IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 28, NO.

3, JULY 2013 1297

A Strategy for Real Power Control in a Direct-Drive

PMSG-Based Wind Energy Conversion System
Omid Alizadeh, Student Member, IEEE, and Amirnaser Yazdani, Senior Member, IEEE

Abstract—As the penetration of wind energy into the power control, rather than the (slow) pitch-angle control presented in
system continues to grow, wind energy conversion systems [4]–[6]. Thus, the pitch-angle control is not exercised for output
(WECSs) are increasingly expected to be able to control their real power control, but it is employed, exclusively and, as per
output real power, while retaining their maximum power-point
tracking (MPPT) capability. This paper proposes a simple real the common practice, for limiting the mechanical power if the
power control strategy, which augments the MPPT feature of turbine overspeeds.
modern WECSs, and is based on rapid torque control as opposed Rapid torque control, however, can excite drive-train tor-
to the traditional pitch-angle control. This paper presents the sional modes, especially since the drive-train stiffness is, in
implementation of the proposed control strategy for a direct-drive general, inversely proportional to the number of generator
WECS that employs the permanent-magnet synchronous gener-
ator, even though the proposed method can also be extended to poles [7] and is therefore low in a WECS with a high-pole
other classes of electronically interfaced WECSs. The paper also PMSG. Moreover, a high-pole PMSG possesses no inherent
presents a parameter-tuning procedure for the proposed control damping [8]. Drive-train oscillations, if not damped, impact the
strategy. The effectiveness of the proposed control strategy is operation and may even lead to instabilities. Thus, a supple-
demonstrated through mathematical analysis and time-domain mentary active damping scheme is designed for the proposed
simulation studies.
power-control strategy, based on a detailed mathematical model
Index Terms—Control, damping, direct drive, eigenvalue anal- and eigenvalue analysis of the WECS. The proposed control
ysis, permanent-magnet synchronous generator (PMSG), wind
strategy and its supplementary active damping scheme enable
energy.
the control of the WECS output real power, from a low value
up to the maximum power corresponding to the prevailing wind
I. INTRODUCTION speed. The active damping strategy, however, is not unique and
may be achieved through other reported techniques [9]–[11].

T HE anticipated large-scale integration of wind energy

conversion systems (WECSs) into the electric power
system indicates that system operators should be able to con-
II. STRUCTURE OF THE DIRECT-DRIVE WECS
Fig. 1 illustrates a simplified schematic diagram of a direct-
trol the output real and reactive powers of the WECSs, to drive PMSG-based WECS. The WECS is composed of a wind
more effectively take part in the control of the power system turbine, a high-pole PMSG, and a power-electronic ac-dc-ac
and to ride through faults and other contingencies [1]. While converter, which interfaces the PMSG to the host utility grid.
the reactive-power controllability of electronically interfaced The ac-dc-ac converter, in turn, consists of two back-to-back
WECSs is widely recognized, their real power controllability voltage-sourced converters, VSC1 and VSC2. The converter
has received insignificant attention and, thus far, been merely VSC1 controls the PMSG torque and, thus, the power that the
utilized for maximum power-point tracking (MPPT). PMSG extracts from the wind turbine, whereas VSC2 regu-
This paper proposes a strategy for controlling the output real lates the dc-link voltage by controlling the real power that it ex-
power of a direct-drive WECS that employs a high-pole per- changes with the grid. The converter VSC2 can also exchange
manent-magnet synchronous generator (PMSG) [2]. The choice reactive power with the grid, to support the grid or enhance
is based on the expectation that PMSG-based WECSs will be voltage stability [12]. In Fig. 1, the composition of the wind tur-
widely deployed in the future, due to their low-loss generators, bine, PMSG, VSC1, and the scheme that controls VSC1 is la-
low maintenance requirements, and quiet drive-trains [3]. The beled as the energy capture subsystem, whereas the composition
proposed real power control strategy is based on rapid torque of the dc-link capacitor, VSC2, the tie reactor , and the con-
trol scheme for the regulation of the dc-link voltage and reactive
power is identified as the controlled dc-voltage power port [13].
Manuscript received June 29, 2011; revised November 04, 2012; accepted This paper exclusively studies the dynamics of the energy cap-
April 04, 2013. Date of publication May 03, 2013; date of current version June ture subsystem; this is possible since the arrangement shown in
20, 2013. This work was supported in part by the Natural Sciences and Engi-
neering Research Council (NSERC) of Canada, and in part by the Faculty of En- Fig. 1 effectively decouples the dynamics of the utility grid from
gineering, the University of Western Ontario. Paper no. TPWRD-00558-2011. those of the wind turbine, drive-train, and PMSG.
O. Alizadeh is with the University of Western Ontario, London, ON N6G 5B9
Canada (e-mail: [email protected]).
A. Yazdani is with Ryerson University, Toronto, ON M5B 2K3, Canada III. MATHEMATICAL MODEL AND CONTROL SCHEMES
(e-mail: [email protected]).
This section presents a mathematical model for the energy
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. capture subsystem of the WECS of Fig. 1. For the sake of com-
Digital Object Identifier 10.1109/TPWRD.2013.2258177 pactness, hereafter, the same notation will be adopted for a vari-

