A scatter search for a multi-type transshipment point location problem with multicommodity flow

Abstract

In this paper, we consider a multi-type transshipment point location problem with multicommodity flow which aims at locating transshipment points (TPs) and determining the type for each open TP. As an extension of the classical two-stage capacitated facility location problem, our problem allows flows of commodities to move among TPs and does not impose restrictions on the number of TPs traversed by flows going from plants to customers. In order to obtain high-quality feasible solutions, we propose a clustering-based scatter search in which the seed solution generation, the diversification technique and the local search are designed using the clusters generated by a data mining technique, the K-means method. The computational results show that the scatter search performs efficiently over different kinds of instances. Moreover, the solution quality is also good since the average gaps to lower bounds range between 1 and 3%.