Optimal switching policy for warm standby systems subjected to standby failure mode

Abstract

Standby is used extensively in mission-critical and safety-critical systems to improve reliability and availability. The dormant period during standby may introduce additional failure modes to the standby components. This is commonly observed in many real systems, yet it has been overlooked in existing research. This study is motivated by a two-motor standby system used in a power plant, in which periodic switching between the two motors is used to mitigate standby failure. We propose a generic system reliability model that captures both the normal aging and standby failures. The long-run average cost of the system can be derived using the technique of semi-regenerative processes. Thereafter, the problem of the optimal switching policy is formulated with the objective of determining the optimal switching period that minimizes the long-run average cost. We further consider a special case where the component failures under normal-use conditions follow a Poisson process and the repair times are exponentially distributed. A numerical study is conducted to demonstrate the proposed methodologies.