New Approach for modelling Doubly-Fed Induction Generator (DFIG) for grid-connection studies

S. K Salman and Babak Badrzadeh

School of Engineering, The Robert Gordon University,
Schoolhill, Aberdeen, AB10 1FR, Scotland, U.K.
Tel :+(0) 1224 262448
Fax :+(0) 1224 262444
Email: s.k.salman@rgu.ac.uk

Summary:

With increased penetration of wind power into electrical grids, DFIG wind turbines are largely deployed due to their
variable speed feature and hence influencing system dynamics. This paper presents simulation results of a Grid-connected
DFIG. A switch-by-switch representation of the PWM converters with a carrier-based Sinusoidal PWM modulation for
both rotor- and stator-side converters has been proposed. Stator-Flux Oriented vector control approach is deployed for
both stator- and rotor-side converters to provide independent control of active and reactive power and keep the DC-link
voltage constant. In order to set synchronous vector controllers, decoupled design based on Internal Model Control
approach is applied, where dynamics of the PWM converters is taken into account.

Keywords:

Grid integration, Variable-speed wind turbine, Doubly-Fed induction generator, Modeling and simulation, fault analysis,
Vector control, back-to-back PWM converter, Internal Model Controller.

Nomenclature:

p is the derivative symbol

Vqs , Vds are the three-Phase supply voltages in d-q reference frame, respectively
iqs , ids are the three-Phase stator currents in d-q reference frame, respectively
λqs , λds are the three-Phase stator flux linkages in d-q reference frame, respectively
Vqr ,Vdr are the three-Phase rotor voltages in d-q reference frame, respectively
iqr , idr are the three-Phase rotor voltages in d-q reference frame, respectively
λqr , λdr are the three-Phase rotor voltages in d-q reference frame, respectively
rs , rr are the stator and rotor resistances of machine per phase, respectively
Lls , Llr are the leakage inductances of stator and rotor windings, respectively
ωe , ωr are the supply and rotor angular frequency (electrical speed), respectively
θ s ,θ r are the stator and rotor flux angle, respectively
Te , Tm are the electromagnetic and mechanical torques, respectively
Ps , Qs are the stator-side active and reactive powers, respectively
Pr , Qr are the rotor-side active and reactive powers, respectively
RON , ROFF are the IGBT ON and OFF resistances, respectively
D, J are the moment of inertia and damping coefficient, respectively
P is the Number of poles
M 1, M 2 is the stator and rotor modulation depths, respectively
Vtri is the triangular Voltage Signal
R, L are the resistance and inductance of input filter, respectively
V1, I1 are the input filter line voltage and current, respectively
E is the DC-link voltage
s is the Laplacian Operator
C is the DC-Link capacitance
PDC is the DC-link active power

1
σ is the leakage factor
k p , ki are the proportional and integral gain, respectively
α is the bandwidth of the closed-loop system
α cs is the Bandwidth of supply-side current controller
α E is the bandwidth of DC-link voltage controller
α cr is the bandwidth of rotor-side current controller
αω is the bandwidth of speed controller

Subscripts and superscripts:

comp stands for compensation term

e identifies synchronously rotating reference frame

1. Introduction:

With increased penetration of wind power into electrical grids, DFIG wind turbines are largely deployed due to their
variable speed feature and hence influencing system dynamics. This has created an interest in developing suitable models
for DFIG to be integrated into power system studies. The continuous trend of having high penetration of wind power, in
recent years, has made it necessary to introduce new practices. For example, grid codes are being revised to ensure that
wind turbines would contribute to the control of voltage and frequency and also to stay connected to the host network
following a disturbance.

In response to the new grid code requirements, several DFIG models have been suggested recently [1-6], including the
full-model which is a 5th order model. These models use quadrature and direct components of rotor voltage in an
appropriate reference frame to provide fast regulation of voltage. The 3rd order model of DFIG which uses a rotor current
[4], not a rotor voltage as control parameter can also be applied to provide very fast regulation of instantaneous currents
with the penalty of losing accuracy. Apart from that, the 3rd order model can be achieved by neglecting the rate of change
of stator flux linkage (transient stability model), given rotor voltage as control parameter [5-7].

