Representative elementary volume for fractured rock masses from geometric, mechanical, and hydraulic perspectives: A review

We present a state-of-the-art review of the representative elementary volume (REV) for fractured rock masses from geometric, mechanical, and hydraulic perspectives. Various fracture modeling approaches, including discrete fracture network (DFN), discrete fracture and matrix (DFM), and pipe network models, are compared in terms of their strengths and limitations. Methods for determining equivalent properties of fractured rock masses, including laboratory experiments, analytical homogenization, numerical simulations, and machine learning (ML) techniques, are systematically reviewed. Key findings reveal that the REV is inherently scale-dependent, anisotropic, and governed by fracture network topology, in situ stress conditions, and flow regimes. Notably, geometric REVs generally serve as a lower bound, while mechanical and hydraulic REVs are typically larger and sensitive to stress and flow conditions. Ongoing challenges include addressing multi-scale heterogeneities, managing computational costs, and effectively integrating field data across disciplines and scales. Future research should focus on integrating multi-source data and developing physics-informed or ML-enhanced predictive frameworks to improve REV estimation and enable reliable field-scale simulations in rock engineering.