17.

0 Release

Module 07: Turbulence

Introduction to ANSYS Fluent

1 © 2016 ANSYS, Inc. April 12, 2016

 Introduction
Lecture Theme:
The majority of engineering flows are turbulent. Simulating turbulent flows in Fluent requires
activating a turbulence model, selecting a near-wall modeling approach and providing inlet
boundary conditions for the turbulence model.

Learning Aims:
You will learn:
• How to use the Reynolds number to determine whether the flow is turbulent
• How to select a turbulence model
• How to choose which approach to use for modeling flow near walls
• How to specify turbulence boundary conditions at inlets

Learning Objectives:
You will be able to determine whether a flow is turbulent and be able to set up and solve
turbulent flow problems.
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

2 © 2016 ANSYS, Inc. April 12, 2016

 Observation by Osborne Reynolds

• Flows can be classified as either :

Laminar
(Low Reynolds Number)

Transitional
(Increasing Reynolds Number)

Turbulent
(Higher Reynolds Number)

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

3 © 2016 ANSYS, Inc. April 12, 2016

 Reynolds Number

• The Reynolds number is the criterion used to determine whether the flow is
laminar or turbulent  .U .L
Re L 

• The Reynolds number is based on the length scale of the flow:
L  x, d, d hyd, etc.
• Transition to turbulence varies depending on the type of flow:
• External flow
• along a surface : ReX > 500 000
• around on obstacle : ReL > 20 000
• Internal flow : ReD > 2 300
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

4 © 2016 ANSYS, Inc. April 12, 2016

 Overview of Computational Approaches
• Three basic approaches can be used to calculate a turbulent flow
DNS LES RANS
(Direct Numerical Simulation) (Large Eddy Simulation) (Reynolds Averaged Navier-
Stokes Simulation)

• Numerically solving the full unsteady • Solves the filtered N-S equations • Solve time-averaged N-S equations
Navier-Stokes equations • Some turbulence is directly resolved • All turbulent motion is modeled
• Resolves the whole spectrum of • Less expensive than DNS, but the • For most problems the time-averaged flow
scales efforts and computational resources (and level of turbulence) are all that is
needed are still too large for most needed
• No modeling is required practical applications
• Many different models are available
• But the cost is too prohibitive! • Available in Fluent but not discussed in
• This is the most widely used approach for
Not practical for industrial flows! Introductory Training
industrial flows
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

5 © 2016 ANSYS, Inc. April 12, 2016

 Turbulence Models Available in Fluent

One-Equation Model
Spalart-Allmaras
Two-Equation Models
RANS based Increase in
models k–ε family (Standard, RNG, Realizable*) Computational
k–ω family (Standard, BSL, SST*) Cost
Reynolds Stress Models Per Iteration

Transition Models
k–kl–ω, Transition SST and Intermittency Models
Detached Eddy Simulation * SST k-w and Realizable k-e are
Large Eddy Simulation recommended choices for standard cases
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

6 © 2016 ANSYS, Inc. April 12, 2016

 Turbulence Model Selection: A Practical Approach
• Fluent offers the choice of many turbulence models
‒ Some are used only for very specific applications
‒ Consider advanced training to learn more about the details of
each model

• For getting started

– The Realizable k-e or SST k-w models are recommended choices for
standard cases
– Where highly accurate resolution of boundary layers is critical, such
as applications involving flow separation or finely resolved heat
transfer profiles, SST k-w is preferred
– If only a crude estimate of turbulence is required, the standard k-e
model can be used
• This might occur in problems where the solution depends more
strongly on other physical models or modeling assumptions than
on the turbulence model

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

7 © 2016 ANSYS, Inc. April 12, 2016

 Turbulent Boundary Layer Profiles
• Near to a wall, in the boundary layer, the velocity changes rapidly.
Velocity, U

Distance from Wall, y

• If we plot the same graph again, where:
– Log scale axes are used
– The velocity is made dimensionless, from U/Ut where

– The wall distance is made dimensionless:

• Then we arrive at the graph on the next page. The shape of this is
generally the same for all flows:
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

