You are on page 1of 50

Workshop on

Computational Fluid Dynamics


(Turbulence Modeling)
By:
Dr. Alam Nawaz Khan Wardag
Department of Chemical Engineering (DCHE)
Pakistan Institute of Engineering and Applied Sciences (PIEAS),
Islamabad.
Tel: +92-51-1111 74327, 1111 PIEAS
Fax: +92-51-9248600
Email: alam@pieas.edu.pk

Definition of CFD
Computational Fluid Dynamics (CFD) is the simulation of fluids
engineering systems using
modeling (mathematical physical problem formulation) and
numerical methods (discretization methods, solvers, numerical
parameters, and grid generations, etc.).

Typical CFD Procedure


Geometry

Choice of
Physical
Models

Mesh
generation

Solving

Postprocessing

Select
geometry

Turbulence
models

Structured/
unstructured

Numerical
schemes

Result
reporting

Select
domain

Flow
properties

Mesh size

Convergence
criteria

Verification &
Validation

Boundary
conditions
Initial
conditions

Importance of CFD
Cost

Simulation (CFD)
Cheap

Experiment
Expensive

Time
Scale
Information
Repeatable
Safety

Short
Any
All
Yes
Yes

Long
Small/Medium
Selected points
Some
Some danger

Objective
To provide an overview of turbulence and its modeling
Mainly cover

What is turbulence?
Types of turbulent flows
How turbulence is generated
Length and time scales in turbulent flow
Prediction methods
Turbulence models
Selecting turbulence model for your application

Motivation
For the Study of Turbulent Flows
An important characteristic of turbulence is its ability to transport and mix fluid

much more effectively than a comparable laminar flow.

The effectiveness of turbulence for transporting and mixing fluids is of prime

importance in many applications.

Turbulence is also effective at 'mixing' the momentum of the fluid.

So the major motivations for the study of turbulent flows is:


The vast majority of flows is turbulent.
The transport and mixing of matter, momentum, and heat in flows is of great

practical importance.
And Turbulence greatly enhances the rates of these processes.

Examples of Turbulent Flows

Wake behind a circular cylinder

Motion of clouds

Wake behind a bullet

Characteristics of Turbulence
Turbulence is very difficult to define.

It is usual to describe turbulence by listing its features:

Turbulent flows are irregular.


Turbulent flows are diffusive.
Turbulent flows are rotational.
Turbulent flows occur at high Reynolds numbers.
Turbulent flows are dissipative.
Turbulence is a continuum phenomenon.
Turbulence is a feature of fluid flows, and not of fluids.
Turbulent flows are non-local.
8

Laminar vs Turbulent Flows


In laminar flow

Adjacent layers of fluid slide past each other without mixing.

Transfer of momentum occurs between layers moving at different speeds because of


viscous stresses.

In turbulent flow

Adjacent layers continually mix.


A net transfer of momentum occurs because of the mixing of fluid elements from layers
with different mean velocity.
This mixing is a far more effective means of transferring momentum than viscous
stresses.
Consequently, the mean velocity profile tends to be more uniform in turbulent flow.

Contd(Laminar vs Turbulent Flows)

10

Types of Turbulent Flows


Two types of turbulent flows:
1) Wall bounded turbulent flows
a) Turbulent boundary layer flows
b) Fully developed flows
2) Free Shear Flows

a) Jet
b) Mixing Layer
c) Wake

11

Free Shear Flows

1. Jet

2. Mixing Layer

3. Wake

Introduction to Transition

13

Physical Processes Involved in Turbulence


Unsteady
Convection
Production
Diffusion

Dissipation

14

Energy Cascade Mechanism


Turbulent flows are characterized by an infinite
number of time and length scales.
Turbulence can be considered to be composed of
eddies of different sizes
These sizes range from the flow length scale L to
the smallest eddies.
Each eddy of length size l has a characteristic
velocity u(l) and timescale t(l)=u(l)/l
The largest eddies have length scales comparable
to L
15

Contd(Energy Cascade Mechanism)


Each eddy has a Reynolds number
For large eddies, Re is large, i.e. viscous effects are negligible.
The idea is that the large eddies are unstable and break up
transferring energy to the smaller eddies.
The smaller eddies undergo the same process and so on
This energy cascade continues until the Reynolds number is
sufficiently small that energy is dissipated by viscous effects.

16

Kolmogorov Scales
The scales of turbulence can be easily established for smallest
eddies if one uses Kolmogorov universal equilibrium theorem
which states:
Rate of transfer of energy from larger eddies to
smaller eddies is approximately equal to the dissipation
of energy to heat by the smallest eddies

These scales are indicative of the smallest eddies present in the


flow, the scale at which the energy is dissipated.
17

Prediction Methods for Turbulent Flows


Direct Numerical Simulation (DNS)
Reynolds Averaged Navier Stokes Equations (RANS)
Large Eddy Simulation (LES)

Hybrid Method.

