A Mixed-Integer Linear Optimization Model for Reservoir Sizing Incorporating Performance Measures: Deterministic and Stochastic Frameworks

Abstract

This paper proposes an optimal sizing strategy for water storage reservoirs that takes into account the minimum allowed number of failure periods (measure of reliability) and the maximum allowed failure magnitude (measure of vulnerability). The optimization model follows the standard operating policy for reservoir release decision—as in typical design procedures that account for failure—and is compared to an alternative mixed-integer linear programming model that uses an open-loop release policy. Deterministic and implicit stochastic versions of the model are presented, which perform efficiently for several example problems, indicating that higher storage capacities are required for higher sustainability (more reliable and less vulnerable operation) and lower risk.