MCDM608L: COMPUTATIONAL

FLUID DYNAMICS (CFD)

M.Tech (WS24-25)

Sivakumar, R
SMEC, VIT Chennai
 Module-7

Turbulence Modeling:
Basic equations of Turbulence: Derivation of
turbulence using non-dimensional analysis,
Reynolds averaging, Reynolds averaged N-S
equations, Eddy viscosity hypothesis, Reynolds
Stress Transport Equations. First order closures: k-ε
two equation models, SST k-ω model. Large Eddy
Simulations (7 hours)
Why to study turbulence?
Why to study Turbulence?

 Understanding turbulence is of great importance

to engineers

 Not able to predict turbulent behavior with any

degree of reliability, even in very simple flow
situations

 Almost all fluid flows which we encounter in daily life

is turbulent

SMBS, VIT Chennai

Why to study Turbulence?

Examples:
 Flow around cars, aeroplanes and buildings
 Boundary layers and wake around and after bluff
bodies such as cars, aeroplanes and buildings
 Flow and combustion in IC engines, gas turbines,
combustors
 Air movements in rooms
 Flow around chimney

SMBS, VIT Chennai

Why to study Turbulence?

Wake of Submarine Swirl & Eddies in a Fast Flowing River

Examples

Ash Plume of Volcano Outer Layer of Sun

SMBS,Source:
VIT Chennai
http://www.cora.nwra.com/~werne/eos/text/turbulence.html
Why to study Turbulence?
Examples

Flow around propeller

Pictures of Jupiter Source: http://www.cora.nwra.com/~werne/eos/text/turbulence.html

Most flows in nature & technical applications are turbulent

SMBS, VIT Chennai
What is turbulence?
 What is Turbulence?

• A Fluid motion in which velocity, pressure, and other flow

quantities fluctuate irregularly in time and space.

• “Turbulent Fluid motion is an irregular condition of flow in which

the various quantities show a random variation with time and
space coordinates, so that statistically distinct average values can
be observed.”
- Hinze

• “Turbulence is due to the formation of point or line vortices on

which some component of the velocity becomes infinite.”
-Jean Leray
What is Turbulence?
 3D, unsteady, irregular motion of particles in a
practically chaotic manner
 Flow field is a random process
 Contrast to the laminar state

u = ū + u’(t)

ū - Mean velocity
u’(t) -Fluctuating component
Probe response in turbulent
flow
Reynold’s Experiment

Laminar Flow
Turbulent Flow
Re < 2300

 Maintains Stability  Dye filament explode

 Mixing is molecular  Rapidly mix across the
tube
 Concentration
gradient is so small
Laminar – Turbulent Transition

Laminar Flow at Transition at Turbulent Flow at

low Re intermediate Re high Re

At a Reynolds number less than the critical, the kinetic energy of flow is not
enough to sustain the random fluctuations against the viscous damping and in such
cases laminar flow continues to exist.
At somewhat higher Reynolds number than the critical Reynolds number, the
kinetic energy of flow supports the growth of fluctuations and transition to
turbulence takes place.
What is Turbulence?

 Turbulent fluctuations are macroscopic in nature

and are subjected to continuum laws

 Identifiable swirling patterns characterize turbulent

eddies

 Turbulence  viewed as superposition of eddies

of varying sizes

 Eddy  Lumps of turbulent fluid masses or Eddies

are essentially some lumps of fluid which are
rotating.
 What is Turbulence?

Flow past a cylinder

Alternate shedding – called Karman

Vortex Street

http://www.grc.nasa.gov/WWW/k-
12/airplane/dragsphere.html
Characteristics of turbulent flow
Characteristics of Turbulent Flow
 Irregularity

 Non-repeatability
 Diffusivity
 High Reynolds number
 Three dimensional vorticity fluctuations
 Dissipation
 Continuum
Characteristics of Turbulent Flow
 Irregularity

 Irregular, random and chaotic

 Flow consists of a spectrum of eddy sizes
 No exact definition of turbulent eddy
 Deterministic & described by NS equation

Small Large
structures structures

SMBS, Source
VIT Chennai
: http://www.biophysics.uwa.edu.au/turbulence/animations/
Characteristics of Turbulent Flow
 Non-repeatability

 Sensitive to initial and boundary conditions

 Averages over large intervals of space and

time may be very well defined and stable

SMBS, VIT Chennai

Characteristics of Turbulent Flow
 Diffusivity  Causes rapid mixing & increased rate of
heat, mass and momentum transfer
 Spreading rate of boundary layers, jets
increases
 Increases the resistance (wall friction) in
internal flows
 This is a single most important aspect of
turbulence from a engineering point of view

SMBS, VIT Chennai

Characteristics of Turbulent Flow

 Momentum Diffusivity  = l + t (m2/s)

 Thermal Diffusivity α = αl + αt (m2/s)
 Mass Diffusivity D = Dl + Dt (m2/s)

 l , αl and Dl  Molecular level fluid properties.

