A robust supply chain design under demand uncertainty and hybrid postponement strategies: Issues in supply chain management

This dissertation covers two research areas in supply chain management. We propose a hybrid postponement strategy for time sensitive customers and a robust supply chain design under demand uncertainty. A hybrid postponement strategy differs from a total postponement strategy in that it allows the stock of both finished goods and generic products. Using a numerical analysis, we show that the hybrid postponement strategy is optimal under certain parameter settings. We compare the total operational cost of the optimal hybrid postponement strategy with that of the best total postponement strategy and find that the savings could be significant in some cases. We investigate the important factors that affect the percentage of products that should be stocked in generic product form. We prove that under certain conditions, it is never optimal to use hybrid postponement strategies for both products.; We also propose a framework for a robust supply chain design under demand uncertainty. Supply Chain (SC) design models emphasize strategic supply chain management decisions. These models determine the most cost-effective location of facilities (including plants and distribution centers), flow of goods throughout the supply chain, and assignment of customers to distribution centers. A robust supply chain design is important because supply chain design involves decisions at the strategic level, and hence it is desirable to keep the SC configuration unchanged over a relatively long period of time once it is determined. A robust supply chain design finds a SC configuration (or a group of SC configurations) that provide(s) robust and attractive performance under demand uncertainty. We provide several performance measures of “robustness” and several solution methods. In particular, we evaluate a two stage stochastic programming based heuristic using carefully designed experiments. The essence of the heuristic is to use samples of the demand scenarios in the stochastic programming model. We compare and evaluate two sampling methods: simple random sampling and uniform design method. The performance of the heuristic is tested against the optimal solution and the solution obtained by fixing the customer demand at the expected level. We conclude from the experiments that our heuristic works reasonably well even with small sample size if the right sampling technique is used. The uniform design method is superior to the random sampling method when sample size is small. We also provide guidelines for the management on when to use the deterministic supply chain design model.