This study proposed a novel hybrid resolved framework coupling computational fluid dynamics (CFD) with discrete element method (DEM) to investigate internal erosion in gap-graded soils. In this framework, a fictitious domain (FD) method for clump was developed to solve the fluid flow around realistic-shaped coarse particles, while a semi-resolved method based on a Gaussian-weighted function was adopted to describe the interactions between fine particles and fluid. Firstly, the accuracy of the proposed CFD-DEM was rigorously validated through simulations of flow past a fixed sphere and single ellipsoid particle settling, compared with experimental results. Subsequently, the samples of gap-graded soil considering realistic shape of coarse particles were established, using spherical harmonic (SH) analysis and clump method. Finally, the hybrid resolved CFD-DEM model was applied to simulate internal erosion in gap-graded soils. Detailed numerical analyses concentrated on macro- -micro mechanics during internal erosion, including the critical hydraulic gradient, structure deformation, as well as particle migration, pore flow, and fabric evolution. The findings from this study provide novel insights into the multi-scale mechanisms underlying the internal erosion in gap-graded soils.

A new fluid-solid coupled numerical approach is developed by combining the CFD (Computational Fluid Dynamics)- DEM (discrete element method) with the pore network model (PNM) to simulate the erosion of the soil-rock mixture. The pore network with pore and pore pipes is constructed based on the particles and updated regularly. A relationship equation is derived between the permeability scalar for micro-scaled pore pipe and the anisotropic permeability tensor for macro-scaled fluid element. By the Delaunay-PorePy-PFC3D program framework, the erosion process of the soil-rock mixture with different fine contents (FCs) is simulated. The results show that the PNM-CFD-DEM model can meet the computational accuracy for simulating the rule-arranged uniform particles. The duration of the erosion stage is different for specimens with different FCs. The PNM-CFDDEM model can reproduce the particle erosion paths in different specimens, as well as the adjustment of the pore network between their coarse particles. The preferential drag forces in the discrete portion take into account the pore network formed by the state of the particle buildup within each fluid element.

This study proposes a resolved framework coupling computational fluid dynamics (CFD) with discrete element method (DEM) to simulate the sedimentation of granular sand. Realistic sand particles were reconstructed by spherical harmonic representation combined with the multi-sphere clump method, and a fictitious domain method for irregular clumps was further developed to solve the fluid-particle interaction. This resolved CFD-DEM offers a direct and robust approach for computing real fluid forces on irregular-shaped granular sands, without relying on any empirical drag force models. Initially, the effectiveness and accuracy of the proposed CFD-DEM were validated through a series of single-particle free settling simulations for various ideal-shaped particles. Critical fluid-particle interacting behaviors in terms of drag force and wake structure were mainly investigated and corroborated with experimental data. The study subsequently progressed to simulate the sedimentation processes of various granular sand assemblies composed of realistic-shaped sand particles, utilizing the proposed CFD-DEM. Detailed numerical analyses concentrated on particle-scale mechanics during sedimentation, including settling trajectories and velocities of particles, as well as the coordination and anisotropy of inter-particle contacts. The results and findings gained from this study provide a novel insight into the micro-mechanisms underlying the sedimentation and accumulation process of granular soils in geological environments.