MODEL DESIGN AND EXPERIMENTAL

INVESTIGATION OF FLOATING WIND

TURBINE

A Thesis Submitted to
the Graduate School of Engineering and Sciences of
İzmir Institute of Technology
in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE
in Mechanical Engineering

by
Ali Arıdıcı

October 2022
İZMİR
 ACKNOWLEDGMENTS

First, I would like to thank my supervisor Assoc. Prof. Ünver ÖZKOL, for in-
spiring knowledge and for including me in this great project. He has been a great mentor
during my thesis study with his suggestions and by leaving me inspired and eager to learn
more.
Secondly, thank all my colleagues with whom I collaborated on the project and
who did not withhold their support and knowledge from me throughout the project and
thesis. My sincere thanks to especially Assoc. Prof. Bergüzar ÖZBAHÇECİ always
considers me as one of her students.
A big thanks go out to especially İsmail Gürkan DEMİRKIRAN and Ezgi
ŞATIROĞLU always being ready to give suggestions and for some good discussions with
me. Also, I owe a big thanks to İsmail for lots of thesis-writing hours and for supporting
me like a big brother; I will never forget those times.
Last but not least, I would like to thank my family and Gülşah YARIMCA for
always giving me fantastic and indispensable support. I can not thank them enough for
the patience they have shown me. That means everything to me.
 ABSTRACT

MODEL DESIGN AND EXPERIMENTAL INVESTIGATION OF

FLOATING WIND TURBINE

Floating offshore wind energy has great potential (which constitutes almost 80%
of total offshore wind energy) to meet electricity demand of the world at the same time to
reach net-zero emission goal by 2050. Floating offshore wind turbines (FOWT) are able to
achieve highest capacity factor since local effects of the offshore terrains are lesser. Thus,
it receives stronger and more stable wind. On the other hand, combined hydrodynamic
and aerodynamic forces with 6 degrees of freedom (DoF) bring unsteadiness and there-
fore, challenges on FOWT design. Furthermore, significant rotational motions, particu-
larly pitch motion, lead the turbine to transient state which can not be simulated through
conventional numerical tools. Therefore, to understand the dynamics of the FOWT, it is
necessary to conduct experimental studies to obtain results by considering all the param-
eters.
The main aim of the thesis is to investigate the dynamic response of the FOWT
under the extreme wind and wave conditions. A 1/40 Froude-scaled version of the Northel
POYRA P36/300 mounted on the spar-type floating platform was developed by colleagues
as a part of TUBITAK (217M451) project. In this thesis, experimental studies were car-
ried out in the wave flume with a wind nozzle in the hydraulic laboratory of IZTECH
Civil Engineering Department. Atmospheric boundary layer (ABL) was scaled, and in-
struments of the experiment were calibrated to characterize wind nozzle and wave maker,
which are vital to obtaining reliable results. The wind nozzle was designed based on
experimental data to reproduce correct Froude-scaled ABL.

iii
 ÖZET

YÜZER RÜZGAR TÜRBİNİ MODEL TASARIMI VE DENEYSEL

İNCELENMESİ

Toplam denizüstü rüzgar enerjisi potansiyelinin neredeyse %80’ini oluşturan

denizüstü yüzer rüzgar enerjisi, dünyanın elektrik talebini karşılama ve 2050 yılı için
sıfır emisyon hedefine ulaşmada büyük potansiyele sahiptir. Denizüstü bölgelerde arazi-
lerin lokal özelliklerinin rüzgara negatif etkisi çok az olduğundan daha güçlü ve sta-
bil rüzgar vardır. Bu yüzden denizüstü yüzer rüzgar türbinleri (YRT) en yüksek kapa-
site faktörüne erişebilmektedir. Diğer taraftan aerodinamik ve hidrodinamik kuvvetler
YRT’nin 6 serbestlik derecesi ile birleşerek sistemi zamana duyarlı hale getirmektedir.
Kuvvetler türbin üzerindeki ivmeyi indükleyerek sistemin salımına sebep olur. Türbinin
maruz kaldığı bu salınım ,özellikle kanat-iz etkileşimi sırasında, kolay bir şekilde kararsız
hale gelmektedir. Bu durum ekstrem koşullar altında daha ciddi yaşanmaktadır. Yunus-
lama hareketi gibi önemli ölçüde yaşanan açısal hareketler, sistemi türbin ve pervane
arasında geçiş durumuna zorlamaktadır. Bu geçiş durumu mevcut sayısal araçlarla model-
lenememektedir. Bu yüzden YRT dinamiğini anlamak için tüm parametrelerin ele alındığı
deneysel çalışma yapılması gerekmektedir.
Bu tezin ana amacı ekstrem rüzgar ve dalga koşulları altında YRT dinamik
davranışını deneysel olarak incelemektir. TÜBİTAK (217M451) projesinin bir parçası
olarak Northel POYRA P36/300 referans türbini ile spar-tip yüzer platformun 1/40
Froude ölçekli tasarımı ve üretimi çalışma arkadaşlarım tarafından gerçekleştirilmiştir.
Bu tezde, deneysel çalışmalar İYTE İnşaat Mühendisliği Bölümünde bulunan dalga
kanalı ve rüzgar nozul’u kullanılarak gerçekleştirilmiştir. Atmosferik sınır tabaka, IEC
standardı ve Froude benzerliğine bağlı kalınarak ölçeklendirilmiştir. Rüzgar nozul’u
ve dalga yapıcı’nın karakterize edilmesi doğru sonuç alınması konusunda büyük önem
taşımaktadır. Deneyde kullanılan sensörler bu karakterizasyonu en doğru şekilde ya-
pabilmek için kalibre edilmiştir. Gerçek boyutlardaki YRT sisteminin dinamiğinin
anlaşılabilmesi için çeşitli koşullar altında testler uygulanmıştır.

iv
 TABLE OF CONTENTS

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

CHAPTER 1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1. Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2. Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

CHAPTER 2. THEORETICAL BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.1. Scaling Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2. Blade Element Momentum Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.1. Impact of Scaling on Aerodynamic Design . . . . . . . . . . . . . . . . . . . . . . 18
2.3. Wave Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1. Regular Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.2. Irregular Wave . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4. Aerodynamic Effects of Floating Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . 25

CHAPTER 3. LITERATURE REVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.1. Full Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2. Simplified Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

CHAPTER 4. EXPERIMENTAL SETUP OF IZTECH FOR SPAR-TYPE FOWT 36

4.1. Model Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.1.1. Wave Flume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1.2. Wind Nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1.3. Traverse Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.4. Model Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.5. Model Spar-Type Floating Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2. Instrumentation and The Tools of Experiments . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1. Hot-wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.2. Strain gauge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

v
 4.2.3. Image processing tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

CHAPTER 5. RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.1. Atmospheric Boundary Layer Reproduction . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.2. Dynamic Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2.1. Free Decay Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.2.2. Only Wind Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.2.3. Wind and Regular Wave Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2.4. Wind and Irregular Wave Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

CHAPTER 6. CONCLUSION, DISCUSSION AND FUTURE STUDIES . . . . . . . . . 86

6.1. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.2. Discussion and Limitation of the Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.3. Recommendation for Future Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

APPENDICES
APPENDIX A. SIGNAL ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
APPENDIX B. CALIBRATION PROCESS OF HOT-WIRE ANEMOMETER . . . . 99
APPENDIX C. DECOMPOSITION CODE OF COUPLED MOTION . . . . . . . . . . . . . . 103
APPENDIX D. WIND TURBINE’S MOTOR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

vi
 LIST OF FIGURES

Figure Page

Figure 1.1 Installed wind energy capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Figure 1.2 Configurations of commonly used bottom-fixed offshore support struc-
ture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Figure 1.3 Offshore wind atlas of Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Figure 1.4 Configurations of the main concepts of floating wind turbine platform . 5
Figure 1.5 Degree of freedoms of Floating offshore wind turbine . . . . . . . . . . . . . . . . . . . 6
Figure 1.6 Illustration of loads on the floating offshore wind turbine . . . . . . . . . . . . . . . 7
Figure 1.7 Transient state between turbine and propeller state during pitch motion 8
Figure 2.1 1D momentum analysis in control volume with actuator disc . . . . . . . . . . . 14
Figure 2.2 Blade element theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Figure 2.3 Scheme of iterative solution technique of BEM . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Figure 2.4 Sketch of the regular wave and boundary conditions . . . . . . . . . . . . . . . . . . . . 23
Figure 2.5 Possibility of unsteadiness along the blade of NREL 5 MW . . . . . . . . . . . . 27
Figure 3.1 (a) surge and (b) pitch response of spar-buoy floating platform under
an Hs = 10.5m wave and three different wind conditions . . . . . . . . . . . . . . . . 29
Figure 3.2 RAOs derived from white-noise wave excitation with Hs = 7.1m
without and with wind at 21.8 m/s; the colored box indicates the wave
frequency range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Figure 3.3 Left : Comparison of Ct vs T SR curves of geometric-match model,
thrust-match model, and NREL 5 MW prototype. Right: Adjustment
of pitch angle of blade to match Ct with prototype . . . . . . . . . . . . . . . . . . . . . . . . 30
Figure 3.4 Comparison of the resuls of experiment and simulation with and with-
out wind: Hs = 4.14m, Tp = 0.94s, Vrated = 1.3m/s . . . . . . . . . . . . . . . . . . . . . 31
Figure 3.5 RAO of experiment and simulation with and without wind . . . . . . . . . . . . . 32
Figure 3.6 RAO of experiment and simulation with and without wind : Hs =
2.64m, Tp = 7.3s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Figure 3.7 Platform motion spectra outputs of experiment under the wind-wave,only-
wave,and only-wind conditions : Hs = 3.6m, Tp = 10.2s, V = 11m/s . 33

vii
Figure Page

Figure 3.8 Various concepts used in scaled model tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Figure 3.9 HexaFloat robot developed at Politecnico di Milano . . . . . . . . . . . . . . . . . . . . . 35
Figure 4.1 Top-sketch view of the test setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Figure 4.2 The dissipating beach with 1/5 gradient placed at the end of the chan-
nel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Figure 4.3 Exploded view of screens, and a contracting nozzle . . . . . . . . . . . . . . . . . . . . . . 39
Figure 4.4 Nozzle Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Figure 4.5 Traverse Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Figure 4.6 The airfoil section used in the model study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 4.7 Geometric model of blade design parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 4.8 Illustration of metacenter and metacentric height . . . . . . . . . . . . . . . . . . . . . . . . . 44
Figure 4.9 Schematic view of spar platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Figure 4.10 Measurement of center of gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Figure 4.11 Ping-pong balls that placed on the tower . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Figure 4.12 Calibration nozzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Figure 4.13 Dwyer Model M1430 Microtector U-manometer . . . . . . . . . . . . . . . . . . . . . . . . . 50
Figure 4.14 Calibration process of hot-wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Figure 4.15 Calibration curve of hot-wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Figure 4.16 Strain gauge sensor on the tower bottom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Figure 4.17 Calibration curve of Strain gauge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Figure 4.18 View from Tracker program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Figure 5.1 Froude-scaled logarithmic wind profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Figure 5.2 Measured wind profiles at x= 0.5D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Figure 5.3 Measured wind profiles at x= 1D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Figure 5.4 Measured wind profiles at x= 2D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Figure 5.5 Measured wind profiles at x= 3D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Figure 5.6 Measured wind points on y-z plane at x= 1D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Figure 5.7 Wind profile at z=190 cm (o = Each measurement, black solid line =
Average of measurements) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Figure 5.8 Wind profile at z=225 cm (o = Each measurement, black solid line =
Average of measurements) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
viii
Figure Page

Figure 5.9 Wind profile at z=260 cm (o = Each measurement, black solid line =
Average of measurements) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Figure 5.10 Experiment wind profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Figure 5.11 Turbulence intensity of wind profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Figure 5.12 Wind speed difference over rotor swept area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Figure 5.13 Free decay tests result in time domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Figure 5.14 Free decay tests result in frequency domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Figure 5.15 Only wind test, from left to right : 1. Initial position, 2. Position under
the wind, 3. Enhanced position by mooring lines, 4. Rotor starts to
rotation, 5. Transient state between starting of rotation and max rpm, 6.
Position when the rotor at max rpm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Figure 5.16 Time series and PSD of measured data from wave gauge during vari-
ous wave experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Figure 5.17 Power spectral densities and time series of surge, heave, pitch motions
of D-H2-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Figure 5.18 Power spectral densities and time series of surge, heave, pitch motions
of D-H4-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Figure 5.19 Power spectral densities and time series of surge, heave, pitch motions
of D-H6-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Figure 5.20 Power spectral densities and time series of surge, heave, pitch motions
of D-H8-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Figure 5.21 Power spectral densities and time series of surge, heave, pitch motions
of D-H10-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Figure 5.22 Power spectral densities and time series of surge, heave, pitch motions
of D-H12-T02-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Figure 5.23 Power spectral densities and time series of surge, heave, pitch motions
of D-H2-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Figure 5.24 Power spectral densities and time series of surge, heave, pitch motions
of D-H4-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Figure 5.25 Power spectral densities and time series of surge, heave, pitch motions
of D-H6-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
ix
Figure Page

Figure 5.26 Power spectral densities and time series of surge, heave, pitch motions
of D-H8-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Figure 5.27 Power spectral densities and time series of surge, heave, pitch motions
of D-H10-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Figure 5.28 Power spectral densities and time series of surge, heave, pitch motions
of D-H12-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Figure 5.29 Power spectral densities of surge, heave, pitch motions of Irregular-
H8-T04-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Figure A.1 Frequency spectrum of data according to time series . . . . . . . . . . . . . . . . . . . . . 96
Figure A.2 Two different amplitude scale for same data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Figure A.3 Condition of aliasing occurrence in the frequency domain . . . . . . . . . . . . . . . 97
Figure A.4 The smeared amplitude in the entire frequency range which is known
as the leakage problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Figure B.1 Schematic view of the IFA 300 Constant Temperature Anemometer
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Figure B.2 Calibration - Probe Data screen of the IFA 300 . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Figure B.3 Communications screen of the IFA 300 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Figure B.4 Block diagram of LabVIEW for the calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Figure D.1 The Pololu Jrk G2 Configuration Utility user interface . . . . . . . . . . . . . . . . . . . 108

x
 LIST OF TABLES

Table Page

Table 2.1 Scale factor table of the parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1
Table 2.2 Reynolds dissimilitude under Froude scaling (λ = 40
) . . . . . . . . . . . . . . . . . . . . 19
Table 4.1 Operational parameters of the turbine (λ = 1/40) . . . . . . . . . . . . . . . . . . . . . . . . . 41
Table 4.2 Mass values of model turbine components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Table 4.3 Structural and hydrostatic properties of spar platform . . . . . . . . . . . . . . . . . . . . . 45
Table 4.4 Scaled parameters of spar platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Table 4.5 Weights and CoG of the FOWT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Table 5.1 Roughness values according to landscape type . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Table 5.2 Full-scaled version of wave parameters of experiment . . . . . . . . . . . . . . . . . . . . 66
Table 5.3 Initial displacement and angular values which applied in free decay test . 67
Table 5.4 Parameters of regular wave of the tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Table D.1 Details of instruments of wind turbine motor’s . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Table D.2 Preliminary study outputs of wind turbine motor’s . . . . . . . . . . . . . . . . . . . . . . . . 107

xi
 CHAPTER 1

INTRODUCTION

The global warming crisis brings unprecedented environmental impacts due to

carbon emissions. Global average temperature has risen 1.5◦ C-2◦ C since the last century,
this increases number of occurrences and the level of hot extremes, droughtiness and ma-
rine heatwaves, also reductions are observed in sea of North and South pole, (permafrost
and snow cover) (23). To provide an effective solution, wind energy -in recent years
offshore wind energy- has gained lots of attention from governments and industry.
In the preceding year, 93.1 GW of new wind energy capacity was installed world-
wide with the largest increase, growing approximately 13%, in renewable energy genera-
tion. Furthermore, new-deployment of offshore wind turbine exceeds previous increments
with 21.31 GW, reaching 55.67 GW capacity (24). Figure 1.1 shows the accumulation of
installed wind energy capacity of last decade.

