Content uploaded by Nicholas A Vovos
Author content
All content in this area was uploaded by Nicholas A Vovos on Oct 03, 2013
Content may be subject to copyright.
An Autonomous System Supplied Only by a
Pitch-Controlled Variable-Speed Wind Turbine
N. A. Schinas, N. A. Vovos, Senior Member, IEEE, and G. B. Giannakopoulos, Senior Member, IEEE
Abstract—This paper presents a control strategy for a variable-
speed pitch-controlled wind turbine (WT) generation scheme for
the supply of an autonomous system with no energy storage units.
The synchronous generator includes two three-phase stator wind-
ings displaced by 30◦that are connected to the transformer load
through two dc links with voltage source inverters (VSI). Follow-
ing priority rules, the load is divided into steps. Each load step can
be supplied by the WT when the wind speed varies between two
predefined speed levels. The first goal of the WT control system is
to supply the load with constant real power under constant voltage
as the wind speed varies between two levels and the second is to op-
erate smoothly interchanging the load steps when the wind speed
breaks through a speed level. There are two controllers: the in-
verter controller that keeps the load voltage constant and the pitch
controller acting on the blade’s angle. Using simulation techniques,
the operation of the WT system and the efficiency of the proposed
control strategy are demonstrated for a wide range of wind speeds.
Index Terms—Converters, synchronous generators (SGs),
variable-speed drives, wind power generation.
I. INTRODUCTION
THE dream of a variable-speed wind turbine (WT) tied
to the electrical grid began to become a viable reality in
the early to mid-1970s. Large wind turbines went online in
the United States and Europe using several different methods
for transforming variable-voltage, variable-frequency outputs to
reliable constant-voltage, constant-frequency outputs. This was
commonly achieved through an ac/dc/ac system. Variable speed
allows the turbine to absorb maximum power from the wind,
but also improves its dynamic behavior, thereby alleviating the
stresses on the mechanical construction. In this way the tur-
bine can be made lighter and therefore cheaper. Furthermore,
the presence of the pitch control can extra-regulate the power
captured from the wind.
In this paper a specific WT generation scheme is proposed
that is able to supply a mixed load (passive and motors) without
the presence of any other source of electric power (autonomous
system). A control system is designed such that the voltage and
the frequency of the load may be preserved in acceptable values
with severe changes of wind speed and with no serious impact
on the power quality absorbed by the load. To increase reliability
and reduce cost and maintenance requirements, no energy stor-
age units or pseudo-loads are used in the proposed autonomous
system. These characteristics are attractive, especially for au-
tonomous systems operating far away from technical support.
Manuscript received February 9, 2005; revised June 13, 2005. Paper no.
TEC-00044-2005.
The authors are with the Department of Electrical and Computer Engi-
neering, University of Patras, Rion, Greece (e-mail: n schinas@hotmail.com;
N.A.Vovos@ee.upatras.gr; G.B.Giannakopoulos@ee.upatras.gr).
Digital Object Identifier 10.1109/TEC.2006.859971
Fig. 1. General configuration of the system under study.
Fig. 2. Cp versus tip speed ratio for pitch angle 0◦,5
◦, and 10◦.
II. WIND TURBINE DESCRIPTION
The WT generation scheme is composed of a WT equipped
with a pitch control for the blades to control the aerodynamic
power and a variable-speed synchronous generator (SG). The
WT is coupled directly (gearless) to the SG, which is made up
of two three-phase stator windings displaced by 30◦, in order to
minimize the torque distortions due to variable speed operation.
Each stator winding is connected to the transformer supplying
the load through a DC link with voltage source inverter (VSI).
Main parameters of the wind turbine are given in the Appendix.
Fig. 1 shows the specific WT generation scheme proposed in
this paper.
III. OPERATING FUNCTIONS OF THE WT SYSTEM
The wind turbine is characterized by its Cp-λcurves, where λ
(tip speed ratio) is the ratio between the linear speed of the tip of
the blade with respect to the wind speed (vw)and Cpis the power
coefficient of the WT [1], [5]. It is known that for a specific wind
speed and pitch angle (β), the WT generates maximum electric
power in a specific mechanical angular frequency ωm, where
Cpis maximum. The power coefficient of the machine Cp(and
so the electric power) versus tip speed ratio (λ)for various pitch
angles is shown in Fig. 2 for a fixed wind speed. Thus, the pitch
angle can regulate the generating WT power for a fixed ωm
and vw.
