voltage control of direct-drive wind turbines for …el-khatib, ahmad jessen, nicholas wachenfeld...

A h m a d E l - K h a t i b , M a r k e l M e s e g u e r S a n M a r t i n , M o h a m e d A m e e n , N i c h o l a s J e s s e n ,

P r a s a n n a V e n k a t e s h A n a n t h a n

SEMESTER PROJECT VOLTAGE CONTROL OF DIRECT-DRIVE WIND TURBINES FOR

OFFSHORE WIND POWER CONNECTED TO DC GRIDS

31/05/2019

Title: Voltage control of direct-drive wind turbines for offshore wind power connected to DC grids. Theme: Dynamic Control of Offshore Electrical Systems Project Period: Fall 2019 Project Group: OES8-3-EN19 Participants: Ameen, Mohamed Ananthan, Prasanna Venkatesh El-Khatib, Ahmad Jessen, Nicholas Wachenfeld Meseguer San Martin, Markel Supervisors: Hajizadeh, Amin Soltani, Mohsen Pages: 74 Date of Completion: May 31, 2019

Abstract: The voltage control of a direct drive wind turbine with a PM synchronous generator is being analysed and discussed in this report which is the outcome of a semester project as a curriculum requirement aiming at application of the related study curriculum courses. This included building mathematical models of the scaled down wind turbine setup and the active rectifiers available at the power electronics laboratory in Aalborg University-Esbjerg, verification of the built models by running the physical setup and comparing output data, analysing potential control strategies and implementing a suitable one, running simulations, tuning of offset parameters and running experiments on the setup to examine the integrity of the built voltage control system. The report gives insight about the modelling of the PMSG, the active rectifier, and a couple of algorithms for the control of the system, as well as the simulations done for their verification. The implementation of the control algorithm in the set-up is also detailed. However, due to different problems with the sensors, the main goal was partially achieved.

TABLE OF CONTENTS

1. INTRODUCTION .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1. USE OF POWER ELECTRONICS IN OFFSHORE WIND ENERGY .. . . . . . . . . . . . 2

1.2. DELIMITATION OF THE PROJECT .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.3. FORMULATION OF THE PROBLEM .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4. GOALS AND STEPS OF THE PROJECT .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2. DEFINITION OF THE TOPOLOGY .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1. USED TOPOLOGY .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1.1. PMSG... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.1.2. Act ive Rect i f iers: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1.3. Ex ist ing l i terature about s imilar topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.2. DESCRIPTION OF THE SET -UP .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.2.1. VLT f requency conver ter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2.2. Induct ion motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2.2.3. Permanent magnet synchronus generator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.2.4. Contro l Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3. MODELLING OF THE SYSTEM .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.1. MATHEMATICAL EQUATIONS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.1.1. d-q reference f rame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.1.2. Equat ions in d-q f rame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.1.3. Per-unit model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.2. SIMULINK MODEL... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4. PREPARATION OF THE SYSTEM .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.1. ESTIMATION OF THE PARAMETERS OF THE PMSG .. . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.2. VALIDATION OF THE PMSG MODEL .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.3. ESTIMATION OF THE PARAMETERS OF THE ACTIVE RECTIFIER .. . . . . . 44

5. CONTROL .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5.1. CONTROL ALGORITHM .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

5.2 STATOR FLUX ORIENTED CONTROL ( SFOC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2.1. PI tuning on d irect ax is . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

5.2.2. PI tuning on quadrat ic ax is . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.3. ROTOR VELOCITY ESTIMATION .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

5.4. SIGNALS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6. ANALYSIS OF THE RESULTS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.1. IMPLEMENTATION OF THE CONTROL LOOP .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.2. RESULTS .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

7. CONCLUSION .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

REFERENCES .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

APPENDIX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

PREFACE

This repor t is re lated to a sc ient i f ic project done by Of fshore Energy Engineer ing

students of Aalborg Univers ity -Esbjerg in their second semester as a curr iculum

requirement . The project durat ion was almost four months ( f rom ear ly Feb,19 to la te

May,19) and i ts subject was about Vol tage Contro l of Direct -Dr ive W ind Turbines for

of fshore wind power connected to DC gr ids. The project main purpose is educat ional

where students wi l l need to apply the informat ion they were exposed to in re lated study

courses and conduct exper iments on a s caled down setup bui l t spec if ica l ly for their

project and p laced at the power e lec tronics laboratory of Aalborg Univers i ty -Esbjerg.

The s tudents part ic ipat ing in th is project would l ike to send their u lt imate apprec iat ion

and sincere regards to their supervisors- Associate Professors Mohsen Sol tani and

Ameen Haj izadeh- for al l the guidance they provided, the lab supervisor -Mr. Henry

Enevoldsen-for bui ld ing the setup, to lerat ing their inf in i te quest ions and doing the

necessary adjustments and repairs due t o their mistakes or improper handl ing of

avai lable faci l i t ies . Apprec iat ion and regards are also extended to their fe l low PHD

student-Mr.Amin Qureshi- for paying regular v is i ts and help ing wi th sett ing up Dspace.

Aalborg Univers ity Esbjerg, May 31, 2019

VOLTAGE CONTROL OF DIRECT-DRIVE WIND TURBINES FOR OFFSHORE WIND POWER CONNECTED TO DC GRIDS

2

1. INTRODUCTION

The project report conta ins seven d if ferent chapters . This introductory chapter g ives a

c loser look in the context of use of e lectr ica l devices in Of fshore wind farms, as wel l

develops the reasons and the goals for such a project . The goals of f this chapter are:

- Br ief ly expla in ing the chal lenges that techno logy faces when us ing power

e lec tronic converters in Of fshore W ind farms.

- In troduce the leading problem of the project .

- L ist the del im itat ions of the project .

- Expla in the d if ferent mi les tones and achievements expected to obta in through the

project .

1.1. USE OF POWER ELECTRONICS IN OFFSHORE WIND ENERGY

The generat ion of energy has changed drast ica l ly over the las t decades into more

susta inable and foss i l fuel - less systems. The main reasons are the g lobal warming

caused (main ly) by CO 2 emiss ions and the decrease in the avai labi l i t y of the non-

renewable resources. Therefore, the development of d if ferent power generat ion methods

is being developed us ing inexhaust ible natura l resources: wind, sun and hydropower [1-

3] .

Among the energy sources ment ioned before, the one that has increased i ts potent ial in

Europe and other W estern Countr ies is of fshore wind energy [4]. The lack of topographic

obstacles in the sea (making the wind more stable and abundant) , the fewer v isua l

impact (of fshore farms cannot normal ly be wi tnessed f rom the coast [5] ) and the

poss ib i l i t y of having enormous unused ocean surfaces avai lable for insta l l ing wind

turb ines ( i f the technology for implant ing f loat ing turb ines keeps on developing) makes

of fshore wind a very interest ing energy source for the future [6].


3

Figure 1 : Sher ingham Shoal wind farm (Source: energynews.es)

One of the most important issues regarding the power generat ion in of fshore wind

turb ines is the conversion of the energy and transmiss ion to onshore. Regarding the

lat ter , AC t ransmission is shown to be inef f ic ient f rom the of fshore farms for large

d istances . One of the main problems is the react ive power generated by the cable shunt

capac itance. Thus, the use of High Vol tage Direct Current (H VDC) l ines to send the

energy generated to land has notably increased [ 7] , thanks to the latest developments

and researches done on the f ie ld. The cost (main problem) and the losses in the

subaquatic wires have been drast ica l ly reduced, as well as increasing the feas ib i l i t y.

Nevertheless, as wind turb ines generate power in AC, i t is impor tant to perform a

convers ion of the e lect r ic i t y to DC. Due to the inherent character ist ics and l im itat ions in

of fshore environment , state-of - the-ar t conversion systems were developed and insta l led

wor ldwide. The land-based wind turb ines are not necessar i ly be the most -sui table ones

when compared to the of fshore in terms of the weight , the s ize, and the re l iabi l i t y [8] .

The weight and the s ize of these convers ion systems may af fect the des ign of the

turb ines or require the insta l lat ion of addit ional p latforms to conta in them, caus ing an

increase in the cost of the plant .

Addi t ional ly, depending on the type of generator, d if ferent convers ion conf igurat i ons are

used to increase the ef f ic iency of the process, inc luding d if ferent k inds of rect i f iers,

transformers, DC-DC conver ters and inver ters. An accurate use of these devices wi l l

lead to obtain ing better results f rom the harvest ing [5] .

Therefore, i t is capita l that p lays a major ro le , both in the p lanning and des ign of the

p lant and dur ing the product ion, to have the most sui table convers ion systems to


4

increase the ef f ic iency and reduce the cost of the farm. A lot of l i terature has bee n

wri t ten about th is topic. Some of the solut ions proposed [9 -10] are d iscussed in the

chapter 2.1.

Another point to be carefu l ly cons idered is the var iabi l i t y on the avai labi l i t y of the wind

energy, one of the character ist ics of the Renewable energy sources. W ind veloc ity is

not constant (nor the speed or the direc t ion) over t ime, and despi te that wind turbines

have been des igned for adapt ing to var iable speed [11] , i t is important to def ine a good

control system for the power conver ters , so the f luc tuat ions regarding the var iable

speed, as wel l as the change in the power product ion does not af fect the behaviour of

the power components. A couple of d if ferent a lgor i thms to maximize the behaviour of

the system are d iscussed and have been implemented in the project.

Some of the poss ib le a lgor i thms, based on the funct ions and the des ired object ive are

shown in F igure 2.

Figure 2 : Dif ferent ex ist ing contro l systems for energy converters based on the

funct ion ( [7]) .


5

1.2. DELIMITATION OF THE PROJECT

As the projects involves a considerable amount of test ing, to val idate the ideas proposed

in the report , i t is necessary to have the appropr iate equipment to do them. However , a

real of fshore wind turb ine is not access ib le for th is purpose .

Then, a set-up has been prepared in the power e lec tronic laboratory located in the

Aalborg Univers i ty Campus in Esbjerg. The set -up is descr ibed in fur ther deta i ls in

chapter 2.2. I t is composed of an induct ion machine, s imulat ing the wind turb ine,

at tached to a permanent magnet synchronous generator (PMSG) and connected to power

convers ion systems. The induct ion machine is connected to the three -phase gr id, which

emulates the wind.

Due to the character is t ics of the set -up, i t is not poss ib le to s imulate a l l the mechanic s

and aerodynamic of a real wind turb ine in the sea. The focus of the project is regarding

the e lectr ica l and e lectronics part . Therefore, the f luc tuat ions in the torque of the wind

turb ine, generated by the changes in the wind speed have not been cons idered. The use

of a VLT converter for feeding the induct ion motor l im its the var iat ion of the veloc i ty to

a g iven reference.

Fur thermore, the avai labi l i t y of the ment ioned devices narrows the focus of the

invest igat ion into the topologies involving PMSG connected to act ive rect i f iers. Another

k ind of conf igurat ion involv ing d if ferent generators (DFIG) or AC/DC convers ion s ystem

(d iode rect i f ier) is not analysed wi th the project .

1.3. FORMULATION OF THE PROBLEM

Taking in account f rom everyth ing explained above, the main goal of th is project is g iv ing

a suitable solut ion to the problem formulated below:

“Formulat ing a sui table contro l system and a lgor i thm for a “di rec t -dr ive” of fshore wind

turb ine using PMSG and an act ive rect i f ier to convert the e lect r ic i t y ; to increase i ts

ef f ic iency and reduce the power losses to the minimum .”

A deep research on previous ly publ ish ed papers and a sequence of exper iments, are

expla ined in deep in the point 1.4. , which has been carr ied out , so the best resul ts are

obtained.


6

1.4. GOALS AND STEPS OF THE PROJECT

The main object ive of the project is to test the poss ib i l i t y of implement i ng a contro l

a lgor i thm us ing power elec tronic converters, to obta in measurements of d if ferent

var iables ( torque, vol tage, cur rent) according to g iven references. For that purpose,

mathemat ical calculat ions and des igns have been done and compared wi th an exis t ing

set-up, descr ibed in chapter 2.2.

The Curr icu lum of the Master studies , in re lat ion to the Project of the 2 n d Semester of

Of fshore Energy Systems, refers to the learn ing requirements for the projec t regarding

the knowledge gained in the subjects taught in the ment ioned semester ; such as non-

l inear control system des ign, power e lec tronics convers ion systems and model l ing of the

e lectr ica l dynamics of of fshore energy systems.

