Detectability of internal partial discharges at an artificial reference failure under different types of on-site test voltages.

Inhalt:
Partial discharge measurement represents an essential part of diagnostics in high-voltage technology and is used to detect failures, e.g., blowholes, in insulations. Three different forms of partial discharges are classified: "External discharges"' "Internal discharges", and "Surface discharges". The theoretical background for the formation of partial discharges is provided by the physical considerations of the individual failure. The aim of this work is to design a defect that generates reproducible internal partial discharges under electrical stress. This PD failure is further used as a reference failure for comparing different on-site test voltages and the partial discharges initiate by them. For this purpose, a test specimen is constructed which has an oblate-shaped cavity within an acrylic glass cast resin body between two spheric electrodes as a high-voltage and a ground electrode. The cavity is completely sealed by a sealing ring but can be opened for cleaning purposes or for ventilation of the defect. In this way, it is possible to revise or avoid any pre-damage or changes to the relevant dielectric properties of the cavity. This arrangement thus enables the mapping of an 'internal partial discharge defect' under reproducible manufacturing criteria and thus serves as a reference defect for the interpretation of the partial discharge physics. Crucial is the reproducibility of the partial discharges, which is demonstrated here with a statistical evaluation. Subsequently, the artificial reference fault will be investigated under different types of on-site test voltages present in high-voltage test engineering. First, the partial discharge insertion voltage is determined under the different test voltage types and statistically evaluated. In the second part of the test, the formation of partial discharges at the internal reference failure is investigated by applying a fixed number of periods of one type of test voltage to the test specimen. The main focus of this investigation are the fingerprints, the amount of the partial discharge events and the distribution of the partial discharge amplitudes. This method demonstrates that partial discharges can be generated at the reference fault in a reproducible manner and is a suitable way to compare test voltage types with respect to their ability to detect this fault.