3D Wing Aerodynamic Multi-Fidelity Foundation Model

Single AI model for a wide, 20-dimensional wing design space — predicting C_L, C_D, C_M in milliseconds.

20Design & flow inputs

4,512RANS CFD cases859 high-fidelity
3,653 low-fidelity

Interactive 3D view

DragRotate Scroll / PinchZoom Right-dragPan

AI Predictions: AoA Sweep

C_L, C_D, and C_M across the full −3° to 9° range for this geometry.

LF-onlyHF-onlyMulti-fidelity

Models. Optuna-tuned MLPs (40 trials each). MF = frozen LF surrogate f_L(x) + HF correction net g([x, f_L]) with residual skip.

HF-test R²	C_L	C_D	C_M
LF-only	0.933	0.953	0.932
HF-only	0.907	0.917	0.840
MF	0.953	0.949	0.927

Surrogate predictions, not CFD. Accuracy drops outside the sampled design space and in regions where CFD often failed (high sweep with high AR).