CHAPTER 1INTRODUCTIONThree characteristics generally provide means for detecting transformer internalfaults.These characteristics include an increase in phase currents, an increase in the differentialcurrent, and gas formation. When transformer internal faults occur, immediatedisconnectionofthefaultedtransformer isnecessarytoavoidextensivedamageandpreserve power system stability. Three types of protection are normally used to detectthese faults: overcurrent protection for phase currents, differential protection fordifferential currents, and gas accumulator for arcing faults.Overcurrent protection with fuses or relays provided the first type of transformer faultprotection.Transformerdifferential protectionisoneofthemost reliableandpopulartechnique for protecting large power transformers. The percentage differential principlewas applied to transformer protection to improve the security of differential protection forexternal faults with CT saturation.ifferential relays are prone tomaloperationinthe presence of transformer inrushcurrents. !nrushcurrentsresult fromtransientsintransformermagneticflux"#$%.Thefirst solution to this problem was to introduce an intentional time delay in the differentialrelay. &nother proposal was to desensiti'e the relay for a given time,to overcome theinrush condition "#(%, "#)%. Others suggested adding a voltage signal to restrain "*% or tosupervise the differential relay "#+%.This research focused primarily on methods of reducing the bloc,ing time of differentialprotection during inrush. These methods included ad-usting the slope of the differentialcharacteristics, ad-ustment of restraining current, and evaluation of current transformersduring saturation. #1.1. Motivation for this work.This wor,was motivated bythe need to reduce the bloc,ing time of differentialprotection during inrush conditions. This is following a number of questions that arisewhile applying differential relays for transformer protection..rotection of large powertransformers is a very challenging problem in power system relaying. /arge transformersare a class of very expensive and vital components of electric power systems. 0ince it isvery important to minimi'e the frequency and duration of unwanted outages, there is ahigh demand imposed on power transformer protective relays1 this includes therequirements of dependability associated with maloperation, security associated with nofalse tripping, and operating speed associated with short fault clearing time "#%. iscrimination between an internal fault and a magneti'ing inrush current has long beenrecogni'ed as a challenging power transformer problem "#%. This research will analy'ethe problem and its effect on transformer differential protection. 2irst, the research willreview the concept of transformer differential protection and then analy'e magneti'inginrush, overexcitation and current transformer saturation phenomena as possible causesof relay maloperation. 0ince magneti'ing inrush current generally contains a large secondharmonic component in comparison to an internal fault, conventional transformerprotection systems are designed to restrain during inrush transient phenomena by sensingthis largesecondharmonic. 3owever, thesecondharmoniccomponent mayalsobegeneratedduringinternalfaults in thepowertransformer"4%. Thismay be dueto CTsaturation, presence of shunt capacitance, or the capacitance in long extra high voltagetransmission lines to which the transformer may be connected.Themagnitude of thesecond harmonic inaninternalfaultcurrentcanbeclose toorgreater thanthat present inthemagneti'inginrushcurrent "#%. Thesecondharmoniccomponents in the magneti'ing inrush currents tend to be relatively small in modern largepower transformers because of the improvements in the power transformer core material.Thecommonlyemployedconventional differential protectiontechniquebasedonthesecond harmonic restraint will have difficulty in distinguishing between an internal faultand an inrush current thereby threatening transformer stability "#%.4Transformeroverexcitationisanotherpossiblecauseofpowertransformerrelaymal5operation. The magnetic flux inside the transformer core is directly proportional to theappliedvoltageandinverselyproportional tothesystemfrequency"6%. Overvoltageand7or underfrequency conditions can produce flux levels that saturate the transformercore. These abnormal operating conditions can exist in any part of the power system, soanytransformermaybeexposedtooverexcitation. Transformeroverexcitationcausestransformer heating and increase exciting current, noise, and vibration "6%. Though it isdifficult, withdifferential protection, tocontrol theamount of overexcitationthat atransformer can tolerate, transformer differential protection tripping for an overexcitationcondition is not desirable. 61.2 Probl! "tat!ntThe basic problems of transformer differential relaying fromthe perspective ofmagneti'ing inrush, overexcitation of the core, internal and external faults are reviewedinthecontext ofmeasurements, security, dependabilityandspeedofoperation. Thisresearch pro-ect investigates methods of reducing the bloc,ing time of differentialprotection during inrush conditions.1.2.1 Inr#sh C#rrnt!nrush current refers to the large amount of current that sometimes occur upon energi'inga transformer. Typically, for steady5state operation, transformer magneti'ation current isslightly less than (8 of the rated current "6%. 