Integrated Optimization in Supply- and Demand-Driven Channels in Two-Stage Supply Chains

Open Access
Li, Xin
Graduate Program:
Industrial Relations and Human Resources
Doctor of Philosophy
Document Type:
Date of Defense:
May 05, 2017
Committee Members:
  • Jose Antonio Ventura, Dissertation Advisor
  • Jose Antonio Ventura, Committee Chair
  • Vittaldas V Prabhu, Committee Member
  • Uday V Shanbhag, Committee Member
  • Terry Paul Harrison, Outside Member
  • two-stage supply chain
  • integrated optimization
  • mathematical model
  • exact algorithm
  • heuristic
  • supplier selection and lot sizing
  • order acceptance and scheduling
  • source selection and supply planning
Essentially, supply chain management (SCM) is the coordination and integration of the flows of materials, information, and finances across the entire supply chain, which encompasses a wide range of important activities in a company’s operation, including supplier management, supply planning, production scheduling, inventory management, etc. Due to the systematic nature of SCM, many of its activities are highly interrelated and should be studied integrally to help the company gain maximum benefit. This dissertation aims at addressing the integrated optimization problems of some critical components in SCM, particularly, for supply- and demand-driven channels in two-stage supply chains. Four specific problems are incorporated in this dissertation. The first problem integrates the decision making process of supplier selection and lot sizing for a manufacturer who orders one type of product from multiple candidate suppliers to fulfill a fixed demand rate. Particularly, we consider the case where suppliers are capacitated and offer a certain type of quantity discount. The product’s perfect rate varies among suppliers and a minimum average perfect rate is required. Besides, in order to facilitate production plans, which are usually made on a daily/weekly/monthly basis, each placed orders is required to cover the demand corresponding to an interval that is multiple of a given time unit. A cyclic order schedule is employed, where the entire planning horizon is divided into repeating order cycles and the set of selected suppliers and corresponding order quantities and frequencies are determined accordingly so that the total cost per time unit is minimized. The second problem is similar to the first one but considers a different discount scheme, where the unit price of a product depends on the total order quantity of a given supplier over a certain period of time, instead of a discount applied to an individual order. Two cases are considered: order quantities are continuous/integer. A joint order acceptance and scheduling problem is studied as the third problem, which considers the scenario where a manufacturer receives multiple orders characterized by their revenue, processing time, due date, and tardiness penalty per time unit. The manufacturer can be represented as a single-machine system that adopts a make-to-order strategy. Due to the capacity limitation, the manufacturer cannot accept all the orders and needs to determine the optimal set of acceptable orders and the corresponding production schedule simultaneously such that the total profit is maximized. In the fourth problem, we investigate a real-world application of the integrated optimization of supplier (source) selection and supply planning in the shale gas and oil industry. We aim at assessing a novel fracking approach, called CO2 fracking, from an economic perspective, by considering the collection, supply, transportation, and storage of CO2. More specifically, a CO2 source selection and supply planning problem is defined and discussed from three aspects: CO2 collection and storage at sources, CO2 transportation, and CO2 storage and usage at well pads. In each of the problems included in this dissertation, real-world scenarios and factors are considered, and mathematical models are built to describe the problem. Due to the complexity of these models, existing software packages are either inefficient, difficult, or even impossible to use directly. To address this issue, properties of the models are analyzed. Based on the properties, we customize the solution procedures for each problem which either solve the problem optimally or produce near-optimal solutions. For the latter case, we provide the approach to evaluate the quality of the produced solutions. Numerical experiments are presented to illustrate the application the models and evaluate the performance of the proposed solution procedures.