Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Industrial and Manufacturing Systems Engineering

First Advisor

Pius J. Egbelu


In today's highly competitive market environment, only companies with a highly efficient supply chain management, which integrates all decisions in various levels of planning and operations, can survive. These decisions must be coordinated and under the same goal, which is to minimize the total systemwide costs of the firm while products are manufactured and distributed to end-customers or retailers. In this study, the focus is on a pull-based supply chain, customer demand driven, multiple products and multiple echelon distribution system consisting of m manufacturing centers, n distribution centers, and p retailers or customers. The objectives of this study can be categorized into two parts. The first objective is to present a general framework of the design and configuration of the supply chain network at strategic and tactical planning levels in a single-product and multi-product multi-echelon supply chain systems. The problems deal with determining the appropriate number, location, and size of each manufacturing facility and distribution center/warehouse that should be used within the logistics network. The second objective of the research is to present a methodology for using a pull-based supply chain system both for a single-product system and multi-product system at the operational planning level. The problems deal with determining which products customers will receive from each available manufacturing facility and distribution center, what production quantities of the products should be manufactured by a particular manufacturing facility, and what quantities of each product and ways of shipment should be used from manufacturing facilities to distribution centers and to customers. Based on the nature of these large-scale mixed integer programming problems, decomposition heuristic algorithms based on relationships between primal and dual decompositions are developed. The mathematical models and heuristic algorithms are then demonstrated and evaluated on several sets of randomly generated problems. Although the heuristic algorithms do not guarantee optimum solutions, their results of the test problems suggest that the heuristics are effective in solving fairly large problems with reasonable computational time. Furthermore, they produce superior performances as compared to the other techniques that are tested.



Digital Repository @ Iowa State University,

Copyright Owner

Pongchai Athikomrattanakul



Proquest ID


File Format


File Size

192 pages