Certification Report — AeroForge v1.0.0

This report states honestly what AeroForge is, what was actually executed and verified, and what is explicitly a seam. The authoritative certification state, Evidence Grade, and Trust Score are assigned by the Forge Proof Layer (proof/PROOF_DECISION.json) — not by this document.

What is claimed

1. The compressible-flow relations (isentropic, normal/oblique shock,

Prandtl-Meyer) reproduce NACA Report 1135 tabulated values to within a max relative error of 3.5e-5 across 37 reference points.

1. The atmosphere model reproduces U.S. Standard Atmosphere 1976 T and p to

within 0.5 % at 0/11/20/32 km.

1. The watertight panelization satisfies the closed-surface identity.
2. The (modified) Newtonian aero method produces physically correct trends

(monotonic CN(α); L/D peak near α≈20°; CD0 with a transonic rise) and the correct hypersonic Cp,max limit.

1. The toolkit runs as one automated workflow with a common JSON interface

across Python, Node, MATLAB, and shell.

Each claim maps to a verification check (verify.py) and to machine-readable evidence (proof/CLAIM_EVIDENCE.json, delivery-package/evidence/).

What was actually executed

• verify.py — 59/59 checks PASS (see verification-report.json).
• scripts/run_workflow.py — produced aerodb_ship.json, trajectory.json,

postflight_reconciliation.json, workflow-summary.md, and six SVG figures.

• scripts/aeroforge.mjs (Node) — exercised against the CLI/JSON contract.

Disclosed seams (DISCLOSED_SEAM)

These are real limitations, stated up front — not hidden in footnotes:

1. High-fidelity CFD is interfaced but not executed. AeroForge is a

*lower-order, engineering-level* tool (Newtonian / perfect-gas / analytic). It is meant to bracket and feed FUN3D / Loci-CHEM / DPLR and ParaView/Tecplot post-processing — it does not run a Navier-Stokes solver, an unstructured mesh generator, or a field visualizer.

1. Post-flight data is synthetic. The reconciliation uses representative,

model-derived "measured" data, not real flight telemetry.

1. Perfect-gas, inviscid physics. Real-gas chemistry, viscous boundary

layers, laminar-turbulent transition, and ablation are not modeled. The Newtonian method captures pressure-dominated hypersonic loads, not skin friction or base flow.

1. Simplified geometry. The vehicles are axisymmetric sphere-cone stand-ins

at the public 9 m stack scale — not the actual finned, control-surfaced, belly-first Starship/Super Heavy OML.

1. Analytic paneling, not CAD. "Watertight" here means an analytic

surface-of-revolution panelization that passes the closure identity, not a CAD kernel or a flight-quality unstructured grid.

1. MATLAB reader not executed. matlab/load_aerodb.m is provided against the

shared JSON schema but is not run in the pipeline (no MATLAB runtime).

Forbidden-language note

This deliverable avoids unproven marketing language. AeroForge is not described as production-ready, flight-validated, or a CFD replacement. Its absolute coefficient values must be anchored by CFD and/or wind-tunnel/flight data before any design use; only the *reference-validated relations* and the *qualitative trends* are asserted here.