An application of FEA and machine learning for the prediction and optimisation of casing buckling and deformation responses in shale gas wells in an in-situ operation.

Bartlett, Mark; Oyeneyin, Babs; Kayvantash, Kambiz; Njuguna, James; Mohammed, Auwalu I.

Abstract

This paper proposes a novel way to study the casing structural integrity using two approaches of finite element analysis (FEA) and machine learning. The approach in this study is unique, as it captures the pertinent parameters influencing the casing buckling and the evaluation of the magnitude of each. In this work, the effect of combined loading using multiple parameters to establish the relationship and effect of each on stress, displacement and ultimately casing safety factor is revealed. The optimised result show remarkable improvement in reducing the total deformation, the von Mises and increasing the safety factor of the casing under combined loading condition. The optimised casing shows 89% reduction in total deformation and 87% reduction in von Mises in comparison to unoptimised simulation result. In addition, the safety factor of 3.3 is obtained against the initial predicted stress of 932.46 MPa with a corresponding safety factor of 0.8129. Real time parametric prediction and optimisation using Lunar and Quasar (ODYSSEE software package) enabled the examination of the casing structural responses based on the pertinent parameters. In effect, a very good agreement was found between 'KNN' and Lunar predictions on parameters influencing casing buckling phenomena and the corresponding Mises stress. Lunar optimisation provided the ideal parameter values for the attainment of pre-define von Mises stress as a function of other factors. This quick approach shows both accuracy and validation of the two independent procedures arriving at the same conclusion. We found that concurrent investigation of the casing buckling attributing factors and optimisation using FEA and ODYSSEE package is sufficient to maintain casing structural integrity during shale gas extraction process.

Citation

MOHAMMED, A.I., BARTLETT, M., OYENEYIN, B., KAYVANTASH, K. and NJUGUNA, J. 2021. An application of FEA and machine learning for the prediction and optimisation of casing buckling and deformation responses in shale gas wells in an in-situ operation. Journal of natural gas science and engineering [online], 95, article 104221. Available from: https://doi.org/10.1016/j.jngse.2021.104221

• Journal Article Type: Article
• Acceptance Date: Aug 23, 2021
• Online Publication Date: Aug 24, 2021
• Publication Date: Nov 30, 2021
• Deposit Date: Aug 24, 2021
• Publicly Available Date: Aug 25, 2022
• Journal: Journal of natural gas science and engineering
• Print ISSN: 1875-5100
• Electronic ISSN: 2212-3865
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 95
• Article Number: 104221
• DOI: https://doi.org/10.1016/j.jngse.2021.104221
• Keywords: Casing deformation; Shale gas well; Prediction; Optimisation; Machine learning; Artificial intelligence
• Public URL: https://rgu-repository.worktribe.com/output/1427479