A Stochastic Planning System for Siting and Closing Public Service Facilities

Three stochastic programming techniques which incorporate uncertain demands in linear programs are discussed. Although the techniques can be combined they are defined separately. They include chance constrained programming, an expected overage and underage cost approach and multi-scenario planning. Next, a methodology for comparing alternative modeling approaches for a given system is presented. Comparisons occur throughout the modeling process, thus steps for building useful models are defined with respect to the comparison of the models. Finally, the topics are tied together through the introduction of a case study which involves studying the location of public libraries in New York City. The purpose of this research is not to compare stochastic programming methods in general, because as is discussed, this debate has occurred in the past. Instead, we are reviewing the three techniques, introducing a comparison methodology, and demonstrating the defined comparison process on a specific implementation of the techniques. The case study which we use lends itself well to the application of stochastic programming techniques because it is significantly affected by uncertain demands.