This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON ENERGY CONVERSION 1

Evaluation of the Performance of a DC-Link

Brake Chopper as a DFIG Low-Voltage
Fault-Ride-Through Device
Graham Pannell, Bashar Zahawi, Senior Member, IEEE, David J. Atkinson, and Petros Missailidis

Abstract—The performance of the doubly fed induction gener-

ator (DFIG) during grid faults is attracting much interest due to
the proliferation of wind turbines that employ this technology. In-
ternational grid codes specify that the generator must exhibit a
fault-ride-through (FRT) capability by remaining connected and
contributing to network stability during a fault. Many DFIG sys-
tems employ a rotor circuit crowbar to protect the rotor converter
during a fault. Although this works well to protect the generator,
it does not provide favorable grid support behavior. This paper
describes an experimental investigation of an alternative FRT ap-
proach using a brake chopper circuit across the converter dc link
to ensure that the dc-link voltage remains under control during a
fault. Two different approaches to chopper control are examined
and the resulting FRT performance is compared with that of a
conventional crowbar approach. The new chopper-based control
methods are experimentally evaluated using a 7.5-kW DFIG test
rig facility.
Index Terms—DC–DC conversion, doubly fed induction gener-
ator (DFIG), induction generators, power conversion, wind power
generation.

Fig. 1. Wind turbine DFIG including conventional crowbar protection.

I. INTRODUCTION
S wind power begins to represent a greater proportion of
A total generation capacity, wind farms are having greater
influence over power system behavior. In this scenario, wind
sion, or, catastrophically, the cascade disconnection of genera-
tors leading to a system underfrequency event and widespread
consumer disconnection. If wind generation is to form a robust
farms are required to contribute to the stability of the system,
component of the electrical power system, wind farms must con-
including the key issue of fault-ride-through (FRT), which refers
tribute to the stability and reliability of the transmission grid,
to the capability of generation plant to remain connected, dy-
including the provision of grid support during grid faults, or
namically stable, and offer network support throughout a serious
voltage dips. Consequently, various national transmission sys-
voltage disturbance on the transmission network. Although the
tem grid codes now require wind farms to remain connected
voltage dips associated with grid faults may last for only a few
during specified voltage dips, and to supply active and reactive
cycles, they can bring about certain undesirable characteristics
power into the network.
of induction-machine generators [1], [2]. These include: uncon-
Doubly fed induction generator (DFIG) technology is
trolled active and reactive power, continued voltage suppres-
presently dominant in the growing global market for variable
speed wind power generation, due to its cost-effective, partially
rated power electronic converters. However, the DFIG is sensi-
Manuscript received July 11, 2012; revised November 19, 2012 and Febru- tive to dips in supply voltage because the internal machine flux
ary 28, 2013; accepted April 16, 2013. This work was supported by the is exposed directly to the electrical grid. The induction generator
U.K.’s Engineering and Physical Sciences Research Council (EPSRC) through very quickly loses internal magnetization in proportion to the
the Engineering Doctorate Programme at Newcastle University. Paper no.
TEC-00361-2012. lost voltage producing large outrush currents on both stator and
G. Pannell is with AC Renewable Energy Systems Limited, Hertfordshire, rotor circuits. Without specific measures to protect against “ride
WD4 8LR, U.K. (e-mail: graham.pannell@res-ltd.com). through” grid faults, a DFIG risks damage to its power electronic
B. Zahawi, D. J. Atkinson, and P. Missailidis are with the School of Elec-
trical and Electronic Engineering, Newcastle University, Newcastle upon Tyne, devices and dc-link capacitors due to the resulting overcurrents
NE1 7RU, U.K. (e-mail: bashar.zahawi@ncl.ac.uk; dave.atkinson@ncl.ac.uk; and/or overvoltages. Conventional converter protection is via
petros.missailidis@ncl.ac.uk). a sustained period of rotor–crowbar application [3], [4] that
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. protects the power converter while allowing the converter to re-
Digital Object Identifier 10.1109/TEC.2013.2261301 sume control at the earliest possible opportunity. Fig. 1 shows
0885-8969/$31.00 © 2013 IEEE
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON ENERGY CONVERSION

