Hydromechanical versus seismic fatigue in progressive failure of deep-seated landslides

Deep-seated landslides are characterized by slow time-dependent movements induced by constant gravitational creep, and periods of acceleration associated with stress changes within the rock and the preexisting fracture network. Natural stress changes are typically caused by seasonal variations in surface temperatures and groundwater pressures, but also by rare and more violent events such as earthquakes. These processes are responsible for cyclic loading and unloading that leads to rock mass damage, fatigue and progressive failure with time, and may explain sudden, unexpected rock slope acceleration phases. In this work, we present a set of distinct-element models to investigate and compare the role of hydromechanical fatigue and seismic fatigue in the history of a deep-seated landslide. Hydromechanical fatigue arises from the fluctuation of pore pressure in fractures accompanying periods of significant groundwater recharge. In alpine areas, there are usually one or two cycles per year associated with snow melt and intense rainfall periods. In the presence of a deep confined aquifer within the rock slope, the amplitude of pressure fluctuations may reach values of 0.25-0.5 MPa. Hydromechanical models considering such pressure changes are able to reproduce observed slope accelerations-decelerations phases. Substantial accelerations may occur even in the absence of a cycle of exceptional amplitude, depending on the current state of fatigue. At a given time, one or more fractures may be critically-stressed, and a small change in applied stress is enough to cause local failure and induce a crisis of enhanced slope acceleration. This is augmented if the rock slope contains planes of weakness associated with its tectonic history, glacial debuttressing, etc. The corresponding behaviour is analogous to laboratory experiments on rock and metal fatigue. Similar fatigue effects have also been proposed for repeated earthquake events in seismically active regions. Unlike hydromechanical stress changes that occur in a more localized portion of the rock slope, those induced by passing seismic waves involve the entire rock mass. Moreover, the frequency and amplitude of stress perturbations due to earthquakes are different than those induced by pore pressure variations, although both likely contribute in combination to the degradation and progressive failure of large deep-seated landslides when both are present. The results presented illustrate that: (1) assessment of slope stability near critical infrastructure (e.g., hydroelectric reservoirs) should not only consider static stability, but also failure mechanisms related to fatigue that evolve over time and may result in an apparently stable slope progressing towards a critical state due to relatively small but repetitive stress changes; and that (2) characterization of fatigue processes affecting a deep-seated landslide requires multiparametric continuous monitoring at different observation points.