Bainite Start Prediction – Supplementary Material

This repository contains the supplementary material for the paper:

"Predictive modeling of the bainite start temperature using Bayesian inference"
Bernd Schuscha, Dominik Brandl, Lorenz Romaner, Ernst Kozeschnik, Reinhold Ebner, Aurélie Jacob, Peter Presoly, Daniel Scheiber

📄 DOI: 10.1016/j.actamat.2025.121131

For related datasets used in other work, see:

CFB Steel Datasets Repository
📎 https://github.com/BerndSchuscha/cfb-steel-datasets

Dataset: data_Bs.csv

This dataset includes bainite start temperature (Bs) and the corresponding alloy compositions from various literature sources.

Columns:

• Nr: Sample ID or serial number
• Source: Citation or dataset source
• C / wt.%, Mn / wt.%, ..., Al / wt.%: Chemical composition in weight percent
• Bs: Observed bainite start temperature in °C

Notes:

• All values are assumed to be in equilibrium or experimentally measured under well-defined conditions.
• Use with caution if doing regression — some elements (e.g., V, Co) may be zero or sparse across rows.

📈 Posterior Samples: Model_TDB.csv

Location: Posterior_Samples/Model_TDB.csv

This files contain posterior samples of regression coefficients obtained via Bayesian inference, representing the influence of each element on bainite start temperature.

Columns

• a_i: Coefficients in atomic percent
• C0: Intercept of the model

Each row corresponds to a sample from the posterior distribution, enabling posterior predictive analysis and credible interval estimation.

📄 Supplementary: Supplemental_Material.pdf

• Details on data curation
• Model specification
• Posterior analysis
• Additional figures and tables