1

State of the Art of Wind Turbines Modelling with

Induction Generator (DFIG)
A. J. Pujante-Lopez, E. Gomez-Lazaro, A. Molina-Garca, J. A. Fuentes-Moreno and S. Martn-Martnez

AbstractIn this paper a bibliographical review about the

state-of-the-art of the electrical and mechanical modelling of
Wind Turbine (WT) with Induction Generator (IG) is realized.
Firstly, a brief overview of WT topologies about fixed and
variable speed operation is included. Then, different orders of
electrical and mechanical models of the Double Fed Induction
Generator (DFIG) have been considered, taking into account
equivalent schematic representations. Regarding the power coefficient of the WT, optimum constants from characteristic curves
are obtained. In this WT with DFIG, the aerodynamic, speed
and power control subsystems are studied. Moreover, several
control strategies into the speed control subsystem when the
WT is operating at partial-load are included. Finally, the power
control system of a DFIG is presented, taking into account two
topologies of power converters and few PWM control schemes for
these converters along with two protection circuits more useful.
Index TermsWind turbine, induction generator, modelling,
parameters, power converter, control strategies, PWM schemes,

I. S TATE - OF - ART WT AND IG TECHNOLOGIES

An overview of wind turbine topologies includes fixed or
variable speed operation as it is explained in [1]. In fixedspeed wind turbines (FSWT), regardless of the wind speed, the
rotors speed is fixed and determined by the frequency of the
supply grid, the gearbox ratio and the generator design (SFIG
or DFIG). In the fixed-speed (FS) topology the IG is directly
connected to the grid, with a soft-starter and a capacitor bank
for reducing reactive power compensation. The FSWT has the
following advantages, it is simple, robust, reliable and wellproven, and also the cost of its electrical parts is low. Its
disadvantages are an uncontrollable reactive power consumption, mechanical stress and limited power quality control. On
the other hand the variable-speed wind turbines (VSWT) are
designed to achieve maximum aerodynamic efficiency over a
wide range of wind speeds. In variable-speed (VS) operation
is possible continuously to adapt (accelerate or decelerate)
the rotational speed of the WT to the wind speed. This way,
the tip speed ratio is kept constant at a predefined value that
corresponds to the maximum power coefficient. Contrary to
a FS system, a VS system keeps the generator torque fairly
The financial support provided by the Ministerio de Ciencia e Innovacion
ENE2009-13106 and Junta de Comunidades de Castilla-La Mancha
PAI08-0145-9976 are gratefully acknowledged.
Emilio Gomez, Antonio J. Pujante and Sergio Martn are with Renewable Energy Research Institute, Universidad de Castilla-La Mancha, Albacete 02071 Spain. (e-mails: [email protected], [email protected], [email protected]). (phone: +34 967 599 200,
fax: +34 967 555 321)

Juan Alvaro
Fuentes and Angel
Molina are with the Department of Electrical Engineering, Universidad Politecnica de Cartagena, Cartagena 30202
Spain; (emails: [email protected], [email protected]). (phones:
+34 968 325 604, +34 968 325 462, fax: +34 968 325 356)

Fig. 1.

Typical wind turbine topologies.

SCIG Squirrel Cage IG; WRIG Wound Rotor IG; DFIG Double
Fed IG; PMSG Permanent Magnet SG; WRSG Wound Rotor SG.

constant and the variations in wind are absorbed by changes

in the generator speed; but the electrical system of a VSWT is
more complicated than that of a FSWT. It is typically equipped
with an IG or synchronous generator (SG) and connected to the
grid through a electric converter. This power converter controls
the generator speed; that is, the power fluctuations caused
by wind variations are absorbed mainly by changes in the
rotor generator speed and consequently in the WT rotor speed.
The advantages of VSWT are an increased energy capture,
improved power quality and reduced mechanical stress on the
WT. The disadvantages are losses in power electronics, the
use of more components and the increased cost of equipment
in the power electronics.
The most commonly applied WT configurations are classified both by their ability to control speed and by the type
of power control they use [2]. Applying speed control as
the criterion, there are four different dominating types of
WT, as illustrated in Figure 1. WT configurations can be

TABLE I

W IND TURBINE CONFIGURATION TYPES .

Speed control
Fixed speed
Variable
speed

Type
Type
Type
Type

A
B
C
D

Stall
Type A0
Type B0
Type C0
Type D0

Power control
Pitch
Active stall
Type A1
Type A2
Type B1
Type B2
Type C1
Type C2
Type D1
Type D2

further classified with respect to the type of power (blade)

control: stall, pitch, active stall. Table I indicates several
types of configurations, taking both criteria into account. Each
combination of these two criteria receives a label; for example,
type B1 denotes the variable-speed pitch-controlled WT.
Next, the typical WT configurations are listing, which
advantages as well as disadvantages are summarizing in [2].
Type A: only fixed speed.
Type B: limited variable speed (VS).
Type C: VS with partial scale frequency converter.
Type D: VS with full-scale frequency converter.
II. DFIG M ODELLING AND OPERATION
Regarding to the electrical and mechanical configuration of
the rotor of the Induction Machine or Generator (IM, IG) in
this case, two different types can be distinguished. The first
one is a short-circuited rotor called squirrel-cage (SCIG) or
Singly Fed Induction Generator (SFIG), and the other one is a
wound rotor (WRIG) named Doubly Fed Induction Generator
(DFIG). The SFIG electrical model is a particular case of
the DFIG electrical model [3], whereas the stator windings
structure and the complete mechanical model are similar in
both configurations. There are also constructed, SFIG and
DFIG with single and double cage rotors, being the double
cage modelling more complex than the other one.
A. Electrical modelling
In an IM, the three-phase stator windings are usually
distributed so that the magnetomotive force (mmf) may be
assumed to be sinusoidally distributed in space around the air
gap periphery. If, in addition, the machine can be assumed
to be magnetically linear, the flux density will also be sinusoidally distributed. For such machines, the space vector
notation and concepts are particularly useful, since space
vectors were originally invented to describe the spatial flux
of an AC machine [4]. A three-phase stator winding, which
is supplied with three-phase currents, forms a rotating flux in
the air gap. The same rotating flux could also be formed with
only two orthogonal phases. In this approach, all variables
are represented by polar vectors indicating the magnitude and
the angular position of the rotating sinusoidal distribution. For
example, the space vector of the stator currents is derived from
the phase currents with the following transformation:
2
i~s = (ia + ib + 2 ic ) = i + ji = Is ej(+1 )
3

(1)

where = ej2/3 , 1 is a phase shift and is the synchronous angle corresponding to the synchronous frequency,

Fig. 2.

Principle of space vectors representation.

as d/dt = , being = t in steady-state. It is also

possible to transform the vector to synchronous coordinates
(dq reference frame) as follow:
i~s = id + jiq = ej Is ej(+1 ) = Is ej1

(2)

The synchronous coordinate system has to be aligned with a

quantity, normally the stator or rotor flux of an IM. However,
it is also possible to align this coordinate system with, for
example, the stator current (Fig. 2). Note that, space vectors
in synchronous coordinates will be DC quantities in the steady
state.
The electrical model of the DFIG is basically represented
by a set of ordinary differential equations which relate the
voltages, currents and magnetic fluxes for both stator and rotor
sides. The transformations into the synchronous coordinate
system can also be applied for phase voltages, phase currents
and flux linkages in the stator and rotor windings.
(

v~s = vsd + jvsq , i~s = isd + jisq , ~s = sd + jsq

v~r = vrd + jvrq , i~r = ird + jirq , ~r = rd + jrq

(3)

The typical detailed model is represented by four first-order

differential equations. This 4th order model is needed in order
to study the transient state of the IG. These four voltage and
four flux linkage equations are representing by two equations
in space vector notation as follows:
(
v~s = Rs i~s jd ~s + d~s /dt
(4)
v~r = Rr i~r jdm ~r + d~r /dt
(
~s = (Ls + Lm )i~s + Lm i~r
(5)
~r = (Lr + Lm )i~r + Lm i~s
being d , dm instantaneous speeds of the dq equivalent
windings, in electrical rad/s; Rs , Rr the stator and rotor resistances, respectively; finally Ls , Lr , Lm the stator leakage,
rotor leakage and magnetizing inductances, respectively.
The reduced model is represented by two first-order differential equations, since this 2nd order model neglects the
dynamics of the stator linked flux in the 4th order model [3].
Therefore, this simplified model is used to analyze the steady
state of the IG. The new voltage equation in space vector
notation is given below, whereas the flux linkage equations
are not modified.

Fig. 3.

Transient-state equivalent circuits: a) T-form, b) -form.

v~s = Rs i~s jd ~s
v~r = Rr i~r jdm ~r + d~r /dt

(6)

One of the common ways of representing the models is

using the equivalent circuit of the DFIG (Fig. 3). That electric
circuit is basically composed by several passive elements as
internal resistances (Rs , Rr ) and leakage inductances (Ls ,
Lr ), the mutual resistance and inductance (Rm , Lm ), as well
as two voltage sources for the stator and rotor windings (Vs ,
Vr ).
This equivalent circuit, in this case without the Rm , is valid
for one equivalent Y phase but if the DFIG is -connected
the machine can still be represented by this equivalent Y
representation. Moreover, there are at least three topologies
for representing the electric circuit, the first one is called Tform which is simple to understand and widely used in the
bibliography [4], the other one is named -form which is
particularly appropriate for analysing and deriving control laws
for the DFIG, for instance the vector control in power converts
[5]; and there also is an inverse 1 -form which is very similar
to -form, as it will be explain after that.
The name of the -form is due to the formation of a
of the equivalent inductances. In the equivalent circuit, the
rotor and the stator leakage inductances have the same effect
from a dynamic point of view. Therefore, it is possible to
use a different representation model in which the equivalent
leakage inductance is placed in the rotor circuit, whereas in
the 1 -form it is into the stator circuit.
The quantities and parameters of the T-form relate to the
simplified -form are summarized in the following equations:

= (Ls + Lm )/Lm , LM = Lm
LL = Ls 2 Lr , RR = 2 Rr

vR = vr , iR = ir /, R = r

(7)

In the transient state of the electric circuit, the voltages,

currents and fluxes include the transient dynamics and all the

Fig. 4.

Drive train models of WT generator system [7].

harmonic frequency components. In the steady state, the analysis is simplified by resolving each variable into a component
at fundamental frequency s , plus the main harmonics [4].
Note that, when the rotor voltage (Vr or VR ) is short
circuited, the equivalent circuit for the DFIG becomes the
ordinary equivalent circuit for a SFIG.
B. Mechanical modelling
The drive-train of a wind turbine generator system (WTGS)
in general consists of a blade pitching mechanism with a
spinner, a hub with blades, a rotor shaft relatively long, and a
heavy gearbox with breaker and generator. The inertia moment
of the wind wheel (hub with blades) is about 90% of the drivetrain total moment, while the generator rotor inertia moment
is about 6-8%. The remaining parts of the drive train comprise
the rest (2-4%) of the total inertia moment. At the same
time, the generator represents the biggest torsional stiffness.
The rotor shaft stiffness is about 100 times less and the
stiffness of the hub with blades is about 50 times less than the
generator stiffness [6]. Because the torsional vibration of the
drive-train elements are inevitable in the low-speed side, and
moreover their character in the high-speed side (e.g. frequency,
amplitude) can highly influence the WTGS performance.
Regarding to the mechanical part of the WT, the general
way for modelling the drive-train treats the rotor as a number

of discrete masses connected together by springs defined by

damping and stiffness coefficients [6]. Therefore, the mechanical model of a WT is represented by differential equations, as
equations as masses, which relate the accelerations, velocities
and angular positions of the different components that make up
the drive-train. The most complex model used in references is
composed of six masses [7]. This 6-mass model considers the
distribution of the aerodynamic torque in the 3 blades of the
WT, including the low and high speed shafts of the drive-train,
the gearbox and the rotor part of the IG as main elements (Fig.
4a). On the other hand, the three-mass model simplifies the
mechanical model by grouping the 3 blades of the WT along
with the primary shaft of the drive-train in a single block and
the other masses keeping equal (Fig. 4b). This 3-mass model
can be easily simplified into the 2-mass model by grouping the
gear-box either in the turbine-blades block or in the generator
block (Fig. 4c), remaining only an equivalent shaft between
both masses [7]. In the one-mass model, all types of windmill
drive train components are lumped together and they work as
a single rotating mass, as shown in Fig. 4d.
Nonetheless, the typical implementation of the drive-train
model used in the transient-state operation analysis is based
on the 2-mass shaft model by method 2, i.e. it is assumed
two lumped-masses: the generator (with gearbox) mass and
the hub with blades (wind wheel) mass. The structure of the
model was presented in Fig. 4c. The model is described by
the two equations where all the variables are expressed with
respect to the generator or the wind-wheel side [6], as follow:
(

g
g
0
00 dw
Jwt
dt = Twt D1,2 (w NGB ) K2M (w NGB )
D

d
1,2
g
g
g
Jg0 dt = Te NGB (w NGB ) K2M (w NGB )
(8)
00
being Jwt
, Jg0 the equivalent inertia constants, K2M the
equivalent shaft spring constant; D1,2 the equivalent damping
0
, Te the mechanical and electrical torques; w ,
coefficient; Twt
g the mechanical angular speeds (elect. rad/s); w , g the
mechanical angular positions (elect. rad); NGB = gn /wn
the gearbox transformation ratio, and wn , gn the windwheel and generator rated speeds, respectively.
The WTGS drive-train is not advisable to be modelled as
a single lumped mass in transient analysis because it is too
simple for representing the dynamics of a WT, and stability
analysis based on this model may give significant errors,
especially in FSWT [7]. However, in the steady-state operation
analysis the minimal realization of the drive-train model is
based on the one-lumped mass [8], whose model can be
described by the following simple equation:

Fig. 5.

Curve of the optimum Cp for typical WT model.

Pmec = Cp (, )0.5R2 Vw3

(10)

Therefore this power depends on the wind speed (Vw ) and

the power coefficient (Cp ), which is a function of the tip speed
ratio () and the pitch angle (). The quantity = Rt /Vw
is defined as the relation between the tip speed and the wind
speed of the WT, being R the rotor radio and the air density.
Figure 5 shows a typical no-lineal relation [9] between Cp and
for different pitch angles ().
The optimum curve CP opt () can be obtained joining the
points where the Cp is maximum for each angle ; in fact
every Cp function has the greatest value in = 0o , that is
CP max and opt is the tip speed ratio at this point. This
opt is assumed as the optimum value that, depending on
the wind speed (Vw ), gives the optimum WT speed (opt )
through the above definition of (). Substituting the wind
speed (Vw ) of equation (10) into () definition, and grouping
the constant terms in the resulting equation, the following
optimum constant (Kopt ) is achieved:
Kopt = 0.5R5

CP max
3opt

(11)

At the optimum point, the maximum active power (Popt )

follows a cubic function of the WT speed (opt ), whereas
the optimum torque (Topt ) follows a quadratic function of the
(opt ), being Popt = Topt opt the relation between them:
s

000
00
Jwt
dr /dt = Twt
Te D1M r

(9)

Popt =

Kopt 3opt , opt

000
Jwt

where
is the wind turbine equivalent inertia constant;
00
D1M is a friction loss coefficient (damping), Twt
is the equivalent mechanical torque; and r is the mechanical angular
speed (elect. rad/s).
C. Wind Turbine Modelling
The mechanical power generated for the WT is given by:

Topt = Kopt 2opt , opt

p
Popt
0
3
= Kopt
Popt
Kopt
s
p
Topt
00
=
= Kopt
Topt
Kopt

These direct and indirect

then used as a reference
of p
electrical converter
p [10],
00
1/ 3 Kopt , Kopt
= 1/ Kopt

(12)
(13)

power and torque curves can

in the active power control
0
once constants Kopt , Kopt
=
are calculated.

without exceeding the rated rotor current. In most practical

applications, the latter operating strategy will be preferred
because this gives a higher power output, resulting in a better
utilisation of the available power generating capacity [11].
Figure 6 shows the power flow diagram in the rotor circuit
for speeds below (a) and above (b) the synchronous speed.
Operating conditions in both subregion I and II can be
summarised as follow:
I)

Sr < s < 0, sPgs = Pgs + Pcur

(15)

II) 1 < s < Sr , sPgs > 0, Pgr > 0, sPgs = Pgs + Pcur
III. OVERVIEW OF POWER CONTROLLABILITY
Power flow in DFIG: a) sub-synchronous, b) supersynchronous mode (Pcur copper losses, Sr rated slip)

Fig. 6.

As can be seen the inverse curves follow cubic and square

root relations, respectively; but this square root can produce
numerical problems with negative quantities that can be translated into fluctuations in the control response.
D. Operating modes
The entire operating range of the WTGS can be divided into
two regions, namely sub and super-synchronous, respectively.
In the sub-synchronous region, the rotor speed is lower than
the synchronous speed and the slip (s) is between zero
and unity. The rotor power component (sPgs ), can be made
negative by injecting the slip-frequency power (Pgr ), into the
rotor circuit. The conventional IM will never be operated
as a generator at sub-synchronous speeds since (sPgs ) is
always positive [11]. At sub-synchronous operating mode the
following conditions are fulfilled:
0 < s < 1, sPgs < 0, Pgr < 0, sPgs = Pgs + Pcur

(14)

In the super-synchronous region, the rotor speed is greater

than the synchronous speed. This region can be divided into
two subregions. In subregion I, the rotor speed is between
the synchronous speed and the rated rotor speed, whereas in
subregion II, the rotor speed is higher than the rated rotor
speed, being this rated speed defined at the rated working
point. In sub-region II, the WTGS needs to be controlled
to avoid the excessive machine currents that would occur,
because the electrical power is obtained from both the stator
and the rotor as sum of them.
On the other hand, in the subregion I the WTGS is
working in the super-synchronous mode, and electrical power
can be either injected to or extracted from the rotor circuit
for generator operation. If power is extracted from the rotor
circuit, the total power output of the machine drops with
respect to the conventional IM. Otherwise, when power is
injected into the rotor circuit, an IM can generate more
power to the grid than in conventional use. This is because
the rotor current can be increased above the typical value

Generally, the most important objectives of the Wind Energy

Conversion System (WECS) control are the following:
controlling the wind captured power for speeds higher
than the rated speed
maximising the wind collected power in partial load as
long as constraints on speed and captured power are met
alleviating the variable loads, in order to guarantee a
certain level of stretchiness of the mechanical parts
meeting strict power quality standards (power factor,
harmonics, flicker, etc.)
transferring the electrical power to the grid at an imposed
level, for wide range of wind speeds
In a DFIG there is three main control subsystems [12].
The first one affects the pitch angle following aerodynamic
power limiting targets. The second implements the generator
control, in order to obtain the variable-speed regime and the
third controls the transfer of the full (or a fraction) of electric
power to the electric grid.
A. Aerodynamic control subsystem
All WTs are designed with a suitable power control. There
are different ways to control aerodynamic forces on the turbine
rotor and thus to limit the power in very high winds in order
to avoid damage to the WT [1]. The simplest, most robust and
cheapest control method is the stall control (passive control),
where the blades are bolted onto the hub at a fixed angle.
The design of rotor aerodynamics causes the rotor to stall
(lose power) when the wind speed exceeds a threshold, and
the aerodynamic power on the blades is limited. Such slow
aerodynamic power regulation causes less power fluctuations
than a fast-pitch power regulation. Some disadvantages of the
method are lower efficiency at low wind speeds, no assisted
start-up and variations in the maximum steady-state power due
to variations in air density and grid frequencies. Another type
is the pitch control (active control), where the blades can be
turned out or into the wind as the power output becomes too
high or too low, respectively. Generally, the advantages of this
type of control are good power control, assisted start-up and
emergency stop.
Some drawbacks are the extra complexity arising from
the pitch mechanism and the higher power fluctuations at
high wind speeds. The instantaneous power will, because of
gusts and the limited speed of the pitch mechanism, fluctuate

Maximum Power Point Tracking (MPPT)

This scheme has the goal to operate the WECS around
the maximum power (within safety limits), using data from
the static power characteristic and a minimum of information
from the system. The WT power characteristic is unknown,
but general features like rated power, rated rotor speed, etc.
are considered known. The high-speed shaft rotational speed
(g ) and the generator active power (P ), are the only available
measurements from the system. Basically, the approach is
based on the computation of the power and rotational speed
gradients. These data are used to determine (P/g ) value,
its sign corresponding to the position of the static operating
point (OP) on the power curve in relation to the maximum
of this curve (zero corresponds to the power maximum). The
variable-speed control system appears in Figure 9a.
A simple control logic aiming at keeping (P/g ) at small
values, updates the speed reference of the generator vector
control block with constant variations (g ). Since the OP
excursions around the optimal operating point (OOP) are quite
large under turbulent winds, versions of this method employing
wind speed estimation or adaptive control input can be needed,
depending on the closeness and its trend of the OP to the OOP.
Even if it is robust subject to WECS parametric uncertainties
and does not need much information, the main inconvenient is
the use of gradient estimations of some variables in dynamic
conditions, [12].
Fuzzy-logic Control
Maximizing the available power from the wind (in the same
modelling assumptions) is aimed at in this case too, and
the control law (evaluation of premises according to control
rules) is an extension of the MPPT method, basically in the
control logic. The rules base is therefore built for keeping
the OP around the OOP (at a small value of P/g ). Value
(g ) is variable for a certain wind speed and depends on the
distance between the optimal and the current OP and on the
speed variation of the latter [12]. In this way small variations
of OP around OOP are obtained in steady- state operation.
The dynamic response is also improved; the fuzzy controller
is more flexible than an MPPT one. The most important
drawback of this method is the context dependence (the wind
site features, the turbine type, etc.), requiring quite consistent
a initial knowledge. The corresponding variable-speed control
system is shown in Figure 9b.
Sliding-mode Control (SMC)
The sliding-mode is a robust control method, suitable for
nonlinear systems; the controller is a variable structure system
which switches with high-frequency between several control
laws. The controller implements a nonlinear control (switching) law, in order to drive and keep the system state trajectory
on a desired (switching) hypersurface, [12]. There is a certain
difficulty about the control design, concerning the definition of
a sliding surface with guaranteed properties of attractiveness
and stability. Flexibility of the sliding-mode approach can be
raised by adopting combined switching surfaces, in order to
allow multi-criteria optimization (i.e., captured power maximization and minimization of the electromagnetic torque variations). This method is effective and quite robust, requiring rel

Fig. 7. Comparison between different stall and pitch control features.

Fig. 8.

Control structure of a VSWT with DFIG [13].

around the rated mean value of the power. The third option
is the active stall control, i.e. the stall of the blade is actively
controlled by pitching the blades. At low wind speeds the
blades are pitched similar to a pitch-controlled WT, in order
to achieve maximum efficiency. At high wind speeds the
blades go into a deeper stall by being pitched slightly into the
direction opposite to that of a pitch-controlled WT. The active
stall WT achieves a smoother limited power, without high
fluctuations. This control type has the advantage of being able
to compensate variations in air density. The combination with
the pitch mechanism makes it easier to carry out emergency
stops and to start up the WT.
The main tasks of the wind wheel (turbine) control system
are to keep the value of the rotor shaft speed and the active
power at a set levels. Therefore, the control system usually
consists of speed, power and blade pitch angle regulators, as
can be seen in the example in figure 8.
B. Speed control subsystem
When the wind speed is under its rated value, the DFIG
is working at partial load, being this state the most regular
operating mode of the WT, because the wind speed is usally
lower than the rated value. In high power WT systems is
very important to implement a specific control system to
improve the energy efficiency of the wind power. There is
three methods, at least, to carry out the task of the control
system. Regarding the optimal power curve, the rotor speed is
related to the maximum power by a optimal constant, as can
be seen in subsection II-C.

QG

/t

P/g

1/s

Logic

PG

Calcul.

IqG

VabcG

Fault
TRF - L

3 --> 2

VdG

R wire

pcc

RL filter

PR, QR

ref

Defuzzification

RSC

RL filter

DFIG

PI

Tg

IdR

S_C

Torque

3 --> 2

IqR

Calqul.

Generator
Reference

IabcR

Crowbar

DC link

GSC

PWM

PWM

VabcR

Equivalent
control
input

ref

eq
+
+

Alternate
Control
input

P*
-

PS, QS

g*

b)

1/s

Control
rules
evaluation

Fuzzification

/t

Tg

PI

PG, QG

IabcG

3 --> 2

VqG

a)

Grid

TRF - H

IdG

PQ

RL wire

Gradient
detection

1/s

T_C
Q_C

VdR

Id_C
+

Iq_C

VqR

2 --> 3

Vdc

GCC

IRn

CVC

Qref

Tg

Fig. 11.

Block diagram of the DFIG control system.

Fig. 12.

Modulation control methods: a) PS: b) HB (c) TC.

c)

Fig. 9.

Fig. 10.

Block diagrams of methods: a) MPPT, b) Fuzzy, c) SMC.