0885-8977/$31.00 © 2013 IEEE

1298 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 28, NO. 3, JULY 2013

Fig. 1. Simplified schematic diagram of the direct-drive WECS.

Fig. 2. Block diagram of the PMSG power control scheme.

able and its Laplace transform. In addition, the superscript sig-
nifies the reference value (setpoint) for a variable.
If the PMSG torque is controlled, the PMSG power can be
A. Mechanical Torque and Drive-Train
regulated. Fig. 2 shows the block diagram of a control scheme
The mechanical torque of a wind turbine is given by: whose main objective is to force the PMSG power to track
the power setpoint . The setpoint is determined based on
(1) the mode of operation, as will be explained in the next section.
Fig. 2 also illustrates the pitch-angle control process whose
where is the turbine torque (in Nm), is the turbine function is to ensure that the turbine and PMSG speeds do not
power (in watts), is the turbine angular speed (in radians per exceed the maximum permissible value, . Thus, if ex-
second), is the turbine radius (in meters), is the air mass den- ceeds , a PI compensator increases to decrease the tur-
sity (in ), is the wind speed (in meters per second), bine power and regulate at ; if is smaller than ,
is the turbine tip-speed ratio (unitless), is the the compensator output is saturated at its lower limit, , to
pitch angle (in degrees), and (unitless) is the so-called tur- maximize the turbine power. Fig. 2 further shows that the PI
bine power efficiency [14]. compensator output passes through a rate limiter which repre-
Drive-train dynamics may be represented by models of dif- sents the limited speed at which the pitch angle can be changed
ferent levels of complexity. However, as discussed in refer- in practice.
ences [15] and [16], a two-mass model is adequate for cap-
turing the dynamics that affect stability; higher-order models IV. PROPOSED CONTROL AND MODES OF OPERATION
are commonly employed for studying the mechanical fatigue of The objective of the proposed control is to enable the WECS
the turbine drive-train. Thus, ignoring the mechanical losses, the to exercise power-flow control in addition to the MPPT. Thus,
drive-train is represented by the following two-mass model: two modes of operation are defined for the WECS of Fig. 1: (1)
the MPPT mode, and (2) the controlled-power (CP) mode. The
(2)
operating mode is determined by the way that the setpoint is
stipulated, as explained next.
(3)
A. MPPT Mode of Operation
(4)
In the MPPT mode, the objective is to maximize the power
where is the PMSG rotor speed (in rad/s); and re- that the turbine extracts. This can be achieved if is maxi-
spectively signify the turbine and PMSG moments of inertia (in mized. To maximize , must be kept constant at its optimum
); is the drive-train stiffness (in Nm/rad); the variable value , regardless of the wind speed. The objective is ful-
represents the torsional displacement of the drive-train (in rad); filled if the PMSG power setpoint is determined based on the
and denotes the PMSG torque (in Nm). Equations (1) through following law [17]:
(4) constitute a state-space drive-train model for the energy cap-
(6)
ture subsystem.
in which the constant is
B. Permanent-Magnet Synchronous Generator (PMSG)
The PMSG torque control is performed in a rotating frame (7)
whose direct axis is aligned with the PMSG rotor flux vector,
as discussed in [13]. The control can be tuned such that It then follows from assuming a fast control that .
responds to its setpoint based on the following first-order Thus, (6) can be rewritten as
transfer function:
(8)
(5)
Fig. 3 illustrates the characteristic curve of a wind turbine
for which the time constant is a design parameter [13]. (heavy solid line), for a wind speed. The figure also plots the
ALIZADEH AND YAZDANI: STRATEGY FOR REAL POWER CONTROL 1299

Fig. 4. Block diagram illustrating the generation of the power setpoint .

than ; otherwise, is equal to and the energy

capture subsystem operates in the MPPT mode. Therefore, to
permanently leave the system in the MPPT mode, it is sufficient
to assign an adequately large value (e.g., larger than the
value of that corresponds to the rated wind speed).