Figure 1.1. Installed wind energy capacity (1)

1
 There are several reasons that makes the offshore wind energy ramp up signif-
icantly. Most of favorable onshore sites have already been used or are not economical
anymore. For instance, the stakeholders in Northern Europe region tend to seize the op-
portunity of offshore wind energy in consequence of the lack of space. Moreover, there
are countries that have mostly rugged, hilly, mountainous or forested terrains where the
wind can not fully harvested (25).
Land-use regulations have been changing since the first installations (26). Size of
the onshore wind turbines getting bigger year by year to efficiently harvest wind energy
resource from the site, whereas transporting big components of the turbine is challenging
in comparison to offshore and due to logistic issues wind turbine manufacturers try to
achieve different technologies such as split blade, modular nacelle (27, 28). Additionally,
noise produced by the turbine has become problematic for both society and governments.
Therefore, less terrain are available for efficient wind power deployment.
According to recent statistics on the ordered wind turbines in 2021, the average
nominal power of the onshore and offshore wind turbines were 4.9 MW and 11.2 MW,
respectively (29). Meanwhile, dynamic surface of sea, without any obstacle, provides less
turbulent and stronger wind which creates advantage in terms of power density. New in-
stalled bottom-fixed offshore wind turbines have a capacity factors ranging from 40-50%,
while the onshore counterpart delivering capacity factor of roughly 40%. Furthermore,
floating wind farm of Hywind Scotland reached highest average capacity factor of 57%
with a record-breaking performance (23).
The Glasgow Climate Pact (COP 26) and the Russian invasion of Ukraine high-
light the urgency of taking action about alternative energy sources; as a result, especially
the European governments account to update their offshore wind energy capacity targets
by 2030. The energy policy may be turned into a security policy in the near future (30).
The offshore turbine support structure technology in the wind energy industry
comes from the oil and gas (O&G) industry (31). The support structures are mainly di-
vided into bottom-fixed and floating platforms, which are derived from various mobile
and fixed drilling platforms used in the O&G industry. The vast majority of current off-
shore wind power generation is provided by bottom-fixed wind turbines. Figure 1.2 shows
five different bottom-fixed platforms commonly used in the offshore wind industry:

2
Figure 1.2. Configurations of commonly used bottom-fixed offshore support structure
(2)

1. Monopile: The structure with cylindrical geometry, connected to the wind turbine
tower with the help of a transition piece or directly, is called monopile. It is gener-
ally used at depths less than 20-30 m since its cost is susceptible to water depth.

2. Gravity platform: Gravity platforms provide stabilization with heavy concrete

consisting of large dimensions on the seabed rather than large piles. It is generally
used on a solid seabed less than 30 meters deep.

3. Tripod: The tripod support structure consists of a central cylindrical tower and
steel frames connecting it to three piles. It has a combination of monopile and
jacket structures.

4. Jacket: The jacket support structure consists of three or four main piles embedded
in the seabed and braces connecting the steel legs extending from these piles in the
form of a lattice.

5. Tripile: It is a 3-legged structure similar to a tripod. The central cylinder is shorter,

with three legs fixed to the seabed. Tripod, jacket, and tripile structures are used at
depths of 20-60 meters.

Bottom-fixed offshore wind turbines account for 99.8% of the installed capacity
of the current offshore capacity. On the other hand, the installed capacity of the floating
wind turbines has reached 121.4 MW, with a new installation of 57.1 MW in 2021 (32).
Around 80% of the world’s offshore wind energy resource potential is available where the
sea is deeper than 50 m, and bottom-fixed offshore wind turbines are economically and
technically unsuitable. European offshore wind energy potential is shown in Figure 1.3.
For many countries in Europe, especially the Mediterranean countries, deep waters are

3
the norm on this country’s coasts. Consequently, bottom-fixed wind turbines are neither
option nor are potentially restricted (4).

Figure 1.3. Offshore wind atlas of Europe (3)

Floating offshore wind turbines (FOWTs) are a promising solution to harness this
resource. Therefore, it is estimated that floating wind energy will constitute 6% of the
total offshore wind energy capacity by 2030. For example, floating projects are increasing
rapidly in countries such as France, the UK, the US, South Korea, and Japan, where sea
conditions are suitable for the floating platform (32). Moreover, wind energy potentials
are available far-offshore where deeper waters are encountered in Turkey.
Floating platforms are divided into three groups in terms of physical principle,
shown in Figure 1.4. It provides stability for all systems (rotor-nacelle assembly, tower)
under various conditions (2). An overview of the principles of floating platforms is given
in the following part:

1. Spar-type: The Spar platform has a long cylindrical structure in the vertical direc-
tion. Since most of the system volume is submerged, the buoyancy force provides
stability against external effects as a restoring force and returns the system to static

4
 equilibrium. The lower center of gravity means, the more stable the system will
be. The mooring lines mounted on the spar are embedded with the anchor to the
seabed, keeping the platform stationary.

2. Tension leg platform (TLP): Tension leg platforms are stabilized by tensioned
mooring lines and anchors well-positioned to the seabed. The taut mooring line
allows for a shorter draft and lighter construction. The robustness of the mooring
line and anchor is vital as they are subjected to higher loads. Tensioned lines pose
a risk in the installation process and operational conditions.

3. Barge : Barge-type platform provides stabilization with a large waterplane area

and shortest draft. Catenary lines are used to prevent it from getting dragged by the
waves. However, the barge is more susceptible to the waves than other platforms.
Therefore, semi-submersible is more widely used in sites. Semi-submersible, a
combination of barge and spar, consists of three columns placed vertically into the
water. It has a longer draft and higher weight than the barge. The columns at equal
distances from each other reduce pitch and roll response compared to the barge
(33).

Floating offshore wind energy is now at the beginning of the commercial phase. A com-
prehensive analysis was made in light of production, commercial innovation, and engi-
neering studies carried out in recent years for cost reduction. As a result, it is predicted
that the cost of offshore wind energy will decrease by 37-49% in 2050 (34). However,
since it is not fully commercialized yet, there are many points that need to be developed
and verified.

Figure 1.4. Configurations of the main concepts of floating wind turbine platform (4)

5
1.1. Problem Definition

FOWTs operate under the combined aerodynamic and hydrodynamic forces. Un-
like onshore and bottom-fixed turbines, they have 6 degrees of freedom (DoF), three trans-
lational (surge, sway, heave), and three rotational (pitch, roll, sway). These DoFs, which
cause a highly dynamic environment (35), are shown in Figure 1.5 .

Figure 1.5. Degree of freedoms of Floating offshore wind turbine (5)

Various loads originating from wind, sea current and waves affect the FOWT sys-
tem, as depicted in Figure 1.6. Hydrodynamic forces bring low frequency and high iner-
tial loads to the tower and nacelle. These loads induce the acceleration of FOWT, and the
system oscillates in different DoFs (7). The thrust force on the rotor causes an additional
pitch angle which can be compensated by the pitch restoring stiffness of the floating plat-
form and the mooring line. Depending on the platform type, this angle can be significant
from an aerodynamic point of view. Dynamic inflow is the aerodynamic effect that occurs
when the loads on the rotor change rapidly during blade pitching, wind gust, and platform
motion. The rotor gradually reaches a new equilibrium state due to the new load condi-
tion, which causes an overshoot in the instant angle of attack and thrust force. There is

6
a delay in the load response of the blade element momentum (BEM) engineering model.
Also, the model’s assumption of momentum balance in its derivation causes deviation in
the FOWT load analysis (10). In their work, Sebastian and Matha (36, 37) show that the
BEM model cannot accurately model this lag response.

Figure 1.6. Illustration of loads on the floating offshore wind turbine (6)

The pitch and yaw motion of the floating platform affect the blades of the wind
turbine through blade-wake interaction which causes transient flow on the rotor. The
blades interact with their own wake whilst the rotor goes towards wake direction, and it
creates highly unsteady aerodynamic around the rotor which called vortex ring state (7),
illustrated in Figure 1.7. Furthermore, the platform’s yaw and pitch rotational oscillations
make the local wind speed on the rotor a motion-dependent variable and consequently
invalidate the uniform wind speed assumption. Besides, it is difficult to model the non-
axial flow on the rotor accurately (38).
In addition to the cases mentioned above, under the circumstances such as extreme
waves and viscous forces, the effect of wave-current interaction on moored floating plat-
form structures is challenging to model in numerical simulations. Therefore, scaled mod-
els and field experiments are essential in verifying nonlinear effects, extreme and complex

7
forces, and dynamic system behavior. Field experiments are not preferred because they
require high cost, long installation, and validation period. Also, the damage it may take
under extreme conditions poses a financial risk. On the other hand, scaled-down model
experiments performed in a controlled laboratory environment are more time-efficient and
have low risk (39).

Figure 1.7. Transient state between turbine and propeller state during pitch motion (7)

Northel POYRA P36/300, suitable for the Izmir Institute of Technology wave
flume conditions, is chosen as the reference turbine. The turbine has a rotor diameter of
36 meters and a nominal power of 320 kW. In order to examine the dynamic behavior
of the wind turbine, mounted on the spar-type platform, under extreme wind and wave
conditions, 1/40 scaling has been made based on the operational conditions at 20 m/s
wind speed where the turbine operates. Unlike the multi-megawatt turbines, scaled-down
testing is much more essential as the small-size FOWT system will exhibit more unstable
behavior under the same extreme conditions.

1.2. Thesis Outline

This chapter explains the importance of floating wind turbines to achieve the
carbon-zero goal by addressing several points. Besides the various advantages of the
floating wind turbine, many technical points still need improvement. These problems are

8
addressed in the problem definition, and it is shown that scaled-down experiments can be
employed as a solution to scrutinize these problems in the initial design of FOWT.
An appropriate scaling methodology has to be applied to accurately model the dy-
namic behavior of the floating wind turbine system under various fluid-based loads. The
second chapter will present the necessary theoretical background to design a scaled-down
wind turbine. In addition, the relationship between Reynolds and Froude, two dimen-
sionless numbers used in fluids scaling, will be given. The effects of Reynolds number
under Froude scale conditions will be discussed. Blade element momentum used in wind
turbine rotor design will be explained, and the hydrodynamic forces acting on the floating
wind turbine are briefly explained. Besides, the theoretical background of regular and ir-
regular waves used in the experimental tests will be described with the linear wave theory.
Last but not least, the aerodynamic effects of floating dynamics on the wind turbine will
be explained and clarified why the scaled model test should be applied for the FOWT.
In chapter 3, the literature will be reviewed. The scaled model wind/wave basin
tests will be divided into two groups and investigated. With the results of the first tests,
various test methods will be presented to improve the performance of the scaled model.
In the 4th chapter, information about the experimental setup in the IZTECH Civil
Engineering hydraulic laboratory and the scaling studies of the FOWT system will be
explained, and the sensors employed during the experiment will be introduced.
In the Results chapter, the atmospheric boundary layer will be scaled according
to the IEC standard and Froude-scaling. In order to get the desired wind characteristic
from the wind nozzle, the comprehensive test study will be explained, and the results
will be given. First, the natural frequency in each decay test for the individual degree of
freedom will be calculated as a result of free decay tests. Then, the wave flume tests with
and without wind conditions with the determined wave parameters will be introduced.
Finally, the time-series and power spectral densities of each degree of freedom will be
given.
Finally, in chapter 6, the experiment results will be scrutinized, the experimental
system will be evaluated, and suggestions will be made for future studies.

9
 CHAPTER 2

THEORETICAL BACKGROUND

Most physical systems can be investigated with a small-scale model that can
achieve the same behavior as a full-scale prototype. A scaling law is required to sim-
ulate the entire physical concept behind the system adequately.

2.1. Scaling Methodology

In fluid mechanics studies, there are three basic scaling laws. First, geometric
similarity: model and prototype geometry must have the same shape to achieve geomet-
ric similarity. All structure lengths used in the model experiment must be reduced by
the same ratio as the prototype. The flow field must also provide similar requirements.
The second is kinematic similitude: the ratio of velocities and accelerations between the
model and the prototype must be the same. With a given geometric similarity, it is also
necessary to satisfy the dynamic similarity, the third similarity, in order to provide kine-
matic similarity or similarity of motion. This is because five different forces act on the
flow field around the structure, creating the fluid-structure interaction. These are surface
tension, viscosity, gravity, pressure, and elasticity. In order to provide dynamic similarity,
the similarity of forces, the ratio between the model and the prototype must be constant
(40).
The system becomes too complicated to satisfy the scaling law if all parameters
are included. Therefore, complete similarity cannot be achieved except for scaling ratios
very close to one (40). For large offshore structures, surface tension and elasticity are
often neglected. Buckingham’s pi theorem, a parametric approach, establishes the rela-
tionship between the model and the prototype by deriving a set of dimensionless quantities
from essential variables that affect the system dynamics. The Reynolds number, which
represents the relationship between inertial and viscous forces, is widely used in wind
tunnel tests for proper rotor scaling in wind turbines. On the other hand, the Froude num-
ber, which represents the ratio of inertial forces to gravitational forces, is employed to
ensure the similarity of dominant hydrodynamic forces in floating structure tests. There-

10
fore, scaling laws can be satisfied by keeping Froude and Reynolds numbers constant.
Reynolds number is defined in equation 2.1:
ρ∗V ∗L
Re = (2.1)
μ
where ρ is the density of the fluid in which the structure is located, V is velocity coming to
the structure, characteristic length of the structure is L, and μ is dynamic viscosity of the
fluid. If the relationship between the prototype and the scaled-down model is re-defined,
after applying the geometric scale factor, using Reynolds number similarity:

V M ∗ L M = V M ∗ LP ∗ λ = V P ∗ L P (2.2)
LM
λ= (2.3)
LP
Dynamic viscosity and density are not taken into account since it is the same fluid. If the
equation is simplified concerning velocity:

VP
VM = (2.4)
λ
The λ was chosen 1/40 for our scaled-down model. Then, 800 m/s is obtained
from the above equation for Reynolds similitude. The model’s speed causes supersonic
flow environments, and correct flow modeling cannot be obtained. Since Reynolds simil-
itude could not be achieved, Froude scaling was applied to create the model environment
to represent the driver loads accurately. The Froude number, a measurement of bulk flow
characteristics like the wave in free surface flows, is defined as follows:

V
Fr = √ (2.5)
g∗L
where g is the gravitational constant, L is hydraulic depth, and V is water velocity. The
same procedure as the previous step is followed to obtain the wind speed employed on
the FOWT.
VM VM VP
√ =√ =√ (2.6)
g ∗ LM g ∗ LP ∗ λ g ∗ LP
If the equation is simplified for velocity:
√
VM = VP ∗ λ (2.7)

Considering Equations 2.4 and 2.7, it can also be seen that it is not possible to utilize
Froude and Reynolds numbers simultaneously to achieve similarity. The Froude-scaled
velocity equation gives the time scale of λ0.5 through unit analysis. For accurate dynamic

11
modeling, the structural and displaced water mass ratios between the prototype and the
model should be preserved:
M assM M assP
= (2.8)
ρ ∗ V olumeM ρ ∗ V olumeP
Since the volume scale is λ3 from geometric scaling, this yields:

M assM = M assP ∗ λ3 (2.9)

Result of the scaling between parameters defined by fundamental quantities, e.g.,

L, M, and T (mass, length, time). The scaling factor of each parameter is determined by
making a unit analysis. Table 2.1 shows the scaling factor of essential parameters in the
FOWT experiment.

Table 2.1. Scale factor table of the parameters

Parameters Unit Scaling Factor
Length L λ
Mass M λ3
Time T λ0.5
Volume L3 λ3
Density M L−3 1
Velocity LT −1 λ 0.5

Acceleration LT −2 1
Frequency T −1 λ −0.5

Force M LT −2 λ3
Moment M L2 T −2 λ4
Power M L2 T −3 λ3.5

Froude-scaling in the model FOWT gives lower wind speeds than the prototype.
As a result, the Reynolds number is significantly reduced with scaled-down characteristic
length and wind speed. Therefore, how the effects of the rotor’s aerodynamic perfor-
mances are modelled will be explained in 2.2.1.

