Professional Documents
Culture Documents
Transmission and
TECHNICAL BROCHURE
distribution equipment
October 2020
TECHNICAL BROCHURE
Members
S. MÖHL DE R. NICOLINI IT
M. RECKER DE
Copyright © 2020
“All rights to this Technical Brochure are retained by CIGRE. It is strictly prohibited to reproduce or provide this publication in any
form or by any means to any third party. Only CIGRE Collective Members companies are allowed to store their copy on their
internal intranet or other company network provided access is restricted to their own employees. No part of this publication may
be reproduced or utilized without permission from CIGRE”.
Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the
accuracy or exhaustiveness of the information. All implied warranties and conditions are excluded to the maximum extent permitted
by law”.
WG XX.XXpany network provided access is restricted to their own employees. No part of this publication may be
reproduced or utilized without permission from CIGRE”.
ISBN : 978-2-85873-522-8
Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
TB 817 - Shunt capacitor switching in distribution and transmission systems
Executive summary
The CIGRE WG A3.38, which was formed in 2016, tried to evaluate shunt capacitor switching
performance of medium voltage and high voltage switching devices. This was achieved on one
side by collecting testing and service experience in distribution and transmission networks by
a survey of utilities in multiple countries. On the other side, the state-of-the-art was obtained
from an analysis of publications on the performance of capacitor switching devices and own
experiments. The survey contained questions on the size of substation and line capacitor
banks, the kind of switching devices, typical switching rates and means for controlling inrush
currents, maintenance practices and age of equipment. Publications on the performance of
switching devices during breaking of capacitive currents were evaluated pertaining to the
inrush current parameters during energizing of capacitor banks. One main target of the
Working Group was to assess the long-term performance of capacitor switching devices, in
particular with respect to the probability of restrikes (classification C1 and C2).
In 1999, CIGRE WG 13.04 published the results of a survey on shunt capacitor bank switching
focusing on transmission networks. This TB extends the survey to distribution networks and
goes into more technical details. The results of the first survey in 1999 were introduced shortly
afterwards into the IEC circuit breaker standard, current IEC 62271-100. Therefore, this TB
also provides an evaluation of the standards IEC 62271-100 and IEEE C37.09 with respect to
test sequences and test parameters such as capacitive current and peak inrush currents. In
particular, the effectiveness of the applied accelerated test procedure focusing on minimum
arcing times and high inrush currents and their relevance for operation in the field is analyzed.
Also, the particularities of synthetic testing are described, which is the only reasonable
procedure to test high voltage capacitive switching devices in the laboratory.
For enhancement of the performance of capacitive switching devices, the state-of-the-art of
alternative devices is also described in this TB covering controlled switching, pre-insertion
resistors and current limiting inductors and the use of semi-conductors. The pros and cons and
benefits of such alternative devices are discussed.
One chapter of this TB discusses the peculiarities of filter bank switching. Parameters of filter
banks are explained. In comparison to shunt capacitor switching, inrush currents during
energizing of a filter bank are much smaller and therefore less demanding for the switching
device. Breaking of currents can be more demanding due to the fact, that the recovery voltage
is modified by the additional voltage contribution from the inductance and a superposition of a
high-frequency transient voltage. This transient is due to the resonance of filter inductances
and load-side cable capacitances. The voltage withstand of the switching device during
recovery therefore needs special attention.
Since 1999, when the standards were revised with new capacitive testing procedures, the
installation of capacitor banks has been extended, mainly driven by power quality requirements
of the network. The present survey on the application and user experience of capacitive
switching devices returned 52 responses from utilities in 18 different countries from which only
38 responses provided sufficient data to be evaluated. Each respondent (utility) provided data
on three to four voltage ranges so that the survey covers 146 different utility-voltage ranges,
from 3.6 kV to 550 kV.
The Working Group commissioned the survey with the intent to discover answers to at least
the following questions:
Survey questions Overall results
Why capacitor banks are The majority applies them for voltage and VAR support.
applied in networks?
What is the average power Capacitive currents at all rated voltages vary between 250 A
of installed capacitor and 390 A with higher mean current values below 17 kV.
banks?
3
TB 817 - Shunt capacitor switching in distribution and transmission systems
Are capacitor banks used Only responses from US indicate they are used for this
on overhead lines? purpose and only at rated voltages up to 38 kV.
What is the classification of The majority claim class C2 and rarely indicate class C1 or
breakers? C0 (C0 is only defined by IEEE).
What is the rate of More than 50 % of the devices are switched at least once
switching? per day.
What is the age of the High voltage switching devices are not older than 20 years,
switching equipment? whereas medium voltage equipment has an age of 10 to 30
years.
What is common 89 % of the respondents maintain on a time-based schedule
maintenance practice? with a majority of 47 % applying intervals > 5 years in HV
networks, whereas in MV networks 42 % claim a one to five
years interval.
What are causes of 50 % of failures are of dielectric nature occurring in the
failures? capacitor bank or in the switching device, and 24% are
mechanical failures of the switching device.
How often surge arresters Only 33 % of the respondents apply surge arresters with a
are used? higher percentage in the high end of MV and HV ranges.
What is the satisfaction of Satisfaction is expressed mainly for SF6 devices.
users with respect to their
switching devices?
Are alternative methods Point-on-wave closing is applied to 60 % of all HV breakers,
applied? whereas more than 63 % do not use any control for MV
devices.
What are the peak inrush For MV applications, the majority of inrush currents peak up
currents? to 10 kA, whereas for HV peaks even above 20 kA are
reached.
One main purpose of this Technical Brochure is to give guidance to users for selecting a
switching device at capacitive currents and inrush currents calculated for the capacitor banks
in their distribution and transmission network. These parameters may differ from the switching
parameters specified and tested according to the standards. Since the main principles of
switching used in today’s switching devices are typically based either on the vacuum or on the
SF6 gas technique, the performance of these devices is treated separately in this TB, where
appropriate. Other principles are only briefly mentioned. However, the aim of the distinction of
the breaking techniques is not to discriminate one against the other but to understand their
different performances. Below some guidance is provided with respect to the most important
switching parameters i.e. capacitive breaking current, inrush current peak and frequency, and
some conclusions given concerning electrical endurance of shunt capacitor switching devices.
- Capacitive breaking current (refer to 4.3.5)
The type tests performed at a rated current of 400 A as specified by the standards are not only
valid for currents equal or lower 400 A but also for larger capacitive currents, since higher
currents exert a beneficial conditioning effect. This effect is more pronounced for vacuum
interrupters, mainly regarding their contact surfaces, than for SF6 interrupters. The Working
Group concludes that even significantly higher capacitive currents above the tested value do
not change the performance class of the device. A conservative approach may be 50 % higher
breaking currents. For SF6 interrupters, a conditioning effect due to breaking currents was not
observed, however in general, there is no negative impact anticipated for higher currents
unless there is a limitation on the total number of switching operations. The breaking
4
TB 817 - Shunt capacitor switching in distribution and transmission systems
performance depends to some degree on the peak inrush currents and the related erosion of
contacts.
- Inrush current peak (refer to 4.3.2 and 4.3.4)
As long as the inrush current peak is smaller than the tested value, it can be assumed that the
deterioration of the contact surfaces in a switching operation is smaller than in type tests where
the highest inrush current peak (rated value) is applied. In particular vacuum interrupters show
an inverse relationship between peak inrush current and breaking performance. Therefore,
peak inrush currents in excess of the rated and tested values cannot be permitted, unless it
can be proven that the deterioration of the surface by higher inrush currents is lower than
tested, which depends on the inrush frequency. Therefore, a new parameter was introduced
i.e. the inrush current integral, ICI, which is an approximate indirect measure of the contact
erosion. Under the premises of equal or smaller ICI, the Working Group believes that an
extension to higher peak inrush currents might be possible without negative impact on the
switching performance. The benefit of alternative devices is to reduce the inrush current peak.
- Inrush frequency (refer to 4.3.3)
Except for oil interrupters, the performance of switching devices does not depend on the inrush
current frequency. Frequencies considerably above or below the frequency as specified in
standards (4250 Hz) can be permitted for vacuum and SF6 interrupters. In any case, the
switching of single capacitor banks implies much lower inrush current frequencies than for
back-to-back capacitors, and the standards do not require separate tests if back-to-back
switching has been verified. In order to reduce the inrush current peak and ICI, additional series
inductances (reactors) can be installed. This automatically lowers the inrush frequency and
has a positive impact on the breaking performance due to the lower inrush current. The
frequency specified in the standards is mainly for defining a common test circuit.
- Electrical endurance (refer to 4.3.6)
Based on responses from the survey, the Working Group concludes that the existing type tests
are an adequate method of evaluating the performance of capacitor bank switching devices.
The survey revealed that users are generally satisfied with the performance of their capacitor
switching devices. This satisfaction has to be seen in the light that more than fifty percent of
the respondents indicate their devices are switched daily, which easily amounts to even more
than 1000 operations in a couple of years. This supports the positive assessment of the
Working Group that the restrike performance of capacitor switches and circuit breakers tested
according to the standards is acceptable in the field or at the very least, if restrikes occur, they
are not noticed and therefore are not destructive.
The back-to-back capacitor switching performance verified in 104 three-phase or 168 single-
phase attempts according to class C2 per standards IEC 62271-100 and IEEE C37.09 can be
extrapolated to at least 500 random operations in the field evaluating the accelerated test
procedure applied in the standards. The performance is probably applicable to a much higher
number of operations. Occasionally in experiments, several thousand making and breaking
operations were performed. The switching contacts exhibited some deterioration up to
considerable erosion and melting upon post-test inspection. An impact on the restrike
probability could not be determined.
Chapter 4 of the TB discusses in detail the specific dependencies determining reignitions and
restrikes after current breaking, separate for vacuum and SF6 interrupters. The inrush current
characteristic mainly determines the breaking performance among other factors. For vacuum
interrupters, where still a lot of open questions on the prevailing physical processes exist, the
deterioration of the contacts due to the rupture of microscopic welds from pre-arcs and large-
area melting of surfaces due to high inrush currents are essential. For SF6 interrupters, the
gradual changing of the geometry of arcing contacts and nozzles due to the erosion of high
inrush currents is important.
5
TB 817 - Shunt capacitor switching in distribution and transmission systems
Contents
Executive summary ............................................................................................................. 3
6
TB 817 - Shunt capacitor switching in distribution and transmission systems
7
TB 817 - Shunt capacitor switching in distribution and transmission systems
8
TB 817 - Shunt capacitor switching in distribution and transmission systems
9
TB 817 - Shunt capacitor switching in distribution and transmission systems
10
TB 817 - Shunt capacitor switching in distribution and transmission systems
11
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 1: Single bank circuit involving a switching device S1 and back-to-back circuit involving S2
Back-to-back switching has some peculiarities compared with single bank switching. As the inductance
L1 and L2 between the capacitors is generally small in comparison to Ls, the inrush current when
switching the second capacitor is much higher than the single bank (or first capacitor) inrush current.
Good technical practice will comprise current limiting reactors for back-to-back applications which will
increase L1 and L2 in Figure 1. Long cables between the capacitors may have a limiting effect as well.
For single core cables, at 36 kV rated voltage and 25 kA short-circuit current, for example, a cable with
a length of 550 m reduces the back-to-back inrush current peak to that of the single capacitor. For 3-
core belted cables, the distance between substation capacitor banks needs to be 1900 m. For a single
-phase circuit or a three-phase solidly grounded network, the dependence is given by1:
𝑈𝑟𝑚𝑠
𝑙𝑐𝑎𝑏𝑙𝑒 =
12𝜋𝑓0 𝐼𝑠𝑐 𝐿𝑐𝑎𝑏𝑙𝑒
Equation 1: Distance between substation capacitor banks connected by cables at which the peak inrush
current of the single capacitor equals that of back-to-back switching
where Lcable is the specific cable inductance equal to 1.4 mH/km for a single core cable and 0.4 mH/km
for a belted 3-core cable, f0 is the power frequency, Isc the short circuit rating and Urms the phase-to-
phase network voltage. The equation assumes that the two substation capacitor banks are equal in size.
From that comparison, back-to-back configurations might even occur between separate substations.
However, in most cases, separately switched capacitors in one substation are concerned.
Generally speaking, the actual required performance of a capacitive switching device is quite close to
its rated and type tested performance and thus poses high stress to the device over the entire life of the
equipment. This not only comprises electric stress due to the high recovery voltage, but also mechanical
stress due to the large number of operations. The kind of switching device i.e. whether it is based on air
or SF6 gas, oil or vacuum has an impact on the particular physical processes occurring during making
and breaking and therefore gain special attention in this TB (chapter 4). This TB only considers vacuum
and SF6 interruption techniques, but not air or oil because those are outdated technologies. Alternative
gases have not been considered as experience on switching is still missing.
Tests have been defined in international standards IEC 62271-100 [4], IEEE C37.09 [5], IEEE C37.100.2
[6] and IEEE C37.66 [7] for verifying the performance of capacitive switching devices including circuit
breakers. Because of the limited number of trials in such type tests, accelerated test procedures have
been developed in the past to assess the performance over the whole life of the device [8]. This
highlights the problem of verifying a device, which is stressed in service with values close to its design
limits.
Some other capacitive switching cases are not considered in this TB such as outrush currents in case
of faults in the system, which were investigated intensively by CIGRE WG A3.26 [9], or the switching of
very long transmission lines. The switching of filter banks, however, is a particular case which is in more
detail described and treated in chapter 6 of this TB.
12
TB 817 - Shunt capacitor switching in distribution and transmission systems
2 This is likely because most devices are < 20 years old, the new ratings were defined in 2000, and most devices
installed since the inception of the ratings are rated C2.
13
TB 817 - Shunt capacitor switching in distribution and transmission systems
The majority of shunt substation banks are switched with SF6 circuit breakers, in particular in HV
substations. In MV substations, vacuum breakers, oil breakers, and IEEE C37.66-style capacitor
switches are almost commonly used. Circuit switchers are mostly applied in HV systems besides
SF6 breakers.
