110 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO.

1, MARCH 2007

Direct Power Control of DFIG With

Constant Switching Frequency and Improved
Transient Performance
Dawei Zhi, Student Member, IEEE, and Lie Xu, Senior Member, IEEE

Abstract—This paper proposes a new direct power control

(DPC) strategy for a doubly fed induction generator (DFIG)-based
wind turbine system. The required rotor control voltage, which
eliminates active and reactive power errors within each fixed time
period, is directly calculated based on stator flux, rotor position,
and active and reactive powers and their corresponding errors.
No extra power or current control loops are required, simplifying
the system design, and improving transient performance. Constant
converter switching frequency is achieved that eases the design of
the power converter and the ac harmonic filter. Rotor voltage limit
during transients is investigated, and a scheme is proposed that Fig. 1. Schematic of a DFIG-based wind generation system.
prioritizes the active and reactive power control such that one re-
mains fully controlled while the error of the other is reduced. The
impact of machine parameter variations on system performance is s, r Stator, rotor.
investigated and found negligible. Simulation results for a 2 MW d, q Synchronous d–q axis.
DFIG system demonstrate the effectiveness and robustness of the
proposed control strategy during variations of active and reactive
power, machine parameters, and wind speed. I. INTRODUCTION
OUBLY-fed induction generator (DFIG) wind turbines
Index Terms—Constant switching frequency, DFIG, direct
power control, pulse width modulation (PWM) converter, wind
energy.
D with converters rated at about 25–30% of the generator
rating are becoming increasingly popular. DFIG-based wind
turbines offer variable speed operation, four-quadrant active
NOMENCLATURE and reactive power capabilities, lower converter cost, and re-
duced power loss compared to wind turbines using fixed speed
θ s , θr Stator flux, rotor angles in the stationary frame. induction generators or fully-fed synchronous generators with
θ Phase angle between the rotor and stator flux full-sized converters. A schematic of a DFIG-based wind energy
vectors. generation system is shown in Fig. 1.
ω 1 , ωr , ωs Synchronous, rotor, slip angular frequency. Control of DFIG wind turbine systems is traditionally based
ψs , ψr Stator, rotor flux vectors. on either stator-flux-oriented [1], [2] or stator-voltage-oriented
I s, I r Stator, rotor current vectors. vector control [3], [4]. The scheme decouples the rotor current
Lm Mutual inductance. into active and reactive power components. Control of the active
Lσs , Lσr Stator, rotor leakage inductance. and reactive powers is achieved with a rotor current controller.
Ls , Lr Stator, rotor self-inductance. One main drawback of this system is that its performance de-
R s , Rr Stator, rotor resistance. pends highly on accurate machine parameters such as stator,
Ps , Qs Stator active and reactive power. rotor resistances, and inductances. Thus, performance degrades
V s, V r Stator, rotor voltage vectors. when actual machine parameters depart from values used in the
Superscripts control system.
s Synchronous reference frame. Direct torque control (DTC) of induction machines, devel-
r Rotor reference frame. oped more than a decade ago, provides an alternative to vector
∗ Reference value. control [5], [6]. DTC minimizes the use of machine parame-
ˆ Conjugate complex. ters and reduces the complexity of vector control algorithms.
Subscripts The DTC method directly controls machine torque and flux by
α–β α–β axis. selecting voltage vectors from a look-up-table using the stator
flux and torque information. One problem with the basic DTC
Manuscript received June 29, 2006; revised October 16, 2006. Paper no. scheme is that its performance deteriorates during starting and
TEC-00227-2006. low-speed operations. Methods proposed to address this prob-
The authors are with the School of Electronics, Electrical Engineering and lem include use of a dither signal [7], a modified switching table
Computer Science, Queen’s University of Belfast, Belfast BT7 1NN, U.K.
(e-mail: [email protected]; [email protected]). to apply available voltage vectors in appropriate sequence [8], or
Digital Object Identifier 10.1109/TEC.2006.889549 predictive techniques [9]. Another problem for DTC is converter
0885-8969/$25.00 © 2007 IEEE
Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
ZHI AND XU: DIRECT POWER CONTROL OF DFIG WITH CONSTANT SWITCHING FREQUENCY 111

