Planejamento da expansão de sistemas de distribuição de energia elétrica considerando restauração do fornecimento
Carregando...
Data
2019-02-25
Autores
Possagnolo, Leonardo Henrique Faria Macedo
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A grande maioria dos sistemas de distribuição de energia elétrica opera de forma radial. Isso significa que cada carga é alimentada por apenas uma subestação por meio de um único caminho. Entretanto, as redes de distribuição apresentam estrutura malhada, de forma que, caso uma contingência ocorra, o restabelecimento do fornecimento possa ser realizado para o maior número possível de consumidores. Os trabalhos que lidam com o problema de planejamento da expansão de sistemas de distribuição, no entanto, geralmente consideram a expansão do sistema para apenas uma topologia radial, sem levar em conta aspectos da restauração do fornecimento para melhoria dos índices de confiabilidade. Nesse contexto, este trabalho aborda o planejamento de sistemas de distribuição considerando aspectos econômicos e de confiabilidade, de forma a incluir a restauração do fornecimento no problema de planejamento da expansão. Na formulação do problema considera-se a expansão de novas subestações, o reforço de subestações existentes, a construção de novos alimentadores em novos caminhos, a troca de condutores existentes e a alocação de geradores distribuídos, além de expansão multiestágio e restauração do fornecimento para melhoria dos índices de confiabilidade. Dois métodos alternativos são propostos para resolver o problema descrito: o primeiro considera modelos matemáticos com diversos graus de precisão, para serem resolvidos por métodos exatos, e o segundo é uma meta-heurística de busca e vizinhança variável, que resolve o modelo completo do problema de forma aproximada, sem garantia de otimalidade. A solução inicial da meta-heurística é gerada por uma estratégia que constrói uma solução para o problema usando busca em vizinhança. Os modelos matemáticos e a meta-heurística foram implementados na linguagem de modelagem matemática AMPL, sendo que os modelos foram diretamente resolvidos com o solver comercial CPLEX, que também foi utilizado para resolver os subproblemas de otimização da meta-heurística. Foram realizados testes nos sistemas de 24 e 54 nós, apresentados na literatura, e os resultados indicam que a inclusão da restauração no problema de planejamento de sistemas de distribuição leva a soluções que apresentam maiores custos de investimento que os modelos tradicionais, entretanto, os índices de confiabilidade da rede são melhorados, e dessa forma as soluções obtidas considerando a abordagem proposta apresentam custos totais de investimento e operação e índices de confiabilidade melhores que os das soluções disponíveis na literatura.
The vast majority of electricity distribution systems are operated radially. This means that each load is supplied by only one substation through a single path. However, distribution networks have a meshed structure so that, in the case of a contingency, the supply is restored to as many customers as possible. The works that deal with the distribution systems expansion planning problem, however, generally consider the expansion of the system for only one radial topology, disregarding the restoration aspects to improve reliability indices. In this context, this work deals with the planning of distribution systems considering economic and reliability aspects, to include the service restoration in the planning problem. In the formulation of the problem, it is considered the expansion of new substations, the reinforcement of existing substations, the construction of new feeders in new paths, the exchange of existing conductors, and the allocation of distribution generation, besides multistage expansion and service restoration to improve the reliability indices of the system. Two alternative methods are proposed to solve the described problem: the first one considers relaxed or approximated mathematical models to be solved by exact methods, and the second one is a variable neighborhood search metaheuristic, which solves the complete model for the problem approximately, without guarantee of optimality. The initial solution of the metaheuristic is generated by a strategy that constructs a solution to the problem using neighborhood search. The mathematical models and the meta-heuristic were implemented in the mathematical modeling language AMPL, and the models were directly solved by the commercial solver CPLEX, which was also used to solve the optimization subproblems of the metaheuristic. Tests were carried out using the 24 and 54 nodes systems presented in the literature, and the results show that the inclusion of the service restoration in the distribution systems expansion planning problem leads to solutions that present higher investment costs than the traditional models. However, the reliability indices of the network are improved, and in this way, the solutions obtained considering the proposed approach present lower overall costs of investment and operation, and better reliability indices than those solutions available in the literature.
The vast majority of electricity distribution systems are operated radially. This means that each load is supplied by only one substation through a single path. However, distribution networks have a meshed structure so that, in the case of a contingency, the supply is restored to as many customers as possible. The works that deal with the distribution systems expansion planning problem, however, generally consider the expansion of the system for only one radial topology, disregarding the restoration aspects to improve reliability indices. In this context, this work deals with the planning of distribution systems considering economic and reliability aspects, to include the service restoration in the planning problem. In the formulation of the problem, it is considered the expansion of new substations, the reinforcement of existing substations, the construction of new feeders in new paths, the exchange of existing conductors, and the allocation of distribution generation, besides multistage expansion and service restoration to improve the reliability indices of the system. Two alternative methods are proposed to solve the described problem: the first one considers relaxed or approximated mathematical models to be solved by exact methods, and the second one is a variable neighborhood search metaheuristic, which solves the complete model for the problem approximately, without guarantee of optimality. The initial solution of the metaheuristic is generated by a strategy that constructs a solution to the problem using neighborhood search. The mathematical models and the meta-heuristic were implemented in the mathematical modeling language AMPL, and the models were directly solved by the commercial solver CPLEX, which was also used to solve the optimization subproblems of the metaheuristic. Tests were carried out using the 24 and 54 nodes systems presented in the literature, and the results show that the inclusion of the service restoration in the distribution systems expansion planning problem leads to solutions that present higher investment costs than the traditional models. However, the reliability indices of the network are improved, and in this way, the solutions obtained considering the proposed approach present lower overall costs of investment and operation, and better reliability indices than those solutions available in the literature.
Descrição
Palavras-chave
Busca em vizinhança variável, Índices de confiabilidade, Meta-heurística, Otimização, Planejamento da expansão de sistemas de distribuição, Programação não linear inteira mista, Restauração de sistemas de distribuição, Distribution systems expansion planning, Distribution systems restoration, Metaheuristic, Mixed-integer nonlinear programming, Optimization, Reliability indices, Variable neighborhood search