Turbine-Driven Centrifugal Compressor Operation 5

[EmmanuelTop]

I have experienced something that I consider strange in regular centrifugal compressor operation. I would be grateful if someone of forum members could clarify these issues:

Turbine-driven refrigerant compressor (isobutane) in alkylation unit is suffering from frequent surging when operating in automatic mode. Automatic operating mode means the following:
- antisurge flow controller is in automatic mode
- suction pressure controller is in automatic mode (suction vessel PC connected to high pressure steam servo assembly, which regulates the RPM of compressor by manipulating steam flow into the turbine; steam is condensed under vacuum)
- suction temperature is always constant, meaning that composition of the compressed gas is also unchanged

At minimum alkylation unit capacity, compressor operates at 95% of maximum RPM, developing polytropic head 80% of design value (?). Lowering the RPM pushes the machine into surge region and raises the suction pressure, so the operators found that it is better to run the compressor with almost maximum RPM in manual mode, in order to have relatively smooth operation of the plant. This somewhat causes suction pressure to vary with time, but with no significant consequences.
What surprised me the most is the following:

1) With this parameters I described, antisurge FCV is open 52%. Polytropic head is 80% of design value, as I said.
2) Lowering the RPM from 7000 to 6800 RPM does not affect suction and discharge pressure (?), but it causes antisurge valve to open further, up to 56%! Moreover, machine goes into surge cycles.
3) Switching from manual to automatic mode of RPM control (via suction PC), makes incredible changes in compressor operation: relatively smooth operation is turned into surging cycles, so the automatic operation is completely abandoned.

My questions are:

1) If actual gas composition, suction pressure and temperature are as designed, why cannot we achieve design polytropic head? Is it possible that there is so little process gas (compared to spillback stream), that antisurge flow (52% valve open) pushes the compressor so much right off the curve, developing less head? Is it possible that machine is mechanically damaged, causing lower polytropic head at 95% of design RPM?
2) Why antisurge valve continues to open further when RPM is reduced, if suction and discharge pressures are unchanged? Isn't it contradictory, practically impossible? Less RPM should require smaller recycle stream (if being far enough from the surge point) in order to achieve the same head - that is what I (thought) I knew about centrifugal compressors.

Can you please throw some light on this.
Thanks in advance.

 

[sshep]

Q: Is your "antisurge flow controller" a characterized antisurge system; just a PID minimum flow controller; or part of a strategy being used to control discharge pressure?

To analyze the second problem of percieved lower than design head, start with a work balance on the turbine side and compare wih compressor side. Verify that what you calculate to be the compressor flow is consistent with the horsepower input and discharge (T,P) conditions.

best wishes,
sshep

 

[EmmanuelTop]

Hello sshep,

Antisurge flow controller is a "characterized antisurge system", as you said. It calculates minimum suction flow at actual RPM (with 10% safety margin) and acts on the spillback valve. It is mostly operated in "manual with backup" mode, because automatic operation pushes the compressor in surge cycles (even with RPM control on manual).

Concerning your second question, what can I conclude from heat/work balance of the compressor assembly? It simply operates far below his working curve conditions at given RPM.

Regards

 

[sshep]

The primary benifit of the force balance is as a very simple consistency check that the compressor flow, inlet and outlet conditions are actually as believed. If any one of Q, Ti, Pi, To, Po is not as you believe (higher than expected recycle, partial blocked suction, etc), the force balance will not close.

As this is a clean system (regrigerant), the system can be easily analyzed. I think you can solve this mystery.

best wishes,
sshep

 

[EmmanuelTop]

Hello again sshep,

I will check the force balance very soon. But still (maybe because of my long explanation), there is one thing that is definitely not in consistency with common sense. Let me describe it again, I 've checked today in the plant.

If there is really so little process gas and 52% antisurge valve opening is required to prevent surging (which means that compressor operates at only 10% higher suction flow - 10% away from surge point, at maximum RPM), why is there so dramatically reduced compressor discharge pressure, if suction pressure and temperature are at design conditions? It should be 6.46barg, and in real operation it is only 5.12barg! Compressed gas composition is as designed, and it has been checked in the laboratory (GC analysis). Refrigerant compressor is delivering much lower polytropic head than it should do, and consequently much lower discharge pressure (measured just at the compressor discharge outlet). Since process conditions, in my opinion, are not the cause of abnormal compressor operation, I was wondering if someone may have a clue what is going on. Also, when running the compressor in manual mode, the whole system operates smoothly; as soon as we switch to automatic suction pressure control (PC connected to turbine motive steam valve), the compressor starts surging. At all times, steam is condensed under vacuum and there are no significant changes of steam condensate outlet parameters.

