Loughorough University Institutionl Repository Influence of spectrl irrdince mesurements on ccurcy of performnce rtio estimtion in lrge scle PV systems This item ws sumitted to Loughorough University's Institutionl Repository y the/n uthor. Cittion: KRAWCZYNSKI, M.... et l, 2010. Influence of spectrl irrdince mesurements on ccurcy of performnce rtio estimtion in lrge scle PV systems. IN Proceedings of the 25th Europen Photovoltic Solr Energy Conference nd Exhiition (EUPVSEC) nd 5th World Conference on Photovoltic Energy Conversion (WCPEC 5), Vlenci, pp. 4710-4714. Additionl Informtion: This is conference pper. Metdt Record: https://dspce.loro.c.uk/2134/9217 Version: Accepted for puliction Pulisher: c WIP Wirtschft und Infrstruktur GmH & Co Plnungs KG Plese cite the pulished version.
This item ws sumitted to Loughorough s Institutionl Repository (https://dspce.loro.c.uk/) y the uthor nd is mde ville under the following Cretive Commons Licence conditions. For the full text of this licence, plese go to: http://cretivecommons.org/licenses/y-nc-nd/2.5/
INFLUENCE OF SPECTRAL IRRADIANCE MEASUREMENTS ON ACCURACY OF PERFORMANCE RATIO ESTIMATION IN LARGE SCALE PV SYSTEMS M. Krwczynski, M.B. Stroel, C.. Hierd, T.R. Betts, R. Gottschlg Centre for Renewle Energy Systems Technology (CREST), Deprtment of Electronic nd Electricl Engineering, Loughorough University, Loughorough, Leicestershire, LE11 3TU, United Kingdom Tel.: +44 1509 635327, Fx: +44 1509 635301, Emil: M.Krwczynski@loro.c.uk ABSTRACT: Understnding the qulity of irrdince mesurements is n essentil prt of PV monitoring. For precise estimtion of solr rdition ll of its properties must e considered. There re two different wys to descrie irrdince rodnd nd spectrl. Brodnd irrdince mesurements re y fr the most commonly pplied technique nd cn e undertken y the use of pyrnometers or clirted reference cells. Brodnd mesurements give integrted power over specified ndwidth. Spectrl irrdince descries not only the integrted energy of the sunlight, ut lso its distriution y wvelength. PV modules re strictly wvelength selective devices. Their spectrl sensitivity depends mostly on the cell mteril technology. Spectrl effects cn e oserved t their most extreme for thin-film -Si modules. Knowing the ccurte intensity nd spectrl distriution of the sunlight my hve significnt influence on ccurte prediction of the ville energy for different types of PV device. Keywords: spectrl response, solr rdition, performnce 1 INTRODUCTION Recovery of the spectrl irrdince from rodnd irrdince mesurements, sed on the stndrdised terrestril solr spectrum with correction for the tmospheric pth length trversed nd the solr ngle of incidence (cused y Sun s position nd site s geogrphic loction nd orienttion), my not e sufficiently precise. The mjority of the influencing prmeters such s cloudiness, reflected irrdince, ir pressure, humidity nd pollution in the tmosphere hve stochstic chrcter nd cnnot e pplied with confidence to spectrl simultion models. To introduce precise spectrl chrcteristion of the irrdince, spectrordiometer mesurements re essentil [1]. PV modules re strictly wvelength selective devices. Their spectrl sensitivity depends mostly on the cell mteril. Spectrl effects cn e oserved t their most extreme for thin-film -Si modules, which hve the widest nd gp [2]. Despite this, spectrl irrdince mesurements re lmost lwys omitted in PV monitoring systems. This is due to the reltively high cost of the spectrordiometers nd the lrge mount of dt tht must e stored nd processed. This pper discusses utilistion of spectrl irrdince dt in photovoltic monitoring systems. Bsed on rel mesurements of the environmentl conditions (irrdince, spectrl irrdince, mient temperture, reltive humidity) initil nlysis of the spectrl influences on different module technologies hs een performed. The EQE responses of the different modules