Presentation: Power Transformer Simulation Laboratory for Proactive Maintenance

Since 2008, Hydro-Quebec, one of the largest electrical companies in North America, has undertaken a major project of online maintenance for its power transformers and other transmission apparatus called IMAGINE. The data gathered from diagnostic devices installed on more and more transformers is now transmitted, collected and centralized in a data historian—the PI System.

The aim of the project is to go from the current preventative maintenance approach to a predictive or proactive maintenance approach.

In 2011, we started a research project on power transformer predictive maintenance called MEET with the object to use the acquired at a to refine our power transformer behavior models. This paper presents the infrastructure, experimental setups and experiments done so far int he research project. One goal of the MEET project is to devise a system to help in the proactive detection of transformer's state degradation in order to avoid fault conditions. The general setup is composed of a power transformer simulator that generates normal condition measurements and abnormal condition measurements based on different problematic situations. These measurements are gathered in the PI Server. The PI Server is further equipped with the Microsoft StreamInsight module to help in the detection of these different degrading conditions using complex event processing. The detection of degrading condition is done by comparison of the measurements received and the theoretical values based on the normal condition of a transformer exposed to the same operational conditions.