368

OPERATING ENVIRONMENT OF VOLTAGE GRADING CAPACITORS

APPLIED

TO HIGH VOLTAGE CIRCUIT-BREAKERS

Working Group
A3.18

February 2009
 Working Group A3.18

OPERATING ENVIRONMENT OF VOLTAGE GRADING CAPACITORS

APPLIED TO HIGH VOLTAGE CIRCUIT-BREAKERS

Members:

M. Runde NO (Convenor)
A. Bosma SE
P.F. Coventry UK
S. Dick AU
E. Kynast DE
C. Paul DE
A. Ribeiro BR
J. Rickmann CH
E. Savary CH
B.R. Sunga CA
D. Yoshida JP
J.K. Schjølberg NO (guest)

Copyright © 2009
“Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right
of use for personal purposes. Are prohibited, except if explicitly agreed by CIGRE, total or partial
reproduction of the publication for use other than personal and transfer to a third party; hence
circulation on any intranet or other company network is forbidden”.

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”.

ISBN :978-2-85873-055-1
 CONTENTS

1 Introduction ...................................................................................................................... 6
1.1 Grading capacitor designs and technologies ........................................................... 6
1.2 Purpose of work and structure of report .................................................................. 8
1.3 References ............................................................................................................. 9
2 Service experience with voltage grading capacitors ....................................................... 10
2.1 Introduction ........................................................................................................... 10
2.2 Problems experienced in service ........................................................................... 10
2.2.1 Corrosion and leaks .................................................................................. 10
2.2.2 Dielectric breakdown ................................................................................. 11
2.2.3 Damaged internal conductors .................................................................... 13
2.2.4 Ceramic grading capacitors ....................................................................... 15
2.3 Diagnostic testing of grading capacitors - methods and practices .......................... 15
2.3.1 Test methods ............................................................................................ 15
2.3.2 Utility practices ......................................................................................... 16
2.4 Discussion and conclusion .................................................................................... 16
2.5 References ........................................................................................................... 17
3 Mechanical stress on grading capacitors ........................................................................ 18
3.1 Introduction ........................................................................................................... 18
3.2 Mechanical stress during circuit-breaker operation ................................................ 18
3.2.1 Background to acceleration measurement and data processing ................. 18
3.2.2 Examples of acceleration measurements .................................................. 22
3.3 Mechanical stress during transport ........................................................................ 30
3.4 Mechanical stress due to seismic activity .............................................................. 32
3.5 Conclusions .......................................................................................................... 32
3.6 References ........................................................................................................... 32
3.7 Bibliography .......................................................................................................... 33
4 Electrical stresses and nominal voltage .......................................................................... 34
4.1 Introduction ........................................................................................................... 34
4.2 Voltage sharing ..................................................................................................... 34
4.3 Nominal voltage .................................................................................................... 37
4.4 Rated insulation level ............................................................................................ 38
4.4.1 Short-duration power-frequency withstand voltage .................................... 38
4.4.2 Switching impulse withstand voltage ......................................................... 39
4.4.3 Lightning impulse withstand voltage .......................................................... 41
4.4.4 Insulation level and nominal voltage .......................................................... 42
4.5 Stresses arising out of switching duties................................................................. 43
4.6 Coordination of withstand voltages between circuit-breaker interrupter and grading
capacitor ............................................................................................................... 43
4.7 Conclusions .......................................................................................................... 43
4.8 References ........................................................................................................... 43
5 Electrical stresses during shunt reactor switching .......................................................... 44
5.1 Background........................................................................................................... 44
5.1.1 Load side oscillation.................................................................................. 44
5.1.2 Current chopping ...................................................................................... 45
5.1.3 Reignition .................................................................................................. 46
2
 5.1.4 Multiple reignitions .................................................................................... 47
5.1.5 Interaction between phases ....................................................................... 48
5.2 Electrical stresses imposed on grading capacitors ................................................ 48
5.2.1 Circuit-breaker in closed position .............................................................. 48
5.2.2 Circuit-breaker opening operation ............................................................. 48
5.2.3 Circuit-breaker in open position ................................................................. 49
5.2.4 Circuit-breaker closing operation ............................................................... 49
5.2.5 Unequal voltage sharing ............................................................................ 49
5.2.6 Capacitor discharge current ...................................................................... 56
5.3 Discussion on levels of electrical stresses ............................................................ 57
5.4 Conclusions .......................................................................................................... 59
5.5 References ........................................................................................................... 59
6 Recommendations for test procedures ........................................................................... 60
6.1 Introduction ........................................................................................................... 60
6.2 Mechanical tests ................................................................................................... 60
6.2.1 General ..................................................................................................... 60
6.2.2 Design and Type Tests ............................................................................. 61
6.2.3 Routine test .............................................................................................. 62
6.2.4 Performance check ................................................................................... 62
6.3 Environmental tests .............................................................................................. 62
6.3.1 General ..................................................................................................... 62
6.3.2 Design tests .............................................................................................. 63
6.4 Electric tests ......................................................................................................... 65
6.4.1 General ..................................................................................................... 65
6.4.2 Nominal voltage of grading capacitors ....................................................... 65
6.4.3 Design tests .............................................................................................. 66
6.4.4 Type tests ................................................................................................. 67
6.4.5 Routine test .............................................................................................. 69
6.4.6 Performance check ................................................................................... 71
6.5 Conclusions .......................................................................................................... 72
6.5.1 Environmental tests................................................................................... 72
6.5.2 Mechanical tests ....................................................................................... 72
6.5.3 Electrical tests .......................................................................................... 72
6.6 References ........................................................................................................... 72

Figure 1 - Two unit circuit-breaker with grading capacitors ..................................................... 6

Figure 2 - Cross section of porcelain insulated grading capacitor for AIS circuit-breakers
(schematic) ............................................................................................................................ 6
Figure 3 - Voltage grading capacitors for GIS circuit-breakers. Metal bellows (front) and
capacitor element stack (right) ............................................................................................... 8
Figure 4 - Ceramic capacitor element (left and middle) and complete grading capacitor
assembly installed on a GIS circuit-breaker (right) .................................................................. 8
Figure 5 - Typical example of an oil leak from an old AIS grading capacitor (left), and a
schematic drawing of a typical flange and sealing arrangement (right) ................................. 11
Figure 6 - Example of circuit-breaker grading capacitor following disruptive discharge ......... 12
Figure 7 - Short-circuited capacitor elements........................................................................ 12
Figure 8 - Carbonisation and arc tracks along the stack (left) and in the supporting
pressboard (right) of a partly short-circuited oil insulated grading capacitor .......................... 13
Figure 9 - Opened grading capacitor with partly broken foil conductor (see arrow). .............. 14
3
Figure 10 - Discolouration of the electric leads between bellows and end plate caused by
discharge activity during disconnector switching operations ................................................. 15
Figure 11 - Thermal image showing heating due to internal discharges in energized grading
capacitor .............................................................................................................................. 16
Figure 12 - General example of acceleration measurement and data processing to velocity
and displacement in three directions during an earthquake ................................................... 19
Figure 13 - Acceleration with respect to time, measured on a TRV control capacitor, mounted
to a dead tank circuit-breaker during a CO-operation ............................................................ 19
Figure 14 - Acceleration with respect to frequency based on measurement shown in Figure
13 20
Figure 15 - Displacement with respect to frequency based on measurement shown in Figure
13 21
Figure 16 - Acceleration (y-axis) with respect to time (x-axis), measured at a TRV control
capacitor, mounted to a dead tank circuit-breaker during CO-operation, original measuring
signal and signal after low-pass filtering ............................................................................... 22
Figure 17 - One pole of circuit-breaker A .............................................................................. 23
Figure 18 - Circuit-breaker B, three poles on a common structure ........................................ 24
Figure 19 – Circuit-breaker C with six units per pole ............................................................. 24
Figure 20 - Attaching the accelerometers to the capacitor .................................................... 25
Figure 21 - Acceleration measurements on circuit-breaker B in axial direction on the driving
side ...................................................................................................................................... 25
Figure 22 - Acceleration versus time during opening operation measured by four
accelerometers simultaneously ............................................................................................. 26
Figure 23 - Acceleration, velocity and displacement of opening operation ............................ 27
Figure 24 - Measurements carried out during circuit-breaker operations (CO) under identical
conditions ............................................................................................................................. 28
Figure 25 - Acceleration in the x-, y- and z-directions during CO operation without filter and
with 200 Hz filter (x = axial direction) .................................................................................... 29
Figure 26 – Displacement in y-direction in log-scale vs. frequency without filtering ............... 30
Figure 27 - Equivalent circuit for voltage sharing, circuit-breaker with 2 interrupters per pole 34
Figure 28 - Equivalent circuit for voltage sharing, circuit-breaker with 4 interrupters per pole 35
Figure 29 - Simplified single phase equivalent circuit ........................................................... 44
Figure 30 - Load side oscillation ........................................................................................... 46
Figure 31 - Reignition ........................................................................................................... 47
Figure 32 - Reignition followed by successful interruption ..................................................... 47
Figure 33 - Equivalent circuit for voltage sharing under static conditions .............................. 50
Figure 34 - Equivalent circuit for voltage sharing under dynamic conditions .......................... 52
Figure 35 - Initial rise in voltage across load side grading capacitor when source side
interrupter prestrikes ............................................................................................................ 53
Figure 36 - V-t curve for coaxial gap in SF 6 (positive polarity) [6] ......................................... 55
Figure 37 - V-t curve for coaxial gap in SF 6 (negative polarity) [6] ........................................ 55
Figure 38 - Current through grading capacitor at prestrike .................................................... 56
Figure 39 - Equivalent circuit for capacitor ............................................................................ 56
Figure 40 – Temperature cycle test ...................................................................................... 64
Figure 41 - Discharge test set-up ......................................................................................... 67
Figure 42 - Wiring diagram of the measuring circuit for the high frequency capacitance of a
capacitor .............................................................................................................................. 69
4
Table 1 - Factor by which voltage is increased compared with an equal voltage sharing, AIS
circuit-breakers .................................................................................................................... 36
Table 2 - Factor by which voltage is increased compared with an equal voltage sharing, GIS
circuit-breakers .................................................................................................................... 37
Table 3 - Rated voltages of circuit-breakers and nominal voltages of grading capacitors ...... 38
Table 4 - Minimum short-duration power-frequency withstand voltage for grading capacitors,
AIS circuit-breakers (IEC 62271-1 [1]) .................................................................................. 39
Table 5 - Minimum short-duration power-frequency withstand voltage for grading capacitors,
GIS circuit-breakers (IEC 62271-203 [2]) .............................................................................. 39
Table 6 - Minimum switching impulse withstand voltage for grading capacitors, AIS circuit-
breakers (IEC 62271-1 [1]) ................................................................................................... 40
Table 7 - Minimum switching impulse withstand voltage for grading capacitors, GIS circuit-
breakers (IEC 62271-203 [2]) ............................................................................................... 41
Table 8 - Minimum lightning impulse withstand voltage for grading capacitors, AIS circuit-
breakers (IEC 62271-1 [1]) ................................................................................................... 42
Table 9 - Minimum lightning impulse withstand voltage for grading capacitors, GIS circuit-
breakers (IEC 62271-203 [2]) ............................................................................................... 42
Table 10 - Comparison between recovery voltage peak and rated switching impulse
withstand voltage of IEC 62271-1 ......................................................................................... 57
Table 11 – Maximum voltage across one grading capacitor of a pair due to unequal sharing
under dynamic conditions (speculative worst case) .............................................................. 58
Table 12 - Rated voltages of circuit-breaker vs. nominal voltages of grading capacitors ....... 66
Table 13 - Range of requirements for grading capacitors for 550 kV circuit-breakers with two
breaking units ....................................................................................................................... 67
Table 14 - Recommended withstand test voltages for grading capacitors for 550 kV circuit-
breakers with two breaking units .......................................................................................... 67

5
1 Introduction
1.1 Grading capacitor designs and technologies
Voltage grading capacitors are accessories for high-voltage circuit-breakers used to control the
voltage distribution across each breaking unit / interrupting chamber of multi-unit circuit-breakers,
both when the circuit-breaker is in open position and during making and breaking operations.
Grading capacitors, if fitted, are always mounted in parallel with each chamber. Figure 1 shows a
modern two unit 300 kV air insulated switchgear (AIS) circuit-breaker with the grading capacitors in
separate porcelain housings next to the bigger porcelain housings containing the SF 6 filled
interrupting chambers.

Figure 1 - Two unit circuit-breaker with grading capacitors

Several technologies and designs for grading capacitors exist, both with regard to dielectric material
being used in the capacitive elements, electric insulation system, internal pressure system, and
housing. Figure 2 shows a schematic drawing of the most common design: Stacked foil wound
capacitor elements impregnated with oil and placed inside a porcelain housing with gas filled
pressure bellows.

Figure 2 - Cross section of porcelain insulated grading capacitor for AIS circuit-breakers
(schematic)

The capacitor elements are made by first winding foils of dielectric and aluminium on a circular
mandrel. This round assembly is then flattened, and the elements are stacked on top of each other.
Electrical interconnection is established either with a metal tab installed within the winding or simply
6
by applying pressure on the mechanical connection between the aluminium foils. Up to few hundred
capacitor elements are connected in series. Typical capacitance values are in the range 300 - 2500
pF.

In foil capacitors manufactured until around 1990, paper was used as the solid dielectric (“all-
paper”), giving a tanδ (loss factor) of 40 - 50×10 -4 (0,4 – 0,5 %). Newer designs use polypropylene
films with a tan δ of 1 - 10×10 -4 , or a combination of polypropylene and paper films, so-called mixed
dielectric, with tan δ typically below 30×10 -4 , depending on the paper to polypropylene ratio. Thin
aluminium foils are used as electrodes.

The capacitor stack is impregnated with oil. Capacitors manufactured up to around 1990 contain
mineral oil. Later synthetic oils became increasingly popular due to their better dielectric
performance, especially at low temperatures. Furthermore, synthetic oils have around 10 % higher
permittivity compared to mineral oil. SF 6 insulated grading capacitors also exist, but are rare.

Oil filled capacitors usually contain a system for compensation of the thermal expansion of the oil
and/or keeping the internal pressure somewhat elevated and fairly constant across a specified
temperature range. Some designs apply one or two large metal bellows filled with Nitrogen at 0,2 –
0,3 MPa overpressure (see Figure 2), whereas others use many, up to around 100, smaller
diaphragm cells. A slight overpressure increases the dielectric strength compared to what it would
have been at atmospheric pressure, and also avoids pressure variations introduced by fluctuations
in the ambient temperature.

Open bellows solutions are also used, in particular for grading capacitors installed in gas insulated
switchgear (GIS). The bellows then equalize the internal pressure of the grading capacitor with that
of the surrounding SF 6 .

SF 6 filled capacitors usually have a higher internal pressure than oil filled, and no bellows.

The stack of capacitor elements is mounted in an insulator housing with metal terminals at both
ends. The housing is made in porcelain or composite with silicone sheds for AIS breakers, or in
reinforced epoxy for GIS circuit-breakers.

Typical dimensions for grading capacitors used in GIS are lengths of 50 – 75 cm and diameters of
10 - 12 cm. AIS grading capacitors can be substantially larger, with lengths up to two metres or
more.

Figure 3 shows a capacitor stack, Nitrogen filled metal bellows and two complete GIS capacitors.

7
 Figure 3 - Voltage grading capacitors for GIS circuit-breakers. Metal bellows (front) and
capacitor element stack (right)

Grading capacitors based on using bulk ceramic dielectrics in the active parts also exist, see Figure
4.

Terminal
Flat electrode

Ceramic

Molded resin

Figure 4 - Ceramic capacitor element (left and middle) and complete grading capacitor
assembly installed on a GIS circuit-breaker (right)

Each capacitor element consists of a cylinder made from a ceramic material with a high dielectric
constant, to which flat metal electrodes and terminals are attached at both ends and the assembly
encapsulated in resin. Several such elements are mounted in series within a SF 6 insulated
cylindrical housing to form a unit, and several such units in parallel are mounted across the
interrupting chamber of the circuit-breaker, see right hand photo of Figure 4.

Ceramic capacitors are very common in GIS circuit-breakers manufactured in the Far East, and
some are also in use in the North America. In Europe they hardly exist.

The big advantage is that there is no oil insulation involved, and thus no leak problems. A major
drawback, and probably also one reason for the limited usage, is that many capacitor elements in
series and in parallel are required, due to the limited voltage and capacitance per element.

1.2 Purpose of work and structure of report

A number of rather disturbing incidents in recent years have brought attention to issues related to
reliability, service environment and test procedures of voltage grading capacitors, at least from
some utilities.

8
Concerning service experience, it was found that very little precise information exists. The voltage
grading capacitors are in many respects considered an integral part of the circuit-breaker, not as a
separate component. Hence, in circuit-breaker reliability surveys the grading capacitors have played
a rather anonymous role, sometimes not even listed among the possible alternatives on the “failed
sub-component” list of the enquiry [1] 1.

In order to obtain some information about failures and flaws experienced on grading capacitors
information has been collected by interviewing twenty utilities from nine countries. The objective
was to get a general and qualitative overview of what types of problems that occur, which parts of
the grading capacitors that fail, what the predominant failure modes are, and what the
consequences are. The outcome of this investigation is presented in Chapter 2.

Chapter 3 deals with the mechanical stresses grading capacitors are subjected to, in particular
during switching operations. The investigation was initiated by discoveries of mechanical fatigue in
the metal bands connecting the capacitor elements in a few rather old capacitors. It became evident
that robustness of these parts could be an important issue. Consequently, it was considered
important to determine what mechanical stresses (acceleration amplitude and frequencies) the
grading capacitors will experience in service and to clarify to what extent the various observed
frequencies / g-levels are harmful. By using accelerometers the mechanical stresses / shocks /
vibrations have been measured directly on the grading capacitors of circuit-breakers of different
technologies during switching operations.

