Lecture 4: Large Eddy Simulation (LES) and

Other Scale Resolving Simulation (SRS)

Models
15.0 Release

Turbulence Modeling Using ANSYS Fluent

1

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Topics to be Discussed
Unsteady Turbulent Flow Calculations
Motivation
Challenges
Hybrid RANS/LES Approaches
Large Eddy Simulation (LES)
Detached Eddy Simulation (DES)
Filtering
Scale Adaptive Simulation (SAS)
Subgrid models
Embedded LES (ELES)
Grid size and time step size
Wall Modeled LES (WMLES)
Boundary Conditions
Wall issues
Modeling approaches that can capture the unsteady motion of a range of different
turbulent scales (turbulent content) are referred to as Scale Resolving Simulations (SRS)
2

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Motivation for Scale Resolving Simulations

Many flows of interest require the simulation to be able to resolve the unsteady
motion of the most important turbulent scales

Large scale separation

Acoustics
Small scale processes like micro-mixing or chemical reactions

This motion is inherently unsteady and three-dimensional

Unsteady RANS (URANS) simulations might be able to capture some mean flow
unsteadiness, such as vortex shedding, but are generally incapable of capturing
unsteadiness arising from turbulence

SST k-w model

3

2014 ANSYS, Inc.

April 23, 2014

LES
ANSYS Confidential

Overview of SRS modeling approaches

Large Eddy Simulation (LES)

Detached Eddy Simulation (DES)
Scale Adaptive Simulation (SAS)
Additional Hybrid RANS/LES approaches
Embedded LES (ELES)
Wall modeled LES (WMLES)

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Large Eddy Simulation (LES)

Recall: Two methods can be used to eliminate the need to resolve small scales
Reynolds Averaging Approach: Temporal averaging
All scales are modeled
Periodic and quasi-periodic unsteady flows

Filtering (LES): Spatial averaging

Transport equations are filtered such that only larger eddies need be resolved
Difficult to model large eddies since they are
anisotropic
subject to history effects
dependent upon flow configuration, boundary conditions, etc.
Only smaller eddies are modeled
Typically isotropic and so more amenable to modeling
Deterministic unsteadiness of large eddy motions can be resolved
5

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES and the Energy Cascade

Energy Cascade (Richardson, 1922)

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Spatial Filtering

Energy spectrum against the length scale

ux, t u x, t ux, t

resolved scale

u x, t

ux, t
2

2014 ANSYS, Inc.

April 23, 2014

2
f

ANSYS Confidential

subgrid scale

LES: Spatial Filtering

A random variable, f (x), is filtered using a space- filter function, G

f (x) f (x)G(x, x)dx

D

With the top-hat filter (among others)

1 / V
G(x, x)
0

The filtered variable becomes

f ( x)

for x
otherwis e

1
f (x)dx,
V

Most LES codes use implicit filters

Filter width determined by grid resolution
8

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

x V

Filtering the Navier-Stokes Equations

Filtering the original Navier-Stokes equations gives filtered
Navier-Stokes equations that are the governing equations in LES
ui ui u j
1 p
ui
N-S

equation
t
x j
xi x j x j
Filter

Filtered N-S
equation

ui ui u j
1 p

t
x j
xi x j

Sub-grid scale (SGS) stress

9

2014 ANSYS, Inc.

April 23, 2014

ij ui u j ui u j
ANSYS Confidential

ui ij

x x
j
j

Needs modeling

Effect of SGS Model

Role of LES model:
Eddies cannot be resolved down to
the molecular dissipation limit
Dissipation of turbulent kinetic energy
at grid-resolution limit requires eddy
viscosity
An important role of the SGS model is
to dissipate the energy of the small
scales
Everything of importance should be
resolved

LES DNS

DNS

Log E

ui ui
x j x j

Generation of eddies
Energy transfer

Viscous
dissipation
LES
dissipation

LES tLES

ui ui
x j x j

Grid Cut-off

Log k

Eddy viscosity calculated by the SGS model should ensure energy at the grid cutoff
is dissipated at the appropriate rate
10

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Decaying Isotropic Turbulence

With SGS Model

Without SGS Model

1.00E-01

Exp T1
WALE C=0.5
DSM

SMAG C=0.18

1.00E-02

1.00E-03

Grid size cutoff

1.00E-04
1.00E+00

1.00E+01

1.00E+02

Without SGS model energy is accumulated at small scales (large wave number k)
With SGS models, energy is dissipated at grid resolution limit
11

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: SGS Stress Modeling

Fluent offers several eddy viscosity
sub-grid scale models
Smagorinsky-Lilly model

Wall-Adapting Local Eddy-Viscosity

(WALE) model

Dynamic Smagorinsky-Lilly model

Dynamic Kinetic Energy Transport model
Wall-Modeled LES (WMLES)
WMLES S-Omega
12

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Smagorinsky-Lilly SGS Model

Simple algebraic (zero - equation) model
1
2
ij kk ij 2 Cs S Sij with V 1/ 3 , S 2Sij Sij
3
Smagorinsky constant Cs= 0.1 ~ 0.2
Model relies on a local equilibrium of the sub-grid scales (i.e. local
production-dissipation of sub-grid scales, no transport)
The major shortcoming is that there is no Cs universally applicable to
different types of flow
Difficulty with transitional (laminar) flows
An ad hoc damping is needed in the near-wall region
13

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: WALE SGS Model

Wall-Adapting Local Eddy-Viscosity model
Algebraic (zero - equation) model retains the simplicity of Smagorinskys
model
d d 3/2
Sij Sij
2
SGS Cs
5/2
d d 5/4
S S
S S
ijijijij

near -w all modification

The WALE SGS model adapts to local near-wall flow structure

Wall damping effects are accounted for without using the damping function

explicitly
Correct asymptotic behavior of eddy viscosity near wall

Does not allow for non-equilibrium or transport effects for turbulence in

sub-grid scales
14

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Dynamic Smagorinsky-Lilly SGS Model

Based on the similarity concept and Germanos identity (Germano et al., 1991;
Lilly, 1992)

Assumes local equilibrium of sub-grid scales, scale similarity between the smallest

resolved scales and the sub-grid scales

The model parameter Cs is automatically adjusted using the resolved velocity field

Fluents implementation

Locally dynamic model

Implemented for unstructured meshes (test-filter)
Constant Cs by default clipped between 0 and 0.23
Dynamic SGS models also available for energy and species

Overcomes the shortcomings of the Smagorinskys model

Can handle transitional flows

The near-wall (damping) effects are accounted for
15

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Dynamic Kinetic Energy Transport SGS Model

1- equation (for SGS kinetic energy) model. Kim & Menon (1997)

1
1/ 2
ij kk ij 2Ck k sgs
Sij
3
k sgs
t

u j k sgs
x j

3/ 2
k sgs
ui

ij
C

x j

x j

sgs k sgs

x
j
k

Transport equation for sub-grid scale kinetic energy allows for history
and non-equilibrium effects
Like the dynamic Sgamorinskys model, the model constants ( Ck, C )
are automatically adjusted on-the-fly using the resolved velocity field
16

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Grid and Time-Step Size

LES requires mesh and the time-step sizes sufficiently fine to resolve the
energy-containing eddies

The cost of resolving near-wall region in high-Re wall-bounded flows is very high
The mesh resolution determines the fraction of turbulent kinetic energy directly
resolved
ln E

Energy
spectrum
against the
length scale
2
ln

f
17

2014 ANSYS, Inc.

April 23, 2014

2
ln

ANSYS Confidential

LES: Grid Size

Suppose we want to resolve
80% of the turbulent kinetic
energy

Cumulative TKE against length-scale of eddies

based on the Kolmogorovs energy spectrum

1.0
0.8

k 0.1k
k 0.5 k
k 0.8 k
k 0.9 k

Then, we need to resolve

the eddies whose sizes are
k
larger than roughly half the
k
size of the integral length
scale l0.

