Locating multiple capacitated semi-obnoxious facilities using evolutionary strategies

Abstract

This paper presents a model for the semi-obnoxious, multiple capacitated facility location problem on a Euclidean plane. Even though this problem arises often in public planning, this is the first known paper solving this class of problem for more than one facility in continuous space where capacity is considered. The problem is solved using a bi-objective evolutionary strategy algorithm that seeks to minimize social and non-social costs. The effects of under and over capacitating the facility are included in the cost functions. Two case study problems are solved, one involving the siting of fire stations in a college town, the other locating solid waste transfer stations in a major metropolitan area. The solutions produced by the model are compared to the real-world placement of the facilities. The algorithm yields many non-dominated solutions covering the whole range of values for both objectives in a short amount of time. Also considered are the effects of adding multiple facilities sequentially as happens in a real-world, phased implementation that spans years, as opposed to placing the facilities all at once.