Abstract

Hub location problems deal with the location of a set of hub facilities and the design of the network so as to provide the most cost-effective way to route a set of commodities through the network. In this thesis we present the Modular Hub Location Problem (MHLP). The MHLP differs from classical hub location problems in the way the economies of scale are modeled. The MHLP considers a step-wise cost function to model the flow dependency of transportation costs at the links of the network. We propose four variants of the MHLP: single allocation and multiple allocation versions with the assumption of having direct connections or not for each case. Computational experiments are performed on benchmark instances in order to evaluate the efficiency and limitations of the considered models.