* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Capacitive current interruption with high voltage air
Survey
Document related concepts
Nanofluidic circuitry wikipedia , lookup
Josephson voltage standard wikipedia , lookup
Valve RF amplifier wikipedia , lookup
Schmitt trigger wikipedia , lookup
Operational amplifier wikipedia , lookup
Electrical ballast wikipedia , lookup
Spark-gap transmitter wikipedia , lookup
Voltage regulator wikipedia , lookup
Resistive opto-isolator wikipedia , lookup
Power electronics wikipedia , lookup
Current source wikipedia , lookup
Power MOSFET wikipedia , lookup
Switched-mode power supply wikipedia , lookup
Opto-isolator wikipedia , lookup
Surge protector wikipedia , lookup
Transcript
Capacitive current interruption with high voltage air-break disconnectors Chai, Y. DOI: 10.6100/IR728755 Published: 01/01/2012 Document Version Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: • A submitted manuscript is the author’s version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher’s website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication Citation for published version (APA): Chai, Y. (2012). Capacitive current interruption with high voltage air-break disconnectors Eindhoven: Technische Universiteit Eindhoven DOI: 10.6100/IR728755 General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 14. May. 2017 CapacitiveCurrentInterruptionwith HighVoltageAir‐breakDisconnectors PROEFSCHRIFT terverkrijgingvandegraadvandoctoraande TechnischeUniversiteitEindhoven,opgezagvande rectormagnificus,prof.dr.ir.C.J.vanDuijn,vooreen commissieaangewezendoorhetCollegevoor Promotiesinhetopenbaarteverdedigen opwoensdag14maart2012om16.00uur door YajingChai geborenteHubei,China Ditproefschriftisgoedgekeurddoordepromotor: prof.dr.ir.R.P.P.Smeets Copromotor: dr.P.A.A.F.Wouters ThisprojectwasfundedbytheDutchMinistryofEconomicAffairs, AgricultureandInnovationinanIOP‐EMVTprogram. AcataloguerecordisavailablefromtheEindhovenUniversity ofTechnologyLibrary. ISBN:978‐90‐386‐3097‐7 ToMilošandRuirui Promotor: prof.dr.ir. R.P.P. Smeets, Eindhoven University of Technology/KEMA Testing,Inspections&Certification Copromotor: dr.P.A.A.F.Wouters,EindhovenUniversityofTechnology Corecommittee: prof.ir.L.vanderSluis,DelftUniversityofTechnology prof.dr‐ing.V.Hinrichsen,DarmstadtUniversityofTechnology prof.dr.E.Lomonova,EindhovenUniversityofTechnology Othermembers: dr.D.F.Peelo,DFPeelo&Associates prof.ir.W.L.Kling,EindhovenUniversityofTechnology prof.dr.ir.A.C.P.M.Backx(chairman),EindhovenUniversityofTechnology Summaryi Summary As low‐cost switching devices in high voltage electrical power supply system disconnectorsbasicallyhaveaninsulationfunctiononly.Nevertheless,theyhavea very limited capability to interrupt current (below one Ampere), e.g. from unloaded busbars or short overhead lines. The present study is a search for possibilitiestoincreasethecurrentinterruptioncapabilitywithauxiliarydevices interacting with the switching arc. In this project the state of the art of disconnectorswitchingisinvestigatedandaninventoryispresentedofmodelsof the free burning arc in air. A series of experiments were arranged at different laboratories.Theswitchingarcandtheinterruptionprocessarestudiedindetail through electrical and optical measurements during the switching process for a disconnector with (without) auxiliary devices under high voltage (0.5 – 30A, 300kV)conditions.Threeoptionsforauxiliarydeviceswereinvestigated:(i)arc coolingbyforcedairflow;(ii)fastinterruptingbyhigh‐velocityopeningcontacts; (iii) reduction of arc energy by adding resistive elements. Finally, a qualitative description is provided on the physical nature of the arc and how the evaluated methodsaffectthearccharacteristics.Allresultsareobtainedbyanalysisofhigh‐ resolutionmeasurementofarccurrent(includingallrelevanttransients),voltages acrossthedisconnectorandhigh‐speedvideoobservation. It was found that, depending on the current to be interrupted, the interruption processisgovernedbythedielectricand/orthermalprocesses. In the dielectric regime, the interrupted current is low (roughly below 1A) and theswitchingarcischaracterizedbyahighrateofrepetitionofinterruptionsand re‐strikesthatonlyceaseafterasufficientgapspacinghasbeenreached.There‐ strikesinteractseverelywiththecircuitryinwhichthedisconnectorisembedded, exciting transients in current and voltage with frequencies up to the megahertz range.Highovervoltagescanbegenerated.Theirmagnitudescanbelimitedbya properchoiceofthecapacitanceatsupplysideofthedisconnector.Thearc‐circuit interaction has been studied and relevant processes have been modelled and verifiedbyexperimentsinfull‐powertest‐circuits. In the thermal regime, the switching arc behaves less vehemently, interrupting andre‐ignitingbasicallyoccurateverypowerfrequencycurrentzero.Becauseof thepresenceofsufficientthermalenergyintheswitchinggapalongthearcpath, the voltage to re‐ignite the arc is limited, and the arc‐circuit interaction is less pronounced.Thoughnotproducingverysevereovervoltages,thearcdurationis longerandthecurrentmaynotbeinterruptedateverycurrentzerocrossing.The ultimate thermal regime is reached when the arc continues to exist after power frequencycurrentzerowithoutanyappreciablevoltagetore‐ignite.Thissituation iiSummary mustbeavoidedbecausearcinggoesonuntilahigherlevelbreakerinterruptsthe current. Before this, the arc can reach far away from its roots and can greatly reduceinsulationclearance. Themainfactorsinfluencingtheinterruptionperformancearethelevelofcurrent tobeinterrupted,thesystemvoltage,theratioofcapacitancesatbothsidesofthe disconnector and the gap length. These factors influence the energy supplied to the arc upon re‐strike. This energy extends the arcing time by lowering the breakdownvoltage.Ithasbeenobservedthatthearcinitsthermalmodealways re‐ignitesinitsformertrajectory.Keytotheinterruptionprocessisthereduction ofbreakdownvoltageinthispath,createdbyhotgasesremainingfromtheformer arc. The existing breakdown models are reviewed in order to understand the influenceofhightemperatureaironthebreakdownprocess. Based on the observed arc behaviour, various methods have been researched to increasetheinterruptioncapability. The most successful methods are those that remove the residual (partially) ionized air from the arc path. Experiments were carried out to demonstrate the effectiveness of air flow directed into the arc’s foot point. A substantial gain in interruption capability is demonstrated, but at the cost of generating re‐ignition transients at a very rapid succession. Specifically, the experiments showed that 7.5Acouldbeinterruptedsuccessfullyat90kVrmsvoltagewithashorterarcing duration(afactorof0.5wasobserved)thanwithoutairflow.Withapplicationof air flow, the frequency of re‐ignitions occurring, and the breakdown voltage are muchhigherthanwithoutairflow. Anothermethod,theassistanceofanauxiliaryswitchabletoproduceaveryfast opening,wasalsosuccessful.Herein,thearcisforcedmechanicallyintoambient coolair,thusavoidingaccumulationofthermalenergyinthearcpath.Specifically, itcaninterruptcurrentsupto7Aat100kVrmssafelyand9Aat90kVrmsinthe experiments with arcing time only a few tens of milliseconds instead of a few seconds. The arc exhibits a "stiff" (linear) character instead of the "erratic" (randomlymoving)arcmodewithadisconnectoralone.Thismethodreducesthe numberofre‐strikes. The possible influence of energy absorbing elements (resistors) is investigated through circuit modelling, supported by some laboratory experiments. Other methods, such as the application of series auxiliary interrupting elements (vacuum,SF6interrupterandablationassistedapproaches)havebeenevaluated. From the practical point of view, the auxiliary fast‐opening interrupter is recommendedduetoitseconomic,simpleandeffectivemerits.Otherapproaches have certain disadvantages. The method with air flow needs a complex constructioninordertointroducethecompressedairflowintothedisconnector, Summaryiii and the hazard for nearby equipments from the overvoltages caused by the interruption is greater. The method of inserted resistor requires very expensive arrangement. Regarding the application of auxiliary interrupters, vacuum interrupters have to be applied in considerable numbers in series and SF6 interrupters have good performance but at very high cost. An ablation assisted approachseemslesspromisingbecausetheleveloftheinterruptedcurrentistoo lowtobeeffective. ivSummary Samenvattingv Samenvatting Scheidingsschakelaars, of kortweg "scheiders", zijn relatief eenvoudige schakelaars in hoogspanningsnetten die in principe enkel een isolerende functie hebben.Nietteminbezittenzeeenzeerbeperktvermogenstromen(totca.1A)te onderbreken, zoals bijv. afkomstig van onbelaste railsystemen of korte hoogspanningslijnen. Deze studie bevat een onderzoek naar mogelijkheden het stroomonderbrekendvermogenvanscheiderstevergrotenmethulpmiddelendie direct ingrijpen in de schakelende lichtboog. Om te beginnen is de stand van de techniekophetgebiedvanschakelenmetscheidersonderzochtenerwordteen overzicht gegeven van het modelleren van vrij brandende lichtbogen in lucht. Tevensiseenserieexperimentenuitgevoerdindiverselaboratoria.Hierinzijnde schakelende lichtboog en het onderbrekingsproces in detail bestudeerd door middel van elektrische en optische metingen gedurende het schakelen onder hoogspanning (0.5 – 30A, 300kV) met en zonder hulpmiddelen. Drie mogelijk hulpmiddelen zijn onderzocht: (i) koeling van de lichtboog door beblazing met eengeforceerdeluchtstroom;(ii)onderbrekingmetzeersnellecontactseparatie snelheid; (iii) reductie van lichtboog energie door toevoeging van resistieve componenten. Tot slot wordt een kwantitatieve beschrijving gegeven van de lichtboog en hoe de boven beschreven methoden ingrijpen in diverse karakteristiekenvandelichtboog. Alle resultaten zijn verkregen met behulp van metingen met hoge resolutie van lichtboog stroom en – spanning (inclusief hun transiënten) alsmede door observatiemetsnellevideotechnieken. Afhankelijk van de grootte van de te onderbreken stroom, wordt het onderbrekingsprocesgekenmerktdoordiëlektrischeen/ofthermischeprocessen. Inhetdiëlektrischeregimeisdeteonderbrekenstroomklein(ruwwegbeneden 1A) en de scheider lichtboog wordt gekenmerkt door een zeer snelle opeenvolging van onderbrekingen en herontstekingen. Dit processtopt wanneer voldoendecontactafstandisbereikt.Als gevolgvandeherontstekingeniser een heftige wisselwerking met het elektrische circuit waar de scheider deel van uitmaakt. Herontstekingen wekken daarbij transiënten op in zowel stroom als spanning met frequenties tot enkele megahertz. Hoge overspanningen kunnen hierbij optreden. De hoogte hiervan kan worden beperkt door een juiste keuze van capaciteit aan de voedende zijde van de scheider. De wisselwerking tussen lichtboog en circuit is bestudeerd; de relevante processen zijn gemodelleerd en getoetstmetexperimenteninhoogvermogenbeproevingscircuits. In het thermische regime gedraagt de lichtboog zich minder heftig, onderbreekt de stroom en herontsteekt in principe pas op iedere nuldoorgang van de netfrequente stroom. Vanwege de aanwezigheid van voldoende thermische viSamenvatting energieinhetlichtboogpadtussendecontactenisdespanningnodigomdeboog te herontsteken beperkt en daarmee is de wisselwerking tussen lichtboog en circuit minder uitgesproken. Hoewel de overspanningen beperkt zijn, is de lichtboogduur langer en wordt de stroom niet bij iedere netfrequente nuldoorgangonderbroken.Demeestuitgesprokenthermischeverschijningsvorm is die waarin (bij verdere verhoging van de stroom) de lichtboog bij elke nuldoorgang herontsteekt zonder meetbare spanning. Deze situatie dient vermeden te worden omdat de lichtboog in deze situatie niet meer dooft en de stroom enkel nog door een vermogensschakelaar onderbroken kan worden. Hieraan voorafgaand kan de lichtboog ver uitwaaieren en de isolatie afstand tot andere,onderspanningstaandedeleninhetstation(tezeer)verkleinen. Debelangrijkstefactorendiehetonderbrekingsprocesbeïnvloedenzijndehoogte van de te onderbreken stroom, de netspanning, de verhouding van capaciteiten ter weerszijden van de scheider en de momentane contactafstand. Deze parameters bepalen de energie die aan de lichtboog wordt toegevoerd op het momentvanherontsteking.Dezeenergieverlengtdelichtboogduurdoordatdeze dedoorslagspanningreduceert.Uitwaarnemingblijktdatinhetthermischregime delichtboogherontsteektinhetvoormaligelichtboogpad.Hieruitwordtafgeleid dat de essentie van het onderbrekingsproces de reductie van doorslagspanning langs dit pad is. Deze reductie wordt veroorzaakt door hete gassen en rest‐ ionisatieafkomstigvande(vorige)lichtboog.Voorhetbegrijpenvandeprocessen die leiden tot reductie van doorslagspanning bij temperatuur verhoging van de luchtzijndebestaandedoorslagmodellenbestudeerd. Op basis van de boven beschreven waarnemingen van het onderbrekingproces zijn diverse methoden onderzocht die kunnen leiden tot vergroting van het stroomonderbrekendvermogenvanscheiders. Het meest succesvol zijn methoden die de gedeeltelijke geïoniseerde lucht verwijderen uit het voormalige lichtboog pad. Experimenten zijn uitgevoerd waarindeeffectiviteitbestudeerdwordtvanbeblazingvandevoetpuntenvande lichtboog met koude lucht. Een significante verhoging van stroom onderbrekend vermogen wordt inderdaad vastgesteld, echter ten koste van de generatie van zeer snel opvolgende en hoge herontstekingen die steile spanningfronten genereren.Deexperimententonenaandatbijeenfase‐aardespanningvan90kV stromen tot ca. 7.5A succesvol onderbroken worden met een 50% kortere lichtboogduurdanzonderbeblazing. Een andere methode bestaande uit het gebruik van zeer snel opende hulpcontacten om daarmee in zeer korte tijd een grote contactafstand te realiseren, blijkt eveneens succesvol. Op deze wijze worden de lichtboog voetpunten zeer snel naar koele lucht getrokken waardoor een te grote thermischeenergiedichtheidvoorkomenwordt.Opdezemanierkunnenstromen van 7A (bij 100kV) en 9A (bij 90kV) onderbroken worden, terwijl de Samenvattingvii lichtboogduur slechts enkele tientallen milliseconden bedraagt in plaats van seconden bij toepassing van conventionele contacten. De uiterlijke verschijningsvorm van de boog verandert hierbij van "dansend" (opwaarts bewegend met vele uitstulpingen) naar "strak" (rechtlijnig brandend als kortste verbinding tussen de contacten). Het aantal herontstekingen blijft hier mee ook beperkt. Ookdemogelijkeinvloedvanenergieabsorberendeelementen(weerstanden)is onderzochtmetmodellering,ondersteundmetenigelaboratoriumexperimenten. Alternatieven, zoals het gebruik van serie geschakelde onderbrekers (vacuüm‐, SF6 onderbrekers en onderbreking gebaseerd op ablatie) zijn geëvalueerd uit de literatuur. Ter vergroting van het stroom onderbrekend vermogen wordt vanuit praktisch oogpuntdemethodemetdesnellehulpcontactenaanbevolenvanwegeprestatie, eenvoud en prijs. Andere methoden hebben duidelijke nadelen. Beblazing met lucht vereist een ingewikkelde constructie om gecomprimeerde lucht in de nabijheidvandeschakelendecontactentebrengenenheeftalsbijkomendnadeel degeneratievansteilespanningsfrontentijdensdeonderbreking.Hetaanbrengen vanweerstandenvereistdureaanpassingen.HulponderbrekersinvacuümofSF6 gasiseffectiefmaarkostbaarenconstructiefingewikkeldvanwegedehoogtevan de spanning die serie schakeling van elementen vaak nodig maakt. Oplossingen met behulp van ablatie zijn minder kansrijk omdat de stroom hiervoor waarschijnlijktelaagis. viiiSamenvatting Contentsix Contents SUMMARY..................................................................................................................................I SAMENVATTING.....................................................................................................................V CONTENTS..............................................................................................................................IX CHAPTER1...............................................................................................................................1 HIGHVOLTAGEAIR‐BREAKDISCONNECTORS.............................................................1 1.1DEFINITIONOFDISCONNECTORS..................................................................................................1 1.2TYPEOFDISCONNECTORS..............................................................................................................1 1.3INTERRUPTINGCURRENTWITHDISCONNECTORS......................................................................3 1.4STANDARDIZATIONSTATUS...........................................................................................................4 1.5OBJECTIVEOFTHESIS.....................................................................................................................5 CHAPTER2...............................................................................................................................9 LITERATUREREVIEW...........................................................................................................9 2.1AWARENESSOFTHECAPACITIVECURRENTINTERRUPTIONWITHDISCONNECTORS...........9 2.2FUNDAMENTALASPECTSOFCAPACITIVECURRENTINTERRUPTION....................................14 2.3TRANSIENTSCAUSEDBYCAPACITIVECURRENTINTERRUPTION..........................................16 2.4APPROACHESTOENHANCETHEINTERRUPTIONCAPABILITY...............................................17 2.5ARCMODELSRELATEDTOCURRENTINTERRUPTIONWITHADISCONNECTOR..................26 2.6CONCLUSION.................................................................................................................................29 CHAPTER3............................................................................................................................35 BASICCIRCUITANALYSIS.................................................................................................35 3.1THEINTERRUPTIONPROCESS.....................................................................................................35 3.2THREE‐COMPONENTSANALYSIS................................................................................................37 3.3SIMULATIONOFRE‐STRIKES.......................................................................................................41 3.4RE‐STRIKETRANSIENTSINDISTRIBUTEDELEMENTLOADCIRCUITS...................................43 3.5CONCLUSION.................................................................................................................................45 CHAPTER4............................................................................................................................47 EXPERIMENTALSETUP.....................................................................................................47 4.1HIGHVOLTAGESOURCE...............................................................................................................47 4.2MEASUREMENTSYSTEM..............................................................................................................51 4.3DISCONNECTORMAINBLADESOPENINGVELOCITY................................................................57 4.4CONCLUSION.................................................................................................................................59 xContents CHAPTER5............................................................................................................................61 CURRENTINTERRUPTIONMECHANISM.....................................................................61 5.1EXPERIMENTALOBSERVATIONS................................................................................................61 5.2ARCINTERRUPTIONCHARACTERISTICS....................................................................................69 5.3DISCONNECTORCONTACTSSPACING.........................................................................................75 5.4RE‐STRIKEVOLTAGE....................................................................................................................76 5.5ELECTRICALARCCHARACTERISTICS.........................................................................................80 5.6ENERGYINPUTINTOTHEARC....................................................................................................82 5.7TRANSIENTSUPONRE‐STRIKE...................................................................................................85 5.8CONCLUSION.................................................................................................................................88 5.9RECOMMENDATIONFORSTANDARDIZATION...........................................................................90 CHAPTER6............................................................................................................................91 INTERRUPTIONWITHAIRFLOWASSISTANCE.........................................................91 6.1EXPERIMENTALSETUP................................................................................................................91 6.2EFFECTOFAIRFLOWONARCING...............................................................................................92 6.3INTERRUPTIONDATAANALYSIS.................................................................................................97 6.4CONCLUSION...............................................................................................................................103 CHAPTER7..........................................................................................................................107 HIGH‐VELOCITYOPENINGAUXILIARYINTERRUPTER........................................107 7.1OPERATINGPRINCIPLE..............................................................................................................107 7.2EXPERIMENTALSETUP..............................................................................................................108 7.3OVERVIEWOFTHEEXPERIMENTALRESULTS........................................................................109 7.4RE‐STRIKEVOLTAGE..................................................................................................................114 7.5ENERGYINPUTINTOTHEARC..................................................................................................117 7.6CONCLUSIONANDRECOMMENDATION...................................................................................120 CHAPTER8..........................................................................................................................123 INTERRUPTIONWITHINSERTEDRESISTORS.........................................................123 8.1SERIESRESISTOR........................................................................................................................123 8.2PARALLELRESISTOR..................................................................................................................128 8.3DISCUSSION.................................................................................................................................130 CHAPTER9..........................................................................................................................133 PRE‐BREAKDOWNPHENOMENAINDISCONNECTORINTERRUPTION...........133 9.1CLASSICALMODELLINGOFBREAKDOWN................................................................................133 9.2PRE‐BREAKDOWNCURRENTINDISCONNECTORINTERRUPTION.......................................137 9.3DISCUSSION.................................................................................................................................138 Contentsxi CHAPTER10........................................................................................................................141 CONCLUSIONSANDRECOMMENDATIONSFORFUTURERESEARCH...............141 10.1CONCLUSIONS...........................................................................................................................141 10.2PROPOSEDFUTURERESEARCH..............................................................................................146 PUBLICATIONSRELATEDTOTHISWORK................................................................147 NOMENCLATURE...............................................................................................................149 ACKNOWLEDGEMENT.....................................................................................................153 CURRICULUMVITAE........................................................................................................155 xiiContents HighVoltageAir‐breakDisconnectors1 Chapter1 HighVoltageAir‐breakDisconnectors 1.1Definitionofdisconnectors High voltage disconnectors are commonly used switching devices in power substations. According to the International Electro‐technical Vocabulary IEV number 441‐14‐05 [1], the definition of a disconnector (DS) is: "A mechanical switching device, which provides, in the open position, an isolating distance in accordance with specified requirements". In IEEE standard C37.100‐1992 [2], insteadoftheterm"disconnector",theterms"disconnecting","disconnectswitch" or "isolator" are used, defined as "A mechanical switching device used for changingtheconnectionsinacircuit,orforisolatingacircuitorequipmentfrom thesourceofpower."Bothdefinitionsaresimilar.Theydefinethattheprinciple function of DSs is to provide electrical and visible isolation from the system. In addition, they must open and close reliably, carry current continuously without overheating,andremainintheclosedpositionunderfaultcurrentconditions[3]. Thedisconnectiongenerallycoverstwoaspects[4]: ‐ Disconnectionrelatedtoordinarydailyoperationofthesystem. ‐ Disconnection related to maintenance of transmission lines or substation equipmentsuchastransformers,circuitbreakers,capacitorbanksandsoon. For power system, personnel safety practices normally require a visible breakasapointofisolation.AnopenDSmeetsthisrequirement. 1.2Typeofdisconnectors DSs in Gas Insulated Substations are out of the scope of this thesis. Only high voltage DSs in the atmospheric air are studied. The HV air‐break DS comes in a variety of types and mounting arrangements. The most common four configurations are: vertical‐break, centre‐break, double‐break, and pantograph type. They can be mounted in various orientations, depending on the space offeredbythesubstation[3],[4]. The vertical‐break DS is presented in Figure 1.1 (left). This type of DS is highly suitableinicyenvironmentsthankstoitsrotatingbladesdesign.Itscontactdesign allows for application in installations where high fault current situations may occur.TheactivepartsofthistypeDSarethehingeendassembly,theblade,and the jaw end assembly. The smaller of the two insulators at the left rotates and rives the blade to open or close. It is usually horizontally mounted as shown in 2Chapter1 Figure 1.1. (left) A three‐phase vertical‐break DS mounted horizontallyin a closed position (CourtesyofHAPAMB.V.);(right)centre‐breakDSinaclosedposition(CourtesyofSiemens). Figure 1.1 (left). It can also be vertically mounted, i.e. the blade is vertically oriented in closed position. The vertical‐break DS requires minimum phase spacing. Anexampleofacentre‐breakDSisillustratedinFigure1.1(right).ThistypeDSis used mainly in locations with low overhead clearances. However, it requires larger phase spacing than a vertical‐break DS. The active parts consist of two blades, disconnecting at the centre. Both insulators rotate in a vertical plane to openorclosetheDS. Adouble‐breakDSisavariationofacentre‐breaktypeandisshowninFigure1.2 (left). The active parts are two jaw assemblies, one at each end, and a rotating blade.ThecentreinsulatorrotatestoopenorclosetheDS.Theycanbeinstalledin minimum overhead clearance locations and require minimum phase spacing. Their field of application includes icy environments due to the rotating blades designandinstallationsinhighfaultcurrentlocationsduetothecontactdesign. This design provides for two gaps per phase, allowing to interrupt significantly highercurrentthanthesingle‐breaktypeDSs[3],[4].Thisisbecausethesystem voltageisdistributedacrosstwogaps. ThepantographDStype,showninFigure1.2(right),isusedworldwide(butonly occasionally in North America) for Extra High Voltage application, 345‐800kV. Theactivepartsconsistofafixedcontactsarrangementattachedtothebusbarat thetop,ascissortypebladeandahingeassemblyatthebottom.Thesmallerone of the two insulators rotates to open or close the DS. This type DS normally provides transitions from high to low busbars, together with providing visible separationatthesametime.Thismethodrequirestheleastspace. HighVoltageAir‐breakDisconnectors3 Figure 1.2. (left) Double‐break DS in an open position; (right) pantograph DS in a closed position(CourtesyofHAPAMB.V). 1.3Interruptingcurrentwithdisconnectors Disconnectors disconnect unloaded circuits in order to accomplish visible isolation. However, they are operated under energized conditions and will therefore interrupt some current. The magnitude of this current depends on the specificsituations.AlthoughDSsdonothaveanycurrentinterruptionrating,they do have a certain interrupting capability, but it is limited due to their slowly movingcontacts.OvertheyearsDSs havebeenappliedtointerrupttransformer excitation currents, capacitive currents from short lengths of bus, cable or overhead line and small load currents. In this case, very small current is interruptedagainstfullsystemrecoveryvoltage(RV).Anothermajorapplication is bus transfer switching. In this case, often the full load current has to be transferredintoaparallelpath.Becauseofmanydifferentconditionsunderwhich theDSsmustinterruptcurrents,nointerruptingratingsareassigned[5]. Themainapplicationswhereacurrentistobeinterruptedare[4]: ‐ Transformermagnetizingcurrent Itisalsocalledexcitationcurrent.Thecurrentisusuallylessthan2A,oftenless than 1A at 100% excitation voltage, for today’s transformers. The current is generally described in terms of an equivalent RMS value derived from the core lossmeasurementbythemanufacturer. ‐ Capacitivecurrent This current, also called charging current, arises from connected short busbars, short unloaded transmission line, instrument transformers and other elements havingstraycapacitance.Theyactasacapacitiveload.Thetypicalrangeofthese currentsforstationair‐breakequipmentisshowninTable1.1[6]. 4Chapter1 Table1.1Typicalcapacitivecurrentrangeinoutdoorsubstations(50Hz). Equipment Capacitivecurrent(A) 72.5kV 145kV 245kV 300kV 420kV 550kV ≤0.04 ≤0.04 ≤0.04 0.05 0.08 0.1 CVT(4μF) 0.05 0.11 0.18 0.22 0.3 0.4 Busbars/m 1.7×10‐4 0.32×10‐3 0.54×10‐3 0.66×10‐3 0.84×10‐3 1.1×10‐3 CT ‐ Loopcurrent Loops are created when the DS commutes or transfers current from one circuit, such as a busbar or transmission line, to a parallel circuit. The loop current can reach up to the full load current in practice. This current has to be switched againstaverylowvoltage(thevoltageacrosstheparallelpath)[4]. 1.4Standardizationstatus A small current interruption capability of the DSs has been recognized by both IEEE and IEC standards. In the IEC standard [7], apart from the definition and basicfunctionoftheDS,thecurrentinterruptingcapabilityisaddressedinthree notes: NOTE 1: A disconnector is capable of opening and closing a circuit when either negligible currentisbrokenormade,orwhennosignificantchangeinthevoltageacrosstheterminals ofeachofthepolesofthedisconnectoroccurs. NOTE 2: "Negligible current" implies currents such as the capacitive currents of bushings, busbars,connections,veryshortlengthsofcable,currentsofpermanentlyconnectedgrading impedancesofcircuit‐breakersandcurrentsofvoltagetransformersanddividers.Forrated voltages of 420kV and below, a current not exceeding 0.5A is a negligible current for the purpose of this definition; for rated voltage above 420kV and currents exceeding 0.5A, the manufacturershouldbeconsulted. NOTE3:Foradisconnectorhavingaratedvoltageof52kVandabove,aratedabilityofbus transfercurrentswitchingmaybeassigned. SimilarlyintheIEEEstandard[8],thefollowingnoteismade: NOTE:Itisrequiredtocarrynormalloadcurrentcontinuously,andalsoabnormalorshort‐ circuit currents for short intervals as specified. It is also required to open or close circuits eitherwhennegligiblecurrentisbrokenormade,orwhennosignificantchangeinthevoltage acrosstheterminalsofeachoftheswitchpolesoccurs. An earlier version of the IEEE standard did include the following note and indicatedthethreetypesofcurrentwithoutexplanation[9]. HighVoltageAir‐breakDisconnectors5 NOTE:Adisconnectingswitchandahorn‐gapswitchhavenointerruptingrating.However,it isrecognizedthattheymayberequiredtointerruptthechargingcurrentofadjacentbuses, supports and bushings. Under certain conditions, they may interrupt other relatively low currents,suchas: 1. Transformermagnetizingcurrent. 2. Chargingcurrentsoflinesdependingonlength,voltage,insulationandotherlocal conditions. 3. Smallloadcurrents. Apartfromtheaboverecognition,therearenostandardsorrequiredratingsfor the current interruption by DSs. Both standards only refer to the current capability of the DS, and both use the word "negligible" to qualify the small current.Bothpointoutthepotentialcurrenttypesandthepossiblesource,which causesthesesmallcurrents.Thedifferencebetweenthetwostandardsisthatthe IEC standard defines the meaning of "negligible current" and quantifies the correspondingcurrentlevels,whereastheIEEEstandarddoesnot.Reference[6] gave the typical range of these currents for station air‐break equipment from 72.5kV to 1200kV. This is the newest IEC technical report, released in 2009, which describes the capacitive current switching duty for high voltage air‐break DSsforratedvoltagesabove52kVandprovidesguidanceonlaboratorytestingto demonstrate the switching capability. This is also the first time to provide an analysisoftheswitchingdutyandtodefinetestingprocedures. 1.5Objectiveofthesis Among these three main applications of current interruption using DSs, loop current interruption has been investigated in detail in [4]. As compared to inductivecurrent(excitationcurrent)andresistivecurrentswitching,acapacitive current is a more challenging to interrupt because of the trapped charge remainingontheloadtobeswitched(Chapters3,5).IntheIEEEstandard[8],the essentialdifferencebetweencapacitivecurrentandexcitation currentispointed outas: "This type of capacitive interruption is similar to excitation current interruption in which there is a succession of interruptions near zero current, each followed by re‐strikes. The difference between the two types of interruptions is that for each capacitive current interruption, a charge is more likely to be retained by the capacitive device. Each of these trappedchargescreatesanadditivebiasvoltagethatincreasestheprobabilityofre‐striking across the open gap of the switch 1/2cycle later when the source voltage has reversed its polarity.Consequently,longerarcreacheshavebeenexperiencedwhenswitchingcapacitive currentsthanwhenswitchingexcitationcurrents." CertainlycomparedwithswitchingonthecapacitivecurrentwithDSs,switching off(interrupting)capacitivecurrentisamoreseveretask. 