See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318562275

Turbulence Modeling - A Review

Technical Report · July 2019

DOI: 10.13140/RG.2.2.35857.33129/2

CITATIONS READS
0 4,056

1 author:

Ideen Sadrehaghighi
CFD Open Series
23 PUBLICATIONS   24 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Independent CFD Reasercher View project

Mesh Generation View project

All content following this page was uploaded by Ideen Sadrehaghighi on 05 July 2018.

The user has requested enhancement of the downloaded file.

 1

CFD Open Series

Revision 1.85.1

Turbulence Modeling
A Review

Ideen Sadrehaghighi, Ph.D.

Turbulence
contemporary abstaract
By Nancy Eckels

ANNAPOLIS, MD
 2

Contents

1 Introduction .................................................................................................................................. 6

2 Turbulence Essentials ............................................................................................................... 8

2.1 Physical Perspectives ........................................................................................................................................... 8
2.2 Components attributing to complexity of physics in Turbulence ...................................................... 8
2.2.1 Enhanced Diffusion ................................................................................................................................... 9
2.2.2 Stability Criteria ......................................................................................................................................... 9
2.2.3 Time Dependency ...................................................................................................................................... 9
2.2.4 Eddies and Spectral Length ................................................................................................................ 10
2.2.5 Non-Linearity Effects ............................................................................................................................ 10
2.2.6 Separation and Drag Reduction ........................................................................................................ 11
2.3 Transition ............................................................................................................................................................... 11

3 Reynolds Averaging and RANS Modeling (Linear Modeling) .................................... 14

3.1 Time Averaging .................................................................................................................................................... 14
3.1.1 Un-Steady RANS (URANS)................................................................................................................... 15
3.2 Reynolds Averaged Naiver-Stokes (RANS) .............................................................................................. 15
3.2.1 Comments on Reynolds Averaged Naiver-Stokes (RANS) ..................................................... 16
3.3 Closure Problem (Boussinesq Assumption) ............................................................................................ 17
3.4 Turbulence Modeling......................................................................................................................................... 17
3.4.1 Turbulent Viscosity, Velocity, and Length Scale ........................................................................ 18
3.4.1.1 Eddy Viscosity (RANS) Models ................................................................................. 19
3.4.1.2 Algebraic Model (Zero Equation) - Mixing Length .................................................. 19
3.4.1.3 Baldwin-Lomax ........................................................................................................ 20
3.4.2 One-Equation Model .............................................................................................................................. 20
3.4.2.1 K Model ................................................................................................................... 21
3.4.2.2 Spallart - Allmaras.................................................................................................... 21
3.4.3 Two Equation Model ............................................................................................................................. 21
3.4.3.1 Standard Transport Equation for k .......................................................................... 22
3.4.3.2 Standard Transport Equation for ε .......................................................................... 23
3.4.3.3 The RNG Model ....................................................................................................... 24
3.4.3.4 The Realizable κ-ε Model ........................................................................................ 24
3.4.3.5 Issues Relating to κ-ε Turbulence Modeling............................................................ 24
3.4.3.6 The κ-ω and SST Models .......................................................................................... 25
3.4.3.7 Wilcox (Standard) κ-ω Model .................................................................................. 25
3.4.3.8 The Baseline (BSL) κ-ω Model ................................................................................. 26
3.4.3.9 SST κ-ω Model ......................................................................................................... 26
3.4.4 Case Study – Subsonic Flow for Turbulence Models of (NACA)-0012 Airfoil................ 28
3.4.4.1 Computational Method ........................................................................................... 28
3.4.4.2 Results and Discussion ............................................................................................ 29
3.4.4.3 Conclusions.............................................................................................................. 32
3.4.5 Near Wall Turbulence Modeling ....................................................................................................... 32
3.4.5.1 Modeling Flow Near the Wall .................................................................................. 33
3.4.5.2 Case Study - Wall-modelling strategies in Large Eddy Simulation of separated high-
Reynolds-number flows ............................................................................................................... 34
 3

4 Beyond the Boussinesq Approximation (Non-Linear Modeling) ............................. 36

4.1 Boussinesq-Approximation Deficiencies .................................................................................................. 36
4.2 Nonlinear Constitutive Relations ................................................................................................................. 36
4.2.1 Case Study – 3D Simulation of Flow Past a Cylinder using Nonlinear Turbulence
Model 37
4.2.1.1 Literature Survey ..................................................................................................... 38
4.2.1.2 Basic Equations ........................................................................................................ 38
4.2.1.3 Numerical Strategies ............................................................................................... 39
4.2.1.4 Geometry and Meshing ........................................................................................... 40
4.2.1.5 Results and Discussion ............................................................................................ 40
4.2.1.6 Conclusions.............................................................................................................. 42
4.3 2 Order Closures (Reynolds Stress Models - RSM) ............................................................................ 42
nd

4.4 LES Model ............................................................................................................................................................... 42

4.4.1 Filter Definition ....................................................................................................................................... 43
4.4.2 Filtered Incompressible N-S Equations ......................................................................................... 44
4.4.3 Numerical Methods for LES ................................................................................................................ 45
4.4.4 Implicit vs Explicit Filtering ............................................................................................................... 45
4.4.5 Quantitative Aspects of Comparison of RANS vs LES Models .............................................. 46
4.4.6 Motivations for Coupling Methods Between RANS and LES ................................................. 47
4.4.6.1 Principal Approaches to Coupling LES with RANS ................................................... 48
4.4.6.2 Results and Discussion for Segregated Modeling ................................................... 49
4.4.6.3 Coupling in RANS/LES Interface............................................................................... 49
4.5 DES Model .............................................................................................................................................................. 50
4.6 DNS Model .............................................................................................................................................................. 52
4.6.1 The Reynolds Number Constraint ................................................................................................... 53
4.6.2 Numerical Considerations ................................................................................................................... 54
4.6.2.1 Spectral Methods .................................................................................................... 54
4.6.2.2 Finite Differencing ................................................................................................... 55
4.7 Strategies for Turbulence Modelling........................................................................................................... 55
4.7.1 Physical Aspects (modification to simple RANS Models)....................................................... 56
4.7.2 Complex RANS Models.......................................................................................................................... 56
4.7.3 Role of Grid Refinement ....................................................................................................................... 57
4.7.4 Outlook ........................................................................................................................................................ 58
4.8 Modeling To Choose From............................................................................................................................... 59

5 Classification of Turbulence Models .................................................................................. 60

5.1 Comparisons of various Turbulence Models ........................................................................................... 61
5.2 Numerical Considerations ............................................................................................................................... 63
5.2.1 Elementary Time-Marching Methods............................................................................................. 63
5.2.1.1 Block-Implicit Methods ........................................................................................... 63

List of Tables:
Table 3.1 Constant values for κ-ε turbulence model equations ............................................................... 23
Table 3.2 Constant values for SST κ-ω turbulence model equations..................................................... 27
Table 4.1 Summery of Turbulence Strategies ................................................................................................. 58
Table 5.1 Advantages and Disadvantages of different Turbulence Models ........................................ 62
 4

List of Figures:
Figure 2.1 Flow Pattern Pass a Circle..................................................................................................................... 8
Figure 2.2 Large Eddies in a Turbulent Boundary Layer ............................................................................ 10
Figure 2.3 Velocity Profile for Laminar vs Turbulent Flows ..................................................................... 11
Figure 2.4 Turbulence Transition Region ......................................................................................................... 11
Figure 2.5 Snapshots of the Stream-Wise Velocity Component for (A) Adiabatic Wall , (B)
Heated Wall, and (C) Cooled Wall ............................................................................................................................. 12
Figure 3.1 Relationship between averaged variables; a) Steady flow b) Un-Steady flow ...... 15
Figure 3.2 Hierarchy of Turbulence Models..................................................................................................... 18
Figure 3.3 Lift Coefficient at Stall (AoA) against Number of Cells........................................................... 28
Figure 3.4 Comparison between experimental data [Abbott et al.]and three different turbulent
models simulation results of the Lift Coefficient for NACA 0012 Airfoil .................................................. 29
Figure 3.5 Comparison Between Experimental Data for Transitional Boundary Layer and
Different Turbulent Models on the Drag Coefficient of NACA-0012 Airfoil ........................................... 30
Figure 3.6 Contours of velocity magnitude at 9° (Top) and 16° (Bottom) AoA with the Spalart-
Allmaras turbulence model ......................................................................................................................................... 31
Figure 3.7 Zones in Turbulent B. L. for a typical Incompressible flow over a smooth flat plate 33
Figure 3.8 Turbulence and near Wall Function .............................................................................................. 34
Figure 3.9 Schematics of hybrid LES-RANS scheme (upper) and two-layer zonal scheme
(lower) ................................................................................................................................................................................. 35
Figure 4.1 Schematic Representation of the Computational Domain ................................................... 40
Figure 4.3 Comparison of Wall Pressure Coefficient (Cp) .......................................................................... 41
Figure 4.2 Streamline Patterns at Four Different Time Intervals for one Vortex Shedding
Cycle ...................................................................................................................................................................................... 41
Figure 4.4 Variation of lift Coefficient with Time ........................................................................................... 41
Figure 4.5 Difference Between the Filtered Velocity and the Instantaneous Velocity .................. 42
Figure 4.6 A velocity field produced by a (LES) of homogeneous decaying turbulence................ 43
Figure 4.7 Vorticity Prompted by the Wake Passing Cycle ........................................................................ 46
Figure 4.8 Time Averaged (RANS) vs Instantaneous (DES) Simulation over a backup step ....... 47
Figure 4.9 Integrated RANS-LES Computations in Gas Turbines: Compressor-Diffuser, ............. 49
Figure 4.11 Compressor and combustor: RANS and LES axial velocity, mid-passage (Courtesy
of Medic et al.) ................................................................................................................................................................... 50
Figure 4.10 DES and effect of Grid Density on the Wake flow ................................................................. 51
Figure 4.12 Comparison of Pressure contours on planner cuts for RANS and DES models ........ 52
Figure 4.13 Vorticity contours from spectral DNS at two Reynolds numbers (Spalart et al.
2008). ................................................................................................................................................................................... 53
Figure 4.14 The sound radiated by a Mach 1.9 circular jet ........................................................................ 54
Figure 4.15 Instantaneous Contours of Stream-Wise Density Gradients from DNS ....................... 55
Figure 4.16 Stream Lines in Square Channel with Nonlinear Constitutive –( Courtesy of
[speziale, 1987]) .............................................................................................................................................................. 56
Figure 4.17 Simulation of Flow Past Circular Cylinder by Various Approaches (Shur et al.,
1996; Travin et al., 2000) ............................................................................................................................................. 57
5
 6

1 Introduction

Turbulence has been the victim of many colorful descriptions over the years, from Lamb’s (1916)
scholarly “chief outstanding difficulty of our subject” to Bradshaw’s (1994) inspired “invention of the
Devil on the seventh day of creation.” This apparent frustration results largely from the mixture of
chaos and order and the wide range of length and time scales that turbulent flows possess1. The three
key elements of CFD are algorithm development, grid generation and turbulence modelling.
Turbulence is inherently three-dimensional and time dependent, and an enormous amount of
information is thus required to completely describe a Introduction Turbulent Flow. This is beyond the
capability of the existing computers for virtually all practical flows. Thus, some kind of approximate
and statistical method, called a turbulence model, is needed.
Complexity of different turbulence models may vary strongly depends on the details one wants to
observe and investigate by carrying out such numerical simulations. N-S equation is inherently
nonlinear, time-dependent, three-dimensional PDE. Turbulence could be thought of as instability
of laminar flow that occurs at high Reynolds numbers. Such instabilities origin form interactions
between non-linear inertial terms and viscous terms in N-S equation. These interactions are
rotational, fully time-dependent and fully three-dimensional. Rotational and three-dimensional
interactions are mutually connected via vortex stretching. Vortex stretching is not possible in two
dimensional space. That is also why no satisfactory two-dimensional approximations for turbulent
phenomena are available. Furthermore turbulence is thought of as random process in time.
Therefore no deterministic approach is possible. Certain properties could be learned about
turbulence using statistical methods. These introduce certain correlation functions among flow
variables. However it is impossible to determine these correlations in advance2.
Another important feature of a turbulent flow is that vortex structures move along the flow. Their
lifetime is usually very long. Hence certain turbulent quantities cannot be specified as local. This
simply means that upstream history of the flow is also important of great importance. In short, the
turbulence is severely restriction the calculation of CFD in aerospace design process where the
inability to reliably predict turbulent flows with significant regions of separation3.
Presently, turbulence modelling based on Reynolds-Averaged Navier Stokes (RANS) equations is the
most common and practical approach for turbulence simulation. RANS are time-averaged
modification of Navier-Stokes equations and turbulence models are semi-empirical mathematical
relations that are used to predict the general effect of turbulence. The objective of turbulence
modelling is to develop equations that will predict the time-averaged velocity, pressure, and
temperature fields without calculating the complete turbulent flow pattern as a function of time.
Unfortunately, there is no single universally accepted turbulence model that works for all flows and
all regimes. Therefore, users have to use engineering judgement to choose from a number of different
alternatives sine the accuracy and effectiveness of each model varies depending on the application.

1 P., Moin and K., Mahesh, “Direct Numerical Simulation: A Tool in Turbulence Research”, Annual Rev. Fluid Mech.
1998. 30:539–78.
2 J., SODJA, “Turbulence models in CFD”, University of Ljubljana, Faculty for mathematics and physics,

Department of physics, March 2007.

3 Dimitri Mavriplis, “Exa-scale Opportunities for Aerospace Engineering“, Department of Mechanical Engineering

University of Wyoming and the Vision CFD2030 Team.

7
 8

2 Turbulence Essentials
2.1 Physical Perspectives
While the fluid elements are smooth and regular for Laminar flow pass a circle as shown in Figure
2.1 (a), the Turbulent elements are irregular in which various quantities show a random variation
with time and space as in Figure 2.1 (b). Therefore, a statically distinct averaging of values can be
distinguished. Also, because of this agitated motion in turbulent flow, the higher energy fluid
elements from the outer region of the flow are pumped close to surface. The diffusion rate of a scalar
quantity is usually greater in a turbulent flow than in a laminar. As the result, the frictional effects
are more severe for a turbulent flow. An outstanding feature of turbulent flow, as opposite to laminar
flow, is that molecules move in chaotic fashion along complex irregular path. The strong chaotic motion
causes the various layers of fluid to mix together intensively. Because of increase momentum and
energy exchange between the molecules and solid walls, turbulent flow leads at some conditions to
higher skin friction and heat transfer as compared to laminar case. It could be argued as that because
of the agitated motion in turbulent flow, the higher energy fluid elements from the outer regions of
flow are pumped close to surface, and hence, average flow velocity near solid surface is larger for
turbulent flow in comparison to laminar as depicted in Figure 2.3.

(a) Laminar flow (b) Turbulent flow

Figure 2.1 Flow Pattern Pass a Circle

2.2 Components attributing to complexity of physics in Turbulence

Before plunging into the mathematics of turbulence, it is worthwhile to first discuss physical aspects
of the phenomenon. It is a rather interesting discussion which encompasses most of the physical flow.
It is intended for interest parties with full text is available in book by4. The following discussion is not
intended as a complete description of this complex topic. Rather, we focus upon a few features of
interest in engineering applications, and in construction of a mathematical model. In 1937, Taylor
and von Karman proposed the following definition of turbulence: "Turbulence is an irregular
motion which in general makes its appearance in fluids, gaseous or liquid, when they flow past
solid surfaces (boundary conditions) or even when neighboring streams of the same fluid flow
past or over one another (i,e. Jet flow)." It is characterized by the presence of a large range of

4 David C. Wilcox, ”Turbulence Modeling for CFD”, 1993, 1994 by DCW Industries, Inc.
 9

excited length and time scales. The irregular nature of turbulence stands in contrast to laminar
motion, because the fluid was imagined to flow in smooth laminae, or layers. Virtually all flows of
practical engineering interest are turbulent. Turbulent flows always occur when the Reynolds
number is large. For slightly viscous fluids such as water and air, large Reynolds number corresponds
to anything stronger than a small swirl or a puff of wind. Careful analysis of solutions to the Navier-
Stokes equation, or more typically to its boundary-layer form, show that turbulence develops as an
instability of laminar flow. The features contributing to complexity of turbulence flow are;

 Enhance Diffusion
 Stability Criteria
 Time Dependency
 Eddies and Spectral Length
 Non-Linearity effects
 Separation

Enhanced Diffusion
Perhaps the most important feature of turbulence from an engineering point of view is its enhanced
diffusivity. Turbulent diffusion greatly enhances the transfer of mass, momentum and energy.
Apparent stresses often develop in turbulent flows that are several orders of magnitude larger than
in corresponding laminar flows.

Stability Criteria
To analyze the stability of laminar flows, virtually all methods begin by linearizing the equations of
motion. Although some degree of success can be achieved in predicting the onset of instabilities that
ultimately lead to turbulence with linear theories, the inherent nonlinearity of the Navier-Stokes
equation precludes a complete analytical description of the actual transition process, let alone the
fully-turbulent state. For a real (i.e., viscous) fluid, the instabilities result from interaction between
the Navier-Stokes equation's nonlinear inertial terms and viscous terms. The interaction is very
complex because it is rotational, fully 3-D and time dependent. The strongly rotational nature of
turbulence goes hand-in-hand with its three dimensionality. Vigorous stretching of vortex lines is
required to maintain the ever-present fluctuating vorticity in a turbulent flow. Vortex stretching is
absent in two-dimensional flows so that turbulence must be three dimensional. This inherent 3-D
means there are no satisfactory 2-D approximations and this is one of the reasons turbulence remains
the most noteworthy unsolved scientific problem of the twentieth century.

Time Dependency
The time-dependent nature of turbulence also contributes to its intractability. The additional
complexity goes beyond the introduction of an additional dimension. Turbulence is characterized by
random fluctuations thus obviating a deterministic approach to the problem. Rather, we must use
statistical methods. On the one hand, this aspect is not really a problem from the engineer's view.
Even if we had a complete time history of a turbulent flow, we would usually integrate the flow
properties of interest over time to extract time-averages. On the other hand, time averaging
operations lead to statistical correlations in the equations of motion that cannot be determined a
priori. This is the classical closure problem, which is the primary focus of this text. In principle, the
time-dependent, three-dimensional Navier-Stokes equation contains all of the physics of a given
turbulent flow. That this is true follows from the fact that turbulence is a continuum phenomenon. As
noted, "Even the smallest scales occurring in a turbulent flow are ordinarily far larger than any
molecular length scale”. Nevertheless, the smallest scales of turbulence are still extremely small.
 10

They are generally many orders of magnitude smaller than the largest scales of turbulence, the latter
being of the same order of magnitude as the dimension of the object about which the fluid is flowing.
Furthermore, the ratio of smallest to largest scales decreases rapidly as the Reynolds number
increases. To make an accurate numerical simulation (i.e., a full time dependent 3-D solution) of a
turbulent flow, all physically relevant scales must be resolved. While more and more progress is
being made with such simulations, computers of the early 1990's have insufficient memory and speed
to solve any turbulent flow problem of practical interest.

