2D Shocks.
Anisotropic meshes are extremely powerfull to accurately capture
shocks. Here is an example where this fact is emphased.
We used our
Discontinuous Galerkin code and adapted the mesh
120 times to reach 1.2 seconds of computation.
Transient anisotropic mesh refinement. Meshes (left) and density
contours (right) after 0.4 (top) and 1.2 (bottom) seconds.
Next pictures show closups of meshes at iteration 120. We
see obviously the good behavior of mesh adaptation at the intersection
of shocks.
Mesh closups.
3D Aniso.
Next picture show a 3D anisotropic mesh that was constructed using
an analytical metric field.
Anisotropic 3D mesh.
Analytic metrics are of course not the final aim of all that.
We then consider the classical problem of a planar shock moving at Mach 10
and hitting a wedge (the double mach reflexion).
For this simulation, we have chosen a wedge of
final height so that the problem is 3D.
This is our first attempt to produce anisotropicly refined
meshes based on flow solutions.
Next pictures show meshes and density profiles at t = 0.3.
Refined 3D meshes as well as a (not so) nice picture
of density isosurface. We still have to work on transparency
on Gmsh.
More infos about anisotropic mesh refinement can be found on
this paper