Development of an Enhanced Actuator Disc Model for the

Simulation of Wind Farms

Asif Zubair

Thesis to obtain the Master of Science Degree in

Energy Engineering and Management

Supervisors: Prof. José Alberto Caiado Falcão de Campos

Dipl.-Ing. Matthias Kretschmer

Examination Committee
Chairperson: Prof. Edgar Caetano Fernandes
Supervisor: Prof. José Alberto Caiado Falcão de Campos
Member of the Committee: Prof. Luís Rego da Cunha Eça

November 2016

I
 ACKNOWLEDGEMENT
First of All, I would like to thank KIC InnoEnergy for giving me the opportunity for valuable M.Sc. ENTECH
program. By completing this program, I have reached an important milestone in my life of doing a master
degree in the field of energy.
Secondly, I would like to extend words of thanks to my thesis supervisor Prof. José Alberto Caiado Falcão
de Campos at IST for approving this thesis and allowing me to conduct this thesis at Stuttgart Wind Energy
(SWE) at Stuttgart University. His lectures in the course of “Offshore wind energy” were a major stimulation
for me to pursue my master thesis in the field of aerodynamics of wind turbines. His immense support,
patience, motivation and knowledge throughout the course of my thesis has helped me a lot in conducting
and finishing the master thesis.

Further, I would like to thank my thesis supervisor Dip.- Ing Matthias Kretchhmer at SWE for giving me the
wonderful opportunity of this master thesis in his institute. I express my sincere gratitude to him for helping
me understand the thesis, guiding me at each and every turn of the thesis, sharing his knowledge with me
over the subject and for being patient in answering all my questions.

Lastly, I would like to thank Prof. Chen and his entire team for allowing me to use the resources of his
institute, including computer labs, for my thesis and providing supportive and friendly atmosphere for work
at all times during the course of my thesis.

II
 ABSTRACT
Computational Fluid Dynamics (CFD) is one of many tools available for predicting the performance and the
flow characteristics around a horizontal axis wind turbine. CFD simulations of a fully resolved turbine
geometry gives good results but it is computationally very expensive. An alternate approach is to use a
simplified CFD-BEMT approach in which the turbine is replaced by an actuator disc and coupled to the
Blade Element Momentum Theory (BEMT). The blades of a wind turbine generate tip vortices. These tip
vortices reduce the lift and hence efficiency, particularly near the tip region. The CFD-BEMT model is not
capable of capturing the effect of the tip vortices on the performance as well as on the flow field. In this
master thesis, three enhanced actuator disc models are presented which are developed by combining
BEMT with Lifting Line Theory (LLT) and coupling them to CFD. The three methods use different
approaches to combine BEMT with LLT. Simulations are performed for CFD-BEMT, with and without the
Prandtl tip loss factor, and with the three enhanced actuator disc models and results are produced for the
performance of the turbine as well as for the flow field. The Performance result are compared with the
results obtained from the FAST V8 code and the local flow field results are compared with the results
obtained from the LLT IST code. The results show that none of the enhanced actuator disc models is
capable of accurately predicting performance and the local flow field simultaneously.

Keywords: Computational Fluid Dynamics (CFD), Blade Element Momentum Theory (BEMT), flow
field, Lifting Line Theory (LLT), tip vortices.

III
 RESUMO

A Mecânica dos Fluidos Computacional (CFD) é uma das ferramentas disponíveis para prever o
desempenho e as características do escoamento de uma turbina eólica de eixo horizontal (HAWT). A
simulação de CFD-RANS de uma turbina de geometria 3D completamente discretizada é muito exigente
em termos computacionais. Uma abordagem alternativa é a utilização de um modelo de disco atuador com
a Teoria do Momento do Elemento de Pá (BEMT). Este modelo de CFD-BEMT não é capaz de considerar
o efeito de vórtices de ponta. Nesta tese de mestrado, três modelos melhorados de discos atuadores são
apresentados,através da combinação de BEMT com a Teoria da Linha Sustentadora (LLT) e combinadas
com o modelo de RANS. Os três métodos usam diferentes abordagens para combinar BEMT com LLT. As
simulações são efetuadas para CFD-BEMT, com e sem fator de perda na ponta de Prandtl, e com os três
modelos melhorados de disco atuador. Os resultados são produzidos para o desempenho da turbina, bem
como para o campo de escoamento. O resultado de desempenho desse modelo é comparado com o
resultado obtido a partir do código FAST V8 e os resultados de campo de escoamento comparados com
os resultados do código LLT. Os resultados mostram que nenhum dos modelos de disco atuador
melhorado é capaz de prever simultaneamente de forma correta o desempenho da turbina e o campo de
escoamento. Quando há uma melhoria na previsão do desempenho, há uma deterioração da previsão do
campo de escoamento, e vice versa.

Palavras-chave: Mecânica dos Fluidos Computacional (CFD), Teoria do Momento do Elemento de Pá

(BEMT), Campo de escoamento, Teoria da linha Sustentadora (LLT), Vórtice de ponta

IV
 TABLE OF CONTENTS
ACKNOWLEDGEMENT.................................................................................................................................... II
ABSTRACT…….. ................................................................................................................................. III
RESUMO............. ................................................................................................................................. IV
TABLE OF CONTENTS ....................................................................................................................................V
LIST OF TABLES ...........................................................................................................................................VIII
LIST OF FIGURES ........................................................................................................................................... IX
NOMENCLATURE ............................................................................................................................................ XI
CHAPTER 1: INTRODUCTION ....................................................................................................................... 1
1.1 Background .............................................................................................................................................. 1
1.2 Motivation and Purpose.......................................................................................................................... 1
1.3 Scope of thesis ........................................................................................................................................ 3
CHAPTER 2: MODELS FOR WIND TURBINE ANALYSIS......................................................................... 5
2.1 Basics ........................................................................................................................................................ 5
2.2 Blade Element Momentum Theory (BEMT) ........................................................................................ 6
2.2.1 Momentum Theory........................................................................................................................... 6
2.2.2 Blade Element Theory..................................................................................................................... 7
2.2.3 Blade Element Momentum Theory................................................................................................ 8
2.2.4 Tip Loss: Effect on Power Coefficient of Number of Blades ..................................................... 9
2.3 Lifting Line Theory (LLT) ................................................................................................................. 10
CHAPTER 3: CFD-MODELS ......................................................................................................................... 14
3.1 The RANS equations ............................................................................................................................ 14
3.2 CFD-BEMT model ................................................................................................................................. 14
3.3 Enhanced actuator disc models (CFD-BEMT+LLT Models)........................................................... 16
3.3.1 BEMT region ................................................................................................................................... 17
3.3.2 BEMT+LLT region.......................................................................................................................... 17
3.3.2.1 Simple Momentum Source Terms: ....................................................................................... 17
3.3.2.2 Additional momentum source terms. ................................................................................... 17
3.3.2.2.1 Method 1 (M1) .................................................................................................................. 18
3.3.2.2.2 Method 2 (M2) .................................................................................................................. 18
3.3.2.2.3 Method 3 (M3) .................................................................................................................. 19
3.4 Calculation of power coefficient (𝐶𝑃) and thrust coefficient (𝐶𝑇) ................................................... 19
CHAPTER 4: COMPUTATIONAL CODES .................................................................................................. 20

V
 4.1 ANSYS-CFX........................................................................................................................................... 20
4.1.1 Introduction to ANSYS-CFX ......................................................................................................... 20
4.1.2 The Structure of ANSYS-CFX...................................................................................................... 20
4.1.2.1 CFX-Pre.................................................................................................................................... 20
4.1.2.2 CFX-Solver .............................................................................................................................. 20
4.1.2.3 CFX-Solver Manager .............................................................................................................. 20
4.1.2.4 CFD-Post.................................................................................................................................. 20
4.2 FAST V8 ................................................................................................................................................. 21
4.3 LLT IST code ........................................................................................................................................ 21
CHAPTER 5: REFERENCE TURBINE FOR CFD SIMULATIONS .......................................................... 22
5.1 Gross properties of turbine .................................................................................................................. 22
5.2 Blade aerodynamic Properties ............................................................................................................ 22
CHAPTER 6: ANSYS-CFX SIMULATION SETUP ..................................................................................... 24
6.1 Overview ................................................................................................................................................. 24
6.2 Boundary Conditions............................................................................................................................. 24
6.2.1 Inlet: ................................................................................................................................................. 25
6.2.2 Outlet: .............................................................................................................................................. 25
6.2.3 Top: .................................................................................................................................................. 26
6.2.4 Ground:............................................................................................................................................ 26
6.2.5 Sides: ............................................................................................................................................... 26
6.3 Mesh structure of fluid domain ............................................................................................................ 26
6.3.1 Actuator disc mesh region (ADMR)............................................................................................. 26
6.3.2 Fine field mesh region (FFMR) .................................................................................................... 27
6.3.3 Coarse field mesh region (CFMR)............................................................................................... 27
6.4 Mesh dependence check ..................................................................................................................... 28
6.5 The FORTRAN Code to calculate momentum source terms. ........................................................ 29
6.5.1 Value_Readin Subroutine ............................................................................................................. 29
6.5.2 Lifting_Line Subroutine ................................................................................................................. 29
6.5.3 Source_Term Subroutine.............................................................................................................. 31
CHAPTER 7: SIMULATION RESULTS AND DISCUSSION ..................................................................... 32
7.1. Overview of the chapter ...................................................................................................................... 32
7.2 Results and discussion for one turbine case..................................................................................... 32
7.2.1 Performance ................................................................................................................................... 33
7.2.1.1 Power Coefficient .................................................................................................................... 33

VI
 7.2.1.2 Thrust coefficient ..................................................................................................................... 34
7.2.2 Local Flow field parameters ......................................................................................................... 35
7.2.2.1 Axial induction factor .............................................................................................................. 35
7.2.2.2 Tangential Induction Factor ................................................................................................... 36
7.2.2.3 Angle of Attack ........................................................................................................................ 37
7.2.2.4 Drag coefficient ....................................................................................................................... 38
7.2.3 Axial velocity field........................................................................................................................... 39
7.2.3.1 Average axial velocity upstream and downstream of the turbine .................................... 39
7.2.3.2 Axial Velocity field displayed on a longitudinal plane (xy-plane) passing through
turbine center ........................................................................................................................................ 41
7.2.3.3 Axial velocity field downstream of turbine at 200m and 600m ......................................... 42
7.2.4 Turbulence Kinetic Energy ........................................................................................................... 46
7.3 Results and discussion for two turbines in a row case .................................................................... 48
7.3.1 Power and thrust of second turbine ............................................................................................ 48
CHAPTER 8: SUMMARY AND CONCLUSION .......................................................................................... 50
REFERENCES…. ............................................................................................................................................ 52

VII
 LIST OF TABLES

Table 1: Gross properties of reference turbine [11] .................................................................................... 22

Table 2: Aerodynamic properties of reference turbine ............................................................................... 23
Table 3 : Mesh Statistics ................................................................................................................................. 27
Table 4: Number of Hexahedra for different mesh regions ....................................................................... 28
Table 5: Performance comparison for different mesh types ...................................................................... 28

VIII
 LIST OF FIGURES

Figure 1: The energy extracting stream-tube of a wind turbine [3] ............................................................. 5

Figure 2: Static pressure and kinetic energy of air between far upstream and far downstream of the
wind turbine ......................................................................................................................................................... 5
Figure 3: Geometry for rotor analysis ............................................................................................................. 6
Figure 4: Velocity triangle at a blade section ................................................................................................. 8
Figure 5: This diagram highlights the components required to define the induced angle of attack .... 11
Figure 6: Velocity triangle at blade section .................................................................................................. 11
Figure 7: Schematics of blade elements ...................................................................................................... 15
Figure 8: Velocity triangle of a blade section ............................................................................................... 15
Figure 9: Regions of actuator disc for CFD-BEMT+LLT model ................................................................ 16
Figure 10: structure of ANSYS-CFX ............................................................................................................. 21
Figure 11: Isometric view of flow field setup with reference coordinate frame ....................................... 24
Figure 12: Front, side and top views of flow field setup showing the meshes and the boundaries ..... 25
Figure 13: Front view of actuator disc mesh. Right diagram shows full mesh. Left diagram shows
zoomed in tip region ........................................................................................................................................ 27
Figure 14 : Performance comparison for different mesh types ................................................................. 29
Figure 15: Flow chart for implementation of momentum source terms .................................................. 30
Figure 16: Variation of power coefficient with TSR ..................................................................................... 33
Figure 17: Variation of thrust coefficient with TSR ...................................................................................... 34
Figure 18: Variation of axial induction factor with dimensionless radial distance at TSR = 8 .............. 35
Figure 19: Variation of tangential induction factor with dimensionless radial distance at TSR = 8 ..... 36
Figure 20: Variation of angle of attack with dimensionless radial distance at TSR = 8 ......................... 37
Figure 21: Variation of the drag coefficient with dimensionless radial distance at TSR = 8 ................. 39
Figure 22: Variation of average axial velocity with axial distance for all CFD models at TSR 4 and
TSR 8 ................................................................................................................................................................. 40
Figure 23: Axial velocity field shown on xy-plane for all CFD models at TSR 4 and TSR 8 ................ 41
Figure 24: Variation of axial velocity along y coordinate at z = 0 at 200m downstream of turbine for
TSR = 4 and TSR= 8 ....................................................................................................................................... 43
Figure 25: Variation of axial velocity along y coordinate at z=0 at 600m downstream of turbine for
TSR = 4 and TSR = 8 ...................................................................................................................................... 44
Figure 26: Variation of tangential velocity along y coordinate at z=0 at 200m downstream of turbine
for TSR = 4 ....................................................................................................................................................... 45

IX
Figure 27: Variation of tangential velocity along y coordinate at z=0 at 200m downstream of turbine
for TSR = 8 ....................................................................................................................................................... 46
Figure 28: Turbulence kinetic energy plotted on xy-plane . Plots on right show different CFD models
at TSR = 8. Plots on left show CFD-BEMT+LLT M3 method at different TSRs ..................................... 47
Figure 29: Variation of power of downstream wind turbine with TSR of upstream wind turbine.......... 48
Figure 30: Variation of thrust of downstream wind turbine with TSR of upstream wind turbine .......... 49

