Professional Documents
Culture Documents
EndofSemesterReport,fall2012
DavidSmith
PreparedtoPartiallyFulfilltheRequirementsofECE401
DepartmentofElectricalandComputerEngineering
ColoradoStateUniversity
FortCollins,Colorado80523
ProjectAdvisor:Dr.GeorgeCollins
IndustrySponsor:DohnSimms,PraxisCorporation
Approvedby:DrGeorgeCollins
Abstract
Formyseniorproject,Ichosetofocusonthesubjectofarcflashhazardsinelectrical
systems,specificallyinutilityscalewindfarmsandphotovoltaicarrays.Asanelectrician,this
subjectcombinesmyinterestsinelectricalworkersafetyandthecontinuedrapidpaceof
thedevelopmentofalternativemethodsofgeneratingelectricity.Thissubjectisofgreat
practicalinteresttoFacilityandProjectManagerswhoareresponsibleforthesafetyoftheir
workersbutmustalsobalancethisconcernwiththeneedtomaintainareliablesystemwith
minimaldowntime.
Anarcflashisasustainedarcingcurrentthatpropagatesthroughtheconductiveplasma
createdbythebreakdownofagaseousdielectricmedium,typicallyair.Giventheright
conditions,thecurrentwillcontinuetoflowunabateduntilinterruptedbyanupstream
overcurrentprotectivedevice.Sucharcsreleaseenormousenergy,resultinginhigh
temperatures,soundlevels,pressuresandtheejectionofhighspeedmoltendebris.Arc
flashesandblastsinelectricaltransmission,distributionandutilizationsystemsarealeading
causeofinjuriesandfatalitiestoelectricalworkersandareestimatedtocosthundredsof
millionsofdollarseachyearindowntimeanddamagedequipment.
Asaspecifichazardtoelectricalworkersthatneededtobeactivelymanaged,arcflashes
andblastswerelargelyignoreduntilthe1985publicationofRalphLeesseminalpaperArc
Flash:TheOtherElectricalHazard,whichpresentedthefirsttheoreticalmodelfor
predictingincidentenergiesworkerscouldbeexposedtoasafunctionofarcingtime,
availableshortcircuitcurrentanddistancefromthearc.
Thismodelandmuchoftheensuingresearchintopredictingarcflashenergiesisspecificto
alternatingcurrent,however.TheempiricallyderivedequationsintheInstituteofElectrical
andElectronicsEngineersGuidetoPerformingArcFlashHazardCalculations(IEEE
Standard1584)arespecifictoarcsinACsystems.Currently,noconsensusstandardexistsfor
calculatingarcenergiesinDCsystems.However,the2012editionoftheNationalFire
ProtectionAssociationsStandardforElectricalSafetyintheWorkplace(NFPA70E)
referencestwopapersthatoffertheoreticalandsemiempiricalmethodsforestimatingDC
arcenergy.
Theprimarygoalofmyprojectwillbetwofold:first,toperformacomparativeevaluationof
theconsensusstandardsforpredictingarcenergyonACsystemsasconductedonawind
farm,andtwo,toperformacomparativeevaluationofthetwomethodsgiveninthe2012
NFPA70EforpredictingDCarcenergyasconductedonaphotovoltaicarray.
Contents
Abstract.........................................................................................................................................................2
ListofFiguresandTables..............................................................................................................................4
Introduction..................................................................................................................................................5
ArcsinACSystems........................................................................................................................................6
ShortCircuitCurrentAnalysis...................................................................................................................7
Utility...................................................................................................................................................10
Transformers.......................................................................................................................................11
Turbines..............................................................................................................................................11
Conductors..........................................................................................................................................11
ArcEnergy...................................................................................................................................................15
LeesMethod..........................................................................................................................................15
IEEE1584Standard.................................................................................................................................16
Conclusions.................................................................................................................................................23
Plansforspring,2013Semester.................................................................................................................24
ArcsinDCSystems..................................................................................................................................24
References..................................................................................................................................................25
AppendixA:Abbreviations.........................................................................................................................26
AppendixB:ExpectedProjectCosts...........................................................................................................26
AppendixC:ProjectOutlines......................................................................................................................27
Acknowledgements.....................................................................................................................................29
List of Figures and Tables
SymmetricalShortCircuitCurrent............7
AsymmetricalShortCircuitCurrent..7
ShortCircuitCurrentDCComponent..8
SystemDiagram..9
ShortCircuitCurrentDiagram..9
PerUnitImpedanceNetwork.13
VectorDiagramsforLeesMethod.15
ArcingCurrentvs.BoltedFaultCurrent<1000V..17
ArcingCurrentvs.BoltedFaultCurrent>1000V17
IncidentEnergyvs.BoltedFaultCurrent<1000V18
IncidentEnergyvs.BoltedFaultCurrent>1000V18
IncidentEnergyDiagram20
TimeCurrentCurvesforXFMRfuseandCollectorCircuitRelay..21
TimeCurrentCurvesofXFMRFuseandTurbineCircuitBreaker..22
Introduction
Thesystemofelectricaltransmission,distributionandutilizationequipmentthatwe
informallyrefertoasthegridconstitutesanenormousandenormouslycomplexnetwork
thatissubjecttomanytypesoffailuresandhencemusthaverobustandwelldesigned
protectionsystems.Thesesystemsmustfulfillthedualfunctionofkeepingthenetworkup
andrunningasconsistentlyandreliablyaspossibleaswellasprotectingthepeoplethat
maintainandrepairit.Findingthebalancebetweenreliabilityandsafetycanbedifficult,and
toooftenthesegoalsareconsideredtobecontradictory.