The devices are generally switched via SCADA (46.9 %), but voltage-controlled switching (19.1 %)
and algorithm-controlled switching (21.7 %) are also common.
Regarding faults in shunt substation banks, no singular cause can be identified. Faults in the bank
itself, in the dielectric of the switching device, and the mechanics of the switching device are
common responses.
The application of inrush mitigation switching techniques is very different in medium and high voltage
networks:
o Lower voltage classes have much fewer mitigation schemes implemented: 63 % of
respondents switch without any control, 22 % use pre-insertion resistors.
o In high voltage networks, SF6 breakers are often used for controlled switching (60 %). Point-
on-wave closing is the most common response, but there is wide variance.
Maintenance is performed based on time intervals, not based solely on a certain number of
operations.
o Time-based maintenance is applied by all respondents with a period of maintenance
between 1 and 5 years over all voltage ranges. In high voltage, 46.5 % utilize maintenance
periods of even greater than 5 years, while in medium voltage, 42 % claim a one to five
years interval.
o Between 33 % (for vacuum) and 54 % (for SF6) of the respondents apply condition-based
maintenance in addition to time-based maintenance with an interval of between 500 and
2000 operations. The specifics of the conditions were not a focus of the survey.
o Maintenance schedules are similarly applied for SF6 and vacuum devices and rarely
differentiated between voltage ranges. It seems that respondents apply maintenance
strategies to their equipment irrespective of device type which indicates that no device is
particularly prone to failures.
14
TB 817 - Shunt capacitor switching in distribution and transmission systems
failure mode, the Working Group recommends more frequent application of surge arresters.
The low application rate requires a closer cost-benefit analysis on surge protection. Clause 7.5
of IEC 60099-5:2018 [77] provides some rules for the selection of surge arresters.
15
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 2: Responses as a function of voltage class with respect to substation and line equipment
Table 1: Scoring of capacitor bank size (power) versus voltage with calculated mean currents (see text)
Umax /kV
Size /MVA 8,26 17,5 27 38 48,3 72,5 100 123 145 245 362 550
1 1 1 1
2,5 2 6 3 1 1
5 2 5 3 3 2 1 1 1
15 1 8 4 5 1 3 1 2 1
30 1 1 5 5 1 3 2
60 1 1 4 4 6 5 3 1
200 5 8 2 3
300 6
lower mean
current /A 361 376 219 288 82 270 96 188 408 344 156 258
higher mean
current /A 628 797 337 406 104 405 133 231 481 582 230 392
16
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 3: Reasons for usage of substation capacitor banks in different regions of the world and for line
banks in US and Canada.
In China, the requirements for a capacitive switching device is typically closer to 800 A. These currents
are present for larger capacitor banks, up to 1600 A in a back-to-back configuration. Compared to typical
continuous currents, this is extremely high as it would mean there exists 1600 A of inductive current for
an installation that has a 3 kA or 4 kA nominal current. These capacitor banks are used on the MV side
of the transformer to compensate on the HV side, because this practice lowers the voltage transients
on the MV side during switching. These switches typically operate multiple times per day and are
designed to handle very large reactive compensation needs. There are no back-to-back banks above
35 kV in China, and two MV ratings are used: 10 kV and 35 kV. This can result in inrush currents around
60 kA, and as such, current limiting reactors are used in these circuits permanently.
Most applications connect capacitor banks in a star configuration. It is most likely that delta
configurations are used mainly in some low voltage applications. It is assumed that this is a matter of
convenient bank arrangement and bank protection. For "can type" capacitors, two bushing capacitors
are only available up to about 18 kV (with some exceptions for fuse-less capacitors). Some designers
may find it unseemly to connect a phase wire to the capacitor case or bank frame. Beyond this, internally
fused capacitors (which are widely used outside North America) are not self-protecting for a large
internal short circuit and require some kind of external relay protection to prevent case rupture and fire.
Such protection is easier to realize in a star configuration. For transmission voltage capacitor banks, it
is easier and more compact to attain the needed voltage rating with a star configuration by connecting
capacitor units in series and stacking racks of capacitors on top of one another.
Line banks – common in the United States – are switched in single-phase mode and extend only to
voltages up to 38 kV. They are smaller than substation banks and have switching currents well below
100 A.
17
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 4: Relative occurrence of the type of switching device separately for MV and HV networks.
3Since respondents did not specify the kind of mechanical failure, the WG assumes that these failures mostly
occurred in the operating mechanism due to large number of operations.
18
TB 817 - Shunt capacitor switching in distribution and transmission systems
bank installation. Further interpretation of Figure 5 is problematic due to lack of details and limited
statistical evidence.
100%
75%
Mechanical failure of
switching device
50%
Fault in capacitor bank
25%
Dielectric failure of switching
device
0%
Damage of protection
elements
Figure 5: Percentage of failure modes vs. voltage classes for shunt substation capacitor banks.
19
TB 817 - Shunt capacitor switching in distribution and transmission systems
80,0%
70,0%
60,0%
50,0%
40,0%
30,0%
20,0%
10,0%
0,0%
fusing no fuse/SDO with SA No SA
50,0%
MV (< 38 kV)
40,0%
HV ( > 48 kV)
30,0%
20,0%
10,0%
0,0%
> 2 /day > 1 /day >1/ > 1/ > 6 / year > 1 / year < 1 /year
week month
Figure 7: Responses indicating the typical switching rate for capacitor bank installations agglomerated
over medium and high voltage range.
20
TB 817 - Shunt capacitor switching in distribution and transmission systems
50,0%
40,0%
MV (< 38kV)
20,0%
10,0%
0,0%
<5 5-10 10- 15- 20- >30
15 20 30
Figure 8: Relative occurrence of most common age in years of medium voltage (blue) and high voltage
(red) equipment
The chart (Figure 9) shows the average age of each type of switching device listed along the X-axis.
The center of each bubble is placed on the y-axis at the value of the mean age of the device type. The
area of the bubble is related to the total number of respondents for that particular type of switching
device and normalized against the device type with the smallest number of responses (n = 1 for Line –
Contactor and Line – Oil Circuit Breaker). SF6 circuit breakers are quite commonly used for the switching
of shunt substation capacitor banks, and their mean age is about 15 years. Fewer utilities are still using
oil circuit breakers for their substation capacitor banks, however, a significant amount of them are still
used with an average age around 18.5 years. Generally, line bank switching devices seem to be younger
than substation bank devices, but this may be cautioned because of the low number of responses.
Figure 9: Bubble chart showing average switching device age, per device type, split out by substation vs.
line bank (larger bubbles indicate higher response count)
21
TB 817 - Shunt capacitor switching in distribution and transmission systems
100%
75%
50% <= 38 kV
> 38 kV
25%
0%
Human Autonomous - Human Autonomous -
Analysis/Action Line Analysis/Action Substation
- Line - Substation
Figure 10: Intent of operation as a percentage of respondents separate for MV and HV; four columns on
the left for line banks; four columns on the right for substation banks
15 Recloser
22
TB 817 - Shunt capacitor switching in distribution and transmission systems
40,0%
MV ( <38 kV)
20,0%
10,0%
0,0%
<5 5-10 10-15 15-25 >25
Figure 12: Relative occurrence of peak inrush currents given in kA (horizontal axis) in medium voltage
networks (blue) and high voltage networks (red)
23
TB 817 - Shunt capacitor switching in distribution and transmission systems
From the responses, it seems that users apply their maintenance strategy to all assets irrespective of
device type and switching rate and do not differentiate for the application. This, however, needs to be
taken with caution, since the survey might have not allowed for sufficient differentiation. Knowing that
most of the capacitor banks are switched on a daily basis, and thus will reach a high number of
operations in a short time, one can nevertheless conclude that time-based maintenance practices are
sufficient to ensure the reliability requirements of the network.
Both the responses to the mean age of the equipment (more than 10 years old) and the generally time-
based maintenance (with 1 to 5 years period) suggest that the failure rate of these devices is in line with
manufacturer specifications and expectations.
50,0%
High voltage
25,0%
12,5%
0,0%
annually 1 - 5 years > 5 years not applic.
Figure 13: Relative occurrence of time-based maintenance (maintenance period) in medium voltage
networks (blue) and high voltage networks (red)
60,0%
Medium voltage
30,0%
15,0%
0,0%
< 200 200 - 500 500 - 2000 > 2000 not applic.
Figure 14: Relative occurrence of condition-based maintenance (operation numbers) in medium voltage
networks (blue) and high voltage networks (red)
24
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 15: Application of time and condition-based maintenance differentiated with respect to device type
(SF6 or vacuum) and switching rate (daily or less than weekly)
2.8. Demographics
Please note that some respondents preferred not to disclose demographic information, so the total count
of responses in these charts may not correlate directly with the information discussed previously in this
chapter. The chart (Figure 16) shows the responses based on location i.e. country. United States has a
relatively high occurrence thanks to advertisement on an IEEE switchgear standards conference.
25
TB 817 - Shunt capacitor switching in distribution and transmission systems
use few line banks above 38 kV. It has to be remarked that the number of responses having substation
banks installed at the specified voltage range was 109 with the majority of 60 coming from US or
Canada. The number of responses having line banks in use was only 34 with 30 coming from US or
Canada.
Figure 17: Proportion of reported banks installed at a particular voltage class by region, separate for line
banks and substation banks.
26
TB 817 - Shunt capacitor switching in distribution and transmission systems
In edition 4 of IEC 56 [12] published in 1987 (and later renumbered to IEC 60056) the number
of tests for each test-duty for capacitive current switching was increased to 48 with the aim to
prove restrike-free performance in the field.
3.1.2. ANSI/IEEE
The first documents containing requirements and test procedures for switching of capacitive
loads for circuit breakers were the following:
– A general purpose circuit breaker is not specifically designed for capacitance current
switching;
– A definite purpose circuit breaker is designed specifically for capacitance current switching.
The difference between the two types of circuit breakers lies in the restrike performance and
associated overvoltages. Multiple restrikes are not allowed for a general purpose circuit breaker.
For definite purpose circuit breakers, there shall be no more than one restrike per operation, and
that restrike shall not be preceded by a current pause in excess of 1/3rd cycle. The requirement
for definite purpose will result in lower overvoltages in the system.
– Tests with 50 Hz cover application for 60 Hz if the test voltage is increased by 110 %;
– The voltage regulation (i.e. the change in voltage when the capacitive load is removed or
inserted) and its effect on the recovery voltage is taken into account;
– Initial voltage jump is not taken into account;
– The connections should be reversed on circuit breakers having a non-symmetrical insulation
path.
3.1.3. Joint IEEE/IEC test procedure
As stated in 3.1.1, the number of tests in IEC 60056 was increased with the aim to improve the
restrike-free performance. Despite the increased number of tests, the test procedure had its
flaws. This became evident when restrike-free circuit breakers were reported to restrike in the
field. The reasons were:
– A misleading definition of the term “restrike-free circuit breaker”. The definition was as
follows: “circuit breaker that interrupts without restrike during the capacitive current breaking
test duties specified in this standard”;
– No focus on interruptions with short arcing times. As a matter of fact, depending on the
chosen angle at the beginning of the test, tests with short arcing times did not occur;
– Limitations of the test circuit, especially the initial voltage jump that has an effect on the
shortest arcing time that can be obtained during the test.
27
TB 817 - Shunt capacitor switching in distribution and transmission systems
During the revision of the 3 rd edition of IEC 60056 a joint Task Force (TF) was formed between
the IEEE Switchgear Committee and IEC/SC 17A Working Group 21 (responsible for the revision
of IEC 60056). The aim of the TF was to elaborate a joint test procedure for switching of
capacitive loads with emphasis on restrike-free performance. The work of the TF lead to
amendments to existing standards, IEEE C37.04a-2003, IEEE C37.09a-2005 and IEC 62271-
100:2001, the new number of the IEC standard for HV circuit breakers. The most significant
changes made to the capacitive test procedures were the following.
– The term “restrike-free” circuit breaker was replaced by introducing two classes: class C1 for
circuit breakers having a low probability of restrike and class C2 for circuit breakers having
a very low probability of restrike. Each of the two classes is associated with a specified test
procedure;
– Class C1 tests are performed on a circuit breaker in new and clean condition, class C2 tests
are performed on an “aged” circuit breaker. Ageing is achieved by performing 3 interruptions
at 60 % of the rated short-circuit current of the circuit breaker;
– The number of tests was increased. For back-to-back capacitor bank switching tests closing
operations are performed with inrush current;
– The initial voltage jump is better defined (see 4.1.2 for further information);
– IEEE C37.09a-2005 defines another class, class C0 for circuit breakers where restrikes are
not a concern.
3.1.4. IEEE standard for common testing requirements for capacitive current
switching
In IEEE a number of standards cover requirements and test procedures for different capacitive
current switching devices:
IEEE C37.100.2 follows the joint procedure for circuit breakers to a high degree. The
preconditioning for class C2 tests applies only to switching devices that have a fault current
breaking capability.
IEEE C37.100.2 also covers a test procedure for random switching operations. This test
procedure consists of 810 three-phase tests or 1200 single-phase tests, which were adapted
from the existing standards IEEE C37.66 and IEEE 1247.
The test and acceptance conditions during capacitor bank switching comprises the following test
duties and procedures as per clause 6.111:
28
TB 817 - Shunt capacitor switching in distribution and transmission systems
Table 3: Capacitor bank switching test duties for class C2 devices adapted from C37.100.2
29
TB 817 - Shunt capacitor switching in distribution and transmission systems
Table 4: Capacitor bank switching test duties for class C1 devices adapted from IEC 62271-100
Total required 24 24
3.3. Test requirements specified in other standards i.e. IEEE C37.66 and
IEEE 1247
Besides the circuit breaker standards IEC 62271-100 and IEEE C37.09, other product standards for
electrical equipment contain test requirements verifying the performance of the product in switching
capacitance currents. Since this TB focuses on switching of capacitor banks, cable and line switching
test procedures are not considered here.
