Test drying energy rate basis

[Mazhar331]

Summary

• Add a focused Drying unit_model regression that drives the real energy-balance call path.
• Verify molar drying rates are converted component-wise to mass rates before the gas sensible-heat and condensed latent-heat terms consume them.
• Pin independently derived gas and condensed temperature-rate results while leaving issue High: [Drying] Condensed-phase energy balance doubles latent heat #24's separate latent-heat multiplier unchanged.

Refs #21, #24, #48, #108.

Acceptance Criteria

• Gas sensible heat combines cpg in J/kg/K with total drying rate in kg/m^3/s.
• Condensed latent cooling combines latent heat in J/kg with component drying rates in kg/m^3/s.
• The regression fails against unchanged origin/master and passes with PR Fix drying gas mass-rate balance #108's shared conversion.
• The unrelated factor-of-two correction remains scoped to High: [Drying] Condensed-phase energy balance doubles latent heat #24.

Dimensional Analysis

• get_drying_rate: mol/m^3/s.
• PR Fix drying gas mass-rate balance #108 conversion: (mol/m^3/s) * (kg/mol) = kg/m^3/s.
• Sensible term: (J/kg/K) * K * (kg/m^3/s) = J/m^3/s.
• Latent term: (kg/m^3/s) * (J/kg) = J/m^3/s.

Coordination Notes

• This PR is stacked on PR Fix drying gas mass-rate balance #108 because that PR owns the single shared molar-to-mass conversion in unit_model. Duplicating the conversion in energy_balance would double-convert after Fix drying gas mass-rate balance #108 lands.
• Closure ownership: PR Fix drying gas mass-rate balance #108 supplies the production fix, while this PR supplies the energy-specific regression. Keep High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 open until both changes reach master, then close it manually with links to Fix drying gas mass-rate balance #108 and this PR.
• Issue High: [Drying] Condensed-phase energy balance doubles latent heat #24 owns removal of the existing * 2 latent-heat factor on the same energy expression. This PR documents and tests the current multiplier without changing its behavior; High: [Drying] Condensed-phase energy balance doubles latent heat #24 will need to update the pinned condensed-rate expectation when it removes the multiplier.
• A full class-material/Purdue-style solve_unit Drying transient example is intentionally deferred. The blocker is not the absence of a checked-in example; it is that open Drying correctness work still affects realistic transient assertions (Critical: [Drying] Gas species balance adds molar source terms to mass-fraction states #21 / PR Fix drying gas mass-rate balance #108, High: [Drying] Condensed-phase energy balance doubles latent heat #24, High: [Drying] Gas molecular weight is computed from mass fractions as mole fractions #28, Medium: [Drying] Gas energy balance multiplies convective enthalpy by density twice #37, Medium: [Drying] Volatile liquid mass-fraction column is forced to zero each RHS evaluation #42, High: [Drying] Gas balance and Darcy velocity produce order-of-magnitude drying transient errors #81, and High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101). Once those land, the class examples can seed broader integration coverage, likely under testing umbrella Expand test coverage across modules (unit + integration) #7 or a dedicated follow-up issue.
• No other open PR adds or changes tests/test_drying_energy_rate_basis.py. PR Fix drying gas mass-rate balance #108 overlaps in PharmaPy/Drying_Model.py and tests/test_drying_model.py, but those changes are the prerequisite base rather than competing hunks.

Tests

• conda run -n pharmapy python -m pytest tests/test_drying_model.py tests/test_drying_energy_rate_basis.py -q — 5 passed.
• conda run -n pharmapy python -m pytest tests/ -m "not assimulo" — 33 passed, 12 deselected.
• conda run -n pharmapy python -m pytest tests/ -m assimulo — 12 passed, 33 deselected.
• Red check on exact origin/master (1a3ab31) — failed as expected: gas energy rates were about 27 times the mass-basis expectations.

Branch Hygiene

• Base branch: fix/issue-21-drying-gas-mass-rate (non-default).
• Source branch point: beb53e40d07bb06a7be0e573858e52947f808cbb, the current head of PR Fix drying gas mass-rate balance #108.
• Stacked status: stacked directly on PR Fix drying gas mass-rate balance #108.
• Prerequisite: PR Fix drying gas mass-rate balance #108 must merge before this PR is retargeted to master or merged through the stack.
• This test-only PR intentionally uses Refs #48; close High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 manually after PR Fix drying gas mass-rate balance #108 and this PR both reach master.

