13763 - Flow Modeling with Machine Learning Modulübersicht

Module Number: 13763
Module Title:Flow Modeling with Machine Learning
  Strömungsmodellierung anhand maschinelles Lernen
Department: Faculty 3 - Mechanical Engineering, Electrical and Energy Systems
Responsible Staff Member:
  • Prof. Dr.-Ing. Schmidt, Heiko
Language of Teaching / Examination:English
Duration:1 semester
Frequency of Offer: Every summer semester
Credits: 6
Learning Outcome:The students are offered an introduction to machine learning in the context of computational fluid dynamics, and turbulent flow analysis and modeling. Elementary definitions and concepts in machine learning will be motivated by CFD applications. A large part of the course is dedicated to the analysis of numerical simulation data using supervised learning approaches. Some aspects of flow feature extraction using unsupervised learning, reduced order modeling, as well as general State-Of-The-Art research issues in machine learning for CFD will also be discussed. At the end of the course, the students are able to understand the key concepts and algorithms in machine learning and how they could be applied in CFD.
Contents:The course contents offer an overview on key machine learning concepts and fundamentals, and how they can be applied in CFD, mainly in the context of analysis of numerical simulations, and modeling of turbulent flows.
  • Review of linear algebra, conservation equations, dimensional analysis, and turbulence modeling
  • Linear regression algorithms and optimization
  • Fourier and Gabor transforms: links to turbulence modeling
  • Model selection
  • Proper Orthogonal Decomposition (POD) and Principal Component Analysis (PCA)
  • Generalities on neural networks
  • Clustering and classification
  • Flow modeling using reduced order models
Recommended Prerequisites:The student should be highly motivated to study numerical simulations of fluid flows using computational methods.
This is a Python-based course demanding some relevant programming background. To that extent, successful completion of the module “Introduction to Computational Thinking and Programming for CFD” is highly desired (but not mandatory). Completion of the module “Turbulence Modeling” is also recommended for a better theoretical overview, but not mandatory.
Mandatory Prerequisites:None
Forms of Teaching and Proportion:
  • Lecture / 2 Hours per Week per Semester
  • Exercise / 2 Hours per Week per Semester
  • Self organised studies / 120 Hours
Teaching Materials and Literature:
  • Duraisamy, Iaccarino and Xiao. “Turbulence Modeling in the Age of Data”, Annu. Rev. Fluid Mech. 51 (2019), 357-377
  • Brunton, Noack and Koumoutsakos. “Machine Learning for Fluid Mechanics”, Annu. Rev. Fluid Mech. 52 (2020), 477-508
  • Brunton & Kutz. Data-driven Science and Engineering. Cambridge University Press, 2019.
  • Chapra & Canale. Numerical Methods for Engineers. McGraw-Hill Higher Education, 2006.
  • Hastie, Tibshirani, and Friedman. The Elements of Statistical Learning: Data Mining, Inference and Prediction. Springer, 2009.
  • MacKay. Information Theory, Inference, and Learning Algorithms. Cambridge University Press, 2003.
Module Examination:Prerequisite + Final Module Examination (MAP)
Assessment Mode for Module Examination:Prerequisite: 
  • Successful completion of the 4-6 exercises discussed in the course
Final Module Examination: 
  • Oral exam (Presentation of short project with consecutive discussion of the results), 30-45 min.
Evaluation of Module Examination:Performance Verification – graded
Limited Number of Participants:None
Part of the Study Programme:
  • Master (research-oriented) / Hybrid Electric Propulsion Technology / PO 2024
  • Abschluss im Ausland / Maschinenbau / keine PO
  • Master (research-oriented) / Maschinenbau / PO 2023
  • Master (research-oriented) - Reduced Semester / Maschinenbau / PO 2023
  • Master (research-oriented) - Co-Op Programme with Practical Place / Maschinenbau - dual / PO 2023
Remarks:The module primarily aims at Master students in the engineering and natural sciences who plan to specialize in computational fluid dynamics.
Module Components:.
Components to be offered in the Current Semester: