Professional Documents
Culture Documents
Product Guide 2015E
Product Guide 2015E
Product Guide 2015E
ABB
High Voltage Products
Surge Arresters
Ludvika, Sweden
2015 Edition
TABLE OF CONTENTS Page
1. Introduction 3
2. Definitions 4
3. Overvoltages 11
8. Arrester Selection 70
9. Installation Guidelines 85
__________________________________________________________________________________________
All electrical systems and equipment are subjected to electrical stresses caused by higher than
normal voltages many times during their lifetime. Such overvoltages are caused by atmospheric
disturbances (lightning), switching phenomena as well as system disturbances, and these cannot
be completely avoided.
It is vital that the electrical equipment operates fault-free during such abnormal conditions.
However, for economic reasons, it is not possible to insulate the electrical equipment with a
sufficiently high withstand level to survive all these overvoltages, particularly those resulting from
lightning or switching surges. Consequently, these pose a very real danger for causing failure of
the electrical equipment. An economical and safe on-line network therefore requires extensive
protection against unacceptable overvoltage loads.
Overvoltage protection is not new, and has been used in one form or another for well over 100
years. Today, overvoltage protection can basically be achieved in two ways (sometimes in
combination):
• Avoid or limit the overvoltages at the point of origin. For example, through the use of
overhead shield earth wires and lower tower footing resistance as countermeasures
against atmospheric overvoltages and pre-insertion resistors and/or controlled switching
against switching overvoltages.
• Limit overvoltages near the electrical equipment with surge arresters
In isolation, shield earth wires and pre-insertion resistors offer a degree of protection. However,
by their nature, surge arresters provide the primary protection against different types of
overvoltages (atmospheric and switching). They are generally connected between each phase
and ground, in parallel with the equipment to be protected and function to divert the surge current
safely to earth; thereby limiting the overvoltage seen by the protected object.
Insulation co-ordination is the art and science of choosing the right insulation strength of electrical
equipment taking into account normal and abnormal service conditions as well as the
characteristics and location of suitable surge arresters. Despite being a well-established
technology, there remains a degree of mysticism about the design, selection and application of
surge arresters in electrical networks. This is not made easier through the continual improvement
and development of the active elements by leading manufacturers as well as the designs and
housing material, ultimately leading to new applications for surge arresters. Resultant ongoing
changes to the Standards just add to the difficulty in understanding for the user.
This guide is intended to clear away some of this mystification, and guide the reader to a better
understanding of how to select and use modern day surge arresters. It is principally limited to the
common application of the protection of transformer insulation between phase and ground in
outdoor air-insulated substations (AIS) with system voltages up to 800kV. Other applications are
briefly discussed, but are, for the most part, considered beyond the scope of this Guide. Instead,
the reader is referred on to additional technical literature which covers the topics in more detail.
In addition, International Standard IEC 60099-5 “Surge arresters - Selection and application
recommendations” is recommended reading.
Finally, the reader is referred to the ABB surge arresters “Arresters Online” web page
(www.abb.com/arrestersonline) for continually updated information on surge arresters.
__________________________________________________________________________________________
To permit the reader to understand the basis for the selection and application of surge arresters,
it is important to make a brief review of some of the common terminology used throughout this
Guide.
The surge arrester standards referred to herein are the prevailing editions of:
Arrester classification
Within IEC 60099-4, station and distribution class arresters are further designated as "high",
"medium" and "low" duty to indicate their specific performance requirements.
Backflashover
Occurs when lightning strikes the transmission line tower structure or overhead shield wire. The
lightning discharge current, flowing through the tower and tower footing impedance, produces
potential differences across the line insulation. If the line insulation strength is exceeded,
flashover occurs, i.e. a backflashover. Backflashover is most prevalent when tower footing
impedance is high.
Charge
Charge is calculated as the absolute value of current integrated over time. For the purpose of the
standard this is the charge that is accumulated in a single event or group of surges lasting for not
more than 2 s and which may be followed by a subsequent event at a time interval not shorter
than 60 s.
__________________________________________________________________________________________
IEC (Uc)
IEC gives the manufacturer the freedom to decide Uc. The value is
verified in the operating duty test. Any uneven voltage distribution in the
arrester shall be accounted for.
IEEE (MCOV)
IEEE lists the maximum continuous operating voltage (MCOV) for all
arrester ratings used in a table. The value is used in all tests specified by IEEE.
Earthing
Energy capability
The energy capability of a surge arrester is different depending on the type, duration and grouping
of applied impulses as well as what occurs afterwards. Arrester standards have historically not
explicitly defined the energy capability of an arrester, and the current editions have specifically
focused on attempting to resolve this deficiency in the following forms (IEC 60099-4 definitions):
__________________________________________________________________________________________
Follow current
The current from the connected power source which flows through an arrester with series gaps
following the passage of discharge current.
Hydrophobicity Classification
The superior electrical performance of composite
insulators and coated insulators stems from the
hydrophobicity (water-repellency) of their surfaces. The
hydrophobicity will change with time due to exposure to
the outdoor environment and partial discharges (corona).
__________________________________________________________________________________________
Maintainability prediction
Mean-time-between-failure (MTBF) is the average time between failures, typically measured in
hours. MTBF is a statistical value and is meant to be the mean over a long period of time and
large number of units. In practice, MTBF is only relevant with reference to repairable items, since
calculations of MTBF assume that a system is "renewed", i.e. fixed, after each failure and then
returned to service immediately after failure. Mean-time-to-failure (MTTF) is the average time
expected to the first failure of a piece of equipment and should be used instead of MTBF in cases
where a non-repairable item is replaced after a failure. However, MTBF is commonly used for
both repairable and non-repairable items. Mean Time to Repair (MTTR) is the total amount of
time spent performing all corrective maintenance repairs divided by the total number of those
repairs. A common misconception about MTBF is that it specifies the time (on average) when
the likelihood of failure equals the likelihood of not having a failure. This is only true for certain
symmetric distributions. For typical distributions with some variance and uncertainty, MTBF only
represents a statistical value and hence is not suitable for predicting detailed time of failure.
__________________________________________________________________________________________
Protective characteristic
The combination of the arrester’s residual voltages for different current impulses. For good
protection, the arrester characteristic should lie well below the equipment insulation withstand
characteristic at all points.
Lightning impulse withstand level (LIWV or BIL) is the equipment’s insulation withstand
level against lightning impulses
Switching impulse withstand level (SIWV or BSL) is the equipment’s insulation withstand
level against switching impulses
Switching impulse protection level (SIPL or Ups) of the arrester is the residual voltage
for a specified switching impulse current.
Parameters Parameters
of the of the
system surge
arrester
Voltage
BIL / BSL
LIWL // SIWV
SIWL Protective margin
LIWV
Protection level
LIPL / SIPL
Upl / Ups
TOV
TOV capability
COV
Protective margin
The protective ratio minus 1 and expressed as a percentage. As an absolute minimum, the margin
should cover the voltage increase due to the connections between the arrester and the protected
equipment as well as the increase in the residual voltage due to the discharge amplitude and
front-time being different from the nominal discharge current of the arrester.
__________________________________________________________________________________________
Response spectra
A response spectrum is a plot of maximum response (displacement, velocity, or acceleration)
versus a system characteristic (frequency or period and damping ratio) for a single degree-of-
freedom oscillator for a particular applied load, such as an earthquake acceleration time history.
Shielding
Protection of phase conductors from direct lightning strokes; generally by means of additional
conductor(s) running on the top of the towers and grounded through the tower structures to earth.
Stations can also be shielded by earth wires or lightning masts.
Shielding failure
Occurs when lightning strikes a phase conductor of a line protected by overhead shield wires.
__________________________________________________________________________________________
For an overhead line, v = 300 m/µs and the surge impedance of a single conductor varies in a
narrow band between about 400 and 500 ohms. The surge impedance of an oil-paper cable
varies from about 30 to 60 ohms and the velocity of propagation is taken to be around 150 m/µs.
In many cases, an equivalent surge impedance or a combined surge impedance of two or more
conductors is desired, leading to the need for more complicated calculations. Furthermore, the
simplified analysis does not take into consideration line geometry, coupling factor (whereby a
travelling wave voltage and current are impressed on only one conductor, leading to a voltage
being induced or coupled to the other conductor), points of discontinuity nor corona effects. More
detailed guidance is given in Cigré 63 and IEC 60071-2.
Travelling wave
Occurs when lightning strikes a transmission line span and a high current surge is injected onto
the struck conductor. The impulse voltage and current waves divide and propagate in both
directions from the stroke terminal at a velocity of approximately 300 m/µs with magnitudes
determined by the stroke current and line surge impedance.
__________________________________________________________________________________________
When lightning strikes a transmission line, the line itself is usually not damaged but the
overvoltage generated may result in flashovers of the line insulators and can also cause insulation
breakdowns in apparatus in sub-stations connected to the line. A lightning impulse has a very
short front time, microseconds (µs), and the voltage on the transmission line can rapidly increase
to several thousands of kilovolts when lightning strikes the line. If the earthing impedances of the
towers are not sufficiently low, a lightning stroke to the tower or to the overhead shield wires (if
any) might cause a so called “backflashover” across the insulator strings to the phase conductors.
Travelling waves are generated at the location where the lightning hits the line and these waves
propagate along the line. The insulation is stressed further if the travelling waves reach an open
end of the line where they are reflected; causing a doubling of the voltage.
It has been acknowledged that many lightning strikes are not a single stroke, but instead consist
of a series of consecutive strokes of varying magnitude and intervals. The incidence of multi-
stroke flashes naturally varies considerably by region and season. In some cases, about
70 - 80 percent of lightning strikes may consist of multiple strokes; 3 - 4 pulses on average, with
an interval between each pulse of less than 50 milliseconds. Distribution lines are generally
unshielded, and hence bear the full force of direct strikes. Consequently, multipulse performance
is a decisive measure of survival ability and reliability of distribution arresters near direct strikes.
__________________________________________________________________________________________
Induced surges are lower in magnitude and the front of the wave is usually substantially longer
than for direct strokes. The front steepness is therefore seldom decisive from a protection point
of view. Further, since lightning flashes to ground rarely exceed 200 kV, their influence is typically
neglected on HV transmission lines due to the relatively high LIWV on these lines.
However, despite the fact that discharge currents are mostly lower than for direct strokes, induced
strokes can nevertheless be decisive for arrester duty requirements. This is especially true for
low voltage systems in areas where the lightning intensity is high. Due to a substantially larger
collection area compared to direct strokes, the number of arrester operations per year can be
substantial. For distribution and low voltage systems it is often the induced overvoltages which
cause the most damage to unprotected equipment.
Due to the common insulation practice with relatively low insulation levels for higher system
voltages, switching overvoltages will normally only be of interest for system voltages above
245 kV.
Switching overvoltages occur in connection with all kinds of switching operations in a network.
The waveshape can be of practically any form, with the fundamental frequency normally in the
order of some hundred to some thousand Hertz.
Large overvoltages can occur in connection with switching operations, particularly with the
following types of loads:
• Interruption of short circuits
• Disconnection of unloaded transformers and shunt reactors
• Switching of long unloaded lines
The switching overvoltages are usually defined in terms of per-unit of the system voltage. The
overvoltage factor is defined as the ratio of the peak value of the overvoltage to the peak value of
the maximum phase-earth voltage. In EHV networks, for example, it is desirable for cost reasons
to reduce the insulation level as much as possible, and thus overvoltages higher than 2.5 p.u. are
usually not accepted.
__________________________________________________________________________________________
These overvoltages can normally be kept to acceptable levels with the help of a high short-circuit
power in the supply network, line compensation with shunt reactors, suitable generator control,
automatic fault clearing, etc. Hence, this type of overvoltage is normally not of concern for the
system equipment itself (although at system voltages of 550 kV and above it may become
significant).
Fig. 3
Um x √2 Classification of
p.u 1 p.u = overvoltages showing
√3 duration and amplitudes of
6 Lightning over-voltages
stress on insulation in
HV networks
5
4 Switching over-voltages 1 p.u = Us x √2
3 √3
Temporary over-voltages
2 System voltage
1
t
10-6 10 -4 10-2 10 0 102 104
In some cases, earthed crossarms or spark gaps have been used close to substations in an
attempt to limit the amplitude of incoming lightning overvoltages. However, such measures tend
to increase the likelihood of flashovers near the station with the consequent generation of fast-
front surges. Special attention should be given to tower earthing near the station to lower the
probability of back flashovers at this location.
__________________________________________________________________________________________
Surge arresters are not normally required to protect against temporary overvoltages (although
special cases exist), but they must survive them. TOV’s can thus be decisive in selection of the
rated voltage for the arresters.
U Ucable = 20% of U
Ureflect = 80% of U
For the purpose of analysis of cables connected to a transformer, the surge impedance of the
transformer is considered as infinite, so that total reflection occurs at the transformer terminals.
To further simplify the matter, the incoming wave is assumed to be rectangular. The result is a
step-wise increase in the voltage, up to potentially double the initial incident voltage, U. In
contrast, for a cable connected between two overhead lines, the voltage reflected in both ends of
the cable will increase to the value of the incoming voltage U.
In practice, the wave front is not rectangular, but has a rate of rise. This rate of rise will change
the way the cable is charged, but not the value. Generally a cable can be considered as a
concentrated capacitance, which will be able to bring down the steepness of the incoming wave,
but if the wave energy is sufficiently high, the cable does not limit the voltage rise.
__________________________________________________________________________________________
For transformers which are manufactured with internally connected cable boxes, the cable is
connected directly to the transformer terminal, without any possibility of connecting arresters at
this point. Therefore arresters can only be installed at the far end of the cable. A certain higher
degree of associated risk for insulation failure may therefore have to be accepted.
In order to achieve the greatest possible protection zone and ensure a sufficient protective margin
between the cable/transformer LIWV and the voltage that may occur, special attention should be
paid to the following for cables operating at 72.5kV and above:
The incoming overhead line connected to the cable should have a zone of about 2km out
from the substation equipped with shield wires for cables lengths of up to 1000m. For
longer cables, the length of line with shield wires should be equivalent to at
least double the length of the cable. The shield wires and tower configuration must
effectively shield the line against direct strikes to the phase conductors as well as back
flashovers.
Inside this zone out from the substation, the footing resistance of the towers must be kept
sufficiently low, i.e. a few ohms.
The incoming lines should be fitted with surge arresters, to further limit the overvoltages
actually transferred into the substation.
The location of the arresters for cable protection must be as close as possible to the
cable terminal in order to achieve the maximum degree of protection.
__________________________________________________________________________________________
Specifically, flashovers are known to occur in cases when the line circuit breaker is open. This
risk is most likely to arise during normal operation as a result of subsequent strokes of the lightning
flash during the open-close cycle; presuming that prolonged opened breaker condition normally
dictates that the breaker disconnecting switches are also opened and grounded. Due to the open
breaker the incoming voltage surge will be doubled. As a result, there may be a flashover across
the open breaker or an insulation breakdown of other line-side equipment; notably instrument
transformers.
A backflashover on the transmission line towers adjacent to the substation can also lead to a
steep voltage wave propagating on the phase conductor towards the substation. The risk of
insulation failures in the substation is proportional to the rate of backflashovers on the lines. Other
important parameters are the steepness and the amplitude of the voltage wave entering the
substation. When selecting the MTBF for the substation insulation, it is important to consider the
consequences of a fault in terms of outage and repair times and costs.
Regardless of the cause, some form of protection should be used to prevent surges impinging on
the line-side equipment. Arresters should always be located adjacent to the power transformers
in the substation. With the line breaker closed, these arresters are often (subject to their effective
protective distance) all that is needed to protect apparatus connected to the same line within the
substation; presuming a low backflashover rate. However, with the breaker opened, no protection
exists for the line side of the breaker, making it and other equipment vulnerable. Occurrence of
overvoltages that may lead to flashovers is related to the key design and protection of the
incoming overhead line.
A Cigré survey made in 1989 showed that almost 60% of all reported violent failures of instrument
transformers were due to lightning. The survey concluded that the number of failures due either
to an inadequate lightning protection or to an inadequate transient withstand voltage is significant
and that this field has to be investigated to reduce the number of failures of Instrument
transformers.
There are a number of advantages with additional arresters located on the line side of the station;
i.e. in front of line breaker
Reduces the overvoltage seen at the station transformer
Reduces overvoltages seen at the closed breaker; especially with long distance to the
station arresters
Protects the open breaker by limiting the effects of multiple lightning strokes
Protects all equipment at the line entrance e.g. instrument transformers
Removes the risk for backflashover; especially when fitted on transmission line towers
with high footing resistance
__________________________________________________________________________________________
A metal-oxide (MO) surge arrester for high voltage applications comprises the following main
components:
• MO resistors (ZnO blocks)
• Internal parts
• Housing of porcelain or polymeric material with end fittings of metal (e.g. flanges)
• A voltage grading arrangement, where necessary (e.g. external grading rings)
The internal parts can differ considerably between a porcelain housed arrester and a polymer-
housed arrester. The only certain commonality between these two designs is that both include a
stack of series connected metal-oxide resistor elements, together with components to keep the
stack together.
__________________________________________________________________________________________
The MO resistor is a densely sintered block, pressed to a cylindrical body. The block consists of
approximately 90% zinc oxide (ZnO) and 10% of other rare earth oxides (additives). During the
manufacturing process a powder is prepared, which is then pressed to a cylindrical body under high
pressure. The pressed bodies are sintered in a kiln for several hours at a temperature in the order
of 1200 °C. During the sintering, the oxide powder transforms to a dense ceramic body with varistor
properties, whereby the additives form an intergranular layer surrounding the zinc oxide grains.
ZnO Grains
10–15 μm
These layers, or barriers, give the MO resistor its non-linear characteristics. Metal is applied on the
end surfaces of the finished varistor to improve the current carrying capability and to secure a good
contact between series-connected varistors. An insulating layer is also applied to the cylindrical
surface to give protection against external flashover and chemical influence.
Before the blocks are assembled in an arrester, they must be subjected to a variety of tests to
verify their protection performance, energy and current capability as well as long term electrical
stability.
In the low current region, called the ”prebreakdown region” (Region 1), the resistivity of the
material is temperature dependant. The normal continuous operating voltage is found in this
region. Here the surge arrester acts as a capacitor, with only small resistive currents through it.
