Baixe o app para aproveitar ainda mais
Prévia do material em texto
PV Systems in Warm and Sunny Climates: Performance Assessment of Commercially Available Solar Photovoltaic Technologies under Different Climatic Conditions in the Brazilian Energy Mix Lucas Rafael do Nascimento, Marília Braga, Ruany Dolla, Rafael Antunes Campos and Ricardo Rüther Universidade Federal de Santa Catarina (UFSC), Campus Universitário Trindade, Caixa Postal 476, Florianópolis – SC, 88040-900, Brazil Abstract — The performance assessment of seven different, commercially-available PV technologies was carried out at eight different climates in Brazil. The eight identical, fully-monitored 70 kWp photovoltaic Evaluation Sites (ESs) have all electrical and environmental parameters measured at one-second intervals. The R&D project, funded by twelve Brazilian Electric Utilities in the scope of the Brazilian National Regulatory Agency’s R&D Program, aims at providing performance information for Utilities to make informed decisions on which PV technology to adopt in the forthcoming utility-scale solar energy auctions taking place in Brazil. The PV technologies evaluated are: thin-film amorphous silicon (a-Si), microcrystalline silicon (a-Si/μc-Si), cadmium telluride (CdTe), copper indium gallium diselenide (CIGS), and bulk mono and multi-crystalline silicon (c-Si and m-Si), all at fixed, latitude tilt, as well as double-axis tracking, concentrated PV (CPV) using InGap/GaAs/Ge at 820 suns concentration. Index Terms — cloud enhancement, PV performance, utility- scale solar, O&M, PID. I. INTRODUCTION With the declining costs of photovoltaics (PV), and the excellent solar energy resource availability in the country, the Brazilian government and the electricity sector have started to evaluate and consider PV as a serious potential contributor to the national electricity mix. Since the late 1990’s, Brazilian electrical utilities are required by the National Electrical Energy Regulatory Agency (ANEEL) to invest 1% of their operational income on R&D. In 2011, ANEEL issued an R&D call dedicated to utility-scale PV. The solar energy research group at the Universidade Federal de Santa Catarina (www.fotovoltaica.ufsc.br) has been actively investigating and promoting PV in Brazil, and operates since 1997 the first grid- connected, thin-film PV generator in the country [1]-[2]. This paper shows the results of this utility-scale R&D project with the PV assessment of six fixed, flat-plate PV technologies, as well as double-axis tracking concentrated PV (CPV) installed at eight different sites, with distinct Brazilian climatic conditions. The results also present some of the peculiarities observed during the continuous and high-temporal-resolution monitoring of PV generators at all these warm and sunny sites. Cloud-edge and cloud-enhancement effects of solar irradiance resulted in operational issues that were not previously described in the literature. This information is of great scientific and economic interest as the PV technology increases its penetration and is part of the strategic planning of the national electricity sector in Brazil. II. EXPERIMENTAL SETUP AND METHODOLOGY Eight identical fully-monitored 70 kWp PV Evaluation Sites (ESs), each using seven different, commercially-available PV technologies (single- and multi-crystalline silicon, two-axis- tracking 820 suns concentrated PV, and the thin-film PV technologies CdTe, CIGS, a-Si and a-Si/μc-Si), were installed in different regions of Brazil between 2013 and 2014. Fig. 1 shows the map of Brazil and the location of the eight ESs, while Fig. 2 shows an aerial view of one of the eight ES. The only difference between the eight ESs is the module tilt angle for the fixed-tilt arrays at each site, which is equal to the local latitude in each case. The ESs are constantly monitored, with irradiance, temperature (ambient and back-of-module) and electrical parameters measured and logged at one-second intervals. High- quality, research-grade and state-of-the-art irradiance, electrical parameters and temperature measurement devices and dataloggers were used. A more detailed description of the eight locations, the experimental set up and equipment specifications were presented elsewhere [3]. Fig. 1. General location in the Brazilian territory of the eight identical R&D ESs, located between 3°S and 28°S, each with 70 kWp of PV installed power evenly distributed among seven different commercially-available PV technologies. 