CFD - Turbulence Modeling

CFD Lecture Note 4

Nov. 2022

N. Hur
Dept. of Mechanical Engineering, Sogang Univ.
Multi-Phenomena CFD Engineering Research Center(ERC)

[1] Versteeg, H.K and Malalasekera, An introduction to computational fluid dynamics the finite volume method
(1991)
[2] Ansys Manual (2009)
[3] Andre Bakker, Applied Computational Fluid Dynamics, Lecture 10-Turbulence Models, Fluent Inc. (2002)
[4] K.D. Squires, Large Eddy Simulation for Particle-Laden Turbulent Flows

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 What is turbulence?
• Unsteady, irregular (non-periodic) motion in which transported
quantities (mass, momentum, scalar species) fluctuate in time and
space
– Identifiable swirling patterns characterize turbulent eddies
– Enhanced mixing (matter, momentum, energy, etc.) results

• Fluid properties and velocity exhibit random variations

– Statistical averaging results in accountable, turbulence related transport
mechanisms
– This characteristic allows for turbulence modeling

• Contains a wide range of turbulent eddy sizes (scales spectrum)

– The size/velocity of large eddies is on the order of the mean flow
• Large eddies derive energy from the mean flow
– Energy is transferred from larger eddies to smaller eddies
• In the smallest eddies, turbulent energy is converted to internal energy by viscous
dissipation
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 What is turbulence?

Great red spot

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Is the Flow Turbulent?
Flows can be characterized by the Reynolds Number, Re

External Flows 𝜌𝑈𝐿

where Re𝐿 =
𝜇
Re𝑥 ≥ 500,000 along a surface
𝐿 = 𝑥, 𝑑, 𝑑ℎ , etc.
Re𝑑 ≥ 20,000 around an obstacle Other factors such as free-
stream turbulence, surface
Internal Flows conditions, and disturbances
may cause transition to
Redh ≥ 2,300
turbulence at lower Reynolds
numbers
Natural Convection
Ra 𝛽𝑔𝐿3 Δ𝑇 𝜌2 𝐶𝑝 𝛽𝑔𝐿3 Δ𝑇
Pr where Ra = = is the Rayleigh number
≥ 109 𝜈𝛼 𝜇𝑘
𝜈 𝜇𝐶𝑝
Pr = =
𝛼 𝑘 is the Prandtl number
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Observation by O. Reynolds

Laminar
(Low Reynolds Number)

Transition
(Increasing Reynolds Number)

Turbulent
(Higher Reynolds Number)

Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulent Flow Structures

Small Large
structures structures

Energy Cascade
Richardson (1922)
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulent Boundary Layer

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulent Boundary Layer - Experiment

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulent Boundary Layer - Simulation

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Governing Equations
Conservation Equations
𝜕𝜌 𝜕
Continuity + (𝜌𝑢𝑖 ) = 0
𝜕𝑡 𝜕𝑥𝑖

𝜕 𝜕 𝜕𝑃 𝜕𝜏𝑖𝑗
Momentum (𝜌𝑢𝑖 ) + (𝜌𝑢𝑖 𝑢𝑗 ) = − +
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑖 𝜕𝑥𝑗

𝜕 𝜕 𝜕𝑃 𝜕 𝜕𝑇
Energy 𝜕𝑡
(𝜌ℎ𝑡𝑜𝑡 ) +
𝜕𝑥𝑗
(𝜌ℎ𝑡𝑜𝑡 𝑢𝑗 ) = +
𝜕𝑡 𝜕𝑥𝑗
(𝑢𝑖 𝜏𝑖𝑗 + 𝜆
𝜕𝑥𝑗
)

𝜕𝑢𝑖 𝜕𝑢𝑗 2 𝜕𝑢𝑖 1

𝜏𝑖𝑗 = 𝜇 + + 𝛿𝑖𝑗 ℎ𝑡𝑜𝑡 = ℎ + 𝑢𝑖2
where 𝜕𝑥𝑗 𝜕𝑥𝑖 3 𝜕𝑥𝑗 2

Note that there is no turbulence equation in the governing

Navier-Stokes equations!
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Prediction Methods

l h = l/ReL3/4

Direct numerical simulation (DNS)

Large eddy simulation (LES)

Reynolds averaged Navier-Stokes equations (RANS)

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Overview of Computational Approaches
Direct Numerical Simulation (DNS)
Theoretically, all turbulent (and laminar / transition) flows can be simulated by numerically
solving the full Navier-Stokes equations
Resolves the whole spectrum of scales. No modeling is required
But the cost is too prohibitive! Not practical for industrial flows

Large Eddy Simulation (LES) type models

Solves the spatially averaged N-S equations
Large eddies are directly resolved, but eddies smaller than the mesh are modeled
Less expensive than DNS, but the amount of computational resources and efforts are still
too large for most practical applications

Reynolds-Averaged Navier-Stokes (RANS) models

Solve time-averaged Navier-Stokes equations
All turbulent length scales are modeled in RANS
Various different models are available
This is the most widely used approach for calculating industrial flows

There is not yet a single, practical turbulence model that can reliably predict all
turbulent flows with sufficient accuracy
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Models

• A turbulence model is a computational procedure to close the

system of mean flow equations.
• For most engineering applications it is unnecessary to resolve
the details of the turbulent fluctuations.
• Turbulence models allow the calculation of the mean flow
without first calculating the full time-dependent flow field.
• We only need to know how turbulence affected the mean flow.
• In particular we need expressions for the Reynolds stresses.
• For a turbulence model to be useful it:
• must have wide applicability,
• be accurate,
• simple,
• and economical to run.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling

Typical point velocity in turbulent flow

𝑢(𝑡) = 𝑈 + 𝑢′(𝑡)

𝑈 : steady mean value

𝑢′(𝑡): fluctuating component

rms of the fluctuations

Mean Φ
of the flow property 𝜙 𝑡 = Φ + 𝜙′ 𝑡

Φ=
1 Δ𝑡
න 𝜙 𝑡 𝑑𝑡
Kinetic energy
Δ𝑡 0

Time average of the fluctuations 𝜙′ 𝑡 is zero

1 Δ𝑡
Turbulence intensity
𝜙′ = න 𝜙′ 𝑡 𝑑𝑡 ≡ 0 2𝑘/3 1/2
Δ𝑡 0 𝑇𝑖 =
𝑈𝑟𝑒𝑓

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Rules for Time Average of Fluctuating Properties

fluctuating properties 𝜙 = Φ + 𝜙′ 𝜓 = Ψ + 𝜓′

fluctuating vector quantity 𝐚 = 𝐀 + 𝐚′

𝜕𝜙 𝜕Φ
𝜙′ = 𝜓′ = 0 Φ=Φ = න𝜑𝑑𝑠 = නΦ𝑑𝑠
𝜕𝑠 𝜕𝑠

𝜙+𝜓 =Φ+Ψ 𝜙𝜓 = ΦΨ + 𝜙′𝜓′ 𝜙Ψ = ΦΨ 𝜙′Ψ = 0

𝑑𝑖𝑣​​𝐚 = 𝑑𝑖𝑣𝐀 𝑑𝑖𝑣 𝜑𝐚 = 𝑑𝑖𝑣 𝜑𝐚 = 𝑑𝑖𝑣 Φ𝐀 + 𝑑𝑖𝑣 𝜑′𝐚′

𝑑𝑖𝑣𝑔𝑟𝑎𝑑𝜑 = 𝑑𝑖𝑣𝑔𝑟𝑎𝑑Φ

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Turbulent Flow Equations for Compressible Flows
𝐮 = 𝐔 + 𝐮′ 𝑢 = 𝑈 + 𝑢′ 𝑣 = 𝑉 + 𝑣′ 𝑤 = 𝑊 + 𝑤′ 𝑝 = 𝑃 + 𝑝′
𝜕𝜌
+ 𝑑𝑖𝑣(𝜌𝐔) = 0
Continuity equation 𝜕𝑡

Reynolds equations
𝜕(𝜌𝑈) 𝜕𝑃 𝜕 𝜌𝑢′2 𝜕 𝜌𝑢′𝑣′ 𝜕 𝜌𝑢′𝑤′
+ 𝑑𝑖𝑣(𝜌𝑈𝐔) = − + 𝑑𝑖𝑣(𝜇𝑔𝑟𝑎𝑑(𝑈)) + − − − + 𝑆𝑀𝑥
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧

𝜕(𝜌𝑉) 𝜕𝑃 𝜕 𝜌𝑢′𝑣′ 𝜕 𝜌𝑣′2 𝜕 𝜌𝑣′𝑤′

+ 𝑑𝑖𝑣(𝜌𝑉𝐔) = − + 𝑑𝑖𝑣(𝜇𝑔𝑟𝑎𝑑(𝑉)) + − − − + 𝑆𝑀𝑦
𝜕𝑡 𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧

𝜕(𝜌𝑊) 𝜕𝑃 𝜕 𝜌𝑢′𝑤′ 𝜕 𝜌𝑣′𝑤′ 𝜕 𝜌𝑤′2

+ 𝑑𝑖𝑣(𝜌𝑊𝐔) = − + 𝑑𝑖𝑣(𝜇𝑔𝑟𝑎𝑑(𝑊)) + − − − + 𝑆𝑀𝑧
𝜕𝑡 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧

Scalar transport equation

𝜕(𝜌Φ) 𝜕 𝜌𝑢′𝜑′ 𝜕 𝜌𝑣′𝜑′ 𝜕 𝜌𝑤′𝜑′
+ 𝑑𝑖𝑣(𝜌Φ𝐔) = 𝑑𝑖𝑣(ΓΦ 𝑔𝑟𝑎𝑑(Φ)) + − − − + 𝑆Φ
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling

Reynolds stresses
- arising from correlations between velocity components.
- 3 normal stresses, 3 shear stresses.

−𝜌𝑢′2 −𝜌𝑢′𝑣′ −𝜌𝑢′𝑤′

−𝜌𝑣′𝑢′ −𝜌𝑣′2 −𝜌𝑣′𝑤′
−𝜌𝑤′𝑢′ −𝜌𝑤′𝑣′ −𝜌𝑤′2

Closure problem – the need for turbulence modeling

- Reynolds averaging results in 6 additional unknowns, the Reynolds stresses
- To solve the equations, we need to link this unknowns to mean flow
properties, which is called turbulence modeling.

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows

Free turbulent flows

• Mixing layer
• jet
• wake

Boundary layer near solid walls

• Flat plate boundary layer

• Pipe flow

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Free turbulent flows

• A mixing layer forms at the interface of two regions

: one with fast and the other with slow moving fluid
• In a jet a region of high speed flow is completely surrounded by stationary
fluid
• A wake is formed behind an object in a flow,
so here a slow-moving region is surrounded by fast-moving region

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Free turbulent flows

𝑈 𝑦 𝑈 − 𝑈min 𝑦 𝑈max − 𝑈 𝑦
= 𝑔( ) = 𝑓( ) = ℎ( )
𝑈max 𝑏 𝑈max − 𝑈min 𝑏 𝑈max − 𝑈min 𝑏

𝑈max 𝑈
and min represent the maximum and minimum mean velocity at a distance x
downstream of the source.
Hence, if these local mean velocity scales are chosen and x in the flow direction.
Such flows are called self-preserving.

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows

Visualization of a jet flow

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Free turbulent flows
The turbulence structure also reaches a self-preserving state albeit after a greater
distance from the flow source than the mean velocity.
Then ′2 ′2 ′2 ′ ′
𝑢 𝑦 𝑣 𝑦 𝑤 𝑦 𝑢𝑣 𝑦
= 𝑓1 ( ) = 𝑓2 ( ) = 𝑓3 ( ) = 𝑓4 )
(
𝑈 2 𝑟𝑒𝑓 𝑏 𝑈 2 𝑟𝑒𝑓 𝑏 𝑈 2 𝑟𝑒𝑓 𝑏 𝑈 2 𝑟𝑒𝑓 𝑏

The velocity scale 𝑈𝑟𝑒𝑓 is, as above, ( 𝑈max − 𝑈min

) for a mixing layer and wake and
𝑈max
for jets.

The precise form of function f,g,h and 𝑓𝑖 varies from flow to flow.

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Free turbulent flows

Mean velocity distributions and turbulence properties for (a) two-dimensional mixing
layer, (b) planar turbulent jet, (c) wake behind a solid strip

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Flat plate boundary layer and pipe flow
If we form a Reynolds number based on a distance y away from the wall Re𝑦 = 𝑈𝑦/𝜐

The mean flow velocity only depends on the distance y from the wall,
fluid density 𝜌 and viscosity 𝜇 and the wall shear stress 𝜏𝑤 .

So 𝑈 = 𝑓(𝑦, 𝜌, 𝜇, 𝜏𝑤 ) 𝑈 𝜌𝑢𝑡 𝑦
Dimensional analysis shows that 𝑢+ = = 𝑓( ) = 𝑓(𝑦 + )
𝑢𝑡 𝜇

This is called the law of the wall and contains the definitions of two important
+
dimensionless group 𝑢+ and 𝑦

Note that the appropriate velocity scale is 𝑢𝑡 = (𝜏𝑤 /𝜌)1/2 , the called friction velocity

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Flat plate boundary layer and pipe flow
The length scale appropriate to this region is the boundary layer thickness 𝛿

In this region we have 𝑈 = 𝑔(𝑦, 𝛿, 𝜌, 𝜏𝑤 )

𝑈 𝑦
Dimensional analysis yields 𝑢+ = = 𝑓 = 𝑔( )
𝑢𝑡 𝛿

The most useful form emerges if we view the wall shear stress as the cause of a
velocity deficit 𝑈max − 𝑈min which decrease the closer we get to the edge of the
boundary layer or the pipe centerline.
𝑈max − 𝑈 𝑦
Thus, = 𝑔( )
𝑢𝑡 𝛿

This formula is called the velocity-defect law

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Flat plate boundary layer and pipe flow
• Linear sub-layer – the fluid layer in contact with a smooth wall
𝜕𝑈
At the solid surface the fluid is stationary. 𝜏(𝑦) = 𝜇 ≃ 𝜏𝑤
𝜕𝑦

After integration with respect to y and application of boundary condition U=0 if y=0
We obtain a linear relationship between the mean velocity and distance from wall

𝜏𝑤 𝑦
𝑈= 𝑢+ = 𝑦 +
𝜇

Because of the linear relationship between the mean velocity and distance from wall
The fluid layer adjacent to the wall is often known as the linear sub-layer

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Flat plate boundary layer and pipe flow
• Log-law layer – the turbulent region close to a smooth wall
+
Outside the viscous sublayer (30 < 𝑦 <500) a region exists where viscous and
turbulent effects are both important

One further assumption regarding the length scale of turbulence

(mixing length 𝑙𝑚 = 𝜅𝑦 ) allows us to derive a dimensionally correct form of the
functional relationship between 𝑢+ and 𝑦 + .

1 1
𝑢+ = ln𝑦 + + 𝐵 = ln( 𝐸𝑦 + )
𝜅 𝜅

This is often called the log-law

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows
Flat plate boundary layer and pipe flow
• Outer layer – the inertia-dominated region far from the wall

In the overlap region the log-law and velocity-defect law have to become equal.

𝑈max − 𝑈 1 𝑦
= ln( ) + 𝐴 where A is a constant.
𝑢𝑡 𝜅 𝛿

This velocity-defect law is often

called the law of the wake.

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Simple Turbulent Flows

Mean velocity and turbulence property distribution data for a flat boundary layer with
a constant imposed pressure (Klebnoff, 1955)

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Boussinesq hypothesis
Many turbulence models are based upon the Boussinesq
hypothesis.
• It was experimentally observed that turbulence decays unless there is
shear in isothermal incompressible flows.
– Turbulence was found to increase as the mean rate of deformation
increases.
– Boussinesq proposed in 1877 that the Reynolds stresses could be linked
to the mean rate of deformation.
Using the suffix notation where i, j, and k denote the x-, y-, and z-
directions respectively, viscous stresses are given by:
𝜕𝑢𝑖 𝜕𝑢𝑗
𝜏𝑖𝑗 = 𝜇 𝑒𝑖𝑗 = 𝜇 +
𝜕𝑥𝑗 𝜕𝑥𝑖

Similarly, link Reynolds stresses to the mean rate of deformation:

𝜕𝑈𝑖 𝜕𝑈𝑗
𝜏𝑖𝑗 = −𝜌𝑢𝑖 ′𝑢𝑗 ′ = 𝜇𝑡 +
𝜕𝑥𝑗 𝜕𝑥𝑖

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulent viscosity

• A new quantity appears: the turbulent viscosity t.

𝜕𝑈𝑖 𝜕𝑈𝑗
𝜏𝑖𝑗 = −𝜌𝑢𝑖 ′𝑢𝑗 ′ = 𝜇𝑡 +
𝜕𝑥𝑗 𝜕𝑥𝑖

• Its unit is the same as that of the molecular viscosity: Pa.s.

• It is also called the eddy viscosity.
• We can also define a kinematic turbulent viscosity: t = t/. Its
unit is m2/s.
• The turbulent viscosity is not homogeneous, i.e. it varies in
space.
• It is, however, assumed to be isotropic. It is the same in all
directions. This assumption is valid for many flows, but not for all
(e.g. flows with strong separation or swirl).
Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Turbulence Models
Classical models : based on (time-averaged) Reynolds equations
1. zero equation model (e.g. mixing length model, Baldwin-Lomax model)
2. one equation model (e.g. Sparlart-Allmaras model etc.)
3. two equation model (e.g. 𝑘 − 𝜀 Model, Model etc.)
4. three equation model (e.g. 𝑘 − 𝜀 − 𝛼 Model etc.)
5. Reynolds stress equation model
6. algebraic stress model
Large eddy simulation (LES) : based on space-filtered equations
Detached eddy simulation (DES) : Hybrid Models
Direct Numerical simulation (DNS)
𝜕𝑢𝑖 𝜕𝑢𝑗
𝜏𝑖𝑗 = 𝜇𝑒𝑖𝑗 = 𝜇 +
Viscous stresses 𝜕𝑥𝑗 𝜕𝑥𝑖

𝜕𝑈𝑖 𝜕𝑈𝑗
Reynolds stress 𝜏𝑖𝑗 = −𝜌𝑢′𝑖 𝑢′𝑗 = 𝜇𝑡 +
𝜕𝑥𝑗 𝜕𝑥𝑖
(Boussinesq, 1877)
Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Mixing Length Model (Prandtl, 1925)
• On dimensional grounds one can express the kinematic
turbulent viscosity as the product of a velocity scale and a length
scale:
𝜈𝑡 (𝑚2 /𝑠) ∝ 𝜗(𝑚/𝑠)ℓ(𝑚)

• If we then assume that the velocity scale is proportional to the

length scale and the gradients in the velocity (shear rate, which
has dimension 1/s):
𝜕𝑈
𝜗 ∝ ℓ
𝜕𝑦

• we can derive Prandtl’s (1925) mixing length model:

𝜕𝑈
𝜈𝑡 = ℓ2𝑚
𝜕𝑦

• Algebraic expressions exist for the mixing length for simple 2-D
flows, such as pipe and channel flow.
Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Mixing Length Model (Prandtl, 1925)
2
𝜕𝑈
𝑣𝑡 = 𝑙𝑚
Prandtl’s mixing length model 𝜕𝑦

2
𝜕𝑈 𝜕𝑈
Turbulent Reynolds stress 𝜏𝑥𝑦 = 𝜏𝑦𝑥 = −𝜌𝑢′𝑣′ = 𝜌𝑙𝑚
𝜕𝑦 𝜕𝑦

𝜕Φ
Mixing length model −𝜌𝑣′𝜙′ = Γ𝑡 Γ𝑡 = 𝜇𝑡 /𝜎𝑡
𝜕𝑦

Prandtl’s mixing length model Rodi (1980)

𝜕𝑈 𝜎𝑡 In near wall flows
2
𝑣𝑡 = 𝑙𝑚
𝜕𝑦 0.9 for jets and mixing layers
= ቌ0.5
0.7 in axisymmetric jets

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Mixing Length Model (Prandtl, 1925)
• Advantages:
• Easy to implement.
• Fast calculation times.
• Good predictions for simple flows where experimental correlations for the
mixing length exist.
• Disadvantages:
• Completely incapable of describing flows where the turbulent length scale
varies: anything with separation or circulation.
• Only calculates mean flow properties and turbulent shear stress.
• Use:
• Sometimes used for simple external aero flows.
• Pretty much completely ignored in commercial CFD programs today.
• Much better models are available.

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Baldwin-Lomax Model (Baldwin and Lomax, 1978)
Baldwin-Lomax model was formulated to be used in applications where the
boundary layer thickness and displacement thickness are not easily determined.
The inner viscosity The mixing length
2
𝑣𝑡 = 𝑙𝑚𝑖𝑥 𝜔 𝑦 ≤ 𝑦𝑚 𝑙𝑚𝑖𝑥 = 𝜅𝑦 1 − exp −𝑦 + /𝐴+
0

