Turbulent Flow Simulations at the Exascale: Application to Wind Energy and Green Aviation

我们的日常生活被流体包围着，甚至被流体维持着。血液在我们体内的血管中流动，空气流入我们的肺部。在动荡的城市和室内环境中，流体流动分散了颗粒物空气污染。流体流动对我们的运输和工业起着至关重要的作用。我们的车辆在空气和水中行驶，由其他液体提供动力，这些液体在发动机的燃烧室中混合。如果不更好地了解流体动力学，我们今天面临的许多环境和能源相关问题就不可能得到解决。从实际的角度来看，与科学家和工程师相关的流体流动是湍流的；乱流是常态，不是例外。迄今为止，由于描述流体运动的完整方程的复杂性，仍然缺乏流体流动现象的完整理论。然而，对它们的理解和控制对于改进技术至关重要，尤其是在对生态影响最小的情况下，以及在许多领域，从工程应用(例如；（工业过程、推进和发电、汽车和飞机设计）到环境科学和技术（例如，空气质量、天气预报、气候预测、洪水灾害监测）。最近使用千万亿次计算取得了重大进展，计算流体动力学现在是实验和理论的重要补充。百亿亿次的湍流模拟将需要对现有的流动求解器进行重大的重新制定，实施新的物理学，并开发更细致的问题制定。然而，它有可能在我们追求更绿色的未来方面取得重大进展，这在很大程度上取决于对湍流这一首要主题的更好理解。为了更好地理解百亿亿次湍流计算带来的机遇和挑战，我们建议创建一个设计和开发工作组（DDWG），致力于百亿亿次湍流模拟，这是英国的一个高优先级研究领域。重点将放在风能和绿色航空应用上，因为百亿亿次计算将改变这些领域的游戏规则，并将有助于使英国成为一个更环保的国家。这个DDWG是建立在英国湍流联盟（UKTC）成员和最近资助的协作计算项目（CCP）湍流的经验和专业知识的基础上的。英国目前正在为pre-exascale和exascale系统设计两种最先进的开源流求解器OpenSBLI和Incompact3d，它们使用一种称为OPS的高级抽象框架，一种带有相关库和预处理器的API，用于在多块结构网格上为应用程序生成并行可执行程序。DDWG的最终产出将是提交一份报告，其中包含战略研究议程，该议程将清楚地阐明在百亿亿次湍流模拟中需要克服的研究挑战、机遇、主要风险和缓解措施。它将制定详细的方法，通过适当的面向应用程序的高级编程抽象，通过概念验证研究来展示OpenSBLI和Incompact3d用于百亿亿级计算的能力，从而实现CFD百亿亿级软件的开发。

Our daily life is surrounded - and even is sustained - by the flow of fluids. Blood moves through thevessels in our bodies, and air flows into our lungs. Fluid flows disperse particulate air pollution in the turbulent urban as well as indoor environments. Fluid flows play a crucial role for our transportation and our industries. Our vehicles move through air and water powered by other fluids that mix in the combustion chambers of engines. Many of the environmental and energy-related issues we face today cannot possibly be tackled without a better understanding of the dynamics of fluids.From a practical point of view, fluid flows relevant to scientists and engineers are turbulent ones; turbulence is the rule, not the exception. To date, a complete theory of fluid flow phenomena is still missing because of the complexity of the full equations describing the motion of a fluid.Their understanding and control is however crucial to improve technologies especially with minimal ecological impact as well as to anticipate events, in many areas ranging from engineering applications(e.g., industrial process, propulsion and power generation, car and aircraft design) to environmental sciences and technologies (e.g., air quality, weather forecasting, climate predictions, flood disasters monitoring).Significant progress has been made recently using petascale computing, and computational fluid dynamics is now a critical complement to experiments and theories. Turbulent flow simulations at the exascale will require significant reformulation of existing flow solvers, implementation of new physics, and development of a more nuanced problem formulation. It has however the potential to produce significant advances in our quest towards a greener future, relying in large parts on a better understanding of the overarching subject of turbulence.To better understand the opportunities and the challenges that will come with exascale computing for turbulent flows, we propose to create a Design and Development Working Group (DDWG) dedicated to turbulent flow simulations at the exascale, a high priority area of research for the UK. The focus will be on wind energy and green aviation applications as exascale computing will be a game changer in these areas and will contribute to make the UK a greener nation. This DDWG is building upon the experience and expertise of the UK Turbulence Consortium (UKTC) members and the recently funded Collaborative Computational Project (CCP) Turbulence. Two state-of-the-art open source flow solvers,OpenSBLI and Incompact3d, are currently being designed in the UK for pre-exascale and exascale systems using a high-level abstraction framework called OPS, an API with associated libraries and pre-processors to generate parallel executables for applications on multi-block structured meshes. The final output of this DDWG will be the delivery of a report with a strategic research agenda that will clearly articulate the research challenges to be overcome, opportunities, key risks and mitigation for turbulence simulations at the exascale. It will set out a detailed approach to enable development of CFD exascale-ready software through appropriate application-oriented, high-level programming abstractions, with proof-of-concept studies to demonstrate the capabilities of OpenSBLI and Incompact3d for exascale computing.