Trace a requirement through analytical models, geometry, numerical simulation, and reviewable evidence.
Converting engineering requirements into bounded analyses: first-principles screening, CAD/model preparation, and CFD or FEA at an appropriate fidelity. Numerical qualification is assessed separately from solver completion, with analytical or independent reference comparisons where available. Retained artifacts support explicit engineering dispositions; failures and unestablished gates remain visible. Human engineering authority stays separate from software and AI execution.
Aero is not an autonomous engineering authority. Solver completion is not numerical qualification, and numerical qualification is not automatically physical validation. A PASS applies only to the documented problem, assumptions, and acceptance criteria; failed and NOT_ESTABLISHED results remain so. Hashes establish artifact identity and integrity, not scientific correctness. LLM confidence or reasoning is not independent computational evidence.
Future work: local engineering pilot — goal, Luna work plan & review gates →
Watch a request being typed, then structured by Aero into reviewable requirements.
Start with your engineering question. Develop the requirements and working document with Aero.
Open Aero →At 20 mL/s through the 200 mm channel, 2.0 mm is the smallest tested gap that passes. Capability study #01 below tests a different flow and budget.
Try these inputs →| Gap | Pressure drop | 70 Pa target |
|---|---|---|
| 1.5 mm | 142.37 Pa | Exceeds |
| 2.0 mm | 60.06 Pa | Meets |
| 2.5 mm | 30.75 Pa | Meets |
9 OpenFOAM solves · 3 grids per gap
Move the throat radius to update the geometry and a first-principles velocity estimate. This does not run CFD.
Parametric illustration · drag to orbit · scroll to zoom
ṁ = ρAV · A = πr²
Nominal dimensions selected. Recorded CFD remains a separate result.
Illustrative full-revolution profile: inlet 20 mm, exit 24 mm, converging length 45 mm, diverging length 110 mm. This browser-generated solid is not the hashed solver CAD.
Continuity screen: 0.01 kg/s air, 101,325 Pa, 300 K, R = 287.05 J/(kg·K), constant density. No compressible expansion, choking, loss or thrust prediction is implied.
Recorded mesh and pressure come from wall-refinement-v6: a one-degree axisymmetric wedge. Display triangles tessellate its boundary faces; they are not the volume-cell topology.
Each study connects an engineering question to a decision, with its method, checks, and limitations.
Real pipe-reference failures, failed nozzle wall-readiness checks, and rejected imported CAD. Inspect the plots and keep the complete OpenFOAM + CalculiX decks.
25 mL/s · 40 Pa pressure-drop budget
Higher flow and a tighter budget than the interactive example. Nine solves across three grids per gap.
Decision: 2.5 mm is the smallest tested passing gap.Idealized parallel-plate flow. To reproduce this study, set 25 mL/s and a 40 Pa budget; the interactive workspace starts at 20 mL/s and 70 Pa.
100 W · 72 °C wall limit · 2 kPa friction budget
Energy balance and heat-transfer calculations compare three layouts under nominal and adverse inputs.
Decision: two 2 mm passages pass both stated limits.Analytical straight-channel screening, not thermal CFD. The report retains the radial heat-transfer verification and model assumptions.
Bracket, pressure tube and thermal restraint: 21 CalculiX solves, with recorded gate states. Explore in Aero → · Read the acceptance report →
Follow Aero's heat-exchanger evidence from the retained two-stream PCHE numerical failure to a blind, publication-grounded thermal-fluid benchmark and a separate high-temperature transfer case.
Compare three grids and the pressure-drop evidence.
Same run · different reviewExplore what numerical convergence does—and does not—establish.
Analytical screening · no CFDStart with an energy balance and identify missing geometry.
Read the retained run, its assumptions and its open validation gates.
Retained run: wall-refinement-v6. The values below come from its result record.
A plain-language request becomes a typed case specification. Assumptions and missing inputs remain visible.
First-principles checks and the live LaTeX notebook establish what the solver should be compared against.
The current CadQuery/OpenCascade manifest records the parametric nozzle, topology checks, dimensions, and source fingerprint.
OpenFOAM checks each generated refinement level before execution. This proof used 6,048, 9,072, and 12,600 cells.
Each rhoSimpleFoam run retained convergence, conservation, monitor, and cleanup evidence.
Compare grid changes, monitor histories, and reference checks in the retained report.
| Grid | Cells | Pressure drop | Mesh study | Solver result |
|---|---|---|---|---|
| Level 2 | 6,048 | 446.189 Pa | Eligible | PASS |
| Level 3 | 9,072 | 439.529 Pa | Eligible | PASS |
| Level 4 | 12,600 | 439.613 Pa | Eligible | PASS |
Public beta / preliminary analysis · 2026
Selected references connect the project description to its public source, technical notes, and evidence records.
Dated local FEA acceptance study, deterministic unit checks, reviewer failures, charts, PDF and actual solver-field renders.
Open ↗Background on CFD foundations, workflow, and aerospace applications.
Open ↗Typed inputs that keep setup intent and solver configuration inspectable.
Open ↗Preflight results, telemetry, manifests, and validation status retained with each workflow.
Open ↗