Skip to content

0.2.7 WP11 - Existing J2 maturity and closure evidence

Decision boundary

WP11 records a controlled maturity extension for the existing small-strain von Mises J2 route. It does not change the material implementation, a solver formulation, a default Newton path, or any Owner threshold. The source under test is 94461602dfd1782be57c20e1801a0d5d8e262ef1; the WP11 execution was started from 4d0ee14f4aa61b9337874a991263a93b4f9a8c73.

The resulting decision is PASS_WITH_LIMITATIONS for the work package and KEEP_QUALIFIED_BOUNDED_WITH_LIMITATIONS for MAT-J2-SMALL, subject to Owner review. MAT-FINITE-J2 remains EXPERIMENTAL/NOT_QUALIFIED.

Controlled evidence

The machine-readable case catalog and evidence are:

  • qualification/0_2_7/wp11_j2_cases.json
  • qualification/0_2_7/wp11_j2_evidence.json
  • qualification/0_2_7/wp11_state.json

The result digest is 5e4825625d40f2363ecefb8f96baa43acac42f23db7195d000ca0c09717ef536. The artifact records the environment, source SHA, input digest, result digest, predeclared policies and all four element-family rows.

Material-point and tangent checks

The direct J2 path covers elastic predictor, first yield, radial return, plastic strain and equivalent plastic strain, stress consistency, unloading, reloading and a simple reversal cycle. All checks are finite and pass. The algorithmic tangent is symmetric as a diagnostic. Central finite differences cover elastic, near-yield, traction, compression, shear, non-proportional, reload and cyclic states at the declared perturbations; the maximum relative error is 2.1204721119376345e-10, below the existing G06 limit 1e-6.

Four-family structural evidence

Family Multi-element Cyclic Energy Rollback Full Newton Modified Newton
TET4 PASS PASS PASS PASS PASS non-converged, diagnostic
TET10 PASS PASS PASS PASS PASS non-converged, diagnostic
HEX8 PASS PASS PASS PASS PASS non-converged, diagnostic
HEX20 PASS PASS PASS PASS PASS non-converged, diagnostic

Multi-element maximum relative residuals are 2.39e-8, 1.93e-8, 3.39e-8 and 7.80e-8 in the table order. The maximum relative energy balance errors are 1.84e-12, 5.49e-9, 8.64e-9 and 1.80e-9. Rollback injects one controlled rejected increment per family and verifies the committed-state digest is restored before retry; no state contamination, NaN/Inf or silent pass is recorded.

Increment and failure characterization

A monotonic J2 path at load scale 0.2 is run with 1, 2 and 4 subdivisions per branch for every family. The finest-level comparison records family- specific displacement and internal-variable sensitivities, but no universal partition-independence threshold is introduced. This evidence does not erase the existing G06 bounded policy, whose strongest independent increment claim remains the declared TET4 study.

Full Newton converges for all four families. Modified Newton non-convergence is reproduced and classified as diagnostic rather than hidden or promoted. Invalid yield stress and hardening inputs fail closed with deterministic ValueError diagnostics. Existing rollback rejection is retained as the transactional failure-path evidence.

External and adjacent capability evidence

The Code_Aster 18.1.0 J2 correlation from qualification/0_2_6/g06_depth_evidence.json is reused as controlled material-level evidence. It is not a new WP11 run and does not extend the structural increment claim. The existing bounded buckling result is referenced by provenance only; WP11 adds no buckling qualification. Modal, Newmark and harmonic/dynamics gaps are outside this work package and remain unchanged.

Maturity and limitations

No capability is promoted or demoted by WP11. The bounded small-strain J2 scope is better characterized on TET4, TET10, HEX8 and HEX20. The following remain explicit limitations:

  • no universal structural increment-independence threshold;
  • algorithmic tangent symmetry is not separately Owner-qualified;
  • full Newton is the accepted nonlinear route for this evidence;
  • cyclic calibration, Bauschinger behavior and physical validation are out of scope;
  • finite-kinematic J2 and coupled nonlinear routes remain experimental or not qualified;
  • no full regression is run in WP11, by policy.

FIXES_APPLIED = NONE; FUNCTIONAL_SOURCE_CHANGED = NO.