Spotlight On... Turbulence

Spotlight
On
Turbulence
STAR-CCM+ v11.06
Turbulence Modeling Overview
In this Spotlight we examine:

What turbulence is;
Why it matters; and
What turbulence modeling options are available in STAR-CCM+
For most engineering analyses, users wish to predict the mean (averaged) flow field
Notable exceptions are aero-acoustics and combustion
These require different flow decomposition (filter)
However, accounting fluctuations about this mean (or filtered) flow is key to calculate:
Loss, drag, heat transfer, acoustics, mixing, and other flow characteristics of interest
Effect of these fluctuations is included via a model
This is turbulence modeling
Unrestricted Siemens AG 2016

Page 2 26.10.2016 Siemens PLM Software
What is Turbulence?
Turbulence occurs when inertial forces greater than viscous forces

This ratio is called the Reynolds number
Characterised by chaotic flow motion
Can be viewed as a superimposition of a mean flow and fluctuations
around this mean state
In CFD, turbulence models provide a means of accounting for the effects of the
turbulent fluctuations.
The definition of this mean flow state and its fluctuations is the basis of
turbulence modeling.

Turbulence Energy Spectrum
Turbulence affects mean flow by transporting and dissipating energy

Cannot be ignored
Energy is withdrawn from mean flow into largest scale eddies
Large eddies break-up into progressively smaller eddies
Energy dissipated at the smallest scales
Process known as the Energy Cascade
Energy
Containing Inertial
Eddies Sub-range
log E(k)
Dissipating
Eddies
log k (wavenumber)

Consequences of Turbulence
Turbulence is instrumental in the momentum mixing and heat transfer between

different parts of the flow field.
Momentum transfer between high-speed free-stream flow and low-speed
boundary layer leads to increased drag
For separated flows, turbulence increases momentum mixing
For rounded bluff objects, turbulence re-energizes the boundary layer, delaying
separation, thus reducing the overall drag.

Turbulence Why Does it Matter?
() = () + ()
Time- and space-dependent flow variables such as velocity can be decomposed,

by averaging or filtering
Component that can be resolved f(x,t)
Component that cannot (f(x,t))
Effect of unresolved structures on resolved modeled as a force
e.g. Reynolds stress for RANS models
Energy of modeled scales significant
Can represent as much as 30% of the total energy of the system
Modeled vs. Resolved Turbulence
Three averaging/filtering methodologies

Reynolds Averaged Navier Stokes (RANS)
All turbulent scales are modeled
Large Eddy Simulation (LES)
The largest eddies are resolved, the smallest eddies are modeled
Direct Numerical Simulation (DNS)
All scales of turbulence resolved (no averaging/filtering, thus no modeling needed)
Kolmogorov
Taylor Microscale, Lengthscale,
k 3
Energy Containing
Eddies Inertial Sub-
log E(k)
range
Dissipating
Resolved Eddies
Modeled
Unrestricted Siemens AG 2016 log k (wavenumber)

Why not Direct Numerical Simulation (DNS)?
DNS is fully resolved CFD

Example: F1 car at 140mph in 10m x 10m x 20m wind tunnel
Kolmogorov length scale (of smallest eddies) = ( 3/ )
Freestream mesh size 0.1mm
Assuming 4 cells in each direction required to resolve an eddy
Need 5.5e13 cells per cubic meter
~100 quadrillion cells
DNS of industrial problems is not possible
Available compute power many orders of magnitude less than required

Turbulence Modeling is Here to Stay
DNS only possible for simple cases

Such as a (very small, low velocity flow) backward facing step
As such it provides a useful tool to calibrate turbulence models
Without DNS for Industrial CFD problems we must use turbulence modeling to
close the Navier-Stokes equations
Good choice of turbulence model critical to achieving accurate results at

acceptable computational cost

Turbulence Model Overview in STAR-CCM+
Reynolds Averaged (Turbulence Modeled)

Spalart-Allmaras (1eqn)
K-Epsilon (2eqn) Standard, Realizable, Low Re, Non Linear
V2F (Elliptic-Relaxation), Elliptic Blending K-Epsilon (4eqn)
K-Omega (2eqn) Standard and SST
Reynolds Stress Transport (many equation)
DES (Turbulence Partially Resolved)

Spalart-Allmaras
K-Omega (SST)
Elliptic Blending K-Epsilon
LES (Turbulence Partially Resolved)

