What is it about?
Two anisotropic output-based adaptation paths on an ONERA M6 wing. Path one: a Pointwise viscous baseline, then FUN3D viscous adjoint adaptation. Path two: the same surface mesh, a cheap inviscid Cart3D adjoint adapt, transfer the adapted spacing back into Pointwise as point sources, rebuild the viscous surface and volume, then FUN3D-adapt from that pre-adapted mesh. After ten cycles the pre-adapted path converges L/D to about 16.04 vs 15.01 for the baseline path, matches shock-region Cp better, and cuts L/D discretization error by 7.4% after only the first flow and adjoint solves relative to a full ten-cycle baseline.
Featured Image
Why is it important?
Viscous output-based adaptation is expensive, and most methods cannot adapt the wall surface because the CAD is already gone. Using a cheap inviscid adapt to place the viscous mesh where the gradients are lets you adapt both surface and boundary layer and spend fewer viscous adjoint cycles to get a better L/D.
Perspectives
Pointwise-era paper with Travis, Nick, and Matt McMullen at Aerion. It is a methods paper on M6, not an S-duct or PAW paper, even though the adaptation theme overlaps 2017-4835.
Zach Davis
Space Dynamics Laboratory
Read the Original
This page is a summary of: Reducing the Computational Cost of Viscous Mesh Adaptation, June 2017, American Institute of Aeronautics and Astronautics (AIAA),
DOI: 10.2514/6.2017-3109.
You can read the full text:
Contributors
The following have contributed to this page
