A 500kW PV GENERATOR Rodrgo Moretón, Eduardo Lorenzo and Javer Muñoz IVCURVE ABSTRACT Ths paper presents the measurement of the IV curve of a 500kW PV generator by means of an ownmade capactve load. It s shown that IV curve analyss can also be appled to bg PV generators and that when measurng the operaton condtons wth reference modules and takng some precautons (especally regardng the operaton cell temperature), t s stll a useful tool for characterzng them and therefore can be ncorporated nto mantenance procedures. As far as we know, ths s the largest IV curve measured so far. KEYWORDS IV curve; nfeld measurements; qualty control; PV generator 1. INTRODUCTION Onste IV curves of PV generators are a useful tool for assessng not only the effectve peak power but also for dagnosng possble performance anomales (shadows, hotspots, polarzaton, connecton falures...). Nevertheless, current state of art s restrcted to relatvely low powers, typcally below lookw, whch s lkely due to the practcal dffcultes of dealng wth large currents. In fact, as far as we know, commercal IV tracers are lmted to 100 A. Nowadays, however, generators of up to 800 kw are found n PV nstallatons, whch mples currents above 1000 A. Ths paper presents the IV curve of a 500kW generator by means of an ownmade capactve load. As far as we know, ths s the largest IV curve measured so far. Once obtaned, the IV characterstcs were extrapolated to standard test condtons (STC) accordng to IEC60891 [1] usng the ncdent rradance, G, and the cell operaton temperature, 7c, regstered by means of two reference modules. 2. IV CURVE MEASUREMENTS 2.1. The PV generator The measurement took place n a 2MW PV plant connected to the grd and located n Lorca (n the SouthEast of Span) on the 18 January 2012. Ths PV plant s formed by four 500kW generators connected to ther respectve nverters. Each generator conssts of 93 parallel connected strngs, each of whch s composed of the seres connecton of 23 monocrystallne slcon modules of 250 W. The nomnal values of the PV generator, resultng from the flashlst nformaton gven by the manufacturer, are the followng: shortcrcut current: 810 A; opencrcut voltage: 876 V; and nomnal power: 535924 W. Every nverter had nne bpolar entres, able to accommodate the cables comng from an equal number of DC boxes n the feld. These entres were paralleled nsde the nverter. The here concerned PV generator ncluded sx DC boxes (3 wth 15 parallel connected strngs and another 3 wth 16 parallel connected strngs), and therefore, three nverter entres were free. We took advantage of ths crcumstance by smply connectng our IV tracer to one of these free entres (Fgure 1). It s worth notng that n ths way, our measurements were made just at the nverter entry. In other words, correspondng results nclude all the losses untl ths pont (possble early degradaton, module msmatchng, DC wrng, etc.). 2.2. The IV tracer We used an upscaled nhousemade capactve load based on nsulated gate bpolar transstors (IGBTs) that has been descrbed n prevous publcatons [2]. Fgure 2 presents a smplfed scheme. Key features are the followng:
precharge compensates ths fact allowng the PV generator to pass through the shortcrcut pont durng the chargng process. To mnmze the nose/sgnal relaton, we used a foursolated channel osclloscope (Metrx Scopx OX7104C, Chauvn Arnoux, Pars, France) for acqurng the current and voltage sgnals comng from the generator, and the rradance and the cell operaton temperature comng from the reference modules. 