3owever, at the time of energisation, thiscurrent mayreach4$timesthenormal ratedcurrent beforequic,lydampingoutandreturning to steady state "6%. This damping effect typically ta,es less than twelve cycles. The practical inrush current magnitudes can range from $.$( to 4$ pu, depending on thepoint on wave of energisation, as well as the residual flux in the transformer core. 1.2.2Rsi$#al %l#& & typical resistive circuit has no memory"6%. The state of the circuit upon de5energisationhas no effect on the state of the circuit upon re5energisation, assuming that the circuit isnot damaged in the process. On the other hand, this does not hold for a transformer, orinductor wound on a ferromagnetic core. 9pon transformer de5energisation, the core remains permanently magneti'ed due to thehysteretic properties of the materials used. The transformer has a residual magnetic flux,themagnitudeofwhichisinfluencedbythetransformer:sspecificpropertiessuchaswinding capacitance and core gap factor "#%. This residual flux is mostly determined bythe opening angle, the point on the incoming voltage sine wave at which the transformerwaspreviously de5energi'ed. ;on5linearandcomplex relationships between theseandother factors ma,e the residual flux hard to predict, and in most cases, it would be easierto measure. *1.'. Rsar(h O#tlin!n Chapter #, the sub-ect and organi'ation of the research are described. The motivationof the wor, and the problem statement of the research are presented. 0ome bac,groundon transformer differential protection during inrush conditions is also presented. !nChapter 4, anoverviewof power systemprotectionandprotectionphilosophyispresented. !n this chapter the protection of power transformers with differential relays isdiscussed. .ercentage restraint differential relays are introduced. 2inally, the protection ofpower transformers with differential relays is presented.!n Chapter 6, simulation of a transformer as applied to differential protection is presented.0imulationswithtransformermodelsarecarriedoutusingboththeoretical andactualtransformer values. !n Chapter *, simulations of current transformers as applied to differential protection aredescribed. Current transformers generally produce negligible distortion undersymmetrical conditions but can become severely distorted under inrush conditions.0imulations with current transformer models under different systemconditions arepresented.Chapter (, simulations of differential relays as appliedtodifferential protectionarepresented.0imulations are carried out to set and ad-ust harmonic restrained differentialrelay to overcome the effects of the presence of inrush current on a power transformer.Chapter ), field case investigations of transformer maloperation are discussed.Chapter ernard.rice developed the firstapproach to differential protection. The advantages of the scheme proposed by ?er' and.rice were soon recogni'ed and the technique has been extensively applied since then "*%.3owever, it soon became apparent that differential protection operated incorrectly due toinrush currents. Over the years, various methods have been developed to ensure correctoperation of differential relays.2.1 Transfor!r Diffrntial Prot(tion & typical differential protectionsystemisshownin2igure4.#. ?ultiplecircuitsmayexist, but the example is sufficient to explain the basic principle of differential protection"4%. !t can be observed from 2igure 4.# that the protection 'one is delimited by currenttransformers. uetoits verynature, differential protectiondoesnot providebac,upprotection to other systemcomponents. 2or this reason, differential protection iscategori'ed as a unit protective scheme.The conductors bringing the current from thecurrent transformers to the differential relay are in some situations called pilot wires.)%i,#r 2.1- @eneral ifferential .rotection .rincipleifferentialrelays perform wellfor externalfaults as long asthecurrent transformersreproduce theprimarycurrents correctly"*%. Whenoneof thecurrent transformerssaturates, or if bothcurrent transformers saturate at different levels, falseoperatingcurrent appears in the differential relay and causes relay maloperation. 0ome relays usethe harmonics caused by the current transformer saturation for added restraint "*%.%i,#r 2.2- ifferential relay currents during normal operation or external, the equation is similar and is as follows.E.#ation 2.2where,>Ais the transformation ratio of current transformer >!>e! is the excitation current of current transformer > on the secondary side&ssuming equal transformation ratios, A =B, the relay operation current Iop is givenbyIop B !&e 5 !>e E.#ation 2.'uring normal system operation and during external faults, the relay operating currentIop is small, but never 'ero. !n the event of a fault in the protection 'one, the input currentis no longer equal to the output current. The operating current of the differential relay isnow the sum of the input currents feeding the fault. +Ae p A AsI I I =Be p Ba BsI I I =2.1.1 Pr(nta, rstraint $iffrntial /rot(tion.ercentage differential protection overcomes the problems related with the identificationof internal faults while ,eeping the advantages of the basic differential scheme "#%. !ngeneral, the operating current in the differential relay is equal to:E.#ation 2.0where,!#, !4 are the currents on the pilot wires of the current transformersue to the complexities associated with transformer differential protection,differentialrelays use a percentage restraint characteristic that compares an operating current with arestraining current. .ercentage restraint increases the operate current needed to actuatethe relay based on the current flowing through the protected transformer. The restraintsetting, or slope, defines the relationship between restraint and operate currents as shownin 2igure 4.6 "(%. The operating current, also called the differential current, !O., can beobtained from the phasor sum of the currents entering the protected element as shown inCquation 4.*.

