Gestão eciente dos novos recursos energéticos advindos das redes inteligentes

Abstract

The increase of smart grids penetration has allowed the development of dierent energy eciency resources, such as demand response programs, integration of electric vehicles into the grid, and the electricity commercialization between the microgrid and the main electric grid. These new distributed resources deal with decision under uncertainty, specially the future price, the microgrid internal demand, the time of connection and disconnection of vehicles or the intermittent generation of solar energy. Since this development of smart grids will generate a substantial increase in the amount of data, several opportunities will open in the upcoming years for the application of resources and techniques that focus on energy eciency and process optimization, such as the analyzes performed in this research. This PhD thesis contributes to the study of uncertainties inherent to these processes, evaluating techniques in the literature that are used to optimize these resources along a decision horizon, and propose models that are intended to foster the success of these distributed energy resources along with consumers. Three main contributions are presented, as well as a literature review of each topic at the beginning of each chapter. The rst contribution is the proposition of a demand response program model with reference load trajectories, considering the prices as the controls to be used to decrease the negative externalities of the program through the penalization of volatility. This proposal presents a trade-o approach between real-time electricity pricing and the principle of minimum tari volatility. The results presented demonstrate how volatility is reduced. The second contribution is the study and analysis of the uncertainties in the electricity commercialization between the main electric grid and a microgrid with distributed generation and connection of electric vehicles batteries. These uncertainties include intermittent solar generation, local microgrid demand, real-time prices and the time of arrival and departure of vehicles. The contribution is the evaluation of this stochasticity and how the participation of the microgrid could be fostered with the consideration of a pre-dened budget scenario to be managed by the microgrid in a time horizon. The third contribution is the stochasticity study of local generation prediction in tropical regions, based on machine learning techniques, and how uncertainty in the estimation of future meteorological variables can impact the predictive ability of the model. The results presented in this chapter show how the prediction ability decreases as we increase the uncertainty of these variables. Finally, we also discuss how each contribution presented in the previous chapters connects with each other and how they jointly impact the microgrid management.