2.3. Measurng condtons Fgure 1. /l/tracer connecton to the entrance of the nverter, followng a fourwre confguraton. It can be observed the /l/tracer n the background, the sx cables from the DC boxes on the rght, and the /l/tracer connecton n a free entry n the foreground. M PV V. + I IGET 1 >r HSHVST y, IGBT: Fgure 2. Smplfed crcut desgn. Both nsulated gate bpolar transstors (IGBTs) functon as swtchers, R D s the capactors dscharge resstance, l/ PRE the precharge battery (acted by swtcher P), and Cthe capactor. An 800V/16.7mF capactor. Roughly, the capactor chargng process s descrbed n terms of a tme constant T = RC, wth C beng the capactance and R beng the resstance determned by the opencrcut voltage (Voc) and the shortcrcut current (/ sc ) of the generator {R = Vodhc) Here, t= 18.1 ms, whch leads to chargng tmes of about 30 ms, large enough to avod transent nfluences [2]. A 400A/1200V IGBT. In ths case, the current measured nearly doubles the lmt of the contnuous IGBT current (400 A), but because of the very low chargng tmes, t causes no damage to the transstor. An 800A/150mV 0.5 class (uncertanty of 0.5%) shunt resstance for measurng the current at the entrance of the load. A fourwre connecton confguraton, wth the voltage taken n the connecton pont, to avod consderng any voltage drop n the /l/tracer cables. A negatve voltage capactor precharge, usng a battery, for assurng the capactor to pass through the shortcrcut pont. Because of the large current value, there was a voltage loss n the IV tracer cables of about 24 V between the connecton pont and the entrance to the load. The negatve voltage Measurng operaton condtons, that s, ncdent rradance (G) and cell operaton temperature (T c ) that were regstered (through / sc and Voc) from two reference modules, prevously stablzed and calbrated outdoors at the IESUPM facltes. The calbraton traceablty s referred to the CIEMAT. The correspondng uncertanty s 2.0% n / sc, 1.0% n Voo and 2.5% n P M. These modules are from the same batches and, therefore, of the same type, as the ones formng the generator. The use of reference modules s the best opton when tryng to reduce uncertanty assocated to spectral, angular, and thermal responses [36]. They were nstalled n the generator's structure n a poston free from shadows. Just as a precauton to prevent the uncertanty assocated wth the effect of dust n the measurements, the reference modules were nstalled more than 15 days n advance, thus guaranteeng smlar dust coverage, as can be observed n Fgure 3. In addton, daly ranfalls of more than 6 mm took place n the days prevous to the measurement, contrbutng to clean homogenously all the modules [7,8]. The man uncertanty source when measurng the IV curve of large PV generators onste s the one assocated wth the T c determnaton [3]. The bgger the generator and the larger the wnd speed, the larger the T c spread among the generator and, therefore, the less representatve the value gven by a sngle reference module. To lmt the correspondng uncertanty, t s worth assurng the followng: chargng tmes of above 20 ms, ncdent rradance larger than 800 W/m, dffuse/global rradance proporton Fgure 3. Vew of the reference modules nstallaton above the generator: there are no apprecable drtness dfferences.
{DIG) lower than 20%, and wnd speed lower than 3 m/s. As presented n Table I, n ths case, the weather condtons easly fulflled these requrements. In any case, a thermographc nspecton of the nstallaton showed mean temperature dfferences between the reference module and the modules formng the generator lower than 2 C. 3. RESULTS Fgure 4 presents the evoluton of the current (lght lne) and the voltage (dark lne) durng the chargng process. As t can be observed, the chargng tme s greater than 20 ms, whch allows to avod fll factor errors n the measurement [2,9,10]. Despte t s not relevant for the fnal results, t s nterestng to note the acute peak current (~ 1400 A), at the begnnng of the capactor chargng process. It occurs because of the dsplacement of majorty carrers, nsde both p and n zones, requred to adapt the length of the depleton zone of