Determine nuclides correctly for DAGMC models in d1s.get_radionuclides

[SteSeg]

Description

While trying to run a d1s simulation on a DAGMC model @eepeterson and I noticed that the model.geometry.get_all_nuclides() method used in get_radionuclides() of d1s.py does not return any nuclide list in the case of a DAGMCUniverse model. We figured it would be useful to add get_all_materials() and get_all_nuclides() directly in the openmc.Model class that can get such info in both csg an dagmc geometries.

If such methods should not be in the Model class, a quick workaround, directly in d1s.py, that would work would be:

# Determine what nuclides appear in the model
model_nuclides = set(model.geometry.get_all_nuclides())

if not model_nuclides:
    model_nuclides = model.materials.get_nuclides()

Also, it seems that as it is now, the get_radionuclides() method does not take into account the naturally occurring radioactive nuclides, which causes an error in a d1s run if model.materials have such nuclides. So we added the capability of including naturally occurring radionuclides with:

# Add naturally occuring radionuclides
if nuclide.half_life is not None:
    radionuclides.add(nuclide.name)

Checklist

• I have performed a self-review of my own code
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)

[paulromano]

Also, it seems that as it is now, the get_radionuclides() method does not take into account the naturally occurring radioactive nuclides, which causes an error in a d1s run if model.materials have such nuclides.

@SteSeg Can you describe what the error is that you run into in this situation? Even if a radionuclide were present in the original material composition, the only way for it to make a contribution to the decay photon flux is if it is produced by a neutron reaction in which case it will get added during get_radionuclides anyway.

[SteSeg]

Also, it seems that as it is now, the get_radionuclides() method does not take into account the naturally occurring radioactive nuclides, which causes an error in a d1s run if model.materials have such nuclides.

@SteSeg Can you describe what the error is that you run into in this situation? Even if a radionuclide were present in the original material composition, the only way for it to make a contribution to the decay photon flux is if it is produced by a neutron reaction in which case it will get added during get_radionuclides anyway.

It was an error that came up when calling either the d1s.time_correction_factors() or d1s.apply_time_correction() method (I don't remember). It was giving back just the name of the nuclide (Ca48 in my case). But I tried to reproduce it commenting out the two lines that included the naturally occurring radionuclides (that I though would solve the issue) but apparently the error is not raised anyways (or anymore). So I just removed those lines right now.

[paulromano]

@SteSeg For now, I'd prefer not to add new methods to Model. I think we need an overall better strategy for dealing with DAGMC and CSG models consistently and adding more methods just adds more technical debt. I've made a change on your branch taking advantage of the _materials_by_id property and this should fix the problem with DAMC models for now.

[pshriwise]

@SteSeg if you do need to get DAGMC cells and their materials, I'd have a look at the Model.sync_dagmc_universes capability. Note it requires one to initialize the model with Model.init_lib first. Once this is called, the Geometry.get_all_materials/cells calls should work more naturally I believe.