Becausethecostsofdowntimeforconsumersofelectricitycanbeextremelyhigh,thereis
enormouspressureonelectricalworkerstoperformtheirdutiesonenergizedsystems,work
thatisextremelyhazardousevenforwelltrainedandqualifiedpersonnel.Clearly,amajor
hazardisthepotentialforinjuryduetoelectricshock,butthereisalsothepotentialfor
seriousinjuryfromelectricalarcflashesandblaststhatcanoccurwhenanunintentional
lowimpedancepathbetweenphaseconductorsorbetweenaphaseconductorandground
iscreated.Ifalowimpedancefaultescalatesintoasustainedelectricarc,tremendous
energiescanbereleasedintheformofintenseheatandconcussiveblasts.Manyofthese
typesoffaultsaretriggeredbyelectriciansandlinemenworkingonenergizedequipment,
workthatnecessarilyinvolvesbeingincloseproximitytothearcandwhichoftenresultsin
devastatingburns,impactsfromexplosiveblasts,fallsandahostofotherequallyserious
injuries.
Untilrelativelyrecently,injuriesfromelectricshockwereconsideredtobemostcommon
andserioushazardfacedbyelectricalworkers,butitisnowrecognizedthatmanyofthe
burnssufferedbyvictimsofelectricalaccidentsthatwereonceascribedtobeingpartofthe
conductionpathofelectriccurrentareactuallycausedbyexposuretoarcflashesandblasts.
Whilethestatisticsonhowmanyinjuriesandfatalitiesarecausedbyshockvs.arcflashes
areamatterofsomedebate,itisgenerallyacceptedthatarcflasheventsarenot
uncommonandconstituteasignificantportionofthe320deathsand4000majorinjuries
thatarecausedbyelectricalaccidentsonaverageeachyearintheUS[1].Whilethecostsof
individualaccidentsvarywidelyandaredifficulttocalculate,electricalaccidentsoftenhave
highercoststhanothertypesofaccidents,andonestudybytheElectricPowerResearch
InstituteinPaloAltoestimatestheaveragetotaldirectandindirectcoststobeashighas15
milliondollarspercase[1].
Further,reducingtheintensityofanarcflashprotectsequipmentaswellasworkers.Ahigh
energy,longdurationarcingfaultinsidethewiringcompartmentofanoilfilledtransformer,
forexample,willtypicallydestroythewiringlugsorpaddles,thefeederconductors,andcan
evenrupturethetank,leadingtothecompletelossofaveryexpensivepieceofequipment,
aswellascostsduetosystemdowntimeandlaborforrepairs.Lessenergeticarcswill
generallycauselessdamage,andsystemswithlowcalculatedincidentenergiesareeasierto
properlymaintainwithlessdowntime,sincethepersonalprotectiveequipmentthatmust
bewornisproportionatelylessburdensome.
Whileshockhazardsmustmostlybemitigatedthroughworkpracticestrainingand
permitting,personalprotectiveequipmentrequirements,andthelikearcflashhazardscan
oftenbesignificantlyreducedbymakingjudiciouschangestotheelectricalsystemitself.To
correctlyidentifywhereandhowimprovementscanbemade,onemustbeabletoestimate
withsomedegreeofaccuracythepotentialarcflashintensityatallpointsintheelectrical
systemwhereelectricalmaintenanceorrepairislikelytooccur.
Arcs in AC Systems
Freeburningarcsinopenairarechaotic,complexandunpredictableinnature,difficultto
accuratelymodelandaretypicallyevaluatedfromablackboxperspective.Themost
commonlyusedtheoreticalmodelforpredictingarcenergyisLeesmethod,whichrelieson
basiccircuittheoryandthemaximumpowertransfertheorem.DevelopedbyRalphLeein
his1982paperTheOtherElectricalhazard:ElectricArcBlastBurns,Leesmethod,along
withvariousbutrelativelyminorrefinements,wastheacceptedmethodologyuntilthe
publicationoftheIEEE1584Standardin2002,whichpresentedempiricallyderived
equationsbasedonextensivelaboratorytestdata.ItshouldbenotedthatLeesmethodis
stilltheconsensusstandardwhenthesystemtobeevaluatedfallsoutsideofthescopeof
the1584standard.
Thedifferencebetweenthetwomethodswillbeexploredindetaillater,butsinceboth
methodsrelyonaccuratelycalculatingthemaximumshortcircuitcurrent(SCC)availablein
theelectricalsystem,adiscussionofshortcircuitstudyprocesswillbehelpfulhere.Itis
generallyacceptedthatthevastmajorityoflinetolineorsinglelinetogroundfaultsina
threephaseACsystemwillquicklyescalateintothreephasefaultsduetothepropagationof
arcingcurrentthroughtheconductiveplasmaandvaporizedconductivematerialcreatedby
theinitialfault.Theenergyradiatedbythearcisafunctionofthecurrentthroughthearc,
whichinturndependsontheontheSCCavailableatthefaultpoint.
