This is a follow-up to the thread "Is fretting detectable".
http://maintenanceforums.com/eve/forums?m=9271040982&a=tpc&f=3751089011

It appears to be another case where bearing slipping on the shaft didn't show on vibration.

This is an 1800 rpm horizointal between bearings pump. Outboard bearings are back-to-back angle contact (where we had the problem), inboard bearing is deep groove/conrad radial bearing.

Originally in September 06, we noticed discolored oil and hot bearing housing. Took vib cuts at that time (as shown in attached) and oil samples at that time (included in attachemnt). Changed the oil in conjunction with oil samples.

Vibration shows a raised noise floor and a lot of non-sync stuff, but we have always seen that on this pump and it's sisters and is attributed to flow conditions. Sister spectra attached for info.

TWF shows somewhat random impacting 2-3 g's true pk/pk perhaps with some 1x modulation... tough to tell. I would like to have had a demod spectrum but we don't have that.

Oil showed lots of wear particles.

Now in January 2007 the bearing remains hot and we went into inspect. These are back-to-back angle-contact bearings on outboard side of a pump.

As-found position of the locknut had locking-tab in-place to prevent motion but there was axial looseness.

There were varnish stains in the bearing from oil overheating. Worst staining on the inner ring where it was apparently hottest. Some varnish stains on the back sides of both outer rings indicate to me they were not in contact. Balls and races were in good condition except for some dull frosting on the races.

I think maybe during installation the bearings were not fully driven in tight and locknut installed before they cooled.

Oil level setting of constant level oiler was reported high. This might have contributed to the overheating.

Maybe it had degraded since last vib check in September? Maybe we would have seen it on vib if we checked it just before we went into the pump? Beats me. This is a challenging application due to all that background broad/non-sync vib due to flow. Also we don't have a great setup for detecting bearing faults on this machine. I'll check with our vib guy about adding a few things to the normal route data for this machine. (A demod spectrum, an acceleration TWF trending true peak, and a high frequency spectrum.)

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

sfp2b_pumpbrg_Jan07.ppt (1,806 Kb, 95 downloads)

If I understood it correctly these tandem bearings were not properly loaded in axial direction which made it feasable for the shaft to have a few mils play axially. Based on the pictures there is also an unrelated fretting corrosion issue. It is not clear from the description whether or not there was detectable bearing inner race-to-shaft radial fit looseness upon inspection. It appears that there was not, just fretting corrosion. If so, not proper grip could have been detected even in normal spectrum but taken with much higher Fmax.

As far as whether or not the raised noise floor, particularly visible in axial direction, could be interpreted as a looseness symptom, I think the answer is no. Remember that back-to-back bearing arrangement completely eliminates axial movement for proper pump operation. For comparison, a deep grove bearing has some normal looseness in axial as well in radial directions although it does not manifest itself in spectrum until clearance gets too big. In the presented case, my guess is that although not properly intalled, the tandem bearings still had minimal axial play which should not have produce a looseness pattern.

TIOMOAICBTW

David

Detecting a bearing slipping on a shaft with envelope signals, unless the measurement path is through the shaft which requires that the shaft be completely stopped and presumably obvious to the eye, is not as easy as detecting outer race slipping because of the signal attenuation through the bearing. There is usually a large signal attenuation across a bearing and very little attenuation from a race outer surface to the corresponding, immediately adjacent, housing or shaft surface.

Epete

I would also be considering the amount of silicon in the sample as a source of heat to add to the varnish deposits. This seems like a lot for an oil sample. Is this the norm?

Good points David and Duncan and Lee.

Duncan - What vibration symptoms would be expected for fretting/slipping between bearing and shaft?

Lee - Good point, I didn't notice that silicon. I'll look a little closer at that.

David - You understood correclty. I agree with your the axial clearance is not likely related, but I am trying to keep an open mind and always interested ot hear opinions.

More info:

The as-found condition is that the shaft is undersized and the bearing bore is oversized with fretting on both surfaces. So the leading theory would definitely be that the poor fit condition existed since the bearing was replaced which was in September 2005 and was the original cause of our problem. But the records of fits measured in September 2005 are unfortunately very ambiguous. At this stage I think it's important to keep an open mind that this observation of poor as-found shaft fit might possibly be a result of the failure, rather than a cause of the failure.

I think I mentioned that the inboard bearing is single deep-groove conrad and the outboard bearing set is back-back-back angle contact pair.