8 © 2016 ANSYS, Inc. April 12, 2016

 Dimensionless Boundary Layer Profiles

• By scaling the variables near the wall the velocity profile data takes on a
predictable form

Using the non-dimensional

velocity and non-dimensional Linear, u+ = y+
distance from the wall results Logarithmic, u+ = 2.5 ln(y+) + 5.45
in a predictable boundary As the system Reynolds number
layer profile for a wide range increases, the logarithmic region
extends to higher values of y+
of flows

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

9 © 2016 ANSYS, Inc. April 12, 2016

 Turbulent Boundary Layer Regions

• Looking more closely at the plot from the u+ = 2.5 ln(y+) + 5.45
previous slide, distinct regions can be identified
based on the form of the dimensionless
velocity profile (e.g. linear or logarithmic)
• For CFD, the most important are the viscous u+ = y +
sublayer, immediately adjacent to the wall and
the log-layer, slightly further away from the
wall
• Different turbulence models require different
inputs depending on whether the simulation
needs to resolve the viscous sublayer with the
mesh
‒ This is an important consideration in turbulent flow
simulation
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

10 © 2016 ANSYS, Inc. April 12, 2016

 Wall Modeling Strategies
• In the near-wall region, the solution gradients are very high, but accurate calculations in
the near-wall region are paramount to the success of the simulation. The choice is
between:

A) Using Wall Functions

B) Resolving the Viscous Sublayer

(details on next slide)

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

11 © 2016 ANSYS, Inc. April 12, 2016

 Wall Modeling Strategies: Using Wall Functions
• Using Wall Functions
– Wall functions utilize the predictable dimensionless boundary layer profile shown on slides 9 & 10 to determine
conditions at the wall (e.g. shear stress) from conditions (velocity, wall distance) at the centroid of the wall
adjacent mesh cell
• This means the cell should be located in the log-layer
• To locate the first cell in the log-layer, it should typically have a y+ value such that
30 < y+ < 300
• This is a very general guideline, not an absolute rule
– for very high Re, y+ can be higher if still in log layer and for very low (but still turbulent) Re, the log-layer may not extend
far enough away from the wall for the use of wall functions to be valid
– Wall functions should never be used if y+ < 30
– Generally speaking, this is the approach if you are more interested in the mixing in the middle of the domain,
rather than the forces on the wall

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

12 © 2016 ANSYS, Inc. April 12, 2016

 Wall Modeling Strategies: Resolving the Viscous Sublayer
• Resolving the Viscous Sublayer
• First grid cell needs to be at about y+ ≈ 1 and a prism layer mesh with growth rate no higher than ≈ 1.2
should be used
– These are not magic numbers – this guideline ensures the mesh will be able to adequately resolve
gradients in the sublayer
• This will add significantly to the mesh count (see next slide)
• Generally speaking, if the forces or heat transfer on the wall are key to your simulation (aerodynamic
drag, turbomachinery blade performance, heat transfer) this is the approach you will take and the
recommended turbulence model for most cases is SST k-w

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

13 © 2016 ANSYS, Inc. April 12, 2016

 Mesh Resolution Near the Wall
• Fewer nodes are needed normal to the wall when logarithmic-based wall functions are
used (compared to resolving the viscous sublayer with the mesh)

y y
u
u

Logarithmic-based Wall functions Viscous sublayer resolving approach

used to resolve boundary layer used to resolve boundary layer
Boundary layer First node wall distance is reflected by y+ value
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

14 © 2016 ANSYS, Inc. April 12, 2016

 Example in Predicting Near-wall Cell Size
• During the pre-processing stage, you will need to know a suitable size for the first layer of grid cells
(inflation layer) so that Y+ is in the desired range
• The actual flow-field will not be known until you have computed the solution (and indeed it is
sometimes unavoidable to have to go back and remesh your model on account of the computed Y+
values)
• To reduce the risk of needing to remesh, you may want to try and predict the cell size by
performing a hand calculation at the start, for example:
The question is what height (y)
Air at 20 m/s should the first row of grid cells be.
y
 = 1.225 kg/m3 We will use SWF, and are aiming for
 = 1.8x10-5 kg/ms
Flat plate, 1m long Y+  50
VL
• For a flat plate, Reynolds number ( Rel  ) gives Rel = 1.4x106

Recall from earlier slide, flow over a surface is turbulent when ReL > 5x105
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

15 © 2016 ANSYS, Inc. April 12, 2016

 Calculating Wall Distance for a Given y+
• Begin with the definition of y+ and rearrange: • Re is known, so use the definitions to
 Ut y y calculate the first cell height
y 
y C f  0.058 Rel0.2  .0034
 Ut 
• The target y+ value and fluid properties are known,
so we need Ut, which is defined as:
t w  12 C f U 2  0.83 kg/ m  s 2 

t tw
Ut  w Ut   0.82 m/s
 
• The wall shear stress ,tw ,can be found from the skin
friction coefficient, Cf: • We know we are aiming for y+ of 50, hence:
y
t w  1 C f U 
2 y  9x10 -4 m
2
• A literature search suggests a formula for the skin Ut 
friction on a plate1 thus:
our first cell height y should be
approximately 1 mm.
C f  0.058 Rel0.2
1 An equivalent formula for internal flows, with Reynolds number based on the pipe diameter is C = 0.079 Re -0.25
f d
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

16 © 2016 ANSYS, Inc. April 12, 2016

 Limitations of Wall Functions
• In some situations, such as boundary layer separation, logarithmic-based wall functions do
not correctly predict the boundary layer profile

Wall functions applicable Wall functions not applicable

Non-equilibrium wall functions have been developed in Fluent to
address this situation but they are very empirical. Resolving the
viscous sublayer with the mesh is recommended if affordable

• In these cases logarithmic-based wall functions should not be used

• Instead, directly resolving the viscous sublayer with the mesh can provide accurate results

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

17 © 2016 ANSYS, Inc. April 12, 2016

 Turbulence Settings for Near Wall Modeling
• If the viscous sublayer is being resolved
– Use k-w models or k-e models with Enhanced Wall Treatment
(EWT)
– No separate input is needed for k-w models
• If wall functions are used
– Use k-e models with wall functions
• EWT can also be used because it is a y+ insensitive method
and will act like a wall function if the first grid point is in
the log-layer
– For k-w models
• The k-w models utilize a y+ insensitive wall treatment and
will act like a wall function if the first grid point is in the log
layer
• However, the advantages of these models may be lost
when a coarse near-wall mesh is used
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

18 © 2016 ANSYS, Inc. April 12, 2016

 Inlet Boundary Conditions
• When turbulent flow enters a domain at inlets or outlets (backflow), boundary conditions
must be given for the turbulence model variables
• Four methods for specifying turbulence boundary conditions:
1) Turbulent intensity and viscosity ratio (default)
Default values of turbulent intensity = 5% and turbulent viscosity ratio = 10 are reasonable
for cases where you have no information about turbulence at an inlet
2) Turbulent intensity and length scale
• Length scale is related to size of large eddies that contain most of energy
– For boundary layer flows: l  0.4δ99
– For flows downstream of grid: l  opening size
3) Turbulent intensity and hydraulic diameter (primarily for internal flows)
4) Explicitly input k, ε, ω, or Reynolds stress components (this is the only method that
allows for profile definition)
Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

19 © 2016 ANSYS, Inc. April 12, 2016

 Guidelines for Inlet Turbulence Conditions
• If you have absolutely no idea of the turbulence levels in your simulation, you could use
following values of turbulence intensities and viscosity ratios:

– Normal turbulent intensities range from 1% to 5%

– The default turbulent intensity value 5% is sufficient for nominal turbulence through a
circular inlet, and is a good estimate in the absence of experimental data
– For external flows, turbulent viscosity ratio of 1-10 is typically a good value
– For internal flows, turbulent viscosity ratio of 10-100 it typically a good value
• For fully developed pipe flow at Re = 50,000, the turbulent viscosity ratio is around
100

