Open-Source Simulation Data for Validation and Machine Learning

PN 1A-1

Project Description

Future development of new modeling strategies for complex flows applications requires simulation data of specific benchmark problems which cover a wide range of flow problems. The need of such data is twofold. First, in classical method development, a wide range of validation data is required to test new models in different flow scenarios. Second, the model development based on machine learning  (ML) not only necessitates an increasing amount of data for validation purposes but also for the training of new ML models. Thus, the goal of this project is to build up an open-source database in addition to the open-source flow solver FLEXI to provide a publicly accessible database of benchmark flow problems.

Project Information

Project Number PN 1A-1
Project Name Open-Source Simulation Data for Validation and Machine Learning
Project Duration January 2021 - December 2022
Project Leader Andrea Beck
Claus-Dieter Munz
Johan Larsson (University of Maryland)
Project Members Jakob Dürrwächter, PhD Researcher
Thomas Kuhn, PhD Researcher
Project Partners Bernhard Weigand
Rainer Helmig

Andrea Beck

Prof. Dr.-Ing.

Numerical Methods in Fluid Mechanics

This image shows Claus-Dieter Munz

Claus-Dieter Munz

Prof. Dr. rer. nat.

Numerical Methods in Fluid Mechanics

[Photo: SimTech/Max Kovalenko]

To the top of the page