The outer viscosity

𝑣𝑡 = 𝜌𝛼𝐶𝑐𝑝 𝐹𝑊𝑎𝑘𝑒 𝐹𝐾𝑙𝑒𝑏 (𝑦, 𝑦max /𝐶𝐾𝑙𝑒𝑏 ) 𝑦 > 𝑦𝑚

where 2
𝐹𝑊𝑎𝑘𝑒 = min 𝑦max 𝐹max , 𝐶𝑤𝑘 𝑦max 𝑈𝑑𝑖𝑓 /𝐹max

1
𝐹max = max( 𝑙𝑚𝑖𝑥 𝜔
𝜅
2
𝑦𝑚𝑥𝑧 𝜔
𝐹𝑊𝑎𝑘𝑒 = 𝑦max 𝐹max 𝛿𝑣∗ =
if then 𝑈𝑒

𝑈𝑑𝑖𝑓
2
𝐹𝑊𝑎𝑘𝑒 = 𝐶𝑤𝑘 𝑦max 𝑈𝑑𝑖𝑓 /𝐹max 𝛿=
if then 𝜔

The constants in this model

𝜅 = 0.40 𝛼 = 0.0168 𝐴+
0 = 26 𝐶𝑐𝑝 = 1.6 𝐶𝐾𝑙𝑒𝑏 = 0.3 𝐶𝑤𝑘 = 1

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Sparlat-Allmaras Model (Spalart and Allmaras, 1992)
• Solves a single conservation equation (PDE) for the turbulent
viscosity:
• This conservation equation contains convective and diffusive transport
terms, as well as expressions for the production and dissipation of t.
• Developed for use in unstructured codes in the aerospace industry.
• Economical and accurate for:
• Attached wall-bounded flows.
• Flows with mild separation and recirculation.
• Weak for:
• Massively separated flows.
• Free shear flows.
• Decaying turbulence.
• Because of its relatively narrow use we will not discuss this
model in detail.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Sparlat-Allmaras Model (Spalart and Allmaras, 1992)
𝜇𝑒𝑓𝑓 = 𝜇 + 𝜌𝑣෥𝑡
Turbulent kinematic viscosity 𝑣෤𝑡

𝜕 𝜌𝑣෤𝑡 𝜕 𝜇𝑒𝑓𝑓 𝜕𝑣෤𝑡 𝑣෤𝑡 𝐶𝑏2 𝜕𝑣෤𝑡

+ 𝜌𝑢𝑗 𝑣෤𝑡 − = 𝐶𝑏1 𝜌𝑆ሚ𝑣෤𝑡 − 𝐶𝜔1 𝜌𝑓𝜔 ( )2 + 𝜌
𝜕𝑡 𝜕𝑥𝑗 𝜎𝑣 𝜕𝑥𝑗 𝑦 𝜎𝑣 𝜕𝑥𝑗

k
0.1335 0.622 0.666 7.1 0.3 2.0 0.41

𝐶𝑏1 1 + 𝐶𝑏2
𝐶𝜔1 = 2 + Turbulent viscosity 𝜇𝑡 = 𝜌𝑣෥𝑡 𝑓𝑣1
𝜅 𝜎𝑣

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- Model (Launder and Spalding, 1974)
𝑘2 Velocity scale 𝜗 = 𝑘1/2
Eddy viscosity 𝜇𝑡 = 𝐶𝜌𝜗𝑙 = 𝜌𝐶𝜇
𝜀 Length scale 𝑙 = 𝑘 3/2 /𝜀

Transport equations for 𝑘 and 𝜀

𝜕 𝜌𝑘 𝜇𝑡
+ 𝑑𝑖𝑣 𝜌𝑘𝐔 = 𝑑𝑖𝑣 𝑔𝑟𝑎𝑑 𝑘 + 𝜇𝑡 𝐸𝑖𝑗 ⋅ 𝐸𝑖𝑗 − 𝜌𝜀
𝜕𝑡 𝜎𝑘

𝜕 𝜌𝜀 𝜇𝑡 𝜀 𝜀2
+ 𝑑𝑖𝑣 𝜌𝜀𝐔 = 𝑑𝑖𝑣 𝑔𝑟𝑎𝑑 𝜀 + 𝐶1𝜀 2𝜇𝑡 𝐸𝑖𝑗 ⋅ 𝐸𝑖𝑗 − 𝐶2𝜀 𝜌
𝜕𝑡 𝜎𝜀 𝑘 𝑘

𝐶𝜇 = 0.09 𝜎𝑘 = 1.00 𝜎𝜀 = 1.30 𝐶1𝜀 = 1.44 𝐶2𝜀 = 1.92

Transport Transport Rate of Rate of

Rate of change
+ of 𝑘 or 𝜀 = of 𝑘 or 𝜀 + production - destruction
of 𝑘 or 𝜀
by convection by diffusion of 𝑘 or 𝜀 of 𝑘 or 𝜀

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- Model (Launder and Spalding, 1974)
• The turbulent viscosity is2calculated from:
𝑘
𝜇𝑡 = 𝜌𝐶𝜇 𝐶𝜇 = 0.09
𝜀

• The Reynolds stresses are then calculated as follows:

𝜕𝑈𝑖 𝜕𝑈𝑗 2 2
−𝜌𝑢𝑖 ′𝑢𝑗 ′ = 𝜇𝑡 + − 𝜌𝑘𝛿𝑖𝑗 = 2𝜇𝑡 𝐸𝑖𝑗 − 𝜌𝑘𝛿𝑖𝑗
𝜕𝑥𝑗 𝜕𝑥𝑖 3 3
𝛿𝑖𝑗 = 1 𝑖𝑓 𝑖 = 𝑗 𝑎𝑛𝑑 𝛿𝑖𝑗 = 0 𝑖𝑓 𝑖 ≠ 𝑗

• The (2/3)ρkδij term ensures that the normal stresses sum to k.

• Note that the k-ε model leads to all normal stresses being equal,
which is usually inaccurate.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- Model (Launder and Spalding, 1974)

• Advantages:
• Relatively simple to implement.
• Leads to stable calculations that converge relatively easily.
• Reasonable predictions for many flows.
• Disadvantages:
• Poor predictions for:
• swirling and rotating flows,
• flows with strong separation,
• axisymmetric jets,
• certain unconfined flows, and
• fully developed flows in non-circular ducts.
• Valid only for fully turbulent flows.
• Simplistic ε equation.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
RNG k- Models (Yakhot, 1992)
: RNG procedure systematically removes the small scales of motion from
the governing equations by expressing their effects in terms of larger scale
motions and a modified viscosity

𝜕 𝜌𝑘
+ 𝑑𝑖𝑣 𝜌𝑘𝐔 = 𝑑𝑖𝑣 𝛼𝑘 𝜇𝑒𝑓𝑓 𝑔𝑟𝑎𝑑 𝑘 + 2𝜇𝑡 𝐸𝑖𝑗 ⋅ 𝐸𝑖𝑗 − 𝜌𝜀
𝜕𝑡

𝜕 𝜌𝜀 ∗
𝜀 𝜀2
+ 𝑑𝑖𝑣 𝜌𝜀𝐔 = 𝑑𝑖𝑣 𝛼𝜀 𝜇𝑒𝑓𝑓 𝑔𝑟𝑎𝑑 𝜀 + 𝐶1𝜀 2𝜇 𝐸 ⋅ 𝐸 − 𝐶2𝜀 𝜌
𝜕𝑡 𝑘 𝑡 𝑖𝑗 𝑖𝑗 𝑘
𝑘2
𝜇𝑒𝑓𝑓 = 𝜇 + 𝜇𝑡 𝜇𝑡 = 𝜌𝐶𝜇
𝜀

