Lars Davidson: MTF270 Turbulence Modelling

http://www.tfd.chalmers.se/doct/comp turb model 42

The SST Model April 23, 2007

The SST (Shear Stress Transport) model of Menter (1994)

is an eddy-viscosity model which includes two main novel-
ties:

1. It is combination of a k − ω model (in the inner bound-

ary layer) and k − ε model (in the outer region of and
outside of the boundary layer);
2. A limitation of the shear stress in adverse pressure
gradient regions is introduced.

The k − ε model has two main weaknesses: it over-predicts

the shear stress in adverse pressure gradient flows because
of too large length scale (due to too low dissipation) and it
requires near-wall modification (i.e. low-Re number damp-
ing functions/terms)

Pressure contours. Red: high pressure; blue:

low pressure

The SST Model

Lars Davidson: MTF270 Turbulence Modelling
http://www.tfd.chalmers.se/doct/comp turb model 43

One example of adverse pressure gradient is the flow

along the surface of an airfoil, see figure above. Consider
the upper surface (suction side). Starting from the leading
edge, the pressure decreases because the velocity increases.
At the crest (at x/c ' 0.15) and further downstream, the
pressure increases since the velocity decreases. This region
is called the adverse pressure gradient (APG) region.
The k−ω model is better at predicting predicting adverse
pressure gradient flow and the standard model of Wilcox
(1988) does not use any damping functions. However, the
disadvantage of the standard k − ω model is that it is de-
pendent on the free-stream value of ω (Menter, 1992)
In order to improve both the k − ε and the k − ω model,
Menter (1994) suggested to combine the two models. Before
doing this, it is convenient to transform the k − ε model into
a k − ω model using the relation ω = ε/(β ∗k), where β ∗ = cµ .
The transformation of the left-hand side (LHS) reads
Dω Dε/(β ∗k) 1 Dε ε Dk 1 Dε ω Dk
= = ∗ − ∗ 2 = ∗ − (69)
Dt Dt β k Dt β k Dt β k Dt k Dt
where D/Dt = Uj ∂/∂xj denotes the material derivative. In-
serting the modelled equations for k and ε yields:
   
Dω 1 ω 1 ω
= ∗ P ε − P k − ∗ Ψε − Ψk +
Dt β k k β k k
| {z } | {z }
Production, Pω Destruction, Ψω
  " 2 2
# (70)
1 T ω T ν ∂ ε νω ∂ k
D − Dk + ∗ +
β ∗k ε k β k ∂x2j k ∂x2j
| {z } | {z }
Turbulent diffusion, Dω
T
Viscous diffusion, Dων

The SST Model

Lars Davidson: MTF270 Turbulence Modelling
http://www.tfd.chalmers.se/doct/comp turb model 44

• Production term
1 ω ε ω
Pω = P ε − P k = C ε1 P k − Pk =
β ∗k k β ∗k 2 k (71)
ω
= (Cε1 − 1) Pk
k

• Destruction term
1 ω ε2 ω
Φω = ∗ Ψε − Ψk = Cε2 − ε =
β k k k k (72)
∗ 2
= (Cε2 − 1)β ω

• Viscous diffusion term

ν ∂ 2ε νω ∂ 2k ν ∂ 2ωk νω ∂ 2k
Dων = ∗ − = − =
β k ∂x2j k ∂x2j k ∂x2j k ∂x2j
  
ν ∂ ∂k ∂ω ω ∂ 2k
= ω +k −ν =
k ∂xj ∂xj ∂xj k ∂x2j
"
2 2
# (73)
ν ∂ω ∂k ∂ k ∂k ∂ω ∂ ω ω ∂ 2k
= +ω 2 + +k 2 −ν =
k ∂xj ∂xj ∂xj ∂xj ∂xj ∂xj k ∂x2j
 
2ν ∂ω ∂k ∂ ∂ω
= + ν
k ∂xj ∂xj ∂xj ∂xj

The turbulent diffusion term is obtained as (the deriva-

tion can be downloaded (Bredberg, 2000))
 
T 2νt ∂k ∂ω ω νt νt ∂ 2 k
Dω = + − +
σε k ∂xj ∂xj k σε σk ∂x2j
    (74)
ω 1 1 ∂νt ∂k ∂ νt ∂ω
+ − +
k σε σk ∂xj ∂xj ∂xj σε ∂xj
In the standard k − ε model we have σk = 1 and σε = 1.3.
If we assume that σk = σε in the second and third term of