This is mainly due to the metal-oxide barrier (intergranular layers) between the zinc-oxide grains
acting as insulating barriers. However, the MO resistor’s temperature influences the insulation
capability slightly, i.e. an increased temperature leads to a higher resistive current.
__________________________________________________________________________________________
At even larger current densities, the arrester is working in the “high current region” (Region 3) and
the curve turns upwards, which determines the impulse behaviour of the surge arrester. The
barriers between the ZnO grains are electrically broken down and the current increase is solely
limited by the resistivity of the ZnO grains themselves.
When the voltage across the arrester is reduced to a normal level, the working point returns again
to Region 1, without delay.
Voltage (p.u.)
Min protection levels in kV (peak)
according IEC60099-4
Icap, capacitive
current (no influence
from temperature)
Current (Ampere)
__________________________________________________________________________________________
An arrester must also be equipped with fastening devices to ease the erection. This is achieved by
assembling flanges (or similar) at one or both ends of the insulator. If the arrester consists of several
series connected units, the flanges are also used to mechanically and electrically secure arrester
units to each other.
Insulators can be manufactured with different mechanical fracture values. The required fracture
value for a specific insulator is determined by the design and intended use of the arrester.
The insulator housings for surge arresters have traditionally been made of porcelain. However,
today there is a strong trend, and even a preference, towards the use of silicone insulators for
arresters at all system voltages.
There are a number of reasons why silicone is seen as an attractive alternative to porcelain,
including:
• Better behaviour in polluted areas
• Better short-circuit capability with increased safety for other equipment and personnel
• Low weight
• Better earthquake withstand capability
It is incorrect, however, to believe that all polymer-housed arresters automatically have these
features just because the porcelain has been replaced by a polymeric insulator. The design must
be scrutinised carefully for each specific type, which can be grouped generally into the following
categories:
• Open or cage design
• Closed design
• Tubular design
__________________________________________________________________________________________
Polymeric materials can potentially be more affected by ageing due to partial discharges and
leakage currents on the surface, UV radiation, chemicals, etc., compared to porcelain, which is a
non-organic material. For this reason, the raw material is often blended with a variety of additives
and fillers to achieve the desired material features: UV stability, anti-tracking, flame-retardancy, etc.
Silicone, as a material, has a natural resistance against these effects, and thus such additives
simply aid in further improving the material’s inherent properties.
In the event that an arrester is required to dissipate more energy than it is capable of, it will
sacrifice itself by failing short circuit. Most commonly, arresters are connected between phase-
ground and the resultant earthfault will immediately collapse the voltage on that phase, thereby
protecting other equipment on the same phase. The upstream protection will initiate a breaker
trip to clear the fault, and the failed arrester can then be replaced.
__________________________________________________________________________________________
Fig. 11
Operating principle of the pressure
relief device of an ABB type EXLIM
porcelain housed arrester.
The sealing cover in ABB’s high voltage EXLIM porcelain-housed arresters also acts as an
overpressure relief device. Other manufacturers may have other solutions; a blast plate for example.
During normal service, the sealing
cover tightens against the porcelain. At
an internal short-circuit of the arrester,
an open arc occurs across the block
column. Due to the heat from the arc,
the internal pressure increases and
would soon reach a value that could
cause an explosion of the insulator if
no pressure relief device was present.
The sealing cover is designed such
that it will open, both at the top and
bottom of the porcelain housing, as
soon as the internal pressure reaches
a certain value (significantly below the
bursting pressure of the porcelain) and Fig. 12
the enclosed gas volume can be The position of the overpressure relief device on an ABB
evacuated to the outside of the type EXLIM T porcelain housed arrester. The figure shows the key
arrester. The internal pressure is thus parts of an arrester with the pressure relief and sealing plate,
relieved, and a violent shattering of the block column, spring device and the cemented metallic flange.
porcelain is avoided.
__________________________________________________________________________________________
Fig. 13
Operating principle of “ pressure relief” for an ABB type PEXLIM
moulded open-cage design.
(2) The gas streams trigger an external flashover and the internal
arc is commutated safely to the outside
The short circuit capability for surge arresters (porcelain and polymer) is verified by tests to
minimize the risk for damage to surrounding equipment and personnel. However, the risks related
to an open arc in service can also be influenced by the physical positioning of the equipment as
well as by the circuit connections.
In order to prevent internal corona during normal service conditions for these type of arresters, the
distance between the block column and insulator must be sufficiently large to ensure that the radial
voltage difference between the blocks and insulator will not create any partial discharges.
For polymer-housed arresters lacking such annular space in their design, the radial voltage
difference is entirely across the rubber insulator. In order to avoid puncturing of the insulator, the
rubber must be sufficiently thick. It is also very important that the insulator is free from voids to
prevent internal corona in the material which might lead to problems in the long term.
The maximum voltage stress occurring across the polymer material is proportional to the length of
the insulator. A longer insulator therefore requires that the thickness of the material is proportionally
increased with respect to the increase in length. Another solution is to reduce the height of the
individual units in a multi-unit arrester, since the maximum voltage across each unit is limited by the
non-linear current-voltage characteristic of the MO resistors.
__________________________________________________________________________________________
The temporary overvoltage and energy handling capabilities are closely related to the temperature
of the MO resistors during normal operation. This temperature depends on the power losses,
which increase rapidly with voltage stress, due to the block material’s non-linear voltage-current
characteristics. Therefore, the most essential parameters to minimize the temperature during
normal operation are inherently low power losses in the zinc oxide material, together with a linear
voltage distribution along the block column.
Under normal operating conditions and voltage, the MO resistors act like a capacitor. The voltage
distribution along the block column then depends on the capacitance of the MO resistors and the
influence of stray capacitances. The stray capacitances are strongly dependent on the height of
the block column. Short arresters - up to about one meter in height - usually have a sufficiently
linear voltage distribution along the block column, as the self-capacitance of the MO resistors is
relatively high. For taller arresters, the influence of stray capacitances makes the voltage
distribution less linear. If no measures are taken to prevent an uneven voltage distribution on a
tall arrester, the local voltage stress at the top may reach (or even exceed) the knee-point of the
voltage-current characteristic of the zinc oxide material. This leads to a localized increase in the
power losses, with high temperatures in the block column as a consequence.
Corona
Above the knee-point of the current-voltage characteristics, the
ring
blocks start to conduct large currents, which would ultimately lead to
the failure of the arrester. The amount of this current is determined
by the applied voltage and the total stray-capacitance of the arrester
to earth and can be considerable; particularly for high-voltage
arresters. Further, the localized heating of the MO resistors (hot-
spots) leads to a reduced energy absorption capability of the
arrester.
__________________________________________________________________________________________
Type tests in accordance with Standards to verify the long-term stability of the MO resistors are
hence not valid if the actual voltage stress on the arrester during service is allowed to exceed the
applied voltage stress proven in the type tests.
When grading arrangements for surge arresters are based on complete electrical field
calculations for each arrester design at the maximum continuous operating voltage and with the
maximum possible three-phase influence taken into account, this guarantees that the voltage
stress remains below the critical level at all points along the block column. This maximum voltage
stress level is then used in accelerated ageing tests on the MO resistors. In this way, the long-
term stability of the MO resistors is verified at the highest possible voltage stress found in any
installation under normal service conditions.
A guide for the determination of the voltage distribution along surge arresters using simplified
representations of arrester geometries and boundary conditions (applied voltage, proximity and
voltage applied to other objects in the vicinity) is given in IEC 60099-4.
This leads to the obvious benefit of lighter structures with subsequent reduced costs, and even the
possible complete elimination of the need for a structure at all if alternative mounting arrangements
are acceptable; e.g. suspended mounting.
Fig. 15
Two examples of possible mounting arrangements for ABB type
PEXLIM silicone housed surge arresters
__________________________________________________________________________________________
There are then several types of mechanical designs in common use: loops or rods, cross-winding
and tubes. These designs can be grouped generally into three basic categories:
• Open or cage design
• Closed or wrap design
• Tubular design
An outer insulator with sheds is required over the inner body, with two common methods for
achieving this being:
• A pre-moulded polymer insulator is made in a separate process, and then slipped over the
internal component assembly (which itself may be enclosed in soft polymer). The boundary
between the internal assembly and the outer polymer insulator is usually filled with grease
or gel, generally of silicone.
• The outer housing is moulded directly onto the internal components to form a void-free,
sealed housing along the entire length of the insulator.
Such designs lack enclosed gas volume. Should the arrester be stressed in excess of its design
capability, an internal arc will be established. Due to the design principle, the arc will easily tear or
burn its way through the polymer material, permitting the arc, along with any resultant gases, to
escape quickly and directly. Hence, special pressure relief vents or diaphragms are not required
for this type of design. However, it is important that the design is not “too open”, otherwise the
internal active elements may be violently cast out.
Penetration of water and moisture must also be prevented, which places strict requirements on the
sealing of the insulator at the metallic flanges (in the case of a pre-moulded housing) and adherence
or bonding of the rubber to all internal parts (in the case where the polymer is directly moulded onto
the inner body). It is equally of great importance that these designs are totally “void-free” and no
air pockets are present, otherwise partial discharges might occur, which would lead to the
destruction of the insulator over time.
ABB employs a unique, patented design for the PEXLIM arrester to enclose the MO resistors of
each module under pre-compression in a cage formed of glass-fibre reinforced loops fixed
between two yokes which form the electrodes. A special mineral-fibre is wound as belts around
the loops resulting in an open-cage design for the module. This achieves high mechanical
strength and excellent short circuit performance, through the belt-windings preventing explosive
expulsion of the internal components.
Each module is then passed through a computer-controlled cleaning and priming process.
Thereafter, the module is loaded in a highly automated vulcanising press, where silicone is
injected at high pressure and temperature to completely bond to the active parts, leaving no
internal voids or air spaces.
__________________________________________________________________________________________
Typical designs include a glass-fibre weave wound directly on the block column or a separate tube
in which the MO resistors are mounted. A soft polymer insulator is then fitted (either
pre-moulded or directly moulded) over this internal component assembly; often together with grease
or gel to fill the interfaces.
In order to obtain a good mechanical strength, the weave/tube must be made sufficiently strong,
which, in turn, might lead to a too strong/closed design with respect to short-circuit strength. The
internal overpressure could rise in the tube design to a high value before cracking the tube, which
may lead to an explosive failure with parts being thrown over a wide area. To prevent a violent
shattering of the housing, a variety of work-around solutions have been utilised, e.g. slots in the
tube. When glass-fibre weave is used, an alternative has been to arrange the windings in a special
manner to obtain weaknesses that may crack. These weaknesses are intended to ensure a
pressure relief and commutation of the internal arc to the outside; thus preventing an explosion.
Note that such alterations do not inherently then make these an “open/cage design”, as the arc path
is not considered to be direct and the internal components are still, in practical terms, completely
surrounded by hard material.
Sealing and partial discharge issues also require consideration in a similar manner as for the open
or cage design.
The internal parts are, in general, almost identical to those used in an arrester with porcelain
housing. In particular, the arrester must be equipped with some type of sealing and pressure relief
devices, similar to what is used on porcelain-housed arresters.
This design has the prime advantage that very high mechanical strength is possible (potentially
even higher than for porcelain) in combination with relatively low mass. Among the disadvantages
compared to other polymeric designs is less efficient cooling of the MO resistors and, if appropriate
precautions are not taken in the design, an increased risk of exposure of the polymeric material to
corona that may occur between the inner wall of the insulator and the block column during external
pollution.
__________________________________________________________________________________________
The unique properties of silicone makes it ideal for use as an electrical insulator due to its superior
properties in the following key areas:
This exceptional pollution performance is achieved due to silicone’s unique dynamic surface
properties resulting in low molecular weight silicone oils diffusing towards the surface to encapsulate
the pollution layer, so that no electrolyte is present that can conduct current. As a result, silicone-
housed arresters exhibit lower leakage and consistently higher flashover values over the total lifetime
of the arrester.
__________________________________________________________________________________________
RTV is the generic group of “Room Temperature Vulcanised” silicone materials, which includes
many of the commercially available household sealants, grease coatings, etc.
RTV 2 component Addition curing with platinum catalyst; permits faster curing at
higher temperatures. Typical applications are casting and
encapsulating.
LSR is the generic group of “Liquid Silicone Rubbers”. These are two component addition cured
by platinum catalyst at high temperature. Typical application is injection moulding.
HCR is the generic group of “High Consistency Rubbers”. These are a one component solid
rubber compound which crosslinks at high temperature (130 – 190 oC) with the aid of either a
peroxide or platinum catalyst. Typical applications are injection moulding and extrusion, and
includes the HTV (“High Temperature Vulcanised”) silicone used for ABB type PEXLIM and
TEXLIM surge arresters.
HCR currently holds around 85% of the insulator market at high voltage, but LSR market use is
growing (particularly in niche markets). HCR is a solid rubber material while LSR has a pasty to
liquid consistency, which lends them to use in different applications and different processing
methods.
__________________________________________________________________________________________
Processing
• Allows the use of low pressure injection systems, leading to lower investment costs
• Can be fully automated
• Fast cycle times at relatively low temperature
4.4.1.2 HCR
Processing
• Needs high pressure & clamp force injection machine, leading to high investment costs
• Platinum cured HCR faster cure rate than peroxide cured HCR, but slower than LSR
• Mould may not need as close tolerances as for LSR, leading to lower cost
4.4.1.3 Application
All groups belong to the silicone family and therefore each permit the fundamental and generic
advantages of silicone to be utilized. However, since there is an expectation of a long in-service
life for HV surge arresters, a high degree of security is required. The manufacturer must carefully
consider the complete design as a whole – including the choice of housing material - to ensure
the final product has adequate performance in all key areas of concern for a given application and
not just excellent in one specific area at the expense of others.
For example, ABB type PEXLIM surge arresters utilize a housing moulded from a specially
formulated HTV-silicone in order to ensure excellent performance in all key areas; including:
__________________________________________________________________________________________
There are a variety of parameters influencing the dimensioning of an arrester, but the demands as
required by a user can be divided into two basic categories:
• Protection against overvoltages
• High reliability and a long service life
Additionally, there is the requirement that the risk of personal injury and damage to adjacent
equipment shall be low in the event of an arrester overloading. Users are also beginning to put
requirements on environmental aspects, for example that arresters should be separable, recyclable
and only contain non-hazardous materials.
The above two main requirements are somewhat in contradiction to each other. Aiming to minimise
the residual voltage normally leads to the reduction in the capability of the arrester to withstand
power-frequency overvoltages. An improved protection level may therefore be achieved by slightly
increasing the risk of overloading the arresters. The acceptance for increase of this risk is, of course,
dependent on how well the amplitude and duration of the temporary overvoltages (TOV’s) can be
predicted. The selection of an arrester is therefore always a compromise between protection levels
and reliability.
A more detailed classification could be based on what stresses a surge arrester is normally
subjected to and what continuous stresses it shall withstand. For example:
• Continuous operating voltage
• Ambient temperature
• Rain, pollution, sun radiation
• Wind and possible ice loadings as well as forces in line terminal connections
For transient overvoltages, the primary task for an arrester is to protect. But it must also normally
be dimensioned to handle the current through it, as well as the heat generated by the overvoltage.
The risk of an external flashover must also be very low.
Detailed test requirements are given in International and National Standards, where the surge
arresters are classified with respect to various parameters such as energy capability, current
withstand, short-circuit capability and residual voltage.
__________________________________________________________________________________________
The continuous operating voltage for AC surge arresters is mainly at power frequency, i.e. 50 Hz or
60 Hz with some percent of superimposed harmonics. For other applications, e.g. HVDC, the
waveform of the voltage might be very complex or even a pure DC voltage. It must therefore be
verified for all applications that the MO resistors are able to withstand the actual voltage under their
technical and commercial lifetime; normally stated to be in the order of 30 years.
The basis for the dimensioning is the result from ageing procedures where possible ageing effects
are accelerated by performing tests at elevated temperature.
Notwithstanding potential brief daily extremes, operating requirements for arresters under
prolonged high ambient temperatures are arguably often overstated by users. It is worth noting
that the ambient air temperature surrounding the arrester is not necessarily the temperature of the
MO resistors themselves. Arresters installed outdoors, for example, will always have a proportion
of their housing in the shade. Thus, even if the ambient air temperature is considered high, it is the
average ambient temperature of the blocks themselves which should be the determining factor for
the evaluation of verifications made for thermal stability during the type tests. Further, this
temperature is normally considered to be the average over a 24 hour period.
In order for a manufacturer to verify that the arrester is capable of withstanding the highest possible
temperatures, certain type tests must be performed. All arresters, according to the IEC standard,
must be designed to withstand an ambient air temperature of -40 °C to +40 °C without impairing the
surge arrester’s function. Due to the MO resistor current-voltage characteristic, high MO resistor
temperatures may be decisive for the arrester’s design as resistive leakage current increases with
higher temperatures at Uc.
__________________________________________________________________________________________
A well-designed arrester should already employ blocks with very low losses and the heat transfer
mechanism should be optimized for these blocks. Thus, in practical terms, pollution performance
is usually improved by using one or more of the first three methods. In this regard, it is important
to note that an increased Uc without corresponding improvement in TOV capability is not effective.
Heating of the arresters due to direct sun radiation and self-heating is normally a minor problem.
The influence from the sun radiation is sometimes thought to be significant, as one might assume
that sun radiation can result in considerably high surface temperatures. However, it is the average
surface temperature of the complete arrester that counts, and sun radiation falls on less than half
of the insulator surface at any point in time. In fact, the closer to the equator an arrester is situated,
the smaller the fraction of the insulator surface that is subjected to direct radiation due to the
sheds. Furthermore, the average block temperature will be well below the localized surface
temperature possible from direct sun radiation and, provided the arrester has low power losses,
self-heating is also negligible. Heating from sources other than sun radiation must be checked
separately for each case.
The effect of direct sun radiation is accounted for in the IEC Operating Duty test of arrester sections,
whereby energy and thermal capacity are verified at a starting temperature of 60 °C; being actual
MO resistor temperature, which adds additional safety margin.
