RENEWED: Extending fluid concept to fit CF & Standard formulations

[vlipovac]

Closes #1240 Closes #1247

Replaces #1244 for reasons of clarity.

Proposed changes

Reworking the concept of model.fluid to be a general fluid with phases and components, instead of just a data collection of constants.

Breaking change with a multiple minor changes throughout the packages. The essential changes are summarized in a self-review in #1244.

porepy.compositional.ChemicalSpecies was merged into FluidConstants, including data class functionality. This led to some changes for SolidConstants as well for inheritance reasons.

The default model set-up is changed to provide a 1-phase, 1-component fluid, where the single component is based on the user-provided FluidConstants in the model.params. The interface to various fluid properties (e.g., fluid_density) is replaced by the general Fluid class.

Types of changes

• [ ] Minor change (e.g., dependency bumps, broken links).
• [ ] Bugfix (non-breaking change which fixes an issue).
• [ ] New feature (non-breaking change which adds functionality).
• [X] Breaking change (fix or feature that would cause existing functionality to not work as expected).
• [X] Testing (contribution related to testing of existing or new functionality).
• [ ] Documentation (contribution related to adding, improving, or fixing documentation).
• [X] Maintenance (e.g., improve logic and performance, remove obsolete code).
• [ ] Other:

Checklist

• [X] The documentation is up-to-date.
• [X] Static typing is included in the update.
• [X] This PR does not duplicate existing functionality.
• [X] The update is covered by the test suite (including tests added in the PR).
• [X] If new skipped tests have been introduced in this PR, pytest was run with the --run-skipped flag.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[vlipovac]

I reworked the concept of constant reference pressure and temperature values:

• The new class is found here. This is a preliminary place, we can also move it to models/abstract_equations.py where the VariableMixin is found.
• Sol.Strat. is adapted to have a new attribute , which represents the reference values. It is in principle analogous to the current solid constants.
• perturbation_from_reference was modified to wrap the reference values into AD and return A-A_ref
• extended material constants for testing were split, having now a dedicated part for reference variable values
• An example usage is given in the tests, like here, and multiple other (all in tests/models and tests/functional)
• VariableMixin.reference_* is removed, and the abomination of model.fluid.reference_component.pressure should be replaced everywhere by model.reference_variable_values.pressure.
• This here was the only place which did not use perturbation_from_reference, I fixed that.
• all other changes should be namespace changes due to above.

I think a test dedicated to the reference values alone, the evaluation of the perturbation from reference and so on, is called for.

[keileg]

@vlipovac: Regarding the tutorials, could you run these on a system with pypardiso installed, so that we don't introduce more error messages in the notebook output?

[vlipovac]

@vlipovac: Regarding the tutorials, could you run these on a system with pypardiso installed, so that we don't introduce more error messages in the notebook output?

Done. There is one last error after the last Python cell in stress_discretization,ipynb coming from a matplotlib.Axes object, which I am not sure how to remove.

[vlipovac]

The latest changes involve:

• renaming MaterialConstants to Constants
• renaming FluidConstants to FluidComponent and adding compositional.base.Component to its inheritance tree
• implementing __setattr__ to imitate the frozen functionality of dataclass, but limited to the constants fields declares with SI units (also, inheriting from Component the Component.fraction must not be frozen)

• Adding generic typing to Fluid, Phase, Compound such that custom material parameters are found by linters and static type checkers:

  • Compound[FluidComponent] says the tracers in the compound are of type FluidComponent
  • Phase[FluidComponent]: says this is a phase composed of FluidComponent
  • Fluid[FluidComponent, Phase[FluidComponent]]: says this is a fluid mixture of fluid-like component and respectively phases.

Minor open todos:

1. Decide on whether to leave compositional.materials.Constants and compositional.materials.ReferenceVariableValues there or find a better spot (maybe porepy.utils?)
2. Add parametrized test for children of Constants, check for freezing of constants and proper unit conversion (though the conversion is to some degree tested separately in each model)
3. Add test for parsing and evaluation of perturbation_from_reference which now accesses the reference variable values.

If you think one of the last two points is not necessary, please let me know.

[keileg]

Thanks @vlipovac. I'll have a look now.

[keileg]

@vlipovac The changes look good to me, I have no further comments to that part.

As for the remaining todos:

1. I prefer not to put things into pp.utils, so for my part, the classes can stay unless someone comes up with a much better suggestion.
2. Yes, some simple testing would be useful here.
3. Yes please.

There are three unresolved conversations that all seem to point to the creation of one (or more?) issue. I suggest we make this and then resolve. Please let me know if I misread some of the conversations.

[vlipovac]

I think we are done.

1. Added test for perturbation from reference method in tests/models/test_constitutive_laws/test_perturbation_from_reference
2. Added parametrized test for all Constants classes available so far in tests/compositional/test_materials (formerly called test_material_constants)
3. New issues created resulting from review here #1259 #1260

Constants and ReferenceVariableValues are left for now in compositional/materials. But in future we might want to put these two somewhere else (also having #1249 in mind)