+ All Categories
Home > Documents > DESIGN AND TESTING OF POLYMER-HOUSED SURGE...

DESIGN AND TESTING OF POLYMER-HOUSED SURGE...

Date post: 09-Jun-2018
Category:
Upload: doancong
View: 217 times
Download: 0 times
Share this document with a friend
19
For presentation at the GCC CIGRÉ 9th Symposium, Abu Dhabi, October 28-29, 1998 1 DESIGN AND TESTING OF POLYMER-HOUSED SURGE ARRESTERS by Minoo Mobedjina Bengt Johnnerfelt Lennart Stenström ABB Switchgear AB, Sweden Abstract Since some years, arresters with polymer-housings have been available on the market for distribution and medium voltage systems. In recent years, this type of arresters have been introduced also on higher voltage systems up to and including 550 kV. However, the international standardisation work is far behind this rapid development and many of existing designs with polymer-housings for high- voltage systems have only been tested according to the existing IEC standard, IEC 99-4 of 1991, which in general only covers arresters with porcelain housings. The existing IEC standard lacks suitable test procedures to ensure an acceptable service performance and life time of a polymer-housed surge arrester. In particular, tests to verify the mechanical strength, short-circuit performance and life time of the arresters are missing. In this report, different design alternatives are discussed and compared and relevant definitions and tests procedures regarding mechanical properties of polymer-housed arresters are presented. Necessary design criteria and tests to verify a sufficiently long life-time as well as operating duty tests to prove the arrester performance with respect to possible energy and current stresses are given. The advantages of silicon insulators under polluted conditions are discussed Finally, this report presents some new areas of applications which open up due to the introduction of polymer-housed arrester designs. One such is protection of transmission lines against lightning/switching surges so as to increase the reliability and security of the transmission system. 1. INTRODUCTION 1.1 SHORT HISTORICAL BACKGROUND Surge arresters constitute the primary protection for all other equipment in a network against overvoltages which may occur due to lightning, system faults or switching operations. The most advanced gapped SiC arresters in the middle of 1970s could give a good protection against overvoltages but, the technique had reached its limits. It was very difficult, e.g., 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. Metal-oxide (ZnO) surge arresters were introduced in the mid of and late 1970s and proved to be a solution to the problems which not could 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 dependent on the installation or vicinity to other apparatus as compared to SiC arresters which sparkover voltage could be affected by the surrounding electrical fields. The ZnO arresters could be designed to meet virtually any energy requirements just by connecting ZnO varistors in parallel even though the technique to ensure a sufficiently good current sharing, and thus energy sharing, between the columns was sophisticated. The possibility to design protective equipment against very high energy stresses also opened up new application areas as, e.g., protection of series capacitors. The ZnO technology was developed further during 1980s and in the beginning of 1990s towards higher voltage stresses of the material, higher specific energy absorption capabilities and better current withstand strengths.
Transcript
Page 1: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

For presentation at the GCC CIGRÉ 9th Symposium, Abu Dhabi, October 28-29, 1998

1

DESIGN AND TESTING OF POLYMER-HOUSED SURGE ARRESTERS

by

Minoo Mobedjina Bengt Johnnerfelt Lennart Stenström

ABB Switchgear AB, Sweden

Abstract

Since some years, arresters with polymer-housingshave been available on the market for distributionand medium voltage systems. In recent years, thistype of arresters have been introduced also on highervoltage systems up to and including 550 kV.However, the international standardisation work isfar behind this rapid development and many ofexisting designs with polymer-housings for high-voltage systems have only been tested according tothe existing IEC standard, IEC 99-4 of 1991, whichin general only covers arresters with porcelainhousings.

The existing IEC standard lacks suitable testprocedures to ensure an acceptable serviceperformance and life time of a polymer-housed surgearrester. In particular, tests to verify the mechanicalstrength, short-circuit performance and life time ofthe arresters are missing.

In this report, different design alternatives arediscussed and compared and relevant definitions andtests procedures regarding mechanical properties ofpolymer-housed arresters are presented. Necessarydesign criteria and tests to verify a sufficiently longlife-time as well as operating duty tests to prove thearrester performance with respect to possible energyand current stresses are given. The advantages ofsilicon insulators under polluted conditions arediscussed

Finally, this report presents some new areas ofapplications which open up due to the introduction ofpolymer-housed arrester designs. One such isprotection of transmission lines againstlightning/switching surges so as to increase thereliability and security of the transmission system.

1. INTRODUCTION

1.1 SHORT HISTORICAL BACKGROUND

Surge arresters constitute the primary protection forall other equipment in a network against overvoltageswhich may occur due to lightning, system faults orswitching operations.

The most advanced gapped SiC arresters in the middleof 1970s could give a good protection againstovervoltages but, the technique had reached its limits.It was very difficult, e.g., to design arresters withseveral parallel columns to cope with the very highenergy requirements needed for HVDC transmissions.The statistical scatter of the sparkover voltage was alsoa limiting factor with respect to the accuracy of theprotection levels.

Metal-oxide (ZnO) surge arresters were introduced inthe mid of and late 1970s and proved to be a solutionto the problems which not could be solved with the oldtechnology. The protection level of a surge arresterwas no longer a statistical parameter but could beaccurately given. The protective function was nolonger dependent on the installation or vicinity to otherapparatus as compared to SiC arresters whichsparkover voltage could be affected by the surroundingelectrical fields. The ZnO arresters could be designedto meet virtually any energy requirements just byconnecting ZnO varistors in parallel even though thetechnique to ensure a sufficiently good current sharing,and thus energy sharing, between the columns wassophisticated. The possibility to design protectiveequipment against very high energy stresses alsoopened up new application areas as, e.g., protection ofseries capacitors.

The ZnO technology was developed further during1980s and in the beginning of 1990s towards highervoltage stresses of the material, higher specific energyabsorption capabilities and better current withstandstrengths.

Page 2: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

2

New polymeric materials, superseding the traditionalporcelain housings, started to be used 1986-1987 fordistribution arresters. At the end of 1980s polymer-housed arresters were available up to 145 kV systemvoltages and today polymer-housed arresters have beenaccepted even up to 550 kV system voltages.

Almost all of the early polymeric designs includedEPDM rubber as an insulator material but during the1990s more and more manufacturers have changed tosilicon rubber which is less affected by environmentalconditions, e.g., UV radiation and pollution.

1.2 DIMENSIONING OF ZNO SURGE ARRESTERS

There are a variety of parameters influencing thedimensioning of an arrester but the demands asrequired by a user can be divided into two maincategories:

• Protection against overvoltages• High reliability and a long service life

In addition there are requirements such as that, in theevent of an arrester overloading, the risk of personalinjury and damage to adjacent equipment shall be low.

The above two main requirements are somewhat incontradiction to each other. Aiming to minimise theresidual voltage normally leads to the reduction in thecapability of the arrester to withstand power-frequencyovervoltages. An improved protection level, therefore,may be achieved by slightly increasing the risk ofoverloading the arresters. The increase of the risk is, ofcourse, dependent on how well the amplitude and timeof the temporary overvoltage (TOV) can be predicted.The selection of an arrester, therefore, always is acompromise between protection levels and reliability.

A more detailed classification could be based on whatstresses a surge arrester normally is subjected to andwhat continuous stresses it shall withstand, e.g.

• Continuous operating voltage• Operation temperature• Rain, pollution, sun radiation• Wind and possible ice loading as well as forces in

line connections

and additional, non-frequent, abnormal stresses, e.g..

