Optimal Size and Location of DistributedGeneration and KVAR Support in Unbalanced

3-<1> Distribution System using PSO

S. Sudhakar ReddyDept. ofElectrical Engg.,

Jadavpur University,Kolkata,India-700032samasr5@gmail.com

Subrata PaulDept. ofElectrical Engg.,

Jadavpur University,Kolkata,India-700032

S. Halder Nee DeyDept. ofElectrical Engg.,

Jadavpur University,Kolkata, India-700032sunitaju@yahoo.com

77

Abstract-This paper proposes a Particle SwarmOptimization based methodology for finding Optimal sizeand location of Distributed Generation and unbalancedReactive power support for unbalanced three phasedistribution network. The improvement in voltage profileand loss saving are presented. The proposed technique istested on IEEE 37 node radial test feeder which is anactual feeder in California. The program is developedusing MATLAB programming software. The resultsobtained show the effectiveness of the method forunbalanced network.

Keywords: Distributed generation (DG); power losses;Particle Swarm Optimization (PSO); Optimal placement.

I. INTRODUCTION

With the ever-increasing demand of electric powerwithout sufficient transmission and generationenhancement, large scale power system black-outs areoccurring. The optimal utilization of existing networkby incorporating the recent technological advances i.e.DG is a rapid developing and the most cost-effectivemeasure in power industry to enhance loadingcapability.

Distribution system is the final link betweenconsumers and generation via the transmission system;it differs from transmission system and has thefollowing characteristics [1].

• It works in radial or weakly meshed topology.• Distribution lines usually have high R1Xratio.• Significant unbalance may be found in both

line parameters and loads.In literature, distribution load flow analysis have

been reported which exploit the topology (radialstructure) of the distribution system [2], [3].

An intensive research on loss minimization hasbeen reported in the literature. S.Elangovan et alpresented sensitivities of line losses with certain controlvariables i.e. generator voltages, transformer tappositions and VAR sources [4]. B.C.G. Shin developed

978-1-4673-4634-4/12/$31.00 ©2012 IEEE

an innovative tool for online power system operationwith optimal power flow based loss minimization [5].Voltage control method has been proposed for lossminimization by using capacitor banks and/or tapchanging transformer [6]. Network reconfiguration forloss minimization with the help of tie lines and switches[7], [8] has been reported.

Electric power generation in regulated environmentwas motivated to use large generating plants to dropdown the cost of per unit output. In recent years severaldriving forces have contributed to the reversal of thistrend and sparked interest in decentralized powerstations [9]. Furthermore, increased customer awarenessand new trends towards "green revolution" generatingtechnologies, have promoted an interest in cleaner,renewable energy conversion safely installed in thedistribution system. Distributed generation is an electricpower source connected directly to distribution networkor on customer side of the meter [10]. Distributedgenerators are either grid connected or stand-aloneelectric generating units located within the distributionsystem at or near end user. J.H.Teng developed themathematical modeling of three phase DG [11], andDGs have become a vital part of distribution system inmany countries and its importance is continuing toincrease. The DGs are classified as follows [10].

Micro distributed generation: 1 Watt to 5 kWSmall distributed generation: 5 kW to 5 MWMedium distributed generation: 5 MW to 50 MWLarge distributed generation: 50 MW to 300 MWJ.Kennedy et al developed the PSG which is based

on Swarm Intelligence i.e. bird flocking, fish schoolingetc. PSG is an extremely simple algorithm to solvecontinuous nonlinear functions with constraints andhighly dependent on stochastic processes [12], [13].PSG technique has been applied for optimal power flowbased loss minimization with optimal size of reactivepower support [14]. M.A.Kashem et al has presented atechnique for the placement of optimal DG size and

International Conference on Emerging Trends in Electrical Engineering and Energy Management (ICETEEEM-2012)

II. PROBLEM FORMULATION

A Three Phase Loss Function

Bus voltages, branch currents, branch losses andtotal active power losses varies with DG size(Sg=Pg+jQg) can be obtained using distributed loadflow analysis. Here Direct Approach Load flow analysisproposed by J.H.Teng has been adopted [1] byconsidering DG as negative load, and the effects of linecharging shunt admittance also included.

