You are on page 1of 37

Modelling Issues with Wind Turbines Wake in

Atmospheric Turbulence

Pierre-Elouan Réthoré Niels N. Sørensen Andreas Bechmann


pire@risoe.dtu.dk nsqr@risoe.dtu.dk andh@risoe.dtu.dk

Wind Energy Division · Risø DTU

The Science of Making Torque from the Wind


Forth · Heraklion · Crete · 29/6/10
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


1 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Introduction
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


2 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Introduction
Background

The goal is a steady-state model of wind


‡

farm wake in atmospheric turbulence


Fast enough for systematic analysis
‡

Réthoré et al. pire@risoe.dtu.dk


2 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Introduction
Background

The goal is a steady-state model of wind


‡

farm wake in atmospheric turbulence


Fast enough for systematic analysis
‡ ‡

The design choices are:


EllipSys (Elliptic CFD)
‡ ‡ ‡

Actuator disc model


Force distribution scaled with local
squared velocities (including wake
rotation and shear effect)
Mesh discretization: 10 cells / diameter
‡ ‡ ‡ ‡ ‡

Steady-state simulation
k-ϵ turbulence model
Neutral boundary layer
Assumed negligible: Coriolis effects,
boundary layer height and temperature
effects

Réthoré et al. pire@risoe.dtu.dk


2 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Introduction
Background
Conway’s Actuator Disc
1

Axial velocity U/ U∞ [-]


The goal is a steady-state model of wind
‡

0.9
Conway
farm wake in atmospheric turbulence 0.8 EllipSys

Fast enough for systematic analysis


‡ ‡

0.7 x = −0.5D
x = 0.0D
The design choices are: 0.6
x = 0.5D
EllipSys (Elliptic CFD) 0.5
‡ ‡ ‡

x = 2.5D
Actuator disc model 0.4

Force distribution scaled with local


squared velocities (including wake Full Rotor Computation
1
rotation and shear effect)

Axial velocity U/ U∞ [-]


Mesh discretization: 10 cells / diameter 0.9
‡ ‡ ‡ ‡ ‡

Steady-state simulation 0.8

k-ϵ turbulence model 0.7


Neutral boundary layer Full Rotor
0.6
Assumed negligible: Coriolis effects, AcDisc 10 c/ D
boundary layer height and temperature 0.5 AcDisc 20 c/ D
x = 3.0D
effects 0.4
1
0 0.5

Laminar cases fully validated (Fig)


‡

y-direction y/ D [-]

Réthoré et al. pire@risoe.dtu.dk


2 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Modelling Issues
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


3 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Modelling Issues
Comparison with Measurements
Nibe B Turbine
1.1

1.0
Axial velocity U/ U∞ [-]

0.9

0.8

0.7 k-ϵ 2.5D


Data 2.5D
0.6 k-ϵ 4D
Data 4D
0.5 k-ϵ 7.5D
Data 7.5D
0.4
−30 −20 −10 0 10 20 30
Relative wind direction [o ]
The standard k-ϵ model performs
Sexbierum badly in comparison with
1.1

1.0 measurementsa b c .
Axial velocity U/ U∞ [-]

0.9 a El Kasmi & Masson, Wind Energy 2008


0.8 b Cabezon et al. EWEC 2009
c Rados et al. EWEC 2009
0.7 k-ϵ 2.5D
Data 2.5D
0.6 k-ϵ 5.5D
Data 5.5D
0.5 k-ϵ 8D
Data 8D
0.4
−30 −20 −10 0 10 20 30
Relative wind direction [o ]

Réthoré et al. pire@risoe.dtu.dk


3 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Modelling Issues
Comparison with LES

Réthoré et al. pire@risoe.dtu.dk


4 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Modelling Issues
Comparison with LES

LES, Variable forces: U−velocity U/UH,∞ [−]

y−direction y/D [−]


2 1

0.8
0
0.6

−2 0.4
−4 −2 0 2 4 6 8 10 12 14
x−direction x/D [−]
LES, Constant forces: U−velocity U/UH,∞ [−]
y−direction y/D [−]

2 1

0.8
0
0.6

−2 0.4
−4 −2 0 2 4 6 8 10 12 14
x−direction x/D [−]
k−ε: U−velocity U/UH,∞ [−]
y−direction y/D [−]

2 1

0.8
0
0.6

−2 0.4
−4 −2 0 2 4 6 8 10 12 14
x−direction x/D [−] pire@risoe.dtu.dk
Réthoré et al.
5 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


6 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Eddy-viscosity: Original Idea

Turbulent momentum transport with 3 assumptions


A1 Velocity conservation during time τ:
U(xp ) = U(x)
xp1
A2 Particles mix during time τ:
xp2
u′ (xp )u′ (x) = 0
x
A3 Mean velocity field is linear over
xp3
turbulent length-scale L:
∂U
U(xp ) − U(x) = Lj ∂x xp4
j

Réthoré et al. pire@risoe.dtu.dk


6 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Eddy-viscosity: Original Idea

Turbulent momentum transport with 3 assumptions


A1 Velocity conservation during time τ:
U(xp ) = U(x)
xp1
A2 Particles mix during time τ:
xp2
u′ (xp )u′ (x) = 0
x
A3 Mean velocity field is linear over
xp3
turbulent length-scale L:
∂U
U(xp ) − U(x) = Lj ∂x xp4
j

Derivation of the Reynolds-stresses with the 3 assumptions


 ‹
∂Uj
Eddy-viscosity concept: ui′ uj′ = 23 kδij − νt ∂U
∂x
i
+ ∂x
?
j i

Réthoré et al. pire@risoe.dtu.dk


6 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Velocity Conservation (A1)

The velocity is not conserved in the region of adverse pressure


gradient surrounding the wind turbine

x
p x
Axial velocity U/ U∞ [-]

U( x p)
Error
U( x )

xp x x-direction x/ D [-]

Réthoré et al. pire@risoe.dtu.dk


7 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Velocity Conservation (A1)

LES shows that the eddy-viscosity concept generates a too high


Normal Reynolds-stress R11 in the area surrounding the turbine

LES: R11
60
LES: u′ u′
50
k-ϵ: R11
R11 / u∗2 [-]

40

30

20

10

0
0 5 10
x-direction x/ D [-]

2
2
where R11 = 3k − 2Cμ kϵ ∂U
∂x
Réthoré et al. pire@risoe.dtu.dk
8 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Velocity Linearity (A3)

A3 velocity linearity over the turbulent length-scale:


U(yp ) − U(y) = L ∂U
∂y
The assumption (A3) is only valid when the velocity changes occur
over a larger distance than the turbulent length-scale L
L
Axial velocity U/ U∞ [-]

L ∂U
∂y U(yp ) − U(y)

yp y y-direction y/ D [-]

Réthoré et al. pire@risoe.dtu.dk


9 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
Velocity Linearity (A3)

However, in the wake boundary region, the velocity is highly


non-linear over the turbulent length-scale.

Axial velocity U/ U∞ [-] L

Error
L ∂U
∂y

U(yp ) − U(y)

yp y y-direction y/ D [-]

Réthoré et al. pire@risoe.dtu.dk


10 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
LES Eddy-viscosity Factor Cμ

The eddy-viscosity factor illustrates where the eddy-viscosity


concept is invalid.

LES Eddy-viscosity factor Cμ [-]


3 0.1
y-direction y/ D [-]

2 0.08
1
0.06
0
0.04
−1
−2 0.02

−3 0
−4 −2 0 2 4 6 8 10 12 14
x-direction x/ D [-]

?
Réthoré et al. pire@risoe.dtu.dk
11 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
Eddy-viscosity
LES Eddy-viscosity Factor Cμ

The eddy-viscosity factor illustrates where the eddy-viscosity


concept is invalid.

LES Eddy-viscosity factor Cμ [-]


3 0.1
y-direction y/ D [-]

2 0.08
1
0.06
0
0.04
−1
−2 0.02

−3 0
−4 −2 0 2 4 6 8 10 12 14
x-direction x/ D [-]

Even if k and ϵ were correct, the Eddy-viscosity concept would still


give the wrong Reynolds-stresses!
?
Réthoré et al. pire@risoe.dtu.dk
11 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Invalid Assumptions
k-ϵ
Wind Turbines Extract Large-scale TKE

The forces on the blades can adapt to large-scale velocity


‡

fluctuations.
TKE Comparison of 2 LES with unsteady and steady forces
‡

LES - Unsteady Forces


LES - Steady Forces
30
k-ϵ - Steady Forces
TKE k/ u∗2 [-]

20

10

0
0 4 8 12
x-direction x/ D [-]

There is a small but visible decrease of TKE at the turbine with the
unsteady forces
Réthoré et al. pire@risoe.dtu.dk
12 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


13 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
El Kasmi-Masson
Compared to LES
Added term in ϵ-equation: Sϵ = Cϵ4 P2t / ρk
‡ ‡

EKM can have the same velocity deficit as LES, but not the
same recovery rate.
The TKE has a different trend compared with LES.
‡ ‡

The dissipation ϵ is in clear contrast with LES.


