Optimal Location for Fixing Fuel Cells in a Distributed Generation Environment using Hybrid Technique

Transcription

1 Internatonal Journal on Electrcal Engneerng and Informatcs - Volume 8, Numer, Septemer 016 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Envronment usng Hyrd Technque T.C.Suramanyam 1, S.S.Tulas Ram, and JBV Surahmanyam 1. JNTU College of Engneerng, Hyderaad, Electrcal& Electroncs Engneerng Chrstu jyoth nsttute of technology scence, Jangaon town,warangal suramanyamtc05@gmal.com Astract: The paper proposes optmal locaton for fxng fuel cells n a dstruton system usng an nnovatve hyrd technque. The novelty of the proposed method s the comned performance of the Genetc Algorthm (GA) and Recurrent Neural Network (RNN) technque, therey ntegratng GA frst phase, RNN technque and GA second phase. The optmum locaton for fxng the fuel cell s attaned y usng the GA frst phase. Here, the GA frst phase utlzes the load flow data at dfferent loadng condtons for determnng the optmum locaton. The RNN s aptly traned y the target fuel cell sze and the correspondng nputs such as load varaton and us numer. Durng the testng tme, the RNN provdes the fuel cell capacty accordng to the load varaton and us numer. By usng the attaned fuel cell capactes, the GA second phase optmzes the fuel cell capacty to mnmze the power loss and the voltage devaton. The ojectve functon manly helps to mprove the us voltage profle and the power loss reducton. Thus, the proposed hyrd technque s mplemented n the MATLAB/smulnk platform and ts effectveness s analyzed y comparng t wth the GA, PSO and other hyrd PSO technques. The comparson results unequvocally demonstrate the superorty of the proposed approach and confrm ts sterlng potental to solve the prolem. Keywords: fuel cell, GA, PSO, voltage, real power, reactve power 1. Introducton The partcular electrcal power system commonly features a generatng system, a transmsson system, power sustatons and the dstruton network [1]. Prncple performance of the ulk generaton and transmsson system would e to provde wth the electrc power and also energy essental for the customers that has a specfc level of relalty and good qualty of support, at least expense []. Currently, the partcular electrcal power market can e n the process of szeale transform regardng desgn, functonng and also regulaton []. Deployment of dstruted generators (DG) wthn just dstruton network s ecomng more nvtng on account of many enefts whch small scale generaton could possly to enhance the electrc power resources [1]. The partcular dstruted generaton (DG) can e consttuted wthn a new emphass for the electrcal power generaton [4]. Dstruted Generaton can e defned as any knd of power generaton that's undled n the dstruton process [5]. Dstruted generaton has extensvely found ther place n power process whch s gong to help the organzaton needs on relalty cost and worth requrements [6]. Consequently, nstallng DG n dstruton system often have mportant mpact on power flow, voltages n addton to relalty ndces that may e postve or maye negatve. It mght e optmstc should they are generally correctly matched usng all of those other system [7]. The actual stng connected wth dstruted generator throughout dstruton feeders proaly wll have an mpact around the operatons and handle the power process, a process created to perform usng large, and core generatng amentes [11]. Apart from, applcaton of DGs offers several advantages, ncludng far etter economy whch have a practcal the actual advancement of huge power plants, decreased envronmental polluton, hgher effcency, ncreased hgh qualty of power supply to the users, decreased loss wthn dstruton systems, ncreased voltage profle, and releasng of network Receved: June 10 th, 015. Accepted: July rd, 016 DOI: /jee

