Developing a Machine Learning System for High Throughput Indentation-based Fracture Testing to Assess Hydrogen Embrittlement - College of Engineering - Purdue University

Project Description

Indentation methods are commonly used to assess local mechanical properties because of their speed and minimal sample preparation, making them well suited to input into machine learning approaches to materials discovery. However, fracture and toughness are not commonly assessed with indentation in metallic materials which suffer from hydrogen embrittlement. With the increasing use of hydrogen in the energy and materials processing landscape, addressing this gap is timely, and likely to lead to high visibility results for a researcher that can develop the next “Rockwell” or “Charpy” test. The successful post-doc will create models of novel indentation tip geometries using multiscale approaches to generate regions of tension which can cause fracture, providing a toughness analog when measurable cracking occurs. The candidate will perform experimental testing on electrochemically charged steel systems, develop an automated data collection system which can be integrated to machine learning platforms, characterize microstructural features that may dominate the compare model predictions to experimental data, and carry out these experiments in situ during charging.

Start Date

Spring/Summer 2027

Postdoc Qualifications

Experience with multiple length scale mechanical modeling.
Experience with mechanical testing.
Automation/programming for image collection.

Co-advisors

Marisol Koslowski, Mechanical Engineering
David Bahr, Materials Engineering

Bibliography

https://doi.org/10.1063/1.5109761
https://doi.org/10.1016/j.jmps.2024.106009
https://doi.org/10.1016/j.ijhydene.2024.06.215
https://doi.org/10.1115/1.4051806
https://doi.org/10.1016/j.actamat.2006.02.007