Solvingthedynamicplantlayoutproblemusinganewhybridmeta-heuristicalgorithm.docVIP

Solving the dynamic plant layout problem using a new hybrid meta-heuristic algorithm Jose M. Rodriguez* Department of Mechanical Engineering, University of New Brunswick, NB, Canada E-mail: jm.rodriguez@unb.ca*Corresponding author F. Chris MacPhee Faculty of Computer Science, University of New Brunswick, NB, Canada E-mail: chris.macphee@unb.ca David J. Bonham Department of Mechanical Engineering, University of New Brunswick, NB, Canada E-mail: bonham@unb.ca Virendrakumar C. Bhavsar Faculty of Computer Science, University of New Brunswick, NB, Canada E-mail: bhavsar@unb.ca Abstract: The quadratic assignment problem (QAP), in the context of manufacturing, can be used to formulate the facility layout problem (FLP) that arranges facilities within a factory depending on material flow between the facilities. The dynamic facility (plant) layout problem (DPLP) extends the FLP to allow for rearrangements of the facilities due to changes in product mix and demand over a given planning horizon. This flexibility over time is an integral part of the agile manufacturing philosophy. Furthermore, the design of agile plant layouts requires finding all optimal permutation sets for FLP/DPLP instances. We compare our algorithm with 21 other algorithms for the DPLP and obtain better solutions than those known to date. Of the attempted 48 DPLP instances, 29 now have new best-known solution. With respect to the 6 department instances that all have a known optimal solution found using dynamic programming, optimal solutions were obtained for 2, which no heuristic had found before, and previously found optimal solutions were matched for the remainder of the instances. With respect to the 15 and 30 department instances, none of which have a known optimal solution, only 3 current best-known solutions could not be obtained. Keywords: quadratic assignment problem; facility layout problem; dynamic plant layout problem; genetic algorithm; tabu search. Reference to this paper should be m