I have an overhang fan ( 1490 rpm) that vibrate excessively, dominated amplitude is at BPFO of SKF 22220EK/C3. Our conclusion is the bearing is damaged and we going to need to replace it. But i think there is another problem beside bearing failure, that is :
1. The amplitude of BPFO is very high, never see this high before.
2. The amplitude is nearly twice from horizontal direction.
3. The clearance between bearing and bearing housing 0.4 mm ( i don't know if there is any connection with the amplitude of BPFO)
I wonder if this BPFO amplification is caused by natural frequency of the fan (resonance).
and is there anyone knows the different between link-belt 22220 and SKF 22220EK/C3, because previously we use link-belt bearing and now we use SKF bearing and it seems that SKF bearing get damaged easily.

thanks and have a good day

This message has been edited. Last edited by: jasmine,

Vibration_spec.doc (90 Kb, 86 downloads)

Jasmine,

Are you 100% confident that this is BPFO? Most of these bearings have BPFO much different from 8.0. Just one has BPFO=8.077. Collect data with higher resolution.

How many blades are in this fan?

David

Jasmine,

Is it possible that the fan has 8-blades? The FOV spectrum shows sides bands around the 8xSS that appear to be spaced at 1xSS, so if if is a bearing fault, then it is probably significant. Resonant amplification is always possible. You need to test to see if it is really present.

Walt

Jasmine,

Is it possible to post a PeakVue or acceleration spectrum and waveform? The spectrums you posted have different RPM ranging from 1481 to 1499.. If your fan is having 8 baldes, you need to find the actual RPM to differentiate the peaks from BPFO to BPF!!

Have a nice day!!

thanks for the replies.
It is 9 blades fan, have check it by my self.
Here is your Peak Vue Jenish.
I am sure that this is BPFO because there is no other excitation near this frequency.

thanks.

Peak_Vue.doc (42 Kb, 52 downloads)

Well, 9 blades, looks like it is an outer race defect with running speed sidebands. The C3 fit might be the reason for the "looseness" sidebands, but probably not alone.

Have you looked to see if there is an inner race defect at 10.9x?

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

[QUOTE]Originally posted by Ralph Stewart:
The C3 fit might be the reason for the "looseness" sidebands, but probably not alone.

What is that mean ralph? should i change the bearing type with tighter fit?. I see spectrum at 10.9 and it's harmonic but it very small compared to BPFO.
Do you think this fan will stand for another month ralph?

thanks.

Although I'm sure everyone knows, to be precise on the terminology, C3 refers to "internal clearance", not "fit". Internal clearance is inside the bearing, fit is between bearing and shaft/housing seats. For example k5 shaft fit and H6 housing fit.

Although the designation C3 is referred to as "looser than normal", C3 in in fact the standard clearance used in virtually all electric motor deep groove bearings and I thought in most other applications as well (as long as no abnormal heating of shaft is expected, in which case you want looser). But I am not as familiar with spherical roller bearings, they may have special considerations.

If these are bearing frequencies then damage is severe.

Jasmine, do you have any historical vibration data in order to make a rough estimation of survival chances for another month? For now I suggest taking readings every week.

I agree with Pete on the C3 thing. This is the most common type of fit for most equipment. The clearance you said you have between the bearing and the housing has nothing to do with the C3 factor.

David is correct in saying "the damage is severe if theses are defects"

I am not saying to put a tighter fitting bearing, what I meant was the C3 might be some of the reason for the 1x sidebands around the outer race, but not all of the reason. There is probable worn extra clearance in the bearing which allows the imbalance/alignment, etc. to add to the 1x sidebands. Usually outer race defects with 1x sidebands are considered more severe than an outer race defect without them. With the inner race frequency also showing, adds to the severity of the bearing problem.

