Vous êtes sur la page 1sur 7

ElectricalEngineeringResearch(EER),Volume3,2015

ComparisonofATPSimulationandMicro
processorBasedFaultLocationUsingDFT
HNouri*1,FJalili2,TBoxshall3
PowerSystems,ElectronicsandControlResearchLaboratory,UWEBristol,UK1&3,SadidPipe&EquipmentCo,
SavehRoad,Tehran,Iran2
*1

hassan.nouri@uwe.ac.uk

Abstract
This paper evaluates the results of a microprocessor based fault locator employed on a laboratory power system model with
those obtained within EMTP/ATP simulations. The Discrete Fourier Transform concept is used to extract the fundamental
frequency components of the voltage and current waveforms under fault and the reactance based method to locate the fault.
Theresultssuggestaclosecorrelationbetweentheexperimentalandsimulationstudiesforallfaulttypes.
Keywords
FaultLocation;DiscreteFourierTransform;Microprocessor;EMTP/ATP

Introduction
Distributionlinefaultshavebeenaprimaryconcerntonetworks.Faultsaffectthereliabilityoftheservicetothe
customer,thestabilityofthepowersystemandthequalityofthepowerdelivered.Distributionlinefaultscaneven
causecompletesupplyoutages.Thesefaultsmustbedetectedquicklyforfastdeactivationofthefaultysectionof
line,inordertoreroutethepowertothecustomertomeettargetrestorationtimes.
Distribution line fault locators can be classified under two categories. The first is the impedance based locators,
which analyse the power frequency (50 Hz or 60 Hz). The second is the travelling wave based locators which
analyse the travel time of high frequency transients. Transient based fault location methods require higher
sampling rates and more responsive voltage or current transducers. They also require multiple fault transient
detectors placed throughout the system, which are clocked using GPS technology and data sent to a central
locationviaacommunicationlink.TheuseofGPScanmakesuchasolutionveryexpensive.Transientbasedfault
location techniques for radial, meshed and distribution lines with tapped load power system configurations are
describedbyanumberofresearchers(Nourietal2014,Magnagoetal1998andNourietal2001),respectively.In
impedance based methods, the ratio of extracted fundamental frequency of the voltage and current waveforms
yields the impedance seen from the generator up to the point of fault. A useful summary of the available
impedancecalculationscanbefoundin(Zimmerman2005).Thereactanceequationthatwillbeusedtoobtainthe
faultlocation(m)inperunitisexpressedas:
v
IM s
is (1)
m
X 1TL

whereIMdefinedasstoringoftheimaginarypartofthevoltage( vs ) andthecurrent( is )ratioatthesendingend


oftheline,andX1TListhepositivesequencereactanceofthedistributionline.
Discrete Fourier Transform
MathematicalBackground
Inpowersystemsthevoltageandcurrentwaveformsareessentiallyshiftedsineorcosinewavesoffundamental
frequency.Thesesignalsareeffectivelyperiodicwhenthesystemisrunningundersteadystateconditions.Under
fault conditions, the fundamentalfrequency of the voltageand current amplitude can be extracted from the non

ElectricalEngineeringResearch(EER),Volume3,2015

periodicsignalsusingtheFourierTransform.ThebasisfunctionschosenfortheFourierTransformisthecomplex
exponential g (t ) e jt .TheFourierTransformisdefinedasfollows:

F ( ) f (t )e jt dt (2)

Oncloserexaminationof(2),itcanbeseenthatFourierTransformintegrationtakesplaceoveralloftime,i.e.from
to . Therefore, the Fourier Transform should be reserved for periodic signals only. A power system
contains many devices such as transformers, loads and switching circuits. These all create harmonic distortion
under steady state conditions. The Fourier Transform is useful to analyse the harmonic content of the signal.
However, when the system is running under a fault condition, non periodic signals exist. Consequently, the
Fourier Transform should not be used to analyse fault conditions. An alternative form of Fouriers analysis
techniquesshouldbeused,knownastheFourierSeries.UnliketheFourierTransform,theFourierSeriesanalyses
thesignaloverafiniteperiodoftime.ThebasicassumptionusedinFourierSeriesisthatthewaveformthatresults
fromthefaultconditionisassumedtobeperiodicwithintheintervalfrom t0 to t0 T ,where T istheperiodof
thefundamentalcomponent.Thefollowingequationsarederivedfrom(Wood1984,PhadKeetal1998,Johnsetal
1995,andCroftetal2001).
Thevoltageorcurrentwaveform s (t ) canberepresentedusingtheFourierSeries.

