INTERNATIONAL JOURNAL OF SCIENTIFIC & ENGINEERING RESEARCH, VOLUME 5, ISSUE 2, FEBRUARY-2014 ISSN

Transcription

1 ITERATIOAL JOURAL OF SCIETIFIC & EGIEERIG RESEARCH, VOLUME 5, ISSUE 2, FEBRUARY Optmal placement for power loss reducton and mprovement voltage profle usng smart methods Mlad Askar Hashemabad, Dr. Mahd Mozaffar Legha, Alreza Khajoee Ravar Abstract- Dstrbuted Generatons (s) are utlzed to supply the actve and reactve power n the transmsson and dstrbuton systems. These types of power sources have many benefts such as power qualty enhancement, voltage devaton reducton, power loss reducton, load sheddng reducton, relablty mprovement, etc. In order to reach the above benefts, the optmal placement and szng of s sgnfcant. In ths regard, ths paper gets use of the Bactera Foragng Algorthm () and Bnary Genetc Algorthm (B) to nvestgate the placement wth the purpose of power loss and voltage devaton reducton. The proposed method s appled on the 33-bus and 69bus IEEE test systems and the optmal place and sze of s from the power losses and voltage devaton mnmzaton are assessed. Also, the performance of the above two algorthms are compared wth each other. Key Words- Bactera Foragng Algorthm () and Bnary Genetc Algorthm (B), Dstrbuted Generaton (), Voltage Devaton, Dstrbuton Systems.. Introducton In the frst days of electrcal networks, the power generaton and transformaton was done locally. Wth the network growth, as the result of several economcal preferences, the local generatons were replaced by the dstrbuton generatons. owadays, accordng to the technology developments as well as changes n the energy polcy especally from the envronmental pont of vew, Dstrbuted Generatons (s) have retaned ther valuable role n the power engneerng. In fact, t s expected that a large porton of the electrcal power would be suppled by s n the near future []. Untl now, many dfferent defntons have been assgned to s. In an overall defnton neglectng the technology and capacty constrants, a s a power generaton source whch s connected to the dstrbuton system or load sde drectly [2]. In ths regard, dfferent power sources such as Wnd Turbnes, Photovoltacs, Fuel Cells, Mcro Turbnes, etc can be utlzed as. The energy of these power sources can be from the fossl fuel, renewable energy or storage devces whch can vary from kw to several 0 MW. Some of the man benefts of s can be named as power generaton cost reducton, power losses reducton, voltage devaton mprovement, power qualty enhancement, the possblty of producng both electrcal and thermal energy n the form of combned heat and power (CHP), reducng Total Harmonc Dstorton (THD), ncreasng the effcency, postve effect on the envronment, etc. The necessary factor to gan the above benefts s the optmal placement and szng of s n the network. It s worth to note that msplacement of s can result to losng the above factors when resultng to nstablty of the system. In ths paper, the two evolutonary algorthms of and B are used to optmze the power losses and voltage devaton objectve functons. The near optmal solutons obtaned by both algorthms show the sutable and accurate mplementaton of these algorthms. In recent years, placement has been done wth dfferent objectve functons. the LMP and MIP technques are used to optmal placement and szng of s [3]. a specfc number of s are used to reduce the actve power losses [4]. Dstrb-uted generaton s obvously a new attracton for power ndustry, commercal and regulatng systems [5]. the Genetc Algorthm () s employed to handle the optmal placemen and szng under dfferent loadng for power loss reducton [6]. The suggested approach s proposed to reach the maxmum balance between the cost and benefts of usng n the network [7]. The man purpose s to fnd the optmal place, sze and type of s n the network [7]. The sze and power factor for were found by usng analytcal expressons to mnmze losses [8]. A mult-objectve performance ndex-based determnaton of sze and locaton of n dstrbuton systems wth dfferent load models was presented by Sngh et al. [9] and mplemented usng. A methodology based on was presented by Sngh and Goswam to accommodate n dstrbuton network by maxmzaton of proft, reducton of losses and mprovement n voltage regulaton [0]. Dasan and Dev used fuzzy adaptaton of evolutonary programmng to fnd sze of s [7]. Mlad Askar Hashemabad, Department of Power Engneerng, Islamc Azad Unversty of scence and Research, Kerman, Iran. (Emal: Mladaskar65@yahoo.com). Mahd Mozaffar Legha, Department of Power Engneerng, Islamc Azad Unversty, Kerman Branch, Iran. (E-mal: Mahd_mozaffar@ymal.com). Alreza Khajoee Ravar, Dstrbuton Electrc Power Company n Kerman South, Kerman, Iran. 2. Types of Dstrbuted Generatons Frst Type: these types of s can just generate and supply actve power. Photovoltacs (PVs) are among the most famous of these types. The output power of PVs wll 204

