Data-driven modeling SIG

andre tomislav
Andre Weiner Tomislav Marić
a.weiner@tu-braunschweig.de maric@mma.tu-darmstadt.de

OpenFOAM-v2206 release webinar

SIG overview

What data-driven entails:

  • machine learning
  • data science
  • data engineering

Short term goals:

  • reduce technical barrier
  • promote available techniques
  • promote third-party work

Long term goals:

  • aid understanding
  • accelerate developments
  • natural inclusion in OF

Current SIG structure (July 2022):

  • 11 members
  • academia - 5
  • industry - 6

Refer to the wiki for a detailed list.

Communication channels:

past/ongoing/upcoming activities

OF workshop training contributions:

Recordings will be available on the Journal's
Youtube channel.

levelset

Idea: solve $||\nabla \psi|| = 1$ using a PINN.

Idea: analyze coherent structures using DMD.

Reconstruction of main vortex shedding mode.

Reconstruction of high-frequency vortex shedding mode.

OpenFOAM-ML hackathon:

  • July 25-27
  • ~20 participants
  • thanks to ESI and AWS!

github.com/OFDataCommittee/OFMLHackathon

Lecture on ML in CFD at TU Braunschweig:

  • available on Github
  • work in progress
  • funded for 6 additional month

Foundation for Innovation in University Teaching

Thanks!