MODEL-BASED CONTROL DEVELOPMENT FOR BINARY PILOT PLANT DISTILLATION COLUMN

The increasingly popular Model Predictive Control (MPC) strategy has been used in many process units either to improve the performance, save utility costs, or create a robust process able to cater to multiple variables. This project focuses on the development of model-based control for a distillation column in the Process Control laboratory at Universiti Teknologi PETRONAS (UTP) separating an ethanol-water and IPA-acetone mixtures. Specifically, the controller inputs are the reflux flow and the reboiler steam flow, while the outputs are distillate and bottom compositions respectively. Previous works have attempted to determine the dynamics of said column, therefore the MPC to be developed in this project is based on two of the derived models, one is a 2 X 2 Wood and Berry model and the other an inferential model. A comparison between the developed MPC controllers with standard PID controller is done to demonstrate the effectiveness and reliability of the MPC controller.