SCIG With Full Power Converter - Book2sides

Modeling and control of Squirrel Cage Induction
Generator with Full Power Converter applied to

windmills
Domı́nguez Garcı́a, José Luis
November 30, 2009

You can not prevent the wind. But windmills can
be built.
(Dutch quote)
The pessimist complains about the wind; the op-

timist expects it to change; the realist adjusts the
sails.
(William Arthur Ward)
If a man knows not what harbor he seeks, any

wind is the right wind.
(Lucius Annaeus Seneca)
SCIG with Full Power Converter 3

applied to windmills
4 SCIG with Full Power Converter
Contents
Abstract 9
Glossary 11
Foreword 17
Aim 19
Reach 21
Introduction 23
1 System’s Analysis 25
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.2 Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.3 Gear Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.4 Squirrel Cage Induction Generator . . . . . . . . . . . . . . . . . . . . . . . . 28
1.5 Full Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.5.1 DC Bus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.5.2 Crow Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.6 Grid connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2 Generator system 35
2.1 Change of abc basis to qd0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.2 qd0 Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.2 Park Transform application on Induction machine equations . . . . . . . . . . 37
2.2.1 Voltage equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.2.2 Torque equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3 State space of SCIG generator . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3 Control system design 43
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2 High level control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.2.1 Constant tip speed ratio scheme control . . . . . . . . . . . . . . . . . 44
3.3 Control of the converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.1 Control of the electrical system of the stator . . . . . . . . . . . . . . 45
3.3.2 Control of the electrical system of the grid side . . . . . . . . . . . . . 49
4 Modeling of the system 53

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Modeling of the Mechanical System . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Modeling of SCIG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Modeling of Full Power Converter . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4.1 Stator Side Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4.2 DC Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4.3 Grid Side Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.5 Modeling of the Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.5.1 Grid Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.5.2 Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5 Simulation of the model and Analysis of the results 57

5.1 Generator connected directly to the grid . . . . . . . . . . . . . . . . . . . . . 57
5.2 Complete system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6 Conclusions 61
A Controllability and Observability of the SCIG Linear System 65

A.1 Controllability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
A.2 Observability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
A.3 A lower bound on the distance to a non-controllable system . . . . . . . . . . 67
A.4 A lower bound on the distance to a non-observable system . . . . . . . . . . . 68
B Space Vector Pulse Width Modulation (SVPWM) 71
C Introduction to current loops 73
D Definition of magnetizing current 75
Bibliography 78

List of Figures
1 Location of windfarms in Finland.[1] . . . . . . . . . . . . . . . . . . . . . . . 17
1.1 Generic Nacelle of a Windmill . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

1.2 Windmill. Source www.winwind.fi . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.3 Illustration of the Area which provide us of the wind power. . . . . . . . . . . 27
1.4 GearBox Input & Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.5 Simplification of the GearBox . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.6 Induction machine’s schematic illustration . . . . . . . . . . . . . . . . . . . . 28
1.7 Full converter illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.8 Simplification of the grid connection . . . . . . . . . . . . . . . . . . . . . . . 33
2.1 Any current vector in abc reference . . . . . . . . . . . . . . . . . . . . . . . . 35

2.2 Representation of the Clarke Transformation . . . . . . . . . . . . . . . . . . 36
2.3 Rotation of the reference (α, β) an angle θ . . . . . . . . . . . . . . . . . . . 36
3.1 Representation of the control system . . . . . . . . . . . . . . . . . . . . . . . 43

3.2 Schematic of the IMC scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Alternative structure of the IMC scheme . . . . . . . . . . . . . . . . . . . . . 46
3.4 Block diagram of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5 Block diagram of the system of the BUS control . . . . . . . . . . . . . . . . 51
4.1 Simulinkr model of the whole windmill . . . . . . . . . . . . . . . . . . . . . 53

4.2 Simulinkr model of the mechanical part . . . . . . . . . . . . . . . . . . . . . 54
4.3 Simulinkr model of the generator . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4 Simulinkr model of the Full Power Converter . . . . . . . . . . . . . . . . . . 54
4.5 Simulinkr model of the Stator Side of the Back-to-Back . . . . . . . . . . . . 54
4.6 Simulinkr model of the DC Link of the converter . . . . . . . . . . . . . . . . 54
4.7 Simulinkr model of the grid side of the frequency converter . . . . . . . . . . 55
4.8 Simulinkr model of the grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.9 Simulinkr model of the grid filter . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.10 Simulinkr model of the grid voltage generation. . . . . . . . . . . . . . . . . . 55
5.1 Mechanical Torque produced by the Generator . . . . . . . . . . . . . . . . . 57

5.2 Stator currents in qd reference . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3 Evolution of Cp(λ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.4 Graphic of characteristic Cp of each turbine. . . . . . . . . . . . . . . . . . . 58
5.5 Graphic of the evolution of the grid filter voltage . . . . . . . . . . . . . . . . 58
5.6 Voltage E of the DC Link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.7 Stator voltage in qd reference . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.8 Stator current in qd reference . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.9 Mechanical Torque produced by the Generator . . . . . . . . . . . . . . . . . 58
5.10 Park Transform of the voltage of the grid . . . . . . . . . . . . . . . . . . . . 59
B.1 Electrical scheme of an inverter . . . . . . . . . . . . . . . . . . . . . . . . . . 71

B.2 Commutation hexagon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

Abstract
The project aims to develop a dynamic model, of a generation system of electrical energy with
a variable speed wind turbine using a squirrel cage induction generator which is connected
to the grid by a back to back frequency converter, for testing purposes.
In the project has been done an analysis of the mathematical equations of the whole system,
separating the mechanical system (turbine and gearbox) and the electrical (generator and
converter).
We study the control design of the machine which allow us to obtain the maximum wind
power extraction, and we also study the control of the converter which couple the generator
with the network.
The proper functioning of the model is checked comparing the obtained results with some
commonly known results.

Glossary
Symbols
ṁ mass flow
vw wind speed
A area where the air flow could pass through
vw1 wind speed before the turbine
vw2 wind speed after the turbine
Pwind mechanical power extracted by the converter
Pwind0 mechanical power that could be converted
cp Power coefficient
ci set of values greater or equal than zero, these are known as turbine’s characteristic
coefficients
R radius of the turbine, that means, the length of blades
c Viscosity constant of the turbine’s shaft
k Elasticity constant of the turbine’s shaft
Jm Inertia moment of the generator
Jt Inertia moment of the turbine
vsabc stator winding’s voltage vector
vrabc rotor winding’s voltage vector
iabc
s stator winding’s current vector
iabc
r rotor winding’s current vector

rs resistance of the stator windings
rr resistance of the rotor windings
Lss self-inductance of the stator windings without the winding owing the dispersion flow
Lrr self-inductance of the rotor windings without the winding owing the dispersion flow
Lsm coupling inductances between stator windings
Lrm coupling inductances between rotor windings
Lsr maximum value reached by coupling inductances between stator and rotor windings
Lls dispersion inductance of the stator windings
Llr dispersion inductance of the rotor windings
L(θr ) Induction machine’s coupling inductance matrix
P number of poles that the Induction Machine has
E0 DC Bus voltage at t=0
C DC Bus capacity
iDCE Current which flows through the condenser
iDCg Current which came from grid’s converter
iDCs Current which came from stator’s converter
Z Value of the impedance
R Value of the resistance parameter
L Value of the inductance parameter
l Value of the length of the wire from converter to electrical transformer
Xccpu Value of the short circuit inductance in PU reference
V1b Base voltage in the primary
V1n Nominal voltage in the primary
Sb Base power
Sn Nominal power

xqd0 xabc after apply Park Transform on it
T (θ) Park Transform
vsqd0 stator winding’s voltage vector in qd0 reference
vrqd0 rotor winding’s voltage vector in qd0 reference
iqd0
s stator winding’s current vector in qd0 reference
iqd0
r rotor winding’s current vector in qd0 reference
C(s) IMC controller.
G(s) Induction Generator’s transfer function.
G(s) Internal model of the Induction Generator’s transfer function.
I Identity or unit matrix
im Magnetizing current
Kcp |θpitch =0 Optimum Torque coefficient of the turbine

Greek Symbols
ρ air density
ωt turbine’s spin speed
λ tip speed ratio
θpitch spin blade angle
θm Orientation angle of the motor’s shaft
θt Orientation angle of turbine
ν Gear ratio of the multiplier
τm Mechanical torque of the engine.
τt Mechanical torque of the turbine
λabc
s stator winding’s concatenated flows vector
λabc
r rotor winding’s concatenated flows vector
ωr generator’s shaft’s orientation angle from electric system
Γr Torque on the rotor shaft
Γm Mechanical Torque of the machine
εcc short circuit voltage
α Controller’s bandwidth in close loop

Acronyms
DFIG Double Fed Induction Generator
SCIG Squirrel Cage Induction Generator
IGBT Insulated Gate Bipolar Transistor
SVPWM Space Vector Pulse Width Modulation
PU Per Unit
MPPT Maximum Power Point Tracking
IMC Internal Model Control
PI Proportional and Integer controller
LTI Linear Time-Invariant

Foreword
The wind power capacity in Finland is 144 MW, 119 wind turbines (May 2009).Wind
power production in 2008 was about 260 GWh which is 0.3% of the Finnish electricity
consumption,[1]. The target of the Finnish Government is to have 2000 MW of installed
wind energy capacity by 2020,[2].
The 90% of Finns would want further investments in wind energy in September 2007,[3].
Therefore there are a lot of ongoing projects of building new windmills and the public desire
of improvement, these are slow because they are too much expensive, and this deceleration
of the construction of the windmills implies do not reach the wind energy installation tar-
get.This fact causes several studies to improve the existing windmills.
This project is thought for to get a dynamic simulation model which allow us to study if
a squirrel cage induction generator with a full power converter connected to its stator side
works better than the conventional control systems.
Figure 1: Location of windfarms in Finland.[1]

Aim
The target of this project is to obtain a dynamic model of the windmill, which is equipped
with a squirrel cage induction generator (SCIG) and a full power converter to do the con-
trol. Moreover, this dynamic model will allow us make some simulations and understand
the behavior of the generator with this kind of a control. Finally, with the results of the
simulations we will be able to conclude if it is good enough and how it works.
Then, we can resume briefly that aim of the project as ” Provide a simulation model of
our SCIG windmill which permit us to test it”.

