The numbers below are the ones recorded in the repo at the time of writing. Test counts are not repeated here because they change. Everything is early access and under verification; none of it replaces a hand check or the engineer's review.
Elastic-plastic limit load
- Closed cylinder Ri 500, t 10 (B1): 5.7229 MPa against 5.7160, +0.12 % on 504 elements, +0.06 % on 990. Open cylinder (B2) +0.11 %. Sphere (B3) +0.04 %. These sit above the target by the faceting of flat elements.
- Simply supported circular plate (B4): +0.00 % on 1216 elements. Clamped plate (B5): +7.8 %, +4.2 % and +2.9 % at 1216, 4864 and 9765 elements. The graded plate mesh is coarse at the rim, where the hinge forms. A clamped flat head needs a fine rim mesh.
- Thick cylinders (B7): +0.12 % at Ro/Ri 1.25 and 1.5 against the shell closed form, and the closed form itself is 1.1 % and 3.4 % low against a solid, so conservative.
- Strip in four-point bending (B6): +1.5 %, the strip's own width (b/t = 10 bends partly as a plate), not a mesh effect.
- CalculiX 2.22: end-capped cylinder slice F1 +0.116 %; strip F2 M/Mp 1.0158 against 1.0158. Nozzle on a cylinder F4 −2.0 % against S8R. A flat plate with nozzle F6 −14.4 % against S8R, outside the ±3 % band (CalculiX has a rigid knot at the junction), −2.9 % against an axisymmetric solid.
- PTB-3 E5.2.2: 630 psig converges; the collapse search gives 811.6 psi against 812.8, about −0.15 %.
- Element and material: consistent tangent against central differences 3.2 × 10⁻¹⁰·E; layered K against the linear shell stiffness to 2 × 10⁻¹⁵; Mp reached to 3 × 10⁻¹⁵; Newton against quasi-Newton within 0.003 %.
Eigenvalue buckling
- Simply supported square plate in uniaxial compression: within 2 % of 4π²D/b², which is 723.05 N/mm for E = 200 000 MPa, ν = 0.3, t = 10 mm, b = 1000 mm; the second mode within 3 % of 6.25/4 times it.
- Thin cylinder in axial compression: between 0.95 and 1.10 of the classical E·t/(R√(3(1 − ν²))) = 1210 MPa for R = 250, t = 2.5. In tension, no positive factor, and the reversed factor matches within 2 %.
- Long cylinder free to go oval under following pressure: within 2 % of 3D/R³ = 0.733 MPa for R = 506, t = 12.
- Skirt: beam-theory stress above the opening within 5 %, and the opening in compression buckles at a lower factor than in tension (an ordering test, not a reference value).
Agitator and skirt hand checks
- The FE check model exported as a CalculiX deck: displacement −2.3 % and peak stress +2.6 % (dev log 2026-10-10). This is for the general model, not an agitator-specific run.
- Skirt 4.5.17.3: the dev log states agreement with a COMPRESS page to 4 decimals. The page is not in the repo.
Not done
- 5.2.4 elastic-plastic with large displacement (waits on finite strain and thinning); 5.3.3 local failure by strain limit; elastic-plastic ratcheting (planned as milestone M11). Ratcheting is checked by the elastic Bree diagram only.
- Method B buckling (5.4.3) with imperfections and plasticity; buckling of a critical location that is not a cylinder, cone, sphere or formed head.
- Hardening in a structure (the job runs perfect plasticity at 1.5S); creep; the Division 1 Appendix 46 route.
- Skirt: base ring, anchor bolt chairs, gussets, stiffeners tight to the sleeve. Agitator: gussets and the flange are not meshed, no dynamics, one stress cycle per revolution.
- Large-model limits: the limit-load job refuses above 15 000 elements. Run times measured in Node: 13 to 14 s on 4178 elements and about 48 s on 12 194.
[UNVERIFIED] Whether the engineer has accepted the design decisions the limit-load route runs on (D3, D5, D8, D11 and the added rules) is recorded as open in the M5 dev log entry, not as closed. The agreement figures above are against closed forms, CalculiX and PTB-3, not an independent review.
FE results are early access and under verification. Treat any figure here as a record of a check, not a guarantee.