HARMONIC INTERACTIONS AND RESONANCE PROBLEMS IN LARGE SCALE LV NETWORKS

Transcription

1 HARMONIC INTERACTIONS AND RESONANCE PROBLEMS IN LARGE SCALE LV NETWORKS M. C. Benhabb, P. R. Wlczek, J. M. A. Myrzk, J. L. Duarte Department of electrcal engneerng, Endhoven Unverty of Technology Den Dolech, CR.1, P.O. Box 513, 5 MB Endhoven, The Netherland Tel.: , Fax: m.c.benhabb@tue.nl Abtract The complexty of lowvoltage (LV) network ncreae wth the ncreang number of connected dpered generaton, uch a photovoltac ytem, and nonlnear load, lke energy avng lamp and peronal computer. Due to the huge number of connected equpment, an accurate analy of LV network dffcult to be realzed, n partcular for tudyng problem related to reonance or to harmonc nteracton generated by the nonlnear load and/or dpered generaton. The man dffculty the necety of carryng out a detaled model of each nonlnear load and/or dpered generaton n order to ntroduce all harmonc a preent n the real cae. Th not traghtforward becaue exceve computer memory and calculaton tme are requred n vew of the large number of equpment n a LV network. On the other hand, f the model of each nonlnear load too mplfed, lo of nformaton concernng the harmonc pectrum unavodable. So, th paper preent a method whch make t poble to olve memory problem n order to be able to tudy power qualty ue (reonance effect and harmonc nteracton) n LV network. The approach lead to mlar reult of a nonlnear modelng decrpton, nce t baed on data from detaled model. Smulaton and expermental reult are gven to demontrate reonance problem and harmonc nteracton n a network that contan houe. Furthermore, an example of mulaton tudy alo preented for a complex network that contan more than 5 houe. Keyword: reonance, harmonc nteracton, equvalent reduced model. 1 INTRODUCTION In general, when tryng to tudy the mpact of nonlnear load and dpered generaton (DG) uch a photovoltac (PV) ytem n LV dtrbuton network, t dffcult to obtan repreentatve mulaton reult that allow analyzng, for example, the nfluence of harmonc current and ther nteracton wth harmonc voltage and network mpedance. The reaon that t neceary to take n conderaton a detaled model for each pece of nonlnear equpment connected to the network. However, the major dffculty when tryng to mulate a large quantty of detaled model manly computer memory lmtaton. To cope wth memory problem the nonlnear model mght be mplfed by ung technque lke average modellng, load flow technque and tate pace theory. But, the fundamental lmtaton of thee method that they can not decrbe the eental nformaton related to the harmonc n the current and/or voltage generated by the nonlnear equpment. An alternatve modellng approach propoed n th paper, whch allow the ntroducton of enough detal n the mulaton wthout requrng exceve computer memory. The reultng model are valdated by comparon of mulaton and expermental reult. It demontrated that the propoed modellng technque can be ued to mulate qute accurately the nteracton between harmonc current generated by a large number of nonlnear load and dpered generaton. Smulaton and expermental reult are gven to valdate the approach. Alo a mulaton tudy preented for a complex network whch contan more than 5 houe wth nonlnear load. PRESENTATION OF THE MODEL In th ecton the ratonale that lead to the contructon of mplfed equvalent model of nonlnear equpment (nonlnear load or dpered generaton) connected to the network preented by mean of two example. The model are valdated by meaured reult from experment..1 Nonlnear load (threephae dode rectfer example) A detaled Smulnk model of a threephae rectfer hown n Fg. 1. The mulaton reult wth th model decrpton for the current and voltage on the grd termnal are hown n Fg.. 1th PSCC, Glagow, Scotland, July 118, 8 Page 1

2 8 va vb vc A a B b C c Three Phae VI Meaurement 1.9 A B C Thyrtor Converter Id Scope 11 8 Fgure 1: Threephae ACDC dode brdge a b Fgure : Threephae current (a) and threephae voltage (b) aborbed by the threephae ACDC Dode Brdge (detaled model) In Fg. 3 the detaled rectfer model hown n Fg. 1 replaced by current ource wth output data n tabular format, whoe value at each mulaton tep are the ame a the one that have been obtaned from the prevou detaled mulaton. conv 1.mat a b Fgure : Threephae current (a) and threephae voltage (b) aborbed by the threephae ACDC dode brdge (mplfed equvalent model) By comparng Fg. to Fg. t can be clearly een that the reult are n perfect match. It can be concluded, that once a detaled mulaton of the rectfer ha been performed and t reult tored n tabular format, an equvalent current ource ung that tabular can repreent the rectfer n other mulaton tude, whch requre much le computaton tme.. Dpered generaton (PV ytem example) A redental grdconnected PV ytem [1] llutrated n Fg. 5. A detaled Smulnk model for th ytem hown n Fg., whch nclude the equvalent crcut model of a photovoltac cell (n our cae the model for the olar cell baed on the wellknown ngledode equvalent crcut taken form [, 3, and 5]) connected to a DC/DC converter, together wth a DC/AC converter that are connected between a PV array and the electrcal power ytem. A a B b conv.mat C c conv 3.mat va 1 vb 1 vc1 Fgure 3: Smplfed equvalent model for the threephae ACDC dode brdge. Smulaton reult wth the mplfed rectfer model n Fg. 3 are hown n Fg.. Fgure 5: Redental PV ytem ncludng nonlnear load. 1th PSCC, Glagow, Scotland, July 118, 8 Page