!O.is proportional to the fault for internal faults and approaches 'ero for any operatingconditions. The differential relay generates a tripping signal if the operating current, !O.,is greater than a percentage of the restraining current, !DT.!O. > 0/.i.!DT E.#ation 2.1where,0/.i is the slope of the ith characteristic of the differential relay=4 #I I IOP =2.1.1.1 Cal(#lation of !ini!#! /i(k #/ (#rrntThe minimum pic,up restraint setting, !p.u EminF ad-usts the sensitivity of the relay. !nnon5numerical relays, the !p.uEminF was fixed at a typical value of $.6( of the relay tap"(%. 0electing a lower !p.uEminF setting needed an increase in the slope setting to maintaina given margin at the ,nee5point of the differential tripping characteristic. Conversely, itis sometimes necessary to accommodate unmonitored loads in the differential 'one. !nthat case, the !p.uEminF setting may be higher. & setting of +.21 per unit of transformerfull loadratingis recommendedfor typical installations wherenounmonitoredloadneeds to be considered. This value is well above the magneti'ing current and provides asafe margin at the ,nee point of the slope characteristic. 2.1.1.2 Cal(#lation of $sir$ !ini!#! /i(k#/ sttin,sTypical differential relay operating characteristic is shown in 2igure 4.6. Thecharacteristic consists of twoslopes, 0/.#and0/.4andahori'ontal straight linedefining therelayminimum pic,upcurrent, !..9. The relayoperating regionislocatedabove the slope characteristic and the restraining region is below the slope characteristic"*%.%i,#r 2.'- ifferential relay with dual slope characteristics#$2.1.2 Transfor!r Prot(tion %a#ltsOvercurrent, differential andgas accumulationarethreetypes of protectionthat arenormally applied to protect power transformers.Overcurrent protection provides the first type of transformer protection, and is used forsmall transformers. ifferential protectionreplacesovercurrent protectionasthemainprotectionfor large power transformers. &nelectric arc inoil decomposes the oil5producing gases. The emission of gas is used in gas accumulator and rate5of5pressure5riserelays to detect internal arcing faults.2.1.2.1 T2/s of fa#lts 2aults can be classified as through faults and internal faults. & through fault is locatedoutsidetheprotection'oneofthetransformer.Theunit protectionofthetransformershould not operate for through faults. The transformer must be disconnected when suchfaults occur only when the faults are not cleared by other relays in pre5specified time.!nternal faults can be phase5to5phase and phase5to ground faults. These internal faults canbe classified into two groups.3ro#/ I-Clectrical faults that cause immediate damage but are generally detectable byunbalance of current or voltage. &mongst them are the following:G .hase5to5earth faultG .hase5to5phase faultG 0hort circuit between turns of high5voltage or low5voltage windingsG 2aults to earth on a tertiary winding or short circuit between turns of a tertiary winding##3ro#/ II- These include incipient faults, which are initially minor but cause substantialdamage if they are not detected and ta,en care of. These faults cannot be detected bymonitoringcurrents or voltages at the terminals of the transformer. !ncipient faultsinclude the following:G & poor electrical connection between conductorsG & core fault which causes arcing in oilG Coolant failure, which causes rise of temperatureG >ad load sharing between transformers in parallel, which can cause overheating due tocirculating currents2or a group ! fault, the transformer should be isolated as quic,ly as possible after theoccurrence of the fault. The group !! faults, though not serious in the incipient stage, maycause ma-or faults in the course of time. !ncipient faults should be cleared soon after theyare detected.2.1.2.2 Probl!s of $iffrntial /rot(tion a//li$ to /owr transfor!rs& number of factors affect adversely the balance of the currents being compared. 0ome ofthese factors are as follows "=%: Two current transformers do not perform equally, even when they are from thesame manufacturer and have the same ratio and type. The remnant magnetic fluxes in the cores of two current transformers may not beidentical and consequently their excitation currents are not identical. The saturationof one of the current transformers affects the waveformandreduces the output of the current transformer. The difference of the outputs of thetwo current transformers manifests as relay operating current. ifference in length of the wiring produces a difference in the resistance of thepilot wires. This difficulty is overcome by connecting ad-ustable resistors to pilotwires.