A number of cases where grading capacitors installed on shunt reactor circuit-breakers have failed
violently (exploded) immediately after opening operations have occurred in several countries. It is
thus speculated that reactive current switching under certain conditions may cause over-voltage or
over-current stresses on the grading capacitors of a large and unforeseen type.

Chapter 4 deals with electrical stresses such as nominal voltage, power frequency, lightning
impulse and switching impulse and defines a set of nominal voltages for grading capacitors.

Chapter 5 reports on comprehensive simulations of transient currents and voltages during shunt
reactor switching, and possible explanations for the failures are put forward and discussed.

At present no internationally accepted standard for testing of grading capacitor exits. Chapter 6
gives, with basis in the findings reported in the previous chapters, recommendations for test
procedures for grading capacitors. Mechanical, environmental and electrical tests are proposed,
most of them to a large extent based on manufacturer test procedures being applied at present. The
recommendations are meant as background material for ongoing and future work by standardizing
bodies such as the IEC.

A number of other topics associated with grading capacitors exist. These include overvoltages
associated with ferroresonance, trapped charges, effect on transient recovery voltages, etc.
However, these and other issues addressing relationships with the entire transmission system or
other types of equipment are considered to be beyond the scope of the present work.

1.3 References
[1] “Final report of the second international enquiry on high voltage circuit-breaker failures and
defects in service”, CIGRÉ Technical Brochure no. 83, 1994

1 Numbers in square brackets refer to the references at the end of each chapter.

9
2 Service experience with voltage grading capacitors
2.1 Introduction
Service problems and failures of grading capacitors do not seem to have been a matter of great
concern for most utilities when considering the overall service experience of their circuit-breakers.
However, exceptions exist. A few utilities have in recent years reported about substantial problems,
including leaks, disruptive dielectric failures and mechanical fatigue problems [1].

The reported malfunctions were not limited to specific designs or equipment provided by certain
manufacturers. It appeared that the majority of grading capacitor manufacturers and most circuit-
breaker vendors were involved, signifying that this might be a broader and more general problem.

But again, the reported problems were limited to a few utilities. For obtaining a wider perspective on
grading capacitor service experience, information should be collected from a broader basis. With
this objective utility representatives from around the world have been contacted and interviewed
about service experience and failure history of their grading capacitors. More than twenty utilities
from nine countries around the world have contributed.

This chapter summarizes the outcome of this mini-survey and provides a general, descriptive and
qualitative overview of what type of problems that have occurred. It is believed that even such
limited and qualitative knowledge about grading capacitor service experience is of value for both
manufacturers and for the electric utilities. It helps the manufacturers improve their products, and
serves as inputs for the utilities when analysing flaws and failures, and in planning diagnostics
testing, maintenance and replacement programs. Furthermore, when preparing international
standards experience from service should be taken into account.

For quantitative failure statistics the reader is referred to the very comprehensive CIGRÉ survey on
reliability of high-voltage equipment that is under way [2].

2.2 Problems experienced in service

2.2.1 Corrosion and leaks
The vast majority of the service problems reported concern leaks and corrosion. Leaks were most
widely reported as oil leaking from grading capacitors in AIS. A significant number of cases of SF 6
leaking into GIS grading capacitors were reported, although these were confined to a few utilities. In
some cases, oil leaks from GIS capacitors had been observed. Moisture ingress into AIS grading
capacitors was reported in a significant number of cases.

There was no correlation with the number of operations or switching duty, but a clear correlation
with age. Leaks were most common for older grading capacitors. Most of the affected capacitors
had been in service for 20 years or more. A very few cases were reported where the grading
capacitors had been in service for between 5 to 10 years.

In some cases, a significant proportion of the grading capacitor population of a certain type of
design was affected. Some utilities reported leaks affecting more than 100 grading capacitors, one
large utility estimated that more than 5 000 out of its total grading capacitor population of
approximately 25 000 had been replaced due to leaks.

Figure 5 shows a typical appearance of a leak from the flange of an AIS grading capacitor and a
common flange design.

10
 Porcelain

Cement
Flange
Oil

End
plate
O-ring Bolt

Figure 5 - Typical example of an oil leak from an old AIS grading capacitor (left), and a
schematic drawing of a typical flange and sealing arrangement (right)

The end-plate is attached to the flange with the bolt and exerts a certain pressure on the o-ring and
thus prevents the oil from leaking out. The flange in turn, is fixed to the porcelain insulator by a
cement joint. There are three well-known causes of leaks and moisture ingress in such a
configuration:

- Crevice corrosion at flange-end plate interface;

- Incompatibility between o-ring material and the end plate metal causing pitting corrosion in the
O-ring groove;
- Cracking of cement causing movement of flange and loss of pressure on the O-ring.
Consequently, mechanisms causing leaks for AIS grading capacitors were reported as seal failures,
corrosion and cracked porcelain. In some cases, seal failure and corrosion were quoted jointly as
the cause of failure.

Leaks for GIS grading capacitors were reported to have been caused by poor workmanship and
quality control during assembly. In particular, poor surface finish and spilt solder on sealing
surfaces were quoted.

Moisture ingress was often associated with leaks in AIS grading capacitors. In some cases, leaks
had lead to disruptive dielectric failure. However, the number of disruptive failures was very small
compared with the large number of leaks reported. Tests on GIS grading capacitors showed that
SF 6 leaks into the oil filled capacitors had not necessarily reduced the strength of the insulation.

In most cases, leaking grading capacitors were replaced. In some cases, all capacitors of a type
had been replaced. Some utilities had repaired leaking capacitors, particularly where seal
replacement was required.

2.2.2 Dielectric breakdown

Some instances of disruptive dielectric failure of grading capacitors were reported. The number of
cases, a few dozens, was small compared to the total number of circuit-breakers covered in the
survey. The failures had occurred on both GIS and AIS circuit-breakers. Most had occurred within
less than 15 years service.

11
 Figure 6 - Example of circuit-breaker grading capacitor following disruptive discharge

A dielectric disruptive failure is a highly disturbing event, usually causing major damages to the rest
of the circuit-breaker (see Figure 6) and in some cases also affecting neighbouring equipment.
In most cases where failure had occurred, the remaining grading capacitors on the affected circuit-
breaker were removed for examination. Some of these were found to exhibit various degrees of
dielectric failure. Dielectric testing revealed increased capacitance and loss angle and the presence
of partial discharge activity. In some cases the capacitors were then opened for visual inspection,
and internal short-circuits were discovered, as shown in Figure 7.

Figure 7 - Short-circuited capacitor elements

In general there was a clear correlation between disruptive dielectric failures and switching duty.
Most of the circuit-breakers concerned had been switching low inductive currents and failure usually
occurred on or shortly after an opening operation. (Although the internal damage may of course
have been created during the previous closing operation.) Most of these were shunt reactor circuits
applications.

The occurrence of failure was apparently not related to the number of operations, although shunt
reactor circuit-breakers are usually operated frequently. The number of operations was typically
thousands. At least one of the failures occurred on a circuit-breaker with controlled switching units
installed; although most were without. The cause of the failures has not been fully explained.

12
One utility reported a number of cases where a change in capacitance value was detected. Some
capacitors were apparently suffering from punctured elements and some from a complete insulation
breakdown. In this case, the failures were not correlated with switching duty.

Examination of one of the partially failed capacitors, see Figure 8, suggests the following failure
development: Dielectric breakdown apparently occurred first at the folded edge of one of the wound
elements, but without causing a complete short-circuit. It is possible that pressing from a circular to
a flat element during manufacture may produce weak areas at the folded edges. Discharge activity
caused gas bubbles to form. The high relative permittivity of the oil and lower dielectric strength of
the gas caused further discharge activity and gas formation under normal service voltage. The gas
bubbles migrated along the interface between the capacitor stack and its pressboard support.
Further discharges occurred, causing carbonisation of the pressboard and the paper insulation.
When an element fails to a short-circuit, the electric stress on the remaining elements increases
and more elements fail.

Figure 8 - Carbonisation and arc tracks along the stack (left) and in the supporting
pressboard (right) of a partly short-circuited oil insulated grading capacitor

It appears, therefore, that failure in a single element is sufficient to initiate the failure process,
which can develop under normal service voltage over time. But again, the reason for the initial
element failure is not clear.

Following a dielectric failure, all capacitors of the affected circuit-breaker were usually replaced.
Some utilities install controlled switching on shunt reactor circuit-breakers in order to eliminate re-
ignition overvoltages. In some utilities, it is practice to open a sequential disconnector following
opening of the shunt reactor circuit-breaker. Sequential disconnection reduces the period of time
during which the grading capacitors are stressed by the system voltage from hours to minutes and
reduces the rate of degradation of any failing capacitors accordingly.

2.2.3 Damaged internal conductors

Capacitance values indicating an open circuit inside the grading capacitor have in some cases been
measured. Some utilities do capacitance measurements on a regular basis as a diagnostic test,
whereas in other cases such measurements are motivated by a faint arcing noise, leaks or other
types of suspicious or irregular behaviour of the grading capacitor.

Open circuit obviously means that the conductors that connect the individual capacitor elements or
the electric connections from the capacitor stack to the terminals are broken or have loosened.
When voltage is applied this of course leads to internal discharges and arcing and the insulation
system of the grading capacitor quickly deteriorates.

When detecting an open circuit grading capacitor the common procedure among most utilities
appears to be to simply replace it immediately without doing any further examinations. However,
occasionally more careful investigations have been carried out in order to elucidate on the failure
mechanisms.

13
Such examinations on grading capacitors installed on GIS circuit-breakers [1] revealed that on a
few designs put into service in the 1980s the aluminium foils or bands used as conductors are
potentially weak points, especially the part connecting the top and bottom of the capacitor stack to
the terminals of the capacitor. Figure 9 shows a dismantled grading capacitor with the foil conductor
partly broken in this area. The oil has here been drained off, and parts of the epoxy housing has
been cut off and removed.

Figure 9 - Opened grading capacitor with partly broken foil conductor (see arrow).

Scanning electron microscopy (SEM) investigations clearly demonstrated that the damage was
caused by mechanical fatigue. The foil conductors are in this case as thin as 0,05 mm (50 µm), and
apparently not sufficiently mechanical robust to over time withstand the mechanical stresses and
shocks generated during circuit-breaker operations. Of this particular grading capacitor design a
substantial fraction the investigated objects (5 out of 12) had incipient fatigue damage, or partly or
completely broken foil conductors.

On another design from the same period a grading capacitor explosion inside a GIS triggered
investigations of internal parts of the remaining 11 capacitors of the circuit-breaker [1]. Also in this
case the aluminium foil between the capacitor stack and the terminals attracted attention. Although
no signs of fatigue or fractures were found, the investigators were not entirely convinced by the
quality of the manufacturing process of the capacitors. It seemed like they were assembled in a
rather hasty and careless way, and in several cases the foil conductor was very sharply bent,
thereby greatly increasing the risk of developing fatigue damage.
Mechanical fatigue is of course not the only possible process that can lead to open circuits in
grading capacitors. For example, conductor solderings, bondings or terminations may overheat and
melt or fail mechanically, and cause an open circuit. However, there are no indications in the
information provided by utilities that such failures are common. On the other hand, the replace-
without-investigation procedure most utilities apply on open circuit capacitors does not allow for
assessments of the underlying failure mechanisms.

Another type of flaw related to the internal electrical connections has also been discovered during
dismantling and inspection of grading capacitors. The electric leads between the bellows and the
end plate have in some cases been found to have become discoloured and brittle, see Figure 10.

Investigations made it clear that this was caused by electric discharges. The conductor loop is
associated with a certain inductance, and during nearby disconnector switching the resulting high
frequency travelling waves induce voltages of sufficiently large magnitude to give discharges from
the leads. Over time the leads became brittle.

The problem was solved simply by replacing the bare leads with insulated ones.

14
 Figure 10 - Discolouration of the electric leads between bellows and end plate caused by
discharge activity during disconnector switching operations

2.2.4 Ceramic grading capacitors

The information presented in the previous sections concerns foil grading capacitors. With regard to
ceramic grading capacitors far less little information about service experience have been available,
presumably since this technology is not much used by the contributing utilities.

One utility reported major problems with cracking of the ceramic elements, tracking problems in
cases where the circuit-breaker was in open position for prolonged periods of time, and also
comprehensive difficulties with capacitance or dissipation factors out of range. As a result of this,
the utility states that it often needs to replace the grading capacitors, and they are not buying new
circuit-breakers equipped with ceramic grading capacitors.

In contrast, one major circuit-breaker manufacturer claims an excellent service record for the
ceramic grading capacitors installed on their breakers. In a population of 7 000 capacitors with more
than 150 000 service years only one failure has occurred.

2.3 Diagnostic testing of grading capacitors - methods and practices

2.3.1 Test methods
A standard method for assessing the condition of a grading capacitor in the field is to measure
capacitance and tan δ . For this purpose there is a large variety of instruments available on the
market, from low-voltage bridges to high-voltage instruments. The latter usually measure at a
voltage of at least 10 kV.

Capacitor manufacturers usually state tolerances on capacitance and tan δ . Measurement of

capacitor parameters in the factory is normally done at 10 % and 100 % of the nominal voltage of
the capacitor. For that reason a measuring instrument that measures at least 10 kV is preferable to
low-voltage bridges.

Measurement should be done on a clean capacitor, under dry weather conditions. Proximity of live
equipment should be avoided; induction in the leads may result in incorrect readings. Care should
always be taken to avoid unnecessary loops in the measuring cables.

If a value outside the acceptable range is measured, it needs to be assessed if this is due to a
measuring error or other circumstances (insulator not clean, induction from nearby equipment,
humid weather, etc.). This can be done by removing the capacitor, cleaning it and repeating the
measurement on ground level. If both measurements give the same result, the capacitor should be
checked at nominal voltage.

In particular, if a capacitor is energized to its full rated voltage between two 10 kV range
measurements, the second one may show a much lower tan δ value than the first. This lack of
reproducibility is attributed to the so-called “Garton effect”, which relates to motion of ions in an
alternating field in the presence of solid boundaries [3]. The effect is observed on modern

15
capacitors with high electric fields. For old grading capacitor designs (“all-paper”) with rated voltage
up to 80 – 100 kV, the effect is far less pronounced and reproducible tan δ measurements can
usually be obtained.

Generally speaking, 10 kV range measurements can be a good and quick check for severe failures,
but minor flaws can be overlooked. Measurements at the nominal voltage of the capacitor are more
reliable, but are impractical in the field.

2.3.2 Utility practices

Some utilities report doing periodic measurements of tan δ and capacitance, typically at an interval
of a few years, and claim that this had effectively detected moisture ingress. Some utilities
performed an insulation resistance check across the circuit-breaker interrupter heads during routine
maintenance. This had also been effective in detecting grading capacitor moisture ingress. Many
utilities reported a routine visual inspection including a check for grading capacitor oil leaks. This
had evidently been effective on account of the large number of leaks that were detected before
failure.

At least one utility monitors the grading capacitors of shunt reactor circuit-breakers using thermal
imaging. The circuit-breaker is left open for three hours while energised before the image is
acquired. Care must be taken to rule out the effects of heating or thermal radiation from other
sources, such as the sun. When comparing the images of the capacitors of the three phases, one
having a higher temperature than the others will be clearly visible. This usually points to an internal
problem in the capacitor. Figure 11 shows the thermal image of a grading capacitor with severe
internal dielectric failures.

Figure 11 - Thermal image showing heating due to internal discharges in energized grading
capacitor

2.4 Discussion and conclusion

It is evident from this simple and relatively limited utility survey that corrosion and leaks on old
grading capacitors are quite common. Fortunately only very few disruptive failures are attributed to
this, but still the costs associated with the leaks are far from negligible.

Largely due to these problems, it appears that the lifetime of many of the old grading capacitor
designs is substantially less than the lifetime of the circuit-breaker itself. The latter may stay in
service for 30 - 50 years. Hence, an increasing number of incidents with leaking old grading
capacitors may be expected in the future if no precautions are taken. Some utilities have fully
recognized this issue and have decided to replace all grading capacitors as a part of a mid-life
revision of their circuit-breakers.

Fortunately, newer grading capacitor designs appear to have better technical solutions with regard
to sealing and corrosion protection, and are also more mechanically robust. It may thus be reason
to believe that the lifetime of capacitors manufactured in recent years is considerably longer.

A detailed and convincing explanation for the disruptive capacitor failures occurring during
switching of small inductive currents has not yet been found. Examinations of partly damaged
reactor breaker capacitors suggest that the failure originates in dielectric stresses of an unforeseen
nature and magnitude. This is an unsatisfactory situation, as understanding the phenomena of
16
course is essential for finding remedies against these very severe incidents. The fact that such
failures have occurred also on circuit-breakers equipped with controlled switching units further
increases the general uneasiness about this.

2.5 References
[1] M. Runde, J. Sletbak, J.-T. Ersdal, A. Mjelve, B. Skyberg, L. Søreide, and E. Åkra, ”Failures
of voltage grading capacitors in GIS circuit breakers”, paper no. A3-306, CIGRÉ session,
Paris, 2004.
[2] C-E. Sölver, “First results from on-going CIGRÉ enquiry on reliability on high voltage
equipment”, paper no. 101, CIGRÉ SC A3&B3 Joint Colloquium, Tokyo, 2005.
[3] G.C. Garton, “Dielectric loss in thin films of insulation liquids”, JIEE, vol. 88, no. 23, pp. 23-40.
1941.

17
3 Mechanical stress on grading capacitors
3.1 Introduction
The present chapter is concerned with the mechanical stresses imposed on circuit-breaker grading
capacitors. The mechanical stresses arising throughout the whole lifetime of the capacitor, including
transport, commissioning and service life, are considered as a basis for test requirements to be
proposed in Chapter 6.

