Measurement Methods for End Winding Vibrations of Large Electrical Machines

Abstract

In this paper measuring methods for the detection of end winding vibrations on large machines are presented and special requirements with regard to monitoring and diagnostic systems are discussed.

Large end winding vibrations due to switching operations, grid disturbances or mechanical loosening may cause partial conductor breaks and damage of the end winding insulation. The damaged insulation can lead to a short circuit with damage or even complete destruction of the electrical machine. The resulting costs for operating failure and repair exceed the costs for continuous monitoring of the end winding condition by a factor of 1000 for large electrical machines like turbo-generators in power plants.

In the first part of this paper, the causes and effects of end winding vibrations are discussed. Afterwards, modern methods for measuring end winding vibrations are presented.

Subsequently, a combined method for modal operating mode analysis is presented, taking into account a learned reference state. This modal view allows to draw conclusions about vibration excitation and phenomena such as operation close to resonance. However, the plausibility and spatial arrangement of the individual bar vibrations used for a modal transformation must be critically examined, since errors in the measurement chain and an unsuitable sensor configuration lead to misinterpretations. The modal consideration of the end winding vibration thus always represents a supplement to other observation forms, such as the Fourier spectrum of a single bar vibration. By learning a reference state with the help of neural networks, even small changes in vibration behavior can be detected.

Finally, the paper gives an outlook on the future developments and requirements of corresponding measurement systems