I have a peculiar high 1x on an overhung rotor. the shaft was replaced last week. The material is stainless steel for the shaft and rotor. The bearings are max mount Link belt (tapered adapter sleeve with nuts). The alignment is perfect. I'll let you look at the data and see what you say. The collected data is on the slide and can be expanded there or copied onto a new slide if you wish. Let me know what you think. Thanks. The phase readings are NOT corrected for orientation.

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

6tdust.ppt (146 Kb, 244 downloads)

quote:

The phase readings are NOT corrected for orientation.

Does this mean the two axial direction phases on the fan bearings are within 12 degrees of each other? And the motor is in phase with the bearings also?
What is the fan bearings sitting on, a fairly strong pedestal base or one that have a tendency to "flex" under a slight amount of force?

If all is strong (pedestal) and the phases are in phase, and the impact test does not show it to be running at or near its critical, I would say balance, in the vertical direction first and then if the axial stayed high, rebalance in the axial direction. You could balance in the axial first, but I know some say you can only balance in the radial direction, but I just wonder where I went wrong by doing it in the axial direction, no where I guess.

Roy,

Does the location of the arrow also indicate the orientation of the accel?

What kind of coupling? Maybe it has a limited end float adjustment?

My first look says imbalance in the fan and maybe some looseness in the back end motor anchoring. There is almost nothing radially on the front end of the motor and then high axial and vertical but low horizontal in the back end. All the axials seem to be phase-locked so I would expect that both rotors are moving as one unit and rocking the back end of the motor.

I would suspect dynamic imbalance in the fan, looseness in the back ancoring of the motor and maybe something in the coupling that is allowing axial transfer to the motor.

Sorry if this is fragmented. Lots of interruptions.

Good Luck,

Danny

Ron,

I see in your data:

1) High axial at all 4-bearings in-phase
2) Out-of-phase motion in both horizontal and vertical directions between motor OB and fan OB bearings
3) Higher vertical than horizontal at fan OB bearing

I agree that source of vibration at 1xSS is from fan rotor unbalance. The motor-fan structure is either very flexible or it has a natural frequency close to 1xSS. I recommend testing for natural frequency when you do balancing job, and correcting as necessary. There may be considerable couple-unbalance as indicated by high axial vibrations that requires two-plane or static-couple balance procedure.

Walt

I agree with the axial being all in phase and I did suspect dynamic unbalance because the H and V are also in phase. I will check the motor mounts when I do this fan plus a bump test on the rotor assembly before balancing. The fan frame itself seems to be very robust. It's .5" plate. I'll try the vertical to axial balancing runs as you suggest Ralph. I can do 2 plane with my box, what do think of placing one sensor in the vertical or horizontal then the second sensor in the axial? Ever tried this?

Why not perform a least square balance using all of the measurements instead of just using the vertical or axial measurements?

quote:

Originally posted by Steve Ciesla:
Why not perform a least square balance using all of the measurements instead of just using the vertical or axial measurements?

Could you expand on this idea? I'm interested. Thanks.

quote:

Why not perform a least square balance using all of the measurements instead of just using the vertical or axial measurements?

Steve,

I also would be interested in more info on this procedure.

If one goes by a text book, there is one component which is not satisfied in order to confirm imbalance, namely, close to 90 deg. difference between V and H for each fan bearing. Based on phase data fan bearings are moving along a straight line rather of something close to circle orbit.

The only symptom supporting unbalance is high axial readings, but not the phase data. Am I wrong?

Should overhung rotor phase readings be interpreted differently from centerhung?

David_G,

On an overhung fan, MOST OF THE TIME, the difference between horz and vert should be near the same for each bearing. In other words, pos. 3 has 7 deg. difference, and pos. 4 has 5 deg. difference. This is near enough to call it the same.
ALong with the high axial, I too would presume some weight is needed on the rotor.

Dave

If one does coastdown plots then one can strengthen the diagnosis. Resonance looks possible and should be investigated.

But what is resonance but the coincidence of a forcing function and a natural frequency. Many resonances are successfully dealt with by balancing. If the machines imbalance changes often (dirt, erosion, etc.) then balancing alone may not suffice as a practical cure, but often balancing is the cost effective solution if only temporarily.

A number of balancing programs now allow for multiple speeds and measurement locations not equal to the number of balance planes. Most of these are least squares solutions to the balance problem. What least squares does is to minimize the (generally weighted) sum of the squares of the residual predicted vibration with a balance correction.

One should pay attention to the predicted amplitudes to determine that the solution is a good one. A single plane balance does not always work.

Dave,

Your statements on phase relationship for an overhung fan match my observations. I just was not sure if it could be generalized. At least I have never read about this distinction.

I wonder what vibration scientist have to say in this regard.

quote:

... A number of balancing programs now allow for multiple speeds and measurement locations not equal to the number of balance planes. Most of these are least squares solutions to the balance problem. What least squares does is to minimize the (generally weighted) sum of the squares of the residual predicted vibration with a balance correction. ...

