Overpredicted Trailing Edge Separation
Hi SU2 users,
I am doing the 2d CFD validation case for the trailing edge separation over the NACA-4series single element airfoil with a low mach number.
The grid is very good C-type, fully structured max-y+ = 0.2, but the solution converges not well. I tried to used initially the JST scheme with multigrid (later on single grid), then switched to upwind scheme ROE_2nd_Order with and without multigrid no success. I also tried with smaller and larger CFL numbers. Last but not the least i used SA turbulence model and i am still using the SU2 v1.1.
I would really appreciate any kind of advise to improve the results, that i will attach later on together with the grid as snapshots.
Thanks in advance.
Thanks a lot for your feedback, we are improving the code everyday, and every month we upload the current version of the code our webpage.
We are very concern about the importance of the V&V, and the SA has been fully validated.
In order to help you, could you please first download the current version of the code
You will find a folder called NACA0012_RANS, where you have the config file and mesh file that we are using for the NACA validation (http://turbmodels.larc.nasa.gov/naca0012_val_sa.html)
Thanks for using SU2
Thanks for sharing the current version of the code and the configuration file for the NACA0012 V&V case. I will try to sought out the problem and let you know about the progress soon.
Many thanks and regards.
How do i compile the earlier attached current code version, as i am using cygwin in win7 to run SU2 v1.1. I have not able to compile this one. Waiting for your reply.
regards and thanks in advance.
Do you have access to Python in your cygwin environment? You could try to use the python build script to run the various makefiles. The script, build_SU2.py, can be found in the SU2Py/ directory.
There are also individual makefiles for the different modules that you can call directly. These are located in the corresponding config/ directories, i.e. SU2_CFD/config/.
Hope this helps!
Hello Fransico and Tom together,
I really appreciate your help and it helps me alot to resolve the problem by using the current SU2 version you provided and the Naca0012 config file to launch SU2 computation.
Well for the validation case i got pretty descent result compared to experiments and CFL3D predictions.
See the attached cp distribution & mesh pictures. The SU2 results have been obtained using SA turbulence model, SST is under progress. I am compiling a kind of user guide kind of presentation which i will share with you guys. I think one can make a very good tutorial out of that and i am working on that.
How did you make this mesh?
Hi Fransico and Tom together,
See the attached file showing the comparison of SU2.SA result with existing CFL3D.SA & Experiment result for the validation of 2D NACA 4412 Airfoil Trailing Edge Separation :)(http://turbmodels.larc.nasa.gov/naca4412sep_val.html).
SU2 results compare very well with the existing CFD CFL3D results and with the experimental results.
I have used my own hybrid grid not the one provided on the website which is fully structured. The farfield is about 20c while the available grids contains about 100c dimensions. Nevertheless the results are very comparable and at the moment the computations have been carried out with SA turublence model.
Thanks and regards for your support and correspondence.
It seems you made this mesh in gridgen...
Thank you for sharing this. We would be interested to see anything else that you prepare for this case.
Hi SU2 team,
Thanks for the reply, for sure i will submit the complete presentation upon completion of this validation study to SU2 developers team.
Thanks and regards.
Dear SU2 Team and Taxalian
I am trying to simulate the case of NACA 4412 at 0 deg angle of attack.
The mesh has a O-grid topology and created by pointwise by normal extrusion.
The conditions are the same as the paper AIAA 2014-0243 case-2 except that the angle of attack is 0 degree.
but i am not able to get convergence even till 10,000 iterations.
There is a lot of oscillation in the lift coefficient.
May i know few things :
1. How much time did it take to simulate the case ? Can you express it in CPU time, machine used,etc. One of my friend left his simulation to run for 150,000 iterations and he got a acceptable convergence.
2. Can i get the mesh and the configuration file for the result which is there in the paper as it seems to have a fairly less number of cells in comparison to the one i am using ( 95,513 cells ) and a good match with the exp results?
For validation purposes we used the meshes that you can find in
My recommendation is to use the 2D PLOT3D files. There is a translator from plot3d to su2 in
something like $./p3d2su2_2D.py -f n4412_113-33.p2dfmt should work.
Once you have the mesh my recommendation is to use an already working config file like
SU2 lead developer
Thank you sir.
I have few doubts :
1. The mesh which i seen in the AIAA 2014-0243 paper does not contain the mesh which is there on the site which you cited. It has the mesh which Taxalian has shown.
2. I used pointwise to open the 2D PLOT3D file of the mesh but there is a problem is setting the boundary condition as the airfoil and the connector connecting it to the downstream is coming up as one connector. Is there a way to circumvent this problem ?
3. Sir, as asked earlier : kindly tell me how much time did it take to simulate the case ? Can you express it in CPU time, machine used,etc. One of my friend left his simulation to run for 150,000 iterations and he got an acceptable convergence.
|All times are GMT -4. The time now is 22:26.|