composite.getReactionRates() doesn't work on the core level

[keckler]

Something appears to be wrong with this method:

armi/armi/reactor/composites.py

Lines 3060 to 3092 in 9d6e2b0

	def getReactionRates(self, nucName, nDensity=None):
	"""
	Get the reaction rates of a certain nuclide on this object.
	Parameters
	----------
	nucName : str
	nuclide name -- e.g. 'U235'
	nDensity : float
	number Density
	Returns
	-------
	rxnRates : dict
	reaction rates (1/s) for nG, nF, n2n, nA and nP
	Notes
	-----
	This is volume integrated NOT (1/cm3-s)
	If you set nDensity to 1 this makes 1-group cross section generation easier
	"""
	rxnRates = {"nG": 0, "nF": 0, "n2n": 0, "nA": 0, "nP": 0, "n3n": 0}
	# not all composite objects are iterable (i.e. components), so in that
	# case just examine only the object itself
	for armiObject in self.getChildren() or [self]:
	for rxName, val in armiObject._getReactionRates(
	nucName, nDensity=nDensity
	).items():
	rxnRates[rxName] += val
	return rxnRates

I recently loaded up a database and tried to examine some reaction rates as follows:

>  r.core.getReactionRates("U235")
Loading microscopic cross section library `ISOTXS`
 {'nG': 0, 'nF': 0, 'n2n': 0, 'nA': 0, 'nP': 0, 'n3n': 0}

> r.core.getAssemblyWithStringLocation('002-001').getReactionRates("U235")
{'nG': 4.219163274625927e+16,
 'nF': 1.5933660772471552e+17,
 'n2n': 86100493155576.52,
 'nA': 0.0,
 'nP': 0.0,
 'n3n': 0}

>  r.core.getAssemblyWithStringLocation('002-001').getReactionRates("U238")
{'nG': 1.3747239159395538e+17,
 'nF': 2.0985393232144132e+16,
 'n2n': 706063666454843.1,
 'nA': 0.0,
 'nP': 0.0,
 'n3n': 0}

It doesn't make sense that the returned reaction rates for the core as a whole are zero, but non-zero values are returned for individual assemblies from the same core.

[john-science]

Okay, it looks to me like Composite.getReactionRates() calls Composite._getReactionRates(), which assumes you are lower than the Core in the hierarchy:

armi/armi/reactor/composites.py

Lines 3040 to 3043 in f6ee150

	return getReactionRateDict(
	nucName,
	self.getAncestorWithFlags(Flags.CORE).lib,
	self.getAncestor(lambda x: isinstance(x, Block)).getMicroSuffix(),

It expects the Core to be an ancestor (parent or above), and when you pass it a Core object it just falls through this exception here:

armi/armi/reactor/composites.py

Lines 3047 to 3049 in f6ee150

	except AttributeError:
	runLog.warning(
	f"Object {self} does not belong to a core and so has no reaction rates.",

And that returns all zeros.

[keckler]

I agree with your assessment.

Another thing to put on your radar, in case we want to guard against it. I have seen that some blueprints name their assemblies something like Inner core fuel. This automagically gives the assembly the CORE flag. And then the logic that you highlight in your comment above will actually get tripped up, in particular the line

self.getAncestorWithFlags(Flags.CORE).lib

The code will try to get a lib from the assembly object, and obviously it will fail in this. The upshot is that more all-zero dicts are returned.