Does anyone know a good source of case studies using PeakVue? I am using PeakVue on some slow speed rolls, approxiately 60 RPM, and I have found bearing problems in the regular spectrum using a low frequency accelerometer, 500 mv/g, but I don't see anything in the Peakvue spectrum or waveform. I am using the same low freq accelerometer, 500 HP Filter, 500 FMax in the PeakVue. Any suggestions would be greatly appreciated.
Thanks,
Ronnie

Try using a regular, not low frequency accelerometer. Low frequency accelerometers usually have low pass filters that may remove out the higher frequency signals that PeakVue uses.

Ronnie,
I am assuming that you are talking about paper machine rolls. Your environment is a huge part of your success with PeakVue. It is the "silver bullet" in some areas of our mill, and barely above useless in other areas. Trying different accels as Duncan mentioned is certainly worth a try, but we have had equal success with the 500 and 100. The first thing I would try is lowering your fmax. At 500 Hz fmax, you are pulling 500 orders in your spectrum. Try cutting down to a 100 Hz fmax, 1600 lines, 1 avg. You may try going into "Analyze" and taking the same setup except use the 100-600 band pass filter.

David Eason

Thanks for the replies. I will try both of your suggestions to see if they help. As to what environment I am talking about, they are rolls in the felt mill in a factory producing shingles for roofs. It reminds me of the wet end of a paper machine if that makes a difference.
David, with the 100 Hz Fmax, should I continue using the 500 HP filter? And the 100-600, is that one broad range or individual settings?
Thanks,
Ronnie

Ronnie,
Yes, continue to use the 500 Hz high-pass. It filters out everything below 500 Hz and keeps everything else from 500 Hz to the top end of the data collector's capability. The 100-600 band pass keeps only this range and filters out everything below 100 and everything above 600. Can you post an example of your typical reading?

David Eason

I tried creating an album of zipped documents but I guess that's illegal. So, I'll try attaching them.

This is a PowerPoint presentation on PV.

PeakVue_2000.zip (3,573 Kb, 103 downloads)

These are some Doctor Know tech notes.

PeakVue_Docs.zip (921 Kb, 59 downloads)

And the granddaddy of all the papers...
PeakVue Reliability 2001 Conference

Thanks Patrick. I will take these to work with me and try to read them this week.
I will also try to post a spectrum David.

Ronnie

Hi Ronnie,

I agree with David on the Fmax length of 100 Hz but I use 800 lines on 60 RPM rolls. True, the time to collect the data is long, but we have to do what we have to do. I always use a HP Filter of 1000 Kz though, I found this will catch a cracked inner race a little quicker than a 500 Hz HP Filter, or at least for me on 5 foot dryer cans, heavy.

How big and heavy are your rolls?

My normal Fmax of the velocity is also at 100 Hz but 800 lines unless I think I need to acquire additional data and then I use 1600 lines in the normal velocity spectrum and Peakvue acceleration.

But using the dual point setup, both can be collected in just a little longer time than one by itself, but not nearly as long as both by themselves. Only thing, both Fmaxes and lines have to be the same length when using dual point collection of Peakvue and velocity data.

About the only thing that I DARE make a call from on these 60 RPM rolls from Peakvue is the cracked inner race. Normally everything else shows up in a normal velocity, again at least for me.

Ronnie,
I would go with as low f-max as possible that includes 3 orders of inner race defect. Peakvue holds the highest amplitude in a period that equals the inverse of the fmax*2.56. With slow speed machines a longer period should make the defects more visable in the spectrum.

CSI recommends Fmax=40*RPM for an unknown bearing, so Fmax=60HZ should also be sufficient.
I agree 500HZ HP should be OK, but playing with a 1000HZ is also worth trying.

I am going to attempt to post three spectrum. One is a spectral comparison from Aug-04 to Sep-04 showing a before and after bearing change. This was prior to setting up PeakVue points. I guess I am going to have to do it one at at time.
Ronnie

Spectral_Comparison.doc (24 Kb, 61 downloads)

This Spectrum is a more recent one that has a PeakVue of the same point and time. This is just for a comparison of my regular spectrum and PeakVue.

Ronnie

Regular_Spectrum.doc (23 Kb, 63 downloads)

This last one is the Peakvue.

Ronnie

PeakVue.doc (24 Kb, 75 downloads)

After posting these spectrum another question comes to mind. If what I am seeing in the Aug-04 spectrum is bearing defects, and there were spalls in the bearing when pulled and inspected, I am seeing this above 300 Hz. Would this be a consideration in setting up my Fmax?
Thanks to all who have replied so far.
Ronnie

quote:

After posting these spectrum another question comes to mind. If what I am seeing in the Aug-04 spectrum is bearing defects, and there were spalls in the bearing when pulled and inspected, I am seeing this above 300 Hz. Would this be a consideration in setting up my Fmax?

Are you saying this bearing was pulled and inspected?

Do you normally look at the regular route data in ACCELELERATION?

What type and number are the bearings in the roll{s} you have posted here?

Yes, the bearing was pulled and inspected. I look at the data in acceleration and velocity. I also look at the waveform but nothing noteable to see there. The bearing is SKF 23036 CC.

Ronnie

Did you set the running speed in the PeakVue spectrum and check to see how many orders the first tall peak is? It looks to be about 10 or 11 Hz, if running speed is 1 Hz it could be the outer race defect peak. I think outer race is a little over 11 orders for this bearing. The mound of energy in the regular acceleration spectrum could have been a resonance excited by the impacting at the spalls.

Jim,
If the mound centered at 300HZ is the carrier frequency then by applying a 500HZ HP filter we eccentually cutting it off from PeakVue processing. That explains why PeakVue does not show a thing. In this case a lower HP is warranted.

Secondly, would a bearing defect show up in PeakVue as a fundamental harmonic only?

Ronnie, what did bearing post mortem analysis reveal?