Publication 97/2

An Introduction to Turbulence Models

Lars Davidson, http://www.tfd.chalmers.se/lada
Department of Thermo and Fluid Dynamics C HALMERS U NIVERSITY OF T ECHNOLOGY G teborg, Sweden, November 2003 o

Nomenclature 1 Turbulence 1.1 Introduction . . . . . . . . . . . . . . . . 1.2 Turbulent Scales . . . . . . . . . . . . . . 1.3 Vorticity/Velocity Gradient Interaction . 1.4 Energy spectrum . . . . . . . . . . . . . Turbulence Models 2.1 Introduction . . . . . . . . . . 2.2 Boussinesq Assumption . . . 2.3 Algebraic Models . . . . . . . 2.4 Equations for Kinetic Energy 2.4.1 The Exact Equation . 2.4.2 The Equation for 2.4.3 The Equation for 2.5 The Modelled Equation . . 2.6 One Equation Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

IY Y &EY UY F IW %

Reynolds Stress Models 5.1 Reynolds Stress Models . . . . . . . . . . . . . . . 5.2 Reynolds Stress Models vs. Eddy Viscosity Models 5.3 Curvature Effects . . . . . . . . . . . . . . . . . . . 5.4 Acceleration and Retardation . . . . . . . . . . . .

38

IW # IW # IW " Y `W IW W IW ' I' %

Low-Re Number Turbulence Models Models . . . . . . . . . . . . . . . . . 4.1 Low-Re 4.2 The Launder-Sharma Low-Re Models . . . . 4.3 Boundary Condition for and . . . . . . . . . . . 4.4 The Two-Layer Model . . . . . . . . . . . . . 4.5 The low-Re Model . . . . . . . . . . . . . . . 4.5.1 The low-Re Model of Peng et al. . . . 4.5.2 The low-Re Model of Bredberg et al. .

28

U' $ I' # I' " I' ' I' '

Two-Equation Turbulence Models 3.1 The Modelled Equation . . . 3.2 Wall Functions . . . . . . . . . 3.3 The Model . . . . . . . . 3.4 The Model . . . . . . . 3.5 The Model . . . . . . . .

. . . . .

22

G' & 2' F %(& C(& #(& #(& #(& "(& ' (&

"


3 5

XH H QP)

AB@4@E421& 7 D5 7 D5 3 ' 0 AB@986421& 7 57 53 ' 0

'

RSP) RSP) R SP) HQ) P

H P QV)

TQ) P R P S) HQ) P

   !   

constants in the Reynolds stress model constants in the Reynolds stress model constants in the modelled equation constants in the modelled equation constant in turbulence model energy (see Eq. 1.8); constant in wall functions (see Eq. 3.4) damping function in pressure strain tensor turbulent kinetic energy ( ) instantaneous (or laminar) velocity in -direction instantaneous (or laminar) velocity in -direction time-averaged velocity in -direction time-averaged velocity in -direction shear stresses uctuating velocity in -direction normal stress in the -direction uctuating velocity in -direction Reynolds stress tensor instantaneous (or laminar) velocity in -direction time-averaged velocity in -direction uctuating (or laminar) velocity in -direction normal stress in the -direction shear stress instantaneous (or laminar) velocity in -direction time-averaged velocity in -direction uctuating velocity in -direction normal stress in the -direction

7t

boundary layer thickness; half channel height dissipation wave number; von Karman constant ( ) dynamic viscosity dynamic turbulent viscosity kinematic viscosity kinematic turbulent viscosity instantaneous (or laminar) vorticity component in

-direction


& EY F 7t

7 s7 s V r R

7 t

7t

 $0 (&$  " 1 ) '% # !

9PGHF&C@ gg7 I E DB

P9HF&CA897342 IG E D B@ 65 q p i bQ V fQTR f hd(UgdQ V QTRc ecb(`dbQ Vb(T RbQ aQ a Y`(XW(URSQ QTVQT V ss wvs vs u V ww D w u V uu Dy x s 7 ys T 7@D 5 D5 7 @5
7  0 H 7s 5

   
)

time-averaged vorticity component in -direction specic dissipation ( ) uctuating vorticity component in -direction turbulent Prandtl number for variable laminar shear stress total shear stress turbulent shear stress

centerline wall

7t

7t

22 )0H Y

    5 ! E &$#6 "P & @

%   T  TT 7R R 7 D

Almost all uid ow which we encounter in daily life is turbulent. Typical examples are ow around (as well as in) cars, aeroplanes and buildings. The boundary layers and the wakes around and after bluff bodies such as cars, aeroplanes and buildings are turbulent. Also the ow and combustion in engines, both in piston engines and gas turbines and combustors, are highly turbulent. Air movements in rooms are also turbulent, at least along the walls where wall-jets are formed. Hence, when we compute uid ow it will most likely be turbulent. In turbulent ow we usually divide the variables in one time-averaged part , which is independent of time (when the mean ow is steady), and one uctuating part so that . There is no denition on turbulent ow, but it has a number of characteristic features (see Tennekes & Lumley [41]) such as: . Turbulent ow is irregular, random and chaotic. The ow consists of a spectrum of different scales (eddy sizes) where largest eddies are of the order of the ow geometry (i.e. boundary layer thickness, jet width, etc). At the other end of the spectra we have the smallest eddies which are by viscous forces (stresses) dissipated into internal energy. Even though turbulence is chaotic it is deterministic and is described by the Navier-Stokes equations. . In turbulent ow the diffusivity increases. This means that the spreading rate of boundary layers, jets, etc. increases as the ow becomes turbulent. The turbulence increases the exchange of momentum in e.g. boundary layers and reduces or delays thereby separation at bluff bodies such as cylinders, airfoils and cars. The increased diffusivity also increases the resistance (wall friction) in internal ows such as in channels and pipes. . Turbulent ow occurs at high Reynolds number. For example, the transition to turbulent ow in pipes occurs that , and in boundary layers at . . Turbulent ow is always three-dimensional. However, when the equations are time averaged we can treat the ow as two-dimensional. . Turbulent ow is dissipative, which means that kinetic energy in the small (dissipative) eddies are transformed into internal energy. The small eddies receive the kinetic energy from slightly larger eddies. The slightly larger eddies receive their energy from even larger eddies and so on. The largest eddies extract their energy from the mean ow. This process of transferred energy from the largest turbulent scales (eddies) to the smallest is called cascade process.

IIIIF & FFFF

D FB G05A

 s D 5 5

"

( 9 #h&@C

 ( v0)

 ( ! 4  8673 5v#g 23##

8 5 G 67h!

8$  % ) $) # 1

#CU70T(7  (

!       " gC 

! " )1'0) ' %# ( & $ 

$ RShQ ' @I  FIP2I H F W DE05A CB

 G 

5 8
$G@H 

D5

. Even though we have small turbulent scales in the ow they are much larger than the molecular scale and we can treat the ow as a continuum.

As mentioned above there are a wide range of scales in turbulent ow. The larger scales are of the order of the ow geometry, for example the boundary layer thickness, with length scale and velocity scale . These scales extract kinetic energy from the mean ow which has a time scale comparable to the large scales, i.e.

The kinetic energy of the large scales is lost to slightly smaller scales with which the large scales interact. Through the cascade process the kinetic energy is in this way transferred from the largest scale to smaller scales. At the smallest scales the frictional forces (viscous stresses) become too large and the kinetic energy is transformed (dissipated) into internal energy. The dissipation is denoted by which is energy per unit time and unit mass ). The dissipation is proportional to the kinematic viscosity ( times the uctuating velocity gradient up to the power of two (see Section 2.4.1). The friction forces exist of course at all scales, but they are larger the smaller the eddies. Thus it is not quite true that eddies which receive their kinetic energy from slightly larger scales give away all of that the slightly smaller scales but a small fraction is dissipated. However it is assumed that most of the energy (say 90 %) that goes into the large scales is nally dissipated at the smallest (dissipative) scales. The smallest scales where dissipation occurs are called the Kolmogorov scales: the velocity scale , the length scale and the time scale . We assume that these scales are determined by viscosity and dissipation . Since the kinetic energy is destroyed by viscous forces it is natural to assume that viscosity plays a part in determining these scales; the larger viscosity, the larger scales. The amount of energy that is to be dissipated is . The more energy that is to be transformed from kinetic energy to internal energy, the larger the velocity gradients must be. Having assumed that the dissipative scales are determined by viscosity and dissipation, we can express , and in and using dimensional analysis. We write

where below each variable its dimensions are given. The dimensions of the left-hand and the right-hand side must be the same. We get two equations,

A & 1& 93

&

# ! )"

A 10 3

HV

 (#' 

P 0! 8 I  E I3 @ 5

&A

(  v70

 

  CC0 #


  9 C #  

# ! $"

#'

D5

&



  H

 

The interaction between vorticity and velocity gradients is an essential ingredient to create and maintain turbulence. Disturbances are amplied the actual process depending on type of ow and these disturbances, which still are laminar and organized and well dened in terms of physical orientation and frequency are turned into chaotic, three-dimensional, random uctuations, i.e. turbulent ow by interaction between the vorticity vector and the velocity gradients. Two idealized phenomena in this interaction process can be identied: vortex stretching and vortex tilting. In order to gain some insight in vortex shedding we will study an idealized, inviscid (viscosity equals to zero) case. The equation for instantaneous vorticity ( ) reads [41, 31, 44]

where is the Levi-Civita tensor (it is if , are in cyclic order, if , are in anti-cyclic order, and if any two of , are equal), and where denotes derivation with respect to . We see that this equation is not an ordinary convection-diffusion equation but is has an additional term on the right-hand side which represents amplication and rotation/tilting of the vorticity lines. If we write it term-by-term it reads

) 8SS`( P 

The diagonal terms in this matrix represent vortex stretching. Imagine a slender, cylindrical uid element which vorticity . We introduce a cylindrical coordinate system with the -axis as the cylinder axis and as the

4 5SSv H

U# 93 A &

A Y 93 &

U" 93 A &

&

& P

Vt

8G 7h! 3 g 8 58 6 3 U5 6! IG # %U5 ! 75 G 65 5 $ B 6 6 B  H H T 2H 3 A R 

)2

2 1

V eR

& V

xt

 

Rt

T T

 R

which gives and so that

%

. In the same way we obtain the expressions for

UW 93 A &

! !

& 5  V& 5 V  R& 5 R &V 5  V& V 5 V  R& V 5 R 5  V & R 5 V  R & R 5 R

&R

x & ) 5 0x 7 ( ) ' 7 x x &"7  x &"@5 x 7 R  7 D 7

Y0 11&  W T QP P "& P

#' 

and one for seconds

U' 93 A &

&

T '



one for meters

# ghS  %6    

" 30 H " 3#Sv H !  ! 