Reach
This project could be applied to reduce manufacturing costs of windmills, changing the gen-
erator typology.
The most used generator is the double feed induction generator (DFIG), and we suggest
to change it by squirrel cage induction generator (SCIG), owing to the robustness of SCIG,
and since it has low cost with an almost null maintenance.
Thanks to this, we are able to reduce the generator expenses and the rest of the costs
are the same because we are coupling the same full converter and applying the same control
what is to be used in DFIG.

Introduction
The windmill, also called as the wind turbine, is a means of exploitation the kinetic energy
of the wind and converting it into electrical energy using an electrical generator.
The windmills can be divided in two essential parts:
I. How we obtain the wind energy. To realize this task we have to use a turbine.
There are two options for the turbine, vertical axis and horizontal axis. We will choose
for our study the horizontal turbine, because this the type which is used more often.
II. How we convert kinetic energy into electrical energy. To make this, there
exist two different approaches, Fixed speed and Variable Speed, where each one im-
plies quite different configurations. We will select for our project the variable speed,
because, again, is the most used method.
In this project we use a squirrel cage induction generator, despite this type of a generator,
usually, is applied by the fixed speed windmills, which is directly connected to the grid,
or can include a condenser between the generator and the grid to compensate the reactive
power.
In our case, we will introduce a full power converter between generator and the grid, thanks
to this we will use the variable speed windmill. The full power converter allow us control
the generator.
It is important to remark that the squirrel cage is rather common as a motor. Therefore,
the control in this way is quite well studied and it also implies a robust machine that need
little, or almost null, maintenance.

Chapter 1
System’s Analysis
1.1 Introduction
In this chapter we will be present the equations that describe each part of our windmill.
This windmill is a variable speed turbine with squirrel cage induction generator with a vec-
tor control.
The windmill could be divided as follows:
Figure 1.1: Generic Nacelle of a Windmill

1.2 Turbine
A windmill is a machine responsible for transforming the kinetic energy of the wind into the
electrical energy. If the wind flow is known, the kinetic power could be expressed, as it is
well known, by means the follow equation:
1 1
Pwind = ṁv 2 = ρAvw
3
(1.1)
2 w 2
where
ṁ mass flow
vw wind speed
ρ air density
A area where the air flow could pass through
Figure 1.2: Windmill. Source www.winwind.fi
The power described in (1.1) is just a mathematical description about kinetic power. How-
ever, this power could not be obtained by a wind turbine. The power that could be achieved
in the best situation, is 0.593 times Pwind . This value is called as the Betz number. It is a
power coefficient and the turbine efficiency limit, for more details how to obtain the Betz
number see [4].
The Power coefficient can be defined as the ratio between the mechanical power extracted
by the converter and the power of the undisturbed air stream (see [5]).
Pwind
cp = Pwind0
2 2
1
4 ρA (vw1 −vw2 )(vw1 +vw2 )
= 1 3
2 ρAvw1
(1.2)
2
= 21 1 − vvw1
w2
1+ vw2
vw1
with

vw1 wind speed before the turbine
vw2 wind speed after the turbine
ρ air density
A area where the air flow could pass through. as could be seen in (1.3)
Pwind mechanical power extracted by the converter
Pwind0 mechanical power that could be converted
1
It should be noted that the quotient between vw2 and vw1 never can overcome 3 because of
the Betz number.
The kinetic power obtained by the turbine can than be defined as:
Pwind = cp Pwind0 (1.3)
where cp is defined in (1.2). However, this coefficient depends directly on each turbine, on
the tip speed ratio λ, which is defined below in (1.5) and, just in the case that the rotor is
equipped with blade pitch control, on the θpitch called pitch. The coefficient’s value can be
found on tables for some specific turbines or determined by analytic function, as follows [6].

1 1
c5
cp (λ, θpitch ) = c1 c2 − c3 θpitch − c4 θpitch − c6 e−c7 Λ (1.4)
Λ
with
ωt R
λ= (1.5)
vw1
1 1 c9
= − 3 (1.6)
Λ λ + c8 θpitch 1 + θpitch
with
ci set of values greater or equal than zero, these are known as turbine’s characteristic
coefficients.
ωt turbine’s spin speed.
R radius of the turbine, that means, the length of blades.
1.3 Gear Box

A transmission or gearbox provides speed and torque conversions from a rotating power
source to another device using gear ratios.
The gearbox in a wind turbine converts the slow, high-torque rotation of the turbine into
much faster rotation of the electrical generator. Usually, it contains three stages to achieve
an overall gear ratio from 40:1 to over 100:1, depending on the size of the turbine. The
first stage of the gearbox is usually a planetary gear, for compactness, and to distribute the
enormous torque of the turbine over more teeth of the low-speed shaft.
The gearbox will be described as a black box that receive a low speed and transforms it to

Figure 1.3: Illustration of the Area which provide us of the wind power.
Figure 1.4: GearBox Input & Output
a faster rotation, as is shown in the illustration (1.4).

There are two ways that are commonly used for the modeling of the gearbox [7]:
1. The Lumped model, which assumes that all the rotating masses can be treated as
one concentrated mass.

2. The Two-mass model, which considers an equivalent system with an equivalent stiff-
ness and damping factor on the wind turbine on the rotor side.
We will use the Two-mass model. Since, if we want to use the model to realize the transient
analysis, Lumped model will give us wrong results. Moreover, it is necessary to remark that
the inertia moment of the turbine is, almost 90% of the inertia of all the drive train (Turbine
and GearBox), and it is known the high efficiency of the transmission. Then, we are able to
neglect friction torque and, only, consider the inertia moment of the turbine and the inertia
of the rotor’s generator. Finally, we get the following equation to describe our model that
could be written as (1.5).
   ν2c νc ν2k νk
   1 
 θ̈m  − Jm J − J J  θ̇ m  Jm 0
  m m m   1 
  νc
θ̈t c νk k   0 τ
− − θ̇
  
t J m
=  Jt Jt Jt Jt  +  t  (1.7)
θ̇   1 0 0 0  θ   0 0 τt
 m   m 

 
θt 0 0
 
θ̇t 0 1 0 0
where
θm Orientation angle of the motor’s shaft.
θt Orientation angle of turbine.
ν Gear ratio of the multiplier.
c Viscosity constant of the turbine’s shaft.
k Elasticity constant of the turbine’s shaft.
Jm Inertia moment of the generator.
Jt Inertia moment of the turbine.
τm Mechanical torque of the engine.
τt Mechanical torque of the turbine.
Figure 1.5: Simplification of the GearBox

1.4 Squirrel Cage Induction Generator
Three-phase induction machines have three windings in the stator and three windings more
in the rotor, although, these can be real or imaginary.
As it is known, all electrical machines can be described as motor and generator as well,
consequently, they can be described with the same set of equations. It is appropriate to
remember that these equations govern the operation of the electrical machines. These equa-
tions are divided in two groups, Voltage equations and Torque equations in machine variables
and other which are expressed in the axes of the reference variables.
With the goal of simplifying these equations, it is common in the technical literature, e.g.
[8], to consider the following hypothesis:
• Symmetric and balanced three-phase induction machine, with a single winding rotor
(Squirrel cage simple) and constant gap.
• Material is assumed to be linear, that is to say, the iron saturation is discarded.
• The iron magnetic permeability is assumed to be infinite in front of the air permeabil-
ity, which means that the magnetic flux density is radial to the gap.
• All kind of losses in the iron are neglected.
• Both the stator windings as the rotor windings represent distributed windings which
always generate a sinusoidal magnetic field distribution in the gap.
All hypothesis that we have explained before, using the induction motor’s illustration, guide
us to the following system of equations which describe the dynamic behavior of the induction
machine.
abc abc abc abc
vs rs 0 is d λs
= + (1.8)
vrabc 0 rrabc iabc
r dt λabc
r
with
vsabc stator winding’s voltage vector
vrabc rotor winding’s voltage vector
iabc
s stator winding’s current vector
iabc
r rotor winding’s current vector
λabc
s stator winding’s concatenated flows vector
λabc
r rotor winding’s concatenated flows vector
The relationship between concatenated flows, rotor and stator’s current is given by
abc abc
Lss Labc
abc
λs sr is
= (1.9)
λabc
r Labc
rs Lrr iabc
r
where each term represents a 3-dimensional matrix or a three-dimensional vector. Then, the

Figure 1.6: Induction machine’s schematic illustration
vectors can be written as

       
vsa vra isa ira
vs =  vsb  , vr =  vrb  , is =  isb  , ir =  irb 
vsc vrc isc irc
and impedance matrices as

 
rs 0 0
rsabc = 0 rs 0  (1.10)
0 0 rs
 
rr 0 0
rrabc = 0 rr 0  (1.11)
0 0 rr
 
Lls + Lss Lsm Lsm
Labc
ss = Lsm Lls + Lss Lsm  (1.12)
Lsm Lsm Lls + Lss

cos(θr + 2π 2π
 
abc t cos(θr ) 3 ) cos(θr − 3 )
Labc
sr = Lrs = Lsr  cos(θr − 2π 3 ) cos(θr ) cos(θr + 2π 3 )
 (1.13)
2π 2π
cos(θr + 3 ) cos(θr − 3 ) cos(θr )
 
Llr + Lrr Lrm Lrm
Labc
rr = Lrm Llr + Lrr Lrm  (1.14)
Lrm Lrm Llr + Lrr
with:
ωr generator’s shaft’s orientation angle from electric system
rs resistance of the stator windings
rr resistance of the rotor windings
Lss self-inductance of the stator windings without the winding owing the dispersion flow
Lrr self-inductance of the rotor windings without the winding owing the dispersion flow
Lsm coupling inductances between stator windings
Lrm coupling inductances between rotor windings
Lsr maximum value reached by coupling inductances between stator and rotor windings
Lls dispersion inductance of the stator windings
Llr dispersion inductance of the rotor windings
The electromechanical conversion theory provides the following equation:
1 t δ[L(θr )]
Γr = [i] [i] (1.15)
2 δ(θr )
where
Γr Torque on the rotor shaft
Lss Lsr
L(θr ) Induction machine’s coupling inductance matrix. L(θr ) =
Lrs Lrr
Usually the induction machines are designed with number of poles over 1. Theoretically,
this could be understood as a ideal multiplier with a transmission ratio P between shaft’s
mechanical angle (θm ) and electrical system’s angle.
P t δ[L(θr )]
Γm = [i] [i] (1.16)
2 δ(θr )
with
P number of poles for the Induction Machine.

