J. G. Pnto, Bruno Exosto, Vtor Montero, L. F. C. Montero, Carlos Couto, João L. Afonso, Comarson of Current and Voltage Shunt Actve Power Flters for Harmonc Comensaton and Reactve Power Control, 8th Annual Conference of the IEEE Industral Electroncs Socety IECON0,. 54-548, Montreal, Canada, Oct. 0, ISBN: 978--467-40-5. Comarson of Current- and Voltage- Shunt Actve Power Flters for Harmonc Comensaton and Reactve Power Control J. G. Pnto, Bruno Exosto, Vtor Montero, L. F. C. Montero, Carlos Couto, João L. Afonso - Centro Algortm Unversty of Mnho Gumarães, Portugal State Unversty of Ro de Janero Electroncs Engneerng Program Ro de Janero, Brazl E-mals: gabrel.nto@agortm.umnho.t, bruno.exosto@agortm.umnho.t, vtor.montero@agortm.umnho.t, lmontero@uerj.br, carlos.couto@agortm.umnho.t, joao.l.afonso@agortm.umnho.t Abstract Ths aer resents the comarson between two three-hase Shunt Actve Power Flters (SAPFs), one wth Current- Inverter (CSI) and the other wth Voltage- Inverter (VSI), whch are used to comensate current harmoncs and to control reactve ower. The control algorthm of both SAPFs s based on the Instantaneous Reactve Power Theory (-q Theory). The comarson here resented focuses on the Total Harmonc Dstorton (THD) and RMS values of the comensated currents, and n the Total Power Factor (TPF) of the nstallaton. Keywords Power Qualty, Harmoncs, Reactve Power, Shunt Actve Power Flter, Phase-Locked Loo, Dgtal Control, Voltage- Inverter (VSI), Current- Inverter (CSI). I. INTRODUCTION The Power Qualty s one of the most mortant characterstcs of the modern electrcal energy dstrbuton systems. Consequently, the oor Power Qualty can cause enormous economc losses to the ndustral facltes. In Euroe, Power Qualty roblems costs more than 50 bllon a year to busness []. The ncreasng use of rectfers, thyrstor ower converters, arc furnaces, swtchng ower sules and other non-lnear loads s known to cause serous roblems n electrc ower systems []. Therefore the use of ths tye of loads moses the develoment of solutons for the mtgaton and reducton of the Power Qualty roblems. Some adoted solutons for these roblems, such as assve flters for harmonc mtgaton, or caactor banks for reactve ower comensaton, can cause other roblems such as resonances, voltage transents and others []. Shunt Actve Power Flters (SAPFs) are a good soluton, and have been develoed over the years [4][5][6]. There are several toologes of SAPFs, but they can be generally classfed n two man grous: Voltage- SAPFs and Current- SAPFs, deendng of the toology of nverter that s used. In ths aer are resented several comuter smulatons results of a Voltage- and a Current- SAPF, controlled usng the -q Theory. II. CONTROL OF THE SHUNT ACTIVE POWER FILTERS The controller of the SAPFs resented n ths aer uses the -q Theory roosed by Akag et al [7]. Ths ower theory has been largely used n the mlementaton of SAPFs over the years, and has rovded good results wth dfferent tyes of electrcal nstallatons and loads [8][9]. The -q Theory works n the α-β reference frame, and therefore, the electrcal grd voltages (v a, v b, v c ) and the load currents ( a, b, c ) must be converted to ths reference frame by alyng the Clarke transformaton, gven by () and (): v v 0 0 v a v b v c a b c The -q Theory ower comonents are calculated usng the exressons () and (4), where s the nstantaneous real ower, and q s the nstantaneous magnary ower (by defnton). () () v v () q v v Each one of the nstantaneous ower comonents can be searated nto an average value and an oscllatng value. The hyscal meanng of each of the nstantaneous owers s: Average value of the nstantaneous real ower. Corresonds to the energy er tme unt transferred from the ower source to the load, n a balanced way through the three hases; ~ Oscllatng value of the nstantaneous real ower. It s the energy er tme unty that s exchanged between the ower source and the load, through the three hases; q The nstantaneous magnary ower. Corresonds to the ower that s exchanged between the hases of the load. Ths comonent does not mly any transference of energy between the ower source and the load, but s resonsble for the exstence of undesrable currents. Normally, only the average value of the nstantaneous real ower ( ) s desrable, and the other ower comonents can be comensated usng a SAPF. In order to calculate the reference currents that the actve flter should nject t s necessary to searate the desred ower comonents from the (4)
