1
Impact of Wind Turbines Equipped with
Doubly-Fed Induction Generators on Distance
Relaying
Simon De Rijcke, Student Member, IEEE, Paula Souto Pérez and Johan Driesen, Member, IEEE
Abstract—Wind generation is evolving from single wind turbines to wind power plants with an installed capacity of tens
of megawatts. Therefore, wind turbines are also coupled to the
transmission grid instead of the distribution grid. Problems arise
during grid faults as the behaviour of wind turbines equipped
with doubly-fed induction generators differs fundamentally from
traditional synchronous generators. The short-circuit behaviour
is vital for protection issues. A distance relaying problem in
a Belgian grid topology is emphasized and studied. This study
concludes that the distance relay is not performing correctly in
case of three phase faults. Moreover, varying parameters such
as the crowbar resistance and the wind turbine loading have a
significant impact on the relay performances.
Index Terms—Doubly-fed induction generator (DFIG), three
phase grid fault, single line-to-ground grid fault, distance relay,
crowbar resistance.
N OMENCLATURE
Symbols
u
i
R
X
ψ
ω
ωn
σ
T
t
Voltage [V]
Current [A]
Resistance [Ω]
Reactance [Ω]
Flux [T]
Frequency [Hz]
Rated grid frequency [Hz]
Leakage factor [-]
Time constant [s]
time [t]
Subscripts
s
r
cb
syn
F
Stator
Rotor
Crowbar
Synchronous
Fault
I. I NTRODUCTION
W
IND energy is increasing its share in the total production capacity of many countries. The IEA1 states that
the global capacity is rising with 20-30 % a year [1]. In 2008,
wind energy accounted for 36 % of all installed capacity, or
The authors are with the Department of Electrical Engineering
(ESAT), Division ELECTA, Katholieke Universiteit Leuven (KUL), Kasteelpark Arenberg 10, 3001 Leuven-Heverlee, Belgium (e-mail: simon.derijcke@esat.kuleuven.be
1 International Energy Agency
978-1-4244-6551-4/10/$26.00 ©2010 IEEE
8484 MW, in the European Union. This number places the
growth of wind energy before any other electricity generating
technology. The growth of wind energy in the electricity grid
goes hand in hand with economical, technical and legislative
evolutions. A better understanding of the behaviour of wind
turbines is needed to fulfill these issues. This paper concentrates on the technical issues. Moreover, the focus is laid upon
the fault-ride through behaviour of wind turbines equipped
with doubly-fed induction generators.
Small wind turbines were traditionally coupled to the distribution grid. This interaction has been investigated extensively
in the past [2][3][4]. The growing number and size of wind
power plants gives rise to a tendency to the coupling of wind
turbines to the transmission grid. Comparison between both
grids is impossible due to fundamental differences and there
is a rising interest in describing the interaction between transmission grid and wind turbines. There is a general consensus
about the challenges system operators face concerning grid
protection when connecting wind turbines to the grid [2].
The parameters influencing the behaviour of the wind turbines, and thus the protection, the most are the generation
technology, the rated power, the operating condition and
the location of the grid fault. Three main technologies are
distinguished on the current market: the squirrel cage induction
generator (SCIG), the doubly-fed induction generator (DFIG)
and the direct-drive synchronous generator (DDSG). During
a short-circuit, the behaviour of the direct-drive synchronous
generator is determined by the converter connecting the generator to the grid. Because of the small thermal capacity of power
electronics, the current may never exceed the imposed limits
of the converter. In contrast with traditional generators, the
short-circuit current is limited by the power electronics. The
converters determine the current behaviour indepently from the
wind turbine. Both the SCIG and the DFIG have an induction
generator directly coupled to the grid. Thus, the short-circuit
behaviour depends highly on the machine characteristics.
This paper contributes to the better understanding of the
implementation of doubly-fed induction generators into the
electrical grid by focusing on protection issues. Although the
behaviour of DFIG is more elaborately described in literature
when applying three-phase faults, single line-to-ground faults
occur more often in transmission grids. This paper considers
both three phase faults and single line-to-ground faults. The
short-circuit behaviour is used for evaluating the impact on a
distance relay. Section II discusses the model of the DFIG that
is used for the simulations. Then, the behaviour during short-
2
circuit is described. Finally, section IV analyses the impact of
a wind power plant, equipped with DFIG, on a distance relay.