a schematic diagram of a wind turbine DFIG with conventional

IGBT based crowbar protection.
The crowbar was developed as a standard rotor circuit pro-
tection device long before the advent of wind turbine grid code
regulations [5]. It involves connecting the rotor phases together
through a designated resistance, diverting current from the rotor-
side converter and rapidly deenergizing the rotor. The crowbar
absorbs the initial energy outflow from the machine, while the
resistance shortens the effective decay timescale of the rotor flux
decay, hence hastening the demagnetization process. Conven-
tionally, the crowbar is applied for an extended duration to fully
demagnetize the rotor. Unfortunately, vector control is lost while
the crowbar is applied and the induction machine must draw its
magnetization from the stator side, producing a high-slip reac-
tive power demand leading to poor power and reactive power
outputs and sustained suppression of the stator voltages. The
FRT performance of wind turbine DFIGs has therefore received
much attention in the literature in recent years. Many schemes
have been proposed to ensure DFIG compliance with grid code Fig. 2. Wind turbine DFIG with dc brake chopper protection.
requirements including the use of a brake resistors [6]–[8], the
application of a controlled resistance crowbar [9], control of the
timing of crowbar application [10], the application of a dynamic
voltage restorer to compensate the faulty line voltage [11], and
the use of novel controller designs to improve the transient
response of the grid-connected DFIG under supply fault condi-
tions [12].
An alternative protection technique obtained by using a dc-
link brake chopper to contain the dc-link voltage while accept-
ing transient rotor overcurrents has also been proposed in the
literature. This method has been demonstrated successfully in Fig. 3. Rotor converter with dc-link brake chopper.
simulation, when used together with a rotor crowbar [13], a se-
ries dynamic brake resistor [14], [15], and even in conjunction capacitance to sink power when the dc-link voltage exceeds a
with a superconducting magnet [16], [17]. The use of the dc fixed predefined threshold.
brake chopper alone as a DFIG protection device has also been
proposed and its performance compared with that of a crowbar
circuit in a simulation study [18], [19] although no experimental II. DC-LINK BRAKE CHOPPER CIRCUIT
investigation of the application of the dc-link brake chopper as The dc-link brake chopper is a simple protection device that
a grid FRT protection device has previously been published. shorts the dc-link through a power resistor when the dc-link
This paper presents an experimental investigation of the use voltage exceeds a fixed threshold. The brake is used to contain
of the dc-link brake chopper as a sole protection device for the dc-link voltage while accepting transient rotor overcurrents.
DFIG grid FRT management. Two different control methods The dc-link brake appears somewhat similar to the rectifier-
are investigated. The first is a relatively simple control strategy insulated gate bipolar transistor (IGBT) crowbar configuration.
in which rotor current control is resumed after a short time delay Instead of a separate rectifier, the six antiparallel diodes in the
initiated once rotor currents have returned to a level compatible DFIG’s rotor-side converter are up-rated to handle short-circuit
with effective current control. A second, more advanced con- currents. A power resistor and series switch are placed in parallel
trol method was also examined where the application of the dc with the dc-link capacitance to sink power as required.
chopper is released and pulsewidth modulated (PWM) control A DFIG dc-link brake chopper is shown schematically in
restored once rotor current had fallen below a given threshold Fig. 3. Only the rotor-side converter is pictured, as this is the
in order to optimize the resumption of power control. Control focus of DFIG fault response. An IGBT chopper circuit is used
of the brake chopper and the FRT performance characteristics to rapidly engage and disengage the resistor. The chopper works
of the dc brake chopper based schemes are investigated using a on a hysteresis band, i.e., the turn-OFF voltage is set below the
7.5 kW laboratory DFIG test rig. turn-ON threshold. The antiparallel diode connected across the
A schematic diagram of a wind turbine DFIG with a dc brake brake resistor is needed to allow for the stray inductance effects
chopper is shown in Fig. 2. The crowbar circuit is dispensed when the chopper IGBT is switched off.
with and the six antiparallel diodes in the DFIG’s rotor-side In case of transient overcurrents as a result of a grid fault, the
converter are up-rated to handle short-circuit currents. A power rotor-side converter PWM control may be disabled and all of the
resistor and series switch are placed in parallel with the dc-link rotor converter’s IGBTs are switched off. Without an alternative
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANNELL et al.: EVALUATION OF THE PERFORMANCE OF A DC-LINK BRAKE CHOPPER 3