2.2. Blade Element Momentum Theory

Fundamentally, a wind turbine converts the kinetic energy from the wind to me-
chanical energy with the rotor, transfers the mechanical power to the generator with the

12
gearbox in the drivetrain, and converts it into electrical energy. The process of harvesting
kinetic energy in the wind to produce mechanical energy at the rotor can be explained
through momentum theory, which is a tool for understanding wind turbine aerodynamics
(41). Conservation of mass, energy, and momentum laws are defined in a control volume
by making steady, axisymmetric, inviscid, incompressible flow assumptions,

ρV.dA = 0 (2.10)
CV

 
uρV.dA = T − pdA.ex (2.11)
CV CV

ruθ ρV.dA = Q (2.12)
CV
  
p 1 2
+ ||V|| ρV.dA = P (2.13)
CV ρ 2
where u, v, uθ , axial, radial, and azimuthal velocity components of V, respectively. A
represents area vector normal to the control volume, T is the thrust force acting on the
rotor in the axial direction, p is the pressure. Torque and power of the rotor are represented
with Q and P, respectively. The main dimensionless parameters used in characterizing the
aerodynamic performance of the wind turbine are given below:

T
Thrust coefficient : CT = 1 (2.14)
2
ρAU02

P
Power coefficient : CP = 1 (2.15)
2
ρAU03
ΩR
Tip speed ratio : λ= (2.16)
U0
where Ω is the rotor’s angular velocity, A is the rotor’s swept area, R is the swept area’s
radius, and U0 is the wind speed.
The rotor can be modeled using an ideal actuator disk, where there is a sudden
pressure drop as the flow passes over it, while the wind speed decreases without disconti-
nuity. The high pressure in front of the actuator disc is named p+ , while the low pressure
just behind it is named p− . The thrust force can be obtained by multiplying the pressure
difference and area,

ΔT = (p+ − p− ).ΔA (2.17)

13
 Figure 2.1. 1D momentum analysis in control volume with actuator disc (8)

According to the continuity equation ( eq. 2.10), each cross-section must have the
same mass flow rate:

ṁ = ρV dA = ρU0 A0 = ρU A = ρU1 A1 (2.18)

The following equation is obtained for thrust, where the axial momentum balance in Equa-
tion 2.11 is applied to the control volume:

T = ṁ(U0 − U1 ) = ρU A(U0 − U1 ) (2.19)

The assumptions made for the axial momentum allow the Bernoulli equation to be
established between far upstream (1) and just in front of the rotor (2), and just behind the
rotor (3) and far downstream (4), as seen in Figure 2.1,

1 1
p0 + ρU02 = p+ + ρU 2 (2.20)
2 2
1 1
p1 + ρU12 = p− + ρU 2 (2.21)
2 2
With the knowledge of the equivalency of pressures in the far upstream and downstream
(p0 = p1 ), these two Bernoulli equations (2.20-2.21) are subtracted from each other:
1
(p+ − p− ) = ΔP = ρ(U02 − U12 ) (2.22)
2

14
 If the thrust in Equation 2.17 is combined with the Equations 2.19-2.22 and sim-
plified:
1
U = (U0 + U1 ) (2.23)
2
An axial induction factor (a) is introduced to appropriately define both the velocity on the
rotor and the velocity downstream with the freestream velocity (42). The axial induction
factor describes how much kinetic energy of air the rotor absorbs.
U0 − U
a= (2.24)
U0
U = (1 − a).U0 (2.25)

U1 = (1 − 2a).U0 (2.26)

Thrust and power extraction can be rewritten in terms of axial induction:

T = 2ρAU02 a(1 − a) (2.27)

P = U.T = 2ρAU03 a(1 − a)2 (2.28)

Considering the equations 2.14-2.15, the thrust and power coefficients can also be rewrit-
ten, respectively:
CT = 4a(1 − a) (2.29)

CP = 4a(1 − a)2 (2.30)

In axial momentum theory, radial and azimuthal variations of the flow are not
taken into account since the actuator disc that gives instant pressure drop is replaced,
and rotational flow is ignored. On the other hand, in general momentum theory, where
radial and azimuthal variations are included (9). The angular velocity (Ω) of the blade,
uθ formed just behind the blade due to blade rotation, and the radial distance of the blade
on the rotor are considered. The Bernoulli equation of both sides of the rotor is rewritten
with the new parameters:
1 1
p+ + ρ(u2 + v 2 + (Ωr)2 ) = p− + ρ(u2 + v 2 + (Ωr − uθ )2 ) (2.31)
2 2
When the above equation is simplified for instant pressure drop:
1
ΔP = −ρΩruθ + ρu2θ (2.32)
2
The local area is separated into annular segments in the radial direction (ΔA =
2πrΔr). According to the axial momentum equation (eq. 2.11), the new thrust equation
is given as follows:

15
 ΔT = (p+ − p− )ΔA = ρU (U0 − U1 )ΔA − (p1 − p0 )ΔA1 + ΔY (2.33)

where ΔA1 is the cross-sectional area in downstream volume and ΔY = (p − p0 )dA.ex
is axial force contribution from lateral pressure. If Equations 2.32 and 2.33 are combined:

ΔA1 ΔY 1
+
ρU (U0 − U1 ) − (p1 − p0 ) = −ρΩruθ + ρu2θ (2.34)
ΔA ΔA 2
When induction factors are introduced, and the last equation is written in terms of these
factors,
U U1  uθ
a=1− , b=1− , a =− , (2.35)
U0 U0 2rΩ

p1 − p0 1 − a 1 ΔY  
2b(1 − a) = 1 .
2 1−b
+ 1 2. + 4λ2 x2 a (1 + a ) (2.36)
2
ρU0 2
ρU0 ΔA
where x = r
R
, Equation 2.36 is the most comprehensive version of the momentum equa-
tion as it includes all parameters. However, since it contains many unknowns, assump-
tions have been applied to solve this equation. For example, the contribution from the
lateral pressure is neglected because it is both a small and a problematic effect to calcu-
late. Glauert model is the most widely used model rotor design and analysis (43). Glauert
has simplified the equation by making the assumptions given below,

p0 ∼
= p1 , u2θ1 ∼
= u2θ , b∼
= 2a (2.37)

Equation 2.36 is simplified with these assumptions and gives the relationship between
axial and tangential induction factors depending on the radial position:
 
a(1 − a) = λ2 x2 a (1 + a ) (2.38)

Under optimum operating conditions of the rotor, it is possible to set the closed system
equations with this equation, allowing the rotor to reach its upper-performance limits.
Besides, its combination with blade element theory forms the pillar of blade element
momentum (BEM) theory (41).
In BEM theory (43), loads are calculated by two independent methods. 1-D mo-
mentum theory is employed with the 2-dimensional airfoil data of the local blade element.
Consider the flow on the rotor with constant wind speed and direction. The flow around
cross-section of the blade is vectorial sum of the free stream wind speed, the angular ve-
locity of the rotor, and the induced velocities due to the presence of the blade. As shown

16
in Figure 2.2b, the angle between the incoming wind and the airfoil chord is called the
local angle of attack. With the given Reynolds number, type of airfoil, and angle of attack,
the lift and drag coefficients on each airfoil can be calculated, as seen in Figure 2.2a. With
this airfoil characteristic, loads along the blade are calculated utilizing momentum theory,
and induced velocities are determined.

(a) Discretization of blade (44) (b) Force diagram of blade element (9)

Figure 2.2. Blade element theory

The axial load and torque are defined as follows:

dT 1 2
= Nb Fn = ρcNb Urel .Cn (2.39)
dr 2

dQ 1 2
= Nb rFt = ρcNb rUrel .Ct (2.40)
dr 2
where Nb is the blade number, Vrel is relative velocity, and c is chord of the blade. Fn and
Ft represent the axial and tangential loads, respectively. These forces can be defined with
non-dimensional coefficients:

Fn Ft
Cn = 1 2
and Ct = 1 2
(2.41)
2
ρcUrel 2
ρcUrel
These coefficients constitute a combination of the airfoil’s lift and drag coefficients de-
pending on the flow angle (φ),

Cn = Cl cos φ + Cd sin φ and Ct = Cl sin φ − Cd cos φ (2.42)

17
The Cl and Cd coefficients, which depend on the geometry of the airfoil, are functions of
Reynolds and angle of attack. Due to the material strength of the rotor, it goes from thick
airfoil to thin from hub to tip. Since there is less angular velocity in the inner part of the
blade, the incoming flow comes with a higher angle of attack. The blade has a locally
varying twist angle to achieve maximum efficiency at the optimum angle of attack during
the operation. As a result, the inner part of the blade has higher twist angle than the outer
part of the blade. The sum of the twist angle and the pitch angle is defined by θp , then the
local angle of attack α = φ − θp . The velocity triangle formed on the airfoil is defined as
follows,

Ωr(1 + a ) U0 (1 − a)
cos φ = and sin φ = (2.43)
Urel Urel
2
When Urel is found from the above relationship and the thrust and torque equation is
rewritten according to that,

U 2 (1 − a)2 U0 (1 − a)Ωr(1 + a )
2
Urel = 0 2 = (2.44)
sin φ sin φ cos φ

dT ρNb cU02 (1 − a)2

= .Cn (2.45)
dr 2 sin2 φ

dQ ρNb cU0 (1 − a)Ωr2 (1 + a )
= .Ct (2.46)
dr 2 sin φ cos φ
If the axial and angular momentum conservation equations are obtained in terms of in-
duction factors; thrust, and torque:
dT
= ρ(U0 − U1 )2πrU = 4πρrU02 a(1 − a) (2.47)
dr
dQ 
= ρruθ 2πrU = 4πρr3 ΩU0 a (1 − a) (2.48)
dr

Recall that uθ = 2Ωra . After combining the final thrust and torque equations and some
algebra, the result is given:
1  1
a= 4 sin2
and a = 4 sin φ cos φ
(2.49)
σCn
φ
+1 σCt
−1

The rotor’s area, where it encounters the wind, is smaller than the actuator disc. Therefore,
it is defined as σ = Nb c/2πr, which is called solidity. Although equations of 2.49 are
explicit, the right side of the equations depends on two different variables (angle of flow
and airfoil geometry). Therefore, the total of system equations is non-linear and implicit.
These system equations are solved using non-linear solution techniques or the simple
iterative loop technique seen in Figure 2.3.

18
 Figure 2.3. Scheme of iterative solution technique of BEM (9)

2.2.1. Impact of Scaling on Aerodynamic Design

In order to achieve aerodynamic similarity in a scaled-down model rotor, three

scaling laws must be met. The kinematic similarity between the model and the prototype
ΩR
can be obtained by keeping the tip speed ratio (T SR = U0
) constant (45). Additionally,
maintaining the scaled mass in the model rotor with the same center of gravity, which will
provide the same mass distribution as the prototype, is very important to correctly match
system dynamics, i.e., gyroscopic effect (46).
To consistently maintain kinematic similarity, maintaining dynamic similarity is
needed for the loads on the rotor. As explained in the previous section, the lift and drag
forces on the blade are the functions of the Reynolds number. Table 2.2 shows that using
Froude scaling leads to a significant decrease in the Reynolds number.

1
Table 2.2. Reynolds dissimilitude under Froude scaling (λ = 40
)

Quantity Prototype Scaling Factor Model

3
Reynolds 6.4 ∗ 104 λ2 252.98

At low Reynolds numbers, especially on the low-pressure side of the airfoil, the
flow displacement thickness is much larger. As a result, the flow separates in the laminar

19
region close to the airfoil’s leading edge, disrupting the optimal pressure distribution. This
effect causes immense changes in the lift (Cl ) and drag (Cd ) coefficients of the airfoil
sections.
One solution to this problem is to modify the blade geometry according to the low
Reynolds number (12). However, the scenario where both thrust and power coefficients
match perfectly is not possible. Therefore, prioritization is required between thrust and
torque force, which is a function of power. Since the reaction force on the mooring
lines connected to the floating platform due to the thrust force is much higher than the
overturning due to torque, keeping the scaled thrust force has been given priority (47).
Furthermore, there is a significant decrease in Cl and an increase in Cd at low Reynolds
numbers. Since the lift coefficient contributes positively to the torque force (eq. 2.42),
it is challenging to obtain scaled torque. In conclusion, even if the model thrust force
is preserved, the dynamic similarity is not fully achieved since all aerodynamic forces
cannot be scaled correctly (6). This imperfection in scaling is called scaling-effect or
Reynolds-effect.

2.3. Wave Theory

This section considers linear wave theory that creates hydrodynamic loads acting
on the FOWT. These are hydrostatic loads, loads from incident waves, radiation loads, and
diffraction loads. Linear wave theory is defined by potential flow using incompressible,
inviscid, and irrotational assumptions (48).

• Radiation load : The force generated on the floating platform as it oscillates in the
sea in various modes without any incident wave. Moreover, it causes waves that
reflect off the floating platform.

• Diffraction load : The load distribution occurs on the surface due to the incident
wave, while the floating platform is stationary without movement.

• Hydrostatic load : The static load occurs perpendicular to the surface due to the
pressure of the surrounding water on the surface of the submerged structure. The
load on the structure is proportional to its draft.
 t
FiHydro = FiW aves + ρgV0 δi3 − CijHydrostatic qj − Kij (t − τ )q̇j (τ )dτ (2.50)
0

20
The combination of linear hydrodynamic forces is represented above as Fhydro . The fourth
term on the right-hand side is the integral of the additional loads from wave radiation that
are not accounted for in Aij . Aij represents the impulsive hydrodynamic added mass;
detailed information is available in Sarpkaya’s ”Wave Forces on Offshore Structures”
book (48). The second and third terms on the right-hand side represent hydrostatic loads
by the combination of buoyancy force, and center of buoyancy (CoB) change from the
movement of the floating platform. The first term on the right-hand side is the total
excitation load on the floating platform due to the incident wave and is closely related to
the wave elevation (49). The following part will explain wave elevation in irregular and
regular waves.

2.3.1. Regular Wave

The Navier-Stokes equations can describe all flows in nature. However, these
equations are difficult to solve and often need computational calculations such as com-
putational fluid dynamics (CFD) tools. Reynolds stresses from the viscid forces formed
in the boundary layer are the main problem for these equations. While describing ocean
waves in the study (50), friction’s effects are ignored, simplified the Navier-Stokes equa-
tion, and obtained Euler equations. As a result of neglecting viscid forces, all external
forces are conservative. Kelvin’s theorem states that the flow will remain irrotational if it
does not have initial vorticity (ω),
⎡ ⎤
− ∂w
⎢ ∂y
∂v
∂z ⎥
⎢ ⎥
ω = curl(u) = ∇ × u = ⎢ ∂u
⎢ ∂z − ∂w ⎥ = 0
∂x ⎥
(2.51)
⎣ ⎦
∂v
∂x
− ∂u
∂y

where u = [u, v, w]T . Since the flow is irrotational, velocity potential (Φ(x, y, z)) can be
defined everywhere in the flow field as:
 T
∂Φ ∂Φ ∂Φ
u = [u, v, w] = grad(Φ) = ∇Φ =
T
, , (2.52)
∂x ∂y ∂z
When the equations 2.51-2.52 are combined with Euler equations, the general version of
Bernoulli’s equation is obtained as follows,
2 2 2 
p 1 ∂Φ ∂Φ ∂Φ ∂Φ
gz + + + + + =0 (2.53)
ρw 2 ∂x ∂y ∂z ∂t

21
where g is the gravitational acceleration, ρ is the density of the water, and p is the pressure.
In order to obtain pressure, the equation can be rewritten below:
2 2 2 
1 ∂Φ ∂Φ ∂Φ ∂Φ
p = −ρw gz − ρw + + −ρw (2.54)
2 ∂x ∂y ∂z ∂t
  
=0 in linearized version

In the assumption of incompressibility, the continuity equation must be satisfied, which

can also be preferred as the Laplace equation:
∂u ∂v ∂w ∂ 2Φ ∂ 2Φ ∂ 2Φ
div(u) = ∇.u = + + = ∇2 Φ = + + 2 =0 (2.55)
∂x ∂y ∂z ∂x2 ∂y 2 ∂z
These equations complement potential flow theory. When appropriate boundary condi-
tions are applied, the differential equation in equation 2.55 can be solved and φ can be
found. The pressure can be calculated when the velocity potential (φ) is substituted in
equation 2.54.
Boundary conditions must be applied at an exact surface elevation, z =
η(x, y, z, t). Since there is no flow towards the bottom of the seabed with depth h, the
kinematic bottom boundary condition is defined as:

∂Φ
= 0 for z = −h (2.56)
∂z
The particle on the free surface has to remain there. In other words, the difference in
material derivation between the free surface elevation (η(x, y, z, t)) and the instantaneous
position (z) of the particle must cancel each other out. This state creates the second
boundary condition, the kinematic free surface,
0 0 0
D ∂Φ ∂η 
>
∂Φ ∂η
 ∂Φ ∂η >

∂Φ ∂η

(z − η) = − −  − −  =0 for z = 0 (2.57)
Dt ∂z ∂t 
∂x ∂x ∂y ∂y 
∂z ∂z
Higher-order terms can be ignored since the free surface elevation derivative is small if
one assumes small wave amplitudes. The dynamic boundary condition expresses that the
free surface’s pressure must equal the atmospheric pressure (p = 0). The Bernoulli’s
equation is rewritten according to this expression:
:0


2 2 
2
1 ∂Φ ∂Φ
  ∂Φ ∂Φ
gz − + + − = for z=0 (2.58)

2 ∂x ∂y ∂z ∂t

Since z = η, the equations 2.57-2.58 can be combined to form a single boundary condi-
tion,
∂Φ 1 ∂ 2 Φ
+ =0 for z=0 (2.59)
∂z g ∂t2

22
Lastly, a periodic boundary condition is defined in regular waves with constant form and
period. It states that the flow at two different points with a difference in length L (wave
length) is the same,  
∂Φ  ∂Φ 
= (2.60)
∂x x=x ∂x x=x+L

Figure 2.4. Sketch of the regular wave and boundary conditions (8)

The sketch formed as a result of the defined boundary conditions can be seen in
Figure 2.4. Below equation can be obtained for the incident wave (Φ0 ); after solving the
potential flow problem with given boundary conditions,

 
igA cosh(k(z + h))
Φ0 = R exp{i(−kx cos(θwaves ) − ky sin(θwaves ))} exp{iωt}
ω cosh(kh)
  
Φ0 (x,y,z)
(2.61)
gA cosh(k(z + h))
Φ0 = sin(ωt − kx cos(θwaves ) − ky sin(θwaves )) (2.62)
ω cosh(kh)
R: Real part of equation
2π
ω= T
: Wave frequency with period of T
2π
k= L
: Spatial wave frequency or wave number
A: Wave amplitude
Φ0 : Complex incident wave potential
θwaves : Angle between waves and reference frame of floating structure
θwaves is chosen as 180 degrees according to the wave flume case in the IZTECH
hydraulic laboratory that results in ei(−kx cos(θwaves )−ky sin(θwaves )) = eikx .The relationship
between time and spatial frequency can be explained by dispersion:

23
 ω 2 = gk tanh(kh) (2.63)

After calculating the velocity potential, the first order wave elevation can be found through
dynamic free surface boundary condition,

1 ∂Φ0   i(ωt−kx cos(θwaves )−ky sin(θwaves )) 
η=− = R Ae (2.64)
g ∂t z=0
 
η = A cos ωt − kx cos(θwaves ) − ky sin(θwaves ) (2.65)