undesred ones. The undesred ower comonents are denomnated x and q x. In addton to the nstantaneous ower comonents defned by the -q Theory, there s also a comonent, reg, whch s used to regulate the caactor voltage and the nductor current, n the DC sde of the Voltage- and Current- SAPFs, resectvely. Ths regulaton, n both of the resented SAPFs, s done wth a Proortonal-Integral (PI) controller, and the error between the reference voltage (V ref ) or the reference current (I ref ) and the voltage or the current measured at the DC sde of the nverter (V DC or I DC ). The comonent reg s an actve ower comonent and s related to the SAPFs steady state oeraton losses. reg must be suled by the electrcal grd, and therefore ths ower comonent s ncluded n the calculaton of x. So, the values of the ower comonents that the SAPFs should roduce are gven by: ~ (5) x reg q x q (6) The undesred ower comonents and reg are then used to determne the comensaton currents n the α-β coordnates usng the exresson (7): c c * * v v v v v x v qx The reference comensaton currents n the a-b-c coordnates ( ca *, cb *, cc *) are determned by alyng the nverse Clarke transformaton to the currents n the α-β reference frame, as demonstrated n exresson (8): ca cb cc * * * c A. Modfed -q Theory for Snusodal -Currents The alcaton of the -q Theory roduces reference comensaton currents to the SAPFs that result n a constant ower at the source. In three-hase ower systems wth dstorted and/or unbalanced voltages t s not ossble at the same tme to have constant ower flow from the source to the load, and also to have snusodal currents at the source. In ths way, when the electrcal grd voltages are dstorted and/or unbalanced, the oeraton of the SAPFs results n dstorted comensated currents at the source, whch are necessary to obtan constant ower at the source. If nstead of constant ower at the source s referred to obtan snusodal currents at the source, t s ossble to modfy the -q Theory n order to acheve t, even when the electrcal grd voltages are not snusodal and balanced. For ths urose, nstead of drectly use the grd voltages to erform the calculatons of the ower comonents defned n the -q Theory, s used the ostve sequence of the fundamental comonent of these voltages. To obtan the ostve sequence of the fundamental comonent of c * * (7) (8) the voltages t can be used a dgtal Phase-Locked Loo (PLL). Ths -q Theory modfcaton was successfully aled both n Current- [0] and Voltage- SAPFs [4]. In Fg. t can be seen the block dagram of the dgtal controller of the SAPFs, based on the -q Theory. Fg. (a) shows the man algorthm and Fg. (b) shows a detaled vew of the dgtal PLL that s used to obtan the ostve sequence of the fundamental comonent of the grd voltages []. III. SHUNT ACTIVE POWER FILTERS TOPOLOGIES The toology of each actve flter s resented n Fg.. As t can be seen, the ower stage of the Voltage- SAPF s comosed by a two-level Voltage- Inverter (VSI) that uses sx IGBTs and a caactor at the DC lnk. The assve flters used n ths toology are LC low ass flters. To dam the resonse of the assve flter n ther resonance frequency, was laced a resstor wth low value, n seres wth the caactors. The Current- SAPF ower stage s consttuted by a Current- Inverter (CSI) that uses sx IGBTs wth reverse blockng dodes n seres wth the IGBTs and an nductor n the DC lnk. The assve flters are also LC assve flters, but the arrangement s dfferent from the Voltage- SAPF. In TABLE I s ossble to see all the