First some more comments about runout.
Attached I tried to find out the change in movement if we have an 11.8" diameter fan sheave with 0.015 mil face TIR and 54.4" center to center distance.

I calculate that the distance from center of bottom sheave to top of the tilted sheave (in the neightborhood of 55.4"+11.8/2 = 60.3" changes by 0.000004". That is four millionths of an inch. It doesn't soudn significant to me. 0.3ips at 2400cpm is somewhere around 0.002" pk/pk movement (measured at a different location = on the bearing housing... but how do you get that much movement originating from a tiny 0.000004" stretch of a 60.3" section of belt)

Just my thoughts. But I am open to comments.

NNcalcs.xls (21 Kb, 28 downloads)

Did you say which end of the fan has the highest 2 x amplitude in the horizontal?

BTW, the design of your fan very closely resembles the one I posted with the broken shaft. Except it was drawing air through a filter on the outboar end. Yours appears to only draw from one, correct?

What about the horizontal amplitudes on the vertical frame just below and under the left and right horizontal frame pieces the outboard bearing, anything there as compared to the bearing?

Scott - keyways play a role in my theory.

Here is my theory, (again, just my theory... may not be the right answer).

Did you guys see how long the keyway was in that photo posted above? Very long. The sheave keyway also very long. Must be they didn't know where the sheave and fan would be mounted when they built the shaft so they left a lot of room for adjustment. Those long keyways magnify the effect of the assymetric stiffness compared to shorter keyways.

Also, remember above when I said that we changed the belts and the vib started out high and then got lower a day later and even lower another day later. What happens after you change belts? ... the new belts loosen as they wear in. We have in fact found that the 2x vibration is very sensitive to belt tension. High belt tenion -> high 2x. Low belt tension -> low 2x. We had the luxury to do a few trial runs to verify that behavior and it is verified (still one more set of data will be coming in and I will post the full results when I have them).

So, my theory was that the assymetric stiffness due to the keyway causes the 2x. And it would naturally be more evident when belt is tight because the belt is pulling against that shaft stiffness (I guess I am assuming it is as a far-below resonance static type of problem in that respect).

The dial indicator readings linked above were an attempt to validate the theory. But they didn't particularly validate it. The external keyway (fan sheave) is marked in those drawings. I don't know where the fan hub keyway is in relation (it is not accessible now). The position of the dial indicator for the sheave rim check at 45 from vertical is exactly in-line with the belts. If the belts were causing a big bend when the sheave was aligned, we would expect to see a shape of deflection varying twice per revolution. But I have plotted those measurements on a circular plot (tab "Circle Plot"), and the major pattern that appears is a simple once per revolution variation.

I took it a step further and did spatial fourier analysis to decompose the dial indicator measurements into their frequency componetns. Overview of those results is in the tab "spatial decomposition". Then in the tab "filter..." I subtracted out the dc/0-frequency component and the 1x component. What is left sort of weakly resembles a 2x, but it is much much smaller magntidue. Now maybe this is due to the facts that:
1 - the 2x is due to stiffness -> sensitive to force applied
2 - we might have a lower force during static check than during running.

#2 is an unknown to me. I have to work through the formula's to try to estimate tensions and forces in static and running. I have the vague idea that there might be twice as much force on the fan sheave during running as static...doesn't seem like big enough difference to explain how the 2x variation during runout check is so small compared to 1x IF that 2x is really causing our vibration while running.

Another factor that's a little odd. The dial indicator spikes negative at 90/270 and if you round the curve, spikes positive at 0/180 (yes, it's not high at 0 but I said round it!). Positive dial indicator corresponds to compression of the dial indicator. This behavior is exactly opposite of what I would expect if the stiffness of the fan sheave keyway were causing it. One unknown is where the fan hub keyway is in relation to fan sheave keyway.

So, the dial indicator is not particularly strong evidence for the keyway/stiffness theory.

One thing we do know is the 2x is very sensitive to belt tension (higher when tight). What does it tell us?
That supports the keway theory.
I guess it could also support a belt resonance theory (although I'm still wondering why we wouldn't see fluttering in conjunction with the 2x if there was a belt resonance).
The impact of these runouts is a little question for me since they seem higher than normal. But I don't see how belt tension factors into those.

NN_Curve1.xls (57 Kb, 23 downloads)

quote:

Did you say which end of the fan has the highest 2 x amplitude in the horizontal?

