[lorenzo95]

Hi all!

I'm using SU2 v.7.0.3 to compute the flow field over a wing of particular shape.
Because of the low Re and the geometry of the wing, a transition model is needed. I already tried Spalart Allmaras with BC model but it does not fit experimental data.
Literature suggests that the gamma- Re_theta model should obviate this problem.

In the SU2 source code I found that classes for LM model are already defined, but when I try to activate it with the option:
KIND_TRANS_MODEL= LM
a bad termination error occurs during the Numerics Preprocessing phase.
I wonder if this is caused by a still incomplete model, or because the option is deactivated for some reason.

Is there a solution to this problem?
Is it possible to activate the current version of the LM model, even if it is not yet tested?

My cfg, if needed, is the following:

[pcg]

Hi,
I had a look and it seems the "creation logic" for that solver is broken, so there is no option to make it work, we need to fix the code.

From what I recall that solver is very experimental (judging from the number of debug statements throughout its code).

I know there are folks working on transition models, your best bet would be to open an issue on the SU2 github and maybe one of them will point you to a working branch.
If no one replies I will try to fix the code.

[lorenzo95]

Thank you very much, I'll try as you suggest!

[REJISH]

Hi Lorenzo,

Did you find solution to your problem..?
I am also having the similar issue.
It would be very useful if you give me some inputs to resolve this.

Best regards,
Rejish

[lorenzo95]

Hi,
yes, in the end I found a solution to my problem: surprisingly for me, it was not related to the transition model.

I found that Bas-Cakmacioglu provided reasonably accurate results when coupled with JST convection scheme. I previously used Roe and that led to completely inaccurate results. I still do not have an explanation why this happened, but I think it was related to the slow speed freestream.

My (unskilled) advice is to try different convection schemes, using the B-C transition model and see which one best fits with your experimental data. However probably you won't be able to obtain a perfect match if your 3d geometry is complex and/or you have large pressure gradient areas.

Hope this helps!

[whizkid]

Hello,

There would be two problems that the BC model does not give you accurate results.

1- Please check your y+ values, they must be below 1.
2- Please check the turbulence intensity you specify. I am guessing it is multiplied by 100 in the code. So for 3%, I think you should enter 0.03 in the config file (did not try myself but someone put a multiplication of 100 in the source code turb_sources.cpp at line 143 which possibly affects the Menter correlation )

In most cases Roe scheme works better.

Regards,

[REJISH]

Hi Lorenzo,

Thank you very much for the details.
Can you point me the literature that has the experimental data which may useful for me to compare the transition results?

Hi Whizkid,

Thank you very much for the details. I will look into the code and test the turbulence intensity setting in the configuration file as you suggested.

[lorenzo95]

Yes of course you must follow also recommendations by whizkid, which I took for granted.
Experimental results to which I referred are those of the particular problem you are trying to simulate and can be in terms of Cp, Cf or other parameters.
Moreover there is a tutorial on SU2 website (https://su2code.github.io/tutorials/...al_Flat_Plate/) and you should read accurately the paper where the model is presented to understand its capabilities and limitations (https://journals.sagepub.com/doi/abs...54406217743537).
Best regards