[wilson]

What are the differences between the Shear Stress, Reynolds BSL and SSG turbulence models. I'm looking at the flow through a butterfly valve and trying to establish which model is best to use and why?

Thanks

[ghorrocks]

I don't know what you mean by the shear stress model. Reynolds BSL and SSG turbulence models are variations on Reynolds Stress Models where the Reynolds Stress tensor is directly simulated rather than modelled such as for the 2-eqn turbulence models.

For flow through a butterfly the SST turbulence model would be my recommendation. I don't think you will get any additional accuracy from any form of RSM model. If SST proves inadequate you should consider SAS/DES models.

[crmorton]

[QUOTE=ghorrocks;239616]I don't know what you mean by the shear stress model. QUOTE]

I think shear stress model is actually referring to the SST model. Doesn't that acronym stand for Shear Stress Transport?
Anyway, SST for sure is the best out of the models suggested.

[Josh]

If you go to the CFX Help file in either Pre, Solver, or Post, they have an excellent section describing each turbulence model, its advantages and disadvantages, and links to the people who created/adapted them.

Just search "turbulence model" in the index of the Help file and underneath you will find every model supplied with CFX.

Additionally, try CFD Wiki - good resource.

[wilson]

So what exactly are your reasons for saying that the SST model would be best? Does it give better accuracy compared with the others?

[Josh]

The k-e model provides mostly accurate resolution for many freestream conditions (nothing complicated, though), but fails to predict accurate separation amounts/locations and other near-wall effects. The k-w model, conversely, models near-wall effects (like separation) effectively, but is oversensitive to inlet conditions.

The SST model uses the freestream resolution of the k-e model and the near-wall effects of the k-w model. Basically, SST is the man.

[crmorton]

unless you have the computer power to complete LES or DNS of course!

[Josh]

And the time. On my 8 processor CPU, an LES/DES would take about a week to run for the simulations I perform.

[Jade M]

A SST model may be more appropriate for separated flows. SST was designed to give highly accurate predictions of the onset and amount of flow separation under adverse pressure gradients. THE SST model is recommended for high accuracy boundary layer simulations. Offers similar benefits as standard k–ω. Dependency on wall distance makes this less suitable for free shear flows.


SSG (Speziale, Sarkar, Gatski) Reynolds Stress – based on k-e; should be better than k-e in theory but may not be better in practice; some of the deficiencies for the simulation of boundary layers are inherited from the underlying k-e equation; particularly accurate prediction of flow separation is problematic
BSL (Baseline) Reynolds Stress – w-based rather than e-based for accurate prediction of flow separation