Enhanced simulation method with multi-channel importance sampling for optimizing maintenance strategies against fatigue crack growth

This study presents an enhanced simulation framework available to derive an optimal inspection strategy against fatigue crack growth under a history-independent strategy, in which an importance sampling method based on a probability measure transformation via the Girsanov theorem is applied to a stochastic fatigue crack growth model involving random fields. The most advantageous point of the proposed method is a multi-channel history-dependent strategy for selecting the most effective probability measure for realizing the dynamic importance sampling, i.e., multi-channel history consisting of scenarios such as pre-inspection failure, post-inspection failure with non-detection or replacement after detection. Numerical examples demonstrate that the proposed method achieves approximately 98% reduction of computing time compared with the so-called crude Monte Carlo simulation, whereas a simple Monte Carlo method can not give an estimation for the optimal inspection strategy because of its insufficient ability for extremely small probability of failure.