Model selection, updating and prediction of fatigue crack propagation using nested sampling algorithm

Mathematical models are often used to interpret experimental data, estimate the parameters and then predictions can be made. In practice, and in several applications, it is common that often more than one model could be used to describe the dynamics of a given phenomenon. Modelling and prediction of fatigue crack growth (FCG) is one of the engineering problems where a number of models with different levels of complexities exist and the selection of the most suitable one is always a challenging task. In this study, model selection, updating and prediction of fatigue crack propagation is carried out under a Bayesian framework. The nested sampling algorithm is selected to estimate the evidence of each competing model using an experimental data set of Aluminum 2024-T3. The obtained results are very encouraging and show the efficiency of the proposed approach when dealing with model selection, updating and prediction issues