We have a family of 4 belt-driven fans. Fan operates at approx 1210 rpm driven by 1790 rpm motor.

In general (historically, among all four fans) the highest vibration on both the fan and motor is at a frequency of 2*fan speed... higher in the horizontal directions than vertical (belt is 45 degree from horizontal). No resonance at 2*fan speed is evident from any bump tests.

The 2x-fan speed vibration tends to jump up and down. 1x and 3x fan speed magnitudes are generally lower. There are not any noticeable harmonics of 2x (i.e. no 4x/6x pattern).

Do you guys have any guesses what would cause this?

Have you checked for belt resonance? I've had a fan exhibit this type of phenomenon in the past. Had a resonance that was exactly two times fan speed. It would pulse like a beat frequency. Tighten the belts and it would go away for a while. Never found a real solution to the problem because the process went down. If given the chance, I'd probably have changed the belt length.

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

I should mention, this is a "quiz" because I know a little bit more information than I am giving you. In fact, I think I know the reason for the vibration (I will provide that info later and you be the judge). But first I thought it would be fun to ask how others would view it.

Billy - thanks for the response.

Belt resonance is something I hadn't thought of and frankly don't know a lot about.

Some info about the belts:
- Belt speed (not resonance frequency) is about 650rpm. We have on on occasion seen high belt speed vibration and harmonics 650, 1300, 1950, 2600).
- We have looked at the belts and on occasions seen some excessive fluttering. I think that corresponds with when we have seen the belt pass frequency harmonics.
- Neither the visual belt fluttering nor the belt pass frequency harmonics seem correlated to 2x fan vibration appearing on motor and fan horizontal positions.
- The highest belt pass frequency harmonic is sometimes 2600... very near 2*fan = 2420. In fact the whole noise floor in this area appears a little elevated. This suggests to me some kind of resonance although nothing shows on bump tests. Could this be indicative of belt resonance? (I don't know).

Is there any way to check for belt resonance other than adjusting tension?

El'Pete,

Have you ruled out sheave alignment? A poor motor base can allow the alignment to change after belts are tensioned and with operating loads!

I have spent a little time with V-belt vibrations. One of the three tension methods is the sonic method. The natural frequency of the belt span is related to belt tension. The frequency can be measured by vibration or microphone by impact method, and it can be calculated from unit weight per length, span, and tension. The two belt spans (top and bottom) have the same tension (hence NF) at rest, but the frequency changes when operating. The taught span has a frequency increase and the slack span has a frequency decrease. I use a microphone close to the belt when running to identify NF. I have also found more than once that the motor structure was resonant and not the fan, so check everything.

Walt

Interesting.
Of all the belt driven fans I monitor, none have the symptoms you mention.
You stated, "The 2x-fan speed vibration tends to jump up and down. ". Do you mean between readings (months apart) or short time (seconds/minutes)?

Is this some type of modulation between 2X-fan and harmonic of belt pass?? (you stated, "highest belt pass frequency harmonic is sometimes 2600.")

Tell us O' wise one :-)

Jim P

It may seem rudimentary, but I’ve determined the “at rest” NF of belts (or a close proximity thereof) by placing my accelerometer on the belts about mid-span and, while in live mode, just give the belts a tap. You’ll get a nice peak at the NF; just don’t hold the sensor to tight against the belts. Be aware though, as Walt pointed out, that as the belts are loaded the NF will be slightly higher on the taught side and slightly lower on the loose side.

Interesting comments.

quote:

Have you ruled out sheave alignment? A poor motor base can allow the alignment to change after belts are tensioned and with operating loads!

We have checked adjusted sheave-to-sheave alignment several times. Often the 2x vibration changes after adjusting alignment... sometimes up and sometimes down. But misalignment should normally give 1x vibration, right?

We do have some buckling on the bar holding the motor tensioning screws used during alignment. Makes the job of aligning a little tougher, but once it's in place at the end of alignment, the motor is locked down with hold-down bolts to sturdy stationary slotted base frame.

We also checked rim runout on sheave, and tried face runout on edjge of sheave and similar runout on the shaft adjacent to sheave. Saw about 10 mils TIR on the shaft and 15 mils on both the sheave rim and face. Seems a little high for shaft runout. But again, I would think that would give 1x, right?

quote:

You stated, "The 2x-fan speed vibration tends to jump up and down. ". Do you mean between readings (months apart) or short time (seconds/minutes)?

Not during a given set of reaidngs. It changes between readings. Most noticeable changes of the fan 2x (up or down) occur after after we have been in there adjusting alignment or tension or replacing belts. In one recent case, the vibration jumped way up immediately after we went in and changed the belts, then was lower a day later, and even lower a day later. During long equipment runs, it tends to stabilize.

quote:

Is this some type of modulation between 2X-fan and harmonic of belt pass?? (you stated, "highest belt pass frequency harmonic is sometimes 2600.")

I don't think so.

I will think some more about belt frequency. Seems like 2400 cpm would be pretty high for a resonant belt frequency from my gut feeling. Maybe I will try to work the sonic tensioner calculation backwards to estimate the frequency from the estimated tension. Or else bump/measure it if we get a chance when the equipment is tagged out.

If there is a belt frequency problem, wouldn't we expect to see the belts flapping? Today we had high 2x fan (0.32 ips on fan horizontal), but belts aren't flapping.

