Realistic Simulation of Jet Engine Noise using Petaflop Computing

使用千万亿次计算对喷气发动机噪声进行真实模拟

• 批准号: 0904675
• 负责人: Gregory Blaisdell
• 金额: $ 129.17万
• 依托单位: Purdue University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2009
• 资助国家: 美国
• 起止时间: 2009-09-01 至 2015-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0904675&HistoricalAwards=false
• 关键词: Realistic; Simulation; Jet; Engine; Noise

Blaisdell0904675The goal of this study is to advance the science of subsonic and supersonic jet noise prediction for modern-day turbofan aircraft engines using petascale computing. Jet noise is an important issue due to increased air traffic, penalties for noisier aircraft, future stringent noise regulations and military operational requirements. Previous experiments have shown that a 50% decrease in jet noise power output can be achieved by certain chevron and lobe mixer designs without essentially changing the net thrust of the engine. However, the physical mechanisms for the reduced noise are not well understood. The effect on noise of mixing devices, in particular chevrons and lobed mixers, is the focus of the present work.The PIs will investigate turbulent mixing by accurately simulating it with advanced computational techniques based on large eddy simulation (LES). Integral acoustic methods will extend the computational fluid dynamics (CFD) results to the far-field. Processing speeds and memory sizes of existing supercomputers limit current simulations to low Reynolds numbers and idealized geometries for the mixing devices. Thus, these simulations do not allow design and optimization of mixing for noise reduction, especially since these mixing devices influence the high frequencies of the noise spectrum, increasing the grid resolution requirements. Modeling at realistic Reynolds numbers and nozzle geometries requires tens of billions of points. The PIs algorithms will be designed to take advantage of multi-level parallelism and, within a node of such an architecture, address the 'memory wall' aspect of multicore architectures where the cost of arithmetic operations is much smaller than memory references. These algorithms will be based on a mixture of the transposition scheme and the multi-block approaches we used in the past. The methodology will be validated by making comparisons with both turbulence and acoustics measurements from high-quality experiments using realistic nozzle geometries. This project represents a computational-engineering activity integrating modern modeling, parallel algorithm design, and fine-tuned implementations on petascale architectures. The project will help promote interdisciplinary research, teaching, training, and learning by training three Ph.D. graduate students. One undergraduate research assistant will be involved during each summer for the five years of the project. Outreach to inner-city high school students from Indianapolis will be done through Purdue's Science Bound Program. The findings of this research will be shared with the aerospace and computer industries. As appropriate, results will also be shared with the general public through the Purdue News Service.

这项研究的目标是利用千万亿次计算技术来推进现代涡轮风扇飞机发动机的亚音速和超音速喷气噪声预测科学。由于空中交通的增加，对噪音飞机的处罚，未来严格的噪音法规和军事操作要求，喷气机噪音是一个重要的问题。先前的实验表明，在不改变发动机净推力的情况下，通过某些v形和叶形混频器设计，可以使喷气噪声功率输出降低50%。然而，降低噪声的物理机制尚不清楚。混合装置对噪声的影响是目前研究的重点，尤其是螺旋形和叶状混合器。pi将通过基于大涡模拟（LES）的先进计算技术精确模拟湍流混合来研究湍流混合。积分声学方法将计算流体力学（CFD）的结果扩展到远场。现有超级计算机的处理速度和内存大小限制了当前对混合装置的低雷诺数和理想几何形状的模拟。因此，这些模拟不允许设计和优化用于降噪的混合，特别是因为这些混合设备影响噪声频谱的高频，增加了网格分辨率要求。在真实的雷诺数和喷嘴几何形状下建模需要数百亿个点。pi算法将被设计为利用多层次并行性，并且在这种架构的节点内，解决多核架构的“内存墙”方面的问题，其中算术运算的成本远小于内存引用。这些算法将基于转置方案和我们过去使用的多块方法的混合。该方法将通过与使用实际喷嘴几何形状的高质量实验的湍流和声学测量结果进行比较来验证。这个项目代表了一个计算工程活动，集成了现代建模、并行算法设计和千兆级架构上的微调实现。该项目将培养3名博士研究生，促进跨学科研究、教学、培训和学习。在项目的五年里，每年夏天都会有一名本科生研究助理参与。普渡大学将通过“科学束缚计划”向印第安纳波利斯市中心的高中生提供服务。这项研究的结果将与航空航天和计算机行业共享。在适当的情况下，结果也将通过普渡大学新闻服务与公众分享。