Optimum Allocation of Distributed Generation using PSO: IEEE Test Case Studies Evaluation

Transcription

1 Optmum Allocaton of Dstrbuted Generaton usng PSO: IEEE Test Case Studes Evaluaton 1 Taha Jabbar Sahb, 1 Mohd Ruddn Ab. Ghan*, Zanarah Jano, 3 Imad Hazm Mohamed 1 Faculty of Electrcal Engneerng (FKE) Center for Language and Human Development, 1, Unverst Teknkal Malaysa Melaka 3 Mddle Techncal Unversty Iraq Emal: dpdruddn@utem.edu.my* *ORCID: /ID: Abstract Dstrbuted Generaton () ntegraton and renewable energy utlzaton n dstrbuton networks have become more attractve as one of the promsng ways n the transton of exstng grd to advance future dstrbuton network. In power system, especally n dstrbuton network, s are able to reduce the total power losses whch have sgnfcant mpact on envronmental polluton. In dstrbuton network, s need to be optmally stuated to avod any unwanted rsks such as voltage rse and losses ncrement. Ths paper amed to perform the optmum allocaton of ntegrated that resulted n the best possble operaton of dstrbuton network. In ths paper, objectve functon was consdered as an nteger and dscrete problem that was formulated wth dfferent objectve functons and constrants. Ths paper proposed Partcle Swarm Optmzaton (PSO) to ascertan the mnmum value of the defned objectve functons. The IEEE 37 bus test system was examned as a test case to demonstrate the effectveness of the proposed approach on the total power losses reducton and voltage profle mprovement. Keywords: Allocaton, Dstrbuton Network Plannng, PSO. INTRODUCTION Dstrbuted Generatons (s) are small generators that are often ntegrated nto the dstrbuton network, provdng the electrcty locally to load customers [1]. mpact on dstrbuton networks s mportant and needs to be well acknowledged to avod any unwanted rsks. An optmal allocaton of s can help to reduce network losses and mprove network performance. Recent studes show that 13% of total generated powers are lost at dstrbuton level [] and can be reduced by nstallng s f optmally placed and szed. Energy losses can also be reduced by a sutable selecton of [3]. The hgh penetraton level of n dstrbuton network can nfluence the power system wth voltage and power losses ncrement. Therefore, voltage mprovement s one of the major ssues that needs to be addressed [4] by allocatng adequate. The technologes adopted n comprse small gas turbnes, mcro-turbnes, fuel cells, wnd and solar energy, bomass, small hydro-power etc. [1]-[6]. In dstrbuton systems, s can provde benefts for the consumers as well as for the utltes, more especally n grds where the central generatons are mpractcable or where there are defcences n the transmsson and dstrbuton system [7]. The man reasons for the ncreasngly wdespread use of can be summed up as follows [8]-[10]. In the process of placement of, several factors need to be consdered, such as the technology to be used, the number and the capacty of the unts, the optmal locaton, and the type of network connecton. However, n order to maxmze benefts, soluton technques for deployment should be obtaned usng optmzaton methods because, nstallng unts at nonoptmal places and n napproprate szes may cause system power losses and cost ncrement. Moreover, nstallng unts s not straght forward, thus, the placement and szng of unts should be carefully selected. Researchers focus on heurstc technques that tend to fnd a good soluton to a complex combnatoral problem wthn a ratonal tme. Sngh et al. [10] asserted that the constant load models have determned the effect of dfferent load and senstvty to voltage and frequency. Authors [11] have used an analytcal method for optmal allocaton. Ths method s based on the equvalent current njecton that uses the bus njecton to branch current (BIBC) and branch current to bus voltage (BCB) matrces whch s developed and broadly acheved for the load flow analyss of the dstrbuton systems whch causes reducton on the search space. In [1], a dynamc programmng applcaton s performed for locaton n terms of loss reducton and voltage mprovement. Many researchers have used evolutonary methods for fndng the optmal placement [13] [15]. In a study [15], an adaptve-weght PSO (APSO) algorthm s used to place multple unts, but the objectve s to mnmze only the real power loss of the system. In [16], unts are placed at the most senstve buses to voltage collapse. The unts have the same capacty and are placed one by one. An analytcal method s utlzed to determne the optmum locaton sze par of a unt n order to mnmze only the lne losses of the power system. In [13], GA based algorthm s presented to locate multple unts to mnmze a cost functon ncludng the system losses and servce nterrupton costs. The optmzaton procedure s formulated usng only voltage profle ndces, and then the effect of ntroducng unts on the lne losses s analysed. In [17] a mult-objectve 900

