Diverging solution for hypersonic flow
 

Hello everyone,

I am new to CFD and I have been trying to simulate 2D viscous hypersonic flow in an external compression scramjet inlet.
I am using version 6 of SU2 and gmsh for meshing.
After multiple trials of mesh adjustment and changes in the configuration file(detailed ahead) also, my solution is diverging.

I have used the JST scheme. For the same problem, inviscid flow solution converged but viscous is not converging.

Due to the classified nature of the work, I cannot share the mesh or the configuration file here but I am attaching a small part of the mesh which I feel could create problem, if any, with meshing.

I have attached 2 (parts of )mesh files...first one is coarse and does not take care of sudden volume changes in element size and boundary layer properly. Second one is fine and boundary layer element size calculated with y+ value in mind.
In both these meshes, the upper part is wall and side ones are shocks. In order to make a shock aligned grid and also to take care of the boundary layer, such a mesh is made. The unstructured part could be causing problems but that's the best I could do to resolve the inner triangle. Skewed elements could be a big issue too but I don't understand how to solve that problem and still have a shock aligned grid.

Trials with cfg file (ultimately it diverged) :

1. Gradual reduction in CFL number - 1,0.5,0.25,0.1,0.05
2. Artificial dissipation increased to 10X, 100X in JST scheme yet solution diverged, although after more reduction in residue initially.
3. In all other schemes, diverged faster than in JST.
4. First order solution for 10k iterations and then second order for next 10k. Up till now residue seemed to come down. But after these 25k iterations when the flow was restarted with 2nd order limiter(this part was done in version 3 of SU2), it started to diverge. This was tried with different CFL numbers too.
5. Strangely, even though second mesh is supposed to be better, the convergence seemed to worsen in it. In the first mesh, residue reduces about 1.5 orders before diverging but in the second mesh, it first oscillates and then starts to diverge.
6. I have tried running it for lower Mach numbers like 2 but it still diverged.

I am not understanding how to solve this problem. Any help would be greatly appreciated. :)

Try a simpler grid [emoji4]