The goals appl ied to th is project are:

- Learn about power e lectronic conver ters used in of fshore energy systems, model

their dynamics and design proper contro l systems ,

- Des igning non- l inear a lgor i thms for contro l l ing the converters and the PMSG ,

- Perform exper iments in the laboratory wi th of fshore energy systems to col lec t

data, implement a contro l ler and emulate the behaviour of a system in real

condit ions,

- Use d if ferent techniques for the detect ion of unknown parameters ,

- Programming the VLT f requency converter to generate the des ired output ,

- Implement contro l lers through Dspace.

In order to fu lf i l these learning object ives, the project has dif ferent steps, each one

re lated to one or more of the goals ment ioned above :

- Chapter two contains a br ief ing about the d if ferent t opologies referr ing to the

generator types and power convers ion systems used in wind turbines that can be

found in the l i terature. Addit ional ly, a descr ipt ion of the devices used in the set-

up for model l ing and control are given.

- Chapter three involves the mathemat ical equations descr ibing the dynamics of

the d if ferent devices that need to be model led, as wel l as their d ig ita l

representat ion in Simul ink .


7

- Chapter four is focused on the val idat ion of the set -up, based on the informat ion

obtained in the two previous chapters , for conf irming the mathemat ical models ,

performing d if ferent exper iments .

- Chapter f ive cons ist of the des ign of an appropr iate control method for the output

VD C and the torque of the generator . Dif ferent approaches are analysed and

compared to f ind the most sui table one. Equal ly, the required sensors for the

implementat ion are expla ined.

- Chapter s ix is based on the analys is of the resul ts obta ined af ter performing the

tes ts wi th the des igned contro l ler in the laboratory.


8

2. DEFINITION OF THE TOPOLOGY

The main goals of th is chapter are:

- Br ief ing and reviewing previous works in which s im i lar topologies of the ones to

be analysed were implemented,

- In troduc ing the devices used in the set-up bui l t in the lab which are going to be

model led and control led.

2.1. USED TOPOLOGY

2.1.1. PMSG

In the recent advances in the DC technology [12], researches have been conducted in

us ing DC col lect ion system on the gr ids ins tead of an AC col lect ion system to reduce

losses and cost [13]. A DC W ind turb ine system cons ists of a Permanent Magnet

Synchronous Generator (PMSG), a generator in which the exc itat ion is induced by the

exc itat ion of the permanent magnets instead of us ing an e lectromagnet in the generator.

The PMSG is a lso cal led as a Magneto.

The major d isadvantage of us ing the PMSG is that the a ir gap f lux ins ide the assembly

is ungoverned, so the voltage of the machine is not eas i ly regulated. Our project a ims

to regulate the voltage us ing act ive rect i f icat ion methods to synchronize wi th the gr id

upon customer requirements . In th is chapter, more wi l l be d iscussed on the dif ferent

W ind Energy Conversion Systems (W ECS) and their generator conf igurat ions ; the

advantages and l im itat ions in us ing the PMSG in the Direct Dr ive W ind Turbine System

(DDWTS), the var ious topologies with in the DDWTS, the power e lectronic convers ions

and their work ing methodologies.

As referred here [14] wi th the promis ing technology and rapid developm ent in the tota l

insta l led wind power capac ity g lobal ly, the WECS is des ired to be more robust and cost -

ef fect ive ; therefore, there is a necess i ty in compar ing the d if ferent aspects in the W ind

Energy Technology. The propi t ious Permanent Magnet Generator S ystem (PMGS) is

presented in deta i l . The need for upscal ing the wind turb ines grows wi th technology and

innovat ion s ince 2004 [15]. W ind Energy is the most cost -ef fect ive solut ion for the

generat ion of energy in the modern industry [16] . Large W ind Turbines lead to h igh net


9

power rat ing and low average Level ized Cost of Energy (LCOE) [17] . However , whi le

cons ider ing upgrading wind turb ines, the mass of the d if ferent components involved

inside the turb ine must a lso be cons idered, volume of the structure and the cost for the

manufactur ing, transportat ion, and COE. The future success of the wind industry might

strongly depend on their abi l i t y to comply wi th both the market expectat ions and the

requirements of the gr id ut i l i t y companies.

The W ind Turbine System can be c lass if ied into 3 types:

( I) Depending on the rotat ion-(a) F ixed Speed W ind Turbine system with a gear

box, (b) L imited Var iable Speed W ind Turbine System with a mult i -stage

gearbox ( i .e) a Doubly Fed Induct ion Generator (DFIG) , (c) Var iable Speed

W ind Turbine System which is a gear less WT system with a Direct Dr ive

Generator (DDG).

( I I ) Depending on the rotat ion mechanism and power convers ion techniques -

Part ia l Power Electronic Converters and Ful l Power Electronic Converters .

( I I I ) Depending on the dr ive train components-Coupled Gearbox and Direct Dr ive

W ind Turbines.

The ear ly wind turb ines had smal ler capac it ies and were equipped with constant speed

Induct ion Generators ( IG). However , th is ear ly concept wi tnessed i ts reject ion due to i ts

pass ive output [18] . Therefore, a change in the turb ine des ign was introduced wi th the

novel concept of Pole Changing Generators wi th Act ive Sta l l Contro l, which led to the

poss ib i l i t y of a l lowing the turb ine to operate at reduced speeds [19] . However, th is

technology d id not f lour ish in the W ind Industry, as i t could damage the generator by

permitt ing h igh wind speed to spread on the dr ive tra in causing v ibrat ions and harmonics

that could be fata l . Later , the Induct ion Generators were equipped wi th a wound rotor

and were current contro l led. However, even the Induct ion Generators d id not las t long

in the ear ly developments of the wind industry even though i t of fered a wide range of

Var iable Speed Operat ions a long wi th Dynamic Load Reduct ion unt i l exc luded the scope

of advancement in the Power Contro l and pol luted the Grid. Subsequent ly, the IG of fered

a s imple, robust , re l iable and economical so lut ions to the W ind Industry, but wi th

compl icat ions in the avai labi l i t y of work ing under low speeds, and heavier mass due to

the presence of the gearbox and other bear ings in the Nacel le.


10

In recent years, PMSG is very promising as manufacturers are moving away f rom Geared

to Gear less Dr ive tra ins. Now, Gear less Direct Dr ive convers ion systems are avai lable

for Mul t i -mega W atts in W ind Turbines (Siemens Gamesa- SG 10.0-193 DD, Hal iade-150

6MW d’Als tom, Vestas V164 -8.0 de). The energy yie ld f rom the PMSG W ECS was found

to be more, and i t a lso of fered re l iabi l i t y in the Power Convers ion System [21]. I t a lso

por trayed some l im itat ions which were bas ical ly concerning the weight of the turb ine,

h igh cost, and found not to be economical ly v iable. The h igh cost of the PMSG is because

of the demand for Rare Earth Magnets (REM ).

Then, came the need for a Hybr id System for the convers ion and has be en introduced

to avoid the drawbacks of both d irect and indirect dr ive t ra ins by coupl ing the PMSG

with a Gearbox having a reduced transmiss ion rat io, conduct ing to h igh ef f ic iency and

l ighter mass.

Regarding the d if ferent k inds of generators , i n the DDW T, they are d irec t ly connected

to the rotor hub. Hence i t operates at low and var iable speeds. The Gear less WT has

many advantages, by removing the gearbox i t increases the ef f ic iency and reduces the

weight of the Magneto. Secondly, the re l iabi l i t y of a ge ar less turb ine is more when

compared to coupled dr ives. F inal ly, there are fewer moving par ts and wi thout the need

for a mechanical gear system both the noise and weight is reduced.

To increase the generator rotor rotat ing speed to gain a h igher power out put, a regular

geared dr ive wind turb ine typical ly uses a mult i -stage gearbox to take the rotat ional

speed f rom the low-speed shaf t of the b lade and transform it into a fast rotat ion on the

h igh-speed shaf t of the generator rotor. The advantages of a geare d generator system

include lower cost and smaller s ize and weight. However, the ut i l izat ion of a gearbox

can s ignif icant ly lower wind turb ine re l iabi l i t y and increase turb ine noise level and

mechanical losses [22].

Among the advantages of PMSG, i t is more re l iable compared to the generator wi th

e lec tr ica l exc i tat ion (current carrying coi ls due to the reduct ion of mechanical e lements

l ike s l ip r ings) . Also, their maintenance cost is fur ther reduced. By having h igh power to

weight rat io wi th low act ive components ( the e lements of the generator that par t ic ipates

in the product ion of torque) which could be expla ined by the fac t that the reduct ion on

the copper sect ion induces a smal l s lot space in the s tator and lessening the weight of

the machine. There is no copper loss present in the PMSG rotor s ince i t provides the


11

f lux dens ity wi thout the f low of current. Also, the c irculat ion of eddy currents in PM

induces losses much lower than copper losses. Thereby, increas ing the thermal

character is t ics. The e lectr ica l ly exc i ted synchronous generators EESG must have a p itch

pole large enough to a l low the arrangement of the exc i tat ion windings. This makes the

EESG heavier and large volume to develop the same f lux dens ity as in the PMSG. The

arrangement of NdFeB (Neodymium) magnets can be customized to make the generator

more f lex ib le.

There are d if ferent k inds of PMSG. Due to economic considerat ions, Radia l F lux

Permanent Magnet (RFPM) is the most common machine topology used for the d irec t

dr ive mul t i -megawatts wind turb ine [23] . I t is mature technology in terms of the

manufactur ing process. In contrast , PMSG machines wi th an outer rotor were used for

medium rat ing wind generators . Axia l F lux Permanent Magnet (AFPM) Generator is

where the machine produces the f lux in the ax ial di rec t ion. This des ign is widely used

in medium and smal l -scale wind turb ines. The machine has one stator and two PM rotors .

AFPM al lows low cogging torque wi th s imple winding and h igh torque volume rat io. Also,

i t has an act ive weight and an ax ia l length smaller than the radia l machine for the same

power rat ing. However , the AFPM has i ts drawbacks- Low torque to mass rat io, structura l

instabi l i t y, and dif f icu l t ies to mainta in the a i rgap for large d iameters. Transverse F lux

Permanent Magnet (TFPM): Produces f lux which is perpendicular to the d irec t ion of rotor

rotat ions. I t has many advantages such as h igh force dens i ty, low copper loss, and

simple winding. The l im itat ions in us ing TFPM is that the f lux path is 3 -Dimensional and

the construct ion is a lso found to be complex.

L imitat ions assoc iated wi th the weight and d iameter : Decreas ing the rotat ing veloc i ty at

a high-power level in the DDPMG wi l l induce a r ise on the torque. I t fo l lows that the

torque is proport ional to the a irgap d i ameter squared to the tangentia l force dens i ty.

Therefore, the DDWT have large d iameters wi th h igh tangent ia l force and a high number

of poles that wi l l result in heavy mass wi th cons iderable augmentat ion on the cost. A

large rotor is required to min imize the use of ironless PMGs which is l ighter than the

Iron core PMG. Because of the large s tructure, i t could become aerodynamical ly

inef f ic ient. To overcome th is problem of the weight, us ing mater ia ls such as a luminium

al loys and composites that may of fer l i ght but expensive s tructura l suppor t could be

cons idered. Fur thermore, by p lac ing a generator bear ing adjacent to the a irgap and

support ing them with ra i ls to reduce the st i f fness requirements of the rotor and stator to

reduce the mass reduct ion. A new concept of magnet bear ing, and mechanical bear ing


12

is being invest igated. In the same way l im itat ions to the pr ice of the PM with regards to

the amount of the Permanent Magnets required for the PMSG is propor t ional to the s ize

of the machine and inversely propor t ional to the speed. The Magnetos are more

at trac t ive than the IG and the EESG for large generators. However, to solve the REM

(Rare Earth Magnets) , researchers have developed a REM free generator f rom Ferr i te

Magnet. Despite being heavier with h igh iner t ia and inductance per phase, the COE for

both machines were a lmost the same with d if ferences in the losses.

The HTSG is equipped wi th the latest technology of wi thstanding h igh-temperature Super

Conduct ing wire, i t is a good way to avoid rare magnets and of fers high ef f ic iency and

reduced weight compared to convent ional generators . Semi -Direct Dr ive W ind Turbine-

The medium speed generators are the same as the Hybr id technology used in the WTS.

