Metaheuristics for homogeneous and heterogeneous machine utilization planning under reliability-centered maintenance

Abstract

Machine utilization is an important factor for the productivity of manufacturing systems. On the one hand, under-utilization would cause low productivity; on the other hand, according to the reliability theory, over-utilization would increase the failure rates of machines and consequently also decrease their utilization rates. In this paper, we present a problem of distributing a predefined workload to a set of available machines under reliability-centered maintenance, the aim of which is to optimize the overall machine utilization in terms of minimizing the total cost caused by maintenance and work delay. We consider two versions of the problem, one with homogeneous machines and the other with heterogeneous machines, both of which are formulated as integer programs. We adapt a set of well-known metaheuristics, including genetic algorithm (GA), particle swarm optimization (PSO), differential evolution (DE), biogeography-based optimization (BBO), and water wave optimization (WWO), to solve this problem. Computational experiments on a variety of problem instances from real-world manufacturers show that the adapted WWO algorithm exhibits the best performance on both versions.