Branch _loss = Ireal {(Branch _ input) - (Branch _ output)} I (2)Where,Branch _input =(¥:a xba* + ¥:b xbb* +¥:c xbc*) (3)

Branch_ output = ("v"a xba* +"v"b xbb*+"v"c xbc*) (4)

where Vsw Vsb and Vsc are sending end voltagesVrw Vrb and Vrc are receiving end voltagesba, b, and beare branch currentsBus voltages and branch currents vary with DG

size hence the total loss may be given and used asfitness function for present study.

location with real and reactive power loss sensitivity[15]. T.Gozel et al developed a MATLAB GraphicalUser Interface package for optimal size and location ofDG and shunt capacitor with help of golden sectionsearch and grid search algorithms [16]. S.R.A.Rahim etal proposed Artificial Immune System optimizationtechnique to find the optimal size of DG and tested it atvarious load conditions [17].

Above discussed optimization techniques areapplied for balanced three phase system with equivalentline modeling, and no attempt has been reported forunbalanced three phase system so far. Whereas thedistribution systems are mostly unbalanced. Forexample, IEEE 37 node radial test feeder whichrepresents the actual feeder located in California [18],[19].

With the concept of PSO available in literature, anattempt has been made in this paper to find optimal sizeand location of reactive power support and/or DGsupport for an existing unbalanced distribution network.In this study, loss saving per unit (KVAR or KVA)support has been considered for finding the bestlocation of KVAR support and/or DG support.

This paper has been organized as follows: SectionII describes problem formulation. Section III presentsPSO application to three phase loss minimization, andSimulation results for IEEE 37 node test feedercompared in voltage profile and losses are described inSection IV. Section V shows IEEE 37 node test feedernetwork, and in Section VI concluding remarks arepresented.

(5)

(10)

total_loss =f(~,Qg)

The method is based on research on swarmssuch as fish schooling and bird flocking.

2. It is based on a simple concept. It works in twosteps, which are calculating the particlevelocity and updating its position. Therefore,the computation time is short, and it requireslittle memory.

PSO is basically developed through simulation ofbirds flocking in multi dimensional space. However, inthis a DG operated in PQ mode is considered as anegative load and represents a point in a twodimensional plane P and Q to indicate swarm particle oragent with the velocity vp along P axis and the velocityvq along the Q axis. Modification of the agent positionis realized by the position and velocity information.

PSO function for three phase total power lossminimization can be written from (5) as

min(total_loss) = min (f(~,Qg)) (9)

Optimal size of the DG can obtainedminimizing the total loss function (9) using PSO. Theposition of each particle (DG size) is determined by thevector Pgi, Qgi initial random within the specifiedrange and its movement by the velocity of the particleas given by.

-- --+ --

~i(t+ 1)=Pgi(t) + VPi(t)

C Basic Concepts

III. PSO ApPLICATION FOR Loss MINIMIZATION

B Constraints

The constraints reflect the limits on physicaldevices in the power system as well as the limits createdto ensure system security. They are given by

~min ~ Pg ~ ~max (6)

z., ~ Qg ~ z., (7)

I~I~ ~max (8)Where, Pgmin and Pgmax are the mmimum and

maximum limits ofDG active power respectively. Qgminand Qgmax are the minimum and maximum limits of DGreactive power respectively. Pijmax maximum powertransfer limit of line between the buses i.j.

PSO is an evolutionary optimization methodmodeled by the social behavior of swarm intelligenceconcept. Swarm intelligence is property of a systemwhereby the collective behaviors of unsophisticatedagents interact locally with their environment to createcoherent global functional patterns. The features ofproposed optimization technique are as follows [12],[13].

1.

(1)total _loss = L Branch _losses

978-1-4673-4634-4/12/$31.00 ©2012 IEEE 78

Optimal Size and Location of Distributed Generation and KVAR Support in Unbalanced 3-<P Distribution System using PSO

79

Qgi(t + 1) = Qgi(t) + vqi(t) (11)The information available for each individual is

based on self best experience and the knowledge ofother individual best performance. These relativeimportances of the two factors can vary from onedecision to another, it is reasonable to apply randomweights to each part, and therefore the velocity updatesare given as [13].

vp;(t+1)=vp;(t)+Mj -R; .(PBi -:Pgi(t))+ ...

+M2 ·1S '(PG-~;(t)) (12)

vq;(t+ 1) =vq;(t) +u, .R,.(QB;-QgJt))+ ...+M2 ·R4 •(QG-Qgi(t)) (13)

Where, vp, vq are active and reactive powervelocity vectors with initial random values.

