Flexible facility design with stochastic data in multiple periods

Abstract

Facility planning is used to design a new facility layout or redesign an existing facility layout. A wide variety of research has been done in the area of facility planning and design. However, most of this research has been done with respect to static facility layout design assuming deterministic data. Although these are important contributions to solving the facility layout problem most real facility layout problems are ongoing, dynamic problems. Due to the dynamic nature of the problem, usually only a portion of the future demand data is known during the design phase. Therefore, it is more realistic to represent and evaluate this problem with a formulation that simultaneously takes' into account the dynamic and stochastic nature of the problem. This thesis develops and models a two-phase heuristic solution procedure that considers both the stochastic and dynamic nature of the facility layout problem. This procedure minimizes total costs while utilizes layouts that have been predetermined to be robust in hopes of developing a robust dynamic layout plan. The expectation of this robust dynamic layout plan is that it will be better able to cope with the dynamic environment of real world applications, which cannot always be fully defined during the facility planning phase. The two-phase heuristic is compared with a modified version of the expected demand formulation (Palekar et al. 1992). The expected demand approach, by design, perforns well for expected demand scenarios. However, this approach has potential to perform poorly for the worst case scenarios. Conversely, the new two-phase heuristic attempts to improve the worst case scenarios while maintaining good expected demand performance. In a situation where there is potential to encounter costly worst case scenarios, it may be better to use an approach like the two-phase heuristic. This thesis provides examples that show instances where this two-phase heuristic might be applied.