V. EIGENVALUE ANALYSIS
Fig. 3. Characteristic curves of a wind turbine for a wind speed and two dif-
ferent values of pitch angle, that is, (heavy solid line) and (light An eigenvalue analysis is performed to reveal the dynamic
solid line). properties of the energy capture subsystem and to tune the pa-
rameters of the proposed power control strategy. To that end,
a linearized model is developed and analyzed. The numerical
PMSG power versus rotor speed, based on (8) (dashed line). It is examples presented hereafter are based on an example WECS
observed that if the WECS is in the MPPT mode, the operating whose parameters are reported in Appendix A. The same ex-
point settles at point A (intersection of the two curves) which ample WECS has also been simulated in time domain for pro-
corresponds to the maximum turbine power at the given ducing the results reported in Section VII.
wind speed. As Fig. 2 indicates, the compensator of the power control
scheme can be described by

B. Controlled-Power (CP) Mode of Operation (9)

In the CP mode, the objective is to regulate the WECS output
power at the command value , regardless of the wind Replacing with in (9) and expressing the resultant in
speed. Thus, is given the value of . Let us assume time domain, one finds
that, initially, the WECS is in the MPPT mode, ,
, and ; then the value of
(and therefore ) is rapidly changed from to (i.e.,
subsequent to a switching from the MPPT mode to the CP (10)
mode). As Fig. 3 indicates, this causes the PMSG power to
drop below the turbine power and results in an increase in As Fig. 2 indicates, if saturation is ignored, in (5) is equal
towards a new value, . Depending on the wind speed, to the compensator output and one finds
can be larger than the maximum permissible rotor speed,
(11)
, as for the example illustrated in Fig. 3. The situation is
circumvented by the pitch-angle control mechanism; thus, once
exceeds , the pitch-angle control scheme increases Eliminating between (10) and (3), and then sub-
and consequently alters the power-speed characteristic of the stituting for from (11) in the resulting equation, one
wind turbine, to the one shown by light solid line in Fig. 3, deduces
such that drops to and the rotor speed settles at
(corresponding to point C in Fig. 3). To ensure that the PMSG
and turbine power-speed curves have at least one crossing
point [see Fig. 3], in the CP mode is limited to the value (12)
. Therefore, if is so large that cannot overtake
it at the given wind speed, then will be limited to Equations (2) through (4), (11), and (12), along with the alge-
and, effectively, the system continues to operate in the MPPT braic (1), constitute the following nonlinear state-space model
mode until either there will be a rise in the wind speed (thus for the energy capture subsystem:
increasing the corresponding ) or the system operator steps
down the command . (13)
Fig. 4 illustrates a mechanism for selecting between the
MPPT and CP modes of operation. As Fig. 4 shows, the set- for which is the vector of state
point is obtained from the output of a hard limiter whose variables (superscript T denotes matrix transposition), and
input and upper saturation limit are and , respec- are the (inter-related) control inputs, and is the
tively (the lower saturation limit is zero). Thus, is equal to disturbance input. is a vector of nonlinear functions of
and the CP mode is exercised, as long as is smaller the state variables and inputs. It should be noted that (13)
1300 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 28, NO. 3, JULY 2013

TABLE I
EIGENVALUES OF THE ENERGY CAPTURE SUBSYSTEM; ,
FOR THE CP MODE

Fig. 6. Block diagram illustrating the implementation of the active damping

control.