3 55 g D S Irradance G1 PVcell Temperature Tc 1 Product Np*Mp N*M STC radaton 1 Controlled Current Source Sere RLC Branch 3 STC radaton Product PV cell Dode v [vdc ] [Vpv ] f(u) Trangle <= Voltage Goto Meaurement 1 From 1 Fcn Mofet m v C [Vpv] Voltage Meaurement 7 Termnator Goto n(teta ) Va [v] C From P.L.L [vdc ] From 3 _ref K T 3.5 z3.7 g ref _nv Product 1 I*peak Zero Order Hold z1 Dcrete Tme Integrator 3 z.99 Dcrete Tranfer Fcn PI Zero Order Hold 1 Hytere Lnv 13. mh. A _nv B Load [v] v AC Voltage Source Goto 1 v. Unveral Brdge Fgure : Detaled model of a PV ytem The DC/DC converter hown n Fg. a boot type ncludng the maxmum power pont tracker (MPPT) of the PV array. The ued bac control gven n []. The DC/AC converter cont of a nglephae nverter operatng wth a unty power factor. The ntantaneou current command for the nverter can be commutated by * multplyng I peak whch the output of the regulator (n our cae a PI regulator) for the DCvoltage control ytem, wth the AC ytem lne voltage wth unty ne wave ampltude obtaned by ung a robut Phae Locked Loop (PLL) [7]. Smulaton reult are hown n Fg. 7 to 1, baed on a varaton of the temperature T and the rradaton G from T= 5 O C and G=1 W/m at t= to t=.15, and from T= 15 O C and G=3 W/m from t=.15 untl t= Fgure 8: Smulated termnal voltage of the nglephae nverter n Fg Fgure 9: Smulated load current of the ytem n Fg Fgure 7: Smulated current on the termnal of the nglephae nverter n Fg Fgure 1: Smulated upply current for the ytem n Fg.. 1th PSCC, Glagow, Scotland, July 118, 8 Page 3

4 Agan, the mulaton reult obtaned wth the detaled model n Fg. for the current njected nto the grd by the PV ytem are tored n tabular format; and a new mulaton run wth a current ource replacng the PV ytem, a hown n Fg. 11. The reult are gven n Fg. 1 to 15. nv.mat Load Fgure 11: Smplfed equvalent model of the PV ytem n Fg Fgure 1: Smulated termnal current of the nglephae nverter baed on the mplfed model n Fg Fgure 13: Smulated termnal voltage of the nglephae nverter baed on the mplfed model n Fg Fgure 15: Smulated upply current wth the mplfed model n Fg. 11. A perfect match found comparng the reult n Fg. 7 to 1 wth the reult n Fg. 1 to Expermental valdaton In order to confrm the valdty of the mplfed equvalent model, two experment performed on a laboratory etup are preented. The frt tet related to reonance ue, and the econd one demontrate current uperpoton effect n a feeder..3.1 Reonance effect The reonance phenomenon wll become an mportant problem n LV network by an ncreaed number of nonlnear load and dpered generaton. To tudy th problem the reonant frequency meaured wth and wthout PV ytem connected to the grd. The laboratory etup repreentng a LV network preented n Fg. 1. It cont of a grd emulator whch feed houe. At frt only a capactor connected and reonant frequency f 1 meaured. Next a PV ytem connected to th LV network (houe No ) a on Fg. 1 and reonant frequency f meaured agan. The dfference n f 1 and f frequency let calculate capactance nerton to the grd of the PV ytem. A the reult, reonant frequency decreae. That mean that many PV ytem connected to the grd decreae the reonant frequency. That why t mportant to tudy th problem. 3V 1 µω 37 µh Fgure 1: Smulated load current wth the mplfed model of Fg. 11. µf Fgure 1: Expermental LV network, etup #1 1th PSCC, Glagow, Scotland, July 118, 8 Page