Eddies and Spectral Length

Turbulence consists of a continuous spectrum of scales ranging from largest to smallest, as opposed
to a discrete set of scales. In order to visualize a turbulent flow with a spectrum of scales we often
refer to turbulent eddies. A turbulent eddy can be thought of as a local swirling motion whose
characteristic dimension is the local
turbulence scale. Figure 2.2 shows as the
flow above the boundary layer has a steady
velocity U; the eddies move at randomly
fluctuating velocities of the order of a tenth
of U. The largest eddy size (l) is comparable
to the boundary-layer thickness (δ). The
interface and the flow above the boundary
is quite sharp [Corrsin and Kistler (1954)].
Eddies overlap in space, large ones carrying
smaller ones. Turbulence features a
cascading process whereby, as the Figure 2.2 Large Eddies in a Turbulent Boundary Layer
turbulence decays, its kinetic energy
transfers from larger eddies to smaller eddy. Ultimately, the smallest eddies dissipate into heat
through the action of molecular viscosity. Thus, we observe that turbulent flows are always
dissipative.
An especially striking feature of a turbulent shear flow is the way large bodies of fluid migrate across
the flow, carrying smaller-scale disturbances with them. The arrival of these large eddies near the
interface between the turbulent region and non-turbulent fluid distorts the interface into a highly
convoluted shape. In addition to migrating across the flow, they have a lifetime so long that they
persist for distances as much as 30 times the width of the flow [Bradshaw (1972)]. Hence, the
turbulent stresses at a given position depend upon upstream history and cannot be uniquely
specified in terms of the local strain-rate tensor as in laminar flow.

Non-Linearity Effects
The nonlinearity of the Navier-Stokes equation leads to interactions between fluctuations of differing
wavelengths and directions. As discussed above, the wavelengths of the motion usually extend all the
way from a maximum comparable to the width of the flow to a minimum fixed by viscous dissipation
of energy. The main physical process that spreads the motion over a wide range of wavelengths is
vortex stretching. The turbulence gains energy if the vortex elements are primarily oriented in a
direction in which the mean velocity gradients can stretch them. Most importantly, wavelengths that
are not too small compared to the mean-flow width interact most strongly with the mean flow.
Consequently, the larger-scale turbulent motion carries most of the energy and is mainly responsible
for the enhanced diffusivity and attending stresses. In turn, the larger eddies randomly stretch the
vortex elements that comprise the smaller eddies, cascading energy to them5.

5 David C. Wilcox, “Turbulence Modeling for CFD”, Copyright © 1993, 1994 by DCW Industries.
 11

Separation and Drag Reduction

Although the frictional effects are more severe in
turbulent flow, both shear stress and aerodynamic
heating are larger for the turbulent flow in comparison
to laminar6. However, turbulent flow has a major
redeeming value; because the energy of the fluid
elements close to surface is larger, it does not
separates from the surface as readily as a laminar flow.
Simply, if a flow is turbulent, it is less likely to separate
from the body surface. And if flow separates, the
separation region is smaller than those for laminar. As
the result the pressure drag Dp will be smaller in
turbulent flow. This leads to great compromise in
aerodynamics. In general, if the body is slender, the
friction drag, Df is much greater than Dp. For this case
since Df is smaller in laminar flow (lower velocity →
lower viscous drag; see Figure 2.3), then laminar flow
is desirable for slender bodies (airfoils). In contrast, for
a blunt body, Dp is much greater than Df. For this case, Figure 2.3 Velocity Profile for Laminar vs
Turbulent Flows
since Dp is smaller in turbulent flow, then turbulent
flow is desirable (gulf balls).

2.3 Transition
The transition from laminar to turbulent is not subtle and consists of several processes, each subject
to intense research interests on their own right. The transition and stability of the laminar flow seem
to be dependent on a critical value of Reynolds number (Figure 2.4) as

ρ  V x cr
Re cr   2100 Eq. 2.1
μ

Figure 2.4 Turbulence Transition Region

6 Anderson, John D. 1984: “Fundamentals of Aerodynamics”, McGraw Hills Inc.

 12

The equations governing a turbulent flow are precisely the same as for laminar flow; however,
solutions is clearly more complicated in this region. That is due to the introduction of new terms and
issues with closure. Two general approach could be envisioned. First a more direct approach with
extreme spatial discretization to capture all the flow eddies near the wall. This is prohibitly CPU
intensive, even with current computing powers. To counter that, the use of some modeling (empirical
or otherwise) in the vicinity of wall region is advocated. Rest of this report organized as subsequent.
Since subject of Turbulence is very involved, therefore, the physical aspects of formulation, namely
Reynolds Stress formulation, and how to derive them represented first. Followed by different aspects
of Turbulence modeling.
The numerous factors contributing to transition from laminar to turbulent flow in a fluid. The intent
has been to provide general background information on the various transition phenomena rather
than to make a study of the problem in depth. Included are the effects on transition of such factors as
pressure gradient, surface temperature, Mach number, and two- and three-dimensional types of
surface roughness [A. L, Braslow]7. The effect of wall heat transfer was investigated by [Shadloo and
Hadjadj]8 through high-resolution DNS , among other factors. As in the case of adiabatic wall
(Figure 2.5-a), stream-wise-elongated streaks are visible in the laminar and transitional regions for
both heated (Figure 2.5-b) and cooled cases (Figure 2.5-c). However, they locally break down and
create turbulent spots further downstream for the heated case when compared with the adiabatic
wall. Therefore, the wall heating stabilizes the flow and postpones the transition.

(A) Adiadatic Wall

(No Heating)

(B) Tw/Tr = 1.5

(Heated)

(C) Tw/Tr = 0.75

(Cooled)

Figure 2.5 Snapshots of the Stream-Wise Velocity Component for (A) Adiabatic Wall , (B) Heated
Wall, and (C) Cooled Wall

7 Albert L, Braslow, “A Review Of Factors Affecting Boundary-Layer Transition”, NASA TN D-3384.

8 M. S. Shadloo & A. Hadjadj, “Laminar-turbulent transition in supersonic boundary layers with surface heat
transfer: A numerical study”, Numerical Heat Transfer, 2017.
13
 14

3 Reynolds Averaging and RANS Modeling (Linear Modeling)

3.1 Time Averaging

Since Most of engineers does not concern with instantaneous values but rather deal with averages.
Therefore, the Reynolds Averaged N-S equations (RANS) are derived by decomposing the dependent
variables in conservative equations into time mean and fluctuating components. Two types of
averaging are greatly used, the classical Time (Reynolds) average, and the Mass Weighted average.
For flows in which density fluctuating could be neglected, the two formulations becomes identical.
The time averaging procedure defined by:

t 0  Δt
1
f
Δt  f dt
t0
and f   0 , u i  u i  ui , ρ  ρ  ρ , p  p  p , T  T  T Eq. 3.1

It is required that Δt be large compared to the period of random fluctuations associated with
turbulence, but small with respect to the time constant for any slow variations in flow field associated
with ordinary unsteady flows. In conventional Reynolds decomposition, the randomly changing flow
variations are replaced by the time average plus fluctuations (see Figure 3.1) and would be discuss
later for Unsteady RANS (URANS) formulations. The resulting of decomposition of variables and time
averaging the entire equation, the continuity, momentum, and energy yield to

ρ 
M ass :  (ρ u j  ρu j )
t x j
 
M omentum : (ρ u i  ρu i )  (ρ u i u j  u i ρu j ) 
t x j
p 
  ( τ  u ρu   ρ u  u   ρu  u  )
x x ij j i i j i j
i j
 (c ρT  c ρ T)
p p   
Energy :   ρc p u j T  
t x  
j
p p p    T 
 uj  u j  k  ρc p Tu j  c p ρTu j   
t x j x j x j  x j 

ui u 
where   τij  τij i
x j x j
 u u j  2 u k  1 i j
τij  μ  i    δ ij δ ij  
 3 x 
and
 
 jx x 0 i j
i  k 

Eq. 3.2
 15

Figure 3.1 Relationship between averaged variables; a) Steady flow b) Un-Steady flow

Un-Steady RANS (URANS)

For unsteady analysis, the Unsteady RANS could be used where of course the unsteady term is
retained in averaging as well. In essence, there are now two time scale involved where the Reynolds
decomposition of U would be as before an average and fluctuated value as


1
U
  U( ) d
0
, U  U  u  Eq. 3.3

Note that the dependent variables are now not only a function of space but also a function of time as
well. Be advised that this concept of time, is different from the time step (t) of mean value. In essence
if you solve your equations with one global time step, which is used in every cell, and if value of the
time step is small enough then you will be able to capture fluctuations, or unsteady behavior in the
MEAN quantities. In other words your solution is time accurate. Steady RANS, or RANS, marches the
solution with a local optimized time step for each cell, and hence is not time accurate, you will get a
faster solution, and it will be steady state. For URANS we have:

Ui  Ui (x, y,z, ) and uiuj  uiuj (x, y,z, ) Eq. 3.4

Even if the results from URANS are unsteady, one is often interested only in time averaged flow as
denoted as <Ū>, which means that we can decompose the results from URANS as a time averaged
part <Ū>, a resolved fluctuation uˈ, and the modeled turbulent fluctuation, u̎ , i.e.

U  U  u   U  u  u Eq. 3.5

3.2 Reynolds Averaged Naiver-Stokes (RANS)

The Reynolds Averaged Navier-Stokes equations (RANS) are obtained from the continuity and
momentum equations by taking the time average of all the terms in the equations. The continuity
equation does not change since it is linear in terms of the velocity. However, the momentum equation
is non-linear, which means that all the fluctuating components do not vanish. An extra term, called
Reynolds stress uiuj, appears in the momentum equation. The result can be written where the
 16

overbar has been dropped from the mean values. This convention will be used throughout here for
obvious reasons.

Closure
Problem

Eq. 3.6
Eq. 3.6 presents the fundamental problem of turbulence. In order to compute all the mean-flow
properties of the turbulent flow we need a reasonably accurate way to compute the Reynolds stress
uʹiuʹj. This is the fundamental reason for the need of the turbulence models. The complete set of the
Reynolds Averaged Navier-Stokes (RANS) equations are not presented here because they can be
found from many references9. Scalar transport equations are also needed, for example to describe
the transport of the concentration of species or the mass fraction of species. Their exact formulation
can be found in10. For incompressible flow with constant properties and no body force the
momentum and energy

Comments on Reynolds Averaged Naiver-Stokes (RANS)

At the first glance the Reynolds equations seemed to be quite complicated to solve in turbulent flow.
Certainly a major problem in fluid mechanic is that they are more equation can be written than
solved. Fortunately, for many important flows, the Reynolds equations can be simplified. But before
turning into the task of simplification, it is prudent to examine the equation themselves. By
considering the incompressible flow, the momentum equation can be written as:

Du̅i ∂p̅ ∂(τ̅ij )Lam ∂(τ̅ij )Turb

ρ = − + +
⏟
Dt ∂x
⏟i ⏟ ∂xj ⏟ ∂xj
Particle Acceleration mean Pressure Laminar like stress Apperant Stree Gradient
of mean motion Gradient Gradient due to transport of momentum
by turbulent fluctuation
where (τ̅ij ) ̅̅̅̅̅
= −ρu ′ u′
i j Similarly for Energy Equation:
Turb

9 P. Kaurinkoski and A. Hellsten, FINFLO: The Parallel Multi-Block Flow Solver, Report A-17, Laboratory of
Aerodynamics, Helsinki University of Technology, Espoo, Finland, 1998.
10 P. Kaurinkoski, “Development of an Equation of State for an Arbitrary Mixture of Thermally Perfect Gases to the

FINFLO flow solver”, Report No B-48, Series B, Laboratory of Aerodynamics, Helsinki University of Technology,
Finland, 1995.
 17

∂ ̅
∂T ∂
−(∇. q)Lam = (k ) and − (∇. q)Turb = (ρC𝑝 ̅̅̅̅̅̅
T ′ 𝑢𝑗′ )
∂xj ∂xj ∂xj
Eq. 3.7
The Reynolds equation cannot be solved in the form given because the new apparent turbulent
stresses and heat-flux quantities must be viewed as new unknowns. To proceed further, we need to
find additional equations involving the new unknowns to make assumption regarding the
relationship between the new apparent turbulence quantities and time-mean flow variables. This is
known as closure problem which is most commonly handled through turbulence modeling to be
discussed next.

3.3 Closure Problem (Boussinesq Assumption)

A linear relation between turbulent shearing stresses and rate of mean strain suggested by
[Boussinesq] which is the fundamental closure for most eddy viscosity models been used.

 u i u j  2  u k 
 ρui uj  μ T     δij  μ T
  ρ κ  κ  ui ui /2 Eq. 3.8
 x x  3
 j i   x k 

Where μT is the turbulent viscosity, κ is the kinematic energy of turbulence are key modeling aspects
associated with so called turbulence closure problem. Reynolds stresses with deformation rate, and
are related to viscosity, mean rate of deformation, and turbulent kinetic energy with Boussinesq’s
proposal expressed can be used to calculate Reynold’s stresses in the final step of turbulence
modelling. It is seen from this equation that the Reynold’s stresses are considered proportional to
the dissipation rate reduced by the eddy turbulent kinetic energy. (The Kronecker delta ensures that
the normal Reynolds stresses are each appropriately accounted for). It can also be seen that the
kinetic energy allocates an equal third for each normal stress component (isotropic assumption).
This is the reason for the inherent inaccuracy of the κ-ε model, making it incapable of describing
anisotropic flow. Other scalar flow properties such as mass and heat can also be modelled using time-
averaged values.
To obtain mathematical closure, the Reynolds stress terms must be related to mean flow properties
either empirically or through flow model which allows calculation of this relationship. References
are also sometimes made to the order of the closure. According to this terminology, a 1st order
closure evaluates the Reynolds stresses through functions of the mean velocity and geometry alone,
as the case above (linear). A 2nd order closure employs a solution to a modeled form of transport
partial differential equations for one or more of the characteristics of turbulence (non-linear). More
complicated quadratic or cubic closures are also available through literature.

3.4 Turbulence Modeling

Turbulence models to close the Reynolds equations can be divided originally into three groups in
general. First models which (directly) use the Boussinesq assumption. Most models currently
employed in engineering are this type. Experimental evidence indicates this valid in many
circumstances. Second are models using the effect of closure to the Reynolds equation without this
assumption. The third category is defined as those that are not based entirely on the Reynolds
equation. An example would be Large Eddy Simulation (LES) where an attempt is made to resolve
the large scale turbulent equation from first principals be numerically solving the filtered set of
equations governing this large scale. Turbulence modeling is used to approximate the effects of the
sub-grid scale turbulence. Although such a calculation have shown great promises, it is
 18

computationally prohibited to be considered as engineer in tool11. There are other references to

turbulent modeling such as order of closure. According to this terminology, a first order closure
evaluates the Reynolds stresses through functions of the mean velocity and geometry alone. A second
order closure employs a solution to a modeled from transport partial differential equation for one or
more of the characteristics of turbulence. Since this straighter forward, we try to adapt this category
as depicted in Figure 3.2. These are some of models mentioned here and by no means is it exclusive.
An ideal model should be with minimum amount of complexity while capturing the essence of the
relevant physics.

Mixing Length
0-Eqaution
Baldwin-Lomax

Spalart-
1-Equation
1st - order Almars Standard
(linear)

κ-ε models RNG

Reynolds Realizable
Average 2-Equation
(RANS) SST
Reynolds
Turbulence Stress κ-ω models
DES 2nd - order Standard
Models
(non-linear) (Wicox)
Algebric Stress
DNS

LES Smagorinsky Quadratic & Cubic κ-ε

Figure 3.2 Hierarchy of Turbulence Models

Turbulent Viscosity, Velocity, and Length Scale

Before we start the derivation for κ-ε and other models, there are some common ground that we
should cover. As viscosity is the central concept for modelling the stresses, it should be noted that the
turbulent viscosity μT is expressed as:

μ T  ρu T l where uT 
2
3
 
1
3

ui ui  Eq. 3.9

It can be seen that the turbulent viscosity is a product of density and two new variables representing
turbulent velocity and turbulent length scale. Turbulent velocity uT can be described as the typical
velocity occurring in the largest eddies and can also be related to the same eddies. Turbulent kinetic
energy according to turbulent length scale is the average length. The new variables, uT and l form the

11Anderson, Dale A; Tannehill, John C; Plecher Richard H; 1984,”Computational Fluid Mechanics and Heat
Transfer”, Hemisphere Publishing Corporation.
 19

basis for the “two-equation‟ k-ε turbulence model, meaning that in addition to the RANS equations,
two more equations are required to solve for turbulent velocity and turbulent length using the model.
The length scale l (for large eddies) is used in the k-ε model to define the length scale ε (for small
eddies), for which a transport equation is used in the model, and represents the dissipation of the
turbulent kinetic energy. The dissipation is expressed as:

κ 2/3 μ T  ui  u j
ε as related to viscosity ε   Eq. 3.10
l ρ x κ x κ

As seen in the second expression for ε, dissipation of the turbulent kinetic energy κ is proportional to
the rate of deformation of eddies. Other scalar flow properties such as mass and heat can also be
modelled using time-averaged values. Similar to the turbulent momentum transport’s
proportionality to average velocity gradients, turbulent scalar transport is proportional to mean
scalar value gradients and can be expressed as


 ρu i   Γ T (9.16) Eq. 3.11
x i

Where ГT refers to turbulent (eddy) diffusivity. As can be seen from above expression, the turbulent
scalar property transport occurs with the same mechanism as in transport of momentum (mixing of
eddies, represented by ГT). For this reason, it can be assumed by Reynold’s analogy that the value of
ГT is similar to μT, the turbulent viscosity. The ratio of μT to ГT is defined as the Prandtl/Schmidt
number σT, and has a value which is normally constant with a value around unity. The next section
presents the κ-ε model and the two extra k and ε transport equations (PDEs) for closing the system
of time-averaged RANS equations. The model is based on the mechanisms causing changes to
turbulent kinetic energy (i.e. turbulent viscosity and velocity fluctuations).