Additionally, in order to model back-to back PWM converters, in the most simplest scenario, it is assumed that the
converters are ideal and the DC-link voltage between the converters is constant [5-11]. Consequently, depending on the
converter control, a controllable voltage (current) source can be implemented to represent the operation of the rotor-side
of the converter in the model. However, in reality DC-link voltage does not keep constant but starts increasing during fault
condition. Therefore, based on the above assumption it would not be possible to determine whether or not the DFIG will
actually trip following a fault.

In a more detailed approach, actual converter representation with PWM-averaged model has been proposed [2], where the
switch network is replaced by average circuit model, on which all the switching elements are separated from the
remainder of network and incorporated into a switch network, containing all the switching elements. However, the
proposed model neglects high frequency effects of the PWM firing scheme and therefore it is not possible to accurately
determine DC-link voltage in the event of fault. A switch-by-switch representation of the back-to-back PWM converters
with their associated modulators for both rotor- and stator-side converters has been proposed [4]. Tolerance-band
(hysteresis) control has been deployed. However, hysteresis controller has two main disadvantages: firstly, the switching
frequency does not remain constant but varies along the AC current waveform and secondly due to the roughness and
randomness of the operation, protection of the converter is difficult [13]. The latter will be of more significance when
assessing performance of the system under fault condition.

In order to resolve the identified problems, this paper proposes a switch-by-switch model of voltage-fed, current
controlled PWM converters, where triangular carrier-based Sinusoidal PWM (SPWM) [13] is applied to maintain the
switching frequency constant. In order to achieve constant switching frequency, calculation of the required rotor voltage
that must be supplied to the generator is adopted.

Various methods such as hysteresis controller, stationary PI controller and synchronous PI controller have been adopted in
order to control current-regulated induction machine [13,20]. Among which, synchronous PI controller has been
acknowledged as being superior [20]. This paper makes use of Internal Model Control which is actually based on the
concept of synchronous PI controllers while different approach is used in order to specify control parameters for a certain
machine instead of using trail-and-error procedure [14]. With regard to IMC applications in DFIG control, [15] adopted

2
this method to determine control parameters of the rotor-side converter of a DFIG, assuming ideal operation of back-to-
back PWM converter. Moreover, the proposed approach may not be accurate enough due to the neglect of leakage factor.
The same approach has been used by [16], while acknowledging leakage factor of induction machine, with DFIG
controlled from the rotor-side. This paper will use the same approach, where dynamics of the PWM converters is taken
into account along with the control of the DFIG from both stator- and rotor-side converters.

2. Principles of Operation:

Fig.1 shows the basic scheme adopted in the majority of systems. The stator is directly connected to the AC mains, whilst
the wound rotor is fed from the Power Electronics Converter via slip rings to allow DIFG to operate at a variety of speeds
in response to changing wind speed. Indeed, the basic concept is to interpose a frequency converter between the variable
frequency induction generator and fixed frequency grid. The DC capacitor linking stator- and rotor-side converters allows
the storage of power from induction generator for further generation. To achieve full control of grid current, the DC-link
voltage must be boosted to a level higher than the amplitude of grid line-to-line voltage.

The slip power can flow in both directions, i.e. to the rotor from the supply and from supply to the rotor and hence the
speed of the machine can be controlled from either rotor- or stator-side converter in both super and sub-synchronous speed
ranges. As a result, the machine can be controlled as a generator or a motor in both super and sub-synchronous operating
modes realizing four operating modes [6]. Below the synchronous speed in the motoring mode and above the synchronous
speed in the generating mode, rotor-side converter operates as a rectifier and stator-side converter as an inverter, where
slip power is returned to the stator. Below the synchronous speed in the generating mode and above the synchronous
speed in the motoring mode, rotor-side converter operates as an inverter and stator-side converter as a rectifier, where slip
power is supplied to the rotor. At the synchronous speed, slip power is taken from supply to excite the rotor windings and
in this case machine behaves as a synchronous machine.

