Collaborative research: simulation and analysis of turbulent jet noise using arbitrary-order Hermite methods

该奖项是根据2009年美国复苏和投资法案（公法111-5）资助的。研究人员与他们的学生和博士后一起，进行一个跨学科的项目，主要有三个目标。第一个目标是开发、优化和传播基于任意阶埃尔米特近似的新型计算方法，这些方法有效地利用千万亿级设施来求解复杂的、多尺度的、时变的偏微分方程系统。其次，Hermite方法被应用于湍流射流噪声的直接模拟，其雷诺数比现有的雷诺数高出一个数量级，为未来与工程设计相关的流动条件下的模拟铺平了道路。最终目标是开发和应用后处理技术，利用仿真数据发现导致喷气噪声的基本物理机制，改进工程模型的输入，并为降噪策略提供信息。Hermite方法的独特特性使其成为高性能计算平台上高分辨率模拟的理想选择。这是他们第一次被充分训练去解决难题。这些方法本身可以在任何需要高精度的应用程序中使用，并且创建了一个优化的构建块库以及用于其应用程序的模板，并免费提供给科学界。理解喷气机噪声是一项巨大的科学挑战。辐射声仅携带流动能量的极小部分，是小尺度湍流和大尺度相干结构复杂动力学的结果。下一代高性能计算系统的有效利用受到硬件异构特性的阻碍。这个项目的重点是新方法，这些方法既足够灵活，可以用于复杂的体系结构，又足够强大，可以提供高分辨率的近似。喷气机噪声是一个环境问题，在世界范围内受到越来越严格的管制。为了实现正在讨论的雄心勃勃的降噪目标，需要大大提高对基础物理学的理解。研究人员将完成的模拟和后续分析使我们更接近发现声辐射的微妙物理机制和合理设计抑制方法的目标。

Hagstrom and Colonius This award is funded under the American Recovery andReinvestment Act of 2009 (Public Law 111-5). The investigators, with their students and postdocs, pursue an interdisciplinary project with three primary aims. The first aim is the development, optimization, and dissemination of novel computational methods based on arbitrary-order Hermite approximations that efficiently utilize petascale facilities to solve complex, multiple-scale, time-dependent systems of partial differential equations. Secondly, Hermite methods are applied to the direct simulation of turbulent jet noise at Reynolds numbers more than an order of magnitude beyond those currently available, paving the way for future simulations under flow conditions relevant to engineering design. The final goal is the development and application of post-processing techniques to use the simulation data to discover the basic physical mechanisms responsible for the jet noise, improve the inputs to engineering models, and inform strategies for noise reduction. The unique properties of Hermite methods make them ideal for high-resolution simulations on high-performance computing platforms. This is the first instance where they are fully exercised to attack a difficult problem. The methods themselves can be profitably used in any application requiring high accuracy, and a library of optimized building blocks along with templates for their application is created and made freely available to the scientific community. To understand jet noise is an immense scientific challenge. The radiated sound, which carries only a minuscule fraction of the flow energy, results from both small-scale turbulence and the complex dynamics of larger-scale coherent structures. The efficient utilization of the next generation of high-performance computing systems is hindered by the heterogeneous nature of the hardware. This project is focused on new methods that are both flexible enough for complex architectures, and powerful enough to deliver high-resolution approximations. Jet noise is an environmental problem subject to increasingly severe regulation throughout the world. To meet the ambitious noise reduction goals under discussion, a greatly enhanced understanding of the basic physics is needed. The simulations and follow-up analysis to be completed by the investigators bring us much closer to the goal of discovering the subtle physical mechanisms responsible for the acoustic radiation and to the rational design of methods to suppress it.