The most suitable way to connect an arrester to the overhead line is to arrange the tee-off vertically
and slack to the line terminal of the arrester to minimize the bending moment on the arrester. Since
surge arresters have a certain maximum bending moment for each design type, expressed in Nm,
the maximum force at the line terminal is lower for a tall arrester than for a shorter one of the same
type.
__________________________________________________________________________________________
Composite materials, such as glass-fibre joined in a matrix with epoxy or other polymeric materials,
can exhibit behaviour changes at high loading. The rate of this material degradation is determined
by temperature, applied force, velocity of the applied force, humidity and the time during which the
load is applied. It is therefore not sufficient to simply dimension the arrester with respect to its
breaking force, but rather consideration must also be taken to how the arrester withstands cyclical
stresses.
Since the resistive leakage current through the arrester is temperature-dependent, achieving
thermal stability is also dependent on the final block temperature. If, for example due to a prior
switching surge, the arrester already has a high starting temperature before being subjected to a
TOV, it will naturally have a lower overvoltage capability. This is exemplified in the TOV
characteristic given below (Figure 19), which shows the ability of a specific MO arrester to withstand
overvoltages with and without a preceding energy absorption.
For MO arresters, the TOV amplitudes are normally at, or immediately above, the knee-point of the
current-voltage characteristic. If a station class arrester is designed to fulfil the IEC standard, it shall
be able to withstand a TOV equal to the rated voltage of the arrester for at least 10 seconds; after
having first been pre-heated to the designated start temperature and immediately prior subjected to
rated thermal energy injection. In the case of distribution class arresters, rated thermal charge
transfer shall be used instead of thermal energy. This voltage level is also designated as the ”rated
voltage” of the arrester in compliance with IEC.
The TOV is generally regarded as a stiff voltage source, i.e. the surge arrester cannot influence the
voltage amplitude. For the dimensioning to fulfil a certain TOV level, the MO resistor characteristic
must be chosen such that the current through the arrester, and consequently the energy dissipation,
will not result in a temperature above the thermal instability point.
The TOV capability given for a certain surge arrester should always be assumed with a stiff voltage
source. However, if this is not the case, the TOV capability of the arrester is, generally, significantly
higher.
__________________________________________________________________________________________
A manufacturer is relatively free to assign any data for the arresters. A given arrester with MO
resistors capable of absorbing a certain amount of high energy (charge) could therefore be assigned
a high energy/charge classification with low TOV capability or, conversely, a low energy/charge
classification with high TOV capability. The ideal should naturally be to assign the highest energy
and charge rating together with the highest possible TOV capability.
The protective characteristic for a MO resistor is slightly dependent on the steepness of the
expected current. The below Figure 20 shows the characteristics for a specific arrester for three
different current shapes given in the Standards.
As can be noted from the diagram, the protection level for currents having a front time of 1µs are
approximately 10% higher compared to currents with a wave form 8/20µs or longer. However, even
more important than this marginal increase for steep current waves, is the effect of positioning the
arrester in relation to the protected equipment and the length of the connections. There is also an
effect as a result of the arrester’s own height (length). These effects add inductance (L) into the
circuit, typically 1µH/m for outdoor arresters, which results in a further increase in the overall residual
voltage against steep current impulses according to the formula U = L.di/dt.
In order to obtain an efficient protection against fast transients, for example caused by
backflashover close to a substation, large margins are therefore required between the protection
level of the surge arrester and the protected equipment’s insulation level.
A MO resistor with larger diameter normally has a better protection level with maintained
overvoltage capability. A better protection level, in this case, also automatically results in a better
energy capability.
Computer programs are used to make accurate calculations of the resulting overvoltages in a
substation originating from lightning and detailed models of the transmission line and substation are
made. In these type of calculations, a MO arrester may be modelled as shown in Figure 18.
__________________________________________________________________________________________
Fig. 20
Example of protective
characteristics for a specific
MO surge arrester.
__________________________________________________________________________________________
High voltage arresters were historically designated according to IEC with a specific line discharge
class 1 through 5, in ascending order of energy capability. The previous “Long duration current
impulse test” and “Switching surge operating duty test” consisted of the application of current
impulses in various forms to simulate discharges through the test sample of a pre-charged line
with defined parameters of line length, surge impedance and overvoltage factor. The testing was
not explicitly intended to represent an actual in-service condition that was likely to occur, but
instead was to be a means to compare performance under the same criteria between different
manufacturer’s designs.
With time, the meaningfulness and purpose with the testing has been neglected, e.g. if a Line
Discharge Class (LDC) 3 arrester was specified, then all LDC 3 arresters were often considered
equal without consideration to relative performance. Furthermore, the specified line discharges
represent very specific situations and these may not be representative to use with results from
system studies for other cases where energy stresses totally different from line discharges will be
imposed on arresters, e.g. capacitor switching, reactor switching, lightning discharges in line
arresters, etc.
There was increasing pressure from users for testing to answer the following sorts of questions:
• What are the energy handling capabilities for long duration currents of different
amplitudes or time durations or protective levels rather than those specified in the five
line discharge classes?
• What is the energy handling capability for any other current shape (single, repeated or
multiple impulse or alternating current)?
• What exactly is the difference between thermal and impulse energy handling capability?
• What is the highest possible “single event” energy stress that will not damage the MO
resistors, even if the thermal energy limit has not been reached?
• How often can these energy “events” be passed without damage to the arrester?
The classification system used in IEC 60099-4 prior to Edition 3.0 was not capable of sufficiently
answering these questions. Furthermore, the line discharge class gave only indirect information
on energy handling capability as it depended on the protective level. Whilst this may be useful
from a system perspective, it does not give adequate information about the arrester design
performance as a whole.
The former line discharge classes (LDC 1 – 5) are removed in Edition 3.0 and instead charge is
used to classify the arresters with respect to discharge capability. A repetitive charge transfer test
replaces the previous long duration current impulse withstand test.
__________________________________________________________________________________________
Aside from withstanding the energy from current impulses, the MO resistors must also have a
sufficiently high dielectric withstand so as to ensure that the voltage across the block will not result
in a puncture or a flashover across the block. To ensure a sufficient insulation withstand margin for
normal stresses, the MO resistors (together with all internal parts in a high voltage arrester) are
dimensioned to withstand current impulses with an amplitude of at least 100 kA, having a wave form
of 4/10 µs.
Requirements for high energy absorption capability can be solved by increasing the block volume -
either by using blocks with larger diameter or by paralleling block columns and/or arresters. To
ensure that the latter designs will operate correctly during service, a very careful procedure is
required to ensure a good current sharing between the block columns and/or arresters connected
in parallel. Furthermore, possible changes of the block characteristic due to the normal applied
service voltage as well as energy and voltage stresses must be extremely small.
In contrast to other HV apparatus, the insulation level for surge arresters therefore does not
inherently need to fulfil a standardised insulation class. Distance effects need not to be
considered. A longer arrester may, in fact, lead to less effective protection for steeper surges for
which the inductance of the arrester itself becomes more significant. Hence, the stipulation of high
external insulation withstand values (e.g. equal to that for the protected equipment) may thus be
disadvantageous for the protected equipment.
Instead, the arrester standards stipulate a specific safety margin between the residual voltage of
the arrester to the voltage withstand level of its external insulation to be sufficient. The voltage
across an arrester can never be higher than that determined by the arrester's protective
characteristics. Only the need for an additional (statistical) safety margin, including correction for
installation altitude, can technically justify a higher external insulation strength. Generally, the risk
of an external flashover less than or equal to 10-3 (0.1 % per year) is considered acceptable (IEC
60071-2); which leads to a factor of approximately 1.10 to 1.15 (excluding altitude correction)
between the arrester protective levels and the SIWV and LIWV of the housing, respectively.
Both the IEC and IEEE arrester standards clearly stipulate that such a margin is sufficient. For
example, IEEE stipulates that the external LIWV of the housing shall be 20 % above the discharge
voltage at 20 kA, 8/20 µs impulse plus an altitude factor of 9% per every 3000 feet (roughly equal
to 10 % per every 1000 m). IEC stipulates a LIWV margin of 15 % above the discharge voltage
at nominal current plus an altitude factor of 13 % for up to 1000 m.
The complete arrester, including possible grading rings, must also be designed to give a
reasonable safety margin against external flashovers. With the specified margins in the IEC
Standard, an acceptable low risk for external flashovers is obtained up to an altitude of 1000 m.
For higher altitudes, special consideration needs to be given on a case-by-case basis.
__________________________________________________________________________________________
RAD then varies directly with pressure and inversely with temperature. The normal value of air
pressure at mean sea level is 101.325 kPa although, depending on meteorological conditions, air
pressure at sea-level may vary from approximately 91% to 107% of this value. In areas above
sea-level air pressure is lower than at sea level and in areas below sea-level higher than at sea
level. Humidity also affects dielectric strength, with the extent dependent on the type of voltage
stress. From a standard humidity of 11 g/m3 , a dryer climate has a tendency to lower the
insulating quality of air while a moister climate tends to increase it due to irregular disruptive
discharge at relative humidity RH >80%.
For determination of the applicable atmospheric correction factor, IEC 60099-4 has concluded it
may be assumed that the effects of ambient temperature and humidity tend to cancel each other
and therefore, for insulation co-ordination purposes, only the air pressure corresponding to the
altitude of the location need be taken into account for both dry and wet insulations.
According to IEC 60071-2, an altitude correction factor, Ka, is then based on the dependence of
the atmospheric pressure on the altitude as given in IEC 60721-2-3. This correction factor can
be calculated from:
𝐻𝐻
𝑚𝑚� �
𝐾𝐾𝑎𝑎 = 𝑒𝑒 8150
Comparison of normal air pressure related to that for altitudes above sea-level at which
substations are anticipated to be installed and the correction factor Ka are given in the Table 1
below.
__________________________________________________________________________________________
This method is deemed the most appropriate method and will sufficiently compensate for air
pressure effects at the respective altitude. The suitability of other methods sometimes proposed
by designers for altitude correction are debatable for use on arresters, since current knowledge
is limited and practice too diverse; for example:
• insulation withstand level required only for the altitude difference over 1000m
• creepage distance adjustment
A surge arrester typically has a much longer housing than “necessary” to fit the internal active
elements. Therefore, even taking into account the down-hang of grading rings (when applicable),
the voltage withstand across the housing is usually much higher than should be needed with
respect to the stipulated safety margin to the protection level of the arrester according to IEC. The
result is that, even though the IEC design-altitude may be 1000masl, many standard arresters
are suitable for installation at significantly higher altitudes; without the need for additional
flashover-length/insulation withstand thanks to their design and inherently low protection levels.
The US-based IEEE Recommended Practice for Seismic Design of Substations (IEEE 693) is
often taken to be the “worst case” earthquake criteria since it provides a comprehensive
prescription for the seismic qualification of equipment. It allows for equipment to be certified as
meeting one of three levels of seismic resistance; with the appropriate level selected by
considering the peak ground acceleration for the 2500-year return period hazard spectrum at the
site, i.e. 2 % probability of occurrence in 50 years. Nevertheless, actual criteria for a specific
location should always be used in any analysis and caution is required when applying earthquake
design spectra from different design codes. True requirements demand consideration with
respect to the applicable zero period acceleration, return period, damping, subsoil type, structural
performance factors, assumed ductility, location, near-fault factors, design methodologies and
expected performance given the nature of the equipment.
When the earthquake is defined as a maximum horizontal acceleration, the bending moment can
be easily calculated when considering the arrester as a rigid body. When more accurate
calculations are necessary, the elasticity and damping of the arrester must first be determined in
a snap-back test; whereby a safe, but significant, force is applied to the terminal and then
suddenly released and the decay of the oscillations evaluated. Knowledge about elasticity,
resonance frequencies and related damping is also required when the earthquake is specified by
a frequency response spectrum. In such cases, a specially developed computer program will
need to be used to perform a Finite Element Method (FEM) response spectrum dynamic analysis.
However, a reliable calculation needs to also have adequate information about the support
structure on which the arrester is erected. For example, mounting the arrester on a rigid structure
which has a sufficiently high natural frequency may reduce or remove the seismic loading on the
arrester.
__________________________________________________________________________________________
Historically, testing has been done using a variety of test methods; with time history considering
a required response spectrum (RRS) being them most widely accepted. Typically a triaxial shake
table is then used to simulate the forces applied during an actual earthquake - including amplitudes,
frequencies, and energy levels that occur in a similar simultaneous manner. A test response
spectrum that envelops the RRS is applied in the two perpendicular horizontal axes of the
equipment together with a reduced response spectrum in the vertical axis (e.g. 80% of that in the
horizontal axes).
Since polymer-housed arresters are more or less elastic, temporary loads – including earthquake
forces and short-circuit forces - can be looked upon differently compared to rigid bodies like
porcelain insulators. The reason for this is that the forces do not have time to act fully due to the
elasticity of the material and mass inertia, i.e. the forces are spread out in time leading to the arrester
not encountering any high instantaneous values. These advantages, combined with a design with
small mass participation, have been fully utilised by ABB for the 550 kV arrester shown in
Figure 21. This arrester withstands a ground horizontal acceleration in excess of the highest
seismic demands as per IEEE standards. Seismic qualification testing has also been successfully
made on other standard ABB surge arresters (without additional bracing), even at the most
arduous 1.0g High Performance level. Alternatively, suspending polymer surge arresters directly
from the overhead line is a viable mounting alternative to eliminate seismic and other large
cantilever loads altogether.
Fig. 22
ABB type TEXLIM surge
arrester undergoing seismic
testing on triaxial shake table.
Fig. 23
IEEE 693 RRS for 0.5g
(High Seismic Level)
with 2% damping.
1,800
1,600
1,400
1,200
1,000
a (g)
0,800
0,600
Fig. 21 0,400
ABB type PEXLIM surge arrester 0,200
for 550 kV purpose- designed 0,000
to meet extreme earthquake 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 8 15 20 25 33 50
requirements in the Los Angeles f (Hz)
area (USA).
__________________________________________________________________________________________
High radial voltage stresses may occur between the block column and the outside of the insulator
during severe external pollution. Generally, external pollution may influence a surge arrester in
the following ways:
• External flashover
• Internal corona
• Heating of the blocks
• Tracking and/or erosion of the insulator
For a given pollution level, the performance of gapless arresters can be generally improved by
employing any or all of the following measures:
• Increasing the creepage distances of the housings to reduce external leakage current
• Using MO resistors of larger volume to improve the energy absorption capability
• Improving the TOV capability, i.e. increase the rated voltage (Ur) for the same arrester type
• Improving the heat transfer mechanism
• Using blocks with lower losses at Uc
To avoid external flashover, the creepage distance of the arrester, i.e. the shed form and the
length of the insulator, is typically designed in compliance with the same criteria valid for other
insulation at the actual site. If this is not sufficient alone, problems for arresters with porcelain
housings installed in extremely polluted areas have historically been solved by greasing the
insulator, thus improving the pollution performance. The aim of the greasing is to reduce the
leakage currents on the insulator surface. Hydrophobic materials, like silicone, give a similar
effect. This is one of the strongest motivations for why silicone has been seen as an attractive
insulator material.
A common belief is that all polymer-housed arresters have better pollution performance compared
to arresters with porcelain housings. However, a more correct statement would be that
hydrophobic materials (like silicone) have better performance in polluted areas due to reduced
external leakage currents. In contrast, EPDM rubber, which can lose its hydrophobic properties
quickly, should be designed in the same manner as porcelain from a pollution performance point
of view.
It is very difficult to avoid internal corona during severe external pollution on arresters containing
an annular gap between the MO resistors and the insulator, irrespective of whether the insulator
is made of porcelain or a polymeric material. The design of such arresters must therefore be able
to withstand corona during such pollution episodes; with some “rules-of-thumb” being:
• ”No” corona during dry conditions
• Minimise the use of organic materials in the arrester. When organic materials are used,
they must have been thoroughly tested and subjected to realistic corona tests
• Prevent the possibility of electrical discharges directly onto the MO resistors
__________________________________________________________________________________________
Possible thermal stresses are determined by the leakage currents that might be present on the
outer surface of the insulator. For porcelain arresters, it has been shown that the integral of the
leakage current, i.e. the charge, can be regarded as independent of the creepage distance, and
instead is approximately linearly dependent on the diameter of the housing. An insulator with a
larger diameter thus may give rise to higher thermal stress during conditions with external pollution,
provided the service conditions are otherwise the same.
Lower leakage currents on the insulator surface is achieved with a hydrophobic surface, eg. the use
of silicone insulators. Figure 24 shows leakage currents as measured on a porcelain insulator and
a polymer-housed arrester having a silicone insulator. The values are taken from test stations at
Dungeness on the English Channel as well as Kelso in South Africa. As can be noted, the
amplitudes of the leakage currents on the silicone insulators are significantly less compared to on
the surface of the porcelain insulator during the specific measuring interval.
Fig. 24
Comparison of the number of peak currents over 30 mA for silicone-housed and porcelain apparatus insulators
during storm periods. The numbers in the legend represent the specific creepage distance (mm/kV Us).
Source: Cigré A3-104
For applications requiring arresters with parallel housings and several units connected in series, the
general rule is that the units should not be connected in parallel except at the top and bottom. This
is because, during pollution episodes, the MO resistors in one unit could conduct the external
leakage current from all of the parallel connected arresters which consequently may give an
increased thermal stress on that unit. Since the MO resistors have a negative temperature
coefficient in the leakage-current region, i.e. the leakage current increases with increased
temperature, the heating of one unit will lead to a reduction of the voltage characteristic with
subsequent increase of the current. An increased current through the unit leads to higher power
losses with increased temperature, and so the cycle continues. Not even a careful current-sharing
test (matching) of the arrester units will be of help below the knee-point of current-voltage
characteristic.
__________________________________________________________________________________________
The concept of thermal stability can be depicted with the help of a heat loss – input balance
diagram, as depicted in the following Figure 25. This shows principally how the ability of an arrester
encapsulation to dissipate heat and the temperature dependent power losses of the MO resistors
result in a working temperature at a certain ambient temperature and chosen voltage stress (”A” in
the Figure). An upper maximum temperature also exists (”B” in the Figure), above which the design
is no longer thermally stable for a given voltage stress.
