Study on Voltage Unbalance Improvement Using SFCL in Power

Feed Network With Electric Railway System

strength of electrical devices, synchronous generators,

protectiverelays,linestransmission,andloads.Inaddition,the
electricalpowerneededbythesystemwhenafaultoccursis
modifiedandinducedinstablestateofthepowergenerators
[2]. Recently, the development of superconducting fault
current limiters (SFCL) offers one of the most attractive
alternativestosolvethefaultcurrentproblems[3][5].Thanks
to their fast transition from a low to a high impedance,
superconductingdevicescanlimit,inaveryshorttime,the
value of any fault current. What is attractive is that the
transitionisduetoanintransitbehaviorofthematerialitself
[6][9].
ThepresenceoftheSFCLinapowersystemcanincrease
thesystemstabilityanddistributedenergyquality[10][15].

Abstract
Thispaperpresentsanovelapproachtodetermine thepower
system stability using resistive superconducting fault current
limiter(SFCL)ofanelectricpowergrid(EPG).Inaddition,the
electricrailwaysystemhasrapidlychangingloadcharacteristics
intime.Anunbalanceisgeneratedowingtotherapidlychanging
large singlephase loads. Subsequently, the unbalanced load
causesanunbalancedtransmissionline.Avoltageunbalancein
thesourceinfluencesthepowerequipmentbycausingareduction
inthepowergenerationcapacityofthegeneratorandadecrease
intheoutputoftheotherfacilitiesinthetransmissionline.In
addition, many flexible ac transmission systems are applied to
transmissionlinestocompensateforandcontrolelectricpower.
Avoltageunbalancecausesacontrolerrorinthesesystems.In
addition,weanalyzedtheeffectsoftheproposedmethodusing
transient simulations. The angular separation of the rotors of
synchronousmachinespresentinthepowersystemisintroduced.
ItisshownthattheSFCLcanhavedifferentimpacts(positive
andnegative)infunctionofitslocationintheEPGwhenafault
occurs.Toevaluatetheeffectivenessoftheproposedmethod,the
IEEEbenchmarkedfourmachinetwoareatestsystemisusedto
carryoutseveralcasestudies.Theresultsshowthatthetransient
stability using SFCL combined with its optimal resistive value
reduces the angular separation of the rotors that improves
effectivelythesystemstabilityduringafault.
Index TermsDamping performance, electric power grid,
superconductingfaultcurrentlimiter,transientstability.

CHAPTER-I
INTRODUCTION
WITH the increasing of system capacity, fault
occurrenceprobabilitybecomeshigher,thatcaninducesevere
damagesinelectricalpowersystem[1].Forexample,thehigh
valueoftheshort circuit current candamage theinsulation

anddecreasesthelowfrequencyoscillationofthegenerators
speedwhenaseveredamageisintroduced(threephasefault).
The advantage of the proposed method is that the selected
locationoftheSFCLtakesintoaccountthefactthatthefault
canoccuranywhereinthestudiedgrid.

WhenanSFCLisintroducedinanelectricpowergrid(EPG),
threeimportantfactorsmustbeconsidered:
[1]optimallocationoftheSFCLintheEPG[16];
[2]optimalresistivevalueoftheSFCL[17];
[3] protectioncoordinationproblem withotherexistingde
vices(circuitbreaker,OCR,etc.).
Thispaperfocusesonfactors1and2.Nowadays,the
optimallocationofSFCLtoimprovethetransientstabilityis
studiedforasimplepowersystemlikesinglemachineinfinite
bus(SMIB)[2],[3],[15].Sofar,theremainingopenquestion
istheselectionofoptimallocationoftheSFCLinalargescale
powersystemtoimproveitstransientstability.Inthispaper
weconsideramultigeneratorsystemandweusetheirrotor
angulardifferencetodefineasensitivityindexwhichleadsto
finding the best location of the SFCL in the grid. This
sensitivity index is calculated with respect to the resistive
value of the used SFCL.The effectiveness of the proposed
method is evaluated on the IEEE benchmark fourmachine
twoarea test system. The toolbox Sim Power Systems of
MATLAB/SIMULINK software is used to carry out
simulationsstudies.
The simulation results show the effectiveness of the
proposedmethod.Infact,theoptimallocationdeterminedfor
theSFCLimprovesthetransientstabilityofthepowersystem

package.

CHAPTER2

II.SFCLMODELAPPLIEDTOPOWER
SYSTEMTheoperationofanSFCLisbasedonthenatural
transition of the superconducting state to normal state by
exceedingthecriticalcurrentIofthematerial.Thistransition
fromthesuperconductingtothenormalstatemustbedonein
averyshorttime,generally,tolimitthefirstcurrentpeaktoa
threshold value not exceeding three to five times the rated
current,belowtheshortcircuitcurrentwithoutlimitation.
TheSFCLisplacedinserieswithacircuitbreaker.During
thefault,thecurrentincreasesuptoreachthethresholdof
transitionfromsuperconductingwire.Thistransitionfromthe
super conducting element to normal state causes the
development ofresistance that limitsortriggers thecurrent
limit.Thetimebetweenthresholdcrossingandthelimitation