Thanks again,

 

[sshep]

Can you share:
steam rate,
steam inlet and condensing pressure,
isobutane flow,
compressor inlet temp and pressure,
compressor outlet temp and pressure

best wishes,
sshep

 

[EmmanuelTop]

P(suction) = 0.105barg
T (suction) = -6C
P (discharge) = 4.600barg
T (discharge) = 54C
Steam rate = 5050 kg/h
Steam inlet pressure = 43.1barg
Steam inlet temperature = 378C
Steam outlet pressure = 0.17bar abs (-0.83barg)
Steam outlet temperature = 58C
Refrigerant flow = 12690 m3/h
Compressor RPM = 6760

Gas (weight) composition:
6.5% C3
91.4% i-C4
2.1% n-C4

Maximum rated compressor RPM is 7346, corresponding to discharge pressure of 6.460barg - at suction volumetric flow 10% higher than surge point flow.

 

[bspasov]

It does not look as valid process measurement - steam fow rate or refrigerant flow (at least they look like most suspicious ones).

 

[zdas04]

EmmanualTop,
I'm a little confused about the compressor discharge pressure. You say it should be 6.4 barg but it is only 5.12 barg. Are you trying to maintain a specific discharge pressure (say with a backpressure valve)? If not, the the discharge pressure will be suction pressure plus the sum of the system resistance to flow at the actual flow rate. For the pressure to be 20% low then either you've got your system resistance or your mass flow rate wrong.

David

 

[EmmanuelTop]

Hello David and bspasov,

Here is additional information: recycle (spillback) stream enters the compressor suction trap without previously being cooled, but after mixing with sub-cooled liquid refrigerant (from liquid refrigerant accumulator drum) in special vapor-liquid mixing nozzle. The bottom line: suction temperature is not changing by time, as well as suction pressure - with compressor operating in manual mode (fixed RPM). So generally, as you can conclude, it is very much stable process.

There is a sketch of refrigeration section available at:

http://www.imagehosting.com/show.php/70646...ection.jpg.html

Note that all downstream control valves maintain resistance to flow practically unchanged: LIC10 is in manual mode, PDIC (vapor bypass) is on automatic mode. Everytime when LIC10 is closed by only 0.2% of valve controller output, compressor starts surging. In my opinion, for LIC10 to operate in the automatic mode, automatic operation of compressor suction pressure controller needs to be provided; otherwise, small changes in LIC10 valve % opening affects compressor load (required polytropic head).

As you can see, there is no backpressure PC valve at the compressor discharge section. Downstream pressure is "fixed" by condenser outlet temperature and pressure drop of compressor downstream equipment. I did not notice any significant changes in this particular pressure over time (PR10 on the sketch). For emergency purposes, or in the case of propane accumulation in the system, there is HCV connected to flare system (the pump for liquid C3/C4 stream transport to depropanizer is currently out of service).

My question regarding compressor operation is: if compressor operates at maximum RPM on manual mode, and if inlet flow is equal to surge point flow + 10% safety margin, will it develop the same polytropic head at the compressor discharge nozzle, regardless of changing downstream resistance? Or not? And why?

My question regarding instruments tuning and antisurge control: what causes unstable operation when switching from manual to automatic mode of suction pressure control, if process is stable while operating in manual mode?

sshep: concerning the steam expansion/isobutane compression work balance, I didn't manage to come up with any conclusion. Calculations show much higher compression work compared to steam expansion work, for given (measured) flows. Do you have any further suggestions?

 

[EmmanuelTop]

Link to the sketch seems to be non-functional.

http://www.imagehosting.com/show.php/706461_RefrigerationSection.jpg.html

 

[dcasto]

You have backpressure valves on the compressor, both LIC10 and PDIC. Lets say LIC10 is wide open as it should always be (why is it there to begin with?). Next PDIC gets a signal to open (I don't know why), the discharge pressure on the compressor will rise because hot gases go into the REF ACC and it pressures up.

The arrangement of air coolers and water coolers has minimized your cooling water and MAXIMIZED your horsepower on the compressor, that usually isn't the most efficient way.

Because you can't give the exact composition, the alkylate's mole wt and its head and actual pressure can cause surging along with the action of LIC10 and PDIC. Most people take the flash spill back as you have drawn it for a new plant under maximum loading. If the unit is way under loaded, then you need some cooling because the hot minimum flow also changes the temperature, which means the density changes, which means the discharge pressure drops and you will cycle up into surge too. So, some of the minimum flow needs to come from the REF ACC tank as a liquid. The amount from the the REF ACC is controlled by a TIC which allows injection of liquid C4 with the hot gas from the spillback valve.