0
6.10
1.6
0.42
0.16

0.1
0.0

0.42
18

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

6.1

LES: Grid Size

Integral length scale l0
Turbulent kinetic energy peaks at integral length scale. This scale must be
sufficiently resolved
Crude estimation for l0
Based on size of bluff body
Estimate from correlations
Perform RANS calculation and compute l0=k1.5/
Generally require 5-10 grid cells per integral length scale to obtain proper
LES content (see next slide)
At least 10 would be preferable

19

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Estimating Grid Size from RANS Calculation

Contour plot showing ratio of integral
length scale to cell size
Higher values imply more grid points
per integral length scale, or higher
spatial resolution

Lower limit of contour scale in the

picture is 10. Areas without contours
may require higher grid resolution
during mesh generation for LES

20

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Time Step Size

The temporal resolution should match or exceed the spatial resolution in LES

Lets say we have a cell with largest edge length of x and local average velocity, U
It takes a time interval t for the flow to travel across the cell
The time step should be small enough to provide an adequate temporal resolution
of the flow as it passes through the cell

x
U

The real velocity can be higher than the averaged velocity

Pre-cursor RANS simulation is used for the assessment of t
Good practice to account for differences between instantaneous and
averaged velocities as well as for the errors introduced from RANS by
Ut
x
0.5
t
2U
x
21

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Inlet Boundary Conditions

It is often important to specify a realistic turbulent inflow velocity
for accurate prediction of the (downstream) flow

ui x, t U i x

time averaged

uix, t

coherent random

Fluent offers two specification methods for inflow perturbations,

available at velocity inlets

Spectral Synthesizer
Vortex Method
22

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Inlet Boundary Conditions

Spectral Synthesizer
Based on the work of Celik et al.(2001)

Able to synthesize anisotropic, inhomogeneous turbulence from RANS

results (k-, k-w, and RSM fields)

The velocity-field satisfies the continuity by design, i.e. it is divergencefree
23

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Inlet Boundary Conditions

Vortex Method
Vorticity transport is modeled by distributing and tracking many pointvortices on a plane (Sergent, Bertoglio)

w x, t

t x x , t
k

k 1

Velocity field computed using the Biot-Savart law

1
ux, t
2
24

2014 ANSYS, Inc.

April 23, 2014

x x w xe z
x x

ANSYS Confidential

dx

LES: Initial Conditions

Initial condition for velocity field is generally not
important for statistically steady-steady flows

Patching a realistic turbulent velocity field can

however help shorten the simulation time
substantially to get to a statistically steady state
The spectral synthesizer can be used to
superimpose turbulence on top of the mean
velocity field
TUI command
/solve/initialize/init-instantaneous-vel

25

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Velocity field generated by

turbulence synthesizer for
homogeneous turbulence

LES: Walls & Near-Wall Resolution (1)

Up to this point walls in LES have not been considered
LES can be prohibitively expensive for complex medium to high Reynolds
number flows involving wall boundary layers
Meaning most technical flows
The problem stems from the mesh resolution required to properly resolve
wall boundary layers in LES, typically reported as
y+ = 1 ( NY = 30 ), x+ = 40, z+ = 20
This is called Near-Wall Resolved LES, sometimes referred to as NWR
Because the streamwise and spanwise resolution scales with wall
coordinates ( x+, z+) there is a strong Reynolds number dependency
resulting in excessive mesh requirements for practical flows
26

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES: Walls & Near-Wall Resolution (2)

The challenge with NWR
As Reynolds number increases and the boundary layer becomes thinner, the size
of important energy bearing eddies decreases
In LES, the important energy bearing eddies must be resolved, thus the cost of
maintaining the grid resolution becomes prohibitive

