Hey guys, got a question about resonance in fan wheels. I've got a fan wheel that I'm trying to determine if resonance could be an issue. I have balanced this fan from around .5 to .6 in/sec radically to around .1 in/sec, but the axial vibration still seems to be a problem. I'm seeing amplitude of around .2 to .3 in/sec axially. I've been researching resonance on a few sites, but decided to come to the best source of information - right here. Can anyone explain to me how to go about checking the resonance associated with a fan impeller? I've always seen looseness issues with this fan and was wondering if it was possible that one of these harmonics was causing a resonance issue for either the shaft critical or the natural freq of the impeller. I'm going to try to find out what the shaft criticals for this fan are tomorrow if possible. Right now, it's still down and the impeller is still accessible. These fans have long been a pain in our side and I've decided to try to get to the bottom of this. We've tried most everything else. The only thing left that I can think of is resonance. So, how 'bout guys, explain this procedure to me like I was a two year old. I'm somewhat new to this field but deal with it eight hours a day so I think I can follow. I am using a CSI 2130 if this helps any explanations.

I can tell you the setup for a bump test with a 2120. I would think the 2130 is very similar.
Press the Analyze Function Key.
Select Acquire Spectrum.
The first of three pages will open.
Frequency: 400Hz
Low Cutoff: 2.0 Hz
Lines: 400
Window: Hanning
SST Control: No
Units/INTG: Accel/Digital
Init Setup: No
Page Two
Average Mode: Normal
Trig Mode: Pre-trg
Trig Level: 1.0
% Pre Trig: 35
FS Range: 10
A-Weight: No
Triax Ctrl: Off
Page Three
Demodulate: NO
PeakVue: No
Prefilter: Off
This following statement is from a paper from CSI on Acquiring a bump test.
On some machines, you may achieve better results by changing some of the entries on page one. The window option, however, should usually remain at Hanning for impact testing, because it prevents leakage errors better than the Uniform window. Digital intergration is recommended to keep the waveform in the sensor units.
You can obtain good results using one-half the waveform peak value as a trigger level and a Hanning window in conjunction with a % pre-trigger of 35. The full scale range should be set to an amplitude above the expected response amplitude. This value is machine specific in the same way as the trigger level. If the analyzer overloads, simply set the full scale value higher. If the vibration response damps out quickly, you may be able to use a uniform window with a 5 or 10 % trigger.

Personally, I have had a lot of success without changing any of the parameters listed above. Place your accelerometer on one of the bearings in the direction you want to check for resonance, i.e. axial for axial vibration. You will need a soft mallet such as a dead blow hammer or even a two by four. Do not use metal hammer. Once you have set all three pages up, press enter. Another screen should appear: "Press enter to start data acquistion." Press enter again. Another screen should appear: "Waiting for trigger." Using the mallet, strike the fan in the direction the accelerometer is mounted. On an overhung fan, you probably can strike the fan downward causing the bearings to rock axially. Do not strike twice during each average. When you have finished, you should have a spectrum with one if not several spikes representing your resonant frequencies. If you do not store the data, you lose it when you make the next measurement.
If you have a photo tach, you could do a coast down test with the monitor peak and phase under the monitor mode from the same analyser function key. If you are running right on top of a resonance, you will get approximately 90 degree phase shift and a significant drop in amplitude over a very short frequency range. If you are running just above resonance, you will get close to 180 degree phase shift with a significant drop in amplitude over a short frequency range.
I hope all of that makes sense.
Ronnie

Thanks for the info Ronnie. I'll try that tomorrow. I do have two modal hammers although I can honestly say that I have no idea how to use them. They've got to be good for something other that killing roaches in the office. I've got a small one and a large one. Not sure what frequencies they excite (Is that correct?). Can someone post a simple explanation on how to use a modal hammer for resonance checks?

Hey Ronnie, the 2130 is, for the most part, the same as the 2120. There are a few subtle differences that I'm not sure about.

*Units/INTG: Accel/Digital - For units, are you using the native units for accel or are you using in/sec.

*Init Setup: No - What is this?

*FS Range: 10 - I'm assuming that this is some sort of "full scale" range. If so, the 2130 has an option to autorange. Am I correct in my assumption?

*A-Weight: No - This is an option on the 2130, but what is it?

*Triax Ctrl: Off - Don't see this anywhere in the setup screen. Not sure what it is, but I assume it has something to do with a triaxial accel. If so, I'm not worried about it.

