[Tobi]

Hi all,

I have just a very simple question.

In which case should I use a low Re turbulence model. Is there any lower Re-number limit where you should use low Re turbulence model like:

(pipe flow)
------------------------

Re < 2300                      laminar
Re > 2300 - 5000               low-Re-turbulence model
Re > 5000                      normal turbulence model

Thanks in advance
Tobi

[nsf]

Hi Tobi,

As far as I understand matters low-Re models are used when you are resolving the fields all through to yplus < 5. When you are using high Re models then your should be yplus > 30.

If you search for "openfoam low re models" you'll find a couple of good threads. Here are some of them,

http://www.cfd-online.com/Forums/ope...-openfoam.html

http://www.cfd-online.com/Forums/ope...-testcase.html

http://www.cfd-online.com/Forums/ope...estigated.html

Best
Nicolas

[lakeat]

Hi Tobi,

Low-Re Number Models (LRN) as compared to High-Re Number Models (HRN) are not used for global low reynolds number flows. I think the name at its first place looks very misleading. It's been asked by my colleagues for many times.

The LRN models means their mathematical formulations are capable to resolve the turbulence flow down to the wall (y^+<1). The low-re is local in this physical sense. High Re models are supposed to work with wall functions. So for a HRN model, even if you refine the wall grid to very fine, it won't work! But for a LRN model, you refine the wall grid and the results will converge and agree very well with the experiments. Of cause, some LRN models are also in practice used with wall functions. They are said to work in HRN mode.

But please keep in mind, for global low/medium Reynolds flows, popular RANS models on the market such as the SA and the SST model and the v2f model, etc., are insufficient to handle them. Not even that, even when the global Re number is high, but whenever say there is strong vortex shedding and huge "low-re" regions, these LRN models are also incapable to predict them well. It seems this is the bottom neck of RANS framework. The best guys in RANS fields are still working on it. Even though it seems hopeless to me. URANS can help in a few situations, but not all. The evolution of Reynolds stress (its time scale) seems just too difficult to model.

For global low re flows or flows with transitions (they are the cases that you HAVE TO account for the transition effect in your simulation), a few "transitional models" are said to predict better, like some three equation models. I think they do not fall into the category of LRN and HRN models, even though they are mostly based on them.

Hope this helps.

[samiam1000]

Hi guys: a question about turbulence models.. ..can I use the standard kEpsilon avoiding the wall functions? How can I get this?

Thanks a lot,
Samuele

[smrutimaya]

Hi All,

A low Re flow flow for INTERNAL FLOWS is below Reynolds number 500 as referred in :

http://www.sciencedirect.com/science...60132310000636

It says a high Re flow is above Re 10,000 .

I have experienced that,
a high Re models like k-epsilon and its variants as well as, k-omega and variants, capture phenomena nicely above Re above say some 15000.

Below Re 10,000 deviation is more. So we can't use high Re models.


For External Flow, Re < 60,000 (Page 5 of below reference)

https://ntrs.nasa.gov/archive/nasa/c...0080023453.pdf


Regards,
Smruti