The Reynolds Averaged Navier Stokes Equations RANS

We start off by writing down the classical Navier Stokes equations for incompressible, constant
viscosity, Newtonian fluids on the form
       
u u u u
D   ∂   −∇P 2 
v = v + (u · ∇) v = + g + ν∇ v  (1)
 
Dt ∂t ρ
     
w w w w

∂
where this notation highlights the fact that the operators , (u · ∇) and (ν∇2 ) must be applied
∂t
to each component of the velocity vector u. To derive the Reynolds Averaged Navier-Stokes
Equations (RANS) we assume that the dependent variables (velocity and pressure) consist of
a mean part φ and a random fluctuating turbulent part φ0 , such that φ = φ + φ0 . This is
called Reynolds decomposition. Before we go on into the algebra, we should note some laws for
averaging.

Laws for averaging

1 T
Z
Let the mean of a time signal φ(t) be defined as φ = φdt, where the integral is taken over
T 0
a time period T which is considerably larger than the period of the turbulent fluctuations. From
this definition we may define several algebraic rules for manipulation of turbulent signals. Let
f = f + f 0 and g = g + g be two turbulent signals:

(1) : f 0 = 0 (2) : f = f (3) : f g = f g

(4) : f 0 g = 0 (5) : f + g = f + g (6) : f g = f g

∂f ∂f
Z Z
(7) : f g = f g + f 0 g 0 (8) : = (9) : f ds = f ds
∂s ∂s

Deriving RANS
The process of Reynolds averaging is now to insert u = u + u0 and P = P + P 0 into (1), and
thereafter take the average of the whole equation term for term, utilizing the laws above. The
result is that the transport equations for u involves additional terms called Reynolds stresses,
that acts as additional unknowns in the equations, and needs proper modelling to find suitable
solutions. As we will see, these Reynolds stresses emerge from the non linear convective accel-
eration term. We will now conduct the Reynolds averaging of the u− momentum equation, and
then generalize the result. Inserting u = u + u0 and p = P + P 0 into the u− momentum, and
then averaging the equation term for term gives:

Time derivative
∂(u + u0 ) ∂u ∂u0 ∂u ∂u0 ∂u ∂u0 ∂u
= + = + = + =
∂t ∂t ∂t ∂t ∂t ∂t ∂t ∂t

1
Pressure gradient
∂(P + P 0 ) ∂P ∂P 0 ∂P ∂P 0 ∂P ∂P 0 ∂P
= + = + = + =
∂x ∂x ∂x ∂x ∂x ∂x ∂x ∂x

Viscous Diffusion
∇2 (u + u0 ) = ∇2 u + ∇2 u0 = ∇2 u + ∇2 u0 = ∇2 u
As seen from the previous calculations, the Reynolds averaging does not produce any additional
term in the linear terms.

Convective acceleration
Before we calculate the non linear convective acceleration term, we should utilize the following
identity
(u · ∇)φ = ∇ · (uφ) − φ(∇ · u)
where of course the term φ(∇ · u) = 0, for an incompressible fluid. Reynolds averaging gives:

([u + u0 ] · ∇)(u + u0 ) = ∇ · [(u + u0 ) (u + u0 ] =

∂ ∂ ∂
((u + u0 )(u + u0 )) + ((v + v 0 )(u + u0 )) + ((w + w0 )(u + u0 )) (2)
∂x ∂y ∂z
Let us work out the first of these terms:
∂ ∂ ∂(u u) ∂uu0 ∂u0 u0
((u + u0 )(u + u0 )) = (u u + 2uu0 + u0 u0 )) = +2 +
∂x ∂x ∂x ∂x ∂x
where the terms uu0 = 0 according to the averaging laws, but the terms u0 u0 6= 0, and this is the
cause of the Reynolds stresses. We may generalize this results into to yield