5 Expermental reult are gven n Fg. 17 to. On fgure 17 we ee the nverter current and voltage. On fgure 18 the upply current and feeder voltage gven. The grd current the um of the nverter and capactor current, n reonance condton. The reonance frequency eeng n fg. whch equal to 175Hz. PV current data when the grd voltage clean 3 µf PV ytem Fgure 1: PV nverter mplfed equvalent model. (1A/dv) (v/dv) Fgure 17: Meaured feeder voltage and njected PV nverter current a a reult of reonance n the etup of Fg. 1. (5A/dv) (v/dv) Fgure 18: Meaured upply current and feeder voltage a a reult of reonance n the etup of Fg. 1. (5A/dv) Fgure 19: Meaured capactor current when reonance occur n etup of Fg Hz Fgure : FFT of the meaured upply current when reonance occur n the etup of Fg. 1. Concernng now mulaton part, two data et are neceary to be ncluded a ource n the reduced mulaton cheme: the polluted grd voltage (Fg.) and the PV nverter current under clean voltage grd condton (Fg.3). Agan, the mplfed equvalent PV nverter model cont of a current ource n parallel wth a capactor a hown n Fg. 1. The reaon of addng th parallel capactor that t correpond to the realtc tuaton of the teted PV nverter. The bac meaured data are hown n Fg. and 3, whle the mulaton reult are gven n Fg. to 7.,, Fgure : Meaured (and tored) polluted grd voltage (.5% of harmonc for the 5 frt harmonc) Fgure 3: Meaured (and tored) current njected by the PV ytem when the grd voltage free from dtorton Fgure : Smulated njected current and feeder voltage (to be compared wth Fg. 17) Fgure 5: Smulated upply current and feeder voltage (to be compared wth Fg. 18). 1th PSCC, Glagow, Scotland, July 118, 8 Page 5

6 The meaured current for each PV connected alone are gven Fg. 9 to 31. (A/dv) (8v/dv) Fgure : Smulated capactor current (to be compared wth Fg. 19). Fgure 9: Meaured current and voltage of PV alone. 5 3 (1A/dv) (8v/dv) h Fgure 7: FFT of the mulated upply current (to be compared wth Fg. ). Fgure 3: Meaured current and voltage of PV3 alone..3. Superpoton of current The econd experment wll how the valdaton of the aumpton that the mplementaton of current ource n the mulaton wth data n tabular format a taken from the ndvdual operaton of nonlnear load, wll reult n a correct uperpoton of effect n a complex LV network. For th purpoe, the meaured data n the tet etup hown n Fg. 8 were taken for each PV ytem connected alone to the network (PV1, PV3 and PV at place 1, 3 and, repectvely, n the expermental branch); and the tored reult were mplemented a current ource n the global mulaton by repectng the orgnal place of connecton of the PV ytem. The ntenton to how that t poble to mulate each houe alone, whch contan dpered generaton and nonlnear load and then to take the ndvdual current and mplement them together n the global mulaton. Th wll enable to mulate complex LV network by takng nto account the real current and voltage of each nonlnear load and/or dpered generaton. 1 µω 37 µh 3V 1 3 (A/dv) (8v/dv) Fgure 31: Meaured current and voltage of PV1 alone. The tored current value are njected n the mulaton cheme preented n Fg. 3. The mulaton reult are hown n Fg \PQ_ Dcrete, T = 5e7. upply nv 1 nv 3 nv to the cope Scope 5 Current 1 Goto R, L 1m Inv 1 Goto 7 17m 17 m Inv 3 Goto 5 17m 17m 17m Inv Goto Goto Vnv 1 Goto v Vnv 3 v Goto 3 Vnv v Load v v Goto 1 nv 1 nv 3 95 \PQ 95 \PQ nv 95 \PQ Fgure 3: Smulaton cheme of the LV network wth equvalent current ource of three PV ytem together. Fgure 8: Expermental LV network, etup # 1th PSCC, Glagow, Scotland, July 118, 8 Page

