An experimental investigation of switching transients in a wind-collection grid scale model in a cable system laboratory

Abstract

With the background of experience from field investigations in cable networks, multiple pre-strikes and re-ignitions are known to occur during switching operations, causing high frequency, steep fronted transient voltages and currents. Off-shore collection grids consist of vast amount of cables, with different length and connection points, which have low surge impedance compared with overhead lines. Consequently multiple pre-strikes and re-ignitions of switching apparatus with cables can cause transient overvoltages with higher time-derivatives than with overhead lines. The validation test circuits specified in the IEC standards developed for impulses due to fulmination on overhead lines do not consider the conditions of large cable grid and repetitive overvoltages due to switching. With the increase of cable grids and particularly wind parks, it has become important to characterize a collection grid cable system and the related transients during manoeuvres with different switching apparatus. The main contribution of this work is the experimental verification of the transient overvoltage phenomena in cable systems. An attempt is made to characterize this phenomenon in terms of number of re-ignitions, the magnitude and the rate of voltage step.

Full text

Page 1

? ? ? ??????????
??????????????????
?????????
?????????????????????????????????????????????????????????????????????????????????
?
??????? !"?
?

??#???????$????????? ??????%???? !"???? ??
AN EXPERIMENTAL INVESTIGATION OF SWITCHING TRANSIENTS IN A WIND-
COLLECTION GRID SCALE MODEL IN A CABLE SYSTEM LABORATORY


Muhamad REZA, Henrik BREDER,
Lars LILJESTRAND, Ambra SANNINO
ABB AB, Sweden
{muhamad.reza, henrik.breder,
lars.liljestrand, ambra.sannino}@se.abb.com
Tarik ABDULAHOVIC, Torbjörn THIRINGER
Chalmers University of Technology, Sweden
{tarik.abdulahovic, torbjorn.thiringer}@chalmers.se

ABSTRACT
With the background of experience from field investigations
in cable networks, multiple pre-strikes and re-ignitions are
known to occur during switching operations, causing high
frequency, steep fronted transient voltages and currents.
Off-shore collection grids consist of vast amount of cables,
with different length and connection points, which have low
surge impedance compared with overhead lines.
Consequently multiple pre-strikes and re-ignitions of
switching apparatus with cables can cause transient over-
voltages with higher time-derivatives than with overhead
lines. The validation test circuits specified in the IEC
standards developed for impulses due to fulmination on
overhead lines do not consider the conditions of large cable
grid and repetitive overvoltages due to switching. With the
increase of cable grids and particularly wind parks, it has
become important to characterize a collection grid cable
system and the related transients during manoeuvres with
different switching apparatus.

The main contribution of this work is the experimental
verification of the transient overvoltage phenomena in
cable systems. An attempt is made to characterize this
phenomenon in terms of number of re-ignitions, the
magnitude and the rate of voltage step.
INTRODUCTION
In overhead line (OHL) systems the transient overvoltages
are often related to overvoltages caused by lightning and
therefore electrical apparatus is currently tested with a
standard lightning impulse – lightning impulse withstand
level (LIWL) – of 1.2/50 µs surge. Surge arresters are used
as a common protection for these transient overvoltages.

In a cable system like an offshore collection grid, transient
overvoltages with multiple pre-strikes and re-ignitions with
high time-derivatives can result from an interaction of
different electrical apparatuses such as transformers, vast
amount of cables – which have a low surge impedance
relative to OHLs – with different length and connection
points and switching devices, for example during circuit
breaker operations in cable system: closing and opening
breakers and switches [1-2]. There, the cables act as
transmission lines for propagating waves with voltages
appearing across the switching contacts or current flowing
through the contacts.

