Damage-based time-dependent simulation of the progressive failure of large alpine rock slopes

Large slopes in alpine environments undergo a complex long-term evolution from glacial to postglacial conditions through a transient period of paraglacial readjustment. During and after this transition the interplay among rock strength, topographic relief and morpho-climatic drivers can promote different types of slope instability, from sudden failures to large, slow (yet potentially catastrophic) rockslides. Deep-seated rockslides are commonly characterised by time-dependent displacements with superposed “long-term creep” components, often considered as evidence of progressive (sub-critical) slope failure (Amitrano & Helmstetter 2006; Lacroix & Amitrano 2013), and “seasonal creep” components resulting from coupling between hydrological triggers and landslides (Cappa et al. 2014; Crosta et al. 2014). Modelling the progressive failure of large rock slopes is key to predict future displacements and potential catastrophic evolution for risk analysis and early warning. Failure forecast for “mature” rockslides with well-developed sliding shear zones and weakened, permeable landslide masses mostly relies on analytical, statistical or numerical models accounting for the hydromechanical response of landslides to rainfall or snowmelt (Guglielmi et al. 2005; Crosta & Agliardi 2003; Crosta et al. 2014). Nevertheless, the geometry, internal structure, strength and hydrology of rock slopes are usually poorly known and likely evolve throughout deglaciation and paraglacial adjustment. Understanding the long-term evolution and progressive failure of large rock slopes requires that we account for time-dependent effects of unloading during deglaciation, permeability and fluid pressure fluctuations, ongoing displacement, and fracture development (Riva 2017; Riva et al, in review). All these aspects are related to a convincing description of rock mass damage processes, from a subcritical (progressive) to a critical (catastrophic) stage. Nevertheless, time-dependent damage evolution under gravitational load and variable external drivers remains poorly explored. Other open issues include: a) difficult use of time-dependent rheological models, unable to account for localization unless a shear zone is pre-defined; b) largely unknown spatial and temporal patterns of water occurrence in the slope during and after deglaciation (Ballantyne 2002); c) limited understanding of the relationships between brittle failure and permeability on the scale and under stress conditions typical of rock slopes; and d) uncertainty in material property upscaling and model calibration for time-dependent models. In this context, we developed DaDyn-RS, a tool simulating the damage-based, time-dependent long-term simulation of real large rock slopes in alpine environments.