shtrausslearning/Data-Driven-CFD

Mixing Computational Fluid Dynamics Topics & Machine Learning

Data-Driven-CFD

• Fundamentally, it is the concept of utilising Computational Fluid Dynamics (CFD) data as inputs to Machine Learning (ML) Models
• The aim of such approaches is to help improve efficiency of a process or assist in solving a particular problem with the help of ML models (eg. regressors, classifier)

ML/CFD literature

• A brief insight into ML applications in the field of CFD suggest there are plenty of applications, especially utilising deep learning approaches:

  • Machine learning–accelerated computational fluid dynamics (https://www.pnas.org/content/118/21/e2101784118)

  • AI-accelerated CFD (Computational Fluid Dynamics): how does byteLAKE’s CFD Suite work? (https://becominghuman.ai/ai-accelerated-cfd-computational-fluid-dynamics-how-does-bytelakes-cfd-suite-work-fea42fd0761e)

  • Accelerating Convergence of Fluid Dynamics Simulations with Convolutional Neural Networks https://www.google.ru/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwi_4-aPkszyAhU5R_EDHdt-BMY4WhAWegQIEBAB&url=https%3A%2F%2Fpp.bme.hu%2Fme%2Farticle%2Fdownload%2F14134%2F8375%2F&usg=AOvVaw2ZZhgQartaf6pNmqTC3Zl0

  • NVIDIA Engineers Apply AI Advances in Engineering Physics Problems https://developer.nvidia.com/blog/nvidia-engineers-apply-ai-advances-in-fluid-mechanics/

  • Geometric Deep Learning for Volumetric Computational Fluid Dynamics https://www.google.ru/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwjwwajYl8zyAhV7SvEDHWmACGM4ZBAWegQIBRAB&url=https%3A%2F%2Finfoscience.epfl.ch%2Frecord%2F267539%2Ffiles%2FMaster%2520Thesis.pdf&usg=AOvVaw2TSQycdI551e0u0bAvw6Jp

Applications

• tradeOffML.md - Application of machine learning in trade-off studies