6Chapter1 It was mentioned that a so‐called "negligible current" does not exceed 0.5A for ratedvoltagesof420kVandbelow.Inthepast,thecurrentinterruptingcapability of air‐break DSs therefore has been taken as 0.5A or less. However, at present with the fast development of power networks, user’s requirement for small capacitive current interruption using air‐break DSs is frequently higher due to complexity of the network or financial reasons. The subject of this thesis "InterruptionofcapacitivecurrentbyHVair‐breakdisconnectors"addressesthe challengingtaskofinterruptingrelativelylargecapacitivecurrents(uptofewtens ofamperes). CapacitivecurrentinterruptionwithaDSconsistsmicroscopicallyofasuccession of interactive events between the power circuit and the AC arc with a repetitive sequence of interruptions, re‐ignitions/re‐strikes. The arc re‐establishment is characterized in terms of oscillations and transients of current and voltage, recovery voltage. The interruption process is characterized in terms of arc duration, arc reach (perpendicular distance of outermost arc position to a line connecting the contacts), arc type (repetitive or continuous), arc brightness, energyinputintothearc,andsoforth. Thisdissertationwillparticularlyfocuson: ‐ TheprinciplemechanismofthecapacitivecurrentinterruptionusingHVair‐ breakDSs.Itincludesthetransientvoltageandcurrent,energyinputofthe circuit,re‐strikevoltage,etc. ‐ Thephenomenaoftheswitchingarc,includingdifferentfeatures,suchasarc voltage,arccurrentandotherphysicalcharacteristicssuchasarcbrightness, arc length, arc duration, arc reach, arc height, based on the analysis of data fromopticalandelectricalexperiments. ‐ Influencing factors on the transient and arc phenomena, which include the ratioofthesourceandloadsidecapacitances,powersupplyvoltage,current leveltobeinterruptedandre‐strikevoltagefromtheelectricalside.Fromthe physical side, the influences of (forced) arc cooling and (forced) elongation areanalyzedexperimentally. ‐ Thebreakdownvoltageandthepre‐breakdownmechanismintheairduring thenumerousre‐ignitionsandre‐strikes. ‐ Approachestoenhancetheinterruptioncapability. i. Anairflowsystemassistingarcquenching ii. Auxiliaryhigh‐velocitycontacts iii. Insertionofresistorsintothecircuit iv. Otherapproaches(discussiononly) HighVoltageAir‐breakDisconnectors7 The research is carried out based on experiments in different laboratories. Development and preliminary experiments are done at the high voltage laboratoryofEindhovenUniversityofTechnology.Fullscaletestsareperformed atKEMAHighPowerLaboratoriesinArnhemandPrague.ThetestobjectDSsare suppliedbyHAPAMB.V.,theNetherlands. References [1] IEVnumber441‐14‐05,[online]. Available:http://std.iec.ch/iev/iev.nsf/display?openform&ievref=441‐14‐05. [2] IEEE Standard Definitions for Power Switchgear, IEEE Standard C37.100‐1992, Oct. 1992. [3] J.D.McDonald,ElectricPowerSubstationsEngineering,London:CRCpress,2007. [4] D. F. Peelo, "Current interruption using high voltage air‐break disconnectors", Ph.D. dissertation,Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven, 2004. [5] IEEE Guide to Current Interruption with Horn‐Gap Air Switches, American National Standard(ANSI)IEEEStandardC37.36b‐1990,Jul.1990. [6] IEC Technical Report IEC/TR 62271‐305, "High voltage switchgear and controlgear– Part 305: Capacitive current switching capability of air‐insulated disconnectors for ratedvoltagesabove52kV",Nov.2009. [7] IEC Standard on "High voltage switchgear and control gear–Part 102: Alternating currentdisconnectorsandearthingswitches",IEC62271‐102,Dec.2001. [8] IEEE Standard Definitions and Requirements for High Voltage air switches, insulators, andbussupports, American National Standard (ANSI) IEEE Standard C37.3Oh‐1978, Jun.1978. [9] AmericanNationalStandardDefinitionsandRequirementsforHighvoltageAirSwitches, Insulators, and Bus Supports, American National Standard (ANSI) IEEE Standard C37.30‐1971,Apr.1971. 8Chapter1 LiteratureReview9 Chapter2 LiteratureReview Fromthehugenumberofpublicationsondevicesappliedinpowersystems,there appearstoexistonlyalimitedamountofpapersonair‐breakDSs.Especially,only afewofthemareactuallyconcernedwiththephenomenarelatedtofairlysmall capacitive current interruption with DSs in high voltage systems. The review presented in this chapter is based on selected papers (partly) related to interruptionofsmallcapacitivecurrent.Inparticularthischapterwillinvolvefive topics: ‐ Engineeringguidelinesoncurrentinterruptionwithair‐breakDSs. ‐ Fundamental aspects of capacitive current interruption with a stand‐alone DS(i.e.withoutanyauxiliarydevicestoaidinterruption). ‐ TransientscausedbyDSswitching. ‐ ApproachestoimprovecapacitivecurrentinterruptingcapabilityoftheDS. ‐ Possiblearcmodelsrelatedtothecurrenttopic. 2.1Awarenessofthecapacitivecurrentinterruptionwithdisconnectors F. E. Andrews et al. belonged to the earliest authors on the field of current interruption with air‐break DSs. They carried out numerous laboratory tests on transformer excitation current, loop current switching at 33kV level and line dropping at 132kV using DSs with horn‐gap on the Public Service Company of NorthernIllinois,USA,inthe1940s.Theresultswerepublishedinathesis[1]and in a subsequent paper [2]. A typical horn gap device (also called arcing horn) is showninFigure2.1.Thehorngapnormallyactsasthelastpointofmetal‐to‐metal contact on the DS when opening. Thus the arc burns between the arcing horn ratherthanbetweenthemainbladesoftheDS.Thegoalofthesetestswastofind the correlation between the voltage across the DS after switching off and the interruptedcurrentononehand,andthearclength,reachontheotherhand.The arclengthwasdefinedasthecompletelengthoftheirregularpathfollowedbythe arc.Thearcreachwasdefinedasthedistancefromapointmidwaybetweenthe bladeendstothemostremotepointofthearcatthetimeofitsmaximumlength (seeFigure2.2).Itwasrecognizedthatthearclengthscaledproportionaltothe arcvoltage.Themainattentionwaspaidto"arcreach",sinceitwasconsideredto bemoresignificantthanthe"arclength"inswitchingoperationsifthephase‐to‐ phaseandphase‐to‐groundclearancewereconsidered[2].Forthisreasonthearc reachhasbeentakenasameasuretoprovideacriterionforswitchingclearance. 10Chapter2 Figure2.1.Arcinghornmountedonavertical‐breakDS(copiedfrom[50]). Figure2.2.DefinitionofarcreachandlengthaccordingtoAndrewsetal.[2]. TheoverallresultsplottedinFigure2.3,copiedfrom[2],presentthearcreachin terms of "feet per kilovolt" as a function of initial interrupted current. From the results,thecriticallimitsdescribedthe"LimitoftheProbableReach"(LPR)ofthe arc,werederivedforexcitationcurrentandloopcurrentinterruption: LPR=5.03UocI,when0<I<100A(2.1) LPR=503Uoc,when100≤I<320A whereLPRisinmillimetres,UocthevoltageacrosstheDSafterswitchingoffisin kilovoltsofrms,andIistheinitialinterruptedcurrentinampereofrms.TheLPR for current levels below 100A was confirmed by later literature [3]‐[6]. In particular, [3] and [4] confirmed that (2.1) was valid for interrupting a magnetizingcurrentof1.73Aat242kV,andforcurrentsofupto9Aatvoltages up to 735kV respectively. In [5] and [6] it was stated that there was good agreementbetweentheresultsand(2.1).Equation(2.1)hasbeenfrequentlycited by utilities and DSs manufactures later on, and is also included in the Standard IEEEC37.36b‐1990[7]. LiteratureReview11 Figure2.3.Arcreachperkilovoltasafunctionoftheinterruptedcurrent,copiedfrom[2]. The test on the capacitive current interruption, performed in [2], showed it was safe to drop a 27km line (about 7A charging current) at 132kV for a 5m horizontalphasespacingowingtoafavourableobliquewind.Inaddition,it was pointed out the arc reach from the capacitive current interruption was clearly largerthanthatfromtheexcitationandloopcurrentinterruptionobservedinthe laboratorytests.Consequently,forthevoltageusedforthecalculationofthearc reach from interrupting capacitive current probably value twice of the actual voltagewastaken.Nofurtherinformationonthisissuewasprovided. Two AIEE committee reports, which are cited frequently, are of interest [8], [9]. The initial goal of the study reported in [8] was to investigate the transformer magnetizing current and its effect on relaying and DS operation. In addition, the interruption of (capacitive) overhead line charging current by a horn‐gap was surveyed in 1949 by distribution of a questionnaire to 250 utilities, of which 59 responded. The main results are presented in Figure 2.4. About 70% of the responsesindicated100%successwithalimitedcurrentrange(e.g.22Aat66kV, 13Aat118kV). A more recent survey on line and cable charging current interruption with DSs was made in 1962 amongst 71 utilities to collect the operating practice and experienceofutilities[9],theresultsofwhichareshowninFigure2.5and2.6. In Figure 2.5 the utility responses were divided into three categories, stating 100%successful,90‐99%successful,andlessthan90%successfuloperationsin linechargingcurrentinterruption.Forcablecharginginterruption(Figure2.6)all operationsreportedweresuccessful,butthetotalnumberofoperationswasvery small(only12operationswerereported). 12Chapter2 Figure 2.4. Interruption of overhead line charging current by a horn‐gap DS (copied from [8]).Note:•:signifies100%successfuloperation;o:signifies90%‐99%successfuloperation; ×:signifies75%‐80%successfuloperation. Figure2.5.Companyexperiencesreportedontheuseofair‐breakDSforinterruptionofline charging current. Note: •: grounded‐100% successful, : grounded‐90%‐99% successful, X: companiesusingquick‐breakorothersimilardevices.(Quick‐break,or "whip"isanauxiliary devicetoaidarcquenching,seeChapter7).”grounded”meansgroundedwyesystem(copied from[9]). TheAIEEreportsweretheearliestthoroughsurveys,whichwerecarriedoutso fartothebestknowledgeoftheauthorofthisthesis.Itwasfoundthatmostofthe utilitiesandcompaniesdidusetheair‐breakDSstointerruptcapacitivecurrentat system voltage lower than 138kV in the 1950s. It was reported that most companies did not have any guide for general use except previous experiences. Someofthemusedmanufacturer’sdata,andothersused(2.1)asaguide. LiteratureReview13 Figure 2.6. Company experiences reported on the use of air‐break DSs for interruption of cable charging current. Note: •: grounded‐100% successful; o: ungrounded‐100% successful (copiedfrom[9]). Asfromthemid1980smostimportantcontributionsarefromPeelo[6],[10]‐[13]. Inhisearlierpapers[10]and[11],thecurrentinterruptingcapabilityofDSswas analyzedbyconsideringthegivenphasespacing,arcreach(basedontheequation fromF.EAndrewsetal.[2]). LPR=2.75×5.03×Uoc×I(2.2) HereLPRisthearcreachinmm,Uocisthevoltageinkilovolt(rms)acrossswitch after interruption, and I is the interrupted current in ampere (rms). Compared with(2.1),afactorof2.75wasintroduced.Thisvaluecamefromthetestresults reported in [2] and [4]. Based on (2.2), safe maximum capacitive current levels thatcanbeinterruptedwerecalculatedandcomparedwithactualcurrentstobe interrupted. A criterion for the application of DSs from a current interrupting pointofviewwasalsoprovidedinTable2.1. BasedmainlyontheresultsfromAndrews[2]andPeelo[10],theIEEEguidefor DStointerruptcapacitivecurrentratingswaspublishedin1990[7].Inthisguide, Table2.1:Maximumcapacitivecurrentlevelsforsafeinterruptingbyair‐breakDSs.Note: dФ is the difference between the minimum metal‐to‐metal phase‐to‐phase clearance,dr is themaximumallowablearcreach,Icismaximumcurrenttobeinterruptedsafely(copied from[11]). Maximum system dФ(m) dr(m) Ic(A) voltage(kV) 15.0 0.31 0.22 1.84 27.5 0.38 0.22 1.02 72.5 0.79 0.37 0.65 145.0 1.35 0.52 0.45 253.0 2.26 0.81 0.40 550.0 4.30 2.20 0.50 14Chapter2 recommendationsforsystemvoltageupto362kVaregivenrelatedto:minimum phasespacingtogroundedobjects(orhorn‐gapswitch‐phasespacing),calculated arc reach and maximum operation voltage, safe capacitive currents to be interrupted with vertical‐break DS equipped with arcing horns and mounted in horizontaluprightposition. Inaddition,in2001IECalsopublishedastandard[14],whereitdefinesthatthe DSissupposedtobeabletodisconnecta"negligible"current(i.e.lessthan0.5A) at system voltage below 420kV. Based on the work of Peelo and Smeets, IEC 62271‐305(2009)Highvoltageswitchgearandcontrolgear–Part305:Capacitive current switching capability of air‐insulated DSs for rated voltages above 52kV was released. Two alternatives circuits for test are recommended for capacitive currentinterruptionwithaDSinthelaboratory. Summarizingthissubsection,alreadytensofyearsagoitwasrecognizedthrough field experience, surveys of utilities and power companies that a DS is able to interrupt the capacitive current. Principally, based on the work of Andrews and Peelo,theIEEEguideincludesthemaximumcapacitivecurrenttobeinterrupted safelybyverticalDSs(seeTable2.1).TheIECstandard,however,alsostatesthe current limitation, being 0.5A for below 420kV, but without supporting information. A test circuit for DS on the capacitive current interruption is proposed. 2.2Fundamentalaspectsofcapacitivecurrentinterruption Most literature sources referred in the previous subsection deal with current interruption from the point of view of the interrupted current level through experiments and surveys. The literature cited in this subsection describes the involvement of the circuit, e.g. by investigating the factors affecting the interruptioncapability[15]‐[20]. The laboratory experiments of Knobloch on interrupting capacitive currents described in [15] were carried out using a pantograph DS with rated voltage 245kV. It was concluded that both the source and load side capacitances play a role during the operations. During the interruption, in the voltage and current wave shapes, transients of two distinct frequencies were observed, i.e. a low frequency (specific data was not provided) and a high‐frequency (750kHz approximately). In [15] all DS tests succeeded in switching off 1A capacitive currentreliably. ApaperbyBoehne[16]wasprobablythefirsttobespecificallyrelatedtosmall capacitive current interruption using a DS to study transients in EHV systems. This paper analyses the field performance of EHV DS operations at 400kV and 505kV in the Pennsylvania Electric Company, through computer simulation. It presents modelling study of various phenomena associated with interrupting LiteratureReview15 small capacitive currents in air with a DS. Particular attention was given to the currentandvoltagetransientmagnitudesasafunctionofbusMVArating,switch resistance (a resistor in series with the DS) and the capacitive load. Specifically, the simulations covered nine MVA‐values (500‐2500MVA), seven values of the interrupted current (0.25A, 0.5A, 1A, 2A, 3A, 4A, 5A), and four separate resistor values (250Ω, 500Ω, 1000Ω, 2000Ω). The results included the line current,bus voltageanditsrateofchange, linevoltage,and thearcvoltage. The main conclusions were: the overvoltage was higher for disconnecting a smaller busbar(intermsofMVA)thanalargerbusbar;therateofswitchingovervoltage rise could reach values in the order of 300kV/µs; two natural frequencies were presentinthevoltageandcurrentduringtheinterruption;thearcvoltageandthe switching resistance can readily assist to reduce the overvoltage below 2.0p.u. (theratioofthetransientvoltageandamplitudeofthephase‐to‐groundvoltage). Aprojectwasinitiatedin1984atGeorgiaPowerCompany[17]todeterminethe influence of parallel lines and line/conductor configuration on the charging currentof115kVlines,whichwereswitchedoffwithanair‐breakDS.Atotalof five types of centre‐break and pole‐mounted DSs, to de‐energize a section of 115kV lines were investigated by computer simulation as well as by field tests. Several line/conductor configurations and parallel line arrangements were included.Theresultsindicatedthatparallellineshadanegligibleeffectonvalues of the charging current. However, the conductor configurations affected the chargingcurrentsignificantly. In Peelo’s Ph.D dissertation [6] and subsequent papers [12], [13] capacitive current interruption was studied through a series of experiments. The main resultscanbesummarizedasfollows: ‐ The interrupted current and the ratio of source and load side capacitance playakeyrole. ‐ The overvoltage was highest and the arcing duration was longest when the ratioofsourceandloadsidecapacitancewassmallest. ‐ Twotypesofappearanceoftheswitchingarc(socalled"erratic"and"stiff" mode) were recognized during the interruption with different ratios of sourceandloadsidecapacitances. As a conclusion, according to the consulted literature, transient phenomena, togetherwithhigh‐frequencyoscillations,onre‐ignitionsorre‐strikesduringthe interruption were found. The capacitances at both sides of the DS play a crucial role. The level of the current to be interrupted influences the interruption capabilityaswell.However,thedetailsarestillnotclear.Forinstance,howdothe capacitancevaluesaffectthearcingtime,thetransientovervoltageandtransient current?Howdothehigh‐frequencyphenomenaariseandhowtoquantifythem? Inthisthesis,theseissueswillbedealtwithindetail. 16Chapter2 2.3Transientscausedbycapacitivecurrentinterruption Research during the last two decades has confirmed that interruption of capacitive currents by DSs can be a major source of interference in secondary circuits of HV substations [21], [22]. DS operation generates high‐frequency transients,whichmayendangersensitivesecondarycomponentsofthenetwork throughelectromagneticcoupling.DuetotheslowmotionoftheDSmainblades duringoperation,anarcoflongduration(uptoseveralseconds)arises.Thearcis characterized by numerous interruptions and re‐strikes and is associated with high‐frequency phenomena. These voltage and current transients have been reportedtointerferewithorevendamagesecondarysystems[23]‐[28].Reported examplesare:(i)PG&ECompanyintheUSAexperiencedproblemswithswitching off the DSs leading to an equipment damage [23]; (ii) transients in substations havebeenidentifiedtointerferewiththecontrolsignalwiringandpowersupply unitsforprotectionrelaysburneddownduringDSoperationinEurope[24]‐[26]; (iii)afaultyrelayperformanceoccurredbecauseoftheDSoperationina500kV switching operation in China [27]; (iv) computer simulations and field measurement confirmed the possibility that normal operation of air DSs could resultinadamagedcurrenttransformerinArgentina[28]. Theliteraturerelatedtotransientphenomenamainlyfocusesonsystemvoltages above 220kV. A major part of the publications either describes field tests in HV substations directly, or was related to investigation of expected surge levels by computersimulationafterhavingobservedfaultsinsecondarysystems[29]‐[32], [40]. Duetovariousfactorssuchasswitchyardconfigurationandbuslength,therange of overvoltage levels caused from operations by air‐break DS is broad. For example,seriesoftestweredoneat362kV,787kVand1200kVsubstations,with anovervoltagelevelofmaximum1.9p.u.duringswitchingoffoperationsofEHV busesinRussia[30].Anearlierfieldtestdoneina500kVsubstationinShanghai, Chinarevealedanovervoltagelevelof1.81p.u.whendisconnectinganunloaded bus with an interrupted current of 0.75A [31]. A maximum overvoltage level of 1.4p.u. was observed in two 220kV substations in Bosnia [21], [32], [33]. A laboratorytestshowedthattheovervoltageisabout2.3p.u.whileinterruptinga capacitivecurrentupto2.3Aat170kVphasevoltage[13].Overvoltagereached 2.6p.u. at the termination of the bus, and 3.1p.u. value at a secondary service circuitinthestationappeared,duringde‐energizinga396mlongbuswithaDSat 345kVintheUSA[40]. Thedominanthigh‐frequencytransientwasalsooneofthefactorsofconcernin previous references. Frequency values reported were in the order of a few hundredkilohertz.Forinstance,itreached720kHzat138kVwhilede‐energizing thetransmissionline[34];itwasfoundthatdominantfrequencywasintherange LiteratureReview17 ofabout400‐500kHzat200kVand500kVsubstationfieldtestsreportedin[21], [31],[35]. In conclusion, most of the work related to transient phenomena associated with capacitive current interruption using air‐break DS focused on the overvoltage level in a specific substation and also focused on methods to protect secondary systems. However, the origin of the transients from disconnecting capacitive currents has hardly been studied yet. The influential factors on overvoltage are alsonotconsideredsystematicallyandindetail.Inaddition,thetransientcurrent was hardly investigated at all. Regarding the dominant frequencies, none of the literature gave detailed analysis of the fast mechanisms involved. Therefore, the fundamental phenomena resulting in high‐frequency behaviour during the switchingprocesswillbestudiedinthisthesis. 2.4Approachestoenhancetheinterruptioncapability Severalapproachestoenhancetheabilitytointerruptthecapacitivecurrentwith air‐breakDSsareaddressedintherelatedliterature.Uptonow,mainattentionis paidtohigh‐velocitycontactopeningdevices,inserteddampingresistorsandgas‐ blastdevices. 2.4.1Fastcontactseparation Ahigh‐velocitycontactopeningdevice(alsocalledwhiptypequick‐break)canbe attachedtotheDS.Itassistsinextinguishingthearcbyachievingalargeairgap withinashorttime.AnexampleofsuchdeviceisshowninFigure2.7. In 1989, a series of laboratory tests was conducted based on 10 different whip type quick‐break devices mainly attached at vertical‐break DSs by the U.S. DepartmentofEnergy[36].Thegoalwastodeterminethemaximuminterrupting capacitive current capability of 115kV DSs with this type of device. The results showed successful interruptions up to 11A at 115kV, managed by the device havingthehighesttipspeed.Itwasfoundthattheinterruptedcurrentmagnitude played a more prominent role than the RV across the DS main blades. However, therewerenofurtherdetailsreportedonthedesignofthedeviceitself. Inastudyreportedin[17]atotaloffivetypesofDS/quick‐breakdevices,together withacentre‐breakandpole‐mountedDStode‐energizeasectionof115kVline, were tested. The results showed that one of the quick‐break DS designs could successfully drop a 29km line at 115kV (corresponding to 6.5A). It was also stated that high tip speed and minimum bounce of quick‐break devices significantly enhance the charging current interruption capability, without mentioninganysupportingdetails. 18Chapter2 Figure 2.7. A centre‐break DS with a high‐velocity auxiliary device (indicated within the ellipse,fromthetestinthisthesis). Reference[37]alsoconfirmedthatahigh‐velocitydevicewasabletosuccessfully interrupt a capacitive current of 8A at 138kV. In [38] both laboratory and field testswerecarriedoutintheAlleghenyPowerSystem.Theresultshowedthatthe high‐velocityopeningdevicesucceededinterrupting17.6Aat80kV. The cited literature revealed that a fast‐opening device usually is employed at 138kVorbelow,anditwasmainlyinstalledwiththevertical‐breakDStype.Itcan be concluded that a high‐velocity opening device extends significantly the capabilityofinterruptingcapacitivecurrentofaDS.Theinterruptionmechanism with this high‐velocity interrupter on a centre‐break DS will be further investigatedinChapter7ofthisthesis. 2.4.2Resistivetransientssuppression A resistor inserted into the circuit can reduce the transients imposed on the system [5], [39]‐[44]. In [39], it was reported that the resistance in series or in paralleltothearccanlimittheamplitudeandthedurationofthetransients. Inordertoevaluatetheswitchingsurgelevelbydisconnectingcapacitivecurrents andmagnetizingcurrents,reference[5]describedaseriesoftestsonanAmerican EHVexperimentalline.Theswitchingsurgewasstudiedwhiledisconnectingwas achievedwithaconventionalswitchwithandwithoutsurgevoltagesuppressing resistors. Charging currents of 2.4A at 400kV and 3.0A at 505kV were interrupted.Comparedtothesituationwithoutresistor,thesurgeleveldecreased byabout15%,whichmeanstheinsertedresistorwasbeneficialfordisconnecting acapacitivecurrent. In Figure 2.8 three remedial approaches to reduce the surge level in a 345kV substation, according to [40], are shown. These methods are: (i) a resistor in LiteratureReview19 serieswiththeDSduringopening;(ii)aporcelain‐enclosedresistorplacedatthe same vertical position as the stationary contact of the vertical‐break DS, which was inserted automatically step by step during the operation; (iii) a parallel combinationoflinetraps(adevicethatblocksthehigh‐frequencycarrier(24kHz to500kHz)andletspower‐frequency(50Hz‐60Hz)passthrough)andaresistor connectedinserieswiththeDS.Itappearedthatmethod(iii)didnotprovevery effective in limiting overvoltages, but methods (i) and (ii) were effective. The overvoltage at the station service (secondary) circuit dropped to 1.7p.u. and to onethird,respectively,comparedwith3.1p.u.fromtheinterruptionwithoutany remedial devices. The resistor, which was bypassed whenever the DS was in closedposition,alsoappearedtobethemosteffectivewaytoreduceovervoltages. Asimilarmethodtograduallyinsertaresistancewasagainemployedin[41]‐[43]. A computer simulation of a three‐phase 550kV system using distributed parameters was carried out to investigate the re‐strike transients upon disconnecting unloaded transmission line. The simulations indicated that the surge‐suppressing resistors were effective in limiting the surge current and significantly contributed to damping of the re‐strike transients [41], [42]. Reference [43] reports a series of field tests, in which a 0.5MΩ resistor was installed in parallel to the main blades of the DS including a small arcing horn, conducted at a 220kV power station (Figure 2.9). The results at switching off a busbarshowedthatthearcingdurationwasreducedfrom2.5swithoutresistor to the value between 1.1s and 1.5s with parallel resistors. It was also reported thatthehigh‐frequencycomponentofthevoltagewaveshapesmeasurednearby was damped. The overvoltage in the antenna of the secondary system in the vicinityreducedfrom3.1p.u.withoutresistortoonly1.6p.u.withresistor.From these results it can be concluded that the damping resistor has a strong suppressing effect on the transient overvoltages caused by interrupting small capacitivecurrent. Further,in[44]theeffectofthearcresistanceontransientssuppressionduetoDS switchingwasstudiedanditwasconfirmedthatarcresistancereducedtransient overvoltages. Insummary,mostoftheworkonemployingresistivedevicesfocusesonthesurge or overvoltage level. However, still the conclusion can be drawn that a resistor enhancesthecapacitivecurrentinterruptioncapabilityoftheDSs.Forinstance,a 396m long bus of 345kV was successfully interrupted according to [40]. Reference [43] showed that arcing time was reduced greatly by a resistor in parallel to the main blades of the DS. In addition, it was mentioned that with an insertedresistorcurrentsupto4Aat230kV,3Aat345kVand1.5Aat500kV wereinterrupted.However,nospecificdetailswereprovided[45]. 20Chapter2 Figure2.8.Sketchof345kVDSwithvariousremedialdevicesmountedinplace(copiedfrom [40]). Figure2.9.InstalledDSwithdampingresistorina220kVsubstation(copiedfrom[43]). Inthisthesis,aresistorinseriesorinparalleltoacentre‐breakDSwillbestudied through simulation and laboratory experiments. The pros and cons will be discussed. 2.4.3Arccooling A gas‐blast device, as a method to improve the DS capability, is not used frequently. Such gas–blast interrupting attachment was mounted on a vertical‐ breakDSasindicatedinFigure2.10.Eachpoleunitwasequippedwithaporcelain LiteratureReview21 Figure2.10.Gas‐blastinterruptertestatTiddtestsite(copiedfrom[47]). tubemountedbetweenthebladeinsulatorstacks,andeachporcelaintubehada convergent‐divergent nozzle, aiming to deliver the high‐speed jet of gas into the arc. This method was found to be a most effective means to interrupt capacitive currents according to [5], [37], [46]‐[48]. It was stated that gas‐blast has strong effectinreducingthearcreachandarclengthbyvisualobservation,butwithout quantitativedatasupport[5];Testsshowedthatavertical‐breakDSwithgas‐blast devicecould drop20Aat aU.S. 138kVstation[37];a135kmlinewas dropped successfullyat230kV(estimatedcurrent25A)withthesupportofthegas‐blast as reported by a test program on a vertical‐break DS at the U.S. Bureau of Reclamation[46];a330kVvertical‐breakDSwithgas‐blastdevicewasdesigned as shown in Figure 2.10 [47]. Specifically the porcelain tube was connected to a gastank,whichwasmountedundereachsingle‐poleswitchunit.Thetestinthe American Gas and Electric 500kV Tidd test site showed that it was capable to interruptcapacitivecurrentupto8.5Aat220kVsuccessfully.Further,anitrogen pufferwasusedinanefforttofacilitatetheDSintheinterruptionofmagnetizing current[48]. The work related to gas‐blast is relatively old. Exclusively, an extra tube was appliedunderthevertical‐breakDS.Sinceitwasprovedthismethodtobemost effective,inthisthesis,thisapproachwillbecontinuedtostudy,butmainlyfocus onthearcextinguishingmechanismandinterruptionperformances. 2.4.4VacuumandSF6interrupter Due to the extraordinary arc interruption performances, vacuum and SF6 interrupters(seeFigure2.11)arealsoimplementedinconjunctionwithanairgap DSinsomedesigns[45].However,inthesecases,thecostsinvolvedarehighand the construction is complex, because the interrupters must be able to withstand the full RV. Most commonly, only vertical‐break DSs are provided with such an auxiliary interrupter. In principle, during the opening operation of the DS 22Chapter2 Figure2.11.(Left)multiplevacuuminterruptersand(right)asinglegapSF6interrupterper phase(copiedfrom[45]). the current is transferred into a loop with the SF6 or vacuum interrupter that performstheswitching.TheSF6orvacuuminterruptersareoutofcircuitduring theclosedpositionoftheDS.Theworkingprincipleofthissortofapproacheswill be described below in detail using vacuum medium as an example. The SF6 interrupterdoesinasimilarmanner. Arcs in vacuum are easier to interrupt than in air, because the vacuum has an electricalstrengthoftypically30kV/mmincaseofwellpolishedcontactsurfaces. Thishighelectricfieldwithstandstrengthandveryfastrecoveryensurethatonce thearcisextinguished,usuallyatthefirstcurrentzero,nore‐ignitionoccursand dielectric recovery is achieved within a few microseconds [51]. The idea of combining the DS with vacuum interrupters is to commutate the current path through the vacuum interrupters first and then open the contacts within the vacuum. In general, multiple series vacuum interrupters are needed for high voltage since a single vacuum interrupter can only function at medium voltage level[52].Theworkingprincipleofaninsertedsinglevacuumbottlewithvertical DSisplottedinFigure2.12.Itworksaccordingtothefollowingsteps: ‐ Step 1: the DS is in closed position. The vacuum interrupter is out of the circuit. The current flows between jaw (metal part connected with main bladesattherightside)andbladecontacts.Thevacuumcontactsareclosed. ‐ Step 2: the DS contacts separate but the circuit is maintained through the movingarchornandthefixedarchorn.Asthebladescontinueopening,the moving arc horn engages the vacuum interrupter’s actuating arm. The vacuumcontactsremainstillinaclosedposition. ‐ Step 3: as the blade movement continues, the fixed and moving horns separate. The current is transferred to a path through the closed vacuum contacts. After an adequate clearance distance is established, the current is interruptedinside ofthe vacuuminterrupterwithnoexternalarcingasthe contactsopen. LiteratureReview23 Figure2.12.InterruptionforDSwithinsertedvacuumbottle(s),(obtainedfrom[52]). ‐ Step 4: the DS moves to the fully open position releasing the actuator arm, whichisspringloadedtoreturntoitsoriginalpositionandclosethevacuum contacts. This approach is actually closer to a load‐break switch than to a DS switch. The mainadvantageisthatitcaninterrupthighercurrentsthananair‐breakDSalone. The level of the current to be interrupted is determined by the number of the vacuummodulesconnectedinseries.Forinstanceitiscapabletointerruptupto 70Acapacitivecurrentforuptoratedsystemvoltage230kV[45].Thedrawbacks ofavacuuminterrupterare: ‐ There is no visual indication of the dielectric integrity prior to operation, creatingapotentialsafetyhazardtopersonnelwhenoperatingtheDS. ‐ EachvacuumbottleinaDSwithamulti‐bottlestackcanhandleonlytheMV voltage level (up to 30 kV). Grading capacitors are required to divide the voltage evenly across the multiple gaps for successful performance. A vacuuminterrupterstackcanonlybeinstalledonavertical‐breakorasingle‐ sidebreakDSsduetoitscomplexconstruction. Sincemostoftheliteratureisrelatedtothevertical‐breakDSasatestobject,the potential interrupting capability of the vertical‐break DS with the aid of various approachesandtheircorrespondingcostsaresummarizedinTable2.2.Vacuum interrupterscanbeusedtointerrupthighercurrent.Issuestobeconsideredare thecostsandtheneedforgradingcapacitors.SF6mayinterruptthehighestlevel currentcomparedwiththeotherapproaches,butwithhighestcost. 24Chapter2 Table2.2.Potentialinterruptionratingsofvertical‐breakDSswithdifferentauxiliary devices.Note:thecostsaretakenfrom[45],assumingthecostofthearcinghornis"1". Typeofauxiliary devices Arcinghorn High‐velocity devices Gas‐blast Multiplevacuum interrupters SF6interrupters Resistors Interruptingcapability (Arms) Interruptsverysmall current,referto[7] Higherthanwitharcing horn(e.g.upto8Aat 138kV[37]) Higherthanwithhigh‐ velocitybreakdevices 70Aatupto230kV Upto300Aatupto 230kV 4 A—230 kV 3A—345kV 1.5A—500kV Cost 1 1.2 Noliterature 2.8 3.6 Noinformation 2.4.5.Ablationassistedapproaches Theuseofarccoolingthroughablation,i.e.evaporatedmaterialsfromcompounds like polyvinyl chloride, polymethylmethacrylate, etc. is a method commonly appliedinloadbreakdevices[53],[54].Theworkingprincipleisconfinementof thehotarcinanarrowtubecausingahighpressureandfastflowinggas,which facilitates rapid arc extinction. Based on this method, a special arcing horn has been developed [55]. This device was designed originally to prevent lightning damage on overhead line, capable of interrupting ground fault currents. It was confirmed experimentally that it can interrupt ground fault current up to 1kA withinhalfcycleat77kVsystemvoltage.Figure2.13showsthearrangementand workingprinciple.Thereisanarrowtube,madeofpolyvinylchloride,mountedat the tip of the arcing horn electrode in order to interrupt the current. The interruption portion, located in the ground side horn, is a tube with an intermediateelectrode.Thiselectrodeisasmallcolumn,whichpiercesataspotin the tube wall. When excessive voltage is applied between the horns at line and ground side, a breakdown occurs between their tips through the intermediate electrode. Then, the arc arises between the horns. The gas during arcing in the interruptiontubehasgainedahigh‐velocityandahigh‐pressure.Thedevicewas notabletointerruptashort‐circuitcurrentabove10kA.Nofurtherinformation about the interruption performance below 445A was reported. Pressure and velocityofthegasinsidethetube,andtheablationofthetubematerialatcurrent levelsaround10Amightbetoolowforeffectiveoperation. LiteratureReview25 Figure2.13.WorkingprincipleofablationassistedarcinghornwithPVCtube(copiedfrom [55]). Arcinterruptionthroughablationisverysuccessfullyappliedinmediumvoltage loadbreakswitches.Oneapplicationisdiscussedhere[56].Figure2.14showsthe structure and working mechanism. Upon disconnecting, the main contacts "1" openfirstandthecurrentisbrieflytakenoverbytheparallelconnectedlagging pins "2". During this breaking motion, the opening springs "3" acting on the lagging pin are tensioned. On reaching a stop, the lagging pin leaves the holding contact "4". The arc occurring between the arcing tip of the holding contact and thetungstentipofthelaggingpinisextinguishedinthearcingchamber"5".The arcing chamber itself is closed. This device contains four sections and has a pressure chamber "6" and an expansion chamber "7". Two extinction plates "8", arranged in the pressure chamber, are forced into the path of the arc by lateral springpressure.Atlowcurrentsthearcisextinguishedduetothecoolingeffectof the walls. In the higher current range arc extinction is achieved by the ablated gases produced in the pressure chamber flowing out of this chamber into the expansion chamber. Owing to this complementary combination of several extinctionprinciplesawidecurrentinterruptionrangeiseffectivelycovered.This device is capable to interrupt cable charging currents up to 20A at low voltage level 38.5 kV. For higher current and higher voltage levels, no information was provided. Based on using ablation assisted arc extinction as commonly used in load break switches was considered. Present designs require high current (for sufficient ablation)orarelimitedtorelativelylowvoltagelevel.Theirapplicabilityonhigh voltageair‐breakDSsmaybequestionablebecausethelonglengthofthearcand the lower energy density. In addition, confining the DS arc in a small space may haveadverseeffectsbecausecoolingfromambientairisobstructed.Thismaybe consideredforfurtherinvestigation. 