026-G08 Active Owner Final Review

Decision

026-G08 remains PASS_WITH_LIMITATIONS under the active Owner decision APPROVED_BOUNDED_WITH_FAMILY_LIMITATIONS. No requirement was lowered and no solver, formulation or eigensolver code changed.

This document and qualification/0_2_6/g08_owner_final_review.json are the active review record. The earlier g08_owner_closeout.json is retained as historical evidence and is not silently rewritten; its family interpretation is superseded by this review where the later corrected Euler and CalculiX evidence applies.

Provenance chain

Evidence role	Artifact	Source SHA
Approved contract	g08_requirements.json	4145f1f42ed5aec513ccf05e215e16e590132546
Controlled execution	g08_execution_evidence.json	6589443e1404a2749ac6c0a9b911f00dd9cb8753
Corrected high-order analytical screen	g08_high_order_analytical_evidence.json	4c45b46cfa7c9a664986bb49a53ef25fa570c030
Active Euler prevalidation	g08_euler_prevalidation_evidence.json	75fbe03a414969d26dee9f0df84ff1c5dcd1661e
HEX20 external rescue	g08_hex20_calculix_rescue_evidence.json	8ee87e5f4093f9152a2e4e5b52cbfafec22c2d7b

The positive-load/tension comparison in g08_euler_reference.json is retained as superseded history and is excluded from active metrics. The review artifact itself is documentary evidence at the review HEAD and does not replace any execution source SHA.

Family decisions

Family	Decision	Active basis	Limitation
TET4	QUALIFIED_BOUNDED	Historical bounded Euler evidence, eligible mesh sequence, residual/mode/replay and bounded CalculiX evidence	Only the declared first-factor/first-mode domain is covered
TET10	PASS_WITH_LIMITATIONS	Corrected Euler trend reaches 3.269%; active mesh change 0.081448%; residual/replay and CalculiX evidence pass	No universal analytical or mesh-convergence claim; transverse diagnostic remains less accurate
HEX8	MORE_EVIDENCE_REQUIRED	C3D8 agrees with QF within 0.0003835% and 0.0004065%; mode/KG diagnostics are coherent	Corrected Euler error remains 298.413%; low-order locking/discretization limitation is likely, not a demonstrated QF bug
HEX20	PASS_WITH_LIMITATIONS	Corrected Euler trend reaches 3.622%; active mesh change 0.912621%; three-level C3D20 correlation passes	Coarse-history sensitivity remains; no universal analytical claim

The HEX8 result is deliberately not promoted. Agreement with C3D8 is useful independent evidence about the discretization behaviour, but it does not make a poor continuum-oracle result a qualified physical result.

Requirement disposition

Disposition	Requirements
OWNER_APPROVED_FULL	G08-001, G08-002, G08-004, G08-006, G08-007, G08-009
OWNER_APPROVED_BOUNDED	G08-003, G08-005, G08-008
DEFERRED	None at gate level
BLOCKING	None for the bounded gate; HEX8 remains blocked at family-claim level

Evidence summary

• Euler: corrected signed compression/load-factor comparison is active; the high-order trends are TET10 27.278% -> 8.465% -> 3.269% and HEX20 22.956% -> 8.072% -> 3.622%. HEX8 remains 298.413% at its final axial screen.
• External: bounded CalculiX evidence is available for the four families across the retained evidence chain. Code_Aster is SKIPPED_NOT_COMPARABLE for this 3D solid eigen-buckling route.
• Mesh: active adjacent changes are TET10 0.081448%, HEX8 0.167113% and HEX20 0.912621%; these are bounded tested-domain observations, not universal convergence laws.
• Mode/residual: finite deterministic first-mode evidence passes; the active HEX8 diagnostic maximum residual is 2.39e-8, below the declared 1e-7 pass band.

Qualified boundary

The bounded claim is limited to linear buckling, the first linearized tangent-instability factor and first mode, homogeneous isotropic 3D material, nodal dead loads, and the sparse SciPy route in the tested domain. TET4 is the qualified family; TET10 and HEX20 have bounded limitations; HEX8 is excluded from the qualified family subset pending further evidence.

Post-buckling, collapse, arc-length, multi-mode qualification, shell/beam/ discrete routes and general physical validation are excluded.