I have attempted to do a 2 plane balancing job on a center hung 1200 rpm hammer mill,
Dia : Width = 5 : 1.
Rotor weight is 4,400 LB
Phase showed almost 180 deg difference in horizonal direction. Mostly 1x was present.
The vertical phase (which had much lower vibration), as in most of the cases, did not make any sense anyway. I've had a good response from both trial weights and the box suggested a 10 LB correction weight on each side about 180deg apart.

Two 10 LB weights sounded little bit too much as similar rotors normally required less then 3 LB. But we do trust computers these days... and I reluctantly follow its recommendation. As a result vibration goes even higher (no errors in weight placement), with still about 180 deg phase difference. The trim weight suggested now is 15 LB.

At this point I don't see any reason to continue with the FastBal II and start two plane balancing on a graph paper using method of separation of couple and force unbalance and succesfully bring vibration down to acceptable levels by using total of about 3 LB.

Have you guys been through something like that?

Thanks,
David

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

A number of algorithms used to balance using computers/calculators have been known to compute large solutions.

What was the data used during the balance including number of probes, balanance planes, and data? Did you sucessfully balance it with a couple shot?

Bill,

I tried 2 balancing planes, 2 HOR probes. I have to dig in order to find actual vibration data during the program procedure. Is that what you need?

Yes, I did successfully balance it using a couple shot. Prior to that I have completely balanced out small force unbalance.

As I have mentioned above, the vertical phase data did not make any sense in terms it did not show classical phase properties indicating presence of unbalance, such as
H1-H2=V1-V2. No phase readings across the coupling was paid attention to as well as they did not indicate any of the classical symptoms. Despite that, the balancing job, as in most of the cases of senseless phase readings, was successful.

I forgot to mention that no resonance was even close.

David

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

David,

Are you certain that all the hammers are swinging freely?

Danny,

The hammers were removed for the balancing procedure.

David

What can happen during a balance is that there is essentially one sensitive direction (set of balance distribution). Darlow (deceased now) in his book on balancing mentions this.

This can be determined by treating the influence coefficients as a matrix and performing a singular value decomposition on it. If you give me the IC's I can do this and explain results.

Of course it could be something else entirely.

David,
I have had a similar problem on a hammer mill. I have to use 1/2 the weight that the computer tells me to use when balancing this machine. The balance programs assume a linear system. I just assumed that the system was not linear. I too could not identify a resonance. I had a fan that reacted the same way.

John J

A crack in the structure or somewhere make anything trying to work linear go nuts. Common in hammer mills and on a fan once in a while. Just a suggestion. I have problems getting people to strip of the hammers for balancing, what tools do you use to persuade them, shotgun? Same for cleaning decanter centrifuges... Olov

Bill,

I thought IC's are transparent when a balancing program is used ( or I am not aware how to retrieve them).

John J,

It does not appear that it was a non-linearity problem since with the weights size obtained from the graphs (which is also a linear algorithm) balancing was successful.

Olov,

We change hammers frequently as they wear out fast and unevenly, so to have just the rotor itself well balanced makes sense to non-vibration people. As far as cracks are concerned they will definitely affect linearity of response but it was not the case here.

David

One can re-creat the IC's with the original data, or some programs save them for use in a future trim or one-shot balance.

This example shows what can happen with a balanceable problem which is mostly couple, just an example.

Original vibration at each end, r0
r0 = [0.711 174.3 degrees end 1
0.711 5.7 degrees end 2]

Add trial weight to end 1 of 1 @ 0 degrees get

r1 = [0.218 @ 161.1 degrees
0.219 @ 18.8 degrees]

Remove first trial weight and add 1 @ 180 degrees to end 2 to get

r2 = [0.219 @ 161.2 degrees
0.218 @ 18.9 degrees]


Remove the second trial weight and compute the correction weight. I get

Wc= [70.71 @ 270.6 end 1
70.71 @ 269.4 end 2]

However adding a correction weight of

0.707 at 0 degrees end1 and
0.707 at 180 degrees to end 2 results in (if your balance program has a predict feature you can check this)

Residual vibration

0.0707 at 90 degrees end 1
0.0707 at 90 degrees end 2

This would be an approximate 90% reduction with 1% of the balance weight required to balance the last little bit. The residual above or any in phase vibration will be difficult to balance.

This message has been edited. Last edited by: William_C._Foiles,

Hello William,

I have never inputed 0 as trial weight position either on my computor balancing program nor any of the portable instruments i have used.

