Nonlinear Model Predictive Control for Combined Cycle Power Plants

Sammanfattning: This master thesis project serves to investigate the possibilities of Nonlinear Model Predictive Control (NMPC) using the example of enthalpy control of the BENSON HRSG (heat recovery steam generator) of a combined cycle power plant (CCPP). The general idea of NMPC is to solve an optimization problem, to nd the next control action, and this optimization problem is based on a model of the system. The models used in the controller implementation are Modelica-based, and the system is described by algebraic dierential equations (DAEs). The controller was implemented in the Python interface of JModelica.org (Modelica-based modeling tool, supporting the Modelica extension Optimica for optimization), together with an extended Kalman lter (EKF) for state estimation. The control algorithm was only evaluated for a setup where the controller model is very similar to the model representing the real process; both models are simplied representations of the real process.