Hi everyone,
I am a student working on a project that focus on improving the control strategy of an ethylene splitter (high purity tower).
I am trying to replace my two PI Controllers (One controls temperature at some tray at the bottom of the tower using reboiler duty and the other controls the composition at some high tray using reflux flow) with nonlinear MPC controller that use the two products purities to modify reflux flow and reboiler duty.
When I added my MPC controller to replace the two PI controllers, it seems like it is making moves that increases the reflux flow and decreases the reboiler duty or the other way around. I know with the two PI controllers setup both my reflux and reboiler duty were moving in the same direction (increase together and decrease together). Is there a problem with my setup or the issue is in the penalties used in my MPC? I know there is high correlation between my two inputs and outputs.
Thanks