Post-combustion carbon capture systems face significant challenges in achieving optimal performance across varying operating conditions, particularly for low and changing CO2 concentration streams. This work presents utilizing a process model to identify key perturbations and study the corresponding effects for post-combustion CO2 capture systems, building upon established steady-state models. The sensitivity analysis results, coupled with machine learning techniques, are used to develop a surrogate model to enable fast performance prediction and optimization under varying conditions. Sensitivity analyses of key parameters—including L/G ratio, temperature and flow operational conditions, loading, and partial pressure—reveal complex interactions affecting energy efficiency. The framework would enable operators and process engineers to visualize these interactions, identify optimal operating windows, and predict system responses without extensive computational demand. Unlike traditional steady-state approaches based on first principle models, this methodology can approximate process responses and complex parameter interactions in a significantly shorter time frame that makes optimization possible for varying CO2 concentration streams. This research advances carbon capture technology by bridging the gap between theoretical process models and practical implementation, providing a valuable tool for both process design and operational optimization, with potential advantages for reducing the overall cost of carbon capture in industrial applications.