Back-to-back connected power converter bridges.

atively few information about the system and being insensitive

to parametric variations. The chattering, specific to SMC, is the
main drawback because it influences negatively the mechanical
subsystem by inducing supplementary stress, and might excite
the unmodelled dynamics (e.g., the oscillating modes of the
blades or of the flexible drive train), thus producing dangerous
oscillations. However, ensuring a sufficiently high chattering
frequency is absolutely necessary. A typical block diagram of
the SMC method is included in Figure 9c.
C. Power control subsystem
Power electronic includes devices such as soft-starters, rectifiers, inverters and frequency converters, and there are different
design topologies for such converters [14]. Self-commutated
converters are either voltage or current source converters
(VSCs or CSCs). They can control both the frequency and
the voltage, moreover with high switching frequencies the
harmonics can be filtered out more easily.
The control system of the DFIG is more complex than
the SFIG control system [6]. That system consists of a
grid frequency side and a rotor side converter (GSC, RSC)
control systems. These power converters (figure 10) enable the
generator to operate above and below the synchronous speed.
The scheme of the control system is included in figure 11.
The GSC controller regulates the intermediate-circuit param-

eters: VDC voltage (VSC) or IDC current (CSC), and the

output currents of the GSC. The reactive power on the GSC
is usually kept close to zero. Then the RSC enables control
of the reactive power consumed/produced by the WTGS. The
RSC control system consists of a set of controllers that allow
regulate the real power/speed and the reactive power [6]. In
general the control system is composed of several subsystem as
Speed Control (S C), Torque Control (T C), Rotor direct and
quadrature axis Current Controllers (Id C, Iq C), Grid Current
Controller (GCC), Circuit Voltage Controller (CVC), ABC
to DQ transformation (3 2), DQ to ABC transformation
(2 3). The scheme presented implements the speed versus
power curve, but other control strategies can be used as speed
vs torque characteristic in [8].
Pulse Width Modulation schemes
In order to drive these power converters, there are many modulation methods [15], but three methods for VSC controlled
rectifiers are the most widely used: Periodical Sampling (PS),
Hysteresis Band (HB), and Triangular Carrier (TC).
The PS method switches the power transistors of the rectifier during the transitions of a square wave clock of fixed
frequency: the PS frequency. In each transition, a comparison
between Iref and Iline is made, and corrections take place.
As shown in Fig. 12a, this type of control is very simple

Fig. 13.

Typical converters for crowbar protection.

to implement: only a comparator and a D-type flip-flop are

needed per phase. The main advantage is that the minimum
time between switching transitions is limited to the period of
the sampling clock. However, the actual Fsw is not clearly
defined. The HB method switches the transistors when the
error between Iref and Iline exceeds a fixed magnitude:
the HB. As seen in Fig. 12b, this type of control needs a
single comparator with hysteresis per phase. In this case the
Fsw is not determined, but its maximum value is knew and
bounded. The TC method, shown in Fig. 12c, compares the
error between Iref and Iline with a triangular wave. This
triangular wave has fixed amplitude and frequency and is
called the TC. The error is processed through a PI gain stage
before the comparison with the triangular carrier takes place.
This control scheme is more complex than PS and HB.
Crowbar protection circuits
When a voltage dip occurs in the DFIG stator voltage,
the rotor current regulators lose the control, and an energy
flow from the stator to the rotor charges the dc-link capacitor.
To avoid the bus voltage reaches the converter limits, it is
necessary to break this energy flow, and the simplest method
is to short circuit the rotor when the dc-link voltage reaches
a limit value [16], [17]. The crowbar protection can be
implemented in many ways depending of the power converter
structure and the desired performances. Figure 13 shows the
most used structures of the crowbar protection circuits. The
first converter consists of a non-controlled 3-phase rectifier
(6 diodes), a controlled power switch (thyristor or transistor)
and a power resistance. This solution is simple, low-cost and
easy to control. The second converter is composed by a totalcontrolled 3-phase rectifier (6 thyristors or transistors), and
Y -connected three power resistances. This solution is robust,
more expensive and more difficult to control than the other.
IV. C ONCLUSION
Once the WT topologies about fixed and variable speed
operation are presented, the paper is focused in the DFIG modelling, since this IM is more implemented currently in Europe
wind farms. Concerning the order of the electrical model of
the DFIG is important for the control system performance. The
simplified model is useful in the initialization of the electrical
variables, when the machine is working at partial or full load,
since for each wind and rotor speeds, the generator reaches
the rated voltages, currents and powers (active, reactive). The
detailed model must be used during start-up of the WT or

