Stochastic Creep Analysis of Hastelloy XR based on MPC-Omega Method and the Extended Kalman Filter

Abstract

A stochastic creep model was developed by incorporating the extended Kalman filter with the MPC-Omega method. The process noise covariance matrix for the filtering algorithm was derived by applying a traditional approach of uncertainty analysis to the empirical relationship of the Omega model’s parameters with stress and temperature. The model was validated with constant stress creep data of Hastelloy XR in an air environment and at various temperatures by predicting the statistical distributions for strain at a given time and the remaining useful life using a Monte Carlo simulation. The mean of the predictions was within a factor of two of the experimental data in almost all cases. In addition, the predicted 95% confidence intervals could cover virtually all of the experimental data.