Capacitor Placement in Radial Distribution System for Improve Network Efficiency Using Artificial Bee Colony

Transcription

1 RESEARCH ARTICLE OPEN ACCESS Capaitor Plaement in Radial Distribution System for Improve Network Effiieny Using Artifiial Bee Colony Mahdi Mozaffari Legha, Marjan Tavakoli, Farzaneh Ostovar 3, Milad Askari Hashemabadi 4 Department of Power Engineering, Jiroft Branh, Islami Azad University, Iran Department of Power Engineering, Jiroft Branh, Islami Azad University, Iran 3 Department of Power Engineering, Shadegan Branh, Islami Azad University, Iran 4 Department of Power Engineering, Islami Azad University of siene and Researh Kerman branh, Iran Abstrat Inreasing appliation of apaitor banks on distribution networks is the diret impat of development of tehnology and the energy disasters that the world is enountering. To obtain these goals the resoures apaity and the installation plae are of a ruial importane. In this paper a new method is proposed to find the optimal and simultaneous plae and apaity of these resoures to redue losses, improve voltage profile. The advantage of ABC algorithm is that it does not require external parameters suh as ross over rate and mutation rate as in ase of geneti algorithm and differential evolution and it is hard to determine these parameters in prior. To demonstrate the validity of the proposed algorithm, omputer simulations are arried out on atual power network of Kerman Provine, Iran and the simulation results are presented and disussed. Keywords: Distribution systems, Loss Sensitivity Fators, Capaitor plaement, Artifiial Bee Colony Algorithm I. Introdution The loss minimization in distribution systems has assumed greater signifiane reently sine the trend towards distribution automation will require the most effiient operating senario for eonomi viability variations. The power losses in distribution systems orrespond to about 70% of total losses in eletri power systems (005). To redue these losses, shunt apaitor banks are installed on distribution primary feeders. The advantages with the addition of shunt apaitors banks are to improve the power fator, feeder voltage profile, Power loss redution and inreases available apaity of feeders. Therefore it is important to find optimal loation and sizes of apaitors in the system to ahieve the above mentioned objetives. Sine, the optimal apaitor plaement is a ompliated ombinatorial optimization problem, many different optimization tehniques and algorithms have been proposed in the past. H. Ng et al (000) proposed the apaitor plaement problem by using fuzzy approximate reasoning. Sundharajan and Pahwa (994) proposed the geneti algorithm approah to determine the optimal plaement of apaitors based on the mehanism of natural seletion. Ji-Pyng Chiou et al (006) proposed the variable sale hybrid differential evolution algorithm for the apaitor plaement in distribution system. Both Grainger et al (98) and Baghzouz and Ertem (990) proposed the onept that the size of apaitor banks was onsidered as a ontinuous variable. Bala et al (995) presented a sensitivity-based method to solve the optimal apaitor plaement problem. In this paper a new method is proposed to find the optimal and simultaneous plae and apaity of these resoures to redue losses, improve voltage profile. The artifiial bee olony algorithm is a new meta heuristi approah, proposed by Karaboga [9]- []. It is inspired by the intelligent foraging behavior of honey bee swarm. The proposed method is tested on atual power network of Kerman Provine, Iran and the simulation results are presented and disussed. II. Objetive funtion The objetive of apaitor plaement in the distribution system is to minimize the annual ost of the system, subjeted to ertain operating onstraints and load pattern. For simpliity, the operation and maintenane ost of the apaitor plaed in the distribution system is not taken into onsideration. The three-phase system is onsidered as balaned and loads are assumed as time invariant. Mathematially, the objetive funtion of the problem is desribed as: Minimize f = Min COST + λδν min () Where COST inludes the ost of power loss and the apaitor plaement, and will be disussed further later. λ is a penalty funtion and (ΔV ) min is the squared sum of the violated voltage onstraint. Moreover, the penalty funtion satisfies the following properties: () If the voltage onstraint is not violated, λ =0; () If the onstraint is violated, a signifiant penalty is imposed to ause the objetive funtion to move away from the undesirable solution. 8 P a g e

2 The voltage magnitude at eah bus must be maintained within its limits and is expressed as: V min V i V max () Where Vi is the voltage magnitude of bus i, V min and V max are bus minimum and maximum voltage limits, respetively. III. Formulation The power flows are omputed by the following set of simplified reursive equations derived from the single-line diagram depited in Fig.. Figure : Single line diagram of main feeder P i+ = P i P Li + R ij + Q i+ = Q i Q Li+ X ij + P i + Q i V i = V i - R ij+ P i +X ij+ Q i + R ij+ +X ij+ V (3) i P i + Q i V (4) i P i +Q i V i (5) Where Pi and Qi are the real and reative powers flowing out of bus i, and P Li and Q Li are the real and reative load powers at bus i. The resistane and reatane of the line setion between buses i and i+ are denoted by R i,i+ and X i,i+ respetively. The power loss of the line setion onneting buses i and i+ may be omputed as P Loss i, i + = R i,i+ P i + Q i V i (6) LOSS The total power loss of the feeder, P T may then be determined by summing up the losses of all line setions of the feeder, whih is given as n P T LOSS = P LOSS i, i + i=0 (7) Considering the pratial apaitors, there exists a finite number of standard sizes whih are integer multiples of the smallest size Q0. Besides, the ost per Kvar varies from one size to another. In general, apaitors of larger size have lower unit pries. The available apaitor size is usually limited to Q max = LQ (8) Therefore, for eah installation loation, there are L apaitor sizes {Q C, Q, 3Q,, LQ} available. Given the annual installation ost for eah ompensated bus, the total ost due to apaitor plaement and power loss hange is written as COST = K p P T LOSS + K f + K i Q i (9) i Where n is number of andidate loations for apaitor plaement, Kp is the equivalent annual ost per unit of power loss in $/(kw-year); K f is the fixed ost for the apaitor plaement. Constant K i is the annual apaitor installation ost, and, i =,,..., n are the indies of the buses seleted for ompensation. The bus reative ompensation power is limited to n Q i Q Li (0) i= Where Q and LQ are the reative power ompensated at bus i and the reative load power at bus i, respetively. IV. Power Flow Analysis Method The methods proposed for solving distribution power flow analysis an be lassified into three ategories: Diret methods, Bakward-Forward sweep methods and Newton-Raphson (NR) methods. The Bakward-Forward Sweep method is an iterative means to solving the load flow equations of radial distribution systems whih has two steps. The Bakward sweep, whih updates urrents using Kirhoff s Current Law (KCL), and the Forward sweep, whih updates voltage using voltage drop alulations []. The Bakward Sweep alulates the urrent injeted into eah branh as a funtion of the end node voltages. It performs a urrent summation while updating voltages. Bus voltages at the end nodes are initialized for the first iteration. Starting at the end buses, eah branh is traversed toward the soure bus updating the voltage and alulating the urrent injeted into eah bus. These alulated urrents are stored and used in the subsequent Forward Sweep alulations. The alulated soure voltage is used for mismath alulation as the termination riteria by omparing it to the speified soure voltage. The Forward Sweep alulates node voltages as a funtion of the urrents injeted into eah bus. The Forward Sweep is a voltage drop alulation with the onstraint that the soure voltage used is the speified nominal voltage at the beginning of eah forward sweep. The voltage is alulated at eah bus, beginning at the soure bus and traversing out to the end buses using the urrents alulated in previous the Bakward Sweep []. Single line diagram of main feeder depited in Fig.. 9 P a g e