Switching transients in three-phase systems can be
extremely complex [3-4]. However, generally while opening
and closing electrical contacts of any apparatus there is a
time interval when the voltage withstand in the apparatus is
low enough to allow re-ignitions during closing operations.
ABB WIND CABLE LABORATORY
Background
Characterization of the transients occurring due to switching
operation can be made by computer simulation in an
electromagnetic transient program provided that the models
are verified. Verification can be obtained by field
measurements or laboratory experiments. In the former,
real-life conditions are recorded, however, it is difficult to
record all necessary parameters to reproduce the exact
conditions, and it is very difficult to test specific conditions
due to commercial aspects (insurance and warranties) and
risks involved. In laboratory environment, on the other
hand, it may be difficult to realize a down-scaled test circuit
that still is representative of the actual real-life conditions.
However, if by realizing such test circuit, it would give full
flexibility to change parameters, define different test
scenarios, and even to test contingencies and faults, which
are practically impossible to test in a real system.
Laboratory Setup
A laboratory environment resembling the collection grid in
a wind park has been realized at ABB Corporate Research.
A small section of a wind park is reproduced in the
laboratory setup. Illustration of a wind park section is found
in Figure 1 and a highlighted part is reproduced in the
laboratory setup. The schematic diagram is shown in
Figure 2. A total of 600 m cable has been installed.
The radial fork structure of the test laboratory presented in
Figure 2 was intended for studies of reflection phenomena.
However, hundreds of recurrent surges from re-ignitions in
combination with multiple reflection points dramatically
complicate the evaluation of the studies. During the study
presented in this report, it was considered to reduce the
number of wave reflection points in order to reduce the

Page 2

? ? ? ??????????
??????????????????
?????????
?????????????????????????????????????????????????????????? ???????????????????????
?
??????? !"?
?

complications during the evaluation of models and
identification of the consequences of parameter changes.
Consequently, the extra feeder SC3 and the extra windmill
cable SC5 within the test setup were disconnected.
??#???????$????????? ??????%???? !"??????

Figure 1 Illustration of a wind park and the part
reproduced in the laboratory setup (highlighted area)

Figure 2 Wind Cable Laboratory schematic diagram
EXPERIMENTAL SETUP
The experimental setup was prepared at the ABB Wind
Cable Laboratory. Figure 3 shows the circuit diagram. The
installation of surge arresters at the protected transformer
terminal is shown in Figure 4.

Figure 3 Circuit diagram of the experimental setup

Figure 4 Installation of surge arresters (ZnO) within the
experimental setup
The experimental setup consisted of the following major
components:
• Power cable: XLPE, trefoil, core diameter for 240 mm2
aluminum:18.1 mm, insulation thickness 5.5 mm,
diameter over insulation 30.7 mm, cross-section of
screen 35 mm2, outer diameter of cable 40 mm, and
capacitance 0.31 µF/km. The nominal voltage is 20 kV.
• Transformers, TX1: 1250 kVA, 20.5/0.41 kV, oil-
immersed, Dyn11. TX2: 1000 kVA, 20/0.69 kV, oil-
immersed, Dyn11.
• Switching apparatus: 12 -24 kV vacuum breakers, e.g.
rated power frequency withstand voltage 28 kV, rate of
rise of transient recovery voltage (TRV) 0.34 kV/µs and
peak TRV 20.6 kV.
• Protective elements: ZnO surge arresters with
continuous operating voltage 14.3 kV, characteristic
points of 1 mA at 17.2 kV and 10 kA at 28.5 kV.
• Voltage measurement: three 10,000:1 voltage dividers,
20 MHz bandwidth and three 2,500:1 voltage dividers,
20 MHz bandwidth.
• Current measurement: Current transformer 100 ns
useable rise time, max. peak current 50 kA and max rms
current 200 A.
• Transient recorder: 50 MHz sampling rate.
• Control system: The laboratory is equipped with a
sequence control system as standard being phased
locked to the fundamental voltage where all sequence
controlled output signals are programmed by means of a
digital controller given as numerical input to the lab
process central controller. The resolution of the
outgoing raw signal from the sequence controller is 50
µs. The total precision is dominated by output relays
with 7 ms operating time before the breaker pulse
output, in the order of ± 0.1 ms.
• Load: A battery of low voltage air core coils configured
to the reactance XL = 0.10 at 50 Hz connected at Node 1
(Transformer TX1).
During the test reported in this paper, the operating voltage
is set at 12 kVL-L, i.e. 60% of the transformer rating.
TX1

Page 3

? ? ? ??????????
??????????????????
?????????
?????????????????????????????????????????????????????????????????????????????????
?
??????? !"?
?