• Temporary overvoltages, TOV• Overvoltages due to transients which affect

-thermal stability & ageing-energy & current withstand capability-external insulation withstand

• Large mechanical forces from, e.g., earthquakes• Severe external pollution

and finally what the arrester can be subjected to onlyonce:

• Internal short-circuit

For transient overvoltages the primary task for anarrester, of course, is to protect but it must normallyalso be dimensioned to handle the current through it aswell as the heat generated by the overvoltage. The riskof an external flashover must also be very low.

Detailed test requirements are given in Internationaland National Standards where the surge arresters areclassified with respect to various parameters such asenergy capability, current withstand, short-circuitcapability and residual voltage.

2. IMPORTANT COMPONENTS OF ZNOSURGE ARRESTERS

A ZnO surge arrester for high voltage applicationsconstitutes mainly of the following components Seefigure a.

• ZnO varistors (blocks)• Internal parts• Pressure relief devices (normally not included for

arresters with polymer-housings since these do notinclude any enclosed gas volume. The short-circuitcapability of a polymer-housed arrester musttherefore be solved as an integrated part of theentire design).

• Housing of porcelain or polymeric material withend fittings (flanges) of metal

• A grading ring arrangement where necessary

Page 3: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

3

Line terminal

Cap

Inner insulator

Outer insulator

ZnO blocks

Spacer

Fibreglass loops

Yoke

Base

Figure A:Principal designs of porcelain- and polymer-housed ZnO surge arresters. The most importantcomponent in the arresters is of course the ZnO varistor itself giving the characteristics of the arrester. Allother details are used to protect or keep the ZnO varistors together

2.1 ZNO VARISTORS

The zinc oxide (ZnO) varistor is a densely sinteredblock, pressed to a cylindrical body. The blockconsists of 90% zinc oxide and 10% of other metaloxides (additives) of which bismuth oxide is the mostimportant.

During the manufacturing process a powder isprepared which then is pressed to a cylindrical bodyunder high pressure. The pressed bodies are thensintered in a kiln for several hours at a temperature of1100 °C to 1 200 °C. During the sintering the oxidepowder transforms to a dense ceramic body withvaristor properties (see figure b) where the additiveswill form an inter-granular layer surrounding the zincoxide grains.

These layers, or barriers, give the varistor its non-linear characteristics. Aluminium is applied on the endsurfaces of the finished varistor to improve the currentcarrying capability and to secure a good contactbetween series- connected varistors. An insulating

layer is applied to the cylindrical surface thus givingprotection against external flashover and againstchemical influence.

Figure B: Current-voltage characteristic for a ZnO-varistor.

Page 4: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

4

2.2 INTERNAL PARTS OF A SURGE ARRESTER ANDDESIGN PRINCIPLES FOR HIGH SHORT-CIRCUITCAPABILITY

For all the different types of housings, the ZnO blocksare manufactured in the same manner. The internalparts, however, differ considerably between aporcelain-housed arrester and a polymer-housedarrester. The only thing common between these twodesigns is that both include a stack of series-connectedzinc oxide varistors together with components to keepthe stack together but there the similarities end.

A porcelain-housed arrester contains normally a largeamount of dry air or inert gas while a polymer-housedarrester normally does not have any enclosed gasvolume. This means that the requirements concerningshort-circuit capability and internal corona must besolved quite differently for the two designs.

There is a possibility that porcelain-housed arresters,containing an enclosed gas volume, might explode dueto the internal pressure increase caused by a short-circuit, if the enclosed gas volume is not quicklyvented. To satisfy this important condition, thearresters must be fitted with some type of pressurerelief system.

In order to prevent internal corona during normalservice conditions, the distance between the blockcolumn and insulator must be sufficiently large toensure that the radial voltage difference between theblocks and insulator will not create any partialdischarges.

Polymer-housed arresters differ depending on the typeof design. Presently these arresters can be found in oneof the following three groups:

I. Open or cage design

II. Closed design

III. Tubular design with an annular gas-gap betweenthe active parts and the external insulator

In the first group, the mechanical design may consistof loops of glass-fibre, a cage of glass-fibre weave orglass-fibre rods around the block column. The ZnOblocks are then utilised to give the design some of itsmechanical strength. A body of silicon rubber orEPDM rubber is then moulded on to the internal parts.An outer insulator with sheds is then fitted or mouldedon the inner body. This outer insulator can also bemade in the same process as used for the inner body.

Such a design lacks an enclosed gas volume. At apossible internal short-circuit, material will beevaporated by the arc and cause a pressure increase.Since the open design deliberately has been madeweak for internal overpressure, the rubber insulatorwill quickly tear, partly or along the whole length ofthe insulator. The air outside the insulator will beionised and the internal arc will commutate to theoutside.figure m illustrates this property vividly.

Surge arresters in group II have been mechanicallydesigned not to include any direct openings enabling apressure relief during an internal short-circuit. Thedesign might include a glass-fibre weave woundeddirectly on the block column or a separate tube inwhich the ZnO blocks are mounted. In order to obtaina good mechanical strength the tube must be madesufficiently strong which, in turn, might lead to a toostrong design with respect to short-circuit strength.The internal overpressure could rise to a high valuebefore cracking the tube which may lead to anexplosive failure with parts thrown over a very largearea. To prevent a violent shattering of the housing, avariety of solutions have been utilised, e.g., slots onthe tubes.

When glass-fibre weave, wound on the blocks to givethe necessary mechanical strength, is used, analternative has been to arrange the windings in aspecial manner to obtain weaknesses that may crack.These weaknesses ensure pressure relief andcommutation of the internal arc to the outside thuspreventing an explosion.

The tubular design finally, is designed more or less inthe same way as a standard porcelain arrester butwhere the porcelain has been substituted by aninsulator of a glass-fibre reinforced epoxy tube with anouter insulator of silicon- or EPDM rubber.

The internal parts, in general, are almost identical tothose used in an arrester with porcelain housing withan annular gas-gap between the block column and theinsulator. The arrester must, obviously, be equippedwith some type of pressure relief device similar towhat is used on arresters with porcelain housing.

This design has its advantages and disadvantagescompared to other polymeric designs. One advantageis that is easier to obtain a high mechanical strength.Among the disadvantages are, e.g., a less efficientcooling of the ZnO blocks and an increased risk ofexposure of the polymeric material to corona that may

Page 5: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

5

occur between the inner wall of the insulator and theblock column during external pollution. This latterproblem can be solved by ensuring that the gapbetween the block column and insulator is very largebut this leads to a costly and thermally even worsedesign.

Polymer-housed arresters lacking the annular gas-gapnormally do not have any problem with corona duringnormal service conditions in dry and clean conditions.The design must be made corona-free during suchconditions and this is normally verified in a routinetest. However, during periods of wet external pollutionon the insulator the radial stresses increaseconsiderably. This necessitates that the insulator mustbe free from cavities to prevent internal corona in thematerial which might create problems in the long run.The thickness of the material must also be sufficient toprevent the possibility of puncturing of the insulatordue to radial voltage stresses or material erosion due toexternal leakage currents on the outer surface of theinsulator. The effects of external pollution are dealtwith later on in the paper. See art. 3.2.5.

2.3 SURGE ARRESTER HOUSING

As mentioned before, the housings of the surgearresters traditionally have been made of porcelain butthe trend today is towards use of polymeric insulatorsfor arresters for both distribution systems as well as formedium voltage systems and recently even for HV andEHV system voltages.

