Multiphase and Multiphysics Modelling of Rainfall Induced Failure in an Experimental Hillslope

Annual precipitation and its intensity have increased worldwide since the start of the 20th century and represent two weather and climate change indicators related to rainfall-induced landslides. Although these landslides can occur in a very short time, the hydro-mechanical conditions that precede them can take several hours or days to develop. In this context, understanding the mechanisms of rainfall-induced landslides and their numerical modelling is topical for reducing risks to human life, facilities and infrastructure and economic loss. In this work, a large-scale experimental hillslope subjected to a controlled rainfall is studied numerically. Sensors and optical fibres were placed in the slope to monitor water pressure and moisture content in the failure layer, as well as axial strain and temperature in the failure surface. The outflow at the toe of the slope was also measured. The experimental hillslope is modelled as a fully coupled variably saturated hydro-thermo-mechanical problem in dynamics. A general geometrically linear finite element model based on Hybrid Mixture Theory and enhanced with Taylor-Hood finite elements is used. The soil response is modelled with the Bolzon–Schrefler model for non-isothermal variably saturated soils. The failure mechanism is further assessed using the global second-order work criterion. The comparison between the experimental and numerical results is analysed using the KGE indicator, showing that the model is capable to correctly reproduce both the hydrological dynamics leading to failure and the strain along the failure surface. The global second-order work criterion proved to predict the proneness to failure.