We have a combustion air fan at our plant, direct coupled, 3589 cpm, overhung rotor, with Fafnir SMN307KS bearings. We recently installed a repaired fan and due to an assembly error, both bearings were running as a fixed pair. One had a stabilizer ring installed and the other was not centered in its housing. In the absence of an actual outer race defect like flaking or a crack, is there something about the design of a self aligning bearing that would make the fundamental BPFO become the dominant peak in the axial plane spectral data while running with this axial preload that the assembly error created?

I would have to guess that the installation error is related to the observed cage frequency. Can I assume that the bearing cited is a double-row spherical roller bearing? These bearings are supposed to accept some amount of misalignment. While doing some previous research, I was surprised, however, how little misalignment this is. The SKF Hotline cited only 1/2 deg total angular misalignment for similar bearings used in our fans. While that may seem totally achievable from a machinist's point of view, it alerted me to the fact that some care must be taken in achieving some acceptable measure of alignment... even with a self-aligning bearing.

So, now... in your situation you have a constant axial preload driving the inner ring and entire roller train against the outer ring. My intuition is that any amount of angular internal mis-alignment in the bearing, once thought acceptable, is now unacceptable... periodically driving the entire roller train into the outer ring at a frequency of once per cage rotation.

I am not, normally, an alarmist... but I would be somewhat uncomfortable with this situation. Without any means to accept axial growth, I suspect the bearing may suffer unexpectedly high loading? I recommend looking very close at whatever means you use to trend early stage bearing degradation... such as peak-vue or envelope detection. Be ready to act if you see evidence of such emerge. I would, also, give your maintenance people a chance to look for some opportunity to correct the installation error at a future convenience.

Are both bearings on one baseplate? Gotta correct the mounting to provide endplay. Shifting one bearing slightly on the shaft would do it. the (SAOL?) housings can stay where they are.

quote:

SMN307KS

The dodge reliant website says this is a wide inner ring ( WIR) single row deep groove ball bearing with an eccentric lock ( grrr) and the OD of the outer race spherically curved to neutralize static misalignment.

http://www.dodgereliancestore....mken/USA/wir-smn.jpg

Commercial shafting is undersized, which lets the eccentric lock tip the inner race even more than if the shaft fit more snugly. The inner race then has a motion with an axial high spot, like a swash plate. The sketch I made suggests that would make the shaft hop axially a little bit as the race high spot climbs over each ball, but at BPFI frequency. So it sounds like maybe an outer race is slightly tipped. Depending on the fit of the nesting spherical surfaces it can take some impact to get them to shift and align.

Here is a little history on the blower. It was installed on a Friday afternoon, (backshift) and no acceptance testing was attempted until Monday. The unusual vibration signature and the rotor speed along with machine criticality left us little choice but to make an attempt to set proper end clearances on this blower shaft. The fan side bearing was sucessfully moved to the measured center of it's housing. The stabilizer ring was left in the coupling side bearing housing and after starting it back up, vibration at the fundamental BPFO was even higher than before. Amplitudes at 184.2 hertz are now at .962" sec peak. We shut the blower off and replaced it with another unit that runs fine now. Note that during the field adjustment attempt the bearing, once unlocked from the shaft, moved easily in the housing. We found no evidence of visible defects once disassembled and still remain confused as to the source of high amplitude vibration at the fundamental outer race frequency. An earlier question about the mounting arrangement, yes the bearing housings are mounted on the same piece of 1" thick steel plate.