Need some 2nd opinions on this please. This is a 22244 CKC3W33 double row spherical bearing on a large 200 rpm synchronous motor, estimated 10,000 HP. Readings are taken twice per month, there are more readings in between the ones shown in the attachments. The peaks hit perfect for some of the Torrington bearings listed in frequency calculations, but the exact manufacturer is not know at this time. This bearing has been in service for years. My question is, do you think that visual defects will be seen in the outer race? Very early, early stage? Thanks

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

C-202_B.a.rtf (119 Kb, 58 downloads)

Here's a single. How do you attach multiple attachments?

C-202_B.b.rtf (40 Kb, 33 downloads)

Some multiple waveforms.

C-202_B.c.rtf (76 Kb, 36 downloads)

Hi Don,

In my opinion/experience, you'll see the defect, which may be really insignificant but, visible. I've had a similar issue with an AC Motor driving a Root Blower, many years ago and the amplitude, if my memory serves me right, was pretty close to what you have. The only difference was, in my case, the indication was an inner race defect and it was during commissioning that it was picked up. Later, when we removed the bearing and inspected it, sure enough there was a visible light score mark on the inner race.

Cheers...Rajan Muthukrishnan

I see this is a 500 Hz HP Filter and has 15 defect peaks in the 400 Hz Fmax, so I am guessing your velocity Fmax is under 500 Hz, so my question is, is the defect showing in the velocity data and is it showing in a 1000 Hz HP Filter? If it is the outer race, and those are 1x sidebands, there is some excessive clearance in the bearing. I would not make a severity call on a 200 rpm bearing without looking at the non Peakvue data, but that is just me.

I wonder why the twf is so clear when the spectrum shows so much.

I almost always use the PeakVue waveform to assess severity for low speed bearings and this one doesn't look like what I would expect from that spectrum.

Is this a special TWF and if so what are the settings?

I agree with Ralph, I wouldn't make any call on something that big too quickly. If you need more data, get it but don't fool around

I would want to collect PeakVue data with some different filter/fmax combinations and look at autocorrelated twfs for both PeakVue and unfiltered acceleration.

quote:

Originally posted by Danny Harvey:
I wonder why the twf is so clear when the spectrum shows so much.

I think CSI has done a poor job plotting multiple TWFs. Nothing except max values could be seen there. Individual TWF plots should be looked at.

It appears to me that amplitudes in TWF is too small even for such a low speed to make a call.

What are those SBs?

David

Can you display your waveform trend and route spectra, non peakvue?

I normally count on trend and especially waveform trending for severity, after pulling taking pictures to compare with the call made.

This gives me a mental as well as physical data I can refer back if needed to have an idea of the severity of defect to expect.

I rarely ever make a severity call based on peakvue data alone, rather I use all data to make the call, which includes waveform pk=pk trending, vHFD trending, the presence of clear peaks being replaced by floor energy.

In our field when is the time to pull a bearing varies depending on who we are working with.
Some one to see the bearing fall apart when the bearing cap is pulled, while others are ok just seeing physical defects.

This goes back to discussions previously on here asking the qyestion, "when is a bearing bad"?

If you wait until the bearing falls apart then its nearly impossible to do a RCFA on the cause of failure, if you pull while the bearing is intact and able to inspect the defect its easier to determine the cause, even though the bearing may have ran another 6 months depending on speed, load and lubrication.

i have rambled...... sorry...

Ralph, yes my Fmax is at 250 hz. In the velocity data, you have to greatly increase the plot vertical scale to find the 1x BPFO. That peak is at .001 ips, the harmonics are basically nothing at all. The 1000 Hz HP reading are still showing the peaks. Took a reading this morning at this setting, see attachment. I should have mentioned in the 1st post, this motor is driving a ethylene recip compressor (1x running speed harmonics).

Danny, this peakvue reading is set for STW. All previous readings were at 500 Hz HP, this mornings is set at 1000 Hz HP.

David, the sidebands are mainly running speed.

C-202_B.d.rtf (40 Kb, 21 downloads)

Here's a velocity reading. Changing this over to acceleration, there's not much difference. The BPFO is still way down, even lower than in velocity.

C-202_B.e.rtf (47 Kb, 24 downloads)

Don,

I forgot to mention yesterday that you can copy the plots to a .doc file and then post them all in one attachment.

The bearing fault is way off from making an issue out of it. The looseness condition is what really stands out from the last plot you posted. If you trend any waveform parameters or VHFD, watch for small, maybe very small increases in acceleration.

Regards,

The 1X peak with the many harmonics in the velocity data is actually normal. Out of the 10 large recips that are here, this data is normal (years worth of data show no change). Thanks for the opinions on the severity. Seems like we feel the same. We plan on doing a shutdown next month. We'll take a close look at this bearing during the PM.

Good data Don.

I think I would let it run for a while. Looks like it is an early defect. Peakvue, IMO, can make one make too early of a call if all avenues are not explored. 200 rpm is not, IMO, a slow speed as far as using Peakvue for slow turning equipment. It is difficult for me to use Peakvue as the sole source of making a call except on a few special cases.

What are the circled patterns in the plot below.

Those peaks in the circled areas are spaced at exactly 200 rpm, shaft speed. After digging in more, there are BPFI peaks with TS sidebands involved in here too. See attachment. That 3rd BPFI peak is within 11 cpm of the 4th harmonic of the BPFO. That's why they show up in previous data pointing out the BPFO.

C-202_B.f.rtf (24 Kb, 14 downloads)

What do you think about turning the outer race 180 degrees in the housing, moving the bottom to the top? I've heard of people doing this, but I've never done it myself. At this time we plan on just an inspection not a replacement. I don't feel comfortable enough to replace just seeing it in peakvue only.

I don't know if this helps much but I have roll bearings in a paper machine that have had levels as high as you are seeing since they were installed. The defect frequencies line up with the BPFO with harmonics. The levels have not increased over the last few months but are still present. I am taking Peak Vue readings with a 1000HZ filter.

quote:

What do you think about turning the outer race 180 degrees in the housing, moving the bottom to the top? I've heard of people doing this, but I've never done it myself. At this time we plan on just an inspection not a replacement. I don't feel comfortable enough to replace just seeing it in peakvue only.

I have heard of some doing that on huge expensive bearing like on King Rolls in a calender stacks.

See what and if the inner race is showing in the velocity data. This may change the outlook on things if it is.

IMO in order to use peakvue accurately you have to calculate your peak waveform amplitudes for anything under 600 rpm. This has worked extremely well for me. if you get to carried away with the spectrum alone, you will tend to call bearings way too soon. The formula that CSI has works OK but still makes the calls a little bit early. The formula that I have found that is more "real world" amplitudes came from TA.
For 200 RPM BPFO alert would be around 3.29Pk and fault would be around 7-8 G's Pk.
Inner race alert level would be around 1.31 g's Pk and fault would be around 2.6-3 g's Pk.
I generally start paying a little closer attention to things for my benefit as it approaches the alert level, then judge from there the speed of degradation in relation to when the next scheduled outage is. Maybe this helps.

Scott