Distributed generation for minimization of power losses in distribution systems

Transcription

1 Unversty of Wollongong Research Onlne Faculty of ngneerng and nformaton Scences - Papers: Part A Faculty of ngneerng and nformaton Scences 006 Dstrbuted generaton for mnmzaton of power losses n dstrbuton systems Kashem M. Muttaq Unversty of asmana, kashem@uow.edu.au An D. Le Unversty of asmana, dtale@utas.edu.au Mchael egnevtsky Unversty of asmana, mchael.negnevtsky@utas.edu.au Gerard Ledwch Queensland Unversty of echnology, g.ledwch@qut.edu.au Publcaton Detals K. A. Kashem, A. D.. Le, M. egnevtsky & G. Ledwch, "Dstrbuted generaton for mnmzaton of power losses n dstrbuton systems," n Power ngneerng Socety General Meetng, 006, pp. -8. Research Onlne s the open access nsttutonal repostory for the Unversty of Wollongong. For further nformaton contact the UOW Lbrary: research-pubs@uow.edu.au

2 Dstrbuted generaton for mnmzaton of power losses n dstrbuton systems Abstract Approprate sze and locaton of dstrbuted generaton () play a sgnfcant role n mnmzng power losses n dstrbuton systems. hs paper represents technques to mnmze power losses n a dstrbuton feeder by optmzng model n terms of sze, locaton and operatng pont of. Senstvty analyss for power losses n terms of sze and operatng pont has been performed. he proposed senstvty ndces can ndcate the changes n power losses wth respect to current njecton. he proposed technques have been developed wth consderng load characterstcs and representng loads wth constant mpedance and constant current models, separately. he optmal sze and locaton of n a dstrbuton feeder can be obtaned through the developed technques, wth mnmum effort. he proposed technques have been tested on a practcal long radal system and results are reported. est results have proven that up to eghty-sx percent of real power loss can be reduced wth a of optmal sze, located at optmal place n the feeder. Keywords power, mnmzaton, generaton, systems, dstrbuted, dstrbuton, losses Dscplnes ngneerng Scence and echnology Studes Publcaton Detals K. A. Kashem, A. D.. Le, M. egnevtsky & G. Ledwch, "Dstrbuted generaton for mnmzaton of power losses n dstrbuton systems," n Power ngneerng Socety General Meetng, 006, pp. -8. hs conference paper s avalable at Research Onlne:

3 See dscussons, stats, and author profles for ths publcaton at: Mnmsng oltage Devaton n Dstrbuton Feeders by Optmsng Sze and Locaton of Dstrbuted Generaton ARCL Source: OA CAOS 7 DOWLOADS 3 WS 98 4 AUHORS, CLUDG: Kashem M. Muttaq Unversty of Wollongong Gerard Ledwch Queensland Unversty of echnology 0 PUBLCAOS,03 CAOS 48 PUBLCAOS 3,484 CAOS S PROFL S PROFL Mchael egnevtsky Unversty of asmana 95 PUBLCAOS,03 CAOS S PROFL Avalable from: Mchael egnevtsky Retreved on: 3 September 05

4 Mnmsng oltage Devaton n Dstrbuton Feeders by Optmsng Sze and Locaton of Dstrbuted Generaton An D.. Le, M.A. Kashem, M. egnevtsky School of ngneerng Unversty of asmana Sandy Bay, asmana, Australa dtale@utas.edu.au G. Ledwch lectrcal and lectronc Systems ngneerng Queensland Unversty of echnology Brsbane, Australa g.ledwch@qut.edu.au ABSRAC A new emergng trend of dstrbuton networks s to use small generatng unts, known as dstrbuted generaton (), operatng n parallel wth the man grd. hs knd of dstrbuton networks has enabled to support power systems n fulfllng ther requrements to ncrease power output as well as qualty of power supply. has potental to alter power flows, system voltages, and the system performance. n order to maxmse benefts from the system, proper plannng s necessary. Determnng an optmal sze and ts locaton are crtcal ssues that are addressed n ths paper. he man purpose of ths research s to maxmse voltage support through optmal szng and locaton of. A new methodology s developed to determne an optmal sze for a certan penetraton and an optmal locaton on the dstrbuton feeder for optmsng system voltages. he developed technque s tested