Turbulence

Products and versions covered

By: Help
Dec 28 2015In-Product View

The Turbulence dialog is for enabling or disabling turbulence, selecting the turbulence model and for
modifying turbulence model parameters.

Select Laminar to simulate laminar flow.

Select Turbulent (the default) to simulate turbulent flow. Most engineering flows are turbulent.

If it is unclear if an analysis should be run as laminar or turbulent, try laminar first. If the flow is actually
turbulent, the analysis will typically diverge within the first ten to fifteen iterations. Change the setting
Turbulent, and start again from iteration 0.

Turb. model
k-epsilon
This is the default turbulence model. It is typically more accurate than the constant eddy viscosity model
but more computationally intensive and slightly less robust. It is not as resource intensive as the RNG
model, but still gives good results. It is a general purpose model that performs well across a large numbe
applications.

SST k-omega
We recommend SST k-omega for external aerodynamics, separated or detached flows, and flow with
adverse pressure gradients.
SST k-omega is robust across a wide range of flow types.
SST does not use wall functions. It simulates turbulence all the way to the wall. To use SST k-omega
effectively, the mesh needs to be very fine in the boundary layer region. You can add up to 10 layers
with the Mesh Enhancement dialog.

Additional notes about SST k-omega

The SST models are a hybrid of the Wilcox k-omega and a k-epsilon model variant. The benefits of this
model include the following:
 1. The SST models exhibit less sensitivity to free stream conditions (flow outside the boundary layer) th
many other turbulence models.
2. Using a shear stress limiter, these models avoid a build-up of excessive turbulent kinetic energy nea
stagnation points.
3. The SST models provide a platform for additional extensions such as SAS and laminar-turbulence
transition.
4. The SST models do not account for wall roughness effects. To simulate wall roughness, it is better to
use the k-epsilon turbulence model .

Scale Adaptive Simulation (SST k-omega SAS)

SST k-omega SAS is recommended for flows with transient turbulence structures such as vortex
shedding and variable wake structures.
You can run steady state simulations with SST k-omega SAS. The turbulent structures cannot be
animated, but the SST k-omega SAS model predicts their formation and shape better than a steady-
state k-epsilon simulation.
To use SST k-omega SAS effectively, the mesh needs to be very fine in the boundary layer region. Y
can add up to 10 layers with the Mesh Enhancement dialog.

Detached Eddy Simulation (SST k-omega DES)

SST k-omega DES is a hybrid between SST k-omega and large eddy simulation (LES) models.
SST k-omega DES produces accurate results for separated, high Reynolds external aerodynamics f
applications.
This model is computationally intensive and is sensitive to the mesh distribution. It works best with a
uniform mesh distribution.

Low Re k-epsilon
This model is well suited for low speed, turbulent flows. The Reynolds number of such flow is typically
between 1,500 and 5,000.

Typical applications include pipe flows and external aerodynamic flow transitioning between laminar and
turbulent, as well as flow situations that have both high speed and low speed areas.

Other flow situations that perform well with the Low Reynolds turbulent model include:

A high-speed jet entering a large room. The jet is highly turbulent when it first enters the room, but th
flow slows down considerably, and the Reynolds number drops. These types of flows can be very
unstable when run with k-epsilon.
Because this turbulent model does not use wall functions, Mesh Enhancement should be always be
enabled. We recommend increasing the number of mesh enhancement layers to 5 (using the Mesh
Enhancement controls on the Meshing dialog).

Note that analyses run with this turbulence model may not be as stable as those run with the k-epsilon
model. Because of this, the Intelligent Solution Control should be enabled (the switch is located in the
Solution Control dialog launched from the Solve task dialog.) Likewise, analyses run with this model may
take more iterations to reach a fully converged solution.

High Reynolds flows that are run with the Low Reynolds turbulence model will generally produce the sam
solution as would the k-epsilon model. Likewise, laminar flows that are run with this model will produce
similar results to a solution run as laminar.

RNG
The RNG turbulence model is more computational intensive, but sometimes slightly more accurate than
k-epsilon model, particularly for separated flows. This model works best for predicting the reattachment
point for separated flows, particularly for flow over a backward-facing step. When using the RNG model,
often recommended to start with the k-epsilon model and after this model is fairly well converged, enable
the RNG model.

Eddy Viscosity
The constant eddy viscosity model is slightly less rigorous than the k-epsilon model, but more numerical
stable. This is a good choice for lower speed turbulent flows and some buoyancy flows. This model is us
if divergence occurs with one of the other models.

Mixing Length
The Mixing Length turbulence model is primarily designed for internal natural convection analyses. Use
the mixing length model, in some cases, has been shown to reduce run times and provide better accura
than the default turbulence model for internal buoyancy-driven flows.

Note that the Mixing Length model is designed for gas flows (such as air), and will not produce good res
when the working fluid is a liquid (such as water).

Additional Parameters
Auto Starup
Auto Startup controls the Automatic Turbulent Start-Up (ATSU) algorithm.
This algorithm goes through a number of steps to obtain turbulent flow solutions. The algorithm starts by
running 10 iterations using a constant eddy viscosity model, so the k and epsilon equations are not solve
With this solution as an initial guess, the two-equation turbulence model is started. At iteration 10, a spik
the convergence monitoring data will appear for the k and epsilon equations. Other steps are then taken
gradually arrive at the converged result. These steps may involve spikes in the convergence monitoring
data at iterations 10, 20 and 50. After 50 iterations, the ATSU is turned off automatically.

If Lock On is selected, the ATSU stays on during the entire analysis until the user manually clicks it off. If
there are convergence difficulties after iteration 50 (divergence within 10 iterations), then you should ena
Lock On. If the ATSU is turned on, you should run at least 200 iterations to ensure convergence of the
turbulent flow solution.

If Extend is selected, an extended version of the ATSU is activated. This method is useful for difficult
analyses, particularly compressible analyses. The minimum number of iterations that should be run with
algorithm is 400.

Turb/Laminar Ratio
The Turb/Laminar Ratio is the ratio of the effective (turbulent) viscosity to the laminar value. It is used to
estimate the effective viscosity at the beginning of the turbulent flow analysis. In most turbulent flow
analyses, the effective viscosity is 2-3 orders of magnitude larger than the laminar value. The default val
is generally suitable for most flows.

For the Mixing Length model, the turb/lam ratio is the upper limit for the eddy viscosity. The free stream
eddy viscosity maxes out at this value.

For the eddy viscosity model, this is the eddy viscosity, even if you change it on a restart now.

For all the other turbulence models (K-Epsilon, RNG, Low Re Number), the specified value is the startin
point or initial value of the eddy viscosity.

It is often helpful to increase the Turb/Lam Ratio to 1000 or even 10,000 for flows that feature a small, h
speed jet shooting into a large plenum. Such flows are typically momentum-driven, and benefit from a la
turbulent viscosity at the beginning of the calculation.

Turbulence Intensity
The Turbulence Intensity Factor controls the amount of turbulent kinetic energy in the inlet stream. Its
default value is 0.05 and should rarely exceed 0.5. The expression used to calculate turbulent kinetic
energy at the inlet is:
I is the Intensity Factor and u, v and w are velocity components.