M, & M 2 are local and global inertia respectively,Rj , R2 are random numbers.

PB & QB are individual best positions of activereactive generations.

PG & QG are the global best positions of activeand reactive generations.

According to the above formulation, the followingsteps can be adopted for implementing the PSGalgorithm.

D Algorithm

Step1: Evaluate fitness function (5) without DGand store the loss and voltage profile.

Step2: Initialize the swarm positions (DG sizes)by assigning random values within theDG specified limits, and the randomvelocities to each swarm position, and setthese swarm positions values as local aswell as global best fitness values.

Step2: Evaluate the fitness function i.e. total lossfor each and every particle.

Step3: For each individual particle, compare theparticle's fitness value with its previousbest fitness. If the current value is betterthan previous best fitness value, then setthis value as the local best & currentparticle's position as local best position.

Step4: Identify best of current local best fitnessvalues, if it is better than global bestfitness then position particle set as globalbest position.

Step5: Update the velocities and positions of allthe particles using (12), (13) and (10),(11). Check with constraints all particleposition (6), (7), (8).

step6: Repeat steps 2-5 until a stopping criterionis met i.e. maximum number of iterationsor a sufficiently good fitness value.

978-1-4673-4634-4/12/$31.00 ©2012 IEEE

Step7: Step 1-6 repeated by placing DG from 2nd

bus to last bus and the global best valuesand position are tabulated. From thetabulation, best bus position for placing ofDG can be identified.

Similarly the above method is applied for balancedreactive support, unbalanced reactive support andbalanced DG with unbalanced reactive support byconsidering 1, 3 and 4 dimensions respectively.

IV. SIMULATION RESULTS

A computer program has been developed in theMATLAB environment to perform the above discussedthree phase loss minimization using PSG techniqueincluding Direct Approach Load Flow Analysis ofJ.H.Teng [1]. The proposed method has been tested onthe standard IEEE-37 node test feeder and the modifiednumbering of feeder is given in Fig.5 [17]. The testsystem has a base load of 2500 KVA and 230kVunbalanced three phase test feeder. In this study, it isconsidered that the DG is operated at an unspecifiedpower factor.

The IEEE-37 node distribution test feeder has beenconsidered for the present study taking the first bus asthe feeder of electric power from thegeneration/transmission network. The remaining busesof the network are load buses. For the placement ofoptimal balanced reactive support, each and every loadbus is considered taking one at a time. The optimalbalanced KVAR size, loss with KVAR support, losssaving with KVAR support and loss saving per unitKVAR support for each are tabulated and given inTABLE-I, and shows that the bus number 30 have bestloss saving per unit KVAR support. Fig.l comparesimprovement in voltage profile by placing the optimalsize of balanced KVAR support at bus number 30which is best loss saving per unit KVAR support.

TABLE-II illustrates optimal unbalanced KVARsupport, loss with KVAR support, loss saving withKVAR support and loss saving per unit KVAR support,and shows that the bus number 35 has best loss savingper unit KVAR support. Fig.2 compares improvementin voltage profile by placing the optimal size ofunbalanced KVAR support at bus number 35 which isbest loss saving per unit support.

Table-III exhibits optimal size of balanced DG(balanced active and reactive power) support, loss withDG support, loss saving with DG support and losssaving per unit DG support (KVA), and shows that thebus number 36 has best loss saving per unit KVAsupport. Fig.3 compares improvement in voltage profileby placing the optimal size of balanced DG at busnumber 36 which is the best loss saving per unit KVAsupport.

International Conference on Emerging Trends in Electrical Engineering and Energy Management (ICETEEEM-2012)

Fig. 2: Bus Voltages with and without KVAR Support(Unbalanced 3-<1»

TABLE I: OPTIMAL REDUCTION IN LOSSES BY BALANCED 3-<1> KVARSUPPORT AT DIFFERENT BUSES

Fig. 1: Bus Voltages with and without KVAR Support (Balanced 3-<1»

TABLE-IV illustrates optimal DG with unbalancedKVAR support, loss saving with DG support and losssaving per unit DG (KVA) support, and shows that thebus number 36 has best loss saving per unit KVAsupport. Fig.4.1, Fig. 4.2 compares improvement involtage profile by placing the optimal size of DG withunbalanced KVAR support at bus number 36 which isthe best loss saving per unit KVA support.

