You are on page 1of 6

Paper accepted for presentation at 2003 IEEE Bologna PowerTech Conference, June 23-26, Bologna, Italy

Aggregated Modelling of Wind Parks in Plower


System Dynamics Simulations
J.G. Slootweg, Member, IEEE, W.L. Kling, Member, IEEE

- Increasing numbers of wind turbines are connected to power plants as would have beeio the case: without any wind
electrical power systems, in order to reduce the adverse turbines present. This is possible, as long as wind power
environmental impact of conventional electrical power penetration is still low. However, a tendency to increase the
generation. A tendency can be observed to erect these turbines in amount of electricity generated &om wind can be observed.
wind parks, connected to the high voltage transmission grid. Therefore, the penetration of wind turbines in electrical power
These parks effect the dynamic behaviour of power systems,
because in wind turbines generator types that are different from systems will increase and they may begin lo influence overall
power system behavior. In that case, it will no longer be
the conventional synchronous generator are used.
To investigate the impact of a wind park on the dynamics of the possible to run a power system by only controlling large scale
power system to which it is connected, an adequate model is power plants. It is therefore important to study the behavior of
required. In order to avoid the necessity of developing a detailed wind turbines in an electrical power system and their
model of a wind park with tens or hundreds of wind turbines and interaction with other generation equipment and with loads.
their interconnectionsand to calculate the wind speed signal for Further, a tendency to concentrate turbines in wind farms can
each individual turbine, aggregated wind park models are needed. be observed in order to use regions with a good wind resource
In the paper, aggregated models for wind parks equipped with
either constant or variable speed wind turbines are presented. It efficiently and to concentrate the visual impact of modem
is shown that results obtained with an aggregated model and with wind turbines, that can easily reach heights of more than 100
a detailed model show a high degree of correspondence, both for m, at certain locations. These wirtd farms are connected to the
high voltage transmission grid and therefore directly influence
normal operation and for disturbances.
the dynamic behavior of an electrical power system. This
Index Terms - wind power, wind park, aggregated modelling, increases the need for adequate models.
simulation,power system dynamics
In this paper, aggregated models of winld parks with both
constant and variable speed wind turbines are presented. The
I. INTRODUCTION
advantage of an aggregated model' is that it eliminates the need
to develop a detailed model of' a wind park with tens or
As a result of increasing environmental concern, the impact of hundreds of wind turbines and their interconnections, and to
conventional electricity generation on the environment is being calculate the wind speed signal 6or each individual turbine in
minimized and efforts are made to generate electricity from advance. As will be shown, the response of the aggregated and
renewable sources. The main advantages of electricity the detailed models shows a high degree of similarity, both
generation from renewable sources are the absence of harmful during normal operation as well as during disturbances.
emissions and the in principle infinite availability of the prime The topic of aggregated wind park modelling has been treated
mover that is converted into electricity. One way of generating earlier, both by the authors of this paper, as well as by other
electricity fiom renewable sources is to use wind turbines that authors [l-41. Nevertheless, it has not been treated extensively
convert the energy contained in flowing air into electricity. Up in the existing literature, because SO far:
to this moment, the amount of wind power integrated into large
For wind parks with variable speed wind turbines, only the
scale electrical power systems only covers a small part of the
response to wind speed changes been investigated [ 11.
total power system load. The rest of the load is for the largest
For constant speed wind turbines, only the fault response
part covered by conventional thermal, nuclear and hydro
has been investigated [2-41.
power plants.
In this paper, the response of detailed and aggregated models
Wind turbines hardly ever take part in voltage and fiequency of wind parks with constant and variable speed wind turbines
control and if a disturbance occurs, the wind turbines are to both wind speed changes and faults will be compared. This
disconnected and reconnected when normal operation has been is important, because presently it is not clear whether
resumed. Thus, notwithstandingthe presence of wind turbines, aggregated models adequately represent a wind park both
frequency and voltage are maintained by controlling the other under normal operating condition3 and during disturbances, as
is required [ 5 ] . Investigations of the response of wind parks
The financial support fiom the Dutch Organization for Scientific Research with constant speed wind turbine!; to wind speed changes and
P O ) is greatly acknowledged.
The authors are with the Faculty of Information Technology and Systems, the fault response of wind parks with variable speed wind
Delft University of Technology, P.O. Box 5031, 2600 GA, DELFT, The turbines have not been carried out yet. These will be carried
Netherlands.
out in this paper, which therefore makes an important
Corresponding author: J.G. Slootweg
contribution to the topic of aggregated wind park modelling.