on a long radal feeder of a practcal system and results are reported. ndex erms -- Dstrbuted Generaton (), Dstrbuton System, Optmal Sze, Optmal Locaton, and oltage Devaton.. RODUCO Utlsng of Dstrbuted Generaton () to produce electrcty has become an ncreasngly attractve choce for both utlty and customers. radtonal optons of power utltes to compensate the rapd growth n electrcty demand are transmsson expanson, substaton capacty upgrade and/or ntegraton []. Among these optons, appears to be the most perspectve one. t does not only releve the burden of supplyng loads from dstrbuton system, but also satsfes the customer s requrements of relable and contnuous power supply, as well as an avalablty of nstantaneous electrcty sources when power nterruptons occur. Moreover, together wth the ongong efforts to reduce captal nvestments and operatng cost of, t s beleved that can potentally become one of the most effectve-cost solutons. n fact, studes have revealed that the electrcty generated by may account for up to 0% of all new generaton by the year 00 []. For decades, small generaton has been used as a backup or stand-by power source to supply electrcty for small personal customers durng grd power outages. he most common type of for ths purpose s desel generaton. owadays, the recent advances n technologes have made ths power soluton possble not only to serve ndvdual customers but also support the entre network n parallel wth the grd. technologes can be categorsed nto two groups: () non-renewable energy technologes and () renewable energy technologes. he frst group conssts of nternal combuston engnes, gas turbnes, mcro turbnes, etc. he second group produces electrcty usng renewable energy sources,.e solar energy, wnd energy, tdal energy, wave energy, geothermal energy, bo-energy, etc. Although has relatvely small sze compared wth central generaton, t s large enough to satsfy electrcty requrements of a group of local customers. Conventonal, purpose of dstrbuton systems s to dstrbute power to the customers. hese customers are desgned to operate as passve network elements and do not generate any power [3]. However, the current trend of ntroducng nto dstrbuton systems makes customers no longer passve they become rather actve. Possbltes of postve mpacts of nclude voltage profle mprovement, system loss reducton, system stablty and relablty mprovement, etc. Among all key ssues, the choce of the sze and locaton s of a great mportance and s addressed n ths paper. szng and sttng problems have been studed by many researchers recently. n [4], a heurstc cost-beneft approach s developed to defne the compettve sze and locaton based on the mnmsaton of nvestment and operatng costs of s. A technque to decde szng and sttng has been proposed n [5], whch allows dstrbuton system planners to nclude an optmal sze of at an optmal locaton of ther desgn. Authors n [6] have developed a new ntegrated dstrbuton system plannng model based on cost optmsaton. Problems related to sze and locaton have been dscussed n [7]. Genetc Algorthms and Decson ree heory are appled n [7] to determne stng and szng for a medum voltage dstrbuton network. n [8], the sze and locaton are selected based on the comparson between the performance of several avalable s on system losses and voltage drop. Authors n [9] have establshed a new method to determne the locaton to mnmse the voltage drop and the number of supervsed. n ths paper, an optmal locaton and ts sze are determned to maxmse mprovements of the voltage profle n dstrbuton systems. A technque has been developed based on the voltage devaton and