26Chapter2 Figure2.14. Working principle for an indoor loadbreakswitchwithtwochambers (copied from[56]). 2.5Arcmodelsrelatedtocurrentinterruptionwithadisconnector Anelectricarcisadischarge,characterizedbyahighcurrentflow(ascomparedto othergasdischarges,likeglowdischarge),resultinginacurrentflowingthrougha usually non‐conductive medium such as air. Understanding the phenomena associated with the electric arc requires combined knowledge of electro‐ magnetism,thermodynamicsandplasmaphysics. The electric arc related to current interruption with a DS is a so called free burning AC arc with relatively low current (up to a few tens of amperes) in the atmosphere. It behaves differently from the well‐studied and well‐documented fault arc in circuit breakers because of its much higher length (metres vs. centimetres), its much lower current (amperes vs. kilo‐amperes) and its unconfined ("free‐burning") nature. Compared with the short arc, where the influence of anode and cathode roots is significant [57], the arc length of DS arc may reach up to a few metres depending on the system voltage and current. Comparedtotheconfinedarc,thefreeburningarcbehavesmorerandomlyandis influenced by external conditions as wind, humidity, and surrounding electromagnetic fields. An overview of the free burning arc literature was presentedin[6]. Arcmodellingmaybeaneffectivecomputationaltoolwhenfieldexperimentsare difficult to manage. Selected models from literature are therefore discussed in relationwiththeirapplicabilityforthearcfromalowcurrentinterruptionorfree burningarcintheair. Duetotherandomnatureofthefreeburningarc,itispracticallynotpossibleto duplicatethegenuinearccharacteristicsbyacomputermodel.Sometimesthearc isrepresentedbyaresistororasavoltagesourcewithperiodicrectangularwave form, changing its sign at each arc current zero crossing; sometimes a more LiteratureReview27 complicatedmodelwithpiecewisecharacteristicsisadoptedtopresentthearcat variablelength,resultinginavariablearcvoltage.However,thesearcmodelsdo not take into account the actual interaction of the arc and the electromagnetic transients in the circuit [58], [59]. No model exists, taking into account the discontinuous character of the arc, which is observed in the present study to be themainfeatureofthelow‐currentDSarc.Therefore,onlymodels,moredistantly relatedtothearcinDScurrentinterruptionwillbeintroducedhere. Anarcmodelsimulatingthebehaviourofarcsdrivenbymagneticforceinagas [57], [60]‐[62], was proposed first in [62]. The arc is assumed to be a chain of small rigid cylindrical current elements. Each element experiences the Lorentz force from the magnetic field and a fluid drag force from the surrounding gas. Each element moves with the velocity determined by these forces. The arc behaviour is determined by the position and the motion of each element. Consequently,thebehaviourandtheshapeoftheentirearcmaybeconstructed. Experimentsin[62]withcurrentof1kAat50HzinSF6showedgoodagreement betweenmodelandtheobservedarcinpractice. Inoverheadline,generally,anarcinghornisusedtoprotecttheinsulatorsagainst damage from arcs after switching, or by lightning induced breakdown to earth. The arcing horn is installed in parallel with the insulator. The arc between the arcinghornsbehavesasafreeburningarcintheair.Basedonthemodel[60],it canbeestimatedhowfarthearcisfromthecolumnoftheinsulator.Asituation foracurrentof160Awasstudiedin[61],resultinginagoodmatchbetweenthe simulationandexperiments.Suchanarcmodelcanpotentiallybeusedtodescribe arc movement related to DS current interruption, and to evaluate the risk of endangeringthesurroundingequipment.However,thecurrentrelatedtotheDS interrupting has much lower values and their arc movement is much more influencedbyconvectionprocessesthanbyelectro‐magneticinteraction. Compared with all other types of arc mentioned before, the secondary fault arc from EHV power overhead line resembles most closely the arc from capacitive current interruption with a DS because of a comparable length and similar environmental(unconfined)conditions.Asecondaryfaultarcfollowstheprimary faultcurrentinsinglephasetrippingandreclosingschemesduetocapacitiveand inductivecouplingwiththesoundphases[63].Thevalueofthesecondarycurrent mayreachuptoafewhundredamperesandisdeterminedbythestructureofthe overheadline,suchaslinearrangement,length,earthing,etc. TheCassiearcmodel[64],frequentlycited,assumesthatthearcchannelhasthe shape of a cylinder and is filled with highly ionized gas with a constant temperature,butwithavariablediameter.Cassiedevelopedhisarcmodelbased ontheassumptionthatahighcurrentarcisgovernedmainlybyconvectionlosses. OneofthemostcommonusedCassiearcequationformulationsis: 28Chapter2 d[ln( g )] 1 u2 ( a2 1) (2.3) u0 dt where u0 is the static arc voltage, ua is the momentary arc voltage, is the arc timeconstant,gistheinstantaneousarcconductance. Throughstudyofexperimentalu‐icharacteristicsofarcs,havingpeakarccurrent values from 68A to 21kA, it was found that the Cassie model may be used to model the behaviour of the secondary arc [59]. Knowing ua ,u0 , from specific experiments, the arc conductance g can be determined to represent the arc for computersimulation[66],[67]. Based on the Cassie model, Kizilcay defined the arc behaviour according [53], [68]‐[70]as: 1 dg d t (G g ) (2.4) ia G (u0 r0 ia ) la whereiaistheinstantaneousarccurrent,latheinstantaneousarclength, u0 isthe arc voltage per unit length, r0 is the arc resistance per unit length, G is the stationaryarcconductance.Parameters , u0 , r0 dependonarclength,andcanbe obtained over a certain time interval due to the required high accuracy of measured arc current and voltage from specific measurements, as described in detailin[59].Thismodelnotonlyconsiderstheheatofthearc,butalsotakesinto accountthesecondaryarcelongation.Theexperimentswithacurrentuptoafew hundred amperes matched the simulated results well. It may be feasible to describethearcwithlowercurrent(fromcapacitivecurrentswitchingaswell). Terzijatookconsiderableefforttomodellongarcsinfreeair[71]–[75].Thearcin his experiments was initiated between two electrodes with a distance of 60cm, the current ranged from 100A to 20kA at a system phase voltage of 20kV. The equation used is a generalization of (2.3): dga f ( ga ,ua , ia ,t ) . By observing the dt arcvoltagewaveform,presentedroughlyasarectangularshapeinphasewiththe arccurrent,thearcwasmodelledthroughthefollowingnonlinearequation: I ua (t ) U a U b 0 R ib (t ) sgn(ia ) (2.5) i ( t ) b LiteratureReview29 where ua(t ),ia(t ) arearcvoltageandarccurrent, I 0 ia (t ) <I 0 (2.6) ib (t ) ia (t ) ia (t ) I 0 Ua ,Ub ,R are parameters defining the shape of the arc voltage, is to account forGaussiannoisewith.Ua=Eala,theaverageofEawastakenbetween1.2kV/m I and 1.5kV/m for the chosen experimental current range. The term U b 0 ib (t ) models the arc re‐ignition voltage, and R ib (t ) is an additional term, which is determined by the arc current ia, R , is part of the arc resistance, at the chosen currentrange,intheorderoftenthsofohm.Theauthoraddedtheeffectofthearc elongationin[75]bymultiplying(2.5)withasuitableelongationfunctionL(t)in [73].Thearccurrentfromtheexperimentswasintherangetakenoffaultcurrent. For instance, the model was applied to a secondary arc in overhead line [74]. However, the method may also be considered to describe the arc related to capacitivecurrentoperationwithaDS. Exceptforthearcmodelbasedonmagneticforces,theothermodelsintroducedin thissectionareexclusivelybasedontheCassiearcmodel.Mostmodelparameters mustbeobtainedfromspecificfieldexperiments.Asaconclusion,althoughthere is no detailed arc model connected to current interruption with a DS, the arc models mentioned above may be considered as a first step. Because of the random,inhomogeneousandtimedependentnatureoftheconductingmedium,it hasnotbeenattemptedtodesignanarcmodelinthepresentstudy. 2.6Conclusion Capacitivecurrentinterruptionusinganair‐breakDShasnotattractedtoomuch attention, and the existing literature provides only a limited insight into the mechanism.However,certainliteraturesourcesdoexist,whicharemoreorless related to this subject. Each from different points of view and with different application purposes, such as surges imposed on secondary systems, disconnectionofmagnetizingcurrents,loopcurrentsandsoforth. It is recognized that DS requirements should include the capability to interrupt small capacitive current. Based on field tests and computer simulations at different system voltage levels, it was found that interrupting even of small capacitive currents can cause large transients, which potentially can damage nearbyequipment.Also,arcswithunacceptabledurationand/orreachcanarise. 30Chapter2 Among many factors, which influence the capability of interruption in practice, such as clearance distance, atmospheric condition (wind, humidity), conductor configurationsandsoon,afewfactorswererecognizedandinvestigated.Therole ofbuslength(sourceandloadsidecapacitance),parallellines,lineconfigurations werepointedout.Inaddition,high‐frequencytransientswererecognizedonlytoa practicalextent,butneitherfundamentallynorsystematicallyinvestigated. Inthisliteraturesurvey,itwasfoundthattherearenointerruptionratingsforDSs exceptthosegivenbyIEEE[7].TheIECstandard[14]alsofailstoclarifythistopic. On the other hand, the present power systems are rapidly increasing in size, requiringperformancesbutalsoenhancingcost‐effectiveness.Incertainlocations, low‐cost DSs having an enhanced current interruption capability could in some situationsofde‐energizationofunloadingtheshortline/cable/busbartoreplace thecircuitbreakers. It is indispensable to understand the phenomena of capacitive current through freeburningarcsinairprofoundlyandtofindapproachestoimprovetheability oftheswitchingdevice:thedisconnector.Thisconsiderationisthemotivationof theworkinthisthesis. References [1] P. A. Abetti, "Arc interruption with disconnecting switches", Master thesis, Illinois InstituteofTechnology,Jan.1948. [2] F.E.Andrews,L.R.Janes,M.A.Anderson,"Interruptingabilityofhorn‐gapswitches", AIEETrans.,vol.69,pp.1016‐1027,Apr.1950. [3] M. W. Anderson, "Power autotransformer current interruption with an air‐break disconnect switch", IEEE Trans. on Power Apparatus and Systems, vol.74, pp. 1157‐ 1163,Jan.1955 [4] CEAProject069‐T‐102report,"Theinterruptingcapabilityofhighvoltagedisconnect switches",CanadianElectricalAssociation,Jul.1982. [5] A.FotiandJ.M.Lakas,"EHVswitchtestsandswitchingsurges,"IEEETrans.onPower ApparatusandSystems,vol.83,pp.266‐271,Mar.1964. [6] D. F. Peelo, "Current interruption using high voltage air‐break disconnectors", Ph.D. dissertation,Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven, 2004. [7] IEEEGuidetoCurrentInterruptionwithHorn‐GapAirSwitches,IEEEStandardC37.36b‐ 1990,Jul.1990. [8] AIEECommitteeReport,"Reportontransformermagnetizingcurrentanditseffecton relayingandairswitchoperation",AIEETrans.,vol.70,pp.1733–1740,Jul.1951. [9] IEEE Committee Report, "Results of survey on interrupting ability of air‐break switches", IEEE Trans. on Power Apparatus and Systems, vol.PAS‐85, pp. 1008‐1019, Sept.1966. [10] D. F. Peelo, "Current interrupting capability of air‐break disconnect switches", IEEE Trans.onPowerDelivery,vol.1,pp.212‐216,Jan.1986. LiteratureReview31 [11] D. F. Peelo, "Current interrupting capability of air‐break disconnect switches", IEEE PowerEngineeringReview,vol.PER‐6,pp.53‐54,Jan.1986. [12] D.F.Peelo,R. P.P.Smeets,L.van DerSluis,S.Kuivenhoven,J.G.Krone,J.H.Sawada and B. R. Sunga, "Current interruption with high voltage air‐break disconnectors", in Proc.CigreConf.,papernr.A3‐301,2004. [13] D.F.Peelo,R.P.PSmeets,J.G.Krone,"Capacitivecurrentinterruptioninatmospheric air",inProc.CigreA3/B3Colloquiun,2005. [14] IEC Standard for High voltage Switchgear and Control Gear –Part 102: Alternating CurrentDisconnectorsandEarthingSwitches,IEC62271‐102,Dec.2001. [15] H.Knobloch,"Switchingofcapacitivecurrentsbyoutdoordisconnectors",inProc.Fifth InternationalSymposiumonHighVoltageEngineeringConf.,Aug.1967. [16] E.W.Boehne,"EHVsurgesuppressiononinterruptinglightcurrentswithairswitches I‐ capacitive currents", IEEE Trans. on Power Apparatus and Systems, vol.PAS‐84, pp.906‐923,Oct.1965 [17] S. G. Patel, W. F. Holcombe, D. E. Parr, "Application of air‐break switches for de‐ energizing transmission lines", IEEE Trans. on Power Apparatus and Systems, vol.4, pp.368‐374,Jan.1989. [18] A. Schutte, H. Rodrigo, "Transient phenomena due to disconnect switching in high voltage substation", in Proc. 11th InternationalHigh Voltage Engineering Symposium Conf.,London,Aug.1999 [19] I. B. Johnson, A. J. Schultz, N. R. Schultz, R. B. Shores, "Some fundamentals on capacitance switching", IEEETrans.onPowerApparatusandSystems, vol.74, pp. 727– 736,Aug.1955. [20] W. M. Leeds, R. C. van Sickle, "The interruption of charging current at high voltage", AIEETrans.,vol.66,pp.373–382,Jan.1947. [21] S. Carsimamovic, Z. Bajramovic, M. Ljevak, M. Veledar, "Very fast electromagnetic transients in air insulated substations and gas insulated substations due to disconnector switching", in Proc. International Symposium on Electromagnetic Compatibility,vol.2,pp.382‐387,Aug.2005. [22] C. M. Wiggins, D. E. Thomas, F. S. Nickel, S. E. Wright, "Transient electromagnetic interference in substations",IEEETrans.onPowerDelivery, vol.9, pp. 1869‐1883, Oct. 1994. [23] R. J. Gavazza, C. M. Wiggins, "Reduction of interference on substation low voltage wiring",IEEETrans.PowerDelivery,vol.11,pp.1317‐1329,Jul.1996. [24] M. A. van, Houten, E. J. M. van, Heesch, A. P. J. van, Deursen, "General methods for protection of electronics against interference, tested in high voltage substations", in Proc8thInternationalSymposiumonElectromagneticCompatibilityConf., pp. 429‐434. Zürich,Mar.1989. [25] A. Rodewald, "Interference generated by switching operations and its simulation", in Proc.IEEEInternationalSymposiumonEMCConf.,pp.133‐138,1984. [26] S.Carsimamovic,R.Gacanovic,Z.Bajramovic,"Switchingovervoltagesinair‐insulated substation(AIS)duetodisconnectorandcircuitbreakerswitching",papernr.C4‐301, CigreConf.,Paris,2006. [27] W. Mao, "Analysis for malfunction of CBF relay insulating switch operation", AutomationofElectricPowerSystems,vol.26,pp.74‐76,Apr.2002. [28] A. Funes, A. Ledesema, "Field measurements and modelling of high‐frequency transientduringdisconnectswitchoperationsinEHVsubstations,assessmentoftheir effectsonCurrentTransformers",papernr.A3‐207,CigreConf.,Paris,2010. [29] G. Carrsco, "Electromagnetic interference in substation Jose up to 400/115 kV", in Proc.IEEETransmission&DistributionConferenceandExposition,pp.1–6,Aug.2006. 32Chapter2 [30] V. S. Rashkes, L. D. Ziles, "Very high‐frequency overvoltages at open air EHV substations during disconnect switch operations", IEEE Trans. on Power Delivery, vol.11,pp.1618‐1623,Jul.1996. [31] Z. Liu, "The test of switching no‐load bus with 500 kV disconnector", High Voltage Engineering,vol.4,pp.58‐61,1990. [32] S. Carsimamovic, Z. Bajramovic, M. Ljevak, M. Veledar, N. Halilhodzic, "Current switching with high voltage air disconnector", in Proc. International Conference on PowerSystemsTransients,papernr.IPST05–229,Jun.2005. [33] S. Carsimamovic, Z. Bajramovic, M. Veledar, O. Hadzic, A. Carsimamovic, "Computer simulationoftransientsduetodisconnectorswitchinginair‐insulatedsubstations",in Proc.InternationalConferenceonComputerasaTool,pp.2051‐2053,Sept.2007. [34] S.G.Lodwig,C.C.Scheutz,"CouplingtocontrolcablesinHVsubstations",inProc.IEEE International Symposium on Electromagnetic Compatibility, vol.1, pp. 249‐253, Aug. 2001. [35] Z. Lu, C. Wang, "The switching transient electric field measurements in a 500 kV substation",inProc.the3rdInternationalSymposiumonElectromagneticCompatibility, pp.194–197,May2002. [36] R. J. Denis, "Tests to determine the interrupting capabilities of 115 kV quick‐break devices", Division of laboratories report from U. S. Department of Energy, Tech. Rep. BonnevillePowerAdministration,Mar.1989. [37] E.C.Rankin,"Experiencewithmethodsofextendingthecapabilityofhighvoltageair‐ break switches", AIEETrans.onPowerApparatusandSystems, vol.78, pp. 1634‐1636, Dec.1959. [38] T.J.Jackson,"138kVairswitchinterruptingdevicetests",presentedatPEAElectrical EquipmentMeeting,Hershey,Pennsylvania,May1977. [39] R.C. van Sickle, "Influence of resistance on switching transients", AIEETrans, vol.58, pp.397‐404,Aug.1939. [40] H. L. Rorden, J. M. Dils, S. B. Griscom, J. W. Skoglund, E. Beck, "Investigations or switching surges caused by 345 kV disconnecting‐switch operation", AIEETrans.on PowerApparatusandSystems,vol.77,pp.838‐844,Oct.1958. [41] D. L. Lott, "Surge‐suppressing resistors applied to EHV disconnector switches", Allis‐ ChalmersEngineeringReview,vol.40,pp.24‐27,1975. [42] K.B.Stump,D.L.Lott,"Controllingsystemtransientswithsurge‐suppressingresistors inairdisconnectswitches",ElectricPowerSystemsResearch,pp.269–277,Feb.1979. [43] M. S. Savic, "Suppression of the high‐frequency disturbances in low voltage circuits caused by disconnector operation in high voltage open‐air substations", IEE ProceedingsGeneration,TransmissionandDistribution,vol.133,pp.293‐297,Jul.1986. [44] S. Carsimamovic, Z. Bajramovic, "Influence of resistance of electric arc on transients due to disconnector switching in air‐insulated substations", IEEE Trans. on Plasma Science,vol.36,pp.2560‐2565,Oct.2008. [45] D.Childress,"Disconnectswitchmountedinterruptingdeviceshowtochoosewhatto usewhenandwhere",[Online].available:www.southernstatesllc.com. [46] P. E. Richardson, A. Foti, "Gas‐blast switch tests on 230‐ kV system", Electrical Engineering,pp.712‐717,Aug.1956. [47] I. W. Gross. C. Killian, J. M. Sheadel, "A 330‐ kV air switch", AIEE Trans. on Power ApparatusandSystems,vol.73,pp.264‐270,Jan.1954. [48] A.J.Mcelroy,W.SPrice,H.M.Smith,D.F.Shankle,"Fieldmeasurementofsurgewhen switching in 345‐ kV stations", AIEE Trans.on Power Apparatusand Systems, vol.83, pp.250‐263,Mar.1964. LiteratureReview33 [49] B. J. Berner, J. R. Rostron, "New improvement in transmission switching using SF6 interrupter technology", in Proc. IEEE/PES Transmission and Distribution Conference andExposition,vol.1.pp.317–322,Oct.2001. [50] J. D. McDonald, "Electric power substations engineering", CRC Press, ISBN‐10: 0849317037,Jun.2003. [51] H M. Ryan, High voltage engineering and testing, 2nd edition, Bungay Suffolk: RefineCatchLtd,p.121,2001. [52] [online].Available:http://www.southernstatesllc.com/. [53] L. Niemeyer, "Evaporation dominated high current arcs in narrow channels", IEEE Trans.PowerAppl.Syst.,vol.PAS‐97,pp.950–958,Jun.1978. [54] E. Z. Ibrahim, "The ablation dominated polymethylmethacrylate arc", J.Phys.D:Appl. Phys.,vol.13,pp.2045–2065,1980. [55] T. Chino, M. Iwata, "Development of arcing horn device for interrupting ground‐fault currentof77kVoverheadlines",IEEEtrans.onpowerdelivery,vol.20,pp.2570‐2575, Oct.2005. [56] [online].Available:http://www.driescher.com/home/index.htm. [57] S.Gu,J.He."Motioncharacteristicsoflongacarcsinatmosphericair",AppliedPhysics Letters,vol.90,pp.0515011‐0515013,2007. [58] A.T.Johns,R.K.Aggarwal,Y.H.Song,"Improvedtechniquesformodellingfaultarcson faulted EHV transmission systems", IEE Proceedings Generation, Transmission and Distribution,vol.141,pp.148‐154,Mar.1994. [59] L.Prikler,M.kizilcay,G.ban,"Improvedsecondaryarcmodelsbasedonidentification of the arc parameters from staged fault test records", in Proc. 14th Power Systems ComputationConf.,papernr.3,p.1‐7,Jun.2002. [60] G. Meunier, A. Abri, "A model for the current interruption of an electric arc", IEEE Trans.onMagnetics,vol.20,pp.1956–1958,Sept.1984. [61] S.Gu,J.He,B.Zhang,G.Xu,S.Han,"Movementsimulationoflongelectricarcalongthe surfaceofinsulatorstringinfreeair",IEEETrans.onMagnetics,vol.42,pp.1359‐1362, Apr.2006. [62] K. Horinouchi, Y. Nakayama, M. Hidaka, "A method of simulating magnetically driven arcs",IEEETrans.onPowerDelivery,vol.12,pp.213‐218,Jan.1997. [63] A.Al‐Rawi,M.Devaney,"Measurementofsecondaryarccurrentintransmissionlines employing single phase switching", in Proc. IEEE Instrumentation and Measurement Technology:Sensing,Processing,NetworkingConf.,vol.1,pp.297‐301,May.1997. [64] L. van der Sluis, Transients in Power Systems, England, John Wiley & Sons Ltd, p.70, 2001. [65] A.P.Strom,"Long60‐cyclearcsinair",AIEETrans.,vol.65,pp.113–118,Mar.1946. [66] A.S.Maikapar,"Extinctionofanopenelectricarc",Elektrichestvo,vol.3,pp.64‐69,Apr. 1960. [67] G. Song, J. Ling, "Research on secondary arc in 1000 kV UHVAC two‐circuit transmissionlinesanditscontrol",inProc.ofChineseSocietyforElectricalEngineering, vol.31,pp.138‐143,Jun.2011. [68] M. Kizilcay, G. Ban, L. Prikler, P. Handl, "Interaction of the secondary arc with the transmission system during single‐phase autorelcosure", in Proc. IEEE Power Tech Conf.,Bologna,vol.2,p.1‐7,Jun.2003. [69] M.Kizilcay,P.LaSeta,"Digitalsimulationoffaultarcsinmedium‐voltagedistribution networks", in Proc. 15th Power Systems Computation Conf., paper nr. 3, p. 1‐7, Aug.2005. [70] M. Kizilcay, T. Pniok, "Digital simulation of fault arcs in power systems", European Trans.onElectricalPower,vol.1,pp.55‐60,Jan.1991. 34Chapter2 [71] V. Terzija, H. Koglin, "Long arc in free air: testing, modelling and parameters estimation: part I", in IEE Proc. Generation, Transmission and Distribution, vol.2, pp.404‐409,May2002. [72] V. Terzija, H. Koglin, "Long arc in free air: testing, modelling and parameters estimation: part II", in IEE Proc. Generation, Transmission and Distribution, vol.2, pp.481‐486,May2002. [73] V. Terzija, Z. M. Radojevic, "Numerical algorithm for adaptive autorelcosure and protection of medium‐voltage overhead lines", IEEETrans.onPowerDelivery, vol.19, pp.554–559,Apr.2004. [74] V.Terzija,H.Koglin,"Laboratorytestinganddynamicmodellingoflongarcinfreeair", in Proc. International Conference Power System Technology, vol.2, pp. 897‐902, Dec. 2000. [75] V. Terzija, G. Preston, M. Popov, N. Terzija, "New Static “AirArc” EMTP model of long arc infree air power delivery",IEEETrans.onPowerDelivery,vol.26,pp.1344–1353, Jul.2011. BasicCircuitAnalysis35 Chapter3 BasicCircuitAnalysis Capacitive current interruption by a DS is considered as a topology with capacitancesateithersideoftheDS.Thecapacitanceatthesupplysidemaycome from (instrument) transformers, busbars or also from the other conductors that areconnectedtothebusbar.ThecapacitanceattheloadsideoftheDS,maycome fromunloadedtransmissionline,busbar,etc.Thecapacitancevaluesrangefrom lessthanonenanofarad,e.g.forshortlines,tohundredsnanofaradincaseoflarge capacitor banks. Switching of DSs is basically governed by charge exchange between these capacitances. This thesis is concerned with switching between capacitances that can be considered as lumped elements capacitances that have valuesupto300nF.Capacitorbanks,havingmuchlargervalue,arenotswitched withDSs. 3.1Theinterruptionprocess CapacitivecurrentinterruptionwithaDSconsistsoftheinteractionbetweenthe circuit and the arc. This interaction results in a repetitive sequence of breaks (interruption of current) and re‐strikes and/or re‐ignitions. Re‐strike is a breakdownoftheinsulatingmedium–theairbetweentheDSblades‐resultingin re‐establishment of the arc, more than one quarter of a cycle after interruption. Re‐ignition is the same physical event, but within one quarter power‐frequency cycle. Re‐strikes and re‐ignitions are characterized in terms of transients in current and voltage due to oscillations in the circuit. The arc is macroscopically characterizedintermsofitsduration,reach(perpendiculardistanceofoutermost arcpositiontoalineconnectingthecontacts),type(repetitiveorcontinuous),the energyinputfromthecircuitduringthere‐strike,arcvoltage,arcbrightness,and so forth. Understanding the basic phenomena upon re‐strike or re‐ignition is essentialintheunderstandingofcapacitivecurrentinterruption.Itwilltherefore beaddressedindetailinthischapter. The equivalent circuit for capacitive current interruption is shown in Figure3.1[1]. The DS is marked with D; Rs, Cs and Cl stand for resistance, capacitanceatsupplyandloadsiderespectively;idisthecurrent;usisthephase‐ to‐groundsupplyvoltageofthenetwork.ud,ucsanduclarevoltageacrosstheD,Cs and Cl respectively. The short‐circuit inductance is based on the short‐time withstandcurrentforwhichtheDSisrated:Ls=Uline/(√3ω.Isc)withIscistheshort‐ timewithstandcurrent;Ulineisthermsofthesourcephase‐to‐phasevoltage.For example,Isc=40kA,Uline=145kV,ω=2πf(f=50Hz),Lsisabout7mH. 36Chapter3 Figure3.1.BasiccircuitdiagramforcapacitivecurrentinterruptionwithaDS. Before the interruption starts, the DS is closed. The circuit of Figure 3.1 is energized by a harmonic source us = Em sin(pt+θ), with P the angular power‐ frequency,Emtheamplitude,andθtheinitialphaseangle.Thecurrentidandthe voltageacrossthecapacitancesCsandCl(denotedasucP)are: ucp E m sin(P t ) (3.1) ucs ucl ucp i C E cos( t ) d l P m P It is assumed that the impedances Rs and ωPLs are much smaller than 1/(PCs,l), meaning that the voltage ucp is very close to us. Theoretically, if Cs is very large, thenidisreducedbecauseofavoltagedropacrossRsandLs(Csisnormallyinthe orderofafewtensofnanofaradinpractice). WhentheDSinterruptsthecurrent,thecircuitinFigure3.1isseparatedintotwo parts abruptly. The supply circuit (left of the DS, consisting of Rs, Ls, Cs) remains energizedwithus.SincetheimpedanceofCsismuchlargerthantheimpedancesof RsandLs,ucsremainsclosetothesourcevoltageus.TherightpartoftheDS,i.e.Cl, hasnodischargepathandthevoltageuclacrossClremains1p.u.duetotrapped charge. After the arc extinguishes at current zero, the RV starts to rise. The RV across the DS ud, is the difference between ucsand ucl arising after interruption (ud=ucs‐ucl).Thedielectricstrengthoftheairgap,denotedasur,startstorecover simultaneously.Whentheincreasingvalueofudexceedsur,thegapre‐ignitesor re‐strikesandthearcre‐establishes.ThisprocessisshowninFigure3.2. In principle the arc lasts no longer than half a power cycle and extinguishes temporarilywhenthearccurrentcrosseszero.Thecircuitisseparatedintotwo partsagainuntilthenextre‐strikeorre‐ignitionoccurs,butgenerallyatahigher voltage because the contacts have separated further after a half cycle. The interruption process therefore consists of repeated arc extinctions and re‐ strikes/re‐ignitions. Finally, this sequence comes to an end and the arc BasicCircuitAnalysis37 Voltage (kV) ucs Arc restrikes 100 u cl 0 u current zero d −100 0 5 10 time (ms) 15 Figure3.2.Illustrationofthemomentswhenthearcextinguishesandre‐strikes. 20 extinguishescompletelywhenthedistancebetweentheDScontactshasbecome sufficientlylargeandtheRVremainsbelowthedielectricstrengthofthegap[2]. Whenthisisnotthecase,thearclastsaslongasthepowersupplyisactiveoras long as external influences (wind) do not extinguish the arc. In practical situations, normal interruption should be achieved before the blades reach a 90 degreeanglewithrespecttotheoriginalclosedpositionofthemainblades.After this moment, due to excessive elongation and reach, the arc may endanger the insulationofthestationarrangement. 3.2Three‐componentsanalysis Uponeachre‐strike,thevoltagesucs,ucl,udandthecurrentidhaveoscillationsat distinctfrequencies.Ahigh‐frequency(HF)componentarisesafterre‐strikewhen the voltages across source and load side capacitances are equalizing. After this process, the voltages ucl,ucs are equal and a voltage drop arises across Ls, which causes a medium‐frequency (MF) oscillation in the circuit. As the HF and MF oscillationsaredampedout,thepower‐frequency(PF)remains.Thesefrequency componentswillbeanalyzedasfollows. 3.2.1High‐frequency(HF)component At the instant of re‐strike the voltages ucs and ucl will equalize through a high‐ frequency oscillation. The voltage UE, remaining across both Cs and Cl after equalizationiscalculatedfromconservationofchargestoredinbothcapacitances: U C U cl Cl [3], where Ucs, Ucl are the initial voltages across Cs, Cl at the U E cs s C s Cl momentoftherestrikerespectively.IntheHFcircuitmodelshowninFigure3.3, the power supply is neglected since it is decoupled by the high impedance of inductanceLsfortheHFoscillation.RH,LHrepresentthehigh‐frequencyresistance and inductance of the circuit formed by the capacitances, the DS and the arc 38Chapter3 Figure3.3.High‐frequencyequivalentcircuitdiagram. withintheHFloop.ucs,uclandiHdenotethevoltageacrossCs,Clandthecurrentin thehigh‐frequencyloopcircuitrespectively.TheHFoscillationlastsonlyashort time(typicallylessthanafewtensofmicroseconds),duringwhichitcanleadtoa hightransientcurrentthroughcircuitandarcandmoderatetotransientvoltages acrossbothcapacitances. The voltage across the air gap just before re‐strike at time approaching t= 0 is denoted as Ur (= Ucs‐ Ucl). At this voltage the air gap breaks down. Assuming for thetotalcircuitresistance RH 2 LH ,andtheequivalentcapacitanceasseenby CH thearc: CH CsCl ,itisfound: Cs Cl 0H Ht Ur ucl 1 C C e sin(Ht H ) UE l s H (3.2) U t r H i t e sin( ) H H LHH Similarly, the voltage ucs across the source side capacitance at HF can be calculated: ucs Ur 0H Ht e sin(Ht H ) UE (3.3) 1 Cs Cl H where H RH 2LH , 0 H 1 LH C H , H 2 0 H 2 H 2 , H = arctan H . H The parameter H is the HF damping constant. The HF oscillation angular frequencyHdependsmainlyontheinductanceandtheseriesconnectionofboth capacitances. This frequency can be up to several megahertz. The current magnitude can reach values up to 7kA in the tests. The oscillation across the BasicCircuitAnalysis39 capacitances eventually ceases and settles at a quasi steady state value UE. The maximumvoltageacrossCsandCl,whichoccursincaseofnegligibleHFdamping, Ur Ur is: max(ucs) = U E and max(ucl) = U E , respectively. The 1 C s Cl 1 Cl C s risespeedofthecurrentdependsonthebreakdownvoltageandtheinductancein HFloop:max(di/dt)=Ur/LH. Atthesmallercapacitance, CsorCl,arisesthe highermaximumvoltage andboth voltages across the capacitances are proportional to the re‐strike voltage Ur. Further, Ur increases with increasing contacts distance, which means the HF voltagesacrossthecapacitanceswillprimarilybecomelargestjustbeforethearc extinguishescompletely.ThemaximumHFvoltagethatcanoccuris3Em,incase Ur=2EmandeitherCs<<ClorCs>>Cl.ThemaximumcurrentthroughtheDSinthe HF loop is Ur CH LH (neglecting HF damping). Obviously, iH increases with increasingUrandCHanddecreaseswithincreasingLH.ThusUrandCs/Cl(orCl/Cs) are key parameters that affect the HF transient behaviour on re‐strikes. The HF transientsinthecurrentarecandidatestocauseelectro‐magneticinterferencein secondarysystems, asitwillbereportedinChapter5.Giventhe high‐frequency and amplitude derivatives of the current, up to a few kiloamperes per microsecondsinlatertests,itcancoupleinductivelywithneighbouringcircuitry. Also,itprovidesasignificantpowerinputintothere‐strikingarc,albeitduringa limitedtime. 3.2.2Medium‐frequency(MF)component Upon re‐strike, also a "medium‐frequency" (MF) oscillation starts. The MF component,whichlastsaboutafewmilliseconds,alsocausesatransientvoltage andcurrent.Atthisstagethevoltageacrossthearcisneglectedandtheanalysis startsafterdecayoftheHFtransient,withvoltageofeachcapacitorequalizedto UE.LHinFigure3.3isneglectedaswell,becauseitsequivalentimpedanceismuch smaller than the capacitance’s impedance at MF. Therefore, Cs and Cl with an identicalinitialvoltageUEareinparallelandtheequivalentcircuitofFigure3.1at MF applies. Because of the charge redistribution during the HF oscillation, the voltagesacrossCsandClhavechanged,andwilldischargeviaLsandRsduringthe durationoftheMFoscillation. The instant of re‐strike is again taken as t = 0. On the time scale of the MF oscillation, us can be treated as a constant (Em sinθ). Similarly as for the HF analysis,uCM(thevoltageacrossCsandCl),andiM(thecurrentthroughtheDS)can bedetermined: 40Chapter3 0M Mt Ur uCM Emsin 1 C C e sin(Mt M ) s l M (3.4) U 1 t r i M e sin(Mt ) M (1 Cs Cl )2 LsM where M RS ,0M 2LS 1 ,M 2 0M 2 M 2 , M arctan M . LS (C s Cl ) M TheoscillationfrequencyMmainlydependsonthesumofbothcapacitancesand the inductance Ls. In general the frequency of this oscillation is in the order of severalkilohertz.Similarlyasforthehigh‐frequencytransient,thevoltagesduring the MFoscillation across both capacitances with initial voltages UE are damped due to the equivalent resistance in the loop and finally reach the value us. The Ur .Itincreaseswith maximumvoltageacrossthecapacitancesis Em sin 1 C s Cl increasing Ur, ratio Cl/Cs and Em Similar to the HF oscillation, the maximum theoretical voltage is 3Em. The maximal current during the MF oscillation is: Ur 1 max(iM) ,whichdependsonUr,Cs/ClandLsaswell,andscales (1 Cs Cl )2 LsM withUr. 3.2.3Three‐componentssynthesis AfterHFandMFcomponentshavedampedout,onlythePFcomponentremains. Sincethetimeconstantsinvolvedarehighlydistinct,theHFcomponentvanishes onthetimescalefortheMFoscillationandtheMFoscillationhasdisappearedon the timescale for PF. The initial voltage, at which the MFoscillationstarts, is the finalsteadystatevoltageafterHFoscillationdecayandtheinitialvoltageforthe PFoscillationisthefinalsteadystatevoltageoftheMFoscillation. Inordertoquantifythecompletetransientbehaviour,thethree‐componentsare combined. The voltage ucl across the load side capacitance and the current id flowingthroughtheDSonre‐strikecanbewrittenas: 0H H t 0M Mt Ur Ur ucl 1 C C e sin(Ht H ) 1 C C e sin(Mt M ) Em sin(Pt ) l s H s l M (3.5) U U 1 t t r i r e H sin( t ) M e sin( ) cos( ) t C E t H M l P m P d (1 Cs Cl )2 LsM LHH BasicCircuitAnalysis41 Voltage (kV) 200 HF MF 100 0 PF −100 −200 u 0 u cl 5 cs u 10 time (ms) d 15 20 Figure 3.4. Simulated wave shapes of ucs, ucl, ud during one power frequency cycle upon re‐ strike. Equation(3.5)containsthethreefrequencycomponentsinthevoltageacrossthe loadsidecapacitanceandinthecurrentthroughtheDSarc.Thevoltageacrossthe sourcesidecapacitancecanbecalculatedinasimilarmanner.Itturnsoutthatthe transientvoltagesandcurrentsdependontheairgapbreakdownvoltage,voltage supply level, the ratio of Cs/Cl and so forth In the following sections the distinct frequency contributions to overvoltages and currents will be discussed on the basisoftheanalysisaboveandcomparedwithexperimentaldata. 3.3Simulationofre‐strikes For illustrating the re‐strike phenomena in voltage and current waveforms a simulation is performed using MATLAB Simulink (SimPowerSystems). The parameters taken are: Em=70√2kV, Ls=10mH, Cs=10nF, Cl=100nF, the resistanceforarcatre‐strikeistakenas20ΩandLH=15µH.Initially,theDSisin closed position. The DS gap starts to recover at t=5ms, and re‐strikes at t=10ms. The simulated wave shapes for the current id, voltages ucl, ucsand ud together with their zoomed in of the HF and MF components, are plotted in Figures 3.4‐3.6. In Figure 3.4, the three‐components are indicated with arrows. According to Section 3.2, the charges of the source and load side capacitance Cs andClexchangeduringthehigh‐frequencyperioduntiltheequalizationvoltageUE is reached, shown in Figure 3.5. The medium‐frequency component, after the high‐frequencyoscillationhasdisappeared,isdepictedinFigure3.6.Inasimilar way, the wave shapes of the current id and its high and medium‐frequency componentsareshowninFigures3.7‐3.9. It can be observed that the high and medium‐frequency components last about 10µsand2‐3msrespectivelyinthissimulation.Duringtheoscillation,thereisa transient current with a peak value up to 2kA upon re‐strike at t=10ms. The observedovervoltageacrossthesourcesidecapacitanceCsis1.5p.u. 42Chapter3 Voltage (kV) 100 UE 0 ucl −100 0 5 10 time (μs) ucs 15 ud 20 25 Voltage (kV) Figure 3.5. Simulated wave shapes of ucs, ucl, ud at high‐frequency (zoomed in from approximately9.96msto10.06msinFigure3.4). ucl 100 u u 0 cs d −100 10 10.5 11 11.5 12 12.5 time (ms) Figure 3.6. Simulated wave shapes of ucs, ucl, ud at medium‐frequency (zoomed in from approximately9.5msto12.5msinFigure3.4). 2 0 −2 −4 0 5 10 15 20 time (ms) Figure3.7.Simulatedwaveshapesofidoveronepowerfrequencycycleuponre‐strike. 10 10.01 10.02 Current (kA) 9.5 Current (kA) 1 0 −1 −2 10.03 10.04 10.05 10.06 time (ms) Figure 3.8. Simulated wave shapes of id at high‐frequency (zoomed in from approximately 10.00msto10.06msinFigure3.7). Current (kA) BasicCircuitAnalysis43 0.2 0.1 0 −0.1 −0.2 9.5 10 10.5 11 11.5 12 time (ms) Figure3.9.Simulatedwaveshapesofidatmedium‐frequency(zoomedinfromapproximately 9.5msto12.0msinFigure3.7). 3.4Re‐striketransientsindistributedelementloadcircuits InthecircuitofFigure3.1onlylumpedcomponentsareconsidered.Asmentioned in Chapter 1, the capacitive current to be interrupted arises from substation components,suchascurrenttransformer,capacitivevoltagetransformer,butalso fromdistributedelements,suchasbusbars,overheadline,andpowercables. If the physical dimensions of the considered components are much shorter than the wavelength corresponding with the current and voltage oscillations, lumped parameters can be used to model the power system. Otherwise, a distributed parameter approach has to be adopted. The transmission line specific (per unit length) parameters L, C and R (G is omitted since the air is considered a perfect insulator) are uniformly distributed over the length of the line. For the steady stateoperation(power‐frequency,50Hzor60Hz),thetransmissionlinescanbe represented by lumped parameters. But for transient behaviour the lines it may benecessarytoberepresentedbydistributedparameters. During the capacitive current interruption by a DS, transient phenomena occur withhigh‐frequencycontent,duetotherepeatedbreaksandre‐strikes.Therefore, a distributed parameter simulation for transmission lines is considered in this section.Thesimulatedresultsofalumpedparameterandadistributedparameter approachofatransmissionlinearecompared. ThevalueofthelumpedelementparametersinFigure3.10aretakentoproduce thesamepower‐frequencyresultsasthoseinFigure3.1.Theunloadedoverhead line is simulated with distributed parameters. A line length of 50km is taken to match the interrupted current Id (PF current of 9A) of the lumped circuit (line capacitanceandinductanceare8.48nF/kmand1.33mH/km,matchingalumped load side capacitance of Cl = 424nF). The simulated results are shown in the Figures 3.11‐3.13, where the wave shapes of the interrupted current id and the voltageacrosstheDSud,arepresentedrespectively:id1,ud1aresimulatedresults from simulation with distributed elements, and id2, ud2 are simulated results are obtainedfromsimulationwithlumpedelements. 44Chapter3 400 i id1 Current (A) d2 200 0 −200 15 20 25 30 time (ms) Current (A) Figure 3.10. Simulated circuit for capacitive current interruption with a transmission line representedbydistributedparameters. 100 0 −100 22.4 22.6 22.8 23 23.2 time (ms) Figure3.11.(Left)waveshapesofthecurrentid1,id2throughtheDSwithdistributedlinesand lumpedcapacitancerespectivelyand(right)expansionofid1between22.4msand23.2ms. 4 d1 2 0 −2 u d2 −4 −6 20 21 22 23 24 25 time (ms) Figure 3.12. Wave shapes of the voltage ud1, ud2 across the DS with distributed lines and lumpedcapacitancerespectively. Voltage (kV) u The wave shapes of current and voltage show that the travelling waves are reflected within 0.33ms. There are transient phenomena on each reflection moment(seeFigure3.11,3.12).Nevertheless,thereishardlyanyhigh‐frequency component at re‐strikes in the voltage and current waveforms with the distributed parameters, which consequently reduces the total amount of energy BasicCircuitAnalysis45 inputintothecircuituponre‐strike.Thisisbecausethehighsurgeimpedanceof the line (in this example 390Ω) limits the HF re‐strike current compared to the much lower surge impedance of the lumped capacitor. Therefore, to interrupt a circuit with lumped parameters is a more severe task as compared to a configurationwithdistributedparameters.Intheanalysisinthischapter,andalso intheexperimentalarrangementdescribedinthefollowingchapters,circuitswith lumpedelementsareapplied.Thispointshouldbepaidattentionto,becauseofits impactontesting,wherenormally(lumped)capacitorbanksareused. 3.5Conclusion Themainobservationswithrespecttocalculationsandsimulationsoncapacitive currentinterruptionwithhighvoltageair‐breakDSsare: CapacitivecurrentinterruptionbyaDSconsistsofrepeatedbreaksandre‐strikes asaconsequenceoftheinteractionbetweenarcandcircuit. The transient upon re‐strike consist of three‐components: high, medium and power‐frequencycomponent.Duringthehigh‐frequencycomponent,thetransient period is in the order of a few tens of microseconds. The medium‐frequency component is in the order of a few milliseconds. The overvoltage across the capacitancesatbothsidesoftheDScanbeupto3p.u.theoretically.Thepeakof thetransientcurrentmaybeuptoafewkiloampereswithinafewmicroseconds. Theratioofsourceandloadsidecapacitance,there‐strikevoltageandthecurrent to be interrupted are the key factors that influence the transient voltage and currentmagnitudesuponre‐strike. Thecircuitwithlumpedelements,whichisadoptedinthisthesis,isaworstcase approach. Capacitive current interruption of transmission lines where a distributedparameterapproachappliesislesssevere. References [1] IEC Technical Report IEC/TR 62271‐305, "High voltage switchgear and controlgear– Part 305: Capacitive current switching capability of air‐insulated DSs for rated voltagesabove52kV",Nov.2009. [2] L.vanderSluis,"TransientsinPowerSystems",Chichester:JohnWiley&Sons,2001. [3] D.F.Peelo,"Currentinterruptionusinghighvoltageair‐breakDSs",Ph.D.dissertation, Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven,2004. 46Chapter3 ExperimentalSetup47 Chapter4 ExperimentalSetup Experimentsinthisthesisareperformedatseverallocationsemployingdifferent typesofvoltagesources.Thebasiccircuitdiagramforalltestsiteswaspresented inFigure3.1.Figure4.1depictsthegeneralsetupincludingthemeasuringsystem schematically. The voltage supply is represented by an ideal voltage source us with series resistance Rs representing losses and inductance Ls. Their values depend on the source type being used (see Section 4.1). Capacitances Cs and Cl stand for the sourceandloadsidecapacitorbankswithaDSinbetween.Thecapacitancevalues canbevariedforthetestseries.Thevoltagesucs,uclarethevoltagesacrossCsand Cl,respectively.ThesevoltagesaremeasuredbythecapacitivevoltagedividersD1 andD2incombinationwiththehighvoltageprobesD3andD4(Section4.2.1).The differential signal is used to determine the voltage across the DS (Section 4.2.2). The current through the DS is indicated by id. Current transformers CT1 and CT2 measure the current through Cl (Section 4.2.3). Two current transformers are applied to cover the dynamic range of different frequency components simultaneously (below 1A to several kiloamperes and from 50Hz to several megahertz),denotedasipfandihf,respectively.Voltagesandcurrentsarerecorded byadataacquisitionsystemincludingfourdigitizers,CH1‐4,eachhavingasingle input channel(Section 4.2.4). In addition to the electrical signals, arc images are recordedbymeansofahigh‐speedcamera,whichisinstalledonthesameheight as the DS blades (Section 4.2.5). The opening characteristics of the DS is determinedinSection4.3. 4.1Highvoltagesource Two different types of high voltage source are employed. Depending on the test sites,apowertransformeroraresonancehighvoltagesourceisused. A short‐circuit generator plus a transformer are available in large scale laboratories,suchasKEMAHighPowerLaboratoriesinArnhemandPrague.The equivalenttestcircuitisshowninFigure4.2.AsourceinductanceLsisusedwitha value up to a few hundreds of millihenry. Air gaps are used to protect the capacitors bank, but are not shown in Figure 4.2. These setups supply up to 173kVrms phase‐to‐ground voltage, and up to 27A current during the experiments. The specific test configurations employed for different series of experimentswillbepresentedinChapters5,6and7. 48Chapter4 Figure4.1.Experimentalcircuitincludingelectricalandopticaltransducers. Figure 4.2. Test circuit with source in KEMA HPL. G: short‐circuit generator; M: master breaker; Ls, Rs: reactor and resistor at source side; S: make switch; TR: short‐circuit transformer. In the high voltage laboratory at the Eindhoven University of Technology a resonantsystem(Hipotronics)isavailableashighvoltagesupply.Itstopviewand equivalentschemeareshownFigure4.3(leftandright,respectively). TheresonantsystemincludesanadjustablehighvoltagereactorLs,acapacitorC and an exciter transformer. The variable auto transformer T1 controls the transformer T2, which supplies power to the resonant circuit, and it isolates the testspecimenfromtheline.Twotuneablereactors,whichcanbeconnectedeither in series or in parallel, make up the source inductance. Each of them has an inductancewithavalueintherangeof400H‐10kH,andresistance(measuredat DC) of about 1kΩ. Each reactor is designed for a voltage of 300kVrms. The AC ExperimentalSetup49 Figure 4.3. (left) Laboratory source for capacitive current interruption with a DS at TU/e; (right)simplifiedschematicsofHipotronicshighvoltagesource. sourcepoweringtheresonantcircuit,indicatedwithuoutinFigure4.3(right)isat maximum 22kVrms. The capacitance Cs is chosen either 2nF or 4nF by series connectionsoffourortwo8nF,150kVcapacitors.Differentcombinationsoften available 16nF high voltage capacitors can be made to adjust the (load) capacitanceCltothedesiredcurrent.Eachofthesecapacitorshasaratedvoltage of150kVrmsandtanδ<5×10‐3.Thetotalcapacitance(denotedasCinFigure4.3 (right)) before opening the DS includes the capacitances at both sides of the DS, including capacitances from voltage dividers, bushings, etc. The reactor Ls is adjusted to compensate the total capacitive reactance and thereby tuning the circuit. The quality factor Q of this system can reach values of 50‐80 for high qualitycapacitiveloads.Thesystemiscapabletosupplya50Hzhighvoltageupto 300kV.Ideally,amaximumcurrentof2Aor4Acanbeachieved,dependingon whetherthereactorsareinseriesorinparallel.Duringtheexperiments,firstthe systemisenergizedwiththeclosedDSatalowlevelofus.Next,theinductanceLs istuneduntilLsandCbecomeresonantat50Hz.Oncethesystemisinresonance, the desired voltage level is chosen and the interruption experiment can commence. A drawback of the resonant system is that it can be tuned into resonance at power‐frequency while the DS is in closed position only. When the DS starts opening, the resonant condition still remains because the load capacitor is connected through the conductive arc at the very beginning. The quality of the circuit drops, however, due to the arc resistivity. Since the arc extinguishes repeatedlyateachcurrentzero,theresonanceislostduringthevoltagerecovery period. After re‐strike/re‐ignition, the resonance condition is met again but the source voltage has started dropping. Figure 4.4 shows the decay of measured source voltage at different initial high voltage levels. It demonstrates that the systemvoltagedecaysrapidlyafter10to20power‐frequencycycles. 50Chapter4 200 85kV 100kV 120kV 140kV 180kV voltage(kV) 150 100 50 0 −100 0 100 200 time(ms) 300 400 500 Figure 4.4. Source voltage decay by loss of resonance and energy dissipation by the arc, measuredresults. u res (kV) 100 without arc R =50Ω a 50 Ra=300Ω 0 −50 −100 0 5 10 15 20 25 30 time (ms) Figure 4.5. Resonant voltage reduction for different values of the arc resistance, simulated results. Thearcresistanceisintheorderofseveralhundredohms(Section5.5.3),i.e.only in the order of 10‐30% of the dc resistance of the resonant circuit (1kΩ). As simulatedinFigure4.5thisisexpectedtoreducethesourcevoltagewiththesame amount, since the circuit quality is inversely proportional to the total circuit resistance.Theinitialvoltagereductioncanbeascribedtotheenergydissipation inthearc,sincethearcinterruptiontimeisstillnegligible.Thevoltagedropsafter tencyclesislargerthancanbeexpectedbasedonthearcenergydissipation.This reduction must mainly be ascribed to the longer arc interruption periods with increasedairgapdistancebetweentheDSbladesresultinginlossofresonance. Anotherdrawbackoftheresonantcircuitisthatthiscircuitdeviatesstronglyfrom thepracticalsituationinwhichcircuitshavealowvalueofsourceinductance(or ahighshort‐circuitpower).Thiscausesthemedium‐frequencycomponentduring the transients to be virtually absent. In this situation, the experiments definitely ExperimentalSetup51 Figure 4.6. Scheme of the voltage dividers (D1 or D2), ui:uo = 10.7:1 and their mounting positionsinparalleltotheDSinsulators. do not match practical situations. Specifically, compared to the experiments performedinthelargetestlaboratories,thearcdurationandlengthisshorterand the arcing/switching is accompanied with lower overvoltage levels. The experimentsdoneattheTU/ehighvoltagelaboratorythereforeareusedforstudy of the principal (microscopic) phenomena, but are not suitable to give realistic macroscopicresults,e.g.arcduration. 4.2Measurementsystem Theelectricalsignalsandopticalarcimagerecordingsystemaredescribedinthis section. 4.2.1Voltagemeasurement The voltage measuring systems at both DS sides are identical from design. Each capacitive divider consists of ten Murata 2nF, 40kV capacitors in series (D1 or D2),anda1000:1NorthStarhighvoltageprobe(D3orD4),seeFigure4.1[1],[2]. The ten divider capacitors for both dividers are carefully selected to realize a divisionratiobeingascloseaspossibletoeachother.Thecapacitivedividers(C1– C10) are shown in Figure 4.6 with division ratio ui:uo = 10.7:1. The dividers are mounted in parallel to the DS insulators. The determination of the divider response up to the cut‐off frequency is accomplished by injecting a unit step voltagewitharisetimeof20ns.Thewaveshapesoftheinputandoutputsignals and their transfer function (ui/uo) are shown in Figure 4.7 and 4.8, respectively. Theresultsshowthatthecapacitivevoltagedividershavebandwidthsuptoabout 5.5MHz.Intheexperiments,thehighesttransients(HFoscillation,seeChapter3) havefrequenciesupto1MHz(seeFigure5.6),whichhasattenuationoflessthan 2%(forcut‐offfrequencyof5.5MHz). 52Chapter4 ui (V) 40 20 rise time 20 ns 0 2 2.5 3 time (μs) 3.5 uo (V) 4 2 rise time 40 ns 0 2 2.5 time (μs) 3 3.5 Figure4.7.Injectedstepfunction(top)andresponse(bottom)ofthedivider. i o |u /u | 15 10 5.77M 1M 10M 1M Frequency (Hz) 10M ∠ui/uo (°) 5 100k 180 90 0 −90 −180 100k 3dB Figure4.8.Frequencyresponseofthecapacitivevoltagedividers(ui/uo). In order to lower the output voltage from the Murata capacitor stack divider furthertobeabletosupplyittothedigitizersystem,commercialvoltagedividers (probes)areused.Thehighvoltageprobe,(typeNorthStar,PVM‐5),isbasicallya parallelRCvoltagedividerdesignedtoproducepreciselyattenuatedsignalsover a wide bandwidth. The circuit diagram is shown in Figure 4.9. The divider network contains a high voltage branch consisting of a parallel capacitor and resistorandalow‐voltagenetwork,whichconsistsofaparallelRCnetworkanda compensationcircuit.Thehighvoltagesectionofthevoltagedivideriscontained in an oil‐filled housing. The low‐voltage section is contained in a separate box underneaththedivider’sbase.Itsmaincharacteristicsare:maxDC/pulsevoltage: 60/100kV; bandwidth: 80MHz; cable impedance: 50Ω; input R/C: 400MΩ/13pF;dividerratio:1000:1. ExperimentalSetup53 Figure4.9.Circuitdiagramofthevoltagedivider. 4.2.2Differentialvoltageaccuracy AccuratevoltagemeasurementsofthevoltageacrosstheDS(ud)areessentialfor arc voltage and re‐strike voltage analysis. Therefore, the differential voltage ud=ucs‐uclmustbeavailablewiththehighestpossibleprecision.Thisisachieved by comparing the voltages ucs and ucl in closed position of the DS. The relative voltagedividerratioscanbeadjustedbycomparingthedivideroutputsbeforeDS openingforeachmeasurement,wherethetwooutputsshouldbeequal. Figure 4.10 shows the wave shapes of voltagesucs, ucl and ud together with the expansions of amplitude and phase of ucs, ucl before interruption. Ideally, they should be completely identical, but due to tolerances in the division ratios of dividers and probes minor differences occur. The zoomed waveforms in Figure 4.10ofucsanduclshowthatthepeakamplitudesandthephaseanglesdiffer.An amplitudecorrectionwithafactor0.9961andaphaseshiftof0.021msisapplied tomatchthewaveshapesofucs,ucltoobtainud.Voltagewaveshapesucs,uclandud expansionsnearvoltagepeakandnearzero‐crossingofucs,uclaftercalibrationare plottedinFigure4.11.Theerrorinvoltagemagnitudeofudhasdroppedfrom1kV toabout0.1kV. 4.2.3Currentmeasurement Insteadofmeasuringiddirectly,thecurrentthroughClismeasuredwiththehigh‐ frequency current transformers CT1, CT2. Because the impedance of Cl (capacitanceintherangeseveralnanofaraduptohundredsofnanofarad)ismuch lower than the impedance of the voltage divider (with a total capacitance in the orderof200pF),theerrorforcurrentdifferencethereforeismaximum2.5%for Cl=8nFfortheTU/elaboratorysetup.Attheotherexperimentalsites,thiseffect isusuallysmallerbecauseofthelargercapacitorvalues(upto900nF). 54Chapter4 1 ud (kV) ucs, ucl(kV) 200 0 0 −1 −200 0 10 20 time (ms) 10 30 20 30 40 time (ms) ucs, ucl(kV) 131.5 130.5 4.2 ucl u 0.2 ucs, ucl(kV) 132 131 132.5 cs 0 ucl ucs −0.2 4.4 4.6 time (ms) 4.8 9.47 9.48 9.49 time (ms) 9.5 cs cl ucs, ucl(kV) u , u (kV) cs cl ud (kV) u , u (kV) Figure4.10.Voltagesucs,uclandudandzoomedinnearvoltagepeakandnearzero‐crossing ofucs,uclbeforeinterruptionwithoutadjustment. 1 200 0 0 −1 −200 0 20 40 0 10 20 30 time (ms) time (ms) 132 0.2 ucl 131.5 ucs 0 131 130.5 4.2 4.4 4.6 time (ms) 4.8 −0.2 9.47 9.475 time (ms) 9.48 Figure4.11.Voltagesucs,uclandudzoomedinnearvoltagepeakandnearzero‐crossingofucs, uclafteradjustmentofamplitudeandphaseshift. Two similar current transformers (CT1, CT2, each connected to a separate digitizer)areemployedformeasuringbothpower‐frequencyandhigh‐frequency components. Each transformer (type: Pearson 110) has the following properties (according to [3]): sensitivity: 0.1V/A; output resistance: 50Ω; maximum peak current: 5kA; maximum rms current: 65A; rise time 20ns; ‐3dB cut‐off ExperimentalSetup55 Figure4.12.Locationofcurrenttransformers. frequencies: approximately 1Hz and 20MHz. Figure 4.12 shows both current transformersandtheirlocationaroundtheleadconnectingCltotheground. 4.2.4Isolateddigitizeranddataacquisition Thevoltageandcurrentsignalsaresuppliedtofourchannelsofahighresolution data acquisition system (LDS Nicolet HV 6600 digital system). It consists of isolated digitizers ("FrontEnd"), optical fibre, Genesis mainframe and software "Perception" (see Figure 4.13). The FrontEnds can be used in a heavily EM polluted high voltage environment owing to their shielding and electrical insulation. Each FrontEnd has an input impedance of 1MΩ/38pF, and its analogue bandwidth ranges up to 25MHz. The sampling frequency is up to 100MS/swithaverticalresolutionof14bit.Theopticalfibrelinktransmitsthe signals in real time to the Genesis (central unit) mainframe, avoiding EM interference in the connection between transducer and central data storage unit and allowing the installation of transducers on locations of any potential. Each FrontEnd is remotely programmable. The Genesis mainframe and a PC are connected via an IP network address. The software "Perception" controls the hardware, records data, allows viewing of the signals, and performs basic signal post‐processing. Final data analysis is carried out with programs in MATLAB script. 4.2.5High‐speedcamera A high‐speed camera (Model Centurio C100, from Lot‐Oriel Europe, shown in Figure4.14)withframememory2GBisemployedtorecordthearcimages[4].It is based on a large format CMOS detector with a maximum resolution of 1280×1024 pixels. A CMOS sensor converts the photons from the light coming throughthelensintoelectriccharges,whicharethenprocessedbytheelectronic circuitry. The resolution ranges from 64x12 pixels, with a maximum recording time of 27s, at a rate of 106 frames/s to 1280×1024 pixels, with a maximum recording time of 3.6s, at a rate of 424 frames/s. The sensitivity is adjustable 56Chapter4 Figure4.13.DataacquisitionsystemwithisolatedFrontEnds,opticalfibrelinks,Genesis mainframeandPC. between50and2000ASA.Theexposuretimeisadjustablefrom2μsto520ms. The camera works in connection with a PC using the USB2.0 interface through software, which allows precise triggering, recording, and post‐processing. The optical format is 17.9x4.3mm. An F‐mount Sigma lens is applied, with filter size 82mm,lensconstruction13‐14,angleofview84.10‐34.30,minimumaperture32, minimumfocusingdistance40cm,magnification1:3.8andfocallength24‐70mm. Duringmeasurements,electricalandopticalsignalsarerecordedsimultaneously. Theelectricalandopticalsignalsaresynchronizedthroughcarefullymatchingthe frame with the highest brightness with the most severe re‐strike moment in the measuredcurrentorvoltagewaveforms.Normally,thefinalre‐strikeischosen. Thedurationoftheelectricaltransientsignalsaccompanyingthebrightframeisa few tens of microseconds, much shorter than the exposure time of the frame, usually taken 1ms for a 1000frames/s recording speed. The timing clock accuracyofthecameraischeckedbymeasuringitssyncoutputandisfoundto deviate 0.06%, i.e. clearly less than variation in the 50Hz power‐frequency. The synchronization between framing rate and power‐frequency therefore is limited bythemomentaryvalueofthepower‐frequency. Thearclengthisestimatedbasedontwodimensionalprojectionofrecordedarc images. In space the arc propagates in all three dimensions due to the blades moving, air flow, magnetic field, presence of objects in the environment. The single high‐speed camera can capture only two dimensions. Ignoring the third dimension leads to certain estimation error. This error can be minimized with prudent camera positioning. The camera is positioned facing the central axis of the frontal view of the DS, in the same height as the main blades. The camera’s position secures that the most dominant directions of arc propagation are captured. ExperimentalSetup57 Figure4.14.High‐speedcameraforrecordingthearcimageinashieldingbox. 4.3Disconnectormainbladesopeningvelocity Theopeningcharacteristics,suchastheopeningvelocityofthemainblades,are based on measured the time dependent distance between the DS contacts. This distancedependsontheanglebetweenthemovingbladesandtheblades’original closed position. The elapsed time is determined by the rotation speed of the drivingshaftoftheDS(approximately40turns/s).Ontheshaftacylindricalsheet is mounted with 25 reflective slots generating 25 pulses per turn (about 1kHz). Thereflectedlightfromasmalllightsourceisdetectedbyasensor.Ittakesabout 210rotations(5s)toopentheDSmainbladesfromzerotoninetydegrees.The light sensor pulses are recorded and related to the mechanical movement of the bladecontactscalculatedfromthedrivingmechanismoftheDS. ThetopviewoftheDSisshowninFigure4.15.Themovementofthearmsofthe rotating parts is indicated in Figure 4.16. The opening mechanism is attached to theshaftviaagearboxandisdrivenbyanelectromotor.Therotationoftheshaft relates linearly to the central rotating strip, indicated as ‘1’ in Figure 4.15. The angleθ2istherotationangleofthestriprelatedtothepositionparalleltotheDS blades. The left end of this strip is mounted to a pipe indicated as ‘2’, which is mountedtoaframeattheleftsidewithontoptheinsulatorindicatedas‘3’.θ1is the rotation angle of this frame related to the position parallel to the DS blades. The pipe pulls this frame of the left insulator and rotates the insulator and meanwhiletheDSblade.Therightinsulatorismountedwithanotherpipetothe left insulator frame and therefore rotates similarly in opposite direction. All distancesandtheinitialangles(forclosedposition)θ1,0(¼πrad),θ2,0(0.186rad) aremeasured. 58Chapter4 θ1 θ2 Figure4.15.SchematicofthetopviewoftheDS. Figure4.16.RotatingangleoftheleftbladeinFigure4.16. The centre of the rotatingstrip is chosen as the origin of the coordinate system. Thepositionoftheconnectionpointtothepipeis: π∙ π∙ ; ∙ cos θ , ; ∙ sin (4.1) , Inthisequation‘p’isthenumberofpulses,withatotalof5250pulsestoopenthe DS completely. A similar expression can be obtained for the position of the connectionbetweenthepipeandtherotatingframe‘3’inFigure4.15: ; A ∙ cos ;B ∙ sin (4.2) , , Speed (m/s) ExperimentalSetup59 0.8 0.6 0.4 0.2 0 0 1 2 3 4 5 time (s) Figure4.17.OpeningspeedoftheDSinTU/eHVlaboratory. Here(‐A,B)indicatesthecentreofrotationwithrespecttothechosenorigin.The lengthofthepipebetween(x1,y1)and(x2,y2)isfixed,l=1305mm: (4.3) Because this holds for every pulse number p this must hold also for every angle . In this way the rotation of the insulator is calculated. From the position at every pulse and the time between the pulses the velocity of the DS blades is calculated.Figure4.17showsthatthecontactseparationspeedisinitially0.2m/s and increases up to 0.6m/s after 2s (this result is based on the 245kV rated voltagecentre‐breakDSfromHapam[5]). 4.4Conclusion The experimental setups with two different power supplies (resonant generator and short‐circuit generator) are described for different test locations. The measuring system includes voltage divider, current transformer, digital data acquisition system, and high‐speed camera. Two identical voltage dividers with 5.5MHzbandwidtharedesignedforthepurposeofmeasuringvoltageacrossthe maincontactsoftheDS.Aftercorrectiontheirdivisionratioisequalwithin0.07%. Thebandwidthofthecurrentmeasurementisupto20MHz.Transientcurrentup to 7kA is measured simultaneously with power‐frequency with sensitivity of 5mA.Ahigh‐speedcameraisemployedanditstakingframeupto5000frames/s, and a resolution of 256x254 pixels. The DS employed in the tests, opens with typicalspeed0.4m/s. References [1] [2] [3] [4] [Online].Available:http://www.murata.com/products/capacitor/index.html. [Online].Available:http://www.highvoltageprobes.com/faq.html. [Online].Available:http://www.pearsonelectronics.com/. [Online].Available:http://www.lot‐oriel.com. [5] [Online].Available:http://www.hapam.nl/. 60Chapter4 CurrentInterruptionMechanism61 Chapter5 CurrentInterruptionMechanism Thecurrentinterruptionofanair‐breakHVDSisinvestigatedexperimentallyby observation of electrical and optical signals in this chapter. From these observationstheinterruptionischaracterizedintermsofarcduration,arcmode, arc movement, arc brightness, and decay of afterglow. The electrical characteristics related to the free burning arc are derived as well. The major factors that control the DSs’ capability to interrupt capacitive current, such as blade contacts spacing, the re‐strike voltages, arc energy, and transient phenomena are studied in detail. In the end, based on experimental data, the principal transient process is described, and the transient voltages and currents arestudied. Experiments are carried out in different laboratories (see Chapter 4.1) with a centre‐breakDS.ThreeDSsaretestedwitharatedvoltageof145kV,245kVand 300kV respectively. Usually at each current level, the operation is repeated at leastthreetimes.ThelevelofthecurrenttobeinterruptediscontrolledbyCl(see Figure 3.1). The main configurations of the capacitors Cs and Cl are listed in Table5.1 with light, medium and dark shaded regions respectively. The experiments from 8A to 27A (at 90kV and Cs fixed at 50nF, Cl ranging from 170nFto935nF)inthedarkshadedregionareperformedwithanexternalfast openingcontacts(seeChapter7),indicatedbytheitalicandboldfontstyle.Some oftheseinterruptionsfailed.Theyareincludedinthistableinordertoillustrate differentarccharacteristicsforthischapter. 5.1Experimentalobservations Based on electrical signals and optical images, a global characterization is made. The interruption process is studied based on voltages across both capacitances andtheDS,andthecurrentthroughtheDS.Theopticalobservationincludesthe developmentofthearc,resultinginarcmotion,arcshape,andarcbrightness.The correlationoftheelectricalandopticalsignalsisalsopresented. 5.1.1Electricalarcsignalobservation Typicalwaveshapesoftherelevanttransientphenomenaduringtheinterruption process are shown in Figures 5.1‐5.8. Specifically, information of ucs (the voltage across the source side capacitance), ucl (the voltage across the load side capacitance)capacitance),id(thecurrentthroughtheDS),ud(=ucs‐ucl,thevoltage 62Chapter5 Table5.1.Interruptingcurrentforsourceandloadsidecapacitancecombinations. Cs Cl Id 4.3 8.0 10.7 16.0 17.0 19.3 32.0 38.6 40.0 48 88 121 133 170 183 218 277 312 346 433 519 588 692 796 865 935 2.0 4.0 1.5 6.0 20.0 60.0 50.0 0.23 0.6 0.60 0.60 1.2 1.0 1.0 2.4 1.0 1.0 2.1 1.1/2.1 2.1 2.1 1.1/2.1 2.1 0.5 1.4 2.6 3.5 3.9 5.0 5.3 6.3 8 9 10 12.5 15 17 20 23 25 27 100.0 2.1 Note:CsandClareinnanofaradsandIdisthecurrentinamperes.Thelightshadedregionis performedwithaDSratingof245kVattheTU/elaboratorysetupatvoltagesupto170kV; the medium shaded region corresponds to a DS rated for 300kV tested at 90kV or 173kV; "1.1/2.1" at Cl = 38.6nF means 1.1A at 90kV and 2.1A at 173kV; the dark shaded region corresponds to a DS rated for 145kV tested at 90kV. For the dark shaded region, the interruptionsupto6.3Aaresuccessful.Theexperimentsindicatedbythemediumanddark shadedregionarecarriedoutatKEMAlaboratories. across the DS) is plotted. The current id is simultaneously measured with a bandwidthof50kHz(ipf,coveringpower‐frequency)andabandwidthof20MHz (ihf, covering high‐frequency transient signals). The experimental parameters for the illustrations in Figures 5.1‐5.8 are: Em = 90√2kV, Cs = 1.5nF, Cl=40nF, Id=1.1A.Timet=0istakenasthemomentwhentheDSbladesstartopening. CurrentInterruptionMechanism63 ucs (kV) 200 0 −200 0 500 1000 1500 2000 Zoomed in ucl (kV) 200 0 −200 0 500 1000 time (ms) 1500 2000 Figure5.1.Voltages(ucs,ucl)crossthecapacitors. 200 ucs Extinction Restrike ucs, ucl (kV) ucl 100 0 −100 −200 1680 Overvoltage 1690 1700 1710 1720 time (ms) 1730 1740 1750 Figure5.2.Zoomedinbetween1680msand1750msofFigure5.1. Steps ud (kV) 200 100 0 −100 −200 0 500 1000 time (ms) 1500 Figure5.3.WaveshapeofthevoltageacrosstheDS:ud=ucs‐ucl. Zoomed in 2000 64Chapter5 Arcing RV rising 100 0 d u (kV) 200 −100 −200 1675 1680 1685 1690 1695 1700 time (ms) 1705 1710 1715 Figure5.4.Zoomedinbetween1675msand1720msofFigure5.3. Step ihf (A) 1000 0 −1000 0 500 1000 1500 time (ms) 2000 Zoomed in 1000 1000 ihf (A) ihf (A) Figure5.5.WaveshapeofthecurrentidscaledtoallowfullcoverageatHFcomponent. 0 −1000 Zoomed in 1880 1900 1920 0 −1000 1906.71906.81906.9 1907 1907.1 time (ms) time (ms) Figure5.6.(left)zoomedinbetween1860msto1930msofFigure5.5,and(right)zoomedin asinglehigh‐frequencycomponent. From the waveforms in these figures the following general phenomena are observed: ‐ The entire interruption process lasts about 2s. During this process the capacitivecurrentinterruptionconsistsofmultiple,periodicalarcextinctions and re‐strikes, which are clearly observed in the expanded wave shapes (Figures5.2,5.4,5.6and5.8).Thearcextinguishesasthearccurrentcrosses zero and re‐establishes during the succeeding half power cycle due to the collapseofthe(power‐frequency)RV.Transientphenomenaoccuruponeach re‐strike. First the HF component appears, followed by the MF and PF components (see Chapter 3). For example, the high‐frequency component showninFigure5.6,withfrequencyapproximately100kHz. CurrentInterruptionMechanism65 20 pf i (A) 10 0 −10 −20 Zoomed in 0 500 1000 1500 2000 pf i (A) time (ms) Figure5.7.WaveshapeofthecurrentrecordedatlowerfrequencyinordertofilteroutHF transients. 2 HF 1 0 MF −1 −2 580 585 590 595 600 605 time (ms) Figure5.8.Zoomedinbetween580msand610msofFigure5.7arccurrentipf. 610 It lasts about 0.1ms. The medium‐frequency component, shown in Figure 5.8, has a frequency of about 2kHz, lasting approximately 6ms. After the high and medium‐frequency oscillations have decayed, only the power‐ frequencycomponentremains. ‐ ‐ ‐ Duringtheinterruption,overvoltagesacrossthecapacitorbanksatbothDS sides and between the DS main blades arise. Upon re‐strike a high peak currentthroughtheDSoccurs.Intheexample,themaximumovervoltageof ucl,uclis1.86p.u.andthemaximumHFtransientcurrentisabout1.6kA(see Figures5.5and5.6). The expansion of the voltage ud (Figure 5.4), and the power‐frequency component of the current (Figure 5.8) show the periods with arc and intervals in between, where the RV builds up at each half power‐frequency cycle. Thearcre‐strikeswhenevertheRV exceedsthedielectricstrengthof the air gap. It extinguishes completely when sufficient air gap distance betweenthemainbladecontactsoftheDShasbeenreachedtowithstandthe RV. Values of udandid(Figures5.3and5.5) oneachmomentofre‐strikedonot rise monotonously with the increasing contacts distance, but "steps" are observed. The re‐strike voltage is not only determined by the gap distance 66Chapter5 betweentwocontactsoftheDS,butapparentlydependsonotherinfluences as well. For example, thermal effects in the air gap can reduce the gap breakdownvoltage,whichwillbediscussedindetaillateroninthischapter. 5.1.2Opticalarcimageobservation The arc extinctions and re‐strikes/re‐ignitions are also visible through arc imaging. The arc length increases due to the increasing distance of the contacts, and the arc moves upwards gradually mainly due to the rising hot air. The arc usually has a single re‐strike resulting in the highest brightness during a short period of time within each half cycle. Both electrical signals and arc images are exemplifiedinFigure5.9foranentireinterruption.Theinstancesintheelectrical signals,wheretheimageframesaretaken,areindicatedbytimeinsidetheoptical images. The arc images show the gradual upward movement as a 2 dimensions projectionofthecomplex3dimensionalshape.Thearclengthisincreasinguntil its final extinction. In the shown example, the arc extinguishes prematurely 100 id(A), ud(kV) u 50 Zoomed in i d d 0 −50 0 50ms 100 200 1ms 100ms 300 400 500 time (ms) 2ms 150ms 600 700 800 3ms 200ms 6ms 8ms 250ms 300ms 350ms 10ms 450ms 650ms 500ms 700ms 550ms 750ms 400ms 11ms 600ms 800ms Figure 5.9. Wave shapes of udand id, and arc image frames at different time instances for 140kV,2AmeasuredattheTU/eHVlaboratory. CurrentInterruptionMechanism67 becauseofthedecayingresonantsourcevoltage(seeChapter4,Figure4.4).The expansion over a full cycle taken between 674ms and 694ms of Figure 5.9, together with arc images taken every 1.1 ms, is shown in Figure 5.10. Time t1 is themomentthatthere‐strikeoccurs;t2isthemomentofthere‐strikeonepower cycle later. As already observed from the electrical signals, also from the brightnessofthearcimages,itisconfirmedthatre‐strikeoccurseachhalfcycle. Thearcbrightnessdropsdrasticallyfromtheframeatre‐striketothesubsequent framebecausethehigh‐frequencycomponentlastsonlyafractionofthecamera exposure time of 0.99ms for 1000frames/s. The arc images show that a re‐ establishedarcfollowscloselythepathoftheprecedingarcbeforere‐strike. ud 20 id d d i (A), u (kV) 40 0 −20 −40 675 680 685 time (ms) 1st f 5th f 9th f 2nd f 3rd f 6th f 10th f 13 th f 14 th f 17th f 18 th f 690 695 4th f 7th f 8th f 11th f 12 th f 15 th f 16 th f 19 th f Figure 5.10. Arc images recorded within 20ms of Figure 5.9 including three re‐strikes (1st, 10thand19thimage). 