Unified modelling framework of flowslide triggering and runout

Abstract

Hydro-mechanical forcing, including sediment deposition, seismic loading, rainfall events and phreatic level rise, can alter the pore water pressure regime and trigger ground failure. In loose sands, such phenomena can manifest with rapid downslope movements displaying signatures of liquefaction and commonly referred to as ‘flowslides’. These can have catastrophic outcomes, yet the use of constitutive models for soils able to quantify the likelihood of liquefaction is usually restricted to triggering analyses, while the quantification of runout distance and velocity tends to be conducted through computationally expensive models based on rheological laws for fluids. To enable a seamless transition from the triggering to the runout stage, in this paper, a spatially condensed modelling framework of low computational cost is proposed, which allows for straightforward use of soil constitutive models in the analysis of the entire life cycle of a flowslide. The proposed framework couples the dynamic motion of a landslide mass with the pore fluid transients taking place within a pre-defined liquefiable zone. The constitutive law modulates the feedback between pore fluid diffusion and landslide movement, encompassing elastoplastic and viscoplastic soil models within the same formulation. Simulations show that partial drainage in the liquefiable zone has considerable effects on the triggering stresses initiating a flowslide. Specifically, the results imply that liquefaction risks are sharply suppressed when the ratio between the characteristic times of loading and pore pressure diffusion falls above a critical threshold influenced by the inelastic deformation response of the soil. Furthermore, it is shown that all the dynamics of flowslide propagation are dominated by the ratio between the characteristic timescales related to fluid consolidation, wave propagation and soil reaction time (i.e. soil viscosity). Although viscosity is not necessary to carry out propagation analyses, ignoring it leads to unrealistic estimates of peak velocity and runout distance. As a result, the model predicts that sand viscosity reflecting the emergence of a microscopic inertial regime is required to explain the flowslide propagation characteristics reported in the literature.