Application of a Matrix Converter for the Power Control of a Variable-Speed

Wind-Turbine Driving a Doubly-Fed Induction Generator

L Zhang*, C Watthanasarn and W. Shepherd

Department of Electronic and Electrical Engineering, University of Bradford,

Richmond Road, Bradford, BD7 lDP, UK
phone: +44-1274-38402 1; fax: +44- 1274-391-521;
email: [email protected]*,[email protected]

Abstract--A grid-connected wind-power generation scheme In this paper, the use of a direct matrix converter [4,5,6] for
using a doubly-fed induction generator @FIG) in conjunction the control of the rotor-side currents of a DFIG system is
with a direct AC-AC matrix converter is proposed. The proposed. Such a configuration offers certain advantages,
analysis employs a stator-flux vector-control algorithm and a notably:
space vector modulated matrix converter to control the
1) the power converter requires no bulky and costly energy
generator rotor current. The system enables optimal speed
tracking for maximum energy capture from the wind and high storage components, like those in the dc-link converter,
performance active and reactive power regulation. The paper 2) the control scheme required by a direct AC-AC
describes the operating principles of this power generation conversion scheme is simpler than that used by a two-
scheme. The matrix converter-based rotor current control stage power conversion.
scheme is highlighted. Simulation studies of the proposed
power generation system were carried out. Results obtained For maximum energy capture from the wind, high
are presented illustrating the good control performance of the performance speed control is desired to enable the generator
system. speed to closely track the value predicted by the wind turbine
power-speed characteristic curve. This is realized by
I. INTRODUCTION regulating the rotor current using a stator-flux, field-oriented
Wind power is widely recognized as a viable source of scheme and a space-vector modulated matrix converter. The
renewable energy. In the UK the percentage of electrical control scheme also enables flexible adjustment of the power
power generated by wind-driven turbines to the total power factor. In this paper, the operating principles of this power
generation increases steadily. Most wind turbines currently generation scheme and the control method used are
operate at constant speeds through grid-connected discussed. Simulation studies were carried out using a
alternators or induction generators. The present system deals 7.5kW induction generator. The results under various
with a variable-speed wind generation scheme. This has a operating conditions are presented. Features of the system
higher energy cap? .? capability than a constant speed and its performance are analysed.
system and reduces mechanical stresses and audible noises.
Integration of the three-phase induction machine and power 11. PRINCIPLE OF A VARIABLE-SPEED
electronic converter provides an effective means to achieve WIND-TURBINE DRIVEN DFIG
variable-speed, constant-frequency (VSCF) wind-power For a horizontal-axis wind turbine, the mechanical output
generation. A number of machine-converter configurations power P, is a function of wind velocity V, and can be
have been proposed in the literature [1,2,3]. One of these, expressed as
considered most attractive by the authors, is that developed
by Pena et a1 [3], which employs a doubly-fed induction pt =+cpp.-R=v;, (1)
generator (DFIG) interfaced to the power grid using a dc- where C, is the coefficient of power, p is the air density and
link voltage-source converter, commonly known as a R is the turbine radius. The turbine torque is the ratio of the
Scherbius drive. In such a configuration, the converter rating output power to the shaft speed at, namely, T,=Pimt.
required is only the slip-fraction of the machine power rating
due to the restricted speed range, and sinusoidal currents can The turbine is normally coupled to a generator shaft through
be obtained from both the stator and rotor windings. a gearbox whose gear ratio <is chosen so as to maintain the
However, the use of PWM dc-link converters requires two- generator shaft speed within a desired speed range.
stage power conversion, namely rectification and inversion, Neglecting the transmission losses, the torque and shaft
which demands a cumbersome control strategy for the speed of the wind turbine, referred to the generator side of
overall power generation system. the gearbox, are given by

0-7803-3932-0
906
 T, r,
= - and w , = <wt, (2)
4-
respectively where T, is the driving torque of the generator
and w, is the generator shaft speed.

