Collaborative Research: Petascale Computing, Visualization, and Science Discovery of Turbulent Sooting Flames

提案标题：合作研究：湍流炭黑火焰的Petascale计算、可视化和科学发现。[牵头]机构：密歇根大学安阿伯提案编号：OCI-0904660主要研究者：Trouve，Arnaud C机构：马里兰州大学学院公园提案编号：OCI-0904480主要研究者：Haworth，丹尼尔C.机构名称：宾夕法尼亚州立大学大学公园提案编号：OCI-0904649主要研究者：Lu，Tianfeng机构：康涅狄格大学提案编号：OCI-0904771主要研究者：Sankaran，Ramanan机构：田纳西大学诺克斯维尔提案编号：OCI-0904818主要研究者：Ma，Kwan-Liu机构：加州大学戴维斯分校提案编号：OCI-0905008本提案将使用2009年美国复苏和再投资法案（公法111-5）提供的资金进行奖励。燃烧产生的烟尘会影响能源效率、环境和人类健康，因此控制烟尘是一个主要的技术和社会问题。本研究旨在利用peta-flop计算实现湍流燃烧碳烟预测。一个由六个研究小组组成的团队正在共同开发一个peta-flop软件包，该软件包将以前所未有的细节和现实主义水平捕捉湍流烟灰火焰的物理和化学。这些研究工作有望使我们对与能源转换和污染物控制有关的许多重要科学问题的基本认识取得突破。这里开发的综合软件包将允许详细考虑气相化学，烟尘形成和辐射传热现象。它将通过增强网格、运算符和方程级别的并行性来扩展之前用于湍流燃烧直接数值模拟（DNS）的万亿次代码。其他目标包括减少与不同碳氢燃料有关的烟尘形成的化学动力学机制;允许任意光学厚度的气体和烟尘颗粒的光谱分辨辐射传热模型;基于截面和矩方法组合的烟尘气溶胶和传输模型;多变量数据集和数据挖掘软件;新的容错和检查点能力;现场可视化;以及自动化特征提取和极限现象的跟踪，例如烟灰火焰中的点燃/熄灭。新的模拟能力将在实验室规模的湍流火焰在高雷诺数和达姆科勒数进行测试。这项合作研究将提高国家？通过吸引新一代的学生参与多学科计算科学和工程，提高竞争力。教师，学生和由此产生的软件开发将受益于计算机科学和领域科学的跨学科交流，以确保有效地设计和优化代码。软件的传播将通过开放源码许可证进行。该项目的多学科，多机构方面自然会导致一些赞助高中和本科生研究项目的机会。这些活动还将补充并受益于与世界各地燃烧研究和计算机科学领域其他研究小组的密切合作。

Proposal Title: Collaborative Research: Petascale Computing, Visualization, and Science Discovery of Turbulent Sooting Flames Principal Investigator: Im, Hong G. [Lead] Institution: University of Michigan Ann Arbor Proposal No: OCI-0904660 Principal Investigator: Trouve, Arnaud C Institution: University of Maryland College Park Proposal No: OCI-0904480 Principal Investigator: Haworth, Daniel C. Institution: Pennsylvania State Univ University Park Proposal No: OCI-0904649 Principal Investigator: Lu, Tianfeng Institution: University of Connecticut Proposal No: OCI-0904771 Principal Investigator: Sankaran, Ramanan Institution: University of Tennessee Knoxville Proposal No: OCI-0904818 Principal Investigator: Ma, Kwan-Liu Institution: University of California-Davis Proposal No: OCI-0905008 This proposal will be awarded using funds made available by the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). Energy efficiency, the environment, and human health can be affected by combustion-generated soot, so controlling soot is a major technological and societal concern. This research is directed toward achieving soot prediction from turbulent combustion by using peta-flop computing. A team of six research groups is working together to develop a peta-flop software package that will capture the physics and chemistry of turbulent sooting flames at an unprecedented level of detail and realism. These research efforts are expected to lead to breakthroughs in our fundamental understanding of many important scientific issues related to energy conversion and pollutant control. The comprehensive software package developed here will allow detailed consideration of gas-phase chemistry, soot formation, and radiative heat transfer phenomena. It will extend a previous teraflop code for direct numerical simulation (DNS) of turbulent combustion by enhancing parallelism at the grid, operator, and equation levels. Other goals include reduced chemical-kinetic mechanisms for soot formation associated with different hydrocarbon fuels; spectrally resolved radiative heat-transfer models for gases and soot particles allowing arbitrary optical thickness; a soot aerosol and transport model based on the combination of sectional and moment methods; multivariate dataset and data-mining software; novel fault tolerance and checkpoint capabilities; in-situ visualization; and automated feature extraction and tracking of limit phenomena such as ignition/extinction in sooting flames. The new simulation capability will be tested in laboratory-scale turbulent flames at high Reynolds and Damköhler numbers. This collaborative research will enhance the nation?s competitiveness by engaging a new generation of students in multi-disciplinary computational science and engineering. Faculty, students, and the resulting software development will benefit from the interdisciplinary interchange of computer science and domain science necessary to ensure that the code is designed and optimized efficiently. Dissemination of the software will occur through an open-source license. The multi-disciplinary, multi-institutional aspects of the project will naturally lead to a number of opportunities for sponsoring high school and undergraduate student research programs. The activities will also complement and benefit from close collaboration with other research groups worldwide in combustion research as well as in computer science.