Modelagem de carga em sistemas de distribuição de energia elétrica

• Ano de defesa: 2018
• Autor(a) principal: Josué Fernando Leal Granados
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ENG - DEPARTAMENTO DE ENGENHARIA ELÉTRICA; Programa de Pós-Graduação em Engenharia Elétrica; UFMG
• Palavras-chave em Português: Modelagem de carga; Algoritmo genético; Fluxo de potência; Sistema de distribuição de energia elétrica; Engenharia elétrica; Algoritmos genéticos; Energia elétrica - Distribuição
• Link de acesso: http://hdl.handle.net/1843/30194

The present work consists in the determination of one load model that allows to calculate correctly of technical losses in electric energy distribution systems. The methodology presented for the calculation of technical losses adopted the standards established by the regulatory entity of the Brazilian electric sector. Currently, Brazilian energy companies perform the calculation of technical losses based on an established regulamentation by the Agência Nacional de Energia Elétrica (ANEEL). In the regulation a model of load ZIP is specified for the representation of all the loads of the system. Thus, this work has two main objectives: i) validating of the calculation of technical losses accordance to the ZIP loading model specified by ANEEL and ii) evaluation of the ZIP load model obtained after a metaheuristic optimization process. Initially, a methodology of load modeling is presented, the purpose of this methodology is to obtain accurate ZIP models of residential and commercial units of the Southeast region of Brazil. After that, the OpenDSS software is used to calculate the technical losses accordance to the models obtained from the modeling methodology, this software executes the power flow and calculates the exact technical losses of the distribution system. The procedure described above is used to generate a Benchmarking function, in which the exact solution of the technical losses accordance to known ZIP loading models is known. Then, an optimization methodology is presented to tackle the problem of load modeling. The methodology is based on a Genetic Algorithm (AG) that interacts with software OpenDSS. The problem is formulated as a mono-objective optimization witch restrictions, which aims at minimizing the error in the calculation of technical losses of the distribution system. This is possible using the Benchmarking function. Finally, the optimization methodology is applied in three distribution systems of different dimensions. In the performed tests different proportions of commercial and residential loads are considered. The results indicate that the optimization methodology obtains load models that improve the calculation of technical losses compared to the calculation obtained by the standard load model.