Iterated great deluge for the dynamic facility layout problem

Abstract

This chapter introduces an iterated great deluge (IGD) heuristic for the dynamic facility layout problem (DFLP). This problem involves the arrangement of manufacturing facilities over time to minimize the sum of the material handling and rearrangement costs. The IGD heuristic combines a great deluge algorithm with a perturbation operator that helps escape from local optima. Our implementation of the IGD heuristic relies on two main steps: (i) we first generate a local optimum solution by running an extended great deluge (EGD) algorithm for N1 iterations; (ii) the second step consists in a loop that perturbs the best current solution, and restarts an improvement by running the EGD algorithm for N2 < N1 iterations. Numerical results for 48 test problems from previous research are reported and compared. The solutions found by our approach are very competitive and a great set of the obtained results are better than or are at par with the well-known best solutions. New solutions (i.e., solutions better than the best solutions existing until now in the DFLP literature) are obtained in 13 problems. The newly developed IGD heuristic is discussed in the context of the general iterated local search paradigm.