This is a centerhung hammer mill. It has a double row spherical roller bearing on both IB and OB sides. Here is the data 3 days before OB bearing cage failure.

The IB bearing (Fig.1) probably raises red flag by showing random TWF impactive pattern with high g's amplitude. As a result, in the spectrum we see multiple harmonics of 1x and a raised floor. Ampllitudes are also relatively high by commonly accepted standards. No historical data is available.
(Note: I am not sure if one can say that g's or velocity amplitudes are excessive as this mill has a rather flimsy frame)

The OB bearing (Fig. 2) causes less concerns in this respect, but it is the one which has actually failed miserably and unexpectedly (Fig.3). Axial readings are unavailable. Looks like it had experienced high axial stress. The question is: was there any clue in the data?

Would you predict this failure?

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

Mill_4cage.doc (130 Kb, 148 downloads)

The time domain data in figure 1 looks bad. Where's an envelope spectrum? If you want to diagnose cage preblems...

Yes, Duncan, Fig. 1 looks bad, but it is not the bearing that failed. Fig. 2 looks better and it did fail.

Unfortunately, my SpikeEnergy spectrum indicated nothing. I have and old version software with predefined "STD envelope" (whatever it means) and 5kHZ filter applied as collection parameter. Fmax=2kHZ.

I wonder where the axial load, being indicated by the failure pattern and axial vibration on IB bearing, comes from? Shaft alignment was verified.

The mill utilizes an elastomeric Paraflex coupling. Is there a possibility of improper coupling installation causing axial vibration?

Thanks,
David

Did you check for axial clearance in the floating bearing? Loss of clearance could produce high acceleration readings in either or both bearings and not necessarily more in the bearing that lost clearance.

I was told that the floating bearing was installed with sufficient room to float.

Your old Spike Energy Spectrum with 5 KHz high pass filter should work fine for this as the effective filter range, 5 KHz to perhaps 10 KHz should have been perfect for this application. Did you look at the data with a dB or log y-axis scale? Of course, if it was a loss of axial clearance, this could be something intermittent, i.e., something that happened only when the temperature was hot or cold enough and not at other times.

quote:

The question is: was there any clue in the data?

What does the IB and OB data look like with a vertical scale of 0.1 in/sec instead of 1.5 and 0.4 scale? Looks like there is something hiding in the "grass" in the higher frequencies.

Personally I always look at my data first with a fixed scale of 0.1 vertical scale. Auto scale "hides" low amplitude high frequency stuff when there are some large peaks (above 0.1) in the spectrum.

Am I reading the top of the plots correctly, one says Peak to peak and one says peak? Why does the OB spectrum stop at 60K rather than 150K like the IB?

"Hiding in the grass" is why DB and log vertical scales are so useful. To see cage frequencies, you probably don't want an Fmax of much more than 3 to 4 times rotating frequency.

quote:

To see cage frequencies, you probably don't want an Fmax of much more than 3 to 4 times rotating frequency.

Good point Duncan. Although sometimes, as we all know, cage shows, whether it's condition is good or bad, in higher frequencies as sidebands around something else as well as in the waveform. This is what I was referring to as "in the grass".
A cage is, as a rule, hard to see in the fundamental range, not impossible, but difficult.
Only my opinion and I could be totally playing left field with no glove.

Here is the plot of the failed cage bearing (Fig. 2) in Db scale. Does not look like there is something pointing to the cage. (Ralph, I have also corrected the p-p issue in Fig. 1)

And if you think about the type of cage failure (the cage just got ruptured axially, not worn out, it is unlikely to expect cage frequencies in the spectra.

I have strong suspicions that it had occurred because of the bearing, which supposed to be floating, actually did not float. Any ideas on how to verify "floating - not floating" situation once the housing cap is removed?

Thanks.

Mill_4cage.doc (58 Kb, 54 downloads)

quote:

the cage just got ruptured axially

Hi David,

Hmmmmm! I am confused on how a cage can explode due to axial thrust of the bearing being unable to "float". Seems like the rollers contact with the races would not allow such a force to occur on the cage.
It may be "plain as day" but can someone explain this?

Also, See the attached plot and question.

Forgive my constant quest for reasons, but I guess that is the way I have always been. When something of mine fails, I always search for what I DID WRONG TO MISS IT, if anything. Upsets me to miss something.

I have had the interesting work the last 8 months to sort out flimsy hammer mills at a couple of sites. Until the various rotor problems was sorted it was found several times on brand new rotors that the clearence was totally off, one brg, tight and one loose despite promise from mech. guys it was ok at assembly. Also axial position was wrong. Seems those problems stopped when rotors was produced to have acceptable vibrations. Can´t say why that would be the case but it is an observation. Olov

Ralph,

The mechanism causing cage stress and following destruction of the cage as oppose to simple excessive wear was always a puzzle to me.

Based on the failed cage pattern I just assumed that there was an axial force applied to the cage. Vibration data also showed high axial vibration. Calculated axial shaft thermal growth is about 0.2inch. One of the scenarious could be a stuck floating bearing causing shaft to bend. Right before the bearing failed the temperature went through the ceiling. The bent shaft could cause bearing to experience excessive axial load and loss of clearance (cold bearing had normal clearance), loss of normal rolling, etc. I don't think cage fault frequencies can be expected with this type of failure.

It would've been good to discuss mechanisms of stresses that cause cage failures.