System Identification and Automatic Control of Distillation Columns

Abstract

This report will discuss system identification and automatic control for a distillation column. This system is nonlinear and multivariable, and the need to control it is crucial for chemical industry. A type of distillation column known as Column A will be used, which was studied by Skogestad at [1]. It separates a binary mixture and has 40 theoretical stages. It consists of one feed stream and two product streams which will be controlled in this report. The model of the system will be identified using MATLAB files, found in [2], which will be linearized and reduced using MATLAB in order to begin design of the controllers. The LV configuration will be used to control the two product compositions 𝑦𝐷 and 𝑥𝐵, using a classical PID controller tuned once by the Ziegler-Nichols closed-loop tuning method and once by the Internal Model Control (IMC) tuning method. In addition, de-couplers will be designed to reduce interaction effects among the system variables. Moreover, the need of the feedforward control to reject the disturbances will be discussed. Finally, evaluation and analysis of the two controllers will be carried out to test which of them is better to control the system.