Yesterday, I was collecting data on a dryer fan and looked at one of the fan bearings in the poorly lit area and thought I saw a loose set collar. After getting my strobe for a better look, I found that it wasn't a locking collar I was looking at, it was the inner race of the bearing. There were no balls left, a piece of the cage was cutting into the shaft and the inner race had fallen down onto the outer race, worn through the outer race, through the housing, and through the support steel. All this and the fan was still running-AT 3550 RPM!!! I can make some guesses as to why it didn't cause a fire or weld some components together, but they would be just guesses. I feel certain that this has been causing some undetected product contamination for quite some time, since there is very little evidence that the balls ever existed.

These bearings are inaccessible (except with a stinger) due to guarding, but I had called this bearing as having wear during my last visit in August. I took another reading yesterday and am pretty sure that the spectrum will show a rub.

I'll post it later today.

Anyone with clients who insist on a 6 month period for readings should feel free to use this as evidence against going that long. I know I certainly plan to.

Amazing that it was still running.

It will be interesting to see those spectra if you get a chance. (and the old ones if you have them). Also if you get a chance to take a picture I'm sure it will be impressive.

Pete,

Sorry, no pictures. It was being repaired before I left and I didn't have my camera with me. It was a real data dog day and I was traveling light.

I'll post the spectra as soon as I finish dumping the routes. About 2000 readings, so it will be a while.

Who want's to make a guess as to what the spectrum will look like?

Velocity spectrum was collected with a 9" stinger to famx 70.5 orders with 1600 lor and 5 averages. special twf in accel to 80 orders with 1024 points. Corrected.

Danny

This message has been edited. Last edited by: Danny Harvey,

Hmmm....no bearing tones left and a cured bearing problem?

Jon
Spintelligent Labs

I have pictures of it.

Well, one just like it.

In this case, I had already called the bearing, and we had switched to the back-up. But for some reason the BAS kept turing the other fan back on.. each month I would find the lag not running, and the lead on again. I would tell them to switch it again.

The warped ribbon in the pic is not a seal, it is the outer race.

The picture was taken after the housing was removed, but the housing was all that was left.
The balls were on the floor.

Martin,

That was bad, but mine was worse

All the way through the outer race, housing and support steel at 3550 and still running. No sign of any balls except some shiny powder. It feeds an injection molder, so I expect they wound up in the product as powder. They couldn't have passed through the molder without being noticed. It does look like yours is bigger though.

BTW, I called this a Level 5 (immediate shutdown) in August, 05. Maybe it would be better if they actually read my report.

Also, I am ready to post data, but will wait a while to see if we have any guesses.

That’s another ugly one Martin. Gruesome failure.

My guess is Danny had a very high raised noise floor with lots of random imapcting in acceleration TWF. Possibly 1x with harmonics.

Next guess?

pete - good guess. Maybe not so random impacting. Perhaps the noise floor would have had a gaussian distribution around zero frequency.

Duncan,

Can you explain a gaussian distribution around zero frequency better?

Thanks,

Danny

I am in agreement with Pete as far as vibration pattern is concerned.

Danny,

Considering a 9" stinger I wonder why
Fmax=4200 HZ has been used. Even a good bearing may drive such a sensor into resonance. This bearing is definitely bad but your data will likely show it is even worth.

David,

It is a long story, but let's just say that I am not paid to cover the fan bearings, just the motors. They have a 1:1 v-belt drive and I added the fan bearings just so that I could tell where the imbalance was. The stinger is the only way to get to them and the ap set is whatever RBMWare assigned when I added the points.

If you only got what you pay for, they wouldn't have been able to ignore my August report.

Danny - adding a v belt drive to the info suggests that the shaft at that bearing location would probably make it more like a rub, perhaps a rub modulated by the assorted belt and pulley related frequencies.

For an example of a gaussian distributed noise floor, look at the fluid film bearing wear example in section 8a of:

section 8.2 in:

http://vibrotek.com/article.php?article=articles/intelect-eng/index.htm

This is one of the methods for diagnosing when rolling elements an their last stage when rolling elements are no longer acting as rollers.

Duncan,

Thanks for the link. I'll read it as soon as I get time.

Sorry, I have should have included the v-belt drive info from the start.

I'll also add that the fan side bearing is the one in question and that it is one of the most acursed breed-the hot (200 deg f), fast, overhung, single inlet, centrifugal fan.

I doubt if the resolution is very good, so I'm not sure what sidebands will be visible. After all, how long will you stand beside a broken bearing that could come flying off at any minute, takinig readings, for free.

Danny

Out of curiosity - grease or oil? Any clue as to what factors led to the failure (other than repair recommendation was ignored)

Pete,

Grease lube.

My guess would be heat, high speed, imbalance, looseness and lack of proper lube combined to kill this one. But is it really a failure? It never stopped until they chose to stop it.

These fans are integral to a plastic dryer that is sitting on top (barely attached to) of standard bar grating. Not an ideal installation in my opinion.

Danny

I'm going to attach some data at 4:00 Eastern, so any other guesses shold be posted before then.

Good call Pete!

The first page is my report from August. Names removed to protect the embarrassed.

The following stuff is from yesterday.

gd06.doc (290 Kb, 61 downloads)