A design and simulation of a Brazilian bid based short-term electricity market

Abstract

In order to overcome a number of difficulties detected on the Brazilian electricity market this paper proposes and tests a bid based short-term market. To simulate the behavior of the hydros in this new market, it was implemented an Agent-Based Model - ABM using the reinforcement QLearning algorithm, Simulated Annealing, and linear programming. In the simulations we used real data from the Brazilian power system encompassing more than 98% of the total hydro installed capacity and three years of market data. The results indicate that the management of (virtual) reservoirs can be under the responsibility of each hydro, which could save water according to their own risk perception, while it is maintained the current efficiency and security levels. Results also suggest that the final monthly short-term market prices can substantially decrease in comparison with the current prices.