X
 NOMENCLATURE

𝑈 Incoming free stream air velocity

𝑈𝑟𝑒𝑙 Relative velocity of air
𝑣𝑎 Axial induced velocity
𝑣𝑡 Tangential induced velocity
𝑎 Axial induction factor
𝑏 Tangential induction factor
𝜆
𝑎 Tip speed ratio
𝜆𝑟 Local speed ratio
𝑎
𝐴
𝑎 Axial force (Thrust force)
𝑎
𝑇 Tangential force
𝑄
𝑎 Torque
𝑃 Power
𝐶𝑇 Coefficient of Thrust
𝐶𝑃 Coefficient of Performance
𝐶𝐿 Coefficient of lift
𝐶𝐷 Coefficient of drag
𝜔 Angular velocity of flow stream
𝛺 Angular velocity of wind turbine rotor
𝜌 Density of air
𝑐 Chord length
𝜓 Pitch
𝐵 Number of blades
𝜙 Undisturbed flow inclination angle
𝜑 Induced flow inclination angle
𝜀 Drag to lift ratio
Г Circulation
𝑟ℎ Hub radius
𝑅 Turbine tip radius
𝐹 Prandtl tip loss correction factor
𝛥𝑥 Thickness of actuator disc
𝑆𝑇𝑎𝑥 Momentum source term axial
𝑆𝑇𝑡𝑎𝑛 Momentum source term tangential
𝐴𝑆𝑇𝑎𝑥 Additional momentum source term axial
𝐴𝑆𝑇𝑡𝑎𝑛 Additional momentum source term tangential
𝐴𝑆 Swept area of rotor
𝑅𝑒 Reynolds number

XI
 CHAPTER 1: INTRODUCTION
1.1 Background

Wind has powered the ships since as early as 5000 B.C [16]. Windmills have been in use for more than
3000 years. Wind mills were mainly used for grinding grains and pumping water [6]. In the start of twentieth
century, the use of electricity started growing and the windmills were slowly transformed into the wind
turbines to generate electricity by connecting the rotor to the generator [10]. Wind power found new
applications in powering buildings at remote sites [9]. Till 1960’s, the wind turbines were of little importance
for the production of electricity due to cheap availability of fossil fuels which were used to generate electricity
from combustion engines [6].

After the oil crisis of 1973, the wind turbines got the attention of many countries which wanted to reduce
their dependence on the imported fossil fuels and numerous research programs were launched across
many countries in the field of wind power in search for harnessing wind power reliably and efficiently. Since
the oil crisis, commercial wind turbines have steadily emerged as a significant business with an annual
turnover in the 1990s of more than a billion US dollars per year. Since then this figure has increased by
approximately 20% a year [6]. Today, the wind turbines are being produced in a variety of sizes. Their size
and operation range from small standalone turbines for battery charging at isolated places to large multi
megawatt machines as part of near-gigawatt-size wind farms that provide electricity to national electric
transmission systems [9].

The wind turbines generate downstream wakes. When the wind turbines are put in a wind farm with limited
area, power losses occur due to wakes. These losses vary with the arrangements of the wind turbines and
the conditions of incoming wind [17]. Field experiments and computational fluid dynamics (CFD) have
unanimously shown that the power produced by a wind farm is smaller than the combined rated powers of
its constituent turbines. This is because a wind turbine extracts energy from freestream atmospheric flow.
This generates a wake at the downstream side of the turbine with reduced velocity. This wake interacts
with a downstream wind turbine and limits its power generation capacity [13]. Therefore, it is an absolute
necessity to optimize the layout design of the wind farms that can minimize the wake losses and improve
the energy production. The optimization of wind farm layout in turn depends on accurate modelling of the
wind turbine performance as well as the flow field around and downstream of a wind turbine.

1.2 Motivation and Purpose

CFD simulations have the potential to simulate and predict the aerodynamic characteristics of a wind as
well as the flow field characteristics around the turbine and in the wake region. CFD is necessary to visualize
the details of the flow. However, simulating fully resolved 3D horizontal axis wind turbine (HAWT) geometry
requires high computational power and time and hence is very expensive. An alternate approach is to use
a simplified and mixed CFD-BEMT approach. In this approach, the Blade Element Momentum Theory
(BEMT) is coupled to the Reynolds-Averaged Navier Stokes equations (CFD-RANS) using an actuator disc

1
concept. This simplified approach requires less computational power and time and hence is much more
efficient [3], [8].

The BEMT assumes infinite number of blades and hence it further assumes that the forces acting on the
rotor are distributed over the complete rotor disc surface. This means that all fluid particles passing through
the rotor disk at a particular radial position would experience the same loss in axial momentum [9]. In
reality, a wind turbine has a finite number of blades. As a result, flow slippage occurs at the tip of the blades
from the pressure side to suction side of the blade due to pressure difference. This flow slippage is
responsible for generating vortices at the tip region [15]. The flow slippage reduces the forces as well as
the power production capacity throughout the length of the blade and particularly at the tip region. This
effect is known as the tip loss effect [12].

BEMT-CFD model does not capture the effect of tip vortices and therefore it has two major shortcomings:
it over predicts the performance and it generates unrealistic flow field. In Simple BEMT, the tip loss effect
can be introduced via some empirical tip loss factor. An example of such a tip loss correction factor is the
Prandtl tip loss factor. The implementation of the Prandtl tip loss factor reduces the forces acting throughout
the length of the turbine blade. The reduction in the force is higher near the tip where the tip vortices have
a pronounced effect [12]. When the Prandtl tip loss factor is applied to the BEMT-CFD, it not only reduces
the forces exerted by the fluid on the blade, but it also reduces the reactive forces exerted by the blade on
the fluid. As a result, the fluid passes through the turbine with little momentum change and the induced
velocities predicted near the tip region are very low. Hence, the effect of the tip vortices is still not captured
in the CFD-BEMT with the Prandtl tip loss model and the flow field prediction is still unrealistic. Further, due
to unrealistic flow field prediction, the performance prediction is also not correct. This is not a problem with
the BEMT with Prandtl tip loss model because the flow field is not simulated and incorporation of the tip
loss model improves the performance prediction significantly as compared to the simple BEMT [7].

For the simulation of wind farms, an accurate prediction of downstream flow field of a wind turbine is as
necessary as the accurate performance prediction of the turbine. This is because the accurate performance
prediction of a downstream wind turbine is very much dependent on the downstream flow field generated
by an upstream wind turbine. Hence the CFD-BEMT models, with and without tip loss correction factor, are
not suitable for the simulation of wind farms and there is a need to develop more sophisticated CFD models
which are capable of accurately predicting both the performance as well as the downstream flow field of a
wind turbine.

Based on the motivations mentioned, the purpose of this thesis is to develop a more sophisticated CFD
model, as compared to CFD-BEMT model, which is capable of accurately predicting both the performance
of a turbine as well as the flow field. In this master thesis, three different CFD models are developed by
combining the BEMT with the LLT and coupling them to a simple actuator disc concept in CFD. The models
differ in the way of combining the LLT with the existing BEMT-CFD model. These models are termed as

2
the enhanced actuator disc models in this master thesis. Two research questions have to be answered by
this thesis, which are:

1. Are the developed enhanced actuator disc models capable of accurately predicting the
performance of a wind turbine as well as the flow field characteristics?

2. What is the effect of an upstream wind turbine on the performance of a downstream wind turbine
using different CFD models?

1.3 Scope of thesis

At Stuttgart Wind Energy (SWE), CFD-BEMT model using actuator disc concept, with and without the
Prandtl tip loss factor, was already incorporated in the ANSYS-CFX code before the start of this master
thesis. The BEMT was coupled to the ANSYS-CFX using a FORTRAN code. The scope of this master
thesis is specified as follows:

1. Incorporate the LLT to the already existing CFD-BEMT model of SWE, using three different
approaches as explained later in chapter 3, to develop enhanced actuator disc models for CFD.
Like the BEMT, the LLT is applied and combined with the BEMT using a FORTRAN code. The
FORTRAN code is called by ANSYS-CFX as a user subroutine.

2. Perform CFD simulations for a single turbine using the three newly developed enhanced actuator
disc models as well as using the older CFD-BEMT actuator disc models, with and without the Prandt
tip loss factor. Compare the results with each other as well as with the reference results obtained
from the FAST V8 code and the LLT IST code. Specifically:

a. Variation of performance, power coefficient and thrust coefficient, with tip speed ratio.

b. Variation of local flow field parameters, angle of attack, coefficient of drag, axial induction factor
and tangential induction factor, with the radial position along the length of blade.

c. Axial velocity field upstream and downstream of the turbine.

d. Tangential velocity field downstream of the turbine.

e. Turbulence kinetic energy in the flow field.

3. Perform CFD simulations for two turbines in a row using the three newly developed enhanced
actuator disc models as well as the older CFD-BEMT actuator disc models, with and without the
Prandtl tip loss factor. For two turbines in a row case, the results will only be compared to each
other since no reference results are available either experimentally or from any commercial code.
The following results will be compared with each other:

a. Variation of the power of downstream wind turbine with tip speed ratio of upstream wind turbine.
Here only the tip speed ratio of upstream turbine is changed while the tip speed ratio of
downstream turbine is always kept the same for all simulations so that the only effect on the
power of downstream wind turbine is due to the tip speed ratio of upstream wind turbine.

3
b. Variation of the thrust of downstream wind turbine with tip speed ratio of upstream wind turbine.
Here only the tip speed ratio of upstream wind turbine is changed while the tip speed ratio of
downstream wind turbine is always kept the same for all simulations so that the only effect on
the thrust of downstream wind turbine is due to the tip speed ratio of upstream wind turbine.

4
 CHAPTER 2: MODELS FOR WIND TURBINE ANALYSIS
2.1 Basics

A wind turbine is a device that extracts some of the kinetic energy from the wind. This extracted energy
may be used to generate electricity. As a result of this extraction of kinetic energy from the wind, the wind
slows down gradually from far upstream to far downstream. In steady axisymmetric flow through the turbine
rotor disc, the portion of the air passing through the turbine constitutes a stream tube of circular cross
section. No flow occurs across the boundary of stream tube. As the air slows down and it does not
compress, the diameter of the stream tube increases by mass conservation. Since no energy is extracted
until the air reaches the turbine, according to Bernoulli’s Equation, the static pressure of the air must rise
[3].

Figure 1: The energy extracting stream-tube of a wind

turbine [3]
As the air passes through the turbine rotor, its pressure decreases suddenly in a step to below atmospheric
level. The air then moves at the downstream side of the turbine with reduced velocity and reduced pressure

Figure 2: Static pressure and kinetic energy of air between far

upstream and far downstream of the wind turbine

until at far downstream, where the pressure again increases to atmospheric pressure to attain equilibrium.
Since no energy is added or extracted downstream of the rotor disc, the increase in pressure occurs on the
expense of kinetic energy [3].

5
2.2 Blade Element Momentum Theory (BEMT)

The blade element momentum theory is a combination of the momentum theory and the blade element
(BE) theory [12].

2.2.1 Momentum Theory

The aerodynamic behavior of wind turbines can be analyzed without considering any specific turbine design
and just by considering the energy extraction process. The simplest device to accomplish this task is called
actuator disc in which the turbine rotor with finite number of blades is replaced by a circular disc through
which the air can flow [9].

Figure 3: Geometry for rotor analysis

Momentum theory is one of the approaches used to determine the performance of the actuator disc. This
analysis is based on the conservation of linear and angular momentum, conservation of mass and
conservation of energy. The fluid flow exerts an axial force (thrust) as well as a torque on the turbine rotor.
The analysis assumes a control volume in which the boundaries of the control volume are the surfaces of
annular stream tube and two cross sections at the start and end of stream tube. Flow can only pass through
the end cross sections and no flow is allowed to cross the surfaces of the stream tube into adjacent stream
tubes [12].

The detailed analysis can be found in [12]. Here only the important and relevant formulas are given together
with brief and relevant description.

The wind velocity of the undisturbed air is represented by 𝑈. 𝑈1 , 𝑈2 , 𝑈3 , and 𝑈4 , represent the velocities of
air at sections 1,2,3 and 4 in Figure 3.

The wind velocity at the rotor plane, using this simple model, is the average of the upstream and
downstream wind speeds.
𝑈1 + 𝑈4
𝑈2 = (1)
2
The axial induction factor is, 𝑎, is defined as

6
 𝑈 − 𝑈2
𝑎= (2)
𝑈
The quantity 𝑈𝑎 often referred to as the induced velocity at the rotor. The velocity of the wind at the rotor is
a combination of the free stream velocity and the induced wind velocity.

An angular induction factor, 𝑏, is defined as:

𝜔
𝑏= (3)
2𝛺
Where 𝜔 is the angular velocity of the flow stream imparted by the wind turbine and 𝛺 is the angular velocity
of the wind turbine.

The induced velocity at the rotor consists of not only the axial component, 𝑈𝑎, but also a component in the
rotor plane, 𝑟𝛺𝑏.

The expression for the axial force, 𝐴, exerted by the fluid on a differential element is given by:

𝑑𝐴 = 4𝑎(1 − 𝑎)𝜌𝑈 2 𝜋𝑟𝑑𝑟 (4)

where 𝜌 is the density of the air.

The tip speed ratio, 𝜆, defined as the ratio of the blade tip speed to the free stream wind speed, is given
by:
𝛺𝑅
𝜆= (5)
𝑈
where 𝑅 is the tip radius of wind turbine.

The local speed ratio is the ratio of the rotor speed at some intermediate radius, 𝑟, to the wind speed:

𝛺𝑟
𝜆𝑟 = (6)
𝑈
The torque, 𝑄, on a differential element is given by:

𝑑𝑄 = 4𝑏(1 − 𝑎)𝜌𝑈𝜋𝑟 3 𝛺𝑑𝑟 (7)

2.2.2 Blade Element Theory

In the BE theory, the forces on the blades of a wind turbine can also be expressed as a function of the lift
and drag coefficients and the angle of attack. For this analysis, the blade is assumed to be divided into a
number of radial elements [12].

