Guest Speaker
Title: |
Rethinking materials simulation with machine-learning strategies |
 |
Speaker: |
Dr. Remi Dingreville |
|
Affiliation: |
Sandia National Laboratories |
|
When: |
Thursday, February 20, 2025 at 11:00:00 AM |
|
Where: |
Boggs Building, Room 03-47 |
|
Host: |
Chaitanya Deo | |
Abstract Materials simulations are ubiquitous across many scientific domains from solid mechanics to biology. Traditional materials simulations based on direct numerical solvers (DNS), such as phase-field or crystal plasticity for instance, offer accurate predictions but are generally computationally expensive. In this talk, I will discuss how we can rethink and accelerate such materials simulations by blending and complementing classical solvers with various machine-learning strategies. These strategies encompass the integration of DNS with history-dependent machine-learned solvers such as recurrent neural networks or neural operators to enable accurate extrapolation and efficient time-to-solution predictions of the predicted dynamics, or the use of physical governing equations directly as the building blocks for constructing specialized neural networks. I will discuss some of the trade-offs one need to pay attention to with such strategies, the type of speed one can gain from such strategies, and how we can harness these emerging techniques for materials design and process optimization. Examples illustrating these points include phase-field modeling, crystal plasticity, and homogenization of heterogeneous materials. SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525. SAND2024-14529A |
||
Biography Dr. Rémi Dingreville is a Distinguished Member of the Technical Staff at Sandia National Laboratories and Staff Scientist at the Center for Integrated Nanotechnologies (CINT). He holds a Ph.D. in Mechanical Engineering from the Georgia Institute of Technology in Atlanta GA. With expertise at the intersection of computational materials and data sciences, his work focuses on bridging the gap between atomic and mesoscale models to understand and characterize process-structure-properties for materials reliability. Dr. Dingreville 's research has wide-ranging applications, from understanding the mechanical properties of nanostructured alloyed materials to designing materials for energy storage and conversion. He has published over 130 peer-reviewed articles on these topics. Dr. Dingreville is the recent recipient of the J. Keith Brimacombe Medal (2025), the Sandia’s Employee Recognition Award (2024), and Sandia’s Postdoc Association Distinguished Mentorship Award (2023). |
||
Notes |
Meet the speaker |
|