Calculation of primary and secondary flow and boundary shear stresses in a meandering channel

Turbulent flow in a meandering channel is computed with two Computational Fluid Dynamics (CFD) codes solving the Navier–Stokes equations by employing different turbulence closure approaches. The first CFD code solves the steady Reynolds-Averaged Navier–Stokes equations (RANS) using an isotropic turbulence closure. The second code is based on the concept of Large Eddy Simulation (LES). LES resolves the large-scale turbulence structures in the flow and is known to outperform RANS models in flows in which large-scale structures dominate the statistics. The results obtained from the two codes are compared with experimental data from a physical model study. Both, LES and RANS simulation, predict the primary helical flow pattern in the meander as well as the occurrence of an outer-bank secondary cell. Computed primary as well as secondary flow velocities are in reasonably good agreement with experimental data. Evidence is given that the outer-bank secondary cell in a meander bend is the residual of the main secondary cell of the previous bend. However, the RANS code, regardless of the turbulence model employed, overpredicts the size and strength of the outer-bank secondary cell. Furthermore, only LES is able to uphold the outer-bank second secondary cell beyond the bend apex until the exit of the bend as turbulence anisotropy contributes to its persistence. The presence of multiple secondary cells has important consequences for the distribution of shear stresses along the wetted perimeter of the channel, and thereby the sediment transport in meandering channels. Consequently, even though LES is expected to compute the bed-shear stresses along the wetted perimeter of the channel with a higher degree of accuracy than the RANS model, comparisons between LES and RANS computed wall shear stresses agree well. These findings are useful for practitioners who need to rely on RANS model predictions of the flow in meandering channels at field scale.