ITR: High Fidelity Simulation for Heterogeneous Civil and Mechanical Systems

摘要：异质民用和机械系统的保真仿真该奖项资助了一项为期三年的研究计划，旨在开发和评估在土木和机械工程中重要的强异质耦合系统的高保真建模和仿真方法。这些系统的特点是相互作用的部件具有广泛不同的机械特性。该方法依赖于多级分解、分系统、接口定位和正则化。提出的研究基于两个创新概念：(1)接口定位方法旨在将系统组件的异构模型连接起来，同时保持非匹配离散化之间的一致性和能量节约。(2)一种正则化方法，该方法保留了在广泛耦合问题的接口上传输的相互作用的全部准确性。本研究将由两个具有不同物理尺度的应用实例驱动并进行评估：(1)地震激励下的大坝，考虑流固相互作用与体空化以及与高度非均质土壤地层的相互作用。(2)一种宽带MEMS频率传感器器件，该器件由安装在层状硅衬底上的谐振器组成。在过去的三十年中，基于模型的物理均匀部件模拟技术取得了巨大的进步。在连接这类模型时会出现困难。这些困难可以追溯到物理，建模或计算方法的不匹配。例如，降解界面或保护性泡沫层在预测组分间通量和力时可能产生非常不准确的结果。本研究采用的方法旨在最大限度地利用现有的孤立子系统模型和方法，同时开发解决这些困难的交互技术。这项研究的结果将应用于处理强异质模型相互作用的其他科学和技术领域。其中包括飞机和海洋工程，电子封装和先进材料的设计。计算多物理场有望成为美国工业正在进行的努力的关键组成部分，通过更快地探索设计和概念，并最终与CADM集成优化，通过信息技术提高生产率。然而，多物理场模拟固有的复杂性必然需要基础学科专家的协作。当最初为解决单个问题而开发模型时，这可能会导致困难和昂贵的改造。研究的目的是促进，通过模块化和重用，及时结合先进的建模和接口技术的仿真。

AbstractHigh Fidelity Simulation for Heterogeneous Civiland Mechanical SystemsThis award funds a three-year research program to develop and evaluate methods for high fidelity modeling and simulation of strongly heterogeneous coupled systems of importance in Civil and Mechanical Engineering. These systems are characterized by interacting components of widely different mechanical characteristics. The approach relies on multilevel decomposition, into subsystems, interface localization and regularization. The proposed research is based on two innovative concepts:(1) An interface localization method designed to link heterogeneous models of system components while maintaining consistency and energy conservation between nonmatching discretizations. (2) A regularization approach that preserves the full accuracy of the interactions transmitted across the interfaces for a wide spectrum of coupled problems.The research will be driven by and evaluated on two application examples that exhibit widely different physical scales:(1) A dam under seismic excitation, accounting for fluid-structure interaction with bulk cavitation as well as interaction with highly heterogeneous soil strata.(2) A wide-band MEMS frequency-sensor device consisting of a resonator mounted on a layered silicon substrate.Techniques for model-based simulation of physically homogeneous components have experienced great advances over the past three decades. Difficulties occur when interfacing such models. These difficulties can be traced to physical, modeling or computational method mismatches. For example, a degrading interface or a protective foam layer can produce highly inaccurate results in the prediction of inter-component fluxes and forces. The approach taken in this research aims to take maximum advantage of existing models and methods for the isolated subsystems, while developing interaction techniques that address those difficulties. Outcomes of this research will have application to other fields of science and technology that deal with the interaction of strongly heterogeneous models. These include aircraft and marine engineering, electronic packaging and the design of advanced materials. Computational multiphysics is expected to be a key component of ongoing US industry efforts to boost productivity through information technology by faster exploration of designs and concepts, and eventually optimization integrated with CADM. However, the complexity inherent in multiphysics simulations necessarily requires the collaboration of experts in the underlying disciplines. This can lead to difficulties and costly retrofits when models originally develop to address the individual problems are interfaced. The research aims at facilitating, through modularity and reuse, the timely incorporation of advances in modeling and interfacing techniques in simulation.