Simulation-Based Optimization with Constrained SPSA for Water Distribution Networks on Military Installations
Abstract
The purpose of this paper is to combine simulation-based optimization and simultaneous perturbation stochastic approximation (SPSA) to create an effective model of a water distribution network and return the optimal diameters for the system. This paper particularly focuses on a distribution network for a military installation. Using a water network simulation that includes random processes to model real world variability, we minimize the monetary cost and amount of the population that receives an inadequate amount of water. We use sequential quadratic programming and projection constraints to add bounds to our model. We conclude by showing that in two case studies, our model using simulation-based optimization performs better than the previously established pipe diameters in the networks.