s (t )

a0
an cos n0 t bn sin n0 t (3)
2 n 1
n 1

wheretheFouriercoefficientsare:
an

2 t T
s (t ) cos n0 t dt (4)
T t
0

And
bn

2 t T
s (t ) sin n0t dt (5)
T t
0

where n = 1, 2, 3 , for the fundamental, second, and third harmonic . It is apparent that the fundamental
frequency can be selected by setting n 1 . Thus, the Fourier coefficients for the fundamental frequency can be
determined:
a1

2 t T
s (t ) cos 0 t dt (6)
T t
0

And
b1

2 t T
s (t ) sin 0t dt (7)
T t
0

Indiscreteanalysisthetimeofthejthsampleis t j j t andtheperiodofthewaveformis T N t ,whereNisthe


numberofsamplespercycleof50Hz.Now(6)and(7)canbedeterminedindiscretetime:
a1

2 N 1
2 N 1
2 N 1
s (t j ) cos 0 t j
s ( j t ) cos 0 j t
s ( j ) cos 0 j t (8)
N t j 0
N t j 0
N j 0

And
b1

2 N 1
2 N 1
2 N 1
s (t j ) sin 0 t j
s ( j t ) sin 0 j t
s ( j ) sin 0 j t (9)
N t j 0
N t j 0
N j 0

TheamplitudeoftherealandimaginarypartsofthefundamentalfrequencysignalarerepresentedbytheFourier
coefficients a1 and b1 respectively. It is necessary to determine the amplitude of the fundamental frequency in
ordertoapplyanimpedanceequationtolocatethefault,

ElectricalEngineeringResearch(EER),Volume3,2015

s a12 b12 (10)

Also it is required to find the phase angle between the real and imaginary components of the fundamental
frequency,
b
s tan 1 1 (11)
a1

PracticalApplicationofDFT
A single line diagram of the laboratory power system model is shown in Fig. 1, with a fault locator system
connectedtobusA.Itcanbeassumedthatthedistributionlinehasalengthof1pu.
AsystemmustbedesignedtogatherthephasevoltagesandcurrentsanduploadthemtotheCPUtoanalysethe
fault conditions. This requires a sensor circuit to measure the voltages and currents, interfaces between the
analogueand digital domains, andan external microprocessor tocontrol the datagathering process. The chosen
externalmicroprocessoristheSiemensC164CIonthePHYTECkitCON164evaluationboard.Ablockdiagramof
therequiredsystemisshowninFig.2.TheADCandDACare12bitdevices.ThedatasentbetweentheC164CI
microcontrollerandtheCPUisviaanRS232busandthemethodfortheC164CIcommunicatingwiththeADC
andDACisaserialperipheralinterface(SPI)bus.
AphotographofthelaboratorypowersystemmodelandmicroprocessorbasedfaultlocatorisshowninFig.3.

