Supersonic flow test, forwards facing step

General
1

Dear all,

I am now testing the supersonic forward step test recently. I also find another guy once trying to do that case, while I follow the suggestions and tried to reproduce the results. I copied the .ini file from Jin Seok https://groups.google.com/d/msg/pyfrmailinglist/bwicPvZb1B8/Z07K4OCjCgAJ, but changed the mesh file as attached. All the conditions are the same.

  1. Test the corase mesh, each grid size is 0.0125, uniformly distributed.The results are like this.

I think this is not correct.
Then I test the finer mesh, each grid size is 0.00625, uniformly distributed. But it gives me minimum sized time step rejected even though timestep is reduced to 1e-10 if “controller” is “pi”, and RuntimeError: NaNs detected at t = 0.009999999999999995 when “controller” is “none”. Peter said this may be due to the doamin is under-resolvable https://groups.google.com/d/msg/pyfrmailinglist/wVJ4pj0unxQ/Jj5taiZyAQAJ. And the suggestion is increase the mesh resolution. I am quite confused now.

  1. When Mach number increases to 3, both coarse and fine meshes would fail.

Could anyone give me a hint to solve this?
Besides, could anyone tell me if the parameters in “solver-artificial-viscosity” or parameters in “solver-interfaces” could affect results very much? If they do, could anyone tell me how to change those parameters correspondingly?

Thank you all in advance.

Regards,
Sam

forward_step_finner.msh (5.09 MB)

forward_step_coarse.msh (1.21 MB)

forward_step.ini (1020 Bytes)

2

Hi Sam,

I think you need to tune artificial viscosity coefficients which relies on solution and mesh size.

From the Persson’s paper (AIAA 2006-0112),

max-artvisc ~ h/p, s0~1/p^4 and kappa is chosen empirically sufficiently large so as to obtain a sharp but smooth shock profile.

I think you may adjust max-artvisc if you refine your mesh.

As I mentioned before, forward facing step problem may be tough to solve with artificial viscosity (https://groups.google.com/d/msg/pyfrmailinglist/bwicPvZb1B8/Z07K4OCjCgAJ). Please try a problem with oblique shock or mild normal shock first.

Regards,

Jin Seok