These depend on type of fluid, temperature level,
etc.,

 t , αt and Dt  Depend on flow geometry, flow

conditions, etc.,
Turbulence is a
 t ≈ αt ≈ Dt for all fluids feature of flow
SMBS, VIT Chennai
Characteristics of Turbulent Flow

 Increased mixing makes the velocity profile much

flatter than in laminar flow

Figure: Comparison of velocity profiles

 Increased velocity and temperature gradients

 Pressure drop & heat transfer in a pipe?
 Drag experienced by a flat plate?
SMBS, VIT Chennai
 Characteristics of Turbulent Flow
 High Reynolds number

Transition to
Turbulent flow occurs
at Re > 2300

 Boundary layer over a flat plate

 Re = 60000 at point A
 Re = 5 x 105 at point B
SMBS, VIT Chennai
Characteristics of Turbulent Flow

SMBS, VIT Chennai

Characteristics of Turbulent Flow
 Three Dimensional Vorticity Fluctuations

 Highly Rotational
 Vortex Dynamics plays an important role
 Energy is transferred from large to small scales by
the interaction of vortices

Vortices

SMBS, VIT Chennai

Source: http://www.math.waikato.ac.nz/~seano/research/turbulence-pictures.html
Characteristics of Turbulent Flow
 Continuum
 Small turbulent scales in the flow are very
much larger than molecular scale

 Flow scales exceed the molecular scales by

three or more orders of magnitude  RANS
equation can be used

 For continuum

0.001 < Knudsen number < 0.01

Characteristics of Turbulent Flow
 Dissipation
 Always dissipative
 Kinetic energy in the small eddies are
transformed into thermal energy due to viscous
shear stress
 Due to steeper gradient near the wall, the
shear stress is greater

SMBS, VIT Chennai

Length Scales in Turbulent Flows
 Used to characterize the flows

 Wide range of length scales (also the time scales)

present
 Eddies:

Smallest eddy
(Molecular Length Scale
Largest eddy
(system length scale
Characteristics of Turbulent Flow
 Dissipation
 Cascading process
Energy mopped up
due to viscous
Extracts energy
dissipation
from the mean flow

Figure: Cascade process with a spectrum of eddies

Characteristics of Turbulent Flow
 Dissipation
Richardson
• L.F. Richardson (“Weather Prediction by Numerical Process.”
Cambridge University Press, 1922) summarized this in the
following often cited verse:

Big whorls have little whorls

Which feed on their velocity;
And little whorls have lesser whorls,
And so on to viscosity
in the molecular sense.
Types of averaging

 Time averaging
 Space averaging
 Ensemble averaging
Types of averaging

 Time averaging
At a given location, measure the property as a
function of time and then take the average
Mathematical representation
 Types of averaging

 Space averaging
 At a given time, average over position

 Ensemble averaging
 Statistical average of repeated experiments under
different condition
Turbulence Modeling
Approaches to predict turbulent flows

 Analytical: No analytical solution exist

 Experimental:
 Expensive
 Time consuming
 Sometimes impossible
 CFD: Promising, the need for turbulence modeling
Why Model Turbulence?
 Most engineering flows we study is turbulence
 Turbulence spans wide spatial and temporal scales
 When described in terms of eddies, non-linear
interaction is complex  cannot be attacked
statistically
 Generating turbulence information by solving the full
Navier-Stokes equations remains incomplete 
Leading to modeling
 Only the effects of the turbulence on the mean flow
are usually sought
 Resolving the details of turbulent fluctuations is not
required for many engineering applications
Types of Turbulence Modeling

 Direct Numerical Simulation (DNS)

