Downloaded from orbit.dtu.

dk on: Nov 05, 2024

Transient Studies in Large Offshore Wind Farms, Tak-ing Into Account Network/Circuit
Breaker Interaction

Glasdam, Jakob; Bak, Claus Leth; Hjerrild, Jesper; Arana Aristi, Iván

Published in:
Proceedings of the 10th International Workshop on Large-Scale Integration of Wind Power into Power Systems
as well as Transmission Networks for Offshore Wind Farms

Publication date:
2011

Document Version
Publisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):
Glasdam, J., Bak, C. L., Hjerrild, J., & Arana Aristi, I. (2011). Transient Studies in Large Offshore Wind Farms,
Tak-ing Into Account Network/Circuit Breaker Interaction. In Proceedings of the 10th International Workshop on
Large-Scale Integration of Wind Power into Power Systems as well as Transmission Networks for Offshore Wind
Farms Energynautics GmbH.

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

 Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
 You may not further distribute the material or use it for any profit-making activity or commercial gain
 You may freely distribute the URL identifying the publication in the public portal

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
 Transient Studies in Large Offshore Wind Farms,
Taking Into Account Network/Circuit
Breaker Interaction
Jakob Glasdam, Claus Leth Bak, Jesper Hjerrild, Ivan Arana

Abstract—Switching overvoltages (SOV) are considered a pos- operation. The prestrike is a consequence of the decreasing
sible source of experienced component failures in existing off- contact gap distance, which facilitates the formation of a low
shore wind farms (OWFs). The inclusion of sufficiently accurate impedance vacuum arc before galvanic contact is established.
and validated models of the main components in the OWF in
the simulation tool is therefore an important issue in order to Because of the working principle of the VCB, there exists a
ensure reliable switching operations. high possibility that the high frequency (HF) inrush current
Transient measurement results in an OWF are compared is interrupted at its zero crossing. Depending on the voltage
with simulation results in PSCAD EMTDC and DigSILENT impressed on the contacts, multiple prestrikes might occur.
Power Factory, where a user-defined model of the vacuum The occurrence of multiple prestrikes is a complex matter
circuit breaker (VCB) is included, capable of simulating multiple
prestrikes during the closing operation. and difficult to predict, as the generated voltages and currents
An analysis of the switching transients that might occur in depend on many factors such as the dielectric and current
an OWF will be made on basis of the validated model and the interruption properties of the VCB. Furthermore, the multiple
importance of the inclusion of a sufficient accurate representation prestrikes depend on the surge impedances of the surrounding
of the VCB in the simulation tool will be described. network, pole scatter, point on wave of closing and so on. This
The inclusion of the VCB model in PSCAD greatly improves
the simulation results, whereas little improvement is found in
sets up a demand for the inclusion of a sufficiently accurate
DigSILENT. VCB representation in the simulation tool, capable of taking
into account network/circuit breaker interaction during the
Index Terms—Cable modeling, circuit breaker modeling,
DigSILENT Power Factory, model validation, transient studies closing operation. A user-defined representation of the VCB
in offshore wind farms, PSCAD. has been implemented in PSCAD EMTDC and DigSILENT
Power Factory in [7] and the usability of the models will be
discussed in section III and the proposed model will be used
I. I NTRODUCTION in order to investigate the possible occurring SOV in OWFs
in section IV.
I N OWF applications, the consequences of component fail-
ure are more severe compared to land based wind farms,
due to higher repair costs and lost revenue. [1] Switching over-
II. S YSTEM D ESCRIPTION
voltages are a possible cause of component failure observed in
Horns Rev OWF 1 as well as in Middelgrunden OWF. [2] Sim- In this paper the Nysted OWF (NOWF) is taking into con-
ulations are widely used to identify the overvoltages that might sideration as an example. A transient measurement campaign
occur in the OWF due to faults and switching operations and has previously been conducted in NOWF, where GPS syn-
in order to verify the design decisions. [1] Validation of OWF chronized, HF measuring systems were installed at different
component models is therefore an important issue in order to locations within NOWF as indicated in Fig. 1.
ensure reliable switching studies. It has been shown in [3], [4],
that insufficient representation of the radial circuit breaker in
H
the simulation tool is the main contributor to discrepancies External
network
between measurement and simulation results for radial ener- G
gization in OWFs. The vacuum circuit breaker (VCB) is the
F
preferred choice as the radial circuit breaker in OWFs due to
its low maintenance requirement and long operation life. [5], 33/33/132 kV
E