Abstract

E-mail:j.g.slootweg@its.tudelft.nl
Tel : +31-15-27 86219
Fax: +31-15-27 81182

0-7803-7967-5/03/$17.0002003 IEEE

11. WINDT u m m TWES

Three different wind turbine types are currently widely


applied. The first is the directly grid coupled squirrel cage
induction generator, used in constant speed wind turbines. The
wind turbine rotor is coupled to the generator through a
gearbox. In most constant speed wind turbines, the power
extracted fiom the wind is limited using the stall effect during
high wind speeds. This means that the rotor is designed in such
a way that its aerodynamic efficiency decreases in high wind
speeds, thus preventing extraction of too much mechanical
power fiom the wind. When the stall effect is used, no active
control systems are necessary. Pitch controlled constant speed
wind turbines have, however, also been built.
The second type is the doubly fed (wound rotor) induction
generator, which allows variable speed operation. The rotor
winding is fed using a back-to-back voltage source converter.
Like in the first type, the wind turbine rotor is coupled to the
generator .through a gearbox. In high wind speeds, the power
extracted fiom the wind is limited by pitching the rotor blades.
The third type is a direct drive synchronous generator, also
allowing variable speed operation. The synchronous generator
can have a wound rotor or be excited using permanent
magnets. It is grid coupled through a back-to-back voltage
source converter or a diode rectifier and voltage source
converter. The synchronous generator is a low speed multi
pole generator, therefore no gearbox is needed. Like in the
second type, the power extracted fiom the wind is limited by
pitching the rotor blades in high wind speeds. The three wind
turbine types are depicted in figure 1. For a more elaborate
description of the various wind turbine types, the reader is
referred to text books [6].

111. AGGREGATED
MODELLING
OF WIND PARKS

A. Power System Dynamics Simulations


The models presented in this paper are meant to be used for
power system simulations, also referred to as fundamental
frequency simulations or electromechanical transient
simulations. This type of simulations can be applied to study
phenomena with fiequencies of about 0.1 to 10 Hz. The typical
problems investigated using power system dynamics
simulation software are voltage and rotor angle stability.
In order to be able to quickly cany out the simulation of
phenomena in the band width of interest, it should be possible
to use a relatively long simulation time step. Therefore, high
fiequency phenomena are neglected in power system dynamics
simulations by using a load flow model for the network and by
neglecting short time constants in the generator and controller
models by assuming that the new steady state is reached
immediately. This approach not only allows the use of a longer
simulation time step, but also reduces the number of
differential equations, because there are no longer differential
equations associated with the network, whereas the number of
differential equations associated with the generators and the
controllers is reduced [7].
In section IILB, the modelling of the individual wind turbines
will be discussed, whereas in section 1II.C the development of
an aggregated park model from the model of an individual
wind turbine is described.
B. Wind Turbine Modelling

1) Constant Speed Wind Turbine Model


A model of a constant speed wind turbine for power system

Squirrel
cage

dynamics simulations consists of a rotor model, a shaft model


and a squirrel cage induction generator model. Here, the
following set of equations is used to describe the rotor of a
constant speed wind turbine

indUct Ion
aeneralor

',P

PCp(A)A,Vw 3

Compensating
capacitors

Rotor

Doublv fed

7-J16.5

cp(31.)=0.44[

ai=

6.94)

-+ 0.002

31.