Losses at 0.9*Uref
2 Losses at 0.8*Uref
Losses at 0.7*Uref
A Losses at 0.6*Uref
1
A = Service temperature at 0
B = Thermal limit at 0.8*Uref
0
40 60 80 100 120 140 160 180 200
It can also be seen from Figure 25 that the instability threshold is very much dependant on the
applied power frequency voltage. As the power losses curve is non-linear, a lower applied service
voltage than verified in test, for example, would shift the upper intersection point further to the
right, thereby increasing significantly the temperature limit at which thermal runaway becomes a
risk.
To explain the concept further: The power losses of a typical MO resistor (curved line) due to a
constant applied power frequency voltage is extremely temperature dependent. At the same time,
the ability of the arrester assembly to dissipate heat is generally linear (straight line) and
proportional to its thermal design and temperature rise above the ambient temperature.
Consequently, there are two intersections of the two curves: one at low temperature – a so called
stable operating point, and the other at high temperature – a so called instability threshold. To
obtain thermal stability, the temperature rise due to power losses in the MO resistors must be
balanced against heat dissipation to the environment.
__________________________________________________________________________________________
Some “rules-of-thumb” for ensuring a design with good heat dissipation, and thereby low risk for
thermal runaway:
• Low-loss MO resistors
• Reduced voltage stress/mm
• Increased MO resistor size
• Homogenous block material
• Non-ageing MO resistors
• Good physical design with regards to thermal heat transfer
__________________________________________________________________________________________
By comparison, classification as per the IEEE C62.11 standard is as shown in Table 3. Station
and Intermediate arresters are to be tested for switching surge energy rating and single-impulse
withstand rating – being similar, but not identical, to the IEC test for W th and Qrs; respectively. All
designs further undergo a Duty-cycle test to verify that the arrester can withstand multiple lightning
impulses of classifying current without causing thermal instability or dielectric failure (a similar
style test was removed from the IEC Operating duty test as it was deemed unnecessary since
lightning impulses give relatively negligible stress compared to that otherwise verified in separate
energy tests).
Arrester classification Max. system voltage (kV) Impulse value crest (kA)
Station 800 20
Station 550 15
Station < 550 10
Intermediate All 5
Distribution, Heavy Duty All 10
Distribution, Normal Duty All 5
Distribution, Light Duty All 5
__________________________________________________________________________________________
The protective function is verified with different measurements of the voltage level for different
current amplitudes and current waveforms (residual voltage tests) and the reliability is checked
through a number of electrical and mechanical tests. An important part of the electrical tests is the
operating duty tests in which an arrester, or a pre-scaled model of the arrester, is subjected to a
combination of stresses representing anticipated service stresses that an arrester might be
subjected to during its lifetime. The lifetime is further verified by subjecting the MO resistors to an
accelerated ageing test procedure.
According to Standards, the tests can be divided into three main categories:
• Type tests (Design tests according to IEEE)
• Routine tests
• Acceptance tests
Type tests are performed after completion of the development of a new arrester design to
establish representative performance and to demonstrate compliance with the relevant
standard. Once made, these tests need not to be repeated unless the design is changed
in a way which may negatively influence the performance. Only the relevant tests need to
be repeated in such a case.
Routine tests are made on each arrester or arrester unit, as well as components, as a
quality control integrated in the production. Their aim is to ensure that the products meet
the design specification.
Acceptance tests are made on a number of randomly chosen arresters from a delivery
lot when it has been specially agreed between the manufacturer and the purchaser at the
time of ordering.
Specifically how surge arresters shall be tested is defined in detail in the Standards, with the two
most widely accepted being IEC 60099-4 (International Standard) and IEEE C62.11
(American National Standard). Since the IEC standards are international and thereby have a wider
scope for use, the coming sections focus primarily on the requirements specified by
IEC 60099-4.
The Standard IEC 60099-4 superseded the old Standard for gapped silicon-carbide surge arresters,
IEC 60099-1. IEC 60099-1 standard is not at all applicable to gapless MO arresters; especially
since there are no series gaps there can be no defining sparkover voltage testing.
IEC 60099-4 reflects a completely different approach on how to select test sections and verify the
arrester characteristics. Ongoing changes to IEC 60099-4 (from Amendment 2 in 2001) deal with
specific issues of importance, including: polymer-housed arresters, short-circuit performance,
accelerated ageing, voltage distribution, environmental and weather ageing tests, mechanical
testing and energy classification. See Table 4.
__________________________________________________________________________________________
Table 4. Type test requirements for AIS gapless MO arresters as per IEC 60099-4. Ed 3.0
Note: Separate and specific criteria apply to arresters for UHV (Us > 800kV), gas-insulated metal enclosed
arresters, separable and dead-front arresters, and liquid-immersed arresters. These are beyond the scope of
this guide.
__________________________________________________________________________________________
Furthermore, it is both customary and accepted that some of the tests are made on scaled-down
models of the arresters, thus making it possible to also scale-down the requirements on the test
equipment. These scaled-down arresters units are called ”section of an arrester” or ”prorated
section”. According to the definition in the IEC Standard, the arrester section intended for a particular
test must correctly represent the performance of the complete arrester during a specific test. An
arrester section may therefore look different depending on the intended tests:
Electrical
Electrical
connection
connection
Electrical
Electrical
connection
connection
Insulation
In both cases, the prorated section may also be a real arrester or arrester unit of the design.
__________________________________________________________________________________________
• Thermal comparison
Verification is made by heating a complete arrester unit and the thermally prorated section to an
elevated temperature by application of voltage. Thereafter, the MO resistor temperatures in the unit
and the section are measured during the cooling time. Cooling curves shall be established and
compared. A correctly designed thermal section shall not cool faster than the arrester unit, otherwise
any corresponding temperature difference shall be added to the start temperature for all thermal
recovery tests.
IEC 60071-1, for insulation co-ordination principles and rules, states that when it has been
demonstrated that one condition (dry or wet) or one polarity or a combination of these produces
the lowest withstand voltage, then it is sufficient to verify the withstand voltage for this particular
condition. Hence, insulation withstand tests shall be wet tests for outdoor arresters where wet
conditions are expected to lower the withstand voltages. Experience shows that this is the case
for power frequency tests and switching impulse tests, but not for lightning impulse tests. If the
arresters are intended for indoor use, dry tests are considered sufficient in all cases.
All distances between the arrester’s own external parts, e.g. grading ring to flanges, must be
checked with respect to voltage stress and withstand. Insulation withstand tests are thus specified
on the complete housing for arresters intended for use on systems of Us > 245 kV; being the
voltage level where multi-unit housings and significantly large external grading rings are most
commonly used. During the test, the assembled shall be equipped with the complete external
grading system, but the internal MO resistors shall be replaced by resistors, capacitors or higher
resistance MO resistors to obtain, approximately, the same voltage grading of the arrester during
high current discharges as would be given by the actual MO resistors used in the arrester.
Nonetheless, formulas are given for insulation withstand strength for lightning, switching and
power frequency voltage and, provided the calculated flashover distances from the formulas are
fulfilled, no tests are then actually required on the complete arrester.
__________________________________________________________________________________________
IEC
(installation < 1000 m) IEEE
Us < 245 kV Us > 245 kV (installation < 1800 m)
d is the dry arcing distance across the arrester housing
Station class:
Power frequency
(1.06 /√2) x Ups
Distribution class: Station class:
IEC:
(0.88 /√2) x Upl 0.82 x Ups
50 Hz, 1 minute Not applicable
No test is required if Distribution class:
(U/859) 0.833
d > [1.82 x (e – 1)] 1.36 x (10 s TOV)
IEEE:
where U = peak value of the
60 Hz, 10 second
power-frequency test voltage
1.25 x Ups
SIWV
No test is required if
250/2500 µs Not applicable Not defined
d > 2.2 x [e(U/1069) – 1]
where U = 1.25 x Ups
1.3 x Upl
LIWV 1.42 x Upl at 20 kA
1.3 x Upl No test required if
1.2/50 µs for all arresters
d > 1.3 x Upl /500
Table 5. Comparison of IEC and IEEE requirements for insulation withstand voltages
Note: IEEE does not use the same correction factors as IEC due partly to the difference
in maximum required design altitude (1800 m for IEEE compared with 1000 m for IEC).
The Standards make a distinction between different current impulses, based on different events in
the network:
__________________________________________________________________________________________
Residual
Residual voltage
voltage for 1/2 current
for 1/2µs currentimpulse
impulse
Residualvoltage
Residual voltage for 8/20
8/20µs current
current impulse
impulse
Voltage (kV) Fig. 27
12
Comparison between residual
8
voltage levels for current
pulses 8/20µs and 1/2µs.
4
The upper curves show the
0 Currentimpulse
Current impulse voltages and the lower the
8/20µs
8/20 currents.
-4
Current
Current impulse
impulse
Note that the curves have
-8
1/21/2µs been misaligned for clarity.
-12
The lower time scale is valid
Current (kA) 0 10 20 30 40 50 for the 1/2µs impulse, while
the upper scale is valid for
0 1 2 3 4 5
Time (microseconds)
the 8/20µs impulse
By testing with different current amplitudes for each of the current-shapes, a complete protection
characteristic is obtained for each waveform. For current impulses with the same amplitude, the
residual voltage level increases slightly for shorter front-times. This frequency dependence is
illustrated in Figure 27 showing results from a test with 10 kA for waveforms 8/20 µs and 1/2 µs.
The steeper front, 1 µs, may be the result of a lightning stroke very close to a substation protected
by surge arresters. Further, inductance effects can become significant with steep current impulses,
and IEC specifies that the steep current impulse residual voltage tests include inductive voltage
contribution – Magnitude of inductive voltage drop for AIS arresters: 2.5, 5, 10 or 20 kV/m of
arrester length for arresters with nominal discharge current of 2.5, 5, 10 or 20 kA, respectively.
In order to generate the specified current pulses an impulse generator is needed with the
capability to create currents up to 40 kA. To be able to create such currents through a complete
arrester at high voltages would require very large impulse generators, since the test equipment
must principally be able to simulate full-scale lightning. Tests on complete arresters are however
not necessary, nor desirable for reasons of accuracy. IEC therefore recommends that the residual
voltage tests are made on scaled-down models of the arrester and specifies also how the measured
values shall be re-calculated to be valid for a complete arrester.
According to IEC, the objective of the residual voltage type tests is to verify the claimed protection
levels by checking the relationship of protection levels at different current waveforms and
amplitudes to a level which is checked in routine tests on all arresters. Normally the residual voltage
at 10 kA with waveform 8/20 µs is used as a reference. This means that the 10 kA level with this
waveform must be verified in a routine test and given for all manufactured arresters. The
requirement for a routine test can be fulfilled by measuring the residual voltage for each individual
block within the arrester and summing up the result. This procedure will be correct, since all blocks
in a single column arrester will be subjected to the same current.
__________________________________________________________________________________________
IEC 60099-4 specifies an accelerated ageing test during 1000 hours at an elevated temperature of
115 oC as a type test. For arresters filled with air, the MO resistors need not be encapsulated during
the test. If the surrounding atmosphere is something else (e.g. nitrogen or other gas) the test must
be performed with the blocks in that particular atmosphere. For polymer-housed arresters, where
the blocks are in direct contact with other materials, the ageing test must be made including all
materials which are in direct contact with the blocks to show that the blocks are not negatively
affected (i.e. aged) due to influence from the other materials.
The accelerating ageing test is based on the Arrhenius law, which provides good confidence on life
expectancy of MO resistors. When tested according to the IEC requirements, the equivalent
minimum demonstrated lifetime is predicted to be 110 years at the conservative average ambient
temperature of 40 oC.
An important parameter during the ageing test is the selection of the voltage stress on the blocks.
The test voltage must reflect the highest possible local voltage stress in the arrester when it is
energized at the highest possible continuous operating voltage, Uc, assigned to the arrester. A
thorough electrical field calculation therefore must be made for each arrester type and rated
voltage which, in turn, is the basis for determining the relevant voltage stress during the
accelerated ageing test procedure. Influence from all phases in a three-phase configuration must
also be taken into account when performing the calculations.
The test shall be considered passed if the following criteria are met:
• any increase of power losses from Pmin is not greater than 1.3 times Pmin during the remaining test period
• all measurements of power losses throughout the ageing period ,including the final measurement, Pend ,is not
greater than 1.1 times Pstart
__________________________________________________________________________________________
It is however not sufficient to check only the characteristics of the blocks, but rather the entire
arrester must be seen as a unity. The ability of the arrester housing to dissipate heat must also be
adjusted to the power losses of the blocks during different service conditions with respect to voltage,
temperature and even frequency. This is necessary to ensure that the average block temperature
will not considerably exceed the ambient temperature, and thereby remain thermally stable.
If the samples pass, then MO resistors, equal in number to those used in parallel to the grading
components in the arrester, shall be connected in parallel to the test sample, and two lightning
impulses shall be applied. In order to pass, there shall not be any evidence of electrical
breakdown and their electrical performance shall be essentially unaffected.
Additional samples shall also undergo a thermal cyclic test, whereby five 48 hour cycles of heating
and cooling to 60 °C and - 40 °C (maintained for at least 16 hours, respectively) shall be applied;
but without any voltage on. In order to pass, there shall not be any evidence of cracking and their
electrical performance shall be essentially unaffected. If the samples pass, then MO resistors,
equal in number to those used in parallel to the grading components in the arrester, shall be
connected in parallel to the test sample and withstand lightning impulses in the same way as for
the similar test described above.
Investigations conducted by Cigré identified differences in the failure energy depending on the
type of energy being injected, or the charge being transferred. With some exceptions, energy
handling capability was generally found to be higher for short impulses of high current density
compared to long lasting events of low current density.
__________________________________________________________________________________________
The classification for the repetitive charge transfer rating value (Qrs) is to be taken from the
following list, with the final claimed rating to be not more than 90 % of the tested level:
• from 0.1 C to 1.2 C in steps of 0.1 C
• from 1.2 C to 4.4 C in steps of 0.4 C
• from 4.4 C up to 10.0 C in steps of 0.8 C
• from 10 C to 20 C in steps of 2 C
• from 20 C upward in steps of 4 C
It is important to note that the test shall be made with 10% higher charge than the rated value.
The test charge shall also be further increased to take into consideration the MO resistors with
highest permitted residual voltage in the arresters. This means that if MO resistors with, say, 10%
higher residual voltage than the nominal value are allowed in the arrester, the test shall be
performed at 1.1 x 1.1 = 1.21 times the rated charge in the case MO resistors with nominal residual
voltage are tested.
__________________________________________________________________________________________
Three prorated sections shall undergo the test procedure to verify the thermal energy rating, Wth,
as per Table 7. If the test sections used for the conditioning part are a dielectric prorated section
then the first high current impulse of the conditioning may be considered to also test the dielectric
withstand of the internal components, otherwise a separate test to verify the internal insulation
strength is required. Thermally prorated sections shall be used for the thermal recovery part of
the test.
For station class arresters, intended for use in a station, long-duration (rectangular) impulses of
2 – 4 ms virtual total duration, or unipolar sine half-wave impulses of 2 – 4 ms total duration, are
to be used. The current amplitudes and number of impulses is not critical, provided the
accumulated energy within a time of 3 minutes is at least equal to 1.0 to 1.1 times the claimed
thermal energy rating (accounting for any current sharing in the case of multi-column designs).
Other styles of impulses are to be used for arresters intended for overhead lines and distribution
class arresters. Each class of arrester has a minimum requirement to meet as stated in Table 2
above.
The classification for the rated thermal energy value (Wth) is to be taken from the following list:
• from 1 kJ/kV to 5 kJ/kV in steps of 0.5 kJ/kV
• from 5 kJ/kV to 16 kJ/kV in steps of 1 kJ/kV
• from 16 kJ/kV to 30 kJ/kV in steps of 2 kJ/kV
• from 30 kJ/kV up in steps of 6 kJ/kV
__________________________________________________________________________________________
Initial tests for sample • Residual voltage test at nominal discharge current
characterization • Reference voltage test at specified reference current
• Check for correct current sharing in case of multi-column arrester design
Determination of continuous operating voltage and rated voltage
Preheating Preheating to start temperature (as defined in the standard dependent on if
Us is < 800 kV or > 800 kV)
With prior duty Rated thermal energy injection, Wth, within three minutes by one or more long-duration
(4 new samples) current impulses or by unipolar sine half-wave current impulses
Application of test voltage and duration according to TOV curve (within 100 ms)
Application of Uc for at least 30 min (until pass or fail is evident)
Without prior duty Application of test voltage and duration according to TOV curve (within 100 ms)
(2 new samples) Application of Uc for at least 30 min (until pass or fail is evident)
Test evaluation • Thermal recovery
• No physical damage
• Change of residual voltage at nominal discharge current within ±5%
An additional temperature cycling and seal pumping test shall verify that the device is secure
against moisture ingress. Further, bending moment, tensile load and torsional load tests shall be
performed on disconnectors used with NGLA.
As a result, special requirements are set on arresters to ensure that a possible arrester failure will
not cause consequential damage to other equipment or injury to personnel. The Standards
therefore require tests where a deliberate internal short-circuit has been made to check the short-
circuit / pressure relief capability.
__________________________________________________________________________________________
For the high current tests, the test samples should be the longest mechanical section with the
highest rated voltage of each different design of arrester. It is accepted that approved high current
tests made on the longest housing also covers all shorter insulators of the same design. For the
low current test, however, the test sample may be a mechanical section of any length with the
highest rated voltage used for each different design and chosen length of test sample.
Arresters with "Design A" have a gas channel running along the entire length of the arrester unit
and fills >50% of the internal volume not occupied by the internal active elements. For this design,
the probability of a failure initiated in the gas volume is much higher than in the solid material.
This type of design makes use of the internal overpressure which is built up due to the internal
arc resulting from the short-circuit of the active elements. The overpressure is created by heating
the enclosed volume of gas, which expands, leading to bursting or flipping of a pressure relief
device (including pre-fabricated weak spots in the housing). In this case the tests are sometimes
called “pressure relief tests”. The arrester housing is dimensioned to not violently break before
the overpressure is relieved. Typically, these arresters are porcelain-housed or polymer-housed
with a composite hollow insulator incorporating an annular gap, i.e. tubular design.