2 T.C.Suramanyam, et al. capacty [1]. Relalty can e nterdependent havng economcs and ncreased expense s essental to accomplsh ncreased stalty or maye to keep up stalty on latest and acceptale levels [8]. Random approaches are now used more roadly wthn power system operatons and plannng due to a varous uncertantes engaged [9]. Loss of load proalty (LOLP) s the prmary nstance that employs ndexes for plannng generaton capacty [10]. The paper proposes optmal locaton for fxng fuel cells n a dstruton system usng an nnovatve hyrd technque. The novelty of the proposed method s the comned performance of the Genetc Algorthm (GA) and Recurrent Neural Network (RNN) technque, therey ntegratng GA frst phase, RNN technque and GA second phase. The optmum locaton for fxng the fuel cell s attaned y usng the GA frst phase. Here, the GA frst phase utlzes the load flow data at dfferent loadng condtons for determnng the optmum locaton. The RNN s aptly traned y the target fuel cell sze and the correspondng nputs such as load varaton and us numer. Durng the testng tme, the RNN provdes the fuel cell capacty accordng to the load varaton and us numer. By usng the attaned fuel cell capactes, the GA second phase optmzes the fuel cell capacty to mnmze the power loss and the voltage devaton. The ojectve functon manly helps to mprove the us voltage profle and the power loss reducton. The rest of the paper s organzed as follows:. Past to that partcular, ths current exploraton works tend to e offered wth secton. The effects along wth the dscusson tend to e offered wth secton 4. Wthn secton 5 the paper s usually concludes.. Lterature Survey: A Recent Related Work From the lterary works, varous connected performs usually are readly avalale whch often usng the est locaton and alty of the mcro grd n the dstruton system surroundngs. Some of them usually are looked at here. Qn et al. [14] have looked nto reactve power aspects wthn power system relalty analyss. A strategy s actually suggested to analyze technque n addton to load pont relalty of power systems wth reactve power shortage on account of falures due to reactve power resources for generators, synchronous condensers, and compensators. Ths relalty ndces on account of reactve power shortage usually are dvded wth those of real power shortage. Reactve shortage s estalshed makng use of reactve power procedure for the nodes whle usng voltage ause to delver more detals pertanng to system plannng and operaton. Mohammad et al. [15] have offered a method usng Partcle Swarm Optmzaton approach (PSO) for placng Dstruted Generators (DG) n the radal dstruton systems to reduce the actual real power losses also to mprove process relalty. A hyrd goal purpose s employed wth the maxmum DG poston. It's got two parts, n frst component the power loss s gven and s named as frst ndex. Power Loss Reducton Index s known as the second component the consequence of DG aout relalty mprovement of process have een thought to e a sngle ndex known as Relalty Improvement Index. Heydar et al. [16] have suggested a prolem formulaton and remedy for the placement and dmensons of DGs optmally. The target s usually to enhance the relalty ndces. The placement along wth sze regardng DGs are generally optmzed havng a Genetc Algorthm (GA). To analyze the recommended algorthm, the partcular IEEE 4 uses dstruton feeder s employed. Kansal et al. [17] get recommended the use of Partcle Swarm Optmzaton (PSO) technque to search for the est dmensons as well as optmum locaton for that placement of DG nsde radal dstruton systems pertanng to actve electrcal power compensaton y lowerng of electrcal power loss as well as development throughout voltage profle. In frst porton, the perfect dmensons of DG can e calculated on each us employng the precse loss formula as well as n the next porton the optmal locaton of DG can e found wth the loss senstvty factor. Ths analytcal expresson wll e ased upon precse loss formula. The optmal dmensons of DG can e calculated on each us y employng the precse loss formula and also the est locaton of DG can e found wth the loss senstvty factor. 568

3 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Mohammad et al. [18] have revewed aout an optmal DG unt placement usng GA. The optmal dmensons from the DG unt s usually computed analytcally usng estmated reason acceptale nodes are motvated ntended for DG unt placement. Relalty along wth power loss reducton ndces of dstruton system nodes are desgned. GA that contan a few rules can e used to look for the DG unt placement. DG unt they ft wth all the hghest sutalty ndex. Smulaton results demonstrate the advantage of maxmum DG unt placement. Palwal et al. [19] have recommended a soluton to fnd optmal dstruted generaton allocaton for loss mnmzaton assocated wth voltage regulaton n dstruton network. The system s usually more examned for elevated numers of Relalty. Dstruted Generator supples the addtonal advantage to maxmze throughout relalty levels snce proposed through the upgrades n numerous relalty ndces such as SAIDI, CAIDI and AENS. Relatve scentfc studes are usually executed and lnked the desred results are resolved. Shayegh et al. [0] have proposed the optmal generaton expanson plannng n restructured power system employng the hyrd coded genetc algorthm and partcle swarm optmzaton. Addtonally, ndvdual power producer s contruton as well as a couple of relalty crtera (LOLP and EENS) are deemed wth GEP ssue. The recommended technque s a quck way of computaton regardng relalty crtera and wll greatly attan maxmum purchase prces for several types of IPP. Dstruted generator s now commonly used n dstruton system to mprove the overall performance of the dstruton system. Major advantages of usng dstruted generator n dstruton system are: t reduces total power losses n the system, mprovement n voltage profle and relalty of the system and many more. There are dfferent types of dstruted generators are n lterature. Some of them are wnd power, solar power, hydroelectrc power, tdal power, small hydro power, photovoltac cells, fuel cells etc. Among the dfferent types of dstruted generators, ths work consders fuel cells for dstruted generaton. The fuel cells are generated ecause of ts renowned advantages over the conventonal resources that are ncreased effcency, ncreased relalty, less mantenance, excellent part- load performance, modularty and fuel flexlty, low chemcal, acoustc and thermal emssons. Whle revewng the recent research works, t s clear that n most of the works any one of the load condton s consdered and some of the works determned the optmal locaton, ut the result s not up to the level and also n most of the works fuel cells are not consdered as dstruted generator. The major prolem n dstruted generator s dentfyng optmal locaton for fxng dstruted generator and also the amount of power to e generated y that dstruted generator depends on the load condtons. The formulated prolem related to the aove mentoned prolems s descred n the followng secton.. Prolem formulaton The fuel cells nstallaton at optmal locaton ultmately leads to varous factors such as lne loss reducton, mproved voltage stalty, relalty and securty. The optmum locaton and szng of the fuel cell are the optmzaton prolems wth nonlnear ojectve functon havng correspondng constrants lke power alance constrant, voltage constrant and fuel cell constrant etc. The man am of the proposed method s focused on reducng the exact power loss, load alancng and voltage devaton of the gven radal dstruton network at the peak load condton. Hence, t utlzes the mult-ojectve functon to determne the optmum locaton and szng of the fuel cell. Here, the mult-ojectve functon s mathematcally formulated as per the followng equaton (1). J Mn f 1, f (1) Where, f 1 and f are the power loss and voltage devaton respectvely. The mathematcal equaton of the mult-ojectve functon s descred as follows: 569

