GAININGVALUEFROMSMARTMETERDATA:POWERQUALITYANDOUTAGEEVENTANALYSISByValdamaE.JohnsonATHESISSubmittedtoMichiganStateUniversityinpartialoftherequirementsforthedegreeofElectricalEngineering-MasterofScience2016ABSTRACTGAININGVALUEFROMSMARTMETERDATA:POWERQUALITYANDOUTAGEEVENTANALYSISByValdamaE.JohnsonElectricutilitycompaniesaremakinginvestmentsinsmartgridtechnologiestoim-provethewaypowerisgeneratedanddeliveredtocustomers.Investmentssuchasthereplacementofelectromechanicalmeterswithsmartmetershasprovidedseveralincludingmoreaccu-rateelectricitybillsandtheabilitytoremotelyconnectordisconnectservice.Inadditiontothesesmartmeterscanalsobeusedtoaddressthelackofvisibilityintotheelectricdistributionsystem.Whilemanyutilitieshavevisibilityatsubstationsthroughsupervisorycontrolanddataacquisitionsystems,muchofthedistributionsystembeyondthesubstationremainsunmonitored.Theobjectiveofthisthesisistoshowhowutilitiescangainmorevisibilityintothedistributionsystembyanalyzingsmartmeterdata.Smartmeteroutageevents,whenintegratedintooutagemanagementsystems,canhelputilitieslocateoutagesfasterandrestorepowertocustomers.Inthisthesis,smartmeteroutageeventsarecomparedwithhistoricaloutageincidentsfromanout-agemanagementsystem.Theresultsofthecomparisonshowthatthereareseveralchallengestoovercomebeforetheseeventscanbeintegratedintooutagemanagementsystems.Thesechal-lengesincludeprocessingmomentarysmartmeteroutagesandensuringthattheelectricmodeliscorrect.Smartmeterscanalsobeusedtoidentifypowerqualityissuessuchasdrasticchangesinvoltagelevels.Anaveragevoltageoverave-dayperiodwasobtainedfor700,000smartmetersandroughly1%hadavoltageabovetheacceptabletolerancerange.Manyoftheseissueswerecausedbythefailureofadistributiontransformerorvoltageregulator.Byanalyzingsmartmeteroutageandvoltagedata,utilitiescanbemoreproactiveinaddressingcustomerandsystemissues.ACKNOWLEDGMENTSFirst,IhavetothankGODforalloftheblessingshehasgivenme.Withouthishelp,thereisnowayIcouldhavecompletedthisthesis.ThereweretimeswhereIwantedtogiveup,butwithGODallthingsarepossible.IwouldliketothankDr.PercyPierreandtheSloanEngineeringProgramforsupportingmethroughoutmytimehereatMichiganStateUniversity.Dr.Pierre,thankyouforbelievinginmewhenIdidnotbelieveinmyself.OnedayIhopetobeinapositionwhereIcanhelpothersthewayyouhavehelpedme.Iwouldliketothankmyadvisor,Dr.JoydeepMitraforallofhisguidanceandpatiencethroughoutthisprocess.Thankyouforbelievinginmeandwelcomingmeintoyourlab.Youradvicehasbeenextremelyvaluabletome.IwouldliketothankDr.SubirBiswasforhissugges-tionsandagreeingtoserveonmycommittee.Tomylabmates:thankyouforyourfriendship,encouragement,andsupport.IwouldalsoliketothankmycolleaguesatConsumersEnergyforprovidingmewiththeresourcesneededtoconductmyresearch.iiiTABLEOFCONTENTSLISTOFTABLES.......................................viLISTOFFIGURES......................................viiChapter1Introduction..................................11.1ResearchObjectives..................................21.2OrganizationofThesis................................3Chapter2Background..................................42.1HistoryofElectromechanicalMeters.........................42.2AdvancedMeteringInfrastructure..........................72.2.1SmartMeters.................................72.2.2CommunicationsSystems..........................92.2.3MeterDataManagementSystems......................102.3GeographicInformationSystems...........................102.4OutageManagementSystems.............................12Chapter3OutageEventComparison..........................133.1Introduction......................................133.2Methodology.....................................133.2.1ObtainingOutageData............................133.2.2DataProcessingwithPython.........................153.3Analysis........................................213.3.1Case1.....................................213.3.2Case2.....................................243.3.3Case3.....................................28Chapter4PowerQualityAnalysis...........................324.1Introduction......................................324.2Methodology.....................................344.2.1ServiceLinesandMeterTypes........................344.2.2ObtainingVoltageData............................354.3Analysis........................................374.3.1Case1.....................................374.3.2Case2.....................................414.3.3Case3.....................................45Chapter5Conclusion..................................47APPENDIX...........................................49ivBIBLIOGRAPHY.......................................52vLISTOFTABLESTable3.1:SmartMeterOutageandRestorationEventCodes................15Table3.2:FeederswithaHighPercentageofSmartMeters.................16Table3.3:OutageEventComparisonResults........................20Table3.4:OMSCustomersforIncidentNumber2884686..................22Table3.5:SmartMeterOutagesforIncidentNumber2884686...............22Table3.6:SmartMeterOutagesforIncidentNumber2900028...............24Table3.7:13MetersthatdidnotcommunicateforIncidentNumber2904879.......26Table3.8:MomentarySMOutagesforIncidentNumber2911524.............29Table4.1:VoltageRangeofMeterswithHighVoltage...................37Table4.2:MeterswithHighVoltageconnectedtoTransformer0812164402........40Table4.3:Top10FeederswithMetersbetween7and10%.................41viLISTOFFIGURESFigure2.1:Shallenberger'sAmpere-HourMeter[10]....................5Figure2.2:SinglePhaseWatt-HourMeter..........................6Figure2.3:SinglePhaseSmartMeter............................8Figure2.4:ElectricDistributionCircuitsinGIS.......................11Figure3.1:SampleOMSIncidentArchive..........................14Figure3.2:FlowchartofOutageEventComparison.....................19Figure3.3:GISMapforIncident2884686..........................23Figure3.4:GISMapforIncident2904879..........................25Figure3.5:GISMapforIncident2906187..........................27Figure3.6:GISMapforIncident2897071..........................28Figure3.7:GISMapforIncident2911524..........................29Figure3.8:GISMapforIncident2899212..........................31Figure4.1:SimpleForm2S..........................34Figure4.2:PISystemScreenshot...............................35Figure4.3:DistributionofMeterswithanAverageVoltage10%.............38Figure4.4:VoltageforMeter30173601.......................38Figure4.5:VoltageforMeter30172317.......................39Figure4.6:DamagedWindingsofTransformer1007343405................40Figure4.7:DistributionofMeterswithanAverageVoltagebetween7and10%......41Figure4.8:GISMapforFeeder054601...........................42Figure4.9:DownstreamMeterVoltageforFeeder054601.............43viiFigure4.10:GISMapforFeeder010603...........................44Figure4.11:DownstreamMeterVoltageforFeeder010603.............45Figure4.12:DistributionofMeterswithanAverageVoltagebetween5and7%.......45viiiChapter1IntroductionElectricutilitiesthroughouttheU.S.aremodernizinganelectricgridthathasbeenprovidingelec-tricityformorethanacentury.Today'selectricpowergridwasoriginallydesignedforaone-waywofelectricityfromcentralizedgeneratorstocustomersthroughtransmissionanddistributionsystems.Thiselectricsystemhasnotseenmuchinnovationoverthepastfewdecades.Infact,muchoftheexistinggridisanalog-basedandreliesonanaginginfrastructuretodeliverelectricity.Recentadvancesingridtechnologiesareallowingutilitiestoimprovereliabilitywhilereducinggreenhousegasemissions.Onewayisthroughtheuseofsmartgridtechnology.Smartgridsarenetworksthatusedigitalandotheradvancedtechnologiestomonitorandcon-trolelectricityasitwsthroughthepowergrid[1].Smartgridtechnologiesenableatwo-waywofinformationbetweenelectricaldevicesandutilityoperatorsthroughadvancedcommunica-tionssystems.Thesetechnologiesalsohelpfacilitatetheintegrationofrenewableenergyresourcessuchaswindandsolarintothegrid.Theelectricpowergridcontainsmillionsofmilesofpowerlines,thousandsofsubstations,protectivedevices,andtransformers[2].Thesecomponentsplayanessentialroleinthedeliveryofelectricpower;however,mostofthemlackthetechnologyneededtoadapttothechangingenergydemandsofthe21stcentury.Smartgridtechnologyallowstransmissionlinesandsubstationstobewithdevicessuchasphasormeasurementunits,xibleACtransmissionsystemdevices,andotheradvancedsensorsthatprovidegreaterlevelsofcontrolandvisibility[3].1Modernizationprojectsarenotonlyoccurringintransmissionsystems,butindistributionsys-temsaswell.Inadditiontodistributedgeneration,utilitiesareupgradingelectricaldeviceswithinthedistributionsystem.Forexample,devicessuchasregulators,capacitors,andswitchesarebe-ingintegratedwithcommunicationssystemsthatenablevoltage/VArcontrolandautomaticfeederswitching[4].Oneofthemostpopularmodernizationprojects,however,isthedeploymentofadvancedmeteringinfrastructures.Anadvancedmeteringinfrastructure(AMI)isacombinationofsmartmeters,communicationsnetworks,anddatamanagementsystemsthatenablesatwo-waywofinformationbetweencustomersandtheutility[5].TheU.S.DepartmentofEnergy,togetherwiththeelectricityindustry,hasinvestedover7.9billiondollarsinto99SmartGridInvest-mentGrant(SGIG)projectsundertheAmericanRecoveryandReinvestmentAct[6].Morethanhalf($4.050billion)oftheallocatedfundswenttotheinstallationofsmartmeters,andthedeploy-mentofadvancedmeteringinfrastructures.Ofthe65SGIGAMIprojects,manyreportedsuchas:fewerphysicalmeterreads,remoteconnectanddisconnectfunctionality,ofenergytheft,andreductionofelectricitydemand.Inadditiontothesesmartmeterscanalsobeusedforoutageaswellasdetectingvoltageorpowerqualityissues.AfewSGIGprojectshaveintegratedAMIsystemswithexistingoutagemanagementsystems(OMS)[7].However,manyelectricutilitiesoutsideoftheSGIGprogramarestillinthedeploymentphaseandhaveyettousesmartmeterstoanalyzeoutageeventsandidentifypowerqualityissues.1.1ResearchObjectivesThisthesispresentsananalysisoftwooftheaforementionedofAMIsystems:outageeventandtheofpowerqualityissues.Thisanalysisisconductedonanexistingelectricsystemthatisintheprocessofdeployingsmartmeterstoits1.8millionelectric2customers.Foroutageeventsmartmeteroutageeventsarecomparedwithhistoricaloutagesinanoutagemanagementsystemtoidentifydifferencesinoutage/restorationtimesandthenumberofcustomersoutforagivenoutage.Thepurposeofthiscomparisonistoidentifypossiblechallengesthatcouldarisewhenintegratingsmartmetereventsintooutagemanagementsystems.Ingeneral,powerqualitycanincludeanumberofissuessuchaschangesinsystemfrequency,voltagelevels,orharmonics.Inthisthesis,powerqualityreferstothedeviationofvoltagelevelsaboveavoltagerange.Smartmetervoltagedatawillbeusedtoidentifyfaultyelectricaldevicessuchasdistributiontransformersandregulatorsbeforedamageisdonetotheelectricsystem.Smartmetersprovidegreaterlevelsofvisibilityintothedistributionsystem,whichallowselectricutilitiestotakeaproactiveapproachtoaddressingbothcustomerandsystemissues.1.2OrganizationofThesisThisthesiscontainsvechapters,includingthisintroduction.Chapter2presentsbackgroundinformationincludingabriefhistoryofelectromechanicalmetersandadescriptionofadvancedmeteringinfrastructures.ThischapteralsodescribestwosystemsthatareessentialtoanytypeofoutageorpowerqualityanalysisusingsmartmetersŠgeographicinformationsystemsandoutagemanagementsystems.Chapter3describestheanalysisofoutageeventsandpresentsaseriesofcasescomparingtheeventsgeneratedbysmartmeterstothosegeneratedbyanoutagemanagementsystem.Chapter4describestheanalysisofsmartmetervoltagedatatoidentifyequipmentissuesandpresentsaseriesofcasesdescribingtheresults.Chapter5providesconcludingremarksandrecommendationsforfutureresearch.3Chapter2Background2.1HistoryofElectromechanicalMetersTheearliestformsofelectricitymeterscanbetracedbacktothelate1800s.Inthelate19thcen-tury,alternatingcurrent(AC)electricsystemswerebeginningtogainpopularityoverdirectcurrent(DC)electricsystems[8].OneofthemajoradvantagesofACsystemswastheabilitytotransmitpoweroverlongdistancesusingtransformers.DCsystems,ontheotherhand,weremoreexpen-siveandrequiredelectricitytobegeneratedneartheenduser.ThewidespreadadoptionofACsystemsincreasedtheneedformetersthatcouldaccuratelymeasurehowmuchenergywasbeingconsumed.ThepreviousmetersthatwereusedinDCsystemswerebasedonanelectrochemicalreactioninwhichanelectriccurrentpassedthroughanelectrolytewithtwozincplatesinsideofajar[9].Attheendofthebillingcycle,theplateswereweighedtoseehowmuchzincwastrans-ferredfromoneplatetotheother,whichdeterminedhowmuchelectriccurrentwassuppliedtothecustomer.Notonlywasitdiftoobtainanaccuratereadingoftheactualamountofenergyconsumed,thesemeterswerenotsuitableforACsystems.Toaddressthisissue,engineersbegandevelopingseveraltypesofelectromechanicalmeters.Themajorityofthesemeterswereeitherampere-hourorwatt-hourmeters.Oneofthemostcom-monlyusedampere-hourmetersinACsystemswasdevelopedbyOliverB.Shallenbergerin1888[9].Thisinductionmetercontainedtwocoils:alargeovalshapedcoilandasmallercoilplacedat4anangleinsideofthelargercoil.Insideofthesmallercoilwasanaluminumdiskattachedtoarodwithaluminumbladesthatwasusedtocontrolthespeedofthedisk.Whenanalternatingcurrentpassedthroughthelargercoil,asecondarycurrentwasinducedintheinnercoil.Thiscreatedamagneticthatinducededdycurrentsintothediskandcausedittorotate.Thespeedatwhichthediskrotatedwasdirectlyproportionaltotheamountofcurrentwingthroughthemeter.Figure2.1:Shallenberger'sAmpere-HourMeter[10]AmajordrawbackofShallenberger'smeterwasthefactthatitonlymeasuredelectriccurrentandnottheactualamountofpowerbeingconsumed.Thismeteroperatedontheassumptionthatthecurrentwasbeingsuppliedtoapurelyresistive(lighting)load.Inthiscase,theamountofenergyconsumedwouldbeproportionaltothecurrentwingthroughthemeter.However,theintroductionofinductivemotorloadscreatedaneedformetersthatcouldaccountforchangesinpowerfactor.In1889,aHungarianengineerbythenameofOtt´oBl´athypatentedtheinductionwatt-hourmeterthatcouldbeusedwithbothresistiveandinductiveloads[11].Bl´athy'smeterconsistedoftwoelectromagnets,arotatingdisk,andapermanentbrakingmagnet.Thetwoelectromagnetswerepositionedinsuchawaythattheycreatedtwomagneticdisplacedinphasewhenacteduponbythelinevoltageandloadcurrent.Thesechangingmagneticinducededdycurrentsintothediskcausingittorotateataspeedproportionaltotheamountofenergybeingconsumed.5Apermanentmagnetwasusedasadampingmechanismtocontrolthespeedofthedisk.Bl´athy'sidealedShallenbergertodevelopaninductionwatt-hourmetertobeusedintheU.S.in1894[9].Overthenextfewdecades,severalmeterswerecreatedtoexpandontheideasofBl´athyandShallenberger.Thesemetersaimedtoreducetheweight,cost,andimprovetheaccuracyofthepreviousmeters.Inaddition,metermanufacturessuchasGEandWestinghousedevelopedpolyphasemetersthatwereusedforcommercialapplications.Figure2.2:SinglePhaseWatt-HourMeterTheelectromechanicalwatt-hourmetersinusetodayoperateonthesamefundamentalprin-ciples.Thesemetersgenerallycontainsimilarcomponents:analuminumdisk;displaydials;permanentmagnets;andavoltagecoilandtwocurrentcoilsthatarewrappedaroundanironcore.Thevoltagecoilisconnectedtothesupplyvoltageandplacedabovethedisk.Thetwocurrentcoilsareconnectedinseriestotheelectricloadandplacedbelowthedisk.Ascurrentwsthroughthevoltagecoilcontainingalargenumberofturns,amagneticiscreatedthatisproportionaltothesupplyvoltage.Thecurrentcoilshavefewerturnsandcreatemagneticthatarepropor-tionaltotheamountofcurrentdrawnbytheload.Thesechangingmagneticinduceeddycurrentsintothealuminumdiskcausingittorotateataspeedproportionaltotheamountofpowerbeingconsumedbytheload.Thealuminumdiskisconnectedtoaseriesofgearsthatturndisplaydialsindicatingtheamountofpowerconsumedinkilowatt-hours.Thepermanentmagnetscontrol6thespeedofthediskandpreventitfromspinningwhenthereisnopowerbeingconsumed.Theintroductionofdigitalelectronicsintotheelectricitymeteringindustryhaseliminatedtheneedforelectromechanicalmeters.Today,manyoftheexistingelectromechanicalmetersarebeingreplacedwithadvancedmetersthatnotonlymeasureenergyconsumptionbutofferawiderangeofThesemeters,alsoreferredtoassmartmeters,willbedescribedinthenextsection.2.2AdvancedMeteringInfrastructure2.2.1SmartMetersAsmartmeterisakeycomponentofanyadvancedmeteringinfrastructure.Smartmeters,unliketraditionalelectromechanicalmeters,havetheabilitytowirelesslytransmitconsumptiondatatotheelectricutilitythroughasecurenetwork.Priortosmartmeters,utilitycompanieshadtosendmeterreaderstothecustomer'slocationeachmonthtoobtaintheamountofenergyconsumed.Withsmartmeters,thisconsumptioninformationcanbesenttotheutilitydailyorinintervalsof15,30,or60minutes[12].Thisinformationcanbeusedtoprovideaccurateelectricitybillsanddevelopprogramstohelpcustomersbettermanagetheirenergyusage.Onemajorofasmartmeterisitsabilitytosendwheneveritlosespowerorisrestored.Thisinformationcanhelputilitiesdeterminethelocationandextentofapoweroutageandidentifycustomerswhoarestilloutofpoweraftercrewshaverestoredpowertoanarea.Smartmeterscanalsomonitorvoltagelevelsatthecustomer'slocation.Thisinformationcanbeusedtoimprovepowerqualityandensurethatcustomersarereceivingvoltagesatlevelssuitableforpowerconsumption.Thedesignofasmartmetercanvarybetweenmanufacturers;however,mostsmartmeterscontainsimilarcomponents.Theseincludedigitalelectronicssuchassensors,microcontrollers,andLCDdisplays.Allsmartmeterscontaincommunicationsmodulesthatallowdatatobesent7fromthecustomertotheutility,andviceversa.AsmartmetercanalsohaveaswitchŠusuallyratedat200ampsŠthatallowsutilitycompaniestoremotelyconnectordisconnectelectricitywithoutsendingaservicepersontothelocation[13,14].Figure2.3:SinglePhaseSmartMeterThereareseveralwaysthatasmartmetercanmeasureelectricity.Thesetechniquesutilizevoltageandcurrentsensors.Thevoltagesensoristypicallyintheformofasimpleresistordividercircuitwithtwolargeresistorsthatreducethelinevoltagetoalevelsuitablefortheinternalelec-tronicsofthesmartmeter[15,16].Themostcommonlyusedcurrentsensorsareshunts,currenttransformers,andRogowskicoils[15,17].Thecurrentshuntisalowresistanceresistorthatisplacedinserieswiththeload.Astheloadcurrentwsthroughtheresistor,asmallvoltagedropiscreatedacrossit.Thisvoltagedropandtheknownresistanceoftheshuntresistorareusedtodeterminethecurrentdrawnbytheload.Acurrenttransformerconsistsofacoilofwirewrappedaroundanironcore.Asthecurrentwsthroughtheprimaryconductor(s)placedinsideofthecurrenttransformer,amagneticiscreatedthatinducesacurrentintothesecondarycoilofthetransformer.Aburdenresistorisusedtocreateavoltagesignalthatisproportionaltothecurrentwingthroughtheconductor[18].ARogowskicoilisacoilofwirewrappedaroundanon-magneticoraircore.Likeacurrenttransformer,aRogowskicoilisplacedaroundaconductorcarryingloadcurrent.TheoutputofaRogowskicoil,however,isavoltagethatisatimederivative8ofthecurrentwingthroughtheconductor.Theinducedvoltagesignalhastobeintegratedtoobtainasignalthatisproportionaltothecurrentwingthroughtheconductor[19,20].Oncetheanalogvoltageandcurrentsignalsareobtained,theyareprocessedbytheinternalelectronicsofthesmartmetertodisplayortransmittheelectricityconsumedbythecustomer.2.2.2CommunicationsSystemsAdvancedmeteringinfrastructuresenabletwo-waycommunicationthroughtheuseofradiofre-quency(RF)technologiessuchasmesh,point-to-point,andcellular[21,22].InaRFmeshnet-work,smartmetersformlocalnetworkswitheachotherandtransmitinformationtonearbyrouters.Theseroutersgatherthemeterdataandcommunicatewitheachotherbeforesendingtheinforma-tiontoacollector.Thecollector(usuallyatower)providesthelinkbetweenthesmartmeterdataandtheutility.Meshnetworksprovidemultiplepathsforasmartmetertosendinformation,whichincreasesthechancesofthedatabeingreceivedbytheutility.However,inruralareaswherecustomerstendtobespreadapart,meshtechnologycanrequireadditionalinfrastructure[21].Inapoint-to-pointnetwork,eachsmartmetertransmitsitsinformationdirectlytothecollector.Sincetherearenointermediatenodesbetweenthesmartmeterandthecollector,thereislessinfras-tructureneededcomparedtomeshnetworks.Thesenetworksareusedinareaswherethereisconsiderabledistancebetweeneachsmartmeter[23].Inacellularnetwork,smartmetersutilizetheexistingnetworksofcellularcompaniestotransmitdata.EachsmartmeterisequippedwithaSIMcardthatenablesthemetertosendamessagecontainingitsdatadirectlytotheutility.Thereareseveraladvantagesassociatedwithcellularnetworks.Tobegin,thesenetworksaremaintainedbythecellularcompanywhichhelpstoreducethecostandtimeneededtosetupthemeteringinfrastructure.Cellularnetworksalsomaintainhighlevelsofreliabilityandprovidewidespreadcoverage.9Powerlinecarrier(PLC)technologycanalsobeusedwithinadvancedmeteringinfrastruc-tures.PLCtechnologyallowsutilitiestoleveragetheexistingdistributionsystemtosendandreceiveinformation[24].Smartmeterscansendinformationthroughmilesofdistributionlinestosubstationswhereitiscollectedandsenttothehead-endsystem.Eachofthesetechnologieshasitsadvantagesanddisadvantages.ThechoiceofcommunicationssystemsforAMIvariesbetweenutilitiesandcandependonanumberoffactors.Theseincludethecosttodeploythesystem,thenumberofcustomersservedbytheutility,theserviceterrain,andtheabilitytoexpandinthefuture[21,22,25].2.2.3MeterDataManagementSystemsTherawinformationfromsmartmeterssuchasconsumptiondata,voltagedata,andoutageorrestorationhastobeprocessedinordertoprovideoperationaltotheutility.Ameterdatamanagementsystemreceivesthisrawdataandpreparesitforusewithintheutility.Meterdatamanagementsystemsvalidateandprocesssmartmeterdataforanumberofapplica-tions.Thesecanincludebillingsystems,outagemanagementsystems(OMS),andgeographicinformationsystems(GIS).2.3GeographicInformationSystemsAgeographicinformationsystemisacomputerbasedsystemthatrelatesassetdatatoageograph-icallocation.Geographicinformationsystemsarebuiltontopofrelationaldatabasesthatcontainavarietyofdatasourcesstoredintables.Thesetablesofinformationcanbeusedtoanswerques-tionsaboutasystemanddisplaytheresultsondigitalmap.ElectricutilitiesuseGIStocreatedynamicmodelsoftheirelectricsystems.PriortoGIS,electricutilitiesreliedonpapermapsto10locateelectricaldevicesintheWithGIS,electricutilitiescannotonlylocateelectricalde-vices,theycanaskquestionsbasedonthatdataandvisualizetheresultsonamap.Forexample,anelectricutilitymaywanttoidentifyallofthecustomerswithsmartmetersthataredownstreamfromaprotectivedeviceonaparticularcircuit.ThiscanbemadepossibleinGISbyqueryingthecustomerandprotectivedevicetablesandcreatingalayerdisplayingtheresults.Figure2.4:ElectricDistributionCircuitsinGISFigure2.4showsaportionofanelectricdistributionsystemmodeledinageographicinfor-mationsystem.Intheabove,eachyellowtrianglerepresentsadistributiontransformerthatfeedsoneormoreelectriccustomers.Thesolidblueandgreenlinesarethethreephaseprimaryconductorswhilethedashedlinesrepresentasinglephase.Eachfusehastwonumbersassociatedwithit.Thetopnumberrepresentstheloadconcentrationpoint(LCP)andthebottomnumberisthecurrentratingofthefuseinamps.Geographicinformationsystemsareessentialtoanytypeofsmartgridanalysis.SmartgridtechnologiessuchasthedeploymentofAMIrequireanelectricmodelthatisuptodatetofullyrealizetheAsaresult,electricutilitiesareinvestingmil-11lionsofdollarsintoGIStocorrectdistributionsystemmodelsincludingcustomertotransformertophaseconnectivity.TheneedforaccuratemodelsoftheelectricdistributionsysteminGISwillcontinuetogrowasmoreelectricutilitiesbegindistributionmodernizationprojects.2.4OutageManagementSystemsAnoutagemanagementsystemisasystemthatallowselectricutilitiestomanagetheprocessofdetectingcustomeroutagesandrestorationsduringastormorinterruption.Anoutagemanagementsystemisusuallyconnectedtoseveralsourcesofinformation.Thisincludesacustomerinforma-tionsystem(CIS)thatprocessescustomertroublecallsandageographicinformationsystemthatprovidesamodelofthedistributionsystem.Ascustomerscalltoreportoutages,anoutageman-agementsystemusesaseriesofpredictionalgorithmstoidentifyelectricaldevicescommontothelocationofthosecustomercalls.Oncetheupstreamdevicesarefound,theelectricutilitycandispatchserviceworkerstothelocationtodeterminetheextentoftheoutageandrestoreservicetothosecustomers.Traditionally,outagemanagementsystemshavereliedoncustomercallsastheprimarysourceofinformationaboutanoutage.However,thedeploymentofadvancedmeteringinfrastructureshasmadeitpossibletousesmartmeterstoidentifycustomeroutagesandrestorations.Whenasmartmeterlosespower,itsendsatotheutilityindicatingthelossofpower.Thesetions,whenintegratedintoOMS,canprovideadditionalinformationtohelputilitiesidentifytheexactnumberofcustomersaffectedbyanoutage.However,inordertosuccessfullyintegratesmartmeteroutageeventsintoOMS,theoutageeventsgeneratedbysmartmetersshouldbecomparedwiththeeventsofanoutagemanagementsystem.12Chapter3OutageEventComparison3.1IntroductionThischapterintroducesthecomparisonofoutageeventsbetweensmartmetersandanoutagemanagementsystem.Asmentionedinsection2.4,smartmeterscanbeusedasanotherwaytoidentifycustomeroutagesorrestorations.Inthischapter,aseriesofcaseswillbepresentedtocompare,foraparticularincidentinOMS,thenumberofsmartmetersthatreportedanoutageorrestorationeventandthetimeinwhichtheeventwasreported.Itisimportanttonotethatthefollowinganalysiswasdoneusinghistoricaloutageeventdata.ComparinghistoricalsmartmeteroutageeventswithOMSincidentshelpsidentifypossiblechallengeswhenintegratingreal-timesmartmeterdataintooutagemanagementsystems.3.2Methodology3.2.1ObtainingOutageDataThestepintheprocessofcomparingsmartmeteroutageeventswithanOMSistoobtainthedata.WheneverasustainedoutageoccursontheelectricsystemŠwhetherit'sascheduledoutageoronecausedbyananimalorstormŠanincidentreportiscreatedinanoutagemanagementsystem.Thisreportincludesinformationsuchastheapproximatetimeoftheoutage,thecause13oftheoutage,thefeeder/circuitimpacted,andthepredictednumberofcustomersaffectedbytheoutage.Aftertheoutagehasbeencleared,therestorationtimeisrecordedandtheincidentreportisarchivedinarelationaldatabase.Figure3.1:SampleOMSIncidentArchiveThenextsourceofdataissmartmeteroutageandrestorationevents.Smartmetersarecapableofsendingfilastgaspfloutagenotwhenacustomerlosespower.Thisisenabledthroughtheuseofinternalcapacitors.Thesecapacitorsstoreenoughenergytoallowthemetertosendoneormoremessagesbacktotheutilityindicatingthelossofpower.Inaddition,wheneverasmartmeterisrestoredafteranoutage,arestorationeventiscreatedandloggedwithinthesmartmeter'smicroprocessor.Smartmeteroutageandrestorationeventsaretransmittedbacktotheutilityeverydaythroughasecurecellularnetwork.Whilethereareseveralmanufacturersofsmartmetersonthemarket,themajorityofthemetersdeployedinthissystemweredevelopedbyItron(Centron)andGeneralElectric(I-210+c).Eachmanufacturerhasitsownuniquesetofcodescorrespondingtooutage14andrestorationevents.However,oncethesecodesarereceivedbythehead-endsystemtheyareconvertedintofourmajorevents:primarypowerdown,registerpowerdown,primarypowerup,andregisterpowerup.Thesefoureventsaregeneratedwheneverthesupplyofpowertothecustomers'meterislostorrestored.Thetwo,primarypowerdownandregisterpowerdownbothindicatethelossofpowerbutcomefromdifferentplaceswithinthemeterandaregenerallywithinafewsecondsofeachother.Similarly,primarypowerupandregisterpowerupbothcorrespondtotherestorationofpowerbutarewithinafewsecondsofeachother.Thisredundancyimprovesthechancesofaneventbeingrecordedwheneveranoutageorrestorationoccurs.Table3.1:SmartMeterOutageandRestorationEventCodesEventCodeDescription18001PrimaryPowerDown18922RegisterPowerDown18002PrimaryPowerUp18923RegisterPowerUpThereareotherincidentsthatcantriggerthelossofpowerforasmartmeter.Theseincludeaninternalfailurewithinthemeteritselforatampereventinwhichacustomerattemptstoremovethemeterfromitssocket.Inbothcases,theseincidentshaveuniqueeventcodesthatproceedapowerdownevent.However,incomparingoutageeventswithknownOMSincidents,thefocusisprimarilyonthefourmetereventsthataretriggeredbyalossofpowertothecustomer.3.2.2DataProcessingwithPythonTheelectricsystemunderstudyprovideselectricitytoover1.8millioncustomers.Thesecustomersaredistributedacross1,145substations(roughly2,200circuitsorfeeders)andareaseitherresidential,industrial,orcommercial.Ontheparticularfeedersthatwereselected,over90%ofresidentialcustomershadsmartmetersinstalled.Table3.2showsthetotalnumberofcustomers15Table3.2:FeederswithaHighPercentageofSmartMetersSubstationFeederIDCustomerswithAMITotalCustomers%ofCustomerswithAMIKnapp1035032,4392,4890.97Becker0475021,8401,9120.96Mccracken0675011,7741,8300.96Dutton0515041,6131,6680.96Norton0299022,0442,1340.95Apple0368042,0462,1380.95Broadway0619011,5871,6660.95Getty0607013,0243,1830.95Apple0368011,2401,2960.95Mccracken0675041,2221,2840.95Broadway0619029861,0300.95MonaLake0731012,4922,6270.94Becker0475012,2192,3370.94Holton0302011,6881,8080.93Breton1242042,0272,1640.93Evanston0128025595970.93Ravenna0373016516930.93Hickory0342029119900.92Saugatuck0519031,3301,4420.92Dutton0515011,1331,2280.92Hamilton0291026647240.91oneachfeederaswellasthenumberofcustomerswithsmartmeters.Thedatausedfortheanalysiswasgatheredfrom16differentsubstations(21feeders)betweenJuneandJulyof2015.Aspreviouslymentioned,thetwomainsourcesofdataforthisanalysisaresmartmeteroutageandrestorationeventsandtheoutageinformationfromOMS.TheoutageandrestorationeventswereextractedfromthemeterdatamanagementsystemandstoredintwoseparatecsvBothcontaininformationsuchasthetimeoftheevent,themeternumberassociatedwiththatevent,thenameandaddressofthecustomer,andthefeederandtransformerthatthemeterisconnectedto.Themaindifferencebetweentheisthatthetwooutageeventcodes,primarypowerdownandregisterpowerdown,arestoredinoneandthetworestorationeventcodes,primarypowerupandregisterpoweruparestoredinanotherTheOMSoutageinformationisalsostoredinseparatecsvOnecontains30outageincidentreportswithavaryingnumberofcustomers16affected.TheotherhasalistofeachcustomeraffectedforagivenincidentID.Thenumberofcustomersaffectedisdeterminedfromtheoutagemanagementsystem'spredictionalgorithmwhichfactorsintheelectricGISmodelandthelocationofcustomertroublecalls.Thepythonprogramminglanguagewasusedfortheoutageeventcomparison.Pythonhasanumberofopensourcepackagesandlibrariesfordataanalysis.Oneofthemostpopularlibraries,andtheonethatwillbeusedforthefollowinganalysis,isthepandaslibrary.Additionalinformationaboutthepandaslibrarycanbefoundin[26,27].Thesourcecodeusedfortheanalysiscanbefoundintheappendix.TheprogramworksbyreadinginthedataandstoringitintotablesorpandasDataFrames.Next,anemptylistiscreatedtostoretheresultsofthecomparison.Foreachincidentoroutage,thepredictednumberofcustomersaffectedaccordingtoOMSisfound.Theoutagetime,restorationtime,andfeederIDoftheoutagefromtheincidentreportisthenextracted.Asmartmeterisgenerallyexpectedtoreportanoutagebeforeacustomercallsin.Ifacustomerwakesup,noticesthatthepowerisout,andcallstheutilityat8:00am,thisisnotnecessarilythetimethatthecustomerlostpower.TheOMSwillrecord8:00amastheoutagetime;however,ifthatcustomerhadasmartmeter,thatmetermayhaverecordedanoutageeventat7:00am.Therefore,inordertocomparesmartmeteroutageeventswithanOMS,thereshouldbeasuftimerange.Forthisanalysis,twohoursbeforeandonehourafteranOMSoutageorrestorationisused.ThenextstepistosmartmeteroutageandrestorationeventsthatarewithinthespetimerangeandonthesamefeederastheOMSincident.Aftertheseeventsarefound,theyaresortedinascendingorderbytimestampandanyduplicatefilastgaspfleventsareremoved.ThesmartmeteroutagesandtheOMScustomersarethencombinedintoasingletable.Ifameter(orcustomeraccountnumber)appearstwiceinthetable,thisisameterthatwasoutaccordingtoboththeOMSandthesmartmeter.Next,gettherestorationtimesofthesmartmetersthatmatchedthe17OMSandcombinetheoutageandrestorationtimesofthemetersthatmatchedintoasingletable.AfterthesmartmetersthatmatchedtheOMSarefound,checkforanymetersthatdidnotmatch.ThesemeterscanbefoundbysearchingthecombinedsmartmeteroutagesandtheOMScustomerstableforuniqueaccountnumbers.Ifameterhasauniqueaccountnumberbutdoesnothaveaneventcodeassociatedwithit,thatmeterwasoutaccordingtotheOMSandwaseitheranelectromechanicalmeterwhichdoesnothaveaneventcodeorasmartmeterthatdidnotcommunicate.Conversely,ifameterhasauniqueaccountnumberandaneventcode,itwasasmartmeterthatseenanoutageduringthetimerangebutwasnotpredictedbytheOMSalgorithm.Theoutageandrestorationtimeofthismeterisstoredandtheresultsoftheoutageeventcomparisonareaddedtotheinitialcomparisonlist.Figure3.2providesawchartoftheoutageeventcomparison.18Figure3.2:FlowchartofOutageEventComparison1920Table3.3:OutageEventComparisonResultsIncidentIDFeederIDOMSCustomers#ofSMsthatmatchedOMS#ofMtrsinOMSthatdidnotcommunicateTotal#ofSMsthatreportedanoutageTotal#ofSMsrestoredOutageCause2884686124204212102121Trees288648303420271683440437Trees2893719061901121111212Animal2893970047501118388Transmission/Generation2894541103503615655656Planned/Scheduled28966680515041131076107107UniqueIncident2897071029102373527373Animal2897614037301211561615Trees289798002990236351307307Animal2898607103503212012020Animal2898628034202716836868Trees2899241036804434038442EquipmentFailure2899786047502131301414Animal2900028103503111101111Animal2900337073101786315882882Animal2901004103503505005353Trees2901113124204131211313Animal2901343036801232032121Trees2901547067504181261312Weather2901600012802141313130Weather2902354029102918929090Weather29036710368041051023102102Trees29048790675014835133737Nocause29061870619012781988UniqueIncident2906343060701131301315Animal2906812061902666247069Trees2909605051903161511515UniqueIncident291003003020111921010Trees29115240731011192179180Weather28992120515011082100100Public3.3AnalysisInthissection,aseriesofcaseswillbepresentedtosummarizetheresultsoftheoutageeventcomparisonshowninTable3.3.Theresultsfromthe30outageincidentreportscanbegroupedintothreecategories.Thecategory,wherethenumberofcustomerspredictedbytheOMSmatchedwiththenumberofsmartmetersthatreportedanoutageduringthetimeframe,iscoveredinCase1.Case2coversdifferentscenarioswherethenumberofOMScustomersandsmartmetersdidnotmatch.ThelastcaseexplorestheimpactofmomentarysmartmeteroutagesŠthatis,meteroutagesthatlastbetween5and10secondsŠontheoutageeventcomparison.3.3.1Case1OneoutageinwhichthenumberofcustomersoutaccordingtotheOMSmatchedwiththenumberofsmartmetersthatreportedanoutageisincidentnumber2884686.Thisoutagewascausedbyatreethatfellonaprimaryconductor,affecting21customers.Table3.4showsthemeternumber,outagetime,andrestorationtimeofeachcustomer.OnethingtonoticeisthateachmeterhasthesameoutageandrestorationtimeaccordingtotheOMS.Table3.5showstheoutagetime,restorationtime,andtransformerofthesmartmetersthatreportedanoutageduringthetimerangeandonthesamefeederastheOMSincident.Thesmartmetersthatreportedoutagesarefedfromthreedifferentdistributiontransformers;tenmetersareconnectedtotransformer0611091302,eightareconnectedto0611091201,andtheotherthreeareconnectedto0611091306.Thethreesmartmetersconnectedtotransformer0611091306werethetoexperienceanoutage.Thesemeterslostpowerat01:40on6/20/2015.Figure3.3showsthelocationofthetrans-formersinrelationtotheupstreamprotectivedevicewhichiscircledinred.Thetwotransformers21Table3.4:OMSCustomersforIncidentNumber2884686MeterOMSOutageTimeOMSRestorationTimeFeederIDIncidentID102679306/20/201503:04:186/20/201506:30:001242042884686102630856/20/201503:04:186/20/201506:30:001242042884686102837386/20/201503:04:186/20/201506:30:001242042884686102796716/20/201503:04:186/20/201506:30:001242042884686102837406/20/201503:04:186/20/201506:30:001242042884686102851036/20/201503:04:186/20/201506:30:001242042884686102827156/20/201503:04:186/20/201506:30:001242042884686102457866/20/201503:04:186/20/201506:30:001242042884686102457876/20/201503:04:186/20/201506:30:001242042884686102457736/20/201503:04:186/20/201506:30:001242042884686102457856/20/201503:04:186/20/201506:30:001242042884686102457746/20/201503:04:186/20/201506:30:001242042884686102457886/20/201503:04:186/20/201506:30:001242042884686102480946/20/201503:04:186/20/201506:30:001242042884686102457756/20/201503:04:186/20/201506:30:001242042884686102480956/20/201503:04:186/20/201506:30:001242042884686102480966/20/201503:04:186/20/201506:30:001242042884686102480936/20/201503:04:186/20/201506:30:001242042884686102630866/20/201503:04:186/20/201506:30:001242042884686102679296/20/201503:04:186/20/201506:30:001242042884686102679326/20/201503:04:186/20/201506:30:001242042884686Table3.5:SmartMeterOutagesforIncidentNumber2884686MeterSMOutageTimeSMRestorationTimeFeederIDTransformer102679306/20/201502:55:396/20/201506:02:411242040611091302102630856/20/201502:55:416/20/201506:02:431242040611091302102837386/20/201502:55:496/20/201506:02:511242040611091302102796716/20/201501:40:086/20/201506:03:091242040611091306102837406/20/201502:55:486/20/201506:02:491242040611091302102851036/20/201501:40:086/20/201506:03:081242040611091306102827156/20/201501:40:086/20/201506:03:081242040611091306102457866/20/201502:55:466/20/201506:02:471242040611091201102457876/20/201502:56:046/20/201506:03:051242040611091201102457736/20/201502:55:466/20/201506:02:471242040611091201102457856/20/201502:55:466/20/201506:02:471242040611091201102457746/20/201502:55:476/20/201506:02:481242040611091201102457886/20/201502:55:476/20/201506:02:491242040611091201102480946/20/201502:55:376/20/201506:02:391242040611091201102457756/20/201502:55:446/20/201506:02:461242040611091201102480956/20/201502:55:476/20/201506:02:481242040611091302102480966/20/201502:55:446/20/201506:02:451242040611091302102480936/20/201502:55:416/20/201506:02:421242040611091302102630866/20/201502:55:426/20/201506:02:431242040611091302102679296/20/201502:55:406/20/201506:02:411242040611091302102679326/20/201502:55:416/20/201506:02:42124204061109130222Figure3.3:GISMapforIncident2884686inredŠ0611091302and0611091201ŠareconnectedtotheZphaseconductor.Transformer0611091306isshowninblueandconnectedtotheXphase.AccordingtotheOMSincidentcalllog,anemergencycallwasmadeat01:50,whichcorrespondstothetimethatthethreemeterslostpower.ItislikelythatthetreefellontheXphaseconductor,sincethesethreemeterslostpoweraroundthesametimethatthecallwasplaced.Toisolatethisfaultfromtherestoftheelectricsystem,theupstreamprotectivedeviceŠwhichisa100AfuseŠwasopened.Theincidentreportmentionsthatanelectricserviceworkeropenedthefuseat03:04torepairtheprimaryconductor.Thisisalsothesameoutagetimeofthe21metersintheOMS.Thesmartmetersontheothertwotransformersdownstreamofthatfuserecordedanoutagetimeof02:55,whichisabout10minutesearlierthantheOMSoutagetime.Aftertherepairsweremade,thefusewasclosedandpowerwasrestoredtotheaffectedcustomers.Anotheroutagethatfallsintothiscategoryisincidentnumber2900028.Thisoutagewascausedbyabirdthatcameincontactwithapolemounteddistributiontransformer,whichresulted23in11customerslosingpower.Sevenoftheelevencustomerscalledtoreportapoweroutage.Ac-cordingtotheOMSincidentcalllog,thecustomercallwasreceivedat08:55:36on7/11/2015andthelastcallwasreceivedat10:21:56.Sincetheoutagemanagementsystemtypicallyusesthetimeofthecustomercallastheoutagetime,all11metersconnectedtothattransformerhadanoutagetimeof08:55:36.TheOMSrestorationtimeofthosemeterswas11:00.Allofthemetersconnectedtothattransformerweresmartmeters.Table3.6showstheoutageandrestorationtimeseenbyeachsmartmeter.Theoutageandrestorationtimeswereroughlythesamebetweenthesmartmetersandtheoutagemanagementsystem.Table3.6:SmartMeterOutagesforIncidentNumber2900028MeterSMOutageTimeSMRestorationTimeFeederIDTransformer102501947/11/201508:52:587/11/201510:53:51103503711171204100565507/11/201508:53:027/11/201510:53:54103503711171204102554747/11/201508:52:557/11/201510:53:48103503711171204102554737/11/201508:52:547/11/201510:53:46103503711171204102554757/11/201508:52:567/11/201510:53:48103503711171204102554507/11/201508:52:577/11/201510:53:48103503711171204102527987/11/201508:53:147/11/201510:54:05103503711171204102554977/11/201508:52:567/11/201510:53:48103503711171204102560017/11/201508:52:597/11/201510:53:51103503711171204102560037/11/201508:52:597/11/201510:53:51103503711171204102501987/11/201508:53:007/11/201510:53:521035037111712043.3.2Case2ThiscasedescribesoutageswherethenumberofcustomersoutaccordingtotheOMSdiffersfromthenumberofsmartmetersthatreportedanoutage.ThreeoutageswillbecoveredthatexplainpossiblereasonsforthedifferencesinTable3.3.Theoneisincidentnumber2904879.Theincidentreportmentionsthatafuseopened,whichresultedin48customerslosingpower,butdoesnotindicatewhatcausedthefusetoopen.Eachoneofthe48customersoutaccordingtotheOMScustomerlisthadanoutagetimeof11:44:10andarestorationtimeof14:00:05on7/14/2015.24Figure3.4:GISMapforIncident2904879Therewere37smartmetersthatreportedanoutageonthesamefeederastheOMSincident.All37lostpoweraround11:33andwererestoredaround13:58onthesameday.Twoofthesemeterswerewithinthetimerange,butwerenotpredictedbytheOMS.Theother35smartmeterswerepredictedbytheOMS.Figure3.4showsthethreedistributiontransformers(0917024403,0917024402,and0917024405)thatpoweredthe35matchingsmartmetersinredandthetwotransformersthatpoweredthetwosmartmetersthatdidnotmatchinblue.FromtheGISmodelshownabove,the35smartmetersthatmatchedOMSareallconnectedtotheYphase.ItispossiblethatthetwosmartmetersthatreportedanoutageareactuallyconnectedtotheYphaseandnottheXphaseasshowninthemodel.IntheOMScustomerlist,therewere13metersthatdidnotcommunicateanoutage.Twoofthesemetersareelectromechanicalmetersandarenotcapableofsendinganoutageevent.Theothermetersaresmartmetersthatdidnotcommunicateanoutage.Table3.7showsthe25Table3.7:13MetersthatdidnotcommunicateforIncidentNumber2904879MeterOMSOutageTimeOMSRestorationTimeTransformerEventCode101205657/14/201511:44:107/14/201514:00:050917024402N/A101248067/14/201511:44:107/14/201514:00:050917024405N/A101247147/14/201511:44:107/14/201514:00:050917024405N/A101247137/14/201511:44:107/14/201514:00:050917024405N/A101247157/14/201511:44:107/14/201514:00:050917024405N/A101247827/14/201511:44:107/14/201514:00:050917024405N/A101247837/14/201511:44:107/14/201514:00:050917024405N/A101247847/14/201511:44:107/14/201514:00:050917024405N/A101217597/14/201511:44:107/14/201514:00:050917024402N/A631547187/14/201511:44:107/14/201514:00:050917024402N/A101220107/14/201511:44:107/14/201514:00:050917024402N/A101070987/14/201511:44:107/14/201514:00:05RemovedfromServiceN/A650363867/14/201511:44:107/14/201514:00:050917024403N/Ameternumberandtransformerassociatedwitheachmeter.Themeternumbersthatstartwith10*indicatethattheyaresmartmeters,andthemetersthatstartwith6*areelectromechanical.Withtheexceptionofonemeter(10107098)thatwasremovedfromservice,allofthesemeterswereconnectedtooneofthethreematchingtransformersinFigure3.4sotheyshouldhaverecordedanoutageeventbutdidnot.Thesecondoutageisincidentnumber2906187.Likethepreviousoutage,afuseopenedcaus-ing27customerstolosepower.Theoutagetimeandrestorationtimeofthe27customersaccord-ingtotheOMSwas08:34:25and09:30respectivelyon7/16/2015.Outofthe27customersthatwereoutaccordingtotheOMS,only8smartmetersreportedanoutage.Thesemeterslostpoweraround08:17andwererestoredaround09:46.Theremaining19metersthatwereoutaccordingtotheOMSaresmartmetersthatdidnotreportanoutage.These19metersareconnectedtofourtransformers:sevenofthemetersareconnectedtotransformer0916041101,sixmetersarecon-nectedto0916041102,threemetersareconnectedto0916041108,andthreemetersareconnectedto0916041103.InFigure3.5,allofthetransformersweredownstreamofthe100Afuse.Whenthatfuseopened,all27smartmetersshouldhavereportedanoutage.However,only8meters26Figure3.5:GISMapforIncident2906187recordedanoutageevent.Thesemetersareconnectedtotheredtransformers.Thethirdoutageisincidentnumber2897071.TheOMSincidentreportmentionedthatananimalwasthecauseofthisoutageandthat37customerswereaffected.However,therewereatotalof73smartmeters(including35thatmatchedtheOMS)thatreportedanoutageonthesamefeederandwithinthesametimerange.Theoutageandrestorationtimeofthe37customersaccordingtotheOMSwas22:23:58on7/5/2015and02:35on7/6/2015,respectively.Allofthe73smartmetersthatreportedanoutagelostpoweraround22:30exceptforonemeterthatlostpowerat22:17:53.Thecustomerassociatedwiththismeterwasthetocallabouttheoutageandmentionedarcingpowerlines.Figure3.6showsthelocationofall73smartmetersthatreportedanoutage.ThesmartmetersthatmatchedwithOMSareconnectedtotheredtransformersandthesmartmetersthatdidnotmatchareconnectedtothelightbluetransformers.Thedarkbluetransformeristhelocationofthe27Figure3.6:GISMapforIncident2897071customerwhocalledtoreporttheoutage.AccordingtoOMS,thefuse(circledinred)openedwhichcaused37customerstolosepower.However,therewere38smartmetersupstreamofthatfusethatalsolostpower.ThesemetershadthesamerestorationtimeofthemetersdownstreamofthefusebutwerenotpredictedbytheOMS.Mostofthe73smartmeterswereconnectedtotheZphase.3.3.3Case3Thiscasecoverstheimpactofmomentarysmartmeteroutagesontheoutageeventcomparison.Oneoutagethatfallsintothiscategoryisincidentnumber2911524.Thisoutageresultedfrombadweatherthatcausedafuseandarrestortoblowaffecting11customers.Theoutageandrestoration28timeofthe11customersaccordingtotheOMSwas18:54:12and20:00respectivelyon7/18/2015.Therewereapproximately179smartmetersthatexperiencedanoutageduringthistimeframeonthesamefeeder.Theoutageandrestorationtimeofthe8metersthatmatchedtheOMScustomerlistwas18:36and19:53respectively.Theremainingsmartmetersthatreportedanoutageweresmartmetersthatexperiencedamomentaryoutagethatlastedafewseconds.Table3.8showstheoutageandrestorationtimeforthreeofthesesmartmeters.Theremainingsmartmetershadsimilaroutageandrestorationtimes.Table3.8:MomentarySMOutagesforIncidentNumber2911524MeterSMOutageTimeSMRestorationTimeFeederIDTransformer101115977/18/201518:36:477/18/201518:36:530731010916154104101115417/18/201518:36:477/18/201518:36:520731010916154104101115437/18/201518:36:507/18/201518:36:550731010916154140Figure3.7:GISMapforIncident291152429Figure3.7showsthelocationofthesmartmetersthatreportedanoutage.Inthisthe11metersthatwereoutaccordingtotheOMSareconnectedtotheredtransformer.Themetersthatexperiencedamomentaryoutageareconnectedtothepurpletransformers.Thereddotsindicatealloftheprotectivedevicesthataredownstreamoftherecloser.Fromthewecanseethatthe200Arecloseropenedandclosedwhichcausedthemomentaryoutages.Anotheroutagethatfallsintothiscategoryisincidentnumber2899212.Thisoutagewascausedbyacarthathitapole.Thecollisioncausedtheprimarylinestotouch,whichresultedin10customerslosingpowerforabout50minutes.TheOMSandsmartmeteroutageandrestorationtimesofthemetersthatmatchedwereroughlythesame.Therewereatotalof100smartmetersthatreportedanoutagearoundthesametime;eightmetersmatchedtheOMS,and92metersexperiencedmomentaryoutagessimilartothepreviousoutage.InFigure3.8,thematchingmetersareconnectedtotheredtransformersandthemetersthatexperiencedmomentaryoutagesareconnectedtothepurpletransformers.30Figure3.8:GISMapforIncident289921231Chapter4PowerQualityAnalysis4.1IntroductionInadditiontoreportingpoweroutages,smartmeterscanalsobeusedtoidentifypowerqualityissues.Thetermpowerqualitycanincludeanumberofissuessuchaschangesinsystemfrequency,voltage,orharmonics.Oneofpowerqualitythatappliestothefollowinganalysisisgivenin[28].Here,theauthorspowerqualityasanypowerproblemmanifestedinvoltage,current,orfrequencydeviationsthatresultsinfailureormisoperationofcustomerequipment.Thischapterwillprimarilyfocusonpowerqualityissuesrelatedtovoltagedeviationsaboveagiventhreshold.Theseissuesarethroughtheuseofsmartmetervoltagedata.TheelectricpowergridprovidespowertomillionsofcustomersandiscomposedofthreemainlevelsŠgeneration,transmission,anddistribution.Atthegenerationlevel,electricgeneratorspro-ducepoweratvoltagesbetween11and35kV.Thesevoltages,however,aretoolowtoeftransmitpoweroverlongdistances.Thetransmissionsystemusestransformersandotherequip-menttoincreasethegenerationvoltagetoalevelbetween60and765kV.Aspowertravelsthroughthetransmissionsystem,thevoltageisgraduallyreducedatsubstationsuntilitreachesthedistri-butionsystem.Manyresidentialandcommercialcustomersareconnectedtothedistributionsystem.Commer-cialcustomersarefedfromprimarydistributionlinesthatoperateatvoltagesbetween4and35kV.32Residentialcustomers,ontheotherhand,requirevoltageslessthanthoseattheprimarydistribu-tionlevel.AdistributiontransformerreducestheprimaryvoltagetoasecondarylevelŠtypicallyat120/240volts.Electricutilitiesarerequiredtoprovidepowertocustomerswithinacertainvoltagerange.ThepreferredservicevoltagerangeaccordingtotheANSIC84.1standardis5%ofthenominalvoltage[29].Ona120voltnominalsystem,thisrangeisbetween114and126volts.Exceptionsaremadeformomentaryvoltagedeviationsthatcanresultfromloadswitching,startingelectricmotors,orweathereventssuchaslightning.Thisstandardappliestovoltagesthatareoutsideofthethresholdforasustainedperiodoftime.Beforetheuseofsmartmeters,electricutilitiestookamorereactiveapproachtoaddressingvoltageissuesatthecustomerlevel.Theonlywayforautilitycompanytoknowifacustomerwasexperiencingvoltageissueswasthroughacustomercomplaint.Inatypicalscenario,acustomerwouldcalltheutilityafterexperiencingequipmentfailureornoticingdimmingoreringlights.Theutilitywouldthensendaserviceworkertothecustomers'hometofurtherinvestigatetheissue.Iftheserviceworkercouldnotdirectlydeterminethecauseoftheissue,apowerqualitymonitormaybeinstalledtorecordthecustomers'voltageoveranextendedperiodoftime.Iftheresultsindicatethatthereisavoltageproblem,stepswouldbetakentotheissue.Amajorofasmartmeterisitsabilitytomonitorvoltageatthecustomers'location.Withsmartmeters,utilitiescanproactivelymonitorvoltagelevelstoensurethatcustomersarereceivingvoltageswithinthetolerancerangeinANSIC84.1.Thenextsectiondescribesthemethodologyusedtoidentifyvoltageissuesusingsmartmeters.334.2Methodology4.2.1ServiceLinesandMeterTypesServicelinesareconductorsconnectedtothesecondarybushingsofadistributiontransformerthatprovidepowertoanelectriccustomer.Atypical120/240voltsingle-phaseresidentialservicehasthreeconductors;twofihotflwiresandoneneutralwire.Thevoltagebetweenanyhotwireandtheneutralis120volts.Thetwohotwiresareinsulatedtoavoidcontactwitheachother.Householdappliancessuchaselectricwaterheatersanddryersareconnectedbetweenthetwohotwires,whichprovideavoltageof240volts.Dependingontheserviceprovidedtothecustomer,therearedifferenttypesofmetersthatcanbeinstalled.Mostoftheresidentialcustomersinthissystemhavea1PH,3W,200A,240V,2Smeter.Theseabbreviationsindicatethatthemeterisasingle-phasemeter(1PH)thathasthreeservicewires(3W)enteringitsmetersocketatacurrentof200ampsandavoltageof240volts.The2Sreferstothefiformfltypeofthemeterwhichexplainsthewiringofthemetersocket.Figure4.1showsabasicform2SmetersocketFigure4.1:SimpleForm2S344.2.2ObtainingVoltageDataThevoltagedatausedinthisanalysiswasextractedfromtheOSIsoftPIsystem.TheOSIsoftPIsystemisadatahistorianthatstoresdatafromutilityassetssuchassmartmeters,capacitors,andothersensors[30].ThePIsystemallowsutilitiestoactivelymonitorassetdataandobtaininsightsfromhistoricaldata.Asmentionedinsection3.2.1,themajorityofsmartmetersdeployedinthissystemweredevelopedbyItronandGeneralElectric.ThePIsystemhasinterfacesforeachmetermanufacturerallowingthedatafromdifferentsmartmeterstobeeasilyintegratedintothehistorian.Figure4.2:PISystemScreenshotFigure4.2showsascreenshotofthedatabasethatcontainsalistofsmartmetersdeployedinthesystem.Eachmeter(orelement)shownhasseveralattributesassociatedwithit.Theseinclude35connectivityinformationsuchasthesubstation,feeder,andtransformerofthemeter,andintervaldatasuchashourlyvoltageorconsumptiondata.Atthetimeofthisanalysistherewereroughly700,000smartmetersinthedatabase.ThePIsystemhasanumberofclienttoolstoextractmeterdatafromthedatabase.ThemostcommonlyusedtoolsarePIProcessBook,PICoresight,andPIDataLink.PIProcessBookallowsuserstographicallyvisualizeattributedata.PICoresightisaweb-basedtoolusedtocreateandsharedisplays.Inthisthesis,PIDataLinkwasusedtoextractthesmartmetervoltagedata.PIDataLinkisaMicrosoftExceladd-inthatallowsuserstoimportPIsystemdataintoExcelforfurtheranalysis.UsingPIDataLink,anaveragevoltageoverave-dayperiod(fromOctober16thtoOctober21stof2015)wascalculatedforeachofthe700,000smartmeters.Themajorityofthesemeterswere120/240voltmetersthathadanominalvoltageof240volts.AccordingtotheANSIC84.1standard,thevoltageofthesemetersshouldremainwithin5%of240voltsorbetween228voltsand252volts.Inthisthesis,onlythecaseswheretheaveragevoltageexceededtheupperthresholdof252voltswereanalyzed.Afterthethemeterswithahighaveragevoltage,thestandarddeviationofthehourlyvoltagedatawascalculatedtoseeiftherewereanyindicationsoferroneousdata.Forexample,ifameterhadanaveragevoltageabove252voltsandalowstandarddeviation,thiswouldindicatethatthehourlyvoltagevaluesofthatmeterwereclosetotheaverage.Ontheotherhand,ifameterhadahighaveragevoltageandahighstandarddeviation,thiscouldindicatethatoneormoreofthehourlyvalueswerefarfromtheaveragevaluewhichwouldwarrantfurtherinvestigation.364.3AnalysisAfterobtaininganaveragevoltageforallofthemetersinthePIsystem,approximately1%ofthemeters(6,873)hadanaveragevoltagegreaterthan252volts.Table4.1showsthevoltagerangeofthemeterswithhighvoltage.Thesemetersweregroupedintothreevoltageranges.Therangecontainsthemetersthathadanaveragevoltagegreaterthanorequalto10%ofthenominal240volts.Thesecondrangecontainsthemeterswithanaveragevoltagebetween7and10%,andthethirdrangecontainsthemetersbetween5and7%.Themetersintherangeareexaminedincase1.Themetersinthesecondandthirdrangesareexaminedincases2and3,respectively.Table4.1:VoltageRangeofMeterswithHighVoltageVoltageRange(VRMS)PercentfromNominalVoltageNumberofMetersV264V10%50256.8V<2647%V<10%891252<V<256.85%<V<7%5,9324.3.1Case1Therewere50metersinthevoltagerange.Theaveragevoltageofthesemetersrangefrom264voltsto291volts.Figure4.3showsthedistributionofmetersinthisvoltagerange.Thestepinaddressingthesehighvoltageissueswastochecktoseeifanyofthemeterswereonthesamefeederorfedfromthesametransformer.Inthiscase,themeterswereconnectedto39circuitsand40differenttransformers.Mostofthesehighvoltageeventswereisolatedandresultedfromthefailureofatransformer.Oneexampleismeternumber30173601.Thismeterhadanaveragevoltageof286.5voltsandwastheonlymeterconnectedtothistransformer.Oncethismeterwasedashavinghighvoltage,thehourlyvoltagedatafromthedatethatthemeterwasinstalledwasobtainedtosee37Figure4.3:DistributionofMeterswithanAverageVoltage10%Figure4.4:VoltageforMeter30173601whenthevoltageexceededthethreshold.Figure4.4showsthevoltageofthismeterfromwhenitwasinstalledinAprilof2015.ThevoltageatthismeterremainedwithintheacceptabletolerancerangeupuntilJune22ndwhensomethinghappenedthatcausedthevoltagetoincrease38toaround268volts.Thevoltageremainedatthisleveluntilitincreasedagaintoaround288voltsinmidSeptember.OnOctober22nd,anelectricserviceworkerwassenttoinvestigatethevoltageandfoundthatthedistributiontransformerwasthecauseofthehighvoltage.Afterthistransformerwasreplaced,thevoltagereturnedtotheacceptablerange.Figure4.5:VoltageforMeter30172317Anotherexampleismeternumber30172317.Thismeterhadanaveragevoltageof279.4volts.Similartothelastexample,thismeterwastheonlymeterconnectedtothetransformer.Figure4.5showsthevoltageofthismeterfromwhenitwasinstalledinAprilof2015.ThevoltageatthismeterremainedwithinthetolerancerangeuntilsomethinghappenedonJuly15ththatcausedthevoltagetoincreasetoaround280volts.Afteritwasdeterminedthatadistributiontransformerwasthecauseofthehighvoltage,itwasreplacedandthevoltagereturnedtotheacceptablerangeonJanuary18th.Afterthebadtransformerwasreplaced,itwastakentoaservicecentertodeterminewhat39causedthetransformertooutputsuchahighvoltage.Theserviceworkerstriedtotesttheturnsratioofthetransformer,butwereunsuccessful.Thistransformer(1007343405)wasdesignedtohaveaturnsratioof20:1toreducethe4.8kVprimaryvoltageto240volts.However,whentheserviceworkerstookthetransformerapart,theyfounddamagetothewindingsofthetransformershowninFigure4.6.Thedamageloweredtheturnsratiofrom20:1toaround17:1whichexplainstheincreaseinvoltagefrom244voltsto280volts.Figure4.6:DamagedWindingsofTransformer1007343405Inadditiontotheexamplesabove,therewerealsoinstanceswheremultiplemeterswithhighvoltagewereconnectedtoasingletransformer.Oneexampleistransformernumber0812164402.Theaveragevoltageandstandarddeviationofthe6metersconnectedtothistransformerisshowninTable4.2.Asexpected,thevoltageofthesemeterswereroughlythesamefromwhenTable4.2:MeterswithHighVoltageconnectedtoTransformer0812164402MeterAverageVoltageStandardDeviationTransformerFeederID30179175274.8122.233081216440208180430179176274.7802.232081216440208180430179177274.7532.230081216440208180430179209274.6042.230081216440208180430179174274.5702.232081216440208180430179227274.3952.228081216440208180440theywereinstalledinAugustof2015.Thevoltageremainedaround270voltsuntilthetransformerwasreplaced,whichreducedthevoltagetowithin5%of240volts.4.3.2Case2Thiscaseexaminesthemetersinthesecondvoltagerangebetween7and10%.Figure4.7showsthedistributionofmetersinthisvoltagerange.Ofthe891meters,801(˘90%)werelocatedon10feeders.Themajorityofmetersonthesefeedershadhighvoltageduetothefailureofavoltageregulator.Theremaining90meterswereconnectedto40differentfeeders.Figure4.7:DistributionofMeterswithanAverageVoltagebetween7and10%Table4.3:Top10FeederswithMetersbetween7and10%FeederIDNumberofMeters05460125002280322501450215402280254069802360106033404100318075402140201018024503841Oneexamplewherethehighvoltageresultedfromthefailureofavoltageregulatorisfeeder054601.Themetersonthiscircuitwerefedfrom120differentdistributiontransformersthatwerealldownstreamofavoltageregulator.InFigure4.8,thetransformersareshownastrianglesandtheFigure4.8:GISMapforFeeder054601voltageregulatorisshownasacircle.ThehourlyvoltagedataatvedifferentmetersdownstreamoftheregulatorwasobtainedtoseeiftheyhadsimilarvoltageIfthemetershadasimilar42voltagethiswouldthattheregulatorwasthesourceofthevoltageissues.Iftheweredifferentateachlocation,thiscouldindicatethattheproblemwasrelatedtoeachtransformer.Figure4.9:DownstreamMeterVoltageforFeeder054601Figure4.9showsthevoltageofthevemeters.Althoughthemeterswereinstalledatdifferenttimes,thevoltageweresimilar.Thetapsofthevoltageregulatorwereadjustedseveraltimes(betweenmidOctoberandJanuary)toattempttoreducethehighvoltage.InJan-uarythisvoltagetrendwasreportedtothecircuitownerwhothefailureofthevoltageregulator.OnJanuary20th,theregulatorwasreplacedandthevoltagereturnedtotheacceptablerange.Anotherexampleisfeeder010603.The34metersthathadhighvoltagewereconnectedto23differenttransformersandhadanaveragevoltagebetween259voltsto263volts.Figure4.10showsthelocationofthetransformersinrelationtothevoltageregulator.The23transformers43Figure4.10:GISMapforFeeder010603werealldownstreamoftheregulatorandconnectedtotheZphase.ThevoltageoffourmetersatdifferentlocationswasobtainedandallfourhadasimilarvoltagewhichisshowninFigure4.11.InNovemberof2015,aserviceworkerwassenttochecktheregulatorandfoundthattheZphasewasstuckonahighertapsetting.OnNovember30th,thevoltageregulatorwasrepairedandthevoltagereturnedtotheacceptablerange.44Figure4.11:DownstreamMeterVoltageforFeeder0106034.3.3Case3Thelastcaseaddressesthemetersinthethirdvoltagerangebetween5and7%.Figure4.12showsthevoltagedistributionofthe5,932meters.TheaveragevoltageofthesemetersrangefromFigure4.12:DistributionofMeterswithanAverageVoltagebetween5and7%45252.0002voltsto256.799volts.Manyofthesemetershadanaveragevoltagethatwasrightattheupperlimitofthe252volt(240V+5%)thresholdintheANSIC84.1standard.Therewere2,786meterswithanaveragevoltagewithin1voltofthethresholdand4,410meterswithanaveragevoltagewithin2volts.Thesemetersaccountforabout75%ofthemetersinthisvoltagerange.Themetersinthisvoltagerangewereconnectedto184differentfeeders.Thirty-twoofthesefeederswerealsoincase2.Thisisattributedtohowclosethelowervoltagelimitofcase2istotheuppervoltagelimitofcase3.46Chapter5ConclusionInthisthesis,ananalysisofsmartmeteroutageandvoltagedatawaspresented.InChapter3,theoutageandrestorationeventsofsmartmeterswerecomparedwiththeoutageincidentsofanoutagemanagementsystem.Theresultsfromtheoutageeventcomparisonshowthatwhilesmartmetereventscanbeusedasanotherwaytonotifyelectricutilitiesofpoweroutagesthereareseveralchallengesthatmustbeovercomebeforetheseeventscanbeintegratedintooutagemanagementsystems.OnechallengeliesinensuringthattheelectricGISmodelisuptodate.ThisincludesmakingsurethatthemetertotransformertophaseconnectivityinthematchestheelectricmodelinGIS.Duringaseverestorm,devicessuchasdistributiontransformerscanbereconnectedtoadifferentphasewhichinsomecasesisnotupdatedintheGISmodel.Anotherchallengeliesinprocessingmomentarysmartmeteroutageevents.Intheeventofanoutagethatcausesareclosertooperate,theoutagemanagementsystemcangetwithsmartmeteroutagesthatonlylastforafewseconds.Theoutagemanagementsystemshouldbecapableofoutthesemomentaryoutages.Onewaytotheseoutagesistohavetheoutagemanagementsystemwaitafewminutesbeforereceivingsmartmeterevents.InChapter4,smartmetervoltagedatawasusedtoidentifyinstanceswherethevoltagedeliv-eredtoacustomerwasaboveagiventhreshold.The1%ofmetersthatwerefoundweregroupedintothreecategories.Thethreevoltagerangeswerechosentogroupthemetersintermsofsever-itysothatthemeterswiththehighestvoltagecouldbeaddressedfollowedbytheremaining47meters.Fromtheanalysis,manyofthemeterswiththehighestvoltagewereconnectedtoabaddistributiontransformer.Therewerealsoinstanceswhereafailingvoltageregulatorwasthesourceofthehighvoltagesensedbythesmartmeter.Manyofthemetersthathadamoreseverehighaveragevoltageduringtheve-dayanalysisperiodhadahighvoltageforasustainedperiodoftime.Inmostcases,thecustomerwasunawareofthehighvoltageissue.Highvoltageissuesaretypicallylessnoticeablethanlowvoltage.Theeffectsoflowvoltagecanincludebrownoutsordimminglightsthatwouldpromptthecustomertocalltheutilitytoreporttheissue.Highvoltagecanalsodamagecustomerequipment,buttheeffectsusuallygounnoticed.Infuturework,bothhighandlowvoltagewillbeanalyzedtoensurethatcustomersreceivepoweratvoltagessuitableforutilization.48APPENDIX49AppendixPythonCode#Importingpythonlibrariesfordataanalysisimportpandasaspdfrompandas.tseries.offsetsimport*fromodoimportodo#Readingsmartmeterdata(outagesandrestorations)fromcsvfiles#"converters={...}"isusedtoconvertselectcolumnstoastring.GD_outages=pd.read_csv("GD_outages.csv",converters={"ACCOUNT":lambdax:str(x),"FEEDER_ID":lambdax:str(x),"TLM":lambdax:str(x),"METER":lambdax:str(x),"Street1":lambdax:str(x)},parse_dates=["DateTime"])GD_restorations=pd.read_csv("GD_restorations.csv",converters={"ACCOUNT":lambdax:str(x),"FEEDER_ID":lambdax:str(x),"TLM":lambdax:str(x),"METER":lambdax:str(x),"Street1":lambdax:str(x)},parse_dates=["DateTime"])#ReadingOMSdatafromcsvfilesOMS_outages=pd.read_csv("incidentlist.csv",converters={"INCIDENT_ID":lambdax:str(x),"FEEDER_ID":lambdax:str(x)},parse_dates=["TIME_OUTAGE","TIME_RESTORED"])predicted_OMS_customers=pd.read_csv("Customers_per_IncidentID.csv",converters={"ACCOUNT":lambdax:str(x),"METER":lambdax:str(x),"FEEDER_ID":lambdax:str(x),"INCIDENT_ID":lambdax:str(x)})#Createanemptylisttostorethecomparisonresultscomparisondatalist=[]#Foreachoutage/incidentinthe"OMS_outages"table:forincidentinOMS_outages["INCIDENT_ID"]:#FindthecustomerspredictedbyOMS:OMS_customers=predicted_OMS_customers[predicted_OMS_customers.INCIDENT_ID==incident]#Getthefeeder_id,outageandrestorationtimefortheOMSincident:OMS_outage_time=OMS_outages[OMS_outages.INCIDENT_ID==incident]["TIME_OUTAGE"]OMS_restoration_time=OMS_outages[OMS_outages.INCIDENT_ID==incident]["TIME_RESTORED"]feeder_id=OMS_outages[OMS_outages.INCIDENT_ID==incident]["