A Case Study on the Classification of Lost Circulation Events During Drilling using Machine Learning Techniques on an Imbalanced Large Dataset

In this study, we present machine learning classification models that forecast and categorize lost circulation severity preemptively using a large class imbalanced drilling dataset with good prediction accuracy. We demonstrate reproducible core techniques involved in tackling a large drilling engineering challenge utilizing easily interpretable machine learning approaches. We utilized a dataset of over 65,000 records with class imbalance problem from Azadegan oilfield formations in Iran. Eleven of the dataset's seventeen parameters are chosen to be used in the classification of five lost circulation events. To generate classification models, we used six basic machine learning algorithms and four ensemble learning methods. Linear Discriminant Analysis (LDA), Logistic Regression (LR), Support Vector Machines (SVM), Classification and Regression Trees (CART), k-Nearest Neighbors (KNN), and Gaussian Naive Bayes (GNB) are the six fundamental techniques. We also used bagging and boosting ensemble learning techniques in the investigation of solutions for improved predicting performance. The performance of these algorithms is measured using four metrics: accuracy, precision, recall, and F1-score. The F1-score weighted to represent the imbalance in the number of instances in the categories of drilling fluid lost circulation events in the dataset is chosen as the preferred evaluation criterion. When the prediction performance of the six basic models was compared, the CART model was found to be the best in class for identifying drilling fluid circulation loss events with an average weighted F1-score of 0.9904 and standard deviation of 0.0015. Upon application of ensemble learning techniques, a Random Forest ensemble of decision trees showed the best predictive performance. It identified and classified lost circulation events with a perfect weighted F1-score of 1.0. Using Permutation Feature Importance (PFI), the measured depth was found to be the most influential factor in accurately recognizing lost circulation events while drilling. The novelty of this work is the reproducibility of the workflow for implementing easily interpretable machine learning classification techniques for predicting lost circulation events. By following the workflow in this study, drilling teams will be able to better plan and conduct corrective actions before drilling fluid losses occur. This capability decreases drilling costs and accelerates resource recovery.