Smagorinsky
Dynamic Smagorinsky
WALE
RANS Turbulence Models
Eddy Viscosity Assumption

Six Reynolds Stresses needed to close the Navier Stokes Equations (in 3D)

Simplification reduces computational expense
Boussinesq Hypothesis used in RANS models

Defines Reynolds Stresses as function of turbulent viscosity and mean strain rate
Tt = 2t S2/3.(t v + k )I
Model turbulent viscosity () to close the NS equations
To model solve for one or more transported quantities, for example:-

Modified turbulent viscosity (Spalart Allmaras)
Turbulent kinetic energy, k, and Turbulent Dissipation, e, = 2/
Turbulent kinetic energy, k, and Specific Turbulent Dissipation, w, = /
.
Inherent assumption of isotropy

Spalart-Allmaras
Summary
A one-equation model
Widely used in aerospace industry
Advantages
Generally simple, economical, robust on good meshes
Valid in the near-wall region
Good predictions for attached flow
Disadvantages
Under-predicts separation
Insufficient scales for some combustion/Lagrangian models (e.g., EBU)
Cannot simulate turbulence decay

Standard K-Omega
Summary
Two equation model
Widely used in aerospace, Formula 1, and turbomachinery industries
Advantages
Performs well for swirling flows (for a two equation model)
Does not require damping functions for resolving sublayer
Performs well for adverse pressure gradients
Disadvantages
Sensitive to inlet/freestream turbulence boundary conditions although modifications to the
original Wilcox model have reduced this sensitivity
Can over-predict separation
Modifications to original 1990 Wilcox model

Wilcox 1998 Modifications for lowRe, vortex stretching, free shear, compressibility effects
Wilcox 2008 Better predictions of separated flows, reduced dependence on freestream/inflow
conditions

SST K-Omega
Summary
Two equation model
Formulated by blending k-Omega near the wall with a transformed k-Epsilon model in the
bulk flow to avoid sensitivity to freestream conditions
Widely used in aerospace, Formula 1, and turbomachinery industries
Advantages
Performs well for swirling flows (for a two equation model)
Does not require damping functions (or a two-layer treatment) for resolving sublayer
Performs well in adverse pressure gradients
Disadvantages
Over-predicts separation
Sensitive to mesh refinement.

SST K-Omega: Non-Linear Constitutive Relations
Quadratic & Cubic Constitutive Relations for SST k-w

New non-linear constitutive relations
In addition to existing linear relations
Improved accuracy
Particularly for anisotropic flows
Correctly predicts secondary flows in square ducts
Unlike Boussinesq models
Base SST
Reduced number of iterations to convergence
Typical Use Cases

Flows with strong swirl, streamline curvature,and boundary layer flows
Unrestricted Siemens AG 2016 RSM + Quad. Press.-Strain SST+QCR

Standard K-Epsilon
Summary
Two equation model
The original general purpose complete model for industry
Advantages
Robust industry standard model
Insensitive to inflow conditions
Disadvantages
Poor accuracy for many classes of problem, including those with swirling and
separated flows
Near wall treatment necessary (low-Re damping or two-layer)
Many variants and modifications available
Results influenced by the wall function implementation

Realizable K-Epsilon
Summary
Two equation model
Developed by Shih at NASA Lewis
Ensures normal stresses ( , , ) are positive
Implemented by varying Cm spatially (constant in standard k-epsilon)
Advantages
More physical and accurate than standard k-epsilon
Performs better than std. k- for separated flows, swirling and
rotating flows, and flows with large streamline curvature
Disadvantages
Near wall treatment necessary
Results can be heavily influenced by the wall function
implementation

Low-Re Damped K-Epsilon Models
Summary
Two equation model
Uses damping function to replicate low Reynolds number near wall behavior
(Two implementations: Abe-Kondoh-Nagano and Lien-Leschziner)
Advantages
Low sensitivity to near-wall mesh resolution
Better prediction of turbulent heat transfer than k-omega models
Disadvantages
Overprediction of separated flows
Many variants of low-Re damped K-epsilon models in the literature
(universality?)

