Subspace Model Identification and Model Predictive Control Based Cost Analysis of a Semicontinuous Distillation Process

Abstract

Semicontinuous distillation is a process intensification technique for purification of multicomponent mixtures. The system is control-driven and thus the control structure and its tuning parameters have crucial importance in the operation and the economics of the process. In this study, for the first time, a model predictive control (MPC) formulation is implemented on a semicontinuous process to evaluate the associated closed-loop cost. A cascade configuration of MPC and PI controllers is designed in which the setpoints of the PI controllers are determined via a shrinking-horizon MPC. The objective is to reduce the operating cost of a cycle while simultaneously maintaining the required product qualities. A subspace identification method is adopted to identify a linear, state-space model to be used in the MPC. The first-principals model of the process is then simulated in gPROMS. Simulation results demonstrate that the MPC has reduced the operational cost of a semicontinuous process by about 11%.