T.Sravani,G.HariKrishna,Sk.JanBasha

81

issmall(afewmicroseconds).Thecircuitbreakerisolatesthelineassoonaspossibleafterthebeginningofthe
limitation.Thesesuperconductingfaultcurrentlimitersuseoneofthefundamentalpropertiesofsuperconductors.
ThequalificationsofSFCLare:
1. Verylowimpedanceduringnormaloperation.
The current limiter must be invisible in this mode.
Sometransientssuchasthosecausedduringtheswitchingofatransformershouldnotinadvertentlycausea
transitionofthelimiter.
2. Highimpedancesystemduringshortcircuit.
Thelimitermustperformitsfunctioninthecaseofmassiveshortcircuitbutalsointhecaseoflowshort
circuitfault.
3. Verygooddynamic.
Thesystemmusttransitinveryquickly(within
millisecond)toeffectivelylimitthevalueoftheshortcircuit.
ThereareseveraltypesofSFCL;themainonesinclude:
1. Theresistivelimiter.
Thisisacoilofsuperconductivematerial,noninductivebyconstruction,mountedinseriesontheline.Incase
offault,thewindinginitiallyatthesuperconductingstatechangestothenormalstate.Itsimpedanceappearsin
theline,whichlimitsthefaultcurrent.ThislimitercanbeappliedinACorDCsystems.
2. Theinductivelimiter.
It consists in its simplest form, two windings connected in parallel. This association is made so that the
impedanceofallisthelowestpossible.Incaseoffault,oneofthecoils(madeofsuperconductingmaterial)
returnstoitsnormalstate.Theotherwindingmadeeithercopperorsuperconductingmateriallimitthecurrent
throughtheinductance.
3. Thelimitingtransformer.
Itconsistsofasuperconductingsecondary shortcircuit andprimarywindingofaconventionalcopper.In
normalOperation,theimpedanceofthistransformerismainlyduetothecouplingbetweenthewindingsand
maybeverylow.Underfault,thesecondarywindingquenchesundertheeffectoffaultcurrentandactsasa
switch that opens the secondary of the transformer. Therefore, we obtain the no load impedance of a
transformerwhichisveryimportant.
Aswesaid,thefirstfunctionofanSFCListolimitthe
faultcurrentinducedinthefaultyline.Inthiscase,thepassagetoitssuperconductingstatetoitsnormalstatecan
onlybeachievedifthefaultappearsinthelinewherethelimiterisplaced.Theprobabilitythatafaultappearsata
specificlocationinthepowersystemisverylow.Consequently,ifthefaultappearsatadifferentlocationinthe
powersystem,theSFCLcannotbeusedforitsfirstfunctionbecausethecurrentvaluethroughtheSFCLwillbenot
sufficienttoensureatransitionofthislast.TheoriginalityofthepresentedworkisthattheSFCLisusedifthefault
appearsanywhereinthesystem.Inthiscase,theSFCLcanbeseenasarapidswitchthatintroducesaresistancein
thepowersystem.Thepresenceofthisadditionalresistanceimprovesthepowersystemstabilityincaseofshort
circuitandthis,regardlessofthepositionofthefault.TheSFCLisnotusedtolimittheFaultcurrentbutto
improvethestabilityofthepowersystem.Ifthefaultappearsinthelinewherethelimiterisplaced,thetwo
functionsoftheSFCLareused(limitationofthefaultcurrentinthelineandimprovementofthepowersystem
stability).Bycons,ifafaultappearsatadifferentlocation,theSFCLintroduceitsresistanceinthepowersystemto
improvethetransientstabilityofgenerators.Inthiscase,thepassagetoitssuperconductingstatetoitsnormalstate
willbeaccomplishedbyapplyingamagneticfieldsuperiortoitscriticalmagneticfieldafterthedetectionofthe
faultintheEPG.Theresultspresentedinthispapershowtheeffectivenessoftheproposedmethod.Wecan

representtheSFCLbyimpedance,whichdependsoncurrent,orbyaresistance,whichvariesduringtime.The
SFCLusedinsimulationrepresentedresistancewhichvarieswithtimeas

(1)Where representsthemaximumresistancethatthe
SFCLcanintroduceinthelinesofthepowersystem.Also, isthetimeoftransitionfromthesuperconductingstate
tothenormalstate,whichisassumed,inthisstudy,tobe1ms[16].

III.DESCRIPTIONOFTHEPROPOSEDMETHOD
ForanEPGwithmorethanonegenerator,anecessaryconditionforsatisfactorysystemoperationisthatall
synchronousmachinesremaininsynchronismwhenaseveretransientdisturbanceappears(lossoftransmission
line,importantincreaseofload,threephaseshortcircuit).Thisaspectofstabilityisinfluencedbythedynamicsof
generatorsrotorangleandpoweranglerelationships.
Whenafaultoccursinthesystem,theoperatingpointofgeneratorssuddenlychanges.Owingtoinertia,the
angularseparation betweentherotorpositionsofeachgeneratorismodifieduntilthefaultiscleared.Atthe
beginningoftheperturbation,thisangularseparationincreaseswithamagnitudethatdependsonthetimenecessary
toclearthefault.Ifthefaultclearingtimeissmall,thepowersystemisperturbedbutastablepointisfoundagain
afterafewseconds.Otherwise,thesynchronismofthepowersystemislost.
Fig.1presentstheevolutionoftheangularseparation ofasimplepowersystem(2generatorsand1load)when
athreephaseshortcircuitoccursat1secnearoneofthegenerators.Inthepresentedfigure,theangularseparation
representsthedifferencebetweentheangularpositionsoftherotorsofthetwogenerators.Ifthefaultisclearedat1,
2sec,thepowersystemremainsstableandtheangle oscillatesbetweentwoextremevalues.Ontheotherhand,if
thefaultisclearedat1,3sec,thekineticenergygainedduringthetimeoffaulthasnotbeenyetcompletely
expendedtothesystem.Theangle continuestoincreaseafterthefaultcleared.Thepowersystemisnotcapableto
returntoastableposition,leadingtolossofsynchronism.Sungetal.in[15]havedemonstrated

Fig.1.Evolutionoftheangularseparation oftherotorswithafaultclearedat1,2s(stablecase)andafault
clearedat1,3s(unstablecase).

Fig.2.IEEEbenchmarkedfourmachinetwoareatestsystem.Whenafaultoccursconsequently,thetransient
stabilityofthesystemisincreased.

