New Grain–Grain Contact Models Eliminate Stiffness Overestimation and Are Validated Using P- and S-Wave Measurements

Voxel-based simulations of elastic properties from micro-computed tomography images frequently predict rocks that are systematically too stiff. Limited image resolution and segmentation smear grain boundaries, enlarge grain–grain contacts, and cause neighboring grains to behave as though they were perfectly bonded. These effects create unrealistically strong load-bearing pathways and bias the calculated effective stiffness tensor.

Recent improvements in GeoDict’s GrainFind module enable individual grains and their contact regions to be identified using watershed-based segmentation. Building on this capability, we present a practical, contact-aware digital rock physics workflow with two complementary correction strategies. The area-dependent approach preserves the original voxel geometry while reducing the stiffness of contact voxels according to their relative contact area. The explicit approach removes sinter-neck-like voxel artifacts from contact regions, thereby reducing artificial mechanical connectivity and simultaneously improving image-derived porosity (Saxena et al., 2019).

Following contact correction, effective elastic properties are calculated in ElastoDict using a homogenization approach. The resulting bulk and shear moduli, together with bulk density, are used to determine compressional- and shear-wave velocities. The workflow is demonstrated on typical sandstones. For the investigated sandstones, the predicted Young’s modulus decreases from a strong overestimation towards a good agreement with laboratory values. Both correction methods produce v_p and v_s values consistent with experimental data. Although the corrected contact regions occupy only approximately 1–5% of the sample volume, they exert a strong influence on the macroscopic mechanical response.

The presentation will guide GeoDict users through grain identification, contact detection, correction-parameter selection, elastic simulation, and wave-velocity calculation. By representing grain contacts more realistically, the workflow provides more reliable inputs for seismic interpretation, reservoir characterization, and geomechanical analysis. It also illustrates how grain-analysis developments motivated partly by other porous materials, including battery electrodes, can be transferred to digital rock applications.