#4 The incoming and outgoing sides of a power transformer have different voltageand current levels. 2or this reason, the ratios of current transformers used on thetwo sides of a differential protection must be different. The power transformer connection produces a phase displacement fromtheprimary voltages and currents to the secondary voltages and currents. The delta5wyeconnection, themost commonoftransformer connections, producesa6$degree displacement. This phase mismatch can also be corrected by the softwareof numerical relays.2.2 Ma,nti4in, Inr#shWhen a transformer is initially energi'ed, there is a substantial amount of current throughthe primary winding called inrush currents. The rate of change of instantaneous flux in atransformer core is proportional to instantaneous voltage drop across the primary winding"%. &s will be discussed in chapter 6, the voltage of the transformer is a derivative of theflux, and the flux is the integral of the voltage. !n a normal operation, the voltage and theflux are phase5shifted by =$ as shown in figure 4.*. %i,#r 2.0 Holtage, ?agnetic 2lux and Current WaveformsWhen the transformer is energi'ed at the moment in time when the instantaneous voltageis at 'ero, the flux and current build up to their maximum level as shown in figure 4.(.!n a transformer that has been sitting idle, both the magnetic flux and the winding currentshould start at 'ero. When the magnetic flux increases in response to a rising voltage, itwill increase from 'ero upwards. Thus, in a transformer that is energi'ed, the flux willreach approximately twice its normal pea, magnitude as shown in figure 4.)#6%i,#r 2.1 Transformer energised when the voltage is at 'ero%i,#r 2.5 Transformer energised when the flux is at 'ero!n an ideal transformer, the magneti'ing current would rise to approximately twice itsnormal pea, value "4%. 3owever, most transformers are not designed withenoughmargins between normal flux pea,s and the saturation limits. uring saturation,disproportionate amounts of mmf are needed to generate magnetic flux. This means thatthe winding current, which generates the mmf to cause flux in the core, willdisproportionately rise to a value exceeding twice its normal pea, as shown in figure 4.esides, the burden of these relays can be high.3owever, electromechanical relays were so extensively employed, tested and ,nown thateven modern relays employ their principle of operation, and still represent a good choicefor certain of applications "#6%."oli$9stat rla2s-With the advances in electronics, the electromechanical technologywas replaced bystatic relays inthe early)$:s. 0tatic relays definedthe operatingcharacteristic based in analog circuitry rather than in the action of windings and coils.The advantages that static relays showed over electromechanical relays were of reducedsi'e, weight and electrical burden. Di,ital rla2s- ?icroprocessors incorporating into the architecture of relays in the +$:s.igital relays incorporatedanalog5to5digital converters E&CsF tosampletheanalogsignals from instrument transformers, and used microprocessors to define the logic of therelay. igital relays presented an improvement in accuracy and control, and the use ofmore complex relay algorithms, extra relay functions and complementary tas,s.N#!ri(al rla2s- The difference between numerical relays and digital relays lies in themicroprocessor used. ;umerical relays use digital signal processors E0.F, which containdedicated microprocessors especially designed to perform digital signal processing.2.0 "#!!ar2The operatingprinciplesof differentialprotectionhavebeendescribedinthis chapter.The differentialprotectionprincipleand thepercentagerestraint differentialprotectionhave been presented. The differential protection of power transformers, together with theproblemsandissuesoftheirapplication, werepresented.& chronologyofrelayswaspresented. 464*CHAPTER ' TRAN"%ORMER MODE= '.+Intro$#(tion!n most power systems, differential protection is applied for transformer capacity above#$?H&, while overcurrent protection is used for transformers below #$?H&.Transformers create large inrush currents when they are energi'ed. This inrush current isrich in harmonics and assumes large initial pea, value of about ( to 6$ times of the ratedvalue "+%. This condition causes maloperation of differential relays. !n order to preventfalse tripping due to the inrush current, a technique using the second harmoniccomponent of the current waveform is commonly used. Therefore,tounderstand thephenomenaofinrushcurrentit wasuseful to firstcreatemodelsthat describetheperformanceofatransformer underinrushconditions. Thischapter will describe the simulation model that was designed using the ?atlab70imulin,program to analy'e the effect of inrush currents on differential protection.'.1 Mo$lin, Transfor!r H2strsis?odelingthe coreofthetransformer is aninvolvedprocess because of thenonlinearbehavior of the flux in the core. To model the hysteresis, an approximate process withlinear elements, resistance and inductance was implemented in ?&T/&>. 2lux can beexpressed as in Cquation 6.# using 2araday:s law.E.#ation '.13ence, the flux density is1E.#ation '.24(dtdN e= = edtABCquation 6.