A multi-device data fusion method for 3D rock fracture reconstruction: Development, comparison, and implications for hydro-mechanical simulations

Fracture void geometry plays a pivotal role in governing the coupled mechanical and hydraulic behavior of rock masses. Inherent limitations from commonly used X-ray CT and three-dimensional scanning hinder the accurate extraction of fracture void geometries, potentially compromising the reliability of mechanical and hydraulic property predictions. In this study, we propose a novel multi-device data fusion approach, the Local Equivalent Volume Method (LEVM), which integrates the high-resolution internal volumetric data from CT with precise surface morphology characterization obtained from 3D scanning. Tensile fractures were generated in fine- and coarse-grained granite specimens and reconstructed using three different methods, which were then compared with the measured true fracture aperture from thin sections. An FFT-based contact model and Modified Local Cubic Law model were employed to simulate fracture normal contact and fluid flow behaviors. Results show that the fracture aperture data obtained using LEVM consistently exhibits the lowest Relative Error for mechanical aperture, generally below 10%, and the smallest Root Mean Square Error (RMSE) among all methods. Different reconstruction methods yield fractures that exhibit fundamentally different hydro-mechanical behaviors. In contrast, the fracture closure and seepage responses simulated using LEVM-reconstructed fractures show close agreement with the experimental observations. Moreover, an index based on the fracture contact area under different confining pressures is proposed to assess the accuracy of fracture reconstruction. This study shows that when a single device cannot reliably reconstruct 3D fracture geometries, multi-device data fusion method provides more accurate models and improves the reliability of subsequent rock joint simulations.