Objective

• Define Reynolds Navier Stokes Equations

(RANS)
• Start with Numerics
 From the last class
Averaging Navier Stokes equations
p  P p ,
ρ ρ ρ , Substitute into Navier Stokes equations
v x  Vx  v x '
v y  Vy  v y ' Instantaneous velocity

v z  Vz  v z ' fluctuation
around
average
T T T' Average
velocity
velocity
Continuity equation: time
v x v y v z  (Vx  v x ' )  (Vy  v y ' )  (Vz  v z ' ) Vx Vy Vz v x ' v y ' v z '
           0
x y z x y z x y z x y z
Average whole equation: 0 0 0 Average
Vx Vy Vz v x ' v y ' v z ' Vx Vy Vz v x ' v y ' v z '
     0      0 Vx Vy Vz
x y z x y z x y z x y z   0
x y z
Average of average = average Average of fluctuation = 0
 From the last class
Example: of Time Averaging
Write continuity equations in a short format:

v x v x v x v x p 2vx 2vx 2vx

ρ(  vx  vy  vz )   μ μ  μ 2  Sx
τ x y z x x 2 y 2 z

   
v  vx i  vy j  vz k

v x v x v x   
vx  vy  vz  div( v x v )  v x div v  div( v x v )
x y z
=0 continuity
2vx 2vx 2vx
 2
 2
 2
 μ div(grad v x )
x y z
Short format of continuity equation in x direction:

v x  p
ρ(  div(v x v ))    μ div(grad v x )  S
τ x x
 Averaging of Momentum Equation
v x  p
ρ(  div(v x v ))    μ div(grad v x )  Sx
τ x
averaging
v x  p
ρ  ρ div(vx v )    μ div(grad v x )  Sx
τ x
0
v x  (Vx  v'x )  (Vx  v'x ) Vx Vx
ρ ρ ρ ρ ρ
τ τ τ τ τ
' '
 
' '
 v ' '
v v x vy v 'x v 'z
div( v x v )  div (Vx V)  div( v x v )  div (Vx V)  x x
 
x y z
      
' ' ' ' ' ' ' ' ' ' ' '
div( v v )  div( v (v i  v j  v k))  div((v v i  v v j  v v k)) 
x x x y z x x x y x z
' '
v 'x v 'x v x v y v 'x v 'z
  
x y z

div(grad v x )  div(grad Vx )  div(grad V x )

 Time Averaged Momentum Equation
Instantaneous velocity
v x v x v x v x p 2vx 2vx 2vx
ρ(  vx  vy  vz )   μ 2
μ 2
 μ 2  Sx
τ x y z x x y z
Average velocities
Vx Vx Vx Vx P  2 Vx  2 Vx  2 Vx v 'x v 'x v 'x v 'y v 'x v 'z
ρ(  Vx  Vy  Vz )   μ 2 μ 2 μ 2 ρ ρ ρ  Sx
τ x y z x x y z x y z

Reynolds stresses
For y and z direction:

Vy Vy Vy Vy

P  2 Vy  2 Vy  2 Vy v 'y v 'x v 'y v 'y v 'y v 'z
ρ(  Vx  Vy  Vz ) μ μ μ ρ ρ ρ  Sy
τ x y z x x 2 y 2 z 2 x y z

Vz Vz Vz Vz P  2 Vz  2 Vz  2 Vz v 'z v 'x v 'z v 'y v 'z v 'z
ρ(  Vx  Vy  Vz ) μ μ μ ρ ρ ρ  Sz
τ x y z x x 2 y 2 z 2 x y z

Total nine
 Time Averaged Continuity Equation
Instantaneous velocities
v x v y v z
  0
x y z
Averaged velocities
Vx Vy Vz
  0
x y z

Time Averaged Energy Equation

Instantaneous temperatures and velocities
T T T T  2T  2T  2T
ρc p (  Vx  Vy  Vz )  k 2 k 2 k 2 Φq
τ x y z x y z
Averaged temperatures and velocities
' ' ' ' ' '
T T T T  2T  2T  2T T v x T v y T v z
ρc p (  Vx  Vy  Vz )  k 2  k 2  k 2  ρ ρ ρ Φq
τ x y z x y z x y z
 Reynolds Averaged Navier Stokes
equations
Vx Vy Vz Reynolds stresses
  0 total 9 - 6 are unknown
x y z

