Online Lecture by Prof. Yoshitaka Adachi

Deciphering the “Feelings” of Steel: 35 Years of the Digital Revolution in Microstructure Design

Department of Materials Design Innovation Engineering,
Nagoya University, Japan

---

Title:
Deciphering the “Feelings” of Steel: 35 Years of the Digital Revolution in Microstructure Design.

Abstract:
This lecture reflects on a 35-year journey to “understand the feelings of steel” by transforming visual microstructures into quantifiable digital data. Traditionally, identifying and segmenting complex steel structures was a grueling manual process taking months; however, the integration of deep learning (such as U-NET) has reduced this task to mere minutes.

Key innovations highlighted in the research include:
Mathematical Quantification: Moving beyond basic metrics like grain size, the author employs advanced mathematical frameworks – Topology (Euler characteristics), Differential Geometry (curvature), and Persistent Homology – to rigorously quantify the connectivity and morphology of steel.

AI-Driven Analysis: The use of Convolutional Neural Networks (CNN) for “image regression” allows researchers to directly link microstructural images to mechanical properties, such as tensile strength, without manual feature extraction.

Generative Material Design: By utilizing Generative Adversarial Networks (GANs) and automated 3D observation tools like “Genus_3D,” the research team has drastically accelerated the cycle of data acquisition and virtual material design.

This session was chaired by prof. Hossein Beladi (Ghent University/TU Delft)

---