[6] Due to the dielectric properties of vacuum, a number of D

so-called prestrikes are almost inevitable during the closing
C
J.G is with DONG Energy Power A/S, Kraftværksvej 53, Skærbæk, 7000
Fredericia, Denmark (e-mail: [email protected]) B
Measurement
C.L.B is with Institute of Energy Technology, Aalborg University, Pontop- point A
pidanstræde 101, 9220 Aalborg, Denmark (e-mail: [email protected])
J.H is with DONG Energy Power A/S, Kraftværksvej 53, Skærbæk, 7000 A1 A9
Fredericia, Denmark (e-mail: [email protected]) Fig. 1. Schematic of Nysted Offshore Wind Farm. The locations of the
I.A is with DONG Energy Power A/S, Kraftværksvej 53, Skærbæk, 7000 measurement system used during the transient measurement campaign are
Fredericia, Denmark (e-mail: [email protected]) also indicated.
 NOWF consists of 72 × 2.3 MW rated Siemens wind properties (CQ) in Aµs−1 of the VCB before the contacts
turbines (WTs). The WTs are arranged in a parallelogram are making.
formed by eight strings or radials. Nine WTs are connected to
each radial. The cable collecting grid is operated at 33 kV and III. VACUUM C IRCUIT B REAKER M ODEL VALIDATION
the voltage is increased to 132 kV through the 90/90/180 MVA
park transformer, located offshore. Each radial is connected to There is no information available for the VCBs installed in
one of the park transformers two medium voltage (MV) bus NOWF, hence the input parameters to the VCB model have
bars through a VCB. been adjusted in each phase in order to best fit the simulation
The transient measurement is done by energizing radial A, results with the measurement results. This is further justified
when all other radial are energized. The WTs were not under by the fact that the parameters are stochastic by nature as the
production during the recording of the transients. closing operation can begin with equal likelihood throughout
one period of the 50 Hz voltage sine wave and rdds can
range from 25 to 100 Vµs−1 . [12], [13] It was observed in
A. System Modeling the measurement results in NOWF, that the current is being
A model of NOWF has been implemented in DigSILENT interrupted each time it crosses zero, hence the value of CQ
and PSCAD based on the information available from the OWF has been set to 600 Aµs−1 , as this is the highest measured
as-built documentation. A schematic of the model can be seen value for this parameter. [14]
in Fig. 2. In Fig. 3 is shown the comparison of measurement (red
curves) and PSCAD simulation results with the build-in switch
(blue) and with the detailed VCB representation (green) for
Lgrid Rgrid
phase B voltages and currents at the platform and the voltages
D at A9.
C
B
Root D
A 2.0
UB,meas
UB,vcb
Voltage [P.U]

1.5 UB
Fig. 2. Schematic of the NOWF network model used in the analysis.
1.0
Only the radials connected to the same bus bar as radial A
(radial B to D) are included in the model and are each repre- 0.5
sented by one long cable. The radials in the cable collecting 0.0
grid are modeled based on geometry and material properties 200 300 400 500 600 700
of the cable according to the guidelines in [8], [9]. However, 1.0
IB,meas
some modifications were found necessary in the cable input 0.8 IB,vcb
Current [kA]

0.6 IB
parameters, as the simulated wave velocity was found higher
0.4
in both simulation tools compared to the measurement. This 0.2
is done by increasing the value of the relative permittivity of 0.0
the main insulation. The SOV were found to be insensitive to -0.2
this parameter, whereas the magnitude of the inrush current -0.4
increased. Furthermore, an unexpected high coupling between 200 300 400 500 600 700
the phases was found in DigSILENT, which was limited by 2.0
UB,meas
separating the phase conductors.
Voltage [P.U]

