Abstract:  The configuration problem of multi-echelon supply chain with limited production capacity is discussed. Based on the relationship of safety stock with respect to both the production capacity and net replenishment time, the supplier selection and the corresponding service times are optimized. A mixed integer programming model has been proposed for solving the supply chain configuration problem under uncertain market demands and limited production capacity, while an improved Genetic Algorithm with higher solution precision is adopted for problem solving. Two observations have been obtained from the results. First, a supply chain should avoid selecting those suppliers with smaller remaining capacities yet without advantageous prices and service times. Second, in case of highly fluctuated demands, a supply chain should allow broader lead time ranges for suppliers to achieve lower SCC（supply chain configuration）cost.

Key words:  capacity constraint, demand uncertainty, supply chain optimization configuration, improved genetic algorithm