Startingfromthisobservation,weproposeamethodtoselecttheoptimallocationoftheSFCLinEPGbasedon
thestudyoftheangularseparation oftherotorsofsynchronousgeneratorsincaseoffault.Manystudieshave
dealtwiththeincreaseoftransientstabilityinpowersystemsthankstoSFCLbutmostofthemhavelimitedthe
studytoasimplepowersystemsuchasgenerator,transmissionline,andinfinitebus.Now,theremainingopen
questionsaretheselectionofoptimallocationoftheSFCLinapowersystemwithmorethan2generators.We
presentamethodwhichisnotlimitedtoasimplesystemlikeanSMIBbutextendedtotheIEEEbenchmarkedfour
machinetwoareatestsystempresentedinFig.2.Thispowersystemiscomposedoffourgenerators,twoloadsand
alongtransmissionline(220km),thatisaadvantagetostudytheoptimallocationofanSFCLinapowersystemin
comparisonwithSMIB.
Thedetailsonthissystem,includingthesystemsparameters,aregivenin[19].Allgenerators(G1toG4)are
equippedwithpowersystemstabilizers(PSSs).
IV.SIMULATIONSRESULTS
Toevaluatetheeffectivenessoftheproposedmethod,tenlocations(L1toL10inthemiddleofeachline)
areconsideredinFig.2.ForagivenpositionoftheSFCL(m=1to10),a100msthreephaseshortcircuitisapplied
at correspondingvalueoftheSFCLresistance usedinsimulation.WhentheSFCLwillbepositionedat each
location,thestudyofindexesSMDmandTSImwillgiveusthebestlocationoftheSFCLinthesystemincasea
threephasefaultcanappearatanypositioninthesystem.Thetimesimulationrequireduntilthesystemisrestored
to its original steadystate operation point after applying the fault is considered to be equal to 20 s. Before
calculatingvaluesofindex,itisnecessarytocomputetheloadflowofthepresentedpowersystem.Theresults
showthatthepowergeneratedbygeneratorsG1andG2ismostlyconsumedbyLoad1.Inarea2,thepower
generatedbygeneratorsG3andG4isconsumedbyLoad2butitisnotsufficient.Therestofthepowerrequiredby
Load2,i.e.,413MW,istransferredfromarea1toarea2bythe220kmtransmissionline.
Therefore,ifthevalueof
(3)isgreaterforagivenresistivevalue,thecorrespondinglocation
becomesmoreoptimal.ThedeterminationofoptimalresistivevalueofSFCLispresentedinthenextsub
section.

A. TCSC-Compensated Transmission

Power transmitted between a sending-end bus and a receiving-end bus in an AC

transmission system is dependent on the series impedance. Further, impedance of a
transmission line consists mainly of inductive reactance, with resistance
accounting for only 510% of impedance [5], [6]. If a series capacitor is inserted
into transmission line, the inductive reac-tance of transmission line could be
compensated by a capacitive supply. This concept of series compensation is
illustrated in Fig. 1 [6]. Typical configuration of a TCSC from a steady-state
perspective involves a fixed capacitor (FC) with a thyristor-controlled reactor
(TCR). TCSC structure is shown in Fig. 2.

Fig. 3. TCSC closed-loop constant current control methodology.

Fig. 5. Transmission system connected with a railway.

CHAPTER 3
MODELING THE SFCL AND FAULT OF THE TRANSMISSION LINE IN AN
ELECTRIC RAILWAY

Control of typically applies open-loop control or closed-loop control. Fig. 3 details a

schematic of a constant current closed-loop control [8]. In Fig. 3, Iref is desired transmission
line current, IM is actual current, and Ierror is difference between Iref and IM . In particular,
Ierror is an important quantity in this control loop. A current unbalance can cause serious
issues for TCSC control.

. Unbalanced Load in an Electric Railway System

An electric railway characteristic that most utilities are concerned with is current
unbalance produced by large single-phase loads These unbalanced currents cause 3-phase
voltage unbalance. To minimize voltage unbalances in 3-phase power feed networks, Scott
transformers are widely used. Nevertheless, an unbalance can be generated owing to large
rapidly changing single-phase loads. Fig. 4 shows an

In order to limit a fault current, many models for the SFCL have been developed: resistortype, reactor-type, transformer-type, etc. In this study, we modeled a resistor-type SFCL
that is mostly basic and used widely which represents the exper-imental studies for
superconducting elements of SFCL. Quench characteristics and recovery characteristics of
a resistor-type SFCL are modeled based on [11] and [12]. An impedance of the SFCL
according to time t is given as follows:

(t < t0)

t
t 02
ex
Zn 1 p

Z(t)

F

a1(t t1) + b1
a2( t2) +

(t0 t < t1)

(2)

(t1 t < t2)

(t t2)

b2

where Zn and TF are the impedance saturated normal tem-perature and time constant. In
addition, t0, t1, and t2 are the quench-starting time, first recovery-starting time, and secondary recovery-starting time, respectively. a1, a2, b1, and b2 are coefficients of the firstorder linear function to denote the experimental results of the recovery characteristics of
the SFCL The values used for the parameters are listed in Table I. The recovery time of the
SFCL is set to the value until fault clearing to protect the TCSC and transmission line.

B. Fault Modeling of the Transmission Line in the Electric Railway

An electric railway is connected to a 154 kV transmission line through a Scott

transformer in Korea. Fig. 5 shows a trans-mission system connected with railway
equivalent model. A

TABLE II
PARAMETERS OF TRANSMISSION LINE FAULT SIMULATION

Fig. 6 Voltage and current results from the simulation: (a) without railway and (b) with
railway.

Fig7. Transmission line three-phase fault simulation equivalent circuit and fault location

TCSC facility is installed to control power flow in transmission line, and an electric railway
includes a single phase load that causes a voltage unbalance. Table II shows the
transmission and electric railway system parameters of simulation. In addition, the self- and
mutual impedance of transmission and rail of the Korean electric railway system were
considered Here, in order to analyze the influence of an electric railway connection on the
transmission line fault, we simulated a fault situation. Fig. 6 shows fault location on the
transmission line with TCSC. Fault starting time is 0.4 s and fault duration time is 0.1 s.
Fig. 7(a) and (b) show simulation results for a 3-phase fault for an electric railway
connection, respectively. In case of fault on the transmission line without electric railway,
3-phase voltage decreased owing to fault and after fault removal, there is a transient
phenomenon that has a small offset voltage but returned to a steady state. In addition, this
simulation showed a small magnitude difference between offset voltages. Current is
generated a large transient current during fault and reach to steady state, load current, after
fault removal. In contrast, large offset voltages are represented in simulation result about
fault on transmission line with electric railway such as Fig. 7(b). Especially, voltage of
phase B is increased up to 350 kV in a moment. In addition, voltage unbalances of
transmission line became more serious compared with above case. Line currents are also
increased and caused unbalance owing to transient voltage.