switching frequency variation that significantly complicates

power circuit design. The hysteresis controller bandwidth must
be chosen carefully [10] to ensure that the switching frequency
stays within the power converter’s maximum limit for all oper-
ating conditions. Modified DTC strategies, incorporating space
vector modulation (SVM), have been used to achieve constant
switching frequency [11]–[14]. Inverter switching duty cycles
in [11] were calculated directly, based on torque and flux er- Fig. 2. Equivalent circuit of a DFIG in the rotor reference frame.
rors within each sample period, whereas in [12] and [13], they
are generated from PI controllers. Output voltage vector was The paper is organized as follows. Section III gives an overview
selected in [14] using the conventional DTC switching table of the original DPC strategy for DFIG control and Section IV
but the voltage vector duration was determined by the torque- describes the proposed strategy. Simulation results for a 2 MW
ripple minimum strategy. Although the switching frequencies DFIG generation system are presented in Section V to demon-
were constant as those DTC strategies, extra drawbacks were strate the performance of the proposed control strategy. Finally,
introduced, such as complicated online calculation [11], addi- conclusions are drawn in section VI.
tional PI controller parameters [12], [13], and weak robustness
on machine parameter variations [11]–[14]. II. PRINCIPLES OF DPC FOR DFIG
Based on the principles of DTC strategy, direct power con-
trol (DPC) was developed for three-phase pulse width modu- In the rotor αr −βr reference frame rotating at an angular
lation (PWM) rectifiers [15]–[17]. Converter switching states speed of ωr , Fig. 2 shows the equivalent circuit of a DFIG.
were selected from an optimal switching table based on in- According to Fig. 2, the stator voltage vector in the rotor frame
stantaneous errors of active and reactive powers and the an- is given as
gular position of converter terminal voltage vector [15], [16], dψsr
or virtual flux that is the integration of the converter output Vsr = Rs Isr +
+ jωr ψsr . (1)
dt
voltage [17]. More recently, DPC control of DFIG-based wind
In the rotor reference frame, the stator and rotor flux are
turbine systems has been proposed [18], [19]. In [18], the con-
expressed as
trol system was based on the estimated rotor flux. Switching
vectors were selected from the optimal switching table using ψsr = Ls Isr + Lm Irr
the estimated rotor flux position, and the errors of the rotor flux
ψrr = Lr Irr + Lm Isr (2)
and the active power/torque. The rotor flux reference was calcu-
lated using the reactive power/power factor reference. Since the where Ls = Lσs + Lm and Lr = Lσr + Lm .
rotor supply frequency, which equals the DFIG slip frequency, Based on (2), the stator current is given as
can become very low, rotor flux estimation is significantly af-
Lr ψsr − Lm ψrr ψsr Lm ψrr
fected by the machine parameter variations. In [19], a DPC Isr = = − (3)
strategy based on the estimated stator flux was proposed. Since Ls Lr − L2m σLs σLs Lr
the stator (network) voltage is relatively harmonic-free with where σ = (Ls Lr − L2m )/Ls Lr is the leakage factor.
fixed frequency, a DFIG’s estimated stator flux accuracy can be Referring to [20], the stator active and reactive power inputs
guaranteed. Switching vectors were selected from the optimal from the network can be calculated as
switching table using the estimated stator flux position, and the 3
errors of the active power and reactive powers. Thus, the control Ps − jQs = Vsr × Iˆsr . (4)
2
system is very simple, and the machine parameters’ impact on
system performance was found to be negligible. However, like If the network voltage is constant, neglecting the stator resis-
a conventional DTC, DPC has switching frequency that varies tance, both the amplitude |ψs | and the rotating speed ω1 of the
significantly with active and reactive power variations, machine stator flux remain constant [19]. Substituting (1) and (3) into (4)
operating speed (rotor slip), and the power controllers’ hystere- yields the stator active and reactive power inputs [19]:
sis bandwidth [18], [19]. Apart from complicating the converter Ps = −kσ ω1 |ψs ||ψr | sin θ
design, the stator side ac filter (Fig. 1), which prevents switching

harmonics from entering the network, needs to be designed to Lr
Qs = kσ ω1 |ψs | |ψr | cos θ − |ψs | (5)
eliminate broadband harmonics. With increased size and power Lm
loss, this significantly reduces filter efficiency. where kσ = 1.5Lm /(σLs Lr ) and θ is the angle between the
This paper proposes a new DPC control strategy for DFIG- rotor and stator flux vectors, as shown in Fig. 3.
based wind energy generation systems with constant switching Differentiating (5) results in the following:
frequency and improved transient performance. The method di-
rectly calculates required rotor control voltage with each switch- dPs d (|ψr | sin θ)
= −kσ ω1 |ψs |
ing period, based on the estimated stator flux, the active and re- dt dt
active powers, and their errors. A method to limit rotor control dQs d (|ψr | cos θ)
voltage and further improve transient performance is described. = kσ ω1 |ψs | . (6)
dt dt

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
112 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 1, MARCH 2007

Fig. 5. Equivalent circuit of a DFIG in the synchronous d–q reference frame.