The following product standards contain capacitor bank switching test definitions and procedures:
a) IEEE C37.66: 2005 - IEEE Standard Requirements for Capacitor Switches for AC
Systems (1 kV to 38 kV),
b) IEEE 1247: 2005 - IEEE Standard for Interrupter Switches for Alternating Current, Rated
Above 1000 Volts,
c) IEEE C37.100.2: 2018 – IEEE Standard for Common Requirements for Testing of AC
Capacitive Current Switching Devices over 1000 V,
d) GB 1984 – High voltage alternating-current circuit breakers (Chinese standard)
e) JEC 2300: 1998 – AC Circuit Breakers (Japanese standard)
f) GOST 52565: 2006 – A.C. circuit breakers for voltages from 3 to 500 kV: General specifications
(Russian standard),
g) British standard (BS), Australian standard (AS), CSA and other local standards.
All standards listed above under d) to g) are based on the capacitor bank switching procedure in IEC
62271-100 or IEEE C37.09a with only slight modifications. Thus, they all apply special test procedures
with an acceleration factor concerning the arcing time comprising around 100 operations.
The Japanese circuit breaker standard JEC 2300 serves as an example. In JEC 2300-2010, the test
conditions and procedures of the capacitive current switching test are defined in subclause 6.12 of this
standard. The contents match with the IEC 62271-100 standard, however, there are some characteristic
differences based on the results of a Japanese users survey. For example, rated values of line-charging
breaking current and cable charging breaking currents are provided in JEC 2300 – and are the same as
for IEC except for 24 kV - but rated values of the capacitor bank breaking current are not provided. Since
the breaking current depends on the capacitance of the capacitor bank used in the field, only the
maximum power of capacitor banks at each rated voltage are provided as reference in this standard
(e.g. 20 MVar at 12 kV and 60 MVar at 36 kV). Also, test parameters for inrush current are not provided
because capacitor banks in Japan are generally installed with 6 % reactor in series so that it is not
necessary or less important to consider the effect of the inrush current.
30
TB 817 - Shunt capacitor switching in distribution and transmission systems
The two IEEE standards a) and b), listed above, apply random test procedures for capacitor and load
interrupter switches instead of the accelerated test procedure in IEC 62271-100. The recently published
IEEE standard C37.100.2, mentioned under item c) has unified the two procedures of these standards
and requires the test duties as listed in Table 5. They require 810 operations for three phase testing and
1200 operations for single phase testing. The original standards specified 1200 operations for single
phase testing.
All tests are CO operations. Inrush currents occur randomly distributed on phases up to the maximum
rated inrush current, and capacitive breaking currents are interrupted at random arcing times. There are
additional requirements, which shall ensure true random operations covering all possible arcing times –
long ones and short ones – and a reasonable distribution of inrush currents. For example, in at least
30% of all operations 85 % of the maximum rated inrush current should be achieved, and in 3 % of the
operations of the C100 test duty (i.e. 8 operations at each polarity) 100 % of the rated inrush current
should be reached. In the test duties C20 and C50, the inrush currents are lower than at C100, since it
is required to adapt the load-side capacitor only to the lower currents.
The performance criteria within these IEEE standards are such that the number of restrikes in a
complete series determines the class of performance of the switch.
Class C2 allows only two restrikes over all test duties.
Class C1 allows up to 24 restrikes over all test duties.
Class C0 is considered to have an unspecified probability of restrikes with more than 25 restrikes
in the test duties.
In contrast to the circuit breaker standards, there is no pre-conditioning required for class C2 switches,
since these are not supposed to interrupt fault currents.
Table 5: Capacitor bank switching test duties for random switching adapted from C37.100.2
Additional CO tests,
376 376 304 246 246 198
random operations
The main example of flexibility in the standards requirements is given by the value of the breaking
current: IEC 62271-100 indicates a preferred maximum current of 400 A (up to a rated voltage of 420
kV) but it is frequently requested by manufacturers, both of MV and HV equipment, to carry out tests at
31
TB 817 - Shunt capacitor switching in distribution and transmission systems
higher values as for instance 800 A, 1200 A and even more than 2000 A. This is an important element
that forces the laboratory to consider the synthetic testing method for the following reason:
The standard prescribes for direct tests that the characteristics of the test circuit should be such that
the power frequency voltage variation, when switching, should be less than 2 % for test-duty BC1
and less than 5 % for test-duty BC2.
This means that, for a test-duty 2 with 400 A of breaking current, the power source shall have at least
8 kA of short circuit current at the test voltage level, becoming 40 kA in the case that a capacitive
breaking current of 2000 A is requested. At HV ratings, but also at the higher levels of MV, this
requirement implies the availability of a source with a considerable short-circuit power.
Another aspect that the laboratory must consider for carrying out direct test is the size of the load
capacitor bank. As an example, a three-phase test on a 145 kV circuit breaker with 400 A breaking
current requires a 100 MVar capacitor bank, and even larger, if higher voltage ratings and/or breaking
currents must be covered.
As a consequence of the above-mentioned reasons, many laboratories resort to the synthetic method
to cover high voltage equipment testing.
The main reference standard for synthetic testing is IEC 62271-101 [16].
IEEE introduced a guide for the synthetic method in 1981, then revised it in 2007, with the document
C37.081 [17] and its supplement C37.081a [18]. However, IEEE C37.09 document [5] refers to IEC
62271-101 when dealing with synthetic tests and C37.081 is only mentioned in the bibliography.
Moreover, synthetic capacitive tests were not treated in IEEE C37.081.
Synthetic capacitive current switching tests are generally carried out using single-phase test circuits,
since three-phase circuits are quite complex to accomplish and not required by the standards. It must
be highlighted that in the former versions of IEC 56, capacitive three-phase tests were prescribed, similar
to short-circuit T100 tests, when testing a three-phase circuit breaker in one enclosure. At that time,
many laboratories were able to perform three-phase capacitive tests only using direct circuits, with
consequent limitation in voltage level, supply source short-circuit power and switching current values for
the reasons described above. Later, this prescription was removed making it possible to perform all
tests with single-phase test circuits but adding requirements on the verification of dielectric performance
(Annex O.3 and O.4 of IEC 62271-100 in the case of dead tank and GIS circuit breakers).
Being carried out as single-phase tests, synthetic capacitive current switching tests require a quite high
number of operations when testing a class C2 circuit breaker: 48 for BC1 and 120 for BC2. To shorten
the test duration when performing line and/or cable switching in addition to capacitor switching BC, the
standard allows carrying out “combined tests” in which it is possible to cover more categories at the
same time provided that:
Test parameters, number of tests and sequences for all the categories are fulfilled
Capacitive current is not less than the highest capacitive current rating to be demonstrated (test
duty 1 and test duty 2 cannot in any case be combined)
Test voltage corresponding to the highest rating to be demonstrated
TRV parameters corresponding to the most severe condition.
The possibility to carry out combined tests is very appreciated by manufacturers to reduce the time and
consequently the cost of a complete test program, even if it may result in a higher probability for a
negative result since it increases the stress to the circuit breaker.
A typical combined test is a BC2 back-to-back capacitor bank sequence covering also BC2 single
capacitor bank, line switching and cable switching test duties 2. The test voltage and TRV in this case
corresponds to the highest capacitive voltage factor kc chosen from the values applied to each category.
In this way, the back-to-back test, normally requiring kc=1.0, can be carried out with kc=1.2 or kc=1.4. A
voltage factor of kc=1.7 is prescribed for breaking tests in non-effectively earthed neutral systems in the
presence of single or two-phase earth faults.
To reduce the complexity of test parameters in synthetic circuits, the laboratory can use simplifications
permitted by the standard. A typical procedure is to separate the CO operations required for back-to-
back capacitor bank switching BC2 into a series of making tests with inrush currents and no-load
opening followed by a series of CO tests with the rated breaking current and no-load closing conditions.
This possibility is given by IEC and IEEE, too, with IEEE prescribing that the C of the CO series shall be
carried out under voltage (no requirement for the inrush current is prescribed). In this way it is
32
TB 817 - Shunt capacitor switching in distribution and transmission systems
significantly easier for the laboratory to use two different circuits and apply and control the correct
parameters for making and breaking operations separately.
The inrush current for BC2 back-to-back switching has the same parameters at all rated voltages of the
circuit breaker, i.e. 20 kA at a frequency of 4 250 Hz. The test voltage value is the only variable when
testing differently rated voltage equipment.
A typical inrush circuit comprises the pre-charging of a shunt capacitor bank connected to the AC source
side of the circuit breaker under test through an auxiliary breaker. When opening the auxiliary breaker,
the capacitor bank remains charged to the test voltage value and then, closing the circuit breaker under
test to ground, the inrush current is obtained. Value and frequency of the inrush current are regulated
using current limiting reactors of proper values. A DC charging of the source side capacitor bank is not
suggested because the same DC voltage is applied to the bushings of the circuit breaker as well,
resulting in a polarization with possible discharge during the inrush current test. Circuits using an
auxiliary breaker can be used but the preferred method is to pre-charge with AC voltage.
IEC 62271-101 suggests possible synthetic circuits to test capacitive current breaking performance. The
laboratory is free to select the most applicable circuit based on the availability of equipment to achieve
the required parameters, on the breaking current value and on the simplicity of test execution. Many test
circuits with different features are indicated in the standard: Annex G gives several examples of general
circuits while Annex N adds examples applicable to metal enclosed or dead tank circuit breakers. Both
annexes are informative.
There are no restrictions in using the indicated circuits or even a different type of circuit, provided that it
can be demonstrated that the correct solicitations are applied to the circuit breaker under test. A
laboratory can develop its own circuit and present it to be included in the standard, after proper
verification. In fact, this is the way how the annexes have been developed.
The widely used test circuit for breaking tests are of two types:
a. Capacitive current injection circuit
The test circuit consists of two combined circuits, a current circuit and a voltage circuit. Both
circuits are of capacitive nature. The application of this type of circuit implies the use of an
auxiliary circuit breaker to isolate the test circuit breaker in the voltage circuit from the current
circuit.
Figure 18: Current injection circuit – example G.2 of annex G of IEC 62271-101 [16]
b. LC oscillating circuit
The test circuit consists of an LC oscillating circuit that provides both the current and voltage
from a single source. The application of this type of circuit does not require the use of an auxiliary
circuit breaker.
33
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 19: LC oscillating circuit – example G.3 of annex G of IEC 62271-101 [16]
The various test circuits can be used without changes for all the test duties and for the various
categories.
An important factor to be considered in synthetic capacitive tests is the dielectric stress of the circuit
breaker, particularly for GIS and dead tank breaker equipment. In the typical synthetic circuit one
terminal of the circuit breaker is connected to ground and, when the TRV is applied, all the voltage is
“seen” by the insulation between the live terminal and the tank. For instance, when testing with kc=1.4,
a voltage of 2.8 p.u. is applied. This does not happen in direct tests, where the dielectric stress of the
insulation is correctly distributed between phases and to ground, which is representative of real
conditions. To avoid overstress, the laboratory normally adopts circuits that insulate the tank to ground
and regulate the distribution of the voltage in order to have a portion of about 50 % on each terminal.
As the synthetic method is not correctly representing the real dielectric stress during breaking
operations, the IEC standard introduced additional requirements on the verification of dielectric
performance with Annex O of IEC 62271-100 in the case of dead tank and GIS circuit breakers. Even
in the case of direct tests, the laboratory can use simplifications that are allowed by the standard to
better control the testing parameters.
For instance, considering the BC2 back-to-back test duty (widely used to cover more categories at a
time, as previously indicated), some laboratories use circuits in which the CO sequence is carried out
with separate sub-circuits, switched through auxiliary breakers. In this case the alternation of C and O
is respected but it is much simpler for the laboratory to control the test parameters for the inrush currents
and for the breaking currents separately. This procedure is possible as the standard does not prescribe
the time between C and O operations. Only some seconds are required to switch between the two
circuits.
There are no particular indications of differences in the performance of a circuit breaker when carrying
out the test sequences by alternating inrush making and breaking operations or by a series of inrush
operations followed by a series of CNO LOADO operations. However, inrush currents are typically wearing
out the contacts of a breaker. In the case of vacuum circuit breakers, emission sites (sometimes called
“spikes”) might be formed on the contact surfaces during making operations, an effect which is normally
mitigated by current breaking operations smoothing the surface. If closing operations are carried out all
in one series, the “spike” effect could be enhanced. In the case of high voltage SF6 circuit breakers, the
inrush current typically wears out the nozzle and contact pin material sometimes creating holes, burnings
or cracks or detachment of small fragments of nozzle material. This could affect the withstand voltage
of a switching device.
34
TB 817 - Shunt capacitor switching in distribution and transmission systems
35
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 20: Comparison of momentary current (top) and current integral (bottom) between inrush current
and short-circuit making current, mind the different time scales.
Besides the total amount of converted energy, the contact area where the energy is deposited is an
important factor related to the current density of the pre-arc. Irrespective of the kind of switching device,
the pre-arc is initially a concentrated arc with high current density able to melt the contact surface under
its arc roots. Afterwards, the arc may spread out reducing the current density considerably or is driven
by electromagnetic forces – like for vacuum circuit breakers equipped with an arc control system – or
stays in concentrated mode due to pinch forces.
Another principal difference between the pre-arc of a high frequency back-to-back and a single bank
inrush current exists because of the frequent polarity reversals in the first case and the unipolar behavior
in the latter case. This is treated in a later section.
Also, there might be differences in the making performance of circuit breakers and that of other
capacitive current switching devices. Load switches or dedicated capacitor switches might have a higher
probability of being welded under inrush currents, since they are not designed for a good short-circuit
current making ability as circuit breakers.
4.1.2. Breaking operation
A single-phase operation of a switching device breaking the capacitive current of a capacitor bank in a
circuit as shown in Figure 21 will stress the interrupter contacts with a high recovery voltage (see Figure
22). Neglecting the impact of the source inductance, the recovery voltage reaches a peak of 2 p.u. at
10 ms (for 50 Hz) or 8.3 ms (for 60 Hz) after current-zero. This is because the load capacitor will
maintain its voltage over seconds whereas the source voltage continues to oscillate with power
frequency. Therefore, the recovery voltage is a quasi-direct voltage. The effect of the source inductance
increases the voltage across the capacitor (Ferranti effect) compared to the actual source voltage. This
leads to an initial voltage jump4 (voltage kick) initiating an additional source side transient recovery
voltage with a frequency of more than 1000 Hz. It is caused by oscillations between the elements CN
and LN which disappear within several milliseconds. The total recovery voltage is shown in Figure 22.