1.5 UB,vcb
Furthermore, only the transformers on radial A are included UB
1.0
in the model using the available standard transformer models
in the simulation tools. The export cable system, consisting 0.5
of a submarine cable an a land based cable, is modeled using 0.0
the lumped π-model in both simulation tools. The external 200 300 400 500 600 700
network is represented by its Thevenin equivalent. Time [µs]
Two model representations of the VCB are considered: Fig. 3. Comparison of measurement and PSCAD simulation results for phase
(i) Build-in switch, which performs one closing operation B voltages and currents at the platform (top and middle plot, respectively) and
timed at the instant when the first prestrike in each phase voltages at A9, when the build-in switch and the VCB model are used.
occurs.
(ii) User-defined VCB model, capable of replicating mul- It is possible to the see that the measured current (IB,meas )
tiple prestrikes. The model has some similarities with the is being interrupted at its zero crossing and therefore appears
existing user-defined model in [10], [11] and is described in as ’half waves’, which is an indication that multiple prestrikes
more details in [7]. Three input parameters are necessary in are occurring. The current interruption causes an energy trans-
each phase: The instant of the closing start time (t0 ), rate fer from the magnetic field associated with the propagating
of decay of the dielectric strength (rdds) in Vµs−1 of the current wave to the electric field associated with the propaga-
decreasing contact gap distance and the current quenching ting voltage wave in order to comply with energy conservation.
The result is an increase in UB,meas at the platform, which 0.0
UA,3
propagates toward the receiving end of radial A, where it is UA,2

Voltage [P.U]
superimposed on the initial wave. IB in the PSCAD simulation -0.5 UA,1
using the build-in switch is not being interrupted at its zero
crossing and hence the voltage build up at the platform as -1.0
well as at A9 is not replicated. IB,vcb is being interrupted at
its zero crossing in the PSCAD simulation using the detailed -1.5
VCB representation. A good agreement can therefore be seen 100 200 300 400 500 600 700 800 900
between the measured and simulated waveforms at the two 0.8
IA,3
locations for the initial part of the transient. It has not been IA,2

Current [kA]
0.4
possible to further improve the simulation results after the IA,1
0.0
initial part of the transient. However, this is considered to be
of little importance as the generated SOV are highest in the -0.4
initial part of the transient. From Fig. 3 it is evident that a -0.8
sufficiently accurate representation of the VCB is required in 100 200 300 400 500 600 700 800 900
the simulation tool in order to replicate the multiple prestrikes Time [µs]
inherent to the VCB technology and hence to OWFs, due to
Fig. 5. Simulated platform voltages and currents for the three study cases
the widespread use of the VCB. for phase A. Label numbering corresponds to the number or radials connected
A similar improvement in the simulation results by the when radial A is being energized.
inclusion of the detailed VCB representation has not been
achieved in DigSILENT. This is due to an unexpected high
rate of simulated cable discharging after the current is being A. Energization of Radial A
interrupted. This is illustrated in Fig. 4, where the measured In Fig. 5 are shown the results of three simulations for phase
voltage for phase B (UB,meas ) at A1 is compared with DigSI- A at the platform, where the number of radials connected to
LENT and PSCAD simulation results (UB,dig and UB,ps , the same bus bar as radial A (radial B to D, see Fig. 1) are
respectively). A good agreement between DigSILENT and varied. The subscript number in the figure denotes the number
measurement results can be seen in Fig. 4 until t ≈ 375 of radials connected during the energization of radial A (i.e
µs, where the current is being interrupted. The voltage on the subscript 3 is for the situation, where radial B, C and D
the isolated cable now starts decreasing and the cable is are connected, 2 is for the situation, where radial C and D are
being completely discharged within a few ms, which is not connected and 1 is when only radial D is connected).
in agreement with real life cable discharging. [7] This high The magnitude of the generated SOV in Fig. 5 is increased
rate of cable discharging has been found for all cable models when more radials are connected. This is due to the impedance
available in DigSILENT. [7] It can therefore be concluded that (Zbus ) at the bus bar side of the VCB is lowered when the
DigSILENT is unsuitable for energization studies in OWFs as number of radials connected is increased. The voltage drop on
well as for cable de-energization studies in general. Zbus is therefore lowered when radial A is being energized.
The peak of the simulated inrush current in Fig. 5 is also
1.5 increased when more radials are connected. The cables in
UB,meas NOWF are equipped with XLPE insulation with a high value
UB,dig
UB,ps of the relative permittivity (r ), hence it is possible for sake
Voltage [P.U]