circuit path, the transient rotor currents are forced to conduct

through the rotor converter diodes. Rotor demagnetization en-
ergy is dumped into the converter’s dc link, causing the dc-link
voltage to rise rapidly. The resulting dc-link voltage rise will
trigger the brake, sinking power through the brake resistor and
helping to prevent a dangerous overvoltage event on the dc link.
When the rotor transients have sufficiently decayed, rotor-side
PWM and rotor current control may be reengaged.
If the rotor converter switches are held off for long enough
for the unsupported rotor flux to decay, then the dc-link voltage
will reverse block the converter’s diodes. The diodes will cease
conducting from the next current zero; the rotor currents are
then held at zero until IGBT switching is resumed.
The change in dc-link voltage is chiefly limited by the dc-link
capacitance. The voltage across the brake resistor will remain
roughly constant. The brake resistor can then be considered as
an approximately constant load, sinking power according to the
equation:
2
Vdc
Pbrake = . (1)
Rbrake
The presence of the capacitance means that the load is in-
versely proportional to the brake resistance. A large brake re-
sistor will draw negligible current, while the lower limit of re-
sistance is related to the maximum current rating of the resistor
and IGBT pair, which, in turn, is determined by the maximum
permissible dc-link voltage:
Vdc,m ax
ibrake ≤ . (2)
Rbrake
Fig. 4. Brake chopper modes of operation (values indicative). (a) PWM ON,
III. DC BRAKE CHOPPER CONTROL brake OFF; (b) PWM OFF, brake OFF; (c) PWM OFF, brake ON.

Brake chopper operating modes are shown in Fig. 4, with

reference to the protection threshold limits used in the test rig
used in this investigation. and the power dissipation in the brake resistor will halt the in-
The brake resistor must be sufficiently rated for to protect the crease in dc voltage. The chopper control circuit switches the
dc link from the worst case of fault induced rotor overcurrents. IGBT device with a duty cycle determined by the magnitude
The antiparallel diodes in the rotor converter must be capable of the overvoltage. The duty cycle of the chopper is increased
of handling rotor fault currents of about 5 pu. The dc-link ca- as the overvoltage increases up to a maximum permanently on
pacitance must be rated to withstand the rectified form of these state. The brake resistor is disengaged independently when the
overcurrents, for periods where the brake resistor is not engaged. dc voltage falls below a lower hysteresis threshold level set at
This translates to roughly 7 pu current (on a machine–rotor pu 795 V in the test rig.
base). The capacitor current rating is unlikely to form an onerous When the rotor current falls below 2.0 pu, resumption of
equipment restriction. The key issue is whether the dc brake can PWM and rotor current control is delayed by 20 ms to ensure
prevent the voltage from exceeding the capacitors’ maximum that, at the end of a given rectification period, the rotor flux
voltage capabilities; this can be secured with an appropriate had sufficiently deteriorated for the vector controller to be able
choice of brake resistor. to maintain the rotor current within safe limits. The rotor-side
Fig. 4(a) illustrates normal operation with the brake chop- converter power/reactive power proportional–integral (PI) con-
per off. In response to a grid fault, transient rotor overcurrents trollers are delayed by a further 20 ms (i.e., 40 ms after the
force the dc-link voltage to rise rapidly. Relying entirely on the current had exceeded 2.0 pu) to allow the current controllers to
dc brake chopper circuit and with no rotor crowbar, the rotor- settle. During control-off periods, the current control PI inte-
side converter PWM control is disengaged (i.e., all switches grators are set to deliver values calculated for normal operation.
are turned “off”), when the measured rotor current magnitude PQ integrators are set to the last measured current to avoid any
exceeded the threshold value of 2.0 pu, allowing the rotor-side possible instabilities when power/reactive power control is re-
converter antiparallel diodes to carry the transient high rotor sumed. As PQ PI control is resumed, the rotor current reference
currents, as shown in Fig. 4(b). The resulting dc-link voltage value changes were rate-limited to no more than 1.5 pu/s to
rise (above 810 V) will trigger the brake, as shown in Fig. 4(c), ensure current feedback control stability.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON ENERGY CONVERSION

1) A wind turbine simulator comprising a 10-kW dc motor

and its drive provided a torque input to the DFIG, to-
gether with a simulated mechanical model executed by
the control hardware to replicate the torque input from a
wind turbine rotor, drawing from industry-supplied wind
turbine data.
2) A 7.5-kW DFIG system including a 50 A-rated IGBT
power stack employed as a back-to-back bidirectional
power converter. The power stack was deliberately over-
rated to accommodate fault conditions without damaging
the equipment.
3) A grid fault emulator allowing a range of balanced voltage-
dip profiles to be applied to the DFIG via a 1:1 Y −
Δ isolation transformer by using 3 three-phase variable
autotransformers to apply three independent voltage levels
(healthy, fault, and recovery voltages) across the generator
terminals.
4) A control hardware assembly consisting of a combination
of analog protection and interface boards, isolated sens-
ing equipment, proprietary controller development equip-
ment, and a dedicated PC. A dSpace control system was
used to execute the DFIG controller, wind turbine simula-
tor and fault test control.
A 1200 V/100 A IGBT was used with a Semikron SKAI100U
Fig. 5. Overview of the 7.5 kW DFIG laboratory test rig employing a dc brake chopper driver board and an 180 Ω power resistor to
brake chopper as an FRT protection device. (a) Schematic of actual DFIG wind limit the dc-link voltage to 810 V. The brake chopper driver
turbine. (b) Schematic of DFIG test rig.
board made its own measurement of the dc-link voltage and
automatically engaged the IGBT if the dc link rises above 810 V.
A more advanced “minimum threshold” control method was The brake chopper cuts out if the dc-link voltage falls below
also developed in which the application of the dc chopper brake 795 V. The IGBT connects a dedicated power resistor in parallel
is released by rotor current threshold (1.9 pu) and feedback across the dc link to sink excess power. The power resistor was
control carefully restored to minimize the rectification period mounted externally to the DFIG control cabinet for better heat
and optimize the resumption of power control, exactly as for dissipation. The brake chopper switch and driver board were
the minimum threshold crowbar method described in a previ- mounted on the converter, as shown in Fig. 6.
ous publication [10]. In this method, the state variables of the A simple simulation of the brake chopper operation was cre-
rotor current PI controller are suspended with last-good-values ated for consolidating the analytical concepts and FRT ideas
to increase stability on the resumption of feedback control. The before the implementation on the test rig. From simulation
outer loop power and reactive power PI controllers are sus- tests, two multiples of operational rotor power dissipated from
pended and then soft-restarted by means of a 10 ms error signal the dc link was sufficient to maintain the dc-link voltage be-
ramp. When the control is resumed, the PQ PI controllers’ inte- low 1000 V in grid fault situations. The test rig brake resistor
gral components are artificially reset to output the most recent was sized to sink twice the maximum operational converter
measurement of rotor current. As a result, the rotor current PI power during grid faults, which equated to 0.5 pu power on the
controllers restart with approximately zero error input, mini- machine pu base, or 3.75 kW. Using (2), with a brake chop-
mizing any kick in current associated with the resumption of per cut-in value of 810 V, this equates to 175 Ω. A 180-Ω,
feedback control, as detailed in [10]. 0.6-kW wire-wound resistor was thus used, housed in a 19 cm
× 12 cm × 15 cm ventilated metal box, and externally mounted
IV. EXPERIMENTAL INVESTIGATION on top of the DFIG control cabinet for better heat dissipation;
the low (essentially thermal) power rating is acceptable because
A 7.5-kW test rig was designed and built to emulate the the grid fault durations are relatively short. Industrial DFIG
elements of a grid-connected DFIG wind turbine (see Fig. 5). wind turbines with ratings of 1.5–2.5 MW still operate with a
The test rig permitted the experimental evaluation of the DFIG typical dc link voltage in the region of 800 V to ensure that
system and its associated dc brake chopper under a range of grid standard 1200 V devices could be employed. This also qualifies
fault conditions. as a low voltage for ease of operation and maintenance as senior
authorized personal status may be required for voltages greater
A. Test Rig than 1 kV, which can be expensive and difficult to organize.
The test rig [20] comprises the following four main elements: Sizing the brake resistor to sink twice the maximum operational
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANNELL et al.: EVALUATION OF THE PERFORMANCE OF A DC-LINK BRAKE CHOPPER 5

Fig. 6. Back-to-back converter stack including brake chopper circuit: (a) front
view; (b) rear view.

converter power with a cut-in dc link voltage of 810 V, (2)

gives a brake resistor value of about 0.7 Ω for a 2-MW DFIG
drive.
The rotor converter’s antiparallel diodes were rated suffi-
ciently high to accommodate the maximum rotor overcurrents.
No crowbar was employed in any of the tests. The brake chop-
per employed on the dc link operated with its own hysteresis
controller using a dc-link voltage measurement independent of
the voltage recorded by the dSpace controller.
In the following discussions, the period for which the rotor
PWM was turned off and vector control suspended is referred
to as the rectification period, because any rotor currents flowing
were rectified by the diodes in the rotor-side converter.

