AERO / PUBLICATION-TARGETED PCHE STUDY · V1
NUMERICAL FAILThe frozen v1 prediction was compared against the accessible publication after the run. No exact experimental point could be matched to the frozen operating condition, so the experimental comparison is honestly left unscored.
3 grids run · 0 cleared every numerical gate · prediction package hash-verified
Numerical disposition: FAIL. All meshes pass OpenFOAM’s region checks, and conservation checks pass their predeclared limits. Residuals remain above the 1×10−6 target on every grid, and medium-to-fine pressure drop changes by more than the 5% allowance.
Experimental agreement: UNSCORED. The source results were reviewed, but no exact measured point matching this frozen condition and definition was established. No validation claim is made.
Samarmad and Jaffal (2023), Results in Engineering 19, 101303. Open the DOI record. The independent reconstruction is one periodic Model 6 cell; it is not author CAD. The analysis uses a frozen 293 K cold inlet, 323 K hot inlet and 0.246 m/s per stream. The source’s 290/293 K and 4/5 mm discrepancies, plus the inferred corrugation-amplitude unit, remain explicit uncertainties.
The solver received no measurement values or result plots. However, the orchestration agent had prior exposure to the abstract-level summary and a Section 4.1 discrepancy, and the solver environment was networked rather than hermetic. This is a reference-withheld deterministic run, not a fully blind benchmark.
For a straight, fully developed semicircular duct only: Re 519.12, hydraulic diameter 1.5275 mm, ΔP 576.2 Pa/channel, and effectiveness 0.365–0.411.
This screening calculation omits corrugation, entrance and manifold effects; it is not presented as a Model 6 prediction.
The author-uploaded full-text record reports CC BY-NC-ND 4.0. This page does not reproduce the article, its figures, tables or measured point values. Local analysis was documented as noncommercial; commercial reuse is not cleared.
The paper contains no structural material model, supports, mechanical loads or structural measurements.
Each row is retained even where a gate fails. Cell counts are tetrahedra in the reconstructed periodic cell.
| Grid | Tetrahedra | Maximum residual | Mass imbalance | Energy imbalance | Every gate |
|---|---|---|---|---|---|
| coarse | 494,670 | 2.138e-4 · FAIL | 0.07% · PASS | 3.67% · PASS | FAIL |
| medium | 1,003,677 | 5.769e-6 · FAIL | 0.09% · PASS | 2.49% · PASS | FAIL |
| fine | 2,770,939 | 6.699e-6 · FAIL | 0.06% · PASS | 1.50% · PASS | FAIL |
High Courant and solid diffusion numbers are retained in the detailed assessment. End-time completion is not treated as convergence, and the frozen acceptance criteria were not changed after the runs.
The publication does not specify a structural case, so it cannot support a paper-correlated FEA prediction. An exploratory, assumption-only transfer plan used the fine-grid solid temperature field and required exactly 100% mapping before creating any structural deck.
| Target solid mesh | Temperature nodes mapped | Coverage | Disposition |
|---|---|---|---|
| coarse | 85,619 / 92,113 | 92.95% | FAIL · no deck |
| medium | 182,061 / 192,543 | 94.56% | FAIL · no deck |
| fine | 504,066 / 504,078 | 100.00% | FAIL · no deck |
A prior single assumption-only CalculiX shakedown used an unconverged coarse CFD field; it is retained separately and is not a source prediction or a mesh-convergence result.
The 653-entry prediction inventory was independently re-hashed with zero mismatches. Manifest SHA-256: e09516c5946190b72e29009f0bc3f99ae3f7586c2c29e89e3f9dc02539e07cfb; inventory SHA-256: 9f34e3f2615437fe15c61b60b36fcddb671ca09cdecd0739689201494437e92d.
The frozen case is 0.246 m/s per stream (solver Re ≈519.105). The accessible paper reports experimental curves over a Reynolds sweep, but the review did not establish an exact pointwise experimental observation matching this operating condition and measurement definition. The preregistered rule prohibits scoring a mismatched point or fitting after reveal; therefore the Aero-versus-experiment error is UNSCORED.
The paper itself reports aggregate Model 6 differences of 11.6% in Nusselt number and 11.5% in friction factor between its own simulation and experiments, with experimental values slightly higher. That is source context—not Aero's score. V1's OpenFOAM numerical disposition remains FAIL; no post-reveal tuning was performed.
OpenTimestamps was submitted to four calendars; Bitcoin confirmation was still pending at reveal. This procedural caveat is disclosed. The freeze is hash-verified, but no confirmed external timestamp is claimed.