II. M ODEL
A. Single wind turbine
The DFIG technology is shown in Fig. 1 and is implemented
in PowerFactory DIgSILENT2 . The stator is directly coupled
to the grid by a three winding transformer. The rotor is connected to the grid through a back-to-back partial scale power
converter in series with an inductance. The power converter
consists of a converter at the rotor side and a converter at the
grid side. Both converters are separated by a DC-bus. In order
to protect the converter against overcurrents or high voltages
in case of strong grid transients, a crowbar is implemented.
When high current transients occur, the crowbar short-circuits
the rotor windings using resistances with a predefined value.
B. Wind power plant
In section IV, an aggregated model is used to study the
impact of a wind power plant during grid faults on a distance
relay. This choice is justified when using a small wind power
plant with negligible cable distances between the wind turbines
and with small differences between the different turbine operating conditions. Moreover, the collective impact of the wind
power plant on the grid protection is the subject of study and
not the individual operating conditions inside the wind park.
Examples of aggregated models can be found in literature [7].
III. B EHAVIOUR OF DFIG DURING SHORT- CIRCUIT
All the simulations are performed with PowerFactory
DIgSILENT. The simulations include three phase shortcircuits and single line-to-ground short-circuits.
A. Three phase short-circuit
Voltage
[p.u.]
DFIG
Gear box
Converter
Crowbar
Fig. 1.
DFIG model
4$402!0-4$*(+$302$*/,45'(!,'(,/15
".3#
Time
Transient simulations require adequate modelling of the
mechanical parts, in order to take well into account the
mechanical oscillations that are resulting from electrical disturbances and that may influence the behaviour of the system.
Therefore, a two mass model is used in this paper [5].
Strong torque fluctuations force the shaft to behave like a
torsion spring. The generator is modelled by a fifth order
model [6]. The model equations for a doubly-fed induction
generator are analogous to an induction machine. Only the
rotor voltages, which are controlled in a DFIG, differ. When
using a synchronous rotating coordinate system, the governing
equations for the stator and rotor voltage are:
s
dψ
ωsyn ,
ψs +
us = Rsis + j
ωn
dt
(1)
r
ωsyn − ωr 1 dψ
ur e−j(ωsyn −ωr )t = Rrir + j
. (2)
ψr +
ωn
ωn dt
The controlling of the converters is of great importance
during steady state operation. However, grid faults occuring
close to the wind turbines activate the crowbar and the
generator control is lost. The controlling units are not further
explained. The full control can be found in reference [6]. The
parameters of the DFIG used in the simulations are added in
Appendix A
2 http://www.digsilent.de
Current
[p.u.] 4$402+$3(522(/4,/15
4$402+$3(522(/4,/15
4$402+$3(522(/4,/15
".3#
Time
Fig. 2. Stator voltage and stator phase currents during a three phase shortcircuit.
The fault, with resistance RF = 0.2 Ω, is applied at time
t0 = 0 s. Large oscillating currents appear in the stator.
Because of the magnetic coupling, large rotor currents are
initiated. The converter at the rotor side must be protected and
the crowbar is activated shortly after the fault. After 500 ms,
the fault is cleared. The crowbar is removed after a predefined
time of 600 ms after the fault initiation.
Fig. 2 illustrates the phase currents generated by the DFIG
and the voltage at the point of common coupling. The current
can be estimated by the expression [8]:
√
is =
2Vs −t/Ts
jωs t −t/Tr
e
.
−
(1
−
σ)e
e
jXs + Rcb
(3)
3
Ssc = 5.2 GVA
Current
HV
[p.u.]
150 kV
A
T
3#3/1*#2'411'.3+.04
3#3/1*#2'411'.3+.04
3#3/1*#2'411'.3+.04
!-2"
36 kV
WT 5 MW X 6
Fig. 3. Stator phase currents during a single line-to-ground fault. Phase A
is short-circuited (red colour).
with Vs the stator voltage and Xs , Rcb , σ generator properties. The time constants Ts and Tr are:
Ts
MV
Time
L
= s,
Rs
Fig. 4.