3.4.1.1 Eddy Viscosity (RANS) Models

Van Driest (1956) devised a viscous damping correction for the mixing-length model. This correction
is still in use in most modern turbulence models. [Cebeci & Smith] refined the eddy viscosity/mixing-
length concept for better use with attached boundary layers12.

3.4.1.2 Algebraic Model (Zero Equation) - Mixing Length

This is invariably the successful this type model which uses the Boussinesq assumption. It was
originally suggested by Prandlt in 1950s a

u
μ T  ρu T l or μ T  ρl 2 Eq. 3.12
y

Where l a mixing length can be thought of as a transverse distance over which the particles maintain
their original momentum. For 3D thin shear layers, Prandtl’s equation is usually interpreted as:

12 David C. Wilcox, “Turbulence Modeling for CFD”, Copyright © 1993, 1994 by DCW Industries, Inc.
 20

1/2

2  u 
2
 w  
2

μ T  ρl       Eq. 3.13

 y   y  

This formula treats the turbulent viscosity as a scalar and gives qualitively correct results, specially
near the wall. The evaluation of l in the mixing length model varies with the type of flow being
considered, wall boundary layer, jet, wakes, etc. For flow along a solid surface (internal or external),
good results are observed by evaluating l according to


/A  y( τ w /ρ w )1/2
linner  κy(1 e  y ) , louter  C1δ and y  Eq. 3.14
νw

Where linner predicts the inner region close to wall and Iouter exceeds linner. The κ is the von Karman
constant as 0.41 and A+ is the damping constant usually set to 26. The expression for linner is
reasonable for producing the inner law-of-the-wall region of turbulent flow and louter produces the
outer “wake-like” region. These two zones are depicted in Figure 3.7 which shows a typical velocity
distribution for an incompressible turbulent boundary layer on a smooth impermeable plate using
“law-of-the-wall” coordinates.

3.4.1.3 Baldwin-Lomax
[Baldwin & Lomax -1978] proposed an alternative algebraic model to eliminate some of the difficulty
in defining a turbulence length scale from the shear-layer thickness. It is a two-layer algebraic 0-
equation model which gives the eddy viscosity μt as a function of the local boundary layer velocity
profile. The model is suitable for high-speed flows with thin attached boundary-layers, typically
present in aerospace and turbo machinery applications. It is also commonly used in quick design
iterations where robustness is more important than capturing all details of the flow physics. The
Baldwin-Lomax model is not suitable for cases with large separated regions and significant
curvature/rotation effects.

One-Equation Model
While employing a much simpler approach than two-equation or second-order closure models, one-
equation models have been somewhat unpopular and have not showed a great deal of success. One
notable exception was the model formulated by [Bradshaw, Ferris, and Atwell -1967], whose model
was tested against the best experimental data of the day at the 1968 Stanford Conference on
Computation and Turbulent Boundary Layers. There has been some renewed interest in the last
several years due to the ease with which one-equation models can be solved numerically, relative to
more complex two-equation or second-order closure models. An obvious shortcoming of algebraic
methods is that μT and uT is zero at the center of for example pipe line cases. The mixing-length model
can be fixed up to overcome this using

μ T  C k ρl (κ)1/2
For 2D incompressible thin - shear layer Eq. 3.15
Dκ   μ T  κ 
2
 u  C ρ(κ) 3/2
   μ    μ T    D
Dt y  Prκ  y   y  l
 21

Where Prκ is defined as Prandtl number for turbulence kinetic energy (≃1.0) and CD ≃ 0.164.

3.4.2.1 K Model
The complete derivation is obtained in two equation model.

3.4.2.2 Spallart - Allmaras

The [Spalart – Allmaras] model adds a single additional variable for a Spalart - Allmaras viscosity and
does not use any wall functions; it solves the entire flow field. The model was originally developed
for aerodynamics applications and is advantageous in that it solves for only a single additional
variable. This makes it less memory-intensive than the other models that solve the flow field in the
buffer layer. Experience shows that this model does not accurately compute fields that exhibit shear
flow, separated flow, or decaying turbulence. Its advantage is that it is quite stable and shows good
convergence.

Two Equation Model

While Kolmogorov’s κ-ω model was the first two-equation model, the most extensive work has been
done by [Daly & Harlow; 1970] and [Launder & Spalding; 1972]. Launder's κ-ε model is the most
widely used two-equation turbulence model; here ε is the dissipation rate of turbulent kinetic energy.
Independently of Kolmogorov, [Saffman ; 1970] developed a κ-ω model that shows advantages to the
more well-known κ-ε model, especially for integrating through the viscous sub-layer and in flows
with adverse pressure gradients. The κ-ε model is very popular for industrial applications due to its
good convergence rate and relatively low memory requirements. It does not very accurately compute
flow fields that exhibit adverse pressure gradients, strong curvature to the flow, or jet flow. It does
perform well for external flow problems around complex geometries.
It was concluded that the algebraic and one equation models did not lend itself to application with
the complex geometries that occur with internal flow calculations. In this context, we define an
internal flow as a flow through turbomachinery blading, as opposed to the external flow that occurs
with isolated airfoil or aircraft wing flow field predictions. When dealing with internal flows, one can
overcome the limitations of a one-equation model with a two-equation model. The currently favored
two-equation models are either the 𝑘-𝜀 model or the 𝑘-𝜔 model13. Both models require one to solve
two transport equations to compute eddy viscosity as an algebraic expression of turbulent kinetic
energy (𝑘) and dissipation rate of turbulent kinetic energy (𝜀) or turbulence frequency (𝜔). In two-
equation models the first equation is typically for turbulent kinetic energy (𝑘) and the second for
either dissipation (𝜀) or turbulence frequency (𝜔). Both the 𝑘-𝜀 model and the 𝑘-𝜔 model rely on an
assumption that one can link eddy viscosity to a time and length scale that characterizes
turbulence that in turn links to the computed flow-field’s characteristics. A feature of the 𝑘-𝜀 and
𝑘-𝜔 models is that the additional transport equations for 𝑘, 𝜀, and 𝜔 share the same form, and,
therefore, for a generic 𝜙 quantity, it reads

D   ν   
 P  ε   ν  T 
 x  Eq. 3.16
Dt x j  σ  j 

On the left-hand side is the quantity’s material derivative. On the right-hand side are one or more
production terms, a dissipation term, a diffusion term dependent on molecular viscosity, and another

13Alessandro Corsini, Giovanni Delibra, and Anthony G. Sheard, “A Critical Review of Computational Methods
and Their Application in Industrial Fan Design”, Hindawi Publishing Corporation, ISRN Mechanical Engineering,
Volume 2013, Article ID 625175.
 22

given as the turbulent viscosity’s function, corrected using the Prandtl number 𝜎𝜙. The primary
difference between the 𝑘-𝜀 and 𝑘-𝜔 models is the different trend of 𝜀 and 𝜔 at the wall and the
definition of the wall boundary conditions for the same variables. When one studies normalized
values of 𝜀 and 𝜔 for an attached flow, it is evident that 𝜔 is less dependent on the Reynolds number
than 𝜀 in the wall’s near vicinity.
There is a general consensus within the computational fluid dynamics community that the 𝑘-𝜀 model
better reproduces the energy cascade of large-scale structures in the main flow core, whilst the 𝑘-𝜔
model performs better near the wall14. A realization that 𝑘-𝜀 models perform better in the
main flow whilst the 𝑘-𝜔 models perform better near the wall leads to the natural conclusion that,
ideally, one would use the two models in combination. It was observed that it is possible to combine
𝑘-𝜀 and 𝑘-𝜔 models as one can reformulate every two-equation model into every other by changing
model coefficients. This realization has enabled engineers to formulate the 𝑘-𝜔 shear stress transport
(𝑘-𝜔 SST) model that solves the equation for 𝜔 near the wall and 𝜀 elsewhere. The use of two-
equation models has become established within the industrial community.

3.4.3.1 Standard Transport Equation for k

To obtain the equation for turbulent kinetic energy k, complicated algebra and rearrangements are
made to the time-averaged continuity equation and the time-averaged Navier-Stokes equations for
momentum. The mathematical manipulations are extensive, and therefore only a short description
of what is done, followed by the resulting equations is presented. The 3 continuity equations are each
multiplied by the respective fluctuating velocity component and then added together. The same
process is carried out for the Reynolds equations for momentum. The two resulting equations are
subtracted and rearranged extensively. Terms for viscous dissipation of turbulent kinetic energy and
the Boussinesq assumption related to Reynold’s stress equation, as shown below in15.

𝛛(𝛒𝛋) 𝛛(𝛒𝐮̅ 𝐢 𝛋) 𝛛 𝐮′𝐢 𝐮′𝐣 ̅𝐢

𝛛𝐮 ̅̅̅̅̅̅̅̅̅̅̅
𝛛𝐮 ′ 𝛛𝐮′
𝐢 𝐢
+ = ′
[𝛒𝐮𝐢 . ( ′ ̅̅̅̅̅̅
′ ′
+ 𝐩 )] − 𝛒𝐮𝐢 𝐮𝐣 . − 𝛍. ′ . ′
⏟𝛛𝐭 ⏟ 𝛛𝐱 𝐢 𝛛𝐱
⏟𝐢 𝟐 ⏟ 𝛛𝐱 𝐣 ⏟ 𝛛𝐱 𝐣 𝛛𝐱 𝐣
𝟏 𝟐 𝟑 𝟒 𝟓
Eq. 3.17
Terms (3) and (5) are replaced using scalar diffusion transport terms for (3) and the time-averaged
term for (5) to result in the following:

𝛛(𝛒𝛋) 𝛛(𝛒𝐮̅ 𝐢 𝛋) 𝛛 𝝁𝑻 𝝏𝜿 ̅𝐢
𝛛𝐮
+ = [ ̅̅̅̅̅̅
] − 𝛒𝐮 ′ 𝐮′ .
𝐢 𝐣 − 𝛒.
⏟ 𝛆
⏟𝛛𝐭 ⏟ 𝛛𝐱 𝐢 𝛛𝐱 𝐢 𝝈𝜿 𝝏𝒙𝒊
⏟ ⏟ 𝛛𝐱 𝐣
𝟓
𝟏 𝟐 𝟑 𝟒
Eq. 3.18
Where is a constant turbulent Schmidt number for κ. The terms (1)-(5) of the turbulent kinetic energy
k transport equation, can be interpreted as the following:

1. Transient term Accumulation of κ (rate of change of k).

2. Convective transport of κ by convection.
3. Diffusive transport of κ by pressure, viscous stresses, and Reynolds stresses (must be
modelled).

14 Alessandro Corsini, Giovanni Delibra, and Anthony G. Sheard, “A Critical Review of Computational Methods
and Their Application in Industrial Fan Design”, Hindawi Publishing Corporation, ISRN Mechanical Engineering,
Volume 2013, Article ID 625175.
15 Tennekes, H.; Lumley, J. L. “A First Course in Turbulence”, MIT Press, Cambridge, MA (1972).
 23

4. Production term Rate of production of k due from the mean flow.

5. Viscous dissipation Rate of viscous dissipation of k (must be modelled) k is a constant
(turbulent Schmidt number) of the k equation.

The above equation for transport of κ along with the equation for transport of ε, constitute the two
additional transport equations to be solved in addition to the RANS equations in the κ-ε turbulence
model. The next section presents the equation for transport of ε.

3.4.3.2 Standard Transport Equation for ε

𝛛 𝛛 𝛛 𝛍𝐓 𝛛𝛆 𝛆 ̅𝐢
𝛛𝐮 𝛆𝟐
+ (𝛒𝐮
̅ 𝐢 𝛆) = [ ] + 𝐂𝟏 (−𝛒𝐮𝐮. ) − 𝐂𝟐 𝛒
⏟ ⏟
𝛛𝐭 𝛛𝐱 𝐢 𝛛𝐱 𝐢 𝛔𝛆 𝛛𝐱 𝐢
⏟ 𝛋
⏟ 𝛛𝐱 𝐢 ⏟ 𝛋
𝟏 𝟐 𝟑 𝟒 𝟓
Eq. 3.19
Similar to the transport equation for κ, the transport equation for ε includes the terms 1 - 5:

1. Accumulation, or rate of change, of ε, Cμ C1 C2 σκ σε

2. Rate of destruction of ε. 0.09 1.44 1.92 1.0 1.3
3. Diffusive transport of ε,
Table 3.1 Constant values for κ-
4. Rate of production of ε, and
ε turbulence model equations
5. Transport of ε by convection,

And C1, C2, σε are constants of the ε equation (see Table 3.1). Alternatively, we inscribe both
equations in a more compact form, using Lagrangian Derivatives terms D/Dt :

Dκ  1  μT  κ  μ T 2
   μ    Sij  ε
Dt x j  ρ  σk  x j  ρ
Dε ε μ  1  μT  ε 
 (C1 T Sij2  C 2 ε)    μ    Eq. 3.20
Dt κ ρ x j  ρ  σε  x j 
κ2 1  u u j 
where μ T  ρCμ and Sij   i  
ε 2  x j x i 

After obtaining κ and ε we are ready to evaluate μT as

Cμ ρ(κ) 2
μT  Eq. 3.21
ε

Despite the enthusiasm which is noted from time to time over two equation model, it is perhaps
appropriate to point out two major restriction on this type of models. Since the two equation model
basically turbulent viscosity models which assumes that the Boussinesq approximation holds.
Therefore, its validates depends to Boussinesq approximation. In algebraic methods, μT is a local
function whereas in two equation model is a more general and complex functioned governing by two
additional PDEs. The second shortcoming is the need to make assumptions in evaluating the various
 24

terms in model transport equation especially third order turbulent correlations16. The same short
coming plagues all other higher order closure attempts, so there is no magic bullet.

3.4.3.3 The RNG Model

The κ-ε model uses “highly abstruse‟ mathematics to extend the eddy viscosity turbulence model and
changes the governing equations by removing smaller scales of motion, replacing them with large
motions and modifying the viscosity term. Results have been good for backward-facing steps, while
other results have been mixed. The RNG model is not very commonly used and has a high
computational overhead.

3.4.3.4 The Realizable κ-ε Model

This is a non-linear version of the κ-ε model. It retains the two-equation κ-ε equations, but expands
the model by including additional effects to account for Reynold’s stress anisotropy without actually
using the seven extra equations used in the RSM (described later) to exactly model the Reynolds
stresses. The turbulent viscosity expression and the rate of dissipation of kinetic energy equation of
the standard κ-ε model are both changed in the realizable κ-ε model to take into account that
turbulence does not always adjust itself instantaneously while moving through the flow domain,
meaning that the Reynolds stress is partially dependent on the mean strain rate itself. This means
that the non-linear realizable κ-ε model allows for the phenomena of the state of turbulence lagging
behind the changes disturbing the turbulence production and dissipation balance.

3.4.3.5 Issues Relating to κ-ε Turbulence Modeling

Acceding to [Menter17], the κ-ε model has been very successful in a large variety of different flow
situations, but it also has a number of well-known shortcomings. From the standpoint of
aerodynamics, the most disturbing of them is the lack of sensitivity to adverse pressure-gradients.
and separation. Under those conditions, the model predicts significantly too high shear-stress levels
and thereby delays (or completely prevents) separation. This could be attributes this shortcoming to
the over prediction of the turbulent length-scale in the near wall region and has shown that a
correction proposed by [Hanjalic and Launder] improves the predictions considerably. However, the
correction is not coordinate-invariant and can therefore not be applied in general coordinates. An
alternative way of improving the results has been proposed by [Chen and Patel] and by [Rodi]. They
replace the ε-equation in the near wall region by a relation that specifies the length-scale analytically.
This also reduces some of the stiffness problems associated with the solution of the model. Although
the procedure is coordinate independent, it has only been applied to relatively simple geometries,
where the change between the algebraic relation and the e-equation could be performed along a pre-
selected gridline. Clearly this cannot be done in flows around complex geometries. Furthermore, the
switch has to be performed in the logarithmic part (the algebraic length-scale is not known in the
wake region), so that the original k - ε model is still being used over most of the boundary layer.
Another problem with the k- ε model is associated with the numerical stiffness of the equations when
integrated through the viscous sublayer. This problem clearly depends on the specific version of the
k – ε model selected, but there are some general aspects to it. All low Reynolds number k – ε models
employ damping functions in one form or another in the sublayer. These are generally highly
nonlinear functions k of dimensionless groups of the dependent variables like Rt= κ2/εν (models
involving y+ are undesirable in separated flows). The behavior of these functions cannot easily be
controlled by conventional linearization techniques and can therefore interfere with the convergence

16Anderson, Dale A; Tannehill, John C; Plecher Richard H; 1984,”Computational Fluid Mechanics and Heat
Transfer”, Hemisphere Publishing Corporation.

Florian R. Menter, “Improved Two-Equation k-ωTurbulence Models for Aerodynamic Flows”, NASA Technical
17

Memorandum 103975, October 1992.

 25

properties of the scheme. A second problem is that ε does not go to zero at a nonslip surface. There
is a significant number of alternative models that have been developed to overcome the shortcomings
of the κ-ε model. One of the most successful, with respect to both, accuracy and robustness, is the κ-
ω model of Wilcox18. It solves one equation for the turbulent kinetic energy k and a second equation
for the specific turbulent dissipation rate (or turbulence frequency) ω. The model performs
significantly better under adverse pressure-gradient conditions than the κ-ε model although it is the
authors experience that an even higher sensitivity to strong adverse pressure-gradients would be
desirable. Another strong-point of the model is the simplicity of its formulation in the viscous
sublayer. The model does not employ damping functions and has straightforward Dirichlet boundary
conditions. This leads to significant advantages in numerical stability.

3.4.3.6 The κ-ω and SST Models

As mentioned before, one of the main problems in turbulence modeling is the accurate prediction of
flow separation from a smooth surface. Standard two-equation turbulence models, such as κ-ε
models, often fail to predict the onset and the amount of flow separation under adverse pressure
gradient conditions. This is an important phenomenon in many technical applications, particularly
for airplane aerodynamics because the stall characteristics of a plane are controlled by the flow
separation from the wing. For this reason, the aerodynamic community has developed a number of
advanced turbulence models for this application. In general, turbulence models based on the ε
equation predict the onset of separation too late and under-predict the amount of separation later
on. This is problematic, as this behavior gives an overly optimistic performance characteristic for an
airfoil. The prediction is therefore not on the conservative side from an engineering stand-point. The
models developed to solve this problem have shown a significantly more accurate prediction of
separation in a number of test cases and in industrial applications. Separation prediction is important
in many technical applications both for internal and external flows. Currently, the most prominent
two-equation models in this area are the κ-ω based models of [Menter]19. The κ-ω based Shear-
Stress-Transport (SST) model was designed to give a highly accurate predictions of the onset and the
amount of flow separation under adverse pressure gradients by the inclusion of transport effects into
the formulation of the eddy-viscosity. This results in a major improvement in terms of flow
separation predictions.