NWR is only currently feasible for low Reynolds number problems

Unlike RANS, both the wall normal and wall parallel spacing (both directions) must also

decrease to resolve smaller eddies, so the number of grid points increases in all 3
directions
As size of boundary layer cells decreases, so also does the size of the time step needed
to maintain local cell Courant numbers less than 1
It is estimated that it should be possible to perform LES with NWR on an airplane wing
by 2045 (Spalart, 2000)

Therefore, some form of near-wall modeling (NWM) is required for many cases in
which it would be desirable to perform LES
27

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES Wall Modeling in Fluent

Wall Functions in LES
The near-wall turbulence is explicitly calculated inside the boundary layer, but not necessarily
down to the laminar sublayer
The first grid point can be located at 20 < y+ < 150
The log law is used to relate instantaneous wall shear stress to wall parallel instantaneous
velocity at the centroid of wall adjacent cells
Werner-Wengle (cheaper) wall functions also available through TUI
/define/models/viscous/near-wall-treatment/werner-wengle-wall-fn?

Wall functions, or other wall layer modeling approaches may be satisfactory in simple
flows, but in many cases improved approaches involving hybrid RANS/LES models are
desirable:

28

Detached Eddy Simulation (DES)

Scale Adaptive Simulation (SAS)
Embedded LES (ELES)
Wall-Modeled LES (WMLES)

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Detached Eddy Simulation (DES)

Goal of DES is to produce a hybrid modeling approach combining the benefits
of RANS and LES while minimizing their disadvantages

RANS
Can achieve good predictions for attached boundary layers
Does not capture unsteady motions of large eddies, even if spatial, temporal resolution
permits

LES
Achieves good prediction of separated regions
Becomes prohibitively expensive in boundary layer with increasing Reynolds number as size
of energy-bearing eddies that need to be resolved decreases
LES region

RANS region

29

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

DES Model Formulation

Example: Spalart-Allmaras model based DES (Spalart et al., 1997)
2

D~
1
~
~~

~
~
Cb1S Cw1 f w

...
Dt
~ x j

x
d

d min d w , CDES

One-equation SGS turbulence model with two regions in flow calculation

Near walls, the flow calculation reduces to unsteady RANS with the S-A model
In the high-Re turbulent core region, where large turbulence scales play a
dominant role, DES recovers LES with a one-equation model for the sub-grid
scale viscosity

DES also available with SST and Realizable k- models

Appropriate turbulent length scale, i.e. d ~ k1/2/w , k3/2/ , for selected model
used to determine RANS/LES boundary.
30

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Delayed Detached Eddy Simulation (DDES)

~
Grid B : d Cdes

~
Grid A : d d

In the original DES formulation, the LES mode can be activated inside the boundary
layer for <<

This can have undesirable side effects such as Grid Induced Separation (GIS) where the flow
separates too far upstream of the actual separation point

In delayed detached eddy simulation (DDES) the length scale is redefined

~
d d f d max 0, d Cdes ;

fd f

, , S
t

ij

, , d

The use of additional physical quantities prevent LES mode activation inside the
boundary layer except in very extreme cases
31

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Delayed Detached Eddy Simulation (DDES)

Implemented for all DES options in
ANSYS CFD codes
The use of DDES is recommended
for all DES simulations with DES-SA
or DES-RKE

With DES-SST the SST blending

functions can be used

The DDES function is

32

recommended in R14 and higher

In earlier versions, use F2

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

IDDES
Improved Delayed Detached Eddy Simulation (IDDES) is a new
formulation of the DES model with the following goals
Provide improved shielding against grid induced separation (GIS)
compared to DDES
Provide a new subgrid length scale definition that will allow the model to
run in wall-modeled LES (WMLES) mode when unsteady inlet conditions
are available, e.g. using vortex method with accurate flow and turbulence
profiles

