Had a customer call about a 'screaming' bearing on a 3600 rpm, overhung fan. This was a new fan and had been in service about 9 months. What they described sounded like a late stage bearing failure which was confirmed when I took vibration data... 1.8 in/sec with 80g peaks in the waveform. The 1x peak was only 0.1 in/sec but there was a huge peak that was almost dead-on the outer race defect frequency with multiple sidebands and harmonics. This was just classic. Interestingly, the motor bearings were also bad, with the inboard being in the worst condition. But what would cause this? The coupling is a Dodge Para-Flex with the one-piece black rubber boot (donut). The hubs can be positioned independently of each other, at any angular position you desire. A series of bolts 'clamp' each end of the boot between the hub and an inner ring (one for each side). How would you set this coupling up? Would you align the bolts? Split the difference (so they were not aligned)? Or just not even think of it?

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

View of bearing....

Rusty,
This looks like too high of a thrust load in the direction of the spalled groove. The track appears to be higher than normal on the outer race. Depending on which way this bearing was oriented, this could be a coupling spacing problem, shaft growth problem, shaft or bearing positioning problem, too high speed operation for the coupling (at high speed the “tire” coupling will try to pull the coupling faces together), wrong bearing or under design.

John J

quote:

each side). How would you set this coupling up? Would you align the bolts? Split the difference (so they were not aligned)? Or just not even think of it?

Page 2 here shows installation sequence:
http://www.dodge-pt.com/pdf/catalog/pt_components/2004_...ge_para_flex_fea.pdf

I wouldn't think that rotating the flanges/hubs to keep bolts on each side in-line or out of line would make much difference. The clamp ring clamps the rubber element uniformly to the flange and the bolts just hold the clamp ring on tight. That's just my thought (but I never saw one of these).

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

John, interesting... the axial loading looked pretty obvious to me. I was told the fault was oriented toward the coupling, which indicates the axial load was toward the coupling, right? (I had to sketch this to visualize it).

Pete, I agree... the orientation of the hubs shouldn't matter, but in this case, the original bolts were 80mm long instead of 60mm as called for. They seem to think the bolts were too long and may have been oriented so the two hubs were almost perfectly aligned and the bolts butted end-to-end. I was buying that scenario, but this would push the shaft away from the coupling, and cause the fault to be on the fan side of the outer race, correct?

John, your comment about the rubber boot deforming outwards and pulling the shafts toward one another is very interesting, because the original boot (and the replacement) is only rated for 2800 rpm.... I thought that was just a balance issue, but now I think I see what the rating is about. I took baseline readings on this fan in February and didn't see much... I was thinking that if the coupling bolts had in fact "bridged" the coupling, making it semi-solid, that I'd have seen that in the data, but I didn't. Very interesting....

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

Did you inspect the inner race? I agree with John J about the tire trying to pull the flanges together. When the flanges are spaced properly, positioning of the bolts may not be a problem. At the same time, I've seen the ends of the flange bolts damaged from contact with the adjacent flange bolt ends. The tapered bushing can be installed in a manner that the "draw up" set screws are to the outside making spacing adjustment possible without having to move the equipment. I don't endorse reversing the bushing due to the tendency of force pulling the flange off the tapered fit. Some mech's cut the short end of an allen wrench off just to be able to get to the set screws between the flange faces. These boots come in very big sizes and are sometimes hard to install in the flanges. A common bad practice is to install longer bolts in the flanges to make assembly easier. I think they're good for lower speed equipment. Some think they're great for higher speed equipment. Easy to remove when they tear up! Like changing a tire at an Indy race. For the most part, 1800 and above, I'd set'em up on the "shelf". In reality, proper spacing with the bolts staggered.

Sid, I don't like them on high speed equipment.. they do seem to be more precisely made than the Omega 'orange' couplings which I don't like at any speed. We were given the wrong dimension for the flange spacing which caused the boot to have a large "gap" at the joint. When reset to the proper spacing, the gap disappeared.

I have recommended a Thomas shim coupling which I see on some 3600 rpm ammonia compressors. They are a bit expensive, and have to be more carefully aligned, but done properly, they are zero maintenance. I stress constantly that you don't have to be constantly working on stuff if you just do everything right.

