Capillary Suction Measurements in Granular Materials and Direct Numerical Simulations Using X-Ray Computed Tomography Microstructure

Abstract

Because of the recent advancements in hardware and reconstruction algorithms, multiphase flow modeling in porous media is experiencing a shift toward using advanced imaging techniques such as X-ray computed tomography in conjunction with direct numerical simulations. This approach captures heterogeneities in soil samples utilized in laboratory testing and results in quick and less tedious predictions compared to the existing methods. In this paper, an experimental setup is developed specifically to validate numerical predictions of the soil water retention curve (SWRC) for two types of sands with identical size but distinct grain morphology of round (Ottawa sand) and angular particle shape (Q-Rok). The complex 3D pore network is captured noninvasively using high-resolution attenuation-based X-ray computed tomography. The numerical predictions are carried out by solving a Young-Laplace equation using the pore morphology method. The experimental results and numerical predictions match well, including the effect of hysteresis in SWRC measurements. The spatial distribution of pore water and pore air corresponding to different capillary suctions is obtained from numerical predictions, providing greater insights into the hydromechanical behavior of partially saturated soils. The sensitivity of the voxel size on simulations is quantified by predicting the pore size distribution of Ottawa sand at three different tomography resolutions.