Development of a Turbulence-Resolving, Three-dimensional, Free Water Surface Numerical Model for Recirculation Eddies in Grand Canyon

In the Colorado River in Grand Canyon, sand bars, which are built in recirculation areas downstream of channel expansions are valuable resources, particularly as natural habitat for endangered native fish and recreation sites for recreation. Since the closure of Glen Canyon Dam in 1963, there has been a reduction in the size of recirculation eddy bars. Simulated floods in the Colorado after tributary flood sediment input from the Paria River are being investigated as a method of rebuilding recirculation eddy beaches. Time-averaged, two-dimensional (and quasi- three-dimensional) numerical models have been employed to predict deposition during these beach/habit-building test flows. However, behind channel expansion areas, flows are strongly three-dimensional and the cross-channel flow is driven primarily by upwelling boils along the eddy fence that are neither stationary in time or space. Furthermore, these strong vertical motions along the eddy fence preclude use of the hydrostatic assumption. In this study, a non-hydrostatic three-dimensional numerical model is presented in order to calculate flow velocity in channels having rapid channel expansions. This model employs the Large Eddy Simulation (LES) turbulence modeling technique. LES uses spatial filtering to separate flows into gird scale and sub-gird scale rather than time averaging, thus it directly calculates large-scale turbulent motions. Also, this model employs a moving grid system and the Body Fitted Coordinate (BFC) system. These grid and coordinate systems allow the model to calculate time-dependent free water surface levels induced by large-scale turbulent motions. The model_fs calculation results are compared to existing experimental results of an open channel flow expansion in a laboratory flume. The comparison shows that the model succeeds to reproduce several key features of the flow, such as the temporally- and vertically-averaged horizontal recirculation eddy structure, and the time-averaged cross-stream vertical flow structure, which has inward flow (into the recirculation zone) at the surface and outward flow near the bed. The calculated water surface level also seems adequate. Its maximum is at the upstream side of the channel expansion and 3% higher than that of the minimum at the downstream end of the recirculation zone. Calculation results show that horizontal eddies are intermittently produced along the separation line between the recirculation zone and the main channel, and those produced eddies move into the recirculation zone while expanding their size.