[Mazhar331]

Maintainer review. I am the PR author, so GitHub only allows a COMMENT review from this account; this is not a formal approval. mergeStateStatus is CLEAN and reviewDecision is empty (the branch has no protection reporting a review gate), so no maintainer approval is being demanded by live GitHub state.

The change is a single new focused regression test (tests/test_drying_energy_rate_basis.py, +99), no production code touched. It is correct, well-scoped, and genuinely guards issue #48.

What I verified

• Changed-file set matches gh pr diff --name-only (one new test file). Merge-base is beb53e4, exactly the base tip / PR #108 head.
• Expected values are independent literals, hand-checked against the code, not recomputations of production expressions:
  • mass conversion mol/m^3/s * (MW/1000) = kg/m^3/s → rates [0.036, 0, 0.184], [0.054, 0, 0.230].
  • gas sensible cpg*(Tg-295)*sum(m_dot) = [1100, 3408]; denom_gas = cvg*eps*rho = [450, 1100] → dTg/dt = [1100/450, 3408/1100].
  • condensed latent sum(m_dot_i * L_i * 2) = [512000, 676000]; denom_cond = 900000 → dTcond/dt = [-512000/900000, -676000/900000].
• Mutation check: deleting the _drying_rate_mass_basis call in unit_model (the pre-#108 / issue #48 defect) makes the test fail — gas rate 66.67 vs expected 2.44 (≈27×), matching the PR's "about 27 times" red-check claim. The test is a real regression guard, not vacuously green.
• Test lanes reproduce the PR body exactly: -m "not assimulo" → 33 passed, 12 deselected; -m assimulo → 12 passed, 33 deselected; targeted test_drying_model.py + test_drying_energy_rate_basis.py → 5 passed.
• The * 2 latent factor is pinned and documented as owned by #24; the test correctly characterizes current behavior rather than changing it. Keeping this a focused unit test (rather than an end-to-end run that would assert #24's unstable multiplier) is the right call.
• Working tree clean after runs; only gitignored .pyc artifacts generated.

Nonblocking

• The production fix for #48 (the dry_rate * MW / 1000 conversion at Drying_Model.py:287) lives in PR #108, which is the base of this stacked PR — this PR's own diff contains only the test. Closes #48 therefore only takes effect once the stack reaches master, and #48 is not actually "resolved with regression coverage" from #116's diff alone. The body already documents the stacking and the manual-close caveat, so this is fine as-is; just confirm the intended close ownership between #108 (fix) and #116 (test) so #48 is neither closed prematurely by #108 nor left dangling.
• The fixture uses number_nodes == num_volatiles == 2. In energy_balance the latent term broadcasts latent_heat (per-node in the real model) against dry_rate[:, idx_volatiles] (per-volatile columns); with both dims equal to 2 the test cannot distinguish per-node from per-component latent semantics. This does not weaken the #48 guard (the mutation check moves both gas and condensed rates ≈27×, and the sensible term via dry_rate.sum is unambiguous), and the underlying broadcasting shape in energy_balance:410 is pre-existing and out of scope here. Worth a note only if a future change makes nodes ≠ volatiles.

Summary

• Blocking: none.
• Nonblocking: Closes #48 close-ownership across the #108/#116 stack; nodes == volatiles == 2 fixture coincidence in the latent term.
• Questions: none.
• Tests run: targeted drying tests (5 passed), not assimulo lane (33 passed), assimulo lane (12 passed), and a mutation check confirming the test fails without the mass conversion. All green.
• Merge-readiness: content is ready on the merits. It targets the non-default base fix/issue-21-drying-gas-mass-rate and is stacked on PR #108, so #108 must land (and this PR be retargeted to master) before merge and before #48 auto-closes. This author COMMENT does not satisfy any branch-protection approval that may apply after those gates clear.

[Mazhar331]

Addressed the review feedback for head d02a597.

Changes made:

• Updated the PR body from Closes #48 to Refs #48.
• Added explicit closure ownership: PR Fix drying gas mass-rate balance #108 supplies the production fix, PR Test drying energy rate basis #116 supplies energy-specific regression coverage, and High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 should be closed manually only after both reach master.
• Kept the stacked base and prerequisite notes current; PR Fix drying gas mass-rate balance #108 remains open with CHANGES_REQUESTED, while PR Test drying energy rate basis #116 remains CLEAN against that base.

Commits pushed: none. This was a metadata-only correction; repository content and head SHA are unchanged.

