Abstract: In complex service environments involving water and soil loads, hydraulic geotechnical structures are highly susceptible to suffusion disasters, potentially triggering structural failures. This study employs the computational fluid dynamics-discrete element method (CFD-DEM) to investigate poorly graded soils, which are commonly encountered in engineering projects such as earth-rock dams and slopes. By considering three different levels of fines content and three distinct particle shapes, we explore the process of suffusion in poorly graded sand. Through an analysis of particle migration trajectories, displacements, and contact numbers, we elucidate the micro-scale characteristics of fines movement during the suffusion process. The migration effects of particles are categorized into four types: clogging, voiding, detouring, and washout. Upon statistical analysis of the evolution of migration effects throughout the entire process, it is evident that during the initial stages of suffusion, most fines are in a detached state. However, as seepage flow takes effect, particles gradually come into contact with each other, resulting in a significant decrease in detachment effects. Consequently, other non-steady-state effects, primarily detours and washouts, gradually increase as moving particles become constrained by their neighbors. The clogging rate of particles escalates, and ultimately, the majority of particles settle into a clogged state, indicating that the sample has reached a steady state.

Key words: CFD-DEM, suffusion, particle morphology, clogging effect, gap-graded soil