Can beating frequency in a srew compressor be at 22 Hz? 2

[mechanicaljw]

Hello All,

We are having a screw compressor that has 3 male lobes and 5 female flutes that runs on a 90kW, 4 pole, 50 Hz motor with synchronization speed of 1480 rpm. However, at a speed of 1350 rpm, we are experiencing a resonance frequency of 22 Hz. We are unable to pin point the source of this frequency and to tell whether it is torsional or structural? Because the driver is the female i computed the difference between 2 times line frequency, which is 45Hz and the excitation frequency and got 22.5 Hz. In our measurements we are getting 22 Hz and error of about +1 as the resonance freuqncy. So at times we are close to 23Hz. Since i know the beating frequency to be result of two frequencies interacting, i am woundering whether this is the cause of our problem? Or am I doing something wrong? Because for beating to occur from what i have read the two interacting frequencies are usually quite close to one another? What is also happening is that at this 1350 rpm where we have this resoance the torque suddenly increases. So we to figure out whether the this is as a result of the resonance?

Any thoughts would be appreciated.

Thanks!

Jimmy

 

[mechanicaljw]

Hi Greg,
Thanks again for the response.I was wondering what specific data you would love to look at? I have data for torque input measurements and FRFs data from vibration analysis that has been done. Kindly let me know if you need any specific data.

Thanks!
Jimmy

 

[GregLocock]

Well OK, let's see your spectra from the running system and some FRFs, in excel or csv format.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[electricpete]

use a stain-telemetry system for both static and dynamic torque (torsional vibrations).

Another critical point (question): We did monitor the input torque and did an FFT of this input torque and clearly we have 22.5 Hz as a resonance frequency in this input torque signal to the compressor.

I have a follow up question about why the torque is increasing at resonance. Can we say that because we are seeing a torque increase at resonance that the resonance is torsional?

What is the ratio of the 22.5hz torque to the static torque?

I agree, more data would be good. Tough to follow.


=====================================
(2B)+(2B)' ?

 

[mechanicaljw]

Hello All,

I am grateful for all of your assistance.

@ electricpete:The compressor torque peak to peak variation is 650 Nm @ resonance compare to 80 Nm @ non-resonance conditions (pls see the attached excel file).

@Greg: I have been trying to figure out what would useful for you. I am attaching this file that shows some post-processing of our torque data together with those that were taken in the vibration analysis. The first stage gear seems to be where we are having a problem. The vibration amplitude there at the critical speeds is consistently very high and during tear down a wear on one side of the gear was observed. Please let me know if you have any thoughts.

Thanks!
Jimmy

 

[electricpete]

Aha. Very good data. LPS for that.

It appears you have some harmonics of running speed in your torque spectrum regardless of speed.
But "vertical" (constant-frequency) line at 22hz in the plot certainly suggests resonance at 22hz as you concluded.
And I can't tell from the plots, you stated 650N-m pk-pk variation at 22hz vs 80N-m at other speeds.
I was interested in ratio of that to the static power as rough indication of the severity. I don't know your static power or torque but can covert 650Nm at 22hz.. equates to 90kw pk/pk power variation or 45kw each way.

Your 90KW motor when operating at reduced speed has thermal limit corresponding to constant torque. So at around half speed, the motor is derated to 45KW. If motor was operating at it's max steady load of 45kw (based on thermal rating), the power would be oscillating 0 to 90KW. I guess that's pretty severe.

fwiw I tend to agree with your diagnosis of torsional resonance.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Thinking through in general ways to attack a torsional resonant condition.

1 - It requires resonance and excitation. I'm not sure what is the source of 1x turning speed torque excitation throughout the speed range. Even 80N-m at non-resonant condition sounds like a bit of a swing. What causes it? Misalignment? Beats me. Might be worth investigating as a first easier approach.

2 - Other approaches I think would seek to detune or dampen the resonance. A pretty big step requiring machine modification.



=====================================
(2B)+(2B)' ?

 

[electricpete]

By the way since you have a variable speed machine... don't know where operation requires the speed but of course detuning may just push the same problem to another speed.

=====================================
(2B)+(2B)' ?

 

[mechanicaljw]

Hi electricpete,

I am grateful for your insights and your thoughts. I have taken them into account and would speak with my colleagues to see if we can look at some of your recommendations. I will also look at the torque ratio that you requested and get back to you. We had concluded that damping might do for a short term fix, just like you suggested. So we are in the process of looking at the appropriate material for doing that. We may want to consider damping the structural vibration and the torsional as well to see which would have the most impart.

