Model-Based General Arcing Fault Detection in Medium-Voltage Distribution Lines

Abstract

Arcing fault is a special fault in medium-voltage distribution lines which can potentially cause shock hazard, apparatuses failure, and wild fire. Due to its short duration or low fault current, the detection of an arcing fault in distribution lines is highly challenging. In this paper, a model-based general detection method to separate an arcing fault from common nonarcing disturbances is proposed. First, an arcing fault model and a nonarcing disturbance model relating the disturbance characteristics, the load parameters, and the substation voltage and current waveforms are derived. For the accuracy of the models, the voltage drop on the distribution line is taken into consideration. Then, the procedure of arcing fault detection is introduced, including disturbance current calculation, parameters estimation, mean-square-error calculation, and decision making. An arcing fault is claimed when the measured voltage and current signals match the arcing fault model better than the nonarcing disturbance model. The method is tested on a modified standard test system using PSCAD/EMTDC considering different fault locations and scenarios. Simulation results show that the proposed detection method has high accuracy and is robust to fault distance, fault resistance, load current, and sampling rate. Laboratory experiments are also conducted to further validate the proposed method.