PERFORMANCE OF GRID COUPLED PV-ARRAYS BASED ON CIS SOLAR MODULES

Transcription

1 PERFORMANCE OF GRID COUPLED PV-ARRAYS BASED ON CIS SOLAR MODULES F. H. Karg, D.Kohake, T. Nierhoff, B. Kühne, S. Grosser and M.Ch. Lux-Steiner : Siemens and Shell Solar GmbH, Otto Hahn Ring, D-79 München, Germany : Fachhochschule Gelsenkirchen, Neidenburgerstraße, D-77 Gelsenkirchen : Fachhochschule Flensburg, Kanzleistraße 9-9, D-9 Flensburg : Hahn-Meitner Institut Berlin, Glienickerstraße, D-9 Berlin ABSTRACT: Grid coupled CIS-PV arrays with total power values between 7 W and W were installed in three different locations in northern Germany and monitored within an elaborate and detailed program over two years in the field. The arrays are configured of commercial ST and ST CIS thin film modules from Siemens & Shell Solar. For every array the module temperature, the solar irradiance in the plane of the array, DC-and AC-power and energy yields were recorded. Based on this data set we calculated the efficiency of the array under normal operating conditions (NOC), its temperature coefficient for power and the extrapolated efficiency under Standard Operating Conditions (STC). For the ST based array average efficiency in the field over the last two years was 7. %. For the larger ST modules it was between.7-. % on the average for the two arrays under test. Part of the difference is due to the lower active-to-total area ratio for the smaller module. Further data analysis have yielded a temperature coefficient for array power between. and. %/K comparable to crystalline silicon modules. Extrapolated efficiency at STC is.-9 % for the ST based array and 9. % for the ST setup. The CIS arrays under test have shown yearly energy yields (DC) of between 9 and kwh/kwp/y. Those Values have been above the crystalline silicon based PV arrays at the same location. Possible reasons of high yields of CIS in the field performance may include their high performance even at lower illumination levels. Keywords: : PV System Monitoring : CuInSe - : Qualification and Testing. INTRODUCTION Grid coupled PV arrays have seen tremendous growth rates in the past years especially in Japan and Germany due to subsidy and market introduction programs For these installations the total amount of electricity per year and installed PV capacity (kwh / kwp) is the key figure of merit and allows estimates on the economic pay back time for the investment. Today most of the installed PV capacity is based on mono- and multicrystalline silicon solar modules. They have shown high reliability and yield, however their high manufacturing costs motivated all major PV producers to develop alternate technologies based on thin films with potentially lower cost. One of the most pomising new technologies that have emerged over the past years is based on chalcopyrite semiconductor absorbers such as CuInSe (CIS) as the most prominent example. First products based on CIS-technology have been introduced to the market by Siemens Solar in 99. Primary advantage of CIStechnology is their high efficiency which is unmatched by any other thin film technology. CIS technology schematics and test results obtained at NREL with various generations of prototype modules over more than ten years has been subject of prior publications [,]. CIS products were initially limited to W modules (ST ) targeted mainly for small scale applications such as battery charging. The introduction of the W CIS (ST ) modules two years ago has opened up the path to some larger gridcoupled PV installations with this new technology. The largest CIS array with kwp was installed on the new congress center in Salzburg /Austria ( the CIS array closest to this conference venue with a total power of 9 kw p is to be seen near the monastery of Benediktbeuren, some km south of Munich. CIS test arrays based on small ST modules and large ST modules were also installed in three german university and research institutions accompanied by a complete monitoring program. Two of the test sites also include other PV arrays based on silicon technology for comparison purposes. Some results on these arrays will be presented for comparison in this paper as well. Primary focus of our CIS data evaluation is on those performance parameters, that are of most practical relevance namely operating efficiency (calculated on the basis of the whole array area) and average daily yield in terms of kwh/kwp. Table: Specifications of CIS ST / Modules used in the PV arrays presented in this work ST * ST Power **..9 ** Voc [V]. Isc [A]..9 dvoc/dt [%].. disc/dt [%].. dp/dt [%]***.. Dimensions [m ]. x..9.. * cell module (early product version, now cells) ** Early product version, now W and 9. % *** this work

