LETE constraint

[benoitpaillard]

Hi all,

pygeo leading edge/trailing constraint is broken, it freezes the points rather than mirror them. It can be fixed easily by adapting v2 scripts.

I can submit a PR for that, is that your preferred way of contribution ?

regards

[friedenhe]

Thanks for the feedback. Before sending us a PR, can you upload a simple case for us to reproduce the issue?

[benoitpaillard]

When running NACA0012 incompressible tutorial, the first 4 and last 4 dvs.shape values remain equal to zero.

I reckon this is due to this

adapting the code from v2 script solves this issue.

[friedenhe]

I managed to reproduce the LE/TE constraint behavior you mentioned for the NACA0012/incompressible case. But I believe this is case-dependent. If you run NACA0012/subsonic, the LE/TE point displacements will not be zeros. Note that NACA0012/incompressible and NACA0012/subsonic have the same LE/TE constraints setup. Therefore, it is the optimizer that decides to have 0 for the LE/TE points in the incompressible case, but the LE/TE points can be non-zeros for other cases.

[benoitpaillard]

When running NACA0012/subsonic tutorial, the first 4 and last 4 digits of dvs.shape remain equal to zero, see log attached.
logOpt.txt

[benoitpaillard]

On top of that, it is possible to confirm that this behavior is an issue by running NACA0012/incompressible case at zero lift, with a very low thickness constraints. The result should be a flat plate, whereas dafoam keeps some thickness at leading edge. See custom runScript.py attached.
runScript.zip

[friedenhe]

, whereas dafoam keeps some thickness at leading edge.

Please post some figures for this.