Reconstructing the microstructure of real gap-graded soils in DEM: Application to internal instability

Abstract

This paper introduces a new technique that enables considering the impact of soil fabric on microstructure of internally unstable gap-graded soil. The influence of soil fabric on erodibility of fine particles was studied by transferring the real soil fabric from a micro-computed tomography (μCT) image to discrete element method (DEM) using a new image processing workflow. The pair correlation function (PCF), a two-point correlation between particles in images, was used as a microgeometrical parameter to assess the impact of image filters on the quality of the μCT image. Comparison between PCF and particle size distribution of the raw μCT and reconstructed images suggest that anisotropic diffusion and non-local mean filters reproduce the real soil fabric samples most effectively. DEM results showed lower coordination number but slightly higher stress reduction factor for the real fabric, indicating that the real contact geometry results in having fine particles more involved in force network.