A robust optimization model for distribution network design under a mixed integer set of scenarios

Abstract

The design of a distribution network under uncertainty is addressed by using the minmax regret paradigm. Economic or technical uncertain factors may have a strong influence on the behavior of a distribution system and should be taken into account in the design process. These factors can affect to the construction and operating costs, origin or destination of the supplied commodities, required demands or arc capacities. In this paper, some of these uncertain parameters are supposed to be continuous while others are clearly discrete in nature, giving rise to a mixed integer set of scenarios under which the system may be assessed. We explore how to design a robust distribution network when the uncertainty in the continuous parameters is modeled through polyhedral sets combined with other sources of uncertainty for the discrete parameters. The corresponding optimization problem is solved using a Benders decomposition framework. Finally, a computational experiment with specialized solvers is carried out in order to test the efficiency of our algorithms.