There are mainly three reasons why polymericmaterials have been seen as an attractive alternative toporcelain as an insulator material for surge arresters:

• Better behaviour in polluted areas• Better short-circuit capability with increased safety

for other equioment and personnel nearby.• Low weight• Non-brittle

It is quite possible to design an arrester fulfilling thesecriteria but it is wrong, however, to believe that allpolymer-housed arresters automatically have all ofthese features just because the porcelain has beenreplaced by a rubber insulator. The design must bescrutinised carefully for each case.

Polymeric materials generally perform better inpolluted environments compared to porcelaininsulator. This is mainly due to the hydrophobicbehaviour of the polymeric material, i.e., the ability to

prevent wetting of the insulator surface. However, itshall be noted that not all of the polymeric insulatorsare equally hydrophobic.

Two commonly used materials are silicon- and EPDMrubber together with a variety of additives to achievedesired material features, e.g., fire-retardant, stableagainst UV radiation etc. Polymeric materials canmore easily be affected by ageing due to partialdischarges and leakage currents on the surface, UVradiation, chemicals etc. compared to porcelain whichis a non-organic material. Both silicon- and EPDMrubber show hydrophobic behaviour when new. Theinsulator made of EPDM rubber, however, will lose itshydrophobicity quickly and is thus often regarded as ahydrophilic insulator material.

Hydrophobicity results in reduced creepage currentsduring external pollution, minimising electricaldischarges on the surface; thereby reducing the effectsof ageing phenomena. The material can lose itshydrophobicity if the insulator has been subjected tohigh leakage currents during a long time due to severepollution, e.g., salt in combination with moisture. Thesilicon rubber, though, will recover its hydrophobicitythrough diffusion of low molecular silicones to thesurface restoring the original hydrophobic behaviour.The EPDM rubber lacks this possibility completelyand hence the material is very likely to lose itshydrophobicity completely with time.

A safe short-circuit performance is not achieved onlyby using a polymeric insulator. The design must takeinto consideration what might happen at a possiblefailure of the ZnO blocks. This can be solved,depending on the type of design, in different ways asdescribed in article 2.2.

Unfortunately, lack of relevant standardised testprocedures for polymer-housed arresters has made itpossible to uncritically use test methods only intendedfor porcelain designs [1,2]. This has led to the belief,incorrectly, that ”all” polymer-housed arresters,irrespective of design, are capable of carryingenormous short-circuit currents.

The work within IEC to specify short-circuit testprocedures suitable for polymer-housed arresters willbe finalised soon [3]. The test procedures most likelyto be adopted will, hopefully soon enough, clean themarket from polymer-housed arresters not having asufficient short-circuit capability.

Page 6: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

6

The possible weight reduction compared to porcelainhoused arresters can be considerable. As an examplean arrester with porcelain insulator for a 550 kVsystem voltage has a mass of approximately 450 kg. Apolymer-housed arrester for conventional up-righterection, with the same rated voltage, can be designedwith a mass of approximately 275 kg. If suspendedmounting is accepted, the weight can further bereduced to a total mass of only approximately 150 kg!

For long arresters for HV and EHV application, thedesired increase in the mechanical strength of thehousing is obtained by using additional stays ofpolymer material as can be seen in figure c.

Since the polymeric insulator, commonly silicon- orEPDM rubber, does not have the mechanical strengthto keep the ZnO column together, other insulatormaterials must be used in the design. The mostcommonly used material is glass-fibre. There areseveral types of mechanical designs, e.g., cross-winding, tubes and loops.

Two main possibilities exist to combine the glass-fibredesign and the insulator; firstly, the glass-fibre designcan be moulded directly into the rubber insulator andsecondly, the boundary between the glass-fibre and therubber insulator is filled with grease or a gel, generallyof silicon. It is of great importance that no air pocketsare present in the design where partial dischargesmight occur leading to destruction of the insulatorwith time. Penetration of water and moisture must alsobe prevented which sets high requirements on thesealing of the insulator at the metallic flanges andadherence of the rubber to all internal parts in case therubber is moulded directly on the inner design.

2.4 GRADING RINGS

Surge arresters for system voltages approximately 145kV and above must normally be equipped with one ormore metallic rings hanging down from the top of thearrester. The function of these rings is to ensure thatthe electrical field surrounding the arrester is as linearas possible. For very high system voltages, additionalrings are used to prevent external corona from theupper metallic flange and from the line terminal.

3. DESIGN

3.1 DESIGNING FOR CONTINUOUS STRESSES

3.1.1 CONTINUOUS OPERATING VOLTAGE

Denoted as Uc in accordance with the IEC standard,

Figure C: Polymer-housed surge arrester for550 kV system voltage. The surge arrester isdesigned to meet extreme earthquakerequirements in the Los Angeles area (USA).

Page 7: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

7

it is the voltage stress the arrester is designed tooperate under during its entire lifetime. The arrestershall act as an insulator against this voltage. Theentire voltage is across the ZnO varistors and thesemust be able to maintain their insulating propertiesduring their entire lifetime.

The continuous operating voltage for AC surgearresters is mainly at power frequency, i.e., 50 Hz to60 Hz with some percent of superimposed harmonics.For other applications, e.g. HVDC, the waveform ofthe voltage might be very complicated. The voltagemight also be a pure DC voltage. It must be verified,therefore, for all applications that the ZnO varistors areable to withstand the actual voltage under theirtechnical and commercial lifetime which normally isstated to be 20 to 30 years.

The basis for the dimensioning is the result fromageing procedures where possible ageing effects areaccelerated by performing tests at an elevatedtemperature of 115 °C. For porcelain-housed arrestersfilled with air (sometimes nitrogen) it is not necessaryto encapsulate the blocks during the test. Forpolymeric arresters, where the ZnO blocks are in directcontact with rubber, silicon grease or any otherpolymeric material, the ageing test must be madeincluding these additional materials to verify that thereare no negative effects, i.e., ageing of the blocks fromthe other materials.

The normal development of power losses for ZnOvaristors is shown in figure d.

At voltage levels below the knee-point the ZnO blockcan be seen as a capacitor which is connected inparallel to a non-linear resistor. The resistance is bothtemperature- and frequency- dependent.

It is not sufficient just to check the behaviour of theZnO varistor alone. The arrester must be seen as anintegrated unit. The ability of the arrester housing totransfer heat must be considered and adjusted to thepower losses of the ZnO varistors. This considerationmust be made for different service conditions withrespect to voltage, temperature and frequency toensure that the continuous block temperature does notconsiderably exceed the ambient temperature.

If the power losses would increase with time, i.e., theZnO blocks “age”, this must be accounted for in thedimensioning of the arrester.

figure e principally shows how the capability of thearrester housing to transfer heat and the temperature-dependent voltage-current characteristic in the leakagecurrent region of a ZnO varistor results in a working-temperature at a certain ambient temperature andcertain chosen voltage stress (A in the Figure).

An upper maximum temperature also exists (B infigure e) above which the design is no longerthermally stable for a given continuous operatingvoltage. If the temperature would increase above thisvalue due to, e.g., transient or temporary overvoltages,the temperature will continue to increase until thearrester fails. The maximum designated Uc for an

0 100 200 300 400 500 600 700 800 900 1000 1100 1200

Time (hours)

0

0.2

0.4

0.6

0.8

1

1.2

Relative power losses P/Po (Po=power losses after 1.5 hour)

Figure D:Typical power losses during anaccelerated ageing test at 115 °C and appliedvoltage ratio 0.97 times the reference voltage. Notethat the test sample includes the polymer insulatormoulded on to the ZnO blocks.

