WIP - Search translated wall images when computing periodic wall distance

[rois1995]

Proposed Changes

Periodic boundaries represent a domain that repeats. Near a periodic boundary, the closest wall can belong to the next repeated copy, rather than the copy stored in the mesh. The current wall-distance search can miss that wall and return a distance that is too large.

For example, imagine a channel with a bump on the floor, repeated end to end. A point near the end of one section may be closer to the bump in the next section. This PR includes those translated copies in the nearest-wall search. It also preserves the identity of the wall that was found.

The testcase makes the error visible at two points representing the same periodic location: develop returns 0.5 on one face, although the correct distance to the repeated sloping wall is about 0.265. With the fix, both faces return 0.5/sqrt(1+1.6^2) = 0.26499947. The test also checks the same query shifted into the next repeated section. Turbulence models use wall distance, so this mismatch can affect their predictions; a complete turbulence-flow comparison is still pending.

The implementation extends the shared nearest-wall search used by FV and FEM. It does not change the mesh or add physical walls.

Draft: currently covers repetition by translation, with independent repeat directions. Rotating sectors, helical repetition, and redundant repeat directions remain unfinished and retain the existing search behavior. Skew/multiple-generator, FEM/multizone, OpenMP, nearest-wall identity and full turbulence-regression validation remain pending. No claim that all periodic wall-distance cases are fixed.

Standalone H build and analytic-distance testcase pass in serial and partitioned MPI2 (4 assertions/rank); broader checks remain pending.

Related Work

• MPI Point-to-Point Refactoring + New Periodic BC Implementation #652 introduced the current periodic implementation.
• Fixes for rotational periodic boundaries: implicit solver, multigrid and limiters #2961 lists the eight related periodic-boundary PRs. Their numbering groups the topics; it does not imply a linear dependency chain.

PR Checklist

• I am submitting my contribution to the develop branch.
• My contribution generates no new compiler warnings (complete branch CI pending).
• My contribution is commented and consistent with SU2 style.
• I ran the repository pre-commit checks on the changed files.
• I have added tests that demonstrate the contribution.
• I have updated appropriate documentation, if necessary.

[rois1995]

Test cases

Reproducers, configurations, numeric logs and scripts: wall_distance. Develop = 6db10127d1; BOX fixtures generate their meshes. The original B–D evidence above was recorded before the expanded follow-ups.

Check	develop	translated-image draft
RANS BOX lower-wall hill, left periodic query	0.5	0.26499947; standalone serial/MPI2 pass
Matched right/shifted query	0.26499947	expected same analytic distance

The develop test executes 3 assertions with 1 failure. The new regression adds an independent analytic-distance check. Rotational/helical/dependent-generator support and broader turbulence/FEM/thread/identity validation remain open.

Standalone H build succeeds; the analytic-distance testcase passes all 4 assertions in serial and on each of two genuinely partitioned MPI ranks. No full turbulence/FEM/OpenMP pass is claimed.

[joshkellyjak]

Is this WIP or ready to be reviewed?

[rois1995]

@joshkellyjak Still WIP. The translated-image fix has an analytic serial/MPI check, but broader validation and the remaining periodic cases are unfinished. I'll keep it as a draft until it's ready for review.

[bigfooted]

@rois1995 please mark WIP in the title and pr:draft as label when it is not ready and you only want to expose your work to the general public.

[rois1995]

@bigfooted Yeah, I forgot about this one, thanks!