Mechanical stresses arise from transportation, seismic activity on site and from operation of the
circuit-breaker. Test requirements addressing transport and seismic activity can be derived from
experience with components or equipment of similar size and weight and are covered by existing
standards.

Concerning mechanical stresses due to operation of the circuit-breaker, no common requirements

are known. The individual behaviour of each type of circuit-breaker influences the mechanical
impact on the grading capacitors. In order to analyze the stresses caused by operation of the
circuit-breaker, several measurements have been collected and are presented in the chapter. The
standard requirements covering transportation and seismic activity are also described.

3.2 Mechanical stress during circuit-breaker operation

To evaluate and interpret the mechanical stresses on grading capacitors acceleration
measurements have been carried out under different circumstances (on site and in laboratories)
with different types of equipment and with different processing of the recorded data.

3.2.1 Background to acceleration measurement and data processing

A force F applied to a body of mass m produces an acceleration a, such that:

F = m × a or a = F / m (1)

The acceleration is a measurable quantity that can be used to study the mechanical stressing of an
object.

The mechanical stresses on a body are usually studied using accelerometers. The accelerometer is
a piezoelectric device that provides an output signal proportional to acceleration. Since acceleration
is the rate of change of velocity, the velocity may be determined by integrating the accelerometer
output signal with respect to time. Similarly, since velocity is the rate of change of displacement, the
displacement may be determined by integrating the accelerometer output signal twice with respect
to time.

In the general case where the acceleration varies with time, the recorded data is integrated
numerically to obtain the velocity and displacement. An example taken from an earthquake
measurement in which acceleration data have been processed to obtain velocity and displacement
in three dimensions is shown in Figure 12.

18
 Figure 12 - General example of acceleration measurement and data processing to velocity
and displacement in three directions during an earthquake

In order to determine the possible mechanical stresses on the object, the mass, acceleration and
displacement must be known.

In addition to the amplitudes of the accelerations, the frequencies must also be considered. The
recorded accelerations in Figure 13, as well as in Figure 12 show high amplitudes that were typical
of those found in the collected measurements of grading capacitor stresses. Figure 13 shows the
acceleration against time measured on a Transient Recovery Voltage (TRV) control capacitor
mounted on a dead tank circuit-breaker during a Close-Open (CO) operation.

Figure 13 - Acceleration with respect to time, measured on a TRV control capacitor, mounted
to a dead tank circuit-breaker during a CO-operation

The acceleration may be determined as a function of frequency by means of the Fourier Transform,
in which a time varying function is expressed as a series of harmonic terms, each of the form:
19
 a (ω ) = Bsin(ωt + ϕB ) (2)

where

a is the acceleration (m/s 2 )

B is a constant (m/s 2 )
ω is de angular frequency (rad/s)
ϕ B is a phase angle (rad)
Due to the mathematical basis of the Fourier Transform, the lowest possible frequency that may be
covered by an analysis is equal to 1/T 0 , where T 0 is the length of the analysed time window. Figure
15 shows the acceleration as a function of frequency for the measurement of Figure 14 with a time
window from 27 to 227 ms.

Figure 14 - Acceleration with respect to frequency based on measurement shown in Figure 13

By integrating the acceleration terms twice with respect to time, the displacement (x) may be
expressed as a series of harmonic terms, each of the form:

x(ω ) = −(B/ω 2 ) × sin(ωt + ϕB ) (3)

Consequently, at higher frequencies the effect of acceleration on the displacement becomes

smaller. This is illustrated by Figure 15, which shows the displacement as a function of frequency
for the measurement of Figure 13.

20
 Figure 15 - Displacement with respect to frequency based on measurement shown in Figure
13

Damage is caused where the displacement results in cracks or delaminating, etc. Excluding
resonances the displacement of parts due to a mechanical shock can only be less than or equal to
the displacement measured on the outside surface.

In Figure 15, frequencies of more than 200 Hz only have values of displacement in the range of
micrometers. Assuming that such displacements do not harm objects like a grading capacitor in any
way, it is possible to neglect such frequencies. This will be true for short lived excitations such as
the mechanical shock associated with circuit-breaker operation.

Where it is possible to neglect the higher frequency components, they may be removed from the
recorded signals by means of low-pass filtering. Figure 16 shows the original curve from Figure 14
together with curves obtained by low-pass filtering with cut-off frequencies of 400 Hz, 200 Hz and
120 Hz. It may be seen that the high measured acceleration values of nearly 30g are associated
with high frequencies and consequently only generate small values of displacement. At frequencies
below 400 Hz, the acceleration does not exceed a level of 10g in the example shown.

21
 Figure 16 - Acceleration (y-axis) with respect to time (x-axis), measured at a TRV control
capacitor, mounted to a dead tank circuit-breaker during CO-operation, original measuring
signal and signal after low-pass filtering

In contrast to the mechanical shock, the more harmful case is a sustained oscillating stress that can
excite mechanical parts to vibrate with their mechanical resonant frequency. Such an oscillating
stress can occur during transport or as a result of the power frequency (50 Hz/100 Hz or
60 Hz/120 z).

3.2.2 Examples of acceleration measurements

Only a few measurements of acceleration on grading capacitors are available from circuit-breaker
manufacturers and independent institutes. The measurements cannot be assumed as being
representative for all circumstances, but will be used as examples and may give an idea of the
mechanical stresses.

The different investigations have been carried out under different focus of evaluation and are
therefore not always comparable. Only limited information is available concerning the measurement
techniques for the recorded data. For interpretation of the recorded signals only general statements
can be made here.

With regard to factors influencing acceleration measurements, the method of fixing the sensors to
the capacitor is assumed to be the same or at least similar for all the collected examples. The
sensors were fixed in a rigid, solid way directly to the end fittings or flange of the capacitor in axial
and radial directions. Measurements at other points on the circuit-breaker are of no use for
obtaining information on the stresses on the capacitor itself.
22
Factors that can be investigated from the available measurements are the following:

• Individual design of circuit-breaker including

- structure of the circuit-breaker to ground;
- type of operating mechanism (spring, hydraulic, pneumatic);
- generally the length and design of the mechanical chain from operating mechanism to
breaking unit.
• Direction of measurements in three dimensions
• Position of measurement: End of capacitor; near the terminal at the outer end of the interrupter;
near to the drive side of the interrupter
• Type of operation (Open (O), Close (C) and combinations) and velocity of operation mechanism
3.2.2.1 On-site measurement at three types of live tank circuit-breakers
The mechanical behaviour of the grading capacitors of three different 300 kV live tank circuit-
breakers was measured during operation.

Two of the investigated circuit-breakers are double-break SF 6 circuit-breakers. Circuit-breaker A

(Figure 17) is single-pole, spring operated. Circuit-breaker B (Figure 18) has a hydraulic drive
mechanism for all three poles on a common structure. Both circuit-breakers have grading capacitors
manufactured by the same manufacturer with capacitances of 1600 pF and 500 pF, respectively.
Circuit-breaker C (Figure 19) is a spring operated minimum oil circuit-breaker with 6 interrupters per
pole, each pole on a single frame. Its grading capacitors have a value of 1000 pF. The
measurements were performed with the circuit-breakers disconnected from the grid. The
manufacturing years of the breakers A, B and C were 1987, 1976 and 1971 respectively.

Figure 17 - One pole of circuit-breaker A

23
 Figure 18 - Circuit-breaker B, three poles on a common structure

Figure 19 – Circuit-breaker C with six units per pole

Measurements were performed with accelerometers attached directly to the capacitors, as shown in
Figure 20. In most cases, the measurements were performed with four accelerometers attached to a
capacitor during one circuit-breaker operation. Two accelerometers were attached at either end of
the capacitor, one in the axial direction and the other in a direction perpendicular to it.

The recorded signals were low-pass-filtered with a cut-off frequency at 16 kHz. The background
noise was recorded several times and a Fast Fourier-Transform (FFT) of the background noise
signal was used as a basis for the digital filters used during post processing.

Signals were recorded and analysed for circuit-breakers A, B and C in a systematic investigation.
Some examples will be shown here to give an impression of the variety of results.

24
 Figure 20 - Attaching the accelerometers to the capacitor

3.2.2.1.1 Effect of operating cycle

Figure 21 shows the difference in recorded signals during C and O operations. Some distinct
amplitude maxima occur during the operations. The highest amplitudes occur when the contacts are
opened. The maximum values recorded here are ± 80g. The maximum amplitudes for the other
circuit-breakers are less extreme but were also in the range of several tens of g’s.

25 80

20 60

15 40

10 20
Acceleration (g)

Acceleration (g)

5 0

0 -20

-5 -40

-10 -60

-15 -80

-20 -100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (s) Time (s)

a) Closing operation b) Opening operation

Figure 21 - Acceleration measurements on circuit-breaker B in axial direction on the driving

side

3.2.2.1.2 Direction of acceleration

Figure 22 shows the pattern recorded simultaneously by the four accelerometers. Accelerometers 1
and 4 are in the axial direction of the grading capacitor with number 1 near the drive side of the
interrupter. Accelerometers 2 and 3 are in a perpendicular direction.

The highest amplitude was measured by accelerometer number 1, corresponding to the axial
direction and nearer the drive side. The acceleration in the radial direction is substantially smaller.

25
 Accelerometer 1
100
Acceleration (g)

-100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Accelerometer 2
100
Acceleration (g)

-100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Accelerometer 3
100
Acceleration (g)

-100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Accelerometer 4
100
Acceleration (g)

-100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Time (s)

Figure 22 - Acceleration versus time during opening operation measured by four

accelerometers simultaneously

3.2.2.1.3 Post-processing
Taking into account that the extreme acceleration values are dominated by high frequencies which
may have a negligible impact on the test object because of the very low displacement, the recorded
signals have been post-processed by integrating twice. The resulting curves of velocity and
displacement are shown below the acceleration over the recorded time window in Figure 23. The
calculated displacement in this example is less than 0,5 mm.

26
 Acceleration
(m/s 2) 1000

-1000
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Velocity
0.05
(m/s)

-0.05
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
-4
x 10 Position
5
(m)

-5
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Time (s)

Figure 23 - Acceleration, velocity and displacement of opening operation

3.2.2.2 Further aspects of acceleration measurements

3.2.2.2.1 GIS circuit-breaker
Measurements were performed on the grading capacitors of two different, double-break 300 kV GIS
circuit-breakers from two manufacturers: Circuit-breaker D was put into service in 1980 with
hydraulic operating mechanism and vertical interrupters and circuit-breaker E was put into service
in 1981, with a pneumatic operating mechanism and horizontal interrupters.

The recorded signals were unfiltered. The signals recorded during C and O operations for the two
circuit-breaker types show the same characteristics as the live tank circuit-breakers: of the highest
acceleration was in the axial direction, generally with amplitudes of 20 to 40g and single peaks up
to 60g in the time domain. Since the signals were unfiltered, the results give little information about
the real stresses on the grading capacitors.

3.2.2.2.2 Repeatability of measurement

Differences between measurements under identical conditions have also to be considered. This is
illustrated in Figure 24, which shows acceleration measurements for circuit-breaker F, a modern
550 kV double-break SF 6 live tank circuit-breaker.

During the measurements, for each switching condition and for identical measuring points five
measurements were recorded for calculation of mean values. For measurements obtained under
identical conditions, visual comparison shows large differences in the traces of the acceleration,
with the peak values differing by more than 50 %. See, for example, the three measurements in
Figure 24. In addition to the acceleration patterns, the position of main contacts, puffer travel and
coil current were recorded.

Due to the large scatter in the measured values, only tendencies can be derived from these results.
The acceleration values are smallest during C operations due to the lower velocity of contact
movement. Acceleration values during O and CO operations are similar and generally higher than
2
those during C operations, reaching values of nearly 300 m/s or around 30g. Here, the acceleration
in the axial direction of the capacitor (z-direction) has the lowest values.

27
The large scatter between the measured records may also originate from the measuring technique.
Acceleration sensors measuring very low levels are very sensitive to changes in the connection
leads and this can influence the measured values. It is not possible to distinguish between changes
in the recorded signal due to the measurement technique and those due to variation in the
measured signal. Consequently, the results have to be interpreted with care.

Traces from the top: Position of main contacts, puffer travel, coil current and signals from four
accelerometers

Figure 24 - Measurements carried out during circuit-breaker operations (CO) under identical
conditions

3.2.2.2.3 Filtering and post-processing

In addition to unfiltered recording of acceleration signals, filtering and post-processing have been
applied to results from circuit-breaker G, a 420 kV double break SF 6 live tank circuit-breaker.

All of the unfiltered measurements show acceleration values in the range of up to 30g or even more
in some cases. No systematic differences due to the different measurement positions or directions
can be distinguished.

Only when the signal bandwidth is limited to a lower frequency (200 Hz in this case) do differences
in the stressing become obvious. This is illustrated in Figure 25. The values recorded in the axial
direction of the capacitors are small in comparison with those recorded in the radial direction. The
radial acceleration values are generally some g, in all cases below 10g.

28
 Figure 25 - Acceleration in the x-, y- and z-directions during CO operation without filter and
with 200 Hz filter (x = axial direction)

The unfiltered signals were post-processed to obtain the displacement as a function of frequency.
The results are shown in Figure 26. The displacement values are in the range of a millimetre at low
frequencies, decreasing rapidly with increased frequency to the micrometre-range at some hundred
Hertz. At frequencies above 300 Hz all displacement values are below 1 µm.

29
 Figure 26 – Displacement in y-direction in log-scale vs. frequency without filtering

Since an exact limit between the values of displacement that are harmless and those that are
harmful cannot be given, a frequency limit can be proposed as an alternative in order to achieve
comparable measurements and test procedures.

Displacements that can result in cracks, delamination, contact problems, etc are important since
these cause damage to the grading capacitor. The displacement of parts inside the capacitor is
normally the same as or smaller than the displacement measured on the outside surface, as long as
mechanical resonances do not occur. A frequency of around 200 Hz may be defined as a limit
above which frequencies have no significant influence on displacements and are not considered
destructive. In the evaluation of acceleration measurements, signals should be limited to this
frequency by low-pass filtering or by post-processing. The result is that the mechanical stress from
operation of the circuit-breaker is apparently below 10g as seen from Figure 16 and Figure 25.

A more harmful case of mechanical stress is an oscillating stress that can stimulate mechanical
parts to vibrate at their mechanical resonance frequency. Such an oscillating stress may occur
during transport. A circuit-breaker switching operation is more similar to a mechanical shock or a
random vibration that only influences the capacitor momentarily. Therefore, for switching
operations, the resonance will not be sustained.

3.3 Mechanical stress during transport

An exact description of the mechanical stresses during transport and handling of grading capacitors
cannot be given due to the diversity of the factors involved. The way in which the capacitors are
packaged is one of the most important factors. The grading capacitors can be shipped as separate
parts, packaged singly or in as a set, or as an integral part of the circuit-breaker assembly when
they are already fitted in the factory. In all cases the packaging method will influence the
mechanical stresses on the grading capacitor and its internal parts along a mechanical chain from
the transport system to the grading capacitor itself.

A German rule for packaging states that an acceleration of 4g has to be considered for stress due to
single shocks. This value corresponds to the maximum assumed acceleration due to a shunting
30
shock on railways. For any kind of transport, many more parameters have to be considered. In
general, these should be covered by IEC standards.

Transportation as part of the environmental conditions has been described in IEC documents
concerning classification and severity (IEC 60721-3-2 [1], IEC TR 60721-4-2 [2]) and environmental
testing (IEC 60068-2-27 [3]). The documents include transports by road, rail, water, air and vertical
transport such as cranes and lifts. The specified environmental conditions apply to the package
containing the product, if it is packed.

The severities of stresses specified in IEC 60721-3-2 are those which will have a low probability of
being exceeded. All specified values may be reached, but they do not occur permanently. Special
values and frequency of occurrence have to be additionally be specified, if applicable.

The mechanical conditions have been classified with their environmental parameters as follows:

• Sinusoidal and random stationary vibrations;

• Non-stationary vibration including shock;
• Free fall;
• Others not relevant to grading capacitors.
For each parameter a classification of severity is given. An idea of the stress values and
corresponding tests is given below. The values given correspond to the highest severity class 2M3:

• Stationary vibration sinusoidal:

- Displacement amplitude: 7,5 mm for 2-8 Hz;
- Acceleration amplitude: 20 m/s 2 for 8-200 Hz;
- Acceleration amplitude: 40 m/s 2 for 200-500 Hz.
NOTE The frequency range may be limited to 200 Hz for transportation on parts of the vehicle with high internal damping.

• Non-stationary vibration including shock:

- Shock response spectrum type I: 300 m/s 2 ;
- Shock response spectrum type II: 1000 m/s 2 .
(The response spectra are dependant on the duration of the impacting half-sine pulse: type I
with 11 ms and type II with 6 ms half sine wave. The concept of shock response spectrum is
described in detail in Appendix B of IEC 60068-2-27 [3]).
• Free fall:
- Mass of 20 kg to 100 kg: 1,2 m height;
- Mass more than 100 kg: 0,5 m height.
All possibilities concerning the mechanical stress due to transport should be considered as far as
possible: Transportation by truck, railway or ship; packaging method and handling requirements.
Shock indicators can be used to record the stresses during transport from factory to site. However,
not all types of stresses can be covered and analyzed by this method. Experience has shown that
damage to equipment during transport cannot always be explained. The experience from transports
of grading capacitors or similar goods will give an indication for selection of the severity of the
mechanical test conditions. Generally applicable guidance for a transport test cannot be given in
this report.

Considering the way in which the grading capacitors are transported, fitted to the circuit-breaker or
packed separately, the values mentioned above seem to be too high and are not recommended for
capacitor testing. In particular, the acceleration values of the non-stationary vibrations including
shock, as well as the free fall tests are outside the ranges normally considered by capacitor and
circuit-breaker manufacturers for transport.