𝐶𝜇 = 0.0845 𝛼𝑘 = 𝛼𝜀 = 1.39 𝐶1𝜀 = 1.42 𝐶2𝜀 = 1.68

∗
𝜂 1 − 𝜂/𝜂0 1/2 𝑘
𝐶1𝜀 = 𝐶1𝜀 − 𝜂 = 2𝐸𝑖𝑗 ⋅ 𝐸𝑖𝑗 𝜂0 = 4.377 𝛽 = 0.012
1 + 𝛽𝜂3 𝜀

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
RNG k- Models (Yakhot, 1992)
• k- equations are derived from the application of a rigorous
statistical technique (Renormalization Group Method) to the
instantaneous Navier-Stokes equations.
• Similar in form to the standard k- equations but includes:
• Additional term in  equation for interaction between turbulence dissipation
and mean shear.
• The effect of swirl on turbulence.
• Analytical formula for turbulent Prandtl number.
• Differential formula for effective viscosity.
• Improved predictions for:
• High streamline curvature and strain rate.
• Transitional flows.
• Wall heat and mass transfer.
• But still does not predict the spreading of a round jet correctly.
Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Realizable k- model (Shih, 1995)
• Shares the same turbulent kinetic energy equation as the
standard k- model.
• Improved equation for ε.
• Variable Cμ instead of constant.
• Improved performance for flows involving:
• Planar and round jets (predicts round jet spreading correctly).
• Boundary layers under strong adverse pressure gradients or separation.
• Rotation, recirculation.
• Strong streamline curvature.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Realizable k- model (Shih, 1995)
Distinctions from standard k- model:
– Alternative formulation for turbulent viscosity:
𝑘2 1
𝜇𝑡 ≡ 𝜌𝐶𝜇 𝐶𝜇 =
𝜀 where 𝑈 ∗ 𝑘 is now variable.
𝐴𝑜 + 𝐴𝑠 𝜀

• (A0, As, and U* are functions of velocity gradients).

• Ensures positivity of normal stresses: u2i ≥ 0

• Ensures Schwarz’s inequality: (ui uj )2 ≤ u2i u2j

– New transport equation for dissipation rate, :

𝐷𝜀 𝜕 𝜇𝑡 𝜕𝜀 𝜀2 𝜀
𝜌 = 𝜇+ + 𝜌𝑐1 𝑆𝜀 − 𝜌𝑐2 + 𝑐1𝜀 𝑐3𝜀 𝐺𝑏
𝐷𝑡 𝜕𝑥𝑗 𝜎𝜀 𝜕𝑥𝑗 𝑘 + 𝜈𝜀 𝑘

Diffusion Generation Destruction Buoyancy

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Realizable k- model (Shih, 1995)
Eddy viscosity computed from.
𝑘2 1
𝜇t = 𝜌 ⥂⥂ 𝐶𝜇 , 𝐶𝜇 =
𝜀 𝑈∗𝑘
𝐴0 + 𝐴𝑠 𝜀

𝑈∗ ≡ 𝑆𝑖𝑗 𝑆𝑖𝑗 + Ω𝑖𝑗 Ω𝑖𝑗

1
𝐴0 = 4.04, 𝐴𝑠 = 6 cos 𝜑 , 𝜑 = cos −1 6𝑊
3

𝑆𝑖𝑗 𝑆𝑗𝑖 𝑆𝑘𝑖

𝑊= , 𝑆ሚ = 𝑆𝑖𝑗 𝑆𝑖𝑗
ሚ
𝑆

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- Model (Wilcox, 1988)
𝑘
Eddy viscosity 𝜇𝑡 = 𝜌
𝜔

Transport equations for 𝑘 and 𝜔

𝜕 𝜌𝑘 𝜕𝑘 𝜕𝑈𝑖 𝜕 𝜕𝑘
+ 𝜌𝑈𝑗 = 𝜏𝑖𝑗 − 𝛽 ∗ 𝜌𝑘𝜔 + 𝜇 + 𝜎 ∗ 𝜇𝑡
𝜕𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗

𝜕 𝜌𝜔 𝜕𝜔 𝜔 𝜕𝑈𝑖 𝜕 𝜕𝜔
+ 𝜌𝑈𝑗 = 𝛼 𝜏𝑖𝑗 − 𝛽𝜌𝜔2 + 𝜇 + 𝜎𝜇𝑡
𝜕𝑡 𝜕𝑥𝑗 𝑘 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗

𝛼 = 5/9 𝛽 = 3/40 𝛽∗ = 9/100 𝜎 = 1/2 𝜎 ∗ = 1/2

𝜀 = 𝛽 ∗ 𝜔𝑘 𝑙 = 𝑘1/2 /𝜔

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- Model (Wilcox, 1988)

• This is another two equation model. In this model ω is an

inverse time scale that is associated with the turbulence.
• This model solves two additional PDEs:
• A modified version of the k equation used in the k-ε model.
• A transport equation for ω.
• The turbulent viscosity is then calculated as follows:
𝑘
𝜇𝑡 = 𝜌
𝜔

• Its numerical behavior is similar to that of the k-ε models.

• It suffers from some of the same drawbacks, such as the
assumption that μt is isotropic.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- SST Model (Menter, 1993)
The k- based SST model accounts for the transport of the turbulent shear
stress and gives highly accurate predictions of the onset and the amount of
flow separation under adverse pressure gradients.
The BSL model combines the advantages of the Wilcox and the k-ε model,
but still fails to properly predict the onset and amount of flow separation from
smooth surfaces
The main reason is that both models do not account for the transport of the
turbulent shear stress.

This results in an over prediction of the eddy-viscosity.

The proper transport behavior can be obtained by a limiter to the formulation

of the eddy-viscosity:
𝑎1 𝑘
𝜈𝑡 = , 𝑎1 = 0.31
max 𝑎1 𝜔, 𝑆𝐹2

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- SST Model (Menter, 1993)
Again F2 is a blending function similar to F1, which restricts the limiter to the
wall boundary layer, as the underlying assumptions are not correct for free
shear flows. S is an invariant measure of the strain rate.

The blending functions are critical to the success of the method.

Their formulation is based on the distance to the nearest surface and on the
flow variables.
𝐶𝜙 = 𝐹1 𝐶𝜙1 + 1 − 𝐹1 𝐶𝜙2

𝐹1 = tanh arg14

𝑘 500𝑣 4𝜌𝑘
arg1 = min max ∗ , , 𝜔 ,
𝛽1 𝜔𝑦 𝑦 2 𝜔 𝜎𝜔2 𝐶𝐷𝑘𝜔 𝑦 2

Where y is the distance to the nearest wall and  is the kinematic viscosity and:
2𝜌 𝛿𝑘 𝛿𝜔
𝐶𝐷𝑘𝜔 = max 𝜔 , 10−20
𝜔𝜎𝜔2 𝛿𝑥𝑗 𝛿𝑥𝑗

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k- SST Model (Menter, 1993)
𝜔
𝜔
𝜎𝑘1 𝜎𝜔1 𝛽1 𝛽1∗ 𝜅

1.176 2.0 0.075 0.09 0.41

𝛽1 1 𝜅2
𝛼1 = ∗ − 𝜔
𝛽1 𝜎𝜔1 𝛽1∗

𝛽2 1 𝜅2
𝛼2 = ∗ − 𝜔
𝛽2 𝜎𝜔2 𝛽2∗

1 1 𝜕𝜔 𝜕𝑘
𝑆𝜔 = 2 1 − 𝐹1 𝜔
𝜎𝜔2 𝜔 𝜕𝑥𝑗 𝜕𝑥𝑗
𝑘 500𝑣
𝐹2 = tanh arg 22 arg 2 = max ,
𝛽1∗ 𝜔𝑦 𝑦 2 𝜔