TEST CASES AND WINDMILL OPERATION
??#???????$????????? ??????%???? !"??????
Switching operations in the medium voltage (MV)
collection grid should be performed in a controlled
sequence involving minimum stress to the windmill system
under all mechanical and electrical considerations. Here
some components such as the low voltage (LV) breakers
and contactors, blade pitch, converters, grid fault ride
through functions, and the MV breaker can be controlled.
Operating the MV breaker located in a windmill connection
point can be made in combination with control of the other
components. From an electrical point of view the operation
is preferably made during minimum load i.e. with the
windmills stopped, minimum load condition or generator
disconnected.
Nevertheless, regardless of the reason for switch/breaker
operations, the corresponding load cases defined for the MV
breaker can be categorized in “no load” and “load” case.
Switching under load can e.g. occur due to faults or
temporary instability. “No load” cases involve the
capacitive currents in an energized system. Capacitive load
has to be considered in particular while operating switches
and disconnectors with low operating speed.
In this study, the opening and the closing of the breaker at
no load or with inductive load connected are considered and
tested. Four test cases are then defined as listed in Table 1.
Table 1 Test cases
Test Case Load Case Switching Operation
1 No load Closing
2 No load Opening
3 Inductive load Closing
4 Inductive load Opening
Studies of pre-strike and re-ignition conditions in circuit
breakers with cables on both sides have shown that all load
currents with both resistive and inductive dominance give
rise to transients, however a mainly inductive load current
provide the worse cases. In the actual experiments, a load
has been selected giving a current essentially larger than the
typical chopping current of a breaker and to a magnitude
that corresponds to the order of equal or less than the
installed power, resulting in 80 % of the installed power at
the selected operating voltage point.
TEST RESULTS AND ANALYSIS
This paper focuses on the occurrence of the transient
overvoltage phenomenon in cable systems. Detail analysis
of the phenomena is not the scope of this report.
Nevertheless, typical results of voltage measured at transfor
mer TX1 terminals (Node 2 in Figure 3) obtained from the
four test cases are listed in Table 1 and also shown in
Figure 5 to Figure 8 and some remarks can still be made.
Figure 5 shows the closing of the breaker at no load. The
switching time is adjusted so the withstand voltage of the
breaker reaches the maximum rated peak voltage when the
voltage is at its maximum at one phase. If that condition is
fulfilled, the pre-strike current is at its maximum value, and
additional pre-strikes will produce the maximum transient
overvoltages. In Figure 6, the switching time of the opening
operation for the no load case is chosen so the no load
current in one phase is at its maximum when the breaker
starts separating contacts. Since the no-load current
represents the magnetizing current of the transformer, the
current lags 90o and the voltage re-strikes are superimposed
on the fundamental frequency voltage when it is at its
maximum. Figure 7 shows the closing of the breaker at
inductive load. The switching time is adjusted as it is in the
case 1. In Figure 8, opening of the breaker with inductive
load is shown. The breaker switching instant in this case is
chosen so the phase current in one phase is at the current
chopping level when the contacts starts to separate. This
ensures that the transient recovery voltage will reach the
breaker withstand voltage very short time after the contact
separation producing a re-strike. It takes much longer time
for the breaker to build up the withstand voltage to the level
higher than the transient recovery voltage produced by re-
strikes compared to the no-load case. This increases the
magnitude and the number of re-strikes compared to the no-
load case. It can be observed that the voltage over the
transformer TX1 is limited by the surge arresters to the level
of 20 kV.
Strikes resulting from breaker operation depend to a large
extent on the time instant of opening and closing of the
contacts of the breaker [4-5]. In this experiment the opening
and the closing of the breaker is performed in such a way
that the worst case scenario is achieved. It was assessed that
the time instant of opening and closing of the breaker
contacts results in the voltage transient of the highest
magnitude. Moreover, the breaker used in the experiment is
also stochastic and probabilistic characteristics (e.g.
probability of arch quenching, etc.). To take this into
account, after the switching time instant resulting in the
highest/steepest voltage transient is obtained for each test
case, 2-4 extra shots are additionally performed with the
exact same instant time. Transients resulting from a typical
test sequence are quantified in terms of number of re-
ignitions, the magnitude and the rate of voltage step, and
shown in the following section.
Measured transient overvoltages versus LIWL
In OHL systems, transient overvoltages often occur as a
result of lightning strikes and according standard equipment
test procedures are designed around the typical 1.2/50 µs
impulse (LIWL = Lightning Impulse Withstand Level) due
to fulmination. When the results of transient overvoltages
measured in the cable laboratory are compared to this
LIWL, Table 2 is obtained.