In a three-phase system, current interruption and therefore the recovery voltage also depends on the
neutral point treatment. In case of a non-solidly earthed neutral system, the recovery voltage of the first-
pole-to-clear is increased by 0.5 p.u. or even more. In addition, special switching arrangements like
4 The standards e.g. IEC 62271-100 do not mention this particular feature of the recovery voltage. However, it is
implicit through the network impedance related to the short-circuit current.
36
TB 817 - Shunt capacitor switching in distribution and transmission systems
controlled switching or mechanically staggered poles will affect the recovery voltage. This subject is not
further discussed here. More information on this topic can be found in IEEE C37.012 [19].
37
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 23: Voltage multiplication due to restrikes in a capacitive circuit, UB is the voltage across the
capacitor, US is the voltage across the switch.
38
TB 817 - Shunt capacitor switching in distribution and transmission systems
assumed for up to 332 or 484 random operations (assuming an acceleration factor of 4 or 6, respectively,
and adding 28 operations with distributed arcing times).
Also, with respect to the inrush current, the type tests pose more stringent conditions than in the field.
First, the maximum inrush current during a class C2 type test shall always occur in the same phase that
experiences the minimum arcing times. Second, the closing phase is specified to lie in an interval of
± 25° around the peak voltage, whereas in the field, the inrush current peak varies in dependence of the
point of closing. Under the conditions specified in IEC 62271-100, an acceleration factor of 1.536
results5. For IEEE C37.09, which describes an interval of ± 15°, the factor is 1.568. Thus, the more
stringent conditions in type tests come mainly from the use of small arcing times and to a lesser extent
from the focus on the highest inrush currents.
Restrikes and associated overvoltages might damage capacitor banks, the switching device or any other
equipment connected to the switch. Therefore, the restrike probability should be as low as possible
and/or appropriate mitigation means applied. These could comprise surge arresters or spark gaps,
controlled switching or a higher rated voltage of the equipment.
The probability of restrike may depend on several circuit parameters such as the peak and frequency of
the inrush currents, the applied voltage and the capacitive currents. Since the impact of these
parameters vary with the switching medium and device type, conclusions regarding the restrike
probability have to be differentiated with respect to the medium used for current interruption in particular
vacuum and gas (SF6).
Conditions originating from the switching device affecting the restrike probability are provided in [2]:
Contact opening speed;
Mechanical wear of contacts and nozzles;
Contact and nozzle erosion after short-circuit current interruption;
Erosion of contacts and nozzles due to prestrike and restrike arcs;
Coating and deterioration of insulating material and extinguishing medium;
Statistical scatter given by the physics of dielectric breakdown;
Scatter of the conditions from one operation to the other.
These parameters and conditions need to be investigated when assessing the performance of a
particular switching device with respect to the restrike probability in the field based on the performance
determined in the laboratory. It has to be remarked that the conditions listed above fit well to gas
interrupters, less to vacuum devices.
5 Considering a sinusoidal voltage and neglecting a preference of higher voltages due to pre-ignitions, the
acceleration factor can be calculated from the mean amplitude of inrush currents in a certain interval of closing
angles φ. Since the voltage varies with cos φ, the mean value is determined by the integral of cos φ (i.e. sin φ
taken at both ends of the interval) divided by the interval 2 φ. The standard prescribes an interval of ± 25° around
the peak voltage, whereas for random switching an interval of ± 90° can be assumed. Thus, an acceleration factor
of 1.536 results.
39
TB 817 - Shunt capacitor switching in distribution and transmission systems
phase controlled and therefore making occurs at random moments in time giving a statistical spread of
inrush current peaks distributed over all three phases. This statistical distribution would allow multiplying
the number of operations of a type test by a factor of 1.536 (see 4.1.3) creating the same impact on
erosion and wear of contacts as observed in this particular type test. Admittedly, this is not a large effect.
However, if the inrush current peak by itself is much smaller than 20 kA, this should have a positive
impact on the performance.
The inrush current frequency (requirement no. 2) depends on the particular capacitive circuit in the field.
As shown in the subsequent paragraph, peak and frequency of the inrush current are closely related to
each other. An inductance in the current path reduces the peak inrush current and coherently decreases
the frequency and vice versa. Therefore, the defined inrush current frequency is more a test specification
than a physical impact parameter. Only for oil breakers it is known that the inrush current frequency has
an impact with respect to the generation of shock waves in the liquid which could damage the nozzle
system. Due to this impact on interrupting parts of oil breakers, the standards do not allow higher
frequencies. With respect to other interrupting media, the sense of an upper limit of the inrush frequency
can be questioned, which is further explained below.
Finally, test requirement no. 3 concerning the damping ratio of the inrush current needs to be
considered. Since the damping factor is determined by the resistance of all elements in the capacitor
circuit, e.g. the cross-section of conductors, it is difficult to predict damping factors in the field. However,
it can be assumed that the standards have introduced values being representative.
Figure 24 shows an example of inrush currents measured during switching of capacitors in back-to-back
arrangement. The oscillogram in the middle displays the complete switching sequence with the source
voltage in the lower trace collapsing at the moment of pre-ignition, the subsequent capacitive current in
the upper trace and the recovery voltage (again in the lower trace) after current interruption and one late
restrike. The oscillogram on the left shows the zoomed inrush current after pre-ignition, and the
oscillogram on the right shows the zoomed discharge current after the restrike extinguished after 2 ms.
In some cases, only observed for vacuum circuit breakers, the inrush current is interrupted during the
pre-arc phase so that the inrush current trace occurs twice.
Figure 24: Oscillogram of a capacitive switching sequence during a single-phase back-to-back test of a
36 kV vacuum circuit breaker [22]. Current is displayed on top [kA] and voltage across the interrupter in
the bottom [kV]); time is counted in [ms].
The standards assume that back-to-back capacitor switching tests cover switching of single capacitor
banks. This is suggesting that the restrike performance of circuit breakers is similar or even better for
single bank switching compared with back-to-back capacitor switching. Indeed, the inrush current peak
during the energizing of single capacitor banks with a capacitive current of less than 400 A is normally
less than 5 kA. It may reach higher values with larger capacitor banks and higher short circuit power of
the network (Figure 25). Nevertheless, it is always smaller than for back-to-back switching where 20 kA
is applied as standard. However, the frequency is very different namely in the range of several 100 Hz
for single bank and several 1000 Hz for back-to-back capacitor switching. Inrush frequencies will be
treated later.
40
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 25: Calculated peak inrush current and trendline as function of capacitive current for single bank
switching; variation comes from different short circuit levels
Figure 26 shows the relative occurrence of capacitive test series in a large power laboratory with
restrikes or with NSDD differentiated by rated voltage and single or back-to-back application [22]. Since
it is known that the vast majority of switchgear up to and including 40.5 kV uses vacuum circuit breakers,
these results are mainly applicable to vacuum interrupters. The similar percentage of restrikes and
NSDD for single and back-to-back banks seems to suggest that the inrush current peak has no impact
on the restrike probability, which opposes common sense. Due to the relatively small number of back-
to-back test series (19), compared with the number of single bank tests (83), the conclusion should be
taken with care [22].
re-strike
occurrence
NSDD
occurrenc
e
Figure 26: Percentage of test-series in which re-strike (top) and NSDD (bottom) occurred [22]
41
TB 817 - Shunt capacitor switching in distribution and transmission systems
In this equation, the pre-arc duration is called tarc, the natural frequency of the inrush current is finrush and
the damping time is τ. Approximately, the integral in the equation can be evaluated by the integral of two
half-cycles of the inrush current multiplied by a series of powers of the corresponding damping factor,
D, i.e. the ratio of two subsequent peaks with same polarity. The series is abbreviated at the nth cycle
counted within the pre-arc duration tarc.
Figure 27: Example of waveshapes of back-to-back capacitor (blue) and single capacitor (red) inrush
currents and calculated integral of inrush current as function of time;
left graph calculated with a damping factor of 0.75, right graph with 0.85.
The value of the ICI depends on the number of half cycles of inrush current before mating of the contacts,
which is determined by the duration of the pre-arc. The integral can be calculated from Equation 2 or
Equation 3, if the pre-arc duration is known. The pre-arc duration is a function of the device type,
interruption medium, closing speed and voltage at the moment of pre-ignition. Typical pre-arc times
range from 1 to 3 ms.
In Equation 3 for the ICI, the ratio of Iinrush to finrush is the proportionality factor. For back-to-back switching,
this ratio is a function of the capacitance C1 of the switched-on capacitor, the capacitance C2 of the
charged capacitor and the rated voltage [19]. The series inductance between the two capacitor banks
drops out of this equation.
The size of the capacitors determines not only the rms value of the capacitive current, but also the peak
inrush current. Therefore, the ICI depends on the capacitive currents, too. Figure 28 shows that the
inrush current integrals for back-to-back and single bank capacitor switching - assuming a pre-arc
duration of 1 ms - are quite similar (50 % higher) if the peak inrush current is in the range of 10 kA, which
is represented by the lower blue trend line in the figure. For a peak inrush current of 20 kA (upper blue
trend line in Figure 28), the ICI is larger for back-to-back switching by a factor of 2 to 3 compared to
single capacitor switching. Table 6 provides ICI values in more detail for different capacitive currents
(related to the size of the switched capacitor bank), inrush current peaks and damping factors and
confirms the conclusions taken from Figure 28. Two remarks should be given to this table: The short-
circuit current, Isc, only has an impact on the ICI for single capacitor switching, and the inrush frequency
results from the adaptation of the series inductance in each capacitor branch to the selected inrush peak
current.
During a making operation in the field and during direct generator testing not only the back-to-back
inrush current outrushing from the already energized capacitor, but also the “single bank” inrush current
from the source is flowing. The superposed single bank current does not change the amplitude of the
back-to-back inrush current but creates a modulation with the smaller frequency and peak of the single
42
TB 817 - Shunt capacitor switching in distribution and transmission systems
bank inrush current. During the pre-arc, the impact of this superposition can be neglected, however,
eventually the generator puts additional charge into the two capacitors.
It has to be remarked that for high inrush current frequencies, the transient currents may fade completely
away during the pre-arc. If the pre-arc re-ignites before the contacts mate, the inrush current will flow
again so that the ICI might even be doubled.
For single capacitor banks, only one or two half-cycles occur during the pre-arc.
Figure 28: Inrush current integral (ICI) displayed for different capacitive currents separate for back-to-
back switching with inrush peaks of 10 kA and 20 kA (blue lines) and single bank switching with 25 kA
and 50 kA rated short-circuit current (red lines); ICI depends on the damping factor (here 0.85) and the
pre-arc duration (here 1 ms)
Table 6: ICI values at a pre-arc duration of 1 ms and calculated inrush frequency f in dependence of
capacitive current Ic, back-to-back inrush current and damping factor D
Isc = 25 kA Isc = 50 kA
43
TB 817 - Shunt capacitor switching in distribution and transmission systems
44
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 29: Three-phase oscillograms of a capacitive breaking operation showing the difference between
non-sustained disruptive discharge (NSDD) and re-strike
Breakdown of the recovery voltage at small contact distance i.e. after short arcing times might be only
temporary, if the high frequency discharge current is interrupted by the device. If not, the 50 Hz
capacitive current will be resumed. After one further half cycle of the power-frequency current, the
contact distance has become much larger and the probability of a re-ignition or restrike is almost zero.
SF6 interrupters have a quite good predictability in this respect, whereas vacuum interrupters sometimes
also show late restrikes after many milliseconds.
The breaking of capacitive currents normally is not a problem for any interrupting medium unless load
switches are considered, since the breaking capability of circuit breakers is designed to interrupt much
higher short-circuit currents. The arc plasma remaining at current-zero, when switching off a capacitive
current, is mostly negligible or vanishes within microseconds. Contrary to the rapid rise of recovery
voltage after short-circuit interruptions, the recovery voltage for capacitive loads increases slowly in 8 to
10 ms. Therefore, normally the “cold” dielectric withstand voltage is applicable during the recovery
phase. Thermal interruption failures do not occur.
Vacuum devices show a much larger scatter of breakdown voltages and restrike occurrences than SF6
devices. In vacuum, breakdown is essentially determined by the contact surfaces i.e. material quality,
roughness, and potential micro-protrusions with high local field strength and associated high field
electron emission. One possible cause for these dielectrically weak points on the surface are ruptured
welds originating from the making operation with high inrush current. One other reason for restrikes
comes from particles released from deteriorated contacts in particular at higher rated voltage, but also
particles ejected from the cathode spots of the breaking arc.
For SF6, the dielectric withstand of the contact gap is mainly determined by the properties of the gas
combined with the electric field strength in the gap. Contact erosion changes the electric field in the gap
and the point of physical contact, but also the flow and local pressure of gas through the nozzle during
an opening operation, which might have an impact on the dielectric withstand capability. Another impact
might come from the sharing and coordination of voltage between main and arcing contacts, involving
the nozzle.
Further details on the restrike process separate for vacuum and SF6 interrupters can be found in chapter
4.4. Oil interrupters are not considered, since they exist only as legacy devices.
45
TB 817 - Shunt capacitor switching in distribution and transmission systems
46
TB 817 - Shunt capacitor switching in distribution and transmission systems
recommends users to only adapt to the inrush current peak. The inrush current frequency results from
the applied inductances and capacitors and does not play a role unless oil breakers are applied6.
VI: There is no influence on the switching performance reported except on the premature extinguishing
of the inrush current and re-ignition before contact touching for high frequencies.
SF6: For SF6 breakers no influence is reported in the range of commonly applied inrush current
frequencies. Damage of design parts is not likely.
Oil: The inrush current frequency has a severe impact on the performance of minimum oil breakers due
to possible shock waves in the liquid and subsequent damage of nozzles. Inrush frequencies should
not exceed the tested one.