 5'

) 0x 7 ' 7 x &"7 x 5

"&

) 2

x &3A 3

!"



radial coordinate (see Fig. 1.1) so that stretch the cylinder. From the continuity equation

. A positive

will

we nd that the radial derivative of the radial velocity must be negative, i.e. the radius of the cylinder will decrease. We have neglected the viscosity since viscous diffusion at high Reynolds number is much smaller than the turbulent one and since viscous dissipation occurs at small scales (see p. 6). Thus there are no viscous stresses acting on the cylindrical uid element surface which means that the rotation momentum

4 5SSv H

 ) #  #

remains constant as the radius of the uid element decreases (note that also the circulation is constant). Equation 1.7 shows that the vorticity increases as the radius decreases. We see that a stretching/compressing will decrease/increase the radius of a slender uid element and increase/decrease its vorticity component aligned with the element. This process will affect the vorticity components in the other two coordinate directions. The off-diagonal terms in Eq. 1.6 represent vortex tilting. Again, take a slender uid element with its axis aligned with the axis, Fig. 1.2. The velocity gradient will tilt the uid element so that it rotates in clock-wise direction. The second term in line one in Eq. 1.6 gives a contribution to . Vortex stretching and vortex tilting thus qualitatively explains how interaction between vorticity and velocity gradient create vorticity in all three coordinate directions from a disturbance which initially was well dened in one coordinate direction. Once this process has started it continues, because vorticity generated by vortex stretching and vortex tilting interacts with the velocity eld and creates further vorticity and so on. The vorticity and velocity eld becomes chaotic and random: turbulence has been created. The turbulence is also maintained by these processes. From the discussion above we can now understand why turbulence always must be three-dimensional (Item IV on p. 5). If the instantaneous ow is two-dimensional we nd that all interaction terms between vorticity and velocity gradients in Eq. 1.6 vanish. For example if and all derivatives with respect to are zero. If the third line in Eq. 1.6 vanishes, and if the third column in Eq. 1.6 disappears. Finally, the

I$ 93 A &

R& R 5

F r 5

R V5 Vt
A F T F `R 3 T
0

0('$#! ) & %    "  

F r 5

# ghS  %6    

!

   & 0

V& R 5 V

&

 

" 30 H " 3#Sv H !  ! 

F &  V 3A V 5 1 3 & 1

V1

F r &"7 5

R
!

V& R 5

 R&

V1

R5



remaining terms in Eq. 1.6 will also be zero since

The interaction between vorticity and velocity gradients will, on average, create smaller and smaller scales. Whereas the large scales which interact with the mean ow have an orientation imposed by the mean ow the small scales will not remember the structure and orientation of the large scales. Thus the small scales will be isotropic, i.e independent of direction.

The turbulent scales are distributed over a range of scales which extends from the largest scales which interact with the mean ow to the smallest scales where dissipation occurs. In wave number space the energy of eddies from to can be expressed as where Eq. 1.8 expresses the contribution from the scales with wave number between and to the turbulent kinetic energy . The dimension of wave number is one over length; thus we can think of wave number as proportional to the inverse of an eddys radius, i.e . The total turbulent kinetic energy is obtained by integrating over the whole wave number space i.e.

UC 93 A &

U% 93 A &

0 1&

A V t3 R 5

0 ! )   

 

' &

 

Fr R & 5 P & R 5 F r & V 5 P V & 5

V V D  h! # P B g 8 5
! !

 U(  CXS h&T! #h


A 3 p



A 3 p

V R

The kinetic energy is the sum of the kinetic energy of the three uctuating velocity components, i.e.

The spectrum of correspond to:

is shown in Fig. 1.3. We nd region I, II and III which

I. In the region we have the large eddies which carry most of the energy. These eddies interact with the mean ow and extract energy from the mean ow. Their energy is past on to slightly smaller scales. The eddies velocity and length scales are and , respectively. III. Dissipation range. The eddies are small and isotropic and it is here that the dissipation occurs. The scales of the eddies are described by the Kolmogorov scales (see Eq. 1.4) II. Inertial subrange. The existence of this region requires that the Reynolds number is high (fully turbulent ow). The eddies in this region represent the mid-region. This region is a transport region in the cascade process. Energy per time unit ( ) is coming from the large eddies at the lower part of this range and is given off to the dissipation range at the higher part. The eddies in this region are independent of both the large, energy containing eddies and the eddies in the dissipation range. One can argue that the eddies in this region should be characterized by the ow of energy ( ) and the size of the eddies . Dimensional reasoning gives

This is a very important law (Kolmogorov spectrum law or the law) which states that, if the ow is fully turbulent (high Reynolds

2U" P W0 A& & 1I& 93

A F & 93 &

0      $  0    " " 0  0& ""   % % "  #    &     % )  "0 " )   ) ($"  $)0&  "  W0 " (! 0(0  ) % $)     )  ) )      & '!%     

0 1&

F&

p 7 s 7 s &'  w  u  s  &' ) V V V

1 ) 23 H b! 0'(Q A 3 p

 U(  CXS h&T! #h
A 3 p

number), the energy spectra should exhibit a -decay. This often used in experiment and Large Eddy Simulations (LES) and Direct Numerical Simulations (DNS) to verify that the ow is fully turbulent. As explained on p. 6 (cascade process) the energy dissipated at the small scales can be estimated using the large scales and . The energy at the large scales lose their energy during a time proportional to , which gives

U(& 93 A' &

0 )

0 1

2U" P W0

I& &

 U(  CXS h&T! #h


0 1

When the ow is turbulent it is preferable to decompose the instantaneous variables (for example velocity components and pressure) into a mean value and a uctuating value, i.e.

One reason why we decompose the variables is that when we measure ow quantities we are usually interested in the mean values rather that the time histories. Another reason is that when we want to solve the Navier-Stokes equation numerically it would require a very ne grid to resolve all turbulent scales and it would also require a ne resolution in time (turbulent ow is always unsteady). The continuity equation and the Navier-Stokes equation read

where denotes derivation with respect to . Since we are dealing with incompressible ow (i.e low Mach number) the dilatation term on the righthand side of Eq. 2.3 is neglected so that

Note that we here use the term incompressible in the sense that density is independent of pressure ( ) , but it does not mean that density is constant; it can be dependent on for example temperature or concentration. Inserting Eq. 2.1 into the continuity equation (2.2) and the Navier-Stokes equation (2.4) we obtain the time averaged continuity equation and NavierStokes equation

A new term appears on the right-hand side of Eq. 2.6 which is called the Reynolds stress tensor. The tensor is symmetric (for example ). It represents correlations between uctuating velocities. It is an additional stress term due to turbulence (uctuating velocities) and it is unknown. We need a model for to close the equation system in Eq. 2.6.

U" 3' A

A 1& 3'

UW 3' A U' 3' A

A Y 3'

U# 3' A

V sR s

x&

  ) & )

x & s 7 s x

xt 5 W 7 x 7 ' P7 & x 5  x "&@5  

x & 7A & x 5  x "&@653  7 & P x & A x 56@5 3  @15 7 7 7

P 7A & x D 5  x "&@D 5 3  7 & D P 7

' (&

s 7 s x

& x 3A 3 1 7& 7 P x & A x 56@5 3  @5 7 7 3BA@5 3  1 & 7

! % 

# 8$  %
s 7 s x

x & x D 5 @D 5  7 1 7  @D 5 1 7 367@D 5 &A 3 

8G 675h!  G 8  5 ) $) 1  D 7 s  @D 5 7@ 5 7

R sV s

This is called the closure problem: the number of unknowns (ten: three velocity components, pressure, six stresses) is larger than the number of equations (four: the continuity equation and three components of the NavierStokes equations). For steady, two-dimensional boundary-layer type of ow (i.e. boundary layers along a at plate, channel ow, pipe ow, jet and wake ow, etc.) where

In the wall region (the viscous sublayer, the buffert layer and the logarithmic layer) the total shear stress is approximately constant and equal to the wall shear stress , see Fig. 2.1. Note that the total shear stress is constant only close to the wall; further away from the wall it decreases (in fully developed channel ow it decreases linearly by the distance form the wall). At the wall the turbulent shear stress vanishes as , and the viscous shear stress attains its wall-stress value . As we go away from the wall the viscous stress decreases and turbulent one increases and at they are approximately equal. In the logarithmic layer the viscous stress is negligible compared to the turbulent stress. In boundary-layer type of ow the turbulent shear stress and the vehave nearly always opposite sign (for a wall jet this locity gradient

((  TS`( gh0( P

denotes streamwise coordinate, and coordinate normal to the ow. Often the pressure gradient is zero. on the right-hand side of Eq. 2.8 To the viscous shear stress appears an additional turbulent one, a turbulent shear stress. The total shear stress is thus

UC 3' A

F&

9@

6 s 7T F  u s 8V

1 0 ) (  uvs P D 5

Vt

45 42 3 3

0 D

Eq. 2.6 reads

  9 CU    S( 

I$ 3' A

 & %

  

) " " &  "  

D 5 0

W (&

t D P

$ " #! 

 & 0

'

 



D5 Dy

u vs P D 5  T

0 D5

'

6 #  S 

T D5

D5 D5

' #

Rt t

Dy

is not the case close to the wall). To get a physical picture of this let us study the ow in a boundary layer, see Fig. 2.2. A uid particle is moving to with (the turbudownwards (particle drawn with solid line) from lent uctuating) velocity . At its new location the velocity is in average smaller than at its old, i.e. . This means that when the particle at (which has streamwise velocity ) comes down to (where the streamwise velocity is ) is has an excess of streamwise velocity compared to its new environment at . Thus the streamwise uctuation is and the correlation between and is negative ( ). positive, i.e. If we look at the other particle (dashed line in Fig. 2.2) we reach the same conclusion. The particle is moving upwards ( ), and it is bringing a decit in so that . Thus, again, . If we study this ow for a long time and average over time we get . If we change the sign we will nd that the sign of of the velocity gradient so that also changes. Above we have used physical reasoning to show the the signs of and are opposite. This can also be found by looking at the production term in the transport equation of the Reynolds stresses (see Section 5). In cases where the shear stress and the velocity gradient have the same sign (for example, in a wall jet) this means that there are other terms in the transport equation which are more important than the production term. There are different levels of approximations involved when closing the equation system in Eq. 2.6. An algebraic equation is used to compute a turbulent viscosity, often called eddy viscosity. The Reynolds stress tensor is then computed using an assumption which relates the Reynolds stress tensor to the velocity gradients and the turbulent viscosity. This assumption is called the Boussinesq assumption. Models which are based on a turbulent (eddy) viscosity are called eddy viscosity mod-

uvs u vs

F vs u

$ 

  " ( )

A 3 E5

F u

) 0

u vs P

F uvs F uvs

s R A 5 R 3 A V 35 A 3D 5 A D 5 R 3 V

"  " " "  ) "   &  #  &

F 0 D5

F u

Y&

F u

F s

 (7C 6

6 #  S 
F s
)

 Cv7   9 

' '

0 D5

 

els. In these models a transport equation is solved for a turbulent quantity (usually the turbulent kinetic energy) and a second turbulent quantity (usually a turbulent length scale) is obtained from an algebraic expression. The turbulent viscosity is calculated from Boussinesq assumption. These models fall into the class of eddy viscosity models. Two transport equations are derived which describe transport of two scalars, for example the turbulent kinetic energy and its dissipation . The Reynolds stress tensor is then computed using an assumption which relates the Reynolds stress tensor to the velocity gradients and an eddy viscosity. The latter is obtained from the two transported scalars. Here a transport equation is derived for . One transport equation has to be added for the Reynolds tensor determining the length scale of the turbulence. Usually an equation for the dissipation is used.

