Application of piece-wise linear system identification to solvent-based post-combustion carbon capture

Solvent-based post-combustion carbon capture (PCC) is currently the most promising method to reduce CO2 emission. To achieve a plant-wide controller for flexible operation, it is necessary to develop a data-driven model to understand the dynamic characteristics of PCC plant. This paper aims to: (i) carry out system identification to develop a data-driven model and (ii) provide insights into the nonlinear dynamics among the key variables from the PCC process in a wide operating range. These key variables include: CO2 capture rate, reboiler temperature, condenser temperature and lean solvent temperature. Pilot-scale PCC process implemented in gCCS was used to generate simulation data for system identification and model comparison. Linear single-input-single-output (SISO) transfer function models were firstly developed at different capture rates. Open loop step tests on identified models were then introduced to report the dynamics of key variables in various operating conditions and to indicate the level of system nonlinearity graphically. The nonlinearity analysis was carried out to investigate the system nonlinearity distribution in a quantitative manner. Based on the nonlinearity analysis, a multi-input-multi-output (MIMO) piece-wise model was proposed to simulate the nonlinear characteristics of PCC plant. The piece-wise model shows a satisfactory agreement with gCCS simulation data. Results of this study successfully demonstrate the nonlinear behavior of the solvent-based PCC process, which can be applied in the design of flexible plant-wide controllers.