I t a ims to avoid the drawbacks of the geared and gear less DT by coupl ing the PMSG

with a gearbox wi th a reduced number of stages. The SDDWT was found to be h ighly

ef f ic ient , l ighter mass, but not widely used . Pseudo Direct Conversion System where the

magnet ic gears could be wor th cons ider ing the solut ion because the magnet ic

transmission presents h igh performance compared to the mechanical gear box.

Electr ical ly exc i ted synchronous generators (eesg) -The EESG is gain ing popular i ty

among the manufacturers and are used in large DDWT systems. The structure of fers

extra contro l of the rotor magnet ic f ield as the machine is e lectr ica l ly exc ited.

2.1.2. ACTIVE RECTIFIERS:

In modern wind turbine ins ta l lat ions, power is produced in the form of AC by the turb ines

themselves. The wind which produces the power is never constant and f luc tuat ions are

present in both voltage and f requency in the AC power output. This requires the power

produced by wind turb ines to be conver ted f rom AC into DC before being int roduced into

the gr id. In the case of th is project, the setup is on a much smal ler scale and the gr id is

replaced wi th a DC load as the f inal output but the theory behind i t is the same.

To put i t as s imple as poss ib le, a rect i f ier is a device that conver ts AC voltage into DC

voltage. Act ive rect i f iers enable a cross -funct ional technology in the convers ion process.

They are known for their inherent h igh switching abi l i t y by contro l l ing the harmonics and

other noises in the system. F igure 3 shows a standard act ive rect i f ier topology.


13

Figure 3 Simple Voltage Source Act ive Rect i f ier .

Act ive rect i f icat ion techniques are used to improve the performance character is t ics of

the convent ional three phase d iode rect i f iers [24] . In general , the rect i f iers are

comprised of mult iple components , with some rect i f iers us ing a transformer . A diode

cannot change i ts polar i ty, thus i f you feed i t f rom the negative s ide, i t wi l l not le t the

e lectr ic i t y pass, creat ing an open c ircuit . Di f ferent types of d iodes have been used in

the manufactur ing of rect i f iers such as sol id d iodes and vacuum tu be diodes. However,

current ly, s i l icon semiconductor rect i f iers are dominant over other types of rec t i f iers,

but th is is s trongly dependent on the s i tuat ion as in th is project a s imple d iode br idge is

used for the back -to-back conver ter . I t is to be noted that the e lectr ic device that

performs an oppos ite funct ion to that of the rect i f iers is cal led an Inver ter ( i t changes

DC current to AC current) . Dif ferent k ind of rect i f iers are shown in f igure 4.

Figure 4 : Types of rec t i f iers [ 26]

Rect i f iers can convert half or the whole per iod (wave) of the AC . As i ts name indicates,

half -wave rect i f ier processes half of the a lternat ing s ignal (e i ther the pos it ive or the

negat ive half ) whi le b lock ing the other half . Such rect i f icat ion process requires jus t one

d iode per phase. However , i t is inef f ic ient for h igh power appl icat ions, as s imply half of

the avai lable power is not being ut i l ized. F igure 5 shows a s ingle -phase s ingle diode

half -wave rect i f ier .


14

Figure 5 : Half -W ave Rect i f ier of a s ingle -phase AC current (source: [27])

To increase the ef f ic iency of power conversion, a ful l wave converter is used which

ut i l izes both s ides of an a lternat ing s ignal and transforms them into a s ingle polar i ty DC

signal. In such type of rect i f iers , more than one d iode p er phase is required (2 d iodes i f

the c ircuit ’s transformer is centre- tapped or 4 d iodes forming a “br idge conver ter” in

case the transformer is not centre-tapped). Examples of these two types are shown in

the f igure 6 (a and b) .

Figure 6 : 2 d iodes fu l l wave a) and br idge b) rec t i f ier of a s ingle-phase AC current

(source: [27])

In larger scale scenar ios or in specia l ized cases, other more compl icated methods can

be used to conver t AC to DC or to increase or decrease the input voltage. However , for

th is project, an act ive rect i f ier was used which is s l ight ly d if ferent f rom the fore

mentioned rect i f iers.


15

An act ive rect i f ier is a s l ight ly more advanced system where the diodes are replaced

wi th act ive ly contro l led swi tches such as trans is tors or, in the case of th is project,

MOSFETs. This is done to increase the performance of the convent ional rec t i f iers which

are load dependent and to reduce the voltage drop that occurs wi th standard

semiconduct ing d iodes. Secondly, the cut in t he costs of act ive power swi tches wi l l lead

to more cost-ef fect ive solut ions. MOSFETs on the other hand, have a very low resis tance

when conduct ing which leads to a smaller voltage drop. A contro l system can be appl ied

to these MOSFETs which a l lows for the e lect r ical topology of the system to be modif ied

to f i t cer tain condit ions. This can lead to power fac tor correct ion, meaning there is a

reduct ion of current harmonics a l lowing for greater overal l sys tem ef f ic iency. In the wind

industry, where the input power is in a state of constant f lux , th is level of contro l

compared to a more pass ive system is much more preferable. In shor t, ac t ive rect i f iers

a l low for a level of contro l over the convers ion process to produce a smoother current

curve by the reduct ion of current harmonics and vol tage drops [27] .

2.1.3. EXISTING LITERATURE ABOUT SIMILAR TOPOLOGIES

A novel sensor - less MPPT Contro l ler for a High Ef f ic iency Microscale W ind Power

Generat ion System is based on the adapt ive contro l to improve the dynamic response

on a microscale W ind Power Generat ion System (W PGS) ; where a wind turb ine wi th

f ixed p itch angle, a Magneto, a d iode br idge rect i f ier , a DC conver ter , a bat tery module

and a DC load is used wi thout the use of mechanical sensors to reduce the costs and

increase rel iabi l i t y [28] . The bas ic pr inc iple is us ing a s ingle s tep AC -DC Converter

which replaces the tradit ional two s tage converters and introduces the MPPT for h igher

ef f ic iency of the turbines and to min imise the Tota l Harmonic Distor t ions (THD). On

improvis ing the operat ion of the converter, a Quasi Synchronous rect i f icat ion topology

is used to contro l the switching for reduc ing the conduct ion loss in the e lectronic

components .

Contro l techniques wi th system ef f ic iency comparison for micro wind turbines i s a s tudy

on the implementat ion of a sensor - less speed control ler for a micro wind turbine wi th

ver i f icat ion on a W ind Emulator Test Rig was performed along wi th a deta i led compar ison

of d if ferent contro l techniques to show the advantages of act ive rect i f i cat ion [29] . The

study is conducted on a micro wind turb ine due to i ts easy avai labi l i t y and cost

cons iderat ions. The designed contro l ler was tested us ing real wind data f rom the Br it ish

Data base at Antarc t ica. The main comparison was on a sensor - less contro l , act ive


16

rect i f ier , a DC-DC Conver ter Regulator wi th a turb ine that is connected to a d iode

rect i f ier . The former was found to be more ef f ic ient by provid ing h igh power coef f ic ients

over the f luctuat ing wind speeds a long wi th good performances at mediu m and low speed

winds.

The paper on Comparat ive s tudy of three phase PW M rect i f iers for W ind Energy

Convers ion provides an a l ternat ive for the use of PWM rect i f ier topologies that can meet

the power requirements on a PMSG known for i ts robustness and cost -ef fect ive dr ives

[30] . The proposed topology uses three b i -d irec t ional swi tches in combinat ion wi th a

three-phase d iode br idge for rec t i f icat ion ; that can a lso be used wi th e ither leading or

lagging power fac tors. This topology is then compared to the con ventional 6-switch PWM

Vol tage source converters by account ing the switching and conduct ion losses. In

addit ion to the compar ison, the common mode vol tage generated by the model was found

to be less than the conventional s ix s tage VSC.

The publ icat ion of the art ic le on Isolated AC-DC Conversion us ing a medium f requency

transformer for an Of fshore W ind turb ine DC Collec t ion Gr id g ives the ins ight of a DC

archi tec ture to synthesis e energy f rom the of fshore gr id and transpor tat ion of power

through High Vol tage DC to the onshore DC col lect ion gr id system [31]. The main

pr inc ip le is the rect i f icat ion and convers ion of the WTG output to medium f requency

fol lowed by transformer iso lat ion and a boost rect i f icat ion s tage. The boost rec t i f icat ion

is governed by emulat ing a res is t ive load for a Medium Vol tage Direct Current (MVDC).

This topology is the future of HVDC connect ions and for f luc tuat ing wind veloc it ies. This

paper a lso refers to the project that is being worked on as i t uses the same conver ter

topology to process the WTG output by us ing back to back three phase inverters to get

the des ired l ine f requenc ies 50Hz or 60 Hz. The vol tage belonging to the l ine f requency

is f i l tered and is fed to the three -phase transformer to obta in the required Medium

Vol tage Alternat ing Current (MVAC). Although the topology is s imple, the components

were found to be bulky and require mul t ip le f i l ters to remove the unwanted harmonics.

The main advantage of us ing a MVDC transmission system is to reduce the react ive

power induced by the AC within the Grid. This paper fur ther compares wi th topologies

where a high f requency t ransformer uses a three - level inverter and a fu l l br idge neutral

point c lamped for a DC-DC convers ion. Another such topology provides the architec ture

for WTG-HVDC connect ion us ing medium f requency generator.


17

Another repor t expla ins how the use of an act ive rect i f ier is becoming more and more

impor tant in the ever -changing energy market . W hile convent ional, pass ive rect i f iers are

useful and cost -ef fec t ive for smal l -scale generat ion, larger power generat ion schemes

(such as a wind farm) require a more contro l led approach [32]. The repor t out l ines how

convent ional three-phase rect i f ier systems lead to a large number of generated

harmonics , especia l ly for large scale g enerat ion, but that ac t ive rect i f icat ion can cur tai l

th is to a cer tain extent and improve power dens ity. Deta i ls regarding the contro l methods

for act ive rect i f iers are expanded upon as wel l . The repor t ’s f ind ings can be summarised

by the s tatement : “Act ive rect i f iers can provide a var iety of features such as s inusoidal

input current, obtain ing h igh-power factor , control l ing act ive and react ive power and

provid ing b id irect ional power f low.”

The contro l a lgor i thm for an act ive rect i f ier is a lso d iscussed in a d if ferent repor t [33].

In th is repor t, a contro l a lgor i thm for an act ive rect i f ier used for charging and d ischarging

batter ies is analysed. An act ive f i l ter is managed by PW M to enable greater harmonic

f i l ter ing. The benef i ts of us ing an act ive f i l ter a s wel l as the control a lgor i thm are

presented and compared to a more industr ia l standard model f rom Siemens. W hile th is

type of rec t i f ier is not used the same way as in th is project , the control methods

presented were re levant and informat ive .

The operat ion of a three-phase act ive rect i f ier under non- ideal condit ions are presented

in the report about a rather self -explanatory case [34]. A three-phase act ive rect i f ier is

operated under lab condi t ions and an analysis is made into how much ef fect certa in

detr imental operat ing condit ions have on the overal l performance of the device. These

ef fects include: long computat ion t ime, presence of acquis i t ion f i l ters , ac phase

imbalances, etc. These undes irable issues as wel l as the necessary contro l s trategies

to reduce their ef fects are analysed and presented. This repor t is understandably

re levant to th is project as the operat ing condit ions in which the lab setup is used, l ike

a l l non- industr ial projects, are less than ideal. A rudimentary contro l scheme, bas ic lab

equipment , a vibrat ing motor are but some of the most obvious examples. Fol lowing

some of the contro l scheme resul ts presented by the repor t may improve the performance

of th is project.

An actual case study for the design and evaluat ion of an act ive rect i f ier for a 4.1MW

Offshore wind turbine is presented in another work [35] . The ef f ic iency, power fac tor

performance, select ion of IGBT as wel l as calculat ion of losses in the power


18

semiconductors of the conver ter are targets of the invest igat ions presented in th is

repor t. Al l of these f indings are for HVDC large scale wind appl icat ion; in short i t could

be cons idered a scaled-up version of th is semester project. While the values used are

much h igher, the contro l s trategies and methods used are s t i l l re levant and cer tain

compar isons can be made. The hardware used is dif ferent but the sof tware (MatLab and

Simul ink) are the same, fur ther a iding in comparing certa in resul ts.

2.2. DESCRIPTION OF THE SET-UP

The set-up in which the exper imental tests have been carr ied out pretends to s imulate

the behaviour of a W ind Power System, s imi lar as the one showed in F igure 7.

Figure 7 : Energy transfer in a DD W ind Power System.

