Abstract
Advanced technologies and monitoring systems are currently receiving significant attention in the drilling industry due to their potential to greatly enhance operational efficiency. Real-time monitoring coupled with predictive analytics enables operators to promptly detect and anticipate deviations from expected parameters, thus preventing potential incidents and ensuring that operations remain within safe operating limits. The approach in this study involves the development of a predictive model to ascertain safe ranges of key operational parameters.
Various machine learning algorithms like Gaussian process, Random Forest, and linear regression are evaluated to identify the most effective solution for predicting surface operating parameters. Hole conditioning operation as a case study is presented, utilizing drilling control parameters such as flowrate, rotation speed, and bit depth to predict the corresponding realtime surface operating parameters, namely, pressure, torque, and hookload, aiming to validate the efficacy of the developed model in real-world drilling scenarios. Additionally, thorough assessments are conducted by simulating multiple random sets to gauge the performance of models, allowing for the identification of potential weaknesses and areas for improvement.
Through rigorous evaluation, it is determined that Gaussian Process outperforms others due to its flexibility and scalability. The training results indicate that more than 90% of data points exhibit errors of less than 5% in prediction corresponding to 150 psi of pressure, less than 0.5 klb-ft in torque, and less than 5 klbs in hookload prediction. The performance of the developed model on the testing set data demonstrates similarly high efficacy in predicting safe windows around operating parameters. This narrow constraint interval indicates a highly effective safe window prediction model capable of predicting with minimal uncertainty, leading to more reliable detection of parameter deviations and aiding in preventing incidents.
This methodology offers a robust framework for leveraging advanced predictive analytics to detect deviations in operating parameters, thereby avoiding potential further non-productive downtime. In addition, the developed methodology is not limited to specific operating parameters and can be applied to a wide range of drilling conditions. It presents a generalized approach that can serve as a protective driver for autonomous drilling operations, ensuring safe and efficient performance across various drilling environments.
Various machine learning algorithms like Gaussian process, Random Forest, and linear regression are evaluated to identify the most effective solution for predicting surface operating parameters. Hole conditioning operation as a case study is presented, utilizing drilling control parameters such as flowrate, rotation speed, and bit depth to predict the corresponding realtime surface operating parameters, namely, pressure, torque, and hookload, aiming to validate the efficacy of the developed model in real-world drilling scenarios. Additionally, thorough assessments are conducted by simulating multiple random sets to gauge the performance of models, allowing for the identification of potential weaknesses and areas for improvement.
Through rigorous evaluation, it is determined that Gaussian Process outperforms others due to its flexibility and scalability. The training results indicate that more than 90% of data points exhibit errors of less than 5% in prediction corresponding to 150 psi of pressure, less than 0.5 klb-ft in torque, and less than 5 klbs in hookload prediction. The performance of the developed model on the testing set data demonstrates similarly high efficacy in predicting safe windows around operating parameters. This narrow constraint interval indicates a highly effective safe window prediction model capable of predicting with minimal uncertainty, leading to more reliable detection of parameter deviations and aiding in preventing incidents.
This methodology offers a robust framework for leveraging advanced predictive analytics to detect deviations in operating parameters, thereby avoiding potential further non-productive downtime. In addition, the developed methodology is not limited to specific operating parameters and can be applied to a wide range of drilling conditions. It presents a generalized approach that can serve as a protective driver for autonomous drilling operations, ensuring safe and efficient performance across various drilling environments.
| Original language | English |
|---|---|
| Title of host publication | International Conference on Offshore Mechanics and Arctic Engineering Proceedings Series |
| Number of pages | 10 |
| Volume | 6 |
| ISBN (Electronic) | 978-0-7918-8895-7 |
| Publication status | Published - 21 Aug 2025 |