Fig. 3. Stator and rotor flux vectors in stationary and rotor reference frames.

Fig. 6. Stator and rotor flux vectors in the synchronous d–q frames.

hysteresis bandwidth, the rotor slip, etc., and can vary signif-
icantly [19]. Therefore, it is difficult to evaluate the rotor-side
converter’s power loss and loading conditions, and to design
a suitable cooling system. Furthermore, because of variable
Fig. 4. Schematic of the original DPC control strategy [19]. switching frequency, the harmonics in the stator current also
vary according to the operating conditions. The ac filter has a sig-
As |ψs | and ω1 remain constant, (6) indicates that the variations nificantly complicated design, since it has to absorb broadband
of active and reactive powers can be controlled by varying the frequency components to prevent their entering the network.
rotor flux components |ψr | sin θ and |ψr | cos θ, respectively.
The relationships between |ψr | sin θ and |ψr | cos θ, the rotor III. PROPOSED DPC STRATEGY
flux, and the stator flux can be seen in Fig. 3. A. DFIG Representation in the Synchronous Reference Frame
As shown in Fig. 2, neglecting the rotor resistance, the rotor
flux variation in the rotor reference frame can be approximated The equivalent circuit of a DFIG in the synchronous d–q
as frame, rotating at the speed of ω1 , is shown in Fig. 5. The d-axis
of the synchronous frame is fixed to the stator flux, as shown in
dψrr
= Vrr − Rr Irr ≈ Vrr . (7) Fig. 6.
dt According to Fig. 5, the stator voltage vector in the syn-
Equation (7) indicates that the rotor flux change is directly con- chronous d–q reference frame is given as
trolled by the applied rotor voltage, i.e., the rotor flux moves in dψss
the direction of the applied rotor voltage vector, and its speed is Vss = Rs Iss + + jω1 ψss . (8)
dt
proportional to the amplitude of the voltage vector. For a three-
phase two-level converter, the operation of the six power devices Under balanced ac voltage supply, the amplitude and rotating
results in eight different voltage vectors [21]. Within each sam- speed of the stator flux are constant. Therefore, in the syn-
pling period, if the stator flux position is known, the impact chronous d–q frame, the stator flux maintains a constant value.
of each voltage vector on the variations of the rotor flux, the Thus,
|ψr | sin θ and |ψr | cos θ components can be determined. Con-
ψss = ψsd
sequently, according to (6), the influence of each voltage vector
on active and reactive power variations can be calculated. An dψss
= 0. (9)
optimal switching table can, then, be arranged to give the most dt
effective rotor voltage vector to reduce the power errors. Considering (9) and neglecting the voltage drop across the stator
Fig. 4 shows the arrangement of the DPC control strategy resistance, (8) can be simplified as
proposed in [19]. Hysteresis control of the active and reactive
powers within each sampling period is achieved by selecting and Vss = jω1 ψss = jω1 ψsd . (10)
applying the optimal rotor voltage vector. In order to guarantee
Similar to (3), the stator current in the synchronous d–q frame
the effective control of the powers, the sampling frequency must
is given as
be sufficiently high, usually in the range of tens of kilohertz.
The converter’s switching frequency depends highly on the Lr ψss − Lm ψrs ψs Lm ψrs
Iss = = s − . (11)
operating conditions such as the active and reactive powers, the Ls Lr − Lm2 σLs σLs Lr

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
ZHI AND XU: DIRECT POWER CONTROL OF DFIG WITH CONSTANT SWITCHING FREQUENCY 113

Thus, the stator active and reactive power inputs can be calcu- Combining (16) and (18), within the Ts period, the rotor volt-
lated as age required to eliminate the power errors in the d–q reference
  frame is calculated as
3 ψ̂ss Lm ψ̂rs
Ps − jQs = jω1 ψsd × − 1 ∆Qs
2 σLs σLs Lr Vrd = − ωs ψrq
  Ts kσ ω1 ψsd
3 ψsd Lm (ψrd − jψrq )
= jω1 ψsd × − 1 −∆Ps
2 σLs σLs Lr Vrq = + ωs ψrd . (19)