B. Test Results
Results for a 15% fault test are given in Fig. 7, showing a
single rectification period following each of fault initiation and
clearance during which transient rotor overcurrents were di-
verted from the converter IGBTs. The dc-link voltage was well
maintained by the brake chopper in this test, scarcely passing Fig. 7. DC brake chopper; controlled delay method (shaded area indicates
820 V on either fault initiation or clearance. Near-dc components rectification operation).
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON ENERGY CONVERSION

of rotor current decayed within the rectification periods, al-

though the stator circuit near-dc decay extends well into the
fault and recovery periods [21]. Good rotor current control is
exhibited roughly 80 ms after each rectification period. The d-
component of rotor current was controlled midfault to 0.67 pu.
The q-component was saturated at −0.70 pu.
Prior to fault clearance, the machine delivered 20% active
power and 18% reactive power generation. The in-fault stator
voltage was steadied at 28 pu. After fault clearance, the reac-
tive power oscillated strongly, while the average value fell from
4 kvar import to zero over roughly 100 ms. This was a dis-
appointing result in terms of local voltage support, which was
suppressed by as much as 10% during this period. As a result of
the reactivation delay, the first rectification period endured for
27.6 ms, starting 1.2 ms after fault initiation, and the second rec-
tification period for 25.6 ms, starting 7.8 ms after fault clearance.
After the first rectification period, the rotor currents approached
zero. During the second period, the rotor currents were “pinched
off”; after reaching zero, the diodes became reverse-biased by
the dc-link voltage and did not conduct until the rotor PWM was
reengaged.
The rotor currents exhibited near-rotor speed oscillations [21]
after fault initiation and clearance reflecting the near-dc decay
on the stator circuit (54.5 Hz after fault initiation and 55.5 Hz af-
ter fault clearance, in accordance with the increased speed). This
oscillation extended further into the fault than normal—roughly
140 ms here. Also, the rotor current magnitude during the recti-
fication periods did not exceed 3 pu, well below the maximum
current seen when employing crowbar methods. These all point Fig. 8. DC brake chopper; minimum threshold rectification method (shaded
area indicates rectification operation).
to a slower decay of machine flux and that the dc link inhibited
the ac fault current decay.
Unlike a standard crowbar circuit, the dc-link brake does
not directly influence the rotor flux decay. Assuming that the C. Minimum Threshold Rectification
internal resistance of the dc-link capacitance is small, the con- The aforementioned results show that the simple control-
verter appears to the rotor as a short-circuit. This produces a delay method successfully managed the rotor currents and re-
cage-induction-machine-type natural flux decay. When the cur- stricted the dc-link voltage to below 850 V in the face of 15%
rent is extinguished, the diodes become reverse-biased and will fault voltages, with a single rectification period at fault initiation
not conduct until the converter’s switches are reactivated. In and at fault clearance.
this manner, the brake resistor behaves very differently from Fig. 8 shows the test results obtained when using the min-
a crowbar resistor. When the rotor current is extinguished, the imum threshold type controller for the dc brake chopper. The
mutual decay contribution to the stator currents is likewise ex- results demonstrate how stable control could not be achieved
tinguished. The stator flux can now only decay through its own when using this control method. The slower demagnetization
circuit, and the effective stator time constant is lengthened. This process of the machine in this case is reflected in these test
effect is reflected in the fault test results, where the turn-OFF results, thus showing that stable vector control could not be
period for the IGBT switches was necessarily elongated in order established immediately after the first peak of rotor current. No-
to be able to resume stable vector control. ticeably, the current controllers did not gain good control after
The slower demagnetization is reflected in the fault test re- the rectification period following fault initiation. Each release
sults. The control-delay method successfully managed the rotor led to substantial oscillations that triggered further rectification
currents and restricted the dc-link voltage to below 850 V in periods; six instances were recorded after fault initiation be-
the face of 15% fault voltages, with a single rectification period fore the currents began to settle, two of which lasted only three
at fault initiation and at fault clearance. The required rectifica- control cycles (600 μs).
tion periods were 26–29 ms. In each test, the fault clearance Noticeably, after each occasion of PWM control resumption,
rectification period preceded a 50% reactive power import that the rotor currents remained high, thereby rendering the control
suppressed the stator voltage by around 11%. The controller environment more difficult. Sharp spikes of rotor current were
required roughly 100 ms in each case to return to unity power produced as the current references pulled sharply to and from
factor. their saturation limits.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