in which P, is the power extracted fiom the wind [W], p is the


air density [kg/m3],cp is the performance coefficient, A, is the
area swept by the rotor [m'], v, is the wind speed [ d s ] and I
the tip speed ratio, equal to the rotor blade tip speed v, [ d s ]
divided by the wind speed v,. Before the wind speed is
inserted in equation (l), it passes through a low pass filter in
order to approximate the smoothing effect of the large rotor
surface for wind speed components with a high frequency.
The shaft is described by the following set of equations, in
which f is the nominal grid frequency [Hz], T is torque [P.u.],
Figure 1. Schematic representation of frequently occurring wind turbine y is the angular displacement between the two ends of the shaft
types: constant speed wind turbine, variable speed wind turbine with doubly [electricalradians], o is rotational speed [P.u.], H is the inertia
fed (wound rotor) induction generator,variable speed wind turbine with direct constant [SI and K, is the shaft stifhess [P.u. torque/electrical
drive synchronous generator.

radians]. The indices wr, m and e mean wind turbine rotor,


generator mechanical and generator electrical respectively.

-daw - T m -

dt

*=

dt

Ksy

2H,

2nf(owr- om)

The generator model in the constant speed wind turbine


concept is a standard induction machine model in which the
dJl/dt terms in the stator equations are neglected in order to be
comply with the assumptions on which power system dynamics
simulations are based. The model's equations can be found in
the literature [7].
2) VariableSpeed Wind Turbine Model
A variable speed wind turbine is a much more complicated
system than a constant speed wind turbine. First, the rotor
model is different, because the performance coefficient
depends not only on the tip speed ratio A, but also on the pitch
angle 8. In nearly all variable speed wind turbines, the power
extracted fiom the wind in high wind speeds is limited by
reducing the aerodynamic efficiency of the rotor through
pitching the blades. Second, a rotor speed controller is needed,
because the variable speed capability leads to the necessity to
control the rotor speed. Third, a pitch angle controller must be
incorporated in the model. To conclude, variable speed wind
turbines are in principle capable of taking part in grid voltage
control because they can vary reactive power output.
Detailed models of the two variable speed wind turbine types
that match the assumptions applied in power system dynamics
simulation have been presented by the authors in earlier papers
[SI,[9]. It has also been shown that it is possible to model both
variable speed wind turbine types with one general variable
speed wind turbine model in power system dynamics
simulations [lo]. Although the two concepts are fimdamentally
different, the differences are mainly related to the different
generator concepts and controller algorithms of the power
electronic converter. These, however, for the largest disappear
when the power electronic converters are modelled as
fhdamental fiequency current sources and are hence hardly
reflected in the frequency bandwidth that is of interest in
power system dynamics simulations. It is therefore possible to
use the general variable speed wind turbine model depicted in

U
Pitch

vodage

angle
i

;et pcini

Figure 2. General variable speed wind turbine model [l], [lo]

T.rmlnal

C. Model Aggregation
I ) Aggregated Wind Speed Modelling
The wind speed can be consideIed to consist of four terms,
namely an average value, a ramp calmponent, a gust
component and turbulence [ll], [12]. In deriving the wind
speed signal for the aggregated wnd park model, it is assumed
that the wind speed can be split up in a filly deterministic and
a fully stochastic part. The stochastic pert consists of the
turbulence. In the aggregated park model, this term is
neglected, because in a wind park the effect of turbulence on
the aggregated output power is reduced due to the smoothing
effect of the large number of wind turbines, as can be
concluded fiom measurements carried oul at existing wind
parks [ 131.
The fully deterministic part consists of the average value and,
if present, the gust and ramp component. 'The average value
can be assumed to be the same tlilroughout the park. The gust
and ramp components travel throiugh the park and the time at
which they arrive at the individual turbines depends on the
average wind speed, the angle of attack and the wind park
layout. The start and stop times (of the gust and the ramp at
each individual wind turbine cart thus be calculated fiom a
single wind speed signal applied to the aggregated wind park
model as a whole, taking into account the wind direction and
the park layout. The wind speed signal is specified by the start
and stop times of the gust and the ramp relative to the centre of
the wind park and the wind direction.
2) Aggregation of Constant Speed Wind Turbines

The way in which constant speed wind turbines can be


aggregated has been discussed in the literature [2]-[4]. In this
paper, it is assumed that the wind park can be represented with
one single constant speed wind turbine. The: characteristics of
this wind turbine can be calculated using the following
equations
n

seq= csi ceq


=
ci
i= 1

i= 1

Pm,eq= ZPm,a

(3)

i= 1

in which S is the MVA rating and C the size of the


compensating capacitor. The index eq means aggregated
equivalent wind turbine, m means mechanical and i indicates
the individual wind turbines in the wind park.
A specific problem in the aggregalion of the turbines is posed
by the internal infiastructure of the wind park. Here, the only
components of the internal wind park infiastructure that are
included in the aggregated model are the transformers at the
wind turbines and, if present, at the point of common coupling
(PCC). The reason for this is that transformers have a
relatively high impedance, whereas the cables within the park
are rather short and therefore have a low impedance when
compared to the transformers. Their impedance is therefore
neglected. The impedance of the cable fioni the PCC to the
point of grid connection is kept, because it c m be quite long,
particularly for off shore wind parks. The resulting aggregated
wind park model is depicted in figure 3.

Active power from n


simplified VS wind
turbine models, one
voltage controller

Turbine transformer
impedance divided
by n

Impedance of PCC
transformer and grid
connection

Figure 5 . Aggregated model of wind park with n variable speed wind turbines

3) Aggregation of Variable Speed Wind Turbines


The instantaneous power generated by a variable speed wind
turbine is dependent on the actual value of the rotor speed,
rather than on the wind speed. Therefore, adding the
mechanical power of the individual wind turbines, as was done
with constant speed wind turbines according to the right
equation of (3), would introduce an error. It would be assumed
that there exists an instantaneous relation between wind speed
and generated power, which is, however, not true. Therefore,
in the aggregated model of a wind park with variable speed
wind turbines, the rotor speed of the individual turbines is kept
track of and the electrical power of the individual turbines is
added, rather than the mechanical power.
The variable speed wind turbine model is simplified before
aggregation, using the following considerations:
When it is assumed that the performance coefficient cp(h,O)
always equals its maximum value, the cp(h,8)characteristic
can be omitted from the model and be replaced by a
constant equal to the maximum of cp. Only a minor error
results from this simplification, as the rotor speed versus
power control characteristic is such, that cp is kept at its
maximum as much as possible. This assumption is thus
equivalent to assuming an ideal rotor speed controller.
The non-linear rotor speed versus control characteristic is
replaced by a fust order approximation.
When the integrator in which the rotor speed is stored is
limited, e.g to 1.1, the pitch angle controller can be
omitted, as it is no longer needed for limiting the speed.
The simplified variable speed wind turbine model that results
from the above, with which each of the turbines in the
aggregated park model is represented, is depicted in figure 4.
The power generated by the aggregated wind park model
equals the sum of the outputs of the simplified models depicted
in figure 4. Thus, as mentioned above, the electrical power is
added, rather than the mechanical power. If the wind turbines
in the park are equipped with voltage controllers, one voltage
controller is attached to the aggregated model. With respect to
the internal park infrastructure, the same approach that was
used in case of the aggregation of the constant speed wind
turbines is applied, i.e. the transformers and grid connection
are kept and the cables within the park are neglected. The
resulting aggregated model is depicted in figure 5.

Wind

Filtered
wnd

Iv. SIMULATION RESULTS


A. Simulated Cases
In order to investigate the accuracy of the aggregated wind
park models presented above, a number of simulation runs has
been carried out. First, two wind park connection schemes
were developed in order to allow a broad comparison of the
detailed and aggregated model. Both consist of ten wind
turbines with a nominal power of 2 MW each, one connected
as a string and the other as a star. The investigated wind park
layouts are depicted in figure 6. Then, simulations are carried
out and the response of the detailed and aggregated wind park
models are compared.
First, the response to a wind speed change is investigated by
applying a wind speed sequence to the detailed and aggregated
models. The wind speed sequence consists of an average value
corresponding to a generated power of 0.75 P.u., a gust with an
amplitude of 3 m / s and a duration of 10 s and a ramp with a
duration of 30 s and an amplitude of 4 m/s. In both cases, the
wind comes from above in figure 6. The simulation run takes
90 s. The gust and the ramp start 30 s and 50 s in the string
connected wind park and at 85 s and 105 s in the star
connected wind park, relative to the parks centre. Different
gust and ramp starting times are used in order to observe the
effect of the wind speed change in case of both connection
schemes within the same time kame.

