Forced Torsional Vibration Analysis of Screw Compressor-Physical Meaning of Negative Torque

[mechanicaljw]

Dear All,
I am at a stage where i am trying to modify the torsional model i developed for free vibration to one for forced vibration. To do this we have measured the motor torque signal on the motor shaft at a critical speed of 1229 rpm. Our first torsional mode (20.5 Hz=1X input speed) is leading to massive torque peak increase and to 1st stage and synchronization gear meshing. While trying to curve fit the torque signal from measurment at 1229 rpm in the form
Asin(Omega*t+alpha)+C, i realized that A is negative in one case (with the highest goodness of fit). The phase in this case is +pi/3. When i change phase angle to -pi/2, A becomes positive but the goodness of fit is slightly lower than in the former. In both cases, C is positive and about the same (C=314 Nm in former & 311 Nm in latter). In the former, A is
-366.2 Nm and in the latter it is +364.11 Nm.

In the raw time series data, you can see that some of the torque values are below zero. This is not the case for non-critical speeds like 1800 rpm.

My first question is which of the two values for torque excitation should i take? The one with +ve amplitude A and negative phase angle or the the one with negative amplitude A and positive phase angle?

If in deed the torque amplitude is negative, what is the physical meaning of this? And is it desirable in the real machine? and what are the consequencies if this happens, in case there are any?

I have attached the time series plot for your kind perusal.

Thanks as usual for your interest and help.

Jimmy

 

[GregLocock]

The physical meaning is that the rest of the system is trying to drive the motor backwards. It is perfectly normal behaviour.

Cheers

Greg Locock


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

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

 

[mechanicaljw]

Hello Greg,

And thank you for the response. I am wondering whether the idea of the motor being driven by the rest of the system would lead to a situation where gears in our machine are forced to run in the reverse direction and hence lead to gear contact/meshing and to noise? Is this something you have encountered in your practice?

Thanks!
Jimmy

 

[GregLocock]

Yes, one solution is scissor gears, as used by Honda on their camshafts.

Cheers

Greg Locock


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

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

 

[rob768]

I am not sure that i reaed it correctly, but a negative torque means that the vibratory torque exceeds static torque (which can occur at lower speed) and at gears this will lead to gear hammer and an increased risk of gear damage. Operation at this speed should be avoided.

 

[mechanicaljw]

Hi Rob768,
And thanks as usual. We do agree with your assessment. I do have follow up questions though. When we repeat the test using MAN truck as drive instead of the electric motor, we still have this 1X input speed torsional frequency, which is our 1st torsional mode. So there is torque peak fluctuation. But we are yet to be able to measure the input torque as we have been able to do on the electric motor. One reason is that in one previous test we attempted to do it and we destroyed the torque flange in the process because the torque peak was too high. So we want to try strain gauging in the future. Do you think torque reversal can also be an issue in this case when the drive is the MAN truck engine? In our noise data we are seeing high peaks in our first stage gear mesh and synchronization gear mesh frequencies. This means we are having contacts? Could it be due to torque reversal?

In addition to the torque peak fluctuation, we are getting pulsating noise or beats at all compressor speeds. In the first test we did in the past, it is seen that the beats was occurring at the 1X input speed. And FFT of the lateral accerelation of lobe passing data found sidebands of 10 Hz, leading ut to believe that maybe these were being excited by our input torque signal from the drive because the 1X input speed was 10 Hz in that test. In the new test we have just done on another MAN truck, the 1X inspeed is is 11 Hz and the pulsating noise is occurring at 11 Hz. I am thinking the N profile rotors we have in our two rotors screw compressor could be the problem. Not stiff enough due to its long length and small diameter. What do you think?

Thanks!

Jimmy

 

[rob768]

sounds to me like you have more than one problem. I would start with a proper torsional vibration calculation of the system to indicate the location of natural frequencies. We don't use a torque flange to measure torsional vibration but always use strain gauges. Sometimes, when no shaft is available, we use a laser-doppler system to meause speed non-uniformety only.
Perhaps the calculation will indicate that use of an elastic coupling (steel or rubber) btween the sets might help.
Its normal to see the gear meshing frequency, depending on the type of tooth. If it is a straigh tooth, it will be there. The better the profile quality or gear type, the less pronounced the meshing frequency will be present.

 

[mechanicaljw]

Hello rob768,

I am looking for a good reference on the design of two rotors screw compressors? Do you or any of your colleague have such information? Something for a beginner would be most welcome.

Thanks!
Jimmy

 

[rob768]

No, sorry. Can't help you there. I Know how they function and some of their problems, and in a failure analysis i can see what goes wrong. Design criteria is something else. I'd most likely get it to work, but that would be basically it.