P s + j Qs
Rotor Grid

Stator

Back-to-back Pr + j Qr
PWM converter
Fig.1.Schematic Diagram of Doubly-Fed Induction Generator

3. Dynamic simulation of DFIG in terms of dq-windings:

The general model for wound rotor induction machine is similar to any fixed-speed induction generator as follows [17]:

3.1. Voltage equations:

3.1.1. Stator Voltage Equations:

Vqs = pλ qs + ωλ ds + rs i qs (1)

Vds = pλ ds − ωλ qs + rs i ds (2)

3.1.2. Rotor Voltage Equations:

V qr = pλ qr + (ω − ω r )λ dr + rr i qr (3)
V dr = pλ dr − (ω − ω r )λ qr + rr i dr (4)

3.2. Power Equations:

3
Ps = (V i + V qs i qs ) (5)
2 ds ds
3
Q s = (V qs i ds − V ds i qs ) (6)
2

3
3.3. Torque Equation:
3P
Te = − (λ ds i qs − λ qs i ds ) (7)
2 2

3.4. Flux Linkage Equations:

3.4.1. Stator Flux Equations:

λ qs = ( Lls + Lm )i qs + Lm i qr (8)

λ ds = ( Lls + Lm )i ds + Lm i dr (9)

3.4.2. Rotor Flux Equations:

λ qr = ( Llr + Lm )i qr + Lm i qs (10)
λ dr = ( Llr + Lm )i dr + Lm i ds (11)

4. Back-to-Back PWM Modeling:

Mathematical modeling of converter system is realized by using various types of models, which can be broadly divided
into two groups: mathematical functional models and Mathematical physical models (either equation-oriented or graphic-
oriented, where graphic-oriented approach is actually based on the same differential equations) [18].

Functional model describes the relationship between the input and output signal of the system in form of mathematical
function(s) and hence constituting elements of the system are not modeled separately. Simplicity and fast time-domain
simulation are the main advantages of this kind of modeling with the penalty of losing accuracy. This has been a popular
approach with regard to DFIG modeling, where simulation of converters has been done based on expected response of
controllers rather than actual modeling of Power Electronics devices. In fact, it is assumed that the converters are ideal
and the DC-link voltage between them is constant. Consequently, depending on the converter control, a controllable
voltage (current) source can be implemented to represent the operation of the rotor-side of the converter in the model.

Physical model, on the other hand, models constituting elements of the system separately and also considers inter-
relationship among different elements within the system, where type and structure of the model is normally dictated by the
particular requirements of the analysis, e.g. steady-state, fault studies, etc. Indeed, due to the importance of more realistic
production of the behavior of DFIG, it is intended to adopt physical model rather than functional model in order to
accurately assess performance of DFIG in the event of fault particularly in determining whether or not the generator will
trip following a fault. This paper proposes a graphic-oriented switch-by-switch representation of the back-to-back PWM
converters with their modulators for both rotor- and stator-side converters (Fig.2), where both IGBT and reverse diode
devices are represented as a two-state resistive switch [19]. The two-state switch can take on two values, RON (close to
zero) and ROFF (very high).
Ios Ior

D1 T1 D T1 D1 T1 T1 D1 T1 D1 T1 D1

S1 S3 S5 R1 R3 R5
500000.0

D1 T D1 T1 D1 T1 T2 D2 T2 D2 T2 D2

S4 S6 S2 R4 R6 R2
Va1

Vb1

Vc1
0.01

0.01

0.01
0.1

0.1

0.1

Ibr

Iar
Icr
Ias

Ibs

Ics

Fig.2 Back-to-back PWM converter structure

4
5. Control of generator and its associated converters:

5.1 Rotor-side converter control:

Aligning the d-axis of reference frame to be along the stator flux linkage (stator flux oriented control) will result in [12]:
λeqs = 0 (12)
And hence from Eq. 8:
Lm
e
i qs =− e
i qr (13)
Lls + L m
Substituting for iqse into Eq. 7 will result in:
3P Lm
Te = − λ eds i qr
e
(14)
2 2 Lls + L m
For λ ds to remain unchanged at zero, pλ ds must be zero [16]. Substituting for pλ ds using Eqs. 1 and 2 will result in
e e e

V dse = rs i ds
e
. Neglecting stator resistance will lead to V ds = 0 . Substituting for Vds = 0 , Eqs. 5 and 6 will be simplified
e e

as follows:
3 e e
Pse = (Vqs i qs ) (15)
2
3
Qse = (Vqse i ds
e
) (16)
2
Therefore, the above equations show that active and reactive powers of the stator can be controlled independently.

