Analysis and Prediction of PV Impacts on Medium Voltage Australian Distribution Networks

Abstract

This paper studies the impact of solar generation on the distribution network voltage and analyses the impact using a number of voltage indices. These voltage indices take into account any deviation or rise from nominal voltage, voltage ramps, and voltage variances. In addition, a machine learning (ML) based model to predict the voltage indices as a function of solar generation has been implemented. A real-life case study considering 18 medium voltage distribution feeders in Australia including residential, industrial, hospital and laboratory feeders with three phase voltage data from phasor measurement units and postcode level solar generation data has been presented. Statistical analysis of the voltage indices shows that most of the feeders experience voltage rise during maximum solar generation periods. The analysis also considers the impact of a generation trip event and finds that the voltage indices violate significantly before a power outage. It can be observed that the ML models can predict the voltage indices more accurately when these indices are discretized.