Rock mechanics–driven interpretable machine learning framework for preliminary seal integrity screening in geological CO2 storage

Muhammad Hammad Rasool , A.K.M. Eahsanul Haque , Numair Ahmed Siddiqui , Hassan Soleimani , Maqsood Ahmad , Nisar Ahmed , Muhammad Mubashir , Venkateshwaran Baskaran , Md Nasirul Islam , Lawal Opeyemi Olajid

Green Energy and Resources ›› 2026, Vol. 4 ›› Issue (3) : 100203

PDF (6291KB)
Green Energy and Resources ›› 2026, Vol. 4 ›› Issue (3) :100203 DOI: 10.1016/j.gerr.2026.100203
Research Article
research-article
Rock mechanics–driven interpretable machine learning framework for preliminary seal integrity screening in geological CO2 storage
Author information +
History +
PDF (6291KB)

Abstract

Seal integrity assessment is essential for ensuring the long-term security of geological CO2 storage; however, early-stage evaluation is often challenged by limited and heterogeneous subsurface datasets. This study proposes an interpretable machine learning framework for preliminary seal integrity screening using limited rock mechanics data from the 31/5–7 Eos well-core of the Northern Lights CO2 storage project. An unsupervised learning workflow was first employed to reconstruct Safe, Transitional, and Risk integrity regimes through cluster validation and ambiguity-based boundary analysis. Multiple supervised learning algorithms, including Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, and XGBoost, were subsequently developed and evaluated under stratified train/test split, Leave-One-Out Cross Validation (LOOCV), and repeated stratified cross-validation strategies. Random Forest demonstrated the most stable predictive behaviour across validation schemes, achieving balanced accuracies exceeding 0.86. Explainability analyses using Random Forest feature importance and SHAP consistently identified porosity and failure related effective axial stress as dominant screening controls, suggesting coupled stress–structure interactions govern integrity behaviour within the studied dataset. Based on these observations, a Stress–Porosity Screening Score (SPSS) was developed as a simplified post hoc screening descriptor for rapid integrity evaluation. Statistical analyses demonstrated significant differentiation of SPSS across reconstructed integrity regimes (p < 0.001). Decision-tree interpretation further enabled extraction of preliminary data-emergent screening pathways and threshold behaviour. Conclusively, the proposed methodology provides an interpretable and data-adaptive screening approach intended to support early-stage caprock assessment and guide subsequent detailed geomechanical investigations.

Keywords

Caprock integrity / CO2 geological storage / Stress-controlled behaviour / Interpretable machine learning / Random forest / Decision tree

Cite this article

Download citation ▾
Muhammad Hammad Rasool, A.K.M. Eahsanul Haque, Numair Ahmed Siddiqui, Hassan Soleimani, Maqsood Ahmad, Nisar Ahmed, Muhammad Mubashir, Venkateshwaran Baskaran, Md Nasirul Islam, Lawal Opeyemi Olajid. Rock mechanics–driven interpretable machine learning framework for preliminary seal integrity screening in geological CO2 storage. Green Energy and Resources, 2026, 4 (3) : 100203 DOI:10.1016/j.gerr.2026.100203

登录浏览全文

4963

注册一个新账户 忘记密码

References

[1]

Ahmed, A., Kamy, S., Delshad, M., Hildebrand, L.M.A., 2025. A comprehensive review of CO2 subsurface storage: integrity, safety, and economic viability. Energy Geosci. 100441.

[2]

Aminaho, E.N., Hossain, M., Faisal, N.H., Sanaee, R., 2025. Application of machine learning in the determination of rock brittleness for CO2 geosequestration. Mach. Learn. Appl. 20, 100656.

[3]

Bhagat, M., Bakariya, B., 2022. Implementation of logistic regression on diabetic dataset using train-test-split, K-Fold and stratified K-Fold approach: M. Bhagat, B. Bakariya,. Natl. Acad. Sci. Lett. 45 (5), 401-404.

[4]

de Lima, A.V.S., do Nascimento Guimar~aes, L.J., de S Beserra, L.B., Horowitz, B., Rojas Tueros, J.A., 2026. Analysis of structural integrity of geological reservoirs for CO2 storage regarding to fault reactivation: a forecasting model based on machine learning techniques. Neural Comput. Appl. 38 (9), 344.

[5]

Griffiths, L., Thompson, N., Smith, H., Skurtveit, E., Grande, L., 2025. Rock mechanical testing of core from the Eos Northern Lights CCS confirmation well and implications for the Geomechanics of North Sea CO2 storage. Int. J. Greenh. Gas Control 144, 104371.

[6]

Hannon Jr., M.J., Esposito, R.A., 2015. Screening considerations for caprock properties in regards to commercial-scale carbon-sequestration operations. Int. J. Greenh. Gas Control 32, 213-223.

