Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing
Abstract The phase-field (PF) method is a physics-based computational approach for simulating interfacial morphology. It has been used to model powder melting, rapid solidification, and grain structure evolution in metal additive manufacturing (AM). However, traditional direct numerical simulation (...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2022-09-01
|
| Series: | npj Computational Materials |
| Online Access: | https://doi.org/10.1038/s41524-022-00890-9 |
| _version_ | 1811264897778974720 |
|---|---|
| author | Tianju Xue Zhengtao Gan Shuheng Liao Jian Cao |
| author_facet | Tianju Xue Zhengtao Gan Shuheng Liao Jian Cao |
| author_sort | Tianju Xue |
| collection | DOAJ |
| description | Abstract The phase-field (PF) method is a physics-based computational approach for simulating interfacial morphology. It has been used to model powder melting, rapid solidification, and grain structure evolution in metal additive manufacturing (AM). However, traditional direct numerical simulation (DNS) of the PF method is computationally expensive due to sufficiently small mesh size. Here, a physics-embedded graph network (PEGN) is proposed to leverage an elegant graph representation of the grain structure and embed the classic PF theory into the graph network. By reformulating the classic PF problem as an unsupervised machine learning task on a graph network, PEGN efficiently solves temperature field, liquid/solid phase fraction, and grain orientation variables to minimize a physics-based loss/energy function. The approach is at least 50 times faster than DNS in both CPU and GPU implementation while still capturing key physical features. Hence, PEGN allows to simulate large-scale multi-layer and multi-track AM build effectively. |
| first_indexed | 2024-04-12T20:12:38Z |
| format | Article |
| id | doaj.art-5dfcc057f34e4eb6887bbde27b77f4d6 |
| institution | Directory Open Access Journal |
| issn | 2057-3960 |
| language | English |
| last_indexed | 2024-04-12T20:12:38Z |
| publishDate | 2022-09-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Computational Materials |
| spelling | doaj.art-5dfcc057f34e4eb6887bbde27b77f4d62022-12-22T03:18:12ZengNature Portfolionpj Computational Materials2057-39602022-09-018111310.1038/s41524-022-00890-9Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturingTianju Xue0Zhengtao Gan1Shuheng Liao2Jian Cao3Department of Mechanical Engineering, Northwestern UniversityDepartment of Mechanical Engineering, Northwestern UniversityDepartment of Mechanical Engineering, Northwestern UniversityDepartment of Mechanical Engineering, Northwestern UniversityAbstract The phase-field (PF) method is a physics-based computational approach for simulating interfacial morphology. It has been used to model powder melting, rapid solidification, and grain structure evolution in metal additive manufacturing (AM). However, traditional direct numerical simulation (DNS) of the PF method is computationally expensive due to sufficiently small mesh size. Here, a physics-embedded graph network (PEGN) is proposed to leverage an elegant graph representation of the grain structure and embed the classic PF theory into the graph network. By reformulating the classic PF problem as an unsupervised machine learning task on a graph network, PEGN efficiently solves temperature field, liquid/solid phase fraction, and grain orientation variables to minimize a physics-based loss/energy function. The approach is at least 50 times faster than DNS in both CPU and GPU implementation while still capturing key physical features. Hence, PEGN allows to simulate large-scale multi-layer and multi-track AM build effectively.https://doi.org/10.1038/s41524-022-00890-9 |
| spellingShingle | Tianju Xue Zhengtao Gan Shuheng Liao Jian Cao Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing npj Computational Materials |
| title | Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing |
| title_full | Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing |
| title_fullStr | Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing |
| title_full_unstemmed | Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing |
| title_short | Physics-embedded graph network for accelerating phase-field simulation of microstructure evolution in additive manufacturing |
| title_sort | physics embedded graph network for accelerating phase field simulation of microstructure evolution in additive manufacturing |
| url | https://doi.org/10.1038/s41524-022-00890-9 |
| work_keys_str_mv | AT tianjuxue physicsembeddedgraphnetworkforacceleratingphasefieldsimulationofmicrostructureevolutioninadditivemanufacturing AT zhengtaogan physicsembeddedgraphnetworkforacceleratingphasefieldsimulationofmicrostructureevolutioninadditivemanufacturing AT shuhengliao physicsembeddedgraphnetworkforacceleratingphasefieldsimulationofmicrostructureevolutioninadditivemanufacturing AT jiancao physicsembeddedgraphnetworkforacceleratingphasefieldsimulationofmicrostructureevolutioninadditivemanufacturing |