31
3.4 Mechanical stress due to seismic activity
Mechanical stresses due to seismic activity have to be considered, since these are often a basic
requirement for circuit-breakers. The requirements for circuit-breakers are covered by standards,
which describe the method of qualification, e.g. by testing or calculation.

According to the rules for testing, the maximum acceleration response of the test object is proven
using a defined frequency range. In this way the response of the test object to an acceleration
applied at ground level and covering all relevant frequencies is obtained, including the condition
causing maximum stress at the circuit-breaker head. These acceleration values will also be
experienced by the grading capacitors.

The maximum acceleration response value in IEC standards is 0,8g and in IEEE standards 1,2g is
stated. The maximum frequency relevant for seismic testing is 35 Hz only.

3.5 Conclusions
As sources of the mechanical stresses on grading capacitors, the operation of the circuit-breaker,
transport and seismic activities on site have been considered. For transportation and seismic
activity, relevant standards can been used to cover the requirements. For the mechanical stresses
during circuit-breaker operations, measurements of acceleration have been collected and
evaluated.

There are only a few measurements of mechanical stresses due to circuit-breaker operation
available. Each of these was conducted under different and special circumstances. The
measurements indicate only certain tendencies in the different parameters. From consideration of
the measured acceleration as a function of frequency and the corresponding values of
displacement, the conclusion can be drawn that the measurements are not sufficient for a
comprehensive analysis of the stresses but can give a hint concerning the stress values.

All measurements show high acceleration values in the range of some tens of g. These high values
are probably harmless since they only occur at high frequencies for which the associated values of
displacement become very low.

Assuming that frequencies above 200 Hz have no significant influence on displacement and are
therefore harmless, the evaluation of acceleration measurements should be limited to this frequency
by low-pass filtering or by post-processing. The result is that the mechanical stress from operation
of the circuit-breaker is apparently below 10g.

The effects of resonances generated during circuit-breaker operation can be excluded since
damping makes the vibrations decay within a few hundred milliseconds and the resonance condition
is not sustained. Hence a switching operation is more similar to a mechanical shock or a random
vibration that only influences the capacitor momentarily.

Mechanical stresses due to transport are well covered by standard test procedures. By defining the
transport class, the ability of a capacitor to withstand mechanical stresses of the relevant severity
can be proven by test, including vibration, shocks and resonances.

Assuming that the mechanical stresses on circuit-breakers due to seismic activity are covered by
the relevant standards, the stressing of the grading capacitors is low in comparison to transport or
operation stressing.

3.6 References
[1] IEC 60721-3-2: Classification of environmental conditions – Part 3: Classification of groups of
environmental parameters and their severities – Section 2: Transportation

[2] IEC 60721-4-2: Classification of environmental conditions – Part 4-2: Guidance for the
correlation and transformation of environmental condition classes of IEC 60721-3 to the
environmental tests of IEC 60068-Transportation

[3] IEC 60068-2-27: Basic environmental testing procedures – Part 2: Tests – Test Ea and
guidance: Shock

32
3.7 Bibliography
IEC 62271-300: High-voltage alternating current circuit-breakers – Part 300: Seismic qualification of
alternating current circuit-breakers

IEC 60068-3-3: Environmental testing – Part 3: Background information. Seismic test methods for
equipment.

33
4 Electrical stresses and nominal voltage
4.1 Introduction
Circuit-breaker voltage grading capacitors are subjected to a range of electrical stresses that occur
when the circuit-breaker is in the open position and that arise during switching operations. When
considering the electrical stresses that are imposed on the grading capacitors and the consequent
insulation levels required of the capacitors, the following factors should be taken into account:

a) Voltage sharing across interrupters

The sharing of voltage across the interrupters of an open circuit-breaker is largely determined
by the values of the grading capacitors, but is also affected by stray capacitances, particularly
from the interrupters to earth, and by capacitor tolerances. The stray capacitances of the
interrupters in the open position are also significant. Consequently, grading capacitors may be
stressed by voltages higher than those that would occur under equal sharing of voltage across
the circuit-breaker pole.
b) Rated withstand voltages required across the open circuit-breaker
The dielectric requirements for the grading capacitor can only be based on the requirements for
the circuit-breaker. These requirements are specified in the relevant equipment standards
(circuit-breaker standard, GIS standard) of IEC and ANSI/IEEE. Special requirements of circuit-
breaker customers have to be considered in addition.
The rated withstand voltage values of the circuit-breaker normally cover all overvoltages in
service. Therefore, special insulation coordination aspects e.g. overvoltage protection or
substation configuration aspects do not need to be considered. Some special operating
practices, such as leaving the circuit-breaker in the open position for a prolonged time under
phase-opposition conditions, impose stresses that are particularly arduous for grading
capacitors and need to be considered separately.
c) Stresses arising out of switching duties
Grading capacitors are subjected to a range of transient electrical stresses that occur as a result
of switching operations. Some of these have significant implications for the grading capacitors
and must be adequately considered.
d) Coordination of withstand voltages between circuit-breaker interrupter and grading capacitor
Withstand voltage values for grading capacitors should be coordinated with those of the circuit-
breaker so as to avoid grading capacitor internal failures.
The above factors will be discussed in the following sections.

4.2 Voltage sharing

The effects of stray capacitance and capacitor tolerances on voltage sharing can be illustrated for a
circuit-breaker with two interrupters per pole using the equivalent circuit of Figure 27.

C p1 C p2

V1 Ce V2

Figure 27 - Equivalent circuit for voltage sharing, circuit-breaker with 2 interrupters per pole

In Figure 27, Ce represents the stray capacitance to earth and C p1 and C p2 represent the total
capacitances across the interrupters comprising the grading capacitor and stray capacitance of the
open interrupter.

The most unequal voltage sharing occurs when one terminal of the open circuit-breaker is
energised, the other terminal is at earth potential, the grading capacitor on the energised side has a
value at the lower end of the tolerance range and the capacitor on the opposite side has a value at
the higher end of the tolerance range. The voltage across the more highly stressed capacitor is

34
 (Cp2 + Ce )
Vp1 = V1 (4)
(Cp1 + Cp2 + Ce )

Expressing the increased voltage V p1 across the capacitor as a ratio compared to the value V 1 /2
under equal sharing gives

Vp1 2(Cp2 + Ce )
= (5)
V1 2 (Cp1 + Cp2 + Ce )

If the nominal capacitance of the grading capacitor C shunted by the stray capacitance C s of an
interrupter in the open position is represented by C p and the range of capacitor tolerance by ±ΔC p ,
then

Cp = C + Cs (6)

Cp1 = Cp − ∆Cp (7)

and
Cp2 = Cp + ∆Cp (8)

For a circuit-breaker with four interrupters per pole, the effects of stray capacitance and capacitor
tolerances on voltage sharing can be illustrated using the equivalent circuit of Figure 28.

C p1 C p2 C p3 C p4

V1 Ce Ce Ce V2

Figure 28 - Equivalent circuit for voltage sharing, circuit-breaker with 4 interrupters per pole

The capacitance to earth Ce at the centre of the circuit-breaker may be less than at the other two
positions if no support column is present, but in the present work it will be assumed to be the same
as this is a worst case. The most unequal voltage sharing occurs when one terminal of the open
circuit-breaker is energised, the other terminal is at earth potential, the grading capacitor nearest
the energised terminal has a value at the lower end of the tolerance range and the remaining
capacitors have a value at the higher end of the tolerance range. The voltage across the most
highly stressed capacitor is

A
Vp1 = V1 (9)
B

where

A = Cp2Cp3Cp4 + Ce (3Cp2Cp3 + 2Cp2Cp4 + Cp3Cp4 ) + Ce2 (2Cp2 + 2Cp3 + Cp4 ) + Ce3 (10)

and

B = Cp1Cp2Cp3 + Cp1Cp2Cp4 + Cp1Cp3Cp4 + Cp2Cp3Cp4 +

Ce (Cp1Cp2 + 2Cp1Cp3 + Cp1Cp4 + 3Cp2Cp3 + 2Cp2Cp4 + Cp3Cp4 ) + (11)
Ce2 (Cp1 + 2Cp2 + 2Cp3 + Cp4 ) + Ce3

Expressing the increased voltage V p1 across the capacitor as a ratio compared to the value V 1 /4
under equal sharing gives

35
 Vp1 4A
= (12)
V1 4 B

C p1 is given by Equation 7 in the previous case,

and

Cp2 = Cp3 = Cp4 = Cp + ∆Cp (13)

Values for the factor by which the voltage across a grading capacitor is increased due to unequal
voltage sharing compared with the value under equal voltage sharing are given in Table 1 for AIS
circuit-breakers and Table 2 for GIS circuit-breakers.

Stray capacitances are strongly dependent on the design of the circuit-breaker. The stray
capacitance to earth is dependent on the geometry of the interrupters, stress shields, etc and may
vary in a wide range. Values in the range 20 to 50 pF for AIS circuit-breakers and 100 to 400 pF for
GIS circuit-breakers have been taken. The stray capacitance across the interrupter in the open
position has a minor effect and tends to improve the voltage sharing. In order not to overestimate its
effect therefore, a modest value of 12 pF has been assumed. This value has been used for both AIS
and GIS circuit-breakers.

Values for the capacitor tolerance ΔC p of ±5 % for AIS circuit-breakers and ±3 % for GIS circuit-
breakers have been taken.

Table 1 - Factor by which voltage is increased compared with an equal voltage sharing, AIS
circuit-breakers

Nominal Number of interrupters per pole

capacitance
2 4
(pF)
Capacitance to earth C e (pF) Capacitance to earth C e (pF)
20 30 40 50 20 30 40 50
500 1,067 1,076 1,085 1,093 1,204 1,263 1,320 1,373
1000 1,059 1,063 1,068 1,072 1,143 1,175 1,206 1,237
1500 1,056 1,059 1,062 1,065 1,122 1,144 1,165 1,186
2000 1,054 1,057 1,059 1,061 1,111 1,127 1,144 1,160

36
 Table 2 - Factor by which voltage is increased compared with an equal voltage sharing, GIS
circuit-breakers

Nominal Number of interrupters per pole

capacitance
2 4
(pF)
Capacitance to earth C e (pF) Capacitance to earth C e (pF)
100 200 300 400 100 200 300 400
500 1,116 1,188 1,249 1,302 1,575 1,918 2,163 2,349
1000 1,075 1,117 1,155 1,190 1,347 1,581 1,770 1,926
1500 1,061 1,090 1,117 1,143 1,256 1,432 1,583 1,713
2000 1,053 1,076 1,097 1,118 1,207 1,348 1,473 1,584

The voltage sharing becomes more unequal as the stray capacitance to earth C e increases. GIS
circuit-breakers have higher stray capacitances to earth than AIS circuit-breakers and hence exhibit
the more unequal voltage sharing. The unequal voltage sharing is significantly greater for a circuit-
breaker with 4 interrupters per pole than for a circuit-breaker with two interrupters per pole. As the
capacitance value of the grading capacitors is increased, the influence of the stray capacitance to
earth reduces and the unequal voltage sharing becomes determined largely by capacitor tolerance.
Thus, in the case of an AIS circuit-breaker with two interrupters per pole, for example, the factor by
which the voltage is increased tends towards a minimum value of 1,05, corresponding to the
capacitor tolerance of 5 %.

It is common practice for circuit-breaker manufacturers to specify a grading factor, F, which is

chosen so as to take into account the possible increase in voltage across a grading capacitor due to
unequal sharing.

From the values shown in Table 1, a grading factor of 1,1 would cover all cases considered for an
AIS circuit-breaker with 2 interrupters per pole. A grading factor of 1,2 would cover cases
considered for an AIS circuit-breaker with 4 interrupters per pole where the grading capacitors have
a value of 1500 pF or more.

From the values shown in Table 2, a grading factor of 1,2 would cover cases considered for a GIS
circuit-breaker with 2 interrupters per pole where the grading capacitors have a value of 1000 pF or
more. A grading factor of 1,1 would cover some cases where the capacitance of the grading
capacitors is higher and the capacitance to earth is lower. For a GIS circuit-breaker with 4
interrupters per pole, the voltage unbalance is greater, with the voltage in some cases exceeding
twice that under equal sharing. Extreme values merit special attention. Higher values of grading
factor may be specified for these cases or, alternatively, different values of grading capacitor may
be selected to achieve a more equal voltage sharing across the circuit-breaker pole.

4.3 Nominal voltage

Currently, the nominal voltages for grading capacitors vary over a wide range with no definition and
no clear relation to the standardised highest voltages for equipment U m (for switchgear, the term
“rated voltage U r ” is used, in accordance with IEC 62271-1 [1]; standards of all other committees
use the term “highest voltage for equipment U m ”).

For a unique designation and furthermore for description of test recommendations, e.g. service
lifetime requirements, a nominal voltage U N will be defined for grading capacitors. The value of U N
implies that this voltage will be the guaranteed maximum voltage stress that the capacitor can
withstand for its whole lifetime. The proposed nominal voltage is higher by a factor equal to the
grading factor than the linear voltage distribution for phase to earth voltage across the circuit-
breaker, thus:

F  Ur 
UN =   (14)
N  3 

37
where N is the number of interrupters per pole and U r is the rated voltage of the circuit-breaker.

Values of nominal voltage for grading capacitors are given in Table 3 for circuit-breakers with 2 and
4 interrupters per pole for grading factors of 1,1 and 1,2.

Table 3 - Rated voltages of circuit-breakers and nominal voltages of grading capacitors

Circuit-breaker Grading capacitor

Rated voltage Phase-to-earth Grading capacitor nominal voltage
voltage UN
Ur U r /√3
kV (r.m.s. value)
N=2 N=4
kV (r.m.s.value) kV (r.m.s. value)
F = 1,1 F = 1,2 F = 1,2
300 173 95 104 52
362 209 115 125 63
420 242 133 145 73
550 318 175 191 95
800 462 254 277 139

4.4 Rated insulation level

Minimum values of insulation levels for grading capacitors may be derived from the rated insulation
levels of the circuit-breaker with which they are to be used, taking into account the number of
interrupters and unequal voltage sharing. The dielectric performance of the grading capacitors
should not be less than that of the circuit-breaker.

In the following sections, minimum values of withstand voltage for grading capacitors are derived
from the withstand voltages for circuit-breakers specified in IEC standards. The requirements of
other standards such as ANSI/IEEE and special customer requirements may not be covered by the
examples given. However, the same methodology can be applied using values specified in other
standards.

Where values specified in the standards are in the form of a combined voltage, i.e. an impulse
applied to one terminal of the open circuit-breaker and a power-frequency voltage applied to the
opposite terminal, the sum of both components of the combined voltage has been used as the basis
for calculation and application to the grading capacitors.

4.4.1 Short-duration power-frequency withstand voltage

Values of rated short-duration power-frequency withstand voltage are given in IEC 62271-1 [1] for
AIS circuit-breakers and IEC 62271-203 [2] for GIS circuit-breakers. Minimum values of rated short-
duration power-frequency withstand voltage U d for grading capacitors can be determined from the
values for the circuit-breaker given in the standards according to the following equation.

F
Ud = × U dCB (15)
N

Values derived from the above are given in

Table 4 for AIS circuit-breakers and Table 5 for GIS circuit-breakers.

38
Table 4 - Minimum short-duration power-frequency withstand voltage for grading capacitors,
AIS circuit-breakers (IEC 62271-1 [1])

Circuit-breaker Grading capacitor

Rated voltage Rated short- Short-duration power-frequency withstand voltage
duration power-
frequency
withstand voltage
Ur Ud Ud

kV (r.m.s. value) kV (r.m.s. value) kV (r.m.s. value)

Across open N=2 N=4
switching device
F = 1,1 F = 1,2 F = 1,2
300 435 239 261 131
362 520 286 312 156
420 610 336 366 183
550 800 440 480 240
800 1 150 633 690 345

Table 5 - Minimum short-duration power-frequency withstand voltage for grading capacitors,

GIS circuit-breakers (IEC 62271-203 [2])

Circuit-breaker Grading capacitor

Rated voltage Rated short- Short-duration power-frequency withstand voltage
duration power-
frequency
withstand voltage
Ur Ud Ud

kV (r.m.s. value) kV (r.m.s. value) kV (r.m.s. value)

Across open N=2 N=4
switching device
F = 1,1 F = 1,2 F = 1,2
300 595 327 357 179
362 675 371 405 203
420 815 448 489 245
550 925 509 555 278
800 1 270 699 762 381

4.4.2 Switching impulse withstand voltage

Values of rated switching impulse withstand level for AIS circuit-breakers are given in IEC 62271-1
[1]. A choice of two values is given for the rated switching impulse withstand voltage across the
open switching device for each rated voltage in range II (300 to 800 kV). In addition, values
specified for rated switching impulse withstand voltage across the isolating distance are also
applicable to some circuit-breakers, as specified in IEC 62271-100 [3].

Values for rated switching impulse withstand level for GIS circuit-breakers are specified in IEC
62271-203 [2]. Single values for rated switching impulse withstand voltage across the open
switching device are given for each rated voltage in range II. As for AIS circuit-breakers, values
specified for rated switching impulse withstand voltage across the isolating distance are also
applicable to some circuit-breakers, as specified in IEC 62271-203 [2].

Minimum values of grading capacitor switching impulse withstand voltage U s have been calculated
according to
39
 F
Us = × U sCB (16)
N

Values derived from the above are given in Table 6 for AIS circuit-breakers and Table 7 for GIS
circuit-breakers.

