Problem in crack gas compressor start up

[Mohammad Ka]

Hi everyone
During start up of an ethylene plant, discharge temperature of the crack gas compressor (fifth stage) was drastically increased, meanwhile very high amount noises like fast opening and closing of a valve was observed. Moreover the compressor faced with insufficient flow. In this case, opening of the antisurge valve was 100%.It should be noted that there is no FE on antisurge loop Of CGC stage 1 and no piping pockets in the suction. However, a pumping trap is installed in upstream of suction line CGC stage 1. Antisurge valve with a difuser is specified to be a low noise
Other trends could be listed as below:
1- Outlet temperature of crack gas from quench tower was increased about 4-5 deg.C.
2-Level of discharge and suction drums of the crack gas compressor (stage 4) were increased during starup.
3-Discharge temperature of 5th stage of the CG compressor was increased to 130 C.
4- Fluctuation of flow was observed in all five stages of the compressor.
5-Alongside the flow fluctuation some small fluctuinations of pressure drop was observed in caustic tower.
6- No vibration in CG compressor was detected by DCS.

The following activities to remedy have been done:
* Antisurge valve was opened and fully inspected . No failure was detected.
* All suction and discharge drums, check valves, orifices, cooling water exchangers and caustic tower in the line were inspected and no chocking or unusual thing was seen. 
* About 1000kg mixed yellow oily water was observed in cooling water exchanger of 5th stage of the crack gas compressor. Antisurge valve is branched from outlet of this exchanger. The water was completely drained. 
*Stage 5 of the compressor was inspected by a borescope camera and no failure was detected.

After these remediations and start the plant for the next three times, the noise, insufficient flow and increased temperature were again repeated.
I was wondering if anyone could propose a solution? 

 

[Mohammad Ka]

Here is a summarized starting sequence of CGC:
1-Before starting check that anti-surge valves are fully opened.
2- When turbine reaches minimum speed, DCS operator adjust opening of anti-surge valves based on the deviation from surge line.
3- Before closing anti-surge valves, open gate valve on discharge line stage 5 to flare to control the discharge pressure.
* At this point, discharge P & T stage 5 increased and flow fluctuated. Anti-surge valve 2 remained fully open while anti-surge valve 1 closed to 60%.
4- Ramp up the speed turbine to normal condition, close the anti-surge valves and close gate valve on discharge line stage 5 to flare.

 

[rotw]

* At this point, discharge P & T stage 5 increased and flow fluctuated. Anti-surge valve 2 remained fully open while anti-surge valve 1 closed to 60%.

After flow fluctuations were noticed (are these fluctuations usual or not?), I understand operator continued turbine ramp up until normal operating speed is reached (questionable?). Is this correct?
Then compressor at nominal speed starts feed forward flow to downstream system and anti-surge valve was gradually and finally closed by operator with anti-surge control system active. Is this correct?

So what did happen next?
Pressure continued to build up, flow at suction reduced and temperature increased at stage #5; this initiated trip (ESD or else?).
Was the trip on discharge temperature high, and was it discharge temperature of stage#5 or of another stage?



Life is not about waiting for the storm to pass. It's about learning dance in the rain.

 

[Mohammad Ka]

In this case the flow fluctuations are unusual. When the flow was fluctuating, the discharge temp stage 5 was increasing drastically so the operator shut down the CGC. It did not reach to normal speed.
Your second point is correct for normal start up to reach normal operation.
Exactly as you mentioned, pressure build up & flow reduction & increasing temp stage 5. However there is ESD on high temp, DCS operator tripped the compressor. The main reason to trip CGC is high temp stage 5.

 

[rotw]

Some other thoughts:

- To me it looks like the gate valve to flare is there to help the compressor relief the pressure (similar to a blow off) and force the compressor to start along the choking line / far from surge. An anti-surge valve fully open that is sized too big can choke the compressor which is not recommended all though it is good for dynamic response anticipation; so the gate valve to flare looks like a way to regulate / fine tune the path of the compressor during ramp up so it goes along the choking line without putting the compressor into deep choke neither. Maybe the capacity of that gate valve was no more sufficient.

