This PDF file attached shows a bearing recently changed and some spectra for this bearing. The double row cylindrical roller bearing has severe damage localized on ~25% of the circumference of one outer race. The rest of the bearing’s surfaces look relatively O.K. This issue developed quickly. One month some high frequency peaks in the spectra and low frequency noise, next month too much vibration for the measuring equipment. This bearing has been in service for about 16 months. It is well maintained in a comfortable environment, an air-handling unit for an office building with a variable speed motor and belt drive.

The spectra taken one month before pulling the bearing shows spikes at harmonics of 3 x running speed. The time waveform shows high frequency noise and some irregular impacts. I believe the bearing is ringing at some frequencies with non-sinusoidal waveform and thus harmonics in spectra.

My first opinion is shaft current damage. This opinion comes from the localized damage area and metal removed from the outer race. The mechanic has measured shaft current around 400 to 500 mVolts that goes to zero with a shaft ground, so there is enough voltage present. However this is the fan shaft so where is the voltage and current coming from?

Is there a second opinion you like better?

Bearing_Damage.pdf (162 Kb, 137 downloads) pdf

Are you certain that the harmonics are 3x running speed and not bearing related? For the amount of damage on the race, I would expect to see some defect frequencies (and harmonics) present.

Steve

Is the bearing on a tapered sleeve? Is it possible the bearing was too tight on the side where the damage was, by too tight I mean not enough clearance between the rollers and the races. Seems like there is a little preload signs showing outside the load zone, is there signs of being loaded 360 degrees? Was the damage on the sheave side or the opposite side?

I'm sorry, but that looks like an angular contact bearing from my persepctive.

Can you post a larger picture that shopws the entire race?

I have myself mistaken an outer race frequency for a 3x. Some bearing frequencies ride so close to this that they are easily mistaken. As Steve was saying, the ammount of damage shown would have to be ringing the outer race.
I have found that bearing faults are usualy better seen in acceleration then velocity. Compare them from time to time and you will see that your higher frequencies show more energy if it is in fact a bearing defect. Also it helps sometimes to strobe the exact shaft speed right when you take your vibration data.

I recently trended a fault in a 22222 skf double roller bearing. It had a harmonic of the outer race with no sidebands. You could hear it from the parking lot as it echoed through the fan housing. I recomended replacement but the plant manager still wanted to run. 2 months is all it took from detection, to increased noise, then it failed and caught on fire! I never saw a raised floor or increase in the lower running speed frequencies. When it was finaly ready to go, it just failed! Bearings can sometimes be that way.

Hey Danny, It's Pete over in West Point! Give me a call sometime when you get a chance. I may have something you could use.

quote:

I'm sorry, but that looks like an angular contact bearing from my persepctive

I'd have put money on it being a spherical (not cylindrical) roller...

The damage on one raceway only might suggest excess axial load - is this the fixed bearing?

If it is a spherical roller, the BPFO is unlikely to be 3.02 orders. Are you sure the speed is correct?

Can you advise the bearing number?

Ian

The bearing is an SKF EXPLORER spherical roller bearing unit (SRY 2 15/16), so Ian would have won his bet. It uses the ConCentra (SKF) technique to fit the shaft.

Load is a concern and will be checked again. There is damage next to the load zone in the (first set) picture that could be attributed to too much preload. This damage may be from the rollers carrying the load on either side of the outer race damage (load zone) and the pounding the rollers were giving the races. The rest of the race (shown in the second pdf file) is relatively clear of damage, except under a magnifying glass small pits and gouges are visible. One of the pictures seems to show deposited material.

The point about acceleration showing bearing faults better is correct, especially if impact spikes are in the time waveform. There are spectra in the attached pdf file showing before (with BPFO marked) and after. The BPFO is slightly above 9 x running speed.

I still think the root cause is shaft current discharging through the bearing, but I am not certain.

Thanks to all for responding, you have been a help with this issue.