As Bill said, some balancing software will allow you to use a number of measurement locations to calculate your balance weight. Let me summarize the procedure below.

Assume you have a two bearing component (fan, turbine, pump, etc.) with one balance plane. You measure vibration in three directions at both bearings for a total of six measurements.
1) Collect the 1X amplitude and phase at each measurement location. These are your original (O) measurements.
2) Install your trial (calibration) weight (W).
3) Collect the 1X amplitude and phase at each measurement location. These measurements result from the vector addition of the original unbalance and the trial weight (O+C).
4) Determine the effect of the trial weight by subtracting the vectors, (C) = (O+C) – (O). To calculate the influence vectors (H) = (C) / (W).
5) If you’re using a single measurement, you would calculate the correction weight as Wcor = (-O) / (H). However, if you do this for all six measurement locations you can get up to six different answers. Which correction weight do you use? Does it provide the best response for all of the points?
6) The least square method is used to minimize the response at all of the measurement points. Knowing the influence vectors, you can predict the response at all locations to a known weight. The least square method calculates the correction weight that minimizes the residual response (R). In this case, 1H, 1V, etc. are the vibration responses at each of the measurement points.
(R) = SQRT (1H^2 + 1V^2 + 1A^2 + 2H^2 +2V^2 + 2A^2)

The largest response has the largest effect on the calculation. However, if you have a number of measurements with similar amplitudes then the least square method incorporates them accordingly.

I’m sure this isn’t the clearest description. Let me know if you want an example with numbers.

Steve,

Do you least square both the amplitude and the phase?? How does phase play into the location of the correction angle?

Thanks in advance to your response.

John Gorman

Thanks for that Steve. I have attached an updated ppt. for your viewing. Tell me if I'm correct here. If, as the plot shows, the amplitude and phase on all 6 readings are close, you could average them. Then where I'm not clear here is how do you calculate the weight to apply in order to achieve the desired result? In this case .080ips at 225deg. I'm not sure how to use the formula to get the answer if it is possible to do this without doing a trial run on the fan. Thanks.

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

6tdust.ppt (152 Kb, 56 downloads)

Sorry to stick my nose in.

I am NOT a balance guy. I don't think he was suggesting to average the vibration vectors before balancing. Here’s the process that I understood to be suggested above:

You have now a set of original vibration measurements: Oi for i=1,2,3 etc correspond to your various measurements

Add trial weight: Utrial

Get another set of vibration measuremetns (Oi+Ti)

Calculate the trial weight vibration effect estimates Ti = (Oi+Ti) – Oi (vector subtraction)

Calculate Influence coefficients ki = Ti/Utrial (vector division)

For a given correction weight, the predicted residual unbalance vibration at each point would be

Ri = Oi + ki * CW (vector arithmetic).

Vary CW and look for a solution that makes all the Ri tolerably small.

One approach to do this would be to combine the Ri into a single number Rtotal = sqrt(R1^2+R2^2+etc). I think there are other ways. For one thing a little trial and error with excel. Try the various solutions you would get from looking at only one point and see how it affects the others.

Seems like a lot of work to me, though.

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

Thanks, I wouldn't use the motor measurements. I'm just going to check the mounting bolts since there is some looseness showing there. I am interested to see if I can calculate a possible solution namely, how much weight and where before the next run. Since the amplitude and phase for each reading are fairly close. I should be doing this job tomorrow or Friday. I will also be doing a frequency response test to see if the fan is running at or near resonance.

quote:

Seems like a lot of work to me, though.

There are standard routines/methods to solve the least squares problem. If this is the measure you want to use to balance (hopefully with weighting), the solution is easier than an exhaustive search in Excel.

It was mentioned that this method is sensitive to relatively large measurements in the calculation. Weighting can help with this.

Also, there can be portions of measurement combinations on which the choosen balance weights have little effect. This can lead to computing large correction weights, but this can be a problem for other 'exact point' balances where the number of measurements equals the number of balanc planes.

Note: The solution that minimizes the square root of the sum of the squares is the same solution that minimizes the sum of the squares, which may be easier to conceptualize.

I have read up on the one shot method and found that I am missing 2 key factors which are mechanical phase lag and balance sensitivity (gm/mil) which I would have if there was prior balance data available for this fan.

quote:

I can do 2 plane with my box, what do think of placing one sensor in the vertical or horizontal then the second sensor in the axial? Ever tried this?

Does you meter multi-plane balance overhung rotors? They are a different animal than a center hung rotor, in that the cross effect from each plane doesn't react the same as a center hung rotor. I have never had much success with the CSI Fastbal II when trying to 2 plane an overhung. Maybe your meter does a better job. Hope so. I think most (95%) field balancing jobs can be done in the single plane mode mainly because they have been dynamically balanced prior to installing. Just my opinion.

How wide is this rotor, front to back?

I definintly would do a coast down ("frequency response") like you said. This could help one not put a weight on the heavy spot for the first trial run.

This "least square" method sounds like it is quite complicated and with one not being experienced over time with it, could result in some bad mistakes.