1.0 of simplicity to represent the cable as a capacitor (C), where

the voltage/current relationship is given by (1). [15]
0.5 dU
I=C·
[A] (1)
dt
0.0 A drop in the voltage at the platform can be seen prior to the
200 300 400 500 600 700
current interruption at t ≈ 250 µs, which is also occurring in
Time [µs]
the current traces due to the linear relationship between voltage
Fig. 4. Comparison of measured phase B voltage at A1 with PSCAD and and current given by the characteristic impedance (Zc ) of the
DigSILENT simulation results using the detailed VCB representation in both cable during the transient.
simulation tools.
The voltage drop at the bus bar when radial A is being
energized propagates on the radials connected to the same
bus bar. The cables on each radial are identical, hence the
IV. T RANSIENT S TUDY IN NOWF difference in the traveling times only depends on the length
The radials in an OWF are supposed to be energized ran- of the radials. The difference in the lengths of radial A to D
domly. [5] Different scenarios are investigated in the following depends on the location of the first WT on each radial relative
in order to give an insight into the SOV that might occur within to the platform, where D is the shortest and A is the longest
an OWF using the validated PSCAD model of NOWF, where radial. This explains the three stepwise decrease in UA,3 in
the network/circuit breaker interaction is included. Fig. 5 before the wave propagating on radial A reappears at the
platform and the current is interrupted. Similarly, two stepwise The frequency of the transients in the case when no radials
decrease in UA,2 are simulated for radial C and D connected are connected is given in (3), where it is evident that Lgrid
and one drop is simulated for UA,1 , when only radial D is will lower the transient frequency.
connected. The total voltage drop due to the interaction with
the other radials is highest when only radial D is connected, 1
which is due to the higher drop in the bus bar voltage at the f= p [Hz] (3)
2π (Lgrid + LA )CA
instant, when radial A is energized, as explained in the above.
As previously explained, the current interruption causes an The simulation results for the case with no radials connected
increase in the platform voltage (Uplat , on the radial side of are shown in Fig. 8, where f = 3.8 kHz, which is 28 % lower
the VCB), which is simulated in all three cases. The magnitude compared to the situation, when other radials are connected.
of the voltage increase is highest for UA,3 and lowest for UA,1 ,
which is due to the different voltage drops as explained in the 0.0
UA,0
above. The frequency of the transients is f = 5.3 kHz for all

Voltage [P.U]
cases, indicating that the energy into radial A is transferred -0.5
from the other radials connected. This is the so-called back- -1.0
to-back energization, which is equivalent to the connection of a
capacitor bank to an already energized capacitor bank. [16] A -1.5
highly simplified schematic of the situation, when energizing -2.0
radial A to the bus bar, where an already energized cable is 100 200 300 400 500 600 700 800 900
connected. Lgrid is the Thevenin inductance of the external
network. The series impedance of the cable is low, hence the 0.4
capacitances CA and CB are almost in parallel and an energy Current [kA]
0.0
transfer from cable B to cable A is taking place.
-0.4
LB -0.8 IA,0
100 200 300 400 500 600 700 800 900
t0 CB Time [µs]
Lgrid LA
Us
Fig. 8. Simulated platform voltages and currents for phase A when no radials
I are connected.
CA

