Multi-model predictive control of post-combustion CO2 capture plants

mono-ethanol-amine (MEA) based post combustion CO2 capture is considered as one of the most promising technology to reduce the CO2 emissions from fossil-tired power plants. With increasing intermittent renewable energy into the electricity grid, the fossil-fired power plants need to operate in a more flexible way. However the CO2 capture plant is a nonlinear multivariable system with large time delay, advance control technics must he applied to the capture process to obtain a satisfactory control performance. To achieve this goal, the gCCS software based on gPROMS simulation platform is firstly used to model the CO2 capture process with MEA solvent. Then this paper uses subspace identification methods to identify the local linear models at different operating points using the input and output data collected from the simulation platform. With the identified model, the multi-model MPC control system is developed to control the capture rate and reboiler temperature. Compared with conventional MPC controller, the closed-loop simulation results show that multi-model MPC controller can track the set point of capture rate quickly and smoothly without any oscillation while keeping the reboiler temperature to its desired operating range, which demonstrates that the multi-model MPC has high control accuracy and better robustness. Thus the multi-model MPC can realize the flexible operation of large-scale CO2 capture rate change.