13 replies to this topic

[tingyik90]

Dear all,

 

I have a question concerning control valve sizing. I have a 3-phase Separator A, separating vapour, HC condensate and heavy phase. MEG is continuously injected into the wellstream, so the heavy phase is a mixture of water and rich MEG.

 

The CV is for the Water/MEG stream. Separator A is at 25 bara, with downstream vessel at 1.2 bar, so the flow is actually choked at the control valve. It has roughly 75%mol water and 25%mol MEG and some dissolved hydrocarbon including 4000ppm CO2, 23 ppm C1, 1 ppm C2 and 1 ppm C3.

 

The CV is located far from Separator A and very near to the downstream vessel. With that, the inlet pressure of CV has dropped to 15 bar due to static and piping head loss. As it contains MEG and some dissolved HC, when the pressure is dropped, these HC components begin to flash, albeit in very low content. Naturally, the saturation pressure is 25 bara, and it is very sensitive wrt the vapour fraction e.g. it drops to 16 bara at 0.00001% mol vapour and 4 bara at 0.0001%mol vapour from HYSYS.

 

The question is, will this vapour pressure be a concern for such small quantity of vapor? I know from masoneilan sizing equation that vapour pressure is used in sizing a choked flow liquid valve. If I use the true saturated pressure of 25 bara (0%mol vapour), the Cv value will be much larger than considering vapour pressure at 4 bara (0.0001%mol vapour).

 

As it is preferable to avoid oversizing the control valve, can I ignore these small amount of vapor and obtain the vapour pressure by removing all these HC condensate from the stream in HYSYS, thus only considering pure water (75%mol) and MEG(25%mol). At my process temperature, the vapour pressure is only 0.25 bara.

 

Looking forward to your reply. Thanks!

 

 

[Zauberberg]

The amount of vapor at flashing conditions seems negligible. I don't think you will have any problem with that. Just avoid to have a control valve which will operate at 2% opening most of the time - this will create problems.

[tingyik90]

Dear Zauberberg,

 

Thank for the reply and sorry for my late reply. Cv value calculated can be quite different for each case e.g. Cv = 160 when at 25 bara vapour pressure and Cv = 130 when at 15 bara vapour pressure, based on my company sizing spreadsheet.

 

Consequently, we decide to go with the balanced design and go with 15 bara vapour pressure which is what we will likely to see at CV inlet although there is only very low volume of vapour. Thanks.

[Zauberberg]

What CV you get if you use vapor pressure of 0.2 bara? The only concern I have (and I have seen many, many similar cases in produced water service lines) that you will end up with an oversized valve, designed for maximum flow and high vapor pressure. Then in reality, when you face e.g. 20% turndown flow and no flashing at all, the valve will end up operating at 2% opening most of the time and wear itself off in a matter of weeks.

[fallah]

tingyik90,

 

I think you can go with the vapor pressure of water (75%) plus MEG (25%) i.e. 0.25 bara at process temperasture. The pressure at CV downstream is 1.2 barg (2.2 bara) well above that vapor pressure and create a reasonable margin will lead to having no flashing at downstream due to the vaporization of such ppm order of magnitude of hyrocarbons...

[serra]

I agree with Fallah,

in case of doubts for two phase flow you may consider a HEM model such that included in Prode Properties,

for this case I would simply calculate bubble point pressure and use that as vapor pressure,

it is not the best option but consider that ISA formulation is for pure fluids, not mixtures...

[Zauberberg]

To be absolutely sure that the valve will work, you can look through several design cases by using ValspeQ or Fisher Specification Manager software. The latter provides you with a convenient overview of design alternatives:

 

http://softadvice.in...an_Valspeq.html

http://www2.emersonp...andownload.aspx

[breizh]

Hi,

 

You may find good tips reading this paper .

 

hope this helps .

 

Breizh

Edited by breizh, 10 March 2015 - 12:35 AM.

[serra]

most manufacturers adopt ISA S75 (derived from old Masoneilan formulations)

ISA (in general) should be adopted only for vaporization of pure liquids,

for mixtures things are complex (something similar to PSV).

 

When (as for the case proposed by tingyik90) bubble point presssure is very low

one can adopt the formulation for liquids without considering vaporization,

 

another possibility (suggested in some old papers and frequently adopted by manufacturers)

is to consider separately vapor and liquid flows (i.e. Cv = Cv Vap + Cv Liq).

