Control Valve Sizing 5

[clementer]

One of the fluid properties required to size a control valve for liquid flashing services is the fluid critical pressure. For a pure fluid, there is only one critical pressure. For a mixture, however, there are two possible options: The true critical pressure and the pseudo-critical pressure. The true critical pressure may be defined as the pressure at which the dew point and the bubble point curves meet in a P-T diagram. The pseudo-critical pressure has not any physical meaning. It is a mole fraction weighted average of the critical pressure of the components in a mixture. The difference between the values of true and pseudo critical pressure may be significant in some cases, which affects the control sizing calculations.

My first question is: Which critical pressure (true or pseudo) should be used for sizing of control valves?

My next question is regarding the validity of the control valves sizing equations for multi-components liquids. According to ANSI/ISA 75.01.01-2002 "Flow Equations for Sizing Control Valves", the sizing equations are valid for pure fluids but not for multi-component mixtures (page 11). Almost all the control valve sizing equations found on vendor’s literature and software are based on the ISA equations. Strictly speaking, none of these equations should be used for sizing control valves for liquid mixtures. So, how should control valves for this kind of services be sized? What errors are introduced by using ISA sizing equations?

Thanks,

Clemente

 

[JLSeagull]

The sizing equations work. Valves come in a wide selection of sizes. Control valve sizing uses a fixed set of conditions as a design basis for selection. The actual inlet pressure changes. The specific gravity changes. Use a sensitivity calculation to see what the critical pressure does to the size. Half the number that you have. Double the number that you have. Note the difference in the calculated coefficient. The critical pressure pertains to the potential cavitation. Mixed fluids are more gentle when cavitating a bit than water or other homogeneous fluids. You can approach valve sizing like a physicist and advise that the math is fuzzy. Or you can use the formulas as a set of tool to select the valves.

 

[JimCasey]

Seems like all I do lately is to follow JLSeagull around to say how right he is.

Critical properties have a lot to do with cavitation prediction, and some effect on noise rediction/choked flow. But as JLS pointed out it's not a life/death thing.

About nine times out of two I will get valves to calculate and the thermodynamic properties are not given or probably even available. I feel lucky when they even name the fluid accurately. Last week they gave me "phosporic acid". OK, what concentration? The specified SG was not within the range of H3PO4 between 0-100% concentration, so maybe something else was in it. What can you believe? In cases like this, I use 999 for the Pcrit so at least I can get an estimate from the sizing program, and when I see 999 it serves as a flag if the valve comes up later.

Cavitation with water can be devastating. Usually water is pretty close to being chemically pure and it changes state at well-defined points, and releases a LOT of energy when it condenses/collapses to a liquid. Other fluids are not like that. Hydrocarbons are mixtures of anything with boiling points in a range. Cavitating hydrocarbons flash to something roughly equivalent to shaving cream, and the bubble collapse is not sharply defined nor does it release much energy. So cavitation ceases to be a problem except to guess at the effect of choked flow on the Cv. Even water/glycol mixture flashes progressively according to the lever law, so the bubble collapse releases the energy over a distance/time period instead of the sharp crackle of a pure water bubble collapse.

Like JLS said: poke different constants in and see how much it doesn't effect the outcome. Then ask yourself how accurate your flow data is anyway. If you don't have a 30% fudge factor, you're probably only the third engineer in history not to. If a valve sizing program calculates the true required Cv to within 5% it belongs in the Smithsonian. The full, vast, overwhelming, array of variables that actually affect calculated Cv are not even asked for in sizing programs. P1, P2, Pv, SG, flowrate, pipesize are usually requested. What about pipe surface roughness, distance to fittings or other pipe disruptions, Pump curve parameters, elevation changes, cumulative system restriction, local altitude, even local gravity variation?

How many time have you seen a valve data sheet like this:

Max NORM MIN
Temp 250 250 250
P1(psig) 100 100 100
P2(psig) 40 40 40
Fluid Proprietary
flowrate 600 gpm 200 0
Viscosity Waterlike
SG 1.2 1.2 1.2
Pv

That's more information than I see most of the time. If I am going even estimate the Cv required, I have to make stuff up (ahem, "estimate")AND I guaran-dam-tee you that if the pump is making 250 psi at the inlet of the valve at max flow the head will be noticeably greater at zero flow. The downstream pressure will also decrease with decreasing flow, too, because the backpressure caused by the system will have a primary component which is a function of flowrare squared.

But that's OK, I can fudge to a safe selection. It is still better than the (frequent) " What valve do I need for 100-pound steam?"

 

[moltenmetal]

Following or no, that was a great post and worthy of a star!