__________________________________________________________________________________________
In “Design A”, a breakdown or flashover in the gas channel will develop an arc very quickly over
the entire length of the arrester. This may generate an intensive shock wave in the gas channel,
stressing the housing over its entire length and imposing high requirements on pressure relief
devices to open quickly. For “Design B”, in case of a breakdown in the solid material, the arc will
develop more slowly.
On the other hand, a fuse wire through holes drilled in the blocks is conversely a too harsh
scenario for this kind of arrester, as it extremely unlikely that all MO resistors of a failing polymer
arrester with this design will be punctured. It is therefore justified to specify the pre-failing method
for “Design B” polymer-housed arresters, which among the alternatives gives a reasonable
compromise with regard to test severeness and realism, and it automatically covers possible
influences of material homogeneity. Conversely, the pre-failing method may be less severe for a
porcelain-housed “Design B” arrester in the case where the arc develops elsewhere than in the
solid material. To cover for the worst-case scenario, “Design B” porcelain-housed should have a
fuse wire drawn along the surface of the block column as far away as possible from the gas
channel.
__________________________________________________________________________________________
Fig. 30 Circuit layout for short circuit testing (source IEC 60099-4)
For porcelain-housed arresters, the C-connection provides the most unfavourable conditions
during the initial phase of the test before venting occurs. Once the arc is externalized, it may then
be kept in close proximity to the arrester housing, resulting in a thermal shock effect causing
excessive chipping and shattering of porcelain sheds. During the remaining arcing time, this
routing forces the arc to move away from the arrester, and thus reduces the risk of the arrester
catching fire.
For polymer-housed arresters, the conductors should be routed as per the Z-connection. In this
way, the arc will stay close to the arrester during the entire duration of the short-circuit current,
thus creating the most unfavourable conditions with regards to fire hazard.
__________________________________________________________________________________________
It had been observed in many cases that after non-violent thermal breaking of the porcelain
housing the active part, completely intact, fell down such that its top end collided with the
enclosure (which previously had a radius equal to the arrester height) and some parts of its top
end fell just outside the enclosure. Though in these cases the short-circuit performance was good,
the arrester did not pass the test. In order to avoid this conflict the radius of the enclosure has
now been chosen to be 20% larger than the arrester height; dimensioned according to the below
formula or, in any case, at least D = 1.8m.
𝐷𝐷 = 1.2 × (2 × 𝐻𝐻 + 𝐷𝐷𝑎𝑎𝑎𝑎𝑎𝑎 )
where D diameter of the enclosure (or side in the case of a square)
H height of the test arrester
Darr diameter of the test arrester
As it usually takes more time than up to the first current peak until the arc has fully commutated
to the outside of the housing on “Design A” arresters, a certain amplitude for the first peak of the
current is defined in the test procedure for the rated short-circuit current - which has to be at least
2.5 times the r.m.s value of the symmetrical component of the prospective short-circuit current.
Many difficulties have arisen in trying to achieving a value of 2.5 for the “asymmetry factor” on
certain polymer arrester designs. These difficulties lead to a limitation of laboratories that could
perform the tests or a modification in the testing procedures (too severe/simple); neither of which
is truly desirable. Specifically for “Design B” arresters, it has been proposed that a symmetrical
current transfers more energy during the first 2 – 3 ms, which is the typical time of the housing to
open for this design. Regardless, skipping the asymmetry factor on these designs permits the
arresters to be tested in full length, and this is considered to be more significant to verifying short
circuit behaviour than any marginal energy differences because of a first peak factor. Thus, for
“Design B” arresters, the peak value of the first half cycle of the prospective current shall be at
least √2 times the r.m.s. value of the rated short circuit current, but otherwise there is no specific
requirement for asymmetry factor with this design.
Similarly, for the reduced short-circuit currents and regardless of design, the peak value of the first
half cycle of the appropriate test current need only be at least √2 times the r.m.s value. In the
case of the low current test, the actual first half cycle peak of the test current must be at least √2
times the r.m.s value. The low current shall flow through the arrester for 1 second or,
for “Design A” porcelain-housed arresters, until venting occurs.
As polymer-housed “Design A” arresters are being produced in increasingly longer units with high
rated voltages, the first asymmetric peak current of 2,5 may not be easily achieved. If this is the
case, IEC 60099-4 proposes testing of an additional (shorter) test sample as well as possible
special adjustments of the test circuit.
__________________________________________________________________________________________
IEC 62271-200 for metal-enclosed substations and IEC 61330 for medium-voltage/low voltage
prefabricated substations explicitly deal with internal arc testing with respect to the safety of
personnel and public in case of failures within such stations. Two levels of accessibility have been
adopted, where the accessibility B refers to stations that are directly accessible to the public.
Although these standards deal with the safety of persons and are therefore very restrictive
concerning possibly arising danger, they permit projection of parts out of the station up to a weight
of 60 g. Since it is advantageous if all standards dealing with similar parts of a power system
specify similar requirements, it was decided that the same criteria be also adopted for the short-
circuit testing of surge arresters. Hence, fragments of ceramic material (MO resistor or porcelain)
of up to 60 g are allowed to be found outside the enclosure, as are pressure relief vent covers
and diaphragms and soft parts of polymeric materials.
These forces will cause a bending moment, which typically has its maximum at the base of the
arrester. The arrester must withstand this moment. In the case of multi-unit arresters, individual
units must also withstand the moment at their length resulting from the applied forces. Bending
moment tests are performed by fixing the housing to the floor and subjecting it to a horizontal
force at the top. The force is then slowly increased until the housing breaks, or in the case of
verification, that the declared value is reached. The test may be performed on complete arresters
or arrester units.
According to IEC 60099-4, several sample tests should be performed on porcelain-unit housings to
determine the mean value of breaking load (MBL). It is then possible to assign the housing a
maximum specified short-term load (SSL), i.e. the 100% value in Figure 31, which can be
considered its maximum withstand moment against dynamic loads such as short circuit forces,
gust winds, earthquake, etc.
SSL is also to be verified by test with the active elements in place to ensure the design really can
withstand this load and not be affected. This should not be confused with the breaking limit proven
during testing of porcelain-housings, which is an average of 20% above this value. The specified
long-term load (SLL), which is the maximum static (continuous) moment, should be limited to 40%
of the SSL on porcelain arresters.
__________________________________________________________________________________________
If potential effects of “mechanical ageing” during continuous loading are not considered, load
levels may be chosen at very high levels compared to what the arrester design can actually
handle. In other words, the short-term load level could be set just a little below the breaking load.
However, it is in doubt whether the arrester would handle these high mechanical loads at the
given load levels (continuous and short-term) under actual conditions during its service lifetime.
There are a number of potential problems that can arise by exposing an arrester to a continuous
load that is too near to the breaking load. These include:
• Damage to the housing of the arrester, which could cause the arrester structure to
completely collapse and break
• Damage to internal parts of the arrester, for example damage to the MO resistors
causing electrical malfunction
• Cracks in the polymer housing or sealing, which could lead to moisture ingress and
electrical failure
Of specific interest is the performance of polymer arresters under continuous loading of a cyclic
nature. Due to their construction, polymer arresters of all designs may flex under mechanical load
and, when this is repeated cyclically (as would occur over their service lifetime), may be the
primary factor which determines the true limit of permissible mechanical loading. A specified
short-term load verified on new arresters not previously subjected to any tests may thus give a
too optimistic value.
__________________________________________________________________________________________
If the arrester passes 1000 cycles at the SLL and subsequent water immersion and evaluation
tests, it is considered likely that the arrester can continuously be subjected to the SLL
Furthermore, the test validates that the SSL is a load which the arrester could be subjected to
even after many years in service. For short polymer arresters, i.e. arresters for system voltages
not exceeding 52 kV, a cyclic load test has not been considered necessary; neither for porcelain-
housed arresters, considering the long experience of this type of arrester.
7.1.11 Environmental
The environmental tests are intended to demonstrate by accelerated test procedures that the
sealing mechanism on porcelain-housed arresters and the exposed metal components, e.g.
flanges and terminals, are not impaired by environmental conditions. For polymer-housed
arresters, it is considered that the weather ageing test imposes sufficient environmental stress,
and hence no additional tests are needed on these types of arresters.
The requirement consists briefly of the below listed individual tests, with the criteria described in
more detail in the relevant IEC 60068-2 documents. The arresters shall be considered
satisfactory provided no degradation in the sealing has occurred.
__________________________________________________________________________________________
Sealing tests on arresters with enclosed gas volume and a separate sealing system should be
made using a sensitive method that can detect very low leakage rates (for example,
max 1W = 1 x 10-6 Pa. m3/s = 1 x 10-5 mbar.litre/s according to IEC 60099-4). Example of test
methods which are commonly used include:
• Helium-mass spectrometer
• Vacuum over water
• Pressure or vacuum decay
• Halogen detection
A moisture ingress test - which applies to polymer arresters only as an integral part of the bending
moment test (see 8.1.10.1) – is performed as a means to demonstrate the ability of the arrester
to resist ingress of moisture after being subjected to specified mechanical stresses.
The test includes subjecting the arrester to both thermal as well as mechanical cycling, as depicted
in Figure 33. After the cycling, the arrester is placed in boiling salt water for 42 hours, and thereafter
moisture is given time to possibility penetrate the arrester (Figure 34). Electrical measurements are
made both before and after the test sequences to verify that the specimen has not absorbed any
moisture. If the electrical characteristic of the arrester has changed during the tests, the most likely
conclusion is that moisture has penetrated inside the housing, which would imply that the arrester
no longer fulfils the original requirements.
Temperature
+60 ºC
+45 ºC
24 h 48h 72 h 96 h
Time
-25 ºC
-40 ºC
__________________________________________________________________________________________
A low corona level – both internal and external – is essential for all surge arresters to achieve
during normal operating conditions. As a consequence, more significance is placed on checking
for corona as a routine test on all produced units rather than as a single type test. Hence a type
test for partial discharge is no longer required by IEC, and instead the partial discharge routine
test is intended to detect internal corona which could otherwise lead to problems in the long term.
The arrester unit must be first pre-stressed at significantly higher than its normal operating voltage
in order to create the potential for initiation of discharges. Thereafter the voltage is reduced to a
value somewhat higher than the arrester unit’s continuous operating voltage, at which the internal
partial discharge level is recorded (measured as apparent charge in pico-coulomb, pC). IEC
60099-4 permits an internal partial discharge level of not more than 10 pC at 1.05 times Uc;
however an arrester manufacturer may choose to set tougher criteria than this.
A number of different methods have been trialed for artificial pollution tests on surge arresters,
with those specifically intended for porcelain-housed arresters having the intention of determining
• risk for external flashover
• effect of partial discharges inside the surge arrester due to radial fields between the
external surface and the internal active elements
• adverse temperature rise of the internal active elements due to a non-linear and
transient voltage grading caused by the pollution layer on the surface of the housing
Different methods are intended to test for one or more phenomena. Further, artificial pollution
tests aimed at determining localised temperature rise are only considered applicable to multi-unit
arresters, since single-unit arresters do not have a direct electrical connection between inside to
outside along their length. However, the risk of puncture exists for very long units. The conclusion
is, of course, that it is necessary to have an arrester design (both internal and external) which
minimizes such stresses and/or their effect under all anticipated conditions.
A problem with many of the pollution test methods is that their relevance to real conditions during
arrester life is questionable. Such methods test the arresters behaviour in more-or-less irrelevant
respects, and thus help neither users nor manufacturers to judge between arrester designs with
respect to pollution performance. A meaningful test program for surge arresters must therefore
start with an investigation of the pollution conditions which arresters can see in real life and what
effect these conditions will actually have on arrester designs.
Consequently, field-tests of arresters in areas with severe natural pollution have been performed
to sort out the relevant mechanisms for arrester performance under polluted conditions. Since
1982, ABB installed MO surge arresters at different sites with known severe pollution conditions
(marine, desert, tropical, industrial, etc.). The testing was carried out in collaboration with
recognized leading power utilities around the world. The field tests included conductivity
measurement of natural-polluted layer, recording of external and internal currents with counting
of current pulses and temperature recordings. The results and experience gained from these field-
tests contributed greatly in the designing of ABB type EXLIM porcelain arresters to ensure their
optimum pollution performance, even under the most severe conditions.
__________________________________________________________________________________________
The weather ageing test in IEC 60099-4 is thus applicable only to polymer housed arresters. The
test is intended as a continuous test with a duration of 1000 hours under salt fog conditions at
constant power frequency voltage equal to Uc. It shall be performed on the longest electrical unit
with the minimum specific creepage distance and the highest rated voltage recommended for the
arrester type.
This test is primarily intended to age the polymer material so as to cause tracking, erosion or
puncture; although other failure mechanisms may also occur. Interruptions due to flashover are
permitted and the arrester shall then be washed with clean tap water and the test restarted with
a lower salt content for the fog. The test is considered passed if no tracking occurs, erosion does
not occur through the entire thickness of the external housing to the next material layer, the sheds
and housing are not punctured, and the electrical performance of the arrester is substantially
unchanged from before to after the test.
Some users may consider their environmental conditions so extreme - intense solar radiation,
frequent temperature inversion with condensation, pollution – that they desire an extended multi-
stress test to validate the design. Such a 5000 hour continuous duration test was an option in
previous editions of IEC 60099-4. However, this has since been removed completely as it was
deemed generally unnecessary after past experience, especially in the case when silicone is used
as the housing material; being the majority of designs today.
Solar radiation simulation has otherwise been incorporated as a separate test to evaluate the
recognized main key element for potential ageing of polymer materials. This is in the form of a
1000 hour ultra-violet (UV) light test on housing-material specimens using either a Xenon-arc or
Fluorescent UV method. In order to pass, the material shall not be visibly degraded by the test.
__________________________________________________________________________________________
• Leakage check of the sealed housing (for arresters with a sealed housing)
o This test proves that the housing hermetically seals the active parts of the arrester.
Common test arrangements are helium-mass spectrometer, vacuum over water, pressure
or vacuum decay and halogen detection. This test is not applicable to certain polymer
arrester designs, when the active parts are directly sealed in the housing material.
IEEE specifies similar requirements for routine tests as per IEC, but uses slightly different
terminology:
• Power-frequency test
• Discharge-voltage test
• Partial discharge (PD) test
• Seal test
• Current sharing test
Manufacturers may also choose to perform additional quality checks. For example, all ABB type
HV arresters are subjected to the above tests (as applicable), plus a measurement of power
losses and grading current at Uc (MCOV). In addition, ABB also routinely undertakes energy
withstand tests on all produced MO resistors (with those having insufficient capability being
automatically rejected), together with accelerated life and impulse current tests as well as a check
of low current characteristics, protection characteristics and capacitance on samples from every
produced batch of MO resistors.
__________________________________________________________________________________________
Additionally, a Special Thermal Stability test is given as an option, which has to be specially
agreed upon. This is, in principle, a shortened version of the Operating Duty test, performed on
blocks from the same batch (or similar) as those used in the arresters from the delivery lot.
That said, the tests made after assembly of a single-unit arrester are then performed on a
“complete arrester”. However, to fulfil the requirements of the Standard, an additional lightning
residual impulse voltage test on the unit may be required if this is not performed as routine. It is
permitted, for example, as a routine test to measure residual voltages on individual MO resistors
for a specific applied current and then sum up the values to give the total residual voltage for the
unit. Because of the lower voltage required at the block level, this permits testing with high
lightning impulse currents (e.g. 10kA) and good measuring accuracy. Conversely, testing a
complete unit (or complete multi-unit arrester) at higher voltages can present problems regarding
circuit capacity to achieve high lightning impulse currents, as well as potential loss of accuracy in
the measured values compared with performing the test on individual blocks.
For a multi-unit arrester, consisting of several individual units, the units may be regarded as
impedances connected in series, where each individual unit has a specific voltage drop (or
residual voltage) for a specific applied current. Thus, measured values on units when summed
up can be regarded as valid for the complete arrester. Specifically in the case of reference voltage
measurement at reference current, provided that the current is high enough to not be affected by
stray capacitances during the measurement, then the summed values on individual units can also
be regarded as valid for the complete arrester. Similarly, for the internal partial discharge test,
provided the pro-rata voltage used on individual units is equal to or higher than the required test
voltage, then assembling the units together will not influence the result with respect to internal PD
measurement.
Acceptance tests made on completely assembled multi-unit arresters will therefore not give any
additional information about the surge arrester characteristics than obtained during tests on
individual units, nor add value or security to the arresters from a delivery lot.
With regards the Special Thermal Stability test, there is a good reason that it is an additional test
to only be made after separate agreement: The test is, in essence, a truncated version of the
Operating Duty type test, and thus if an Operating Duty test applicable to the surge arresters has
been performed in recent years, additional testing should not be necessary as undertaking such
a test will not give any additional information about the ordered arrester’s inherent performance.
__________________________________________________________________________________________
This section contains only a brief guidance for selection of the most important parameters of HV
surge arresters used for standard applications such as transformer protection. For a more
comprehensive guide, reference is made to IEC 60099-5 and ABB Application Guides. For
specialized applications, a more detailed system analysis or insulation co-ordination study may
be necessary to permit selection.
It is vital that the correct arrester is selected which will provide the desired protection, as well as
withstand normal and specified abnormal service conditions. The basic selection is carried out in
two major steps:
• Match the electrical characteristics of the arrester to the system’s electrical demands
• Match the mechanical characteristics of the arrester to the system’s mechanical and
environmental requirements
The simplified process for selection of the electrical characteristics is depicted in the following
flowchart (Figure 35).
__________________________________________________________________________________________
The most commonly occurring TOV is that at a single line earth fault. The amplitude is given by
multiplying Us/√3 by the earth-fault factor ke, which in turn is determined from the earthing
conditions. Figure 36 gives the value for ke based on the system sequence reactances and
resistances for the most unfavourable fault resistance. Should these system parameters be
unknown, ke is often assumed to be 1.4 for directly earthed systems and 1.73 for resonant earthed
or isolated neutral systems.