[7]

Hawez, H.K., Bakir, H., Jamal, K., Kakakhan, M., Hussein, K., Omar, M., 2025. Numerical investigation of CO2 injection effects on shale caprock integrity: a case study of opalinus clay. Gases 5 (3), 15.

[8]

Isah, A., Tariq, Z., Mustafa, A., Mahmoud, M., Okoroafor, E.R., 2025. A review of data-driven machine learning applications in reservoir petrophysics. Arabian J. Sci. Eng. 50 (24), 20343-20377.

[9]

Karpfinger, F., Dubost, F.-X., Samir, M., Blinov, V., Khatri, G., Finko, A., Thomas, D.P., 2025. Geomechanical evaluation for cap rock integrity. In: SPE Annual Caspian Technical Conference. SPE p. D031S019R003.

[10]

Khadka, N., 2023. How leave-one-out cross validation (LOOCV) improve’s model performance. Dataaspirant-A Data Sci.Port. Beginn. 5, 542-548.

[11]

Kim, H., Vilarrasa, V., Makhnenko, R.Y., 2025. Laboratory-scale assessment of CO2 sealing potential of heterogeneous caprock. Water Resour. Res. 61 (7) p. e2025WR040339.

[12]

Lee, S.-Y., Mohamed, F.R., Lee, K.-H., McPherson, B., Balch, R., Yoon, S., 2023. Probabilistic evaluation of geomechanical risks in CO2 storage: an exploration of caprock integrity metrics using a multilaminate model. Energies 16 (19), 6954.

[13]

Rafiei Renani, H., Cai, M., 2022. Forty-year review of the Hoek-Brown failure criterion for jointed rock masses. Rock Mech. Rock Eng. 55 (1), 439-461.

[14]

Rasool, M.H., Ahmad, M., Ayoub, M., 2023. Selecting geological formations for CO2 storage: a comparative rating system. Sustainability 15 (8), 6599.

[15]

Rasool, M.H., Jaffari, R., Ahmad, M., Siddiqui, N.A., Junejo, A.Z., Abbas, M.A., 2025. Novel deep learning framework for efficient pressure zone detection via analysis of pore pressure profiling. Arabian J. Sci. Eng. 50 (1), 237-248.

[16]

Ross, A., Alexander, D., 2026. A caprock integrity assessment to optimize CO2 storage using coupled geomechanics: a case study of the Angostura field. Deep Resourc. Eng. 100252.

[17]

Sadiq, M., Sharma, R., Bisht, B.S., Chaudhary, M., 2025. Ensuring long-term caprock integrity in carbon capture, utilisation and storage: a comprehensive review of geomechanical and monitoring approaches. Curr. Sci. 129 (2).

[18]

Skerbisch, L., et al., 2025. Caprock integrity: a critical factor for carbon capture and storage in the Vienna Basin. Int. J. Greenh. Gas Control 146, 104434.

[19]

Suhaidi, N.N.S., Kaylyn, F.F., Ismail, N., 2025. Assessment of caprock integrity utilising rock failure prediction methodology for carbon capture and storage (CCS): a review. Adv. Mater. Res. 1184, 37-47.

[20]

Sun, Z., Salazar-Tio, R., Wu, L., Bostrøm, B., Fager, A., Crouse, B., 2023. Geomechanical assessment of a large-scale CO2 storage and insights from uncertainty analysis. Geoenergy Sci. Eng. 224, 211596.

[21]

Tillero, E., 2024. Machine learning-based modelling for geologic CO2 storage in deep saline aquifers. Case study of bunter sandstone in Southern North Sea. Int. J. Greenh. Gas Control 133, 104077.

[22]

Venkateshwaran, B., Haque, A.E., Ahmed, N.S., Ramkumar, M., Olutoki, J.O., 2023. A hybrid approach to caprock potential assessment and uncertainty estimation: bagging SVM and random forest. In: 2nd EAGE Workshop on Quantifying Uncertainty in Depth Imaging, vol 2023. European Association of Geoscientists & Engineers, pp. 1-7 no. 1.

[23]

Widodo, S., Brawijaya, H., Samudi, S., 2022. Stratified K-fold cross validation optimization on machine learning for prediction. Sinkron: jurnal dan penelitian teknik informatika 6 (4), 2407-2414.

[24]

Yao, P., Yu, Z., Zhang, Y., Xu, T., 2023. Application of machine learning in carbon capture and storage: an in-depth insight from the perspective of geoscience. Fuel 333, 126296.

[25]

Younessi, A., Kårstad, E., Basu, P., Larsen, M., Burachok, O., 2024. Geomechanical analysis of caprock integrity and fault stability for Greensand CO2 storage project feasibility. In: ARMA US Rock Mechanics/Geomechanics Symposium. ARMA p. D032S038R001.

PDF (6291KB)

0

Accesses

0

Citation

Detail

Sections
Recommended

/

〈 〉