Application of Nonlinear Optimization Models to Long- and Medium-Term Hydrothermal Scheduling of the Brazilian Power System

This study applies nonlinear optimization to the long- and medium-term hydrothermal scheduling in the Brazilian Power System (BPS). The long-term problem is addressed with a monthly stage simulation model employing Model Predictive Control (MPC) as the operating policy. Thus, in each simulation stage, generation decisions for hydro and thermal plants are determined by solving a multistage deterministic nonlinear optimization model derived by substituting inflow random variables with deterministic values based on expectations. The medium-term problem is formulated as a multistage deterministic nonlinear optimization model, encompassing weekly stages followed by monthly stages until the end of the operation planning horizon. Numerical results from case studies performed with BPS are presented, as well as comparisons with the corresponding linear models in use based on SDDP. The advantages of MPC, which derive from the preservation of the nonlinear modeling of hydropower generation, are the maintenance of high levels of storage in the hydroelectric system and the operation of cascading reservoirs preserving storage in downstream reservoirs, which increases hydro generation and consequently reduces thermal generation and operating costs.