Storage efficiency prediction for feasibility assessment of underground CO2 storage: Novel machine learning approaches

Underground CO2 storage is a critical strategy for reducing atmospheric greenhouse gases and mitigating climate change by sequestering carbon emissions in various geological formations. Machine learning has emerged as a pivotal tool in enhancing the efficiency and accuracy of CO2 storage, particularly in the context of carbon capture and storage (CCS) projects. In this study, data from various CO2 storage sites (including salt caverns, saline aquifers, depleted oil and gas reservoirs, coal seams, and basalts) was used to predict storage efficiency. The data comprises a comprehensive set of parameters including technical, economic, safety, environmental, and other aspects that are effective in CCS projects. Powerful machine learning and novel hybrid methods were employed to predict storage efficiency. According to the results, the methods achieved high accuracy. The analysis revealed that the storage depth parameter has the least effect on the algorithms, while the CO2 storage capacity parameter has the greatest importance score on the results. The study's findings demonstrated the high accuracy of intelligent methods, including machine learning, in predicting the efficiency of CO2 storage. These results are used to assess the feasibility of CO2 storage in various types of sites, helping to mitigate risks and minimize costs.