K-Epsilon: V2F Model
Summary
Four equation model (k, , v2, f) extension of k-epsilon
Handles non-local (elliptic) effect of walls via specific equation (f), and near
wall correct near-wall asymptotic limit (v2, or wall-normal fluctuations)
Advantages
More accurate prediction of near wall turbulence, heat transfer, skin friction,
and separation
Disadvantages:
Robustness (elliptic equation sensitive to near-wall modelling),
Expensive (4 equations)

K-Epsilon: Elliptic Blending Model
Summary
Four equation model (k, , , ) extension of k-epsilon
Similar to V2F in nature
Handles non-local (elliptic) effect of walls via an elliptic blending approach
rather than the elliptic relaxation approach of V2F
Advantages
More robust than the V2F model
Accurate prediction of near wall turbulence, heat transfer, skin friction, and
separation
Disadvantages:
Expensive (4 equations)

Std. K-Epsilon with Non-linear Constitutive Relations
Summary
The Boussinesq approximation used by the k-epsilon model assumes a linear
relationship between the Reynolds-stress and the strain rate tensors
The standard k-epsilon model with non-linear constitutive relations instead
uses a higher order expansion to give a quadratic or cubic relationship
Advantages
Anisotropy effects accounted for via algebraic formulation
No further transport equations
At low strain rates recovers standard k-epsilon
Shows potential where secondary flows are important
Disadvantages
Robustness (requires finer mesh to predict secondary re-circulation),
Not sensitive to frame-rotation.

Reynolds Stress Transport
Summary
Seven equation model (six Reynolds stresses + epsilon)
The most complete and complex RANS model
Advantages
Able to capture anisotropy (swirling flows, secondary flows, etc.)
For applications such as cyclones, RSM is the only model that produces good results
Disadvantages
Computationally expensive
Near wall treatment necessary
Less robust than 2 eqn. models on poor quality meshes
Variants and Options:

Linear (Gibson/Launder)
Linear two-layer (Launder/Sharma)
Quadratic (Sarkar, Speziale & Gatski)
Low Reynolds Number EB-RSM
Option to use the Generalized Gradient Diffusion Hypothesis (Daly-Harlow)
Reynolds stress diffusion is treated anisotropically for improved accuracy

Transition Modeling
Summary
Three models available
Turbulence Suppression Model Specified laminar region
Gamma ReTheta transition model Predictive four equation model
extension of SST-kw, additional transport equations for intermittency
and transition momentum thickness
Gamma transition model Predictive three equation model extension
of SST-kw achieves similar levels of accuracy to Gamma ReTheta at
reduced computational cost
Advantages
Gamma and Gamma ReTheta models have predictive (correlation
based) capability for transition
Disadvantages
Computationally expensive
Can be less stable

Transition Modeling Gamma Transition Model
Reduce computational time

Gamma Re Theta Gamma Model
Compared to Gamma Re Theta model
Typically ~1.2-1.25x speed-up
One equation transition model

Option under SST-kw
Achieves similar levels of accuracy to Gamma Re

Theta model
Testcase # cores Speed-up of Gamma
Typical Use Cases over GammaReTheta
F1 Wings Airfoil Case 11 from ERCOFTAC database 1 1.25x
Gas Turbines Airfoil Case 11 from ERCOFTAC database 5 1.24x
Aircraft Wings
Ahmed body slant angle 25 degrees 16 1.19x
Sickle Wing 16 1.21x
Prolate Spheroid 16 1.07x

Transition Modeling Cross Flow Modification for Re
Improved accuracy
Particularly for high speed compressible cases
Modification to Gamma Re Theta Transition Model

Takes account of cross flow effects
Secondary flows normal to the streamlines
More prevalent in swept wings
Chordwise pressure gradient has component
normal to streamlines
Significantly modifies onset of transition
Typical Use Cases

Transonic swept wings Skin Friction for Prolate Spheroid in STAR-CCM+ Compared to
Other Codes

Transition Modeling Cross Flow Modification for Re
For swept transonic wings

Crossflow effects are significant
Re +
Sickle wing test case Crossflow
Has varying sweep angles from 30 degrees a the Modification
root to 55 degrees at the tip
Petzold and Radespiel, AIAA Paper 2013-2466
Results with crossflow modification match
experiment well Re
Without, transition region confined to trailing edge
Note that STAR-CCM+ geometry (including tunnel) is

reconstructed from available data in the paper and is not
exact so comparison is qualitative only
Sickle wing testcase Comparison of skin friction with
and without crossflow mod. and experiment (red line)

