High Frequency vibration in a dry screw compressor

[MFeng]

First let me say that I am new to screw compressors so some of my questions may be pretty basic.

Background:
A large screw compressor (~1200 HP) is processing natural gas. The machine has four male lobes and six female lobes. It is geared down from 1200 RPM to ~1164 RPM. It discharges into NPS 16 pipe increases to 20" goes through a 48" silencer and then back down to 20".

The discharge line (20+ ft off skid) seems to be vibrating significantly at higher frequencies, a few measurements were on the order of 32g's RMS overall. The main frequency peaks start around 910 Hz and increase in multiples of ~228 Hz up to ~4330 Hz.

Measurements taken on the compressor itself show similar frequency content but far lower amplitudes.

I have felt other screw compressors and they always feel "buzzy" but I've never seen the spectra. I have also heard about pipe and vessel shell modes being excited by the screw compressors.

Questions:
What frequencies should I be on the look out for?

It seems only likely that the source of excitation has to be flow induced (pulsation, turbulence, vortex shedding etc etc) or else I would expect the compressor skid itself to be vibrating more. Is this reasonable?

Is there a good reference for troubleshooting vibrations in screw compressors I could read?

Is there a decent acceleration based severity chart for these frequencies? I have come across a DLI chart that seems to indicate ~1000in/s^2 (2.58g's) as "tolerable".

Thanks

MF

 

[GregLocock]

32 g is a lot, but that does depend a fair bit on the frequency content, you have very high frequencies so that figure is less alarming than if it had been at 200 hz.

Yes, any decent vibration book should have a chart of allowable vibration levels for machinery. Harris Handbook of Shock and Vibration would be a good place ot start (sorry I don't have any to hand to check).

I'd expect to see multiples of shaft speed for balance problems etc, and 4*1164/60=78 Hz and odd multiples of that for your gas forces.

So your measured frequencies are very high in comparison, how many teeth are on the reduction gears?




Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MFeng]

Thanks for quick reply.

I don't know exactly how many teeth are on the reduction gear but its very near a 1:1 gear ratio.

My feeling is that it isn't unbalanced forces or another sort of mechanical excitation simply because the energy required to vibrate 20" pipe 20 ft off skid would be expressing itself most strongly around the sources (engine, gear box, compressor).

Since the compressor and on-skid vibration levels are an order of magnitude lower I find it unlikely that the vibration source is mechanical in nature.

It makes me wonder if the silencer they placed on the discharge line is doing its job. Its very presence suggests that the initial designers were concerned with pulsations, but I've always heard that screw compressors put out very low amplitude (but higher frequency) pulsations.

 

[ronfrend]

The main sources of HF vibration (and noise) on screw compressors are

1) Pulse from the rotor
2) Standing wave in the discharge piping

The pulse frequency from the rotor is 1164 * 4 = 41,164cpm which is 686Hz. That does not appear to be your problem although the difference between lobe pass and the 910 you are seeing is 224Hz (near enough to your 228 which 1/3 male rotor speed). Perhaps there is some damage or wear on one or two of the female lobes? Check in time domain to verify.

I would suspect that you have some sort of flow induced vibration at 910Hz which is interacting with the 228/686 frequencies.

Use c=fL
c-gas velocity (or in some cases speed of sound in the gas)
f-frequency
L-wavelength

Calculate your gas velocity and see if that coincides with f=910Hz and L=any length of discharge pipe between bends or other flow direction changes (where vibration is highest).

I have had cases where I have had to repipe the discharge to get rid of this type of vibration.

Keep us informed and let us know how you get on.

Regards,
Ron Frend

Ron Frend

http://www.predicon.net

 

[GregLocock]

I think you misread his rpm

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[MFeng]

Update:

So upon further investigation it turns out that there is internal gearing in the compressor and that the male rotor is supposed to be turning at ~2887 RPM (48.1 Hz) which would make the lobe mesh frequency ~192.5 Hz. Still pretty far from 910 Hz.

Ron, it is a dry gas compressor and the rotors do not touch so my suspicion is that rotor wear isn't a major factor, but I could be mistaken in that assumption.

In talking to plant operators I learned that the gear ratio on the external gearbox was changed to speed up the compressor so they could squeeze more flow though it. This makes me think that the male rotor may be turning faster now than the 2887 that was on the original performance specs.

Another gut hunch is that by speeding up the compressor, the silencer has moved beyond its original design envelope and is no longer providing the benefit it is supposed to. We were able to slow down the machine and the magnitude of the vibrations were reduced by an order of magnitude, but the spectra content remained the same.

So this is all pointing me in the direction of:
Excitation caused by vortex shedding-> Acoustical resonance -> Shell Mode resonance on the pipe.

Now to find some theory to back me up.

 

[unclesyd]

Your mention of the silencer stirred up the windmills in my mind.
We have a two stage SS rotary screw compressor of about the same HP and speed where the silencer did play a role in a big vibration problem after some piping changes. To correct the problem it required the silencer to moved closer to outlet flange, if fact I think it was mounted on the outlet flange, and silencer internals changed.

 

[MFeng]

Another Update:

It turns out that the performance runs I was looking at were old. The compressor was re-geared to speed it up and get more flow. The rotor is now turning at 3422 RPM which translates to a lobe pass frequency of 228 Hz.

All the readings were indicating multiples of 228 Hz. I am sure that we are seeing acoustic harmonics of the lobe pass frequency, and it would appear that for whatever reason the silencer is ineffective at this point.