Fault Detection and Localization in LV Smart Grids

Abstract

In this paper, fault detection and localization methods for a Low Voltage (LV) smart distribution grid are presented. Two fault detection approaches were examined based on current measurements at the beginning of the feeder and on the highest voltage drop across the feeder branches. The localization method was based solely on nodal rms voltage measurements. Five influencing factors were considered: location of the fault, time of the day, fault resistance value (ranging from $0.1 \Omega $ to $1 k\Omega $.), type of the fault (single-phase to ground short-circuit (SC) and three-phase SC faults) and the available measurements. The method performed best for three-phase faults with fault resistance lower that $50 \Omega $ presenting an accuracy of 96.03% in fault distance estimation. The effects of the above factors in the accuracy are also analyzed. The results have been validated by simulation means on a real semi-rural LV distribution grid of Portugal.