Discharge measurement has shown itself to be the most problematic electrical method of electrical machine CM. It requires special sensors, wide bandwidth (>100 kHz) and very complex analysis for fault detection. It can only be recommended where a specific and costly high-voltage failure mode is being searched for in a known location on a large machine. It addresses one of the most vital parts of the electrical machine, it can detect global effects, including possibly remanent life of the machine insulation and it does give a long warning before failure occurs. Yet, the analysis of the previous chapter shows that, with modern materials, the proportion of machine failures due to insulation faults are now less than a third. Furthermore, the detection methods rely on the most advanced signal processing to extract useful indications, which are then open to wide interpretation by PD measurement experts. This has made it extremely difficult to increase the confidence of machine operators in the value of this type of monitoring because of their need to refer to differing expert opinion. PD monitoring was first applied to isolated insulated components, such as bushings and cable stop joints where it had and continues to have a vital role to play. Its greatest impact to date has been on transformers, substation plant, gas and air-insulated switchgear, where specific failure modes in particular locations are being searched for using both wide-band RFI and narrow-band EMI techniques, with support from acoustic measurement. However, when applied to the distributed, multi-path, multi-connection, variably stressed insulation system of an electrical machine winding, it has a much more difficult task. These methods have been valuable on large machines such as hydro-generators where stator winding fault locations are limited to particular machine ends and their high-voltage failure mode connections, allowing the precise location of sensor couplers, the tailoring of signal processing to that failure mode and where the asset value justifies the application of complex techniques. Work still continues to develop this method, including the use of AI, in these chapter, to determine the overall deterioration of a winding insulation system but that objective has not yet been reached. However, the most thorough treatise to navigate this complex but promising technology.

Chapter Contents:

• 10.1 Introduction
• 10.2 Background to discharge detection
• 10.3 Early discharge detection methods
• 10.3.1 RF coupling method
• 10.3.2 Earth loop transient method
• 10.3.3 Capacitive coupling method
• 10.3.4 Wide-band RF method
• 10.3.5 Insulation remanent life
• 10.4 Detection problems
• 10.5 Modern discharge detection methods
• 10.6 Conclusions

Inspec keywords: machine windings; partial discharge measurement; machine insulation; artificial intelligence; electric machine analysis computing

Other keywords: PD measurement; winding insulation system; sub-station plant; signal processing; stator winding fault locations; sensor couplers; artificial intelligence; high-voltage failure mode; machine failures; machine insulation; online partial discharge; air-insulated switchgear; acoustic measurement; electrical machine winding; fault detection; wideband RFI; hydro-generators; discharge measurement; electrical monitoring; PD monitoring; narrow-band EMI techniques

Subjects: d.c. machines; Charge measurement; a.c. machines; Power engineering computing; Dielectric breakdown and discharges