PANNELL et al.: EVALUATION OF THE PERFORMANCE OF A DC-LINK BRAKE CHOPPER 7

V. CONCLUSION
The use of a dc brake chopper circuit as a DFIG FRT device
in response to grid faults was investigated in this paper using
a 7.5 kW experimental test facility. Using a simple, delayed
control method for the resumption of PWM control, the dc-link
brake was shown to successfully maintain the dc-link voltage
within safe limits, while potentially dangerous rotor overcur-
rents were allowed to conduct safely through (overrated) ro-
tor converter diodes. Power control was restored within about
80 ms of each disengagement of the rotor converter’s switches,
which occurred once for each voltage step of fault initiation and
clearance.
However, the periods of converter disengagement and the
length of time required to restore power control were each sig-
nificantly longer than what could be achieved with a rotor crow-
bar circuit. This is because the dc-link brake chopper fundamen-
tally does not assist the demagnetization process of the machine,
whereas the crowbar can shorten the rotor decay timescale. The
dc brake fully demagnetizes the rotor before forcing the charac-
teristic slow stator flux decay. The result is a delayed resumption
in control.
A more sophisticated minimum threshold type dc brake con-
troller in which the application of the dc chopper brake is re-
leased by a predefined rotor current threshold and feedback
control carefully restored to minimize the rectification period
was also developed and tested but failed to deliver satisfactory
results. The slower machine demagnetization meant that the
Fig. 9. Minimum threshold crowbar test results (shaded area indicates crowbar current controllers could not gain good control after the rectifi-
operation). cation period following fault initiation and clearance triggering
further multiple rectification periods.
As a stand-alone protection device, the dc brake chopper
A similar pattern emerged from voltage recovery where three could be relatively easily retrofitted to an existing DFIG con-
rectification periods were triggered by the rotor currents and the verter. The total unit costs may also be less than the crowbar:
power output became highly oscillatory. Each time the PWM the IGBT brake switch need only be rated for little more than
was disengaged, the PQ controllers were reset, which helps to 1 pu rotor current (determined by the maximum dc-link voltage
explain the extended duration of poor control through the fault and the brake resistor). Up-rating individual diodes is relatively
period and beyond. cheap but the requirement to house the nonstandard diodes may
The dc link experienced a rapid overvoltage on fault initiation add a significant overhead.
of 126 V in 12.8 ms. Above 810 V, the brake resistor was However, in terms of its overall electrical performance as a
engaged. After 13 ms, the transient overcurrents fell, the power stand-alone FRT device, the dc brake chopper offers inferior
dumped on the dc link fell away, and the dc-link brake chopped electrical characteristics when compared with crowbar based
the voltage back down to under 800 V. At fault clearance, in methods; particularly in terms of the rapidity with which full
a similar manner a peak of 862 V was experienced. Again the control may be restored. This is because the dc-link brake funda-
chopper returned the voltage below 800 V before the line-side mentally does not assist the demagnetization process of the ma-
converter, controlling power export via the recovery voltage, chine following a fault event, whereas the crowbar can shorten
restored a settled 750 V operational dc voltage within 300 ms the rotor decay timescales with the appropriate choice of crow-
of stator voltage recovery. bar resistance, thereby leading to much quicker resumption in
In contrast, the successful operation of the minimum thresh- control. A dc brake may nonetheless be considered valuable for
old controller when used for crowbar operation is demonstrated converters with low values of dc-link capacitance; these may be
in Fig. 9. The stable performance of the controller is reflected more sensitive to incoming overcurrents.
by the fact that the crowbar circuit was engaged only twice.
The first crowbar application, 1.2 ms after the fault, lasted for
13.0 ms. The second application, 6.6 ms after fault clearance, REFERENCES
lasted for 15.6 ms. During the first instance of crowbar applica-
[1] S. K. Salman and A. L. J. Teo, “Windmill modeling consideration and
tion, the dc-link voltage barely registered the event. The second factors influencing the stability of a grid-connected wind power-based
led to a safe excursion of only 30 V. The rotor currents flowing embedded generator,” IEEE Trans. Power Syst., vol. 18, no. 2, pp. 793–
through the converter remained below 2 pu for the whole test. 802, May 2003.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON ENERGY CONVERSION

