Real-time Drilling Event Detection Based on Mud-logging data Using Long Short Term Memory Neural Networks

Badinloo, Mehran; Farashiani, Jamal; Khoozan, Davood; Mansoori, Mehdi; Daneshkhah, Vahid

Real-time Drilling Event Detection Based on Mud-logging data Using Long Short Term Memory Neural Networks

Document Type : Original Research

Authors

M. Badinloo ¹

J. Farashiani ¹

D. Khoozan ¹

M. Mansoori ²

V. Daneshkhah ²

¹ Faculty of Chemical Engineering, Tarbiat Modares University, Tehran, Iran

² Dana Energy Company, Tehran, Iran

Abstract

Research Subject:Drilling operations frequently encounter numerous challenges that can lead to significant financial, human, and environmental losses. Therefore, predicting potential problems before they occur and implementing necessary preventive measures is crucial to minimizing risks. In this context, this study investigates the impact of employing artificial intelligence (AI) algorithms to forecast drilling complications using real-time mud logging data collected from existing wells in an Iranian oilfield.

Research approach: A hybrid architecture combining Long Short-Term Memory (LSTM) and Fully Connected neural networks was developed for the identification and detection of anomalies such as kicks and stuck pipe. Given the scarcity of these anomalies in the dataset, which could adversely affect model accuracy and performance, the Synthetic Minority Oversampling Technique (SMOTE) was applied to balance class distribution and enhance the overall effectiveness of the network. Furthermore, the influence of varying hyperparameters on reducing network error was systematically analyzed.

Main Results: Various network architectures and structures were examined. The experimental results indicated that the optimal model achieved an accuracy of 94.45% on the testing dataset with the following hyperparameters: a lookback of 7, a learning rate of 0.001, a dropout rate of 0.2, a batch size of 32, and a four-layer network architecture with 512, 256, and 256 units in the first, second, and third hidden layers, respectively. This configuration yielded higher accuracy and fewer false alarms in anomaly detection compared to other tested models. Based on the obtained results, this approach demonstrates significant potential for real-time anomaly detection in drilling operations.

Keywords

Mud logging

Long-short term memory network

Kick

Stuck pipe

SMOTE

Subjects

drilling

Hosseini, P. (2017). Estimation of lost circulation amount occurs during under balanced drilling using drilling data and neural network. Egyptian Journal of Petroleum, 26(3), 627-634.
Abbas, A. K., Al-haideri, N. A., & Bashikh, A. A. (2019). Implementing artificial neural networks and support vector machines to predict lost circulation. Egyptian Journal of Petroleum, 28(4), 339-347.
Hou, X., Yang, J., Yin, Q., Liu, H., Chen, H., Zheng, J., ... & Liu, X. (2020, May). Lost circulation prediction in south China sea using machine learning and big data technology. In Offshore Technology Conference (p. D041S053R005). OTC.
Hou, X., Yang, J., Yin, Q., Chen, L., Cao, B., Xu, J., ... & Zhao, X. (2019, October). Automatic gas influxes detection in offshore drilling based on machine learning technology. In SPE Gas & Oil Technology Showcase and Conference (p. D021S010R003). SPE.
Muojeke, S., Venkatesan, R., & Khan, F. (2020). Supervised data-driven approach to early kick detection during drilling operation. Journal of Petroleum Science and Engineering, 192, 107324.
Nhat, D. M., Venkatesan, R., & Khan, F. (2020). Data-driven Bayesian network model for early kick detection in industrial drilling process. Process Safety and Environmental Protection, 138, 130-138.
Magana-Mora, A., Gharbi, S., Alshaikh, A., & Al-Yami, A. (2019, March). AccuPipePred: A framework for the accurate and early detection of stuck pipe for real-time drilling operations. In SPE Middle East Oil and Gas Show and Conference (p. D032S077R003). SPE.
Alshaikh, A., Magana-Mora, A., Gharbi, S. A., & Al-Yami, A. (2019, March). Machine learning for detecting stuck pipe incidents: Data analytics and models evaluation. In International petroleum technology conference (p. D021S021R003). IPTC.
Li, Y., Cao, W., Hu, W., & Wu, M. (2020). Diagnosis of downhole incidents for geological drilling processes using multi-time scale feature extraction and probabilistic neural networks. Process Safety and Environmental Protection, 137, 106-115.
Zhang, Z., Lai, X., Wu, M., Chen, L., Lu, C., & Du, S. (2021). Fault diagnosis based on feature clustering of time series data for loss and kick of drilling process. Journal of Process Control, 102, 24-33.
Hairani, H., Anggrawan, A., & Priyanto, D. (2023). Improvement performance of the random forest method on unbalanced diabetes data classification using Smote-Tomek Link. JOIV: international journal on informatics visualization, 7(1), 258-264.
Passos, D., & Mishra, P. (2022). A tutorial on automatic hyperparameter tuning of deep spectral modelling for regression and classification tasks. Chemometrics and Intelligent Laboratory Systems, 223, 104520.
Wang, J., & Ozbayoglu, E. M. (2022, June). Application of Recurrent Neural Network Long Short-Term Memory Model on Early Kick Detection. In International Conference on Offshore Mechanics and Arctic Engineering (Vol. 85956, p. V010T11A008). American Society of Mechanical Engineers.