Multistep model predictive control of ethyl acetate reactive distillation column

Abstract

An unconstrained model predictive control (MPC) strategy is presented for single input-single output (SISO) control of a reactive distillation column. The predictive model of the MPC is based on the autoregressive moving average (ARX) model structure, whose parameters are updated on-line using the process measurement information. The design and performance of the control strategy is evaluated by applying to a highly nonlinear ethyl acetate reactive distillation column with double feed column configuration involving an esterification reaction with azeotropism. The controller performance is studied with respect to the effect of tuning parameters involved in the ARX model parameter adaptation algorithm and also towards the influence of model prediction and control horizons. The simulation results illustrate the superior performance of the multistep model predictive control strategy over the conventional PI controller for both setpoint tracking as well as disturbance rejection.