were tken under considertion nd were used together with the spectrl irrdince mesurements to clculte current density ville for specific module. Photon Utiliztion Rtio (PUR) hs een introduced to express the influence of spectrl vritions for given module type. 2 APPROACH In theoreticl considertions light is descried s wve prticle clled photon. The energy of photon depends only on its frequency (1). E hc = (1) λ ( λ ) hν = [ ev ] Where h is Plnck s constnt, c is the speed of light in vcuum, υ is the frequency in Hertz nd λ is the wvelength in µm. Brodnd irrdince sensors mesure rdint power, which is the sum of spectrl irrdince received from the light source over the specified ndwidth (2). W H = E( λ ) 2 (2) Where E( is spectrl irrdince. The sme rdint power my e received y rodnd sensor s result of interction with different mounts of photons, depending on their energy (i.e. different spectrl distriutions). Rdint power does not provide ccurte informtion on the numer of photons tht interct with sensor. Genertion of electron-hole pirs depends on the sorption coefficient of the mteril, the numer of intercting photons nd thickness of the mteril (3). G = 1 xα α Ne (3) 3 Where α is the sorption coefficient, N is the numer of photons nd x is the depth in the mteril If the photon energy is equl to or greter thn the nd gp of semiconductor, photon cn excite n electronhole pir. In conventionl solr cell one photon cn generte only one electron. Photon energy greter thn the nd gp is wsted, s the excited electron will thermlize down to the nd gp edge. Although new concept of multi-excittion genertion in orgnic mterils hs een proposed, it is still fr from ppliction [3]. For currently ville photovoltic devices it is importnt to know the numer of photons tht re le to generte crriers insted of the sunlight s rdint power. Spectrl irrdince descries the distriution of rdint power over the spectrum. If the rdint power is known
for defined wvelength intervl, photon flux cn esily e clculted (4). E ( ) ( photons φ λ = ( ) 2 (4) Ep λ s m µ m Where φ( is the photon flux, E( is the spectrl irrdince nd E p ( is the energy of photon in oules (1 = 6.24150974 x10 18 ev). The spectrl response of photovoltic cell cn e descried y n externl quntum efficiency (EQE). Externl quntum efficiency is rtio etween the numer of crriers collected outside of the device to the numer of photons of given energy incident on the device (5). electrons / s EQE ( λ ) = [%] (5) photons / s By comining photon flux φ( with EQE( the spectrl current density of cell cn e clculted. ( (6) µ m A ( ) = EQE φ( λ 2 Integrting ( through the wvelengths under given spectrl photon flux gives the resultnt density of the current ville from the cell (7). A = ( λ ) 2 (7) Moreover, if the sme type of cross-clirted spectrl irrdince sensor hs een used to chrcterize module/cell EQE( nd to mesure spectrl irrdince F(, the solute clirtion ccurcy of the sensors will not hve significnt influence on the ccurcy of the photocurrent clcultion. To demonstrte the influence of the spectrl irrdince on the density of current generted y modules, the photon utiliztion rtio (PUR) is introduced (8). given device () to the current density ville under the experienced spectrum for n idel device ( T ), normlized y the rtio of the current density under AM1.5 illumintion for the given device ( AM1.5 ) to the current density ville under AM1.5 illumintion for the idel device ( AM1.5 ). 