M. Merakli, H. Yaman
In this study, we consider a capacitated multiple allocation hub location problem with uncertainty in demand. We model uncertainty using a polyhedral set where the traffic demand adjacent at each node is bounded above. This model is known as the hose model. We aim to find a hub network that is feasible for all possible demand scenarios and that has a minimum worst case cost. We first present a mathematical formulation of the problem. Our initial computational experiments showed that this model is much harder to solve compared to its deterministic counterpart. To handle large size instances, we propose two different Benders decomposition schemes. We develop an algorithm to solve the dual subproblems using complementary slackness. In our computational experiments, we test the efficiency of our approaches and we analyze the effects of uncertainty. The results show that ignoring demand uncertainty may result in high routing costs and congested hubs.
Keywords: Hub location, robust optimization, Benders decomposition
Scheduled
WE1 Logistic optimization
June 1, 2016 4:30 PM
Salón de actos