2 formulaton for the stng and szng of resources nto exstng dstrbuton networks has been proposed. Fgure 1 shows the mportance of Integraton to the power dstrbuton network. populaton loop and evaluated n order to fnd the best ntal random soluton for PSO man loop. The mutaton operator was performed once the PSO terated n locally. The best global result of the algorthm s the optmum soluton wth mnmum power losses and best possble voltage profle. Start Model ntalzaton Iteraton.no., Populaton.no., Mutaton rate, Max number, Max sze power flow equaton solve usng open DSS engne Fnd the mnmum voltage buses PSO ntalzaton and create randomly soluton for locaton and sze subjected to constrant Analyze the obtaned results by solvng the power flow usng open DSS Fgure 1: mportance n future power dstrbuton network Update global best and personal best No Last populaton number METHODOLOGY Partcle swarm optmzaton (PSO) theory has been developed through a smulaton of smplfed socal models. PSO s an optmzaton method establshed by Kennedy and Eberhart n 1995, nspred by socal compartment of brd flockng or fsh schoolng [18]. PSO s consdered as one the recent developments n combnatoral meta-heurstc optmzaton whch s a populaton-based stochastc search algorthm [19]- [0] basng on a smple concept. It works n two steps, calculatng the partcle velocty and updatng ts poston. Therefore, the computaton tme s short, and lttle memory s requred. The process of PSO algorthm n fndng optmal values follows the work of ths anmal socety. PSO conssts of a swarm of partcles, where partcle represents a potental soluton. Recently, there are several modfcatons from the orgnal PSO. In order to accelerate the achevement of the best condtons. The development wll provde new advantages and also resolve the dversty of problems. One of the weaknesses of PSO algorthm s that sometmes optmum solutons wll converge nto the local area of the search space. Ths phenomena s called local mnma. Ths paper has utlzed modfed-pso algorthm that mproves mutaton features on the percentage of populaton of the partcles. The mutaton on populaton may help to avod local mnma phenomena. In other words mutaton can mprove PSO algorthm n terms of preventng to converge locally and fnd an optmum soluton. Fgure shows the PSO algorthm that has been proposed n ths paper for optmum allocaton. The algorthm was ntalzed wth PSO ntal parameters such as maxmum teraton, populaton sze, and mutaton rate. A test case was modelled and power flow calculaton performed usng OpenDSS engne. The random solutons were generated n Mutaton Yes PSO man loop Update poston and velocty Evaluate cost Compare mutated wth non-mutated to choose the best cost Update the global best and personal best No Last number of teraton End Yes Fgure : PSO methodology a. Impacts of s on oltage Drop Wth out mutaton Due to the small rato of X to R n dstrbuton networks and the radal structure of these grds, the mpact of s on dstrbuton network voltage s sgnfcant. By consderng ths ssue, the voltage drop for the network can be wrtten as follows [3], [1], []: ( R jx ) I (1) 1 P jq I () 901

3 ( RP XQ) ( RP XQ) ( RP XQ) (3) where, s the lne voltage drop, R+jX s the lne mpedance, Q s the reactve power, P s the actve power, 1 & and I, are the Bus 1 and voltage ampltude and the current flow through the lne, respectvely. The above equatons should be consdered as one of the constrants of the optmzaton problem. Ths constrant can be consdered as penalty factor nto the objectve functon. The transformer tap postons can be nfluenced by voltage mprovement n dstrbuton networks whch s mportant for voltage regulaton. Once the voltage at secondary termnal of transformer has mproved as close as possble to one per unt, the tap poston steps can be stuated n ntal poston. Ths act can gves more opportunty to voltage regulaton by tap changer and wll ncrease the lfecycle of transformers that would be cost effectve [3]. Hence, the voltage mprovement through allocatng unts nto the dstrbuton network wll cause more flexblty of transformer tap changer to regulate the voltage. c) Mnmzaton of total power losses and voltage profle mprovement wth maxmum numbers and szes of : Mn; OF. Losses... Number Sze Number Sze olaton volaton olaton olaton MAX ( 0,( N Max )) olaton MAX ( 0,( Szes MaxTotal )) Where all of the objectve functons are subject to the followng constrants: (6) (7) (8) b. Objectve Functon Formulaton and Case Study Assessment Ths study has performed three dfferent scenaros assessment wth objectve functons n order to fnd the optmum placement and szng of nto an IEEE 37 buses test network as shown n Fgure 3. Ths test network has been utlzed n ths paper n order to demonstrate the functonalty of the proposed algorthm to fnd the optmum placement and szng of the. The objectve functon and constrants are formulated n three deferent scenaros. placement and szng has sgnfcant mpacts on the total network losses and voltage profle n dstrbuton networks. In ths regards, three types of scenaros have been elaborated n ths paper as follows: a) Mnmzaton of total power losses and voltage profle mprovement wthout any constrants for maxmum sze and numbers: Mn; OF.Losses. (4) volaton b) Mnmzaton of total power losses and voltage profle mprovement wth maxmum number of number and wthout sze constrants: Bus voltage Current feeders mn (9) t max rated I f I f (10) unts are consdered as constant output generaton wth unt power factor and there are a few ntal assumptons of the szes and constrants whch are specfed as follows: Maxmum number of ; Scenaro #1, 37 unts (all nodes) Scenaro #, 9 unts Scenaro #3, 9 unts Maxmum Sze of total unts; Scenaro #1, not lmted Scenaro #, not lmted Scenaro #3, 1500kW Mnmum sze of each unt s Maxmum sze of each unt s 1MW oltage constrant (0.95 node 1.05) Mn; OF. Losses.. Number volaton olaton (5) 90

