Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms
Abstract Lightweight aluminum alloy is one of the widely used structural materials for various industries due to its low density, high strength-to-weight ratio, good corrosion resistance, and excellent recyclability. However, complex service conditions often result in material degradation due to sim...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | npj Materials Degradation |
| Online Access: | https://doi.org/10.1038/s41529-024-00549-4 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832585472818282496 |
|---|---|
| author | Yucong Gu Kaiwen Wang Zhengyu Zhang Yi Yao Ziming Xin Wenjun Cai Lin Li |
| author_facet | Yucong Gu Kaiwen Wang Zhengyu Zhang Yi Yao Ziming Xin Wenjun Cai Lin Li |
| author_sort | Yucong Gu |
| collection | DOAJ |
| description | Abstract Lightweight aluminum alloy is one of the widely used structural materials for various industries due to its low density, high strength-to-weight ratio, good corrosion resistance, and excellent recyclability. However, complex service conditions often result in material degradation due to simultaneous mechanical and corrosion attacks on the metal surfaces, such as tribocorrosion. This phenomenon represents a complex multiphysics challenge, wherein the tribocorrosion-induced material loss emerges as a function of varied environmental, mechanical, and electrochemical descriptors, each entailing distinct yet interlinked physical processes. The pursuit of simultaneous optimization across multiple material properties to enhance the overall tribocorrosion resistance is hampered by the inherent trade-offs between wear and corrosion resistance. Addressing this complexity, our study develops a novel methodology fusing machine-learning (ML) and genetic algorithm (GA)-based optimization techniques to tailor aluminum-based alloys for enhanced tribocorrosion resistance. Leveraging an experimentally validated multiphysics finite element analysis (FEA) model, we have used six key material parameters to model the tribocorrosion performance of Al alloys over a large property space. The ML model employs an ensemble method of artificial neural networks (ANNs) to predict the tribocorroded surface profile and total material loss based on FEA simulation results, significantly reducing computational time compared to conventional FEA methods. Crucially, our high-throughput screening pinpoints corrosion current density and yield strength as two pivotal parameters influencing tribocorrosion behavior. Harnessing GA optimization alongside the ML model, we efficiently identify a suite of optimal material properties—encompassing both mechanical and electrochemical aspects—for aluminum alloys, resulting in superior tribocorrosion resistance. This selection is substantiated through validation against high-fidelity FEA simulation results. This data-driven framework holds promise for tailoring tribocorrosion-resistant materials beyond aluminum alloys, adaptable to a wide range of metals and service environments. |
| format | Article |
| id | doaj-art-09aa0ddd95ab418aa256c2a78279d43a |
| institution | Kabale University |
| issn | 2397-2106 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Materials Degradation |
| spelling | doaj-art-09aa0ddd95ab418aa256c2a78279d43a2025-01-26T12:45:56ZengNature Portfolionpj Materials Degradation2397-21062025-01-019111510.1038/s41529-024-00549-4Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithmsYucong Gu0Kaiwen Wang1Zhengyu Zhang2Yi Yao3Ziming Xin4Wenjun Cai5Lin Li6School for Engineering of Matter, Transport and Energy, Arizona State UniversityDepartment of Materials Science and Engineering, Virginia Polytechnic Institute and State UniversityDepartment of Materials Science and Engineering, Virginia Polytechnic Institute and State UniversitySchool for Engineering of Matter, Transport and Energy, Arizona State UniversityDepartment of Materials Science and Engineering, Virginia Polytechnic Institute and State UniversityDepartment of Materials Science and Engineering, Virginia Polytechnic Institute and State UniversitySchool for Engineering of Matter, Transport and Energy, Arizona State UniversityAbstract Lightweight aluminum alloy is one of the widely used structural materials for various industries due to its low density, high strength-to-weight ratio, good corrosion resistance, and excellent recyclability. However, complex service conditions often result in material degradation due to simultaneous mechanical and corrosion attacks on the metal surfaces, such as tribocorrosion. This phenomenon represents a complex multiphysics challenge, wherein the tribocorrosion-induced material loss emerges as a function of varied environmental, mechanical, and electrochemical descriptors, each entailing distinct yet interlinked physical processes. The pursuit of simultaneous optimization across multiple material properties to enhance the overall tribocorrosion resistance is hampered by the inherent trade-offs between wear and corrosion resistance. Addressing this complexity, our study develops a novel methodology fusing machine-learning (ML) and genetic algorithm (GA)-based optimization techniques to tailor aluminum-based alloys for enhanced tribocorrosion resistance. Leveraging an experimentally validated multiphysics finite element analysis (FEA) model, we have used six key material parameters to model the tribocorrosion performance of Al alloys over a large property space. The ML model employs an ensemble method of artificial neural networks (ANNs) to predict the tribocorroded surface profile and total material loss based on FEA simulation results, significantly reducing computational time compared to conventional FEA methods. Crucially, our high-throughput screening pinpoints corrosion current density and yield strength as two pivotal parameters influencing tribocorrosion behavior. Harnessing GA optimization alongside the ML model, we efficiently identify a suite of optimal material properties—encompassing both mechanical and electrochemical aspects—for aluminum alloys, resulting in superior tribocorrosion resistance. This selection is substantiated through validation against high-fidelity FEA simulation results. This data-driven framework holds promise for tailoring tribocorrosion-resistant materials beyond aluminum alloys, adaptable to a wide range of metals and service environments.https://doi.org/10.1038/s41529-024-00549-4 |
| spellingShingle | Yucong Gu Kaiwen Wang Zhengyu Zhang Yi Yao Ziming Xin Wenjun Cai Lin Li Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms npj Materials Degradation |
| title | Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| title_full | Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| title_fullStr | Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| title_full_unstemmed | Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| title_short | Accelerating the design and discovery of tribocorrosion-resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| title_sort | accelerating the design and discovery of tribocorrosion resistant metals by interfacing multiphysics modeling with machine learning and genetic algorithms |
| url | https://doi.org/10.1038/s41529-024-00549-4 |
| work_keys_str_mv | AT yuconggu acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT kaiwenwang acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT zhengyuzhang acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT yiyao acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT zimingxin acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT wenjuncai acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms AT linli acceleratingthedesignanddiscoveryoftribocorrosionresistantmetalsbyinterfacingmultiphysicsmodelingwithmachinelearningandgeneticalgorithms |