18

Direct Numerical Simulation (DNS)


In DNS, all the length and time scales are resolved.

We make direct use of the NS equations


DNS is highly informative regarding the physics of fluid flow

However, Computer cost (memory, CPU time, hardware)


increases with the Reynolds number
The grid should be as fine as possible
For 2D flows, the number of nodes is proportional to Re(3/4)
For a general 3D flow, the total number of nodes is proportional to Re(9/4)

Therefore DNS currently applied to simple flows.


19

Reynolds Averaged Navier-Stokes Equations (RANS)


Mathematically the instantaneous flow quantity can be written as

Where

Contd.(Reynolds Averaged Navier-Stokes Equations)


In order to carry out mathematical details, the following averaging
rules are applied:

Contd.(Reynolds Averaged Navier-Stokes Equations)


The continuity equation for 2D flow is

The velocity and density components are replaced by sum of mean and fluctuating
components of velocity and density respectively as:

Once this equation is expanded, it is time averaged according to the rules defined
earlier to provide

Contd.(Reynolds Averaged Navier-Stokes Equations)


The x component of the momentum equation is

The instantaneous values are replaced by time averaged mean and fluctuating
values to provide

Contd.(Reynolds Averaged Navier-Stokes Equations)


The entire equation is time averaged and rearranged as

Writing this equation in terms of shear stress one obtains

Contd.(Reynolds Averaged Navier-Stokes Equations)


The energy equation in terms of total energy per unit mass is expressed as

Mathematical procedure to include turbulence is similar to previous manipulation of


momentum equation and the final result for energy equation is

Contd.(Reynolds Averaged Navier-Stokes Equations)


The above equations can be written in the following form

Turbulent Shear Stress and Heat Flux


The governing equations for steady incompressible boundary layer are

It is customary to combine the laminar and turbulent shear stress terms in above
momentum equation as

For laminar flow one may write

Contd(Turbulent Shear Stress and Heat Flux)


In order to express turbulent shear in similar form Boussinesq approximation is
used, thus one writes

Hence momentum equation is expressed as

In similar fashion, turbulent conductivity is defined to combine laminar and


turbulent heat fluxes

In terms of turbulence parameters the energy equation is rearranged as

Contd(Turbulent Shear Stress and Heat Flux)


Prandtl Mixing length

The Prandtl hypothesis is expressed as

Similarly

In terms of turbulent viscosity and turbulent conductivity, the following may be


written

Turbulence Models
A turbulence model is a computational procedure to close the system of mean
flow equations.
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:
o must have wide applicability;
o be accurate;
o Simple;
o and economical to run.
30

Common Turbulence Models


Classical models. Based on Reynolds Averaged Navier-Stokes (RANS) equations
(time averaged):
1. Zero equation model
2. One equation model
3. Two equation models:
k- style models
standard k- model,
k- RNG model,
Realizable k- model
k- model.
4.Algebraic Stress model
5. Seven equation model: Reynolds stress model.

The number of equations denotes the number of additional PDEs that are being
solved.

31

Zero Equation Turbulence Models


The underlying assumption in zero equation models is that the local rate of production of turbulence
and the rate of dissipation of turbulence are approximately equal.
Generally most models employ an inner region/outer region formulation to
length.

represent mixing

A commonly used models utilizes an exponential function for inner region, whereas the outer region
is proportional to the boundary layer thickness. Mathematically they are expressed as

A turbulence model which is not written in terms of the boundary layer quantities was introduced by
Baldwin and Lomax. For this the inner region is approximated as

Contd..(Zero Equation Turbulence Models)


The nondimensional space coordinate

can be written as

or

The outer region is approximated by

Finally the turbulent viscosity distribution across the boundary layer is determined
from

Advantages
Mathematically simple
Cheap in terms of computing resources
Well established
Good predictions for thin shear layer,
jets, wakes, mixing layer and B.L

Disadvantages
Completely incapable of
describing flows with separation
and recirculation.

One Equation Models


One equation models employ a partial differential equation for velocity scale,
whereas the length scale is specified algebraically.
The velocity scale is typically written in terms of turbulent kinetic energy k defined
as

The turbulent viscosity is written as:


Economical and Accurate for:
Attached wall bounded flows
Flows with mild separation and
recirculation
Weak for:
Massively separated flows
Free shear flows

Two Equation Models


A commonly used two equation turbulence model is the k- model.
The standard k- two equation model expressed by the turbulent kinetic energy
equation

And the dissipation rate equation

The turbulent viscosity is related to by

Contd(Two Equation Models)


Advantages
Can handle complex flowfields which include massive separations.
Excellent performance for many industrially relevant flows.
well established: The most widely validated turbulence model.
Leads to stable calculations that converge relatively easily.
Disadvantages
More expensive to implement than zero and one equation models.
Poor prediction for swirling flows, axisymmetric jets, unconfined flows and fully
developed flows in non-circular ducts.

Contd(Two Equation Models)


RNG k- Model:
k- equations are derived from the application of a rigorous statistical
technique (Renormalization Group Method) to the instantaneous NavierStokes 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.

37

Contd(Two Equation Models)


Realizable k- Model:
Shares the same turbulent kinetic energy equation as the standard k- model.
Improved equation for .
Variable C 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.
Recirculation.
Strong streamline curvature

38

Reynolds Stress Model (RSM)


RSM closes the Reynolds-Averaged Navier-Stokes equations by solving additional
transport equations for the six independent Reynolds stresses.
Transport equations derived by Reynolds averaging the product of the
momentum equations with a fluctuating property.
Closure also requires one equation for turbulent dissipation.
Isotropic eddy viscosity assumption is avoided.
Resulting equations contain terms that need to be modeled.
RSM is good for accurately predicting complex flows.
Accounts for streamline curvature, swirl, rotation and high strain rates.
o Cyclone flows
o Rotating flow passages,
o Flows involving separation.
39

Comparison of RANS Turbulence Models


Model

Strengths

Weaknesses

SpalartAllmaras

Economical (1-eq.); good track


record for mildly complex B.L. type
of flows.

Not very widely tested yet; lack of submodels


(e.g. combustion, buoyancy).

STD k-

Robust, economical,
reasonably accurate; long
accumulated performance
data.

Mediocre results for complex flows with


severe pressure gradients, strong streamline
curvature, swirl and rotation. Predicts that
round jets spread 15% faster than planar jets
whereas in actuality they spread 15% slower.

RNG k-

Good for moderately complex


behavior like jet impingement,
separating flows, swirling
flows, and secondary flows.

Subjected to limitations due to isotropic eddy


viscosity assumption. Same problem with
round jets as standard k-.

Realizable
k-

Offers largely the same benefits as


RNG but also resolves the round-jet.

Subjected to limitations due to isotropic eddy


viscosity assumption

Reynolds
Stress
Model

Physically most complete


model (history, transport, and
anisotropy of turbulent
stresses are all accounted for).

Requires more CPU effort (2-3x); tightly


coupled momentum and turbulence
equations

40

Large Eddy Simulation (LES)


All the effort in a DNS is directed towards the resolution of small scales.

As 99% of the energy is contained outside the dissipation range (the smallest scales).
Therefore, one thinks of modelling these small scales that have a universal character
while fully resolving the larger scales: This is Large Eddy Simulation.
In LES, large (Grid) Scales (GS) are resolved and the small (Sub-Grid) Scales (SGS) are
modelled.
Subgrid model for the SGS turbulent scales

A filter is introduced that would act as an automation technique that tells the
equations what to fully resolve and what to model.
41

42

Example
Consider a two dimensional CFD case of the flow between two parallel plates to
demonstrate the laminar and turbulent nature of the fluid flow. Using CFD observe the
velocity and viscosity profiles in the fully developed region with a working fluid taken
as air for inlet velocities of 0.02m/s and 1m/s.

Contd..(Example)

Velocity profile

Viscosity profile

Turbulent K. E profile

Dissipation Rate profile

Guidelines for Computing Turbulent flows


Simulation of turbulent flows require decision based on
Flow physics
to characterize the flow features (turbulence, high gradients, etc.)

Accuracy requirement
to evaluate the grids resolution required for a certain accuracy

Resources requirement
to evaluate the need for sophisticated turbulence models

45

ContdGuidelines for Computing Turbulent flows


Modeling procedure:
Determine relevant Reynolds number to estimate if the flow is
turbulent
Select a turbulence model option and a near-wall treatment
Estimate the physical dimension of the first grid point off the wall
(y+)
Generate the grid
Select a suitable numerical scheme
Perform the simulation
Reality check (experiments, literature, model consistency, grid
resolution)
46

Guidelines for Selecting Turbulence Model


Some of the recommendations:
Identify flow features present in the problem under consideration.
E.g. Free shear layers, separated flows etc.

Take advantage of specialized models developed to capture specific flow


features.
In case of no idea, use SA model for external flows otherwise K-epsilon
model is better choice.
Apart from turbulence model, solution accuracy is also dependent on
quality of grid/mesh.

47

Contd..
Effect of Turbulence Models on Flow over an Airfoil

48

Contd..

49

Concluding Remarks
Major worry about turbulence modeling is that one cannot
estimate the errors of computer results
There is no better way to calculate turbulent flows unless we deal with very
low Re and simple geometries.
This may change in future with improvements in computer technology

One could use experimental data instead


Expensive and slow process
Has its own limitation

Another important issue is the identification of numerical versus modeling


errors

What is the answer then????


Compare information from various sources i.e. experimental,
numerical, theoretical and industrial
50