Grid topology with zoned distance relay.
DIgSILENT
Power Factory
Comtrade
OMICRON
U,I (analog)
Relay
7SA612
log files
(4)
Lr
=
.
(5)
Rr + Rcb
Due to the short-circuit, the flux stops rotating and causes
a DC-component in the stator with a time constant Ts . This
component is represented by the first term between brackets
in equation (3). The second term between brackets represents
a DC-component in the rotor, causing an AC-component in
the stator with time constant Tr . Time constant Tr is 30 times
smaller than Ts and the AC-component is rapidly damped in
comparison with the DC-current. A stationary current must
be added to the current in equation (3). Because the residual
voltage is very low, almost no excitation is left and the
stationary currents drop to negligible values. During the fault,
the power exchange drops to zero because of the voltage loss.
DIGSI
Tr
Fig. 5.
Data stream for simulations
A. Simulations
All grid fault simulations are performed in PowerFactory
DIgSILENT. The simulation data is converted by COMTRADE data exchange to digital current and voltage signals
with a constant sample frequency of 2 kHz. These digital
signals are converted to analog signals for injection in the
distance relay. All data is processed and studied using log
files. The complete data stream is illustrated in Fig. 5.
B. Grid topology and relay setting
B. Single line-to-ground fault
The phase currents are shown in Fig. 3 . In contrast with
the three phase fault, the single line-to-ground fault shortcircuits only one phase and keeps the two other phases excited.
Due to this excitation, the current amplitudes do not drop
to zero which was the case for the three phase fault. The
short-circuited phase has an analogous behaviour as the phase
currents for the three phase fault. Both a large DC-component
and a fast decaying AC-component occur in the stator current.
All the phases are magnetically coupled and the exact phase
currents are difficult to calculate. A full explanation of the
current requires the analysis of direct, inverse and homopolar
components and is beyond the scope of this paper.
IV. S TUDY CASE : I MPACT ON A DISTANCE RELAY
This section describes the impact of a wind power plant
using doubly-fed induction generators on a distance relay in
Belgian grid topology. First the grid topology is explained.
Thereafter, a three phase short-circuit and a single line-toground fault are simulated. Finally, a parameter analysis is
performed.
The topology of the grid is illustrated in Fig. 4. Six DFIG
wind turbines, with a total capacity of 30 MW, are connected
to the Medium Voltage (MV) bus. This MV bus is connected
to the transmission grid at 150 kV by transformer T. The
transmission grid is represented by an equivalent with shortcircuit power Ssc = 5.2 GVA.
The impact of the wind turbines on distance relay A3 is
studied, considering faults on the High Voltage (HV) bus.
Only in section IV-D2, faults on a line emanating from the
HV bus are considered. All these faults are situated in the
forward direction of the distance relay A. The operation zones
of this relay are illustrated on Fig. 6, as well as the delay
time for each zone. Assuming faults on the HV bus, the bus
protection is the primary protection system. In case of failure,
all lines emanating from bus HVf rom need to be disconnected
by the distance relays located on these lines, acting as a backup
protection. For the wind turbines, distance relay A acts as a
backup and needs to disconnect the wind turbines from the bus.
When functioning correctly, the relay detects the fault on the
bus and trips after 500 ms, which is the correct delay time to
3 The
relay under consideration is distance relay type 7SA612 of Siemens4 .
4
120%
70%
Zone 2
0,100
2.5 s
0,075
1.3 s
Zone 1
T
Start zone
Forwards
A
X/Ohm(sec ondary)
MV
HVto
HVfrom
150 ms
2.5 s
Start zone
Backwards
Fig. 6.
0,050
500 ms
Zone 3
50%
0,025
-0,000
-0,025
-0,050
-0,075
Distance zones with delay times.
