Thanks Danny. That's some good info to think about. Ultra-V I assume is some other vendor's designation for the "narrow" belts (3V/5V/8V). Gates call this their Super HC series. I think they are all the same as long as they say 5V without another letter letter after the V (like VX, VP). I remember reading that three of the narrow belts do the job of 5 conventional belts. I guess if the tension is 167% higher on 3 of the narrow belts, it would still give the same pulley loading as 5 of the conventional belts.

I have a hard time selling our guys to use laser alignment tools on V belts. Most of the literature I have seen gives a tolerance around 0.5 degree for sheave alignment of V belts (much tighter for sync belts). Our skilled guys say they can get that using straight-edges or string. In round numbers that is 1" per 120" length, or 0.5" per the 60" center distance. I think they are probably right that lasers aren't needed to meet that. Should we be shooting for better than that?

=================

Back to our trouble machine:
Attached are the spectra for the fan/motor 22 which is the one that has higher 2x Fan than the others, and appears possibly increasing.

The order of the slides are: 1A Feb 08, 1A Feb 07, 1H Feb 08, 1H Feb 07....

So as you page through the slides every other slide is recent (2008) and every other slide is old (2007). Also the positions move from motor outboard (1) to motor inboard (2) to fan (3). As I mentioned we don't have access to measure fan outboard bearing because it is in the plenum box.

Harmonics of motor speed are in green. Harmonics of fan speed are in red.
Harmonics of calculated belt pass are in blue. And in purple, we have odd harmonics of belt pass (i.e. 0.5*belt pass, 1.5*belt pass, 2.5*belt pass). That's a little strange (does anyone know if it means anything?) , but we had some of those purple frequencies in 2007 as well. The big difference I notice from 2007 is that the 2x fan is now much higher, even though we have detensioned belts to the minimum..

===========
Edited to add - there is a frequency 3729 which shows up on the fan positions 3H and 3V. It is BPFO for the fan bearing - Link Belt 339. But the BPFO harmonics of this do not show up in the higher frequencies and no other bearing patterns evident in the high frequencies at least out to our Fmax of 30kcpm (I know... that's high enough... we will look a little more). And our peak acceleration TWF is relatively low (<1.5 g's true pk).

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

NN22_0to10K_VS_HISTORICAL_small.ppt (1,240 Kb, 23 downloads)

Attached is some additional info which is not essential to understanding our problem machine #22. (The spectra above are probably more relevant). It is just general background for anyone who really wants to dig into it.

It is waterfall history for all four of the machines. The first batch of waterfalls are for 22. The next batch for 21, then 11, then 12.

The recent history of 22 in Feb 08 is hard to see on these waterfalls, because we had a period between May 07 and early Feb 08 when the belts were tight and the vib was very high... it hides the most recent data (Feb 21 08) due to the difference in magnitude. That's why I created the powerpoint in the previous post so we can easily compare how it's doing now to how it used to be doing.

If you are already in data overload, I would recommend skipping this file.

NN_WaterfallsFeb08_small.ppt (3,374 Kb, 12 downloads)

Gates has a nifty belt design program you can download for free (I think you have to register on their website though). It lets you design a belt drive system very quickly and includes component costs. Or you can model the installation you already have. Also calculates belt tension, etc. I have not installed the "pro" version yet, but the older program I have works great!

Gates DesignFlex Pro software

Pete, in my mind "allowable" doesn't really mean anything. Allowable runnout on a sheave is kind of like "allowable misalignment" for a coupling.... who determines whats allowable, and how? If eccentricity creates a problem, it's too much, no matter how small the eccentricity is.

The sheave runnout on the driven roll I described was 0.007" TIR. For a 60 lb. sheave that results in 95 gram-inches of imbalance (60 x 454 x 0.007 x 1/2). I used 26.5 grams @ 4.75" radius to balance it... that's 125 gm-in which is fairly close. Note the weight is not directly inline with the shaft bend, but it's fairly close. And there was probably a little imbalance left in the roll (from the roll shop) or the sheave was probably not perfectly balanced.
Click this link for Photos

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

Rusty
You balanced sheave runout? That cures the unbalance but didn't you still have pulsing forces on belts?? or was that not a big problem on this machine?

Jim P
p.s. Thanks for the link to Gates Belts. Handy

Runnout often balances easily, but yes, there was a 'pure' 2x left. Sheave runnout can and often does give you a pure 2x vibration.

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

Aha. I was picturing that a roll was driven by some kind of very wide belts around a wide section of the roll (I don't know where I got that idea... maybe I was confusing the product belts/sheets with the drive belts). But from the pictures, the roll is driven by a a vee-belt sheave very similar to ours. So, I guess it does provide a valuable reference point to support the notion that sheave runout can cause 2x vibration.

Sorry, I am a little slow sometimes
Thanks for being patient.

I still can't picture how the 2x occurs...

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

quote:

Our skilled guys say they can get that using straight-edges or string.

(In reference to belt alignment)

Perhaps they 'can' but I doubt they 'do.' And they are only talking about offset alignment. I doubt they are even thinking about the two types of angular misalignment, and it's very hard to correct that without a laser.

You might be right. It brings up another question similar to the last one: can poor sheave to sheave alignment cause 2x ?

=========

I did a math excercize to estimate the tensions and belt resonant frequencies. I assumed we tensioned our belts to the middle of the band (10 pounds). I don't think there is a formula to predict the tension during operation exactly. But I think there is a lower bound for slack side tension based on preventing gross belt slip at a given transmitted power level. And the reasonable upper bound for slack side tension is the value that it was tensioned to in static condition (surely slack side tension must be lower than that during operation).