3.4.3.7 Wilcox (Standard) κ-ω Model

The starting point of the present formulation is the κ-ω model developed by [Wilcox]20. It solves two
transport equations, one for the turbulent kinetic energy, κ , and one for the turbulent frequency, ω.
The stress tensor is computed from the eddy-viscosity concept.

18 W'llcox, D. C., "Reassessment of the Scale-Determining Equation for Advanced Turbulence Models," AIAA
Journal, Vol.26, Nov. 1988, pp.1299-1310.
19 Menter, F.R., “Two-equation eddy-viscosity turbulence models for engineering applications”, AIAA-Journal.,

32(8), pp. 1598 - 1605, 1994.

20 Wilcox, D. C. (2008), “Formulation of the k–ω Turbulence Model Revisited”, 46 (11), AIAA Journal, pp. 2823–

2838, Bibcode:2008AIAAJ..46.2823W, doi:10.2514/1.36541.

 26

Dκ   ρκ  κ  ui
 ρP  β*ρωκ   μ  σ κ   with P  τ ij
Dt x j  ω  x j  xj
Eq. 3.22
Dω γω   ρκ  ω  ρσ d κ ω
 P - βρω 2   μ  σ ω  
Dt κ x j  ω  x j  ω x j x j

For recommendations for the values of the different parameters, see Wilcox (2008)21.

3.4.3.8 The Baseline (BSL) κ-ω Model

The main problem with the Wilcox model is its well-known strong sensitivity to freestream
conditions [Menter]22. Depending on the value specified for ω at the inlet, a significant variation in
the results of the model can be obtained. This is undesirable and in order to solve the problem, a
blending between the κ-ω model near the surface and the κ-ε model in the outer region was
developed by [Menter]23. It consists of a transformation of the κ-ε model to a κ-ω formulation and a
subsequent addition of the corresponding equations. The Wilcox model is thereby multiplied by a
blending function F1 and the transformed κ-ε model by a function (1 - F1), F1 is equal to one near the
surface and decreases to a value of zero outside the boundary layer (that is, a function of the wall
distance). At the boundary layer edge and outside the boundary layer, the standard κ-ε model is
therefore recovered24.

3.4.3.9 SST κ-ω Model

The Shear-Stress Transport (SST) κ-ω turbulence model is a type of hybrid model, combining two
models in order to better calculate flow in the near-wall region. It was designed in response to the
problem of the κ-ε models unsatisfactory near-wall performance for boundary layers with adverse
pressure gradients. It utilizes a standard κ-ε model to calculate flow properties in the free-stream
(turbulent) flow region far from the wall, while using a modified k-ε model near the wall using the
turbulence frequency ω as a second variable instead of turbulent kinetic energy dissipation term ε.
Since the air in this projects heat exchanger is flowing between two flat fins very close to each other,
it is expected that the boundary layer flow has a strong influence on the results, and properly
modelling this near-wall flow could be important for accuracy of the calculations. Therefore the SST
κ-ω turbulence model has also been chosen for CFD simulations in this project. This SST κ-ω model
is similar to the κ-ε turbulence model, but instead of ε as the second variable, it uses a turbulence
frequency variable omega, which is expressed as ω = ε/k [s-1]. The SST κ-ω model computes Reynolds
stresses in the same way as in the κ-ε model. The transport equation for turbulent kinetic energy k
for the k-ω model is:

21 Wikipedia.
22 Menter, F.R., “Multiscale model for turbulent flows”, In 24th Fluid Dynamics Conference. American Institute of
Aeronautics and Astronautics, 1993.
23 Menter, F.R., “Two-equation eddy-viscosity turbulence models for engineering applications”, AIAA-Journal.,

32(8), pp. 1598 - 1605, 1994.

24 Training Manual, ANSYS CFX.
 27

𝛛(𝛒𝛋) 𝛛 𝛛 𝛍𝐓
+ (𝛒𝐮
̅ 𝐢 𝛋) = [(𝛍 + ) 𝛁𝛋 ] + 𝐏
⏟ 𝛃∗ 𝛒𝛋𝛚
𝛋−⏟
⏟𝛛𝐭 𝛛𝐱 𝐢
⏟ 𝛛𝐱 𝐢
⏟ 𝛔𝛋
𝟒 𝟓
𝟏 𝟐 𝟑
̅ 𝐢 𝛛𝐮
𝛛𝐮 ̅𝐢 𝟐 ̅𝐢
𝛛𝐮
𝐰𝐡𝐞𝐫𝐞 𝐏𝛋 = (𝟐𝛍𝐓 . − 𝛒𝛋 𝛅 )
𝛛𝐱 𝐣 𝛛𝐱 𝐣 𝟑 𝛛𝐱 𝐣 𝐢𝐣
Eq. 3.23
The terms (1) - (5) in above expression, the turbulent kinetic energy k transport equation for the SST
Omega turbulence model, can be interpreted as the following:

1. Transient term - Accumulation of k (rate of change of k)

2. Convective transport - Transport of k by convection
3. Diffusive transport - Turbulent diffusion transport of k
4. Production term - Rate of production of k
5. Dissipation Rate of dissipation of k
Where σκ and β* are equation constants. The transport equation for turbulent frequency ω for the k-
ω model is:

𝛛(𝛒𝛚) 𝛛(𝛒𝐮̅ 𝐢 𝛚) 𝛛 𝛍𝐓
+ = [(𝛍 + ) 𝛁𝛚] +
⏟𝛛𝐭 ⏟ 𝛛𝐱 𝐢 𝛛𝐱
⏟𝐢 𝛔𝛚,𝟏
𝟏 𝟐 𝟑
̅ 𝐢 𝛛𝐮
𝛛𝐮 ̅𝐢 𝟐 ̅𝐢
𝛛𝐮 𝛒 𝛛𝛋 𝛛𝛚
𝛄𝟐 (𝟐𝛒 . − 𝛒𝛚 𝛃𝟐 𝛒𝛚𝟐 + 𝟐
𝛅𝐢𝐣 ) − ⏟
⏟ 𝛛𝐱 𝐣 𝛛𝐱 𝐣 𝟑 𝛛𝐱 𝐣 𝛔
⏟𝛚,𝟐 𝛛𝐱 𝛋 𝛛𝐱 𝜿
𝟓
𝟒 𝟔
Eq. 3.24
The general description for each of the terms in (1) to (6) are the usual terms for accumulation,
convection, diffusion, production, and dissipation of ω. The last term (6) is called a “cross-diffusion‟
term, an additional source term, and has a role in the transition of the modelling from ε to ω. The
constants for the Mentor SST κ-ω turbulence model are
listed in β* β2 σκ σω,1 σω,2 Υ2
Table 3.2. Additional modifications have been made to 0.09 0.083 1.0 2.0 1.17 0.44
the model for performance optimization. There are
blending functions added to improve the numerical Table 3.2 Constant values for SST κ-ω
stability and make a smoother transition between the turbulence model equations
two models. There have also been limiting functions
made to control the eddy viscosity in wake region and adverse pressure flows25. The κ-ω model is
similar to κ-ε, but it solves for ω, the specific rate of dissipation of kinetic energy. It also uses wall
functions and therefore has comparable memory requirements. Additionally, it has more difficulty
converging and is quite sensitive to the initial guess at the solution. Hence, the κ-ε model is often used
first to find an initial condition for solving the κ-ω model. The κ-ω model is useful in many cases

25Versteeg, H K; Malalasekera, W.”An Introduction to Computational Fluid Dynamics, The Finite Volume
Method”, Second edition, Pearson Education Limited, Essex, England (2007).
 28

where the κ-ε model is not accurate, such as internal flows, flows that exhibit strong curvature,
separated flows, and jets26.

Case Study – Subsonic Flow for Turbulence Models of (NACA)-0012 Airfoil

The analysis of the 2D subsonic flow over a (NACA) 0012 airfoil at various angles of attack and
operating at a Re = 3×106 is presented by [Eleni et al.]27. The flow was obtained by solving the steady-
state governing equations combined with one of three turbulence models (Spalart-Allmaras,
Realizable κ-ε, and κ-ω Shear Stress Transport (SST). The aim of the work was to show the behavior
of the airfoil at these conditions and to establish a verified solution method. The computational
domain was composed of 80000 cells emerged in a structured way, taking care of the refinement of
the grid near the airfoil in order to enclose the boundary layer approach. Calculations were done for
constant air velocity altering only the angle of attack for every turbulence model tested. This work
highlighted two areas in (CFD) that require further investigation: transition point prediction and
turbulence modeling. The laminar to turbulent transition point was modeled in order to get accurate
results for the drag coefficient at various Reynolds numbers. In addition, calculations showed that
the turbulence models used in commercial CFD codes does not give yet accurate results at high angles
of attack.

3.4.4.1 Computational Method

The NACA 0012, with Reynolds number of Re = 3x106, same with the reliable experimental data from
[Abbott and Von Doenhoff]28, in order to validate the present simulation. The free stream
temperature is 300 K, which is the same as the environmental temperature. The density of the air at
the given temperature is ρ=1.225kg/m3 and the viscosity is μ=1.7894×10-5 kg/ms. For this Reynolds
number, the flow can be described as incompressible. This is an assumption close to reality and it is
not necessary to resolve the energy equation. A segregated, implicit solver was utilized where the
calculations were done for angles of attack ranging from -12 to 20. The first step in performing a CFD
simulation should be to
investigate the effect of the
mesh size on the solution
results. The appropriate
number of nodes can be
determined by increasing the
number of nodes until the
mesh is sufficiently fine so that
further refinement does not
change the results (mesh
independence). Figure 3.3
shows the effect of number of
grid cells in coefficient of lift at
stall angle of attack (16°). This
study revealed that a C-type
grid topology with 80000
quadrilateral cells would be Figure 3.3 Lift Coefficient at Stall (AoA) against Number of Cells

26 Walter Frei, “Which Turbulence Model Should I Choose for My CFD Application? “, COMSOL Blog, September
16, 2013.
27 D. C. Eleni, Tsavalos I. Athanasios and Margaris P. Dionissios, “Evaluation of the turbulence models for the

simulation of the flow over a National Advisory Committee for Aeronautics (NACA) 0012 airfoil”, Journal of
Mechanical Engineering Research, March 2012.
28 Abbott IH, Von Doenhoff AE. “Theory of Wing Sections”. Dover Publishing, New York.
 29

sufficient to establish a grid independent solution. The domain height was set to approximately 20
chord lengths, and the height of the first cell adjacent to the surface was set to 10-5, corresponding to
a maximum y+ of approximately 0.2. A y+ of this size should be sufficient to properly resolve the inner
parts of the boundary layer. In order to include the transition effects in the aerodynamic coefficients
calculation and get accurate results for the drag coefficient, a new method was used. The transition
point from laminar to turbulent flow on the airfoil was determined and the computational mesh was
split in two regions, a laminar and a turbulent region. To calculate the transition point the following
procedure was used. A random value for the transition point (xtr) was chosen and the computational
domain was split at that point with a perpendicular line. The problem was simulated by defining the
left region as laminar and the right as turbulent zone.

3.4.4.2 Results and Discussion

Simulations for various angles of attack were done in order to be able to compare the results from
the different turbulence models and then validate them with existing experimental data from reliable
sources. To do so, the model was solved with a range of different angles of attack from -12 to 20°. On
an airfoil, the resultants of the forces are usually resolved into two forces and one moment. The

Figure 3.4 Comparison between experimental data [Abbott et al.]and three different turbulent
models simulation results of the Lift Coefficient for NACA 0012 Airfoil

component of the net force acting normal to the incoming flow stream is known as the lift force and
the component of the net force acting parallel to the incoming flow stream is known as the drag force.
The curves of the lift and the drag coefficient are shown for various angles of attack, computed with
three turbulence models and compared with experimental data. Figure 3.4 shows that at low angles
of attack, the dimensionless lift coefficient increased linearly with angle of attack. Flow was attached
to the airfoil throughout this regime. At an angle of attack of roughly 15 to 16°, the flow on the upper
 30

surface of the airfoil began to separate and a condition known as stall began to develop. All three
models had a good agreement with the experimental data at angles of attack from -10 to 10° and the
same behavior at all angles of attack until stall. It was obvious that the Spalart-Allmaras turbulence
model had the same behavior with the experimental data as well as after stall angle.
Near stall, disagreement between the data was shown. The lift coefficient peaked and the drag
coefficient increased as stall increased. The predicted drag coefficients were higher than the
experimental data (Figure 3.5). This over prediction of drag was expected since the actual airfoil has
laminar flow over the forward half. The turbulence models cannot calculate the transition point from
laminar to turbulent and consider that the boundary layer is turbulent throughout its length. From
theory, the turbulent boundary layer carries more energy and is much greater than at the viscous
boundary layer, which carries less energy. The computational results must be compared with
experimental data of a fully turbulent boundary layer. This was done only for CD as CL is less sensitive
to the transition point.
[ Johansen]29 contained experimental data of CD for the NACA 0012 airfoil and Re = 3×106, where the
boundary layer formed around the airfoil is fully turbulent. Figure 3.5 shows the curves of CD for
various angles of attack, compared with experimental data for fully turbulent boundary layer30. The
values of from the three turbulence
models were very close to
experimental data for the fully
turbulent boundary layer. The most
accurate model was the κ-ω SST
model, next came the Spalart-
Allmaras, and latest in precision
was the Realizable κ-ε .
In order to get more accurate results,
the computational domain could be
split into two different domains to run
mixed laminar and turbulent flow. The
disadvantages of this approach were
that the accuracy of simulations
depends on the ability to accurately
guess the transition location, and a
new grid must be generated if the
transition point had to change
[Silisteanu-Botez]31. If the transition Figure 3.5 Comparison Between Experimental Data for
point is known, the grid can easily be Transitional Boundary Layer and Different Turbulent Models
split in two with a vertical line that on the Drag Coefficient of NACA-0012 Airfoil
passes through this point and then
laminar and turbulent zones are defined. The results of this method at angle of attack a=0 and
operating at Re = 1×106, 2×106, 3×106, 4×106 and 5×106. Initially, was calculated for a fully turbulent
boundary layer and compared with CD experimental data from NASA [McCroskey]32. Then,

29 Johansen J. “Prediction of Laminar/Turbulent Transition in Airfoil Flows”. RISE National Laboratory, Roskilde,
Denmark, 1997.
30 Comparison Between Different Turbulent Models and Experimental Data obtained by [Abbott & Von

Doenhoff] and [Johansen] for Transitional Boundary Layer on the Drag Coefficient of NACA-0012 Airfoil.
31 Silisteanu PD, Botez RM. “Transition flow occurrence estimation new method”. 48th AIAA Aerospace Science

Meeting. Orlando, Florida, 2010.

32 McCroskey WJ, ”A Critical Assessment of Wind Tunnel Results for the NACA 0012 Airfoil”. U.S. Army Aviation

Research and Technology Activity, Nasa Technical Memorandum, 42: 285-330, 1987.
 31

simulations were made with the split grid for the five Reynolds numbers. The computational results
for the fully turbulent boundary layer agreed very well with the corresponding experimental data.
The discrepancy between the Drag Coefficient and experimental data from [McCroskey] for fully
turbulent boundary layer was up to 5.6%. On the other hand, the comparison between the simulation
results with the split grid and the experimental data from [McCroskey] for transitional boundary
layer showed an excellent agreement, with maximum error of about 3.6%. It was also observed that
as the Reynolds number increased, the Drag Coefficient decreased. When the boundary layer was
fully turbulent the reduction of was more intense and when there was a transition from laminar to
turbulent, was reduced to a much
lower rate. It is worth noting that
this process of calculating the
transition point is quite simple
when the angle of attack is zero
because the flow is symmetric and
the transition point is the same
above and below the airfoil. At
nonzero angles of attack the process
is more complicated because
transition points are different for
the upper and lower surface of the
airfoil. With different AoA, the
pressure on the lower surface of the
airfoil was greater than that of the
incoming flow stream and as a
result it effectively “pushed” the
airfoil upward, normal to the
incoming flow stream. On the other
hand, the components of the
pressure distribution parallel to the
incoming flow stream tended to
slow the velocity of the incoming
flow relative to the airfoil, as do the
viscous stresses.
Contours of velocity components at
angles of attack 9 and 16° are also
shown in Figure 3.6. The trailing
edge stagnation point moved
Figure 3.6 Contours of velocity magnitude at 9° (Top) and 16°
slightly forward on the airfoil at low
(Bottom) AoA with the Spalart-Allmaras turbulence model
angles of attack and it jumped
rapidly to leading edge at stall angle.
A stagnation point is a point in a flow field where the local velocity of the fluid is zero. The upper
surface of the airfoil experienced a higher velocity compared to the lower surface. That was expected
from the pressure distribution. As the angle of attack increased the upper surface velocity was much
higher than the velocity of the lower surface. For further information, please consult the [Eleni et
al.]33.

33D. C. Eleni, Tsavalos I. Athanasios and Margaris P. Dionissios, “Evaluation of the turbulence models for the
simulation of the flow over a National Advisory Committee for Aeronautics (NACA) 0012 airfoil”, Journal of
Mechanical Engineering Research, March 2012.
 32

3.4.4.3 Conclusions
This paper showed the behavior of the 4-digit symmetric airfoil NACA 0012 at various angles of
attack. The most appropriate turbulence model for these simulations was the κ-ω SST two-equation
model, which had a good agreement with the published experimental data of other investigators for
a wider range of angles of attack. The predicted drag coefficients were higher than the existing
experimental data from reliable sources. This over prediction of drag was expected since the actual
airfoil has laminar flow over the forward half. The computational results from the three turbulence
models were compared with experimental data where the boundary layer formed around the airfoil
is fully turbulent and they agreed well. Afterwards, the transition point from laminar to turbulent
regime was predicted, the computational grid split in two regions, a laminar and a turbulent region,
and then new simulations were realized. By this method, the computational results agreed very well
with corresponding experimental data34.

Near Wall Turbulence Modeling

Most turbulence models requires special algebraic formula, often called Wall Function to represent
the distribution of velocity, temperature, turbulence, energy, etc. within boundary layers. This
practice is necessary because these models are not valid in the region within the layer where the
molecular and turbulent effects are comparable in magnitude. It is also expedient because it avoids
the need for employing a fine mesh within boundary layer. The wall function representation of near-
wall turbulent behavior is inexact, the accuracy being dependent on the degree to which the
assumption and approximations embodied in the function correspondent with reality of the
application. The Standard wall function with following characteristics:

 Variation in velocity etc. are predominantly normal to the wall, leading to one dimensional
behavior.
 Effects of pressure gradients and body forces are negligible, leading to uniform shear stress
in the layer.
 Shear stress and velocity vectors are aligned and unidirectional through the layer.
 A balance exist between turbulence energy production and dissipation.
 There is a linear variation of turbulence length scales.