According to the theory of aerodynamics, lift and drag forces are perpendicular and parallel respectively to
relative wind. The relative wind is the vector sum of the wind velocity at the rotor, 𝑈(1 − 𝑎), and the wind
velocity due to rotation of the blade, 𝛺𝑟(1 + 𝑏) [12].

7
Figure 4 shows the relationships between the various angles, velocities and forces at a blade section. Here,
𝜓 is the section pitch angle, which is the angle between the chord line and the plane of rotation; 𝛼 is the
angle of attack, which is the angle between the chord line and relative velocity of wind; 𝜙 is the undisturbed
flow inclination angle; 𝜑 is the induced flow inclination angle; 𝑑𝐿 is the incremental lift force; 𝑑𝐷 is the
incremental drag force; 𝑑𝐴 is the incremental axial force (normal to the plane of rotation); and 𝑑𝑇 is the
incremental tangential force (parallel to the plane of rotation). This is the force is responsible for creating
useful torque. Finally, 𝑈𝒓𝒆𝒍 is the relative wind velocity [12].

Figure 4: Velocity triangle at a blade section

Here again, only the relevant and most important relations are given for the BE theory. The details can be
found in [12].

The induced flow inclination angle is given by:

𝑈(1 − 𝑎)
tan 𝜑 = (8)
𝛺𝑟(1 + 𝑏)

The total axial force on the section at a distance, 𝑟, from the center is:

1 2 (𝐶
𝑑𝐴 = 𝐵𝜌𝑈𝑟𝑒𝑙 𝐿 cos 𝜑 + 𝐶𝐷 sin 𝜑)𝑐𝑑𝑟 (9)
2
where 𝐵 is the number of blades, 𝐶𝑳 is the lift coefficient, 𝐶𝐷 is the drag coefficient and 𝑈𝑟𝑒𝑙 is the relative
velocity of wind.

The differential torque due to the tangential force acting at a distance, r, from the center is given by:

1 2 (𝐶
𝑑𝑄 = 𝐵𝜌𝑈𝑟𝑒𝑙 𝐿 sin 𝜑 − 𝐶𝐷 cos 𝜑)𝑐𝑟𝑑𝑟 (10)
2
2.2.3 Blade Element Momentum Theory

BEMT is the combination of BE and momentum theories. In BEMT, axial force and torque from BE are
forced to match axial force and torque from momentum theory [12].

In the calculation of induction factors, a and b, accepted practice is to set 𝐶𝐷 equal to zero (see Wilson and
Lissaman, 1974). So, when the torque equations from momentum and blade element theory are equated
(Equations (7) and (10)), with 𝐶𝐷 = 0, one gets

8
 𝑏 𝜎 ′ 𝐶𝐿
= (11)
(1 − 𝑎) (4𝜆𝑟 sin 𝜑)

Where 𝜎′ is the local solidity, defined by:

𝐵𝑐
𝜎′ = (12)
2𝜋𝑟
By equating the axial force equations from momentum and blade element theory (Equations (4) and (9)),
one obtains:

𝑎 𝜎 ′ 𝐶𝐿 cos 𝜑
= (13)
(1 − 𝑎) (4 sin2 𝜑)

After some algebraic manipulations, these equations are solved iteratively to determine the flow conditions
and forces at each blade section.

The power contribution from each annulus is

𝑑𝑃 = 𝛺𝑑𝑄 (14)

The total power from the rotor is:

𝑅
𝑃 = ∫ 𝛺𝑑𝑄 (15)
𝑟ℎ

Where 𝑟ℎ is the hub radius.

2.2.4 Tip Loss: Effect on Power Coefficient of Number of Blades

Because the pressure on the upper side of a blade is lower than that on the lower side, air tends to flow
around the tip from the lower to upper surface, reducing the lift and hence the power production near the
tip. [12]. A number of methods have been suggested for including the effect of the tip loss. The most
straightforward approach to use is one developed by Prandtl (see de Vries, 1979). According to this method,
a correction factor, 𝐹, must be introduced into the previously discussed equations

2 (𝐵/2)[1 − (𝑟/𝑅)]
𝐹 = ( ) cos −1 [𝑒𝑥𝑝 (− { })] (16)
𝜋 (𝑟/𝑅) sin 𝜑

Note that 𝐹 is always between 0 and 1. This tip loss correction factor characterizes the reduction in the
forces at a radius 𝑟 along the blade that is due to the tip loss. Thus Equations (4) and (7) become:

𝑑𝐴 = 𝐹4𝑎(1 − 𝑎)𝜌𝑈 2 𝜋𝑟𝑑𝑟 (17)

And

𝑑𝑄 = 4𝐹𝑏(1 − 𝑎)𝜌𝑈𝜋𝑟 3 𝛺𝑑𝑟 (18)

9
2.3 Lifting Line Theory (LLT)

In a wind turbine, vortices are generated at the tip region due to pressure difference between upper and
lower surface of the blade. Tip vortices reduce the angle of attack relative to an idealized infinitely long 2D
airfoil which in turn causes reduction in lift force.
The lift force on a wing is produced due to pressure differential between the lower and upper surfaces of
the wing. The associated difference in flow speed above and below the wing section can be characterized
as circulation (or vorticity). The induced circulation makes the flow to deflect downwards. This downward
deflection is called as downwash [7]. According to circulation theory, see [1], the lift force of a wing of
constant section and unit span is calculated as:

𝐿 = 𝜌𝑈Г (19)

where Г is the circulation.

Figure 5 shows that the geometric angle of attack 𝛼 is composed of two further components; induced angle
of attack 𝛼𝑖 and the effective angle of attack 𝛼𝑒 . The induced angle of attack is the angle of downward
deflection resulting from induced downwash [11]. The induced angle of attack can be computed by:

𝛼𝑖 = tan−1 𝑤/𝑈 (20)

The effective angle of attack is given by:

𝛼𝑒 = 𝛼 − 𝛼𝑖 (21)

Wing tip vortices are explained in [15] and [1]. These wing tip vortices downstream of the wing induce an
additional small downward component of air velocity in the neighborhood of the wing itself. This additional
downward component, additional downwash (or additional induced velocity), varies across the length of a
finite wing. The span wise component of air vector on the upper surface is directed from tip to hub, and on
the lower surface the span wise component of velocity is directed from hub to tip. The physical effect of the
resultant shear at the trailing edge is the development of a trailing vortex sheet which contains vorticity [7].
Prandtl’s classical lifting line theory, as detailed in [15], describes this phenomenon in terms of an infinite
number of horseshoe vortices distributed across the wing span, compounded towards the root of the span.

Prandtl’s classical lifting line theory is an approach that uses this system of horse shoe vortices (bound
wing vortices and trailing wing vortices) to determine the induced velocities and forces on the wing.
According to the LLT, for a wing of finite length, αe at the wing tip is constrained as the angle of attack at
which lift force reduces to zero. This then influences the distribution of downwash across the span [7].
Hence when LLT is applied to wind turbines, it is capable of predicting the induced velocities generated by
the horse shoe vortex system and further calculate the forces on the blades of the turbine.

10
 U

Figure 5: This diagram highlights the components required to

define the induced angle of attack

The following Lifting line theory for the designing of turbine till the end of this section is taken from [4]. It is
assumed that the turbine, having 𝐵 number of blades, is subjected to a uniform axial airstream of velocity
𝑈. Flow is steady in the reference frame of turbine. Each turbine blade is assumed to be replaced by a
radial line vortex having circulation strength 𝛤(𝑟) which varies along the radial line from the hub radius 𝑟 =
𝑟ℎ to the blade tip 𝑟 = 𝑅.

The effect of turbine hub is incorporated by considering the potential flow disturbance by an approximate
image method assuming an infinitely long cylinder of radius 𝑟 = 𝑟ℎ . In this case, the lifting line circulation
may be different from zero at the hub radius. At the tip, the circulation reduces to zero Г(𝑅) = 0. A trailing
vortex sheet is emitted from each lifting line. It is assumed that the trailing vortices are helical in nature with
constant radius and pitch in the stream wise direction.

Figure 6: Velocity triangle at blade section

The velocity triangle for a blade section at radius 𝑟 is shown in Figure 6. The axial and tangential induced
velocities at the lifting line may be written as:

11
 𝑅
1 𝑑Г 𝑖𝑎,𝑡
νa,t (r) = ∫ 𝑑𝑟 (22)
4𝜋 𝑟ℎ 𝑑𝑟′ 𝑟 − 𝑟′

where

𝑟 𝑟′ 𝑟 𝑟
ia,t = ( , , 𝜑𝜈 , 𝐵) = 𝑖𝑎,𝑡 ( ′ , 𝜑𝜈 , 𝐵) − 𝑖𝑎,𝑡 ( ′(𝑖𝑚) , 𝜑𝜈𝑖𝑚 , 𝐵) (23)
𝑟′ 𝑟ℎ 𝑟 𝑟
are axial and tangential induction factors due to a set of B semi-infinite helical vortices of radius r’ and
pitch angle φv , which are modified to take into account the velocity induced by approximate image
vortices generated due to presence of infinitely long cylinder of radius 𝑟ℎ . The approximate image
vortices are semi-infinite helical vortices and with radius 𝑟 ′(𝑖𝑚) = 𝑟ℎ2 /𝑟 ′ and the pitch angle satisfying
tanϕv(im) =(r‘/r‘(im))tanϕv. ia,t (r/r‘,ϕv,B) are the usual axial and tangential induction factors for helicoidal
vortices in the lifting line theory [13]. For their evaluation asymptotic formulas of Morgan and Wrench
(1965) can be used. In the moderately loaded theory, we can determined the induced flow inclination
angle as:

𝑈 − 𝜈𝑎
𝜑𝜈 (r) = 𝜑(r) = tan−1 (24)
𝛺𝑟 + 𝜈𝑡

Kutta-Joukowski law is used to calculate the inviscid forces on the lifting line. The viscous effects on the
axial force and torque can be included by introducing the drag to lift ratio 𝜀 = 𝑑𝐷/𝑑𝐿 = 𝐶𝐷 /𝐶𝐿 of the blade
section forces, see Figure 6. The axial force and the torque on the rotor are obtained by integration along
the radius and summing on the number of blades:
𝑅
A = ρB ∫ (𝛺𝑟 + 𝜈𝑡 )(1 + ε tan 𝜑)Г(𝑟)𝑑𝑟 (25)
𝑟ℎ

𝑅
Q = ρB ∫ (𝑈 − 𝜈𝑎 )(1 − ε cot 𝜑)Г(𝑟)𝑟𝑑𝑟 (26)
𝑟ℎ

Where 𝜌 is the fluid density

Blade section lift and drag coefficients are defined by

𝑑𝐿/𝑑𝑟 𝑑𝐷/𝑑𝑟
C𝐿 = 2 , 𝐶𝐷 = 2 (27)
1/2𝜌𝑈𝑟𝑒𝑙 𝑐 1/2𝜌𝑈𝑟𝑒𝑙 𝑐

where 𝑈𝒓𝒆𝒍 is the relative velocity of wind and 𝑐 is the chord of the blade section. Using Kutta-Joukowski
law the coefficient of lift can be calculated as
2Г
C𝐿 = (28)
𝑈𝑟𝑒𝑙 𝑐

The section pitch angle is related to induced flow inclination angle by

ψ=𝜑− 𝛼 (29)

12
Where 𝛼 is the section angle of attack at the corresponding lift coefficient (Figure 6).The lift and drag
coefficients of the blade section are assumed to be a function of the angle of attack and of the section
Reynolds number, 𝑅𝑒.

Dimensionless quantities are introduced by using the turbine tip radius, 𝑅, as reference length and the
incoming free stream air velocity, 𝑈, as reference velocity. The turbine thrust and power coefficients are
defined by

𝐴
CT = (30)
1 2 2
𝜌𝑈 𝜋𝑅
2
𝑃 𝛺𝑄
CP = = (31)
1 3 2 1 3 2
𝜌𝑈 𝜋𝑅 𝜌𝑈 𝜋𝑅
2 2
Which are functions of the turbine tip-speed-ratio 𝜆 = (Ω𝑅)/𝑈 and the turbine Reynolds number.

13
 CHAPTER 3: CFD-MODELS
In this chapter, the description of CFD models used in this thesis is provided. First, the governing equations
of CFD are discussed. Then the already exiting CFD-BEMT model and newly developed enhanced actuator
disc models are discussed.

3.1 The RANS equations

Simulations were performed using ANSYS-CFX with the Reynolds-Averaged Navier-Stokes (RANS)
equations, which are obtained by averaging the fluctuating quantities in the original transport equations. In
the following steady-state RANS equations, the bar is dropped for averaged quantities, except for products
of fluctuating quantities.

Continuity equation is given as:

𝜕
𝜌(𝑈𝑗 ) = 0 (32)
𝜕𝑥𝑗

Momentum equation is given as:

𝜕 𝜕𝑝 𝜕
𝜌(𝑈𝑖 𝑈𝑗 ) = − + ̅̅̅̅̅̅
(𝜏 − 𝜌𝑈 𝑖 𝑈𝑗 ) + 𝑆𝑖 (33)
𝜕𝑥𝑗 𝜕𝑥𝑖 𝜕𝑥𝑗 𝑖𝑗
where 𝜌 is the density, 𝑈𝑖 (𝑖 = u, v, w) is the averaged velocity, xi (i = x, y, z) is the position, 𝑝 is the mean
̅̅̅̅̅̅
pressure, 𝜏 is the molecular stress tensor, 𝜌𝑈 𝑖 𝑈𝑗 are the Reynolds stresses, and 𝑆𝑖 are additional source

terms in the i = x, y, z momentum equations. The Reynolds stresses are calculated from the k-ω SST
model.

