Multi-component inventory policies in assembly systems with demand and supply uncertainties

Abstract

This dissertation considers multi-component inventory problems in an assembly system with demand and supply uncertainties. Since we focus on necessary components for the single-type final product, the inventory levels of all components are synchronized by the demand of the final product. Moreover, fixed ordering cost and inventory holding cost factors are different for each component. With considering the inter-dependency and characteristics of component inventory, we solve the three problems defined with different assumptions about demand and supply uncertainties. At first, we study multi-component inventory policies for an assembly system with stochastic demand. The objective of the problem is to determine the replenishment plan of components that minimizes inventory related costs. We prove optimal solution properties and develop an algorithm to find the optimal solution for a single-period problem. We also find some properties for a multi-period case and provide efficient heuristics to solve problems with a relatively large number of components. Secondly, we find an economic replenishment plan for components with random lead-times in an assembly system. Due to the uncertainty of replenishment lead-times, we should find the optimal order quantities and timings of components simultaneously to minimize the expected sum of total costs. We establish a stochastic programming model to find the optimal solution for a small sized problem, and study the cost function and solutions properties by using a simple model with two components in two periods. Based on the analysis, we suggest an algorithm to find heuristic solutions within a reasonable computation time. Finally, we consider multi-component inventory problem for an assembly system with both supply and demand uncertainties. To solve the problem, we study the structures of inventory control policies found in previous studies and suggest a new structured inventory policy that reflects the characteristics of component inventory, including stochastic replenishment lead-times of components and stochastic demand of the final product. We also provide a method to find the best parameters of the suggested inventory policy. We design a variety of experiments to evaluate the suggested inventory policies and the algorithms for finding solutions and policy parameters. We also generate test instances with various conditions to test the robustness of solutions. The inventory policies and algorithms proposed in this dissertation are also tested with large-sized experiments that reflect practical situations. In the results of computational experiments, the suggested algorithm and inventory policies of this dissertation show good performances under various conditions. Through this dissertation, we can provide managerial insights that can be applied to operate assembly systems efficiently in real situations.