 Ts kσ ω1 ψsd
Lr ψsd
= kσ ω1 −ψsd ψrq + jψsd − ψrd . The rotor flux in the d–q frame can be calculated using a
Lm method similar to that shown in (2). However, its accuracy could
(12) be affected by the variation of Lm as will be discussed later in
the paper. An alternative method is based on (13) as
Splitting (12) into real and imaginary parts yields
Qs Lr
Ps = −kσ ω1 ψsd ψrq ψrd = + ψsd

 kσ ω1 ψsd Lm
Lr −Ps
Qs = kσ ω1 ψsd ψrd − ψsd . (13) ψrq = . (20)
Lm kσ ω1 ψsd
As the stator flux stays constant, according to (13), the active
Substituting (20) into (19) results in the required rotor voltage
and reactive power changes over a constant period of Ts are
in the d–q reference frame as
given by
1 ∆Qs Ps
∆Ps = −kσ ω1 ψsd ∆ψrq Vrd = + ωs
Ts kσ ω1 ψsd kσ ω1 ψsd
∆Qs = kσ ω1 ψsd ∆ψrd . (14)

1 −∆Ps Qs Lr
Vrq = + ωs + ψsd . (21)
Equation (14) indicates that the stator reactive and active power Ts kσ ω1 ψsd kσ ω1 ψsd Lm
changes are determined by the changes of the rotor flux compo-
The first terms on the right-hand side reduce power errors while
nents on the d–q axis, i.e., ∆ψrd and ∆ψrq , respectively.
the second terms compensate the rotor slip that causes the differ-
ent rotating speeds of the stator and rotor flux. As can be seen,
B. Active and Reactive Power Control
calculations require only simple multiplications and divisions,
The proposed active and reactive power control calculates the and no complicated mathematics.
required rotor voltage that will reduce the active and reactive
power errors to zero during a constant sampling time period of C. Rotor Output Voltage Limit and PWM Modulation
Ts . A PWM modulator is, then, used to generate the applied
rotor voltage for the Ts period. For a DFIG, the maximum output voltage from the rotor-side
At the beginning of the sampling period Ts , the active and converter is usually within the range of 30% of the stator voltage.
reactive power errors are calculated as Under steady-state operation, the required rotor control voltage
is unlikely to exceed the voltage limit of the rotor-side converter.
∆Ps = Ps∗ − Ps However, during transient conditions, large variations of active
∆Qs = Q∗s − Qs . (15) and/or reactive power references can result in large power errors
in one sampling period Ts . Consequently, the rotor voltages Vrd
According to (14), in order to reduce the power errors shown in and Vrq , calculated using (21), may exceed the voltage capability
(15) to zero, the rotor flux changes in the d–q axis must follow of the converter. Therefore, Vrd and Vrq must be limited to
∆Ps improve the transient response. In the case of large active power
∆ψrq = − error, the calculated Vrq is likely to exceed the voltage limit,
kσ ω1 ψsd
while Vrd remains within the limit. Traditionally, the two voltage
∆Qs
∆ψrd = . (16) components were scaled proportionally. However, the scaling
kσ ω1 ψsd of Vrd in this case could result in reactive power control being
As shown in Fig. 5, in the synchronous d–q reference frame, temporarily lost. The approach adopted here maintains Vrd and
the rotor flux is given by scales Vrq according to the maximum voltage limit such that
dψrs reactive power remains controlled while active power is driven
= Vrs − Rs Irs − j(ω1 − ωr )ψrs . (17) toward the desired value. This process can be represented as
dt
Neglecting the rotor resistance, within the period of Ts , the 
Vrd = Vrd
changes of rotor flux in the d- and q-axis are given by 

∆ψrd = Vrd Ts + ωs ψrq Ts Vrq = sign(Vrq ) Vr2max − Vrd
2 (22)
∆ψrq = Vrq Ts − ωs ψrd Ts (18) where Vr max is the maximum voltage the rotor-side converter
where ωs = ω1 − ωr is the slip frequency. can produce.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
114 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 1, MARCH 2007

Fig. 8. Schematic of the simulated system.

