Flash Drum Control Valve Sizing 1

[chemdesign]

I am analyzing a control valve on the liquid effluent from a two phase (liquid/vapor) flash drum. The liquid is a physical solvent (Selexol) and contains approx. 8% (wt.) CO2. The pressure at the valve’s inlet flange is 185 psig, which is basically the flash drum’s pressure. The pressure at the outlet of the valve is 15 psig. The liquid in the upstream drum is stored at its bubble point due to the dissolved CO2, but the solvent itself has a very low vapor pressure (<1 psia). The true critical pressure of the gas rich solvent is not known, but is the pseudocritical pressure is approx. 1350 psia.

Operating data from the plant indicates that the control valve is only running at 30% open. My calcs using Fisher’s liquid sizing equation of dP allowable = Km*(P1-rc*Pv) indicate the valve is choked (dP allowable of 24 psi vs. dP actual of 170 psi). Using the dP allowable as the pressure drop in the valve sizing equation shows that the valve should be running at 70% open. All indications are the trim in the valve has not been replaced since the valve was shipped from the factory.

Finally, the Question: Is there something special about sizing control valves for this application (physical solvent with a large amount of absorbed gases) that would make equation of dP allowable invalid? If so, please explain. The only other thing that I can think of is that the sever flashing service has caused plug erosion and changed the valve trim characteristic from linear to something resembling quick opening.

Would appreciate any feedback on this.

 

[Montemayor]

chemdesign:

First, I strongly recommend you go to

http://www.documentation.frco.com/groups/public/documents/book/cvh99.pdf

and download the free, 295-page Fisher Control Valve Handbook, reading thoroughly the section on Flashing control valves and their design.

Next, I would refer you to your statement: “The liquid in the upstream drum is stored at its bubble point due to the dissolved CO2, but the solvent itself has a very low vapor pressure (<1 psia). Without knowing anything else other than knowledge that the CO2 is physically absorbed in dimethylether of polyethylene glycol according to Henry's law and then regenerated using either or both heat or pressure reduction, it is my belief that your liquid is not at the bubble point. This may be an important point if you are using a computer program that prompts for this effect – and you are confirming it. Assuming you are a Chem E, you should recognize this as an important point because what it means is that the liquid solvent is essentially not flashing into vapor due to the pressure drop. Rather, the dissolved CO2 gas contained in the volume of liquid solvent is released (“flashed”) and expands rapidly, probably achieving choked flow – due to its high specific volume at the lower pressure of 15 psig.

So, what you have is a given volume flowrate of liquid solvent being level-controlled and releasing its dissolved gas across the valve plug due to Henry’s law. This is flashing, but I suspect it is essentially all due to the dissolved CO2 and receives no vapor contribution from the solvent. I don’t know the volume % of the CO2 in the resultant 2-phase flow, but I suspect it is very high. 8% by weight seems to be a lot of CO2. I suggest you follow the Fisher Control Valve guide lines and run your calculations on the above logic. You don’t have to use a computer program. In fact, just for curiosity, I’d run a calculation on the CO2 alone, assuming that the ultimate %volume of solvent is minimal – and see what results this gives and how that compares with your first calculation.

Depending on the service life of the existing control valve, you may be very correct in suspecting that the present plug condition is eroded and causing your valve stroke to “hunker down”. Your observation and analysis of this effect is very good.

I hope this helps you out.


Art Montemayor
Spring, TX

 

[chemdesign]

Mr. Montemayor:

Thank you for your feedback, your line of thought seems right on target.

A couple of comments. First, I guess I don’t understand your statement that the liquid in the upstream flash drum may not be at its bubble point. That flash drum is separating flashed CO2 from the Selexol, and any drop in that drum’s pressure increases the volume of flash gas. Therefore, I am assuming the liquid’s vapor pressure is equal to the drum pressure.

Second, I have read the Fisher Control Valve Handbook (some salesman was kind enough to give me a hardbound copy years ago) and I cannot find anything in the section on flashing valves that explains how to deal with sizing a low vapor pressure liquid solvent containing a significant quantity of dissolved gases. The sizing equations provided are basically the ones cited in my original post. I am thinking that they do not apply to flashing dissolved gases from low vapor pressure liquids.

