Inherent Control Valve Characteristics 1

[cornejoi]

Can anyone provide me with a a web link, book refrence, or any experience in how to choose an inherent valve characteristic when designing a control loop?

I've posted this question in other forums only because I didn't know this forum exsisted until now.

Anyway, this is what I know:

Control valves have inherent and installed characteristics. The inherent characteristics can be one of three:

1) Linear
2) Equal Percentage
3) Quick Opening

I understand the curves that each one of these inherent characteristics produces, and some very general guidlines for when one should choose what. I know that in most cases an equal percent characteristic will be specified.

However, I haven't grasped the concept of why most valves need to be equal percent. I know that is has something to do with the installed characteristic being linearized, in most cases.

I've looked everywhere for a praticle approach for choosing valve characteristics and so far I've only found "guidelines" that valve vendors provide. Moreover, most valve vendors tell me to stick to =% since you can now characterize in the positioner. My question is then "Why have different valve cages if characterization can be done in the positioner?"

Anyway, what I'm looking for is a praticle method of determining which valve characteristic is best suited for a specific application and why. The Fisher "Control Valve Handbook" (aka: The Fisher Black Book) and the Fisher "Control Valve Sourcebook (aka: The Fisher Red Book) have some good generic information but nothing to the exetent that I'm looking for.

If anyone is a guru in this area or knows where I can find good, regarded, information on this topic I'd greatly appreciate it.

I know that there are many schools of thought on this topic, I'm not looking for a "be all end all answer" but some direction and refrences on how to understand a process so that I can apply what I've learned to better specify a control valve.

Thanks guys.

--Igor

 

[bmsg]

The way I understand it, the inherent flow characteristic of a control valve is no longer valid when it is put in a process, because the pressure drop across the control valve changes (in most applications) with the amount of flow through it - if the flow is more (i.e., the valve is more open), the frictional loss in the pipeline is more and only the balance pressure drop is available for the control valve.
As such, as the control valve opens more and more in a typical process, the pressure drop across it decreases progressively.

In this situation, the inherent characteristic of the control valve 'deteriorates'. That is to say, the curve for the equal percentage valve becomes flatter, tending to the curve of the linear valve - and the curve of the linear valve tends more towards that of the quick opening type valve.

Moreover, the 'deterioration' of the flow characteristiv of the control valve is dependent on the ratio of the pressure drop across the valve to the total pressure drop in the line having the valve.

When the pressure drop across the control valve is about 20 - 30% of the overall drop in the pipeline (including the valve), the valve characteristic becomes linear. If the ratio is less than 20%, the curve bends further upwards - in the direction of the curve of the quick opening type valve.

Most control valve manufacturers recommend about 30% of the overall pressure drop of the pipeline to be assigned to the control valve - then the installed characteristic is more or less linear.

The lesser the percentage pressure drop assigned to the control valve, the more its installed characteristic resembles that of a quick opening type valve (that is, an on-off type valve), and the lesser controllability the control valve will have.

I am not an expert on control valves - I have a working knowlegde of them. I think that it is generally safe to use an equal percentage type valve for most applications where the DP across the valve varies with the opening of the valve.

A typical application where I would not specify an equal percentage valve is for level control in a tank (say) - where the pressure acros the valve is more or less constant - no matter how much the valve is opened.

BMSG

 

[4carats]

Cornejoi,

I advise you to buy the following book:
Principles and Practice of Automatic Process Control, by Carios A. Smith and Armando B. Corripio. Important: get the 1st edition (not the 2nd edition). In the 1st edition, it shows derivation of "Installed Characteristics" from Inherent Characteristics of control valves. Vigorous proofs were given.

Sizing of control valves always start from process data. Instrument Engineers must consider the fluid energy handled by the control valve trim. [ISA book 'Practical Guides for Measurement and Controls, Control Valves, by Borden, ed., 1998, has given trim exit fluid Kinetic Energy criteria to users of control valves.]. If the trim cannot handle the fluid energy, vibrations can occur which will affect the linkage movements and the pilot valve of the positioner, resulting in poor control. Therefore priority of characterization is as follows:
1st --- control valve trim charaterization
2nd --- positioner characterization, cam or electronic
3re --- DCS characterization

Process Engineers decide the fraction of the dynamic pressure to be taken by the control valves.

Instrument Engineers decide the oversized factor and the rangeability of the control valves. [In this regard, both the Process Engineers and the Instrument Engineers are usually Chemical Engineers from academic point of view.]

The three parameters above (fraction of the dynamic pressure, oversized factor, and rangeability of the control valves) & the Inherent Characteristic define the shape of the "installed" characteristic curve. The "installed" characteristic should be chosen in such as way that an overall linear process [process + valve] is achieved.

For further information, just ask.

4carats

 

[jsummerfield]

Good post 4carats. I do not have the books referenced and may buy them myself. As with other books, the actuator-positioner linkage stuff my be somewhat obsolete with the smart positioners. Any of the second-order lag information would be valuable. However, I do not think that smart positioners have a bypass. Bypassing the positioner was a fix for a fast valve where stability was a problem.


John