Becauseofthisloadindependence,wefindthattheequivalentimpedancelookingbackinto
atypicalsystemfromthefaultpointispredominantlyinductive,sothefaultcurrent
waveformwilllagthevoltagebycloseto90degrees.Thismeansthatthesymmetryofthe
currentwaveformaboutthezeroaxiswilldependonthepointinthevoltagecycleatwhich
thefaultoccurs.Ifthevoltageisatitspeakvaluewhenthefaultoccurs,thecurrentwillstart
atzeromagnitudeandfollowthevoltagedownthroughthezerocrossingandbe
symmetricalaboutthezeroaxis.If,however,thevoltageisatzerowhenthefaultoccurs,
thecurrentwillagainstartatzero,butsinceitmustfollowthevoltageby90degrees,the
currentwillriseabovethezeroaxisandvaryasasinusoidaboutaDCoffsetvalue:
Figure1:SymmetricalSCC[2]Figure2:AsymmetricalSCC[2]
Itiscommontoanalyzetheasymmetricalfaultcurrentasthesumofasymmetricalsinusoid
andaDCcomponentthatdecaysasafunctionofthecircuitsX/Rratio:
Figure3:SCCasSinusoidandDCComponent[2]
ThemaximumasymmetricalSCCcanbefoundtheoreticallybyanalyzingthetransient
responseofanequivalentseriesRLcircuitjustafterthetimeoffaultinception.This,
however,isarelativelycomplicatedcalculationinarealsystemandsimplifiedmethodshave
beendevelopedwherethesteadystatesymmetricalSCCisadjustedbyatableofmultiplying
factorsbasedontherelativeX/RratioatthefaultpointandthetestX/Rratioofthe
interruptingdevice.ThesemethodsaredescribedinseveralIECandANSIstandardsfor
differentinterruptingdevices,mostnotablyANSIC37.04andANSIC37.010forhighvoltage
circuitbreakersratedonasymmetricalbasis.
However,thesestandardsareintendedtoguidetheproperselectionofaninterrupting
devicecapableofopeningunderworstcasefaultconditions,forwhichthemaximum
asymmetricalSCCmustbeknown.Forthepurposesofestimatingarcenergy,the
symmetricalRMSSCCisthevaluewewant,sincethepowerinanarcisfoundastheproduct
oftheeffective(RMS)voltageacrossthearcandtheeffectivecurrentthroughthearc.The
currentthroughthearcwillbelowerthantheavailablesymmetricalSCCduetoarc
impedance,butwebeginbyfindingthesymmetricalSCCandadjustingitaccordingly.
Thisisaccomplishedbyreducingthesystemtoaperunitimpedancenetworkatacommon
baseMVAandacommondrivingvoltage,afterwhichtheavailablesymmetricalSCCateach
nodecanbecalculatedsimplyasI=V/Z,whereVisthesystemdrivingvoltageandZisthe
equivalentimpedanceofthenetworkasseenbythefaultpoint.Asanillustrativeexample,
wewilluseasmallsystemconsistingoftwo2MWwindturbinesfeedinga115KVsubstation,
allprotectedbyavarietyofovercurrentdevices.Thefollowingdiagramsdepictthesystem
schematic(Figure4)andthesystemwiththeavailablesymmetricalRMSshortcircuitcurrent
shownascalculatedbyESAsEasyPowersoftware,basedonactualdevicedata(Figure5):
Figure4:SystemDiagram
Figure5:ShortCircuitCurrentDiagram
Toverifytheabovevaluesandseehowtheyareobtainedwewillchooseabaseof
10MVAandaperunitdrivingvoltageof115KV/115KV=1.
Nextweneedtofindthebaseimpedanceateachvoltagelevelinthesystem.Fromthe
relationshipsV=IZandS=IV,wecanobtainanduseZ=V2/Stofindthebaseimpedances:
Z115kv=(115,000)2/10MVA=1322.5
Z34.5KV=(34,500)2/10MVA=119.03
Z0.6kv=(600)2/10MVA=0.036
Next,eachelementinthesystemmustbeconvertedtoperunitimpedanceusingthe
relationshipZpu=Zactual/Zbase:
Utility
SCCcontributionsfromtheelectricutilitythatfeedsasystemmustbeobtainedinorder
accuratelycalculateSCCvaluesinthesystem.Acommonmistakeinshortcircuitanalysisis
toassumethattheutilityisaninfinitebusanidealvoltagesourcewithzerointernal
resistancecapableofsupplyingunlimitedSCCundertherationalethatthisisaworstcase
approach.Theproblemisthatoverestimatingtheshortcircuitcurrentataparticularnodein
asystemcanleadtobelievingthatagivenovercurrentprotectivedevicemaydetectand
clearafaultfasterthanitwouldinreality.Manyovercurrentprotectivedeviceshave
inversetimecurrentcharacteristics,meaningthatthehigherthecurrentdetectedis,the
fasterthedevicewillopen.Overestimatingtheavailableshortcircuitcurrentcanleadto
significantlyunderestimatingthedeviceopeningtimeandhence,theincidentenergy.