4 T.C.Suramanyam, et al. A. Power loss ( f 1 ) The dstruton systems uld nstalty durng the peak loadng condtons, whch leads to constrant volatons. The otaned fuel cell locaton and capacty mnmze the power loss and realze the lmts of the constrants. The requred exact loss of the dstruton system s calculated y the followng equaton [6] (). n 1 PL I R () 1 f Where P L s the exact loss of the dstruton system, R the resstance etween us and us j, and I, the lne current. The voltage devaton s determned as detaled n the followng secton.1. B. Voltage devaton ( f ) The radal dstruton network voltage profle management s the man factor. When the fuel cell s connected to a dstruton network, the voltage profle s changed. It can e evaluated at all of the uses n the radal dstruton systems. Here, the dentfed fuel cell has to mnmze the dfference etween the normal us voltage and the rated us voltage for mprovng the voltage stalty. The requred voltage devaton equaton s descred n the followng equaton (5). f N V V rated 1 (5) V the voltage at us and N the numer of uses. Where, V rated s the specfed voltage; The ojectve functon s dependent on the constrants such as the power alance constrants, us voltage constrants and DG capacty constrants, whch are descred n the followng secton.. C. Constrants (). Power alance constrant The power alance equaton manly descres the fact that the generated fuel cell power has to satsfy oth the demand and the loss of the system. The requred power alance equaton s descred n the followng equatons (8) and (9). P Q D D D D Where, P D G Q D G Y Y D P G and N V V j j1 N j j1 of the lne etween and j, j cos( ) (8) j j j V V sn( ) (9) D Q G are the power generatons of generators at us, the phase angle of the us, at us j whch le n the followng lmts Q Q Q. D(mn) D D(max) D D D D P D and P Y j the admttance D Q D the load demands P P D(mn) D D(max) D D D and 570

5 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton (). Bus voltage constrant The voltage lmts of each us must le wthn the prescred lmts, whch are gven n the followng equaton (10). mn max V V V (10) Where, V mn and normally the us voltage les etween max V are the mnmum and maxmum values of voltage at us ; 0.95 V pu. (). Fuel cell constrant The fuel cell constrant manly conssts of the allowale capacty for the uses and the correspondng power factor of the fuel cell. The capacty of the fuel cell s descred n the followng equaton (11). mn max P P P (11) DG DG DG The proposed method utlzes the constrants to fnd the mnmum power loss attanng us durng the faulty condton, whch s known as optmum locaton. Dependng on the loss functon the correspondng capacty of fuel cell s dentfed and added to the partcular locaton. The ref process aout the determnaton of fuel cell locaton and capacty usng the proposed method s descred n the followng secton Optmal Locaton and Optmal Capacty of Fuel Cell usng Hyrd method The optmal locaton and capacty of the fuel cell are formulated as mult-ojectve constraned optmzaton prolems. Ths paper uses a novel comned GA and RNN for solvng the optmal locaton and capacty of the fuel cell. Here, the proposed hyrd method utlzes GA n two stages, such as the GA frst phase and GA second phase wth the comnaton of RNN technque. The process to dentfy the tranng dataset usng GA frst phase s explaned n the followng secton 4.1. A. GA Frst Phase Based Optmal Locaton Determnaton Of late, GAs have assumed supreme sgnfcance as the unversal purpose optmzaton algorthms n accordance wth the procedure of natural selecton and genetcs. They functon on strng structures such as chromosomes, characterstcally a concatenated lst of nary dgts symolzng a programmng of the control constrants lke phenotype of a specfed dlemma [1]. In the proposed method, GA frst phase s utlzed for attanng the optmum locaton of the fuel cell. Here, the ojectve of the GA frst phase s focused on mnmzed power loss attaned us durng the dfferent types of loadng condtons. The ackward sweep and forward sweep technque of supply load flow s effectvely employed to perform the preferred target []. The crossover and mutaton s neglgle for the optmum locaton determnaton process. The steps to dentfy the optmum locaton for radal dstruton network are descred elow: Steps of the GA frst phase Step 1: Run the load flow equaton for normal condton and dfferent types of loadng condtons. Step : Intalze the requred parameters of GA such as radal dstruton network us data set as N numer of uses, us voltage V ), real power loss and reactve power loss ( P and Q ) etc. L L ( 571