arameter values of both SAPFs. Three dfferent tyes of balanced three-hase loads were used to test the erformance of the SAPFs: - three nductors (L L ) and three resstors (R L ) connected n Y; - a full-brdge rectfer wth a resstor (R DC ) and an nductor (L DC ) n the DC sde, and three nut nductors (L I ); - a full-brdge rectfer wth a resstor (R DC ) and a caactor (C) n the DC sde, and three nut nductors (L I ). The smulatons were erformed usng the PSIM software. va vb vc a b c a-b-c a-b-c v α v β α-β α-β vα vβ α β x x PLLα α β PLL q PLLβ q VDC_ref VDC IDC_ref k I IDC k P - - LPF (a) cos(ωt) sn(ωt) (b) PI PI - Fg.. Block dagram of the SAPFs dgtal controller: (a) Imlementaton of the -q Theory; (b) Phase-Locked Loo. ~ ω cα* cβ* reg cα* cβ* ωt α-β a-b-c ca* cb* cc*
(a) Fg.. SAPFs ower nverters toologes: (a) Voltage- SAPF; (b) Current- SAPF. (b) TABLE I SIMULATION PARAMETERS OF THE SHUNT ACTIVE POWER FILTERS VS-SAPF Parameters CS-SAPF Parameters V l-l 00 V V l-l 00 V L sx 0. mh L sx 0. mh Rsx 0. Ω Rsx 0. Ω C DC mf L DC 8 mh V DC 400 V I DC Varable C Fx 0 μf C Fx 60 μf L Fx 5 mh L Fx.5 mh R Fx Ω R Fx.5 Ω : (RL load) TABLE II LOAD SIMULATION PARAMETERS Parameters : (Full-Brdge Rectfer wth RL load) : (Full-Brdge Rectfer wth RC load) L L 40 mh L I mh L I mh R L 7 Ω R DC 0 Ω R DC Ω L DC 4 mh C 0.6 mf In both of the SAPFs was used Perodc-Samlng modulaton technque. The mlementaton of ths swtchng technque for the Voltage- SAPF s done by erodcally samlng the nverter outut current ( fx ) and then makng a comarson wth the reference comensaton current ( cx *). The IGBTs are then swtched accordng wth the result of ths comarson. In the mlementaton of the Perodc-Samlng swtchng technque n the Current- SAPF, nstead of drectly swtchng the IGBTS, the result of the comarson s sent to a combnatonal logc crcut that generates the correct ulse atterns to ths tye of nverter. The IGBTs are then swtched accordng wth the result of ths combnatonal logc crcut. The dfferences n the swtchng atterns resde n the fact that n the Voltage- SAPF two IGBTs from the same nverter leg cannot be conductng smultaneously, otherwse the DC sde caactor s shortcrcuted and the current can destroy the IGBTs. In the Current- SAPF the current of the DC sde nductor must never be nterruted, otherwse the voltage of the DC sde nductor ncreases to values that can destroy the IGBTs of the nverter. To avod that, all the states nvolve two IGBTs conductng at the same tme. In the case of the Voltage- SAPF are ntroduced dead-tmes to ensure that the DC sde caactor s not short-crcuted. In the case of the Current- SAPF t s necessary to ensure the overlang of the swtchng ulses. In the smulatons of both of the SAPFs, the samlng frequency of the dgtal controller was fxed n 0 khz and the mnmum tme between swtchng transtons was lmted by ths samlng frequency. Therefore, the swtchng frequency s not fxed and has a maxmum value of 5 khz. IV. SIMULATION RESULTS In Fg. t s ossble to see smulaton results of the source, load and actve flter currents wth the SAPFs comensatng three dfferent loads. As t can be seen, after the comensaton wth both SAPFs, the source currents waveforms become almost snusodal and n hase wth the electrcal grd voltage. Ths ndcates that the SAPFs are oeratng correctly. The harmonc sectrum of the source and load currents can be seen n Fg. 4. The harmonc sectrum of the source currents n all the smulated cases s ndcatve that both of the SAPFs are oeratng well and that they comensate the current harmoncs correctly. Ths s notceable when the SAPFs comensate the non-lnear loads. In TABLE III s ossble to see a summary of the smulaton results obtaned wth the SAPFs comensatng the three dfferent loads. The results from the comensaton of the RL load show that the RMS values of the source currents are nferor to the values of the load currents. The fact that the total ower factor after the comensaton s nearly untary ndcates that both of the SAPFs are erformng well.
