Hydromechanical factors influencing seepage-induced suffusion in internally unstable soils

Internal erosion, and in particular suffusion, poses an escalating threat to the stability of water-retaining earth structures such as embankment dams and levees globally, due to increased pressures from climate change and growing populations. While it is well-established that the hydromechanical regime within the soil body influences this process, existing studies conducted under varying testing configurations have come to conflicting conclusions on the exact mechanisms in process. This study addresses this knowledge deficit by introducing a novel rigid-walled centrifuge permeameter, purpose-built for investigating a wide range of seepage path lengths and effective stress conditions. This apparatus facilitates the manipulation of established centrifuge scaling laws via a parametric modelling of models approach, allowing the spatial onset and progression of suffusion to be tracked through interstitial pore pressure measurements and post-test particle size analysis.

In total, seven centrifuge permeameter tests are conducted, accompanied by a 1-g test undertaken in the same cell. An under-filled, gap-graded soil comprising of a mixture of silt and sand is used in all tests. The suite of tests allows for the influence of both macroscale and granular seepage path length to be considered, as well as the effective stress gradient across the specimens. Notably, the hydraulic loading history emerges as a paramount factor governing the spatial progression of suffusion. This insight underscores the importance of considering not just the initiation of fine particle migration but, more crucially, the ongoing evolution of suffusion.

The results of this study reveal that suffusion is a phenomenon characterised by substantial spatial variability, thereby challenging conventional modelling approaches based on macroscale assumptions of homogeneity and Darcy's flow. Through a mesoscale interpretation of the results, it becomes evident that variations in hydraulic loading regimes, boundary conditions, and measurement point locations have historically contributed to the discrepancies observed between tests conducted in different apparatuses. Moving forward, a more unified approach between particle scale and element scale modelling is needed, so that suffusion can be accurately characterised within real structures.