Multi-Resolution CEV CFD Database
Computational fluid dynamics dataset for a Crew Exploration Vehicle (CEV) geometry at hypersonic conditions. Each sample is a steady-state solution on a sequence of four successively refined unstructured meshes produced by an anisotropic mesh-adaptation pipeline. The database spans 164 samples across a range of freestream Mach numbers and angles of attack.
Contents
| Samples | 164 |
| Cycles per sample | 4 (cycle0 – cycle3) |
| Mesh | Unstructured, tetrahedral, 3-D |
| Solver fields | 5 (temperature, Mach, density, pressure, velocity) |
| Adaptation fields | 4 (shock flags, velocity gradient, size tensor) |
Layout
metadata/
samples.parquet one row per sample (inflow conditions)
cycles.parquet one row per (sample_id, cycle): file presence flags
fields.json field names, units, components, descriptions
data/
NNNN/ four-digit zero-padded sample ID
cycle0/ … cycle3/ successively refined adaptation cycles
volume.vtu flow solution on the cycle mesh (solver fields)
supp.vtu adaptation fields on the same mesh
surface.vtm multiblock boundary surfaces (VTK composite)
Samples
Samples are drawn from the following inflow parameter space:
| column | units | description |
|---|---|---|
sample_id |
Four-digit ID; matches the data/NNNN/ folder |
|
mach |
Freestream Mach number | |
aoa |
deg | Angle of attack |
Adaptation cycles
cycleK holds the solution on the mesh after K adaptations; cycle0 is the starter mesh. Minimum mesh size in adaptation cycle N is h_min = 0.006 / (4 · 2^(N-1)), maximum size h_max = 0.2, and maximum requested aspect ratio 32. The adaptation is driven by the static temperature Hessian, modified at shocks flagged from the temperature and pressure fields.
| cycle | description |
|---|---|
| 0 | Starter mesh solution |
| 1 | First adapted mesh |
| 2 | Second adapted mesh |
| 3 | Third (finest) adapted mesh |
The adaptation fields in supp.vtu for cycle K were used to generate the mesh
for cycle K+1.
Fields
volume.vtu
Flow-solution fields on the adapted mesh. Arrays are named Field_N where N
matches the gc.register_field_data() ID in the solver source.
| field | label | components | assoc. | units |
|---|---|---|---|---|
Field_0 |
temperature | 1 | cell | K |
Field_20 |
Mach_number | 1 | cell | |
Field_21 |
density | 1 | cell | kg/m³ |
Field_22 |
pressure | 1 | cell | Pa |
Field_24 |
velocity | 3 | cell | m/s |
supp.vtu
Mesh-adaptation fields computed by the adaptation pipeline. Arrays are named
Field_N using the same ID convention.
| field | label | components | assoc. | description |
|---|---|---|---|---|
Field_5 |
temperature_flagField | 1 | point | Shock flag (temperature) |
Field_10 |
pressure_flagField | 1 | point | Shock flag (pressure) |
Field_11 |
velocity_gradient | 9 | point | Velocity gradient tensor (row-major) |
Field_12 |
size_field | 9 | point | Anisotropic sizing tensor (row-major) |
The size_field (Field_12) is a 3×3 symmetric positive-definite tensor stored
in row-major order. It encodes the target edge lengths and principal directions
for the next adaptation cycle. Each vector indicates a direction, and the magnitude
of each vector indicates requested mesh size in that direction.
surface.vtm
Multiblock VTK composite file. Each block is a boundary surface patch
(surface/<name>.vtu files are referenced by the .vtm index). Solver fields
are interpolated to surface nodes.
Loading
from huggingface_hub import snapshot_download
import vtk
from vtk.util.numpy_support import vtk_to_numpy
# Download one sample (cycle 1 only) — avoids fetching the full dataset
local = snapshot_download(
repo_id="sahni-lab/multi-res-cev-cfd-database",
repo_type="dataset",
allow_patterns=["metadata/*", "data/0000/cycle1/*"],
)
# Read volume.vtu with VTK
reader = vtk.vtkXMLUnstructuredGridReader()
reader.SetFileName(f"{local}/data/0000/cycle1/volume.vtu")
reader.Update()
grid = reader.GetOutput()
# Extract Mach number (cell data, Field_20)
arr = vtk_to_numpy(grid.GetCellData().GetArray("Field_20"))
print("Mach range:", arr.min(), arr.max())
Or with pandas for the metadata only:
import pandas as pd
from huggingface_hub import hf_hub_download
samples = pd.read_parquet(
hf_hub_download("sahni-lab/multi-res-cev-cfd-database",
"metadata/samples.parquet", repo_type="dataset")
)
print(samples.head())
Citation
A citation will be added once the accompanying paper is published.
- Downloads last month
- 27