6 T.C.Suramanyam, et al. Step : Generate the random populaton of load value and apply t to the uses. The requred random populaton chromosomes are descred n the followng equaton (1). 1 d X [ X, X X ] (1) Where, 1,... n, and d s specfed as the dmensons of the populaton space. Step 4: Set the count k =k+1 Step 5: Fnd the ftness as follows ftness Mn( P L ) (1) Step 6: Select the chromosome est X, whch has the mnmum ftness. Step 7: Apply the load changes and go to step 4, untl the requred termnaton crtera s acheved. At the end of the process, the GA frst phase develops the optmum locaton to place the fuel cell. Then RNN gves the capacty of the fuel cell to solve the prolem. The process to get the fuel cell capacty usng RNN s explaned n the followng secton 4.. B. RNN Based Fuel Cell Ratngs predcton Recurrent neural network (RNN) s, n fact, analogous to feed-forward neural network (FNN), though RNN comprses feedack loop around neuron. Feedack loop, on the other hand, also comprses unt delay operator (z -1 ) [4]. The presented neurons have the nteror connectons and each neuron n RNN receves a numer of nputs, dependng on the actvaton functons of the RNN results n the output level of the neuron. The learnng task s gven n the form of examples, whch s known as tranng examples. Normally RNN has three layers lke nput layer, hdden layer and output layer. Here, the RNN s traned y the target fuel cell capacty wth the correspondng nputs such as load varaton and the us numer. The RNN structure s explaned n the fgure 1. The supervsed learnng process s utlzed for tranng the RNN, whch s refly descred as follows. Fgure 1. Structure of the RNN 57

7 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Supervsed Learnng and Tranng Process Ths secton descres the tranng process of the RNN. Here, the supervsng learnng law of the gradent descent s used to tran the RNN durng the end of the ntalzaton process. The dervaton s smlar to the ack propagaton algorthm. It s used to ensure the weght adjustments w, w and w of the RNN y usng the tranng datasets. By usng the chan a rule, error etween the actual output and target output s calculated and updated. The man purpose of the supervsng learnng algorthm s to mnmze the error functon, whch s explaned n the followng equaton (14). 1 1 E ( Wa Wt ) es (14) Where, W s the actual output of the network, a W t the target output of the network and E the error functon. The error calculaton and weght updatng are explaned as follows: Layer 1: Ths layer s used to update the weght of the w. Here the updated weght s gven y the followng equaton (15). w ( N 1) w ( N) w ( N) (15) Where, w E R propagates the error term, Layer : E R c cr wth c Rc Net c s the learnng rate for adjustng the parameter Ths layer performs multplcaton operaton and the updated rule for y the followng equaton (16) and (17). Wth E E e s c s Rc es Rc w. w and w s gven w ( N 1) w ( N) w ( N) (16) w a ( N 1) w ( c) w ( N) (17) a a c c Where, w w P a w a E w a E w E R c E R R R c c a R R R R R cw wa and are the learnng rates for adjustng the parameter c Q a a w and w respectvely, w, w and w are the tunng parameters. We can derve a learnng algorthm that drves E a to zero. Once the process s fnshed, the RNN s ready to gve the fuel cell capacty. But the optmum capacty of the fuel cell to mprove the voltage profle and mnmze the power loss of the system s determned y the GA second phase. The detaled process of the fuel cell optmum capacty determnaton s descred n the followng secton 4.. a 57