3.2 CFD-BEMT model

This method is based on the solution of RANS equations in combinations in combination with the BEMT.
When an airfoil moves through air, aerodynamic force is exerted on the airfoil. This force can be resolved
into two mutually perpendicular components, the lift force, 𝐿, and the drag force, 𝐷, defined as follows.

2
𝐿 = 0.5𝜌𝑈𝑟𝑒𝑙 𝐴𝑟 𝐶𝐿 (34)

2
𝐷 = 0.5𝜌𝑈𝑟𝑒𝑙 𝐴𝑟 𝐶𝐷 (35)

where 𝜌 is the density of air, 𝑈𝑟𝑒𝑙 is the relative velocity of air, 𝐴𝑟 is the area of the object, 𝐶𝐿 is the lift
coefficient and 𝐶𝐷 is the drag coefficient. The lift and drag coefficients are dependent on the angle of attack,
𝛼, the Reynolds number, 𝑅𝑒, and the geometric properties of airfoil.

In the BEMT, the effect of multiple blades is averaged over one complete rotation. In the CFD-BEMT model,
the effect of blade on the fluid is introduced through momentum source terms in Navier-Stokes equations
(see Equation (33)).

14
BEMT discretize the blade in a number of elements as shown in Figure 7. The forces on each blade section
are as shown again in the Figure 8.

Figure 7: Schematics of blade elements

Figure 8: Velocity triangle of a blade section

The induced flow inclination angle is defined as:

𝑈(1 − 𝑎)
𝜑 = tan−1 (36)
𝛺𝑟(1 + 𝑏)
Relative wind velocity is given by:
2
𝑈𝑟𝑒𝑙 = [𝑈(1 − 𝑎)]2 + [𝛺𝑟(1 + 𝑏)]2 (37)

Where 𝑈 is undisturbed wind velocity, 𝛺 is angular velocity of turbine, 𝑟 is the radius of blade element, 𝑎 is
axial induction factor and 𝑏 is tangential induction factor. Here it is important to mention that in CFD-BEMT
model, 𝑎 and 𝑏 are calculated directly from the RANS calculations.

The lift force and drag force on a blade element are given by:

2
𝑑𝐿 = 0.5𝜌𝑈𝑟𝑒𝑙 𝐶𝐿 𝑐𝑑𝑟 (38)

2
𝑑𝐷 = 0.5𝜌𝑈𝑟𝑒𝑙 𝐶𝐷 𝑐𝑑𝑟 (39)

where 𝑐 is the chord length.

Using Figure 8, lift force and drag force are related to axial force and tangential force as follows

𝑑𝐴 = 𝑑𝐿 cos 𝜑 + 𝑑𝐷 sin 𝜑 (40)

𝑑𝑇 = 𝑑𝐿 sin 𝜑 − 𝑑𝐷 cos 𝜑 (41)

Substituting Equations (38) and (39) into Equations (40) and (41) gives:

15
 2
𝑑𝐴 = 0.5𝜌𝑈𝑟𝑒𝑙 𝑐𝑑𝑟(𝐶𝐿 cos 𝜑 + 𝐶𝐷 sin 𝜑) (42)

2
𝑑𝑇 = 0.5𝜌𝑈𝑟𝑒𝑙 𝑐𝑑𝑟(𝐶𝐿 sin 𝜑 − 𝐶𝐷 cos 𝜑) (43)

Which when converted to force per unit will give the axial and tangential momentum source terms, 𝑆𝑇𝑎𝑥
and 𝑆𝑇𝑡𝑎𝑛 to be incorporated in Navier-Stokes Equations.
𝐵𝑑𝐴
𝑆𝑇𝑎𝑥 = (44)
2𝜋𝑟𝛥𝑥𝑑𝑟
𝐵𝑑𝑇
𝑆𝑇𝑡𝑎𝑛 = (45)
2𝜋𝑟𝛥𝑥𝑑𝑟
where 𝐵 is the number of blades and Δx is the thickness of the actuator disc
When the Prandtl Tip loss correction is used, Equations (44) and (45) becomes
𝐵𝐹𝑑𝐴
𝑆𝑇𝑎𝑥 = (46)
2𝜋𝑟𝛥𝑥𝑑𝑟
𝐵𝐹𝑑𝑇
𝑆𝑇𝑡𝑎𝑛 = (47)
2𝜋𝑟𝛥𝑥𝑑𝑟
, Where 𝐹 is the Prandtl tip loss factor given as
2 (𝐵/2)[1 − (𝑟/𝑅)]
𝐹 = ( ) cos −1 [𝑒𝑥𝑝 (− { })] (48)
𝜋 (𝑟/𝑅) sin 𝜑

3.3 Enhanced actuator disc models (CFD-BEMT+LLT Models)

To address the issue of simultaneously improving performance prediction and flow field, in this model the
BEMT and the LLT are combined. The LLT has the capability of taking tip vortices into account and
accurately predicting the local flow field. Since the tip vortices has strong impact only near the tip region,
the LLT is coupled with the BEMT in the last 10% of the blade length at the tip side. From now on, this
region will be referred as the BEMT+LLT region. In the remaining 90% of the blade length starting from
center of turbine, simple BEMT is used. From now on, this region will be referred as BEMT region. This is
shown in Figure 9.

KEY:
BEMT region
First 90% of blade length
Actuator starting from centre
Disc
BEMT + LLT region
Last 10 % of blade length
at the tip side

Figure 9: Regions of actuator disc for CFD-BEMT+LLT model

16
3.3.1 BEMT region
In the CFD-BEMT model discussed earlier, the BEMT was incorporated to CFD by calculating and applying
axial and tangential momentum source terms derived using the BEMT, as shown in Equations (36) to (45).
In the CFD-BEMT+LLT models, the same formulations and procedure is used for the CFD-BEMT region
as in that region only the BEMT is coupled with CFD.

3.3.2 BEMT+LLT region

In the BEMT+LLT region, both for axial and tangential directions, two momentum source terms are
calculated. They will onwards be referred as simple momentum source terms and additional momentum
source terms. These two are combined to give cumulative momentum source terms for the BEMT+LLT
region.

3.3.2.1 Simple Momentum Source Terms:

Tip vortices generate additional induced velocities. These additional induced velocities reduce the forces
on the blade particularly in the tip region. The LLT captures the effect of the tip vortices on the induced
velocities. Simple momentum source terms will impart the effect of reactive forces of the turbine blade on
the fluid. In the calculation of simple momentum source terms for the BEMT+LLT region, induced velocities
calculated from the LLT are used in the CFD-BEMT model (discussed earlier) in contrast to axial and
tangential induction factors obtained from the RANS calculations of the current flow field iteration. The result
is that, the forces on the blades and hence the momentum source terms are reduced as compared to the
simple CFD-BEMT model.

To calculate simple momentum source terms, LLT is used to calculate axial and tangential induced
velocities, 𝑣𝑎 and 𝑣𝑡 , in the BEMT+LLT region. These induced velocities are used to find the induced flow
inclination angle as below:
𝑈 − 𝑣𝑎
𝜑 = tan−1 (49)
𝛺𝑟 + 𝑣𝑡
The relative velocity of the wind is calculated as:
2
𝑈𝑟𝑒𝑙 = [𝑈 − 𝑣𝑎 ]2 + [𝛺𝑟 + 𝑣𝑡 ]2 (50)

After finding the flow inclination angle and relative velocity using induced velocities calculated from the LLT,
simple momentum source terms are calculated using the same formulations and procedure as that of CFD-
BEMT model from Equation (38) to (45).

3.3.2.2 Additional momentum source terms.

Additional momentum source terms represent the effect of the tip vortices on the fluid in terms of induced
velocities. The induced velocities calculated by the LLT contain the effect of the tip vortices. The idea used
to calculate the additional momentum source terms is that to calculate the difference in induced velocities
from the LLT and the RANS calculations and transform that difference in velocity into a force. Three different

17
approaches, Method 1, Method 2 and Method 3, are used to transform the difference in induced velocity to
the force.

3.3.2.2.1 Method 1 (M1)

This method is taken from Edmunds et al [7]. According to this method, the additional reactive force, 𝑑𝐹𝑣 ,
at the tip required to produce zero lift condition at the tip is given by

𝑑𝐹𝑣 = 0.5𝜌𝑤𝑎2 𝑐𝑑𝑟 (51)

Where 𝑤𝑎 is the additional downwash (additional induced velocity) required to achieve zero lift condition at
the tip. In Method 1 of calculating additional momentum source terms, additional axial and tangential
induced velocities are calculated which are required to bring the induced velocities from the RANS
calculations to the induced velocities from the LLT. These additional induced velocities are calculated by
taking the difference of induced velocities from the LLT and the RANS calculations. This difference in the
induced velocities is used to calculate the additional reactive forces in the axial and tangential direction,
𝑑𝐹𝐴 and 𝑑𝐹𝑇 , separately as follows:
2
𝑑𝐹𝐴 = 0.5𝜌(𝑣𝑎𝐿𝐿𝑇 − 𝑣𝑎𝑅𝑆 ) 𝑐𝑑𝑟 (52)
2
𝑑𝐹𝑇 = 0.5𝜌(𝑣𝑡𝐿𝐿𝑇 − 𝑣𝑡𝑅𝑆 ) 𝑐𝑑𝑟 (53)

where 𝑣𝑎𝐿𝐿𝑇 and 𝑣𝑡𝐿𝐿𝑇 are the axial and tangential induced velocity from the LLT and 𝑣𝑎𝑅𝑆 and 𝑣𝑡𝑅𝑆 are
the axial and tangential induced velocities from RANS calculations. The additional reactive force when
converted to force per unit volume gives the additional momentum source terms as follows
2
𝐵𝑑𝐹𝐴 𝐵𝜌(𝑣𝑎𝐿𝐿𝑇 − 𝑣𝑎𝑅𝑆 ) 𝑐
𝐴𝑆𝑇𝑎𝑥 = = (54)
2𝜋𝑟𝛥𝑥𝑑𝑟 4𝜋𝑟𝛥𝑥
2
𝐵𝑑𝐹𝑡 𝐵𝜌(𝑣𝑡𝐿𝐿𝑇 − 𝑣𝑡𝑅𝑆 ) 𝑐
𝐴𝑆𝑇𝑎𝑥 = = (55)
2𝜋𝑟𝛥𝑥𝑑𝑟 4𝜋𝑟𝛥𝑥
The additional momentum source terms calculated from this method are considerably smaller than those
calculated from M2 and M3.

3.3.2.2.2 Method 2 (M2)

This method is based on momentum balance and uses Newton’s second Law of motion. Difference in
induced velocities from the LLT and the RANS calculations is obtained. This difference is used to calculate
the rate of change of momentum, required to produce this change of velocity, which is equivalent to force.
Axial and tangential forces, dFA and dFT are calculated as

dFA = (𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )𝑑𝐴𝑟 ) ∗ (𝑣𝑎𝐿𝐿𝑇 − 𝑣𝑎𝑅𝑆 ) (56)

dF 𝑇 = (𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )𝑑𝐴𝑟 ) ∗ (𝑣𝑡𝐿𝐿𝑇 − 𝑣𝑡𝑅𝑆 ) (57)

18
where 𝑑𝐴𝑟 is the cross section area of differential control volume, 𝑣𝑎𝐿𝐿𝑇 and 𝑣𝑡𝐿𝐿𝑇 are the axial and
tangential induced velocity from the LLT and 𝑣𝑎𝑅𝑆 and 𝑣𝑡𝑅𝑆 are the axial and tangential induced velocities
from RANS calculations. Additional momentum source terms are calculated by finding the force per unit of
volume. The procedure is applied for both axial and tangential source terms. The axial and tangential source
terms, 𝐴𝑆𝑇𝑎𝑥 and 𝐴𝑆𝑇𝑡𝑎𝑛 , for this method are given as

𝑑𝐹𝐴 𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )(𝑣𝑎𝐿𝐿𝑇 − 𝑣𝑎𝑅𝑆 )

𝐴𝑆𝑇𝑎𝑥 = = (58)
𝑑𝐴𝑟 𝛥𝑥 𝛥𝑥

𝑑𝐹𝑇 𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )(𝑣𝑡𝐿𝐿𝑇 − 𝑣𝑡𝑅𝑆 )

𝐴𝑆𝑇𝑡𝑎𝑛 = = (59)
𝑑𝐴𝑟 𝛥𝑥 𝛥𝑥

The source terms calculated from this method are higher than M1 and lower than M3.

3.3.2.2.3 Method 3 (M3)

This method is similar to Method 2. The only difference is that the induced velocities from the flow field are
assumed to be zero, which holds strictly at the tip. Hence, difference in induced velocities from LLT and
flow field reduces to absolute value of induced velocities from LLT. This method is not physical but is used
to increase the source terms and observe the effects of increased additional momentum source terms. The
axial and tangential source terms for this method are given as

𝑑𝐹𝐴 𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )(𝑣𝑎𝐿𝐿𝑇 )

𝐴𝑆𝑇𝑎𝑥 = = (60)
𝑑𝐴𝑟 𝛥𝑥 𝛥𝑥

𝑑𝐹𝑇 𝜌(𝑈 − 𝑣𝑎𝑅𝑆 )(𝑣𝑡𝐿𝐿𝑇 )

𝐴𝑆𝑇𝑡𝑎𝑛 = = (61)
𝑑𝐴𝑟 𝛥𝑥 𝛥𝑥

3.4 Calculation of power coefficient (𝑪𝑷 ) and thrust coefficient (𝑪𝑻 )

𝜔𝑄 𝜔𝛿𝑄
𝐶𝑃 = =∑ (62)
1 3 1
𝜌𝑈 𝐴𝑠 𝜌𝑈 𝐴
2 2 3 𝑠
𝐴 𝛿𝐴
𝐶𝑇 = =∑ (63)
1 2 1 2
𝜌𝑈 𝐴𝑠 𝜌𝑈 𝐴𝑠
2 2
Where 𝐴𝑠 is the rotor swept area which is corresponding to the sectional area of the disc, 𝑄 is the torque
obtained from the product of the radius (𝑟) and the tangential force (𝑇) in the BEM-CFD simulation, and 𝐴
is the axial force acting on the turbine which is equivalent to the thrust force. Thus, the equivalent 𝐶𝑃 and
𝐶𝑇 are calculated by summation over each element with width 𝑑𝑟.

