Date of Degree
PhD (Doctor of Philosophy)
Timothy J. Lowe
Barrett W. Thomas
The hub location problem seeks to find the best location for hubs and the assignment of non-hub nodes to hubs. The resulting structure is a hub-and-spoke network. In this dissertation, we study the hub location problem within the context of three different application areas.
We first consider the air travel industry, where many airline companies (especially the legacy airlines) operate hub-and-spoke networks. Important considerations in designing such a hub-and-spoke network are the cost of establishing the hubs, the cost of transporting passengers through the network, and the operating costs and maximum flying ranges of the different aircraft types in the fleet. We introduce the hub covering flow problem, which accounts for all these design considerations.
We next look at time-definite delivery services in the small package delivery industry, where a customer's package is guaranteed to be delivered within a certain time-window. With possible delays along the transportation route, a package may not be delivered on time. We introduce the stochastic p-hub center problem, which seeks to design a network where the longest origin-destination path in the network is minimized. The solution provides an upper bound on the delivery times, which can be used to design time-constrained service offerings, or to ascertain if current service guarantees are feasible. A service-level constraint is included into the model to ensure a high likelihood of on-time delivery.
In communication networks, two important service-level considerations are robustness (the ability of the network to perform when components become unavailable) and response time (the time required to transfer information through the network). We propose another version of the stochastic p-hub center problem, which employs a two-stage stochastic programming formulation, that addresses the possibility of unavailable components or transmission delays in the network.
Copyright 2007 Thaddeus Kim Teck Sim