Simultaneous Reconfiguration with DG Placement using Bit-Shift Operator Based TLBO

Transcription

1 Smultaneous Reconfguraton wth Placement usng Bt-Shft Operator Based TLBO Ayan Ghosh, *Deepak Kumar EEE Department B.I.T Mesra, Ranch Jharkhand, Inda S. R. Samantaray School of Electrcal Scences IIT Bhubaneswar Odsha, Inda Abstract Dstrbuted Generatons (s) are most commonly used n Advanced Power Dstrbuton System (APDS) for mprovng the voltage level and mnmzng the power loss n dstrbuton system. Ths paper represents a modfed Teachng Learnng Based Optmzaton Technque (TLBO) to reconfgure the dstrbuton system and fnd the optmal placement and szng of s to mnmze the total system loss smultaneously. In modfed TLBO radal trees are consder as a ntal populaton and possble paths between nodes are consdered as desgn varables for fndng the optmal reconfgured route. A shft operator s used to change the swtch poston n order to fnd the desred swtch combnaton. A smultaneous reconfguraton wth placement s done by the proposed method wth dfferent szes of s n search of optmal soluton. Proposed approach s llustrated on the standard IEEE-33 bus test system consderng real-tme desgn practces. Keywords Prmary power dstrbuton system, Reconfguraton, Teachng Learnng Based Optmzaton Technque, Dstrbuted Generaton. I. INTRODUCTION Network Reconfguraton, capactor placement, nstallaton are broadly used to desgn an advanced power dstrbuton system. Due to hgh densty of loads the amount of loss n dstrbuton system make a good mpact n the total system loss. Through a decade power system planners are reconfgured the dstrbuton system to decrease the system loss. Network reconfguraton problem stated as a combnatoral problem whch can be solved by the dfferent combnatons of swtches. By changng the on-off status of sectonalzng and te swtches the dstrbuton network reconfguraton can be done by satsfyng all the system constrants. Loss mnmzaton s not only the crtera for desgnng an APDS, there are few more lke voltage mprovement, system securty, load shftng etc. Therefore, n present condton the power system planners are tred to satsfy more than one obectve to make a cost effectve, relable mnmum loss system wth mproved voltage profle. The end voltage level of radally reconfgured system s not so mproved, therefore for the purpose of voltage mprovement and mnmum loss system; s are placed n the prmary dstrbuton network. To fnd optmal locaton and szng of s, planners use dfferent optmzaton technques. Heurstc methods lke Ant Colony Optmzaton(ACO)[1], Genetc Algorthm(GA)[6], Partcle swarm Optmzaton(PSO)[3],Seeker Optmzaton Algorthm(SOA)[4], Smulated Annealng(SA)[5], are used for solvng the plannng problem of power dstrbuton system. Reference [6] ntroduces a conventonal technque for optmal placement of n power system. Reference [7] shows the mpact of s on voltage profle, relablty and losses to desgn an APDS. An analytcal process s done by [8] to fnd the optmal locaton and szng of s to mnmze the system loss. Reference [9] proposes a metaheurstc method named Artfcal Bee Colony (ABC) and modfed t for placement wth network reconfguraton. Another meta-heurstc method called Harmony Search Algorthm (HSA) s used for smultaneous reconfguraton wth placement on IEEE-33 and 69 bus systems [10]. s work same lke a capactor but the only dfference between two s capactor only supply the reactve power whereas s supples both actve and reactve power. But must be placed on the optmal postons because desred output s totally changed (system loss wll be ncrease) wth the msplacement of s. Ths paper mplements a new verson of bt shft operator based Teachng Learnng Optmzaton algorthm (BS- TLBO) to desgn prmary power dstrbuton system smultaneously n the presence of s. The new verson of TLBO s manly based on locaton of on-off condton of swtches by changng the populaton n bnary array and ntroducng a shft operator such that t changes the poston of swtch to make a new populaton n both teacher and student phase to fnd optmal result of problem. Feasblty of the proposed technque s valdated on IEEE 33 bus system [10] and compared wth a well-known and effectve method HSA[1]. The man contrbuton of the paper s the development of a modfed verson of TLBO called bt-shft operator based TLBO for the frst tme n ths work for smultaneous reconfguraton wth placement on prmary power dstrbuton system. The organzatons of the paper are as follows: Problem model (Secton II), Problem formulaton (Secton III), Proposed methodology (Secton IV), Results and analyss (Secton V) and the Conclusons (Secton VI). II. PROBLEM MODEL Reconfguraton problem generally done on the radal network to optmze the path of power flow by optmzaton technque wthout volatng the system constrants. In mesh network there are many dfferent scenaros of power flow between two branches. A reconfguraton problem fnds the optmal path of power flow between two nodes. The

