Solving the design of distributed layout problem using forecast windows: A hybrid algorithm approach

Abstract

In today's competitive environment, manufacturing facilities have to be more responsive to the frequent changes in product mix and demand by realigning their organizational structure for minimizing material handling cost. However, manufacturing firms are reluctant to modify the layout as it leads to operation disruption and excess rearrangement cost. In this paper, we present an alternative approach for designing a multi-period layout (i.e., distributed layout) that maintains a tradeoff between re-layout cost and cost of excess material handling. Obtaining an optimal solution to distributed layout problem is generally a difficult task, owing to larger size of quadratic assignment problem. In order to overcome the aforementioned drawback, a meta-heuristic, named 'CSO-DLP' (Clonal Symbiotic Operated-Distributed Layout Planning) is developed for designing a distributed layout that jointly determines the arrangement of department and flow allocation among them. It inherits its trait from Symbiotic algorithm and Clonal algorithm. In addition to these; the concept of 'forecast window' is used, which evaluates the layout for varying number of periods at a given time. The proposed meta-heuristic is applied on a benchmark dataset and the effect of system parameters, such as rearrangement cost, department disintegration, and duplication are investigated and benchmarked in this paper. Highlights? An alternative approach for designing a multi-period layout (i.e., distributed layout) that maintains a tradeoff between re-layout cost and cost of excess material handling. ? A meta-heuristic, named 'CSO-DLP' (is developed for designing a distributed layout that jointly determines the arrangement of department and flow allocation among them. ? The proposed meta-heuristic is applied on a benchmark dataset and the effect of system parameters, such as rearrangement cost, department disintegration, and duplication are investigated and benchmarked in this paper.