ABSTRACT
Simulations of flow in river bends, especially at the outer bank, can differ based on the treatment of the air-water interface and on the turbulence model used. Here, four detached eddy simulation (DES) models and one large eddy simulation (LES) model are compared based on their ability to reproduce experimental results in a channel bend, with a specific focus on the reproduction of the outer bank cell. The Multidimensional Universal Limiter with Explicit Solution (MULES) interface model and the LES rigid lid models are able to reproduce the outer bank cell of secondary circulation, while the rigid lid models employing DES are not. The combination of the k-omega SST turbulence model (which is employed at the wall boundaries of the DES simulations) together with the symmetry condition applied at the rigid lid boundary, appears to disrupt the production of the outer bank cells in these simulations.