2 98 ITERATIOAL JOURAL OF SCIETIFIC & EGIEERIG RESEARCH, VOLUME 5, ISSUE 2, FEBRUARY-204 enter the storage batteres drectly and so t can be stored as actve power. Second Type: these types of s can just generate reactve power. Synchronous condensers are from ths category. Synchronous condensers are synchronous machnes whch connect to the network under no load condton to enhance the network stuaton. The amount of reactve power produced s determned by the control of the exctng system. Thrd Type: these types of s can generate both actve and reactve powers. Wnd Turbnes (WTs) belong to ths category whch gets use of nducton motors to produce electrcal power. Here reactve power s utlzed to produce actve power. Forth Type: these types of s can adjust the bus voltage. In ths type, the power unt wll generate or consume reactve power to regulate the bus voltage. The actve and reactve power sources used n ths paper are as follows: 0 MW 0 MVAR P MW Q 3 actve power MVAR reactve power 3. Problem Formulaton (Power losses, voltage devaton, objectve functon) Reducng the actve power losses s a sgnfcant factor n the proper performance of the power system. The power losses n a system can be evaluated as follows (): Aj(P Pj Q Q ) Bj (Q Pj P Q ) j j j The objectve functon s as follows (2): FtnessFun cton Mnmze ( w. w 2. ( P P D P mn V max V V max I j I j Where V s the voltage of the th bus and Ij s the current flowng between the th and jth buses. 4. Test systems Ths paper uses the 33-bus and 69-bus test systems as the case studes. The total actve and reactve loads for 33bus system are MW and MVar, respectvely. Also, the ntal actve and reactve power losses neglectng s are 8 kw and 2 kvar, respectvely. In the case of 69bus test system, the amounts of actve and reactve power loads are 3.8 MW and MVAR respectvely. The total actve and reactve power losses neglectng s are 225 kw and 02 kvar respectvely. Ths paper uses the thrd type whch was mentoned above. PL ( V ) ( V ) P P L L, ormal ) ormal The frst term of above equaton s the objectve functon whch s calculated as follows (3): P L Loss k k The second term of the objectve functon s the voltage devaton. The coeffcents w and w2 are found expermentally to balance the objectve functon whch are and.8 for the 33-bus and and 26.8 for the 69-bus system [5]-[8]. In Eq., P & Pj are the actve power of the th and jth buses; Q and Qj are the reactve power of the th and jth buses, respectvely. The power and voltage constrants are as follows (4,5,6): 5. The proposed algorthms 5.. Bactra foragng algorthm atural selecton tends to elmnate anmals wth poor foragng strateges and favor the propagaton of genes of those anmals that have successful foragng strateges. The Eschercha col (E. co bactera that are present n our ntestnes, also undergo ths foragng strateges. The socal foragng behavor of E. col bactera has been used to solve optmzaton problems. The optmzaton n comprses the followng process: chemotaxs, swarmng, reproducton, elmnaton and dspersal The chemotaxs s the actvty that bactera gatherng to nutrent-rch area naturally. The characterstc of E. col bactera s: the dameter s µm, the length s 2µm, under approprate condtons can reproduce (splt) n 20 mn. The move of the E. col s done wth flagellum CHemotaxs step The process n the control system s acheved through Swmmng and tumblng va flagellum. To represent a tumble, a unt length random drecton, Say ( j ), ths wll be used to defne the drecton of movement after a tumble, then: ( j, k, l ) ( k, l ) c() ( j ) Where ( j, k, l ) the poston of the th bacterum at jth chemo s taxs step, kth reproducton step and lth elmnaton and dspersal step