FIG.1SINGLELINEDIAGRAMOFTHELABORATORYPOWERSYSTEMMODEL

FIG.2ABLOCKDIAGRAMOFTHEDATAACQUISITIONSYSTEM

FIG.3PHOTOGRAPHOFLABORATORYPOWERSYSTEMMODELANDMICROPROCESSORFAULTLOCATOR

ElectricalEngineeringResearch(EER),Volume3,2015

VL VL
IL IL

V
Im L
I
m L
Z TL

V
Im LL
I
m LL
Z TL
FIG.4BLOCKDIAGRAMOFSOFTWAREFUNCTIONALITY

FaultLocationSoftware
ThefaultlocationsoftwareisafunctionbasedprogramasshowninFig.4.Therearethreemainfunctionswhich
determinethefaultlocation:
i.
Thefunctiongetsamples()usestimer2interruptservicesubroutinetogenerateasamplingperiodof625s.
Ateachsamplingintervalthefunctioncollectstheinstantaneousvalueofthelinevoltagesandcurrents.Itrepeats
thisuntiltherequirednumberofsamplesisgathered.
ii.
This data is passed on to the function DFT(), which extracts the amplitude and phase angle of the
fundamentalfrequency.
iii.
The analyse() function uses a fault current threshold determined during program initialisation to decide
which phases are faulty. A variation of reactance equation is then applied to locate the fault depending on the
numberoffaultyphases.Thereactanceequationgivesafaultlocationin m pu.
EMTPATPSimulation
AlinetogroundonphaseAissimulatedusinganEMTP/ATPsimulationmodelshowninFig.5.Thefaultoccurs
inthemiddleofthedistributionlinewithafaultlocationof0.5pu.TheEMTP/ATPmodelparametersarebasedon
thelaboratorypowersystemmodelshowninFig.1.
Distribution Line

Load

415 : 37

FIG.5ATPSIMULATIONMODELOFAFAULTY415VDISTRIBUTIONLINE

Results & Discussion


EMTPATPSimulation
Afaultissettooccuratt=0.06secondsandlastsfortheremainingdurationofthesignal(5cycles).Theprofilesof
theATPvoltageandcurrentwaveformscanbeseenforphasesA,BandCinFig.6andFig.7,respectively.
The DFT is used to extract the amplitude and phase angle of fundamental frequency line voltages and currents
from Fig. 6 and Fig. 7. All transmission line signals are sampled at a rate of 32 samples per 50 Hz cycle, which
producesasamplingperiodof Ts 625 s .TheDFTfundamentalfrequencyextractionofphaseAforvoltageand
currentcanbeseeninFig.8andfigureFig.9,respectively.Byexaminingthesewaveformsitcanbeseenthatthere

ElectricalEngineeringResearch(EER),Volume3,2015

are64prefaultsamples(64to0)and160samplesafterthetimeoffaultoccurrence.Duringtheprefaultsamples,
thefundamentalfrequencyvoltageandcurrentamplitudeis1pu.Theapplicationofafaultonthelinecausesthe
currentatthesendingendtoincreasetoapproximately11pu.Also,thereisadecayingexponentialDCoffsetof
short duration. It should be noted that the DFT ignores this DC offset and just tracks the amplitude of the
fundamental frequency, as required. The fundamental frequency voltage amplitude does not change by a
significantamount.
EMTP/ATP
Va(t)
Vb(t)
Vc(t)

EMTP/ATP

Current(pu)

Voltage(pu)

Ia(t)
Ib(t)
Ic(t)

Time
Time

Time

FIG.6ATPPHASEVOLTAGESWAVEFORMSFIG.7ATPPHASECURRENTSWAVEFORMS
EMTP/ATP

EMTP/ATP

FIG.8EMTPATPFUNDAMENTALFREQUENCYVOLTAGEFIG.9EMTPATPFUNDAMENTALFREQUENCYCURRENT
EXTRACTIONUSINGDFTEXTRACTIONUSINGDFT

These results show that the DFT can extract the fundamental frequency in approximately 32 samples, i.e. a
completecycleof50Hz.Theresultantfaultlocationinpucanbefoundbyusingthereactanceequation(discussed
earlier)asshowninFig.10.
Thiswaveformcorrectlyshowsthatthefaultlocationconvergesto0.5puaftertheapplicationofthefault.
TheresultantfaultlocationinperunitcanbefoundusingthereactanceequationasshowinFig.13.

