2 Comparative Study of DFIG Power Control Using Stator

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Comparative Study of DFIG Power Control Using StatorVoltage and Stator-Flux Oriented Frames

Shuhui Li, Senior Member, IEEE, Rajab Challoo, Member, IEEE and Marty J. Nemmers

Abstract This paper compares real and reactive power control


for a DFIG (doubly-fed induction generator) based wind turbine
system using stator-voltage and stator-flux oriented frames. The
paper presents DFIG steady-state and transient models in d-q
reference frame. The steady-state model is used to obtain the
general relationship between the rotor d/q current and stator
real/reactive power references using stator-flux and stator-voltage
oriented frames. The transient model, together with the analysis
based on DFIG steady-state d-q equivalent circuit, is used to
develop DFIG power control mechanisms. In the evaluation of
DFIG controllers, the wind turbine driving torque is modeled by
considering typical wind turbine aerodynamic characteristics
under variable wind and pitch angle conditions. Comparisons are
made to evaluate the theoretical difference of the DFIG
controllers using stator-voltage and stator-flux oriented frames.
Simulation results for a 2 MW DFIG system demonstrate the
performance of the controllers for variable wind speed and active
and reactive power references.
Index Terms doubly-fed induction generator, d-q vector
control, stator-flux-oriented frame, stator-voltage-oriented frame,
wind power control.

I.

INTRODUCTION

doubly-fed induction generator (DFIG) is an adjustablespeed induction machine widely utilized in the modern
wind power industry [1, 2]. The reasons for using variable
speed wind turbines are fourfold: a higher energy yield, a
reduction of mechanical loads and a simpler pitch control, an
extensive controllability of both active and reactive powers,
and less fluctuation in output power [2].
However, the performance of a DFIG depends not only
on the induction generator but also on the d-q vector control
applied to the generator as well as where the d-axis is aligned
in the controller design. The normal orientation frames used
in a DFIG system are the stator-flux and stator-voltage
oriented frames. In most traditional schemes, the control of
DFIG active and reactive powers is achieved through a nested
power- and current-loop controller using the stator-fluxoriented frame [3]. Under a close-loop d-q vector control
condition, it is normally believed that a DFIG can generate
electricity in a wide speed range both above and below the
synchronous speed [2]. In [3] and [4], DFIG control
mechanisms are reported using the stator-flux-oriented frame
Shuhui Li and Marty J. Nemmers are with the ECE Department, The
University of Alabama, Tuscaloosa, AL 35487.
Rajab Challoo is with the Department of Electrical Engineering and
Computer Science, Texas A&M University-Kingsville, Kingsville, TX 78363.

978-1-4244-4241-6/09/$25.00 2009 IEEE

with the position of the stator-flux space estimated through


the measurement of the stator-flux space vector in -
reference frame. In [5], a stator-flux-oriented DFIG control
strategy is proposed, in which the position of the stator-flux
space vector is estimated through the measurements of statorvoltage and rotor-current space vectors in - reference
frame. In [6], a stator-flux-oriented control of a cascaded
doubly-fed induction machine is proposed, in which one of
the main approaches used to estimate the position of statorflux space vector is to add a delay angle of 90 to the statorvoltage space vector. Other control approaches are also
proposed recently, such as direct-power-control strategies
using the stator-flux-oriented frame [7, 8]. Although the
stator-voltage-oriented frame is normally not used for general
DFIG real and reactive power control, [9] and [10] report
special control approaches to improve DFIG stability under
unbalanced conditions using the stator-voltage-oriented
frame. In [11], a cascaded DFIG configuration and a relevant
control mechanism are presented, in which one DFIG (power
machine) is controlled by the power converter indirectly
through the other DFIG (control machine). Although the
system configuration is a little bit different, most techniques
as reported in [11] are similar to those of traditional DFIGs.
As various approaches are being developed for control of
a DFIG system, it becomes important to investigate and
compare DFIG performance under general d-q control
strategy using stator-flux and stator-voltage oriented frames.
A variety of techniques have been developed to study the
behavior of a DFIG under different d-q control conditions.
Those techniques can be divided into two categories: 1)
transient approaches [2-11], and 2) steady-state methods [12,
13]. Transient approaches are essential to study DFIG
dynamic performance in a short time period. Steady-state
methods are important to examine DFIG characteristics under
different control rules in a broader spectrum.
This paper studies and compares DFIG power control
under general d-q vector control strategy through combined
steady-state and transient simulation techniques using statorvoltage and stator-flux oriented frames. In the sections that
follow, the paper first describes the fundamental operating
principles and a general d-q control mechanism of the DFIG.
Then, a DFIG steady-state model and equivalent circuit in d-q
reference frame is developed, which is then used to calculate
the rotor d and q current reference for a desired real and
reactive power reference. The rotor current d-q loop control
strategies using stator-voltage and stator-flux oriented frames
are developed based upon DFIG d-q transient model as well

as the analysis obtained from the d-q steady-state equivalent


circuit. Both analytical and performance studies are
conducted to compare the control techniques developed using
the two different orientation frames. Finally, the paper is
summarized in the conclusion section.