Table 6 - Minimum switching impulse withstand voltage for grading capacitors, AIS circuit-
breakers (IEC 62271-1 [1])

Circuit-breaker Grading capacitor

Rated Rated switching Switching impulse withstand voltage
voltage impulse withstand
voltage
Ur Us Us

kV kV (peak value) kV (peak value)

(r.m.s.
value) Across Across Across open switching Across isolating distance
open isolating device
switching distance
device N=2 N=4 N=2 N=4
F = 1,1 F = 1,2 F = 1,2 F = 1,1 F = 1,2 F = 1,2
300 750 700 413 450 225 520 567 284
(+245)
850 468 510 255
362 850 800 468 510 255 602 657 329
(+295)
950 523 570 285
420 950 900 523 570 285 685 747 374
(+345)
1 050 578 630 315
550 1 050 900 578 630 315 743 810 405
(+450)
1 175 646 705 353
800 1 425 1 175 784 855 428 1 004 1 095 548
(+650)
1 550 853 930 465

40
 Table 7 - Minimum switching impulse withstand voltage for grading capacitors, GIS circuit-
breakers (IEC 62271-203 [2])

Circuit-breaker Grading capacitor

Rated Rated switching Switching impulse withstand voltage
voltage impulse withstand
voltage
Us Us
Ur
kV (peak value) kV (peak value)
kV
(r.m.s. Across Across Across open switching Across isolating distance
value) open isolating device
switching distance
device N=2 N=4 N=2 N=4
F = 1,1 F = 1,2 F = 1,2 F = 1,1 F = 1,2 F = 1,2
300 850 700 468 510 255 520 567 284
(+245)
362 950 800 523 570 285 602 657 329
(+295)
420 1 050 900 578 630 315 685 747 374
(+345)
550 1 175 900 646 705 353 743 810 405
(+450)
800 1 425 1 100 784 855 428 963 1 050 525
(+650)

4.4.3 Lightning impulse withstand voltage

Values of rated lightning impulse withstand voltage are given in IEC 62271-1 [1] for AIS circuit-
breakers, where a choice of two values is given for each rated voltage in range II, and in IEC
62271-203 [2] for GIS circuit-breakers.

Minimum values of grading capacitor lightning impulse withstand voltage have been calculated
according to

F
Up = × U pCB (17)
N

Values derived from the above are given in Table 8 for AIS circuit-breakers and Table 9 for GIS
circuit-breakers.

41
 Table 8 - Minimum lightning impulse withstand voltage for grading capacitors, AIS circuit-
breakers (IEC 62271-1 [1])

Circuit-breaker Grading capacitor

Rated voltage Rated lightning Lightning impulse withstand voltage
impulse
withstand voltage
Ur Up Up

kV (r.m.s. value) kV (peak value) kV (peak value)

Across open N=2 N=4
switching device
F = 1,1 F = 1,2 F = 1,2
300 950 (+170) 616 672 336
1 050 (+170) 671 732 366
362 1 050 (+205) 690 753 377
1 175 (+205) 759 828 414
420 1 300 (+240) 847 924 462
1 425 (+240) 916 999 500
550 1 425 (+315) 957 1 044 522
1 550 (+315) 1 026 1 119 560
800 1 800 (+455) 1 240 1 353 677
2 100 (+455) 1 405 1 533 767

Table 9 - Minimum lightning impulse withstand voltage for grading capacitors, GIS circuit-
breakers (IEC 62271-203 [2])

Circuit-breaker Grading capacitor

Rated voltage Rated lightning Lightning impulse withstand voltage
impulse
withstand voltage

Ur Up Up

kV (r.m.s. value) kV (peak value) kV (peak value)

Across open N=2 N=4
switching device
F = 1,1 F = 1,2 F = 1,2
300 1 050 (+170) 671 732 366
362 1 175 (+205) 759 828 414
420 1 425 (+240) 916 999 500
550 1 550 (+315) 1 026 1 119 560
800 2 100 (+455) 1 405 1 533 767

4.4.4 Insulation level and nominal voltage

Table 4 through Table 9 highlight that for each rated voltage for the circuit-breaker there are a
number of sets of withstand voltage values for the grading capacitors, depending on the grading
factor and number of interrupters. If the requirements of additional standards, such as ANSI/IEEE,
had also been considered, then the number of sets of withstand voltage would have increased
further. It makes sense to seek a smaller number of standardised values that will cover the broadest
range of applications.

42
4.5 Stresses arising out of switching duties
In most cases, the electrical stresses to which grading capacitors are subjected during switching
operations are less severe than those associated with the withstand voltages described above.
However, some switching operations give rise to stresses that have significant implications for the
grading capacitors. In particular, a number of dielectric failures of grading capacitors have occurred
during switching of low inductive currents associated with shunt reactors, as reported in Chapter 2.
The electrical stresses to which grading capacitors are subjected during shunt reactor switching will
be described in detail in the next chapter.

4.6 Coordination of withstand voltages between circuit-breaker interrupter and grading

capacitor
The standard impulse withstand test procedures are based on a flashover probability of 10 %. To
prevent an internal failure of the grading capacitor during the test of a circuit-breaker, the withstand
voltage value should be chosen to be somewhat higher for the grading capacitor compared to the
interrupter. The rationale for this is that the internal flashovers in grading capacitors are non-self
restoring.

The short-duration power frequency test has no statistical background since testing is normally
without flashover. Nevertheless, to prevent an internal failure of the grading capacitor during the
test of a circuit-breaker, the withstand voltage value should be chosen to be somewhat higher for
the grading capacitor compared to the interrupter.

4.7 Conclusions
The factors which determine the electrical stresses on circuit-breaker voltage grading capacitors in
service have been described and a methodology for deriving the insulation levels required of the
grading capacitors has been demonstrated. Recommended withstand voltages based on the above
will be given in Chapter 6.

4.8 References
[1] IEC 62271-1: High-voltage switchgear and controlgear – Part 1: Common specifications

[2] IEC 62271-203: High-voltage switchgear and controlgear – Part 203: Gas-insulated metal-
enclosed switchgear for rated voltages above 52 kV

[3] IEC 62271-100: High-voltage switchgear and controlgear – Part 100: Alternating current
circuit-breakers

43
5 Electrical stresses during shunt reactor switching
As described in Chapter 2, a number of disruptive dielectric failures of circuit-breaker grading
capacitors have been reported. In general, there was a clear correlation between disruptive
dielectric failures and switching duty. Most of the circuit-breakers concerned had been switching low
inductive currents associated with shunt reactors and failure usually occurred on or shortly after a
breaking operation. The cause of the failures has not been fully understood.

This chapter is concerned with the electrical stresses imposed on circuit-breaker voltage grading
capacitors during shunt reactor switching. The relevant aspects of shunt reactor switching are
summarised. The corresponding electrical stresses imposed on the grading capacitors are
discussed and possible explanations for dielectric failures are proposed. Conclusions are drawn for
specification and test requirements for grading capacitors for shunt reactor switching applications.

5.1 Background
Working Group 02 of Study Committee No. 13 (A3) has studied the interruption of small inductive
currents. Its report was published in a CIGRE brochure [1], of which Chapters 1 and 2 present the
general theory of small inductive current switching, while Chapter 4 deals specifically with reactor
switching. Test requirements and an application guide for high-voltage circuit-breakers for shunt
reactor switching are published in IEC 62271-110 [2]. More recently, CIGRÉ Working Group A3.11
have compiled an application guide for IEC 62271-100 and IEC 62271-1, which includes a detailed
analysis of switching of inductive loads [3].

The salient phenomenon associated with shunt reactor switching may be illustrated using the basic
equivalent circuit shown in Figure 29. The circuit can be extended as required to take into account
neutral reactor grounding, interphase coupling or other features that may be applicable to the case
under consideration.

Lp Cp

Ls Lb

CB

V0 Cs CL L

Key
V0 source voltage (peak value) CL load capacitance
Ls source inductance Lb stray inductance
L load inductance Lp stray inductance in parallel with circuit-breaker
Cs source capacitance Cp capacitance in parallel with circuit-breaker
CB circuit-breaker

Figure 29 - Simplified single phase equivalent circuit

5.1.1 Load side oscillation

As the circuit breaker opens, it will tend to interrupt as the current reaches current zero. Since, for a
shunt reactor, the load is predominantly inductive at the power frequency, the current zero
coincides with the peak of the voltage wave. When interruption occurs, therefore, the capacitive
elements on the load side of the circuit-breaker are charged to a level corresponding to the voltage
peak. The trapped energy discharges through the inductance of the shunt reactor, resulting in the

44
load side oscillation. The frequency of the load side oscillation is largely determined by the
capacitance C L and inductance L of the load:

1
fL = (18)
2π LCL

The frequency is typically in the range 0,5 – 5 kHz. The load side oscillation decays within a few
tens or hundreds of milliseconds.

5.1.2 Current chopping

In practice, current chopping occurs as instability forces the arc current to zero, causing interruption
to occur before the natural power frequency current zero. The power frequency current i ch flowing
through the inductance of the reactor at the moment of interruption then contributes to the trapped
energy, increasing the amplitude of the load side oscillation above that of the power frequency
voltage. The first peak of the oscillation is of the same polarity as the power frequency voltage at
the moment of interruption and is referred to as the suppression peak. The second peak of the
oscillation is referred to as the recovery voltage peak and usually corresponds to the maximum
stress across the circuit-breaker.

Making the simplifying assumptions that interruption occurs close to the power frequency voltage
peak, that the arc voltage drop may be neglected and that damping does not significantly affect the
suppression peak, the amplitude V m of the suppression peak (or chopping overvoltage) may be
calculated from the trapped energy, thus:

Vm = [V
0
2
+ ( L / CL )ich 2 ] (19)

Except in the case of vacuum circuit-breakers, the level of chopped current may be calculated using
the ‘chopping number’, which is a characteristic of the circuit-breaker and is usually determined
during testing. The level of chopped current is dependent on the chopping number λ for a single
interrupter, the number N of interrupters in series per pole and the effective capacitance C t in
parallel with the circuit-breaker terminals:

ich = λ Ct (20)

Where
Cs × CL
Ct = Cp +
Cs + CL

The amplitude of the suppression peak therefore becomes:

Vm = [V0 2 + NLλ2 (Ct CL )] (21)

From Equation 21 it will be appreciated that the chopping level of the circuit-breaker, the inductance
of the shunt reactor and the capacitance of the grading capacitors all tend to increase the level of
the suppression peak. For SF 6 puffer circuit-breakers, the chopping number generally increases
with arcing time, while for oil circuit-breakers and some air-blast breakers it is fairly independent of
arcing time [2].

The voltage waveform at the load side of the circuit-breaker and the resultant voltage across the
circuit-breaker, or recovery voltage, are of the form shown in Figure 30.

45
 Vm
V0

Figure 30 - Load side oscillation

5.1.3 Reignition
Following interruption, if the rising voltage across an interrupter exceeds the recovering dielectric
strength of the contact gap at a given moment, the arc will reignite. When interrupting small
inductive currents, the dielectric recovery of the circuit-breaker is given by its cold recovery
characteristic [4, 5]. The likelihood of reignitions and the voltage magnitudes at which they occur
are dependent on the type of circuit-breaker. Reignition may occur at any voltage up to the recovery
voltage peak.

Reignition re-establishes a conducting path through the circuit-breaker and initiates three kinds of
oscillations:

- the first parallel oscillation occurs as the capacitance C p across the circuit-breaker, which
includes the grading capacitors, is discharged through the circuit-breaker;
- the second parallel oscillation occurs as the capacitances C s and C L on either side of the circuit-
breaker are equalised in voltage;
- the main circuit oscillation occurs as the voltage on C s and C L returns to a steady state value.
An example of the voltage waveform on the load side of the circuit-breaker during reignition is
shown in Figure 31.

46
 Vm
V0

Figure 31 - Reignition

Following a reignition, the circuit-breaker will normally run in arc until interruption occurs at the next
power frequency current zero. The load side voltage during reignition followed by interruption at the
power frequency current zero is shown in Figure 32.

Successful interruption
Reignition

Figure 32 - Reignition followed by successful interruption

5.1.4 Multiple reignitions

Following a reignition, interruption may occur at a natural zero crossing of the current during the
second parallel oscillation or the main circuit oscillation. Whether or not interruption occurs depends
on the characteristics of the circuit-breaker and the rate of change of current with time at the zero
crossing.

Where interruption following reignition does occur, conditions may be such that the trapped energy
on the load side of the circuit-breaker is greater than it was just before the reignition. A new load
side oscillation is therefore established with a greater prospective peak voltage and a further
reignition may occur. Such a reignition is likely to occur at a higher voltage than the previous
reignition due to the increased contact gap. In some cases, a condition is possible where multiple
reignitions and interruptions give rise to voltage escalation.

Vacuum circuit-breakers are able to interrupt high frequency currents and may give rise to multiple
reignition and voltage escalation. Other types of circuit-breaker are generally not able to interrupt
high frequency currents and the risk of multiple reignitions is low.

47
5.1.5 Interaction between phases
Coupling between phases may occur due to stray capacitance and mutual inductance. The
interruption behaviour in one phase may be influenced by events in other phases: Induced current
chopping may occur where an induced current creates a current zero before the power frequency
current zero; induced reignition may occur where an induced voltage causes a sufficient increase in
voltage across an open circuit-breaker pole. In addition to induced transients, a transfer of energy
between phases during the load side oscillation may occur after the interruption of at least two
phases.

5.2 Electrical stresses imposed on grading capacitors

5.2.1 Circuit-breaker in closed position
While the circuit-breaker is in the closed position, the grading capacitors are not exposed to any
dielectric stress other than that due to the surrounding electric field, which will be greatest at the
edges. The electric field around the capacitor is significant and is an important consideration in
circuit-breaker design.

5.2.2 Circuit-breaker opening operation

5.2.2.1 Recovery voltage
Following interruption, the grading capacitors are stressed by the recovery voltage as shown in
Figure 30. The voltage waveform is the difference between the phase-to-earth power frequency
voltage of the source and the load side oscillation. The maximum value of the recovery voltage
occurs at the recovery peak. Assuming the source voltage does not change appreciably during the
first half cycle of the load side oscillation and if damping is neglected, the recovery peak has a
value of:

VR = V0 + [V0 2 + NLλ2 (Ct / CL )] (22)

The value of the recovery peak exceeds 2 times the peak value of the source voltage V 0 by an
amount depending on the level of chopped current and hence the chopping number of the circuit-
breaker and the characteristics of the circuit in which it is applied. It is therefore difficult to be
precise about an upper limit to the value of the recovery peak. CIGRÉ Technical Brochure 50 [1]
states that suppression peak overvoltages for reactors with earthed neutral are below 2 p.u. for
normal reactor sizes if switched with a modern circuit-breaker. It seems unlikely, therefore, that the
recovery peak will exceed 3 times the peak value of the source voltage for a modern circuit-breaker.
However, for small reactors or reactors earthed through a reactance, the suppression peak
overvoltage may exceed 2 p.u. and the recovery peak may therefore exceed 3 p.u.

5.2.2.2 Reignition
Where a reignition occurs, the voltage across the grading capacitors collapses within a fraction of a
microsecond and the first parallel oscillation is established. The grading capacitors are discharged
through the near-short circuit formed by the circuit-breaker interrupters and the reignited arc. The
frequency of oscillation is dependent on the capacitance in parallel with the circuit-breaker and the
inductance of the loop:

1
fp = (23)
2π LpCp

The frequency is typically in the range 1 – 10 MHz. The amplitude I p of the current is dependent on
the voltage V p across the grading capacitors at the instant of reignition:

I p = Vp Cp Lp (24)

The resulting current is of short duration but may reach a value of several kiloamperes.

The grading capacitors are not directly influenced by the second parallel oscillation or the main
circuit oscillation, since the circuit-breaker is running in arc during these oscillations and the
capacitors are effectively short-circuited.
48
Where multiple reignitions occur, the grading capacitors will be exposed to repeated cycles of
stressing by the load side oscillation and discharging through the first parallel oscillation.

In some utilities, circuit-breakers switching shunt reactors are provided with controlled opening to
reduce the likelihood of reignition.

5.2.3 Circuit-breaker in open position

Following successful interruption, the load side oscillation decays leaving the capacitors stressed
by the phase-to-earth power frequency voltage of the source. In some utilities, it is practice to open
a sequential disconnector in series with the shunt reactor circuit-breaker thus eliminating the stress
on the grading capacitors while the circuit-breaker is in the open position. Elsewhere, the grading
capacitors may remain stressed by the power frequency voltage until the next closing operation,
typically a period of several hours or days.

5.2.4 Circuit-breaker closing operation

Prestriking during a making operation imposes similar stresses on the grading capacitors to those
imposed by reignition, with a rapid voltage collapse, stressing of the inductive elements of the
grading capacitor and high instantaneous currents associated with the first parallel oscillation.

The current is dependent on the voltage across the grading capacitors at the instant of prestrike,
the maximum value of which corresponds to the peak of the power frequency voltage V 0 . Although
this is less than the maximum voltage across the capacitors that may occur at reignition, the current
may still reach a value of a few kiloamperes.

In some utilities, it is the practice to use controlled closing for shunt reactor circuit-breakers. The
closing operation is timed to occur around the peak of the power frequency voltage in order to
protect the shunt reactor from the effects of an excessive magnetic flux.

5.2.5 Unequal voltage sharing

5.2.5.1 Unequal voltage sharing under static conditions
The distribution of voltage across the open interrupters of a circuit-breaker pole is affected not only
by the capacitance of the grading capacitors, but also by stray capacitances, particularly the
capacitance to earth, and unequal sharing of voltage between the grading capacitors.

The effect of stray capacitance on voltage sharing can be illustrated in the case of a circuit-breaker
with two interrupters per pole. When the circuit-breaker is in the open position, the impedances of
the load and the source have negligible influence on the voltage distribution at low frequencies and
the equivalent circuit can be simplified to that of Figure 33. C p is the capacitance of the grading
capacitors and C e is the stray capacitance to earth of the circuit-breaker.

