Maintenance decision-making model for gas turbine engine components

Abstract

When designing gas turbine engine components, the inspection and maintenance (I&M) plan is prepared using the safe life. However, the I&M plan determined using safe life may be costly since all components are replaced at designated life. Therefore, it is important to make maintenance decisions considering the time-dependent deterioration process of gas turbine engine components for a cost-saving I&M plan. In this study, we proposed a maintenance decision-making model for gas turbine engine components based on a partially observed Markov decision process (POMDP). Using dynamic Bayesian networks, a decision-making model integrating a reliability analysis model, and a decision model for I&M planning was constructed. The signal amplitude data resulting from non-destructive inspection according to operation hour was used as partially observed data. The total cost obtained from the proposed model were compared with the results using a fixed I&M plan. The proposed model resulted in more cost-effectiveness I&M planning within affordable risk levels by considering the interaction between risk cost and I&M cost.