Gas Compressor Performance Deviation and Evaluation

[danschwind]

Hi everyone,

I’ve been comparing my compressors with the expected performance curves supplied by the OEM (I have multiple MW curves) and things are going well, where I often see a degradation on the compressor both on a polytropic head basis as well as on a polytropic efficiency, as expected for a compressor being subjected to a dirty gas that is operating for quite a few years.

For some compressors, I find that the current operating point is very close or even slightly above the theoretical curve supplied by the OEM, which I relate to a normal variance in the expected to real performance curve. I even expect that the theoretical curve to be a little bit conservative (less performance) than the real due to performance guarantees and whatnot.

However, for a particular set of compressors (the second stage of a 2-stage bundle machine), I’m finding values that are too good to be true. For other identical set, the values are within what I expect: some degradation compared to the theoretical curve. For the particular set, the current operational point seems to be so much greater than the theoretical curve that it seems I have a super-compressor in hand, where the current operating point in the performance curve is closer to the curve for a gas 5 kgmol/kg heavier than my current gas (!!!), even though the polytropic efficiency itself (which I calculate with the temperatures) is degraded. Note that the first stage of the same bundle has a behavior of a degraded compressor as I would expect.

I’ve asked for calibration of the suction flowmeter and the results were the same. Only thing pending related to measurements would be an orifice plate inspection to check any blockage, but it will take some time for this to happen (probably a few months).

I’ve been wondering if anyone has gone through the rabbit hole of evaluating their compressors and found results like this as well…was it due to bad field measurements or was the compressor really a super-performer?

I can post the overall data here if you guys really want to do some calculations but that is not really my point…just curious about others experience with this.

Best regards,


Daniel
Process Engineer
Rio de Janeiro - Brazil

 

[georgeverghese]

"where the current operating point in the performance curve is closer to the curve for a gas 5 kgmol/kg heavier than my current gas"

You mean 5kg/kgmole heavier.

Presume you're observing the derived polytropic head actually developed on this 2nd stage at this speed and flow is greater than that predicted on the poly. head - flow curve ?

Where is the suction flow meter, upstream or downstream of the tee in from the stage 2 antisurge recycle line ? Is this line completely free draining or could entrained liquids be held back at the upstream side of the FE ? Maybe the little drain hole at the bottom of this FE may be bunged up ?

 

[danschwind]

"You mean 5kg/kgmole heavier."

Yes, sorry for the brainfart.


"Presume you're observing the derived polytropic head actually developed on this 2nd stage at this speed and flow is greater than that predicted on the poly. head - flow curve ?"

Exactly. For this particular stage I'm computing around 4.6 kJ/kg more poly head than what the performance curve for this MW predicts.

It is kinda of balancing out in the end, because for the first stage I'm getting a very high downgrade from the predicted curve.

For the other 1st stage/2nd stage machine, I'm computing values that make way more sense. Both stages delivering 2.6 to 3.0 less poly head than what is expected from the curve.

I know that every machine is different and the predicted curve is only that, a prediction, but nevertheless this being so much more than what the original curve predicts is bugging me out.


"Where is the suction flow meter, upstream or downstream of the tee in from the stage 2 antisurge recycle line ? Is this line completely free draining or could entrained liquids be held back at the upstream side of the FE ? Maybe the little drain hole at the bottom of this FE may be bunged up ?"

The flowmeter is downstream of the recycle, so it is reading recycle+non-recycled gas at all times (so it is reading what the compressor is seeing).
The flowmeters are at a higher point that should free drain, yes. It is also downstream of the scrubber.
The lines are dirty though. 1st and 2nd stage are seeing whatever is coming from the wells and I'll admit the scrubbers are a bit lacking in the performance deparment (I'm already revamping them). The thing is, all compressors have pretty much the same layout, pipework, etc. This is the only one that is over-performing by a lot.

Oone of our 3rd stage compressors is also overperforming, but by just a tiny margin. And this guy at least already gets clean gas (after Amine and Glycol, at least 3 scrubbers and 2 coalescing filters, plus the columns, including a wash water section on amine contactor top). This is not bugging me at all.

Particularities with the flowmeter will only be found when/if we inspect it...dunno when this will be possible. Everything is a possiblity here, and I'm definetely not trying to pinpoint 'exactly' what is. Just wondering really if this over-performing was observed by others or if this is either the only in the world.

Thanks for the feedback, George.


Daniel
Process Engineer
Rio de Janeiro - Brazil

 

[georgeverghese]

The suction flow to the stage 2 compressor would obviously be press / temp and mole weight corrected to make this comparison.

From memory, a difference of only 4.6kJ/kg on poly head is typically only a small fraction of the total developed, possibly even less than 1%.

 

[danschwind]

Im calculating the flow rigorously through the orifice plate dP. Everything is corrected, yes.

It is around 4% of the total poly head...a small number, yes. Still, the difference in potential flow (at same suction and discharge pressure) is pretty much an additional oil well, if looking this in isolation (as mentioned, the first stage brings everything down).

Daniel
Process Engineer
Rio de Janeiro - Brazil

 

[georgeverghese]

If CO2 and N2 content in feed gas are somewhat different compared to the design case, then the polytropic head curve will shift slightly also - this is related to the change in the polytropic coefficient (not the same as polytropic exponent). Generalised compressibility correction for feed gas variations in mol wt built into the antisurge controller also may be slightly in error for CO2/ N2 composition changes. So may the isentropic coeff k. It would be possible to make manual corrections for changes to k and z, but I dont think there is a way to get the corrected value for the polytropic coeff. - usually obtained by the compressor manufacturer from actual tests.

 

[danschwind]

CO2 and N2 are not the exact same as design but have not varied wildly. They are the same for all machines though and this over-performance is only observed for this particular one. Don't think this is it!

Daniel
Process Engineer
Rio de Janeiro - Brazil