2.3.2. Irregular Wave

The actual sea state in nature is irregular and cannot be represented by a single
regular wave (48). Nevertheless, the linearity in potential flow allows for superposition.
Therefore, the irregular wave can be defined as the sum of N regular waves with randomly
shifted phases (ψj ). Surface elevation can then be defined at x = y = 0 as:

 

N 
N
η(x = 0, y = 0, t) = R iψj
Aj e e iωj t
= Aj cos(ωj t + ψj ) (2.66)
j=1
   j=1
Aj

A is the complex amplitude. The waves are stochastic due to their random phases. When
examining a floating structure exposed to stochastic force, the statistical properties of the
stochastic force must be constant over time. Pedersen et al. (51) state that the sea state is
considered stationary for over 3-4 hours period. For stationary processes, the mean square
of surface elevation is constant. When applying it to the wave with no phase at x=y=0 in
the T period, the surface elevation is defined as equal to its variance:
  
2 1 T 2 1 2 cos(T ω) sin(T ω) + T ω 1
η = σn =
2 η dt = A = A2 (2.67)
T 0 2 Tω 2
2π
Since ω = T
, the equation is simplified with sin(2π) = 0. It is also proof that these
waves contain mean energy per unit area,
1 1
Etot = Epotential + Ekinetic = 2. ρgσn2 = ρgA2 (2.68)
2 2
It is seen that there is a relationship between the amplitude of the wave and its energy. The
discrete version of the spectral density function Sη (fj ), which represents the frequency of

24
 ωj
each wave (fj = 2π
) in terms of energy contribution, can be defined as:

N 
N
1
Sη (fj )Δf = σn2 = A2j (2.69)
j=1 j=1
2

Both Sη (fj ) and Aj can be rewritten according to each other,


1 2
Sη (fj )Δf = Aj and Aj = 2Sη (fj )Δf (2.70)
2
The wave’s amplitude can be found with a given spectrum, and the irregular sea
state can be defined using equation 2.66. The JONSWAP spectrum, which is widely used
to represent ocean waves, is defined as follows:
−5  −4 
f f
Sη (fj )Δf = 0.3125Hs2 Tp exp −1.25
fp fp
⎧
⎪
2 ⎫
⎪ (2.71)
⎨ f
−1 ⎬
fp
exp −0.5
  ⎪
⎩
σ ⎪
⎭
. 1 − 0.287 ln(γ) γ
1
where the frequency with the most energy is called peak frequency (fp = Tp
). Hs is
named significant wave height, the mean of the highest 1/3 of the waves, and Hs = 4σn in
conjunction with a standard deviation. If f =< fp , σ = 0.07 else σ = 0.09. γ is the peak
shape parameter that expresses how fully developed the sea state is. γ = 1 represents
the fully developed sea state, in which the JONSWAP spectrum would be the same as the
Pierson-Moskowitz spectrum.

2.4. Aerodynamic Effects of Floating Dynamics

As a result of the motion of the FOWT system in 6 DoF, there are instantaneous
changes in the wind coming to the rotor, especially during the pitch motion. These in-
stantaneous changes, combined with the tilt angle of the shaft, directly affect the angle
of attack of the wind coming to the blade sections. This effect affects aerodynamic per-
formance with phenomena called dynamic stall and violates the basic assumptions of the
BEM model (7). Although advanced turbine control systems and certain DoF have ac-
tive damping applications, FOWT systems have high speeds and accelerations at the rotor
blade sections.
The thrust force acting on the rotor is balanced by the pitch restoring stiffness of
the floating platform and the forces by mooring lines connected to the platform. There-
fore, depending on the floating platform configuration, significant additional pitch angles

25
can occur on the platform. Active damping systems are also used to mitigate this ad-
ditional angle. In the study carried out on the NREL 5 MW turbine (10), the reference
turbine of the National renewable energy laboratory (NREL), a decrease of less than 1%
is observed in the annual production of the turbine until the floating platform angle of
θpitch ≈ 4.5◦ . When an onshore wind turbine with a capacity factor of Cp,onshore is used
as a FOWT, the effect of the pitch angle of the system on power production is approxi-
mated as follows,

1
PF OW T = Cp,onshore ρA(U0 . cos(θpitch ))3 (2.72)
2
As mentioned in the problem definition, the interaction of the blade with the wake
created by itself, which is called blade-wake-interaction, creates unsteadiness in the flow.
Reduced frequency analysis evaluates the unsteadiness created by the platform movement
at a specific frequency (ωplatf orm [rad/s]) on the blade (38),

ωplatf orm c
k= (2.73)
2U
where k represents the regime of the flow around the oscillating airfoil and c is the chord
length of the airfoil. When k is less than 0.05, the flow is steady, and the circulatory
contributions to the airfoil to lift dominate. When k is greater than 0.05, the flow is
unsteady, and the mass contribution from the acceleration effects drives the flow. The
local relative wind speed over the blade section, which is the combination of the free
stream wind speed and the rotational speed of the blade, without taking into account the
induction factors, is as follows:


U= U02 + (rΩ)2 (2.74)

When the local velocity equation is placed in the reduced frequency equation, the flow
regimes formed on the blade are defined by the following equation,

ωplatf orm c(r)

k= (2.75)
2 U02 + (rΩ)2
Figure 2.5 shows the possibility of NREL 5 MW turbine unsteadiness, depending
on different floating platform configurations and operational conditions. The grey areas
represent unsteadiness, and it is inferred from the Figure 2.5 that the inboard sections of
the blade are sensitive to unsteadiness at lower wind speeds.

26
 Figure 2.5. Possibility of unsteadiness along the blade of NREL 5 MW (10)

Last but not least, the axial movement of the platform (Uplatf orm ) affects the blade
tip region. It creates recirculation at the blade tip due to high speeds, known as the vortex
ring state. This recirculation breaks the momentum balance assumption by creating a
highly unsteady transient flow, and the rotor acts as a propeller. Therefore, the following
equation must be satisfied for the BEM model to be valid (10),

U0 − Uplatf orm => 2|aU0 | (2.76)

Under extreme wave loads, the FOWT movement highly depends on the frequency
and height of the wave. Especially a small turbine that tends to exhibit highly unsteady
behavior is considered in this thesis. In order to understand the dynamics of the system,
experimental studies should be carried out to obtain accurate results by taking into account
all parameters.

27
 CHAPTER 3

LITERATURE REVIEW

There are two main approaches in the scaled testing of FOWTs, which are back-
bones of the scaled model testing, full approach and simplified approach. These ap-
proaches have separate ways to represent loads coming from wind or wave. The first one
is full approach which deals with a scaled model of whole wind turbine (i.e., rotor, na-
celle, floating platform) with all its complexity. The second one is simplified approach
which includes the minimum number of components to obtain accurate dynamic behavior
of the FOWT. In simplified approach, one or more actuators are used instead of rotor or
floating platform. Based on measured motion of the floating platform, the turbine loads
are calculated by a software and the software runs the actuators instantaneously during the
tests which is called software-in-loop (SIL) (52). In this study, full approach is considered
for scaled model testing of the reference turbine with spar-type platform.

3.1. Full Approach

The first work in this field is done by testing a 5 MW turbine with the Hywind
spar-buoy concept at the Froude-scale factor of 1/47 (53). In the test, which is performed
to verify the numerical simulation results, active pitch control and active pitch control with
active damping are compared under the above-rated wind speed operating conditions.
The second Froude-scaled FOWT model with a full-approach approach is devel-
oped by Goupee A. et al. (54), known as test campaign of DeepCwind (2010-2013). The
NREL 5 MW reference turbine (55) with a rotor diameter of 126 meters is modeled at
a 1/50 scale, and details can be found in Martin’s thesis (6). The model wind turbine is
mounted on three floating platforms: a tension-leg platform, a semi-submersible, and a
spar buoy (56). Wind speed is not Froude-scaled to capture the overall thrust that con-
tributes to the motion of the FOWT. The platform’s motion and tower load responses are
investigated in various wave and wind conditions, one of the test results can be seen in
Figure 3.1 where it can be seen that the wind speed changes the response at the natural
frequency of the surge. On the other hand, the amplitude in the wind speed frequency

28
range (< 0.02 Hz) raised with increasing wind speed in the pitch response.

Figure 3.1. (a) surge and (b) pitch response of spar-buoy floating platform under an
Hs = 10.5m wave and three different wind conditions (11)

Second phase of DeepCwind experimental test campaign (57), is performed to

validate and calibrate NREL 5 MW wind turbine simulated with FAST (aero-hydro-servo-
elastic code). Froude-scaled wind is not used in this study as well. The discrepancy
between simulation and experiment gets large when the second-order effects of the wave
and dynamic mooring forces are present in the tests.

Figure 3.2. RAOs derived from white-noise wave excitation with Hs = 7.1m without
and with wind at 21.8 m/s; the colored box indicates the wave frequency
range (12)

Summary of the DeepCwind results (12) show that the TSR must remain constant
to get the correct damping effect from the wind in the Froude scale environment. There-
fore, Martin et al. (58) proposed two ways to achieve the desired overall thrust with the

29
turbine Froude scale wind in the study. These options are roughened leading edge and
redesigned blade for low Reynolds. With unpredictable results through roughened lead-
ing edge, Martin has shown that the better way is the thrust-matched redesigned blade.
According to that study, Fowler et al. (46) demonstrated that the thrust-matched blade
could capture full-scale overall thrust at Froude-scale wind speeds in the study.
The NREL 5MW turbine utilized in the DeepCwind project is revised for Froude-
scaled wind and named as Marin Stock Wind Turbine (MSWT) by using CFD and BEM
tools, as well as wind data from the first experiments (59). In addition, the system is
integrated to enable active blade pitch control during the experiment. It is shown in
wind/wave basin tests that MSWT performs much better than the geometric-matched
model in emulating the behavior of the full-scale prototype (13), as shown in Figure 3.3.
The thrust-matched blade, which has approximately three times longer chord length than
the geometric-match blade, reaches 98.1% of the desired CT value at TSR of nearly 7,
which is rated wind speed operating condition. Four different pitch angles are tried in the
model to obtain the prototype CT value under various operating conditions. Ultimately,
with +3.5 degrees pitch angle, it is achieved at a TSR of 5.

Figure 3.3. Left : Comparison of Ct vs T SR curves of geometric-match model, thrust-

match model, and NREL 5 MW prototype. Right: Adjustment of pitch
angle of blade to match Ct with prototype (13)

The WINDFLO project operates the 1 MW two-bladed Vergnet turbine mounted

on a semisubmersible with two thrust-matched blades in the Froude scale factor of 1/40
for hydrodynamic tests and 1/25 for aero-hydro tests (60). The test is carried out to verify
the FastHydro code, which is developed to make numerical simulations in which second-
order wave and dynamic mooring effects are considered. Rotational motions, especially

30
pitch motion, of the model FOWT are investigated under various conditions.
Within the scope of the INNWIND.EU project (61), the 1/45 scale OC4-
DeepCwind semisubmersible floating platform with a 5 MW turbine is scaled up to 10
MW, represented by Robertson et al.(62), and the thrust-matched rotor is produced us-
ing a 1/60 Froude-scaled factor based on that design. Adjustments are made to catch the
overall weight and center of gravity to achieve the same dynamic behavior. Numerical
simulation is validated according to the experimental tests under various conditions at
Ecole Centrale de Nantes (ECN) (63).
1/60 Froude-scale DTU 10 MW reference wind turbine (64) mounted on TLP
platform is tested in Danish Hydraulic Institute (DHI) as part of the INWIND.EU project.
The model was performed under various wave and wind conditions to validate the nu-
merical model (65) implemented in the Flex 5 tool, which is based on unsteady BEM.
The results increased the confidence in the numerical simulation (14). The probability of
exceedance curves with and without wind are given in Figure 3.4; the mean value is sub-
tracted from the whole signal. According to the experimental results, while the amplitude
slightly decreased in the surge through wind, it increased slightly in pitch. It is seen that
the wind increases the heave motion at a high rate.

Figure 3.4. Comparison of the resuls of experiment and simulation with and without
wind: Hs = 4.14m, Tp = 0.94s, Vrated = 1.3m/s, (14)

31
3.2. Simplified Approach

Roddier et al. (66) introduce an innovative version of semisubmersible as a new

floating platform to mitigate pitch motion. All technical details about the early stage of
the product (such as design, site assessment, fabrication, installation, and commissioning)
are assessed in detail. After the aerodynamic simulation is solved with FAST, it is cou-
pled with the hydrodynamic software TIMEFLOAT (TF) to compute the dynamics and
kinematics of the FOWT. A 1/105 scale model of the prototype is produced and tested in
the wave basin at UC Berkeley to validate this simulation (15). The drag disc, consisting
of a foam board, is used instead of a rotor, and an electric motor rotates an aluminum rod
with weights to emulate the gyroscopic effect created by the rotor during operation. The
simulation and experiment results are shown in Figure 3.5 where it is seen that the wind
does not affect the heave, the surge response of the system has slightly decreased through
the wind with the increasing wave period. In addition, the pitch response decreased to
about 12 seconds with the wind, then increased almost linearly with the increase in the
period.

Figure 3.5. RAO of experiment and simulation with and without wind (15)

As an alternative way, Azcona et al. (16) employ a ducted fan to provide thrust
force from the rotor during testing. The semisubmersible platform with a 6 MW wind
turbine is modeled at a 1/40 scale, and several tests are performed at the wave tank of

32
ECN. In order to introduce variable thrust force by the fan, FAST is utilized to provide
real-time feedback with a controller, known as hybrid testing with SIL. The results are
given in Figure 3.6 where peak frequencies are formed at the natural frequency of surge
and pitch. The second peak is occurred by the frequency of the wave. Moreover, it can be
said that the wind increases the system energy at low frequencies, according to the wind
frequency of 0-0.1 Hz in the experiment.

Figure 3.6. RAO of experiment and simulation with and without wind : Hs = 2.64m,
Tp = 7.3s, (16)

Sauder et al. (67) propose a new way for wind/wave model testing of FOWT,
called the real-time hybrid model (ReaTHM). While the aerodynamic forces are emulated
by cables connected to the actuator and a frame placed on the tower top, the hydrodynamic
parameters are measured from the physical test at MARINTEK’s ocean basin . A 5MW-
CSC wind turbine and braceless semisubmersible floating platform are scaled at 1/30 for
basin testing. The measured data from sensors is given as input to the FAST code, and the
full-scale simulation outputs are first converted to model scale, then fed to the actuators
for the cable tensions. The results in Figure 3.7 show that the presence of wind only
slightly affects pitch motion (17).

Figure 3.7. Platform motion spectra outputs of experiment under the wind-wave,only-
wave,and only-wind conditions : Hs = 3.6m, Tp = 10.2s, V = 11m/s,
(17)

33
 Connecting actuators with many cables is a complex and challenging system to
implement. In addition, since the ducted fan and drag disc cannot cover rotational and
stall effects, they cannot fully represent the prototype rotor. Therefore, Meseguer and Raul
(68) proposed a multi-fan system consisting of 6 motors to provide a rapid aerodynamic
response to floating dynamic motion and rotational effects. While four fans only work in
the wind direction to produce thrust, the other two fans create roll moment to mimic the
effects of torque. The same SIL scheme in the ReaTHM is used in this study to feed the
multi-fan actuators. Tested with a 1/40 scale NREL 5 MW turbine, the system catch the
desired aerodynamic forces with a 1.5% deviation and show almost the same performance
as the full-scale prototype.

(b) Scaled model test

(a) Scaled model of the WindFloat model with the ducted fan at
(15) ECN (16)

(d) Scaled model with the

actuator-cable system for
(c) The multi-fan actuator system (68) ReaTHM test (17)

Figure 3.8. Various concepts used in scaled model tests

34
 Aerodynamic and mechatronic designs are considered for a 1/75 scaled model of
DTU 10 MW at Politecnico di Milano in the study of Bayati et al. (69). Due to the
lack of a wave basin facility that creates gravity-based wave loads, length and velocity
scale factors are considered separately instead of Froude-scaling. Applying a higher ve-
locity scale factor (such as 2,3) with an individual pitch control system reduces Reynolds
dissimilitude. The results show good aggregation between the desired and the measured
aerodynamic forces. A 6-DOFs HexaFloat robot is developed in the following work to
simulate aerodynamic performance under floating motion (18). The tests are carried out in
the atmospheric boundary layer wind tunnel. With SIL, measured data or hydrodynamic
simulation data or hydrodynamic simulation based on measured data is given as input to
the system, resulting in global motion output being physically reflected by the robot.

Figure 3.9. HexaFloat robot developed at Politecnico di Milano (18)

35
 CHAPTER 4

EXPERIMENTAL SETUP OF IZTECH FOR SPAR-TYPE

FOWT

The generation of a high-quality wind and wave environment in a model flume

experiment is crucial for the correct coupling between the aerodynamic and hydrodynamic
forces on the floating wind turbines. In this study, aerodynamic and hydrodynamic design
are considered for a 1/40 Froude-scaled model of Northel Poyra P36/300 turbine with
spar-type platform. An experiment campaign is used to investigate the dynamic behavior
of the spar-type floating offshore wind turbine (FOWT) under various wave loads and
extreme wind condition. The spar platform’s, wind nozzle characteristics and effect of
the wind on the FOWT motion have been investigated. Experimental study are carried
out in the wave flume in the hydraulic laboratory of IYTE Civil Engineering Department.
A sketch of the flume’s is provided in Figure 4.1.

