Laboratory Experiments on the Failure of Coarse Homogeneous Sediment Natural Dams on a Sloping Bed

We present the results of a systematic series of 168 laboratory experiments that examine the critical conditions for the failure of landslide dams, which obstruct the course of mountain rivers. The experiments were carried out by using three different sediment materials with a quasiuniform grain size distribution and with a flume bed slope angles that ranged between 0 degrees and 5.71 degrees (0-10%). Three main typologies of dam failure were observed for increasing values of the dam's downstream-face angle: (1) overtopping; (2) headcutting, which led to the formation of an erosion channel on the dam's downstream face that progressively migrated up to the dam crest; and (3) initial slide of large part of superficial layer of the dam's downstream face, which was followed by headcutting. The experiments focused on the second type of failure to provide a safety criterion based on the upstream reservoir level. The quantities that govern this phenomenon have been identified and a functional relationship is proposed based on the dimensional analysis and curve fitting of the minimum level of the upstream reservoir leading to dam failure. A comparison of both the experimental findings and field data that are available in literature shows that the proposed relationship generally provides a conservative estimate for landslide dams in which comminution effects (due to fragmentation of material in the landslide process) are negligible (dam volume lower than 10(6) m(3)). Finally, the applicability to larger volume landslide dams or to geometric configurations other than those investigated is discussed.