Description
This repository contains experimental and numerical data from fatigue tests
on welded steel specimens with different conditions (e.g. as-welded, deep-rolled, HFMI).
The data are stored in open, text-based formats (UTF-8 encoded, tab-delimited or CSV)
to ensure long-term accessibility and compatibility with MATLAB, Python, and other tools.
Each file includes human-readable headers and can be directly imported.
Structure
scan_geometry/
3D point cloud data of weld geometries (front/back)
raw/
Original scan data without modifications
processed/
Transformed and post-processed scan data:
- coordinate systems aligned
- weld profiles smoothed
metadata.txt / .csv
General specimen information:
- SpecimenName, Material, Condition, Geometry, Thickness_mm
fatigue_test_data.txt / .csv
Fatigue test results:
- StressRatio, NominalStressRange_MPa, LoadCyclesToFailure,
LoadCyclesToCrack, RunOut, Outlier, FailedNotch, Date
imperfections.txt / .csv
Measured welding imperfections:
- AngularMisalignment_mm, AxialMisalignment_mm
scan_files.txt / .csv
Mapping between specimen names and scan files:
- Raw and processed scans (front/back)
fe_results.txt / .csv
Finite element results per specimen and notch (1–4):
- ClampingStressNominal_MPa
- ClampingStressNotch_MPa
- NotchFactor_TCD
weldtoe_geometry.txt / .csv
Weld toe geometry per specimen and notch (1–4):
- Radius_IM_mm, Radius_LSM_mm, Radius_CM_mm
- Angle_MAX_deg
The values represent averaged weld toe geometry parameters.
Radii and angles are computed as mean values along the weld toe line
for each notch and measurement side, excluding invalid values (NaN).
import_examples/
Example scripts for reading and processing the data in MATLAB and Python
Data Format
- TXT files: tab-separated ("\t")
- CSV files: comma-separated
- Missing values are indicated by "NA"
Example Usage
Example scripts are provided in the folder:
import_examples/
MATLAB:
T = readtable('fatigue_test_data.txt', 'Delimiter', '\t');
Python:
import pandas as pd
df = pd.read_csv('fatigue_test_data.txt', sep='\t')
Notes
- Each row corresponds to one specimen (or one notch for FE/geometry data).
- Specimens are uniquely identified by "SpecimenName".
- FE and weld toe data contain multiple rows per specimen (one per notch).
- Processed scan data have been aligned and smoothed for further analysis.
Reference
This dataset is associated with the following dissertation:
Dänekas, C. (2026).
Influence of deep rolling on the fatigue resistance of butt-welded joints.
Leibniz University Hannover.
Contact
Christian Dänekas
LinkedIn: www.linkedin.com/in/christian-daenekas
Mail Insitute: stahlbau@stahl.uni-hannover.de
Website: https://www.stahlbau.uni-hannover.de/en/institute