Large eddy simulations for quasi-2D turbulence in shallow flows: A comparison between different subgrid scale models

Abstract

In this study, the performance of the horizontal large eddy simulation module, developed at the University of Leuven (HLES-KULeuven module) is assessed. A comparison between different subgrid scale models has been carried out. The study is concerned with the non-rotating and unstratified flows. The results of the simulation for an oscillatory backward facing (BFS) flow are presented in case of an expanding flume based on a one-length scale approach and a two-length scale approach. Three subgrid scale (SGS) models have been tested: Smagorinsky SGS model (Smagorinsky, J., (1963). General circulation experiments with the primitive equations, I. the basic experiments. Monthly Weather Review, 91(3), 99–164), Uittenbogaard SGS model (Uittenbogaard, R.E., and van Vossen, B., (2004). Subgrid-scale model for quasi-2D turbulence in shallow water. Shallow Flows. Jirka and Uijttewaal (Eds.), Taylor & Francis Group, London, ISBN 90 5809 700 5) and a proposed two-length scale approach. The first two models are considered to be a one-length scale models. A simulation without a subgrid scale model for the horizontal mixing has also been conducted. In all simulations, a quadratic friction model parameterizes the dissipation produced by the 3D-subdepth scale turbulence. The two-length scale concept uses a newly mixing length formulation for the quasi-2D turbulence and doesn't depend on the filter width in contrast to the one-length scale approach, in which the mixing length is function of the filter width. The outputs of the HLES-KULeuven module have been compared with the experimental data taken from Stelling, G.S., and Wang, L.X., (1984). Experiments and computations on separating flow in an expanding flume. Dept. Civil Engineering, Delft University of Technology, Report 2–84.). The two-length scale approach has been validated with experimental data from SERC Flood Channel Facility at HR Wallingford. In general, there is a qualitative agreement with the experimental data. It has also been found that the two-length scale approach produces more elongated and less isotropic vortex than the one-length scale models.