Collaborative Research: Multiscale Modeling of Amorphous Solids - Energy Landscapes to Failure Prediction

合作研究：非晶固体的多尺度建模 - 能源景观到故障预测

• 批准号: 1909733
• 负责人: Christopher Rycroft
• 金额: $ 21.98万
• 依托单位: Harvard University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-12-01 至 2022-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1909733&HistoricalAwards=false
• 关键词: Collaborative; Research; Multiscale; Modeling; Amorphous

NONTECHNICAL SUMMARYThis award supports theoretical and computational research, and education to advance understanding of how amorphous materials fail or break under mechanical stress. Amorphous solids are a category of materials that share the distinction that they do not exhibit any crystal structure. Examples include colloidal pastes, foams, gels, silicate glasses, like window glass and Gorilla(R) glass, and many consumer plastics. This research will lay the groundwork for understanding how these materials respond to stresses and eventually fail so as to guide the use of existing materials as well as the development of new materials. The research focuses on metallic glass as an exemplary amorphous solid. Metallic glasses are an extremely promising emerging class of high strength metallic materials, but their application is limited by their failure mechanisms and the current inability to predict their behavior when subjected to mechanical loads. Improved predictive capabilities are necessary for guiding the design and utilization of failure tolerant metallic glass alloys that will have broad application in medicine, defense and consumer goods. This research will build mathematical descriptions of metallic glasses based on simulations performed on the atomic scale. Computational methods to implement these mathematical descriptions will then be developed so as to estimate both the material behavior and the reliability of the predictions that result. The techniques developed will be broadly disseminated through open source computer codes. Additionally, the project will be integrated with educational research and outreach activities of the investigators, which include broad systemic engagement within Baltimore City elementary schools, improvement of introductory computing education, outreach focused on improving participation of women and URM students at both the high-school and undergraduate levels in engineering research, and public lectures at libraries.NONTECHNICAL SUMMARYThis award supports theoretical and computational research, and education to advance understanding of how amorphous materials fail under mechanical stress. The research team aims to develop physics-based multiscale models of failure processes in amorphous solids, with metallic glass taken as an exemplar material. The physics accessible via atomic scale models of glass structure will be analyzed to obtain statistics of the shear transformation zone defects that control plastic deformation. These will be related to statistical mechanics models of effective temperature that characterize the degree of glass disorder so as to build a constitutive model of plastic deformation. The constitutive model will in turn be incorporated into a high-fidelity 3D viscoplastic finite differencing scheme that adapts techniques originally developed for solving the Navier-Stokes equation. A novel machine learning algorithm will guide the parameterization of the constitutive model from atomistic data so as to inform the continuum method and provide uncertainty quantification. The simulation methods will be pushed to increase the scales on which failure can be modeled by developing a statistical representative-volume element approach incorporating adaptive meshing and resulting in true multiscale simulations of failure in amorphous solids. In doing so, this research is aimed to broadly advance fundamental understanding of how the macroscopic mechanical response of amorphous solid materials is informed by their atomic scale structures. The research will result in the development of methods for making direct connections between atomic scale data and theories applicable on the continuum scale. These will include novel machine learning methods and new numerical schemes for failure prediction that incorporate uncertainty quantification. Ultimately, this work will result in an improved understanding of how amorphous microstructure controls failure on the macro scale.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.

该奖项支持理论和计算研究，以及教育，以促进对非晶材料在机械应力下如何失效或断裂的理解。无定形固体是一类材料，它们的区别在于它们不表现出任何晶体结构。实例包括胶体糊剂、泡沫、凝胶、硅酸盐玻璃（如窗玻璃和Gorilla（R）玻璃）和许多消费塑料。这项研究将为了解这些材料如何响应应力并最终失效奠定基础，从而指导现有材料的使用以及新材料的开发。研究的重点是金属玻璃作为一种典型的非晶固体。金属玻璃是一种非常有前途的新兴类别的高强度金属材料，但它们的应用受到其失效机制和目前无法预测其在承受机械载荷时的行为的限制。改进的预测能力对于指导失效容限金属玻璃合金的设计和利用是必要的，所述失效容限金属玻璃合金将在医学、国防和消费品中具有广泛的应用。这项研究将基于在原子尺度上进行的模拟建立金属玻璃的数学描述。然后将开发实现这些数学描述的计算方法，以估计材料行为和预测结果的可靠性。所开发的技术将通过开放源码计算机代码广泛传播。此外，该项目将与调查人员的教育研究和外联活动相结合，其中包括巴尔的摩市小学内的广泛系统参与，改进计算机入门教育，重点提高高中和本科阶段妇女和URM学生参与工程研究的外联活动，NONTECHNICAL SUMMARY该奖项支持理论和计算研究，以及促进对非晶材料在机械应力下如何失效的理解的教育。该研究小组的目标是开发基于物理学的非晶固体失效过程的多尺度模型，并以金属玻璃为范例材料。通过玻璃结构的原子尺度模型可访问的物理将被分析，以获得控制塑性变形的剪切转变区缺陷的统计数据。这些都将涉及到统计力学模型的有效温度，表征玻璃无序程度，从而建立一个本构模型的塑性变形。本构模型反过来将被纳入一个高保真的三维粘塑性有限差分格式，适应技术最初开发的解决Navier-Stokes方程。一种新的机器学习算法将指导原子数据的本构模型的参数化，以便通知连续方法并提供不确定性量化。模拟方法将被推到增加的规模上，可以通过开发一个统计的代表性体积元素的方法，结合自适应网格，并导致真正的多尺度模拟的非晶固体故障建模。在这样做的过程中，这项研究的目的是广泛地推进对非晶固体材料的宏观力学响应如何由其原子尺度结构所告知的基本理解。这项研究将导致发展方法，使原子尺度的数据和适用于连续尺度的理论之间的直接联系。这些将包括新的机器学习方法和新的数值方案，用于故障预测，其中包括不确定性量化。最终，这项工作将导致对非晶微观结构如何在宏观尺度上控制失效的更好理解。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

Manifold learning for coarse-graining atomistic simulations: Application to amorphous solids

• DOI: 10.1016/j.actamat.2021.117008
• 发表时间: 2021-03
• 期刊: Acta Materialia
• 影响因子: 9.4
• 作者: Katiana Kontolati;Darius D. Alix-Williams;Nicholas M. Boffi;M. Falk;C. Rycroft;M. Shields