2D mesh adaptation to solution - poor quality mesh

Hi all,

I am investigating a backward facing step periodic in the z direction. To test the possibilities of MMG3D, I performed a coarse computation and extracted the wall distance from Paraview, then I tried to refine the mesh accordingly to the wall distance using .sol file. I manage to obtain a mesh but its quality is very poor. May you help me to improve it? I tried to play with the -hgrad and -hausd parameters but I cannot manage to produce a nice mesh (see attached pictures). Why am I obtaining very heterogeneous cells size at different streamwise locations for the same wall distance?

Here are the command line: mmg2d_O3 channel_with_sudden_expansion_entree_perio_2D.mesh -sol wall_dist_2Dfrom3D.sol -hausd 0.01 -hgrad 2. -hmax 0.2

I attach my .mesh and .sol files.wall_dist_2Dfrom3D.sol (45.6 KB)
channel_with_sudden_expansion_entree_perio_2D.mesh (641.7 KB)

Thank you very much in advance!