Impact of dam and reservoir parameters on peak breach discharge predictions for two models

Scope and method of study. Two computational models for embankment dam breach are reviewed, NWSB (National Weather Service BREACH) and SIMBA (Simplified Breach Analysis). The models' predictions of peak breach discharge, Qp, were evaluated against two contrasting, well-documented physical breach tests and against a synthetic data set. Physical test embankments were approximately 6 ft in height with contrasting material properties. The first was highly erodible achieving peak discharge in about 0.5 hrs of overtopping, the second was erosion resistant and never breached into reservoir. The synthetic set was developed from a prediction equation and historical cases of dam failure. The synthetic set dam heights, hd, ranged from 5 to 400 ft with variations of storage volume relative to height, reservoir shape, and material rate parameters. The material rate parameter for NWSB was median particle diameter, D50; SIMBA's was erodibility, kd.;Findings and conclusions. While observations were more of a comparative nature for the synthesized set, the laboratory breaches provided known Qb as a basis. NWSB proved wholly incapable of modeling the material properties of the laboratory breaches, predicting near instantaneous breach for both experiments. SIMBA was able to predict Qp and even timing with remarkable accuracy. For the synthesized data sets, both models exhibited sensitivity to changes in height and relative storage volume. NWSB responded more to changing storage for higher dams, while altering this parameter had more effect on lower dams for SIMBA. NWSB was sensitive to changes in D50 only at a mid-range of dam heights; while SIMBA was sensitive to most variations in k d, especially for smaller, less erodible dams. Discontinuities in the estimates of Qp were noted at or near hd = 50 ft, and can likely be attributed to height-dependent processes in both models. The slope of Qp as plotted again hd, closely matched that of the prediction equation: rather than a validation of the equation, it is actually a function of the height to storage relationship. While NWSB uses obtainable material properties, they were inadequate to describe cohesive behavior of the material. Erodibility, kd, was in these cases a more appropriate material property for modeling dam breach.