Figure 8 and 9 show the set-up and i ts schema re lated to Figure 7. In the set-up, the

wind is emulated us ing the 3 -phase gr id . The induct ion motor generates the mechanical

power that feeds the PMSG, which provides power to the act ive rect i f ie r . The funct ion

of the HVDC gr id is performed by a DC load connected in the output of the rect i f ier .


19

Figure 8 : Schema of the set -up

Figure 9 : Set-up bui l t in the laboratory for carrying the exper iments .

The dr ives used in the set -up are descr ibed below.


20

2.2.1. VLT FREQUENCY CONVERTER

Figure 10 shows the VLT FC 301 f requency conver ter . The VLT is a revers ib le elec tr ica l

device, used for veloc ity contro l and wi th the abi l i t y of behaving equal ly as an inverter

and a rect i f ier , depending on the feed and the convers ion object ive. I t has a range of

power up to 3 kW.

Figure 10 : VLT FC 301 Automat ion Dr ive (source: Danfoss) .

Two VLT form the set -up. The goal of the f irst one, fed f rom the gr id, is applying a

constant veloci ty to the PMSG. The power supplied by the gr id is constant, unless

f luc tuat ions are manually in troduced, so the lack of ve loci ty contro l would suppose the

introduct ion of a l l the power supply f rom the gr id di rec t ly to the set -up. Regulat ing and

adjust ing th is value is the goal of the f irst VLT. The converter has a c losed- loop system

for determining the rotat ion veloc ity of the motor.

Addi t ional ly, the exis tence of a f requency conver ter between the gr id and the induct ion

motor wi l l equal ly a l low a sof t s tart ing of the dr ive and set -up, not a l lowing the current


21

peaks formed when s tart ing the motor . Us ing a pre insta l led manua l control , i t is possib le

to modify the input power , to modify the rotat ion or torque.

The second VLT rect i f ies the AC voltage produced in the PMSG into a DC vol tage. The

goal is to output a constant V D C , regardless of the ampl i tude of the input three-phase

voltage. To achieve that, i t is necessary to contro l the trans istor br idge ins ide the

rect i f ier , provid ing i t f rom the contro l ler .

Figure 11 shows the e lectr ic inner schema of the VLT. Both c ircuits have d if ferent

character is t ics, corresponding to their object ives. F igure 11 a) shows the schema of the

speed contro l VLT, whereas f igure 11 b) shows the schema of the act ive rect i f ier . As the

object ive is to output DC power, the inductors have been removed for the act ive rect i f ier .

Figure 11 : Elect r ic schema of both VLT f requency converter (source: Danfoss).


22

Table 1: Name Plate Detai ls for Inver ter/Rect i f ier

Type 2 in 1 Inverter/Rect if ier

Manufacturer Danfoss

Input Voltage 3 x 380-500 V

Input Frequency 50/60Hz

Input Current 6.5/5.7 A

Output voltage 3x0-Vin

Output frequency 0-590Hz

Output Current 7.2/6.3 A

Power Rating 3.0 kW (400 V) /4.0 HP (460 V)

2.2.2. INDUCTION MOTOR

Figure 12 shows the squirre l cage induct ion motor used for s imulat ing the behaviour of

the wind turbine. I t is a MT71A-14, manufactured by ABB. F igure 13 shows a table wi th

the main character is t ics of the motor . The motor has been connected in star

conf igurat ion. Further detai ls can be found in the datasheet.

Figure 12 : ABB MT71A-14 Induct ion motor .


23

Table 2: Character ist ics of the IM .

Parameter Value (Unit)

Rated power 0.25 kW

Rated voltage 380-420 V

Rated current 0,95 A

Power factor 0.66

2.2.3. PERM ANENT MAGNET SYNCHRONUS GENER ATOR

Figure 13 shows the PMSG used in the set -up. I t has been coupled wi th the induct ion

motor, so the torque produced by the IM induces the exc itat ion in the s tator to produce

AC power . The generator has an output of data measurement of the rotor pos it ion . This

data has been used to provide information about the s ituat ion of the system to the d ig ita l

control ler .

Figure 13 : SEW Eurodr ive PMSG (source: SEW Eurodr ive) .


24

Table 3: Name Plate Detai ls for PMSG

Type Permanent Magnet Synchronous

Generator , 3-phase.

Manufacturer SEW -Eurodr ive Grove, DK.

RPM 0-3000 rpm

Input Current 0 .95 / 6 A

Operating Voltage 169V

Voltage 400V

Frequency 150Hz

Attached between the induct ion machine and the PMSG, a torque meter was insta l led.

The torque meter is a DR-3000 manufactured by Lorenz Messtechnik , wi th a rank of

measurement unt i l 2 N·m. The measurements of the torque were used as a reference for

the contro l ler . F igure 14 shows the torque meter.

Figure 14 : Contact less torque meter sensor DR-3000 (source: Lorenz Messtechnik

Gmbh).

Table 4: Name Plate Detai ls for Torque Meter

Type DR-3000

Manufacturer Lorenz Messtechnik

Output Signal ±25000 dig its

Control Signal 25000 dig its

Imp./Rev 360


25

2.2.4. CONTROL HARDWARE

The main purpose of the project is to implement a dig ita l contro l ler for the PMSG; an

interfac ing hardware is to connect the PMSG and the d ig ita l contro l ler is required.

Modern HAWT systems are steered us ing computers wi th advanced control systems.

dSPACE (Dig ita l Signal Process ing and Contro l Engineer ing) is a Hardware In the Loop

(HIL) s imulat ion system that is used for the development and test ing of contro l systems

ut i l ised in the operat ion of complex machines and systems. The HIL s imulators mimic

the actual system and helps in the wide study of the system as a whole.

Figure 15 : dSPACE hardware used for contro l purpose (source: dSPACE).

W ith the recent trends, the HIL s imulat ion is widely used in the Of fshore industry and

other marine appl icat ions. By integrat ing the var ious components in the system, the HIL

can test the control systems help ing in avoid ing r isks in the phys ical environment. The

main benef i ts of us ing the HIL is to conduct tests that require r igorous test ing and

chal lenges the safety of the real t ime machines.

The dSPACE uses a user interface (UI) , ca l led the Contro l Desk. I t is an exper iment

sof tware that incorporates seamless Electronic contro l uni ts (ECU) developed by

performing the ins truct ions by the user on a s ingle work ing p latform f rom the s imulat ion

to the actual des ign system in the phys ical environment . The sof tware has been


26

connected wi th MATLAB/Simul ink , in order to become the transmissio n system f rom the

d ig i ta l contro l ler into the dr ives.

Table 5 shows the character is t ics of the dSPACE Microlabbox hardware. F igure 16

shows the interre lat ionship between al l the components af fect ing the hardware and the

system.

Figure 16: Hardware Diagram


27

Table 5: Dspace Micro lab Box Detai ls

MicroLab Box Processor NXP (Freescale) Qor lQ P5020, dual -core, 2 GHz 32 KB L1 data cache per core ,

Real t ime processor 32 KB L1 instruct ion cache per core, 512 KB L2 cache per core, 2 MB L3 cache tota l

Memory 1 GB DRAM 128 MB f lash memory

Boot t ime

Autonomous booting of appl icat ions f rom f lash

(depending on appl icat ion s ize) , ~5 s for a 5 MB

appl icat ion

Interface In tegrated Gigabi t Ethernet host interface.

Analog input 8 14-bi t channels , 10 Msps, d if ferent ia l; funct ional i t y:

f ree running mode 24 16-bi t channels, 1 Msps,

d if ferent ia l; funct ional i ty: s ingle convers ion and burst

convers ion mode with d if ferent tr igger and interrupt

opt ions

Input voltage range -10V to +10V

Analog output 16 16-bit channels , 1 Msps, sett l ing t ime: 1 µs

Output voltage range -10V to +10V

Output current ± 8 mA

Physical connect ions 4 x Sub-D 50 I /O connectors

4 x Sub-D 9 I /O connectors


48 x BNC I /O connectors



27 x spr ing-cage terminal b lock connectors wi th 8

p ins each

3 x RJ45 for Ethernet (host and I /O)

USB Type A ( for data logging)

2 x 2 banana connectors for sensor supply

Power supply


28

3. MODELLING OF THE SYSTEM

In th is chapter, the mathemat ical model of the set -up ment ioned in chapter 2.2. is g iven.

In order to develop an adequate contro l system, i t is important to def ine an accurate

model, as the contro l des ign is based on the model. Cons ider ing th is, the goals for th is

chapter are:

- Giv ing the mathemat ical equat ions regarding to each of the e lectr ica l device s

composing the system,

- Develop ing the equivalent model of the set-up in Simul ink ,

- Simulate and compare the resul ts of the model regarding the theoret ica l va lues.

3.1. MATHEMATICAL EQUATIONS

3.1.1. D-Q REFERENCE FRAME

The phys ical model of the e lectr ical machines can be descr ibed in the three -phase ABC

space vector system. In most of the c lassic AC dr ives the generat ion of the three s ine

waves is based on motor elec tromechanical character is t ics and on an equivalent model

for the motor in i ts s teady s tate. However, the s inusoidal references are d if f icul t to use

for contro l regulat ion purposes. L inear contro l lers (PID) are unable to handle the contro l

of th is k ind of references wi thout a lterat ion [ 36].

To transform the s inusoidal va lues to constant var iables , two-step matr ix transformat ion

is used [37]. The f irst transformat ion converts the s inusoidal waves (ABC) into a

stat ionary two-phase f rame (αβ) . The formula of th is transformation, cal led “Clarke ’s

t ransformat ion ” , can be found in equat ion (1). I t must be noted that, for th is paper , the

poss ib i l i t y of having an unbalanced phase has not been cons idered.

[𝐾𝛼

𝐾𝛽] = [

1 0 01

√3

2

√30] [

𝐾𝑎

𝐾𝑏

𝐾𝑐

] (1)

Where K is the space vector represent ing e lectr ica l var iables (current, vo l tage, f lux,

etc) .


29

The main purpose of the second transformation, “Park ’s transformation” , a ims at

conver t ing the s inusoidal waves calculated wi th “Clarke ’s” in to constant var iables, by

t rans lat ing the s tat ionary f rame components (αβ) to a rotat ing reference f rame

components (dq) which rotates wi th an angular speed 𝜔𝐾. Then, i t is possib le to use

these values as references for the purpose of contro l, based on the re lat ionship with the

var iables that need to be contro l led ( torque/f lux) . The formula of the “Park ’s

t ransformat ion ” can be found in equat ion (2).

[𝐾𝑑

𝐾𝑞] = [

𝑐𝑜𝑠(𝜑) 𝑠𝑖𝑛(𝜑)

−𝑠𝑖𝑛(𝜑) 𝑐𝑜𝑠(𝜑)] [

𝐾𝛼

𝐾𝛽] (2)

Where φ is the angle between "α" ax is and "d" ax is (see below f igure 1 7).

Figure 17 shows the space vector def in i t ion for the s tator vol tage in the three reference

f rames. F igure 18 shows the representat ion over t ime of the measurements of a var iable

of a unitary ampli tude and a f requency of 50 Hz in the three d if ferent reference systems

(abc, a lpha/beta and dq).

Figure 17 : Space vector representa t ion of the voltage in ABC, αβ and d-q f rames [38] .


30

Figure 18 : Graphical representat ion of a var iable in ABC, αβ and d-q f rames.

3.1.2. EQUATIONS IN D-Q FRAME

As it has been ment ioned , the induct ion motor pretends to s imulate the wind turb ine

rotor . I t is therefore important to des ign the model of the induct ion motor proper ly, in

order to s imulate the amount of energy transferred f rom the VLT into the mechanical

torque feeding the PMSG.


31

Figure 19 shows the equivalent per-phase c i rcuit of the induct ion machine. The energy

is transferred f rom the stator to the rotor through the mutual inductance.

Figure 19 : Per-phase equivalent c ircu i t in the d-q f rame of an induct ion motor [36]

The d if ferent ia l state equations of the induct ion motor for the d-q f rame are g iven in the

equat ions below (3-7) . I t is poss ib le to obta in the generated torque and f lux for the

PMSG at a spec if ic appl ied power [37].