To your comment: "By the way since you have a variable speed machine... don't know where operation requires the speed but of course detuning may just push the same problem to another speed"-I agree that when you detune you may only shift the energy in the system to another speed. I think if it were to shift to higher frequencies that would be good but we are constrained by the operating range of the machine. And because it is fixed we cannot do anything about this. I'll let you know what happens when the damping is done.

Thanks!
Jimmy

 

[electricpete]

Do you have any thoughts about what source of excitation is creating the 80N-m torque oscillation at 1x running speed, even during off-resonant conditions?

=====================================
(2B)+(2B)' ?

 

[mechanicaljw]

Hello

Can you comment further on your previous question "Do you have any thoughts about what source of excitation is creating the 80N-m torque oscillation at 1x running speed, even during off-resonant conditions?" Do you mean that we still see peaks @ 1x rotational speeds in the torque FFT data even when we are running at non-critical speeds?

I was thinking about what you also meant by static power? in one of your previous questions?

Thanks!
Jimmy

 

[mechanicaljw]

The previous post I sent was intended for Electricpete. Sorry that i forgot to address it to him.

Thanks!
Jimmy

 

[mechanicaljw]

@ electricpete
The major problem with the torsional vibration theory is that we cannot change the 1st mode by varying prop shaft stiffness. Can you give this some thoughts. Why do you think this is happening?

Thanks as usual.
Jimmy

 

[mechanicaljw]

@ electricpete

For the last question i am thinking because we used the static stiffness instead of the dynamic stiffness? Can that be possible explanation for why the natural frequency did not change with change in the stiffness?

Thanks!
Jimmy

 

[electricpete]

Can you comment further on your previous question "Do you have any thoughts about what source of excitation is creating the 80N-m torque oscillation at 1x running speed, even during off-resonant conditions?" Do you mean that we still see peaks @ 1x rotational speeds in the torque FFT data even when we are running at non-critical speeds?

Yes, 1x and several harmonics of 1x are present at all speeds in your torque waterfall/campbell plot labeled “Autospectrum(torque) – Input”. We can’t judge the magnitude from those plots, but the text underneath it states that there is 80 N-m pk/pk of torque oscillation even at non-resonant conditions.

Is it a lot? I’m not sure but I tend to compare it to the “static” = steady = non-varying = dc component of the torque.
The rated torque of the motor is P / w = 45,000W / (2*Pi*25hz) = 286N-m.
I assume you are operating at or below the peak torque rating. So the peak-to-peak torque variation you are seeing even at non-resonant conditions is at least 80/286 ~28% of the steady torque. It seems like a lot to me. Certainly would not be expected for centrifugal machines ... I don’t know about lobe compressors.

Regardless of whether it’s expected or not, if you have this non-insignificant excitation present at non-resonant conditions, it’s not surprising it will increase by a factor of 10 or so when you vary speed to resonant conditions.

Since varying speed is part of the application, seems worthwhile to investigate the source of the 1x torque oscillation and see if it can be reduced.

The major problem with the torsional vibration theory is that we cannot change the 1st mode by varying prop shaft stiffness. Can you give this some thoughts. Why do you think this is happening?

Again the torque waterfall plot seems to provide compelling evidence of torsional resonance in 2 respects:
1 – you stated the highest torsional variation occurred when speed was 22.5hz and torsional variation decreased away from that point
2 – there is a white band at 22.5hz for all speeds.

I think buried in your question is a statement that you have attempted to stiffen the shaft, but saw no change in behavior, is that what you’re saying?

It may be that the torsional stiffness of the shaft is much higher than other torsional stiffnesses in the system (like the coupling). In that case, doubling the stiffness of the shaft does nothing if you don’t change the coupling (I’m neglecting any weight changes associated with this stiffening). As a simplistic example: imagine you have a series combination of springs with spring constants 1E6, 1E3, 1E6. The effective spring constant of the series combination is about 1E6. Now double the 1E6 stiffness and leave the 1E3 alone. The effective spring constant of the series combination is still about 1E3.


=====================================
(2B)+(2B)' ?

 

[electricpete]

Sorry, typo correction. Change 1E6 to 1E3 as follows:

As a simplistic example: imagine you have a series combination of springs with spring constants 1E6, 1E3, 1E6. The effective spring constant of the series combination is about 1E3. Now double the 1E6 stiffness and leave the 1E3 alone. The effective spring constant of the series combination is still about 1E3.

=====================================
(2B)+(2B)' ?