Above the different types of turbulence models have been listed in increasing order of complexity, ability to model the turbulence, and cost in terms of computational work (CPU time).

In eddy viscosity turbulence models the Reynolds stresses are linked to the velocity gradients via the turbulent viscosity: this relation is called the Boussinesq assumption, where the Reynolds stress tensor in the time averaged Navier-Stokes equation is replaced by the turbulent viscosity multiplied by the velocity gradients. To show this we introduce this assumption for the diffusion term at the right-hand side of Eq. 2.6 and make an identication

which gives

where is the turbulent kinetic energy (see Eq. 1.10). On the other hand the continuity equation (Eq. 2.5) gives that the right-hand side of Eq. 2.9

2  1

If we in Eq. 2.9 do a contraction (i.e. setting indices side gives

) the right-hand

U% 3' A

x &  7A & x D 5  x "&@D 5 3 A 0 7

" (&

P 7 x 3  x &  s 7 s @ 7A & x D 5  x &"@D 5 3 

8 65 PP  G 7&# D "XP 89XXP  G 6P

 7CTS`( 8673 4 @HI  ( 6 ((  ( !

#U CST(  (
 7C ( 6  (7C 6

d 7A & x D 5  x &"@D 5 3 0 P s 7 s 7 x

s 7 s x

 # dR $

# dRS 

( 0T()$   (
) ) r I' 7 s 7 s

 ##

 #

is equal to zero. In order to make Eq. 2.9 valid upon contraction we add to the left-hand side of Eq. 2.9 so that

Note that contraction of

gives

Above we have used and which are characteristic for the large turbulent scales. This is reasonable, because it is these scales which are responsible for most of the transport by turbulent diffusion. In an algebraic turbulence model the velocity gradient is used as a velocity scale and some physical length is used as the length scale. In boundary layer-type of ow (see Eq. 2.7) we obtain

where is the coordinate normal to the wall, and where is the mixing length, and the model is called the mixing length model. It is an old model and is hardly used any more. One problem with the model is that is unknown and must be determined. More modern algebraic models are the Baldwin-Lomax model [2] and the Cebeci-Smith [6] model which are frequently used in aerodynamics when computing the ow around airfoils, aeroplanes, etc. For a presentation and discussion of algebraic turbulence models the interested reader is referred to Wilcox [46].

The equation for turbulent kinetic energy is derived from the Navier-Stokes equation which reads assuming steady, incompressible, constant viscosity (cf. Eq. 2.4)

C6 " 0( 7U7)1 ! ( ! 6 )86

In eddy viscosity models we want an expression for the turbulent viscosity . The dimension of is (same as ). A turbulent velocity scale multiplied with a turbulent length scale gives the correct dimension, i.e.

A F & 3'

U(& 3' A'

U(& 3' AW

A& 1I& 3'

7 s7 s V ) R

B g 8

P9I G ! 03 E   6 I W "&  &  V V  R R 77 & x7 7 x D 5  x &"@D 5 3 0 P s 7 s x

) x 7 'W

"

# (&

# )  g 4 Q  P 75 896 GeP G 75& 6 8 6 3

x x &"@5  7 & P x 3A x 5 @5 3 7 & 7

 7A &

! !

(7C 6

5 F 7 V 

 Cv7     9

 0

) x 7 2U' W0

 

!"%

$ 1&
The terms in Eq. 2.23 have the following meaning.

 

UI' 3' AW
0

  1 0 ) x1 "&7 Wx0s "&) 7 s ( P x &  x & ) P 7 s 7 `s &' sx



s P x1 &"@D 5 0x s )7 s P ( x1 3&AE) 0x ) D 5 3 ( 7 x 

 

The last term on the right-hand side of Eq. 2.15 is zero. Now we can assemble the transport equation for the turbulent kinetic energy. Equations 2.17, 2.18, 2.20, 2.21,2.22 give

UI' 3' A'

0

' x x & ) x x & AB7 s 7 s 3 & x & A67 sWx &"7 s 3

A& 1G' 3'

0

A 2' 3' F
0

U(& 3' A%
0

U(& 3' AC
0

I1& 3' A$
0 0

U(& 3' e7 s x &  s 7 s  x D 5 7 s  s @15D  A# x x 7 U(& 3' A"

0

A Y & 3'
0

For the rst term we use the same trick as in Eq. 2.18 so that

A x "&7 Wx "&7 s P x & B7 sWx &"7 s 3 7 sWx x &"7 s  s 7 3B7 s P 7 s 7 & P &A 3

The second term on the right-hand side of Eq. 2.15 reads

The third term in Eq. 2.16 can be written as (using the same technique as in Eq. 2.18)

x &"@D 5 s 7 s 7 s x &  s @D 5  7 x x 7 F x 3A xgs & 3 7 s x &  x D 5 @D 5 P A s  x D 5 3 BA7 s  @D 5 3  7 7 x

7 `x 3A s 7 s 3 7 P7 s& x  7 s x x &  @D 5 @5   7 s 7 &  D P 7 s x & x D 5 @D 5 P x 65@5  7  7 7s

& x 7 & D5 7 x 3A s 7 s 3 P x x &"@D 5  7 & D P x 3A x G@D 5 3

! #h 

# v9)#  ( 

Using the continuity equation

The left-hand side can be rewritten as

Subtract Eq. 2.14 from Eq. 2.13, multiply by obtain

The time averaged Navier-Stokes equation reads (cf. Eq. 2.6)

The rst term on the right-hand side of Eq. 2.15 has the form

x & AB7 s 7 sWs 3 &' x

x &  7 Wx D 5  7 s x &"7 s 7 s  x &"7 s 7 s x D 5 &' s x&  7 ' s 7 s & x D 5 "x &  ) x D 5  

 x 3A x D 5 3 F &

We obtain the second term (using

, the rst term is rewritten as

and time average and we

) from

. The large turbulent scales extract energy from the mean ow. This term (including the minus sign) is almost always positive.

The two rst terms represent by pressure-velocity uctuations, and velocity uctuations, respectively. The last term is viscous diffusion. . This term is responsible for transformation of kinetic energy at small scales to internal energy. The term (including the minus sign) is always negative. equation read

Note that the dissipation includes all derivatives. This is because the dissipation term is at its largest for small, isotropic scales where the usual boundary-layer approximation that is not valid.

The equation for the instantaneous kinetic energy is derived from the Navier-Stokes equation. We assume steady, incompressible ow with constant viscosity, see Eq. 2.13. Multiply Eq. 2.13 by so that

The term on the left-hand side can be rewritten as

where . The rst term on the right-hand side of Eq. 2.25 can be written as

The viscous term in Eq. 2.25 is rewritten in the same way as the viscous term in Section 2.4.1, see Eqs. 2.21 and 2.22, i.e.

UI' 3' A%

Now we can assemble the transport equation for 2.27 and 2.28 into Eq. 2.25

by inserting Eqs. 2.26,

UI' 3' AC

UI' 3' A"

A `' x "&3' 7 Wx "&7 s P  ) P7 s 7 u &' Y s s

UI' 3' A#

IU' 3' A$

7 5 7@6@5 V 57 R

AB@59@6421& 7 7 53 ' 0 ' &A & 7 7 x 3Q x 5 3 x 3AB@65@653 x 5 & ' & 7 7 & 7 7 7 7 & 7 7 x 3BA@65@653 x 5 & P x 3BA@5 @653 x 5 x &"@5 x 5 @5 P x 3A x 65@5 @5 3

607 s

x "&@5 x "&@5 P7 3A @563 P x x & x 3Q x 5 3 7 7 & 7 &A

'

t 607 s

x x "&@65@5  7 & @P x 3A x 5 7@5 3 @5 7 7 75 & 7

)

In boundary-layer ow the exact

0) (

& '%6C #  9  

A7 57 53 ' 0  B@9@6421& v9 # Q  P

C (&

! #h 

P u D 5 vs P ) D y

x "&@5 x &"@5 P x x & x x &"@56@5 7 7 7 7

# v9)#  ( 

7 3A @56P 7 & @P & 7 3 7 5

#   # 6   # h1   ( #CU70T(7

) D5

$ @HI 

  

 ##

We recognize the usual transport terms due to convection and viscous diffusion. The second term on the right-hand side is responsible for transport of by pressure-velocity interaction. The last term is the dissipation term which transforms kinetic energy into internal energy. It is interesting to compare this term to the dissipation term in Eq. 2.23. Insert the Reynolds decomposition so that

This shows that the dissipation taking place in the scales larger than the smallest ones is negligible (see further discussion at the end of Sub-section 2.4.3).