[2] Z. Chen, Y. Hu, and F. Blaabjerg, “Stability improvement of induction Graham Pannell received the B.A. degree in physics
generator-based wind turbine systems,” IET Renewable Power Genera- from the University of Cambridge, Cambridge, UK,
tion, vol. 1, no. 1, pp. 81–93, Mar. 2007. in 2001, and the Ph.D. degree in engineering from
[3] S. Seman, J. Niiranen, S. Kanerva, A. Arkkio, and J. Saitz, “Performance Newcastle University, Newcastle upon Tyne, UK, in
study of a doubly fed wind-power induction generator under network 2008.
disturbances,” IEEE Trans. Energy Convers., vol. 21, no. 4, pp. 883–890, He is currently a Senior Engineer with AC Renew-
Dec. 2006. able Energy Systems Limited, Hertfordshire, UK. He
[4] O. Gomis-Bellmunt, A. Junyent-Ferre, A. Sumper, and J. Bergas-Jan, was a Research Engineer with Econnect Ltd. where
“Ride-through control of a doubly fed induction generator under un- he was involved in cost-effective improvements in
balanced voltage sags,” IEEE Trans. Energy Convers., vol. 23, no. 4, doubly fed induction generator performance.
pp. 1036–1045, Dec. 2008.
[5] J. Morren and S. W. H de Haan, “Short-circuit current of wind turbines
with doubly fed induction generator,” IEEE Trans. Energy Convers.,
vol. 22, no. 1, pp. 174–180, Mar. 2007.
[6] A. Causebrook, D. J. Atkinson, and A. G. Jack, “Fault ride-through of
large wind farms using series dynamic braking resistors,” IEEE Trans. Bashar Zahawi (M’96–SM’04) received the B.Sc.
Power Syst., vol. 22, no. 3, pp. 966–975, Aug. 2007. and Ph.D. degrees in electrical and electronic engi-
[7] J. Yang, J. E. Fletcher, and J. O’Reilly, “A series-dynamic-resistor-based neering from Newcastle University, Newcastle upon
converter protection scheme for doubly-fed induction generator during Tyne, UK, in 1983 and 1988, respectively.
various fault conditions,” IEEE Trans. Energy Convers., vol. 25, no. 2, From 1988 to 1993, he was a Design Engineer
pp. 422–432, Jun. 2010. with a UK manufacturer of large variable speed drives
[8] X. Yan, G. Venkataramanan, Y. Wang, Q. Dong, and B. Zhang, “Grid- and other power conversion equipment. In 1994, he
fault tolerant operation of a DFIG wind turbine generator using a pas- was appointed as a Lecturer in electrical engineering
sive resistance network,” IEEE Trans. Power Electron., vol. 26, no. 10, at the University of Manchester. In 2003, he joined
pp. 2896–2905, Oct. 2011. the School of Electrical and Electronic Engineering,
[9] Y. Ren and W. Zhang, “A novel control strategy of an active crowbar for Newcastle University, UK, as a Senior Lecturer. His
DFIG-based wind turbine during grid faults,” in Proc. IEEE Int. Conf. current research interests include small-scale generation, power conversion, and
Electric Machines Drives (IEMDC), Niagara Falls, ON, Canada, May the application of nonlinear dynamical methods to electrical circuits and sys-
2011, pp. 1137–1142. tems.
[10] G. Pannell, D. J. Atkinson, and B. Zahawi, “Minimum-threshold crowbar Dr Zahawi is a Chartered Electrical Engineer and the recipient of the Cromp-
for a fault-ride-through grid-code-compliant DFIG wind turbine,” IEEE ton Premium awarded by the Institution of Electrical Engineers (IEE) and the
Trans. Energy Convers., vol. 25, no. 3, pp. 750–759, Sep. 2010. Denny Medal awarded by the Institute of Marine Engineering, Science & Tech-
[11] C. Wessels, F. Gebhardt, and F. W. Fuchs, “Fault ride-through of a DFIG nology (IMarEST).
wind turbine using a dynamic voltage restorer during symmetrical and
asymmetrical grid faults,” IEEE Trans. Power Electron., vol. 26, no. 3,
pp. 807–815, Mar. 2011.
[12] J. P. da Costa, H. Pinheiro, T. Degner, and G. Arnold, “Robust controller
