Wing Optimizer
A Python tool that searches wing geometry and airfoil pairs, evaluates them in Flow5, and refines a shortlist against a three-speed mission.
What it does
Set mass, airspeeds, geometry bounds, airfoil coverage, and optional pitching-moment preferences. The search tests candidates with Flow5, learns from completed evaluations, and promotes promising designs to a finer mesh.
The two saved flying-wing searches used 0.70 kg, 10 / 20 / 30 m/s, and an extra drag coefficient of 0.012. Mission L/D weights are 25% / 55% / 20%. Shortlist order also includes pitch preferences.
Engineering workflow
What has been measured
An independent Flow5 VLM2 check used the UIUC rectangular AR = 3 wing at Re = 80,000. Seventeen points from −4° to 4° were paired with digitized wind-tunnel curves.
| Mesh | CL RMSE | Cm RMSE |
|---|---|---|
| 20 × 12 | 0.01297–0.02196 | 0.00329–0.00345 |
| 32 × 18 | 0.01373–0.02246 | 0.00326–0.00341 |
| 48 × 24 | 0.01407–0.02269 | 0.00331–0.00347 |
Predicted lift slope: 0.0554–0.0560 per degree; measured: 0.0590–0.0593. This benchmark is inviscid: it does not validate total drag, L/D, or either flying-wing winner. Reference curves were digitized; the test geometry was reconstructed.
The run audit checked 7,321 raw-output hashes and 34 native-project hashes. Lift, drag allowance, and L/D arithmetic matched saved values. These checks establish data consistency, not flight accuracy.
How the runs compare
| Design / source | Geometry and conditions | Reported L/D |
|---|---|---|
| 15 Oct · 000521 | 0.80 m span · 35° sweep · 0.70 kg · cruise 20 m/s | 8.59 cruise, predicted |
| 17 Oct · 000858 | 0.80 m span · 19.25° sweep · 0.70 kg · cruise 20 m/s | 7.96 cruise, predicted |
| ETERNITY · Bronz et al., 2013 ↗ | 1.00 m span · 0.145 m² · 0.93 kg · complete conventional aircraft | 13.9 maximum glide, measured |
| Flying-wing MAV · Shams et al., 2020 ↗ | 0.50 m span · 40° sweep · tunnel at 20 m/s | 17 peak, measured |
The saved cruise points include extra CD = 0.012 and use fixed controls. Published peak values describe different wings, Reynolds numbers, and trim conditions. They provide design context; their differences cannot be read as prediction errors.
Inspect the two winners
Lofts reconstructed from the saved dimensions and exported root/tip airfoil coordinates. These are engineering geometry, not AI-generated models. The loft is not the exact Flow5 analysis mesh; no pressure field is implied.
Results and geometry
Select a saved run and candidate. The planform and operating-point table update together.
Selected candidate
| Point | Speed (m/s) | L/D | Drag (N) | Cm | α (°) |
|---|
| Rank / select | Root / tip | Span (m) | Mission L/D | Cruise L/D |
|---|
Two runs, shown side by side.
Mission L/D · JX-GS-15 / MH30
Cruise L/D 8.59 · Cm −0.0317
Mission L/D · E171 / RG15
Cruise L/D 7.96 · Cm −0.0242
The mission mass and speeds match, but other requirements and scoring settings changed. These are recorded runs, not a controlled benchmark of algorithm improvement.
Mission L/D includes the declared extra drag coefficient of 0.012. Mission weights are 25% slow, 55% cruise, and 20% fast. The displayed winners have saved applied-settings, viscous-drag, and numerical-agreement checks.
Physical elevon trim and flight readiness are not verified. Both searches ended at a 14-hour limit; neither establishes a global optimum or general aerodynamic accuracy bound. Raw wing-only polars can show higher L/D because they omit the extra drag allowance.
Project record
The app is at version 2.6.1. The results above preserve the original 15 and 17 October runs.
Result downloads contain mission settings, candidate geometry, polars, and source-summary hashes. Private chat transcripts and local paths are excluded.