Multiscale insights into Sliding Surface Liquefaction through DEM simulations

Abstract

Recognizing the mechanisms that trigger liquefaction is critical for developing reliable models to prevent landslides. The tendency for liquefaction to occur generally decreases with increasing soil density. However, when grain fragmentation occurs, the material becomes more contractive, making liquefaction possible even in relatively dense samples. This phenomenon was first recognized and named Sliding Surface Liquefaction (SSL) by Kyoji Sassa’s research group (Soils Found, a=Vol 36, 1996, pp.53-64 ), who reported comprehensive laboratory studies on the topic. Yet, the mechanisms at the grain scale remain poorly understood. To advance in the understanding of SSL and support the development of predictive models, we investigate the links between micro- and macromechanical behavior in crushable granular materials subjected to constant volume shearing. We perform two-dimensional simulations using the Contact Dynamics Discrete Element Method, focusing on the effects of particle fragmentation strength and grading evolution during undrained shearing until liquefaction. The results reveal that higher densities and particle strength delay the onset of liquefaction. At high densities, regardless of the strength of the particles, grading during crushing asymptotically approaches an ultimate distribution, which depends on the initial density and is not associated with the occurrence of liquefaction. Although the amount of grain fragmentation is lower in looser samples, liquefaction occurs in earlier stages than in denser cases.