FACILITY LOCATION PROBLEM WITH BALANCING FACILITY WORKLOADS AND PROBABILISTIC CUSTOMER DEMANDS

Abstract

This study concerns a variant of a facility location problem with the aim of ensuring that demand points, or workloads, are not concentrated on certain facilities. Moreover, instead of assigning each demand point to the nearest facility only, the problem is modeled in such a way that customers at each demand point may choose to go to any facility. Customer behaviour is modeled in such a way that the probability of choosing a facility is related to distances between demand points and facilities. We attempt to distribute workloads among facilities by minimizing the maximum expected number of demand points served by each facility. A computational experiment is conducted to investigate performance of an exact method on problem instances of various sizes with varying numbers of potential facility locations a7id varying numbers of facilities to be located. Experiment results suggest that the exact 7nethod may struggle to find an opti7nal solution when the nu7nber of facilities to be located is relatively small co7npared to the number of available facility locations.