# Aero FEA implementation and acceptance — Rev 1.0

12 September 2026. Seven configurations across three proof studies; 21 real CalculiX solves.
These are preliminary numerical studies, not experimental validation or design certification.

| Configuration | Coarse / medium / fine elements | Fine peak MPa | Comparison displacement mm | Numerical | Requirements |
|---|---|---:|---:|---|---|
| bracket | 3150 / 6138 / 10623 | 29.2343 | 0.284587 | PASS | PASS |
| tube | 48 / 108 / 192 | 6.10381 | 0.000835022 | NOT_ESTABLISHED | PASS |
| tube3d | 4532 / 9918 / 17889 | 6.12857 | 0.000834988 | NOT_ESTABLISHED | PASS |
| port | 9311 / 16988 / 26133 | 71.4077 | 0.000877595 | FAIL | NOT_ESTABLISHED |
| thermal-free | 442 / 1844 / 3650 | 1.52738e-11 | 0.115 | PASS | PASS |
| thermal-fixed | 442 / 1844 / 3650 | 80.4992 | 4.49986e-17 | PASS | PASS |
| thermal-gradient | 442 / 1844 / 3650 | 110.293 | 0.009085 | PASS | FAIL |

## First principles and interpretation

### bracket
Analytical expectation: 30 MPa, 0.285714 mm. Euler-Bernoulli gross section; hole and fixed-face local effects excluded.
Measured comparison displacement: 0.284587 mm (transverse tip displacement).
Medium-to-fine displacement change: 0.0647436%; selected stress change: 0.0573696%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### tube
Analytical expectation: 6.10717 MPa, 0.000835065 mm. Lame open-ended tube: sigma_z=0; bore von Mises and radial expansion. No port or end-cap load.
Measured comparison displacement: 0.000835022 mm (bore radial expansion; 3D least-squares midspan fit removes arbitrary rigid translation).
Medium-to-fine displacement change: 4.89554e-05%; selected stress change: 0.0297194%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### tube3d
Analytical expectation: 6.10717 MPa, 0.000835065 mm. Lame open-ended tube: sigma_z=0; bore von Mises and radial expansion. No port or end-cap load.
Measured comparison displacement: 0.000834988 mm (bore radial expansion; 3D least-squares midspan fit removes arbitrary rigid translation).
Medium-to-fine displacement change: 0.0300506%; selected stress change: 0.240446%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### port
Analytical expectation: 6.10717 MPa, 0.000835065 mm. Lame open-ended tube: sigma_z=0; bore von Mises and radial expansion. No port or end-cap load.
Measured comparison displacement: 0.000877595 mm (bore radial expansion; 3D least-squares midspan fit removes arbitrary rigid translation).
Medium-to-fine displacement change: 8.70769%; selected stress change: 55.856%.
Local stress: LOCAL_SINGULARITY_SUSPECTED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### thermal-free
Analytical expectation: 0 MPa, 0.115 mm. Uniaxial thermal strain, axial linear temperature profile; not triaxial constraint.
Measured comparison displacement: 0.115 mm (axial displacement (not vector magnitude)).
Medium-to-fine displacement change: 0%; selected stress change: 0.000292769%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### thermal-fixed
Analytical expectation: 80.5 MPa, 0 mm. Uniaxial thermal strain, axial linear temperature profile; not triaxial constraint.
Measured comparison displacement: 4.49986e-17 mm (axial displacement (not vector magnitude)).
Medium-to-fine displacement change: 0%; selected stress change: 4.99795e-13%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: PRELIMINARY_ONLY. Design readiness: NOT_ESTABLISHED.

### thermal-gradient
Analytical expectation: 104.65 MPa, 0.008625 mm. Uniaxial thermal strain, axial linear temperature profile; not triaxial constraint.
Measured comparison displacement: 0.009085 mm (axial displacement (not vector magnitude)).
Medium-to-fine displacement change: 0.00197983%; selected stress change: 0.208803%.
Local stress: LOCAL_STRESS_REVIEW_REQUIRED. Disposition: REJECT_OR_REDESIGN. Design readiness: NOT_ESTABLISHED.

The side-port case is deliberately NOT accepted: peak stress and displacement do not meet the mesh-sensitivity criteria. It demonstrates escalation and refusal, not a successful manifold design. Its cut-port surfaces are unpressurized and require a revised physical load/support model before design use.
The gradient thermal case demonstrates numerical PASS with engineering FAIL against a 100 MPa limit. Pressure cases retain missing reaction-equilibrium evidence as NOT_EVALUATED. Free-expansion roundoff stress is not treated as a singularity.

## Software, tooling and security

CalculiX: This is Version 2.20; Gmsh: 4.15.2; CadQuery: 2.7.0; OpenCascade/OCP: 7.8.1.1.
All solve artifacts were SHA-256 checked before publication. No image, Docker control, host paths, raw host logs or private credentials are projected.
Worker: 10001:10001, network none, read-only root, dropped capabilities, no-new-privileges, 2 CPU / 2 GiB / 64 PID limits, fixed executables and disposable scratch. Controller timeouts remove only their own named container.
12 semantic MCP-compatible tools cover classification, fidelity, case proposals, geometry, mesh, preflight, submission, status, results, evidence, comparison and escalation. Human approval is a separate non-MCP endpoint with a different credential.
FEA reuses the existing engineering workbench and scientific viewport. Public examples are recorded; public chat does not acquire a solver tool. The private gateway and existing-app registration are implemented; production supervision and private-profile deployment/configuration remain operational setup, not something these proof runs establish.

## Capability answers

- A. Analytical instead of FEA: yes, including a solver-free executable analytical path.
- B. Axisymmetric instead of 3D: yes, demonstrated by the open-ended pressure tube.
- C. Escalate to 3D: yes; the side port invalidates axisymmetry and requires a new approved candidate.
- D. Strategy without arbitrary execution authority: yes; strict schemas and fixed worker commands.
- E. Missing structural evidence as PASS: no; missing required evidence blocks acceptance.
- F. Numerical versus engineering satisfaction: separate gates; demonstrated by the gradient thermal case.
- G. Future CFD provenance: strict coupling envelope/validators exist; actual mapping and coupled execution remain disabled.
- H. Same workbench/evidence philosophy: yes; common saved cases, stages and scientific projection.

## Deferred / not claimed

Beam, shell and plane models are selectable planning recommendations, not executable mesh/solver adapters in this revision. Modal, body-force templates, buckling, nonlinear/contact/transient/plasticity, arbitrary imported CAD/temperature fields, validated CFD field transfer, probe and section planes remain deferred.
This is a bounded first structural release, not completion of every future mode named in the roadmap.

## Source and verification

Core files: Aero/fea/bounded/{contracts,worker,controller,evidence,tools,routes,server,integration}.py; worker Dockerfiles; test_bounded_fea.py; acceptance and packaging runners; AERO_FEA_PIPELINE.md; existing app/workbench registrations.
UI files: existing pressure-viewer.ts extended for structural fields; structural-viewer.ts; aero-structural-cases.ts; current.astro; conversation gateway discipline context.
Hermetic tests are separate from live integration. Existing cooling/study and conversation regressions are also rerun. See test-receipt.json alongside this document for the final counts.
Primary references: https://www.dhondt.de/ and https://gmsh.info/doc/texinfo/gmsh.html.