Fig. 7. Schematic of the proposed DPC for a DFIG system.

calculate the stator flux angle θs and its rotating speed ω1 . The
rotor d–q control voltage references are, then, calculated using
The opposite is true during reactive power steps when Vrq
(21). They are passed to the voltage limiter and transformed to
remains unchanged while Vrd is scaled. Within the sampling
the rotor frame using the rotor angle that is obtained by a shaft
period, if simultaneous step changes of active and reactive pow-
encoder. Finally, PWM switching pattern is generated and used
ers cause both Vrd and Vrq to exceed the voltage limit, Vrd and
to control the rotor-side converter.
Vrq are scaled proportionally as
 Vr max
Vrd = Vrd  E. Impact of Parameter Variations on System Performance
2 +V2
Vrd rq Stator flux estimation requires knowledge of stator resistance
Vr max that has negligible impact on system performance [19]. Using

Vrq = Vrq  . (23) (21), rotor voltage calculation accuracy is mainly influenced by
2 +V2
Vrd rq the constant kσ and the inductance ratio Lr /Lm that are deter-
Once the required rotor output voltage in the synchronous mined by the stator and rotor leakage and mutual inductance.
reference frame is obtained, it must be transformed to the rotor Since the leakage flux magnetic path is mainly air, the varia-
αr –βr reference frame. This is achieved using the following: tion of the leakage inductance during operation is insignificant.
However, mutual inductance variation needs to be considered
Vrr = Vrs ej(θs −θr ) . (24) due to possible variation of the magnetic permeability of the
The desired rotor voltage in the rotor frame can, then, generate stator and rotor cores under different operating conditions. As
switching signals for the rotor-side converter using either SVM described in Appendix A, considering the relatively small leak-
[21] or sinusoidal PWM [22] whose switching frequency is fixed age inductance Lσs and Lσr compared to the mutual inductance
constantly at 1/(2Ts ). A typical switching frequency for this Lm , the required parameters can be simplified as
converter ranges within a few kilohertz, and therefore, the pro-
3 1 Lr Lσs
posed DPC sampling frequency is within a few kilohertz range. kσ ≈ , =1+ ≈ 1. (26)
2 Lσs + Lσr Lm Lm
D. System Implementation
As can be seen from (26), the variation of Lm has little
As shown in (21) and (24), to calculate the required rotor impact on kσ and Lr /Lm , and therefore, its influence on the
voltage, the values of the active and reactive powers and their performance of the proposed control strategy would also be
errors, the stator flux amplitude, rotating speed and angle, and insignificant.
the rotor angle are required.
The stator instant active and reactive powers can be directly IV. SIMULATION RESULTS
calculated from the measured stator voltages and currents. In
the stationary reference frame, the stator flux is estimated using Simulations of the proposed control strategy for a DFIG-
the following: based generation system used PSCAD/EMTDC, and Fig. 8
 shows the schematic of the implemented system. The DFIG
ψs = (Vs − Rs Is ) dt. (25) is rated at 2 MW, and its parameters are given in Appendix
B. The nominal converter dc link voltage was set at 1200 V.
Since the stator voltage is relatively harmonic-free and its Based on the proposed DPC strategy, the rotor-side converter
frequency is fixed, the previous equation can provide an accurate controls the DFIG stator’s active and reactive power. The
estimation of the stator flux. sampling period during simulation was 250 µs, which refers
The schematic of the proposed DPC strategy is shown in to a rotor-side converter switching frequency of 2 kHz. The
Fig. 7. The three-phase stator voltages and currents are measured grid-side converter has to maintain a constant dc link voltage,
and transformed into the stationary α–β reference frame. The and it is controlled by a method similar to the dc voltage
stator active and reactive powers are calculated and the stator controller in a VSC transmission system [23] with a switching
flux is estimated. A phase-locked loop (PLL) is, then, used to frequency of 2 kHz. As shown in Fig. 8, a high frequency ac

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
ZHI AND XU: DIRECT POWER CONTROL OF DFIG WITH CONSTANT SWITCHING FREQUENCY 115

Fig. 9. Simulated results under various stator active and reactive power steps and constant rotor speed. (a) Three-phase stator current (kA). (b) Three-phase rotor
current (kA). (c) Stator active power (MW) and reactive power (MVar). (d) dc link voltage (pu). (A) Rotor angular speed: 1.0 pu. (B) Rotor angular speed: 1.2 pu.

