NUMERICAL MODELING OF AN EARTHQUAKE-INDUCED LANDSLIDE CONSIDERING THE STRAIN-SOFTENING CHARACTERISTICS AT THE BEDDING PLANE

The damage caused by an earthquake-induced landslide can generally be classified as either a limited deformation or a catastrophic failure. From an engineering point of view, the latter can be much more dangerous because the sliding mass may continue moving until it collides with another object. If a catastrophic failure occurs near a river, the debris may block the river, causing serious damage to the adjacent area. Therefore, examination of the mechanism of such catastrophic slope failures is important with respect to the mitigation of earthquake disasters in mountainous districts, although numerical modeling of such phenomena is rather difficult. In the present study, a new numerical model is developed to simulate an earthquake-induced catastrophic landslide that occured at a typical dip slope, namely, the Yokowatashi Landslide in Japan. In this case, the upper part of the bedrock on the planer tectonic dip surface slid more than 70 m. Only shear-strength degradation at the bedding plane could cause such a long-distance traveling failure. To investigate the strain-softening characteristics of the materials that filled the bedding plane, a series of laboratory tests involving undisturbed block samples was performed. The measured stress-displacement relationships under cyclic loading were numerically modeled as a newly proposed elasto-plastic constitutive model to be used in numerical simulations of landslide, based on the dynamic finite element method. The observed phenomena were appropriately simulated by the proposed method. The mechanism of catastrophic failure is discussed in detail in this paper in order to clarify the relationships between the strain-softening characteristics and the global slope stability. Our newly proposed method to evaluate the possibility of a catastrophic failure was applied to the landslide, and the moment when the slope becomes unstable was able to be predicted. The results confirm that the proposed method can predict the catastrophic failure of a slope.