RADIO-FREQUENCY SENSING OF INCIPIENT ARCING FAULTS WITHIN LARGE TURBINE GENERATORS

Sensing of incipient faults within large turbine generators (≥ 500 MW) is very important because a persistent minor abnormality may suddenly develop into a major fault and cause extensive damage before conventional early warning sensors can respond. Consequently, there is a need for a reliable sensing system which will respond to the early arcing stage of a fault, when insulation surfaces are not at a sufficiently high temperature to emit particulates and activate the conventional Core or Generator Condition Monitor. A promising technique being explored is radio frequency (RF) sensing of arcing faults, which is based on the premise that abnormalities, such as, a broken coil strand, will cause arcing which produces R.F. currents that flow in the machine neutral connection with a magnitude and spectrum signature which can be recognized and differentiated from other radio frequency signals which may be present. Although the R.F. technique is not fully developed, sufficient data have been obtained from machines in service to yield an average spectrum signature which is representative of a generator free from arcing faults. Also, a machine with an order of magnitude greater than normal R.F. level at the neutral was removed from service for inspection even though no particulates had been sensed by the Core Monitor. Clear evidence of an arcing condition was found and repaired, and when the machine was returned to service the neutral R.F. level was in the normal range. In this paper the theory and application of the R.F. technique is outlined and discussed. This paper is based on work performed under EPRI Contract RP 970–1, “On Line Monitoring and Diagnostic Systems for Generators.” Copyright © 1979 by The Institute of Electrical and Electronics Engineers, Inc.