Fig. 6. Simplified schematic of the connection of cable A to the already

energized cable B, where losses are neglected. B. Comparison of Different OWF Configurations
The influence of the OWF configuration on the SOV will be
The frequency of the inrush current in Fig. 6 can be investigated in the following, where three cases are considered:
calculated as in (2). [7], [15] (i) Radial A energization, (ii) Radial D energization and
(iii) Radial A energization, where the length of radial A is
1 doubled and the lengths of radials B to D are increased in
f= √ [Hz] order to maintain the radial inter spacing. Three radials are
2π LA CA
1 (2) connected in each case. Case (i) is the same as in the previous
≈ √ section. Case (iii) is the hypothetical case, where the step up
4lcab µr r
transformer is located onshore and hence the radial lengths are
where µr and r are the relative permeability and permit- increased. In Fig. 9 are shown the initial part of the simulated
tivity of the cable, respectively. lcab is the length of the cable. phase A voltages and currents at the sending end of the radial
Equation (2) indicates that the frequency of the transients when being energized.
other cables depends only on the length and materials of the The initial parts of the transients in Fig. 9 are identical
cable. in the three cases, which is due to the identical types and
The situation is different when no cables are connected to number of cables connected to the bus bar in the three cases.
the bus bar, as cable A will interact with the external network Differences can be seen for the traveling wave times and hence
as shown in Fig. 7. the transient frequency, which is due to the different length
of the radial being energized in the three cases. The three
stepwise decrease in the waveforms for case (i) (red curves)
t0
due to interaction with the other radials connected are also
Lgrid LA occurring in case (iii) (blue curves), whereas they are more
Us attenuated, which is due to the longer traveling distances in this
CA
case. The radial interaction is not occurring in case (ii), where
Fig. 7. Simplified schematic of the connection of cable A, when no cables radial D is being energized. Because of the shorter length of
are connected. radial D compared to the other radials, the wave propagating
on this radial will reappear at the platform, where the current 1.8
1.7 OV, vcb
is interrupted before the waves on the other radials reappear OV
1.6 Ratio [-]

Voltage [P.U]
at the platform. 1.5
1.4
1.3
C. Point on Wave Influence on the Generated SOV 1.2
The most important parameter on the generated SOV is the 1.1
1.0
point on voltage wave (POW). The instant of the contact start 0 2 4 6 8 10 12 14 16 18 20
closing time is uniformly distributed over one period of the 1.9
1.8 OV, vcb
fundamental frequency, which results in an infinite number 1.7 OV

Voltage [P.U]
of switching combinations. Different methods to capture the 1.6 Ratio [-]
1.5
highest possible SOV do exist, namely the systematic and the 1.4
1.3
statistical switch. [17] However, none of the existing methods 1.2
are taking into account the possibility of the occurrence 1.1
1.0
of multiple prestrikes, inherent to the VCB technology. A 0.9
0 2 4 6 8 10 12 14 16 18 20
design procedure for a statistical switch, being able to take
Instant of closing [ms]
into account network/VCB interaction is proposed in [7] and
further work is expected to be done in this area. Fig. 10. Simulated case SOV using the build-in switch and the detailed VCB
description. Top plot at the platform and bottom plot at A9.
A number of simulations have been carried out in order to
investigate the influence of the POW on the generated SOV.
This is done by systematically varying the instant t0 , when the has been found that the multiple prestrikes are occurring for
closing operation begins by an increment of 1 ms over one a negative value of dU /dt for one of the phases during the
period of the fundamental frequency. The simulations have closing operation. [7] The author therefore proposes that t0
been done for both VCB representations investigated in the should be confined to an interval of 10 ms.
current paper in order to compare the simulation results. The It can be seen from the top plot in Fig. 10, that there are
simulations are done by energizing radial A, when radial B to large differences in the simulated case OV at the platform
D are connected. The highest case SOV has been captured for using the two VCB representations, the highest difference
each simulation at the platform and at A9 as shown in Fig. being 60 %. This therefore stresses out the importance of
10. The time axis in Fig. 10 indicates the instant, when the the inclusion of a suffiently accurate VCB representation in
closing operation begins in the simulation. the simulation tool, as it was found from Fig. 3 that the
A half wave symmetry of the highest case SOV is evident detailed VCB representation greatly improves the accuracy of
for both VCB representations in Fig. 10. It is therefore possible the simulation results. Little difference can be observed in the
to confine t0 in an interval of 10 ms in a 50 Hz system in order simulation results at A9 in the bottom plot in Fig. 10, which is
to reduce the computational time. It is suggested in [18], that also in good agreement with Fig. 3. A similar agreement is also
t0 can be confined to the peak portion of the voltage wave and found at A1. It can therefore be concluded that a sufficiently
for positive values of dU /dt of the voltage wave, and hence accurate VCB representation is required in the simulation tool
further reduce the computational time. However, this is not in order to predict the SOV at the platform, whereas the build-
in agreement with the measurement results in NOWF, as it in switch can be used if only the SOV at the wind turbines
are investigated.
From Fig. 10 it is furthermore evident that the selected
0.0
UA,c1 resolution of 1 ms for t0 is too low, as there are large
-0.4 UA,c2 differences between some of the successive simulation results
Voltage [P.U]