∂ ∂u2 ∂u0 u0
((u + u0 )(u + u0 )) = +
∂x ∂x ∂x

∂ ∂(u v) ∂u0 v 0
((v + v 0 )(u + u0 )) = +
∂y ∂y ∂y

∂ ∂(u w) ∂u0 w0
((w + w0 )(u + u0 )) +
∂z ∂z ∂z
such that
∇ · [(u + u0 ) (u + u0 ] = ∇ · (u u) + ∇ · u0 u0
We are now in a position to write the full form of the u−momentum RANS equation which
becomes
∂u 1 ∂P
+ (u · ∇)u = − + g + ν∇2 u + ∇ · u0 u0
∂t ρ ∂x
This results may be generalized in 3D to form the complete RANS equations
       
u u u [−u0 u0 −u0 v 0 −u0 w0 ]
Du ∂   −∇P 2 
=  v  + (u · ∇)  v  = + g + ν∇  v  + ∇ ·  [−v 0 u0 −v 0 v 0 −v 0 w0 ]  (3)
   
Dt ∂t ρ
w w w [−w0 u0 −w0 v 0 −w0 w0 ]

2
The last term of this equations is the divergence of the Reynolds stress tensor. Written in this
form, it illustrates that for each of the momentum equations, the divergence operator should be
applied to the row vector of the corresponding momentum equation. For example if we were
to write out the v−momentum, we must apply the divergence operator to the 2nd row of the
Reynolds stress tensor.

Continuity
Performing Reynolds averaging of continuity gives us ∇ · (u + u0 ) = 0 →

∇ · u = 0, ∇ · u0 = 0 (4)

So we see that both the mean and fluctuation parts of velocity both satisfy continuity individ-
ually.

RANS on tensor notation

The equations in turbulence are notoriously long, tedious and cumbersome to write out in full
vector notation. Therefore it is an absolute must to utilize tensor notation (also called index
notation). The important thing to remember is to know the Einstein summation convention
that says that if a term consist of variables with repeated index, the term must be summed over
that index. An example is continuity
∂ui ∂u ∂v ∂w
∇·u= = + +
∂xi ∂x ∂y ∂z
Since the i appears in both ui and xi , we say that that index is repeated, and therefore the
term must by summed over i = 1, 2, 3. If a term has both i and j repeated, the term must be
summed over both i and j giving a total of 9 terms if i, j = 1, 2, 3. Using this convention, we
may now write the RANS equations on tensor notation

∂ui ∂ui ∂P ∂ 2 ui ∂
ρ( + uj )=− + ρg i + µ − ρu0 u0 (5)
∂t ∂xj ∂xi ∂xj 2 ∂xj i j
∂ 2 uj ∂ 2 uj
Note that the Laplacian term = must be summed over i since the i is repeated.
∂xi 2 ∂xi ∂xi
Note also that since the Reynolds stress tensor is symmetric, we have a total of 6 additional
∂
unknowns. The question now is how to model the additional unknowns − (ρu0i u0j ). If we
∂xj
∂
assume 2D flow where w = = 0 but w02 6= 0, we are left with the additional terms −ρu02 ,
∂z
−ρu0 v 0 and −ρv 02
It is the aim of turbulence models to be able to relate the unknown components of the Reynolds
stress tensor to the mean flow quantities and thus providing closeure to the equations. One line
of modelling known as eddy viscosity models is to relate the Reynolds stress tensor components
to a the mean rate of strain tensor through a turbulent (eddy) viscosity. This is known as the
Boussinesq approximation for turbulence, and takes the form
!
∂ui ∂uj 2
− u0i u0j = νt + − kδij (6)
∂xj ∂xi 3

where δij is the Kronecker delta function δij = 1 if i = j, and δij = 0 if i 6= j. k is the turbulent

3
 ui ui
kinetic energy defined as k = . A transport equation of k can be derived by forming the
2
dot product of ui with the RANS equations (3), but the result involves many terms.

k -  turbulence model
Now we need to model the turbulent viscosity νt . The k −  model, uses

