Model predictive control of post-combustion CO 2 capture system for coal-fired power plants

The technique of post-combustion CO 2 capture is one of the most direct and effective means of reducing CO 2 emissions from coal-fired power plants. However, owing to the complex behavior of CO 2 capture system, such as large inertia, strong coupling among multi-variables, and strict constraints, the conventional PID controllers cannot meet the flexible operation requirement. For this reason, this paper proposes the use of model predictive control to improve the control performance of the CO 2 capture system. Subspace identification is firstly utilized to identify a model from input-output data of the CO 2 capture system. The power plant flue gas is considered in the modeling as measured disturbance, so that its influence can be quickly removed by the predictive control system. The model predictive controller is then developed on the identified model to regulate the CO 2 capture level and re-boiler temperature. To alleviate the effect of unknown disturbances and modeling mismatches, and achieve an offset-free tracking of the capture system, integral action is added by using an augmented state-space model in the controller design. Simulation results on an amine solvent-based post-combustion CO 2 capture system demonstrates the effectiveness of the proposed controller.