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0103 Measuring environmental and operational PV electrical parameters at the individual PV string level, and at one-second time resolution at all these sites leads to very large amounts of data, which have to be verified and qualified every day. This high time-resolution was chosen in order to enable the detection and evaluation of fast-changing environmental conditions (particularly irradiance, which was measured with Kipp and Zonen SMP11 pyranometers and IMT reference cells), and their influence on the response and performance of PV devices operating in a utility-scale PV power plant environment. The decision of measuring all parameters at the one-second time resolution was made in the context of our previous experience with solar irradiance variations in a sunbelt region, in contrast with that in higher latitude climates [4]-[6], and the effects of a larger fraction of solar energy availability at higher irradiances on PV generator inverter sizing [7]-[9]. Fig. 2. Aerial view of one of the eight identical R&D ESs, with 70 kWp of PV using seven different commercially-available PV technologies. IV. RESULTS AND DISCUSSION: PV PERFORMANCE ASSESSMENT In a warm and sunny environment, temperature is the main cause of energy losses in a PV system. Table I shows the temperature coefficient of maximum power point for the PV module technologies used in this study and Table II presents the irradiance-weighted back-of-module average temperature of the six fixed PV module technologies at each of the eight ESs during the analyzed period. The AC Performance Ratio (PR) assessment of the PV technologies installed in the ESs is presented in Fig. 3. The analyzed period includes only simultaneous valid data for all the PV technologies on each ES; this is the reason why some sites have only a few months of available data. TABLE I TEMPERATURE COEFFICIENT OF MAXIMUM POWER POINT FOR THE PV MODULE TECHNOLOGIES USED IN THIS STUDY, SORTED FROM LOWEST TO HIGHEST. TABLE II IRRADIANCE-WEIGHTED BACK-OF-MODULE TEMPERATURE AVERAGE OF THE SIX FIXED PV MODULE TECHNOLOGIES AT EACH OF THE EIGHT ESS DURING THE ANALYZED PERIOD. For the presented data it can be observed that the thin-film a- Si technology resulted in a superior PR than the other technologies for most of the ESs. This behavior is due to the lower temperature coefficient of this technology, which results in good performance in warm climates [10]. A bluer spectral content of sunlight in the region might also have been beneficial for these blue-biased PV devices. The analyzed period was after stabilization of the Staebler-Wronski effect (SWE), which affectsthe output performance of thin-film a-Si PV modules strongly during the first year of outdoor operation and stabilizes after some 1000 kWh/m2 of sunlight exposure [11]. For c-Si and m-Si technologies, a good performance is observed in most of the ESs, with both technologies presenting very similar PR. However, a few ESs have presented a marked discrepancy between the two technologies. Field evaluations revealed intense PID degradation in coastal areas with high relative temperate and humidity at such sites. Fig. 4 shows the effect of PID in two p-type multicrystalline silicon strings from the Capivari de Baixo ES (28°S, 49°W), located on the coast of southern Brazil. 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0104 Fig. 3. AC Performance Ratio (PR) of the PV technologies installed at the eight ESs in different climates in Brazil. Fig. 4 (a) shows the mapping of the two strings: string M outlined in blue, and string P outlined in orange, with 13 and 16 modules, respectively. In the image, modules in green belong to the positive end of the string, while modules in red belong to the negative end. Fig. 4 (b) shows the electroluminescence (EL) images made for the same two strings. Finally, Fig. 4 (c) shows the measured power relative to the healthiest module on the string, obtained from individually measured I-V curves. The modules marked in black on images (a) and (c), and shaded and outlined in red on image (b), are modules that were blown away from the metallic structure during an extreme meteorological event that hit the region on October 16th, 2016 with strong winds (up to 200 km/h gusts), prior to the EL imaging and the measuring of the modules I-V curves. Many modules were severely damaged, but not all of them could be replaced, and therefore were put back on the structure but not connected back to the strings they belonged to, hence should be disregarded in this analysis. In Fig. 4 (b), it can be easily seen that the modules located towards the negative end of the strings present more darkened cells on the EL image and, therefore, presented a lower peak power when measured with the I-V curve tracer, as seen in Fig. 4 (c). Fig. 5 shows the individually measured