3 99 c () s the sze of step taken n the random drecton that specfed by the tumble(run length unt). If the new poston of bacterum ( j, s better than the old poston, then the bacterum wll keep takng successve step n that drecton. The number of teraton chemotaxs step s C. And the maxmum number of permssble successve steps s s Swarmng step The bactera n tmes of stresses release attractants to sgnal bactera to swarm together. It however also releases a repellant to sgnal others to be at a mnmum dstance from t. Thus all of them wll have a cell to cell attracton va attractant and cell to cell repulson va repellant. The mathematcal representaton for swarmng can be represented by: ( ) = ( ) = Where: dattract: depth of the attractant wattract: measure of the wdth of the attractant hrepellant: heght of the repellant effect wrepellant: measure of the wdth of the repellant p: number of parameters to be optmzed S : total number of bactera Jcc: the cost functon to be added to the actual cost functon to be mnmzed, to present a tme varyng cost functon Reproducton After C chemotaxs step, a reproductve step s occurs. The ftness of bactera s calculated, that s, durng all chemotaxs steps: c J J health (, Then ths ftness s sorted n ascendng order. The least healthy bactera de and the other bactera, each bacterum splt nto two bactera, thus the sze of the populaton s constant Elmnaton and dspersal: The chemo taxs step provdes a bass for local search, and the productve step speeds the convergence. Whle to a large extent, only chemotaxs and reproducton are not enough for global optma searchng. Then an elmnaton and dspersal event s necessary. For each elmnaton and dspersal event each bacterum s elmnated wth a probablty Ped, and dspersed them to a new envronment. The selecton of Ped, play an mportant role n convergence of the algorthm. If Ped s large, the algorthm can degrade to random exhaustve search. If however, t s chosen approprately, t can help the algorthm jump out of local optma and nto a global optmum. PSEUDO CODE FOR : The algorthm s dscussed here. [step]: Intalzaton. p: number of parameter that be optmzed 2. S: the total number of bactera 3. C, re, ed: the number of chemo taxs steps, the number of reproducton steps, the number of elmnaton and dspersal events, respectvely. 4. s: the maxmum number of permssble successve steps 5. the s of : dattract, wattract, hrepellant, wrepellant 6. Ped: the probablty of elmnaton and dspersal event 7. C() : the step sze [step2]: Elmnaton and dspersal loop: l=l+ [step3]: Reproducton loop: k=k+ [step4]: chemotaxs loop: j=j+ [a]. for =,2,, take a chemo taxs step for each bacterum as follows: [b]. compute ftness functon, (, + J Let J(, J(, Jcc [c]. let J J (, to save ths snce we last may fnd a better cost va a run. [d]. Tumble: generate a random vector () that ( ) [e]. Move: Let ( ) ( j, k, ( k, c( ) T [f]. compute J(,j+, and let J(, J(, Jcc ( ) ( ) [g]. swm ) Let m = 0(counter for swm length) ) Whle m<s Let m=m+ If J(,j+, < Jlast (f dong better), Let Jlast=J(,j+, and ( ) ( j, k, ( k, c( ) T ( ) ( ) Else let m = s [h]. Go to next bacterum (+) f [step5]. If j<c, go to step 3. [step6]. Reproducton [a]. for the gven k and l, and for each =,2,,, Let c J(, J health Sort ths ftness n order of ascendng.