0885-8969/$25.00 © 2006 IEEE
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
Fig. 3. Pversus vwfor various ωmand β.
Fig. 4. Electrical layout of the dc link.
In Fig. 3 we can see the power output of the WT versus the
wind speed for (a) ωm=4r/s and β=0
◦,(b)ωm=4.5r/s and
β=2
◦, and (c) ωm=5r/s and β=3
◦. It is obvious that as the
wind speed increases, the power can be changed or kept at the
same value (e.g., 140 kW) for suitable values of the mechanical
angular frequency and the pitch angle. As a result, according to
the wind speed the pitch angle can be adjusted in such a way that
the mechanical angular frequency of the WT is regulated suit-
ably so that the desired amount of power will be captured from
the wind and finally be delivered to the load. This type of control
also ensures that the rotor speed lies between its rated limits.
Fig. 4 depicts the general layout of the DC link for each
of the two three-phase stator windings. The load is supplied
with the voltage Vsand the terminal stator voltage of the WT
is V1. The variable-frequency ac current of the generator is
transformed into dc through the rectifier and then back into
constant-frequency ac through the inverter for the supply of the
load.
IV. GENERAL CONTROL SYSTEM DESCRIPTION
The main objective of this article is to develop a suitable
control method by which an autonomous electric system can be
supplied by the above-mentioned WT system as it is depicted in
Fig. 1.
In such a case, our first priority is to keep the power output of
the WT constant and equal to the load defined by a load shedding
system, according to the available wind speed. To achieve this
under various wind speeds the control capabilities of the pitch
control system must be used in cooperation with the control of
the ac/dc/ac system.
So the first criterion in designing a system of this kind is
that depending on the wind capabilities of the region where it is
installed, the rated power of the WT must exceed the maximum
load supplied to satisfy the load demand for a longer period of
time. The real power produced by the WT depends on the wind
Fig. 5. Load shedding with three load steps.
speed at the height of the hub, which means that it varies over
a range of values according to its power curve. For a system
of this kind to be practical, we assume that it is installed in a
place where the wind speed is above a minimum value most of
the time and the kind of load is such that it can be supplied in
several steps with no batteries or pseudo-loads.
The steps are predecided using priority rules. Each load step
defines a minimum wind speed, vws, above which the WT is
capable of supplying the specific load and corresponds to the
maximum power the WT can produce under vws. The load
shedding system monitors the wind speed and if it is above one
of the defined values vws for a specific period of time, it switches
on the appropriate load. Then the WT, through the pitch control
system, regulates the supplied electric power to match the new
load. Therefore, each load step corresponds to a value of real
power that the WT is able to produce for a specific wind speed
(vws), a specific mechanical frequency (ωms )for optimum wind
power capture, and a small value of the pitch angle (βs).Asthe
wind speed increases beyond vws, the control of the WT must be
able to hold the produced real power to its previous value using
the pitch control system; otherwise, the voltage of the load will
increase beyond its rated value, which is unacceptable. In this
region of operation, the machine works at a power coefficient
(Cp)smaller than its maximum value (Fig. 2), so that the right
amount of real power is produced.
When the wind speed decreases, the WT power must again
be adjusted to the load up to the point at which it becomes equal
with vws. If the wind speed decreases more, the load cannot be
supplied and the load shedding system changes the load to the
lower load step.
The above concept of control is depicted in Fig. 5. The first
load step power is considered to be 120 kW. The WT can produce
this power for vw1 =7m/s, ω1= 100 rad/s, and β1≈0◦.The
actual WT power output must remain the same irrespective of
the wind speed (for vw≥vw1). This is achieved through the
inverter and the pitch angle control as follows:
The inverter control keeps the load voltage constant as the
wind speed deviates from vw1 and this results in constant real
power supplied to the load.
This is achieved through the fast adjustment of the modu-
lation index (m)of the pulsation of the inverter devices. The
modulation index can be changed of course within some lim-
its. If the wind speed changes steadily too much, the inverter
control can no longer keep the load voltage constant as there
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
is a significant imbalance between the power captured from the
wind and the power being absorbed by the load. In this case, the
pitch control system adjusts the pitch angle so that the power
produced by the WT almost equals the power being absorbed
by the load. When this is true, the voltage V1of the SG, the dc
voltage Vdc, and the load voltage Vshave their nominal values.
