Comparison between standard and non-linear k-ε turbulence models for three-dimensional simulation of turbulent flow in a meandering open channel

A home-made CFD code was used to simulate complex flows in a compound meandering open channel with two bends. The first set of simulations solves the unsteady ReynoldsAveraged Navier-Stokes equations (RANS) using an isotropic turbulence closure (standard k-ε model). The second set solves unsteady RANS using a non-linear k-ε model which takes into account anisotropic effects. The transport equations for turbulent kinetic energy and dissipation are solved for both models, but a non-linear description of Reynolds stresses was added for the anisotropic model. CPU time affection is negligible compared to isotropic k-ε, however numerical convergence is harder to achieve. The FiniteVolume code solves the non-hydrostatic 3-D unsteady RANS equations.The code is layer-structured and can simulate free surface movement.Wall function approach was used near solid walls and bottom. The test case is a 12 m long and 0.25 m wide meandering open channel with two 180? bends taken from literature. Strong curvature on successive bends creates important non-hydrostatic flow behaviour. Hydrostatic 1-D or 2-D models cannot realistically represent this kind of three dimensional features. Flow in meandering channels is characterized by helical cells generated by centrifugal forces. This secondary velocities configuration is important in order to predict practical issues in engineering like shear stresses on the banks and bed. Both turbulence modelling approaches were tested in this domain. The results obtained were compared with experimental data and previous LES simulations performed by other authors. Anisotropic turbulence model was found to pick up turbulent flow patterns better than usual isotropic models do. Good agreement was found between numerical secondary velocities and measured ones. Three variants of the anisotropic turbulence model were tested and discussion was made about their suitability.