Control of a homogeneous stirred reactor: Trajectory Piecewise-Linear model for NMPC

As a consequence of increasing computational power and improved algorithms, Model Predictive Control (MPC) nowadays is becoming applicable to high-dynamic processes. If using nonlinear MPC strategies (NMPC), e.g. Sequential Quadratic Programming-solvers (SQP), it is reasonable, if not mandatory to reduce the prediction model in order to cope with real-time. In this work the Trajectory Piecewise-Linear approach (TPWL) as a Model Order Reduction (MOR) scheme is investigated in simulation on its applicability on controlling combustion instabilities in an open homogeneous reactor using a SQP strategy. The corresponding nonlinear model exhibits, like the experiment, temperature oscillations during combustion. The temperature oscillations are due to chemical kinetics and typical for homogeneous low-temperature combustion. Since chemical kinetics are highly nonlinear, NMPC is a promising approach to damp these oscillations and to control the system in different operating points.