Deep/ultra-deep oil and gas resources are abundant at vertical depths of more than 3,500 m, which is an important succeeding field for future oil and gas exploitation. However, a lack of understanding of the multi-scale mechanical behavior of deep reservoirs under in situ conditions, as well as an insufficiently accurate prediction of engineering sweet spots, restricts the effectiveness of hydraulic fracturing in deep shale gas exploitation. In this study, the application of cross-scale rock mechanics, digital rock core modeling, and machine learning in characterizing reservoir geomechanical properties and assessing engineering sweet spots was summarized. The challenges and future development directions of the above research elements were explored. To achieve efficient deep-resource exploitation, it’s essential to clarify the mechanical behavior of shales with different mineral compositions at micro- and macro-scales. Numerical models incorporating mineral spatial heterogeneity were developed to analyze the multifactorial synergistic mechanism influencing shale brittle failure. Finally, intelligent fracability prediction methods for deep shale were proposed to accurately identify engineering sweet spots. The research findings have identified the key research and development directions for deep-resources development from a rock mechanics perspective. Document Type: Perspective Cited as: Zhao, G., Guo, Y., Yang, C., Chang X., Zhang X. Evaluation of the cross-scale mechanical behavior and fracability of deep shales: How innovations benefit the exploitation of deep resources. Advances in Geo-Energy Research, 2025, 17(2): 91-94. https://doi.org/10.46690/ager.2025.08.01

Deep-learning-based (DL-based) seismic inversion has emerged as one of the state-of-the-art research areas in exploration geophysics with the development of artificial intelligence technology. Due to its good portability and high computational efficiency, this method has emerged as a data-driven approach for estimating subsurface properties. However, most of the current DL-based methods rely solely on seismic data, lacking the incorporation of prior information. In addition, these methods are usually performed trace-by-trace, resulting in insufficient horizontal constraints. These limitations make traditional methods less robust, particularly when dealing with high noise levels or limited data. To address these challenges, we propose a multi-input deep learning network for pre-stack inversion, which combines data-driven and model-driven approaches for optimization. The proposed method separately extracts features from the model and data, merging them to improve feature utilization. Moreover, by adopting a 2-D training unit, rather than a trace-by-trace approach, the method improves the horizontal continuity of the results. Tests on synthetic and real seismic data confirmed the robustness and improved stability of the proposed method, even under challenging conditions. This dual-driven approach significantly enhances the reliability of seismic inversion.

Fracability, a key indicator for reservoir stimulation assessment, is directly related to shale gas production and resource extraction efficiency. In this study, a comprehensive fracability evaluation model for the shale reservoirs of the Wufeng-Longmaxi Formation in the Sichuan Basin was developed by taking into account factors such as brittleness, geostress differences, and natural fractures. The fracability of the Wufeng-Longmaxi Formation shale was evaluated quantitatively from the log data, and analyzed in relation to the actual fracturing of the deep well. The results show that the established model can effectively predict the fracturing potential of the reservoir. Shale brittleness is not equivalent to fracability. The brittleness index alone is not sufficient to characterize the ability of deep shale reservoirs to produce complex fracture networks after fracturing, but it is still a key factor in fracability evaluation. The established model can effectively predict the fracturing potential of the reservoir, which can provide a scientific basis for the fracturing modification and sweet spot preference of deep shale reservoirs in the Sichuan Basin. The purpose of this study is to use the comprehensive evaluation model to more accurately predict the fracturing potential of the Wufeng-Longmaxi Formation shale reservoirs in the Sichuan Basin, providing theoretical support for fracture stimulation and sweet spot optimization.