The IDDES implementation is based on SST-DES

The IDDES subgrid length scale definition is also used in the
WMLES model
33

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Detached Eddy Simulation (DES)

Useful to relax the strong LES mesh constraints close to the wall
Mesh constraints relaxed in wall-parallel directions
The near-wall cell should still have y+ = 1 and a sufficient number of grid
points with moderate expansion ratio in the wall normal direction to
resolve the boundary layer

RANS/LES interface depends on the mesh

Need sufficient cells close to wall for resolving RANS layer
All the boundary layer should be covered by RANS
Production mechanisms far from wall must be correctly resolved with LES

34

Grid spacing must be small enough to resolve turbulent eddies with CFL ~ 1
Need to check the RANS/LES interface a posteriori
See next slide

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Checking RANS and LES Regions in DDES

When performing Delayed DES, the variable DES TKE
Dissipation Multiplier can be used to check where the
model is operating in RANS mode

The entire boundary layer should be covered by RANS

Depending on the RANS model, there are different

possibilities

DDES-SA and DDES-RKE

The value of the shielding function fd (Slide 31) is

displayed and is equal to 0 in the RANS region (boundary

layer) and 1 in the LES region
DDES-SST
The value of FDES (below) is displayed and is >1 in the LES
region and 1 in the RANS region.

Dissipation term in TKE equation : Yk *kw FDES

35

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Unsteady Flow Calculations Without LES/DES

RANS models, even in unsteady mode (URANS), are only able to resolve
unsteady structures that are comparable with typical characteristic length
scale

Even if calculation is performed with a grid and time step size that are
capable of resolving smaller scales
RANS with SST k- model

LES

This is often attributed to the time/ensemble averaging procedure

Is that really the case?
36

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Scale Adaptive Simulation (SAS) Model

Transport equations for length scale determining equations (, w) are not
exact scale equations, but simply modeled in a manner analogous to the
k-equation

In the SAS model, an exact transport equation for the integral length scale
is used instead

The exact integral length scale equation is transformed by change of variables

into w for implementation within SST k-w framework, resulting in:

w U j w

S 2 w 2
t
x j
x j

t w 2 1 k w k w w ~

2S 2 L
2
L
x w x x w x x

j
j
j
j
vK
w j

Wilcox Model
37

2014 ANSYS, Inc.

April 23, 2014

BSL (SST) Model

ANSYS Confidential

SAS Model

von Karman Length Scale

The von Karman length scale in the

transformed w equation allows the

model to detect which scales are already
resolved structures in the flow field

LvK

U y
2U y 2

This allows the evolution of a turbulent

spectrum in unstable flow regions whereas
conventional URANS models only predict a
single scale
Turbulent structures in the wake of
a cylinder in crossflow
38

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

SAS Model
If the grid is too coarse, or the time step size is too large, the model may
revert to steady RANS.

The model may not go unsteady for all flows

Can give steady results in attached and mildly separated wall bounded flows
At worst, if this happens, the end result is still a valid RANS solution, which is
not necessarily the case with other hybrid approaches

Unsteady solutions with turbulent content are obtained for flows with
large separation and mixing zones

Many industrial applications are characterized by strong mixing in largely

separated regions

39

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Additional Model Combinations for SRS

SAS can be selected together with any w-based
model except the k-kl-w transition model

Standard k-w, SST, Stress Omega, Transition SST and

Intermittency Transition models

(D)DES can be enabled with the Transition SST and

Intermittency Transition models

First select the transition model in the Viscous Model

panel, then DES appears as an option

40

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Embedded LES (ELES)

Models such as DES and SAS are global approaches which use a single model
formulation in the entire domain

There is a RANS region and a LES region, but they are determined by the model itself
based only on the solution and the grid spacing.
Successful when the flow instability is sufficiently strong to produce unsteadiness in
the calculation

Embedded LES (ELES) is a slightly different concept in that there are still RANS
regions and LES regions, but now the user explicitly defines the boundary
between them

