Análise Probabilística das Variações de Tensão de Curta Duração em Redes de Distribuição de Baixa Tensão Considerando a Inserção de Geração Distribuída Fotovoltaica

Abstract

Driven by the high indices of solar irradiation in Brazil, the increasing reduction on the price of the photovoltaic solar panels and the high tariffs charged by the utility companies, the photovoltaic distributed generation (PVDG) shall considerably increase its participation percentage in the Brazilian energy matrix in the next years. Due to this tendency, it is important to know the technical impacts of the connection of PVDG to electric grid. It is known that the insertion of DG affects the short circuit levels. One of the most important power quality disturbances in sensitive loads are the short duration voltage variations (SDVV) caused mainly by short circuits (faults) in the electric network. Thus, the insertion of PVDG has the potential to affect quality indices associated with SDVV. However, the power output of the PVDG depends on the stochastic environmental conditions, i.e., the solar irradiance and the environmental temperature. Due to this, a precise assessment of the impact of PVDG on the indices associated with SDVV must be carried out considering uncertainties related to these environmental conditions. This work assessed the impact of PVDG in the SDVV indices of the secondary distribution system proposed by the CIGRÈ using a probabilistic approach. For this purpose, it was developed a model based on the following techniques: network representation in phase coordinates, admittance summation method, failure rate model based on loading condition, state enumeration method and data clustering. Uncertainties in the fault scenario (for example, fault location, type and resistance) and in the output power variation of the PVDG due to random variations of the solar irradiance and the environmental temperature were considered. The fault scenario uncertainties were modeled according to their known probability distributions, while the one related to the input parameters of the PVDG were modeled using data clustering techniques. The state enumeration method based on fault position was chosen to perform the probabilistic prediction of the SDVV indices after a careful comparison between this method and the Monte Carlo simulation method considering both precision and computational effort. The admittance summation method was used to obtain the pre-fault and the post-fault states of the network. It was proposed an improvement of this method in order to reduce the CPU time of the short-circuit states evaluation. It was also proposed a model to obtain the failure rate of a feeder section according to the loading condition of its conductors. This model intended to investigate the variation on SDVV indices due to changing in the reliability data of the feeders caused by the variation on their loading which is caused by the insertion of PVDG in the system. Tests carried on the CIGRÈ low voltage system showed that the insertion of PVDG has little direct effect in the SDVV indices, but it can significantly reduce the indices by the indirect effect of the failure rate reduction caused by the decrease in the distribution network loading.