We think that closing phase angle control of TCSC system is influenced by generated
transient voltage and current as the cause of these results. These phenomenon will cause
problem about voltage stability and malfunction of protection scheme in the transmission
grid. These results show that an electric railway connection aggravates voltage unbalance in
the transmission line.

CHAPTER -4
METHOD FOR IMPROVING THE VOLTAGE UNBALANCE

USING THE SFCL FROM THE LINE FAULT

We propose a method to improve voltage unbalance using an SFCL. If a transmission line

fault occurs, fault current is decreased by the SFCL. After fault is cleared, voltage unbalance produced by railways single-phase loads can quickly be reduced. As a result, a
transmission and electric railway system return to the steady state. In particular, SFCL has
advantage of fast operation within 1/4 cycle. The operation characteristics of conventional
relay and breaker are greater than 515 cycles [1].

A. Case Study

As mentioned in the fault modeling, a single-phase load in an electric railway is one of

the causes of voltage unbalance in the transmission system. However, the proposed method
using SFCL can alleviate this problem. SFCL was installed in the transmission line between
TCSC and electric railway in the case study simulation using the PSCAD/EMTDC
software package. SFCL decreased the voltage unbalance.

Fig. 8 shows the transmission system including an elec-tric railway and SFCL. Fig. 9(a)
and (b) show the results

when SFCL has an impedance of 10 and 20 , respectively. The results showed fault
current limiting characteristics and improvement in voltage unbalance. In addition, we
discov-ered following features: 1) the larger resistance of a resistor-type SFCL, the more
voltage unbalance was improved. 2) In SFCL operating process, there are differences of
recovery time between superconducting elements owing to unbalance fault current shown
in Fig. 7(b). If a fault occurs, the proposed method clears voltage unbalance and protects
TCSC. Thus, transmission system can quickly return to operating in a con-ventional state.
As a result, we will expect improvement effect for the problems about voltage stability and
protection scheme malfunction.

CHAPTER 5
MATLAB PROGRAMS

4.1 Introduction to Mat Lab:

MATLAB

stands for MATrix LABoratory which has been developed by Math Works Inc. Mat

Lab provides matrix as one of the basic elements. It is dynamically decided depending on what value we
assigned to it.
MATLAB is case sensitive and so we have to be careful about the case of the variables while using
them in our programs. The combination of analysis capabilities, flexibility, reliability and powerful
graphics makes MATLAB the main software package for power system engineers

MATLAB gives an interactive environment with hundreds of reliable and accurate built in
function. These functions help in providing the solution to a variety of mathematical problems including
matrix algebra, linear system, differential equation, optimization, non - linear system and many other types
of scientific and technical computations. The most important feature of MATLAB is its programming
capability, which supports both types of programming object oriented and structured programming and
is very easy to learn and use and allow user developed functions.

4.2 Program

clc
close all
clear all
tic

%
gendata = [

1.2

0.0025

0.00032

1.15 0.0022

0.00028

1.1

genn=[1
2
3];

genno=gendata (:,1);
A=gendata(:,1);
B=gendata(:,2);
C=gendata(:,3);

0.002

0.00026];

gn=max(genn);
fprintf( ' pd lamda pg( 1) pg(2) pg(3)\n ');
j=1;
for lamda=1.873:0.05:4.35
for n=1:gn

p(n)=-(B(n)/C(n))+sqrt((B(n)/C(n))^2-4*(A(n)-lamda)/C(n));
pg(n)=p(n)/2;
pd=sum(pg);
end
disp([pd,lamda,pg( 1 ),pg(2),pg(3)]);
pg1(j)=pg(1);
pg2(j )=pg(2);
pg3(j )=pg(3);
pdn(j )=pg( 1 )+pg(2)+pg(3);
lam(j )=lamda;
j =j +1;

end
toc

plot (pg1, lam,'--', pg2, lam,'-', pg3, lam,'--', pdn, lam,'--');

xlabel ('LOAD IN MW');

ylabel ('INCREMENTAL FUEL COST (LAMDA)');
title ('INCREMENTAL COST CURVES TO MEET THE TOTAL SYSTEM LOAD DEMAMD AT
MINIMUM COST');
grid on

(a)Case Study 1:
clc
%nntwarn off
%ECONOMIC LOAD SCHEDULING WITHOUT TRANSMISSION LOSSES BY ANN

input = [0.3580 0.3919 0.4233 0.4527 0.4804 0.5067 0.5318 0.5558 0.5789 0.6011

0.6226

0.6434 0.6635 0.6831 0.7022 0.7208 0.7390 0.7567 0.7741 0.7911 0.8077 0.8240 0.8401 0.8558
0.8712 0.8864 0.9014 0.9161 0.9306 0.9449 0.9589 0.9728 0.9865 1.0000];

Pd=input*910;

tt = [0.3407 0.3764 0.4093 0.4399 0.4686 0.4958 0.5217 0.5465 0.5702 0.5931
0.6151 0.6364 0.6571 0.6772 0.6967 0.7157 0.7343 0.7524 0.7701 0.7874
0.8044 0.8210 0.8373 0.8533 0.8691 0.88450.8998 0.9147 0.9295 0.9440
0.9583 0.9724 0.9863 1.0000;
0.3577 0.3917 0.4231 0.4525 0.4802 0.5065 0.5316 0.5556 0.5787 0.6009
0.6224 0.6432 0.6634 0.6830 0.7021 0.7207 0.7389 0.7566 0.7740 0.7910
0.8076 0.8240 0.8400 0.8557 0.8712 0.8864 0.9013 0.9161 0.9306 0.9448

0.9589 0.9728 0.9865 1.0000;

0.3736 0.4060 0.4361 0.4644 0.4911 0.5166 0.5409 0.5642 0.5867 0.6084
0.6293 0.6496 0.6694 0.6886 0.7073 0.7255 0.7433 0.7607 0.7777 0.7944
0.8108 0.8268 0.8425 0.8580 0.8732 0.8882 0.9029 0.9174 0.9316 0.9457
0.9595 0.9732 0.9867 1.0000];

pg1=tt(1,:)*467;
pg2=tt(2,:)*320;
pg3=tt(3,:)* 140;

disp('

