Author(s): Pedro Xavier Ramos, Laurent Schindfessel, Tom De Mulder, Joao Pego

Linked Author(s): Pedro Xavier Ramos, Laurent Schindfessel, Tom De Mulder, Joao Pego

Keywords: Open-channel confluences; Large-Eddy Simulation; Secondary flow; Upwelling;

Abstract: Right-angled confluences of open-channels with rectangular cross-sections are common configurations used to study the complex confluence hydrodynamics by means of lab experiments and numerical simulations.
Large-Eddy simulations (LES) of confluences have traditionally been validated using bulk quantities such as water levels and streamwise mean velocities. Given the cost of LES, simplifications are usually adopted (wall-models, rigid-lids as water surface, etc.). However, recent studies have shown that the predicted secondary flow turns out to be sensitive to the curvature of the top boundary of the numerical domain when it comes to secondary motions and, therefore, the secondary flow shall be used within the validation of the numerical simulations. Weber et al. (2001) provide an experimental dataset in a confluence with a main channel and a tributary having equal widths and concordant beds, at a moderately low downstream Froude number (Frd≈0.37) and at different discharge ratios. between the upstream main channel and the downstream main channel, q.
In his paper, the authors found that the sensitivity of secondary flow to the rigid-lid treatment reported in Ramos et al. (2019) for the case with q≈0.25, does not show up for the concordant bed case of Weber et al. (2001) with the more moderate q≈0.58, despite the two cases sharing the same Froude number.
Moreover, it has been shown, for the present case, that the discordance between the bed levels of the tributary and the main channel, which often occurs in natural confluences, alters the secondary flow patterns significantly, namely, distorting the mixing layer.

DOI: https://doi.org/10.3850/38WC092019-1454

Year: 2019