2 . ARRAY MONITORING SCHEMATICS Module specifications for the two types of CIS modules used in the arrays under investigation are listed in table. The ST modules wired in the arrays are of the standard configuration i.e. as glass-glass laminates with black anodized aluminum frames. Since deployment of these early product generations module configuration and performance has been improved. Today s modules have a reduced number of series connected cells ( instead of ) and an associated higher Jmpp to better tune module voltage for V applications. Also the module (total area) efficiency for the ST product was increased from.9 to 9.. Table with the CIS specifications demonstrate, that temperature coefficients are quite similar to crystalline silicon. Also in a second aspect - namely their spectral response - the two technologies show similar properties as displayed quantitatively in Fig.. individual PV array, on-line data and a web camera view of the test site. Flensburg (University of Applied Science) The PV test site is placed on the roof of a university building (fig. ). The CIS array comprises ST Modules with a total of 9 W STC power according to specification (see table ). strings of modules in series are cabled into a Siemens Inverter SPN. As with the array described above Hall compensation sensors are used for DC current measurements. To compare CIS performance in Flensburg with traditional Si based technologies we use one of the PV arrays on the same roof (fig., up front) based on mono crystalline SM / Volt modules (Siemens) with a total array power of. kw. Further details on all arrays under test and an on-line visualization of their present day performance can be directly obtained via internet under Quantumefficiency.... Cu(InGa)(SSe) CZ-Si Wavelength [nm] Fig. : Typical Spectral Response of monocrystalline silicon modules in comparison to CIS (i.e. Cu(InGa)(SSe) ) modules as used for this work on field performance. The configuration of the grid coupled CIS arrays set up in three locations will subsequently be described in more detail. To make data more comparable and independent from the BOS components used in various locations we present exclusively DC yields. The corresponding AC yields should be roughly % below the DC yields ( % of which attributed to the inverter and % to the cabling.) Gelsenkirchen (University of Applied Science) At a test site near the Bocholt branch of this university various PV and combined cycle power technologies are monitored and evaluated (Fig. ). The PV section has mono-, multicrystalline and amorphous silicon as well as CIS modules in test. The CIS array comprises 7 ST Modules consisting of strings with two modules in series each. DC and AC power is monitored by separate sensors independently from the inverter (WE, Würth Electronic). In addition solar radiation in plane of the array (by an ESTI sensor), global irradiation and module backside temperature (with a PT thin film sensor) are recorded. Five minute averages of all data are stored by a Data Taker DT module and fed into a TCP/IP network by an embedded web server. A visualization of the complete system is available via internet ( including descriptions of every Fig. : Overview of PV test array in Bocholt comprising crystalline (tracking and fixed), amorphous silicon and CIS modules Hahn- Meitner Institute The third CIS array is located in Berlin at the Wannsee division of the Hahn-Meitner Insitute (Fig. ). The array consists of ST modules distributed on strings adding up to a total power of Wp In addition to the electrical parameters of the array and the backside temperature of the modules a complete set of weather data (wind, ambient temperature) for this location. These additional data would allow corrections of the module temperature with ambient temperature and wind speed to better approximate the real junction temperature of the CIS modules. However in this work, we omitted these corrections in order to allow for direct data comparisons with the other arrays. DC current values in this array are taken directly from the NEG inverter (Solon). Preceding measurments have made shure that those current measurements are subject to less than % measurement errors for currents above % of rated power.

