*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Title: Large Scale Optimization Methods for Fleet Management in Long-Haul Transportation Networks
Advisor: Dr. Alan Erera
Committee Members:
Dr. Natashia Boland
Dr. Martin Savelsbergh
Dr. Alejandro Toriello
Dr. John-Paul Clarke
Date and time: Wednesday, November 1st, 3:00 PM.
Location: Groseclose Building, room 226A.
Abstract:
This thesis addresses modern challenges in transportation and logistics planning from both methodological and practical standpoints. First, we present a robust optimization model for facilitating empty repositioning of mobile resources in large-scale transportation networks, where demand is uncertain. We extend the "budget of uncertainty" idea to realistically represent empty repositioning settings, and propose a rolling horizon framework featuring this model. In the second part of this thesis, we introduce a novel integrated modeling approach for introducing alternative fuel trucks (AFTs) into long-haul diesel fleets, while taking into consideration strategic and operational aspects of the transportation network. We then provide exact and heuristic solution methodologies for finding optimal or good fleet replacement strategies using the integrated model. Finally, in the third study of this thesis, we provide a method for computing dual bounds for multistage stochastic mixed integer programming problems with a finite number of scenarios. These dual bounds, called “partition bounds”, are based on solving group subproblems on partitions of the scenario set. We develop a scenario set partition sampling method to obtain effective bounds, and demonstrate the power of this method against another sampling-based approach on a wide range of test instances.