When I calculate the required Cv for only the flashed solvent (free of dissolved gases) and the Cg for only the flash gas and then them add together, I get that the valve should be 30% open which matches the data from the field. Does this approach hold water? It seems overly simplistic, but matches the approach that used to be used (pre Diers)for sizing flashing relief valves.

FYI, the volume % of the flashed CO2 in the downstream 2 phase mixture is 97%.

 

[Montemayor]

chemdesign:

Thanks to your response and additional data, I now have a better picture of what is concerning you. First, my definition of the “Bubble Point” is the text book description that states: A liquid at its Bubble Point is a liquid that is just about to vaporize at the pressure and temperature of its system; it is a saturated liquid, by definition. You stated, in the original post, that the solvent has a very low vapor pressure (<1 psia) while being under a system pressure of 185 psig. This, by definition, makes it a super-cooled liquid – not a saturated liquid at its Bubble Point. Now, you state “I am assuming the liquid’s vapor pressure is equal to the drum pressure”. The two statements seem contradictory, but I’m going to assume that the condition in the flash drum is not that of a saturated solvent. A saturated solvent would be found in the reboiler of the solvent stripper. I hope I am correct in following your description.

I’m elated that you have a hard copy volume of the Fisher Control Valve Handbook. This makes things go a lot faster and easier in explaining the methodology employed. I’m presuming that you have observed the definition of “Cv” and understand that it represents a numerical value for a volumetric (not mass!) quantity of fluid flowing through the valve in a unit of time. Therefore, bear in mind that control valves are sensitive to and rated by the volume, rather than the mass, passing through them. This is the reason to be mindful and specific in determining what phase is going through the valve. In your application we find a very unique situation that is not normally found in most applications: the inlet fluid is essentially a super-cooled liquid saturated with dissolved gas and as it traverses the plug opening at the valve’s seat, it doesn’t vaporize but it releases a very large quantity of gas – mixed with its parent liquid solvent (essentially, a 2-phase flow). In this special case I have used the sizing criteria expressed and detailed in page 72 of the Fisher Control Valve Handbook titled “Sizing for Liquid-Gas Mixtures”. My copy of the Hand Book is the Second Edition, 3rd Printing. Yours may be a later version with a different page number – but it should be in there anyway.

I have not sized control valves according the Dier’s method of additive Cvs, and don’t see anything wrong with the method as of now. It might turn out that dealing with Fisher’s recommended 2-phase calculation will give similar results. I hope the above comments are of some help and you reach a successful result. As I suspected, you have verified that the volumetric % of the CO2 is very high. In other words, from a practical point of view the valve is seeing essentially a pure CO2 gas stream and the use of a Cg is called for if this is the basis of capacity design.


Art Montemayor
Spring, TX

 

[chemdesign]

Mr. Montemayor:

You are correct in your assumption that the upstream flash drum is not at the solvent's bubble point. If dissolved gases were not present, the liquid would definately be a sub-cooled. My statement that the liquid was at its bubble point was referring to the rich solvent (i.e. dissolved gas present). In that sense, any reduction in pressure or increase in temperature would cause the formation of vapor; but the vapor composition would be nearly 100% dissolved CO2.

I could not find the sizing method Liquid-Gas Mixtures in the current edition of Fisher's control valve handbook. One concern I would have on this method would how it deals with the fact that most of the vapor is not formed until the liuqid enters the valve's orifice. I did try sizing this valve assuming a two phase mixture at the inlet to the valve. This results in a greater Cv than if you use the liquid sizing equations and assume the liquid solvent has the same properties as the dissolved gas/solvent mixture.

I am going to send the process data to Fisher and ask them to help resize the valve. Thanks.

 

[jsummerfield]

Some trap and tricks may still exist. Many flash drum applications result in a small flow and a small control valve. If you are dealing with a small volumetric flow application and a large vessel you may want to operate the valve on gap control. If the level control operates the valve close to the seat the valve trip will wire-draw. You may find this in the Fisher book. A solution to this is to operate the valve on/off as a dump valve; or throttle only at higher rates and on/off below a certain output percentage.

If Fisher quotes a one-inch valve with reduced trim assure that it is hard such as alloy 6 (stellite). Also, look up differential gap control among the DCS algorithms - simililarly available in a pneumatic displacer control too.

John