3 Power [W/m ] / [W/kW p ] 9 7 CIS-DC-Power 9 [W/kWp] Irradiation [W/m²] 7 CIS-Temperature [ C] Module Temperature [ C] : : [h] 7: : Fig.: Irradiation, DC-Power and module backside temperature over the course of a sunny day in Bocholt (June /) Fig. : Overview of PV test array at the roof of University of Applied Science in Flensburg comprising monocrystalline (front), CIS (middle) and multicrystalline Si modules (back). Subsequent three graphs show average daily energy yields within a particular month for the three arrays and the corresponding operating efficiency. Months with no data are due to problems with the data tracking system or with BOS components (for instance due to lightning damage). If only few days in the data recording have been missing, a linear extrapolation to the full monthly yield was made. The resulting average efficiency of the ST array was 7. % and of the two larger ST arrays consistently between.7 resp.. %. Fig. : CIS PV array at the Wannsee site of the Hahn- Meitner Institute in Berlin. RESULTS A typical example of raw data for PV array DC power, module temperature and insolation taken over the course of a sunny day in Bocholt is shown in Fig.. Integrated energy values of electric power and irradiation (kwh) will be used to determine daily yield and conversion efficiency of the array. In addition the measured module temperature is used to extract array power under standard conditions ( C, W/m ) from the measured DC power at lower irradiation levels and higher temperatures (see below). kwh/kwp/d CIS [ST ] January March May Operating Eff. July September November January March May July Fig.: Normalized daily yield (kwh/kwp/d) and operating efficiency over the course of two years for the CIS array in Bocholt based on ST modules. Average operating efficiency during the observation period was 7. %. A summary table on CIS field performance is given below. We have chosen a month period where all arrays had a high uptime and nearly complete data sets are available. Operating efficiency of the two ST arrays is in good agreement. As explained previously, the lower efficiency of the ST module is due to the lower activeto-total area ratio.

4 kwh/kwp/d CZ-Si-Reference CIS [ST ] Operating Eff. CIS Februar April Juni August Oktober Dezember Februar April Juni August Fig.7: Normalized daily yield (kwh/kwp/d) and operating efficiency for the CIS array in Flensburg based on ST modules in comparison to a crystalline Silicon array. Average operating efficiency for CIS during this time window was. %. amorphous silicon arrays. The origin of the widely different performance of the two (multijunction) amorphous silicon arrays is subject of current investigations. Preliminary I/V measurements of these arrays seem to indicate that one a-si array is performing significantly above, the other significantly below specification. Table : Summary of yearly DC yield for the three CIS arrays (may april ) Location Av. Operating KWh/kWp/y Efficiency Bocholt, ST 7. 9 Flensburg, ST. 97 Berlin, ST.7 Bocholt: Total Yield [kwh/kwp] /99-/ kwh/kwp/d 7 CIS: ST Apr Jun Aug Operating Eff. Okt Dez Feb Apr Jun Aug Fig.: Normalized daily yield (kwh/kwp/d) and operating efficiency over the course of two years for the CIS array in Berlin based on ST modules. Average module efficiency during the observation period was.7 %. Yearly energy yields (DC) of 9- kwh/kwp would correspond to routhly and 9 kwh/kwp of AC power fed into the grid. These performance values are certainly above average in this part of Germany accoording to previous experience with the -roof-top-program results []. For comparison we now also had a look on alternative, silicon-based PV technologies and their performance. Fig. 7 above has already demonstrated a direct comparison with a mono-silicon array array next to the CIS-array. In this particular example, the Si array performed consistently more than % below the CIS array. As a second example multi and monocrystalline arrays have been compared with the CIS installation in Bocholt. In Fig. 9 a summary chart on DC energy yield of the different technologies is given. CIS shows higher performance than the multicrystalline and one of the mc-si a-si_ a-si_ CIS Fig.9: Total Yield of different PV technologies at the test site in Bocholt. (Array power as specified.) Preliminary I/V characterization of the two a-si arrays (from different vendors) indicate, that their widely different performance is due to under- resp. overspecification. The efficiency under normal operating conditions (NOC) is certainly of the most practical value to plan energy yield for a given installed nominal power. We determined in addition array power under Standard Test Conditions ( C, W/m ) though to better assess stability and conformity of the array power with its specifications even after several years of outdoor exp osure. In order to extrapolate from NOC to STC we first determined the temperature coefficient of a normalized array power (P ). This has been obtained by a linear correction of measured array power up to a standard irradiation of W/m at different temperatures but always high irradiance values (> W/m ). The whole set of P values for the arrays in Bocholt and Berlin is plotted in fig. over their module temperature. The temperature coefficient was extracted from a linear least square fit to these data. Results for both arrays are between. and. [%]/ K. The temperature coefficient determined by this data set can also be plotted for the same data set as a function of time to detect potential long term drifts in arra power. This has been done for the ST array in Bocholt as demonstrated in Fig. below. Although there is some scattering in this data especially in the first year period the array power seems to perform very stable.

