Meshing is often the most time-consuming part of a CFD project. For complex hydraulic structures, generating a high-quality mesh can easily take longer than running the simulation itself (and generate plenty of headaches along the way!).
I’m experimenting with one of the alternative approaches: embedded-boundaries (EB). Take the raw STL, drop it into a Cartesian grid, and let the solver cut the cells. No snappyHexMesh dictionaries. No spending days tweaking mesh quality or wondering how many cells you’ll end up with after the meshing finally finishes.
The video shows a weir going from a raw STL to a running free-surface simulation—without creating a body-fitted mesh or touching snappyHexMesh.
There are still trade-offs, especially regarding near-wall accuracy, but I think there are some promising ways to address them.
It’s an exciting direction that could significantly simplify our hydraulic CFD workflows; after all, this is similar to what some of those big-brand 😉 CFD packages do under the hood.
Has anyone else experimented with embedded boundaries for free-surface CFD? What do you see as the biggest remaining challenges?