A Data Driven Approach for Monitoring Puncture Events in Capacitor Voltage Transformers

In 2014 a number of Capacitor Voltage Transformers (CVTs) exploded in the Swedish transmission grid. To address this problem, data driven methods for continuously monitoring CVTs were developed. Thanks to these methods, CVT capacitor punctures are now regularly noticed and addressed before causing serious damage. This paper describes the algorithms used to make this possible. Detecting all CVT capacitor punctures as they occur increases the transmission system stability, avoids equipment damage, lowers risks to personnel when visiting the substations and makes it possible to correct measurement errors in the energy measurement systems. Automatic supervision scripts, using energy meter data, are running continuously and will alert the service personal immediately when problems occur. With the algorithms proposed in this article, the Swedish National Grid (Svk) has detected many old punctures and expects no future capacitor punctures to go undetected. Detecting all new punctures enables Svk to save substantial future cost since planned actions now are possible.