978-1-4673-4634-4/12/$31.00 ©2012 IEEE 80

KVAR Reactive Loss with Loss LossPlaced Support KVAR Saving Savingat Bus Qa=Qb=Qc Support (KW) with Per unit

no. KVAR KVARSupport Support

(KW)1 No support 60.575 .. ..2 232.014 57.127 3.447 0.004953 202.780 55.939 4.636 0.007624 116.955 57.668 2.907 0.008285 92.399 58.247 2.328 0.008406 87.703 58.217 2.358 0.008967 159.361 56.489 4.086 0.008558 126.825 56.862 3.713 0.009769 116.711 57.076 3.498 0.0099910 78.368 58.129 2.446 0.0104011 99.710 57.199 3.375 0.0112812 71.738 57.617 2.958 0.0137413 55.261 58.186 2.389 0.0144114 69.279 57.642 2.933 0.0141115 81.581 57.767 2.808 0.0114716 70.767 58.102 2.473 0.0116517 158.047 56.061 4.514 0.0095218 123.993 56.861 3.714 0.0099919 109.612 57.214 3.361 0.0102220 87.578 57.830 2.744 0.0104521 94.794 57.603 2.972 0.0104522 129.165 56.046 4.529 0.01169

Table I (Contd.) ..

Optimal Size and Location of Distributed Generation and KVAR Support in Unbalanced 3-<P Distribution System using PSO

...Table I Contd.)23 123.130 56.048 4.527 0.0122524 98.808 56.843 3.732 0.0125925 114.795 56.199 4.376 0.0127126 92.542 57.020 3.555 0.0128027 106.548 56.252 4.323 0.0135328 96.019 56.400 4.175 0.0144929 76.766 57.103 3.472 0.0150830 51.598 58.110 2.465 0.0159231 70.954 57.343 3.232 0.0151832 84.361 56.688 3.887 0.0153633 78.367 56.905 3.670 0.0156134 72.388 57.149 3.426 0.0157835 67.423 57.362 3.213 0.0158936 66.733 57.396 3.179 0.0158837 58.260 58.496 2.079 0.01189

Fig. 3. Bus Voltages with and withoutDG Support (Balanced 3-<1»

V. TEST SYSTEM

The proposed technique was tested on IEEE 37node test feeder which is located in California [16] andthe program was developed using MATLAB

978-1-4673-4634-4/12/$31.00 ©2012 IEEE 81

programming software. The schematic diagram for thetest system with modified bus numbering is shown inFig. 5.

TABLE II: OPTIMAL REDUCTION IN LOSSES BY UNBALANCED 3-<1>KVAR SUPPORT AT DIFFERENT BUSES

KVAR Optimal Size of Reactive Loss Loss LossSupport Support with Saving SavingPlaced Qa Qb o. KVAR with Perat Bus (KVAR) Suppor KVAR unit

No. t(KW) Suppo KVARrt Suppo

(KW) rt1 No Support 60.575 .. ..2 607.25 391.26 261.95 54.535 6.040 0.00483 505.29 353.75 248.76 52.704 7.871 0.00714 288.29 183.87 150.27 55.841 4.734 0.00765 232.83 142.91 117.70 56.703 3.872 0.00786 201.25 127.21 118.00 57.003 3.571 0.00807 379.51 264.93 213.42 54.026 6.549 0.00768 282.39 201.43 182.99 55.076 5.499 0.00829 254.12 183.54 168.31 55.460 5.115 0.008410 164.96 126.96 110.91 57.039 3.536 0.008811 210.58 142.54 153.75 55.943 4.632 0.009112 134.27 84.13 119.48 57.021 3.554 0.010513 96.79 60.61 93.92 57.803 2.772 0.011014 125.50 78.95 116.55 57.108 3.467 0.010815 170.80 115.30 127.52 56.762 3.813 0.009216 146.83 97.60 110.94 57.266 3.309 0.009317 399.62 303.77 188.74 52.409 8.166 0.009218 310.21 237.60 147.72 53.917 6.658 0.009619 270.58 212.05 133.74 54.585 5.990 0.009720 212.59 167.15 106.35 55.737 4.837 0.010021 232.49 181.15 115.92 55.341 5.234 0.009922 316.33 247.80 158.72 52.384 8.191 0.011323 299.41 238.68 152.41 52.395 8.180 0.011824 234.07 185.59 128.17 54.092 6.483 0.011825 284.08 235.83 139.96 52.413 8.162 0.012426 231.15 187.59 110.78 53.925 6.650 0.012627 261.01 221.34 130.09 52.439 8.136 0.013328 236.46 206.78 115.66 52.529 8.046 0.014429 190.23 159.60 91.06 53.987 6.588 0.014930 121.86 98.51 63.57 56.266 4.309 0.015231 177.43 146.74 83.23 54.404 6.171 0.015132 206.09 186.81 101.70 52.918 7.657 0.015533 191.78 176.93 93.39 53.260 7.315 0.015834 179.13 162.56 86.04 53.690 6.885 0.016135 168.16 150.71 79.04 54.087 6.487 0.016336 166.09 147.58 80.25 54.174 6.401 0.016237 148.51 109.11 59.30 56.845 3.730 0.0118

