A dam or levee breach caused by overflow erosion is difficult to evaluate during an overtopping event due to difficulty in accessibility and quickly changing conditions. However, for assessment of risk associated with breach time and ensuing downstream consequences, the erosion rate of embankment soils during this process needs to be evaluated. Soil erosion and water depth measurements were taken during flume tests using a Shallow Water Lidar (SWL) system scans. The tests were conducted in a 1-m-wide tilting flume on three cleans and gravel soil mixes with a median grain size D50of 2, 5,and 20 mm. The box measured 0.45-m-wide ×1.2-m-long ×0.25-m-deep. Due to the confined environment of the flow in the flume, the acting bed shear changes with hydrodynamics of the flow differently from under a uniform flow. The scour hole generated in the test box reaches equilibrium when the acting bed shear is equal to the critical shear. Standard machine learning techniques were used to image soil and water profiles from noisy Lidar data. First, the data are filtered using zonal-averaging and then based on the filtered data; the methodology selects the best profiles from a competing set based on the minimum error each profile produces on the data. Once the profiles are obtained, erosion rates and bed shear are computed, and a qualitative assessment is carried out to understand the relationship between temporal and spatial dependence of erosion rate on bed shear and soil particle size. Erosion rate and shear stress reached their maximum value within the first 60–70 seconds of the test and spatially within 0.3 m from the upstream end of the test box. The erosion rate decreased by about 4 times from 0.13 cm/s to 0.03 cm/s as D50increased from 2 mm to 20 mm at the same acting bed shear. The erosion rate for both mixes is reduced over time; however, the rate of reduction for D50of 20 mm is much higher over the same test duration. The erosion rate was shown to be strongly correlated to the acting bed shear nonlinearly. The results indicate that the calculated spatial variation of shear stress over the duration of the tests is consistent with the formation of maximum depth of scour hole.
Scholarly Commons Citation
Ellithy, G. S., & Parida, S. S. (2022). Using Machine Learning in Estimating Changing Bed Shear over a Flume Test Box. Geo-Congress 2022, (). Retrieved from https://commons.erau.edu/publication/2175