In the transient period, this type of control ensures that the rotor
speed is inside its range of operation.
There is no need to specify exactly the mechanical speed
of the machine because we do not wish to capture maximum
power from the wind for every vw. The mechanical speed of
the machine is the result of the wind speed and the value of
the pitch angle, as from Fig. 3 there is a right combination of
ωmand βwith every vwfor the machine to produce the desired
amount of power. According to the analysis of the previous
paragraph, the action of the pitch actuator under high wind-
speed values ensures that the mechanical speed of the machine is
restrained between its rated limits: 1) the available wind power
is better exploited and 2) a wider range and different kinds
of loads can be supplied according to the time percentage of
the necessary wind speed for each level. For instance, public
lighting should be in the first level and entertainment devices in
the second.
The economical advantages due to the absence of any energy
storage devices are obvious. Although there might be better
performance of the system operation or better wind power ex-
ploitation if such devices were present, the goal of this paper is
to develop a system that needs no extra units for the load to be
supplied properly.
A. Low-Level Control System
1) General Description: The control system consists of two
control loops:
1) An inverter controller that uses a pulsation scheme based
on the PWM technique. The output of the inverter is a con-
stant frequency ac voltage whose amplitude is regulated by
acting on the modulation index (m)of the pulsation pro-
cess. The mchanges according to the difference between
the reference value of the rms load voltage (Vs,ref)and
the real value (Vs). Any wind speed fluctuations cause
alterations of the dc voltage across the capacitor (Vdc).
The fast electronic action of the inverter control, com-
pared to the slow electromechanical pitch actuator, keeps
the load voltage almost constant under small wind speed
fluctuations.
2) A pitch controller that acts on the pitch blades’ angle ac-
cording to the difference (Vdc,er)between the reference
value of the dc voltage (Vdc,ref)and the real value (Vdc )
(Fig. 4). Any imbalance between the WT generating out-
put power and the power absorbed by the load alters the
Vdc from its nominal value. The task of the slow pitch
actuator is to keep Vdc constant under large alterations of
the wind speed by eliminating this imbalance and so the
inverter control does not reach its limits.
If, for example, the wind speed starts to increase steadily
while the load remains constant, the dc voltage will also start to
Fig. 6. (a) Block diagram of the speed controller. (b) Block diagram of the
pitch controller.
Fig. 7. Membership functions of the input.
increase. The pitch controller will change the pitch angle suit-
ably so the energy from the wind will be limited to its previous
value as the machine works with a new frequency and the Vdc
will be decreased again to its nominal value. At the same time
small variations on the ac voltage Vs(Fig. 4) are regulated by
the inverter controller.
Fig. 6 shows the general block diagrams of the two controllers.
2) Design of the Pitch Controller Based on Fuzzy Logic:
The design of the pitch controller was based on the principles
of fuzzy logic with a minimum of rules. From the level of Vdc,
the controller must determine the appropriate pitch angle rate
of change (y). The concept is simple: if Vdc is much smaller
than Vdc,ref (Vdc,er positive) then more power must be delivered
to the load for the particular wind speed, which means that the
pitch angle should be decreased (ynegative). On the contrary, if
Vdc is much greater than Vdc,ref (Vdc,er negative) then the angle
must be increased for the energy from the wind to be limited (y
positive). As the wind speed is never constant, small fluctuations
must be ignored by the control system, so when Vdc is either a
little less than or a little more than Vdc,ref (Vdc,er zero) then the
angle remains the same (yzero). Fig. 7 depicts the membership
functions of the input Vdc,er of the controller and Fig. 8 the
functions of the output y.
There are only three rules of the controller and are shown in
Fig. 9.
Finally, the block diagram of the whole pitch controller is
shown in Fig. 10.
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
Fig. 8. Membership functions of the output.
Fig. 9. Rules of the fuzzy controller.
Fig. 10. Block diagram of the pitch controller.
One crucial point in judging the behavior of the pitch con-
troller in a real system is the maximum rate by which the pitch
angle can change, according to the capability of the pitch motor.
This is the reason a rate limiter of ±10 deg/s has been added in
the simulation program.
3) The Inverter Controller: The inverter controller is a sim-
ple PI controller with the following parameters:
KP=0.4,K
I= 500.
