Hey guys, I've got a 3600 rpm fan with some unusual symptoms (at least unusual to me)and would like some opinions on what to do next. I've included all of my information with amplitudes and phase in the attached Word document. This fan was just rebuilt and the fan impeller, shaft, and drive coupling were all balanced in our machine shop. As is stated in the document, the alignment appears to be within tolerances. I have bumped the structure and the impeller and there does not appear to be any natural frequencies around run speed. A coastdown did not reveal resonance as a cause either. The phase on the bearings has them close to 180 degrees out of phase in the axial direction. This led me to believe that maybe we had a bowed shaft. I indicated the shaft today, however, and this does not appear to be present. I'm curious about what you guys think. Bring on the questions.

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

Fan_speed_is_3600_rpm.doc (732 Kb, 189 downloads)

Billy,

Although you have not provided spectra (would be helpful), this is likely just an unbalance problem IMO. Did you run the motor solo?

Sorry about not including the spectra. I've sorta been rushing around this evening and I guess it slipped my mind. Pretty much all of the spectra is dominated by run speed. Check out the attached Word document for the data.

Sorry that I don't have spectral data for the vertical and axial planes on the fan and the vertical planes for the motor. I have them somewhere but can't seem to find them now. Not that is any consolation, but the spectra for these points are extremely similar the one's in the attached document.

David_G, I would agree with the unbalance scenario except that the phase for the bearings is not pointing to that. Shouldn't the bearings in the axial direction be in phase if this a balance issue. My machinist are assuring me that this fan was balanced correctly in the shop. I did run the motor solo and it's fine. Hardly no vibration present at all.

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

M1H.doc (56 Kb, 104 downloads)

When you took the axial phase data, was the accelerometer oriented in the same direction... if not, there's your 180 phase difference. Did you take the axial readings on the same side of the shaft?

If you actually have a 180 difference, that's very unusual for an overhung fan.... I've never seen that before.

For a fan running 3600 rpm, it doesn't take much imbalance to cause excessive vibration. If the shaft, wheel, & coupling wasn't balanced as an 'assembly' you can have significant imbalance when all the components are mounted. Even an improper coupling key length, or a fan hub key, could be the problem.

Often the easiest "dignosis" is just to field balance it and see it it responds as it should.

Rusty, the axial phase readings that are noted with the horizontal and vertical points in the attached document were captured at the top of each bearing in the same direction. They are 180 degrees out of phase, or somewhat close to it. That's what has me so confused. Also, in the attachment, the bearing diagrams at the bottom were all collected at the points shown and also in the same direction (away from coupling).

The impeller, shaft and coupling were balanced as a complete assembly with all of the same parts. The only part removed for fan assembly was the coupling which can only be reinstalled one way.

I have tried to balance this fan, but I can't seem to get the 5% change that FastBAL is looking for to give me a correction weight. I have added some very menacing weight on this thing too. It has already scared me once so that's why I wanted to get a second opinion. It was so bad that the accel wouldn't stay still. I'm sure some of you guys may have seen this in the past, but I haven't and it was somewhat alarming.

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

Billy,

Now after I looked at the spectra I'd have again verified good shaft alignment. What kind of coupling is there?

Are both bearings bolted to a pedestal base with a common top surface? What kind of bearings? Spherical rollers with taper adapters?

I'd mount a dial indicator grounded to the base aways from the bearings and measure axially at the top of each housing, similar to the vibration measurement points, while rolling the fan by hand. I'd kind of expect that one or both bearing races are installed cocked on the shaft, so that the shaft is shucked axially 1X. 1 bearing pushes, and the other resists, with the result that the bearing housings move in opposite directions. The same indicator referencing the bearing inner race face should make it obvious with runout.

Shaft alignment was checked twice. Once by the mechanic and once by myself. That's not saying much but it did check out pretty good. Less that .003" for both parallels and less than .002" for both angulars. This was done with a laser (ROTALIGN) and I've always had misgivings about lasers(not because they don't work, but because we don't know when they don't). I sometimes think we put too much faith in them.

David, the coupling is a Woods Type 10.

Dan, sound like a good idea about checking the bearings for squareness to the shaft. Will run that by the mechanics on Monday. Yes, these bearings are bolted to a pedestal base with a common top surface (picture below). The bearings installed are Link Belt P335 pillow block bearings.