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

20 © 2016 ANSYS, Inc. April 12, 2016

 Summary – Turbulence Modeling Guidelines
• To perform a turbulent flow calculation in Fluent
‒ Calculate the Reynolds number and determine whether flow is turbulent.
‒ Decide on a near-wall modeling strategy
• The choices are A) Resolve the viscous sublayer or B) Use wall functions
• Create the mesh with y+ suitable for the selected approach
‒ Choose turbulence model and near wall treatment (if necessary) in the Viscous Models panel
• Realizable k-e or SST k-w are recommended choices for standard cases
• SST k-w is preferred for cases where the viscous sublayer needs to be resolved (flow separation,
detailed heat transfer)
‒ Set reasonable boundary conditions for the turbulence model variables

Introduction Reynolds Number Models Near-Wall Treatments Inlet BCs Summary

21 © 2016 ANSYS, Inc. April 12, 2016

 Appendix

22 © 2016 ANSYS, Inc. April 12, 2016

 RANS Turbulence Model Descriptions
Model Description
Spalart – A single transport equation model solving directly for a modified turbulent viscosity. Designed specifically for aerospace applications involving
wall-bounded flows on a fine near-wall mesh. Fluent’s implementation allows the use of coarser meshes. Option to include the strain rate in the
Allmaras production term improves predictions of vortical flows.

The baseline two-transport-equation model solving for k and ε. This is the default k–ε model. Coefficients are empirically derived; valid for fully
Standard k–ε turbulent flows only. Options to account for viscous heating, buoyancy, and compressibility are shared with other k–ε models.

A variant of the standard k–ε model. Equations and coefficients are analytically derived. Significant changes in the ε equation improves the
RNG k–ε ability to model highly strained flows. Additional options aid in predicting swirling and low Reynolds number flows.

A variant of the standard k–ε model. Its “realizability” stems from changes that allow certain mathematical constraints to be obeyed which
Realizable k–ε ultimately improves the performance of this model.
A two-transport-equation model solving for k and ω, the specific dissipation rate (ε / k) based on Wilcox (1998). This is the default k–ω model.
Standard k–ω Demonstrates superior performance to k–ε models for wall-bounded and low Reynolds number flows. Options account for low Reynolds number
effects, free shear, and compressible flows.

A variant of the standard k–ω model. Combines the original Wilcox model for use near walls and the standard k–ε model away from walls using a
SST k–ω blending function. Also limits turbulent viscosity to guarantee that τT ~ k.

Reynolds stresses are solved directly using transport equations, avoiding isotropic viscosity assumption of other models. Use for highly swirling
RSM flows. Quadratic pressure-strain option improves performance for many basic shear flows.

* Realizable k-e or SST k-w are the recommended choice for standard cases
23 © 2016 ANSYS, Inc. April 12, 2016
 RANS Turbulence Model Usage
Model Behavior and Usage
Economical for large meshes. Good for mildly complex (quasi-2D) external/internal flows and boundary layer flows under pressure
Spalart-Allmaras gradient (e.g. airfoils, wings, airplane fuselages, missiles, ship hulls). Performs poorly for 3D flows, free shear flows, flows with strong
separation.
Robust. Widely used despite the known limitations of the model. Performs poorly for complex flows involving severe pressure gradient,
Standard k–ε separation, strong streamline curvature. Suitable for initial iterations, initial screening of alternative designs, and parametric studies.

Suitable for complex shear flows involving rapid strain, moderate swirl, vortices, and locally transitional flows (e.g. boundary layer
Realizable k–ε* separation, massive separation, and vortex shedding behind bluff bodies, stall in wide-angle diffusers, room ventilation).

Offers largely the same benefits and has similar applications as Realizable. Possibly harder to converge than Realizable.
RNG k–ε
Superior performance for wall-bounded boundary layer, free shear, and low Reynolds number flows compared to models from the k-e
Standard k–ω family. Suitable for complex boundary layer flows under adverse pressure gradient and separation (external aerodynamics and
turbomachinery). Separation can be predicted to be excessive and early.
Offers similar benefits as standard k–ω. Not overly sensitive to inlet boundary conditions like the standard k–ω. Provides more accurate
SST k–ω* prediction of flow separation than other RANS models.
Similar to SST k-w. Good for some complex flows if SST model is overpredicting flow separation
BSL k–ω
Physically the most sound RANS model. Avoids isotropic eddy viscosity assumption. More CPU time and memory required. Tougher to
RSM converge due to close coupling of equations. Suitable for complex 3D flows with strong streamline curvature, strong swirl/rotation (e.g.
curved duct, rotating flow passages, swirl combustors with very large inlet swirl, cyclones).