#shows that the flux is directly proportional to the integral of the voltageacross thewinding. The magnetic fieldintensityinthetransformer is alsodirectlyproportional to the current. 3ence, the flux density, >, versus the magnetic field intensity,3, can be approximated by the voltage integral versus current.-1500 -1000 -500 0 500 1000 1500-150-100-50050100150CurrentVoltage IntegralVoltage integral versus current of resistive element%i,#r '.1-Holtage !ntegral versus Current of Desistive Clement2igure 6.# shows that the integral of voltage and resistive current are phase5shifted by =$.uetothephaseshift, therelationshiphasanelliptical shapewithtworadiithatarefunctions of the resistance and the angular frequency "#%.4)%i,#r '.2- Holtage !ntegral versus Current of !nductive ClementWhen the integral of voltage and inductive current are in phase, they form a straight linerelationship as shown in 2igure 6.4.When the two elements are added together in parallel as shown in 2igure 6.6, the totalcurrents are given by Cquation 6.6E.#ation '.'-1500 -1000 -500 0 500 1000 1500-1500-1000-500050010001500CurrentVoltage IntegralApproximate Representation of Transformer Hysteresis%i,#r '.'- &pproximate representation of Transformer 3ysteresisTransformer excitation is shown in 2igure 6.*. This is used in this study as anapproximate representation of the transformer core. -1500 -1000 -500 0 500 1000 1500-1500-1000-500050010001500Inductance CurrentVoltage Integral versus current of the inductive element4 Not a vali$ link.%i,#r '.1 Transformer Cquivalent Circuit'.2.1 Anal2sis of th E.#ivalnt Mo$lThe equivalentcircuitof 2igure6.* can further be reduced to thatof 2igure 6.(. Theprimary current of the transformer is given by Cquation 6.*E.#ation '.0The current !4is equal to the load current as seen from the primary side. This is also,nown as the reflected load current. The relationship between !4 and !4 is the turns ratioof the transformer as given by &mpere5Turns Cquation "#%.!4;# B !4;4 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02-1.5-1-0.500.511.5excitation current (pu)excitation fux (pu)4++ =4 $ #I I IE.#ation '.1The voltage equations of the primary and secondary circuits are:E.#ation '.5E.#ation '.6Cquation 6.) can be re5written by substituting Cquation 6.< into Cquation 6.)E.#ation '.7E.#ation '.8H4is the reflected voltage of the secondary windingD4 is the reflected resistance of the secondary windingI4 is the reflected inductive reactance of the secondary windingE.#ation '.1+The transformer equivalent circuit is redrawn as per equation 6.#$ and is shown in 2igure6.).4=#444NNII=F E# # # # #jX R I E V + + =F E4 4 4 4 4jX R I V E + + =( ) ( )# # # 4 444#44#4 #jX R I jX RNNINNV V + + + + == 4#4 4NNV V44#4 4= NNR R44#4 4= NNX X( ) ( )# # # 4 4 4 4 #jX R I X j R I V V + + + + =%i,#r '.5: Transformer Cquivalent Circuit with values refered to the primary'.2.2 Matlab?"i!#link Mo$l of th Transfor!r.Thetransformer equivalent circuit wasmodeledusingCquation6.#$.Themodel wasdesigned to simulate the current inrush phenomenon upon transformer energisation. Thetransformer was modeled as a series resistance and lea,age inductance and by a nonlinearmagneti'ing inductance. The core loss was represented by a shuntresistive branch D$TRANSFORMER EQUIVALENT CIRCUITe(2.7womag curveSine WaveScope.47Re.37*(u+u^9)MagCurve-K-Le1sIntegratordu/dtDerivative1/501/RcVin VinlambdaIcimIeIe%i,#r '.6 The ?atlab70imulin, circuit for the equivalent transformer.2igures 6.+ to 6.#6 show the transformer magneti'ing curves forvarious incident angles6$%i,#r '.7- ?agneti'ing Curve at $J .hase &ngle%i,#r '.8- Ma,nti4in, C#rv at 01@ Phas An,l6#%i,#r '.1+- ?agneti'ing Curve at =$J .hase &ngleTh rs/(tiv inr#sh (#rrnts ,ra/hs ar shown in a//n$i&'.' Mo$l Aali$ation!t was found necessary to fit parameters of the ?&T/&>70!?9/!;K model to a realtransformer. This allowed comparison of theoretical and practically obtained results andalso to evaluate the elimination of transformer inrush currents through controlled closing.Todetermine the parameters of the transformer, open andshort circuit tests wereperformed'.'.1 O/n Cir(#it TstsThe open5circuit tests were performed in order to determine exciting branch parameters,of the equivalent circuit, the no5load loss, the no load exciting current, and the no5loadpower factor. The experiment setup is shown in 2igure 6.##. & rated voltage was appliedto the primary side of the transformer while the secondary winding was open5circuited. 64%i,#r '.11: Cxperimental setup of the open5circuit test'.'.2 "hort Cir(#it TstsThe short5circuit test was conducted by short5circuiting the secondary terminal of the transformer, and applying a reduced voltage to the primary side, as shown in 2igure 6.#4, such that the rated current flowed through the windings. %i,#r '.12- Cxperimental 0etup of the 0hort5Circuit Tests66Continuouspowerguiv+-Vol tmeterScope0Real& Reacti ve Powersignal rmsRMS3signalrmsRMS2signalrmsRMS1signalrmsRMS1 2i near !rans"ormer0#i spla$30#i spl a$20#i spl a$i+-Current i+-%mmeterV&P'%cti ve & ReactivePower%C Voltage Source%i,#r '.1' Transformer short circuit test simulationThe results of the open and short circuit tests are shown in Table 6.4 Test Holts, H Current !#, & Current !4 .ower, WOpen Circuit 4*$$ $.