3 EXPERIMENTAL SETUP Three similr sets of reference irrdince sensors were instlled in vrious loctions. Ech of the sets contined two second clss rodnd pyrnometers, VIS spectrordiometer, up to five different PV modules (working in short circuit mode), module temperture sensors, mient temperture nd humidity sensors nd wind speed nd direction sensor. The VIS spectrordiometers were instlled in the plne of the rrys to provide ccurte reference spectrl irrdince mesurements for instlled modules. Spectrl irrdince mesurements should llow n increse in the ccurcy of prediction of the energy ville for vriety of PV device. Dedicted short-circuited modules were used for nonised reference mesurements. Such module hs the sme ngulr nd spectrl response s the modules tht re generting power. It will lso degrdes in similr wy. Short circuit current vries more or less linerly with irrdince, so it cn e used s mesure of reference irrdince. For ech of the investigted technologies, one dedicted reference short circuited module ws instlled. A module s temperture hs n influence on most of its electricl prmeters. Therml energy trnsport etween the module nd environment depends on mounting structure, irrdince, mient temperture, humidity, wind speed nd direction. All those prmeters were monitored to provide n ccurte reference for system performnce nlysis. Detiled PV plnt specifiction nd ll mesurements were stored in dtse to provide n effective method of dt nlysis. Most of the nlyses were performed y custom developed softwre, with the help of sttisticl pckges. A simplified lyout of the mesuring system is shown in Figure 1. N EQE( E( E( A PUR = = = NAM 1.5 EQE( EAM 1.5 ( E AM 1.5 ( T AM 1.5 TAM 1.5 Where N A is the rtio etween the numer of photons tht generte crriers in the device to the totl numer of photons reching the device under the experienced spectrum, N AM1.5 is the rtio etween the numer of photons tht would generte crriers in the device under AM1.5 illumintion to the totl numer of photons reching the device under AM1.5 illumintion. In terms of current, PUR cn e descried s the rtio etween the current density under the experienced spectrum for (8) I SC, T module Air T,H,S G horizontl G inplne G spectrl Dt Aquisition System Figure 1: Mesurement system lyout. Dtse Processing
Figure 2: Reference modules nd irrdince sensors instlled onsite. 4 ANALYSIS The min motivtion for the nlysis reported here ws to estlish to wht extend irrdince mesurements cn e improved to llow more ccurte estimtion of the performnce rtio (PR) of photovoltic system. The used irrdince sensors were mnufctured y different compnies. To void influences of the mnufcturer s specific clirtion, the spectrordiometer nd reference module responses were clirted with the rodnd pyrnometer. The pyrnometer ws used s n solute reference for two resons: it is clssified nd clirted in ccordnce to ISO stndrds (ISO 9060 nd ISO 9846) nd it is used s reference sensor for irrdince mesurements y module mnufcturers. Moreover, the pyrnometer sensitivity does not significntly depend on wvelength (within its sensitivity rnge), rdition intensity, mient temperture nd time. 4.1 Spectrordiometer cross-clirtion. Records used for solute clirtion were chosen on criteri given y stndrd test conditions (H=995-1005 W/m 2 nd T mient =25 0 C. From selected spectr, the men spectrl response of the spectrordiometer under STC ws clculted. Spectrl irrdince mesurements were tken over the rnge of 300-1150nm. In clcultions rnge of 350-1050nm ws used due to high mesurement uncertinty outside of this intervl. Mesured spectr were integrted over the rnge of 350-1050nm nd compred with the corresponding reltive frction (74%) of the integrted AM1.5 spectrum. The rtio etween the mesured nd stndrd integrted spectrl irrdince ws used to scle the spectrordiometer mesurements. Figure 3 shows spectrum mesured under STC efore nd fter reclirtion, in comprison to the AM1.5 spectrum. irrdince thn the reference AM1.5 spectrum specifies. Such lrge error in solute clirtion ws verified during spectrordiometer indoor tests [4]. In [4] the uthors hve demonstrted tht the shpe of the spectrum mesured y the type of spectrordiometer used here follows the spectrum of clirted reference lmp within 2% difference in given intervl. Moreover, vritions oserved in Figure 3, etween mesured nd AM1.5 spectr shpes