0 (zero) computes as nill on most computer programs. I've never trusted any portable instruments that were computing reference angles of a value of 0 (zero).
Perhaps, nowadays they are more intelligent and 0=360...

I may be wrong... but...

Markoleo

Markoleo,

With the exception of those fans I balance all the time, and want to one shot it the next time, I put "0" zero in for the trial weight angle a bunch. It doesn't matter, it's just a place to reference every other move.
When the portable gives a solution, it is "so many degrees from zero"

Dave

quote:

r2 = [0.219 @ 161.2 degrees
0.218 @ 0.18.9 degrees]

Bill there is a typo for your second run. I assume it is 18.9 degrees

Regards

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

Corrected the typoo. Thankks.

Markoleo,

The computer has no issue with 0 degrees. The goal is to get 0 amplitude, but this is a singularity in polar coordinates; at 0 amplitude, what is the angle? Any angle will do. Try finding the atan(y/0); you have to check for this. Of course one doesn't get 0 vibration all the time.

David,

Instead of using single plane balance balancing techniques, have you tried multi-plane? When balancing on a “long” rotor, multi-plane is much easier and typically provides better results than single plane.

Treat each coupled pair (equal weights 180 degrees apart on each end) and static pair (equal weight in phase at each end) as single plane weight placements, with the appropriate resolved static and coupled 1X filtered vector component.

You can then use simple graphical methods to determine the results. Use the computer program only to check your hand calculations or if you need to do more than one weight (either single or multi-plane) placement at one time (typically you won’t need to do this).

zackman,

My rotor is actually short, but had predominantly couple unbalance. The whole problem I had was that 2 plane computer balancing program had produced weird numbers that made me switch to a couple-static graphical method, which was successful and required much smaller correction weights.

Bill,

I'll did not have a chance to go through your numbers yet.

David

Occassional calculation of large correction weights from multiplane balances using either exact point or least squares balance programs has been noticed since their inception. The example I gave illustrates this.

Some have done as David did, reduce the number of balance planes. Sometimes choosing more or different speeds can help. The balance program tries to do its best (find optimal solution or exact solution to drive vibration to zero if an exact point balance program). Getting that last little bit may be difficult.

Sometimes if you just balance the important part, like the couple that David balanced, you can achieve acceptable vibration levels, and after all this is the purpose of balancing.

There are other balance methods, but I don't know if any have commercial implementations. I have balanced many machines where this could have been a problem, but either I put constraints on the balance weights, response levels, or reduced the IC's to only the important directions to eliminate calculating large (and sometime not that useful) balance weights.

quote:

Originally posted by William_C._Foiles:
, but either I put constraints on the balance weights, response levels, or reduced the IC's to only the important directions to eliminate calculating large (and sometime not that useful) balance weights.

Bill,

I found your example balance problem intriguing. I ran your numbers in Rotorbal with similar results. The prediction was just like your prediction.

The fact that you can reduce the vibration level from .711 to .07 with only .707 weight additons rather than trying to take it to "0" with weights 100 times larger is really impressive!

I do not have a program that will allow conditions / restrictions to be placed on weight amounts or vibration levels. What software has those features?

Thanks for your example and for sharing your thoughts and experience.

Hudson Smith

Good to see you on the board Hudson. I don't guess you've been to the Fudrukers in Dharan recently.

I have some files I wrote for Matlab and for Scilab (a freeware program). It allowed me to do a few tricks with good results.

Currently, ROMAC has a code or two that has a number of features like this. I gave a lab two weeks ago at UVA where they used the ROMAC code (simple but tricky balance problem on a 3 mass rotor with tilt pad bearings - they fell for the traps and got 20+ mils pp through the first critical with their first balance shot. The grad student looked worried that I might bow his rotor further than it already is bowed.) If you are still with the double cross company, you are a member; write and I shall give you a contact within the xx company if that is where you are.
--------

For clarity -- I did not operate the ROMAC code and do not know directly how easy to use it is; it looked ok from a distance. Some of the members wanted to use the code; although, that was not my main purpose. I didn't pay too much attention to the program, because I don't have it. I have spoken to the author about its capabilities and seen a presentation of his work.

Also, I would recommend joining ROMAC to those in a positon to do so. I am not a member, at least yet, as I have been reminded.

This message has been edited. Last edited by: William_C._Foiles,