As an excercize in calculation with not much value, we could try to calculate the axial bearing force which arises from centrifugal force in this coupling.

Assume the elastic element is like a cylinder with inner diameter 5” and outer diameter 6” and length 1”.

Assume the elastic element has density similar to water (62.4 lbm/ft^3).

Mass = PI*(Router^2-Rinner^2) * L* Rho
= Pi * (3^2 – 2.5^2) *1 (inch^3) * (1ft^3/12^3/inch^3) * 62.4 lbm/ft^3
= 0.31 lbm (does that sound about right for the weight?)

Centrifugal Force = Mass * Radius * (2*Pi*speed)^2
= 0.31 lbm * 0.25 ft * (2*Pi*60 / sec)^2 * [lbf / (32.2 lbm*ft/sec^2)]
= 340 lbf.

Assume half the centrifugal force is transmitted to bearing on each side.

Further assume force is transmitted to the hub at 45 degree angle. (looks reasonable from brochure). The radial forces cancel. The axial force would be 0.707 times the centrifugal force.

Estimated axial force on bearing = (1/2) * 0.707 * 340 lbf = 120 lbf.

That actually doesn’t sound tremendously high. If you knew the exact coupling model and dimensions you could tweak the estimate of mass and diameter for a better estimate (also check my math while you’re at it.).

If you knew the bearing number, you then could compare that to the bearing load rating (with correction for axial vs radial load) and develop an estimate of effect on bearing life.

It sounds like there are 2 competing theories:
1 – The bolts contacted and created forces pushing bearings outwards. (although this doesn’t agree with what they told you about which side the damage was on).
2 – The coupling centrifugal force pushed the bearings inward.

Even though it conflicts with what they told you, my vote would be for #1 for two reasons:
A - Information that you see with your own eyes is more reliable that info you hear from other people. You can see with your eyes there was something seriously wrong with those coupling bolts. Who knows what was really the position of the bearing before they pulled it out.
B – The calculated force above seems pretty darned small. While we could judge it a little better if we knew the bearing number and coupling size, I suspect even an order of magnitude increase would not give the results you saw on bearings on both sides.

We have Thomas Shimpacks installed on a number of 2-pole 100hp machines. We ignored them for 15 years and didn't have any problems. After 15 years we inspected them and they were in pretty bad shape. In fact one had every single shim in the pack cracked (and we never picked up anything on vibration even when we went back and looked at the data with a microscope!). But they still worked. With a little bit of inspection more often than 15 years, these seem like a good product.

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

Thinking about it a little more, the 45 degree angle means the axial force will be equal to the radial force (not a factor of 0.707 times radial force). If you delete the 0.707 from the above calculation the result goes up to something like 170 lbf (still pretty small).

El'Pete,
If you measured ultrasound emissions from coupling, then you would have detected cracked disks in the Thomas coupling. It works far better than vibfration measurements for this type of fault detection.

Regarding the faulty bearing disassembly: if it was done by someone else and not match-marked to confirm orientation, then you can't trust the description of how it was arranged.

Walt

Thanks Walt. I’ll keep that in mind. We have an ultrasound gun but it sits on the shelf most of the time. Sooner or later we will start taking more advantage of it.

I agree with you on the bearings. Unless given a special request, it would be a rare mechanic who would mark the bearing position and orientation before removal. I think a lot of folks install bearings with the outer-ring letters facing out, but you can't count on that unless you talk to them.

I had some pulp dryer flakt bearings where we were using the same style bearing as shown above. They never held up. I think the load was more than this style bearing could handle.

They were installed because someone sold them to someone for this application to reduce heat since the area they were installed in was a very hot environment. I am thinking the reduce surface contact of the balls to the outer race is thought to reduce heat.

We continued having problems until the bearings were changed back to a spherical roller bearing.

I have some,and had probably 15-20 pictures of these bearings with the same defect wear pattern. Load surface for balls on outer race just wasnt enough, balls kept digging a ditch in the outer race just as shown in the pic above.