We replaced inboard/outboard oilers in Augut 2006. The reason was to upgrade the oiler to provide an equalizing configuration. Both inboard/outboard oilers indicated darkened oil shortly after that changeout and the outboard also became hot. (Later we deduce that oil levels had been misadjusted during installation of these oilers). Then we changed the oil itself in Sept 2006 and drew a sample at that time with results attached showed wear particles in outboard but not inboard bearing. Then operated until Jan 07 and disassembled for inspection.

It seems certainly a contributing factor might be the high oil level which cause extra heating in the bearing which might have caused or increased the looseness between inner ring and shaft. (Especially since the discolored oil first appeared after changing the oiler and presumably that was the time that the high level was created at both bearings).

Analysis of temperature difference and interference: The angle bearings are 7308 bearing with ~1.5" bore. The pump pumps cold water so it seems reasonable the large-mass shaft might easily transfer heat away from the bearing and keep the shaft cool under the bearing. I don't know off-hand the shaft fit requirements for an angle bearing, but for a similar deep groove bearing 6308 with a k5 shaft fit, the room temperature interference tolerance is 2 - 25 microns = 0.08 mils to 1 mils. The amount of temperature increase of the steel inner ring temperature above the shaft temperature in order to bring that interference to 0 on a 1.5" steel inner ring would be 8F if we were initially at the low end (2 microns) and 100F if we were initially at the high end (25 microns). (By the way does anyone want to check my math on that?) Certainly a 8F temperature difference doesn't seem unreasonable to me and I'm not positive that 100F temperature difference would be unreasonable given that the temperature was high enough to cause the oil to coat out as varnish (although this may have been after the bearing begun spinning). Also due to surface roughness and dimensional irregularity of course 0 interference doesn't mean no restraining force.... I'm not sure exactly what number of looseness to use to predict where spinning might occur (any comments?), but if we had the worst case only 2 microns room-temperature interference combined with an assumed 100F operating temperature difference, that puts us almost to 1 mil operating clearance which doesn't sound good.

The only ways I can think of where extra axial clearance played a role would be 1 - if it caused increase in temperature due to skidding or 2 - if it caused abnormal loading pattern. I admit these seem pretty remote and I don't see any evidence of either on the races.

Any more comments?

What is the resolution on your spectal data? I see the fmax at 20k. The data seems a little broad. Not very defined.

Good comment Ron. The frequency resolution / bin width dramatically affects the appearance of a broadband signal such as our raised noise floor that I called "flow noise". I forgot about that.

The spectrum setup in those slides is our normal setup for this machine Fmax=72000, lines = 1600, Bin width = 72000/1600 = 45 cpm. I used our normal setup so I could show comparison with sister pumps. I zoomed in because of the unusual behavior that background noise is much larger than running speed harmonics which makes those harmonics tough to spot.

We also have a higher resolution spectrum taken on this one machine back in September as follows: Fmax=60000, lines = 6400, Bin width = 60,000/640 = 9 cpm

Attached slide 1 is the normal spectrum.
Attached slide 2 is the higher resolution spectrum at the same point and time as slide 1.
Comparing from slide 1 to slide 2 I see that as we decreasing bin width by this factor of 9/45 cpm ~ 20%, the broadband spectrum goes down by this same factor. Only the running speed harmonics are mostly unchanged.

On slide 3 I have also included an off-route spectrum with higher Fmax 600kcpm... I don't see any pattern in these peaks.

Any comments?
The conclusion as related to this machine... do you think the noise floor is unusually high or it was just our "normal" spectrum setup that causes it?

sfp2b_pumpbrg_Jan07excerpt1.ppt (212 Kb, 28 downloads)

We had a bearing seize on the shaft about 6 hours after i took a vibration reading on it. It did show some raised floor looseness, but no impacting. It was on a paper machine press roll running 3015FPM and 352RPM. does anyone have any examples with this? We had to scarf the bearing off as it had welded itself to the shaft. The bearing nut and keeper were still in place. After examing the journal, it looked as though at one time, it hade been sprayed to build up the metal. The metal had peeled itself off the original journal in different sections.

One of the main causes of bearings spinning is flaws on two or more bearing surfaces. Depending on fit and clearance, you could have a bearing that doesn't spin as long as the bearing is good. but once the bearings have flaws on more than one surface, the forces that cause the bearing to spin can increase dramatically.