Abstract On-line partial discharge testing and monitoring have been widely applied by petrochemical industries on large motors and generators to determine the need for maintenance of the stator winding insulation system. Over the past 20 years, a single consistent method has been used to collect on-line PD data from over 22 000 motors and generators equipped with the required sensors. Of these, PD data from 8500 machines collected to the end of 2021 have been assembled into a single database, along with machine ratings and machine operating data. For each machine, the PD magnitude for each phase from the most recent test when the motor or generator was operating at normal load and stator winding operating temperature was statistically analyzed. The cumulative probability of occurrence for any PD activity level for any particular machine rating, manufacturing method and winding design etc. could then be produced. It has become clear that the probability distributions for different stator winding operating voltages produce statistically significant distributions for machines of various voltage ratings. Over the years, these probability tables have been correlated with visual inspections of hundreds of stator windings as well as off-line test results. This analysis indicates that when the PD magnitude is higher than about 90% of similar machines tested with the same method, then there is a very high probability of significant stator winding aging. This, combined with the evolution of PD over time, can be used to determine when maintenance is advisable.
