High: [Reactors] Coil heat transfer, sensitivity estimation, and steady-state PFR modes crash

[sergey-gusev94]

Maintainer triage (June 29, 2026)

• Status: Verified bug against the current master source in this fork.
• Severity: High.
• Area labels: area:reactors.
• Tackle batch: P3 - documented-mode crashes and tooling breakages.

The source review confirmed that the cited code path and failure mode are still present. The original audit details are preserved below for reproduction notes and suggested fixes.

---

Original audit details

Severity: High / Medium · Category: crash (control-flow + API misuse) · Part of #67

Three Reactors.py defects, each of which makes a documented reactor configuration crash. Grouped as one "make the reactor module runnable in all advertised modes" fix.

---

1. (High) PlugFlowReactor.solve_steady reads non-existent Inlet.concentr

Location: PharmaPy/Reactors.py:1604-1607

c_inlet = self.Inlet.concentr (1604), then indexed. No phase/stream stores concentr: LiquidPhase.__set_amounts sets only mole_conc/mass_conc/mole_frac/mass_frac; LiquidStream.updatePhase maps its concentr argument positionally onto the base mole_conc. There is no concentr property anywhere.

• Trigger: build a PFR, assign a LiquidStream inlet, call reactor.solve_steady(vol_rxn) (also feeds plot_steady/concProfSteady/tempProfSteady).
• Wrong result: AttributeError: 'LiquidStream' object has no attribute 'concentr'; the entire steady-state PFR solver and plot_steady are unrunnable.
• Fix: use self.Inlet.mole_conc (the dynamic path already uses self.Liquid_1.mole_conc).

2. (Medium) CSTR / SemibatchReactor solve_unit(eval_sens=True) leaves problem undefined

Location: PharmaPy/Reactors.py:1134-1144 (and 1326-1336)

if eval_sens: pass (1134-1135) never builds problem; only the else branch creates Explicit_Problem. Execution then unconditionally reaches solver = CVode(problem) (1144). paramest_wrapper calls solve_unit(eval_sens=True) whenever self.return_sens is True, which defaults True in the CSTR/Semibatch constructors.

• Trigger: sensitivity-based parameter estimation on a CSTR/SemibatchReactor with default return_sens=True.
• Wrong result: NameError: name 'problem' is not defined; CSTR/Semibatch parameter estimation crashes immediately.
• Fix: implement the eval_sens problem setup (mirror BatchReactor.solve_unit 800-813 with get_sundials_callable, problem.jac, problem.rhs_sens), or raise NotImplementedError instead of pass.

3. (Medium) heat_transfer ht_mode='coil' leaves heat_transf undefined

Location: PharmaPy/Reactors.py:294-302

if self.ht_mode == 'coil': pass (296-297) never assigns heat_transf, yet return heat_transf (302) is reached unconditionally. The tank reactors' energy balances call self.heat_transfer(...) for any non-isothermal run (BatchReactor 711; CSTR 1060/1066). 'coil' is an advertised ht_mode (docstring 603-605).

• Trigger: BatchReactor(isothermal=False, ht_mode='coil') (or CSTR), solve_unit.
• Wrong result: UnboundLocalError: local variable 'heat_transf' referenced before assignment on the first RHS evaluation.
• Fix: implement coil-mode area/duty, or raise NotImplementedError for ht_mode=='coil' rather than falling through.

---

Related (not duplicates): #55 (sensitivity Jacobian block sizing), #54 (temp_ht metadata), #33 (PFR specific area diam/4) are distinct defects.

[parkyr]

Triage: reproduced

All three defects reproduce on unchanged base source; the root causes and fix locations in the audit above are confirmed as written. One correction: defect 2 raises UnboundLocalError, not NameError — problem is bound in the else branch, so it is a local name, not a global lookup.

