Liquid ring compressor issues

[Guest078960594]

Hi,
first time poster, long-time lurker. This seems like a good place to bounce ideas since we have already contacted the mfg and together we have come up with theories but no root cause.

Our flare gas recovery liquid ring compressor had two vibration alerts of over 10mm/s RMS, seemingly out of nowhere in the middle of regular operation (3 weeks between events). We didn't have a vibration trip then yet and after the "shock" the machine kept chugging as usual at 4-5mm/s of vibration, except for increased mechanical seal leakage (pressurized seal). After the machine was opened for inspection and seal change, there were marks of shaft bending and the vanes touching the casing cover. The vibration spectrum showed clearly the VPF as the highest peak with VPF+running speed harmonics during the events.
The compressor is a 2 stage machine pressurizing from 0 barg -> 5,5barg, inlet pipe is 8". The whole unit is 4 months old.

The prevailing theory is a liquid slug ("heavy" HC condensation? water?) which flooded the compressor. During the events the discharge separator level was recorded rising, which would indicate incoming liquid from outside the unit. We have a knockout drum without a demister net in front of the machine which didn't register any level change and there are no low bends in the piping from the drum to the compressor to collect. The system has a gas recycle valve from after the discharge separator and cooler back in to the suction line. As is probably normal with flare gas systems, we have many units feeding this recovery unit so hard to pin any process changes as the cause.

So, does anyone have experience with liquid slugs in LRVCs, or in flare gas recovery in general? Or is there a better explanation?
If anyone is actually interested I can describe in more detail

 

[zdas04]

I've seen these machines jump around a bit, but it has always been due to boiling in the seal fluid (restricted suction usually). I suppose a slug of hydrocarbon liquid could enter a process that is at max compression ratios and boil. I would look very closely at heat of compression. It is a very messy calculation for liquid-flooded machines. I've got an example for oil-flooded screws in Chapter 8 of my book, but I can't find an example that you can download for free.

[bold]David Simpson, PE[/bold]
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[zdas04]

I did find the arithmetic Oil Flooded Screw Temp Calcs so you don't have to buy my book to see how to calculate it.

[bold]David Simpson, PE[/bold]
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[Guest078960594]

but it has always been due to boiling in the seal fluid (restricted suction usually)

Interesting for you to say that, because the suction pressure did momentarily increase from 0kpa to 7kpa before going down again, around 1 minute before the shocks. I didn't want to lead on too much on my first post. This is the meat of our problem, and it sounds like a blockage between the pressure measurement and the compressor. Between the measurement and machine there is the recycle line junction and a non-return valve.

We use water as the seal fluid and it's recycled. The seal water is replaced at a rate of around 100kg/h. The seal water flow is 24t/h and temp 23c, and after the suction pressure bounce, the seal water amount increased with the vibrations simultaneously. So after whatever happened (blockage?), the machine started sucking in more seal water. The seal water amount went back to normal after the large vibration peak so it seems the machine somehow emptied itself, but not without damages.

I checked your link, if I understood correctly I can answer (trying to theoretically calculate discharge temp?). We do measure the KO drum temp. and discharge temp (normal: 30degC->55degC).
Now that I got an answer I'll look if I can get a trend graph attached here. Thanks for responding David and please hang on for a bit yet.

Trend

EDIT: attached a graph. Most interested what happened before the machine sucked in loads of water (orange line) and got shook. I wasn't able to add the suction pressure to the graph but it peaked inside the red rectangle. There is a clear amp draw increase before the event but we don't know why. I will try to get a better graph done as soon as possible.

 

[georgeverghese]

So you've got seal water mixed with some process condensate and this is being separated at some 2 phase separator - interface level detection ( condensate - water ) is always a challenge. Perhaps there are times when the seal water level valve has drawn down too much and pulled in condensate also. A sketch or PID of this unit showing this interface loop also may help.

 

[Guest078960594]

interface level detection ( condensate - water ) is always a challenge

Again very close. There were lots of issues with this measurement right after startup and they had to recalculate some parameters and change a float and whatnot. My knowledge on instruments is limited but I know it was exactly this interface level we were having issues with (showed constantly 0%, but has been working now (or at least showing a logical value) for a few months). The seal water has been sampled afterwards and has been clean every time though. I am having some issues with my local DCS software from where I draw the trends so I hope our IT department can get it fixed for me soon (IT dept? soon? yeah right).

I'll see what I can do with the PID... and I will ask about the interface instrument again.
Thank you George for responding.