EKM: Cϵ4 = 0.001
1
EKM: Cϵ4 = 0.005
U/ U∞,H [-]

0.8 EKM: Cϵ4 = 0.01


Std k-ϵ
0.6
LES
0.4
40
k/ u∗2 [-]

30

20

10

0
50
ϵ/ ϵ ∞,H [-]

40
30
20
10
0
2
22 0 2 4 6 8 10 12

13 of 26
Réthoré et al. x-direction x/ D [-] pire@risoe.dtu.dk
Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
El Kasmi-Masson
Compared to Sexbierum Measurements
Not enough turbulence
‡ ‡

Cannot satisfy all the met.masts measurements


[−]

x=2.5D x=5.5D x=8D


H,∞
Normalized axial velocity U/U

1 C =0.005
ε4
C =0.01
ε4
0.8
C =0.03
ε4
0.6 C =0.37
ε4

0.4 Std−k−ε
−20 0 20 −20 0 20 −20 0 20 data
o
Relative wind direction [ ]

x=2.5D x=5.5D x=8D


Turbulence Intensity k/U [−]
2

0.1

0.05

0
−20 0 20 −20 0 20 −20 0 20
o
Relative wind direction [ ]
Réthoré et al. pire@risoe.dtu.dk
14 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
Realizability
Compared to LES
2
Realizability criteria (hacked): ui′ uj′ ≤ CSwz ui′ ui′ · uj′ uj′
‡ ‡ ‡

Only physical for CSwz = 1.0.


It can have the same velocity deficit as LES, and the same
recovery rate.
Similar TKE and dissipation trends.
‡

RM: CSwz = 1.0


1
RM: CSwz = 0.5
U/ U∞,H [-]

0.8 RM: CSwz = 0.4


Std k-ϵ
0.6
LES
0.4
40
k/ u∗2 [-]

30

20

10

0
30
ϵ/ ϵ ∞,H [-]

20

10

0
Réthoré et al. 2
22 0 2 4 6 8 10 12 pire@risoe.dtu.dk
15 of 26 Risø DTU x-direction x/ DTurbine
Wind [-] Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
Realizability
Compared to Sexbierum Measurements
Not enough wake spreading on the side
‡ ‡ ‡

Too much turbulence in the far wake region


Cannot satisfy all the met.masts measurements
[−]

x=2.5D x=5.5D x=8D


H,∞
Normalized axial velocity U/U

1 C =1.0
Swz
CSwz=0.5
0.8
CSwz=0.3
0.6 CSwz=0.2

0.4 Std−k−ε
−20 0 20 −20 0 20 −20 0 20 data
o
Relative wind direction [ ]

x=2.5D x=5.5D x=8D


Turbulence Intensity k/U [−]
2

0.1

0.05

0
−20 0 20 −20 0 20 −20 0 20
Réthoré et al. o
Relative wind direction [ ] pire@risoe.dtu.dk
16 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
k-ϵ Modifications
Wind Turbine Canopy
Compared to LES
Added terms in k and ϵ-equation: Sk = −CF βd Uk and
‡

Sϵ = −CF Cϵd βd Uϵ
It is not enough to correct the k-ϵ model
‡

WTCM: βd = 1, Cϵd = 1
1
WTCM: βd = 1, Cϵd = 0
U/ U∞,H [-]

0.8 WTCM: βd = 2, Cϵd = 0


Std k-ϵ
0.6
LES
0.4
40
k/ u∗2 [-]

30

20

10

0
50
ϵ/ ϵ ∞,H [-]

40
30
20
10
0
2
22 0 2 4 6 8 10 12
x-direction x/ D [-]
Réthoré et al. pire@risoe.dtu.dk
17 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Beyond the Eddy-viscosity
Outline

1 Introduction

2 Modelling Issues

3 Invalid Assumptions
Eddy-viscosity
k-ϵ

4 k-ϵ Modifications
El Kasmi-Masson
Realizability
Wind Turbine Canopy

5 Beyond the Eddy-viscosity

Réthoré et al. pire@risoe.dtu.dk


18 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Beyond the Eddy-viscosity
How to model turbulence, then?

Design constraints
The Reynolds-stresses should be transported.
‡ ‡

The pressure gradient has to be taken into account.

Réthoré et al. pire@risoe.dtu.dk


18 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Beyond the Eddy-viscosity
How to model turbulence, then?

Design constraints
The Reynolds-stresses should be transported.
‡ ‡

The pressure gradient has to be taken into account.

Reynolds-stresses Transport Model (RSM)?