TheSCCautilitycanprovideataparticularpointiseasilyobtainedfromtheutilityitself,and
istypicallygivenasamperesatparticularvoltageandX/Rratio.Inourexample,theutility
contributionis3003amperesat115KVwithanX/R=6.08.Tousetherelationshipabove,we
needtofindtheactualimpedancelookingbackintotheutility:
=22.1
0.0167
Thenweneedtoseparatetheimpedanceintoitsrealandimaginaryparts:
tan
cos 80.66
sin 80.66
6.08
80.66
0.00271
0.0165
Hencetheperunitutilityimpedanceis:Zpu=0.00271+j0.0165
Transformers
Practicalpowertransformersaredesignedsuchthatthereactanceismuchlargerthanthe
resistanceofwirethatmakesupthetransformercoils.Transformerimpedancesaregivenby
themanufacturerinperunitvalues(%Z),oftenwithoutanaccompanyingX/Rratioorpower
factor,sothisimpedancecanbeassumedtobepurelyreactiveforthepurposescalculating
SCC.The%Zisvalidfortherated(nameplate)VAandmustbescaledtothearbitrarily
chosenbaseMVA:
TX1
%
9.04 10
.
100
100 48
XFMR1andXFMR2
%
5.66 10
.
100
100
2
Turbines
Theimpedanceofthewindturbinelikeanygeneratorwillbegivenasthreevalues:Xd
(subtransientreactance),Xd(transientreactance)andXd(synchronousreactance).Sincewe
areinterestedinmaximumsymmetricalSCC,wewanttouseXd,thevaluethatwillgivethe
SCCinthefirsthalfcycle.Inpractice,thegeneratorwillbemuchlikeatransformerandhave
anX>>R,eventhoughthenameplatemayprovideanX/Rorpowerfactor.Inourexample,
theturbinenameplategivesanX/R=24.97,sowewillneglecttherealpartoftheimpedance.
Theperunitimpedancemustagainbescaledbythebaseimpedance:
%
16.7 10
.
100
100
2
Conductors
TheimpedanceoftheconductorsinthesystemcansignificantlyreduceSCCatthefault
pointandmustbeconsidered.Thisisparticularlytrueinanexamplelikeours,awindfarm,
wherethecollectorcablesbetweenturbinescanbethousandsoffeetlong.Tofindtheper
unitconductorimpedances,weneedtofirstfindtheactualimpedance,whichwecanobtain
fromthemanufacturerorfromareferencetable.Wellusetable9fromtheNational
ElectricalCode:
1000kcmil:0.019+j0.037(ohms/1000)
500kcmil:0.029+j0.039(ohms/1000)
4/0awg:0.091+j0.041(ohms/1000)
Forthe4/0and1000kcmillines,weonlyhave1conductorperrun,witheachrun1000long,
sowecansimplydividetheaboveimpedancesbythebaseimpedanceat34.5KV:
1000kcmil:
4/0:
.
.
Forthe500kcmillines,wehavefourparallelconductorsperrunatatotalof350long,so
theimpedancesofeachconductorwilladdinversely:
1
4
1
1
0.00254
0.00341
0.35 0.029
0.039
Thenwedividetheactualimpedancebythebaseimpedanceat600voltstofindtheperunit
impedance:
0.00254
0.00341
.
.
0.036
Finally,wecandrawtheperunitimpedancenetwork:
Figure6:perunitimpedancenetwork
TofindtheSCCatnodeA,wefindtheequivalentimpedanceatthatpointastheparallel
combinationofthethreebranches:
0.0708
1.213
/
0.0700
1.213
0.00363
0.03598
/
1
1
1
1
.
.
Then,wecanfindtheperunitSCC:
1
.
2.875
29.15 29.30
0.00335
0.03397
TofindtheactualSCCat34.5KV,weneedtocalculatethebasecurrent:
10
167.35
3
3 34.5
Andfinally,theactualSCCatnodeAistheproductofthebasecurrentandtheperunit
current:
WeseefromtheSCCdiagramabovethatthevaluecalculatedbytheEasyPowersoftwareat
thatnodeis4917Awithanangleof83.2,differencesthatareprobablyduetorounding
errorsinthehandcalculations.TheSCCatanyothernodeinthesystemcanbefound
similarly,thoughitisclearlyeasiertousethesoftware!However,itiscriticaltobeableto
assesstheSCCvaluesgivenbythesoftwaretoverifytheaccuracyoftheequipmentdata.If,
forexample,theimpedanceofTX1wereinadvertentlyenteredas3.04%ratherthan9.04%,
theSCCatnodeAwouldbe7,353A.
Arc Energy
Theprimarydifficultyinpredictingtheenergyradiatedbyanelectricarcisaccurately
determiningarcimpedance.Sincethearcpowerorenergypertimeistheproductofthe
drivingvoltageandthearcingcurrent,thearcimpedancewouldneedtobeknownto
calculatethedifferencebetweentheboltedfaultcurrentandthearcingcurrent.In
applicationswherearcsaredeliberatelyformed,suchasweldersandhighintensity
dischargelighting,thearcgap,pressureandchemicalcompositionofthedielectricis
preciselyknowntoachievethedesiredeffect.Intheinadvertentarcsthatweareinterested
in,struckbyacarelesslydroppedtoolorbyoperatingamalfunctioningswitch,these
parametersmayvarygreatly,andaccuratelypredictingthearcimpedanceisnearly
impossible.
Lees Method
Inordertoworkaroundthisdifficulty,RalphLeebeganwiththeassertionthatthesources
ofSCCinapracticalelectricalsystemareprimarilyinductiveandthattheimpedanceofan
arcisprimarilyresistive,suchthatthesourcevoltageandarcingcurrentwouldbeshiftedby
90degrees,withthevoltageacrossthearcnecessarilybeingphasewiththearcingcurrent.
