Multi‐factor coupling effects in hydraulic fracturing of laminated shale: Experimental insights and physics‐informed neural network‐driven optimization
Deep Underground Science and Engineering · 09.07.2026
DOI: 10.1002/dug2.70108 ↗Kurz gefasst
Fachartikel in der Zeitschrift Deep Underground Science and Engineering, online erschienen am 09.07.2026. Verfasst von Tao Wang und 6 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Beijing Nova Program.
Abstract (Originalsprache)
Against the backdrop of deep shale gas development, the unclear multi‐factor coupling mechanisms in laminated shale fracturing limit exploitation efficiency. To reveal the influencing mechanism of the comprehensive effects of differential stress, the bedding dip angle, and the injection rate on fracture network formation, this study conducted a series of hydraulic fracturing experiments using shale cores with a diameter of 25 mm and a length of 50 mm. Through macro‐mechanical response testing and computed tomography scanning‐based micro‐characterization, the overall damage characteristics and internal micro‐damage specificity of the rock mass after fracturing were systematically analyzed. Combined with an orthogonal test design, a physics‐informed neural network (PINN) surrogate model, and Sobol global sensitivity analysis, a multi‐factor coupling framework was established to quantify the factor importance and interaction effects on fracturing performance within the experimental domain, providing qualitative implications for field‐scale hydraulic fracturing in laminated shale. The results indicate that when the axial pressure is adjusted, the fracturing response of specimens with bedding dip angles of 30° shows a more significant improvement. After the initiation of fractures in laminated specimens, the fractures undergo angular deviation in the vertical direction due to the bedding effect, propagate along the bedding planes, and gradually extend toward the direction of the maximum principal stress. The trained PINN prediction model shows excellent robustness and generalization ability, and further identifies the differential stress, the bedding dip angle, and the injection rate as the key factors governing shale breakdown pressure, with their first‐order Sobol sensitivity indices quantified as 0.87, 0.74, and 0.48, respectively. Finally, a targeted parameter optimization strategy is proposed, providing important references for onsite hydraulic fracturing construction.
Förderung
- National Natural Science Foundation of China
- Beijing Nova Program
Zitieren
APA
Wang, T., Xu, G., Liu, L., Zhang, Y., Zhu, S., Chu, H., & Zhao, K. (2026). Multi‐factor coupling effects in hydraulic fracturing of laminated shale: Experimental insights and physics‐informed neural network‐driven optimization. Deep Underground Science and Engineering. https://doi.org/10.1002/dug2.70108
Vancouver
Wang T, Xu G, Liu L, Zhang Y, Zhu S, Chu H, et al. Multi‐factor coupling effects in hydraulic fracturing of laminated shale: Experimental insights and physics‐informed neural network‐driven optimization. Deep Underground Science and Engineering. 2026. doi:10.1002/dug2.70108
BibTeX
@article{Wang2026,
author = {Wang, Tao and Xu, Guoyao and Liu, Liyuan and Zhang, Yao and Zhu, Sitao and Chu, Hongyang and Zhao, Kaikai},
title = {Multi‐factor coupling effects in hydraulic fracturing of laminated shale: Experimental insights and physics‐informed neural network‐driven optimization},
journal = {Deep Underground Science and Engineering},
year = {2026},
doi = {10.1002/dug2.70108}
}
EUVIMED