PG1

PG2

PG3

PD=PG1+PG2+PG3 ')

disp([pg1' pg2' pg3' Pd'])

disp ('DATA USED FOR TRAING THE NEURAL NETWORK')
pause
clc

% INTIALIZATON OF NEURAL NETWORK %

net = newff ([minmax (input)], [10 3], {'tansig','purelin'}, 'trainlm');
disp ('NEURAL NETWORKS WEIGHTS AND BIAS BEFORE TRAINING')
net.iw {1, 1}

net.b {1, 1}
net.b {2, 1}
pause
clc

% TRAINING PARAMETERS%
net.trainParam.epochs = 1000;
net.trainParam.goal = 0;
%
net= train (net, input, tt);
clc
disp
net.iw {1, 1}
net.lw {2, 1}
net.b {1}
net.b {2}
pause
clc

% SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT %

ysim = sim (net, input);
disp ('SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT')
pgs1 = ysim (1, :)*467;
pgs2 = ysim (2, :)*320;
pgs3 = ysim (3,:)* 140;
disp ([pg1' pgs1' pg2' pgs2' pg3' pgs3'])
pause
clc
%TESTING OF ANN%
test = [0.4358 0.4959 0.5560 0.6162 0.6763 0.7363 0.7965 0.8566 0.8867];
tic
a=sim (net, test);
toc
pgt1=a(1,:)*467;
pgt2 = a (2,:)*320;
pgt3 = a (3, :)* 140;
test = test*910;
disp ('SIMULATION FOR TEST INPUT')
disp ([test' pgt1 pgt2 pgt3])
save annstu net

(b)Case Study 2:

warning off
close all
clc
lsg;
%nntwarn off
input=pdn;
Pd=input/max(input);
Pgs= [pg1' pg2' pg3' pdn'];
tt= [pg1/ 94.9605;pg2/102.5964;pg3/ 107.5583];

disp('

PG1

PG2

PG3

PD=PG1+PG2+PG3')

disp([pg1' pg2' pg3' input'])

disp('DATA USED FOR TRAING THE NEURAL NETWORK')
pause
clc

% INTIALIZATON OF NEURAL NETWORK %

net=newff([minmax(input)],[10 3],{'tansig','purelin'},'trainlm');
disp('NEURAL NETWORKS WEIGHTS AND BIAS BEFORE TRAINING')
net.iw{1,1}
net.lw{2,1}
net.b{1,1}
net.b{2,1}
pause
clc
% TRAINING PARAMETERS%
net.trainParam.epochs = 1000;
net.trainParam.goal = 0;
%ANN TRAINING %
net= train(net,input,tt);
clc
disp
net.iw{1,1}
net.lw{2,1}
net.b{1}
net.b{2}
pause

clc
% SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT %
tic
ysim=sim(net,input);
toc
disp('SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT')
pgs1=ysim(1,:)*94.9605;
pgs2=ysim(2,:)*102.5964;
pgs3=ysim(3,:)* 107.5583;
pds=pgs1+pgs2+pgs3;
pdact=input*305.1152;

format short g
disp([pg1' pgs1' pg2' pgs2' pg3' pgs3'])
pause
clc

CHAPTER 5
RESULTS

5.1 Graphical Method

PD

LAMDA

PG (1)

PG (2)

PG (3)

139.9728

1.8730

42.1197

47.0379

50.8153

145.0944

1.9230

43.7869

48.7606

52.5469

150.0535

1.9730

45.3977

50.4288

54.2269

154.8646

2.0230

46.9576

52.0472

55.8597

159.5403

2.0730

48.4711

53.6202

57.4490

164.0915

2.1230

49.9420

55.1513

58.9982

168.5276

2.1730

51.3738

56.6437

60.5100

172.8569

2.2230

52.7694

58.1003

61.9871

177.0868

2.2730

54.1315

59.5234

63.4319

181.2238

2.3230

55.4624

60.9152

64.8462

185.2739

2.3730

56.7640

62.2779

66.2320

189.2423

2.4230

58.0384

63.6130

67.5909

193.1337

2.4730

59.2870

64.9222

68.9

PG (1)

PG (2)

PG (3)

60.5114

66.2070

70.2340

PD

LAMDA

196.9525

2.5230

200.7025

2.5730

61.7130

67.4687

71.5208

204.3875

2.6230

62.8930

68.7085

72.7860

208.0106

2.6730

64.0524

69.9275

74.0306

211.5748

2.7230

65.1925

71.1267

75.2557

215.0831

2.7730

66.3140

72.3070

76.4621

218.5378

2.8230

67.4179

73.4693

77.6506

221.9414

2.8730

68.5050

74.6144

78.8220

225.2961

2.9230

69.5760

75.7431

79.9771

228.6040

2.9730

70.6316

76.8560

81.1164

231.8669

3.0230

71.6724

77.9538

82.2407

235.0866

3.0730

72.6991

79.0370

83.3505

238.2649

3.1230

73.7123

80.1063

84.4463

241.4032

3.1730

74.7123

81.1622

85.5287

244.5031

3.2230

75.6999

82.2051

86.5982

PG (1)

PG (2)

PG (3)

PD

LAMDA

241.4032

3.1730

74.7123

81.1622

85.5287

244.5031

3.2230

75.6999

82.2051

86.5982

247.5659

3.2730

76.6753

83.2355

87.6551

3.3230

77.6390

84.2539

88.7000

253.5854

3.3730

78.5915

85.2607

89.7332

256.5445

3.4230

79.5332

86.2562

90.7552

259.4713

3.4730

80.4643

87.2408

91.7662

262.3668

3.5230

81.3852

88.2150

92.7666

265.2320

3.5730

82.2963

89.1789

93.7568

268.0679

3.6230

83.1979

90.1330

94.7370

270.8753

3.6730

84.0902

91.0775

95.7076

250.5929

273.6551

3.7230

84.9736

92.0127

96.6688

276.4080

3.7730

85.8483

92.9388

97.6209

279.1349

3.8230

86.7146

3.8562

98.5641

281.8365

3.8730

87.5726

94.7651

99.4988

PG (2)

PG (3)

PD

LAMDA

279.1349

PG (1)

3.8230

86.7146

3.8562

98.5641

281.8365

3.8730

87.5726

94.7651

99.4988

284.5134

3.9230

88.4227

95.6656

100.4251

287.1663

3.9730

89.2650

96.5581

101.3432

4.0230

90.0998

97.4428

102.2533

4.0730

90.9272

98.3198

103.1558

289.7959

292.4027

294.9874

4.1230

91.7475

99.1893

104.0506

297.5504

4.1730

92.5608

100.0515

104.9381

300.0924

4.2230

93.3672

100.9067

105.8185

302.6139

4.2730

94.1671

101.7549

106.6918

305.1152

4.3230

94.9605

102.5964

107.5583

Figure 5.1

Elapsed time is 0.531000 seconds.

