[Haolin Zhi]

Hi everyone,

I've been currently getting familiar with SU2 (version 7.3.1) according to the tutorial cases. I built up the solver from source code and there is not problem with the running of SU2.

Now the question is, I found the tutorial cases about unsteady flow (e.g., NACA0012 in compressible flow case and in shape optimization case) were written for version 7.0.1. I wonder whether there is big difference between version 7.0.1 and 7.3.1 (but they are both version 7....).

For example, the unsteady case of NACA0012, the time step presented on the website is 5e-3, while it is 5e-4 in the downloaded config file. I tried both the two time steps, and I cannot get the same aerodynamic history as presented on the website (the drag curve I derived is smooth and fluctuates near 0.22, while the website shows it is obviously periodical and fluctuates from 0.31 to 0.39). I didn't change any features in .cfg file except the time step.

For the case of unsteady discrete-adjoint optimization of NACA0012, there is the same difference of time step between the website and the .cfg file. When the SU2 runs this case, it always reports the error like "linear system orthogonality fails". I know the FGMRES diverges, but I didn't change the .cfg file.

So, I wonder, are the tutorial cases, especially for the unsteady ones, still usable for the SU2 version 7.3.1?

I really appreciate any help you could give me.

Kind regards,
Zhi

[pcg]

I think someone reported some difficulties with that same testcase not developing unsteady behavior some time ago. I'll have a look.

[pcg]

Make the following changes and the case will converge better:
CFL_NUMBER= 1e12
LINEAR_SOLVER_ERROR= 0.1
LINEAR_SOLVER_ITER= 10

I see the drag and lift oscillating as shown on the website.

[Haolin Zhi]

Originally Posted by pcg

Make the following changes and the case will converge better:
CFL_NUMBER= 1e12
LINEAR_SOLVER_ERROR= 0.1
LINEAR_SOLVER_ITER= 10

I see the drag and lift oscillating as shown on the website.

Hi Pedro,

Thank you for your reply. I tried what you said and it works! Also, I find increasing the internal iterations also helps.

So it seems that the default settings of INNER_ITER and CFL_NUMBER in the config file cannot produce a nice convergence during each time iteration, therefore the unsteady behavior cannot be developed.

Thanks again for your kind help!

Regards,
Zhi