How can we better predict flow resistance in rough-bed rivers? A numerical (CFD) approach

Predicting flow in rough-bed rivers, which are a common and important feature of upland river networks, is crucial for improved river management, as changes in flow discharge can affect sediment transport, cause flooding, and disrupt habitats. Standard grain-size approaches that predict flow resistance (e.g., D₈₄) do not perform well in rough-bed rivers because they are unable to account for the coarse topographic features that rough-bed rivers possess, such as exposed bedrock, potholes, bedrock ribs and boulders. The flow resistance derives from the interaction of the flow and the bed topography, which causes spatial variations in pressure and form drag, and modulates the shear stress available for sediment transport. However, the extent to which different components of the bed topography affect the flow resistance is not well understood. Thus, improvements in flow resistance prediction are necessary.

One feasible approach to fill this gap is to use computational fluid dynamics (CFD), as it is able to numerically resolve the pressure at the bed, at the resolution of the grid representing the bed topography. In this work, we developed a robust, validated numerical model to simulate a series of flow discharges over three different rough-bed river beds. Our CFD model fully resolves the Navier-Stokes equations in a three-dimensional, cartesian-gridded domain. The model captures the adjustment of the free surface at the air-water interface with the volume-of-fluid method, such that the simulated flows are not constrained by low Froude numbers. The three different bed sections were reconstructed from high-resolution topographic data from bedrock rivers with smooth, intermediate and rough topography (standard deviation of the bed elevation equal to 0.043 m, 0.083 m and 0.131 m at the chosen sections, respectively). As the CFD are replicating scaled flume experiments performed with the same bed topography, the topography in the CFD has also been scaled by 1:10 from the field dimensions. For each bed, we assess the pressure distribution, pressure gradient and form drag over the beds under five different flow depths and discharges.