Bearing_Damage_2.pdf (256 Kb, 66 downloads)

Looking at this bearing I don't see shaft current being the problem.
Its a preload or an excess axial load what Ralph and Ians have previous noted.
Here is what I would look for.
1. Check if you don't have 2 fixed bearings on this shaft
2.Check that when bolting the housing down to the base you are not preloading this bearing.
3. If this is belt driven check for excessive belt tension and sheave misalignment

One side of the raceway still shows the grind marks from manufacture, the other has wear and damage.

Likewise the rollers; the rollers on the left hand side (in the picture) show wear and damage, the other side are in much better condition.

I'd still be strongly suspicious of excess axial load.

The small pits and gouges could be debris impaction from the spalled area.

Could the voltage mentioned in the first post be a static charge??

ian

Hi,
The rollers look spalled to me. Load zone in the OR also is not uniform and appear to be towards the side.
I would take a look at the installation of the bearing once. Normally fans with spherical rolling elements have two locating rings on either side of the bearing. If there is a mismatch between the locating rings and the shaft step, an axial pre-load can get generated resulting in this type of failure.You will need to carry out a mock assembly without the locating rings and measure the gap between the shaft step and the bearing and then provide spacers of adequate width.
I would also take a look at the grease bucket being used (Have seen such behavior when some impurities went in along with the grease)

What was the PeakVue waveform amplitude on the 08-JUL-06 data in the 2nd pdf doc?

Now that we know that it is a spherical roller bearing rather than a cylindrical, it looks like axial loading is the cause. I would take a look at the assembly and make sure that the spacer rings are located in the proper area and installed correctly and check alignment of the bearings.

It appears that there is significant axial preloading.

Hello,

This discussion is appreciated. Overloading (and overspeed) are still on the table. The mechanic is certain the other bearing is not axially restrained, the bearing fit is correct, and I am certain he knows more about belt tension than I do. However the belt tension is a concern and the bearing loads (weight, belt, axial, and aero) need to be checked.

The pdf file attached contains the peakvue waveform and spectra (repeated).

Thank you,

Bearing_Damage_3.pdf (46 Kb, 35 downloads)

If axial preload is a concern, the belt "alignment" is probably a bigger suspect than belt tension.

Also, when the bearing clearance is "set", the bearing lower half should be bolted to the base. Otherwise, the bearing can be mispositioned relative to the base bolt holes, and when the bolts are tightened the bearing can be cocked. Saw this the other day... the bearing was already set and bolted up, but they'd left the base bolts loose to aid in alignment. However the bolts were sitting at a slight angle to the bearing foot. When tightened, the bolt heads pulled down flat, so something had to give. (Schedule was critical & I couldn't talk them into spending a couple more hours correcting the problem). Just something to watch for.

Another possibility is the bearing was dropped sometime before or during installation. This "dings" the bearing pretty good and can start localized spalling.

I know Peakvue is generally regarded as a great tool, but for normal speed rolling-element-bearings, I strongly encourage you to take your route waveforms in acceleration with a 3 - 4 kHz Fmax and to set up a trend parameter of the Max Peak Waveform value.

It looked to me from the pictures available that the wear was very close to the edge of the race in the load zone. If the roller track is further from the load zone on the opposite side of the race you have probably got a bearing alignment issue. If this is a pillow block housing look at the outside of the race to see where the housing split was located and then you can tell where your load zone was when the bearing was mounted. If you know where your load zone was when the bearing was mounted you will know the angle of misalignment. If your bearing is a taper bore the numbers will usually be to the outside (both inner and outer race). I have used this method to diagnose and correct misaligned bearings with success.
If you suspect electrical current damage you can usually tell by the fact that the rollers and roller track will have a frosty appearance. If you do much bearing failure analysis I strongly recommend a training course, It will do you a world of good to learn the correct methodology and you get lots of good reference materials. I took mine with SKF but there are other providers that may be just as good.

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