Wall modeled computational fluid dynamics using machine learning - Using CFD coupled machine learning to create wall models for increased LES knowledge for future industrial use

Aug 2, 2026

Upload by Ezra

Machine learning is increasingly used in engineering and computational fluid dynamics, where faster yet accurate turbulence simulations remain a key challenge. The thesis studies wall-modeled Large Eddy Simulation by addressing limitations of classical wall models, which become inaccurate under adverse pressure gradients. A new CFD-coupled training strategy for machine-learning wall models is proposed, combining supervised learning with unsupervised elements to correct numerical inaccuracies. Models are trained on channel flow and boundary-layer flow, yielding accurate results and outperforming classical wall functions in comparison, while training time and coupling effects pose practical issues. Future work targets more complex adverse-pressure flows and full LES investigations.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages41

Reading Time1 h 43 min

AccessFree to download

Abstract

Contents

1 Introduction

2 Theory

3 Method

4 Results

5 Conclusions

Bibliography

Wall modeled computational fluid dynamics using machine learning Using CFD coupled machine learning to create wall models for increased LES knowledge for future industrial use Master’s thesis in Applied Mechanics RASMUS OLAUSSON DEPARTMENT OF MECHANICS AND MARITIME SCIENCES DIVISION OF FLUID DYNAMICS CHALMERS UNIVERSITY OF TECHNOLOGY Göteborg 2024 [www.chalmers.se](www.chalmers.se) Master’s thesis 2024 Wall modeled computational fluid dynamics using machine learning Using CFD coupled machine learning to create wall models for increased LES knowledge for future industrial use RASMUS OLAUSSON Department of Mechanics and Maritime Sciences Division of Fluid Dynamics Chalmers University of Technology Göteborg 2024 Wall modeled computational fluid dynamics using machine learning Using CFD coupled machine learning to create wall models for increased LES knowledge for future industrial use RASMUS OLAUSSON © RASMUS OLAUSSON, 2024 . Supervisor: Magnus Carlsson, SAAB Aeronautics Examiner: Lars Davidson, Division of Fluid Dynamics Master’s thesis 2024 Department of Mechanics and Maritime Sciences Division of Fluid Dynamics Chalmers University of Technology SE-412 96 Göteborg Telefon +46 31 772 1000 Typeset in LATEX Göteborg 2024 Wall modeled computational fluid dynamics using machine learning Using CFD coupled machine learning to create wall models for increased LES knowledge for future industrial use RASMUS OLAUSSON Department of Mechanics and Maritime Sciences Division of Fluid Dynamics Chalmers University of Technology Abstract Machine learning is becoming a useful tool in many parts of engineering and science and computational fluid dynamics is no exception. At the same time the need for more accurate simulations with resolved turbulence is also increasing. But to make turbulence resolving methods such as Large Eddie Simulations (LES) accessible for industrial use computational speed up is still required. One method is to employ wall modeled LES. Classical wall models works well for simple flows but they get inaccurate for flows containing adverse pressure gradient. A solution to this problem is to use a machine learning based wall model. In this project a machine learning based method for wall models is investigated anda new proposed way of coupling a CFD solver to the training process is tested. The method combines a supervised learning method with an unsupervised learning approach to take numerical inaccuracies of the CFD solver into the wall model. The models in this project were trained on channel flow and boundary layer flow. The results shown that machine learning based methods do work and give accurate results. These models are compared to classical wall functions and they can be more accurate. Some problems in the training process was found and especially the training time could become unreasonable if coupled with a LES solver. Solutions to these problems are discussed and in the future more complicated flows with adverse pressure gradients and LES simulations should be investigated. Nyckelord: CFD, LES, RANS, Machine Learning, Wall models. Acknowledgements I want to give a massive thanks to my supervisor Magnus Carlson at SAAB who has been great. He has guided me throughout thus project and his advice and support has been invaluable. I also want to thank Sebastion Arvidsson at SAAB for giving me this opportunity and showing his interest in the project his input and support has been greatly appreciated. Lastly I want to thank my examiner Lars Davidson who’s input and knowledge in the field has been needed in many of the cross roads faced in this project. Rasmus Olausson, Göteborg, June 2024 Contents List of Figures xi 1 Introduction 1 1. 1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 Theory 3 2.1 The governing equations . . . . .

FAQs

Why are classical wall models insufficient for industrial wall-modeled LES?

Classical wall models work well for simple flows but can become inaccurate when flows include adverse pressure gradients.

How does the project couple CFD with machine learning during training?

The thesis investigates a machine learning wall-model approach with a proposed coupling method that integrates CFD solver behavior into the training process.

What flows were used to train and evaluate the machine learning wall models?

The models were trained and tested on channel flow and boundary-layer flow, and the results were compared with classical wall functions.

Aug 2, 2026

Upload by Ezra