Fig. 36
Curves showing relationship between
R0/X1 and X0/X1 for constant earth fault
factor ke and zero fault resistance
(Source: IEC 60071-2)
The duration of the applied TOV during earthfault depends on the fault-clearance time. If this is
not known, it may usually be estimated to be in the range of 1 to 3 seconds for directly earthed
HV systems and 3 to 10 seconds for directly earthed distribution systems. For isolated neutral or
resonant earthed systems, the duration is important to determine more specifically, as it may
vary from a few seconds to some hours or even days; depending on whether fault-clearing is used
or not. For an anticipated fault duration over 2 hours, the TOV should generally be considered
(in most cases) as continuous, and the arrester rating chosen accordingly.
For the most common 3-phase systems, specific TOV and durations are proposed in Table 11.
The assumptions made for directly earthed neutral systems include some combined effects of
earth faults and load rejection; considering if an earth fault occurs during a load rejection, the TOV
on the healthy phases tends to rise further than it would if the events occurred separately.
Table 11. Common choice for earth fault factor and fault duration
__________________________________________________________________________________________
In some cases, efforts are made to reduce the earth-fault current by selectively earthing the
neutrals of only a few transformers, yet maintaining an effectively-earthed system overall. In such
cases, there is a possibility that some parts of the system may become non-effectively earthed
(i.e. increase in value of ke) for certain periods of time when one or more of the earthed-neutral
transformers are out of service. An earth fault during this time may lead to higher TOV and
subsequent arrester failure if this contingency is not taken into account. Since such occurrences
are rare, it may be justified to accept a risk of arrester failure instead of selecting an arrester with
higher TOV capability and thus a higher protective level.
Special consideration applies to an arrester on the delta tertiary winding of a transformer where
one corner of the delta is permanently connected to ground. In such applications, the continuous
operating voltage applied to the arrester will be the full phase-phase voltage even though the
arresters are connected phase-ground.
Should a considerable amount of harmonics (> 10%) be present in the system, a safety factor of
1.05 (i.e. 5%) is recommended (IEC 60099-5) to account for the increase in peak value of Uca.
However, in systems with short automatic fault-clearance times, a safety factor of 1.0 is often
sufficient nevertheless since the limited duration is normally covered by the TOV characteristic of
the arrester.
The manufacturer should be consulted if abnormal service conditions exist which are outside of
those specified by the Standards: such as ambient temperature below –40 °C or above +40 °C,
frequencies under 48Hz or above 62Hz, presence of heat sources (e.g. furnaces) near the arrester,
etc. Such abnormal service conditions may lead to the need for selection of higher Uc and/or rated
voltage (Ur), unless the arrester has been designed and verified to withstand the specified service
conditions. All ABB arresters, for example, can withstand wider ranges of temperature (–50 °C to
+45 °C) and frequency (15 Hz to 62 Hz), without the need for special consideration.
Factors affecting the TOV capability of an arrester are pre-energy absorbed (i.e. the initial
temperature of the blocks) prior to the application of TOV and the applied voltage following the
TOV.
__________________________________________________________________________________________
The following procedure is suggested to select an arrester with sufficient TOV capability:
• Select a preliminary rated voltage (Ur0) based on Uc, with Ur0 = Uc/0.8
where 0.8 is the “design factor” for MO arresters
• Determine other temporary overvoltages TOV1, TOV2, … TOVn with amplitude and
duration as calculated or estimated, e.g. using the common choice guide in Table 11.
• Consider the possible energy absorption W (in kJ) prior to the TOV and calculate W/Ur0.
For each TOV, determine the minimum required rated voltages Ure, Ur1, Ur2, … Urn by
dividing the determined TOV amplitude by the temporary overvoltage strength factor Tr for
a selected type of arrester for the actual duration of the TOV and the calculated energy
absorption W/Ur0. If the calculated specific energy absorption W/Ur0 is higher than the
value given for the first choice of arrester type, then increase Ur0 or select an arrester type
with a higher energy capability.
o Thus Ure = TOVe/Tre, Ur1 = TOV1/Tr1, Ur2 = TOV2/Tr2, etc
• Select a final rated voltage, Ur, which is the highest of the values Uro, Ure, Ur1, Ur2, etc.
If this is a non-standard rating, choose the next higher rating.
Important parameters affecting the selection of the nominal discharge current are:
• the importance of the protected object
• number of lines connected to the station
• the line insulation
• ground flash density in the area
• line performance with respect to backflashes and shielding failures some spans out from
the station
__________________________________________________________________________________________
To avoid expensive computer studies, a simplified method can be applied as a first attempt to
estimate the arrester stresses caused by switching as per the previous edition of IEC 60099-4. A
simple single-phase model is useful in many cases; although if these calculations reveal high
energies and the need for more qualified studies than had been considered initially, a more
accurate study would be justified. In order to use the simplified method, the parameters for
evaluation according to the previous Line Discharge Class must be determined in some way for
comparison.
If the surge travel time of the line is short compared with one cycle of power frequency and the
line presents a low impedance, Z, the current through the arrester will have a rectangular shape
with a duration equal to twice the travel time T of the wave on the line. The current through the
arrester and its residual voltage at this current are given by the intersection of the relevant
switching surge characteristics and the load line, and can be determined by plotting a load
diagram, as depicted in Figure 37.
Fig. 37
Load Diagram
UL Prospective overvoltage
Z Line surge impedance
Ia Surge arrester current
Ups Surge arrester switching surge
protection level (residual voltage)
__________________________________________________________________________________________
When no such studies are unavailable in any form, a good fallback practice is always to utilize
positive service experience. If, for example, an arrester of a specific Line Discharge Class has
served well, then a new arrester with similar capability should be the first choice. The difficulty is
then limited to comparing the old Line Discharge Class system with the new classification. This
is discussed in more detail in separate document 1HSAT2014-15-001, to arrive at the following
suggested comparative performance for station class arresters:
kJ/kV (U r ) C
2 5 1.1
3 8 2
4 11 2.8
5 15 4
Table 13. Suggested comparison between old and new classification system and given requirements
(reference: 1HSAT2014-15-001)
If the chosen energy or charge capability is not sufficient, the most economical solution is to
increase the arrester rated voltage. If this leads to an unacceptable protection level, then select
another type with a higher energy/charge capability. For very high demands, parallel MO block
columns and/or arresters may be needed to meet the energy requirements. In these cases,
proper and careful matching must be undertaken to ensure sufficiently equal current sharing, as
full current sharing is not necessarily assured with standard arresters.
At lower system voltages (below 245 kV), the energy due to switching will generally be low. At
the same time, less attention is often paid to effective grounding and shielding of such systems.
Hence the design capability will be determined by lightning stresses. A conservative estimate for
the arrester energy capability for lightning surges is obtained in the high current test using a
4/10 µs impulse with standardized peak amplitudes of 100 kA or 65 kA (IEC and IEEE, respectively).
This wave subjects the arrester to high energy during a very short time and hence to a thermal
shock as well. It is worth noting that discharges of the amplitudes stipulated in the tests rarely occur
in reality, and the real impulse durations seen in service may be longer than the stipulated test
impulse duration.
__________________________________________________________________________________________
Hence, it is advantageous to choose an arrester with larger diameter blocks (and consequently
higher discharge capability) for
• areas with high lightning activity
• important installations and apparatus
• lines and stations where grounding or shielding conditions are inadequate
Special consideration may need to be given to protection levels for a very steep time to voltage
crest, e.g. 1µs. For this type of very steep voltage wave, the effect of connection leads as well as
the distance between the arrester and the protected object must be carefully considered in order to
accurately determine the voltage stress on the equipment. Withstand curves can then be plotted
to check that a sufficient safety margin exist.
𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿
• Margin for lightning impulses = �� 𝑈𝑈𝑝𝑝𝑝𝑝
� − 1 � × 100
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
• Margin for switching impulses = �� 𝑈𝑈𝑈𝑈𝑈𝑈
� − 1 � × 100
Margins may typically be acceptable for arresters with inherently low Upl and Ups, and also the
fact that most equipment at present has high LIWV and SIWV. However, depending on the
electrical distance between the arrester and the protected equipment, the margin for lightning
impulses can become reduced, and thus arresters fail to protect equipment that is not in close
vicinity (i.e. within their protection zone). The flexible erection alternatives for polymer arresters
may be of benefit in reducing the distance effect. Additional line-entrance arresters may also
help.
__________________________________________________________________________________________
Fig. 38
Insulation withstand with time for paper and oil insulated power transformers.
Ageing reduces insulation withstand of equipment and thus the protection margin.
v L
a Upl U
h Surge arrester
Fast-fronted overvoltages spread out along a line in the form of travelling waves. When a
travelling wave reaches a point where the surge impedance changes, reflections and refractions
take place. If the surge impedance is considered infinite – for example a transformer winding or
an open circuit breaker – then a total reflection will occur. The positive instantaneous sum of the
resultant oscillations cause the voltage at the remote end to increase step-wise to as much as
double the value of the initial incoming voltage.
__________________________________________________________________________________________
The generally used formula to estimate the voltage increase due to distance effect is:
(2 × 𝑆𝑆 × 𝐿𝐿)
𝑈𝑈 = 𝑈𝑈𝑝𝑝𝑝𝑝 +
𝑣𝑣
where
U voltage at the protected object (kV)
Upl lightning impulse protection level of the arrester (kV)
S steepness of the incoming voltage wave (kV/µs)
L electrical distance between arrester and protected object
including connection leads (a + b) and arrester height (h)
v velocity of wave propagation (m/µs); approximately equal to the velocity
of light 300 m/µs, except for cables for which 150 m/µs may be used
Note! The distance effect reduction does not apply to the Ups margin since the front-time
of a switching surge impulse is longer.
The protection margin will therefore dramatically reduce with increased separation distances, as
well as with increased steepness of the incoming wave. The latter is a function of how close to
the substation the lightning strikes the transmission line and risk for backflash or shielding
penetration.
Steepnesses of 1200 kV/µs and 2000 kV/µs have been well established in Standards and practical
insulation co-ordination studies for HV sub-stations, and are often used as reference surge
steepnesses. Nevertheless, the determined strike rate leads to the choice of actual steepness for
a given application.
This simplified method must be used with caution as it is only an approximation. It does not take
into account any capacitance of the protected object, nor inductance effects nor the initial voltage
at the instant of surge. This simple method may not be sufficient in the case of small margins
between the arrester protection level and the object’s LIWV; whereby more complex computer
modeling may then be necessary. In all cases, the importance of short distances and connection
lead lengths cannot be over-emphasized.
For neutral-ground arresters protecting fully insulated transformer neutrals, the recommended
rated voltage is approximately the maximum system voltage (Us) divided by √3, assuming a
relatively long fault duration. Short or very long fault durations may warrant selection of a different
rated voltage, after taking into account the specific TOV requirements. In addition, special
considerations must be taken for resonant-earthed systems with long radial lines, as a higher
rated voltage may be necessary. Alternative selection criteria may also apply in cases with
unusually low insulation withstand or for neutrals of transformers with non-uniform insulation.
__________________________________________________________________________________________
For neutral-ground arresters specifically, it is a further advantage that the insulation withstand of
the arrester is approximately equal (or even below) that of the transformer neutral so that, in the
unlikely event of a flashover, it should occur in preference at the arrester. Consequently, these
arresters would typically be assembled in the shortest housing possible.
Protection of special electrical equipment such as motors, generators, capacitor banks, etc,
typically require more detailed evaluation than afforded by the simplified approach, and are
therefore beyond the scope of this Guide.
__________________________________________________________________________________________
Note. IEC 60815 subsequently has been split into three parts to cover definitions (-1), porcelain-housings
(-2) and polymer-housings (-3). IEC 60815-1 then introduced the term “unified specific creepage distance”
(USCD) defined as the creepage distance of an insulator divided by the r.m.s. value of the highest operating
voltage across the insulator. This definition differs from that of specific creepage distance (SCD) where the
line-to-line value of the highest voltage for the equipment is used (for a.c. systems usually Us/√3). For line-
to-earth insulation, this definition will result in a value that is √3 times that given by the definition of specific
creepage distance in IEC/TR 60815 (1986). There is thus no physical difference, but the basis for the
specification must be understood and given to avoid misinterpretation.
For porcelain-housed arresters, select the housing to give the desired creepage - generally the
same as for the other equipment in the same location. If the specific creepage demand exceeds
31 mm/kV, a special design may be required. Silicone-housed arresters, being highly
hydrophobic, are better suited for extremely polluted areas than porcelain- or EPDM-housed
arresters. Based on the experience from long-term testing of silicone apparatus insulators, it is
considered possible to reduce the creepage requirements for silicone housings by at least one
step lower specific creepage, i.e. 20 - 30% less than for porcelain/EPDM. The background to this
conclusion is presented in the Cigré document A3-104 “Optimized use of HV composite apparatus
insulators: field experience from coastal and inland test stations”.
IEC 60815-3 furthermore accepts that reduction of creepage distance is possible and permissible
from a pollution withstand or flashover point of view on specific polymeric insulators based on
field trials, test stations or historic data with the same design, materials and electric stress. Hence,
for example, if a specific creepage distance of 31mm/kV (Us) is deemed necessary at a site based
on porcelain-criteria, then 25mm/kV is considered adequate with the silicone used in conjunction
with the ABB’s PEXLIM design. With such reasoning, 31mm/kV SCD would only be necessary
on PEXLIM arresters in the case that the site conditions dictate higher than 31mm/kV for
porcelain.
Mechanical loads on surge arresters can be divided into either static or dynamic loads. Static
loads are those which are applied continuously (e.g. weight of line conductors, normal wind, etc),
whereas dynamic loads are often higher in magnitude, but need only be withstood for short
periods (e.g. short-circuit current forces, gust winds, earthquake, etc.). Consideration should also
be given to the fact that some loads may act alone or in combination.
__________________________________________________________________________________________
The maximum permissible horizontal load for individual forces is calculated as the maximum
moment which the arrester can withstand, divided by the distance between the base of the
arrester and point of the applied force. Loads at the line terminal connections can be considered
to act at the centre of the terminal, whilst wind loads are assumed to act generally about the
arrester’s centre of gravity. For areas with high seismic risk, different specifications and
verification methods exist, and the manufacturer should be consulted to verify the arrester’s
withstand capability. Notably, mechanical strength requirements are different for seismic loads
and it may be permissible to exceed SSL.
In the case of multiple loads acting in combination, the horizontal loads from individual forces
should be used to calculate the vector sum of the bending moments acting about the base, to
determine if the arrester housing can withstand them when applied simultaneously. Importantly,
this calculation should consider realistic combinations of loads (as would be applied in service)
against appropriate safety margins to the arrester’s static and dynamic mechanical strength, and
not simply as a sum of the maximum cantilever load applied at the top terminal.
This is illustrated in Figure 41, which shows that the arrester in question can withstand the
specified loads (safety factor > 1.0) when like loads are correctly considered to act about
appropriate moment arms.
Loads resulting from tensile and compression forces are not usually of concern, as these are
normally limited for standard applications and arrester housings are also typically strong in these
directions. Torsional loading on the arrester is also considered an abnormal service condition,
but may need closer consideration should it exist.
For connecting arresters to the line, a common solution is to use the same conductor as for
current-carrying equipment connected to the same line. However, this is often unnecessarily
large and over-dimensioned for the purpose - the continuous total current through an arrester is
of the order of only a few milli-Amps. The result is undue mechanical loading on the arrester.
Connecting the arresters to the line instead by light, vertical and slack tee-offs, can considerably
reduce the demand for mechanical strength, without requiring significant deviation from common
practice. See Figure 42.
Due to their flexible construction, there may be a visible deflection at the line-end of polymer
arresters under mechanical load. This may ultimately determine the limit of loading which is able
to be applied. However, since polymer arresters are light compared to equivalent porcelain-
housed arresters, they permit innovative erection alternatives which could reduce the loading; for
example suspended or under-hung erection or special bracing.
__________________________________________________________________________________________
TERMINAL LOAD
Static terminal load Fts N 250
Max. dynamic terminal load Ftd N 500
Short circuit force at terminal Fsc= 0.2*Ik 2*(L/r) N 845
Static terminal moment Mts=Fts*H Nm 818
Dynamic terminal load Mtd=Ftd*H Nm 1637
Short circuit load Msc= Fsc*H Nm 2766
Short circuit current Ik kA 65
Conductor clearance r m 3,1
Conductor length L m 3,1
SEISMIC LOAD
Maximum ground acceleration a g 0,3
Damping % 2
Torsion of the insulator C Nm/rad 93000
Moment of inertia (around base) I=(m*H2)/3 kgm2 464
Natural frequency (eigenfrequency) f=(C/I)0,5/2π Hz 2,3
Arrester acc. at cog. (from IEC RRS) aas g 0,81
Arrester acc. at cog. (from peak of RRS) as g 0,85 Fig. 41 Example of
Static analysis factor
Structure factor
k1
k
1,5
1,0
Estimated Loading Table
Max. seismic load Fs =m*as*g*k1*k N 1626
Moment due to seismic loads Ms =Fs *h Nm 2661
70% TERMINAL + SEISMIC + STATIC WIND ##) 0,7*Mts +Ms +Mw*0,12 Nm 3238
Dynamic Safety factor > 1,0 to Md Exceptional load case 3 1,54
§)
For seismic load, higher than SSL may be considered
#)
At 30% of max.dynamic
##)
At 10% of max.dynamic
𝑆𝑆𝑘𝑘
𝐼𝐼 =
�√3 × 𝑈𝑈𝑠𝑠 �
where
I prospective symmetrical short–circuit current (kA)
Sk 3-phase short-circuit power in MVA
at the point where the arrester is to be installed
Us maximum system voltage (kV)
If Sk is not known, the breaking capacity of the associated circuit breaker can be used as a guide
for the short-circuit current.
__________________________________________________________________________________________
Noteworthy is that some users define altitude correction in terms of increased creepage distance.
For surge arresters – and specifically silicone-housed designs – this is not as critical and instead
the required physical clearance in air across the arrester with consideration to its protective
performance is deemed the defining criteria (see also 5.2.4).
Each family utilizes ABB’s own MO resistors that fulfil or exceed the demands for respective IEC
station class. These are in turn given a letter designation based on block type which forms part of
the arrester type.
For ABB surge arresters, the type designation furthermore gives detailed information of the rating
and its application, as indicated below.
Suffix letters:
E Non-standard electrical data
M Non-standard mechanical data
P Parallel columns
H Underhung mounting
L Line arrester
Specific applications may dictate the need for a certain type outside of the common choice.