I-V curves for each of the modules on the two strings shown in Fig. 4. A reduction in shunt resistance can be seen, as well as a decrease in Voc, reflecting the junction to be less capable of separating holes and electrons [12]. For the a-Si/μc-Si and CIGS technologies, a slight negative discrepancy between the measured power and the power declared by the manufacturer was observed at the commissioning of these technologies. (a) (b) (c) Fig. 4. PID effect on two p-type multicrystalline strings from the Capivari de Baixo ES (28°S, 49°W), located on the coast of Southern Brazil. (a) Shows the mapping of the two strings connected to independent MPPT channels: string M (outlined in blue, with 13 modules) and string P (outlined in orange, with 16 modules), with positive modules in green and negative modules in red. (b) Shows the EL images made for all the m-Si modules within the two strings. (c) Shows the measured power, obtained using an I-V curve tracer, relative to the healthiest module on the string. The modules shaded in red and black in the pictures are modules that were damaged and were thus disregarded in the study. For the CdTe technology, important advances have been made by the manufacturer First Solar since the installation of these systems [13], and the output characteristics of Series 2 modules installed at our eight ESs were considerably improved in the current Series 4 product. AC performance ratios 2-3% higher than multicrystalline silicon have been reported in the literature for Northern Brazil in a similar R&D project using CdTe Series 3 modules from the same manufacturer [14]. At the end of the project, all installed double-axis tracking concentrated PV (CPV) were inoperative or intermittent. In view of the faced problems, it was not possible to assess the long-term evaluation of this PV technology. Despite the state of maturity of CPV systems, until the deployment of this project (2013-2016) no such system had been installed in Brazil. The equipment was supplied by the company for the first time to the Southern hemisphere, which required changes in system control and hardware, which were not sufficient for the acceptable operation of the systems. A detailed energy loss analysis was also carried out through simulations regarding the Itiquira ES (17°S, 54°W), using PV sizing software PVsyst 6.72 with default parameters values, with exception of soiling losses, which were assessed on site through the measurement of I-V curves before and after cleaning the PV modules. The simulation was performed using on-site measured GHI (secondary standard pyranometer) and ambient temperature. The total measured loss and the simulated losses are presented in Fig. 6 and in Table III a summary of the results obtained in this part of the study is shown. 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0105 (a) (b) Fig. 5. Individually measured I-V curves for each of the PV modules on the two strings shown in Fig. 5. Fig. 6 (a) displays the I-V curves measured for the modules on string M, while Fig. 6 (b) displays the same for string P. Measured losses were calculated using the Performance Ratios for each technology in this specific ES using GTI data measured with secondary standard pyranometers, and energy data obtained through high-quality, research-grade energy meters. Fig. 6. Measured loss (left bars in orange) and simulated losses (stacked bars on the right) for the Itiquira ES (17°S, 54°W). Most of the differences between simulation and measured results, as presented, are a consequence of PID problems, marked initial degradation and not necessarily only due to simulation uncertainties. TABLE III SUMMARY OF RESULTS FOR MEASURED LOSSES AND TOTAL SIMULATED LOSSES FOR THE ITIQUIRA ES (17°S, 54°W). III. RESULTS AND DISCUSSIONS: OVERIRRADIANCE EVENTS The highest irradiance measurement reported in the literature so far was 1891 W/m2 [15], measured at a 1829 m altitude site in Colorado - USA. The highest irradiance peak measured in Brazil (Itiquira ES 17°S, 54°W), at 522 m altitude, and where large-scale PV power plants are currently being installed, was 1823 W/m². While this extreme overirradiance event only lasted a few seconds, other overirradiance events lasting several minutes have also been recorded, as detailed in Table IV, and which can lead to deleterious consequences to the operation and maintenance of PV power plants, as will be further discussed. Overirradiance events can only be detected if solar irradiance is measured at a high time resolution with high-quality, fast pyranometers, and the suspicion is that such events might go unnoticed in many other low-latitude sites throughout the globe. When