For flow along solid surfaces (internal or external), a good approximation is observed by evaluating
the expressions for linner is responsible for producing the inner law of the wall region of turbulent flow
and louter produces the outer wake-like region. These two zones are illustrated in Figure 3.7 which
depicts a typical velocity distribution for an incompressible turbulent boundary layer on a smooth
impermeable plate.

34 See Previous.
 33

Figure 3.7 Zones in Turbulent B. L. for a typical Incompressible flow over a smooth flat plate

3.4.5.1 Modeling Flow Near the Wall

Two approaches are commonly used to model the flow in the near-wall region:

1 - Wall Function Method

uses empirical formulas that impose suitable conditions near the wall without resolving the
boundary layer, thus saving computational resources. All turbulence models address a suitable wall
function method. The major advantages of the wall function approach is that the high gradient shear
layers near walls can be modeled with relatively coarse meshes, yielding substantial savings in CPU
time and storage. It also avoids the need to account for viscous effects in the turbulence model. (see
Figure 3.8).

2 - Low-Reynolds-Number Method
Resolves the details of the boundary layer profile by using very small mesh length scales in the
direction normal to the wall (very thin inflation layers). Note that the low-Re method does not refer
to the device Reynolds number, but to the turbulent Reynolds number, which is low in the viscous
sublayer. This method can therefore be used even in simulations with very high device Reynolds
numbers, as long as the viscous sublayer has been resolved. The computations are extended through
the viscosity-affected sublayer close to the wall. The low-Re approach requires a very fine mesh in
the near-wall zone and correspondingly large number of nodes. Computer-storage and run-time
requirements are higher than those of the wall-function approach and care must be taken to ensure
good numerical resolution in the near-wall region to capture the rapid variation in variables. To
reduce the resolution requirements, an automatic wall treatment was developed which allows a
gradual switch between wall functions and low-Reynolds number grids, without a loss in accuracy.
 34

Figure 3.8 Turbulence and near Wall Function

3.4.5.2 Case Study - Wall-modelling strategies in Large Eddy Simulation of separated high-
Reynolds-number flows
If ReL (say, that based on the boundary layer thickness) exceeds roughly 105, the resource
requirements are entirely dominated by the need to resolve the near-wall region (the inner layer),
with the number of nodes rising roughly as N ~ Re2.5L. Simulations for practical configurations would
not, in most circumstances, be undertaken today with meshes exceeding 10-50M nodes,
corresponding to ReL ~ 5*105, yet this is still a very modest Reynolds number in practice [Tessicini,
and Leschziner]35. The general approach taken in recent years towards addressing the problem of
near-wall resolution has been to combine RANS modelling near the wall with LES in the outer flow.
The key premise underpinning this strategy is that it should allow near-wall numerical cells to be
used that have far higher aspect ratios than those required by LES, typically 500-1000, relative to 50
in wall-resolving LES. Substantial savings could thus be made by using much coarser stream wise and
span wise meshes than are dictated by the LES constraints. Efforts currently have focused on two
particular methods, one being a hybrid LES-RANS scheme and the other being a two-layer zonal
schemes. The difference between them is explained by reference to Figure 3.9, which conveys the
manner by which the LES and RANS regions communicate numerically.
The hybrid method uses a single computational domain. Within a predefined layer near the wall, that
can be prescribed in terms of y+, RANS equations are solved using one-equation or two-equation
eddy-viscosity models that are dynamically adjusted so to comply with continuity of eddy viscosity
across the interface, νRANS= νLES, beyond which a LES sub-grid-scale model is used. To achieve this
compatibility, the RANS model coefficients at the interface is determined by comparison of the RANS
viscosity, containing the RANS-determined turbulence energy and dissipation rate at the interface,
to the LES viscosity at the interface. The variation of the coefficients from the interface to the wall is
then prescribed analytically, based on observations derived from a-priori wall-resolved LES
performed in channel flows. The zonal method uses two overlapping grids across the near-wall layer.
The LES grid extends to the wall, but is relatively coarse, maintaining cell-aspect-ratio constraints

35F. Tessicini, M.A. Leschziner, “Wall-modelling strategies in large eddy simulation of separated high- Reynolds-
number flows”.
 35

appropriate to LES. Within the near-wall layer, a separate grid is inserted, which is refined towards
the wall, typically to a wall-nearest node located at y+=O(1). Within that layer, parabolized RANS
equations are solved for the wall-parallel-velocity components, using a simple algebraic turbulence
model - for example, a mixing-length model.

Figure 3.9 Schematics of hybrid LES-RANS scheme (upper) and two-layer zonal scheme (lower)
 36

4 Beyond the Boussinesq Approximation (Non-Linear Modeling)

The Boussinesq eddy-viscosity approximation assumes the principal axes of the Reynolds-stress
tensor, τij, are coincident with those of the mean strain-rate tensor, Sij, at all points in a turbulent flow.
This is the analog of Stokes approximation for laminar flows. The coefficient of proportionality
between τij and Sij is the eddy viscosity, μT, which is linear. Unlike the molecular viscosity which is a
property of the fluid, the eddy viscosity depends upon many details of the flow under consideration.
It is affected by the shape and nature (e.g., roughness height) of any solid boundaries, freestream
turbulence intensity, and, perhaps most significantly, flow history effects. Experimental evidence
indicates that flow history effects on τij often persist for long distances in a turbulent flow, thus
casting doubt on the validity of a simple linear relationship between τij and Sij. Next, we outline
several flows for which the Boussinesq approximation yields a completely unsatisfactory description.

4.1 Boussinesq-Approximation Deficiencies

While models based on the Boussinesq eddy-viscosity approximation provide excellent predictions
for many flows of engineering interest, there are some applications for which predicted flow
properties differ greatly from corresponding measurements. Generally speaking, such models are
inaccurate for flows with sudden changes in mean strain rate and for flows with what Bradshaw
refers to as extra rates of strain. It is unsurprising that flows with sudden changes in mean strain rate
pose a problem. The Reynolds stresses adjust to such changes at a rate unrelated to mean flow
processes and time scales, so that the Boussinesq approximation must fail. Similarly, when a flow
experiences extra rates of strain caused by rapid dilatation, out of plane straining, or significant
streamline curvature, all of which give rise to unequal normal Reynolds stresses, the approximation
again becomes suspect. Some of the most noteworthy types of applications for which models based
on the Boussinesq approximation fail are:

1. flows with sudden changes in mean strain rate;

2. flow over curved surfaces;
3. flow in ducts with secondary motions;
4. flow in rotating and stratified fluids ;
5. three-dimensional flows;
6. flows with boundary-layer separation .

4.2 Nonlinear Constitutive Relations

One approach to achieving a more appropriate description of the Reynold stress tensor without
introducing any additional differential equations is to assume the Boussinesq approximation is
simply the leading term in a series expansion of functional. Proceeding with this premise it can be
shown that for incompressible flow the expansion must proceed through second order according to

 ui u j  2
 ρu i uj  μ T     δ ij ρ κ  or
 x x  3
 j i 
Eq. 4.1
2 1  u u j 
 ρu i uj   TSij  δ ij κ where Sij   i  
3 2  x j x i 

The above isotropic relation assumes that the principal axis of the Reynolds stress tensor S͞ ij coincides
 37

with that of the mean strain rate. The standard κ-ε model does not take into account the anisotropic
effects and fails to represent the complex interaction mechanisms between Reynolds stresses and
the mean velocity field. For example, the linear model fails to mimic the effects related to streamline
curvature, secondary motion, or flow with extra strain rates. These anisotropic effects can be
predicted by introducing a nonlinear expression for the Reynolds stresses as given in the following
expression36:

𝟐 ̅𝛋
𝛒 ̅𝛋
𝛒
̅̅̅̅̅̅̅
−𝛒𝐮 ′ 𝐮′ = − 𝛒
𝐢 𝐣 ̅ 𝛋𝛅 𝐢𝐣 + 𝟐𝛍 ̅
𝐒
𝐓 𝐢𝐣 − 𝐁 ̅
𝐒 ̅
𝐒 𝛅 − 𝐂 𝐒 𝐒 −
⏟𝟑 ⏟𝛚𝟐 𝐦𝐧 𝐧𝐦 𝐢𝐣 𝛚𝟐 𝐢𝐤 𝐤𝐣
𝐁𝐨𝐮𝐬𝐬𝐢𝐧𝐞𝐬𝐪 𝑵𝒐𝒏−𝑳𝒊𝒏𝒆𝒂𝒓 𝑻𝒓𝒆𝒎 𝟏
̅𝛋
𝛒 ̅𝛋
𝛒 ̅𝛋
𝛒
𝐃 𝟐 (𝐒̅𝐢𝐤 𝛀
̅ 𝐤𝐣 + 𝐒̅𝐣𝐤 𝛀
̅ 𝐤𝐢 ) − 𝐅 ̅ 𝐦𝐧 𝛀
𝛀 ̅ 𝐧𝐦 𝛅𝐢𝐣 − 𝐆 ̅ 𝛀
𝛀 ̅
⏟𝛚 𝛚𝟐 𝛚𝟐 𝐢𝐤 𝐤𝐣
𝑵𝒐𝒏−𝑳𝒊𝒏𝒆𝒂𝒓 𝑻𝒆𝒓𝒎 𝟐
𝛋 𝛋𝟑 ̅ 𝛛𝐮
𝟏 𝛛𝐮 ̅ ̅𝐣
̅ 𝐢 𝛛𝐮
𝟏 𝛛𝐮
𝐰𝐡𝐞𝐫𝐞 ≈ , ̅𝐢𝐣 = ( 𝐢 + 𝐣 )
𝐒 ̅ 𝐢𝐣 =
𝐚𝐧𝐝 𝛀 ( − )
𝛚𝟐 𝛆𝟐 𝟐 𝛛𝐱 𝐣 𝛛𝐱 𝐢 𝟐 𝛛𝐱 𝐣 𝛛𝐱 𝐢
Eq. 4.2
Where B, C, D, F, and G are closure coefficients. These coefficients of the non-linear terms should be
carefully determined because they are expected to influence the physical accuracy and numerical
performance of the model. Here, the coefficients are adjusted through the consideration of the
anisotropy in simple shear flows detailed by [Champagne et al.]37 and [Harris et al.]38 Solving this
equation is a daunting task, but there are assumptions (as the case with all turbulence models) which
can be made, depending to the case. We don’t get into different modeling but for an excellent
discussion, readers should refer to39. (See Eq. 4.2).

Case Study – 3D Simulation of Flow Past a Cylinder using Nonlinear Turbulence Model
The flow past a cylinder of circular cross section has been the subject of interest for industrial
researchers as well as scientists, because of its wide range of applications. To cite a few examples,
flow in bridge piers, chimney stacks, and tower structures in civil engineering; electrodes in chemical
engineering; nuclear fuel rods in the atomic field and heat exchanger tubes in thermal engineering,
etc., fall under this subject of study. Although the geometry is simple, the flow has complicated
features such as stagnation points, laminar boundary-layer separation, turbulent shear layers,
periodic vortex shedding, and wakes. Even though there is much literature available on numerical
simulation of laminar flow past a two-dimensional circular cylinder at low Reynolds number, a focus
on practical high Reynolds numbers is less. This could be due to the complexity of formulating
Reynolds stresses in turbulent flows. The majority of turbulent flow calculations carried out in earlier
days used two equation models such as the standard κ-ε model (hereafter referred as SKE) and the
κ-ω model, because of their robustness, computational efficiency, and completeness. In the classical
SKE model, the turbulent kinetic energy (κ) and the turbulent kinetic energy dissipation rate (ε) were

36 Ichiro Kimura, and Takashi Hosoda, ”A non-linear κ-ε model with realizability for prediction of flows around
bluff bodies”, Int. J. Numerical Meth. Fluids 2003; 42:813–837 (DOI: 10.1002/_d.540).
37 Champagne FH, Harris VG, Corrsin S. ,”Experiments on nearly homogeneous turbulent shear flow”. Journal of

Fluid Mechanics 1970; 41:81–139.

38 Harris VG, Graham JAH, Corrsin S. “Further experiments in nearly homogeneous turbulent shear flow”. Journal

of Fluid Mechanics 1977; 81:657–687.

39 David C. Wilcox, ”Turbulence Modeling for CFD”,1993, 1994 by DCW Industries, Inc.
 38

calculated using modeled transport equations separately for k and e along with the Boussinesq eddy
viscosity approximation.

4.2.1.1 Literature Survey

Subsequently, nonlinear models were proposed by many researchers such as [Gatski and Speziale]40,
[Craft et al].41, and Shih et al42. Some researchers have tested the appropriateness of nonlinear
turbulence models for simple flows such as homogeneous shear flow, separated flow over a
backward-facing step, and flow in a confined jet. They obtained encouraging results in the prediction
of mean and turbulent quantities. The two-dimensional, unsteady, Reynolds-averaged Navier-Stokes
(URANS) simulation of subcritical flow past a circular cylinder at Re = 1.4x105 was carried out by
[Tutar and Holdo]43 using the SKE, nonlinear κ-ε formulation and large-eddy simulation (LES)
technique. The wake centerline velocity recovery predicted by the nonlinear κ-ε turbulence model
showed better agreement with the LES and experimental results than those predicted by the SKE
model. [Saghafian et al.]44 simulated the two dimensional flow past a circular cylinder using a
nonlinear eddy viscosity model and tested the mean drag coefficient (Cd mean), root-mean-square
(RMS) lift coefficient (Cl RMS), and Strouhal number value (St = f D/U∞) for the range of Reynolds
numbers from subcritical laminar separation to supercritical turbulent separation.
The standard and renormalized group versions of the κ-ε model were examined by Jennifer 45 for
two-dimensional flow past a circular cylinder at subcritical Reynolds number Re=5,232. With this
increasing interest in nonlinear turbulence models, [Kimura and Hosoda]46 proposed a cubic
nonlinear κ-ε model by accounting for the effect of anisotropy. They tested the model for two-
dimensional flow around a square cylinder and a surface-mounted cubic obstacle. The model, which
included the realizability condition, performed better than linear models when compared with
experimental results. Recently, the above model was used by [Ramesh et al.]47 for simulating three
dimensional flow around a square cylinder, and they found it performs better than results by the
standard κ-ε and RNG κ-ε models.

4.2.1.2 Basic Equations

The ensemble-averaged RANS equations for an incompressible flow are Continuity and Momentum
equations given by:

40 T. B. Gatski and C. G. Speziale,“On Explicit Algebraic Stress Models for Complex Turbulent Flows”, J. Fluid Mech.,
vol. 254, pp. 59–78, 1993.
41 T. J. Craft, B. E. Launder, and K. Suga, “Development and Applications of a Cubic Eddy-Viscosity Model of

Turbulence”, Int. J. Heat Fluid flow, vol. 17, pp. 108–115, 1996.
42 T. H. Shih, J. Zhu, and J. L. Lumley, “A New Reynolds Stress Algebraic Equation Model”, Computer. Meth. Appl.

Mech. Eng., vol. 125, pp. 287–302, 1997.

43 M. Tutar and A. E. Holdo, “Computational Modeling of Flow around a Circular Cylinder in Subcritical Flow

Regime with Various Turbulence Models”, Int. J. Numerical Methods Fluids, vol. 35, pp. 763–784, 2000.
44 M. Saghafian, P. K. Stansby, M. S. Saidi, and D. D. Asplay, “Simulation of Turbulent Flows around a Circular

Cylinder Using Non-Linear Eddy-Viscosity Modeling: Steady and Oscillatory Ambient Flows”, J. Fluids Structure,
vol. 17, pp. 1213–1236, 2003.
45 R. B. Jennifer, “Verification Testing in Computational Fluid Dynamics: An Example Using Reynolds-Averaged

Navier-Stokes Methods for Two Dimensional Flow in the Near Wake of a Circular Cylinder”, Int. J. Numerical Meth.
Fluids, vol. 43, pp. 1371–1389, 2003.
46 L. Kimura and T. Hosoda, “A Non-linear κ-ε Model with Reliability for Prediction of Flows around Bluff Bodies”,

Int. J. Numerical Method Fluids, vol. 42, pp. 817–837, 2003.

47 V. Ramesh, S. Vengadesan, and J. L. Narasimhan, “3D Unsteady RANS Simulation of Turbulent Flow over Bluff

Body by Non-linear Model”, Int. J. Numerical Meth. Heat Fluid Flow, vol. 16, pp. 660–673, 2006.
 39

u i u i u i u j P (ui uj ) 1  u i
2
0 ,     Re Eq. 4.3
x i t i x j x i x j x i x j

where xi is the spatial coordinate, t is the time, u͞i is the ensemble-averaged velocity, ͞u´i is the
fluctuating velocity, and P is the averaged pressure divided by the density. As a result of ensemble-
averaging process, further unknowns are introduced into the momentum equations by means of
Reynolds stresses ͞u´iu´j . In engineering flows, closure approximation using two-equation models
for u
͞ ´ju´j have gained popularity because of their simplicity. In this article the study is confined to the
κ-ε model, which employs additional transport equations for turbulent kinetic energy κ and its
dissipation rate ε, and they are given as

κ κu j u   ν t  κ 
  ui uj i  ε    ν  
t x j x j x j σ
 κ  x j 
Eq. 4.4
ε εu j ε ε2   ν t  ε 
  C1 u i u j  C 2    ν  
t x j κ κ x j σ
 ε  x j 

Where κ is the turbulent kinetic energy, is the turbulent kinetic energy dissipation rate, ν is the fluid
kinematic viscosity, and νt is the eddy viscosity. C1, C2, σκ, and σε are the model constants given by
Table 3.1.