this range represent very high and low wind speeds,


respectively, that are normally outside the continuous rating
of the machine [15].
C p ( , ) =

i =0 j =0

(3)

ij

0.5

II. DFIG WIND TURBINE AND CONTROLS

= 3

= 1

0.4

= 5
0.3

= 7

Cp

A DFIG wind turbine primarily consists of three parts: a


wind turbine drive train, an induction generator (doubly fed),
and a power electronic converter (Fig. 1) [4]. In the wind
turbine drive train, the rotor blades of a wind turbine catch
wind energy that is then transferred to the induction generator
through a gearbox. The induction generator, converting the
mechanical energy into the electrical energy, is a standard,
wound rotor induction machine with its stator windings
directly connected to the grid and its rotor windings connected
to the grid through a frequency converter [2-4]. The
frequency converter is built by two self-commutated voltageregulated PWM converters [3], a rotor-side converter and a
grid-side converter, with an intermediate dc voltage link.

= 9
= 11

0.2

= 13

0.1

= 15

0.0
0

10

12

14

16

18

20

tip-speed-ratio

Fig. 2. A 1.5MW wind turbine Cp curves

For each pitch angle, there is an optimal tip-speed-ratio opt


under which Cp takes a maximum value, i.e., maximum power
extraction from the wind for that pitch angle. Therefore, at a
low wind speed vw, the rotational speed of the rotor blades m
is normally regulated to a value of m_opt through the
integrated controls at DFIG generator and wind turbine levels
so that = Rbladem Vw = opt . As the captured wind power
exceeds the rated power at a high wind speed, a power
limitation control is activated to keep the generated power at
the rated value by adjusting the pitch angle.

Fig. 1. Configuration of a DFIG wind turbine


A. Power Extracted from the Wind
The power extracted by a wind turbine from the wind is
expressed by the well-known cube law equation [14]:
1
(1)
Pw = air Ablade C p ( , )vw3
2
= Rbladem vw
(2)
where air is the air density [kg/m3], Ablade is the area covered
by the rotor blades [m2], Cp is the turbine performance
coefficient, vw is wind speed [m/s], Rblade is the radius of the
rotor blades [m], and m is the angular speed of the blades.
The air density depends on factors, such as field elevation and
air temperature, and may change from 1.07 kg/m3 in a hot and
high elevation region to 1.395 kg/m3 in a cold and low attitude
place. The performance coefficient, Cp, is a function of the
tip-speed-ratio (2) and the pitch angle of the rotor blades .
It is determined by aerodynamic laws and thus may change
from one wind turbine type to another.
Figure 2 shows the Cp curves for a DFIG wind turbine
[15]. The mathematical representation of the Cp curves is
obtained through curve fitting as shown by (3) where aij
coefficients are given in [15]. The curve fit is a very good
approximation for values of 2 < < 13. Values of outside

B. Generator Control Configuration of DFIG


The speed control of a DFIG is usually transformed into
power control. The controller of the rotor-side converter,
commonly a two-stage controller, consists of a reactive power
controller and an active power controller [4, 16].
Traditionally, the stator-flux oriented frame is utilized for the
controller designs [3-8, 16], in which the q-axis current
component is for active power control and the d-axis
component is for reactive power control (Fig. 3) [4, 16]. To
maintain a constant frequency within the converter, this
current control concept is typically implemented indirectly
through a voltage regulated PWM converter [17]. The d-q
voltage control signals of the converter, the final control
action applied to the converter, are obtained by comparing the
rotor d- and q-current setpoints to the actual rotor d and q
currents as shown by the second stage controller in Fig. 3.

Fig. 3. DFIG Rotor side controller

The general strategy for transformation from the d and q


signals to the three-phase sinusoidal signal applied to the
converter is illustrated in Fig. 4, where the d and q reference
voltages, vrd* and vrq*, represents the d and q voltage control
signals from the controller (Fig. 3). The and reference
voltages, v* and v*, are obtained from the d-q reference
voltages through a vector rotation of e j ( s r ) , where s
represents the stator-flux or stator-voltage space vector
position and r represents the rotor position. The two and
voltages together are then used to generate a three-phase
sinusoidal reference voltage, va*, vb*, and vc*, to control the
rotor-side converter. Using the converter-average model, the
three-phase sinusoidal voltage vA, vB, and vC, injected into the
grid by the converter, is linearly proportional to the threephase sinusoidal reference voltage va*, vb*, and vc* in the
converter linear modulation mode. The gain between the two
is k PWM = Vdc Vtri , where Vdc is the dc link voltage, and Vtri is
the amplitude of the bipolar triangle carrier waveform.