Figure 4.1. Top-sketch view of the test setup (19)

36
4.1. Model Setup

Physical and numerical models are used to investigate the stability and aero-
hydrodynamic responses of floating offshore wind turbines. There are many different
numerical tools on this subject and it is stated in the literature that they give very good
results (70). However, since numerical models are created by considering certain assump-
tions and simplifications, they must be validated with physical model experiments. On
the other hand, the field experiment with very large dimensions of the systems is usually
financially prohibited. Duration of the tests, uncontrollable and unrepeatable environmen-
tal conditions, and the product rights of full-scale commercial prototype are some of the
constraints of the field experiments (71). For these reasons, one of the preferred methods
in experimental studies is scaling in a controlled environment. It can still be seen as an
fundamental validation step in the system behavior.

4.1.1. Wave Flume

Dimensions of the wave flume of IZTECH Hydrodynamics Laboratory is 40 m

(length) × 1 m (width) × 1.4 m (depth). The wave flume was constructed from steel
except middle section of one side, which is 8-m-long thick glass. The glass section helps
to observe the dynamics of the FOWT during the experiment. A piston type wave-maker
with a 5kW servo motor drives the wave maker which is aimed to generate sea states
with irregular (variable period and wave height) waves that observed in nature as well as
regular (sinusoidal, constant height and period) waves to be used in basic experiments. A
dissipation beach, which is made of rocks of various sizes with a 1/5 gradient, is placed at
the opposite end to absorb the waves that otherwise reflect backward, as shown in Figure
4.2. Moreover, the waves propagate to the back of the piston, creating splashes and the
risk of damaging the motor when the generated and reflected waves interfere. For this
reason, a steel cage filled with absorbent material (plastic sponge) were placed to damp
the waves in this area.

37
 Figure 4.2. The dissipating beach with 1/5 gradient placed at the end of the channel

4.1.2. Wind Nozzle

The constraint that must be complied with in the design of the wind nozzle is
the width of the wave flume which is equal to 1 meter. The wind nozzle was designed
with intentions of maintaining air flow for the wind turbine considering its motion un-
der the waves (as explained in section 2.1). In this study, the length scale factor was
chosen as 40, therefore the velocity scale factor must be around 6.5 according to Froude
scaling. Since the primary goal is to accurately capture motion of the turbine under the
extreme conditions, the range of 15 − 25m/s -which is close to cut-out wind speed- are
mainly examined. Accordingly, the wind speeds generated by wind nozzle are between
2.3 − 3.8m/s. Analytical studies often rely on potential flow theory for the design of
contraction nozzle curves. On the other hand, numerical simulation gives more realis-
tic results because of modelling the boundary layer. Moreover, in contraction regions of
small wind tunnels, it has been seen that 5th or 4th order polynomials give very good
results (72). For this reason, 4th order polynomials were used in the design of the nozzle,
considering the simplicity of fabrication process.
To reduce turbulence intensity and obtain uniform flow, the contraction nozzle is
commonly used in wind tunnel applications. Turbulence intensity, the simplest form of
turbulence measurement, is obtained by dividing the standard deviation of the wind speed

38
by the mean wind speed.

Figure 4.3. Exploded view of screens, and a contracting nozzle

Two different designs of the nozzle ( as shown in Figure 4.4) were manufactured
with the purpose of creating model wind flow. The first nozzle was manufactured in 2018
as part of the M.Sc. thesis of Serkan EROL (73), Figure 4.3. The nozzle is 3.6 m long
and has a cross-sectional inlet area of 1.8x1.8m2 and outlet area of 0.9x0.9m2 (0.25 con-
traction ratio). Honeycomb and perforated were placed between fans and inlet section of
the nozzle to obtain low turbulence flow by decomposing large eddies. To characterize its
real performance, someintensive experimental measurements have been conducted using
hot wire anemometer and pitot tube with U-tube micro manometer. The results shows
that the range of generated wind speed is not enough to meet the expectations when heave
motion of the FOWT -due to the wave- is taken into account.

(a) First contraction nozzle design (b) Second contraction nozzle design

Figure 4.4. Nozzle Designs

39
 Therefore, a new nozzle was designed by switching length, outlet area based on
outcome of the experiment. In addition, the contraction curve was re-defined with new
dimensions. A contraction ratio of 1/2 has been chosen, resulting in an outlet area of
1.2x1.4 m2 and 1.8 m long, Figure 4.4b.

4.1.3. Traverse Mechanism

Positioning precision is only required along height of the wave flume to acquire
air flow properties accurately with measurement sensors. The translational movement
are provided automatically by a stepper motor with precision of 0.1 mm and motor is
controlled using LabVIEW software. Schematic of the traverse mechanism shown in
Figure 4.5. On the other hand, to reduce the cost of the traverse mechanism, it is moved
along the wave flume direction(x) and in the horizontal (y) axis manually with sigma
profiles. 45x45x1700mm heavy sigma profiles called ”A” shown in Figure 4.5 are fixed
to the carrier frame through brackets. Along the wave flume axis, 45x45x1800mm heavy
sigma profiles named as ”C” were connected with the same 300mm long heavy sigma
profiles called ”B”. The connection bolts are tightened when positioning is done. In the
same way, the 45x180x500mm heavy sigma profile called ”D” was moved in a similar
way and manual positioning was made on the horizontal axis of the wave flume.

Figure 4.5. Traverse Mechanism

40
4.1.4. Model Turbine

The wind turbine is a 1/40 scaled version of Northel Poyra P36/300 reference wind
turbine and due to the Reynolds dissimilitude and dynamic scaling issues, the blades of
the wind turbine were redesigned to preserve drag coefficient and obtain max torque.
All details about the model turbine design and manufacturing are given in Serkan Erol’s
master thesis (73). Some important information from the reference above is summarized
below for completeness.
Although the study is novel in terms of turbine nominal power and size, this nov-
elty brings important constraints to fabrication of the components and instrumentation. It
is extremely difficult to make a mold for a blade of this size since the thickness becomes
lower than 1 mm at the tip region of blade. A hollow and light blade structure is required
to satisfy desired moment of inertia. For these reasons, 3D printing method was chosen
to produce the blades.
Low Reynolds number drastically reduces the aerodynamics performance of blades,
thus, the blades were redesigned to match overall thrust coefficient. Due to the open jet
flow generated by the nozzle on the wave flume, the unsteady aerodynamic behavior of
the scaled rotor is not the main interest of the experiment unlike the wind tunnel tests.
When the scaled thrust is obtained under the Froude scaled wind conditions and opera-
tion, the rotor is said to perform at model scale which is called performance-scaled rotors.
Furthermore, motor was employed to reach desired rotational speed due to inadequate
torque generated by rotor. The motor is controlled instantaneously by a PID controller.

Table 4.1. Operational parameters of the turbine (λ = 1/40)

Parameters Reference Scaling Factor Model

Wind Speed 20 m/s λ0.5 3.16 m/s
Rotor Speed 50 rpm λ−0.5 316.22 rpm
Power 320 kW λ3.5 0.79 W
Power Coefficient 0.064 1 0.064
Thrust 18.9 kN λ3 0.296 N
Thrust Coefficient 0.076 1 0.076

Both Computational Fluid Dynamic (CFD) and Blade Element Momentum The-

41
ory (BEMT) simulations were performed for the specified operational conditions given in
Table 4.1, during the rotor design stage, to select best design in terms of performance. Re-
sults shows that Ishii airfoil (Figure 4.6) with a given configuration in Figure 4.7, creates
thrust force of 0.303 N, 0.32 N in BEMT and CFD respectively while the scaled thrust is
0.296 N. These values considered to be close enough for our experimental purposes.

Ishii Airfoil
0.4
Suction Side
Pressure Side
0.3

0.2

0.1
y/c [-]

-0.1

-0.2

-0.3

-0.4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x/c [-]

Figure 4.6. The airfoil section used in the model study

0.8
t/c [-]

0.6

0.4

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

0.08
Chord [m]

0.06
0.04
0.02
0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45

40
Twist [Degree]

30

20

10
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
r [m]

Figure 4.7. Geometric model of blade design parameters

Center of gravity (CoG) of the reference turbine at 26.24 meters high on tower. To
model the dynamic behavior of the reference turbine, model turbine must have the same

42
scaled mass and center of gravity (0.65 m) as well. Therefore, each component have to
be designed respecting the mass reduction of the scaling. Table 4.2 shows that, although
total mass was obtained correctly, desired individual masses were not achieved for each
component.

Table 4.2. Mass values of model turbine components

Components Reference [kg] Desired Mass[gr] Manufactured Mass[gr]
3x Blade mass 5322 83.15 112.52
Tower mass 34900.7 545.32 495.65
Hub mass 4420 69.06 105.47
Nacelle mass 21796 340.56 321.76
Total mass 66438.7 1038.10 1035.4

4.1.5. Model Spar-Type Floating Platform

The spar-type floating platform, which is well-known ballast-stability scheme in

the literature, was chosen since it is a suitable type of platform in terms of geometry for
using in wave flume of IZTECH and also there are plenty of data in the open literature.
There are two forces which drives the spar platform stability in water during its static
state. One of them is downward gravity forces (weight of system) acting at center of
gravity and the other one is buoyancy force which is equal to weight of water displaced by
immersed volume of spar, acting at centroid of the immersed volume. Rotational motion
changes the center of volume, and buoyancy force act as a restoring force to return the
floating platform it’s first equilibrium position. This only happens if metacentric height is
positive, which is defined as the distance between metacenter and the center of gravity, and
by the following equation (4.1), and shown schematically in the Figure 4.8 . Otherwise,
metacentric height provoke unstable condition.

GM = M − G (4.1)

43
 Figure 4.8. Illustration of metacenter and metacentric height (20)

A parametric analysis was performed considering mass, inertia, mooring lines,

hydrodynamic and aerodynamic forces on the reference turbine. Since water depth is 1
meter in the setup, numerical studies were done with 40 meters depth and scaled down
for 1/40 Froude-scaled model. The dimensions of the scaled model are shown in Table
4.4 and 4.3, with parameters given in Figure 4.9.
In the production process, to satisfy the dynamic similitude, stainless steel ma-
terial was used to fabricate mooring lines and the shell (thickness of 2 mm) of the spar
platform. A cap on the spar platform was designed for adjustment of the ballast weight.
The balance between buoyancy and weight must be accurately matched to preserve plat-
form stability (71). There is a material difference between practice and theory, which
case a shift in the CoG. Ballast of the model is adjusted by adding sand and pebble mix.
Density calculations of the mixture were made and the necessary density was obtained
as 1.6gr/cm3 . According to this calculation, 5430 grams of mixture was placed in the
spar which creates a height of 10.4cm . However, in the calculations, it was seen that
more weight should be added to ensure the total weight and center of gravity met the
requirements, for this purpose, a 2 cm thick iron plate was mounted under the spar model.

44
Table 4.3. Structural and hydrostatic properties of spar platform
Parameter Value
Mass 17.565 kg
Draft 0.6336 m
The volume of displaced water 0.0176 m3
Roll inertia 1.919 kgm2
Pitch inertia 1.919 kgm2
Yaw inertia 0.04 kgm2
Center of mass below SWL 0.411 m
Center of buoyancy below SWL 0.363 m
Metacentric height GM 0.0481 m
Static buoyancy force 172.936 N
Heave hydrostatic stiffness 31.068 N/m
Roll hydrostatic stiffness 0.145 N.m/◦
Pitch hydrostatic stiffness 0.145 N.m/◦

Figure 4.9. Schematic view of spar platform (19)

45
 Table 4.4. Scaled parameters of spar platform
Parameter Value [mm]
Df 204
D2 204
D1 63.5
Hf 4
Hb 125
H3 504.5
H2 50
H1 75
H0 50
Hf 1 25
t 2

The weight and center of gravity information of all components in the FOWT
system which is employed in the experiment are given in Table 4.5.

Table 4.5. Weights and CoG of the FOWT

Parameter Weight [gr] CoG on Z-axis [mm]
Turbine + Turbine-Spar Connection Part 1243.5 617.37
3 x Ping pong ball 10.33 475.33
Empty spar 6484 -432.6
Ballast Mix (Sand+Pebble) 5430 -648.7
Iron plate (under the spar) 4088 -714.6
Mooring lines (attached the fairled) 310 -486
Total mass 17565 -491.13
Center of mass below SWL -416.13

After the weights measurement process, first the center of gravity of the empty
spar and the turbine were both numerically calculated and experimentally measured for
validation of the center of gravity. The measurement of the center of gravity was made by
hanging the empty spar and turbine model with the wire, parallel to the ground, as shown
in Figure 4.10, and location was recorded during balance state. The center of gravity is
262.5 mm below from the connection surface of turbine and spar which origin point of

46
the coordinate system. Secondly, the FOWT was put into the water, draft was measured.
The draft was 63 cm, the result is very close to the target value of 63.3 cm which given in
Table 4.3.

Figure 4.10. Measurement of center of gravity

According to the results, there are very small differences between design and
model, hence it is assumed that the model satisfies the requirements.

4.2. Instrumentation and The Tools of Experiments

An accurate measurement is an essential part of an experiment. It helps to quantify

conditions before and during the test. Characterization of wave maker and wind nozzle
were performed through measuring the wave height, wave period, outlet air velocity at
different positions from the nozzle, and turbulence intensity of the air profile before the
test. Moreover, wave probes were also used in wave measurement, a strain gauge for the
thrust measurement, Go-Pro Hero 5 camera with three ping pong balls, as seen in Figure
4.11, were used to capture the 2-dimensional motion of the FOWT during the test. In
this study, main interest are wind, thrust and motion thus, interested reader can find more
information in the study of Kadir (74). Following paragraphs will give information about

47
the calibration of sensors. The sensors are as follows: hot-wire, strain gauge and image
processing tools.

Figure 4.11. Ping-pong balls that placed on the tower

4.2.1. Hot-wire

The hot-wire anemometer was used to measure the velocity of the wind nozzle at
different distances and the velocity differences, such as wind shear, that occur around the

48
model wind turbine. Hot-Wire anemometer gives precise and fast results. In order to use
the hot wire anemometer, it is necessary to calibrate the sensor against a known veloc-
ity. The tools which were needed for the calibration: calibration nozzle, DC motor, DC
power supply, hot-wire probe, pitot tube, micro U-manometer, computer and anemometer
system.
The reference flow for the calibration is needed to be laminar steady flow. In order
to obtain such a flow, a nozzle as shown in Figure 4.12, is used. The grid and wire mesh
was employed to obtain a lower turbulent flow at the nozzle exit.

Figure 4.12. Calibration nozzle

The anemometer system is a constant temperature type, more information about

it can be found in the reference (75) . This voltage is measured with the help of a 16-bit
DAQ card and recorded on a computer.
The working principle of the hot-wire can be explained simply as the heat transfer
between the heated wire and the cold air flow around it. This heat transfer is a function
of the air velocity. A relation is defined between the air velocity and the electrical voltage
through this principle. The main purpose of the hot-wire electronic circuit is to provide a
controlled amount of electric current to the wire part (75).
The wind speed given to the anemometer was measured against a High Wycombe
L type pitot tube connected to the Dwyer Model M1430 Microtector U-manometer seen in
Figure 4.13. Before the calibration, the consistency of the pitot tube at very low (0−2m/s)
wind speeds was measured at outlet of the calibration nozzle. The height value ”h” of the

49
Dwyer manometer, which shows the pressure difference between the two outlets of the
pitot tube, was taken many times for the same condition, and the wind speed at the nozzle
outlet was calculated with the Bernoulli equation given at Eqn 4.2 .
1 1
p1 + ρair V12 = p2 + ρair V22 (4.2)
2 2

Figure 4.13. Dwyer Model M1430 Microtector U-manometer

When the static pressure is accepted as the 1st state and the stagnation pressure as
the 2nd state, it is simplified as the equation 4.3 and the velocity is found by calculating
the pressure difference with the height ”h” which is measured in the U-manometer. It has
been observed that the Pitot tube gives very high consistency results until the minimum
(0.72 m⁄s) wind speed.
!
2ΔP
ΔP = ρf luid ∗ g ∗ h =⇒ V2 = (4.3)
ρair
A calibration curve with high consistency was obtained between the voltage value
that changes according to the wind speed and measured manometer data, through 4th
order polynomial fit, Figure 4.15.

50
 Figure 4.14. Calibration process of hot-wire

V elocity(V ) = −0.0003559 ∗ V 4 + 0.002151 ∗ V 3 + 0.07404 ∗ V 2 + 0.8124 ∗ V + 4.196

(4.4)
Changing environmental conditions affect the sensitivity of the hot-wire anemome-
ter, and therefore it needs to be calibrated regularly.

2
Measured data
1 4th order polinomial fit

0
R-square: 0.9994
hot_voltage [Volt]

-1 RMSE: 0.03704

-2

-3

-4

-5

-6
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
hot_vel [m/s]

Figure 4.15. Calibration curve of hot-wire

51
4.2.2. Strain gauge

The measurement of the thrust on the turbine tower ca not be obtained directly,
but instead total dynamic force on the tower was obtained by measuring the strain at
the bottom part of the tower by means of a strain gauge. In this application, it is aimed to
measure all the forces on the turbine first and to find only the thrust force coming from the
wind. In this study, BF350-3AA / 1.5AA model strain gauge module, which consists of
1 gauge and 1 amplifier, were used. This module provides analog output voltage between
0-3.5 V. The gauges are placed on the tower at angle of 90 ◦ as shown in Figure 4.16.