- Consequence of surge is not only vibations and noise but a surge cycle could lead to a drastic increase of temperature at the suction. So I do not know if this can be correlated to your statement in the OP: "fast opening and closing of a valve was observed"

- It is also needed to know the design of the anti-surge control system. Do you have control line that moves dynamically? Do you have safety line that would put the valve fully open if a certain threshold of the deviation is crossed. Is there a surge counter?

I maybe on the wrong track, but just as brainstorming exercise might be good to have a broad look at things.

Life is not about waiting for the storm to pass. It's about learning dance in the rain.

 

[georgeverghese]

Re your responses,
c)I find it strange that your recycle line take off to ASV on stage 5 is taken from aftercooler discharge and not from downstream of stage 5D drum. So there would be lots of liquid in this recycle line, causing slugging and pressure fluctuations etc, especially when you are injecting BFW or wash oil in to stage 5 casing - is this ASV recycle line a free draining line? Standard practice is to install this take off from downstream of stage 5D drum, so you will then have only gas being recycled.
3)Yes I was asking about heat tracing - okay.
Your response to (1) about the chattering on the check valve on caustic tower exit line brings up another suspect cause: Maybe the cracked gas feed to CGC is too low in mol wt to enable getting to the final discharge pressure, and this is most apparent in stage 5 - the compressor has pushed it self up to max speed on startup and is then too close to the surge line. Hence the stage 5 ASV at 100% open. This also matches the high temp on stage 5D - low mol wt gas causes high temp. Compressor is close to or on the surge line and pressure fluctuations occur on stage 5 and stage 4 ( and then reflected back all the way to stage 1). Have you checked the mol wt of this cracked gas? If this is found to be true, then maybe there is something wrong with the exit gas from the furnace. Is it exiting at the correct temp range or is there too much H2?

 

[Mohammad Ka]

rotw
I believe the capacity of gate valve is so enough based on the process data sheet and previous runs.
The suction temp stage 5 did not increased so much. Field operators reported "fast opening and closing of a valve" which it was apparently the check valve after caustic tower and before suction drum stage 5.
There is a control line, a safety line and a surge line in the anti-surge control system. The control line is specified as being at a distance of 5 % of the volume flow along a performance curve represential constant speed. When this line is reached, the controller opens the limit recycle valve. The safety line which is located between the surge line and the control line and when is reached, the control system is overridden by the valve jump.
Can you explain more by surge counter.

 

[Mohammad Ka]

georgeverghese
Its an important point that you mentioned but the temperature of stage 5D drum is 15C. The anti surge line goes to suction drum stage 5 and the liquid may remove in this drum. However, there is a drain in low point of after cooler discharge stage 5.
The mol wt of the CG and H2 conc. in feed during the previous start-up are 20.05 and 32.53 mol%, respectively. The recommended values by vendor are 22.14 (MW) and 31.9 mol% (H2). Moreover, the temperature of the exit gas from the furnaces is in the normal range.

 

[rotw]

For surge counter, you may want to view some description on page 33 of 35 of the following article:

https://oaktrust.library.tamu.edu/bitstream/handle/1969.1/159801/12_Jacobson.pdf

Life is not about waiting for the storm to pass. It's about learning dance in the rain.

 

[georgeverghese]

Agreed, 15degC is too cold for this recycle valve feed on stage 5. In standard practice, a new hot discharge drum would be required on stage 5D (upstream of PR kettle chiller). You can connect / relocate the stage 5 recycle takeoff to be just downstream of this hot 5D drum.
Mol wt and H2 content for cracked gas feed in the previous startup appears to be within spec range. Last few suspect causes would be :
a) There is some defect in the operation of the stage 5 ASV
b) There is some recent modification or adjustment made to stage 5 antisurge algorithm and /or antisurge controller P/I/D settings in DCS
c) Some deterioration in compressor polytropic efficiency on stage 5 and 4 - but you have checked impellers / stator diffuser vanes and all are clean.

 

[Mohammad Ka]

a) ASVs were sent to shop. No failure or plugging for both valves has been seen.
b)as far as I know from the last normal operation to now, modification or adjustment has not been made to CGC antisurge control or setting. However, it should be checked.
c) CGC stage 4 & 5 were opened. Every thing is ok.