The equation for is derived in the same way as that for . Assume steady, incompressible ow with constant viscosity and multiply the time-averaged Navier-Stokes equations (Eq. 2.14) so that

The term on the left-hand side and the two rst terms on the right-hand side are treated in the same way as in Section 2.4.2, and we can write

where

. The last term is rewritten as

Inserted in Eq. 2.33 gives

On the left-hand side we have the usual convective term. On the right-hand side we nd: transport by viscous diffusion, transport by pressure-velocity interaction, viscous dissipation, transport by velocity-stress interaction and loss of energy to the uctuating velocity eld, i.e. to . Note that the last term in Eq. 2.35 is the same as the last term in Eq. 2.23 but with opposite sign: here we clearly can see that the main source term in the equation (the production term) appears as a sink term in the equation. It is interesting to compare the source terms in the equation for (2.23), (Eq. 2.29) and (Eq. 2.35). In the equation, the dissipation term is

UIW 3' A#

A& 1GW 3'

UIW 3' A'

UIW 3' AW

A `W 3' Y

UIW 3' A"

A7 5 53 ' 0 6@97 6421&

7 x "&@5 x "&7 s

x "&@D 5 A s 7 s 3 7 x x 7 3 x 7 D5  x 3&A s 7 s @D 5 P x 3&A s 7 s 3 @SP AB@G@G421& D 7 D5 7 D5 3 ' 0 & x 7 7 7 & 7 &A T x 3A s 7 s 3 @D 5 P x "&@D 5 x "&@D 5 P7 3A D @D 5 3 P x x & D x 3QD x D 5 3

x 3A s 7 s 3 @D 5 P x x "&@D 5 @D 5  7 & D @D 5 P x 3A x D 5 7@D 5 3 @D 5 & x 7 7 7 7 & 7

7s

x "&@D 5 s 7 s 7 x x 7  x 3&A s 7 s @D 5 3 P & A 7 7 & 7 x "&@D 5 x "&@D 5 P7 3A D @D 5 3 P x x & D x 3QD x D 5 3

x "&7 Wx "&7 s P x &"@D 5 x "&7@D 5 P x &"@5 x &"@5 P 7 7 7 s

AB@47@4421& 7 D5 D5 3 ' 0 A 7 D5 7 D5 3 ' 0  B@G@G421& v9 # Q  P

% (&

x &"7 sWx &"7 s D x &"@5 x &"@5 7 7

D5

As the scales of

is much larger than those of

, i.e.

we get

A 2W 3' F

! #h 

x "&7 Wx "&7 s  x &"@D 5 x &"@D 5 x &"@5 x &"@5 7 7 7 7 s

# v9)#  ( 

  

In the equation the dissipation term and the negative production term (representing loss of kinetic energy to the eld) read

The dissipation terms in Eq. 2.36 appear in Eqs. 2.37 and 2.38. The dissipation of the instantaneous velocity eld is distributed into the time-averaged eld and the uctuating eld. However, as mentioned above, the dissipation at the uctuating level is much larger than at the time-averaged level (see Eqs. 2.30 and 2.31).

so that

XhS  #C7U0T7(%6  (

where is the turbulent Prandtl number for . There is no model for the pressure diffusion term in Eq. 2.23. It is small (see Figs. 4.1 and 4.3) and thus it is simply neglected. The dissipation term in Eq. 2.23 is basically estimated as in Eq. 1.12. The velocity scale is now

XhS  0) '&%6 ( C #  9 

Note that the last term in Eq. 2.39 is zero for incompressible ow due to continuity. The triple correlations in term III in Eq. 2.23 is modeled using a gradient law where we assume that is diffused down the gradient, i.e from region of high to regions of small (cf. Fouriers law for heat ux: heat is diffused from hot to cold regions). We get

XhS  #  6CU 

In Eq. 2.23 a number of terms are unknown, namely the production term, the turbulent diffusion term and the dissipation term. In the production term it is the stress tensor which is unknown. Since we have an expression for this which is used in the Navier-Stokes equation we use the same expression in the production term. Equation 2.10 inserted in the production term (term II) in Eq. 2.23 gives

A& 1EY 3'

U' Y 3' A

UIW 3' AC

UIW 3' A%

A UY 3' F

7 s )@D 5 7 5  7

7 "&@D) 'W P x "&@D 5  7 & x D 5 75 7

2' F

8 6 3 G 75&

 x "@G & 7 D5

 0

x &"7 sWx &"7 s P x &"@D 5 s 7 s P 7 x

x& )

x &"@D 5 s 7 s P 7 x

s ) x &"7 Wx &"7 s r H

 9II G

P 7 s 7 sWs &' x

and in the

equation the production and the dissipation terms read

IUW 3' A$


0

 # )  C Q 6  6 P
7 x 7 7 T x &"@D 5 s 7 s  x &"@D 5 x &"@D 5 P

D
)

 %

We have one constant in the turbulent diffusion term and it will be determined later. The dissipation term contains another unknown quantity, the turbulent length scale. An additional transport will be derived from which we can compute . In the model, where is obtained from its own transport, the dissipation term in Eq. 2.43 is simply . For boundary-layer ow Eq. 2.43 has the form

In one equation models a transport equation is often solved for the turbulent kinetic energy. The unknown turbulent length scale must be given, and often an algebraic expression is used [4, 48]. This length scale is, for example, taken as proportional to the thickness of the boundary layer, the width of a jet or a wake. The main disadvantage of this type of model is that it is not applicable to general ows since it is not possible to nd a general expression for an algebraic length scale. However, some proposals have been made where the turbulent length scale is computed in a more general way [14, 30]. In [30] a transport equation for turbulent viscosity is used.

UW Y 3' A

A IY 3' Y

) P

D5

) P

0 )

G' &

HP QV)

 ) 

x&   ) x & ) 0

9PI G G 67&  8 53

) Dy

The modelled

equation can now be assembled and we get

( 6 7C

# 

x 3AE) x D 5 3 &

) D5

An exact equation for the dissipation can be derived from the Navier-Stokes equation (see, for instance, Wilcox [46]). However, the number of unknown terms is very large and they involve double correlations of uctuating velocities, and gradients of uctuating velocities and pressure. It is better to derive an equation based on physical reasoning. In the exact equation for the production term includes, as in the equation, turbulent quantities and and velocity gradients. If we choose to include and in the production term and only turbulent quantities in the dissipation term, we take, glancing at the equation (Eq. 2.43)

For boundary-layer ow Eq. 3.2 reads

The natural way to treat wall boundaries is to make the grid sufciently ne so that the sharp gradients prevailing there are resolved. Often, when computing complex three-dimensional ow, that requires too much computer resources. An alternative is to assume that the ow near the wall behaves like a fully developed turbulent boundary layer and prescribe boundary conditions employing wall functions. The assumption that the ow near the wall has the characteristics of a that in a boundary layer if often not true at all. However, given a maximum number of nodes that we can afford to use in a computation, it is often preferable to use wall functions which allows us to use ne grid in other regions where the gradients of the ow variables are large. In a fully turbulent boundary layer the production term and the dissipation term in the log-law region ( ) are much larger than the

U' 3W A

UW 3W ) H ebQ P A Vc

Note that for the production term we have write the transport equation for the dissipation as

. Now we can

A 1& 3W


0

7 x "&@D 5

s 7 s x

)0 AE22 3H

D5

! %  


2H ebQ A Vc

Rc dbQ c

0 H dbQ Rc

IF & 9 2W F F

)


) H VecbQ P $ "0(  " V 7 7 D5 ) R x "&@D 5  7 & x D 5  x "&@4 H dcbQ P c

dRbcQ 3

# 8$  % ) 1$)

c 0

I' '

) &   x x H c & 0

8 6 3 G 75&


P 8G 67h! 8  I903  5  I H D  D t5 y H

  9II G 

  $)  '


x 32H x D 5 3 &A

%!

%!

20

10

10

20 0 200 400 600

800

1000

1200

other terms, see Fig. 3.1. The log-law we use can be written as

Comparing this with the standard form of the log-law

we see that

where we have replaced the dissipation term by . In the log-law region the shear stress is equal to the wall shear stress , see Fig. 2.1. The Boussinesq assumption for the shear stress reads (see Eq. 2.10)

Using the denition of the wall shear stress 3.15 into Eq. 3.8 we get

, and inserting Eqs. 3.9,

U% 3W A

UC 3W A

A F & 3W

7T 6 Vs

02 )

In the log-layer we can write the modelled

equation (see Eq. 2.44) as

0 #

I$ 3W A

U# 3W A

U" 3W A A Y 3W

 g
WY
0 0 0 0 

  )  

CI

) % 0 !()

 

&%$#"!CBA@9&%(8 # &%&%$$##&%4!"!$23#$)1 #7)0)65(''  # 

# $)  

H   

 

I' W

F G

u vs P

W 2F F Y

 7H P V D 5  0 F H0) & S p

S T

s p 8

s    D  8   & 0   " ( )

6 u D 5 0 vs P T

( )#

s 8

)0


V 8V s
5U0


F h% p 8s 5

 CC v


&

)0

&

B D EIFFUIYW Y D CP5A

QR

bQ i

  

For the velocity component parallel to the wall the wall shear stress is used as a ux boundary condition (cf. prescribing heat ux in the temperature equation). When the wall is not parallel to any velocity component, it is more convenient to prescribe the turbulent viscosity. The wall shear stress is obtained by calculating the viscosity at the node adjacent to the wall from the log-law. The viscosity used in momentum equations is prescribed at the nodes adjacent to the wall (index P) as follows. The shear stress at the wall can be expressed as
6

where denotes the velocity parallel to the wall and is the normal distance to the wall. Using the denition of the friction velocity we obtain
8

Y `'

&

&

where

&

&

8 &

p3

s D5 0

 

0 9 E 8 s 

) 0

Substituting

with the log-law (Eq. 3.4) we nally can write

 v9

" 301 !    9

U(& 3W AW

0 5 vs P u

&

where the velocity gradient in the production term computed from the log-law in Eq. 3.4, i.e.

has been

v9   

where the friction velocity is obtained, iteratively, from the log-law (Eq. 3.4). Index denotes the rst interior node (adjacent to the wall). The dissipation is obtained from observing that production and dissipation are in balance (see Eq. 3.8). The dissipation can thus be written as

R 7

v9   9   vS`(

From experiments we have that in the log-law region of a boundary layer so that . When we are using wall functions and are not solved at the nodes adjacent to the walls. Instead they are xed according to the theory presented above. The turbulent kinetic energy is set from Eq. 3.10, i.e.

