An Application of Decision Trees Algorithm to Project Hourly Electricity Spot Price as Support for Decision Making on Electricity Trading in Brazil

Estimating the price of a financial asset or any tradable product is a complex task that depends on the availability of a reasonable amount of data samples. In the Brazilian electricity market environment, where spot prices are centrally calculated by computational models, the projection of hourly energy prices at the spot market is essential for decision-making, and with the particularities of this sector, this task becomes even more complex due to the stochastic behavior of some variables, such as the inflow to hydroelectric power plants and the correlation between variables that affect electricity generation, traditional statistical techniques of time series forecasting present an additional complexity when one tries to project scenarios of spot prices on different time horizons. To address these complexities of traditional forecasting methods, this study presents a new approach based on Machine Learning methodology applied to the electricity spot prices forecasting process. The model’s Learning Base is obtained from public information provided by the Brazilian official computational models: NEWAVE, DECOMP, and DESSEM. The application of the methodology to real cases, using back-testing with actual information from the Brazilian electricity sector demonstrates that the research is promising, as the adherence of the projections with the realized values is significant.

Enabling Hybrid Projects through Appropriate Market Design

Large-scale hybrid power plants, composed of two or more generation sources and with the participation of energy storage systems, have driven important electricity Market Design regulation discussions worldwide. Regulatory framework ought to be adapted to support technical particularities of these new generation arranges. This paper presents an assessment of the main requirements to be met by Market Design to enable hybrid power plants by means of assertive market incentives. Assessing regulatory adjustments promoted in Australia, United States, India, China, and Brazil, emphasizing the latter one, the authors presents a case study by applying specific computational simulation and optimization model to a hybrid Hydro-Solar plant, that supports the findings for the necessary evolution needed in the national regulatory framework in order to enable hybrid projects. The evaluation of international experiences indicates that the insertion of hybrid projects is associated with the design of the market they belong to and demand regulatory adjustments so that their attributes can be properly valued for the benefit of all stakeholders, especially for the electricity consumer.