B. High-Level Control System Description
The previous control loops are driven by a higher level con-
trol system. This system has one input, the measured wind speed
(vw), which is then filtered (vwa)so the actual average alter-
ations of the wind speed will be used. Fig. 11 depicts the general
flowchart of the proposed high-level control system.
The software of this system also includes the load shedding
of our autonomous system, which is considered to have many
load steps. Generally speaking, the supply of a load step n
corresponds to a minimum wind speed vwn.Ifvwa overcomes
vw(n+1) for a time period tpl, the new load step n+1is inserted.
Otherwise, if vwa becomes smaller than vwn then a lower load
step n−1is supplied.
V. S IMULATION PROCE SS
The proposed autonomous system has been studied for var-
ious disturbances using the SIMULINK simulation program.
The simulation of the synchronous generator includes a de-
tailed model based on Park’s equations in thed-qplane of all the
quantities of the machine [1], [3], [4]. The inverters are made
Fig. 11. Flowchart of the high-level control system.
up of insulated gate bipolar transistor (IGBT) semiconductor
components. The pulsation scheme is based on the pulsewidth
modulation techniques. The wind is simulated as a random sig-
nal that varies around a specific mean value. The simulation
model of the WT is based on the following nonlinear equation
that expresses the relationship between the power coefficient
(CP), the tip speed ratio (λ), and the blades’ angle (β)[2], [5]:
CP=(0.44 −0.0167β) sin π(λ−3)
15 −0.3β−0.00184(λ−3)β
The tip speed ratio λis given by [2]:
λ=ΩR
VW
where Ris the WT rotor radius, Ωis the mechanical angular
frequency of the WT, and vwis the wind speed. The output
mechanical torque Tmis calculated by [2]
Tm=1
2ρARCP
V2
W
λ
where ρis the air density and Ais the swept area of the blades.
The response of the WT is studied when it is connected to an
autonomous load containing passive components and motors.
The load has been divided into two steps. The load in the first
step consists of a passive component with a rated real power of
70 kW and a 30 kVA motor both at a rated voltage of 400 V. In
the second level there is an extra passive load of 250 kW and
35 kVar. The inputs of the control system are the wind speed
and the references of the ac rated voltage of the load and the dc
link voltage.
VI. SIMULATION RESULTS
In the first case we test the system under a fast increase of
the wind speed from 7 m/s to 10 m/s at t=8s and back to
about 7 m/s at t=15 s. Fig. 12(a) shows the waveform of
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
Fig. 12. System response for step wind speed changes with constant load.
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
Fig. 13. System response for a wind speed increase that activates the second load step.
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
the wind for the simulation time. The value for vw2that has
been defined for the supply of the second load step is 10.5 m/s
and thus the high-level controller decides that the system will
continue to supply only the first load step. Due to the increase of
the vw, the WT rotor mechanical power increases [Fig. 12(b)]
and so do the angular frequency [Fig. 12(c)] and the voltage Vdc
[Fig. 12(d)]. The inverter controller almost simultaneously alters
the pulsation [Fig. 12(e)], the dc current idc1 from the rectifier is
decreased [Fig. 12(f)], and so the power to the load is constant
[Fig. 12(g)]. As the pitch controller slowly (much less than the
rate limit of ±10 deg/s) alters the pitch angle [Fig. 12(h)] the WT
mechanical rotor power is regulated to its previous value under
the new vw, with a new frequency due to the initial extra power
from the wind and the dc voltage starts to return to its nominal
value. Then the inverter controller alters again the modulation
index mand the idc1 returns to its previous value. The rms load
voltage has small variations [Fig. 12(i)]. The system operates
similarly in the decrease of the wind speed. The result reveals
that the power delivered to the load and its voltage are kept
almost constant despite the large variations of the wind speed.
In the second case, we test the system for a greater change in
the wind speed, which permits the next load step to be activated.
So, at t=8s the wind speed is again increased from 7 m/s to
11.5 m/s [Fig. 13(a)] and the second load step is inserted when
vwbecomes greater than 10.5 m/s. At t=15s the wind speed is
decreased at about 8 m/s. As before, when vwbecomes less than
10.5 m/s, the second load step is not supplied with power. The
mechanical rotor power initially increases [Fig. 13(b)], as do the
angular frequency [Fig. 13(c)] and the voltage Vdc [Fig. 13(d)].