Fromthis,thevoltagedropsacrossthesourceinductanceandthearcresistancecanbe
drawninquadrature:
Figure7:VectorDiagramsforLeesMethod[3]
Fromthis,wecanseethatthemaximumpowerwillbetransferredtothearcunder
condition2,whenEs2=Ea1orXsource=Rarc,accordingtothemaximumpowertransfertheorem.
Fromthevectorsforcondition2,wecansolveformaximumpower:
cos 45
0.707
,andbythesamelogicI2=0.707Isc
HencePmax=V2*I2=0.5*V0*Ibf=0.5*VAbf
Then,treatingthearcasasphericalradiantheatsource,thefractionofenergyabsorbedby
asphericalsurfaceperunitareawillbeinverselyproportionaltothesquareofthedistance
betweenthearcandthatsurface:
2.142
Where:
EisincidentenergyinJ/cm2
MVAistheboltedfaultSCC
tisarcingtimeinseconds
Disdistancefromthearcinmm
Thismethodhasseverallimitations,themostsignificantofwhichisthatatthemaximum
powerpoint,wewillalwayshavethatthevoltageacrossthearcandthecurrentthroughitare
equalto70.7%ofthesystemvoltageandtheboltedfaultcurrent,withtheresultthatpredicted
arcenergywillincreaselinearlywiththesystemvoltageandtheavailableboltedfaultcurrent.
ThisisproblematicgiventhenonlinearnatureofarcVIcharacteristics,wherepastacertain
point,thevoltageacrossthearcisprimarilyafunctionofarclength,andcurrentthroughthe
arcmayincreasealmostindependentlyofarcvoltage.Afurtherissueisthatthemaximum
powermethoddoesnotprovideuswithawaytoaccuratelypredictthearcingcurrent,which
maybesignificantlydifferentthanthe70.7%ofboltedfaultcurrentpresentatmaximumpower
point.
currentdeliveredbythesystemunderavarietyofconfigurationandprovidedequationsfor
predictingarcingcurrentasafunctionofsystemvoltage,arcgapandboltedfaultcurrent
(equationsforarcingcurrentandincidentenergyarereproducedintheappendix).Thetest
resultsdemonstratedthatarcingcurrentasapercentageofboltedfaultcurrentincreased
quicklyasthesystemvoltagewasincreased.Belowisaplotofarcingcurrentvs.boltedfault
currentusingthe1584equationatthefivevoltagesunder1000Vthatthegrouptestedat,with
allothervariablesheldconstant:
Figure8:ArcCurrentvs.BoltedFaultCurrentfor<1000V
At208V,weseethatthearcingcurrentisroughly20%ofboltedfaultcurrent,whileat600V,
arcingcurrentiscloserto50%boltedfaultcurrent.
Thegroupalsotestedat2.3KV,4.16KVand13.8KV,findingthatatthesevoltages,arcing
currentwasalmostentirelyafunctionofboltedfaultSCC.Theequationprovidedforarcing
currentisthesameforallthreesystemvoltagesandweseethatthearcingcurrentisalmost
equaltotheboltedfaultcurrent:
Figure9:ArcCurrentvs.BoltedFaultCurrent>1000V
WhatthismeansisthatwhenwecompareIncidentEnergycomputedbythetwomethodsover
arangeofboltedfaultcurrentvalues,weseethatLeesmethodwillbeconsistentlybutnot
excessivelyconservativeatthelowersystemvoltages:
Figure10:IncidentEnergyvs.BoltedFaultCurrent<1000V
For1584Curves:ArcGap=32mm,ArcTime=0.1sec,CalcFactor=1.5,Dist.Exp.=1.473,WorkingDist.=18
ForLeesCurves:ArcTime=0.1sec,WorkingDist.=18
However,above1000volts,Leesmethodprovidesveryconservativeresults.Intheplot
below,weseethatthe1584equationsreturnthesameresultsforallthreeofthetest
voltages,whilethevaluespredictedbyLeesmethodwillcontinuetoincreaseasadirect
functionofincreasingsystemvoltage:
Figure11:IncidentEnergyvs.BoltedFaultCurrent>1000V
For1584Curves:ArcGap=153mm,ArcTime=0.1sec,CalcFactor=1,Dist.Exp.=0.973,WorkingDist.=38
ForLeesCurves:ArcTime=0.1sec,WorkingDist.=38
Thisisproblematicbecausethe1584grouponlytestedupto13.8KV,andhenceanysystem
withvoltagesabovethislevelmustbeevaluatedusingLeesmethod.Whatthismeansis
thatelectricalworkersperformingenergizedworkonsystemsabove13.8KVarelikely
donningPersonnelProtectiveEquipment(PPE)ratedfarhigherthanwhatmightactuallybe
necessarytopreventanincurableburn.PPEratedfor40cal/cm2isheavy,uncomfortable,
restrictsmotionandvisibilityandhencelowersworkerproductivity,andcouldconceivably
increasethelikelihoodofanaccidentoccurring.
