Intermittent Temperature Spikes Prior to Bearing Failure 1

[mellomain]

I have 4kV vertical motor. From Feb 6, 2019 until "the event" on April 9, 2019, the UMB (spherical roller thrust type) indicated high temperature spikes on numerous occasions. The spikes (+300F) were detected via remote indications and most lasted less than 30 minutes. On April 6, 2019 the motor was run in this condition for about 8 hours before being shut down. The motor was restarted on April 9, 2019 and within 20 minutes temperatures rose and smoke emitted from the UMB vent (the motor was shut down and sent to repair shop where bearing failure was confirmed. In ALL of the high temperature incidents, teams investigated only to find vibrations and sounds to be normal (temperatures to be normal once they arrived on the scene). Historical reviews of vibration checks and oil analysis results also found no evidence of a degrading trend. Vibrations and UMB oil samples are conducted every 3 months on this motor; more often if a problem is detected or troubleshooting is required.

So first, is a series of temperature "spikes" a common precursor to motor bearing failure? Secondly, is it common for a motor bearing to fail with only signs of high temperature spikes and no other symptoms (like oil or vibration problems)?

Thanks in advance,
mmain

Man is troubled by what might be called the Dog Wish, a strange and involved compulsion to be as happy and carefree as a dog --- James Thurber

 

[electricpete]

A series of temperature spikes to 300F+ is certainly unusual and cause for concern. I would view it as either a serious bearing condition or a temperature indication anomaly. Lack of seeing anything in vibration might steer me more toward indication anomaly. I might examine how fast the temperature rose and fall to see if it was realistic.

It is more common for early stages of rolling bearing degradation to show up in vibration than in temperature.

Temperature changes aren’t common until the later stages of failure.

It’s rare that a bearing will end up failing without showing up in vibration first.

There are a few scenarios that come to mind to explain what you saw:
1 – The temperature spikes were not due to degrading bearing, but were indication of some other factor (lubrication anomaly, load anomaly, alignment anomaly, foreign material contamination of the bearing lubricant) that later led to bearing degrdataion.

2 – The high-frequency content of the bearing defect vibration signal may have been attenuated if the machine construction doesn't transmit vibration well from the bearing to the point of measurement. (One factor is if the bearing housing is buried deep within the reservoir, another is how the bearing housing is supported to the frame and how many interfaces there are). If your vibration guys are relying on magnitude-based threshholds to evaluate bearing defect (i.e. peak g’s), then this attenuation might trip them up. However in this case, I would still expect that evidence of bearing defect could be found in hindsight when examining the spectra in log scale, and how any low level bearing frequencies changed over time.



I have seen on BALL bearings where a ball defect will seem to come and go as the ball spin axis changes. But that doesn't apply to spherical roller bearing.


=====================================
(2B)+(2B)' ?