Redundancy allocation of heterogeneous components by simulation-based optimization

The present study proposes a novel simulation-based optimization model for a series-parallel redundancy allocation problem (RAP) under heterogeneous components for reliability maximization subjected to system-level constraints through the identification of the optimal redundancy strategy, component type, and subsystem component count. To obtain higher practicality, active, cold-standby, mixed, and K-mixed redundancy strategies are incorporated as the decision variables. In general, as it is difficult to determine a closed-form (excluding active redundancy strategies), it is not possible to analytically assess system reliability. Earlier studies on system reliability optimization applied convenient lower bound approximation. This limits higher reliability levels. In order to tackle this problem, this study adopted simulation sampling to make unbiased efficient reliability estimates. To this end, 4Dscript interpreting programming was utilized to develop a simulation model. As RAP has a combinatorial nature and is a random and NP-hard problem, this study adopted the genetic algorithm (GA) optimization. To validate the model and assess GA efficiency, the numerical findings of some benchmark tests were employed. The proposed approach outperformed earlier approaches in reliability and confidence.