* Realizable k-e or SST k-w are the recommended choice for standard cases
24 © 2016 ANSYS, Inc. April 12, 2016
 Example #1 – Turbulent Flow Past a Blunt Flat Plate

• Turbulent flow past a blunt flat plate was simulated using four different
turbulence models.
– 8,700 cell quad mesh, graded near leading edge and reattachment location.
– Non-equilibrium boundary layer treatment
xR
U0
ReD  50,000

Recirculation zone Reattachment point

N. Djilali and I. S. Gartshore (1991), “Turbulent Flow Around a Bluff Rectangular Plate, Part
I: Experimental Investigation,” JFE, Vol. 113, pp. 51–59.
25 © 2016 ANSYS, Inc. April 12, 2016
 Example #1 Turbulent Flow Past a Blunt Flat Plate
Contours of Turbulent Kinetic Energy (m2/s2)
0.70
0.63
0.56
Standard k–ε RNG k–ε
0.49
0.42
0.35
0.28
0.21
0.14
Realizable k–ε Reynolds Stress
0.07
0.00
The standard k-e model greatly overpredicts the production of turbulence at stagnation
points, which can lead to qualitatively inaccurate predictions, as seen on the next slide.
26 © 2016 ANSYS, Inc. April 12, 2016
 Example #1 Turbulent Flow Past a Blunt Flat Plate

Predicted separation bubble:

Skin
Standard k–ε (SKE) Friction
Coefficient
Cf × 1000

Realizable k–ε (RKE)

Distance Along Plate, x / D

SKE severely underpredicts the size of the

separation bubble, while RKE predicts the Experimentally observed reattachment
size exactly. point is at x / D = 4.7

27 © 2016 ANSYS, Inc. April 12, 2016

 Example #2 : Pipe Expansion with Heat Transfer

• Reynolds Number ReD= 40750

• Fully Developed Turbulent Flow at Inlet

• Experiments by Baughn et al. (1984) q"=const

H q"=0
D Outlet
Inlet d
axis
H 40 x H
28 © 2016 ANSYS, Inc. April 12, 2016
 Example #2 : Pipe Expansion with Heat Transfer
• Plot shows dimensionless distance versus Nusselt Number
• Best agreement is with SST and k-omega models which do a better job of capturing flow
recirculation zones accurately

29 © 2016 ANSYS, Inc. April 12, 2016

 Example #3 Turbulent Flow in a Cyclone

• 40,000-cell hexahedral mesh 0.1 m

0.12 m

• High-order upwind scheme was used.

Uin = 20 m/s

0.2 m
• Computed using SKE, RNG, RKE and RSM
(second moment closure) models with the 0.97 m
standard wall functions

• Represents highly swirling flows (Wmax = 1.8 Uin)

30 © 2016 ANSYS, Inc. April 12, 2016

 Example #3 Turbulent Flow in a Cyclone
• Tangential velocity profile predictions at 0.41 m below the vortex finder
The cyclone is dominated by strongly
swirling flow. The Reynolds stresses
become anisotropic in swirling flows.
Two equation turbulence models such
as k-e and k-w based models assume
that turbulence is always isotropic
and thus the accuracy of these
models suffers in swirling flows. The
Reynolds stress model makes no
assumption of isotropic behavior and
is thus better suited for applications
where swirl is the dominant flow
feature

31 © 2016 ANSYS, Inc. April 12, 2016

 Example 4: Diffuser
Shear Stress Transport (SST) Model
• It accounts more accurately for the transport of the turbulent shear stress, which
improves predictions of the onset and the amount of flow separation compared to k-e
models

Standard k-e fails to predict separation

SST result and experiment

Experiment Gersten et al.