40 60 80 100 120 140 160 180 200

Varistor temperature - degrees C

0

1

2

3

4

5

Thermal characteristics of housingPower losses at 0.6*UrefPower losses at 0.7*UrefPower losses at 0.8*UrefPower losses at 0.9*UrefRelative power losses

A

B

Figure E: Thermal characteristics of a surge arresterhousing and power losses for a ZnO varistor atdifferent relative voltage stresses (ambienttemperature +40 °C, Uref = reference voltage)

Page 8: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

8

arrester must thus be chosen with respect to possiblepower losses due to ageing, maximum ambienttemperature, estimated energy absorption capabilityfor transient overvoltages and temporary overvoltage(TOV) capability after the energy absorption.

When losses and possible ageing of the ZnO blocksare judged, a consideration of the complete arresterdesign must be made. The local voltage stress along along arrester for high system voltages might deviateconsiderably from the average voltage stress. This, inturn, might lead to local heating of the upper part ofthe arrester and possible ageing of the ZnO blockssubjected to this high voltage.

It is essential, therefore, to distinguish between whatthe ZnO blocks can be subjected to without anyencapsulation and how the design actually can bemade taking into consideration that the ZnO blocks areencapsulated in a long arrester.

To ensure that the maximum stresses does not exceedgiven design criteria, the necessity of a suitable voltagegrading must be considered. This is best accomplishedwith computer programs for electrical fieldcalculations.

3.1.2 VOLTAGE GRADING

During normal operation conditions and operationvoltages the ZnO blocks act like capacitors. Thevoltage across the ZnO blocks, therefore, will bedetermined by the self-capacitance of the blocks as

well as stray capacitance to the surroundings. For along ZnO column, the self-capacitance of the ZnOblocks quickly becomes insufficient to ensure an evenvoltage distribution between the blocks. The surgearrester, therefore, must be equipped with some type ofvoltage grading. This can be achieved by additionalgrading capacitors and/or grading rings. Provision ofgrading rings is the most common way improving thevoltage distribution.

The risk of local heating of the ZnO blocks (hot-spots),with consequent reduced energy absorption capabilityof the arrester, increases if the voltage distribution isnot reasonably uniform along the whole arrester. Typetests in accordance with standards, to verify that theZnO blocks are stable during sufficiently long time,are not valid either if the actual voltage stress on thearrester during actual service is allowed to exceed theapplied voltage stress in the type tests.

An actual surge arrester installation constitutes athree-dimensional problem with three-phase voltagesinvolved together with certain stipulated minimumdistances between phases and to grounded (earthed)objects. All this must be considered when making acalculation. Not to consider the influence of adjacentphases, for example, will lead to an underestimation ofthe maximum uneven voltage distribution by up to 10%.

System voltage 145 kV System voltage 245 kV System voltage 420 kV System voltage 800 kVFigure F: Examples on different grading ring arrangements for different system voltages. Note that the arrestersare not shown to scale.

Page 9: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

9

figure f shows the typical grading ring arrangementfor arresters for different system voltages ( 145 to 800kV).

Without using any components at all to improve thevoltage grading, e.g., grading capacitors or suspendedgrading rings, the voltage across individual ZnOblocks at the line-end of a long arrester will be abovethe knee-point of the current-voltage characteristics,i.e., where the blocks start to conduct large currents.This current is determined by the applied voltage andthe total stray-capacitance of the arrester to earth andcan, for high voltage arresters, be considerable.

Big metallic electrodes, e.g., metallic flanges or ringsto reduce corona without any suspension from itselectrical contact point to the arrester, increases thestray-capacitance to earth amplifying the unevenvoltage distribution.

3.1.3 MECHANICAL DESIGN OF POLYMER-HOUSED

ARRESTERS

Continuous stresses on polymeric materials must beselected with respect to the material behaviour of thepolymer. Many of these characteristics are stronglydependent on temperature and load time. Polymericmaterials becomes softer at higher temperatures with ahigher degree of creeping (cold flowing), at coldtemperatures the material becomes brittle.

It therefore is of great importance that the arresterdesign is tested with different temperature and load

combinations to verify that all possible sealingsoperate adequately in the entire temperature interval.

Composite materials, e.g., glass-fibre joined in amatrix with epoxy or other polymeric materials,exhibit behaviour changes at high loading. The rate ofthis material degradation is determined by temperature,applied force, velocity of the applied force, humidityand the time during which the load is applied. It is notsufficient, therefore, just to dimension the arrester withrespect to its breaking force but consideration mustalso be taken to how the arrester withstands cyclicalstresses.

Up to a certain mechanical load, the fibres of thecomposite material will not break (degrade). This isthe maximum load, defined in terms of the maximumusable bending moment (MUBM), that can be appliedcontinuously in service. This value has very littlespread between different housings of the same type

unlike that for porcelain for which large safety marginsare recommended due to the spread in the breakingmoment.

The MUBM limit is best verified by measuring theacoustic emission to determine what forces might beapplied on the arresters without long-term degradationof the composite materials. The MUBM value shouldbe compared with the “static load” limit for porcelainswhich is 40% of the minimum breaking moment (asdefined in DIN 48113).

At a value slightly above the MUBM, some fibres maystart to break. When enough fibres break, there is asmall change in the mechanical properties whenstressed above MUBM again. A permanent deflectionresults when sufficient number of fibres are broken.Thus small overloads beyond MUBM have nosignificant impact on the service performance.

The new IEC standard, [3] will include a test wherethe arrester is subjected to both thermal as well asmechanical cycling. After the cycling, the arrester isplaced in boiling water for 42 hours where moisture isgiven time and possibility to penetrate the arrester.Electrical measurements are made both before andafter the test sequences to verify that the specimen hasnot absorbed any moisture. If the electricalcharacteristic of the arrester has changed during thetests, the most likely conclusion is that moisture haspenetrated into the design which might imply that thearrester no longer fulfils the original requirements.

Since the polymeric arresters are elastic, temporaryloads, like short-circuit forces and earthquake forces,can be looked upon differently compared to rigidbodies like porcelain insulators. The reason for this isthat the forces do not have time to act fully due to theelasticity of the material and mass inertia, i.e., theforces are spread in time leading to that the arresterwill not encounter any high instantaneous values.These advantages , combined with a design with smallmass participation, have been fully utilised for the 550kV arrester shown in figure c. This arrester withstandsa ground horizontal acceleration of 0.5 gcorresponding to the highest seismic demands as perIEEE/ANSI standards without any problems at all.

3.1.4 INTERNAL PARTS

A low corona (partial discharge, PD) level is desirablefor all apparatus designs intended for high voltageapplications during normal service conditions.Porcelain arresters, though, will have large voltage

Page 10: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

10

differences between the outside and inside of thearrester during external pollution and wetting of theporcelain surface. To fully avoid corona under suchconditions will not give technically and economicallydefensible designs. Instead the internal parts includingthe ZnO blocks must be able to withstand theseconditions.

For polymeric arresters, lacking such annular space inthe design, the voltage difference is entirely across therubber insulator. In order to avoid puncturing of theinsulator the rubber must be sufficiently thick. It isalso very important that the insulator does not haveany air pockets which might give internal coronawhich, with time, may destroy the insulator.

The allowable voltage stress across the material isproportional to the length of the insulator. A longerinsulator, therefore, requires that the thickness of thematerial is proportionally increased with respect to theincrease in length.