for DFIGs of grid-connected wind turbines,” IEEE Trans. Ind. Electron., David J. Atkinson received the B.Sc. in electrical and
vol. 58, no. 9, pp. 4023–4038, Sep. 2011. electronic engineering from Sunderland Polytechnic,
[13] I. Erlich, H. Wrede, and C. Feltes, “Dynamic behaviour of DFIG-based Sunderland, UK, in 1978, and the Ph.D. degree from
wind turbines during grid faults,” presented at IEEE Power Convers. Conf. Newcastle University, Newcastle upon Tyne, UK, in
(PCC 2007), Nagoya, Japan. 1991.
[14] K. E. Okedu, S. M. Muyeen, R. Takahashi, and J. Tamura, “Wind farms In 1987, he joined Newcastle University, where he
fault ride through using DFIG with new protection scheme,” IEEE Trans. is currently a Senior Lecturer in the Power Electron-
Sustainable Energy, vol. 3, no. 2, pp. 242–254, Apr. 2012. ics, Drives and Machines Research Group, School of
[15] B. Gong, D. Xu, and B. Wu, “Cost effective method for DFIG fault Electrical and Electronic Engineering. For 17 years,
ride-through during symmetrical voltage dip,” in Proc. 36th Annu. IEEE he was with NEI Reyrolle Ltd and British Gas Corpo-
Ind. Electron. Soc. Conf. (IECON 2010), Phoenix, AZ, USA, Nov. 2010, ration. His current research interests include control
pp. 3269–3274. of power electronics systems including electric drives and converters.
[16] J. Shi, Y. Tang, Y. Xia, L. Ren, and J. Li, “SMES based excitation system Dr Atkinson is a Chartered Electrical Engineer.
for doubly-fed induction generator in wind power application,” IEEE
Trans. Appl. Supercond., vol. 21, no. 3, pp. 1105–1108, Jun. 2011.
[17] W. Guo, L. Xiao, and S. Dai, “Enhancing low-voltage ride-through ca-
pability and smoothing output power of DFIG with a superconducting
fault-current limiter–magnetic energy storage system,” IEEE Trans. En- Petros Missailidis received the B.S degree in me-
ergy Convers., vol. 27, no. 2, pp. 277–295, Jun. 2012. chanical engineering from Aristotelian University of
[18] T. Kawady, C. Feltes, I. Erlich, and A. I. Taalab, “Protection system be- Thessaloniki, Thessaloniki, Greece, in 1986, the M.S.
havior of DFIG based wind farms for grid-faults with practical considera- degree in mechanical engineering from the University
tions,” in Proc. IEEE Power Energy Soc. General Meeting, Minneapolis, of Buffalo, NY, USA, in 1990, the M.S. degrees in
Minnesota, USA, Jul. 2010, pp. 1–6. electrical and electronic engineering and biomedical
[19] T. Kawady, H. Shaaban, and A. El-Sherif, “Investigation of grid-support engineering from Rensselaer Polytechnic Institute,
capabilities of doubly fed induction generators during grid faults,” in Proc. Troy, NY, in 1993, the M.S. degree in mechanical
IET Conf. Renewable Power Generation (RPG 2011), Edinburgh, U.K., engineering from the same institute in 1995, and the
Sep., pp. 1–7. Ph.D. degree in mechanical engineering from Rens-
[20] D. J. Atkinson, G. Pannell, W. Cao, B. Zahawi, T. Abeyasekera, and selaer Polytechnic Institute, in 2000.
M. Jovanovic, “A doubly-fed induction generator test facility for grid He was with ComHouse Wireless Inc., M/A-COM Inc., General Electric
fault ride-through analysis,” IEEE Instrum. Meas. Mag., vol. 15, no. 6, Schenectady NY, Erie County Medical Centre, Buffalo, NY, and the University
pp. 20–27, Dec. 2012. of Toronto. He is currently with the School of Electrical and Electronics Engi-
[21] G. Pannell, D. J. Atkinson, and B. Zahawi, “Analytical study of grid-fault neering, Newcastle University, Newcastle upon Tyne, UK, where he is involved
response of wind turbine doubly fed induction generator,” IEEE Trans. in conducting research in the areas of system identification, nonlinear control,
Energy Convers., vol. 25, no. 4, pp. 1081–1091, Dec. 2010. and control of electric drives.