Turbulence Model for Depth-Averaged Flows in Navigation Installations

A turbulence-based computational procedure was developed for the modeling of depth-averaged, time-dependent flows with recirculation zones and applied to the prediction of flow fields in navigation installations induced by hydropower releases. The computational procedure uses a depth-averaged version of the k -ε model and a fractional step method for the solution of the flow and transport equations. The advection terms are treated using a characteristics approach with high-order Hermite bicubic spatial interpolation to minimize numerical diffusion. The model is not restricted by the rigid-lid assumption and can handle the flooding and drying of sloping boundaries. Model performance is evaluated by comparisons with measured flow fields and the significance of discrepancies is discussed. The predictive capability improves significantly when results are compared to those of a constant eddy viscosity model; the improved predictability is explained by the variability in calculated eddy viscosities—and consequently in the relative importance of the turbulent transport terms—in the different flow regions.