5 penetraton n the networks to obtan an optmal sze and locaton of for any type of dstrbuton systems.. MPROM OF WORK CODOS B CURR JCO FROM An applcaton of can mprove the voltage profle of a power system, especally on tradtonal dstrbuton networks (whch are usually radal), where voltages are very close to ther lower lmts at buses close to remote ends. One of the advantages of the usage s that can nject both actve and reactve power of any combnaton to mprove system condtons and satsfy customer s demands nstantaneously. Let us assume that dstrbuton system conssts of load buses, where the bus number refers to the source bus and the bus number refers to the remote bus at the end of the feeder. A wll be connected to the feeder and the locaton may be vared to determne ts optmal locaton. he ntegraton ntroduces an nternal bus, +, n the system. he can move along the feeder wth ths nternal bus, wthout changng the arrangement of the other buses n the system. A current of + s njected nto the system from the. he resultng voltage changes at all buses n the network due to ths current njecton are calculated as: bus () where bus s the mpedance matrx for (+) bus system; s the vector of voltage changes for bus to +; s the vector of current changes for bus to +. By expandng q.() nto the matrx form, we obtan: +,,,,,,, +,,,, +,, +, +, + +, () hen by parttonng q.() nto sub-matrces, we can obtan: A B C B D (3) C F + + where, A, B [ ] D,,,, and,, C,+, + F +,+, + Because n ths test system, the source s connected at bus and the at bus +, we have 0, S, + and + When we substtute those values nto q.(3), we obtan: A B C B D C S S 0 F From q.(4), we obtan: ( B S ) ( + ) (4) + (5) F D C S (6) By substtutng from q.(5) nto q.(6), we get: S D ( ) + + F C oltage at the utlty s kept constant, thus there s no change n S. n other words, S 0. Substtutng S 0 nto q. (7) and rearrangng the equaton, we obtan: ( F D ) B S (7) (8) hen by substtutng q.(8) nto q.(5), we get ( ) F D (9) D he voltage changes from bus to bus of the system due to the current njecton by the can be computed usng q.(9). q.(9) can be rewrtten n matrx form as: a (0) where a s the coeffcent matrx of sze -, used for calculatng voltage changes at bus through bus. An ntroducton of nto the system results n new system s voltages. he new voltages can be obtaned by superposton as follows: a + a S 0 where and wth, respectvely. () are voltages of the system wthout and 3. MAMSG OLAG MPROM B Modern electrcal equpment s desgned to operate wthn a specfc range of voltages. qupment of both utlty and customers can only tolerate the fluctuatons of voltage n a small perod of tme. f the voltage remans unstable for a longer tme, the equpment has a hgh probablty to get damaged. he acceptable range of

6 voltages vares dependng on the regulaton complance of dfferent regons n dfferent countres. n asmana, for example, the lmts of voltage varatons are ± 6 percent n normal operatng condtons. he voltage varatons may occur n dstrbuton systems because of many dfferent reasons. Lne mpedances cause a sgnfcant drop n voltage. Moreover, when the avalable reactve generaton cannot meet the growng demand for reactve power at customer s sdes, a voltage drop may occur n the system. Also, for long radal feeders, whch are very common n rural areas, the transmsson of reactve power may not be possble and therefore voltage drop wll also be ncreased at the customer s connecton ponts of loads. herefore, the voltages of load buses at the remote ends are usually lower than the voltages of load buses close to the utlty substaton. oltage devaton can be defned as the dfference between the nomnal voltage and the actual voltage. he smaller the devaton of bus voltage from the nomnal voltage, the better the voltage condton of the system. A voltage devaton ndex (D) as defned as the sum of the squared value of the absolute voltage dfference between the nomnal voltage and the actual voltage for all buses n the system: D n () where s the total number of buses, n s the nomnal voltage, and s the actual voltage at bus. 4. DSCRPO OF S SSM A part of the actual practcal dstrbuton system shown n Fg. s selected as a test system to determne an optmal sze and locaton of the. he test system s derved from a dstrbuton network of Aurora nergy, a