Figure 4.16. Strain gauge sensor on the tower bottom

In order to read the resistance change in the strain gauges, these gauges were
first connected to the Wheatstone-bridge and the amplifier. The signals at the output of
the amplifier were sent directly to the computer via a 16 bit data acquisition card (PCI-
1710HGU) and record at 100 Hz.
Since the resistance change in the strain gauge is converted to voltage with the
whitstone-bridge, the device must be calibrated in order to understand the force that
causes the voltage change. The calibration measurement was made by recording the volt-
age values obtained by applying a known forces to the tower. A trendline was created
with the voltage values and a curve fitting equation was derived as seen in Figure 4.17.

52
 3.5
Measured Data
1st order linear fit
3

2.5

Force [N]
1.5
y = -23.77*x+77.96
R-square = 0.9741
1

0.5

-0.5
3.14 3.16 3.18 3.2 3.22 3.24 3.26 3.28 3.3
Average_Voltage

Figure 4.17. Calibration curve of Strain gauge

4.2.3. Image processing tool

Image processing technique was employed to analyze the motion of the floating
wind turbine model in wave and wind conditions over time. For this purpose, colored ping
pong balls were glued on the turbine tower and video recorded at 120 frame per second
with a high resolution camera (GoPro Hero 5) during the experiments. The processing of
recorded videos is carried out with an open source program called Tracker (76). Tracker
receives the video and process it as successive frames. For example, a 60-second video
with a 120Hz camera consists of 120x60=7200 frames. In the Tracker program, steps are
as follows:

• The first frame of the video is chosen as a reference and an origin point is defined
in this frame.

• A calibration line is drawn from the top reference point (red ball) to the bottom
reference point, and the length of this line is defined as the real distance between
the balls, 0.725 meter. The program calibrates itself by proportioning the length in
pixels of the calibration line in the first frame to the length in defined meters.

• Then, by following these balls in each frame of the video, the horizontal and verti-
cal (time-dependent) distances with respect to the origin point are recorded by the

53
 program.

The Tracker GUI shown in Figure 4.18. Here, the intersection of the purple lines
represents the origin, and the blue line represents the calibration line. The graph of ball’s
positions depend on time in the horizontal and vertical planes, as it is shown in the upper
right window, and the same values are shown as a table in the lower right window. The
processed position data is then decomposed into surge, pitch, and heave movements with
a written Matlab code which will explained following part.

Figure 4.18. View from Tracker program

The distance (1446 mm) of the center of gravity from Point 1 (red reference point
in the Figure 4.18) is used to separate translation from rotation. Point 1 is the point close
to the nacelle, Point 2 (white reference point in the Figure 4.18) is the point close to the
water level. The coordinates obtained as a result of image processing and the center of
gravity can be created virtually with the following equations.

YiP oint−1 − YiP oint−2

Anglei = atan( ) (4.5)
XiP oint−1 − XiP oint−2

Distance = 1446mm

i = Xi + (Distance ∗ cos(Anglei ))
P oint−1
CoGX (4.6)

54
 CoGYi = YiP oint−1 + (Distance ∗ sin(Anglei )) (4.7)

Since the center of gravity is also the center of rotation. If CoG motion is cal-
culated and subtracted from entire motion, the difference will only give the rotational
motion. When each time step is specified in i:

Surgei = CoGX
i+1 − CoGi
X
(4.8)

Heavei = CoGYi+1 − CoGYi (4.9)

P oint−1
Yi+1 − Heavei − YiP oint−1
P itchi = atan( ) (4.10)
P oint−1
Xi+1 − Surgei − XiP oint−1
Summation of all these instantaneous surge and heave motions will give the total
translational motion.

55
 CHAPTER 5

RESULTS

5.1. Atmospheric Boundary Layer Reproduction

The lowest layer of the troposphere, which is directly affected by the presence of
the Earth’s surface, is defined as the atmospheric boundary layer. The surface layer (a.k.a.
atmospheric boundary layer), where the forces occur due to the turbulent motions of the
air. The height of the layer is about 100 meters, and the wind speed increases rapidly with
height. The wind profile in the surface layer is approximated by the logarithmic equation
as follows (77):

u∗ z
u(z) = ln( ) (5.1)
κ z0
where z is the mean height above ground level, u is the time average wind speed

at that height , u∗ is the friction velocity (u∗ = τρ ) and κ is the Von-Karman constant,
which is taken as 0.40 in boundary layer meteorology. Due to the no-slip condition at-
tached to the wall, the wind speed in the surface boundary layer should decrease and
become zero as it approaches the surface. The z0 height ,where the wind speed is zero, is
defined as the roughness or aerodynamic roughness and its value is not equal to zero.

Table 5.1. Roughness values according to landscape type (22)

Landscape Type Water Snow Grass Farmland Forest City
Roughness [m] 2 ∗ 10−4 10−3 0.03 0.1 0.8 1.0

Over 2/3 of the earth’s surface consists of water surfaces such as oceans, seas and
lakes, and the marine boundary layer plays major role in small-scale air-sea interaction.
There are great differences between the soil and sea surface due to thermodynamic and
dynamic characteristics. Roughness of sea is calculated as 2 ∗ 10−4 meter as seen in Table

56
5.1. This is acceptable for long-term climate events. When calm and stormy sea states are
compared, it is seen that the roughness depends on the wind speed. It has been observed
that in severe stormy weather conditions over the open oceans, waves reach frightening
dimensions of 20-30 meters (78). For terrains where air interacts with sea, the Charnock
roughness equation is defined based on friction velocity (79):

u2∗
z0 = a (5.2)
g
where ”a” is the Charnock constant, and g is gravity acceleration. In the absence
of any field data regarding turbulence, IEC 61400-3 “Design requirements for offshore
wind turbines” standard recommends that the roughness and standard deviation of the
turbulence should be calculated by the Charnock equation with the following equations
(80):
 2
a κ ∗ Vhub
z0 = (5.3)
g ln( zhub
z0
)
The turbulence intensity above the sea is higher at low speeds and decreases with
increasing wind speed, reaching a minimum value at 8 m/s – 12 m/s and continues to
increase slowly for speeds higher than 12 m/s (81).
According to the operational conditions of the reference turbine (20m/s ) wind
speed and 1/40 scaling factor, the desired wind speed at the hub height of the model
turbine (36 meters) was calculated to be (3.16m/s). The hub height is 41 meters together
with the height above still water level of the floating platform.
For neutral atmospheric conditions, Wang et al. (82) presented a new model using
the Charnock constant based on the IEC standard. The study divides the sea state into
three groups in terms of friction velocity:

• u∗ < 0.11m/s =⇒ aerodynamically calm sea surface

• 0.11m/s < u∗ < 0.26m/s =⇒ transient sea state from calm to rough

• 0.26m/s < u∗ =⇒ completely rough sea state

In the study, the Charnock constant is set at 0.011 for calm sea, 0.016 for tran-
sient sea, and 0.072 for completely rough sea. Considering the extreme conditions, the
constant was taken 0.072. Equation 5.3 was solved implicitly with given inputs, and the

57
roughness was found to be 0.006 meter. According to Equation 5.2, friction velocity (u∗ )
was calculated as 0.9066m/s, and this value proves the completely rough sea state.
The relationship between the friction velocity and the longitudinal variance of the
wind speed was defined by Stull (83) as follows:

σu2 = 6.25 ∗ u2∗ (5.4)

Turbulence intensity is found correspondingly by dividing the longitudinal vari-

ance by the mean wind speed,

σu
TI = (5.5)
u
The longitudinal variance and turbulence intensity were found 2.266 m/s, 11.33% respec-
tively. Target Froude-scaled atmospheric boundary layer was obtained theoretically, can
be seen in Figure 5.1, for our experiment.

160

140

120
Height [cm]

100

80

60

40 Logaritmic wind profile

Upper Limit of Rotor
20 Lower Limit of Rotor
Rotor Hub Height

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Velocity [m/s]

Figure 5.1. Froude-scaled logarithmic wind profile

Performance Test of Wind Nozzle : In order to understand the wind character-

istics and to find optimum position for the FOWT, measurement tests were performed. D
= 120 cm characteristic length was taken from the nozzle exit in the wind direction axis.

58
Three different fan performances for each position in 4 different positions were consid-
ered and data collected with the pitot tube. The origin point moved from sea level of the
FOWT to sea level of the nozzle exit plane, and it is accepted as the origin point for this
test.
The nozzle was placed at height of 155 cm from the ground. It has a length of 120
cm in the y-axis (y = -60: +60) and a length of 140 cm in the z-axis (z =155:295). The
result of the measurements are given in the Figures below.

Figure 5.2. Measured wind profiles at x= 0.5D

59
Figure 5.3. Measured wind profiles at x= 1D

Figure 5.4. Measured wind profiles at x= 2D

60
 Figure 5.5. Measured wind profiles at x= 3D

As a result of the test, it was decided that the position at x=1D is the most suitable
position for the wind turbine system. In order to bring this profile closer to the atmo-
spheric boundary layer profile, different tests were considered as well.
A separate experiment was conducted to observe how homogeneously the flow on
the turbine is distributed when looking in the direction of flow from the wind turbine. The
Figures given below help to understand how homogeneous the flow is, especially in the
Y axis. The experiment was carried out with the upper fans at 80% and the lower fans at
100% performance.

61
 Figure 5.6. Measured wind points on y-z plane at x= 1D

In Figure 5.6, the coordinates of the points where measurements were taken from
120 cm (1D) away from the x-axis of the nozzle exit area are given. In addition, the rotor
area projection on the nozzle exit plane is seen as circular. Wind speed measurements
were taken at 11 points (-50:+50) at 10 cm intervals on the y-axis at z=190,225,260 cm
heights. Measurements were taken 6 times in succession for each point.

Figure 5.7. Wind profile at z=190 cm (o = Each measurement, black solid line = Av-
erage of measurements)

62
Figure 5.8. Wind profile at z=225 cm (o = Each measurement, black solid line = Av-
erage of measurements)

Figure 5.9. Wind profile at z=260 cm (o = Each measurement, black solid line = Av-
erage of measurements)

63
 In Figure 5.7 and Figure 5.9, a decreasing trend in wind speed from y=+10cm to
+50 is seen. In Figure 5.8 there is an almost homogeneous flow, except for the higher-
than-average velocity profile between y=+10:+40. To solve this problem, three identical
wing are placed in the nozzle.
As a result of the optimization studies, the profile closest to the desired wind
profile was obtained when the lower fans and the upper fans were operated at 80.5% and
83% performance, respectively. The wind speed distribution in the rotor swept area is
shown in Figure 5.10.
Atmospheric boundary layer is very important in terms of observing dynamic
events such as overturning moment of the floating platform. The realistic case for the
wind turbine can be achieved by using ABL.

Figure 5.10. Experiment wind profile

The turbulence intensity was calculated, Figure 5.11, based on the hot-wire data
which has 500 Hz sampling rate and 120 seconds long data were collected for each point.
Figure 5.12 shows the difference between the wind profile that will create the rotor thrust
and the desired wind profile. The maximum wind speed difference over the rotor sweep
area was 0.213 m/s, the average wind speed difference was 0.03 m/s, and the average
turbulence intensity was 14.78%. Although the in-homogeneous wind difference affects
the moment of the system, as it can be understood from the average wind speed difference,
the total thrust force coming from the rotor was provided with high accuracy.

64
 Figure 5.11. Turbulence intensity of wind profile

Figure 5.12. Wind speed difference over rotor swept area

65
5.2. Dynamic Tests

In the experiments, it was planned to take the wave height from 2 m to 12 m in the
prototype, taking into account the wave height range that may occur in reality. To find the
wave period corresponding to these wave heights, the wave steepness ”s”, which gives the
relationship between the wave height and the wave period, was used. The wave steepness
is given in Equation 5.6:
H
s= (5.6)
gT 2 /2π
H is the wave height, T is the wave period, and g is the gravitational acceleration. As the
wave steepness increases, the wave period gets shorter. Two different wave steepnesses
of 0.02 and 0.04 were taken in the experiments. The steepness of 0.02 was used to see the
effect of long-period waves, and the more encountered periods in nature was represented
with 0.04 in the experiments. Increasing the wave steepness further was not considered
since it causes wave breaks in the channel. The wave parameters and value ranges used
in the experiments are given in Table 5.2.

Table 5.2. Full-scaled version of wave parameters of experiment

Parameter Range
Wave Height 2 - 12 m
Wave Steepness 0.02, 0.04
Wave Period 5.69 - 19.6 s

Regarding the wave parameters, tests were carried out in the wave channel for
wave-wind,only wind and only wave conditions. During the tests, the motion was cap-
tured by camera and post-processed through Tracker for the heave, pitch and surge degrees
of freedom. The motion was decomposed with in-house MATLAB code. Thrust data was
collected from the strain gauge which was attached to the bottom of tower. The results
are given in next subsections.

66
5.2.1. Free Decay Tests

Free decay tests are performed by giving a certain initial translational or rotational
displacements to the model in the wave flume and releasing the model. The model os-
cillates with gradually decreasing amplitude until it reaches the static equilibrium state.
This set of experiments is important in terms of providing information about the damp-
ing characteristics of the model. Within the scope of these experiments, free decay tests
with 3 degrees of freedom were applied (surge, pitch, and heave). Since the wave and
wind come from one direction, the movement in other degrees of freedom is neglected in
this study. The initial translational distances and rotational degree which applied in free
damping tests are given in Table 5.3. The results are given in Figure 5.13.

Table 5.3. Initial displacement and angular values which applied in free decay test
Degree of Freedom Value
Surge 23 cm
Pitch 14◦
Heave -36 cm

Figure 5.13. Free decay tests result in time domain

67
 Fast Fourier Transform analysis was performed to find the natural frequency in
each decay test for the individual degree of freedom. Since moored floating platforms are
highly coupled systems, it is impossible to measure only one mode. Hence it contains the
frequencies of other DoFs. Figure 5.14 shows single and coupled responses formed from
surge, heave and pitch free decay tests. In the middle subfigure, it is seen that the heave
response is coupled with pitch

X 0.0771208
6 Y 5.31758
Surgeplatform - FFT

0
0 0.05 0.1 0.15 0.2 X 0.207294
0.25 0.3 0.35 0.4 0.45 0.5
Y 1.49614
1.5
Heaveplatform - FFT

X 0.276392
0.5 Y 0.303086

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
X 0.273973
4 Y 3.52416
Pitchplatform - FFT

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Frequency

Figure 5.14. Free decay tests result in frequency domain

As a result of free decay tests, surge, heave, and pitch natural frequencies were
calculated as 0.077 Hz, 0.2 Hz, and 0.276 Hz, respectively.

5.2.2. Only Wind Tests

In floating platforms, the aerodynamic thrust acting on the rotor gives the average
additional pitch angle depending on the platform concept. Although this effect is dimin-
ished by the stability of the platform itself, secondly, the mooring lines tends to return to
their initial state, it can result in a significant additional average pitch angle (47). In order
to observe the dynamics created by these effects on the system, experiments were carried
out under the atmospheric boundary layer.

68
Figure 5.15. Only wind test, from left to right : 1. Initial position, 2. Position under the
wind, 3. Enhanced position by mooring lines, 4. Rotor starts to rotation, 5.
Transient state between starting of rotation and max rpm, 6. Position when
the rotor at max rpm.

The behavior of the system can be seen step by step during the experiments in
Figure 5.15. With occurrence of the wind, the drag force causes the surge. The logarith-
mic profile exerts more force on the upper half of the rotor,thus, the pitch angle and draft
of the system was increased. The longitudinally shifted platform tended to return to its
initial state by tension of mooring lines. The working moment of the rotor creates tran-
sient state, instead of extracting energy from the flow, it gives energy to the flow. Hence,
maximum surge and pitch angle were occurred. The rotor was accelerated gradually until
it reaches the desired rpm value, and steady state. In the final state, there were +3 cm
surge, -1.2 cm heave, +1.35◦ pitch angle differences from the initial position. The strain
gauge results shows that the thrust force values acting on the tower vary between -0.2 N
and 0.9 N. However,the average thrust force was found to be 0.29 N.

5.2.3. Wind and Regular Wave Tests

The wave parameters of model were determined by preliminary tests in regular

wave experiments. The wave heights and periods used in the experimental tests are given
in Table 5.4. The test are named according to the full-scaled value of the wave height and
the wave steepness. For example, the experimental condition of wave height of 4 m and
wave steepness of 0.02 is called D-H4-T02.

