centrifugal compressor antisurge loop (return line tie-in) 1

[rotw]

Hello,

I have a flow diagram situation for centrifugal compressor line as follows;

From upstream to downstream the gas flows first through a suction knock out drum, gas is compressed and then cooled (after cooler); after that the export gas goes through a final knock out drum and is then exported to another unit downstream (for treatment).

The process gas is rich natural gas and the machine is feed service / boosting basically.
This is a downstream segment oil and gas application. At the compressor suction knock out drum location there is enough margin to dew point when looking at temperature and pressure conditions and gas composition(s).

Currently the anti surge line return has the tie-in point upstream the suction knock out drum. While the take off is downstream the compressor after-cooler.

I am thinking to move the tie-in point at the compressor suction line from upstream the knock out drum (present configuration) to downstream the knock out drum. Thinking is that the gas which is recycled, when it is recycled (anti-surge line), shall return to approximately the operating temperatures and pressures of compressor at suction therefore margin to dew point should be okay so if there is any condensation occurring due to after-cooling it will recover to the vapor phase area soon after the pressure is dropped in the anti-surge valve.

Why doing this ? we do have two identical trains (main and stand by) but one common knock out drum for both trains at suction (poor design, difficult and late to correct now) - I want each machine to have a dedicated anti-surge loop with dedicated automatic isolation valves per loop (and outside the loop) so that one machine can run normally while the other is operated with the isolation valves in blocked-in configuration. With a common knock out pot at suction there seems to be stability issue if the stand by machine is started and main machine is running as the later will absorb the flow of the stand by machine while being started pushing it to surge till it comes online during a switch between main and stand by machines.

Back to anti surge tie-in issue, I feel I am overlooking something by doing his and is not good practice to by pass he suction knock out drum in the anti-surge line return ; can anyone help ? Thanks

 

[georgeverghese]

Before we get any further with this proposal;
At the aftercooler, have we checked what the recycle gas composition / phases will be like with the lowest coolant supply temp, max corresponding coolant flow and lowest gas flow through the compressor, with gas being rich initially - this would be a rate check on the aftercooler with all tubeside and shellside fouling resistances removed.

 

[EmmanuelTop]

Are you absolutely positive that under no operating scenario there can be a condensed phase downstream of the discharge cooler? Once when liquid phase appears in the system, dropping pressure is not a guarantee that all liquid will re-vaporize. Particularly if water gets condensed - the amount of vaporization on the lower pressure side will be negligible. Similar might happen with some of the Hydrocarbons. Vaporization is not only thermodynamics - there needs to be a sufficient time for all liquid to vaporize, and antisurge circuits are very fast circuits with small retention time.

Can you provide a dedicated KO drum just for the antisurge stream? Definitely not a common configuration, but this will ensure no liquid will reach the compressor suction.

Also, how the emergency isolation valves (if any) for the two compressor trains will be arranged in this configuration? They are normally located upstream of the suction drum(s) and downstream of the antisurge take off point. You need to look at the sequence of events if any or both of the compressor trains undergo process or emergency shutdown and see whether the machines are adequately protected in all these scenarios, given the position and sequence of opening/closing of SDV's and BDV's surrounding the system.



Dejan IVANOVIC
Process Engineer, MSChE

 

[JJPellin]

I would not recommend this change. I have seen the results of liquid ingestion in a centrifugal compressor. Even relatively small amounts of liquid can cause a severe failure. EmmanuelTop is absolutely correct. We used to deliberately inject liquid into the suctions of our Alky refrigeration machines to cool the stream. Based on thermodynamics alone, the liquid should have all vaporized. We believed this to be the case until one of the compressors suffered a catastrophic wreck. A relatively small amount of liquid trapped behind a closed spill-back valve hit a running Coker wet gas compressor, causing the drive coupling to disengage from both shafts. Our high speed Penex hydrogen recycle compressors suffered a number of catastrophic failures when liquid condensed downstream of the knockout in a long run of un-insulated pipe.

Johnny Pellin

 

[georgeverghese]

Agreed, one amongst many pre requisites for any further work would be that the recycle line should be completely free draining, including the reducers u/s and d/s the recycle valve.

You would also need good characterisation of the C5-C6plus components in this rich gas for this heat transfer rate check on this aftercooler.

 

[LittleInch]

The thing I'm having difficulty with in your OP is the second last para.

The suction valve to the machine starting up needs to be open with the outlet closed and th A/S valve modulationg to prevent operation outside the surge line.

difficult to visualise what your conern is and wheterht his is theoretical or actual - "there seems to be stability issue.."

Needs a diagram and a better description of the start-up operation - which valves closed, flows, pressures etc

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[EmmanuelTop]

To summarize - it is a big "no no" [flush]

Dejan IVANOVIC
Process Engineer, MSChE

 

[rotw]

Hi Guys,

Sorry that I did not follow up ealier, ( too buzy with work...)
and thanks for your contributions.

EmmanuelTop :

- Are you absolutely positive that under no operating scenario there can be a condensed phase downstream of the discharge cooler?
Answer: no. In anycase the recommendations provided here are definitely to be followed, so I would not consider connecting the antisurge return line downstream the Knock out drum, it is very true that antisurge is a fast circuit so this is very risky.

- Can you provide a dedicated KO drum just for the antisurge stream?
Answer: it is an option but in reality I am afraid it won't happen.


- Also, how the emergency isolation valves (if any) for the two compressor trains will be arranged in this configuration? They are normally located upstream of the suction drum(s) and downstream of the antisurge take off point.
Answer: it is arranged as you have described.

Littleinch:

Again sorry for delay.

"The suction valve to the machine starting up needs to be open with the outlet closed and th A/S valve modulationg to prevent operation outside the surge line.

difficult to visualise what your conern is and wheterht his is theoretical or actual - "there seems to be stability issue.."

Answer: To describe the configuration in short (I could post a diagram but need to find more time):
There is one suction Knock out drum common for two identical machines arranged in parallel.
Each machine has its dedicated anti-surge line BUT both antisurge return lines connects at the same tie-in point which is upstream the suction knock out drum.
So problem is: When the first machine is already running and the second machine needs to be started - which is normally done by recycling the flow till machine develops sufficient pressure to open the discharge check valve - we could be facing some dynamic problems as the recycle paths of both machines are not completely independant.

 

[awhicker84]

What is everyone's thoughts on the risks of having heat tracing on the piping anti-surge line?

Would that be sufficient to allow him to move the tie in point?

 

[georgeverghese]

No amount of heat tracing will help to clear the deficit in superheat created by the stage cooler in this recycle line, when it is operating at the duty point described earlier.

 

[engrjalbani]

Can you provide estimate how far ASV loop is from compressor? I assume there is no hot-bypass valve,which is always short loop connecting compressor downstream to upstream with no gas flowing thru cooler and suction KO. ASV can be considered at downstream of suction KO only if there is no probability of condensation/liquid formation, and it is very near to compressor.

I am also confused, how do you manage isolation of each machine, whether isolating valves of each train are installed upstream of suction KO drum or otherwise!