Environment

• Base commit: 1a3ab31 (origin/master)
• Command: pytest tests/test_reactor_modes.py
• Python 3.11, assimulo present (PharmaPy/Reactors.py imports CVode at module scope, so the test carries pytest.mark.assimulo)

Observed vs. expected

Path	Observed on 1a3ab31	Expected
PlugFlowReactor.solve_steady (Reactors.py:1604)	AttributeError: 'LiquidStream' object has no attribute 'concentr'	Returns; concProfSteady populated
CSTR.solve_unit(eval_sens=True) (Reactors.py:1144)	UnboundLocalError: local variable 'problem' referenced before assignment	Sensitivity problem built, or explicit NotImplementedError
SemibatchReactor.solve_unit(eval_sens=True) (Reactors.py:1336)	UnboundLocalError: local variable 'problem' referenced before assignment	Sensitivity problem built, or explicit NotImplementedError
heat_transfer(ht_mode='coil') (Reactors.py:302)	UnboundLocalError: local variable 'heat_transf' referenced before assignment	Duty returned, or explicit NotImplementedError

Failing test (red on unchanged source)

tests/test_reactor_modes.py — builds real LiquidStream / LiquidPhase / RxnKinetics / CoolingWater objects from tests/integration/data/pfr_test_pure_comp.json; no stubs or monkeypatching.

FAILED tests/test_reactor_modes.py::test_pfr_solve_steady_reads_inlet_mole_conc
  PharmaPy/Reactors.py:1604: in solve_steady
      c_inlet = self.Inlet.concentr
  E   AttributeError: 'LiquidStream' object has no attribute 'concentr'

FAILED tests/test_reactor_modes.py::test_sensitivity_mode_does_not_fall_through[CSTR]
  PharmaPy/Reactors.py:1144: in solve_unit
      solver = CVode(problem)
  E   UnboundLocalError: local variable 'problem' referenced before assignment

FAILED tests/test_reactor_modes.py::test_sensitivity_mode_does_not_fall_through[SemibatchReactor]
  PharmaPy/Reactors.py:1336: in solve_unit
      solver = CVode(problem)
  E   UnboundLocalError: local variable 'problem' referenced before assignment

FAILED tests/test_reactor_modes.py::test_coil_ht_mode_does_not_fall_through
  PharmaPy/Reactors.py:302: in heat_transfer
      return heat_transf
  E   UnboundLocalError: local variable 'heat_transf' referenced before assignment

4 failed in 0.62s

The two fall-through tests accept NotImplementedError as a pass, so they do not prescribe implementing coil heat transfer or CSTR sensitivities — only that the mode stops leaking an internal control-flow error.

Sanity check that the test is not vacuously red: with defect 1 changed to self.Inlet.mole_conc and the two pass branches replaced by raise NotImplementedError, all 4 tests pass (4 passed in 0.59s).

Notes for the fix

• The one-line fix for defect 1 is sufficient: with Inlet.mole_conc substituted, solve_steady runs to completion (CVode converges in 109 steps) and populates concProfSteady. No further defect is hiding behind the AttributeError.
• Defect 2 is reachable in normal use, not only via an explicit eval_sens=True: paramest_wrapper (Reactors.py:474-477) calls solve_unit(eval_sens=True) whenever self.return_sens is true, and return_sens=True is the constructor default for CSTR (Reactors.py:956) and SemibatchReactor (Reactors.py:1260).

Acceptance criteria

1. PlugFlowReactor.solve_steady(vol_rxn) completes against a LiquidStream inlet and populates concProfSteady; plot_steady / concProfSteady / tempProfSteady are reachable.
2. CSTR.solve_unit(eval_sens=True) and SemibatchReactor.solve_unit(eval_sens=True) never raise UnboundLocalError / NameError: either a sensitivity problem is constructed, or the mode is explicitly refused.
3. heat_transfer with ht_mode='coil' never raises UnboundLocalError / NameError, under the same either/or.
4. If a mode is refused rather than implemented, the refusal is raised at call time and names the unsupported mode, so paramest_wrapper and the non-isothermal energy balances fail loudly rather than at an unrelated line.
5. The existing suite still passes: pytest tests/ -m "not assimulo" and the assimulo integration job.