2 proposed methodology strategcally placed normally open (NO)/ normally closed (NC) swtches n all branches to represent the total system as a combnaton of one or zero. To solve a network reconfguraton problem numbers of parameters are to be handled, therefore for purpose of easy operaton bnary shftng of branches are to be done. By changng the status of swtches ('1' or '0'), the proposed method fnds the optmal soluton for desgnng a sutable PDS. Suppose a normal dstrbuton system havng u number of sectonalzng swtches and v number of te swtches, then the total swtches sss (sss = u+v) represents the radal dstrbuton system, that can be represented as [ sss 1, sss, sss3... sss n ]. The total swtch sequence s called as S''S''S'' n the rest of ths paper. The proposed methodology generates a number of radal networks whch are represented by number of bnary array. For example, the swtch sequence s represented as [ ] n Fg.1.(assumng '1' as a normally open swtch and '0' as a normally closed swtch), where [1,,3,4,5,7,8,9,10,11] are sectonalzng swtches and 6 s the te swtch. total 37 connectng branches n whch 3 are closed and 5 are open. So, the total system s represented as a sequence of swtch [sw1, sw1, sw3... sw37] or as a bnary array lke [ ]. III. PROBLEM FORMULATION In the proposed work the man goal of the optmzaton functon s to mnmze the total system real power loss whle satsfyng all type of constrants n the presence of unt. Fgure 4 shows the two bus system n the presence of a. Fg.1.Smple Radal Dstrbuton System Another radal system s shown by the swtch sequence [ ] n Fg.. Fg.3: Example of a prmary mesh dstrbuton system Fg. 4: Sngle lne dagram of two bus system wth a source. The mathematcal formulaton of the obectve functon s as follows: Fg..Smple Radal Dstrbuton System Fg.3 shows a graphcal representaton of small prmary meshed dstrbuton system. The system conssts of 33 nodes and 1 sub-staton. In ths paper, the proposed methodology turns the whole system nto a bnary array by representng each branch as a NO/NC swtch. Lke there s f = mn n = 0 P + Q V r + 1 (1) The proposed problem s subected to some equalty and non-equalty constrants such as:

3 Equalty Constrants [9]: ( P ) Q r P + λ P P 0 + P α P + Q Q 1 x QL + λqβp V V + 1 L = + p V ( ) Q = = V ( r + 1P + x+ 1Q ) + ( r x+ 1) Where, λ p ( P + Q ) Real power multpler. It s 1 f there s actve source n lne or 0 f there s no actve power source n lne. λ Q Reactve power multpler. It s ± 1 f there s reactve source n lne or 0 f no reactve source n lne. α At bus +1, the total actve component of power. P +1 β P +1 At bus +1, the total nected reactve power. Inequalty Constrants [9]: a) Voltage lmt: System voltage must be n the lmt of ± 5% of the nomnal voltage value spec sys spec Vmn V Vmax = 0,1,..., n(5) b) Thermal capacty lmts sys rated sys S, + 1 S, + 1 s + 1,1 = 0,1,..., n c) unt sze and power factor lmts φ S mn mn where S ϕ ϕ S ϕ S max max sze of a power factor of a The practcal concerns about the sze and power factor are consdered. The pre-defned unt szes are from 30-40% of total system demand. The power factor sets to operate at 0.85, 0.9, 0.95 and unty and may be dfferent to the bus load power factor where the unt s placed. IV. PROPOSED METHODOLOGY A. Basc Teachng Learnng Based Optmzaton algorthm: Teachng learnng based optmzaton technque manly based on unversal phenomena of transfer of knowledge between teacher and learners. In a classroom the teacher s most knowledgeable person and the students are enhancng ther knowledge by two ways. One of them s the experence of student by sharng the knowledge wth the teacher and another one s by mutually co-operaton between themselves means one student share hs/her knowledge wth the all V (6) (7) (8) () (3) (4) other students. As metaheurstc technque TLBO also depend upon the ntal populaton where each student s taken as a one populaton. The subects taught n the classroom are taken as the desgn varables of TLBO. The exstng soluton n terms of knowledge s updated n teacher phase + D N, = O, where, D M s the dfference between the result of teacher and the mean result of whole class [13]. D = r( M T * M ) (10) M N F Student phase s the second stage of the TLBO algorthm where student enhances ther knowledge by mutual collaboraton. In ths stage the exstng soluton s agan updated by the equaton = + r ( ) f ) < f ( ) (11) N, O, M ( = + r ( ) f ) > f ( ) (1) N, O, ( The maor part of a typcal TLBO s dscussed n [13]. (9) B. Modfed Teachng Learnng based optmzaton Technque: An example of swtch representaton of prmary network s shown n the problem model where normally closed (NC) swtches are fndng the route of optmal power flow. Some number of bnary arrays s chosen as ntal populaton n the modfed TLBO technque. A shft operator named bt shft drver s used to change the normally open swtch poston to update the populatons. 1) Bt shft drver Bt shft drver shft the poston of swtches and gve a new path for power flows. The shft operator shfts the swtches ether left or rght.e. n both drectons. Also, the poston of shftng or number of bt shftng can be controlled by ths operator. The drecton of operaton and the steps of shftng are dependent on the equaton of the basc TLBO. The operators operate on the both teacher and student phase of TLBO and update the exstng populaton. Fg.5 shows the general bt shft operaton of bt shft drver. The frst case shows the rght shft where the all swtches shft two postons at the rght drecton and the second case shows a left shft where all swtches shft one poston left. Lke ths the operator shfts the swtches left or rght at any postons.

