Date of Award

Winter 12-2015

Embargo Period

8-30-2016

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Chemical Engineering

Advisor(s)

Ignacio E. Grossmann

Abstract

Supply chain models describe the activities carried out in the process industry. They are used to design and operate complex sequences of tasks that transform raw materials and deliver final products to markets. Many optimization models have been developed for supply chain planning because they offer the possibility of finding strategies that lead to greater economic benefits. The traditional models have focused on finding the optimal decisions of the supply chain planner in a deterministic context. However, it is widely recognized that uncertainty and external decisionmakers play a fundamental role in the economic success of industrial supply chains. This thesis proposes mathematical programming models for supply chain optimization that consider uncertainty and external decision-makers in a variety of industrial settings. Chapter 1 provides the motivation and the necessary background for our models. In Chapter 2, we study the design of resilient supply chains with risk of disruptions. Disruptions is a type of uncertainty that has not received much attention for supply chain planning, but it is known to have a significant effect in the performance of supply chains. We develop a stochastic programming model for supply chain design that includes disruptions at distribution centers, and a tailor-made solution method to address industrial instances of the problem. In Chapter 3, we present a novel cross-decomposition algorithm for investment planning under uncertainty. The algorithm integrates Benders and Lagrangean decomposition for two-stage stochastic programming formulations. Our computational experiments on instances of the resilient supply chain design problem show the superior performance of the cross-decomposition algorithm over Benders decomposition and direct solution with commercial MILP solvers. In Chapter 4, we propose a new approach for production planning and inventory management in process networks. Inventory management in these networks is a very challenging task because of the close interaction between production activities and the presence of diverse sources of uncertainty. Our planning strategy is based on implementing basestock policies iii to control production rates and inventory levels. Our results show the benefits of using a policybased approach for inventory planning in comparison to other stochastic programming approaches. In Chapter 5, we address the capacity planning problem with rational markets. Our model considers potential customers as rational decision-makers in a bilevel optimization formulation. We propose two reformulation techniques that transform the bilevel model into a single-level problem by replacing the lower level with constraints that guarantee its optimality. The reformulations are based on the Karush-Kuhn-Tucker conditions and the strong duality property of the lower-level linear program; the examples show better computational performance for the duality-based reformulation. The results also demonstrate the benefits of considering markets as rational decision-makers for capacity expansion planning, since it allows developing expansion plans according to the needs of the consumers. In Chapter 6, we extend the capacity planning model to include competitors that optimize their own capacity expansion plans. The resulting trilevel formulation considers as rational all decision-makers present in a competitive environment. We analyze the properties of the trilevel formulation and develop two algorithms to solve this challenging problem. The results reveal the complex interactions that take place in decision-making problems with multiple players and show the importance of considering them in the model. Finally, in Chapter 7 we present the conclusions of this thesis. We demonstrate that uncertainty and external decision-makers have significant impact in supply chain operations, and that our models can be used to anticipate their influence in supply chain performance. The application of these models for industrial supply chain planning has a remarkable potential to increase efficiency.

Share

COinS