Hi Ari,

Thanks for your input.
I'm relieved to hear the 67 Hz natural frequency is about right.

I've attached some rundown spectra taken with 1 second increments and notes as to when I think the reverse power is applied, and what I think is happening.
I will appreciate your thoughts.

A mechanical source of the 2x vibration has always been a problem to be specific about given the very soft and forgiving couplings.

I think I've addressed your concern about the rotating stator within it's housing but maybe not so any comments will be appreciated.

The interesting thing is that there are 50 Hz vibrations precisely during the reverse power time and they remain constant for that time, despite a higher inrush current than the following few seconds.

What could be voltage dependant and not current dependant about reverse power ?

Cheers

John

BWest1f.zip (29 Kb, 6 downloads)

John,
1)
That is good coast-down info. Let's have a look at the plots first. At 14:06:11, indeed it looks there is resonance at 49 hz. I wonder: is this also observed at 14:05:53 - 14:05:55, with excitation from 1*rpm HS? Difficult for me to see it on the over-layered plot.
And what about 14:06:30 and later, it is not plotted. At this time point the 3*rpm HS (or 3*rpm LS/RS, does not matter) is passing the 49 hz region, is then a resonance effect detectable as well?
2)
BTW: Is the data acquisition system that fast to be able to produce a accurate spectrum each second whilst HS speed is changing so rapidly? I think you are aware it isn't, refer to 'sidebands' at 2*rpm HS and 3*rpm HS at 14:05:59 par example. Therefore we must take care not to rely too much on accuracy of amplitude and frequency readings.
3)
You have selected clutch ring as a measuring point. Perhaps the resonant condition is more pronounced in other directions and/or on measuring points. As outlined previously, I suspect resonance in axial direction on stationary IC endshields. Do you have coast down data available on this point/direction? Could help us to get more close to the source of 2*rpm RS/LS excitation.
4)
With reference to often used text "reverse power is applied".
I think it is the DC winding that is activated for a while, 9 secs, to decellerate the HS rotor. During this time the RS/LS rotor keeps on running at 1500 rpm because the synchronous machine is still synchronized to the mains power supply. This time of 9 seconds is enough to have the HS rotor arrive at 1500 rpm too and both rotors are magnetically sticked to each other with small slip speed. Then at 14:06:10 the whole unit is taken off-line, because K41 opens, and both rotors coast-down to 0 rpm independeantly of each other and without any electrical supply.
Thus the AC winding remains inactivated all the time during coasting down, starting at 14:05:52.
I hope my recollection don't let be down here or have you arranged something like a temporarily 2-phase AC reversal, Maurice won't permit this I guess? .
This could be an answer to your last question about voltage/current dependancy, with reference to that mysterious 50 hz appearance. I think the answer should be: don't worry about it, it is a temporary DC winding related electric effect and has nothing to do with the beat problem.
5)
Refer to A): The IC RS/LS bearings are ball bearings, not roller bearings.
6)
I agree A) or B) could be existent.
I would like to add another point of view:
What could happen if the thermal expansion does not function properly (interference fit of free bearing outer ring and housing bore has become a shrink fit)? Then the axial/radial clearance may not be zero at all temperatures / operational conditions. This could have an effect on stiffness of endshield resulting in a resonance frequency drop from well above 50 .. 60 hz down to 49 hz. This finally results in amplification of 2*rpm RS/LS rotor and the beat along with the adjacent 1*rpm HS rotor. To me this sounds like the most likely scenario at the moment. This would call for replacement of an IC and subsequently a thorough bearing arrangement diemsional check.
7) Point 6) would inevitably result in following consideration:
In fact the beat is not the problem regarding UPS system reliability, at least the vibration levels are acceptable. The beat happens to be a signal that the bearing arrangement is not optimal and may on day result in a bearing problem.
8)
Enough thoughts for tonight, time for a beer and to go to bed, have a good breakfast !
Arie Mol