WP11B HEX8 buckling experimental-readiness contract

This contract is separate from WP06/WP06B. It does not rewrite their Euler or mesh-convergence failures and does not change any WP06 tolerance.

Experimental scope

The candidate scope is HEX8 linear eigenvalue buckling from a linear-elastic prestress state, on valid bounded geometries, with explicitly tested boundary conditions and the first positive instability factor/mode only. It is for exploratory/research use.

The frozen benchmark is a rectangular solid cantilever block (L=4.0, b=1.2, h=1.0) with all DOFs fixed on x=0 and uniform nodal dead compression on the x=L face (P=1.0). The material is homogeneous isotropic linear elasticity (E=1000, nu=0.3). Levels are (2,2,2), (4,4,4) and (6,6,6) HEX8 cells.

Frozen gates

Preload residual must be at most 1e-8, geometric-stiffness symmetry at most 1e-12, and recomputed eigen residual at most 1e-7. The critical factor must be finite and positive; the associated mode must be finite, physically lateral (UZ dominant in this weak-axis setup), and free of NaN/Inf.

Doubling E must double the critical factor within 5%; doubling preload P must halve it within 5%. Doubling L must produce a finite positive factor lower than the reference. A moderate geometry perturbation must remain finite and within 10% of the level-4 reference. Two replay digests must match exactly.

Refinement is characterization, not a convergence-to-Euler gate. Factors, adjacent changes, modes and residuals must be reported honestly; a non-stable trend cannot be presented as general convergence.

Explicit exclusions

Euler prediction, certification, production use, imperfections, nonlinear/post-buckling, plasticity, arbitrary meshes, universal accuracy, multi-mode or mixed-mesh buckling, HEX8R, SRI, B-bar and hourglass control are outside this contract. An unavailable external oracle is recorded as such and is never invented.