Vx Vx Vx Vx P  2 Vx  2 Vx  2 Vx v 'x v 'x v 'x v 'y v 'x v 'z
ρ(  Vx  Vy  Vz )   μ 2 μ 2 μ 2 ρ ρ ρ  Sx
τ x y z x x y z x y z
same
Vy Vy Vy Vy P  2 Vy  2 Vy  2 Vy v 'y v 'x v 'y v 'y v 'y v 'z
ρ(  Vx  Vy  Vz )   μ 2 μ 2 μ 2 ρ ρ ρ  Sy
τ x y z x x y z x y z

Vz Vz Vz Vz P  2 Vz  2 Vz  2 Vz v 'z v 'x v 'z v 'y v 'z v 'z
ρ(  Vx  Vy  Vz )   μ 2 μ 2 μ 2 ρ ρ ρ  Sz
τ x y z x x y z x y z

Total 4 equations and 4 + 6 = 10 unknowns

We need to model the Reynolds stresses !
Modeling of Turbulent Viscosity
μ Fluid property – often called laminar viscosity

μt Flow property – turbulent viscosity

 constant  t
 MVM  MVM: Mean velocity models
  mixing length
TKEM: Turbulent kinetic energy equation models
 One - Eq.
 
     Free
    
     1  Layer
  High Re  wall 
  2  Layer
   bounded 
 
  k -    3  Layer
  
Models based on μ t  
TKEMTwo   Low Re
   Buoyancy
  
 Eq.  Curvature
 
 k - 
 
 
 k - l

   Additional models:
 k - kl
 k  f


   LES: Large Eddy simulation models
  ......
RSM: Reynolds stress models
 Modeling of Reynolds stresses
Eddy viscosity models
'
v v '(incompressible flow)
ρ x x
 (ρv 'x v 'x )
x x
Average velocity
Boussinesq eddy-viscosity approximation

ρv i' v 'j Is proportional to deformation  Vi  Vj 

 
 x j x i 
μt Coefficient of proportionality

 V   V   V 
ρ v x v x  μ t  2 x  ρ v y v y  μ t  2 y  ρ v z v z  μ t  2 z 
 x   y   z 
 V Vy 
ρ v x v y  ρ v y v x  μ t  x  
k = kinetic energy
 y x 
of turbulence
 V V 
ρ v x v z  ρ v z v x  μ t  x  z  ' '
v 'x v 'x v y v y v 'z v 'z
 z x  k  
2 2 2
 V Vy 
ρ v z v y  ρ v y v z  μ t  z  
 y z  Substitute into Reynolds Averaged equations
 Reynolds Averaged Navier Stokes
equations
Continuity:
Vx Vy Vz
1)   0
x y z
Momentum:
Vx V V V P  V  V  V
2) ρ(  Vx x  Vy x  Vz x )    [(μ  μ t ) y ]  [(μ  μ t ) y ]  [(μ  μ t ) y ]  ST x
τ x y z x x x y y z z

V
y yV y V y V P  y V  V
y  Vy
3) ρ( τ  Vx x  Vy y  Vz z )   x  x [(μ  μ t ) x ]  y [(μ  μ t ) y ]  z [(μ  μ t ) z ]  ST y

Vz V V V P  V  V  V
4) ρ(  Vx z  Vy z  Vz z )    [(μ  μ t ) y ]  [(μ  μ t ) y ]  [(μ  μ t ) y ]  ST z
τ x y z x x x y y z z

 v x v y v
S Tz  Sz  stz  Sz  [(μ  μ t )  (μ  μ t )  (μ  μ t ) z ]
z x y z
Similar is for STy and STx

4 equations 5 unknowns → We need to model μt

 One equation models:
Prandtl Mixing-Length Model (1926)
Vx

y
Vx
μ t ρl2 x
l
y

Characteristic length (in practical applications:

distance to the closest surface)
-Two dimensional model
-Mathematically simple
-Computationally stable
-Do not work for many flow types

There are many modifications of Mixing-Length Model:

- Indoor zero equation model: t = 0.03874  V l Distance to the closest surface
Air velocity
Kinetic energy and dissipation of energy
Kolmogorov scale
Eddy breakup and decay to smaller length scales where dissipation appear
 Two equation turbulent model
k -  model

