1、93FTM9_UGear Tooth Bending Fatigue Crack Detection byAcoustic Emissions and Tooth ComplianceMeasurementsby: Jeffrey Wheitner, Donald Houser and Craig BlazakisOhio State UniversityAmerican Gear Manufacturers AssociationI ITECHNICAL PAPERGear Tooth Bending Fatigue Crack Detection by AcousticEmissions
2、and Tooth Compliance MeasurementsJeffrey Wheitner, Donald Houser and Craig Blazakis, Ohio State UniversityThe statementsandopinionscontained hereinarcthose oftheauthorandshouldnotbeconstruedasanofficial action oropinionof theAmericanGear ManufacancrsAssociation.ABSTRACT:Early detection of cracksin g
3、earteeth subjectedto standardbendingfatigue testshaslongbeen adiffic_ _L Thepurpose of this paper is to present the results of gear tooth bending fatigue tests in which the combination of acousticemissions measurements andtooth compliancemeasurementsare exploredas tools for early crack detection. Si
4、ngletooth bending fatigue tests for severaI different gear materials were performed on an eleclrohydraulic fatigue testingmachine using a standardSAE fatigue test fixture and geargeometry. Throughout testing, acoustic emissions weremonitored t/siflgan acoustic emissions lransduceraffixed to the surf
5、aceof the gear. Toothcompliance was measuredusing an accelerometer affixed to the lower platform of the fatigue test fixture. The combination of these two crackdetection methods servesto describe suchfatiguetestcharacteristicsas theprobable timeof crack ir6_a_on, rate of crackpropagation, and percen
6、t of total fatigue life spent in crack propagationphase. The effects of materials, metallurgicalprocesses suchas carbmSzationand shot peening, and the resultant surfacefinish on the fatigue process as revealed bythese crack detection methods axealso presented.Copyright 1993American Gear Manufacturer
7、sAssociation1500 King Street, Suite 201Alexandria, Virginia, 22314 :_October, 1993ISBN: 1-55589-621-9Gear Tooth Bend:rag Fatigue Crack Detection ByAcoustic Emissions and Tooth Compliance MeasurementsC. A. Blazakis, Product Design Center, Newark, OhioD. tL Houser, The Ohio State University, Dept. of
8、MechanicalEngineeringJ.A.Wheitner, CumminsEngine Company, Inc., Columbus, IndianaIntroduction methodsincludedyepenetrantinspection,magneticparticleinspection,eddycurrentinspection,ultrasonicinspection,andBecausegeartoothbendingfatiguefailurestendtobe radiography.moreseriousthananyothergearfailuremod
9、eAlban,1985,an Two methodsfordetectingandmonitoringthegrowthextensiveeffortisunderwayamong researcherstodevelop offatiguecracks,however, havebeenfoundto be veryaccuratemodels for the prediction of gear tooth bendingfatigue compatible with single tooth bending fatigue test conditions:life as well as
10、to generate experimental data for comparison acoustic emissions measurements and tooth compliancewith such models, measurements. Furthermore, the combination of these twoAs described by Collins Collin.%1981, “Fatiguemay be methods applied in parallel during testing has produced reliablecharacterized
11、 as a progressive failure phenomenon that characterization of the fatigue process for various gearproceeds by the initiation and propagation of cracks to an materials, surfacefinishes, etc.unstable size.“ This implies that fatigue life prediction models The remainder of this paper presents the resul
12、ts ofmust be designed to address both the crack initiation and the research aimed at exploring the use of these crack detectioncrack propagation regimes of the total fatigue process such that methods on gears of various materials, surface finishes, andmetallurgical processes.N-Ni+ NpTest Fixture and
13、 Gear Geometrywhere N is the total fatigue life, Ni is the number of cycles toinitiation,and Np is the number of cycles of propagation. In order to facilitate consistent and reproducibleThis entire fatigue process, including both the crack fatigue test results, the Society of Automotive Engineers (S
14、AE)initiation and crack propagation phases, can be simulatedin the has developed a standard fatigue test fixture, test gearlaboratory using an electrohydraulic testing machine fitted with geometry, and testing procedure to be used throughout thea fixture which allows a cyclic bending load to be appl
15、ied gearing industry Buenneke, et al, 1982. The standard SAEdirectly to a single gear tooth. Such testing, known as single test gear geometry has 34 teeth, a 20 pressure angle, atooth bending fatigue testing, allows the exploration of the diametralpitch of six, and an outer diameter of six inches. A
16、effects of materials, metallurgical processes, surface finish, detailed description of this test gear and fixturegeometry can begeometry, etc. on the fatigue process, found in the above reference. Throughout this research, theDue to the nature of this type of testing, however, the standard SAE geome
17、try and testing procedure were used.abilityto detect fatigue crack formation duringeach test can be Figures 1 and 2 illustrate the test fixture and the test geardifficult. Safety concerns prohibit the close-up visual inspection geometry.of the tooth fillet during the test. Also, the geometric Gear M
