Loss of slope support due to base liquefaction: comparison of 1g and centrifuge landslide flume experiments

Recent physical model studies of landslide behaviour in loose, granular slopes have illustrated that the soil at the base of a slope may be of heightened risk of liquefaction as this area is located directly in the path of monotonic shearing events (i.e. localised toe failures) and is often saturated by groundwater. The objective of this paper is to further investigate the factors that influence the possible transition of a small toe slide to a larger and higher mobility liquefaction flow failure. Specifically, this paper investigated the influence of stress-level and soil-layer thickness between two different laboratory testing methods in controlling whether a slide transitions to a flow in physical model flume tests. Observations from two flume experiments were contrasted: one in which suction stresses were designed to dominate (a 0.33 m soil layer in a large-scale 1 g test) and a second enhanced gravity test performed in a geotechnical centrifuge where the body stresses are more representative of field scale landslides (i.e. behaviour representative of a 1.5 m thick soil layer). While the centrifuge test was brought to widespread liquefaction failure under increasing seepage flow, the 1 g flume test experienced liquefaction, but only displaced 25 mm. The results from this study highlight the dominant effect matric suction can have on the mobility of landslides, and provides a clear demonstration of the danger of misinterpreting the behaviour observed in small scale 1 g models as being fully accurate representations of larger field-scale landslide behaviour.