 

[EmmanuelTop]

dcasto,

Can you make a hand-drawing of that modified scheme and post it somewhere, for example at

http://www.imagehosting.com

?

I do not understand exactly what do you mean, by only reading your post.
What happens with compressor discharge pressure (measured at the compressor discharge nozzle) when you close LIC10 a little bit (1-2%, for example)? Does it develop the same polytropic head?

P.S. You have refrigerant gas composition in one of my previous messages. It is very close to design composition.

Thanks,

 

[sshep]

I assumed that the compressor flow was actual volumetric inlet (please verify). I also made a slight adjustment to the inlet composition to get the inlet temp per your text data above (not the dwg data). From this I calculated:

Compressor Polytropic Effi= 0.725
Compressor Power= 925kW
Turbine Power= 1000kW (assumed 75% isotropic effi)

best to make a check of my calcs yourself.

This doesn't seem to be an unreasonable balance or compressor efficiency. I am curious about what the design effi was because the statement that the "polytropic head is only 80% of design" is not so clear to me as efficiency. Also due to compression ratio, if the actual suction pressure is slightly below what your instrument shows, the discharge pressure will appear to be much lower for the same power. If the suction pressure is really lower by even 0.15 bar, it would explain your lower discharge pressure. You should verify this pressure with a gage as close to the suction as possible.

The lower discharge pressure does not seem like a serious operational problem. As was pointed out above, the design looks like it would waste energy because you compress to higher pressure than would be needed to condense against air. It also seems like a lot of possible interactions among controllers- including some process considerations due to the open loop nature of the system. It doesn't seem surprising to me that that there could be instability controlling the suction pressure by speed for reasons unrelated to the compressor efficiency observations.

With regard to the surging, can you clarify how you calculate the compressor inlet flow is measured (or calculated), and the flow inputs to/from the surge controller?

best wishes,
sshep

 

[EmmanuelTop]

sshep:

Suction pressure is actually somewhat higher than designed (0.105barg VS 0.08barg). This should move the compressor away from surge conditions.

Surge control: total discharge flow is measured and then recalculated for suction conditions (p, T). This flow is compared to the minimum suction flow (at given RPM) plus 10% safety margin, which acts as the software controller output on antisurge valve. There is an option to run antisurge (spillback) valve in 3 modes of operation:
- automatic
- manual with backup (you can raise spillback flow above minimum calculated flow, but not below)
- manual (it is possible to open/close spillback valve in the full range 0-100%)

You said: "I am curious about what the design efficiency was because the statement that the "polytropic head is only 80% of design" is not so clear to me".
According to compressor ACFM-Hp chart, at maximum RPM compressor should develop 6.46barg discharge pressure (also given as tabular data), instead of 5.1 or 5.3bar - which is actually the case (remember, we have even slightly higher suction pressure than designed). Gas composition is almost exactly the same as designed, so I really cannot find the process-side reason for so much smaller discharge pressure at maximum RPM of the compressor.

When analyzing compressor problems in the past few days, actually I became so confused with the facts (and measurements), that I have suspicious about - at least I think so - fundamental principles of centrifugal compressor operation.

I am in love with this Alkylation unit, that is for sure!

 

[dcasto]

here is the image: note how I dropped the discharge pressure which is lower head and volume required by the compressor along with energy.

http://www.imagehosting.com/show.php/710554_refrig.bmp.html

 

[sshep]

With respect to possible reasons why the compression ratio on a centrifugal compressor seems to be low we have seen everything from demister pads jammed in the suction to fouled (or mechanically damaged) rotors and diaphrams. My emphasis previously was to insure that the data was consistent in order to draw a conclusion.

There is nothing at this point to suggest that any plant data is bad, on the otherhand there is no reason to shutdown early to check the compressor for damage.

The surging is a seperate issue. Can you get speed on control by using extra spillback (manual with backup)?

========

This is my own view of the operation if it helps. NOTE: I am speculating about your system as most of my refrigeration experience is with self regulating variable area condensing (as Decasto's looks to be). There is nothing in your sketch to indicate a liquid level in the finfan.