overloaded, inverters limit the output power to nominal power. This is done by the inverter’s MPPT (Maximum Power Point Tracking) increasing the DC voltage from maximum power point voltage towards open circuit voltage. Assuming this inverter strategy and considering that most of utility-scale projects use undersized inverters, one might be lead to the conclusion that, in the occurrence of an overirradianceevent, inverters will sweep MPPT and all the strings will operate with a higher voltage and a lower string current than during normal MPPT operation. However, infield measurements in this R&D project support what has already been suggested in the literature: overirradiance events have only a small spatial footprint, because of their direct dependence on the suncloud edge-surface geometry [15]. This implies that depending on cloud geometry, inverter maximum power and Inverter Loading Ratio (ILR), during an overirradiance event part of the array could be operating with currents proportional to overirradiance magnitudes, while the other part of the array could be in the cloud shadows, resulting in an instantaneous inverter output power below inverter nominal power. In this case, the inverter will not limit the output power using the strategy described above, which will 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0106 result in high currents, which associated with high fuse operating temperatures might cause string fuses to blow. TABLE IV OVERIRRADIANCE EVENTS AND EVENT DURATION FOR IRRADIANCE PEAKS AT OR ABOVE THE EXTRATERRESTRIAL IRRADIANCE OF 1367 W/M2 REGISTERED IN BRAZIL (ITIQUIRA ES 17°S, 54°W) BETWEEN JANUARY 2015 AND MARCH 2017. Typically, technical specifications provided by PV module manufacturers include information on the “maximum series fuse rating”, and designing a PV array with a fuse rated above this maximum value will lead to voiding the PV module warranty. At most sites where PV plants in Brazil will operate, ambient temperatures and irradiation levels are high, and the string boxes where the above-mentioned fuses are installed can reach temperatures higher than 70°C [16], as has been routinely measured throughout this project. Fuse maximum currents are usually rated at 20°C operating temperatures, and at 70°C the typical derating factor is 0.7. This fact, associated with overirradiance values in excess of 1.4 to 1.6 suns for time spans in the minute-range, will result in conditions that will lead to the blowing of even slow-blow or time-delay fuses, if they are designed as per the PV module manufacturer’s maximum series fuse rating specifications. No previous reports of such a combination of effects has been found in the literature. Based on these findings, it can be foreseen that a number of such events might take place in many of the large-scale PV power plants being installed in Brazil since 2017. It is therefore recommended that PV module manufacturers reassess their maximum series fuse ratings in the light of this information, and utility-scale projects could size string boxes in a manner to keep fuses from reaching such high temperatures. In Brazil, current guidelines require that a minimum of one- year irradiance measurements in 10 minutes averaging intervals should be presented for a PV power plant to be eligible for the regulated market energy auctions. These requirements are made so that site-adaptation techniques between long term satellite- derived data and ground measured solar radiation can be performed, aiming at the reduction of the uncertainty on the energy production estimation [17]-[18]. This 10-minute averaged data, however, prevents the better assessment of ILR losses. In Fig. 7, a cloudy day’s global horizontal irradiance measured at one-second (blue and orange spiky curve) and 10- minute averages (light-orange shadow smooth curve) at the Cachoeira Dourada ES (18°S, 49°W) are presented. Averaging one-second measurements at 10-minute intervals hides valuable information on irradiance peaks, and leads to misleading conclusions on ILR losses and the true irradiation resource availability at a particular site. Fig. 7. One-second instant (blue and orange spiky curve) and 10- minute averages (light-orange shadow smooth curve – Brazilian PV power plant standard data acquisition resolution) solar irradiance measurements at the Cachoeira Dourada ES (18°S, 49°W). The high irradiance peaks tend to disappear when data is averaged at 10-minute intervals. However, the use of one-second time resolution leads to very large amounts of data and requires high-performance computing. Data averaging of one minute is suggested to be an optimum resolution without requiring further improvements