UA,c3
-0.8 at both locations. This is true for both VCB representations.
-1.2 This finding implies the importance of the inclusion of a VCB
representation, capable of taking into account the randomness
-1.6
of the closing instant, as described in the above.
-2.0
150 200 250 300 350 400 450 500 550
0.0 V. C ONCLUSION
Current [kA]

-0.2 Transient measurement results in NOWF have been com-

-0.4 pared with simulation results in this paper with emphasis
-0.6 IA,c1 on the radial vacuum circuit breaker, which has previously
IA,c2 been found to be the main limitation in the accuracy of
-0.8 IA,c3
simulation results in OWFs. A user-defined representation of
150 200 250 300 350 400 450 500 550 the VCB has been implemented in PSCAD EMTDC and
Time [µs] DigSILENT Power Factory and the simulation results have
Fig. 9. Simulated sending end voltages and currents for the three study cases been compared with transient measurement results in NOWF,
for phase A. where one of the radials is being energized. The inclusion
of the VCB in PSCAD greatly improves the accuracy of the [10] J. Helmer and M. Lindmayer, “Mathematical modeling of the high fre-
simulation results, whereas little improvement was found in quency behavior of vacuum interrupters and comparison with measured
transients in power systems,” in Discharges and Electrical Insulation in
the DigSILENT simulation results, which was found to be Vacuum, 1996. Proceedings. ISDEIV., XVIIth International Symposium
due to an unexpected high rate of cable discharging after on, vol. 1, pp. 323 –331 vol.1, jul 1996.
current interruption. DigSILENT is therefore found unsuitable [11] B. K. Rao and G. Gajjar, “Development and Application of Vacuum
Circuit Breaker Model in Electromagnetic Transient Simulation,” Power
for energization studies in OWFs as well as for cable de- India Conference, IEEE, 2006.
energization studies in general. [12] J. Cornick and A. Tleis, “Computer simulation of three-phase prestriking
A transient analysis has been carried out in order to in- transients in cable-connected motor systems,” Electric Power Applica-
tions, IEE Proceedings B, vol. 138, pp. 105 –114, May 1991.
vestigate the possible SOV, that might occur in large OWFs [13] T. Abdulahovic, “Analysis of High-Frequency Electrical Transients in
during radial energization. It has been found that a sufficiently Offshore Wind Parks,” Master’s thesis, Dept. of Energy and Environ-
accurate VCB representation is required in the simulation tool ment. Division of Electric Power Engineering Chalmers University of
Technology, 2009.
in order to predict the SOV at the platform, whereas the build- [14] S. Wong, L. Snider, and E. Lo, “Overvoltages and reignition behavior
in switch can be used if only the SOV at the wind turbines of vacuum circuit breaker,” Conference on Advances in Power System
are required. Control, Operation and Management, 2003. ASDCOM 2003. Sixth
International (Conf. Publ. No. 497), pp. 653–658, 2003.
The magnitude of the SOV is found to increases with in- [15] A. Greenwood, Electrical transients in power systems. John Wiley &
creased number of radials connected. The transient frequency Sons Ltd, 2 ed., 1991.
is found to be insensitive to the number of connected radials, [16] F. Faria da Silva, C. Bak, and M. Hansen, Back-to-Back Energization
of a 60kV Cable Network - Inrush Currents Phenomenon. IEEE, 2010.
as long as at least one radial is connected. This indicates the [17] J. Martinez, R. Natarajan, and E. Camm, “Comparison of statistical
little influence of the external network. The accuracy of the switching results using gaussian, uniform and systematic switching
external network becomes important with no radials connected approaches,” in Power Engineering Society Summer Meeting, 2000.
IEEE, 2000.
as the frequency is lowered by 28 % compared to the situations [18] IEEE PES Switching Transients Task Force, Task Force Report: Mod-
with one or more radials are connected. This is due to the eling Guidelines for Switching Transients. 1997.
