Fracture-surface images, descriptors and tensile data for PA12 specimens manufactured by selective absorption fusion
Description
Desktop USB-microscope images of all 124 fracture faces from 62 PA12 tensile specimens manufactured by selective absorption fusion (Stratasys H350) in three orientations (XZ, YX, YZ), at seven build angles (0–90°) and three crosshead speeds (1, 10, 100 mm/min). Includes segmentation masks, face and core polygons, fracture-surface descriptors (core area, outside whitening, total whitening), the original tensile records and stress–strain curves, and Python code (reproduce.py) that regenerates every number and figure of the associated article. Supports: "Low-cost fractography links fracture-surface descriptors directly to tensile properties of PA12 specimens manufactured by selective absorption fusion", Additive Manufacturing Letters.
Files
Steps to reproduce
To reproduce: Install Python 3.11 or later. Unzip PA12_AML_Reproducibility.zip. From the PA12_Reproducibility folder, run python -m pip install -r requirements.txt. This installs numpy, scipy, Pillow, matplotlib and xlrd. Run python reproduce.py. It takes about one minute and needs no network access. reproduce.py regenerates the masks, descriptors, mechanical data matching, all fits and tests, Figures 2–7, Supplementary Figures S1–S4 and the Fracture-Surface Atlas. It then checks every pixel count and reported statistic against reference_results/ (relative tolerance 1e-9). It also runs an independent audit that recomputes all values from the per-face data shown in the atlas. A successful run ends by printing "pixel_counts_match": true and "fit_statistics_match": true, and writes reproduction_validation.json. README.txt describes each script and file.