Investigation of dormant landslides in earthquake conditions using a physical model

The response of a loess-mudstone landslide model to seismic activity is examined using centrifuge shaking table tests and 2D seismic input waves. The physical model of a dormant landslide is subjected to simulated seismic signals of different input amplitudes, and the dynamic response is analyzed in terms of relative peak ground acceleration (PGA), which is used to assess deformation and failure mechanisms of the slope. Relative PGA increases with increasing landslide height, reaching maximum values at the crest, while values at the toe are slightly larger than those in the middle of the slope. Relative PGA is the highest along the outer surface and weakest along the sliding surface. Horizontal values are generally larger than vertical values, and the amplitude of the seismic input strongly affects the landslide dynamic response. The PGA amplification factors increase considerably with increasing input amplitude, reaching maximum values at 0.3-0.4 g, and then decrease. The largest displacement is observed at the crest, while displacements in the middle part and at the toe increase steadily with higher input amplitudes, and displacement at the crest increases at a higher, irregular rate. Deformation is most apparent at the crest, developing from cracks into collapsed soil deposited at the toe. Local shallow sliding is observed along the landslide surface, and the dormant sliding body may generate displacement along the sliding surface caused by earthquake reactivation. A numerical model of the landslide provides results consistent with centrifuge model testing; both indicate similarity in seismically induced dynamic responses and failure characteristics.