Predicting and avoiding hazardous occurrences of stuck pipe for the petroleum wells at offshore Vietnam using machine learning techniques

Abstract

Over many years, the petroleum industry has been facing the hazardous occurrences associated with drilling stuck pipe. It is actually one of the most challenging and costly problems at offshore drilling rigs, especially in the high-risk, high-cost environments as the ultra-deep or deep water. Nowadays, with the development of Artificial Intelligence these problems can be predicted and avoided by applying machine learning techniques. The purpose of study is to apply the two most powerful machine learning techniques including Artificial Neural Network (ANN) and Support Vector Machine (SVM) to predict and avoid stuck pipe occurrences for the petroleum wells at offshore Vietnam. Modeling ANN and SVM using Matlab programming language based on many actual drilling parameters to estimate the risk of stuck pipe. The output of the models of both ANN and SVM machine learning techniques is the probability of non-stuck or stuck pipe at different measured depth points. Predictive results from both ANN and SVM techniques are compared to real measurements with the accuracy of all models above 85%. These machine learning techniques can help predicting stuck pipe occurrences as well as avoiding them by considering the changes of some important drilling operational parameters.