Evaluation of municipal water supply operating rules using stochastic dominance

Abstract

A procedure for evaluating and selecting among alternative rules for operating a municipal water supply system is outlined in this study. It is assumed that monthly water demands and supplies are random. The total cost, however, is affected by both current month and future water allocation decisions with respect to the operation of facilities. A perfect foresight model using mixed integer programming is developed and applied to 36 years of historical demand and supply data. Using the solutions to this model, several simple operating rules are derived. These rules are applied to the historical data to simulate system operation, and cumulative distribution of net revenue for each rule is derived. Based on first‐ and second‐degree stochastic dominance criteria, the performance of alternative rules are evaluated. The procedure is also repeated with a set of generated data sequences to check the consistency of the solutions. Average reductions of up to 11% in annual net revenues from those of a perfect foresight model are observed, for various operating rules. Using stochastically dominant rules, annual revenues can be increased by 5% on the average from a commonly used rule based on unit cost.