Advantages of deterministic optimization in long-term hydrothermal scheduling of large-scale power systems

This paper investigates the advantages of deterministic optimization in long-term hydrothermal scheduling of large-scale power systems. A specialized nonlinear optimization model was developed and coupled with a simulation model, taking into account expected values of inflows, in a model predictive control (MPC) approach. It is compared to stochastic dual dynamic programming (SDDP), as currently in use by the Brazilian ISO. Two distinct case studies were conducted for the monthly scheduling of the Brazilian power system on critical hydrology conditions: first, MPC is enforced to equally dispatch the thermal generation as calculated by SDDP; secondly, MPC is free to compute the optimal thermal dispatch, as well as hydro. The results show that MPC provides lower operation costs and less energy shortages due to better hydrothermal coordination and improved cascade reservoir management resulting from the deterministic optimization model, even under unfavorable scenarios.