Kinetic energy Energy dissipation

k2
μ t  ρ Cμ From dimensional analysis
ε

constant
We need to model k and 
Two additional equations:

k
 
ρ(  div(k V))  div μ  μ t /  μ grad k  2t Eij Eij - ρ

kinetic energy
τ
  2
ρ(  div( V))  divμ  μ t /   grad   C1 f 12 t Eij Eij - C2 f 22ρ dissipation
τ k
 Reynolds Averaged Navier Stokes
equations
Continuity:
Vx Vy Vz
1)   0
x y z
Momentum:
Vx V V V P  V  V  V
2) ρ(  Vx x  Vy x  Vz x )    [(μ  μ t ) y ]  [(μ  μ t ) y ]  [(μ  μ t ) y ]  Sx
τ x y z x x x y y z z

V
y yV y V y V P  y V  V
y  V y
3) ρ( τ  Vx x  Vy y  Vz z )   y  x [(μ  μ t ) x ]  y [(μ  μ t ) y ]  z [(μ  μ t ) z ]  Sy

Vz V V V P  V  V  V
4) ρ(  Vx z  Vy z  Vz z )    [(μ  μ t ) z ]  [(μ  μ t ) z ]  [(μ  μ t ) z ]  Sz
τ x y z z x x y y z z

General format:

 
ρ  ρ divV  div(Γ  ,eff grad  )  S
τ
 General CFD Equation
 
ρ  ρ divV  div(Γ  ,eff grad  )  S
τ
Values of , ,eff and S
 ,eff S
Equation
Continuity 1 0 0
x-momentum V1 + t -P/x+Sx
y-momentum V2 + t -P/y-g(T∞-Twall)+Sy
z-momentum V3 + t -P/z+Sz
T-equation T /l + t/t ST
k-equation k (+ t)/k G- +GB
-equation  (+ t)/ [ (C1G-C2)/k] +C3GB(/k)
Species C (+ t)/c SC
Age of air  + t 
 t =Ck2/ , G= t (Ui/xj +Uj/xi) Ui/xj ,
GB=-g(/CP)( t/T,t) T/ xi
 Finite Volume Method
- Conservation of for the finite volume
Divide the whole computation domain into sub-domains

One dimension:
n
h
W x P E
w e x

s
x
l e w

- Finite volume is a fixed space in the flow domain with imaginary boundaries
that allow the fluid to flow in and out.

- Integral conservation of the quantities such as mass, momentum and energy.


General Transport Equation -3D problem
steady-state
H N 
ρ div(V )  div(Γ  ,eff grad  )  S
W P E
S L

Equation for node P in the algebraic format:

a P Φ P  a E Φ E  a W Φ W  a SΦ S  a N Φ N  a H Φ H  a L Φ L  f
 1-D example of discretization of
general transport equation
 
ρ  div(ρV )  divΓ  ,eff grad   S
τ
Steady state 1dimension (x): W xw P xe E


ρVx Φ    (Γ Φ, eff Φ )  SΦ w
x
e
x x x
Point W and E represent the cell center of the west and east neighbors of cell P
and w, e the neighboring surfaces. Integrating with Gaussian theorem on this control
volume gives: 
 divdV   ndA
V A
ρVx Φ e  ρVx Φ w  { Γ Φ, eff Φ    Γ Φ, eff Φ  }  SΦ x
 x  e  x  w

To obtain the equations for the value at point P,

assumptions are used to convert the surface values to the center values .
 1-D example of discretization of
general transport equation
 
ρ  div(ρV )  divΓ  ,eff grad   S
τ
Steady state 1dimension (x): W xw P xe E


ρVx Φ    (Γ Φ, eff Φ )  SΦ w
x
e
x x x
Point W and E represent the cell center of the west and east neighbors of cell P
and w, e the neighboring surfaces. Integrating with Gaussian theorem on this control
volume gives: 
 divdV   ndA
V A
ρVx Φ e  ρVx Φ w  { Γ Φ, eff Φ    Γ Φ, eff Φ  }  SΦ x
 x  e  x  w

To obtain the equations for the value at point P,

assumptions are used to convert the surface values to the center values .
 Convection term

ρVx Φ    (Γ Φ, eff Φ )  SΦ W xw P xe E
x x x x
w e

– Central difference scheme:

ρVx Φ e  ρVx Φ w  (ρVx Φ E  ρVx Φ P ) / 2  (ρVx Φ P  ρVx Φ W ) / 2  ρVx Φ E  ρVx Φ W

- Upwind-scheme:

Φ e  Φ P and Φ w  ΦW ρVx Φ e  ρVx Φ w  ρVx Φ P  ρVx Φ W

Φ e  Φ E and Φ w  Φ P ρVx Φ e  ρVx Φ w  ρVx Φ E  ρVx Φ P