I am looking for advice regarding fault detection in a Jones refiner using vibration analysis. Routing is done only 90-120 days with radial and axial points collected on inboard and outboard bearing of the refiner. No early warning signs were detected before present bearing failure. 400 RPM, 2000 Hz fmax used up to this point. Am I missing the range with this? Could really use some input,...Thanks!

Rodney,

First off, IMO, 90-120 days is a little long to go without checking a critcal high $ piece of equipment. But that is just me and my thoughts.

What type data are you taking (CSI or ?) 2000 Hz is not too high if the resolution is high, 1600 to 3200 lines and several revs in the waveform. If you use CSI and want to, you can send me the data for a look see. Are these bearings in a housing under another housing, making access to the actual bearing very difficult?

Thank you Ralph.
Yes it is a CSI 2120 with RBMWare.
1600 lines with 5 plus revs.
will attempt to send you an attachment with spectrum when time permits but in the meantime I will mention that the measurements do not show anything unusual in the way of bearing frequencies even when compared with the follow up taken from the newly installed quill. The bearings are inside an enclosed quill and I have never opened one to see geometry.
Lastly I agree about the time period for inspection frequency but that is not my choice.
Thanks again.
Rodney

Rodney,

If you want to extract out the machine into a small database and send me to look over, you can send it to
ralph_stewart@alertanalytical.com

Routing is done only 90-120 days...No early warning signs were detected before present bearing failure. Hmmm...

John from PA

Rodney,

Naturally, I agree with all that 90-120 days is too long. A quick study of the costs associated with this failure alone (use dollar signs not Hz) might change some minds about that.

With only 400 rpm, like Ralph said, 120000 cpm fmax is high, but as long as you have adequate resolution, it won't hurt. A rule of thumb for setting fmax on rolling element bearings is 70 x running speed, but I generally err on the high side. I would probably be using an fmax of 60000, 3200 lor and 5 averages and collect a velocity spectrum and acceleration twf. I would also collect a PeakVue with 60000 fmax, 3200 lor, 1000 hz hp filter and 1 average. (I'm a recent convert to just 1 average and am still not too comfortable with that.) Make sure that the waveforms have at least 3, but preferably 6-10 shaft revs. Know what bearings are in the refiner and monitor the waveform to make sure you get the best location and mark it. Make sure it is clean and flat and if it isn't, either fix it or glue a target on.



Remind the bosses that you can cover a refiner in about 5 or 10 minutes per month, so it's not a real big risk either.

Thanks for the info.
I also collect a 50 hz at 800 lines for low end information.
The area covered includes five paper machines and support equipment.
Down from three men to two men to one man to no man and presently back to one man.
Rodney

Wow! That settles it-1 PeakVue average. There I just saved you 100 milliseconds. Good Luck!

I can show you some data collection and analysis things that might save you some significant time. E-mail me if you want to talk about it.

quote:

Down from three men to two men to one man to no man and presently back to one man.

WOW! Looks like maybe they (management) don't care about reliability or monitoring or sudden unscheduled shutdowns, huh?.

Ralph,
I am doing the best I can.
I am creating a database to send to you but am not certain how to email it. Does it need to be zipped?
Rodney

No if it is just the one machine with several sets of data history. But using the zip function in rbm will work, but I don't like it. Just attach the .RBM file to the email if it is less than 8 or 9 meg in size.