18、aterialsconstraintsimposed by the test fixture make conventional crackdetection methods difficult to apply during testing. Such Several materials were used for fatigue testing,1JTable 2. 4118 Root Surface Finishes#4118A Ra = 32 lainGroupCnoup#4118B Ra =42 lain qFGroup #4118C Ra = 64 lainall of the g
19、ears except the 9310 had lead crowning.Fatimae Testin_ ProcedureAll fatigue testing was performed on an MTS SystemsCorporation Model 810 fatigue testing machine using a 55 kiphydraulic actuator which was fined with the SAE fixture. AllFigure 1. SAE Standard Test Gear testing was performed using a si
20、nusoidal applied load with afrequency of 10Hz. A load ratio, R, wherel Lo_d_1 a l _Cell _ of 0.1 Was maintained for all tests. Each test was allowed torun continuously until finalfracture occurred, at which time themachine automatically shut down, or until the established run- - - out life was reach
21、ed. Due to time constraints, a run-out life of106 cycles was usually observed.V- I III “ II-“ I, ,I it Aco.stio mi sionsM rementsL-_ Pi_o, _ _ Acoustic emissions are small amplitude transient elasticActuatorn I stress waves resulting from the sudden release of energy during t_-Figure 2. SAE Standard
22、 Fatigue Test Fixture deformation and failure processes in stressed materials. The V,primary sources of acoustic emissions are crack growth andincluding: carburized 4118 alloy steel carburized 8620 alloy plastic deformation. Sudden movement at the source producessteel, carburized 9310 alloy steel, c
23、arburized and shot peened a stress wave which radiates to the surface of the structure andexcites a piezoelectric transducer.9310 alloy steel, and carburized 4320 alloy steel.The carburized 9310 gears were all manufactured and In order to monitor acoustic emissions within theheat treated together, w
24、ith four of these gears being gear during each fatigue test, a Physical Acoustics modelsubsequently shot peened to the specifications listed in Table 1. #NAN030 acoustic emissions sensor which is resonant at 300These gears are referred to as groups #9310 and #9310P, kHz is used. Prior to each fatigu
25、e test, the transducer isattached to the gear just below the root of the tooth to be testedrespectively. using ordinary super glue. Emissions are amplified using aPhysical Acoustics model #1220 preamplifier set at 40 dB gain,Table 1. 9310 Shot Peening Spees and are then monitored using a Physical Ac
26、oustics mode/#1200A crack detector. The crack detector is able to record a -_zcumulative total of the emission counts from the growing crack,Specification: MIL-S-13165which is output to a dual channel strip chart recorder. EachShot Size: MI-330-Hcount represents a single incidence of the acoustic em
27、issions -_Intensity: 12-16A signal crossing a pre-defined threshold. A detailed descriptionCoverage: 200% of acoustic emissions can be found in Metals HandbookPollock, 1989.Nine of the 4118 gears were manufactured by onecompany using three different cutting speeds, feeds, and tools, Gear Tooth Stiff
28、nessMeasurementsto give three different root surface finishes, listed in Table 2.These gears are referred to as groups #4118A, #4118B, and Tooth stiffness is a measure of the force developed in#4118C. Ra is the measured peak to valley surface roughness, the tooth per unit displacement. Because the i
29、ntroduction of aThe 4320 gears are referred to as group #4320, and the fatigue crack into the tooth fillet reduces tooth stit_ess,8620 gears as group #8620. Most of the gears tested were observations of changes in tooth stiffness taken during a fatiguedonated by companies so gear finishing was chara
30、cteristic of test serve as reliable indicators of the presence and size of sucheach particular companies manufacturing process. Therefore, a crack.2 fIn order to monitor changes in the stiffness of thesystem due to the propagation of a crack in the test tooth, an_ accelerometer is screw mounted to t
31、he base of the fixture. A 500 /PCB model #302B03 piezoelectric accelerometer with a _ 46o , Jscnitivitv of 299.9 mW_ along with a I(i_ler model _504E4 t_ 400. _ _ - . 350 tdual mode amplifier is used. Accelerometer output during a _ 300fatigue test can be used to calculate the inenance, I, of the _
32、250 igear/fixture system using the equation tu 200= 150 i /“_100 i / :l= F _ 60- 0 _ _ i50000 60000 70000 80000 90000where F is the load applied to the fixture, x is the fixture Cyclesdisplacement, and _c“is the acceleration of the fixture as Figure 3. Acoustic Emission Count vs. Fatigue Life for an
33、measured by the accelerometer. Wheitner Wheitner, 1992 8620 Gear Toothdemonstrated that based upon a lumped parameter model of thesystem, the dynamic effects of the fixture na_s have no effect been calibrated to show the corresponding number of cycles ason the fixture stiffness, and thus the fixture