Smulaton Results of the Voltage- SAPF Comensaton of (RL load): Smulaton Results of the Current- SAPF Flter Comensaton of (Full-brdge rectfer wth RL load): Flter Comensaton of (Full-brdge rectfer wth RC load): Flter Fg.. Waveforms of grd voltage and currents at the load, source and actve flter wth Voltage- SAPF and wth Current- SAPF, for three dfferent tyes of loads.
Smulaton Results wth the Voltage- SAPF Smulaton Results wth the Current- SAPF Comensaton of (RL load): 0 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz Comensaton of (Full-brdge rectfer wth RL load): 0 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz Comensaton of (Full-brdge rectfer wth RC load): 0 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz 0 5% 0 Hz 50 Hz 500 Hz 750 Hz 000 Hz Fg. 4. Harmonc sectrum of the currents at the load and source wth Voltage- SAPF and wth Current- SAPF, for three dfferent tyes of loads.
TABLE III SIMULATION RESULTS OF THE SHUNT ACTIVE POWER FILTERS VS-SAPF (RL load) CS-SAPF P S 60 W P S 608 W S S 6 VA S S 600 VA TPF S 0.998 TPF S 0.997 THD IS.% THD IS.% I S (RMS) 8.58 A I S (RMS) 7.90 A P L 590 W P L 5908 W S L 0957 VA S L 0970 VA TPF L 0.58 TPF L 0.58 THD IL 0.4% THD IL 0.4% P L/P S 9.% P L/P S 97.% I L (RMS).0 A I L (RMS).0 A (Full-brdge rectfer wth RL load) P S 654 W P S 6084 W S S 66 VA S S 60 VA TPF S 0.998 TPF S 0.996 THD IS.7% THD IS.4% I S (RMS) 8.0 A I S (RMS) 7.9 A P L 59 W P L 595 W S L 654 W S L 6540 VA TPF L 0.908 TPF L 0.907 THD IL 8.9% THD IL 8.9% P L/P S 96.4% P L/P S 97.5% I L (RMS) 9.0 A I L (RMS) 9. A (Full-brdge rectfer wth RC load) P S 5796 W P S 58 W S S 587 VA S S 58 VA TPF S 0.986 TPF S 0.997 THD IS.% THD IS.% I S (RMS) 7.0 A I S (RMS) 7.0 A P L 557 W P L 557 W S L 607 VA S L 606 VA TPF L 0.897 TPF L 0.897 THD IL 8.8% THD IL 8.68% P L/P S 96.% P L/P S 95.8% I L (RMS) 8.0 A I L (RMS) 8.0 A When comensatng the full-brdge rectfer wth RL load the erformance of the SAPFs s also smlar. The source currents THD after the comensaton s slghter hgher n the case of the Voltage- SAPF, nevertheless the dfference between the two SAPFs erformance s neglgble. In the comensaton of the full-brdge rectfer wth RC load, the erformance of both SAPFs s once more very smlar. The source current THD values after the comensaton are much nferor to the load currents THD, and there s almost no dfference n the source currents THDs of both SAPFs. The RMS values of the currents n ths case are slghtly nferor n the source, after comensaton, than n the load. V. CONCLUSIONS In ths aer was resented a comarson between a Voltage- and a Current- Shunt Actve Power Flter (SAPF) when comensatng three dfferent tyes of loads. The comarson between the two SAPFs addressed the Total Harmonc Dstorton (THD) and RMS values of the comensated currents and the Total Power Factor (TPF). As the results resented n ths aer show, the erformance of both SAPFs s very smlar. The values of the source