Insufficient testing on buReducingAssemblyRotation

[drewj-tp]

ARMI provides functionality for a fuel handler to do "burnup reducing" assembly rotation

armi/armi/physics/fuelCycle/assemblyRotationAlgorithms.py

Lines 38 to 40 in f1f3dbe

	def buReducingAssemblyRotation(fh):
	"""
	Rotates all detail assemblies to put the highest bu pin in the lowest power orientation.

armi/armi/physics/fuelCycle/hexAssemblyFuelMgmtUtils.py

Lines 32 to 41 in f1f3dbe

	def getOptimalAssemblyOrientation(a, aPrev):
	"""
	Get optimal assembly orientation/rotation to minimize peak burnup.
	Notes
	-----
	Works by placing the highest-BU pin in the location (of 6 possible locations) with lowest
	expected pin power. We evaluated "expected pin power" based on the power distribution in
	aPrev, which is the previous assembly located here. If aPrev has no pin detail, then we must use its
	corner fast fluxes to make an estimate.

The testing for this functionality is lacking, in my opinion; it checks that that the orientation is different, not "optimal"

armi/armi/physics/fuelCycle/tests/test_assemblyRotationAlgorithms.py

Lines 30 to 50 in f1f3dbe

	class TestFuelHandlerMgmtTools(FuelHandlerTestHelper):
	def test_buReducingAssemblyRotation(self):
	fh = fuelHandlers.FuelHandler(self.o)
	hist = self.o.getInterface("history")
	newSettings = {CONF_ASSEM_ROTATION_STATIONARY: True}
	self.o.cs = self.o.cs.modified(newSettings=newSettings)
	assem = self.o.r.core.getFirstAssembly(Flags.FUEL)
	# apply dummy pin-level data to allow intelligent rotation
	for b in assem.getBlocks(Flags.FUEL):
	b.breakFuelComponentsIntoIndividuals()
	b.initializePinLocations()
	b.p.percentBuMaxPinLocation = 10
	b.p.percentBuMax = 5
	b.p.linPowByPin = list(reversed(range(b.getNumPins())))
	addSomeDetailAssemblies(hist, [assem])
	rotNum = b.getRotationNum()
	rotAlgos.buReducingAssemblyRotation(fh)
	self.assertNotEqual(b.getRotationNum(), rotNum)

So that could use some improvement.

I propose we add more testing, specifically on the internal getOptimalAssemblyOrientation function, to check that

1. We get the correct, not just different, orientation rotating an assembly "in place" (a is aPrev in the function)
2. We get the correct orientation rotating an assembly in the location of a different assembly (a is not aPrev)

Addendum

This last sentence

armi/armi/physics/fuelCycle/hexAssemblyFuelMgmtUtils.py

Lines 40 to 41 in f1f3dbe

	aPrev, which is the previous assembly located here. If aPrev has no pin detail, then we must use its
	corner fast fluxes to make an estimate.

is wrong and should be removed. If we don't have any pin information, the getOptimalAssemblyOrientation raises an error

armi/armi/physics/fuelCycle/hexAssemblyFuelMgmtUtils.py

Lines 119 to 122 in f1f3dbe

	raise ValueError(
	"Cannot perform detailed rotation analysis without pin-level "
	"flux information."
	)

[john-science]

If we had unit tests for this that were of very small reactors (~three rings), would that make it easier to hand-calculate the optimum rotation?

(And yes, ditch that last sentence.)

[drewj-tp]

If we had unit tests for this that were of very small reactors (~three rings), would that make it easier to hand-calculate the optimum rotation?

It's probably good to have. I'm working on a test that picks up a test reactor like

armi/armi/physics/fuelCycle/tests/test_fuelHandlers.py

Lines 64 to 73 in b436537

	def setUp(self):
	"""
	Build a dummy reactor without using input files.
	There are some igniters and feeds but none of these have any number densities.
	"""
	self.o, self.r = test_reactors.loadTestReactor(
	self.directoryChanger.destination,
	customSettings={"nCycles": 3, "trackAssems": True},
	)

So we already have something that's halfway decent. The trick is setting the block data such that you have a known input state (pin powers, max burnup pin index, etc.) such that you can compare the rotation you get vs. the rotation you expect