19
 CHAPTER 4: COMPUTATIONAL CODES
In this master thesis, three computational codes are used; ANSYS-CFX. FAST V8 and LLT IST code. The
CFD simulations are performed in ANSYS-CFX. Brief details of these codes are given below.

4.1 ANSYS-CFX

4.1.1 Introduction to ANSYS-CFX

ANSYS CFX is a general purpose software suit that uses Computational Fluid Dynamics (CFD) to solve
the fluid problems. ANSYS combines an advanced solver with powerful preprocessing and post processing
capabilities. ANSYS-CFX is based on the control volume fixed in space fluid model. [2].

4.1.2 The Structure of ANSYS-CFX

ANSYS-CFX consists of four software modules:

4.1.2.1 CFX-Pre
CFX-Pre is used to define simulations. Multiple meshes may be imported, allowing each section of complex
geometries to use the most appropriate mesh. Analyses, which consist of flow physics, boundary
conditions, initial values, and solver parameters, are also specified. A full range of boundary conditions,
including inlets, outlets and openings, together with boundary conditions for heat transfer models and
periodicity, are all available in ANSYS-CFX through CFX-Pre [2].

4.1.2.2 CFX-Solver
CFX-Solver solves all the solution variables for the simulation for the problem specification generated in
CFX-Pre. One of the most important features of ANSYS-CFX is its use of a coupled solver, in which all the
hydrodynamic equations are solved as a single system which results in a faster converged solution as
compared to traditional segregated solver [2].

4.1.2.3 CFX-Solver Manager

The CFX-Solver Manager module provides greater control to the management of the CFD task. Its major
functions are to specify the input files to the CFX-Solver, Start/stop the CFX-Solver, monitor the progress
of the solution and set up the CFX-Solver for a parallel calculation [2].

4.1.2.4 CFD-Post

CFD-Post provides interactive post processing graphics tools to examine the ANSYS CFX simulation
results. Important features include quantitative post-processing, report generation, user-defined variables,
generation of a variety of graphical objects where visibility, transparency, color, and line/face rendering can
be controlled and power syntax to allow fully programmable session files [2].

20
 Figure 10: structure of ANSYS-CFX

4.2 FAST V8
The FAST (Fatigue, Aerodynamics, Structures, and Turbulence) is an open source computer aided
engineering (CAE) tool developed by NREL (National Renewable Energy Laboratory). It is a comprehensive
multiphysics engineering software tool used to design and analyze horizontal axis wind turbines. The novel
FAST v8 contains modules for aerodynamics (AeroDyn); hydrodynamics (HydroDyn); control and electrical
drive dynamics (ServoDyn); rotor, drivetrain, nacelle, tower, and platform structural dynamics (ElastoDyn);
multimember substructure structural dynamics (SubDyn); mooring statics and dynamics (MAP); and ice
loads (IceFloe) [5].

AeroDyn is a set of routines used in conjunction with an aeroelastic simulation code to predict the
aerodynamics of horizontal axis wind turbines. AeroDyn has the capability to determine the aerodynamics
characteristics of wind turbines using two types of models; the BEMT and the generalized dynamic-wake
theory. For this master thesis, AeroDyn is used to obtain some reference results using BEMT model, both
with and without the Prandtl tip loss correction [14].

4.3 LLT IST code

LLT IST code is developed in Instituto Superior Técnico, Portugal. This code uses the LLT to determine the
aerodynamic characteristics of a wind turbine. The code is based on the induction factor method of Lerbs
(1952) combined with a vortex-lattice method for the discretization of the lifting line (Falcão de Campos,
2007, Baltazar et al, 2011). The implementation of the induction factors follows the formulation of Morgan
and Wrench (1965). The local flow field parameters, which include axial and tangential induced velocities,
α, and CD, obtained from CFD simulations are compared with those obtained from LLT IST code. Further,
the LLT is applied in the CFD simulations via a FORTRAN subroutine. This subroutine is derived from the
same LLT IST code.

21
 CHAPTER 5: REFERENCE TURBINE FOR CFD SIMULATIONS
For this thesis, turbine specifications of “NREL offshore 5-MW baseline wind turbine” are used. The
specifications of the turbine are developed by NREL of The U.S. Department of Energy (DOE) to support
the concept studies for assessing offshore wind technology. This wind turbine is a conventional three-
bladed upwind variable-speed variable blade-pitch-to-feather-controlled turbine. NREL has given the
specifications for gross properties of turbine, blade structural properties, blade aerodynamic properties, hub
and nacelle properties, drivetrain properties, tower properties and baseline control system properties [11].
Here only gross properties of turbine and aerodynamic properties of turbine will be provided as only these
properties are relevant to this thesis.

5.1 Gross properties of turbine

The gross properties of NREL offshore 5-MW baseline wind turbine are given in Table 1.

Rating 5 MW

Rotor Orientation, Configuration Upwind, 3 Blades

Control Variable Speed, Collective Pitch

Drivetrain High Speed, Multiple-Stage Gearbox

Rotor, Hub Diameter 126 m, 3 m

Hub Height 90 m

Cut-In, Rated, Cut-Out Wind Speed 3 m/s, 11.4 m/s, 25 m/s

Cut-In, Rated Rotor Speed 6.9 rpm, 12.1 rpm

Rated Tip Speed 80 m/s

Overhang, Shaft Tilt, Precone 5 m, 5º, 2.5º

Rotor Mass 110,000 kg

Nacelle Mass 240,000 kg

Tower Mass 347,460 kg

Table 1: Gross properties of reference turbine [11]

5.2 Blade aerodynamic Properties

The aerodynamic properties of the turbine used in the CFD simulations are given in Table 2. The table
shows pitch, chord and airfoil distribution along the blade. Eight airfoils are used for this turbine. The two
innermost airfoils represent cylinders with drag coefficients of 0.50 and 0.35 and zero lift. The effect of hub
is neglected by using both lift and drag coefficients equal to zero. The airfoil data (𝐶𝐿 and 𝐶𝐷 against 𝛼) for
all the airfoils used is given in [11]. Here it is important to note that in this thesis the original airfoil distribution
provided by NREL is not used due to a mistake that was realized after the majority of the simulations were

22
conducted. The airfoils used in this thesis are same as provided by NREL but their distribution along the
radial position is different as compared to the distribution provided by NREL. However, this does not affect
the purpose of this thesis since the thesis is based on the comparison of results and all the analysis (CFD
simulations as well as reference result from FAST V8 and LLT IST code) are done with the same airfoil
distributions.

Radius Pitch (˚) chord(m) Airfoil

0 13.308 3.542 Hub
1.5 13.308 3.542 Cylinder 1
4.2334 13.308 3.698006 Cylinder 1
6.9666 13.308 4.010494 Cylinder 2
9.7 13.308 4.323002 DU21_A17
11.75 13.308 4.557 DU21_A17
13.8 12.394 4.6045 DU25_A17
15.85 11.48 4.652 DU25_A17
17.9 10.821 4.555 DU25_A17
22 9.5865 4.3535 DU30_A17
26.1 8.403 4.128 DU35_A17
30.2 7.1695 3.8775 DU35_A17
34.3 5.9525 3.625 DU40_A17
38.4 4.7745 3.379 DU40_A17
42.5 3.6565 3.133 NACA64_A17
44.55 3.125 3.01 NACA64_A17
46.6 2.722 2.887 NACA64_A17
50.7 1.9225 2.641 NACA64_A17
54.8 1.128204 2.395001 NACA64_A17
57.5334 0.616491 2.199496 NACA64_A17
58.9 0.37 2.086 NACA64_A17
60.2666 0.238005 1.752512 NACA64_A17
61.6333 0.106 1.419 NACA64_A17
63 0.053 0.7095 NACA64_A17

Table 2: Aerodynamic properties of reference turbine

23
 CHAPTER 6: ANSYS-CFX SIMULATION SETUP
6.1 Overview

In this section description of flow domain and simulation setup is given. Simulations are performed for single
turbine case and two turbines in a row case. The flow domain used is 1500 X 600 X 600 m for single turbine

Figure 11: Isometric view of flow field setup with reference coordinate frame

case and 2350 X 600 X 600 m for two turbines in a row case. The flow domain is surrounded by six
boundaries and contains three types of mesh regions. These boundaries and mesh regions will be
explained in later sections. The actuator disc is built according to original dimensions i.e. on a scale of 1:1.
In two turbines in a row case, the turbines are placed 849.5m apart. Figures 11 and 12 show the setup for
the single turbine case in the ANSYS-CFX. Axial velocities are parallel to x-axis while tangential velocities
are in planes parallel to yz-plane. The origin of the reference coordinate frame is at the center of turbine as
shown in Figure 11. The fluid used in simulations is air at 25˚C. The magnitude of inlet velocity is set at 8
m/s for all the calculations.

6.2 Boundary Conditions

The flow domain has 6 boundaries which are named inlet, outlet, top, bottom and 2 sides. The
characteristics of these boundaries are as follows.

24
 TOP TOP

CFMR
FFMR
ADMR
SIDE
SIDE INLET OUTLE
FFMR
T
ADMR
CFMR

GROUND GROUND

Front View Side View

SIDE

ADMR

FFMR
INLET CFMR
OUTLE
T

SIDE

Top View

Figure 12: Front, side and top views of flow field setup showing the meshes and the boundaries

6.2.1 Inlet:

The boundary type for this boundary is set to “Inlet” in the ANSYS-CFX Pre. The direction constraint of
“Inlet” boundary type requires that the flow direction is parallel to the boundary surface normal. Uniform
velocity is applied to this boundary with no turbulence.

6.2.2 Outlet:

This boundary has a boundary type of “outlet” defined in ANSYS-CFX. At this boundary, a relative static
pressure of 0 Pa is applied.

25
6.2.3 Top:

This boundary is the top surface of the fluid domain. It is defined by boundary type of “Opening” in ANSYS-
CFX Pre. An opening boundary condition allows the fluid to cross the boundary surface in either direction.
For example, all of the fluid might flow into the domain at the opening, or all of the fluid might flow out of the
domain, or a mixture of the two might occur. At this boundary, a relative static pressure of 0 Pa is applied.

6.2.4 Ground:

This boundary defines the bottom of the fluid domain. It is defined by boundary type of “Wall” in ANSYS-
CFX Pre. The fluid cannot cross this boundary. Free slip condition is used for this wall. In this case, the
velocity component parallel to the wall has a finite value (which is computed), but the velocity normal to the
wall, and the wall shear stress, are both set to zero.

6.2.5 Sides:

These boundaries define the two sides of the fluid domain. They are defined by boundary type of
“SYMMETRY” in ANSYS-CFX Pre. The symmetry boundary type imposes constraints that ‘mirror’ the flow
on either side of it. For example, the normal velocity component at the symmetry plane boundary is set to
zero and the scalar variable gradients normal to the boundary are also set to zero

6.3 Mesh structure of fluid domain

The six boundaries defined above confines 3 types of mesh regions. These are a fine field mesh region, a
coarse field mesh region and the actuator disc mesh region. These 3 mesh regions are shown in the Figure
12.

6.3.1 Actuator disc mesh region (ADMR)

The actuator disc, replacing the fully resolved turbine geometry, is inserted in the flow field not as a solid
body but as a mesh. This is because the solid body will prevent the flow to pass through it. On the other
hand, the mesh will allow the flow to pass through it. But to impart the same effect on the fluid as that of a
fully resolved turbine geometry, momentum source terms are applied in this region of the mesh. The radius
of the actuator disc mesh complies exactly with the radius of the turbine, which is 63 m. The thickness of
the actuator disc mesh is set to 0.1 m. Thickness less than 0.1 m causes problems in the convergence of
the analysis.

Figure 13 shows the front view of the actuator disc mesh. It can be seen from this figure that the actuator
disc mesh a has a finer resolution near the tip region as compared to the region between the hub and the
tip. This is because some of the flow variables, like axial and tangential induced velocities and angle of
attack, change steeply near the tip region and a finer mesh resolution at the tip region helps to a more
accurate prediction of flow characteristics. For the mesh statistics of this region, we refer to Table 3.

26
 Figure 13: Front view of actuator disc mesh. Right diagram shows full mesh.
Left diagram shows zoomed in tip region

6.3.2 Fine field mesh region (FFMR)

As seen in Figures 11 and 12, there is a fine field mesh region around the actuator disc mesh. This region
of the fluid domain has a finer mesh resolution as compared to rest of the fluid domain (coarse field mesh
region). The flow characteristics change rapidly in the vicinity, both upstream and downstream, of the
actuator disc and to accurately capture these rapidly changing flow characteristics, a finer mesh resolution
is used. For the mesh statistics of this region, we refer to Table 3.

Mesh regions Actuator disc Fine field mesh Coarse field mesh
Number of nodes mesh region
26789 region
239465 region
599432
Hexahedra 23872 226152 573860
Volume (m3) 1258.72 4.0734e+6 5.35954e+8
Table 3 : Mesh Statistics

6.3.3 Coarse field mesh region (CFMR)

The fluid domain around the actuator disc mesh region and the fine field mesh region is the coarse field
mesh region. It has a coarser mesh resolution as compared to the mesh resolution of fine field mesh region.
A coarser mesh resolution is used in this region to make the simulations computationally more efficient.
Flow characteristics change rapidly only around the actuator disc and hence a finer mesh resolution is used
in that region. In the rest of the fluid domain, the change in flow characteristics is relatively gradual and the

27
use of coarser resolution yields both accurate results as well as efficient computation. For the mesh
statistics of this region, we refer to Table 3.