could hve een cused y sitespecific environmentl conditions (i.e. pollution). If n externl quntum efficiency (EQE) for given module technology is known, the current of the device cn e clculted for given spectrum. Figure 4 presents spectrl photon flux clculted from the AM1.5 spectrum nd externl quntum efficiency of stndrd module technologies. photon flux density 5x10 18 4x10 18 3x10 18 2x10 18 1x10 18 0 AM1.5 spectrum mono-si -Si c-si 400 600 800 1000 1200 wvelength [nm] Figure 4: Photon flux density nd EQE of the photovoltic modules used in this study. ( (7) is the spectrl current density nd represents the numer of photons effectively intercting with device nd is shown in Figure 5 for the modules used in this work. Integrtion of ( llows clcultion of the current generted y module. spectrl current density [A/m 2 nm] 0.6 0.5 0.3 0.2 mono-si -Si multi-si 1.0 0.8 0.6 0.4 0.2 0.0 EQE E([ W m -2 µm -1 ] 1.6 1.4 1.2 1.0 0.8 0.6 0.4 AM1.5 stndrd mesured clirted 0.0 400 600 800 1000 1200 wvelenght [nm ] Figure 5: Spectrl current density for different modules. By integrting spectrl current density through the nd of 300nm to 1200nm, STC current densities for the investigted module types ws clculted nd compred with dt from mnufcturers dtsheets. (Tle I) 0.2 0.0 300 400 500 600 700 800 900 1000 1100 λ[nm ] Figure 3: Mesured nd stndrd glol spectrl irrdince (tilted). As shown, non-clirted spectrordiometer mesurements correspond to pproximtely 12% lower [A/m 2 ] EQE Dtsheet mono-si 366.5 371.0 -Si 142.4-262.5 260.0 multi-si 337.8 333.3 AM 1.5 464.5 464.5
Tle I: Current density for different cell technologies under stndrd test conditions. The mesurement rnge of the used spectrordiometer lys etween 350 1050nm. Current densities shown in Tle I were reclculted for tht ndwidth (Tle II). [A/m 2 ] EQE sed Spectrl Irrdince sed Err [%] mono-si 341.78 341.5-0.08 -Si 141.1 143.1 1.41 260.9 261.3 0.15 multi-si 325.0 324.7-0.9 AM 1.5 409.24 409.24 0 Tle II: Current density for different cell technologies under stndrd test conditions. Tle II shows tht n estimtion of the current density (for conditions close to STC) sed on module EQE nd mesured spectrl irrdince gives similr results to estimtes sed on module EQE nd reference AM1.5 spectrl irrdince. This vlidtes the spectrordiometer clirtion. 4.2 Reference module clirtion For most of the used reference modules, the mnufcturer`s flsh dt were ville. Flsh tests re performed with solr simultors. The prmeters of the light emitted y the simultors does not fully correspond to the prmeters of sunlight. Module short circuit current (I sc ) ws mesured outdoors t STC to check if the mnufcturer s prmeters corresponded with those mesured under rodnd reference irrdince. In most cses stndrd test conditions do not correspond with rel operting conditions (i.e. temperture of the loded module under STC irrdince of 1000W/m 2 will e rrely equl to 25 0 C, especilly in southern loctions). Depending on site specific conditions, different method s of short circuit current extrpoltion should e used. For this nlysis the northern method ws used. Conditions pplicle for record extrction re given in Tle III. Device Reference module (southern country method) Reference module (northern country method) T module H T mient [W/m 2 ] [ 0 C] [ 0 C] 995-105 X X 950-1050 X 24-26 Tle III: Record extrction condition for Isc extrpoltion. Tle IV presents short circuit currents extrpolted from outdoor rodnd irrdince mesurements t STC nd mnufcturer supplied flsh test dt. Module type I sc I sc Difference (flsh) (extrcted) (%) µ-si 2.20 2.192 0.36 m-si 5.79 5.651 2.40 p-si (1) 8.37 8.069 3.60 p-si (2) 8.07 8.027 0.53 2.39 2.408 0.75 Tle IV: Short circuit current for different modules (mnufcturer specified vs. mesured t STC). 