4 Fgure 3: IEEE 37 node test network RESULTS AND DISCUSSION Table 1 shows the comparson of IEEE 37 nodes standard case were three types of scenaros are desgned for allocaton. All of the scenaros for optmum allocaton of resulted n the total power losses reducton and voltage profle mprovement. The scenaro #1 ntegrated 16 unts wth 700kW total generaton whch resulted n the sgnfcant losses reducton about 50% wth 30.68kW power losses compared to the standard case whch was 60.56kW. In scenaro #, the ntegraton was lmted to 9 unts whch have concluded to 650kW total generaton wth 31.55kW power losses. In other words, scenaro # has chosen the larger sze of unts n order to mnmze the power losses by ntegratng only 9 unts. In scenaro #3, the number of unts and sze of the total generaton were lmted to 9 unts and 1500kW, respectvely. As shown n Table 1, the scenaro #3 had acceptable total power loss reductons 47.79kW compared to the standard case. It should be noted that n all of the scenaros, the voltage profles has been controlled wthn IEEE standard range (0.95< node <1.05) as shown n Fgures 4-7. In addton, the voltage profle has been mproved after the ntegraton. The comparson of total power losses between standard case and all scenaros s shown n Fgure 8. It ndcates the power losses per elements of networks, before and after the ntegraton. Table 1: The comparson table between standard case and three scenaros of s allocaton Standard case After optmum s allocaton Scenaro #1 Scenaro # Scenaro #3 Total Losses[kW] Mn.voltage[p.u.] Max.voltage[p.u.] Numbers of unts No unts Bus701: Bus 705: Bus 713: Bus 703: Bus 706: Bus 7: Bus 708: Bus 731: Bus 736: Bus 741: Bus718: Bus 734: Bus 737: Bus 738: Bus 78: Bus79: Total 16 unts 700 KW 700KW 00KW 100KW 150KW Bus 701: Bus 70: Bus 77: Bus 704 : Bus 706: Bus 7: Bus 73: Bus 735: Bus 738: 9 unts 650 KW 750KW 350KW 50 KW 50 KW 400 KW Bus 74: Bus 733: Bus 731: Bus 735: Bus 740: Bus 718: Bus 734: Bus 738: Bus 78: 9 unts 1500 KW 300 KW 903

5 Fgure 4:oltage drop n dstance before ntegraton Fgure 5:oltage drop n dstance for scenaros #1 Fgure 6:oltage drop n dstance for scenaros # Fgure 7:oltage drop n dstance for scenaros #3 Fgure 8: Power losses comparson for allocaton before and after scenaros 904

