Transient analysis for ground fault distance estimation in electrical distribution networks: Dissertation

The report discusses post fault transient analysis for single phase to ground fault distance estimation in electrical distribution networks. The networks are assumed to be radially operated, and the neutral either isolated or compensated. The composition of ground fault transients, i.e. charge, discharge and interline compensating com ponents, is first discussed. For transient analysis, a new two frequency model is introduced, which takes into account the interdependence of charge and discharge components. According to the theory, the charge transient, which is primarily due to the voltage rise of the two fault free phases, is most suitable for fault location, since it is the lowest in frequency and in most cases also the largest in amplitude. Of the various factors affecting the transients, fault resistance and loads were found to be the most important ones. When the magnitude of these is increased, the transients become damped, which makes their parameter identification difficult. Load impedances can also cause large errors to the distance estimates, although the parameters were identified accurately. Of significance also is the fault moment. If the instantaneous voltage is zero, the transient amplitudes are smaller than that of the uncompensated fundamental frequency fault current. Also errors of measurement transformers are discussed, and models proposed for the analysis of their transient response. From the other error sources, the effect of fault arc nonlinearity was studied using a simulation model. Of crucial importance are the methods, by which the transient parameters are identified from the measured signals. In the prototype system developed, other signal components, except the charge transient, are first filtered out numerically. The transient parameters are then estimated using a least squares modification of Prony's method. Three methods were developed for fault location. The first one is based on the correlation of transient frequency and fault distance. Unfortunately the method is in practice too sensitive to variation of ground capacitances and impedances of low voltage loads. The other two methods are based on the transient line terminal impedance. The measurements needed are from the phase voltages and phase currents of the faulty line. The location accuracy is, according to the field tests, about one kilometer. However, only the faults with resistances lower than about 50 ohms can be located reliably. Finally other transient based methods, differential-equation algorithms and frequency domain methods, are discussed for comparison.