Hi all,

I'm having problems with a system that I would normally control using a cascaded PID loop. Unfortunately, I do not have a lot of instrumentation to work with.

Here is the system, I have to maintain air concentration levels in a tank by constantly blowing air into it. The amount of air that is entering the tank is modulated by a MOV. The air supply is kept constant by another loop which also supplies air to other systems. The concentration of air in the tank is very slow to change.

Normally I would cascade this system into two loops. The primary loop would use the concentration as the PV and output a set point to the second loop. The second loop would modulate the MOV based on a flow reading from the outlet pipe hunting around the setpoint from the primary loop.


User____
| ___
| | |
[PID] |
_________| |
| __ |
| | | |
[PID] | [%]
| | T | T
[M] [F] T | T
-----(\)--=-->T________T

In my case I don't have any flow measurement from outlet pipe. The only thing I have is valve position, which I don't want to use as a Process Value.

Any Suggestions?

 

Did you say what the problem was?

Does the supply pressure of the air vary such that you get too high variation without the flow loop?

Does the downstream pressure vary considerably?

Does your valve lack a positioner such that the actual position for a given output is highly uncertain? High hysteresis? Highly non-linear?

If not, you would probably get little benefit from cascade anyhow. Many cascaded loops will perform adequately without an "inner" or slave loop. I look at a many of ours and wonder why we made them cascade loops - must have been in Luyben or something.

John Rezabek
ISP