I’m a mechanical engineering undergraduate in my final year.
My final year project is to predict steel alloys’ composition/category from their mechanical properties using machine learning. For this, I needed a database that contains different steel alloys’ chemical compositions (or steel alloys’ category, i.e., Carbon Steel, or low-alloy steel, etc.), as well as their mechanical properties to train the machine learning model.
After some research, I came across your dataset (matbench_steels), and it seemed like the perfect match. I was just hoping to know if there is a dataset which contains other mechanical properties alongside the steels’ yield strength (for example: Hardness, toughness, young’s modulus, shear modulus, bulk modulus, poison ratio)
Inquiry on matbench_steels Dataset Scope for Machine Learning-Based Steel Alloy Design
I am a final-year undergraduate student in Mechanical Engineering developing a machine learning model for my thesis project. The objective is to predict steel alloy composition/category (e.g., carbon steel, low-alloy steel) based on mechanical properties through inverse design methodology.
Your dataset matbench_steels appears highly relevant to this work. According to the documentation, it includes:
Chemical composition
Yield strength (Rp0.2)
Tensile strength (Rm)
Elongation (A%)
To enhance model robustness, I seek confirmation if the dataset also contains any of the following additional mechanical properties:
Hardness (e.g., Vickers, Brinell, Rockwell)
Fracture toughness (KIC)
Elastic moduli:
Young’s modulus (E)
Shear modulus (G)
Bulk modulus (K)
Poisson’s ratio (ν) (Note: Corrected from “toxicity ratio” based on engineering context)
These properties are critical for establishing multi-property design constraints in alloy optimization. If available, please specify:
Measurement standards (e.g., ASTM E8, ISO 6892-1)
Data completeness (% of samples with these properties)
Should such extensions not exist in the current dataset, I would appreciate guidance on potential complementary databases (e.g., ASM Handbooks, NIMS MatNavi).
Thank you for your expertise. This dataset will significantly advance my work in ML-driven materials design.