*Prefilter: Off - Also don't see a selection for this. Not sure if it's important.

Anyway, I used this on a cabinet in my office to see if I would get any readings and it seems to work pretty well. I know that I don't want anything vibrating around 2930cpm on or near my favorite file cabinet.

Is there another way of doing this? I'm sure this way if the best, but I'm just curios about checks for natural frequencies.

What's the best way to find shaft criticals? Coast down with peak and phase? Cascade plot while coasting down?

BWZE,
Units/INTG: Accel/Digital, This is in acceleration. It is recommended to use this but I have accidently set it for in/sec and got successful results.

Initial Setup: If you leave this at yes, it will set all the parameters to whatever your route setup is for the point you are on.

FS Range: This is a full scale range and if you set it to autorange it takes a lot longer to acquire the spectrum. Use 10 as a starting point and adjust if necessary.

A-Weight: Not really sure what this means?

Triax Ctrl: Your guess sounds plausible to me

I also do this test with the 2115 and some of the selections are not there either.
As for your filing cabinet, hopefully it will never have to run that fast.
I have never done cascades. If I am already setup for a balance job, it seems easier to do a coast down. Usually you will notice a shudder when something coast through a resonance but if it is running right on the resonance that would not be the case. I guess depending on the situation as to which is more applicable.
One other method for resonance testing that I am familiar with is negative linear averaging. This can be done while the machines is running. If you would like, I can fax you a paper showing the setup for this. If you do not want to post your number, email me at ronniecl@aol.com.

Ronnie

Bwze:
I hate to seem remedial, but have you addressed the coupled imbalance within the rotor, or just forced? Along with this, were your weights in two planes on the rotor? If so forget me and my apologies for second guessing you.

I can't really address the couple unbalance in this fan because the impeller design and placement. I believe this to be the case. I don't see a logical way to place the weights on the back plate of the impeller. The front is very accessible, but the back is recessed in the housing and hard to get to. With this said, I'll try to attach some photos of the impeller in place so you guys can see what I'm talking about.

On another note, I just got back from trying this bump test in the field. I tried many different accel placements and hit in radial and axial directions. I seem to have gotten some interesting results. I'll try to attach these spectrums also. See what you guys think. In most all the tests I got a peak very close to run speed. The peak is not necessarily within 10% run speed,but I wonder if this natural frequency doesn't change slighly when at full speed. If I am doing this correctly and this is truly a somewhat clear representation of a natural frequency of this impeller, what will it cause in my normal spectrum.

I could only post one picture to the previous post so here is the bump test spectrum.

Sorry,didn't have the peak marked with a cursor.

So is 3333 CPM close to your fan turning speed? Anything closer than + or - 10% is what I have read is too close.

Ronnie

Is this a 3600 rpm machine?

Does the entire "thing" (where the "62521988 0020 X 3482") turn or just the area where you have the weights?

Yeah, turning speed is around 3575.

Ralph, the entire "thing" turns. There is a plate that mounts to the front of this housing on the studs that you see in the picture. This plate contains the "duct" that delivers nitrogen to the eye of the impeller.

Bwze,

How much does the weights weigh you have on the fins? Appears to be "3" "nuts", huh?

Why are we so close to the center of the wheel?

What is the total diameter of the wheel?

Did you say it was running around .5 or .6 in/sec? Would this be around 3 to 3.5 mils?

The small weight is .49 ounces, the large one is 1.6 ounces. They are "nuts" that have been altered slightly to serve another purpose. They are close to the center of the wheel because I was informed that since I can't perform a proper two plane balance that the weights should be installed as close to the back plate of impeller as possible. The wheel is approximately 34" in diameter and around 4" in width. Before I started to balance this fan, the plate over the impeller was removed to ease access. There are eleven blades on this impeller, and nine blades had weights of some type on them. I asked maintenance to remove all weights so that I could start from scratch. When the blower was restarted, I got the following readings:

F1H - .4847 in/sec or 2.47 mils
F1V - .2169 in/sec or .6509 mils
F1A - .5676 in/sec or 2.62 mils
F2H - .3766 in/sec or 1.93 mils
F2V - .5718 in/sec or 2.77 mils
F2A - .7598 in/sec or 1.97 mils

I know that those values don't seem to add up, but that's the overall for each point with no weights on the impeller at all. I deriving this info right from the spectrum that I collected. The mil levels are just the calculated levels from changing in/sec to mils in RBMware.