the pn juncton to the solar cell appled voltage. Swtchng the capactor on mples the PV generator suddenly changng from opencrcut to shortcrcut condtons. Hence, the length of the depleton zone must sharply reduce. Ths process lasts typcally for less than 0.2 ms and does not affect the measurement. In fact, t only appears n I V curves when they are captured wth a relatvely hgh samplng frequency. Here, we have used 12.5 khz. On the other hand, as a precauton to prevent any dfference between the voltage reached at the capactor termnals and the real opencrcut voltage, the latter s also measured just before the chargng process. Once obtaned, the curves were extrapolated to STC n accordance wth IEC60891 (procedure 1) usng the current and voltage temperature coeffcents gven by the manufacturer, whch are a = 0.04%/ K and /? = 0.33%/ K, respectvely. V oc and / sc have been calculated by lnear extrapolaton from the ponts around them [1,11]. The seres resstance (R s ) was supposed constant through all the operatng condtons and estmated assumng a varable fll factor, followng the equatons proposed by Green [12]. Ths method has demonstrated a good approxmaton [13]. The curve correcton factor (K) was fxed at 1.25 x 10 2/ C, whch s a typcal value for crystallne slcon cells [14]. Fgure 5 shows the I V curve under real operaton condtons (dark lne) and once extrapolated to STC (lght lne). Table I presents the operaton condtons and the man characterstcs of one of the IV curves, whereas Table II summarzes the mean values obtaned after sx measurements. The average maxmum power of the PV generator resulted P^ mc6089 = 502761 W. Table I. Operaton condtons and man results of one of the /l/curves obtaned. Measured values Standard test condton values Locaton Date Hour G [W/m 2 ] D [W/m 2 ] 7c [ C] Wnd speed [m/s] Ar mass CHARGE [ms] Lorca (SE of Span) 18 J anuary 2012 13:24 899 70 47.2 <1.5 1.98 26.2 fee [A] V oc [V] /MIA] l/ M[V] PM, IEC60891 [W] FF PM.S [W] K [Of C] flsiol 709.0 780.0 650.0 614.4 400550 0.724 400550 1.25x 0.091 10~ 3 791.5 846.1 731.8 687.6 503216 0.751 495133 1.25x10~ 3 0.071 Mf 1 [ '.;:: " 2 " 4:; 4? f H 1 X X H *. Fgure 4. Evoluton of / and l/durng the chargng process. The dsplacement current peak can be observed at f = 0.
» ~t... J. mc e: t»" 3 Hfl xa j» <m Hfl MO TOQ Hfl Fgure 5. /I/curve measured (dark lne) and extrapolated to standard test condtons (STC) (lght lne). Table II. Mean and standard devaton values of the /l/curves fee* [A] V oc [V] 'M [A] V M [V] PM, IEC60891 FF PM,S [W] [W] obtaned. 797.2 841.4 737.1 682.1 502761 492182 0.750 M 3anA 5.1 5.6 13.7 12692 12497 4.9 0.010 It s worth mentonng that the uncertanty of the measured power n one curve s lower than 1.4% and the one of the extrapolated power s lower than 3.6%. These values have been calculated followng a type B evaluaton as establshed by the "Gude to the Expresson of the Uncertanty n Measurement" [15]. The man uncertanty factors have been the calbraton of the reference modules and the temperature coeffcents [16]. Another way of obtanng the maxmum power at STC s to calculate the maxmum power from the measured curve and then translate t by usng PMX l x r S(T C where subscrpt "M" means measured, superscrpt "*" means STC, and c5 = 0.45%/ K s the power temperature coeffcent gven by the manufacturer [3,17]. Despte ts smplcty, there s expermental evdence of ths equaton beng as good as more complex ones [18]. Ths way, the average maxmum power resulted P Mt = 492182W. The dfference between both extrapolaton methods (2.1%) s small and gves an dea of the uncertanty assocated wth these procedures and of the coherence of the temperature coeffcents. As t has been presented n prevous works, IV curve measurements can be compared wth the ones made usng a wattmeter [3]. For ths PV generator, the analyss wth the wattmeter results n an extrapolated maxmum power of PM, watt = 501407 W. The small dfference between both procedures (0.3%) ndcates coherence and valdates the results obtaned. 