Are the pressure gradient and body forces accounted for?
‡ ‡

How are modelled the transport of higher order terms? Do they


rely on the linearity over length scale assumption?

Réthoré et al. pire@risoe.dtu.dk


18 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Summary and future work

None of the k-ϵ wake models studied are physical: they need to
‡

be recalibrated in every types of inflow conditions, wind farm


spacing etc..
It is not possible to fix the invalid assumptions of the
‡

eddy-viscosity concept by only acting on k and/or ϵ.


The extraction of large-scale turbulence or “turbulent cascade
‡

short-circuit” concept is not enough to explain the errors of the


standard k-ϵ.
We need another way to model the Reynolds-stresses for the
‡

wind turbine wake in atmospheric turbulence.


Future studies:
‡

Reynolds-stresses transport models


‡ ‡

Non local turbulence closure models (e.g. Stull’s TTT)

Réthoré et al. pire@risoe.dtu.dk


19 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Thank you for your attention!

Réthoré et al. pire@risoe.dtu.dk


20 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Eddy-viscosity: Original Idea
Based on turbulent momentum transport with 3 assumptions:
‡

A1 Velocity conservation during time τ: xp1


U(xp ) = U(x)
xp2
A2 Particles mix during time τ: x
u′ (xp )u′ (x) = 0
xp3
A3 Mean velocity field is linear over turbulent
∂U
length-scale L: U(xp ) − U(x) = Lj ∂x xp4
j

Derivation of the Reynolds-stresses based on the 3 assumptions


‡

€ Š € Š A1
:
A1 u′ (x) = u′ (xp ) + U(xp ) − U(x) + U(x) − U(xp )
: A2 € Š
A2 u′ (x)v′ (x) = v′ (x)u′ (xp ) + v′ (x) U(xp ) − U(x)
∂U
A3 u′ (x)v′ (x) = v′ (x)Lj ∂x
j
 ‹
∂Uj
Eddy-viscosity concept: ui′ uj′ = 23 kδij − νt ∂Ui
+
‡

∂xj ∂xi
Back

Réthoré et al. pire@risoe.dtu.dk


21 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Conway’s exact actuator disc model
Conway (1998) has derived an analytical solution for an axis
‡

symmetric non-uniform and heavily loaded actuator disc,


including the slipstream contraction.
The vorticity in the slipstream of a turbine is equivalent of a
‡

volume distribution of ring vortices


The flow induced by a ring vortex or radius r ′ and strength Γ
‡

can be derived Ras:


Γrr ′ ∞ ′
Ψ(r, z) = J (sr ′ )J1 (sr)e−s|z−z | ds
‡

2 0 1
−Γsign(z−z ′ )r ′ R ∞ ′
Vr (r, z) = sJ1 (sr ′ )J1 (sr)e−s|z−z | ds
‡

2 0
′ ∞ ′
Vz (r, z) = Γr2 0 sJ0 (sr ′ )J1 (sr)e−s|z−z | ds
R
‡

By superposition, the flow fields induced by the vorticity


‡

distribution ω ϕ (r, z) can be derived


r ∞ R(z) ∞ ′
R R R
Ψ(r, z) = ω ϕ (r ′ , z ′ )r ′ J1 (sr ′ )J1 (sr)e−s|z−z | ds dr ′ dz ′
‡

2 0 0 0
∞ R(z) ∞ ′
Vr (r, z) = 12 0 0
R R R
±ω ϕ (r ′ , z ′ )r ′ sJ1 (sr ′ )J1 (sr)e−s|z−z | ds dr ′ dz ′
‡

0
∞ R(z) ∞ ′
Vz (r, z) = 12 0 0
R R R
ω ϕ (r ′ , z ′ )r ′ sJ0 (sr ′ )J1 (sr)e−s|z−z | ds dr ′ dz ′
‡

Réthoré et al. pire@risoe.dtu.dk


22 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Conway’s exact actuator disc model for heavily
loaded turbines
Simplest expression for the vorticity: ω ϕ (r, z) = ar. This
‡

corresponds to a parabolic wake profile: Vz (r, ∞) = 2a R2w − r 2 .