Fig. 4.1: Bus Voltage with and without DG & KVAR Support(Unbalanced three Phase) Vab

International Conference on Emerging Trends in Electrical Engineering and Energy Management (ICETEEEM-2012)

28 402.82 101.30 24.360 36.812 0.029529 321.06 81.20 30.706 30.466 0.030730 199.90 52.31 40.257 20.915 0.033731 297.50 74.99 32.674 28.498 0.031032 349.36 86.75 26.208 34.964 0.032433 335.02 82.68 27.832 33.340 0.032234 310.99 76.49 29.821 31.351 0.032635 290.16 71.21 31.657 29.515 0.032936 280.37 68.89 31.921 29.251 0.033837 229.05 64.85 43.687 17.484 0.0245

TABLE IV: OPTIMAL REDUCTION IN LOSSES BY BALANCED 3-<1> DGIN CO-ORDINATION WITH UNBALANCED KVAR SUPPORT AT

DIFFERENT BUSES

Fig. 4.2: Bus Voltages with and without DG & KVAR Support(Unbalanced three Phase) Vbc, Vca

TABLE III: OPTIMAL REDUCTION IN LOSSES BY BALANCED 3-<1> DGSUPPORT AT DIFFERENT BUSES

DG Optimal DG Size Loss Loss LossPlaced Pa=Pb=Pc Qa=Qb=Qc with Saving Savingat Bus DG withDG Per unit

No. (KW) (KVAR) (KW) (KW) KVA1 No Support 60.575 .. ..2 873.89 247.93 34.732 26.439 0.00973 776.66 216.78 25.373 35.799 0.01484 442.36 123.57 39.050 22.121 0.01615 351.58 97.27 43.171 18.001 0.01646 310.89 89.83 43.953 17.219 0.01777 605.19 169.25 30.363 30.809 0.01638 472.84 134.34 34.260 26.912 0.01829 424.29 121.44 35.994 25.178 0.019010 279.79 80.12 43.304 17.868 0.020511 360.23 105.34 38.106 23.066 0.020512 241.42 75.72 43.100 18.072 0.023813 172.69 55.88 46.967 14.204 0.026114 224.65 71.46 43.570 17.602 0.024915 294.53 86.35 42.000 19.172 0.020816 253.41 74.78 44.418 16.753 0.021117 636.43 170.44 23.552 37.620 0.019018 497.85 132.48 30.271 30.901 0.020019 441.86 116.92 33.098 28.074 0.020520 344.92 91.61 38.262 22.910 0.021421 380.45 100.80 36.465 24.707 0.020922 501.60 132.61 23.275 37.897 0.024323 478.59 125.88 23.202 37.970 0.025624 382.59 101.38 30.178 30.994 0.026125 474.10 121.92 23.503 37.669 0.025726 383.32 98.22 30.350 30.822 0.026027 431.84 110.28 23.686 37.486 0.0280

978-1-4673-4634-4/12/$31.00 ©2012 IEEE 82

DG Optimal Size of DG & Reactive Loss LossPlaced SUPPORT Saving Savingat Bus Pa=Pb o. Qb o. with per

