| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.04 MB | Adobe PDF |
Authors
Abstract(s)
Face ao caminho que é percorrido com o objetivo de promover a descarbonização das indústrias e da economia mundial, ao longo dos últimos anos tem vindo a ser estabelecidas várias metas e objetivos com o propósito de reduzir a dependência dos combustíveis fósseis. Neste âmbito, a estrutura do ISEL decidiu desenvolver um conjunto de medidas, contempladas na iniciativa ISELGreen. A implementação destas iniciativas previa a instalação de um sistema solar fotovoltaico para a produção de energia para autoconsumo, concluído no final de 2019.
Tendo em mente a otimização das necessidades energéticas do ISEL, para poder delinear estratégias num futuro próximo, serão desenvolvidos modelos de previsão associados ao consumo e à produção solar fotovoltaica, através de redes neuronais artificiais (RNA). Desta forma ao longo deste estudo foram definidas as variáveis que constituem estes dois modelos de previsão.
Com base nos modelos de previsão foi desenvolvido um sistema de otimização que visa a implementação de um sistema a de armazenamento (SAE). O desenvolvimento deste modelo tem como foco minimizar o custo da energia consumida da rede, otimizando as trocas de energia entre o campus e a rede elétrica. Este sistema será desenvolvido através de um modelo de programação linear.
O objetivo será entender o impacto da implementação de um modelo de
otimização com recurso a um SAE, onde o conjunto de entrada deste modelo tem como base os modelos de previsão desenvolvidos com recurso às RNA. No final do estudo será verificada a robustez/validade do modelo de previsão, comparando os resultados do modelo de otimização, com os valores de consumo e produção solar reais/medidos.
Combinando o modelo de previsão com o modelo de otimização foi possível estimar que a implementação de um SAE poderá resultar na redução de 1.1% no custo da energia consumida da rede, valor que foi obtido tendo em consideração os valores previstos. Considerando um modelo de previsão com erro nulo (valores medidos), foi conseguida uma redução no custo da energia consumida da rede de 1.6%.
In light of the efforts being made with the objective of promoting the decarbonization of industries and of the global economy, multiple goals and objectives have been set in the past few years with the purpose of reducing fossil fuel dependence. Within this scope, the structure of ISEL has decided to develop a set of measures, contemplated in the ISELGreen initiative. The application of these measures predicted the installation of a photovoltaic system for the production of energy for self consumption, concluded by the end of 2019. Keeping in mind the optimization of the energetic demands of ISEL, in order to delineate strategies within the near future, forecasting models associated to the consumption and to the photovoltaic solar production will be developed, through artificial neural networks (ANN). As such, throughout this study, the variables that constitute these two forecasting models will be defined. Based on the forecasting models, an optimization system will be developed aiming to implement an an energy storage system (ESS). The development of this model will focus on minimizing the cost of energy consumption in the network, optimizing the energy exchanges between the campus and the electric network. This system will be developed using a linear programming model. The final objective will be to understand the impact of the execution of an optimization model resorting to a ESS, where the origin set of this model is based on the forecast models developed with ARN’s. At the end of the study the validity/robustness of the forecast model will be verified, by comparing the optimization model when the predicted values had a null prediction error. In a more practical sense, these consumption and solar production values correspond to the real/measured values. By combining the forecast model with the optimization model, it was possible to estimate a 1.1% energy consumption cost reduction when implementing an ESS, a value obtained taking into consideration the forecast values. Considering a null error forecast model (measured values), a 1.6% energy consumption reduction was obtained.
In light of the efforts being made with the objective of promoting the decarbonization of industries and of the global economy, multiple goals and objectives have been set in the past few years with the purpose of reducing fossil fuel dependence. Within this scope, the structure of ISEL has decided to develop a set of measures, contemplated in the ISELGreen initiative. The application of these measures predicted the installation of a photovoltaic system for the production of energy for self consumption, concluded by the end of 2019. Keeping in mind the optimization of the energetic demands of ISEL, in order to delineate strategies within the near future, forecasting models associated to the consumption and to the photovoltaic solar production will be developed, through artificial neural networks (ANN). As such, throughout this study, the variables that constitute these two forecasting models will be defined. Based on the forecasting models, an optimization system will be developed aiming to implement an an energy storage system (ESS). The development of this model will focus on minimizing the cost of energy consumption in the network, optimizing the energy exchanges between the campus and the electric network. This system will be developed using a linear programming model. The final objective will be to understand the impact of the execution of an optimization model resorting to a ESS, where the origin set of this model is based on the forecast models developed with ARN’s. At the end of the study the validity/robustness of the forecast model will be verified, by comparing the optimization model when the predicted values had a null prediction error. In a more practical sense, these consumption and solar production values correspond to the real/measured values. By combining the forecast model with the optimization model, it was possible to estimate a 1.1% energy consumption cost reduction when implementing an ESS, a value obtained taking into consideration the forecast values. Considering a null error forecast model (measured values), a 1.6% energy consumption reduction was obtained.
Description
Dissertação para obtenção do grau de Mestre em Engenharia Eletrotécnica – Ramo de Energia
Keywords
Modelos de previsão Otimização Redes neuronais artificiais Sistemas de armazenamento de energia Artificial neural network Forecasting models Optimization Energy storage systems
Citation
DIAS, Gustavo Afonso Burgoa – Sistema de otimização para gestão de armazenamento de energia elétrica em baterias. Lisboa: Instituto Superior de Engenharia de Lisboa. 2023. Dissertação de Mestrado.
Publisher
Instituto Superior de Engenharia de Lisboa