4.4 Photon utiliztion rtio The min ide tested y this work is tht, since module output current depends on the spectrl chrcteristics of the illumintion, spectrl mesurements could e pplied to plnt monitoring to provide more ccurte predictions of reference irrdince. This prmeter hs the most significnt influence on the ccurcy of PR estimtion. The spectrl chrcteristics of sunlight vry over the course of dy s well s with the time of the yer. These vritions hppen due to the thickness of tmosphere tht the light hs to trvel through overcst conditions nd erosols contined in the tmosphere. Figure 6 shows mesured men spectr for two different months nd for specific dy. normlised men spectrum [ritrry units uly( m en) uly(one dy men) nury (men) 300 400 500 600 700 800 900 1000 1100 wvelength [nm ] Figure 6: Spectrl irrdince vrition over time. As shown in Figure 6, differences in spectrl chrcteristics cn e significnt nd will hve n influence on module performnce. To demonstrte this, current densities for different module technologies were clculted under spectr experienced during the mesurement cmpign. To quntify utiliztion of the spectrum y specified technology PUR ws introduced (8). Dily distriutions of PUR were clculted for different dys nd re shown in Figure 7 (cler dy) nd Figure 8 (fully overcst dy). PUR[ritry] rodnd irrdince[w/m 2 ] 1.15 1.10 1.05 1.00 0.95 0.90 1000 500 0 Irrcince -Si mono-si multi-si 15/08/2009 10:54:00 15/08/2009 15:55:00 15/08/2009 10:54:00 15/08/2009 15:55:00 Figure7: Distriution of PUR over the course of dy for cler dy.
rodnd irrdince[w/m 2 ] PUR[ritry] 1.10 mono-si multi-si -Si 1.05 1.00 0.95 300 200 100 0 05/10/2009 09:57:05 05/10/2009 11:57:05 05/10/2009 09:57:05 05/10/2009 11:57:05 Figure 8: Distriution of PUR over the course of dy for hevily overcst dy. [4] M. Krwczynski, M. B. Stroel, R. Gottschlg, Intercomprison of Spectrordiometers for Outdoor Performnce Monitoring, Proceedings 24 th Europen Photovoltic Solr Energy Conference, Vol. 1 (2009) 3406 3408 5 CONCLUSIONS Spectrl effects cn e oserved for photovoltic devices. However t the present time it is reltively difficult to use spectrordiometers y themselves s reference irrdince sensors for PV systems. This is due to lck of clssifiction nd clirtion stndrds tht spectrordiometers could e mnufctured to. Additionl clirtion sed on stndrdised rodnd irrdince sensors hd to e followed to chieve ccurte solute spectrl mesurements. Short circuited reference modules seem to e good lterntive for spectrl irrdince mesurements. When the sensing nd generting devices re of the sme technology spectrl effects should e utomticlly compensted. It ws shown tht module prmeters specified y mnufcturers re not lwys precise. Without dditionl clirtion reference modules will not provide ccurte estimtion of ville irrdition. Moreover ech technology to e monitored requires seprte reference module or cell. To quntify the influence of spectrl irrdince on module the photon utiliztion rtio hs een introduced. It llows presenttion of the utiliztion of the mesured spectrum y different types of modules in reltion to the AM1.5 spectrum. It does not depend on the solute vlue of irrdince nd presents the spectrlly dependnt response of the module s singulr numer. The lrgest spectrl influences were oserved for -Si modules. Amorphous silicon hs reltively wide nd gp, which mkes its spectrl response correspondingly nrrow. Due to this, vritions in the spectrl irrdince hve stronger influence on this technology thn on ny of the others reported here. 6 REFERENCES [1] T.R.Betts, Investigtion of Photovoltic Device Opertion under Vrying Spectrl Conditions, PhD Thesis, Loughorough University (2004) [2] R.Gottschlg, T.R.Betts, D.G.Infield, M.. Kerney, The effect of spectrl vritions on the performnce prmeters of single nd doule junction morphous silicon solr cells, Solr Energy Mterils nd Solr Cells Vol. 85 (2005) 415-428 [3] Lurens D. A. Sieeles, Orgnic solr cells: Two electrons from one photon, Nture Chemistry, Vol.2 (2010) 608-609