Why the near-wall region is special
Almost everything engineers care about — drag, heat transfer, separation, transition — is set in a thin layer next to the wall where velocity climbs from zero to the freestream. Resolving it demands cells that are extremely thin normal to the wall but long along it: high-aspect-ratio, anisotropic cells. That is exactly what prism (or inflation) layers provide — stacked, wall-parallel layers that capture steep normal gradients without wasting cells in the tangential direction.
Why it is the hardest part
Ask most CFD engineers where a mesh job stalls and they point at the boundary layer. A few reasons:
- Conflicting constraints: you need a specific first cell height (see y⁺), a gentle growth ratio, and enough layers to span the boundary layer — all at once.
- Geometry sensitivity: prism generation is a local marching process that is fragile wherever the surface is not smooth.
- Iteration: it usually takes several remesh-and-check cycles to get y⁺ and layer quality right.
Layer collapse and where it happens
The classic failure is layer collapse or self-intersection. As prisms march inward, neighboring fronts converge in concave regions, sharp corners, narrow gaps, and trailing edges. When opposing fronts meet, layers pinch, invert, or get squeezed out entirely, leaving gaps the mesher backfills with poor-quality tets or, worse, invalid cells.
The common mitigations — reducing layer count locally, shrinking total thickness, or stepping the layers back — all trade away near-wall resolution in exactly the spots where the flow is most complex.
How AutoMesh-Geo helps
Because the boundary layer is where meshes fail and time is lost, robust near-wall handling matters most here. AutoMesh-Geo builds feature-conforming cells that follow concave corners and tight features instead of colliding in them, keeping near-wall resolution intact through the regions that usually force manual repair.