32

𝑑𝑖𝑠𝑑

𝑑𝜏=

𝑅𝑠𝐿2𝑟 + 𝑅𝑟𝐿2

𝑚

𝐿𝑟𝑤𝜎

𝑖𝑠𝑑 +𝑅𝑟𝐿𝑚

𝐿𝑟𝑤𝜎

𝜓𝑟𝑑 + 𝜔𝑘𝑖𝑠𝑞 + 𝜔𝑟

𝐿𝑚

𝑤𝜎

𝜓𝑟𝑞 +𝐿𝑟

𝑤𝜎

𝑢𝑠𝑑 (3)

𝑑𝑖𝑠𝑞

𝑑𝜏=

𝑅𝑠𝐿2𝑟 + 𝑅𝑟𝐿2

𝑚

𝐿𝑟𝑤𝜎

𝑖𝑠𝑞 +𝑅𝑟𝐿𝑚

𝐿𝑟𝑤𝜎

𝜓𝑟𝑞 − 𝜔𝑘𝑖𝑠𝑑 − 𝜔𝑟

𝐿𝑚

𝑤𝜎

𝜓𝑟𝑑 +𝐿𝑟

𝑤𝜎

𝑢𝑠𝑞 (4)

𝑑𝜓𝑟𝑑

𝑑𝜏= −

𝑅𝑟

𝐿𝑟

𝜓𝑟𝑑 + 𝜔𝑟𝜓𝑟𝑞 +𝑅𝑟𝐿𝑚

𝐿𝑟

𝑖𝑠𝑑 (5)

𝑑𝜓𝑟𝑞

𝑑𝜏= −

𝑅𝑟

𝐿𝑟

𝜓𝑟𝑞 + 𝜔𝑟𝜓𝑟𝑑 +𝑅𝑟𝐿𝑚

𝐿𝑟

𝑖𝑠𝑞 (6)

𝑑𝜔𝑟

𝑑𝜏=

𝐿𝑚

𝐽𝐿𝑟

(𝜓𝑟𝑑𝑖𝑠𝑞 − 𝜓𝑟𝑞𝑖𝑠𝑑) −1

𝐽𝑡0 (7)

Where:

iS is the s tator ’s current vector Lm is the mutual inductance

uS is the s tator ’s voltage vector ψS is the s tator ’s f lux vector

RS is the s tator ’s res is tance ψ r is the rotor ’s f lux vector

R r is the rotor ’s res is tance L r is the rotor ’s inductance

ω k reference f rame angular speed

The main equat ions of a PMSG can be expressed d irec t ly f rom the equat ions of a DC

exc ited synchronous generator , wi th the s impl i f icat ion that PMSG does not have damper

windings. Equat ions in steady-state for a PMSG in dq f rame are given below (8 -14) [39-

40]. F igure 20 shows the e lectr ica l c ircu i t of the connect ion between the PMSG and the

act ive rect i f ier

𝑑𝑖𝑑

𝑑𝑡=

−𝑅𝑠𝑖𝑑 + 𝐿𝑞𝑝𝜔𝑟𝑖𝑞 + 𝑢𝑑

𝐿𝑑

(8)

𝑑𝑖𝑞

𝑑𝑡=

−𝑅𝑠𝑖𝑞 − 𝐿𝑑𝑝𝜔𝑟𝑖𝑑 − 𝑝𝜓𝑓𝜔𝑟 + 𝑢𝑞

𝐿𝑞

(9)

𝜓𝑑 = 𝐿𝑑𝑖𝑑 + 𝜓𝑚 (10)

𝜓𝑞 = 𝐿𝑞𝑖𝑞 (11)


33

𝑇𝑒 = 1.5𝑝(𝜓𝑞𝑖𝑑 − 𝜓𝑑𝑖𝑞) (12)

𝜔𝑒 = 𝑝𝜔𝑟 (13)

𝑑𝜔𝑟

𝑑𝑡=

1

𝐽𝑒𝑞

(𝑇𝑒 − 𝐵𝜔𝑟 − 𝑇𝑚) (14)

Where:

iS is the s tator ’s current vector Tm is the mechanical t ime constant

uS is the s tator ’s voltage vector ψR is the rotor veloc ity

RS is the s tator ’s res is tance ωe is the rotat ional ve loc ity of the stator

ψ f is the permanent magnet f lux ω k reference f rame angular speed

T e is the electromagnetic torque p,B,J refer to pole pairs, f r ic t ion and iner t ia

Ld / q is the inductance in d and q ψd / q is the magnet f lux in d and q

Figure 20: Electr ic c ircu it between the PMSG and the act ive rect i f ier (source: [ 40]) .

3.1.3. PER-UNIT MODEL

A per-unit model is a system of express ion for quant i t ies in re lat ion to a g iven base

value. I t , therefore, represents the propor t ion of the value in respect to an arb itrar i l y

selected value. The unit is p.u (per-uni t) .

Per-unit models are widely used in power electronics . The main purpose is having a l l

the var iables regarding reference uni tary va lues. In the case of the project, the use of


34

per -units makes the implementat ion of the control a lgor i thm eas ier ( fur ther deta i ls are

g iven in chapter 5). Table 6 shows the re lat ion of the g iven per -uni t base values.

Table 6: Table of the selected base var iables

Base variable Value

Voltage 400V

Current 10A

3.2. SIMULINK MODEL

This sect ion of the repor t looks at expla in ing the implementat ion in Simul ink of the

mathemat ical equations descr ibed in the previous sect ion. The equations involv ing each

of the dr ives, descr ibed in sect ion 3.1, have been independently model led and p laced

on interconnected subsystems. The f inal model can be found in F igure 21.

Figure 21: Model of the set-up des igned in Simul ink .

Figure 22 shows the VLT used for contro l l ing the induct ion machine and input the

f luc tuat ions. The design has been done based on the real c ircu it of the f requency

conver ter ( f igure 11) . The contro l device of the VLT inc ludes d if ferent parameters and

est imat ion of systems that have been simpl i f ied for the s imulat ion.


35

Figure 22: Model of the VLT created in Simul ink .

Figure 23 shows the schema of the induct ion motor generated with Simul ink . I t has been

model led fo l lowing equat ions (3 -7) . Two independent subsystems comprise i t . The f irst

one ( f igure 23 a) , transforms the measured stator voltage into the dq f rame us ing

Clarke ’s and Park ’s t ransformations. Af terwards, these values are used to compute the

accelerat ion of the rotor and the produced electromagnet ic torque ( f igure 23 b).


36

Figure 23: Model of the IM created in Simul ink .


37

The torque generated by the induct ion motor feeds the PMSG, as i t can be seen in

f igures 24 a-d. The subsystem of the PMSG is formed by many d if ferent subsystems,

comprised of d if ferent funct ions. The “Electr ical model” and “Mechanical model”

subsystems conta in the equat ions (8 -14) .

Other subsystems have been used in order to transform the s ignals f rom one reference

f rame to another and to conver t the d ig ita l s ignals into e lec tr ic i t y.

Figure 24 a-b: Model of the PMSG.


38

Figure 24 c-d: Model of the PMSG.

Figure 25 shows the model of the act ive rect i f ier implemented in Simulink . I t cons ists o f

6 MOSFET semiconductors fed wi th a PW M s ignal. The des ign of the s ignal and the

control a lgor i thms used for the contro l are expla ined in chapter 5. In order not to short

the power supply, the control s ignal g iven to the bottom leg of the rect i f ier is a lways the

complement of the s ignal g iven to the top leg.


39

Figure 25 : Model of the act ive rect i f ier implemented in Simulink .


40

4. PREPARATION OF THE SYSTEM

Previous chapters have g iven theoret ica l ins ights about the ex ist ing set -up and the

governing equat ions . In order to complete the model and get some parameters required

for developing the contro l system, some exper iments have been done. The goals of th is

chapter are:

- Expla in the development of tes ts to f ind the unknown parameters ,

- Est imate the values .

4.1. ESTIMATION OF THE PARAMETERS OF THE PMSG

One of the most important th ings when def in ing a mathemat ical model is having the

values of the real parameters integrat ing i t , so the model can be val idated by compar ison

or use wi th contro l purposes.

Among the character is t ics g iven about the PMSG, i t is not possib le to f ind the values for

the impedances in the stator (LS , RS) and the pole pairs. Therefore, i t was necessary to

develop a method through which i t is poss ib le to get the values. This method consis ts

on a ser ies of measurements of d if ferent var iables, to calculate the internal parameters .

The procedure is fu l ly descr ibed in [41] .

For obta in ing the pair of poles, the f requency of the stator vol tage was observe d in an

osc i l loscope for a constant veloc ity. Us ing equation (15), the pole pair was computed.

𝑝𝑝 =

60 𝑓

𝑛𝑅

(15)

The resis tance s tator was obta ined us ing a mul t imeter and measur ing the l ine

res istance. One of the consequences of th is measurement was real izing that the one of

the phases of the PMSG was unbalanced. The causes are unknown. Consequent ly, i t

was necessary to have i t replaced by another one.

Equal ly, the mult imeter was used for measur ing the res istance of the IM and adjusted in

the opt ions of the VLT. Consequent ly, the appl ied veloc ity showed in the screen of the

conver ter converged at the one measured by the torque meter .


41

For calculat ing the inductance, an al ignment of the rotor was done with the

corresponding axes, to calculate the value of the inductance for the d and q f rames.

When the rotor is a l igned wi th the A ax is , the obta ined inductance value wi th the

mult imeter corresponds to the value of L d . Similar ly, i t is poss ible to calculate the valu e

of the q ax is inductance when the rotor is locked at 90 º.

Other parameters, such as the f lux l inkage and the iner t ia were obta ined f rom the

datasheet .

Table 7 shows the values obta ined wi th the performed test for the PM SG.

Table 7: Obta ined parameters of the PMSG.

Variable Measured value (Unit )

Pole pair (pp) 3

Stator resis tance (RS) 21,5 Ω

D ax is inductance (L d) 26 mH

Q ax is inductance (L q) 31,7 mH

Permanent magnet f lux (ΨM) 0,15 W b

Inert ia (J) 0,16 kg·cm 2

4.2. VALIDATION OF THE PMSG MODEL

The next s tep af ter f ind ing the d if ferent parameters of the set-up is pursuing a val idat ion

of the ex ist ing model. This val idat ion can be done by doing d if ferent measurements and

f inding i f the obta ined values match the ones obta ined wi th the Simulink model .

The procedure used is very s imple. Us ing one of the VLT s for applying a constant

veloc i ty, a three-phase res ist ive load was connected in delta to the output of the PMSG.

The value of the load was measured us ing a mult imet er as wel l as the phase vol tage,

the voltage and the current in the load.

Table 8 shows the rates and values obtained dur ing the measurements. Three d if ferent

samples were taken, f ix ing the veloc i ty of the rotor at three d if ferent speeds .


42

Figure 26 : Exper iment for val idat ion of the parameters of the PMSG.

Table 8: Data obta ined in the test .

Sample nª ωR ( rad/s) VA (V) VB (V) VC (v) IA (A) IB (A) IC (A)

1 26 6,95 6,08 6,09 0,189 0,188 0,162

2 50 13,36 11,62 11,69 0,365 0,362 0,312

3 80 20,87 18,12 18,25 0,567 0,569 0,49

The same condit ions were appl ied in a vers ion of the Simul ink model in open - loop. Given

the constant value for the veloc i ty, the measurements in the load were evaluated. The

model is shown in f igure 27. F igure 28 and 29 shows the graphs for the voltage and

current obta ined wi th the model for the same condi t ions as the tes t .

I t can be apprec iated that the d if ferences are smal l , but u lt imately insignif icant in the

balanced phases , val idat ing the values obta ined for the model .


43

Figure 27 : Used model in open- loop.

Figure 28 : Voltage in the res istance for the d if ferent phases in the Simulink model .


44

Figure 29 : Phase current in the Simul ink model .

Table 9 : Compar ison between measured and s imulated values .

ωR (rad/s) M/S VA (V) VB (V) VC (v) IA (A) IB (A) IC (A)

26 Measured 6,95 6,08 6,09 0,189 0,188 0,162

Simulated 6,91 6,91 6,93 0,198 0,199 0,1983

50 Measured 13,36 11,62 11,69 0,365 0,362 0,312

Simulated 13,24 13,25 13,30 0,380 0,382 0,380

80 Measured 20,87 18,12 18,25 0,567 0,569 0,490

Simulated 21,04 21,07 21,14 0,604 0,607 0,605

4.3. ESTIMATION OF THE PARAMETERS OF THE ACTIVE RECTIFIER

Fi l ters of some sort are essent ia l to the operat ion of most e lec tronic c i rcui ts, to mit igate

harmonic and EMI [42] .