32 © 2016 ANSYS, Inc. April 12, 2016

 Turbulent Flow Characteristics
• Turbulent flow characteristics :
• Unsteady, three-dimensional, irregular, stochastic motion in which transported quantities (mass,
momentum, scalar species) fluctuate in time and space
• Enhanced mixing of these quantities results from the fluctuations
• Unpredictability in detail
• Turbulent flows contain a wide range of eddy sizes
• Large scale coherent structures are different in each flow, whereas small eddies are more
universal

33 © 2016 ANSYS, Inc. April 12, 2016

 Energy Cascade

• Energy is transferred from larger eddies to smaller eddies

• Larger eddies contain most of the energy
• In the smallest eddies, turbulent energy is converted to internal energy by viscous
dissipation

Energy Cascade Richardson

(1922), Kolmogorov (1941)

34 © 2016 ANSYS, Inc. April 12, 2016

 Backward Facing Step
• As engineers, in most cases we do not actually need to see an exact snapshot of the
velocity at a particular instant.
• Instead for most problems, knowing the time-averaged velocity (and intensity of the
turbulent fluctuations) is all we need to know. This gives us a useful way to approach
modelling turbulence.
Instantaneous velocity contours

Time-averaged velocity contours

35 © 2016 ANSYS, Inc. April 12, 2016

 Mean and Instantaneous Velocities
• If we recorded the velocity at a particular point in the real (turbulent) fluid flow,
the instantaneous velocity (U) would look like this:
u Fluctuating velocity

Velocity U Time-average of velocity

U Instantaneous velocity

Time
• At any point in time: U  U  u
• The time average of the fluctuating velocity must be zero: u  0
• BUT, the RMS of u  is not necessarily zero: u 2  0
1

• The turbulent energy, k, is given by the fluctuating velocity components as:k  u2  v2  w2
2

36 © 2016 ANSYS, Inc. April 12, 2016
 RANS Modeling : Averaging

• After decomposing the velocity into mean and instantaneous parts and time-
averaging, the instantaneous Navier-Stokes equations may be rewritten as the
Reynolds-averaged Navier-Stokes (RANS) equations:
 ui ui  p   ui   Rij Rij  uiuj
  uk      
 
 t xk  xi x j  x j  x j (Reynolds stress tensor)
• The Reynolds stresses are additional unknowns introduced by the averaging
procedure, hence they must be modeled (related to the averaged flow quantities) in
order to close the system of governing equations
   u '2   u ' v '   u ' w ' 
 t xx t xy t xz   
 
 6 unknowns … Rτ ij t yx 
t yyuiu     u ' v '   v '2   v ' w ' 
t yzj 
t   
 zx t zy t zz     u ' w '   v ' w '   w ' 
2
 
37 © 2016 ANSYS, Inc. April 12, 2016
 RANS Modeling : The Closure Problem
• The Reynolds Stress tensor Rij  uiuj must be solved
• The RANS equations can be closed in two ways:
Reynolds-Stress Models (RSM) Eddy Viscosity Models
• Rij is directly solved via transport equations • Boussinesq hypothesis
(modeling is still required for many terms in the Reynolds stresses are modeled using an eddy (or
transport equations) turbulent) viscosity, μT


t

uiuj 

xk
  
 uk uiuj  Pij  Fij  DijT   ij  eij
 u u j  2 uk
Rij   uiuj   T  i    T
 x x  3 x
2
ij   k ij
3
 j i  k

• RSM is advantageous in complex 3D turbulent flows with • The hypothesis is reasonable for simple turbulent shear flows:
large streamline curvature and swirl, but the model is more boundary layers, round jets, mixing layers, channel flows, etc.
complex, computationally intensive, more difficult to
converge than eddy viscosity models
• Note: All turbulence models contain empiricism
• Equations cannot be derived from fundamental principles
• Some calibrating to observed solutions and “intelligent guessing” is contained in the models