4 Fg. 7(a) and (b) shows the dfferent radal topologes on the bass of dfferent bnary array made by bt shft drver. Lkewse, there are many more dfferent trees are generated by bt shft drver whch works on the bass of TLBO equatons. ) Constructng a Modfed TLBO: Fgure 5: Basc operaton of Bt Shft Drver A smple example s taken to descrbe the operaton of the bt shft drver n reconfguraton problem. A dstrbuton system shown n Fg. 6(a) have 7 nodes and 10 connectng paths of power flow. Proposed methodology make a radal network n Fg. 6(b) and change t n a bnary array [ ] The formulas of both teacher phase and student phase for TLBO are updated based upon the bt shft drver to make a modfed verson of TLBO. The redefned equaton of the teacher phase s modfed as = Θ D ' (13) wher,e N, O, M,, N O, and DM can be defned from the basc TLBO [13]. The operator Θ s the bt shft drver used to update the prevous populaton to a new one. DM s also D ' M modfed to for the purpose of modfcaton, n modfed TLBO as (a) D ' = r sgn( M T * M ) M N F (14) (b) Fgure 6: A 7 node dstrbuton system (a) and the radal topology (b) The bt shft operator changes the swtch poston at the left or rght drecton on the bass of shftng. If a left shft s occur the radal topology of Fg.6(b) s changed to [ ].e. Fg.7(a) or f rght shft occurs than the radal topology changes to [ ].e. Fg.7(b). (a) (b) Fgure 7: Radal topology for left shft (a) and rght shft(b) Here, the sgn functon s used to express the value of dfference as {-1,0,1} for the bnary shft operaton or help to choose the drecton and poston of bt operaton. ' Lkewse, f the value of D M s -1 then left shft occurs f the value s +1 then rght shft occur or no shft wll occur f the value s 0. The student phase equaton s also changed, t s redefned as: = O, Θ r sgn ( ) f ( ) < f ( ) (15) N, = Θr sgn ( ) f ) > f ( ) (16) N, O, ( Here, two bnary arrays whch s expressed as two students are subtracted as bt wse. The sgn functon converts the bt wse subtracton n bnary. After that proposed methodology make a summaton to fnd the total value of bt wse subtracton for the shft operaton. Lke f the summaton value become {...,-,-1,0,1,,3...} then the sgn represent the drecton and the value represent the poston of shftng. For the value of the bt shft operaton shows n Fg.5 (frst case). 3) Applcaton of Modfed TLBO for the sad problem: The modfed verson of TLBO not only mnmzes the total loss but also fnd the optmal locaton and szng of s. The steps of modfed TLBO wth presence of s are shown n fg 8.

5 IV. RESULTS AND ANALYSIS The proposed algorthm s appled on an IEEE 33-bus test system whch s shown n Fg. 9. The test system conssts of 37 branches out of whch 3 branches are sectonalzng lnes and rests 5 are te lnes. The system contans a sngle substaton at node 1 and rest nodes are energzed through ths substaton only. The total real and reactve power loads on the system are 3715 KW and 300 KVAR. The load s assumed to be constant and the base power s consdered as 100 MVA and base voltage as 1.66 KV. The bt shft operator based TLBO have been developed and run for 0 ndependent runs where each ndependent run consst of 0 populatons and 0 teratons. Modfed verson of TLBO have no modal parameters. The teachng factor (TF) value s chosen randomly between one and two. The proposed methodology fnds the optmal locaton of wth mnmzng the loss. The optmal soluton havng mnmum loss of 73.98KW wth 3 placement at the nodes 17,11,13 respectvely. Fgure 10 shows the optmal reconfgured tree wth optmal locatons. Fgure 9: Test Case Fgure 8: Flowchart of Modfed TLBO Fgure 10: Optmal reconfgured network wth placement