3 Input Data The model oeffiients are omputed one information is based on the omparison of food soure positions. When the netar of a food soure is abandoned by the bees, a new food soure is randomly determined by a sout bee and replaed with the abandoned one. Flowhart of the proposed method depited in Fig. 3. Bakward forward Sweep load flow omputation Calulation of real and reative power Calulate the branh urrent of the bus and the No Auray < ΔV Yes Calulate the branh urrent of the bus and the first bus Figure : Single line diagram of main feeder V. Artifiial Bee Colony Algorithm (ABC) Artifiial Bee Colony (ABC) algorithm, proposed by Karaboga for optimizing numerial problems in [6], simulates the intelligent foraging behavior of honey bee swarms. In ABC algorithm, the olony of artifiial bees ontains three groups of bees: employed bees, and unemployed bees: onlookers and souts. In ABC, first half of the olony onsists of employed artifiial bees and the seond half onstitutes the artifiial onlookers. The employed bee whose food soure has been exhausted beomes a sout bee. In ABC algorithm, the position of a food soure represents a possible solution to the optimization problem and the netar amount of a food soure orresponds to the quality (fitness) of the assoiated solution. The number of the employed bees is equal to the number of food soures, eah of whih also represents a site, being exploited at the moment or to the number of solutions in the population. In the ABC algorithm, first half of the olony onsists of employed artifiial bees and the seond half onstitutes the onlookers. For every food soure, there is only one employed bee. In the ABC algorithm, eah yle of the searh onsists of three steps: sending the employed bees onto the food soures and then measuring their netar amounts Hene, the dane of employed bees arrying higher netar reruits the onlookers for the food soure areas with higher netar amount. After arriving at the seleted area, she hooses a new food soure in the neighborhood of the one in the memory depending on visual information. Visual Figure 3: Flowhart of the proposed method VI. Test Results To study the proposed method, atual power network of Kosar feeder of Kerman Provine, Iran is simulated in Cymedist. Figure 3 illustrates the singleline diagram of this network. The base values of the system are taken as 0kV and 0MVA. The system onsists of 0 distribution transformers with various ratings. The details of the distribution transformers are given in table. The details of the distribution ondutors are given in table. The lengths of the feeder segments are given in table 3. The total onneted load on the system is 550 KVA and the peak demand for the year is 0 KVA at a PF of 0.8 lag. The onneted loads on the transformers are listed in table 4. Table : Details of transformers in the system Rating [KVA] Number No load losses [watts] Impedane [%] P a g e

4 Table : Details of ondutors in the system Type R [Ω/km] X [Ω/km] Cmax [A] A [mm] Hyena Dog Mink Table 3: Distribution System Line Data from To Length (meters) Table 4: Details of the onneted loads Transformer no Load [Kva] In addition the total network loss, whih was MW before installing apaitor, has diminished to the 4.55MW whih shows 45.8% derease. Table 5 and 6 depits the Results of power flow before and after installation of apaitor. The simulation results are given in Table 7. These results reveal that the inlusions of apaitor redue the line losses as expeted. It an be shown from the graphs that, LRI dereases marginally, sine the ore losses of the transformers and the LV side losses remain onstant being independent of the presene of v. It an also be seen that with the inrease in the reative power of apaitor, LRI, derease Table 5: Results of power flow before installation of apaitor Bus Number V (pu) Table 6: Results of power flow after installation of apaitor banks Bus Number V (pu) P a g e

5 Voltage M Mozaffari Legha et al Int. Journal of Engineering Researh and Appliations Table 7: Variation of LRI and Optimal plae and apaity of apaitor banks Number plae,, 6 7,3, 5,6,7,3, 5 7,8,9,, 0 5,7,3,5,6,8, 0,4,9,0,4,8, 0 Piked apaity [Mvar] Presumabl e Capaity Range [Mvar] LRI [%] The detailed pu voltages profile of all the nodes of the system after apaitor plaement are shown in the Figure Voltage profile before apaitor plaement Voltage profile after apaitor plaement Bus Number Fig 4: Voltage profile of 0 bus system before and after apaitor plaement VII. Conlusion In the present paper, a new population based artifiial In the present paper, a new population based artifiial bee olony algorithm (ABC) has been proposed to solve apaitor plaement problem and quantifying the total line loss redution in distribution system. Simulations are arried on atual power network of Kerman Provine, Iran. The simulation results show that the inlusion of apaitor, marginally redue the losses in a distribution system. This is beause; the line losses form only a minor part of the distribution system losses and the apaitor an redue only the line losses. The other losses viz. the transformer losses and the LV side distribution losses remain unaltered. Hene this fat should be onsidered before installing a apaitor into a system. The results obtained by the proposed method outperform the other methods in terms of quality of the solution and omputation effiieny. Referenes [] C. Lyra, C. Pissara, C. Cavellui, A. Mendes, P. M. Frana(005), Capaitor plaement in largesized radial distribution networks, replaement and sizing of apaitor banks in distorted distribution networks by geneti algorithms, IEE Proeedings Generation, Transmision & Distribution, pp [] Ng H.N., Salama M.M.A. and Chikhani A.Y(000), Capaitor alloation by approximate reasoning: fuzzy apaitor plaement, IEEE Transations on Power Delivery, vol. 5, No., pp [3] Sundharajan and A. Pahwa(994), Optimal seletion of apaitors for radial distribution systems using geneti algorithm, IEEE Trans. Power Systems, vol. 9, No.3, pp [4] Ji-Pyng Chiou et al(006), Capaitor plaement in large sale distribution system using variable saling hybrid differential evolution, Eletri Power and Energy Systems, vol. 8, pp [5] J. J. Grainger, S. H. Lee (98), Optimum size and loation of shunt apaitors for redution of losses on distribution feeders, IEEE Trans Power Apparatus Systems, vol. 00, pp [6] S. H. Lee, J. J. Grainger (98), Optimum plaement of fixed and swithed apaitors on primary distribution feeders, IEEE Trans PAS, vol. 00,pp [7] Baghzouz. Y and Ertem S(990), Shunt apaitor sizing for radial distribution feeders with distorted substation voltages, IEEE Trans Power Delivery,vol. 5, pp [8] J. L. Bala, P. A. Kuntz, M. Tayor(995), Sensitivity-based optimal apaitor plaement on a radial distribution feeder, Pro. Northon 95, IEEE Tehnial Appliation Conf., pp [9] B. Basturk, D. Karaboga(006), An artifiial bee olony (ABC) algorithm for numeri funtion optimization, IEEE 3 P a g e