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

Billy,

Fans and blowers operating at 3600 RPM all too often have very high vibration due to structural resonance. From the photos it appears that the fan and motor are mounted on relatively thin flat plates. These plates will have bending, twisting and circular modes if the natural frequencies are excited to vibrate. It may be that the plate supporting the fan bearings is resonanting at the 1st circular mode which would cause the bearings to move out of phase as the plate moved up and down in the center.
The motor support plate may have a 2nd bending mode natural frequency at run speed which would cause the motor to have a rocking motion.

Amplitude and phase data at several points on the mounting surfaces should tell you if this is what is happening. The fix is adding stiffeners under the plates to detune the modes. Using much thicker mounting plates, minimum 1", will usually eliminate these type problems. This would be a good application for operating deflection anlaysis (ODS)

According with the attach blog, wich is a resume that I wrote from a Ludecas paper, your problem look to be Dynamic unbalance (a combination of force and couple unbalance).

It refer to :
1.- Radial fase difference between 0º to 180º
2.- Vertical and horizontal fase difference around +/- 30º
3.- Is suggested multiplane correction

http://vibrobalance.blogspot.com/

Billy,

Has this type coupling always been on this machine?
Personally I prefer a grid or gear type on a 3600 rpm machine, but that's just me.

I am not saying there is misalignment at the coupling, but with such high amplitudes in the axial direction of the fan and motor, there might be something wrong there. Has the end clearance between the coupling halves been confirmed? Is there any 1X "impacting" in the waveform in the axial direction? One thing using a laser alignment, it does not know whether there is coupling to shaft runout or not, nor can it see a bent shaft or any other problems that might be wrong in that area. It only aligns the center lines of the two shafts. Have you checked for any problems with the hubs and shafts?

In your picture of the fan, I notice how close the coupling is to the frame of the fan, was the laser mounted on the fan shaft or the hub of the coupling? What about the motor, shaft or hub?

[quote]I have tried to balance this fan, but I can't seem to get the 5% change that FastBAL is looking for to give me a correction weight. I have added some very menacing weight on this thing too.{/quote}
Are you saying Fastbal will not give you an "answer" for the correction weight? Absolutely refuses?
Which direction are you trying to balance in?

There may be two problems, one in the fan balance and another in the drive end. I would run the motor solo and make sure it is ok. What about the pedestal? Is it moving in phase with the bearings? Is it tight on the floor?

I had a difficult time understanding the phase across the coupling in the axial direction from the pictures. Sorry. Is it in or out?

Quote posted by Ralph
ote]What about the pedestal? Is it moving in phase with the bearings? Is it tight on the floor?[/quote]
Would by interesting to see a shut down, if pedestal is not tight on the floor the amplitude decay response is proportional (named dry damping), not logarithmic.

There are lots of good checks mentioned, but unless I missed it, there is another couple to verify. I see both bearings are cam-lock pillow blocks that are slid onto the shaft and locked by set screws. First, care should be taken so as not to create runout by the way the bearings are mounted. A dial indicator can verify runout of the shaft at the bearings. Second, it is generally accepted that one bearing is held while the other is floating to allow for any shaft expansion. Bearing manufacturers make pillow blocks that are both fixed and floating with a code identifying which is which. Don't know how much heat is there, but it is a consideration.

Gary B

Ken, will a circular mode at the bearing pedastal be readily visible with a bump test?

Ralph, as a matter of fact, no this is not the original coupling. This Woods coupling was my idea. The original coupling was a Falk gear type. I have not had alot of experience with these couplings and thought that maybe it was not a good idea for the application. We decided to try this Woods due to a good history with them. The original Falk gear coupling had a fair amount of axial play. I could shift the coupling all the way in one direction and roll the shaft by hand for a few revolutions and the coupling would "walk" to the center of the axial play. I felt like this was a good indication of misalignment, but when the misalignment was checked, everything was fine. I'm not sure what you mean about end clearance between the coupling halves. If you are refering to whether the rubber insert has any axial play, then yes it does. I didn't notice any impacting in the axial waveforms. I connected the mounting hardware for the laser to the shaft and not the coupling. I learned long ago that you can never be sure if the coupling is concentric or not. Fastbal would not give me a correction weight at all, just a warning that the there was not enough of an amplitude or phase change. The motor has been run solo and there appear to be no problems with it. I don't have any phase data for the for pedastal, but can gather some tomorrow. The motor inboard axial and fan inboard axial are ~100 degrees out of phase.

Gary, I wondered the same thing about the held and floating bearing, but noone from maintenance could answer this question for me. I assume that both are held and this could end of being a bad thing.