68Chapter5 5.1.3Combinedopticalandelectricalsignals The changes in the arc over time are characterized with two types of measurements. The electrical measurements provide information about changes in the arc current and voltage. On the other hand the optical measurements provide a visual perception of the arc properties through the recorded images. Thepresenceofthearcinanimageresultsintheincreasedimagebrightness.In ordertoinvestigatethecorrelationbetweenopticalandelectricalarcinformation qualitatively,amethodissoughttoquantifythebrightnessofasinglearcimage frame,whichareextractedfromhigh‐speedarcvideorecording. Brightnessisanattributeofvisualperceptioninwhichasourceappearstoemita given amount of light. In order to quantify brightness in this application, it is definedhereasthenormalizedintegratedgreyscalevalueofallpixelsinacertain imageofthehigh‐speedvideocameraoutputfile.Thegreyscalevalueofapixelin thedigitalimagehasavalueintherangeof0‐255.Thevalue0correspondstothe black, while the value 255 represent white. Such definition has no verified/calibrated relationship to any physical quantity of the light emitting properties of the arc. Therefore, brightness measurement results thus have only indicativevalue. The brightness of a pixel with coordinates (i, j) in frame k of a digital image recording is denoted as Bi(,kj) , where 1≤i≤n and 1≤j≤m (n×m is the spatial resolutionofthecamera).Theimagebrightnessofthekthframeisdefinedasthe sumofbrightnessfromallpixels: n m Bk Bi(,kj ) (5.1) i 1 j 1 The values of Bk of different images can only be compared directly if they are takenwithidenticalcamerasettings.Inordertocompareandevaluateeachimage in a recording, the brightness is normalized to the brightness Nk of a full white image: N k 1 N n m i1 j 1 w (k ) B i , j , (5.2) whereNkisthebrightnessafternormalizationonNw=n×m×255(255beingthe maximum 8 bit pixel intensity value). The arc brightness is estimated as an increaseintheimagebrightnessduetoappearanceofarc.Thus,itiscalculatedas a difference between the brightness of the image with the arc and the image withoutarc(backgroundimage): – .ThebackgroundbrightnessN0is takenfromarecordedframejustbeforeopeningtheblades. CurrentInterruptionMechanism69 Practical imaging systems have a limited dynamic range, i.e., the range ofluminositythatcanbereproducedaccuratelyislimited.Partsofthesubject(in thiscasearc)thataretoobrightarerepresentedaswhite,withnodetailswhichis often referred to as saturation;parts that are too dark are represented as black. The levels of the luminosity in the recorded arc were too high for the dynamic rangeofthecamera.Therefore,certainlevelofthesaturationispresentinalmost allimages.Thenumberofsaturatedpixelsishigherintheimagescorresponding tothemomentoftherestrikes,especiallyfortheimageswithdarkenvironment. Inthegreyscale,pixelswhichexhibitsaturationeffectarerepresentedwithapixel intensity of 255. The presence of the saturation effect in the images prevents quantitative analysis. However, the recorded images can provide an indication aboutarcbehaviourandallowthatthemostimportanttrendsareobserved. An example of arc brightness as a function of time is shown in Figures 5.11 and 5.12 for Id=1.3A with a resolution of 256×254 pixels, recorded with 5000frames/s.Thearcimageissynchronizedwiththemeasuredwaveformsfor voltageudandcurrentid(referSection4.2.5).InFigure5.11thearcbrightnessand its expansions of three time periods (120ms‐160ms, 440ms‐490ms, 650ms‐ 780ms)togetherwiththesynchronizedudandidtracesareshown.Figure5.12is a further expansion taken between 670ms and 716ms to show the precise synchronization. Thefollowingobservationsaremade: ‐ Higharcbrightnessoccurseachhalfpower‐frequencycycle,directlyatthe momentofre‐strike.Figure5.12showsthatthebrightarconlycovers2‐3 images corresponding to a few tenths of a millisecond. The fast decay is relatedtothedurationofthehigh‐frequencyoscillations. ‐ The arc brightness varies synchronously with the arc current. It implies that in a steady burning arc (between re‐strikes) the arc current is the dominatingfactor,whichdeterminesthearcbrightness. 5.2Arcinterruptioncharacteristics The arc characteristics of the interruption include arc duration, appearance, brightness, movement and remnant decay. The electrical parameters influencing on the arc characteristics will be analyzed. Besides the electrical circuit parameters,environmentalconditionsuchaswind,humidityplaysaroleaswell. Buttheireffectisnotconsideredinthisthesis. 5.2.1Arcduration Arc duration is a key factor in evaluating the interruption performance. The arc duration is defined as the duration from the moment when the arc ignites between the two contacts of the blades until the moment when it extinguishes 70Chapter5 0.2 0.06 0.18 x 10 −4 0.04 Brightness 0.16 8 6 0.14 4 2 0.12 0.02 130 140 150 0 −3 0.1 x 10 0.08 10 0.06 5 0.04 0 440 0.02 680 450 460 470 480 700 720 740 760 490 0 0 100 200 300 400 500 600 700 800 30 ud (kV) id (A) 10 d d u (kV), i (A) 20 0 −10 −20 −30 0 100 200 300 400 time (ms) 500 600 700 800 Figure 5.11. Arc brightness with the synchronized voltage ud and current id waveforms for Id=1.3A. completely.Theglowemittedbyhotairremnantsafterthefinalarccurrentzero crossingisnottakenintoaccount.Figures5.13and5.14chosenfromthemedium shadedregioninTable5.1showthearcdurationasafunctionoftheinterrupting current Id and Cs/Cl, respectively. The arc duration depends on Id (with equal Cs and us) as indicated by the regression lines in Figure 5.13. It decreases with increasing ratio Cs/Cl at fixed us and Id, see the regression lines in Figure 5.14. Furthermore, it increases with the system voltage when the current Id and the ratio Cs/Cl are taken constant. The results confirm that the system voltage, interrupting current and ratio of the capacitances at source and load side are indeedthemainfactorsaffectingthearcduration. CurrentInterruptionMechanism71 Brightness 0.02 0.015 0.01 0.005 Current (A) 0 670 2 675 680 685 690 695 700 705 710 715 675 680 685 690 695 time (ms) 700 705 710 715 1 0 −1 −2 670 Figure5.12.Expansionofbothelectrical(lower)andoptical(upper)signalsbetween670ms and716msfromFigure5.11. 1.5nF 20nF 100nF Linear(60nF) Arc duration (ms) 2100 1900 6nF 60nF Linear(1.5nF) 1700 1500 1300 1100 900 0.5 1.0 1.5 2.0 Current (A) Figure5.13.ArcdurationversusinterruptingcurrentIdwithCsasaparameterat173kV. 1.0A at 173kV 2.1A~2.3A at 173kV 1.0A at 90kV Linear (1.0A at 173kV) Linear (2.1A~2.3A at 173kV) 2100 Arc duration (ms) 1900 1700 1500 1300 1100 900 0.0 0.5 1.0 1.5 Cs/Cl 2.0 2.5 3.0 3.5 Figure5.14.ArcdurationversustheratioCs/Clat90kVand173kVwithIdasaparameter. 72Chapter5 5.2.2Arcappearance Fortheexperimentsupto2.3Athearcexhibitstwodistinctvisualappearances, or "modes", as shown in Figure 5.15. These modes are denoted as "erratic" and "stiff"modein[1].Fortheerraticmodethearcdevelopsinacomplexshapeand itslengthismuchlongerthanthebladetipspacing(Figure5.15a).Thestiffmode isacontractionfromtheerraticmodetoamoreorlessstraightpathbetweenthe contacts (Figure 5.15b). It is observed that the arc shape remains erratic during thecompleteinterruptionforCs/Cl<1.5intheexperiments.Theerraticarcmode is associated with longer arcing time. The arc changes from erratic to stiff mode afteracertaintime,andremainsstifftilltheendforCs/Cl>1.5.Thisisassociated with shorter arcing time. The experimental results show no convincing relationshipbetweenthearcmodeandIdorus.However,thearcwithstiffmodeis clearlyrelatedtotheratioCs/Cl.WithsmallerCs/Cl,themedium‐frequencyvoltage transientsarehigher(referSection3.4).Theincreasedheatdissipationfacilitates the next breakdown along the same path despite its complex curled shape. The influenceofCs/Clwillalsobediscussedinthesection5.6. 5.2.3Arcbrightnessversusarccurrent In a number of experiments arc interruption failed resulting in continuation of arcing after the DS blades have reached their final open positions. These failed interruptions allow to investigate the relation between entire brightness and current,sincethesituationwithfullyopenedbladesensuresthatthearccurrentis theonlyvariablefactor,andnotthegapspacing.Thebrightnessatcurrentlevels between12.5Aand27Aof90kVisdisplayedinFigure5.16.Itisobservedthatthe entirearcbrightnessincreaseswithincreasingId.Itisalsoobservedfromthearc video that the arc burning with fixed air distance with higher current exhibits muchmorecurlsthanitdoeswithlowercurrent. Figure5.15.Arcin(a)erraticmodeand(b)stiffmode. Brightness CurrentInterruptionMechanism73 0.12 12.5A 27A Last restrike 0.1 0.08 0.06 50 100 150 200 250 time (ms) Figure5.16.Thearcbrightnessagainsttimewiththecurrentasaparameter. 5.2.3Arcmovement In absence of wind influences, the arc development upon the increasing gap distance is governed by the convection of the heated air in between. The horizontal extension depends on opening velocity of the air gap, which is in the range0.2to0.6m/s.Theupwardspeedcanbebestanalyzedfromcameraimages based on failed interruptions, because of the long arc lengths. Moreover, the arc burnscontinuously,whichismostsuitableforstudyingitsmovement. Figure5.17 shows the arc motion from a failed interruption with Id =10A, Em=90√2kV.TheheightHinmmreferstothedistanceabovetheoriginalclosed positionofthemainblades.Afterabout70halfcycles,thetestisstoppedbythe master breaker in the laboratory. For each half cycle, a single frame is selected. The sub‐figures are taken at corresponding phase angles for 10 subsequent half cycles;thesevensub‐figurescoverthecompletearcduration.Thetimedifference betweentwocurvesthereforeis10ms(halfpowercycle)andbetweentwosub‐ figures it is 100ms (ten half power cycles). It is observed, that the arc moves upwards exhibiting a seemingly random path. Arcs in successive frames are slightlyverticallydisplacedrelativetothepreviousone. This behaviour suggests that the heat produced by the arc creates favourable conditions a little upward from the present position. The actual arc trajectory thereforeismainlydeterminedbytheconditionoftheairinthegapratherthanof theactuallocationofthearcroots. Thelastsub‐figureinFigure5.17showsthehighestpoint(arcreach)ineacharc frame versus time for interrupted currents in the range of 12.5A to 27A. The height increases nearly linearly with time, approximately 2.3m/s. This speed is rather uncorrelated with current level in the tests. Too wide extent of the arc ("reach") as a result of failed interruptions may cause dielectrically hazardous situationsinsubstations. 74Chapter5 700 600 12.5A 20A 23A 27A 500 H(mm) 400 300 200 100 0 −100 0 100 200 time (ms) 300 Figure5.17.Upwardarcmovement(framestakeneach10ms).Thelastsubfigureshowsthe arcreachincreasingwiththearcingtimeusingtheinterruptedcurrentasaparameter. CurrentInterruptionMechanism75 5.2.4Arcremnantdecay The light emission does not vanish immediately after final arc extinction. The decay in brightness of these arc remnants is investigated as a function of the interrupted current. The normalized brightness at four distinct values of Id is plotted in Figure 5.18 (left) for successful interruptions. Herein, t = 0 is the moment of arc extinction. The curves can be fitted to match an exponential behaviour,Aexp(‐t/τ),withτthedecaytimeconstant.Thistimeconstant,plotted versus Id in Figure 5.18 (right), scales approximately linearly with Id: τ = 0.9*Id. Theobtainedresultimpliesthatwithhigherinterruptedcurrent,thearcafterglow decays slower. Apparently, the hot ionized discharge channel cools down more slowlyafterhigherarccurrent. 5.3Disconnectorcontactsspacing RV is the voltage impressed across the DS by the circuit after arc extinction at currentzero.Itpossiblyprecedesre‐strikethatre‐establishesthearc.ThisRVis maximumifitoccursathalfapower‐frequencycycleafterinterruption.Thenthe source voltage (at 1p.u.) is equal but in opposite polarity with the trapped DC voltageacrosstheloadsidecapacitance(alsoat1p.u.).Thus,RVbuildsupto2p.u. ThisisalsothemaximumRVthatthe(singlephase)circuitiscapabletosupply. The tips between the blades of a DS can be considered as rod‐rod electrodes. In the"IEEEStd4‐1995,TechniquesforHighvoltageTesting"[2],rod‐rodgapspark‐ overvoltages at power‐frequency versus the air spacing are listed. From this standard, the required contact gap spacing based on the centre‐break DS blade length [3], and its corresponding blade angle can be derived [1]. To allow for a 10%margin[4],thesystemvoltageistaken1.1p.u.andthusaRVpeakvalueof 2.2p.u.isapplicable.TheconditiontobesatisfiedfortheRVnolongerbeingable to breakdown the air gap is: Ugap ≥ 2.2×Em, where Ugap is the air gap withstand voltageandEmistheamplitudeofphase‐to‐groundvoltage. 0.12 25 9A 12A 20A 27A 0.06 20 τ (ms) Brightness 0.09 15 10 0.03 5 0 5 10 15 20 25 40 60 Current (A) time (ms) Figure 5.18. Brightness decay and its time constant τ dependency with the interrupted current. 0 20 76Chapter5 Table5.2showstheminimumrequiredgaplengthandbladeanglewithrespectto the close position at different system voltage levels for a centre‐break DS. The calculated results show the blade angle is less than 60 degree, considering +8% margin,takenfrom[2]. Thebladeanglesobtainedfromexperimentscarriedoutona245kVratedvoltage centre‐break DS differ greatly from the values in Table 5.2. In Figure 5.19, the bladeanglesatthemomentoffinalarcinterruptionisgiven.Allanglesatfinalarc extinction exceed 50 degrees; some of them reach up to the final 90 degree openingangle.Evidently,theresultsfromtheexperimentsexceedsignificantlythe calculatedidealvalues,basedon(cold)air‐breakdownvoltageonly. For estimation of the minimum required gap distance in Table 5.2, two assumptionsweremade[1]: The thermal energy of the arc is not significant. This means that the ‐ breakdown voltage of the gap essentially is equal to a steady state, "cold" situationwithoutarcingshortlybeforebreakdown. ‐ The opening contact gap is approximated as a rod‐rod type gap. In the presentexperiments,nocoronaringsorsimilarconstructionswereinstalled on the jaw assembly. This can cause field enhancements at the end of the contacts. 5.4Re‐strikevoltage There‐strikevoltage,denotedasUr,isdefinedasthevalueofudacrosstheDSat themomentofre‐strike.There‐strikevoltageismainlydeterminedbytheairgap distance,butitisalsoinfluencedbythephysicalairgapconditionedbythearc. 5.4.1Re‐strikevoltagedevelopment The development of Ur is analyzed for measurements with different currents Id, andcapacitancesratiosCs/Cl.There‐strikevoltagesobtainedatasupplyvoltage of173kVaregiveninFigure5.20,withCs/ClandIdasparameters.Thefollowing conclusionsaredrawn: ‐ The maximum re‐strike voltage level is indeed found to be twice the peak supplyvoltage.The re‐strikevoltagedoesnot show amonotonousincrease withincreasingairgaplength.Instead,suddendownwardstepsareobserved ateachcurrentlevel.However,withhighercurrentlevelthe"steps”aremore numerous and deeper. Apparently, besides the air gap length, also other factorsshouldbeconsidered. ‐ There‐strikevoltageincreaseswithdecreasingIdandincreasingCs/Cl.Larger Id and smaller Cs/Cl are associated with a higher energy input into the arc upon re‐strikes (Section 5.6). The arc path needs more time to recover its dielectric stress during the next RV period resulting in a lower re‐strike voltagefortheconsecutivebreakdown. CurrentInterruptionMechanism77 Table 5.2. Minimum contact gap distance and blade angle needed for isolating distance withacentre‐breakDS. Contactgap Bladeangle System Bladelength RVpeak (mm) (0) voltage(kV) (mm) (kV) mean +8% mean +8% 72.5 900 130 220 238 41 43 145 1650 260 481 519 45 47 245 2600 440 835 902 47 49 362 3500 650 1253 1353 50 52 420 4000 754 1465 1582 51 53 550 4400 988 1936 2091 56 58 Angle (degree) 1.5nF 85 6nF 20nF 60nF 100nF 75 65 55 45 0.0 0.5 1.0 1.5 2.0 Current (A) Figure5.19.BladeangleversusIdwithCsasparameterat173kV. 2.5 400 U (kV) 0 1.0 0.08 0 r r 2.1A −200 −400 3.1 0.57A 200 U (kV) 500 0.23A 0 500 1000 1500 time(ms) 2000 −500 0 500 1000 time (ms) 1500 Figure5.20.EffectofcurrentId(left,withCs=1.5nF)andcapacitorratioCs/Cl(right,with Id=2.1A)onre‐strikevoltage. ‐ Positive and negative re‐strikes do not occur symmetrically. For example, Figure5.20left,mostapparentlyat2.1Athepositiveandnegativere‐strike voltagesdiffer.Are‐strikeusuallyoccurseachhalfcycle.However,becauseof asymmetricalovervoltageacrossCsandCl,onlyonere‐strikewithineachfull power‐frequency cycle may occur. It will be explained in the following subsection. pf d i (A), u (kV) 78Chapter5 0 −50 ipf=0 −100 i pf −150 ud −200 Re−strike moment 30 40 50 60 70 time (ms) 80 90 100 Figure 5.21. Wave shapes of ud and ipf (id measured at lower frequency) with only one arc extinctionperpowerfrequencycycle. 5.4.2Unipolarre‐strikevoltage Generally,oncethecontactsoftheDSseparate,re‐strikesoccurineachhalfpower cycle.Figure5.21showsanexampleofthewaveshapesofthevoltageudandthe current id versus time of an interruption process, for which the last three re‐ strikes are shown. Here the arc extinguishes and re‐strikes only each full power cycle. According to the high‐speed arc image recordings with 5000frames per second,noarcextinctionsareobserved.Apparently,thearcdoesnotextinguishas longasthehigh‐andmedium‐frequencyoscillationisofsufficientmagnitude.The phenomenon is always observed, if the corresponding re‐strike voltage exceeds the source peak voltage. Once the re‐strike voltage is high enough, the medium‐ frequency components last sufficiently long to prevent interruption at power‐ frequency current zero. The oscillation in the current and voltage still continues whenthepower‐frequencycurrentcrosseszero.Inthissituation,thearccanfail toextinguish.Thisisduetotheinfluenceofthethermalenergyprovidedbythe oscillations. If this phenomenon occurs repeatedly, re‐strike will happen only at onepolarityofthepower‐frequency,eitherpositiveornegative. Themissedarcextinctioncanalsobeobservedthrougharcimageanalysis.Figure 5.22 shows the arc images corresponding to consecutive frames, taken every millisecond,withinahalfcycle.Thexaxisindicatesthehorizontalextensionofthe arc.Theyaxisisthearcheightwithrespecttothemainblades.Inordertodisplay the arcs distinctly, a vertical offset is added to the consecutive images. The 1st imagecorrespondstoare‐strikemomentwhilethe10thframecorrespondstothe endofthehalfcycle,i.e.themomentwhenthearccurrentreacheszeroagain.The images after the 10th frame correspond to the next re‐strike. In Figure 5.22a the 10th image shows a discontinuous arc path, whereas in Figure 5.22b this discontinuity, measured at a later stage of the same arc development, is not observed. Thepolarityeffectnormallyappearsinthelaterstageoftheinterruptionprocess. At the beginning of the interruption, the arc extinguishes at each current zero CurrentInterruptionMechanism79 Figure5.22.TypicalarcimagesfromexperimentswithId =0.7AandEm =130√2kVatTU/e HV laboratory, (a) with obvious re‐strike frame and (b) arc does not extinguish at power frequencyzerocrossing.D=0isthemiddlepointofthetwoDSbladesinaclosedposition. because air gap conditions, i.e. temperature and ionization, are not sufficient to sustain arcing. However at a later stage, the gap conditions can sustain arcing during current zero without an appreciable re‐strike voltage and visible interruption. 5.4.3Re‐strikevoltagewithairgapdistance After re‐strike the arc resumes a path closely following the "traces" left by the previousarcthathasjustdisappeared.Thearcchoosesthedielectricallyweakest pathforwhichthebreakdownvoltageislowest.Theremaininghotpartlyionized airenablesthearctore‐igniteatalowerbreakdownvoltagecomparedtoa‘cold’ situation.Themeasuredbreakdownvoltagethatre‐establishesthearcisplotted versustheairgapspacingatId=0.3Aand0.7AinFigure5.23.Thelengthofthe arcpathisestimatedthroughthearcimageframes.Thereferenceunconditioned breakdown voltage, also plotted in Figure 5.23, is taken from the IEEE standard [2]forarod‐rodconfiguration.Clearly,thebladeshapeismuchmorecomplicated, but the comparison provides a notion to what extent the actual breakdown voltage is lowered. Despite the strong variation in breakdown voltages and the relativelyrougharcpathestimate(arclengthisbasedona2Dprojectioncaptured by the camera), the reduction in breakdown voltage by the remaining residual ionization can be observed clearly. Temporary arc extinction at current zero occurs in a situation with relatively low pre‐heating and ionization of the arc channel.Continuousarcingoccurswhensufficientheatandchargedparticlesare leftintheairgap.Theresidualgapconditionhasamajoreffect,sinceitlowersthe re‐strikevoltageneededtore‐establishthearcandprolongsthearcduration. 80Chapter5 Reference 0.4A 120kV 0.7A 130kV 160 Voltage (kV) 120 80 40 0 0 100 200 Air gap length (mm) 300 Figure5.23.Breakdownvoltageofthegapversusairgaplength. 400 5.5Electricalarccharacteristics The free burning arc during air‐break DS interruption has a small current compared to the fault arc in circuit breakers. For understanding this type of arc behaviour, the arc current, voltage, resistance, and U‐I characteristics are analyzed.Arccurrent(denotedasia)andarcvoltage(denotedasua)areequalto theidandudwaveforms,respectively,duringthearcingperiod.Therefore,thearc voltage and current waveforms are not continuous curves, but their absence during the voltage recovery phase can be short or hardly visible when the breakdown voltage is still relatively low. This occurs at the early stage of the interruptionforlowcurrents,orduringtheentirearcdurationwithhighcurrent whentheinterruptionfails. 5.5.1Arccurrent Figure 5.24 shows examples of arc current and voltage waveforms for a clearly visibleandrelativelylong,anon‐visibleperiodofcurrentzero.Itisobservedthat when the re‐strike voltage is large enough, the arc current has clear transient components at the beginning of each re‐strike. When the arc is continuously burningnotransientbehaviourisobservedinthewaveforms. 5.5.2Arcvoltage Similartothearccurrent,thearcvoltageexhibitsanobvioustransientcomponent atthebeginningofeachre‐strike(Figure5.24left).Itis,however,verydifficultto distinguish between the arc voltage and the induced voltage due to the severe current transients with very high di/dt. When the breakdown voltage is low, eitherbecauseofastillsmallDScontactdistanceorbecauseofapreconditioned gap,therearehardlytransientvoltagesatthebeginningofeachre‐strike(Figure 5.24 right). The arc voltage has a distorted shape, due to the non‐linear arc u 20 10 0 −10 −20 ia ua a ua (kV), ia (A) ua (kV), ia (A) CurrentInterruptionMechanism81 characteristic.Ifthemedianvalueofthevoltageshapeistakenasameasureofthe arcvoltage,itreachesuptoapproximately10kV. Fortheentireinterruption,there‐strikevoltagedoesnotincreasemonotonously as shown in Figure 5.25 (left). This is because the periodically shortening of the arcbyshort‐circuitsofloopsandcurls,suddenlyloweringthearcvoltage.Incase the arc is not interrupted at full open blade position, the air gap distance is constant,andthearckeepsburningwithoutobviousre‐ignitionsorre‐strikes.The arcvoltagecontinuestoincreaseduringthearcingtime,whichcorrespondswith the increasing arc length for an undisturbed arc, referring Figure 5.25 right. As shown in Figure 5.26, the arc voltage increases nearly linearly, with about 3.2kV/matId=27A. 5.5.3U‐Icharacteristic Thearcvoltageandcurrentareinphase(Figure5.24),implyingthatthearchasa resistivenature.Sincethearcvoltageisintheorderofafewkilovolts,andthearc currentisupto27Ainthisexampletest,thearcresistanceisuptoafewhundred ohms. The arc current remains constant, the arc resistance increases with increasingarcvoltage.Duetothenon‐linearnatureofthearcvoltage‐current(U‐ I)characteristic,thearcresistancevariesstronglyduringapowercycleasshown inFigure5.27(left). i a 20 10 0 −10 −20 520 540 560 1610 1620 1630 1640 1650 time (ms) time (ms) Figure 5.24. Arc current and voltage waveforms, left: Id=3.9A with clear current zero periods;right:Id=27Awithacontinuousarccurrent. 40 10 20 5 0 −20 −40 ud (kV) ud (kV) 0 −5 −10 200 300 400 500 2000 2100 2200 2300 time (ms) time (ms) Figure5.25.Arcvoltagewithandwithoutstepsintransientmagnitudesat27A. 82Chapter5 Arc voltage (kV) 20 16 12 8 4 0 0.8 1.8 2.8 3.8 Arc length La (m) 4.8 5.8 Figure5.26.ArcvoltageversusarclengthatIdof27A. 40 1 20 u (kV) 0.5 0 a Arc resistance (k Ω) 0 −20 −0.5 450 500 550 time (ms) −40 −20 0 i (A) a Figure5.27.ExampleofthearcresistanceandarcU‐Icharacteristicsat27A. 20 AtypicalU‐IcharacteristicisplottedinFigure5.27(right),whichisderivedfrom a long duration arc at Id = 27A. It is also observed that the arc voltage increases rapidly immediately following current zero and reaches a somewhat constant valuewithinthehalfcycle.ThehysteresisloopconfirmsthetypicalarcU‐Icurve. Thisloopalsoexplainsthatthearcvoltagehasvariablevaluesatidenticalcurrent, butisdeterminedbythearclength. 5.6Energyinputintothearc The energy input w into the arc is obtained by integrating the product of the voltageuaandthecurrentiaoverthearcingtime: w t2 t ( u a ia ) dt (5.3) 1 Here t1 is the moment of re‐strike and t2 is the moment of extinction at power‐ frequencycurrentzero.Figure5.28illustratesthewaveshapesofthevoltageua, thecurrentia,andthecorrespondingenergywoverthetimeofasinglere‐strike. CurrentInterruptionMechanism83 Figure5.28.Waveshapesofud,idandthecorrespondingenergyw;t1isthere‐strikemoment andt2isthemomentoftemporarilyarcextinction. Uponare‐strike,threefrequencycomponentsaredistinguished(seeChapter3):a high‐frequency component due to charge equalization of Cs and Cl, a medium‐ frequency component from the oscillation between circuit inductance, both side capacitances in parallel, and the power‐frequency (PF) component. The solid rectangleinFigure5.28indicatesthedurationwheretheHFoscillationprevails, thedottedrectangleisthedurationofMFoscillation,andthedashedrectangleis theareaoftheremainingPF.Itcanbeobservedthattheenergyincreasesrapidly duringtheinitialHFoscillationandcontinuestoincrease,butatalowerrateuntil the MF oscillation is damped. It reaches an almost constant value at power‐ frequency.Forinstance,intheexampleofFigure5.28itreaches1.10kJduringthe first0.3ms,increasesgraduallyto2.18kJduringthefollowing8ms,untilitfinally reaches2.0kJ.TheenergycontributionsintheshownexamplefortheHFandMF oscillationsarecomparable,butthePFphasehardlycontributes.Thusitmaybe concludedthat(here)re‐striketransientscontributevery significantlytoenergy inputthearc. 5.6.1Energydistributionintransients The energy after each re‐strike in terms of percentages contributed during the phaseswithdifferentdominantfrequenciesisplottedinFigure5.29.Thecurrent isvariedfrom9Ato27A. 84Chapter5 HF MF PF Linear(HF) Linear(MF) Linear(PF) Energy percentage 80 60 40 20 0 8 12 16 20 24 Current (A) Figure 5.29. Energy contribution upon re‐strike by various current frequency components (HF,MF,PF)asafunctionofcurrentId. The dependence on current level is roughly indicated by fitted lines. The results showthatthecontributionoftheHFoscillationisrelativelyhighatlowvaluesfor Id and drops at higher Id. For instance, the HF energy contribution lies between 30% and 80% at Id < 12.5A but is less than 55% for 15A< Id <27A. The MF contribution increases with current level with an energy fraction between 20% and 50% at Id < 12.5A and over 40% for 15A< Id <27A. The energy at PF is usuallysmallerthan30%atalltestedcurrentlevels.Onaveragethecontribution ofPFisabout10%. For the experiments with a fixed value of Cs (Id in the range of 9A to 27A), the interruptedcurrentdependsonthevalueofCl.SinceClismuchlargerthanCsthe MFcontributiondependsonClwhereastheHFcontributiondependsonCs,which istakenconstant.Therefore,theenergyinputfromtheMFcomponentincreases withhigherCl.Asaresult,therelativeimportanceoftheMFcomponentincreases with Id. The energy contribution from the power‐frequency current is always relatively low. There is a large spread in the relative contribution of the HF and MFoscillations,buttheybothdominatetheenergyinput. 5.6.2InfluenceofCs/ClandIdonenergy TherelationbetweenfactorsCs/Cl,IdandthearcenergyisshowninFigure5.30 for a current Id = 2.1A. In the upper three figures (indicated by ‘a’), the developmentofarcenergyagainsttimewithCs/Clasaparameterisgivenineach arcing period. The lower three figures (indicated by ‘b’), show the arc energy againsttimewithIdasaparameter.Itisobservedthat: ‐ Theenergyinputintothearcuponre‐strikeistypicallyafewhundredJoule going up to a few thousands Joule at lower ratio of Cs/Cl and longer arc duration. CurrentInterruptionMechanism85 ‐ Itincreasesgradually(withoccasional"steps")becauseoftheincreasingre‐ strikevoltagecausedbythemotionoftheDScontacts.Itreachesthelargest value,justbeforecompletearcextinction. ‐ ItishigherwithhigherinterruptioncurrentandlowerCs. 5.6.3Energyversusre‐strikevoltageUr Thecorrelationbetweentheenergywandre‐strikevoltageUratdifferentcurrent Id is presented in Figure 5.31. Four different re‐strike voltage (Ur) ranges are selectedwithIdbetween9Aand27A.SinceCs(50nF)andEm(90√2kV)arefixed, IddependsonClonly.Figure5.31showsthatatthesamecurrentlevel,theenergy supplied into the arc increases with increasing re‐strike voltage. Therefore, with higherinterruptedcurrentandhigherre‐strikevoltage,moreenergyisdelivered to the arc upon re‐strikes. Because of the higher energy input into the arc, the breakdownvoltageofthegapislowerandthegaprecoversslower,thearcisnot more difficult to interrupt but more easily to re‐strike, even at longer gaps, indicating that the current to be interrupted and the breakdown voltage on re‐ strikearecriticalfactorsforinterruptionperformance. 5.7Transientsuponre‐strike DuetotheslowmovementoftheDSmainbladesanarcofalongdurationarises, characterized by transient phenomena upon re‐strike. The resulting transient voltage and current may endanger sensitive secondary components of the networkthroughelectromagneticcoupling. From the frequency components initiated by re‐strike, the maximum transient Ur voltage across both capacitances, um ax Em , occurs at medium‐ 1 Cs Cl frequency(seeChapter3);EmisthepeakvalueofthepowersourceandUristhe re‐strikevoltage.Takingthere‐strikevoltageasaworstcaseequalto2*Em,with Cs/Cl << 1, the maximum transient voltage may reach 3p.u. The maximum Ur transientcurrentuponeachre‐strikethroughtheDSis imax LH ( 1 Cs 1 Cl withLH ) theequivalentinductanceofHFloop.Themaximumvalueofthetransientcurrent dependsonthere‐strikevoltageandthecircuitparametersofthehigh‐frequency loop and can reach values up to a few kiloamperes within a few microseconds (7kAinthetest). Energy (J) 86Chapter5 2000 2.1A C /C =1.5/40 s l 1000 0 0 200 400 600 800 Energy (J) 1200 1400 1600 1800 2000 2200 1400 1600 1800 2000 2200 1400 1600 1800 2000 2200 1.0A C /C =1.5/19.3 500 0 1000 time (ms) 1000 s 0 l 200 400 600 800 1000 1200 time (ms) Energy(J) 400 0.57A C /C =1.5/10.7 s l 200 0 0 200 400 600 800 1000 1200 time(ms) (a) Energy (J) 2000 1000 0 Cs/Cl=1.5/40 0 200 400 600 800 Energy (J) 1200 1400 1600 1800 2000 2200 1400 1600 1800 2000 2200 1400 1600 1800 2000 2200 time (ms) 400 Cs/Cl=20/40 200 0 1000 0 200 400 600 800 1000 1200 Energy (J) time (ms) 200 Cs/Cl=100/40 100 0 0 200 400 600 800 1000 1200 time (ms) (b) Figure5.30.Energyinputtothearconre‐strikeversustimewithparameter:(a)capacitor ratioCs/ClatId=2.1A,and(b)currentIdwithfixedCs=1.5nF. Energy (kJ) CurrentInterruptionMechanism87 6 5 4 3 2 1 0 200~250kV 150~180kV 50~80kV 10~20kV 8 13 18 Current (A) 23 28 Figure5.31.Arcenergyuponre‐strikeatdifferentinterruptioncurrentlevel.Parameterisre‐ strikevoltageinterval. 5.7.1Transientvoltageuponre‐strike Figure5.1showstheovervoltagesacrosstheloadsidecapacitancethatappearin the end phase of the interruption process. Figure 5.32 shows the maximum observedovervoltageuponre‐strikeacrossClasafunctionofthere‐strikevoltage atdifferentvaluesoftheratioCs/Cl,withasourcevoltageof173kVandacurrent of 2.1A. The overvoltage across Cl can reach up to 2.4p.u. It increases with increasingre‐strikevoltageanddecreasingratioCs/Cl.Itcanalsobeobservedthat theovervoltagereachesthehighestvalueatthelowestCs/Clratio:Cs/Cl=0.04. The dominant frequency of the overvoltages is in the order of a few tens of kilohertz. Since most of the overvoltages occur at the end of the interruption, where the re‐strike voltage is high, it would be advisable to put effort in interruption of the circuit at an earlier stage. At higher ratio of Cs/Cl, the overvoltagesreducesignificantly. Apart from the high peak values, overvoltages are characterized by steep fronts. Althoughtheserepeatedsteepfrontshavealowerfall‐timethanthatobservedin GIS‐DSswitching,theresultingtravellingwavesinopenairstationsstillcanlead toconsiderablestressestoequipment(notablytransformers)directlyconnected totheDS[5].Thetransientmayreach2p.u.withinafewtensofmicrosecondsin theperformedexperiments. 5.7.2Transientcurrentuponre‐strike The current through the disconnector in the experiments at 9A < Id < 27A are recorded with a bandwidth of 20MHz. For the experiments, the value of Cs is 50nF, and the source RMS voltage is taken as 90kV. Figure 5.33 shows that the transient peak current value increases with the re‐strike voltage. The dominant frequency of current transient is a few tens of kilohertz to a few megahertz. according to the measured data. It was also found that the rate of rise of the transientcurrentisafewkiloamperespermicrosecond. 88Chapter5 Figure 5.32. Overvoltage across Cl against re‐strike voltage with the ratio of Cs/Cl as a parameteratcurrentof2.1A. Current (kA) 8 6 4 2 0 0 0.5 1 1.5 2 Re-strike voltage (p.u.) Figure5.33.MeasuredHFtransientcurrentpeakvalueagainstre‐strikevoltage. 