Near Wall Modeling
For most industrial CFD problems the mesh resolution is insufficiently fine near the wall to
resolve the viscous-affected (pseudo-laminar) sublayer.
A wall function is used to determine the relationship between the first cell center and the
wall (y+ of the first cell in the 100s)
For cases requiring accurate predictions of heat transfer and separation it is necessary to
resolve the viscous sublayer with a fine prism layer mesh (y+ ~1)
Not all turbulence models can resolve down to the wall and require special near wall treatment (some
k-epsilon models, RSM)
U+ = U+

The first cell center should not be placed in the
buffer layer (5 < y+ < 30)
Fully Turbulent
Log-Law Region
Viscous Buffer Defect Layer
Sublayer Layer

+ =

Unrestricted Siemens AG 2016 5 30 500-1000 log y+
Wall Treatment
Three basic approaches:

1) high y+ uses std. wall function approaches
2) low y+ uses low Re approach (no wall function)
3) all y+ uses hybrid approach
valid for models that can be solved through sublayer
blends turbulence source terms between wall function and low Re
uses continuous wall laws (Reichardt, Kader)

Two-Layer Models
Realizable two-layer with hybrid wall treatment has been the default since STAR-CCM+'s inception
Key to robustness is the blending of linear equation coefficients

Discretized equation + =

Algebraic equation = (1 )
Key to accuracy is using the minimum possible Re for the crossover location

Curvature Correction
Curvature correction is typically used to overcome the premature dissipation of vorticity

Vortices shed from the tip of wings or other lifting surfaces will tend to dissipate early with two-equation
models
Augments the turbulent energy production term according to local rotation and vorticity rates to offset this
early dissipation
It supplies the effects of strong (streamline) curvature and frame-rotation
Available for the following turbulence models

SST K-omega
Realizable k-epsilon

Curvature Correction Validation
Rotating channel flow, Ro = 0.5

DNS of Kristoffersen & Andersson (1993) for Re = 194 and 0 < Ro <0.5
Velocity Turbulent kinetic energy

Curvature Correction Validation
NACA 0012 at incidence

Exp. Chow et al. (1993), a = 10, U = 46m/s
CC factor
Crossflow velocity Streamwise velocity


Largest energy containing eddies are resolved
The mesh is used as a filter
Turbulent length scales smaller than the grid are modeled by a sub-grid scale model
The Taylor Microscale can be used as an indication of an appropriate mesh size
Kolmogorov
Taylor Microscale, Lengthscale,
k 3
Energy Containing
Eddies Inertial Sub-
log E(k)
range
Dissipating
Resolved Eddies
Modeled
log k (wavenumber)
LES Sub-Grid Scale (SGS) Models
Choice of Sub-Grid Scale model generally less important than choice of primary turbulence model
Provided the mesh is suitably resolved (and the SGS models are more basic)
Three SGS models available in STAR-CCM+:-
Smagorinsky
Uses mixing length hypothesis to model the sub-grid scale stresses
Smagorinsky constant, Cs fixed
However experimentally found to vary from one flow type to another and spatially
Dynamic Smagorinsky
Extended version of Smagorinsky
Determines Cs locally using bootstrap method and test filter
Wall Adapting Local Eddy (WALE)

Algebraic sub-grid scale model
Computationally cheapest model
WALE constant, Cw, is fixed
However experimentally found to vary

Application Example LES Jet Noise
Simulation of jet/nozzle configuration at Mach

0.75
Near field simulated with LES

Resolves acoustic pressure
Propagation to far-field using permeable FW -H

integration.
Unrestricted Siemens AG 2016 Instantaneous Pressure and Vorticity Field

LES/DES Boundary Conditions & Initialization
Synthetic Eddy Method (SEM)

For LES models rather than specify turbulence quantities for inflow boundaries and initialization it is
necessary to provide appropriate fluctuations from the mean flow
If this is done as random noise, there are no coherent turbulent structures and the turbulence will quickly
dissipate
SEM generates synthetic eddies for inflow boundaries and initialization
Option to scale mass flow to a constant value at SEM boundaries (otherwise synthetic turbulence causes
mass flow to fluctuate)

Detached Eddy Simulation (DES)
Hybrid method between RANS and LES

If the grid is fine enough turbulence model becomes a sub-grid scale model to a locally applied LES model
If the grid is not fine enough, the turbulence model is the underlying RANS model (e.g. Spalart Allmaras)
Variants for RANS portion