Beyond these points, some questions have still remained:
1. Why the discharge temp stage 5 increased if ASV stage 5 is 100% opened?
2. The capacity of orifice RO in stage 5D to flare is 40 t/h and the FE in stage 5D represented 100 t/h. ASV stage 5 is fully open. Temp stage 5S is almost constant but temp & press stage 5D rise.
What happened to the other 60 t/h flow? If ASV is fully open, T&P stage 5 should decrease and CGC moves from surge line to right.
3. Why DP of the caustic tower is fluctuated? Why the check valve is chattered?

 

[georgeverghese]

Re your questions:
1. If stage 5 suction pressure and temp and cracked gas mol wt is all within spec, then stage 5D temp should be within spec also,since you say there is no fouling of the impellers/diffuser vanes on stage 5. So I cannot explain this also.
2. FE on stage 5D is within stage 5 recycle loop, so it will read higher than gas to flare. It just means that 60t/hr is going through stage 5 recycle line.
3. These are signs that stage 5 is operating inside the surge line for stage 5, if you ask me.

Presumably the steam turbine speed is operating in a stable manner. You also said earlier that stage 5D cooler was checked and there is no sign of leaks or fouling on water side.

Presume you are not over injecting BFW and wash oil into the compressor to the extent that it is interfering with compressor operation. Is there some injection quill for these?

Have you checked the TT on stage 5D? When was it last calibrated?

 

[Mohammad Ka]

TT on stage 5D and other transmitters are checked. All calibrated.

There are some paradoxes in operating manual and process data sheets documents:
1- When 3 furnaces are fed with feed stock and one in hot stand by conditions with all effluents diverted to quench water tower, the crack gas compressor can be started.
The capacity of each furnace is 30 t/h, for 3 furnaces is 90 t/h.
2-The CGC shall be designed to operate satisfactorily at a 40% turn down of the feed capacity i.e. 102.88 t/h. This condition will occur during start-up, shutdown or cracking furnace worst operating condition. This value is accordant to the compressor process datasheet for turn down case with 94.45 t/h as recycle flow in stage 1.
3-In the compressor datasheet, for start-up case it is clearly stated 175.8 t/h feed to stage 1 CGC and 55.1 t/h recycle flow.
4- Composition of crack gas for start-up case is 22.95% mol. H2, 42.78% mol. Ethylene (!) and 26.2% mol. Ethane. and for normal operation case is 35% mol. H2, 32.62% mol. Ethylene and 20% mol. Ethane.
In one of the previous posts I noted that " The recommended values by vendor are 22.14 (MW) and 31.9 mol% (H2)." but the corrected value is 22.95% mol during start-up. While the H2 composition from lab is 32.53% mol.

Here is the questions:
a) why the conc. of ethylene during start-up is high and H2 is low compared to normal ope. ?
b) The CGC shall be started by which flow, 90 t/hr or 102.88 t/h or 175.8 t/h ?

 

[georgeverghese]

a) Think this has to do with the average tubeskin temp in the cracking furnace : tubeskin may be cooler during startup and hotter during normal operation ( to get from low thermal duty to high thermal duty). Hence further cracking of ethylene at the tubeskin surface may be significant during normal operation (conversion of ethylene to propylene and C4, C5 = / C4,C5 dienes), which results in further formation of H2, despite the lower residence time. Hence H2 increases and ethylene decreases when going from startup to normal ops ( and as a result other cracked gas products increases).
b)Since you have a stage 4 to stage 1 recycle which is also controlled by low stage 1S pressure, in practice, you should be able to operate down to close to zero net feed flow from the quench tower ( even when you have only one furnace with say 50% design flow ie 15t/hr), assuming the stage 4 ASV is suitably sized to provide this recycle flow. You may pls check that the ASV on stage 4 can do this (ie. be able to recycle approx say 170-200t/hr when at say 90% of full speed-also refer to the performance curve for the compressor). The same should be checked for stage 5 ASV. There is a delicate balance between this high recycle flow during emergency ramp down in cracker furnace output vs the low normal operating recycle flow - turndown on control valve is approx 20:1 and can be stretched to 30:1).