0000000000000000000000000000000000000000000000000000000000000
Case Study 1:

PG1

PG2

PG3

PD=PG1+PG2+PG3

159.1069

114.4640

52.3040

325.7800

175.7788

125.3440

56.8400

356.6290

191.1431

135.3920

61.0540

385.2030

205.4333

144.8000

65.0160

411.9570

218.8362

153.6640

68.7540

437.1640

231.5386

162.0800

72.3240

461.0970

243.6339

170.1120

75.7260

483.9380

255.2155

177.7920

78.9880

505.7780

266.2834

185.1840

82.1380

526.7990

276.9777

192.2880

85.1760

547.0010

287.2517

199.1680

88.1020

566.5660

297.1988

205.8240

90.9440

585.4940

306.8657

212.2880

93.7160

603.7850

316.2524

218.5600

96.4040

621.6210

325.3589

224.6720

99.0220

639.0020

334.2319

230.6240

101.5700

655.9280

342.9181

236.4480

104.0620

672.4900

351.3708

242.1120

106.4980

688.5970

359.6367

247.6800

108.8780

704.4310

367.7158

253.1200

111.2160

719.9010

375.6548

258.4320

113.5120

735.0070

383.4070

263.6800

115.7520

749.8400

PG1

PG2

PG3

PD=PG1+PG2+PG3

391.0191

268.8000

117.9500

764.4910

398.4911

273.8240

120.1200

778.7780

405.8697

278.7840

122.2480

792.7920

413.0615

283.6480

124.3480

806.6240

420.2066

288.4160

126.4060

820.2740

427.1649

293.1520

128.4360

833.6510

434.0765

297.7920

130.4240

846.8460

440.8480

302.3360

132.3980

859.8590

447.5261

306.8480

134.3300

872.5990

454.1108

311.2960

136.2480

885.2480

460.6021

315.6800

138.1380

897.7150

467.0000

320.0000

140.0000

910.0000

Ans =
-43.6137
-43.6137
43.6137
-43.6137
43.6137
43.6137
-43.6137
43.6137
-43.6137
-43.6137
Ans =
43.6137
40.5026
-37.3915
34.2804
-31.1693
-28.0582
24.9470
-21.8359
18.7248
15.6137

Ans =
0.3853
-0.8318
-0.0913

Figure 5.3 (a)

WEIGHTS AND BIAS AFTER TRAINING

Ans =
-18.3671
-6.9487
21.3496
-25.2423
16.1732
10.4797
-10.3463
12.5184
-38.8156
-42.8783

Ans =
Columns 1 through 7

-0.0163 -0.1359

0.0072 -0.0062

0.0197

0.0585 -0.0584

-0.0163 -0.1335

0.0070 -0.0062

0.0193

0.0573 -0.0567

-0.0162 -0.1313

0.0070 -0.0061

0.0189

0.0561 -0.0552

Columns 8 through 10

0.0540 -0.0066 -0.0225

0.0520 -0.0062 -0.0213
0.0500 -0.0059 -0.0202

Ans =

18.8009
6.4658
-18.1833
20.2644
-12.1753
-7.0560
5.8538
-5.8304
16.8740
16.4497

Ans =

0.7128
0.7210
0.7285

SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT

PG1

PG1

PG2

PG2

PG3

PG3

Actual

Obtained

Actual

Obtained

Actual

Obtained

159.1069

159.1054

114.4640

114.4655

52.3040

52.3038

175.7788

175.7834

125.3440

125.3397

56.8400

56.8405

191.1431

191.1394

135.3920

135.3954

61.0540

61.0537

205.4333

205.4322

144.8000

144.8021

65.0160

65.0154

218.8362

218.8351

153.6640

153.6602

68.7540

68.7564

PG1

PG1

PG2

PG2

PG3

Obtained

Actual

PG3

Actual

Obtained

Actual

231.5386

231.5435

162.0800

162.0807

72.3240

72.3205

243.6339

243.6447

170.1120

170.1125

75.7260

75.7272

255.2155

255.1895

177.7920

177.7876

78.9880

78.9892

266.2834

266.3055

185.1840

185.1887

82.1380

82.1403

276.9777

276.9620

192.2880

192.2940

85.1760

85.1700

287.2517

287.2514

199.1680

199.1638

88.1020

88.1032

297.1988

297.2125

205.8240

205.8229

90.9440

90.9497

306.8657

306.8587

212.2880

212.2781

93.7160

93.7115

316.2524

316.2517

218.5600

218.5689

96.4040

96.4049

325.3589

325.3656

224.6720

224.6766

99.0220

99.0212

334.2319

334.2275

230.6240

230.6182

101.5700

101.5673

342.9181

342.9182

236.4480

236.4479

104.0620

104.0662

351.3708

351.3654

242.1120

242.1176

106.4980

106.4972

Obtained

359.6367

359.6347

247.6800

247.6728

108.8780

108.8802

367.7158

367.7250

253.1200

253.1139

111.2160

111.2154

375.6548

375.6553

258.4320

258.4521

113.5120

113.5079

383.4070

383.4026

263.6800

263.6685

115.7520

115.7497

391.0191

391.0236

268.8000

268.7992

117.9500

117.9561

398.4911

398.4936

273.8240

273.8271

120.1200

120.1194

405.8697

405.8466

278.7840

278.7761

122.2480

122.2493

413.0615

413.0828

283.6480

283.6472

124.3480

124.3458

420.2066

420.1998

288.4160

288.4390

126.4060

126.4081

427.1649

427.1742

293.1520

293.1358

128.4360

128.4296

PG1

PG1

PG2

PG2

PG3

PG3

Actual

Obtained

Actual

Obtained

Actual

Obtained

434.0765

434.0645

297.7920

297.7772

130.4240

130.4274

440.8480

440.8632

302.3360

302.3587

132.3980

132.3996

447.5261

447.5130

306.8480

306.8420

134.3300

134.3300

454.1108

454.1090

311.2960

311.2922

136.2480

136.2468

460.6021

460.6094

315.6800

315.6819

138.1380

138.1384

467.0000

466.9976

320.0000

319.9999

140.0000

140.0000

Elapsed time is 0.016000 seconds.