49
 Cp Cp

V0 Ce Vm

Key
V0 source voltage
Vm load voltage
Cp capacitance of grading capacitor
Ce capacitance to earth of circuit-breaker

Figure 33 - Equivalent circuit for voltage sharing under static conditions

In general, the voltage on the source side capacitor is determined by

dVp1
dV0
d t
(
Ce + Cp −
dVm
dt
Cp)
= (25)
dt 2Cp + Ce

and the voltage on the load side capacitor is determined by

dVp2
dV0
Cp −
dVm
Ce + Cp ( )
= dt dt (26)
dt 2Cp + Ce

Under steady state conditions, V m = 0 and the voltage on the source side capacitor becomes

Cp + Ce
Vp1(t ) = V0 (t ) (27)
2Cp + Ce

and the voltage on the load side capacitor becomes

Cp
Vp2 (t ) = V0 (t ) (28)
2Cp + Ce

Thus, at steady state, the voltage is shared between the grading capacitors in the ratio

1 + Ce Cp (29)

with the greater voltage across the source-side grading capacitor.

Under transient conditions, i.e. during the load side oscillation, the voltages on the source and load
side grading capacitors may be determined from Equations (25) and (26) taking into account the
changes ΔV 0 in V 0 and ΔV m in V m from the initial condition where the voltage on both capacitors is
zero. The voltage on the source side capacitor becomes

∆V0 (Ce + Cp ) − ∆VmCp

Vp1 = (30)
2Cp + Ce

50
and the voltage on the load side capacitor becomes

∆V0Cp − ∆Vm (Ce + Cp )

Vp2 = (31)
2Cp + Ce

Particularly during the first half cycle of the load side oscillation, the change ΔV m in V m will be
larger than the change ΔV 0 in V 0 and hence the voltage across the load side grading capacitor will
be greater.

Both at steady state and during the load side oscillation, the inequality in voltage sharing is
dependent on the value of the stray capacitance to earth compared to the capacitance of the
grading capacitors. Since the stray capacitance is normally significantly less than the capacitance
of the grading capacitors, the inequality in voltage sharing will be modest. If, for example, the stray
capacitance has a value of 20 % of that of a grading capacitor, the voltage across the more
stressed grading capacitor at steady state may be increased by about 9 % compared to an equal
division. The increase in voltage across the more stressed capacitor during the early part of the
load side oscillation is similar. The values given correspond to a rather extreme case of a GIS
circuit-breaker with high stray capacitance and low value grading capacitors. The increase in
voltage across a grading capacitor due to unequal sharing under static conditions will usually be
less than this.

5.2.5.2 Unequal voltage sharing under dynamic conditions

Unequal voltage sharing may arise under dynamic conditions where reignition or prestriking does
not occur simultaneously across all interrupters of a pole. In practice, this is likely to occur in all
cases to a greater or lesser extent. When an interrupter reignites or prestrikes before the other
interrupters of the pole, the collapse of voltage across the interrupter causes an instantaneous
redistribution of voltage around the adjacent components. The voltage across the grading
capacitors of the other interrupters will tend to increase until each interrupter reignites or prestrikes.

The redistribution of voltage and the subsequent transient response as each interrupter reignites or
prestrikes is dependent on the stray impedances associated with the circuit-breaker and its
connections. The resulting voltages will be difficult to calculate unless the stray impedances are
known in detail. There is also little information available on the possible intervals between reignition
or prestriking of the interrupters. Furthermore, WG A3.18 knows of no studies in which this
phenomenon has been investigated. The following sections attempt to predict the stresses that
might result due to unequal voltage sharing under dynamic conditions. Much of that which follows is
speculative, since it has not been possible to confirm the predictions by measurement.
Nevertheless, some general conclusions about the voltage sharing under dynamic conditions may
be drawn.

The effects on the grading capacitor of a delay between closing of interrupters during prestriking or
reignition are illustrated in the following example of a circuit-breaker with two interrupters per pole
switching a 400 kV 200 MVA shunt reactor. The circuit is shown in Figure 34. The component
values used in the example are given in the key to the figure.

51
 Ls R b1 L b1 L b2 R b2

V0 Cs Cp Rp Cp Rp CL L RL
Lp Lp
Ce

Key
V0 Source side voltage (peak value) 326,6 kVpeak

Ls Source inductance 12 mH
Cs Source capacitance 60 nF
R b1 Damping, source side connections 2Ω
L b1 Inductance, source side connections 6 μH
Lp Inductance, grading capacitor 840 nH
Cp Capacitance, grading capacitor 750 pF
Rp Resistance, grading capacitor 7,75 Ω
Ce Capacitance to earth, circuit-breaker 130 pF
L b2 Inductance, load side connections 6,4 μH
R b2 Damping, load side connections 2Ω
CL Load capacitance 2,6 nF
L Load inductance 2,4 H
RL Load damping 1,0 MΩ

Figure 34 - Equivalent circuit for voltage sharing under dynamic conditions

Due to the short time window in which voltage sharing under dynamic conditions takes place, the
events will be dominated by C s , C L and the components in the loop between them.

5.2.5.3 Prestriking
The voltage stress across grading capacitors during a circuit-breaker making operation where there
is a delay between prestriking of interrupters is illustrated for the two cases where the source side
interrupter prestrikes before the load side interrupter and where the load side interrupter prestrikes
before the source side interrupter. In both cases, the voltages across the grading capacitors have
been simulated for the circuit described above. The simulations represent the worst-case scenario
where the prestrike occurs at the peak of the source voltage waveform.

When the source side interrupter prestrikes before the load side interrupter, the voltage across the
load side capacitor rises from its initial value, as described by Equation 28, towards the
instantaneous value of the source voltage and can be expected to overshoot, since the resistances
around the circuit contribute little damping. The frequency of the initial transient is dominated by the
capacitance of the grading capacitor C p and the inductance L b1 + L p + L b2 and approximates to

1
fp = (32)
2π ( Lb1 + Lp + Lb2 )Cp

The value of the source capacitance C s is generally much larger than that of the grading capacitor,
C p , and its voltage tends not to vary during the timescale of the initial transient. The load side
capacitance C L shunted by the inductance L of the reactor does not respond significantly during this
timescale and remains essentially at earth potential during the period of interest.

Figure 35 shows the transient rise in voltage across the load side grading capacitor when the
source side interrupter prestrikes at time t = 0 as simulated for the circuit of Figure 34.

52
 Source side interrupter prestrikes

-2,0 0,0 2,0 4,0 6,0 8,0

time (µs)

Figure 35 - Initial rise in voltage across load side grading capacitor when source side
interrupter prestrikes

In the simulation, the initial voltage across the load side grading capacitor when the source side
interrupter prestrikes is 150 kV. The frequency of the transient voltage across the capacitor
following the prestrike is 1.8 MHz. The voltage reaches a peak value of 399 kV, equal to 1,22 times
the peak value of the source voltage and occurring around 300 ns after the first prestrike.

When the load side interrupter prestrikes before the source side interrupter, the voltage across the
source side capacitor increases from its initial value, as described by Equation 26, towards the
instantaneous value of the source voltage and overshoots. The behaviour of the transient voltage is
similar to that described above for the previous case.

In the simulation, the initial voltage across the source side grading capacitor when the load side
interrupter prestrikes is 176 kV. The voltage reaches a peak value of 380 kV, equal to 1,16 times
the peak value of the source voltage and occurring around 300 ns after the first prestrike.

When calculating the transient rise of voltage across the grading capacitors, the values of R b1 and
R b2 will generally be difficult to determine. However, the effects of damping may be neglected from
the calculation to obtain a worst-case result for either grading capacitor of approximately 1,5 times
the peak value of the source voltage or 3 times the initial voltage across the capacitor.

5.2.5.4 Reignition
During a breaking operation, the voltage across the grading capacitors is zero until the circuit-
breaker interrupts, following which the recovery voltage is shared between the source and the load
side grading capacitors according to Equations 30 and 31. Where a reignition occurs, the
redistribution of voltage shows similar behaviour to that when a prestrike occurs, except that the
amplitude of the transient may be larger.

The voltages across grading capacitors have been simulated using the circuit of Figure 34 for the
two cases where the source side interrupter reignites before the load side interrupter and where the
load-side interrupter reignites before the source-side interrupter. The simulations represent the
worst-case scenario where reignition occurs at the recovery voltage peak. Since reignitions occur
when the contact gap is small, interruption at current zero has been assumed and the effect of
current chopping has not been taken into account.

When the source side interrupter reignites before the load side interrupter, the voltage across the
load side grading capacitor increases towards the instantaneous value of the recovery voltage and
53
overshoots. In the simulation, the initial voltage across the load side grading capacitor when the
first reignition occurs is 349 kV. The voltage across the capacitor reaches a peak value of 765 kV,
equal to 2,34 times the peak value of the source voltage.

When the load side interrupter reignites before the source side interrupter, the voltage across the
source side grading capacitor increases towards the instantaneous value of the recovery voltage
and overshoots. In the simulation, the initial voltage across the source side grading capacitor when
the first reignition occurs is 298 kV. The voltage across the capacitor reaches a peak value of 779
kV, equal to 2,39 times the peak value of the source voltage.

Where the effects of damping are neglected from the calculation, a worst-case result is obtained for
either grading capacitor of approximately 3 times the peak value of the source voltage.

5.2.5.5 Discussion on unequal voltage sharing under dynamic conditions

From the above it will be seen that when one interrupter of a pair prestrikes or reignites before the
other, the voltage across the second interrupter, and hence its associated grading capacitor, rises.
The rate of rise of the voltage is rapid and largely determined by the capacitance of the grading
capacitor, the inductance of the connections to the circuit-breaker and the associated resistive
damping. Other components have little influence on the shape of the initial transient. The
prospective amplitude of the transient is dependent on the voltage across the circuit-breaker pole
when the first prestrike or reignition occurs. The level that the transient voltage may reach before
the second interrupter prestrikes or reignites is dependent on the interval that elapses until the
second prestrike or reignition occurs. If the interval is sufficient, the voltage across the grading
capacitor may approach a value of up to 1,5 times the peak value of the source voltage in the case
of a prestrike and up to 3 times the peak value of the source voltage in the case of a reignition.

An interval between prestriking or reignition of the interrupters of a pole may be caused by

mechanical differences in operation of the interrupters and by the time lags affecting gas breakdown.
The time lag between the application of a sufficiently high voltage to a gas gap and the subsequent
breakdown comprises the statistical and formative time lags. The statistical time lag is the time that
elapses until an electron capable of initiating breakdown appears in the gap, while the formative
time lag is the time required for the breakdown to develop.

Boeck et al [6] studied the insulating behaviour of SF 6 in the presence of fast transients. Tests were
performed in which a fast oscillating voltage with rise time of about 50 ns was applied to a coaxial
gap in SF 6 at 0,3 MPa and the voltage at which breakdown occurred recorded. The results are
shown in Figure 36 for positive polarity and in Figure 37 for negative polarity. Results obtained for
double exponential impulse voltages 0,4/50 μs are also shown for comparison. The latter are shown
as points within an envelope. For both polarities, the lowest measured breakdown voltage under
impulse voltage is shown by a dotted line.

54
 Figure 36 - V-t curve for coaxial gap in SF 6 (positive polarity) [6]

Figure 37 - V-t curve for coaxial gap in SF 6 (negative polarity) [6]

The results show that the voltage at which breakdown occurred was dependent on the rise time of
the applied voltage. Results obtained for the fast oscillating voltage show considerable scatter, with
some values of breakdown voltage significantly higher than those for the double exponential
envelope. A distinct polarity effect can be seen, with breakdowns under the positive polarity fast
oscillating voltage occurring predominantly at around 50 ns, while those for the negative polarity
occurred predominantly in the range 100-200 ns.

The values of breakdown voltage and time to breakdown that were observed under fast oscillating
voltages support the view that unequal voltage sharing under dynamic conditions may be significant.
The investigations were performed under cold, steady state conditions and did not reproduce those
of a circuit-breaker interrupter, where hot gases, ions and gas flow will have an influence. There is
a clear need for further studies in this area.

55
5.2.6 Capacitor discharge current
A simulation of the current through a grading capacitor following a prestrike is shown in Figure 38.
The example corresponds to the circuit of Figure 34, with the prestrike occurring at the voltage peak.
The current reaches a peak value in excess of 4000 A. The frequency of oscillation is approximately
6 MHz.

-1,0 -0,5 0,0 0,5 1,0 1,5

time [μs]
Figure 38 - Current through grading capacitor at prestrike

In the case of a reignition at the recovery voltage peak, if the interval between reignition of the two
interrupters is such that the second interrupter reignites when the transient voltage across it is near
its peak value, the peak value of the current would increase to around 19 000 A.

In general, the transient current that flows when a grading capacitor is discharged following a
prestrike or reignition may be determined using the equivalent circuit for the capacitor and its
associated interrupter shown in Figure 39.

Lp Cp Rp

Figure 39 - Equivalent circuit for capacitor

If the capacitor is discharged from an initial voltage V p , the transient current at time t from the
discharge is

i = Ae −αt sinβt (33)

where

Rp 1 Vp
α= , β = ω 0 2 − α 2 , ω0 2 = and A =
2Lp LpCp βLp

On account of the high resonant frequency of the loop formed by the grading capacitor and the
associated interrupter, a test performed by discharging the capacitor through a rod-rod gap in air,
as recommended in some standards, may not be able to fully excite the resonance. The current in
such a test may therefore be less than that which occurs in service.
56
Since many shunt reactors are switched daily, the grading capacitors may be subjected to the
stresses associated with prestrikes hundreds of times per year and the stresses associated with
reignitions several tens of times per year.

5.3 Discussion on levels of electrical stresses

Grading capacitors experience electrical stresses during shunt reactor switching as follows:

Following successful interruption, the open circuit-breaker and hence the grading capacitors are
stressed by a recovery voltage comprising the difference between the power frequency phase-to-
earth voltage of the source and the load side oscillation. The load side oscillation has a frequency
in the range 0,5 to 5 kHz and decays within a few tens or hundreds of milliseconds. The peak value
of the load side oscillation exceeds the peak value of the source voltage by an amount depending
on the level of chopped current and hence difficult to specify precisely. IEC 62271-110 specifies
values of the prospective transient voltage of the load circuit to be used in circuit-breaker shunt
reactor switching tests. The peak values of the prospective transient voltage specified in the
standard have been used in the present work to determine maximum values of the recovery voltage
across the circuit-breaker. The maximum values of the recovery voltage peak so derived are
compared with the rated switching impulse withstand voltage across the open switching device of
IEC 62271-1 [7] in Table 10.

Table 10 - Comparison between recovery voltage peak and rated switching impulse withstand
voltage of IEC 62271-1

Rated Peak voltage of load Recovery voltage Rated switching impulse withstand voltage across
voltage circuit peak open switching device
Ur Vm VR Us

kV (r.m.s. kV kV kV (peak value)

value)
300 465 710 750
850
362 562 857 850
950
420 652 997 950
1 050
550 853 1303 1 050
1 175
V m corresponds to u c as defined in 6.115.6 of IEC 62271-110.
NOTE
The recovery voltage peak is assumed to be the sum of the source voltage peak and the peak voltage of the load circuit

2
VR = U r + Vm
3

The recovery voltage peak is comparable to and may exceed the rated switching impulse withstand
voltage across the open switching device. Therefore, the recovery voltage must be taken into
consideration and allowance made for current chopping when specifying grading capacitors for
shunt reactor switching applications.

Following the decay of the load side oscillation, the open circuit-breaker and hence the grading
capacitors remain stressed by the phase-to-earth power frequency voltage of the source. The stress
persists typically for several hours or days until the circuit-breaker is closed, or for a few minutes
where sequential disconnection is used. The maximum power-frequency voltage that may exist
across the open circuit-breaker is well below the rated short-duration power-frequency withstand

57
voltage across the open switching device. However, the long-term performance of the grading
capacitors if stressed with the power-frequency voltage for long periods must be addressed.

The sharing of voltage between the grading capacitors of a pole is affected by stray capacitances.
Under static conditions, a grading capacitor may be stressed by a few percent more than its
nominal share of the voltage and this must be taken into account by the use of an appropriate factor
when determining test levels for grading capacitor units.

Under dynamic conditions, where one interrupter of a pole prestrikes or reignites momentarily
before the others, the stress across the grading capacitor of the other interrupters will begin to
increase rapidly. Depending on the interval that elapses before an interrupter prestrikes or reignites,
the increase in voltage across the interrupter and, consequently, across its grading capacitor may
be significant. There is, however, very little information available on the possible intervals that may
occur or the consequent levels of voltage across the grading capacitors.

For a circuit-breaker having two interrupters per pole, under worst-case conditions the stress across
the grading capacitor may, in principle, reach a level of approximately 1,5 times the peak value of
the source voltage in the case of a prestrike and 3 times the peak value of the source voltage in the
case of a reignition, as given in Table 11. It should be emphasised that the values given are
speculative worst-case values, since no experimental data is available on the actual stresses that
might occur due to unequal voltage sharing under dynamic conditions.

Table 11 – Maximum voltage across one grading capacitor of a pair due to unequal sharing
under dynamic conditions (speculative worst case)

Rated voltage Maximum voltage across grading capacitor due to unequal

Ur sharing under dynamic conditions
kV (peak value)
kV Prestrike Reignition
300 367 735
362 443 887
420 514 1028
550 674 1347

It is recommended that the stress due to unequal voltage sharing under dynamic conditions must
therefore be taken into account when specifying grading capacitors for shunt reactor switching
applications.

Following prestriking or reignition, a grading capacitor is discharged from its initial voltage through
the low impedance loop closed by its interrupter. There are two ways in which the discharge might
impose a stress on the grading capacitor:

a) Firstly, the rapid collapse of voltage causes an equal and opposite voltage to appear across the
inductive elements of the grading capacitor. The distribution of this voltage is dependent on the
geometry of the capacitor elements and the connections between them and may result in
transient dielectric stresses appearing in regions not normally stressed under steady state
conditions.
b) Secondly, the capacitor is discharged through minimal impedance and the current may reach a
value of some kiloamperes. Where resistive contact is present within the grading capacitor
assembly, the current may result in local heating.
The highest discharge currents are associated with reignition where the initial voltage across the
capacitor may be greater. The ability of the capacitor to be discharged through a short circuit from
such an initial voltage must therefore be taken into account when specifying grading capacitors for
shunt reactor switching applications.