 

[georgeverghese]

Based on this recent info, think the prime suspect for this is now the ASV on stage 5 - it is too small to allow stable operation at full discharge pressure with a net fresh feed of 40t/hr. Reasons for these are :
a) The furnace total effluent of 90t/hr seems completely out of order with the compressor capacity of approx 230-260t/hr.
b) Operation of stage 5 close to or inside surge line has resulted in high stage 5D temp due to low polytropic eff close to the surge line and possibly also low pressure on stage 5
c) Surging of stage 5 is evident from chattering of exit valve on caustic tower.

To address this problem we have 2 possible solutions
1) Live within the recycle capacity constraints of the current stage 5 ASV. So in case you wish to operate with a startup flow of 40t/hr, operate the flare valve such that 5D pressure is deliberately kept low and stage 5 ASV is not fully open. You can only ramp up pressure on 5D when there is sufficent fresh feed from the furnaces.
2)If you wish to ramp up pressure at 40t/hr and lower flow, then it is obvious we have to change out the trim on stage 5 ASV to allow more recycle. Ask a local consultant to resize this valve for you for the lowest furnace feed flow you wish to operate at. It is also better to enable a control scheme similar to stage 4 ASV ie. allow the ASV to recycle back to 5S also on detection of low pressure on 5S. Then stage 5 ASV will be acting as both capacity recycle and antisurge service.

It is also strange that though stage 5 is operating very close to or inside the surge zone, the compressor has not tripped-some one has disabled the surge trip on stage 5?

 

[Mohammad Ka]

Dear georgeverghese
After opening and checking the CGC, opening and fully cleaning the caustic tower, the plant was started. This time the compressor started with no problem and the plant is in normal operation now.
since no obvious evident has been detected in the CGC, I checked the trend of caustic tower DP. During the normal operation the DP in middle of the tower (DP=0.2bar) is above the spec (DP=0.05bar). Before cleaning the tower and during the startup when the speed was increased to min. speed the DP rose to 0.152 barg & fluctuated trend (First start-up), 0.151 barg & fluctuated trend (Second start-up). After using wash oil to clean the tower, DP rose 0.304 barg & smooth trend(Third start-up), 0.492 barg & smooth trend (Fourth start-up). After opening the tower ad fully drain the tower with water and wash oil, in the recent successful start-up the DP rose to 0.04 barg & normal trend.
It seems the middle of the tower was fully plugged during normal operation and using wash oil could not remove the plugging and more ever compact the fouling inside the middel packing. The channeling of crack gas inside the tower was still remained. When the tower was opened, and fully drained by water and wash oil, the significant amount of polymer was removed and the packing was cleaned and the channeling problem was solved.
Do you agree with this hypothesis?

 

[georgeverghese]

It is good you have found this fouling of the packing in the tower, and have cleaned up the packing. But what is the worst case total dp you observed previously when this packing was fouled? Even the worst case dp you have now quoted ie 0.5bar is small in comparison to the stage 5 compressor normal dp, so it cannot be the reason why the compressor was operating in the surge zone for stage 5, it is only a small contributing factor.
Suspect that in this recent smooth startup, the plant operators must have waited longer till there was sufficient flow ( ie >40t/hr)from the furnaces before they ramped up pressure on stage 5D at the flare valve.

 

[Mohammad Ka]

Dear georgeverghese
I checked the specifications of the ASVs and found these valves are suitably sized for recycling 170-200 t/h gas flow at 90% of full speed.
It should be noted that the plant has been producing since 2004 and the CGC has successfully started many times. I prepared the trends of speed, flow, P, T and dP of these previous runs which may help us to better undrestand the problem but because of some limitations I could not upload here publically. I was wondering if you could give me your email to send you these trends.

 

[rotw]

I am interested to have the trends, gas composition and the maps for each section, if at all possible!

Life is not about waiting for the storm to pass. It's about learning dance in the rain.

 

[Mohammad Ka]

rotw,
How can I give you the data?

 

[hacksaw]

At the bottom of "Reply to this Thread" you'll see "Attachment"
next
"Click Here to upload your file to ENGINEERING.com"

you will have a search window to the files