Date of Graduation

12-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level

Graduate

Department

Industrial Engineering

Advisor/Mentor

Rainwater, Chase E.

Committee Member

Nachtmann, Heather L.

Second Committee Member

Pohl, Edward A.

Third Committee Member

Kent, John

Keywords

Constraint Programming; Genetic Algorithm; Heuristics; Hours Of Service Regulations; Scheduling; Transportation

Abstract

This research proposes novel solution techniques to optimize two real-world problems in the area of scheduling and transportation. We first consider a model for optimizing the operations of dredges. In this problem, scheduling and assignment decisions are integrated across a finite planning horizon. Additional constraints and problem elements explicitly considered include, but are not limited, to environmental work window restrictions, budget limitations, dredge operation rates and schedule-dependent dredge availability. Our approach makes use of Constraint Programming (CP) to obtain quality and robust solutions within an amount of time small enough to be useful to practitioners. The expanded feature set of the methodology presented makes our solution tool the most comprehensive and flexible decision-making framework for dredge scheduling in existence.

The second transportation and logistics problem considered in this dissertation considers a unified variation of the Vehicle Routing Problem (VRP). This work offers a powerful yet flexible tool to model and solve real-world problems, each with their specifications, constraints, and requirements. We review existing VRP problems from the literature and propose new VRP variants that differ from the existing ones by the consideration of hours of service regulation on the active and drive hours of drivers in a single or multiple shifts. Real-world instances of these problems consist of thousands of customer locations and hundreds of vehicles. To ensure the quality of the solutions, we compare the performance of our approach with CPLEX on several benchmark instances from the literature.

Finally, the third chapter of this work focuses on a comprehensive analysis of the methodology presented in Chapter 4. Specifically, sensitivity analysis regarding the parameters driving the performance of the heuristics is performed. Also, we propose a Genetic Algorithm (GA) to solve the VRP variants in Chapter 3 and provide a computational study of its performance against CPLEX and the approaches in Chapter 3.

Download

Share

COinS