Alocação de Capacitores em Sistemas de Distribuição de Energia Elétrica: uma Abordagem Baseada em Fluxo de Potência Trifásico Probabilístico Ótimo

Abstract

The Power Flow is a vital tool for power systems analysis, either because it provides the determination of the most diverse electrical parameters in steady state, or because it supports other studies, such as continuous power flow or stability analysis. Its traditional formulation requires deterministic input variables, such as those that represent the generation and load configurations of a given topology. However, analyzing the behavior of a system from a single perspective only assesses a specific operating condition – which does not necessarily reflect the reality. If such limited conjecture is used in planning studies, important errors can be caused due to the discrepancy between the considered input variables and those verified in practice. Seeking to improve these evaluations, the present work proposesthe development of an Optimal Power Flow problem that takes into account a multi-scenario formulation through the Probabilistic Power Flow, in which the demand is represented by a random variable. To model the different scenarios, a clustering technique called K-Means is applied. Optimal positioning and sizing of fixed and switched capacitors in unbalanced distribution systems is sought, considering a daily planning horizon. For this, an optimization framework is developed that deals with each hour of the day individually, in an approach that mixes classical and evolutionary techniques. The determination of the optimal switching strategy of the switched capacitors for each hour of the day is also targeted for optimization, to minimize the expected value of total daily losses and avoid voltage violations in critical scenarios, according to Brazilian standards. To apply the proposed method, tests were carried out in unbalanced topologies: IEEE 13, IEEE 37 and IEEE 123. In all simulations, improvements were found in relation to the total lost energy and the voltage profile. Furthermore, comparisons of the Probabilistic Power Flow technique developed with the traditional Monte Carlo Method were raised, through which conclusions were provided about the statistical precision of the method and its favorable computational performance.