PROJECT 01 · BRANCH C · PLACEHOLDER

NACA0012 · U × AoA Full Sweep

The future two-dimensional parameter branch combining lessons from the independent speed and AoA studies into a controlled operating-envelope benchmark.

PYTHON REFERENCE IMPLEMENTATION
/projects/naca0012/speed-aoa
BRANCH C · PURPOSE

Combine speed and AoA only after the two one-dimensional problems are understood

This branch is the first true multi-parameter operating map. It should inherit lessons from both one-dimensional sweeps rather than treating U × α as a brute-force rectangular dataset.

PARAMETER SPACE

Sparse U × α anchors

Start from validated speed and AoA anchors. Add density near regime changes, high uncertainty or demonstrated held-out error.

GENERALIZATION

Hold out combinations

Test unseen (U, α) pairs, not merely speeds or angles already represented elsewhere in training.

FILL-IN ROADMAP

Planned evidence layers

1Merge branch knowledge into U × α design spaceafter Speed + AoA branches
2Lock off-grid 2-D validation combinationsbefore model selection
3Benchmark POD/local-ROM familyglobal basis only if justified
4Benchmark PhysicsNeMo learned representationsame held-out map
5Calibrate uncertainty / OODcompare confidence to real held-out error
6Run adaptive CFD acquisition loopmeasure before / after information gain
EXPECTED USER VIEW

Eventually this becomes the operating-envelope interface

A future interactive surface can expose speed, AoA, predicted flow/load response, nearest training conditions, confidence/OOD state and the recommended next CFD point when support is insufficient.

← Back to current Speed Sweep case study