when the wind speed changes suddenly either when the grid
connection of the WT is affected by some type of electrical
fault too. Regarding the order of the mechanical model, it is
proved that as wind speed is intermittent and stochastic in
nature, the torques acting on blades of wind turbines are not
always equal. This effect can only be analysed through the
6-mass drive train model. But this mechanical model slows
the simulations due to complex and lengthy mathematical
computation with small time-steps. Moreover, the unequal
blade torque distributions do not have effect on the transient
stability of WTGS, which encourages the consideration of the
reduced order 3-mass model [7], although 2-mass model is
the most used in the references. Finally, control systems of
the WT and DFIG are included, but they usually vary, since
their structures and parameters depend on the WTGS specific
model, the rated power and the rotor speed range variation [6].
R EFERENCES
[1] T. Ackermann, Wind Power in Power Systems. John Wiley & Sons,
Ltd,, 2005.
[2] A. Hansen, Generators and power electronics for wind turbines, John
Wiley & Sons, Ltd, p. 24, 2004.
[3] A. Feijoo, J. Cidras, and C. Carrillo, A third order model for the doublyfed induction machine, Electric Power Systems Research, vol. 56, pp.
121127, 2000.
[4] G. Slemon, Modelling of induction machines for electric drives, IEEEJ-IA, vol. 25, no. 6, pp. 11261131, 1989.
[5] A. Petersson, L. Harnefors, and T. Thiringer, Evaluation of current
control methods for wind turbines using doubly-fed induction machines,
IEEE Transactions on Power Electronics, vol. 20, pp. 227 235, 2005.
[6] Z. Lubosny, Wind Turbine Operation in Electric Power Systems. Advanced modeling. Berlin, Germany, Springer-Verlag, 2003.
[7] S. Muyeen, M. Hasan Ali, and R. Takahashi, Comparative study
on transient stability analysis of wind turbine generator system using
different drive train models, Renewable Power Generation, IET, vol. 1,
pp. 131141, 2007.
[8] J. Ekanayake, L. Holdsworth, and W. XueGuang, Dynamic modeling
of doubly fed induction generator wind turbines, IEEE Transactions on
Power systems, vol. 18, pp. 803809, 2003.
[9] R. Datta and V. Ranganathan, Variable-speed wind power generation
using doubly fed wound rotor induction machine-a comparison with
alternative schemes, IEEE Transaction on Energy Conversion, vol. 17,
no. 3, pp. 414421, Sept. 2002.
[10] A. Pujante, E. Gomez, and J. Fuentes, A dfig dynamic model for its
use on wind power generation studies, International Conference on
Sustainable Energy Technologies (ICSET), pp. 1028 1033, 2008.
[11] I. Cadirci and M. Errniq, Double-output induction generator operating at subsynchronous and supersynchronous speeds: steady-state
performance optimisation and wind-energy recovery, IEE ProceedingsElectric Power Applications, vol. 139, pp. 429 442, 1992.
[12] I. Munteanu, A. I. Bratcu, N.-A. Cutululis, and E. Ceanga, Optimal
Control of Wind Energy Systems: Towards a Global Approach, 1st ed.
Springer, March 2008, iSBN-13: 978-1848000797.
[13] E. Hau, Wind turbines. fundamentals, technologies, application, economics, Springer, 2006.
[14] D. Novotny and T. Lipo, Vector control and dynamics of ac drives,
Clarendon Press, Oxford,, 1996.
[15] N. Mohan, T. Undeland, and W. Robbins, Power electronics: Converters,
Applications, and Design. 2nd Edition, John Wiley & Sons, Inc., 1995.
[16] M. Rodriguez, G. Abad, I. Sarasola, and A. Gilabert, Crowbar control
algorithms for doubly fed induction generator during voltage dips, EPE
2005 Dresden, pp. 110, 2005.
[17] J. Morren, W. Sjoerd, and H. de Haan, Short-circuit current of wind
turbines with doubly fed induction generator, IEEE transactions on
Energy conversion, 2007.