Another solution is to reduce the height of theindividual units in a multi-unit arrester, since themaximum voltage across each unit is limited by thenon-linear current-voltage characteristic of the ZnOblocks. In order to verify the withstand against thesetype of stresses, IEC has proposed a long-time testunder continuous operating voltage with continuouslyapplied saltfog [3]. The test must be made on thelongest arrester housing for at least 1 000 hours.

3.2 DESIGNING FOR NON-CONTINUOUS STRESSES

3.2.1 TEMPORARY OVERVOLTAGES

TOV may occur in networks at, e.g., earth-faults. Thisis a voltage which, by definition, is above Uc and

normally will last from some few periods up to someseconds. In certain isolated systems, the duration of anearth-fault may last some days. The TOVs arenormally preceded by a switching surge.

A ZnO arrester is considered to have withstood a TOVif:

a) the ZnO-blocks are not destroyed due to energyunder the TOV i.e. cracking, puncturing orflashover of the blocks does not occur.

b) the surge arrester is thermally stable against Ucafter cessation of the TOV

Since the leakage current through the arrester istemperature-dependent, see also figure b, fulfilling b)above is also dependent on the final blocktemperature. If, for example, due to a switching surge,the arrester already has a high starting temperaturebefore being subjected to a TOV, it will naturally havea lower overvoltage capability.

This is exemplified in figure g showing the ability of aZnO arrester to withstand overvoltages with or withouta preceding energy absorption. The lower curve isvalid for an arrester which has been subjected tomaximum allowable energy, e.g., from a switchingsurge prior to the TOV. The upper curve is valid for anarrester without prior energy duty.

With ZnO arresters the TOV amplitudes are normallyat, or immediately above, the knee-point of the current-voltage characteristic. If the arrester is designedfulfilling the IEC standard, it shall be able to withstanda TOV equal to the rated voltage of the arrester for atleast 10 seconds after being subjected to an energyinjection corresponding to two line discharges as perrelevant line discharge class of the arrester.

The TOV is generally regarded as a stiff voltagesource, i.e., the surge arrester cannot influence thevoltage amplitude. For a dimensioning to fulfil acertain TOV level, the varistor characteristic must bechosen so the current through the arrester, andconsequently the energy dissipation, will not result in atemperature above the thermal instability-point.

The TOV capability given for a certain surge arrestershould always be assumed with a stiff voltage source.However, if this is not the case, the TOV capability ofthe arrester, in general, is significantly higher.

0.1 1 10 100 1000 10000 100000

Duration of TOV in seconds

0.7

0.8

0.9

1

1.1

1.2

1.3Without prior energyWith prior energy = 4.5 kJ/kV (Ur)

TOV Strength factor (Tr)

Uc(MAX)=0.8xUr

Figure G: TOV capability for polymer-housed linedischarge class 3 arrester as per IEC

Page 11: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

11

An important parameter concerning the dimensioningfor TOVs is to accurately control the knee-pointvoltage since the non-linearity of the characteristic isin its extreme in the TOV range. This is best made bydefining a reference voltage close to the knee-point onthe voltage-current characteristics and then, in routinetests, checking that every arresters has a referencevoltage above a guaranteed minimum voltage.

A distinct advantage with polymer-housed arresters isthe superior heat transfer which leads to shortercooling times and possible higher Uc or acceptance ofa higher ambient temperature (above IEC stipulations)as is often the case in tropical desert climates. This isillustrated in figure h. The voltage after the energyinjection was purposely increased to induce a thermalrunaway in the porcelain-housed sample. At the sameconditions, the polymer-housed sample was thermallystable.

A manufacturer is free to assign any data for thearresters. A given arrester with ZnO blocks capable toabsorb high energies, therefore, could be assigned avery high line discharge class with low TOV capabilityor, on the contrary, a low line discharge class withhigh TOV capability.

3.2.2 TRANSIENT OVERVOLTAGES - ENERGY

CAPABILITY - CURRENT WITHSTAND STRENGTHS

A surge arrester may in service be subjected todifferent energy impulses originating from, e.g.,lightning, faults in the net-work and switching of linesand/or capacitor banks.

The arresters must be designed in such a way that theZnO blocks will withstand the energy or currentwithout failing. Additionally the arrester must be ableto withstand the energy thermally, i.e., it must be ableto cool against Uc after an energy absorption.

High voltage arresters are normally designed for aspecific line discharge class. figure i shows relativeenergies in kJ/kV rated voltage for the different linedischarge classes. The intention with the classificationis naturally that a higher class should represent ahigher energy capability for a given arrester. This istrue, however, only if the ratio between the switchingimpulse residual voltage to the rated voltage of thearrester is approximately a factor of two. If the residualvoltage is much higher, the line discharge class willbecome a useless quality measure.

The rated energy is often given in catalogues in kJ/kVrated voltage. Since the ZnO blocks normally are able

to withstand sufficiently higher energies for longertimes, seconds, compared to shorter times, e.g., milliseconds, the expression itself is meaningless if, at the

same time, the shortest time for which the arrester canbe subjected to the given energy is not stated.

A surge arrester may contain a large number of ZnOblocks and if just one of these blocks fails during anovervoltage the probability for a failure of thecomplete arrester is significant. The failure rate for asingle ZnO varistor, therefore, must be extremelysmall to obtain a high reliability of the completearrester. One way to guarantee a low failure rate is toroutine-test all manufactured varistors with an energyconsiderably exceeding the corresponding varistorenergy at the given rated energy for the arrester.

1.0 1.4 1.8 2.2 2.6 3.0

RELATIVE PROTECTIVE LEVEL Ua/Ur

0

1

2

3

4

5

6

7SPECIFIC ENERGY kJ/kV (Ur) (IEC)

CLASS 1

CLASS 2

CLASS 3

CLASS 4

CLASS 5

Figure I: Relative energy stresses for different linedischarge classes according to IEC

0 5 10 15 20 25 30 35

Time (minutes)

60

100

140

180

220Porcelain housingPolymer housing

Temperature (degrees C)

First discharge

Second discharge

Figure H: Oscillogram from an operating duty testshowing the superior cooling properties of polymerhousing.

Page 12: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

12

As mentioned before, a high voltage arrester isnormally designed in compliance with a chosen linedischarge class as per IEC with respect to energy. Fornon-standard stresses, e.g., capacitor discharges orhigh energies due to lightning, the design is normallymade with a lower energy stress per varistor.

The ZnO blocks, apart from withstanding the energyfrom current impulses, also must have a sufficientlyhigh dielectric withstand ensuring that the voltageacross the block will not result in a puncture of or aflashover across the block. To ensure a sufficientinsulation withstand margin for normal stresses, theZnO blocks, including all internal parts in a highvoltage arrester, are dimensioned to withstand currentimpulses with an amplitude of at least 100 kA having awave form of 4/10 µs. Requirements with very highenergy absorption capability cannot be solved byusing ZnO blocks with ever larger volumes but mustbe solved by connecting ZnO varistor columns andarresters in parallel.

To ensure that such designs will operate correctlyduring service, a very careful procedure is required toensure a good current sharing between the blockcolumns connected in parallel. Furthermore, possiblechanges of the block characteristic due to the normalapplied service voltage as well as energy- and voltagestresses must be extremely small.

From protection perspective, it is acceptable that theresidual voltage decreases due to repeated currentimpulses. When parallel connection of ZnO blocks isutilised, the acceptable deviations, however, are muchlower than what the IEC standard permits (+/- 5%).