45

Regarding to the connect ion between the act ive rect i f ier and the f requency converter,

commonly, two d if ferent types of f i l ters are ut i l ized at the supply -s ide of the act ive f ront -

end [24].

The s implest f i l ter is the L -f i l ter . The problem assoc iated wi th the L-f i l ter is to obta in the

required damping , i t wi l l result in a large value which not only impairs power dens ity but

a lso results in a voltage drop, thereby increas ing losses in the f i l ter and the converter

as the converter gain rat io is increased ( i .e . , boost rect i f ier) . Thereby, us ing an LC

conf igurat ion would be much more ef f ic ient .

However , due to the lack of sk i l ls of the authors and the impossib i l i ty of adding the f i l ters

to the s imulat ion, no f i l ter was implemented between the rect i f ier and the PMSG.

The VLT circuit inc ludes in the output a capac itance, as i t can be observed in the e lectr ic

schema shown in F igure 11 b. The funct ion of th is capaci tance is reduc ing the r ipples

and var iance of the output DC voltage. The value of th is e lement was ob served f rom the

inner c ircuit . Table 10 shows the values.

Table 3: Values of the inductance of the wires.

Variable Measured value (Unit )

LV L T 7 mH

CV L T 142 µF


46

5. CONTROL

Once the model has been def ined, i t is necessary to choose the control a lgor i thm and

def ine i ts implementat ion. The goals of th is chapter are:

- Def in ing the a lgor i thm invest igated for the topology of the e lectr ica l devices ,

- Deta i l a l l the necessary components for the model to work ,

- Show the cal ibrat ion of the sensors .

5.1. CONTROL ALGORITHM

Many dif ferent contro l a lgor i thms have been used for contro l l ing a PMSG. F igure 30

shows a c lass if icat ion of the d if ferent ex is t ing contro l a lgor i thms that can be found in

the l i terature [43].

Figure 30 : C lassif icat ion of contro l a lgor i thms for PMSG [43].


47

As i t has been ment ioned in Chapter 2, many of these have been researched in the

l i terature for the given topology. For this purpose, Stator Flux Or iented Control has been

implemented.

5.2 STATOR FLUX ORIENTED CONTROL (SFOC)

One of the most used contro l methods is Fie ld Oriented Contro l (FOC). I t requires

obtain ing the currents of the PMSG in the rotat ing reference f rame via the Park ’s

t ransformat ion and control l ing the d axis and q ax is current for contro l l ing the

generator ’s torque and speed respect ive ly. F ie ld Or iented Control can a lso be

implemented by obta in ing the currents of the PMSG in the stat ionary reference f rame

via the Clarke ’s transformation and contro l l ing the resultant currents . The rotat ing

reference f rame was chosen as the contro l s trategy due to i t s ease of implementat ion.

A c losed loop contro l strategy was appl ied when contro l l ing the q ax is current wi th

respect to the generated e lectromagnet ic torque and the d ax is cur rent is contro l led so

the amount of react ive power generated is min imised. The bas ic contro l d iagram is

shown in the f igure below.

Figure 31 : Schema of the contro l a lgor i thm implemented on the set -up [43] .


48

For th is contro l to be implemente d the three-phase s tator current and the rotor pos i t ion

need to be measured. For the contro l of the q ax is cur rent , the outer contro l loop is for

the torque generated by the PMSG and the inner contro l loop is the current contro l. The

re lat ionship between the torque and the current is g iven in equat ion (12). To contro l the

currents independent ly of each other , the decoupl ing term must be added to both

currents. The decoupl ing terms can be observed in equat ions (8 -9) .

For contro l of the d ax is current, the constant torque angle contro l was implemented. I t

s imply cons is ts on dr iv ing the current to i ts min imum poss ib le value to reduce the amount

of react ive power generated by the PMSG.

An impor tant requirement for the contro l syste m to be ef fec t ive is an accurate tuning of

the PI contro l lers . The data about the set -up was col lec ted using empir ica l exper iment

methods. Equations (8-14) were rearranged to f ind the values of the l inear contr o l lers,

us ing Laplace transform ation.

𝑢𝑑

∗ = (𝑖𝑑∗ − 𝑖𝑑)(𝐾𝑑𝑝 +

𝐾𝑑𝑖

𝑠) − 𝐿𝑞𝑝𝜔𝑟𝑖𝑞 (16)

𝑇𝑒∗ = 𝐾𝑡𝑒𝑝𝑖𝑞((𝐿𝑞 − 𝐿𝑑)𝑖𝑑 − 𝜓𝑚) (17)

𝑢𝑞

∗ = (𝑖𝑞∗ − 𝑖𝑞)(𝐾𝑞𝑝 +

𝐾𝑞𝑖

𝑠) − 𝐿𝑑𝑝𝜔𝑟𝑖𝑑 − 𝑝𝜓𝑓𝜔𝑟 (18)

As i t has been ment ioned, the value of the d irect current was def ined to be as smal l as

poss ib le. On the other hand, i t was necessary to g ive a reference value for the produced

e lectromagnet ic torque.

The d and q ax is reference voltages are generated and then t ransformed to their three

phase values before going to the PW M generator which g ives the switching s ignals for

the act ive rect i f ier based on the contro l developed.

5.2.1. PI TUNING ON DIRECT AXIS

The tuning of the PI control lers involved in the above descr ibed c ontro l scheme is done

by analys ing the var ious transfer funct ions that the current s ignals are put through and


49

then choos ing the appropr iate constants for the system to a stable system us ing the

opt imum modulus method of tuning, which is expla ined in more deta i l below.

Figure 32 : Block d iagram of transfer funct ions for d irect ax is current

The error between the reference and the measured d irec t ax is cur rent is fed to the PI

control ler with a transfer funct ion G P I , the transfer funct ion can be descr ibed as:

𝐺𝑃𝐼(𝑠) =

𝑈(𝑠)

𝐸(𝑠)= 𝐾𝑑𝑝 +

𝐾𝑑𝑖

𝑠 (19)

Where 𝐾𝑑𝑝 is the proport ional gain and 𝐾𝑑𝑖 is the integral gain.

The next transfer funct ion encountered is that of the pulse width modulat ion or PW M

that is being used before g iv ing the s ignals to the rect i f ier , the f requency of the PW M

signals is 10000 Hz so the t ime constant would be the inverse of that . Therefore :

Ts = 1/f = 10 - 4 s.

Using the t ime constant value, the transfer funct ion of the pulse width modulat ion G P W M

is def ined as:

𝐺𝑃𝑊𝑀(𝑠) =

𝑈(𝑠)

𝐸(𝑠)=

1

𝑠𝑇𝑠 + 1 (20)

Af ter the pulse width modulat ion , the s ignal is then g iven to the p lant . The complete

transfer funct ion of the p lant involves adding some of fset or feedforward terms in order

to obta in the compensated tr ansfer funct ions as expla ined in the previous sect ion. The

transfer funct ion of the p lant GPL can be obta ined as:

𝐺𝑃𝐿(𝑠) =

𝑈(𝑠)

𝐸(𝑠)=

1

𝑠𝐿𝑑 + 𝑅𝑠

(21)

Where R s and L d are s tator res istance and d ax is inductance respect ive ly.


50

Now three transfer funct ions have been obta ined, so the open loop transfer funct ion is

acquired by mult ip lying them al l together s ince they a l l fo l low one another in d irect

success ion or they are connected in ser ies . The open loop transfer funct ion G O L is

def ined as:

𝐺𝑂𝐿(𝑠) = 𝐺𝑃𝐿(𝑠) . 𝐺𝑃𝑊𝑀(𝑠) . 𝐺𝑃𝐼(𝑠) =

𝐾𝑑𝑖 + 𝑠𝐾𝑑𝑝

𝑠(𝑠𝑇𝑠 + 1)(𝑠𝐿𝑑 + 𝑅𝑠) (22)

As explained previous ly, the transfer funct ion was made to fo l low the opt imum modulus

des ign cr i ter ia wi th the damping factor ζ =√2

2. To make i t fo l low the s tated des ign cr i ter ia ,

then the open loop t ransfer funct ion must be of the fo l lowing form :

𝐺𝑂𝑀(𝑠) =

1

2𝜏𝑠(𝜏𝑠 + 1) (23)

Af ter rearranging the open loop transfer funct ion, i t was found that in order for the

system to fo l low the opt imum modulus des ign cr i ter ia, the PI gains have to be as fo l lows:

𝐾𝑑𝑖 =

𝑅𝑠

2𝑇𝑠

𝐾𝑑𝑝 =𝐿𝑞

2𝑇𝑠

(24)

5.2.2. PI TUNING ON QUADRATIC AXIS

Figure 33: Block d iagram of transfer funct ions for quadrat ic ax is current


51

The error between the reference and the measured d irec t ax is cur rent is fed to the PI

control ler with a transfer funct ion G P I , the transfer funct ion can be descr ibed as:

𝐺𝑃𝐼(𝑠) =

𝑈(𝑠)

𝐸(𝑠)= 𝐾𝑞𝑝 +

𝐾𝑞𝑖

𝑠 (25)

Where 𝐾𝑞𝑝 is the proport ional gain and 𝐾𝑞𝑖 is the integral gain.

The next transfer funct ion encountered is that of the pulse width modulat ion or PW M

that is being used before g iv ing the s ignals to the rect i f ier . The f requency of the PWM

signals is 10000 Hz so the t ime constant would be the inverse of that . Therefore :

Ts = 1/f = 10 - 4 s

Using the t ime constant value, the transfer funct ion of the pulse width modulat ion G P W M

is def ined as:

𝐺𝑃𝑊𝑀(𝑠) =

𝑈(𝑠)

𝐸(𝑠)=

1

𝑠𝑇𝑠 + 1 (56)

Af ter the pulse width modulat ion , the s ignal is then g iven to the p lant . The complete

transfer funct ion of the p lant involves adding some of fset or feedforward terms in order

to obta in the compensated transfer f unct ions as expla ined in the previous sect ion. The

transfer funct ion of the p lant GPL can be obta ined as:

𝐺𝑃𝐿(𝑠) =

𝑈(𝑠)

𝐸(𝑠)=

1

𝑠𝐿𝑞 + 𝑅𝑠

(27)

Where R s and L q are s tator res istance and d ax is inductance respect ive ly.

Now three transfer funct ions have been obta ined, so the open loop transfer funct ion is

acquired by mult ip lying them al l together s ince they a l l fo l low one another in d irect

success ion or they are connected in ser ies . The open loop transfer funct ion G O L is

def ined as:

𝐺𝑂𝐿(𝑠) = 𝐺𝑃𝐿(𝑠) . 𝐺𝑃𝑊𝑀(𝑠) . 𝐺𝑃𝐼(𝑠) =

𝐾𝑞𝑖 + 𝑠𝐾𝑞𝑝

𝑠(𝑠𝑇𝑠 + 1)(𝑠𝐿𝑞 + 𝑅𝑠) (28)


52

As expla ined previous ly the transfer funct ion was made to fo l low the opt imum modulus

des ign cr i ter ia wi th the damping factor ζ =√2

2. To make i t fo l low the s tated des ign cr i ter ia ,

then the open loop t ransfer funct ion must be of the fo l lowing form :

𝐺𝑂𝑀(𝑠) =

1

2𝜏𝑠(𝜏𝑠 + 1) (29)

Af ter rearranging the open loop transfer funct ion, i t was found that in order for the

system to fo l low the opt imum modulus des ign cr i ter ia, the PI gains have to be as fo l lows:

𝐾𝑞𝑖 =

𝑅𝑠

2𝑇𝑠

𝐾𝑞𝑝 =𝐿𝑞

2𝑇𝑠

(30)

5.3. ROTOR VELOCITY ESTIMATION

Nowadays, i t is of ten seen that reducing the number of sensors corresponds d irec t ly to

a reduct ion in cost. Never theless, knowing the pos i t ion and the angular veloci ty of the

rotor are st i l l fundamental for the implementat ion of the contro l a lgor i thms. W ith that

purpose, d if ferent est imations methods are used [ 44] .

The most popular solut ions can be categor ized into the ones that es t imate the back -

EMFs of rotor f lux and the ones that rely upon the magnet ic anisotropy of the machine.

The use of the f irst k ind of a lgor i thm impl ies the calculat ion of the vol tages generated

by the PMSG employing the machine model to compensate the voltage drop across the

motor impedances [45] .

