Detection of high-impedance faults. Final report
S.J. Balser, Kevin A. Clements, Eugene Kallaur · 1982
High impedance faults usually involve one or two phases of a three phase distribution system. As a consequence, system current unbalance will change during a fault. A statistical algorithm for the detection of high impedance faults was developed based on changes in sequence current unbalance. Staged fault tests were combined with measurements of unfaulted system normal operations to test the algorithm. A Chi-squared test statistic is formed by measuring the sequence current unbalance of the first, third and fifth harmonics. Under balanced conditions the fundamental contains only positive sequence, the third contains only zero sequence, and the fifth contains only negative sequence. In the proposed method, the remaining two sequences are normalized with respect to the fundamental primary sequence and compared statistically to the past history of sequence unbalance. If an abnormal unbalance is present, it is flagged as a possible high impedance fault. The method requires a real time estimation of the normalized sequence current statistics. This will most likely require a microprocessor based device. Similar techniques have been proposed for digital relaying schemes and the proposed method for the detection of high impedance faults could most likely be incorporated into an overall digital relaying scheme.