5 Array Power [W] HMI [PSTC: W; dp/dt: =. %] Bocholt [P STC: 77 W; dp/dt: =. %] 7 Module Back Side Temperature [ C] Fig.: Array power normalized to W/m irradiation for installations in Berlin () and Bocholt (/) as a function of backside of module temperature. The PSTC values obtained in this extrapolation would correspond to efficiencies of 9 % (ST ) resp. 9. %. As a final cheque for the consistency of our data we determined the STC power of the complete array a with a portable I/V curver tracer. The measured I/V characteristics under NOC (about W/m ) was transformed to the I/V under standard test conditions according to a method first introduced by Blaesser []. In this case the extrapolated power is slightly lower (-%) as compared to previously described methods, however still within expectations. An overview of the results for Pstc obtained by the various methods in comparison to the specified power is given in the subsequent table. With all three methods the extrapolated STC power of the array is above the specified minimum value of 7 W. STC-Power [W] <7 W> Mrz Jan Nov Fig. 9: ST array power (Bocholt) normalized to STC. Average array power according to this evaluation is 7 W which would correspond to a STC efficiency of. %. Table :Summary of array power under STC determined by extrapolating array power resp. measured I/V characteristics. For comparison also the specified power is given. Bocholt ST Specified power 7 I/V measurement (extrapolated from 77 NOC) P extrapolation 77 P STC average 7. CONCLUSION AND OUTLOOK Grid coupled CIS arrays installed in three different locations in Germany have been characterized over.- years by their daily and yierly energy yield, operating efficiency and power under standard test conditions (STC, C, W/m ). STC efficiency before and after two years of field exposure exceed the specified minimum power guaranteed by the vendor. Although temperature coefficients and spectral response of CIS closely matches those of crystalline silicon, energy yields have been higher for CIS in this test. The results clearly show, that high CIS efficiency at STC transforms into high energy yields that are above average of existing technologies. Most recent CIS products have increased efficiencies still further as compared to the first product generation under test in this publication. In this investigation average efficiency under normal operating conditions (NOC) achieved between and 9 % of the specified efficiency under STC. Two likely reasons for this high performance ratio will have to be taken into account:?? Module and array efficiencies in practice are above minimum specified value (see table ).?? High module efficiency is maintained even at low illumination intensities. The latter has been confirmed by a preliminary survey among fabricated modules. Typically fill factor of modules goes through a maximum at W/m. This module characteristics will have to be further examined under different illumination intensities and spectra. ACKNOWLEDGEMENT Siemens & Shell Solar greatly acknowledge funding from the European Union Joule III program, the German Ministry of Economics, the Bavarian Research Foundation and Ministry of Economics, Traffic and Technology. REFERENCES [] Tarrent, D. E., Wieting, R. D. th Sunshine Workshop Tokyo, Japan, Feb., [] Karg, F. H., Sol. Energy Mat. Sol. Cells (), - [] Fraunhofer ISE, -Dächer Meß- und Auswerteprogramm, Jahresjournal 997 [] Blaesser, G., Krebs, K., Ossenbrink, H., Verbaken, J., Proc. th EC PVSEC, (9)