Author(s): Milad Abdollahpour; Paola Gualtieri; Carlo Gualtieri

Linked Author(s): Milad Abdollahpour, Paola Gualtieri, Carlo Gualtieri

Keywords: BFSF; Vegetation; Numerical simulation; OpenFOAM

Abstract: Aquatic vegetation affects flow dynamics in natural channels, reducing flow velocity and increasing turbulence intensities. Due to the complex geometries of real plants, aquatic vegetation is commonly modeled as rigid cylinders which are used to represent plants with few branches and leaves. Vegetation is often found inside lateral cavities of natural channels. The flow in such cavities could be investigated using the well-known geometry of the backward-facing step (BFS), which is frequently applied as the benchmark in CFD studies. In the present study, the flow over a BFS where a cylinder was placed immediately downstream of the step was investigated through numerical simulation. This geometry is intended to represent a cavity with a vegetation stem in a natural channel. Numerical simulations were carried out with the open-source CFD toolbox OpenFOAM. The RANS turbulence models used were k-ε, RNG k-ε, k-ω, SST k-ω, RSM. The Reynolds number based on the step height ranged from 75 to 9000, hence covering both laminar and turbulent flow. The results of laminar BFSF without cylinder were compared with the literature numerical and experimental. The error for streamwise velocity profiles and reattachment lengths was less than 8.1% and 3%, respectively. The cylinder significantly modified the structure of recirculating flow over the BFSF. Also, the cylinder increased the skewness of the velocity profiles, and the location of the maximum velocity shifted towards the upper wall. The performance of RANS models was compared with literature experimental data (PIV) and numerical data (DNS) of BFSF without cylinder. The errors in predicting reattachment length and streamwise velocity profiles ranged from 2.3% to 28.5% and from 7.8% to 14.5%, respectively. The most accurate model in predicting velocity profiles was the SST k-ω, followed by the k-ω, RNG k-ɛ, the standard k-ɛ, and RSM. Such better accuracy of the SST k-ω model was expected as this model was already recommended for cases with adverse pressure gradient and flow separation because it addresses the advantages of k-ω and k-ε models.

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221153

Year: 2022