Bayesian uncertainty quantification and propagation for validation of a microstructure sensitive model for prediction of fatigue crack initiation

Abstract

A microstructure and deformation mechanism based fatigue crack initiation and life prediction model, which links microstructure variability of a polycrystalline material to the scatter in fatigue life, is validated using an uncertainty quantification and propagation framework. First, global sensitivity analysis (GSA) is used to identify the set of most influential parameters in the fatigue life prediction model. Following GSA, the posterior distributions of all influential parameters are calculated using a Bayesian inference framework, which is built based on a Markov chain Monte Carlo (MCMC) algorithm. The quantified uncertainties thus obtained, are propagated through the model using Monte Carlo sampling technique to make robust predictions of fatigue life. The model is validated by comparing the predictions to experimental fatigue life data.