8 T.C.Suramanyam, et al. 4.. Fuel Cell Capacty Optmzaton Usng GA Second Phase Ths secton descres the fuel cell capacty determnaton usng GA second phase. The optmum locaton parameters are used as the nputs of the GA second phase. Here, the fuel cell capacty [5] s randomly generated and appled for the otaned optmum locaton. From the appled capactes, we can fnd the optmum capacty of fuel cell y usng the mult-ojectve functon,.e., mnmzaton of oth power loss and voltage devaton. The step y step process to otan the optmum capacty of the fuel cell s explaned elow: Steps to fnd the capacty of fuel cell Step 1: Intalze the nput parameters lke requred radal dstruton system us data such as us voltage ( V ), power loss ( P ), fuel cell capacty lmt etc. L Step : Set the tme counter t=0 and randomly generate the n chromosomes,.e., fuel cell szes wthn the searchng space dmenson Y, Y ]. [ mn max 1 d Y [ Y, Y Y ] (18) Where, 1,... n, and d s specfed as the dmensons of the populaton space. Step : Evaluate each chromosome n the ntal populaton usng the ojectve functon (1) and search for the est value of the ojectve functon J. Ths step fnally Sets the chromosome proportonally to the J est as the est. Step 4: Update the tme counter t = t+1. Step 5: Create a new populaton y repeatng the followng steps untl the new populaton s completed. Selecton: select the two parent chromosomes from the populaton accordng to the ftness functon. Crossover: The crossover operaton s acheved etween the two chromosomes, leadng to generaton of a new set of chromosomes. Mutaton: By usng the mutaton proalty, the method mutates a new chld at each chromosome. Acceptance: Place new chld n a new populaton. Step 6: Run the algorthm wth the new set of populaton. Step 7: If one of the stoppng crtera s acheved, stop the operaton,. else go ack to the step. Once the aove mentoned steps are fnshed, the system s ready to gve the optmum capacty of the fuel cell wth mproved voltage profle and mnmum power loss. The proposed method workng structure s gven n the followng fgure. The proposed method s mplemented n the MATLAB platform through IEEE standard ench mark system and the performance s evaluated y usng the comparson studes, whch s descred n the followng secton 5. est 574

9 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Fgure. Structure of the proposed hyrd technque 5. Numercal Results and Dscusson The proposed mutual method s mplemented n MATLAB/smulnk R01a platform, 4GB RAM and Intel(R) core(tm) -100 CPU wth.10 GHz. The IEEE us radal dstruton system wth.7 MW and. MVAr s utlzed for the testng of the proposed hyrd method [7,8]. The mentoned testng system conssts of nodes and ranches. Here, the ackward sweep and forward sweep method of dstruton load flow [9] s used to fulfll the desred ojectve. The effectveness of the proposed method s dentfed y usng the 575

10 T.C.Suramanyam, et al. comparatve analyss wth the GA, PSO and hyrd PSO technques. The results are dsplayed as follows. Fgure. Structure of the IEEE us dstruton system The followng fgure 4 descres the IEEE us radal dstruton system normal voltage profle. It s seen that the radal dstruton network mantans the us voltage wthn the specfed lmt at all the uses,.e., the nomnal voltage of the sustaton s 1 pu. The testng system lne loss at normal condton s descred n the fgure 5. The maxmum lne losses are present n the us system at normal condton s 58 kw. The fgure 6 shows the us voltage of the radal dstruton system at fault tme. Due to the load ncreasng at the range of 0%, the us voltage s volated from the normal condton. The power loss of the IEEE us radal dstruton system at fault tme s gven n the fgure 9. It s oserved that the power loss s ncreased due to the load varaton of the dstruton system. In the faulty condton the us power loss s lkely to go up to 6kW, whch s a hgh loss compared to the normal condton. So t s essental to fnd the optmum locaton to place a fuel cell at rght capacty. The proposed method s utlzed for determnng the optmal locaton to acheve mnmum power loss y placng the optmum capacty of fuel cell. The us voltage profle after locatng the fuel cell s explaned n the fgure 8, at the same tme the lne losses of the system are descred n the fgure 9. Form the fgure, t s crystal clear that the proposed method voltage profle s effectvely mantaned near the normal condton and the power loss s mnmzed at 58kW. The fgure 10 descres the proposed method ftness at 0% load varaton, whch s dependent on the teratons. Fgure 4. Normal us voltage 576

11 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Fgure 5. Normal power loss Fgure 6. Bus voltage at 0% load varaton Fgure 7. Power loss at 0% load varaton 577

12 T.C.Suramanyam, et al. Fgure 8. Bus voltage usng proposed method at 0% load varaton Fgure 9. Power loss usng proposed method at 0% load varaton Fgure 10. Ftness of the proposed method at 0% load varaton 578

13 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Fgure 11. Bus voltage at 0% load varaton Fgure 1. Power loss at 0% load varaton Fgure 1. Bus voltage usng proposed method at 0% load varaton 579

14 T.C.Suramanyam, et al. Fgure 14. Power loss usng proposed method at 0% load varaton Fgure 15. Ftness of the proposed method at 0% load varaton Fgure 16. Bus voltage at 40% load varaton 580

15 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Fgure 17. Power loss at 40% load varaton Fgure 18. Bus voltage usng proposed method at 40% load varaton Fgure 19. Power loss usng proposed method at 40% load varaton 581

