QF Solver 0.2.8 WP04 — MITC3 and MITC4 V&V

This record freezes the WP04 contract from baseline 2a6a02859e3f19ddf26903167de1db6fdd3f3c7c. The machine-readable contract is wp04_mitc_vnv.json.

Historical evidence audit

WP01 classified MITC3 static as NEEDS_MINOR_VNV, MITC3 modal/Newmark/harmonic as NEEDS_MAJOR_VNV, and all four MITC4 routes as NEEDS_MINOR_VNV. The historical Owner review documents remain useful provenance, but their cited qualification/vnv, qualification/evidence, results and PDF artifacts are not all present in this clone. They are therefore not treated as current executable correlations and are not recreated or rewritten.

Frozen scope and decision rule

The campaign promotes only bounded, reproducible scopes. MITC3 static uses affine membrane/bending/shear patches, rigid-body and energy/reaction invariants, locking and reference-shell trends. MITC3 modal, Newmark and harmonic remain experimental because their current independent same-mesh oracles are missing.

MITC4 static uses element mechanics, patch, orientation/objectivity, drilling, locking and declared structural convergence. MITC4 modal uses a slender cantilever analytical reference, but its final relative-residual observation (2.6317020430528298e-08) does not meet the stricter frozen WP04 limit (1e-08), so it remains EXPERIMENTAL. MITC4 Newmark uses an exact first-mode time history with temporal refinement and energy checks. MITC4 harmonic uses a closed-form first-mode FRF across several frequencies, including zero-frequency and resonance checks.

All limits are predeclared in the JSON record before the WP04 campaign. A technical PASS is not a universal claim: each accepted route remains bounded by the exact geometry, material, mesh, loading, damping and observable scope listed in the record.

Local orientation, permutations and thickness

The executable shell checks include local-frame/objectivity and orientation contracts, consistent node ordering, rigid-body invariance and finite positive thickness cases. Reversed or inconsistent shell orientation remains rejected by the mesh contract; it is not silently repaired into a qualification pass.

External correlation policy

No absent historical Code_Aster, CalculiX or PDF artifact is counted as a live correlation. The approved MITC4 dynamic scopes use independent analytical time/frequency references because their claims are explicitly limited to a single verified mode. The MITC3 dynamic routes do not receive that promotion: same-assembled-model internal references alone do not close their route-level oracle requirement.

Executed decision snapshot

Combination	WP04 decision	Evidence status
MITC3 linear static	QUALIFIED_BOUNDED candidate	Two deterministic replays; patch, equilibrium, energy, orientation, permutation and locking gates pass. Reference-shell trends remain bounded warnings.
MITC3 modal / Newmark / harmonic	EXPERIMENTAL	No executable independent shell oracle is present in the clone.
MITC4 linear static	QUALIFIED_BOUNDED candidate	Two deterministic replays; mechanics, patches, drilling sensitivity and structural convergence gates pass. Cook is retained as a bounded trend warning.
MITC4 modal	EXPERIMENTAL	Two deterministic replays pass the study's internal checks, but the predeclared WP04 residual gate fails; no tolerance was relaxed.
MITC4 Newmark	QUALIFIED_BOUNDED candidate	Two deterministic replays pass the declared time-step, phase/period, energy and residual gates.
MITC4 harmonic	QUALIFIED_BOUNDED candidate	Two deterministic replays pass the declared multi-frequency amplitude, phase, resonance, static-limit and residual gates.

The machine-readable replay digests, observed values and test references are recorded in wp04_mitc_vnv.json. The eight-route decision delta is summarized in the WP04 maturity matrix. The final Owner decisions are recorded in wp04_owner_gate_final.json. The four approved bounded routes are now QUALIFIED_BOUNDED within the scopes below; no claim is broadened beyond the evidence.

Across the 46-record technical matrix after the WP03 and WP04 deltas, the state is 31 QUALIFIED_BOUNDED, 14 EXPERIMENTAL, and 1 NOT_QUALIFIED, for a total of 46. The sole not-qualified combination is COMB-HEX8-linear_buckling. COMB-WEDGE6-linear_static remains a distinct EXPERIMENTAL combination and is not confused with that status.

Final Owner gate

Combination	Owner decision	Resulting state	Limitation or rejected gate
MITC3 linear static	APPROVE_WITH_LIMITATIONS	QUALIFIED_BOUNDED	Executed patch, equilibrium, energy, orientation, permutation and locking scope only.
MITC3 modal	REJECT	EXPERIMENTAL	No executable independent shell modal oracle.
MITC3 Newmark	REJECT	EXPERIMENTAL	No executable independent same-mesh transient shell oracle.
MITC3 harmonic	REJECT	EXPERIMENTAL	No executable independent same-mesh shell FRF oracle.
MITC4 linear static	APPROVE_WITH_LIMITATIONS	QUALIFIED_BOUNDED	Executed mechanics, patch, equilibrium, drilling and reference-mesh scope only.
MITC4 modal	REJECT	EXPERIMENTAL	Relative residual 2.6317020430528298e-08 exceeds frozen 1e-08 gate.
MITC4 Newmark	APPROVE_WITH_LIMITATIONS	QUALIFIED_BOUNDED	Verified first-mode, undamped Newmark temporal scope only.
MITC4 harmonic	APPROVE_WITH_LIMITATIONS	QUALIFIED_BOUNDED	Verified first-mode, seven-frequency, mass-proportional damping scope only.

The Owner gate found no numerical-core change, no opportunistic tolerance change and no alteration of 0.2.7 evidence. The machine-readable gate also reconciles all 46 source combination records directly.