Spalart Allmaras (original)
SST k-omega
Elliptic Blending k-epsilon
These variants are available in the following forms

Delayed Detached Eddy Simulation (DDES)
Ensures that the whole boundary layer is RANS even if the grid is fine in the boundary layer
Improved Delayed Detached Eddy Simulation (IDDES)
Only inner portion of boundary layer is RANS, determined based on wall distance

Elliptic Blending DDES
Benefits
Better prediction of the near-wall behavior compared to
other DES models
Feature
A DES turbulence model based on Elliptic-Blending k-
Compatible with all y+ Wall and low-Re y+ wall treatment
Available as DDES variant
Model has the property of transitioning to turbulence
almost naturally and predicting better separation

Elliptic Blending DDES Test Case
Ahmed body test case

Rear slant angle 25
Critical angle for separation is 30
Inflow velocity 40 m/s
Complex flow sensitive to underlying RANS model
Ideal test case to distinguish between DES models
Two DES models tested

SST kw IDDES

SST kw IDDES predicts flow separated from the slanted surface

This over-prediction of separation results from the underlying SST k IDDES
RANS model (SST-kw)
EB-DDES predicts attached flow

In near wall region, Elliptic Blending RANS model active
Leads to better prediction of near wall turbulence and any
separation

62 mm ahead 162 mm
of trailing ahead of
edge trailing edge
238 mm
downstream
of trailing
edge
438 mm
downstream
of trailing
edge
Unrestricted Siemens AG 2016 Comparison of new Elliptic Blending based DES with SST kw based DES on
Page 41 26.10.2016 the Ahmed Body : improved separation prediction Siemens PLM Software
Turbulent Heat Flux
Classical heat flux based on Bousinessq approximation

Heat flux assumed to be proportional to the turbulent viscosity
Not valid for buoyant flows or in close proximity to the wall
Temperature Heat Flux or Algebraic Heat Flux Model (AHFM)

Algebraic formulation for turbulent heat flux
Additional transport equation solved
Requires low Reynolds number turbulence model
Low Reynolds number standard k- model

Turbulence Models Available in STAR-CCM+
High-Reynolds number (High Y+)

Wall treatments Low-Reynolds number (Low Y+)
All Y+ Hybrid treatment
Standard formulation (low-Reynolds number)
Spalart-Allmaras
High Reynolds number
Standard K-
Standard two-layer K-
Standard K- low-Reynolds number
Realizable K-
Realizable two-layer K-
K- Abe-Kondoh-Nagano low-Reynolds number
V2F low-Reynolds number
Elliptic Blending K -
RANS Models Multiphase standard K-
Multiphase standard two-layer K-
Multiphase realizable K-
Standard K- (Wilcox 2008)
K-
SST K-
Linear pressure strain
Quadratic pressure strain
Reynolds Stress Transport
Linear pressure strain two-layer
Low Reynolds EB-RSM
User specified turbulent suppression
Transistion Modeling Laminar-Turbulent Transition Gamma transition model
Gamma ReTheta transition model
Smagorinsky subgrid scale model
Large Eddy Simulation (LES) Dynamic Smagorinsky subgrid scale model
Wale subgrid scale model
Delayed Detached Eddy SST K- Delayed Detached Eddy Simulation
LES & DES Models
Simulation (DDES) Spalart-Allmaras Delayed Detached Eddy Simulation
Elliptic Blending k- Delayed Detached Eddy Simulation
Improved Delayed Detached SST K- w Improved Delayed Detached Eddy Simulation
Unrestricted Siemens AG 2016 Eddy Simulation (IDDES) Spalart-Allmaras Improved Delayed Detached Eddy Simulation
Page 43 26.10.2016 Resolved Turbulence Inflow Conditions Synthetic Eddy Method (SEM) Siemens PLM Software
Choosing a Turbulence Model (Summary)
Turbulence Model Usage

