Apply decusping to individual assemblies.

[mgjarrett]

Description

---

Grouping assemblies by type does not always work because assemblies of a single type can vary significantly in height over their lifetime in a reactor.

The particular example that motivates this bug fix is axial expansion caused by irradiation-driven swelling.

Checklist

• This PR has only one purpose or idea.
• Tests have been added/updated to verify that the new/changed code works.

• The code style follows good practices.
• The commit message(s) follow good practices.

• The release notes (location doc/release/0.X.rst) are up-to-date with any bug fixes or new features.
• The documentation is still up-to-date in the doc folder.
• The dependencies are still up-to-date in setup.py.

[mgjarrett]

I believe the coverage decrease is only because I deleted covered lines. No new uncovered lines were introduced, and unit tests were enhanced to provide more coverage of the new feature.

[keckler]

There are a couple questions that'd I'd like to get clarified before I approve this.

But nominally I think it looks fine.

Is there a way that you can show me how you are convinced that this is solving the problem that you're trying to fix?

And finally, are there performance or other issues that this will now cause that were leading us to not just go this route in the first place?

[mgjarrett]

Is there a way that you can show me how you are convinced that this is solving the problem that you're trying to fix?

I will send you some evidence.

And finally, are there performance or other issues that this will now cause that were leading us to not just go this route in the first place?

I don't think there's a significant performance hit, but it does require paring down from a much longer initial list of material boundaries. The meshList length here is now going to be determined by the total number of assemblies instead of unique assembly types. In most cases all but 1 or 2 of them will eventually get pop'ed.

armi/armi/reactor/converters/uniformMesh.py

Lines 266 to 294 in 320778d

	while True:
	for i in range(len(meshList) - 1):
	difference = abs(meshList[i + 1] - meshList[i])
	if difference < minimumMeshSize:
	if meshList[i] in anchorPoints and meshList[i + 1] in anchorPoints:
	errorMsg = (
	"Attempting to remove two anchor points!\n"
	"The uniform mesh minimum size for decusping is smaller than the "
	"gap between anchor points, which cannot be removed:\n"
	f"{meshList[i]}, {meshList[i+1]}, gap = {abs(meshList[i]-meshList[i+1])}"
	)
	runLog.error(errorMsg)
	raise ValueError(errorMsg)
	if meshList[i + 1] in anchorPoints:
	removeIndex = i
	else:
	removeIndex = i + 1
	if warn:
	runLog.warning(
	f"{meshList[i + 1]} is too close to {meshList[i]}! "
	f"Difference = {difference} is less than mesh size "
	f"tolerance of {minimumMeshSize}. The uniform mesh will "
	f"remove {meshList[removeIndex]}."
	)
	break
	else:
	return sorted(meshList)
	meshList.pop(removeIndex)

[mgjarrett]

The coverage decrease is again from removing previously covered lines. All of the new lines added are covered: