I HAVE EDITED THE TITLE TO RELFECT A NEW FOCUS OF THE THREAD. 1/8/08.

Can anyone give me the fault frequencies for the following bearing:
Torrington 240BA51 or 240BH51, MRC 148R

I have a the following for MRC 148R
FTF 0.35
BSF 1.45
2xBSF 2.9
BPOR 2.8
BPIR 5.2


What we are seeing on this machine appears to be harmonics of 12.38 orders with running speed sidebands. Although the sidebands seem often higher than the main harmonic peak.

I can't really pick ou the periodicity in the time waveform (even though I suspect it should be 12.38 orders).

The machine is 324rpm. True pk/pk accleration (from TWF) is 25 g's. The ringdown frequency of the impacts on the TWF is around 100,000cpm which is around where the harmonics of 12.38 orders reach their highest level.

It sure sounds like a bearing fault, but I would be more comfortable if I could match it up with bearing frequencies.

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

Pete,

Nothing on the Torrington but for MRC 148R I have (csi database):

ftf-.440
bsf 3.910
bpfo 9.661
bpfi 12.339

Yours looks pretty close to bpfi.

I'll keep looking for the Torrington numbers.

Where did your information come from?

I'll bet you have made a good catch despite bad information.

You da man.

Pete,

It appears that for such a low speed shaft as 324 RPM you have collected the TWF with relatively high, at least Fmax=100,000 CPM, which allowed you to see both high frequency and high magnitude in the TWF. That is good, since normally recommended
Fmax=70xRPM = 70 x 324 = 23,000 CPM
What kind of machine is it, gearbox?

David

Thanks Danny. Your timely information helped fill in the picture very well before the big meeting we had today on this motor.

The fault frequencies above are for MRC 148R from our Entek database. Our Entek database doesn't list those Torrington bearings. There are interchangeable bearings. While I don't expect the fault frequencies to be exactly the same, I'm surprised they are that much different.

David - This is a vertical 3500 hp motor rigid-coupled to an axial flow deep draft pump (circ water pump). Kingsbury thrust bearing on top of the motor. The bearing of concern is a rolling bearing at bottom of the motor which from the drawing looks like an angle contact bearing according to the drawing cross-section (can anyone tell by those bearing part numbers?).

We will be be following this condition very closely over the next two months. So I will be posting some more details soon and then asking some questions about severity.

Epete,

Did someone save that bearing to your database? I don't show it in either of my Odyssey DB's (laptop and desktop).
I looked and it wasn't there. I don't think I have deleted any from either my laptop or desktop.

Dave

Pete,
What about the first part of my question. Do you set such a high Fmax=100,000cpm for such a slow machine? Do you collect spectrum and TWF with different parameters?

Thanks Dave. I'm not sure whether it was added or original... is there a way to tell?

Attached are waterfalls, spectra, twf's

cw11vib.ppt (1,392 Kb, 68 downloads)

Attached are thermal images of lower bearing for this motor (#11) and 5 sister motors. This motor is running a little hotter, although not at a level that would normally concern us if not for the vib pattern.

cwlowerbrgtemperatures.ppt (2,000 Kb, 65 downloads)

Pete,

What Manufacturer on motor and model # if you have it?

Thanks,
Sean

It is a Peeble's Parson's frame 180 vertical motor manufactured around 1980. Rewound in 1996. Refurbished in March/April 2006.

Pete,

The closest Torrington numbers I can match are for a 240BH02, this may be in the same series as the 240BH51, I don't know. The information I have on the 240BH02 is as follows:
Single Row Maximum Capacity Ball Bearing - Open Type,
Dimensions are: ID = 9.4488 in or 240.00 mm,
OD = 17.3228 in or 440.00 mm,
Width = 2.8346 in or 72.00 mm

I hope this helps, it may get you in the ball park. Have you tried contacting a Torrington Rep?

Travis

Thanks Travis. I have not been able to locate any of those numbers on-line yet. Will keep trying.

================================

Now I'd like to change the subject to developing a monitoring strategy for the bearing until it can be repaired.

I realize that forecasting failure is not a precise business. But we are asked to watch this motor very carefully over the next two months until a maintenance availability at which time the motor will be removed and the lower bearing replaced/inspected (we will also check as-found alignment).

