Modified cluster boundary search technique for improved layout designs

Published techniques for Facility Layout Problem (FLP) can be broadly classified as either constructive placement or iterative improvement techniques. For the former category, the quality of solution is dependent upon the ordering of facilities for placing them one at a time, while for the latter category the quality of solution depends on the initial design used for iterative improvement. This paper presents a constructive placement technique, based on modified cluster boundary search algorithm, to optimize the layout of unequal-area facilities on a continual planar site using a multiple firing order mechanism along with an efficient search procedure to minimize the impact of ordering. Since the dimensions of the facilities are modeled in the mathematical formulation of the problem, there is no overlapping of facilities at any stage of the optimization process. The effectiveness of the presented technique is determined in terms of the quality of optimal solution and its computation cost. Test results are presented for benchmark problems of VIP-PLANOPT, a well-known commercially available software package for obtaining optimal layout designs. It is shown that the technique generates high-quality layout designs with minimal computation cost, especially for large-size problems. For the benchmark problem involving 100 unequal-area facilities, the presented technique achieved a layout design with nearly 10% less cost as compared to that obtained by VIP-PLANOPT. Results are also presented for layout designs with specified upper bound on white space (area) in the final layout.