Controls on the Breach Geometry and Flood Hydrograph During Overtopping of Non-cohesive Earthen Dams

Abstract:

To gain insight into overtopping failure of natural and constructed earthen dams, large-scale experiments were done with dams built of damp, compacted, fine-grained sand with nearly trapezoidal cross sections. Both the height and top width were varied, so that the cross-sectional shape ranged approximately from that of embankment or moraine dams to that of landslide dams. Breaching was initiated by cutting a shallow notch across the dam crest and allowing water escaping from a finite upstream reservoir to form its own channel. The channel invariably developed a stepped profile, with steps migrating upstream and coalescing into a headcut. Hydraulic control became established at the channel head, or breach crest, an arcuate erosional feature. Intersection of the migrating headcut with the breach crest triggered rapid, runaway downcutting. Step development and headcut retreat proceeded much more slowly in low dams with wide tops than in tall dams with narrow tops. Photogrammetry and underwater videography showed that the slope of the retreating headcut stayed near the friction angle of the sand and that the evolving flow cross section at the breach crest maintained a geometrically similar shape in 3D. Contrary to a common assumption in dam-breach models, the hydraulic control section was rarely perturbed by slope failures. The idea that hydraulic control reflects tradeoff between sediment delivery by slope failures and sediment removal by fluvial processes is wrong.

Hydrographs were quite reproducible when the time datum was chosen as the time that the migrating headcut reached the breach crest. Peak discharge increased systematically with the initial reservoir water level Z₀, a puzzling fact given that that flow depth at the breach crest was nearly always much smaller than Z₀, implying that the hydraulics at the breach crest, and thus the local shear stress, should have been nearly independent of Z₀. This conundrum may be resolved if breach-crest erosion rate, and hence the flood hydrograph, are primarily controlled not by shear stress at the breach crest, but rather by headcut retreat, which is known to depend upon both and tailwater depth. Because slope failures into the breach channel cause temporary tailwater impoundment, the actual mechanism by which slope failures affect the flood hydrograph is thus identified.