A wind turbine can only generate a certain percentage of

power associated with the wind. This percentage is
represented by Cp which is a function of wind speed, turbine
rotational speed and the pitch angle of specific wind turbine
blades.
Figure 2 Torque-speed characteristics of the wind turbine
It is convenient to combine V, and utin a single variable (referred to generator side with gearbox ratio=5.05)
called the tip-speed ratio h where h = Rwfl,. A non-linear
Control of a turbine shaft speed is through regulating T,
relationship between Cp and iL can be obtained
which is directly determined by the DFIG generated power.
experimentally for any given wind turbine. A measured Cp-A
This consists of both the stator and rotor power components.
curve for a 7.5kW turbine [2] is shown in Fig.1 and is the
The rotor power P, may be drawn from or returned to the
basis of the wind turbine studied in this work. As shown in
utility grid depending on the operating mode of the DFIG.
Fig.2, for a particular wind speed within the range from cut-
For the subsynchronous mode (wr<us), the rotor power is
in to rated wind velocity, the turbine rotational speed can be
positive indicating that the power flows from the grid to the
adjusted so that the maximum Cp can be obtained. This, in
rotor. On the contrary, negative power in the super-
turn, causes the wind turbine to generate a maximum power
at that wind speed. With a variable-speed wind-turbine synchronous mode (wr>ws)means that rotor power is fed to
driven DFIG, the turbine shaft speed can be regulated the grid.
electrically.
Neglecting the stator copper loss and the motor core loss, the
The input driving power to the DFIG is equivalent to the stator power is approximately equal to the air-gap power
turbine power subtracted by the friction losses in the turbine which is defined as Pg=P,tP,,,h'. P , The DFIG generated
electrical power supplied to the grid is, thus, equal to PA-P,
shaft Pmech= P, - D o s which can also be expressed as
To extract the maximum possible power from a wind turbine
Pmech= Tmw,=Tm(1-S)~,, (3) the control scheme for a DFIG should attempt to enable the
where T, is the generator driving torque and per unit slip wind turbine to track its optimal power curve, as given in
s = ( w s - IDr) I u s . Fig 2 Commonly applied control strategies include DFIG-
rotor speed control and direct power control In this work the
The driving torque T, is thus related to turbine torque T, by latter is used
T,=T,-Dw, T, is balanced in steady-state operation by the
electromagnetic torque T, of the DFIG if the friction and 111. A MATRIX CONVERTER CONTROLLED DFIG
windage losses Dw: are accounted for in terms of system A. System ConJiguration
power losses In the generation system, an AC-AC matrix converter may
be used to supply the variable-frequency voltages to the rotor
terminals of the induction machine. Fig.3 shows schematics
Coefficient of power Cp of the matrix converter-DFIG configuration and its
05 r
simplified control scheme. The stator of the generator is
connected directly to the utility grid. A matrix converter is
inserted in the rotor circuit, giving direct AC-AC power
conversion between the rotor circuit and grid. The grid-side
connection is made via a three-phase LC filter to suppress
i
high-order harmonics. A matrix converter provides
t
01 I bidirectional power-flow control thereby enabling the DFIG
* 't to operate in either subsynchronous (or<ws) or super-
01 \
0 5 10
Tip speed ratio 1l 5
20 synchronous modes (w,>o,). In both modes the stator active
Figure 1 Coefficient of power for the wind turbine model power is generated from the DFIG and delivered to the grid.
On the other hand, the rotor active power is either supplied