Verification on the unchanged head:

• Core tests: pass.
• Assimulo integration tests: pass.
• Prior local checks remain applicable: 5 focused drying tests, 33 core tests, and 12 Assimulo-marked tests passed.

Intentionally not changed:

• The number_nodes == num_volatiles == 2 fixture coincidence is a nonblocking note about a pre-existing latent-heat broadcasting contract. Changing dimensions would expose or require a separate energy-shape fix outside this PR's mass-basis scope; it does not weaken the verified High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 mutation guard.

No unresolved inline threads exist, so no thread replies were required. reviewDecision remains empty; the remaining merge dependency is PR #108.

[Mazhar331]

Addressed the latent-heat fixture-shape review note on head cb18bc6.

Changes made:

• Added commit cb18bc6 (Clarify drying energy test shape).
• Changed tests/test_drying_energy_rate_basis.py from 2 nodes / 2 volatiles to 3 nodes / 2 volatiles.
• Kept latent_heat as a length-2 vector and added a short comment that those values are per volatile component, not per node.
• Updated the third row's hand-computed expected gas and condensed temperature rates.

Why this is OK for the current PR:

• It keeps PR Test drying energy rate basis #116 focused on issue High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48's mass-rate energy-basis regression.
• It removes the accidental nodes == volatiles coincidence without pulling the broader VaporPhase.getHeatVaporization shape contract into this PR.
• The broader provider/consumer shape follow-up remains appropriate for issue High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 rather than High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48/Test drying energy rate basis #116.

Verification:

• conda run -n pharmapy python -m pytest tests/test_drying_energy_rate_basis.py -q -> 1 passed.
• conda run -n pharmapy python -m pytest tests/test_drying_model.py tests/test_drying_energy_rate_basis.py -q -> 5 passed.
• conda run -n pharmapy python -m pytest tests/ -m "not assimulo" -> 33 passed, 12 deselected.
• conda run -n pharmapy python -m pytest tests/ -m assimulo -> 12 passed, 33 deselected.
• GitHub Actions on cb18bc6: Core tests and Assimulo integration tests both passed.

No unresolved inline review threads exist. reviewDecision remains empty and mergeStateStatus is CLEAN; the remaining stack dependency is still PR #108.

[Mazhar331]

Coordination follow-up for the latent-heat shape note:

• The local ambiguity in this PR's test fixture has been removed by commit cb18bc6 (3 nodes, 2 volatiles).
• The broader provider/consumer shape contract is now recorded on High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101: High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 (comment)
• I am not adding this to High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 separately because High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48/Test drying energy rate basis #116 are scoped to molar-to-mass drying-rate basis in energy terms; High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 is the cleaner owner for the VaporPhase.getHeatVaporization shape contract and the associated Drying consumer audit.

[Mazhar331]

Read-only resolution check of my earlier maintainer review (COMMENT on head d02a597), now re-verified against current head cb18bc6 and cross-checked with the coordination notes on this PR and on #101.

Both review findings addressed

• closes-48-fix-in-108 (Nonblocking) — Addressed. PR body now reads Refs #21, #24, #48, #108 and states close High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 manually only after PR Fix drying gas mass-rate balance #108 and this PR both reach master. Close-ownership between Fix drying gas mass-rate balance #108 (production fix) and Test drying energy rate basis #116 (regression coverage) is explicit.
• nodes-eq-volatiles-coincidence (Nonblocking) — Addressed by commit cb18bc6. The fixture is now number_nodes=3 with num_volatiles=2, so nodes != volatiles and the length-2 latent_heat stub can only be read as per volatile component (the added comment makes this explicit). The third node's expected gas/condensed rates are independently hand-derivable (sensible = 1500*(310-295)*0.110 = 2475, denom_gas = 1400*0.25*2 = 700; latent = (0.018*2e6 + 0.092*1e6)*2 = 256000, denom_cond = 900000).

Verification performed (read-only)

• test_drying_energy_rate_basis.py at head cb18bc6 → 1 passed; test_drying_model.py + test_drying_energy_rate_basis.py → 5 passed.
• Re-ran the mutation check at the 3-node head (scratch copy, real checkout untouched): deleting the _drying_rate_mass_basis conversion makes the test fail with gas rates [66.7, 87.3, 96.4] vs expected [2.44, 3.10, 3.54] (≈27×). The test still genuinely guards issue High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 after the fixture change.
• No approval gate: master is unprotected (protection endpoint 404), reviewDecision empty, mergeStateStatus: CLEAN. No inline review threads exist. The only remaining merge dependency is PR Fix drying gas mass-rate balance #108 (currently CHANGES_REQUESTED against master).