-0,100
1,25
-0,125
1,00
-0,10
0,00
0,75
0,50
X/Ohm(secondary )
(a)
0,25
0,025
0,00
-0,25
0,10
R/Ohm(secondary)
0,020
0,015
-0,50
X/Ohm(secondary)
-0,75
-1,00
-1,25
0
1
0,005
0,000
-0,005
2
R/Ohm(secondary)
0,010
-0,010
-0,015
Fig. 7.
fault.
Close-up of the zone on a R-X plane for a single line-to-ground
-0,020
-0,02
-0,01
0,00
0,01
0,02
0,03
R/Ohm(secondary)
give the bus protection sufficient operating time. Thus, faults
on the HV bus are situated in zone 1 (Fig. 6). The second
zone in the forward direction is zone 2. The first zone in the
backward direction is zone 3. Both startzones are combined in
one bi-directional zone. The zones of interest are represented
on a close-up of the R-X plane on Fig. 7.
C. Base case
Before the fault initiation, the six DFIG are producing 4.5
MW and 0.2 MVar each. The generator speed is 1.08 p.u. The
detection time is represented by the pick-up time (PU time).
1) Three phase fault: The calculated impedance by the
distance relay is illustrated on Fig. 8(a). The impedance is
only illustrated for only one phase. The circulating impedance
is the result of turning current phasors produced by the
DFIG. This behaviour causes the impedance location to switch
between zones 1 and 3, respectively a zone in the forward and
backward direction. The relay detects the fault immediately in
the forward direction. After 34 ms, the current drops to less
than 0.2 A (on the secondary side of the TP), resulting in
an interruption of the detection. A substitution of the delay
time of 500 ms in zone 1 by 0 ms results in a fault detection
and removal. One can conclude that a fault on the HV bus is
correctly detected but is not removed after 500 ms due to the
detection stop after 34 ms. Because of the fault, the DFIG lost
its excitation resulting in a very rapidly decaying short-circuit
current.
2) Single line-to-ground fault: Fig. 8(b) shows the
impedance location of the faulted phase calculated by the
(b)
Fig. 8. Impedance location in the R-X plane for a three phase fault (a) or a
single line-to-ground fault (b).
relay in case of a single line-to-ground fault. Although the
location is situated in the backward direction several times,
the relay detects the fault in the forward direction correctly.
The relay uses a memory and tolerates short interruptions
of the detection in the forward direction. In contrast with a
three phase fault, a single line-to-ground fault is detected and
removed correctly.
D. Parameter analysis
This section describes the impact on the distance relay
caused by varying parameters.
1) Fault resistance: A higher fault resistance prevents detection of three phase faults and single line-to-ground faults.
This malfunctioning is caused by the injection factor. This
phenomenon is illustrated on Fig. 9. The strong grid (5.2
GVA) supplies a high short-circuit current Ig . Even with a
small fault resistance, this high current pushes the bus voltage
Ur up. In combination with the small current Ir coming from
the generators, the measured impedance Zr = Ur /Ir is not
located in the detection zones of the relay and detection is
prevented.
2) Faults on a line in zone 1: Unlike any other simulation in
this paper, this section assumes line faults instead of bus faults.
The main protection for zone 1 is a distance relay emanating
from bus HVf rom in Fig. 6. Relay A acts as a secondary
5
TABLE II
R ESULTS OF THE INJECTION SIMULATIONS FOR VARYING LOAD
Ir Ur
Zd
Zg
Rf
DFIGs
Ig
Ir
Fig. 9.
Fault
GRID
Phenomenon of injection factor.
TABLE I
R ESULTS OF THE INJECTION SIMULATIONS FOR VARYING
RF
Three phase
PU time [ms]
B
34
C
/
Single phase
PU time [ms]
A
600
C
/
TABLE III
R ESULTS OF THE INJECTION SIMULATIONS FOR A WIND POWER
Crowbar resistance [p.u.]
0.01
low
Legend: A = fault detected and removed; B = fault detected and
not removed; C = fault not detected.
