idaholab
Add IC initialization for newly-activated nodes in `ElementSubdomainModifier`
close #30627
Add IC initialization for newly-activated nodes in ElementSubdomainModifier
This PR is added upon #30300
Reason
The ElementSubdomainModifier dynamically activates elements in the domain by changing their subdomain IDs in the case of Directed Energy Deposition (DED) or manufactured process. We would use ElementSubdomainModifier to turn on solving on some materials to mimic the deposition process. However, when a new set of elements becomes active from non-solved state, the nodes shared exclusively by either newly-activated or non-solved elements (i.e., elements not yet part of the solution) are called newly-activated nodes. These nodes do not carry prior solution history and must be assigned appropriate initial conditions before being incorporated into the simulation.
Simply assigning zero values to these nodes (e.g., zero temperature or displacement) often causes numerical difficulties, especially when using nonlinear material models like those provided by the neml2 library. Poor convergence or unstable behavior can result.
Design
Add initial condition strategies for newly-activated nodes/elements during subdomain activation
We use Polynomial patch recovery to assign physically meaningful initial conditions:
- Samples solution values from nearby solved elements.
- Performs least-squares fitting (e.g., using a quadratic basis).
- Evaluates the fitted polynomial at the coordinates of newly-activated nodes or the centroids of newly-activated elements.
Impact
Improves stability and convergence in additive manufacturing simulations by providing physically reasonable initial guesses on newly-activated nodes.
@roystgnr when you get a chance, could you take a look at this PR (mostly at the concept level)? I think the PR description is fairly thorough, but basically we are trying to construct a polynomial over a patch of elements, then use the constructed polynomial to perform extrapolation. User will have control over the polynomial order.
This should be a better approach than our previous attempt which simply finds the nearest dof. I think this approach works for any function spaces.
@lindsayad sorry this was confusing -- this PR is based on the other PR for default block restriction. So I guess you meant to leave the above comments there. I've addressed your comments with this commit https://github.com/idaholab/moose/pull/30300/commits/ebfec5e1a47351b98df595d86cf377780ebb66dd. The commit right after that removes those deprecated parameters from input files.
Job Documentation, step Docs: sync website on ba72242 wanted to post the following:
View the site here
This comment will be updated on new commits.
Job Precheck, step Clang format on d16c82c wanted to post the following:
Your code requires style changes.
A patch was auto generated and copied here
You can directly apply the patch by running, in the top level of your repository:
curl -s https://mooseframework.inl.gov/docs/PRs/30630/clang_format/style.patch | git apply -v
Alternatively, with your repository up to date and in the top level of your repository:
git clang-format 3459da3c998c2fa3175a986bfb9edbaa3bf566b8
@GiudGiud @lindsayad or @roystgnr can we get a review on the framework level changes?
The main beef is in ElementSubdomainModifierBase, and I've already reviewed that part fairly thoroughly. The rest of the changes in framework are pretty light.
The PR has adapted quite a lot since its creation, so I took the liberty of rewriting the PR description. @ChengHauYang see if what I wrote makes sense and reflects what you have now.
The documentation failure is because NodalPatchRecoveryVariable is missing class description. You probably forgot to do that in the validParams method.
Job Coverage, step Generate coverage on ba72242 wanted to post the following:
Framework coverage
| 3d7413 | #30630 ba7224 | ||||
|---|---|---|---|---|---|
| Total | Total | +/- | New | ||
| Rate | 85.64% | 85.64% | +0.01% | 87.83% | |
| Hits | 116895 | 117254 | +359 | 433 | |
| Misses | 19601 | 19653 | +52 | 60 | |
Modules coverage
Coverage did not change
Full coverage reports
Reports
-
framework -
chemical_reactions -
combined -
contact -
electromagnetics -
external_petsc_solver -
fluid_properties -
fsi -
functional_expansion_tools -
geochemistry -
heat_transfer -
level_set -
misc -
navier_stokes -
optimization -
peridynamics -
phase_field -
porous_flow -
ray_tracing -
rdg -
reactor -
richards -
scalar_transport -
solid_mechanics -
solid_properties -
stochastic_tools -
subchannel -
thermal_hydraulics -
xfem
Warnings
-
frameworknew line coverage rate 87.83% is less than the suggested 90.0%
This comment will be updated on new commits.
The last commit https://github.com/idaholab/moose/pull/30630/commits/62f5ce1e85a2fbc93168b641d2a2d3b0c71da35a is a follow up of #31306 and should be able to close #31305
Sorry, I thought I turned on all of the parallel sweeps, but perhaps I forgot the debug test.