Without the loss of generality we may suppose the number of poles is exactly one. However,
we want to remark the results obtained generalize to the situation of multiple poles.
The equation (1.16) expresses the torque developed by the induction machine at any time,
depending on the instantaneous currents circulating by each one of the six windings, and
the separation angle between the stator winding 1 and the rotor winding 1. This equation
is obtained by the electrical system energy balance.
Developing the equation (1.15), it is easily simplified due to Lss and Lrr are not θr depen-
dent. Then the derivative of this constant vanishes. So the new equation can be written as
t
iabc iabc

1 s 0 Nsr s
Γr = (1.17)
2 iabc
r Nrs 0 iabc
r
where
sin(θr + 2π 2π
 
sin(θr ) 3 ) sin(θr − 3 )
abc abc t
= −Lsr  sin(θr − 2π sin(θr + 2π

Nsr = Nrs 3 ) sin(θr ) 3 ) (1.18)

2π 2π
sin(θr + 3 ) sin(θr − 3 ) sin(θr )
Also it is possible to make the same operation, finding the mechanical torque produced by
the generator. This can be achieved just by multiplying by the transmission ratio P . Then,
we obtain the following equation:
t
iabc iabc

P s 0 Nsr s
Γm = (1.19)
2 iabc
r Nrs 0 iabc
r
The generator which we will be the studied object is a SCIG (Squirrel Cage Induction Gen-
erator), this type of generator is known also as Short-Circuit Induction Generator, owing
to rotor windings are then connected in short−circuit. So, the only part of the generator
connected to the grid will be the stator. Because of this connection to the rotor windings,
we are able to get the next simplification vrabc = 0. Then, the equation described before
(1.17) could be written as follows:
abc abc abc abc
vs rs 0 is d λs
= + (1.20)
0abc 0 rrabc iabc
r dt λabc
r
In the next chapter, we will be describe the most common way to work with these equations
and how to obtain them.
1.5 Full Converter

If the induction generator is connected directly to the grid, then a capacitor bank is connected
in stand−alone systems to the stator to provide the magnetizing current for the reactive
power. However, for grid connected systems the reactive power is drawn from the grid.
Induction generators also are able to use back to back converters. Then, this fact let us make

several configurations. Therefore it is a good idea to couple this converter to SCIG, because
this decoupling of the generator and the grid allows us to work with different frequencies on
each side. This leads us to the advantage to change the usual operation typology of SCIG,
because as we are able to work with different frequencies, is not necessary to be worried
about the rotor speed. Then we will study a variable speed windmill.
As was said before, to take advantage of the variable speed operation a power electronic
interface must be provided between the machine terminals and the grid. The back to back
converter is a suitable option for Cage induction machine in wind Power application. The
back to back is formed by a rectifier, an inverter and a condenser between both. The rectifier
and the inverter, are composed by IGBT and these are controlled by Space Vector Pulse
Width Modulation (SVPWM)1
Figure 1.7: Full converter illustration
1.5.1 DC Bus
DC Bus is connected between rotor’s converter and grid’s converter. DC Bus’ voltage V has
the following expression:
Z t Z t
1 1
E = E0 + iDCE dt = E0 + (iDCg − iDCs )dt (1.21)
C 0 C 0
with
E0 DC Bus voltage at t=0
C DC Bus capacity
iDCE Current which flows through the condenser
iDCg Current which came from grid’s converter
iDCs Current which came from stator’s converter
1.5.2 Crow Bar
The Crow Bar is a protection circuit used to prevent an overvoltage of a power supply unit
from destroying the IGBTs and diodes of the converter.
1 The description of this technique is described in the appendix B

The Crow Bar operates, putting a short circuit or low resistance path across the voltage
source, when excessive currents or voltages are detected. It means that when the power is
too big and the converter is not able to absorb it all. It usually happens in voltage sags.
Crow Bar will not be included in our model because it is not the aim of the project to study
the behavior of the generator during a sag.
1.6 Grid connection

The grid connection will be described just until the electrical transformer, because after
these the rest is part of the wind farm.
The cable which are between the converter and the electrical transformer could be called
”short cable” because the length of the wire is shorter than 50 km. [9] Then, it is possible
make to the simplification:
• Neglect the effect of capacity and focus the effect of their parameters in a series
impedance equal to the total impedance of the line.
Z t = (R + jωL)l (1.22)
where
Z Value of the impedance.
R Value of the resistance parameter.
L Value of the inductance parameter.
l Value of the length of the wire from converter to electrical transformer.
Then, the electrical circuit which connects the grid with the grid side converter can be de-
scribed by the Ohm law (∆V = Z · I). Applying this equation and the impedance of the
line, the electrical circuit can be written as:
d abc
vzabc − vlabc = rl iabc
l + Ll i (1.23)
dt l
The Electrical transformer will be described with just an inductance. It can be approx-
imated in this way in the PU (per unit) reference, and the inductance has the following
equation:
V1n 2 Sb

xccpu = εcc (1.24)
V1b Sn
where
εcc short circuit voltage.
Xccpu Value of the short circuit inductance in PU reference.
V1b Base voltage in the primary.
V1n Nominal voltage in the primary.
Sb Base power.
Sn Nominal power.

Figure 1.8: Simplification of the grid connection

Chapter 2
Generator system
On electrical system studies, it is usual to utilize mathematical transform of variables that

are used to remove the time dependence of some parameters.
2.1 Change of abc basis to qd0

Quadrature−direct−zero (qd0) transformation is a mathematical transformation, used to
simplify the analysis of three-phase circuits. In the case of balanced three-phase circuits,
application of the qd0 transform reduces the three AC quantities to two DC quantities.
Simplified calculations can, then, be carried out on these imaginary DC quantities before,
performing the inverse transform to recover the actual three-phase AC results. It is often
used in order to simplify the analysis of three-phase synchronous machines, or to simplify
calculations for the control of three-phase inverters.
2.1.1 Introduction
The main goal of this introduction to give a brief resume of how to reach the qd0 transfor-
mation that will be used in rest of the document.
Mathematical theory
Let A, B be two square matrices, if these matrices could be describe as T (A) = S −1 AS it is

called Similarity Transformation, and S is known as the change of basis matrix. The change
of basis matrix can be obtained by the eigenvectors of A.
Clarke Transformation
Clarke Transformation allow us to change the 3-dim system to 2-dim system (abc to αβ).
For example, the current vector that is shown in (2.1).

Figure 2.1: Any current vector in abc reference
 
1  a 
− −√21

α 2 1 √2
= 3 b (2.1)
β 3 0 2 − 23 
c

Figure 2.2: Representation of the Clarke Transformation

Park Transformation
The Park Transformation describes a rotation of an orthogonal system ( (α, β) to (q, d)).

q −sin(θf ield ) cos(θf ield ) α
= (2.2)
d cos(θf ield ) sin(θf ield ) β
Figure 2.3: Rotation of the reference (α, β) an angle θ
2.1.2 qd0 Transformation

1
For the calculations, the
√ Park Transform be used which is proportional to Direct−qua-
drature−zero with a 3 ratio. Because of this, it is usual to describe the Park Transform
as qd0 too. These equations are the ”shortcut” for the Clark and Park Transformations,
instead of use both matrices with just one is enough.
The Park Transformation with an angle θ of a vector either (xabc R) is defined as:
xqd0 = T (θ)xabc (2.3)
with
2π 2π
 
cos(θ) cos(θ − 3 ) cos(θ + 3 )
2 2π 2π
T (θ) = sin(θ) sin(θ − 3 ) sin(θ + 3 )
 (2.4)
3 1 1 1
2 2 2
1 Throughout the document, qd0 Transformation will be called Park Transform, although it could be
known as Clarke-Park Transformation

Then could be said that the vector xqd0 is the vector xabc in dq0 on the angle θ reference.
As a change of basis matrix T (θ) is invertible. So, it’s true that:
xabc = T −1 (θ)xqd0 (2.5)

with
 
cos(θ) sin(θ) 1
T −1 (θ) =  cos(θ − 2π3 ) sin(θ − 2π3 ) 1  (2.6)
cos(θ + 3 ) sin(θ + 2π
2π
3 ) 1
Basis change qd0 to other reference in qd0 variables
From one vector xabc could be obtained xqd0 using the Park Transformation with the θ, and
from the same vector could be calculated x̂qd0 using the Park Transformation with the θ̂.
Then, the relationship between xqd0 and x̂qd0 is:
xqd0 = P (θ − θ̂)x̂qd0 (2.7)
where:
 
cos(θ − θ̂) −sin(θ − θ̂) 0
P (θ − θ̂) =  sin(θ − θ̂) cos(θ − θ̂) 0  (2.8)
0 0 1
2.2 Park Transform application on Induction machine

equations
To be able to apply Park Transform to Induction machine equations, it is necessary to define
a matrix which includes 2 matrices of Park. The first one transforms the stator variables
and the second converts the rotor variables, both into synchronous reference.

T (θ) 0
T (θ, θ − θr ) = (2.9)
0 T (θ − θr )
2.2.1 Voltage equations
From the equation (1.8) is known that:

d
vsabc = rsabc iabc Labc iabc + Labc abc

s + sr ir (2.10a)
dt ss s
d
vrabc = rrabc iabc Labc iabc + Labc abc

r + rr ir (2.10b)
dt rs s

Applying the Park Transform to (2.10a) and (2.10b) and defining some parameters
Ls = Lss − Lsm + Lls
Lr = Lrr − Lrm + Llr (2.11)
3
M= Lsr
2
the equation is:
vsqd0 θ̇[Y ]Lqd0 qd0

θ̇[Y ]Lqd0 iqd0

ss + rs sr s
=
vrqd0 qd0
(θ̇ − θr )[Y ]Lrs (θ̇ − θr )[Y ]Lqd0 qd0
rr + rr iqd0
r
(2.12)
Lqd0 Lqd0 iqd0

ss sr d s
+
Lqd0
rs Lqd0
rr
dt iqd0
r
where
 
rs 0 0
−1
rsqd0 = Tqd0 (θ)rsabc Tqd0 (θ) =  0 rs 0  (2.13a)
0 0 rs
 
rr 0 0
−1
rrqd0 = Tqd0 (θ − θr )rrabc Tqd0 (θ − θr ) =  0 rr 0  (2.13b)
0 0 rr
 
Ls 0 0
−1
Lqd0
ss = Tqd0 (θ)Labc T
ss qd0 (θ) =  0 Ls 0  (2.13c)
0 0 Lss + 2Lsm + Lls
 
M 0 0
−1
Lqd0
sr = Tqd0 (θ)Labc T
sr qd0 (θ − θr ) =  0 M 0  (2.13d)
0 0 M
 
M 0 0
−1
Lqd0
rs = Tqd0 (θ − θr )Labc T
rs qd0 (θ) =  0 M 0  (2.13e)
0 0 M
 
Lr 0 0
−1
Lqd0
rr = Tqd0 (θ − θr )Labc T
rr qd0 (θ − θ r ) =  0 Lr 0  (2.13f)
0 0 Lrr + 2Lrm + Llr
 
0 1 0
Y =  −1 0 0  (2.13g)
0 0 0
By the development of this expression, we are able to get an equation of voltage where the
component 0 only depends on the current 0 and its derivatives. Meanwhile, voltages q and
d are depend only on the currents and their derivatives. Therefore, they could be expressed

as follows:
     

 vsq 
 Ls 0 M 0 
 isq 

vsd  0 Ls 0 M  isd
   
=   d
 vrq
 

 M 0 Lr 0  dt 
 irq 

vrd 0 M 0 Lr ird
   
 
rs Ls θ̇ 0 M θ̇  
isq
−Ls θ̇ rs −M θ̇ 0

  

i
 
sd
 
+
 0 M θ̇ − θ̇r rr Lr θ̇ − θ̇r   irq

  

ird
 
−M θ̇ − θ̇r 0 −Lr θ̇ − θ̇r rr
(2.14a)
dis0
vs0 = (Lss + 2Lsm + Lls ) + rs is0 (2.14b)
dt
dir0
vr0 = (Lrr + 2Lrm + Llr ) + rr ir0 (2.14c)
dt
Applying the same hypothesis as in the equation (1.20), it is possible for us to write (2.14a)
as follows:
     

 vsq 
 Ls 0 M 0 
 isq 

vsd  0 Ls 0 M  isd
   
=   d
 0
 

 M 0 Lr 0  dt 
 irq 

0 0 M 0 Lr ird
   
 
rs Ls θ̇ 0 M θ̇  
isq
−Ls θ̇ rs −M θ̇ 0

  

i
 
sd
 
+
 0 M θ̇ − θ̇r rr Lr θ̇ − θ̇r   irq

  

ird
 
−M θ̇ − θ̇r 0 −Lr θ̇ − θ̇r rr
(2.15)
2.2.2 Torque equations
Taking into account the equation (1.19):

t
iabc abc
iabc

1 s 0 Nsr s
Γm = 2 iabc
r
abc
Nrs 0 iabc
r
(2.16)
t abc abc abc t abc abc
1
= 2 iabc
s Nsr ir + ir Nrs is
and applying the Park Transform on it, we will obtain the following expression:
3
Γm = M (isq ird − isd irq ) (2.17)
2

2.3 State space of SCIG generator
The dynamic behavior of the machine can be studied by means of a dynamic linear system, as
we saw in the equation (2.15). Although, most adequate mathematical expression to realize

the system simulation is the state space, which usually is expressed as Ẋ = A · X + B · U .
Then, by the reordering of the equations (2.15), (2.14b),(2.14c) we are able to obtain our
goal with just the few next steps:
First, it is necessary to put on the different side of the equal sign derivative variables and
non-derivative.
   
" #  isq   vsq 
Ls 0 M 0    
isd vsd
   
0 Ls 0 M d
M 0 Lr 0 dt  i =
0 M 0 Lr  rq 
 
 0 

ird 0
   
 
 rs Ls θ̇ 0 M θ̇
 
 isq 

−Ls θ̇ −M θ̇ isd
rs 0
 
− 0 M (θ̇−θ̇r ) rr Lr (θ̇−θ̇r )

 irq 
−M (θ̇−θ̇r ) −Lr (θ̇−θ̇r )
0 rr
 
ird
 
(2.18)
Second, we have to determine the inverse of the matrix which is with the derivative. It is
needed because of A−1 ∗ A = I.
 −1  
Ls 0 M 0 Lr 0 −M 0
0 Ls 0 M  = 1  0 Lr 0 −M 

M

−M
 (2.19)
0 Lr 0 Ls Lr − M 2 0 Ls 0 
0 M 0 Lr 0 −M 0 Ls
Third, to leave only the derivative variables without multiplying constants, we should mul-
tiply the inverse in both sides.
   
 isq  " # rs Ls θ̇ 0 M θ̇
  isq 
  Lr 0 −M 0  
i −Ls θ̇ −M θ̇ isd
  rs 0
 
d sd 1 0 Lr 0 −M
dt  i = − Ls Lr −M 2 −M 0 Ls 0
 0 M (θ̇−θ̇r ) rr Lr (θ̇−θ̇r )

rq  0 −M 0 Ls
 irq 
−M (θ̇−θ̇r ) −Lr (θ̇−θ̇r )
  0 rr
 
ird ird
   
 
" # vsq 
Lr 0 −M 0  
v
 
0 Lr 0 −M sd
+ Ls Lr1−M 2 −M 0 Ls 0
0 −M 0 Ls
 0
 

0
 
(2.20)
Finally, we have to multiply the matrices and discard unnecessary elements of the system.
Then, we get the following equation:

 
 isq
 

d  isd 
=
 irq
dt  

ird
 
| {z }
Ẋ
 
isq 
M 2 θ˙r +(Ls Lr −M 2 )θ̇ M Lr θ˙r

Lr rs −M rr 
 
1 2 ˙ 2
 −M θr −(Ls Lr −M )θ̇ L r rs −M Lr θ˙r −M rr  isd

− −M Ls θ˙r (Ls Lr −M 2 )θ̇−Ls Lr θ˙r
Ls Lr − M2 −M rs Ls rr  irq 
M Ls θ˙r −(Ls Lr −M 2 )θ̇−Ls Lr θ˙r

 
−M rs Ls rr i

| {z } rd
A
| {z }
X
 
" # vsq 
Lr 0  

1 0 Lr vsd

+ −M 0
Ls Lr − M 2
0 −M 
 0  
| {z } 0 
B | {z }
U
(2.21a)
2
dis0 rs 1
= is0 − vs0 (2.21b)
dt Lss + 2Lsm + Lls Lss + 2Lsm + Lls
dir0 rr 1
= ir0 − vr0 (2.21c)
dt Lrr + 2Lrm + Llr Lrr + 2Lrm + Llr
2 In the appendix A will be studied the controllability and the observability of the system.

Chapter 3
Control system design
3.1 Introduction
Control plays a very important role in modern wind energy conversion systems. In fact,
the principle target of control is to enable us to obtain as much quantity of energy from
the wind as is possible during certain weather conditions, and deliver it to the grid at the
best possible conditions. As well as, to keep the system on a safe working area to reduce
aerodynamic and mechanical loads, and to protect the electrical devices, for example, pitch
controlled that spin the blades in his own axis to regulate the position of it, or stall controlled
which makes us loose aerodynamical power, this type of control will have constant θpitch .
Additionally, the control can be used to perform as reactive power suppliers or consumers
according to the power system requirements.
The control can be divided in 2 levels, high level control system and control of the converter
system.
3.2 High level control
Usually in the windmills there are distinguished two different work regions: partial load and
total load.
When the windmill are working in total load, the high level control system keep the nominal
rotation regime and it orients the blades to extract the nominal power.
Instead, when the windmill is working on partial load the control keep the orientation of the
blades and it regulates the rotational speed to get the maximum power that is possible from
the wind. There are, basically, two ways to control the system when it is in partial load:
constant tip speed ratio constant scheme and the scheme of maximum power point tracking
(MPPT).
In our study we assume a partial load working zone controlled by constant tip ratio con-
stant scheme, which is the most common used control scheme, supposing pitch angle constant
(θpitch = 0).

Figure 3.1: Representation of the control system
3.2.1 Constant tip speed ratio scheme control
Constant tip speed ratio control consist in reach an optimal torque curve which is a function
of the rotation speed in design step of the control, and consign it to the converter so that
the mechanical system is stable in the optimal point of wind power extraction.
Fundamentals of the control
Remembering the equation (1.3), it is clear that the unique variable parameter which we are
able to control is cp , because the wind speed is completely variable and uncontrollable, the
area is determined by our structure and the air density is, almost, constant. Moreover, we
know how this coefficient can be described (1.4) and this is a function of the tip speed ratio
(λ), θpitch and some constant parameters which are defined depending on the windmill.
As our target is to obtain as much power as possible, always working in a security bounds
and we are able to regulate cp . Then, a good solution will be to find the optimal value of
cp , which can be calculated by deriving cp (θpitch , λ) by λ. It is important to remark that
the calculation are done with θpitch = 0 already, because for θpitch 6= 0, the system is quite
hard to solve.
! !
1 c9
d 1 c9 −c7 −
c5 λ+c8 θpitch 1+θ 3
c1 c2 − 3 − c3 θpitch − c4 θpitch − c6 e pitch =0
dλ λ + c8 θpitch 1 + θpitch
(3.1)

From the equation (3.1) we can get the tip speed ratio for θpitch = 0 λopt |θpitch =0 , which
expression is:
c2 c7
λopt |θpitch =0 = (3.2)
c2 c7 c9 + c6 c7 + c2

Replacing this expression in the equation of cp (1.4), we get the optimal value of cp when
θpitch = 0:
c
c4 c7 θ 5 +c3 c7 θpitch +c6 c7 +c2
pitch
c1 c2 e− c2
cpopt = (3.3)
c7
Including both equations (3.2) and (3.3) into the expression of the power obtained by the
turbine we are able to get an equation of the turbine torque what is needed to extract as
much as possible for each wind speed. So, the equation of the Torque for θpitch = 0 is:
c c +c2
− 6 7
c1 e c2 (c2 c7 c9 +c6 c7 +c2 )3 1 3 2
Γθpitch =0 = c22 c27 2 ρAR ωt
(3.4)
= Kcp |θpitch =0 ωt2
where Kcp |θpitch =0 is called the Optimum Torque coefficient of the turbine.
Assuming that the control of the converter is able to force the generator to the give us the
set-point torque with a bigger speed of answer (some magnitude orders) than mechanical
dynamics, and neglecting the flexion of the axis, if we want to get this optimum torque, the
expression of the angular acceleration of the windmill is:
v3

dωt 1 1
= (Γt − Γm ) = ρA cp (vw , ωt ) w − Kcp |θpitch =0 ωt2 (3.5)
dt J 2J ωt
3.3 Control of the converter

The control of the converter system has 2 subsystems: the control of the electrical system
of the stator, and the control of the electrical system of the grid side.
3.3.1 Control of the electrical system of the stator
The control of the electrical system of the stator can be realized through different current
loops.1
To make the control of Torque and induction generator’s speed, will be used by the IMC
(Internal Model Control) because it is very insensitive to changes of parameters, and can
easily tune controllers [10].
This control strategy allows us to use a control structure which encapsulates the process
system, as is described later. Then, as Induction Machine is in synchronism reference has
a transfer function which is a first order system, as it will be shown in 3.3.1, the control
structure that we get is quite similar to PI regulators. The main advantages are:
• Setpoint values are constant in the synchronism reference. We are able to get a zero
static error if we choose this error, and will be neglecting any noise or perturbation,
1 In the appendix C are shown some of them.

as could be seen on (3.3).
• The control parameters are obtained based on known parameters of the machine, and
of a parameter (the bandwidth desired closed loop control).
Brief description of IMC
The control strategy based on IMC was developed for the chemical engineering applications
[11]. The Internal Model Control (IMC) concept relies on the Internal Model Principle which
says that the control can be achieved only if the control system includes some representation
of the process to be controlled.
The illustration (3.2) show the IMC scheme.
Figure 3.2: Schematic of the IMC scheme
where
C(s) IMC controller.
G(s) Induction Generator’s transfer function.
G(s) Internal model of the Induction Generator’s transfer function.
As can be deduced from the 3.2, if G∗ (s) = G(s) (it means that the approximation of the
Induction generator model is perfect), then doesn’t exist feedback and the transfer function’s
matrix close loop is: Gc (s) = G(s)C(s). So, the system will be stable, only, if both transfer
function are stable.
For that, it seems to be a good option to choose the transfer function as C(s) = G∗−1 (s),
because the rise time will be instantaneous. But, this gives us a lot of disadvantages, for
example, detune of the parameter of the model or too high variables which, maybe, are not
possible to apply.

To improve robustness, it is possible for us to include a modification of the model. The idea
is to define the transfer function matrix equal that system transfer function in series with a
low-pass filter. Then the controller’s function matrix becomes:
α
C(s) = G∗−1 L(s); L(s) = I (3.6)
s+α
where
I Identity or unit matrix.
α Controller’s bandwidth in close loop.
In the firstorder systems, the rise time, is related to the bandwidth by the following equation:
tr = ln α9
Current’s Control Loop of the SCIG
Known the IMC propose, it is needed to apply it on current’s control loop of the induction
generator. To do this, we need the transfer function of the control loop, equation which
could be reached directly from Induction generator’s state equation on synchronism refer-
ence (2.15). If we suppose current constant for all the time, then it is correct to neglect the
terms which are affected by magnetizing current, due to it will be, just, a little disturbance
that could be corrected easily with a integer. The transfer function becomes:

vsq rs Ls θ̇ isq L 0 d isq
= + s (3.7)
vsd −Ls θ̇ rs isd 0 Ls dt isd
Applying Laplace transform to (3.7) we obtain

vsq r + Ls s Ls θ̇ isq r + Ls s Ls θ̇
= s ⇒ G∗−1 (s) = s (3.8)
vsd −Ls θ̇ rs + Ls s isd −Ls θ̇ rs + Ls s
To reach controller’s transfer function, we apply the equation (3.6).

α rs + Ls s Ls θ̇
C(s) = G∗−1 (s)L(s) = (3.9)
s + α −Ls θ̇ rs + Ls s
It can be observed from the scheme, that the transfer function F (s) is the transfer function
resulting of the control loop which contains C(s) and G∗−1 .

∗−1 −1
α rs + Ls s Ls θ̇
F (s) = 1 − C(s) · G · C(s) ⇒ F (s) = (3.10)
s −Ls θ̇ rs + Ls s
The system is a control (F), a plant (G) and a unity feedback, and can be represented as it
is shown in the figure 3.4.
Then, the transfer function of the whole system can be obtained simplifying the diagram.
The final transfer function is L, introduced in 3.6, which is a first order system with an

Figure 3.3: Alternative structure of the IMC scheme
Figure 3.4: Block diagram of the system
unitary gain and time constant α−1 .

α
0
CS(s) = [I + F (s) · G(s)]−1 · F (s) · G(s) = s+α
α
2
(3.11)
0 s+α
How to define the currents of reference
To be able to realize the control of the converter of the rotor’s side, we need to define the
reference currents, as is shown in the following control equation
α ref
isq s+α 0 isq
= (3.12)
isd 0 α
s+α iref
sd
To find these references, first of all, we should make a change of variables which is frequently
used in the study of Induction machines in steady state. This new variable is called the mag-
netizing current3 which is the image of the magnetic field in the gap, in our case, with a
magnetic field seen from the rotor. It is described for the following relationship:
Lr
im = is + i (3.13)
M r
Applying this change of variable to the equation system of SCIG (2.15) whose reference and
orientation is synchronized by the flux of the rotor, we obtain:
 2
 
M2
Ls − M
  

 vsd  Lr 0 Lr 0 
 isd 

vsq
  M2 M2  
isq

= 
 0 Ls − Lr 0 Lr 
 d
0  dt imd 

   0 0 M 0    
0 imq
   
0 0 0 M
(3.14)
 
M2 2
rs − Ls − Lr θ̇ 0 −M
Lr θ̇  
 2
 isd
M2

 Ls − M θ̇ rs Lr θ̇ 0
  
 isq

Lr
+ 
 − Lrrr M

0 rr LMr −M θ̇r  

 imd 
 imq
 
rr M
0 − Lr M −M θ̇r rr Lr
Assuming that there is a current control voltage inverter (CCVSI), we can dismiss the equa-
tions which are referred to stator because we do not need to calculate the voltage because
of the CCVSI control will do. Furthermore, there are infinite synchronism references, if we
choose the reference which its direct component is oriented by the maximum of the magnetic
field in the gap (seen from rotor), then the quadratic component of the magnetizing current
will be zero (imq = 0).
So, finally the equation (3.14) can be written as:
 M rr M d
 0 = −rr Lr isd + L r
imd + M dt imd
(3.15)
0 = −rr LMr isq + M θ̇r imd


a a12
2 The inverse of a two-dimensions matrix is easilly calculated. From a matrix A = 11 , its inverse
a21 a22

1 a22 −a 12
is defined as A−1 = det{A}
−a21 a11
3 In the appendix D is explained how it is defined.

Applying the same changes and assumptions to the torque equation (2.17), it becomes:
M2
Γ= · imd · isq (3.16)
Lr
Supposing an steady state of the system and the system in a save zone of work, we are able
to neglect the derivative of the magnetizing current due to it is, almost, null.
Then, we are able to find our desired references as:
Lr ref
iref
sq = θ̇r imd (3.17a)
rr
iref ref
sd = imd (3.17b)
Can be seen that our control system, only, has one degree of freedom, which is iref md .
Replacing the value of iref
sq (3.17a) in the equation of the torque (3.16), we can find a equa-
ref ref
tion which give us imd from the desired torque (Γ ) what will be defined by the wind
speed because we are able to include, also, the hypothesis of the tip speed ratio constant
which offer us a constant relationship between wind speed and turbine rotational speed, and
assuming the perfect conversion of turbine rotational speed and rotor rotational speed 4 .
s s
rr Γ ref rr 12 Cpopt ρAR2 vw
iref
md = = (3.18)
M 2 θ̇r M 2 λopt |θpitch =0 · ν
3.3.2 Control of the electrical system of the grid side
The control of the electrical system of the grid is configured to control the DC Bus voltage
and the reactive power which is consumed or supplied by the converter of the grid side.
The control of the this side of the converter can be defined by the same way that for the
others machines which are using the same converter back to back. This is possible because
the converter decouple both sides, so the grid side is a single element by itself.[12]
Current’s Control Loop of the grid
Applying Park transformation to the equations of the circuit of the grid (1.23) in vz refer-
ence, the equations are:
−Ll θ˙e

vzq vlq rl ilq L 0 d ilq
− = + l (3.19)
0 vld Ll θ˙e rl ild 0 Ll dt ild
where
θ˙e It is the frequency of the grid (2πf)
The evolution of this differential equations system depends on the frequency of the grid.
Applying a feed-forward 5 to the system, we delete the dependence of the dynamics with the
4ω · ν =
ωr = θ̇r
t
5 Method
used to keep constant the state of the system. The system answer to the perturbations in a
known - defined way.

frequency of the grid. We define the feed-forward as:
vblq + vzq − Ll θ˙e ild

vlq
= (3.20)
vld −bvld − Ll θ˙e ilq
Moreover, after the incorporation of the feed-forward, the system becomes decoupled be-
tween q and d. With the following system:

vblq rl 0 ilq Ll 0 d ilq
= + (3.21)
vbld 0 rl ild 0 Ll dt ild
Performing the Laplace Transform to our system (3.21), we obtain the following transfer
function:

vlq r + Ls s 0 ilq r + Ls s 0
= s ⇒ S ∗−1 (s) = s (3.22)
vld 0 rs + Ls s ild 0 rs + Ls s
To make the control of this system we will use again the IMC method, and the whole
system is:
α ( ref )
ilq s+α 0 ilq
= α (3.23)
ild 0 s+α iref
ld
How to define the currents of reference
The control loops of Bus voltage and Reactive power are responsible of generate the refer-
ences for the current loops.
The reference iref
ld is easy to define because reactive power expression of the grid side is
simple in our reference:
3
Qz = − vzq ild (3.24)
2
then, the reference can be found from the referenced reactive power.
2Qref
iref
ld = −
z
(3.25)
3vzq
Otherwise, first of all, the goal of the Bus voltage control is obtain iref
lq holding constant the
DC Link voltage (E). Moreover, to define the Bus voltage control it is necessary make some
assumptions:
1. The grid side converter has an efficiency of 100% and it makes perfect the conversion
between DC /AC.
2. Bus condenser capacity is large that implies a slow voltage evolution, the condenser
voltage is almost constant and we can neglect the grid side inductance looses.

3. The current loop is fast enough to consider that we have an ideal current source.
Making a balance of power through converter of the grid and taking account of the first
hypothesis that we had defined before, we get the following expression:
3
PDCl = Plq ⇒ vDCl · iDCl = 2 (vld · ild + vlq · ilq )
(3.26)
3
⇒ vDCl · iDCl = 2 (vlq · ilq )
Then, we obtain an equation for iDCl current:

3 vlq
iDCl = ilq (3.27)
2 E
Using the second assumption which explain us E ' constant and neglecting voltage drop in
the inductance of the grid side (vlq ' vzq ). Then, equation becomes:
3 vzq
iDCl = ilq (3.28)
2 E
where, now, vzq and E are values always constants.

Making a little review of our equations and of our target, we can say that: We have a DC
1
Link between 2 converters, the DC Link follow the equation (1.21): E = C·s (iDCl − iDCs )6
and we have obtained an expression that relates iDCl and ilq (3.28).
Finally, the resultant system is linear and, then, can be easily designed a controller what
be able to give us a current reference iref
lq which hold Econstant (steady error null) for an
current input (iDCs ) with step shape and reply speed slow enough to be agree with the third
assumption.
Then, the block diagram of the system described before is:
Figure 3.5: Block diagram of the system of the BUS control
The error expression of this system can be expressed as:

1
s
e(s) = 3vzq E ∗ (s) + C
3vzq iDCs (s) (3.29)
s+ 2EC Gc (s) s+ 2EC Gc (s)
Due to we want to assure a null error, we must choose a controller which help us to reach it.
6 Is the equation (1.21), in the Laplace variables

To calculate the steady error of the system we use the Final Value Theorem7 . Assuming
the inputs E ∗ and iDCs as a constant, we can define the controller to reach our desire. As
is quite easy to see for the input E ∗ is only needed a proportional controlled, but for iDCs
is necessary a PI controller. Then we choose the more restrictive of both,
PI controller.
To determine the values of the parameters of the controller Kp + Ksi , we introduce this
definition to our system.
1 s
C C
e(s) = 3vzq = 3vzq Kp 3vzq Ki
(3.30a)
s + 2EC Kp + Ksi s2 + 2EC s + 2EC
s s
C C
⇒ 3v K 3vzq Ki
(3.30b)
s2 + 2ζωn s + ωn2 s2 + 2EC
zq p
s + 2EC
3vzq Ki 2EC 2
ωn2 = 2EC ⇒ Ki = 3vzq ωn
3vzq Kp 4EC (3.30c)
2ζωn = 2EC ⇒ Kp = 3vzq ζωn
We can check if our PI controller works:

s2
C 18
lim s · e(s) = 3vzq Kp 3vzq Ki
⇒ → 09 (3.31)
s→0 s2 + C
2EC s + 2EC
7 x(t→∞)= lims→0 s C(s)

8 Because there is the same maximum potence on numerator and denominator.
9 Applying the second assumption of the large capacity of the Bus Link condenser.

Chapter 4
Modeling of the system
4.1 Introduction
The equations developed in previous chapters have been implemented in Simulinkr 1 to

simulate the system response for a given situation. In this chapter could be seen the details
of the system implementation.
The main idea of the model is keep the same basic structure of the system, at least, the
most important groups: Mechanical system (Turbine and GearBox), Generator, Converter
with his control and Grid.
The complete model is implemented as is shown in the figure (4.1).
Figure 4.1: Simulinkr model of the whole windmill
4.2 Modeling of the Mechanical System
The model of the mechanical system includes both Turbine system and Gear Box.
To determine this model have been necessary some simplifications of the original equations
obtained before, which are:
1 Simulinkr , (Registered trademark of MathWorks) is an environment for multidomain simulation and
Model-Based Design for dynamic and embedded systems.

• The Gear Box model is replaced by only the transmission ratio. We can make this
simplification because we are mainly interested in the electrical behavior of the system.
• The dynamics of the windmill is described in the model as dω 1

dt = J (Γt − Γm ), where
t
J will be only the Inertia of the turbine because is some orders of magnitude bigger
than the Inertia of the generator, as can be seen in [13].2
The implemented mechanical system (4.2)requires three inputs: wind speed (vw ), pitch an-
gle (θpitch ) and Generator Torque (Γm ). And give us as output: Cp value, mechanical angle
(θm ) and mechanical speed (ωm or θ̇m ).
Figure 4.2: Simulinkr model of the mechanical part
4.3 Modeling of SCIG
The model of the generator (figure 4.3) have been defined using the equations of the gener-
ator obtained before. In this box, it is calculated the electrical torque of the generator, and
the active and reactive power of the stator.
4.4 Modeling of Full Power Converter
The model of the full power converter (figure 4.4)have been divided in 3 parts ( Generator
side, DC Link and Network side), as is indeed the converter (rectifier , dc bus and inverter).
2 Typical Inertias Range: Jturb [3 · 106 − 9 · 106 ] and Jgen [65 − 130] [kgm2 ]

Figure 4.3: Simulinkr model of the generator
Figure 4.4: Simulinkr model of the Full Power Converter
4.4.1 Stator Side Converter
The block of the stator side converter (figure 4.5) includes the Torque control of the gener-
ator, through a stator voltage control.
Figure 4.5: Simulinkr model of the Stator Side of the Back-to-Back

4.4.2 DC Link
The block of the DC Link (figure 4.6) is defined directly by the equation (1.21).
Figure 4.6: Simulinkr model of the DC Link of the converter
4.4.3 Grid Side Converter

The block of the grid converter (figure 4.7) includes the DC Link voltage control and the
reactive power control delivered to the grid.
Figure 4.7: Simulinkr model of the grid side of the frequency converter
4.5 Modeling of the Grid

The model of the grid (figure 4.8)have been divided in 2 parts ( Grid Filter and Grid).
4.5.1 Grid Filter

The block of grid filter (figure 4.9) represents the little voltage sag which is produced by the
impedances of the wire, as is defined in the equation (1.23).

Figure 4.8: Simulinkr model of the grid
Figure 4.9: Simulinkr model of the grid filter
4.5.2 Grid
The block of grid (figure 4.10) represents all the grid, which could be defined as an ideal
voltage source. In this block can be found a Park transformation box to change the abc
voltage to qd.
Figure 4.10: Simulinkr model of the grid voltage generation.

Chapter 5
Simulation of the model and

Analysis of the results
To realize the simulation of the model have been defined some parameters. The mechan-
ical parameters (Inertia, transmission ratio, radius and ci [constants of the turbine power
curves]) have been determined as a mean of the range of values that can be found in [13].
The electrical parameters (resistances, inductances and rated power) have been used as in
[14].
We do two different simulations: Generator connected directly to the grid and Complete
system.
5.1 Generator connected directly to the grid

As we explain before, this typology of connection is the most common in SCIG generators
(Constant-speed wind Turbine). So, by one hand, this simulation show us how work this
type of turbine and, by other hand, allow us know if the model of the generator is working
correctly.
We will study how is working the generator, and to determine this we will plot the evolution
of the Torque (figure 5.1)and, also, the evolution of the currents (figure 5.2).
Figure 5.1: Mechanical Torque produced by the Generator

Figure 5.2: Stator currents in qd reference
As can be seen in the figures (5.1) and (5.2), the Torque and the currents reach a steady
state, which implies that the generator model is correct. Can be detected a transitory which
produce a large spike in the q stator current (∼
= 3000 A).
Is necessary look how is working the mechanical system. With this purpose is plotted how
change the power coefficient (Cp) as function of tip speed ratio (figure 5.3).
Figure 5.3: Evolution of Cp(λ)
Figure 5.4: Graphic of characteristic Cp of each turbine.

Comparing the figures (5.3) and (5.41 ), we are able to say that the system is working cor-
rectly due to the values are inside the range of the three blades values that are shown.
After verifying the results obtained, we can say that the generator and the mechanical sys-
tem are working properly.
5.2 Complete system

In this section is checked the validity of our object of study, a SCIG connected to the grid
through a back to back converter.
Using the information obtained in the previous section, we know that the model of the
generator and the mechanical system are correct. Then, we need to test the model of the
converter.
Figure 5.5: Graphic of the evolution of the grid filter voltage
Figure 5.6: Voltage E of the DC Link
In the graphics (5.5) and (5.6) can be seen as the system tries to achieve the values referenced.
Although they do not reach the values, we can interpret as positive this attempt because it
means that, probably, if the system does not became saturated they will arrive to a steady
state.
1 Graphic coming from [5]

Figure 5.7: Stator voltage in qd reference
Figure 5.8: Stator current in qd reference
Figure 5.9: Mechanical Torque produced by the Generator
By one hand, in the graphic (5.7) can be seen how the stator voltage, after the transient,
become little but never constant. By the other hand, in the graphic (5.8) can be seen the
stator current which is always growing up, in principle, following the reference current of-
fered by the control. The oscillating torque occurs with some upper and lower limits (5.9),
which means no existing control on the generator.
Then, in the light of these data, can be determined that the stator side converter model
is not working as we want. It can occur because of several factors: the current references
are not calculated correctly, the transient affect directly to the system which can not reach

the steady state, there are too many simplifications (system without crowbar or filter in
the stator side) or the chosen parameters are wrong, guiding the system to other solution
(infinite or minus infinite).
Park
This point is included to show (figure 5.10)how the Park Transform acts on sine waves.
As can be seen in the graphic, the Park Transform change a 3-dimension time dependent
vector to a 2-dimension constant. Moreover, if the angle of definition correct, one of the
components could be null.
Figure 5.10: Park Transform of the voltage of the grid

Chapter 6
Conclusions
In this project, have been described deeply the equations formulation of the dynamic model
of a squirrel cage induction generator connected to the grid by a back to back converter.
Have been studied, also, the control strategy to extract the maximum power of the wind,
and the control strategy to regulate the voltage and current which will be delivered to the
grid in good conditions.
Analyzing the obtained results can be concluded that the model of the SCIG generator, the
model of the mechanical system, the model of the DC Link and the model of the grid side
converter are correct. But, in the model of the stator side converter there are something
wrong. However, the mistake is bounded. Then is recommended make more simulations,
and a deep study of the model to fix it.
For further work, are proposed follow different lines of study. Is proposed realize the study
of how the system behavior change when are replaced the simplifications by the complete
system. Can be also interesting include in the model the grid and the transformer, and
analyze how react the grid when we connect a Farm of windmills. Is suggested study the
behavior of the model when occurs a sag. Finally, can be necessary realize a prototype of
our model.

Acknowledgements
I would like to acknowledge the guidance and suggestions offered by Dr. Keijo Ruotsalainen,
who has helped me understand many things during the project. And thank to the University
of Oulu who gave me the option of realize this project.
I would like to thank CITCEA (Centre d’Innovació Tecnològica en Convertidors Estàtics

i Actuadors) of the UPC (Universitat Politècnica de Catalunya) who introduce me to this
beautiful world of the wind power. Specially, I thank to Dr. Oriol Gomis and Adrià Junyent,
of whom I have learned and they, always, were there when I need some help.
Finally, I would like to thank to my family and my girlfriend who, always, gave me the
strength to go ahead in each moment of my life.

Appendix A
Controllability and
Observability of the SCIG
Linear System
It is well known that an objective in control theory, is to be able to transfer the movement
from one point in the state space to another manipulating to chosen control variables and
to know which points in the state space can be reached from a given point, this leads to
two concepts controllability and reachability. Another point of interest is the concept related
to determining the state of a system from the knowledge of the input-output data leading
to the concept of observability. In 1960 The engineer and mathematician R.E. Kalman
[15], laid down the foundation of modern control theory. In this appendix we analyze the
controllability and observability system properties [16], which allow us be sure of we are able
to control, and know our system. We have the following system:
Ẋ = A · X + B · U (A.1a)
Y =C ·X +D·U (A.1b)
Where we assume A, B, C and D like a constant. This assumption allow us define the system
as LTI (Linear Time-Invariant).
In our generator we could get this system including the equation of the SCIG (2.20) and the
equation of the flows (1.9) in qd reference. Then our system is:

 
 isq
 

d  isd 
=
dt 
 irq 

ird
 
| {z }
Ẋ
 
isq 
M 2 θ˙r +(Ls Lr −M 2 )θ̇ M Lr θ˙r

L r rs −M rr 
 
1 −M 2 ˙
θ −(L L −M 2
) θ̇ L r rs −M Lr θ˙r −M rr  isd

−  r s r
−M Ls θ˙r (Ls Lr −M 2 )θ̇−Ls Lr θ˙r
Ls Lr − M2 −M rs L s rr  irq 
M Ls θ˙r −(Ls Lr −M 2 )θ̇−Ls Lr θ˙r

 
−M rs Ls rr i

| {z } rd
A
| {z }
X
 
" # vsq 
Lr 0 −M 0 
 
1 0 Lr 0 −M vsd

+ −M 0 Ls 0
Ls Lr − M 2 0 −M 0 Ls  0 
 
| {z } 0 
B | {z }
U
(A.2a)
    

 λsq 
 Ls 0 M 0   isq 
 
λsd 0 Ls 0 M i
  
=  sd 1 (A.2b)

 λrq 

M 0 Lr 0 
irq 
λrd 0 M 0 Lr ird
   
| {z } | {z }| {z }
Y C X
A.1 Controllability
The concept of controllability denotes the ability to move a system around in its entire
configuration space using only certain admissible manipulations. The system can be defined
as controllable if, for all initial times and all initial states, there exists some input function
that drives the state vector to any final state at some finite time.
For continuous time invariant linear systems the concept of controllability is equivalent the
reachability, as a consequence we can study controllability computing a rank of a certain
matrix (reachability matrix) that in this case it is known as controllability matrix, which is
given by:
A2 · B An−1 · B

R= B A·B ... (A.3)
If the rank of the matrix is maximum, then the system is controllable.

As can be realized easily, our system is controllable because B has a maximum rank.
" #
Lr 0 −M 0
1 0 Lr 0 −M
−M 0 Ls 0 (A.4)
Ls Lr − M 2 0 −M 0 Ls
| {z }
B
And it is quite easy to calculate its determinant, which is different than zero.
1D has not included because is a null matrix.

L2r L2s − 2Ls Lr M 2 + M 4
det(B) = |B| = = Ls Lr − M 2 6= 0 (A.5)
Ls Lr − M 2
So, as well known, det(B) 6= 0 ⇒ maximum rank of B ⇒ maximum rank of R
We are able to say that our system is controllable.
A.2 Observability
A system is observable if, for any possible sequence of state and control vectors, the current
state can be determined in finite time using only the outputs. This means that from the
outputs of the system it is possible to determine the behavior of the system. If a system
is not observable, the current values of some of its states can not be determined through
output sensors.
Ẋ = AX + BU,
The notion of observability is dual reachability, that is to say a system
Y = CX + DU

Ẋ = At X + C t U
is observable if and only if the dual system is reachable.
Y = B t X + Dt U
Taking into account the equivalence into reachability and observability we can compute the
observability by means the rank of a certain matrix called observability matrix given by:
 
C
 C ·A 
2
 
O = C ·A (A.6)
 

 .. 
 . 
C · An−1
If the rank of the matrix is maximum, then the system is observable.

As can be realized easily, our system is observable because C has a maximum rank.
 
Ls 0 M 0
 0 Ls 0 M 

M 0 Lr 0 
 (A.7)
0 M 0 Lr
| {z }
C
And it is quite easy to calculate its determinant, which is different than zero.
det(C) = |C| = L2r L2s − 2Ls Lr M 2 + M 4 6= 0 (A.8)

So, as well known, det(C) 6= 0 ⇒ maximum rank of C ⇒ maximum rank of O
We are able to say that our system is observable.
A.3 A lower bound on the distance to a non-controllable

system

Ẋ = A · X + B · U,
Given a system we are interested in to obtain a lower bound for
Y =C ·X +D·U
the distance between a controllable system and the nearest non-controllable one, this bound
is given in terms of the singular values of a certain matrix associated to the system.

The problem to study controllable systems being those whose behavior does not change
when applying small perturbations, in the qualitative theory of dynamical systems has been
widely studied by many authors in control theory and under different points of view, (see
[17] for example).
Our goal is to obtain a bound for the value of the radius of a ball which is neighborhood of
a controllable system, containing only systems which are also controllable.
In our particular setup the matrix B has full rank, so if we perturb a system and the new
matrix B 0 = B + εB has full rank, the the perturbed system is also controllable because of
rank B 0 A0 · B 0 . . . A0 n−1 · B 0 = rank B 0 = 4

So, if the perturbed system is non controllable necessarily rank B 0 < 4. Then, now the
question is how far is a system such that rank B 0 < 4.
The distance we will deal with is that deduced from the Frobenius norm. Weqrecall
P that
given a matrix K = (kij ) ∈ Mn×m (C), its Frobenius norm is defined as kKk = 2
ij kij .
The Eckart-Young and Minkowski theorem (see [18] for example), states that the smallest
perturbation in the Frobenius norm that reduces the rank of a matrix K with rank K = r
from r to r − 1 is σr (K), the smallest non-zero singular value of K.
Computing the singular values of matrix B we have
L2r + M 2
 
0 −M (Lr + Ls ) 0
1 2 2
0 L + M 0 −M (L r + Ls )
BtB =
 r

 
(Ls Lr − M 2 )2  −M (Lr + Ls ) 0 M 2 + L2s 0 
0 −M (Lr + Ls ) 0 M 2 + L2s
The characteristic polynomial of this matrix is

2
L2r + L2s + 2M 2

1
p(t) = t2 − t + = ((t − t1 )(t − t2 ))2
(Ls Lr − M 2 )2 (Lr Ls − M 2 )2
√ √
L2 +L2 +2M 2 + (L2r −L2s )2 +4M 2 (Lr +Ls )2 L2r +L2s +2M 2 − (L2r −L2s )2 +4M 2 (Lr +Ls )2
with t1 = r s 2(Ls Lr −M )2 2 ≥ t2 = 2(Ls Lr −M 2 )2 .
√
Consequently, the smallest non zero singular value is σ2 = t2 and if kδBk < σ2 , the
perturbed system is controllable.
In fact we can refine this bound computing the smallest non-zero singular value of the
controllability matrix
A2 · B A3 · B

B A·B
getting more margin of safety.
A.4 A lower bound on the distance to a non-observable

system

Ẋ = A · X + B · U,
Analogously, given a system we are interested in to obtain a lower
Y =C ·X +D·U
bound for the distance between a observable system and the nearest non-observable one, that

is to say we try to obtain a bound for the value of the radius of a ball which is neighborhood
of a observable system, containing only systems which are also observable.
Computing the singular values of matrix B we have
L2s + M 2
 
0 M (Lr + Ls ) 0
2 2
t
 0 L s + M 0 M (Lr + Ls ) 
C C= 
 M (Lr + Ls ) 0 M 2 + L2r 0 
0 M (Lr + Ls ) 0 M 2 + L2r
The characteristic polynomial of this matrix is

2
p(t) = t2 − (L2r + L2s + 2M 2 )t + (Lr Ls − M 2 )2 = ((t − t1 )(t − t2 ))2
√ √
L2r +L2s +2M 2 + (L2r −L2s )2 +4M 2 (Lr +Ls )2 L2r +L2s +2M 2 − (L2r −L2s )2 +4M 2 (Lr +Ls )2
with t1 = 2 ≥ t2 = 2 .
√
Consequently, the smallest non zero singular value is σ2 = t2 and if kδBk < σ2 , the
perturbed system is controllable.
In fact we can refine this bound computing the smallest non-zero singular value of the
controllability matrix  
C
 C ·A 
 
 C · A2 
C · A3
Obviously, taking the minimum between the controllability bound and the observability
bound we obtain a bound for both characters the controllability and observability.

