Thermo-porp-medchanical modelling of the creeping and failure phases of catastrophic landslides

After the sadly famous Vajont landslide in 1963, many attempts have appeared in the literature to give an explanation for the surprisingly high speed reached by the slide. One possible explanation is that thermo-poro- mechanical effects dominated the process: the loss of strength and the consequent high velocity can be explained by taking into account the generation of heat and excess pore pressures within the shear band during the sliding process. On the other hand, in common with other slides, Vajont exhibited an initial stage of creeping behaviour, where movement at slow and relatively constant velocity took place over a long period of time, many orders of magnitude longer than the final phase of catastrophic acceleration. What affects the transition between the two regimes and why this transition may occur in some landslides but not others is not fully understood. The first comprehensive thermo-poro- mechanical landslide models for the final catastrophic phase employed a two-yield- surface model for the soil, consisting of the Coulomb line and a cap, and took into account friction-velocity softening on the basis of ring shear test observations. The governing equations of the problem are the equation of pore pressure diffusion-generation, where a pressurization coefficient linking pore pressure generation to temperature changes is introduced, the heat equation and the dynamics equation. Using a similar mathematical setting, it is also possible to use a variant of the Modified Cam clay model for the soil, in which case the pressurisation coefficient arises as a function of other, better established material parameters. The creeping phase of the landslide is also possible to model, using ad hoc assumptions for the dependence of the soil’s frictional properties on temperature and/or strain rate. Here we consider the possibility that frictional heating of the slip zone, combined with non-linear, strain-rate friction hardening, suffices to explain the whole range of behaviour observed: the creeping phase can be modelled as the combined result of friction hardening and thermal softening, whereas the final phase of catastrophic acceleration kicks in if thermal softening comes to dominate the process. Whether this happens or not depends on local conditions, like the distance of and the conditions at the boundaries, as well as material parameters. In a departure from the assumptions of linear elasticity made in literature, we model soil behaviour using a variant of Modified Cam Clay that allows shrinkage of the elastic domain with increasing temperature. The thermal softening and strain/strain rate friction softening mechanisms previously considered are now not necessary for modelling the observed phenomena, leading to an overall simpler formulation.