4. CONCLUSIONS Ths paper presents the measurement of the IV curve of a 500kW PV generator by means of an ownmade capactve load. It has been shown that IV curve analyss can also be appled to bg PV generators and that, when takng some precautons (especally regardng the 7c), t s stll a useful tool for characterzng them and therefore can be ncorporated nto mantenance procedures. ACKNOWLEDGEMENTS The authors would lke to acknowledge Gehrlcher Solar, owner of the plant, for gvng us the possblty of measurng these curves. REFERENCES 1. IEC 60891. Photovoltac Devces. Procedures for Temperature and Irradance Correctons to Measure IV Characterstcs. 2nd edn, Internatonal Electrotechncal Commsson: Geneva (Swtzerland), 2009; ISBN: 2831810731. 2. Muñoz J, Lorenzo E. Capactve load based on IGBTs for onste characterzaton of PV arrays. Solar Energy 2006; 80: 14891497. 3. MartnezMoreno F, Lorenzo E, Muñoz J, Moretón R. On the testng of large PV arrays. Progress n Photovoltacs: Research and applcatons 2011. DOI: 10.1002/pp.ll02 4. Moretón R., Lorenzo E., Martínez F. Feld performance of PV modules qualty control process. 23rd
European Photovoltac Solar Energy Conference, 28752877, Valenca, Span (2008). 5. Caamaño E, Lorenzo E, Zlles R. Qualty control of wde collectons of PV modules: lessons learned from the IES experence. Progress n Photovoltacs: Research and applcatons 1999; 7: 137149. 6. Kng DL, Boyson WE, Kratochvl JA. Photovoltac array performance model. Sanda Natonal Laboratores, Report SAND20043535. (2004) 7. Garca M, Marroyo L., Lorenzo E, Perez M. Solng and other optcal losses n solartrackng PV plants n Navarra. Progress n Photovoltacs: Research and Applcatons 2010; 19(2): 211217. DOI: 10.1002/ pp.1004 8. http://eportal.magrama.gob.es/websar/seleccon ParametrosMap.aspx?dst=l. 9. Blaesser G, Munro D. Gudelnes for the assessment of photovoltac plants. Document C: Intal and perodc tests on photovoltac plants. Jont Research Centre of the European Communtes, Ispra Establshment, Report EUR 16340 EN. (1995). 10. Fabero F, Vela N, AlonsoAbella M, Chenlo F. Characterzaton of recent commercal technologes of PV modules based on outdoor and ndoor IV curve measurements. 20th European Photovoltac Solar Energy Conference, 20592062, Barcelona, Span.(2005) 11. Luque A, Hegedus S. Handbook of Photovoltac Scence and Engneerng. 2nd edn. Chapter 18, Wley: Hoboken, NJ, USA, 2011; 817821. 12. Green M. Solar Cells. Prentce Hall: Kensgton. Chapter 5, 1982; 9598. 13. Polvern D, Tzamals G, Mllejans H. A valdaton study of photovoltac module seres resstance determnaton under varous operatng condtons accordng to IEC 60891. Progress n Photovoltacs: Research and Applcatons 2012; 20: 650660. DOI: 10.1002/pp.l200 14. IEC 60891. Procedures for Temperature and Irradance Correctons to Measure IV Characterstcs of Crystallne Slcon Photovoltac Devces, 1st edn, Internatonal Electrotechncal Commsson: Geneva (Swtzerland), 1994. 15. ISO/IEC Gude 100:2008. Evaluaton of measurement data Gude to the expresson of the uncertanty n measurement. 2008. 16. Makrdes G, Znsser B, Norton M, Georghou GE, Schubert M, Werner JH. Error sources n outdoor performance evaluaton of photovoltac systems. 24th European Photovoltac Solar Energy Conference, 39043909, Hamburg, Germany. 2009. 17. Osterwald CR, Translaton of devce performance measurements to reference condtons. Solar Cells 1986; 18: 269279. 18. Fuentes M, Nofuentes G, Agulera J, Talavera DL, Castro M. Applcaton and valdaton of algebrac methods to predct the behavour of crystallne slcon PV modules n Medterranean clmates. Solar Energy 2007; 81: 13961408. DOI: 10.1016/j. solener.2006.12.008