16 T.C.Suramanyam, et al. Fgure 0. Ftness of the proposed method at 40% load varaton Fgure 1. Bus voltage comparson at 0% load varaton Fgure. Power loss comparson at 0% load varaton 58

17 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton Fgure. Ftness comparson at 0% load varaton The IEEE 0 radal dstruton system us voltage at 0% load varaton s descred n the fgure 11. It clearly shows the voltage devaton on account of the load varaton compared to the normal condton. Also the power loss at the 0% load ncrement s descred n the fgure 1, whch shows that the maxmum power loss of ths load condton s 64kW. In ths condton the proposed method optmzed results are appled and the correspondng voltage and power loss are explaned n the fgures 1 and 14. After the arrangement of fuel cell, t mproves the voltage profle n the specfed lmt and reduces the power loss nto 55kW. The proposed method ftness evaluaton dependent on the teraton s explaned n the fgure 15. The IEEE us system s allowed to meet the 40% load ncrement, whch affects the power flow quanttes of the system. Durng the load ncrement perod the system normal voltage profle gets collapsed, whch s shown n the fgure 16. Smlar load condton the system power loss s descred n the fgure 17. In the stuaton, to restore the normal condton of the us system wth the help of optmum ratng of fuel cell at the optmum locaton, the proposed method fnds the locaton and capacty of the fuel cell at the 40% load ncrement. After the fuel cell placement the us voltage profle s analyzed, whch s shown n the fgure 18. Due to the voltage stalty the constrants are mantaned n the stale lmt, whch reflects mnmum power loss. The attaned mnmzed power loss s explaned n the fgure 19. The ftness of the proposed method durng the 40% load varaton of the us system s explaned n the fgure 0. Then the proposed method results are compared to the other optmzaton technques such as GA, PSO and hyrd PSO. Here, we fnd the optmum locaton and capacty of the fuel cell usng the aove mentoned technques for 0% load varaton. Intally the proposed method us voltage profle for 0% load varaton s compared wth the aove mentoned technques, whch s descred n the fgure 1. From the comparson, we oserve that the proposed method voltage profle s well mproved near the normal voltage profle compared to the other technques. The power loss of the dfferent methods s compared n the fgure. Here, the proposed method effectvely reduces the power loss y selectng the optmum locaton and capacty of the fuel cell. Fnally the proposed method ftness evaluaton durng the teraton process s compared wth the other technques, whch s descred n the fgure. For attanng the optmum locaton and capacty of fuel cell t s essental to mnmze the voltage devaton and power loss. 58

18 T.C.Suramanyam, et al. Load n % Bus no. Tale 1. Performance analyss of the GA technque Power loss n kw Voltage Fuel cell capacty Normal Power loss After fault After fuel cell placement Mn Max CPU tme (sec) Load n % Bus no. Tale. Performance analyss of the PSO technque Power loss n kw Voltage Fuel cell Normal After fuel After capacty Power cell Mn Max fault loss placement CPU tme (sec) Load n % Bus no. Tale. Performance analyss of the hyrd PSO technque Power loss n kw Voltage Fuel cell Normal After fuel capacty After Power cell Mn Max fault loss placement CPU tme (sec) Load n % Bus no. Tale 4. Performance analyss of the proposed method Power loss n kw Voltage Fuel cell Mn Max capacty Normal Power loss After fault After fuel cell placement CPU tme (sec) The aove tales explan the performance analyss of the dfferent technques lke GA, PSO, hyrd PSO and the proposed method. Tale 1 and tale show the performance analyss of oth the GA technque and PSO technque respectvely. The hyrd PSO performance s analyzed n the tale and the proposed method effectveness s analyzed n the tale 4. The aove mentoned technques are tested aganst varous load condtons lke 0%, 0% and 40% load varatons. Durng the respectve load condtons the power loss, voltage devaton and tme for attanng the optmum results are taulated. From the tales we can conclude that the proposed method effectvely attans the mnmum power loss and mproved voltage profle, therey, optmally selectng the locaton for fxng the fuel cell effectvely. 584