No. =Pc DG& Unit(KW) (KVAR) KVAR KVA

Suppor Supppot(KW) rt

1 No support .. ..2 855.23 526.74 384.67 339.17 27.318 0.0103 762.94 448.13 349.68 308.53 37.160 0.0154 437.75 274.19 184.81 173.00 22.918 0.0165 348.55 225.29 144.19 133.36 18.661 0.0166 308.02 187.68 127.53 129.86 17.581 0.0177 597.55 352.01 263.85 249.47 31.915 0.0168 468.82 270.32 201.95 206.72 27.660 0.0189 420.61 239.08 183.77 186.91 25.813 0.01810 277.86 155.26 127.01 120.18 18.176 0.02011 358.22 206.62 143.41 169.65 23.480 0.02012 240.86 135.07 85.35 129.63 17.987 0.02213 172.10 92.48 62.34 97.10 13.929 0.02414 224.03 122.69 80.53 123.70 17.455 0.02315 293.21 169.54 116.31 140.26 19.414 0.02016 252.46 146.67 98.64 121.48 16.872 0.02017 625.79 359.31 303.36 236.27 39.384 0.01918 491.28 288.26 239.00 178.14 32.433 0.02019 436.73 255.01 213.92 158.53 29.462 0.02020 341.27 199.43 167.32 122.80 23.975 0.02121 376.71 222.14 183.38 135.30 25.893 0.02122 493.40 280.15 242.67 185.61 39.942 0.02423 471.25 266.38 233.51 176.35 40.075 0.02624 377.88 213.07 183.54 145.67 32.553 0.02625 469.27 269.97 239.69 163.56 40.067 0.02626 380.27 223.94 192.15 128.07 32.808 0.02627 426.89 242.06 220.42 149.21 39.940 0.02828 399.77 228.39 211.98 132.40 39.482 0.03029 319.35 187.38 165.30 103.26 32.625 0.03130 198.48 114.70 97.77 70.07 21.971 0.03331 296.06 175.69 152.39 94.14 30.537 0.03132 346.19 193.91 186.39 115.03 37.579 0.03333 333.22 187.84 182.98 106.06 35.971 0.03234 309.59 176.66 168.81 97.50 33.865 0.03335 289.02 166.65 156.94 89.40 31.899 0.03336 278.26 158.81 148.58 90.67 31.510 0.03437 229.02 152.40 117.78 73.01 18.443 0.024

VI. CONCLUSION

This paper presents a method for finding optimalsize and location of DG and reactive powercompensation with the help PSG technique. In thispresent study, the aim was to optimize the size of DGand KVAR allocation for loss reduction in existingunbalanced distribution network. Minimizing the losses

Optimal Size and Location of Distributed Generation and KVAR Support in Unbalanced 3-<D Distribution System using PSO

BIOGRAPHIES

Sudhakar Reddy Sarna obtained his B.E fromOsmania University, M.Tech Degree in Electrical andElectronics Engineering at JNTU, Hyaderabad in 1995 and2004 respectively. He is working as Senior Lecturer, Govt.Polytechnic, Adoni, Andhra Pradesh. He is currently aresearch scholar at Jadavpur University, Kolkata, WestBengal, India. His current research interest includesunbalanced Distribution System and DistributedGeneration applications.

Dr. Sunita Halder Nee Dey received B.E fromJalpaiguri Govt. College, NBU, M.E from Bengal Engg.College (DU) and Ph.D from BESU, Kolkata in ElectricalEngg in 1998, 2001, and 2006 respectively. She is SeniorLecturer in Dept. of Electrical Engg., Jadavpur University,Kolkata, West Bengal, India. She has published severalresearch papers in international and national journals andconferences. Her research interests are mainly in PowerSystem Operation and Control, OPF, Voltage Stability,FACTS applications.

Dr. Subrata Paul is Professor with the Department ofElectrical Engineering at Jadavpur University, Kolkata,West Bengal, India. His work has been primarilyconcerned with Power System Operation and Controls,Robust Controls and FACTS applications

[8] M.A.Kashem, V.Ganapathy, G.B. Jasmon, M.I.Buhari, "ANovel Method for Loss Minimization in Distribution Networks,"IEEE, DRPT-2000, pp-251-256

[9] A.C.Neto, M.G.D.Silva, A.B.Rodrigues, "Impact of DistributedGeneration on Reliability Evaluation of Radial DistributionSystems under Network Constraints," International Conferenceon Probabilistic Methods Applied to Power Systems KTH,Stockholm, Sweden - Jun. 2006.

