Data models for large aircraft aerodynamics using next-generation computational fluid dynamics

通过计算流体动力学模拟和大规模实验活动实现的高保真空气动力学数据对于满足飞机性能的严格限制，同时满足航空业最雄心勃勃的可持续性目标至关重要。该项目的重点是空气动力学数据生成的数值方面，使用并共同开发一种流代码，该代码利用为高度并行计算架构设计的最新算法，将开发生成能够封装许多关键动态现象的高保真空气动力学模型的工具和方法。这些现象会影响飞机的性能及其在高速跨音速流态中的环境足迹。这一挑战激发了对最先进的计算流体动力学（CFD）工具和在频域中运行的内置快速算法的需求，同时使用高性能计算系统。尽管拥有一流的CFD技术和计算设施，飞机机翼空气动力学设计周期中的快速周转时间要求使用有监督或无监督的机器学习算法，这些算法可以表示在巨大的参数空间中与真实世界数据的内插和外推相关的不确定性。在高性能计算系统上进行飞机机翼空气动力学的计算流体动力学代码和模拟，这是由于需要产生模拟复杂的跨音速空气动力学现象所需的数据。此外，与工业合作伙伴的领域专家合作，对最新的CFD技术在合适的用例上进行关键评估，以促进最终用户流程的接受和整合，同时挑战当前的工业实践。最后，机器学习算法的探索，以获得实用的工具，机翼设计问题，可以结合来自不同来源的数据。

High-fidelity aerodynamic data, enabled through computational fluid dynamics simulations and large-scale experimental campaigns, is critical in satisfying stringent constraints on aircraft performance while meeting the most ambitious targets on sustainability in the aviation sector. The project focus is on the numerical aspects of aerodynamic data generation using, and co-developing, a flow code that exploits the latest algorithms devised for highly parallel computing architectures.Tools and methods to generate high-fidelity aerodynamic models capable of encapsulating a number of critical dynamic phenomena will be developed. These phenomena affect the aircraft's performance and its environmental footprint in the high-speed transonic flow regime. This challenge then motivates the need for state-of-the-art computational fluid dynamics (CFD) tools and built-in fast algorithms operating in the frequency domain, while using high-performance computing systems. Despite best-in-class CFD technology and computing facilities, quick turnaround times in the design cycles for aircraft wing aerodynamics necessitate the use of supervised or unsupervised Machine Learning algorithms that can represent the uncertainty associated with interpolation and extrapolation across real-world data in vast parameter spacesThe main focuses of the project include the further development of a state-of-the-art CFD code and simulation of aircraft wing aerodynamics on high-performance computing systems necessitated by the need to generate the required data for modelling the intricate transonic aerodynamic phenomena. Also, the critical assessment of the latest CFD technology on suitable use cases in collaboration with the industrial partner's domain experts to foster the acceptance and integration into end-user processes, while challenging the current industrial practice. Finally, the exploration of Machine Learning algorithms to derive practical tools for wing design problems that can incorporate data from disparate sources.