38 © 2016 ANSYS, Inc. April 12, 2016

 Two-Equation Models
• Two transport equations are solved, giving two independent scales
for calculating t
– Virtually all use the transport equation for the turbulent kinetic energy, k
Dk    t  k 
    
 
   P  e ; P   S 2
( ske) S  2Sij Sij
Dt x j 
  k  x j  t

production dissipation

– Several transport variables have been proposed, based on dimensional arguments, and used for
second equation. The eddy viscosity t is then formulated from the two transport variables.
– Kolmogorov, w: t  k / w, l  k1/2 / w, k  e / w
• w is specific dissipation rate
• defined in terms of large eddy scales that define supply rate of k
– Chou, e: t  k2 / e, l  k3/2 / e
– Rotta, l: t  k1/2l, e  k3/2 / l

39 © 2016 ANSYS, Inc. April 12, 2016

 Standard k-e Model Equations

k-transport equation
Dk   t  k 
  
   
 x    S 2
 e ; S  2 Sij Sij
Dt x j 
 k  j

t

production dissipation
e-transport equation

De


x j


  
t
e
 e  e

 x   C1e t S 2
 C2e e 
Dt 
  
j k

inverse time scale

coefficients
 k ,  e , Cie , Ce 2 Empirical constants determined from benchmark
experiments of simple flows using air and water.0
k2
turbulent viscosity  t   C
e

40 © 2016 ANSYS, Inc. April 12, 2016

 RANS : EVM :Standard k–ε (SKE) Model
• The Standard K-Epsilon model (SKE) is the most widely-used engineering turbulence
model for industrial applications
– Model parameters are calibrated by using data from a number of benchmark
experiments such as pipe flow, flat plate, etc.
– Robust and reasonably accurate for a wide range of applications
– Contains submodels for compressibility, buoyancy, combustion, etc.

• Known limitations of the SKE model:

– Performs poorly for flows with larger pressure gradient, strong separation, high swirling
component and large streamline curvature.
– Inaccurate prediction of the spreading rate of round jets.
– Production of k is excessive (unphysical) in regions with large strain rate (for example,
near a stagnation point), resulting in very inaccurate model predictions.

41 © 2016 ANSYS, Inc. April 12, 2016

 RANS : EVM: Realizable k-epsilon
• Realizable k–ε (RKE) model (Shih):
– Dissipation rate (ε) equation is derived from the mean-square vorticity fluctuation, which is
fundamentally different from the SKE.
– Several realizability conditions are enforced for Reynolds stresses.

– Benefits:
• Accurately predicts the spreading rate of both planar and round jets
• Also likely to provide superior performance for flows involving rotation, boundary layers
under strong adverse pressure gradients, separation, and recirculation

OFTEN PREFERRED TO STANDARD K-EPSILON.

42 © 2016 ANSYS, Inc. April 12, 2016

 RANS : EVM : Spalart-Allmaras (S-A) Model
• Spalart-Allmaras is a low-cost RANS model solving a single transport equation for a
modified eddy viscosity

• Designed specifically for aerospace applications involving wall-bounded flows

– Has been shown to give good results for boundary layers subjected to adverse pressure gradients.
– Used mainly for aerospace and turbomachinery applications

• Limitations:
– The model was designed for wall bounded flows and flows with mild separation and recirculation.
– No claim is made regarding its applicability to all types of complex engineering flows.

43 © 2016 ANSYS, Inc. April 12, 2016

 k-omega Models
• In k-w models, the transport equation for the turbulent dissipation rate, e, is replaced
with an equation for the specific dissipation rate, w
– The turbulent kinetic energy transport equation is still solved

• k-w models have gained popularity in recent years mainly because:

– Much better performance than k-e models for boundary layer flows
• For separation, transition, low Re effects, and impingement, k-w models are more accurate than k-e models

– Accurate and robust for a wide range of boundary layer flows with pressure gradient

• Two variations of the k-w model are available in Fluent

– Standard k-w model (Wilcox, 1998)
– SST k-w model (Menter)

44 © 2016 ANSYS, Inc. April 12, 2016

 k-omega Model
• k-w models are RANS two-equations based models
 ui u j  2
Rij   uiuj   T      T uk ij  2  k ij
 x xi  3 xk 3
k  j 
t    
w
Dk ui   t  k 
ω = specific dissipation rate
  t ij     f k w  
    