Speed control is normal in a variable area condensing case as it regulates the system so that the compressor is handling only the flow needed for the refrigeration load and the refrigeration temp is also controlled (vaporizing pressure). Your system runs at constant speed so suction ressure floats. My speculation is that the condensing pressure is set by the finfan air flow (and temp) and exchanger area (assumes area fixed=no liquid level in finfan). Consider the the fixed position finfan control valve pipe as extra pipe resistance since the position is fixed. There is a pressure (around 5barg) at which the required flow will be completely condensed over the available cooler area- i.e. if pressure were lower not all would condense and pressure would rise, if discharge pressure were higher then all would condense and the pressure would drop. The system in your constant speed case becomes self regulating through small (but important) suction pressure changes against the refrigerent condenser. If I am correct about the condensing pressure at a given refrigerent load being fixed by the finfan area and air flow, it may not ever be possible to also control the suction pressure. Such a case seems over specified since refrigerant flow (process heat load), inlet (speed controller) and outlet pressure (finfan area) are all set independently.

The net result of this to efficiency is that I think you should be conceptually wondering why the suction pressure is higher for the constant speed case rather than asking yourself why the discharge pressure is low. This was my confusion about what the "only 80% of polytropic head" really met. I know that you believe the suction pressure is low, but this would be one of the measurements that I would field check as it wouldn't need to be that much off to make everything ok. In addition my estimate of dew pt temp for your vapor mix at 0.105 barg was about 5C higher than your data, suggesting maybe you really do have a lower suction pressure (on the otherhand it may mean nothing but that my vapor pressure properties are a bit off).

Thanks for the sketches and all. I am sure that someone can give a better explaination about the speed vs pressure profile than I gave. If I am wrong then I would like to know what you think sets the compressor discharge pressure in your case. Anyway don't give up on your compressor theory, just try and put it into the context of the process (refrigeration loads and condensing conditions).

best wishes,
sshep

 

[EmmanuelTop]

dcasto,

Could you please post a larger picture of that modified flow scheme? I'm having problems in figuring out certain numbers and streams. Also, about new TIC - where does it take the temperature signal from? I apologize in advance, but something is missing in my analysis of this conception you proposed.
Wouldn't be better to install a PC valve instead of HIC valve in depropanizer feed drum (refrigerant blowdown drum) and fix the condensing pressure of compressor discharge stream? This service (refrigerant blowdown to De-C3 in FCC Gas plant) is not functioning because of depropanizer charge pump minimum flow problems and choking of the depropanizer with additional feed, which essentially does not contain C3 when blowdown is continuous. Without fixed discharge pressure, pressure rise in the refrigerant blowdown drum during day is causing compressor suction pressure to rise equally - because compressor is operating on manual, remember? Propane accumulation is very slow but consistent, so I think some sort of purging is necessary - through PC valve instead of HCV... With small losses of i-C4 to the flare system. Could you please explain your concept in more details?


sshep:

Yes, we are using manual with backup type of antisurge control. I do agree with everything you have written in your last post.
About discharge pressure what do I think? Here is my concept:
- Compressor is always on manual (fixed RPM)
- Antisurge valve is in "manual with backup" mode, which means that surge conditions should always be avoided, if system is well-tuned
- Equipment downstream pressure (refrigerant blowdown - depropanizer charge drum pressure) is set by condenser outlet temperature and C3 content in compressor gas. Without purging, C3 content gradually rises regardless of condenser outlet temperature. This means that downstream pressure also swings (rises) by time, because it is impossible to hold C3 in liquid
- Since compressor operates with fixed RPM, suction pressure rise is the consequence of downstream pressure rise
- Having the same suction temperature and higher suction pressure, ACFM drops (I am very suspicious about fast-acting of antisurge controller) and, probably, moves the compressor toward surge point. This is something that happens always in AUTOMATIC mode of suction pressure control, but not in manual mode
- I think without downstream pressure control (PCV) it is impossible to run the plant in regular way, since propane accumulation (even in 7-day cycle) is a fact.

 

[EmmanuelTop]

One additional observation, I don't know if you will agree with me:

In existing equipment configuration, if compressor discharge pressure floats (depending on system resistance and backpressure from the blowdown drum), having suction pressure controller in the automatic mode will continuously speed-up the compressor untill it trips off. Because of C3 accumulation and non-functioning of the liquid purge to De-C3 tower, I think it would be very difficult to maintain constant suction pressure without PC valve to flare system. Do you agree with that?

Regards,

 

[dcasto]

http://www.imagehosting.com/show.php/729943_refrig.GIF.html

The TIC watches the temperature in the REF suction drum. You need to set the tempearture about 5 to 10 C above the dew point temp of the feed stream. I'd never seen this before until I was troubleshooting a 20,000 HP propane refrigeration system with 2 9,000 HP gas turbine single shaft 6 stage centrifugal compressors with a 2000 HP steam helper/starter turbine. They had a CCC antisurge system in manual mode, We called CCC in for training and returned the system to automatic and all the operational problems went away.