An experimental study of embankment dam breaching

There are thousands of embankment dams around the world which represent a significant economic and ecological risk should any of them fail. In previous failure events such as the Baldwin Hills Reservoir near Los Angeles which failed in 1964, destroying 277 homes and killing five people and causing $73 million in property damage. Other failures include the Buffalo Creek coal tailings dam in West Virginia which failed in 1972, killing 125 people, injuring over 1000 more and leaving thousands homeless and the Teton Dam in Idaho which failed in 1976 killing 14 people and causing over $1 billion in damages, and the recent Situ Gintung Lake Dam near Jakarta, Indonesia which failed March 26, 2009 killing almost 100 people. The goal of this study is to provide the experimental basis for a new numerical dam breach model which will help to address many of the shortcomings identified in the literature which are due primarily to lack of knowledge. Shortcomings have been identified in the existing models in the areas of breach initiation, breach location, breach morphology, hydraulics of flow over an embankment, sediment transport equations, geomechanics of the breach, and modeling composite embankments. For example, it has been said that there is an uncertainty of about 50% which can be expected from estimates of peak discharge for the existing numerical models (Franca and Almeida, 2004). To address the existing limitations, two embankment test series were developed. The first series of tests was conducted in a smaller flume in which compaction was varied. These tests shed light on the effects of compaction as well as the effects of side walls in testing and some of the effects that unsaturated soils and slope stability have on the breach process. The second series of tests was conducted in a larger flume with varying initial breach geometry as well as varying the filtration characteristics of the embankment. These tests demonstrate the role of initial breach geometry in dam breach mechanics and also help with understanding the importance of slope stability and unsaturated soil mechanics. One of these runs also included a clay core to study the effects of composite construction. These two test series were supported by data obtained by a series of geotechnical investigations as well as a series of erosion characteristics tests which were both conducted using the dam construction materials. From these test series, the following conclusions have been drawn: (i) A four phase breach process is proposed based on quantitative analysis of breach morphology. (ii) Initial breach geometry does not appear to have any effect on the peak breach outflow. (iii) The density of the embankment plays an important role in determining the shape of the outflow hydrograph as well as the peak flow. (iv) The breach is widest at the downstream toe of the embankment and smallest at the crest, but widens again upstream of the crest. (v) The slope failure mode does not appear to have any impact on breach characteristics. (vi) Tailwater conditions have a notable effect on breach morphology and reservoir volume also appears to have an important effect on sediment deposition characteristics. It should be noted that these conclusions have not been drawn for the purposes of improving dam construction techniques. Rather, they are useful for developing a more advanced numerical breach model in the hope of improving the accuracy of risk assessments in downstream flood zones. A new set of criteria could then be developed to define the new flood elevations and improve warning times for people who may be at risk