Towards a new partial discharge calibration strategy based on the transfer function of machine stator windings

Individual partial discharge (PD) detection and measurement is well recognised as a methodology for providing data on which assessments of the insulation integrity of high power plant can be made. It relies, as a technique, on the accurate acquisition of PD pulse data, including pulse magnitudes. Unfortunately, due to calibration difficulties with partial discharge detectors in static or rotating machine applications, pulse magnitude measurements may be erroneous. Calibration difficulties arise when the PD pulse is not simulated by the calibration pulse. This occurs when the PD pulse is 'distorted' as it propagates from its site of origin to the machine/detector terminals. To aid in the development of a new calibration strategy, the paper reports the results of two techniques. In the first, the transfer functions of different sections of stator winding are investigated to determine whether a frequency band exists in which a PD pulse would suffer minimal distortion as a function of location, in its propagation from site of origin to detector. Such a frequency band would indicate the appropriate bandwidth to which a detector could be tuned to avoid errors which were location dependent. This work builds on previously reported investigations. In the second, a technique of detected PD pulse reconstruction is investigated. The distorted, detected pulse postpropagation is reconstructed to its original magnitude via transfer function techniques. Given that the transfer functions from different locations in a winding are known and the location of the discharge can be identified or inferred, the technique offers an empirical approach to minimise calibration errors.