I am guessing this fan may find similiar experiences.

my 2 cents.

Mike

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

Walt, I'm still trying to learn what to listen for in a failing coupling with ultrasound. Can you describe the noise difference between good and bad? I've had really good success monitoring temperature in high speed gear couplings, but have yet been able to discern a problem with ultrasound. Is it the same noise that one may hear once the coupling degrades to an humanly audible tone, just lower amplitude? Can you describe? One problem I have is the sound of the coupling bolts passing the receptor tube. What does a worn coupling sound like? Is there clicking, ticking, or cracking? It must be an experience issue and ultrasound for me (as far as couplings go) will need to be explored more. Thanks in advance. By the way, Rusty, the coupling should come with a cardboard spacer gage for each size. I totally agree with you about perference of coupling, and that Pete is one smart knocker!

Sid,

There are several possible ultrasound sound sources when measuring a coupling:
1) Friction -- from insufficient lubrication (if applicable), wear, and shaft misalignment
2) Impacting -- from loose, cracked, or worn parts
3) Windage -- flow noise from air passing over bolts, nuts, holes, protruding gaskets, and keys
4) Background Sound -- either direct path or reflected off coupling surface from same machine (bearings or seals) or from other machines or sources

The best measurement point for microphone is in radial direction at coupling split joint at a distance of about 1". Obviously, coupling guard design and other access issues including safety could alter the microphone location. A remote microphone on a flexible gooseneck is usefull for reaching under a coupling guard. A 3/8" diameter hole in the coupling guard is often adequate for a direct sound path. More than one microphone position is suggested if backgound sound is an issue. If coupling has a spacer or spool piece, then measure at each flexible coupling joint.

Measure ultrasound level (dB) and use headphones to listen for distinctive sound characteristics. Friction has high sound level with no unique sound character (white noise). Impacting has high sound level with distinctive clicking/popping sound. The "normal" sound level for a good coupling tends to increase as shaft speed increases.

I use an SDT-150 meter (39 kHz), I have had excellent results on all types of flexible couplings, including vee belts and chain drives.

Walt

Walt - can I collect the ultrasound data on the vibration data collector for later analysis/review?

Sid – I give up....what’s a knocker?

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

El'Pete,
By your own admission, vibration did not indicate a coupling fault. I know you are an analytical guy who likes to quantify everything, but ultrasound technology, right now, is not that well developed. You are a "smart knocker"; what ever that is!

Here a few datalogging options for ultrasound:
1) Use vibration analyzer as datalogger with Manual entry of dB value and Note Code for audio observations [I have done this with CSI 2120 and MasterTrend, but it is combersome. Why use a $15,000.00 datalogger?]
2) Use a handheld PDA with E-form [I use an iPAQ Pocket PC]
3) Use paper pad for logging and then transfer to vibration database [I have used MS Excel SS for making paper forms and then for plotting data]
4) A few meters on market have built-in datalogging [I am not impressed with this feature unless a note about "audio" can also be logged]

I consider it a waste of time and resources to use a vibration analyzer connected to an ultrasound meter for bearing or coupling measurements.

Walt

El'Pete,

If you're using a UE gun and DataPac, yes you can collect it with the analyzer. I forget right now what the sensitivity I used was, but I collected quite a bit with them a few years ago. The crystal (pickup) wasn't all that good, and the resolution sucked. The newer ones may have a better (read that more expensive) crystal and electronics now though.

Dave

Dave

Was there any grease in there?

Rusty
Here are a couple more thoughts.

1. Is this a taperd bore bearing?
If it is I would be check that they are using the correct taperd sleeve.
2.I would also check if this bearing should have a lubrication groove on the outer ring.
(w 33) I seen the storeroom miss some numbers.

Mark

I'm with Mike H on this one.

Almost every self aligning ball bearing I've ever pulled out has looked like that. If it's on a tapered sleeve, you've only got to breathe on it too hard, and the outer race will look like that. If a sperical roller is up to the speed rating, that would be the go.

I used to see a very similar thing on some 3000 RPM overhung fans (secondary air fans for a boier), they were changed to spherical rollers, no more problems.