[quote]I wondered the same thing about the held and floating bearing, but noone from maintenance could answer this question for me.[quote]

MAN!!! That doesn't look good for someone in maint., not knowing if both are held or both floating or one of each!! WOW! Where was the leadman!! Can the bearing tag numberrs be read while it is running?

Another question that could be asked is, is the vibration and phase the same when cold as it is when hot(when the floating bearing must perform its duty) and are the bearings running hot due to not having room to take up the expansion?

As far as "Fastbal" goes, what were the amplitudes and phases when you added weights? Which direction were you balancing in? Were you single plane or two plane? No 5% change and being scared when weights were added sounds like maybe "fastbal" is asleep at the wheel.

Bearing tag info showed that the bearings were 2 3/16" P335 Link Belt bearings.

Not sure about vibration and phase from cold to hot, but this fan runs rough from the get go so I would think that it doesn't change (only and assumption though).

Not sure about the amplitudes and phase angles when I was balancing. They are still in the analyzer at work. I really hate to second guess my machinist though. They balance stuff all the time at the plant and have never had any complaints. The balance machine was just recalibrated too. Of course, anything is possible as far as mistakes are concerned, but both of these guys checked it after one another and they both came up with the same results. I actually had the mechanic on the job get very angry with me because I told him the fan shaft, impeller, and coupling were already balanced and there was nothing else that I could do. He said, and this is where the light began to be shed on the subject, that "The guy that did this job before you always had to rebalance this thing after we rebuilt it. No matter how many times the machine shop balanced it, it was always rebalanced in the field." I tried to explain to him that it didn't make any sense to me that if an assembly is balanced in a machine shop to whatever tolerance they used and then taken to the field and installed, would somehow mysteriously become unbalanced. It just didn't add up to me. And the fact that the guy before me had so much trouble with this same machine is leading me to believe that there is something else going on. Of course the phase angle descepancies from the bearings and the fact that they are moving opposite of one another only, from what I have picked up in training, compounds the fact that balance doesn't seem to be the issue here. I have been wrong before, however (have lost count there), and would love to be proved wrong.

I was trying to single plane using the horizontal and verticals for both bearings...two measurment planes and one weight plane.

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

Well!

Seems like you are into it!

As far as rebalancing after installed in the field, that is no newbie. To question your machinst would be critical. I have seen things balanced in the shop and when installed, feel like they are coming off the frame. No option but to rebalance. The stiffness, mass and all those high tech terms comes into play a lot of times.

If the amplitudes and all are the same from "the get go", then floating bearings would not be in play yet, in theory. But if the bearings are mounted in a "pre-loaded" condition against one another and neither is floating, then that may be a problem.

Have you got access to the "other guys" records of balancing this demon?

quote:

I could shift the coupling all the way in one direction and roll the shaft by hand for a few revolutions and the coupling would "walk" to the center of the axial play. I felt like this was a good indication of misalignment, but when the misalignment was checked, everything was fine

Are you talking about this Woods coupling or the old Falk? Being a 3600 rpm and a very loose fitting coupling to shaft and only held with a set screw, is not the best game plan there is for a higher speed machine, in my opinion. I would check the run out on the coupling halves.

I think it may be more than one problem involved here.

I am a little puzzled about the balancing program not giving you permission to install a correction weight. I have never seen that before. Or maybe I have always forced the program to give me an answer of some kind and let me decide if it were good or not. Show us the numbers from the meter and see if we can see what is happening.

But like I said, balance may not be the ruler here in this case.

Of course this is only my opinion and I could be totally wrong one more time.

At first glance, high 1x amplitudes almost always indicate that imbalance is at least part of the problem. I wouldn't worry about the 'shop balance' -- just balance it in the field if possible. The 180 axial phase difference is very unusual and probably indicates a "cocked" bearing. It's not easy to tell if these bearings are square to the shaft. First take an inside measurement from foot-foot on both bearings. This tells you if the bearings are sitting parallel to each other. Then mount a dial indicator in a magnetic V-base and attach it to the shaft near the bearing. Set the indicator against the inside face of one of the bearing feet and note the reading. Then rotate the shaft 180 degrees and indicate the other foot of the same bearing. Even with machining/casting errors of the bearing housings, the readings should be within a few thousanths of each other. Repeat this procedure on the other bearing. Self-aligning bearings have a much smaller margin of error than is genearlly thought.