Itisalsosometimesthecasethatfacilityorprojectmanagerschoosetoevaluateincident
energyusingthemaximumpowermethod,reasoningthatthiswillalwaysgiveaworstcase
value.Inasystemwheretheovercurrentinterruptingdevicesareinstantaneouscircuit
breakersordefinitetimerelayswherethecurrentatwhichthedevicewilltripis
independentoftimethisisprobablyagoodnotion;theaboveplotsprovideaprettygood
indicationofthedifferencesonewouldexpecttoseebetweenmethods.Giventherelatively
minormarginbywhichLeesmethodoverestimatesincidentenergyonlowvoltagesystems,
onecouldcertainlyusethismethodwithahighdegreeofconfidencethatelectricalworkers
wearingPPEratedtowithstandtheestimatedenergywouldbeadequatelyprotected
withoutthatPPEbeingunnecessarilyburdensome.
However,aspreviouslymentioned,Leesmethoddoesnotprovideawaytopredictarcing
current.Insystemswhereovercurrentdeviceshaveinversetimecurrentcharacteristics
wherethetimeittakesforthedevicetoopendecreasesasthecurrentthroughthedevice
increasesanaccurateestimateofthedeviceclearingtimeisessentialtoproperlypredict
incidentenergy.SinceLeesmethodcalculatesthemaximumpowerinanarcusingthe
boltedfaultSCC/squarerootof2,thisistheamountofcurrentwemustlogicallyassumeis
presentinthearc,andhenceistheamountofcurrentsensedbytheclearingdevice.Asthe
1584testresultsshow,thearcingcurrentisastrongfunctionofsystemvoltageunder
1000V,andonthesesystems,Leesmethodmayvastlyoverestimatethearcingcurrentthat
ispresent.Thiswouldleadtounderestimatingthearcclearingtimeandhencethearc
energy.
Returningtoourpreviousexampleofthesmallwindfarmandcalculatingtheincident
energyusingbothLeesmethodandthe1584equations,wegetvaluesshowninthe
followingdiagram:
Figure12:IncidentEnergy;allvaluesforopenair
SincetheportionsofthesystemfromtheutilityuptotheprimarysideofXFMR1and
XFMR2areat115KVand34.5KV,andhenceareoutsidetherangeofthe1584model,the
onlyvaluesshownarethosegivenbyLeesMethod(thefaultclearingtimeisgivenbythe
pickuptimeofthedifferentialrelaynotshownplustheopeningtimeofthe115KVmain
breaker.Givenwhatwehaveseenatthetestedmediumvoltages(2.3KV,4.16KV,13.8KV),it
islikelythatthesevaluesarequiteconservative.Itis,however,difficulttoquantifyjusthow
conservative,andaprudentadherencetobestworkpracticesdictatesthatelectrical
workersdonPPEratedforthelevelsshown.
Betweenthe600VsideoftheXFMRsandtheMCCBbreakers,weseethatLeesMethod
predictsincidentenergywillbeapproximately25%lowerthanwhatisgivenbythe1584
equations.IncidentenergyatthesenodesisdeterminedbythetripcharacteristicsoftheFS
1andFS2fuses.Asseeninfigure13below,the1584equationspredictthatthearcing
currentforfaultsatthesenodeswillbeapproximately18KA,while70.7%oftheboltedfault
SCCavailableontheXFMRsecondaryisapproximately21.4KA.Duetotheinversetime
currentcharacteristicsofthefuse,thisrelativelysmalldifferenceresultsinLeesMethod
predictingthatthefusewillclear8cyclesfasterthanthe1584prediction,resultinginthe
lowerincidentenergyvalues:
Figure13:TimeCurrentCurvesforXFMRfuseandCollectorCircuitRelay
Attheterminalsoftheturbine,thesituationispotentiallymuchworse.Here,the
InstantaneouspickupoftheMCCB1breakerthelowerrightportionofthepurplecurveis
setto14KA,meaningthatthedeviceshouldopenimmediatelyifanycurrentissensed
abovethatlevel.However,asisalwaysthecase,thereissomeuncertaintyhere:the
manufacturergivesarangeofcurrentsatwhichthedevicehasbeenobservedtooperate
duringtesting.Thisrangeisgivengraphicallybythewidthofthetripcurveineachregionof
operation.Further,thisrangeisonlyvalidfordevicesintheconditiontheywereintheday
theyleftthefactory.Oldandpoorlymaintaineddevicesareconsiderablymore
unpredictable.
AsseeninFigure14below,thecalculatedarcingcurrentisapproximately14.5KA,while
70.7%oftheboltedfaultSCCseenbyMCCB1is16.9KA.Giventheuncertaintydescribed
above,EasyPowercorrectlyassumesthatthearcingcurrentmaynotbesufficienttotripthe
breaker,sothetotalarcclearingtimeisbasedontheShortTermsettingof0.3secondsplus
thebreakeropeningtimeof0.04seconds.UsingLeesMethodestimatingthearcing
currenttobe70.7%oftheboltedfaultSCCwewouldthinkthatthereissufficientarcing
currenttooperatethebreakerinstantaneouslyintheeventofafault,resultinginan
incidentenergyvaluefully10timeslowerthanthatpredictedwhenusingthe1584
methodology.
Figure14:TimeCurrentCurvesofXFMRFuseandTurbineCircuitBreaker
Ideally,facilitymanagersandengineerswouldsettheInstantaneouspickupsontheirdevices
tothelowestpossiblesettingthatstillpermitsnormal,reliableoperationoftheirsystem,
andthesmallthesmalldifferenceinestimatedarcingcurrentshowninthisexamplewould
notberightonthetransitionzonebetweentripregions,butbewellabovethe
Instantaneoussetting.Wewouldseeamuchsmallereffectifweassumethatbothcurrent
valuesaresufficienttooperatethedeviceintheInstantaneousregion.Withatotalclearing
timeof0.06seconds,theestimatedincidentenergyusingthe1584equationsis2.6cal/cm2
comparedtothe3.1cal/cm2usingLee.
