Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning
Abstract Fast prediction of microstructural responses based on realistic material topology is vital for linking process, structure, and properties. This work presents a digital framework for metallic materials using microscale features. We explore deep learning for two primary goals: (1) segmenting...
Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-08-01
|
| Series: | npj Computational Materials |
| Online Access: | https://doi.org/10.1038/s41524-025-01718-y |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849343327288164352 |
|---|---|
| author | Shahed Rezaei Kianoosh Taghikhani Alexandre Viardin Reza Najian Asl Ali Harandi Nikhil Vijay Jagtap David Bailly Hannah Naber Alexander Gramlich Tim Brepols Mustapha Abouridouane Ulrich Krupp Thomas Bergs Markus Apel |
| author_facet | Shahed Rezaei Kianoosh Taghikhani Alexandre Viardin Reza Najian Asl Ali Harandi Nikhil Vijay Jagtap David Bailly Hannah Naber Alexander Gramlich Tim Brepols Mustapha Abouridouane Ulrich Krupp Thomas Bergs Markus Apel |
| author_sort | Shahed Rezaei |
| collection | DOAJ |
| description | Abstract Fast prediction of microstructural responses based on realistic material topology is vital for linking process, structure, and properties. This work presents a digital framework for metallic materials using microscale features. We explore deep learning for two primary goals: (1) segmenting experimental images to extract microstructural topology, translated into spatial property distributions; and (2) learning mappings from digital microstructures to mechanical fields using physics-informed operator learning. Loss functions are formulated using discretized weak or strong forms, and boundary conditions-Dirichlet and periodic-are embedded in the network. Input space is reduced to focus on key features of 2D and 3D materials, and generalization to varying loads and input topologies are demonstrated. Compared to FEM and FFT solvers, our models yield errors under 1–5% for averaged quantities and are over 1000× faster during 3D inference. |
| format | Article |
| id | doaj-art-a61d3333d54a4a8b9a2968dee35e4c0b |
| institution | Kabale University |
| issn | 2057-3960 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Computational Materials |
| spelling | doaj-art-a61d3333d54a4a8b9a2968dee35e4c0b2025-08-20T03:43:01ZengNature Portfolionpj Computational Materials2057-39602025-08-0111111810.1038/s41524-025-01718-yDigitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learningShahed Rezaei0Kianoosh Taghikhani1Alexandre Viardin2Reza Najian Asl3Ali Harandi4Nikhil Vijay Jagtap5David Bailly6Hannah Naber7Alexander Gramlich8Tim Brepols9Mustapha Abouridouane10Ulrich Krupp11Thomas Bergs12Markus Apel13ACCESS e.V.ACCESS e.V.ACCESS e.V.Technical University of MunichInstitute of Applied Mechanics, RWTH Aachen UniversityInstitute for Metal FormingInstitute for Metal FormingSteel InstituteSteel InstituteInstitute of Applied Mechanics, RWTH Aachen UniversityManufacturing Technology InstituteSteel InstituteManufacturing Technology InstituteACCESS e.V.Abstract Fast prediction of microstructural responses based on realistic material topology is vital for linking process, structure, and properties. This work presents a digital framework for metallic materials using microscale features. We explore deep learning for two primary goals: (1) segmenting experimental images to extract microstructural topology, translated into spatial property distributions; and (2) learning mappings from digital microstructures to mechanical fields using physics-informed operator learning. Loss functions are formulated using discretized weak or strong forms, and boundary conditions-Dirichlet and periodic-are embedded in the network. Input space is reduced to focus on key features of 2D and 3D materials, and generalization to varying loads and input topologies are demonstrated. Compared to FEM and FFT solvers, our models yield errors under 1–5% for averaged quantities and are over 1000× faster during 3D inference.https://doi.org/10.1038/s41524-025-01718-y |
| spellingShingle | Shahed Rezaei Kianoosh Taghikhani Alexandre Viardin Reza Najian Asl Ali Harandi Nikhil Vijay Jagtap David Bailly Hannah Naber Alexander Gramlich Tim Brepols Mustapha Abouridouane Ulrich Krupp Thomas Bergs Markus Apel Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning npj Computational Materials |
| title | Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| title_full | Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| title_fullStr | Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| title_full_unstemmed | Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| title_short | Digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| title_sort | digitalizing metallic materials from image segmentation to multiscale solutions via physics informed operator learning |
| url | https://doi.org/10.1038/s41524-025-01718-y |
| work_keys_str_mv | AT shahedrezaei digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT kianooshtaghikhani digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT alexandreviardin digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT rezanajianasl digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT aliharandi digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT nikhilvijayjagtap digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT davidbailly digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT hannahnaber digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT alexandergramlich digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT timbrepols digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT mustaphaabouridouane digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT ulrichkrupp digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT thomasbergs digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning AT markusapel digitalizingmetallicmaterialsfromimagesegmentationtomultiscalesolutionsviaphysicsinformedoperatorlearning |