CROWBAR RESISTANCE
Fault
load
high (base case)
0.05
0.1
0.2
0.5
1
2
0mΩ
1mΩ
Three phase
PU time [ms]
B
30
B
160
B
30
B
10
B
34
C
/
C
/
C
/
Single phase
PU time [ms]
A
151
A
155
A
600
A
600
A
600
A
600
C
/
C
/
Legend: A = fault detected and removed; B = fault detected and
not removed; C = fault not detected.
protection for faults in zone 1. The same conclusions of section
IV-D1 for a higher fault resistance are valid. Even a fault at a
distance of 1% of the total distance of zone 1, is not detected in
zone 1 but in zone 2. This would not be the case for a distance
relay located on the line emanating from bus HVf rom . This
relay would see the current contribution of the grid coming
from other lines connected to bus HVf rom .
3) Crowbar resistance: The results are summarized in
Table I. For the three phase short-circuit, the higher crowbar
resistance in series with the rotor windings lowers the peak
current and prevents the detection of a fault. The quick
insertion of the crowbar affects the detection properties of
the relay. A lower crowbar resistance causes an unpredictable
detection behaviour of the relay. The relay switches between
forward and backward detection until the current value drops
below the pick-up threshold. For a single line-to-ground fault,
a higher resistance damps the current and prevents detection. A
lower value implies a switching detection behaviour between
the forward and backward direction. With a fault resistance
RF = 1 mΩ, the PU time of 155 ms is preceded by a period
of backward and forward detection. The fault is cleared after
335 ms by a final detection in the backward direction. The
situation is even worse when RF = 0 mΩ because the time
between fault initiation and removal is 151 ms. When a single
line-to-ground fault occurs, the generator is excited by the grid
using the two healthy phases. The generator absorbs a lot of
reactive power which can cause the apparent power flow to
turn around leading to a backward detection.
4) Turbine loading: The load is reduced to 0.3 MW / 0.01
MVar and the slip is -15%. Both the three phase and the single
line-to-ground fault stay undetected because of the reduced
peak value. The results are summarized in Table II.
5) Number of generators: Four wind turbines are added to
the wind power plant of the base case. The results are summarized in Table III. The detection behaviour stays unchanged
for the three phase fault. Only a small increase is noticed in
PLANT WITH EXTRA WIND TURBINES
Fault
6 wind turbines (base case)
10 wind turbines
high load
low load
Three phase
PU
time
[ms]
B
34
B
39
C
/
Single phase
PU
time
[ms]
A
600
A
595
A
589
Legend: A = fault detected and removed; B = fault detected and
not removed; C = fault not detected.
the PU time because of the higher currents produced by the
extra generators. The higher short-circuit power causes the
detection of a single line-to-ground fault when the load of the
wind turbines is reduced to 0.3 MW / 0.01 MVar and with a
slip equal to -15%.
V. C ONCLUSION
First, one can conclude that for a grid fault close to the
generator, control is lost and the interaction with the grid is
determined by the machine characteristics.
Second, the impact of a wind power plant using doubly-fed
induction generators on a distance relay is studied. In case of
a three phase fault, the detection is interrupted due to small
currents resulting from the loss of excitation. A single lineto-ground fault is detected and removed correctly unless the
crowbar resistance or the load is lowered. A higher number of
generators increases the short-circuit power and has a positive
influence on the detection behaviour. A lower crowbar resistance causes an unpredictable detection behaviour resulting in
a backward detection and an early, uncorrect fault removal.
From these results, it follows that this distance relay with
its actual settings is not suitable as a backup protection for the
bus differential protection in this grid topology including wind
turbines equipped with DFIG. Further research should focus
on this problem and suggest possible solutions. Moreover,
other wind energy conversion systems like the direct-drive
synchronous generator should be included in further research
as the behaviour of this technology differs fundamentally from
the doubly-fed induction generator.
A PPENDIX A
This appendix gives the parameters of the DFIG that have
been used in Table IV
6
TABLE IV
PARAMETERVALUES OF THE DFIG
Parameter
Rated power
Voltage (stator)
Inertia
Crowbar reactance
Crowbar resistance
Number of Poles
Resistance (stator)
Resistance (rotor)
Reactance (stator)
Reactance (rotor)
Mutual reactance
Symbol
Pnom
Us
Jgen
Xcb
Rcb
p
Rs
Rr
Xs
Xr
Xm
Value
5 MVA
3.3 kV
101.7156 kgm2
0.1 p.u.