2 MW
*1 Mvar

Maximum
rotors eed

.Mrcha..- -. ._

nical

M i d u r n 1
rotor Speed
Active

power
[PU

Figure 4. Simplified variable speed wind turbine model for use in an


aggregated
model of a wind park with variable meed wind turbines
-_ -

Figure 6.Investigatedwind park connection schemes

Then, the fault response of the aggregated model is


investigated by comparing it to that of the star scheme. A
simulation run of 10 s is used and the wind speed is assumed
to be constant. After 1 s, a fault with a duration of 150 ms is
applied at the PCC.
B. Simulation Results
In figure 7, the simulation results are depicted. Because the
interaction of the wind park with the power system is the main
point of interest in power system dynamics simulations, the
quantities on which the comparison of the detailed (solid lines)
and aggregated (dotted lines) models is based, are the active
and reactive power flowing from the PCC to the system and
the voltage at the PCC. The left and middle figures depict the
response to a wind speed change, the right the fault response.
The upper figures correspond to the constant speed wind
turbines, the lower to the variable speed wind turbines.
It can be concluded from the figures that in all cases, there is a
high degree of correspondence between the detailed and the
aggregated model. The observed differences in case of the
constant speed wind turbines are mainly caused by the neglect
of turbulence. However, the effect of turbulence on the
aggregated output power of a wind park will become smaller,
the more wind turbines the park contains. In practice, most
wind parks will consist of far more than ten turbines. It is
therefore not considered necessary to modify the aggregated
model of the park with constant speed wind turbines in order
to include the effect of turbulence. In case of variable speed
wind turbines, turbulence is hardly reflected in the output
power, due to the functioning of the rotor as an energy buffer.
The difference between the active power in case of the
variable speed wind turbines is mainly due to the
simplifications applied in deriving the simplified variable
speed wind turbine model depicted in figure 4 from a detailed
model. The different rotor speed versus power characteristic
leads to small differences in generated active power. In turn,
these lead to differences in reactive power and PCC voltage.
In case of the fault response, the differences between the
detailed and the aggregated model can hardly be observed.
They are similar to those occurring in case of a wind speed
change but are hided by the larger scale of the graphics.
V. CONCLUSIONS

In this paper, aggregated models of a wind park with constant


speed and with variable speed wind turbines for use in power
system dynamics simulations are presented and verified by
comparing their response to that of detailed wind park models.
The use of aggregated models reduces the modelling effort for
the user and the amount of data to be entered, because no
longer a detailed model of the wind park infrastructure and of
the individual turbines is required. Further, it eliminates the
need to specify the wind speed at each individual wind turbine
within the park.
When the response of the aggregated model was compared
with a detailed model, it was concluded that notwithstanding
the applied simplifications, the agreement between the
responses of the aggregated and detailed wind park models is
very close, both for a wind speed change and a fault.

