Text

theses

We study a supply chain operations scheduling problem subject to both renewable and non-renewable resources. After an operation has been completed, the non-renewable resource is consumed whereas the renewable resource may be resumed for the next operation. Of both the renewable and the non-renewable resources, limited amounts are available and they need to be delivered to the locations where they are needed. The operations have deadlines, and the availability of the renewable resources depends on the sequence of the operations. Our problem is to find a coordinated operations schedule for the non-renewable and renewable resources so that the total tardiness across all the customers in the given network is minimized. Part 1 of this dissertation presents an overview of existing solution methodologies for integrated supply chain operations scheduling/planning problems involving production, inventory, distribution, and routing. We take into account problems dealing with operational decisions and classify them according to their characteristics, such as time constraints and routing decisions that are directly related to our research problem. Various methodologies are presented and discussed, and their possible integrations, combinations, and extensions are discussed. In Part 2 of the dissertation, we build a mixed integer-programming model, present a complexity classification for our problem, and show where the borderline lies between NP-Hardness and polynomial time solvability. We analyze the structural properties of our problem, provide strongly polynomial-time solutions for several special cases that have practical applications, and identify the cases that are computationally intractable. Finally, we propose a framework of heuristic procedures for solving more general versions of this problem. In Part 3 of this study, we introduce a mathematical programming based rolling horizon heuristic that is able to locate near optimal solutions within ten-minute of CPU time for networks up to 80 customer service operations. This heuristic solves the problem through an iterative process. In each iteration, a subset of customers and a subset of batches of non-renewable resources, together with the travel teams (renewable resources), are scheduled by solving a respective optimization problem of a much smaller size. Through an extensive empirical study with over 5,000 randomly generated test cases under various parameters, the empirical error gaps of this proposed solution approach, when compared to the best solution obtained by a commercial optimizer within one-hour of CPU time, are constantly within 5%. This work can be extended in several directions. One of them is to conduct a thorough simulation study to assess the impact of management policies on the effectiveness of emergency logistics involving bottleneck renewable and non-renewable resources. Another one is to design and evaluate meta-heuristics for solving a more general version of our problem.

ETD_5229

Ph.D.

Includes bibliographical references

Includes vita

by Shengbin Wang

doi:10.7282/T3KP808W

ETD doctoral

The author owns the copyright to this work.