4.2.1.3 Numerical Strategies

The governing equations for velocities and turbulent quantities are solved using the finite-volume-
based commercial solver FLUENT® 6.2. The equations are discretized on a collocated grid in fully
implicit form. Momentum equations are solved using the QUICK scheme, and the SIMPLE algorithm
is used for coupling the pressure and velocity terms. The second-order upwind scheme is used to
discretize convective terms and also the terms in equations for turbulent quantities. The second-
order implicit scheme is used for time integration of each equation. The present nonlinear model is
incorporated in FLUENT through User-Defined Functions (UDFs). The nonlinear stress term is added
as a source term in equations for κ and ε.
The turbulent viscosity is also made to vary. The implementation of the present Nonlinear κ-ε model
(hereafter referred to as NLKE) was validated for turbulent flow around a square cylinder. A box-
type computational domain (Figure 4.1) with structured grid in Cartesian coordinates having origin
at the center of the cylinder is used. Stream wise direction is along the x axis with x ¼ 0 at the center
of the cylinder, the y axis is the vertical axis with y ¼ 0 the wake centerline, and the z axis is the span
wise direction with z ¼ 0 being the mid span of the cylinder. The upstream boundary with uniform
inlet velocity is placed at 7D from the center of the body, where D is the diameter of the cylinder. This
extent is less than that used in the LES simulation of FF02, where the inlet boundary was kept at 10D
from center of the body and the URANS, and the LES simulation of LU01, who used 8D from the center
of the body. At the inlet boundary, 2% turbulence intensity is specified. The convective boundary
condition at the outlet is specified at 20D downstream of the body. The same domain size was used
by FF02, whereas LU01 used 24D. In the y direction, slip boundary condition is applied at 8D from
the center of the body, which is the same as the one used by previous references. Most other reported
LES simulations have used either an O-grid or a C-grid with larger boundary dimensions. All the LES
simulations were done with span wise domain size of PD, whereas the URANS simulations have been
reported only for two-dimensional flow. The three-dimensional stability analysis and the experiment
 40

on a subcritical Reynolds number showed the dominant span wise scales having wavelengths of
approximately three to four cylinder diameters in the Reynolds number range 180 < Re < 240. After
this Reynolds number, the wavelength shortens to nearly one diameter. In our simulation, we have
taken a span wise length of 4D along the z direction, and periodic boundary condition is enforced on
the boundary. This extent is slightly larger than that used in LES and DNS calculations. A non-
equilibrium wall function approach is used to capture the adverse pressure gradient effect.

4.2.1.4 Geometry and

Meshing
The present NLKE model was
simulated with five different
grids for grid dependence
studies. In all grids, along the
circumference of the cylinder,
144 mesh points were placed,
and in the cross-stream
direction, 85 mesh points were
used. For all grids, the mesh
points were placed uniformly
in the span wise direction.
Different grids were achieved
by systematically changing the
resolution in the stream wise
and span wise directions.
Figure 4.1 shows the
schematic representation of
the computational domain
used. Due to ease of domain, a
simple structured grid being Figure 4.1 Schematic Representation of the Computational Domain
used. Of course, to be noted
that the overall grid size is
coarser than those used by LES and DNS. The same grid is used to simulate three-dimensional flow
by the standard κ-ε model (referred to as 3DSKE) as well as two-dimensional flow by the nonlinear
κ-ε model (2DNLKE) for comparison purposes.

4.2.1.5 Results and Discussion

The solution is initiated and allowed to march in time with a non-dimensional increment of dt (Δt
U∞/D =5x103) until the vortex shedding becomes periodic. The time variation of the lift coefficient is
shown in Figure 4.4, and periodicity is observed. The simulation is continued for five more vortex
shedding cycles to advect all the numerical errors to downstream. In Breuer’s LES work 48, in the time
history of the lift coefficient, there was a low-frequency component over a regular periodic
component. Hence, in order to achieve reproducible statistics, averaging was done over 22 vortex
shedding cycles. In the present work, as there are no such features, the time averaging is done over
10 vortex shedding cycles to obtain both bulk and mean field quantities. Here T (t U∞/D) is non
dimensional time for one vortex shedding cycle.
The present three-dimensional simulation with the NLKE model (referred as 3DNLKE) shows better
agreement with the experimental results. The present 3DNLKE simulation predicted the mean drag

48M. Breuer, “Large Eddy Simulation of the Subcritical Flow past a Circular Cylinder: Numerical and Modeling
Aspects”, Int. J. Numerical Meth. Fluids, vol. 28, pp. 1281–1302, 1998.
 41

coefficient (Cd mean) and Strouhal number

(St) within the experimental uncertainty, but
the two-dimensional simulation (2DNLKE)
under predicts these values. The
recirculation length (Lr=D) and negative
velocity (U1 min=U∞) in the wake predicted by
the present 3DNLKE simulation agreed well
with those of experimental and LES results.
Simulation by 3DSKE predicts longer
recirculation length (Lr=D) and less negative
velocity in the wake. Location of the
minimum velocity (rmin=D) by the present Figure 4.4 Variation of lift Coefficient with Time
two-dimensional simulation with the NLKE
model (2DNLKE) and the two-
dimensional URANS simulation result
from LU01 showed an under predicting
trend. However, the present 3DNLKE
results match well with experiment. The
mean stream wise velocity recovery along
the wake centerline is shown in
Figure 4.3. It can also be observed that
the experimental data showed some
scatter. Except for these locations, the
present 3DNLKE results are in good
agreement with experiment. However,
slower wake velocity recovery is
observed with 3DSKE model. at this
Reynolds number and beyond ineffective,
as variations in the span wise direction
are neglected. Figure 4.2 provides a Figure 4.2 Comparison of Wall Pressure Coefficient (Cp)
comparison of the wall pressure
coefficient with experimental and
different numerical simulations. The
LES results agree well with the
experimental results, but all our three
simulations (3DSKE, 2DNLKE, and
3DNLKE) show different trends. The
over prediction of stagnation-point
value by the SKE model is a well-
known fact and is attributed to the
steep velocity gradient on the
upstream side. The 2DNLKE and
3DNLKE models predict the
stagnation point correctly, but the
maximum negative pressure and
base pressure coefficient differ from Figure 4.3 Streamline Patterns at Four Different Time
the experimental results predicts the Intervals for one Vortex Shedding Cycle
trend.
 42

4.2.1.6 Conclusions
In the present work, 3-D unsteady computation of flow past a circular cylinder at subcritical Reynolds
number has been performed using a nonlinear κ-ε model to evaluate its applicability. The same test
case was simulated with 2DNLKE and its 3D counterpart 3DSKE to understand the effectiveness of
the present model. The bulk parameters and the wake velocity recovery match well with
experimental data and LES results. Since the grid requirement is not as severe as in LES and the
number of cycles required to do averaging is also less, computational cost associated with the present
model is very much less. For high-Re flows and flows encountered in practical engineering
applications, there is a restriction on the mesh size and the LES technique is prohibitively expensive.
Encouraging performance of the present NLKE model suggests that it could be used as an alternative
tool in these situations. Further improvement in the prediction may be possible by making the model
fully cubic form.

4.3 2nd Order Closures (Reynolds Stress Models - RSM)

Due to the increased complexity of this class of turbulence models, 2nd order closure models do not
share the same wide use as the more popular two-equation or algebraic models. The most
noteworthy efforts in the development of this class of models was performed by [Donaldson and
Rosenbaum (1968), Daly and Harlow (1970), and Launder , Reece, and Rodi (1975)]. The latter has
become the baseline 2nd order closure model, with more recent contributions made by [Lumley
(1978)], [Speziale (1985, 1987), and many other thereafter, who have added mathematical rigor to
the model formulation. The first three belong to so called 1st order closures. LRR R-ε and Launder-
Gibson R-ε models, calculates the anisotropic Reynolds stresses present in typical flows of complex
strain fields or significant body forces. RSM is the most complex of the turbulence models, and was
designed to address the problems of the κ-ε model (which cannot predict flows in long non-circular
duct because of the isotropic modelling of the Reynold’s stresses). The RSM can therefore accurately
account for the Reynold’s stress field directional effects. Because of the many Reynold’s stresses to
model, there are seven extra partial differential equations to solve, making computing costs very
high. The Reynolds Stress models, sometimes called stress-equation models, are enforcing to those
models which do not assume the Boussiqes assumption. Thus it would seems that these are perhaps
the ultimate models. Nevertheless these models still utilize approximation and assumption in the
modeling terms.

4.4 LES Model

The Large-Eddy Simulation (LES), divides the turbulent flow into two parts: large-scale and small-
scale motion. Large-scale motion is calculated in LES, while the small-scale motion needs to be
modeled because of the effects of large-
scale motion. The most important aspect
in application of LES is the use of suitable
sub-grid scale model. The implication to
[Kolmogorov's (1941)] theory of self-
similarity is that the large eddy of
simulation are dependent on geometry,
while the smaller scales more universal.
This feature allows one to explicitly solve
for the large eddies in a calculation and
implicitly account for the small eddies by
using a Sub-Grid-Scale (SGS) model. LES
always solves three-dimensional, time Figure 4.5 Difference Between the Filtered Velocity
dependent flow, calculating a mean of and the Instantaneous Velocity
time-dependent flow fields. Therefore, it is
 43

best suited for transient simulations. The main difference between conventional turbulence
modeling and LES is the averaging procedure. The LES technique does not involve the use of
ensemble average; rather it consists in applying a spatial filter to N-S equations.
Filter Definition
Mathematically, one may think of separating the velocity field into a resolved and sub-grid part. The
resolved part of the field represent the "large" eddies, while the sub-grid part of the velocity
represent the "small scales" whose effect on the resolved field is included through the sub-grid-scale
model49. This is called explicit filtering and Figure 4.5 illustrates the difference between the filtered
velocity ūi and the instantaneous velocity ux. formally, one may think of filtering as the convolution
of a function with a filtering kernel G:

𝐮𝐢 (𝐱⃗) = ∫ 𝐆(𝐱⃗ − 𝛏)𝐮(𝛏)𝐝(𝛏) ̅ 𝐢 + 𝐮′𝐢

𝐮𝐢 = 𝐮
Eq. 4.5
Where ūi the resolvable scale, and u’i is the sub-grid scale part. However, most practical (and
commercial) implementations of LES use the grid itself as the filter (the box filter) and perform no
explicit filtering. This is best shown in series of Figure 4.6 where L defines the largest eddies and Δ
is the box filter based on mesh size. It is important to note that the large eddy simulation filtering
operation does not satisfy the properties of a Reynolds operator. The expression most often used for
sub-grid scale part as a geometric mean is

Δ = L3 Δ = L /32 Δ = L /16

Figure 4.6 A velocity field produced by a (LES) of homogeneous decaying turbulence

 g  ( x  y  z )1/3 Eq. 4.6

or its generalization, the cubic root of the cell volume. In case of anisotropic grids, definition tends to
provide a fairly low value50. For this reason, the quadratic mean is used as following

49From Wikipedia.
50Jochen Frӧhlich , Dominic von Terzi,” Hybrid LES/RANS methods for the simulation of turbulent flows”,
Progress in Aerospace Sciences 44 (2008) 349– 377.
 44

1/2
 2x  2y  2z 
g    Eq. 4.7
 3 
 
which is advocated in some publications. Other authors favor the maximum

g  Max ( x ,  y ,  z ) Eq. 4.8

Filtered Incompressible N-S Equations

Using Einstein notation, the Navier–Stokes equations for an incompressible fluid in Cartesian
coordinates are:

u i u i u i u j 1 p  2ui
0 ,   ν Eq. 4.9
x i x j x j ρ x i x jx j

Filtering the Momentum equation results in

u i u i u j 1 p  2u i
  ν or
t x j ρ x i x jx j
Eq. 4.10
 u i u i u j 1 p 2 ui
  ν
t x j ρ x i x jx j

This equation models the change of time of the filtered variable u͞i . Since the unfiltered variable ui
are not known, it is impossible to directly calculate

𝛛𝐮𝐢 𝐮𝐣 ̅𝐢𝐮
𝛛𝐮 ̅𝐣
≠
⏟𝛛𝐱 𝐣 ⏟𝛛𝐱 𝐣
𝐍𝐨𝐭 𝐊𝐧𝐨𝐰𝐧 𝐊𝐧𝐨𝐰𝐧
Eq. 4.11
However the quantity on the right is known. Substituting:
 45

̅𝐢𝐮
̅ 𝐢 𝛛𝐮
𝛛𝐮 ̅𝐣 𝟏 𝛛𝐩 𝛛𝟐 𝐮̅𝐢 ̅̅̅̅̅
𝛛𝐮 𝐢 𝐮𝐣 ̅𝐢𝐮
𝛛𝐮 ̅𝐣
+ =− +𝛎 − −
𝛛𝐭 𝛛𝐱 𝐣 𝛒 𝛛𝐱 𝐢 𝛛𝐱 𝐣 𝛛𝐱 𝐣 ⏟𝛛𝐱 𝐣 𝛛𝐱 𝐣
𝛛𝛕𝐢𝐣
( 𝛛𝐱 𝐣 )
𝐰𝐡𝐞𝐫𝐞 𝛕𝐢𝐣 = ̅̅̅̅̅
𝐮 ̅𝐢𝐮
⏟𝐢 𝐮𝐣 − 𝐮 ̅𝐣
𝐒𝐆𝐒 𝐌𝐨𝐝𝐞𝐥𝐢𝐧𝐠
Eq. 4.12
The result is a set of LES equations. In addition, the filtered governing equation for a passive scalar φ
such as mixture fraction or temperature, can be written as

q ij

t


x j

u j 
 J
x j
x j
Eq. 4.13

Where the diffusive flux of and is the sub-filter stress tensor for the scalar. The filtered diffusive flux
is unclosed, unless a particular form is assumed for it (e.g. a gradient diffusion model ∂

q ij   u j  u j Eq. 4.14

And can similarly be split up into contributions from interactions between various scales. This sub-
filter tensor also requires a sub-filter model.

Numerical Methods for LES

Large eddy simulation involves the solution to the discrete filtered governing equations using CFD.
LES resolves scales from the domain size down to the filter size Δ and as such a substantial portion
of high wave number turbulent fluctuations must be resolved. This requires either high-order
numerical schemes, or fine grid resolution if low-order numerical schemes are used. [Pope]51
addresses the question of how fine a grid resolution is needed to resolve a filtered velocity field u(x).
[Ghosal]52 found that for low-order discretization schemes, such as those used in finite volume
methods, the truncation error can be the same order as the sub-filter scale contributions, unless the
filter widths considerably larger than the grid spacing. While even-order schemes have truncation
error, they are non-dissipative,53 and because sub-filter scale models are dissipative, even-order
schemes will not affect the sub-filter scale model contributions as strongly as dissipative schemes.

Implicit vs Explicit Filtering

The filtering operation in large eddy simulation can be implicit or explicit. Implicit filtering
recognizes that the sub-filter scale model will dissipate in the same manner as many numerical
schemes. In this way, the grid, or the numerical discretization scheme, can be assumed to be the LES

51 Pope, S. B. (2000). Turbulent Flows. Cambridge University Press.

52 Ghosal, S. (April 1996). "An analysis of numerical errors in large-eddy simulations of turbulence". Journal of
Computational Physics 125 (1): 187–206.
53 Randall J. Leveque (1992), “Numerical Methods for Conservation Laws (2nd Ed.)”, Birkhäuser Basel. ISBN 978-

3-7643-2723-1.
 46

low-pass filter. While this takes full advantage of the

grid resolution, and eliminates the computational cost
of calculating a sub-filter scale model term, it is
difficult to determine the shape of the LES filter that is
associated with some numerical issues. Additionally,
truncation error can also become an issue54. In
explicit filtering, an LES filter is applied to the
discretized Navier–Stokes equations, providing a
well-defined filter shape and reducing the truncation
error. However, explicit filtering requires a finer grid
than implicit filtering, and the computational cost
increases with [Sagaut]55 covers LES numeric in
greater detail. [Sarkar and Voke]56 carried out an LES
study of interactions of passing wakes and in flexional
boundary layer over a low-pressure turbine blade.
Figure 4.7 shows flow structures due to the complex
interactions of passing wakes and the separated shear
layer where the iso-surface of vorticity at an instant of
time through the wake passing cycle. Figure 4.7 Vorticity Prompted by the Wake
Passing Cycle
Quantitative Aspects of Comparison of RANS
vs LES Models
The biggest difference between LES and RANS is that, contrary to LES, RANS assumes that ūi = 0 (see
the Reynolds-averaged Navier–Stokes equations). In LES the filter is spatially based and acts to
reduce the amplitude of the scales of motion, whereas in RANS the time filter removes ALL scales of
motion with timescales less than the filter width. What distinguishes LES from RANS is the
definition of the small scales. While LES assumes the small scales to be smaller than the mesh
size Δ, RANS assumes them to be smaller than the largest eddy scale L. So the quality of LES model
is directly dependent in mesh size Δx. The mathematical similarity of LES and RANS equations, as you
can see in the equations (9.13), are being solved essentially the same. However, the physics are
different as evident in Eq. (9.13) using vector notation:

RANS
u
t
 
   ( u u )    ν  ν T   u   u
T
 
 p 
Eq. 4.15
LES
u
t
 
   (u u )    ν  ν SGS   u   u
T
 
 p 
It is obvious, the only change is in the dynamic viscosity determination, νT and νSGS. Or, the main
difference being that in RANS the unclosed term is a function of the turbulent kinetic energy and the
turbulent dissipation rate whereas in LES the closure term is dependent on the length scale of the
numerical grid. So in RANS the results are independent of the grid resolution!
and usually the DES needs more refine mesh that RANS. Another point of view is that RANS can only
give a time averaged mean value for velocity field since it is based on time averaging. In fact velocity

54 Grinstein, Fernando, Margolin, Len, Rider, William, “Implicit large eddy simulation”, Cambridge University
Press. ISBN 978-0-521-86982-9, 2007.
55 Sagaut, Pierre, “ Large Eddy Simulation for Incompressible Flows”, (3rd Ed.), Springer, 2006.
56 “Large-eddy simulation of unsteady surface pressure over a LP turbine due to interactions of passing wakes and

in flexional boundary layer” , Journal of Turbomachine, pp. 221–23, 2006.

 47

field in this method is averaged over a

time period of "Δt" which is
considerably higher than time constant
of velocity fluctuations. An example
RANS
would be the flow in backward step
using both RANS and DES models, (see
Figure 4.8) where the difference is
obvious. While DNS resolves all scales
of motion, all the way down to the
Kolmogorov scale, LES is next up and DES
resolves most of the scales, with the
smallest eddies being modeled. RANS is
on the other end of the spectrum from Figure 4.8 Time Averaged (RANS) vs Instantaneous (DES)
DNS, where only the large-scale eddies Simulation over a backup step
are resolved and the remaining scales
are modeled.

