• Tidak ada hasil yang ditemukan

Electric Power Systems Research

N/A
N/A
Protected

Academic year: 2023

Membagikan "Electric Power Systems Research"

Copied!
10
0
0

Teks penuh

(1)

ContentslistsavailableatScienceDirect

Electric Power Systems Research

jou rn al h om e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / e p s r

Modeling power networks using dynamic phasors in the dq0 reference frame

Yoash Levron

a,∗

, Juri Belikov

b

aTheAndrewandErnaViterbiFacultyofElectricalEngineering,Technion—IsraelInstituteofTechnology,Haifa3200003,Israel

bFacultyofMechanicalEngineering,Technion—IsraelInstituteofTechnology,Haifa3200003,Israel

a r t i c l e i n f o

Articlehistory:

Received26July2016 Receivedinrevisedform 20November2016 Accepted28November2016

Keywords:

Quasistatic Averagesignals Dynamicphasors Inter-areaoscillations Multi-machine Stability

a b s t r a c t

Thedynamicbehavioroflargepowersystemshasbeentraditionallystudiedbymeansoftime-varying phasors,undertheassumptionthatthesystemisquasi-static.However,withincreasingintegrationoffast renewableanddistributedsourcesintopowergrids,thisassumptionisbecomingincreasinglyinaccurate.

Inthispaper,wepresentadynamicmodelofgeneraltransmissionanddistributionnetworksthatuses dynamicphasorsinthedq0referenceframe.Themodelisformulatedinthefrequencydomain,andis basedonthenetworkfrequencydependentadmittancematrix.Wealsopresentasimplifiedversionof thismodelthatisobtainedbyafirst-orderTaylorapproximationofthedynamicequations.Theproposed modelsextendthequasi-staticmodeltohigherfrequencies,whileemployingdq0signalsthatarestatic atsteady-state,andthereforecombinetheadvantagesofhighbandwidthandawell-definedoperating point.Themodelsareverifiednumericallyusingthe9-,30-,and118-bustest-casenetworks.Simulations showthatfrequencyresponsesofallmodelscoincideatlowfrequenciesanddivergeathighfrequencies.

Inaddition,responsesofthedq0modelinthetimedomainandintheabcreferenceframeareveryclose tothoseofthetransientmodel.

©2016ElsevierB.V.Allrightsreserved.

1. Introduction

Dynamicprocessesoccurringinlargeelectricpowersystemsaremodeledatvaryinglevelsofabstraction[1,2].Insimplestaticmodels, voltagesandcurrentsareassumedtobesinusoidal,andarerepresentedbyphasors.Thenetworkistypicallydescribedinthiscaseby thepowerflowequations.Ontheotherextreme,transientmodelsdescribethesystemusingnonlineardifferentialequations,whichare solvednumericallyinthetimedomain[3–8].Athirdtypeofmodel,basedontime-varyingphasors,iscalledtheaverage-value,dynamic orquasi-staticmodel[2].Thismodelissufficientlyaccurateaslongasphasorsareslowlychangingincomparisontothesystemfrequency [9,10].Akeyadvantageofthequasi-staticmodelisthatitenableslongnumericsteptimes,andhencecanacceleratethesimulationprocess.

Inaddition,sincethequasi-staticmodelemploysphasorsinsteadofsinusoidalACsignals,thesystemoperatingpointiswell-defined,a propertywhichconsiderablysimplifiesstabilitystudies.Duetotheseproperties,quasi-staticmodelshavebeenusedextensivelyinthe analysisofdynamicinteractionsthatoccurintimeframesofsecondstominutes,andhavehistoricallyenabledstudiesofmachinesstability, inter-areaoscillation,andslowdynamicphenomena[1–3,11].

Inrecentyears,withtheemergenceofsmalldistributedgeneratorsandfastpowerelectronicsbaseddevices,theassumptionofquasi- staticphasorsisbecomingincreasinglyinaccurate.Duetotheseemergingtechnologies,voltageandcurrentsignalscancontainhigh harmoniccomponents,andcanexhibitfastamplitudeandphasevariations[9].Thesedevelopmentshaveledtoanewclassofmodels basedondynamicphasors.Theconceptofdynamicphasorshasemergedfromaveragingtechniquesemployedinpowerelectronics,and hasbeenextendedtothree-phasesystems[12].Dynamicphasorsgeneralizetheideaofquasi-staticphasors,andrepresentvoltageand currentsignalsbyFourierseriesexpansionsinwhichtheharmoniccomponentsareevaluatedoveramovingtimewindow[13].This approachoffersthebenefitsofaphasorbasedanalysis,highaccuracy,andfastsimulations[14,15].

Duetotheseproperties,dynamicphasorshavebeenusedintheanalysisofsynchronousandinductionmachines[12,16],HVDCsys- tems[14,17,18],FACTSdevices[10],sub-synchronousresonance[19,20],asymmetricsystems[12,21],andasymmetricfaults[16,22].For

Correspondingauthor.

E-mailaddresses:[email protected](Y.Levron),[email protected](J.Belikov).

http://dx.doi.org/10.1016/j.epsr.2016.11.024 0378-7796/©2016ElsevierB.V.Allrightsreserved.

(2)

Fig.1.Aunitinthenetwork,showingsignalsintheabcreferenceframe.

instance,adynamicphasormodelofalinecommutatedHVDCconverterispresentedin[14].Themodelrepresentsthelow-frequency dynamicsoftheconverter,andhaslowercomputationalrequirementsthanaconventionaltransientmodel.Adynamicphasorrepresen- tationofanunbalancedradialdistributionsystemispresentedin[21],wherevariouscomponentsincludingaphotovoltaicsourceandan inductionmachinearemodeled,anddynamicinteractionofthesecomponentsisshown.Asimulatorbasedondynamicphasorshasbeen developedin[15,23],whichusesdifferentialswitched-algebraicstate-resetequationstodescribethesystemcomponents.Thesimulation toolcombinesadvantagesoftransientstabilityandelectro-magneticsimulationprograms,andisdemonstratedontheIEEE39bustest- casenetwork.Work[24]extendsthedynamicphasorconcepttomulti-generator,multi-frequencysystems.Thetheorypresentedin[24]

enablesstudyofelectricpowersystemswithoutassumingasinglefrequency.Theproposedapproachisvalidatedonatwin-generator system,andcanbeextendedtolargernetworks.Inaddition,dynamicphasorsareutilizedinstateestimation[25–29],insystemswith varyingfrequencies[30],andalsoinmicrogrids[31].

Theserecentworksaremainlyfocusedonmodelingspecificgeneratorsand loads,and donotprovidea completemodeloflarge transmissionnetworks.Tobridgethisgap,thispaperpresentsadynamicmodelofgeneraltransmissionanddistributionnetworksusing dynamicphasorsinthedq0referenceframe,andshowshowthismodelcanbeimplementedinpracticeusingtime-domainstateequations.

Thefirstmodelpresentedinthisworkisformulatedinthefrequencydomain,andisbasedonthenetworkfrequencydependentadmittance matrix.Sinceadirectnumericimplementationofthismodelischallengingingeneral,wepresentasimplifiedversionofitcalledthefirst- orderdq0model,whichisobtainedbyafirst-orderTaylorapproximationofthedynamicequations.Theproposedfrequency-domainmodel extendsthequasi-staticmodeltohigherfrequencies,whileemployingdq0signalsthatarestaticatsteady-state,andthereforecombines theadvantagesofhighbandwidthandawell-definedoperatingpoint.Thestandardquasi-staticmodelfollowsasaspecialcasefromthe modelatlowfrequencies.

Thepapercontinuesasfollows.Section2recallsbasicconceptsofdynamicphasorsandexplainshowtoformulatequasi-staticmodelsin termsofdq0quantities.Section3presentstheproposeddq0model.Thefirst-orderdq0modelispresentedinSection4.Section5provides simpleillustrativeexamples,followedbyvariousnumericalresultsprovidedinSection6.ConcludingremarksaredrawninSection7.

2. Dynamicphasorsinthedq0referenceframe

Forsystemsinsteady-state,thedq0transformationmapssinusoidalsignalstoconstantquantities.Thistransformationiscompatible withstandardelectricmachinemodels,andwithemergingmodelsofrenewableanddistributedgenerators[10,15,32,33].

AssumeageneralpowernetworkcontainingNthree-phaseunits,withvoltagesandcurrentsaspresentedinFig.1.

Thedq0transformation(asdefinedin[26])isgivenby

⎢ ⎣

xd(t) xq(t) x0(t)

⎥ ⎦

=23

⎢ ⎢

⎢ ⎢

⎢ ⎣

cos(ωst) cos

ωst−2

3 cos

ωst+2 3

−sin(ωst) −sin

ωst−2

3 −sin

ωst+2 3 1

2

1 2

1 2

⎥ ⎥

⎥ ⎥

⎥ ⎦

⎢ ⎣

xa(t) xb(t) xc(t)

⎥ ⎦

, (1)

whereωsisthenominalsystemfrequency,andxisreplacedwitheitheravoltage(vn)oracurrent(in)signal.Aprominentfeatureof thedq0transformationisthatforbalancedandstaticsystems,thedirectandquadraturecomponentsxd,xqareconstants,andx0=0.Ina quasi-staticsystemxa,xb,xcarenearlysinusoidal,withslowlyvaryingamplitudesandphases.Inthiscase,thesystemcanbemodeledin termsoftime-varyingphasors,whicharedefinedintermsofthedq0componentsas

Vn(t)= 1

√2

vdn(t)+jvqn(t)

, In(t)= 1

√2

idn(t)+jiqn(t)

. (2)

Underthequasi-staticapproximationthesedqcomponentsareidenticaltothexycomponents,asdescribedin[34].Thefollowinganalysis definesthestandardquasi-staticmodelandpowerflowequationsintermsofdq0signals.Quasi-staticnetworksareusuallydescribedby theadmittancematrixYbus.Definethevectors:

Vd(t)=

vd1(t),...,vdN(t)

T

, Id(t)=

id1(t),...,idN(t)

T

, Vq(t)=

vq1(t),...,vqN(t)

T

, Iq(t)=

iq1(t),...,iqN(t)

T

, V0(t)=

v01(t),...,v0N(t)

T

, I0(t)=

i01(t),...,i0N(t)

T

. (3)

Thenthecomplexvoltageandcurrentphasorsrelateby Id(t)+jIq(t)=Ybus

Vd(t)+jVq(t)

(4)

(3)

andtherealandimaginarypartsofthisequationsare Id(t)=Re

Ybus

Vd(t)−Im

Ybus

Vq(t), Iq(t)=Im

Ybus

Vd(t)+Re

Ybus

Vq(t). (5)

Theseequationsconstituteaquasi-staticmodel.Equivalently,quasi-staticnetworksarealsodescribedbythepowerflowequations[33]

Pn(t)=|Vn(t)|

N k=1

|ynk||Vk(t)|cos

∠ynkk−ın

, Qn(t)=−|Vn(t)|

N k=1

|ynk||Vk(t)|sin

∠ynkk−ın

. (6)

InwhichtheconstantsynkareelementsofthenetworkadmittancematrixYbus,andpowers,amplitudesandphasesareformulatedin termsofdq0signalsas

Pn(t)=Re

Vn(t)In(t)

= 1 2

vdn(t)idn(t)+vqn(t)iqn(t)

, Qn(t)=Im

Vn(t)In(t)

= 1 2

vqn(t)idn(t)−vdn(t)iqn(t)

,

Vn(t)

=1

2

vdn(t)

2

+

vqn(t)

2

, ın(t)=∠Vn(t)=atan2

vqn(t),vdn(t)

. (7)

Ifthesystemisnotquasi-static,thephasorsabovecanbegeneralizedbydynamicphasors.ThesecanbedefinedbyaFourierseries expansion,asin[16].However,tosimplifynotations,wehavechoseninthispapertodefinedynamicphasorswithrespecttotheFourier transformas

⎢ ⎣

Vnd(ω) Vnq(ω) Vn0(ω)

⎥ ⎦

=

−∞

⎢ ⎣

vdn(t) vqn(t) v0n(t)

⎥ ⎦

ejωtdt,

⎢ ⎣

Ind(ω) Inq(ω) In0(ω)

⎥ ⎦

=

−∞

⎢ ⎣

idn(t) iqn(t) i0n(t)

⎥ ⎦

ejωtdt. (8)

3. Largenetworkdynamicsinthedq0referenceframe

Thepowerflowequationsandtheequivalentdq0modelin(5),(6)areonlyvalidunderthequasi-staticapproximation,assumingslow variationsinamplitudesandphases[34].Thefollowingtheoremextendsthequasi-staticmodeltohigherfrequencies,anddescribesthe systemdynamicsforgeneraldq0signals.

Theorem1. Insymmetricpowernetworks,adynamicmodelbasedondq0dynamicphasorscanbedescribedas

⎢ ⎣

Id(ω) Iq(ω) I0(ω)

⎥ ⎦

=

U(ω)jR(ω) jR(ω)U(ω) 00

0 0 Ybus(ω)

⎢ ⎣

Vd(ω) Vq(ω) V0(ω)

⎥ ⎦

. (9)

Thefollowingdefinitionsareusedin(9):

U(ω):=1 2

Ybus(ω+ωs)+Ybus(ω−ωs)

, R(ω):=1 2

Ybus(ω+ωs)−Ybus(ω−ωs)

, (10)

and

Vd(ω)=

V1d(ω),...,VNd(ω)

T

,Id(ω)=

Id1(ω),...,INd(ω)

T

, Vq(ω)=

V1q(ω),...,VNq(ω)

T

,Iq(ω)=

Iq1(ω),...,INq(ω)

T

, V0(ω)=

V10(ω),...,VN0(ω)

T

,I0(ω)=

I10(ω),...,I0N(ω)

T

. (11)

Proof. Followingthedefinitionofthedq0transformationin(1),voltagesandcurrentscanbeexpressedassumsofthreeelements Va(t)=Vd(t)cos(ωst)−Vq(t)sin(ωst)+V0(t), Ia(t)=Id(t)cos(ωst)−Iq(t)sin(ωst)+I0(t). (12) Similarexpressionsapplytophasesb,cwithaproperphase-shiftof±2/3.Transformationtothefrequencydomainusingthemodulation propertyyields

Va(ω)= 1 2

Vd(ω−ωs)+Vd(ω+ωs)+jVq(ω−ωs)−jVq(ω+ωs)

+V0(ω),

Ia(ω)=1 2

Id(ω−ωs)+Id(ω+ωs)+jIq(ω−ωs)−jIq(ω+ωs)

+I0(ω). (13)

Accordingtodefinition

I1a(ω) ··· IaN(ω)

T

=Ybus(ω)

V1a(ω) ··· VNa(ω)

T

. (14)

Substituting(14)in(13)yields

Id(ω−ωs)+Id(ω+ωs)+jIq(ω−ωs)−jIq(ω+ωs)+2I0(ω)

=Ybus(ω)

Vd(ω−ωs)+Vd(ω+ωs)+jVq(ω−ωs)−jVq(ω+ωs)+2V0(ω)

. (15)

(4)

Usingrelation(10),equation(15)canberewrittenas

Id(ω−ωs)+Id(ω+ωs)+jIq(ω−ωs)−jIq(ω+ωs)+2I0(ω)=2Ybus(ω)V0(ω)+U(ω−ωs)Vd(ω−ωs)+jR(ω−ωs)Vq(ω−ωs) +U(ω+ωs)Vd(ω+ωs)+jR(ω+ωs)Vq(ω+ωs)+j(−jR(ω−ωs)Vd(ω−ωs)+U(ω−ωs)Vq(ω−ωs))−j(−jR(ω+ωs)Vd(ω+ωs)

+U(ω+ωs)Vq(ω+ωs)), (16)

whichaftercomparisonofcoefficientsresultsin(9).䊐

NotethatTheorem1holdsforsymmetricnetworks,inwhichthematrixYbus(ω)equallyappliestoeachofthethreephases.However, thegeneratorsandloadsconnectedtothenetworkarenotnecessarilylinearorsymmetric.

Thefollowingcorollaryrelatestheproposeddq0modeltothestandardquasi-staticmodel.

Corollary1. WhenYbus(ω+ωs)=YbusandYbus(ω−ωs)=(Ybus),equation(9)reducestothequasi-staticexpressionsin(5).

ThiscorollarystatesthatifthegeneralYbus(ω+ωs)canbeapproximatedbyaconstantmatrixwhenω→0,thenthedynamicmodelis quasi-static.

4. Thefirst-orderdq0model

Sincethefulldq0modelin(9)is,ingeneral,hardtoimplementnumerically,thissectionpresentsasimplifiedversionofitcalledthe first-orderdq0model,whichisobtainedbyafirst-orderTaylorapproximationofthedynamicequations.Inthiscaseitisassumedthatthe dq0signalsarebandlimited,andhavesignificantenergyonlyatlowfrequencies.Thisenablesrepresentationoftheadmittancematrixby aTaylorseriesaroundω=0as

Ybus(ω+ωs) ≈

M k=0

ωk k!

∂Ybus(ω+ωs)

∂ωk |ω=0, Ybus(ω−ωs) ≈

M k=0

ωk k!

∂Ybus(ω−ωs)

∂ωk |ω=0,=

M k=0

(jω)kjk k!

∂Ybus(ω+ωs)

∂ωk

|ω=0,

Ybus(ω) ≈

M k=0

ωk k!

∂Ybus(ω)

∂ωk |ω=0,

(17)

whereintheexpressionforYbus(ω−ωs)wehaveutilizedYbus(ω)=(Ybus(−ω))andemployednestedfunctionsderivatives.Thestandard quasi-staticapproximationisobtainedbyazero-orderTaylorserieswithM=0suchthat

Ybus(ω+ωs)≈Ybuss), Ybus(ω−ωs)≈

Ybuss)

. (18)

Iftheseexpressionsaresubstitutedintotheoriginalmodel(9),thenthemodelreducestothequasi-staticequationsin(5).Amoreaccurate dynamicmodelisobtainedbythefirst-orderTaylorseriesapproximation(M=1)as

Ybus(ω+ωs)≈Ybuss)+jωF, Ybus(ω−ωs)≈(Ybuss))+jωF, Ybus(ω)≈Ybus(0)+jωF0, F=−j∂Ybus(ω+ωs)

∂ω |ω=0, F0=−j∂Ybus(ω)

∂ω |ω=0. (19)

Substitutionoftheseexpressionsin(9)yields

⎢ ⎣

Id(ω) Iq(ω) I0(ω)

⎥ ⎦

=(M2+M1jω)

⎢ ⎣

Vd(ω) Vq(ω) V0(ω)

⎥ ⎦

(20)

with

M1=

ReIm{{FF}} ReIm{F{F}} 00

0 0 F0

, M2=

⎢ ⎣

Re

Ybuss)

−Im

Ybuss)

0 Im

Ybuss)

Re

Ybuss)

0

0 0 Ybus(0)

⎥ ⎦

. (21)

Thismodelcanbeimplementedinpracticeusingalinearstate-spacemodel dx

dt =Ax+Bu y =Cx+Du,

(22)

(5)

Fig.2.Example1:anetworkwithasingleinductor.

Fig.3. Example2:longtransmissionlinefeedingamatchedload.

whereu=(Vd(t),Vq(t),V0(t))T,y=(Id(t),Iq(t),I0(t))T.Thematrices A,B,C,Darechosensuchthatatlowfrequenciesthefrequency responseofthestate-spacemodel(22)isequalto(20).Thegeneralfrequencyresponseisgivenbyy(ω)=(C(jωI−A)−1B+D)u(ω),andis approximatedatω→0by

y(ω)≈

C

−A1

I+jωA1

B+D

u(ω)

=

−CA1B+D−jωCA2B

u(ω).

(23)

Comparisonof(23)and(20)provides

−CA2B=M1, −CA1B+D=M2. (24)

From(24),C,Dcanbecomputedforanyfull-rankmatricesA,Bas

C=−M1B1A2, D=M2−M1B1AB. (25)

ThematricesA,Bcanbearbitrarilychosen.ItistypicallybeneficialtoselectA,BsuchthatD=0,toobtainzerogainatω→∞.Thiscriterion resultsin

D=M2−M1B1AB⇒A=BM−11 M2B1. (26)

Ingeneral,ifthematrixM1isnotfull-rank,thenAcanbecomputedbysolvingastandardleast-squaresminimizationproblem A=BPB−1, P=argmin

X M1X−M222. (27)

ThematrixBcanbearbitrarilychosen,forinstanceB=I.

5. Dynamicdq0models:simpleexamples

ThesimplenetworkinFig.2demonstratesthedifferencebetweenthequasi-staticmodelandthefulldq0model.

Inthisexample,thequasi-staticmodel(5)provides

Vd(ω)=−ωsLIq(ω), Vq(ω)=ωsLId(ω). (28)

Incomparison,thedq0modelcanbecomputedby(9),andresultsin

Vd(ω)=jωLId(ω)−ωsLIq(ω), Vq(ω)=ωsLId(ω)+jωLIq(ω), V0(ω)=jωLI0(ω). (29) Notethatwhenω→0thequasi-staticandthedq0modelsareequivalent.

Anotherexample(Fig.3)isthelongtransmissionline.Inthisexamplethedq0dynamicphasorsareshowntorelatethroughtimedelays.

Assumealongpowerlineoflengthl.Thelineislossless,withinductanceLxandcapacitanceCxperunitlength.Itfeedsamatchedloadwith animpedanceequaltothecharacteristicimpedanceZc=

Lx/Cx.Theresultingloadcurrentisdelayedinrespecttothesourcevoltage [33]as

iabc2 (t)= 1

Zcvabc1 (t−d), (30)

wherethetimedelayisgivenbyd=l

LxCx.

(6)

Fig.4.Comparisonoffrequencyresponsesinthe9-busnetworkfordqquantitiesfrominput1tooutput4.

Fig.5. Comparisonoffrequencyresponsesinthe118-busnetworkfordqquantitiesfrominput76tooutput77.

Thecorrespondingdq0modelcanbecomputedby(9).DefiningYbus(ω)=(1/Zc)ejωdandapplyingtheinverseFouriertransform,the resultis

id2(t)= 1 Zc

cos(ωsd)vd1(t−d)+sin(ωsd)vq1(t−d)

, iq2(t)= 1 Zc

−sin(ωsd)vd1(t−d)+cos(ωsd)vq1(t−d)

, i02(t)= 1

Zcv01(t−d). (31) Thedirectandquadraturecurrentcomponentsaredelayedinrespecttothesourcevoltage.

6. Numericalresults

Inthissectionthevariousmodelsarecomparedintermsoffrequencyandtransientresponses.Threemodelsaretested:thefulldq0 model,thefirst-ordermodel,andthequasi-staticmodel.Thesearecomparedtothefulltransientmodel,whichisconstructedinstate- spaceusingMatlab’sSimscapePowerSystemssoftwarepackage.The9-,30-,and118-busnetworksfromtheMatPowerdatabase[35]are used.

(7)

Fig.6. Comparisonofthefirst100largestHankelsingularvaluesofthedq0modelandthefirst-ordermodelinthe118-busnetwork.

Fig.7.Comparisonofstep-responsesinthe9-bussystem.Theinputs(bottomplot)arestepsinphasesa,b,catbus1.Theoutputisthecurrentofphaseaatbus4.

AnarrayofBodeplotsrepresentingthedq0,first-order,andquasi-staticmodelsisgraphicallyillustratedinFigs.4and5for9-and 118-busnetworks.Theplotsdepictthemagnitude(indB)andphase(indegrees)forseveralinput/outputpairs.Thestepinputisapplied toageneratorlocatedatbus1andtheoutputismeasuredatbus4forthecaseof9-busnetwork.Similarly,for118-busnetworkthestep inputisappliedtoageneratorlocatedatbus76andtheoutputismeasuredatbus77.Thefiguresshowthatthefrequencyresponses coincideatlowfrequencies(ω→0)anddivergeathighfrequencies.Specifically,thequasi-staticmodelisrepresentedbyaconstantgain, andhasaconstantgainandphaseatallfrequencies.Thefirst-ordermodelresponsecoincideswiththedq0modelresponseoverawider frequencyrangeanddeviatesonlyathighfrequencies.ThesefiguresillustratethemainresultpresentedinTheorem1thatdq0model extendsbothquasi-staticandfirst-ordermodelsbytakinghigh-frequencydynamicphenomenaintoaccount.

Fig.6comparesthelargestsingularvaluesofthedq0modelandthefirst-ordermodelforthe118-busnetwork.Itcanbeseenthat themostenergetic(largest)valuesaresimilar,meaningthatthemajorityofenergyispreserved.Thisinturnverifiesthequalityofthe first-ordermodel.

Figs.7–11comparethetransientresponseofthevariousmodelsforthe9-,30-and118-busnetworks.Allinputsandoutputsarethree phasesignals.Thetransientmodelissimulateddirectlyintheabcreferenceframe.Thedq0model,quasi-staticmodel,andthefirst-order modelarerepresentedinthedq0referenceframe,soinputandoutputsignalsareconvertedaccordinglyusing(1).Tothisend,thedq0 modelisrepresentedinthestate-spaceformusingimplementationavailablein[36].Thefirst-ordermodelisimplementedusingequations (22)–(27).

Figs.7and8providesimulationresultsforthecaseof9-busnetwork.InFig.7thestepinputsignals(shiftedintimeforeachofthephases) areappliedtothegenerator1atbus1andtheoutputsaremeasuredatbus4.InFig.8amorecomplicatedscenarioisconsidered.Before thetransient,theinputsarebalancedthreephasesinusoidalsignalsua1=sin(250t)atbus1.Then,thetransientistriggeredbyshorting phaseatoground,i.e.,ua1=0.Notethatuc,d1 =sin(250t±2/3).Theoutputisthecurrentofphaseameasuredatbus4.Thesimulation resultsconfirmtheabovefrequencydomainanalysispresentedinFig.4andthefactthatthequasi-staticmodelbecomesinaccurateat highfrequencies.Thesimilarsettingswereusedtoobtainsimulationresultsforthe30-and118-busnetworks(seeFigs.9–11).Thefigures showthattransientresponsesofthedq0modelsaresimilartothoseofthetransientmodels.Thequasi-staticmodelisimplementedby constantgains,andisinaccurate,exceptforbalancedsinusoidalinputs.Thefirst-ordermodelismoderatelyaccurate.

(8)

Fig.8.Comparisonoftransientsinthe9-bussystem.Beforethetransient,theinputsarebalancedthreephasesinusoidalsignalsatbus1.Thetransientistriggeredby shortingphaseatoground.Theoutputisthecurrentofphaseaatbus4.

Fig.9.Comparisonofstepresponsesinthe30-busnetwork.Theinputsarestepsatt=0inphasesa,b,c.(a)inputatbus1,outputatbus2,(b)inputatbus5,outputatbus 7,(c)inputatbus8,outputatbus28,(d)inputatbus11,outputatbus9.

Table1

Errorsintransientresponses.

Fig.7 Fig.8(a) Fig.10 Fig.11(a)

MSE

dq0 0.9595×10−8 0.9572×10−9 0.6557×10−8 0.9340×10−7

fo 0.0035 3.4425×10−5 4.6116×10−5 4.9407×10−4

qs 0.1929 0.0020 0.0039 0.0170

ISE

dq0 0.7922×10−8 0.9706×10−9 0.3570×10−8 0.8491×10−7

fo 0.0035 3.4446×10−5 1.1498×10−5 1.2363×10−4

qs 0.1913 0.0020 9.6554×10−4 0.0042

(9)

Fig.10.Comparisonofstep-responsesinthe118-bussystem.Theinputs(bottomplot)arestepsinphasesa,b,cofbus76.Theoutputisthecurrentofphaseaatbus77.

Fig.11.Comparisonoftransientsinthe118-bussystem.Beforethetransient,theinputsarebalancedthreephasesinusoidalsignalsatbus76.Thetransientistriggeredby shortingphaseatoground.Theoutputisthecurrentofphaseaatbus77.

ErrorsinthemodelstransientresponsesarepresentedinTable1,wheretwotypicalmetrics,themeansquarederror(MSE)andthe integralsquarederror(ISE),areused.Theerrorsarecalculatedincomparisontothetransientmodel,whichservesasareference.Itcanbe seenthatthemagnitudesoferrorsforthedq0modelaresignificantlylowerthanthoseoffirst-orderandquasi-staticmodels.

7. Conclusion

Dynamicbehaviorandstabilityoflarge-scalepowersystemshavebeentraditionallystudiedbymeansoftime-varyingphasors,under theassumptionthatthesystemisquasi-static.However,withincreasingintegrationoffastrenewableanddistributedsourcesintopower grids,thisassumptionisbecomingincreasinglyinaccurate.Thispaperusesdynamicphasorsinthedq0referenceframetodescribethe dynamicsoflargetransmissionanddistributionnetworks.Theproposedmodelsdonotemploytheassumptionofquasi-staticphasors, andthereforeremainaccurateoverawiderangeoffrequencies.Thefulldq0modelisbasedonthefrequencydependentadmittance matrix,andisapproximatedusingtheTaylorseriesexpansiontogeneratemodelsoflowercomplexity:theclassicquasi-staticmodel isobtainedbyazero-orderapproximation,whilethemoreaccuratefirst-ordermodelisobtainedbyafirst-orderapproximation.The developedmodelscombinetwopropertiesofinterest.Ononehand,theyprovidetheadvantagesofthedq0referenceframe.Specifically,

(10)

themodelsemploynonrotatingdq0signalsthatarestaticatsteady-state,andthusenableawell-definedoperatingpoint.Ontheother hand,theproposedmodelsimprovetheaccuracyofthequasi-staticmodelathighfrequencies,enablingrepresentationoffastdynamic phenomenainnetworksthatincludesmalldistributedgeneratorsandfastpowerelectronicsbaseddevices.Inparticular,theseproperties cancontributetoaccurateandlowcomplexitystabilityanalysisinsuchnetworks.Simulationsshowthatthefrequencyresponsesofall modelscoincideatlowfrequenciesanddivergeathighfrequencies.Inaddition,responsesofthedq0modelinthetimedomainandinthe abcreferenceframeareveryclosetothoseofthetransientmodel.

Acknowledgment

TheworkwaspartlysupportedbyGrandTechnionEnergyProgram(GTEP)andaTechnionfellowship.

References

[1]Y.Liu,K.Sun,Y.Liu,Ameasurement-basedpowersystemmodelfordynamicresponseestimationandinstabilitywarning,Electr.PowerSyst.Res.124(2015)1–9.

[2]L.Miller,L.Cibulka,M.Brown,A.V.Meier,Electricdistributionsystemmodelsforrenewableintegration:Statusandresearchgapsanalysis,Tech.rep.,California EnergyCommission,CA,USA,2013.

[3]P.M.Anderson,A.A.Fouad,PowerSystemControlandStability,JohnWiley&Sons,2008.

[4]C.Dufour,J.Mahseredjian,J.Bélanger,Acombinedstate-spacenodalmethodforthesimulationofpowersystemtransients,IEEETrans.PowerDel.26(2)(2011) 928–935.

[5]U.N.Gnanarathna,A.M.Gole,R.P.Jayasinghe,EfficientmodelingofmodularmultilevelHVDCconverters(MMC)onelectromagnetictransientsimulationprograms, IEEETrans.PowerDel.26(1)(2011)316–324.

[6]R.Majumder,Someaspectsofstabilityinmicrogrids,IEEETrans.PowerSyst.28(3)(2013)3243–3252.

[7]T.Nishikawa,A.E.Motter,Comparativeanalysisofexistingmodelsforpower-gridsynchronization,NewJ.Phys.17(1)(2015).

[8]R.Zárate-Minano,T.V.Cutsem,F.Milano,J.A.Conejo,Securingtransientstabilityusingtime-domainsimulationswithinanoptimalpowerflow,IEEETrans.Power Syst.25(1)(2010)243–253.

[9]M.Ili ´c,J.Zaborszky,Quasistationaryphasorconcepts,in:DynamicsandControlofLargeElectricPowerSystems,Wiley,NewYork,2000,pp.9–60.

[10]P.C.Stefanov,A.M.Stankovi ´c,ModelingofUPFCoperationunderunbalancedconditionswithdynamicphasors,IEEETrans.PowerSyst.17(2)(2002)395–403.

[11]R.Yousefian,S.Kamalasadan,ALyapunovfunctionbasedoptimalhybridpowersystemcontrollerforimprovedtransientstability,Electr.PowerSyst.Res.137(2016) 6–15.

[12]A.M.Stankovi ´c,S.R.Sanders,T.Aydin,DynamicphasorsinmodelingandanalysisofunbalancedpolyphaseAmachines,IEEETrans.EnergyConvers.17(1)(2002) 107–113.

[13]S.Almer,U.Jonsson,Dynamicphasoranalysisofperiodicsystems,IEEETrans.Autom.Control54(8)(2009)2007–2012.

[14]M.Daryabak,S.Filizadeh,J.Jatskevich,A.Davoudi,M.Saeedifard,V.K.Sood,J.A.Martinez,D.Aliprantis,J.Cano,A.Mehrizi-Sani,ModelingofLCC-HVDCsystemsusing dynamicphasors,IEEETrans.PowerDel.29(4)(2014)1989–1998.

[15]T.Demiray,G.Andersson,L.Busarello,Evaluationstudyforthesimulationofpowersystemtransientsusingdynamicphasormodels,in:TransmissionandDistribution Conf.andExpo,Bogotá,Colombia,2008,pp.1–6.

[16]A.M.Stankovi ´c,T.Aydin,Analysisofasymmetricalfaultsinpowersystemsusingdynamicphasors,IEEETrans.PowerSyst.15(3)(2000)1062–1068.

[17]C.Liu,A.Bose,P.Tian,ModelingandanalysisofHVDCconverterbythree-phasedynamicphasor,IEEETrans.PowerDel.29(1)(2014)3–12.

[18]H.Zhu,Z.Cai,H.Liu,Q.Qi,Y.Ni,Hybrid-modeltransientstabilitysimulationusingdynamicphasorsbasedHVDCsystemmodel,Electr.PowerSyst.Res.76(6-7)(2006) 582–591.

[19]M.C.Chudasama,A.M.Kulkarni,DynamicphasoranalysisofSSRmitigationschemesbasedonpassivephaseimbalance,IEEETrans.PowerSyst.26(3)(2011) 1668–1676.

[20]P.Mattavelli,A.M.Stankovic,G.C.Verghese,SSRanalysiswithdynamicphasormodelofthyristor-controlledseriescapacitor,IEEETrans.PowerSyst.14(1)(1999) 200–208.

[21]Z.Miao,L.Piyasinghe,J.Khazaei,L.Fan,Dynamicphasor-basedmodelingofunbalancedradialdistributionsystems,IEEETrans.PowerSyst.30(6)(2015)3102–3109.

[22]S.Chandrasekar,R.Gokaraju,Dynamicphasormodelingoftype3DFIGwindgenerators(includingSSCIphenomenon)forshort-circuitcalculations,IEEETrans.Power Del.30(2)(2015)887–897.

[23]T.H.Demiray,Simulationofpowersystemdynamicsusingdynamicphasormodels.Ph.D.thesis,TUWien,2008.

[24]T.Yang,S.Bozhko,G.Asher,Multi-generatorsystemmodellingbasedondynamicphasorconcept,in:EuropeanConf.onPowerElectronicsandApplications,Lille, France,2013,pp.1–10.

[25]T.Bi,H.Liu,Q.Feng,C.Qian,Y.Liu,Dynamicphasormodel-basedsynchrophasorestimationalgorithmforM-classPMU,IEEETrans.PowerDel.30(3)(2015) 1162–1171.

[26]D.-G.Lee,S.-H.Kang,S.-R.Nam,Modifieddynamicphasorestimationalgorithmforthetransientsignalsofdistributedgenerators,IEEETrans.SmartGrids4(1)(2013) 419–424.

[27]R.K.Mai,L.Fu,Z.Y.Dong,K.P.Wong,Z.Q.Bo,H.B.Xu,DynamicphasorandfrequencyestimatorsconsideringdecayingDCcomponents,IEEETrans.PowerSyst.27(2) (2012)671–681.

[28]R.K.Mai,L.Fu,Z.Y.Dong,B.Kirby,Z.Q.Bo,AnadaptivedynamicphasorestimatorconsideringDCoffsetforPMUapplications,IEEETrans.PowerDel.26(3)(2011) 1744–1754.

[29]J.Ren,M.Kezunovic,Anadaptivephasorestimatorforpowersystemwaveformscontainingtransients,IEEETrans.PowerDel.27(2)(2012)735–745.

[30]T.Yang,S.Bozhko,J.M.Le-Peuvedic,G.Asher,C.I.Hill,Dynamicphasormodelingofmulti-generatorvariablefrequencyelectricalpowersystems,IEEETrans.Power Syst.31(1)(2016)563–571.

[31]X.Guo,Z.Lu,B.Wang,X.Sun,L.Wang,J.M.Guerrero,Dynamicphasors-basedmodelingandstabilityanalysisofdroop-controlledinvertersformicrogridapplications, IEEETrans.SmartGrids5(6)(2014)2980–2987.

[32]A.E.Fitzgerald,C.Kingsley,S.D.Umans,ElectricMachinery,6thed.,McGraw-Hill,NewYork,2003.

[33]J.J.Grainger,W.D.Stevenson,PowerSystemAnalysis,McGraw-Hill,NewYork,1994.

[34]T.V.Cutsem,C.Vournas,VoltageStabilityofElectricPowerSystems,KluwerAcademicPublishers,Norwell,MA,1998.

[35]R.D.Zimmerman,C.E.Murillo-Sánchez,R.J.Thomas,MATPOWER:steady-stateoperationsplanningandanalysistoolsforpowersystemsresearchandeducation,IEEE Trans.PowerSyst.26(1)(2011)12–19.

[36]Y.Levron,J.Belikov,Toolboxformodelingandanalysisofpowernetworksinthedq0referenceframe,MATLABCentralFileExchange,2016http://www.mathworks.

com/matlabcentral/fileexchange/58702(retrieved15.08.16).

Referensi

Dokumen terkait

Charger Energy Storage (battery) Power Converter Electric Motor Transmission, driveshaft Wheels Drive Control Signals Powertrain.. Figure 1: An Electric

Cara kerja Sistem Electric Power Steering (EPS) adalah saat kunci diputar ke posisi ON, Control Module memperoleh arus listrik untuk kondisi stand-by, bersamaan dengan itu

List of Tables TABLE I EXPERIMENTAL DESIGN CONDITION TABLE II BIDIRECTIONAL FLYBACK LOSS ANALYSIS TABLE III BIDIRECTIONAL FLYBACK DESIGN PARAMETERS TABLE IⅤ EXPERIMENTAL SET-UP

20 2 THREE PHASE CIRCUITS: POWER DEFINITIONS AND VARIOUS COMPO- NENTS Lectures 9-18 29 2.1 Three-phase Sinusoidal Balanced System.. 29 2.1.2 Three Phase Instantaneous Active

1, the energy stockpiling capability is given by module cross breed electric vehicle PHEV or EV batteries through bidirectional dc chargers coordinated inside a customary PV framework

Some of the functional highlights and differentiations from classical dispatch are: • Extension for price-based, distributed, less predictable resources • Active, dynamic demand •