filter is connected to the stator side to absorb the switching

harmonics generated by the two converters.
In a practical system, voltages and currents are sampled at
the start of the sampling period. The required rotor control
voltage for the sampling period is, then, calculated and passed
to the PWM modulator. Inevitably, there is a time delay
between the instant sampling and PWM modulator’s receiving
the required rotor control voltage. The proposed DPC strategy’s
rotor voltage calculation is relatively simple, and the time delay
should be fairly small. Nevertheless, the simulated output rotor
voltage is delayed by 50 µs to closely represent a practical DPC
control system. During the simulation, the grid side converter is
enabled first, such that the converter dc link voltage is regulated.
The DFIG stator is, then, energized with the rotor rotating at
a fixed speed, and with the rotor-side converter disabled. This
starting process is not shown in the following results.
Initial studies with various active and reactive power steps
were carried out to test the dynamic response of the proposed
DPC strategy. The DFIG was assumed to be in speed control, i.e.,
with the rotor speed set externally, since in a practical system, the
wind turbine’s large inertia results in slow rotor speed change. Fig. 10. Harmonic spectra of the rotor and stator current ωr = 1.2 pu, Ps =
Simulated results are shown in Fig. 9(A) and (B) for rotor speed 1.5 MW, Qs = −0.5 MVar). (A) Rotor current. (B) Stator current.
of 1.0 and 1.2 pu, respectively, where the synchronous speed
is defined as 1 unit. The rotor-side converter was enabled at
0.2 s with the initial stator active and reactive power references active and reactive powers are within a few milliseconds. The
being −2 MW and −0.5 MVar, respectively (“−” refers to step change of one control variable, i.e., active or reactive power,
generating active power and absorbing reactive power). The does not affect the other due to the way the rotor voltage limit
active and reactive power references were step changed from was applied. There is no overshoot of either the stator/rotor
−2 to −1 MW at 0.4 s and from − 0.5 to + 0.5 MVar at 0.6 s, currents or the active/reactive powers.
respectively. The effectiveness of the proposed control strategy The harmonic spectra of the rotor and stator currents under
is clearly indicated in Fig. 9. The dynamic responses of both the proposed DPC strategy are shown in Fig. 10(a) and (b),

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
116 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 1, MARCH 2007

Fig. 11. Simulated results under various stator active and reactive power steps and rotor speed variation. (a) Three-phase stator current (kA). (b) Three-phase
rotor current (kA). (c) Stator active power (MW) and reactive power (MVar). (d) Rotor speed (pu). (A) With nominal Lm . (B) With 20% Lm error.

respectively. The dominant harmonics for both currents are

around the switching frequency of 2 kHz. For different operating
conditions, such as different rotor speed and active and reactive
powers, the system responses and current harmonic spectra were
found to be similar to those shown in Figs. 9 and 10.
Studies with various power steps during rotor speed and
machine parameter variations were carried out to further test
the proposed DPC schemes. Fig. 11(A) and (B) shows the
simulations with the mutual inductance used in the controller
having 0% and 20% errors, respectively. As shown, during
the period of 0.3–0.7 s, the rotor speed increased from 0.8
to 1.2 pu. Various power steps were applied, i.e., active and
reactive power references were changed from −2 to −1 MW
at 0.4 s, and from −0.5 to +0.5 MVar at 0.6 s, respectively.
As seen in Fig. 11(A), the system response with rotor speed
variation is satisfactory. Comparing Fig. 11(A) and (B), there
is hardly any difference, and even with such large inductance
errors, the system maintains superb performance under both
steady-state and transient conditions.
The proposed DPC robustness was also tested with grid-side
converter performance degraded to introduce significant dc volt-
Fig. 12. Simulation results of a complete DFIG-based generation system with
age variation during transients. The impact of such dc voltage step change of wind speed. (a) Wind speed (m/s). (b) DFIG stator active and
variation on the dynamic and steady-state performances of the reactive power input (MVar). (c) Rotor speed (pu). (d) dc link voltage (pu).
proposed DPC strategy is negligible. The results are not shown
here due to space limitations.
Further tests for a complete generation system including a maximum power tracking curve [4]. Fig. 12 shows the simulated
typical 2 MW wind turbine and the DFIG were carried out. results when the wind speed changed from 11 to 13 m/s at 0.5 s,
The DFIG was set in torque control, i.e., the speed is the result and then to 10 m/s at 1.5 s. The lumped inertia constant of the
of stator/rotor voltage/current and the mechanical torque. The system is set to a relatively small value of 0.2 s in the study to
active power reference for the DFIG was calculated from the reduce the simulation time. As can be seen from Fig. 12, the

