heat-exchanger, "bulk" valve and "fine" valve - is it " 1

[ipupkin]

Hi, All,

I have a typical PID control application and I'm still not sure what's the right name for it.

Process - heat exchanger (oil cooled by water)
Controlled variable - temperature of oil after the heat exchgr.
Manipulated variable - "water in heat exchgr." inlet valve position

There are two valves acting in parallel, large one for bulk changes, another small one for fine tuning. the small one maintains the flow, and the large one moves so that the small one is usually at 50% open.

1) Is it still called "split-range control"?
2) Is there a typical way how to use one (or two) PI[D] controller in this case?

 

[FrancisL]

No this is not split range control as you describe it, although split range could be used in such a way that the flow is only through the small valve when a low flow is required, thus achieving a greater turn down than the big valve alone.
If so then you only need one PID controller, followed by a split range function that determines the signal to each valve. I suspect that is what you really need.
Are you sure that statement "the large one moves so that the small one is usually at 50% open" is correct?
Because if you try to do that you will most likely have tuning problems. If the control on the large valve acts slowly (which it would have to or the small valve would bounce around like crazy) then if there were a large process upset the large valve, which should respond quickly will be unable to.

Francis

 

[controlnovice]

Francis is corect in

a) that it is not split range
b) questioning the small valve at 50%. Although I think the goal is the same as what follows.

I think the answer to your question 'b' is:

There is a typical way to do this with two PIDs. I don't have my books in front of me, but if I remember correctly, the small valve is controlled by the temperature. The large valve is controlled by the position of the small valve.

The large valve will look at the position of the small valve. There has to be some range though, so that the small valve can move between, say, 20-80% without the large valve moving. Once the small valve hits 80%, then the large valve should start to open. When the small valve hits 20%, the large valve will start to close. So, in essence, you are keeping the small valve at around 50% (kind of - I guess you could say it's in the middle of 20-80% right? And with the large valve opening/closing as needed, the small valve might position itself at 50%). The reason you want a range, 20-80%, before the big valve moves is to reduce the oscillation of the big valve - let the small valve do as much of the work as posible.

Again, this is without my books but I also seem to remember that the large valve should only use the Proportional(gain) parameter of PID. Since the small valve is performing the 'fine tuning', it is not imperative that the large valve be exact (using integral parameter of PID). It only needs to see an error above 80% or below 20% of the small valve position and start to move.

This configuration is simple in a DCS or PLC - more tricky with single loop controllers.

 

[rmw]

I thought of this thread earlier today when reviewing valve specs with a client. In his control scheme, for spray water for Reheater sprays and Superheater sprays on a large supercritical boiler, the control valves are paralleled with the larger one having a flow rate of 180K lb/hr, and the smaller one having a flow rate of 20K lb/hr.

The large one has a characterizable cam for the actuator that converts the linear command signal into a "quick opening" characteristic, while paralleling that the smaller flow valve had a cam that characterized it as an "equal percentage" trim.

Both valves get the same command signal simultaneously.

I had never seen this particular scheme of parallel vs split flow, but have you checked you actuators or positioners for characterized cams??

rmw

 

[FrancisL]

The scheme suggested by controlnovice looks possible, but you would not use PID control on the big valve, rather some simpler algorithm to ramp the large valve open or closed at a rate slow enough for the PID on the small valve to respond.
I suspect that the loop gain would be different for different positions of the large valve, which could cause tuning problems.
rmw is correct that the valve characteristics need to be considered so as to get a reasonably linear overal response.

 

[ipupkin]

FrancisL, controlnovice, rmw,
thanks for replies!

Actually, I was a bit confused by

http://www.controlartsinc.com/split.htm

My big ball valve is not purposed to be a control valve, and has strongly nonlinear characteristics. I can [try], of course, to eliminate this in PLC software.

and thanks for the idea to use a simple ramp for the big valve.

 

[controlnovice]

FrancisL - Good idea on the ramping of the big valve!

 

[25362]

What about the advantages or disadvantages of a three-way valve ?

 

[jsummerfield]

I have not applied a control scheme as described. A few ways to accomplish the scheme might include a gap-acting control or a fuzzy logic algorithm for the "Big Valve" with PID for the "Small Valve". The objective of the big valve control would not be 50 % but to just get within the mid range of the small valve and then stop processing the output. This valve would be controlling based upon the output to the small valve as the measured variable. The scheme would be set to only change the control output once the small valve is outside limits, perhaps 20 and 80 percent.

You would have to think through what the big valve should do if you have to place the small valve on manual and still want the big valve to control the process.

John

 

[m777182]

You can easily operate your big valve by hand; if you process is under control, it will not oscilate very much because all other controllers will maintain approx. steady state(flows, temperatures,...) You will use the big valve for quick start up/shut down or some other extreme conditions and you don't need a complex software and experts to tune it.
m777182