Data-driven Multiscale Damage and Failure Prediction

数据驱动的多尺度损坏和故障预测

• 批准号: 1762035
• 负责人: Wing Liu
• 金额: $ 53.7万
• 依托单位: Northwestern University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-07-15 至 2022-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1762035&HistoricalAwards=false
• 关键词: Data; driven; Multiscale; Damage; Failure

Damage and failure of materials is commonplace; the ability to predict damage and subsequent failure in engineered systems is foundational to design, and critically important when failures are expensive and even life-threatening. As manufacturing technologies become more advanced, particularly with the advent of additive manufacturing where nearly any shape or form can be made by local application of material and heat, so too must the methods used to predict the mechanical response of these components. The computational modeling framework in this research will enable a wider application of these advanced manufacturing technologies thorough a rigorous understanding of the material performance of parts made with these methods. An extensive experimental characterization and validation effort will form the basis of this computational framework. As such, this research will promote manufacturing sciences and knowledge for the fields where shape and form considerations outweigh production rate concerns, e.g., in biomedical and aerospace industries. The manufacturing advances enabled by this research will directly benefit the U.S. economy, advance national health, prosperity, and welfare, and secure national defense through technological innovations, e.g., through reduced aircraft fuel consumption from lighter additively manufactured parts. The intersection of domains required for this research, including: manufacturing, mechanical engineering, materials science, and computational sciences, will support interdisciplinary collaboration that can lead to crosscutting improvements in engineering education for the modern age. As part of this project, outreach to high school students will be performed to foster interest in engineering, undergraduate summer interns will be recruited to conduct state-of-the-art research, and specialized graduate student projects will be created related to advanced modeling and simulation.The anticipated outcome of the research is a predictive computational theory for damage and failure of complex, hierarchical materials such as metal alloys. The effort builds on data-driven, reduced order, and multiscale principles under the traditional framework of mechanics with the potential for a transformative new theory. Initially, fundamental characterization experiments (including x-ray tomography and diffraction) will be conducted to understand the relationship between material microstructures and mechanical properties in additively manufactured metals. This information will be used to calibrate micromechanical models, and simulations will be used to populate a database of synthetic microstructures and their mechanical response. From this, a new concurrent multiscale theory based on reduced-order methods will be developed, capable of capturing nonlinearity both in geometric and material response. This method will query the database constructed in the first phase for mechanical information and use that data to predict damage and failure, particularly for metals parts made with additive manufacturing.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

材料的损坏和失效是常见的;预测工程系统中的损坏和后续失效的能力是设计的基础，并且在失效代价高昂甚至危及生命时至关重要。随着制造技术变得越来越先进，特别是随着增材制造的出现，几乎任何形状或形式都可以通过局部应用材料和热量来制造，因此用于预测这些部件的机械响应的方法也必须如此。本研究中的计算建模框架将使这些先进制造技术的更广泛的应用，通过严格的理解与这些方法制成的零件的材料性能。广泛的实验表征和验证工作将形成这个计算框架的基础。因此，这项研究将促进制造科学和知识的领域，形状和形式的考虑超过生产率的关注，例如，在生物医学和航空航天工业中。这项研究所带来的制造业进步将直接有益于美国经济，促进国家健康，繁荣和福利，并通过技术创新确保国防安全，例如，通过更轻的增材制造部件减少飞机燃料消耗。这项研究所需的领域的交叉，包括：制造，机械工程，材料科学和计算科学，将支持跨学科的合作，可以导致现代工程教育的横向改进。作为该项目的一部分，将对高中生进行宣传，以培养他们对工程的兴趣，将招募本科生暑期实习生进行最先进的研究，并将创建与高级建模和模拟相关的专业研究生项目。研究的预期成果是金属合金等复杂分层材料的损伤和失效的预测计算理论。这项工作建立在传统力学框架下的数据驱动、降阶和多尺度原则的基础上，具有变革性新理论的潜力。最初，将进行基本表征实验（包括X射线断层扫描和衍射），以了解增材制造金属中材料微观结构和机械性能之间的关系。这些信息将用于校准微机械模型，模拟将用于填充合成微结构及其机械响应的数据库。由此，一个新的并发多尺度理论的基础上降阶方法将开发，能够捕捉几何和材料响应的非线性。该方法将查询在第一阶段构建的数据库中的机械信息，并使用该数据预测损坏和故障，特别是对于使用增材制造的金属部件。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

An inverse modeling approach for predicting filled rubber performance

• DOI: 10.1016/j.cma.2019.112567
• 发表时间: 2019-12
• 期刊: Computer Methods in Applied Mechanics and Engineering
• 影响因子: 7.2
• 作者: Jiaying Gao;M. Shakoor;H. Jinnai;H. Kadowaki;E. Seta;Wing Kam Liu

HiDeNN-TD: Reduced-order hierarchical deep learning neural networks

HiDeNN-TD：降阶分层深度学习神经网络

• DOI: 10.1016/j.cma.2021.114414
• 发表时间: 2022
• 期刊: Computer Methods in Applied Mechanics and Engineering
• 影响因子: 7.2
• 作者: Zhang, Lei;Lu, Ye;Tang, Shaoqiang;Liu, Wing Kam
• 通讯作者: Liu, Wing Kam

Predictive multiscale modeling for Unidirectional Carbon Fiber Reinforced Polymers

• DOI: 10.1016/j.compscitech.2019.107922
• 发表时间: 2020-01
• 期刊: Composites Science and Technology
• 影响因子: 9.1
• 作者: Jiaying Gao;M. Shakoor;G. Domel;Matthias Merzkirch;Guowei Zhou;D. Zeng;X. Su;Wing Kam Liu

Reduced Order Machine Learning Finite Element Methods: Concept, Implementation, and Future Applications

降阶机器学习有限元方法：概念、实现和未来应用

• DOI: 10.32604/cmes.2021.017719
• 发表时间: 2021
• 期刊: Computer Modeling in Engineering & Sciences
• 作者: Lu, Ye;Li, Hengyang;Saha, Sourav;Mojumder, Satyajit;Al Amin, Abdullah;Suarez, Derick;Liu, Yingjian;Qian, Dong;Kam Liu, Wing
• 通讯作者: Kam Liu, Wing

Mechanistic data-driven prediction of as-built mechanical properties in metal additive manufacturing

• DOI: 10.1038/s41524-021-00555-z
• 发表时间: 2021-06
• 期刊: npj Computational Materials
• 影响因子: 9.7
• 作者: Xiaoyu Xie;Jennifer L. Bennett;Sourav Saha;Ye Lu;Jian Cao;Wing Kam Liu;Zhengtao Gan