CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   SU2 (http://www.cfd-online.com/Forums/su2/)
-   -   nan "tke + specific dissipation" while restarting the SST solution (http://www.cfd-online.com/Forums/su2/115032-nan-tke-specific-dissipation-while-restarting-sst-solution.html)

taxalian March 22, 2013 06:30

nan "tke + specific dissipation" while restarting the SST solution
 
Hi SU2 users,
I am using the current version "SU2_Rev1206". I have a nicely converged solution and i want to run the solution for another flow angle by using the previous solution as restart one, it works well with SA turbulence model. But know i am testing the SST model. So i turned on the parameter:


% Restart solution (NO, YES)
RESTART_SOL= YES % NO


The solver successfully reads the "restart_flow.dat" but it cannot initialize/obtain the residual values of turbulent kinetic energy and specific dissipation, see the error below:

------------------------------ Begin solver -----------------------------

Iter Time(s) Res[Rho] Res[kine] Res[omega] CLift(Total) CDrag(Total)
1 8.322500 -0.069912 nan nan 0.469435 0.042631

------------------------------------------------------------------------


Even though the turbulence values are very nicely converged.
I also don't find any configuration parameter that can be seen during the solver run as follows:

------------------------------------------------------------------------
Free-stream turb. kinetic energy (non-dimensional): 4.2525e-05
Free-stream specific dissipation (non-dimensional): 60.699
------------------------------------------------------------------------

May be these values are obtained internally or they might be available in the next version release.
Sometimes it also helps by changing these values in case of convergence problem or restart solutions.

Thanks and Regards.

fpalacios March 23, 2013 23:54

Hi,

The SST implementation is in a V&V stage. Please follow our twitter account @su2code if you want to be informed about the most recent activity in turbulence modeling.

Best,
Francisco

taxalian March 24, 2013 06:40

Hi Fransico,
Thanks for your reply, well as you know from my earlier thread titled "Overpredicted Trailing Edge Separation", where you also posted some replies. I have successfully validated the trailing edge separation test case, i got pretty descents result compared to experiments and CFL3D predictions.

http://turbmodels.larc.nasa.gov/naca4412sep_val.html.

I would also like to perform the V&V for the SST turbulence model too, but due to this problem, i am not able to run the simulation at desired flow angle.

I will post my complete results on the forum and let me know how can i submit these results in your V&V stage of NACA airfoil trailing edge separation.

Amrita Lonkar May 2, 2013 13:06

Hi,

We are very interested in displaying your results of the V&V of the trailing edge separation over a NACA 4412 airfoil. If you have some time, could you please send us your results along with a short write along the lines of this test case:

http://adl.stanford.edu/docs/display...ulent+RAE+2822

We will add it to the test case folder under your name.

Thank you for offering to help with the V&V of the SST model. I will need to look into the issues you're getting in more detail and get back to you when we have it resolved.

Thanks,
Amrita

taxalian May 3, 2013 14:09

Hi Amrita,
Sure, i will send you the results. I will give you the SA results for this test case, as i still don't have the time to perform the SST computations. Let me know how do you want to have the results. In the form of presentation or what else.

Thanks and regards.

taxalian May 5, 2013 15:26

Hi Amrita,
I would like to say that the SST problem is resolved and i am finished with the SST computation and the results seems to be looking very good in terms of the Cp distribution, but of course i need to the similar post-processing i did for the SA turbulence model. I will try to provide you the test case details according to the template as you mentioned in your earlier post.

Thanks.

Amrita Lonkar May 5, 2013 15:34

Hi,

If you could email us the results with a short description of the problem, the configuration options you used to solve it, and a picture of the mesh you used with the boundary conditions, that will be very nice. We will then add the test case on our website. Thank you.

I am glad that you have been able to fix the issue with SST restart, and we are extremely interested in a validation study using the SST turbulence model, so if you have any luck, please contact us and we will add a test case on the website for this too.

Thanks again for your interest in SU2,

Amrita

taxalian May 18, 2013 16:19

SU2 SST Verification and Validation Test Case
 
3 Attachment(s)
Hi SU2 Users,

It is worth mentioning, that in the V&V stage of SU2 code with the SST turbulence model. I have successfully completed the validation of 2D NACA 4412 Airfoil Trailing Edge Separation with SU2 code using SST turbulence model. (http://turbmodels.larc.nasa.gov/naca4412sep_val.html).

SU2 results compare very well with the existing CFD CFL3D results and with the experimental results.

The similar study has been already carried out with SU2 code using SA turbulence model. For details see the following thread:

http://www.cfd-online.com/Forums/su2...eparation.html

In can be observed from the above mentioned thread, that the SA results predicted a smaller trailing edge separation compared to the current SST results which predicted much better trailing edge separation compared to the experiment.

fpalacios May 21, 2013 00:18

Quote:

Originally Posted by taxalian (Post 428474)
Hi SU2 Users,

It is worth mentioning, that in the V&V stage of SU2 code with the SST turbulence model. I have successfully completed the validation of 2D NACA 4412 Airfoil Trailing Edge Separation with SU2 code using SST turbulence model. (http://turbmodels.larc.nasa.gov/naca4412sep_val.html).

SU2 results compare very well with the existing CFD CFL3D results and with the experimental results.

The similar study has been already carried out with SU2 code using SA turbulence model. For details see the following thread:

http://www.cfd-online.com/Forums/su2...eparation.html

In can be observed from the above mentioned thread, that the SA results predicted a smaller trailing edge separation compared to the current SST results which predicted much better trailing edge separation compared to the experiment.

Thanks!
This validation is very useful for us. We are collecting all this data, and we will update the webpage with the validation for the next major release (v2.1).

For the time being we have upload the new developers release 2.0.4. We have improved a lot the stability of the code. If you have a while give it a shot and increase the CFL number of your simulations (Krylov based methods are highly recommended).

Thanks again for your help,
Best,
Francisco

taxalian May 22, 2013 14:20

Quote:

Originally Posted by fpalacios (Post 428781)
Thanks!
This validation is very useful for us. We are collecting all this data, and we will update the webpage with the validation for the next major release (v2.1).

For the time being we have upload the new developers release 2.0.4. We have improved a lot the stability of the code. If you have a while give it a shot and increase the CFL number of your simulations (Krylov based methods are highly recommended).

Thanks again for your help,
Best,
Francisco

Hi Francisco,
I tried the release 2.0.4 with higher CFL number i.e 50 to 100 in combination with Krylov based methods, but after 10 to 15 iterations the simulation diverges abruptly. Even though my SST computations were carried out with CFL = 20 without Krylov method.


All times are GMT -4. The time now is 23:43.