Cashen/optiwing-airfoil-2d-surface-v1

OptiWing Airfoil 2D — surface fields

As the basic dataset, plus the per-node surface fields (pressure, velocity, skin friction, …) needed to reconstruct the primal flow state on the airfoil surface.

This is the 2D OptiWing aerodynamic shape-optimization dataset, formatted to match the EngiBench airfoil convention (one row per optimization case). The three OptiWing-2D datasets share case_id, so a row in one joins to the same case in the others. All three are built from a single raw pass, so their geometry and fields are identical where they overlap.

Geometry is the native CFD surface contour at physical chord (c ≈ 0.98): coords are (x, y) at the solver's surface nodes (~200 per section, ordered TE → upper → LE → lower → TE), and every surface field is the unmodified solver value at those same nodes — no resampling, no normalization. The contour re-meshes to reproduce the simulation, and each field value corresponds exactly to its coordinate. For a one-to-one comparison with the 192-point 3D dataset, resample onto a common grid. The 2D area is the analogue of the 3D paper's volume constraint; area_ratio is the achieved area / initial area.

Splits

split	examples
train	878
val	49
test	107

Splits are at the case level and shared with the OptiWing 3D companion data, so paired 2D/3D cases land in the same split.

Features

field	type	description
case_id	int	optimization case id (shared across all three datasets)
mach	float	freestream Mach number (sampled input condition)
reynolds	float	Reynolds number (sampled input condition)
temperature	float	freestream static temperature [K] (sampled input condition)
cl_target	float	target lift coefficient (input condition / constraint)
area_ratio_min	float	minimum allowable area ratio (input constraint)
area_initial	float	chord-normalized area of the initial section (shoelace)
cd, cl, cl_con_violation, area_ratio	float	optimized objectives / constraints
initial_design, optimal_design	struct	geometry and per-node surface fields (below)

Each design struct holds values at the native CFD surface nodes (~200 points, ordered TE→upper→LE→lower→TE; arc ∈ [0,1] is the per-node arc-length fraction). Coordinates and fields are the unmodified solver values at the same nodes, so each field value corresponds exactly to its coordinate.

sub-field	description
angle_of_attack	section angle of attack [deg]
coords	(x, y) pairs, physical chord ≈0.98 (~200×2)
arc	per-node arc-length fraction (~200)
cp	pressure coefficient (~200)
pressure, density, temperature	primal thermodynamic state at the wall (~200 each)
velocity_x, velocity_y	in-plane velocity components (~200 each)
cf_x, cf_y	in-plane skin-friction coefficient (~200 each)
yplus	wall y+ (~200)

Loading

from datasets import load_dataset
ds = load_dataset("Cashen/optiwing-airfoil-2d-surface-v1")
print(ds["train"][0].keys())

Citation

This dataset does not yet have a companion paper or DOI. Until one exists, please cite the dataset directly by its Hugging Face URL and cite the prior-version works in "Built upon".

@misc{optiwing_airfoil_2d_2026_surface,
  title        = {OptiWing Airfoil 2D (surface): a 2D aerodynamic shape-optimization dataset},
  author       = {Cashen Diniz and Mark Fuge},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {Hugging Face Datasets},
  url          = {https://huggingface.co/datasets/Cashen/optiwing-airfoil-2d-surface-v1},
  note         = {Version v0}
}

Built upon

This is a new dataset (wider parameter bounds, full surface fields, and complete optimization trajectories), but its lineage and format derive from prior work — please cite these as well:

@misc{diniz2025optiwing3d,
  title         = {OptiWing3D: A Diverse Dataset of Optimized Wing Designs},
  author        = {Diniz, Cashen and Fuge, Mark D.},
  year          = {2025},
  eprint        = {2512.12867},
  archivePrefix = {arXiv},
  doi           = {10.48550/arXiv.2512.12867},
  url           = {https://arxiv.org/abs/2512.12867}
}

@inproceedings{felten2025engibench,
  title     = {EngiBench: A Framework for Data-Driven Engineering Design Research},
  author    = {Felten, Florian and Apaza, Gabriel and Br\"{a}unlich, Gerhard and
               Diniz, Cashen and Dong, Xuliang and Drake, Arthur and Habibi, Milad and
               Hoffman, Nathaniel J. and Keeler, Matthew and Massoudi, Soheyl and
               VanGessel, Francis G. and Fuge, Mark},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS),
               Datasets and Benchmarks Track},
  year      = {2025},
  eprint    = {2508.00831},
  archivePrefix = {arXiv},
  url       = {https://arxiv.org/abs/2508.00831}
}

The EngiBench airfoil format this dataset mirrors is airfoil_v0: https://huggingface.co/datasets/IDEALLab/airfoil_v0.

License

cc-by-nc-sa-4.0