, via illustrated actual case studies, the mechanisms by which these failures occur, themannerinwhichtheyprogressandmethodsfortestingfinishedgearsforthepossibilityofinternalproblems.In addition, precautio
5、nary steps that can be taken during design, manufacture, heat treatment and qualitycontrol to minimize the possibility of these problems occurring in a finished part along with similar stepsrequired to prevent any flawed gears from entering service are also presented and discussed.Copyright 2008Amer
6、ican Gear Manufacturers Association500 Montgomery Street, Suite 350Alexandria, Virginia, 22314October, 2008ISBN: 978-1-55589-943-13Hydrogen and Internal Residual Stress Gear Failures - Some Failure Analysesand Case StudiesRaymond J. Drago and Roy J. Cunningham, Drive Systems Technology, Inc.Introduc
7、tionUnlike most fatigue failures, hydrogen and internalstress related failures almost always occur withvirtually no warning at all. No debris is generatedand,sincethe cracksprogress fromdeep withinthepart to the surface, there isgenerally noobservablechange in noise or vibration level, until the ver
8、y lastmomentwhena spectacularfracture failureoccurs.The very large, double helical, carburized pinionshown here (figure 1), for example, operated satis-factorily for many months before suffering a cata-strophic failure (the shaft fractured into two halves,stopping the mill immediately and causing an
9、cillarydamage as well). While this fracture occurred aftera period of successful service time, it is more com-monforsuchfailurestooccurveryearly inthe lifeofthe part. In some instances, dramatic failures haveoccurred while post heat treatment processing wasunder way in the factory!Figure 1. Large mi
10、ll gear failure(note leg at left for scale)We tend to think of steel as a solid medium,however, given the right conditions, nascenthydrogen will actually migrate interstitially throughthe steel matrix to gather in a common location thatwill then form the nucleation site for a crack thatleads tothe d
11、ramaticfailure eventsthat typifytheseproblems. Internalresidual tensilestresses canactalone to cause similar failures or they can work inconcertwithhydrogentoacceleratethe“fatal”crackpropagation.The hydrogen problemControlling the hydrogen content in gear steels isimportant because hydrogen entrapme
12、nt can havea very detrimental effect on service performance ofsteelgears. Onlyafewpartspermillionofhydrogendissolved in steel can cause internal hairline cracks(flakes), hydrogen embrittlement, hydrogen blister-ing and loss of tensile ductility, particularly in largesteel gears with thick cross-sect
13、ions.When the hydrogen content of the molten steelused to cast the ingot from which a gear blank willultimately be fabricated is in excess of the solubilitylimit of hydrogen in solid iron, the hydrogen will berejected during the solidification process. The fail-ure mode which results is often called
14、 a “hydrogenburst” because of the visual characteristics of thefracture face, as will be described below.Hydrogen can be entrapped in a steel gear blankduring the initial steel melting process, duringprocessing operations during the manufacturingprocess(e.g.,acidetch inspectionfor hardfinishingburns
15、) and during subsequent electroplatingprocesses. Internal residual tensile stresses aregeneratedduringsolidificationofthemoltensteelaswellasduringheattreatment,particularlyduringthequench operation. It is also quite possible, and insome cases likely, that both hydrogen entrapmentand internal residua
16、l tensile stresses may occur to-gether.As the thick slab that will be used to fabricate aheavy section steel gear is rapidly cooled, there willbe little diffusion of hydrogen out of resultant bar.This type of cooling can also result in increased in-ternalresidualtensile stresses. Hydrogensolubility4
17、decreases with adecreasing temperature,but ifthehydrogen is internal it will migrate to tensile stresslocations,andformhydrogengas. Thusifhydrogenispresentinthesteelasitis beingpoured, therewillbe a build up of H2 gas pressure in the steel matrixduring rapid cooling. For the limiting case of no hy-d
18、rogen diffusion, the pressure generated by theen-trapped hydrogen willbe asshown inFigure 2(dataobtained from Reference 1). Clearly, as even verysmallhydrogenconcentrationsaslowas4ppmcangenerate significant internal gas pressures. As thehydrogen content increases, the internal gas pres-sure increase
19、s at an exponential rate. Even atcon-centrations as low as 8 ppm, the pressure increaseisobvious. Atdoubledigitconcentrations, thepres-sure can literally “blow” a steel gear apart.Temperature,DegreesFFigure 2. Hydrogen pressure buildup duringrapid coolingWhilethetemperaturerangeofthisdataislimited,i