Shunt reactor switching is typically performed on a daily basis and the grading capacitors will, in
general, be exposed to the associated stresses hundreds of times per year. The capacitors may be

58
exposed to the stresses associated with reignitions during one in every few O operations or several
tens of times per year.

5.4 Conclusions
Shunt reactor switching imposes significant stresses on circuit-breaker grading capacitors. It is
necessary, therefore, to specify additional requirements and type tests for grading capacitors for
shunt reactor switching in order to ensure their strength and capability for this application.

The additional requirements and type tests are required to cover the stresses associated with the
recovery voltage, making allowance for the contribution of current chopping, and the stresses
associated with unequal voltage sharing under dynamic conditions, particularly those that might
arise during reignition. The ability of a capacitor to support continuous power frequency stressing
must be addressed.

The specified ratings and test levels should make allowance for unequal voltage sharing under
static conditions, which may cause the stress across a capacitor to be increased by a few percent
from its nominal share of the voltage across the circuit-breaker.

Account needs to be taken of the ability of a capacitor to be discharged through the near short
circuit loop closed by its associated interrupter, during prestriking and during reignition, due to the
rapid rate of change of voltage and the high amplitude currents that will result. Due to the high
resonant frequency of the short-circuited capacitor, a test using a rod-rod gap in air to discharge the
capacitor is unlikely to be sufficient to excite resonance.

It has been shown that unequal voltage sharing under dynamic conditions is likely to have a
significant effect on the electrical stresses experienced by grading capacitors. It has not been
possible to quantify the levels of voltage and current that might result since experimental data is not
presently available. There is a need for more research in this area.

5.5 References
[1] S. Berneryd et al – (CIGRE WG 13.02), Interruption of small inductive currents, CIGRE
Technical Brochure 50, December 1995.
[2] IEC 62271-110, High-voltage switchgear and controlgear – Part 110: Inductive load switching
[3] CIGRE WG A3.11, Guide for application of IEC 62271-100 and IEC 62271-1: Part 2 – Making
and breaking tests, CIGRE Technical Brochure 304, October 2006.
[4] I. Uglesic et al, Transients due to switching of 400 kV shunt reactor, Proceedings of
International Conference on Power Systems Transients, Rio de Janeiro, Paper No. 45, 2001.
[5] Z. Ma et al, “An investigation of transient overvoltage generation when switching high voltage
shunt reactors by SF 6 circuit breakers”, IEEE Transactions on Power Delivery, Vol. 13, No. 2,
April 1998.
[6] W Boeck et al, “Insulating behaviour of SF 6 with and without solid insulation in case of fast
transients”, Paper 15-07, CIGRE 1986 Session
[7] IEC 62271-1: High-voltage switchgear and controlgear – Part 1: Common specifications

59
6 Recommendations for test procedures
6.1 Introduction
As previously mentioned, no internationally accepted testing standards exist for grading capacitors.
The lack of standards has been an obvious short-coming, both as seen from the point of view of
manufacturers of grading capacitors, of circuit-breaker manufacturers, and of the end users in the
electric utility sector. The objective with this chapter is to give recommendations with regard to
standards for testing.

Considering the described service experiences discussed in Chapter 2 and the dielectric,
mechanical and environmental stresses grading capacitors are subjected to, a set of test
procedures has to be selected. The objective with the tests is to qualify a certain capacitor design,
to prove the requirements in a type test and to check the manufacturing process. Consequently,
electrical, mechanical and environmental qualification tests have to be selected for:

• Design tests: These are carried out on a sample or a full size capacitor. The results can be
extended to a range of designs by modelling and extrapolation. Moreover, design tests can be
carried out on a one-of-a-kind development stage device;
• Type tests: These carried out on a full sized capacitor manufactured in an industrial production
line. In principle, the type tests are only valid for the capacitor type tested, but in practice the
results are extended to types of similar technology, size and rating;
• Routine tests: These carried out on each capacitor before delivery and can be used as
verification test after type tests;
• Performance check: These checks are carried out after the grading capacitor has been
subjected to various types of stresses, to prove the condition.
It should be noted that the terms “design test” and “type test” are not very accurately defined. The
most important difference is that type tests should be performed on a complete, full scale unit.

The tests should take into consideration that grading capacitors not only are purchased by circuit-
breaker manufacturers. Utilities are also buying grading capacitors for replacement on their circuit-
breaker during revisions or after failures. (It is however, recommended that utilities consult the
manufacturer of the circuit-breaker before replacement is initiated.)

The tests described in this chapter are grouped into mechanical, environmental and dielectric tests.
In each of these sections common practices are reviewed first. Then recommendations of the
working group are given. The recommendations are based on current practices, assessments of
stresses occurring during service, service experience and failures, and existing circuit-breaker test
standards.

6.2 Mechanical tests

6.2.1 General
The properties to be checked are primarily the mechanical strength of the housing and the
mechanical robustness of the internal parts, such as the capacitor stack, electrical connections,
mechanical fixations etc. Stresses can originate from mechanical operation, transportation, handling
etc., as discussed in Chapter 3.

The housing can be tested mechanically in three different ways:

• Bending;
• Pressure, if necessary (depends on design);
• Vibration;
Several types of housings are used as outlined in Chapter 1, and several types of internal pressures
are used. The requirements for the housing are quite different for the different types, e.g. outdoor
air insulated grading capacitors with porcelain or composite housing, and GIS grading capacitors
installed inside a pressurized SF 6 ambient, and testing must take the application into consideration.

60
It is foreseeable that the capacitor occasionally might be subjected to its own weight when
supported at one end only during, for example, removal. Such a procedure may be in violation of
the manufacturer’s recommended procedures.

The mechanical integrity of the internal parts is typically tested by mechanical vibration /
mechanical shock tests.

The best way to qualify the overall mechanical behaviour of the grading capacitor is to test it on the
circuit-breaker as specified in the mechanical endurance test of IEC 62271-100 [1] with 2 000 or
10 000 mechanical operations. However, in many contexts this is very impractical.

6.2.2 Design and Type Tests

6.2.2.1 Bending and pressure test on housing
6.2.2.1.1 Common practice
Because of different developments or switchgear manufacturer specification, the different grading
capacitor manufacturers have introduced different test procedures, but these tests are not
technically relevant, the grading capacitor being usually fixed on both sides.

GIS capacitors with epoxy housing are typically not subjected to static tests.
6.2.2.1.2 Recommendations of the WG
The service experience survey carried out (see Chapter 2) did not disclose any failures that were
due to mechanical failure of the housing.

a) Bending test
Grading capacitors usually do not see high bending stresses since they are fixed between the
two ends. One of the highest bending stresses could be a 100 kg assembly crew worker
standing in the middle of a 2 m long grading capacitor (not a recommended practice). Such load
results in a bending moment of approximately 1 kNm. Considering this 1 kNm as bending load,
IEC 62155 [2] requires a test with 2,1 times the routinely expected loads as type test value. This
would result in a bending test with 2,2 kNm for long insulators.
b) Pressure test
IEC 62155 specifies the tests for hollow pressurised and unpressurised ceramic and glass
insulators. For pressure type test the IEC 62155 specifies the following:
1) hollow insulators intended for general use
"The insulator shall withstand a test pressure which is higher than the design pressure for 5
min without failure. The test pressure is dependent on equipment."
2) hollow insulators intended for use with permanent gas pressure
"The type test withstand pressure is dependent on equipment design. The insulator shall
withstand 4,25 times the design pressure for 5 min without failure. The design pressure shall
be determined by the equipment manufacturer."
The WG recommends to follow IEC 62155 regarding pressure tests.

6.2.2.2 Mechanical vibration test

6.2.2.2.1 Common practice
Two types of vibration tests are currently in practice.

a) A frequency sweep vibration test in which the capacitor is subjected to frequency sweeps in the
range of 8 to 80 Hz or 10 to150 Hz with 100 sweeps at 10 octave/min. The amplitude (d) is 1,6
mm in the radial direction test and 0,6 mm in axial direction test with an acceleration limited at
25g. This corresponds to accelerations (a) in the range 1,5 – 25g (a = ω 2 × d, where ω = 2πf is
the radial frequency);
b) A half cycle sinus impulse “shock” per axis. Among the shock durations being used are 5 - 7 ms
or 14 ms (corresponding to frequencies of approximately 30 - 100 Hz). The accelerations being
applied are up to 15g in radial direction and 30g (15g for GIS) in axial direction. Typically 500 to
1 000 shocks are applied in each direction.
61
The pass criteria include the performance of an extended frequency sweep to determine the
mechanical resonance frequency before and after the test. No significant changes should be seen.
A second criterion is doing tightness tests before and after the vibration tests. The third criterion is
to carry out the dielectric tests and a performance check (see 6.4.6) after the vibration test.

Another procedure that is used is performing vibration tests on a complete circuit-breaker pole, i.e.
during a mechanical endurance test.

The seismic test according to IEC 62271-300 [3] is less demanding for the grading capacitor and
only rarely applied on the grading capacitor itself. The transportation tests described in 3.3 are not
considered adequate and are not generally used.
6.2.2.2.2 WG recommendations
As described in Chapter 3, the most severe mechanical vibration stresses are those originating from
the operation of the circuit-breaker and the frequencies of interest are those below around 200 Hz.
Accelerations up to around 15g in all directions are assumed to cover the stresses that may occur
on the grading capacitors.

The mechanical impulse test using a shaking table on a complete capacitor assembly is
recommended. Impulses should be made in all three directions.

Prior to the impulse tests, the lowest mechanical resonance frequency should be determined by
performing a frequency sweep. Typical resonance frequencies are in the range of 50 - 80 Hz. Half
sine shocks should be applied with an amplitude of 15g. The duration of the half sine should be as
long as 1 cycle of the lowest mechanical resonance frequency (corresponding to a frequency of 0,5
times the resonance frequency), but not longer than 20 ms. The reason for the choice of 0,5 times
the resonance frequency is that the maximum displacement is generated between external
components and internal components when the duration is 1 cycle. If the test with the duration of 1
cycle is impractical, the test should be done with a duration that is as long as possible. The number
of impulses should be 1 000 in each direction. While this number is lower than that corresponding to
the mechanical endurance class M2, experience shows that any failure will have occurred within the
recommended number of impulses. Moreover, the test stress level is greater than the stress under
normal service conditions.

The acceptance criteria should be based on checking if the lowest mechanical resonance frequency
has changed, followed by the tightness test (see 6.3.2.2) and the electrical performance test (see
6.4.6). Visual inspection of the internal parts should also be carried out.

6.2.3 Routine test

6.2.3.1 Common practice
The routine test performed is a check of dimension of the housing. No other tests are presently
carried out.

6.2.3.2 WG recommendations
No routine tests regarding the mechanical properties are recommended by the WG.

6.2.4 Performance check

6.2.4.1 Common practice
No tests are presently being applied, just a visual inspection.

6.2.4.2 WG recommendations
No performance checks are recommended by the WG.

6.3 Environmental tests

6.3.1 General
The environmental tests comprise tests that verify the resistance against corrosion and tightness
tests.

62
Corrosion tests for AIS grading capacitors should consider the usual environmental stresses that all
outdoor equipment is subjected to. Grading capacitors in GIS have to withstand SF 6 by-products.

Tightness tests should take account of oil and gas tightness both when the external pressure is
higher than the internal pressure of the grading capacitor, and vice versa. See the description in
Chapter 1 of the various pressure systems being used.

As described in Chapter 2 oil leaks were by far the most common problem. However, most of the
leaks were found on rather old capacitors. It appears that the improvements with regard to corrosion
protection, sealing designs and quality assurance have made this a lesser problem on newer
designs.

6.3.2 Design tests

6.3.2.1 Corrosion tests
6.3.2.1.1 Common practice
One circuit-breaker manufacturer specifies salt fog tests and Sulfur Dioxide (SO 2 ) tests for grading
capacitors used in AIS. The test is made on representative samples which are put into climatic
chamber.

• Typically the salt fog test consists of 500 hours in a NaCl (50 g/l) atmosphere at 35 °C
(IEC 60068-2-11 [4]);
• The SO 2 test may comprise 20 cycles each with 8 h at 40 °C under normal humidity with SO 2 for
16 h at 23 °C and with relative humidity of 75 % (IEC 60068-2-42 [5]).
For grading capacitors to be installed in GIS no corrosion test is done.

Generally speaking, unless some materials or combinations of materials that have not been used
previously in such application, no particular corrosion tests are performed.

The acceptance criteria are that the samples pass the tightness test after the corrosion test and that
the contact surfaces between flanges and covers have no significant traces of corrosion.

6.3.2.1.2 WG recommendations
For grading capacitors used in AIS it is recommended to perform a corrosion tests according to IEC
60068-2-11 or IEC 60068-2-52 [6] and IEC 60068-2-42. It is however, recognized that it may be
difficult to find a laboratory that is willing (or still has an installation) to perform the test with SO 2 .

The acceptance criteria are that the samples pass the tightness test after the corrosion test and that
the contact surfaces between flanges and covers have no significant traces of corrosion.

When introducing new materials for GIS grading capacitors their ability to withstand SF 6 by-
products have to be verified; No standard test is available, the one way to check the compatibility
against SF 6 by-products is to store a test sample in an atmosphere of used gas including a certain
amount of SF 6 by-products for a suitable long time.

The stressed materials have to be qualified after that by visual inspection, measurement of surface
resistivity or similar measures.

6.3.2.2 Tightness test

6.3.2.2.1 Common practice
Tightness is normally checked by subjecting a grading capacitor or a smaller (shorter) model with
the same flange design to elevated temperatures and/or by thermal cycling, with visual leakage
control during the thermal cycling, the routine leakage test is applied before and after the thermal
cycling. The temperature range for the cycles is the same as the operating temperature range,
typically from -40 °C up to 90 °C for GIS and from -55 °C up to 75 °C for grading capacitors used in
AIS. The number of cycles is typically around 10.

63
For oil filled capacitors leakage tests are done by visual inspection, i.e. looking for oil coming out.
For SF 6 impregnated capacitors, SF 6 or Helium (He) leakage detection systems are typically used.
When SF 6 leakage detection is performed, the filled capacitor is placed in a plastic bag and the SF 6
concentration is checked at the beginning and after a suitable time and the leakage range is
calculated. When He leakage detection is applied He is sprayed around the capacitor seals while
the capacitor is evacuated and a He detector is connected to the vacuum pump.

6.3.2.2.2 WG recommendations
The cycle test with min 10 cycles at minimum temperature (10 h) – ambient temperature (2h) -
maximum temperature (10 h) according to the graph in Figure 40 should be applied.

Maximum
temperature

Ambient Time
temperature

Minimum
temperature

Figure 40 – Temperature cycle test

The acceptance criteria are that the sample passes the tightness test according routine test
paragraph hereunder after this program.

6.3.2.3 Gas tightness test for oil impregnated capacitors used in GIS circuit-breakers
6.3.2.3.1 Common practice
One circuit-breaker manufacturer specifies to check tightness by installing a grading capacitor in an
SF 6 environment at elevated temperature and pressure (maximum values of the rated ranges),
typically at 90 °C / 0,9 MPa abs (based on a filling pressure of 0,7 MPa abs at 20 °C). After 40 days,
a dissolved gas analysis is performed.

The acceptance criterion is that the maximum SF 6 content of the oil shall not exceed a level
corresponding to a leakage rate of 10 -6 cm 3 /s at 100 kPa for 1 dm 3 oil.

This test is considered as design test, and it is made on one capacitor which will cover the complete
range of grading capacitors.

6.3.2.3.2 WG recommendations
Apply, as design test, the common practice to all oil impregnated capacitors for GIS application.

6.3.2.4 Type test

6.3.2.4.1 Common practice
As the sealing system is tested in full size during design testing, no type test is necessary.

6.3.2.4.2 WG recommendations
The WG does not recommend any environmental test as type test.

6.3.2.5 Routine test

6.3.2.5.1 Common practice
The tightness of oil filled capacitors is routine tested with a pressure of the liquid above the
operating pressure, depending on the type of the expansion device for the capacitor units for 24 h.
A visual check with powder or paper is made when the capacitor is removed from the oven and
64
repeated after 24 or 48 hours at ambient temperature and atmospheric pressure. A Helium test
before the impregnation has also proved very good results.

For capacitors used in GIS, both vacuum (1 Pa) and air pressure (0,8 MPa) is applied during the
routine test at the highest temperature of its application range, typically 90 °C. The reason for this
test is that capacitor is subjected to vacuum (by the circuit-breaker manufacturer before the SF 6
filling) and SF 6 pressure in service.

The acceptance criteria for Helium test is a leakage rate of 10 -4 Pa×l/s at ambient temperature. For
the pressure test, when the grading capacitor is removed from the oven an absorbent is applied to
the flanges. No trace of oil is allowed on the absorbent immediately after the grading capacitor has
been removed from the oven or 24 or 48 hours after.

For SF 6 impregnated capacitors SF 6 or He leakage detection systems are typically used. When SF 6
leakage detection is performed, the filled capacitor is placed in a plastic bag and the SF 6
concentration is checked at the beginning and after a suitable time and the leakage range is
calculated. When He leakage detection is applied He is sprayed around the capacitor seals while
the capacitor is evacuated and a He detector is connected to the vacuum pump.

6.3.2.5.2 WG recommendations
The WG recommends applying the common practice.

6.3.2.6 Performance check

6.3.2.6.1 Common practice
The repetition of the routine test is made currently to check the tightness of the capacitor.

6.3.2.6.2 WG recommendations
The WG recommends applying the common practice.

6.4 Electric tests

6.4.1 General
The aim of the electric tests is mainly to cover the dielectric requirements of the active parts of the
capacitor, i.e. the internal performance. The external insulation is normally not the subject of
qualification, because the circuit-breaker design determines the external behavior and the capacitor
is only an integral part of the circuit-breaker.