I don't have the readings from after the balance yet, but using Fast Bal II, I was able to reduce the horizontal vibration by approx. 95%. The mil readings in the balance software showed approx. 2.5 mils in the horizontal with .001 mils or so in the vertical. No, I don't have any phase data, and yes I realize that would help alot, but I don't have any excuses. I just forgot to do it because I don't get to do alot of balancing.

I've included a picture of the weights that I use.

Have you tried balancing using the axial 1x measurements to make your calculations? Imbalance on an overhung rotor may have the highest levels in the axial direction rather than the radial direction.

We had some turbines with oversped trip mechanisms that had this characteristic. Balancing, using calculations based on the axial levels could reduce axial levels by 70% without making any difference in radial levels!

Jon
Spintelligent Labs

Could you explain this method? When I do my balancing now, I use a two channel analyzer with two accels, one in the horizontal position and one in the vertical position. I take measurement one at the first bearing in the horizontal postion and the next measurement at the first bearing in the vertical position. Then I repeat the cycle for the second bearing. How do you do this with axial measurements?

quote:

I take measurement one at the first bearing in the horizontal postion and the next measurement at the first bearing in the vertical position. Then I repeat the cycle for the second bearing. How do you do this with axial measurements?

I am confused as to why you do this during your "single plane" balance procedure. What do you do after the above described procedure? Are the measurements for just determining what the amplitudes are in these directions or do you use all of them (the readings) to perform the balance?

Seems like you said you can't do a 2 plane balance on this rotor due to the lack of access to the back plane, huh? Are you using the 2 plane program in the meter to perform a single plane balance? I have never tried this, but -------.

I think what Jon is saying is do a single plane balance using 1 transducer placed on the bearing with the greatest amplitude in the axial direction. Some say one can only balance using the radial direction, but I have to agree with Jon, the axial amplitude and phase work on an overhung rotor when the axial is a problem before or after correcting the radial with a single plane balance. Of course, the axial amplitude may be caused by some other force than imbalance, but if the amplitude can be lowered through balancing, until proper corrections can be made, that is what the game is all about.

I would say place 1 transducer in the axial direction on the worse bearing (hopefully the one next to the fan), record the "original reference" phase and amplitude into the meter, determine where you want to place the first trial weight and how much, enter this weight and position into the meter, make your trial run, enter the new trial run data and allow the meter to perform the calculations for the "correction" weight.

One must still watch the radial amplitudes after "lowering" the axial.

Of course this is only my opinion and carries no warranty.

Ralph,
If I understand your question right, you want to know why you would want to use both bearings and two radial directions, horizontal and vertical, in a single Plane balance job instead of the one bearing closest to the fan and,or, just horizontal or vertical readings?
I am very familiar with the CSI balancing program. I have been using it for many years. You have the option of doing a single plane balance job with one or two measurement planes and two or four measurement points. I use the two measurement planes and four measurements points even when I single plane to make sure both bearings and both horizontal and vertical readings come down while balancing. What I find when doing this is in a single plane unbalance, with no other problems other than unbalance, the phase readings on both bearings will be very close. And what I normally see is both bearings in both directions react to my balance procedure. This is still a single plane balance and it only allows for correction in one plane. I have had the vertical readings go up or stay high when balancing only in the horizontal direction. And and if I use four measurement points I get better overall results horizontally and vertically. The only thing I can think of that I would need to take into consideration with this, and I should be taking this into consideration even using one bearing, is do I have any run out on the pulley end, or misalignment that could affect the drive side bearing more than the fan side. If for some reason problems other than unbalance are not an option to correct at that given time, I will use just the bearing closest to the fan and get it as low as possible.
I always want to correct all other problems before balancing but I find that is not always possible. Cost versus down time versus production lost versus ect.

BWZE,
If I am doing the math right, I figure you are running within about 7% of your resonance at 3333 CPM. I think this would have some amplification factor in your readings. I do think Jon and Ralph have a good point about trying to balance in the axial plane if it doesn't make your radial readings go up.

Thanks,
Ronnie

Thanks Ronnie.

I have also used CSI Fastbal for many years (since ~~ 1989)and never really used that function or even knew it would work for single plane.
Goes to show we are never too old to learn, huh?

Does the program take into consideration and use these other amplitudes and phases or still just use a single chosen direction for balancing single plane?

I always went the simple way::: single plane, single transducer, then recheck everything after balancing in my original chosen direction. Guess I am too simple minded to complicate things. Your way may be easier, I'll have to give it a try.

Thanks again.