Abstract— In the past few years, as manufacturers have reduced stator endwinding support to lower costs, stator endwinding vibration has emerged as an important failure mechanism of large motors and generators. Endwinding vibration, which is primarily driven by magnetic forces in normal operation and much larger forces in fault conditions, leads to high voltage insulation abrasion and copper conductor cracking from high cycle fatigue. Many catastrophic generator and motor failures have resulted. The most effective method to monitor if dangerous endwinding vibration is starting to occur is to continuously monitor the vibration levels and frequencies using fiber-optic accelerometers. Such sensors have been available for over 20 years, but it is apparent that incorrect results are sometimes obtained, leading to false indications on the condition of the stator endwinding support structure. False indications could be the result of multiple causes, and one of them is improper location of the sensors, i.e. the sensors are installed in locations of minimum vibration, making one believe there is no vibration problem. Modal testing is sometimes used to determine the optimum location of the sensors. However, since this test can only be performed at ambient temperature, not at winding operating temperature, it is possible that with a temperature increase, the optimum location positions could be changed. This paper will discuss the requirements for selection of sensors and effect of temperature on modal test results.