5.8Conclusion Both recorded electrical and optical signals show that in the few seconds of capacitivecurrentinterruptionthearchasfrequentextinctionsandre‐ignitions.It extinguishes temporarily at the power‐frequency current zero, but re‐ignites by the circuit impressed recovery voltage and ultimately extinguishes when the electricalstrengthoftheairgapislargerthantherecoveryvoltagethecircuitcan provide. During this process, the breakdown voltage of the air gap along the arc pathissignificantlyreducedduetothethermalinfluenceofthearc.Asaresult,it may take seconds for the gap to recover and withstand the highest recovery voltage. Arcbehaviourandtheinterruptionperformancearehighlyaffectedbythecircuit parameters. These parameters include capacitance ratio Cs/Cl, stray and short‐ circuit inductance and power supply voltage. All these affect the current to be CurrentInterruptionMechanism89 interrupted, the re‐strike voltage, associated transients as well as the energy suppliedtothearcandultimatelytheheatingoftheairinclosevicinityofthearc. Thesourceandloadsidecapacitancesarethemainfactorsthatinfluencethearc behaviourandtheinterruptionperformance.WithlowerratioCs/Cl,(specifically lower than 0.1 in the tests), the arc duration, overvoltage across the load side capacitance,contactsgaplength,bladeangleatfinalinterruption,andtheenergy inputintothearcaremuchlarger,whereasthere‐strikevoltageislowerdueto slowerrecovery.WithhigherCs/Cl,thearcexhibitsanerraticmodethateitheris sustainedorchangedtoastiffmode. The current to be interrupted is another governing factor, which affects the interruption performance. With higher current, the arc duration, overvoltage acrosstheloadsidecapacitance,contactsgapspacing,bladeanglesandarcenergy aremuchlargerbecausegaprecoveryisslowerandthere‐strikevoltageislower. With higher current, the arc is brighter; the arc remnants decay slower, the arc upwards velocity is not affected much. The arc overall brightness is also increasingwiththeinterruptedcurrent. Withlowerre‐strikevoltage,theenergyinputintothearcuponre‐strike,andthe overvoltage, over current are smaller. The re‐strike voltage depends on the momentary air gap length and its corresponding physical conditions. The arc energy, overvoltages and the transient current in the circuit reach the largest valuejustbeforethearcextinguishescompletely. Whenthegaprecoversrelativelyfast(atlowercurrent,largerratioCs/Cl)thearc durationisshort.Whenhighercurrentthermallyconditionsthegapbyreducing itsbreakdownvalue,thearcdurationcanbemuchlonger,oreveninfinite. The value of re‐strike voltage can never be higher than 2p.u., but the moment when this maximum is reached depends on the gap dielectric condition. The breakdown reduction mechanism will be discussed further in more detail in Chapter9. Whenthere‐strikevoltageishigherthantheamplitudeofthesourcevoltage,the high and medium‐frequency components are still strong when the next current zeroappears.Insuchacase,thearcdoesnotstopuntilanotherhalfcyclecurrent zero. If this phenomenon occurs continuously, the re‐strike voltage and the voltageacrosstheloadsideappearunipolar. Theopticalarcbehaviourassiststounderstandtheinterruptionmechanismmore profoundly. The re‐ignited arc follows the "traces" left by the just extinguished arc. That means the arc chooses the dielectrically weakest path for which the breakdownvoltageislowest.Theionizationremainingafterthedisappearanceof the past arc makes the arc re‐ignite or re‐strike at a lower breakdown voltage. 90Chapter5 Ways to shorten the arc duration, or equivalently, to increase the current interruptioncapability,mustbesoughtalongthisline. 5.9Recommendationforstandardization Themacroscopicarcbehaviourisstronglydependentonthecircuitasquantified by Cs/Cl. This observation implies that for testing of the DS switching capability, the circuit plays a major role (this also applies to the testing of auxiliary interruptingdevices).Sincenotest‐circuithasbeendefinedyet,oneofthetasksof theIECmaintenanceteam,elaboratinganamendmenttotheIECstandard62271‐ 102wastodefineacircuit.Itwasdecidedthat20CO(close‐open)testshavetobe performedwithCs/Cl=0.1,adoptingatest‐circuitasinFigure3.1.Alternative,but yetadequatesupplycircuits,supplyingmuchlessthantheshort‐timecurrentare being discussed. It was decided to give the document the status of a technical reportandallowtimeforcollectingexperience.ThetechnicalreportIEC/TR62271‐ 305wasissuedin2009. References [1] D. F. Peelo, "Current interruption using high voltage air‐break disconnectors", Ph.D. dissertation,Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven, 2004. [2] IEEEStandardTechniquesforHighvoltageTesting,IEEEStandard4‐1995,Aug.1995. [3] [Online].Available:http://www.hapam.nl/. [4] IECStandardon"IECstandardvoltages",IEC60038:Jun.2009. [5] S. Arigye‐Mushabe, K. A. Folly, "Evaluation of switching capacitive currents by disconnect switches using DIgSILENT software tool", in Proc. 2005 The Australasian UniversitiesPowerEngineeringConf.,vol.1,papernr:S036,Sept.2005. InterruptionwithAirFlowAssistance91 Chapter6 InterruptionwithAirFlowAssistance Air flow is an effective approach for arc quenching during current interruption withair‐breakDSs,asalreadyreportedin[1]‐[3].Researchonforcedairinjection into the arc ("air flow device") is rather old, and the studies were exclusively basedontheairflowinaverticaldirectionfacingthearcbodyfromunderneatha vertical‐breakDS.Thetestresultsprovedthismethodtobeeffective.However,all literature only focused on the basic macroscopic test results. A detailed analysis on the mechanisms improving the interruption capability is lacking. In this chapter an experimental study is presented, specifically focusing on arcing time, re‐strikevoltage,re‐strikefrequencyandderivativeofinterruptedcurrent(di/dt). Themechanismsthroughwhichairflowassiststheinterruptionarediscussed. 6.1Experimentalsetup ThetestcircuitdepictedinChapter4isextendedwithairflowoutletsasshownin Figure6.1.Theairflowiscontrolledthroughtwoseparatenylonairhosesending ateachtipofthemainblades, directedtowardsthearcroots. Thehoseopening area is approximately 50mm2, directed about 5cm from the expected arc root. Theoutletis mountedon themovingblades, ensuringasteadyairflowtowards both arc roots. Two levels of air flow have been injected into the arc: 215litre/minute and 325litre/minute. The experiments are conducted with current values Id of 0.5A, 1.4A, 2.6A, 3.9A, 5.0A, 6.3A and 7.5A at 90kV. The valueofCsistakenas50nFandClrangesfrom18nFto265nF.ThevalueofIdis determinedbyClandtheotherparametersinthecircuitarefixed. Switchingoperationsareperformedforeachappliedcurrentlevel: ‐ withoutairflow(denotedascaseA). ‐ withairflowrate215litre/minute,airvelocity72m/s(denotedascaseB). ‐ withairflowrate325litre/minute,velocity108m/s(denotedascaseC). The air flow rate on each side is measured by an air flow meter. Three opening operationsareperformedateachcurrentlevelandeachairflowlevel.However, case A was not applicable at 7.5A, because the DS main blades reached the maximum position just after final arc extinction already at 6.3A, which implied that interruption at 7.5A without air flow would fail. Case B tests were not performedforcurrentlevelsabove2.6A,becauseoflaboratorytimerestrictions. 92Chapter6 Figure6.1.DSincludingvoltagedividersandnylonhosesinjectingairflowfromunderneath theDSbladesatbothsides. In all switching operations, the arc extinguished completely before the main bladesoftheDSreachedthefinalopeningposition.Thevoltagesacrossthesource andload‐sidecapacitors(ucs,ucl),andthecurrentflowingthroughtheDSidwere recorded. The measuring and data acquisition systems are described in detail in Chapter 4. The experiments were performed at KEMA High Power Laboratory Arnhem,whereastrongvoltagesourceisavailable. 6.2Effectofairflowonarcing The effect of air flow on the interruption is strongly reflected in the measured currentandvoltagewaveshapes,andintherepetitivere‐strikepatterns. 6.2.1Measuredcurrentandvoltage The current and voltage wave shapes from case A (no flow) differ significantly fromcasesB(moderateflowrate)andC(highflowrate).Figure6.2illustratesthe wave shapes of the voltages ud across the DS (ud=ucs‐ucl) and the current id throughtheDSforthethreedifferentcasesatcurrentId=2.6A.Experimentsat other current levels show similar qualitative behaviour in spite of a varying arc duration. Figure 6.3 contains the expansions of the wave shapes taken over the periodt=1000~1150ms(t=0indicatesthearcstart).Theairgaplengthhardly changesduringthis150mstimespan. Inallcasesitisobservedthat:(i)thearcingdurationisintheorderofseconds; (ii)theentireinterruptionconsistsofnumerousinterruptionsandre‐strikes/re‐ ignitions; (iii) the electrical transients occur at each re‐strike/re‐ignition, which causes overvoltages and (local) current spikes in the local circuitry. However, differencesbetweentheinterruptionswith‐andwithoutairflowareapparent. InterruptionwithAirFlowAssistance93 Looking at Figure 6.2, in case A, the arc duration is 2.3s. Mostly, there is only a single re‐strike within each half power‐frequency cycle as shown in Figure 6.3. Oncethere‐strikeoccurs,thearccontinuestillthenextpower‐frequencycurrent zero. ThearrowsinFigure 6.3illustratere‐strikeandsubsequentarc extinction. Comparedtothewaveshapesobtainedwithairflow(casesB,C)there‐strikesfor case A occur at lower breakdown voltage and are associated with smaller transients. For instance, in Figure 6.2 the re‐strikes between 0 and 1000ms for case A are hardly visible (re‐strike voltage remains below 20kV); the re‐strikes after 1000ms are less frequent compared to cases B and C. These observations reflecttheslowrecoveryofthedielectricstrengthoftheairgap. ComparedtocaseA(2.3sarcduration),thearcdurationismuchshorterinboth caseBandC(1.6sforcaseBand1.4sforcaseC);inthelatersituations,thereare multiple re‐ignitions occurring within each half power cycle (compared to only onere‐strikefollowingcurrentzero);there‐ignitionsoccurmorefrequentlyand showhighervoltagetransientamplitudes. Compared to case B the arc duration in case C is even shorter; the re‐ignitions occur more frequently and the breakdown voltages are higher (see the period from0to500msinFigure6.2).Theexpandedwaveshapesofvoltageudbetween 800msand950msareplottedinFigure6.4.ComparedtocaseAthebreakdown voltages in case B and C are much higher for identical air gap length. The breakdown voltages for case C are usually larger than for case B as well. This means that applying a higher amount of air flow helps to interrupt the circuit more successfully. The air flow cools the former arc path and removes localized patches of ionization and thereby increases the dielectric strength of the same volumeofairinthegap.ThiswillbeexplainedfurtherinChapter9.Ontheother hand,thenumberandmagnitudeoftransientsincurrentandvoltagearehigher. 6.2.2Recoveryvoltage Recoveryvoltageisthedifferencebetweenthevoltageacrossthesourceandload side capacitance after arc extinction at current zero. Without air flow RV rises right after current zero as a (1‐cos) wave shape in one polarity until re‐strike occurs,asdiscussedinChapters3and5.Ifthearcisforcedtoextinguishbefore power‐frequencycurrentzero,asinthepresenceofairflow,RVwillmaintainits originalpolaritybeforere‐ignitionandcannotreachthemaximumvalueof2p.u. asinthecaseofinterruptionatcurrentzero.Figure6.5showsasinglerecovery event in both situations. Without air flow, the RV increases directly to negative values(Figure6.5left).Withairflow,theRVmayfirstchangepolaritybeforere‐ ignition(Figure6.5right).Withoutairflow,atcurrentzerothearcextinguishesat the maximum voltage across the load side. However, the arc can be forced to extinguishbeforepower‐frequencycurrentzerobecauseofsuper‐imposedhigh‐ frequencyoscillations,whicharemoreapparentinarcinterruptionwithairflow. 94Chapter6 id (A) 20 0 Case A −20 200 d u (kV) 100 0 Case A −100 −200 0 500 1000 1500 2000 20 Case B id (A) 10 0 −10 −20 200 ud (kV) Case B 0 −200 0 500 1000 1500 2000 20 Case C id (A) 10 0 −10 −20 200 ud (kV) Case C 0 −200 0 500 1000 time (ms) 1500 2000 Figure6.2.Waveshapesofid,udforcasesA(noflow),B(215litre/min)andC(325litre/min). InterruptionwithAirFlowAssistance95 30 Restrike occurs i (A), u (kV) 20 d 10 0 d −10 −20 Arc extinction at current zero −30 Case A 1000 1050 1100 1150 id (A), ud (kV) 100 50 0 −50 −100 Case B −150 1000 1050 1100 1150 id(A), ud (kV) 50 0 −50 −100 id u d Case C −150 1000 1050 1100 time (ms) Figure6.3.Expansionovertherange1000~1150msinFigure6.2. 1150 Thevoltageacrosstheloadcapacitancemaynothavereacheditsmaximumvalue (thesourcevoltageamplitude). Figure6.6presentsresultsfromsimulations,showingthewaveshapesofidandud assuming various power‐frequency phase angles where the arc current is chopped.Att=20msthearccurrentisatmaximum.Att=25msthearccurrent crosseszero.Thecurrentsidi(i=1…6),andvoltagesudi(i=1…6)arethesimulatedresults when the arc is chopped at 20 ms, 21 ms, 22 ms, 23 ms, 24 ms, 25 ms. 96Chapter6 100 case B case A 0 d u (kV) 50 −50 −100 case C −150 830 840 850 860 870 time (ms) 880 890 900 Figure6.4.Expansionofudbetween800ms~950msforcasesA,BandC. 30 re−strike re−strike ud u (kV), i (A) extinguish 0 i 20 d d d u (kV), i (A) 5 −5 d extinguish 0 i d −10 10 d d u −10 0 1 time (ms) 0 2 1 2 3 time (ms) 4 Figure6.5.Singlerecoveryevent,(left)withoutairflow,(right)withairflow. 12 i d1 i −100 d2 id3 8 Voltage (kV) Current (A) 10 i d4 i 6 d5 id6 4 2 −150 ud1 u d2 −200 u d3 ud4 −250 ud5 u 0 d6 20 21 22 time (ms) 23 20 25 30 time (ms) Figure6.6.Simulatedwaveshapesforidandudatvariousarc‐choppingphaseangles. 35 ThesimulatedresultsshowthattheRVacrosstheDSreacheslowervalueswhen arc current is chopped closer to its peak value. The RV is equal to ucs plus the offsetUcl.Sincethesourcesideucsremainsnearlyequaltous,itmeansthattheRV dependsonUcl.Thelargerthearccurrentatchopping,thesmallerUclis,andthe smaller the maximum value that the RV may reach. With air flow, the arc is InterruptionwithAirFlowAssistance97 choppedatvariousphaseanglesapartfromthepower‐frequencyarccurrentzero. Thus,thevaluestheRVcanreacharesmallerthanwithoutairflow.Withasmaller RV,itismoredifficulttobreakdowntheairgapforthenextre‐strike. 6.2.3Arcmodes Basedonthepreviousobservations,itmightbestatedthatthearcappearsintwo distinct modes. One mode is referred to as "thermal arc", usually occurring for case A. It is associated with a sequence of arc extinctions at (power‐frequency) currentzeroandre‐strikesuntilsufficientgapdistancehasbeenreached.During every half cycle the arc remains stable. Transients are limited to re‐strikes that occur every half cycle. The other mode is referred to as "dielectric arc", usually occurringincaseswithsufficientairflow(casesBandC)orothercooling.Here, thearcisaveryconcentratedsuccessionofarcinterruptions(mostlyawayfrom naturalcurrentzero)andre‐ignitions.Thearcischoppedatvariousphaseangles withinapowercycle.Apparently,theairgapdielectricrecoveryspeedisgreatly increasedcomparedtocaseA,becauseofcoolingofthe(former)arcpathbythe air flow. Visually, the "thermal" arc duration is longer, contains more curls than thedielectricarc,hasalessviolentappearanceandaudiblenoisethanthearcin the "dielectric" mode. Examples of arcs in both modes and their re‐strikes are showninFigure6.7. 6.3Interruptiondataanalysis For the analysis of the experimental data the main focus will be on the arc duration,whichisobviouslydifferentforthecaseswithandwithoutairflow.The re‐strikevoltage,thenumberofre‐strikespercycle,andtheinterruptedcurrent derivativedi/dtarestudiedaswellinordertounderstandtheeffectofairflowon theinterruptionmechanism. 6.3.1Arcduration ThearcdurationisinfluencedsignificantlybytheinterruptedcurrentIdandthe airflowrateintheexperiments.Figure6.8presentsthearcdurationasafunction of the interrupted current with the air flow as a parameter for all interruption operations.Obviously,thearcdurationforcaseAismuchlongerthanforcasesB andC;thearcdurationincaseCisnearly50%reducedwithrespecttocaseA.In addition,atidenticalinterruptedcurrentlevel,thearcdurationdecreaseswitha higheramountofairflowrate.ItshouldbenotedthatthecaseBtestswereonly performedupto2.6A.ThelinearinterpolationlinesforcasesAandCshowthat thearcdurationtendstoincreasewithincreasinginterruptioncurrentIdinboth cases (however with a large spread). This is in agreement with the results presentedinChapter5forinterruptionswithoutairflow. 98Chapter6 Thermal arc Thermal arc re-strike Dielectric arc Dielectric arc re-strike Figure6.7.Arcmodes:(left)steadystateoperation;(right)attheirre‐strikemoments. Arc duration (s) 2.9 case A case B case C Linear (case A) Linear (case C) 1.9 0.9 0 2 4 Current (A) 6 8 Figure6.8.ArcdurationversusthecurrentIdwithairflowrateasparameter. 6.3.2Re‐strikevoltageversustime Measuredvaluesfromthreerepeatedoperationsatthesamecurrentlevelandair flow rate are combined to get statistically relevant re‐strike voltage values in order to detect a possible trend. The values of the re‐strike voltages are divided into 10 time intervals (bins), equally distributed over the interval from the momentthatthearcstartsuntilthemomentthatthearcextinguishescompletely. InterruptionwithAirFlowAssistance99 Themedianvaluesofthepositivere‐strikevoltagesineachbinversustimeusing airflowlevelasaparameterarepresentedinFigures6.9and6.10.Negativere‐ strikevoltagesshowsimilarbehaviour.Timet=0isthemomentofarcinitiation. FromFigures6.9and6.10,thefollowingconclusionsaredrawn: ‐ Foragiventime(orDSgaplength),normallythere‐strikevoltagesincasesB andCaremuchhigherthanincaseAforeverycurrentlevel. ‐ Atalowcurrent(0.5Ainthesetests),however,there‐strikevoltageforthe caseswith‐andwithoutairflowaresimilar.Theairflowdoesnotaffectthe interruptionasmuchasitdoesforacurrentof1.4Aorhigher.There‐strike voltage at 0.5A for the three cases increases approximately linearly with time (i.e. with the air gap length). This suggests that thermal influences on therecoverydonotplayamainroleyetduetothesmallcurrent. Forcurrentlevelsupto2.6A,whereexperimentsforbothcaseBandCare ‐ carried out, the re‐strike voltages of cases B (moderate air flow rate) and caseC(highairflowrate)aresimilar. ‐ There‐strikevoltageismainlydeterminedbytheinterruptedcurrentincase A. With higher current, the re‐strike voltage is lower, which was also observedinChapter5,wheretheinterruptedcurrentwasbelow2.3A. ‐ Without air flow, the re‐strike voltages clearly depend on the current level during the entire interruption process. On the other hand in cases B and C, during the first 1500ms of the interruption process, the influence of the currentislimitedsincethedifferencebetweenre‐strikevoltagesatdifferent currentlevelsishardtobeobserved(inparticularforcaseCinFigure6.10). However, at the end of the interruption process, the re‐strike voltages correlatebetterwithcurrentlevel.Itisalsoobservedthattheinfluenceofthe currentisnotsoobviousforcaseCasitisforcaseB;thisimpliesthatwith higherairflowrate,thecurrentinfluenceissmaller. It is concluded that both air flow and interrupted current play key roles in the interruption process. For rather small current (roughly < 1A), the thermal influencedoesnotaffectthere‐strikevoltagevaluesignificantly,andtheairgap recoversmainlydielectrically.Whenthecurrentislarger(roughly>2Ainthese experiments),thethermalinfluencesstarttodominatetheinterruption.Iftheair flowrateishigh,there‐strikevoltagevaluesatdifferentcurrentlevelsaresimilar at the most stages of the interruption, which means that the air flow mainly dominates the interruption, and the current level plays a less prominent role in currentrangechosenintheseexperiments. 6.3.3Breakdownfrequency Obviouslythenumberofbreakdowneventsperunittime(breakdownfrequency) dependsonairflowrateandinterruptedcurrent.Figures6.11and6.12showthe CumulativeNumber(CN)ofbreakdowneventsversustimeatallappliedcurrent 100Chapter6 200 200 1.4A 0.5A 100 100 0 0 200 500 1000 0 2000 0 200 1500 2.6A 100 1000 1500 2000 100 0 0 200 1000 2000 0 3000 0 1000 2000 3000 1000 2000 3000 200 6.3A 5.0A 100 100 0 0 200 7.5A 100 0 0 Ur (kV) 500 3.9A 1000 2000 1000 0 3000 2000 0 0 l/m 325 l/m 215 l/m 3000 time (ms) Figure6.9.Re‐strikevoltage(inkilovolts)versustime(milliseconds)withairflowrateasa parameteratcurrentrange0.5~7.5A. 300 300 case B 0.5A 200 1.4A 2.6A 3.9A 100 0 6.3A 0 500 200 case C 150 Voltage (kV) Voltage (kV) voltage (kV) case A 1000 1500 2000 time (ms) 2500 0.5A 1.4A 5.0A 100 3000 0.5A 1.4A 200 2.6A 100 0 0 500 1000 time (ms) 1500 2000 6.3A 7.5A 50 0 0 500 1000 1500 time (ms) 2000 2500 Figure6.10. Gaprecoveryreflectedinre‐strikevoltage(kilovolts)versustime(milliseconds) withcurrentIdasaparameterforcasesA,BandC. levels.AcontinuoustheoreticalCNlineofonere‐strikeper10msasareferenceis addedatthebottomofFigure6.12.Thefollowingbehaviourisobserved: ‐ Sincethearcdoesnotonlyextinguishatpower‐frequencycurrentzero,but isalsochoppedatotherphaseangles,atanidenticalarcingtimetheCNsfor casesBandCaremuchlargerthanforcaseA,ateverycurrentlevel. ‐ At 0.5A, the CN per unit time is more or less similar for cases B and C. However,at1.4Aand2.6AtheCNforcaseCislargerthanforcaseB.Ahigh airflowrateresultsinmorefrequentbreakdownevents. InterruptionwithAirFlowAssistance101 ‐ AtallcurrentlevelsforcaseA,theCNincreasesrapidlyattheverybeginning of the interruption process. This rapid increasing stage lasts longer in case the current is low compared to higher current. After this stage the CN increases more slowly and linearly. The rate of increase is about 0.1 per millisecond. The initial high re‐strike frequency is because the thermal influenceplaysnomajorroleyet.Lateron,thearcexistscontinuouslytillthe nextcurrentzerocrossingduetothethermalinfluences. ‐ For case C in Figure 6.12 the CN is decreasing with the increasing current level.Thisimpliesthattheeffectofthecurrentbecomesmorepronouncedat highercurrentlevel. Asaconclusion,atalowcurrent(0.5Ainthetest)andintheveryearlystageof theinterruptionprocessesathighercurrents,thecurrentinfluenceonbreakdown frequency is relatively small. The interruption mode is mainly dielectrically determined. In a later stage of the interruption process with higher current, the currentleveldominatestheinterruptioninabsenceofairflow.However,athigher currentlevels(>1.4A),theairflowbecomesakeymechanism.Bothairflowand currentleveldeterminethearcduration. 6.3.4Currentderivativedi/dt The current derivative di/dt at the moment of current zero is (for circuit breakers) a key factor to evaluate the interrupting performance. Since the arc extinguishestemporarilywhenthecurrentcrosseszeroateachhalfcycleincase A, the derivative di/dt is calculated from the current waveform where is the rms value of the arc current; is the angular √2 sin frequency (2 ∗ 50Hz), and is the initial phase angle. The derivative di/dt at current zero is √2 .In case A, the current ranges from 0.5A to 6.3A, hence 0.2~3 A/ms. However, the arc can be chopped at any moment for caseBandC,whichmeansthatcurrentzeros,leadingtointerruption,arecaused by the high‐frequency current transients with much higher di/dt than quoted above. di/dt is calculated from the measurements at each arc extinction point. The calculateddi/dtvaluesarestatisticallydistributedvariables.Theirpropertiesare analyzed with the aid of Figures 6.13 and 6.14. Results are shown for different currentlevelsandforthecasesA,BandC.Noticeably,thedi/dtvaluesat0.5Aare notdependingontheairflowrate,whichconfirmstheconclusiondrawnearlier on thermal influences not being relevant yet. When the current exceeds 1.4A, di/dtishigherwithairflowthanwithoutairflowateachcurrent level.At1.4A and 2.6A, the di/dt in case B is lower than that in case C, which means di/dt is largerwithhigherairflowrates. 102Chapter6 2000 0 0.5A 1000 CN 3000 1000 2000 0 CN 2000 3000 0 7.5A 0 1000 2000 3000 time (ms) 1000 2000 3000 4000 time (ms) case A case B case C 1000 0 1000 6.3A time (ms) 2000 3000 500 0 1000 2000 3.9A 1000 5.0A 0 1000 1000 3000 time (ms) 1000 0 0 time (ms) 0 0 1.4A 2000 2.6A 2000 CN 2000 time (ms) 1000 0 1000 0 0 2000 CN CN 1000 CN CN 2000 2000 3000 Figure 6.11. Cumulative number of breakdown events versus time with the current as parameterforcasesA,BandC. time (ms) 1500 case B case A 1500 1000 1.4A CN CN 0.5A 2.6A 3.9A 500 1.4A 0.5A 1000 2.6A 500 6.3A 0 0 1000 case C 1500 CN 500 1500 2000 time (ms) 2500 3000 0 0 1000 2000 time (ms) 3000 1.4A 2.6A 3.9A 1000 7.5A 500 0 0 500 1000 1500 2000 time (ms) 2500 3000 Figure6.12.Numberofre‐strikesversustimeforcasesA,BandCatdifferentvaluesofId. InterruptionwithAirFlowAssistance103 100 0 case A case B case C 0 100 500 1000 di/dt (A/ms) Percentile 50 500 1000 di/dt (A/ms) Percentile 100 500 1000 di/dt (A/ms) 1500 1500 500 1000 di/dt (A/ms) 1500 7.5A 50 0 0 0 100 5.0A 1000 di/dt (A/ms) 50 1500 50 0 500 3.9A 0 0 0 100 2.6A 0 50 0 1500 Percentile 50 1.4A Percentile 0.5A Percentile Percentile 100 0 500 1000 di/dt (A/ms) 1500 Figure6.13.Cumulativedistributionofinterrupteddi/dtforthecasesA,BandCatdifferent currentvalues. SincethetestsforcaseBarenotdoneatcurrentabove2.6A,thevalueofdi/dtis onlyevaluatedforcasesAandC.The50‐percentile(medianvalue)di/dtvaluefor case A, is in the range 4 to 25A/ms (with low current of 1.4A). This value is comparable with the estimated values before, assuming only arc extinction at power‐frequency current zero. The median values of di/dt in case C range from 100 to 300A/ms, several ten times higher than for case A. Obviously, the interruptibledi/dt with air flow is drastically higher than without air flow. This againconfirmsthestronginfluenceofairflowontheinterruptionprocess. 6.4Conclusion Two interruption modes are observed, a dielectric and a thermal mode. The dielectric mode interruption mainly occurs in the case with a high air flow rate (here 315 litre/minute) and with arc current roughly below 1A. The thermal modeoccursinthecasewithoutairflowathigherinterruptedcurrent(>1.4Ain the present experiments). The main differences between these modes can be summarizedasfollows: Duringthedielectricmode,thearccurrentmaybechoppedatanymomentinthe power‐frequency cycle for the case with sufficient air flow. Multiple breakdown 104Chapter6 100 Percentile 80 1.4A 2.6A 3.9A 60 40 Case A 6.3A 20 0 0 60 0.1 0.2 0.3 0.4 di/dt (A/μs) 0.5 0.6 0.7 0.8 50% line 40 20 0 0 0.005 0.01 0.015 0.02 0.025 100 0.5A 1.4A 2.6A 3.9A 5.0A 6.3A 7.5A Percentile 80 60 Case C 40 20 0 0 0.2 0.4 0.6 0.8 di/dt (A/μs) 1 1.2 1.4 1.6 80 60 50% line 40 20 0.1 0.15 0.2 0.25 0.3 Figure 6.14. Cumulative distribution of interrupted di/dt for cases A and C with Id as a parameter. eventswithinahalfpower‐frequencycycleareverylikelyandthenumberofthe breakdowneventsperunittimeismuchhighercomparedwiththethermalmode interruption.Atthesameinterruptedcurrentlevelthebreakdownvoltage,mainly dominatedbytheairgaplength,ishigherthanforthethermalinterruptionmode. InterruptionwithAirFlowAssistance105 Both the higher breakdown repetition frequency and the higher value of the breakdownvoltagemakethismodepotentiallymorehazardousforneighbouring substationequipment(specificallypowertransformers).Theinterruptedcurrent derivativedi/dtatthecurrentzeroismuchhigher(aboutafactor20)thanforthe thermal interruption mode. This implies that the arc current can also be interrupted at current zero from higher frequency oscillations than power‐ frequency,causedbythetransientsinthecircuit. During the thermal interruption mode, the arc can initially be chopped at any moment at the very beginning of interruption process due to the very short air gap length. However, after (1‐2 half cycles), the arc extinguishes at the power‐ frequency current zero only. The interrupted current derivative di/dt increases roughly linearly with the interrupted current, except in the small gap initial arc phase. Due to thermal influences, the breakdown voltage is much lower than in thecaseofthedielectricinterruptionmodeatthesameairgapdistance.Thisis,in principle, less hazardous for voltage transients striking nearby equipment. However, ultimately the overvoltage is at the same level, but reaches at a much longergapthaninthedielectricmodeandoccurslessfrequently. Since air flow may convert thermal interruption into dielectric interruption, the dielectric strength of the air gap increases, as well as the corresponding interruptibledi/dt.Theairflowisaneffectivemethodtoimprovetheinterrupting capability of the air‐break DS to interrupt small capacitive currents. However, significant values (here 315litre/minute) at both arc roots during the complete arcing process have to be realized. This limits its practical application because compressorsandhigh‐pressurecontainersshouldbeavailable.Alternatively,one could imagine creating the necessary pressure simultaneously with the blade rotating motion. A better directed air flow could reduce the necessary air flow rate. References [1] E.C.Rankin,"Experiencewithmethodsofextendingthecapabilityofhighvoltageair‐ break switches", AIEETrans.onPowerApparatusandSystems, vol.78, pp. 1634‐1636, Dec.1959. [2] I. W. Gross, C. Killian, J. M. Sheadel, "A 330‐ kV air switch", AIEE Trans. on Power ApparatusandSystems,vol.73,pp.264‐270,Jan.1954. [3] A. Foti, J. M. Lakas, "EHV switch tests and switching surges," IEEE Trans. on Power ApparatusandSystems,vol.83,pp.266‐271,Mar.1964. 106Chapter6 High‐velocityOpeningAuxiliaryInterrupter107 Chapter7 High‐velocityOpeningAuxiliaryInterrupter A device with fast opening contacts called high‐velocity auxiliary interrupter, attachedtothemainbladesoftheDS,hasbeenincreasinglyemployedsincethe 1950s.Thisdevicehasbeenusedmostlyonvertical‐breakDSandmainlyinNorth America. According to [1] a high‐velocity interrupter using a mechanical spring can greatly reduce the arcing time by achieving a large air gap in a short time intervalstartingattheinstanttheauxiliarycontactsrelease. 7.1Operatingprinciple The working principle of a centre‐break DS with high‐velocity auxiliary interrupterisillustratedinFigure7.1.Inparalleltothemaincontactbladesofa DS, there are two auxiliary steel springs, which are also referred to as "whips". They are short‐circuited by the main contacts when the DS is in closed position. When the DS main contacts separate the current commutates from the main blades to the auxiliary contacts and the arc does not yet start. The whips bend (thereby tensioning the springs) during opening until the contact at the hook releases. At that moment, the whips act as released springs and the arc arises betweentheauxiliarycontacts.Thepotentialenergystoredinthebended"spring" contacts,togetherwiththeirlowmass,causeafastaccelerationandprovidesfora rapid opening. Because of the high opening speed the arc extinguishes rapidly. Thecircuitisinterruptedandavisiblecontactseparationisprovided[2]. Theincreasedinterruptioncapabilitybythehighcontactopeningspeedofquick‐ break devices can be explained qualitatively. Assuming the arc re‐strikes at the momentofmaximumRVandcontinuestoburnforanotherhalfcycle(T½c)tillthe nextcurrentzerocrossing,thearcdurationTarccanbeestimatedroughlywith: T a rc 2 .2 U lin e T 1 c (7.1) 2 3 v a ir E Here,vairistheopeningvelocityoftheairgap;Eistheelectricfieldstrengththe air gap can withstand; Uline is the phase to phase voltage. To allow for a 10% margin,thesystemvoltageistakentobeat1.1p.u.,soaRVpeakvalueof2.2p.u. is adopted[3]. Equation (7.1) shows that with larger E and vair, the arc duration becomesshorter.ThebreakdownfieldstrengthEisaffectedbymanyfactors,such asthermaleffectsfromthearccurrent,bladegeometry,gapdistance,andsoforth. 108Chapter7 Figure 7.1. Operating principle of a centre‐break DS with a high‐velocity auxiliary interrupter. With higher vair the required air gap distance is reached faster and the arc durationisshorter. There are only a few publications related to the innovation by the high‐velocity auxiliary interrupter [2], [4]. Moreover, there is hardly any information on the interruptionprocessitselfandonpotentialcausesofoccasionalfailuressincethe limited available literature merely focused on test results rather than their explanations. In this chapter a series of experiments is described in order to understand the interruption phenomena. The reason for occasional interruption failureisdiscussedintermsofarcenergyandre‐strikevoltage. 7.2Experimentalsetup The experimental setup, together with the measuring system is identical to the simplifiedcircuitdiagramdepictedinChapter4.Thesetupandaclose‐upofthe DS,withtheattachedauxiliaryinterrupter,arepresentedinFigure7.2.Thetested DS is a centre‐break type with a rated voltage of 145 kV. The experiment is performedat90kVandafixedvalueforCsof50nF.ThecurrentIdiscontrolled withthevalueofClanditsrangeischosenbetween5.3Aand27A. The opening velocity of the auxiliary contacts is derived from the momentary distanceoftheseparatingcontactsasafunctionoftimebyanalyzingthearcimage frames. Figure 7.3 (left) shows the determination of the air gap length d by comparing distance D1 between the contacts in each image with the known distanceD2betweenthefixedsidesoftheDS(1650mm:acentre‐breakDSfrom High‐velocityOpeningAuxiliaryInterrupter109 Figure 7.2. Laboratory setup (left) and DS (right) with high‐velocity auxiliary device (its contactsareinsidethedrawnellipse). Distance (mm) 400 300 200 100 0 0 10 20 time (ms) 30 40 Figure7.3.Distancebetweentheauxiliaryinterruptercontacts;arcimageexample(left),and contactdistanceasafunctionoftime(right,releasesatt=0). HAPAM). Data obtained from all experiments are combined in Figure 7.3 (right) togetherwithaquadraticregressionaccordingtod=1/2at2+v0t,whereaisthe acceleration and v0 is the initial velocity when the interrupter contacts release. Thefitresultsinanaccelerationof260m/s2andaninitialvelocityof2m/s.The average speed during the complete opening process is 11m/s, more than ten timeshigherthantheseparationvelocityoftheDSmainblades(referChapter4, and[2]). 7.3Overviewoftheexperimentalresults The experiments have been arranged such that there are at least three times repeated(closingand)openingoperationsperformedforeachcurrentlevel.Only the opening operation is studied. The results are summarized in Table 7.1. Successful interruption means that the arc only burns between the auxiliary 110Chapter7 Table 7.1. Summary of the experiments at 90 kV on interruption with rapid auxiliary contacts. CO Id(A) 5.3 8 9.5 10 12.5 15 17 20 23 25 27 1st 2nd 3rd 4th 5th 6th ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ Note:‘CO’meansclose‐openoperation, :successfulinterruption; :failedinterruption;‐:no test. contactsandextinguishesbeforetheyreturntotheiroriginalpositions,parallelto themainblades.Aninterruptionfailswhenthearccommutatesfromtheauxiliary contactstothemainbladesandthearcdoesnotextinguish. Table 7.1 shows that the interruption has a higher likelihood to be successful at lower current levels. Below 10A the interruptions always succeeded. However, the failure probability does not relate directly to the current Id. For example, interruption sometimes failed at 10A and 12.5A, but always succeeded at 15A and17A.InadditiontothetestsindicatedinTable7.1,aseriesof20repeatedCO tests was performed at 7A with 100kV supply voltage. All interruptions are successful,butfor10Acurrentonlyoneoutoffsixissuccessfulat100kV.With increasing Id, there is an overall tendency of current interruption to fail. The reasonsforfailurewillbediscussedinthefollowingsections. 