Hi,

I wonder if anyone come across any good paper discussing in what circumstances (e.g. effect of boundary condition etc) & why Reynolds Stress Model (RSM) isn't predicted the flow better than two-equation model? I read quite some CFD papers comparing different turbulent models & RSM seems not predicting flow very well.

I start wondering if solving more equations really make things better....

I do not know about particular papers because you have not described the flows you are studying.

An RST model will work better than a two equation model when terms that are accurately handled in the RST model are modelled in the 2 equation model and those terms are important to the physics of the flow being considered. Those terms are essentially the convection and production terms of the Reynolds stresses.

A physical process where this difference is very apparent is the generation and suppression of turbulence due to streamline curvature. This is caused by the structure of the stress field being convected while the rate of strain field rotates due to the curvature. An RST model handles this accurately but a two-equation model will rotate the stress field to be aligned with the rate of strain field due to the Boussinesq model used for the stresses. The process is effectively unrepresentable by the modelling used in the two equation model.

The answer is always to use models appropriate for the physics being studied. For a range of flows there is not much difference between RST and two equation models, for some two equations models are poor and RST models are OK (above example) while for others both are poor.

Hi Adam,

although the RSM models do more accuratly model the transport processes of Reynolds stresses there still is modelling required for dissipation, pressure strain, and diffusion of Reynolds stresses. As pointed out above the RSM now needs no modelling for the convection and production of Reynolds Stresses.

Remember in the RSM (most anyhow) there are 7 equations....6 Reynolds stresses and a modelled dissipation equation.

For the dissipation equation most RSM models use the standard epsilon equation which is generally believed to be the weakest part of the 2 equation models as well.

That is what I believe to be the biggest problem to overcome.

You can find various papers which discuss specific deficiencies ie one pressure-strain closure vs another, etc. but I do not have the references handy.

Regards,

Bak_Flow