4.3.4. Inrush current integral
The inrush current integral (ICI) plays a role in the erosion and wear of contacts during the making
operation. The ICI multiplied by the arc voltage is the energy converted into heating and melting of the
contact surfaces in the vicinity of the prestrike. From Table 6, it is obvious that the ICI generally
decreases with lower breaking currents and increases with higher inrush current peaks. This justifies
that inrush current peaks higher than type tested cannot be permitted in the field, unless it can be proven
that the deterioration of the surface is smaller than determined for the type test. The ICI as indirect
measure of contact erosion might be applied to prove equivalence. However, there are only rare
combinations of capacitive current and peak inrush current where this could be utilized7.
VI: The ICI is a measure of the amount of contact material melted during the pre-arc and may give an
indication of potential deterioration of the contact surface.
SF6: Extended test series show a relation between the inrush current integrated over the duration of the
pre-arc and the local erosion and wear of the arcing contacts and subsequent change of its
dielectrically relevant shape. Erosion of contacts can contaminate the arcing system and influence
the restrike probability. Also, nozzle erosion has an impact.
4.3.5. Capacitive breaking current
While most type tests are carried out with a capacitive current of 400 A as recommended by the
standards, the actual capacitive breaking currents in the field may be higher or lower. Lower currents
are covered by the type test procedure described in the standards and therefore can be applied in the
field without concern. Higher currents deserve special consideration. The working group concludes that
up to 50 % above tested values should give the same performance as verified for the rated current
value. It should nevertheless be ensured that the ICI value related to the applied inrush current is smaller
than tested to ensure the same or smaller wear of contacts. Arcing durations of the breaking current will
be longer than in type tests, where a lot of attempts are performed with minimum arcing times.
VI: The cathode spots of the breaking arc erode and thereby condition the contact surface by
smoothening dielectrically weak points. Field emission currents originating from such weak points
are known to be considerably reduced. Thus, for vacuum interrupters, the restrike probability is
reduced with higher capacitive currents and/or longer arc durations and the switching performance
is thus improved.
SF6: The erosion and wear due to inrush currents during the making operation and the resulting metallic
dust mainly determine the dielectric withstand of the contact gap and thereby the restrike probability
during the breaking operation. The amount of contact erosion caused by breaking currents is much
smaller than that coming from inrush currents. Higher breaking currents even have a beneficial effect
since minimum arcing times tend to become longer so that the contacts are farther apart when the
peak of the recovery voltage appears resulting in a lower restrike probability. However, an impact
might come from the inrush current integral increasing with higher breaking currents (see Table 6)
which therefore should be evaluated.
6 The capability of circuit breakers to handle inrush currents is in some IEEE standards expressed in terms of the
product of inrush current peak times the inrush current frequency (kA*kHz). However, this parameter lacks
physical rationale except for oil breakers.
7 For example, the ICI at 400 A and 20 kA inrush current is slightly larger than the ICI at 300 A and 22,7 kA (9,61
47
TB 817 - Shunt capacitor switching in distribution and transmission systems
48
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 30: Breakdown events during the sequence of closing and opening of a vacuum interrupter
It is generally observed that the electric field strength at which a pre-ignition occurs has first a
considerable scatter or statistical variation, which is much larger for vacuum interrupters than for SF6
interrupters, and second can achieve quite low values during frequent operations [20], [24], [25]. The
pre-ignition field strengths scatter over more than a factor of three of the lowest field strength (Figure
31). The mean of all pre.ignition field strengths steadily decreases in the course of the first 50 operations
[26], which has the consequence that the pre-arc duration is increasing. After these 50 operations, the
observed lowest field strength does not decrease further even after hundreds of operations. The further
decrease of the trendline in Figure 31 is not caused by a further reduction of the minimum pre-ignition
field strength, but by the lack of higher field strengths. The smallest breakdown fields in the tested
vacuum interrupters were as low as 3.3 kV/mm whereas the minimum withstand electric field strength
achieved in technically processed vacuum interrupters for field equipment is larger than 8 kV/mm (e.g.
80 kV AC withstand voltage applied on a 14 mm gap). From other experiments using a lot of different
contact materials [23] it is concluded that these low values only occur after no-load opening of contacts
subject to inrush currents. After switching of currents of several 100 A, the minimum field strength is in
the range of 5 kV/mm. The thesis [23] provides a comprehensive overview over most of the phenomena
occurring during capacitive switching.
Figure 31: Scatter of pre-ignition field strength as function of number of operations at different vacuum
contact gaps [20]
In a vacuum interrupter the decreasing field strength results from the deterioration of the contact surface
by the pre-arc of the inrush current. It is anticipated that the higher the pre-arc current peak and the
longer the pre-arc duration are (i.e. the larger the inrush current integral, ICI), the deterioration of the
contacts becomes more severe. The pre-arc might be quite concentrated with a current density in the
arc root of several 108 A/m2, which is able to melt both contact surfaces locally in a short time.
On the other side, one also has to consider that the pre-arcs spread out rapidly with the speed of cathode
spots of approximately 10 m/s and may have lower current densities in less than 1 ms [27]. For a 2 kA/
168 Hz inrush current, the latter paper measured a relationship between the pre-arc duration and the
DC breakdown electric field after opening of the gap to a distance of 1 mm. The breakdown field is used
here as a measure for the amount of contact melting and welding due to the inrush current. The
breakdown field shows a minimum at 1 ms pre-arc duration, which coincides with a maximum surface
49
TB 817 - Shunt capacitor switching in distribution and transmission systems
melting depth and width calculated from the recorded spatial extension of the pre-arc and associated
arc current. Also, the cathode spot traces photographed on the contacts of a demounted vacuum
interrupter show a considerable spreading and diffusion of the pre-arc [28]. The severe melting spot
originating at the pre-ignition spot obviously lasts less than 1 ms. This justifies using a pre-arc duration
of only 1 ms in the evaluation of the ICI in Table 6.
Such a melting spot is not only present during the low frequency, low peak current for single bank
switching, but also during the high frequency, high peak inrush current for back-to-back bank switching.
It is anticipated that also these inrush currents are only concentrated at the beginning and then spread
out on the contacts during the decreasing current pulse. However, most published experiments use
single capacitor bank inrush currents with low frequency instead of back-to-back capacitor inrush
currents. Some papers exist where high frequency inrush currents were applied such as [29] and [30].
In the latter paper, the arc appearance was recorded by a high-speed video camera showing the
expansion of the arc in the first half-wave of a 10 kA / 3800 Hz inrush current. The expansion speed
reaches 100 m/s and gives current densities not much higher than calculated for single bank inrush
currents. Thus, heating of the surface could be in the same order of magnitude as for single bank
switching. However, this needs further experimental work.
After the contacts touch, the melted areas engage, solidify and might form a welding spot. Sometimes
bouncing of the contacts is observed with durations up to 5 ms after galvanic contact, which might
rupture and re-melt these welding spots. It seems that the pre-ignition always occurs at the same spot
on the contacts where the surface was deteriorated by previous pre-ignitions. However, sometimes also
two or even three affected surface areas are detected [23], which are beneficial in terms of the
performance since they distribute the deterioration of the contacts to a larger area. Also, the high current
pre-arcs move on the surface driven by the electro-magnetic forces generated by the contact structure
[23] distributing the heat to a larger area. It is expected that after hundreds or even thousand operations,
the whole contact surface shows extensive melting traces in dependence of peak inrush current and
inrush current integral (ICI).
The question is to which extent a considerably melted surface of the contacts impacts the switching
performance after 100 or even more operations and in a typical application in the field where perhaps
more than 1000 operations are performed. Besides small melted areas of several 10 up to 100 mm2, an
essential material build-up is often observed on one of the contacts correlated with an excavation on the
opposite contact [23], [31], [32], [33]. This macroscopic protrusion can reach several millimeters in height
and grows only on one of the two contacts irrespective of the polarity of voltage during the making
operation. The reason for this asymmetric behavior is that the welding spot is not ruptured in the center
plane of the melt volume, but rather at one of the boundaries to the bulk, non-melted contact material.
Once a preference of one side has been established, it remains for the whole switching series. The
electric field at this protrusion increases with the height of the protrusion. [26] has indicated an increase
of the restrike probability after approximately 200 operations with the build-up of a 4 mm protrusion in a
12 mm gap perhaps related to the increase of the macroscopic field strength at this protrusion. In other
publications such an increase of the restrike probability was not observed in spite of many test series
[23]. The authors of [34] found no increase of the restrike rate even after the application of more than
2000 making operations with an inrush current peak of 4.8 kA at 240 Hz.
In general, the build-up rate of the observed macroscopic protrusion seems to be unrealistically high
compared with modern vacuum interrupters and is perhaps related to flat contacts with no arc control
and favored by no-load opening operations or very small breaking currents (< 1A). Also, the closing
speed of the mechanism seems to play a role [26]. Nevertheless, some melting of the interrupter
contacts is anticipated after numerous making operations. Since contact melting due to arcing is
simulated by the application of test duty T60 according to IEC 62271-100, i.e. three short-circuit
interruptions for class C2 circuit breakers, the impact on the restrike probability should be covered by
type tests.
4.4.2. Breaking phenomena observed in vacuum interrupters
In vacuum interrupters, voltage breakdown is determined essentially by surface effects. Mainly two
breakdown effects are considered by the experts quite controversially: instability (explosion) of a
protrusion of sub-µm size on the contact surface or the impact of sub-mm particles on the surface. The
sub-µm protrusions, called micro spots in the following, are not the same as the protrusions due the
build-up of contact material at the pre-ignition site, but are small spots distributed on the surface. Such
micro spots might be metallic protrusions, exposed grain boundaries, impurity inclusions or machining
remainders which might have been created or uncovered from the bulk contact material and which could
50
TB 817 - Shunt capacitor switching in distribution and transmission systems
induce voltage breakdown. These micro spots increase the overall microscopic field strength locally by
orders of magnitude and may reach amplification factors of 1000 [29], [23].
However, also particles can be a potential cause of breakdown by either the impact itself as proposed
by Cranberg [35] or by modifying the surface topology on impact. Observations of glowing particles on
video films or of artificially produced particles in back-scattered laser light show particles bouncing
between contacts or temporarily detaching from a surface [36]. These experiments disclose breakdown
events in the moment particles hit the cathode, leave the cathode or are close to the cathode. The mass
of the particles is far too large and the applied voltage is too small (50 kV) to achieve speeds way above
10 m/s, which would allow breakdown according to the Cranberg hypothesis. Instead, these particles
seem to trigger breakdown of micro spots on the surface, which would otherwise not become unstable.
These breakdown processes are not yet fully understood.
When the contacts are opened, the solidified melting spot on the interrupter contacts due to the
preceding making operation is ruptured initiating cathode spots of a vacuum arc. Depending on current
peak and duration of the breaking arc [37], the ruptured weld is eroded. The erosion of the surface due
to cathode spots is considerable and can be recognized from numerous traces on the surface. In
absolute values, however, it is quite negligible. A typical erosion rate of 20 µg/As [38] - related to the
current integral of the breaking arc - results in an erosion depth of 32 nm for a 500 A diffuse arc. Thus,
some parts of the weld fractures normally remain. These fractures might become the origin of a
breakdown of the recovery voltage i.e. a re-ignition or restrike. They might also be the origin of a pre-
ignition in the subsequent closing operation. Indeed, the coincidence of pre-ignition and restrike sites
was macroscopically observed on interrupters with optical access [20]. However, this does not mean
that microscopically the same micro spots were involved.
The most important question is whether there is a correlation between the inrush current peak or integral
(ICI) of the pre-arc and the probability of restrikes. In [20], a correlation has been claimed between the
cumulated distribution Ppre(E) of pre-ignition field strengths E – evaluated from the ratio of applied DC
voltage and observed pre-arc contact distance dpre - and the restrike probability Rre(Ûre,d) at the peak
recovery voltage Ûre and full contact gap d. The author compares this restrike probability with the
cumulative probability of pre-ignitions at an electric field strength given by the highest possible
breakdown field during recovery i.e. Ûre/d.
𝑈 𝑈
𝑅𝑟𝑒 ( ) ≈ 𝑃𝑝𝑟𝑒 (𝐸 < )
𝑑 𝑑
Equation 4: Equivalence of restrike probability and cumulated probability of pre-ignitions at the same
field strength
This correlation would suggest the relevance of electron field emission induced processes for restrikes
assuming that the breakdown electric fields during the making and breaking operations are the same.
However, the breakdown processes even at the same field strength are probably different at the large
gaps during recovery from those at small contact distances where pre-ignitions take place. The
numerous tests performed in [23] reveal restrike probabilities being below or above the distribution of
pre-ignition field strengths. Equation 4 only proves that field emission induced breakdown processes
are feasible during capacitive switching, but they are not necessarily the only ones.
On the other side, high field emission (FE) currents were indeed measured after making of inrush
currents [29], [39], [22]. The FE currents increase with the peak of the inrush current, however, not
necessarily with the number of making operations occurring with the same peak current [39]. A lot of
basic investigations in particular those allowing visible access were performed in evacuated housings,
others were performed in commercially produced sealed-for-life interrupters. The peak of FE currents
was generally a factor of 2 lower in commercially produced interrupters compared with non-regular
interrupters of the same design, however, manufactured intentionally under non clean-room conditions
[39]. Also, only the non-regular devices showed large fluctuations of the FE current caused by
mechanical operations and restrikes with a probability of approximately 15 % irrespective of the FE
current peak, whereas the commercial interrupters showed stable FE currents and no breakdown at all.
The same difference between a device evacuated by a vacuum pump and commercially processed
interrupters was revealed in [40], where the evacuated device showed voltage breakdown due to
mechanical shock waves produced by the operating mechanism, whereas the vacuum interrupter didn’t
show any breakdowns at all. Mechanical shocks due to the opening mechanism which may last for more
than 100 ms seem to initiate voltage breakdown in particular for devices which have not been processed
under the best clean room conditions as for commercial vacuum interrupters.