5.2 Decoupling control scheme:

Using stator flux oriented approach, implementation with current controlled PWM inverter requires decoupling scheme
[13]. In fact, the system is coupled because the inductance matrix is not diagonal. It means that any changes on voltage
component in d or q axes results in changes in both current components [13]. In order to combat this problem, equations
will be re-developed in order to compensate for these cross coupling between d and q axes [12, 20]. Let us first define the
following parameter called the leakage factor of the induction machine as [12]:
L2m
σ = 1− (17)
( Lls + Lm )( Llr + Lm )
Substituting for i ds from Eq. 9 into Eq. 11 for λ dr and from Eq. 13 into Eq. 10:
e e

Lm
λ edr = σ ( Llr + Lm )i dr
e
+ λ eds (18)
Lls + L m
λeqr = σ ( Llr + Lm )i qr
e
(19)
e
Substituting these equations into Eqs. 3 and 4 for V qre and V dr :

V dre = V dr'e + V dre ,comp (20)

V qre = V qr'e + Vqre ,comp (21)

V dr'e = e
rr i dr + σ ( Llr + e
L m ) pi dr (22)
V qr'e = e
rr i qr + σ ( Llr + e
Lm ) pi qr (23)
Lm
V dre ,comp = pλ eds − (ω e − ω r )σ ( Llr + L m )i qr
e
(24)
Lls + L m
Lm
V qre ,comp = (ω e − ω r ) λeds + (ω e − ω r )σ ( Llr + Lm )i dr
e
(25)
Lls + Lm
Adding these compensating terms to the corresponding uncompensated voltage terms makes it possible to achieve
decoupled performance of the stator flux-oriented control of the rotor-side converter. Fig.3 shows stator field oriented

5
 control of rotor-side converter [20], where PI controllers are applied to control rotor current and shaft speed which will be
explained further. It is also worth noting that VR and T represent rotation and phase transformations, respectively [21].

V qre ,comp
C
ω rref
ref ,e
i qr Vqrref ,e Varref , s
PI PI
e
i qr Vbrref , s Rotor
ωr -1
VR (θsr) -1
T (θsr) PWM
Side
converter
ref ,e Vdrref ,e Vcrref , s
i dr PI
e
i dr
Vdre ,comp
e
i qr
i abc
e
idr VR(θsr) T(θsr)

θr
d/dt Encoder

DFIG
Fig.3 Stator flux oriented control of rotor-side converter

5.3 Stator-side (Front-End) converter (FEC) control:

Fig.4 shows the schematic of FEC control. The main objective of stator-side converter is to maintain the DC-link voltage
constant regardless of the magnitude and direction of the slip power. A current-regulated PWM scheme is used, where q
and d axes currents are used to regulate DC-link voltage and reactive power, respectively. Based on Fig.2 and taking into
account the inductance and resistance of the input filter, voltage equations across the inductor can be written as follows:
⎡Vas ⎤ ⎡i as ⎤ ⎡i as ⎤ ⎡Va1 ⎤
⎢V ⎥ = R ⎢i ⎥ + Lp ⎢i ⎥ + ⎢V ⎥ (26)
⎢ bs ⎥ ⎢ bs ⎥ ⎢ bs ⎥ ⎢ b1 ⎥
⎢⎣Vcs ⎥⎦ ⎢⎣ics ⎥⎦ ⎢⎣ics ⎥⎦ ⎢⎣Vc1 ⎥⎦
Applying phase and rotation transformations [21] to the above equation results in:
Vdse = Rids
e
+ Lpi ds
e
− ω e Li qs
e
+ Vde1 (27)