Cµ k 2
νt = (7)

where Cµ is an empirical constant, and  is the rate of turbulent dissipation. A transport
∂u0 ∂
equations for  may be derived by applying the operator ν i to the RANS equations
∂xj ∂xj
(3), but again the result involves many complicated terms. Instead of incorporating all the
complexities of the analytically derived k and  equations, the k −  equation uses approximated
versions of them involving only the most important terms:
! !
Dk ∂ νt ∂k ∂ui ∂ui ∂uj
= + νt + − (8)
Dt ∂xj σk ∂xj ∂xj ∂xj ∂xi
! !
D ∂ νt ∂  ∂ui ∂ui ∂uj 2
= + C1 νt + − C2 (9)
Dt ∂xj σ ∂xj k ∂xj ∂xj ∂xi k
with empirically determined constants

Cmu = 0.09 C1 = 1.44 C2 = 1.92 σK = 1.0 σ = 1.3

These constants should ideally be tuned to the particular flow type at and. Let us now expand
the k equation to get a feeling of their complexity. Expanding the k equation term for term
gives:
!
∂ νt ∂k ∂ νt ∂k ∂ νt ∂k ∂ νt ∂k νt
       
1. = + + =∇· ∇k
∂xj σ ∂xj ∂x σ ∂x ∂y σ ∂y ∂z σ ∂x σ

∂ui ∂ui ∂u ∂u ∂v ∂v ∂w ∂w
2. = + + = [∇u · ∇u] + [∇v · ∇v] + [∇w · ∇w]
∂xj ∂xj ∂xj ∂xj ∂xj ∂xj ∂xj ∂xj
∂ui ∂uj ∂u ∂uj ∂v ∂uj ∂w ∂uj ∂u ∂u ∂u
3. = + + = [∇u · ] + [∇v · ] + [∇w · ]
∂xj ∂xi ∂xj ∂x ∂xj ∂y ∂xj ∂z ∂x ∂y ∂z

In 3 dimensions (1) consists of 3 terms, (2) consists of 9 terms, and (3) consists of 9 terms, so
the RHS of equations (8) consists of 22 terms including the additional  term. In 2D this is
reduced to a total of 10 terms. We also see that the  equation contains exactly the same tensor
terms.

Boundary treatment
The k −  equations written above are only valid for fully turbulent high Reynolds number flow,
and are thus not valid near the wall. Therefore these equations are not integrated all the way
to the walls. Instead we use one of two options: 1) patch a turbulent wall-law (a logarithmic
velocity profile) onto the solution near the walls, and 2) add damping functions to the equations

4
wall law treatment
let up be a velocity component at a node p closest to the wall. We then impose that the velocity
at this node should follow the logarithmic wall law according to
up 1 v ∗ yp
 
= ln +B (10)
v∗ κ ν

v∗2
Kp = p (11)
Cµ

v∗3
p = (12)
κyp
1/2
τw

where v ∗ is the wall friction velocity v ∗ = , κ ≈ 0.40 is the Von Karman constant
ρ

Implementation of the k -  model in openFoam

Initial and Boundary Condition
To set initial and boundary data for turbulence we need to specify the turbulence intensity
I which is the ratio of the magnitude of the turbulent fluctuations to the magnitude of the
characteristic mean velocity.
q q
1 02 2
3 (u + v 02 + w02 ) 3k 3
I= = → k = (Uref I)2
Uref Uref 2

I is a dimensionless quantity and its value typically has a value ranging from [0, 20]. I = [0, 1]
is considered low intensity, [1, 5] is medium intensity, and [5, 20] is considered high turbulent
intensity. Note also that if one has access to fluid dynamical experimental equipments, one may
measure the turbulent fluctuations quite accurately. For fully developed pipe flow we may use
the relation
1
I = 0.16Re− 8
As boundary condition on  we may use

0.164k 1.5
=
0.07L