69
 Table 5.4. Parameters of regular wave of the tests
Test name Height [cm] Period [s] Steepness
D-H2-T02 4.88 1.22 0.2128
D-H4-T02 10.16 1.79 0.2058
D-H6-T02 15.66 2.19 0.2119
D-H8-T02 19.77 2.53 0.2004
D-H10-T02 24.98 2.82 0.2038
D-H12-T02 29.8 3.1 0.2012
D-H2-T04 5.17 0.9 0.4142
D-H4-T04 9.65 1.26 0.3945
D-H6-T04 14.82 1.55 0.4003
D-H8-T04 19.79 1.77 0.4099
D-H10-T04 24.92 2 0.4043
D-H12-T04 29.5 2.09 0.4383

The regular wave that propagate along the flume loses most of its energy by break-
ing at the dissipation beach, but as determined by the preliminary studies, about 12% is
reflected back from the end of channel. The waves which reflected from the end are di-
rected towards the end of the channel by reflecting again from the wave maker. This event
continues for each wave until the wave loses its energy completely, and leads to distorted
wave propagation. Therefore, duration time for regular wave experiments was determined
as about 60 seconds at most.
The measurements collected from the wave gauge sensor during regular wave ex-
periments are given in Figure 5.16, where it is seen that harmonic frequencies occur in
the wave flume at multiples of the incident wave frequency during the regular wave ex-
periment. Consequently, this situation appears as harmonic frequencies in the results of
regular wave experiments.
After turned on the fans of the nozzle, regular wave tests were repeated under the
extreme wind. The experiments were carried out by rotating the blades at 316 rpm. The
positions obtained by processing the video recordings were separated into surge, heave
and pitch movements. The results are given in Figures 5.17-5.28.

70
 Wave gauge measurement of H6-T02 wave 5
Power Spectral Density of H6-T02 wave
10 X100.451434
Y 864.218
X 0.902869
5 Y 12.5145

0 100
X 1.3787
Y 0.8146

PSD [m2/Hz]
-5

wave experiments
10-5

Wave Displacement [cm]

-10
0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
Time [sec] Hz
Wave gauge measurement of H8-T04 wave X 0.561243 Power Spectral Density of H8-T04 wave
105
Y 1545.9
10
X 1.12249
Y 8.58799
5
100
0 X 1.67153
Y 0.702301

PSD [m2/Hz]
-5

10-5

Wave Displacement [cm]

-10
0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [sec] Hz
Wave gauge measurement of H10-T02 wave X 0.353827 Power Spectral Density of H10-T02 wave
20 105
Y 2057.33 X 0.707654
Y 225.385

10
100 X 1.06148
Y 13.953
0
PSD [m2/Hz]

Wave Displacement [cm]

-10 10-5
0 10 20 30 40 50 60 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time [sec] Hz

Figure 5.16. Time series and PSD of measured data from wave gauge during various

71
Figure 5.17. Power spectral densities and time series of surge, heave, pitch motions of
D-H2-T02-Test

72
Figure 5.18. Power spectral densities and time series of surge, heave, pitch motions of
D-H4-T02-Test

73
Figure 5.19. Power spectral densities and time series of surge, heave, pitch motions of
D-H6-T02-Test

74
Figure 5.20. Power spectral densities and time series of surge, heave, pitch motions of
D-H8-T02-Test

75
Figure 5.21. Power spectral densities and time series of surge, heave, pitch motions of
D-H10-T02-Test

76
Figure 5.22. Power spectral densities and time series of surge, heave, pitch motions of
D-H12-T02-Test

77
Figure 5.23. Power spectral densities and time series of surge, heave, pitch motions of
D-H2-T04-Test

78
Figure 5.24. Power spectral densities and time series of surge, heave, pitch motions of
D-H4-T04-Test

79
Figure 5.25. Power spectral densities and time series of surge, heave, pitch motions of
D-H6-T04-Test

80
Figure 5.26. Power spectral densities and time series of surge, heave, pitch motions of
D-H8-T04-Test

81
Figure 5.27. Power spectral densities and time series of surge, heave, pitch motions of
D-H10-T04-Test

82
Figure 5.28. Power spectral densities and time series of surge, heave, pitch motions of
D-H12-T04-Test

83
5.2.4. Wind and Irregular Wave Tests

Irregular waves are synthesized by sea state spectrums such as JONSWAP and
Pierson-Moskowitz. Displacement values are obtained as time series by inverse Fourier
transform and sent as inputs to the wave maker within the scope of experiments. In the
literature, it has been suggested that these time series should be long enough to cover 1000
waves for sensible desired spectral properties. Moreover, the data acquisition frequency
of the wave-meter should be sufficient to measure at least 10 discrete data points of one
wave (84). For example, the test duration of irregular wave series ,which have average
period of 1 s, must be at least 1x1000=1000 s, and the sampling frequency of the wave-
meter must be at least 10Hz.
It has been observed that the wave generator stops the motor by touching the limit
switches during the production of some extreme waves in the series, which have high
wave heights. Considering these constraints, only one wave condition were performed
in the channel. The number of data has been chosen as 2n = 4096 to facilitate power
spectral density calculations to be used in the processing of wave data.
The preliminary studies of irregular waves were completed in the channel. The
obtained data were processed with written Python code. Both the time-dependent and the
frequency-dependent data were examined, and the frequency-dependent spectrum was
extracted and compared with the spectrum of the theoretical wave. The result shows that,
wave maker can achieves the desired irregular wave.
The D-H8-T04 test condition was tested under the only wave and wind + wave
conditions. The power spectral density of the motion are shown in Figure 5.29.

84
 Irregular-H8-T04 Response
Only Wave
Wind+Wave

T04-Test
m2 /Hz
100

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Hz
105
Only Wave
Wind+Wave

m2 /Hz
100

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Hz

Only Wave
Wind+Wave

100

Degree 2 /Hz
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Hz

Figure 5.29. Power spectral densities of surge, heave, pitch motions of Irregular-H8-

85
 CHAPTER 6

CONCLUSION, DISCUSSION AND FUTURE STUDIES

6.1. Conclusions

Within the scope of the thesis, the dynamic behavior of the turbine system in
extreme wind and various wave conditions was investigated by using the Froude-scaled
model of Northel Poyra P36/300 turbine and spar type floating platform. An accurate
geometric, dynamic and kinematic similarities are vital in the experiment. In this context,
the consistency of the pitot tube was measured at low wind speeds for wind nozzle charac-
terization. The hot-wire sensor has been calibrated with this pitot tube to investigate the
wind characteristics in more detail. Electrical and computer connection of the traverse
mechanism has been established and it has been made ready for hot-wire and pitot tube.
The atmospheric boundary layer profile that should come to the turbine was calculated
theoretically by considering the IEC standard. As a result of the measurements, it was de-
cided that the first nozzle was not suitable for the desired wind profile, and the new nozzle
was designed. After the new nozzle was produced, measurements were taken at various
distances and fan performances for the desired atmospheric boundary layer profile and the
most appropriate distance was selected. Optimization studies were carried out to achieve
the most accurate profile in 1D, which is the most suitable distance. While the lower fans
were operating at 80.5 percent and the upper fans at 83 percent, great accuracy has been
achieved with average of 0.03 m/s difference from the desired wind over the rotor sweep
area.
In order to achieve dynamic similarity, the weight and center of gravity of the each
components are important. The ballast of the spar platform was calculated theoretically
considering this situation, and as a result of the experimental studies, the sand-pebble
mixture was placed in the spar and the steel plate was placed under the spar, and the
model matched with design requirements.
Image processing method was chosen to collect system motion data. In the im-
age processing method, both the in-house code developed in python and the open source

86
Tracker program used, which is common program in both academy and industry. As a
result of the preliminary studies, it was seen that the results of the two programs matched
each other and the experiments were analyzed with Tracker, which is easier to use. For
the RGB algorithm of the Tracker, a black curtain was hung behind the turbine system
and 3 red ping-pong balls were attached to the turbine for detection of the ping-pong balls
in program. The total motion obtained as a result of the program was divided into degree
of freedoms with the in-house code written in Matlab, and the results were obtained.
Within the scope of dynamic test, free decay tests, only wind tests, only wave
tests and wind + wave tests were carried out. The damping characteristics of the displace-
ments and rotations in the model matched the theory and provided preliminary validation
in terms of dynamic behavior. Only wind tests were carried out to measure the thrust
force. The average thrust force was measured as 0.29 N, that providing high match with
the numerical analysis. In order to see the effect of the wind on the oscillation of the
floating platform, the system was first subjected to wave tests only, and then the same
wave conditions were repeated under the extreme wind.
A combination of 6 different wave heights and 2 different wave periods, 12 dif-
ferent regular waves and one irregular wave test were conducted. Studies show that hy-
drodynamic forces have dominant role on floating wind turbine oscillation. On the other
hand, aerodynamic forces changes only the amplitude of these oscillations. While the
wind increases the oscillation in the system at low wave heights, it is seen that the wind
dampens the oscillation as the wave height increases.

6.2. Discussion and Limitation of the Tests

Although using the wave flume instead of wave basin brings novelty, the wave
flume constrains the experiment. Using a scale factor greater than 40 greatly reduces the
weight of the scale model and limits both production and the sensors to be used. A turbine
model with a scale factor of less than 40 does not fit into the flume dimensions. Because
of these limitations, obtaining the model weight was taken as a priority. Although the
overall center of gravity matched, the weight distribution on a component basis did not
match due to the manufacturing issues. Since it is very difficult to place a sensor on the
turbine in terms of weight distribution, motion tracking was done with image processing.
Moreover, the strain gauge was placed at the bottom of the tower to measure the thrust

87
force. In terms of applicability, only one atmospheric boundary layer has been created.
The experimental tests cover the situations where the wind and the wave come in the same
direction.

6.3. Recommendation for Future Studies

In nature, wind and waves usually do not come from the same direction. There-
fore, expanding the experimental work on various wave directions can represent better
real-world scenarios. Analyzing the dynamic behavior of the turbine with a fixed sys-
tem consisting of more than one camera and accelerometers to be placed on the floating
platform will increase both the scope and accuracy of the study. In future studies, the
operation can be expanded under various operation conditions by increasing the number
of wind profiles. In real world, wind turbines actively change the pitch angle of the blades
according to the wind speed. Using active pitch sensors together with the controller on
the turbine will provide better aerodynamic loads that expected from the rotor during the
experiment.

88
 REFERENCES

[1] (2022). ”wind energy data”. https://www.irena.org/wind.

[2] Castro-Santos, L. and V. Diaz Casas (2015, 07). ”Floating Offshore Wind Farms and
Their Application in Galicia (NW Spain)”, pp. 455–465.

[3] (1993). ”Wind resources part I: The European wind climatology In A. D. Garrad, W.
Palz, S. Scheller (Eds.)”, Bedford. H.S. Stephens and Associates.

[4] GWEC (2022a). ” floating offshore wind – a global opportunity”.

[5] Sebastian, T. and M. Lackner (2012). ”development of a free vortex wake method code
for offshore floating wind turbines”. Renewable Energy 46, 269–275.

[6] Martin, H. (2011, 12). ”Development of a Scale Model Wind Turbine for Testing of
Offshore Floating Wind Turbine Systems”.

[7] Tran, T. and D.-H. Kim (2015, 07). ”the platform pitching motion of floating offshore
wind turbine: A preliminary unsteady aerodynamic analysis”. Journal of Wind Engi-
neering and Industrial Aerodynamics 142.

[8] Lomholt, A. K. (2019). ”numerical study in bhawc and validation against experimental
data of a model scaled 10 mw floating wind turbine with active pitch control using
the tetraspar floater”. Master’s thesis, Lyngby Denmark.

[9] Branlard, E. (2017, 01). ”Wind Turbine Aerodynamics and Vorticity-Based Methods”,
Volume 7.

[10] Matha, D., J. Cruz, M. Masciola, E. Bachynski-Polić, M. Atcheson, A. Goupee,

S. Gueydon, and A. Robertson (2016a, 08). ”Modelling of Floating Offshore Wind
Technologies”, pp. 133–240.

[11] Goupee, A., B. Koo, R. Kimball, K. Lambrakos, and H. Dagher (2012, 07). ”experi-
mental comparison of three floating wind turbine concepts”. Volume 136.

[12] Robertson, A., J. Jonkman, A. Goupee, A. Coulling, I. Prowell, J. Browning, M.

Masciola, and P. Molta (2013, 06). ”summary of conclusions and recommendations
drawn from the deepcwind scaled floating offshore wind system test campaign”.

[13] Kimball, R., A. Goupee, M. Fowler, E.-J. Ridder, and J. Helder (2014, 06). ”wind/wave
basin verification of a performance-matched scale-model wind turbine on a floating
offshore wind turbine platform”. Volume 9, pp. V09BT09A025.

89
[14] Pegalajar-Jurado, A., A. Hansen, R. Laugesen, R. Mikkelsen, M. Borg, T. Kim,
N. Heilskov, and H. Bredmose (2016, 09). ”experimental and numerical study of
a 10mw tlp wind turbine in waves and wind”. Journal of Physics: Conference Se-
ries 753, 092007.

[15] Roddier, D., C. Cermelli, A. Aubault, and A. Weinstein (2010, 05). ”windfloat: A
floating foundation for offshore wind turbines”. Journal of Renewable and Sustain-
able Energy 2.

[16] Azcona, J., F. Bouchotrouch, M. González, J. Garciandı́a, X. Munduate, F. Kelber-

lau, and T. Nygaard (2014, 06). ”aerodynamic thrust modelling in wave tank tests of
offshore floating wind turbines using a ducted fan”. Journal of Physics: Conference
Series 524, 012089.

[17] Bachynski-Polić, E., M. Thys, T. Sauder, V. Chabaud, and L. Sæther (2016, 06). ”real-
time hybrid model testing of a braceless semi-submersible wind turbine: Part ii —
experimental results”. pp. V006T09A040.

[18] Bayati, I., A. Facchinetti, A. Fontanella, H. Giberti, and M. Belloli (2018, 06). ”a wind
tunnel/hil setup for integrated tests of floating offshore wind turbines”. Journal of
Physics: Conference Series 1037, 052025.

[19] Alkarem, Y. (2020). ”numerical examination of floating offshore wind turbine and de-
velopment of an innovative floating platform design”. Master’s thesis, Izmir.

[20] Doerry and A. Doerry (2008, 02). ”ship dynamics for maritime isar imaging”.

[21] Company, H.-P. (1985). ”the fundamentals of signal analysis”.

[22] Landberg, L. (2015). ”Meteorology for Wind Energy: An Introduction. Wiley.

[23] GWEC (2021). ”global offshore wind report-2021”.

[24] IRENA (2022). ”renewable energy statistics 2022”. Available at www.irena.org/

Statistics/Download-Data.

[25] Bingöl, F. (2010, March). Complex Terrain and Wind Lidars. Ph. D. thesis.

[26] Unnewehr, J. F., E. Jalbout, C. Jung, D. Schindler, and A. Weidlich (2021). ”getting
more with less? why repowering onshore wind farms does not always lead to more
wind power generation – a german case study”. Renewable Energy 180, 245–257.

[27] ”Siemens Gamesa launches 5.8 MW wind turbine”.

https://www.rechargenews.com/wind/siemens-gamesa-launch\
-es-5-8mw-wind-turbine/2-1-579633. Accessed: 24th of August 2022.

90
[28] ”The next step in our modularisation journey”.
https://www.vestas.com/en/media/blog/technology/
the-next-step-in-our-modularisation-journey. Accessed: 24th of
August 2022.

[29] Europe, W. (February 2022). ”wind energy in europe - 2021 statistics and the outlook
for 2022-2026”.

[30] IEA (2022). ”a 10-point plan to reduce the european union’s reliance on
russian natural gas”. Available at https://www.iea.org/reports/
a-10-point-plan-to-reduce-the-european-unions-reliance\
--on-russian-natural-gas.

[31] Marijuán, A. R. (2017). ”offshore floating platforms : Analysis of a solution for motion
mitigation”.

[32] GWEC (2022b). ”global offshore wind report-2022”.

[33] Robertson, A. and J. Jonkman (2011, 01). ”loads analysis of several offshore floating
wind turbine concepts”. Proceedings of the International Offshore and Polar Engi-
neering Conference.

[34] Wiser, R., J. Rand, J. Seel, P. Beiter, E. Baker, E. Lantz, and P. Gilman (2021, 05). ”ex-
pert elicitation survey predicts 37% to 49% declines in wind energy costs by 2050”.
Nature Energy 2021.

[35] Jonkman, J. (2007, 01). ”dynamics modeling and loads analysis of an offshore floating
wind turbine”.

[36] Sebastian, T. (2012). The Aerodynamics and Near Wake of an Offshore Floating Hori-
zontal Axis Wind Turbine. Ph. D. thesis.

[37] (2013, 06). ”Variations in Ultimate Load Predictions for Floating Offshore Wind
Turbine Extreme Pitching Motions Applying Different Aerodynamic Methodolo-
gies”, Volume All Days of International Ocean and Polar Engineering Conference.
ISOPE-I-13-102.

[38] Sebastian, T. and M. Lackner (2013, 04). ”characterization of the unsteady aerodynam-
ics of offshore floating wind turbines”. Wind Energy 16, 339–352.

[39] Chen, P., J. Chen, and Z. Hu (2020, 10). ”review of experimental-numerical method-
ologies and challenges for floating offshore wind turbines”. Journal of Marine Science
and Application.

[40] Chakrabarti, S. K. (1994). ”Offshore Structure Modeling”. WORLD SCIENTIFIC.

91
[41] Sørensen, J. (2016). General Momentum Theory for Horizontal Axis Wind Turbines.
Ph. D. thesis.

[42] Betz, A. (1926). ”Wind-Energie Und Ihre Ausnutzung Durch Windmuhle”. Gottingen:
Vandenhoe.