SIMULATION OF ANN FOR TEST INPUT

Pd

PG1

PG2

PG3

396.5780

197.0694

139.2940

62.6943

451.2690

226.3212

158.6186

70.8542

505.9600

255.2858

177.8516

79.0164

560.7420

284.1897

197.1187

87.2297

615.4330

312.9971

216.3887

95.4713

670.0330

341.6280

235.5822

103.6950

724.8150

370.3041

254.8496

111.9607

779.5060

398.8754

274.0841

120.2300

806.8970

413.2253

283.7432

124.3871

Case Study 2:

PG1

PG2

PG3

PD=PG1+PG2+PG3

42.12

47.038

50.815

139.97

43.787

48.761

52.547

145.09

45.398

50.429

54.227

150.05

46.958

52.047

55.86

154.86

48.471

53.62

57.449

159.54

49.942

55.151

58.998

164.09

51.374

56.644

60.51

168.53

52.769

58.1

61.987

172.86

54.132

59.523

63.432

177.09

55.462

60.915

64.846

181.22

56.764

62.278

66.232

185.27

58.038

63.613

67.591

189.24

59.287

64.922

68.924

193.13

60.511

66.207

70.234

196.95

61.713

67.469

71.521

200.7

62.893

68.709

72.786

204.39

64.052

69.928

74.031

208.01

65.192

71.127

75.256

211.57

66.314

72.307

76.462

215.08

67.418

73.469

77.651

218.54

68.505

74.614

78.822

221.94

69.576

75.743

79.977

225.3

70.632

76.856

81.116

228.6

71.672

77.954

82.241

231.87

72.699

79.037

83.35

235.09

73.712

80.106

84.446

238.26

PG1

PG2

PG3

PD=PG1+PG2+PG3

74.712

81.162

85.529

241.4

75.7

82.205

86.598

244.5

76.675

83.236

87.655

247.57

77.639

84.254

88.7

250.59

78.592

85.261

89.733

253.59

79.533

86.256

90.755

256.54

80.464

87.241

91.766

259.47

81.385

88.215

92.767

262.37

82.296

89.179

93.757

265.23

83.198

90.133

94.737

268.07

84.09

91.077

95.708

270.88

84.974

92.013

96.669

273.66

85.848

92.939

97.621

276.41

86.715

93.856

98.564

279.13

87.573

94.765

99.499

281.84

88.423

95.666

100.43

284.51

89.265

96.558

101.34

287.17

90.1

97.443

102.25

289.8

90.927

98.32

103.16

292.4

91.747

99.189

104.05

294.99

92.561

100.05

104.94

297.55

93.367

100.91

105.82

300.09

94.167

101.75

106.69

302.61

94.96

102.6

107.56

305.12

DATA USED FOR TRANING

WEIGHTS AND BIAS BEFORE TRAINING

ans =

0.16955
0.16955
-0.16955
0.16955
0.16955
0.16955
-0.16955
0.16955
0.16955
-0.16955