U(& 3W A'

A& 1I& 3W


0 0

&

& 5 & 0

& 

&

D 5 & 0

%UF F iWQ

8 

Vs

( )#

D5

&

) V s VeR i Q

&

Vs

 CC v 5 0
8 8

&  

&

& 

5 & 0

W F
&

D5

  
) 20 s 8V

In the model the modelled transport equations for and (Eqs. 2.43, 3.2) are solved. The turbulent length scale is obtained from (see Eq. 1.12,2.42)

The turbulent viscosity is computed from (see Eqs. 2.11, 2.41, 1.12)

The dissipation and production term can be estimated as (see Sub-section 3.2)

In the logarithmic layer we have that , but from Eqs. 3.17, 3.18 we nd that . Instead the diffusion term in Eq. 3.16 can be rewritten using Eqs. 3.17, 3.18, 3.15 as

U(& 3W AC

U(& 3W A%

 i g   Q c c V eRi Q V c   ) ) 0 VV V e F 2H 0 F E) 0 i g   Q

I' "

! 

which together with

! 

gives

R 7

 S v `( 

We have ve unknown constants and , which we hope should be universal i.e same for all types of ows. Simple ows are chosen where the equation can be simplied and where experimental data are used to determine the constants. The constant was determined above (Subsection 3.2). The equation in the logarithmic part of a turbulent boundary layer was studied where the convection and the diffusion term could be neglected. In a similar way we can nd a value for the constant . We look at the equation for the logarithmic part of a turbulent boundary layer, where the convection term is negligible, and utilizing that production and dissipation are in balance , we can write Eq. 3.3 as

A Y & 3W

U(& 3W A"

U(& 3W A#

I1& 3W A$


0 0 0 0

dUQ Rc

Xeb(`db(Tv`Q TVcQTRcQ i

C 1 ) c 0 ) Vc H A ebQ P RdcbQ 3  H 0  V H )

`Q i

H ) bQ ) ibQ  i VeR 

I G   V

C QV) Q     6 H P P

T s 8

V e

H)

V e

H P QV)

!"%!

Inserting Eq. 3.19 and Eq. 3.17 into Eq. 3.16 gives

Experimental data give [43], and is chosen. We have found three relations (Eqs. 3.10, 3.20, 3.23) to determine three of the ve unknown constants. The last two constants, and , are optimized by applying the model to various fundamental ows such as ow in channel, pipes, jets, wakes, etc. The ve constants are given the following values: , , , , .

The model is gaining in popularity. The model was proposed by Wilcox [45, 46, 36]. In this model the standard equation is solved, but as a length determining equation is used. This quantity is often called specic dissipation from its denition . The modelled and equation read

UI' 3W AW

A `' 3W Y

UI' 3W A"

I% X& eUQ ' Vc

GW X& c F X& &

R ) hdQ Vf

7 2H t 0

Assuming that the decay of is exponential . Insert this in Eq. 3.21, derivate to nd Eq. 3.22 yields

, Eq. 3.21 gives and insert it into

R 7

 S v `( 

The ow behind a turbulence generating grid is a simple ow which allows us to determine the constant. Far behind the grid the velocity gradients are very small which means that . Furthermore and the diffusion terms are negligible so that the modelled and equations (Eqs. 2.43, 3.2) read

A& 1G' 3W

) UI' 3W A'

A 2' 3W F


0 0 0

F y

F 

R gdfQ 3

) R

UF F #

I G  I% X& ebQ IY X& dbQ %UF F WQ ' Vc Y Rc i

I' #

) R H T ) R 0 8 x&   f x 3A R x D 5 3 & x & R 0  

22 R )0H R

x&  x& )

I' X& "

e`Q Vc

 ) 

f 0

R P P 6 C SV) Q

c eRi Q P V VebQ dbQ c Rc V V

)

t 7 V H ecbQ P H D 5

t H P )7 D 5

& 

x 3AE) x D 5 3 &

VebQ c

R P)

t P




%!

The constants are determined as in Sub-section 3.3: , , , and . When wall functions are used and are prescribed as (cf. Sub-section 3.2):

In regions of low turbulence when both and go to zero, large numerical problems for the model appear in the equation as becomes zero. The destruction term in the equation includes , and this causes problems as even if also goes to zero; they must both go to zero at a correct rate to avoid problems, and this is often not the case. On the in the contrary, no such problems appear in the equation. If equation in Eq. 3.24, the turbulent diffusion term simply goes to zero. Note that the production term in the equation does not include since

One of the most recent proposals is the model of Speziale [39] where the transport equation for the turbulent time scale is derived. The exact equation for is derived from the exact and equations. The modelled and equations read

The constants are: .

and

are taken from the

model, and

IU' 3W A$

UI' 3W  )T 1& P ebQ 3 AC A Vc

H QPV) Vc ebQ T0 `E) H  T  x & T x & T V 02  ' P x & T x & ) R 02  & P 93 )T  x  x & T  V 2 & 0

 )

A dRbcQ

T P )

T ) P

U' $

x &   2) x & ) 0

I G 

# IW X& V 2 R 2 Rc dbQ iWQ T T G) ibQ 0

H0 `E) T

' 

R P )

In Ref. [35] the ow.

model was used to predict transitional, recirculating

UI' 3W A#

2U" gdQ UF F %0 Rf %
0

@D 5 7

xt 7t x D5

20 H )

7 @D 5

xt

t RgdQ @Dx7 5 f

A 3 v 6 R T 8V s VeR A 3 s eR 8 V 8

 7 t x D5

' f 7 @D 5

xt

H P Q)

6 T P P C QV) Q
) R f 0(gdQ R

'

x 3A T x D 5 3 &

x 3AE) x D 5 3 &

f)

R gdfQ

FY0 Vf U1UW hdQ

6

 %!

In the previous section we discussed wall functions which are used in order to reduce the number of cells. However, we must be aware that this is an approximation which, if the ow near the boundary is important, can be rather crude. In many internal ows where all boundaries are either walls, symmetry planes, inlet or outlets the boundary layer may not be that important, as the ow eld is often pressure-determined. For external ows (for example ow around cars, ships, aeroplanes etc.), however, the ow conditions in the boundaries are almost invariably important. When we are predicting heat transfer it is in general no good idea to use wall functions, because the heat transfer at the walls are very important for the temperature eld in the whole domain. When we chose not to use wall functions we thus insert sufciently many grid lines near solid boundaries so that the boundary layer can be adequately resolved. However, when the wall is approached the viscous the ow is viscous domeffects become more important and for inating, i.e. the viscous diffusion is much larger that the turbulent one (see Fig. 4.1). Thus, the turbulence models presented so far may not be correct since fully turbulent conditions have been assumed; this type of models are often referred to as highnumber models. In this section we will discuss modications of highnumber models so that they can be used all the way down to the wall. These modied models are termed low Reynolds number models. Please note that high Reynolds number and low Reynolds number do not refer to the global Reynolds number (for example , , etc.) but here we are talking about the local formed by a turbulent uctuation turbulent Reynolds number and turbulent length scale. This Reynolds number varies throughout the computational domain and is proportional to the ratio of the turbulent and physical viscosity , i.e. . This ratio is of the order of 100 or larger in fully turbulent ow and it goes to zero when a wall is approached. We start by studying how various quantities behave close to the wall when . Taylor expansion of the uctuating velocities (also valid for the mean velocities ) gives

F 0u

I' C

where slip, i.e.

are functions of space and time. At the wall we have nowhich gives . Furthermore, at the wall , and the continuity equation gives so that

A 1& 3Y

(
0

7s

! %  

"

# 8$  % ) $) 1
9

B5A B 5A

0 1

0 

B 5A

B 5A

`` `` ``

& 1

F 5A F 5A 5A B B B

5 R  V VWQ 5 SQ  V V 5 R  V V R

7 @D 5

F 0 w t 0 s F w u s V WQ ``

! ) 0


 Q   (  




I' %

In Fig. 4.2 DNS-data for the fully developed ow in a channel is presented.

There exist a number of Low-Re number models [32, 7, 10, 1, 27]. When deriving low-Re models it is common to study the behavior of the terms when in the exact equations and require that the corresponding terms in the modelled equations behave in the same way. Let us study

H P QV)

9PI G  2)0(42  1 'G

A 3


     

0 D5 )

UW 3Y A
0

A 3 A1V 3 A 3
!

A V 3 A  3

`` `` 9A R Q  R R 3 V ``VC V `` V R R Q `` V VV ``  R V V V
 

u vs V w Vu Vs

U' 3Y A
0

&  ! " ) $ #  # & " $  %  8 & %0     " ! %" #   A  "   "      ) ( "( "( 0 %)  H  "  ) ()   %  '  $%  $# ) "( "(  ) " )  0(  ) ) !((  #  3  ! 0 2UF  F #  U0 s 0 2F I"% 2"C $ 0  5 %   0 0 ) #  G '     )   F" 5  &  )  B   0) !(% "  # 0%'$ $ ) #!%  0 "   8     %  ) $ % $ # 7 5 A & 0 Y ()"  % )      
0

9 

H


 

(7C 6

0.3

From Eq. 4.2 we immediately get

`` `` ``

V VWQ VV V V
%

R SQ w u R s


. Equation 4.1 can now be written

F  R

0 s

CI

0.2

0.1

0.1

0.2

10

15

20

25

30

H P QV)

4R

2W F
Comparing Eqs. 4.4 and 4.5 we nd that the dissipation term in the modelled equation behaves in the same way as in the exact equation when

U# 3Y A
0

A 3


AA 33


H ) ibQ  V

U" 3Y A
0

A 3

) 0

The pressure diffusion term is usually neglected, partly because it is not measurable, and partly because close to the wall it is not important, see Fig. 4.3 (see also [28]). The modelled equation reads

A Y 3Y
0

A 3 x1 "&7 Wx0s "&) 7 s A 3 1 0 )' ( 7 P P s 7 u & s u 



V  F" 5 F%C #    !% 0 0 2UF F U0 s 0 2I2$ 0 5    $%  " $0%0)'$  (#! %  0 "   8    %  ) $  % $ # )   
0

 D 5 0 a w

s 0a w
8

s 0a s
4R

(7C 6

H P QV)

2.5

0.12

D 5 0 a s

1.5

0.04

 D 5 0 a u

s 0a u

0.5

0 0

20

40

60

80

100

0.02

0.06

0.08

0.14

0.1

0 0

0.2

0.4

0 E

When arriving at that the production term is

A 3

7 s

a7 s

0 # 0

7 %

B 5A 

9 

" ) & G')  % ' Y ()"  



the exact

equation near the wall (see Eq. 2.24).

we have used
0.6 0.8 1

0 ) )

A 3 1 H ) ( P 0 V) A 3 V  ) 0
( 1

D5

( 0 ) D

) D5

0u

( P V)  A 3 V 0 1 ) ( u D 5 vs P ) D y

) D5

0.25 0.2

0.15 0.1 0.05

0.1 0

10

15

. However, both the modelled production and the diffusion term are of whereas the exact terms are of . This inconsistency of the modelled terms can be removed by replacing the constant by where is a damping function so that when and when . Please note that the term damping term in this case is not correct since actually is augmenting when rather than damping. However, it is common to call all low-Re number functions for damping functions. Instead of introducing a damping function , we can choose to solve for a modied dissipation which is denoted , see Ref. [25] and Section 4.2. It is possible to proceed in the same way when deriving damping functions for the equation [39]. An alternative way is to study the modelled equation near the wall and keep only the terms which do not tend to zero. From Eq. 3.3 we get

where it has been assumed that the production term has been suitable modied so that . We nd that the only term which do not vanish at the wall are the viscous diffusion term and the dissipation term

8  & 0    ) $ #  #  " " %"   "  & " $    )  ( "!( ( 0 # )    "   ) )  "     "()  2UF % F '  U0 $%  0 # 2I2$ ) "( "0! ( 5 ) " # 0 F "   5 %  s F C#   # % #  B  )0 # !"()  0%'$  $) !%  0 "   8    %  ) $  % $ # 7 5A %  ) #   

I$ 3Y A

i q WQ i

A 3

i SR Q

1 3 A

iq

iq

XH

H 0

0.05

20

25

30


A 3 V R c 0 )

 9

 

A 3 A 3 V 0 1 ) ( 0 1 ) ) ( Vc H ebQ P V H V A 3 R )0 ) (

GW &

iq

 )

1 3 A

H RdbcQ

(7C 6


i q F2"  9

A 3 A 3 R R 1 0) ( H D  1 H D t50) ( y

H P QV)
)

0 s

A 3

4R

 # ) ( "(   "  ! & ) I G'   C W Y ()"  

& 

0 #'

iq

&

! 

iq

" %

so that close to the wall the dissipation equation reads

The equation needs to be modied since the diffusion term cannot balance the destruction term when .