Fig. 5. Absolute value of the real part of the dominant eigenvalue of as a

function of and .

assumes a constant pitch angle (i.e., it ignores the dynamics

of the pitch-angle control mechanism) to keep the mathemat-
ical model tractable. The approximation is plausible since Fig. 7. Control block diagram of the active damping scheme.
the pitch-angle is normally settled at its minimum value, and
varies only if the turbine speed exceeds its maximum value.
Moreover, the ignored dynamics are, by design and by nature, where and are matrices whose elements are functions of
remarkably slower than those of the state variables in . the system steady-state operating point; the matrices are intro-
However, the simulation model used for the assessment of the duced in Appendix B.
proposed control strategy includes the pitch-angle control loop Table I reports the eigenvalues of and . It is observed
(Section VII). that in the MPPT mode, the system has two pairs of complex-
In the MPPT mode, is determined based on (6). Taking conjugate eigenvalues that correspond to two stable but poorly
derivatives with respect to time from both sides of (6), and elim- damped eigenmodes. The situation is even worse for the CP
inating between the resulting equation and (3), one finds mode; in the CP mode, the energy capture subsystem has an
unstable oscillatory mode. Thus, both modes of operation call
for an active damping mechanism, which will be described in
(14)
the next section.

Substituting in (13) for and , respectively, from VI. ACTIVE DAMPING STRATEGY
(6) and (14), and linearizing the resulting set of equations, one
deduces A. Structure
The active damping scheme designed hereafter augments the
(15) PMSG torque setpoint with a high-pass filtered measure of the
rotor speed, through the scheme illustrated in Fig. 6. As Fig. 6
where and are matrices whose elements are functions of shows, first the ac component of the rotor speed is extracted
the steady-state operating point of the system, and “~ ” denotes by passing through a high-pass filter, . Then, a compen-
the small-signal perturbation of a variable; the matrices are in- sator processes the error between zero and the filter output
troduced in Appendix B. , and determines the supplementary component for the
For the example WECS, Fig. 5 plots the absolute value of PMSG torque setpoint . As illustrated in Fig. 7, the active
, that is, the real part of the dominant eigenvalue of , as a damping scheme of Fig. 6 results in a control loop whose ob-
function of and ; the dominant eigenvalue is defined as jective is to (rapidly) force the ac component of to zero.
the eigenvalue with the smallest real part (in absolute value), and Let be a second-order high-pass filter of the form
has been calculated for the operating point that corresponds to
9.0 m/s. As Fig. 5 illustrates, is maximized if 1.0 (17)
and 2.4; these values are adopted for subsequent analyses.
In the CP mode, and, thus, the linearized version
of (13) takes the form for which and are the corner frequency and quality factor,
respectively. For the example WECS, Fig. 8 shows a family of
curves that plot the imaginary part (frequency) of the unstable
(16) complex-conjugate eigenvalues in the CP mode (see Table I),
for a corresponding set of wind speeds, as a function of the ratio
ALIZADEH AND YAZDANI: STRATEGY FOR REAL POWER CONTROL 1301

Differentiating (3) with respect to time, eliminating

between the result and (4), and expressing the result in the small-
signal form, one deduces

(21)

Substituting for from (19) in (21), one finds

(22)

Eliminating between (22) and the expression of

Fig. 8. Variation of the damped natural frequency as a function of the normal- (17) in the small-signal time-domain form, one concludes that
ized power command in the CP mode.

. As the figure indicates, the frequency of the un-

stable mode varies over a fairly narrow range about 4.05 rad/s.
Considering this frequency band, the choices of and
rad/s result in a small difference between and , (23)
in terms of magnitude and phase angle. As will be discussed in
the next subsection, the choice of a pure gain for results
in stable operation. Equations (19), (20), (23), and the small-signal versions of
(2)–(4) constitute a state-space model for the energy capture
subsystems augmented with the presented active damping
B. Eigenvalue Analysis scheme. Replacing and in the state-space model
with their expressions corresponding to each mode of opera-
To assess the effectiveness of the presented active damping tion, the following linear state-space models are obtained:
scheme, and to find the optimum value for the gain ,
the state-space model of Section V is modified.
If , as Fig. 6 indicates, the supplementary compo- for the MPPT mode (24)
nent of the PMSG torque setpoint is
for the CP mode (25)
(18)