Motivations for Coupling Methods Between RANS and LES

As indicated previously, (RANS) models provide results for mean quantities with engineering
accuracy at moderate cost for a wide range of flows. In other situations, dominated by large-scale
anisotropic vortical structures like wakes of bluff bodies, the average quantities are often less
satisfactory when a RANS model is employed. Then large eddy simulation (LES) performs generally
better and bears less modeling uncertainties. Furthermore, LES by construction provides unsteady
data that are indispensable in many cases: determination of unsteady forces, fluid–structure
coupling, identification of aerodynamic sources of sound, and phase-resolved multiphase flow, to
name but a few issues [Frohlich & Terzi]57. Unfortunately, LES is by a factor of 10 to 100 more costly
than RANS computations; LES requires a finer grid, cannot benefit from symmetries of the flow in
space, and provides mean values only by averaging the unsteady flow field computed with small time
step over a long sampling time. Hence, it seems natural to attempt a combination of both turbulence
modeling approaches and to perform LES only where it is needed while using RANS in regions where
it is reliable and efficient.
Another and somewhat different motivation for LES/RANS coupling stems from wall-bounded flows.
Close to walls, the LES philosophy of resolving the locally most energetic vortical structures requires
to substantially reduce the step size of the grid since the dominating structures become very small in
this region. Furthermore, when increasing the Reynolds number, the scaling of the computational
effort is similar to that of a DNS in its dependence on Re just with a smaller constant. That makes the
approach unfeasible for wall-bounded flows at high Re, such as the flow over a wing . As a remedy,
some sort of wall model can be introduced to bridge the near-wall part of the boundary layer and to
make the scaling of the required number of grid points independent of Re. A RANS model depends
on physical quantities describing the entirety of the turbulent fluctuations. For the sequel it is
necessary to define the specifics of LES models and RANS models. Using an unsteady definition of a
Reynolds average as discussed above, the transport equations for the Reynolds-averaged velocity
<ui> read:

57Jochen Frӧhlich , Dominic von Terzi,” Hybrid LES/RANS methods for the simulation of turbulent flows”,
Progress in Aerospace Sciences 44 (2008) 349– 377.
 48

 ui  ui u j p   ν u i  τ ijRANS
   
t x j x j x j  x j 
 x j
Eq. 4.16
u i u i u j p   νu i  τ ij
DES

   
t x j x j x j  x j  x j

For example, the κ–ε model determines

  ui   u 
τ ijRANS  f  , κ, ε, C  , τ ijDES  f  i , Δ g , C  Eq. 4.17
 x i   x i 
where C is a model constant, κ the turbulent kinetic energy, and ε the turbulent dissipation rate. The
latter two are determined from other relations. For LES based on the Smagorinsky model uses a
relation like where Δg is a length scale related to the numerical grid, Since there exist many variants
of LES and RANS models we define the following: a model qualifies as an LES model if it explicitly
involves in one or the other way the step size of the computational grid. RANS models, in contrast,
only depend on physical quantities, including geometric features like the wall distance.

4.4.6.1 Principal Approaches to Coupling LES with RANS

The similarity of the equations and the considerations in Section 2.4 suggest the concept of unified
modeling. This approach is based on using the same transport equation for some resolved velocity
ūi, yet to be specified in its meaning:

u i u i u j p   νu i  τ ijModel

  
t x j x j x j  x j 
 x j Eq. 4.18

A transition from LES to RANS can be achieved in several ways. One possibility is blending, i.e. by a
weighted sum of a RANS model and an LES the models according to

 ijModel  f RANS ijRANS  f LES ijLES Eq. 4.19

In this fashion, f RANS and f LES are local blending coefficients determined by the local value of a given
criterion. Another strategy is to use a pure LES model in one part of the domain and a pure RANS
model in the remainder, so that a boundary between a RANS zone and an LES zone can be specified
at each instant in time. The transport equation for the velocity, however, is the same in both zones
with no particular adjustment other than switching the model term at the interface. This way the
computed resolved velocity is continuous. We term this strategy Interfacing LES and RANS.
Furthermore, if the interface is constant in time, it is called a hard interface. If it changes in time
depending on the computed solution, it is termed a soft interface.
Segregated modeling is the counterpart to unified modeling as LES is employed in one part of the
computational domain, while RANS is used in the remainder. With segregated modeling, however,
the resolved quantities are no more continuous at the interfaces. Instead, almost stand-alone LES and
RANS computations are performed in their respective subdomains which are then coupled via
 49

appropriate boundary conditions. Except for laminar flows, the solution is discontinuous at these
interfaces. This avoids any gradual transition in some gray area characteristic of unified turbulence
models. Segregated modeling allows for embedded LES by designing a configuration where in an
otherwise RANS simulation a specific region is selected to be treated with LES with full two-way
coupling between the zones58.

4.4.6.2 Results and Discussion for Segregated Modeling

As an example of segregated modeling, is integrated RANS-LES of the NASA stage 35 compressor and
a diffuser. While the compressor stage is computed with the RANS flow solver, the subsequent
diffuser with the LES flow solver (see Figure 4.9). Here, we look at the vorticity magnitude
distribution at the 50% clip plane of the stator. Again, we can identify the wakes of the stator passing
the interface. The different description of turbulence in the two mathematical approaches is
apparent. While on the RANS side the
turbulence is modeled in a turbulence
model and cannot be seen in the
vorticity distribution, on the LES side
the fine scale turbulence is regenerated
and can be identified as small-scale
structures in the LES solution59.
Another example in the feasibility of a
hybrid RANS–LES approach is the
numerical simulation of aircraft wing-
tip vortices which has been studied by
[Kolomenskiy et al]60. Mesh sensitivity
tests of our RANS solver and
comparisons between two different
turbulence models indicate that the
RANS approach adequately describes Figure 4.9 Integrated RANS-LES Computations in Gas
the flow upstream from the trailing Turbines: Compressor-Diffuser,
edge, but overestimates the rate of
decay of the wing-tip vortex. A hybrid RANS–LES method is presented that results in a better
agreement with the wind tunnel experiment for numerical simulation of the wake of an airliner.

4.4.6.3 Coupling in RANS/LES Interface

A fully coupled solution requires that all flow variables must be exchanged at the interface. When
some engine components are computed with LES and others with RANS approximations will have to
be made to couple instantaneous and averaged variables. To simplify the problem, we consider only
the one-way coupling of the velocity and turbulence variables. One-way coupling means that
information is passed only downstream; the variables at the inlet of the downstream domain are
computed from the variables at the outlet of the upstream domain. For the RANS/LES interface,
turbulent fluctuations need to be added to the velocity from an upstream RANS computation. For the
LES/RANS interface, such as the combustor/turbine interface, a simple time average of the velocity
provides a mean velocity at the inlet of the RANS domain. This velocity distribution is again highly

58 Jochen Frӧhlich , Dominic von Terzi,” Hybrid LES/RANS methods for the simulation of turbulent flows”,
Progress in Aerospace Sciences 44 (2008) 349– 377.
59 J. U. Schlüter, X. Wu, S. Kim, J. J. Alonso, and H. Pitsch, AIAA-2004-369, 42nd Aerospace Sciences Meeting and

Exhibit Conference, January 2004.

60 Dmitry Kolomenskiy, Roberto Paoli, and Jean-Franc¸ois Boussuge, “Hybrid Rans–Les Simulation of Wingtip

Vortex Dynamics”, FEDSM2014-21349.

 50

non-uniform, which allows to

describe turbulence at the inlet with
the local turbulence generation from
the mean velocity.
The flow in the compressor is
computed with unsteady RANS using
the k-ω model and the flow in the
combustor is computed with LES. The
last blade row of the compressor is a
row of stators and does not rotate.
The blade row upstream is a row of
rotors that rotate counter-clockwise.
The mean flow is highly complex,
with the wakes originating from the
last stage of the compressor (see Figure 4.10 Compressor and combustor: RANS and LES axial
Figure 4.10). These wakes are velocity, mid-passage (Courtesy of Medic et al.)
unsteady due to the rotation of the
rotors in the compressor. Larger values of turbulent kinetic energy are in the regions with strong
velocity gradients. The large values of k near the hub might be spurious; they are highly dependent
on the quality of the grid. This illustrates the fact k from the k-ω model usually fails to accurately
represent the true turbulent kinetic energy in complex flows. The flow field in the downstream LES
domain is highly dependent on the conditions at the inlet. To generate an inflow profile for the LES
in the combustor, the mean velocity at the combustor inlet is set equal to the RANS velocity at the
compressor exit and appropriate fluctuations need to be added [Medic, et al]61.
The aim of segregated modeling is to compute all models in their regime of validity: steady RANS for
flows with stationary statistics and unsteady LES with high resolution where it is needed. Therefore
one can choose the best suited method for each subdomain without considering their compatibility
and without fear of inconsistencies in their use. Furthermore, any gray zone where the model is left
alone with generating fluctuations in some transition process is avoided. The price to pay is the need
for comparatively complex coupling conditions. For block-structured solvers, however, the routines
for data exchange required anyway facilitate a straightforward implementation. Inappropriate
coupling conditions lead to contamination of the results in the LES or RANS subdomains. Depending
on the type of the interface, the requirements on the coupling conditions (inlet and outlet) differs
[Frohlich & Terzi]62

4.5 DES Model

An alternative between LES and RANS models would the Detached Eddy Simulations (DES) which
attempts to combine the best aspects of RANS and LES methodologies in a single solution strategy.
This is more pronounce in near-wall regions in a RANS-like manner, and treat the rest of the flow in
an LES-like manner. The model was originally formulated by replacing the distance function d in the
Spalart-Allmaras (SA) model with a modified distance function D as:

D  Min (d, CDES , ) Eq. 4.20

61 G. Medic, D. You AND G. Kalitzin, “An approach for coupling RANS and LES in integrated computations of jet
engines”, Center for Turbulence Research, Annual Research Briefs, 2006.
62 Jochen Frӧhlich , Dominic von Terzi,” Hybrid LES/RANS methods for the simulation of turbulent flows”,

Progress in Aerospace Sciences 44 (2008) 349– 377.

 51

Where CDES is a constant and Δ is 13 M cells

the largest dimension of the grid
cell in question. This modified
distance function causes the model
to behave as a RANS model in
regions close to walls, and in a
Smagorinsky-like manner (LES)
away from the walls. This is usually
justified with arguments that the
scale-dependence of the model is
made local rather than global, and
that dimensional analysis backs up
this claim. In summary, an example
16 M cells
of a hybrid technique, detached
eddy simulation (DES) is a
modification of a RANS model in
which the model switches to a sub-
grid scale formulation in regions
fine enough for LES calculations.
Therefore, the grid resolution is not
as demanding as pure LES, there by
considerably cutting down the cost
of the computation. A study in CD
Adapco® shows that for wake flow
drag calculation, DES is more
accurate than RANS and highly
dependent in grid resolution (see Figure 4.11 DES and effect of Grid Density on the Wake flow
Figure 4.11, top 13 M cells vs
bottom-16 M cells). Another example would be the presentation of flow in rectangular ogive fore-
body (Aircraft fore body)63. In that environment we set close to the wall D is by the wall parallel
spacing’s, i.e., D = d; d = << Δ for RANS and away the wall CDESΔ < D, i.e., D = CDES Δ for LES. Using an
unstructured generated using VGRID. Simulation details can be found in [Viswanathan, Squires and
Forsythe 2006] where the results shown in Figure 4.12.

 Re = 2.21 x 106, Mach number = 0.21

 Grid sizes from 2.1 x 106 cells to 8.75 x 106 cells

63Kyle D. Squires, Les Application in Aerodynamics”, School of Mechanical, Aerospace, Chemical and Materials
Engineering Arizona State University Tempe, Arizona, USA.
 52

 Angle of attack: 60 degrees

Figure 4.12 Comparison of Pressure contours on planner cuts for RANS and DES models

4.6 DNS Model

The DNS is a research tool, and not a brute-force solution to the Navier-Stokes equations for
engineering problems. It is a method which is CPU intensive, as it tries to resolve N-S equations
without any turbulence approximation, requiring a very fine numerical resolution to capture all the
details of turbulence. The theory is that the whole range of spatial and temporal scales of the
turbulence must be resolved. All the spatial scales of the turbulence must be resolved in the
computational mesh, from the smallest dissipative scales (Kolmogorov microscales), up to the
integral scale L, associated with the motions containing most of the kinetic energy. It is not hard to
prove that for turbulent Reynolds number, we have:

u L
N 3DNS  Re9/4 Re  Eq. 4.21


Where N is the number of points along a given mesh direction with increments h and ú is the root
 53

mean square (RMS) of velocity. Hence, the memory storage requirement in a DNS grows very fast
with the Reynolds number64. In addition, given the very large memory necessary, the integration of
the solution in time must be done by an explicit method. This means that in order to be accurate, the
integration, for most discretization methods, must be done with a time step, Δt, small enough such
that a fluid particle moves only a fraction of the mesh spacing h in each step. Therefore, it remains
limited to very simple cases. Filtering requires a finer grid than implicit filtering, and the
computational cost increases with [Sagaut (2006)] covers LES numeric in greater detail65. [Sarkar
and Voke]66 carried out an LES study of interactions of passing wakes and in flexional boundary layer
over a low-pressure turbine blade and Figure 4.7 shows flow structures due to the complex
interactions of passing wakes and the separated shear layer. Evidently, as it appears in Eq. (9.2) and
above discussion, the DNS method is highly dependent to Reynolds number which will be debated in
the next section.

The Reynolds Number Constraint

Turbulence contains eddies with a wide range of sizes. These eddies interact with each other in a
non-linear fashion through their induced velocity fields, changing the orientation and shape of their
neighbors. As first described by Richardson (1922) and quantified by Kolmogorov (1941), the net
effect of this change-of-shape (i.e. straining) process is to ‘cascade’ kinetic energy from the largest to
the smallest scales of the turbulence. As a result, the largest eddies are the most energetic, and their
size, shape and speed are set by the details of the flow configuration, and not directly affected by the
viscosity of the fluid. The size of the smallest eddies, on the other hand, is determined both by how
much energy enters the cascade at the large scales and by the viscosity. The primary role of viscosity
is to define the scale at which the energy is dissipated. Therefore, the Reynolds number of the flow is
determines how small the smallest scales are, relative to the largest eddies67. This behavior, known
as Reynolds-number similarity, can be observed in Figure 4.13, which presents DNS results from
the same boundary-layer flow at two Reynolds numbers that differ by a factor of two. This illustrates
the challenge faced by DNS, which must use a domain large enough to comfortably include the largest
naturally-occurring eddies while using a grid fine enough to fully resolve the dissipation scales. Using
today’s most capable computers, this can only be done for Reynolds numbers orders of magnitude
smaller than found, for example, in full-scale aeronautical flows.

Figure 4.13 Vorticity contours from spectral DNS at two Reynolds numbers (Spalart et al. 2008).

64 From Wikipedia.
65 Sagaut, Pierre (2006). “Large Eddy Simulation for Incompressible Flows”, (3rd ed.), Springer.
66 “Large-eddy simulation of unsteady surface pressure over a LP turbine due to interactions of passing wakes and

in flexional boundary layer”, J Turbo machine, 128 (2) (2006), pp. 221–231
67 Gary N. Coleman and Richard D. Sandberg, “A Primer on Direct Numerical Simulation of Turbulence – Methods,

Procedures and Guidelines”, Technical Report AFM-09/01a - March 2010.

 54

Numerical Considerations
The obligation of having to resolve all spatial and temporal scales of the turbulence requires that
numerical errors be monitored and controlled. As a result, DNS has historically not used commercial

Figure 4.14 The sound radiated by a Mach 1.9 circular jet

CFD packages, but specially-written codes, optimized for the flow-types of interest68. The need for
DNS algorithms to be efficient; that is, to have a high ratio of accuracy to computational cost which is
particularly important. There are a number of strategies that DNS codes have employed to do this,
including finite-volume, finite-element, and discrete-vortex as well as B-spline methods. But central
to development of an efficient algorithm for DNS are two methods; Spectral and Finite Difference.

4.6.2.1 Spectral Methods

Spectral methods approximate the flow variables as linear combinations ‘expansions’ of global basis
functions that involve complex-exponential or orthogonal polynomial Eigen solutions of an
appropriate problem. (See Figure 4.13). An exciting new development has been the field of
computational aeroacoustics, where both the fluid motion and the sound it radiates are directly
computed (see Tam69, Lele70). Computational aeroacoustics is still in its infancy; the sound from a
canonical flow such as a perfectly expanded supersonic jet has only just been computed [Freund, Lele
& Moin] as a unpublished information71. Figure 4.14 shows results for a Mach 1.9 jet from their
massive computations using 640x270x128 mesh points. Contours of vorticity (black) are overlaid on
dilatation contours (gray). The computations were performed by [ Freund, Lele & Moin].

68 Gary N. Coleman and Richard D. Sandberg, “A Primer on Direct Numerical Simulation of Turbulence –
Methods, Procedures and Guidelines”, Technical Report AFM-09/01a (March 2010).
69 TamCKW. “Computational aeroacoustics: issues and methods”, AIAA pap. 95–0677.
70 Lele SK. 1992, “Compact finite difference schemes with spectral-like resolution”, J. Comp. Phys. 103:16–42.
71 Parviz Moin and Krishnan Mahesh, “ Direct Numerical Simulation: A Tool in Turbulence Research”, Annu. Rev.

Fluid Mech. 1998. 30:539–78. Copyrightc 1998 by Annual Reviews Inc.

 55

4.6.2.2 Finite Differencing

Because of their ease of implementation, suitability to parallelization, and possible high-order
accuracy, finite-difference (FD) schemes for DNS have become increasingly popular, especially for
the emerging area of computational aeroacoustics (CAA) (see Figure 4.15 where Instantaneous
contours of stream wise density gradients from high-order finite-difference DNS of a supersonic
axisymmetric wake at Mach 2.46 and Re= 100, 000 (based on freestream velocity and diameter of
body; Sandberg 2008). A wide range of options is available. Low-order FD methods allow complex
geometries and irregular grids, and in this sense are similar to finite-volume methods. The
computational efficiency of low-order (especially upwind) FD approximations (FDA) is often
unacceptable, requiring many more grid points than a spectral method to achieve the same accuracy.

Figure 4.15 Instantaneous Contours of Stream-Wise Density Gradients from DNS

4.7 Strategies for Turbulence Modelling

According to [P.R. Spalart]72 turbulence predictions in aerodynamics face two principal challenges:
(I) growth and separation of the boundary layer, and (II) momentum transfer after separation.
(I) is simpler, but makes very high accuracy demands, and appears to give models of higher
complexity little advantage. (II) is now the arena for complex RANS models and the newer strategies,
by which time-dependent three-dimensional simulations are the norm even over two-dimensional
geometries. In some strategies, grid refinement is aimed at numerical accuracy; in others it is aimed
at richer turbulence physics. In some approaches, the empirical constants play a strong role even
when the grid is very fine; in others, their role vanishes. For several decades, practical methods will
necessarily be RANS, possibly unsteady, or RANS/LES hybrids, pure LES being unaffordable. Their
empirical content will remain substantial, and the law of the wall will be particularly resistant.
Estimates are offered of the grid resolution needed for the application of each strategy to full-blown
aerodynamic calculations, feeding into rough estimates of its feasibility date, based on computing-
power growth. The numerical strengths of CFD increase by the year thanks to the progress of
computers, whereas turbulence modelling almost stagnate. A primary purpose here is to provide a
viewpoint on these relatively new, evolving and misunderstood methods, besides predicting that
they will proliferate and make a major contribution.

