Locating facilities with resiliency to capacity failures and correlated demand uncertainty

Abstract

This paper studies a capacitated facility location problem under disruptions, where disruption risk not only diminishes facilities’ capacity partially or completely but also affects customers’ demand patterns. We use a budgeted uncertainty set to simultaneously capture facilities’ disruptions and customers’ uncertain demand caused by the failures of nearby facilities. Three two-stage robust optimization formulations with different objectives and performance bounds are built to model the problem. Both column-and-constraint generation algorithm and Benders decomposition method are implemented to solve the models. Numerical tests are conducted to study the impact of disruption correlated demand uncertainty and the performance of models and algorithms.