It then follows from replacing by in (5) [see where ; the matrices

Fig. 6], expressing the result in the small-signal time-domain , , , and are introduced in Appendix B.
form, and substituting for from (18) in the final form that For the example WECS, Fig. 9 plots the migration of the (ini-
tially) unstable complex-conjugate eigenvalues in the CP mode,
(19) when the active damping mechanism is enabled and is varied
from zero to . The migration plot is sketched for the
operating point that corresponds to 9.0 m/s and
Eliminating between (10) and (3), linearizing the re-
1.5 MW. It is observed that the eigenvalues, which
sult, and substituting for from (19) in the final form, one
are unstable for , migrate toward the left-half plane as
deduces
is increased, but move back toward the imaginary axis once
surpasses a certain value. This behavior indicates the exis-
tence of an optimum value for . The optimum value is com-
puted such that the complex-conjugate eigenvalues possesses
the maximum damping ratio. This, for the example WECS, cor-
responds to the choice of , which results in the
smallest angle between the real axis and the tangent to the mi-
gration plot.
Table II reports the eigenvalues of the energy capture sub-
system under the MPPT and CP modes of operation. A com-
(20) parison of the results reported in Table I confirms the improved
1302 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 28, NO. 3, JULY 2013

Fig. 9. Migration plot of the (initially) unstable eigenvalues in the CP mode

for different values of .

TABLE II
EIGENVALUES OF THE ENERGY CAPTURE SUBSYSTEM WITH ACTIVE DAMPING
CONTROL; 9 m/s, , FOR THE CP MODE 1.5 MW

Fig. 11. Response to the operation mode change from MPPT to CP.

shows, for , the response is oscillatory and unstable,

while results in sustained oscillations. How-
ever, as expected, results in a damped re-
sponse. Fig. 11 further indicates that the frequency of oscil-
lations closely match those predicted by the eigenvalue anal-
ysis. For example, Fig. 11(b) indicates that the period of oscil-
lations for is about 1.56 ( 4.69/3) s, which
corresponds to an angular frequency of about 4.03 rad/s. This
frequency is very close to that indicated by Fig. 9 for
.
Fig. 10. Migration plot of the dominant eigenvalues for , when
the wind speed varies from 6 to 12 m/s. (a) and (b) VII. SIMULATION RESULTS
.
To further demonstrate the effectiveness of the proposed
power control strategy, the detailed switched model of the
damping of the complex-conjugate eigenvalues with higher un- example WECS has been subjected to various operating con-
damped natural frequency, under the presented active damping ditions. In the graphs to follow, the angular velocities are
scheme. More important, the active damping scheme has stabi- expressed in rad/s, the torques are expressed in MNm, the
lized the system in the CP mode, as all of the eigenvalues lie in powers are expressed in MW, the dc-link voltage is expressed
the left-half plane. in kV, the wind speed is expressed in m/s, and the pitch angle
For the example WECS, Fig. 10 plots the migration of the is expressed in degrees.
dominant eigenvalues for a constant and two different values Figs. 12 and 13 illustrate the WECS response to changes in
of the ratio , when the wind speed varies from 6 the operating mode and wind speed. Before s,
to 12 m/s. It is observed that some eigenvalues approach the MW and the wind speed is assumed to be 9.0 m/s. In this case,
imaginary axis as the wind speed increases. Nonetheless, the the WECS is operating in the MPPT mode and the turbine yields
system remains stable over the entire wind speed range, and the the maximum power. The example WECS then experiences the
eigenmodes are well damped. following sequence of events: (1) at s, is stepped
To verify the accuracy of the developed mathematical model, from 6.0 MW down to 1.5 MW and, since the maximum power
the response of to an abrupt switching from the MPPT mode corresponding to m/s is about 2.1 MW, the operation
to the CP mode is depicted in Fig. 11. The response is obtained mode is changed from the MPPT mode to the CP mode; (2) at
from a detailed switched model of the example WECS, devel- s, the wind speed assumes a step change from 9.0 to 12
oped in the PSCAD/EMTDC environment [18], for the gains m/s and thus the WECS remains in the CP mode; (3) at
, , and . As Fig. 11 s, is stepped further down to 0.5 MW and thus the WECS
ALIZADEH AND YAZDANI: STRATEGY FOR REAL POWER CONTROL 1303

Fig. 13. Response to changes in the operation mode and wind speed (cont.).