RANS and LES regions are separately defined and use different models
Synthetic turbulence is generated at the interface to convert RANS turbulence to
unsteady, turbulent resolved scales entering the LES region

ELES is not a new turbulence model, but the combination of RANS and LES
models joined by appropriate interface conditions
41

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Coupled Zonal Modeling for ELES

Example: zonal definition for backward facing step
RANS-LES Interface
Acts as B.C. for LES model in Zone 2
LES-RANS Interface
Acts as B.C. for RANS model in Zone 3

ZONE 1
RANS Model

Zonal interfaces can be nonconformal (most common) or

conformal
Definition of model interaction
at interfaces is critical
42

2014 ANSYS, Inc.

April 23, 2014

ZONE 2
LES Model

ANSYS Confidential

ZONE 3
RANS Model

ELES Model Combinations (1)

RANS Models
Any RANS model can be selected in the RANS
zone except Spalart-Allmaras

Does not have two distinct turbulent scales

The same RANS model must be used in all

RANS zones
Selected in Viscous Models panel
LES Models
All algebraic LES models can be selected in the
Embedded LES tab in the Cell Zone conditions panel
Dynamic Smagorinsky model also available

Bounded central differencing recommended for

momentum

Central differencing also available

Discretization used in LES zone is selected in Cell Zone
Conditions panel
43

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

ELES Model Combinations (2)

Zonal Interfaces
RANS-LES Interfaces modeled turbulence needs to be converted to resolved
turbulence

Vortex method (recommended)

Spectral synthesizer

LES-RANS Interfaces
Select No Perturbations in b.c. panel

There is only a single boundary condition type (RANS/LES Interface) available. The
user defines whether it is LES-RANS or RANS-LES through specification of the
Fluctuating Velocity Algorithm
44

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES-RANS Interfaces (1)

At LES-RANS interfaces, the RANS transport equations need to obtain
information from the upstream LES zone. The following possibilities are
available

The RANS model from a precursor RANS simulation is frozen in the LES zone
Recommended for flows where the LES region does not change the overall flow topology
and the LES-RANS interface is outside the region of interest

The SAS model is run passively in the LES zone and then reactivated at the LES-RANS
interface

In the LES zone, the unsteady velocity field is calculated using the LES eddy viscosity
The SAS transport equations are solved too and are based on the LES velocity field. The SAS
t is not used in the momentum equation. The SAS t is reactivated at the interface and
used as the boundary condition for the downstream RANS zone

45

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

LES-RANS Interfaces (2)

One additional possibility for running ELES models

The SAS or DES model is run actively in all zones. Synthetic turbulence is generated
at the interface and propagates downstream
Can be used to trigger SAS into unsteadiness

It is recommended to locate RANS-LES interfaces where there is no backflow

46

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Non-conformal Interfaces in ELES (1)

To create a RANS/LES interface
boundary type at a non-conformal
interface

Define the interface in Define > Mesh

Interfaces
Left click on the interface in the list on
the left of the panel after it has been
created
Note the name of the zone which then
appears under Interface Interior
Zone
47

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

(see next slide)

Non-conformal Interfaces in ELES (2)

(Continued from previous slide)

Select LES zone in cell zone conditions

panel
Or activate SAS or DES models in the
Viscous Models panel
Select the interface interior zone in the
boundary conditions
The list of zone types will now contain
rans-les-interface. Change the type
and enter the appropriate interface
settings
48

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

(see previous slide)

Wall Modeled LES (WMLES) Simulation

WMLES is designed to reduce the mesh resolution requirements of LES for medium
to high Reynolds number wall bounded flows

Boundary layers, channel and pipe flows

WMLES is based on the concept of modeling the flow close to the wall (inner part
of the boundary layer) with RANS and resolving the central part of the boundary
layer with LES