dstrbuton utlty of asmana, whch conssts of a long radal feeder connected to Smthton substaton. Smthton substaton s one of response centres n asmana. t s located on the orth-west of asmana. he substaton has two ncomng feeders of 0 k and supples fve -k long radal feeders. hese are Woolnorth (48 km), Arthur Rver (64 km), dth Creek (4 km), Roger Rver (8 km), and Smthton ownshp (300m). Fg. shows one-lne dagram of the substaton. Fgure : Smthton substaton s one-lne dagram WOOLORH Dstrbuton feeders he ncluson nto the system can mprove voltage profle of the system and reduce the voltage devaton. Let us assume that a s ncluded n the system, the new voltage devaton ndex (D ) wll then become: By replacng D by ( ) D By substtutng the values of q.(4), we obtan: D n (3) +, we get: n n ( + ) (4) from q.() nto ( + a ) (5) he change n the voltage devaton ndex of the system due to the njecton can be calculated by subtractng q.(5) from q.(), and obtaned as D [ n n ( + a ) ] (6) he should be desgned so that t wll mnmse the voltage devaton n the system. As can be seen from q.(6), the voltage mprovement can be acheved by njectng current from the. he maxmum voltage mprovement can be obtaned by determnng an optmal value of the current njecton. SMHO Fgure : Smthton - Woolnorth test feeder he 48-km radal feeder from Smthton substaton to Woolnorth shown n Fg. s chosen as a test feeder to valdate the developed technque. he test feeder conssts of load buses. he lne mpedance of the feeder s Z l j Ω/km. omnal substaton voltage S s k and hevenn source mpedance Z S s j.680 Ω. he total load of the feeder s MA. For smplfcaton, we assume that the feeder has unformly dstrbuted loads. 5. SMULAO RSULS AD DSCUSSOS Smulatons are conducted n MALAB 7.0 wth a ntegrated n the test system to nvestgate an mprovement n the voltage profle of the system. he MA base of MA and voltage base of k are used n the smulaton. he s moved along the feeder to determne ts optmal locaton. Also, the level of the penetraton s ncreased from mnmum to maxmum to obtan an optmal sze of the. he value of the voltage devaton ndex of the system wthout the was t s known that when the system has hgh R/ rato of lne, t requres mostly real power for

7 voltage mprovement [0]. For the test system, the rato of R/ s.78. herefore, for smplcty, we can chose only real power njectons from the. hus, the current njecton from the has the same phase as the phase of the local voltage at the connecton pont. 5.. FFCS OF H LOCAO O H SSM OLAG Prevous studes have proven that the locaton s one of the most mportant ssues n dstrbuton system plannng. A proper locaton of the wll maxmse ts potental support for mprovng the system voltage profle. Smulatons are carred out to determne the locaton, whch provdes the mnmum voltage devaton ndex for a requred level of the penetraton. Consder a that can support the full load n the system. Such stuaton s defned as the 00% penetraton. Fg.3 shows the voltage devaton ndex of the system. he ndex value s calculated wth the placed at each load bus, one at a tme. lowest value, and therefore more voltage support s needed. When the level of penetraton s ncreased, an optmal locaton s moved toward the mddle of the feeder. hs reveals that when the penetraton s hgh, t s more effectve to place the away from the remote end, so that all load buses n the system can beneft from the. 5.. FFC OF SZ O SSM OLAG Dfferent szes provde dfferent levels of voltage support to the system. Our purpose s to determne the proper sze to maxmse the system voltage mprovement, whch s evaluated by voltage devaton ndex. Smulatons are carred out va ncreasng the sze at dfferent locatons of the system. Fg.4 examnes how the system has responded to varyng the sze. Load buses selected to test the performance of are buses 68, 6, 54, 47 and 40. From ths graph we can see that, the changes of the system voltage devaton ndex due to the changes of the penetraton level has a parabolc