| In order to make effective operational decisions and retain competitive advantage in the marketplace, companies have begun to discard the traditional operational way of decision-making, where production and distribution functions were handled separately, and adopt a new approach for the analysis of integrated production and distribution systems. By using the integrated view of the entire production and distribution system, companies can make more sound operational decisions to improve their operational performance and provide better customer service.; This dissertation aims to design an integrated production and distribution (IPD) model that considers a multi-echelon structure consisting of one manufacturing plant, multiple distribution centers, and multiple customer zones. The objective of the model is to minimize the total cost of manufacturing, inventory, and transportation, while satisfying customer demands simultaneously. To achieve this objective, the model is first formulated as a 0-1 mixed integer problem with considerable decision variables and constraints, and then a hybrid genetic algorithm is proposed to obtain feasible and satisfactory solutions to the problem.; To apply the proposed hybrid genetic algorithm, the formulated problem is transformed to ensure the satisfaction of equality constraints during the genetic search. In particular, to enhance the search performance and cope with binary decision variables, the hybrid genetic algorithm incorporates three heuristic techniques: (1) a heuristic method to generate an initial population with elitism to ensure a better search result; (2) a repairing procedure to handle infeasible offspring and guarantee their feasibility, and (3) a local search heuristic to reduce the transportation cost and improve the results of genetic search.; The experimental results show that the proposed hybrid genetic algorithm is very promising. As compared with the branch-and-bound method, the computation time required by the proposed algorithm can be improved substantially while the average difference of the genetic search solutions varies only from 1.42 percent to 2.28 percent of the optimal solutions in the case studies. This suggests that our algorithm can be an efficient and effective approach for providing feasible and satisfactory solutions to large-scale integrated production and distribution systems. |