4 200 [b]. The Sr bactera wth the hghest Jhealth s de, the remanng Sr bactera wth the best s splt [step7]. If k < re, go to step 3. [step8]. Elmnaton and dspersal For =,2,,, wth probablty Ped, elmnate and dsperse each bacterum, and ths result n keepng the number of bactera n the populaton constant. To do ths, f a bacterum s elmnated, smply one to a random locaton on the optmzaton doman. If l< ed, then go to step 2; otherwse ED. 6. Bnary genetc algorthm (B) Ths algorthm mmcs the behavor of natural evolvement [9]. At frst, some bnary strngs are generated randomly. Each of these strngs smulates a chromosome and a set of chromosomes construct the populaton. By formng the populaton, the search for the optmal soluton starts. Accordng to the ftness of the chromosomes, the most sutable chromosomes are chosen to generate the next generaton. The B has specal operators to ncrease the algorthm performance whch are descrbed below. The Genetc Algorthm operators: Reproducton: In ths step, the prevous chromosomes wth the most ftness are utlzed as the next parents. The most famous method for ths process s the roulette wheel mechansm. Crossover: After selectng parents, the chldren chromosome should be constructed. The crossover operator should mx the parents chromosome to produce new chldren. Generally, the crossover coeffcent s selected around 0.8 to. Mutaton: The mutaton operator s utlzed to avod the premature convergence and escapng from local optma. Ths process s mplemented by convertng some bnary numbers from 0 to and vce versa. The mutaton coeffcent s usually between 0.00 and Generaton of the ntal populaton. 2- Evaluatng the ftness of each chromosome accordng to the objectve functon. 3- Constructng new chromosomes by the use of old ones. 4- Applyng the mutaton and crossover operators 5- Omttng some chromosomes to make space for the new ones. 6- Evaluatng the ftness of the new chromosomes accordng to the objectve functon and transferrng them to the populaton. 7- If the termnaton crteron s satsfed then fnsh the algorthm else return to step smulaton results Ths paper uses the and B to reduce the power losses and enhance the voltage devaton objectve functons. Both of the algorthms are appled to the 33-bus and 69-bus IEEE test systems and the optmal places and szes are evaluated. The smulatons are n the MATLAB package usng ewton-raphson load flow method. The output power generaton capactes of the s are as follows: MW MW Actve power 0 P 3 MVAR MVAR 0 Q Table : The smulaton results for the 33-bus IEEE test system Reactve power In the frst part, the optmal allocaton of one n the 33- bus system s assessed. In the next part, 2 s are assessed. The smulaton results are shown n Table. Test System The proposed algroth m Operaton status place (MW) (MVA place (MW) (MVA Ploss (KW) Qloss (KVA Actv e% Power loss reducton Reactve % ormal bus ormal 8 2 B

5 20 As t can be seen from Table, the amount of power losses n the 33-bus network usng 2 s s reduced effectvely. Also, t can be seen that has better performance n searchng the optmal operatng pont than B. In the case of usng 2 s by, the percent of actve and reactve power reducton are 84.4% and 83.3% respectvely. In the case of B, the percent of actve and reactve power reducton are 82.7% and 80.8%.00 respectvely. The results of voltage profle enhancement can be deduced from Fgures (3) and (4). As t can be seen, utlzaton of has resulted to better voltage level. In Fgure (5), the voltage devaton after usng one n the 33-bus test system by both and B s shown. As t can be seen from Fgure (5), the voltage profle n both cases are to some extent smlar Voltage Profle Wthout Wth Wth Bus umber(case 33 bus) Fg.3. The voltage devaton n the 33-bus IEEE test system usng B Voltage Profle Wthout Wth Wth Bus umber(case 33 bus) Fg.4. The voltage devaton n the 33-bus IEEE test system usng 204

6 Voltage Profle Bus umber(case33 wth ) and Fg.5. Comparson of the voltage devaton of the 33-bus test system n the case of usng both and B.00 Voltage Profle Bus umber(case33 wth 2 ) and Fg.6. Comparson of the voltage devaton of the 33-bus test system n the case of 2 s usng both and B Fgure (6) shows the voltage profle after usng 2 s for the 33-bus test system for and B. As t can be seen, at some ponts, s better when B s better n some others. For the 69-bus test system, 2 s are consdered. Table 2 shows the results assocated wth the exstence of and 2 s n the network. The total power losses of the 69-bus test system consderng 2 s have reduced notably. Ths power loss reducton s much for the than the B. For example, by the use of 2 s n the networ the percent of actve and reactve power loss reducton for are 94.6% and 90.29% respectvely when for B are 93.2% and 89.8% respectvely. As t can be seen, the amount of power loss reducton s more usng than B. The smulaton results for the voltage profle for both B and are shown n Fgures (7) and (8). In Fgure (9), the voltage devatons of the system after usng n the 69-bus test system for both and B are shown. As t can be seen, utlzaton of has reduced the voltage devaton

7 Voltage Profle Wthout Wth Wth Bua umber(case 69 bus) Fg.7. The voltage devaton of the 69-bus IEEE test system usng B Wthout Wth Wth 2 Voltage Profle Bus umber(case 69 bus) Fg.8. The voltage devaton of the 69-bus IEEE test system usng 204