WMLES decides on the wall-parallel RANS/LES interface automatically based on

grid resolution and boundary layer properties
min max CW dW , CW hmax , hWn , hmax ,

dW
hmax
hWn
49

Cw 0.15

distance from the wall

length of the longest edge cell
wall normal edge length of cell

2014 ANSYS, Inc.

April 23, 2014

A hybrid RANS-LES approach with delayed-DES and wall-modelled LES

capabilities, Mikhail L. Shur , Philippe R. Spalart , Mikhail K. Strelets ,
Andrey K. Travin , Int. J of Heat and Fluid Flow 29, 2008
ANSYS Confidential

WMLES Eddy Viscosity

RANS / LES blending

t min dW , CSMAG
S
CSMAG

1 exp y / 25 S

Strain rate magnitude

Smagorinsky constant CSMAG = 0.2
Modified grid spacing (see previous slide)

RANS Model Prandtl mixing length model with van Driest damping
LES Model Smagorinsky model with van Driest damping
The WMLES S-Omega option replaces S with abs(S-W) where W is the vorticity
magnitude

Allows prediction of transitional effects

Improved predictions of separating shear layers
50

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

WMLES Grid Requirements

For a flow with boundary layer thickness, , each boundary layer volume, x
x , in the computational domain, the grid used for WMLES should be as
follows
x = / 10, z = / 20, NY = 40, y+ at first grid point should be less than 1

In wall resolved LES, as Reynolds number increases, it is required to maintain

x+ and z+, such that the total cell count increases (significantly) with
Reynolds number
In WMLES, it is only necessary to maintain x = / 10 and z = / 20 Thus
the required number of grid points is much less sensitive to increasing
Reynolds number

The number of grid points in the wall normal direction will increase to maintain
y+ < 1 at the first grid point
51

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

WMLES Summary
WMLES can be used with IDDES or as a standalone subgrid model for LES
WMLES reduces grid resolution requirements (compared to wall-resolved LES) for

medium to high Reynolds number flows

WMLES grid resolution requirements
x

10

, z

20

, N y 30 40

The time step size should be defined so that the cell Courant number in the boundary

layer is 0.3
WMLES is still very expensive compared to RANS

10 x 40 x 20 grid points per boundary layer volume compared with 1 x 30 x 1 for RANS
Unsteady calculation with low cell Courant number compared to steady state calculation

Keep resolved boundary layer regions as small as possible

52

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Summary of SRS Models (1)

LES

Not suitable if wall boundary layers at high Re

Resolution requirements are excessive for such flows

SAS

Safest hybrid model

Fallback RANS solution if model does not go unsteady

No explicit grid sensitivity in RANS zone

Needs flow instability to generate resolved turbulence

Stays in RANS/URANS mode if instability is missing or too weak

DES

53

Explicit combination of RANS and LES models

For many flows results are similar to SAS
Allows computation of flows with weaker instabilities than needed for SAS
Grid sensitivity in RANS regions requires shielding use DDES as default

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Summary of SRS Models (2)

ELES

Useful to limit LES zone to critical area

Allows manual definition of LES zone

Unlike SAS or DES which depend on both the grid and the solution

Synthetic turbulence generated at RANS-LES interface

Can also be used to trigger SAS/DES into unsteady mode
WMLES

The inner part of the boundary layer is covered by RANS and the central part by LES,

drastically reducing the LES grid requirements for high Re flows

Boundary layer resolution

10 x 40 x 20 grid points for x x boundary layer

Cell courant number in boundary layer = 0.3
Potentially 1.0 for free shear flows

All models are necessary. The question is not Which model is best? but
Which model is best for a given type of application?
54

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential

Best Practices for Scale Resolving Simulations

A Best Practices document for SRS is available from support.ansys.com
Detailed monograph covering theory, model selection and numerical
settings
Enter bpg in the search field on the main landing page

55

2014 ANSYS, Inc.

April 23, 2014

ANSYS Confidential