*+*< $ #loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 0.5 Current Transformer .olarity diagram0.1.1 Conn(tionsThere are three ways that current transformers are connected on three5phase circuits1 wye, opendelta and delta.0.1.1.1 )2 Conn(t$!n wye connection a current transformer is placed in each phase with phase relays to detect phasefaults. !nthisconnectionsecondarycurrentsareinphasewithprimarycurrent asshownin2igure *.loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 1.2 Three5phase !nternal 2ault%i,#r 1.'a 0imulation of differential currents for normal operation and after fault inceptionC/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 1.'b .i!erential currents during normal operation%i,#r 1.'( i!erential currents after fault inceptionThe first part is the behavior of the differential relay currents during the normal operation of thepower transformer. The second part is the behavior of the differential relay currents after theoccurrence of the fault. Oooming of the simulations of the differential currents during normaloperationandaftertheoccurrenceofthefault areshownin2igure(.6EbFand2igure(.6EcFrespectively. The relevant values for ad-usting purposes of the unfiltered restraining current andthe operating current are summari'ed in Table (.#.C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush ConditionsDiffrntialC#rrntNor!alO/ration Aftr %a#lt!nitial 0table !nitial 0tableOperating #.(( $.+ 6$.* (+Destraining ).loc,ing Time of ifferential .rotection uring !nrush ConditionsIn /igure 0.1 are shown the di!erential currents 2ust after the fault inception,and below it, the second harmonic magnitude of the operating current forthe same period of time. In can be observed that the second harmonic of theoperating current had pic values from the time of the fault inception t=0.55seconds up to t=0.562 seconds. The di!erence between operating currentand restraining current during the pic values of the second harmonic wasnot large enough to avoid maing the restraining current temporarily largerthantheoperationcurrent 2ust after thefault inception, whichcausedadelaying in the identi"cation of the fault condition by the di!erential relay.%i,#r 1.7- i!erential currents and second harmonic current 2ust after fault inceptionC/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush ConditionsIn /igure 0.3 shows the operating current andthemodi"ed restrainingcurrent. /igure 0.3a shows the entire simulation, /igure 0.3b shows a zoomedsimulation of the di!erential currents during the a!ect of the inrush current.In /igure 0.3c isthe zoomed simulation of the di!erential currents 2ust afterfault inception.It can be observed that the restraining current was larger than the operatingcurrent during the e!ect of the inrush current and for the rest of the normaloperation. 4owever, in/igure0.3citcan beobservedthattherestrainingcurrent was larger than the operating current during the "rst 5 samples afterfault inception, timethat representsthedelayof thedi!erential relaytoidentifythefaultcondition. Thismeansthatthedi!erential relaymadeacompromise by delaying the identi"cation of a fault, loss of dependability, inorder to avoid false tripping during the presence of inrush current, increasein reliability.:a< 6omplete simulation graph showing the normal operation and fault eventC/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions:b< 7oomed di!erential currents during the e!ect of the inrush current:(< 7oomed di!erential currents after fault inception%i,#r 1.8- i!erential currents ad2usted to overcome inrush current issuesC/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions1.'Diffrntial /rot(tion of #n$r intrnal fa#ltsThe purpose of this study was to investigate the response of differential relays to internal faultsintheprotectedtransformer.Thedifferential relayad-ustedinprevioussectionswasusedtocarry out this study. The study was divided in the simulation of internal fault on the side of C#s5B and in the simulation of internal faults on the side of C#s 8B as shown in 2igure (.#$%i,#r 1+- !nternal 2ault 0imulationThe response of the differential currents for a three5phase internal fault in front of the CT:s of theside of C#s 5 Qconsidering that the CT:s RseeS toward the transformer location. The response ofthe differential currents for a phase &5to5ground internal fault in front of the CT:s on C#s 5isshown in 2igure (.##.C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 1.11- ifferential currents, three5phase internal fault, >us ) side%i,#r 1.12- ifferential currents, phase &5to5ground internal fault, >us ) side!t can be observed that the differential relayshowed correct operation for the simulated faults.>oth responses also showed the delay in the identification of the fault condition.C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush ConditionsThe response of the differential currents fora three5phase internal fault and a phase to groundfault infront of theCT:sof thesideof>us =is shownin2igures (.#6and2igure(.#*respectively. The differential relay showed correct operation for both simulated faults. 3owever,it was also observed that the responses to the internal faults of side of C#s 8showed a shorterdelay in the identification of the fault conditions, which improved the differential relayperformance.