Abstract
Most industrial control systems consist of a significant number of control loops. Generally, large processes are divided into many interconnected subsystems affecting each other, thus creating multivariable systems known as multiple–input multiple–output (MIMO) process. Generalized Predictive Control (GPC) based control algorithm is most suitable for MIMO systems. In this study, multivariate nonlinear (NL) GPC and discrete–time PID control of calcium oxide catalyst–packed reactive distillation (RD) column used for biodiesel production from waste cooking oil were investigated. Temperatures of reaction and reboiler sections were controlled by using non–decoupled and decoupled MIMO NLGPC and discrete–time MIMO PID algorithms. Feed flow rate with constant molar ratio and reboiler heat duty parameters were selected as manipulating variables. All recommended control methods, except for non-decoupled MIMO NLGPC, have been found to perform satisfactorily reference tracking and disturbance rejection in RD column. Consequently, the best control results were obtained in the decoupled MIMO NLGPC.