3.2.3 TRANSIENT OVERVOLTAGES - EXTERNAL

INSULATION

In contradiction to other apparatus, the insulation levelfor surge arresters does not need to fulfil astandardised insulation class since the arrestereffectively will protect its own insulation againstovervoltages. Distance effects need not be considered.Instead, the Standards stipulate a specific safetymargin between the residual voltage of the arrester andthe voltage withstand level of its external insulation.The complete arrester, including possible gradingrings, therefore must be designed to give a reasonablesafety margin against external flashover during anovervoltage.

IEC requires the following minimum externalinsulation levels for an arrester housing:

Arresters with a rated voltage < 200 kVa) For a standard lightning impulse, 1.3 times the

residual voltage at the nominal current with a waveshape 8/20µs

b) For power frequency, 50/60 Hz (peak value), 1.06times the residual voltage at the classifying currentwith a wave shape 30/60µs

Arresters with a rated voltage ≥ 200 kVa) For a standard lightning impulse, 1.3 times the

residual voltage at the nominal current with a waveshape 8/20µs

b) For a standard switching impulse, 1.25 times theresidual voltage at the classifying current with awave shape 30/60µs

The tests with switching impulses and powerfrequency are made as wet tests if the arresters are tobe installed outdoors. With the specified margins tothe protection characteristic of the arrester, anacceptable low risk for external insulation failure isobtained up to an installation altitude of 1 000 mabove sea level as required by IEC.

All distances between the different parts of a surgearrester, e.g., grading rings to flanges or betweenflanges of the individual units or distances to earthed(grounded) equipment and to adjacent phases, must beverified with respect to voltage stress and voltagewithstand. The complete arrester should preferably betested to verify the withstand values even though thepresent IEC standard does not so stipulate [2].

The ZnO blocks cannot be included during these testssince test equipment capable of generating the requiredhigh currents does not exist. In order to emulate actualservice conditions as much as possible, the ZnO blocksshould, for a multi-unit arrester, be replaced by gradingcapacitors. If the ZnO blocks are removed without anyreplacement for the voltage grading, the test result maynot be conservative.

3.2.4 TRANSIENT OVERVOLTAGES - PROTECTIVE

FUNCTION

The arrester shall, for an expected maximum current,limit an overvoltage to a level well below theinsulation withstand level of the protected equipment.

The protective characteristic for a ZnO varistor isslightly dependent on the steepness of the expectedcurrent. figure j shows the characteristic for a specificarrester for the three different current shapes given in

Page 13: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

13

the arrester Standards. As can be noted from thediagram, the protection level for currents having afront time of 1 µs are approximately 10% highercompared to currents with a wave form 8/20 µs orlonger. However, even more important than thismarginal increase, for currents in the µs region, is theeffect of positioning the arrester in relation to theprotected equipment and the length of the connections.There is also an effect due to the arrester height.

In order to obtain an efficient protection against fasttransients, e.g., backflashover close to a substation,large margins, therefore, are required between theprotection level of the surge arrester and the protectedequipment’s insulation level.

ZnO blocks with larger diameter has normally a betterprotection level with maintained overvoltagecapability. A better protection level gives alsoautomatically a better energy capability.

3.2.5 EXTERNAL POLLUTION

External pollution may influence a surge arrester asfollows:

• Possibility of internal corona• External flashover• Heating of the ZnO blocks• Tracking and erosion of insulator (polymer-housed

arresters)

0.1 1 10 100

Current (kA)

70

80

90

100

110

120

130

140Lightning (8/20 micros. current wave) Switch (30/60 micros. current wave)Steep (1/2 micros. current wave)

Max residual voltage in per cent of residual voltage at 10kA 8/20 impulse

Figure J: Protective characteristic for a polymer-housed surge arrester with nominal discharge current20 kA. The protection level is given in % of the 10 kAlevel which is checked in a routine test

The problems for arresters with porcelain housingsinstalled in extremely polluted areas have been solvedby greasing the insulator thus improving the pollutionperformance. The aim of the greasing is to reduce theleakage currents on the insulator surface. Hydrophobic

materials, like silicon rubber, give a similar effect.This is one strong motive why silicon rubber has beenseen as an attractive insulator material.

A common conception is that polymer-housedarresters have a better pollution performance comparedto porcelain. However, a more correct statementshould be that hydrophobic materials have betterperformance in polluted areas due to reduced leakagecurrents. EPDM rubber, that loses its hydrophobicproperties quickly, must be designed in the samemanner as porcelain from pollution point of view.

It is very difficult to avoid internal corona, asdiscussed previously, during severe external pollutionon arresters containing an annular gap between theZnO blocks and the insulator as in the case ofarrangements similar to porcelain-housed arresters.The design of such arresters, therefore, must be able towithstand corona during such occasions.

Some rules-of-thumb for designs like these are:

• "No" corona in dry conditions• Minimise the use of organic materials. When

organic materials are used they must have beenthoroughly tested and subjected to a realisticcorona test

• Prevent the possibility of electrical dischargesdirectly on to the ZnO blocks

Concerning polymer-housed arresters, large radialvoltage stresses may occur between the blocks and theoutside of the insulator during severe externalpollution. It is very important, therefore, that therubber insulator is thick enough to avoid a puncture ofthe insulator. If such a design includes large airpockets or cavities, corona might occur that eventuallyleads to an arrester failure. As mentioned before, asupplement to the IEC standard will most likely beissued with requirements on a 1 000 hours test withcontinuous saltfog to verify the long-term stability ofthe insulation [3].

To avoid external flashover the creepage distance ofthe arrester, i.e. the shed-form and the length, isdesigned in compliance with the same criteria valid forother insulation at the actual site.

Possible thermal stresses are determined by theleakage currents that might be present on the outersurface of the insulator. For porcelain arresters it hasbeen shown that the integral of the leakage current, i.e.the charge, can be regarded as independent of the

Page 14: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

14

creepage distance but it is approximately linearlydependent on the diameter of the porcelain. Aninsulator with a larger diameter thus may give rise tohigher thermal stress during conditions with externalpollution, provided the service conditions otherwiseare the same.

For applications requiring arresters with parallelhousings and several units connected in series, thegeneral rule is that the units should not be connected inparallel except at the top and bottom. This is because,in such an event, the ZnO blocks in one unit couldconduct the external leakage current from all of theparallel connected arresters. Since the ZnO blockshave a negative temperature coefficient in the leakage-current region, a heating of one unit will lead to areduction of the voltage characteristic with subsequentincrease of the current. An increased current throughthe unit leads to higher power losses with increasedtemperature etc. Not even a careful current-sharing testof the arrester units will be of any help below the knee-point of current-voltage characteristic. However, abovethe knee-point the characteristic has a slightly positivetemperature coefficient.

Improvement in a ZnO arrester’s external pollutionwithstand, during otherwise similar conditions, isobtained by:

• Higher rated voltage, i.e., a higher TOV capability• Higher energy capability, i.e., normally a larger

block volume• Improved heat conduction - higher thermal stability

point• Lower power losses at continuous operating

voltage• Lower leakage currents on the insulator surface

Lower leakage currents on the insulator surface isachieved by a hydrophobic surface. figure k showsleakage currents as measured on a porcelain insulatorand a polymeric arrester for 145 kV systems having asilicon rubber insulator. The values are taken from anon-going test at NGC’s test station at Dungeness at theEnglish Channel. As can be noted, the amplitudes ofthe leakage currents are roughly half to a third of thecorresponding leakage currents on the surface of theporcelain insulator during this specific measuringinterval.

