Re-run affected load cases, combinations and complete nonlinear or staged histories. Compare displacements, member forces, reactions and design demands with the results already used. Review delivered calculations and sizing based on them. Follow each advisory for its affected inputs and user action. A refusal or an unavailable quantity supplies no replacement answer: it is not zero demand or a passing check.
Safety advisories
| Advisory | Severity | What changed and what to do |
|---|---|---|
| SA-2026-048 | High | CIRSOC 301 omitted end moments from Mmax for any section with a shortened unbraced segment; channels and tees in double curvature could also miss a sign change. Re-check all CIRSOC 301 members with a declared Lb shorter than the member, including steel checks and optimiser proposals. |
| SA-2026-049 | High | Rigid restraint plus spring could become an elastic support in 3D. Choose the intended boundary condition explicitly and re-run affected analyses. |
| SA-2026-050 | Medium | Failed slenderness could become zero, and multi-case results could contain non-finite numbers. Discard those values, review inputs and re-run. |
| SA-2026-051 | High | 3D moving loads missed interior peaks and axial axle weight. Re-run envelopes and review member sizing and design demands. |
| SA-2026-052 | High | 2D/3D moving-load axial envelopes missed distributed-load boundaries. Re-run and check interior axial extrema; reducing the train step alone is insufficient. |
| SA-2026-053 | Low | Moving-load envelopes included exterior force limits at member-end loads. Re-run axial/shear envelopes and review sizing, retaining support reactions in equilibrium checks. |
| SA-2026-054 | High | Linear solves could return unresolved stiffness responses or miss a load component. Re-run static cases and combinations; independently verify demands when the corrected solver refuses. |
| SA-2026-055 | High | P-Delta understated amplification under axial span loads. Re-run affected analyses and derived checks, verify mesh convergence and reassess stability. |
| SA-2026-056 | High | Unfixed: constrained mixed frame/shell dynamics change under equivalent model descriptions. Independently verify affected demands before using them; no conservative correction factor is established. |
| SA-2026-057 | Medium | Plastic displays could call an outside factor inside its interval. Re-run certificates and compare exact endpoints. A resource refusal does not prove that no mechanism exists. |
| SA-2026-058 | High | Large harmonic models retained inaccurate omitted-mode responses. Re-run sweeps and review amplitude, phase and peaks, including frequencies outside the truncation warning. |
| SA-2026-059 | High | Advanced web analyses gave cables compressive resistance. Use the supported dedicated cable solver or independent analysis; advanced cable workflows now refuse explicitly. |
| SA-2026-060 | High | Staged artificial rotational stiffness changed translations and member actions. Re-run full staged histories with releases or support springs and review derived design demands. |
| SA-2026-061 | High | Path reactions were omitted or recovered at the wrong load; staged and dynamic forces could be missing. Re-run and review support, connection and foundation demands. Unavailable forces are not zero; total dynamic reactions remain unavailable. |
| SA-2026-062 | High | An initial elastic refusal could become a successful zero raw 3D plastic collapse factor. Re-run affected plastic analyses; a numerical refusal is not evidence of collapse at zero load. |
| SA-2026-063 | High | Constraints dropped prescribed restrained-master motion. Re-run complete histories and review displacements, member forces and reactions together with SA-2026-061. Unsupported cable/time-history support motion now refuses. |
Other changes that affect answers
The results registry ends at version 49 in 0.5.0 and reaches 78 in 0.6.0. The table selects other changes to review alongside the advisories above. Re-run every analysis covered by an applicable notice and compare its outputs. The full list, including superseded and withdrawn checks, is in the release notes.
| Results version | Result that changes |
|---|---|
| 50 | 2D: rotations that only connectors hold, left free by them, are refused |
| 52 | 3D: node rotations a connector couples but leaves free are refused |
| 53 | Buckling no longer takes geometric stiffness from round-off axial forces. In 2D and 3D, members whose axial forces cannot be distinguished from round-off contribute no geometric stiffness. The analysis is refused if their possible compression could lower the critical factor by more than 1 %. |
| 54 | The net area of a staggered bolted end takes its worst chain, and an all-elements bolted end needs bolts in the web |
| 56 | 3D inclined supports count restraints in a common frame |
| 58 | Modal bases use the rounding scale of each mass Gram entry |
| 59 | Checked 3D plastic analysis returns the certified lower bound |
| 61 | Concrete fibre load control refines its stopping interval |
| 63 | Linear analysis refuses a mechanism or an unresolvable stiffness instead of solving it with a warning |
| 65 | Invalid frozen contacts and cable bar analyses are withheld |
| 66 | Buckling round-off bounds follow actual constraint coupling |
| 67 | Buckling factors use the original stiffness pencil |
| 68 | Plastic balance uses solved model IDs and certificates budget actual candidates |
| 69 | Nonlinear convergence follows physical equilibrium |
| 71 | Include transverse truss actions in conditional member balance |
| 72 | Arc-length continuation regroups reference-load arithmetic |
| 73 | Modal analysis preserves the physical mass range |
| 75 | SSI corrects elastic residual plateaus |
| 76 | Constraint reduction retains directional rounding uncertainty |
| 78 | Compatible loads on released rotations |
Two notices describe behavior already in 0.5.0
These two notices were registered later. Their underlying changes were already in v0.5.0, so they are not new numerical changes in this release. They can still appear in a project re-check notice.
| Results version | Result that changes |
|---|---|
| 72 | The public slab gallery uses hard simple supports on walls |
| 72 | SSI returns support reactions and includes soil springs in constraint forces |
Known limitations
- SA-2026-056 remains unfixed: independently verify dynamic demands for constrained mixed frame-and-shell models that change materially after renumbering, rotation or unit conversion. No conservative correction factor is established.
- 2D plastic collapse remains refused pending #758, as described in SA-2026-044. A certified 3D lower bound does not certify the raw hinge history.
- Time history does not provide total dynamic support demands. Use an independent analysis including inertia, damping and support loading. Unsupported cable analyses and nonzero prescribed support motion in cable/time-history workflows refuse explicitly.
- Spatial arc-length and displacement-control solvers are not added. Existing warnings and restrictions remain applicable.
Other changes
- Generated planar structures receive explicit restraints, and sheds default to physical bracing. Regenerate and review the supports and bracing; saved models are unchanged.
- The Fehér, Sipos and Várkonyi bent-strip shape-control experiment has an interactive solver, actuator paths and a paper comparison in English, Spanish and Portuguese. The cut-strip summary is also translated into Spanish and Portuguese.
- The 0.5.0 blog release article now has full German, French and Turkish translations. Plastic certificate warnings and resource refusals are translated into every offered language.