Coordination notes are consistent and technically accurate

The note on #101 (gh-coord:source-pr=116) and its backlink here are mutually cross-linked and correctly scoped. I verified its technical claim against the source:

• VaporPhase.getHeatVaporization(..., basis="mass") multi-temperature branch (Phases.py:653) reassigns deltahvap = deltahvap[:, idx] / self.mw[idx] * 1000, reducing columns to the subcritical idx. For the drying path that subset equals the volatile columns, so Drying_Model.py:410 (dry_rate[:, self.idx_volatiles] * latent_heat) works today only because the High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 bug pre-selects those columns.
• After High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 preserves full (num_nodes, num_components) columns, Drying_Model.py:410 will hit a (nodes, 2) * (nodes, 3) shape mismatch unless it selects latent_heat[:, self.idx_volatiles] first — exactly as the High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 note prescribes. Deferring the provider/consumer shape contract to High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 rather than folding it into High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48/Test drying energy rate basis #116 is the right split.

One follow-up for #101 (not blocking #116)

This PR's getHeatVaporization stub is a 1-D per-component vector, which is a simplification of the real multi-temperature return (a 2-D (num_nodes, num_volatiles) array today, (num_nodes, num_components) after #101). When #101 changes the consumer to latent_heat[:, self.idx_volatiles], indexing this 1-D stub with [:, idx_volatiles] will raise, so #101's Drying-consumer audit should explicitly update this existing test's stub to 2-D, not only add a new nodes != volatiles case. Recording so the coupling isn't lost.

Assessment

All review feedback on #116 is resolved at head cb18bc6. The fixture change is test-only and stays within the exact scope my finding anticipated, so no new automated review round is warranted. Merge-readiness is unchanged: content is ready, gated only on PR #108 landing and this PR being retargeted to master.

[Mazhar331]

Addressed the latest review feedback on head cf940d9.

Changes made:

• Added commit cf940d9 (Document drying energy test units).
• Added a module-level description to tests/test_drying_energy_rate_basis.py explaining that the test is a compact synthetic Drying.unit_model regression, not a full solve_unit drying example.
• Documented that no checked-in PharmaPy example currently constructs an end-to-end Drying run from full phase/cake classes.
• Added bracketed unit annotations throughout the test fixture and assertions: dimensionless fractions/indices, pressure, length, heat capacity, viscosity, latent heat, drying rate, state layout, and expected temperature rates.

Verification:

• conda run -n pharmapy python -m pytest tests/test_drying_energy_rate_basis.py -q -> 1 passed.
• conda run -n pharmapy python -m pytest tests/test_drying_model.py tests/test_drying_energy_rate_basis.py -q -> 5 passed.
• conda run -n pharmapy python -m pytest tests/ -m "not assimulo" -> 33 passed, 12 deselected.
• conda run -n pharmapy python -m pytest tests/ -m assimulo -> 12 passed, 33 deselected.
• GitHub Actions on cf940d9: Core tests and Assimulo integration tests both passed.

The inline review thread has been replied to. reviewDecision still reports CHANGES_REQUESTED until the reviewer updates or dismisses that review. The remaining stack dependency is still PR #108.

[Mazhar331]

Addressed the corrected end-to-end Drying test rationale on head 0010799.

Changes made:

• Added commit 0010799 (Document drying integration deferral).
• Updated tests/test_drying_energy_rate_basis.py to clarify that external class-material examples can seed future integration coverage, but a full Drying transient regression is deferred until open Drying correctness issues are resolved.
• Updated the PR body from broad “end-to-end” wording to a focused unit_model energy-balance regression and added the same deferral rationale to the coordination notes.
• Added the matching High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 coordination note: High: [Drying] Energy terms multiply molar drying rates by mass-basis heat quantities #48 (comment)

Why no full class-style solve_unit example is added here:

• Critical: [Drying] Gas species balance adds molar source terms to mass-fraction states #21 / PR Fix drying gas mass-rate balance #108 is still open, so the shared mass-rate conversion has not reached master.
• High: [Drying] Condensed-phase energy balance doubles latent heat #24 still owns the latent-heat * 2.
• High: [Drying] Gas balance and Darcy velocity produce order-of-magnitude drying transient errors #81 still tracks order-of-magnitude Drying transient errors.
• High: [Drying] Gas molecular weight is computed from mass fractions as mole fractions #28, Medium: [Drying] Gas energy balance multiplies convective enthalpy by density twice #37, Medium: [Drying] Volatile liquid mass-fraction column is forced to zero each RHS evaluation #42, and High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 are still open Drying/latent-heat correctness blockers.