907
to the machine in the subsynchronous mode or delivered to B. Principle ofActive and Reactive Power Control
the grid in the supersynchronous mode. To provide independent control of the stator active power P,
and reactive power Qs of the DFIG, by means of rotor
Now the output power of a wind turbine at a specific wind current regulation, it is necessary to define the dq
speed varies with change of the turbine shaft speed. The components of the rotor currents in the stator-flux oriented
control objective is to ensure that the power developed by the reference frame and show that P, and Q, can be represented
turbine is a maximum at any wind speed. The control as functions of the individual current components.
scheme must also maintain continuous power flow from the Subsequently, the P, and Qs commands can be used to
DFIG to the grid. To achieve this, a turbine shaft speed determine the reference rotor currents.
which results in a maximum turbine power must be
determined and the DFIG is controlled so as to obtain the Stator-flux oriented control is used to regulate the rotor
desired shaft speed. The desired shaft speed can be current. In this scheme the d components
determined by an optimal power tracking algorithm which is of the rotor current vector is aligned with the stator-flux
not fully investigated in this present work. Instead, the stator linkage vector I,,hence the active and reactive currents
active power is controlled directly assuming that a maximum
generator developed power is known. The ideal machine supplied to the power grid become linear functions of the
iutor current d and q components, given as
stator power, denoted by P,*,is used as the reference value
for the DFIG power control loop. In the inner current control (4)
loop, the stator-flux vector position is used to establish a -
reference frame that allows the d and q axis components of The magnitude of the stator magnetising current vector z,,f.
the rotor current to be controlled independently. Adjustment is a constant determined by the supply voltage. The stator
of the q-axis component of the rotor current. i,, controls active and reactive power components may be given as
either the generator developed-torque or the stator-side
active power of the DFIG (Ps).Regulating the d-axis
component, jdr, controls directly the stator-side reactive 3 3 L;
power flow (Qs). Q, = $ ( V ~ , I : ~- ~ k " , )= 2 v e l ds
gs e =Tu L,
E -ImS('ms -i:r)

(6)
When a matrix converter is used to control the DFIG, Knowing L,,, L, and ideal values of Ps- and Q,-, the
regulation of the rotor currents and phase-angle control of
the grid-side currents are a combined task which can be
reference values for J;~
and I:,. can be calculated directlj
accomplished by a single modulation scheme From the rotor from the above equations
current control loop, Fig 4, the desired rotor voltage vector is
generated and taken as the reference output voltage for the C. Rotor Current Regulation Using a Matrix Converter
matrix converter The space vector modulation technique [6] As the reference rotor current components are in stator-flux
is used to achieve the above control Objective oriented coordinates, these must be transferred to the same
reference frame as the DFIG rotor current vector. This is
achieved by rotating the rotor reference current vector by an
angular position p = (w - w )dt . Due to the rotor speed
variation, p is updated at every sample interval. Once the
reference frame for both the command and measured current
vectors are conformed, simple proportional plus integral (PI)
regulators can be used to control the d and q components of
power the rotor current. The cross-coupling effect of the d-g
components must be compensated at the output of the
controllers. In stator-flux oriented coordinates, these may be
expressed as

nv vir' = - - m 8 , d r i i r , (7)

The rotor voltage, thus obtained, defines the required line-to-

line voltage o f a matrix converter. Meanwhile, the converter
Figure 3 Configuration of the matrix converter-DFIG wind-
input voltage can be derived from the monitored values of
power generation scheme stator voltages Vds and vqs.Once these values are specified for
each sampling period, the matrix converter can be controlled kept below the synchronous speed resulting in the
to provide the rotor current regulation. A block diagram subsynchronous mode of operation. Fig. 5(c) shows the
illustrating the power control of the DFIG in stator flux waveforms of the converter input voltage vUsand filtered
oriented coordinates is shown in Fig.4. input current i,, indicating that the rotor power is supplied
by the grid. The stator active power P, follows the current
command iTr* as shown in Fig.5(d).

When the DFIG is operated in the supersynchronous mode, a

set of results, corresponding to those in Fig.5, is shown in
Fig.6. These differ from the results in Fig.5 in that the rotor
current tir is temporarily decreased hence allowing the
generator to accelerate. It can be seen, from Fig.6(c), that the
direction of the rotor power flow is reversed compared with
that of Fig.S(c).