If the trend before then indicates that the motor is in imminent danger of failing and won't make the full two months, we have a strategy to take the motor off-line if it is really necessary (it is preferred to wait to the scheduled availability if possible).

I will be looking to develop a monitoring strategy based on temperature and vibration. For example, if temperature increases to 150F on bearing housing or true peak acceleration doubles, or we see an increase in the number of fault frequency families, we would ask to secure the motor.

Any suggestions or comments?

Attached is a comparison of TWF and spectrum for the last two measurements (12/11/07 and 1/7/08). Additional vib data was posted previously.

CW11_Comparison.ppt (252 Kb, 48 downloads)

Pete,

It looks from the twf that amplitudes have dropped slightly since December. Was the bearing recently lubricated? I would increase the frequency of the readings to at least once a week and consider lubrication when assessing the peak to peak value in the twf.

I would also collect Spike Energy readings and auto-correlate the twf if you have that capability.

I'm also with David on trying a lower fmax.

I'm betting it makes it to shutdown, but I don't have to hear the $25,000/hour lecture from the Plant Manager.

ps I don't have much thermography experience but the fact that the shaft is hotter than the bearing housing would seem to support an inner race defect.

IMO there is a possibility that (based particularly on 12/11/07 TWF: single sharp impact every revolution ) there is a crack in the inner race. Significant temperature increase ( +25*F) observed in the bearing area could be attributed to inner race creeping on the shaft rather then to a spall/pit in the race. If this is the case then fault progression is non-linear.

David

Do you have enough historical data to calculate some statistical information? Instead of relying on a 100% increase (doubling), you could use the average plus 3 standard deviations. You could consider using data from the sister motors as well.

Pete,
This is a very interesting case. I'm with danny here, and suggest you shorten the measurement frequency. Once a week would be good for starters but i'll consider taking more radings when you approach the 2 months limit you speak of.

You should read also the Energy Spike levels, if they change rapidly over time, you'll be in trouble.

In the meantime my advice is to lubricate the bearing to soften the spectrum, and if its lubricated by oil, take a sample and do a particle count, i bet it's really high

"I'm not sure whether it was added or original... is there a way to tell?"

I don't know if there is a way to tell if a bearing has been added to the DB or not. I add them all the time when I find info I am certain about. The fact that what you found in yours is so far out from what Danny found would make me suspect someone added one with the wrong info (maybe calculated from physical dimensions or something). However, I don't know for sure. Just curious.

As far as monitoring, I'm with Steve. I would set up statistical alarms, and put a sister machine in the list of those measurements I was setting the statistics with. 2 sigma + 10% and 3 sigma + 10% or something similiar. Manually set the storage spec to "no stat" on those measurements you know are bad if the auto "Exclude data outliers" doesn't catch them. I find it doesn't always work real well, no matter who ole Grubbs was , or what the "Median of Absolute Deviation" is supposed to catch. Collect data like Danny says, weekly.

My 2 cents.

Dave

Grubbs is a tricky fella. A lot of times, I will just export the data to Excel and calculate the average and standard deviation. Then I'll sort the data and see how many measurements exceed the avg + 3 sigma threshold, toss those measurements, and recalculate the average and standard deviation. If you don't have a lot of data, this gets rid of the outliers pretty quickly. If you do have a lot of data, the outliers don't affect the values very much.

Dave - Do you find that the extra +10% helps you much? I can see where you are coming from (having some extra buffer), but justifying it (technically or to management) may be difficult depending on your circumstances.

No Steve, I don't think it helps that much.

There are some trends that the plain old 2 and 3 sigma just doesn't seem to give much wiggle room. They look tight to me. A good running 1800 rpm pump in the 0.0? ips vibe range looks like it needs a little more room to roam, so I give it some percentage.

It's just a mind thing with the "extra buffer". I've always done it.
And you're right, if I had to explain a rationale for setting them that way to someone who was beyond "bs'n", I'd have trouble.

Dave

Pete, I believe that "interchangeable" bearings can sometimes vary in the number of balls/rollers..... which brings to mind that I still can't figure (intuitively) why the BPFI and BPFO add up to the number of balls/rollers.