Date of Award

3-10-2021

Publication Type

Doctoral Thesis

Degree Name

Ph.D.

Department

Mechanical, Automotive, and Materials Engineering

First Advisor

Ahmed Azab

Keywords

Cellular Manufacturing System, Dynamic Facility Layout Problem, Dynamic Programming, Metaheuristics, Multi-Objective Optimization, Robust Layout

Rights

info:eu-repo/semantics/embargoedAccess

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Abstract

The facility layout problem is one of the most classical yet influential problems in the planning of production systems. A well-designed layout minimizes the material handling costs (MHC), personnel flow distances, work in process, and improves the performance of these systems in terms of operating costs and time. Because of this importance, facility layout has a rich literature in industrial engineering and operations research. Facility layout problems (FLPs) are generally concerned with positioning a set of facilities to satisfy some criteria or objectives under certain constraints. Traditional FLPs try to put facilities with the high material flow as close as possible to minimize the MHC. In static facility layout problems (SFLP), the product demands and mixes are considered deterministic parameters with constant values. The material flow between facilities is fixed over the planning horizon. However, in today’s market, manufacturing systems are constantly facing changes in product demands and mixes. These changes make it necessary to change the layout from one period to the other to be adapted to the changes. Consequently, there is a need for dynamic approaches of FLP that aim to generate layouts with high adaptation concerning changes in product demand and mix. This thesis focuses on studying the layout problems, with an emphasis on the changing environment of manufacturing systems. Despite the fact that designing layouts within the dynamic environment context is more realistic, the SFLP is observed to have been remained worthy to be analyzed. Hence, a math-heuristic approach is developed to solve an SFLP. To this aim, first, the facilities are grouped into many possible vertical clusters, second, the best combination of the generated clusters to be in the final layout are selected by solving a linear programming model, and finally, the selected clusters are sequenced within the shop floor. Although the presented math-heuristic approach is effective in solving SFLP, applying approaches to cope with the changing manufacturing environment is required. One of the most well-known approaches to deal with the changing manufacturing environment is the dynamic facility layout problem (DFLP). DFLP suits reconfigurable manufacturing systems since their machinery and material handling devices are reconfigurable to encounter the new necessities for the variations of product mix and demand. In DFLP, the planning horizon is divided into some periods. The goal is to find a layout for each period to minimize the total MHC for all periods and the total rearrangement costs between the periods. Dynamic programming (DP) has been known as one of the effective methods to optimize DFLP. In the DP method, all the possible layouts for every single period are generated and given to DP as its state-space. However, by increasing the number of facilities, it is impossible to give all the possible layouts to DP and only a restricted number of layouts should be fed to DP. This leads to ignoring some layouts and losing the optimality; to deal with this difficulty, an improved DP approach is proposed. It uses a hybrid metaheuristic algorithm to select the initial layouts for DP that lead to the best solution of DP for DFLP. The proposed approach includes two phases. In the first phase, a large set of layouts are generated through a heuristic method. In the second phase, a genetic algorithm (GA) is applied to search for the best subset of layouts to be given to DP. DP, improved by starting with the most promising initial layouts, is applied to find the multi-period layout. Finally, a tabu search algorithm is utilized for further improvement of the solution obtained by improved DP. Computational experiments show that improved DP provides more efficient solutions than DP approaches in the literature. The improved DP can efficiently solve DFLP and find the best layout for each period considering both material handling and layout rearrangement costs. However, rearrangement costs may include some unpredictable costs concerning interruption in production or moving of facilities. Therefore, in some cases, managerial decisions tend to avoid any rearrangements. To this aim, a semi-robust approach is developed to optimize an FLP in a cellular manufacturing system (CMS). In this approach, the pick-up/drop-off (P/D) points of the cells are changed to adapt the layout with changes in product demand and mix. This approach suits more a cellular flexible manufacturing system or a conventional system. A multi-objective nonlinear mixed-integer programming model is proposed to simultaneously search for the optimum number of cells, optimum allocation of facilities to cells, optimum intra- and inter-cellular layout design, and the optimum locations of the P/D points of the cells in each period. A modified non-dominated sorting genetic algorithm (MNSGA-II) enhanced by an improved non-dominated sorting strategy and a modified dynamic crowding distance procedure is used to find Pareto-optimal solutions. The computational experiments are carried out to show the effectiveness of the proposed MNSGA-II against other popular metaheuristic algorithms.

Available for download on Thursday, March 10, 2022

Share

COinS