VI. REFERENCES
[l] J.G. Slootweg, S.W.H. de Haan, H. :Polinder, W.L. Kling, Aggregated
modeling of wind parks with variable sp:ed wind tuirbines in power system
dynamics simulations, 14th Power Systems Computation Conference,
Sevilla, Spain, 24-28 June 2002.
[2] R.M.G. Castro, J.M. Ferreira de Jesus, A wind park reduced-order model
using singular perturbations theory, IEEE Transactions on Energy
Conversion, v.11, n.4, December 1996, p.735 -741.
[3] R.M.G. Castro, J.M. Ferreira de Jesus, An aggregated wind park model,
13th PSCC Power Systems Computation Conference, Trondheim, Norway, v.
2, p. 1302-1307,28 June 28-July 2, 1999.
[4] V. Akhmatov, H. Knudsen, An aggregate model of a grid-connected,
large-scale, offshore wind farm for power stability inirestigations-importance
of windmill mechanical system, International Journal of Electrical Power &
Energy Systems, v. 25, n. 9, p. 707-719, July 2002.
[5] J.G. Slootweg, W.L. Kling, Modeling of Large Wind Farms in Power
System Simulations, Proceedings of the IEEE PES Summer Meeting,
Chicago, July 25-29,2002.
[6] S. Heier, Grid integration of Wind Energy Conversion Systems, Chicester,
UK: John Wiley & Sons Ltd., 1998.
[7] P. Kundur, Power system stability ancl control, New York: McGraw-Hill,
Inc., 1994.
[8] J.G. Slootweg, H. Polinder, W.L. Klirig, Dynamic Modelling of a Wind
Turbine with Direct Drive Synchronous Cenerator and Back to back Voltage
Source Converter and its Controls, 2001 European Wind Energy Conference
and Exhibition, Copenhagen, Denmark, July 2-6,2001,
[9] J.G. Slootweg, H. Polinder, W.L. Kling, Dynamic Modelling of a Wind
Turbine with Doubly Fed Induction Generator,, 2001 IEEE Power
Engineering Society Summer Meeting, Vancouver, Canada, July 15-19,2001,
[lo] J.G. Slootweg, S.W.H. Haan, H. Polinder, W.L. Kling, General Model
for Representing Variable Speed Wind Tiirbines in Power System Dynamics
Simulations, IEEE Transactions on Power Systems, v. 18, n. 1, February
2003, p. 144-151.
[l I] 0. Wasynczuk, D.T. Man, J.P. Sullivan, Dynamic behavior of a class of
wind turbine generators during random wind fluctuations, IEEE
Transactions on Power Apparatus and Systems, v.100, n.6, June 1981,
p.2837-2845.
[12] P.M. Anderson, A. Bose, Stability simulation of wind turbine systems,
IEEE Transactions on Power Apparatus and Systems, v.102, n.12, Dec. 1983,
p.379 1-3795.
[13] J. Cadogan, M. Milligan, Y. Wan, B. Kirby, Short-term output
variations in wind farms: implications for ancillary services in the united
states, Wind Power for the 21st Century, Kassel, Germany, September
25-27,2000,

VII. BIOGRAPHIES
J.G. Slootweg (h4 01) received his MSc degree in
electrical engineering f k m Delft University of Technology
on September 23rd, 1998. During his education he stayed
in Berlin for SIX months, to hear lectures at TU Berlin and
to conduct research at the Dynamowerk of Siemens AG
He is currently working towards a PhD on the effects of
large scale integration of new technology on power system
dynamics. The research is carried out at the Electrical
Power Systems Laboratory of Delft IJT.

W.L. Kling (M 95) received his MSc degree in electrical


engineering from the Technical University of Eindhoven in
1978. Currently he is a part time professor at the Electric
Power Systems Laboratory of Delft UT.His experience lies
in the area of planning and operation of power systems He
is involved in scientific: organizaticlns such as Cigrk and
IEEE. He is the Dutch representative in the Cigrk Study
Committee C 1 System Lbvelopment and Economics.

Wind speed change


Star connected

17 5

Wind speed change


String connected

9
'

1-50rns fault at PCC

40

17

16 5

p
I

16

T35 5

15

14.5

14

20

40
Time 131

60

80
Time [s]

1'-

I
0

1006

20

40
Time IS]

60

80

20

40

60

-40

80

Time [SI

i
1

0994 ,-'

20

145l

20

40
Time

60

098

80

20

[51

40

60

80

60

80

40
Time [SI

60

80

60

80

Time Is1

-1 6

20

40

20

40

Timeis]

m e 151
1.0027

1
I

...

1 003

- 1

10021 0

20

40

60

80

1001 0

20

40

60

80 o - o o

[SI
Time [SI
Figure 7. Simulation results. From left to right: response of star connected wind park to a wind speed change, response of string connected wind park to a wind
speed change and fault response of wind park. The upper three figures depict active and reactive power and PCC voltage of a wind park with constant speed
wind turbines, the lower three figures depict active and reactive power and PCC voltage of a wind park with variable speed wind turbines. The solid lines
correspond to the detailed model and the dotted lines to the aggregated model of the wind park.
mne 151

Time