[43] Glauert, H. (1935). ”Airplane Propellers”, Volume Vol. IV. New York.

[44] (2011). ”Aerodynamics of Horizontal Axis Wind Turbines”, Chapter 3, pp. 39–136.
John Wiley Sons, Ltd.

[45] Castro, I. (2017). ”design of a 10mw wind turbine rotor blade for testing of a scaled-
down floating offshore support structure”.

[46] Fowler, M., R. Kimball, D. Thomas, and A. Goupee (2013, 06). ”design and testing of
scale model wind turbines for use in wind/wave basin model tests of floating offshore
wind turbines”. Volume 8.

[47] Matha, D., J. Cruz, M. Masciola, E. Bachynski-Polić, M. Atcheson, A. Goupee,

S. Gueydon, and A. Robertson (2016b, 08). ”Modelling of Floating Offshore Wind
Technologies”, pp. 133–240.

[48] Sarpkaya, T. S. (2010). ”Wave Forces on Offshore Structures”. Cambridge University

Press.

[49] Jonkman, J. (2009, 07). ”dynamics of offshore floating wind turbines-model develop-
ment and verification”. Wind Energy 12, 459 – 492.

[50] Svendsen, I. and I. Jonsson (1976). Hydrodynamics of coastal regions. Den Private
ingeniørfond, Technical University of Denmark.

[51] Pedersen, P., C. Aage, and E. o. K. Danmarks Tekniske Universitet. Institut for
Mekanik (2001). ”Grundlæggende skibs- og offshoreteknik”. Danmarks Tekniske
Universitet, Institut for Makanik, energi og konstruktion, Maritim teknik.

[52] Gueydon, S., I. Bayati, and E. de Ridder (2020). ”discussion of solutions for basin
model tests of fowts in combined waves and wind”. Ocean Engineering 209, 107288.

[53] Skaare, B., T. Hanson, F. Nielsen, R. Yttervik, A. Hansen, K. Thomsen, T. Larsen,

B. Skaare@hydro, No, T. David, F. No, Gunnar, R. Com, A. No, and Melchior (2007,
01). ”integrated dynamic analysis of floating offshore wind turbines”. European Wind
Energy Conference and Exhibition.

[54] Goupee, A., B. Koo, K. Lambrakos, and R. Kimball (2012, 04). ”model tests for three
floating wind turbine concepts”. Offshore Technology Conference, Proceedings 3.

92
[55] JM, J., S. Butterfield, W. Musial, and G. Scott (2009, 01). ”definition of a 5mw
reference wind turbine for offshore system development”. National Renewable En-
ergy Laboratory (NREL).

[56] Jonkman, J. (2010, 01). ”definition of the floating system for phase iv of oc3”. NREL
technical report.

[57] Coulling, A., A. Goupee, A. Robertson, J. Jonkman, and H. Dagher (2013, 03). ”vali-
dation of a fast semi-submersible floating wind turbine numerical model with deep-
cwind test data”. Journal of Renewable and Sustainable Energy 5.

[58] Martin, H., R. Kimball, A. Viselli, and A. Goupee (2012, 07). ”methodology for
wind/wave basin testing of floating offshore wind turbines”. Volume 136.

[59] Ridder, E.-J., W. Otto, G. Zondervan, F. Huijs, and G. Vaz (2014, 06). ”development
of a scaled-down floating wind turbine for offshore basin testing”. Proceedings of the
International Conference on Offshore Mechanics and Arctic Engineering - OMAE 9.

[60] Le Boulluec, M., J. Ohana, A. Martin, and A. Houmard (2013, 06). ”tank testing of a
new concept of floating offshore wind turbine”. Volume 8.

[61] Lemmer, F, A. J. M. D. A. F. C. F. B. H. and P. Montinari (2014). ”innwind.eu d4.24:

Floating wind model tests ecole centrale de nantes 2014”. Technical report, Nantes.

[62] Robertson, A., J. Jonkman, M. Masciola, H. Song, A. Goupee, A. Coulling, and

C. Luan (2014, 9). ”definition of the semisubmersible floating system for phase ii
of oc4”.

[63] Borisade, F., C. Koch, F. Lemmer, P. W. Cheng, F. Campagnolo, and D. Matha (2018,
03). ”validation of innwind.eu scaled model tests of a semisubmersible floating wind
turbine”. International Journal of Offshore and Polar Engineering 28, 54–64.

[64] (2013). ” the dtu 10-mw reference wind turbine ”.

[65] Kumari Ramachandran, G. (2013). A Numerical Model for a Floating TLP Wind Tur-
bine. Ph. D. thesis.

[66] Roddier, D., C. Cermelli, and A. Weinstein (2009, 01). ”windfloat: A floating
foundation for offshore wind turbines—part i: Design basis and qualification
process”. Volume 4.

[67] Sauder, T., V. Chabaud, M. Thys, E. Bachynski-Polić, and L. Sæther (2016, 06). ”real-
time hybrid model testing of a braceless semi-submersible wind turbine: Part i — the
hybrid approach”. pp. V006T09A039.

[68] Urbán, A. M. and R. Guanche (2019). ”wind turbine aerodynamics scale-modeling for
93
 floating offshore wind platform testing”. Journal of Wind Engineering and Industrial
Aerodynamics 186, 49–57.

[69] Bayati, I., M. Belloli, L. Bernini, H. Giberti, and A. Zasso (2017, 04). ”scale model
technology for floating offshore wind turbines”. IET Renewable Power Genera-
tion 11.

[70] Jonkman, J. M. and M. L. Buhl, Jr (2005, 10). ”fast user’s guide - updated august
2005”.

[71] Canet, H., P. Bortolotti, and C. L. Bottasso (2021). ”on the scaling of wind turbine
rotors”. Wind Energy Science 6(3), 601–626.

[72] Mehta, R. and P. Bradshaw (1979). ”design rules for small low speed wind tunnels”.
The Aeronautical Journal (1968) 83(827), 443–453.

[73] Erol, S. (2020). ”scaled down modelling of a horizontal wind turbine for a floating
wind turbine research”. Master’s thesis, Izmir.

[74] Aktaş, K. (2020). ”wave generation and analysis in the laboratory wave channel to
conduct experiments on the numerically modeled spar type floating wind turbine”.
Master’s thesis, Izmir.

[75] Bruun, H. (2009, 07). ”hot-wire anemometry: Principles and signal analysis”. Mea-
surement Science and Technology 7.

[76] Brown, D. (2009). ”tracker video analysis and modelling tool”.

[77] Emeis, S. (2018, 04). ”Wind Energy Meteorology - Second Edition”.

[78] (1988). ”chapter 13 marine atmospheric boundary layer”. In S. P. Arya (Ed.), In-
troduction to Micrometeorology, Volume 42 of International Geophysics, pp.
197–222. Academic Press.

[79] Charnock, H. (1955). ”wind stress on a water surface”. Quarterly Journal of the Royal
Meteorological Society 81, 639–640.

[80] IEC (2019). ”iec 61400-3-1, design requirements of offshore wind turbines”.

[81] Türk, M. and S. Emeis (2010). ”the dependence of offshore turbulence intensity on
wind speed”. Journal of Wind Engineering and Industrial Aerodynamics 98(8), 466–
471.

[82] Wang, H., R. Barthelmie, S. Pryor, and H.-G. Kim (2014, 10). ”a new turbulence model
for offshore wind turbine standards”. Wind Energy 17.

94
[83] Stull, R. (1988). ”An Introduction to Boundary Layer Meteorology”. Atmospheric and
Oceanographic Sciences Library. Springer Netherlands.

[84] Goda, Y. (2010). ”Random Seas and Design of Maritime Structures” (3rd ed.).
WORLD SCIENTIFIC.

[85] ”jrk g2 18v27 usb motor controller with feedback”. https://www.pololu.com/

product/3148. Accessed: 16th of October 2022.

[86] ”20.4:1 metal gearmotor 25dx65l mm hp 12v with 48 cpr encoder”. https://www.
pololu.com/product/4843. Accessed: 16th of October 2022.

95
 APPENDIX A

SIGNAL ANALYSIS

In signal analysis, the variable can be examined in both the time and frequency
domains without data loss, as shown in Figure A.1. Time-dependent variation of the 2-
dimensional motions was obtained during the experiment in this study. In order to evaluate
this data in detail and detect hidden effects that may not be visible in the time series, the
data has also been translated into the frequency domain (21).

Surgeplatform - FFT
20
Surge [cm]

4
0
2

-20 0
0 20 40 60 80 100 0 0.1 0.2 0.3 0.4 0.5

20 1.5
Heaveplatform - FFT
Heave [cm]

0 1

-20 0.5

-40 0
0 20 40 60 80 100 120 0 0.1 0.2 0.3 0.4 0.5

20 4
Pitchplatform - FFT
Pitch [Degree]

0 2

-20 0
0 20 40 60 80 100 120 0 0.1 0.2 0.3 0.4 0.5
Time [Second] Frequency

Figure A.1. Frequency spectrum of data according to time series

Furthermore, the y-axis that show amplitude is scaled logarithmically to make

minor effects even more visible, which compresses large signal amplitudes and expands
small amplitudes. Figure A.2 shows same data in log and linear scale for comparison.

96
 (a) Linear amplitude scale (b) Logarithmic amplitude scale

Figure A.2. Two different amplitude scale for same data (21)

Issues must be considered when converting time series data to the frequency do-
main (21). The first of these is the ”aliasing” problem. When the sampling rate fs of our
sensor is not greater than twice the maximum frequency of our signal fmax , which occurs
during the experiment, undesired aliasing appears. Figure A.3 depicts the aliased signal,
and removing this signal from the spectrum is not possible with filtering since it is an
artifact of Fourier Transformation (21). This minimum sample requirement is known as
Nyquist criteria, and the formula is given below:

fs > 2.fmax (A.1)

Figure A.3. Condition of aliasing occurrence in the frequency domain (21)

As another precaution, leakage shall be considered. Leakage occurs in the spec-

trum if the combination of the edges (beginning, ending) of the time-recorded signal is not
periodic. Therefore, instead of single or multiple peaks in the spectrum, the signal ampli-
tude is spread into the entire spectrum range, as seen in Figure A.4. Various windowing

97
schemes are used to avoid leakage (21).

Figure A.4. The smeared amplitude in the entire frequency range which is known as
the leakage problem (21)

98
 APPENDIX B

CALIBRATION PROCESS OF HOT-WIRE ANEMOMETER

In order to calibrate the hot-wire, first of all, the hot-wire is connected to the
IFA300 cabinet in one of the first two channels, which is a constant temperature anemome-
ter channel in 8 channels; the schematic view is shown in Figure B.1.
Since the current BNC connecter board does not work, the signal is collected with
the Advantech PCI-1710 HCG DAQ card from the A/D converter board output of the
hot-wire channel.

Figure B.1. Schematic view of the IFA 300 Constant Temperature Anemometer System

1. Complete the connections of Hot-wire, IFA300 cabinet, and DAQ card, as depicted
in Figure B.1. Next, connect the GND and signal outputs from the A/D board to

99
 channels 60 (60th channel: AI GND) and 68 (68th channel: Analog Input) on the
DAQ card.

2. In the ”ThermalPRO” interface, open the IFA300 Communication tab shown in

Figure B.3. After selecting the single wire type, which is the hot-wire type used
in the experiment, from the probe types, measure the probe resistance from the
Measure box.

3. After making sure that the hot-wire and BNC connector are disconnected, connect
the closed connector to the free end of the BNC cable and measure the cable resis-
tance from the measure box.

4. Go to Calibration → Probe Data tab (Figure B.2) and enter probe type, no of IFA
channel, measured cable resistance and operate resistance, then set offset to 0 and
gain to 1.

5. After entering the information, create a dummy calibration file with the ”Save as”
button and go directly to the Acquisition-Probe tab instead of the ”Next screen”
button. Then, run hot-wire with the saved calibration file.

6. Open the block diagram created in the LabVIEW program to collect the data on the
DAQ card, control algorithm of the program is shown in Figure B.4. Next, calculate
the offset and gain values within the ± 10 Volt limits of the DAQ card by giving the
hot-wire the minimum and maximum velocities it encounters during the experiment
at the calibration nozzle.

7. After repeating steps 4 and 5 with the new offset and gain values, measure at least 14
different velocities, which are also measured externally with a pitot tube, between
the minimum and maximum velocities through LabVIEW.

8. The hot-wire can be used to measure the velocity of air after it has been calibrated
against the pitot tube with measurements from LabVIEW.

100
Figure B.2. Calibration - Probe Data screen of the IFA 300

Figure B.3. Communications screen of the IFA 300

101
Figure B.4. Block diagram of LabVIEW for the calibration

102
 APPENDIX C

DECOMPOSITION CODE OF COUPLED MOTION

clc;
clear;
close all;
%% Decomposition of coupled motion

%Arrange the readtable input based on the experiment output

Data = readtable('rotated_GOPR2817.csv');

dt=1/119.88; %Camera fps

x1=Data(:,2); %Point 1 x coordinate

y1=Data(:,3); %Point 1 y coordinate
x2=Data(:,6); %Point 2 x coordinate
y2=Data(:,7); %Point 2 y coordinate

x1=table2array(x1);
y1=table2array(y1);
x2=table2array(x2);
y2=table2array(y2);

len = length(x1);
dist = 1.40343; %Real distance between Point 1 and CoG

for i=1:len
%Calculation of distance between Point 1 and Point 2
Distance_point1_point2(i)=sqrt((x1(i)-x2(i))ˆ2...
+(y1(i)-y2(i))ˆ2);

%Calculation of angle between Point 1 and Point 2

Angle(i) = atan((y1(i)-y2(i))/(x1(i)-x2(i)))*(180/pi);

103
 if Angle(i)<=0
Angle(i)=Angle(i)+180;
end
Angle(i)=-Angle(i);
end

mean_distance =mean(Distance_point1_point2);

for i=1:len
%CoG can be calculated mathematically.
CoG_x(i)= x1(i) + (dist * cos(deg2rad(Angle(i))));
CoG_y(i)= y1(i) + (dist * sin(deg2rad(Angle(i))));
Center_of_gravity(i,:) = [CoG_x(i) CoG_y(i)];
end

for i=1:len-1
time(i)=dt*i;
surge(i) = CoG_x(i+1)-CoG_x(i);
heave(i) = CoG_y(i+1)-CoG_y(i);
pitch(i) = atan((y1(i+1)-heave(i)-y1(i))...
/(x1(i+1)-surge(i)-x1(i)))*(180/pi);
end

k =0;
u= 0;

for i=1:len-1
k=k+surge(i);
total_surge(i) = k; %Add all individual surge respectively.
u=u+heave(i);
total_heave(i) = u; %Add all individual heave respectively.
end

figure
subplot(1,3,1);
plot(time,total_surge,'k-','LineWidth',0.5)

104
ylim([-0.5 0.5]) %Limit adjustment can be made according to test
grid on;
title('Surge [m]')
subplot(1,3,2);
plot(time,total_heave,'k-','LineWidth',0.5)
ylim([-0.2 0.2]) %Limit adjustment can be made according to test
grid on;
title('Heave [m]')
subplot(1,3,3);
plot(time,pitch,'k-','LineWidth',0.5)
ylim([-10 10]) %Limit adjustment can be made according to test
grid on;
title('Pitch [Degree]')
suptitle('Irregular Wave Experiment')

105
 APPENDIX D

WIND TURBINE’S MOTOR

It is seen that it is impossible (with a very low torque coefficient) for the model
turbine to reach 316.22 rpm at a wind speed of 3.16 m/s on its own; thus, the rotor must
be driven by the motor to achieve the desired speed. Therefore, the motor and driver with
PID controller were used in this study, and Table D.1 shows the name of the instruments.

Table D.1. Details of instruments of wind turbine motor’s

Instrument Part Number Part Name
Motor Pololu #4843 20.4:1 Metal Gearmotor 25Dx65L mm HP 12
with 48 CPR Encoder
Driver Pololu #3148 Jrk G2 18v27 USB Motor Controller with Feedback

After the motor-driver-computer connections are completed, the driver’s recom-

mended voltage (24 Volt) is provided with a power supply. Details about instruments,
connections, and setup can be found on Pololu’s official webpage (85, 86). With the
”Pololu Jrk G2 Configuration Utility” program, which is downloaded to the computer,
the PID settings of the motor are tuned and driven with this interface.
Since direct rpm control cannot be done through the interface, the rpm value,
which is a function of the target value in the interface, is recorded based on the target
value, as shown in Table D.2. Rotor rotation direction may change depending on where
motor power terminals are connected. According to the last connection configuration,
Target (counterclockwise< 2048, clockwise> 2048) is raised above 2048.

106
 Table D.2. Preliminary study outputs of wind turbine motor’s
Target Measured rpm Duty Cycle Target Duty Cycle Target [%]
2051 66 67 11
2052 82 89 15
2053 99 102 17
2054 113 120 20
2055 132 141 24
2056 150 169 28
2057 163 184 31
2058 182 211 35
2059 200 236 39
2060 212 253 42
2061 232 283 47
2062 250 312 52
2063 261 341 57
2064 281 383 64
2065 298 400 67
2066 316 428 71

107
Figure D.1. The Pololu Jrk G2 Configuration Utility user interface

108