Model predictive control of vinyl chloride monomer process by Aspen Plus Dynamics and MATLAB/Simulink co-simulation approach

Characteristics of a vinyl chloride monomer (VCM) process is complex and nonlinear due to interactions between units in a reaction-separation network, multiple process streams, and multiple control loops involved. A fluctuation of the thermal cracking unit could result in a difficulty in maintaining downstream units at the setpoints. In this work, an approach to develop a model predictive control (MPC) for the VCM process by deploying a co-simulation between MATLAB/Simulink and Aspen Plus Dynamics is presented. The co-simulation provides more capability and comfortability to design the MPC controller, and evaluate controllability. The VCM process consisting of thermal cracking, quench and distillation is modelled by Aspen Plus Dynamics. The MPC is developed by integrating the concept of plant-wide control and subsystem partitioning. To reduce the burden of mathematical modeling, the MATLAB system identification toolbox is used to develop a multivariable linear model for the MPC controller by reconciling dynamics data from the VCM plant model. Performances of the developed MPC is evaluated under the regulatory test of the EDC feed disturbance. By comparison with the multiple single-input-single-output proportional-integral controllers through the efficiency indexes-an integral squared error, an overshoot and a settling time. Simulation results supported that the MPC controller outperforms proportional-integral controllers.