 Reynolds-averaged Navier-Stokes Equations (RANS)
RANS
 Zero equation model
 One Equation model
 Two equation model
 Reynolds stress (Seven equation) model
 Algebraic stress model
Computational Cost
Increases
 Space filtered equations
 Large Eddy Simulation (LES)
 Detached Eddy Simulation (DES) LES
 Types of Turbulence Modeling

http://www.personal.psu.edu/lnl/
Choosing a Turbulence Model

 No single model is universally accepted for all classes of

problems
 Depends on
 Physics encompassed in the flow
 Level of accuracy required
 Available computational resources
 Amount of time available for simulation
Choosing a Turbulence Model
Choosing a Turbulence Model

 Turbulence modeling comes in varying degrees of

complexity
 There is not yet a single, practical turbulence model
that can reliably predict all turbulent flows with
sufficient accuracy
 Determining the right choice of turbulence model
depends on the detail of results expected
Direct Numerical Simulation (DNS)
 Direct Numerical Simulation (DNS)
 Strictly speaking, DNS is not a turbulence model at all
 Most straightforward approach to the solution of
turbulent flows
 Governing equations are discretized and solved
numerically
 Completely free of modeling assumptions
 Does not use any averaging technique
 Errors introduced only by numerical approximation
 Useful tool for the study of transitional and turbulent flow
physics
Direct Numerical Simulation (DNS)

 Eddy sizes  metres to microns

 Fluctuations  seconds to microseconds
 Thus, for capturing the effects of the whole range of
eddies,
mesh size < smallest significant eddy
time step < smallest eddy time scale
 For any practical problem, the number of grid points
required may be of the order of 1012 and several
million time steps may be required
 Computer resources are immense. Very powerful and
very fast and parallel computing are needed
Direct Numerical Simulation (DNS)
 To resolve all scales - too many grids and short time
marching needed
 Number of grid points required  (Re)9/4
 Number of time-steps ≈ (Re)-1/2
 Cost of the computation scales  (Re)3
 Grid requirements increases rapidly with Re
 Limited to simple geometries at low Re
Channel flow simulations at Re ≈ 40000:
 These were performed on a 2048 processor machine,
required 6×106 processor hours (the entire machine for
around 4 months), and produced around 25TB of raw
data
Direct Numerical Simulation (DNS)
 . . . but this is the best way of gathering detailed
information on turbulent interaction; mean properties, first
and second moments, two-point correlations etc. in full fields
 Unrealistically expensive for flows of engineering interest
 . . . but still has a role: gathering data for model evaluation
and fundamental understanding of turbulent interaction
 Compared to experiments, much more reliable and
complete data sets, including correlations and
visualization
 Can be used to guide turbulence model development
 It is not a tool for everyday engineering use
Reynolds Time Averaging
Reynolds Time Averaging
 For turbulent flow, because of the fluctuations, every velocity and
pressure term in the continuity and momentum equations is a
rapid varying function of time and space
 Cannot handle such instantaneous fluctuating variables
 Attention is towards the average or mean values of velocity,
pressure, shear stress etc., in the turbulent flow
 Fine details of all the turbulent eddies are not required
Reynolds Time Averaging

when T  ∞

T is the averaging period taken to be longer than any

significant period of the fluctuations themselves
Governing equations (Incompressible flow)

u v w
Continuity equation:   0
x y z

u u u u p
x-mom.: ( u v  w )     2 u  g x
t x y z x

v v v v p
y-mom.:  (  u  v  w )     2 v  g y
t x y z y

w w w w p
z-mom.: (  u v  w )     2 w  g z
t x y z z
T T T T
Heat balance: C p ( u v  w )  k 2T  Q
t x y z

418
Governing equations (Compressible flow)

Mass balance:   ( u )  ( v)  ( w)

   0
t x y z
 ( u )  ( u 2 )  ( uv)  ( uw)  xx  xy  xz
       g x
Momentum t x y z x y z
Balance  ( v)  ( uv)  ( v 2 )  ( vw)  yx  yy  yz
       g y
(x,y,z) t x y z x y z
equations:  ( w)  ( uw)  ( vw)  ( w 2 )  zx  zy  zz
       g z
t x y z x y z
u 2  v 2 w v
 xx   p  2  (.V )  yy   p  2   (.V )  yz   zy   (  )
x 3 y 3 y z

w 2  u v u w
 zz   p  2  (.V )  xy   yx   (  )  xz   zx   (  )
z 3 y x z x
Energy  ( e)  ( uH )  ( vH )  ( wH )
    .(kT )    Q
equation t x y z
SMBS, VIT Chennai 419
Reynolds Time Averaged Conservation Equations

