A Two-Stage Stochastic Optimization for Robust Operation of Multipurpose Reservoirs

Robust reservoir operation has long been considered a promising solution for addressing water allocation problems in the absence of reliable hydroclimatic forecasts. This study aims to evaluate the performance of this solution using a novel two-stage stochastic optimization model. The model maximizes economic benefits from reservoir deliveries while integrating stochastic inflows into a water allocation system with multiple demands and various constraints. The outcome of the model is a robust set of monthly reservoir releases that perform well under a wide range of hydroclimatic conditions. The model has been applied to the case of the Big Bend Reach of the Rio Grande/Bravo, a transboundary river basin of high importance for Mexico and the United States. The performance of the robust operation policy was assessed by comparing its outcome to those obtained under observed historical operations and an operation policy derived from a deterministic version of the optimization model that assumes perfect hydroclimatic knowledge. The results of this study indicate that the set of robust releases developed here outperforms historical reservoir operations and performs similarly to operations under perfect knowledge. These results show the effectiveness of robust reservoir operation and the usefulness of the proposed optimization model for decision-making under increasing hydroclimatic uncertainty.