Since the generator is driven by a constant mechanical

torque, constant shaft speed can be obtained by setting i, to
a constant value while id,.* may be set to any desired value.
Step increase and decrease in rotor current ii, during which
iir is maintained constant is illustrated in Fig.7(a) Rotor
Figure 4 Block diagram of the power control of a DFIG
current id,' is set equal to the magnetising current ,i which
means that the excitation current is supplied entirely to the
IV. SIMULATION STUDY AND DISCUSSION machine rotor. This results in unity power factor on the grid
A 7.5kW wind-turbine driven DFIG was simulated. The as the stator reactive power Qsis zero. The rotor current I,,
rating and parameters of this machine are given in the is sinusoidal as shown in Fig.7(b) while the stator power and
Appendix. Taking the stator d and q axes as the stationary reactive power are shown in Fig.7(c).
reference frame, the machine model used can be written
The rotor current waveform when the generator shaft speed
PLm 0
is gradually increased through synchronous speed is shown
0 Rs t p L s 0 pLm in Fig 8 It is seen that smooth transition of the rotor
PLm w,L, Rr + P L r w,Lr currents from sub to supersynchronous modes is achieved
PLm - w J r Rr + PLr The results in this section indicate that stable operation of
the DFIG can be achieved by means of stator-flux oriented
The mechanical part of the machine model is given as control This can be effectively implemented using a matrix
2 - T, = J p w r +- D w ~ , (10)
converter which is capable of producing sinusoidal input and
output currents
3
and 2 = -2 PPLm - ldsis, 1 (1 1)
Figures 9(a)-9(c) demonstrate the power control operation of
the DFIG. In this, the P, and Qs controllers (PI type) are
The matrix converter is modelled using ideal switches. The used to instruct the rotor current controllers iir and iJr
speed of the DFIG is controlled by a PI controller which
respectively. The sampling time for the power PI controllers
provides the rotor current command l y r * while idr* is was set to 5ms while the current controller setting was
determined by the reactive current requirement.
5 0 0 p . The command P,' is derived from the reference
power-speed curve. The DFIG is driven by the turbine
The sampling time for the current control loop and the
operated under the time-varying wind-speed. For a given
switching period of the converter are both set to 500ps while
wind-speed variation, the generator shaft speed response is
the speed loop sampling time is 5 ms. Results shown in Fig.5
plotted in Fig.g(a). The power and reactive power of the
were obtained when the current command iqLvis increased to
machine stator are shown together with the rotor power in
6.5A and subsequently decreased while idr is set to zero.
Fig.9@). The grid power and reactive power representing the
Waveforms of the rotor currents i4er and I $ are shown in resultant output power of the VSG system are plotted in
Fig.5(a). Correspondingly, the rotor current i, and the shaft Fig.9(c). As can be seen, the waveform of the grid power
speed are shown in Fig.5@). In this case, the shaft speed is contains a series of the steady-state operating points defined

909
in the ideal turbine characteristics. It can be shown that the
optimal output power derived from the reference power
curve is achieved for any wind speed.

V. CONCLUSION
A variable-speed wind-power generator using a DFIG in
conjunction with a matrix converter is proposed. Stable
operation of the DFIG was achieved by means of stator-flux
oriented control technique. The operational principle of the
proposed wind-power generator and the validity of the
control scheme were illustrated by the steady-state and
transient responses of the power and currents associated with ,
A A A A
the machine stator and rotor. Simulation results demonstrate 6r sped 2 940

that the proposed wind-power generator is feasible and has

certain advantages.

VI. APPENDIX
. U. P,‘
Doubly-fed induction machine parameters: 3 760
7.5 kW, 415V, 6 poles, rated speed 970 rpm, N,/N,= 1.7, , \

0 02 04 06 08 1 12 14 16 18 2(SeC)
q=l.O20, K=0.8ln, L3=0.093H, Lr=0.081H, Lm=0.0664H
(all referred to rotor). (b) rotor currents i, and shaft speed
_____. - 04
VI. REFERENCES J 300
[ 11 R. Jones and G. A. Smith, “High quality mains power r’ 200
from variable-speed wind turbines,” Renewable Energy, ,. 1M)