TABLE III
WIND TURBINE PARAMETERS

TABLE IV
PMSG PARAMETERS

Fig. 12. Response to changes in the operation mode and wind speed.

continues to operate in the CP mode; (4) at s, the wind TABLE V

COMPENSATORS AND OTHER PARAMETERS
speed changes stepwise from 12 to 9.0 m/s. However, since the
maximum power corresponding to m/s is larger than
MW, the WECS retains its CP operating mode; (5)
at s, is stepped up to 3.5 MW. This command is
larger than the maximum power for m/s, that is, 2.1
MW. Therefore, the WECS experiences a change from the CP
mode to the MPPT mode and, as such, its output power settles
at 2.1 MW; and (6) at s, the wind speed again rises the output power transiently overshoots, but reverts to its steady
stepwise from 9.0 to 12 m/s and, since the maximum power state value of 2.1 MW (that is, the maximum power for
corresponding to m/s (about 5.0 MW) is larger than the wind speed of 9.0 m/s). The reason for the overshoot is the
3.5 MW, the operating mode reverts back to the CP stored mechanical energy of the drive-train inertia, which is
mode and the output power settles at 3.5 MW. momentarily released.
Fig. 12 shows that the output power rapidly tracks Fig. 13 shows the waveform of the supplementary compo-
during the periods when the WECS operates in the CP mode, nent of the PMSG torque setpoint [Fig. 13(a)], and the pitch
that is, from to 150 s, and from s onwards. angle waveform [Fig. 13(b)]. It is observed that tran-
The figure also indicates that the rises in the wind speed, at siently responds to each disturbance, but settles down at zero.
s and 110 s, when the system is in the CP mode, result By contrast, the pitch angle only responds to those disturbances
in transient excursions in the output power, but have no effects that cause the drive-train speed to exceed (and to be in need of
on the steady-state command following. It is interesting to note regulation at) the maximum permissible value of 1.35 rad/s [see
that at s when is changed from 0.5 to 3.5 MW, Fig. 12(a) and (b).
1304 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 28, NO. 3, JULY 2013

(26)

(27)

VIII. CONCLUSION (30)

This paper presented a simple real power control strategy
based on rapid control of the generator torque. The implementa-
(31)
tion of the proposed control was demonstrated for a direct-drive
WECS that employs a PMSG. It was shown that the proposed
strategy enables rapid control of the WECS output power, from
small values up to the maximum power that corresponds to REFERENCES
the prevailing wind conditions, but results in the instability of [1] W. Qiao and R. G. Harley, “Grid connection requirements and solu-
tions for dfig wind turbines,” in Proc. IEEE Energy 2030 Conf., At-
the drive-train torsional modes. Therefore, the paper also pre- lanta, GA, USA, Feb. 2009, pp. 1–8.
sented a supplementary active damping scheme and a procedure [2] Y. Chen, P. Pillary, and A. Khan, “PM wind generator topologies,”
for parameter tuning. The effectiveness of the proposed con- IEEE Trans. Ind. Appl., vol. 41, no. 6, pp. 1619–1626, Nov./Dec. 2005.
[3] L. Soderlund and J. T. Eriksson, “A permanent-magnet generator for
trol strategy was demonstrated by mathematical analyses and wind power applications,” IEEE Trans. Magn., vol. 32, no. 4, pp.
time-domain simulation studies. 2389–2392, Jul. 1996.
[4] E. Muljadi and C. P. Butterfield, “Pitch-controlled variable-speed
wind turbine generation,” IEEE Trans. Ind. Appl., vol. 37, no. 1, pp.
APPENDIX A 240–246, Jan./Feb. 2001.
PARAMETERS OF THE EXAMPLE WECS [5] P. K. Keung, P. Li, H. Banakar, and B. T. Ooi, “Kinetic energy of
wind — turbine generators for system frequency support,” IEEE Trans.
The parameters of the example WECS are reported in Power Syst., vol. 24, no. 1, pp. 279–287, Feb. 2009.
Table III and Table IV. The other parameters are introduced [6] F. A. Bhuiyan and A. Yazdani, “Multimode control of a dfig — based
in Table V. The saturation limits on the PMSG torque setpoint wind —power unit for remote applications,” IEEE Trans. Power Del.,
vol. 24, no. 4, pp. 2079–2089, Oct. 2009.
(see Figs. 2 and 6) are set to 4.8 MNm. [7] C. Jauch, “Transient and dynamic control of a variable speed wind tur-
bine with synchronous generator,” Wind Energy, vol. 10, no. 3, pp.
APPENDIX B 247–269, Feb. 2007.
[8] G. Michalke, A. Hansen, and T. Hartkopf, “Control strategy of a vari-
ANALYTICAL FORMS OF MATRICES able speed wind turbine with multipole permanent magnet synchronous
The analytical forms of the matrices , , , , , and generator,” presented at the Eur. Wind Energy Conf. Exhibit., Milan,
Italy, May 2007.
are presented in (26) through (31). Matrices and can [9] A. Hansen and G. Michalke, “Modeling and control of variable —
be obtained by eliminating 6th and 7th rows and columns from speed multi — pole permanent magnet synchronous generator wind
matrices and , respectively. The entries of these matrices turbine,” Wind Energy, vol. 11, no. 5, pp. 537–554, 2008.
[10] J. M. Mauricio, A. E. Leon, A. Gomez-Exposito, and J. A. Solsona,
are functions of the steady-state values of the variables (denoted “An electrical approach to mechanical effort reduction in wind energy
by the overline). See (27) at the top of the page. conversion systems,” IEEE Trans. Energy Convers., vol. 23, no. 4, pp.
1108–1110, Dec. 2008.
[11] H. Geng, D. Xu, B. Wu, and G. Yang, “Active damping for PMSG-
(28) based WECS with DC-link current estimation,” IEEE Trans. Ind. Elec-
tron., vol. 58, no. 4, pp. 1110–1119, Apr. 2011.
[12] H. W. Kim, S. S. Kim, and H. S. Ku, “Modeling and control of PMSG-
(29) based variable-speed wind turbine,” Elect. Power Syst. Res., vol. 80,
no. 1, pp. 46–52, Jan. 2010.
ALIZADEH AND YAZDANI: STRATEGY FOR REAL POWER CONTROL 1305