V qse = Ri qs
e
+ Lpi qs
e
− ω e Li ds
e
+ V qe1 (28)
Therefore, reference values for front-end converter can be written as:
,e
V dref
1 = − Ri ds
e
− Lpi ds
e
+ (ω e Li qs
e
) (29)
,e
V qref
1 = − Ri qs
e
− Lpi qs
e
+ (ω e Li ds
e
− V qse ) (30)
Where the terms in brackets represent the decoupling terms. Neglecting harmonics due to the switching, the capacitor
leakage and the losses in the inductor resistance and converter itself [20]:
3
Eios = Vqse iqs
e
(31)
2
M1
Vqse = E (32)
2 2
3
ios = M1iqs e
(33)
4 2

6
 CpE = ios − ior (34)
Assuming rotor-side converter DC-link current as disturbance and combining Eqs. (33) and (34):
3M 1
E= ios (35)
4 2

E ω e Li ds
e

,e C
ref , e Vqref ,s
E ref i qs 1
Varef
1
PI PI
,s
e
i qs Vbref
1
VR-1(θe) T-1(θe) PWM FEC
ref , e ,s
V d1
Vcref
1
ref ,e
ids PI
e
i ds R, L
ω e Li qs
e

V qse Vabc
VR(θe) T(θe)

iqse i abc
idse VR(θe) T(θe)

Fig.4 Stator flux oriented control of stator-side converter

5.4 Current-Regulated (CR) modulation:

In this paper, a voltage-source, current controlled PWM converter is chosen. The main task of the control scheme in the
Current Controlled (CC)-PWM converter is to force the current to follow their reference signals. By comparing desired
and actual values of the phase currents, the CC generates the switching states for converter which decreases the current
errors. There are various ways to obtain the switching signals for the inverter switches in order to control the inverter
output current. The simplest is tolerance-band (hysteresis) control, where the actual current is compared to the desired
current in each phase leg of the converter output. However, hysteresis controller has two main disadvantages: firstly, the
switching frequency does not remain constant but varies along the AC current waveform and secondly due to the
roughness and randomness of the operation, protection of the converter is difficult [13]. The latter will be of more
significance when assessing performance of the system on fault condition.

In this paper, SPWM method is adopted, which is based on a triangular carrier signal [13]. By comparison of the common
carrier signal with three reference sinusoidal signals the switching instants of the IGBTs are defined. Utilizing SPWM
modulation technique makes it possible to resolve the identified disadvantages of hysteresis control. In order to maintain
the switching frequency within the switch-mode converter constant in a SPWM-modulated converter, it is a common
practice to calculate the required rotor voltages that the converter must supply to the machine, making the rotor currents
equal to their reference values [12]. This concept will be described in section 6.3.1. The error between reference and
actual current, then, is amplified or fed through a PI controller. This scheme of modulation is usually referred to as fixed-
frequency current control [19].

6. Controller Design:

6.1 Basics of Internal Model Controller (IMC):

∧
The idea behind IMC is to augment the error between system, G(S), and the model of the system, G (S ) , by a transfer
function C(S). More details can be found in [13, 14, 16]. For a first-order system the controller is an ordinary PI controller
which can be represented as follows [22]:

7
 −1
ki α ∧
F (s) = k p + = G ( s) (36)
s s
∧
Where G (s ) is the transfer function of the model of the system [20].

6.2 Supply-Side Converter

In this simplified design the small time constants such as power converter dead time, feedback filter and digital signal
processing delay are neglected [13]. Thus, only the dynamics of supply-side filter is taken into account:

6.2.1 Current Controller:

+ PI Plant
I sref Converter Is
Controller
-

Fig.5 Current-control loop of stator-side converter

Fig.5 shows the current-control loop of stator-side converter. Based on Eqs. 29 and 30 and considering compensation
terms as disturbance, the plant for the current control loop is given by [20]:
e
^ e
ids (s) iqs ( s) 1
G ( s) = = = (37)
( R + Ls )ids
e
(s) ( R + Ls )iqs
e
(s) R + Ls
Given converter model as [13]:
M 1E
k cs = (38)
2Vtri
And substituting for Eq. 35 :
α cs L
kp = (39)
k cs
And
α cs R
ki = (40)
k cs

