The attached doc shows the normal spectrum of a bearing with an outer race defect. At first glance one might think there are sidebands of something modulating the outer race frequencies. But upon closer examination, the "sidebands" are not sidebands at all, but a frequency peak of its own being that is a fraction faster than the suspected BPFO. This can be seen as the spectrum length "grows" and the number of harmonics increase and the distance between the peaks increase.

What do you guys think it might be? I have my guess but would like to hear from you. Do you think maybe it is just a glitch in the FFT from the waveform to the spectrum (not my guess ) or is really something?

The roll is scheduled for removal in 2 days.

sidebands_that_are_not.doc (156 Kb, 156 downloads)

When you tear down the roll, check to see if the inner race was creaping on the journal. I have been able to identify 2 turning speeds on a roll before where the bearing was loose on the journal.

Just my guess,

Gary B

Hi, I would suggest that the frequency in question is a 'felt barring' frequency; ie the number of bars (or compression marks) on the felt times the rotational speed of the felt. These are caused after by minor imperfections in the nip creating bars in the felt over a period of time. These will also be a function of the one of the press roll diameters.
Cheers
Steve

Ralph,
Does it show up on the tending side bearing also, or just the drive side bearing?

Is it reproducible over several measurements or points?

If not, maybe it could represent a sudden change in machine conditions when data was being collected? Maybe the axial loading which would change the contact angle and fault frequency. (just a swag).

Ralph,

Possibly a defect frequency from the pilot bearing?

Ralph,

Could it be that RPM was varying during data acquisition?

David

quote:

but a frequency peak of its own being that is a fraction faster than the suspected BPFO. This can be seen as the spectrum length "grows" and the number of harmonics increase and the distance between the peaks increase.

Hi Ralph, I trust you are confident of the above since we can't really check that on the 1st spectrum. Your blow up shows it better but I would like to see greater resolution. If the difference between which ever harmonic of BPFO and the neighbouring higher peak grows with increasing harmonic numbers then what you are looking at is 2 different sets of harmonics. One relating to damage on each of the 2 raceways in the double row spherical rolling element bearing. The equations used to derive fault frequencies include variables such as pitch diameter, rolling element diameter & contact angle, if any of these are marginally different (dimensional changes due to spalling)compared to design then slightly different fault frequencies will result. That includes differences between in & outboard races.

Alternatively if the "sidebands" prove to be 1x then vibration cannot be from the pilot bearing since 12.25 would be need to be inner race (I'm guessing 241** is the bearing number, and inner race is stationary - and won't produce sidebands. This leaves the tender side bearing which could have a rotating outer race but frequencies would need to match and sidebands 1x. Levels seem low but it is a slow machine speed / roll and the peaks protrude well above the noisefloor, so I would also change the roll. Regards.

Answers to questions thus far, to the best of my ability:

Gary,
"check to see if the inner race was creaping on the journal". Will do

UK,
"I would suggest that the frequency in question is a 'felt barring' frequency" Might be, huh? Will check to see.

JimB,
"Does it show up on the tending side bearing also, or just the drive side bearing?" Just on the drive side

Electricpete,
"Is it reproducible over several measurements or points?" Yes it is. Same from Febuary 8 through the 11.

Danny,
"Possibly a defect frequency from the pilot bearing?" I checked that 22216. It is not the pilot bearing or does not appear to be.

David_G,
"Could it be that RPM was varying during data acquisition?" Good point, but it has shown on more than one days data.

Vibe-Rater,
"I trust you are confident of the above since we can't really check that on the 1st spectrum. Your blow up shows it better but I would like to see greater resolution." I am confidient. I guess I could get the guy to take a higher res but the current res is 200 Hz and 1600 Lines.

One other thing, this roll was installed with a new bearing on or after February 2nd. The first data set on the new bearing was showing the outer race defect frequency. Had not been in storage but a few days, so brinelling might can be ruled out. I have not seen the machine, only the data, since this roll was installed. I was hoping someone had seen this some time in the past and had a real live case history.

