PID tuning help

P

Thread Starter

pk

Hello,

We are controlling the negative pressure inside a vessel by controlling a valve in suction line. I am trying to adjust the PID loop and was having too much oscillation. Anyhow it has improved now. But finally I am getting around +/-14% towards positive/negative side from the set point And the time of one oscillation is around 18 to 20 sec. My gain is at .4 and integral is 5.2 sec.

Is it possible to improve it more? Please help me fine tune it.
 
G

George Younkin

pk,

If it will be of any help, I can email documents on the basics of PID to you. Contact me if interested.

Regards,

George Younkin, P.E.
gwyounkin at charter. net
 
Hi there,

Have you tried using the offline tuning method? If possible, try to do a step test and give me the data (PV, OP, SP trending). I will help you do the math.

Best Regards,
Morpheus
Process Control
Shell
 
Yes, of course it should be possible to get better control than this. However, please don't assume that the oscillation is due to bad tuning!

You can usually get rid of oscillation caused by the PID controller by using a smaller gain or a longer reset time. Yes, I know this is not scientifically or mathematically correct, but if you are "tuning by feel" then it will help get the loop stable.

If you cannot get rid of the oscillation you need to try 2 things.

1. Put the loop in Manual. If it still oscillates, the the problem is probably coming from somewhere outside your control loop. Maybe you have another badly tuned loop somewhere else in the plant.

2. Get your boots on and go and look at the valve. A sticky valve or valve with a worn actuator mechanism will cause cycling. The only way to resolve this is to fix the valve!

There are good loop tuning tools out there that will help you to solve this problem quickly & correctly if you know how to use them.

My 2 C

Rob

www[.]lymac.co.nz
 
H

Hans H. Eder

Since you are controlling a pressure, I-action must be applied carefully and a Ti of 5 sec is quite harsh. Look at the phase shift between the PV and the OP. Most likely the maxima and minima of the PV curve will be somewhere in the middle between the maxima and minima of the OP curve. If so, increase the integral time - in steps – substantially. After that, you should be able to increase your gain.

Hans H. Eder
[email protected]
www.act-control.com
 
R
Another possibility, your control valve may be too large for the application. Try cutting your gain in half and increase the reset time by 5 (temporary measure), if it still oscillates only more slowly I would suspect the valve either oversize or sticky as Kiwi Rob suggested.

Roy
 
thanks frinds for your valuable information.
i increased the integral factor from 5 to 7 sec.
now it is quite better. still there are some oscillations but within range. It is not actually a valve but a damper operated by auma positioner.
 
P

Peter Nachtwey

Controlling gas pressures or suction is very non-linear. So what are you really controlling? There are gas laws that can be used to give a clue like PV=nrT. We need to know what is really being removed from the gas volume. Until we know that we are only guessing. This is not like controlling a piece of machinery or a simple temperature controller.
 
H
Integral time still seems very low to me (of course, without having seen your response curves). Again: Look at the phase shift between the PV and the OP. If you want, you can send me the curves.

Hans H. Eder
hans.eder at act-control. com
www.act-control.com
 
yes , we are sucking air to get negative pressure.
I need one more information from you guys.

If my control loop is well tuned and everthing is very good. What will be the effect if
1) i increase or decrease the P.
2) i increase or decrease the I.
 
Top