Improvement of a Hydrogenerator Prognostic Model by using Partial Discharge Measurement Analysis

Normand Amyot, Claude Hudon, Mélanie Lévesque, Mario Bélec, and Olivier Blancke
Submission Type: 
Full Paper
phmc_17_056.pdf424.26 KBAugust 18, 2017 - 11:06am

Availability and performance of hydrogenerators are key features that have driven electrical utilities to implement monitoring and diagnostic methods in order to evolve to condition based maintenance (CBM). Ten years ago, Hydro-Quebec has implemented a home-built web-based application, called MIDA, to cover most of its power plants. MIDA centralizes diagnostic data from several tools, aggregates all diagnostic results and calculates a health index for each hydrogenerators. Data from MIDA used in conjunction with PHM techniques can feed a prognostic model that will provide useful equipment information and lead to the implementation of predictive maintenance. The prognostic framework used for hydrogenerators is based on a failure mechanism and symptom analysis (FMSA) approach. For the stator, a major component of hydrogenerators, more than 100 failure mechanisms have been consigned in the form of causal trees or graphs. A large number of these failure mechanisms involve the presence of partial discharges (PD) before failure occurs. At Hydro-Quebec, PD measurements on hydrogenerators have been carried out over the past 30 years and a significant PD database is integrated in MIDA. The analysis of this huge amount of data is of paramount importance to understand the behavior and evolution of the discharge activity in order to build a robust prognostic approach using physics as well as data driven models. To that end, this paper presents case studies that shed some light on key features related to the evolution of PD activity in hydrogenerators. The paper discusses how to use this data in the prognostic model to assess warning signs before failure occurs.

Publication Year: 
Publication Volume: 
Publication Control Number: 
Page Count: 
Submission Topic Areas: 
Model-based methods for fault detection, diagnostics, and prognosis
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed