substation smart meter integration - flexmeter...

FLEXMETER 646568 D3.1 Substation smart meter integration H2020-LCE-2014-3

Public

H2020-LCE-2014-3 FLEXMETER

Flexible smart metering for multiple energy vectors with active prosumers

Project Duration 2015-01-01 – 2017-12-31 Type CP

WP no. Deliverable no. Lead participant

WP3 D3.1 IREN

Substation smart meter integration

Prepared by IREN

Issued by FLEXMETER Project Office Document Number/Rev. FLEXMETER/D3.1/V2.0 Classification Public Submission Date 2016-02-02 Due Date 2015-12-31

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no. 646568

©Copyright 2010 POLITECNICO DI TORINO, IREN ENERGIA SPA, STMICROELECTRONICS SRL,TELECOM ITALIA, RHEINISCH-WESTFAELISCHE TECHNISCHE HOCHSCHULE AACHEN, INSTITUTPOLYTECHNIQUEDEGRENOBLE,UNIVERSITATEAPOLITEHNICADINBUCURESTI,SIVECOROMANIASA,ALMAMATERSTUDIORUM–UNIVERSITA’DIBOLOGNA,E-ONSVERIGEAB.This document and the information contained hereinmay not be copied, used or disclosed inwholeor inpartoutsideof theconsortiumexceptwithpriorwrittenpermissionof thepartnerslistedabove.


Document Title Substation smart meter integration

Type Deliverable

Ref D3.1

Target version V1.0

Current issue V0.1

Status Final

File D3.1_flexmeter_v2.0.pdf

Author(s)

Reviewer(s) All consortium, UPB

Approver(s) Andrea Acquaviva

Approval date 2016-02-01

Release date 2016-02-02

Distribution of the release

Dissemination level PU

Distribution list All consortium

History of Changes Date Version Comments

2015-12-12 0.1 Creation

2016-18-01 1.0 First version

2016-31-01 1.1 Second version

2016-02-01 2.0 Final version


1. Executivesummary..............................................................................................................................4

1.1. General..............................................................................................................................................4

1.2. Deliverablestructure.........................................................................................................................4

2. SmartmeterintegrationintheMV/LVsubstationsinTurin..............................................................4

2.1. Smartmeterscharacteristics.............................................................................................................4

2.1.1. Criteriaforchoosingthebasetechnologyforthesmartmeter....................................................4

2.1.2. Smartmetersoverview.................................................................................................................5

2.1.3. STCOMETsmartmeterandpowerlinecommunicationsystem-on-chipdescription..................6

2.1.4. Safetyrecommendations..............................................................................................................7

2.1.5. Three-phasemetrologyevaluation...............................................................................................7

2.2. ObjectivesoftheintegrationofthesmartmeterintheMV/LVsubstation.....................................9

2.3. ApplicationofNIALMtechnology......................................................................................................9

3. SmartmeterintegrationintheMV/LVsubstationsinMalmö.........................................................11

3.1. Smartmeterscharacteristics...........................................................................................................11

3.2. ObjectivesoftheintegrationofthesmartmeterintheMV/LVsubstation...................................13

4. Meterdataaggregation......................................................................................................................13

4.1. Introduction-dataaggregationasinformationcompression........................................................13

4.2. Aggregationintimedomain............................................................................................................16

4.3. MeasurementQualityandReportingRate.....................................................................................18

4.4. RandomenssPower–ameasureofheterogenousreportingrates[15]........................................19

4.5. Aggregationofwideareameasurements.......................................................................................20

4.6. AggregationparadigmandSmartMeteringtechnology[19].........................................................21

5. UsecasesrelatedtotheintegrationofsmartmetersintheMV/LVsubstations............................22

5.1. TheusecasesforoutagedetectionandoutagelocationinLVnetworks.......................................22

5.2. TheusecasesforfaultlocationinMVnetworks............................................................................23

4.3 TheusecasesforNetworkmanagementforloadbalancing..........................................................24

6. References..........................................................................................................................................25


1. Executivesummary1.1. General

FLEXMETERprojecthastheaimtodevelopandtoinstallaflexiblesmartmeteringarchitectureformultipleenergyvectors.AnimportantstepoftheprojectistointegratesmartmetersintheMV/LVsubstation,inordertoperformalgorithmsenablinganenergybalanceinthesubtendednetwork,loadnon-intrusivemonitoring(NILM),LValarmstestingincaseoffaultsandoutagesorfeeders’overload.

1.2. Deliverablestructure

Deliverable3.1hastheobjectivetocarryoutanoverviewofthemainobjectivesreachedbytheintegrationofthesmartmetersintheMV/LVsubstation.

Thedocumentisstructuredin4Chapters:

• Chapter1introducestheobjectiveoftheFLEXMETERprojectandthestructureofthedocument.• Chapter2and3willmakeanoverviewofthesmartmeterintegrationintheMV/LVsubstationsin

termsofsmartmeterscharacteristics,ofthegoalsreachedthroughouttheirinstallationandoftheapplicationofNIALMtechnology,fortheItalianpilot.

• Chapter4makesanoverviewofthemeterdataaggregation.• Chapter 5 reports the different use cases about the installation of smart meters in the MV/LV

substation.• Chapter6reportsthereferencesofthedocument.

Atthemomenttheprojectisatanearlystageinwhichthemeters,mainlyintheItaliantestsite,havenotyetbeeninstalled,sothisdocumentwillbeupdatedwiththeresultsachieveddownstreaminstallationofthemetersoranewdocumentwillbedrawnupwhichcontainingtheseresults.

2. SmartmeterintegrationintheMV/LVsubstationsinTurin2.1. Smartmeterscharacteristics

2.1.1. Criteriaforchoosingthebasetechnologyforthesmartmeter

BasedonD.1.1,thatisgivingthespecificationonthedistributionnetwork,itwasdecidedtobasetheSmartMeters at substation level on the ST Comet technology. STwill provide the Comet evaluation board onwhich tobuildaround the smartmeter thatwillbeused for theproject. STwillprovide,nevertheless, afirmware instance of communication while ST and POLITO will contribute to customize the providedfirmware to match the technical specifications stated on deliverable D1.1. Further it is described theenablingtechnologyofferedbySTtotheproject,inordertheachievealltheabovespecifications.


2.1.2. Smartmetersoverview

StartingfromEVLKSTCOMET10-1,anevaluationboardofSTCOMET,whichisactuallyasmartmeter,STwillgiveschematics,project,Gerberfiles,andallisneededforthedesignofthefinalboardoftheSmartMeter.The STCOMET platform development environment is a set of SW, HW and FW tools and associateddocumentationavailablefromSTorthirdparties,conceivedtoeasetheevaluationanddevelopmentflowofanSTCOMETbasedsmartmeterdesign.TheEVLKSTCOMET10-1,showninFigure2.1,isadevelopmentkitfortheSTCOMETplatform,exploitingtheperformancecapabilityofthefull-featureSTCOMET10device.TheSTCOMET10isasingledeviceintegratingaflexiblepowerlinecommunication(PLC)modemwithafullyembeddedanalogfrontend(AFE)andalinedriver,ahighperformance3-channelmetrologyfunctionandaCortex™-M4applicationcore.

Figure2.1:STCOMET10developmentkit(EVLKSTCOMET10-1)

Fromthe functionalpointof view, theSTCOMETdevelopmentkit (EVLKSTCOMET10-1) is composedbyamainboard implementing a complete smartmeterwithpower line communication (PLC) andmetrologyfunctionalityandaseparatepowersupplyboardbasedontheVIPER26HAC-DCconverter.Withacompleteaccess to the STCOMETGPIOs andperipherals and theon-board SEGGER® J-Link®probe it is specificallydesignedtofacilitateHW/SWdevelopment.InFigure2.2isshowntheSTCOMETdevelopmentkit.


Figure1.2:STCOMETdevelopmentkit-functionalblockdiagram

Pleaserefertothislink:

http://www.st.com/web/catalog/tools/FM116/CL2207/SC1078/PF262863?s_searchtype=partnumber todownloaddatasheet,applicationnoteandUserManual.

2.1.3. STCOMETsmartmeterandpowerlinecommunicationsystem-on-chipdescription

The STCOMET is a device that integrates a narrow-band power line communication (NB-PLC)modem, ahigh-performanceapplicationcoreandmetrologyfunctions.Themetrologysub-systemissuitablefortheEN50470-1,EN50470-3,IEC62053-21,IEC62053-22andIEC23compliantclass1,class0.5andclass0.2ACmetering applications. With its multi core fully programmable architecture, the STCOMET platform cansupport a range of PLC standard protocols. The PLC line coupling interface is designed to allow theSTCOMET device to transmit and receive on the AC mains line using any narrow-band PLC modulation(singlecarrierorOFDM)upto500kHz,mainlyforautomaticmeterreading(AMR)applications.ThedefaultconfigurationofthePLClinecouplingtargetstheG3-PLC(ITUG.9903)andPRIME(ITUG.9904)CENELECA-bandprotocolstandards.,TheSTCOMETdevelopmentkitcanbealsoadjustedtofitothernarrow-bandPLCprotocolsinCENELECA-bandorFCCband(e.g.:S-FSKIEC61334-5-1,IEEE1901.2,G3-PLCFCC,METERSANDMORE®).Ifnecessary,aspecificcustomer'smodulecanbedesignedandplacedinsteadoftheLCDmodule,foradifferentperipheralsconfiguration.InFigure2.3isshowntheSTCOMETblockdiagram.


Figure2.3:STCOMETblockdiagram

2.1.4. Safetyrecommendations

TheSTCOMETdevelopmentkitmustbeusedbyexperttechniciansonly.Duetothehighvoltage(85-265Vac)presentonthenon-isolatedparts,specialcaremustbetakeninordertoavoidelectricrisksforpeoplesafety.Therearenoprotectionsagainsthighvoltageaccidentalhumancontact.Afterdisconnectionoftheboard from the mains all the live part must not be touched immediately because of the energizedcapacitors.Itismandatorytouseamainsinsulationtransformertoperformanytestsonthehighvoltagesections,usingtestinstrumentslike,forinstance,spectrumanalyzersoroscilloscopes.One should not connect any probe to high voltage sections if the board is not isolated from themainssupply,inordertoavoiddamaginginstrumentsanddemonstrationtools.Whenconfiguredformeteringevaluation,theSTCOMETdevelopmentkitisnotisolatedandgroundwillbetiedtotheline.Onshouldnotconnectinstrumentprobesthatcanbringtheearthconnectiontotheline,thuspotentiallydamagingtheSTCOMETdevelopmentkitandtheinstrumentsandcreatingelectricalrisk.

2.1.5. Three-phasemetrologyevaluation

TheSTCOMETdevelopmentkitprovidesanSPI/UARTinterface(J2connector)andgeneralpurposesignals(J3 connector) on the LCDmodule (Figure 2.5) to connect STPMxxmetrology boards in order to build athree-phasemeterdevelopmentkit.Figure2.4reportstheJ2andJ3pinoutplustheconfigurationjumpersto select between the UART and SPI connection to the external STPMxx board. Table 1 describes thejumperconfigurationtoselectSPIorUARTconfiguration.


Figure2.4:Three-phasemetrologyevaluation-digitalconnectionstoSTPMxxevaluationboards

Table1:Three-phasemetrologyevaluation-SPI/UARTconfiguration

Jumper SPIconfiguration UARTconfigurationJ15 Close OpenJ35 Open CloseJ36 Close OpenR50 NP 1KΩ

Figure2.5:LCDmoduledrawingwithindicationofthevarioussections


2.2. ObjectivesoftheintegrationofthesmartmeterintheMV/LVsubstation

ThroughouttheintegrationofthesmartmetersintheMV/LVsubstation,chosenfortheproject,itispossibletoreachthefollowinggoals:

• disaggregatetransformerloadinordertocarryouttheloadbalancingofthenetwork;• analyzetheabnormalenergyflowsinordertodetectenergytheftsalmostinrealtime;• detectandlocatefaultsandoutagesintheLVnetwork,throughoutappropriatealarms;• analyzethepossiblelossesinthegrid;• supplyinformationaboutthemonitoredsubstationsandthepowerquality;• monitorinquasi-realtimeandmoreoptimallythesystem.

2.3. ApplicationofNILMtechnology

TheprojectrequirestheinstallationofmeasurementequipmentoneachLVderivation,namelyupstreamofallLVcostumerssuppliedbythederivation.

Throughtheanalysisofdatagatheredtheobjectiveisdevelopingalgorithmsableto:

• disaggregatetransformerload;• automaticallydetectandlocatefaultsandoutagesinthelowvoltagedistributiongrid;• provideinformationaboutthemonitoredsubstationsandpowerquality.

AlgorithmswillbedevelopedonaMatlabsimulator,oncethealgorithmswillbesufficientlyreliabletheywillbeconvertedintoanon-linetool.

Referringtothedisaggregatetransformerload,loaddisaggregationwillbeobtainedapplyingNILM-based(Non-Intrusive Load Monitoring) signal processing techniques at the data gathered by power meterinstalledon theLVderivationof theMV/LV transformer.NowadaysNILM iswidely studiedathouseholdlevel: itallowstheenergydisaggregationandthe loadsrecognitionthroughtheanalysisoftheaggregatepowerconsumptionmeasured inasinglepoint.The loadrecognition ispossible thanks to loadsignaturewhich is defined as the electrical behavior of an individual appliance/piece of equipment when it is inoperation. Similar to any human’s signature, each electrical device contains unique features in itsconsumptionbehavior.

For the purpose of the Flexmeter project the extension of these techniques to higher levels in thedistributiongridbecomesnecessary.

Energydisaggregationallows,foreachtransformerderivation,tobreakdowntheloadsamongeachtypeofconsumers.Byknowingthepercentagerespecttothetotalitispossibletoidentify,foreachtypeofload,thedaily, seasonalandyearlyconsumptiontrends.Theknowledgeof loadtrendsallows tomake furtherconsiderationsaboutthesubstationcapabilitytomeetdemandandplanningcorrectiveactionsinordertoupgradecomponents toensure systems long-termability tomeetdemand (e.g. if ina certain substationthe demand is becoming greater than the transformer capacity, it is possible to take corrective actionbefore fault happen). It is also possible to correlate load trends with environment variables (at leasttemperature)toforecastthevariationofloadsbehaviorincaseofexternalchanging.Loaddisaggregation


also allows to highlight the portion of demand manageable load and the effect of demand sidemanagementonpowersystemperformance.

FromFigure2.6youcannoticethateachfinalcostumerownsacertainquantityofindividualloadtype,i.e.theindividualdevice.Asthenumberofindividualloadtypesfarexceedsthereasonablelevelwhichcanbeconsidered for the application ofNILM, it is not practical disaggregating the consumption among all theindividualloadtypesinstalledattheconsumerlevel,consequentlywedecidedtosubdividetheminasmallnumberofloadcategories.

Aloadcategoryisdefinedasagroupofelectricaldeviceswhichmayhavedifferentend-usesbut,forourpurpose,devicesbelongingat thesame loadcategoryhavethesame,orsimilar,electricalcharacteristics[1].Atthemoment,forthepurposeofthestudyitwasdecidedtoconsidertheloadcategoriesshowedinFigure2.7.

Figure2.6:organizingindividualloadtypesintoloadcategories

Figure2.7:mainLoadCategories[1]

Electricalloadsinallsectorsofelectricityconsumers(industrial,commercial,andresidential)aretypicallyamix of linear (e.g. resistive heaters, directly connectedmotors, incandescent lamp) and non-linear loads


(e.g.adjustablespeeddrives,powerelectronics,compactfluorescent lamps).The latterareallsourcesofcurrentharmonicshencewewillconsideracombinationofthetimedomainsignaturesnormallyusedfordomesticNILMwithfrequencydomainfeature.

The load harmonic spectrum could be obtained by applying Fourier Transform at themeasurements ofcurrent.Thisinformationconstitutesasignatureusefultoidentifyloadcomposition.Inordertoidentifytheharmonic spectrum it is necessary toalsouse the sampling frequency for current andvoltage to severalkHz.Aminimalqualityofthemeasurementset-up, includingthereportingrate, ismandatory inordertohavemeaningfulresults.Flexmeterwillensurethisbyusingtheproposedsmartmeteringplatform

HarmoniccontainmentisusedforNILMalsoincontextmorecomplicatedi.e.industrialbuildings.In[2]isstatedthattheharmonicdistortionsprovideatremendouslyimprovementinrecognitioncapacity.

However,itisnecessaryverifyingthefeasibilityofextendingthisapproachalsoatthesecondarysubstationlevel.Infactliteraturereviewallowshighlightingsomeissuesrelatedtothepropertiesofharmonicswhenalotofnon-linearloadareworkingatthesametime.Thesepropertiescouldmakedifficulttogobacktotheactualloadcomposition(intermsofloadcategories)bytheonlymeansofharmonicmeasurementatthebeginningofLVfeeder.Theseissuesaresummarizedbelow:

1 Theharmoniccurrentproducedbyaggregatenon-linear loads isusuallysignificantlysmallerthanthearithmeticalsumoftheharmoniccurrentsproducedbythe individualnon-linear load,mainlyduetophasecancellation.Consequentlyinformationabouttheoperationofcertainloadscouldbelostduetotheharmoniccancellation.

2 Thenumberandpowerratingofdevicesconnectedtothefeederareimportantparameterswhichhaveanimpactontheharmoniccompositionofaggregateload,buttheseparameterscouldnotbeeasilyavailable.

3 Underpollutedvoltagecondition,prochangesinthedevicecurrentspectrummayhappen.

Finally, considering the abovementionedpoints, in order toutilize the information contained in currentharmonicsasloadsignaturefortheloadidentification,itisnecessary,foreachloadcategory,todeterminetheshapeofthecurrentspectrumattheMV/LVtransformerlevel.

Itisnecessarytoinvestigatethepossibilityofovercomingtheabovementionedissues,inordertorealizeareliable load signature for each load category. Starting from the harmonic behavior of single devices, astudy on the modified current harmonic content when multiple devices, belonging to different loadcategories,areoperatingsimultaneouslybecomesnecessary.

3. SmartmeterintegrationintheMV/LVsubstationsinMalmö

3.1. Smartmeterscharacteristics

The electricity grid of the pilot area in Hyllie today essentially consist of two 10 kV radials withinterconnectionpossibilities,seefigurebelow.


Figure3.1–electricitygridofthepilotarea

Substation nr 4 and 10 is used for the Flexmeter project with a total of approx. 350 metering pointsconnectedtothesetwosubstations.Itisresidentialcustomersandsmallcommercialcustomers.

Inthesubstationnr4and10there is installedEchelonCTmeterthatmeasuretheenergydata intheLVnetwork. Thesemeterswill registermeasurement data every 10-20 seconds and sends the data to theFlexmeterMDMdatabasecommunicatingviaGPRS.

The quantities that are provided by the Echelon CTmeter includemeter energy (summation of phases,totals,orperindividualtariff),instantaneousRMSvaluessuchasvoltageandcurrent,andpulseinputs.Allvalues are updated within the meter once per second for usage such as viewing, data logging, orcommunications.

Themeasuredelectricalvaluesare:

• Activepower(kW),summation:forward,reverse;• Active energy (kWh), summation: forward, reverse, forward + reverse, forward – reverse Note:

“forward – reverse” is limited to zero at the bottom of the range, negative values are notaccumulated;

• Reactivepower(kvar),summation:import,export;• Reactiveenergy(kvarh),summation:import,export;• RMSvoltage,perphase;• RMScurrent,perphase;• Powerfactor,perphase;• Frequency;• Sineofphaseangle,perphase;• Pulsecountsfrompulseoutputdevices;• Apparentpower(VA),summation.

Due to the limitation in the communication module that sends the data to the E.ON Flexmeter MDMsystemtheresolutionwillbevaluesevery10-20secondsintheMDMdatabase.


3.2. ObjectivesoftheintegrationofthesmartmeterintheMV/LVsubstation

Theobjectivesofinstallingasmartmeterinthesubstationarethefollowing:

• MeasurethetotalactiveenergyintheLVnetworkandcompareittothesumofalltheenergyfortheendusers.Theresultofthecomparisonistoanalyzeenergylossesandfindenergythefts.

• Measure and monitor the phase and total currents in the LV network to be able to act (senddemandresponsecommands)ifthecurrentreachesthehigherlimitationofthetransformerinthesubstation.

• Measureandmonitor the frequency in theLVnetwork tobeable toact (senddemandresponsecommands)topreservetheproperfrequency.

• MeasureandmonitorthevoltageintheLVnetworkandcompareittotheenduserstobeabletodetermineifapowerfailureislocatedattheendcustomerorintheLVnetworkcable.

4. Meterdataaggregation4.1. Introduction-dataaggregationasinformationcompression

The presently emerging power systems –intrinsically included in the smart grids paradigm –reflect thesynthesis of an unique technological evolution: on one side, the increased complexity of the activedistributiongridsaredemandingtheuseofmoreelaboratemodelsofthegridinfrastructure:intermittentgeneration, prosumer - type loads with patterns highly variable in time (constant impedance, constantactive power, profiled reactive power), andmobile storage units require new control algorithms, whichhavetorelyondifferentmodelsfortheenergytransferitself.Arelevantexampleisgivenbyhybrid(ac/dc)activedistributiongrids.Ontheotherside,theevolutionoftheICT-enabledsolutionshasadirect impacton both the customers and signal-processing techniques embedded in commercially available dataacquisition systems. Active distribution grids exhibit several differences from active distribution grids:firstly, low inertia due to the power electronics interfaced generation units and rectifier based loadstranslates into one order of magnitude lower constant times associated to power transfer in bothsteadystateanddynamicoperationandsecondly:ahighercontrolflexibilityandpotentialacceptabilityofdifferentlevelsofqualityofsupplyaccordingtocustomizedenergycontracts.

Stateestimationisoneofthemainapplicationsthatcanbeusedindistributionnetworks,forprovidingthebestestimationofthevoltagemagnitudeandphaseforallnetworknodesbasedonthenetworktopologyinformation, the network parameters, and available measurement data. Further, a specific task ofdistributionnetworks is thepower flowmanagementprovidingdirectbenefitondeferralof investmentsduetonetworkcongestions.Theadventofsmartmeteringandassociatedserviceslikedemandresponseand demand side management will contribute to the development of a quasi-real time responsivenessschemeformonitoringandcontrolofthedistributionnetworks.Itisexpectedthatfeatureslikeautomaticalarmingoflowvoltagenetworkoutageswillimprovelowvoltagenetworkreliability.

Activedistributiongridsoperationwillrequirenewformulationofoptimalenergytransferandthereforeitisneeded toanalyze in greaterdetail the LVnetwork. Further, a specific taskofmicrogrid control is thepower flowmanagement. Smartmetersandassociated services likedemand response (DR)anddemand


sidemanagement (DSM) contribute to the development of a quasi-real time responsiveness scheme formonitoringandcontrol.Themeasurementreportingrateassociatedwiththesecontrolschemesisusually10 minutes. Monitoring the steady-state power transfer in LV and MV networks is mainly done in thepower quality (PQ) framework. Monitoring (and control) is to be equally determined by the quality ofmeasurementsystemandbytheprocessingtechniquesofthemeasurementresults(actingasinformationretrieval).Standards [3]aresilentlyassumingsteadystateoperationwithnearlyconstant frequency.Theupperlimitationsfordeviationsoffrequencyandvoltage(rmsvalues)areappliedtovaluesobtainedfromasuccessive aggregation of measurement data available from long-term observation. A compromisebetween the end-user needs for information and the amountof data resulting fromaveragingover oneperiod of the observed quantities waveforms (i.e. rms values) has been up to now done by dataaggregationoverthetimeaxis.

Inthefollowing,weuseacombinationoftheBusinessIntelligencedefinitonofaggregation:

Dataaggregation isatypeofdataandinformationminingprocesswheredata issearched,gatheredandpresentedinareport-based,summarizedformattoachievespecificbusinessobjectivesorprocessesand/orconducthumananalysis.

WiththeclassicaloneusedbyStatistics:

Aggregateddataaredatacombinedfromseveralmeasurements;groupsofobservationsarereplacedwithsummarystatisticsbasedonthoseobservations.

InPowerSystems,aggregationalgorithmshavebeenpromotedfortheconsiderationof:

i. informationconcentrator,datacompression;ii. noisefilteringiii. dc-acenergytransfermodelsequivalence.

Further,thedataconcentratorparadigm–especiallyforWAMCSapplicationsconsidersdataaggregationasthe activity of forming a set of data from multiple inputs. The later approach is corresponding to anaggregationinspace,afunctionrelevantformoderndeploymentofsmartmeters.

A first example of formalizing the information concentrators in power systems – in the sense of datacompression -isgivenby theuseof rmsvalues fordescribingquantitieswith [quasi-] sinusoidalvariation(currents,voltages,activeandreactivepowers):

( )∫+

=Tt

tdttx

TX 0

0

21 (1)

wherex(t) istheperiodical(periodT)quantitymappedintoasinglepointX,itsrmsvalue.Fordiscretizedsignals,thermsalgorithmscanbeimplementedineithersynchronouswayorusingamovingwindowwithpre-definedwidth (usually one period of the original signal). The algorithm of computing rms values onnon-overlapping windows with fixed length (N samples) is a powerful compression method. In [4] it isshown that such algorithms are equivalent to low-pass filtering; if the required response time of themonitoring process is less than tens ofmilliseconds,more information on the actual implementation isneeded.Thistypeofdatacompressioncanbeseenasamoving-rmsprocessappliedtothediscretesignalg[n]wherethecomponentsarethermsvaluesobtainedfromsampledvaluesx[n]followedbydecimationwithfactorNk,whereNk=10,15,299,12,…accordingtotheaggregationlevelproposedin[3].


Figure4.1showsthesignalsinvolvedinthetimeaggregationprocess[5].

Figure4.1-Timeaggregationprocessrepresentedasrmsfilteringfollowedbydecimation[6].

Forexample,thefirstlevelofaggregationisperformedonatimeperiodcorrespondingto10cyclesoftheinitialwaveform (N1=10) resulting d[n], the output of the decimation applied to g[n] over exactlyN=N1samplesandforwhichthez-transformis:

( ) [ ] [ ] [ ] ( )∑ ∑∑= =

−−

=

− ====0 ,...2,,0

/

0 k NNk

NNkk

k

k zHzkgzkNgzkdzD (2)

In(2),H(z)isthez-transformofanadditionalsignalh[n],definedsuchas:

[ ] [ ]⎩⎨⎧ =

=otherwise

Nnwhenngnh

,00mod,

Itcanbeshown[5]that

( ) ∑−

=

⎟⎠

⎞⎜⎝

⎛ +=1

0

21 N

k Nk

NG

ND πωω (3)

Forthemoving-rmsfiltering[4]appliedtothediscretesignal

2rmsrms XG =

where

1;2 ==Ω ss

fff

π

onecanwrite[6]:

( )

( ) [ )∑−

= ⎟⎠

⎞⎜⎝

⎛+

Ω

=

Ω

∈Ω

⎟⎠

⎞⎜⎝

⎛ −−

=Ω

⎯⎯ →⎯

⎟⎟⎠

⎞⎜⎜⎝

⎛−−

=Ω

1

022

1

,0,111

11

111

11

N

k Nk

Nj

NN

j

eN

ND

eN

NG

ππ

(4)


4.2. Aggregationintimedomain

Accordingtothestandard[3], theaggregationmethodrelatedtovoltagemeasurement istherootmeansquare(rms)algorithmperformedonquantitiesdescribingenergytransferunderstationaryconditions.Theaggregationmethoduses4differenttimeintervals:200ms(10cyclesfor50Hzsystemfrequency–asbasicintervalforthemeasurementprocess),and3s,10minutes,2hoursasaggregationintervals.Aggregationsareperformed,withoneexceptionforthefrequency,usingthesquarerootofthearithmeticmeanofthesquaredinputvalues[3].Figure1showsthetheoreticalstructureoftheaggregationprocess.

Typically three categories of aggregation are encountered: package aggregation, cycle aggregation, andtime-clockaggregation.Thepackageaggregationisthe10-cycletimeintervalaggregation.

According to thestandard, thetimetag is thetimeat theendof the10-minaggregation [3].The2-hourintervalsmeasureddatashouldbecomputed fromtwelve10-minute time intervals. Inconsequence, theformulaforthefirstaggregationlevel(0.2stimeinterval)is[7]:

(5)

wherevkisthequantityassociatedwithonefundamentalperiodofthequasi-sinusoidalwaveforms;usuallythisquantity isthermsvalueofvoltagesorcurrents,whichalready istheresultofansecond-orderfilter[4]. Inorder toassessa stationaryprocess,whichhas the sameprobabilitydistribution forall timesandpositions,parameterslikemeanandvarianceshouldnotchangeovertime.Itisexpectedthatsomeofthequantities describing the power system operation and observed during 0.2 s time intervals meet thestationarity condition. With this assumption we can make the hypothesis that the 10 values could betreatedasmeasurementsoverthesamemeasurand[3].Similarly,dataaggregatedinthenextstepshouldinclude theassociateduncertainty form theprevious step. The formulaeused for the further three timeaggregationintervals(3s,10minutesand2hours)are[7]:

; ; (6)

Althoughstationarityisalwaysassumed,recentstudiesshowthatthisconditionisnotalwaysmetforthe3stime interval.Moreover,stationarityfora10-minuteperiod inpowersystems isalreadyaquestionablehypothesis[8].BecausefrequencymeasurementdidnotneedanaggregationmethodasthatdefinedforclassA instruments [3], theproposeddata concentrator is practically a simple arithmetic averagingoveradjacentapproximately500waveformtimeperiods.

Figure4.2 shows [9] the twoaggregationalgorithmsproposedby thepowerquality series standards IEC61000-4-30[1].

( ) ∑=

=10

1

2

2.0 10k

ks

vv

( )( )∑

=

=15

1

22.0

3 15m

mss

vv ( )

( )∑=

=200

1

23

min10 200p

psvv ( )( )∑

=

=12

1

2min10

2 12p

ph

vv


a)

b)

Figure4.2-Timeaggregationprocessaccordingto[3]:a)timeintervalsfortherms-algorithmappliedtovoltage-,current–,distortion-,andvoltageunbalancedata;b)timeintervalsforthearithmeticaveragingalgorithmappliedtofrequencydata.

Presentlythereisagapbetween(i)thelevelofapproximationusedformodelingthecurrentandvoltagewaveformswhich is implicitlyassumedbymostof themeasurementdevicesdeployed inpower systemsand(ii)thecapabilitiesandfunctionalitiesexhibitedbythehighfidelity,highaccuracyandhighnumberofpotentialreportingrates[10]ofthenewlydeployedsynchronizedmeasurementunits(SMU).Thisgapcandetermineasignificantdepreciationoftheinformationmediatedbythecontrolsystemswhicharerelyingonreal-timemeasurementsdeliveredbyequipmentwithheterogeneousaggregationandreportingrates[7].Moreover,real-timedefinitionisevolvingfollowingchangingpatternsofenergytransfer:forexample,theincreaseduseofpowerelectronicsconvertersastheinterfaceforbothgenerationandloadsresultsinlowersysteminertiaandconsequentlyrequiresafastercontrol.

Severalapplications,amongwhichofincreasedimportanceistheclassofstateestimators,needbothtypesofmeasurements: theso-called“classical”one,adapted forade facto steady-stateparadigmof relevantquantities and the “modern” one, i.e. with fewer embedded assumptions on the variability of samequantities. Measurements in power systems are physical measurements, which embed the twoindependent features (objectivity and intersubjectivity) directly in the structure of the measuringinstrument[11,12].


The main constraint so far is put by the power quality (PQ) standards [3] where several aggregationalgorithmsarerecommended,withspecificfocusontheinformationcompression.Thefurtherprocessingofrmsvalues(alreadytheoutputofafilteringalgorithm)resultsinsignificantsignaldistortion[13].

4.3. MeasurementQualityandReportingRate

Theneedofmergingmeasurementdatamadeavailablewithdifferentreportingratesand/orembeddingstandard aggregation algorithms requires an evaluation of the quality of measurements. Therefore it isneeded to considerdefinitionaluncertainty,which is stronglydependentnotonlyonmeasurementdatapost-processingbutalsoonthephysicalmodeloftheenergytransferintheconsideredgridsection.

LetusconsideraclassicalmeasurementequipmentE1whichfulfillstheIEC61000-4-30recommendations[3]formeasuringtheparametersofasignalx(t).LetbeTNtheaveragingwindow,whichformostofthePQmeasurementequipmentisidenticaltothereportingperiod.Accordingto[3]:

quantitiesother for 2h}, min; 10 3s; ms; {200 Tfrequencyfor 2h}, min; 10 10s; ms; {200 T

N

N

all∈

∈

(7)

The algorithm embedded in E1 can be [3] either the simple average of the information provided byN=[TN/T0]samples(forfrequencydata,whereT0istheinverseofthereported(average)frequency),orthermsaverageforotherquantities.Moreover, for the in-usereportingrate,E1 iscertifiedwithanaccuracyclassc1(classA,BorCinstruments)or, ingeneral,hasassociatedthemeasurementstandarduncertaintyu1. For example, considering a uniform distribution ofmeasurement errors, for themeasurement rangeXmax,andformostanalogicinstruments,thecorrespondingstandarduncertaintyis:

3100Xc max1

1⋅

=u (8)

Now let’s consider a synchronized measurement device E2 (for example PMU) for which the maximalrelative error is ePMU; usually ePMU is derived from the maximal uncertainty associated with the localmeasurementchannels,whichforfrequencyislessthan1ppm(0.0001%)ofreading,plustimebaseerror;whileforthetimestampis1μsplustimebaseerror;timebaseerrorislessthan1μs,whenlockedtoatleastonesatellitewithcorrectposition.ThisdeviceisdeployedwithaRNreportingrate.Accordingto[14]:RN∈{50,10,1,…}frames/s.

Thequestion is how to assess (and compare) theirmeasurementquality equivalent accuracywhen suchtwo devices E2 and E1 are simultaneously used in a common application. To this, we distinguish twoscenarios[15]:

First scenario:Theapplication requires the information tobeprocessed in thepowerquality framework.This translates into an approximation of the energy transfer process through either simple or quadraticaveragingalgorithmappliedtothemonitoredquantity(voltage,frequencyetc.).PMUdataisavailablewithhighestreportingratewhichhastobefurtheraggregatedfollowingthesameaveragingalgorithm(overthewindowTN).Thereference information inthiscase is representedbytheE1data.EquipmentE2willhaveassociatedastandarduncertaintyu2Case1,derived fromtheE2standarduncertaintyu2 (madeavailable forthemaximalreportingrate)andthestandarduncertaintyassociatedwiththesignalprocessingunitwhich


computes the requiredaverageoverTN. In this case, theerrorderived frommodel approximation (PMUmeasures the “true” data series of measured quantities, while E1 reports an approximation) is to beaccountedbyE1onlyasadefinitionaluncertaintyanditscontributiontothemeasurementuncertainty isalready taken intoaccountby theaccuracyclassofE1.Therefore the twosetsofmeasurementsbecomecompatiblebyreferringtothesameapproximationwhereE1hasstandarduncertaintyu1andE2has[16]:

Nuu Case221

2 = (9)

Taking this into consideration, the relative standard uncertainty can be computed for all quantities andreportingrates.

Secondscenario: in this caseE2 isusedat lower reporting rates,with thegoal toemulatea synchronousinformation merging from E1- type of equipment; i.e. the “real” information is made available after adecimationprocess(forexample,1frame/sselectionforanequipmentE2abletodeliverinformationwithamaximalreportingrateof50frames/swouldresultinadecimationfactorofinformation=50).

Data fusionof the informationprovidedbyE1 (reportingperiodTN followinganapproximationalgorithmapplied to N instants of the ”theoretical” data stream) and E2 (actual information provided at instantsseparatedbyatimewindowTN2=TN,i.einformationdecimation)hastoconsiderthestandarduncertaintyofE2asprovidedbythemanufacturer(forexample,TVE=1%)andforE1,acompounduncertaintyderivedfrombothdefinitionaluncertaintyandtheequipmentE1standarduncertainty:

u1Case2 = uM

2 +u12 (10)

Here the model (definitional) uncertainty uM is a measure of “distance” between the actual descriptorimplicitly used for the variation in time of a quantity and its single values resulted from applying theembeddedaggregationalgorithm.Thisdistanceisdependentnotonlyfromthetimeaggregationformulaandlengthofthetimeintervalbutalsofromthephenomenadescribedbythequantityinuse.Thislattercomponent cannot be derived theoretically, as it is dependent on the actual power network operationconditions.Anextensivemonitoringcampaignneeds tobepursued inorder toquantify this component.The following sectionpresents specific caseswhere the information is fused fromPMUs installed in fourdifferent nodes of the Romanian power system and set on the highest available reporting rate (50frames/s).

4.4. RandomenssPower–ameasureofheterogenousreportingrates[15]

Let’sconsidera linear(resistive)circuit.ForthiscircuitU, Iareevaluated(rms)overeachT1 interval(rmsaggregation)andcorrespondingPkmeasured(firstlevelofaggregationaccordingtoIEC61000-4-30).DoingasimplesummationweobtainttheaveragepowercorrespondingtoTAobservationtime.Forthepurposeof this excerciseT1=200msandTA=3s theobservation interval; TA=15T1. The challenge is to compare thisresultwith adirectmeasurementover TA.Asit canbe seen in the tablebellow theobtained results aredifferent.

ForTA,theequivalentvoltage(aggregatedonfirstlevel),thecurrentandtheapparentpowerare:


( ) ( )

VAIUS

AI

IVU

U

AAA

kk

Ak

k

A

5.2411

4849.1015

;9984.22915

15

1

215

1

2

=⋅=

⇒

====∑∑==

AAAR

Ak

k

A

SVAPSD

SWP

P

⋅≈=−=

⇒

<==∑=

07.0 8528.166

7.240515

22

15

1

k 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Uk[V] 222 225 236 221 239 235 230 232 225 229 239 231 230 225 230

Ik[A] 11 11 10 10 10 9 11 10 10 11 111 11 11 11 10

Pk[W] 2442 2475 2360 2210 2390 2115 2530 2320 2250 2519 2629 2541 2530 2475 2300

4.5. Aggregationofwideareameasurements

Aggregation inspacedenotesanalgorithmtoencode informationprovidedbysensors located invariousnodes of a network for which accurate models are available. Applications based on measurementinformationretrievedfromseveralnodesofthesystemtobecontrolledneedasynchronizationprotocol.The required clock precision is dependent on the application, the standards maturity and availabletechnology.

Inseveralcontrolapplications(forexample,activedistributiongridsoperation),whicharebasedonpowerquality assessment, both time and space data aggregation is needed. A good model of such dualaggregation is given by synchronized measurements. Such a platform should allow the connection ofmeasurement equipment from different vendors, full control over the measurement chain and itsassociateduncertainties,areliablecommunicationinfrastructure,andadatahub(cloud)astheunderlyinglayer for concentrating thedata and the environment for extracting,managing andprocessing thedata.Suchanenvironmentcanbeseenasacomplexcodingapplication(inspaceandtime)forwhichaminimalstandardizationeffort is required.Datahugrelatedstandardization (space-relateddataaggregation) isatthe beginning [14], while for time aggregation only the PQ algorithms are standardized.Moreover, themeasurement equipement-data hub communication layer should be governed by standardizedcommunication. Integration on the same platform of PQ-information simultaneously delivered bymeasurementdeviceswithdifferentreportingratesisstillunderdevelopment[18].Difficultiesrelyinthedifferenttypeofinformationcompressionincaseoftimeaggregationofphasor–andPQdata,whichisnotalwayspossibletorelatetothesteady-stategridoperation.


4.6. AggregationparadigmandSmartMeteringtechnology[19]

Smartmeters (SM)deployment,asbasis forSmartGridsupportaswellas for flexibleenergyandenergyservicesmarkets arebecoming themostpresent installedequipment in LVdistributionnetworks, in linewith the aim of European Union for 80% SM in 2020, as a politically endorsed target which shows thesocietyengagementtowardsmoreefficiencyandsustainability.

DistributionnetworksandespeciallytheLVnetworksarealsolivingadramaticchangeintheeraofSmartGrids deployment, through high penetration of distributed generation, in an environmentwhich ask forboth high network efficiency and flexible localmarkets, including neighborhoodsmarkets of energy andenergyservicessuchasdemandresponse.

Thenewmeterchipsaremoreandmoreperformantanduseinthelatestsolutionsarchitecturescomplexarchitecturessuchasdual-coreARMCortexprocessorswithDSPfunctionalities,allowing intensivedigitalcomputations for measuring the energy but also additional quantities based on voltage and currentcomputations,suchasPowerQuality(PQ)monitoring.

ThemainPQareasofinterestfortheDSOaswellasfortheuserare:

• Voltagelevel,whichiscurrentlydescribedbyEN50160,givinglegallimitsinastatisticalapproachbasedonP95%levels[20,21];

• Voltageharmoniclevel,throughthesyntheticTHDandthroughindividualharmonicsupto50,withthesamestatisticalapproachofP95%[3,21]andwithnormsforthestatisticallevel

• Currentharmonic level, analyzedwith the samealgorithms [3,21],butwithouta legally requiredlimit

• Non-symmetries of voltage and current, calculated for the fundamental or for the RMS value,withinastatisticalapproach;

• Flickereffect,meaninglowfrequencydeviationsofvoltage,withthesamestatisticalmethod[22],• Voltagedips,asshortvoltagedepreciations,tobeclassifiede.g.accordingtoUnipederules;• Energycontinuityinsupply,throughnumberandtime-lengthsofoutages,tobeusedforcalculating

SAIDIandSAIFI.

ParticularlyPQaspectsa),b)ande)havestandardswithhighrequirementsintermsofcomputation,whichaskforhighcostspecializedequipment.

Existing power quality (PQ) standards [3,21] use several aggregation algorithms in order to retrievestatisticalresultsP95%ofdifferentharmonics.

According to current standards andPQpractice, statistics of harmonics is basedon a complex statisticalmethodology,withthefollowingsteps:

• harmoniclevelforeachsetof10periodsiscontinuouslycalculatedthroughDFTcomputationsandaremediatedeach3seconds(15setsconsideredforexact50Hz,but14upto16setsforfrequencydeviations under or over the 50 Hz) or each 10 minutes (15x600 sets); moreover, overlappingwindows can bring the necessity to make 2 different DFT/FFT analysis in the same time, to becoveredby the computational powerof thePQanalyzer; thismay request, for classA, a doublecomputation power in some cases and requires corresponding microprocessor based platform(usuallyincludingaDSPpart)inthemeasurementequipment;


• anadditional2hoursintervalisalsomentioneddinthestandard,byaggregating12timeintervalsof10minutes;

• Finally, a week of such 10minutes results,meaning 1440/10 x 7 = 1008 values are analyzed interms of distribution of probability, in order to get a 95% level of probability, to be used asinformativelimitsfortheharmonicdistortionlevel.

This procedure has been recently questioned and more efficient solutions are to be addressed in theprojectNobelGrid[23]:

ThestatisticalLightPQAssessmentMethodisnotchallengingthetodaystandards,butisintendedtobringamethodforPQassessmentwhichhasacceptableerrorscomparingwithtodaystandardsbutallowscost-effective implementation in each piece of a new generation of Smart Meters to be connected in LVnetworks. In the Smart Grid oriented PQ framework, this method is desired to become a standard tocomplementtodaystandards, applicable inSmartMetersdesignsandrecognizedaspracticalmethodofPQassessmentatlowvoltagelevel,dealingwithsmallconsumers,prosumersandactiveLVnetworks.

5. UsecasesrelatedtotheintegrationofsmartmetersintheMV/LVsubstations

Distribution networks, because of their geographical dispersion in urban and rural areas, can besignificantly affected by faults arising from different sources such as adverse weather conditions, andequipmentfailure.Inordertominimizeoutagetimesandimprovethecontinuityofsupply,developmentofpracticalmethodfordetectionand locationofoutages ismandatory.Powercompaniesnormallydependon customer trouble calls for outagemapping. However, inmodern distribution networks, smartmeterdatawouldbeanavailableoptiontodeterminetheimpactedpartofthenetwork.

5.1. TheusecasesforoutagedetectionandoutagelocationinLVnetworks

At themoment usuallyDSOs use customer calls to detect the LV network outages, this process has thefollowingdisadvantages:

1. Duringthenightstherewouldbealittlenumberofcustomercallswhichmakestheoutagelocationprocessdifficultorimpossible.

2. Falseorfakereportsarehardtorealize.3. Thewholeprocessistimeconsumingandinefficient.

OntheLVgrid,severalmonitoringdeviceswillbeinstalled,intheFlexmeterproject,whichwouldpreparethefollowingdata:

• RMSvaluesofVoltage,Current,ReactiveandActivePower;• LoadprofilesfromLVcustomersandproductionpoints.;• Statusofremote-controllableelementsinsecondarysubstations(RTU1-IED2);• Alarmsrelatedtoeventsandfaults(IED-PQM3-RTU).

1 Remote terminal unit

2 Intelligent electronic device


Anoutagedetectionandlocationplatform(Figure5.1)canusetheseLVmeasurementanddatacollectedfrom different sources to detect and locate LV outages in a fast and efficient manner with no or littlehumaninterventions.

Figure5.1–outagedetectionandlocationplatform

Incaseofoutage,all themeters intheoutageareadetecttheabsenceofvoltageandsenda“lastgasp”signaltothecloudsystem.Noparticularstringentrequirementsareneededfromthepointofviewofthecommunicationnetwork.Themainissueisthatthecommunicationnetworkshouldworkalso inapoweroutagecondition,thankstoaUPSsystem.

5.2. TheusecasesforfaultlocationinMVnetworks

After thedetectionofa fault in theMVdistributionnetworkand reactionof theprotection system, it ispossibletodesignanalgorithm,toidentifythefaultlocationontheMVlines(Figure5.2).

3 Power quality meter


Figure5.2–faultdetectionandlocationplatform

TheHV/MVsubstationsarealreadyequippedwithvoltageandcurrentmeasurementsforthepurposeofprotection. MV network fault location is an additional feature which could be implemented in HV/MVsubstationsor inthemaincontrolcenter. Inthefirstcase just thefault location is reported,while inthesecondcasethevoltageandcurrentwaveformsor their (calculatedduring-fault)phasorsshouldbesentfrom the primary HV/MV substation to the cloud. The requirements from the communication networkpointofviewarethesameofthepreviouspoint.Alsointhiscasenoparticularstringentrequirementsareneededfromthepointofviewofthecommunicationnetwork.Theamountofdatatobesentisquitesmallandtherearenostringent latencyrequirements.Even if thedataarrivetothecloudsomesecondsafterthe event, itwould bemuch faster thanwith conventional fault locationmethods. Also in this case themainissueisthatthenetworkshouldworkalsoinapoweroutagecondition,thankstoaUPSsystem.

4.3 TheusecasesforNetworkmanagementforloadbalancing

Theusecaseabout“Networkmanagement”aimstobalancenetworkloadprofilebyusingenergystorages.Thegoalistowriteanalgorithmthatmanagesthechargeanddischargeoperationsofabatterysituatedinthemediumtolowvoltagecabins.Inthesequel,wewillcallthisalgorithmstoragemanager.Thistoolwillworkbyusingreal-timeconsumptionandproductionpowermeasurementsfromthedatabaseinthecloud,andbyusingthe loadprofilescenariosgeneratefromthe ING.Theoutputofthestoragemanager is, foreachtimestep,thecontroloftheenergystorage.Byusingthistool,theenergynetworkwillgainstabilityandfurtherrenewableenergywillbemorecompetitive.


TheusecasediagramisshowninFigure5.3.

Figure5.3–Usecasediagram

Intheusecasediagram,wehavecalledStoragemanagementinsteadofStoragemanagerbecausewestilldonotknowifwearegoingtouseoneormorealgorithms.TheinformationflowtotheStorageManagerconsistsof:

• Realtimeconsumptionofthebuildingsconnectedtothemediumtolowvoltagecabinwherethebatteryislocated.

• Real time production of energy of the buildings connected to themedium to low voltage cabinwherethebatteryislocated.

• Areasonableamountofloadprofilescenarios.

TheinformationflowfromtheStorageManagerconsistsof:

• Foreachtimesteptheaction(charge-discharge-donothing)thatthebatterymustdo.

6. References

[1] Advanced load modelling for power system studies, PhD Thesis, Collin, Adam John, 2013-11-28,http://hdl.handle.net/1842/8890.

[2] Design a Neural Network for Features Selection in Non-intrusive Monitoring of Industrial electricalLoads.Hong-TzerYang,Hsueh-HsienChang,Ching-LungLin.IEEE.

[3] IEC 61000-4-30 ed3.0, Electromagnetic compatibility (EMC) - Part 4-30: Testing and measurementtechniques-Powerqualitymeasurementmethods,2015

[4]AlbuM.,Heydt,G.T., "On theuseofRMSvalues inpowerquality assessment " IEEETransactionsonPowerDelivery,2003

[5]MihaelaAlbu,“OnPowerQualityAssessment:accuratevs.meaningful”,GeraldT.Heydt,Festschrift,13October2013,ArizonaStateUniversity,Tempe,AZ


[6]MihaelaAlbu, 2014, “DualDataAggregation for PowerQualityAssessment”, Proc. of the I2MTC '14.IEEE InstrumentationandMeasurementTechnologyConference,Montevideo,Uruguay,13-15May2014,pag.620-624

[7]MihaiCalin,AnaMariaDumitrescu,MarkosAsprou,EliasKyriakides,MihaelaAlbu,“MeasurementDataAggregationforActiveDistributionNetworks”,Proc.of the IEEEWorkshoponAppliedMeasurements forPowerSystemsAMPS2013,Aachen,25-27Sept.2013

[8] Albu, M., Postolache, P., Golovanov, C. "Non-stationary signals' analysis in power systems. A time-frequencyapproach"10thMediterraneanElectrotechnicalConference,2000.MELECON2000

[9]AnaRuxandraToma,Ana-MariaDumitrescu,MihaelaAlbu,2015,„AssessmentofrapidvoltagechangesusingPMUdata”;Proc.2015IEEEInternationalWorkshoponAppliedMeasurementsforPowerSystems”;Aachen,Germania,23-25Sept.2015,pg.126-131

[10] Paolo Castello, Marco Lixia, Carlo Muscas, Paolo Attilio Pegoraro, “Impact of the Model on theAccuracyofSynchrophasorMeasurement”,IEEETransactionsOnInstrumentationAndMeasurement,Vol.61,No.8,August2012,pp.2179-2188

[11] Luca Mari, Paolo Carbone, Dario Petri, “Measurement Fundamentals: A Pragmatic View”, IEEETransactionsOnInstrumentationAndMeasurement,Vol.61,No.8,August2012,pp.2107-2115

[12] Alessandro Giordani, Luca Mari, “Measurement, Models, and Uncertainty”, IEEE Transactions OnInstrumentationAndMeasurement,Vol.61,No.8,August2012,pp.2144-2152

[13]MihaelaAlbu,AnaMariaDumitrescuandRoxanaPopovici, “RateofChangeofFrequency -aPowerQuality Descriptor”, Proc. of 16th International Conference on Harmonics and Quality of Power (ICHQP2014),Bucharest,Romania,May2014,pp312–316

[14]C37.118.2-2011-IEEEStandardforSynchrophasorDataTransferforPowerSystems

[15] Mihaela Albu, Ana-Maria Dumitrescu, „Applicability of synchronised measurements in moderndistribution grids. A discussion”, Proc. of IEEE Applied Measurements for Power Systems AMPS2014,Aachen,24-26,Sept.2014,ISBN:978-1-4799-6823-7

[16] JCGM 100:2008 “Evaluation of measurement data-guide to the expression of uncertainty inmeasurement,GUM2008”.

[17]Emanuel,A.E.,An introductiontotheconceptofRandomnessPower,Proceedingsofthe 201014thInternationalConferenceonHarmonicsandQualityofPower(ICHQP),2010

[18]1. AnaMaria Dumitrescu,Mihai Calin,MihaelaM. Albu, 2013, „Measurement Data Aggregation inPowerSystems–ChallengesandOpportunities”,Proc.ofthe2ndConferenceonthePowerOptionsfortheEasternMediterraneanregion,POEM,Nicosia.,6-8October2013

[19] Mihai Sanduleac, Mihaela Albu, JoaoMartins, Mª Dolores Alacreu, Carmen Stanescu, 2015, PowerQuality Assessment in LV networks using new SmartMeters design, Proc. of the CPE 2015, Lisbon, June2015

[20]EN50160,Voltagecharacteristicsofelectricitysuppliedbypublicdistributionsystems


[21]IEC61000-4-7,Electromagneticcompatibility(EMC)-Part4-7:Testingandmeasurementtechniques–General guide on harmonics and interharmonics measurements and instrumentation, for power supplysystemsandequipmentconnectedthereto

[22]IEC61000-4-7,Electromagneticcompatibility(EMC)–Part4-15Testingandmeasurementtechniques,Flickermeter–Functionalanddesignspecifications

[22]H2020NobelGrid–Smartenergyforpeopleproject,www.nobelgrid.eu

substation smart meter integration - flexmeter...

Documents