In a new cross-disciplinary project, researchers from Aarhus University will develop an entirely new way of modelling complex turbulent systems.

Accurate simulations are crucial for understanding turbulent flows and are key to designing more efficient engineering systems. This includes systems for energy production, such as wind farms, as well as those for transportation, like ships, trains, and aircrafts.  Additionally, these simulations are crucial for addressing environmental concerns, notably in predicting greenhouse gas emission and pollutant dispersion.

In a research project called "Data-driven model discovery for turbulent flows", researchers will use progressive machine learning as an entirely new approach to develop groundbreaking turbulence models.

"With this project, we want to use progressive machine learning as a new approach to develop turbulence models that can predict complex flows in turbulent systems. Our approach in the project is different from traditional machine learning methods because it mimics the progressive development of empirical models based on physical hypotheses, rather than blindly using machine learning to forget established laws of physics,” says Associate Professor Mahdi Abkar from the Department of Mechanical and Production Engineering at Aarhus University, who is heading the project with Professor Alexandros Iosifidis from the Department of Electrical and Computer Engineering.

He continues:

"Our vision is to develop tools that can very accurately predict turbulent flows and thus be applied in engineering design and decision-making in both academia and industry."

The project aims to create a long-term collaboration between research communities in computer science and fluid mechanics, and its results will be included in research-based teaching activities at Aarhus University.

"We believe that society needs a new generation of engineers with solid physics and computer engineering knowledge who can understand and adapt machine learning tools to solve physics problems such as turbulence modelling," adds Mahdi Abkar.

The project has received DKK 3 million in funding from the Villum Foundation's Synergy pool.

Contact

Professor Alexandros Iosifidis Aarhus University, Department of Electrical and Computer Engineering
​​​​​​ai@ece.au.dk
Tel.: +45

Associate Professor Mahdi Abkar
Aarhus University, Department of Mechanical and Production Engineering
Mail: abkar@mpe.au.dk 
Tel.: +4593521694