Ask an engineer how long a CFD study takes and they’ll usually quote the solve. Ask how long the project took, and a different number appears — one dominated by a step that rarely makes it onto the schedule: building the mesh.
The work that hides in “meshing”
Meshing is not one task. It’s geometry repair, feature defeaturing, boundary-layer setup, quality checking, and — when a case won’t converge — going back to do it all again. Most of that is manual, and most of it is invisible to a project plan that has a single line item called “mesh.”
The near-wall region is where it concentrates. Resolving the boundary layer means hitting a first-cell-height target, keeping a gentle growth ratio, and stacking enough layers to span the profile — all while the layers try to collapse in every concave corner and tight gap. It is fiddly, it is fragile, and it is exactly the part that decides whether your drag and heat-transfer numbers mean anything.
Why it resists automation
Traditional meshers put the burden on the operator. Two engineers, same geometry, different settings — different meshes. That operator dependence is why meshing is hard to schedule and hard to reproduce: the result lives partly in someone’s head.
It also makes meshing expensive in a second way. Every design iteration pays the tax again. In workflows where compute is capped — restricted-CFD motorsport, or large ML training sweeps — the hours lost to setup are hours you can’t spend on the runs that matter.
A different starting point
The way out isn’t a faster manual mesher — it’s a mesh that conforms to the geometry by construction and comes out the same every time. That’s what a conforming Voronoi approach makes possible: polyhedral cells that honor faults and features as hard constraints, near-wall layers grown as part of the mesh, and a deterministic result you can put in a report.
Budget the mesh honestly and it’s often the biggest line on the project. Automate it well and it stops being a line at all.
If meshing is the step that keeps eating your schedule, we’d like to show you a different one. Book a technical walkthrough.