ans =

Columns 1 through 7

0.31106

0.39816

0.31043

-0.14949

0.43308

-0.21619

-0.20563

0.67517

0.18933

0.022623

0.25463

-0.17274

-0.25678

0.13148

0.5528

Columns 8 through 10

-0.030073

-0.26925

0.64602

-0.77077

-0.71991

0.3479

0.32971

0.13355

0.99889

-0.021309

0.96542

-0.62819

0.61328

0.40127

0.40714

ans =

-51.732
-48.621
45.51
-42.399
-39.288
-36.177
33.066
-29.955
-26.844
23.73

ans =

0.92327
-0.88228
-0.27938

WEIGHTS AND BIAS AFTER TRAINING

ans =

0.0683
-0.0627
0.0525
-0.0428
0.0401
0.0368
-0.0265
-0.0307
0.0220
-0.0358

ans =

Columns 1 through 9

0.0307 -0.0175

0.0210 -0.0310

0.0251

0.0213 -0.0671 -0.0203

0.1081

0.0301 -0.0172

0.0206 -0.0305

0.0245

0.0208 -0.0655 -0.0198

0.1042

0.0296 -0.0169

0.0202 -0.0299

0.0240

0.0204 -0.0640 -0.0193

0.1005

Column 10

-0.0448
-0.0418
-0.0392

ans =

-21.4224
18.7187
-15.0221
11.6138
-10.1758
-8.6791
5.6815
5.7437
-3.4178
4.3214

ans =

0.6809
0.6916
0.7013

SIMULATION OF ANN FOR TRAINED INPUT-OUTPUT

PG1
Actual

PG1

PG2

PG2

PG3

PG3

Obtained

Actual

Obtained

Actual

42.12

42.12

47.038

47.038

50.815

50.815

43.787

43.786

48.761

48.76

52.547

52.547

45.398

45.398

50.429

50.429

54.227

54.227

46.958

46.958

52.047

52.047

55.86

55.86

PG1

PG1

PG2

PG2

PG3

PG3

Actual

Obtained

Obtained

Actual

Obtained

Actual

Obtained

48.471

48.471

53.62

53.62

57.449

57.449

49.942

49.942

55.151

55.151

58.998

58.998

51.374

51.374

56.644

56.644

60.51

60.51

52.769

52.77

58.1

58.1

61.987

61.987

54.132

54.132

59.523

59.524

63.432

63.432

55.462

55.462

60.915

60.915

64.846

64.846

56.764

56.764

62.278

62.278

66.232

66.232

58.038

58.038

63.613

63.613

67.591

67.591

59.287

59.287

64.922

64.922

68.924

68.924

60.511

60.511

66.207

66.207

70.234

70.234

61.713

61.713

67.469

67.469

71.521

71.521

62.893

62.893

68.709

68.709

72.786

72.786

64.052

64.052

69.928

69.927

74.031

74.031

65.192

65.192

71.127

71.127

75.256

75.256

66.314

66.314

72.307

72.307

76.462

76.462

67.418

67.418

73.469

73.469

77.651

77.651

68.505

68.505

74.614

74.615

78.822

78.822

69.576

69.576

75.743

75.743

79.977

79.977

70.632

70.632

76.856

76.856

81.116

81.116

71.672

71.672

77.954

77.954

82.241

82.241

72.699

72.699

79.037

79.037

83.35

83.35

73.712

73.712

80.106

80.106

84.446

84.446

74.712

74.712

81.162

81.162

85.529

85.529

PG1

PG1

PG2
Actual

PG2

PG3

Obtained

Actual

PG3

Actual

Obtained

Obtained

75.7

75.7

82.205

82.205

86.598

86.598

76.675

76.675

83.236

83.235

87.655

87.655

77.639

77.639

84.254

84.254

88.7

88.7

78.592

78.591

85.261

85.261

89.733

89.733

79.533

79.533

86.256

86.256

90.755

90.755

80.464

80.464

87.241

87.241

91.766

91.766

81.385

81.385

88.215

88.215

92.767

92.767

82.296

82.296

89.179

89.179

93.757

93.757

83.198

83.198

90.133

90.133

94.737

94.737

84.09

84.09

91.077

91.077

95.708

95.708

84.974

84.974

92.013

92.013

96.669

96.669

85.848

85.848

92.939

92.939

97.621

97.621

86.715

86.715

93.856

93.856

98.564

98.564

87.573

87.573

94.765

94.765

99.499

99.499

88.423

88.423

95.666

95.666

100.43

100.43

89.265

89.265

96.558

96.558

101.34

101.34

90.1

90.1

97.443

97.443

102.25

102.25

90.927

90.927

98.32

98.32

103.16

103.16

91.747

91.747

99.189

99.189

104.05

104.05

92.561

92.561

100.05

100.05

104.94

104.94

93.367

93.367

100.91

100.91

105.82

105.82

94.167

94.167

101.75

101.75

106.69

106.69

94.96

94.96

102.6

102.6

107.56

107.56

CHAPTER 5
CONCLUSION

This paper proposed a method to reduce voltage unbalance for a TCSC-compensated

transmission line using an SFCL. First, the configuration and operation of a compensated
trans-mission line and connected electric railway system were mod-eled and detailed. Next,
voltage unbalance in transmission line was studied when line fault occurs. Finally, the
method for alleviating this problem with SFCL was considered.

The proposed method showed the following improvements for transmission line faults: 1)
the fault current was decreased as compared to the existing system fault current and 2)
voltage unbalance in the transmission system was quickly improved after the fault was
removed. In future, we will study a protection scheme using an SFCL for a compensated
transmission system to improve system stability.

REFERENCES

[4]

H.-S. Shin, S.-M. Cho, J.-S. Huh, J.-C. Kim, and D.-J. Kweon, Applica-tion on of

SFCL in automatic power changeover switch system of elec-tric railway, IEEE Trans.
Appl. Supercond., vol. 22, no. 3, p. 5600704, Jun. 2012.

[5]

S. T. Senini and P. J. Wolfs, Novel topology for correction of unbalanced load in

single phase electric traction systems, in Proc. IEEE Power Electron. Spec. Conf.,
2002, vol. 3, pp. 12081212.

[6]

T. Uzuka, S. Ikedo, and K. Ueda, A static voltage fluctuation compen-sator for AC

electric railway, in Proc. IEEE Power Electron. Spec. Conf., 2004, vol. 3, pp. 1869
1873.

[7]

A. D. Del Rosso, C. A. Canizares, and V. M. Dona, A study of TCSC controller

design for power system stability improvement, IEEE Trans. Power Syst., vol. 18, no.
4, pp. 14871496, Nov. 2012.

[8]

S. S. N. Singh and A. K. David, Optimal location of FACTs devices for congestion

management, Elect. Power Syst. Res., vol. 58, no. 2, pp. 71 79, Jun. 2001.

[9]

J.-U. Lim, J.-C. Seo, and S.-I. Moon, Selection of optimal TCSC location to keep

the steady-state voltage profile within limits, presented at the KIEE Conf., 1998.

[10]

A. Kazemi and B. Badrzadeh, Modeling and simulation of SVC and TCSC to study

their limits on maximum loadability point, Int. J. Elect. Power Energy Syst., vol. 26,
no. 8, pp. 619626, Oct. 2004.

[11]

Y. Tang, R. Yu, and M. Yan, Research on the synchronous voltage reversal control

model of TCSC, in Proc. Int. Conf. Power Syst. Technol., Oct. 2428, 2010, pp. 16.

[12]

T.-H. Chen, Criteria to estimate the voltage unbalances due to high-speed railway

demands, IEEE Trans. Power Syst., vol. 9, no. 3, pp. 16721678, Aug. 1994.

[13]

S.-H. Chang, K.-H. Oh, and J.-H. Kim, Analysis of voltage unbalance in the electric

railway system using two-port network model, KIEE Trans., vol. 50A, pp. 248254,
May 2001.

[14]

J.-S. Kim, S.-H. Lim, J.-F. Moon, J.-C. Kim, and O.-B. Hyun, Analysis on the

protective coordination on neutral line of main transformer in power distribution

substation with superconducting fault current limiter, KIEE Trans., vol. 58, no. 11, pp.
20892094, Nov. 2009.

[15] H. R. Kim, S.-W. Yim, O.-B. Hyun, J. Sim, and S.-Y. Oh, Analysis on recovery
characteristics of superconducting fault current limiters, presented at the Proc. Conf.
Magn. Technol., Philadelphia, PA, USA, Aug. 2731, 2007, MT-20.

[16]

S.-H. Lim, S.-R. Lee, H.-S. Choi, and B.-S. Han, Analysis of operational

characteristics of flux-lock type SFCL combined with Power compen-sator, IEEE

Trans. Appl. Supercond., vol. 15, no. 2, pp. 20432046, Jun. 2005.