Inpractice,however,itisoftenthecasethattheinrushcurrentsnecessarytostart
transformersandmotorsarequitelargeanddifficulttocalculateprecisely,andthemargins
canindeedbeasnarrowasshowninthisexample.Thekeytakeawayhereisthatwecannot
assumethatLeesmethodwillalwaysprovideconservativeresults,andfurther,thatas
muchroomaspossibleneedstobegivenbetweenrelaysettingsandarcingcurrent
regardlessofwhatmethodisused,sincethereisconsiderableuncertaintyinboththe
clearingdevicethresholdandtheestimationofthearcingcurrent.
Conclusions
Muchoftheforegoinganalysisiswellknownintheindustryanddoesnotpresentnew
methodsforconductinganArcFlashHazardAnalysisoutsidethosegiveninthetwo
consensusstandardsIhaveevaluated.Itdoes,however,amplydemonstratethecarethat
mustbetakenindecidingwhatmethodtoapplyaftercarefulconsiderationofsitespecific
conditions.SinceLeesmodelprovidesatheoreticalmaximumarcenergybutnomethodfor
findingarcingcurrent,theempiricalequationsforbotharcenergyandarcingcurrentare
likelytogivevaluesthatmostcloselyapproximatereality,butitiscriticaltorealizethatthe
valuesobtainedareindeedapproximations.
Further,thereisconsiderablevalueintakingacloselookatthedisparitybetweenthe
maximumpowermethodandempiricaldatainACcaseasanentrypointtoinvestigating
howwemightpredictarcenergiesinDCsystems.ExistingresearchhasshownthattheVI
characteristicsofarcsinACandDCsystemsarenonlinear,behaviorthatLeesmaximum
powermethoddoesnotaccountfor,soitisreasonabletopredictthatthesituationmaybe
muchthesameinDCsystems,andthatmoreaccurateresultswillbereturnedbyan
empiricalmodel.
Arcs in DC Systems
Asofthiswriting,thereisnoconsensusstandardforcalculatingarcflashhazardsonaDirect
Currentsystem.However,the2012editionoftheNationalFireProtectionAssociations
StandardforElectricalSafetyintheWorkplace(NFPA70E)[5]referencestwopapersthat
offertheoreticalandsemiempiricalmethodsforestimatingDCarcenergy.Thetheoretical
methodisbasedonapaperbyDanDoan[6]thatpresentsamaximumpowerapproachto
determiningarcenergyverysimilartothatgivenbyLee.Thesemiempiricalmethod,
presentedinapaperbyDr.RavelAmmerman[7]andseveralcolleagues,isbasedonasurvey
ofextanthistoricaldatagoingbackover100years.
Myplanforthecomingsemesteristoperformacomparativeevaluationofthesetwo
methodsinmuchthesamefashionasforLeesmethodandtheIEEE1584model,andapply
bothmethodsinattemptingtopredictthearcenergiesthatmayexistontheDCbusofa
photovoltaic(PV)array.Thereare,however,twocomplicatingfactorsspecifictoapplying
eithermodeltotheDCbusofaPVsystem.First,theresistanceofaPVcellisnonlinearin
nature,hencethecurrentthecellscanprovidetoagivenfaultpointmaychange
significantlyandrapidlyastheloadresistancegoesfromthatoftheinvertertothe
resistanceatthefaultpoint.Inotherwords,thesourceresistancethatofthePVcellsis
notfixedtothesameextentastheresistanceofastringofbatteriesis.
Second,overcurrentprotectiononPVarrayisfundamentallydifferentthaninatypicalAC
system.TheshortcircuitcurrentaPVcellcansupplyisonlyslightlyhigherthanthecurrentit
willproduceatstandardtestconditionswithaloadmatchedtothecharacteristicresistance
ofthearray.Thismeansthatthefusesprotectingindividualstringsinthearrayareintended
primarilyforoverloadprotectionasdistinctfromfaultprotection,andmaynotserveto
interruptfaultsatall.Identifyingthefaultclearingdevicesandtheirassociatedclearing
timesandaccuratelypredictingthearcingfaultcurrentwillbetheprimarychallengesin
completingthestudy.
References
[1]ChicagoElectricalTraumaResearchInstitutewebsite[Online].Availableat:
http://www.cetri.org/statistics.html
[2]GeneralElectricApplicationInformation,ShortCircuitCurrentCalculations[online].Availableat:
http://www.geindustrial.com/publibrary/checkout/GET3550F?TNR=White%20Papers|GET
3550F|generic
[3]TheotherElectricalHazard:ElectricArcBlastBurns,RalphH.Lee,1982
[4] GuideforPerformingArcFlashHazardCalculations,IEEE15842002,2002.