The Woods coupling in use will work OK. If it's new, a good straightedge (Starrett 6" scale) laid across the hubs will verify alignment to within about 0.015" which will allos this coupling to run.... it'd be better if it were much closer than that, but I think even that much misalignment is not going to cause the high 1x you are seeing.

My vote is "cocked bearing".....

As far as balancing, I'd be glad to look at your numbers and see if anything looks amiss. I've been using FastBal for 17 years now. I can also give you a trial weight and location that should get you at least a 50% reduction in vibration if you know the wheel weight and diameter. What does the coastdown data (monitor amplitude/phase) look like?

I'm actually glad to hear that balance shouldn't be completely ruled out. Even though my guys are agreeing that their balance techniques are working, there was still a small part of me that was hoping it was a balance issue so that I could find a cause for the problem.

I will admit also that I have not had proper training to balance machines. The extent of my balance training was a once over of the software and a general explanation of the process. I have successfully balanced overhung fans in the past, but maybe it was just luck. If luck has anything to do with, then it's sure rearing is ugly head at the wrong time because I'm scheduled to go to a field balancing class next week. If only this fan could have waited until then.

I have a contact that works with Rexnord/Link Belt and I think I'll give him a call about the fixed and floating bearing issue. Maybe he can explain what can be done to make one of these bearings a floater.

Ralph, the coupling I was referring to was the original Falk. I haven't checked runout of the coupling yet, but will try to do that today.

I will post the numbers from the analyzer for the initial run of the fan when I get to work this morning.

Rusty, I have measured from the inside of one bearing foot the inside of the other bearing foot and they are very close. Less that 1/32" out. I didn't have access to a V block on Friday to check fan to bearing squareness, but I got one from my shop here at the house that I'm going to take to work with me today....I'll take a look at this today when I get back to the fan. I did mount a mag base to the bearing pedestal and checked axial runout of the bearing at the locking collar. There was a pretty substantial amount...maybe .010" for each one, but I wasn't sure if this was OK or not because it is only a locking collar and I felt that it could actually be this much out but not represent the true condition of the bearing. I didn't have a way to measure anything on the bearing itself....no actuall bearing surfaces available...only the cast pillow block housing material. I'll also check the readings from the shaft to each side of the bearing feet to see if they are within a few thoundsandths of each other.

Rusty, I do have coastdown peak and phase avaiable. I will post this data also once I get to work.

My balance job is set up as follows:
Two measurement planes - IBH/IBV,OBH/OBV
One weight plane - 12 discrete positions w/weight placement radius at 5.75 inches against rotation.

Here are the results from the reference and trial run with Fastbal.
Reference Run
IBH 3571.2 2.8657 mils 134.4
IBV 3571.7 0.0355 mils 16.2
OBH 3572.6 8.7235 mils 255.5
OBV 3571.9 0.0343 mils 15.8

Here is the warning message that Fastbal is giving me after the reference run:

"USER ALERT !!
Reference data at 1x RPM is <50%
of total vibratory energy at:
Plane 1:IBH - 24%
IBV - OK
Plane 2:OBH - OK
OBV - OK

There is an indication of substantial vibratory energy due to other causes. Correcting unbalance will primarily reduce the vibration magnitude of the 1X RPM frequency peak. Acquire an FFT at the points using the same sensor setup as the Balance Job to determin what other vibration frequencies are present."

I decided to continue with adding the trial weight to see what affect it would have. I added .4 oz to discrete postiion 2 and .2 oz to discrete position 3 at a radius of 5.75 inches. These are the results:

Trial Run
IBH 3571.0 2.5441 25.3
IBV 3570.8 0.0158 8.8
OBH 3569.7 5.7150 142.4
OBV 3572.9 0.0259 13.4

Then I get this message:

"User Alert !!
Trial Run Data End to End 1X
Phase Difference >= 100 Degrees

Plane 1:IBH
Plane 2:OBH

On a two bearing single weight plane system, if the phase of the Trial Weight Vector on the first bearing is 100 degrees or more from the Trial Weight Vector on the second bearing, then the pivot point of the rotor is between the bearings. Thus the unit will be difficult or impossible to single plane balance. If balancing does not satifactorily resolve the 1xRPM vibration, consider performing a two plane balance instead of a single plane balance."


P.S. Rusty, I can't find my coastdown peak and phase data. It's quite possible that I didn't save it.


I'm taking this to mean that that there are other peaks in the spectrum that has Fastbal thinking that there are other problems.

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