 

[mechanicaljw]

Hello electricpete,

Thanks again for your valuable contribution. I in deed saw the typo and was just about to ask when you wrote to correct it.

The last point you made about the stiffness not changing because of the possibility of connection to other flexible joints that have lower stiffnesses was spot on. To further investigate this we have a test coming up that would look at a rigid or a solid drive shaft with no flexible joints. We hope that this will raise the first mode frequency away from the lower frequencies we have been having.

To look at the effect the prop-shaft torque might be having on the system we have repeated the measurements with a hydraulic drive motor and the compressor sound was smoother when compared to the prop-shaft driven unit. The torque increase at 1350 rpm vanished and from the run up data there was no resonance freq below 100 Hz and the over all vibration level was lower. So it is safe to say that the compressor itself is not the source of the problem. The only other observation that we need explanation for is the fact that there was still large fluctuation in the torque. The question now is why? Could there be some form of coupling between the torque signal and the pressure inside the compressor? Kind of a flow induced problem?

Your thoughts would be appreciated as usual.

Jimmy.

 

[electricpete]

To look at the effect the prop-shaft torque might be having on the system we have repeated the measurements with a hydraulic drive motor and the compressor sound was smoother when compared to the prop-shaft driven unit. The torque increase at 1350 rpm vanished and from the run up data there was no resonance freq below 100 Hz and the over all vibration level was lower. So it is safe to say that the compressor itself is not the source of the problem. The only other observation that we need explanation for is the fact that there was still large fluctuation in the torque. The question now is why?

So, if I understand correctly, when using hydraulic motor, you still had torque oscillations (on the order of tens of N-m) accross a broad range of frequencies? The only difference is that you didn't run into a frequency where this oscillation increased to hundreds of N-m?

If the answer to my two questions above is yes, then I would think the excitation remains and the resonance was removed or damped.

How removed - perhaps moved in frequency above 100hz due to much lower rotating inertia of the hydraulic motor than the electric motor possibly combined with shorter shaft (?).

How damped - perhaps the hydraulic motor provides significant torsional damping (?)

Could there be some form of coupling between the torque signal and the pressure inside the compressor? Kind of a flow induced problem?

Beats me.


=====================================
(2B)+(2B)' ?

 

[electricpete]

How removed - perhaps moved in frequency above 100hz due to much lower rotating inertia of the hydraulic motor than the electric motor possibly combined with shorter shaft (?).

also of course stiffer coupling?



=====================================
(2B)+(2B)' ?

 

[GregLocock]

Hi Jimmy

Can you restate exactly which problem you are trying to solve?

Looking at your torque campbell's plot I have a hard time believing the 22 Hz vertical 'resonance' line is a mechanical resonance. Could you plot a narrowband, say 20-25Hz amplitude vs rpm for that torque data?



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[mechanicaljw]

Hello Greg,
I am working on getting the narrowbank plot and will send it as soon as it is done.

To your question about what we want to do. We have a screw compressor that is run by a diesel engine from a tractor. The drive is a propshaft. When the compressor runs, we are getting this strange noise @ 13 Hz. When the drive is changed to an electric motor, the noise we are getting is @ 23 Hz. The campbell's plot that i sent was for the latter. Just recently, we replaced the electric motor with a hydraulic drive and campbell's plot of the torque showed no resonance below 100 Hz. The resonance freq we had was @ 700 Hz. The machine was not noisy and the torque increase was not significant, which is not the case for the others that i have mentioned above. The task we have on our hand is to do away with this noise and to reduce the torque @ the resonance frequency. We first must answer whether the problem is a mechanical resonance or a torsional problem. The line diagram attached may help clarify the problem. In this case the electric motor is the drive and frequency or speed of the motor is identical to the frequency of the noise we are getting in the system. The same is true for the diesel engine drive. As a last resort we are working on using a torsional damper or a mechanical damper to see if this would elimniate the problem.

@ electricpete: I agree that in the case of the hydraulic drive the resonant frequency was probably shifted from the 23 Hz to 700 Hz or damped. The peak torque that existed in the case of the electric motor or the diesel engine no longer existed, only that it does fluctuate.

Recent test with Solid propshaft:
280 Nm peak torque amplitude (much lower than standard propshaft)
40 Hz Resonance frequency

For the standard or original propshaft:
700+ Nm peak torque amplitude
Resonance freq (believed to be first mode torsional) is @ 24 Hz

Care was taken to ensure the same shaft length and dia.

What are your thoughts? Can we say the frequency change of the 1st mode when the standard propshaft is changed to a solid one is an indication that we have torsional resonance?

Thanks!
Jimmy