72P.R. Spalart, “Strategies for turbulence modelling and simulations”, International Journal of Heat and Fluid
Flow 21 (2000) 252±263.
 56

Physical Aspects (modification to simple RANS Models)

The field of classical RANS turbulence modelling is active. We are referring here primarily to eddy-
viscosity models. Improvements will be made to the simpler transport models, typically by adding
new empirical terms aimed at compressibility, streamline curvature or better anisotropy of the
Reynolds-Stress tensor (non-Boussinesq constitutive relations). Which can, for instance, create
secondary flows of the second kind in a square pipe [Speziale]73. An example is given in Figure 4.16.
The SA model was used with the non-linear constitutive relation, as a first attempt to improve on the
eddy-viscosity approximation. The SA model exceeded expectations, and is now useful to a rather
large user base. The model is not sophisticated, but it is not inaccurate.

Complex RANS Models

The DES on two grids also depicts the inclusion of smaller eddies allowed by the finer grid. The flow
is at a Reynolds number of 50,000. It led us to propose Detached-Eddy Simulation (DES), a further
step in the hybridization of LES (Spalart et al., 1997). The idea is to entrust the whole boundary layer
(populated with ``attached'' eddies) to a RANS model, and only separated regions (“detached''
eddies) to LES. It is aimed primarily at external flows. solutions are under-resolved in the separated
regions.
Figure 4.17 illustrates the difference in
description for the flow past a circular cylinder.
The runs on the right are DES, but the visible
eddies are in the LES region. Of course the DES
figures show only a plane out of the 3D domain.
Each step from SRANS through 2D URANS to
LES/DES adds a dimension: first the time and
then the third space dimension. The cost increase
is of an order of magnitude each time. In return,
the SRANS drag is too low at Cd ≈ 0.9, the URANS
drag is much too high at Cd ≈1.7, and the DES drag
is in much better agreement with experiment,
although grid effects are still solutions are under-
resolved in the separated regions. present: Cd ≈
1.05 on the coarse grid but 1.32 on the fine grid
[Travin et al.]74. The experiment gives 1.2. For a
compete discussion, please refer to [P.R. Figure 4.16 Stream Lines in Square Channel
Spalart]75. with Nonlinear Constitutive –( Courtesy of
[speziale, 1987])

73 Speziale, C.G.,” On nonlinear K- l and K - ԑ models of turbulence”. J. Fluid Mech. 178, 459, 1987.
74 Travin, A., Shur, M., Strelets, M., Spalart, P.R., 2000. “Detached-eddy simulations past a circular cylinder”.
Flow, Turb. Combust, to appear.
75 P.R. Spalart, “Strategies for turbulence modelling and simulations”, International Journal of Heat and Fluid

Flow 21 (2000) 252±263.

 57

Figure 4.17 Simulation of Flow Past Circular Cylinder by Various Approaches (Shur et al., 1996;
Travin et al., 2000)

Role of Grid Refinement

In RANS, the equations possess a smooth exact solution, and the numerical solution approaches that
solution as we refine the grid. The aim of grid refinement is numerical accuracy. In contrast, in LES
as it is practiced and in DES, the Sub-Grid-Scale (SGS) model adjusts to the grid so that the smallest
resolved eddies match the grid spacing. In a finer grid, resolving eddies to a smaller size gives the
large energy-containing eddies more eddies for genuine nonlinear interactions, making them more
accurate. The aim of grid refinement is now physical instead of numerical, to use simple words. This
distinction is tracked in Table 4.1 (several methods had to be labelled “hybrid”, as their aim is
different in different flow regions). Another description of it is that when the aim is numerical, the
turbulence model does not depend on the grid spacing but when the aim is physical, it does. A
consequence is that in URANS, no amount of grid refinement will override the influence of the
empirical content of the turbulence model. In contrast, in a method with the ``physical'' aim, grid
refinement weakens the role of the modelled eddies and thus improves the fidelity of the simulation.
It has been proposed not to automatically link the width of the LES filter and the grid spacing, in order
to obtain solutions of the filtered equations free of significant numerical errors. Then, the numerical
errors remain the same fraction of the SGS kinetic energy, and it is fair to write that “numerical and
SGS effects cannot be separated”. This situation is disturbing to some careful people, who would prefer
to understand the physical system of the filtered equations, and then obtain very accurate solutions
to it. Widening the filter raises the magnitude of the sub-grid is effect can be offset by the reduction
of numerical errors, or even by the design of a superior stresses, which are notoriously inaccurate
locally. It is unlikely that SGS model, possibly gained from the better physical understanding. On a
given grid, an LES with a wide filter would cost almost the same as one with a narrow filter: the cost
 58

per step would be the same, and the time step could only rise by a modest amount. Conclusive tests
of the filter-grid relationship would crucially depend on the definition of a figure of merit.

Re-
Name Aim Unsteady Empiricism Grid Size
dependence
2DURANS Numerical Yes Weak Strong 105
3DRANS Numerical No Weak Strong 107
3DURANS Numerical Yes Weak Strong 107
DES Hybrid Yes Weak Strong 108
LES Hybrid Yes Weak Weak 1011.5
QDNS Physical Yes Strong Weak 1015
DNS Numerical Yes Strong None 1016

Table 4.1 Summery of Turbulence Strategies

Outlook
Progress in numerical methods and computers is intensifying the challenge for turbulence
treatments, to provide a useful level of accuracy in slightly or massively separated flows over fairly
complex geometries at very high Reynolds numbers. This is desirable in the near future, especially
the jet-engine industry. In addition, the needs of non-specialist users and automatic optimizers
dictate a very high robustness. Flows with shallow or no separation appear to be within the reach of
the current steady RANS methods or their finely calibrated derivatives, incorporating modest
improvements such as nonlinear constitutive relations. For such flows, transition prediction with
generality, accuracy, and robustness may prove more challenging than turbulence prediction. With
massive separation, it appears possible we will give up RANS, steady or unsteady. This will probably
be the major debate of the next few years. The alternative is a derivative of LES, in which the largest,
unsteady, geometry-dependent eddies are simulated and (for most purposes) “discarded” by an
averaging process. We have to balance our ambitions with cost considerations, and a table brief the
issue was tentatively provided.
A major consideration is whether LES is practical for the entire boundary layer, and it was strongly
argued that this will not be the case, in the foreseeable future. This forces hybrid methods, with quasi-
steady RANS in the boundary layer. In this paper, LES was effectively defined as a simulation in which
the turbulence model is tuned to the grid spacing, and RANS as the opposite. Other more subtle
definitions probably exist, but this one seems to classify almost all the studies to date. Speziale's
hybrid proposal involves the grid spacing and the Kolmogorov length scale but, surprisingly, not the
internal length scale of the RANS turbulence model; thus, it is difficult to classify [Speziale]76. The
proposal of [Aubrun et al.]77. is also hybrid, as it leads to combining “modelled” and “resolved”
Reynolds stresses, but the modelled stresses do not scale (and vanish) with the grid spacing as they
do in LES 78. Variations on the now-running DES proposal clearly have a wide window of opportunity.

76 Speziale, C.G., “Turbulence modeling for time-dependent RANS and VLES”, a review.AIAA J. 36 (2), 173, 1998.
77 Aubrun, S., Kao, P.L., HaMinh, H., Boisson, H, “The semi deterministic approach as way to study coherent
structures. Case of a turbulent flow behind a backward-facing step”, Proceedings of the Fourth International
Symposium on Engineering Turbulence Modelling and Measurements, Corsica, Elsevier, Amsterdam, 1999.
78 See Previous.
 59

The plausible spread of hybrid methods highlights the permanence of a partnership between
empiricism and numerical power in turbulence prediction at full-size Reynolds numbers. This
demands a balance in funding and in publication space. Since hybrid methods offer flexibility when
setting the boundary between “RANS region” and “LES regions”, the more capable the RANS
component is, the lower the cost of the hybrid calculation will be. Therefore, the switch to LES in
some regions does not remove the incentive to further the RANS technology. This scene also raises
the issue of which core of experiments and DNS will be the foundation of the empirical component of
the system. As ever, we will need simple flows for calibration of the RANS sub-system, and more
complex flows for validation of the full CFD system.

4.8 Modeling To Choose From

To recap, there are in general four general categories of turbulence models to choose from, They are:

 Reynolds-Averaged Navier-Stokes (RANS): Choosing an appropriated models, the

governing equations are solved in ensemble-averaged form, including appropriate models
for the effect of turbulence.
 Large Eddy Simulation (LES): Large turbulent structures in the flow are resolved by the
governing equations, while the effect of the sub-grid scales (SGS) are modelled. The scale
separation is obtained by applying a filter to the governing equations which also influences
the form of the SGS models.
 Detached Eddy Simulation (DES): Hybrid method that treats near-wall regions with a RANS
approach and the bulk flow with an LES approach, (CPU intensive).
 Direct Numerical Simulation (DNS): Resolves all scales of turbulence by solving the Navier-
Stokes equations numerically without any turbulence modelling, (CPU intensive).
 60

5 Classification of Turbulence Models

In general, the approach for solving turbulent flow equations can roughly be divided into four classes
as Direct Numerical Simulation (DNS), Reynolds Average N-S equations (RANS), Large Eddy
Simulation (LES) and Detached Eddy Simulation (DES), as 79:

1. RANS-based turbulence Models

a. Linear Eddy Viscosity models
i. Algebraic Models
1. Cebeci-Smith model
2. Baldwin-Lomax model
3. Johnson-King model
4. A roughness-dependent model
ii. One equation models
1. Prandtl's one-equation model
2. Baldwin-Barth model
3. Spalart-Allmaras model
4. Rahman-Siikonen-Agarwal Model
iii. Two equation models
1. κ-ε models
a. Standard κ-ε model
b. Realizable κ-ε model
c. RNG κ-ε model
2. κ-ω models
a. Wilcox's κ-ω model
b. Wilcox's modified κ-ω model
c. SST κ-ω model
b. Nonlinear Eddy Viscosity models
i. Explicit nonlinear constitutive relation
1. Quadratic κ-ε
2. Cubic κ-ε
3. Explicit Algebraic Reynolds Stress Models (EARSM)
ii. v2-f models
1. v2-f Model
2. ζ-f Model
c. Reynolds Stress Model (RSM)
2. Large Eddy Simulation (LES)
a. Smagorinsky-Lilly model
b. Dynamic sub-grid-scale model
c. RNG-LES model
d. Wall-adapting local eddy-viscosity (WALE) model
e. Kinetic energy sub-grid-scale model
f. Near-wall treatment for LES models
g. Structural modeling
3. Detached Eddy Simulation (DES)
4. Direct Numerical Simulation (DNS)

79 Wikipedia.
 61

5.1 Comparisons of various Turbulence Models

In the previous sections, we have presented turbulence models that are commonly encountered in
commercial CFD codes. In these sections the physical and mathematical principles underlying the
turbulence models were presented together with discussions about their limitations. An overview of
turbulence models, sorted by descending complexity, depicted on Table 5.1 which gives a short
summary of the advantages and shortcomings of these models80. In general the turbulence models
are developed to predict velocities accurately. The parameters in the models, (e.g., k and ε), may very
well be off by a factor of 3. Using these parameters in other models, e.g. for mixing or bubble break-
up, should be done with the awareness that the parameters do not have exact physical relevance but
only show the trend.

80 Cambridge University Press

 62

Table 5.1 Advantages and Disadvantages of different Turbulence Models

Turbulence
Models Advantages Disadvantages

Zero-equation When ∂U/∂y = 0 ⇒ μ T = 0. Lack of

models, e.g. Cost-effective model applicable for a transport of turbulent scales. Estimation
mixing-length limited number of flows. of the mixing length is difficult. Cannot be
Model used as a general turbulence model.
One-equation The use of an algebraic equation for the
Cost-effective model applicable for a
models, e.g. length scale is too restrictive. Transport
limited number of flows.
k-algebraic Model of the length scale is not accounted for.
Complete models in the sense that velocity
Two-equation
and length scales of turbulence are Limited to an eddy-viscosity assumption.
models, k–ε
predicted with transport equations. Good Turbulent viscosity is assumed to be
Group standard,
results for many engineering applications. isotropic. Convection and diffusion of the
RNG, Realizable, k–
Especially good for trend analysis. Robust, shear stresses are neglected.
ω and Mentor SST
economical and easy to apply.
Not good for round jets and flows
The most widely used and validated involving significant curvature, swirl
Standard k–ε Model
model. (isentropic), sudden acceleration,
separation and low-Re regions.
Modification of the standard k–ε model
Not as stable as the standard k–ε model.
RNG k–ε Model gives improved simulations for swirling
Not suited for round jets.
flows and flow separation
Modification of the standard k–ε model
Realizable k–ε gives improved simulations for swirling
Not as stable as the standard k–ε models
Model flows and flow separation. Can also handle
round jets.
Works well at low Re. Does not need wall
Needs fine mesh close to the wall, first
κ–ω Model functions. Works well with adverse
grid point at y+ < 5.
pressure gradients and separating flow.
Uses k–ε in the free stream and k–ω in the Needs fine mesh close to the wall. Over
wall-bounded region. Works well with predicts turbulence in regions with large
adverse pressure gradients and separating normal strain, e.g. stagnation regions and
SST Model
flow. Many authors recommend that the regions with strong acceleration, but is
SST model should replace the k–ε model as better than k–ε.
the first choice.
Applicable for complex flow where the Computationally expensive with 11
turbulent-viscosity models fail. Accounts transport equations. Several terms in the
Reynolds Stress
for anisotropy. Good performance for many transport equations must be closed. Poor
Models, (RSM)
complex flows, e.g. swirl, flow separation Performance for some flows due to the
and planar jets. closures introduced in the model.
Applicable to complex flows. Gives High computational cost. Large amount of
Large-eddy information about structures in turbulent data that must be stored and post-
Simulation (LES) flows. And a lot of information that cannot processed. Difficult to find proper time-
be obtained otherwise. resolved boundary conditions for flow.
No turbulence models are introduced. Extreme computational cost for practical
Useful at low Re Numbers, especially for engineering flow Simulations. Huge
DNS
gaseous flows. Useful to develop and amount of data
validate turbulence models
 63

5.2 Numerical Considerations

Modern turbulence model equations pose special numerical difficulties that must be understood in
order to obtain reliable numerical solutions, even for boundary-layer flows where the equations are
parabolic. For one-equation, two-equation and second-order closure models, these difficulties
include stiffness caused by the presence of an additional time scale, singular behavior near solid
boundaries, and non-analytical behavior at sharp turbulent/non-turbulent interfaces . This chapter
focuses on these difficulties and on the solution methods for turbulence-model equations that have
evolved.

Elementary Time-Marching Methods

The examples of time marching methods and their effect on solution will not be discussed here since
it is already detailed in average flow analysis. Just a mentioned method will be suffice here.

 Explicit Schemes such as DuFort-Frankel, Godunov, Lax-Wendroff and Mac-Cormack

methods.
 Implicit methods such as Crank-Nicolson, Euler, and Alternating Direction Implicit (ADI)
schemes.

5.2.1.1 Block-Implicit Methods

The most efficient numerical methods currently available for complex flow fields are block-implicit
methods. They differ from elementary implicit methods in one very important respect. Specifically,
when an elementary implicit scheme is applied to a coupled set of equations, each equation is solved
in sequence. In the context of a system of equations, this is usually referred to as a sequentially-
implicit method. By contrast, a block-implicit scheme solves all of the equations simultaneously at
each grid point. The block-implicit formulation, generally requiring inversion of block-tridiagonal
matrices, entails more computational effort than a sequentially-implicit method. The additional
computation at each grid point and time step is usually compensated for by a dramatically improved
convergence rate. Block-implicit solvers can achieve CFL numbers in excess of 100, and often
converge in 100 to 200 time steps for flows including boundary-layer separation. For example, using
a block-implicit method, a supersonic two-dimensional shock-separated turbulent flow can be
simulated on an 80486-based microcomputer in about 3 hours of CPU time [Wilcox (1991)] . On the
same computer, a similar computation would take about 25 hours using a sequentially-implicit
method [Wilcox (1990)] and 75 hours using an explicit method [Wilcox (1974)].
As in the preceding section, we begin with a brief overview of block implicit methods. For simplicity,
we focus on a one-dimensional system. The primary concern in this section is, of course, upon how
turbulence model source terms impact such methods. Now, suppose we choose to use a two-equation
turbulence model to determine the Reynolds stress. The following three points that must be
considered in modifying a block-implicit solution scheme.

1. Decide whether to solve all equations simultaneously or to solve the model equations and
mean-flow equations sequentially.
2. If the preferred option is to solve all equations simultaneously, determine the changes to the
flux-Jacobian matrices.
3. Make provision for handling source terms.

In principle, solving all equations simultaneously will yield the most rapidly convergent scheme in
the number of iterations, but not necessarily in CPU time. However, the coupling between the
turbulence-model equations and the mean-flow equations appears to be relatively weak. The
primary coupling is through the diffusion terms, and the eddy viscosity is usually treated as a
constant in forming the viscous-flux Jacobian matrix. Limited experience to date seems to indicate
 64

there is little advantage to solving all equations simultaneously as opposed to solving the model
equations and mean-flow equations sequentially. If all equations are solved simultaneously, the basic
system of equations for 1-D κ-ω model would be as follows:

Q 
 (F  Fv )  S
t x
 0 
 4 u   0 
ρ  ρu   μ  τ   
3 x
xx
ρu   ρu 2  p   0 
  4 u  
    u  μ  τ xx   q̂ x  
 0 
Q  ρ E  , F  ( ρ E  p)u  , Fv    3 x   , S   u 
 ρκ   ρ uκ   κ   τ xx x  β ρ ωκ 
     (μ  σ μ T )    ω  u 
ρ ω  ρ uω   x   
α   xx
τ  β ρ ω 2

 (μ  σμ T ) ω    κ  x 
 x 
1 4 u 2
E  e  u 2  κ and τ xx  μ T  ρκ
2 3 x 3
I  F Fv  S  F Fv
  δx     ΔQ  δ x (F  Fv )  S , where
n n n
and are the Jacobian
 Δt  Q Q  Q  Q Q

Eq. 5.1
Since dS/dQ is a diagonal matrix and its diagonal elements are always negative, its contribution is
guaranteed to enhance diagonal dominance of the matrix multiplying ΔQ.

View publication stats