A Bayesian approach for the quantification of strength model uncertainty factor in ultimate limit state

This paper presents a method for quantifying the model uncertainty factor X_r associated with the hull girder ultimate strength prediction of ship structures. As real-scale observations from hull collapses are not available, the quantification of X_r usually involves the combination of engineering judgment along with the consideration of data from advanced numerical models, such as non-linear finite element analysis (NLFEA). This fact motivates the use of a Bayesian approach for the determination of X_r . In this study, Bayesian statistical inference is employed for estimating the parameters that characterize the probabilistic model of X_r . The effect of various NLFEA samples, non-informative priors, and different levels of error in the NLFEA results is examined. Finally, a distribution is proposed to model X_r which can be used both on the development of design criteria for double hull oil tankers in ultimate limit state and the reliability assessment of existing vessels as well.