Full-field displacements and strains obtained by digital image correlation during fatigue crack growth experiments
- 1. German Aerospace Center (DLR)
- 1. German Aerospace Center (DLR)
Description
This data publication contains full-field displacements and strains obtained by 3D digital image correlation (DIC) using a GOM Aramis 12M system including three fatigue crack propagation (fcp) experiments of AA2024-T3 aluminium sheet material.
The repository consists of three datasets of different experiments named
- S950,1.6 - MT950 specimen, 1.6 mm sheet thickness, load ratios R=0.3, 1.0
- S160,2.0 - MT160 specimen, 2.0 mm sheet thickness, load ratios R=0.1, 0.25, 0.5, 0.75, 1.0
- S160,4.7 - MT160 specimen, 4.7 mm sheet thickness, load ratios R=0.1, 0.25, 0.5, 0.75, 1.0
where Sw,t denotes a middle tension (MT) specimen with width w and thickness t. The nodal DIC measurements at different times during the experiments are provided as .txt files we call "nodemaps" and stored in subfolders "Nodemaps". Each nodemap consists of a header containing meta data information like a running number (current stage index) or the applied force, followed by the nodal displacements and strains in tabular form. Additionally, the dataset S160,4.7 contains crack path and crack tip labels for each nodemap in the subfolder "GroundTruth". The ground truth is provided as arrays of size 256x256. Each pixel of the array contains the label "2" for the class "crack tip", "1" for the class "crack path", or "0" for the class "background". These labels were created in a semi-manual fashion and can be used for machine learned crack detection using supervised training.
These datasets were recently used to evaluate neural attention of convolutional neural networks trained on fatigue crack tip detection in Melching et al. (Sci Rep, 2022). Additional guidance on data loading and usage can also be found at https://github.com/dlr-wf/explainable-crack-tip-detection.
The experiments S160,2.0 and S160,4.7 were conducted and analysed by Strohmann et al. (FFEMS, 2021).
The experiment S950,1.6 was conducted and analysed by Breitbarth et al. (FFEMS, 2020).
Files
S_160_2.0.zip
Additional details
Related works
- Is cited by
- Journal article: 10.1038/s41598-022-13275-1 (DOI)
- Software: 10.5281/zenodo.6605441 (DOI)
- Software: https://gitlab.com/dlr-wf/explainable-crack-tip-detection (URL)
References
- Melching et al. (2022) Explainable machine learning for precise fatigue crack tip detection. doi:10.1038/s41598-022-13275-1
- Strohmann et al. (2021) Automatic detection of fatigue crack paths using digital image correlation and convolutional neural networks. doi:10.1111/ffe.13433
- Breitbarth et al. (2020) High-stress fatigue crack propagation in thin AA2024-T3 sheet material. doi:10.1111/ffe.13335