Improving configuration of complex production lines via simulation-based optimization

Abstract

Optimizing the configuration of a complex production line is an NP-hard problem in various machine settings. Solving real-life-size instances of this problem becomes a more common challenge because the current trend of reshoring induces multi-national firms to transfer manufacturing facilities from workforce-intensive to capital-intensive production environments which usually require re-configuration of the transferred manufacturing systems according to the availability of better machinery in a capital-intensive environment. This paper focuses on the problem of optimizing production line configuration and proposes several simulation-based optimization approaches based on myopic search, ant-colony, simulated annealing, and response-surface methodologies. We investigate the relative performances of these proposed algorithms on a real-life manufacturing system transfer case in automotive industry according to solution quality and computation-time metrics under different parameter scenarios. Thus, our numerical results may guide the decision makers in choosing a suitable solution approach for this problem depending on the problem size and time availability. Our results also illustrate that ant-colony optimization, a methodology not widely applied in simulation-based optimization, provides high-solution quality for this problem when matched-up with a myopic search to find a good initial solution.