We can only measure the inboard bearing. The outboard bearing is in the plenum...closed up durign oepration.

quote:

BTW, the design of your fan very closely resembles the one I posted with the broken shaft. Except it was drawing air through a filter on the outboar end. Yours appears to only draw from one, correct?

Yes, only draws from the outboard end. Are there some unique features about this fan that might make it susceptible to shaft problems?

quote:

What about the horizontal amplitudes on the vertical frame just below and under the left and right horizontal frame pieces the outboard bearing, anything there as compared to the bearing?

I do know that if you put your hand anywhere on the fan housing you feel vibration. Our vib guy measured it and says it is more 1x fan than 2x fan. Also during the unbelted motor solo run, we had 0.05ips of 2*fan on the motor! It must have been was coming from the adjacent running fan which shares a fan discharge header and shares a base.

quote:

Yes, only draws from the outboard end. Are there some unique features about this fan that might make it susceptible to shaft problems?

What I was wondering about is that the theory from our shaft breakage was from the keyway being cut too long which weakened the shaft. I think one of the pictures I posted showed this, maybe not.

Guess it was not in the first picture. Here is one picture. The kewyway was WAY TOO LONG. This is showing the end that was not in the hub of the fan, but the much too much long keyway.

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

broke.ppt (144 Kb, 80 downloads) broke

Reasonable theory.

quote:

2 - we might have a lower force during static check than during running.

The tension on one side of the belt certainly increases to drive the fan. The tension from driving the fan should be simple to figure out, power/speed = torque; torque/radius = force(or tension difference); approximate what static tension is with the assumption that you can't push on a rope (one side of belt has tension one doesn't).

Pete,

I'll have to give it some more thought because I looked across the room at my TA wall chart and it says 1 x rather than 2 x and your description seems accurate.

I didn't catch your sheave diameters but I assume they are under 12". OD Eccentricity tolerance would be about .010" and side wobble and runout .001"/1" of sheave diameter.

What type of sheave is this? Does it use a QD type bushing or a Taperlok? If it is a QD sheave, is the face of the hub along the axial split smooth on both sides?

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

Pete,

I only see the non-drive end of the fan. Am I missing another photo that shows the drive end?

I had a case in 2001 where I suspected that a high 2 x running speed was being caused by excessive tension on a v-belt drive. In this case, the driven shaft was stainless steel shafting supplied by Redi-Keyed Shafting Co. It had a keyway cut the entire length and came in 8 foot lengths. This was on a 3600 rpm drive.

I couldn't prove it was the shaft, but we really had to fine tune the belt tension to get enough tension to drive the pump but not enough to cause a large increase at 2 x running speed.

I'll try to find the data, but I doubt if I have if anymore.

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

El-Pete,

Regarding page 2 of your PPT: The sheave will rub the belt sideways (axial to shaft) at the 1/4 and 3/4 positions you showed. Does that not constitute a 2 per rev event affecting both sheave and belts? I am not saying that this is a primary cause for your 2xSS vibration. You noted a chenge in vibrations (up/down) when changing belt tension. Belt tension can have several effects:

a) Changes NF of belt span
b) Changes sheave alignment, if mounting has flexibility
c) May change boundry/support conditions affecting fan shaft NF, if there is a bending vibration mode close to 2xSS

Walt

Walt - It looks to me that the axial (sideways in the figure) forcing function created in slide 2 is at 1x fan speed frequency since the direction of the axial push is opposite at 3/4 revolution from where it was at 1/4.

There is also a radial (upwards in the figure) forcing function created at 2x fan speed since the push is min downwards at 1/4 revolution and min downwards again at 3/4 revolution (max downwards at 0 and 1/2 revolutions). BUT, again the magnitude of this radial force seems miniscule as discussed above.

You are right that the belt tension could have a tuning effect on resonance amplifying vibration from any forcing function whether small or large. My gut feel is still that the 2x forcing function from sheave runout is too small to be a factor, but I will try to keep an open mind.

Danny - I will post some more photo's.

Ralph - Did they actually do a bump test on that test before the shaft failure? That would have been good fortune (usually it seems like a shaft failure takes everyone by surprise).

To all - I was also wondering about the applicability of bump test to my rotor. I think since the squirrel cage is overhung from the hub and the hub is not in the axial center of the shaft, then the fan squirrel cage will tilt and create a gyroscopic effect. That means the actual resonant frequency could be much higher the bump test frequency. I think the bump test will establish a lower bound for shaft critical, but not an upper bound. (do you guys agree?). Is it ok to bump it on the sheave? Or do we need to have access to bump it between the bearings.

