System Reliability Optimization Considering Uncertain Future Operating Conditions and Usage Stresses

This paper develops a new system reliability design optimization model based on a series-parallel system, where each subsystem is composed of components that are chosen to optimize the system reliability, while considering uncertainty of future usage conditions. In previous research, component reliability is generally assumed to be known with certainty because the usage and operating stresses are either not changing or they are known with certainty. However, this paper proposes a more general perspective in analyzing the uncertainty of actual system usage and operating stresses. In industry, predictions of component reliability often have inaccuracy or uncertainty due to unplanned variation, or changing environments and operating stresses. Sensitivity of particular component reliability occurs due to a relative increase/decrease of operating forces or stresses. A riskneutral design is considered for the system reliability optimization model with a probability associated with defined possible future usage conditions. The optimal redundancy allocation problem for each subsystem is composed of multiple choices of components with system-level constraints. The system is designed to maximize expected system reliability while considering the uncertainty of future component usage conditions and stresses. Nonlinear integer programming and a neighborhood search heuristic was used to solve this new problem formulation.