trend. As the output of the current ncreases, the voltage devaton ndex decreases. However, t wll start to ncrease after a certan level of the penetraton. he pont where the voltage devaton obtans ts mnmum value s defned as the optmal sze for maxmum voltage mprovement. Also, the mportance of locaton s clearly llustrated n Fg.4 dfferent locatons gve dfferent optmal szes. Fgure 3: oltage devaton wth the 00% penetraton he mnmum value of the voltage devaton ndex of the system s found to be , whch corresponds to the located at bus 35. Smulatons are also carred out wth dfferent levels of the penetraton. Optmal locatons for dfferent penetratons are presented n able. able : Optmal locaton for dfferent levels of penetraton penetraton Optmal locaton Dstance (km) 0% 0% 30% 40% 50% 60% 70% 80% 90% 00% he results show that wth the smaller penetraton, system voltages are mproved the most when the s located at the end of the feeder. hs can be explaned by the fact that the voltage at the remote load bus has the Fgure 4: Changes of voltage devaton wth ncreasng penetraton he mnmum voltage devaton ndex can also be obtaned for dfferent locatons and dfferent penetraton. he results are shown n Fg.5. he threedmensonal graph here represents the mnmum voltage devaton ndex of the system, correspondng to the respectve penetraton and locaton. From Fg.5, we observe that the mprovement of the system voltages does not only depend on the sze, but also on the locaton. Maxmum voltage support by the s acheved wth the 80% penetraton at bus 45, whch s the optmal locaton. q.(6) s used to determne an optmal sze at a specfed locaton. Results are shown n able. Snce

8 the results obtaned n the prevous part ndcate that the operates more effectvely when t s located at a load bus close to the end of the feeder, we only consder buses from 40 to. located closer to the remote end, the optmal szng s decreasng. he proposed method can also be used to defne an optmal sze for a specfc locaton. n realty, a fxed locaton s a common practce, especally wth those s that requre large space for nstallaton or need to be sted near the energy resources. Fgure 5: Mnmum voltage devaton ndex at dfferent levels of the penetraton Bus able : Optmal sze for specfed buses Dstance (km) Mn D D D, max From able, we can see that the mnmum voltage devaton ndex (D) or the maxmum voltage devaton reducton ( D) s obtaned when the s located at bus 45 wth the output current of.53 p.u., whch corresponds to the approxmately 80% penetraton. t can also be noted that the at dfferent stes on the feeder gves dfferent levels of voltage support to the system (local maxmum values). Fg.6 shows how the locaton affects the voltage devaton ndex and the voltage devaton reducton of the system. An optmal sze s changng wth the changes n locaton, whch s shown n Fg.7. When the s Fgure 6: oltage devaton ndex and voltage devaton reducton wth located at dfferent buses Fgure 7: Optmal szes for dfferent locatons Fg.8 shows the voltage profle of the feeder wth and wthout the. System wthout has the voltage profle of a typcal radal system, where the voltage s decreasng along the feeder. he voltages at some downstream load buses drop below the acceptable range of 0.94 p.u. he lowest voltage of 0.96 p.u. occurs at the remote load bus. However, the voltage profle of the system wth shows a sgnfcant mprovement n voltage. he voltages of all sectons n the feeder are wthn the voltage lmt. Lowest voltage for the system wth ntegraton s 0.99 p.u. Fg.9 llustrates the reducton n voltage devaton at each load bus n the system. he has effectvely reduced the voltage devatons and keeps the voltages of all load buses close to the nomnal value. t s desrable to keep voltages at all sectons as close to.0 p.u. as possble. However, such approach can only be used when the cost of s s not an ssue. he cost can be consderably reduced by operatng the network wth acceptable voltage range of ( ± 0.06 p.u.).