The respective arresters are nevertheless normally able to be tailored to cater for virtually any
rating or application.
__________________________________________________________________________________________
__________________________________________________________________________________________
Upon arrival to site, the contents of all packages should be checked against the respective
packing lists and any shortages identified. During unpacking, a visual inspection should be made
for any obvious signs of transport damage.
Reference shall be made to the assembly and special instructions provided for details of correct
installation and these shall always be followed and take precedence.
Since ABB undertakes such extensive routine, batch and sample tests on the MO resistors, surge
arresters and counters/monitors (in excess of the requirements of the applicable standards),
additional testing or commissioning checks are not considered warranted or necessary at
installation or before taking ABB arresters and counters into service.
__________________________________________________________________________________________
• Above 200 oC, copper anneals and loses about 70 percent of its strength, thus
compromising its ability to withstand short circuit forces. This increases the risk of
breakage from whipping.
• Above 90 oC, standard PVC softens. With an ambient temperature of 40 oC, this permits
only 50 oC temperature rise. PVC begins to melt at 80 oC temperature rise, at which point
the risk of catching fire is very high.
The required insulation level for this connector is based on foreseen lightning levels. The voltage
drop due to the internal resistance and inductance in the cable itself will be negligible in the case
of lighting impulses and what dominates is the circuit inductance. In the general case, the
lightning surge current generates a magnetic flux in the circuit comprising the insulated base, the
support pedestal and the insulated conductor. The voltage induced is proportional to the magnetic
flux in the closed loop and is little affected by the size of the conductor. For this reason the same
insulation level is usually required for all earth connectors, regardless of their cross sectional area.
• The earth conductor between the arrester and counter should be insulated for at least
5 x L kV (LIWV), where L is the conductor length in metres between the arrester earth
terminal and the surge counter terminal. Note that the maximum permissible length L of
the earth conductor between arrester and surge counter is determined by the LIWV of the
insulated base which the arrester is mounted on as well as the counter itself
• The LIWV of the insulating base and the counter must also withstand this induced voltage;
otherwise it will flashover and the impulse will be earthed through the structure without
passing through the counter.
• Even if the LIWV of the insulated cable is sufficient, this lead must in any case be kept as
short as practicable since its inductance-drop adds to the protection level of the arrester.
__________________________________________________________________________________________
Single-unit porcelain arresters have been mounted horizontally for some years (commonly as
cable terminations) without any obvious distress having been reported. Issues may arise during
actual installation - mainly relating to the logistics of placing a heavy porcelain on a pole - but as
long as care is taken, it seems to have been able to be done successfully. Nevertheless, it is not
generally recommended to mount arresters utilizing hollow insulators (porcelain or composite)
horizontally, since the internal stack of MO blocks is typically held under spring-compression and
there is then the risk that they may become displaced if not kept vertical. Also, inclining the
arrester makes its own weight work against the inherent mechanical strength, since the cantilever
loading capability is reduced by an amount equivalent to the arrester mass times gravity applied
about the centre of gravity. This can significantly reduce the permissible load when heavy and
long porcelain housings are involved.
In contrast, direct-moulded silicone-housed arresters (e.g. PEXLIM) by design are not affected in
the same way as for hollow-insulator arresters. Therefore, these designs of arresters may be
mounted at any angle, regardless of length and number of units. Normal consideration of
mechanical loading nevertheless continues to apply. In particular, since the "self-weight" of the
arrester will apply a permanent moment this needs to be deducted from the permissible loading
(both SLL and SSL). Furthermore, with significant ”self-weight", a polymer arrester may also have
considerable deflection. Even if loads are kept within acceptable limits, it can look banana-shaped
which will likely cause concern for the user; a factor which may restrict its practical use. Hanging
lightweight polymer arresters from the line is another option, which then removes concerns
regarding cantilever loading.
When the arrester is to be installed completely upside down (inverted mounting) on an overhead
structure, this is possible provided the design is arranged during production to ensure, amongst
other matters, that the sheds are "right way up" in order to assist with water runoff. It is therefore
not usually possible to mount a standard arrester upside down. As a rule, standard arresters may
be used for mounting angles from 0 – 90o. For angles between >90 –180o, arresters specially
designed for inverted mounting are to be used.
__________________________________________________________________________________________
Individual designs may be more secure than others; but on the whole, users now have a high
degree of security that the risk is low of a modern arrester failing as a result of an inherent design
breakdown. But this is only part of the story - notwithstanding improvements in design, arresters
do still actually sacrifice themselves on occasion. These “failures” (in actual fact a justifiable
overload) are typically due to an unforeseen event in the system which is beyond the design
parameters used to make the initial selection of a suitable arrester. Breaker failure, stolen or
forgotten earths, mismatch in line switching co-ordination, inappropriate relay setting, etc., can all
place undue stresses on the arrester and lead to its overload. Naturally the arrester will “fail”
short-circuit, and it is expected to do so safely.
The risks are real when using an arrester as a mechanical support – prolonged outages,
subsequent damage, high consequential costs, system instability and unacceptable danger to
public safety. Users need to understand they are accepting such risks by relying on mechanical
strength measured after just one successful short circuit test; especially if it is not representative
of the actual installation.
Taking the bus support option may be seen as an easy fix or even the only way with a retro-fit.
Such a solution should nevertheless be the exception, not the rule and then only after all other
feasible alternatives have been seriously evaluated. For example, if clearance space is the issue,
suspending a lightweight solid design polymer arrester from the bus has been used as a viable
and effective solution; electrically, mechanically and cost-wise.
__________________________________________________________________________________________
A surge arrester does not contain any moving parts or items that can wear out. Consequently
there is nothing to maintain, adjust, correct or repair, which is why there is normally no need to
perform any form of periodical checking or monitoring. In general, a correctly chosen and installed
arrester is regarded as maintenance free during its entire lifetime. A correctly chosen arrester in
this context means that its electrical and mechanical characteristics are matched to actual service
conditions.
10.1 Mean Time Between Failure (MTBF) and Mean Time To Repair (MTTR)
The question of MTBF and MTTR is often asked for substation equipment. However, such an
analysis has no meaning or direct relevance for surge arresters. The design-life of a modern
gapless MO arrester can reasonably be expected to be at least as long as the equipment it is
protecting (nominally accepted to be 30+ years). However, this does not mean that it will
necessarily last as long as the primary plant. It must be remembered that a surge arrester is, in
principle, a sacrificial protective device, designed to operate to protect other electrical equipment.
During the normal course of events, it may need to sacrifice itself at any time to protect the primary
plant. It is therefore often very difficult to distinguish between a failure and a correct operation
as, depending on the nature of the electrical surge, an arrester could "fail" internally during correct
operation. But this should not be considered as a "failure" if it occurs for genuine reasons; rather
it is the function of a surge arrester.
Finally, an arrester which has “failed” must be taken out of service and disposed of according to
local regulations. It cannot be repaired or reworked.
10.2 Cleaning
Periodical cleaning of porcelain-housed arresters is usually only necessary after periods of heavy
marine or industrial pollution. Surge arresters may be washed under voltage (live-washing),
following the same safety regulations as for any other high voltage equipment, plus with the
following additional precautions:
• surge arresters normally employ shorter flashover distance compared to other insulators,
leading to an increased risk for external flashover during the washing
• surge arresters with series connected units must have all units washed simultaneously to
avoid overheating of any unit
Arresters with silicone housing should, in general, not need to be washed at all. Nevertheless, it
is acknowledged that silicone insulators exposed to heavy pollution for long periods may become
discoloured and appear dirty over time. This is as a result of low molecular-weight silicone oils
diffusing to the surface, ultimately encapsulating the pollution layer and making the housing
appear dirty and difficult to clean. This function permits the housing to ultimately recover its
hydrophobicity, even after a temporary loss; a unique feature amongst insulators. Of importance
is that, unlike other types of insulators, this discolouration does not necessarily mean that a
silicone insulator's in-service pollution performance is affected. In fact, cleaning of a silicone
insulator can actually have the disadvantage of washing away the silicone oils deposited on the
surface, thereby reducing its hydrophobicity for a period.
__________________________________________________________________________________________
Regardless of how dirty the insulator appears, what is of interest is whether or not the surface of
the housing is hydrophobic or not. A class scale exists for measuring the degree of
hydrophobicity, and tests can be undertaken for determining the extent to which the surface of
the arrester has become hydrophilic. Seven wettability (hydrophobicity) classes (WC) have been
defined with a value between 1 and 7. WC 1 corresponds to a completely hydrophobic (water-
repellent) surface and WC 7 is a completely hydrophilic (totally wetted filmed) surface. By
definition, a composite insulator can be considered hydrophobic in the range WC 1 - 4. Silicone-
housings typically exhibit WC 1 – 2 when new. In contrast, a porcelain insulator exhibits WC > 5
when clean and new and WC 7 after a time in service, i.e. completely hydrophilic, without the
ability to recover. If desired, this class scale provides a coarse value of the wetting status and is
particularly suitable for a fast and easy check of insulators in the field. Refer IEC TS 62073 for
further details.
Note! This is not a simple test to perform in the field. Fig. 45 Wettability class
However, it can be undertaken on an individual (source IEC TS 62073)
insulator removed from service as a means to
evaluate the pollution performance of silicone
insulators under specific site conditions.
__________________________________________________________________________________________
For system voltages above approximately 100 kV, surge counters are often
installed in series with the arresters. The main reason for the use of surge
counters is to check if a particular transmission line or phase suffers from
an exceptionally high number of overvoltages leading to arrester operation
- lightning faults on a line, for example. If this is the case, some preventative
counter-measures may be necessary to limit the number of surges.
However, surge counters tell only part of the story, as they simply register
the number of surges according to their operating characteristic. The user
therefore has no way of telling the magnitude of the surge and if it was
Fig. 46
significant, nor when it occurred and if it was coincident with a system Surge counter mounted
event. on support stand
__________________________________________________________________________________________
Since the capacitive component of the current dominates so greatly, the total leakage current
measured on a simple mA-meter will be very sensitive to the installation; making interpretation of
the readings difficult. Further, there is no evidence that the capacitive current would change
significantly due to deterioration of the voltage-current characteristic of the surge arrester.
Consequently, measurement of capacitive current cannot reliably indicate the condition of MO
arresters. Although increasing values may be of some use in indicating that cleaning of the
insulators is necessary.
Instead, it is generally recognised (IEC 60099-5) that the only reliable indicator for the condition
of a gapless arrester which can be assessed during normal service is to measure the resistive
component of the leakage current (or estimate it from the 3rd harmonic), and compare it with the
maximum allowable resistive current, as given by the manufacturer, under prevailing service
conditions i.e. temperature and applied voltage. Ageing of the MO resistors will generally cause a
gradual increase of the resistive leakage current with time.
Because of the order of magnitude difference (µA vs. mA), a significant change in the resistive
current would be required before it could be noticed on a milliamp meter. Therefore, special
measuring apparatus is necessary to separate out the two components, and give a reliable
detection method for the analysis of the leakage current through gapless MO surge arresters.
Such devices are ABB Surge Arrester Monitor EXCOUNT-II or EXCOUNT-III and Doble Leakage
Current Monitor LCM.
__________________________________________________________________________________________
The earliest overvoltage protection devices were introduced during the last decade of the
19th Century and consisted of a simple air gap for which the sparkover voltage changed with weather
conditions, i.e. temperature, air pressure and humidity. One major disadvantage with this device
was that its operation led to a power arc and consequent interruption of power supply on systems
having earthed neutral points.
The next significant step in the development was the so called conventional arrester, or gapped
arrester, developed during the 1930’s. The arrester comprised of voltage dependent silicon carbide
(SiC) resistor blocks in series with spark gaps, mounted together in a porcelain housing.
The most advanced gapped SiC arresters in the middle of the 1970’s gave good protection against
overvoltages, but the technique had reached its limits. It was difficult, for example, to design
arresters with several parallel columns to cope with the very high energy requirements needed for
HVDC transmissions. The statistical scatter of the sparkover voltage was also a limiting factor with
respect to the accuracy of the protection levels.
The metal-oxide (also called MO, zinc-oxide or ZnO) surge arrester was introduced in the mid to
late 1970’s and proved to be a solution to the problems which could not be solved with the old
technology. The protection level of a surge arrester was no longer a statistical parameter, but could
be accurately given. The protective function was no longer dependant on the installation or vicinity
to other apparatus - as compared to SiC arresters, whose sparkover voltage could be affected by
surrounding electrical fields. The MO arrester could be designed to meet virtually any energy
requirements by connecting MO resistors in parallel (even though the technique to ensure a
sufficiently good current sharing, and thus energy sharing, between the columns is sophisticated).
The possibility to design protective equipment which could handle extremely high energy stresses
also opened up new application areas; protection of series capacitors, for example.
__________________________________________________________________________________________
Many of the early polymeric designs utilized EPDM rubber as an insulator material, but during the
1990’s more and more manufacturers changed to silicone, which is less affected by environmental
conditions, including UV radiation and pollution.
The operating principle for SiC arresters with passive (non current limiting) gaps and active (current
limiting) gaps differs. For the passive gaps, the overvoltage wave creates an increasing voltage
across the gaps until sparkover occurs and, during a short period of time, an impulse current rushes
through the arrester. Thereafter, the normal power frequency voltage will force a follow current
through the arrester of several hundreds of amperes. Due to the non-linearity of the resistor blocks,
the current is reduced much faster than the voltage, and when the voltage approaches zero, the
current is choked and the arc extinguishes.
Practically the entire voltage is across the blocks, with only some small percentage being across
the gaps as an arc voltage drop. When the current has been reduced to below about 1 ampere,
occurring some electrical degrees before zero, the arc voltage drop is suddenly increased since the
low current cannot support any plasma. The arc is transformed to a corona discharge and is
extinguished some hundred microseconds before the zero crossing.
__________________________________________________________________________________________
A. Stack of spark-gaps
B. Coil
C. Shunt resistor
D. Valve resistor
E. Grading resistor
Fig. 51
Section of an active-gap
arrester
The magnetic field results in an electromagnetic force acting on the arc, which is forced from the
initial ignition point out into a narrow chamber where the arc is lengthened 50 - 100 times. The arc
is cooled against the walls and starts to take up voltage. The resulting voltage reduces the follow
current and, as soon as the momentary value of the power frequency voltage falls below the arc
voltage, the follow current ceases. This is in contrast to a passive gap, which must wait until the
voltage is almost zero before it can interrupt the current.
Voltage distribution for steeper waves is determined by the capacitance of the arrester. The
function of the grading resistors in gapped arresters is to distribute the voltage evenly across the
gaps in the event of relatively slow voltage variations. The sparkover voltage at power frequency
and for switching surges is then determined by these grading resistors. There are two kinds of
grading resistors, those with linear resistance and those with non-linear resistance. Generally,
the sparkover voltages for this frequency range needs to be fairly high to prevent false operations
for normal service voltage variations.
Service under polluted conditions has always been a problem for gapped arresters. The formation
of so called “dry bands” on the porcelain surface under such conditions leads to a disturbed
voltage gradient, which affects the internal gaps by means of coupling capacitance between gaps
and porcelains. As a result, some arresters may then even sparkover at service voltage during
periods of heavy pollution. Repeated sparkover may result in overheating when the gaps fail to
reseal, leading to complete failure of the arrester. Improved reliability under conditions of high
contamination requires a strong grading, which can be achieved with highly non-linear grading
resistors.
__________________________________________________________________________________________
Metal-oxide (MO) resistors, in contrast, represent a high impedance at normal service voltage and
draw only a small leakage current (predominantly capacitive), with the resistive component of the
current in the order of only 50 to 250µApeak (depending on the MO resistor diameter). Such a low
“leakage” is neither dangerous to the MO resistor nor uneconomic for the system. Therefore MO
resistors can be placed directly on voltage, and it is possible to remove the series gaps entirely
from the arrester.
MO resistors have an extremely non-linear, but well defined, volt-amp operating characteristic. The
working principle of a gapless MO arrester is therefore very simple: When an impulse occurs, the
arrester’s impedance reduces via its operating characteristic and subsequently changes over from
conducting a small, predominantly capacitive current to a large resistive current. Due to the
passage of the impulse current, a voltage is consequently built up across the arrester (residual
voltage), the magnitude of which is determined by the volt-amp operating characteristic of the
arrester for the applied impulse current and waveshape. Once the impulse has been dissipated,
the arrester thereafter immediately returns back along its operating characteristic to its non-
conducting state.
Even though a lightning overvoltage causes an impulse current through a gapless MO arrester as
for the gapped arrester, the normal power frequency voltage after the discharge is not high enough
to force a follow current through the arrester. Hence, a MO arrester is only subjected to the energy
from the lightning, in contrast to the SiC arrester, where a large energy contribution is obtained from
the follow current. See Figure 52.
Protection levels for gapless MO arresters depend only on the residual voltages determined by the
operating characteristic for the respective waveshapes and currents and thus are better defined
and more stable compared with gapped types. In general, the protection levels are lower (i.e. better)
than for gapped SiC arresters of equal rated voltage. This improvement is particularly marked when
steep-fronted impulses and switching surges are considered.
Table 18. Summary of the major differences between gapless MO and gapped SiC arresters
__________________________________________________________________________________________
Fig. 52 Comparison in operation of a silicon carbide (SiC) arrester with passive gaps
and a gapless metal-oxide (MO) arrester without series gaps
__________________________________________________________________________________________
Further, when systems expand, there may be a need to upgrade the arresters connected to them;
a fact that is often overlooked. The result is heavier than designed operating duty and increased
failure risk. Arresters manufactured even as late as 1960 to 1970 may not be provided with any
suitable pressure-relief mechanism for safe operation during internal short circuit. Even where
such mechanism exists, it may not function satisfactorily if the short-circuit capacity of the line has
been increased after the original installation and is now higher than the arrester capability. Such
arresters almost certainly would not fulfil today’s tough requirements for short circuit safety and
would fail violently in the event of their malfunction; causing damage to equipment nearby as well
as posing a serious risk of injury to any personnel in the vicinity.
