Density determination of CO2-Rich fluids within CCUS processes

Abstract

Since accurate density determination is a crucial parameter for carbon dioxide (CO2) fluid measurement and transportation pipeline designing, this study focuses on predicting the density of CO2-rich fluids in transportation pipelines for carbon capture, utilisation, and storage (CCUS) processes. In this regard, two supervised machine learning (ML) models, including the extra trees regressor (ETR) and Gaussian process regression (GPR), were developed using a diverse database of field-scale simulation samples. The models were compared based on their accuracy using statistical and graphical analysis. The GPR model performed better than the ETR model, achieving a smaller root mean square error (RMSE). The GPR model provides valuable insights for pipeline design, flow meter reliability, and uncertainty assessment in carbon storage. Experimental validation confirmed the robustness and practical applicability of the GPR model. This study demonstrates the potential of ML techniques to enhance the efficiency and reliability of CO2-rich fluid transportation in CCUS processes.