[13] A. Yazdani and R. Iravani, Voltage-Sourced Converters in Power Sys- Amirnaser Yazdani (M’05–SM’09) received the
tems. Piscataway, NJ: IEEE/Wiley, 2010. Ph.D. degree in electrical engineering from the
[14] S. Heier, Grid Integration of Wind Energy Conversion Sys- University of Toronto, Toronto, ON, Canada, in
tems. Hoboken, NJ: Wiley, 1998. 2005.
[15] S. M. Muyeen, M. H. Ali, R. Takahashi, T. Murata, and J. Takahashi, He was with the University of Western Ontario
“Comparative study on transient stability analysis of wind turbine gen- (UWO), London, ON, Canada. Currently, he is
erator system using different drive train models,” IET Renew. Power an Associate Professor with Ryerson University,
Gen., vol. 1, no. 2, pp. 131–141, Jun. 2007. Toronto, ON, Canada. He is a co-author of the
[16] G. Ramtharan, N. Jenkins, O. Anaya-Lara, and E. Bossanyi, “Influence book Voltage-Sourced Converters in Power Systems
of rotor structural dynamics representations on the electrical transient (IEEE/Wiley, 2010). His research interests include
performance of FSIG and DFIG wind turbines,” Wind Energy, vol. 10, modeling and control of electronic power converters,
pp. 293–301, Jul./Aug. 2007. renewable electric power systems, distributed generation and storage, and
[17] S. Morimoto, H. Nakayama, M. Sanada, and Y. Takeda, “Sensorless microgrids.
output maximization control for variable-speed wind generation
system using IPMSG,” IEEE Trans. Ind. Appl., vol. 41, no. 1, pp.
60–67, Jan./Feb. 2005.
[18] PSCAD/EMTDC v. 4.2, Manitoba. Winnipeg, MB, Canada: HVDC
Research Centre.

Omid Alizadeh (S’12) received the B.Sc. degree

in electrical engineering from Khajeh-Nasir Toosi
University, Tehran, Iran, in 2005, the M.Sc. degree in
electrical engineering from the University of Tehran,
Tehran, Iran, in 2008, and is currently pursuing
the Ph.D. degree in electrical engineering at the
University of Western Ontario (UWO), London,
ON, Canada.
His research interests include design, modeling,
and control of wind energy conversion systems.