Real Deep Drawing and Cutting (RDDAC) Dataset - Teaser
Description
A ~174 MB sample of the RDDAC dataset: 18 physical deep-drawing and cutting experiments (one per parameter category) with press force signals, sheet-thickness and oil-film traverses, and high-resolution 3D laser scans, plus the Croissant 1.1 manifest, the process parameter table for all 9,000 experiments, and the dataset documentation PDF.
RDDAC is the experimental counterpart to the DDACS simulation dataset (doi:10.18419/DARUS-4801): same cup geometry, same DP600 steel, same two-stage process. Use it to quantify the simulation-to-reality gap or train models on real process data.
This record is a discovery sample. The full release (9,000 experiments, ~87 GB of lossless HDF5, predefined 7,200/900/900 train/val/test split) is hosted on DaRUS. Please cite the dataset there: doi:10.18419/DARUS-5589.
Quickstart: pip install rddac · Docs: rddac.readthedocs.io · Source: github.com/BaumSebastian/RDDAC
License: data CC BY 4.0, package code MIT.
Layout note: to use this bundle with the rddac package, place sample.zip under data/h5/ next to process_parameters.csv and the manifest. This is the layout that rddac download --small creates. Zenodo stores files flat, so the folder cannot be preserved here.
Methods
9,000 deep-drawing and cutting experiments on DP600 dual-phase steel blanks, forming a modified quadratic cup in two operations (OP10 deep drawing, OP20 cutting) across a full parameter grid of 2 geometries x 3 blankholder forces x 3 lubrication patterns with up to 500 repetitions each. Per experiment: press force signals from four load cells sampled at 300 Hz together with punch temperature and position; a sheet-thickness traverse and an oil-film traverse (area density) on the flat blank; and Keyence LJ-X8400 laser scans (3200 x 2000 px height and luminescence buffers) after each operation, stored as raw sensor data in HDF5. The Croissant 1.1 manifest in this record declares the complete schema.
Files
rddac_documentation.pdf
Additional details
Related works
- Is documented by
- Software documentation: https://rddac.readthedocs.io (URL)
- Is supplement to
- Journal article: 10.1007/s12666-026-03870-5 (DOI)
Dates
- Collected
-
2025-09-01/2025-11-20Measurement campaign
Software
- Repository URL
- https://github.com/BaumSebastian/RDDAC
- Programming language
- Python
- Development Status
- Active