Fig. 4. Decoupled d-q vector control structure for DFIG rotor-side converter

III. DFIG TRANSIENT AND STEADY-STATE MODEL IN d-q


REFERENCE FRAME

vrq = Rr irq + d rq dt + r rd

(8)

Vrd = Rr I rd r ( Lls + Lm ) I rq + Lm I sq

(9)

Vsq = Rs I sq + s ( Lls + Lm ) I sd + Lm I rd
Vrq = Rr I rq + r ( Lls + Lm ) I rd + Lm I sd

Using space vectors and considering r = ss, (8) and (9)


can be represented by (10) and (11), respectively, where Vs_dq,
Vr_dq, Is_dq, and Ir_dq are the steady-state d-q space vectors of
stator and rotor voltages and currents.
Vs _ dq = Rs I s _ dq + js Lls I s _ dq + js Lm ( I s _ dq + I r _ dq )

R
= r I r _ dq + js Llr I r _ dq + js Lm ( I s _ dq + I r _ dq )
(11)
s
s
The d-q steady-state equivalent circuit is obtained from
(10) and (11) as shown by Fig. 5. From the equivalent circuit,
DFIG stator, rotor and converted power can be obtained.
Using the motor convention, the stator real and reactive power
is shown by (12). The rotor loss power is shown by (13). The
air gap power is shown by (14). There is an additional
complex power item (15), which is absorbed by the rotor from
the grid through the frequency converter. The air gap power
(14) is composed of the power converted to mechanical form
(Pconv), the rotor copper losses (13), and the power absorbed
by the injected rotor voltage source. Thus, (16) allows the
computation of the power converted to mechanical form
based upon the defined sign conventions. Note that a positive
converted power implies motoring, while a negative converted
power implies generating. Ps and Pr can be positive or
negative, i.e., absorbing or generating power from the stator
or rotor path.
Ps + jQs = Vs _ dq I s*_ dq
(12)
PRCL = I r2_ dq Rr

(13)

Is_dq

(6)

rd = ( Llr + Lm ) ird + Lm isd , rq = ( Llr + Lm ) irq + Lm isq

(7)

Vs_dq

where Rs, Rr, Lls, and Llr are the resistances and leakage
inductances of the DFIG stator and rotor windings; Lm is the
mutual inductance; vsd, vsq, vrd, vrq, isd, isq, ird, irq, sd, sq, rd,
and rq are the d and q components of the space vectors of
stator and rotor voltages, currents, and fluxes; and s and r
are the angular frequencies of stator and rotor currents. In
steady-state condition, (8) and (9) result, where Vsd, Vsq, Vrd,
Vrq, Isd, Isq, Ird, and Irq are the d and q steady-state components
of the space vectors of stator and rotor voltages and currents.

(14)
(15)

Pconv = PAG + Protor PRCL

(16)

Rs

Ir_dq

Lls
2
1

Ims_dq

Llr
1

Lm

Rr

Rr(1-s)/s

+
Vr_dq

Ems_dq
2

Pr + jQr = Vr _ dq I r*_ dq

sd = ( Lls + Lm ) isd + Lm ird , sq = ( Lls + Lm ) isq + Lmirq

uuuuuur

PAG = Re Ems _ dq I r _ dq

(5)

(10)

Vr _ dq

A commonly used model for the DFIG is the Park model.


Using the motor convention, the space vector theory yields
stator and rotor voltage and flux equations in the form [17]:
vsd = Rsisd + d sd dt s sq
(4)
vsq = Rsisq + d sq dt + s sd
vrd = Rr ird + d rd dt r rq

Vsd = Rs I sd s ( Lls + Lm ) I sq + Lm I rq

Vr_dq(1-s)/s

Fig. 5. DFIG steady-state dq equivalent circuit

The relationship between the steady-state stator-voltage


and stator-flux space vectors can also be estimated using the
equivalent circuit. Assuming that stator winding resistance
and leakage inductance are negligible, then, the relations
among the stator voltage (Vs_dq), magnetizing current (Ims_dq),
and stator flux (s_dq) space vectors are
Vs _ dq
I ms _ dq =
,
s _ dq = Lm I ms _ dq .
(17)
js Lm

