[jentink]

I've been running several test cases of 2D blunt leading edges. Starting with a fairly coarse grid, I'm trying to get a resolved leading edge shock.

The GRAD_FLOW option does something, but doesn't appear to be picking up the shock very well. And the SUPERSONIC_SHOCK option doesn't do anything.

Are there factors in the code I can tweak to get better shock detection?

Thanks.

[jentink]

Another thing I'd like to be able to do is limit adaption to certain areas. IE, not the structured layer near the surface of the leading edge.

[copeland]

Hey Tom,

The GRAD_FLOW option for SU2_MAC refines the mesh at locations where the flow density gradient is highest. The total number of cells added to the refined grid is controllable using the NEW_ELEMS tag in the configuration file. As the mesh is refined, there are many cases where shocks may change location -- this frequently happens along stagnation lines for supersonic flows over blunt bodies or airfoils. 2nd order spatial discretization schemes will help mitigate this, but be aware that the refinement can "chase" the bow shock around.

To get good shock capturing, it will require playing with the number of mesh adaptation cycles you perform and the number of elements that you add at each adaptation cycle. My recommendation, especially if you are starting from a very coarse grid, is to perform one or two adaptation cycles with 30-50% new elements at each cycle -- then reduce the number to ~10% and perform several more cycles. This is often problem dependent, so try some different approaches and please share your findings with the community.


Regards,
Sean

[jentink]

Is the supersonic_shock option not implemented?

Also, w/out looking at the source code, and not knowing a great deal about this algorithm, how hard would it be to prevent adaptation in rectangular cells?

[copeland]

If you open grid_adaptation_structure and look for SetSupShock_Refinement, you can see exactly what the SUPERSONIC_SHOCK option does. The adaptation method refines within the Mach cone of the freestream Mach number set in the configuration file. That particular method was implemented for a specific problem the development team has been working on and some values have been hard-coded. If you want to use it, try playing with some of those values to get it to fit your problem of interest.

[jwang1_09]

Hello jentink, Can you tell me how use the mesh adaptation of su2? What is the command about implementing SU2_CFD and SU2_MAC (SU2_MSH) simultaneously？ Thank you very much.

[akun646]

Hi there!

I have also been running some cases of hypersonic sphere and cylinder, and trying to do some adaptations either by GRAD_FLOW or by WAKE option. For 2D models, the adaptations were done without any problems, but it didn't work in the 3D meshes for me. Has anyone ever tried adaptation on 3D mesh? By the way, I have structured grid. Thanks a lot!

Facundo

[jwang1_09]

I have the same problem.