tunning PID 1

[zonnegod]

I am dealing with a pid controller which has to regulate a flow as follows:
Set Point: 60 mc/h
Manipulated variable is a flow control valve.

Now, in practice I have noticed that the reaction time is very slow. Considering I can make a tunning and I can change, the "gain", the "integration time", "derivation time", please tell me if you know, how can I set these parameters of the PID. Should I increase the gain? Now the gain is around 30.

Thank you

 

[pmover]

without actual data (input and feedback or response), a worthy response cannot be given.

attached is a document that may provide guidance.

if this does not help, i've another PID tuning document that I've used many times for tuning PI control loops. it worked nearly every time I used it.

post a request and i'll search for it and attach.

good luck!
-pmover

 

[zonnegod]

Thanks, I will have a look on it. keep in touch!

 

[LeoChem]

You can carry out your simulations with SCILAB. There's a apllication inside called XCOS that works like SIMULINK (from MATLAB) and make it easy to simulate the response for any kind of system.

You can download it for free at:

http://www.scilab.in/

I actually used to simulate the response of a second order system to a degree function with different values of "gain", "integration constant" and "derivation constant"

 

[MortenA]

Control theory is actually a subject in itself and nobonny can just "..please tell me if you know, how can I set these parameters of the PID"

I would recommend that you either find somebody who know - or let it be.

Sometimes "slow response" is needed in order not to get a unstable system. It may work fine for a while untill something else changes...

I dont think there will be much use in looking at simulation tools unless you know and understand the theory.

Best regrads

Morten

 

[LeoChem]

Zonnegod

Is important to understand that the gain of the PID is not always the key to adequate the response of your system. Can make it unstable

The best thing is knowing your system behavior and equationate it, but if it is not possible because of the complexity you can infer the behavior by changing control parameters in a simulator with a simpler equation.

The derivation time can actuate fastly over the system's reaction time, but is dangerous because of its sensitivity to fluctuations

The use of integration time is to eliminate the offset of the final reponse, but its not so useful if you are trying to lower your reaction time.

 

[danw2]

To make the controller's response more aggressive

- for P, increase the gain value or decrease the proportional band value, depending on which your device uses [probably gain according to your post]

- for I, decrease the value if units are minutes (or time units) [increase the value if the units are repeats/minute, but your post indicates integral time]

- for D, increase the value

It might prudent to make a record of the values you started with and increase only one at a time.

 

[MortenA]

Ziegler-Nichols must be turning in his grave....

Best regards (and luck)

Morten

 

[LeoChem]

Morten

You are no dealing with children...

Your last comment was totally unnecessary. If you don't want to help we don't need your help

 

[JackWhite]

And one more thing, Mr. Morten,...Ziegler and Nichols are two different people...so "he" isn't turning anywhere. Just wanted to support danw2s' method, but i would keep 2 variables constant and vary just one, dont't tune 3 at a time or you won't understant wich influences the system how.

Good luck

 

[zonnegod]

JackWhite and Danw2, I did it in this way. I increased the gain, but not too much because if I did it too much the controller got a little crazy...

thanks

 

[FrancisL]

Basically you can just twiddle all the parameters, hastily returning to manual when it all goes wrong.
Or you can learn some basic control theory, just a little can help.

Francis

http://www.controldraw.co.uk

http://www.s88control.blogspot.com

 

[MortenA]

Leochem, been offline for a while: I dont know him, and you dont. You should be carefull giving specific advice when you dont know the full set-up as you dont (and neither do I).

Francist, cant really determine if you are ironic?

Best regards

Morten

 

[FrancisL]

Morten. My point was serious, but expressed with a drop of irony .
If you don't know what integral and derivative do it is unlikely that you will be able to tune the loop reasonably.

Francis

http://www.controldraw.co.uk

http://www.s88control.blogspot.com