Dynamic facility layout problem based on open queuing network theory

Determining the location of machines for given periods of time, depending on changes in the material flow between the machines, is known in the literature as a dynamic facility layout problem (DFLP). While most approaches focus on reducing handling and rearrangement costs, this article also considers the amount of work-in-process (WIP) for a particular class of problems. WIP results from the queuing phenomenon and depends on the availability of transportation units. To solve these types of problems, we suggest the use of an analytical approach which uses open queuing network theory and is based on a quadratic assignment problem formulation. Since a queuing model approximates the studied system, the accuracy of the results is evaluated through a comparison with simulation results. Considering both the NP-hardness and the multi-objective nature of the problem, a meta-heuristic optimization approach is proposed. It aims at determining the Pareto front using cloud-based multi-objective simulated annealing (C-MOSA). The performance of C-MOSA is evaluated against other published multi-objective approaches. Computational experiments are performed and the results show that C-MOSA is capable of obtaining efficient results to solve the multi-objective dynamic facility layout problem.