Search

Flexible thermal power plants integrated with CO 2 capture systems can balance the intermittent power generation of renewable energy sources with low-carbon electricity. Among these power systems, natural gas combined cycles will play a fundamental role because of their faster operation and higher efficiency. Optimisation-based control strategies can enhance the flexible power dispatch of these systems and improve their performance during transient operation. This work proposes a model predictive control (MPC) strategy to stabilise these power plants with post-combustion CO 2 capture based on temperature swing chemical absorption and provide offset-free reference tracking. A delta-input formulation with disturbance modelling is proposed, as it provides more efficient computation with offset-free control. Data-based models were developed to replicate the performance of the actual power and capture plants. Prediction of nonlinear behaviour was accomplished by creating a network of local linear models, which allowed the formulation of the dynamic optimisation program in the MPC strategy as a convex quadratic programming problem. A case study demonstrated the effectiveness of the proposed MPC to balance drastic changes on power demand and keep specified capture ratios. Furthermore, the reduced deviations achieved in the reboiler temperature suggest that the nominal value of this parameter could be increased to improve the desorption process without risks of reaching temperatures where the solvent would degradate. [ABSTRACT FROM AUTHOR]

A dynamic model of the post-combustion CO 2 capture process based on chemical absorption is used to investigate the transient behavior and dynamic responses of the process and to detect stabilization time when various disturbances are introduced. Plant dimensions and parameter settings are based on the SINTEF CO 2 capture pilot plant at Tiller in Norway, and the overall process model is validated using two sets of steady state pilot plant data. A deviation between model and pilot plant results of −0.8% and −4.5% in absorbed CO 2 and 2.6% and 1.2% in desorbed CO 2 is seen for the two cases used in validation, respectively, which is within the observed pilot plant CO 2 mass balance error of ±6%. The simulated absorber and desorber temperature profiles show also adequate agreement to the pilot plant measurements. The process model is further used to simulate set-point changes in flue gas flow rate, reboiler duty and solvent flow rate in order to investigate typical stabilization times at various locations in the process. As expected, mixing models such as the absorber sump and reboiler will introduce time constants that affect the dynamic response profiles, while plug flow models such as the cross heat exchanger and lean cooler causes pure transport delays and no additional settling time. Mass transfer and chemical reaction rates causes some process inertia, but it is relatively small compared to the inertia of larger mixing vessels such as the absorber sump, reboiler and buffer tank and transport delay caused by plug flow. Changes to the solvent flow rate are also seen as a larger disturbance to the process compared to changes in flue gas flow rate and reboiler duty, reflected by longer process stabilization time to reach new steady state conditions. The estimated 90% settling times for the response in CO 2 capture rate in the Tiller pilot plant are less than 1 h, 3.5–6 h and 3.5–4 h for step changes in flue gas flow rate, solvent flow rate and reboiler duty, respectively. [ABSTRACT FROM AUTHOR]