In other words, when assuming Rs=0 and Lls=0mH, the


stator-voltage space vector leads the stator-flux space vector
by 90. If the stator winding resistance and leakage
inductance are considered, the steady-state stator-flux space
vector, according to (6) and Fig. 5, is
s _ dq = ( Lls + Lm ) I s _ dq + Lm I r _ dq = Lls I s _ dq + Lm I ms _ dq
(18)
which is different from (17) by a stator leakage flux. But, it is
found from numerical computation that when considering the
stator resistance and leakage inductance, the stator-flux space
vector still lags the stator-voltage space vector by about 90
and their influence to the magnetizing current and stator-flux
space vectors is normally small especially for a DFIG
operating within the rated power. Table 1 gives the
parameters for a typical 2MW DFIG. Table 2 illustrates the
delay angle of the stator-flux space vector over the statorvoltage space vector, obtained by solving Fig. 5 and (18), for
several different generator slip values and rotor control
voltages. As it can be seen from Table 2, the stator-voltage
space vector usually leads the stator-flux space vector by
about 90 even when the stator winding resistance and
leakage inductance are considered.
Table 1. Typical DFIM data used in the simulation study
Parameter
Value
Units
Apparent Power
2000
kVA
Rated Voltage
690
V
RS (stator resistance)
0.0043
pu
XlS (stator reactance)
0.0809
pu
Rr (rotor resistance referred to
0.0048
pu
stator side)
Xlr (rotor reactance referred to
0.0871
pu
stator side)
Xm (magnetizing reactance)
3.459
pu
Frequency
50
Hz
Table 2. Angle between stator-flux (s) and stator-voltage
(sv) space vectors (stator-voltage vector as the reference)
Generator slip,
sv - s
Rotor control voltage
s=0.05, Vrd_sv=20V, Vrq_sv=2V
-89.886
s=0.22, Vrd_sv=100V, Vrq_sv=5V
-90.15
s=0.362, Vrd_sv=150V, Vrq_sv=7V
-90.013
s=0.638, Vrd_sv=250V, Vrq_sv=7V
-89.937

estimated from DFIG d-q steady-state equivalent circuit for a


given real and reactive power reference. Instead of using a
power-loop controller, this paper develops a direct
computational approach for fast estimation of rotor d or q
current reference from a real or reactive power reference
using the steady-state d-q equivalent circuit. In order to
distinguish the difference of d and q components of a space
vector in different orientation frames, the paper uses a
subscript sv representing stator-voltage-oriented frame and a
subscript sf signifying stator-flux-oriented frame in the
following sections.
A. Stator-Voltage Orientation Frame
In the stator-voltage orientation frame, the d-axis of the
reference frame is aligned along the stator-voltage position.
Assuming that the grid voltage applied to the stator is
constant, then, stator q-axis voltage is zero, and stator d-axis
voltage is constant, i.e., Vs _ dq _ sv = Vsd _ sv + j0 . Therefore,
according

to

(17),

I ms _ dq _ sv 0 jI ms

and

s _ dq _ sv 0 + jsd _ sv 0 + jLm I ms where I ms = Vsd _ sv (s Lm ) ,


i.e., the q component of the magnetizing current is constant
and the d component of the magnetizing current is zero. Then,
the stator d-q current space vector I s _ dq _ sv , according to Fig. 5,
is
I s _ dq _ sv = I ms _ dq _ sv I r _ dq _ sv = I rd _ sv j ( I ms + I rq _ sv ) (19)
In terms of the magnetizing and rotor currents, the stator real
and reactive powers according to (12) are
Ps = Vsd _ sv I rd _ sv , Qs = Vsd _ sv ( I rq _ sv + I ms )

(20)

Hence, given stator real and reactive power references Ps_ref


and Qs_ref, the steady-state rotor d and q current references are
I rd _ sv _ ref = Ps _ ref Vsd _ sv
(21)
I rq _ sv _ ref = Qs _ ref Vsd _ sv I ms
B. Stator-Flux Orientation Frame
The stator-flux orientation frame is widely used in DFIG
control designs [8-12]. In the stator-flux orientation frame, the
d-axis of the reference frame is aligned along the stator-flux
position so that stator q-axis flux linkage is zero, and d-axis
flux linkage is constant, i.e.,
s _ dq _ sf = sd _ sf + j 0 Lm ims + j 0
(22)
In the stator-flux oriented frame, the steady-state magnetizing
current space vector is I ms _ dq _ sf I ms + j 0 . Thus, according to