FEEDER_ID"]#Specifytimerangeforsmartmeteroutage/restorationevents(2hoursbeforeand1houraftertheOMS#outage/restoration):time_b4_outage=OMS_outage_time-DateOffset(hours=2)time_aftr_outage=OMS_outage_time+DateOffset(hours=1)time_b4_restoration=OMS_restoration_time-DateOffset(hours=2)time_aftr_restoration=OMS_restoration_time+DateOffset(hours=1)#Findthesmartmetersthatreportedoutages/restorationsbetweenthetimerange#andonthesamefeederastheOMSincidentoutages_frm_GD=GD_outages[(GD_outages.DateTime>time_b4_outage.values[0])&(GD_outages.DateTime<time_aftr_outage.values[0])&(GD_outages.FEEDER_ID==feeder_id.values[0])]restorations_frm_GD=GD_restorations[(GD_restorations.DateTime>time_b4_restoration.values[0])&(GD_restorations.DateTime<time_aftr_restoration.values[0])&(GD_restorations.FEEDER_ID==feeder_id.values[0])]#Sorttheoutage/restorationtimesanddropany"lastgasp"events:outages_frm_GD_sorted=outages_frm_GD.copy().sort_values("DateTime",axis=0,ascending=True)restorations_frm_GD_sorted=restorations_frm_GD.copy().sort_values("DateTime",axis=0,ascending=True)50smart_meter_outages=outages_frm_GD_sorted.copy().drop_duplicates("ACCOUNT")smart_meter_restorations=restorations_frm_GD_sorted.copy().drop_duplicates("ACCOUNT")#CombinetheOMScustomersandthesmartmeteroutagesintoonetable:combined_outages=pd.concat([OMS_customers,smart_meter_outages])#Findthenumberoftimesanaccountnumberappearsinthecombinedtable:out_acct_number_count=combined_outages.ACCOUNT.value_counts()#Findtheaccountnumbersthatappearedtwiceinthetable.Ifitappearstwice,#itmeansthatthemeterwasinitiallyinboththeOMSandSmartMeterOutagetables:find_dupl_outage_acctnums=out_acct_number_count[out_acct_number_count==2].indexoutages_that_matched_OMS=combined_outages[(combined_outages.ACCOUNT.isin(find_dupl_outage_acctnums))&(combined_outages.Code.notnull()==True)]#CombinetheoutagesthatmatchedOMSwiththesmartmeterrestorationstofindthe#restorationtimesofthemetersthatmatched:combined_restorations=pd.concat([outages_that_matched_OMS,smart_meter_restorations])rest_acct_numb_count=combined_restorations.ACCOUNT.value_counts()find_dupl_rest_acctnums=rest_acct_numb_count[rest_acct_numb_count==2].indexrestor_that_matched_OMS=combined_restorations[(combined_restorations.ACCOUNT.isin(find_dupl_rest_acctnums))&(combined_restorations.Name.notnull())]#Combinetheoutageandrestorationtimesofthematchingsmartmetersintoonetable:Smart_Meters_that_Matched_OMS=outages_that_matched_OMS.merge(restor_that_matched_OMS,on="ACCOUNT")[["ACCOUNT","METER_y","DateTime_x","DateTime_y","FEEDER_ID_x","TLM_x"]]#Renamecolumns:Smart_Meters_that_Matched_OMS.columns=["ACCOUNT","METER","OUTAGE_TIME","TIME_RESTORED","FEEDER_ID","TLM"]#Findtheuniqueaccountnumbers,(i.e.theaccountnumbersthatonlyappearonce,#eitherintheOMStableortheSmartMeteroutageTable):unique_acct_num=out_acct_number_count[out_acct_number_count==1].index#Ifitdoesn'thaveaeventcode,itwasoriginallyintheOMSoutageTable.Mtrs_in_OMS_not_in_SmrtMtrtbl=combined_outages[(combined_outages.ACCOUNT.isin(unique_acct_num))&(combined_outages.Code.notnull()==False)][["ACCOUNT","METER","TIME_OUTAGE","TIME_RESTORED","Code"]]#FindtheSmartmetersthatwereintheSmartMeterOutageTablebutnotintheOMScustomertable#IfithasaCode,itwasoriginallyintheSmartMeterOutageTableMtrs_in_SmrtMtrtbl_not_in_OMS=combined_outages[(combined_outages.ACCOUNT.isin(unique_acct_num))&(combined_outages.Code.notnull()==True)]#Findoutageandrestorationtimesoftheremainingsmartmetersremaining_outages=Mtrs_in_SmrtMtrtbl_not_in_OMS.copy()remaining_restorations=smart_meter_restorations[smart_meter_restorations.ACCOUNT.isin(remaining_outages.ACCOUNT)]#Combinethetwotablestogether:Smart_Meters_that_didnt_Match_OMS=remaining_outages.merge(remaining_restorations,on="ACCOUNT")[["ACCOUNT","METER_y","DateTime_x","DateTime_y","FEEDER_ID_x","TLM"]]#Foreachincident,appendthecomparisonresultstotheinitialempty#list("comparisondatalist"):comparisondatalist.append((incident,feeder_id.values[0],len(OMS_customers),len(Smart_Meters_that_Matched_OMS),len(Mtrs_in_OMS_not_in_SmrtMtrtbl),len(smart_meter_outages),len(smart_meter_restorations)))#StorethecomparisonresultsofeachincidentinanewtableComparison_Results=pd.DataFrame(comparisondatalist,columns=["IncidentID","FeederID","OMSCustomers","#ofSMsthatmatchedOMS","#ofMtrsinOMSthatdidnotcommunicate","Total#ofSMsthatreportedanoutage","Total#ofSMsrestored"])#Writetheresultsofthattabletoa.csvfileodo(Comparison_Results,"Comparison_Results.csv")51BIBLIOGRAPHY52BIBLIOGRAPHY[1]U.S.DepartmentofEnergy.WhatistheSmartGrid?Available:https://www.smartgrid.gov/thesmartgrid/smartgrid.html[2]U.S.DepartmentofEnergy.fiEnablingModernizationoftheElectricPowerSystemflinQTR:AnAssessmentofEnergyTechnologiesandResearchOpportunities.Available:http://energy.goTechnology-Review-20150.pdf,pp.53,Sept.2015.[3]InternationalEnergyAgency.TechnologyRoadmap:SmartGrids.Available:https://www.iea.org/publications/freepublications/publication/smartgridsroadmap.pdf,pp.18-19,2011.[4]U.S.DepartmentofEnergy.2014SmartGridSystemReport.Available:http://energy.gov/SystemReport2014.pdf,pp.7-8,2014.[5]AdvancedMeteringInfrastructureandCustomerSystems.Available:https://www.smartgrid.gov/recoveryact/deploymentstatus/sdgpamisystems.html[6]U.S.DepartmentofEnergy.SmartGridInvestmentGrantProgramProgressReportII.Avail-able:https://www.smartgrid.goprogressreport2013.pdf,Oct.2013.[7]U.S.DepartmentofEnergyOfofEnergyDeliveryandEnergyReliability.SmartGridInvestmentsImproveGridReliability,Resilience,andStormResponses.Available:https://www.smartgrid.goMaster-File-with-edits120114.pdf,pp.8-9,2014.[8]J.Cowdrey,fiTheWaroftheCurrentsflinHomePowerMagazine,vol.111,Available:http://www.uvm.edu/˘AE06/readings/ACDC.pdf,pp.88-92,2006.[9]C.R.Riker,fiTheEngineeringEvolutionofElectricalApparatus-XXXV:TheHistoryoftheWatt-HourMeterinAmericaflinTheElectricJournal,vol.15,pp.504-509,1919.[10]A.Madrigal,MeasuringtheHistoryofElectricity,Image:LibraryofCongress.Available:http://www.wired.com/2008/09/gallery-electrical-meter/[11]O.T.Bl´athy,fiElectricMeterforAlternatingCurrents,flU.S.Patent423210,March11,1890.[12]U.S.EnergyInformationAdministration.AnAssessmentofIntervalDataandTheirPotentialApplicationtoResidentialElectricityEndUseModeling.Available:https://www.eia.gov/consumption/residential/reports/smartmetering/pdf/assessment.pdf,pp.15-19,Feb.2015.53[13]Itron.OpenWayCentronMeter.Available:https://www.itron.com/PublishedContent/OpenWay%20Centron%20Meter.pdf,2016.[14]GEDigitalEnergy.ResidentalElectricalMetering.Available:http://www.choptankelectric.com/sites/choptank2015.[15]K.S.KWeranga,S.Kumarawadu,andD.P.Chandima,fiEvolutionofElectricityMeters,flinSmartMeteringDesignandApplications.Singapore:Springer,2014,ch.2,pp.17-33.[16]J.Ekanayake,fiSmartMeteringandDemand-SideIntegration,flinSmartGrid:TechnologyandApplications,1sted.Chichester,UK:Wiley,ch.5,sec.5.3.1,pp.87-88,2012.[17]M.Conner,Tamper-resistantsmartpowermetersrelyonisolatedsensors.Available:http://www.edn.com/design/power-management/4313844/Tamper-resistant-smart-power-meters-rely-on-isolated-sensors,2009.[18]X.Jiang,fiDesignandImplementationofaHighFidelityACMeteringNetwork,flinInt.Conf.onInformationProcessinginSensorNetworks,SanFrancisco,CA,pp.3-4,2009.[19]D.A.WardandJ.LaT.Exon,fiUsingRogowskicoilsfortransientcurrentmeasurements,flEng.Sci.Edu.J.,vol.2,no.3,pp105-113,Jun.1993.[20]V.SkendzicandB.Hughes,fiUsingRogowskiCoilsInsideProtectiveRelays,flinProc.66thAnn.Conf.forProtectiveRelayEngineers,CollegeStation,TX,pp.1-10,Apr.2013.[21]L.DallyandR.Edwards,fiEvaluatingtheCommunicationsComponentofYourSmartGridSolution,flLandis+Gyr,Available:http://www.landisgyr.com/resources/evaluating-the-communications-component-of-your-smart-grid-solution/,Dec.2013.[22]U.S.DepartmentofEnergy.OperationsandMaintenanceSavingsfromAdvancedMeteringInfrastructure-InitialResults.Available:http://energy.goSavingsDec2012Final.pdf,pp.4-5,Dec.2012.[23]PublicTechnologyInstitute.SmartGrid101forLocalGovernments.Available:http://www.cadmusgroup.com/wp-content/uploads/2013/03/SmartGrid101forLocalGovernments201112.pdf,p.6,2011.[24]EdisonElectricInstitute.SmartMetersandSmartMeterSystems:AMeteringIndustryPerspective.Available:http://www.eei.org/issuesandpolicy/grid-enhancements/documents/smartmeters.pdf,pp.11-12,Mar.2011.54[25]NationalEnergyTechnologyLaboratory.NETLModernGridStrategyPoweringour21st-CenturyEconomy:AdvancedMeteringInfrastructure.Available:https://www.smartgrid.goSGInteropReportPostcommentperiodversion200808.pdf,pp.8,11,Feb.2008.[26]PythonDataAnalysisLibrary,availableat:http://pandas.pydata.org/[27]W.Mckinney,fiGettingStartedwithpandas,flinPythonforDataAnalysis:DataWranglingwithPandas,NumPy,andIPython,1sted.Sebastopol:O'Reilly,2013,ch.5,pp.107-153.[28]R.C.Dugan,fiIntroductionflinElectricalPowerSystemsQuality,3rded.NewYork:McGraw-Hill,2012,p.3.[29]AmericanNationalStandardforElectricPowerSystemsandEquipmentŠVoltageRatings(60Hz),ANSIStandardC84.1-2016.[30]OSIsoftPISystem,availableat:http://www.osisoft.com/pi-system/#tab255