Centrifugal Compressor Performance 1

[AndreChE]

Dear all

We have been operating our steam cracker for 5 years and the turnaround was delayed one year due to a plant revamping preparation.

I, as a process engineer, am responsible for the compressor follow-up. Suddenly, in the last two months the first two (of five) stages have been loosing performance with the (aprrox) the same flowrate.
The compression ratio decreased and the temperature ratio increased --> fouling... but..suddenly? not progressive

It has lost 9% efficiency in two months and it continues with decreasing trend. We have an antifouling injection for 4 stages (except the 4th one). The rate was increased twice but without sucess. The suction temperature is quite low (32ºC) compared with the past (for volume optimization).

Other, bigger, problem is exactly in the 4 stage. It is "highly efficient". Suction temperatures are quite low and upstream heat exchanger is water limited due to the low T. Very low RT at this stage. I cannot understand what is happening in these pressure range (10 bar) for this cooling. I have been watching Joule-Thomson efect in a downstream drum LIC but this is not the reason. Amazing behaviour but not understandable. Last time this condenser was oppened (due to very high DP) the fouling, after plant shutdown and HE opening, was very cold, like 2-4ºC.
I am now collecting raw gas lab analysis to check isentropic exponents and to try conclude something more.
What can be happening? Every help...is precious to troubleshoot.

AndreChE

 

[UmeshMathur]

You might also check the compressor seals in case excessive leakage is happening which would bypass a fraction of the discharge gas back to suction. That would be one explanation for loss of compresion ratio, increase in temperature ratio, and corresponding decrease in efficiency.

Also, a thorough check of the calibration of ALL your instrumentation would be very useful to eliminate measurement errors as a source of the confusion.

Generally, fouling of charge gas compressors in ethylene plants does increase noticeably after a year of operation and, especially for a liquids cracker, a five year turnaround would be considered much too long.

 

[AndreChE]

That's a reason, I agree.
What do you think about the low suction temperature?

AndreChE

 

[UmeshMathur]

Hi, Andre:

It is hard for me to understand the problem as you perceive it without some of the following details regarding the application:

(1) First stage suction gas T, P, estimated flow and composition.
(2) All interstage suction T&P.
(3) Final stage discharge P.
(4) Cooing water supply T.
(5) Vendor's polytropic head and polytropic efficiency curves (new machine) for all stages. The curves for polytropic head v/s actual suction flow should be from the surge line to stonewall for each RPM. Similarly, the polytropic efficiency should be plotted v/s suction flow
for each RPM.

Based on this, one could calculate both the expected performance for a new machine and the actual performance for comparison.

I also don't understand your reference to the Joule-Thompson effect. This generally refers to the cooling of a fluid (usually vapor) when undergoing an isenthalpic expansion, e.g., through a valve. Saturated liquid flashing through a valve (your LIC?) to a 2-phase mixture is generally NOT considered Joule-Thompson cooling.

Finally, once you describe your system's performance in a bit more detail, we can discuss the apparent abnormal behavior of the 4th stage. It is hard to imagine that the performance of any charge gas compressor stage 5 years after the last turnaround can exceed its design basis.

I still think that a big part of your dilemma stems from inaccurate measurements of T&P.

However, if you do take the trouble to send me this data, I promise to help you analyze it properly.