Nonlinear Model Predictive Control for Power Generation Gas Turbines

Various industrial applications rely on Gas Turbines (GTs) for power generation. The control task of power generation GTs is a challenging task considering the GT nonlinear multi time-scale dynamics as well as the critical process requirements. This implicates huge experimental as well as engineering efforts. In particular, the dynamic performance of a GT being a highly nonlinear and complex system strongly depends on the applied control system. The operation of GTs is still controlled using classical control methods to this day, where classical PID loops along with nonlinear empirically produced mappings are used for the different control tasks of the gas turbine. An advanced control method is to be investigated in this work for the application at hand. Namely Nonlinear Model Predictive Control (NMPC), where a model of the process (Compressor, Combustion, Turbine, Generator) is used in an optimization problem in finding the optimal value of the manipulated variables achieving the control task. The NMPC technique is directly applied in simulations to achieve the different control tasks of an industrial scale gas turbine (Siemens SGT-800 50MW) with emphasis on the real time capability of the algorithm. More broadly, this demonstrates that the proposed controller can be successfully employed in real time for achieving the control goals of a gas turbine operation.