Prediction of natural fracture network patterns using feature engineering and machine learning approaches

Abstract

In this paper, we present a study of machine learning algorithms for predicting patterns of natural fracture network. The dataset used originates from the Teapot Dome field, USA. Initially, fracture azimuths were categorized into eight classes, each representing a 45-degree segment. Various machine learning models were then employed, ranging from traditional boosting algorithms to more recent approaches to predict the fracture classes. The K-Nearest Neighbors (KNN) algorithm was used to produce the best initial results with an accuracy of approximately 70%. After applying data augmentation techniques, we improved the model performance, achieving an accuracy of 88%. In addition, with feature engineering, we achieve 98%. This work highlights the potential of machine learning models in predicting fracture paths, contributing to the broader application of ML in the geomechanical model.

Document Type: Original article

Cited as: Kurmanbek, B., Merembayev, T., Amanbek, Y. Prediction of natural fracture network patterns using feature engineering and machine learning approaches. Computational Energy Science, 2024, 1(4): 167-174. https://doi.org/10.46690/compes.2024.04.02

References

Amanbek, Y., Merembayev, T.,  Srinivasan, S.  Framework of fracture network modeling using conditioned data with sequential gaussian simulation. Arabian Journal of Geosciences, 2023, 16(3): 219-241. 

 Amanbek, Y., Singh, G., Wheeler, M. F.,  et al.  Adaptive numerical homogenization for upscaling single phase flow and transport. Journal of Computational Physics, 2019, 387: 117-133.

 Bishop, C. M.,   Nasrabadi, N. M.   Pattern recognition and machine learning. New York, USA, Springer-Verlag New York, 2006.

 Breiman, L. Random forests. Machine Learning, 2001, 45(1): 5-32.

Chandna, A.,  Srinivasan, S.   Mapping natural fracture networks using geomechanical inferences from machine learning approaches. Computational Geosciences, 2022, 26(3): 651-676.

Chandna, A.,  Srinivasan, S.  Modeling natural fracture networks and data assimilation using multipoint geostatistics and machine learning-based geomechanical inferences. In Developments in Structural Geology and Tectonics, 2023, 6: 57-82.

Chen, T., He, T., Benesty, M., et al.  Xgboost: Extreme gradient boosting. R Package Version 0.4-2, 2015, 1(4): 1-4.

Cooper, S. P., Goodwin, L. B., Lorenz, J. C.  Fracture and fault patterns associated with basement-cored anticlines: The example of teapot dome, wyoming. American Association of Petroleum Geologists Bulletin, 2006, 90(12): 1903-1920.

Cover, T., Hart, P. Nearest neighbor pattern classification. IEEE transactions on information theory, 1967, 13(1): 21-27.

Dramsch, J. S.  70 years of machine learning in geoscience in review. Advances in Geophysics, 2020, 61: 1-55.

Feng, R., Mosegaard, K., Mukerji, T., et al. Estimation of reservoir fracture properties from seismic data using markov chain monte carlo methods. Mathematical Geosciences, 2024, 56(6): 1161-1184.

Freites, A., Corbett, P., Rongier, G., et al. Automated classification of well test responses in naturally fractured reservoirs using unsupervised machine learning. Transport in Porous Media, 2023, 147(3): 747-779.

Ke, G., Meng, Q., Finley, T., et al.  Lightgbm: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 2017, 30: 3146-3154.

Li, K., Ren, B., Guan, T., et al. A hybrid cluster-borderline SMOTE method for imbalanced data of rock groutability classification. Bulletin of Engineering Geology and the Environment, 2022, 81: 1-15.

Merembayev, T., Amanbek, Y.  Natural fracture network model using machine learning approach.   International Conference on Computational Science and Its Applications, 2023, 14107: 384-397.

Nick, T. G.,   Campbell, K. M.  Logistic regression. Topics in Biostatistics, 2007, 404: 273-301.

 Prokhorenkova, L., Gusev, G., Vorobev, A., et al. CatBoost: Unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 2018, 31: 6639- 6649.

 Rokach, L., Maimon, O.&nbspDecision trees, in Data mining and knowledge discovery handbook, edited by  O. Maimon and L. Rokach,  Springer, Boston,&nbsppp. 165-192, 2005.

Schwartz, B. C.   Fracture pattern characterization of the tensleep formation, Teapot Dome, Wyoming. Morgantown, West Virginia, Eberly College of Arts and Sciences, 2006.

Srivastava, R. M., Frykman, P.,  Jensen, M.   Geostatistical simulation of fracture networks.   Geostatistics Banff 2004, 2004, 14: 295-304.

Valera, M., Guo, Z., Kelly, P., et al. Machine learning for graph-based representations of three-dimensional discrete fracture networks. Computational Geosciences, 2018, 22: 695-710.