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
ZHI AND XU: DIRECT POWER CONTROL OF DFIG WITH CONSTANT SWITCHING FREQUENCY 117

TABLE I REFERENCES
PARAMETERS OF THE SIMULATED DFIG
[1] W. Leonhard, Control of Electrical Drives. London, U.K.: Springer,
2001.
[2] 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,” Inst. Elect. Eng. Proc. Elect. Power Appl.,
vol. 143, no. 3, pp. 231–241, May 1996.
[3] H. Akagi and H. Sato, “Control and performance of a doubly-fed induction
machine intended for a flywheel energy storage system,” IEEE Trans.
Power Electron., vol. 17, no. 1, pp. 109–116, Jan. 2002.
[4] R. W. De Doncker, S. Muller, and M. Deicke, “Doubly fed induction
generator systems for wind turbines,” IEEE Ind. Appl. Mag., vol. 8, no. 3,
pp. 26–33, May/Jun. 2002.
[5] I. Takahashi and T. Noguchi, “A new quick-response and high-efficiency
control strategy of an induction motor,” Inst. Elect. Eng. Trans. Ind. Appl.,
vol. 22, no. 5, pp. 820–827, 1986.
[6] M. Depenbrock, “Direct self-control (DSC) of inverter-fed induction ma-
system operation is satisfactory and maximum power tracking chine,” IEEE Trans. Power Electron., vol. 3, no. 4, pp. 420–429, Oct.
1988.
is achieved during wind speed variation. [7] M. P. Kazmierkowski and A. Kasprowicz, “Improved direct torque and
flux vector control of PWM inverter-fed induction motor drives,” IEEE
Trans. Ind. Electron., vol. 42, no. 4, pp. 344–350, Aug. 1995.
V. CONCLUSION [8] P. Vas, Sensorless Vector and Direct Torque Control. Oxford, U.K.:
Clarendon, 1998.
This paper presented a new direct active and reactive power [9] U. Baader, M. Depenbrock, and G. Gierse, “Direct self-control (DSC) of
control strategy for a DFIG-based wind generation system. The inverter-fed induction machine—A basis for speed control without speed
measurement,” IEEE Trans. Ind. Appl., vol. 28, no. 3, pp. 581–588,
controller directly calculates the required rotor control volt- May/Jun. 1992.
age within each fixed time period based on the stator flux, [10] D. Casadei, G. Grandi, G. Serra, and A. Tani, “Effects of flux and torque
the rotor position, and the values of active and reactive pow- hysteresis band amplitude in direct torque control of induction machines,”
in Proc. 20th Int. Conf. Ind. Electron., Control Instrum., 1994, pp. 299–
ers and their errors. Constant converter switching frequency 304.
is achieved that eases the design of the power converter and [11] T. G. Habetler, F. Profumo, M. Pastorelli, and L. M. Tolbert, “Direct
the ac harmonic filter. Due to the relatively small power rating torque control of induction machines using space vector modulation,”
IEEE Trans. Ind. Appl., vol. 28, no. 5, pp. 1045–1053, Sep. 1992.
of the rotor-side converter, a method to limit the rotor voltage [12] Y. S. Lai and J. H. Chen, “A new approach to direct torque control of in-
is proposed to improve the system transient performance. The duction motor drives for constant inverter switching frequency and torque
impact of machine parameter variations on system response ripple reduction,” IEEE Trans. Energy Convers., vol. 16, no. 3, pp. 220–
227, Sep. 2001.
is analyzed and found to be negligible. Simulation confirms [13] N. R. N. Idris and A. H. M. Yatim, “Direct torque control of induc-
the effectiveness and robustness of the proposed DPC strategy tion machines with constant switching frequency and reduced torque
during various operating conditions and variations of machine ripple,” IEEE Trans. Ind. Electron., vol. 51, no. 4, pp. 758–767, Aug.
2004.
parameter. [14] J. Kang and S. Sul, “New direct torque control of induction motor for
minimum torque ripple and constant switching frequency,” IEEE Trans.
Ind. Appl., vol. 35, no. 5, pp. 1076–1082, Sep./Oct. 1999.
APPENDIX [15] T. Noguchi, H. Tomiki, S. Kondo, and I. Takahashi, “Direct power control
of PWM converter without power-source voltage sensors,” IEEE Trans.
A. Parameters Affect the Rotor Voltage Vector Calculation Ind. Appl., vol. 34, no. 3, pp. 473–479, May/Jun. 1998.
[16] G. Escobar, A. M. Stankovic, J. M. Carrasco, E. Galvan, and R. Ortega,
Considering Lσs  Lm and Lσr  Lm , the constant kσ and “Analysis and design of direct power control (DPC) for a three phase syn-
Lr /Lm can be simplified as chronous rectifier via output regulation subspaces,” IEEE Trans. Power
Electron., vol. 18, no. 3, pp. 823–830, May 2003.
[17] M. Malinowski, M. P. Kazmierkowski, S. Hansen, F. Blaabjerg, and
3 Lm 3 Lm G. D. Marques, “Virtual-flux-based direct power control of three-phase
kσ = =
2 σLs Lr 2 Ls Lr − L2m PWM rectifiers,” IEEE Trans. Ind. Appl., vol. 37, no. 4, pp. 1019–1027,
Jul./Aug. 2001.
3 Lm [18] K. P. Gokhale, D. W. Karraker, and S. J. Heikkila, “Controller for a wound
= rotor slip ring induction machine,” U.S. Patent 6 448 735 B1, Sep. 2002.
2 (Lσs + Lm )(Lσr + Lm ) − L2m [19] L. Xu and P. Cartwright, “Direct active and reactive power control of DFIG
for wind energy generation,” IEEE Trans. Energy Convers., vol. 21, no. 3,
3 Lm pp. 750–758, Sep. 2006.
=
2 Lm (Lσs + Lσr ) − Lσs Lσr [20] H. Akagi, Y. Kanazawa, and A. Nabae, “Generalized theory of the in-
stantaneous reactive power in three-phase circuits,” in Proc. Int. Power
3 Lm 3 1 Electron. Conf., 1983, pp. 1375–1386.
≈ = (A1) [21] H. W. Van De Broeck, H. C. Skudelny, and G. V. Stanke, “Analysis and
2 Lm (Lσs + Lσr ) 2 Lσs + Lσr
realization of a pulsewidth modulator based on voltage space vectors,”
Lr Lσs + Lm Lσs IEEE Trans. Ind. Appl., vol. 24, no. 1, pp. 142–150, Jan./Feb. 1988.
= =1+ . (A2) [22] S. R. Bowes, “Advanced regular-sampled PWM control techniques for
Lm Lm Lm drives and static power converters,” IEEE Trans. Ind. Electron., vol. 42,
no. 4, pp. 367–373, Aug. 1995.
B. Parameters of the DFIG Used for Simulation [23] J. L. Thomas, S. Poullain, and A. Benchaib, “Analysis of a robust DC-bus
voltage control system for a VSC-Transmission scheme,” in Proc. 7th Int.
See Table I for listing of the DFIG Parameters. Conf. AC-DC Power Trans., London, U.K., 2001, pp. 119–124.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.
118 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 22, NO. 1, MARCH 2007