6.4 Mesh dependence check

For a mesh dependence check, three meshes are used, coarse mesh, medium mesh and fine mesh. The
mesh statistics are given in Table 4. Simulations are performed with these three meshes using CFD-BEMT
model for TSR = 4 and TSR = 6. The power and thrust coefficients are compared for different meshes to
check mesh sensitivity. The results are shown in tabular form (Table 5) as well as graphical form (Figure
14).

Clearly, the performance remains fairly constant for the different meshes. The variation in performance is
more pronounced at higher TSR. In this thesis, a medium mesh resolution is used.

Coarse Medium Fine

ADMR 14792 23872 41864

FFMR 141956 226152 549286

CFMR 235420 573860 1990440

Total 392168 823884 2581590

Table 4: Number of Hexahedra for different mesh regions

TSR = 4 TSR = 6

Mesh CP CT CP CT

Coarse 0.224493 0.378085 0.459224 0.639462

Medium 0.224858 0.377768 0.461813 0.643123

Fine 0.224707 0.37795 0.461459 0.642109

Table 5: Performance comparison for different mesh types

28
 Coefficient of Performance Coefficient of Thrust
0.5 0.7
0.6
0.4
0.5
0.3
0.4
TSR 6 TSR 6
0.2 0.3
TSR 4 TSR 4
0.2
0.1
0.1
0 0
Coarse Medium Fine 1 2 3

Figure 14 : Performance comparison for different mesh types

6.5 The FORTRAN Code to calculate momentum source terms.

The BEMT and the LLT are applied using FORTRAN code coupled to ANSYS CFX. The flow chart showing
the working of the code is given in Figure 15. There are three important subroutines in the FORTRAN code
which are used for the implementation of the model. These three subroutines are named Value_readin,
Source_Term and Lifting_Line subroutines. A brief description of these subroutines are also given below.

6.5.1 Value_Readin Subroutine

This subroutine is called only once at the start of simulation. This subroutine reads the input files. Input data
include TSR, hub radius, tip radius, number of blades, number of lifting line elements, airfoil geometry
distribution (distribution of chord, pitch and airfoil used along the radius) and airfoil data (α, CL and CD data)
of the 8 airfoils used in the simulation. After reading the data, this subroutine stores the data in the memory
management system (MMS) of ANSYS-CFX. The data is later retrieved from the MMS in the Source_Term
subroutine and is used in further calculations.

6.5.2 Lifting_Line Subroutine

This subroutine is called by Source_Term_subroutine only if the control volume lies in the BEMT+LLT
region. This subroutine applies Lifting line theory and gives as an output, a matrix of dimensions (lifting line
elements X 3). Three columns of the matrix are non-dimensional radial locations of lifting line elements,
axial Induction factors and tangential induction factors. This matrix is used by the Source_Term subroutine
to calculate momentum source terms and additional momentum source terms in BEMT+LLT region. Lifting
Line theory is applied using a standalone FORTRAN code already developed at IST, Lisbon. The structure
of the code was adapted to make it compatible with ANSYS-CFX.

29
 ANSYS-CFX: Simulation Started

Value_Readin Subroutine Memory

Called only once at start of
Management
simulation
System
parameters
Flow field
Iterations continue if residuals are
greater than a limit value

Input Data

Source_Term Subroutine
Called in every flow field
iteration

from LLT
va and vt
Source Terms
Momentum

Lifting_Line Subroutine
Called if control volume is in
BEMT+LLT region

ANSYS-CFX: Source terms applied on the actuator disc, iterations are performed
on the flow field and residuals calculated

Iterations stops if residuals KEY

are smaller than a limit value
FORTRAN SUBROUTINE

ANSYS-CFX PROCESS

DATA TRANSFER

FORTRAN SUBROUTINE CALLED

ANSYS-CFX ITERATIONS

Figure 15: Flow chart for implementation of momentum source terms

30
6.5.3 Source_Term Subroutine

This subroutine calculates the momentum source terms for every control volume of actuator disc. It
calculates the momentum source terms using the data stored in MMS by Value_readin subroutine, the
current flow field data provided by ANSYS-CFX and the data provided by Lifting_Line subroutine.

31
 CHAPTER 7: SIMULATION RESULTS AND DISCUSSION

7.1. Overview of the chapter

In this chapter, results of simulations conducted with five different CFD models are presented and
discussions are made for single turbine and two turbines in a row. The five models are:

1. CFD-BEMT model which is a hybrid of CFD and BEMT.

2. CFD-BEMT TL model which is a hybrid of CFD and BEMT with the Prandtl tip loss factor.

3. CFD BEMT+LLT M1 model which is an enhanced actuator disc model with the additional
momentum source terms derived from Method 1.

4. CFD BEMT+LLT M2 model which is an enhanced actuator disc model with the additional
momentum source terms derived from Method 2.

5. CFD BEMT+LLT M3 model which is an enhanced actuator disc model with the additional
momentum source terms derived from Method 3.

In the first part of this chapter, the results obtained from simulations for the single turbine case will be
presented for all five CFD models. Results for performance are compared with the reference result obtained
from FAST V8 code. Results for the local flow field parameters at the turbine are compared with the results
obtained from LLT IST code.

In the second part of this chapter, performance results for the downstream wind turbine obtained from
simulations of two turbines in a row are presented. The idea is to find out how the performance of a
downstream wind turbine is effected by the presence of an upstream wind turbine. The results obtained
from the five different CFD models are only compared with each other because reference results are not
available, either experimental or from any commercial software tool, for downstream wind turbine.

7.2 Results and discussion for one turbine case

The following results and discussions are included in this section.

1. Performance: 𝐶𝑃 and 𝐶𝑇 .

2. Local Flow field parameters; axial and tangential induction factors, α and 𝐶𝐷 , along the span
of the turbine blade.

3. Axial velocity field.

4. Tangential velocity field.

5. Turbulence Kinetic Energy.

32
7.2.1 Performance

7.2.1.1 Power Coefficient

Figure 16 shows the variation in 𝐶𝑃 with TSR. The reference results from FAST V8 are also presented in
this figure. FAST- BEMT and CFD-BEMT models have higher 𝐶𝑃 throughout the range of TSR presented
except for TSR = 9 where the 𝐶𝑃 predicted from CFD-BEMT+LLT M3 is higher than CP predicted from Fast
BEMT.

Power Coefficient
0.55 FAST BEMT
FAST BEMT TIP LOSS
CFD-BEMT
0.5 CFD-BEMT TL
CFD-BEMT+LLT M1
CFD-BEMT+LLT M2
CFD-BEMT+LLT M3
0.45

0.4
CP

0.35

0.3

0.25

0.2

0.15
3 4 5 6 7 8 9 10
TSR

Figure 16: Variation of power coefficient with TSR

𝐶𝑃 predicted from Fast BEMT and CFD-BEMT are in good correlation from TSR = 4 to TSR = 7. Above
TSR = 7, CFD-BEMT over predicts 𝐶𝑃 compared to FAST BEMT and the difference increases with the
increase in TSR.

33
CFD-BEMT TL under predicts CP as compared to CP from FAST BEMT with tip loss from TSR = 4 to TSR
= 6.5. At TSR = 6.5 𝐶𝑃 predicted from CFD-BEMT TL and FAST BEMT with tip loss almost coincides. Above
TSR = 6.5, CFD-BEMT TL over predicts 𝐶𝑃 compared to FAST BEMT with tip loss and the over prediction
further increases with increase in TSR.

𝐶𝑃 predicted from CFD-BEMT+LLT M1, CFD-BEMT+LLT M2 and CFD-BEMT+LLT models show good
correlation from TSR = 4 to TSR = 7. Above TST = 7, 𝐶𝑃 values predicted from these methods start to
deviate from each other with the 𝐶𝑃 from CFD-BEMT+LLT M3 model being the highest and from CFD-
BEMT+LLT M1 being the lowest. These three enhanced actuator disc models under predict 𝐶𝑃 compared
to FAST BEMT with tip loss from TSR = 4 to TSR = 5.5. From TSR = 5.5 to TSR = 7, 𝐶𝑃 predicted from
enhanced actuator disc models and from FAST BEMT with tip loss shows good correlation. Above TSR =
7, the enhanced actuator disc models over predict 𝐶𝑃 as compared FAST BEMT with tip loss.

For all models except CFD-BEMT+LLT M3, maximum 𝐶𝑃 lies either at TSR = 7.5 or at TSR = 8. But for
CFD-BEMT+LLT M3 model, maximum 𝐶𝑃 lies at TSR = 9 and it may further increase above TSR =9.

7.2.1.2 Thrust coefficient

Thrust Coefficient
1
FAST BEMT
FAST BEMT TIP LOSS
CFD BEMT
0.9 CFD-BEMT TL
CFD-BEMT+LLT M3
CFD-BEMT+LLT M2
CFD-BEMT+LLT M1
0.8

0.7
CT

0.6

0.5

0.4

0.3
3 4 5 6 7 8 9 10
TSR
Figure 17: Variation of thrust coefficient with TSR

34
Figure 17 shows variation of 𝐶𝑇 with TSR. Reference results from FAST V8 are also presented in this figure.
It is observed that for all five CFD models and for reference results, 𝐶𝑇 increases with increase in TSR. It is
very clear from this figure that at a particular TSR, the CFD model with higher momentum source terms will
have a higher 𝐶𝑇 and vice versa. CFD-BEMT TL has reduced axial force throughout the blade span due to
the tip loss effect as compared to other methods, hence its axial momentum source terms and its 𝐶𝑇 is
lowest for all TSRs presented here. CFD-BEMT+LLT M3 model has the highest axial source terms in the
BEMT+LLT region, therefore it has highest predicted values of 𝐶𝑇 as compared to all other methods at all
TSRs presented. Similar is the case with CFD-BEMT+LLT M2 model. The momentum source terms for
CFD-BEMT+LLT M2 model in tip region are higher than those for CFD-BEMT, CFD-BEMT TL and CFD-
BEMT+LLT M1 models, so 𝐶𝑇 predicted from CFD-BEMT+LLT M2 model is also higher than CFD-BEMT,
CFD-BEMT TL and CFD-BEMT+LLT M1 models.

Further it is observed that 𝐶𝑇 predicted from FAST BEMT and CFD-BEMT are in very good correlation
throughout the range of TSR presented. However, 𝐶𝑇 predicted from FAST BEMT with tip loss effect is not
in good correlation with 𝐶𝑇 predicted from CFD-BEMT TL. 𝐶𝑇 calculated from CFD-BEMT TL is lower as
compared to 𝐶𝑇 calculated from FAST BEMT with tip loss effect throughout the range of TSR presented.
Rather 𝐶𝑇 predicted from CFD-BEMT+LLT M1 and FAST BEMT with tip loss are in strong correlation
throughout the range of TSR.

7.2.2 Local Flow field parameters

7.2.2.1 Axial induction factor

Axial induction factor (TSR = 8)

1.2
CFD-BEMT CFD-BEMT TL CFD-BEMT+LLT M2
CFD-BEMT+LLT M1 LLT IST CODE CFD-BEMT+LLT M3
1

0.8
va/U

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
r/R
Figure 18: Variation of axial induction factor with dimensionless radial distance at TSR = 8

35
Figure 18 represents the variation of the axial induction factor with dimensionless radial distance at TSR =
8. Results from the LLT IST code are used for comparison. As expected, all CFD models predict lower axial
induction factors throughout the span of turbine blade in comparison with the Lifting line. In the tip region,
CFD-BEMT TL has the lowest axial induction factors due to reduced forces resulting from the tip loss effect.
The axial induction factor in the tip region approaches the Lifting line for CFD-BEMT+LLT M2 and CFD-
BEMT+LLT M3 models. CFD-BEMT+LLT M3 model has the highest additional momentum source terms
and hence has the highest axial induction factors in the tip region. Even axial induction factors in BEMT+LLT
region for this model are higher than those from LLT IST except at the very tip end. Still at the very tip end,
CFD-BEMT+LLT M3 model is not capable to predict high axial induction factors as are predicted with LLT
IST code. One reason for this is that, ANSYS-CFX is based on control volume approach and hence instead
of calculating axial induction factor at the very tip end, it calculates an average of the axial induction factors
in the last cell towards the tip end. The value of the axial induction factor at the very tip end will be higher
than the average value calculated for the last control volume. For CFD-BEMT+LLT M1, flow field prediction
in the tip region is rather poor and axial induction factors are smaller even as compared to those obtained
from BEMT-CFD model. This is because additional momentum source terms for CFD-BEMT+LLT M1
method are very small and hence do not have a strong impact. Near the blade root, all CFD models behave
opposite to LLT IST code. LLT IST code predicts high axial induction factors at the root, to achieve zero lift
condition like the tip end, while the CFD models predict low axial induction factors near the root.

7.2.2.2 Tangential Induction Factor

Tangential induction factor (TSR = 8)

0.8
CFD-BEMT CFD-BEMT TL
CFD-BEMT+LLT M2 CFD-BEMT+LLT M1
0.7
LLT IST CODE CFD-BEMT+LLT M3

0.6

0.5
vt/U

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
r/R

Figure 19: Variation of tangential induction factor with dimensionless radial distance at TSR = 8

36
Figure 19 shows the variation of tangential induction factor with dimensionless radial distance at TSR = 8.
Here it is important to mention that the tangential induction factors are obtained by dividing tangential
velocity with incoming freestream air velocity (not by dividing tangential velocity by 2Ω, which is the usual
case). From 𝑟/𝑅 = 0.2 to 𝑟/𝑅 = 0.7, all CFD models slightly over predict tangential induction factor as
compared to LLT IST code. However above 𝑟/𝑅 = 0.7, particularly at the tip region, CFD models over
predict tangential induction factors compared to LLT IST code to large extents. For enhanced actuator disc
models, the effect of additional momentum source terms is to reduce the tangential induced velocities near
the tip region and match them to those obtained from LLT IST code. That is why there is a dip in the values
of tangential induction factors predicted from CFD-BEMT+ LLT M3 model near 𝑟/𝑅 = 0.9. After the dip the
tangential induction factors rise sharply and at the tip it is even larger than the tangential induction factors
predicted from CFD-BEMT and CFD-BEMT TL models. The reason for this rising trend of tangential
induction factor near the tip is still unclear. CFD-BEMT+LLT M2 model also follows similar trend as CFD-
BEMT+LLT M3 model but the dip and rise trend in induction factor near the tip is very small as compared
to the trend from CFD-BEMT+LLT M3 model.

