Data-Driven Model for the Drilling Accidents Prediction

Abstract

Majority of the accidents while drilling have a number of premonitory symptoms notable during continuous drilling support. Experts can usually recognize such symptoms, however, we are not aware of any system that can do this job automatically. We have developed a Machine learning algorithm which allows detecting anomalies using the drilling support data (drilling telemetry). The algorithm automatically extracts patterns of premonitory symptoms and then recognizes them during drilling. The machine learning model is based on Gradient Boosting decision trees. The model analyzes real time drilling parameters within a sliding 4-hour window. For each measurement, the model calculates the probability of an accident and warns about anomaly of particular type, if the probability exceeds the selected threshold. Our training sample comes from 20+ oilfields and consists of sections related to 80+ accidents of the following types: stuck pipe, mud loss, gas-oil-water show, washout of pipe string, failure of drilling tool, packing formation, that occurred while drilling, trip-in, trip-out, reaming. We have designed the prediction model to work during drilling new wells and to distinguish the normal drilling process from the faulty one. One can configure the anomaly threshold to balance amount of false alarms and the number of missed accidents. To evaluate quality of the model we measure such data science metrics as ROC AUC score and confusion matrices. While testing model can identify 24 accident from 30 with high confidence, whereas for the others there is still a room for improvement. Our findings suggest that including more accidents of underrepresented types will improve quality. Other data science metrics also support aptitude of the model. Finally, having data from multiple heterogeneous oilfields, we expect that the model will generalize well to new ones. This paper presents a good practice of development and implementation of a data-driven model for automatic supervision of continuous drilling. In particular, the model described in the paper will assist specialists with drilling accidents prediction, optimize their work with data and reduce the nonproductive time associated with the accidents by up to 20%.

Forecasting the abnormal events at well drilling with machine learning

“…Conducted tests suggest that the current model is not sensitive to the changes in parameter setting. The Bag-of-features model was also compared to the other state-of-the-art models such as the CNN model, the Breakdown model, described in paper [6], and the Random model. The developed model's quality metrics exceed the quality of other models and ensure good performance during real-time drilling operations.…”

“…For the training, we use SGD optimizer and cross-entropy as loss functions. In addition, to compare the current model with previous studies, we retrained and validated the Breakdown model, presented in paper [6], on the same samples used during the Bag-of-features quality estimation procedures. The breakdown model uses aggregated features, such as mean, slope, etc., over the interval and predicts the accident with the Gradient Boosting algorithm.…”

“…This approach was based on summary statistics (mean, standard deviation, slope, and ratios) of each interval as features. However, it does not show the high values of metrics during the accident forecasting [6]. The described approach can be used as the model for the state-of-the-art comparison for our solution.…”

Quantitative Evaluation of Pre-Drilling Safety by Combining Analytic Hierarchy Process with Alternating Condition Expectation

“…Drilling-related underlying accidents, such as blowouts, lost circulation, sticking, fish and wellbore instability, can be attributed to several causes, including unexpected geologic variances, improperly executed procedures and unadaptable equipment [5][6][7][8]. To evaluate drilling safety, two steps are usually employed.…”

Making the black-box brighter: Interpreting machine learning algorithm for forecasting drilling accidents

“…In papers (Gurina et al, 2022b;Antipova et al, 2019) authors consider the problem of minimizing non-productive time (NPT) during the well construction process using AI techniques. At each time moment, the proposed ML model returns the probability of whether the corresponding time interval contains anomaly behavior that can lead to the specific type of accident or not.…”