34、 stiffness can be determined using the strip chart recorder speed and the testmeasured statically. The inertance can therefore be related to frequency.the stiffmess of the gear tooth, k, by the equation In this case, crack initiation appears to occur atapproximately 61000 cycles. It should be noted,
35、 however, thatbecause a physical definition of crack initiation has yet to bek = Ice2 = Fee2 defined, and because the acoustic emi._sions detection system has a finite sensitivity, the term “crack initiation“ simply refersto the time at which the first detectable emissions are recorded.where co= 20
36、7r rad/sec is the test frequency. This implies that It may be possible to further refine the acoustic emissionsdetection system in order to increase sensitivity. Suchrefinement may include adjustments to the natural frequency ofFee: the transducer, the threshold value, the amplifier gain, or the5_=k
37、 number and placement of transducers.Notice that prior to the initiation point, the number ofThus, for a constan.t applied load and a constant frequency, the recorded acoustic emissions counts is non-zero. This is themeasured accelerometer output is inversely proportional to the result of background
38、 noise which is generated by the testingsti_ess of the gear tooth. Based on this analysis, it was machine. Because it is difficult to separate the noise signal fromdetermined that the accelerometer output could be used as a the signal generated by the growing crack, the sensitivity of thedirect meas
39、ure of the change in stiffness of the gear tooth detection system must be maintained at levels which minimizeduring a fatigue test without actually ealcalating the tooth the amount of background interference.stiffness. In some cases the accelerometer output was recorded As seen in the figure, the em
40、issions slowly increase until,using a PC data acquisition system, while other times it was at approximately 87000 cycles, the rate of emissions increasesrecorded in parallel with the acoustic emissions output on a dual dramatically, leading to final fracture at 89000 cycles. Thischannel strip chart
41、recorder. This allows for easy comparison acoustic emissions footprint is typical of the 4118 and 4320of the data from both crack detection methods, gears as well.In both the peened and unpeened 9310 gears, however,the crack initiation mode was very different, as illustrated inFigure 4. As the gear
42、reached the critical point, itResults instantaneously cracked along the thickness. When thishappened, the gear let out an audible pop that sounded like theThe remainder of this paper describes the results that snapping of ones fingers. At the same time this pop was heard,were obtained using the prev
43、iously described equipment and the acoustic emissions rose very rapidly. After this abrupt jumpprocedures. Results are presented for a representative cross- in emissions, the carve flattens out during propagation, thensection of the materials which were tested, in addition to cases once again increa
44、ses very rapidly until failure.in which unusual behavior was observed. It is proposed that in the 9310 gears, a smalldiscontinuity is growing below the surface near the case-coretransition. After this crack reaches a critical length, it quicklyAcoustic Emissions ResuJt. propagates to the surface res
45、ulting in the sudden burst ofacoustic _emissions. This Wpe of failure is consistent withA typical acoustic emissions plot for an 8620 gear is failures that occur below the surface in earbufized steel in whichillustrated in Figure 3. The graph consists of strip chart the endurance limit of the core i
46、s much lower that therecorder output in which the y-axis represents acoustic endurance limit of the case. The endurance limit in directlyemissions counts, and the x-axis represents time, which has proportional to the hardness of the material in steels. This31400_o120 o _ _“ $_-1000 _ 9z2_o,_ 800 / _
47、 _:600 : -“ _ “14.1 , _ 86-u 400 ,_“_, ,; , E 200 = 5, , r_30 i , l14000 14200 14400 14600 14800 15000 7G.CyclesFigure 4. Acoustic EmissionCount vs. Fatigue Life for a 9310 CyclesFigure 6. System Stiffness for a 4118 Gear Tooth at 170 ksiGear Tooth Root Stresso_ -= -oI ._ m i-2000 0 E _/- 1,IGO =_ _
48、 1lO.2O -12oa _ _ G=“- m|_ _ _ .,ooou .,o -_ 0.IS _ UEII400=0.00 ,_. -_“ I “0 82 _(;000 I000 $0000 12000 14000 1l_000 18000 20000 SlflO0 S17000 S$7200 .m;ll7400 5_ _600 _ 171100 S 18000Cycles _ lad_ Le_l_-o- co=t= CyclesFigure 5. Acoustic Emissions and Crack Length _s. Fatigue Figure 7. System Stiff
49、ness for a 9310 Unpeened Gear ToothatLife for a 4118 Gear Tooth 262 ksi Root Stress Vacoustic emissions failure mode was consistently observed with technique to mean the time at which the first measurable changeallofthe9310gearstested, in stiffnessisrecorded.Acoustic emissions work very well for predicting the Figure 7 illustrates a similar stiffness plot for a 9310onset of a fatigue crack in 8620 steels. Figure 5 illustrates a unpeened gear. In this case the curve appears to be muchplot of the acoustic emission counts and the through-
