[delsolman]

Hello, I am trying to mesh a channel by snappyHexMesh for LES without wall function.
To capture small eddies near to the wall, which is crucial to prevent the velocity overshoot in velocity profile (caused by superstreaks) I am refining the mesh in near wall region in all three directions.
That's the idea... But when I go with delta_x+, delta_y+ and delta_z+ to 1, then I go crazy with total # of cells...
1.) So I am happy with corser cells but I WANT TO SPLIT the first cell layer in normal direction...
How can I do this? To remain whole mesh and just split the first layer? My idea was, to define boundary layer of two layers of total hight same als the first cell layer. But I have this feeling that snappy shift some cell layers over the boundary layer when I create them.
You can see that behaviour on my pictures. (oh...on the picture there is the boundary layer made by two layers with expansionRation 2 - this was my best solution to keep the cell hight in first two layers)

2.) How can I prevent to colapse the boundary layer in the corner please? I set featureAngle 180 but I am not happy :-)

[delsolman]

...could I have a additional question: is it possible to generate a boundary layer higher than one cell layer please? Let's say two times higher with 6 layers?
So higher than this one on the picture...

[delsolman]

I made some progress...
the key is to switch from medial-axis algorithm (default) to meshShrinker:

// maxThicknessToMedialRatio 0.3; //medial-axis
// minMedialAxisAngle 90; //medial-axis

meshShrinker displacementMotionSolver;
solver displacementLaplacian;
displacementLaplacianCoeffs
{
diffusivity quadratic inverseDistance (wall1 wall2 etc);
}
additional:
- in fvSolution:

solvers
{
cellDisplacement
{
solver GAMG;
smoother GaussSeidel;
tolerance 1e-7;
relTol 0.01;
}
}

- in fvSchemes:

gradSchemes
{
default Gauss linear;
}
laplacianSchemes
{
laplacian(diffusivity,cellDisplacement) Gauss linear corrected;
}

But could someone help me please how can I prevent collapsing of the boundary layer in the corner? I would really appreciate that... :-)

[delsolman]

see picture...