Development of prediction models for storage efficiency factor to estimate volumetric CO₂ storage capacity in saline aquifer

An accurate estimation of the CO₂ storage capacity in saline aquifers is critical for the successful implementation of geological CO₂ sequestration projects. Although volumetric methods provide quick and accurate estimates for the storage capacity, the storage efficiency factor (SEF) relies on empirical correlations or computationally expensive numerical simulations. To overcome these limitations, this study proposed a novel regression-based prediction approach for directly predicting SEF in vertical and horizontal well scenarios using simulation-derived data. Four regression models were developed: multi-linear regression (MLR), exponential (Exp), and response surface methodology models (RSM-1 and RSM-2). Among them, the RSM-2 models, which incorporate linear, interaction, and quadratic terms, demonstrated the highest accuracy, achieving mean absolute percentage errors of 1.1% and 3.4% for vertical and horizontal wells, respectively. Compared to prior empirical approaches, the RSM-2 model provided more precise and consistent predictions, with a narrower SEF distribution range and reduced uncertainty, enhancing its suitability for site screening and storage planning. The models were applied to two reservoirs in the Pohang Basin, Southeast Korea, and yielded storage capacity estimates within 5% of full-physics simulation results, confirming their validity and reliability. These models provide a cost-effective and time-efficient alternative to conventional simulations and are particularly advantageous in early-stage feasibility studies and large-scale CO₂ storage assessments. Future work will expand these models to accommodate more complex reservoir conditions and enhance their generalizability across diverse geological settings.