All the tests carried out and the operating experiencegained so far indicate that the external creepagedistance for polymer-housed arresters could be shorter

than that for equivalent porcelain-housed arresters byone class (as defined in IEC 815). This would be ofgreat advantage for use in desert climates where theneed for the necessary high creepage leads, at present,to expensive and difficult designs in porcelainhousings.

0

5

10

15

20

25

30

Cu

rren

t (m

A)

Arrester with silicone insulatorPorcelain insulator

Daily maximum currents in a 16 days period at Dungeness test station

Figure K: The leakage currents for 145 kV polymer-housed surge arrester and porcelain insulator atDungeness test station. The leakage current for thearrester includes an internal leakage current ofaround 1 mA. The creepage distance for the polymericarrester is 5 148 mm and 4 580 mm for the porcelaininsulator.

3.3 DIMENSIONING FOR HIGH SHORT-CIRCUITPERFORMANCE

As mentioned previously, the primary duty of a surgearrester, viz. to protect other equipment under allcircumstances, gives a slightly higher risk of failurecompared to other high voltage apparatus, which isaccepted generally.

Since the risk of failure is not negligible, specificrequirements are set on arresters to ensure thatpossible failures will not give consequential damageson other equipment, or, lead to unacceptable risk forpeople. Tests, where the internal parts are deliberatelyshort-circuited, are also required, therefore, in theStandards. From design point-of-view, the aim is toensure that the arrester housing is not scattered after apossible overloading.

In the existing Standard dealing with short-circuittests, IEC 99-1 (being the old surge arrester Standardfor gapped SiC arresters), it is taken for granted that an

Page 15: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

15

arrester fulfilling a certain current class, with respectto short-circuit performance, automatically also fulfilslower current requirements. Recently it has been foundthat this is not always the case. A design might have”grey zones” if only tested with the highest possiblecurrent amplitude. A test made on the longest insulatorused for a specific arrester design, is also considered tocover shorter insulators.

Discussions are going on within IEC on how theinternal short-circuit shall be made before applying ofthe short-circuit current. A thin fuse-wire, arbitrarylocated, might not represent an actual fault-event,especially if the design is non-symmetrical withrespect to arrangement of the pressure relief devices. Ithas been discussed, therefore, to place the wire in alocation where it would represent the worst case fordifferent design types and this requirement will beincluded in the Standard.

How to perform short-circuit tests on polymericarresters, with no internal channels for a pressurerelief, is another question discussed within IEC. Asmentioned previously, it is not possible to uncriticallyapply test methods intended for porcelain arresters onpolymeric designs. To perform tests by arbitrarilyshort-circuiting a polymeric arrester with a fuse-wirelocated alongside the block column, inside the externalinsulator, could result in that unsafe arresters arebelieved to be completely safe.

A suggested revision of the IEC Standard will mostprobably lead to tests on arresters at approximately25%, 50% and 100% of the classifying short-circuitcurrent. How the tests are performed are so far onlydefined in IEC 99-1 but a working group within IEC(IEC TC37 WG4) is working to revise the testprocedure. The present tests shall be made with a highcurrent, 16 kA to 80 kA, as well as with a low current,400 A to 800 A.

The test duration is 0.2 seconds during the high-current test which reflects the time it takes a circuitbreaker to disconnect a fault. To avoid an explosion ofthe arrester housing the internal arc must, in mostcases, be commutated to the outside of the arresterwithin the first half-period of the short-circuit current.Since this time is critical, a certain current amplitude isdefined for the first major loop of current, being 2.6times the prospective symmetrical fault current.For thelow current test, 600 A to 800 A, the current ismaintained until opening of the pressure relief deviceoccurs, which shall take place within 1 second.

Figure L: A polymer-housed arrester prior to a short-circuit test.

Figure M: The same arrester after a short-circuit testat 50 kA sym

Page 16: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

16

The most likely test procedure, according to IEC, willgive two possibilities, two test methods, to obtain aninternal short-circuit. The first method is to provoke ashort-circuit of the ZnO blocks by applying asufficiently high voltage on the arrester leading to anelectrical failure in two to eight minutes where afterthe arrester shall be subjected to the short-circuit test(high current) within five minutes. The secondalternative is to short-circuit the arrester with a thinfuse-wire through a pre-drilled hole between the centreand the periphery of the blocks. This latter method isconsidered to be the worst-case model.

The pictures in Error! Reference source not found.and Error! Reference source not found. show theresults of a short-circuit test at 50 kA , performed inaccordance with the proposed IEC standards.

4. VERIFICATION OF SURGEARRESTER DESIGN

Set requirements on a surge arrester and the design ofthe same are considered to be satisfactory verified byhaving the arrester subjected to the following tests:

• Residual voltage measurement at different currentamplitudes and wave-shapes

• Current impulse withstand tests• Operating duty test• Accelerated ageing test• Artificial pollution test• External insulation test• Short-circuit test• Mechanical test

The above tests are considered to be type tests (designtests) but some of these may also be performed duringthe manufacturing process and/or assembly as a part ofa manufacturer’s quality assurance. The protectivecharacteristic is verified during the various residualvoltage tests.

The reliability is checked through a number ofelectrical and mechanical tests. An important part ofthe electrical tests is the operating duty test in whichan arrester, or a pre-scaled model of the arrester, issubjected to a combination of stresses representinganticipated service stresses that an arrester might besubjected to during its lifetime.The lifetime is finally verified by subjecting the ZnOblocks to an accelerated ageing test procedure.

Within IEC, TC37 is responsible for thestandardisation of surge arresters. The working groupresponsible for the new Standard for gapless metal-oxide arresters, IEC 99-4, is named IEC TC37 WG4.This working group will continue its work also afterpublishing of the new standard. The group shallpropose, among others, a test method for artificialpollution on ZnO arresters, something that still islacking in the new Standard.

In the forthcoming Standard on polymer-housedarresters, the test procedures will differ considerablyfrom previous tests on porcelain designs. A tightnesscheck will e.g., be required to verify that polymericarresters will not absorb moisture [3]. According to thesuggested test procedure, the arrester shall besubjected to both mechanical and electrical tests beforeimmersed in boiling salt water. After the boiling, theelectrical tests will be repeated to verify that thecharacteristic has not changed, something which couldindicate penetration of water.

5. SPECIAL APPLICATIONS OFPOLYMERIC ARRESTERS -LIGHTNING & SWITCHINGPROTECTION OF TRANSMISSIONLINES

5.1 LIGHTNING PROTECTION OF TRANSMISSIONLINES

Transmission lines in the lower system voltage range,70 kV - 245 kV, are often sensitive to lightningovervoltages due to that:

• the insulation withstand is relatively low• the transmission line often lacks shielding wires• the footing impedance of the towers is high• the transmission line lacks a continuous

counterpoise (earth wire)

Despite this, meshed networks with rapid re-connection of faulty lines give satisfactory operationsafety. Short-time disturbances (around 0.5 seconds)must be ignored, however,in radial nets as well as thevoltage drop during the fault time (around 0.1second) occurring also in the meshed nets.

There are, however, some types of loads where eventhe shortest disturbance is of greatest importance;e.g. process industries as steel mills, paper mills andrefineries. For these loads, even a very shortdisruption or voltage drop could lead to unacceptable

Page 17: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

17

interruption of the on-going processes. The cost forsuch an interruption is both the value of lostproduction and the costs to re-start the production.The accumulated sum for these costs can be veryhigh. In a de-regulated energy market such costs willbe more visible to the network operator than before,since the buyer could set new, higher demands ondelivery security.