0.1 p.u.
2
0.00299 p.u.
0.004 p.u.
0.125 p.u.
0.05 p.u.
2.5 p.u.
A PPENDIX B
This appendix gives the settings of the distance relay type
7SA612 in Table V.
Simon De Rijcke (S 08) was born in Belgium, on
May 26, 1986. He graduated as an electrotechnical
engineer from the Katholieke Universiteit Leuven
(KULeuven) in 2009 and received a master in
energy. Since 2009, he is working as a research
assistant towards a Ph.D. in the research group
ELECTA, department of Electrical Engineering of
the KULeuven. His field of interest is the integration of renewable energy sources in the electricity
system.
Paula Souto Pérez graduated in Electrical Engineering in 2005 in the University of Vigo, Spain, with
intensification in Electrotechnics. Since 2006 she is
a research assistant in the research group ELECTA
of the K.U. Leuven (Belgium), working towards a
Ph.D. Her fields of interest are wind power and its
relation with grid stability in Europe.
TABLE V
S ETTINGS OF THE DISTANCE RELAY
Zone
Zone 1
Zone 2
Zone 3
Start zone forward
Start zone backward
Reactance [Ω]
Resistance[Ω]
three phase
single phase
2.65
10
/
20
/
4.41
12
/
20
/
0.4417
4.44
5.6
15
200
R EFERENCES
[1] International Energy Agency, “Renewable Energy Essentials: Wind,”
International Energy Agency, Tech. Rep., 2008.
[2] N. Hadjsaid, J. F. Canard, and F. Dumas, “Dispersed generation impact
on distribution networks,” Computer Application in Power, vol. 12, pp.
22–28, 1999.
[3] G. Hodgkinson, “System implications of embedded generation and its
protection and control. PES perspective.” in System Implications of
Embedded Generation and Its Protection and Control. Birmingham,
UK: IEE, Februari 1998, pp. 1–15.
[4] P. P. Barker and R. W. de Mello, Determining the Impact of Distributed
Generation on Power Systems: Part 1 - Radial Distribution Systems.
Power Technologies, Inc., 2000.
[5] Anca D. Hansen and Gabriele Michalke and P. Sørensen and T. Lund
and Florin Iov, “Co-ordinated Voltage Control of DFIG Wind Turbines in
Uninterrupted Operation during Grid Faults,” Wiley Interscience, vol. 10,
pp. 51–68, 2006.
[6] A. D. Hansen and F. Iov and P. Sørensen and N. Cutululis and C.
Jauch and F. Blaabjerg, Dynamic wind turbine models in power system
simulation tool DIgSILENT. Ris-R-1400(ed. 2)(EN), 2007.
[7] M. Pöller and S. Achilles, “Aggregated Wind Park Models for Analyzing
Power System Dynamics,” in Fourth International Workshop on LargeScale Integration of Wind Power and Transmission Networks. Billund,
Denmark: DIgSILENT, October 2006, pp. 99–104.
[8] J. Morren and S. W. H. de Haan, “Short-Circuit Current of Wind Turbines
with Doubly Fed Induction Generator,” Electrical Power and Energy
Systems, vol. 22, pp. 174–180, 2007.
Johan Driesen (S 93-M 97) was born in 1973 in
Belgium. He received the M.Sc. degree in 1996 as
Electrotechnical Engineer from the K.U. Leuven,
Belgium. He received the Ph.D. degree in Electrical
Engineering at K.U.Leuven in 2000 on the finite
element solution of coupled thermal-electromagnetic
problems and related applications in electrical machines and drives, microsystems and power quality
issues. Currently he is an associate professor at
the K.U.Leuven and teaches power electronics and
drives. In 2000-2001 he was a visiting researcher in
the Imperial College of Science, Technology and Medicine, London, UK. In
2002 he was working at the University of California, Berkeley, USA. Currently
he conducts research on distributed generation, including renewable energy
systems, power electronics and its applications, for instance in drives and
power quality.