Variable neighbourhood search for redundancy allocation problems

A variable neighbourhood search (VNS) algorithm has been developed to solve the redundancy allocation problem (RAP). The VNS method is perfectly suited to those combinatorial problems with potential neighbourhood structures, as in the case of the RAP. The most studied configuration of the RAP is a series system of s-independent k-out-of-n:G subsystems, the so-called series–parallel system. The RAP is to select the optimal combination and redundancy levels of components to meet system-level constraints. Two types of objectives are considered in this study – system reliability maximization and system cost minimization. The VNS algorithm is tested on sets of benchmark problems and compared to the best heuristics in the literature such as tabu search, multiple weighted objective heuristic, ant colony optimization and genetic algorithm. Computational results show the advantages and benefits of VNS for solving both types of RAP while considering both solution quality and computational efficiency.