6.2.2 DC-link Voltage controller:

+
E ref PI DC-Link Plant E
Controller Constant
-

Fig.6 DC-link voltage-control loop

Based on fig.6 and given DC-Link transfer characteristic from Eq. 35 and plant transfer function as [20]:
3M 1
E= ios (41)
4 2
And

8
 ^ 1
G(S ) = (42)
Cs
And substituting for Eq.36 leads to:
4 2α E C
kp = (43)
3M1
There will be a remaining error when the induction machine is loaded and active power flows between DC-link and the
machine. The remaining error can be eliminated by adding an integrator to the DC-link voltage controller [22]. The
following is often adopted for the selection of the controller integration time:
10
Ti = (44)
αE
kp
Where Ti = (45)
ki

6.3 Rotor-Side Converter:

6.3.1 Current Controller:

As mentioned previously, in order to keep the switching frequency constant, it is necessary to calculate the required rotor
voltages. As shown in the block diagram of Fig.7, it is possible to generate the reference voltages from given reference
currents. In fact, the i qr and i dr errors are processed by the PI controller to give Vqr and Vdr , respectively. Using Eqs. 22
and 23, the plant can be represented by transfer functions below:
1 e
e
idr ( s) = V ' ( s) (46)
rr + sσ ( Llr + Lm ) dr
1 e
e
iqr ( s) = V ' ( s) (47)
rr + sσ ( Llr + Lm ) qr
Therefore, the transfer function of the Plant can be given by:
^ 1
G(s) = (48)
rr + sσ ( Llr + Lm )

+ Vr'
I rref PI Plant Ir
Controller
-

Fig.7 Designing current-control loop for constant switching frequency

Incorporating dynamics of PWM converter makes the above block diagram shown below.

+ Plant
I rref PI Converter Ir
Controller
-

Fig.8 Current-control loop of rotor-side converter including dynamics of PWM

Where converter can be represented as [13]:

9
 M 2E
k cr = (49)
2Vtri
This will determine control parameters as follows:
α cr σ ( Llr + Lm )
kp = (50)
k cr
And
α cr rr
ki = (51)
k cr

6.3.2 Speed Controller:

+
ω rref PI Plant ωr
Controller
-

Fig.9 speed-control loop of DFIG

Assuming single-mass representation of the mechanical drive train and using Fig.9, plant transfer function is given by
[20]:
^ 1
G(S ) = (52)
JS + D
Therefore:
α ω JP
k pω = (53)
2
And
k iω = α ω D (54)

7. Results and discussion:

7.1 Test system:

Simulation is carried out using PSCAD/EMTDC simulation tool. The performance of a 2.2 MVA, 690 V, 50 Hz, 4- pole
DFIG is investigated before, during and after fault by applying a constant -1 p.u mechanical torque. Mechanical damping
is set to 0.02 p.u to compensate for friction and windage losses. Reference values are set in such a way that Sub
synchronous generating mode with +0.28 slip is realized. For sake of the simplicity, shaft is represented as a single-mass
model using the swing equation. In order to limit fault currents/voltages, stator-to-rotor turn ratio is set to 0.4333.

Fig.10 shows the power system used to investigate the performance of the DFIG under fault condition. A boost-up 0.69/11
kV, 2.5 MVA transformer with leakage inductance of 0.06 p.u connects the generating system through the distribution
cables to the consumers and the rest of the network, where the latter is represented as a three-phase voltage source with a
short circuit level of 100 MVA and X to R ratio of 10. A 3LG short-circuit fault with a fault resistance of 0.01 Ω is
applied to the network at t=7 sec for 470 msec. As can be seen in Fig.11, the dip in the stator voltage is about 85 %
representing one of the most severe fault conditions under which DFIG should withstand.

