libCEED
libCEED copied to clipboard
WIP: Fluids - Automate mesh generation for boundary layer
Hi @amandashack :wave: I hope you're enjoying break. Can you please let us know if you plan to get back to this, and if so, on what timeline?
@jedbrown Note that now you can do that with a DMPlexTransform. You mark the cells you want to split with a label, and you can refine them normal to the boundary, or you can extrude. I have several tests of this now.
@knepley Yeah, this branch is intended to create a mesh with element size versus fraction of the computational domain that looks similar to the figure below, but with the transition points (from viscous sublayer to turbulent BL to far-field) and aspect ratios controllable in intuitive ways using plus-units. This would be a nice feature to upstream to PETSc once it's shown to be useful here.

Hi @jedbrown yes I plan to work on this over break. My goal is to have it completed by the time classes start back up!
:wave: Hi, @amandashack. Just checking in on this, if you're still planning to finish and if you have what you need to do so? Completing this would make it easier to conduct accuracy-efficiency experiments for turbulent flow like you see in this clip.
Hi @jedbrown, Leila is on vacation so we are meeting next week to work on this. I have shared with her some notes on where I am at. I think after we meet on Monday or Tuesday I will reach out to you to let you know where we're at, ask any questions we may have, and give you a timeline for finishing.
Sounds good. The high order solvers are much improved now so this design space will get interesting.
Closing in preparation for the HONEE move, but re-open (here or there) if we want to revive this (its been more than a year)