Real-Time Optimal Load Dispatch Recommendation System for the Santo Antônio Hydroelectric Power Plant

The real-time optimal load dispatch (OLD) problem in a hydropower plant aims to determine the generation level of each generating unit (GU) in the most efficient way and a few seconds. Due to the hydro production function (HPF), it is a nonlinear programming (NLP) problem that can be efficiently handled via mixed-integer linear programming (MILP). However, solving the real-time OLD of the Santo Antônio Hydro Plant (SAHP) is particularly challenging, both in terms of modeling and execution time, due to the following issues: (1) there are 50 GUs with different hydraulic efficiency curves; (2) the gross head is dependent of the GU dam location; and (3) significant (and potentially dangerous) losses in the trash racks produced by the Madeira River debris. In this scenario, this paper presents a recommendation system for the real-time OLD of the SAHP based on MILP. In real-time, the system employs gross head and trash rack losses measures to model and solve a MILP problem that obtains the GUs power level to be used by the hydro plant operator. We compare the system solution with the strategy currently used by the plant operator and with the equal dispatch between the identical GUs. The results emphasize the improvement in plant productivity. In the 2019 year, for example, the OLD real-time recommendation system presents a 9852.63 MWh increasing in power energy generation.