6 CONCLUSION In ths paper, three dfferent scenaros for s allocaton have been appled n order to study the mpacts of these unts on network performance. The Partcle Swarm Optmzaton (PSO) has been utlzed n ths paper n order to solve the objectve functons of optmum s allocaton n dstrbuton network. The results show that the total power losses can be mnmzed sgnfcantly by allocatng the optmum sze of n the optmum placement. Moreover, the lmtatons of number and sze have been determned as constrants of the objectve functon and the results show that the constrants could be optmally appled. In addton, the voltage profle of the network has been mproved and power losses sgnfcantly reduced. In short, usng unts n dstrbuton networks s one the proper solutons n future dstrbuton network plannng n order to mprove the network performance provded, unts are allocated adequately. More studes need to be conducted on and ther mpacts on dstrbuton networks. ACKNOWLEEMENTS The authors gratefully acknowledge the fundng support provded by the Unverst Teknkal Malaysa Melaka under the research grant no. PJP/016/FKE/HI5/S REFERENCES [1] Ackermann, T., Andersson, G., and Söder, L., Electrcty market regulatons and ther mpact on dstrbuted generaton, 000, Electr. Utl. Deregul. Restruct. Power Technol Proceedngs. DRPT 000. Int. Conf., no. Aprl, pp ,. [] Azm, S. and Swarup, K. S., 005, Optmal capactor allocaton n radal dstrbuton systems under APDRP, n Internatonal IEEE Inda Conference (INDICON), pp [3] Ab Ghan, M.R., Hasan, I. J., Gan, C. K. and Z. Jano, 015, Losses Reducton and oltage Improvement wth Optmum Allocaton usng GA," MAGNT Research Report,ol. 3, No.6, pp [4] Loa,., accaro, A. and K. asakh, 013, A selforganzng archtecture based on cooperatve fuzzy agents for smart grd voltage control, IEEE Trans. Ind. Informatcs, " ol. 9, No. 3, pp [5] Blaabjerg, F., Teodorescu, R., Lserre, M.,and Tmbus, A.., 006, Overvew of control and grd synchronzaton for dstrbuted power generaton systems, IEEE Trans. Ind. Electron., ol. 53, No. 5, pp [6] Lasseter, R. H., 010, Control of Dstrbuted Resources,. [7] Grffn, T., Tomsovc, K., Secrest, D. and Law, A., 000, Placement of dspersed generaton systems for reduced losses, Proc. 33rd Annu. Hawa Int. Conf. Syst. Sc., ol. 00, No. c, pp [8] Banerjee, R., 006, Comparson of optons for dstrbuted generaton n Inda, Energy Polcy, ol. 34, No. 1, pp [9] Carpnell, G., Cell, G., Plo, F. and Russo, A., 001, Dstrbuted Generaton stng and szng under uncertanty, 001 IEEE Porto Power Tech Proc., vol. 4, pp [10] Sngh, R. K. and Goswam, S. K., 010, Optmum allocaton of dstrbuted generatons based on nodal prcng for proft, loss reducton, and voltage mprovement ncludng voltage rse ssue, Int. J. Electr. Power Energy Syst., ol. 3, No. 6, pp [11] Gözel, T. and Hocaoglu, M. H., 009, An analytcal method for the szng and stng of dstrbuted generators n radal systems, Electr. Power Syst. Res., ol. 79, No. 6, pp [1] Khales, N. and Haghfam, M.-R., 009, Applcaton of dynamc programmng for dstrbuted generaton allocaton, Electrcal Power & Energy Conference (EPEC), 009 IEEE. pp [13] Km, K.-H., Lee, Y.-J., Rhee, S.-B., Lee, S.-K., and You, S.-K., 00, Dspersed generator placement usng fuzzy-ga n dstrbuton systems, n Power Engneerng Socety Summer Meetng, 00 IEEE, ol. 3, pp [14] Nknam, T., Ranjbar, A. M., Shran, A. R., Mozafar, B., and Ostad, A., 005, Optmal operaton of dstrbuton system wth regard to dstrbuted generaton: A comparson of evolutonary methods, n Conference Record - IAS Annual Meetng (IEEE Industry Applcatons Socety), ol. 4, pp [15] Prommee, W. and Ongsakul, W., 008, Optmal mult-dstrbuted generaton placement by adaptve weght partcle swarm, 008 Int. Conf. Control. Autom. Syst. ICCAS 008, pp [16] Hedayat, H., Nabavnak, S. A. and Akbarmajd, A., 006, A new method for placement of unts n dstrbuton networks, 006 IEEE PES Power Syst. Conf. Expo. PSCE Proc., pp [17] Cell, G., Ghan, E., Mocc, S., and Plo, F., May 005, A Multobjectve Evolutonary Algorthm for the Szng and Stng of Dstrbuted Generaton, IEEE Trans. Power Syst., ol. 0, No., pp [18] Eberhart, R. and Kennedy, J., 1995, A new optmzer usng partcle swarm theory, Proc. Sxth Int. Symp. Mcro Mach. Hum. Sc., pp [19] an Den Bergh, F. and Engelbrecht, A. P., 006, 905

7 A study of partcle swarm optmzaton partcle trajectores, Inf. Sc. (Ny)., ol. 176, No. 8, pp [0] Santosa, B., 009, Tutoral Partcle Swarm Optmzaton, pp [1] Yu, T. C., 1996, " Prncples and desgn of low voltage systems". Sngapore: Byte Power Publcatons. [] Glan, S. A. and Ahmad, B., 015, Analyss of oltage Drop n a Low oltage Electrcal System for Statstcal Control Process, MAGNT Research Report, ol. 3, No. 1, pp [3] Clark, S. W., 014, Improvement of step-voltage regulatng transformer effcency through tap changer control modfcaton, n T&D Conference and Exposton, IEEE PES, pp