9 Dependng on the requrements of the utlty and customers, dfferent levels of lower lmt of the voltage may be requred. able 3 provdes optmal szes and locatons of the for dfferent reference voltages. he sze s reduced when the reference voltage s reduced. he last column n able 3 shows the placement as a rato of the dstance of the from the substaton to the total feeder length. able 3: szes and locatons for dfferent reference voltage levels Reference sze locaton voltage ka sze Penetraton Bus Dstance Placed at 0.94 p.u % km 97.% 0.95 p.u % km 9.5% 0.96 p.u % km 86.7% 0.97 p.u % km 8.% 0.98 p.u % km 76.0% 0.99 p.u % km 68.8% usng the proposed technque, the voltage devaton can be mproved consderably for any dstrbuton systems. Also, a new method to compute the voltage mprovement at each load bus n the system wth s ntroduced. Smulatons are carred out wth a practcal system to verfy the valdty of the proposed method. Optmal szes for mnmsng the system voltage devatons are computed for dfferent canddate locatons, and thus the optmal locaton can be selected. he outcome of ths research helps not only to determne the optmal sttng and szng n the system, but also to determne the optmal szng for a specfc locaton of. 7. ACKOWLMS hs research has been funded by the Australan Research Councl under ARC Lnkage Grant K0043 ntegraton of Dstrbuted and Renewable Power Generaton nto lectrcty Grd Systems. he authors also would lke to thank Aurora nergy personnel for provdng data for case studes. RFRCS Fgure 8: oltage profle of system wth and wthout Fgure 9: he voltage devaton for each load bus n the system before and after ncluson 6. COCLUSOS n ths paper, mprovement of the system s voltage profle by utlsaton of s s dscussed. A novel approach has been developed to determne an optmal sze and a locaton to maxmse the voltage support n dstrbuton systems. oltage varatons of the system are evaluated by the voltage devaton ndex. By [] R.. Brown, Pan Jupng, Feng aomng, and K. Koutlev, Stng Dstrbuted Generaton to Defer &D xpanson, ransmsson and Dstrbuton Conference and xposton, 00, /PS, 8 Oct.- ov. 00, ol., pp [] P.P. Barker, and R.W. de Mello, Determnng the mpact of Dstrbuted Generaton on Power Systems: Part - Radal Dstrbuton Systems, 000 Power ngneerng Socety Summer Meetng, Seattle, Washngton, 000, pp [3] S.K. Salman, he mpact of mbedded Generaton on oltage Regulaton and Losses of Dstrbuton etworks, Colloquum on the mpact of mbedded Generaton on Dstrbuton etworks (Dgest o. 996/94), London, 5 Oct. 996, pp. / - /5. [4] W. l-khattam, K. Bhattacharya,. Hegazy, and M.M.A. Salama, Optmal nvestment plannng for dstrbuted generaton n a compettve electrcty market, ransactons on Power Systems, Aug. 004, ol. 9, ssue 3, pp [5] G. Cell,. Ghan, S. Mocc, and F. Plo, A multobjectve evolutonary algorthm for the szng and stng of dstrbuted generaton, ransactons on Power Systems, May 005, ol. 0, ssue, pp [6] W. l-khattam,.g. Hegazy, and M.M.A. Salama, An ntegrated dstrbuted generaton optmzaton model for dstrbuton system plannng, ransactons on Power Systems, May 005, ol. 0, ssue, pp [7] G. Carpnell, G. Cell, F. Plo, and A. Russo, Dstrbuted generaton stng and szng under uncertanty, 00 Porto Power ech Conference, Porto, 0-3 Sept. 00, ol. 4. [8] C.L.. Borges, and D.M. Falcao, mpact of dstrbuted generaton allocaton and szng on relablty, losses and voltage profle, 003 Bologna Power ech Conference, Bologna, taly, 3-6 June 003, ol.. [9] R. Care,. Retere,. Morn, M. Fontela, and. Hadjsad, oltage management of dstrbuted generaton n dstrbuton networks, Power ngneerng Socety General Meetng, 003,, 3-7 July 003, ol., pp [0] M.A. Kashem, and G. Ledwch, Dstrbuted generaton as oltage support for sngle wre arth return systems, ransactons on Power Delvery, July 004, ol. 9, ssue 3, pp