20、tisclearthathigherconcentrationsofnascenthydro-gen can produce very significant internal hydrogengas pressure as the steel cools, especially as itreaches room temperature. As nascent hydrogenmigrates these stresses build up and over time andmaynotproduceanyeffectatalluntilthegeariswellalongineitheri
21、tsmanufacturingprocessingorevenafter being placed in loaded service. This is one oftheprimaryproblemswithahydrogentypeoffailure there is no known time limit for when fracture willinitiate.The authors have participated in failure investiga-tions where massive fractures occurred while alarge gear (usu
22、ally a solid on shaft pinion) wassimplysittingonthefloorawaitingthe nextprocess-ing operation (most frequently after tooth finishingby grinding or hard cutting has been accomplished)and after the gear had been in loaded service forseveral years. The sound of a hydrogen burst onthe shop floor is some
23、thing similar to the report of ahigh powered hunting rifle discharge. It does nottake much imagination to realize that such a failurein the shop is, to say the least, a bit unnerving!In service hydrogen burst failures, while not asreadily observable as those that occur on the shopfloor, are equally
24、spectacular in that they usuallyresult in a sudden, complete gear blank fracturefailure. This type of failure is most often observedon large solid on shaft pinions and when the “shaft”fractures, consequential damage can be and oftenis very extensive. The pinion shown in Figure 3, forexample, fractur
25、ed due to a hydrogen burst afterabout one year of loaded service.Figure 3. Complete pinion “shaft” fracturedue to hydrogen burst5The red arrows (1) in Figure 3 indicate the initialcrackduetotheinternalhydrogenburst. Thatcrackoriginated near the centerline of the shaft sectionunder the teeth and prop
26、agated outward until itreached the toothed section of the pinion. At thatpoint,thepinionshaftfracturedintwo. Thatfractureprecipitated the tooth fractures visible on the leftside of Figure 3. It should be obvious that a large(over 4 feet outside diameter and just less than 16feet long) pinion such as
27、 this one carries with it a lotof energy that must be dissipated quickly when itfractures in half. As the pinion grinds to a suddenstop, an enormous amount of energy must beabsorbed by the supporting bearings and the sur-rounding housing structure. In this case, the toothfractures precipitated addit
28、ional cracking as shownby the green (2) arrows in Figure 3. In addition, thebearing supporting this pinion just adjacent to thepurple arrow (3) in Figure 3, was immediately andcompletely shattered as Figure 4 shows.Figure 4. Fractured bearing on pinion shaftIn this case, the housing structure remain
29、ed intactand no external damage occurred. Where thehousing structure was not sufficiently strong toretain the fractured components, external damagehas occurred. In those cases, large fracturedsectionswereejectedfrom thehousing. Obviously,thepotentialforpersonalinjuryexists;however,nei-ther authorhas
30、 directknowledge ofsuch asituationactually occurring.Numerous sources of hydrogen exist during thesteel making process. One of the primary sourcesof hydrogen is water associated with the steelmaking ingredients such as an impurity in the alloyadditions or the scrap used or the carburizers (car-bon i
31、ncreasing materials). Water can also beintroduced during the ingot pouring operation if thetundish or the ceramic mold lining has not beensufficiently dried, or feed is contaminated with evenvery small amounts of water. In one case, theauthors were able to track a hydrogen related gearfailure to wat
32、er left over from flooding during asevere rain storm during the days preceding thetime the ingot/billet used to make the gear forgingwas produced.While hydrogen introduced during the basic steelproductionprocessistheusualsourceofhydrogen,it can also be introduced during the finishing and/orelectropl
33、ating operation as well. The latter processis well understood and has been clearlydemonstrated on large through hardened steelcomponents which require corrosion protection byplating materials such as cadmium. Hydrogenembrittlementfailuresareverycommon,e.g.,whenhigh strength steel such as 4340 in the
34、 220+ KSIstrength level was electroplated for corrosionprotectionpurposes. Asaresult,greatcaremustbeexercisedduringthedesignstagetoprecludethesetypes of failures.Hydrogen introduced by either the raw materialsteel melting or gear heat treatment processes isgenerallyresponsibleforthemostcatastrophict
