[maysmech]

Dear Foamers,
I want use LRR turbulent model but i didn't find any sample in tutorial files.
I don't know how is its properties which should be set.
Any suggestion will be appreciated.
Best,

[romant]

Hej,

did you have any success?

[maysmech]

Hi,
I used tut files which were k-e. their settings are almost the same. you need to change turbulentProperties from k-e to LRR.

[romant]

what is the field R that the calculation requires?

[maysmech]

R is reynolds stress tensor. i set it zero fixedValue for inlets, zeroGradient for outlets.
i used kqRWallFunction for walls. i am not sure about this one.

[romant]

thank you very much, I will try this and come back to this here when I have a result.

[romant]

it seems to work with those boundary conditions, givig approximately the same results as the non-reynolds stress models. the results are slightly improved over the standard k-epsilon

[maysmech]

I have same problem. i don't know why its result is approximately same as k-e!

Any idea?

[skyinventorbt]

Dear Maysam,

Kindly refer to the RSM Assessment, Disadvantages in Versteeg & Malalasekara "Introduction to CFD", where it has been mentioned.

"In some flows RSM performs just as poorly as k-epsilon model "
--
BTK

[skeptik]

I used to LRR in cyclone, so it had showed better results than k-eps both k-omega models.

I think it's a good model for special cases.

[skyinventorbt]

Dear skeptik,

Thats good to hear.

If you don`t mind, can you upload your LRR case in zipped format for reference.

I am using LRR for Jet flows and I am finding difficulty in convergence and some other problems. So it would be helpful if I go through some converged cases. By that hope I can figure out what is going wrong in my simulations.

Thanks in advance.
--
KANNAN

[cfdivan]

Hi maysmech,

The evaluation of the turbulence model is not that straightforward!

Mesh resolution and wall modelling strategy is quite critical for the conclusions of the simulations results. In any case, if there is no high strain in your flow, I cannot see the advantage of using a highly expensive turbulence model like RSM. For a large number of applications, the isotropic assumption is fair enough.

Regards,

[skeptik]

Here you can download cyclone- testcase

[skyinventorbt]

Dear Skeptik,
I used almost the same conditions as you used, but for axisymmetric jet case i am finding difficulties in convergence and getting physically possible solutions
--
KANNAN

[skeptik]

Originally Posted by skyinventorbt

Dear Skeptik,
I used almost the same conditions as you used, but for axisymmetric jet case i am finding difficulties in convergence and getting physically possible solutions
--
KANNAN

As i know, it's a common problem of the RSM-models.
But, possibly, bad convergence and non-physically solutions were reasoned by wrong boundary conditions. What is your case? Can you publish it?

[skyinventorbt]

Dear Skeptik,

Here is the first case which i tried.

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

--
KANNAN

[jason]

Hi there skeptik,

Thank you very much for you cyclone download, I also wanted to try out LRR for a cyclone so your post was very useful.

I was wondering, did you ever manage to verify the results on your cyclone? I ran your case first with simpleFoam with no turbulence and then with LRR activated. The velocity plot in the z-plane (which Im assuming is the tangential velocity profile?) still appears to be a non Rankine Vortex like this one for K-e.
http://www.waset.org/journals/waset/v59/v59-328.pdf

Whereas I am hoping RSM/LRR looks like this one...which is good enough for what I am currently doing
http://www.cfd.com.au/cfd_conf12/PDFs/040KAR.pdf

What do you think? Any comments appreciated.

Best Regards

Jason

[skeptik]

Originally Posted by jason

I was wondering, did you ever manage to verify the results on your cyclone? I ran your case first with simpleFoam with no turbulence and then with LRR activated. The velocity plot in the z-plane (which Im assuming is the tangential velocity profile?) still appears to be a non Rankine Vortex like this one for K-e.
http://www.waset.org/journals/waset/v59/v59-328.pdf

Whereas I am hoping RSM/LRR looks like this one...which is good enough for what I am currently doing
http://www.cfd.com.au/cfd_conf12/PDFs/040KAR.pdf

Hi Jason,

About my case. The velosity profile was not correct. But at least RSM results were better than k-eps. Sorry, I can't remember what was right and what was wrong exactly.

My cyclone was just a test-case, i've used third-party mesh. May be problem in initial conditions or in the mesh discretisation. At least about that case i'd say that mesh needs to be refined in near-wall region. For instance by 'refineWallLayer" function. Also as i remember there was tetra-mesh. It could be converted by polyDualMesh which reduces mesh size. Finally i didn't try to change fvSchemes.

[skeptik]

Here you can see differences.
Vortexes are little bit bigger and have little bit higher frequency. But not so high as in experimental case.

a10_comp2.jpg

[jason]

Hi again,
Thanks for the reply, I'm testing a snappy mesh with layers but I think the mesh is a little too coarse in the middle for my liking. So, I will build a pure hex mesh similar to the two cited papers and begin again.
BR
Jason