51
TB 817 - Shunt capacitor switching in distribution and transmission systems
In these investigations [40], it was also shown that the voltage breakdown probability decreased by a
factor of 2 to 4, when the interrupters were supposed to several short-circuit current interruptions. This
would imply that the impact of the application of test duty T60 (60 % of rated short-circuit current) as
pre-conditioning for class C2 type testing is rather beneficial than deconditioning. Melting of the contact
surface seems to rather increase the dielectric voltage withstand than decreasing it.
The observations above show the importance of a high-quality interrupter manufacturing. The restrike
probability to some extent depends on the manufacturing process of the device i.e. on the cleanliness
and composition of contact materials, but also on clean room conditions and soldering at high
temperatures. The results also indicate that a restrike during capacitive switching and a voltage
breakdown after mere touching of the contacts and no-load opening have the same origin, which is
reasonable since the plasma of the breaking arc has vanished for long before restrikes happen.
Publications [39] and [22] show that the arcing duration and the amplitude of the capacitive breaking
current have a large impact on the amplitude of the FE currents and therefore on the building of micro
spots. The vacuum arc ignited by opening the contacts erodes micro spots and diminishes the FE
current considerably. A capacitive current of 20 A has almost no effect, since it comprises only one
single cathode spot, whereas a current of 500 A with more than ten cathode spots reduces the FE
current by orders of magnitude in dependence of the arcing time. A 2.7 ms arc duration is less effective
than a 11 ms arc duration, however, still significant. The same observation was made in [29], where the
cumulative breakdown probability taken at a voltage of e.g. 30 kV decreased from 54 % after no-load
openings to 10 % after breaking of 1000 A currents with 7 ms arcing time. In parallel, the FE currents
were reduced by a factor of 20.
Also, the field emission current itself conditions the surface. In [20], it was observed that the prestrike
distribution was dependent on the fully open contact gap, which was either 8, 12 or 14 mm, though the
largest distance where pre-ignitions occurred was 6 mm. Therefore, the fully open gap should have had
no influence on the pre-ignition process. The conditioning effect removes micro spots with low pre-
ignition field strengths and therefore shifts the distribution in the mean to higher field values. This effect
is obviously caused by the high field emission currents flowing at smaller contact distances in the time
between an opening operation and the subsequent closing operation. The time under voltage application
is in the order of seconds for the experiments and may last for hours for applications in the field. This
conditioning effect also had a large impact on the restrike probability in these experiments [20], since
the extrapolation from the pre-ignition distribution obtained at a stroke of 14 mm (with small conditioning
effect because of small FE currents) would have resulted in a restrike probability of almost 50 % at a
gap of 8 mm. Instead, the restrike probability was only 12 %. This is understandable for late restrikes
because of the removal of micro spots with low pre-ignition field strengths by high FE currents before
the breakdown. However, the reasoning is difficult to apply for the first recovery voltage peak, since
there is too little time for conditioning. Also as explained above, there are other breakdown processes
than field emission induced ones.
Indeed, a one to one correlation between pre-ignition and restrike process was never observed as stated
in several papers [39], [26], [23]. Restrikes do not necessarily occur after long pre-arcing times which
should produce a strong welding spot and a particularly ragged weld rupture. Also, a restrike does not
necessarily improve the surface quality and result in an exceptionally short pre-arcing time. On the
contrary, it was observed [23] that FE currents even increased after a breakdown indicating a further
de-conditioning of the surface. In a lot of cases, such a breakdown caused subsequent restrikes in the
same operation. Thus, the micro spots responsible for pre-ignitions and for restrikes must be different,
if they are responsible for breakdowns. This is not yet an argument in favor of particle induced
breakdown, since it is known that surfaces contain hundreds of micro spots each of them could create
FE currents but also breakdown. On the other side, there are hundreds or thousands of particles
produced during the breaking of capacitive currents [41], [42]. The restrike probability would be far too
high unless only a very small percentage of the particles had the ability to cause breakdown. And this
hypothesis is contrary to the observation that higher capacitive currents even reduce the restrike
probability compared with no-load operations due to the erosion effect of the same cathode spots
producing numerous particles.
It has to be admitted eventually that the origin of restrikes after capacitive switching is still uncertain as
are the physical processes causing in general voltage breakdown in vacuum.
More information on capacitive switching with vacuum interrupters, which hasn’t been addressed above,
can be found in references [48 to 58].
52
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 32: Erosion marks after 2400 operations with pre-arcs with exceptionally 100 kA /25 kHz on arc
extinguishing contacts of an SF6 circuit breaker [44]
For pin – tulip style contacts, the geometry is progressively modified with successive making operations.
Since the pin contact is typically of a rounded shape and the tulip contact of a cylindric one, the first
breakdown happens at the closest distance between the contacts, not necessarily at the tip of the pin,
but rather at the circumference of the pin towards the tulip contact. For the high inrush current
frequencies, the pre-arcs remain small in diameter so that the current density is increased. This results
in a high temperature of the arcing spot. The position of the pre-arc varies around the pin and tulip from
operation to operation. The breakdown pattern has a statistical scatter. This means for typically
concentric contact systems as shown in Figure 32 that the pre-arc will pre-ignite at the circular area of
lowest dielectric withstand strength. Each making operation will cause a slight erosion on both contacts
involved. The arc extinguishes when the contacts touch. The total wear of the arcing contacts is not only
due to material erosion or evaporation, but also – and sometimes to an even larger extent – by
mechanical abrasion of material melted and solidified after the pre-arc [45]. The closing speed and
alignment of the contacts play an important role. The extreme off-center wear of the pin shown in Figure
32 is believed due to misalignment of the contact system.
53
TB 817 - Shunt capacitor switching in distribution and transmission systems
The pre-arcing duration can increase with progressing geometrical modification. The pre-ignition field
strength doesn’t change much by contact modification, but the time between first pre.ignition and full
galvanic contact might be extended compared to the as new geometry. Thus, the pre-arcing duration
increased in the above experiment from 0.25 ms to 1.1 ms after 2500 operations though this resulted
only in a 23 % increase of the ICI and pre-arc energy due to the fast damping of the current.
As a first approximation, erosion can be expressed as the consumption of the pre-arc energy (i.e. the
ICI multiplied with the cathode or anode voltage drop) in the evaporation energy of the material. The
theoretical mass loss in this case can be obtained from the ratio of pre-arc energy to the specific
evaporation energy of the contact material (4820 kJ/kg for Tungsten). This would result in a specific
mass loss of approximately 3.1 mg/As proportional to the ICI assuming a cathode voltage drop of 15 V.
Typical erosion rates for concentrated SF6 arcs are in the order of 500 µg/As for WCu arcing contacts
as determined in [45]. The total mass loss in the above experiment was 20 g after 2500 operations with
an ICI of approximately 10 As per discharge, which would result in a similar specific mass loss of 800
µg/As. However, this amounts to less than 25 % of the theoretical erosion rate estimated above, which
only takes evaporation and even not mechanical abrasion into account. Therefore, it must be assumed
that most of the pre-arc energy is consumed in heating up the surface without reaching boiling
temperature when the pre-arc spreads out from the ignition point.
The insulating material of the nozzle can be eroded by the intense plasma and radiation of the pre-arc,
but also during breaking operations with capacitive current. The wear of nozzle parts might make the
voltage coordination between the main and arcing contacts unsafe with a possible breakdown across
the main contacts during current breaking, but also current making. The pre-arc sometimes can occur
between one main contact and one arcing contact puncturing and damaging the nozzle, due to lack of
dielectric coordination during the closing movement.
For class C2 type tests, contact erosion is simulated by the application of T60 i.e. interruption of three
short circuits with 60% of rated short circuit current before the capacitive test duties. The total amount
of erosion is less than that observed after the 2400 inrush current making tests described above. The
integral of arcing current during T60 (e.g. 525 As for three 25 kA current interruptions with a mean
arcing time of 7 ms) is much lower than the cumulated inrush current integral in the tests of more than
10 kAs. Also, T60 mostly erodes the arcing tip instead of the cylindrical arcing contact, since the contacts
are quite distant during interruption. However, for the normally applied 104 operations in type tests, the
current integrals are comparable.
It can be concluded that pre-arcs erode the geometry of the arcing contacts (pin and tulip) after more
than 1000 operations considerably in dependence of the applied inrush current and inrush current
integral. The erosion can be estimated from the pre-arc energy, which is evaluated from the ICI and a
reasonable cathode voltage drop. However, a direct correlation is difficult to achieve because of the
uncertainties with respect to the spreading (current density) of the arc, pre-arc duration and applicable
voltage drop. Standards up to now only consider the peak of the inrush current, which does not provide
the whole picture either.
54
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 33: Voltage withstand (BIL) between the arcing contacts of an SF6 breaker after electrical
endurance tests with different total pre-arc energies [44]
In particular the erosion of the arcing contacts decreases the withstand voltage between the contacts
after numerous making operations. In Figure 33, the withstand voltage measured at different contact
distances has been converted into times after contact separation applying the known opening speed of
the contacts. The figure shows the lightning impulse withstand voltage determined after a complete
series of making tests for different pre-arc energies accumulated over the whole endurance test with
different inrush current peaks and frequencies. The two smooth cosine functions in the figure represent
the recovery voltage after current-zero in an ungrounded and grounded network, respectively. The black
curve (25 kA, 6.6 kHz, 2500 CO) resulting in an ICI of almost 20 kAs intersects the recovery voltage
curve at 3.0 ms. It is anticipated that the capacitive switching test would have generated re-ignitions,
perhaps restrikes in this case. Due to the small capacitive current, SF6 breakers are able to interrupt it
with very small arcing time, implying that the 1-cos recovery voltage develops across a small gap.
Publications [46] and [47] calculate the electric field distribution at all contact distances during an
opening operation and apply the streamer criterion for the breakdown field. Due to erosion of the
contacts, the electric field is changed, and the recovery voltage might exceed the withstand voltage
leading to a re-ignition or a restrike. In addition, the flow of gas during the opening needs to be
considered, since the erosion of the arcing contacts, but also that of the nozzle, might change the local
gas density during opening with an impact on the streamer criterion. Endurance tests with different types
of breakers showed a first restrike due to contact erosion after 200, 1000 and 1200 operations
depending on the design of the interrupters [43]. Applying point-on-wave control for the closing operation
reduced the erosion of the contacts of these breakers considerably and resulted in the appearance of
the first restrike not before 3000 and 5000 operations.
The material evaporated by the pre-arc might have an additional impact on the insulating behavior of
SF6 interrupters. Metal vapor can settle on insulating parts and reduce the withstand voltage. Also,
space charges could accumulate on insulating surfaces, which might change the local voltage
distribution and initiate spontaneous breakdown at later times. Solid insulation materials like nozzles
can hold space charges even from previous switching operations with an impact on the dielectric
behavior of the nozzle. Compared to the situation during testing, an SF6 circuit breaker in the field
discharges space charges slowly over time.
For the nozzle close to the arcing contact system, the erosion of previous operations plays a role, also
regarding early breakdowns i.e. re-ignitions. They are caused by a too small contact distance after short
arcing times or by a small opening speed. The higher the erosion of the nozzle, affecting the insulating
parts between the arcing contacts, the higher the probability of re-ignitions and restrikes during
capacitive breaking operations.
55
TB 817 - Shunt capacitor switching in distribution and transmission systems
Summarizing, the pre-arc during the inrush current causes erosion of arcing contacts and erosion and
possible coating of insulating materials including nozzles. After more than 1000 operations this erosion
can reduce the withstand voltage of an interrupter, which in combination with a different gas flow
because of the eroded parts might cause re-ignitions and restrikes. These voltage breakdowns then
become probably permanent and require maintenance of the interrupter.
56
TB 817 - Shunt capacitor switching in distribution and transmission systems
57
TB 817 - Shunt capacitor switching in distribution and transmission systems
There are multiple benefits of controlled switching for circuit breakers. The controlled energization of
capacitor banks reduces the magnitude of inrush currents and limits the associated electrical wear.
Consequently, the contact erosion and nozzle ablation are decreased which positively impacts the
maintenance interval and costs as well as extend the circuit breaker life-time [59]. Point-on-wave closing
is applied to 60 % of all high voltage breakers according to the survey. Moreover, controlled de-
energization can be used as performance improvement by choosing an optimum arcing time that will
reduce the probability of a restrike. In this meaning it can be used to increase the basic performance
and to safely operate the circuit breaker outside the proven strength and capability without an increase
of failure probability [59], [60]. Although applied in fewer cases, the controlled de-energization of
capacitor banks reduces the restrike probability of the circuit breaker as by controlling the opening
instant a relatively large contact gap can be ensured [59], [60], [62].
In addition to benefits for circuit breakers, there are several benefits of controlled switching for capacitor
banks as well. Assuming that switching transient effects are a dimensioning criterion for the design of
capacitor banks, the controlled switching principle supports a more efficient use of the resources. In this
sense it helps to minimize the bank design as it theoretically allows to use a design with reduced margins
as well as higher steady-state and design stresses whilst acceptable performance levels are maintained
[59].
Furthermore, controlled switching helps to decrease failure rates of the capacitor units due to minimized
electrical and mechanical stress. An increased availability and reliability can be achieved whilst
unplanned maintenance costs should be lower. Also, a reduction of recertification costs for new
constructions or capacitor bank extensions can be a consequence [59].
Main benefit of the controlled switching principle for the power-system is the improved power quality as
well as an increased reliability and availability of the network components. Controlled switching reduces
the impact by transients in the network. Control and protection wirings are less affected by transient
surge coupling. Magnitude, shape, symmetry and frequency of the voltage wave is less impacted.
Switching overvoltages and travelling waves can be minimized. In succession the electrical and
mechanical stress, the re-strike probability and consequently the failure risk of the network components
is reduced. In addition, the transient potential rise of the substation earthing mesh is limited which means
an increased safety for personnel [59].
The reference [59] gives more detailed information and explanations about the multiple benefits for
circuit breakers, capacitor banks and power systems. The Table 7 is taken from [59] and shows a
succinct summary.
