A Multivariable Model Predictive Control Project in a Computer Aided Master’s Degree Course

Abstract

Abstract This paper presents a multivariable Model Predictive Control (MPC) simulation case study project performed individually or in groups of 2 students, which accounts for one of the assessment components (70%) of the computer aided “Advanced Process Control” course at the 4th year of Bologna Master’s degree of Chemical Engineering at the department of Chemical Engineering, Instituto Superior Tecnico/University of Lisbon. This simulation control project provides a realistic platform for applying various control strategies such as feedback control, ratio and/or cascade control, feedforward control, decoupling multivariable control, and centralized multivariable MPC. Considering that the initial current control configuration of the case study “real” naphta cracker process is a MIMO multi-loop PID feedback base case, as control engineer the students were able to improve the base case control performance and robustness by designing and implementing an alternative centralized multivariable MPC control strategy. The MPC controller was designed in MATLAB®/ Simulink with MPC Toolbox. The tuning of the implemented controllers was investigated, by examining the effect of the MPC tuning parameters such as the sampling period, the prediction horizon, the control horizon and the weighting matrices.