Under those assumptions, I calculated
slack side tensions = 30 to 120 pounds
tight side tensions = 150 to 240 pounds
Total sheave force = 530 - 1060 pounds
Resonant freuqneices of slack side: 560 - 1130 cpm
Resonant freuqneices of tight side: 1255 - 1600 cpm

It doesn't tell me much, but at least gives an idea how much force on the sheaves. If I get energetic, I might try to use that force as an input to a swag static calculation to guesstimate whether it is credible that keyway stiffness can cause the vibration.

Also it seems that the fundamental resonant frequencies of the belts are far below 2x fan speed (2400 cpm).

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

TensionCalcsR1.xls (119 Kb, 22 downloads)

Pete,

Since you have gone this far, you might want to see how much pull from tensioning it would require to bend the shaft, accounting for the long keyways. Would the slide base bend before or after the shaft is really all you need to answer since you know the base is bent. The location of the bend appears to be the logical place for the base to bend due to over-tensioning. The sharp vertical component of the tension forces makes that worse.

Have you put a dial indicator on the shaft with the belts tensioned? I would think that if you measured that and incorporated your min. and max tight side belt pull estimations you would get a pretty good picture of what is going on.

The belt tension is going to vary according to alignment and so is slip, more so after the initial stretching of the belts. With badly worn belts and sheaves (and believe me, in my work in the brick plant, I have seen some very badly worn sheaves and belts) the slack side can have almost no tension with the belts flopping around unevenly due to their differing natural frequencies. This is usually when one gets tangled up in another and throws them all off. That's when the old worn-out sheaves get new belts and the accelerated wear cycle starts anew.

BTW, I thought you knew the answer to this quiz. If you don't who does?

Also, I agree with Rusty that the mechanics "can" do it with a string. I can make gravel with a hammer, too.

Arrange for a loaner and have them use it once. They will probably change their minds.

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

Pete, you could verify your calculations using the Gates DesignFlex program. Personally, I don't think belt resonance is that much of a problem, and I am pretty certain that belt resonance will not produce 2X vibration of the driven component. Neither will belt/sheave misalignment.

The only thing that will produce 2X vibration of your fan shaft, in the absence of 1x or other harmonics (rubs, looseness, etc.) is a mechanical fault that produces 2 pulses per rev of the shaft. It really is that simple.

Thanks Danny and Rusty. Those are good comments and a lot to think about.

May be I am trying to overthink the issue, but I am still trying to understand how/why runout or a bent shaft would give 2x.

Attached is a picture of a bent shaft. The distance between sheaves changes at 1x as shown in the discussion of the red box. If we pictured the belt as a simple rubber band (ignoring the rotation and torque transmission aspects), then I would conclude that the force should vary at 1x.

So, what is different about a belt that would make it vary at 2x instead? As shown in the blue box, the angle changes at 2x. Maybe this affects the coefficient of friction (for example higher coefficient twice per revolution when sheaves are aligned and lower twice per revolution when sheaves are not aligned)? With a varying coefficient of friction, maybe the belt seeks a different slip resulting in oscillating torque and therefore oscillating radial force? That seems kind of a longshot to me but just trying to understand the mechanism.

BentShaft.ppt (56 Kb, 27 downloads)

Pete,
Instead of the rubber band analogy, think of a 60hz AC current generating a 120hz vibration. It's the change in the direction of the force that creates the 2X. In a belt driven machine, axial run out in the sheave moves the belts in and out once each revolution. A bent shaft moves the belts up and down once each revolution and in and out each revolution.

I have not read every post and would like to explore only the possibility of belt resonance that may be very close to 2x. See attached a bump test plot on a belt.

Many clean resonances are present. If I had a wider Fmax, probably a few more could be seen. A small size accelerometer was glued to the belt.

Surprisingly 1sta nd 2nd mode are harmonically related
Unfortunately single channel test was performed, Uniform window has been used

David

Belt_bump_test.doc (50 Kb, 29 downloads)

Charlie - Thanks. I can easily understand why 60hz current in a 60hz field makes 120 force. Assume the current and field (flux) are in-phase. When current peaks positive, flux peaks positive, force peaks positive because force = current * flux desnity... + = +*+. When current peaks negative, flux peaks negative, force peaks positive again (+ = -*-). I can't see any analogous relationship in a belt.

David - Thanks. Interesting bump test. The theory does give us a reason to expect harmonics for the geometry of a string/belt under tension supported at both ends. Den Hartog page 139 gives the resonant frequencies
w = n* Pi/L * sqrt(T/mu). The resonant frequencies come in harmonic series F, 2F, 3F etc. The situation is identical to a guitaor string. 1F is fundamental. 2F is one octave above. 3F is a fifth above that. 4F is a fourth above that. 5F is a major third above that. 6F is a minor third above that. In music it is known as an overtone series.

Back to the word of belts, does anyone have an example where belt resonance has caused increased vibration? I would like to know what the spectrum looked like. I would also like to know if the belts were flapping wildly (ours aren't)

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

quote:

Originally posted by electricpete:
I am still trying to understand how/why runout or a bent shaft would give 2x.

I agree, an eccentric sheave or a bent shaft will produce 1x due to varying length of the belt at 1x or due to unbalance resulting from a bent shaft.

On another hand, ECCENTRIC mass in presence of a unidirectional force, such as gravity and/or belt tension, will produce 2x.
Variation of shaft stiffness at 2x in presence of a unidirectional force will also produce 2x (Den Hartog p. 248).

David

Pete,

In the case of belts flapping wildly (which commonly occurs especially in dirtier environments) I would guess that you would get a high floor energy at lower frequencies in a velocity spectrum and random impacting in the waveform.

I would only guess because if I see anything- v-belts, arms, legs, lips, etc.- wildly flapping, I try to keep my distance.