Carnegie Mellon University


May 20, 2021

Holm, NETL Design Heat-Resistant Alloys Using Computer Vision

By Liz Rosevear

Liz Rosevear

Professor of Materials Science and Engineering (MSE) Elizabeth Holm and the National Energy Technology Laboratory (NETL) develop microstructural designs for heat-resistant alloys using computer vision.

Project Title: Computer vision and machine learning making the processing-microstructure-property connection in heat resistant alloys

Abstract: A fundamental tenet of materials science is that Processing generates the microStructure that mediates material Properties – the PSP connection. Given its ability to find relationships in large, complex data sets, machine learning seems tailor-made for exploring PSP connections. In this project, we develop and apply computer vision tools to create quantitative representations of microstructural images and apply machine learning methods to discover correlations between these visual data and property metadata. We develop these tools to be relevant to the performance of heat resistant alloys, specifically 347 stainless steel. Our objectives are:

  • Collect microstructural image data and property metadata for heat resistant alloy systems
  • Develop material-agnostic CV techniques to extract knowledge from microstructural images.
  • Create ML systems to find relationships between microstructures and property metadata.
  • Analyze and interpret the results to discover new PSP connections.

This development is critical in producing reliable electricily supplies while producing fewer emissions and helping meet U.S. decarbonization goals.