A Stochastic Hydrothermal Decision Support System for Planning Operation: New Developments for the HIDROTERM Model

The Brazilian hydrothermal system consists of a completely linked network of 143 medium and large hydropower plants, 888 small hydropower plants, and 1,602 thermal and 82 wind power plants, with a total installed capacity of 128,570 MW. The hydropower plants produce, on average, 91% of the total electricity consumed in the country, while the complementary thermal system is dispatched mainly during periods of drought. In our previous studies, we developed a nonlinear programming model, HIDROTERM, to optimize the management and operation of the hydrothermal system, considering individual hydropower plants, thermal generation and exchange, multiple uses of the water, and system expansion. The model is deterministic. In this paper, we develop a new version of HIDROTERM to incorporate the stochasticity of the inflows in the formulation. Additionally, hourly load variation is represented by three levels: peak, base, and low. The proposed approach is based on two-stage stochastic programming with recourse. In the first stage (first time period) the system passes from its initial state to the final state based on deterministic inflow forecasts. From the second time period on (the second stage), the state of the system branches out according to inflow scenarios. The branching can be scaled up gradually until the end of the planning horizon. New developments are made in the formulation, interface, database, and the underlying simulation model. The newly developed integrated decision support system is being tested on the entire Brazilian hydrothermal system.