There are at least a dozen different low Re models presented in the literature. Most of them can be cast in the form [32] (in boundary-layer form, for convenience)

 P Xg R 6 ThC)Cv 2

Different models use different damping functions ( , and ) and different extra terms ( and ). Many models solve for rather than for where is equal to the wall value of which gives an easy boundary (see Sub-section 4.3). Other models which solve for use condition no extra source in the equation, i.e. . Below we give some details for one of the most popular low-Re models, the Launder-Sharma model [25] which is based on the model of Jones & Launder [20]. The model is given by Eqs. 4.9, 4.10, 4.11 and 4.12 where

UC 3Y A

U% 3Y A

A F & 3Y

U(& 3Y A' A& 1I& 3Y

U(& 3Y AW

H P QV)


0 0 0 0 0 0

Vq

Rq iq

D5

XH

9PI G  2)0'(42 3 D 3 S) g  8  42  1 G @ 3

0 X H R q c SQ R

( 6 7C

D5

H P Q)

H P S)


F H

 XH 

F XH p  XH H XH i ) i q bQ 0 V ) p  XH V q VebcQ P  c V t  XH D y  XH D 5 0  ) D  ) D t 5 y

IW '

 ) 

4R

2
F


XH ) V  VD 5 ' V  V  0 ' ) F  V A P V ! W QP & & F & V A 2"20 Y h W A P  93 ! 

  P R 6 Xg ThC)Cv 2 Q   P

) H VebQ P V H F c

p Vq Rq iq

The term was added to match the experimental peak in around [20]. The term is introduced to mimic the nal stage of decay of turbulence behind a turbulence generating grid when the exponent in changes from to .

In many low-Re models is the dependent variable rather than . The main reason is that the boundary condition for is rather complicated. The largest term in the equation (see Eq. 4.4) close to the wall, are the dissipation term and the viscous diffusion term which both are of so that

From this equation we get immediately a boundary condition for as

From Eq. 4.14 we can derive alternative boundary conditions. The exact form of the dissipation term close to the wall reads (see Eq. 2.24)

In the same way we get an expression for the turbulent kinetic energy

Comparing Eqs. 4.17 and 4.19 we nd

U(& 3Y A%

A 2' 3Y F

IW W

``  R Q  R  &' V V

 '

so that

I1& 3Y A$

U(& 3Y AC

where and Taylor expansion in Eq. 4.1 gives

have been assumed. Using

A Y & 3Y

U(& 3Y A"

U(& 3Y A#

v

A 3

  803  eG G 67 65  G B & 8  G 8 8 3

X H )v H v9 # %) 0gC)C$   6  6 ! 6 

D

" h'

XH

V w

``  R Q  R  &' ) V V V

``

" I' X&

H SP)

7H

t 0  V s

 RV Q  RV  H

V)

P V) F

Vq

!"

F2'

In the Sharma-Launder model this is exactly the expression for in Eqs. 4.12 and 4.13, which means that the boundary condition for is zero, i.e. . In the model of Chien [8], the following boundary condition is used

which gives Eq. 4.21.

Near the walls the one-equation model by Wolfshtein [48], modied by Chen and Patel [7], is used. In this model the standard equation is solved; the diffusion term in the -equation is modelled using the eddy viscosity assumption. The turbulent length scales are prescribed as [15, 11]
#

( is the normal distance from the wall) so that the dissipation term in the -equation and the turbulent viscosity are obtained as:

The Reynolds number

and the constants are dened as

The one-equation model is used near the walls (for ), and the standard highin the remaining part of the ow. The matching line could either be chosen along a pre-selected grid line, or it could be dened as the cell where the damping function

takes, e.g., the value . The matching of the one-equation model and the model does not pose any problems but gives a smooth distribution of and across the matching line.

UI' 3Y A'

UI' 3Y AW

2I' A F"

A c Q 0 A 93 P

F % F iQ Q ' c Q T `$ i Q T  i g Q T UF  bgT ) )  Q

!

P &

Q c T A i Q 0 A 93 P

I G  g 42`) G '  B3

! 

Y `W

This is obtained by assuming

in Eqs. 4.17 and 4.18 so that

A& 1G' 3Y

F  X H
0

XH

R SQ R

H P ! P  C6 QV) h3 2 R $ Q

i )

"I% F A i Q 0 A P93

biQ

HQV) P

!

0 T c ) H

V ) '

P &

B 5A

) V RV RV ' H

Ve

Q i

!

P &

H P0 QV)

A model which is being used more and more is the model of Wilcox [45]. The standard model can actually be used all the way to the wall without any modications [45, 29, 34]. One problem is the boundary condition for at walls since (see Eq. 3.25)

tends to innity. In Sub-section 4.3 we derived boundary conditions for by studying the equation close to the wall. In the same way we can here use the equation (Eq. 3.25) close to the wall to derive a boundary condition for . The largest terms in Eq. 3.25 are the viscous diffusion term and the destruction term, i.e.

The solution to this equation is

The equation is normally not solved close to the wall but for is computed from Eq. 4.26, and thus no boundary condition actually needed. This works well in nite volume methods but when nite element methods are used is needed at the wall. A slightly different approach must then be used [16]. Wilcox has also proposed a model [47] which is modied for viscous effects, i.e. a true low-Re model with damping function. He demonstrates that this model can predict transition and claims that it can be used for taking the effect of surface roughness into account which later has been conrmed [33]. A modication of this model has been proposed in [36].

A `' 3Y Y

UI' 3Y A"

UI' 3Y A#


R " h' 9@
0 0 0

R P)

IW "

I G 
R P QV)

R P C6 SV)  A 3

R Vf P F V hdQ V R V

1) 20'( IG
)


RP )

4R

) 8H R

f V h#V dQ R



P Q
R R

IW #

I' " P
A 

UI' 3Y A%
0

 V


!

P &


!

&A


&G% F  UF  i q % F  R ) i q i % 

xt
R

f 

xt

  ) 

UI' 3Y AC
0

xt
)

xt xt  ) ) R )    f % ) &A R x D 5 3 x t  R hf P ) V V % R Rgf % xt A &E) x D 5 3 x t

 @R ) ) %

R P #h&98$ C6 SV)  6 

IU' 3Y A$
0

" X&
A 

eR V

!

xt xt
R

) ) A R ) hfdQ P ) f RgdQ 3 V q f R f  dQ ) R Q P )
)q)

IW X& f T C  ) T 2$  dQ " F " F f 2$ F  hdQ T ' Y  gdQ T UF  ) Q " F Vf F Rf % F   " X& W Y  & f P ! W Y  & f q qT  A    I2e'VF R F " P & F A  2$U% F A  ! 4IF F  V  F& P P &  "IUF  i q ' F  A  !   F& P ! 'I2$ QP & ) q ' F  A  R ) i q   t f  t t  x x A R x D5 3 x R   t  )   x t AE) x D 5 3 x t  x

R P ) C6 SV)  

The turbulent viscosity is given by


1

4R



P Q  

4R


4R

P Q

R P C6 SV) 



P Q

The

R P SV)

A new

 ) R P)

model of Peng et al. reads [36]

model was recently proposed by Bredberg et al. [5] which reads

 
1

  

UW $

A 2W 3Y F
0

F T % Y  Rgf
%

C X& F V f T & ) T '2$ F  hf % F T & X& f % T & i % T UF  )

% %

A R GE) 30
A

with the turbulent Reynolds number dened as in the model are given as

. The constants

R P C6 SV) 

4R



P Q

In Reynolds Stress Models the Boussinesq assumption (Eq. 2.10) is not used but a partial differential equation (transport equation) for the stress tensor is derived from the Navier-Stokes equation. This is done in the same way as for the equation (Eq. 2.23). Take the Navier-Stokes equation for the instantaneous velocity (Eq. 2.4). Subtract the momentum equation for the time averaged velocity ((Eq. 2.6) and multiply by . Derive the same equation with indices and interchanged. Add the two equations and time average. The resulting equation reads

where

is the pressure-strain term, which promotes isotropy of the turbulence; is the dissipation (i.e. transformation of mechanical energy into heat in the small-scale turbulence) of ;

Note that if we take the trace of Eq. 5.1 and divide by two we get the equation for the turbulent kinetic energy (Eq. 2.23). When taking the trace the pressure-strain term vanishes since

due to continuity. Thus the pressure-strain term in the Reynolds stress equation does not add or destruct any turbulent kinetic energy it merely redistributes the energy between the normal components ( , and ). acts to Furthermore, it can be shown using physical reasoning [19] that reduce the large normal stress component(s) and distributes this energy to the other normal component(s).

U' 3" A

V w

s 7 s x

and

are the convection and diffusion, respectively, of

x 7

Vu Vs

77

x s 7 s

IW C

x s 7 s

is the production of )];

[note that

A 1& 3"

 VV

7 5

7 @D 5

R R 77

)1 &

7 x (H 0 &" )1 & x s ) 7) s ' ( P x7 ( ) 0  ) & A s 7 s 3 P 0x 7  ) 7  ) `s 7 s P ) x sx s s x  x7 x7 x7 % 1 0 ) ( 7A & s  x "&7 s 3  )1 "&7 D 5 ) Ws P ) ) & x D 5 ) s 7 s P ( )1 & A s 0) s ) D 5 3 ( x sx 0 x 7

) RV

! %  

!! 1

F 7 &"7 s ' 77

!

x s

x7

"V % 
)


! !

x7

x7

x7

7 x (H

The pressure diffusion term is for two reasons commonly neglected. First, it is not possible to measure this term and before DNS-data (Direct Numerical Simulations) were available it was thus not possible to model this term. Second, from DNS-data is has indeed been found to be small (see Fig. 4.3). The dissipation tensor is assumed to be isotropic, i.e.

