Poor quality of mesh after deformation (SU2-5.0.0)
 

I used SU2_DEF to evaluate CRM wing grid deformation, but the quality of my structured mesh is pretty bad. My question is whether the FFD box settings could affect the grid deformation,for example the distance of FFD box from the wall. Another question is whether the strctured mesh doesn't fit SU2_DEF?
Could anyone help with this ? thank you!
Here are my FFD boxes, configuration file and slice of my wing (red color is deformed grid, the black one is original grid)

Thanks for your question

There are some deformation settings that can be altered to get a smoother deformation, see the GRID DEFORMATION PARAMETERS section of the config_template.cfg file provided with the code.

You may also want to check whether the FFD box boundary intersects the geometry at any point (which would freeze a plane of the box).

Thanks for your suggestions. I have already checked the FFD box and it doesn't intersect into the geometry. And i also tried different grid deformation settings, but none of them could get a nice grid. I noticed that all of the grid in the testcases of SU2 are unstructred, is there any chance that the grid deformation method (for example spring method in SU2) doesn't fit structred mesh?

Please check that the deformation is converging, by using DEFORM_CONSOLE_OUTPUT=YES to check how many iterations are being run and whether it exits due to meeting a maximum iteration or a tolerance.

The grid deformation method uses the spring analogy, which does mean that shapes that would be unstable in a physical structure (ie, a square vs a triangle) suffer from the same problems in the mesh deformation.
One thing to try is using the stiffness parameters, ie using wall distance which will better preserve the shape of the mesh near the wall, or increasing the constant stiffness (with DEFORM_ELASTICITY_MODULUS, which defaults to 2E11)

Thanks for your suggestions. I have handled this problem by setting DEFORM_LINEAR_SOLVER_PREC= JACOBI, it was originaly ILU_SGS method. I don't know exactly why but it works well. The deformation is fully converged with residual equals 1.2096e-13 and the final grid is also smooth.

Thank you again

prisms vs tets
 

Does this mean that mixed meshes with prisms in boundary layers will be unstable during deformation compared to pure tetrahedral mesh?