Abstract
In recent years, stator endwinding vibration has developed into an important failure mechanism of large motors and generators. This can be attributed partly to the efforts by manufacturers driven by end users to reduce costs. Consequently machines are being operated with insufficient stator endwinding support. This lack of support leads to excessive motion between parts resulting in dusting/greasing and ultimately cracked conductors due to high cycle copper fatigue. Additionally, load cycling machines, especially in the deregulated market and with demand fluctuations experience additional forces that can lead to endwinding vibration. In order to avoid premature failure, this excessive motion during operation should be monitored. Stator endwinding vibration has been monitored for over 20 years using fiber optic accelerometers, but it has become apparent that the results of these monitoring programs are sometimes incorrect. An effective online monitoring system should incorporate offline testing to identify optimal locations for accelerometer installation, capture a wide frequency range of data, have the ability to view the data in various measures of vibration, and correlate the vibration with operating parameters.