The SST Model

Lars Davidson: MTF270 Turbulence Modelling
http://www.tfd.chalmers.se/doct/comp turb model 45

the right-hand side, we can considerably simplify the tur-

bulence diffusion so that
 
2ν ∂k ∂ω ∂ ν ∂ω
(75)
t t
DωT = +
σε k ∂xj ∂xj ∂xj σε ∂xj

We can now finally write the ε equation formulated as

an equation for ω
  
∂ ∂ νt ∂ω ω
(Ūj ω) = ν+ + α Pk − βω 2
∂xj ∂xj σ ∂xj k
 ω
(76)

2 νt ∂k ∂ω
+ ν+
k σε ∂xi ∂xi
α = Cε1 − 1 = 0.44, β = (Cε2 − 1)β ∗ = 0.0828
Since this equation will be used for the outer part of the
boundary layer, the viscous part in the last term is omitted.
In the SST model the coefficients are smoothly switched
from k − ω values in the inner region of the boundary layer
and k − ε values in the outer region. Functions of the form
" ( √ ) #
k 500ν 4σω2k
F1 = tanh(ξ 4), ξ = min max , , (77)
β ∗ωy y 2 ω CDω y 2

are used. F1 = 1 in the near-wall region and F = 0 in the

outer region.
At p. 42 it was mentioned that the k − ω model is bet-
ter than the k − ε model in predicting adverse pressure-
gradient flows because it predicts a smaller shear stress.
Still, the predicted shear stress is too large. This brings us
to the second modification (see p. 42). When introducing
this second modification, Menter (1994) noted that a model
(the Johnson - King model [JK]) which is based on trans-
port of the main shear stress uv, predicts adverse pressure

The SST Model

Lars Davidson: MTF270 Turbulence Modelling
http://www.tfd.chalmers.se/doct/comp turb model 46

gradient flows much better than the k − ω model. In the

JK model, the uv transport equation is built on Bradshaw’s
assumption (Bradshaw et al., 1967)

−uv = a1 k (78)

where a1 = cµ = β ∗1/2. In boundary layer flow, the Boussi-

1/2

nesq assumption can be written as

k ∂U k cµ k 2
−uv = = Ω= Ω=
ω ∂y ω ε
 1/2  1/2  1/2 (79)
cµ k 2 2 Pk Pk
cµ1/2k Ω 1/2
= cµ k = a1 k
ε2 ε ε
where Ω is the vorticity (in boundary layer flow Ω = ∂U/∂y).
It is found from experiments that in adverse pressure gra-
dient flow the production is much larger than the dissipa-
tion (Pk > ε), which explains why Eq. 79 over-predicts the
shear stress; the model works poorly in this type of flow.
To reduce |uv| in Eq. 79 in adverse pressure gradient flow,
Menter (1994) proposed to re-define the turbulent eddy vis-
cosity as (cf. Eq. 78)
a1 k
−uv = νt Ω ⇒ −νt = (80)
max(a1 ω, F2Ω)
where F2 is a damping function (similar to F1) which is 1
near walls and zero elsewhere. When the production is
large (i.e. when Ω is large), Eq. 80 reduces νt . It is im-
portant to ensure that this limitation is not active in usual
boundary layer flows where Pk ' ε. It can be seen that
the shear stress is reduced only in regions where Pk > ε,
because if Pk > ε then Ω > a1 ω since
ωPk ωε
Ω2 = > = cµ ω 2 = a21 ω 2 (81)
k k
The SST Model
Lars Davidson: MTF270 Turbulence Modelling
http://www.tfd.chalmers.se/doct/comp turb model 47

In regions where Pk ≤ ε, Eq. 80 returns to νt = k/ω as it

should.
Presently, the SST model has been slightly further de-
veloped. Two modifications have been introduced (Menter
et al., 2003b). The absolute vorticity Ω in Eq. 80 has been
replaced by S = 2Sij Sij . This limits νt in stagnation regions
similar to Eq. 42. The production term is in the new SST
model limited by 10ε, i.e.

Pk,new = min (Pk , 10ε) (82)

The final form of the SST model is given in Eq. 165 at p. 95.

The SST Model