Reliability optimization of dynamic k-out-of-n systems with competing failure modes

摘要

Designing a reliable system under different working conditions is a challenging task for system designers. This paper introduces different approaches to cope with the uncertainty of working conditions in redundancy allocation problems. A novel dynamic k-out-of-n system is developed, in which possible working conditions are modelled using discrete scenarios. The proposed dynamic strategy takes advantage of changing the value of k as a response to the different working condition. The decision making process consists of two stages: in the first stage, when the working condition is uncertain, the total number and the type of the redundant components are determined. In the second stage, after realizing the working condition, the decision about the value of k is made. The proposed strategy is capable of modelling the systems consisting of components with two competing failure modes: open and short mode. To make the model more realistic, the possibility of mixing the redundancy choices as well as the effect of each failure mode at the system level are considered. The model is solved using a genetic algorithm for a fire detection system. It is demonstrated that the model outperforms the classic approaches and provides a higher level of system reliability without imposing extra cost on the system.