Bayesian method for estimating initial crack size distribution for probabilistic risk analysis of repairable structural details

Abstract

Abstract A Bayesian method for estimating the initial crack length distribution is proposed for probabilistic risk analysis (PRA) of repaired structural details. Non‐informative prior is updated using a likelihood function, which is constructed using inspection capability, structural geometry, and material properties. The likelihood function based on the equivalent initial flaw size method was used for the left tail to avoid calculating the small‐crack growth. The right tail was estimated based on the inspection capability and the maximum possible crack length. The proposed method alleviates the difficulties in selecting the initial crack‐length distribution parameters. As an application, the KT‐1 aircraft risk after the repair was analyzed. The PRA was conducted using the initial crack length distribution estimated via the proposed method. The result was compared with the conventional lognormal initial crack‐length distribution.