~ G
Grid 11kV line 11/0.69 DFIG
Transformer
Load
Fig.10 Block diagram of grid-connected induction wind generator

10
7.2 Simulation results:

As can be observed in Figs. 11 and 12, immediately after fault, stator flux and consequently stator voltage will drop,
depending on several factors such as fault type, fault location, fault impedance, short circuit level, X to R ratio ratio, etc.
Flux linkage Grid Voltage
Lambda (mWb) Vg (p.u)
0.0030 1.20

1.00
0.0020
0.80

0.0010 0.60

0.40
0.0000
0.20

-0.0010 0.00
6.0 7.0 8.0 9.0 10.0 11.0 12.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

Fig11. Stator Flux Fig12. Stator Voltage

Using Eq.14, it can be concluded that Te is proportional to λ ds yet with negative sign. Therefore, using stator flux
orientation and given behavior of λ ds = λ in Fig.10, an increase in electromagnetic torque, shortly after fault occurrence,
is expected to observe. This characteristic can be confirmed by Fig.13. This increase in electromagnetic torque will result
in the acceleration of the rotor based on the swing equation, as mechanical torque is assumed to be fairly constant. As
depicted in Fig.14, upon fault clearance, rotor speed starts decreasing indicating appropriate action of the control system.
However, in contrast to the electromagnetic torque, rotor speed does not immediately restore its pre-fault value, but
decelerating for a few seconds until being stable. The main reason is that current control loops are designed to be much
faster than speed control loop, e.g. 100 times faster, as fast speed control loops are susceptible to the noise [20].

Electromagnetic Torque Rotor speed

TE (p.u) Omega (p.u)
0.50

0.900
0.00

0.850
-0.50
0.800
-1.00
0.750

-1.50
0.700

-2.00 0.650
6.0 7.0 8.0 9.0 10.0 11.0 12.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

Fig.13 Electromagnetic torque Fig14 Rotor speed

Clearly, immediately after the occurrence of fault condition rotor current and hence DC-link current increase. However,
due to low voltages of DFIG terminals, stator-side converter can not exchange this extra current to the network and
therefore active power falls toward zero (Fig.15). This will, in turn, lead to the accumulation of the charges and
consequently DC-link voltage will rise. Shortly after fault clearance, rotor and stator current will restore their pre-fault
values and so does DC-link voltage ensuring stable performance of the DFIG under transient conditions as shown in Fig.
16.

Stator and rotor fault currents are depicted in Figs. 17 and 18, showing that fault inception produce large currents at stator
terminals and consequently, due to the existence of magnetic coupling between rotor and stator, at rotor terminals.
However, in contrast to conventional fixed-speed induction generator, fault currents do not decay very fast. This might be
due to the action of controllers along with the impact of PWM modulators.

11
 Total active power DC-Link voltage
P_GEN (kW) Ecap (kV)
3.0k
0.850
2.5k
0.825
2.0k 0.800
0.775
1.5k
0.750
1.0k 0.725
0.700
0.5k
0.675
0.0 0.650
6.0 7.0 8.0 9.0 10.0 11.0 12.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0

Fig.15 Total active power Fig.16 DC-Link voltage

Stator curent Rotor current

Isa (kA) Ira (kA)
4.0 2.00
3.0 1.50
2.0 1.00

1.0 0.50
0.0 0.00
-1.0 -0.50
-2.0 -1.00

-3.0 -1.50
-4.0 -2.00
6.80 6.90 7.00 7.10 7.20 7.30 7.40 7.50 7.60 7.70 6.50 7.00 7.50 8.00 8.50 9.00

Fig.17 Stator current Fig.18 Rotor current

8. Conclusions:

With increased penetration of wind energy and moving toward active networks, grid codes are being revised to reflect the
new requirements. This has created an interest in developing more detailed models particularly with respect to fault
analysis. This paper has proposed a detailed 5th order model of grid-connected DFIG and its associated controllers.
Incorporation of stator flux transients into the induction machine model along with the switch-by-switch representation of
sinusoidally modulated power electronics converters allows a more realistic prediction of the behavior of DFIG in the
event of fault. Owing to the fact that DFIG controls have a significant influence on the system dynamics, vector control is
applied for both stator- and rotor-side converters to increase the degree of controllability, where fixed-frequency Internal
Model Controller approach is adopted to design the controllers precisely. In this way, a fairly robust fault-tolerant system
is achieved ensuring transient stability of the system.