interaction with the short circuit inductance of the external
Jakob Glasdam was born in 1984. He received
network. B.Sc. and M.Sc. degrees in electrical power engi-
The analysis has shown that all possible scenarios should neering from Southern University of Denmark and
be investigated in the design phase of future OWFs. from Aalborg University, respectively.
He is currently an Industrial PhD student in coopera-
tion with DONG Energy and Aalborg University.
ACKNOWLEDGMENT The main direction of his research is related to
The transient measurement results used in this study were harmonics in large offshore wind farms.
obtained under a project titled ’Voltage conditions and tran-
sient phenomena in medium voltage grids of modern wind
farms’, contract 2005-2-6345, supported by the Danish TSO Claus Leth Bak was born in Djursland, Denmark,
in 1965. He received B. Sc. in Electrical Power
Energinet.dk. Engineering from the engineering college in Århus
in 1992, he received M.Sc. in Electrical Power
R EFERENCES Engineering in 1994. He is an Associate Professor
at Aalborg University with experience on high vol-
[1] L. Christensen, M. Ulletved, P. Sørensen, T. Sørensen, T. Olsen, tage engineering, relay protection for transmission
H. Nielsen, P. Sørensen, and O. Holmstrøm, “GPS synchronized high systems and substation automation and dynamic
voltage measuring system,” Nordic Wind Power Conference, Nov. 2007. analysis (PSCAD/EMTDC) of large power systems.
[2] W. Sweet, “Danish wind turbines take unfortunate turn,” Spectrum,
IEEE, vol. 41, no. 11, pp. 30, 34, 2004.
[3] P. Sørensen, A. D. Hansen, T. Sørensen, C. S. Nielsen, H. K. Nielsen, Jesper Hjerrild was born in 1971. He received the
L. Christensen, and M. Ulletved, “Switching transients in wind farm M.Sc. and Ph.D. degrees in electrical engineering
grids,” European Wind Energy Conference and Exhibition, 2007. from the Technical University of Denmark, Lyngby,
[4] I. Arana, J. Holbøll, T. Sørensen, A. H. Nielsen, P. Sørensen, and in 1999 and 2002, respectively. Currently he has
O. Holmstrøm, “Comparison of Measured Transient Overvoltages in the been employed at Dong Energy. His main technical
Collection Grid of Nysted Offshore Wind Farm with EMT Simulations,” interest is electrical power systems in general, in-
International Conference on Power Systems Transients (IPST2009), volving a variety of technical disciplines including
2009. modelling of power system including wind power
[5] L. Liljestrand, A. Sannino, H. Breder, and S. Thorburn, “Transients in and power system control, stability and harmonics.
collection grids of large offshore wind parks,” Wind Energy, vol. 11, Furthermore, he also works with designing of the
pp. 45–61, 2008. wind farm.
[6] I. Arana, L. Kocewiak, J. Holbøll, C. Bak, A. Nielsen, A. Jensen, From 2002 until 2004 Jesper Hjerrild was employed at DEFU (The Associa-
J. Hjerrild, and T. Sørensen, How to improve the design of the electrical tion of Danish Energy Companies R&D).
system in future wind power plants. Technical University of Denmark,
2009. Iván Arana was born in Mexico City in Mexico,
[7] J. Glasdam, “Development, validation and application of a vacuum 1983. He received the bachelor degree from ITESM,
circuit breaker model for time domain analysis in large offshore wind Mexico in 2005 and MSc. degree from DTU, Den-
farms,” Master’s thesis, Institute of Energy Technology, Aalborg Uni- mark in 2008. He is now an Industrial PhD student
versity, Denmark, 2011. at the CET in cooperation with DONG Energy and
[8] B. Gustavsen, J. Martinez, and D. Durbak, “Parameter determination for Siemens Wind Power.
modeling system transients-part ii: Insulated cables,” Power Delivery,
IEEE Transactions on, vol. 20, pp. 2045 – 2050, july 2005.
[9] B. Gustavsen, “Panel session on data for modeling system transients
insulated cables,” in Power Engineering Society Winter Meeting, 2001.
IEEE, vol. 2, pp. 718–723, 2001.