Crest Factor? 1

[mtigos]

Hello everyone.
I have a problem with analysing vibration data with regard to "Crest Factor". All other measurements (Overall vibration, velocity and acceleration) seem to be well below the alarm limits. BUT when looking at the Crest Factor, its above the limits. These measurements were taken from a Soda-Ash compactor (20-tons) bearings. Its driven through a gearbox with a 400KW motor.
With all the foregoing, the Compactor was disassembled for repairs (but for a failed oil seal) and it was discovered that the DE bearing had colapsed (cracked inner and outter race as well as the bearing sleeve).

 

[GregLocock]

Is your definition of crest factor the same as mine = ratio of peak value to the rms ?

If so be very careful. High crest factor signals cause fatigue out of all proportion to the RMS level.

I havee never seen a crest factor of 8 before.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[mtigos]

Thanks GregLocock. Yes the crest factor=peak/RMS. We have two production streams and the other Compactor is showing similar values of crest factor (please see attached for comparison), with all other parameters below alarm limits though.
So, really what does the Crest Factor tell someone?
Thanks for your help though.
By the way we are using CSI 2120. I am a graduate and have been placed with the condition monitoring team. Your help will be most appreciated.

 

[GregLocock]

Since it is happening on two machines, I think you need to look a bit deeper. Can you get a time history from the machine, during one of these events?



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[mtigos]

I hope we are on the same page, time history as in trend right? Attached is the trend. Remember I mentioned that we stopped one compactor for replacing seals but found that the DE bearing had collapsed. Now I was just wondering if the other compactor could be going thru the same.

 

[spciesla]

For time history, I believe Greg is referring to the acceleration timewaveform (TWF). The only one that looks "real" is on page 4 of the doc2.doc file you attached. Both the peak and rms acceleration values seem rather low. It would be nice to know more information about your application and measurements.

Bearings (Manufacturer and Model)
Operating Speed
Fmax for TWF collection
# Lines of resolution

Where are you making the physical vibration measurements? How often?
What kind of mount (magnet, stud, etc.)?

 

[GregLocock]

I've got to say the waveform looks a bit noisy.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[BobM3]

On that last page I can see the approximately 13-15 hertz signal (about 11 cycles in the .75 seconds shown) in the time domain plot. I also see that in the frequency domain plot. But where, in the frequency domain plot, is the high frequency noise (evident in the time domain plot, as Greg points out)?

That doesn't explain the crest factor problem though. It does look like the crest factor calculation is about right in the bottom plot (with the "noise" contributing to the peaks).

 

[FeX32]

I wonder if there are any real time or passive filters being used. May alleviate some unnecessary problems common to analyzing noisy signals.
Although, not knowing much about the bearings this may not even be noise?.




Fe