HR: 16:30h
AN: H54B-03    [Abstracts]
TI: Three-dimensional Modeling of Flow Over Two- and Three-dimensional Dunes
AU: * Burman, A
EM: arburman@usgs.gov
AF: US Geological Survey, Box 25046, MS-413, Lakewood, CO 80225 United States
AU: Nelson, J M
EM: jmn@usgs.gov
AF: US Geological Survey, Box 25046, MS-413, Lakewood, CO 80225 United States
AU: Shimizu, Y
EM: yasu@eng.hokudai.ac.jp
AF: Hokkaido University, Department of Civil Engineering, Sapporo, Hok 062 Japan
AB: To predict the growth of bedforms from an initial bed configuration to finite-amplitude form for a wide variety of flows and sediment characteristics, we believe that the details of the near-bed turbulence field must be resolved, especially with regard to the interaction of the turbulence field with spatial accelerations. Unfortunately, most closure-type models are notoriously poor for resolving this type of interaction, requiring the specification of empirical coefficients for different types of flows. Avoiding the use of turbulence closure requires the application of a direct numerical simulation or at least a large-eddy simulation model. To investigate the feasibility of using such an approach in a morphological evolution model, we tested a turbulence-resolving computational model. The model can be used with extremely small grid spacing (i.e., Kolmogorov scale) as a direct numerical simulation, or with larger grid spacing and a sub-grid-scale closure as a large-eddy simulation. To evaluate the performance of the approach, we compared computations of velocity, pressure, and various turbulence quantities to values measured over both two- and three-dimensional bedforms in a laboratory setting. Even using relatively large grid spacing, the model predicts mean velocity fields quite accurately, with almost no dependence on the sub-grid-scale closure. Similarly, pressure at the bed is well predicted and, as a result, form drag estimates are accurate. As expected, turbulence quantities such as Reynolds stress components are more sensitive to grid spacing, with smaller grids yielding improved, but not perfect, comparison to measurements. The frequency structure of vortex shedding from the bedform crest is accurately reproduced, as is unsteadiness in the shape and streamwise extent of the separation eddy in the lee of the bedforms. Model convergence is good using a rigid-lid approximation, but is problematic for free-surface calculations. Future work is directed at improving the treatment of the free surface and combining the flow model with grain-scale computations of sediment movement.
DE: 1815 Erosion and sedimentation
DE: 1824 Geomorphology (1625)
SC: Hydrology [H]
MN: 2004 AGU Fall Meeting