CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   SU2 Shape Design (https://www.cfd-online.com/Forums/su2-shape-design/)
-   -   SU2_DOT blow-up during Shape Design (https://www.cfd-online.com/Forums/su2-shape-design/212222-su2_dot-blow-up-during-shape-design.html)

Oliver November 28, 2018 14:00

SU2_DOT blow-up during Shape Design
 
1 Attachment(s)
Dear all,


I've ran into some problems while running an optimization process for an airfoil. More specifically, the SU2_DOT component blows-up upon first starting. The error is "Initial Mesh Sensitivity Residual= NaN".



I'm attaching the configuration file used for the procedure. The discrete adjoint solver has no issues with this configuration and the mesh, successfully converging to a stable value for the geometry sensitivity, and roughly 8 orders of magnitude drop in the adjoint variables. Similarly, the flow solver converges using a Cauchy condition on the drag (up to 1e-7), and 7-8 orders of magnitude drop for pressure and turbulence residuals.


The grid I used has only QUAD cells, and is relatively well refined, with Y+ < 1, a 1.15 growth ratio for the estimated BL region and around 300 grid points around the wall. I've also tried to maintain high grid quality throughout, so I don't believe this is the source of problems (and the good converge of SU2_CFD confirms).



So far, the changes I've attempted in the grid deformation procedure have been unsuccessful, but they have been rather trial and error, not having much knowledge related to grid deformation algorithms.



Does anybody now a more robust setup so as to avoid blow-up? Any help in this direction would be much appreciated.


Cheers,
Oliviu

tomp1993 April 26, 2019 09:26

Hi
Try:
listing all of your DV's in DV_KIND (separated by commas and spaces) and DV_PARAM (separate each set of brackets with semicolons and spaces).
using KIND_ADAPT=NONE and ADAPT_BOUNDARY=YES (this might do nothing but i use it and it doesn't hurt)
also just be aware that some of your surface marker names in DEFINITION_DV are 'AIRFOIL' whilst others are 'airfoil'. SU2 is case-sensitive.
i tthink i may have started a thread on the same issue as this previously so have a scroll through the forum and that might be more helpful than what i'm suggesting now.
best wishes
Tom

tomp1993 April 26, 2019 09:27

P.S. it might help if you post the output of SU2_DOT as it may give some insight.

angupraveen July 27, 2019 02:24

Sensitivity analysis failure
 
Dear all,

I'm trying to do a 3D SE2A Aircraft wing optimization using SU2. Unfortunately, I am unable to get the following things working.

1. The solution doesn't converge even after 5000 iterations both in ADJOINT and DIRECT solvers. I mean it keeps computing until the max iter value.

2. After the iterations are done, I tried to compute the sensitivity using SU2_DOT_AD and it gives of_grad.dat file with '0' as the value with the following message before

"Design variable (FFD_SETTING) number 0."

Any help would be highly appreciated

Thanks in advance

Angu


All times are GMT -4. The time now is 01:49.