Numerical Investigation of Seismically Induced Rock Mass Fatigue as a Mechanism Contributing to the Progressive Failure of Deep-Seated Landslides

Abstract

The importance of earthquakes in triggering catastrophic failure of deep-seated landslides has long been recognized and is well documented in the literature. However, seismic waves do not only act as a trigger mechanism. They also contribute to the progressive failure of large rock slopes as a fatigue process that is highly efficient in deforming and damaging rock slopes. Given the typically long recurrence time and unpredictability of earthquakes, field-based investigations of co-seismic rock slope deformations are difficult. We present here a conceptual numerical study that demonstrates how repeated earthquake activity over time can destabilize a relatively strong rock slope by creating and propagating new fractures until the rock mass is sufficiently weakened to initiate catastrophic failure. Our results further show that the damage and displacement induced by a certain earthquake strongly depends on pre-existing damage. In fact, the damage history of the slope influences the earthquake-induced displacement as much as earthquake ground motion characteristics such as the peak ground acceleration. Because seismically induced fatigue is: (1) characterized by low repeat frequency, (2) represents a large amplitude damage event, and (3) weakens the entire rock mass, it differs from other fatigue processes. Hydro-mechanical cycles, for instance, occur at higher repeat frequencies (i.e., annual cycles), lower amplitude, and only affect limited parts of the rock mass. Thus, we also compare seismically induced fatigue to seasonal hydro-mechanical fatigue. While earthquakes can progressively weaken even a strong, competent rock mass, hydro-mechanical fatigue requires a higher degree of pre-existing damage to be effective. We conclude that displacement rates induced by hydro-mechanical cycling are indicative of the degree of pre-existing damage in the rock mass. Another indicator of pre-existing damage is the seismic amplification pattern of a slope; frequency-dependent amplification factors are highly sensitive to changes in the fracture network within the slope. Our study demonstrates the importance of including fatigue-related damage history—in particular, seismically induced fatigue—into landslide stability and hazard assessments.