8 Voltage Profle Bus umber(case69 wth ) and Fg.9. Comparson of the voltage devaton of the 69-bus test system n the case of usng both and B Test System As t can be seen from Fgure (2), after usng, the amount of power loss s reduced effectvely. For nstance, utlzaton of (by ) n the 69- bus test system has reduced the ntal actve and reactve power losses from 225 kw and 02 kvar The proposed algrothm to the sutable s of 26 kw and 6 kvar respectvely. Smlar results can be deduced for the B case for both actve and reactve power losses. Table 2: The smulaton results for the 69-bus IEEE test system Operato n status place (MW) (MVA place (MW) (MVA Ploss (KW) Qloss (KVA Actve % Power loss reducton Reactve % 69-bus ormal ormal B

9 COCLUSIOS Ths paper utlzes and B to optmze the total power losses as well as the voltage devaton of buses on both 33-bus and 69-bus IEEE dstrbuton test systems. Usng n the network results to notable reducton n the power losses and enhancng the voltage profle smultaneously. The near optma solutons found by both algorthms show the proper and accurate mplementaton and ablty of and B n optmal placement and szng problem. Techncally, has better performance and could acheve to more optmal solutons n regard to B. If the optmal placement s not mplemented properly, the fnal result can be a network wth more power losses and even voltage nstablty. However, n ths paper, the total stuaton of the networks from power loss and voltage profle ponts of vew s enhanced effectvely. ssue, Electrcal Power and Energy Systems, 32,pp ,200. [] Hung D.Q., Mthulananthan. and Bansal R.C., Analytcal Expressons for Allocaton n Prmary Dstrbutonetworks, IEEE Transactons on Energy Converson, vol. 25, no. 3, pp ,200. References [] Salvader, L.An, nternatonal perspectve on the future of power generaton and transmsson worldwde: the Italan case, IEEE Transacton on Energy Converson, Marc [2]G. Pepermans, J.Dresen,D.Haeseldonckx,R. Belmans and W.D' haeseleer, Dstrbuted generaton: defnton,benefts and ssues, Energy Polcy, In Press, Corrected Proof, Avalable onlne 20 ovember, [3]Khanabad, M.; Doostzadeh, M. Esmaelan, A. Mohsennezhad, M., Transmsson Congeston Management through Optmal Dstrbuted Generaton's Szng and, Internatonal Conference on Envronment and Electrcal Engneerng (EEEIC),20. [4] T.. Shukla, S.P. Sngh, K. B. a Allocaton of optmal dstrbuted generaton usng for mnmum system losses n radal dstrbuton networks, Internatonal Journal of Engneerng, Scence and Technology Vol. 2, o.3, pp ,200. [5] R. K. Sngh and S. K. Goswam, Optmum Allocaton of Dstrbuted Generatons Based on odal Prcng for Proft, Loss Reducton and Voltage Improvement Includng Voltage Rce Issue, Internatonal Journal of Electrcal Power and Energy Systems, Vol. 32, o. 6, pp ,200. [6]Deependra Sngh, Devender Sngh, and K. S. Verma, based Optmal Szng & Placement of Dstrbuted Generaton for Loss Mnmzaton, Internatonal Journal of Intellgent Systems and Technologeys, [7] J.H. Teng, T.S. Luor and Y.H. Lu, Strategc dstrbuted generator placements for servce relablty mprovements, IEEE Power Engneerng Socety Summer Meetng, vol 2, pp , July2002. [8] M. Mozaffar Legha, (20) Determnaton of exhauston and juncton of n dstrbuton network and ts loss maxmum, due to geographcal condton, MS.c Thess. Islamc Azad Unversty, Saveh Branch, Markaz Provnce, Iran. [9] Sngh D., Sngh D., Verma K.S., Multobjectve optmzaton for plannng wth load models, IEEE Transactons on Power Systems, vol. 24, no., pp ,2009. [0] Sngh R.K. and Goswam S.K., Optmum allocaton of dstrbuted generatons based on nodal prcng for proft, loss reducton, and voltage mprovement ncludng voltage rse 204