Appendix B
Space Vector Pulse Width

Modulation (SVPWM)
An three-phase inverter feeded with voltage, represented in the figure (B.1), is composed
by DC Bus, also called(DC Link), with 2 condensers of high capacity and 3 switch pairs 1
which are linked to each phase of the grid between each pair. The inverter use a binary
reference, which describe in what position are the switches.
Figure B.1: Electrical scheme of an inverter
As is shown in the figure (B.1) the inverter has 6 switches, which can be only in two different
positions (as we told before binary reference): blocking state or going-on state. Otherwise,
both switches of the same branch had 2 freedom degrees as they have to be in opposite
states. Then, the inverter are able to generate 8 (23 , 3 different branches which have 2
freedom degrees) different three-phasic voltages in its output.
If the 8 different voltages vectors are represented in a steady Park reference with θ=0, we
obtain 6 vectors which are separated 60o between each other, and two vectors with null
longitude. So, six of these vectors represent the vertices of an hexagon and the other two
just the center, as it is shown in the figure (B.2) and described in the table (B).
1 The switches are, in fact, IGBT with a protection diode

State IGBT activated Vector van vbn vcn vqn vdn v
√0n
− 2
0 Q4 Q6 Q2 (0 0 0) 0 0 0 0 0 2
√
E
2 1 1 2 − 2
1 Q1 Q6 Q2 (1 0 0) 3 E −3 E −3 E 3 E √
0 √6
E
2 Q1 Q3 Q2 (1 1 0) 1
3 E
1
3 E − 32 E 1
3 E − √33 E 6
√
2
E
− 2
3 Q4 Q3 Q2 (0 1 0) − 13 E 2
3 E − 31 E − 13 E − 33 E √6
E
2
4 Q4 Q3 Q5 (0 1 1) − 23 E 1
3 E
1
3 E − 23 E √
0 6
√
E
3 − 2
5 Q4 Q6 Q5 (0 0 1) − 13 E − 31 E 2
3 E − 13 E √3
E √6
E
1 3 2
6 Q1 Q6 Q5 (1 0 1) 3 E − 32 E 1
3 E
1
3 E 3 E √6
E
2
7 Q1 Q3 Q5 (1 1 1) 0 0 0 0 0 2 E
Table B.1: Table of commutation states of the inverter
Figure B.2: Commutation hexagon
The SVPWM method to generate any three-phasic voltage, consists of alternating different
commutation configurations of the switches which corresponding to the two vertices of the
hexagon where there is the desired voltage and one of the zero states in time periods cal-
culated so that resultant voltage signal has a continuous component equal to the desired
voltage and some high frequency components which are easy to filter.
∗
Known a desired voltage vector vqd , the periods [ti , ti+1 , t0 ] in per unit, when the states
should be activated, are as follow:
∗
ti 1 v(i+1)d −v(i+1)q vq
= (B.1a)
ti+1 viq v(i+1)d − vid v(i+1)q −vid viq vd∗
t0 = 1 − (ti + ti+1 ) (B.1b)

Appendix C
Introduction to current loops
This appendix is a brief introduction to few techniques which allow the control by current
of a voltage inverter.[19]
The most common techniques are:
• Hysteresis control, also called Bang Bang control. It was the most useful in analog-
ical control time, as it had a fast control (high bandwidth). In contrast, it presents a
commutation frequency variable, and large current ripple.
• PI Regulator in stationary reference. The main idea of this regulator was to im-
plement a PI loop for each current of the triphasic inverter. But in this way, working
independent each regulator, the system is not able to reach the voltage zero state, and
the error in one current have influence in the other currents. So, its behavior is not
optimal.
• PI Regulator in synchronism reference. This controller is based in two PI regu-

lators in synchronism reference ( moreover, the reference is oriented by the maximum
of the magnetic field of the rotor. The problems of this control are that it depends
hardly on the estimation of the parameters of the machine, and is not able to control
a over-modulation.
• Predictive regulators. It is based on the analysis of the trajectories that will be

followed by current vector.
• IMC method. Our selection

Appendix D
Definition of magnetizing
current
In the study of induction machine in steady state is customary to define a new current
variable called magnetizing current I m , to be a representation of the magnetic field in the
gap.
Otherwise, the target of the bigger part of the control strategies is make, independently, the
control of the torque and the control of the induction machine, though in the equations with
the stator and rotor current variables the flux does not appear as a explicit variable (but it
is the sum of the flux generated by stator and the flux generated by rotor).
The flux can be defined in three different ways [20].
• Magnetic field seen from stator: Ls im = Ls is + M ir
• Magnetic field seen from rotor: M im = M is + Lr ir
• Magnetic field in the gap: M im = M is + M ir
To apply this change of variables there are a change of variable generic matrix, which will
be function of a and b. This matrix is defined as follow:
         

 isd 
 1 0 0 0   isd 
 
 isd  1 0 0 0   isd 
isq 0 1 0 0 isq isq 0 1 0 0 isq
         
=  ⇒ =  −a 1
 (D.1)
 imd  a 0 b 0  ird   ird  
b 0 b 0 
 imd 
−a
      1 

imq 0 a 0 b irq irq 0 0 b imq
      
b
Depending on which definition of magnetizing current we want to use, we a and b will get
the following values:
M
• Magnetic field seen from stator: a = 1 and b = Ls
Lr
• Magnetic field seen from rotor: a = 1 and b = M

• Magnetic field in the gap: a = 1 and b = 1
Finally, applying this change of variables in our system, choosing magnetic field seen from
rotor, we obtain:
 
M2 M2
   

 vsd 
 Ls − Lr 0 Lr 0 
 isd 


vsq
  M2 M2  
isq

=
 0 Ls − Lr 0 Lr  d

0
 dt  i

 
  0 0 M 0   md 

0 i
   
0 0 0 M mq
(D.2a)
 
M2 2
rs − Ls − Lr θ̇ 0 −M
Lr θ̇  
 2
 isd
M2

 Ls − M θ̇ rs Lr θ̇ 0
  
 isq

+ Lr 
 − Lrrr M

0 rr LMr −M θ̇r  

 imd 
 imq
 
rr M
0 − Lr M −M θ̇r rr Lr
M2
Γ= · imd · isq (D.2b)
Lr

Bibliography
[1] VTT. Wind energy statistics in finland. http://www.vtt.fi/proj/windenergystatistics/,

2009. 7, 17
[2] Suomen Tuulivoimayhdistys. Current political issues in finland.

http://www.tuulivoimayhdistys.fi/in+english, 2009. 17
[3] Helsingin Sanomat. Valtaosa suomalaisista kannattaa tuulivoiman lisäämistä. 2007. 17
[4] Frank White. Fluid Mechanics. McGraw-Hill, 4th edition, 1998. 26
[5] Erik Hau. Wind Turbine: Fundamentals, Technologies, Applications, Economics.

Springer, 2006. 26, 58
[6] S. Heier. Grid Integration of Wind Energy Conversion Systems. John Wiley and Sons,
1998. 26
[7] Slavomir Seman. Transient performance analysis of the Wind-Power Induction Gener-
ators. PhD thesis, Helsinki University of Technoloy. 27
[8] P.C. Krause. Analysis of Electric Machinery. McGraw-Hill, 1986. 28
[9] Juan A. Martinez Velasco. Sistemas Eléctricos de Potencia. CPDA, 2000. 32
[10] H.P. Nee L. Harnefors. Model-based current control of ac machines using internal model
control method/ f. IEEE Transactions on Industry Applications, 34:133–141, 1998. 45
[11] Ming T. Tham. Introduction to Robust Control. PhD thesis, University of Newcastle
upon tyne, 2002. 45
[12] Adrià Junyent. Modelització i control d’un sistema de generació elèctrica de turbina de
vent. Master’s thesis, ETSEIB- UPC. 49
[13] Thomas Ackermann. Wind Power in Power Systems. Wiley, 2005. 53, 57
[14] Dafeng Fu ; Yan Xing. Study on linear dynamic model and analysis of operating
characteristics of high-power vscf wind energy conversion system. World Non-Grid-
Connected Wind Power on Energy Conference, 2009. 57
[15] R.E. Kalman. Control System Analysis and Design via the Second Method of Liapunov:
I. Continuos-Time Systems; II. Discret-Time Systems. 1960. 65
[16] M Isabel Garcia-Planas. Notas sobre Teoria Matemática de Sistemas. 1994. 65
[17] M Isabel Garcia-Planas. Estimating the sensivity of the structure of the generalized
systems under feedback and output injectino equivalence. WSEAS trans. Circuits and
Systems, 5:298–304, 2006. 68

[18] C. Van Loan G. Golub. Matrix Computations. University Press, 1989. 68
[19] Joan Bergas. Control del motor d’inducció considerant els lı́mits del convertidor i del
motor. PhD thesis, UPC, 2000. 73
[20] D. W. Novotny R.W De Doncker. The universal field oriented controller. IEEE Trans-
actions on Industry Applications, 30:92–100, 1994. 75


SCIG With Full Power Converter - Book2sides

Uploaded by

Copyright:

Available Formats

SCIG With Full Power Converter - Book2sides

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SCIG With Full Power Converter - Book2sides

Uploaded by

Copyright:

Available Formats

Modeling and control of Squirrel Cage Induction

Generator with Full Power Converter applied to

Domı́nguez Garcı́a, José Luis

November 30, 2009

The pessimist complains about the wind; the op-

If a man knows not what harbor he seeks, any

SCIG with Full Power Converter 3

SCIG with Full Power Converter 5

4 Modeling of the system 53

5 Simulation of the model and Analysis of the results 57

A Controllability and Observability of the SCIG Linear System 65

B Space Vector Pulse Width Modulation (SVPWM) 71

C Introduction to current loops 73

D Definition of magnetizing current 75

6 SCIG with Full Power Converter

1 Location of windfarms in Finland.[1] . . . . . . . . . . . . . . . . . . . . . . . 17

1.1 Generic Nacelle of a Windmill . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.1 Any current vector in abc reference . . . . . . . . . . . . . . . . . . . . . . . . 35

3.1 Representation of the control system . . . . . . . . . . . . . . . . . . . . . . . 43

4.1 Simulinkr model of the whole windmill . . . . . . . . . . . . . . . . . . . . . 53

SCIG with Full Power Converter 7

5.1 Mechanical Torque produced by the Generator . . . . . . . . . . . . . . . . . 57

B.1 Electrical scheme of an inverter . . . . . . . . . . . . . . . . . . . . . . . . . . 71

8 SCIG with Full Power Converter

SCIG with Full Power Converter 9

A area where the air flow could pass through

vw1 wind speed before the turbine

vw2 wind speed after the turbine

Pwind mechanical power extracted by the converter

Pwind0 mechanical power that could be converted

R radius of the turbine, that means, the length of blades

c Viscosity constant of the turbine’s shaft

k Elasticity constant of the turbine’s shaft

Jm Inertia moment of the generator

Jt Inertia moment of the turbine

vsabc stator winding’s voltage vector

vrabc rotor winding’s voltage vector

SCIG with Full Power Converter 11

rr resistance of the rotor windings

Lsm coupling inductances between stator windings

Lrm coupling inductances between rotor windings

Lls dispersion inductance of the stator windings

Llr dispersion inductance of the rotor windings

L(θr ) Induction machine’s coupling inductance matrix

P number of poles that the Induction Machine has

E0 DC Bus voltage at t=0

iDCE Current which flows through the condenser

iDCg Current which came from grid’s converter

iDCs Current which came from stator’s converter

Z Value of the impedance

R Value of the resistance parameter

L Value of the inductance parameter

l Value of the length of the wire from converter to electrical transformer

Xccpu Value of the short circuit inductance in PU reference

V1b Base voltage in the primary

V1n Nominal voltage in the primary

12 SCIG with Full Power Converter

T (θ) Park Transform

vsqd0 stator winding’s voltage vector in qd0 reference