Another poss ib i l i t y re l ies on the use of Extended Kalman F i l ters (EKF) for the est imation

of the rotor angle, which a l lows for very good t rack ing performance, even in the presence

of h igh levels of noise in the measurements . However, i ts appl icat ion is l im ited by the

computat ional ef for t needed to perform the calculat ions, as wel l as the dif f icu lty of the

EKF calculat ions [46].

For the project, the PMSG incorporates a resolver for measur ing the angle of the rotor .

Knowing the reference point of the angle and being able to measure the d if ference in

between two sample t imes, i t is poss ib le to compute the d if ferent speeds through a


53

Forward Euler c ircui t . The c ircuit is shown in F igure 34. No computat ional ly heavy

a lgor i thms are therefore needed. Alternat ive ly, tak ing in to account that the rotat ion

veloc i ty is constant , i t would also be poss ib le to input i t manual ly.

Figure 34 : Est imat ion of the veloc i ty.

5.4. SIGNALS

As it is ment ioned in chapter 2.2.4, the hardware used for in terfacing the model in

Simul ink wi th the set -up is dSPACE Micro labBox. The hardware p lays two fundamental

ro les in the system: on the one hand, i t was used to send the PW M signals to the rect i f ier ;

on the other hand, i t was used for sending the measurement s of cur rents and rotor angle

needed for the feedback of the contro l.

Among the s ignals that the project managed, two categor ies can be d ist inguished: inputs

to the control ler and outputs .

One of the most important inputs is the resolver , a type of rotat ional angle sensor for

control l ing the dr ive that produces the rotat ional movement in the PMSG. To contro l the

generator, i t is necessary to detect the magnetic pole pos it ions of the generator and

measure i ts rotat ional speed. The resolver serves a s a sensor and performs the fo l lowing

funct ions. I t is connected to the dr ive tra in which rotates with the rotor of the magneto.

The angle is detected by means of an e lectr ic pulse generated due to the d if ference in

the reactance of the rotor and f ixed sta tor of the PMSG. The resolver has a f requency

over 70 kHz and can measure the pos it ion at rotat ional speeds up to 30000 rpm.

Equal ly important are the current transducers, used for obta in ing the current values

generated by the PMSG. A transducer is a devi ce that converts an energy into i ts


54

equivalent output . Usually a current transducer consis ts of a current t ransformer and a

s ignal processor made into a s ingle combinat ion for measur ing and monitor ing purposes.

Typical ly, a current transducer operates at v ar ious input ranges and provides analog ue

output s ignals that are most compat ib le with PLCs, dataloggers and SCADA systems.

These current sensors are known for their h igh performance, f lex ib i l i t y and durabi l i t y.

The character is t ics of the sensors are g iven in Table 11. Due to i ts range and the

sens i t ivi t y, i t is su itable enough for catching the changes in the current.

Table 4: Character ist ics of the transducers .

Load Maximum Current ±50A

Load Frequency Bandwidth DC ~ 120 kHz

Isolat ion Vol tage 4800 VA C

Sens i t iv i t y 40 mV/A.

Operat ing Voltage Regulated 5VD C , or 8 ~ 35 VD C

Operat ing Current 20mA(max)

Load No Current Output Terminal Vol tage 2.5 VD C

For th is project, two t ransducers were used. Assuming that the phases are balanced,

the homopolar component of the current should be 0 [ 37]. Therefore, the remain ing

phase current can be est imated f rom the combinat ion of the other two. The output of the

transducer was fed to the dSPACE hardware analog ue input channel . They are shown

in f igure 35.


55

Figure 35 : Current transducer.

Regarding the output s ignals, the PW M has been sent to the act ive rect i f ier us ing opt ic

f iber wires connected to the dSPACE hardware d ig i ta l input /output channel by a d ig i ta l

board. F igure 36 shows the d ispos i t ion of the equipment needed to interconnect dSPACE

Dig ita l output and the rect i f ier .


56

Figure 36 : Board wi th the opt ic f iber

Before implement ing the sensors in c lose loop, a check up was done ; in order to see the

accuracy and needed cal ibrat ion. F irst , the generator was run in a constant low speed

for 45 seconds. Figure 37 shows the measurements obta ined f rom the resolver. The

values corresponded to a range between 0 and 360 (degrees). I t can be seen that the

resolver detects a reference and i t g ives a very accurate value for the pos it ion.


57

Figure 37 : Resul ts obta ined f rom the measurement of the rotat ion angle

Addi t ional ly, the pos it ion of the resolver was used for comput ing the rotat ional speed ,

as i t is expla ined in chapter 5.4. Figure 38 shows the obta ined results for the value of

the veloc i ty imput ing a constant value of 240 rpm. The obta ined veloc i ty is about 25

rad/s (239 rpm). Despi te the ex istence of peaks, the system could be considered prec ise

enough, espec ia l ly for h igh speeds (where the var iance of the est imation is smal ler) .


58

Figure 38 : Measured angle (rad) and veloc ity ( rad/s) for a 180 RPM constant veloc ity.

Current transducers proved to have some of fset and gain d if ference s. Therefore, i t was

necessary to des ign a c ircui t that would compensate the values. Both transducers had

d if ferent of fsets, so two independent c ircuits were developed. For that , a constant

current was appl ied to a load and i t was measured wi th the t ransducers . F igure 3 9 and

40 show the measured and est imated real current for both transducers. Nevertheless, i t

must be ment ioned that the of fset of the transducers changed f rom test to test . I t was

therefore imposs ible to get accurate data in the f inal test .


59

Figure 39 : Measured current wi th the transducer A before and af ter cal ibrat ion.

Figure 40 : Measured current wi th the transducer B before and af ter cal ibrat ion.

To ver i fy the output PW M signal sent to the rect i f ier , an osc i l loscope was used to

measure the s ignal , as i t can be seen in f igure 41.


60

Figure 41 : Measurement of PW M with an osci l loscope.


61

6. ANALYSIS OF THE RESULTS

After val idat ing the model and f ix ing the sensors, the control was supposed to be tested

in c losed- loop in the set -up. However , the object ive was not achieved. The goals of th is

chapter are:

- Expla in the implementat ion of the control ler in the set -up,

- Say which contro l system wi l l be chosen ,

- Expla in why the implementat ion was not done.

6.1. IMPLEMENTATION OF THE CONTROL LOOP

Figure 42 : SFOC algor i thm des igned for being implemented in the set -up

Figure 42 shows the a lgor i thm ready to be implemented in dSPACE. The analog ue inputs

were connected to the transducers , to obta in the measurements of the currents . The

subsystems attached to the inputs are the cal ibrat ion b locks for the transducers . V q and

Vd subsystems conta in the PI contro l lers and the reference values for I q and Id


62

6.2. RESULTS

Once the set-up was ready, the sensors were cal ibrated, and the PI tuned, the

implementa t ion was carr ied out. I t was necessary to give some references to the system.

The values used are shown in table 12:

Table 12: G iven values for the references in the implementat ion .

Rotat ional ve loc i ty 600 rpm

Reference of the d current 0 A

Reference of the q current 0.5 A

Switching f requency 10 kHz

The implemented speed was ass igned wi th in the range of the motor and the generator.

The switching f requency was selected to be lower than the maximum accepted by the

act ive rect i f ier . The select ion for d reference current was made in order to decrease the

generated react ive power , whereas the reference for q current is an arb itrary value that

would regulate the e lectromagnet ic torque (equation (12)) . The b igger the q current , the

b igger the generated e lectromagnet ic torque. In most of the wind turb ine contr ol lers , the

value of the torque is selected based on the veloc ity of the wind and the required/rated

power of the turb ine [1, 9, 11] .

Figure 43-46 show the values obtained in the s imulat ion for the currents and vol tages

in d and q f rame. An overshoot can be appreciated when the system starts to rotate (0,1

second). Despi te a few r ipples that can be observed in the current , the references are

tracked wi th an inf in i tes imal error . The value matched the mathematical ca lculat ions

done af terwards.

Figure 47 shows the re lat ionship between the mechanical power transmit ted to the

generator and the DC power consumed in the load. I t can be seen that despi te there are

some losses (res istance in the s tator, wires and voltage drops in the switches) , the rat io

is s ignif icant ly h igh.


63

Figure 43 : Id obta ined in Simulink

Figure 44 : Iq obta ined in Simulink


64

Figure 45 : Vd obta ined in Simul ink

Figure 46 : Vq obta ined in Simul ink


65

Figure 47 : Mechanical power and power in the load

Figure 48 shows the set -up ready for the implementat ion of the contro l ler .

Figure 48 : Set-up ready for the implementat ion.

The resul ts obtained f rom the implementat ion were inaccurate. The sensors d id not

measure a proper value for the current . F igure 4 9 shows the values of current a, b and


66

c. Current a and b were measured wi th the transducers, whereas c was computed f rom

the other two. The f igure expresses changes that could not correspond wi th real i t y,

because the current decreased when the veloc i ty of the motor increased. Moreover,

currents are not purely s inusoidal wi th an average of 0.

Figure 49 : Current values measured dur ing the f inal tes t.


67

Figure 50 show the values of the PWM signals sent to the VLT. I t can be seen that s ince

the current values are not what was expected, the PWM signal does not look l ike i t

should. That is why i t is not al igned and phased wi th each other .

Figure 50 : PW M signals generated in the tes t.

In order to measure the DC voltage, a mul t imeter was connected in the DC load. For a

rotat ional ve loc ity of 600 rpm the measured DC vol tage was 38 Vol ts ( f igure 51) . The

simulat ions carr ied out shows that , when the generator is given an uncontro l led

reference the currents fol low an open- loop system. I t can be seen in f igure 52. Thus,

due to the impossib i l i ty of feeding the contro l ler wi th the correct current value, the

generated PW M signals were random and, thus, the obtained value in DC was lower than

wi th a contro l ler .


68

Figure 51 : DC voltage measured dur ing the f inal test .

Figure 52: Results for the s imulat ion for 600 rpm, uncontro l led and contro l led.


69

7. CONCLUSION

The main goal of the project was des igning and implement ing a suitable contro l ler for a

set-up s imulat ing a real wind turb ine us ing dSPACE and Simul ink . In terms of the

des igning par t, the contro l ler has proved to be able to track the references g iven ( in a

range of real is t ic values) wi th PI contro l lers tuned fo l lowing the obta ined parameters

f rom the model . In terms of the implementat ion ho wever, the acquirement of data was

not poss ib le.

Addi t ional ly, another of the goals of the project, the model l ing of the PMSG has a lso

been proven to be successfu l. Despi te the lack of al l the whole dynamics of the system

(such as Hal l ef fec t) , i t has prove n to be a good approx imat ion, as i t can be seen in the

exper iments in open- loop descr ibed in sect ion 4.

Among the points that could be improved, the qual i t y, sens i t ivi t y and cal ibrat ion of the

sensors can be ment ioned. The imposs ibi l i ty of gett ing any results f rom one of the

sensors prevented the possib i l i t y of having complete resul ts. On the other hand, the

tuning of the PI could be improved in order to remove the over -shoot ing f rom the

currents. This would mean giv ing d if ferent values tha n the theoret ica l ones. I t m ight be

poss ib le that some unforeseen ef fects may have not been cons idered in the des ign of

the l inear contro l lers.

For future projects, another k ind of contro l ler could be tes ted (Direct Torque Contro l,

Sl id ing Mode Contro l…) and compare them with the obta ined results f rom th is project .

Moreover, d if ferent ways to re late the references to real parameters (such as torque or

rotat iona l speed) could be implemented, such as systems for referr ing the torque

reference according to the power required.

VOLTAGE CONTROL OF DIRECT -DRIVE WIND TURBINES FOR OFFSHORE WIND

POW ER CONNECTED TO DC GRIDS

70

REFERENCES

[1 ] E. Hossain, J. Hossain, N. Sak ib, R. Bayindir (2017) . “Modell ing and Simulat ion of

Permanent Magnet Synchronous Generator W ind Turbine: A Step to Microgr id

Technology” , In ternat ional Journal of Renewable Energy Research.

[2 ] E. Hossain, E. Kabalc i , R. Bayindir , R. Perez (2014). “Microgr id tes tbeds around the

wor ld: State of ar t ” , Energy Convers ion and Management 86 .

[3 ] F. Mal ik , M. Lehtonen (2016) . “A review: Agents in smart gr ids, ” Electr ic Power

Systems Research. ISSN 0378-7796.

