Evidence-based framework for real-time life-cycle management of fatigue-critical details of structures

Abstract

Ship structures are subjected to repetitive wave loads that can lead to fatigue damage. Accumulation of fatigue damage may eventually result in structural failure. In order to mitigate fatigue hazard to ship structures, timely inspections and maintenance actions must be planned during their life-span. As delays in the process of decision-making could result in the increase of the risk of failure, decision-supportive frameworks working in a real-time fashion are extremely important. In this paper, a life-cycle management framework for fatigue-critical details is proposed on the basis of Dynamic Bayesian Network (DBN) and Influence Diagram (ID). The proposed framework, established based on a stochastic fatigue crack growth model, leverages exact inference and discretisation of continuous random variables to enable real-time Bayesian updating and utility-based decision-making. Compared with previous life-cycle management studies, the new framework is capable of making decisions regarding different inspection techniques and appropriate repair actions in a holistic manner. The relatively simple structure of DBN and ID also makes the proposed framework especially attractive to stakeholders and practitioners, who may have limited knowledge of statistics, Bayesian updating and utility theories.