Course on advanced turbulence and transition modelling: Reynolds-Averaged Navier-Stokes (RANS), Large Eddy Simulation (LES) and laminar-to-turbulent transition models
Date and Time
19-20 September 2019, Warsaw, POLAND
Short course description
This 2-day course is intended for engineers whose work concerns CFD modelling of turbulent flows and convective heat transfer. The course aims to imbue the participants with a thorough understanding of eddy-viscosity based RANS models, RANS-based laminar-to-turbulent transition models, and LES approaches. A detailed explanation of RANS-based techniques will be given with emphasis on the interpretation of the results. Limitations and advantages of RANS-based strategies will be discussed. An explanation of the influence of inlet boundary conditions for modelled scalars will be provided; cases in which RANS models exhibit a strong sensitivity to the prescribed inlet values of the modelled variables will be reviewed.
The second part of the course is meant to address the deficiencies of RANS models. Examples of applications of the Large Eddy Simulation (implicit LES or quasi-Direct Numerical Simulation, DNS) and the increasingly popular lattice Boltzmann methods will be given. We will provide the theoretical background and discuss applications of LES for massively separated flows, where the use of standard RANS models might lead to large errors. We will teach how to evaluate the quality of a LES solution by means of two-point and autocorrelations and how to estimate the required mesh cell and time step sizes to resolve the turbulence statistics with sufficient accuracy.
The last part of the course will focus on modelling laminar-to-turbulent transition of a flow. We will explain the physics of natural, bypass, and separated flow transitions and describe the most popular transition models (available in many commercial CFD packages). Finally, we will show and discuss the results of transition simulations and compare them with experiments, DNS, and results obtained using fully-turbulent RANS models.
The lectures will consist of a theoretical background on each of the topics and will be illustrated with simulation results. Limitations and advantages of various turbulence modelling and convective heat transfer modelling strategies will be discussed. It will be demonstrated how to overcome or mitigate turbulent flow and heat transfer modelling errors by proper selection of closure strategies.
9:00 – Introduction to physics of turbulent flow, introduction to the Reynolds-averaged Navier-Stokes (RANS) equations.
10:30 – Description of the selected two-equation turbulence models based on practical examples. Why does RANS fail in the prediction of massively separated flows? In which models and flow regions are the values of turbulent kinetic energy and turbulent length scale overestimated? How is this correlated with mesh resolution?
- discussion of the k-epsilon and realizable k-epsilon models,
- discussion of the k-omega model,
- discussion of the k-omega SST model.
13:00 – Lunch
14:00 – Inflow boundary conditions for modelled scalars in RANS – when and in which models are the results strongly/mildly affected by the inlet values of modelled variables?
14:30 – RANS-based modelling of convective heat transfer: energy equation, gradient hypothesis, turbulent Prandtl number. Examples of results where heat transfer is largely under- or overestimated due to inappropriate selection of RANS models.
16:00 – Introduction to Large Eddy Simulation (LES).
9:00 – LES continued:
- How to evaluate the quality and accuracy of a LES solution and how to perform the postprocessing of results? (averages, stresses, two-point correlations, auto-correlations, postprocessing).
- How to choose proper mesh density and time step size of the simulation to reproduce turbulent eddies sufficiently?
11:00 – LES of massively separated flows. Discussion of convective heat transfer and comparison of results with RANS: does heat transfer always need to be treated as unsteady in massively separated flows, where do RANS methods completely fail to capture heat transfer characteristics?
12:00 – Introduction to the physics of the transition from laminar to turbulent flow in boundary layers. Modelling of the transition process.
13:00 – Lunch
14:00 – Modelling of the transition process (continued).
- discussion of the selected models and results
15:30 – An application of transition models for industrial cases. Differences between fully-turbulent and transitional flow results.
17:00 – End of the course
Tea and coffee during the breaks and lunch will be provided on each day of the course.
Takeaways from the course
- Participants will understand the physics of turbulence and laminar-to-turbulent transition,
- Participants will understand the underlying theoretical formulation of selected RANS models,
- Participants will have analysed real-life engineering cases in order to be able to discern which problems, discrepancies, and inaccurate solutions may stem from the inappropriate selection of turbulence models,
- Participants will understand, based on practical examples, how an incorrect solution of the flow and turbulent quantities affects the solution of the heat transfer problem,
- Participants will know how to post-process LES simulation results and how to choose required spatial and temporal resolution (mesh element size and time step size),
- Participants will know how the closure-terms in each of the discussed transition models are defined,
- Participants will have learned, based on conclusions drawn from the discussed cases, how to employ their CFD packages of choice to simulate turbulent flows with required accuracy.
The course is intended for industry-active engineers working on CFD simulations of turbulent flows. It may also be useful for PhD students or researchers working in the field.
Sławomir Kubacki, PhD, Doctor of Science (habilitation) – Head Lecturer of the Course
Dr Sławomir Kubacki is a recognized expert in the field of the hybrid RANS/LES methods and laminar-to-turbulent transition models for turbomachinery flows with more than 20 years of experience. His key achievement is the new transition intermittency model added to the k-omega equations. He has authored more than 50 journal papers and conference presentations in that field cited by over 350 authors (Scholar). In 2000, he worked at AEA-Technology, Otterfing, Germany, on testing and implementation of prototype intermittency model into the CFD code. His supervisor was Dr Florian Menter, the chief CFD scientist of ANSYS. After obtaining a PhD degree in 2005 from Czestochowa University of Technology (Poland), Dr Kubacki continued his research at the University of Ghent with Erick Dick. Dr Kubacki also led his projects on experimental and numerical investigation of turbulence transition models and CFD simulations for a low-pressure turbine, in cooperation with Avio Aero group (a GE Aviation Business). In 2017 he was granted Doctor of Science degree (habilitation) from Warsaw University of Technology, where he continues his research on turbulence modelling.
Bartosz Górecki, PhD – will lead the part of the course related to numerical methods in turbulence modelling
Dr Bartosz Górecki is the co-founder and the CEO of QuickerSim and the author of the first version of the QuickerSim CFD Toolbox for MATLAB. His adventure with CFD started over 10 years ago at RTWH Aachen (Germany), where he implemented chemical kinetics models to a hypersonic flow solver as his bachelor thesis. His master thesis on the unconditionally stable numerical time stepping schemes implemented with the spectral element method won him the Polish national competition for the best CFD-related work. His PhD thesis concerned the adjoint-based control of flow problems. In parallel to pursuing his PhD at Warsaw University of Technology, he worked as a CFD-engineer. He contributed to several aerospace projects. One of his tasks concerned the aerodynamics of the MiG-29 fighter aircraft. He kept extending his modelling skills during several summer and winter schools on turbulence modelling and HPC (the Chalmers University of Technology, CINECA in Bologna, LRZ in Munich, and HLRS in Stuttgart).
CZIiTT (Rektorska 4, 00-614 Warsaw, Poland) – Centre of Innovation Management and Technology Transfer, Warsaw University of Technology
Registration and Fees
IBAN account number (in EUR): PL 89114020040000351203916053
SWIFT code: BREXPLPWMBK
ul. Stanisława Noakowskiego 4/8
00-666 Warsaw, Poland
Telephone number: +48 514 021 542