OpenMC/openmc/deplete/operator.py

629 lines
22 KiB
Python

"""OpenMC transport operator
This module implements a transport operator for OpenMC so that it can be used by
depletion integrators. The implementation makes use of the Python bindings to
OpenMC's C API so that reading tally results and updating material number
densities is all done in-memory instead of through the filesystem.
"""
import copy
from collections import OrderedDict
from itertools import chain
import os
import time
import xml.etree.ElementTree as ET
import h5py
import numpy as np
import openmc
import openmc.capi
from openmc.data import JOULE_PER_EV
from . import comm
from .abc import TransportOperator, OperatorResult
from .atom_number import AtomNumber
from .reaction_rates import ReactionRates
def _distribute(items):
"""Distribute items across MPI communicator
Parameters
----------
items : list
List of items of distribute
Returns
-------
list
Items assigned to process that called
"""
min_size, extra = divmod(len(items), comm.size)
j = 0
for i in range(comm.size):
chunk_size = min_size + int(i < extra)
if comm.rank == i:
return items[j:j + chunk_size]
j += chunk_size
class Operator(TransportOperator):
"""OpenMC transport operator for depletion.
Instances of this class can be used to perform depletion using OpenMC as the
transport operator. Normally, a user needn't call methods of this class
directly. Instead, an instance of this class is passed to an integrator
function, such as :func:`openmc.deplete.integrator.cecm`.
Parameters
----------
geometry : openmc.Geometry
OpenMC geometry object
settings : openmc.Settings
OpenMC Settings object
chain_file : str, optional
Path to the depletion chain XML file. Defaults to the
:envvar:`OPENMC_DEPLETE_CHAIN` environment variable if it exists.
prev_results : ResultsList, optional
Results from a previous depletion calculation. If this argument is
specified, the depletion calculation will start from the latest state
in the previous results.
Attributes
----------
geometry : openmc.Geometry
OpenMC geometry object
settings : openmc.Settings
OpenMC settings object
dilute_initial : float
Initial atom density to add for nuclides that are zero in initial
condition to ensure they exist in the decay chain. Only done for
nuclides with reaction rates. Defaults to 1.0e3.
output_dir : pathlib.Path
Path to output directory to save results.
round_number : bool
Whether or not to round output to OpenMC to 8 digits.
Useful in testing, as OpenMC is incredibly sensitive to exact values.
number : openmc.deplete.AtomNumber
Total number of atoms in simulation.
nuclides_with_data : set of str
A set listing all unique nuclides available from cross_sections.xml.
chain : openmc.deplete.Chain
The depletion chain information necessary to form matrices and tallies.
reaction_rates : openmc.deplete.ReactionRates
Reaction rates from the last operator step.
burnable_mats : list of str
All burnable material IDs
local_mats : list of str
All burnable material IDs being managed by a single process
prev_res : ResultsList
Results from a previous depletion calculation
"""
def __init__(self, geometry, settings, chain_file=None, prev_results=None):
super().__init__(chain_file)
self.round_number = False
self.settings = settings
self.geometry = geometry
if prev_results != None:
# Reload volumes into geometry
prev_results[-1].transfer_volumes(geometry)
# Store previous results in operator
self.prev_res = prev_results
else:
self.prev_res = None
# Clear out OpenMC, create task lists, distribute
openmc.reset_auto_ids()
self.burnable_mats, volume, nuclides = self._get_burnable_mats()
self.local_mats = _distribute(self.burnable_mats)
# Determine which nuclides have incident neutron data
self.nuclides_with_data = self._get_nuclides_with_data()
# Select nuclides with data that are also in the chain
self._burnable_nucs = [nuc.name for nuc in self.chain.nuclides
if nuc.name in self.nuclides_with_data]
# Extract number densities from the geometry / previous depletion run
self._extract_number(self.local_mats, volume, nuclides, self.prev_res)
# Create reaction rates array
self.reaction_rates = ReactionRates(
self.local_mats, self._burnable_nucs, self.chain.reactions)
def __call__(self, vec, power, print_out=True):
"""Runs a simulation.
Parameters
----------
vec : list of numpy.ndarray
Total atoms to be used in function.
power : float
Power of the reactor in [W]
print_out : bool, optional
Whether or not to print out time.
Returns
-------
openmc.deplete.OperatorResult
Eigenvalue and reaction rates resulting from transport operator
"""
# Prevent OpenMC from complaining about re-creating tallies
openmc.reset_auto_ids()
# Update status
self.number.set_density(vec)
time_start = time.time()
# Update material compositions and tally nuclides
self._update_materials()
self._tally.nuclides = self._get_tally_nuclides()
# Run OpenMC
openmc.capi.reset()
openmc.capi.run()
time_openmc = time.time()
# Extract results
op_result = self._unpack_tallies_and_normalize(power)
if comm.rank == 0:
time_unpack = time.time()
if print_out:
print("Time to openmc: ", time_openmc - time_start)
print("Time to unpack: ", time_unpack - time_openmc)
return copy.deepcopy(op_result)
def _get_burnable_mats(self):
"""Determine depletable materials, volumes, and nuclids
Returns
-------
burnable_mats : list of str
List of burnable material IDs
volume : OrderedDict of str to float
Volume of each material in [cm^3]
nuclides : list of str
Nuclides in order of how they'll appear in the simulation.
"""
burnable_mats = set()
model_nuclides = set()
volume = OrderedDict()
# Iterate once through the geometry to get dictionaries
for mat in self.geometry.get_all_materials().values():
for nuclide in mat.get_nuclides():
model_nuclides.add(nuclide)
if mat.depletable:
burnable_mats.add(str(mat.id))
if mat.volume is None:
raise RuntimeError("Volume not specified for depletable "
"material with ID={}.".format(mat.id))
volume[str(mat.id)] = mat.volume
# Make sure there are burnable materials
if not burnable_mats:
raise RuntimeError(
"No depletable materials were found in the model.")
# Sort the sets
burnable_mats = sorted(burnable_mats, key=int)
model_nuclides = sorted(model_nuclides)
# Construct a global nuclide dictionary, burned first
nuclides = list(self.chain.nuclide_dict)
for nuc in model_nuclides:
if nuc not in nuclides:
nuclides.append(nuc)
return burnable_mats, volume, nuclides
def _extract_number(self, local_mats, volume, nuclides, prev_res=None):
"""Construct AtomNumber using geometry
Parameters
----------
local_mats : list of str
Material IDs to be managed by this process
volume : OrderedDict of str to float
Volumes for the above materials in [cm^3]
nuclides : list of str
Nuclides to be used in the simulation.
prev_res : ResultsList, optional
Results from a previous depletion calculation
"""
self.number = AtomNumber(local_mats, nuclides, volume, len(self.chain))
if self.dilute_initial != 0.0:
for nuc in self._burnable_nucs:
self.number.set_atom_density(np.s_[:], nuc, self.dilute_initial)
# Now extract and store the number densities
# From the geometry if no previous depletion results
if prev_res is None:
for mat in self.geometry.get_all_materials().values():
if str(mat.id) in local_mats:
self._set_number_from_mat(mat)
# Else from previous depletion results
else:
for mat in self.geometry.get_all_materials().values():
if str(mat.id) in local_mats:
self._set_number_from_results(mat, prev_res)
def _set_number_from_mat(self, mat):
"""Extracts material and number densities from openmc.Material
Parameters
----------
mat : openmc.Material
The material to read from
"""
mat_id = str(mat.id)
for nuclide, density in mat.get_nuclide_atom_densities().values():
number = density * 1.0e24
self.number.set_atom_density(mat_id, nuclide, number)
def _set_number_from_results(self, mat, prev_res):
"""Extracts material nuclides and number densities.
If the nuclide concentration's evolution is tracked, the densities come
from depletion results. Else, densities are extracted from the geometry
in the summary.
Parameters
----------
mat : openmc.Material
The material to read from
prev_res : ResultsList
Results from a previous depletion calculation
"""
mat_id = str(mat.id)
# Get nuclide lists from geometry and depletion results
depl_nuc = prev_res[-1].nuc_to_ind
geom_nuc_densities = mat.get_nuclide_atom_densities()
# Merge lists of nuclides, with the same order for every calculation
geom_nuc_densities.update(depl_nuc)
for nuclide in geom_nuc_densities.keys():
if nuclide in depl_nuc:
concentration = prev_res.get_atoms(mat_id, nuclide)[1][-1]
volume = prev_res[-1].volume[mat_id]
number = concentration / volume
else:
density = geom_nuc_densities[nuclide][1]
number = density * 1.0e24
self.number.set_atom_density(mat_id, nuclide, number)
def initial_condition(self):
"""Performs final setup and returns initial condition.