Page 4

? ? ? ??????????
??????????????????
?????????
?????????????????????????????????????????????????????????? ???????????????????????
?
??????? !"?
?

Table 2 Recorded transient overvoltages* during laboratory
experiment versus standard LIWL (Un = 12 kV) [6]
Number
of strikes
(#)
??#???????$???????????????%???? !"??????
Voltage
step
magnitude
(kV)
Voltage
step 10%-
90%
steepness
(kV/µs)
13-30 1 Closing at no
load
2 Opening at no
load
3 Closing with
inductive load
4 Opening with
inductive load
5 LIWL
* Only voltage steps with the magnitude more than 5 kV and the 10%-90%
value steepness more than 5 kV/µs are identified as strikes.
** Only one shot (from several) in this case resulted in recorded strikes
(more than 5 kV in magnitude and 5 kV/µs in steepness), a result only a
single value displayed.
2-3 10.5-11
3** 7.5** 12**
10-30 10-13 14-27.5
230-323 41.5-43 82.5-88
1 75 62.5
CONCLUSIONS
The experiments verify the occurrence of repetitive pre-
strikes and re-ignitions determined by the electrical circuit
and the point on wave of switching. With a ZnO-type surge
arrester alone as protection device, the experiments show
that a high number of repetitive voltage transients below the
arrester protection level can occur at the incoming terminals
of the transformer. Repetitive voltage transients with higher
time derivatives than the standard lightning impulse test
have been recorded and analyzed.
ACKNOWLEDGMENT
This work was supported by the Swedish Energy.

REFERENCES

[1] J.P. Eichenberg, H. Hennenfent, L. Liljestrand, 1998,
“Multiple re-strikes phenomenon when using vacuum
circuit breakers to start refiner motors”, Proceedings
Pulp & Paper Industry Tech. Conference, 266- 273.
[2] L. Liljestrand, A. Sannino, H. Breder, S. Johansson,
2008, “Transients in Collection Grids of Large
Offshore Wind Parks”, Wind Energy. vol. 11, 45-61.
[3] T.E. Browne, 1984, Circuit Interruption, Marcel
Dekker, Inc., New York.
[4] L. van der Sluis, 2001, Transients in Power Systems,
John Wiley & Sons, Ltd, Chichester.
[5] A. Greenwood, 1991, Electrical Transients in Power
Systems, John Wiley & Sons, Inc., New York.
[6] IEC 60694, 2002, “Common Specifications for High-
Voltage Switchgear and Controlgear Standards”, Ed.
2.2 2002-01 (In conjunction with of IEC 62271, “High-
Voltage Switchgear and Controlgear”, 2003).
1520253035
−15
−10
−5
0
5
10
15
Time (ms)
UTX1 (kV)


phase a
phase b
phase c

Figure 5 Closing the breaker at no load (case 1)
05101520
−15
−10
−5
0
5
10
15
20
Time (ms)
UTX1 (kV)


phase a
phase b
phase c

Figure 6 Opening the breaker at no load (case 2)
1015202530
−15
−10
−5
0
5
10
15
Time (ms)
UTX1 (kV)


phase a
phase b
phase c

Figure 7 Closing the breaker with inductive load
connected at transformer TX1 (case 3)
05101520
−25
−20
−15
−10
−5
0
5
10
15
20
25
Time (ms)
UTX1 (kV)


phase a
phase b
phase c

Figure 8 Opening the breaker with inductive load
connected at transformer TX1 (case 4)