Understanding the turbulence models

Video: Perform a Large Eddy Simulation (LES) turbulence modeling

Modeling fluid flow turbulence is important for the accurate simulation of fluid flow and convective heat transfer. The flow governing equations can be solved directly only for simple cases of flow. For real life turbulence flow simulation, the turbulence models are used.

The flow solver uses a number of different turbulence models, which add viscosity term to the Navier-Stokes governing equations, that can be classified as follows:

• Zero-equation models compute the viscosity term algebraically.

• One-equation models compute the viscosity term through one additional equation that is solved in parallel to the Navier-Stokes equations.

• Two-equation models compute the viscosity term through two additional equations that are solved in parallel to the Navier-Stokes equations.

All these models use simplified governing equations that don't take into account small turbulence scale. Thus, these equations are computationally less expensive to solve than using Direct Numerical Simulation. For more information, see Flow Solver Reference Manual .

Zero-equation models

Fixed Turbulent Viscosity

Provides fast, less computationally intensive solutions then one and two-equations models. The fixed turbulent viscosity model is robust and fast. Use it when you need to quickly obtain preliminary results for any model. It can also be useful in determining general trends and troubleshooting problem areas in the model. The biggest disadvantage of this model is that it is very sensitive to the specified characteristic turbulence scales. Once you have successfully run an analysis using this model, you can improve precision by restarting the analysis with the other turbulence models.Not available for the parallel flow solver and in Multiphysics.

Mixing Length

Provides good results in applications such as atmospheric science and oceanography. This model is sometimes called the Algebraic model. Although it is less accurate than the two-equation models for some problems, it is more robust and less computationally intensive than the two-equation models. This model provides accurate predictions for thin shear layer flows such as jets, mixing layers, wakes and boundary layers. It may fail in flows with separation and recirculation, such as back eddies behind a circular obstruction. In these cases, the two-equation models are better choices.

LES — Large Eddy Simulation

Solves the filtered Navier-Stokes equations. All other turbulence models use Reynolds stress terms in the Reynolds-Averaged Navier-Stokes (RANS) equations to model turbulent structures. This turbulence model is recommended when you want to visualize the actual eddies in the flow and when you are interested in the instantaneous flow fields and their statistics. For more information, see LES — Large Eddy Simulation.Not available for the serial flow solver.

One-equation models

SA — Spalart - Allmaras

It is used for predictions of aerodynamic flows such as wing profile and turbomachinery. It solves a transport equation for modified viscosity variable referred as the Spalart - Allmaras variable without calculating the length scale related to the shear layer thickness. It gives satisfactory predictions of boundary layers subjected to adverse pressure gradients. Mainly calibrated for aerospace applications, this model is s not a general purpose turbulence model. It is not suitable for turbulent jet or free shear flow modeling.Not available for the serial flow solver.

Two-equation models

Standard K-Epsilon

Is a well established and widely validated. However, it displays poor performance for unconfined flows, flows with rotation, and flows with strong adverse pressure gradients and very low y+(<11). For these flows, consider using the shear stress transport or k-omega models.Although the standard k-epsilon model, like many two-equation models, provides a more accurate description of the effects of turbulence on the mean flow, it also adds significant computational time to the solution. In addition, a finer 3D flow mesh is generally required to actually observe any improvement in turbulence simulation. The total solution time may double or triple by using the more accurate standard k-epsilon model. In general, you should first solve models using the fixed turbulent viscosity model before you use the standard k-epsilon model.

RNG K-Epsilon

Is derived from the application of the Re-Normalization (RNG) method to the Navier-Stokes equations. Compared to the standard k-epsilon model, this model has an additional term in the turbulence dissipation rate equation that accounts for the different scales of motion of turbulence flow. This gives a better accuracy for rapidly strained flow and swirling flows, high streamline curvature flows, transitional and separated flows than the standard k-epsilon model. It might be harder to get a converge solution with RNG k-epsilon model than the standard k-epsilon model.Not available for the serial flow solver.

Realizable K-Epsilon

Uses a new model for the dissipation rate equation and a new realizable eddy viscosity formulation compares to standard k-epsilon model. In this model, the Cμ quantities which is involved in the standard k-epsilon eddy viscosity formulation is no longer a constant but a variable. This model performs well in the same range of applications as the RNG k-epsilon model but has the benefit to perform better for rotational flows, jet flows and for calculation of boundary layers under strong adverse pressure gradients. It is recommended to use this model for high Reynolds number turbulent flow conditions.Not available for the serial flow solver.

K-Omega

Provides better results in transitional flows and in flows with adverse pressure gradients and for modeling boundary layer problems than the k-epsilon models. The k-omega model is one of the most common of turbulence models. It can be too sensitive to the inlet free-stream turbulence properties.In general, you should first solve models using the fixed turbulent viscosity model as the k-omega model adds significant computational time to the solution and requires a finer 3D flow mesh.

SST — Shear Stress Transport

Blends the standard k-epsilon model with k-omega model. Away from walls, the SST model behaves like the standard k-epsilon model. Near the walls, the SST model behaves like a k-omega model. The SST model displays better performance for unconfined flows and flows with strong adverse pressure gradients.

How do I

Define Advanced Parameters

Define Generic Entities

Learn more

Solution options

Understanding thermostat options for steady state analyses

Thermal initial conditions

3D flow initial conditions

Understanding ambient conditions

Joining fluid meshes

Wall distance calculation methods

Understanding out-of-bound options for time-dependent tables

Setting turbulence scale values

Turbulence Characteristics

Including additional input files for thermal analysis

Restarting a solution

Understanding time-varying time step

Target Temperature and Target Temperature Change

Solving the model

Wall distance calculation methods

Look up more details

LES — Large Eddy Simulation

Adaptive time stepping method

Parallel flow solver schemes

Computed quantities for multiple convective BCs

Meshing for turbulence modeling

Quick links

Command reference

Pre/Post video examples

Bulk Entry Descriptions

Simcenter 3D tutorials

Browse Simcenter 3D help by product area

Simcenter 3D Thermal/Flow, Electronic Systems Cooling, and Space Systems Thermal boundary conditions

Understanding the turbulence models, Simcenter 3D 2021.1 Series

© 2020 Siemens

window.mainLanguage="en_US"

window.delivId=""

window.projectId=""

MathJax.Hub.Config({ TeX: { extensions: ["autoload-all.js"] }, tex2jax: { displayMath: [ ] }, "SVG": { scale: 125 } });