With the high Gs (near 40) amplitude in the Peakvue spectrum, even though some of the data in the Peakvue data is coming from between 40.5x and 70x of your velocity data, IMO, with the use of a 1000 Hz HP Filter, the defect appears to extends out passed your 70x Fmax and is getting more severe, if indeed this is a defect.

To say it will run another month would probably be a 65/35 chance it will, but do not take my word for that, that is only my opinion and I could be totally wrong, but seeing what the data actually looks like out past 70x in conventional velocity or acceleration and acceleration waveform data might help on making a call. At this point in time the bearing is stll responding as it should based on the original geometric configuration (still acting like a bearing with the fault frequencies staying at the calculated position}. Checking it, as David suggested, every week, might give some insight as to how rapidly it is failing, as well as some history during the time the bearing has been in there and adding a few shots of grease a couple of times a week might help. One question or statement not mention, or I think so, is the temperature in air around the fan and in the fan. The hotter the surrounding air and the air it is moving, the more severe things might get.

Why don't you extract out the machine's data (all of it) into a small database and either post it here or sent it to me and a few of the other guys so we can look at it closer?

thanks for the replies

David, here is your waterfall data. note there is a big increase -especially at vertical direction- between 21 Sept 2007 to 21 Feb 2008, in fact this fan was not operating during that time (sister fan was in operation -since Oct 07 until Jan 08). would it be false brineling causing this phenomenon?

Ralph, this is 100 HP motor with 6217 bearing directly connected using grid coupling with 9 bladed overhang fan supported by two SKF 22220EK/C3 spherical roller bearing.

Guys, need explanation about this, top clearance between bearing and bearing housing is 0.4 mm, but i don't see clear harmonic of 1x running speed If this OK, i will not make any maintenance for this bearing housing during next outage.

thanks and have a great day

waterfall.doc (76 Kb, 24 downloads)

Per SKF, Housing fit should be H7/H8 or if heat thru shaft then G7
For a 180mm OD bearing,
H7 - 0-65L, H8 = 0-88L G7 14L - 79L
The highest clearance listed above is about 88 microns or 0.088 mm

You said 0.4 mm which is way beyond that tolerance.
I agree it's surprising that you don't have looseness indication.

Two ideas
1 - Since you measured clearance at the bottom, maybe weight of machine is holding it down
2 - ** More likely - the shape of your assembled housing seat is irregular (not a pure cylinder or taper). This could be due to the way it was assembled, foreign material on the seat, or distortion from some other factor. So the bearing is not held uniformly by the housing seat, but held tighter at some points than others. This could be the cause of your BPFO. Since a non-uniform housing support to the bearing can create a distortion of the outer ring which causes vibration as the balls roll over it. Since the irregularity is on the outside of the ring, the race on the inside of the ring is not an abrupt irregularity like a defect, but a smoother irregularity. This tends to give rise to the fundamental and low-order BPFO (smooth variation as the balls roll over the smooth high spot), rather than the high harmonics (sharp impacts as the balls roll over a defect on the outer race).

quote:

This could be the cause of your BPFO. Since a non-uniform housing support to the bearing can create a distortion of the outer ring which causes vibration as the balls roll over it.

I was thinking along the same lines until I saw the high Gs in the Peakvue Waveform data. True, some of the data in the Peakvue data is coming from the same area as the upper section of the velocity data, and would be nice to see a 2K HP Filter Peakvue to see if the high Gs are past the 70x (1740 Hz) position.

If I read this correctly, the fan has been down for nearly 5 months? What kind of enviroment is it sitting in? Possibly could have gotten a little water or condensate in the bearing causing rust problems and rapid increases in amplitude.

Wonder why, with a clearance so large (0.4 mm or 15 mils) between the outer race and the housing, hasn't the outer race crawled out of the load zone, what with a defect to hammer against constantly. A quick thing "makeshift" that might allow it to run another month might be in opening the housing and rotating the outer race 180 degrees out of the present load zone. Not a recommended fix, but a possible "get by" until time is avaiiable for correct replacement.