Dawei Zhi (S’07) received the B.Sc. and M.Sc. de- Lie Xu (M’03–SM’06) received the B.Sc. degree in
grees in electrical and electronic engineering from electrical and electronic engineering from Zhejiang
Zhejiang University, Hangzhou, P. R. China, in 2000 University, Hangzhou, China, in 1993, and the Ph.D.
and 2003, respectively. He is currently working to- degree in electrical and electronic engineering from
ward the Ph.D. degree at Queen’s University of the University of Sheffield, Sheffield, U.K., in 1999.
Belfast, Belfast, U.K. From 1999 to 2000, he was with the Centre
From 2003 to 2005, he was with Delta Power Elec- for Economic Renewable Power Delivery (CERPD),
tronics Center, Shanghai, China. His current research University of Glasgow, Glasgow, U.K., and from
interests include power electronics, machine drives, 2001 to 2003, he was with ALSTOM T&D, Stafford,
and control and operation of DFIG for wind energy U.K. He is now with the School of Electronic, Elec-
generation. trical Engineering and Computer Science, Queen’s
University of Belfast, Belfast, U.K. His current research interests include power
electronics, renewable energy, and application of power electronics to power
systems.

Authorized licensed use limited to: Consortium - Algeria (CERIST). Downloaded on March 18,2024 at 13:14:49 UTC from IEEE Xplore. Restrictions apply.