6 Swarm Intelligene Symposium 006, May -4, Indianapolis, IN, USA. [0] D. Karaboga, B. Basturk(007), A powerful and effiient algorithm for numerial funtion optimization: artifiial bee olony (ABC) algorithm,journal of Global Optimization, vol. 39, pp [] D. Karaboga, B. Basturk(008), On the performane of artifiial bee olony (ABC) algorithm, Applied Soft Computing, vol. 8 pp [] Baran ME, Wu FF (989), Optimal sizing of apaitors plaed on a radial distribution systems, IEEE Trans Power Deliver, vol. 4, pp [3] Prakash K. and Sydulu M (007), Partile swarm optimization based apaitor plaement on radial distribution systems, IEEE Power Engineering Soiety general meeting 007, pp. -5. [4] S. F. Mekhamer et al (00), New heuristi strategies for reative power ompensation of radial distribution feeders, IEEE Trans Power Delivery, vol. 7, No. 4, pp [5] D. Das(00), Reative power ompensation for radial distribution networks using geneti algorithms, Eletri Power and Energy Systems, vol. 4, pp [6] K. S. Swarup (005), Geneti Algorithm for optimal apaitor alloation in radial distribution systems,proeedings of the 6th WSEAS Int. Conf. on EVOLUTIONARY COMPUTING, Lisbon, Portugal, June 6-8, pp5-59. [7] M.Chis, M. M. A. Salama and S. Jayaram(997), Capaitor Plaement in distribution system using heuristi searh strategies, IEE Pro-Gener, Transm, Distrib, vol, 44, No.3, pp [8] Das et al (995), Simple and effiient method for load flow solution of radial distribution network, Eletri Power and Energy Systems, vol. 7, No.5, pp [9] H.N.Ng, M.M.A. Salama and A.Y.Chikhani(000), Capaitor Alloation by Approximate Reasoning: Fuzzy Capaitor Plaement, IEEE Trans.Power Delivery, vol. 5, no., pp Saveh Branh. He is interested in the stability of power systems and eletrial distribution systems and DSP in power systems. He has presented more than 8 journal papers and 35 onferene papers. Marjan Tavakoli; Trainer of department of Power Engineering, Islami Azad University Jiroft Branh. She is reeived MS degrees of Department of ommuniation engineering, Shahid Bahonar University, Kerman, Iran. She is interested in the ommuniation, DSP, DIP and stability of power ommuniation and power distribution systems. Farzaneh Ostovar; Trainer of department of Power Engineering, Islami Azad University Shadegan Branh. She is reeived MS degrees of Department of Power engineering, Islami Azad University Dezful Branh. She is interested in the stability of power systems and power distribution systems. Milad Askari Hashem Abadi: MS Student of Department of Power Engineering, Islami Azad University of siene and Researh Kerman branh, Iran. He is interested in the stability of power systems and power quality in distribution systems. Mahdi Mozaffari Legha; PhD student of Power Engineering from Shiraz University, He is Trainer of department of Power Engineering, Islami Azad University Jiroft Branh and reeived the M.S. degrees from Islami Azad University 33 P a g e