Advanced Modeling and Control of a Solid Sorbent-Based CO2 Capture Process

Solid sorbent-based CO2 capture from flue gas provides a novel alternative to traditional solvent-based processes due to potentially lower energy consumption for regeneration. Optimal operation of such a gas-solid flow process requires an advanced control framework for efficient disturbance rejection and setpoint tracking. This work focuses on the development of computationally fast and accurate dynamic reduced models (D-RMs) and the application of these models to nonlinear model predictive control (NMPC) algorithms. Two different types of D-RMs are developed: the first type is a data-driven model where step-test data are utilized to generate the D-RM through the Decoupled A-B Net (DABNet) model, and the second type of D-RM is a temporal and spatial reduction of the original process model to improve its computational efficiency while still retaining its physics. Disturbance rejection and setpoint tracking characteristics of the NMPC and its computational performance are studied for both types of DRMs. In addition, performance of both NMPC formulations are compared with linear model predictive control (LMPC) formulations.