Multi-Stage Stochastic Models for Hydropower Generation in a Target-Priority Reservoir with Quasi-Dependent Inflows

In this paper, we investigate the water reservoir operation problem under conditions of stochastic water inflows and exogenous operational constraints. The objective is to maximize the firm energy level while maintaining several operational reliabilities satisfied at a high level. A multistage stochastic programming model is formulated to specify release policies that are adapted to the uncertainty in the problem for shorter time periods. The model is designed to be flexible enough to be a planning or an operational tool. It is then solved by using stochastic dynamic programming (SDP) formulations with custom tailored algorithms that exploit the problem structure for a more efficient solution. The results and the simulation analysis show that the model solutions provide high levels for operational reliability while generating an optimal level of firm energy. The optimization-simulation framework provides energy authorities with tractable, yet efficient, models for short term water release schedules.