7.2.2.3 Angle of Attack

Angle of attack (TSR = 8)

18
CFD-BEMT CFD-BEMT TL
CFD-BEMt+LLT M2 CFD-BEMT+LLT M1
16 LLT IST CODE CFD-BEMt+LLT M3

14

12

10
α (˚)

0
0 0.2 0.4 0.6 0.8 1
-2
r/R

Figure 20: Variation of angle of attack with dimensionless radial distance at TSR = 8

37
Figure 20 represents the variation of angle of attack, α, with dimensionless radial distance at TSR = 8.
Results from LLT IST code are used for comparison. According to LLT, the angle of attack at the tip region
should reduce to a value that generates zero lift condition. That is why the angle of attack from LLT IST
codes shows a sharp decrease at the tip region. Except CFD-BEMT+LLT M3 model, all CFD models predict
higher α throughout the length of turbine blade as compared to LLT IST code. In most of the BEMT+LLT
region, CFD-BEMT+LLT M3 slightly under predicts the angle of attack compared to LLT IST code.

Although α for zero lift condition (as predicted by LLT IST code) is not achieved with any CFD model, clearly
the prediction of α in the tip region is improved with CFD-BEMT+LLT M2 and CFD-BEMT+LLT M3 models
as compared to CFD-BEMT, CFD-BEMT TL and CFD-BEMT+LLT M1 models. CFD-BEMT+LLT M1 model
has very small additional momentum source terms in the tip region and hence prediction of α with this
method is poor even as compared to CFD-BEMT method. α prediction from CFD-BEMT TL model is highest
in the tip region because the tip loss reduces the blade forces which in turn results in poor prediction of
induced velocities and hence α.

Overall, in the tip region, predictions of α from CFD-BEMT+LLT M3 model follow the results from LLT IST
code to highest degree and predictions from CFD-BEMT TL model follow these results to lowest degree.

Further, again it can be observed that the predictions from CFD models and LLT follow opposite trend in
the root region because in the LLT, the condition of zero lift at the root must be met.

7.2.2.4 Drag coefficient

Figure 21 represents variation of 𝐶𝐷 with dimensionless radial distance at TSR = 8. Results from LLT IST
code are used for comparison. Predictions of 𝐶𝐷 form all CFD models show good correlation with 𝐶𝐷
obtained from LLT IST code for most part of the blade length from 𝑟/𝑅 = 0.4 to 𝑟/𝑅 = 0.9.

For 𝑟/𝑅 smaller than 0.3, all CFD models considerably over predict 𝐶𝐷 compared to LLT IST code. This is
because at the hub region, angle of attack prediction from CFD models are high as compared to angle of
attack predictions from LLT IST code and higher angles of attack leads to higher 𝐶𝐷 for the airfoils used in
that region (DU21_A17 and DU25_A17).

At the tip region 𝐶𝐷 predicted from CFD BEMT, CFD-BEMT+LLT M3 and CFD-BEMT+LLT M2 models show
very good correlation with 𝐶𝐷 predicted from LLT IST code. CFD-BEMT TL and CFD-BEMT+LLT M1 slightly
over predicts 𝐶𝐷 compared to LLT IST code with the predictions from CFD-BEMT TL model being the
highest. These over predictions of 𝐶𝐷 from CFD-BEMT TL and CFD-BEMT+LLT M1 models are due to
higher angles of attack predicted from these models as compared to angles of attack predicted from other
CFD models in the tip region

38
 Drag coefficient
0.06
CFD-BEMT CFD-BEMT TL CFD-BEMT+LLT M1
0.05 CFD-BEMT+LLT M2 CFD-BEMT+LLT M3 LLT IST code

0.04

0.03
CD

0.02

0.01

0
0 0.2 0.4 0.6 0.8 1
r/R

Figure 21: Variation of the drag coefficient with dimensionless radial distance at TSR = 8

7.2.3 Axial velocity field

7.2.3.1 Average axial velocity upstream and downstream of the turbine

Figure 22 shows the variation of average axial velocity plotted against axial distance (x-coordinate)
upstream and downstream of the turbine. The graph shows two sets of curves. One set of five curves (light
colors) is for TSR = 8 and the other set of five curves (dark colors) is for TSR = 4. First of all, it can be
clearly observed from the graph that, the higher the TSR, the greater the reduction in velocity from the
upstream to the downstream side. This is because for higher TSRs, axial (thrust) forces are higher, axial
reactive forces are higher and hence a larger reduction in axial velocity takes place.

Further, it can also be noted that generally, the higher the axial momentum source terms from a CFD
method, the larger will be the reduction in velocity from the upstream side to downstream side. For TSR =
4, CFD-BEMT+LLT M3 has the highest axial momentum source terms in the tip region, therefore causing
the largest reduction in velocity in this region (refer to Figure 24) and hence the greatest reduction in
average velocity among all five CFD methods for TSR = 4. CFD-BEMT TL has the lowest axial momentum
source terms throughout the length of blade due to tip loss effect, hence the reduction in velocity is the
lowest. Similar is the case with the remaining three methods. It can also be seen that this graph also
complies with the graph of 𝐶𝑇 against TSR. The method with higher 𝐶𝑇 has higher thrust, higher axial
reactive force and hence greater reductions in axial velocity.

39
 Average axial velocity
CFD-BEMT+LLT M1 TSR 4 CFD-BEMT TL TSR 4
8 CFD-BEMT+LLT M3 TSR 4 CFD-BEMT+LLT M2 TSR 4
CFD-BEMT TSR 4 CFD-BEMT+LLT M1 TSR8
CFD-BEMT TL TSR 8 CFD-BEMT+LLT M3 TSR 8
CFD-BEMT+LLT M2 TSR 8 CFD-BEMT TSR8
7.5

7
Average axial velocity (m/s)

6.5

5.5

4.5

4
-400 -200 0 200 400 600 800
Axial distance (m)

Figure 22: Variation of average axial velocity with axial distance for all CFD models at TSR 4 and TSR 8

For TSR = 8, generally the trend is same as that of TSR = 4. But there is one important behavior for CFD-
BEMT+LLT M3 model that needs to be discussed. The decrease in average axial velocity is highest for this
method till about 50 m downstream of the turbine, after that velocity starts to increase again. In figure 23, it
can be seen in the plot of CFD-BEMT+LLT M3 model at TSR = 8 that negative velocity is achieved near
the tip region at the downstream side very close to turbine. High reduction in velocity generates high shear
between the wake boundary and undisturbed free stream air. This leads to turbulence generation which
causes mixing of the wake with undisturbed free stream air and hence increases the velocity in the wake
region. Turbulence generation can also be verified from turbulence plots in figure 28. The plot of CFD-
BEMT+LLT M3 at TSR = 8 clearly shows high turbulence.

40
7.2.3.2 Axial Velocity field displayed on a longitudinal plane (xy-plane) passing
through turbine center

CFD-BEM - TSR 4 CFD-BEM TSR 8

CFD-BEM TL – TSR 4 CFD-BEM TL TSR 8

CFD-BEMT+LLT M1 TSR 4 CFD-BEMT+LLT M1 TSR 8

CFD-BEMT+LLT M2 TSR 4 CFD-BEMT+LLT M2 TSR 8

CFD-BEMT+LLT M3 TSR 4 CFD-BEMT+LLT M3 TSR 8

Figure 23: Axial velocity field shown on xy-plane for all CFD models at TSR 4 and TSR 8

41
Figure 23 shows the axial velocity field plotted on xy-plane for all CFD methods at TSR = 4 and TSR = 8.
The scales for TSR = 4 plots and TSR = 8 plots are different for clarity purpose and to cover full range of
velocities involved. TSR = 4 plots clearly show that velocities downstream of the turbine for CFD-BEMT TL
model are higher than those obtained from CFD-BEMT model. This is obvious because in CFD-BEMT TL
model, axial forces on blades are reduced due to tip loss effect throughout the length of blade and hence
axial source terms are also small. This difference is not observable for TSR = 8 because TSR = 8 plots
have scale of wider range.

Further, since the additional momentum source terms are very small for CFD-BEMT+LLT M1, there is not
much observable difference between the plots for CFD-BEMT model and CFD-BEMT+LLT M1 at TSR = 4.
This observation also complies with Figures 17, 18 and 22 where the differences in CT, axial induction factor
and average axial velocity respectively for these two models are very small for TSR = 4. For CFD-
BEMT+LLT M3 model, axial momentum source terms are of highest magnitude at the tip region
(BEMT+LLT region), hence there is a dark blue region at the downstream side of tip region clearly indicating
high reduction in axial velocity in this region. For the CFD-BEMT+LLT M2 model, the additional momentum
source terms are of medium level among the three enhanced actuator disc models, hence there is light
blue color at the downstream side of tip region which indicates that velocity is decreased in this region but
not as much as that for CFD-BEMT+LLT M3 model.

Lastly, the plot for CFD-BEMT+LLT M3 model at TSR = 8 shows that axial velocity at the downstream side
near the turbine at the tip region has achieved negative value of about -0.3 m/s. This negative value is
achieved because the additional momentum source terms are very high leading to huge reduction of
velocity. But the velocity does not remain negative further downstream at tip region and it increases
because as already mentioned in previous section of results, turbulence is generated which causes the
velocity to increase.

7.2.3.3 Axial velocity field downstream of turbine at 200m and 600m

Figure 24 represents the variation of axial velocity along the y coordinate at z = 0 at a distance of 200m
downstream of the turbine. The graph shows two sets of curves. One set of curve (dark colors) is for TSR
= 4 and the second set of curves (light colors) is for TSR = 8. Here in these graphs, it is again experienced
that generally the greater the TSR, the lower will be the value of velocity. The set of curves for TSR = 4
generally shows higher velocity than the set of curve for TSR = 8. Also it can be observed that when
additional source terms are applied in the tip region for CFD-BEMT+LLT M2 and CFD-BEMT+LLT M3
models, velocity is more reduced in the tip region as compared to rest of the blade region. Velocity is lowest
in the tip region for CFD-BEMT+LLT M3 as it has the highest axial momentum source terms at the tip
region. Furthermore, CFD-BEMT+LLT M1 does not show reduction in velocity in the tip region as compared
to CFD-BEMT since the additional momentum source terms are very small for this method. CFD-BEMT TL,
as expected, shows the highest velocities in the outer blade region as compared to rest of the CFD models

42
 Axial velocity along y coordinate at 200m downstream at TSR =
4 and TSR = 8
100

80

60

40
y-coordinate (m)

20

0
2 3 4 5 6 7 8 9

-20

-40

-60

-80

-100
axial velocity (m/s)
CFD-BEMT TSR 4 CFD-BEMT TL TSR 4 CFD-BEMT+LLT M1 TSR4
CFD-BEMT+LLT M2 TSR4 CFD-BEMT+LLT M3 TSR 4 CFD-BEMT TSR 8
CFD-BEMT TL TSR 8 CFD-BEMT+LLT M3 TSR 8 CFD-BEMT+LLT M2 TSR8
CFD-BEMT+LLT M1 TSR 8

Figure 24: Variation of axial velocity along y coordinate at z = 0 at 200m downstream of turbine for TSR
= 4 and TSR= 8
since the axial forces are reduced due to implementation of tip loss effect. For all CFD models, the axial
velocities are highest in the center region because for Hub, Cylinder 1 and Cylinder 2 used in the center
region of turbine, 𝐶𝐿 is zero which leads to reduced axial reactive forces.

Furthermore, it is important to mention here that for CFD-BEMT+LLT M3 model at TSR = 8, the velocity in
the tip region has already started to recover at 200 m downstream of the turbine due to turbulence. That is
why in the tip region, axial velocities predicted from CFD-BEMT+LLT M3 model are higher than those
predicted from rest of the CFD models. The same thing can be observed in Figure 25 which representing

43
 Axial velocity along y coordinate at 600m downstream at TSR = 4 and
TSR =8

80

30
y coorinate (m)

2 3 4 5 6 7 8
-20

-70

axial velocity (m/s)

-120

CFD-BEMT TAR 4 CFD-BEMT TL TSR 4 CFD-BEMT+LLT M3 TSR 4

CFD-BEMT+LLT M2 TSR 4 CFD-BEMT+LLT M1 TSR 4 CFD-BEMT TSR 8
CFD-BEMT TL TSR 8 CFD-BEMT+LLT M3 TSR 8 CFD-BEMT+LLT M2 TSR 8
CFD-BEMT+LLT M1 TSR 8

Figure 25: Variation of axial velocity along y coordinate at z=0 at 600m downstream of turbine for TSR =
4 and TSR = 8

the variation of axial velocity with the y coordinate at z = 0 at a distance of 600 m downstream of the turbine.
The axial velocity profile for CFD-BEMT+LLT M3 model at TSR = 8 shows large recovery from velocity
deficit generated due to extraction of energy by the turbine. Hence turbulence is very helpful in recovering
the reduced velocities in the turbine wake.

Another observation can be made from Figures 24 and 25. The stream tube expands more at higher TSR.
At TSR = 4, stream tube diameter is around 140 m in diameter at 200 m downstream of turbine while at
TSR = 8, stream tube diameter is about 160 m at 200 m downstream of turbine. This is because at higher
TSR, more energy is extracted from the wind and hence, the stream tube expands more in the process of
recovering the negative pressure to atmospheric level at downstream side of turbine.