[4] U.S. Depar tment of Energy (2008). “20% W ind Energy by 2030: Increas ing W ind

Energy’s Contr ibut ion to U.S. Electr ic i t y Supply. ”

[5 ] E. Vei l leux and P. W. Lehn (2014). " Interconnect ion of Direct -Dr ive W ind Turbines

Using a Ser ies-Connected DC Gr id," in IEEE Transact ions on Susta inable Energy, vol .

5, no. 1, pp. 139-147, IEEE.

[6] H. Zhang, F. Gruson, D. F lórez, C. Saudemont (2016). “Design and Contro l of a DC

Ser ies Of fshore W ind Farm Based on HVDC-MMC , ” Symposium de Genie Elect r ique.

[7 ] N. Hol tsmark and M. Mol inas (2013) . "Thermal analys is of matr ix and Back -To-Back

conver ters for ser ies -connected wind turb ines," IECON 2013 - 39th Annual Conference

of the IEEE Industr ia l Electronics Soc iety, Vienna, pp. 4967 -4972, IEEE.

[8] M. L iserre, R. Cardenas, M. Mol inas and J. Rodr iguez (2011) . "Overview of Mul t i -MW

Wind Turbines and W ind Parks," IEEE Transact ions on Industr ia l Electronics , vol . 58,

no. 4, pp. 1081-1095, IEEE.

[9] S. Lundberg (2006) . “W ind farm conf igurat ion and energy ef f ic iency s tudies - ser ies

DC versus AC layouts , ” Chalmers Univers ity of Technology.

[10] P. Monjean, J. Delanoe, J. Sprooten, C. Saudemont , B. Robyns (2011) . “Compar ison

Methodology of Of fshore W ind Farm Architec tures , ” Electr imacs.

[11] M. H. Mughal and L. Guoj ie (2015), "Review of Pi tch Contro l for Var iable Speed

W ind Turbine," 2015 IEEE 12th Int l Conf on Ubiqui tous Inte l l igence (UIC -ATC-ScalCom),

Bei j ing, pp. 738-744, IEEE.

[12] M. Cal lavik , M. Bahrman and P. Sandeberg (2012), "Technology developments and

p lans to solve operat ional chal lenges fac i l i tat ing the HVDC of fshore gr id, " IEEE Power

and Energy Soc iety General Meet ing, San Diego, pp. 1 -6, IEEE.

[13] H. Taghizadeh, A. M. Cross, R. Whitehouse and C. Barker (2015) , "Switched

capac itor DC-DC converters for HVDC applicat ions," 11th IET Internat ional Conference

on AC and DC Power Transmission, Birmingham, pp. 1 -9, IEEE.

[14] H. L i and Z. Chen (2008). "Overview of d if ferent wind generator systems and their

compar isons," in IET Renewable Power Generat ion, vol. 2 , no. 2, pp. 123 -138, IEEE.



71

[15] Bensalah, A & Joly and P & Merc ier (2018) . “W ell-posedness of a general ized t ime-

harmonic transpor t equat ion for acoust ics in f low , ” Mathemat ical Methods in the Applied

Sc iences, ResearchGate.

[16] A. Ahadi, X . Liang and W . Li (2017) . "An analyt ica l method for wind energy potent ial ,

re l iabi l i t y, and cost assessment for wind generat ion systems," 2017 IEEE/IAS 53rd

Industr ia l and Commerc ia l Power Systems Technical Conference ( I&CPS), Niagara

Fal ls, pp. 1-11, IEEE.

[17] A. B. Abrahamsen et al . (2018). "Comparison of Level ized Cost of Energy of

Superconduct ing Direct Dr ive Generators for a 10 -MW Offshore W ind Turbine," in IEEE

Transact ions on Applied Superconduct iv i ty, vo l. 28, no. 4, pp. 1 -5, IEEE.

[18] M. Pal and K. S. Sandhu (2014) . "Behavior of gr id connected induct ion generator

due to load d is turbances across power network ," Students Conference on Engineer ing

and Systems, Al lahabad, pp. 1 -4, IEEE.

[19] K. Sakai, M. Suzuk i and K. Tak ishima (2017) . "Induct ion machines wi th novel

concentrated windings," 2017 IEEE Internat ional Electr ic Machines and Dr ives

Conference ( IEMDC), Miami, pp. 1 -7, IEEE.

[20] S. Kumar, Ch & S. Kumar (2018) . “A Look Over on Prosperous Power Electronic

Converters for W ind Turbine System , ” pp. 842-847, ResearchGate.

[21] R. Blasco-Gimenez, S. Añó-Vi l lalba, J. Rodríguez-D'Der lée, F. Morant and S.

Bernal-Perez (2010). "Dis tr ibuted Voltage and Frequency Contro l of Of fshore W ind

Farms Connected W ith a Diode-Based HVdc L ink," Transact ions on Power Electronics,

vo l. 25, no. 12, pp. 3095-3105, IEEE.

[22] W . Tong (2010) . “Wind power generat ion and W ind Turbine Des ign . ”

[23] J . Faiz, Z. Val ipour, M. Shokr i-Kojour i , M. Azeem Khan (2016). Design of a Radia l

Flux Permanent Magnet Wind Generator wi th Low Coercive Force Magnets . IEEE.

[24] P. Davar i , F . Zare, A. Abdelhak im (2018) . Act ive Rect i f iers and their Contro l.

Aalborg: Elsevier.

[25] I . David (2017) . “How does a rect i f ier work , ” Sc ienc ing.

[26] D. M. Vi la thgamuwa and S. D. G. Jayas inghe (2012) . "Rect i f ier systems for var iable

speed wind generat ion - a review," 2012 IEEE Internat ional Symposium on Industr ia l

Electronics , Hangzhou , pp. 1058-1065, IEEE.

[27] M.D. Gous iya Begum and M. Kiran Jyoth i (2016) . “Act ive Rect i f ier wi th Integrated

System Contro l for Micro W ind Power Systems , ” Vol . 4, pp. 2487-2493, IJ ITech.

[28] C. Pan and Y. Juan (2010) , "A Novel Sensor less MPPT Control ler for a High -

Ef f ic iency Microscale W ind Power Generat ion System," IEEE Transact ions on Energy

Convers ion, vol . 25, no. 1, pp. 207-216, IEEE.



72

[29] D. X. L lano and R. A. McMahon (2017). "Contro l Techniques wi th System Ef f ic iency

Compar ison for Microwind Turbines," IEEE Transact ions on Susta inable Energy, vol. 8 ,

no. 4, pp. 1609-1617, IEEE.

[30] S. K. T. Mil ler and J. Sun (2006) . "Comparat ive study of three-phase PW M rect i f iers

for wind energy convers ion," Twenty-F irs t Annual IEEE Applied Power Electronics

Conference and Expos it ion, Dallas , pp. 7 , IEEE.

[31] H. Kr ishnamoor thy, M. Danie l , J. Ramos -Ruiz, P. Enjet i , L . L iu and E. Ae lo iza

(2017). "Iso lated AC–DC Conver ter Us ing Medium Frequency Transformer for Of f -Shore

W ind Turbine DC Col lect ion Gr id," IEEE Transact ions on Industr ia l Electronics, vo l . 64,

no. 11, pp. 8939-8947, IEEE.

[32] Davar i , P. Zare, F. Abdelhak im (2018) . “Active Rect i f iers and Their Contro l , ” Contro l

of Power Electronic Conver ters and Systems , Vol . 2, pp. 3-52, Aalborg Univers i ty.

[33] U. Rechkoska-Shikoska, E. Lazarevska, V. Kat ic, J. Shikosk i (2003) . ”Contro l

a lgor i thm of an act ive rect i f ier ,” IFAC Proceedings Volumes, Volume 36, Issue 7, pp.

217-222, Sc ienceDirect.

[34] M. L iserre, A. Del l 'Aqui l a, F. Blaabjerg (2002). "Des ign and contro l of a three-phase

act ive rect i f ier under non- ideal operat ing condit ions," Conference Record of the 2002

IEEE Industry Appl ica t ions Conference, Pit tsburgh, vol .2, pp. 1181-1188, IEEE.

[35] S. Keshavarz (2011) . “Design and Evaluat ion of an Act ive Rect i f ier for a 4.1 MW

Off -Shore W ind Turbine , ” Göteborg, Chalmers Univers i ty of Technology.

[36] Texas Inst rument Europe (1998) . “Field Orientated Control of 3 -Phase AC-Motors . ”

[37] H. Abu-Rub, A. Iqbal , J . Guzinsk i (2012) . “High Performance Contro l of Ac Dr ives

wi th Mat lab/Simul ink Models , ” W iley Intersc ience.

[38] A. Cimpoeru (2010) . “Encoder less Vector Contro l of PMSG for W ind Turbine

Appl icat ions , ” Inst i tu te of Energy Technology AAU.

[39] A.D. Hansen, G. Michalke (2018). “Modell ing and Contro l of Var iable speed Mult i -

pole Permanent Magnet synchronous Generator W ind Turbine , ” W iley Intersc ience.

[40] P. Bouwer (2013) . “Model l ing, Des ign and Implementat ion of a Small -Scale, Pos it ion

Sensor less, Var iable Speed W ind Energy Convers ion System Incorporat ing DTC -SVM of

a PMSG Dr ive wi th RLC Fi l ter , ” Ste l lenbosch Univers i ty.

[41] V. Bobek (2013) . “PMSM Electr ical Parameters Measurement , ” NXP.

[42] K. Lacanette (1991). “A Basic Introduct ion to F i l ters - Act ive, Passive, and Switched-

Capaci tor , ” Nat ional Semiconductor Appl icat ion .

[43] K. Belmokhtar, H. Ibrahim, M.L. Doumbia (2016) . “A Maximum Power Point Track ing

Contro l Algor i thms for a PMSG ‐based W ECS for Iso lated Appl icat ions: Cr it ica l Review , ”

In techOpen.



73

[44] G. But icchi , E. Lorenzani , F. Immovi l l i and C. Bianchin i (2015) . "Act ive Rect i f ier

W ith Integrated System Contro l for Microwind Power Systems," IEEE Transact ions on

Susta inable Energy, vol. 6 , no. 1, pp. 60 -69, IEEE.

[45] P. P. Acarnley and J. F. Watson (2006) . "Review of pos it ion-sensor less operat ion

of brushless permanent -magnet machines," IEEE Transact ions on Industr ia l Elect ronics,

vo l. 53, no. 2, pp. 352-362, IEEE.

[46] Y. Shi , K. Sun, L. Huang and Y. Li (2012). "Onl ine Ident i f icat ion of Permanent

Magnet F lux Based on Extended Kalman F i l ter for IPMSM Dr ive W ith Pos i t ion Sensor less

Contro l, " IEEE Transact ions on Industr ia l Electronics , vo l. 59 , no. 11, pp. 4169-4178,

IEEE.



74

APPENDIX

[A1] Danfoss (2013). “Des ign Guide VLT® AutomationDrive FC 301/302” ,

ht tp:/ / f i les.danfoss.com/download/Dr ives/MG33BF02.pdf

[A2] SEW Eurodr ive (2013). “Operat ing Inst ruct ions Synchronous Servomotors”,

ht tps:/ /www.qimarox.com/media/upload/or ig inal /74/synchrone -cmp-servo-motor-eng-

1419424474.pdf

[A3] ABB (2014) . “Low voltage Motors Motor Guide”,

ht tps :/ /new.abb.com/docs/ l ibrar iesprovider53/about -downloads/ low-vol tage-motor -

guide.pdf

[A4] Lorenz Messtechnik GmbH (2008). “Operat ion Manual for “Plug and Play“ USB

Torque Sensor” , http:/ /www.shintron.com.tw/pro images/a3 -2/090307_DR_3000.pdf

[A5] dSPACE (2017) . “MicroLabBox",

ht tp:/ /www.cokesen.com/resimler /1521101701_Dokuman1.pdf

http://files.danfoss.com/download/Drives/MG33BF02.pdf

https://www.qimarox.com/media/upload/original/74/synchrone-cmp-servo-motor-eng-1419424474.pdf

https://www.qimarox.com/media/upload/original/74/synchrone-cmp-servo-motor-eng-1419424474.pdf

https://new.abb.com/docs/librariesprovider53/about-downloads/low-voltage-motor-guide.pdf

https://new.abb.com/docs/librariesprovider53/about-downloads/low-voltage-motor-guide.pdf

http://www.shintron.com.tw/proimages/a3-2/090307_DR_3000.pdf

http://www.cokesen.com/resimler/1521101701_Dokuman1.pdf

voltage control of direct-drive wind turbines for …el-khatib, ahmad jessen, nicholas wachenfeld...

Documents