%i,#r 1.1'- ifferential currents, phase &5to5ground internal fault, >us = side%i,#r 1.10- ifferential currents, three5phase internal fault, >us = sideC/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions1.0Diffrntial /rot(tion #n$r &trnal fa#ltsThe purpose of this study was to investigate the response of differential relays to external faults.The differential relay ad-usted in previous sections was used in this study. The study was carriedout for external faults before C#s 5, and for external faults after C#s 8, as shown in 2igure (.#(%i,#r 1.11- 0imulation of Cxternal 2aultsThe response of the differential currents for a phase &5to5ground external fault located directly atC#s 5 is shown in 2igure (.#). The response of the differential currents for a three5phase externalfault, located directly atC#s 5is shown in 2igure (.#loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 1.15- i!erential currents, three&phase e8ternal fault, at C#s 5%i,#r 1.16- i!erential currents, phase A&to&ground e8ternal fault at C#s 5The response of the differential currents for a three5phase external fault, located directly at C#s 8is shownin2igure(.#+. Theresponseof thedifferential currentsforaphase&5to5groundexternal fault located directly atC#s 8is shown in 2igure (.#loc,ing Time of ifferential .rotection uring !nrush Conditions%i,#r 1.17- i!erential currents, three&phase e8ternal fault, at C#s 8%i,#r 1.18-ifferential currents, phase &5to5ground external fault C#s 8C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions1.1Diffrntial /rot(tion with CT sat#rationThe purpose of this study was to observe the behavior of the numerical differential relay modelwith CT saturation. 0econdary currents for normal and current transformers saturation for CT)and CT= with ratios #loc,ing Time of ifferential .rotection uring !nrush Conditions6.' Diffrntial Prot(tion Rla2 "i!#lations3istorically, different means of delaying differential protectionwere usedtoprevent falsetripping during inrush conditions. !n most cases, the relay was disabled for a given time whenswitchingatransformer. This inmodernpower systempracticeis nolonger consideredanacceptable means of restrainingthe differential relayduringmagneti'inginrushconditionsespecially for large power transformers. There are several means of restraining differential relaysduring magneti'ing inrush. The research used the second harmonic restraint method. 6.'.1 Diffrntial /rot(tion a,ainst inr#sh (#rrnt?agneti'ing current inrush appears as an internal fault to differential relays. 0imulations werecarried out to set and ad-ust a harmonic restraint differential relay during inrush. The behaviourof the differential currents during the inrush current conditions was developed. On energisation,it wasfoundthat theoperatingcurrent waslarger thantherestrainingcurrent. This causedmaloperation of the differential relay. This is unwanted operation because inrush current is not afault current.The slope of second harmonic restraint current was ad-usted and set so that therelay did not trip on energisation. 6.'.2 Diffrntial /rot(tion to intrnal fa#lts0imulations were carried out to find the response of the differential relay to internal faults. Whenthe second harmonic restraint of the differential relay is properly ad-usted and set1 the relay candiscriminate between internal and external faults. Three5phase fault and a phase to ground faultwere simulated. The relay showed correct response to both fault simulations. When a differentialrelayiscorrectlyad-ustedandsetitcancorrectlydiscriminatebetweeninternal andexternalfaults. This alsoincreases theresponseof therelaytointernal faults anddifferential relayperformance.C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions6.'.' Diffrntial /rot(tion to &trnal fa#lts0imulations were also carried out to find the response of the differential relay to external faults.Therelayremainedstabletoexternal faults. 0incedifferential protectionis aunit type ofprotection, it should remain stable for all faults outside its 'one of protection.6.'.0 Diffrntial /rot(tion #n$r (#rrnt transfor!r sat#ration (on$itions0imulationswerecarriedout tofindtheresponseoftherelaywhenthecurrent transformersaturate. 3igh primary currents will result in the creation of a high flux density in the currenttransformer ironcore. Whenthis densityreaches or exceeds thedesignlimits of thecore,saturation results. When the current transformers reach that point, their accuracy becomes verypoor andthe output is distortedbyharmonics. This affects the secondarycurrent output.0imulation results show maloperation of the relay due to current transformer saturation. 0incedifferential relays use second harmonic to discriminate between inrush and harmonics, currentsaturation adversely affects the relay in discrimination between inrush and a fault. C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions6.0 CONC=U"IONWhen a transformer is energi'ed, there is large amount of inrush current generated in its primarywinding.This current appears only on one side of the transformer and is not reflected on theother side of the transformer. This causes an imbalance of the currents appearing at thetransformer differential relay. This imbalance will be seen as a differential current and will causethe differential relay to trip. 0ince an inrush condition is not a fault condition, the operation of adifferential relay during an inrush condition must be prevented.There are several ways of restraining the differential relay from operating during inrush. Theseinclude desensiti'ing of relays1 wave shape recognition techniques and harmonic based methods.esensiti'ation method is no longer being practised. Wave shape recognition methods are stillrelatively new and not widely practised. 