The modulation index alters suitably [Fig. 13(e)] as the extra
load decreases the voltage, which means that the idc1 must
increase [Fig. 13(f)] in order for the load power to be increased
[Fig. 13(g)]. The pitch angle regulates the mechanical power to
the WT rotor and is shown in Fig. 13(h). The rms voltage at the
load is not heavily disturbed by the introduction of the second
level of the load [Fig. 13(i)].
VII. CONCLUSION
A variable-speed, pitch-controlled WT and its control system
have been proposed for the supply of an autonomous system
with no energy storage units. The system has been tested by
simulation with analytical models. The philosophy of the con-
trol strategy, which is based on a hierarchical architecture, is de-
scribed. The proposed scheme includes a pitch controller based
on fuzzy logic, which acts according to the level of the wind
speed and the load being supplied. Simulation results show good
performance of the proposed system even under large variations
of the wind speed. The active power and the rotational speed
of the WT are adjusted in a smooth way. The distortions of the
load voltage are within acceptable limits.
APPENDIX
Wind Turbine: Pn= 500 kW, R=21 m, Cpmax =0.44,
λopt =10.5rad.
Synchronous Generator: Sn= 500 kVA, Vn= 400 V, p=
60,R
s=2.9mΩ,Rkq =2.24 mΩ,Rkd =44.8mΩ,Rfd =
0.8mΩ,Xls =4.6mΩ,Xlm =40.65 mΩ,Xlkq =8.96 mΩ,
Xlkd =14.75 mΩ,Xlfd =19.61 mΩ,Xmd = 600 mΩ,Xmq =
61.22 mΩ.
REFERENCES
[1] N. Schinas, N. A. Vovos, and G. B. Giannakopoulos, “Embedded wind
turbine generation to weak distribution AC grid,” Wind Eng. ,vol. 26, no. 6,
pp. 397–409, Nov. 2002.
[2] E. S. Abdin and W. Xu, “Control design and dynamic performance anal-
ysis of a wind turbine-induction generation unit,” IEEE Trans. Energy
Conversion, vol. 15, no. 1, pp. 91–96, Mar. 2000.
[3] S. D. Sudhoff, “Analysis and average-value modelling of dual line-
commutated converter: 6-phase synchronous machine systems,” IEEE
Trans. Energy Conversion, vol. 8, no. 3, pp. 411–417, Sep. 1993.
[4] P. C. Krause, Analysis of Electric Machinery. New York: McGraw Hill,
1986.
[5] E. Muljadi and C. P. Butterfield, “Pitch-controlled variable-speed wind
turbine generation,” IEEE Trans. Ind. Appl., vol. 37, no. 1, pp. 240–246,
Jan./Feb. 2002.
N. A. Schinas was born in Athens, Greece, in 1972.
He received the diploma in electrical engineering
from the University of Patras, Rion, Greece, in 1996.
As an electrical engineer he has worked with in-
dustry automation and medium-voltage installations,
as well as the development and the construction of
wind farms in Western Greece. He is currently a
postgraduate student in the Electrical and Computer
Engineering Department at the University of Patras
and his main fields of interest are the transient stabil-
ity study of integrated ac/dc systems, power quality,
wind generation, and renewable energy sources generally.
N. A. Vovos (M’76–SM’95) was born in Thessa-
loniki, Greece, in 1951. He received the diploma and
Ph.D. degree from the University of Patras, Rion,
Greece, and the M.Sc. degree from the University
of Manchester Institute of Science and Technology
(UMIST), Manchester, UK, in 1974, 1978, and 1975,
respectively.
He is Professor and Head of the Electrical and
Computer Engineering Department of the Univer-
sity of Patras and his main fields of interest are the
transient stability study of integrated ac/dc systems,
FACTS, power quality, and renewable energy sources.
G. B. Giannakopoulos (M’95–SM’96) was born in
Volos, Greece, in 1950. He received the diploma and
Ph.D. degree in electrical engineering from the Uni-
versity of Patras, Rion, Greece, in 1975 and 1978,
respectively.
He is currently Professor in the Electrical and
Computer Engineering Department at the University
of Patras and his main fields of interest are HVDC
transmission, computer techniques in power system
analysis, FACTS, power quality, and renewable en-
ergy sources.
Authorized licensed use limited to: King Mongkuts Institute of Technology Thornburi. Downloaded on October 16, 2008 at 00:40 from IEEE Xplore. Restrictions apply.
