Decision Sciences

Abstract

This thesis studies the decision making problem in the supply chain management. We first studies optimal capacity expansion for a risk-averse decision maker over finite horizon. The capacity can be adjusted in bilateral ways: It can either be purchased or salvaged. Purchasing capacity incurs both fixed purchasing cost per order and variable cost per unit, similarly, salvaging capacity brings variable revenue per unit and incurs fixed salvaging cost per order as well. The capacity is then used in production to satisfy period demand, and the demand that exceeds the capacity is satisfied from an expediting supplier. We first consider the case when the capacity is carried over to the next period without deterioration incurring maintenance cost. We propose a framework to incorporate risk preference in multi-period capacity expansion models. Specifically, consumption model is employed to capture the risk sensitivity. We show that capacity decision and consumption decision in each period can be separately determined. Besides, we characterize the structure of the optimal capacity strategy and compare the optimal capacity strategy for a risk-averse decision maker with the optimal capacity strategy for a risk-neutral decision maker. We also extend our results to the cases when capacity deteriorates with both deterministic deterioration rate and stochastic deterioration rate from one period to the next period, or when the demand and financial parameters are world-driven, or when the manager can satisfy demand strategically, i.e., to determine whether or not to satisfy all demand and how many to be satisfied. Numerical tests are presented to study the impact of financial parameters and demand characterizers on the optimal capacity strategy. According to our numerical examples, the optimal capacity strategy can be easily characterized by (b,B, s, S) under certain conditions, whereby the capacity is purchased up-to B when the initial capacity is below b, is salvaged down-to S when it is above s, and stays put otherwise.

As inventory is an important component of capacity expansion, I have included in this thesis a detailed chapter on stochastic inventory in the context of assemble to order environment. In particular, I present the analysis and optimality results regarding the joint decision of the inventory replenishment and common component allocation rule for an assemble-to-order N-system.