currents THDs after the comensaton are smlar n both cases. The RMS values of the currents n the source are reduced n all the cases and more substantally when the SAPFs comensate the RL load. In both cases the -q Theory used n the control system erforms well and successfully dentfes the currents comonents that must be comensated by the SAPFs. For future work t wll be done a comarson of the SAPFs dynamc erformance n resonse to varatons n the loads. ACKNOWLEDGMENT Ths work s fnanced by FEDER Funds, through the Oeratonal Program for Comettveness Factors COMPETE, and by Natonal Funds through FCT Foundaton for Scence and Technology of Portugal, under rojects: PTDC/EEA-EEL/04569/008; FCOMP- 0-04-FEDER-0674 and MIT-PT/EDAM-SMS/000/008. REFERENCES [] Poor Power Qualty costs Euroean busness more than 50 bllon a year, Leonardo Power Qualty Intatve - Euroean Coer Insttute, 008. [] A. Bachry and Z. A. Styczynsk, An Analyss of Dstrbuton System Power Qualty Problems Resultng from Unbalance and Harmoncs, n IEEE PES Transmsson and Dstrbuton Conference and Exoston (IEEE Cat. No.0CH7495), 00,. 76-766. [] J. G. Pnto, R. Pregtzer, L. F. C. Montero, C. Couto, and J. L. Afonso, A Combned Seres Actve Flter and Passve Flters for Harmoncs, Unbalances and Flcker Comensaton, n 007 Int. Conference on Power Engneerng, Energy and Electrcal Drves, 007,. 54-59. [4] J. Afonso, C. Couto, and J. Martns, Actve Flters wth Control Based on the -q Theory, IEEE Industral Electroncs Socety Newsletter, vol. 47,. 5-0, 000. [5] B. Sngh, K. Al-haddad, and A. Chandra, A Revew of Actve Flters for Power Qualty Imrovement, IEEE Transactons on Industral Electroncs, vol. 46, no. 5,. 960-97, 999. [6] R. Pregtzer, J. G. Pnto, M. J. Seúlveda, and J. L. Afonso, Parallel Assocaton of Shunt Actve Power Flters, n ISIE 007. IEEE Int. Symosum on Industral Electroncs, 007, no.,. 49-498. [7] H. Akag, Y. Kanazawa, and A. Nabae, Generalzed Theory of the Instantaneous Reactve Power n Three-Phase Crcuts, n IPEC 8 - Int. Power Electroncs Conf., 98,. 75-86. [8] J. G. Pnto, P. Neves, D. Goncalves, and J. L. Afonso, Feld results on develoed three-hase four-wre Shunt Actve Power Flters, IECON 009. 5th Annual Conference of IEEE, 009,. 480-485. [9] B. Exosto, H. Gonçalves, J. G. Pnto, J. L. Afonso, and C. Couto, Three Phase Four Wre Shunt Actve Power Flter from Theory to Industral Faclty Tests, n th Int. Conference on Electrcal Power Qualty and Utlzaton - EPQU, 0, no.,. - 5. [0] B. Exosto, H. Carnero, G. Pnto, C. Couto, and J. L. Afonso, Smulatons of a current-source Shunt Actve Power Flter wth Carrer-Based PWM and Perodc Samlng Modulaton Technques, n EPE 0 4th Euroean Conference on Power Electroncs and Alcatons, 0,. -8. [] L. G. Barbosa Rolm, D. Rodrgues da Costa Jr., and M. Aredes, Analyss and Software Imlementaton of a Robust Synchronzng PLL Crcut Based on the q Theory, IEEE Transactons on Industral Electroncs, vol. 5, no. 6,. 99-96, Dec-006.