From the denition of (see 5.1) we nd that the assumption in Eq. 5.5 is equivalent to assuming that for small scales (where dissipation occurs) the

# A Y 3" v7   7US 

The object of the wall correction terms and is to take the effect of coordinate system, the wall into account. Here we have introduced a with along the wall and normal to the wall. Near a wall (the term near may well extend to ) the normal stress normal to the wall is damped (for a wall located at, for example, this mean that the normal stress is damped), and the other two are augmented (see Fig. 4.2). In the literature there are many proposals for better (and more complicated) pressure strain models [40, 23]. The triple correlation in the diffusion term is often modelled as [9]
0

We nd that there are terms which are unknown in Eq. 5.1, such as the triple correlations , the pressure diffusion and the pressure strain , and the dissipation tensor . From Navier-Stokes equation we could derive transport equations for this unknown quantities but this would add further unknowns to the equation system (the closure problem, see p. 12). Instead we supply models for the unknown terms. The pressure strain term, which is an important term since its contribution is signicant, is modelled as [24, 17]

UW 3" A

U" 3" A

0 A 0)x 7 s 

 )7

7 x `H x gs 3

R & Y Ya

F t

R & a

V & x a7

IW %

 R&

I2' 9 FF

9PI G XP 7 5C@ P &&B 1  P IG8 q ) V s H R a Q q V s H R a Q ' P  W ) x 7 ' P x 7 VWQ P   ) x 7 'W P s 7 s H) SQ P x R x a7  V & x 7   R & x 7
& 

( 6 7C

)&

H A x s 7 s 3 ) s ) s Y`Q x 7 

7 x `H

7 x `H

x7 s`s 7 s x

((  TS
2 3

H t" I" h' )

7H x 7 'W x (H 7

(  ! 8673 4

R& x7 x7

q R & Y Ya R & a V& x7

Vu

two derivative and are not correlated for . This is the same as assuming that for small scales and are not correlated for which is a good approximation since the turbulence at these small scale is isotropic, see Section 1.4. We have given models for all unknown term in Eq. 5.1 and the modelled Reynolds equation reads [24, 17]

where should be taken from Eq. 5.3. For a review on RSMs (Reynolds Stress Models), see [18, 22, 26, 38].

Whenever non-isotropic effects are important we should consider using RSMs. Note that in a turbulent boundary layer the turbulence is always non-isotropic, but isotropic eddy viscosity models handle this type of ow excellent as far as mean ow quantities are concerned. Of course a model give very poor representation of the normal stresses. Examples where non-isotropic effects often are important are ows with strong curvature, swirling ows, ows with strong acceleration/retardation. Below we present list some advantages and disadvantages with RSMs and eddy viscosity models. Advantages with eddy viscosity models: i) simple due to the use of an isotropic eddy (turbulent) viscosity; ii) stable via stability-promoting second-order gradients in the meanow equations; iii) work reasonably well for a large number of engineering ows. Disadvantages with eddy viscosity models: i) isotropic, and thus not good in predicting normal stresses ( ); ii) as a consequence of i) it is unable to account for curvature effects; iii) as a consequence of i) it is unable to account for irrotational strains. Advantages with RSMs: i) the production terms need not to be modelled; ii) thanks to i) it can selectively augment or damp the stresses due to curvature effects, acceleration/retardation, swirling ow, buoyancy etc. Disadvantages with RSMs:

W U# 3" H x 7 ' P x 7 A

V wTV uTV s

9PI G U5 X"9P6  B 6PG!

( 7C6

H P QV)

" 0( 7U7( !
2 )& 

x7 H &3A s 7 s 3 s ) s `Q  ) & A s 7 s 3  x ) Y x  sx Ws P ) & x D 5 ) s 7 s P ) & A s 7 s ) D 5 3 x

 

H )86 ! 6

 9PI G XP 7 5C@ P &&B 1 P IG8

& ) "7 D 5 )

s x

7(C 6 UY F 7s
)&

xs

((  TS

) &"7

(  ! 8673 4


%

A 0 B r B 0

U (r)

i) complex and difcult to implement; ii) numerically unstable because small stabilizing second-order derivatives in the momentum equations (only laminar diffusion); iii) CPU consuming.

Curvature effects, related either to curvature of the wall or streamline curvature, are known to have signicant effects on the turbulence [3]. Both types of curvature are present in attached ows on curved surfaces, and in separation regions. The entire Reynolds stress tensor is active in the interaction process between shear stresses, normal stresses and mean velocity strains. When predicting ows where curvature effects are important, it is thus necessary to use turbulence models that accurately predict all Reynolds stresses, not only the shear stresses. For a discussion of curvature effects, see Refs. [12, 13]. When the streamlines in boundary layer type of ow have a convex (concave) curvature, the turbulence is stabilized (destabilized), which dampens (augments) the turbulence [3, 37], especially the shear stress and the Reynolds stress normal to the wall. Thus convex streamline curvature decreases the stress levels. It can be shown that it is the exact modelling of the production terms in the RSM which allows the RSM to respond correctly to this effect. The model, in contrast, is not able to respond to streamline curvature. The ratio of boundary layer thickness to curvature radius is a common parameter for quantifying the curvature effects on the turbulence. The work reviewed by Bradshaw demonstrates that even such small amounts of convex curvature as can have a signicant effect on the turbulence. Thompson and Whitelaw [42] carried out an experimental inves-

0 )

1 ) ! 0( 1 ) '

1 )  

%  7(

EY &

$ 

& 4F F A 0

  " ( )

P5 7h!

( 0

h & 0$    1 

#

!

H P Q)

5 7  3

F 5 A I5 3 " T1 

 & 0

!"%

y x

streamline

tigation on a conguration simulating the ow near a trailing edge of an airfoil, where they measured . They reported a 50 percent decrease of (Reynolds stress in the normal direction to the wall) owing to and was also substantial. In addition curvature. The reduction of they reported signicant damping of the turbulence in the shear layer in the outer part of the separation region. An illustrative model case is curved boundary layer ow. A polar coordinate system (see Fig. 5.1)) with locally aligned with the streamline is introduced. As (with and ), the radial inviscid momentum equation degenerates to

Here the variables are instantaneous or laminar. The centrifugal force exerts a force in the normal direction (outward) on a uid following the streamline, which is balanced by the pressure gradient. If the uid is displaced by some disturbance (e.g. turbulent uctuation) outwards to level A, it encounters a pressure gradient larger than that to which it was accustomed at , as , which from Eq. 5.7 gives . Hence the uid is forced back to . Similarly, if the uid is displaced inwards to level B, the pressure gradient is smaller here than at and cannot keep the uid at level B. Instead the centrifugal force drives it back to its original level. It is clear from the model problem above that convex curvature, when , has a stabilizing effect on (turbulent) uctuations, at least in the radial direction. It is discussed below how the Reynolds stress model responds to streamline curvature. Assume that there is a at-plate boundary layer ow. The ratio of the normal stresses and is typically 5. At one station, the ow is deected upwards, see Fig. 5.2. How will this affect the turbulence? Let us study the effect of concave streamline curvature. The production terms owing to rotational strains can be written as

0        ) $ )  " )0('& %   t )   0 0  %  7A    3  %  # $%  % " !!%   ) "       &    0)     # ()  %  (  %    ' "00 "  )  '$  " ( #    "  )  ) &% 0 '0(& " 00 " &  )   

I$ 3" A

x7




A 1 0 3 A 1 0 3

F 5

0 15

uvs P

UF F W

'Y

D A

Vs

3 1 I5

Vu

(5 3 (5 3 A A

( 0

h & 0$    1  Vs

F 

Vu

P 15

0 (5

As long as the streamlines in Fig. 5.2 are parallel to the wall, all pro. However as soon as the streamlines are deduction is a result of ected, there are more terms resulting from . Even if is much smaller that it will still contribute non-negligibly to as is much larger than . Thus the magnitude of will increase ( is negative) as . An increase in the magnitude of will increase , and . This means that and will which in turn will increase be larger and the magnitude of will be further increased, and so on. It is seen that there is a positive feedback, which continuously increases the Reynolds stresses. It can be said that the turbulence is owing to concave curvature of the streamlines. However, the model is not very sensitive to streamline curvature (neither convex nor concave), as the two rotational strains are multiplied by the same coefcient (the turbulent viscosity). If the ow (concave curvature) in Fig. 5.2 is a wall jet ow where , the situation will be reversed: the turbulence will be . If the streamline (and the wall) in Fig. 5.2 is deected downwards, the situation will be as follows: the turbulence is stabilizing when , and destabilizing for . The stabilizing or destabilizing effect of streamline curvature is thus dependent on the type of curvature (convex or concave), and whether there is an increase or decrease in momentum in the tangential direction with ). For convenience, radial distance from its origin (i.e. the sign of these cases are summarized in Table 5.1.