 

There are a few papers (Samson) discussing applications of control valves with liquid mixtures

at conditions where a large fraction of liquid vaporizes,

the procedures proposed for solving such cases are based on HNE-DS,

as alternative one could adopt a methodology based on HEM

 

I adopt HNE-DS or HEM only for difficult cases (mixture of liquids flashing)

these methods are available in Prode Properties and easily coded in Excel

[tingyik90]

Dear all,

 

Sorry for my late reply and thank you very much for your constructive feedback. Pdf by breizh and software by Zauberberg are very useful.

 

Below are the calculation result and conclusion:

1. By letting down the stream from 25 bara to 15 bara in HYSYS, a mixed stream with 5.1 kg/hr of vapour and 190,000 kg/hr of MEG/water stream is obtained. This stream has roughly 0.2%vol vapour only. Using corresponding mixed phase properties and vapour pressure of 15 bara (saturated), required Cv = 180. This is sized using 2 methods in my company spreadsheet, including one by serra (Cv vap + Cv Liq). Fisher Specification Manager software gave similar Cv.

 

2. Another way as discussed is to exclude the gas compenents in the stream, and obtain the stream properties as pure MEG/water only in HYSYS (ratio between MEG/water is kept the same). This stream will have a vapour pressure of 0.33 bara. The required Cv = 65 only.

 

3. However, by checking in HYSYS, letting down to 1.2 bara, the stream will have up to 82%vol of vapour, with up to 71%mol CO2 and 29%mol H2O in the vapour stream (rather unexpected for me). It is a flashing application and therefore, Cv = 180 should be applied to be conservative.

 

4. Note that if only considering pure MEG/water in HYSYS, the stream indeed does not flash upon pressure let down since the vapour pressure is lower than 1.2 bara.

 

I hope this clears the issue. Thank you very much for your help.

[serra]

3) "the stream will have up to 82%vol of vapour, with up to 71%mol CO2 and 29%mol H2O in the vapour"

 

a 82% vapor fraction with 71% mol CO2 seem quite strange considering cases 1 and 2,

are you sure about the results of isothermal (or, maybe, with specified H) flash ?

which is the feed composition ?

as said in my previous post in case of doubts for two phase flow you may consider models based on HEM methodology

[tingyik90]

Serra,

 

The inlet stream has a lot of components which are at ppm level. Thinking that it might be due to correlation limit of the simulator, I stripped down components which are less than 50 ppm, but that practically left me with water, MEG and CO2 only. Still, the flashing is significant and I can confirm it is isenthalpic flashing via a control valve unit operation. This is tested with HYSYS using Peng Robinson, Kabadi Danner and the Glycol package.

 

You might want to verify my results using other simulator. The conditions are as follows. 

 

Pressure: 25 bara --> 1.2 bara

Temperature: 77.81 degC

H2O: 74.8296%mol

EGlycol: 24.6995%mol

CO2: 0.4710%mol

[serra]

I think there is something strange in your result 3) see above message #10

 

for personal interest I solved your flash operation with Hin = Hout,

with two models (CPA-PR and Extended PR)

Pin 25 Bar.a  Pout 1.2 Bar.a

Tin 77.81 C Tout about 77.8 C

feed % (mol) H2O 74.8295 MEG 24.6995 CO2 0.471

 

with both models I get limited vapor fractions,

with CPA-PR 

Vapor Fraction (mol) 0.0053

vapor % (mol) H2O 21.9 MEG 0.06 CO2 78

 

if you compare this value (0.53 %) with your 82% of vapour

it seems there is some difference,

a possibility could be that you are calculating volume fraction as

Vfract  = volume vapor / (volume vapor + volume liquid)

but also for this case with values for liquid volume about 0.0011 m3/kg and vapor volume about 0.63 m3/kg

the values seem different...

Edited by serra, 19 March 2015 - 06:45 AM.

[Padmakar Katre]

Hi,

Can you confirm if there is any change in MEG to water ratio and thereby resulting different loadings of these traces in the heavy phase stream from separator? How confident are you about predictions of these loadings with selected thermo model in simulator?

 

You can assess these things first before you make final decision. Good luck