Uncertainty quantification on the prediction of creep remaining useful life

Accurate prediction of remaining useful life (RUL) under creep conditions is crucial for the design and maintenance of industrial equipment operating at high temperatures. Traditional deterministic methods often overlook significant variability in experimental data, leading to unreliable predictions. This study introduces a probabilistic framework to address uncertainties in predicting creep rupture time. We utilize robust regression methods to minimize the influence of outliers and enhance model estimates. Sobol indices-based global sensitivity analysis identifies the most influential parameters, followed by Monte Carlo simulations to determine the probability distribution of the material's RUL. Model selection techniques, including the Akaike and Bayesian information criteria, ensure the optimal predictive model. This probabilistic approach allows for the delineation of safe operational limits with quantifiable confidence levels, thereby improving the reliability and safety of high-temperature applications. The framework's versatility also allows integration with various mathematical models, offering a comprehensive understanding of creep behavior.