IEE Conf. Pub. No.385, November 1993, pp.202-206. /: 0

-100
[2] R Spee, S. Bhowmik and J. HR.Enslin, “Adaptive
control strategies for variable-speed doubly-fed wind
1 -200
’ .-?a0
power generation system,” IEEE IAS Annual Meeting ~~~

Conf, 1994, pp.545-552. 08 0.85 09 095 (sec)

[3] R Pena, J. C. Clare and G.M. Asher, “Doubly-fed (c) converter input current I , , and voltage vu,
induction generator using back-to-back PWM converters (kW) - ~ ~ - -
and its application to variable-speed wind-energy
30 -?-..-^-XI-

generation,” IEE Proc. Part B, May 1996, pp.231-241. I Pc

2 5
[4] M. G. B. Venturim dnd A. Alesina, “Analysis and
Design of Optimum-Amplitude Nine-Switch Direct AC- 20‘

AC Converters,” IEEE Trans on Power Electronics, 1 5 -

__cc

1989. I - \
0 05 1 1 5 2 (sec)
[5] L. Huber, D. Borojevic, N. Burany : “Analysis, Design
and Implementation of the Space-Vector Modulator for (d) stator active power
Force-Commutated Cycloconverters,” IEE Proc. Part B, Figure 5 Step responses to current command zgrs
Vo1.139, No.2, March 1992. in subsynchronous mode
[6] L. Zhang, C. Watthanasarn and W. Shepherd, “Analysisand
Implementationof a Space Vector Modulation Algorithm for
Direct AC-AC Matrix Converters”,European Power
Electronics Journal, Vol. 6, No. 1, May 1996, pp 7-15.
[7] W. Shepherd, L. N. Hulley, D. T. W. Liang, Power
Electronics andMOtor Control (2nd edition), Cambridge
University Press, Cambridge, England, 1995.
[8] P.Vas, Vector Control ofAC Machines, Oxford Science
Publishng, 1990.
ow
I

I
I__d

-1
1

0 0.5 1 15 (sec)
[9] C. Watthanasarn, “Optimal Control and Applications of
AC-AC Matrix Converters,” PhD Thesis, 1997, (a) synchronous rotor currents ziTand i;,.
University of Bradford, UK.
 (kW/kVAR) (rpm1
3 p,
2- 1_1_1$1200
1 r speed 3 1000
I
01 i 800
-1 ' y"-- 4 600
-2
I
1 i Q, Ikh..--.--.- i 400
I 1
-3
--------
-4 0
' 200

05 1 15 2 \sec)

0 0.5 1 1.5 2 (sec) (c) stator active power Ps,

reactive power Qs and shaft speed
(b) rotor current i, and shaft speed Figure 7 Step responses to the rotor current command idr
_______ 0

._________
1.I

(kW) 2 5 j-
1.15 1.2
-.
1.25
(c) converter input current I , , and voltage vas
--'
1.3
(sec) -10
0
- 1 2

Figure 8 Rotor current I , , during shaft acceleration

through synchronous speed
3
__
4 5
.a
i

__j

20' 7pr i
wind SDeed ( m k )
15

10'
5.
0 5 ; '.-.-..-.---.-.-
L- ~~ 0 -L
. .___ -
0 0 5 1 1 5 (sec) 0 1 2 3 4 5
time (sec)
(d) stator active power P, shaft speed (rpm)
Figure 6 Step responses to current command i q r
1500
/
/----,
in supersynchronous mode 1000 -____ \
------ -

(a) wind speed variation and shaft response

(kW/kVAR!
60
5 0 ' p, n,
34 00 ; /I
(a) synchronous rotor currents ?;z and i:r Q S

'OO 1 2 3 4 5 6 tlme(sec,

(b) stator and rotor power

(kWlkVAR)

:: 9 '\

-L
3 0
2 0 1
10; Qnet
-10 L--
-___J (sec) or
0 0.5 1.o 1.5
-'O 0 1 2 3 4 5 6 time ( s e c )
(b) rotor currents i,
(c) resultant output power to utility grid
Figure 9 Optimal power tracking operation of a DFIG