6 TABLE I :PERFORMANCE COMPARISON OF MODIFIED TLBO WITH OTHER TECHNIQUES Modfed TLBO Solutons Base case (scenaro1) Only Reconfguraton (scenaro) Reconfguraton wth smultaneous nstallaton (scenaro3) Swtchs opened 33,34,35,36,37 7,11,34,36,37 7,11,8,34,36 Mnmum Loss(KW) Sze of n MW(Bus Number) (17) (11) (13) % Loss Reducton Mnmum voltage(n Pu) HSA [10] Base case Only Reconfgurat on Reconfguraton wth smultaneous nstallaton Swtchs opened 33,34,35,36,37 7,14,9,3,37 7,14,10,3,8 Mnmum Loss(KW) Sze of n MW(Bus Number) (3) (31) (33) % Loss Reducton Mnmum voltage(n Pu) The proposed methodology s compared wth another technque called HAS [10] n Table I. The proposed methodology also mprove the voltage profle of the system by allocaton. Fgure 11 shows the voltage mprovement of the prmary network done by the proposed methodology. prmary dstrbuton system and fnd the optmal placement and szng of s smultaneously wth an obectve of mnmzng the total system power loss. Extensve tests are carred out n order to fnd the effcacy of the proposed algorthm. The test results obtaned clearly shows the effectveness of the proposed algorthm. Nonetheless, the technque hghlghted n ths work s effectve and useful to the planners for co-ordnated plannng by combnng reconfguraton wth placement smultaneously. REFERENCES [1] J. F. Gómez, H. M. Khodr,P. M. De Olvera,L. Ocque and J. M. Yusta, Ant Colony System Algorthm for the plannng of prmary dstrbuton network, IEEE Trans. On Power Systems,vol. 19,no, pp , May 004. [] I. J. Ramírez-Rosado and J. L. Bernal-Agustín, Genetc Algorthm Appled to Desgn of Large Power Dstrbuton System, IEEE Trans. On Power Systems,vol. 13,no, pp , May [3] S. Ganguly, N. C. Sahoo, and D. Das, Mult-obectve plannng of electrcal dstrbuton systems ncorporatng sectonalzng swtches and telnes usng partcle swarm optmzaton, Swarm Evol Comput., Vol. 3, pp. 15 3, 01. [4] D.Kumar, S. R. Samantaray, I.Kamwa, Mult-obectve desgn of advance power dstrbuton networks usng restrcted-populaton based multy obectve seeker optmsaton-algorthm and fuzzy operator, IET Gener. Transm.Dstrb., vol. 9,no. 11, pp , 015. [5] K.Nara and M.Ktagawa, Dstrbuton system loss mnmum reconfguraton by smulated annealng method, Proc. IEE Internatonal Conference on Advances n Power System Control, Operaton and Management, Hong Kong,, pp , Nov [6] W. Rosehart and E. Nowck, Optmal placement of dstrbuted generaton, n Proc. 14th Power Systems Computaton Conf., Sevllla, pp. 1 5, Secton 11, paper, 00. [7] C. L. T. Borges, D. M. Falcao, Impact of Dstrbuted Generaton Allocaton and Szng on Relablty, Losses and Voltage Profle, IEEE Power Tech Conference Proceedngs, vol., 003. [8] C. Wang and M. H. Nehrr, Analytcal approaches for optmal placement of dstrbuted generaton sources n power systems, IEEE Trans. Power Syst., vol. 19, no. 4, pp , Nov [9]F. S. Abu-Mout, and M. E. El-Hawary. "Modfed artfcal bee colony algorthm for optmal dstrbuted generaton szng and allocaton n dstrbuton systems." In Electrcal Power & Energy Conference (EPEC), 009 IEEE, pp. 1-9, 009. [10] R. Srnvasa Rao, Kumudhn Ravndra, K. Satsh, and S. V. L. Narasmham. "Power loss mnmzaton n dstrbuton system usng network reconfguraton n the presence of dstrbuted generaton." IEEE Transactons on Power Systems, vol. 8, no. 1, pp , 013. [11] D. Kumar, and S.R. Samantaray, Desgn of an advance electrc power dstrbuton systems usng seeker optmzaton algorthm, n Internatonal Journal of Elect. Power and Energy Syst., vol. 63, pp , 014. [1] Rung-Fang Cang,, Ya-Chn Chang, and Chan-Nan Lu. "Feeder reconfguraton for accommodatng dstrbuted generatons nterconnecton." In Intellgent System Applcaton to Power Systems (ISAP), th Internatonal Conference on, pp. 1-6, 011. [13] R.Venkata Rao, Vmal J. Savsan, and D. P. Vakhara. "Teachng learnng-based optmzaton: an optmzaton method for contnuous nonlnear large scale problems." Informaton Scences 183, no. 1, pp.1-15, 003. Fgure 11: voltage profle of IEEE 33 bus test system VI. CONCLUSIONS The proposed research work have developed a new approach called bt shft operator based modfed Teachng Learnng Based Optmzaton Technque for reconfgurng the