7 oppote drecton and negatve current hown n Fg.33 and Fg.3. 3 LV NETWORK SIMULATION Fgure 33: Smulated reult of the LV network wth three PV ytem together. From top to bottom: the total current and the feeder voltage; current and voltage of PV1, PV3 and PV, repectvely. (A/dv) (8v/dv) (A/dv) (8v/dv) (1A/dv) (8v/dv) A an example of applcaton of the propoed modellng approach by mean of equvalent current ource, a complete LV network havng 9 houe mulated n uch way that each houe repreented by mplfed equvalent model of a PV ytem n parallel wth one nonlnear load comprng a nglephae dode rectfer. Each nglephae rectfer connected to a RL load and produce ndvdually a current THD equal to.%, whle the PV current ha a THD =.8%. A typcal Dutch LV network taken from [8], and compre two feeder a hown n Fg. 35, beng connected to a MV network va a 1kV/V tranformer. Each feeder dvded n fve ecton. The dtance between each ecton hown n Fg. 35 (ecton 1:1m, ecton : 1m, ecton 3: 18m, ecton : 5m, ecton 5: 5m). Each ecton contan houe, whch are located at every 1m from each other. So, the total number of houe n one feeder 8. Supply voltage 1 kv / V Cable 15Al 8 houe 1 m 1 m 18 m 5 m 5 m Houe n1 1 m Cable 5Al Fgure 35: Decrpton of the LV network. Feeder 1 Feeder Houe n (A/dv) (8v/dv) Smulaton reult are hown n Fg. 3 to 1. 5 Fgure 3: Meaured reult of the LV network wth three PV ytem together. From top to bottom: the total current and the feeder voltage; current and voltage of PV1, PV3 and PV, repectvely. The reultng current n Fg. 33 can be compared wth the expermental data n Fg. 3, where the three PV are connected multaneouly n the laboratory etup. It clear that the experment and mulaton are n good match. Remark1: In the experment (Fg. 3) a hgh frequency current wa meaured (more then 3kHz). Th current flowng between two nverter. The frequency not relevant for th nvetgaton. Remark: The grd current meaurng probe clamped to meaure the potve current flowng from the ource to the houe. Snce n the houe only PV ytem are connected the current flowng n the Fgure 3: Smulated voltage n the begnnng of feeder Fgure 37: Smulated threephae current n the begnnng of feeder 1. 1th PSCC, Glagow, Scotland, July 118, 8 Page 7

8 Fgure 38: Smulated voltage n the begnnng of feeder Fgure 39: Smulated threephae current n the begnnng of feeder Fgure : Smulated threephae voltage at the LV termnal of the dtrbuton tranformer Fgure 1: Smulated total threephae current (feeder 1 feeder ) near the tranformer. Table 1 ummarze the THD of the LV network. THD Current n feeder1 19.% Current n feeder 19.18% Current n um of 19.1% feeder1 and feeder Voltage n feeder1.5% Voltage n feeder.3% Voltage n um of.5% feeder1 and feeder Table 1: THD n the LV network A hown n Table 1, the current THD of the network hgher than the allowed 5%, whch requre the ntroducton of (actve) flter n the dtrbuton grd. CONCLUSION Th paper propoe a way for tudyng complex LV network whch contan a large amount of nonlnear load and dpered generaton. The man dea to ue tored real data (expermentaton) or data from detaled model to contruct mplfed equvalent current ource, and to ntroduce thee ource n tabular form n the complex LV network. By th way, problem related to computer memory lmtaton are avoded. Smulaton and expermental reult how that the approach vald. The propoed model gve good reult when compared to experment, and let to tudy problem concernng reonance effect and harmonc nteracton between dpered generaton and nonlnear load oberved n LV network. REFERENCES [1] Lm Boon Teong: Smulaton of a Redental Grd Connected Photovoltac Sytem A the of bachelor of engneerng, department of electrcal and computer engneerng, Curtn unverty of technology, 5. [] Abou ElMaaty Metwally Metwally Aly Abd El Aal Modellng and Smulaton of a Photovoltac Fuel Cell Hybrd Sytem PhD the at Faculty of Electrcal Engneerng, Unverty of Kael, 5. [3] ACHIM%S.%Y/the%1.pdf [] Abd ElShafy A. Nafeh, F. H. Fahmy, E. M. Abou ElZahab Maxmumpower operaton of a tandalone PV ytem ung fuzzy logc control Internatonal journal of numercal modelng:electronc Network, devce and feld, Vol. 15 pp , John Wley & Son, Ltd.. [5] C. Hua, J. Ln, and C. Shen, Implementaton of a DSPcontrolled photovoltac ytem wth peak power trackng, IEEE Tran. on Indutral Electronc, Vol. 5, No. 1, February 1998, pp [] Matu, M.; Ktano, T.; Dehong Xu; Zhongqng Y.: A new maxmum photovoltac power trackng control cheme baed on power equlbrum at DC lnk ThrtyFourth IAS Annual Meetng, IEEE Indutry Applcaton Conference, [7] M. C. Benhabb; S. Saadate: A New Robut Expermentally Valdated Phae Locked Loop for Power Electronc Control EPE Journal Volume 15. [8] J. F.G. Cobben, W.L. Klng and J. M.A. Myrzk Power Qualty apect of a future mcro grd Internatonal Conference on Future Power Sytem, pp. 5, Nov. 5. 1th PSCC, Glagow, Scotland, July 118, 8 Page 8