Real-time turbine-generator rotor torsional vibration monitoring

This paper presents a novel method for turbine-generator rotor torsional vibration monitoring. Torsional vibrations can be evaluated from an accurate knowledge of the angular velocity of the rotor. The input information for calculating the angular velocity is the data measured by an incremental encoder. However, the analysis of torsional vibrations of a turbine-generator rotor is such a specific task that it is not possible to use traditional methods for incremental encoder data processing. The novelty of the described method lies just in the way the measured data are processed. For this purpose, a multi-step algorithm was built, including a Kalman filter. Thanks to this, the resulting algorithm could be recursive and thus able to adapt to changes of operational conditions. At the same time, a result error is also given. The suitability of the chosen approach is confirmed by a series of case studies, in which the evaluation of the torsional vibrations of the rotors of large steam and gas turbines is shown.