Variable amplitude loading in the very high-cycle fatigue regime S. E. STANZL-TSCHEGG 1 , H. MAYER 1 and A. STICH 2 1 Institute of Meteorology and Physics, University of Agricultural Sciences, Tu Èrkenschanzstraûe 18, Vienna, Austria, 2 and AUDI AG, D-85045 Ingolstadt, Germany Received in final form 11 April 2002 ABSTRACT Ultrasonicfatiguetestingisappropriatetoperformrandomloadingtestsintheregime ofveryhighnumbersofcycles.Ithasbeenshownthatneitheranendurancelimitnora thresholdstressintensityexistsunderloadingwithrandomlyvaryingamplitudesevenfor materials that do show these limits under constant amplitude loading conditions. The technicalfeaturesoftheultrasonictestingtechniqueinordertoperformrandomfatigue tests are shortly described. Endurance tests were performed on smooth specimens of AlSi7Mg(A356.0)aluminiumalloyandonnotchedAISI4142andC45steelspecimens. ThepreviousstudiesofcrackpropagationandthresholdbehaviouronAISI420ferritic chromiumsteelandGGG100-Bcastironareincluded.Experimentalresultsonlifetime andfatiguecrackgrowthmeasurementsunderrandomlyvaryingamplitudes,aswellas lifetimepredictions,basedonconstantamplitudemeasurementsanddamageaccumula- tioncalculationsarereported. Keywords fatigue limit; lifetime prediction; random fatigue; threshold; ultrasonic fatigue;variableamplitudeloading. NOMENCLATURE a  cracklength E  Young'smodulus e max  Maximumstrainamplitudeofacycle K max  Maximumvalueofthestress-intensityfactorofacycle W  Specimenwidth Y v (a/W)  Correction function for calculation of stress-intensity factor in velocity- controlledultrasonictests INTRODUCTION Cyclic loading of machine parts or other moving tech- nicaldevicesusuallydoesnottakeplaceunderconstant, butrandomlyvaryingamplitudes.Experimentaltesting, however, mostly is performed at constant amplitudes andS±Ncurvesorda/dN vs. DK-curvesaredetermined. Lifetime predictions for varying amplitude loading are based on these measurements by assuming some kind of damage accumulation. For variable amplitude loading experiments, exerting-defined random-loading sequences,whichapproximatetheactualloadingcondi- tionsbest,simulatein-serviceloadingconditions.Prob- lems, however, arise for applications with very high number of cycles, as in automobile components, where lifetimesofbetween10 8 and10 9 cyclesareexpected,or for railway components or offshore structures, where even10 10 cycleshavetobeendured.Economicconsider- ationsmakeitimpossibletouseconventionalhydraulic machinesforsystematicstudieswithsuchhighnumbers of cycles. Methods, therefore, have been developed in ordertoreducethetestingtimes±thatis,smallampli- tudesareomittedorthesequencesaremade`harder'in fatigue experiments. The question, however, remained: how safe lifetime predictions for actual loading are, whicharebasedonsuchchangedloadingconditions. 1,2 Fromascientificpointofview,especiallythehigh-cycle regime of lifetime curve (smooth specimen) and da/dN vs. DK-curve(notchedspecimen)ismostinteresting.Itis usuallyunknownoratleastnotwelldocumented,ifatall oratwhichnumberofcycles(Fig.1)afatiguelimitora threshold value exists under variable amplitude loading ß 2002 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 25, 887±896 887 Correspondence: S. E. Stanzl-Tschegg, Institute of Meteorology and Physics, University of Agricultural Sciences, Türkenschanzstraße 18, Vienna, Austria. E-mail: stefanie.tschegg@mail.boku.ac.at