Jasmine,

Looking at the PeakVue TWF one can see something that appears to be cage modulation. Couple of FTF harmonics are also appear to be present in the spectrum. Could you confirm that?
Also post Autocorrelated TWF.

The above does not necessarily mean the cage is damaged but the bearing could be starving for lube. At the same time lubrication starvation could have accelerated outer race damage.

I suggest pumping some grease into the bearing and checking the effect immediately after.

David

Good point Ralph. I didn't look at those high G's in TWF (35 g's true pk/0!). Those would not come from irregularity at the housing / OD of the outer ring, but indicate some defect on the outer race (ID of outer ring).

So why do we have both symptoms? Maybe we have both.

An irregularity in the housing support of the bearing creating that low frequency stuff, and defect on the race creating the high frequency and high G's.

What are possible ties between them?
A poorly supported housing fit would cause flexing/fatigue of the ring.... may have accelerated fatigue at the race. Or not.... you'll have plenty more clues to think about for determining the cause when you inspect the bearing.

In any case those high TWF g's have me concerned. Perhaps there is small comfort from the fact that the pattern is simple and doesn't show obvious inner race or ball or cage defects yet (although maybe they are hiding in there...Dave's comments), but still it looks bad from the high g's, the strange clearance measurement, and the strange high BPFO fundamental.

Have you looked at the temperature. Increasing temperature on a bearing showing defects can be a sign of the final stages of failure (I have a case study on that). Although absence of high temeprature doesn't prove much.

A pattern where low range BPFOs are prevailing in magnitude suggests in most cases rather outer race distortion ( as Pete mentioned ) then a race defect. But there are cases when it is still a race defect but the pit is so deep and long that this results in a pattern which we see. This situation will reconcile the apparent controversy Pete has mentioned. In any case this may result in appearance of cage frequencies which are likely present in this case.

Jasmine, both bearings show almost same pattern in velocity spectra. Can you post PeakVue data on the other bearing? PeakVue should localize the culprit.

David

thanks for the clearance pete.
Ralph this fan is placed at outdoor and dusty environment mainly from coal. To overcome this clearance problem we insert shim at upper housing, but we can't change the bearing yet, don't know if this action would give any effect on vibration spectrum, haven't check it yet. Sorry ralph i don't have the PV that your ask, i'll try to get some. By the way, what is happening if high Gs past 70x position?
David, you are right, i have FTF frequency at PV spectrum at both bearing, but i don't see BPFO at inboard bearing PV, would it be too small to see and masked by other frequency?
Here is the auto correlated spectrum and PV spectrum

autocorelated_and_PV.rtf (433 Kb, 6 downloads)

quote:

Ralph this fan is placed at outdoor and dusty environment mainly from coal. To overcome this clearance problem we insert shim at upper housing, but we can't change the bearing yet,can you stop long enough to rotate the outer race 180 degrees in the housing to possibly move the sospected defect out of the load zone? don't know if this action would give any effect on vibration spectrum, haven't check it yet. Sorry ralph i don't have the PV that your ask, i'll try to get some. By the way, what is happening if high Gs past 70x position?

A 2k Hz HP Filter probably would show if the Gs were high in the area past the 70x area. The 70x appears to be where the velocity data stops storing data and this data from 40x to 70x is also included in the 1K HP Filtered Peakvue. In theory, Peakvue is only showing less than 300 Hz past the 70x area? IMO it would say whether or not there are 35+ Gs amplitude in the higher frequency area and probably could help in determining the severity. If the Gs are limited to being below 70x IMO the severity would be less.

Or why not just take an acceleration spectrum and waveform out to 5000 Hz and 6400 lines or higher and see what is actually there in conventional data? This would be less confusing to some who are not too well acquainted with Peakvue data. Better yet, take both.

IMO you can never have too much data so long as you know where it came from.

Wish I could open the Peakvue "can" again, but not yet, not yet.