Method to Profile the Maintenance Needs of a Fleet of Rotating Machine Assets using Partial Discharge Data

Abstract

Online partial discharge (PD) measurements have long been used as an effective means to evaluate the condition of the stator windings of generators and motors. Those who manage a fleet of such assets have the responsibility to minimize the risk of disruption in service. The efficient way of managing is to allocate maintenance funds only to the asset(s) that needs immediate attention and to delay or skip maintenance for the rest in the population. The common practice to shortlist the worst performing assets is to use preset criteria based on PD pulse magnitude, its repetition rate or its derivatives. These metrics are unreliable as PD activity inside electrical insulation can accelerate or decelerate without a change in the physical condition of the electric insulation. Using them to shortlist the worst performing assets often results in incorrect identification and wastage of time and resources investigating the wrong bunch. Therefore, a better method to identify and profile the maintenance needs of an asset is needed. In the paper, a method to identify the worst performing assets in a fleet and determine if maintenance action is needed using PD measurement data is described. The fleet screening tool is based on the estimation of a sampling of destructive energy absorbed by the electrical insulation from PD activity and comparing its longterm accumulated values against a base distribution which is effectively a historical database of annual averages of actual power dissipated by a large population of similar assets. This tool provides a quick and usable result: what percent of similar assets in the population have suffered more damage than any given asset. This allows the asset owner to prioritize the asset for maintenance and minimize the risk of disruption in service. An example of the implementation is illustrated.