Hi all,
My group and I recently conducted a level control experiment, where after getting the open loop response curve, we retrieved the PI controller tuning parameters via the Cohen Coon tuning relation. We further optimized the PI controller through observing the controller performance, by changing either the proportional gain and integral time and implementing a fixed set point change in water level. Our group realised that when decreasing the proportional gain below a certain value, while keeping the integral time constant, overshoot occurred.
This is very contradictory to what is expected, as it should be that only when increasing the proportional gain, the controller response is faster hence causing overshoot? Does anyone have an explanation for why a decrease in proportional gain might result in a larger overshoot in this case?
I have attached the response curves to this post. The red curve represents the water level output curve, while the yellow represents the step change in the setpoint.
Thank you.
Best regards,
Sharon Lam