Simulation based optimization for supply chain configuration design

Supply chain management promotes the integration of organizational functions of production systems ranging from ordering of raw materials throughout manufacturing processes, to distribution and delivery of products to customers. Its applications demonstrate that this idea enables organizations to achieve higher quality products, better customer service, and lower overall cost.; Supply Chain Configuration Design is the first issue managers face when establishing a new supply chain or re-designing of existing ones. Several challenges associated with Supply Chain Configuration Design could be listed: (1) relationships among companies in supply chains are complex, dynamic and uncertain, (2) supply chain networks in practice are both broad and deep, and (3) the decision-making process involves many decision variables most of which are qualitative and policy in nature. With a focus on Supply Chain Configuration Design this dissertation has developed a new approach to solve real-world problems. Applying the approach, we were able to address several strategic problems simultaneously: location, allocation, supplier selection, production planning and distribution problems.; This dissertation constructed a user-friendly, graphical tool for modeling supply chain configurations so that logic properties of supply chain processes can be analyzed and verified; users are able to precisely describe parameters and relationships among the supply chain's processes, to obtain better understanding of decision making issues in complex systems, to evaluate particular policies and to predict the outcomes.; The core of this research concentrated on the development of a hybrid optimization approach that combined Simulation, Mixed Integer Programming and Genetic Algorithm for Supply Chain Configuration Design. GA provides a mechanism to optimize qualitative and policy variables. MIP reduces computing efforts by manipulating quantitative variables. Simulation is used to evaluate performances of each configuration with non-linear, complex relationships and under more realistic assumptions. The approach was designed to be general and robust so that it could be applied to any supply chain.; A case study, performance evaluation and time complexity analysis for the new approach were implemented. The proposed approach was found empirically to outperform the random sampling approach and the pure Genetic Algorithm.