Original v5
13.094% maximum grid change
2,880 → 23,040 → 77,760 cells. The 1% mesh gate failed.
Inspect v5 valuesAERO / TWO-FLUID CHT QUALIFICATION
FAILED_NUMERICAL_GATESHX-CFD-QUAL-01: acceptance repair complete; qualification still failed.
Alex Blythe · September 21, 2026. The latest audit repairs the evaluator and recovers native flux, energy and interface evidence. Cold-side field/flux settling and unresolved acceptance definitions remain. No new CFD was run.
Decision: HX-03 and HX-04 Stage B remain blocked
CURRENT STATUS · 2026-09-21
What the completed audit established
Fail-closed acceptance checks now have 40 behavioral tests. Native flux, energy/storage and interface heat rates were recovered; 36 field comparisons have exact topology correspondence. Cold-side flux/velocity settling and unresolved contract definitions still block qualification. No new CFD or historical rescoring.
Matched 800–1000 s solid-storage energy closure: 0.00766%. The tube U change over that full window remains 0.001853 against the 0.001 comparison limit. A small signed-average mass error hides larger alternating instantaneous errors. Mesh independence remains NOT_EVALUATED.
What this status means
Real solver evidence, not a marketing pass. These are retained `chtMultiRegionFoam` fixture results with three regions: shell fluid, tube fluid and solid wall.
Not physical validation. The fixture uses constant/linear reference properties, has no verified CAD-to-solver identity, and has no experimental comparison in this record. A numerical gate pass does not establish design readiness.
Historical record: original versus v7
| Gate / quantity | Original v5 | Remediation v7 |
|---|---|---|
| Mesh quality | PASS | PASS |
| Run completion | PASS | PASS |
| Hot/cold mass conservation | PASS | PASS |
| Energy conservation | PASS | PASS |
| Mesh independence | FAIL | PASS |
| Residual convergence | PASS_THRESHOLD | NOT_ESTABLISHED |
| Monitor stability | NOT_ESTABLISHED | NOT_ESTABLISHED |
| HX-03 Stage B | STOPPED | STOPPED |
The mesh result changed; the conclusion did not get inflated
Remediation v7
0.995% maximum grid change
184,320 → 360,000 → 622,080 cells. The same 1% gate passes.
Inspect v7 valuesWhat remains open
- Current numerical evidence: cold-side flux errors persist and the full-window tube U change remains above the stationarity comparison limit.
- Current contract gaps: mass temporal aggregation, absolute-pressure/property coverage, interface-temperature tolerance, and geometry-backed local/history definitions require independent predeclaration.
- Recovered evidence: monitors, native energy/storage and integrated interface heat transfer are now quantified. They do not override failed or unestablished hard gates.
- Historical residual result: v7 remains unchanged at 0.04778247 against its historical 1e-3 criterion.
No HX-03 Stage B work is authorized by this result. The correct public status is still FAILED_NUMERICAL_GATES.
Solver-control diagnostic
An isolated coarse-case diagnostic changed only nOuterCorrectors from 1 to 2 and ran to 100 s. The maximum initial residual was 0.00144013 in the retained v7 comparison block and 0.00220704 in the control block. This did not establish an improvement, and final linear residuals were not substituted for the qualification gate.
DIAGNOSTIC_ONLY_NOT_QUALIFIED. The residual and monitor-stability gates remain open.
ParaView fields and retained telemetry
Download the derived visualization package to inspect temperature, velocity, and p_rgh in ParaView. Each bundle contains a .pvd collection for the shell, tube, and solid latest-time snapshots.
Charts and diagnostic figures below are click-through slides. Playback is reserved for the actual 3-D flow-line views; opening a chart will never start a recording.
Retained shell and tube outlet-temperature telemetry through 800 s.
Open flow lines in ParaViewFlow-line reportFlow-entry reportCombined story reportFlow-line manifestFlow-line checksumsDownload combined MP4
| Run | ParaView bundle | Additional view |
|---|---|---|
| Original v5 fine | v5_fine_vtk.zip | Outlet monitor chart Residual decomposition Transverse localization Combined CFD flow story |
| v6 fine | v6_fine_vtk.zip | |
| Retained v7 fine | v7_fine_vtk.zip | |
| Unchanged continuation | continuation_v1_fine_vtk.zip | |
| Solver-control coarse | solver_control_v1_coarse_vtk.zip |
The fields are latest-time snapshots exported from retained OpenFOAM cases. The combined recording is a presentation of those retained fields and the entry reveal; it is not a substitute for a time-resolved solver dataset. Its final segment uses the retained T values on a fixed 435–600 K display scale to make the gradient easier to see, without changing the solver data. The solid-equation observability repair is retained as a patched-copy report with numeric e residuals.
Retained local evidence
The public files are a compact status record. The canonical evidence packages remain in the local Aero reports and retain solver logs, control dictionaries, mesh checks, monitor series and SHA-256 manifests.