[10] T.Ackermann, G.Andersson, L.Soder, "Distributed generation: adefinition," Electric Power Systems Research, vol. 79, noA Dec.2000,pp.195-2000.

[11] J.H.Teng, "Modeling distributed generations in three-phasedistribution load flow," lET Generation Transmission &Distribution, Apr. 2007, vol. 2, no. 3, pp. 330-340.

[12] J.Kennedy, R.Heberhart, "Particle Swarm Optimization," IEEEInternational conference, NN, volA, 1995, pp.1942-1948.

[13] T.Niknam, M.R.Narimani, J.Aghaei, R.A.Abarghooee,"Improved particle swarm optimization for multi-objectiveoptimal power flow considering the cost, loss, emission andvoltage stability index," lET Generation, Transmission &Distribution, sep.2011, Vol. 6, No.6, pp.515-527.

[14] A.A.A.Esmin, G.Lambert-Torres, "Loss Power MinimizationUsing Particle Swarm Optimization," IEEE trans. vol.27, no.2,2006, pp-1988-1994.

[15] M.A.Kashem, A.D.T.Le, M.Negnevitsky, G.Ledwich,"Distributed Generation for Minimization of Power Losses inDistribution Systems," IEEE PES general meeting, 2006, pp-1-8.

[16] T.Gozel, U.Eminoglu, M.H.Hocaoglu, "A tool for voltagestability and optimization (VS&OP) in radial distributionsystems using matlab graphical user interface (GUI)," ElectricPower Systems Research, vol.19, no.5, 2008, pp.505-518.

[17] S.R.A.Rahim et al, "Implementation of DG for LossMinimization and Voltage Profile in Distribution System," IEEE,2010, pp-490-494.

[18] W.H.Kersting, "Radial Distribution Test Feeders," IEEEDistribution System Analysis Subcommittee Report, PESsummer meeting 2000.

[19] M.L.Baughman et al, "IEEE-37 Test feeder," DistributionSystem Analysis Subcommittee, IEEE Power EngineeringSociety, 2004.

13724

72516

11

15 706

-------70712

a-------4--.=-------il--------l-------1IIt--------720

799

722712 2 145 701

742 4 713 87046

705 3 702 79 714

729 744 727 703 1020 718

19 18 1721

728

732 708 2426 731

25~7736

30 27 733 77529

710 28 734 35740

31 32 33 34735 36

737 738 711 741

Fig. 5: Single Line Diagram of three Phase Unbalanced IEEE 37Node Test Feeder

REFERENCES

in the system would bring two types of advantages, oneenhances the loading capacity of existing networkwithin the thermal limits and the other one is energysaving. The proposed method successfully tested usingMATLAB program for IEEE 37 node radial test feeder.The simulation results clearly demonstrate that voltageprofile has improved considerably and losses getreduced.

[1] J.H.Teng, "A Direct Approach for Distribution System LoadFlow Solutions," IEEE Trans. on Power Delivery, vol. 18, no. 3,Jul 2003, pp-882-887.

[2] A.G.Bhutad, S.V.Kulkarni, S.A.Khaparde, "Three-phase LoadFlow Methods for Radial Distribution Networks," IEEETENCON-2003, vol. 2, pp.781-785, 2003.

[3] W.H.Kersting, Distribution System Modeling and Analysis,CRC Press, 2001.

[4] S.Elangovan, Sen.Mem, "New approach for real power lossminimization," IEEE proc. on Generation, Transmission andDistribution, vol. 130, no. 6, Nov 1983, pp-295-299.

[5] B.C.G.Shin, F.Lauderdale, "Development of the lossminimization function for real time Power System Operation: Anew tool," IEEE Transactions on Power Systems, vol. 9, no. 4.Nov 1994, pp-2028-2034.

[6] D.Lukman, T.R.Blackburn, "Loss Minimization in Load FlowSimulation in Power System," IEEE International Conference,New South Wales Univ, vol.1, 2001, pp-84-88.

[7] J.C.Wang, H.D.Chiang, G.R.Darling, "An Efficient Algorithmfor Real-Time Network Reconfiguration in Large ScaleUnbalanced Distribution Systems," IEEE Transactions onPower Systems, vol. 11, no. 1, Feb 1996, pp-511-517.

978-1-4673-4634-4/12/$31.00 ©2012 IEEE 83