@article{Weng2026, 
author = {Cheng-Kai Weng and Hong-Wei Yang and Jun Li and Xian-Jun Chen and Gong-Hui Liu and Shu-Sheng Guo and Zhen-Yu Long and Wang Chen},
title = {Integrating rock mechanics and optimized mechanical specific energy for real-time pore pressure estimation in high-temperature and high-pressure drilling},
year = {2026},
journal = {Petroleum Science},
volume = {23},
number = {8},
pages = {4829-4841},
keywords = {Pore pressure, High-temperature and high-pressure well, Mechanical specific energy, Machine learning, Safe drilling},
url = {https://www.sciopen.com/article/10.1016/j.petsci.2026.03.062},
doi = {10.1016/j.petsci.2026.03.062},
abstract = {Accurate real-time estimation of pore pressure (Pp) is essential in high-temperature and high-pressure (HTHP) wells to prevent blowouts and lost circulation, given the complexity of their pressure regimes. However, conventional methods are inadequate: seismic-based and logging-based models are hindered by geological uncertainties, the dc-index principle is incompatible with PDC bits, and reliable while-drilling acoustic measurements remain prohibitively expensive. To overcome existing limitations, a surface-based and real-time Pp estimation framework is proposed, in which a direct Pp equation is derived by integrating an approximation using friction-corrected mechanical specific energy as the confined compressive strength (CCS) into the Mohr-Coulomb failure criterion. To ensure high-fidelity inputs for this equation, ridge regression is employed to invert rock strength parameters from drilling data, while a transient thermo-hydraulic model accurately calculates dynamic downhole pressure instead of relying on the static assumption. Validation on five HTHP wells in the Ying-Qiong Basin demonstrates that after accounting for thermo-pressure coupling, the method reduces the mean absolute error (MAE) in Pp equivalent density by 0.085 g/cm3 compared to the hydrostatic assumption. Furthermore, the proposed method achieves an MAE of 4.12%, outperforming the dc-index method, which achieves an MAE of 5.78%. Notably, the new method is more stable, with its prediction error envelope remaining within ±5%, whereas the dc-indexʼs error extends to ±10%. Given its theoretical compatibility with modern PDC bits and its demonstrated high accuracy, this surface-based and real-time scheme has the potential to overcome the conventional limitations of Pp estimation from surface data, providing a robust safeguard for well control in HTHP environments.}
}