Returns
-------
list of numpy.ndarray
Total density for initial conditions.
"""
# Create XML files
if comm.rank == 0:
self.geometry.export_to_xml()
self.settings.export_to_xml()
self._generate_materials_xml()
# Initialize OpenMC library
comm.barrier()
openmc.capi.init(intracomm=comm)
# Generate tallies in memory
self._generate_tallies()
# Return number density vector
return list(self.number.get_mat_slice(np.s_[:]))
def finalize(self):
"""Finalize a depletion simulation and release resources."""
openmc.capi.finalize()
def _update_materials(self):
"""Updates material compositions in OpenMC on all processes."""
for rank in range(comm.size):
number_i = comm.bcast(self.number, root=rank)
for mat in number_i.materials:
nuclides = []
densities = []
for nuc in number_i.nuclides:
if nuc in self.nuclides_with_data:
val = 1.0e-24 * number_i.get_atom_density(mat, nuc)
# If nuclide is zero, do not add to the problem.
if val > 0.0:
if self.round_number:
val_magnitude = np.floor(np.log10(val))
val_scaled = val / 10**val_magnitude
val_round = round(val_scaled, 8)
val = val_round * 10**val_magnitude
nuclides.append(nuc)
densities.append(val)
else:
# Only output warnings if values are significantly
# negative. CRAM does not guarantee positive values.
if val < -1.0e-21:
print("WARNING: nuclide ", nuc, " in material ", mat,
" is negative (density = ", val, " at/barn-cm)")
number_i[mat, nuc] = 0.0
# Update densities on C API side
mat_internal = openmc.capi.materials[int(mat)]
mat_internal.set_densities(nuclides, densities)
#TODO Update densities on the Python side, otherwise the
# summary.h5 file contains densities at the first time step
def _generate_materials_xml(self):
"""Creates materials.xml from self.number.
Due to uncertainty with how MPI interacts with OpenMC API, this
constructs the XML manually. The long term goal is to do this
through direct memory writing.
"""
materials = openmc.Materials(self.geometry.get_all_materials()
.values())
# Sort nuclides according to order in AtomNumber object
nuclides = list(self.number.nuclides)
for mat in materials:
mat._nuclides.sort(key=lambda x: nuclides.index(x[0]))
materials.export_to_xml()
def _get_tally_nuclides(self):
"""Determine nuclides that should be tallied for reaction rates.
This method returns a list of all nuclides that have neutron data and
are listed in the depletion chain. Technically, we should tally nuclides
that may not appear in the depletion chain because we still need to get
the fission reaction rate for these nuclides in order to normalize
power, but that is left as a future exercise.
Returns
-------
list of str
Tally nuclides
"""
nuc_set = set()
# Create the set of all nuclides in the decay chain in materials marked
# for burning in which the number density is greater than zero.
for nuc in self.number.nuclides:
if nuc in self.nuclides_with_data:
if np.sum(self.number[:, nuc]) > 0.0:
nuc_set.add(nuc)
# Communicate which nuclides have nonzeros to rank 0
if comm.rank == 0:
for i in range(1, comm.size):
nuc_newset = comm.recv(source=i, tag=i)
nuc_set |= nuc_newset
else:
comm.send(nuc_set, dest=0, tag=comm.rank)
if comm.rank == 0:
# Sort nuclides in the same order as self.number
nuc_list = [nuc for nuc in self.number.nuclides
if nuc in nuc_set]
else:
nuc_list = None
# Store list of tally nuclides on each process
nuc_list = comm.bcast(nuc_list)
return [nuc for nuc in nuc_list if nuc in self.chain]
def _generate_tallies(self):
"""Generates depletion tallies.
Using information from the depletion chain as well as the nuclides
currently in the problem, this function automatically generates a
tally.xml for the simulation.
"""
# Create tallies for depleting regions
materials = [openmc.capi.materials[int(i)]
for i in self.burnable_mats]
mat_filter = openmc.capi.MaterialFilter(materials)
# Set up a tally that has a material filter covering each depletable
# material and scores corresponding to all reactions that cause
# transmutation. The nuclides for the tally are set later when eval() is
# called.
self._tally = openmc.capi.Tally()
self._tally.scores = self.chain.reactions
self._tally.filters = [mat_filter]
def _unpack_tallies_and_normalize(self, power):
"""Unpack tallies from OpenMC and return an operator result
This method uses OpenMC's C API bindings to determine the k-effective
value and reaction rates from the simulation. The reaction rates are
normalized by the user-specified power, summing the product of the
fission reaction rate times the fission Q value for each material.
