Advancing Reynolds-Averaged Navier–Stokes Turbulence Models with Machine Learning and Data-Driven Methods - Doctoral Thesis

Aug 5, 2026

Upload by Eden

Reynolds-averaged Navier–Stokes (RANS) methods form the basis of turbulent-flow simulations in industrial and technical contexts because they deliver computational efficiency while retaining engineering usefulness. This doctoral work advances RANS turbulence modeling by integrating machine-learning ideas and data-driven methods, targeting improved prediction quality for key flow phenomena. The thesis connects methodological developments with results reported in peer-reviewed venues and conference proceedings, culminating in a coherent set of contributions evaluated through the presented research outcomes and discussions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages186

Reading Time7 h 49 min

AccessFree to download

Preface

Acknowledgements

Abstract

Technische Universität Darmstadt Fachbereich Maschinenbau Darmstadt 2025 Advancing Reynolds-Averaged Navier–Stokes Turbulence Models with Machine Learning and Data-Driven Methods Zur Erlangung des akademischen Grades Doktor-Ingenieur (Dr.-Ing.) genehmigte Dissertation von Felix Rüdiger Köhler M.Sc . Referent: Prof. Dr. rer. nat. Michael Schäfer Korreferent: Apl. Prof. Dr.-Ing. habil. Suad Jakirlić Tag der Einreichung: 24.02.2025 Tag der Disputation: 28.05.2025 Advancing Reynolds-Averaged Navier–Stokes Turbulence Models with Machine Learning and Data-Driven Methods Submitted doctoral thesis by Felix Rüdiger Köhler 1. Review: Prof. Dr. rer. nat. Michael Schäfer 2. Review: Apl. Prof. Dr.-Ing. habil. Suad Jakirlić Date of submission: 24.02.2025 Date of thesis defense: 28.05.2025 Technical University of Darmstadt – D17 Department of Mechanical Engineering Darmstadt 2025 Please cite this document as: URN: urn:nbn:de:tuda-tuprints-306847 URL: [https://tuprints.ulb.tu-darmstadt.de/30684](https://tuprints.ulb.tu-darmstadt.de/30684) [Year of publication on TUprints: 2025](Year of publication on TUprints: 2025) This document is provided by TUprints, the E-Publishing Service of TU Darmstadt [https://tuprints.ulb.tu-darmstadt.de](https://tuprints.ulb.tu-darmstadt.de)[ ](https://tuprints.ulb.tu-darmstadt.de)[tuprints@ulb.tu-darmstadt.de](tuprints@ulb.tu-darmstadt.de) This work is licensed under a Creative Commons License: CC BY 4.0 International [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/) Preface It is with great satisfaction that I present this doctoral thesis, representing the culmination of seven years of dedicated research and exploration. The results presented in this thesis go beyond the work published during my tenure as a research assistant at the Institute for Numerical Methods in Mechanical Engineering at the Technical University of Darmstadt. Partial results from this thesis have been published in: 1. F. Köhler, J. Munz, and M. Schäfer, "Data-driven augmentation of RANS turbulence models for improved prediction of separation in wall-bounded flows", In: AIAA Scitech 2020 Forum, 2020 . 2. F. Köhler, S. Jakirlić, S. Wegt, and M. Schäfer, "Data-driven Modeling of the Pressure strain Term in Second-Moment Closure RANS Models", In: Conference Proceedings: 13th International ERCOFTAC symposium on engineering, turbulence, modelling and measurements, Greece, Rhodes, pp. 515–520, 2021 . Additional research published during my time as a research assistant includes: 3. F. Köhler, R. Maduta, B. Krumbein, and S. Jakirlić, "Scrutinizing conventional and eddy-resolving unsteady RANS approaches in computing the flow and aeroacoustics past a tandem cylinder", In: Symposium der Deutsche Gesellschaft für Luft-und Raumfahrt, Cham: Springer International Publishing, pp. 586–596, 2018 . 4. Z. Kraus, F. Köhler, J. Friedrich, and M. Schäfer, "Deep Learning Approach for Curvature Prediction in Algebraic Volume of Fluid Methods", In: 9th International Conference on Computational Methods for Coupled Problems in Science and Engineering, 2021 . 5. M. Kannapinn, F. Köhler, and M. Schäfer, "A Validated Thermal Computational Fluid Dynamics Model of Wine Warming in a Glass", Applied Sciences 14, no. 19, 2024 . Acknowledgements This work would not have been possible without the support of many individuals, to whom Iam deeply grateful. I would like to express my sincere gratitude to Prof. Michael Schäfer forgiving me the opportunity to work as a research assistant and for fostering a straightforward and positive working environment. I am also grateful to him for granting me complete freedom to conduct my research independently, allowing me to explore ideas and complete this thesis. I am also sincerely thankful to Prof. Suad Jakirlić for his continuous support, which began during my master’s studies and extended throughout my doctora

FAQs

What is the main subject of the doctoral thesis?

The thesis focuses on advancing Reynolds-averaged Navier–Stokes (RANS) turbulence models using machine learning and data-driven methods to improve turbulent flow predictions.

Why are RANS methods important for turbulent-flow simulations?

RANS methods are fundamental because they provide computational efficiency for turbulent-flow simulations in industrial and technical applications.

What kinds of research outputs are referenced in the thesis?

The preface lists prior and related publications and conference papers, indicating that the thesis builds on multiple published partial results and research developed during the author’s tenure.

Aug 5, 2026

Upload by Eden