7.3.1Electricalandopticalobservations Figure7.4ashowsthewaveshapesofthevoltagesucs,uclandthecurrentid(with low and high‐frequency bandwidth, denoted as ipf and ihf respectively) for a successful interruption at Id = 12.5A. The voltage ud across the DS is shown in Figure7.4btogetherwiththecurrentidinhighverticalresolution.Timet=0isthe moment when the auxiliary contacts release and the arc ignites. The arc does extinctafter73ms.Itisobservedthatthereareseveralre‐strikeswiththeusual high‐frequency oscillations during the arcing period. A few re‐strikes with clear oscillationsareindicatedwith‘1’,’2’,’3’inFigure7.4b.Themaximumamplitude of the transient current reaches 5.9kA as shown in Figure 7.4. The overvoltages acrossCs,ClandtheDSare2.6p.u.,2.5p.u.and2.0p.u.,respectively. High‐velocityOpeningAuxiliaryInterrupter111 The arc image is recorded with a speed of 1000frames/s. A few frames from characteristic instances directly after the arc initiation are shown in Figure 7.5 with the time indication inside of the pictures. The images show that the arc mainlyextendsstraightbetweentheauxiliarycontacts.Theframeat63mswith thehighestbrightnesscorrespondstothemostseverere‐strike,indicatedbyR1in Figure7.4a.Theframetakenatt=73ms,aboutahalfcyclelater,showsthelast frame before the arc extinguishes definitely and the frame taken at t=83ms, againahalfcyclelater,showstheafterglowatthearcremnants. Figure7.6apresentsthewaveshapesofthevoltagesucs,uclandthearccurrent(ipf, ihf)forafailedinterruptionatId=27A.Figure7.6bzoomsinonthewaveshapeof udtogether with the current ipf. Preceding the severe re‐strike at t=88ms, there areseveralre‐strikesaftertheauxiliarycontactsarereleased,indicatedwiththe numbers‘1’,’2’,’3’.TheRVbeforethelastre‐strike,indicatedbyR2,isclosetothe maximumRVof254kV(2Em)followedbyatransientcurrentpeakat6.5kA.The overvoltages across Cs, Cl and the DS reach 2.6p.u., 2.6p.u., and 2.0p.u., respectively. After that, the arc is re‐established between the main blades (not between the auxiliary contacts) until it is finally interrupted by the master breaker after 600ms. After the severe re‐strike at R2 there are no apparent re‐ strikes and no obvious arc current‐less periods are observed; the arc burns continuously, implying thermal re‐ignition after every current zero, with no measurablevoltageneededtore‐ignitethearc.Theaveragearcvoltageincreases graduallywithtime(andarclength)uptoalevelof10kV. ImagestakenatinstancesdirectlyafterthearcinitiationareshowninFigure7.7, wherethearcingtimeisindicated.Itisobservedthatthearcisfirstburningina more or less straight appearance between the auxiliary contacts and then transferssuddenlytothemainbladesofDSafteraseverere‐strike(R2)(brightest frameinFigure7.7).Atthatinstancethestiff,lineararcisconvertedtoanerratic, curlingarcmovingupwards,drivenbytheupwardmotionoftheheatedair. While burning between the auxiliary contacts, for both successful and failed interruptions,thearctendstofollowtheshortest,straightpathbetweenthe(fast moving)auxiliarycontacts.Itgetsacomplexerraticthree‐dimensionalstructure after commutating to the (slow moving) main blades of the DS in a failed interruption.Comparingsuccessfulandfailedinterruptions,thetransientcurrent andovervoltagevaluesamplitudeshardlydiffer.Theovercurrentandovervoltage also show no significant differences between a DS equipped with fast opening contacts and a DS only having the main blades. The observed overvoltages reported in [4] reach 2.3p.u., but this value was obtained at lower interrupted currents Id≤2.3 A. That means the high‐velocity interrupter does not contribute to a reduction of the overvoltage and transient current. However, since the duration of a successful interruption is an order of magnitude shorter even at significantlyhighercurrent,thenumberofre‐strikesismuchlower. i (kA) ipf (A) ucs(kV) ucl (kV) hf 112Chapter7 400 200 0 −200 400 200 0 −200 50 R arc extinguishes 1 0 −50 6 2 0 −2 0 20 40 60 80 100 120 140 time (ms) (a) d pf u (kV), i (A) 300 ipf 200 ud 1 2 3 100 0 Arc extinguishes 0 10 20 30 40 50 60 70 80 90 100 time (ms) (b) Figure7.4.(a)Waveshapesofucs,ucl,ipf,ihfandudforasuccessfulinterruption.R1indicates the most severe re‐strike; (b) ‘1’, ’2’, ’3’ indicate clearly observable re‐strikes just before arc extinction.Thecirclesinthelowerfigureindicatethere‐ignitions/re‐strikes. Figure7.5.Arcimagesatcharacteristicinstancesfromasuccessfulcurrentinterruption.The timeoftheFigure7.5correspondstothetimeofFigure7.4a. 200 0 cs cl u (kV) 200 0 −200 u (kV) High‐velocityOpeningAuxiliaryInterrupter113 0 pf i (A) −200 100 6 hf i (kA) −100 0 −2 R2 0 100 200 300 400 500 600 time (ms) (a) ipf (A) , ud (kV) 100 Arc extinguishes 0 −100 ipf −200 ud −20 0 20 Thermal Re−ignition 1 2 3 40 60 80 100 120 150 time (ms) (b) Figure7.6.(a)Waveshapesofucs,ucl,ipf,ihfandudforafailedinterruption.R2indicatesthe most severe re‐strike; (b) ‘1’, ’2’, ’3’ are the re‐strikes before the arc transfers to the main contacts. Figure7.7.Arcimagesfromafailedinterruption.ThetimeoftheFigure7.7correspondsto thetimeofFigure7.6a. 114Chapter7 Two distinct types of arc continuation are distinguished. One is defined as dielectricallydominatedre‐strike,whichmeansthatthebreakdownoftheairgap occurs in (largely) unconditioned air (degree of ionization and thermal effects hardlyaffectthedielectricstrengthofthegap)atahighfieldstress.Thesecond type is defined as thermally dominated re‐ignition, which means that the breakdown occurs at a lowered voltage due to a relatively high thermal energy densityinthearcregion.Therefore,thedielectricre‐strikevoltageismuchhigher thanthethermalre‐ignitionvoltageforanidenticalairgapdistance.Theclearly visiblere‐strikeswithnumbers‘2’,’3’inFigure7.4band‘1’,’2’,’3’inFigure7.6b are dielectrically dominated. Thermally dominated re‐ignition appears in the failed interruptions at higher current. Thermal re‐ignition occurs in failed interruptions after the arc has transferred to the main blades. There are no furtherclearobservablere‐ignitionsorre‐strikes,whilethearccontinuesburning between the slowly opening main blades (see Figure 7.6a). Although the arc currentperiodicallycrosseszero,thearccontinuestoexist,sustainedbythermal energy. The arc always re‐ignites immediately along its prior path just after the currentzerocrossing. 7.3.2Arcduration Figure 7.8 shows the arcing time versus the current Id for all successful interruptions. There is no obvious correlation between arc duration and DS current Id (up to the maximum tested current level Id = 27 A). The arc duration rangesfrom40msto90ms,comparablewithsimilarexperimentsreportedin[6] forId=7.3A,wherethearcdurationwasintherangeof40to70ms.Thereasonis relatedtotherapidcoolingofthearcpaththankstotheveryfastelongationofthe openinggap.Breakdownvoltagereductionduetoionizationisthereforebelieved tobeofminorimportanceintheentireinterruptionprocessbecauseoftherapid mixingofcoolairwiththepocketsofionizedair. 7.4Re‐strikevoltage The analysis of re‐strike voltage is based on the last three re‐strikes before the final interruption, since they are clearly observable in all cases. For each single experiment,thevaluesofthere‐strikevoltagesUrareplottedversusthecurrentId in Figure 7.9. The filled bars belong to failed interruptions and the open bars correspond to successful interruptions. The numbers on the right side of Figure7.9 indicate the experimental sequence. From the results summarized in Figure7.9thefollowingobservationsaremade: ‐ Inallexperimentsclear(atleastthree)observablere‐strikesoccur,although they can differ greatly in magnitude of the re‐strike voltage. The earlier re‐ strikes,indicatedwith‘1’,usuallyoccuratlowvoltagebecausetheairgapis still short. The re‐strikes can be qualified as thermally influenced with reducedre‐strikevoltageduetohotairleftbythearc.Ontheotherhand,the High‐velocityOpeningAuxiliaryInterrupter115 finalbreakdowns,indicatedwith‘3’,usuallyhavehighre‐strikevoltages(up to 2p.u.), since a large air gap is reached and the re‐strike is dielectrically dominated,whichwillbediscussedlaterinthissamesection. ‐ Considering only successful interruptions (open bars in Figure 7.9), the probability of occurrence of re‐strike voltage (Ur) higher than 1p.u. decreaseswithincreasingId.ForId<10A,mostofthethreeconsecutivere‐ strikes have values above 1p.u. However, only a single high Ur value is observed (in each test series) for Id > 12A. This suggests that for higher interrupted currents, the thermal influence on the interruption process becomes more prevailing, causing a lower re‐strike voltage. This phenomenon was also observed in previous experiments for a DS without auxiliary contacts during the entire interruption process [2], [5]. However, here it mainly occurs at the end phase of the interruption process. Apparently, despitetherapidly increasingairgap, theaccumulatedthermal Arc duration (ms) 90 80 70 60 50 40 7 12 17 Current (A) 22 27 Re−strike voltage (kV) Figure7.8.ArcdurationversusthecurrentIdat90kVsupplyvoltage. 200 100 0 200 100 0 200 100 0 200 100 0 200 100 0 200 100 0 3 1 2 1st 2nd 3rd 4th 5th 6th 5.4 8 9.5 10 12.5 15 17 20 23 25 27 Figure7.9.Re‐strikevoltageUrversustheinterruptedcurrentIdforsuccessivetests(atleast threeforeachcurrentlevel). Current (A) 116Chapter7 energyfromthepreviousarcre‐strikestillaffectsthenextone,especiallyat relativelyhighId(≥10A).Thissuggeststhattheinterruptioncapabilitycan be further improved with a higher opening velocity, since faster opening reducesthenumberofre‐strikesfurtherandhelpscoolingthearcremnants duringrecovery. Consideringonlyfailedinterruptions(filledbarsinFigure7.9),thelastthree ‐ consecutivevaluesofUralwaysexceed1p.u.atId=10,12.5A(evenreaching 2p.u., the maximum value according to [7]). However, only two re‐strike voltage values exceed 1p.u. for Id larger than 17A. This implies that at a lowerId(<12.5A), higher (consecutive) re‐strike voltages need to occur to cause a failed interruption. At larger Id (> 20A), only a single dielectrically dominatedre‐strikecanalreadyresultinaninterruptionfailure.Therefore, avoiding even a single dielectric re‐strike at larger interrupted current is neededfortheinterruptiontobesuccessful.Mostofthedielectricre‐strikes occur at the end of the interruption process, so it is important to interrupt thecircuitsuccessfullyalreadyquicklyaftercontactseparationbyrealizinga longgapinaveryshorttime. Themeasuredre‐strikevoltageversustheairgaplengthisplottedinFigure7.10. Theresultsareobtainedfromboththeexperimentalseriesat7A,100kVwitha totalof20operations,andtheexperimentssummarizedinTable7.1with9Ato 27A at 90kV utilizing the same DS with the same auxiliary device. Similar as in Chapter5,theplottedreferencebreakdownvoltageasafunctionofgaplengthis takenfromtheIEEEstandard[8].Itshowsthat,althoughthebreakdownvoltage from the experimental results increases with increasing gap length, they are considerablylowerthanthereferencevalues.Uptoadistanceofaround350mm, the reference electric field strength (0.5kV/mm) exceeds 10 times the experimentallyobtainedresults(upto0.05kV/mm).Foranairgaplengthabove 550mm, the reference electric field increases approximately linearly with increasingairgaplength.However,themeasuredvaluestendtorisemuchmore steeplyandaremuchclosertothereferencevalues.Probably,whilethedistance between the auxiliary contact tips becomes larger, the influence of the residual ionizationissmaller,causingthere‐strikevoltagetobeintheorderofhalfofthe referencevalues. Thevaluesofthere‐strikevoltagessuggestabreakupintotworegions:thermal re‐ignition breakdown region and dielectric re‐strike region, indicated in Figure 7.10byahorizontaldashedline.Inthedielectricre‐strikeregion,thevoltagesare larger than roughly 0.5p.u. (45‐50kV), and for the thermal breakdown region theyarebelowthisvalue. Figure 7.11 shows the relation between re‐strike/re‐ignition voltage, current Id, andairgaplength.Thedashedlineseparatesthethermallyandthedielectrically dominatedregions. High‐velocityOpeningAuxiliaryInterrupter117 Breakdown voltage (kV) 700 600 9-27A 7A Reference 500 400 Dielectric breakdown 300 200 100 Thermal 0 0 200 400 600 800 Air gap length (mm) 1000 1200 1400 Gap length (mm) Figure7.10.Breakdownvoltageversusairgaplength. 1400 1200 1000 800 600 400 200 0 >0.5 p.u. <0.5 p.u. Dielectrical breakdown Thermal breakdown 0 5 10 15 20 25 Current (A) Figure7.11.RelationbetweenbreakdownvoltageandcurrentId,airgaplength. Figure 7.11 suggests that the distinction between thermal and dielectric breakdown is related to the energy input per volume. Both higher current and smallergapdistancefavourthermalbreakdown. In Figure 5.22 with 0.7A, 130kV and 0.4A, 120kV, 99% of the observed breakdownvoltagelevelsarebelow40kVwithin400mmairgaplength.InFigure 6.10withexperimentsupto6.3A,thebreakdownvoltagesarelowerthan100kV (1.1p.u.) in the entire interruption process. In contrast to this situation with no fast opening device, most breakdown voltages values with the high‐velocity interrupterareintheregionofdielectricregions(seeFigure7.10),althoughthe numberofbreakdownsisrelativelysmall. 7.5Energyinputintothearc Theenergyinputintothearciscomparedforfailedandsuccessfuloperations,and itsdependencyonre‐strikevoltageandcurrentsisstudied. 118Chapter7 7.5.1Energyuponre‐strikeforfailedandsuccessfulinterruptions In order to compare the energy supplied to the arc upon re‐strike in failed and successfulinterruptions,thearcenergyassociatedwiththelastthreeconsecutive re‐strikes is calculated (see Figure 7.9). The method to calculate the arc energy has been described in Section 5.6. Taking Id = 12.5A as an example, the energy calculatedforthethreeexperimentsisshowninFigure7.12.Theenergiesw1,w2, w3 correspond to the 1st, 2nd, 3rd re‐strike in each test. Among these three interruption experiments, the 1st and 3rd succeeded, but the 2nd one failed. The energyinputintothearcatthe2ndexperimentismuchlargerthanfortheother two;inthe2ndexperimentthetotalenergyinputbythethreere‐strikesisalmost 4kJcomparedwithapproximately2kJand40Jatthe1standthe3rdexperiment, respectively. Thetotalarcenergysuppliedintothearcresultingfromthelastthreere‐strikes versusthecurrentIdisdepictedinFigure7.13.Thecumulatedarcenergyforthe failed interruptions clearly exceeds the values for successful interruptions. The energycorrelatesonlymarginallywithcurrentlevel.Thisobservationalignswith the observation in Section 7.3 on the weak relationship between current and likelihoodoffailureincaseofauxiliarycontactsprovidinginterruption. 7.5.2Energyversusre‐strikevoltage Theenergyinputtothearcisdeterminedbythe(transient)voltageandcurrent, thelatterdependingonthecurrentId(throughthecircuitparameters)andthere‐ strike voltage (refer to Section 5.6). The calculated energy is plotted versus re‐ strike voltage and current in Figure 7.14. The energy is calculated from the experimental series with Id from 5.3A to 27A summarized in Table 7.1. 2000 Energy (J) 1000 0 30 2000 w1 w2 w3 1st experiment 40 2nd experiment 50 60 70 80 90 50 60 70 80 90 50 60 70 80 90 1000 0 30 2000 40 3rd experiment 1000 0 30 40 time (ms) Figure7.12.Temporaldevelopmentoftheenergywinputintothearcuponre‐strikeforthe threeexperimentsatId=12.5A.Theverticalscalesaretakenequaltoindicatethewiderange inenergysupplybydifferentre‐strikes. High‐velocityOpeningAuxiliaryInterrupter119 8000 failure Energy (J) 6000 successful 4000 2000 0 4.5 9.5 14.5 Current (A) 19.5 24.5 Figure 7.13. Arc energy by last three re‐strikes versus current for failed‐ and successful interruptions. 6000 Energy (J) 5000 4000 3000 20-27 A 2000 12.5-17 A 1000 5.3-10 A 0 0 50 100 150 Breakdown voltage (kV) 200 250 700 Energy (J) 600 500 400 300 200 100 0 0 10 20 Breakdown voltage (kV) 30 40 Figure7.14.Arcenergyversusre‐strikevoltage(Ur)withIdasparameteranditsexpansion upto40kV. 120Chapter7 The current values are grouped in three ranges. With identical re‐strike voltage, however, it depends on the interrupted current. That may explain why the interruptionhasthetendencytobemorelikelytofailathigherId. 7.6Conclusionandrecommendation Compared to the interruption without high‐speed auxiliary contacts, the arc duration is significantly reduced and the number of re‐strikes during the entire interruptionislimitedtoonlyafewre‐strikesorre‐ignitions.Thearcisburning between the auxiliary contacts, which exhibits a "stiff" arc for successful interruption. Application of fast opening contacts does not reduce the transient overvoltageandcurrentamplitudes.However,thenumberofvoltagetransientsis muchsmaller. The energy, supplied to the arc upon re‐strike, is closely related to re‐strike voltageandcurrent.Thus,bothre‐strikevoltageandcurrentdeterminethefailure or success of interruption. This implies that reducing the re‐strike voltage and avoidingthebreakdownatlargerairgapcanenhancetheinterruptioncapability. The experiments show that the DS with fast opening auxiliary contacts is very effective in increasing the DS current interruption capability. It can interrupt currentsupto7Aat100kVsafelyand9Aat90kVinourexperimentsandthe arcingtimehasdroppedsignificantlytoonlyaseveraltensofmillisecondsinstead ofseveralseconds. Residual ionization has a major effect, since it lowers the breakdown voltage neededtore‐establishthearcandprolongsthearcduration.Fastopeningofthe gapmixescoldairwiththearcresidueandreducesthethermalenergydensityby diluting the ionized air. Also it cools down the air gap, which increases the gas molecule density and thereby reduces the mean free travelling path for free electrons. It is recommended that the contact opening speed must be as fast as possible withoutanybouncingandthatitismaintainedaslongaspossible.Thisreduces accumulation of heat in the air gap. The length of the straight arc before the arc starts an upward motion depends only on the air gap between the tips of the auxiliarycontacts.Thatmeansthehigherthecontactseparationspeed,thefaster thearcisforcedtoelongate,andthefasterthearcextinguishes,becauserecovery of the gap breakdown voltage towards the unconditioned "cold air" situation is fast. The moment that the whip releases, however, should be chosen such that dielectric re‐strike of the arc between the main blades is avoided. Before the auxiliary contacts release, the distance between main blades must meet the requirement in Table 5.2 in Chapter 5. Due to these various factors, it is High‐velocityOpeningAuxiliaryInterrupter121 recommended to test the specific auxiliary interrupter before it is put into practice. References [1] E. L. Luehring, J. P. Fitzgerald, "Switching the magnetizing current of large 345‐ kV transformers with double‐Break air switches", IEEE Trans. on Power Apparatus and Systems,vol.pas‐84,pp.902‐906,Oct.1965. [2] Y.Chai,P.A.A.F.Wouters,S.Kuivenhoven,P.vanderWal,M.Vanis,R.P.P.Smeets, "Current Interruption Phenomena in HV DSs with High‐Speed Opening Auxiliary Contacts",inProc.IEEEPESGeneralMeetingConf.,2010,pp.1‐8. [3] Y. Chai, P. A. A. F. Wouters, R. T. W. J. van Hoppe, R. P. P. Smeets, D. F. Peelo, "Experiments on capacitive current interruption with air‐break high voltage DSs", in Proc.Asia‐PacificPowerandEnergyEngineeringConf.,2009,p.1‐6. [4] D.F.Peelo,"Currentinterruptionusinghighvoltageair‐breakDSs",Ph.D.dissertation, Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven,2004. [5] Y.Chai,P.A.A.F.Wouters,R.T.W.J.vanHoppe,R.P.P.Smeets,D.F.Peelo,"Capacitive currentinterruptionwithair‐breakhighvoltageDSs",IEEETrans.PowerDelivery,vol. 25,pp.762‐769,Apr.2010. [6] KEMAtestreport411‐05,KEMA,Arnhem,theNetherlands,Dec.2005. [7] L.vanderSluis,TransientsinPowerSystems,NewYork:Wiley,2001,p.49. [8] IEEEStandardTechniquesforHighvoltageTesting,IEEEStandard4‐1995,Aug.1995. 122Chapter7 InterruptionwithInsertedResistors123 Chapter8 InterruptionwithInsertedResistors In addition to the methods discussed in the previous chapters, the option of insertingaresistorwillbediscussedinthischapter.Theresistivedeviceisusedto dissipatearcenergy.Theapproachwillbediscussedbasedonitsmechanismand applicability.Lessattentionispaidtopracticalrealizationandontheeconomicsof theconsideredsolutions. Insertion of a resistive component into the circuit aims to reduce switching overvoltage, e.g. a closing resistor often applied in a circuit breaker. Several comparable approaches are reported in [1]‐[9]. According to [1], an inserted resistorimprovesthedampingoftheoscillationsincapacitivecircuits.References [2]‐[4]focusontheapplicationofaresistorinserieswithDSsinGIS.References [5]‐[8]pointoutthatinsertedresistorsareabletoreducethedurationoftheDS switching arc. In [6], linear and nonlinear resistors are adopted to reduce switching surges. Both nonlinear and linear resistors are reported to have a similar beneficial effect on switching capacitive currents for gas‐blast switches (which are not employed at the present time) although without any details provided. Inthissection,theinterruptionperformanceofaDShavingaresistorinsertedin series with the main blades contacts is discussed first: 1) simulations show the potential effect on the interruption of parameters such as arc current, RV, high‐ frequency oscillation, and energy dissipation by the resistor; 2) laboratory experiments on the effect on overvoltage damping are presented. Next, interruptionperformanceofaDShavingaresistorinsertedinparalleltothemain bladescontactsisdiscussedaswell.Finally,someremarksaremadeonselection criteriaforresistiveelements. 8.1Seriesresistor AssumingthatinthecircuitofFigure8.1,aresistorRisinsertedinserieswiththe DS just before the DS starts opening. The parameters taken for simulations are: sourcevoltageus=Emsin (whereEm=√2/√3UlinewithUlinesystemvoltage inrms:rangingfrom72.5kVto550kV),Rs=50Ω,Ls=480mH,Cs=10nF.The initial current to be interrupted in absence of R is taken 10 A. The effect on interruptedcurrent,RV,transientandenergydissipationarestudied. 124Chapter8 Figure8.1.EquivalentcircuitforaDSwitharesistorinsertedinseries. 8.1.1EffectoncurrentId TheimpedancesofCsandClappliedinthetestcircuitsaremuchlargerthanthe sourceimpedanceofRsandLs.TheDScurrentIdreducesaccordingto: Id Uline 3 R2 1 (Cl )2 (8.1) Figure8.2showsthiscurrentasafunctionofRfordifferentsystemvoltagelevels. Thehorizontallineindicatesthevalueoftheresistortolimitthecurrentto1A. Since the value for the capacitance at DS load side is in the range of tens to hundreds of nanofarad, the current Id does not decrease significantly below a resistancevalueof10kΩ.Ontheotherhand,notmuchisgainedbyusingvalues exceeding 100kΩ. These values determine the resistance range if a reduction of thepower‐frequencycurrentduringswitchingisaimedfor.Obviously,theeffect ofinsertingaresistoralsodependsonthesystemvoltagelevel. 8.1.2Effectonrecoveryvoltage For a source voltage given by us Em sin( t ) , the RV and its maximum value RVmaxaregivenby(8.2)and(8.3),respectively. RV Em sin( t ) Em 1 1 R Cl 2 2 2 (8.2) RVmax Em Em 1 1 R Cl 2 2 2 (8.3) InterruptionwithInsertedResistors125 10 72.5kV(C =760nF) l Current (A) 8 system voltage increasing direction 6 145kV(C =380nF) l 245kV(Cl=225nF) 362kV(Cl=152nF) 4 550kV(C =100nF) l 2 0 0 50 100 150 200 250 Resistance (kΩ) 300 Figure8.2.InterruptioncurrentIdversusinsertedresistancevalue. 350 400 With a fixed voltage Em, the maximum RVmax across the DS decreases with increasingresistancebecausepartofthevoltagedropsacrossR.Comparedtothe situationwithoutinsertedresistor,theRVmaxishalvedwhen R Cl 1 ,because notrappedchargeisaccumulatedinCl.InthesimulationofFigure8.3,fivevalues of R (0.1kΩ, 1kΩ, 10kΩ, 100kΩ, 1000kΩ) are considered, showing the wave shapesofRVatsystemvoltagesof72.5kVand550kV.TheRVamplitudereduces withincreasingR.At72.5kVtheeffectonRVbecomessignificantasfrom10kΩ and for 550kV as from 100kΩ. Therefore, the inserted resistor should have a valuebetweenafewtensofkilo‐ohmsandafewhundredofkilo‐ohms,depending onthevoltagelevel. 8.1.3Effectontransientoscillationuponre‐strikes A resistor may have a significant effect on high and medium‐frequency components. The effectiveness of its value depends on the inductance and capacitance in the high and/or medium‐frequency loop. For the damping high‐ frequencycomponent,theresistanceRshouldexceed 2 LH ,whereLHandCHare CH theequivalentinductanceandcapacitanceoftheHFloop,indicatedinFigure3.1. SinceLHisintheorderoftensofmicrohenryrangeandCHistheseriesvalueofCs and Cl, a value typically of a few hundred ohms is sufficient to avoid the high‐ frequency oscillation. To avoid medium‐frequency oscillations, typical values as from a few kilo‐ohms are needed since the source side inductance in the circuit usuallyistensofmilohenriesandtheequivalentcapacitanceistheparallelvalue ofCsandCl. Figure 8.4 shows simulations of oscillations for various values of the inserted resistor. The simulation parameters are: Em = 300√2kV, Id =2A, Cs = 18.4nF, Cl=36.8nF,LH=20µH,Ls=13.8mH,Rs=1kΩ,thevaluesofRaretakenas0.01kΩ, 126Chapter8 150 1000 RV (kV) 100 50 RV (kV) 0.1kΩ 1kΩ 10kΩ 100kΩ 1000kΩ 0 0.1kΩ 1kΩ 10kΩ 100kΩ 1000kΩ 500 0 −50 −100 0 20 time (ms) 30 40 −500 0 5 10 time (ms) 15 20 Current (A) Figure8.3.RVwaveshapeforvariousresistorvalues(left)at72.5kVand(right)at550kV. 1000Ω 400 100Ω 200 10Ω 0 −200 5 5.1 5.2 time (ms) Figure8.4.Waveshapesofidfordifferentresistancevaluesshowingitseffectondampingof thehighandmedium‐frequencycomponents. 0.1kΩ, 1kΩ. ForR = 0.01kΩ oscillations are observed in the medium and high‐ frequency range; atR = 0.1kΩ the high‐frequency component has disappeared, butthemedium‐frequencycomponentstillremains;atR=1kΩalloscillationsare damped. 8.1.4Energydissipationthroughresistorduringinterruption TheenergydissipatedintheinsertedresistorbytheDScurrentisobtainedfrom 1 ( U line ) 2 3 wR 2 R t (8.4) R (1 Cl ) 2 Therelationshipbetweenthedissipatedenergy(over1second)andtheresistor value is shown in Figure 8.5. The energy dissipation, i.e. the energy the resistor shouldbecapabletowithstand,hasamaximumwhentheimpedanceofClequals R,i.e.whenR=1/(ωCl). Takinganexample,supposingtheinitialcurrentis10Aat145kVsystemvoltage (Cl=380nF), R is 8.4kΩ. With the value of R = 8.4kΩ, the current reduces from InterruptionwithInsertedResistors127 2000 72.5kV 145kV Energy (kJ) 1500 245kV 550kV 1000 500 0 0 20 40 60 Resistance (kΩ) 80 100 Figure8.5.EnergydissipationversusinsertedresistorvalueatCllistedintheFigure8.2. 10Ato7A, RVreducesfrom2Emto(1+1/√2)Em.Obviouslyitisnotpossibleto interrupt successfully when Id=7A at 145kV. Similar results may be obtained fromothersystemvoltagelevels.Therefore,aresistorisrecommendwithavalue largerthan1/(ωCl),butrequireddissipationenergylowerthemaximumvalueof energy in Figure 8.5. In such a case, the supply voltage will mainly drops across theresistor. It should also be mentioned that insertion of a high‐ohmic current limiting resistorjustbeforeDSopeningcreatesdifficultiesthatare,infact,comparableto those associated with the opening of the DS itself. Therefore,how to insert such high resistor into the circuit is motivation for further work. On the other hand, resistor values suitable to damp the switching oscillations have much smaller valueandlimiteddissipation.Insertionoflow‐ohmicresistorsisthereforeeasier toachieve. 8.1.5Experimentalresults A series of experiments to verify the effect of an inserted series resistor on oscillation damping is carried out at the TU/e HV laboratory. The chosen parametersare:Em=100√2kV,Cs=4nF,Cl=32nF,Id=1A,LH=17μH.Figure8.6 presents the wave shapes of the current id and their expansions around a single re‐strike for R=0Ω, 200Ω and 1kΩ. The oscillations observed with inserted resistor disappear for 200Ω, which can be expected since the corresponding quality factor of the HF circuit is 0.4. The amplitude of the transient currents is reduceddrastically,uptoafactor10,forR=1kΩ. The optical observations of the arc are presented in Figure 8.7, showing the cumulatedarcbrightness(seeChapter5)withandwithoutseriesresistor."Stars" indicatethearcbrightnesswithR=200Ωand"Dots"showtheresultwithR=0Ω. Withoutdampingresistorahigherbrightnesstendstooccur,especiallyattheend of the interruption. The damping resistor prevents severe re‐strikes, not only fromtheperspectiveofarcbrightness,butalsofromtheaudiblenoiseassociated 128Chapter8 500 500 0 0 R=0 −500 R=0 −500 0 100 200 300 400 500 550.838 550.840 550.842 550.844 550.846 550.848 400 600 Current (A) Current (A) 500 0 R=200Ω −500 0 500 100 200 300 400 500 300 200 100 R=200Ω 0 566.350 100 600 566.360 566.370 566.380 566.390 50 0 0 R=100kΩ −500 0 100 200 300 400 R=100kΩ −50 500 600 579.105 579.115 time (ms) 579.125 579.135 time (ms) Figure 8.6. (left) Wave shapes of current id and (right) their final re‐strikes zoomed in for differentresistorvalues. 6 Brightness 4 x 10 2 R=0 0 −2 Brightness 0 6 x 10 4 100 200 time (ms) 300 400 500 2 0 R=200Ω −2 300 320 340 360 380 400 420 time (ms) 440 Figure8.7Arcbrightnessagainsttimewithandwithoutseriesresistor. 460 480 with these re‐strikes. However, the effect on arc duration could not be analyzed duetothegradualvoltagedecay. 8.2Parallelresistor Figure 8.8 shows the circuit with a resistor including an additional switch, indicatedbyD1,inparalleltotheDSmainblades.Inclosedposition,theresistor InterruptionwithInsertedResistors129 Figure8.8.Thecircuitforaparallelinsertedresistor. branchisshort‐circuitedbythemainblades.WhentheDSstartstoopen,anarc arisesbetweenitscontacts.However,oncethearcextinguishesatthenextcurrent zero, the voltage across R that equals the voltage across the air gap, determines thepossibilityofsubsequentre‐strikes.IncasethevoltageacrosstheresistorRis lowerthanthemomentarybreakdownvoltageoftheairgap,therewillbenore‐ strikesbetweenthecontactsofthemainblades.ThecurrentwillflowthroughR anditsseriesswitchD1.Theresistorhastobedisconnectedfromthecircuitafter commutation,asituationsimilartothecircuitofFigure8.1.Puttingaresistorin paralleltotheDSinitiallymaybeapracticalwaytoinsertaresistor,whichafter current commutation becomes a series component. For this series resistor, a relatively large value is needed for an appreciable reduction in interruption current. For preventing re‐strikes between the main contacts, a lower value is required. These opposing requirements prevent this method to be effective for practicalapplication. It is also possible to put a nonlinear resistor, such as MOV, a component with a nonlinearcurrent–voltagecharacteristic,inparalleltothedisconnector.Obviously the current to be interrupted is similar to the situation without the nonlinear resistor.Figure8.9showsthesimulationresultsofthevoltageswaveshapesofucs, ucl and the voltage across the disconnector. The simulation parameters are us = 120sin(ωt+θ), Cs = 10 nF, Cl = 380 nF. The voltage at 1mA for the nonlinear resistoris70kV.TheresultsgivethebasicideathattheMOVcouldfixthevoltage acrossthedisconnector,whichreducetheRV.Asaconsequencethechancetobe restrikeisless.However,furtherinvestigationisrecommendedasfuturework. 130Chapter8 Voltage (kV) 200 100 0 −100 u u cl −200 0 5 10 cs u d 15 20 25 time (ms) Figure8.9.Simulatedresultsofthewaveshapesofucs,uclandtheirdifference. 30 8.3Discussion An inserted resistor can theoretically facilitate current interruption by reducing the current and lowering the RV (relatively high values needed), or by suppressing transient currents during DS re‐strike (lower values of resistances). ThehighvalueoftheresistorneededforlimitingthecurrentandRVcausesheat dissipationissuesaswell ascommutationproblems andthereforeisin question forpracticalapplication.Thedissipationrequirementstosuppresstransientsare mucheasiertocomplywith,buthavenotproventobeeffectiveinshorteningarc duration. An inserted resistor in parallel to the main contacts of the DS may improvethecurrentinterruptioncapabilitybylimitingtheRVacrosstheDS,but only relatively low values can be applied for successful commutation. Nonlinear resistorisalsoapotentialchoice. Resistor type and value strongly depend on the current the DS is intended to interrupt.Aresistorwithavaluesufficienttoreducetheinterruptioncurrentand the RV should be capable to dissipate energy in the order of tens of kilojoule during the time the current flows through the resistive element. This is in the order of a few seconds. The allowed temperature rise during that time is an importantdesigncriterion,whichdeterminesthesizeandmaterialoftheresistor. Considering the high‐frequency phenomenon, a resistor with low inductance is preferred which at the same time is capable to dissipate sufficient energy. Resistorsshouldhavehighheatcapacityandlowthermalexpansioncoefficientto limit the thermal and mechanical stresses on the component. Ceramic resistors areapreferredoption,buttheyareratherexpensive. References [1] R.C.VanSickle,"Influenceofresistanceonswitchingtransients",AIEETrans.,vol.58. pp.397‐404,Aug.1939. [2] Y.Hiroshi,M.Tetsuo,"Disconnectingswitch",U.S.Patent5225642,Jul.1993. [3] H.Yamamoto,J.Hirata,"Disconnectingswitch",U.S.Patent5091614,Feb.1992. InterruptionwithInsertedResistors131 [4] Y.Yamagata,K.Tanaka,S.Nishiwaki,"suppressionofVFTin1100kVGISbyadopting resistor‐fitted disconnector", IEEETrans.onPowerDelivery, vol. 11, pp.872‐880, Apr. 1996. [5] M. S. Savic, "Suppression of the high‐frequency disturbances in low‐voltage circuits caused by disconnector operation in high voltage open‐air substations", IEE Proceedings,vol.133,pp.293‐297,Jul.1986. [6] A. Foti, J. M. Lakas, "EHV switch tests and switching surges," IEEE Trans. on Power ApparatusandSystems,vol.83,pp.266‐271,Mar.1964. [7] D. F. Peelo, "Current interruption using high voltage air‐break disconnectors", Ph.D. dissertation,Dept.ElectricalEngineering,EindhovenUniv.ofTechnology,Eindhoven, 2004. [8] R. P. O’leary, R. H. Harner, "Evaluation of methods for controlling the overvoltages producedbytheen‐generationofashuntcapacitorbank",inProc.LargeHighVoltage ElectricSystemsinternationalConf.,Aug.1988. [9] J. Tao, S. Chen, C. Yang, "Simulation of disconnect switch operation of air insulated Substation",InProc.theXIVinternationalSymposiumonHighVoltageEngineering,Aug. 2005. 132Chapter8 Pre‐breakdownPhenomenainDisconnectorInterruption133 Chapter9 Pre‐breakdownPhenomenainDisconnectorInterruption SmallcurrentinterruptionwithDSsisasuccessionofarcingandbreakdown(re‐ strike/re‐ignition). The physics behind the interruption process occurring in switchingwithanair‐breakDSisstronglyrelatedtorepetitivebreakdowns,thus tothebreakdownmechanismsinairgapswithpresenceofpocketsofhot,ionized air.TheACrecoveryvoltageduringthebladesopeningresultsinavaryingelectric field over a gradually increasing gap length. This chapter gives a qualitative discussiononthe breakdownprocesses,duringtheconditionspresent inthe air gap.Thebasicbreakdownmechanism,anditsapplicationtoanon‐uniformfield, will be described first. Next, experimental evidence will be presented related to currentinterruption, mainlyfocusingonthepre‐breakdowncurrent.Finally, the implications for understanding of the current interruption behaviour with air‐ breakDSswillbediscussed. 9.1Classicalmodellingofbreakdown Atambienttemperatureandpressure,airisanexcellentinsulator.Itiscommonly used in power system components such as overhead lines, switch yards, etc. Electric breakdown in air occurs when the dielectric strength of the air is exceeded;itisanevolutionfromnon‐conductivetoconductivestatus. Gasbreakdownmechanismshavebeeninvestigatedovermorethanacenturyand most phenomena can basically be understood in terms of the Townsend and streamerdischargemodel[1]. 