TitleColumn generation and network modeling in large-scale logistics networks
NameLiang, Zhe (author), Chaovalitwongse, Wanpracha Art (chair), Albin, Susan L. (internal member), Elsayed, Elsayed A. (internal member), Pham, Hoang (internal member), Resende, Mauricio G. C. (outside member), Rutgers University, Graduate School - New Brunswick,
SubjectIndustrial and Systems Engineering,
Air traffic control--Computer programs,
Routing (Computer network management)
DescriptionDecomposition has been used in solving numerous problems in mathematics, computer science, engineering, management, and operations research. In this dissertation, we use decomposition methods to solve three practical combinatorial optimization problems arising in telecommunication and airline planning. In the first part of the dissertation, we study a redundant multicast routing problem with group diverse constraint (RMRGD) that arises in many network applications such as communication systems, power supply distribution systems, transportation networks, etc. We propose three mixed integer programming (MIP) models, an edge-based, a path-based, and a tree-based model, to solve RMRGD. We proposed two decomposition methods based on the column generation and branch-and-price to solve the path-based and tree-based models. Our empirical results suggest that the edge-based model is superior in solving small and mid-sized problems, whereas the tree-based model performs better for large problems. In the second part of the dissertation, we study the flight conflict resolving problem (FCR). The purpose of flight conflict re-scheduling problem is to provide a flight schedule that minimizes the total penalty cost of schedule changes, while maintaining the FAA separation standard between aircrafts. We propose a set-partitioning-based flight sequence model (FSM) that selects an optimal set of flight sequences to minimize the total penalty cost. We also extend the FSM to consider equity among airlines because such corporate decision making (CDM)-feature is necessary and critical for the future aviation systems. The computation results show the proposed solution methods outperform other solution methods, and solve the real life test cases optimally within reasonable time In the third part of the dissertation, the aircraft maintenance routing problem is studied. The aircraft maintenance routing problem is aimed at scheduling the aircraft rotations so that adequate maintenance opportunities are provided to every aircraft in the fleet. In this dissertation, we present two new compact rotation-tour network representations for the daily aircraft maintenance routing problem (AMR) and the weekly aircraft maintenance routing problem (WAMR), and propose new mixed-integer linear programming formulations to solve these two problems. The computational study suggests the proposed models are able to solve large real-life test instances optimally in reasonable time.
NoteIncludes bibliographical references
Noteby Zhe Liang
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.