Dt x j x j 
  k  x j 
 e 1
w 

Dw w
  t ij
ui
   f w 2 
 

  
t  w 
 k t
x j x j w  x 
Dt k 
  j 


• One of the advantages of the k-w formulation is the near wall treatment for low-
Reynolds number computations
– designed to predict correct behavior when integrated to the wall
• the k-w models switches between a viscous sublayer formulation (i.e. direct resolution of the boundary layer) at low y+ values and
a wall function approach at higher y+ values
– while k-e model variations require Enhanced Wall Treatment to capture correct viscous sublayer behavior

45 © 2016 ANSYS, Inc. April 12, 2016

 SST Model
• Shear Stress Transport (SST) Model
• The SST model is a hybrid two-equation model that combines the advantages of both k-e and
k-w models
– The k-w model performs much better than k-e models for boundary layer flows
– Wilcox’ original k-w model is overly sensitive to the freestream value (BC) of w, while the k-e model is not prone to
such problems

k-e
k-w
Wall

• The k-e and k-w models are blended such that the SST model functions like the k-w close to the wall and the k-e model in
the freestream

SST is a good compromise between k-e and k-w models

46 © 2016 ANSYS, Inc. April 12, 2016

 RANS: Other Models in Fluent
• RNG k-e model
– Model constants are derived from renormalization group (RNG) theory instead of empiricism
– Advantages over the standard k-e model are very similar to those of the RKE model

• Reynolds Stress model (RSM)

– Instead of using eddy viscosity to close the RANS equations, RSM solves transport equations for the individual
Reynolds stresses
• 7 additional equations in 3D, compared to 2 additional equations with EVM.
– Much more computationally expensive than EVM and generally very difficult to converge
• As a result, RSM is used primarily in flows where eddy viscosity models are known to fail
• These are mainly flows where strong swirl is the predominant flow feature, for instance a cyclone

47 © 2016 ANSYS, Inc. April 12, 2016

 Enhanced Wall Treatment (EWT)
• Need for y+ insensitive wall treatment

• EWT smoothly varies from low-Re to

wall function with mesh resolution

• EWT available for k-e and RSM models

• Similar approach implemented for k-w

equation based models, and for the
Spalart-Allmaras model
The term "y+ insensitive wall treatment" does not mean the results will be identical
no matter what the value of y+ is at the wall-adjacent cell. It means that as you
refine the mesh, the solution will tend gradually towards grid independence. This is
in contrast to wall functions, where the solution can be extremely sensitive to y+
values and grid independent solutions can be difficult to achieve.
48 © 2016 ANSYS, Inc. April 12, 2016
 y+ for the SST and k-omega Models
• The SST and k-w models were formulated to be near-wall resolving models
where the viscous sublayer is resolved by the mesh
– To take full advantage of this formulation, y+ should be ≈ 1
– This is necessary for accurate prediction of flow separation

• These models can still be used with a coarser near-wall mesh and produce
valid results, within the limitations of logarithmic wall functions
– The first grid point should still be in the logarithmic layer (y+ < 300 for most
flows)
– Many advantages of these models may be lost when a coarse near-wall
mesh is used

49 © 2016 ANSYS, Inc. April 12, 2016

 Turbulent Scales Related to k and e
• Characteristics of the Turbulent Structures:
• Length scale : l [m]
• Velocity scale : k [m/s] ui x, t   U i x, t   uix, t 
l
• Time scale : [s] Instantaneous Time-average Fluctuating
k component component component

- Turbulent kinetic energy :


1 2
k  u '  v '2  w '2
2
 [m2/s2]

- Turbulent kinetic energy dissipation : e [m2/s3] ~ k3/2/l (dimensional analysis)

- Turbulent Reynolds : Ret = k1/2l/n ~ k2/ne [-]

u 1 2 k [-]
- Turbulent Intensity : I   20 %
U U 3

50 © 2016 ANSYS, Inc. April 12, 2016