€ Š

After integration with respect to z’ and including the additional


‡

stream function due to the free stream U∞


U∞ r 2
R∞
Ψ(r, z) = + ar R2 (z ′ )I(−1,2,1) (R(z ′ ), r, z − z ′ ) dz ′
‡

2R 2 0
a ∞
Vr (r, z) = 2 0 ±R2 (z ′ )I(0,2,1) (R(z ′ ), r, z − z ′ ) dz ′
‡

R∞
Vz (r, z) = U∞ + ar R2 (z ′ )I(0,2,0) (R(z ′ ), r, z − z ′ ) dz ′
‡

2 0
The wake width R(z) can be determined recursively by noticing
‡

that the stream function is constant on the slipstream boundary


Ψ(RT , 0) = Ψ(R(z), z)
‡

The Bessel-Laplace integrals


‡

R∞
I(λ,μ,ν)(R, r, z) = 0 sλ Jμ (sR)Jν (sr)e−s|z| ds are related to one
another using recursive rules, and can be evaluated in terms of
complete elliptic integrals.
The load distribution is directly related to the stream function:
‡

L(r) = aρ (Ψ(RT , 0) − Ψ(r, 0))


Réthoré et al. pire@risoe.dtu.dk
23 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Solving Bessel-Laplace integrals
It is possible to derive recursive relations between
‡

Bessel-Laplace integrals:
I(λ,μ,ν)(R, r, |z|) = I(λ,ν,μ)(r, R, |z|)
‡ ‡

4(ν−1)ω
I(0,ν,ν) = I
2ν−1 (0,ν−1,ν−1)
− 2ν−3 I
2ν−1 (0,ν−2,ν−2)
,
(2ν−1)|z|k 4 € Š
I(1,ν,ν) = 8Rr(1−k 2) I(0,ν−1,ν−1) − ωI(0,ν,ν) ,
‡

I(0,ν+1,ν) = Rr I(0,ν,ν−1) + 2ν
r
€ Š
I(1,ν+1,ν+1) − I(1,ν−1,ν−1) for ν 6= 0,
‡

2(ν+1)|z| 2(ν+1)R (ν+1+μ)


I(0,μ,ν) = r(ν+1−μ) I(0,μ,ν+1) − r(ν+1−μ) I(0,μ−1,ν+1) + (ν+1−μ) I(0,μ,ν+2),
‡

R
€ Š
I(λ,μ,ν) = 2μ I(λ+1,μ+1,ν) − I(λ+1,μ−1,ν) for λ < 0,
‡

It is possible to express some basic Bessel-Laplace integrals in


‡

terms of complete elliptic functions:


kK(k)
I(0,0,0)(R, r, |z|) = p .
‡

π hrR i
1 |z|kK(k) Λ (|β|,k)
I(0,1,0)(R, r, |z|) = R 1 − p − 0 2 (r < R)
‡

h 2π irR
|z|kK(k) Λ (|β|,k)
= R1 − p + 0 2 (r > R)
2π rR
[(2−k 2 )K(k)−2E(k)]
I(0,1,1)(R, r, |z|) = p
‡

πk rR

Réthoré et al. pire@risoe.dtu.dk


24 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
Comparison between Conway’s and EllipSys’
actuator disc models
The load distribution L(r) found with Conway’s model for a factor
a = − UR∞2 is used as an input in the actuator disc model of EllipSys.
T

Axial velocity at various axial coordinate z (CT=−0.4484) Radial velocity at various axial coordinate z (CT=−0.4484)
1.1 0.16
Conway: z=−0.5D
Normalized axial velocity V /U [−]

1 0.14 EllipSys: z=−0.5D

Normalized axial radial V /U [−]


Conway: z=0D


0.12
z

0.9 EllipSys: z=0D

r
0.1 Conway: z=0.5D
0.8 EllipSys: z=0.5D
0.08 Conway: z=2.5D
0.7 EllipSys: z=2.5D
0.06
0.6
0.04

0.5 0.02

0.4 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Normalized distance from center [−] Normalized distance from center [−]

Back

Réthoré et al. pire@risoe.dtu.dk


25 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10
LES Eddy-viscosity factor Cμ
LES Eddy-viscosity factor Cμ [-]
3 0.1
y-direction y/ D [-]

2 0.08
1
0.06
0
0.04
−1
−2 0.02

−3 0
−4 −2 0 2 4 6 8 10 12 14
x-direction x/ D [-]

||R||
where Cμ = ||S|| kϵ2 is here defined as the ratio between:
Reynolds-stress tensor norm
‡

q  
2 ′ ′ 2 ′ ′
||R|| = 3 kδ ij − u u
i j 3 kδ ij − u u
i j
ǁ ‹ ‹
∂Ui ∂Uj ∂Ui ∂Uj
Strain-rate tensor norm ||S|| = ∂x + ∂x ∂x + ∂x
‡

j i j i
back

Réthoré et al. pire@risoe.dtu.dk


26 of 26 Risø DTU Wind Turbine Wake in Atmospheric Turbulence · 29/6/10

You might also like