5.2 SURGE ARRESTERS FOR TRANSMISSION LINEPROTECTION AND THEIR DESIGN

What could then be done to increase the deliverysecurity with respect to faults caused by lightning?The traditional methods to reduce the number offaults caused by lightning have been:

• installation of shield wires• improvement of the earthing impedance of the

towers• increasing the insulation level

Unfortunately, implementing the above gives onlymarginal improvements of the delivery security,especially if the earthing conditions are difficult dueto a high earth resistivity.

A new possibility to reduce the number of line faultscaused by lightning is to install metal-oxide surgearresters with polymeric insulators in parallel withthe line insulators. These transmission line arresters(TLA) normally consist of standard polymer-housedarresters together with a disconnecting device andfastening equipment for installation on the line itselfor on the tower.

Transmission line arresters give complete protectionagainst lightning flashovers for the actual lineinsulator. Insulators in adjacent phases and in othertowers, however, are not protected; why TLA shouldbe installed on all phases on the towers that areintended to be protected.

In reality, TLA are seldom installed throughout anentire line length but only in areas where lightninggives most problems due to exposed position, badearthing conditions etc. Modern localisation systemsfor lightning-storms in combination with traditionalfault statistics are excellent tools to identify towerswhere TLA should be installed to be of the bestpossible use.

The dimensioning of a TLA generally follows thesame criteria as for an arrester in a substation. It is of

great importance that the TLA is designed correctlywith respect to energy capability since the stresses onthe arrester at lightning are highly dependent on theearthing conditions, presence of shield wires etc. The

selection of the energy capability for TLA has beendiscussed at several International conferences duringthe last years [4,5].

5.2.1 PRACTICAL USE OF TRANSMISSION LINE

ARRESTERS

figure n shows how a TLA with polymeric housinghas been installed in a 145 kV transmission line. Thearrester is secured to the line with standard

suspension line brackets. At the bottom of thearrester, a disconnecting device is attached to give anautomatic disconnection of the earth connection inthe event of an arrester failure due to over-stressing.

Figure N: Transmission line arrester withdisconnecting device in a 145 kV-network

Figure O: Transmission line arrester for a 420 kVcompact line installed below insulator strings.Notethe disconnecting device on the high-voltage end atleft.

Page 18: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

18

Another example is given in figure o where anarrester for 420 kV system is installed in a compactline tower.

As an alternative to the disconnecting device, anexternal gap can be used connected in series with thearrester. At a possible arrester failure the operationcan be maintained without a need to disconnect thearrester. An external gap requires, however, a verycareful adjustment to the actual tower type,movements of the line due to wind etc.. TLA withoutseries gaps are preferable from the practical point-of-view since such easily can be designed to fit variousdifferent tower types.

TLA are preferably installed continuously along theline sections which are exposed to the most problemsdue to lightning strokes. Along these protectedsections, the earthing impedance of the towers can beaccepted to be very high without any risk forflashovers. The last towers of the line sectionsprotected by TLA, however, must have adequateearthing conditions otherwise there is a risk thatlightning strokes on the protected section will causeflashovers on adjacent towers on the unprotected linesections. This protection philosophy is illustrated infigure p.

3 4 5 6 7 8 9 10 11

Tower location

1

2

3

4

5

6

7

8

9

10

11

Vo

ltag

e ac

ross

in

sula

tors

-

p.u

.

No arresters in first 2 towers with low TFIArresters in first 2 towers with low TFINormal line insulation strength

High tower footing impedance (TFI). Low TFILow TFI

Figure P :The effect of transmission line arrestersalong line section with high TFI, demonstrating theneed for arresters at the low TFI towers at the endsof the section.

5.3 SWITCHING SURGE CONTROL

For long EHV lines, pre-insertion resistorstraditionally are used to limit switching overvoltagesat closing and reclosing operations. Surge arresters,as a robust and efficient alternative, could be locatedat line ends and along the line at selected points. To

locate arresters along the line has previously not beena practical solution due to the fact that onlyporcelain-housed arresters with high dischargeenergy capability have been available. Now with theintroduction of polymer-housed arresters of IEC linedischarge class 3 and 4 up to and including 550 kVsystems, a very efficient overvoltage control alonglong transmission lines is possible which isillustrated in figure q.

0 20 40 60 80 100

Distance, percentage of line length

1

1.5

2

2.5

3

3.5

4

4.5

5

2 %

o

verv

olt

age

valu

es (p

.u.)

No overvoltage controlSurge arresters at line endsSurge arresters at line ends and two additional locations along the line

Figure Q: Overvoltages phase to ground by three-phase reclosing of 550 kV, 200 km transmission linewith previous ground fault.

6. CONCLUSIONSExisting standards have to be revised to meetnecessary requirements from the manufacturers andusers regarding arrester designs with polymerichousings.

Utilising polymer-housings results in arrester designswith lower weight and better pollution performancethan conventional porcelain arresters. Thermalperformance, in general, will be better which couldbe used to improve protection levels and/oracceptance of higher ambient temperatures aboveIEC stipulation. A high short-circuit capability couldbe obtained as well.

Silicon rubber with necessary fillers so far seems tobe a better insulator material than EPDM.

It is possible to design polymer-housed surgearresters for EHV voltages and to meet very highrequirements on mechanical strength. Special designcan give highly improved seismic performancecompared to porcelain-housed arresters.

Polymer-housed arresters give new applicationpossibilities like transmission line arresters for

Page 19: DESIGN AND TESTING OF POLYMER-HOUSED SURGE …fs1.gongyeku.com/data/default/201212a/20121212144714262.pdf · arrester. In particular, tests to verify the mechanical strength, short

19

limiting lightning and switching surges ontransmission lines.

7. REFERENCES[1] IEC Standard 99-1, ”Non-linear resistor typegapped surge arresters for a/c. systems.”, 1991-05.

[2] IEC Standard 99-4, ”Metal-oxide surge arresterswithout gaps for a/c. systems”, 1991-11.

[3] IEC Committee Draft TC37/154/CD, ”Non-linearmetal-oxide resistor polymeric housed surge arresterswithout spark gaps”, March 1996.

[4] L. Stenström, J. Lundquist, ”Selection,Dimensioning and Testing of Line Surge Arresters”,presented at the CIGRÉ International Workshop onLine Surge Arresters and Lightning, Rio de Janeiro,Brazil, April 24 -26, 1996.

[5] L. Stenström, J. Lundquist, ”Energy Stress onTransmission Line Arresters Considering the TotalLightning Charge Distribution”, presented at theIEEE/PES Transmission and Distribution Conferenceand Exposition, Los Angeles, September 15-20,1996.

Bengt Johnnerfelt (M ‘85) was born in Sweden in1951. He received a M.S. degree in ElectricalEngineering from Chalmers University ofTechnology, Göteborg, Sweden, in 1976, from whichdate he has been with ABB. From 1978, he has beeninvolved in arrester development and is responsiblefor R&D in this field since 1987. He is active in IECTC37, WG4 and several working groups inANSI/SPDC.

Minoo Mobedjina was born in India in 1937. Hereceived a Master’s Degree in Electrical PowerEngineering from Indian Institute of Science,Bangalore, India in 1959. He has been working since1960 with ABB in India and Sweden. Since 1980, hehas been involved with technical marketing of metal-oxide surge arresters all over the world.

Lennart Stenström (M ‘86) was born in Sweden in1951. He received a M.S. degree in ElectricalEngineering from Chalmers University ofTechnology, Göteborg, Sweden, in 1975. From 1975,he has been with ABB, working on metal-oxide surgearrester design, development and application.


Recommended