[5]StandardforElectricalSafetyintheWorkplace,NFPA70E,2012
[6] ArcFlashCalculationsforExposurestoDCSystems,DanielR.Doan,2007
[7] DCArcModelsandIncidentEnergyCalculations,Ammerman,Gammon,Sen,Nelson,2009
Appendix A: Abbreviations
IEEE:InstituteofElectricalandElectronicsEngineers
NFPA:NationalFireProtectionsAssociation
SCC:ShortCircuitCurrent
IE:IncidentEnergy
RMS:RootMeanSquare
TCC:TimeCurrentCurve
XFMR:PowerTransformer
Travel:
2tripstoAlamosafordatacollection.255mileseachway
At$0.555/mi(gsa.gov)
255*4*$0.555=$566.10
Researchpapercosts:
$10eachfromIEEEexplore
Estimate6totalfor$60
Edaysposter:
$20
Totalestimatedexpenses:$646.10
ProjectOutline,Revision#1,September9,2012
1)
2)
3)
4)
5)
ResearchstandardmethodsforcalculatingarcenergiesinACsystems.Identifyandsummarizesignificantdifferencesusingasmall
windfarmasarealworldexample.EvaluatewhythestandardmethodologiesdontapplytoDCsystems:
a.
Leesmethod
b. IEEE1584
c.
NFPA70Etablemethod
Deliverables:
1) Shortcircuitcurrentandpotentialarcenergyvaluesforsmallwindfarmusingabovemethods
Estimate60hoursOctober15,2012
2) Reportevaluatingdifferencesbetweenmethods,strengthsandshortcomingsofeachandanexplanationofwhythey
cannotbeappliedtoDCsystems
Estimate20hoursOctober29,2012
SeekoutandevaluateexistingresearchintomethodsforcalculatingarcenergiesonDCsystems:
a.
IEEEworkinggroup
b. ESA
c.
DanDoanpaper
Deliverables
1) Reportsummarizingandevaluatingassumptionsandscopeofexistingresearch
Estimate40hoursNovember30,2012
Developtheaboveintoanalgorithmthatcanbeappliedspecificallytoworkeraccessiblesystemnodesonautilityscalephotovoltaic
array
Deliverables
1) Equationsforcalculatingarcenergyasfunctionofsystemvoltage,availableshortcircuitcurrent,arcimpedanceandarc
gap
Estimate60hoursFebruary28,2012
2) Reportexplainingtheprocessofdevelopingtheequationsandjustifyingtheassumptionsmade
Estimate20hoursMarch15,2012
VisitIberdrolaRenewablesSanLuissolarfarm(30MW):
a.
CollectData(OCPD,conductors,PVpanels,inverters,powertransformers,etc.)
Deliverables
1) Onelinediagramoffarm
Estimate20hoursMarch29,2012
ImplementequationsusingacombinationofExcelorMatlabandESAsEasyPowersoftwareandevaluatetheSanLuisSolarFarm:
Deliverables
1) Availableshortcircuitcurrentvaluesatallsignificantnodesinthesystem
Estimate10hoursApril8,2012
2) EquipmentDutyStudy
Estimate10hoursApril15,2012
3) TimeCurrentCurvesofOCPDfoundonfarm
Estimate10hoursApril22,2012
4) Potentialincidentenergyvaluesoverarangeoflikelyworkingdistances
Estimate10hoursApril29,2012
6)Iftimeallows,evaluateaboveresultsandofferthefollowing:
1) Recommendationsforsystemchangesorworkpracticestolowerpotentialincidentenergy
2) Recommendationsforsystemchangesorworkpracticestoreduceshockexposures
ProjectOutline,Revision#2,December7,2012
1)
CompletealgorithmforfindingiterativesolutiontoequationforarcVIcharacteristicbasedontheStokesand
OppenlanderdatagiveninDr.Ammermanspaper
2)
Deliverables
a. Matlabcodeandreportexplainingmethodologyandfitcomparisontoempiricalcurves
January30,2012
UnderstandresistanceofthePVcellandhowitchangesasafunctionofloadresistance.Thisiscriticalforapplyingthe
equationsfrompart1above.Ideally,IwouldliketousetheMaxPowerPointTrackingintheinvertertorapidlyvary
theloadresistanceseenbythePVarrayandrecordtheresults,butitisunlikelyIwouldbeallowedtodoso.Iwill
probablyhavetousethecharacteristicresistanceasdefinedbyVmaxpower/Imaxpower.Thisshouldgiveworstcasevalue
3)
Deliverables
a. ReportjustifyingthevalueusedfortheequivalentresistanceofthePVarraylookinginfromthefaultpoint
February28,2012
VisitIberdrolaRenewablesSanLuissolarfarm(30MW):
b. CollectData(OCPD,conductors,PVpanels,inverters,powertransformers,etc.)
Deliverables
2) Onelinediagramoffarm
March29,2012
4)
ImplementequationsusingacombinationofExcelorMatlabandESAsEasyPowersoftwareandevaluatetheSanLuis
SolarFarm:
Deliverables
5) Availableshortcircuitcurrentvaluesatallsignificantnodesinthesystem
April8,2012
6) EquipmentDutyStudy
April15,2012
7) TimeCurrentCurvesofOCPDfoundonfarm
April22,2012
8) Potentialincidentenergyvaluesoverarangeoflikelyworkingdistances
April29,2012
Acknowledgements
IwouldliketothankMr.SimmsandDr.Collinsfortheirguidancethroughoutthesemester
andfortheirpatienceasIslowlylearnedthebackgroundmaterialnecessaryfor
understandingthissubject.IwouldalsoliketothankJohnKolakforhisconstantoptimism
andtirelessencouragementwhenitwasmostneeded.