If I am reading the data correctly, there is an impacting type event occurring at 12.99Hz, and another impacting event occurring at 13.08Hz (difference of 0.09Hz, or 0.085 orders)

The roll speed is 1.06 Hz (63.6RPM)

About the only thing I can think of that would give that sort of result, which has ben mentioned above, is that there is damage across both outer raceways but due to axial load, or the bearing being cocked, the two rows of rollers are not running centrally in the outer ring, and are therefore giving slightly different frequencies of the same defect.

Or, of course, I could be wandering right down the garden path to where all the fairies live.

Waveform?

Hi Ralph, as previous replies suggests I am with IanS on this. Could you post an extract of your CSI db so we can play with it, might be able to refine analysis....Regards....display in orders to make more sense....oops, mentioned unmetionable...and sideband seperation display.

Vib-Rater and Ians are thinking along the lines as are the analyst who sent me the data and myself. Seems like the logical explanation. But we could be totally wrong.

Vib-Rater,

I will see if the analyst who sent me the data will allow me to send you the data to "play with". I had rather send it via email.

Planty,
Attached is the waveform in 4 different displays.

This message has been edited. Last edited by: Ralph Stewart,

waveform.ppt (348 Kb, 67 downloads) Waveform

Ralph,

I've got to tell you, I have never seen a spectrum like that before! My first take was like Danny, a pilot. Then I thought of one of those wild ones with a third raceway...not real common even for a papermill. I guess this one has my head scratching. Can't answer them all. Here is what I can tell though, the amplitudes are so low, that you are not even at an initial pitting stage yet. You can easily wait until the defect peaks hit .02 in/sec. The waveform does not have a lot of random or periodic impacting and the crest factor isn't high enough ~5.5.

Whew, I need a break now.

Later

quote:

I've got to tell you, I have never seen a spectrum like that before! My first take was like Danny, a pilot. Then I thought of one of those wild ones with a third raceway...not real common even for a papermill. I guess this one has my head scratching. Can't answer them all. Here is what I can tell though, the amplitudes are so low, that you are not even at an initial pitting stage yet. You can easily wait until the defect peaks hit .02 in/sec. The waveform does not have a lot of random or periodic impacting and the crest factor isn't high enough ~5.5.

Thanks Mike,

Waht worries us is that this is a new installation and this is showing from the get go.
They are supposed to pull the roll today. Hope to know something by this afternoon.

I and the analyst who sent me the data thing maybe the housing was cocked when installed and pulled down in a bind and caused some sort of distortation.

Hope to know soon.

Hey Ralph, gotta ask is this bearing during operation have rotating outer or inner race ?

I am guessing you dont believe the peaks are a separate fault but are sidebands of running speed, slightly changing due to bearing crawling.

It being BPFO with any sign of sidebands makes me wonder if this setup has rotating outer race,

I know for sure your on top of it !!!

Take care,

Mike

Picture

Ralph,

Thanks for following up. Does not happen too often unfortunately. So, apparently this is a rare case of a bearing damage - in both races, which probably have slightly different parameters.

IMO the damage (although not pitting ) is substantial. In this regard this case also demonstrates another important thing - no amplitude or CF is too small to indicate a severe problem.

Was the root cause lack of lubrication?

David

Great followup. Grease was contaminated by moisture - you can see the corrosion marks in lines where rolling elements came to stop. Although both races look the same, the inboard side (exposed to spray from process entering through seal) would be worst. What's the bet that side nearest on photo was inboard?? Deep scuffing / wear (is where the rollers enter loadzone, slowed down by grease, then as clearance reduces, they chatter and carry on as they get up to speed again, only to exit at hte other end of the loadzone), the shiny bit just to the right of bottom dead centre as displayed will be the reason for the 2 sets of harmonics - ref my earlier reply. Ralph 1 - paper machine 0.

Another Picture