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

I once collected data from a bank of 3 belt driven pumps. One of the pumps always displayed what I thought was a large belt frequency and we tried a number of times to change belts align sheaves, etc with little success. On one particular day I was obviously collecting data on auto-pilot and was halfway through collection when I realised that the pump wasn't even running (please tell me it's not just me who does this), but the vibration level had hardly changed. After a little investigation, it turned out the the pipework was in resonance from a nearby machine and was transmitting into the pump.

Could this be a possibility in your case?

We did do an solo de-belted run of the motor. The frequency 2*fan was still present on the motor, but at a much lower level 0.05 ips. It was being transmitted through the foundation from the adjacent running fan. The high levels of 2*fan are only present on one of these machines when that particular machine is running.

Is the fan rotor overhung type? Is axial reading also high?May be overhung rotor unbalance.

The fan shaft is not overhung, but the fan squirrel cage is overhung (if that makes sense).

Attached is a view standing in the inlet plenum looking at the suction side (outboard side) of the fan. You can see the outboard bearing. You can see through the overhung squirrel cage (no spider on outboard end) to the hub at the inboard end. On the other side of that hub would be the inboard bearing and then the sheave (those are not visible in this picture).

2* fan is high on the motor axial and horizontal. On the fan the axial 2x is low.

ViewFromInside.ppt (826 Kb, 99 downloads)

E'Pete

Ok first you had me then you lost me. Let me get this straight. The fan is between bearings and the sheave is not? At least that is what I get from the picture. But you mention the fan is overhumg. How is this possible when the bearing is outside in your picture???

Weak base. Fan is sitting on springs. DIfferent air velocities (static pressures) causes inlet differentials that will vary and rock fan base at 2X RPM.

Dave

Don't know what might be wrong with yours but here is a case history from a belt driven fan that had a high 2x and finally broke the shaft.

This message has been edited. Last edited by: Ralph Stewart,

2x_before_Shaft_Broke.doc (696 Kb, 53 downloads) 2x before shaft broke

How about non-uniform shaft giving variation of stiffness at 2xRPM?

I'm late getting into this one, but I haven't read anything other than Pete's original post.

My guess would be and eccentric fan sheave or maybe a fan shaft bent at the sheave end. (And possibly hammer marks on the end of the shaft.)

After seeing the picture, I had the exact same reaction as Dave. Is there anything stiffening the bearing pedestal that we can't see? It looks like a long, small diameter shaft, too.

Any hammer marks show up yet?

Gates makes a sonic belt tuner. You tell it the details of the belt drive and it tells you what frequency to tune it to. It's a great way to tension belts, but you need to retension just like with any other method. I used it on an NHRA top fuel dragster to tension the blower belt and it really helped. You can also tape your accelerometer to belts to get the natural frequency. It adds some damping, but not as much as your arm. It definitely changes quite a bit when the motor is running.

If it is a bend at the drive end of the shaft or an eccentric sheave, it should show up highest in the direction of the drive and at 2 x fan speed as the belts tighten and slacken during each revolution.

Do you have any phase data?

El-Pete,

I don't agree with your premise that sheave eccentricity and misalignment would only cause 1xSS vibration frequency. Both situations would pass by two belt spans per revolution. So why not a possible 2xSS dynamic force?

Billy,

Be careful with placing accelerometer at mid span of belts unless it is very tiny. The added mass to the belt will give a lower measured NF than actual. I use a tiny (2-grams) accelerometer placed on belt (with petro wax) close to the sheave, so mass-loading will not be an issue. A microphone works well also.

Walt

Ralph - shaft crack is not an encouraging possibility. But we have it on 4 sister machines, present since day 1. There are variations up and down but no long-term increase. So I don't think it would be a shaft crack. But how would one check this?

Dave - what would give a non-uniform shaft stiffness ? (other than cracked shaft). (btw - your theory is close to mine.)

Danny: "If it is a bend at the drive end of the shaft or an eccentric sheave, it should show up highest in the direction of the drive and at 2 x fan speed as the belts tighten and slacken during each revolution".

We do in fact have fairly high runouts on the sheave and the shaft drive as mentioned above. The exact readings are in slide 1 attached.

I am trying to imagine how sheave tilt or shaft runout might result in 2x. If Walt or Danny or anyone can explain it I would appreicate it.

The only thing I can think of where sheave tilt gives 2x is the scenario shown in slide 2: If the sheave is tilted about a diameter-line through center, then twice per revolution that diameter line will be in-line with the belts and there will be max tension (L=C+D) . Twice per revolution, that diameter is perperndicular to the sheave, the top of the sheave is tilted away from the belt, and there will be min tension (L<C+D). BUT, if you do the geomtery of a 10 mil tilt accross a distance of 55", I think this effect would be miniscule.

I assume you are talkign about phase data at 2x? Do we need a 2-channel analyser to do that? Would it be worth the trouble?

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

TiltedSheave.ppt (162 Kb, 61 downloads)

If memory serves me right, I remember Art Crawford saying something about 2X being inherent on center hung fans due to the shaft flexing because it is actually weakened where the keyway is cut in the shaft, but I would tend to think that the amplitude would be fairly stable from read to read.
Have you monitored this in the live mode to see if the amplitudes are jumping up and down at this frequency? If so, I would tend to lean towards a resonance somewhere as some of the others have stated.

Scott