PDA

View Full Version : Tuning PID Controller in InfoWorks ICM



nkleekamp
July 25, 2016, 03:02 PM
Hi,

I have recently discovered the use of PID controllers and have been using them to control sluice gates based on the water level in a downstream interceptor in a sanitary sewer system. I have a bit of trouble tuning the P,I, and D variables though. Although I can typically get my curve to smooth out after much effort and trial/error, I was wondering if anyone had any tips on tuning the variables in an efficient way. I can usually get the curve to oscillate centered on my target interceptor level with the P variable, but from there it goes out of control until I get the D variable tuned back in. If I change some parameters in my model or change up the rainfall event I find the need to re-tune the D variable, which is not very time friendly the way I am doing it now. I have attached an image of one of my tuned controllers as well as one that occurs after I have changed to a different rainfall event and have not tuned.

Any help is greatly appreciated!

Thanks,

Nick

Robert Dickinson
July 25, 2016, 05:46 PM
Hello Nick,
Can you show the constants you are using for the PID controller?
Regards,
Robert Dickinson

nkleekamp
July 26, 2016, 05:29 AM
Absolutely! For the tuned image in my original post I am using a time step measurement of 240s and my constants for P, I, and D are -0.5, 0.0, and 0.5, respectively.

For the un tuned image in my original post the time step measurement is 180s and my constants for P, I, and D are the same as the tuned image - -0.5, 0.0, 0.5, respectively.

nkleekamp
July 26, 2016, 06:55 AM
I did get my curves to smooth out. Though, I did not changed my time step measurement or constants. All I did this time was match the crown of the pipe flowing into the interceptor with the same elevation as the depth set for the sensor. I am still interested in your input though as I am sure that I will need to re-tune once again if I change something.

Thanks,

Nick

nkleekamp
July 26, 2016, 07:34 AM
I do need to make a correction on my first post from today. My D constant is 5 not 0.5...