Acknowledgement:

The authors would like to thank The Robert Gordon University for providing facilities. Babak Badrzadeh is thankful for
the Robert Gordon University for their financial support to undertake this research.
References:

1. Sun T, Chen Z, Blaabjerg F. Transient analysis of grid-connected wind turbines with DFIG after an external short-
circuit fault. 4th Nordic wind power conference, 2004, Chalmers University of technology, Sweden.
2. Koessler RJ, Pillutla S, Trinh LH. Integration of large wind farms into utility grids (Part 1-Modeling of DFIG). IEEE
Power Engineering Society general meeting, 2003; Toronto, Canada.
3. Pourbeik P, Koessler RJ, Dickmander DL. Integration of wind farms into utility grids. ibid; Toronto, Canada.
4. Kazakchov YA, Feltes JW, Zavadil R. Modelling wind farm for power system stability studies. ibid; Toronto, Canada
5. Ekanayake, J.B, Holdsworth, L, Wu, X., Jenkins, N. Dynamic modeling of Doubly Fed Induction generator wind
turbines. IEEE Transaction on Power Systems, 2003, 2:803-809.
6. Holdsworth L, Wu XG, Ekanayake JB, Jenkins N. Comparison of fixed speed and doubly-fed induction wind turbines
during power system disturbances. IEE Proceedings: Generation, Transmission, Distribution, 2003, 3: 343-352

12
7. Feijoo A, Cidaras J, Carrillo C. A third order model for the doubly-fed induction machine. Electric power systems
research journal,2002, 3:26-33
8. Slootweg JG, Polinder H, Kling WL. Dynamic modeling of a wind turbine with doubly fed induction generator. IEEE
Power Engineering summer meeting, 2001; Vancouver, Canada.
9. Rodriguez, JM, et al. Incidence on power system dynamics of high penetration of fixed speed and doubly fed wind
energy systems: Study of the Spanish case, IEEE Transaction on Power Systems, 2002, 4:1089-1095.
10. Thiringer T, Peterson A, Petru T. Grid disturbance response of wind turbines equipped with induction generator and
doubly-fed induction generator. IEEE Power Engineering Society general meeting, 2003; Toronto, Canada.
11. Tapia A, Tapia G, Ostolaza JX, Saenz JR. Modeling and control of a wind turbine driven Doubly Fed Induction
Generator. IEEE Transaction on Energy Conversion, 2003, 2:194-204.
12. Mohan N. Advanced electric drives: Analysis, control and modeling using SIMULINK; MNPERE, 2001.
13. Kazmierkowski MP, Krishnan R, Blaabjerg F. Control in Power Electronics; Academic press, 2001.
14. Harnefors L, Nee HL. Model-based current control of AC machines using the internal model control method. IEEE
Transactions on Industry Applications, 1998; 1:133-141.
15. Morren J, De Han SWH, Bauer P, Pierik JTG. Comparison of complete and reduced models of a wind turbine using
doubly-fed induction generator.10th European Conference on Power Electronics and Applications, 2003;Toulouse,France.
16. Peterson A. Analysis, modeling and control of Doubly-Fed Induction Generators for wind turbines. Licentiate thesis;
Chalmers University, Gutenberg, Sweden, 2003.
17. Ong CM. Dynamic Simulation of electric machinery using MATLAB/SIMULINK; Prentice Hall, 1998.
18. Lubosny Z. Wind turbine operation in electric power systems; Springer Verlag, 2004.
19. PSCAD/EMTDC V4.0 user’s manual, 2003.
20. Pena R, Clare JC, Asher GM. Doubly fed induction generator using back-to-back PWM converters and its application
to variable-speed wind-energy generation. IEE Proceedings on Electrical Power Applications, 1996; 3:231-241
21. Novotny DW, Lipo TA. Vector control and dynamics of AC drives; Clarendon Press, 1996.
22. Ottersen R. Vector control of a double-side PWM converter and induction machine drive. Licentiate thesis; Chalmers
University, Gutenberg, Sweden, 2000.

13