𝜔 𝜔
𝜎𝑘2 𝜎𝜔2 𝛽2 𝛽2∗ 𝜅

1.0 1.168 0.0828 0.09 0.41

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
𝑘−𝜀−𝛼
Turbulence Modeling
k-- Model (Ghirelli, 2007)
k-- model is a development of the k- model (Launder and Spalding, 1974)
2 𝑘2 𝛾
𝑛
Eddy viscosity 𝜈𝑡 = 𝑆𝑐𝑇 𝑐𝜏 1−𝛾 𝑐𝛼 𝛼𝜀
3 𝜀 = exp −
𝑐𝜏 𝑘 σ𝑛−1
𝑖=0 𝛾
𝑖

Transport equations for 𝑘, 𝜀 and 𝛼

𝐷𝑘 𝜕 𝜈𝑡 𝜕𝑘
= 𝜈+ +𝐺−𝜀
𝐷𝑡 𝜕𝑥𝑖 𝜎𝑘 𝜕𝑥𝑖

𝐷𝜀 𝜕 𝜈𝑡 𝜕𝜀 𝜀 𝜀2
= 𝜈+ + 𝑐1 𝐺 − 𝑐2
𝐷𝑡 𝜕𝑥𝑖 𝜎𝑘 𝜕𝑥𝑖 𝑘 𝑘

𝐷 𝑘𝛼 𝜕 𝜈𝑡 𝜕 𝑘𝛼
= 𝜈+ + 𝑘 − 𝛼𝜀
𝐷𝑡 𝜕𝑥𝑖 𝜎𝑘 𝜕𝑥𝑖

𝑐𝜏 = 0.19 𝑐𝛼 = 0.2 𝑐1 = 1.21 𝑐2 = 1.77 𝜎𝑘 = 0.6 𝜎𝜀 = 𝑛 = 16

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
k−ε−v ′ 2 Model (Durbin, 1995)
The turbulence model was adapted to consider the wall proximity and anisotropic effect.

1 𝜕 𝑣 ′2 :Normal turbulent intensities

𝑗
𝜌𝑢𝑚 𝜑 − 𝑞𝑚 𝛽𝑚 = 𝑆𝜙 𝑓22 : Elliptic relaxation equation
𝐽 𝜕𝑥 𝑗
𝑧
0.25
𝐶𝜀1 = 1.3 +
[1 + 𝑑/2𝑙𝑠𝑐𝑎𝑙𝑒 8 ]

𝜕𝑈𝑖 𝑛 𝜕𝑈𝑗 𝑛 𝜕𝑈𝑖 𝑛

𝜑 Γ𝜑,𝑒𝑓𝑓 𝑆𝜑 𝑃𝑘 = 𝜇𝑡 𝛼 + 𝑚 𝛼𝑗 𝛼
𝜕𝑥 𝑛 𝑗 𝜕𝑥 𝜕𝑥 𝑛 𝑗
𝜇𝑡
𝜇+ 𝑃𝑘 − 𝜌𝜀
𝑘 𝜎𝑘
Length and time scale
𝜇𝑡 𝑧
𝐶𝜀1 𝑃𝑘 − 𝐶𝜀2 𝜌𝜀
𝜇+ 𝑘 3/2 𝜈 3/4
𝜀 𝜎𝜀 𝑙𝑠𝑐𝑎𝑙𝑒 = max , 𝐶𝜂 1/4
𝑇𝑠𝑐𝑎𝑙𝑒 𝜀 𝜀
𝜇𝑡 𝜀
𝜇+ 𝜌𝑘𝑓22 − 𝜌𝑣 ′ 2 𝐿𝑡 = 𝐶𝐿 𝑙𝑠𝑐𝑎𝑙𝑒
𝑣 ′2 𝜎𝑘 𝑘
𝑇𝑠𝑐𝑎𝑙𝑒
1
𝑘 𝜈 2
= max , 6
2
1 𝜕 𝑗 𝜕𝑓22 𝑛 [2/3 − 𝑣 ′ 2 /𝑘] 𝑃𝑘 𝜀 𝜀
𝐿 𝜌𝐽𝛼 − 𝛼 = 𝜌 𝐶1 − 1 + 𝐶2 − 𝜌𝑓22
𝐽 𝜕𝑥 𝑗 𝑚
𝜕𝑥 𝑛 𝑚 𝑇𝑠𝑐𝑎𝑙𝑒 𝑘

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Reynolds Stress Equation Models (Launder et al., 1975)
𝐷𝑅𝑖𝑗 1 2 3 4 5 Reynolds stress
= 𝑃𝑖𝑗 + 𝐷𝑖𝑗 − 𝜀𝑖𝑗 + Π𝑖𝑗 + Ω𝑖𝑗 𝑅𝑖𝑗 = −𝜏𝑖𝑗 /𝜌 = 𝑢′𝑖 𝑢′𝑗
𝐷𝑡

𝜕𝑈𝑗 𝜕𝑈𝑗
𝑃𝑖𝑗 = − 𝑅𝑖𝑚 + 𝑅𝑗𝑚
1 Rate of production of 𝑅𝑖𝑗 𝜕𝑥𝑚 𝜕𝑥𝑚

𝜕 𝑣𝑡 𝜕𝑅𝑖𝑗 𝑣𝑡
2 Transport of 𝑅𝑖𝑗 by diffusion 𝐷𝑖𝑗 = = 𝑑𝑖𝑣 𝑔𝑟𝑎𝑑 𝑅𝑖𝑗
𝜕𝑥𝑚 𝜎𝑘 𝜕𝑥𝑚 𝜎𝑘

2
3 Rate of dissipation of 𝑅𝑖𝑗 𝜀𝑖𝑗 = 𝜀𝛿𝑖𝑗
3

4 Transport of 𝑅𝑖𝑗due to turbulent pressure-strain interactions

𝜀 2 2
Π𝑖𝑗 = −𝐶1 𝑅 − 𝑘𝛿 − 𝐶2 𝑃𝑖𝑗 − 𝑃𝛿𝑖𝑗 𝐶1 = 1.8 𝐶2 = 0.6
𝑘 𝑖𝑗 3 𝑖𝑗 3

5 Transport of 𝑅𝑖𝑗due to rotation Ω𝑖𝑗 = −2𝜔𝑘 𝑅𝑗𝑚 𝑒𝑖𝑘𝑚 − 𝑅𝑖𝑚 𝑒𝑗𝑘𝑚

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Reynolds Stress Equation Models (Launder et al., 1975)

The various terms are modeled as follows:

– Production Pij is retained in its exact form.
– Diffusive transport Dij is modeled using a gradient diffusion assumption.
– The dissipation εij, is related to ε as calculated from the standard ε
equation, although more advanced ε models are available also.
– Pressure strain interactions πij, are very important. These include pressure
fluctuations due to eddies interacting with each other, and due to
interactions between eddies and regions of the flow with a different mean
velocity. The overall effect is to make the normal stresses more isotropic
and to decrease shear stresses. It does not change the total turbulent
kinetic energy. This is a difficult to model term, and various models are
available. Common is the Launder model. Improved, non-equilibrium
models are available also.
– Transport due to rotation Ωij is retained in its exact form.

Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Algebraic Stress Equation Models (Naot and Rodi, 1982)
: Assume that the sum of the convection and diffusion terms of the Reynolds
stresses is proportional to the sum of the convection and diffusion terms of
turbulent kinetic energy

Du 'i u ' j u 'i u ' j  Dk 

− Dij   −  k − transport _ terms  
Dt k  Dt 

( )
u 'i u ' j
=  −u 'i u ' j  Eij − 
k

2 𝐶𝐷 2 𝑘
Algebraic stress model 𝑅𝑖𝑗 = 𝑢′𝑖 𝑢′𝑗 = 𝑘𝛿𝑖𝑗 + 𝑃𝑖𝑗 − 𝑃𝛿𝑖𝑗
3 𝐶1 − 1 + 𝑃/𝜀 3 𝜀

𝐶𝐷 = 0.55 𝐶1 = 2.2

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence Modeling
Algebraic Stress Equation Models (Naot and Rodi, 1982)
• The same k and ε equations are solved as with the standard k-ε
model.
• However, the Boussinesq assumption is not used.
• The full Reynolds stress equations are first derived, and then
some simplifying assumptions are made that allow the
derivation of algebraic equations for the Reynolds stresses.
• Thus fewer PDEs have to be solved than with the full RSM and
it is much easier to implement.
• The algebraic equations themselves are not very stable,
however, and computer time is significantly more than with the
standard k-ε model.
• This model was used in the 1980s and early 1990s. Research
continues but this model is rarely used in industry anymore now
that most commercial CFD codes have full RSM
implementations available.
Ref. : Applied Computational Fluid Dynamics, Lecture 10 –Turbulence models, André Bakker (2002-2005), Fluent Inc.(2002)
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall
• The velocity profile near the wall is important:
– Pressure Drop
– Separation
– Shear Effects
– Recirculation
• Turbulence models are generally suited to
model the flow outside the boundary layer
• Examination of experimental data yields a wide
variety of results in the boundary layer

The above graph shows non-

dimensional velocity versus non-
dimensional distance from the
wall. Different flows show
different boundary layer profiles.

Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall
• By scaling the variables near the wall the velocity profile data takes on
a predictable form (transitioning from linear to logarithmic behavior)

Scaling the non-dimensional

velocity and non-
dimensional distance from
the wall results in a 𝑦+ , 𝑦 + ≤ 𝑦𝑚
+

predictable boundary layer 𝑢 + = ቐ1

ln 𝐸𝑦 + , 𝑦 + ≥ 𝑦𝑚
+
profile for a wide range of 𝜅
flows

• Since near wall conditions are often predictable, functions can be used to
determine the near wall profiles rather than using a fine mesh to actually
resolve the profile
– These functions are called wall functions
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall

• High Reynolds Model : Wall function

• Low Reynolds Model : Damping function

• Two-layer model:
• Viscous sublayer :Damping function
• Region except boundary layer : Wall funtion

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall
High Reynolds No. Model : Wall Function
• y+ is the non-dimensional distance from the wall
– It is used to measure the distance of the first node away from the wall

y
u
Boundary layer
Wall functions used to
resolve boundary layer

y+

• Wall functions are only valid within specific y+ values

• If y+ is too high the first node is outside the boundary layer and wall functions
will be imposed too far into the domain
• If y+ is too low the first node will lie in the laminar (viscous) part of the boundar
y layer where wall functions are not valid
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall
High Reynolds No. Model : Wall Function
𝑦+ , 𝑦 + ≤ 11.63 𝑦𝑢𝜏 𝜏𝑤 1/4
𝑢 + = ቐ1 𝑦+ = , 𝑢𝜏 = = 𝐶𝜇 𝑘 1/2
ln 𝐸𝑦 + , 𝑦 + ≥ 11.63 𝜈 𝜌
𝜅

𝜏𝑤
1/4 1/2
𝜎𝜅𝐶𝜇 𝑘𝑃 𝑢𝑃 − 𝑢𝑤
, 𝑦 + ≤ 11.63
ln 𝐸𝑦𝑃+
=
𝜇 𝑢𝑃 − 𝑢𝑤
, 𝑦 + ≥ 11.63
𝑦𝑝

𝜕𝑢 𝜕𝑢 𝑢𝑝 𝑢𝑝
𝑃 = 𝜏𝑤 = 𝜌𝑢𝜏2 = 𝜌𝑢𝜏 +
𝜕𝑛 𝜕𝑛 𝑢 𝑦𝑝
𝑝
3/2
𝑢𝜏3 3/4 𝑘𝑝
𝜀𝑃 = = 𝐶𝜇
𝜅𝑦𝑝 𝜅𝑦𝑝

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Turbulence near the wall
High Reynolds No. Model : Wall Function
• In some situations, such as boundary layer separation, wall functions do
not correctly predict the boundary layer profile

Wall functions applicable Wall functions not applicable

• In these cases wall functions should not be used
• Instead, directly resolving the boundary layer can provide accurate
results
• Not all turbulence models allow the wall functions to be turned off
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Wall Treatment
Low-Reynolds No. Model : Damping Function

Damping functions
y 2 5.29
𝑓𝜇 = 1 − 𝑒 −0.0198 R𝑡 1+
u Re𝑦

𝑦 𝑘 : Reynolds number
Re𝑦 =
𝜈
𝑘2
R𝑡 = : Turbulent Reynolds number
𝜈𝜀
2𝜈𝑘
𝜀𝑃 = : Dissipation at the near-wall
Damping functions used to 𝑦2

resolve boundary layer

• Low Reynolds model is applied in the laminar (viscous) part of the

boundary layer where damping function are valid. Generally y+ at first
node is used for about 1~2.
Ref. : ANSYS, Inc. Proprietary © 2009 ANSYS, Inc. All rights reserved.
Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Wall Treatment
Two-layer Model :
Viscous sublayer : Damping Function
Region except boundary layer : Wall function

Ref. STAR-CD Ver3.24 manual

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Cross Flow Fan RANS Simulation

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)

• Dependent variables decomposed into large-scale

(resolved) and subgrid-scale (unresolved) parts
• The large scales (~ integral scale), which depend on the
boundary conditions and are flow dependent are
computed
• The small scales, which are more universal (less
dependent on boundary conditions), are modelled
• Large scales contribute most of the Reynolds stresses
• Yield the 3D, instantaneous velocity field due to the large
scales, and statistical information
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Resolution requirement : LES

Large (integral)
scale L

Only resolve the integral length scale (energy containing eddies)

Integral scale varies only weakly with Reynolds number
Computational cost of LES away from solid boundaries varies
slowly with Reynolds number
Infinite Reynolds number computations are possible
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Resolution requirement : LES
Boundary layers
Outer layer (Chapman 1979)

• Only scales of order d need to be resolved

Inner layer (Chapman 1979)

• Must resolve near-wall eddies

Computational cost ~ Re2.4
While not as demanding as DNS, LES still poses considerable computational
cost

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Governing Equations

Separate the large and the small scales by the filtering operation:

Filter width
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Governing Equations
Apply the filtering operation to the Navier-Stokes equations

Filtered equations of motion…

Subgrid-scale (SGS) stress (which needs to be modeled)

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Modeling considerations
• LES velocity fields contain substantially more information than RANS
solutions (frequency, wavenumber).
• This information can be used to improve SGS models

K = cutoff wavenumber ~ 1/D (filter width)

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Subgrid modeling

are SGS stresses that require closure

Subgrid stress due mostly to the small scale, but may also contain some large-scale
contributions as well

