Designing Facility Layouts with Hard and Soft Constraints by Evolutionary Algorithm

Abstract

Facility layout exerts a significant effect on factory operations and performance. Layout problems and floorplanning problems have been studied in the past decades. These problems typically involve locating objects of various sizes in a space for a certain objective. This paper presents a novel layout problem formulation, called the constrained facility layout problem (CFLP), aiming to find the layout with minimal covering area and connection length. In addition, the CFLP includes a hard constraint on the spatial clearance and a soft constraint on the geometrically relative order. The movability of objects is further considered in the CFLP. The formulation of CFLP is pertinent to industrial cases. To solve the CFLP, we adopt the covariance matrix adaptation evolution strategy (CMAES), a powerful evolutionary algorithm on numerical optimization. The proposed CFLP formulation and CMAES are utilized to solve the real-world facility layout problems of CTCI Corporation, which is a world-leading engineering services provider. The results show the high capability and advantages of the proposed approach in producing satisfactory layouts within very competitive time cost, in comparison to the layouts generated by human experts.