Drilling Speed Enhancement Method for Extended Reach Wells Based on Machine Learning and Bayesian Optimization

The complex wellbore trajectories and significant horizontal displacements of offshore extended reach wells lead to increased downhole friction, severely affecting drilling efficiency. By leveraging drilling and logging data, this study proposes a novel method for rate of penetration (ROP) enhancement in extended reach wells with ROP prediction and drilling parameter optimization based on machine learning. Firstly, the original field data were pre-processed by filtering and normalization, followed by correlation analysis, revealing that ROP has strong correlation with drilling parameters such as weight on bit (WOB) and rotary speed, as well as wellbore trajectory parameters such as inclination angle and horizontal displacement. Based on these findings, ROP prediction models were developed using BP neural networks, random forests, and support vector machines. The results show that the BP neural network model outperforms the others, providing relatively accurate ROP predictions for offshore extended reach wells. Finally, the Bayesian optimization algorithm was employed to optimize parameters such as WOB, rotary speed, and displacement for ROP enhancement. The optimization results show that the ROP increases by 18.86% on average after the optimization of drilling parameters. The research results reveal the influence of drilling parameters and wellbore trajectory parameters on the ROP of extended reach wells and provide a theoretical basis for increasing the ROP of extended reach wells.