0 D5

U% 3" A

UC 3" A

A F & 3"

A& 1I& 3"

 u V s 9VR uvs P V 9VR V s t 9VR 0 D y

0 0 0 0

3 4 @ 9 76 54 0EDC@CBA280(23

3 4 @ 9 76 0EDC@CBA2805

" % 0 % "   % ( &$ #!     & 0 "  ) ' % "    

F 0 D5

0 0 # " )  0 0 # " )  

0  1

1D 5 0

t D y uvs ' P V V t V D 5 V u P D y V s P 9R

9VR t 0 Dy

 t D V y

)   0 0 # " 0 0 # " ) 

0  1

D 5 uvs ' P R R

D5

VV

WY

9VR

 1  1 0377  1  0 451 7 RR

)

0 D5

( 0

h & 0$    1 

H QP)

F 0 D5

F t 0 Dy V u 0 D 5

C I C I CI

H P QV)

P s T 2 V 1

P vs T 2 u 1

P u T 2 V 1

x2 y x1 x x2 x1

When the ow accelerates and/or decelerate the irrotational strains ( , and ) become important. In boundary layer ow, the only term which contributes to the pro( denotes streamwise direcduction term in the equation is tion). Thompson and Whitelaw [42] found that, near the separation point as well as in the separation zone, the production term is of equal importance. This was conrmed in prediction of separated ow using RSM [12, 13]. In pure boundary layer ow the only term which contributes to the production term in the and -equations is . Thompson and Whitelaw [42] found that near the separation point, as well as in the separation zone, the production term is of equal importance. As the exact form of the production terms are used in second-moment closures, the production due to irrotational strains is correctly accounted for. the proIn the case of stagnation-like ow (see Fig. 5.3), where duction due to normal stresses is zero, which is also the results given by second-moment closure, whereas models give a large production. In order to illustrate this, let us write the production due to the irrotational strains and for RSM and :

If we get The production term in be of the two strains.

since due to continuity. model, however, will be large, since it will

t 0 D5

t 0 5 A V u P V s3 P

Vu

Vs

t D 5 A V u P V s 3 P 0 D 5 uvs P 0 H

0 D y P t D 5 0

  t H P  0 ' )  Q) V Dy  V D5  t u P s P &A V V  R R 3 " F  A yD V D 5 V H P QV) 0 Dy t 0 D5

%  7(

t 0 5 uvs P

    )   )

8 6 3 G 75& 3 5 1  03 G 673 g I ! 8 8 5

H P QV)

 6  #C#gS
IY Y W "
0

HQ) P F D VV

 

4 )v 6

RR

0 D

  #h77

Vu

0 Dy

Vs

Academic Press, 1974. [7] [8]

Near-wall turbulence models for complex ows including separation. AIAA Journal 26 (1988), 641648. Predictions of channel and boundary layer ows with a low-reynolds-number turbulence model. AIAA Journal 20 (1982), 33 38. Transport equation in turbulence. Physics of Fluids 13 (1970), 26342649. Calculation of the turbulent buoyancy-driven ow in a rectangular cavity using an efcient solver and two different low reynolds number turbulence models. Numer. Heat Transfer 18 (1990), 129147. Reynolds stress transport modelling of shock/boundary-layer interaction. 24th AIAA Fluid Dynamics Conference, AIAA 93-3099, Orlando, 1993. Prediction of the ow around an airfoil using a Reynolds stress transport model. ASME: Journal of Fluids Engineering 117 (1995), 5057.

"Y

gY73ua  9  I 5 v"

[12]

g9 

 I 5 v" Y73ua

[11]

H P QV)

ggY73ua  9  I 5 v"

[10]

 V  2 F "   "  0 2 " GR4Q&DXua

[9]

 5 A   "  5 c&Hf7BHp&!&yhxhXGw

[6]

Analysis of Turbulent Boundary Layers.

R P)

gY73ut"QEsr q#XBQ#HXXQG0 9   I 5 v " a  D  p '  S  i ' F  F

[5]

An improved turbulence model applied to recirculating ows. International Journal of Heat and Fluid Flow 23, 6 (2002), 731743.

Edh3XE4hgedcb`YHQX8W&SUR3QPHG0 $ 2 2 f  " a I I5 F F V  4" I  " F

[4]

Calculation of boundary-layer development using the turbulent energy equation. Journal of Fluid Mechanics 28 (1967), 593616.

 2 f S   " D   ThXHX"EXGw

&TR3QPHG0  S 4" I  " F

[3]

Effects of streamline curvature on turbulent ow. Agardograph progress no. 169, AGARD, 1973.

 " A  9   "  0 5 4 2 " ED8#CB#@876310

[2]

Thin-layer approximation and algebraic model for separated turbulent ows. AIAA 78-257, Huntsville, AL, 1978.

   " $       ()#"'&%"#!

[1]

A new turbulence model for predicting uid ow and heat transfer in separating and reattaching ows - 1. Flow eld calculations. Int. J. Heat Mass Transfer 37 (1994), 139151.

S 4 (
 5  X7QGw

! 8$  &(  # 1 

[20]

The prediction of laminarization with a two-equation model of turbulence. Int. J. Heat Mass Transfer 15 (1972), 301314. The collaborative testing of turbulence models (organized by P. Bradshaw ). Data Disk No. 4 (also available at Ercoftacs wwwserver http://fluindigo.mech.surrey.ac.uk/database), 1990. Second-moment closure: Present ... and future? Int. J. Heat and Fluid Flow 10 (1989), 282300. On the elimination of wall-topography parameters from second-moment closures. Physics of Fluids A 6 (1994), 9991006. Progress in the development of a Reynolds-stress turbulence closure. Journal of Fluid Mechanics 68, 3 (1975), 537566. Application of the energy dissipation model of turbulence to the calculation of ow near a spinning disc. Lett. Heat and Mass Transfer 1 (1974), 131138.

#Y

 0 #"BAQQHU)X)9 F "  p   "  0 F    "

[25]

 "y5u!T hX&)X)9     " x  0 F    "

[24]

  p 5   "  F    " &S r qy)9 q0q)X)9

[23]

 0 F    " 8)X)9

[22]

 i q A75

[21]

0)FX")9t"Q `hhi   I    i 5 q 7&D

[19]

 x 2   "  5`" &D

[18]

Advanced turbulence closure models: A view of current status and future prospects. Int. J. Heat and Fluid Flow 15 (1994), 178203. Turbulence, second ed. McGraw-Hill, New York, 1975.

 0 F   " 9   "     I 5 )X)YH7

[17]

Ground effects on pressure uctuations in the atmospheric boundary layer. Journal of Fluid Mechanics 86 (1978), 491511.

 S 5 v A &7gB

[16]

The Full Multigrid Method Applied to Turbulent Flow in Ventilated Enclosures Using Structured and Unstructured Grids. PhD thesis, Dept. of Thermo and Fluid Dynamics, Chalmers University of Technology, Goteborg, 1997.

cgy5 " ggY73ua 5    9   I 5 v "

[15]

Navier-Stokes stall predictions using an algebraic stress model. J. Spacecraft and Rockets 29 (1992), 794800.

  II &YY2

[14]

Calculation of some parabolic and elliptic ows using a new one-equation turbulence model. In 5th Int. Conf. on Numerical Methods in Laminar and Turbulent Flow (1987), C. Taylor, W. Habashi, and M. Hafez, Eds., Pineridge Press, pp. 411422.

   9  I 5 v" " gBY73ua

g#Y73ua  9  I 5 v"

[13]

Reynolds stress transport modelling of shock-induced separated ow. Computers & Fluids 24 (1995), 253268.

S 4 (
4

T )XQX Qp   yu F F  x   " 5  

[37]

Calculation of curved shear layers with two-equation turbulence models. Physics of Fluids 26 (1983), 14221435.

 F A2  qpB#'HX13D " g9BY73ugEsr q1#XBS     I 5 v " a D  p ' 

[36]

A modied lowReynolds-number model for recirculating ows. ASME: Journal of Fluids Engineering 119 (1997), 867875.

 p A 2 D   9   I 5 v " a  D  p '  q#'HFX13"QgY73uEsr q#XBS

[35]

The two-equations model applied to recirculating ventilation ows. Tech. turbulence Rep. 96/13, Dept. of Thermo and Fluid Dynamics, Chalmers University of Technology, Gothenburg, 1996.

 p A2  qW#'HFX13D dgdYI73uWEsr qt#XBS  " 9  5 v" a D p ' 

[34]

Performance of two-equation turbulence models for numerical simulation of ventilation ows. In 5th Int. Conf. on Air Distributions in Rooms, ROOMVENT96 (Yokohama, Japan, 1996), S. Murakami, Ed., vol. 2, pp. 153160.

$Y

#6( T)hXHXS i      " 2 f "

[33]

Application of turbulence models to separated ows over rough surfaces. ASME: Journal of Fluids Engineering 117 (1995), 234241.

T)FXQX xQ@#  yu ThXHXS F  p  " 5    2 f "

[32]

Turbulence models for near-wall and low Reynolds number ows: A review. AIAA Journal 23 (1985), 13081319.

R P QV)

R P)

   f  " 6 QS

[31]

 V F f GH)XHX

[30]

On the connection between one- and two-equation models of turbulence. In Engineering Turbulence Modelling and Experiments 3 (1996), W. Rodi and G. Bergeles, Eds., Elsevier, pp. 131140. Incompressible Flow. John Wiley & Sons, New York, 1984.

 V F f G)XHX

[29]

Two-equation eddy-viscosity turbulence models for engineering applications. AIAA Journal 32 (1994), 15981605.

  &S 75 

@i A75 E$@)Yt  "   F   I  "

[28]

Reynolds-stress and dissipation-rate budgets in a turbulent channel ow. Journal of Fluid Mechanics 194 (1988), 1544.

 )X75 Yh9 F   xI

  "   5 @GV X7)9

[27]

Low-Reynolds-number eddyviscosity modelling based on non-linear stress-strain/vorticity relations. In Engineering Turbulence Modelling and Experiments 3 (1996), W. Rodi and G. Bergeles, Eds., Elsevier, pp. 91100.

 F  x  )X75 YIh9

[26]

Modelling engineering ows with Reynolds stress turbulence closure. J. Wind Engng. and Ind. Aerodyn. 35 (1991), 2147.

S 4 (
4

CY

 5 f I 2  7HP 

[48]

The velocity and temperature distribution in onedimensional ow with turbulence augmentation and pressure gradient. Int. J. Mass Heat Transfer 12 (1969), 301318.

 C  x caq#HH275 

[47]

Simulation of transition with a two-equation turbulence model. AIAA Journal 32 (1994), 247255.

 C  x caq#HH275 

[46]

Turbulence Modeling for CFD. DCW Industries, Inc., 5354 Palm Drive, La Canada, California 91011, 1993.

 a C  x c#HH275 

[45]

Reassessment of the scale-determining equation. AIAA Journal 26, 11 (1988), 12991310.

 V 5 GHf7Q 

[44]

   I  4  cQ#XPE&

[43]

The Structure of Turbulent Shear Flow, second ed. Cambridge University Press, New York, 1976. Fluid Mechanics. McGraw-Hill, Inc., New York, 1994.

ER3`XH75Q  "60EYIh i 4"2 f      A  

[42]

Characteristics of a trailingedge ow with turbulent boundary-layer separation. Journal of Fluid Mechanics 157 (1985), 305326.

u321A 9 GEG`hX i   " D I  

[41]

A First Course in Turbulence. The MIT Press, Cambridge, Massachusetts, 1972.

 5 I " !QyQfX "qpQ)Q3" 5 Xp   F " F " p  w 2

[40]

Modelling the pressurestrain correlation of turbulence: an invariant dynamical system approach. Journal of Fluid Mechanics 227 (1991), 245272.

   F   "   w 2 &YIX%6#75h 83" 5 Xp

[39]

Critical evaluation of two-equation models for near-wall turbulence. AIAA Journal 30 (1992), 324331.

E#"Q D ' 

  " ( 5 " 9  y)6#Qp

[38]

Second-order near-wall turbulence closures: a review. AIAA Journal 29 (1991), 18191835.

S 4 (
4