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Energy Transfer Mechanism

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Subgrid modeling

Two types of energy-exchange

mechanisms are important:
Local (in wave-number)
interactions
Distant interactions

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Interaction between scales
P(k) T(k)
Kraichnan (1976);
 Energy transfer from
Domaradzki et al wavenumbers above k to
(1993): wavenumbers above Kc
1.0 3

75% of the energy 0.8

transferred to 2
0.6
scales smaller than D Normalized
0.4 eddy viscosity
comes from scales 1
between D and 2D 0.2

0.0 0
0.0 0.1 1
1 k/Kc

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Subgrid-sclae modeling
Most models introduce an SGS eddy
viscosity

Express the eddy viscosity as the product between the velocity and length scale

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Equilibrium assumption
Production of SGS energy:
balances viscous dissipation:

Leads to:

Obtain the Smagorinsky (1963) model

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Smagorinsky model
•
• Since the constant CS (the Smagorinsky constant) is real, the model
is absolutely dissipative:

• To evaluate CS assume a spectrum with an inertial range:

• Integrate the dissipation spectrum k2 E(k) over all resolved

wavenumbers:

• With Ko=1.41 this gives CS @ 0.18

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Subgrid modeling for the fluid

Smagorinsky model suffers from some well known deficiencies

• Model coefficient requires adjustment in sheared turbulence, transitional
flows, near solid surfaces, etc.
• Does not account for local interactions
• Absolutely dissipative (cannot account for backscatter of energy)
Many improvements possible…
• Two-point closures (Kraichnan 1976, Chollet and Lesieur 1981, Metais and
Lesieur 1992)
• Scale-similar and mixed models (Bardina, Ferziger and Reynolds 1980,
Anderson and Meneveau 1999, … )
• Dynamic models (Germano, Piomelli, Cabot and Moin 1991, Moin, Squires,
Cabot and Lee 1991, Meneveau, Lund and Cabot 1996)
• Deconvolution models (Shah and Ferziger 1995, Domaradzki and Loh 1999,
Stoltz and Adams 1999… )
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Scale-similar and mixed models

Scale-similar models are based on

the assumption that the most active
subgrid scales are those closer to
the cutoff, and that the scales with
which they interact most are those
right above the cutoff (Bardina et al.
1980).
The largest subgrid scales can be obtained by filtering the SGS
velocity to obtain .
A Smagorinsky model is added to represent the dissipative effect of
the small scales, to give

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Bardina et al.Computational
Multiphysics (1983), Fluid
Meneveau 2001, Department
Dynamics Laboratory, … of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Dynamic models

In dynamic models for the subgrid-scale

stresses, the model coefficients are
computed dynamically as the calculation
progresses
Define a test filter with width
Consider the identity

The resolved turbulent stresses Lij are the contribution from the region
between test-filter and grid-filter scale
The subtest stresses are obtained by applying the test
filter to the filtered Navier-Stokes equations
Ref, Germano et al. (1991)
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES
Vast majority of LES calculations performed
today resolve the wall layer
Enforce no-slip condition, have fine wall-
parallel grids
Resolution of the inner layer

• An increasing percentage of points goes into the inner layer with

increases in Reynolds number
Alternative for high Reynolds number applications
• Bypass the wall layer
• Model its effect in a global (RANS-like) sense

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES

Purpose: relate the wall stress to the outer layer velocity locally and
instantaneously

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES
Purpose: relate the wall stress to the outer layer velocity locally and
instantaneously

Contours of tw in a plane channel.

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES

Equilibrium models
• Deardorff (1970), Schumann
(1974), Piomelli et al. (1988);
Balaras et al. (1995)
• Some form of generalized
law-of-the-wall is imposed to
relate ut to uo
• Solve for ut, given uo.
• Modifications to the velocity profile
can account for:
– Low Reynolds-number flows
– Moderate pressure gradients
– Transpiration

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES

Zonal approaches

Solve a different set of differential equations in the inner layer

Typically, hybrid LES/RANS approaches
Inner-layer structures:
Time-scales shorter than time-step
Length-scales smaller than grid size
Their effects can be taken into account globally only
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES

Backscatter modeling
TKE Production
The force
generates Energy input
significant by the model
energy and
enstrophy
The force
generates
rotational small Enstrophy
scales in the with model
interface region Enstrophy
w/o model Curl of the
force

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES Backscatter modeling

• The small scales support additional Reynolds shear stress.

Contours of streamwise velocity fluctuations
Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Large Eddy Simulation (LES)
LES (Balaras and Benocci, 1994)
Wall layer modeling for LES

Backscatter modeling
w/o model

with model

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Cross Flow Fan LES Simulation

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Hybrid RANS/LES
DES (Spalart, Jou, Stretlets, and Allmaras, 1997)

𝜕 𝜇𝑡 2
𝜌𝑘 + ∇ ⋅ 𝜌𝑢𝑘 = 𝜇 + ∇ 𝑘 + 𝜌𝑃𝑘 − 𝐷𝑘
𝜕𝑡 𝜎𝑘

𝜕 𝜇𝑡 2 𝜀 𝜀2
𝜌𝜀 + ∇ ⋅ 𝜌𝑢𝜀 = 𝜇 + ∇ 𝜀 + 𝜌𝛼𝑃𝑘 − 𝜌𝐶𝜀2
𝜕𝑡 𝜎𝜀 𝑘 𝑘
𝜌𝐶𝜇 𝑘 2
Turbulent viscosity 𝜇𝑡 =
𝜀

Dissipative source term : , The length scale

3 3
𝜌𝑘 2 𝑘2
𝐷𝑘 = 𝑙ሚ = min( , 𝐶𝐷𝐸𝑆 Δ)
𝑙ሚ 𝜀

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Hybrid RANS/LES
DES (Spalart, Jou, Stretlets, and Allmaras, 1997)
Wall layer modeling for LES

 Applied as a wall model by Nikitin et al (2000).

 Channel flow, Re t = 180-80,000.

 Sustained turbulence with a smooth inner layer

 Transition between RANS and DNS zones characterized

by "superstreaks" and increased velocity gradient.

 LES log layer had high intercept.

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Hybrid RANS/LES
DES (Spalart, Jou, Stretlets, and Allmaras, 1997)
Wall layer modeling for LES

Ref ,Nikitin et al. (2000)

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Hybrid RANS/LES
DES (Spalart, Jou, Stretlets, and Allmaras, 1997)
Wall layer modeling for LES

Contours of streamwise velocity fluctuations

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Hybrid RANS/LES
DES (Spalart, Jou, Stretlets, and Allmaras, 1997)
Wall layer modeling for LES

Detached-Eddy Simulation applied “as is” leads to an additional

buffer layer between the modeled wall layer and resolved outer
layer
Piomelli, Balaras, Pasinato, Squires and Spalart (2003) addressed
this deficiency using a backscatter model
Backscatter model in the interface region to represent the
energy feedback due to the unresolved small scales
Divergence-free random forcing fi is added to the RHS of the
momentum equations

Ref. : Large Eddy Simulation for Particle-Laden Turbulent Flows, Kyle D. Squires, Dispersion of Particles in Turbulent flow .

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University
 Detached Eddy Simulation (DES)
DES Method

Multiphysics Computational Fluid Dynamics Laboratory, Department of Mechanical Engineering, Sogang University