Parameters
----------
power : float
Power of the reactor in [W]
Returns
-------
openmc.deplete.OperatorResult
Eigenvalue and reaction rates resulting from transport operator
"""
rates = self.reaction_rates
rates[:, :, :] = 0.0
k_combined = openmc.capi.keff()[0]
# Extract tally bins
materials = self.burnable_mats
nuclides = self._tally.nuclides
# Form fast map
nuc_ind = [rates.index_nuc[nuc] for nuc in nuclides]
react_ind = [rates.index_rx[react] for react in self.chain.reactions]
# Compute fission power
# TODO : improve this calculation
# Keep track of energy produced from all reactions in eV per source
# particle
energy = 0.0
# Create arrays to store fission Q values, reaction rates, and nuclide
# numbers
fission_Q = np.zeros(rates.n_nuc)
rates_expanded = np.zeros((rates.n_nuc, rates.n_react))
number = np.zeros(rates.n_nuc)
fission_ind = rates.index_rx["fission"]
for nuclide in self.chain.nuclides:
if nuclide.name in rates.index_nuc:
for rx in nuclide.reactions:
if rx.type == 'fission':
ind = rates.index_nuc[nuclide.name]
fission_Q[ind] = rx.Q
break
# Extract results
for i, mat in enumerate(self.local_mats):
# Get tally index
slab = materials.index(mat)
# Get material results hyperslab
results = self._tally.results[slab, :, 1]
# Zero out reaction rates and nuclide numbers
rates_expanded[:] = 0.0
number[:] = 0.0
# Expand into our memory layout
j = 0
for nuc, i_nuc_results in zip(nuclides, nuc_ind):
number[i_nuc_results] = self.number[mat, nuc]
for react in react_ind:
rates_expanded[i_nuc_results, react] = results[j]
j += 1
# Accumulate energy from fission
energy += np.dot(rates_expanded[:, fission_ind], fission_Q)
# Divide by total number and store
for i_nuc_results in nuc_ind:
if number[i_nuc_results] != 0.0:
for react in react_ind:
rates_expanded[i_nuc_results, react] /= number[i_nuc_results]
rates[i, :, :] = rates_expanded
# Reduce energy produced from all processes
energy = comm.allreduce(energy)
# Determine power in eV/s
power /= JOULE_PER_EV
# Scale reaction rates to obtain units of reactions/sec
rates *= power / energy
return OperatorResult(k_combined, rates)
def _get_nuclides_with_data(self):
"""Loads a cross_sections.xml file to find participating nuclides.
This allows for nuclides that are important in the decay chain but not
important neutronically, or have no cross section data.
"""
# Reads cross_sections.xml to create a dictionary containing
# participating (burning and not just decaying) nuclides.
try:
filename = os.environ["OPENMC_CROSS_SECTIONS"]
except KeyError:
filename = None
nuclides = set()
try:
tree = ET.parse(filename)
except Exception:
if filename is None:
msg = "No cross_sections.xml specified in materials."
else:
msg = 'Cross section file "{}" is invalid.'.format(filename)
raise IOError(msg)
root = tree.getroot()
for nuclide_node in root.findall('library'):
mats = nuclide_node.get('materials')
if not mats:
continue
for name in mats.split():
# Make a burn list of the union of nuclides in cross_sections.xml
# and nuclides in depletion chain.
if name not in nuclides:
nuclides.add(name)
return nuclides
def get_results_info(self):
"""Returns volume list, material lists, and nuc lists.
Returns
-------
volume : dict of str float
Volumes corresponding to materials in full_burn_dict
nuc_list : list of str
A list of all nuclide names. Used for sorting the simulation.
burn_list : list of int
A list of all material IDs to be burned. Used for sorting the simulation.
full_burn_list : list
List of all burnable material IDs
"""
nuc_list = self.number.burnable_nuclides
burn_list = self.local_mats
volume = {}
for i, mat in enumerate(burn_list):
volume[mat] = self.number.volume[i]
# Combine volume dictionaries across processes
volume_list = comm.allgather(volume)
volume = {k: v for d in volume_list for k, v in d.items()}
return volume, nuc_list, burn_list, self.burnable_mats