6.4.2 Nominal voltage of grading capacitors

6.4.2.1 Current practice
As discussed in Chapter 4, the nominal voltages (U N ) for grading capacitors are varying in a wide
range without any definition and no clear relation to the standardized highest voltages for
equipment U m . The common practice of circuit-breaker manufacturers is to specify a grading factor
between 1,05 to 1,3 (e.g. a voltage of 170 – 210 kV for a two break 550 kV circuit-breaker),
sometimes up to 1,75 (280 kV) but with different relations to the test voltages.

6.4.2.2 WG recommendation
The value of U N is the maximum voltage the capacitor should be capable of withstanding during its
entire lifetime. Based on the considerations given in Chapter 4, the WG recommends, a nominal
voltage of 20 % above the linear voltage distribution for the line-to-ground voltage across the
circuit-breaker.

U N = 1,2 × (U r / 3 ) / N

with N = the number of series connected breaking units per phase and U r = rated voltage of the
circuit-breaker. The factor 1,2 covers most unbalanced voltage distributions and is a simple and
general approach of handling the large number of possible cases.

65
It should be noted that the grading factor 1,2 does not necessarily apply to the test voltages. The
reason for this is that the test voltages are derived from the requirements given in IEC 62271-1
applying a voltage grading factor of 1,2. The calculated values are rounded to the nearest
standardized voltage for the particular waveshape.

Table 12 - Rated voltages of circuit-breaker vs. nominal voltages of grading capacitors

Circuit-breaker Grading capacitor

Rated voltage Phase-to-earth voltage Circuit-breaker with 2 Circuit-breaker with 4
breaking units/pole breaking units/pole
Ur U r /√3 UN UN
kV kV kV kV
300 173 105 50
362 209 125 65
420 242 145 75
550 318 190 95
800 462 280 140

6.4.3 Design tests

6.4.3.1 Ageing test – Life time test
The ageing of the dielectric of the capacitor is following Arrhenius’ rule. By means of an accelerated
ageing test using increased voltage, temperature and/or frequency manufacturers determine the
lifetime of a sample of the grading capacitor.

6.4.3.1.1 Common practice

Common practice is an accelerated ageing test as described above.

6.4.3.1.2 WG recommendation
A sample capacitor should be tested at 2,3×U N during 500 hours and 2,0×U N during 10 000 hours in
an oven such that the dielectric is maintained at a temperature of 90 °C (maximum ambient
temperature including self-heating of the capacitor).

The 500 and 10 000 hours test represent the life time of 100 000 to 1 000 000 hours, respectively at
nominal voltage. This rule applies to capacitors based on “mixed dielectric” and “polypropylene film”
technology.

6.4.3.2 Discharge test

This test covers the daily duty of the circuit-breaker when switching capacitive loads. The switching
of disconnectors near to the circuit-breaker is also covered by this test.

6.4.3.2.1 Common practice

Some manufacturers perform discharge tests on capacitor samples during development. When such
a test is performed, common practice is to charge the capacitor and discharge it through an air gap
a large number of times. Such a test would demonstrate the long term reliability of the dielectric
material and internal connections of the capacitor.

6.4.3.2.2 WG recommendation
For the verification of the reliability of the dielectric material and internal connections, the WG
recommends discharge testing using for example the following procedure. A capacitor is charged
with a DC voltage and it is discharged through a air gap. 10 000 discharges are applied on a
sample at U DC = 1,5 × U N √2 at an interval of 3 s. It is sufficient to perform the test with one polarity
only. The inductance in the discharge loop should be as small as possible. The factor 1,5 is
considered to also cover the application of a line circuit-breaker closing on a trapped charge, as the
probability that the voltage during closing is at its maximum (2 p.u.) is considered low. The routine
test should be repeated after the discharge test and there should be no change in capacitance.
66
For this test a sample capacitor can be used to verify the design of a capacitor type. Figure 41
shows a suggested test set-up.

DC supply

Resistive
voltage
divider

Figure 41 - Discharge test set-up

6.4.4 Type tests

6.4.4.1 Withstand voltage tests
6.4.4.1.1 Common practice
The withstand voltages for the grading capacitors vary in a wide range for the same or similar
applications. It depends on the requirements of the purchaser who, in most cases is the circuit-
breaker manufacturer, but who also can be an end user. Table 13 shows an example of the range in
requirements for grading capacitors for 550 kV circuit-breakers having two breaking units. The data
have been derived from data obtained from capacitor manufacturers.

Table 13 - Range of requirements for grading capacitors for 550 kV circuit-breakers with two
breaking units

Type of voltage Levels

Lightning impulse 970 kV -1500 kV
Chopped wave 1350 kV – 2000 kV
Switching impulse 700 kV – 850 kV
Short duration power frequency up to 5 min 460 kV – 720 kV

6.4.4.1.2 WG recommendation
The choice of withstand voltages for a grading capacitor is based on the considerations given in
Chapter 4. An example of selected values for 550 kV is given in Table 14. The calculated values
have been rounded to the standardized values given in IEC 60071-1.

Table 14 - Recommended withstand test voltages for grading capacitors for 550 kV circuit-
breakers with two breaking units

Type of voltage Withstand voltage test levels for grading capacitors

kV
General purpose Special purpose
Lightning impulse 1050 1175, 1300, (1425)
Chopped wave 1300 1425, 1550
Switching impulse 750 850, 950, (1050)
Short duration power frequency 460 510, 570, (630)

NOTE The values in brackets are to be considered only for exceptionally high requirements across the

67
 open circuit-breaker

The lowest values cover the dielectric withstand requirements of the open circuit-breaker given in
IEC 62271-1. The highest values have been derived from the shunt reactor switching stresses
corresponds to the highest requirements. Values in between have to be chosen according to the
special requirements of each application.

For the impulse tests 15 impulses with both polarities should be applied. For the power frequency
type test a voltage application of 5 minutes is recommended.

6.4.4.2 Flashover test

During dielectric testing of circuit-breakers flashovers cannot be avoided. With flashovers the
stressing of these sudden discharge or charge reversals of the capacitors shall be simulated here.
Also the current carrying capability of switching operation discharge processes in service are
covered with this. Furthermore, a flashover along the outside of the capacitor should not harm the
porcelain or composite housing.

6.4.4.2.1 Common practice

Some manufacturers perform a special test comprising 20 flashovers with a 50 Hz test voltage prior
to the withstand voltage tests to check the insulation coordination of the capacitor. After these 20
flashovers and the withstand test the capacitor has to be partial discharge (PD) free, and no
capacitance change and no perforation is permitted.

6.4.4.2.2 WG recommendation
The common practice should be followed, i.e. a flashover test on the complete grading capacitor,
but not necessarily with power frequency voltage, impulse voltage is also suitable.

6.4.4.3 Long duration power-frequency voltage withstand test

Concerning the power-frequency application in service two situations have to be considered. First
for the circuit-breaker in open position (with one terminal at service voltage and the other one
earthed) the line-to-ground voltage is stressing the open circuit-breaker. The second situation is the
out-of-phase condition which periodically appears with 2 times the line-to-ground voltage across the
open circuit-breaker. The out-of-phase condition during an automatic synchronising operation in the
grid normally last some seconds or minutes only, but special requirements based on operation
practice or other ideas lead to prolonged time requirements in some specifications of the end user
of the circuit-breaker. Requirements of out-of-phase durations up to the whole lifetime of the circuit-
breaker are known, but are considered as not realistic service application.

6.4.4.3.1 Common practice

Common practice is to verify the increased service voltage by the short duration power-frequency
voltage test (1 or 5 minutes) or by application of defined voltage values for long time periods (8 or
24 hours) covering the out-of-phase conditions.

6.4.4.3.2 WG recommendation
The aim during the type test is to prove the thermal stability of the complete grading capacitor.

To cover out-of-phase condition a long time power-frequency voltage withstand test is

recommended with 2 times the nominal voltage of the grading capacitor (2,0 × U N ) in a type test.
The time duration has to be chosen according the requirements. Eight hours is the recommended
test duration. Special applications may lead to longer testing time requirements.

Special attention has to be paid to “old” capacitors of the “all paper” design with high tan δ values.
These capacitors tend to have increased self-heating due to the relatively high losses.

68
6.4.4.4 Resonance frequency measurement
6.4.4.4.1 WG recommendation
The resonance frequency has to be higher than 700 kHz, so that the capacitor is still effective with
a front time up to 0,5 μs (500 kHz - transient voltage).

The measurement can be made either using analysis of the discharge current, or a measurement of
the high frequency capacitance as given in IEC 60358 [8], the measured values of the capacitance
between the line and low-voltage terminals shall not deviate by more than –20 % to +50 % from the
rated capacitance.

When measuring the resonant frequency of the grading capacitor, it is essential to keep the stray
inductance of the connecting leads to a minimum in order not to influence the resonant frequency
significantly. In the test arrangement of Figure 42 a number of parallel leads have been distributed
around the circumference of the grading capacitor and arranged as close to its axis as possible.

Key
HT Line terminal
LT Low voltage terminal

Figure 42 - Wiring diagram of the measuring circuit for the high frequency capacitance of a
capacitor

6.4.5 Routine test

6.4.5.1 Short duration power-frequency withstand voltage test
This test should eliminate every risk of breakdown coming in the beginning of the lifetime of a
capacitor.

6.4.5.1.1 Common practice

The voltage is specified at values between 1,25 and 1,8 × U 50Hz-CB /N. The duration is 60 seconds,
historically the capacitors are tested during 72 s at 50 Hz so that the capacitor sees the same
number of oscillations as 60 s at 60 Hz (which is not specified in IEC 60060 [9]).

Two very often used test values are either 1,35 or 1,6 × U 50Hz-CB /N.

6.4.5.1.2 WG recommendation
For routine tests the same power frequency voltage test values can be chosen as for the type tests
With this the recommended short-duration power frequency withstand test voltages for grading

69
capacitors intended to be used on 550 kV circuit-breakers with two breaking units are 460 kV, 510
kV, 570 kV and 630 kV.

The capacitor is considered having passed the test when the subsequent measurement of
capacitance, tan δ and PD measurements do not deviate significantly from those measured before
the test.

6.4.5.2 Capacitance measurement

The measurement of the capacitance is to check that its value lies in the specified tolerance. The
tolerance of the capacitance is depending on the design of the housing and the capacitor stack..

6.4.5.2.1 Common practice

The used values of tolerance for AIS capacitors are in the range of -+06 % up to ±5 % and for GIS
capacitors ±2 % up to ±5 %.

The measurement of capacitance is normally specified at a nominal voltage. The capacitor

manufacturers propose to measure first at 10 % of the nominal voltage to check that no breakdown
occurs or has occurred during the first application of voltage.

6.4.5.2.2 WG recommendations
The measurement of capacitance should be done at least at the voltage levels 0,1 × U N and
1,0 × U N and again at 1,0 × U N after to the power-frequency test.

The recommended maximum tolerances are for

- AIS grading capacitors ±5 %

- GIS grading capacitors ±3 %.
6.4.5.3 Loss factor measurement (tan δ )
The loss factor measurement will confirm a good impregnation and that all contacts are correct. The
value of tan δ is dependent on the used technology of the capacitor elements.

The values of tan δ measured are depending on the capacitor technology and are as follows (at
nominal voltage):

- Oil impregnated paper: tan δ < 50 × 10 -4

- Oil impregnated mixed dielectric: tan δ < 30 × 10 -4
- Oil impregnated Polypropylene: tan δ < 10 × 10 -4 (see IEC 60044 [10])
6.4.5.3.1 Common practice
The tan δ measurement is performed at nominal voltage by the capacitor manufacturers.

Some utilities perform a tan δ measurement on site at 10 kV. The results of such a test should be
approached with caution, as the measurement is not repeatable. This is due to the low voltage
stress (an accepted capacitor at the exit of the factory can be refused 2 months later due to the
“Garton effect”, see Chapter 2).

6.4.5.3.2 WG recommendations
The loss factor should be measured at a voltage level of at least 1,0 × U N .

The capacitance as well as tan δ , measured and documented during the routine test, are important
values to check the integrity of the grading capacitor after any kind of stressing by repeating these
measurements.

6.4.5.4 Partial discharge measurement

The Partial Discharge (PD) measurement will detect any flaw in a capacitor element (like crack in
the aluminium foil, bad contact between 2 elements, conductive parts in the capacitor, etc.). The
measurement is specified in 60270 [11].
70
For PD measurement of grading capacitors the following matters have to be considered:

a) The quality check of the capacitor itself after the manufacturing process or after a certain
stressing to check the capacitor performance;
b) The PD test procedure of the complete GIS circuit-breaker or of GIS compartments including the
circuit-breaker. There the power frequency voltage test has to be performed before the PD test
as a pre-stress or as withstand test. During this time PD monitoring can be done for failure
detection in the tested GIS section. A similar situation is the on-site testing including PD
monitoring. No PD noise from anywhere is allowed during this time. Therefore concerning the
grading capacitors the extinction voltage is generally of high importance.
c) For AIS circuit-breakers there should be no PD which could disturb the required radio
interference voltage test.
6.4.5.4.1 Common practice
Sometimes the inception and extinction voltage are specified, but usually the capacitor
manufacturers measure the PD at a fixed voltage value before and after the power frequency
voltage test.

The voltage level for PD measurement is varying between one and two times the linear fraction of
line-to-ground voltage of the circuit-breaker. The permissible maximum PD values are between 3 pC
and 10 pC for GIS and AIS grading capacitors.

For comparison the standard maximum permissible partial discharge level for GIS is 5 pC at
1,2 × U m /√3 after pre-stress with the power frequency withstand voltage.

6.4.5.4.2 WG recommendation
The proposed test voltage for partial discharge measurement for GIS-capacitors should be 2,0 × U N
with a maximum permissible level of 3 pC and recording of the inception and extinction voltages.
For AIS capacitors 1,5 × U N is considered sufficient.

Special for type tests only: Additionally, it may be useful to know the behaviour of the capacitor
when exposed to different voltages. Therefore, the PD-values should be recorded during the voltage
up and down for inception/extinction to check the behaviour of the capacitor when the circuit-
breaker test is exceeding the 2 p.u. level.

A further aspect is that for GIS the PD test is required to be performed after all withstand voltage
tests (impulse and power-frequency). Therefore, this test should be done directly after a series of
dielectric tests (important for type tests). For example if switching impulse, lightning impulse and
chopped wave tests have been performed in one day, than the PD test should be done at the end
on the same day, but only once at the end, not between each impulse test.

6.4.6 Performance check

These measurements give the possibility to check the capacitor after any kind of stressing and
provide acceptance criteria or control criteria concerning the performance of the grading capacitor
by comparison with the first measurements (normally routine test).

Three measurements are recommended (at nominal voltage):

- capacitance 2;
- tan δ ;
- partial discharge.
If possible a power-frequency voltage test can be added. Generally a maximum voltage of 80 % of
that of the routine test voltage is applied.

2 Generally, the capacitance measured at 10 kV has a good repeatability, in contradiction to the tan δ measurement at 10
kV.

71
6.5 Conclusions
Based on assessments of the operating environment as well as the utility service experience with
voltage grading capacitors, the following recommendations with regard to test procedures are given.

6.5.1 Environmental tests

AIS grading capacitors should be subjected to the same corrosion tests as specified by the IEC for
other outdoor components. Similarly, GIS grading capacitors should be tested for their ability to
withstand SF 6 by-products in the same way as other GIS components are.

To check the tightness of the capacitor housing the design test should include thermal cycling
across the entire specified service temperature range. For routine tests grading capacitors should at
its maximum operating temperature be pressurized to a level above the maximum operating
pressure. GIS grading capacitors should also be tested in vacuum. In all cases the results should
be qualified with leakage tests.

6.5.2 Mechanical tests

Mechanical design and type testing of capacitors should be carried out using a shaking table and
subject the capacitor to half sine mechanical shocks with a duration corresponding to a frequency
lower than the mechanical resonance frequency, as determined by a frequency sweep recording.

In addition, to verify the mechanical strength and integrity of the capacitor housing, it is
recommended to carry out bending tests and pressure tests according to the IEC standard for
hollow ceramic or glass insulators.

6.5.3 Electrical tests

The dielectric tests level and procedures should be based on those applied for circuit-breakers, but
by taking the voltage sharing between the grading capacitors into account. The voltage sharing
differs substantially for different circuit-breaker designs and applications as it is determined by the
values of the capacitors, the tolerances and stray capacitances.

Grading capacitors used on circuit-breakers applied for shunt reactor switching deserve special
consideration. Increased dielectric requirements are recommended.

6.6 References
[1] IEC 62271-100: High-voltage switchgear and controlgear – Part 100: Alternating current
circuit-breakers
[2] IEC 62155: Hollow pressurized and unpressurized ceramic and glass insulators for use in
electrical equipment with rated voltages greater than 1 000 V
[3] IEC 62271-300: High-voltage switchgear and controlgear – Part 300: Seismic qualification of
alternating current circuit-breakers
[4] IEC 60068-2-11: Environmental testing - Part 2: Tests. Test Ka: Salt mist
[5] IEC 60068-2-42: Environmental testing - Part 2-42: Tests - Test Kc: Sulphur dioxide test for
contacts and connections
[6] IEC 60068-2-52: Environmental testing - Part 2: Tests - Test Kb: Salt mist, cyclic (sodium,
chloride solution)
[7] IEC 62271-1: High-voltage switchgear and controlgear – Part 1: Common specifications
[8] IEC 60358: Coupling capacitors and capacitor dividers
[9] IEC 60060: High-voltage test techniques (all parts)
[10] IEC 60044: Instrument transformers (all parts)
[11] IEC 60270: High-voltage test techniques - Partial discharge measurements

72