IV. COMPUTING ROTOR d-q CURRENT REFERENCES FROM


REAL/REACTIVE POWER REFERENCES

(17), the stator-voltage space vector is Vs _ dq _ sf 0 + jVsq _ sf ,

The general approach for DFIG active and reactive power


control is the conventional two-stage decoupled d-q control
strategy as shown in Fig. 3. In the power control loop, the
error signal between a reference and actual power generates a
rotor d or q current reference. In the current control loop, the
d or q component of the injected voltage is generated by
comparing the actual rotor d or q current to the rotor d or q
current setpoint. The rotor d or q currents can also be

voltage is constant and the d component of the stator voltage


is zero. Therefore, the stator d-q current space vector I s _ dq _ sf ,

where Vsq _ sv I mss Lm , i.e., the q component of the stator

according to Fig. 5, is

I s _ dq _ sf = I ms _ dq _ sf I r _ dq _ sf = ( I ms I rd _ sf ) jI rq _ sf

(23)

In terms of rotor and magnetizing currents, the stator real and


reactive powers according to (12) are

Ps = Vsq _ sf I rq _ sf , Qs = Vsq _ sf ( I ms I rd _ sf

(24)

Thus, given stator real and reactive references Ps_ref and Qs_ref,
the steady-state rotor d and q current references are
I rd _ sf _ ref = Qs _ ref Vsq _ sf + I ms
(25)
I rq _ sf _ ref = Ps _ ref Vsq _ sf
V. CONTROLLER DESIGN IN STATOR-VOLTAGE AND STATORFLUX ORIENTED FRAMES
The real and reactive power of a DFIG is controlled by the
rotor-side converter, in which the reactive power reference is
generated for either maximum reactive power compensation
or a need for grid voltage support, and the real power
reference is generated for either the maximum power
extraction from the wind or a real power demand from the
wind power plant controller [16]. According to Section IV,
the real and reactive power references can be transferred into
rotor d or q current references. Therefore, control of DFIG
real and reactive power can be implemented by regulating
rotor d or q current depending on the orientation frame used
in the controller design. It is important to specify that this
indirect power control approach through the rotor d-q current
is important in maintaining a high power quality of the
system, i.e., reduced system imbalance and harmonics.
Although direct-power-control strategies were proposed
recently [7, 8] for DFIG power control, removal of the rotor
current loop could result in more current and flux imbalance
in the generator as shown by some simulation results in [7, 8].
A. Controller Design in Stator-Voltage-Oriented Frame
According to Section IV-A, in the stator-voltage-oriented
frame, ims _ dq _ sv = 0 + j ims , where ims = sq _ sv Lm , depending
primarily on stator voltage, is almost constant so that the
derivative of the magnetizing current is zero. Thus, (5) can be
expressed in (26). In the stator-voltage-oriented frame, the
position of the stator voltage space vector sv can be obtained
directly from the stator voltage measurement in - reference
frame (27), making it easy and accurate to compute the
transformation from abc to dq or dq to abc that are needed for
a d-q control mechanism.
di

vrd = Rr ird + Llr rd r Llr irq r Lm ims


dt

(26)
dirq

vrq = Rr irq + Llr


L
i
+

r lr rd
dt

sv = tan 1 ( vs vs )

(27)

In (26), the component of the bracket is treated as the state


equation between the voltage and current on d and q loops,
and the other parts are treated as compensation items.
Therefore, a design of the current controller can be obtained
from the system block diagram as shown by Fig. 5, in which
1 ( L s + R ) represents plant transfer function for d or q
current loop, kPWM is the gain of the power electronic
converter (Fig. 3), kFB is a gain on the feedback path such as a

gain of a sensor, and the PID block stands for the controller to
be designed [17].

Fig. 6. A system block diagram for design of current-loop controller

The current-loop controller of the d or q loop outputs a


voltage signal, vrd and vrq, based on the error signal between
the reference and actual rotor current on the d or q loop. The
d or q reference voltage, vrd* or vrq*, applied to the rotor-side
converter is acquired from (26) by replacing the bracket by d
or q voltage from d or q-loop controller, respectively, as
shown by (28), in which a magnetizing current compensation
item is added to the rotor d voltage equation in the statorvoltage-oriented frame.
vrd* = vrd' r Llr irq r Lm ims

(28)

vrq* = vrq' + r Llr ird

B. Controller Design in Stator-Flux-Oriented Frame


In the stator-flux oriented frame, ims _ dq _ sf = ims + j0 , where
ims = sd _ sf Lm is almost constant if the stator voltage keeps

unchanged. Thus, (5) can be expressed by (29). Traditionally,


the position of the stator-flux space vector sf is estimated by
using (30), which requires measurements of both statorvoltage and stator-current space vectors in - reference
frame. Since the integration is involved, the computational
error for the estimation of sf is normally higher than that for
the estimation of sv [18].