Spalart-Allmaras Widely used in aerospace for external aerodynamics, economical, and produces good results for attached flow s. Under-predicts
separation.
Standard k-w Performs w ell for sw irling flow s, does not require damping functions or tw o layer treatment to resolve sublayer. Over-predicts
separation, sensitive to freestream/inflow turbulence.
SST k-w Has the near w all performance and benefits of standard k-w w ithout the sensitivity to freestream/inflow conditions, w ith added
mesh sensitivity .
Standard k-e The industry standard w orkhorse, stable, but diffusive, not satisfactory for high fidelity results. Requires near w all damping or
tw o layer treatment.
Realizable k-e More accurate and physical than standard k-e. Performs better than std. k- for separated flow s, swirling and rotating flow s, and
flow s w ith large streamline curvature. Requires near w all tw o layer treatment.
Low-Re Damped k-e model Arguably low -Re damped k-epsilon models have a place for simulating cases w here the low Reynolds number region extends
over much of the domain.
V2F Model Accounts for the effects of near w all anisotropy. More accurate prediction of near w all heat transfer, skin fiction, and separation.
Elliptic Blending Model Accounts for the effects of near w all anisotropy w ith similar benefits to V2F. More robust than V2F.
Standard k-e with Non-Linear Constitutive Relations Allow s the effects of anisotropy to be modeled w ithout requiring further transport equations. Show s potential for capturing
secondary flow s. Requires near w all treatment.
Reynolds Stress Transport The only RANS model that fully models anisotropy. Essential for some sw irling flow s such as in cyclones. Expensive and less
stable
Transition Model (Gamma-ReTheta) Able to predict laminar-turbulent transition. Expensive as tw o additional transport equations added.
Large Eddy Simulation (LES) Able to predict the largest turbulent eddy structures on a sufficiently fine mesh. Sub-grid models are inferior to RANS models,
and near w all mesh size w ill be insufficient to capture much of the turbulent energy spectrum. Expensive due to mesh resolution
required
Detached Eddy Simulation (DDES, IDDES) A blend of LES in the bulk flow and RANS (SA, SST-kw , or EB-ke) near the w all intended to be the best of both w orlds.
Expensive due to mesh resolution required

Summary
CD-adapco is committed to providing the best in class turbulence models in STAR-CCM+

A good choice of turbulence model is key to accurate CFD results
STAR-CCM+ has a wide range of turbulence models suitable to all classes of flow and computational
budgets, from laminar, through transitional flow, to modeled and resolved turbulence

Spotlight
On
Turbulence
STAR-CCM+ v11.06

Spotlight On... Turbulence

Uploaded by

Copyright:

Available Formats

Spotlight On... Turbulence

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Spotlight On... Turbulence

Uploaded by

Copyright:

Available Formats

Spotlight

In this Spotlight we examine:

Unrestricted Siemens AG 2016

Turbulence occurs when inertial forces greater than viscous forces

Unrestricted Siemens AG 2016

Turbulence affects mean flow by transporting and dissipating energy

Unrestricted Siemens AG 2016

Turbulence is instrumental in the momentum mixing and heat transfer between

Unrestricted Siemens AG 2016

Time- and space-dependent flow variables such as velocity can be decomposed,

Three averaging/filtering methodologies

Unrestricted Siemens AG 2016 log k (wavenumber)

DNS is fully resolved CFD

Unrestricted Siemens AG 2016

DNS only possible for simple cases

As such it provides a useful tool to calibrate turbulence models

Good choice of turbulence model critical to achieving accurate results at

Unrestricted Siemens AG 2016

Reynolds Averaged (Turbulence Modeled)

DES (Turbulence Partially Resolved)

LES (Turbulence Partially Resolved)

Eddy Viscosity Assumption

Boussinesq Hypothesis used in RANS models

Model turbulent viscosity () to close the NS equations

To model solve for one or more transported quantities, for example:-

Inherent assumption of isotropy

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Modifications to original 1990 Wilcox model

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Quadratic & Cubic Constitutive Relations for SST k-w

Reduced number of iterations to convergence

Typical Use Cases

Unrestricted Siemens AG 2016 RSM + Quad. Press.-Strain SST+QCR

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Variants and Options:

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Reduce computational time

One equation transition model

Achieves similar levels of accuracy to Gamma Re

Unrestricted Siemens AG 2016

Modification to Gamma Re Theta Transition Model

Typical Use Cases

Unrestricted Siemens AG 2016

For swept transonic wings

Note that STAR-CCM+ geometry (including tunnel) is

Unrestricted Siemens AG 2016

Three basic approaches:

Unrestricted Siemens AG 2016

Unrestricted Siemens AG 2016

Curvature correction is typically used to overcome the premature dissipation of vorticity

It supplies the effects of strong (streamline) curvature and frame-rotation