58
TB 817 - Shunt capacitor switching in distribution and transmission systems
Current and voltage have a shift of 120 degree between the three phases of the three-phase power
system. This is why controlled switching can be effectively optimized only in case of single-pole operated
circuit breakers as each pole is independently controlled and operated to the optimum switching
moment. Three-pole gang-operated breakers are mostly not suitable for controlled switching [63] unless
a mechanical delay between the poles or a specific staggering is adapted to the application. They have
limited added value.
5.2.2. Controlled closing
Controlled closing refers to controlling the point of conduction of each pole of the circuit breaker with
respect to the phase angle of the voltage. The technical effect is the reduction of inrush currents and
minimization of over-voltages in the system caused by reflected traveling waves. The benefits are
minimized stress on the switching devices and capacitor banks and consequently longer maintenance
intervals, reduced failure probability and expected longer life-time of the equipment [59], [60], [62].
The point-on-wave-controller monitors the source voltage for a controlled closing operation. Depending
on the mechanical closing time (Tclosing) and the prestrike characteristics (Tprearcing) of the circuit breaker
as well as the optimum closing instant – e.g. voltage zero-crossing for capacitive loads or voltage peak
for inductive loads – a randomly given closing command (tcommand) is delayed by the controller (Twaiting +
Tcontroller). After this calculated time delay the controller starts the energization of the closing coils and
the circuit breaker contacts starts to approach each other. Since the dielectric withstand of the interrupter
is decreasing with the distance between its contacts, the current may start flowing slightly before the
mechanical contacts touch. The electrical circuit is then closed through a pre-arc (tmake) [63].
59
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 35: Schematic timing sequence of a controlled closing operation (Source: [63])
The optimum point for energization of a shunt capacitor bank is the instant of the system voltage zero-
crossing. In this case the making operation would minimize capacitor inrush currents and network
voltage disturbances. In reality circuit breakers are not ideal mechanical switches as their operating
times have some scattering e.g. due to coupling of linkages, translations or other moving parts inside
the mechanical chain. Therefore, operating times typically vary between ±0.5 ms up to several
milliseconds. Circuit breakers used for controlled switching of capacitor banks should have a mechanical
scatter of less than ±1.0 ms, preferably less than ±0.5 ms [62], [63], [64].
In the opened position, the contacts are fully separated, and the dielectric strength of the circuit breaker
depends on the distance between the contacts and the properties of the used insulating medium (SF6
and vacuum). During the closing operation and the reduction of the contact distance, the contact velocity
as well as the design of the circuit breaker interrupter determine the slope of the dielectric strength which
is defined as « Rate of Decrease of Dielectric Strength » (RDDS). As soon as the voltage across the
approaching contacts exceeds the dielectric withstand strength, an arc will be ignited and conducts
current (Tprearcing). At the moment of contact touch the arc is extinguished and the metallic conduction
starts [64].
For a proper adjustment of the point-on-wave-controller the circuit breaker characteristics like the scatter
of closing times or the RDDS must be known. As visible in the diagram in Figure 36, there is a non-
neglectable probability that the making does not occur exactly at the voltage zero-crossing. Therefore,
the point-on-wave-controller is adjusted to have the instant of energization slightly after the zero-
crossing to compensate mechanical but also dielectric scatter (see Figure 37) [64].
60
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 36: Effect of the RDDS and scattering on the optimum closing moment
61
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 38: Schematic timing sequence of a controlled opening operation (Source: [63])
The aim of the controlled de-energization of a shunt capacitor bank is to achieve a sufficient dielectric
withstand of the open contact gap against the recovery voltage after a predetermined arcing time. In
case of a malfunction or failure of the point-on-wave-controller, the worst dielectric stress occurs when
the circuit breaker interrupts at the current zero-crossing with a minimum arcing time of 0.0 ms, since
the contact gap at the peak of the recovery voltage is the smallest compared to longer arcing times. The
risk is an increased restrike probability of the circuit breaker as well as higher dielectric stress and
disturbance to the installed electrical equipment in case of a restrike.
Figure 39 shows the voltage across the circuit breaker after current interruption [64]. At current zero, the
phase voltage is at the peak value as it is lagging the current by 90°. The capacitor bank will stay charged
at the maximum voltage whilst the supply-side will alternate to the opposite polarity of the phase voltage.
Consequently, the recovery voltage across the open contact gap of the circuit breaker will rise with a 1-
cos function up to a maximum value of 2 p.u. [62], [63], [64].
Figure 39: Voltage across the circuit breaker after interruption at current zero-crossing (Source: [64])
Capacitors are not ideal devices with a gradual self-discharge. Depending on the size of the capacitor
bank and the circuit conditions the charge will be around 30 - 40 % of the initial voltage after 1 - 2
minutes. After 5 times the discharge time constant (the time to close after de-energizing of a capacitor
bank normally will be much longer), the capacitor bank is considered to be fully discharged [64].
CIGRE WG A3.35 has extensively studied the controlled opening of shunt capacitor banks. The
reference [62] gives more detailed information and explanations.
62
TB 817 - Shunt capacitor switching in distribution and transmission systems
8 Few designs apply pre-insertion reactors instead of resistors with similar effects as resistors.
63
TB 817 - Shunt capacitor switching in distribution and transmission systems
64
TB 817 - Shunt capacitor switching in distribution and transmission systems
65
TB 817 - Shunt capacitor switching in distribution and transmission systems
Solid state devices maintain a certain resistance in the ON condition which results in a temperature rise
(or power loss) significantly higher than experienced in mechanical switches. The OPEN condition is
characterized by high impedance which causes a small current (leakage current) to remain. The
resulting power flow can be neglected, however, an electrical connection still exists and can never justify
a disconnect position.
Solid state switching devices act extremely fast, typically within microseconds. Their main application is
found within converters, where they operate in the kHz range. The fast reaction time allows a number
of special functions such as point-on-wave switching and has the additional benefit that the device is
not dependent on mechanical performance, which often limits the application of operating mechanisms.
Solid state switches can not only be controlled for turn-ON and turn-OFF but can also be used to control
the impedance. This feature can be utilized for current limiting, i.e. may have the same effect as resistors
or current limiting inductances used in conjunction with mechanical switches.
Power electronic devices are susceptible to overload. They can only withstand a certain level of
overcurrent or short circuit current. For example, thyristors can withstand a current surge of a limited
peak only for one half-cycle of current. They need to be turned off at the next current zero to avoid
destruction. IGBT and IGCT can only handle even smaller overcurrents, however, could interrupt
developing short-circuit currents before being damaged. Prerequisite is a very fast detection of the fault
in order to issue the control signal in due time. With respect to the handling of load and inrush currents
of capacitor banks, solid state switches should not have any limitations. However, if the location may
experience short-circuits currents, these have to be taken into account. Mechanical back-up switches
would not be helpful in this case.
The functionality of power electronics devices to control the duration of conducting state in relation to
the cycle time allows for the direct control of capacitive currents. This feature is widely used for FACTS
(flexible AC transmission systems) featuring different solutions – for example – thyristor switched
capacitor (TSC), thyristor controlled reactor (TCR) or static var compensator (STATCOM). These
devices are used instead of (mechanical) switches, which can just switch a capacitor ON or OFF. FACTS
allow for adjusted capacitive currents and react as fast as needed. FACTS will not be considered here.
The basic design of FACTS is in a way comparable to the design of solid state switches, however, the
control is different. This explains, why solid state switches are hardly used for capacitor switching
applications: With the expenditures for a solid state switch, a FACTS can be realized which provides
flexible control of currents and not just switching of currents.
66
TB 817 - Shunt capacitor switching in distribution and transmission systems
5.5.4. Cost
Because of their size and the need for additional cooling and protection equipment, power electronics
devices are expensive. They can only be justified when features and performance are required which
cannot be obtained with mechanical switches. Based on today's costs and maturity of technology, only
solid state converters including FACTS play an important role and have world wide applications, while
pure solid state switches are extremely rare in MV and HV networks.
To achieve the correct sequence, a synchronizing relay is used to send a start command to the operating
mechanism at the correct phase angle of the source voltage. The switch has one single fast operating
mechanism with the three phases mechanically coupled by a common shaft featuring a built-in
mechanical offset between the phases. The shaft is operated by a digitally controlled servomotor [69].
The speed of the rotating shaft is so high that the commutation between contacts and diodes is achieved
within a fraction of a half cycle of the power frequency. In principle, the phase-shifted operation of the
three poles can also be achieved by three separate drives. A resolver attached to the motor provides
feedback to a closed loop controller so that a pre-determined trajectory for each close or open operation
is ensured.
5.6.3. Ratings
In dependence of the voltage rating of the single diodes and the number of diodes in series (stack),
different voltage ratings can be achieved. The system should be applicable up to a voltage of 38 kV with
reasonable size and costs. The rated capacitive current determines the size of the diodes via their
thermal capacity. Without considerable extension of the thermal capacity of the diodes, the device is not
able to close into a full short-circuit.
67
TB 817 - Shunt capacitor switching in distribution and transmission systems
68
TB 817 - Shunt capacitor switching in distribution and transmission systems
69
TB 817 - Shunt capacitor switching in distribution and transmission systems
In [71], it is stated: “Very large inductors convert the capacitor bank to a true harmonic filter branch (often
13th harmonic or lower)”. According to this statement, filter banks considered in this Technical brochure
are 13th and/or lower harmonic filter banks. With increasing network voltage and harmonic order, the
filter inductors become smaller and eventually reach the size of the network inductance of several mH.
𝐶1 1
i(𝑡) = 𝑉𝑚 √ sin ( 𝑡)
𝐿𝑒𝑞 √𝐿𝑒𝑞 𝐶1
Equation 5: Temporal development of the inrush current for a single filter bank
The peak and frequency of the inrush current for a single filter bank are the same as those in [19]:
70
TB 817 - Shunt capacitor switching in distribution and transmission systems
𝐶1
𝑖𝑝𝑒𝑎𝑘 = 𝑉𝑚 √
𝐿𝑒𝑞
(a)
(b)
Figure 45: Single bank energization: (a) inrush currents (i in Figure 44) and (b) transient overvoltages
(VBUS to ground in Figure 44); comparison of filter bank and single capacitor bank
71
TB 817 - Shunt capacitor switching in distribution and transmission systems
Equation 8: Natural response of inrush current of a filter bank in the frequency domain
(a)
(b)
Figure 47: Circuits for back-to-back filter banks energization: (a) equivalent circuit and (b) frequency-
domain circuit.
The natural response of the inrush current in the time domain is:
72
TB 817 - Shunt capacitor switching in distribution and transmission systems
The inrush current includes two natural components: one flowing between the bank being energized (C2
in Figure 47) and the bank already energized (C1 in Figure 47), and the other flowing between the source
impedance and the bank being energized.
If the source inductance 𝐿𝑆 is much greater than the tuning current limiting reactors (𝐿𝑆 ≫ 𝐿1 and 𝐿2 ), the
inrush current is mainly the resonant current between the two banks and is expressed as:
𝐶𝑒𝑞 1
𝑖2ℎ (𝑡) = 𝑉𝑚 √ sin ( 𝑡)
𝐿𝑒𝑞 √𝐶𝑒𝑞 𝐿𝑒𝑞
The peak and frequency of the inrush current for two banks in parallel, which are the same as those in
[73], can be written as:
𝐶𝑒𝑞
𝑖𝑝𝑒𝑎𝑘 = 𝑉𝑚 √
𝐿𝑒𝑞
𝐶2 1
𝑖2ℎ (𝑡) = 𝑉𝑚 √ sin ( 𝑡)
𝐿𝑆 + 𝐿2 √𝐶2 (𝐿𝑆 + 𝐿2 )
𝐶2
𝑖𝑝𝑒𝑎𝑘 = 𝑉𝑚 √
𝐿𝑆 + 𝐿2
73
TB 817 - Shunt capacitor switching in distribution and transmission systems
1
𝑓𝑖𝑛𝑟𝑢𝑠ℎ =
2𝜋√𝐶2 (𝐿𝑆 + 𝐿2 )
(a)
(b)
Figure 48: Back-to-back energization: (a) inrush currents (i2 in Figure 47) and (b) transient overvoltages
(VBUS to ground in Figure 47); comparison of filter bank and back-to-back capacitor bank
The inrush current of a back-to-back filter bank with 150 MVAr with a configuration as shown in Figure
49 is presented in Figure 50 together with a measured waveform to prove coincidence between
74
TB 817 - Shunt capacitor switching in distribution and transmission systems
measurement and calculation. The inrush current for the switching of the filter branch HF5 is measured
under the condition of closed circuit breakers Q1, Q2, Q3 and Q4.
Figure 49: Schematic diagram of four multiple single-tuned filter banks for electric arc furnace. The
parameters in this system are: source inductance LS = 1.84 mH at 33 kV, inductance in 5th harmonic filter
(HF5) LHF5 = 2.34 mH and capacitance in HF5 CHF5 = 122.5 μF.
The measured current is compared with the EMTP simulation. In Figure 50, the temporal development
of frequency and peak of the inrush current in test and calculation are very alike. Some deviation
between the two waveforms is due to slightly different parameters in the real and simulation circuits.
Figure 50: Inrush currents for 5th harmonic filter energization: blue line – measured waveform and red
dotted line – EMTP simulation waveform [76]
With the EMTP simulation, the inrush current contribution of each feeder branch is analyzed and
displayed in Figure 51. The analysis shows that the inrush current of low harmonic filter banks is mainly
given by the source impedance and the higher harmonic filter banks (i.e. HF4 and HF3 in Figure 51)
that include relatively low inductances.
75
TB 817 - Shunt capacitor switching in distribution and transmission systems
Figure 51: Total inrush current through Q5, current coming from source and current contributed by each
feeder of the filter bank for HF5 energization.
Figure 52: Circuit diagram for single phase filter current interruption.
76
TB 817 - Shunt capacitor switching in distribution and transmission systems
For a filter bank, the trapped voltage at Cl (ul) can be calculated as:
𝑍𝐶𝑙
𝑢𝑙 = 𝑢′
𝑍𝐿𝑙 + 𝑍𝐶𝑙 𝑠
Table 10: Trapped capacitor voltage for filter banks; the parameters in Table 9 are used for the
calculation.
