Fix NS-VOF stong residual formulation

hepap commented 3 months ago

Description

The term associated with the viscosity jump in the strong residual for the NS-VOF assembler added in PR https://github.com/chaos-polymtl/lethe/pull/1149 led to a ill-posed formulation in phase change cases where the solid is represented by a highly viscous fluid. Hence, we don't consider this term in the current formulation.
There was no appropriate application test for lpbf/phase change with vof cases that could have allowed us to identify this ill-posed formulation prior to the merge of PR https://github.com/chaos-polymtl/lethe/pull/1149.
The LPBF benchmark example prm files were almost up to date, but did not exactly matched the one we used to generate the benchmark results. The mesh files were missing from the repo.

Solution

This PR removes the term associated with the viscosity jump.
A new test is now included to avoid future mistake in lpbf/phase change with vof cases.
LPBF benchmark example prm files were modified and are now up to date with the current working version. The mesh files are now available in the repo.

Testing

The new test simulate the lpbf benchmark case using the extra-extra-coarse mesh. Temperature statistics and mass conservation monitoring are the outputted metrics (to track both the change in HT and NS-VOF solvers).
Other NS-VOF tests were updated, and results for those tests returned to their previous values (or almost with small differences).

Documentation

No modification in the documentation.

Miscellaneous (will be removed when merged)

I guess the tests will fail because of the CI issue. I have still opened the PR since the formulation won't change and I will update the restart of the new test when I have updated my P4est version :)

Checklist (will be removed when merged)

Code related list:

[x] All in-code documentation related to this PR is up to date (Doxygen format)
[x] Lethe documentation is up to date
[x] Fix has unit test(s) (preferred) or application test(s)
[x] The branch is rebased onto master
[x] Changelog (CHANGELOG.md) is up to date
[x] Code is indented with indent-all and .prm files (examples and tests) with prm-indent

Pull request related list:

[x] Labels are applied
[x] There are at least 2 reviewers (or 1 if small feature) excluding the responsible for the merge
[x] If this PR closes an issue or is related to a project, it is linked in the "Projects" or "Development" section
[x] If the fix is temporary, an issue is opened
[ ] The PR description is cleaned and ready for merge

blaisb commented 3 months ago

The debug test still seems to crash?

AmishgaAlphonius commented 3 months ago

The debug test still seems to crash?

It's a bit weird, the load from a checkpoint function for the triangulation has an assertion that checks if all cells are at the same level of refinement. In Release there is no issue, but in Debug, it crashes because of this. I tried it locally also and I had the same issue. Is it that assertions are only checked on Debug?

blaisb commented 3 months ago

The debug test still seems to crash?

It's a bit weird, the load from a checkpoint function for the triangulation has an assertion that checks if all cells are at the same level of refinement. In Release there is no issue, but in Debug, it crashes because of this. I tried it locally also and I had the same issue. Is it that assertions are only checked on Debug?

You need to make sure the restart files does NOT refine the triangulation. Generally this is the case for all the solvers (we do not do mesh refinement when we are reading a restart file), however we don't have any restart tests for VOF I guess,so maybe this issue was there all along. You can check a restart with any VOF cse in the debug to see if that's the issue. But otherwise, the triangulation/ mesh must have the same properties as the one used in the restart (aka the coarse grid has to be the same)

Assertions are not run in DEBUG mode, so that it works in release does not mean anything.

hepap commented 3 months ago

I will fix the restart with the box refinement feature on another branch, and I will come back to this PR so that it stays a coherent PR.

hepap commented 3 months ago

So I removed the box refinement for the new test in this PR, and I fixed the restart problem with box ref in PR #1184.

This PR (#1180) can be merge before PR #1184 (I added a test in PR #1184 to verify restart with box ref).

blaisb commented 3 months ago

@hepap can you rebase? Then this will be ready to go :)

chaos-polymtl / lethe