vrd = Rr ird + Llr

vrq

= Rr irq

+ Llr

dird
r Llr irq
dt
dirq
dt

+ r Llr ird

(29)
+ r Lm ims

s = ( vs Rs is )dt , s = ( vs Rs is )dt

sf = tan 1 ( s s )

(30)

The controller design strategy using the stator-fluxoriented frame is the same as that using the stator-voltageoriented frame (Section V-A). However, according to (29),
the d and q reference voltage, vrd* and vrq*, applied to the
rotor-side converter are (31), in which the magnetizing current
compensation item is added in the rotor q-axis voltage
equation rather than the d-axis voltage equation.
vrd* = vrd' r Llr irq
vrq* = vrq' + r Llr ird + r Lm ims

(31)

Figure 7 shows the overall d-q vector control structure


using either the stator-voltage or the stator-flux oriented
frame. In the figure, the d or q reference voltage, vrd* and vrq*,

Fig. 7 DFIG decoupled d-q vector control structure using stator-voltage and stator-flux oriented frames

consists of the d or q voltage, vrd or vrq, from the controllers


plus the compensation items. The magnetizing current
compensation item is added to d or q loop depending on the
position of the two switches in the figure. If switch S1 turns to
the left and switch S2 turns to the right, the magnetizing
current, calculated using ims = sq _ sv Lm , is added to d-axis
voltage equation, which represents the control structure using
the stator-voltage-oriented frame. If switch S1 turns to the
right and switch S2 turns to the left, the magnetizing current,
calculated using ims = sd _ sf Lm , is added to q-axis voltage
equation, which represents the control structure using the
stator-flux-oriented frame. Angle is an adjusting angle
depending on what is the orientation frame used in the
controller design. is 0 for stator-voltage-oriented frame and
is nonzero for the stator-flux-oriented frame. In the statorflux-oriented frame, represents the delay angle of the statorflux space vector over the stator-voltage space vector (See
Section VI-B).
VI. POWER CONTROL EVALUATION USING STATOR-VOLTAGE
AND STATOR-FLUX ORIENTED FRAMES
A. Power Control Using Stator-Voltage-Oriented Frame
In the stator-voltage-oriented frame, the controller of d or
q loop is designed for a wound rotor induction machine
according to Fig. 6. The parameters of the induction machine
are shown in Table 1. The frequency response technique is
used in the controller design, where the crossover frequency
of the controller is two orders smaller than the switching
frequency and the phase margin is 60.

To evaluate the performance of the d-q controller, an


integrated transient simulation system with the feedback
control is built using SimPowersystems (Fig. 8). The grid is
represented by a three-phase 50Hz, 690V line-line voltage
source that is directly applied to the stator. The rotor voltage,
a three-phase controlled voltage source, is regulated by vrd*
and vrq* using PWM converter average model. The rotor
voltage, having the slip frequency, is obtained through dq-toabc transformation based upon DFIG stator-voltage space
vector position and rotor position. The turbine driving torque
is calculated using (1)-(3) based on the information of wind
speed, turbine blade pitch angle, and generator rotational
speed. The wind speed is 7m/s, air density is 1.17kg/m3, and
pitch angle is 1. The stator-voltage space vector position is
estimated using (27), which only requires present
measurements of the stator voltage in - reference frame.
Major measurements in the transient environment include
speed, torque, and three-phase voltage, current, and real and
reactive power at the stator and rotor paths. For each power
measurement point, passive sign convention is used, i.e.,
power absorbed toward the generator is positive. In the
transient environment, the performance of the d-q controller is
examined for different wind speeds and real and reactive
power references.
Figure 9 demonstrates the performance of the controller
for a set of real and reactive power references. The initial real
and reactive power references are -300kW and -100kVar,
respectively, i.e., a condition that both real and reactive power
are generated from the stator. At t=4s, the real power
reference changes to -500kW while reactive power reference
is unchanged, i.e., more real power generation. At t=7s, the

more fluctuations in output power and high flux and stator


current imbalance.
135

S
p
e
e
d
(r
a
d
/s
)

130

125

120

115

110

5
6
Time (s )

10

a) Rotor speed
100
data 1
data 2

-100
P
o
w
e
r(kW
/kV
a
r)

real power reference changes to 50kVar while real power


reference remains 500kW, i.e., reduced reactive power
generation. As shown by the instantaneous power waveforms
(Fig. 9b), both real and reactive power can be regulated
around the reference values effectively before and after t=4s
and t=7s, showing that the controller works properly for grid
real and reactive power control and the effectiveness of the
transformation from the real and reactive power references to
rotor d-q current references using (21). The speed and power
transient results are also consistent with the steady-state
control characteristics as shown in [19]. The stator and rotor
currents show effective power quality during the transition
from one power reference to another (Fig. 9c&d). It is needed
to point out that the proposed estimation approach for rotor dq current references using (21) and (25) could be used with
traditional power loop controller (Fig. 3) to reduce the
response time greatly if more accurate estimation rotor d-q
currents references are needed.

