Basedonloadseparationtheory,theloadseparationparameterSpbmethodisaneffectiveapproachforestimatingtheJ-resistancecurvefromrecordsofloadversusdisplacementdirectly,usingonesharpcrackedspecimenandanadditionalreferencebluntcrackedspecimen.However,theeffectofthereferencebluntcrackedspecimenonJ-resistancedeterminationwasnotexplicitlyconsideredinpastwork.Inthispaper,amodifiedloadseparationparameterSpbmethodwasdevelopedtoeliminatethiseffect,andthenauniqueestimationofinstantaneouscracklengthforonesharpcrackedspecimencouldbeobtained.Furthermore,aforcedbluntingcalibrationmethodwasalsoadoptedtodeterminetheinstantaneouscracklengthintheloadinseparableregion,referringtoanormalizationmethod.ExperimentsonsteamturbinerotatorsteelCr2Ni2MoVwerecarriedouttoestimateJ-resistancecurvesusinganunloadingcompliancemethod.Byremovingunloadandreloaddatafromload-displacementrecords,theJ-resistancecurveforthesamesharpcrackedspecimenwasestimatedusingthemodifiedseparationparameterSpbmethod.TheresultsindicatethatthemodifiedSpbmethodcompletelyeliminatestheeffectofthereferencebluntcrackedspecimenontheinstantaneouscracklengthdeterminationofthesharpcrackedspecimen.However,differentJ-resistancecurvesinasmallrangeofcrackextensionarepresentwhendifferentbluntingcoefficientsareusedinthebluntinglineequation.TheJ-resistancecurveobtainedfromthemodifiedSpbmethodagreeswellwiththatobtainedfromthecompliancemethod.
