Centrality based solution approaches for median-type incomplete hub location problems

Abstract

Hub location problems are of those main issues which are focused on by researchers from different aspects for the last three decades especially along with the growth of transportation networks in the world. However, several methods developed for addressing hub location problems do not perform well for large-scale networks because of their computational complexity. This study presents heuristic methodologies based on characteristic features which affect the design of incomplete hub networks. The main idea of this methodology is predicated upon the analysis of characteristics of hub locations. In this respect, the focus was placed on centrality measures, which are frequently used in social networks. Capacity constraints were not addressed, and single-allocation hub location problems were analyzed in this study focused on p-median problems which are the most common problems in the literature. Such characteristics of hub locations as the allocation across the distribution network distance and demand quantities were assessed based on centrality measures, and simple heuristic methods were developed. These methods evaluate nodes across distribution networks from diverse aspects such as centrality, distance, flow, and specify the level of importance of nodes to the networks. Candidate hub nodes are divided into sub-sets in terms of the level of importance and are inserted into the original model as constraints, and then the model is solved under these constraints, which are produced based on certain specifications. In order to test the performance of the proposed methods, data sets such as CAB, AP, URAND and TR which were frequently used in the literature were employed. It was ascertained that the proposed methodology provided good quality solutions in a short solution time. Moreover, the proposed method enables the detailed analysis of hub locations across the network and reduce the problem size. That being the case, it offers valuable opportunities in terms of both the quality of solutions and the solution time for p-hub median problems.