9.1.1Townsenddischarge The current through the air gap between two parallel plate electrodes as a functionoftheappliedDCvoltagewasinvestigatedbyTownsendintheearly20th century. It was found that initially the current increased proportional with the applied voltage and then remained approximately constant at a value i0. As the applied voltage was increased further, the current started increasing exponentially until complete breakdown of the gap. The general pattern of the current‐voltagerelationshipissketchedinFigure9.1[1].Uptothevoltageu1the currentincreases,becauseofanincreasingdriftvelocityofthechargedparticles insidethegapundertheinfluenceoftheappliedelectricfield,untilabackground current (saturation current) i0 is reached. The rate of continuously created free electrons and ions by e.g. natural radioactive decay, cosmic and UV radiation 134Chapter9 limitsthecurrent.Theincreaseinthecurrentpastu2uptothebreakdownatu3is ascribedtoionizationoftheinsulatinggasmoleculesbyelectronimpact. Withanelectricfieldappliedacrossthegap,aninitialfreeelectronwillaccelerate and gain kinetic energy. Upon impact with a gas molecule, if the kinetic energy exceeds the ionization energy of the molecule, there is a probability that an electron is liberated. The energy gained between two successive collisions is proportional to the electric field and the travelled distance between consecutive collisions,whichisdenotedasthefreetravellingpath.Itsmeanvalueisinversely proportional to the concentration of gas particles, i.e. the gas density. This free electron generating process is called "Electron Impact Ionization" [2]. Both electrons will accelerate and may gain sufficient energy for further ionizing collisions,resultinginanavalancheoffreeelectrons.Thisprocessisquantifiedby the first Townsend coefficient, representing the average number of electrons producedbyoneelectronperunitlengthinthedirectionoftheelectricfield. Acompletebreakdownrequiressecondaryprocesses.IntheTownsenddischarge model, the positive ions left behind drift to the cathode. If sufficient energy is releasedupontheirimpactandneutralizationatthecathode,thereisaprobability that a secondary electron is released from the cathode. Alternatively, photons emittedbyneutralgasparticles,whichareexcitedupontheelectronimpact,can account as well for secondary electrons. The second Townsend ionization coefficient gives the probability that an ion or photon results in a secondary electron. The Townsend criterion for complete gap breakdown states that, if on averageperinitialfreeelectronmorethanonesecondaryelectronisliberated,the gapionizationgrowstoafullbreakdown. Paschenstudiedexperimentallythebreakdownvoltageofagasinahomogeneous field as a function of gas pressure and gap spacing in 1889. His findings are nowadaysreferredtoasthewell‐knownPaschenlaw,statingthatthebreakdown voltageUbisafunctionoftheproductgaspressurepandgaplengthd:Ub=f(pd). Figure 9.2 depicts the Paschen curve for air at a temperature of 20℃. The minimumofthiscurvecanbeunderstoodintermsoftheTownsendmodel.Atthe leftsideoftheminimumeitherthepressureorthegaplengthhasbecomesolow, that few ionizing collisions occur during the crossing of the gap. The electron avalanche is too small to have sufficient probability of creating secondary electronstocontinuethebreakdownprocess.Attherightside,theenergygained by free electrons is low, either due to a reduced mean free travelling path at higher pressure (i.e. higher density) or low electric field because of large gap length.Thus,alsotheelectronimpactionizationisreduced. The pressure of an ideal gas in equilibrium is described by its state equation p=NTk_B,(p:gaspressure,N:gasdensity,k_B:Boltzmannconstant;T:absolute temperature). This relation shows that pressure is affected by the density or temperature of the gas. Therefore sometimes the Paschen law is presented such Pre‐breakdownPhenomenainDisconnectorInterruption135 Figure9.1.Relationshipbetweenthecurrentandthevoltageinpre‐breakdownregion. Figure9.2.Paschencurveforairinlog‐logscale,copiedfrom[2]. that the breakdown voltage is a function of the product of gas density and gap length. AlthoughthePaschenlawwasderivedfromexperimentsonuniformelectricfield, itdoesnotreallyapplyforthenon‐uniformfieldbetweentheDSmainbladeswith concentratedfieldatthetips.However,theimpactionizationbyacceleratedfree electronsinfluencedbytheelectricfieldstrengthisalsohereconsideredasamain mechanism. Deviations from predictions based on the Townsend model were observed for large air gaps with inhomogeneous fields or at high pressure. It also fails to explain the electron transit time in practice often being shorter than can be expectedfromtheavalanchemodel. 9.1.2Streamerbreakdown The streamer breakdown mechanism is an alternative approach, also based on primaryelectronavalanches.ItwasproposedbyRaether,Meek,andLoeb,based 136Chapter9 Figure9.3.Fielddistortioninagapcausedbyspacechargeofanelectronavalanche(copied from[5]),E0istheappliedelectricfield,E(x)istheelectricfieldafterdistortion. on numerous experiments [3], [4]. In the streamer model the effect of the space chargefieldbytheavalancheelectronsandphotonionizationinthegasvolumeis takenintoaccount.Astheelectronsmovetotheanode,thepositiveionsareleft behindinarelativelyslow–movingtail.Theavalanchehasaccumulated,especially intheheadoftheavalanche,negativespacecharge.Thetailcontainspositiveions. These space charges distort the applied electric field distribution, i.e. enhancing the electric field in front of the head and behind the tail of the avalanche; in between the electric field is reduced as shown in Figure 9.3. Electrons will gain moreenergybecauseoftheenhancedfield,causingincreasedionizationandalso more energetic (UV) electrons from highly excited gas atoms. These UV‐photons startnewavalanchesfurtheronbyphoto‐ionization.Thissignificantlyaccelerates the breakdown development. A streamer is a thin channel formed from the primaryavalanche. When streamersbridgethegapwithplasma, thebreakdown occurs. The streamer mechanism is specifically important in an inhomogeneous field,whereinregionswithfieldenhancementalsorelativelargelocalionization occurs,causinghighspacechargefields. In a non‐uniform gap, there is a divergent field. The electric stress near one or bothelectrodeswithhighcurvatureisenhancedwithrespecttotheremainderof the gap [5]. Consequently the ionization is enhanced. Partial ionization starts in thegaswithhighestelectricfieldintensity,butdoesnotdirectlycausecomplete breakdown. At higher applied voltage the field strength over the complete gap reachavaluesufficienttoinitiatecompletebreakdown. Pre‐breakdownPhenomenainDisconnectorInterruption137 9.2Pre‐breakdowncurrentindisconnectorinterruption The experimental data presented in Chapter 5 showed that each interruption involvesnumerousrepetitivearcextinctionsandre‐ignitions.Forthestudyofthe pre‐breakdownprocessofcurrentinterruptionwithaDS,thefocusisonthearc current after arc extinction at the power‐frequency current zero until the succeedingbreakdown. Typical wave shapes of the current id through the DS at Id of 2.6A and its expansions at different stages of the interruption process are plotted in Figures 9.4and9.5,respectively.Fourexamples,eachindicatedbyacircle,areextracted from the full current waveform shown in Figure 9.5. Duration Δt1 denotes the periodaftercurrentzerocrossing,themeasuredcurrentremainszero(belowthe measurementthresholdof0.5mA).Afterthisperiod,asmall"step"inthecurrent is observed (see Figures 9.5b and 9.5c). The period Δt2 denotes the duration between the moment of starting the "step" until the moment of complete breakdown. Regarding the intervals Δt1 and Δt2 the following observations are made: ‐ The current zero duration Δt1 decreases gradually with time, i.e. with increasinggapdistance(compareFigures9.5a,b,candd).Inthefinalstage oftheinterruption,thecurrentzeroperiodisevenhardlydiscernible(Figure 9.5d). ‐ Atthebeginningoftheinterruption,afterΔt1thebreakdownoccurswithout anobservablepre‐breakdowncurrent(Figure9.5a).Inthelaterstageofthe interruptionprocess,thereisa"step"inthecurrent,followedbyagradually risingcurrent(Figure9.5b,c).Intheendstageoftheinterruption,the"step" is also not clearly visible, but instead directly after interruption a pre‐ breakdowncurrentarises(Figure9.5d).AstheRVincreasesfurther,thegap breaksdowncompletely. ‐ TheperiodΔt2changesalsowithincreasingairgaplength.Inearlystageof interruption, Δt2 is not noticeable; in the end stage, it nearly occupies the entireperiodbetweencurrentzerocrossingandbreakdown. Manyfactorsareinvolveddeterminingthedevelopmentoftheentirebreakdown, suchastheelectricfield,residualchargesandtemperatureinsidethegap(degree ofionizationoftheair).Theelectricfieldinbetweenthemainblades,resembling a rod‐rod electrode configuration, is inhomogeneous. Further fragments of residualplasmaareleftinthehotairgapfrompreviousarcing.Thisisexemplified inFigure5.18showinglightemissionfromhotspotsinthearcpathaftercurrent interruption. Theobserved"steps"inthecurrentoccurringafterafewcycles(about10cycles atIdof2.6A)aftertheoperationstartscanbeunderstoodintermsoftheelectric andthermodynamicconditionofthearcpath.Oncethecurrentcrosseszero,the 138Chapter9 residualplasmagraduallyvanishesasthegapiscoolingdown.Atthesametime theRVstartsrising.Theconductivityoftheplasmaofthepreviousarcpathistoo weak for a detectable current, resulting in a measured current zero stage (Δt1). With increasing RV, the electric fields at the blade contacts (the arc roots) are enhancedbythenon‐uniformfield.NeartheDSbladesrootsthesituationismost favourable to initiate new ionization, apparently via a sudden onset of a current resulting in a step‐wise increase. The current path will follow the previous arc routebecauseofthestillpresentchargecarriersandtheenhancedtemperature. TheelectricfieldincreaseswiththefurtherRV.Inaddition,thedissipatedenergy by the current heats the channel. This leads to re‐ionization until complete breakdownoccursaftertimeΔt2. Δt1 decreases with the gap length means the "step" is initiated earlier with the sameRVevenwithlongerarcroute.Thereasoncanhighlyprobablybeattributed to the remaining partly ionized hot air. There is more accumulative residual charge(ionizedhotair)afterlongertimebecausethehotairneedslongduration fordiffusion.InFigure9.6thedurationΔt2isplottedversusthearcingtimeduring which the interruption develops. Measurements at three interruption current levels(Idof1.4A,2.6A,5A)arecompared.ThetimeΔt2islowerforlargercurrent Idandsmallerarcingtime(smallergaplength).Itmaybeexplained: Δt2 is larger for larger gap length. Longer arc length at larger gap length, ‐ resulting in lower air density. In such a case, it required longer duration to accomplishfullbreakdown. ‐ Withhighercurrent,thepathleftbyapreviousarchasahighertemperature. As the gas temperature is inversely proportional to the gas density at constant pressure (assuming the ideal gas law still applies), higher temperaturemeanslowerdensity,resultinginlongerfreetravellingpathfor electrons, stronger ionization, and consequently lower breakdown voltage. Lower breakdown voltages correspond to shorter time Δt2 for the RV to reachthebreakdownvalue. Theinfluenceoftemperatureonthebreakdownvoltagecanroughlybeestimated usingthefactor293/T(K)correctingthedensityfortemperature[6].FromFigure 5.23 it was concluded that the actual observed breakdown voltage was about a factor 6 lower than expected for unconditioned air at the same gap distance according to IEEE Standard. If this difference could be exclusively ascribed to temperature a value of 1500°C is obtained. This temperature is not arc temperature,butthechanneltemperatureatthebreakdown. 9.3Discussion The pre‐breakdown process of the small current interruption with a DS is qualitativelyanalyzedalongtheclassicalelectronavalanchebreakdowntheory.A pre‐breakdowncurrentofafewamperescanbeobservedwithdurationofseveral Pre‐breakdownPhenomenainDisconnectorInterruption139 Current (A) 20 10 0 −10 (a) −20 (c) (b) 0 500 (d) 1000 time (ms) 1500 2000 (a) 1 Current (A) Current (A) Figure9.4.WaveshapeofthecurrentthroughtheDSatIdof2.6A;circlesindicatethe instanceswhereexpansionsaretakenforFigure9.5. 0.5 0 −0.5 50 1 0.5 0 52 time (ms) complete breakdown current zero "step" Δt2 1 0 1330.5 1331 1331.5 1332 1332.5 1333 1333.5 time (ms) (b) Δt (c) 0.5 53 Current (A) Current (A) 1.5 51 1 0.4 (d) 0.2 0 −0.2 −0.5 830.5 831 831.5 time (ms) 832 832.5 2070 2071 2072 2073 time (ms) 2074 Figure9.5.Pre‐breakdowncurrentatdifferentmomentsofthecurrentinterruptionprocess throughtheDSatIdof2.6A,seeFigure9.4. Duration (ms) 3 1.4A 2.6A 2 1 0 5A 0 500 1000 1500 time (ms) 2000 2500 Figure9.6.BreakdowndevelopmentdurationΔt2againsttimeforIdof1.4A,2.6Aand5A. hundredsofmicrosecondspriortobreakdown.ThiscurrentisinitiatedbytheE‐ field at the arc root, but drawn through the still weakly conducting former arc channel, and supplies energy to it. If this supply of energy exceeds the loss of energy mainly by natural (or forced) convection cooling, the former arc channel 140Chapter9 will become highly conductive again and breakdown follows. Practical examples offorcedconvectionareairflowandwhip,draggingthearcfastintocoolambient air. For instance, there is no “steps” phenomenon observed for the entire interruptionwithairflowbecauseresidualchargesareforcedtoleavetheairgap by the air flow. Instead the complete breakdown occurs suddenly (see Figure 9.4a).Thepre‐breakdowndischargedevelopsfasterwithincreasingcurrentId(or shorterdistance)intobreakdown.Thisisrelatedtohighertemperatureinthearc path giving rise to a larger mean free travelling path for liberated electrons and the higher residual ionization. The pre‐conditioned path from an earlier breakdownisthereforeapreferredpathforsubsequentre‐strikes.Thismodelis consistent with the observed beneficial effect of an applied air flow through coolingthebreakdownchannelandremovingionizedgasremnants. References [1] J.S.Townsend,Electricityingases,OxfordUniversityPress,p.2,1915. [2] E. Kuffel, W. S. Zaengl, J. Kuffel, High Voltage Engineering Fundamentals, Second Edition,Newnes,Aug.2000. [3] H.Raether,Electronavalanchesandbreakdowningases,S.l.:Butterworth,1964. [4] J. M. Meek, J. D. Craggs, ElectricalBreakdownofGases, John Wiley & Sons, Chichester, 1978. [5] G.J.J.Winands,Efficientstreamerplasmageneration,Dissertation,EindhovenUniv.of Technology,Eindhoven,2007. [6] C. L. Wadhwa, HighVoltageEngineering, 2nd edition, New age international limited publisher,2007. ConclusionsandRecommendationsforFutureResearch141 Chapter10 ConclusionsandRecommendationsforFutureResearch Limited literature sources recognize that disconnector requirements should include the capability of interrupting capacitive currents up to a certain current level. For safety and economical reasons, it is desirable to find approaches to improve the current interrupting capability. For this, it is indispensable to understand the related interruption phenomena and its influencing factors profoundly.Thisisthemainobjectiveofthisthesis. Capacitive current interruption by a disconnector involves the interaction between the arc and the circuit. The transient phenomena caused by this interaction, cover different frequency ranges. Transient voltages and currents, that may have an impact on neighbouring equipment, arise during the interruption. Avarietyofexperimentalsetupsweredesignedandinstalledinthisstudy.They includethetestdisconnector,high‐resolutionandwide‐bandopticalandelectrical measurement and data acquisition. The high voltage sources used a resonant generator for detailed observation of the arcing process as well as high voltage power transformers in two different power laboratories to make quantitative analysis during the entire interruption process. Aided by computer modelling, experimentaldataobtainedfromthreedifferenttest‐siteshavebeenanalysedin order to better understand the interruption in terms of repetitive arcing and breakdown phenomena. This understanding can be exploited to improve the interruptioncapabilityofadisconnectorwiththeassistancefromauxiliarymeans suchasairflow,high‐speedcontactseparationanddampingresistors. Themainconclusionsofthisthesisandrecommendationsforfutureworkwillbe addressedinthischapter. 10.1Conclusions Recordedelectricalandopticalsignalsshowthatinthefewsecondsofcapacitive currentinterruption,thearchasfrequentextinctionsandre‐ignitions/re‐strikes. It extinguishes temporarily at the (power‐frequency) current zero crossing, but re‐ignites due to the recovery voltage impressed by the circuit. This process comes to a halt when the dielectric stress of the air gap exceeds the recovery voltagevalue.Duringthisprocess,thebreakdownvoltageoftheairgapalongthe arcpathissignificantlyreducedduetothethermalinfluencescausedbythearc, 142Chapter10 and is lower than the direct shortest distance connection between the disconnector blades. For slow opening main blades (as usual in a standard disconnector), it may take seconds to reach sufficient gap distance to withstand thehighestrecoveryvoltage.Theassociatedtransientovervoltagehasreacheda value up to 2.5p.u. in the experiments, and could theoretically reach 3p.u., with therateofriseintheorderoftenthsperunitpermicrosecond.Currenttransients with peak values up to a few kiloamperes were observed, rising with a few kiloamperespermicrosecond. Theobservedarcbehaviour,transients,re‐strikevoltageandsoforth,arehighly affected by the circuit parameters. These parameters involve the capacitances Cs andClatbothsidesofthedisconnector,strayandshort‐circuitinductance,andthe powersupplyvoltage.Thesefactorsallaffectthecurrenttobeinterrupted,there‐ strike voltage with associated transients as well as the energy supplied into the arc,whichultimatelyheatstheairinthevicinityofthearc. Thesourceandloadsidecapacitancesarethemaincircuitfactorsthatinfluence theinterruptionperformance.WitharatioCs/Cl,smallerthan0.1inthepresented experiments, the arc duration, overvoltage in the circuit, contact gap length and bladeangleatfinalinterruption,andenergyinputintothearcareincreasing.Slow dielectrically recovery of the (former) arc trajectory against circuit impressed recoveryvoltageisultimatelythecauseforlongarcduration. Thecurrenttobeinterruptedisanothergoverningfactor.Withhighercurrent,the arc duration, the magnitude of overvoltage and the energy dissipation upon re‐ strike increase, resulting in larger blade contact gap spacing when final interruption is achieved. With increasing current, the gap recovery becomes slowerandthere‐strikevoltagedecreases,untilacurrentlevelisreachedbeyond whichinterruptionisnolongerpossibleandthearcre‐ignitespurelythermally.In that mode, the arc will generally no longer extinguish, irrespective of the gap space. It has been observed that at increasing current the overall arc brightness increases, the decay of arc remnant light emission is slower, and the post arc energyinputduringrecoveryislarger. With decreasing re‐strike voltage, the energy input into the arc on re‐strike reduces. The arc energy, overvoltages and transient current in the circuit reach theirmaximum(duringtheinterruptionprocess)justbeforethearcextinguishes completely. Therefore interrupting the circuit successfully at an early stage at a relativelyshortgapimpliesalowerre‐strikevoltage,lowertransients,andlower energyinputintothearc. It is observed that the re‐ignited arc follows the "traces" left by the just extinguished arc. The arc prefers the dielectrically weakest path for which the breakdown voltage is lowest. The ionization left behind from the arc in the previouspower‐frequencycyclemakesthesubsequentre‐ignitionorre‐striketo ConclusionsandRecommendationsforFutureResearch143 occuratareducedvoltage,comparedtotheactualcold‐aircontactdistance.Apre‐ breakdown current, drawn through the still weakly conducting former arc channel, is supplied by the rising recovery voltage. It supplies energy to the channel, creating the conditions for pre‐breakdown until complete breakdown. Upon re‐strike, there is energy input from the transient components heating the arcfurther.Bythisthesubsequentlevelofvoltageatwhichre‐strike/re‐ignition willoccurisaffected. With the purpose of removing the residual charges from previous arc path, thus avoiding space charge to accumulate at all, two approaches are investigated in detail experimentally: removal of residual plasma by air flow and drawing the residual plasma into cooler air flow by a high‐speed auxiliary interrupter. Two interruptionmodesareobserved,indicatedasadielectricandathermalmode. Thedielectricmodemainlyoccursinthecasewithahighairflowrate(directedto thearcfootpoints,inthetestsupto315litre/minute)andwithlowarccurrent (roughlybelow1A).Inthismode,thermaleffectsplayaverylimitedrole.During the dielectric mode, the arc current may be chopped at any phase angle of the power‐frequencycycle.Multiplebreakdowneventswithinahalfpower‐frequency cycle are common and the number of the breakdown events per unit time is significantlyhigherthaninthethermalmodeinterruption.Atthesamelevelofthe interrupted current, the breakdown voltage (mainly dominated by the air gap length),ishigherthanforthethermalinterruptionmode.Bothhigherbreakdown repetitionfrequencyandhighervalueofthebreakdownvoltagemakesthismode potentially more hazardous for neighbouring equipment due to the frequent transients. The interrupted current derivative (di/dt) at current zero is much higher (about a factor 20) than in the thermal mode. This implies that the arc current can also interrupt at current zero by high‐frequency oscillations, caused bythetransientsinthecircuit. The thermal mode occurs at higher interrupted current (>1.4A in the present experiments) without air flow. During thermal interruption, the arc in principle extinguishes at every power‐frequency current zero. The interrupted current derivative increases roughly linear with current. Due to the residual ionization frompreviousarcs, the breakdownvoltageismuch lower ("thermally reduced") than in the dielectric interruption mode. In principle, the voltage transients are lesshazardoustonearbyequipment.However,eventuallytheovervoltagereaches averyhighlevel,butatamuchlargergapdistancethaninthedielectricmode.Re‐ strikesoccurlessfrequently(oneperpowerfrequencyloop). Figure10.1depictsqualitativelythefrequencyofre‐strikesversusthedurationof the interruption process for the standard centre‐break disconnector with and without air flow. The continuous thin lines indicate the successful interruptions without air flow. The continuous bold line indicates the case of a failed interruptioninwhichthearccontinues"forever"inthethermalmode.Thedashed 144Chapter10 lines indicate interruptions with air flow assistance. At the earliest stage of the interruption, with and without air flow, the frequency of interruption is higher than during the remainder of the interruption process. With increasing arcing time, the frequency of re‐strike decreases until 100Hz, i.e. once per half 50Hz power‐frequency cycle. The decrease is faster with higher current. With air flow and also at low current, the frequency is always higher than 100Hz until final interruption. The air flow results in the overall effect of increasing the re‐strike frequencyatthesamecurrent.Threeregionscanbedefined: ‐ A region where the frequency clearly exceeds 100Hz. Re‐strikes are less influencedbytheresidualplasma.There‐strikevoltageismainlydetermined bythegaplength(leftregioninFigure10.1). ‐ A region where the frequency is 100Hz. The re‐strikes are increasingly influencedbytheresidualplasma(dependingonthecurrentlevel).Highre‐ strike voltage is still required in order to breakdown the air gap, but the current plays an important role in the recovery (centre region in Figure 10.1). ‐ Aregionwherenore‐strikesoccur.Thebreakdownisthermallydominated completely.Re‐strikevoltageisnotnecessaryinmaintainingofarcing.This situationappearswhentheinterruptionfails(rightregioninFigure10.1). Applicationofhigh‐speedauxiliarycontactsreducesthearcdurationsignificantly and the number of re‐strikes during the entire interruption is limited to only a few. The arc is burning between the auxiliary contacts, bridging the gap in a straight manner ("stiff arc"). This is different from interruption with a disconnector with slow moving main contacts, where thermally dominated breakdown occurs during a major period of the entire process with an "erratic" arc shape containing significant upward motion and curling. Application of fast opening contacts does not reduce the transient overvoltage and current amplitudes.However,thenumberofvoltagetransientsismuchsmaller,because oftheveryshortarcduration. It is recommended that the contact opening speed is as fast as possible without anybouncing.Withhighercontactseparationspeed,thearcisforcedtoelongate, coolingwillbemoreefficient,anditwillextinguishinashortertime.Themoment ofthewhiprelease,however,shouldbechosensuchthatdielectricbackre‐strike ofthearcbetweenthemainbladesisavoided. Actually the effect of a high‐speed interrupter and the air flow can also be understoodfromthebreakdownvoltages(seeFigures5.20and6.10).Figure10.2 conceptually depicts the breakdown voltage versus time with high‐speed interrupter,andwithcoldandhotairqualitativelyatthesamegaplength. ConclusionsandRecommendationsforFutureResearch145 Figure10.1.Frequencyofre‐strikeswithandwithoutairflowwiththeinterruptedcurrentas aparameter. Figure 10.2. Breakdown voltage against time with high‐velocity contacts, standard disconnectorandrecoverywitharcatvaryingcurrents. Inthecoldair,thedielectricrecoveryvoltageincreaseswithcoolingandremoval of the residual plasma by convection. The breakdown voltage is proportional to the gap distance. However, for higher currents the breakdown voltage is lower duetotheinfluencesfromapreviousarc.Withairflow,thebreakdownvoltageis 146Chapter10 higherduetothestrongerconvectiveplasmaremoval(seeFigure6.10);withthe assistanceofahigh‐speedinterrupter,thedielectricrecoverydevelopsevenfaster becauseofforcedconvectioncomingfromthesurroundingcoolairinaveryshort time. In this sense, cooling the arc either by dragging it into a cold environment (high‐speed auxiliary switch) or by supplying cold air (air flow) is the essential process. Since the maximum recovery voltage the circuit is able to supply is limited to 2p.u.,thearcwillstopwhenthebreakdownvoltageofthemediumhasreached this voltage. There is minimum arc duration (tm) for high‐velocity and standard disconnector, shown in Figure 10.2. For the interruptions with disconnecting underarcing,thearcdurationincreaseswithhigherinterruptedcurrent. 10.2Proposedfutureresearch Theconditionofthe(ionized)airleftfrompreviousre‐strikesisamainissuein understanding capacitive current interruption with a disconnector. The physical phenomenadiscussedinChapter9onlyprovidearoughbackground. Quantitative understanding of the influence of arc temperature, degree and amount of ionization in the air gap on the breakdown development and the dielectricrecoverycouldbeworthwhileforfurtherinvestigation. Forthefreeburningarcinairthereisnoproperarcmodelcoveringtherangeof currents occurring in the disconnector. This point is addressed in Chapter 5 but research, e.g. along the line of the approach proposed in the review given in Chapter2maybecarriedoutfurther. Interruption with the assistance of a pre‐inserted resistor is theoretically and experimentally only basically studied in Chapter 8. Future experimental work couldbecarriedoutinordertoaddresstheresistancevalueintermsofcurrent reduction, heat dissipation and difficulties related with the insertion of (high value) resistors into the circuit. The merits of applying non‐linear components couldalsobeconsidered. Ablationassistedarcextinctionispresentlyonlyusedforrelativelyhighcurrent interruption. The feasibility for low current levels, studied here, can be investigated. Such study could include new materials, which suit best with the applicationondisconnectorsforrelativelylowcapacitivecurrents. Publicationsrelatedtothiswork147 Publicationsrelatedtothiswork [1] Y.Chai,P.A.A.F.Wouters,R.P.P.Smeets,"CapacitiveCurrentInterruptionby HVAir‐BreakDisconnectorsWithHigh‐VelocityOpeningAuxiliaryContacts", IEEETrans.onPowerDelivery,vol.26,pp.2668‐2675,Oct.2011. [2] Y.Chai,P.A.A.F.Wouters,R.T.W.J.vanHoppe,R.P.P.Smeets,D.F.Peelo, "Capacitive current interruption with air‐break high voltage disconnectors", IEEETrans.onPowerDelivery,vol.25,pp.762‐769,Apr.2010. [3] Y.Chai,P.A.A.F.Wouters,R.P.P.Smeets,"Arcimagingoncapacitivecurrent interruptionusingadisconnectorwithanauxiliaryinterrupter",inProc.IEEE Asia‐PacificPowerandEnergyEngineeringConf.,Wuhan,4p,Mar.2011. [4] Y. Chai, P. A. A. F. Wouters, R.T.W.J. van Hoppe and R. P. P. Smeets, "The influence of air flow on current interruption with air‐break HV disconnectors", in Proc. the 17th International Symposium on High Voltage EngineeringConf.,CapeTown,6p,Aug.2011. [5] Y.Chai,P.A.A.F.Wouters,S.Kuivenhoven,P.vanderWal,M.Vanis,R.P.P. Smeets, "Current interruption phenomena in HV disconnectors with high‐ speedopeningauxiliarycontacts",inProc.IEEEPowerandEnergySocietyGM Conf.,Minneapolis,8p,Jul.2010. [6] Y.Chai,P.A.A.F.Wouters,R.T.W.J.vanHoppe,R.P.P.Smeets,"Arcimages during small capacitive current interruption with HV air‐break disconnectors", in Proc. the 18th International Conference on Gas Discharges andTheirApplicationsConf.,Greifswald,pp.158‐161,Sept.2010. [7] Y.Chai,P.A.A.F.Wouters,R.T.W.J.vanHoppe,R.P.P.Smeets,D.F.Peelo, "Experiments on capacitive current interruption with air‐break high voltage disconnectors",inProc.IEEEAsia‐PacificPowerandEnergyEngineeringConf., Wuhan,6p,Mar.2009. [8] Y. Chai, P. A. A. F. Wouters, R. T. W. J. van Hoppe, R. P. P. Smeets, "Arc behaviourinsmallcapacitivecurrentinterruptionwithhighvoltageair‐break disconnector", in Proc. the 16th International Symposium on High Voltage EngineeringConf.,Hannover,papernr.G‐19,6p,Aug.2009. Publicationsrelatedtothiswork148 Nomenclature149 Nomenclature Abbreviations AC CT CVT CN CO DS DSs DC HF,MF,PF HV GIS G M p.u. Rms RV S TR Symbols p H,M H,M θ θ1 θ2 θ1,0 θ2,0 τ a Δt1 Δt2 C Alternatingcurrent Currenttransformer Capacitivevoltagetransformer Cumulativenumber Closeandopen Disconnector Disconnectors Directcurrent High‐,medium‐,power‐frequency Highvoltage Gasinsulatedsubstation Short‐circuitgenerator Masterbreaker Perunit Rootmeansquare Recoveryvoltage Makeswitch Short‐circuittransformer Angularvelocity Angularpower‐frequencyofus OscillationangularfrequencyatHF,MF DampingconstantatHF,MF Initialangleofus Rotationangleoftheframerelatedtotheposition paralleloftheDSblades Rotationangleofthestriprelatedtotheposition paralleloftheDSblades Initialvalueoftheθ1 Initialvalueoftheθ2 Timeconstantofthedecayofhotgasleftfromarc Accelerationofthereleasedwhip Currentzerolastingperiod Current(notzerovalues)lastingperiodbefore thebreakdownoccurs Capacitance 150Nomenclature CH CapacitanceofCs,Clinseries Cs,Cl Capacitanceatthesourceandloadside C1…10 Capacitorsforvoltagedivider Bk Sumofthebrightnessofaframe (k) B(i,j) Brightnessofthekthframe FrontEndchannels CH1…4 d Whipcontactseparation D1…4 Voltagedivider di/dt Currentderivativeatcurrentzerocrossing Em Amplitudeofus E Electricfieldstrengthoftheairgap f Frequency H Archeightoftheagainstthebladesatoriginal position Short‐timewithstandcurrentofacircuitbreaker Isc id Currentthroughthedisconnector i,j Coordinatesofaimagepixel Id rmsofid Imax AmplitudeofthecurrentthroughtheDS iH,iM CurrentthroughthedisconnectoratHFandMF ipf CurrentthroughDSmeasuredatalowbandwidth Ihf Current through DS measured at a high bandwidth ia Arccurrent k_B: Boltzmannconstant Ls Inductanceatthesource LH EquivalentinductanceofHFloop La Arclength N Gasdensity(specifymolespervolumeornumber ofmoleculespervolume) Nk Normalizedbrightnessofaframe N0 Normalizedbrightnessofthefirstframeina videorecording Brightnessofthearcaftercorrectionwitha Narc backgroundframe Pressure P RVmax Peakoftherecoveryvoltage R s Resistanceatthesource RH EquivalentresistanceofHFloop R Resistor Tarc Arcingtime T Absolutetemperature ucs,ucl VoltageacrossCsandCl Nomenclature151 ud Voltageacrossthedisconnector Ur Re‐strikevoltageoftheairgap Uline rmsvalueofthesystemvoltage EqualizationvoltageafterHFoscillation UE uCM VoltageacrossCsandClatMF VoltageacrossCs,Clatthemomentofarestrike Ucs,Ucl Ugap Gapwithstandvoltage InputvoltageofthevoltagedividerofD1during ui calibration uin Inputvoltageoftheadjustabletransformer uo OutputvoltageofthevoltagedividerofD1during calibration uout Outputvoltageoftheadjustabletransformer Voltageacrossthedisconnectorafterswitching Uoc off ua Arcvoltageinstantaneousvalue Ua Arcvoltagemeanvalue vair Openingvelocityoftheairgap Instantaneoussourcevoltagevalue, us phase‐to‐groundvoltage v0 Initialspeedofthereleasedwhip w Energyinputthecircuituponrestrike wR Energydissipationoftheresistor 152Nomenclature Acknowledgement153 Acknowledgement Thisthesiswouldnothavebeenfinishedwithouttheeffortsfrommanypeople.I wouldliketomakeuseofthisopportunitytoshowmypersonalgratitudetothose whosupportedmeduringthecourseofthiswork. Firstofall,Iamindebtedtomypromotor,prof.dr.ir.RenéSmeetsforacceptingme as his Ph.D. student. I am very much appreciative for his enthusiastic encouragement,guidanceandsupportfromtheverybeginningtothefinalstage ofthisworkandmystayatTU/e.Heisintelligent,analytical,constructive,aswell asdiplomatic,kind,considerateandcaringinsociallife. Ialsoowemydeepestgratitudetomyco‐promotordr.PeterWouters.Hisdown‐ to‐earth attitude, high level knowledge, always fast and very carefully reviewing all my paper work within last four years are indeed remarkable. On many weekends,holidays,twilighthours,Isawthelightoninhisofficewhichwassuch aninspirationtome.Ienjoyedalotworkingwithhim,travellingandevensharing someimportantfestivals.IalsothankhiswifeFenforherfabulousfoodandtheir daughterYanfortakingcareofRuirui. I thank the sponsors for this project: IOP‐EMVT for funding, HAPAM for contributingtestobjectsandallowingtomeasureduringtestinginPrague,Enexis foritsgeneroussupplyofcapacitorsloadsandKEMAforsupplyinglaboratorytest space. Further, I would like to thank all the committee members, prof. dr‐ing. V. Hinrichsen,prof.ir.L.vanderSluis,prof.dr.E.Lomonova,dr.D.F.Peeloandprof. ir. W. L. Kling. Thanks for their time on reading my thesis and giving me very constructive feedbacks. Special thanks go to dr. David Peelo for his recommendationsandhistwovisitsallthewayfromCanada. Furthermore,alotofthanksgotoing.RenévanHoppe.Hedidalotofworkformy experimental setup. Without his smart ideas and great help the experiments wouldnotbefulfilled.IalsothankNillesVrijsenforhisinterestincalculatingthe disconnectorbladesmovingspeedduringhisinternshipwithme.Specialthanks go to ir. Sander Kuivenhoven from KEMA, ing. Paul van der Wal from HAPAM, Sjoerd van Driel from TU/e for their help during the tests in KEMA laboratories andothersuggestionsforthisproject.Ialsothankir.AdvanVenrooijforhishelp onthecircuitdrawingsinthelastmomentoffinalizingthethesis. Itisanhonourformetoshowmythankfulnesstomanycolleaguesandstafffrom theEESgroupfortheirgeneralsuggestionsandcaringtheprogressofthisthesis fromtimetotime.Ithankallmy(former)fellowPh.Ds.,Sharmistha,Shima,Zhen, Acknowledgment154 Rong,Vlada,Petr,Peter,Totis,Lei,Jasper,Anton,Paul,Arjan,Geesje,Glenn,Pavlo, Phuong,Vuong,Ioannis,Yan,Yu,Jin,Ballard,Jerom,HelderandmanyIstillhave forgottentomention,Ienjoyedverymuchtothetimespentwithyounotonlyin ourofficialcoffeeroombutalsointheexternalcoffeecorner.Thankyouguysfor beingpartofmyPh.D.life. Big thanks go to my personal friends from TU/e, and also those friends I met at RegionalInternationalSchoolinEindhoven.Thankyouforsharingallusefuland 'gossipy'informationabouteverything.Youknowhowimportantitistohaveall ofyouaround. Finally I am grateful to my Chinese and Serbian family, especially my beloved husband Miloš for his never‐ending love, support, understanding and encouragement. Of course, our always positive and joyful daughter Ruirui is a powerful source of inspiration and energy. Life becomes so colourful being with youtwotogether. YajingChai 24th,Jan.2012 CurriculumVitae155 CurriculumVitae Yajing Chai was born on 08‐01‐1975 in Hubei, China. She received her B.Sc. and M.Sc. degree from Wuhan University in 1996 and 2001 respectively at the High Voltage Group of Electrical Power Engineering. From 1996 till 2004, she was a laboratoryassistantandalecturerwiththeDepartmentofElectricalEngineering ofWuhanUniversity.ShewasanexchangeyoungscholaratBelgradeUniversity, Serbiabetween2004and2005.From2005to2007,shewasaresearcherwiththe High Voltage Institute of Wuhan University. In 2008 she joined the Electrical EnergySystemsgroupatEindhovenUniversityofTechnology,theNetherlands,as aPh.D.candidateunderthesupervisionofprof.dr.ir.R.P.P.Smeetsanddr.P.A.A.F. Wouters.