Probabilistic Sensitivity Analysis in Minimum Fatigue Life Prediction of a Shot Peened Titanium Alloy

Incorporating surface-treatment induced residual stresses (RS) in life prediction has been hampered by the lack of a physics-based approach that accounts for the variability in the RS profiles. Hence, the fatigue variability behavior of the alpha+beta titanium alloy, Ti-6Al-2Sn-4Zr-6Mo, was studied under shot-peen surface conditions. The variability in the RS profile and its influence on the location and size of the failure initiation sites were characterized. This distribution of initiation sizes was coupled with the variability in micro-scale crack growth and RS profiles to probabilistically predict the minimum fatigue life. To use this approach within the ICME framework, the relative importance of the model parameters has to be quantified. Probabilistic sensitivities using the score function method were developed for the lifing analysis. This presentation will discuss the sensitivities with respect to the key statistical inputs in the probabilistic approach, including initiation size, crack growth behavior and RS profile.