19 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton 5. Concluson Ths paper ntroduces a hyrd technque for locatng the capacty of fuel cell n the dstruted generaton system. In the proposed method, the optmal locaton for fxng the fuel cell s determned y the GA frst phase and the fuel cell capacty s predcted y the RNN. The optmum capacty of the fuel cell to reduce the power loss and voltage devaton s attaned y usng the GA second phase. The advantage of the proposed method s ts enhanced capacty to acheve mproved us voltage profle, reduced power loss, transmsson and dstruton relef capacty for oth utltes and the customers. Ths process s tested n the IEEE standard radal dstruton enchmark systems and the effectveness s analyzed wth dfferent algorthms. Here the comparson analyss s made etween the radal dstruton system power loss and voltage at varous condtons lke normal condton, durng the fault tme, GA, PSO, hyrd PSO and proposed technques. From the comparson results we are convnced that the proposed method s the well effectve technque to dentfy the optmum locaton and capacty of the fuel for the radal dstruton system, whch s superor to the other technques. 6. References [1]. Had Zayandehrood, Azah Mohamed, Hussan Shareef and Marjan Mohammadjafar, "Impact of dstruted generatons on power system protecton performance", Internatonal Journal of the Physcal Scences Vol. 6, No. 16, pp , Aug 011. []. Seastan Ros. M, Vctor Vdal. P and Davd L. Kguel, "Bus-Based Relalty Indces and Assocated Costs n the Bulk Power System", IEEE Transactons on Power Systems, Vol. 1, No., pp , Aug []. A. A. Chowdhury, Sudhr Kumar Agarwal and Don O. Koval, "Relalty Modelng of Dstruted Generatonn Conventonal Dstruton SystemsPlannng and Analyss", IEEE Transactons on Industry Applcatons, Vol. 9, No. 5, pp , Oct 00. [4]. F. Gharedagh, M. Deys, H. Jamal, A khall, "Investgaton of Power Qualty n Presence of Fuel Cell Based Dstruted Generaton", Australan Journal of Basc and Appled Scences, Vol. 5, No. 10, pp , 011. [5]. Akash T. Davda, M. D. Desa and B. R. Parekh, "Impact of Emeddng Renewale Dstruted Generaton on Voltage Profle of Dstruton System: A Case Study", ARPN Journal of Engneerng and Appled Scences, Vol. 6, No. 6, pp , June 011. [6]. Moen Moen-Aghtae, Payman Dehghanan and Seyed Hamd Hossen, "Optmal Dstruted Generaton Placement n a Restructured Envronment va a Mult-Ojectve Optmzaton Approach", 16th Conference on Electrcal Power Dstruton Networks (EPDC), Iran, pp. 1-6, 011. [7]. R. Yousefan and H. Monsef, "DG-Allocaton Based on Relalty Indces y Means of Monte Carlo Smulaton and AHP", 10th Internatonal Conference on Envronment and Electrcal Engneerng (EEEIC), Iran, pp. 1-4, 011. [8]. Lmu, Tka R. and Saha, Tapan K., "Investgatons of the mpact of powerformer on composte power system relalty", Proceedngs of the IEEE Power Engneerng Socety General Meetng, Unted States, pp , 005. [9]. Lngfeng Wang and Chanan Sngh, "Adequacy Assessment of Power-generatng Systems Includng Wnd Power Integraton Based on Ant Colony System Algorthm", IEEE Power Tech, Lausanne, pp , 007. [10]. Saket R K, Bansal and R C, Sngh G, Generaton capacty adequacy evaluaton ased on peak load consderaton, The South Pacfc Journal of Natural Scence Vol. 4, pp. 8 44, 006. [11]. Bndeshwar Sngh, K.S. Verma, Deependra Snghand S.N. Sngh, "A Novel Approach for Optmal Placement of Dstruted Generaton & FACTS Controllers In Power Systems: An Overvew and Key Issues", Internatonal Journal of Revews n Computng, Vol. 7, pp. 9-54,

