SCALE:Industry empowerment to multiphase fluid dynamics simulations using Artificial Intelligence & statistical methods on modern hardware architectur

规模：使用人工智能进行多相流体动力学模拟的行业授权

• 批准号: EP/Y034686/1
• 负责人: Manolis Gavaises
• 金额: $ 66.43万
• 依托单位: City, University of London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2023
• 资助国家: 英国
• 起止时间: 2023 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FY034686%2F1
• 关键词: SCALE; Industry; empowerment; multiphase; fluid

Multi-phase, trans/supercritical and non-Newtonian fluid flows with heat and mass transfer are critical in enhancing the performance of energy production, propulsion and biomedical systems. Examples include: hydraulic turbomachines, ship propellers, CO2-neutral e-fuels and e-motor cooling systems, particleladen flows in inhalers and focused ultrasounds for drug delivery. What all these cases have in common is the high level of complexity which makes Direct Numerical Simulations impossible. State-of-the-art LES simulations rely on simplified assumptions but do not have yet the desired accuracy, while often require enormously expensive CPU resources. The aim of project (acronym 'SCALE') is to develop simulation methods and reduced-order models using physics-informed and data-driven Machine Learning and optimisation methods for such flow processes. These will be trained against 'ground-truth' databases that will be generated for the first time using both DNS and experimentally validated, industry-relevant LES and multi-fidelity RANS simulations. The new simulation tools will be applied for the first time to industrial problems and their ability to accelerate design times and improve accuracy will be jointly pursued and evaluated with the non-academic partners of SCALE. These are international corporations and market leaders in the aforementioned areas. Holistic training by experts from science and industry includes broad reviews on relevant scientific topics, modern high performance computing architectures suitable for performing such simulations, big data analytics as well as extensive support for mastering scientific tasks and transferring the knowledge acquired to industrial practice. SCALE will also deliver transferable soft skills training from a well-connected cohort of leaders with the ability to communicate across disciplines and within the general public. This coupling of research with industry makes SCALE a truly outstanding network for doctoral candidates to start their careers.

具有传热和传质的多相、反式/超临界和非牛顿流体流动对于提高能源生产、推进和生物医学系统的性能至关重要。示例包括：液压涡轮机、船舶螺旋桨、二氧化碳中性电子燃料和电动机冷却系统、吸入器中的微粒流和用于药物输送的聚焦超声。所有这些情况的共同点是高度的复杂性，这使得直接数值模拟是不可能的。最先进的LES模拟依赖于简化的假设，但还没有达到所需的精度，同时往往需要非常昂贵的CPU资源。该项目的目的（首字母缩写为“SCALE”）是开发模拟方法和降阶模型，使用物理信息和数据驱动的机器学习和优化方法，用于此类流程。这些将针对“地面实况”数据库进行训练，这些数据库将首次使用DNS和实验验证的行业相关LES和多保真度RANS模拟生成。新的仿真工具将首次应用于工业问题，其加速设计时间和提高精度的能力将与SCALE的非学术合作伙伴共同追求和评估。这些都是上述领域的国际公司和市场领导者。来自科学和工业的专家提供的全面培训包括对相关科学主题的广泛审查，适合执行此类模拟的现代高性能计算架构，大数据分析以及对掌握科学任务和将所获得的知识转移到工业实践的广泛支持。SCALE还将提供可转移的软技能培训，这些培训来自一群关系密切的领导者，他们有能力跨学科和在公众中进行沟通。这种研究与工业的结合使SCALE成为博士候选人开始职业生涯的真正优秀的网络。