Computational Coupled Physics

Oak Ridge National Laboratory’s (ORNL’s) Computational Coupled Physics Group aims to advance scientific discovery through the development of scalable, performance-portable computational tools and AI methods that enable breakthroughs in materials and manufacturing research. The group combines high-performance computing (HPC), ML, and materials science to accelerate the design and deployment of next-generation materials, with particular focus on sustainable manufacturing, clean energy technologies, and improved industrial processes that benefit society.

The group’s overall mission focuses on several key themes:

  • Computational methods and AI/ML for materials science
  • Performance and scalability for HPC systems
  • Applications in advanced manufacturing and materials design
  • Sustainability and practical industrial impact
  • Cross-disciplinary integration of computing and materials science

This dynamic research group is advancing the frontiers of materials science and manufacturing through innovative computational methods and AI. This work spans three interconnected areas: advanced manufacturing processes, sustainable materials development, and HPC tools that accelerate scientific discovery.

A major focus for the group is improving metal additive manufacturing (3D printing) by providing methods to better understand and control how materials solidify and form their internal structures. The group has developed sophisticated simulation tools (e.g., ExaCA, Toucan) that can predict how metals will behave during printing, thereby helping manufacturers optimize their processes. The group’s research is particularly relevant for critical applications in automotive and aerospace industries, in which material performance and reliability are essential.

The group is also tackling sustainability challenges by investigating how to better use recycled aluminum in manufacturing. The group’s research shows how impurities in recycled materials affect performance and how advanced manufacturing techniques might help overcome these limitations. This work is especially important for electric vehicles, for which new aluminum alloys are being developed to replace heavier copper components and improve vehicle efficiency.

What sets this group apart is their development of cutting-edge computational tools that harness the power of supercomputers such as ORNL’s Frontier. The group has created innovative software frameworks (e.g., HydraGNN, MDLoader) that use AI to predict material properties and simulate manufacturing processes much faster than traditional methods. Their tools are designed to be performance portable, meaning that they can run efficiently on different types of computer architectures, making them widely accessible to the scientific community.

Looking ahead, the group is focused on bridging the gap between computational prediction and real-world manufacturing. By combining AI, HPC, and materials science, the group is working toward a future in which manufacturers can reliably predict and control material properties, leading to more efficient, sustainable, and innovative products. This work could revolutionize how materials are designed and manufactured—from electric vehicle components to aerospace parts—while supporting the transition to more sustainable manufacturing practices.

Key projects include modeling of advanced manufacturing processes through the Exascale Computing Project ExaAM project (https://www.exascaleproject.org/research-project/exaam, POC: Matt Bement) and our strong coordination with the ORNL Manufacturing Demonstration Facility and its Digital Factory effort (group POC: Matt Rolchigo), the Kokkos performance portability library (https://kokkos.org, POC: Damien Lebrun-Grandie), the ArborX geometric search library (https://github.com/arborx/ArborX, POC: Andrey Prokopenko) , and advanced surrogate modeling through the ORNL AI Initiative (https://www.ornl.gov/ai, group POC: Massimiliano Lupo Pasini).