20 T.C.Suramanyam, et al. [1]. Fara Gharedagh, Haneh Jamal, Mansoureh Des and Atefeh Khall, "Investgaton of a new slandng detecton method for dstruted power generaton systems", Internatonal Journal of the Physcal Scences Vol. 6, No., pp , Oct 011. [1]. Seyed Al Mohammad Javadan and Maryam Massael, "Impact of Dstruted Generaton on Dstruton System s Relalty Consderng Recloser-Fuse Mscoordnaton-A Practcal Case Study", Indan Journal of Scence and Technology, Vol. 4, No. 10, pp , Oct 011. [14]. Wenpng Qn, Peng Wang, Xaoqng Han, and Xnhu Du, "Reactve Power Aspects n Relalty Assessment of Power Systems", IEEE Transactons ON Power Systems, Vol. 6, No. 1, pp. 85-9, Fe 011. [15]. Mohammad Mohammad and M. Akar Nasa, "PSO Based Multojectve Approach for Optmal Szng and Placement of Dstruted Generaton", Research Journal of Appled Scences, Engneerng and Technology, Vol., No. 8, pp. 8-87, 011. [16]. Morteza heydar, Amn Hajzadeh and Mahd Banejad, "Optmal Placement of Dstruted Generaton Resources", Internatonal Journal of Power System Operaton and Energy Management, Vol. 1, Issue., pp , 011. [17]. Satsh Kansal, B.B.R. Sa, Barjeev Tyag and Vshal Kumar, "Optmal placement of dstruted generaton n dstruton networks", Internatonal Journal of Engneerng, Scence and Technology, Vol., No., pp , 011. [18]. Mohammad Mohammad and M. Akar Nasa, "DG Placement wth Consderng Relalty Improvement and Power Loss Reducton wth GA Method" Research Journal of Appled Scences, Engneerng and Technology, Vol., No. 8, pp , 011. [19]. Pryanka Palwal and N.P. Patdar, "Dstruted Generator Placement for Loss Reducton and Improvement n Relalty", World Academy of Scence, Engneerng and Technology, Vol. 69, pp , 010. [0]. H. Shayegh, H. Hossen, A. Shaan and M. Mahdav, "GEP Consderng Purchase Prces, Profts of IPPs and Relalty Crtera Usng Hyrd GA and PSO", World Academy of Scence, Engneerng and Technology, Vol. 44, pp , 008. [1]. D. E. Golderg, "Genetc Algorthms n Search, Optmzaton and Machne Learnng, Readng", MA: Addson-Wesley, []. Haque.MH, "Effcent load flow method for dstruton systems wth radal or mesh confguraton", In Proceedngs of IEEE Generaton Transmsson and Dstruton, Vol.14, No.1, pp.8-, []. Chung Hsng Chen, Chh-Mng Hong and Fu-Sheng Cheng,"Intellgent speed sensorless maxmum power pont trackng control for wnd generaton system", Electrcal Power and Energy Systems, Vol.4, pp , 01. [4]. B. Purwahyud, M.H. Purnomo, Soeago, M. Ashar, and T. Hyama, Desgn of Recurrent Neural Network Oserver for Inducton Motor, The rd Internatonal Student Conference on Advanced Scence and Technology, pp.05-06, 009. [5]. Taher Nkam and Bahman Bahman Frouz, "A practcal algorthm for dstruton state estmaton ncludng renewale energy sources", Renewale Energy, Vol.4, pp.09-16, 009. [6]. Sarang Pande and J.G.Ghodekar, "Computaton of Techncal Power Loss of Feeders and Transformers n Dstruton System usng Load Factor and Load Loss Factor", Internatonal Journal of Multdscplnary Scences and Engneerng, Vol., No. 6, pp.- 5, June 01. [7]. Kashem MA, Ganapathy V, Jasmon GB and Buhar MI, "A novel method for loss mnmzaton n dstruton networks", In Proceedngs of nternatonal conference on electrc utlty deregulaton and restructurng and power technology, London, Aprl 000. [8]. Satsh Kansal, Vshal Kumar and Barjeev Tyag, "Optmal placement of dfferent type of DG sources n dstruton networks", Electrcal Power and Energy Systems, Vol.5, pp ,

21 Optmal Locaton for Fxng Fuel Cells n a Dstruted Generaton [9]. Haque MH, "Effcent load flow method for dstruton systems wth radal or mesh confguraton" In Proceedngs of IEEE Generaton Transmsson and Dstruton, Vol.14, No.1, pp. 8, T.C. Suramanyam s a research scholar n JNTU college of Engneerng and he s a faculty n Electrcal & Electroncs Engneerng department of School of Engnnerng, Nalla Narasmhareddy group of nsattutons, Hyderaad, Inda, wth 17 years of rch experence n teachng, tranng, research. Hs research nterest s n electrcal power dstruton, power system relalty and automaton of power systems. He pulshed 11 research papers n reputed nternatonal & natonal journals and 4 papers n natonal conferences S.S.Tulas Ram receved B.Tech, M.Tech and Ph.D degrees n Electrcal Engneerng from JNTU College of Engneerg, Kaknada Currently, he s workng as Professor of Electrcal& Electroncs Engneerng n JNTU College of Engneerng, Hyderaad, wth three decades of rch experence n teachng, tranng, research, projects and admnstraton. Hs area of nterest nclude hgh Voltage engneerng, power system analyss and control. He pulshed 15 research papers n reputed nternatonal & natonal journals and 8 papers n nternatonal and natonal conferences. JBV Surahmanyam s a Doctorate n Electrcal Engneerng from JNTU- Hyderaad, Inda, wth two decades of rch experence n teachng, tranng, research, ndustry, projects and consultancy. He pulshed 5 research papers n reputed nternatonal journals and 0 papers n nternatonal and natonal conferences. Hs research nterest s n automaton of power systems. He s an expert n condton montorng of ndustral equpment through modern dagnostc technques. He mplemented the latest GPS and GIS technologes n many power utltes n Inda successfully. He executed many nternatonal and natonal level techncal projects effectvely funded y Power Fnance Corporaton, Mnstry of Power, Government of Inda, APDRP, DRUM, USAID and DFID-UK. 587