As there are still many gapped silicon-carbide (SiC) surge arresters in service worldwide, it is
worthwhile mentioning what can be done to assess their condition, since aged gapped arresters
can malfunction due to a number of reasons, including:
• sealing failures
• arc erosion
• grading component failures
Monitoring may be undertaken on-line as a first step by scanning the arrester with an infrared
camera to reveal any unusual hot spots.
After the arrester is disconnected from the supply source, additional information can be gained
off-line by the following tests:
• Physically examine the arrester units externally to see if the gaskets have deteriorated or
there is any sign of moisture ingress.
• Megger each unit separately to detect any shorted units. However, when grading
components are present, the readings should not tend to infinity, otherwise a discontinuity
may be suspected in the unit.
• If the grading current of an arrester is known at the time of its manufacture or installation,
this figure can be used to compare with the value after it has been in service for some
time. Considerable deviations from the original recording should motivate further
investigation or replacement of the arrester.
• Perform a spark-over test at power frequency (50 or 60 Hz) and compare the results with
the values obtained during routine tests. If the results are more than +10% from that given
in the data sheet, the unit should be replaced. The sparkover voltage for an arrester must
be measured in a high voltage laboratory to obtain the necessary sensitivity and control,
and thus this test cannot be performed on site.
With consideration to the age and residual life of most gapped arresters, versus the time and cost
to remove them from site, install replacements, perform tests in a HV lab, interpret the results and
then possibly reinstate them in service, many users decide it is better to simply undertake a
replacement program of all installed gapped arresters of a certain age without further analysis.
__________________________________________________________________________________________
With increased focus on system reliability, together with ongoing developments occurring in the
field of overvoltage protection, new and innovative applications are continuing to be found for the
use of surge arresters.
Many of these are, however, quite specialised and require more in depth discussion than is
considered possible within the scope of this Guide. Nevertheless, this section briefly discusses
a number of these topics, and refers the reader to other ABB technical information for further
reading should they be of interest. See References.
The insulation withstand of the surge arrester itself has to be co-ordinated with its own protective
characteristics. The arrester has to be positioned with respect to grounded objects and surge
arresters in adjacent phases, without increasing the total risk for insulation failure. The insulation
withstand properties of surge arresters in a substation can be divided into:
• insulation withstand of the surge arrester itself, including the insulation between flanges
and grading rings, etc.
• insulation withstand between the surge arrester and grounded objects
• insulation withstand between the surge arrester and other equipment connected to the
same phase, e.g. bushings
• insulation withstand between surge arresters in adjacent phases
The insulation withstand should be the only constraint when selecting suitable clearances for
properly dimensioned surge arresters. Any effects which various phase-to-ground and phase-to-
phase clearances may have on the voltage distribution along the MO block column should have
already been accounted for in a well-made design.
The insulation withstand of the surge arrester itself should also have been thoroughly considered
at the design stage. Spacing between metal flanges, as well as spacing between flanges and
grading rings, should be designed to be sufficiently large to withstand overvoltages appearing
during current discharges; at least up to the design altitude (and perhaps more).
a) Phase-to-ground clearance
The phase-to-ground clearance in substations is usually based on the selected standard rated
lightning and switching impulse withstand voltages. International Standard IEC 60071-2, for
example, recommends minimum clearances.
In general, the clearance between a grounded object and a surge arrester should be the same as
the phase-to-ground clearance selected for other high voltage equipment in a substation. If it is
not possible to use the normal phase-to-ground clearance in special applications, a smaller
clearance may be chosen, considering the protective characteristics of the arrester, and after
correction for altitude. However, this is generally only possible if there is a fairly big margin
between the standard rated withstand voltage for a substation and the protective level of the
arresters.
__________________________________________________________________________________________
c) Phase-to-phase clearance
The phase-to-phase clearance for high-voltage equipment in a substation is normally based on
the selected standard rated lightning and switching impulse phase-to-phase withstand voltages.
IEC 60071-2, for example, recommends minimum phase-to-phase clearances. Note that the
normal selection of surge arrester protective levels does not directly protect the phase-to-phase
insulation.
In general, the clearance between surge arresters in adjacent phases should be the same as the
phase-to-phase clearance selected for other high-voltage equipment in the substation. If it is not
possible to use the normal phase-to-phase clearance in a special application of surge arresters,
the minimum clearance with regard to lightning overvoltages can be derived, and should include
altitude correction.
Similarly, the minimum phase-to-phase clearance for arresters with respect to switching
overvoltages should also, if possible, always be based on the selected standard rated switching
impulse phase-to-phase withstand voltage for the substation. If a special application requires a
minimized phase spacing, a favourable electrode configuration established by the grading rings
may permit a reduction of the phase-to-phase clearance in certain cases.
More information on the selection of surge arresters for this application is available in the ABB
Technical Information document “Insulation withstand and clearances with EXLIM and
PEXLIM surge arresters”.
__________________________________________________________________________________________
The normal arresters in the station cannot protect the breaker against this event, and instead a
separate set of arresters on the line side of the breaker are required. Such additional breaker
arresters give the additional benefit of improving the overall overvoltage protection of the station.
__________________________________________________________________________________________
Despite this, meshed networks with rapid re-connection of faulted lines for the most part give
satisfactory operational continuity. Short-time disturbances (around 0.5 seconds) must be
tolerated in radial networks, as well as the voltage drop during the fault time (around 0.1 second)
occurring in the meshed networks.
There are, however, some types of loads where even brief disturbances can have a severe impact
for the on-going process - e.g. steel mills, paper mills, refineries, etc. The cost for such an
interruption, both in terms of value of lost production and the costs to re-start the production, are
unacceptable. In today’s deregulated energy market, such costs will be more visible to the
network operator than before, since the buyer can set high demands on delivery security.
The traditional methods to reduce the number of faults caused by lightning have been:
• installation of shield wires
• improvement of the earthing impedance of the towers
• increasing the insulation level
Unfortunately, implementing these methods gives only marginal improvements of the delivery
security, especially if the earthing conditions are difficult due to a high earth resistivity.
A better alternative to reduce the number of line faults caused by lightning is to install MO arresters
with polymeric insulators in parallel with the line insulators. These line surge arresters (LSA)
normally consist of standard polymer-housed arresters together with a disconnecting device and
fastening equipment for installation on the line itself or on the tower.
The dimensioning of a LSA generally follows the same criteria as for an arrester in a substation.
However, it is of particular importance that the LSA is designed correctly with respect to energy
and TOV capability, since the stresses on the arrester at lightning are highly dependent on the
earthing conditions, presence of shield wires, etc. Similarly, the complete solution must be
physically and mechanically robust, since it is fully exposed to the elements and typically
inaccessible for close inspection or maintenance.
__________________________________________________________________________________________
The final point is perhaps one of the most commonly overlooked. It is promoted that the series
gap will hold against power frequency voltage in the event of an arrester overload; which may be
true. However, the gap cannot be dimensioned to withstand lightning impulses. Hence, a
subsequent lightning strike to the line nearby the failed arrester will cause the gap to fire and a
permanent earthfault will occur, with consequential line outage. Without some kind of signalling
device (rarely provided) this failed arrester will be practically impossible to locate and repeated
flashovers and line trips will occur. The problem will be multiplied in the case that several arresters
along the line have failed.
In contrast to EGLA, an LSA can be suspended from the line or mounted on the tower in a way
that best suits the overall configuration. By so doing, the arrangement can be made to ensure
that, in the event of an arrester overload, adequate clearance is assured to withstand not only
PFWL but also LIWL and SIWL. This is achieved through the clever use and placement of the
self-disconnecting device.
More information on the selection of surge arresters for this application is available in the ABB
brochure “PEXLINK: Transmission-line protection for disturbance-free system operation”.
__________________________________________________________________________________________
Optimum overvoltage control requires correct choice of the resistor value in relation to the source
impedance level, the line length and the line parameters. Although a well-proven technology,
pre-insertion resistors can lead to a number of problems in mechanical design and operation; with
adverse impact on overall system reliability. As robust and efficient alternatives, used either alone
or in combination, the microprocessor-based ABB point-on-wave controller type SwitchsyncTM and
PEXLINK Line Arresters could be substituted instead.
More information on the selection of surge arresters for this application is available in the ABB
Technical Information document “Application guidelines for transmission line switching
overvoltage control” and ABB brochure “PEXLINK: Transmission-line protection for
disturbance-free system operation”.
__________________________________________________________________________________________
It is common practice today to use ”restrike-free” breakers. However, since many banks are
switched on a daily basis, the probability of obtaining high transients associated with capacitor
switching increases. Furthermore, the standardized procedure to verify that the breaker is
restrike-free includes only a limited number of tests. The use of arresters in this application not
only gives protection if a restrike does occur, but also decreases the probability of multiple
restrikes since the trapped charge on the capacitors is reduced.
Generally speaking, capacitor protection by surge arresters has been a difficult task before MO
arresters became available. The high discharge currents and possible energies associated with
an arrester operation at a capacitor bank heavily stressed the spark gaps in a SiC gapped
arrester. The possible high energies could also result in overstressed SiC blocks. Once a
sparkover occurred, the arrester which sparked-over had to discharge the whole energy stored in
the capacitor bank and also carry a power-frequency follow current before a resealing at the next
voltage zero was possible.
With the introduction of MO surge arresters, it is possible to meet any energy demand by simply
paralleling the necessary number of blocks, even if the procedure to ensure current sharing is
quite sophisticated.
Many capacitor banks are operated without surge arresters. However, there are a variety of
beneficial reasons to install arresters:
More information on the selection of surge arresters for this application is available in the ABB
Technical Information document “Guidelines for selection of surge arresters for shunt
capacitor banks”.
__________________________________________________________________________________________
In addition to the old, but still valid, arguments for series compensation, the possibilities to use
adjustable capacitors for load-flow control and balanced loading between parallel lines make
series compensation even more interesting for the future.
An extremely vital component for the series compensation scheme is its overvoltage protection.
Historically, it comprised a single spark gap (for moderate demands on capacitor reinsertion
speed) or a dual spark gap protective scheme (for faster reinsertion or other tougher
requirements). With the availability of MO resistors, the protective schemes have been further
improved by using the varistors in parallel with the spark gaps, and ultimately even without the
spark gaps. This has led to simple and robust protection with ultra-fast re-insertion speeds, low
re-insertion transients and low protection levels.
Modern all-film capacitors have low losses, but their overvoltage withstand capability is less than
that for the old type of paper-film capacitors. This leads to requirements of low protection levels
to obtain an economical capacitor design. Low protection levels, however, may be difficult to
achieve with spark gaps alone, since reinsertion transients can give unwanted gap operations.
With MO resistors, this problem is easily solved and, in addition, the reinsertion of the capacitor
will be instantaneous as soon as the voltage across the capacitor decreases below the conduction
”knee-point” of the MO resistor. The spark gap is used as overload protection for the varistor and
is also usually used to quickly by-pass the capacitor/varistor for internal faults in order to limit the
necessary design energy capability for the varistor. For higher protection levels, it may be
necessary to use two gaps in series. With further improvements in varistor energy capability and
faster by-pass breakers, it is possible in most cases to dispense completely with the spark gaps.
More information on the selection of surge arresters for this application is available in the ABB
Technical Information document “Overvoltage protection of series capacitors”.
Static Var Compensation (SVC) schemes often require arresters with energy capability exceeding
that provided by a single MO block column. However, they are much smaller than those for SC
schemes. Standard surge arresters are normally easily adapted to cope with these requirements
after due consideration to ensure adequate current and energy sharing.
__________________________________________________________________________________________
__________________________________________________________________________________________
IEC 60060-1 High-voltage test techniques. Part 1: General definitions and test requirements
IEC 60068-2-11 Environmental testing - Part 2: Tests. Test Ka: Salt mist
IEC 60068-2-14 Environmental testing - Part 2: Tests. Test N: Change of temperature
IEC 60068-2-42 Environmental testing - Part 2: Tests. Test Kc: Sulphur dioxide test for contacts
and connections
IEC 60071-1 Insulation co-ordination - Part 1: Definitions, principles and rules
IEC 60071-2 Insulation co-ordination - Part 2: Application guide
IEC 60099-1 Surge arresters - Part 1: Non-linear resistor type gapped surge arresters for a.c. systems
IEC 60099-4 Surge arresters - Part 4: Metal-oxide surge arresters without gaps for a.c. systems
IEC 60099-5 Surge arresters - Part 5: Selection and application recommendations
IEC 60507 Artificial pollution tests on high-voltage insulators to be used on a.c. systems
IEC 60815 -1,2,3 Selection & dimensioning of high-voltage insulators intended for use in polluted conditions
IEC 61166 High-voltage alternating current circuit-breakers - Guide for seismic qualification
of high-voltage alternating current circuit-breakers
IEC 61462 Composite insulators - Hollow insulators for use in outdoor and indoor electrical
equipment - Definitions, test methods, acceptance criteria and design
recommendationsIEC TS 62073 Guidance on the measurement of wettability of insulator surfaces
IEEE C62.11 IEEE Standard for Metal-Oxide Surge Arresters for AC Power Circuits (>1 kV)
IEEE C62.22 IEEE Guide for the Application of Metal-Oxide Surge Arresters for
Alternating-Current Systems
IEEE 693 Recommended Practices for Seismic Design of Substations
Cigré Technical Brochure 57 “The paper oil insulated measurement transformer”, 1990
Cigré Technical Brochure 60 “Metal Oxide Surge Arresters in AC Systems”, 1991
Cigré Technical Brochure 63 “Guide to Procedures for Estimating the Lightning Performance of
Transmission Lines”, 1991
Cigré WG33.11 Task Force 6 Application Procedures for Station and Overhead Line Insulation
Coordination, “Flashovers of open circuit breakers caused by lightning
strokes”, 1997
Cigré Technical Brochure 544 “MO surge arresters: Stresses and Test Procedures”, 2013
__________________________________________________________________________________________
[6] S. Vitet, L. Stenström, J. Lundquist. ”Thermal Stress on ZnO Surge Arresters in Polluted Conditions – Part I:
Laboratory test methods”, presented IEEE, PES 1991 T&D Conference and Exposition, Dallas, Texas
September 22-27, 1991
[7] S. Vitet, A. Schei, L. Stenström, J. Lundquist. ”Thermal Behaviour of ZnO Surge Arresters in Polluted
Conditions – Part II: Field test results”, presented IEEE, PES 1991 T&D Conference and Exposition, Dallas,
Texas September 22-27, 1991
[8] S. Vitet, M. Louis, A. Schei, L. Stenström, J. Lundquist. ”Thermal Behaviour of ZnO Surge Arresters in
Polluted Conditions”. Cigré 1994 Session August 30 to September 5, Technical Paper 33-208
[9] L. Stenström, J. Lundquist. ”New Polymer-housed ZnO Arrester for High Energy Applications”. Cigré 1994
Session August 28 to September 3, Technical Paper 33-202
[10] L. Stenström, J. Lundquist, ”Selection, Dimensioning and Testing of Line Surge Arresters”, presented at the
Cigré International Workshop on Line Surge Arresters and Lightning, Rio de Janeiro, Brazil, April 24 -26, 1996
[11] L. Stenström, J. Lundquist, ”Energy Stress on Transmission Line Arresters Considering the Total Lightning
Charge Distribution”, presented at the IEEE/PES Transmission and Distribution Conference and Exposition,
Los Angeles, September 15-20, 1996
[12] C. Neumann, V. Aschendorff, G. Balzer, H. Gartmair, E. Kynast, V. Rees, “Performance of the switched gap of
SF6-HV circuit-breakers stressed by lightning overvoltages”, Cigré 1996 Session, Technical Paper 13-102.
[13] M. Mobedjina, L. Stenström, “Limitation of Switching Overvoltages by use of Transmission Line Surge
Arresters”, Cigré SC-33 International Conference, Zagreb, 1998, Technical Paper P.30
[14] M. Mobedjina, B. Johnnerfelt, L. Stenström, “Design and Testing of Polymer-housed Surge Arresters”,
presented at GCC Cigré 9th Symposium Abu Dhabi, October 28-29, 1998
[15] A.R. Hileman, Insulation Coordination for Power Systems, Marcel Dekker, Inc. 1999
[16] M. Mobedjina, L. Stenström, “Improved Transmission Line Performance using Polymer-housed Surge
Arresters”, presented at CEPSI Seminar, Manila, October 23-27, 2000
[17] L. Gutman, L. Stenström, D. Gustavsson, D. Windmar, W.L. Vosloo. ”Optimized use of HV composite
apparatus insulators: field experience from coastal and inland test stations”. Cigré 2004 Session, Technical
Paper A3-104
[18] R.P.P. Smeets, H. Barts, W.A.Van Der Linden, L.Stenström. ”Modern ZnO surge arresters under short-circuit
current stresses: Test experiences and critical review of the IEC Standard”. Cigré 2004 Session, Technical
Paper A3-105
[19] S. Narita, A. Sawada, H. Watanabe, B. Johnnerfelt, L. Strenström, “Design and testing of polymer-housed
surge arresters with special emphasis on seismic stresses and selection of specific creepage in coastal
areas”, Cigré Tokyo, 2005, Technical Paper A3-108.
[20] J. Taylor, ABB Power Products, Ludvika, Sweden, “Short circuit behaviour – Surge arresters and counters”,
presented at INMR world congress and exhibition, Brazil, May 13 -16, 2007.
[21] L. Stenström, J. Taylor, F. Persson, N.T. Osiptsov, “Installation of LSA on a 400kV double-circuit line in
Russia”, presented at Cigré colloquium “Application of Line Surge Arresters in Power Distribution and
Transmission Systems”, Cavtat 2008
[22] ABB AB, High Voltage Products, Ludvika, Sweden, “Mechanical testing and demands on HV surge arresters”,
June 2008
[23] D. Wu, M. Li, M. Kvarngren, “Uncertainties in the application of atmospheric and altitude corrections as
recommended in IEC standards”, 2009 International Symposium on High Voltage Engineering.
__________________________________________________________________________________________
[25] M. Martin, J. Taylor, M. Popic, “Optimisation of air-insulated substations by consideration of surge arrester
protective characteristic”, presented at International council on large electric systems – Cigré Croatian
National committee, Cavtat 2013
__________________________________________________________________________________________
ABB
High Voltage Products
Surge Arresters
__________________________________________________________________________________________