Advanced Step Nonlinear Model Predictive Control for Two-stage Thermo Mechanical Pulping Processes

Abstract

AbstractIn this work, we use a general nonlinear model predictive control (NMPC) technique for the two-stage (primary and secondary refining) thermo-mechanical pulping (TMP) refining process. The NMPC strategy is based on empirical and first-principle models which describe dynamic behavior and interactions among process variables. The computational burden is one of main drawbacks of NMPC controllers when applied to real systems. In this work, we handle the computational burden of the resulting nonlinear programming (NLP) problem using the IPOPT (Interior Point OPTimizer) solver. This is further improved with the advanced step NMPC (asNMPC) controller concept, a sensitivity based approximation to the solution of the resulting NLP. The simulation study compares the performances of the ideal-NMPC and the asNMPC strategies. The asNMPC controller reduces the CPU time and performs well in the presence of disturbances.