Screw Compressor

[DaveSco]

I am currently looking at the vibration monitoring on a couple of gas screw compressors, with a 5/7 ratio. There have been a number of problems with these units in the past, where instrusive maintenance has revealed worn screws and bearings. Very little of which has shown up in the vibration monitring. Trending is difficult due to the fact that the unit is on and off load frequenctly, however analysis of the spectra on (or as close as possible)indicated multiples of 5x rpm in the acceleration envelope spectra (with apparent sidebands of 30 and 50 Hz). Latest readings (two months following) indicated a increase in carpet level and no or little sign of the impacts. Could someone please explain what it is I am seeing? The compressor does sound a little noisy when on-load.

 

[DaveSco]

Further to previous, it should be noted that overall vibration levels rarely exceed 1g rms. Velocity readings in all directions <4mm/s rms with levels higher in horizontal direction (as would be expected)

 

[rob768]

< 4 mm/s RMS?? I'd leave the thing alone...
What you see in the probably the meshing frequency which, since it is not a since function, produces sidebands. The carpet level or noise is created by the random gnerated small inpacts due to surface irregularites or wear. Monitor that, and you monitor the increase in wear.

 

[DaveSco]

Rob, Thanks for your response, unfortunately I have followed this method in the past, and when the client decided that intrusive maintenance was required. The bearings and screws were found to be worn. There is also currently an issue with the slide valve, which seems to be stuck at 40%...I am currently looking into the possibiliy of a resonance issue with this. If you have ay further thoughts i would be interested in hearing them.

 

[MachineryWatch]

Dave,

My experience with screw compressors is that you need to do anything you can to get the data at max load and consistant loads to make valid diagnosis of the vibration data. Also, you must take a close look at all rotor/shaft speeds and Gear Mesh and Rotor Mesh frequencies. Document these and list all the mesh frequencies out to 3.5X the mesh frequency. Your overall levels that you mentioned mean nothing if they are not measuured out to high enough frequencies to include these higher frequencies, so you need to know how your instrument actually measures the overall level. After you know the frequencies you are dealing with, you need to make sure your instrumentation is up to the task. Does your accelerometer have a flat (valid) response out to about 3.25X Gear Mesh Frequency? Can your analyzer/data collector acquire data out to this same high frequency? Then you need to be sure you are using a method of sensor attachment that allows the sensor to function properly. At the high gear mesh frequency harmonics produced by some screw compressors, the standard practice of using a handheld probe or a magnetic base WILL NOT PROVIDE ACCURATE, REPEATABLE DATA. The only way to get good repeatable data at these frequencies (if you have a sensor that is up to the task) is to either stud mount it or use a good adhesive.

Finally, many screw compressors are very heat sensitive. Restricted gas intakes due to poor filter maintenance can cause internal temperatures to skyrocket. This leads to excessive rotor expansion and reduced rotor/rotor and rotor/housing clearances and even rubbing, not to mention bearing problems resulting from heat damage and poor lubrication.

Use spectrum analysis and waveform analysis for these machines. Take care to mount your sensors in the place where they have the most direct path to the bearings.

Skip Hartman

http://www.machinerywatch.com

 

[DaveSco]

Skip,
I can confirm that the data is collected using a epoxy-resin attached stud and accelerometer with a linear range up to 10kHz. This covers multiples of mesh on this unit. The positions at DE are as close to bearing locations as I can acertain from the drawings.

I have actually noted something interesting regarding the load period. I was under the impression that the unit was on load for only a short period of time (increased meshing noise), however monitoring of the motor current suggests that the unit is under load for a longer period of time than the noise exists. This is estimated, as the motor current readings are off-skid.
As far as I know the increased noise can be caused by four conditions;
Lubrication Failure (Modified Gauge Installed to Ensure Correct Pressure drop and hence flow. Also recent oil analysis suggests oil in excellent condition)
Liquids in Gas (Condensate has been found in oil line on previous occasion, but unsure how to check when in operation)
Excessive rotor end play (unsure how to check - again would expect increased vibration on bearing housing)
Loose or worn bearings (would expect bearing defects characteristics in spectra)

I plan to take the following actions;
1. Confirm via radio link when unit is on/off load and compare to noise emitted.
2. Change oil filter (DP 0.6) and get it checked for wear metals.
3. Speak to process personnel in attempt to clarify if condensate could be 'lost' into the compressor.

Any further veiws or suggestions would be appreciated.

 

[sms]

Screw compressors are very sensitive to slide valve problems, you need to get that fixed.....

-The future's so bright I gotta wear shades!

 

[DaveSco]

Sms, thanks for the post.
Could you, or others, confirm what a problem with the slide valve would actually involve in terms of the process etc, and therefore mechanical concequences of slide not moving? I would like to clarify this before suggesting remedial action. Cheers.

 

[DaveSco]

Further to previous reports on the compressor slide valve - I have managed to clear the position gauge glass enough to see movement. The valve moves between approximately zero and 42%. It is only when the value reaches 42% that the broadband vibration (accel up to 10kHz) and noise can heard. Is it possible that this is resonance condition? What steps can be taken to eliminate this condition?
Thanks in advance. Dave

 

[crjones]

The slide valve is meant to unload or load the unit changing the internal compression ratio. So when your compression requirement goes above the compression ratio you will generate more heat. Usually the noise that you hear is a transitioning noise as it rebalances itself.