A study of the propagation behaviour of partial discharge pulses in the high-voltage winding of hydro generators

Abstract

Rotating electrical machines are one of the key assets for a reliable power supply. An unplanned outage of such components can lead to high follow-up costs. To prevent such unexpected occurrences the partial discharge (PD) diagnosis can be a reliable instrument for detecting failures in the insulation system before major damage or a breakdown occurs. However, there have been repeated instances where the measurable apparent charge on the terminals does not correlate with damages found during inspection or repair. One important reason is the fact that PD-measuring systems with high bandwidth can only accurately measure PD-sources, which are located close to the terminals. The investigation described in this contribution is intended to enhance the understanding of the propagation behaviour of PD pulses in a stator winding. The goal of these tests is to contribute to a reliable identification of hidden PD symptoms in the stator winding insulation. To reach this goal, artificial PD impulses are injected at different parts of the winding of two different hydro generators. They are different in their physical size and electrical parameters. These artificial PD impulses are compared to “real” PD impulses triggered by an electrode arrangement, as close to the artificial injection as possible, in the smaller of the two hydro generators. Measurements at the terminals of the stator windings are performed using both, the frequency selective PD system and a wide band digital oscilloscope. The result of the frequency selective measurement is a so-called “attenuation matrix”, where the dependence of the location of injection and the corresponding measurement of the apparent charge recorded at the terminals is presented for different filter bandwidths. The analysis of recorded signals at the terminals of the stator winding shows strong damping phenomena for the high frequency content of the pulse and demonstrates the benefits of the frequency selective measurement.