 Continuity Equation

Reynolds
 X- Momentum Equation stress

Reynolds Time Averaged Conservation Equations

 The additional terms


called
Reynolds stresses or turbulent stresses
 These must be modeled in order to close the momentum
equation
 Called closure problem
 “Closure”  Sufficient number of equations for all the
unknowns

The function of turbulence modeling is to devise approximations for the unknown

correlations in terms of flow properties that are known so that a sufficient number of
equations exist. In making such approximations, we close the system.
SMBS, VIT Chennai
Modeling the Reynolds Stress

 There are many RANS models available in the

literature, and in CFD codes
 Some are fairly simple algebraic forms, whilst others
involve solving additional transport equations
 In principle, the more complex forms can have more
flow physics embedded in them, and might therefore
be expected to be more reliable over a wider range of
flows
Boussinesq Approach
 Many turbulent models based on this approach
 Relates Reynolds stress to mean velocity
gradients
 Turbulence was found to increase with mean
velocity gradient

 Important parameter that needs determination

will be the eddy viscosity [µt](Pa.s)
 Turbulent viscosity is used to close the
momentum equation
Boussinesq Approach
 Low computational cost

 Spalart-Allmaras model  One transport

equation to solve μt
 k-ε and k-ω models  Two transport equations
to solve μt as a function of
turbulent kinetic energy (k) and
turbulence dissipation rate (ε)
 µt varies in space
 Assumes μt is isotropic (non-directional) which is not
strictly true
One Equation Model
One Equation Model
 One-equation models based on turbulence kinetic energy are
incomplete as they relate the turbulence length scales to some
typical flow dimension. They are rarely used
 One-equation models based on an equation for the eddy
viscosity are complete such as Baldwin-Barth model and
Spalart-Allmaras model
 Spalart-Allmaras model can predicts better results than
Baldwin-Barth model, and much better results for separated
flow than Baldwin-Barth model and algebraic models
 Also most of DES simulations are based on the Spalart-Allmaras
model
One Equation Model: Spalart Allmaras Model
 Simple one equation model
 Solves a modeled transport equation for the kinematic
eddy (turbulent) viscosity
 Designed specifically for aerospace applications involving
wall bounded flows.
 Produce good results for boundary layers subjected to
adverse pressure gradients
 Suitable model for relatively crude simulations on coarse
meshes where accurate turbulent flow computations are
not critical
 Not suitable for massive separated flows, free shear flows,
decaying turbulence
 One Equation Model: Spalart Allmaras Model

Transport equation

Storage Convective

constant

Production of
Destruction of
turbulent turbulent
viscosity
viscosity

 Destruction of turbulent viscosity that occurs in the near-wall

region due to wall blocking and viscous damping
 One Equation Model: Spalart Allmaras Model

Turbulent viscosity

Damping
Model constants Function

where
One Equation Model: K-Model
Transport equation for k

Pk + Dk

Storage Convective Diffusive

transport of transport of k Rate of
k production
of k
Rate of
dissipation of
The production and k
dissipation terms are
suitably formulated

SMBS, VIT Chennai

One Equation Model: K-Model

Production of k Dissipation of k

 Turbulent kinetic  Large eddies break

energy increases up
 Energy goes from  Energy is transferred
mean flow to from eddies to mean
eddies flow
SMBS, VIT Chennai
One Equation Model: K-Model

 Eddy viscosity μt = ρ lm k

SMBS, VIT Chennai

Summary
 Only the effects of the turbulence on the mean flow
are usually sought
 Resolving the details of turbulent fluctuations is not
required for many engineering applications
 No single model is universally accepted for all
classes of problems
 DNS is not a turbulence model at all but it is the
most straightforward approach to the solution of
turbulent flows
 Boussinesq Approach relates Reynolds stress to
mean velocity gradients
Summary
 One-equation models based on an equation for the eddy
viscosity are complete such as Baldwin-Barth model and
Spalart-Allmaras model
 Spalart-Allmaras model can predicts better results than
Baldwin-Barth model, and much better results for
separated flow than Baldwin-Barth model and algebraic
models
End of Module-7