-200

Real power

-300

Reactive power

-400

-500

-600

5
6
Time (s)

10

b) Real and reactive power


800

Three-phaseabccurrents(A)

600
400
200
0
-200
-400
-600
-800
3.9

3.95

4.05
Time (s)

4.1

4.15

4.2

c) Stator a, b, and c currents

B. Power Control Using Stator-Flux-Oriented Frame


In the stator-flux-oriented frame, the PI controller of the d
or q current loop is the same as that of stator-voltage-oriented
frame. However, the overall control structure after adding the
compensation items is different (Fig. 7). Another difference is
the estimation of the stator-flux space vector position.
Traditionally, the position of the stator-flux space vector is
estimated using (30), which normally has a high
computational error due to the multiplication of the stator
current with a low stator winding resistance and the numerical
integration involved. The stator-flux space vector position can
also be estimated by adding a delay angle (Fig. 7) to the
stator-voltage space vector position. The delay angle can
either be 90 (Table 2) or estimated more accurately using
(18). Figure 10 shows the delay angle of the stator-flux space
vector over the stator-voltage space vector computed using
(30) (Approach 1) and (18) (Approach 2) in the transient
simulation system of Fig. 8 after the system is stable for a
power reference. As it can be seen, the delay angle estimated
using both approaches is about 90. However, estimation
using approach 2 is more stable than that using approach 1,
which implies that the estimation of stator-flux space vector
position using (30) could let the DFIG controller result in

1000
500
0
-500
-1000
-1500
3.9

3.95

4.05
Time (s)

4.1

4.15

4.2

c) Rotor a, b, and c currents


Fig. 9. DFIG control using the stator-voltage-oriented frame
-88
Approach 1
Approach 2

-88.5
Stator-flux space vector angle

Fig. 8. Front-end converter with feedback control

Three-phaseabccurrents(A)

1500

-89
-89.5
-90
-90.5
-91
-91.5
-92

2.05

2.1

2.15

2.2

2.25
Time (s)

2.3

2.35

2.4

2.45

2.5

Fig. 10. Comparison of the angle between the stator-flux and stator-voltage
space vectors: estimated using two different approaches

The performance of the d-q controller using the statorflux-oriented frame is evaluated using the same conditions of
Section VI-A. The position of the stator-flux space vector is
estimated by adding a delay angle to the stator-voltage space
vector (Fig. 7). The delay angle is either 90 (Table 2) or

DCiv,link
eq

e je

estimated using (18). Both have the similar performance. The


performance of the controller using the stator-flux-oriented
frame is presented in Fig. 11 for the same real and reactive
power references used in Fig. 9. By comparing Fig. 11 with
Fig. 9, it demonstrates that 1) the controller designs using
stator-voltage and stator-flux oriented frames have equivalent
performance, 2) it is effective and stable to estimate statorflux space vector position by simply adding -90 to the statorvoltage space vector position, and 3) it is effectual to estimate
rotor d-q current references using (25) in the stator-fluxoriented frame for a desired power reference. Although the
controller performance using the stator-voltage and stator-flux
oriented frames is equivalent, the theoretical implementation
of the controller is different for the two different orientation
frames as shown by Fig.7.
135

illustrate that the controller design using both stator-voltage


and stator-flux oriented frames has equivalent performance.
VIII. REFERENCES
[1]

[2]

[3]

[4]

[5]

Speed(rad/s)

130

[6]

125

120

[7]
115

110

5
6
Time (s)

10

[8]

a) Rotor speed
100

[9]

Real Power
Reactive Power

Pow
er(kW
/kVar)

-100

Reactive power

-200

[10]

Real power

-300

[11]

-400

-500

-600

5
6
Time (s)

10

[12]

b) Real and reactive power


Fig. 11. DFIG control using the stator-flux-oriented frame

VII.