44
7.2.4 Tangential velocity downstream of the turbine

Tangential velocity at 200m downstream at TSR 4

80.00

60.00

40.00
y coordinate (m)

20.00

0.00
0 0.1 0.2 0.3 0.4 0.5 0.6

-20.00

-40.00

-60.00

-80.00
tangential velocity (m/s)
CFD-BEMT CFD-BEMT TL CFD-BEMT+LLT M3
CFD-BEMT+LLT M2 CFD-BEMT+LLT M1

Figure 26: Variation of tangential velocity along y coordinate at z=0 at 200m downstream of turbine for TSR = 4

In Figures 26 and 27, the distribution of tangential induced velocities along y coordinate at 200 m
downstream of the turbine for all CFD models are given for TSR = 4 and TSR = 8. Since the results are
almost symmetric about the line y = 0 m, discussion will only be made for results along positive y-coordinate.
For TSR = 4, the tangential induced velocities from all CFD methods almost coincide in the center region
of wake from y = 0 to y = 40 m. In outer wake region beyond y = 40 m, CFD-BEMT TL clearly shows lower
tangential induced velocities as compared to CFD-BEMT. This is evident as due to tip loss effect CFD-
BEMT TL model has lower forces, particularly at the tip region of the turbine, as compared to CFD-BEMT
model. From LLT IST code, tangential induced velocities are lower in magnitude as compared to those
calculated from CFD-BEMT models (see Figure 19). Therefore, in CFD-BEMT+LLT M1, CFD-BEMT+LLT
M2 and CFD-BEMT+LLT M3 models, the effect of additional momentum source term is to reduce the total
momentum source terms as compared to CFD-BEMT model. Reduced total momentum source terms would
generate lower tangential velocities. This effect can be clearly seen in the tip region of the wake for CFD-
BEMT+LLT M3 model as additional momentum source terms are highest for this model. The CFD-
BEMT+LLT M1 model has additional momentum source terms of very small magnitude, therefore tangential

45
 Tangential velocity at 200 m downstream at TSR = 8
100.00

80.00

60.00

40.00
y coordinate (m)

20.00

0.00
0 0.2 0.4 0.6 0.8 1 1.2 1.4
-20.00

-40.00
CFD-BEMT
-60.00 CFD_BEMT TL
CFD_BEMT+LLT M1
-80.00 CFD-BEMT+LLT M2
CFD-BEMT+LLT M3
-100.00
tangential velocity (m/s)

Figure 27: Variation of tangential velocity along y coordinate at z=0 at 200m downstream of turbine for TSR = 8

velocities from this model, like other parameters, coincide with the tangential velocities from CFD-BEMT
model.
At TSR = 8, similar behavior is observed for tangential velocities of CFD models except for a few differences.
Generally, the tangential velocities are higher at TSR = 8 as compared to those obtained at TSR = 4. CFD-
BEMT+LLT M3 shows significantly reduced tangential velocities as compared to those obtained from CFD-
BEMT+LLT M1, CFD-BEMT and CFD-BEMT TL in the wake region outwards beyond y = 50 m. This is
again due to turbulence mixing. Generally, the wakes for different models at TSR = 8 have expanded more
as compared to those at TSR = 4.

7.2.4 Turbulence Kinetic Energy

Figure 28 shows turbulence kinetic energy plotted on xy-plane. Plots on left show different CFD models at
TSR = 8. Plots on right show CFD-BEMT+LLT M3 method at different TSRs. For CFD-BEMT, CFD-BEMT
TL, CFD-BEMT+LLT M1 and CFD-BEMT+LLT M2 models, there is no observable turbulence kinetic energy
present at TSR = 8. However, for CFD-BEMT+LLT M3 model, turbulence kinetic energy is seen clearly at
TSR = 8

46
 CFD-BEMT TSR 8 CFD-BEMT+LLT M3 TSR 4

CFD-BEMT TL TSR 8 CFD-BEMT+LLT M3 TSR 5.5

CFD-BEMT+LLT M1 TSR 8 CFD-BEMT+LLT M3 TSR 7

CFD-BEMT+LLT M2 TSR 8 CFD-BEMT+LLT M3 TSR 8

CFD-BEMT+LLT M3 TSR 8 CFD-BEMT+LLT M3 TSR 9

Figure 28: Turbulence kinetic energy plotted on xy-plane . Plots on right show different CFD models at
TSR = 8. Plots on left show CFD-BEMT+LLT M3 method at different TSRs

For CFD-BEMT+LLT M3 model at TSR = 8, turbulence kinetic energy is maximum at downstream side near
the turbine. As the distance downstream of the turbine increase, turbulence kinetic energy decreases and
penetrates towards the center. The plot of axial velocity field at xy-plane for CFD-BEMT+LLT M3 model at
TSR = 8 shown in Figure 23 also supports this turbulence graph of CFD-BEMT+LLT M3 model at TSR = 8.

47
The high turbulence is originated in the region where the axial velocity is negative. Negative velocity is
caused by high source terms.

From the plots on right side of Figure 28. It is clearly seen that the turbulence level increases with increase
in TSR. This is because higher TSRs generate higher magnitudes of additional source terms leading to
greater reduction in axial velocity. For TSR = 4 and 5.5, no observable turbulence kinetic energy is achieved
for the selected scale. For TSRs = 7, 8 and 9, we can see observable levels in turbulence which increases
with TSR.

7.3 Results and discussion for two turbines in a row case

For two turbines in a row case, it is interesting to see how the power and thrust of the downstream wind
turbine is effected by the presence of an upstream wind turbine. For this purpose, the second turbine is
always set to same TSR of 8 while the TSR of first turbine is changed. The simulations are performed for
all five CFD models and the results are compared with each other.

7.3.1 Power and thrust of second turbine

Power of second Turbine vs TSR of First Turbine

2.0E+06 CFD-BEMT CFD-BEMT TL
CFD-BEMT+LLT M3 CFD-BEMT+LLT M2
1.8E+06
CFD-BEMT+LLT M1
1.6E+06

1.4E+06

1.2E+06
Power (W)

1.0E+06

8.0E+05

6.0E+05

4.0E+05

2.0E+05

0.0E+00
0 1 2 3 4 5 6 7 8 9
TSR of upstream wind turbine

Figure 29: Variation of power of downstream wind turbine with TSR of upstream wind turbine

Figure 29 represents the variation of power of the downstream wind turbine with the TSR of the upstream
wind turbine. It is important to notify here that for comparison purpose, results for TSR = 0 corresponds to
the cases with single wind turbine (i.e. no upstream wind turbine is present). For all CFD models, the power

48
of the downstream turbine follows the same general trend with the change in TSR of upstream wind turbine.
The power of the downstream wind turbine decrease with the increase in TSR of the upstream wind turbine.
For different CFD models, there is not much noticeable difference in the decreasing trend of power with
increase in TSR of the upstream wind turbine except that the difference in power predicted from different
CFD models diminishes with increase in TSR of first turbine.

Thrust of second turbine vs TSR of first turbine

4.5E+05
CFD-BEMT CFD-BEMT TL
4.0E+05 CFD-BEMT+LLT M3 CFD-BEMT+LLT M2
CFD-BEMT+LLT M1
3.5E+05

3.0E+05
Thrust (N)

2.5E+05

2.0E+05

1.5E+05

1.0E+05

5.0E+04

0.0E+00
0 1 2 3 4 5 6 7 8 9
TSR of upstream wind turbine
Figure 30: Variation of thrust of downstream wind turbine with TSR of upstream wind turbine

Figure 30 represents the variation of the thrust of the downstream wind turbine with the TSR of the upstream
wind turbine. Here again, results for TSR = 0 corresponds to the cases with single wind turbine. For all CFD
methods, like power, the thrust of downstream wind turbine shows the same general decreasing trend with
increase in TSR of upstream wind turbine. Similarly to the power, for different CFD models, there is not
much noticeable difference in the decreasing trend in thrust with increase in TSR of upstream wind turbine
except that the difference in thrust predicted from different CFD models diminishes with the increase in TSR
of first turbine.

49
 CHAPTER 8: SUMMARY AND CONCLUSION

Three enhanced actuator disc models for CFD analysis of HAWT, CFD-BEMT+LLT M1, CFD-BEMT+LLT
M2 and CFD-BEMT+LLT M3, are developed with the intention of simultaneously improving performance
and flow field (at tip region) as compared to already existing CFD-BEMT models, with and without tip loss
effect. The models are developed by combining the BEMT with the LLT and coupling them to a CFD-RANS
model via actuator disc concept. Simulations for a single turbine and two turbines in a row are performed
for all five models with the commercial CFD code ANSYS-CFX, which is based on control volume fixed in
space approach. Simulation results include performance of wind turbine, local flow field parameters around
wind turbine and flow field parameters upstream and downstream of the wind turbine. For comparison of
performance results obtained from simulations of single turbine, results from FAST V8 code are used. For
comparison of local flow field parameters from the simulations of single turbine, results from LLT IST code
are used. Performance results of downstream wind turbine from simulations of two turbines in row are not
compared to reference results. In this case, the performance results of the downstream wind turbine with
different CFD models are only compared with each other.

None of the enhanced actuator disc model is capable of achieving both the tasks (improving performance
and flow field predictions) simultaneously. The three enhanced actuator disc models show very different
prediction of flow field at the tip region. Among the three enhanced actuator disc models, for flow field
prediction, CFD-BEMT+LLT M1 model shows the worst correlation while CFD-BEMT+LLT M1 model shows
the best correlation with the reference results.

CFD-BEMT+LLT M2 and CFD-BEMT+LLT M3 models show too much over prediction in 𝐶𝑇 as compared
to results obtained from FAST BEMT and FAST BEMT with tip loss throughout the range of TSR presented.
There is very narrow range of TSR, TSR = 5.5 to TSR = 7, where the 𝐶𝑃 predicted by all the enhanced
actuator disc models strongly correlate with the 𝐶𝑃 predicted from FAST V8 with tip loss. Outside that range,
the enhanced actuator disc models either under predict or over predict 𝐶𝑃 . Among the three enhanced
actuator disc models, for performance prediction, CFD-BEMT+LLT M1 model shows the best correlation
while CFD-BEMT+LLT M3 model shows worst correlation with the results obtained from FAST BEMT with
tip loss.

Performance predictions from CFD-BEMT TL model show significant difference from the reference
performance predictions from FAST BEMT with tip loss. Furthermore, prediction of flow field by CFD-BEMT
TL model at the tip region is the worst among all five CFD models used. Clearly, tip loss correction is not a
good approach to be incorporated with CFD-BEMT models.

Hence, it can be said that with enhanced actuator disc models, simultaneous improvement in predictions
of performance and flow field is not possible. Improvement in performance prediction occurs at the expense
of unrealistic predictions of flow field and vice versa.

50
Only CFD-BEMT+LLT M3 model is capable of producing noticeable turbulence level. Turbulence helps to
recover the velocity deficit in the wake. However, the CFD-BEMT+LLT M3 model is not realistic and is only
used in this thesis to observe the effect of high momentum source terms in the tip region. But clearly it can
be seen that high source terms cause high reductions in axial velocities which lead to turbulence. The
enhanced diffusion caused by turbulence is likely to be responsible for the early recover of the wake velocity
observed with this model at TSR = 8.

51
 REFERENCES

1. Anderon, J. (2010). Fundamentals of Aerodynamics (5 ed.). McGraw-Hill Education.

2. ANSYS, Inc. (2013). ANSYS CFX Introduction. Canonsburg, USA.
3. Bai, C. -J., & Wang, W. -C. (2016). Review of computational and experimental approaches to analysis of
aerodynamic performance in horizontal-axis wind turbines (HAWTs). Renewable and Sustainable
Energy Reviews, Volume 63, 506-519.
4. Baltazar, J., Machado, J., & Falcão de Campos, J. (2011). HYDRODYNAMIC DESIGN AND ANALYSIS OF
HORIZONTAL AXIS MARINE. Proceedings of the ASME 2011 30th International Conference on Ocean,
Offshore and Arctic Engineering. Rotterdam, France.
5. Barahona, B., Jonkman, J., Damiani, R., Robertson, A., & Hayman, G. (2014). Verification of the New
FAST v8 Capabilities for the Modeling of Fixed-Bottom Offshore Wind Turbines. United States: National
Renewable Energy Laboratory.
6. Burton , T., Sharpe, D., Jenkins, N., & Bossanyi, E. (2001). Wind Energy Handbook. Wiley.
7. Edmunds, M., Williams, A. J., Masters, I., & Croft, T. N. (2015). BEM-CFD: A Revised Model for Accurate
Prediction. Proceedings of the 11th European Wave and Tidal Energy Conference. Nantes.
8. Esfahanian, V., Salavati Pour, A., Harsini, I., Haghani, A., Pasandeh, R., Shahbazi, A., & Ahmadi, G.
(2013). Numerical analysis of flow field around NREL Phase II wind turbine by a hybrid CFD/BEM
method. Wind Engineering and Industrial Aerodynamics, 120, 29-36.
9. Hansen, M. O. (2008). Aerodynamics of Wind Turbines (Second Edition). London: Earthscan.
10. Iowa Energy Center. (2016). History of Wind Energy. Retrieved from Iowa Energy Center:
http://www.iowaenergycenter.org/wind-energy-manual/history-of-wind-energy
11. Jonkman, J., Butterfield, S., Musial, W., & Scott, G. (2009). Definition of a 5-MW Reference Wind
Turbine for Offshore System Development. National Renewable Energy Laboratory.
12. Manwell, J. F., Mcgowan, J. G., & Rogers, A. L. (2009). Wind Energy Explained: Theory, Design and
Application, 2nd Edition. London: Wiley.
13. Mehta, D., van Zuijlen, A. H., Koren, B., Holierhoek, J. G., & Bijl, H. (2014). Large Eddy Simulation of
wind farm aerodynamics: A review. Wind Engineering and Industrial Aerodynamics, 1-17.
14. Moriarty, P. J., & Hansen, A. C. (2005). AeroDyn Theory Manual. National Renewable Energy
Laboratory.
15. Von Mises, R. (1959). Theory of Flight. Dover Publications.
16. Wind Energy Foundation. (2016). History of Wind Energy. Retrieved from Wind Energy Foundation.
17. Wu, Y. -T., & Porté-Agel, F. (2015). Modeling turbine wakes and power losses within a wind farm using
LES: An application to the Horns Rev offshore wind farm. Renewable Energy, 945-955.

52