1.5
Va(t)

EMTP/ATP

Experimental

VA(t)
1

Voltage(pu)

0.5

0.5

Pre-fault

Fault
1

Pre-fault

Fault

1.5
64

32

32

64

96

128

160

Samples

FIG.10EMTPATPFAULTLOCATIONWAVEFORMFIG.11MICROPROCESSORBASEDFUNDAMENTAL
FREQUENCYVOLTAGEEXTRACTIONUSINGDFT

ElectricalEngineeringResearch(EER),Volume3,2015

15

1.8

Ia(t)

10

Experimental

1.4

FaultLocation(pu)

Current(pu)

10

Pre-fault

15

Experimental

1.6

IA(t)

1.2
m(t)
1
0.8
0.6

Pre-fault

0.4

Fault

Fault

0.2
0

20

64

64

32

32

64

96

128

32

32

64

96

128

160

160
Samples

Samples

FIG.12MICROPROCESSORBASEDFUNDAMENTALFIG.13MICROPROCESSORFAULTLOCATIONWAVEFORM
FREQUENCYCURRENTEXTRACTIONUSINGDFT

VerificationofResultsUsingtheMicroprocessorBasedFaultLocator
Results produced by EMTP/ATP simulation are verified using the microprocessor based fault locator on the
laboratorypowersystemmodel.TheDFTisusedtoextractthefundamentalfrequency,amplitudeandphaseangle
forallthreelinevoltagesandcurrents.SimilarlytotheEMTP/ATPsimulationasamplingconditionof32samples
per50Hzcycleisused.TheDFTfundamentalfrequencyextractionofthevoltageandcurrentwaveformscanbe
seeninFig.11andFig.12,respectively.
Conclusions
The results produced in the EMTP/ATP environment are compared with experimental results produced by
microprocessor based fault locator on a laboratory power system model. The DFT extracts the fundamental
frequency inless than a complete cycle of 50Hzand when combined with the reactanceequation produces very
accuratefaultlocation.TheDFTeffectivelyignoresDCoffsetespeciallywhenpresentinheavilyinductivelines.
A collapse in line voltage or an increase in line current is a useful indicator of a faulty phase. The fault locator
allowsquickdetectionforfastdeactivationofthefaultyline.Thisallowsthepowertobereroutedtothecustomer
to meet target restoration times. Typically, distribution line faults have a fault impedance of unknown quantity,
causing error in fault location. The reactance method significantly reduces this error to a tolerable amount.
Generally,faultimpedanceisresistiveinnature.Thereactanceequationignoresallresistancesinthesystemand
locatesthefaultpurelybasedonreactiveimpedance.
TheDFTrequiresalargeamountofprocessingpowerduetotheneedforittoanalyseacomplete50Hzcycleat
everysampleperiod.Ithasthemajoradvantageoverothermethodsofextractingthefundamentalfrequencythat
nopriordistributionlineinformationisrequiredapartfromthereactivepartofthelineimpedance.
REFERENCES

[1]. Croft,R.Davison,andM.Hargreaves,EngineeringMathematics:AFoundationforElectronic,Electrical,Communications,
andSystemsEngineers:PrenticeHall,2001.
[2]. G.PhadkeandJ.S.Thorp,Computerrelayingforpowersystems:JohnWiley&Sons,Inc.NewYork,NY,USA,1988.
[3]. J.WoodandB.F.Wollenberg,Powergeneration,operation,andcontrol:WileyNewYork,1984.
[4]. T.JohnsandS.K.Salman,DigitalProtectionforPowerSystems:PeterPeregrinusLtd,1995.
[5]. Wang,Methodologies&AlgorithmsforFaultLocatorsinModernPowerSystems,PhDThesis,UWEBristol,2002.
[6]. F.H.MagnagoandA.Abur,Faultlocationusingwavelets,PowerDelivery,IEEEransactionson,vol.13,pp.14751480,
1998.

ElectricalEngineeringResearch(EER),Volume3,2015

[7]. H. Nouri and C. Wang, A Fast Fault Classification Technique for Power Systems, International Journal of Advanced
ResearchinElectrical,ElectronicsandInstrumentationEngineering,Vol.3,Issue11,November2014.
[8]. H. Nouri, C. Wang, and T. Davies, An Accurate Fault Location Technique for Distribution Lines With Tapped Loads
UsingWaveletTransform,Proc.2001IEEEPowerTech,Porto,vol.3,pp.1013,2001.
[9]. K.ZimmermanandD.Costello,Impedancebasedfaultlocationexperience,ProtectiverelayEngineers,200558thAnnual
Conferencefor,pp.211226,2005.