For medium-voltage applications, filter banks are commonly connected to an ungrounded star point to
block the flow of 3rd harmonic current into the system through the grounded neutral [73]. On the other
hand, a grounded star point with the tuning reactor located on the neutral side may be used for high-
voltage applications [73]. A delta-connection is only used for lower voltage (e.g. 2400 V) [1] and therefore
this connection is not considered.
77
TB 817 - Shunt capacitor switching in distribution and transmission systems
The equivalent circuit for ungrounded filter banks is shown in Figure 54. In Figure 54, the ungrounded
filter bank is represented by capacitors Cf combined with inductors Lf while Ceq is the capacitance
between the circuit breaker (mostly cable capacitance) and the filter bank, and Cn is the capacitance to
ground of the neutral.
Under steady-state conditions, Cn is at ground potential. When phase A is interrupted first, it makes the
three-phase system unbalanced and causes the rise of the potential of the neutral point by half of the
trapped capacitor voltage (VC). The peak of the power-frequency RV in phase A is then:
𝑉𝐴𝐴′ = 𝑉𝑆 + 1.5𝑉𝐶
Equation 17: Peak recovery voltage (RV) in phase A in an ungrounded filter bank
In the case of capacitor banks, the trapped capacitor voltage VC is almost the same as the source voltage
VS. By assuming VC = VS, the RV in phase A of capacitor banks can reach 2.5 pu (2.5 VS) [74]. As shown
in Table 10, the trapped capacitor voltage VC for filter banks is higher than the source voltage. Therefore,
filter banks experience higher RV than mere capacitor banks. For example, the RVs of 2nd and 3rd
harmonic filters in Table 10 can reach up to 3.08 pu and 2.71 pu, respectively.
Figure 55 shows measured transient capacitor voltages during the interruption of the current of a 2nd
harmonic filter (Figure 55a) and the transient recovery voltage across phase A (Figure 55b), which
interrupted first, recalculated from the measured phase to ground voltages. 1 pu is equivalent to the
source voltage.
(a) Transient voltages of phases A, B and C measured phase to ground during the interruption of the
current in the 2nd harmonic filter branch
78
TB 817 - Shunt capacitor switching in distribution and transmission systems
(b) Transient recovery voltage (TRV) across the interrupter contacts calculated for phase A from
measured signals in a)
Figure 55: Transient recovery voltage for three-phase filter current breaking [76]
Figure 55b is confusing at first glance, since the smooth 1 – cos shape of the RV known from capacitive
switching seems to exist, however, superimposed by a high-frequency transient voltage better known
from inductive switching. Indeed, the power frequency recovery voltage with a mean peak at
approximately 3 pu is superimposed by a transient voltage due to the resonance between the filter
inductance and the load-side cable capacitances denoted as Ceq in Figure 54 increasing the peak
voltage up to 3.5 p.u.
Neglecting the first millisecond of the transient recovery voltage (which probably indicates multiple re-
ignitions), the oscillating capacitor voltage consists of two high-frequency parts: one part before and a
second part after the interruption of phases B and C (vertical dotted line in Figure 55b). Before the
interruption of phases B and C, the transient overvoltage originates from the unbalanced circuit
configuration shown in Figure 56a below and its frequency (f1) can be calculated with the fact that Cf »
Ceq and Cn:
1
𝑓1 =
2𝜋√1.5𝐿𝑓 𝐶𝑒𝑞
Equation 18: Frequency of TRV after interruption of phase A before phases B and C interrupt
The equivalent schematics of the circuit after the interruption is shown in Figure 56b and its frequency
(f2) is:
1
𝑓2 =
2𝜋 √𝐿𝑓 𝐶𝑒𝑞
(a) (b)
Figure 56: Circuit diagrams for three-phase filter current interruption: (a) in one phase only and (b) in all
three-phases.
79
TB 817 - Shunt capacitor switching in distribution and transmission systems
It is also observed in Figure 55a that the time span between the interruption of phase A and the current
zero-crossings of phases B and C should be a quarter of a cycle (90 degrees) later in the power
frequency. However, the interruption of phases B and C happened around 30 degrees earlier i.e. around
60 degrees due to the harmonics from the thyristor rectifier installed at the same bus with the filter banks,
as shown in Figure 49.
The circuit parameters for the measured voltage waveforms in Figure 55 are: Lf = 39.1 mH and Cf = 52.5
μF. It is expected that Ceq i.e. the capacitance of the cable connection to ground has a value of several
10 nF depending on the length of the cable. It is difficult in practice to know the exact value. From the
measurement, the value of Ceq can be calculated reversely as 55 nF and with this value, the frequencies
of this circuit are the same as the frequencies evaluated from Figure 55b.
So far, ungrounded filter banks were treated. For grounded filter banks, there is no capacitance between
the neutral point and the ground. Therefore, the RV in phase A of the grounded filter banks is smaller
than that of the ungrounded case and its analytical expression is:
𝑉𝐴𝐴′ = 𝑉𝑆 + 𝑉𝐶
Equation 20: Peak recovery voltage (RV) in phase A in a grounded filter bank
The circuit diagram of the grounded filter bank is the same as in Figure 56b for both the single-phase
and three-phase interruptions due to their balanced condition. The high-frequency component of the
TRV can be calculated as:
1
𝑓=
2𝜋√𝐿𝑓 𝐶𝑒𝑞
𝐿𝑓
𝑈𝑚𝑎𝑥 = √ 𝐼
𝐶𝑒𝑞 𝑐ℎ𝑜𝑝
80
TB 817 - Shunt capacitor switching in distribution and transmission systems
𝐶1 1
i(𝑡) = 𝑉𝑚 √ sin ( 𝑡)
𝐿𝑒𝑞 √𝐿1 𝐶1
Equation 25: Temporal development of the outrush current of a filter bank in case of a fault
where, 𝐿𝑒𝑞 = 𝐿1 + 𝐿𝑙𝑖𝑛𝑒 .
Therefore, the peak and frequency of the outrush current are:
𝐶1
𝑖𝑝𝑒𝑎𝑘 = 𝑉𝑚 √
𝐿𝑒𝑞
81
TB 817 - Shunt capacitor switching in distribution and transmission systems
82
TB 817 - Shunt capacitor switching in distribution and transmission systems
https://cigregroups.org/display/SCA3/Study+Committee+A3+Home
# Question Text
Select the voltage classes used within your system(s). Note: Please include any voltage levels where
1
capacitors are installed on the tertiary winding of a transformer
Do you utilize shunt substation capacitor banks in your system(s)? Note 1: series compensation
2 capacitor banks are outside the scope of this Working Group - please only answer for shunt substation
capacitor banks at this time Note 2: overhead line banks will be queried later
3 Select system frequency(ies)
Select all the voltage classes that have at least one shunt substation capacitor bank in your
4 system(s). Reminder: series compensation capacitor banks are outside the scope of this Working
Group - please only answer for shunt capacitor banks
5 Why are substation capacitor banks used?
6 For each switching device, select the typical substation shunt capacitor bank size.
7 What is the typical substation shunt capacitor bank close-open switching rate?
Please indicate the types of capacitor bank configurations used. If "other," please describe in the
8
comments.
13 Please select the category that best describes the most common age of your substation shunt
capacitor bank switching equipment. Estimation is acceptable if exact data is not available.
How often is inspection and/or maintenance completed on the substation shunt capacitor
14
bank switching device, in terms of time?
15 How often is inspection and/or maintenance completed on the substation shunt capacitor bank
switching device, in terms of total count of close-open switching operations?
83
TB 817 - Shunt capacitor switching in distribution and transmission systems
# Question Text
16 Please describe standard maintenance interval with respect to switching duty, this time by device type.
17 Please describe standard maintenance interval with respect to time, this time by device type.
Please describe how substation shunt capacitor bank switching devices are operated. Note: multiple
18
choice allowed
Are there applications where substation capacitor bank protection is done with a separate switching
19 device than substation capacitor bank switching? If yes, please describe the situation(s) and any
reasons why.
Indicate the type(s) of devices used for protection of substation capacitor banks. If "other," please
20
describe in comments.
21 Are substation capacitor banks switched or protected with any other device types not listed above?
22 Do you use devices to improve substation shunt capacitor bank switching performance?
Select the controlled switching techniques used depending on substation capacitor bank switching
23
device type
24 Are you knowledgeable about the inrush characteristics of your substation shunt capacitor banks?
List the typical inrush current peak of your substation shunt capacitor banks (kA) – only for back-to-
25
back banks
List the typical inrush current frequency of your substation shunt capacitor banks (kHz) - only for back-
26
to-back banks.
For the cases where controlled switching is implemented, please indicate the observed reduction in
27 peak inrush current for substation shunt capacitor banks (single-bank switching only). Note: this
question is inquiring about the percentage reduction in inrush from the prospective peak
Please indicate if you use reactors in series with the capacitor bank to limit the inrush or outrush
28 current (not a filter reactor). Please indicate the most typical inductance range of the reactor. Values
are in microhenries (uH)
What factors force a change or replacement in switching equipment? Consider factors like age,
29
number of switching operations, or catastrophic event.
Are design choices changing for substation capacitor banks? For example, are new installations using
30 back-to-back configurations instead of single bank configurations? If so, please provide as much detail
as possible in the comment field below.
Are you satisfied with the switching performance of existing devices to switch substation capacitor
31
banks? Please, provide any reasoning for your response
Do you utilize overhead line capacitor banks in your system(s)? Note 1: as noted previously, series
32 compensation capacitor banks are outside the scope of this Working Group - please only answer for
line capacitor banks
84
TB 817 - Shunt capacitor switching in distribution and transmission systems
# Question Text
Select all the voltage classes that have at least one overhead line capacitor bank in your
33 system(s). Reminder: series compensation capacitor banks are outside the scope of this Working
Group - please only answer for shunt capacitor banks
34 Select the system frequency(ies) where overhead line capacitor banks are installed
35 Why are overhead line capacitor banks used?
36 Indicate the typical overhead line capacitor bank sizes
37 What is the typical overhead line capacitor bank close-open switching rate?
Please indicate the types of overhead line capacitor bank configurations used. If "other," please
38
describe in the comments.
Have your overhead line capacitor banks suffered faults? If yes, please provide any relevant
39 information about the nature of these faults. This could include rate of faults, fault current
magnitudes, fault causes, in addition to any other information you may have.Please also describe what
type(s) of switching device(s) see these faults and why those devices are used.
Indicate the type(s) of devices used for the switching of overhead line capacitor banks. If "other,"
40
please describe in comments.
41 Select the typical overhead line capacitor bank switching device class rating
42 Please select the category that best describes the most common age of your overhead line capacitor
bank switching equipment. Estimation is acceptable if exact data is not available.
How often is inspection and/or maintenance completed on the overhead line capacitor bank switching
43
device, in terms of time?
44 How often is inspection and/or maintenance completed on the overhead line capacitor bank switching
device, in terms of total count of close-open switching operations?
Please describe standard maintenance interval with respect to switching duty, this time by device type,
45
for overhead line capacitor banks.
Please describe standard maintenance interval with respect to time, this time by device type, for
46
overhead line capacitor banks.
Please describe how overhead line capacitor bank switching devices are operated. Note: multiple
47
choice allowed
Are there applications where overhead line capacitor bank protection is done with a separate
48 switching device than overhead line capacitor bank switching? If yes, please describe the situation(s)
and any reasons why.
49 Are overhead line capacitor banks switched or protected with any other device types not listed above?
Indicate the type(s) of devices used for protection of overhead line capacitor banks. If "other," please
50
describe in comments.
51 Do you use devices to improve overhead line capacitor bank switching performance?
85
TB 817 - Shunt capacitor switching in distribution and transmission systems
# Question Text
Select the controlled switching techniques used depending on overhead line capacitor bank switching
52
device type
53 What factors force a change in switching equipment used for overhead line capacitor banks? Consider
factors like age, number of switching operations, or catastrophic event.
Are design choices changing for overhead line capacitor banks? For example, are new installations
54 using back-to-back configurations instead of single bank configurations? If so, please provide as much
detail as possible in the comment field below.
Are you satisfied with the switching performance of existing devices to switch overhead line capacitor
55
banks? Please, provide any reasoning for your response
56 Do you utilize harmonic filter banks in your system?
57 Select all the voltage classes that have at least one filter bank in your system(s).
58 Select the system frequency(ies) where filter banks are installed.
59 Where are filter banks installed?
60 Why are filter banks installed?
61 Select the typical substation shunt capacitor bank size, switched each by one switching device.
68 Please provide a contact email address, so the survey results can be shared with you directly
69 Select country
70 I am answering on behalf of and/or in my capacity as...
Are you willing to share further information about your experiences with the CIGRE A3.38 working
71 group? If so please provide the requested contact information, and a member of the working group
will contact you. At minimum please provide a name and email address.
86
TB 817 - Shunt capacitor switching in distribution and transmission systems
87
TB 817 - Shunt capacitor switching in distribution and transmission systems
88
TB 817 - Shunt capacitor switching in distribution and transmission systems
89
TB 817 - Shunt capacitor switching in distribution and transmission systems
90
TB 817 - Shunt capacitor switching in distribution and transmission systems
91
TB 817 - Shunt capacitor switching in distribution and transmission systems
92
TB 817 - Shunt capacitor switching in distribution and transmission systems
93
TB 817 - Shunt capacitor switching in distribution and transmission systems
[74] A. Greenwood, Electrical Transients in Power Systems, John Wiley & Sons, 1971.
[75] IEEE C57.142: “Guide to Describe the Occurrence and Mitigation of Switching Transients
induced by Transformers, Switching Device, and System Interaction”, 2010
[76] By courtesy of Kim Seongil, Korea, 2019
[77] IEC 60099-5: 2018 “Surge arresters – Part 5: Selection and application recommendations”
94
CIGRE
21, rue d'Artois
75008 Paris - FRANCE
© CIGRE
ISBN : 978-2-85873-522-8