UCLA

Stochastic programming is a mathematical model used to resolve the uncertainty of random variables in optimization problems. In reservoir management and operation, the reservoir inflow is typically regarded as a random variable as it brings most of the operation uncertainty. Although stochastic programming has been successfully applied to many reservoir managements cases, the pursuit of the improvement on its accuracy, efficiency, and applicability never ceases. This dissertation consists of five chapters. The first introductory presents the classical stochastic model and describes the challenges. Then, the second chapter develops a statistical model that focuses on improving the distribution fitting accuracy for the monthly average inflow as the random variable. The third chapter discusses a method aiming at streamflow scenario tree reduction, which is essential for alleviating the computational burden of a two-stage stochastic programming with recourse model. The fourth chapter expands the applicability of stochastic programming model, by introducing a multi-objective, multi-stage stochastic programming with recourse model. The final chapter offers conclusions, discussions, and potential future research opportunities.