[13]

CONCLUSIONS

This paper compares real and reactive power control for a


DFIG based wind turbine using stator-voltage and stator-flux
oriented frames. The paper presents both DFIG steady-state
and transient models in d-q reference frame. The steady-state
model is used to obtain the general relationship between rotor
d/q current and stator real/reactive power references using
stator-flux and stator-voltage oriented frames. The transient
model, together with the analysis based on DFIG d-q steadystate equivalent circuit, is used to develop and design DFIG
controller. Analytical study demonstrates that mathematical
representations of the controller are different using the two
different orientation frames. Another difference is the
estimation of the positions of stator-voltage and stator-flux
space vectors. It is easier to estimate a stator-voltage space
vector position than a stator-flux space vector position.
Performance study shows that stator-flux space vector
position can be estimated the by simply adding -90 to the
stator-voltage space vector position. The simulation results

[14]
[15]

[16]

[17]

[18]
[19]

R. Zavadil, N. Miller, A. Ellis, and E. Muljadi, Making Connections:


Wind Generation Challenges and Progress, IEEE Power & Energy
Magazine, Vol.3, No. 6, Nov. 2005.
S. Muller, M. Deicke, R.W. De Doncker, Doubly Fed Induction
Generator Systems for Wind Turbines, IEEE Industry Applications
Magazine, Vol. 8, No. 3, 26-33, May/June 2002.
R. Pena, J.C. Clare, and G. M. Asher, Doubly fed induction generator
using back-to-back PWM converters and its application to variable speed
wind-energy generation, IEE Proc.-Electr. Power Appl., Vol. 143, No 3,
May 1996.
Chowdhury BH, Chellapilla S, Double-fed induction generator control for
variable speed wind power generation, Electric Power Systems Research,
Volume: 76, Issue: 9-10, Pages: 786-800, JUN 2006.
L. Morel, H. Godfroid, A. Mirzaian, J.M. Kauffmann, Double-fed
induction machine: converter optimization and field oriented control
without position sensor, IEE Proc.-Electr. Power Appl., Vol. 145, No 4,
May 1998.
B. Hopfensperger, D. Atkinson, and R. A. Lakin, Stator flux oriented
control of a cascaded doubly-fed induction machine, IEE Proc. Electr.
Power Appl., vol. 146, no. 6, pp. 597605, Nov. 1999.
Xu L., Cartwright P, Direct active and reactive power control of DFIG
for wind energy generation, IEEE Transactions on Energy Conversion,
Volume: 21, Issue: 3, Pages: 750-758, Sep. 2006.
Zhi D.W., Xu L., Direct power control of DFIG with constant switching
frequency and improved transient performance, IEEE Transactions on
Energy Conversion, Volume: 22, Issue: 1, Pages: 110-118, Mar. 2007.
L. Xu, and Y. Wang, Dynamic Modeling and Control of DFIG Based
Wind Turbines under Unbalanced Network Conditions, IEEE
Transactions on Power Systems, Vol. 22, No. 1, Feb. 2007.
Brekken TKA, Mohan N, Control of a doubly fed induction wind
generator under unbalanced grid voltage conditions, IEEE Transactions
on Energy Conversion, Volume: 22, Issue: 1, Pages: 129-135, Mar. 2007.
Protsenko K, Xu D, Modeling and control of brushless doubly-fed
induction generators in wind energy applications, IEEE Transactions on
Power Electronics, Volume: 23, Issue: 3, Pages: 1191-1197, May 2008.
Smith, S.; Todd, R.; Barnes, M.; Tavner, P.J., "Improved energy
conversion for doubly-fed wind generators," IEEE Transactions on
Industry Applications, Vol. 42, No. 6, pp. 1421-1428, Nov. 2006.
Vicatos, M.S.; Tegopoulos, J.A., "Steady state analysis of a doubly-fed
induction generator under synchronous operation," IEEE Transactions on
Energy Conversion, vol.4, no.3, pp. 495-501, Sep 1989.
L. L. Freris, Wind energy conversion system (Upper Saddle River, NJ:
Prentice Hall, 1990).
Nicholas W. Miller, William W. Price, and Juan J. Sanchez-Gasca,
Dynamic Modeling of GE 1.5 and 3.6 Wind Turbine-Generators, GE
Power Systems, October 27, 2003.
Hansen, A.D., Iov, F., Srensen, P., and Blaabjerg, F., Overall control
strategy of variable speed doubly-fed induction generator wind turbine,
In: Grid integration and electrical systems of wind turbines and wind farms
(CD-ROM). Nordic wind power conference 2004 (NWPC 04), Gteborg
(SE), 1-2 Mar 2004. (Chalmers University of Technology, Gteborg,
2004) 7 p.
N. Mohan, Advanced Electric Drives Analysis, Modeling and Control
using Simulink, MN: Minnesota Power Electronics Research & Education,
ISBN 0-9715292-0-5, 2001.
G. Franklin, J.D. Powell, A. Emami-Naeini, Feedback Control of Dynamic
Systems, 5th ed., Prentice Hall Upper Saddle River, New Jersey, 2005.
S. Li, T. H., and S. Mazari, Integrating Electrical and Aerodynamic
Characteristics for DFIG Speed Control Study, accepted by IEEE PES
2009 Power Systems Conference and Exposition, Seattle, WA, March 1518, 2009.

You might also like