[JinZhiyi]

Dear developers and users,

I am using SU2 version 3.0 "eagle", when I am running the rotating testcase of "caradonna_tung", it seems can not be convergence, and when it runs to 22 iterations, some errors occur like follows:

CsysSolve::modGramSchmidt:: w[i+1] = NaN
libc++abi.dylib: terminating with uncaught exception of type int
Abort trap: 6


Can any nice guy, tell me what happened??
Or figure out a solution, I want to compare the numerical results from SU2 with the classic experiments.


Thanks in advance




Jin

[Emily1412]

If you ran in parallel, try changing the np (e.g. from -p 14 to -p 16) and see if the error disappear.

[JinZhiyi]

Originally Posted by Emily1412

If you ran in parallel, try changing the np (e.g. from -p 14 to -p 16) and see if the error disappear.

Sorry， I don't know how to install parallel....
Is there any way for a serial one run successfully.

[Emily1412]

Sorry, I don't know the exact solution to this error. It seems that the error occurred frequently in both serial and parallel run. So far, I haven't figured out the reason of the error...

[Santiago Padron]

Hi Jin,

We are aware of this problem, this is due to the SU2 version 3.0 "eagle" having new options that are not compatible with the test case you are trying to run.
We will be releasing an updated version of the code and test cases shortly.
In the mean time you can try using an older version of the code that should be compatible with your configuration file.

Also, a comparison to experiment of this case in SU2 can be found here
http://adl.stanford.edu/papers/AIAA-2012-3018.pdf

Santiago

[Hennet_m]

Dear all,

I have download and setup the new version of your very good CFD solver SU2: the 3.1 version.

Moreover, I have seen that the Caradonna & Tung case has been presented in the 2014 AIAA "Open-source Analysis and Design Technology for Turbulent Flows".

I would like to know if the tutorial case (which is located in the new archive test cases of the 3.1 version) worked?

In fact, I have tried and the solver seems to diverge. In a first step, I have generated the mesh with the cyclic boundaries using SU2_PBC which correctly generated the periodic mesh.
Then, I have launch CFD simulation using SU2_CFD and the calculation seems to diverge.

Have I done a mistake? I would really like to make comparison between SU2 and OpenFOAM on this case, my results using OpenFOAM seems to be not too bad.

Best regards,

Martin

[JinZhiyi]

Originally Posted by Hennet_m

Dear all,

I have download and setup the new version of your very good CFD solver SU2: the 3.1 version.

Moreover, I have seen that the Caradonna & Tung case has been presented in the 2014 AIAA "Open-source Analysis and Design Technology for Turbulent Flows".

I would like to know if the tutorial case (which is located in the new archive test cases of the 3.1 version) worked?

In fact, I have tried and the solver seems to diverge. In a first step, I have generated the mesh with the cyclic boundaries using SU2_PBC which correctly generated the periodic mesh.
Then, I have launch CFD simulation using SU2_CFD and the calculation seems to diverge.

Have I done a mistake? I would really like to make comparison between SU2 and OpenFOAM on this case, my results using OpenFOAM seems to be not too bad.

Best regards,

Martin

Hi Martin,

Can you please tell me how to use SU2_PBC to generate the periodic mesh with cyclic boundaries. I would like to try your method in my machine.


Best regard


Jin

[Hennet_m]

Hi JinZhiyi,

As it doesn't work, I am not sure at 100% that I am not doing mistake.
- In a first step, open the rot_caradonna_tung.cfg and set the MESH_FILENAME=mesh_caradonna_tung.su2
- Execute the command: SU2_PBC rot_caradonna_tung.cfg. It will create you the periodic mesh: mesh_caradonna_tung_periodic.su2.
- Change the MESH_FILENAME=mesh_caradonna_tung.su2 and execute the CFD flow by: SU2_CFD rot_caradonna_tung.cfg

Best regards,

Martin

[JinZhiyi]

Originally Posted by Hennet_m

Hi JinZhiyi,

As it doesn't work, I am not sure at 100% that I am not doing mistake.
- In a first step, open the rot_caradonna_tung.cfg and set the MESH_FILENAME=mesh_caradonna_tung.su2
- Execute the command: SU2_PBC rot_caradonna_tung.cfg. It will create you the periodic mesh: mesh_caradonna_tung_periodic.su2.
- Change the MESH_FILENAME=mesh_caradonna_tung.su2 and execute the CFD flow by: SU2_CFD rot_caradonna_tung.cfg

Best regards,

Martin

Hi, Martin, I follow your advice on generation of Periodic Boundary Mesh, also I noticed the rotating angular velocity in config file is wrong, according to the mesh provided, the blade should rotate around the Z-axial, so when i modify the rotating angular velocity to"Angular velocity about x, y, z axes: ( 0, 0, 130.9 ) rad/s", I got a converged result, despite initially you may see some warning message like "The solution contains 241 non-physical points." and the Residual can be vibrate for several steps.

However, when checking the surface-flow.dat, compared with the AIAA paper the developer provided, it seems the pressure coefficient don't match the data showed on the paper well.


Have a try

Jin

[economon]

Hi,

For the time being (we will likely be overhauling the periodic boundary condition soon in the code), I would recommend creating a mesh of the full computational domain for the two-bladed Caradonna & Tung case without periodic boundaries.

Also, note that the results given in the paper here

http://adl.stanford.edu/papers/AIAA-2012-3018.pdf

are based on the Euler equations. In this case, due to the inviscid assumption, the numerically computed pressures and shock location do not perfectly agree with experiment (although they are similar to inviscid results from other codes). For a more accurate validation of the C & T rotor with SU2 using viscous flow, please see the case mentioned above that was presented in this recent paper

http://su2.stanford.edu/documents/SU...ciTech2014.pdf

Hope this helps,
Tom