Applications of ML in CCS

The use of machine learning (ML) in carbon capture and storage (CCS) is an evolving field, with many potential technologies in the pipeline for implementation. This paper considers the most prevalent ones, including ANN, DT, SVM(R), XGBoost, and clustering, with their unique toolsets, and how they can be implemented in hybrid models to ensure effective CCS. It considers how these technologies are trained, and how they can be implemented in various parts of the CCS process, including storage site selection, real-time monitoring and optimisation, leakage detection, predictive maintenance, and enhancing CO2 absorption materials. The investigation of the XGBoost algorithm in this study has confirmed ML's effectivity, whilst identifying areas of further improvement which can be worked upon to enhance the model's accuracy (root squared, or R2, score).