Verification:

• conda run -n pharmapy python -m pytest tests/test_drying_energy_rate_basis.py -q -> 1 passed.
• conda run -n pharmapy python -m pytest tests/test_drying_model.py tests/test_drying_energy_rate_basis.py -q -> 5 passed.
• GitHub Actions on 0010799: Core tests and Assimulo integration tests both passed.

The inline review thread has been replied to. reviewDecision still reports CHANGES_REQUESTED until the reviewer updates or dismisses that review.

[Mazhar331]

Read-only review-resolution check of @bernalde's CHANGES_REQUESTED review, verified against current head 0010799f (commits cf940d9, 0010799f).

Comment-by-comment status

@bernalde's inline thread on tests/test_drying_energy_rate_basis.py (still isResolved: false) had three asks:

• Add a file description — Addressed. cf940d9 adds a module docstring stating the file is a compact synthetic unit_model regression, not a full solve_unit example. Present at head.
• Concern that the test is too time-consuming — Addressed. The test is a single unit_model right-hand-side evaluation with every heavy collaborator stubbed (material_balance, high_resolution_fvm, get_drying_rate), so it is inherently fast (~1s; the cb18bc69..0010799f delta is docstring + unit comments + a black-style reformat, with no numeric literal or expected value changed). Green Assimulo CI on 0010799f confirms it runs and passes.
• Add an end-to-end drying example (if any) from the classes — Documented deferral (declined for now). No checked-in end-to-end Drying example exists, and a full transient regression is deferred because the realistic-transient path depends on open Drying-correctness work — Critical: [Drying] Gas species balance adds molar source terms to mass-fraction states #21 / PR Fix drying gas mass-rate balance #108, High: [Drying] Condensed-phase energy balance doubles latent heat #24, High: [Drying] Gas molecular weight is computed from mass fractions as mole fractions #28, Medium: [Drying] Gas energy balance multiplies convective enthalpy by density twice #37, Medium: [Drying] Volatile liquid mass-fraction column is forced to zero each RHS evaluation #42, High: [Drying] Gas balance and Darcy velocity produce order-of-magnitude drying transient errors #81, and High: [Phases] Multi-temperature VaporPhase mass-basis latent heat drops component columns #101 are all currently OPEN. This is documented in the docstring and PR body (matching the practice of keeping a focused regression test when broader coverage would assert unstable behavior tied to open defects).

What this pass did not change

reviewDecision is CHANGES_REQUESTED from @bernalde, a different reviewer than the PR author. GitHub does not clear that gate when the comments are addressed — it stays until @bernalde submits a new review or dismisses the prior one; the PR author cannot clear another reviewer's CHANGES_REQUESTED. @bernalde has not re-reviewed since the cf940d9/0010799f fixes. mergeStateStatus is CLEAN (no conflicts), but the review gate blocks merge.

The only substantively open item is the end-to-end example, which was deferred rather than added. The rationale is factually accurate (all cited blockers verified OPEN), but whether the documented deferral is acceptable — versus requiring an example in this PR — is @bernalde's decision, not something the author can self-resolve.

Assessment

• Documentation asks: addressed at head.
• End-to-end example: reasonably deferred with a sound, verifiable rationale; awaiting @bernalde's accept/reject.
• No new automated review round is warranted (the fixes are documentation plus a semantically-inert fixture reformat, within the scope of the original comment).
• Next step is a human one: @bernalde to re-review 0010799f and either accept the deferral (dismiss or submit a fresh review to release the gate) or state that an end-to-end example must land in this PR.

Verification-only pass; no code, reviews, or thread states were modified.

[Mazhar331]

Coordination note from PR #119 / issue #81:

The end-to-end Drying example/test discussed on this PR should remain deferred until the open Drying correctness work is solved and merged. PR #119 fixes #81's gas-velocity, gas-source holdup, and single-node initialization defects, but a realistic Drying.solve_unit transient still depends on the rest of the drying correctness cluster.

Before turning the class/Purdue-style example into an integration assertion, wait for #21 / PR #108, #24, #28, #37, #42, #81 / PR #119, and #101 to land. Until then, PR #116's focused unit_model -> energy_balance regression is the right level of coverage for #48.