CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   SU2 (https://www.cfd-online.com/Forums/su2/)
-   -   SU2 Falcon error in using fsi_computation.py (https://www.cfd-online.com/Forums/su2/201689-su2-falcon-error-using-fsi_computation-py.html)

rocco.bom May 8, 2018 10:15

SU2 Falcon error in using fsi_computation.py
 
Good morning,

I'm trying to make fsi_computation.py run with the new Falcon release. I had no problem making it run with the previous Raven version. Now, instead, the code gives error when running the wrapped SU2 function SetInitialMesh(). The last thing the code is able to print before the error message is:

Min. volume: 1.73089e-09, max. volume: 2.64949.
Min. distance: 1.16009e-06, max. distance: 24.9477.

So, going back to the source I targeted where the code actually prints this message, and it happens in file /Common/src/grid_movement_structure.cpp at line 541 (I underline it is in case of version Falcon).

Now, comparing the .cpp fine with the one belonging to the previous version I notice a lot of changes. Is it possible the actualized version has some bug or it is just a problem of retro-compatibility of the fsi_computation.py tool I'm using, which requires some changes in the python interface?

I'll be glad to anyone who can give me some tips about this issue. :)
Thanks

rocco.bom May 15, 2018 09:28

Possible bug in CILUPreconditioner
 
I found the problem (which may actually be a bug ). It occurs when the wrapped function SetInitialMesh() is run. This directly involves the core of SU2 as I found out the error is given during the run of the function CILUPreconditioner (inside matrix_structure.cpp) which, as much I understood, builds the preconditioning matrix for the solution of the linear system at the base of the mesh deforming procedure.

SU2 uses by default ILU precoditioner whose relative function (see above) has remarkable differences from the Raven release (I didn't go too much into details). Same error I get if using Jacobi preconditioner.
At contrary, everything seems fine if using LU_SGS preconditioner and the fsi_computation.py finally works. Maybe it's the case to point it out to the developers involved in this part of the code.


All times are GMT -4. The time now is 02:54.