3armonic based methods are widely practised and thisresearch used the second harmonic restraint method.The inrush current has a large harmoniccomponent which is not present in fault currents. !nrush currents generate harmonics with secondharmonicamplitudesashighas)(8ofthefundamental.Thisisusedbyharmonicrestraintrelays to distinguish between faults and inrush. Transformer models were designed to give an in5depth understanding of the inrush phenomenon.These simulations were developed using ?atlab70imulin,. ;o load transformer simulations werecarriedout. Thesesimulations showedhighmagnitudeof asymmetrical current withahighharmonic content. The magnitude of the inrush current was found to be depended on the point ofvoltage at which switching in occurred. The greatest inrush current occurred when the incidentvoltage was at $J and 6)$J and least occurred when the voltage was at =$J and 4loc,ing Time of ifferential .rotection uring !nrush Conditionsue tocurrent transformer saturation duringinrushconditions, the amount of the secondharmonic current may drop considerably affecting protection relays that use second harmonicrestraint method.0imulations werecarriedout todeterminetheperformanceof thedifferential relayduetointernal faults, external faults and during current transformer saturation conditions. Thesesimulations were developed using .0C&. The model provided valuable insight into thebehaviour of a differential relay in a wide range of field events. 0imulations were first carried outto ad-ust and set the second harmonic restraint slope. !nrush currents were created by opening and closing the circuit brea,er causing energi'ation ofthe transformer and inrush currents. The relay restrained the effect of the presence of the inrushcurrent. 0ince inrush is not a fault, it showed correct operation. !nternal faults were simulated to investigate the response of the relay to internal faults.The relayshowed correct operation for the simulated faults. 0ince differential protection is a unit type ofprotection, it showed correct operation by remaining stable to all external faults. Current transformer saturationwassimulatedtoinvestigatethebehaviour ofthedifferentialrelay. The differential relay was adversely affected by current transformer saturation. ependingonthe degree of saturation, differential relays maymaloperate due tocurrent transformersaturation.Theharmonicrestraint methodadds theharmoniccomponent of theoperatecurrent tothefundamental component of the restraint current, providing dynamic restraint during transformerinrush. 3armonic restraint methods ensure relay security for a very high percentage oftransformer inrush currents. .roperly setting and ad-usting the second harmonic restraintpercentage reduces the bloc,ing time of differential protection during inrush. !t also providesrelay reliability to internal faults and stability to external faults.C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions3armonic restraint methods may not be adequate to prevent differential element operation foruniquecaseswithverylowharmoniccontent intheoperatingcurrent. ?odernmethodsfordifferentiatinginrushcurrent fromfault current mayberequiredtoensuresecuritywithoutsacrificing fast and dependable operations when energising a faulted transformer. 2urtherresearch is required in methods such as wavelet5based techniques for discrimination of internalfaults from magneti'ing inrush currents in power transformers. C/C; ($(Desearch .ro-ect ?arch 4$$loc,ing Time of ifferential .rotection uring !nrush Conditions7. Rfrn(s"#% C.?. Ong Rynamic 0imulation of Clectical ?achinery using ?atlab70imulin,S .rentice3all, #==+"4% V. 3. >run,e, V.K. 2rWhlich, XCliminationof Transformer !nrushCurrentsbyControlled0witching, X!CCC Transactions On .ower elivery, vol. #), no. 4, &pril 4$$#."6% &.?. @u'man and 0. Oocholl, R.erformannce analysis of traditional and improvedtransformer differential protective relaysS"*% & @u'man and 0.Oocholl R & current based solution for transformer differential protection Qpart #:.roblem 0tatementS ,S !CCC Trans. .ower elivery, volume #) no.* pp. *+(*=#, 4$$#"(% ?.Thompson and V.D. ClossonR9sing !opCharacteristics to troubleshoot transformerdifferential relay misoperationS >alser Clectric, 4$$(")% ..C. 0utherland R&pplication of transformer ground differential protection relaysS,S !CCCTrans. .ower elivery, volume 6) no.# pp. #)54#, 4$$$".Kas'tenny and ?.Ke'unovic, R!mproved .ower Transformer .rotection 9sing ;umericalDelaysS, !CCC Computer &pplications in .ower, Hol.##, ;o.*, October #==+, pp.6=5*(."+% K.Karsai, .Kerenyi and /. Kiss , R/arge power transformers+, Clsevier, ;ew Yor,, #=+. Campbell, ?. .o''uoli, ROperate and Destraint 0ignals of aTransformer ifferential DelayS, $,th Ann-a. /eor0ia %e1hProte1tive Re.a2in0 *on3eren1e, ?ay4$$$."#*% W. &. Clmore, R.rotective Delaying. Theory and &pplications,S 0econd Cdition,?arcel e,,er !nc., 4$$*."#(% / 2. Kennedy and C. . 3ayward, R3armonic5Current5Destrained Delays for ifferential.rotection,S AIEE %ransa1tions, Hol. (