View Issue Details

IDProjectCategoryView StatusLast Update
0003788OpenFOAMFeaturepublic2022-01-24 09:58
Reporterwyldckat Assigned Tohenry  
PrioritynormalSeverityminorReproducibilitysometimes
Status closedResolutionsuspended 
Product Versiondev 
Summary0003788: Finally a case that demonstrates the recent option extendRefinementSpan
DescriptionFollowing up on issue #3361 and commit 5d93da3aed0d03b38081717e29cfd506eba294a9: https://github.com/OpenFOAM/OpenFOAM-dev/commit/5d93da3aed0d03b38081717e29cfd506eba294a9

Attached is the image "extendRefinementSpan.png" that demonstrates how the castellatedMeshControls:extendedRefinementSpan option can affect the deployment of refinement levels on surfaces/shells...

About the provided case(s):
- These use an excerpt of a larger geometry, which is why the surface is not entirely closed... but not being closed does not cause this issue.

- The larger cells are level 0.

- The openings/reveals on the left and top have the same refinement level 3 defined in castellatedMeshControls::refinementSurfaces.

- Said reveals have nearly the same geometrical dimensions.
   - Although the one on the left is split in two, namely the surface we see on the top is split along X, which we can spot a very small snapping error there.

- Both cases use a level 1 refinement box, but only on the top reveal, not on the reveal on the left.

So what's sort-of heuristically weird in this?
- The top reveal should have been refined without any problems, without the need for a refinement box, nor need the flag extendRefinementSpan.
- At least based on what I know about how the snappyHexMesh algorithm 'sees' the geometries to refine and mesh them.

Now, the flag extendRefinementSpan is nice enough to help use get the top reveal to be visible, but the level 1 refinement leak to the right of the top reveal should not have happened.
Steps To ReproduceAttached are two cases, with just the flag set differently:
- case_with_extendedRefinementSpan.tar.gz
- case_without_extendedRefinementSpan.tar.gz

Running and clean up is the same as OpenFOAM's tutorials.
TagsNo tags attached.

Activities

wyldckat

2022-01-21 16:58

updater  

extendRefinementSpan.png (20,858 bytes)   
extendRefinementSpan.png (20,858 bytes)   

henry

2022-01-21 17:46

manager   ~0012421

It is not clear what your are proposing, should the extendRefinementSpan option be removed?

wyldckat

2022-01-21 18:09

updater   ~0012422

My apologies for not making this clear right away, but the base idea here is to provide a better documentation of this feature, which is technically incomplete.

Therefore, the option should stay and will be useful in these kinds of specific situations.

That said, even though this feature is useful in said situations, it's flawed in the way it was implemented. And given that it was not clear why this feature was needed in the first place, when I managed to track down this near-perfect example case, I had to share it.

To me, I'm fine with this issue being closed as-is, namely identifying what this feature was for in the first place, specially when it was originally not optional.

henry

2022-01-21 18:17

manager   ~0012423

It is not clear what needs to be done, are you planning to work on it or leaving it for someone in the community to work on it?

wyldckat

2022-01-21 18:30

updater   ~0012424

I'm planning to work on it, but I have no idea when I'll be able to do so. And it will be done at work, not on my free time.

I'm still looking a bit more into this and am planning to share my findings on this, as well as then solution when I come to it.

So far, I've only managed to deduce that:
- If the refinement level was shared from the shells to the feature edges, it would then allow the feature edges increase the visibility on the reveal at the top.

Either way, if you prefer, this issue can be closed as suspended. The primary objective of this report was to try and contribute to this feature, even though it's not yet the solution.

henry

2022-01-24 09:58

manager   ~0012427

Pending contribution.

Issue History

Date Modified Username Field Change
2022-01-21 16:58 wyldckat New Issue
2022-01-21 16:58 wyldckat File Added: extendRefinementSpan.png
2022-01-21 16:58 wyldckat File Added: case_with_extendedRefinementSpan.tar.gz
2022-01-21 16:58 wyldckat File Added: case_without_extendedRefinementSpan.tar.gz
2022-01-21 17:46 henry Note Added: 0012421
2022-01-21 18:09 wyldckat Note Added: 0012422
2022-01-21 18:17 henry Note Added: 0012423
2022-01-21 18:30 wyldckat Note Added: 0012424
2022-01-24 09:58 henry Assigned To => henry
2022-01-24 09:58 henry Status new => closed
2022-01-24 09:58 henry Resolution open => suspended
2022-01-24 09:58 henry Note Added: 0012427