35、ypeoffailureinwhichthelargegearisfracturedintotwopieces, as shown in Figure 5.Figure 5. Double helical pinion fractured dueto hydrogen burst6Other processes, such as the acid etch inspectionfrequently performed on hard finished toothsurfaces to insure that burning has not occurred,however,canalsoint
36、roducehydrogen,althoughthemechanism tends to be more localized. The spiralbevel pinion shown in Figure 6 failed due to a largecrack that was initiated in a micro-hydrogen burstthat initiated near the tooth surface as shown inFigure 7.Figure 6. Spiral bevel pinion failure due tohydrogen micro-burstFi
37、gure 7. Crack visible in figure 4 originatedin this hydrogen micro-burstNote the scale in Figure 7 which indicates that thespecific hydrogen burst from which the crackinitiated was only about 0.030 inch by 0.010 inch insize. The hydrogen that caused this local micro-burst was likely due to thelack o
38、fan adequatebakecycle after the acid etch inspection process wascompleted. Baking the part for several hours atabout 275 degrees F will cause any residual hydro-gen to leach from the tooth surfaces that had beenetch inspected.The residual internal tensile stressproblemDuring the quench operation tha
39、t is associated withall gear heat treatment processes several adverseconditions may occur. The two most obvious aredistortion and the occurrence of residual tensilestresses. Residual compressive stresses can alsooccur; however, the presence of such compressivestresses, particularly on the tooth surf
40、aces is abeneficial result and thus quite desirable. Con-versely,internalresidualtensilestressescaninitiatefailures that are similar to those initiated by ahydrogen burst in extent.Figure 8 shows a transverse cross-sectionmachined through a double helical pinion near theedge of the apex gap. A large
41、 crack, extendingnearly completely around the circumference in theapex gap between the LH and RH helical teeth wasfoundafterabout13monthsofserviceloading. Thecrack appeared quite suddenly with little warning.Figure 8. Cross-section through crackinitiated by internal residual tensile stressesAs part
42、of the failure investigation, several crosssectionswerecutalongitslengthandperpendicularto the centerline of the pinion. The part had a pitchdiameter of about forty eight inches and a length ofabout 16 feet, along the length of the pinion. Thissection revealed a series of cracks similar to thatshown
43、inFigure3. Eachofthedefectsinitiatednearthe center of the pinion shaft section and prog-ressedinagenerallyslowspiral,asindicatedbytheredarrowsinFigure8,fromthecentertotheoutsidediameter. Total catastrophic fracture of the pinionoccurred when the cracks came close to the toothroots.In the case of thi
44、s very large pinion, high internalresidual tensile stresses were believed formed dur-7ing the quench operation after carburization. Be-causeofthesizeofthepinion,alargethermalgradi-entexistedforanextendedperiodoftimeduringthequench operation. That is, the exterior of the partcooled much more quickly
45、than the interior of thepart,thusresultinginlockedinthermallyinducedre-sidual tensile stresses. Such stresses can be gen-erated due to poor quenching procedures, such asinadequate agitation during quench, and by prob-lemswiththe quenchmedia (oil,water, orpolymer),such as contamination, that limits e
46、ffective heattransfer during the quench.Specific Case StudyThe pinion shown in Figure 9 failed after approxi-mately 7 months of loaded service. The crackshownbytheredarrowsresultedinacompletefrac-ture of this doublehelical pinion. Unfortunately,thisoccurred during normal operation. No anomalousopera
47、tingconditionsor unusualloading wererepor-ted. The failure was sudden and occurred withoutanywarningsymptomsofanykind. Thesteelmillinwhich this pinion was operating performed well fortheentiretimeleadinguptothefailureandthensud-denly ceased running in a matter of seconds. Thepinion fracture resulted
48、 in a complete stoppage ofthe mill which engendered a large amount of collat-eral damage and financial loss.Figure 9. Fractured pinion still in gearbox(Operating time approximately 7 months)The pinion, which is about 84 inches in overalllength,wasmanufacturedfroman18NiCrMo5steelforging. The teeth we
49、re carburized to an effectivecase depth requirement of 0.160 - 0.172 inch. Theface width of the helical teeth is about 7 inches perhand and the apex gap is about 9 inches long. Thefracture occurred in the 19 inch diameter apex gapshaftsection. EachsetofhelicalteethAandBwereseparated by mating fracture surfaces A and B asshown in Figure 10. After removal from the gear-box, the double helical pinion was found to becompletely fractured into the two segments Figure10 shows.Figure 10. Fractured pinion segments fromfigure 9Figure 11 shows a close up view of the segment onthe left