I notice the fan nameplate states "maximum RPM: 1281". Would that type of nameplate limit typically be based on providing a margin to critical speed? Or something else. (It is an "American Standard" fan. Nameplate photo attached.

I will post some calculations/questions about belt tensions later this weekend.

If you think you have a large asymmetry stiffness in the shaft bump in 2 directions. The frequencies should shift with speed. If you have asymmetric supports the backwards mode can be excited in addition to the forward mode.

If the warning of 1281 has to do with resonance, hopefully they have considered the backward mode, too.

Pete
Thanks for clarifying timing of vibration "jumps".
I was thinking of beats (vs. modulation as I incorrectly called in earlier post). But since very little response from other machines, I agree this is not likely case.
Curious - have you taken time domain data at times when its low vs. high?
I think its a key point that it is not "consistent" - but what that means I too am stuck.

quote:

This is due to the fact that at operating speed the unbalance force is in play causing higher magnitude of shaft deflection at twice per rev rate.

Imbalance causing 2X?

One must know the spring rate for the belt to look at forces on the shaft. The first approach I gave seems a little simplistic.

quote:

Ralph - Did they actually do a bump test on that test before the shaft failure? That would have been good fortune (usually it seems like a shaft failure takes everyone by surprise).

Yes I did Pete. Wondering what was causing such a high 2x in the bearing readings. We had the paper machine down for routine maintenence so I decide to do some test. The outboard end of the fan looked almost identical to the picture you have posted except for what appears to be an extremely flimsy frame in your picture. Shaft was easy to get to with a transducer.

I see I completely missed some earlier responses.

Jason - It is very much a between bearings machine. The fan is between bearings and the sheave is not (very normal). The aspect I point out that I called overhung (maybe it's the wrong terminology) is simply that the squirrel cage is supported by a hub on one sde and nothing on the other side.

RRS Dave - I haven't heard of "inlet differentials" but I gather it's an air effect. Can it cause 2x?

Ralph - your rotor resonant at 2x must still have had an exciting function at 2x. Do you have an idea what that wasy? Maybe the assymetric stiffness from the keyway?

In his book Den Hartog describes the effect of gravity force and unbalance causing 2x.

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

quote:

Ralph - your rotor resonant at 2x must still have had an exciting function at 2x. Do you have an idea what that wasy? Maybe the assymetric stiffness from the keyway?

I am sorry Pete, but I do not what was exciting it. I know after the new shaft was built and installed the 2x was not there.

I'm going to take another stab here along the lines of the keyway allowing the shaft to flex.
Would not a Modal analysis performed on the shaft show the various flexure modes at given frequencies? I believe that this would give you a better indication of the direction of the flex in relation to the keyway versus the dial indicator method. You would also be able to see what the flexure mode of the shaft should be @ 2X TS frequency. The in-sitsu method woud probably be preferred here instead of the free-free since the belt tension appears to have an effect on your readings.
The steel composition of the shaft in question could be slightly different from the like-kind fans thus affecting the stiffness of this one fan shaft more so than the others.

Scott

Pete, I'm late to this dance, and I have not read the rather lengthy discussion yet, but it's a waste of time to even begin to analyze this problem without knowing the belt rotational speed? What is the belt speed?

Normally it's around 1/2 to 1/3 of the motor speed. I usually find the belt speed with the SpeedVue sensor and then confirm with a strobe light. You'll be miles ahead of the pack if you actually put a piece of tach tape on one of the belts to allow time synchronus averaging (TSA).

Yesterday I saw an odd frequency on a sanding roll in a TimeSaver 3-head drum sander. Doing a TSA on the motor, the roll speed, and finally the belt speed clearly showed it was related to belt rotational speed. A bump test on the roll confirmed a resonance at 2685 cpm, vs. the dominant vibration of 2521 cpm which is the 5th harmonic of the 524.5 rpm belt speed. I have no idea why the 5th harmonic is there since this is a 3-belt sheave... sometimes it doesn't take much to excite a resonance.

Long story short, one should have all the relevant data - at least that which is easily attainable - before starting analysis of any machine.

Ralph - thanks. From my thinking, it could be a resonance or critical, excited by the 2x variation in stiffness due to keyway.

Vibescott - I will think about modal analysis.

Rusty - The belt speed is approx 650 cpm as indicated in my post 20 February 2008 06:30 PM. That was calculated from the belt/sheave dimensions and confirmed through the presence of peak at expected frequency in the spectrum. If there is additional data I should be considering, I would be grateful for suggestions.