in computational capacity. V. CONCLUSIONS In this work, the performance assessment of seven different PV technologies for utility-scale applications in eight warm and sunny sites in Brazil has been described and evaluated on a long-term project. The results presented in this paper give a concise overview of the main aspects (cloud-enhancement effects, blowing fuses, PID, simulation uncertainties) that will have to be considered as utility-scale PV power plants start to emerge in large amounts in warm and sunny countries. The results also present the performance assessment of the PV technologies installed at each ES. Thin-film PV technologies with a low temperature coefficient of power presented superior output performance. Mono- and multi- crystalline revealed intense PID degradation in coastal areas with high temperature and relative humidity. Extreme overirradiance events (which lasted from seconds to several minutes) were measured and shown, and it can be speculated that high time resolution irradiance measurements in many sunny sites with considerable cloud movement will reveal this aspect to be a common feature at such sites. 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0107 ACKNOWLEDGEMENTS The authors wish to acknowledge with thanks the Brazilian Electrical Energy Regulatory Agency ANEEL, as well as ENGIE Brasil Energia and the 11 other cooperating Electrical Utility companies involved in project PE-0403-0027/2011 for sponsoring this project. L. Nascimento acknowledges the Brazilian Post-Graduate council CAPES for a doctoral scholarship. REFERENCES [1] R. Rüther, “Experiences and operational results of the first grid- connected, building-integrated, thin-film photovoltaic installation in Brazil.” in 2nd World Conference on Photovoltaic Solar Energy Conversion – WCPEC2, 1998. [2] L.R. Nascimento and R. Rüther, “Fifteen years and counting: The reliable long-term performance of the first grid-connected, building-integrated, thin-film photovoltaic installation in Brazil.” in 40th IEEE Photovoltaic Specialist Conference, 2014, pp. 3372- 3377. [3] R. Rüther, L.R. Nascimento, R. Campos, A.A. Montenegro, ANEEL R&D Project Report PE-0403-0027/2011. [4] R.C. Andrade and C. Tiba, “Extreme global solar irradiance due to cloud enhancement in northeastern Brazil.” Renewable Energy, vol. 86, pp. 1433-1441, 2016. [5] M.P. Almeida, R. Zilles and E. Lorenzo, “Extreme overirradiance events in São Paulo, Brazil.” Solar Energy, vol. 110, pp. 168-173, 2014. [6] R.D. Piacentini, G.M. Salum, N. Fraidenraich, and C. Tiba, “Extreme total solar irradiance due to cloud enhancement at sea level of the NE Atlantic coast of Brazil.” Renewable Energy, vol. 36, pp. 409-412, 2011. [7] J. Luoma, J. Kleissl and K. Murray, “Optimal inverter sizing considering cloud enhacement.” Solar Energy, pp. 421-429, 2012. [8] B. Burger and R. Rüther, “Inverter sizing of grid-connectedphotovoltaic systems in the light of local solar resource distribution characteristics and temperature.” Solar Energy, pp. 32-45, 2006. [9] M. Kratzenberg, E. Deschamps, L.R. Nascimento, R. Rüther and H. Zürn, “Optimal Photovoltaic Inverter Sizing Considering Different Climate Conditions and Energy Prices.” Energy Procedia, 2014. [10] R. Rüther, G. Tamizh-Mani, J. Del Cueto, J. Adelstein, A.A. Montenegro and B. Von Roedern, “Performance test of amorphous silicon modules in different climates: higher minimum operating temperatures lead to higher performance levels.” in PV Conversion, 2003. Proc. of 3rd World Conference on, 2003. [11] R. Rüther, J. Del Cueto, G. Tamizh-Mani, A.A Montenegro, S. Rummel, A. Anderberg and B. Von Roedern, “Performance test of amorphous silicon modules in different climates - year four: progress in understanding exposure history stabilization effects.” in 33th IEEE Photovoltaic Specialist Conference, 2008. [12] S. Pingel, O. Frank, M. Winkler, S. Daryan, T. Geipel, H. Hoehne and J. Berghold, “Potential Induced Degradation of solar cells and panels.” in Photovoltaic Specialists Conference (PVSC), 2010 35th IEEE, 2010. Anais., 2010.v.p. 002817-002822. [13] N. Strevel, L. Trippel, C. Kotarba and I. Khan, “Improvements in CdTe module reliability and long-term degradation through advances in construction and device innovation.” Photovoltaics International, vol. 22, 2013. [14] M. Braga, R. A. Campos, L. R. D. Nascimento, E. M. Santos and R. Rüther, “Avaliação de desempenho de diferentes tecnologias fotovoltaicas em Irecê, BA.” in VII Congresso Brasileiro de Energia Solar, 2018. Anais. Gramado - SC, 2018.v.p. [15] C. A. Gueymard, “Cloud and albedo enhancement impacts on solar irradiance using high-frequency measurements from thermopile and photodiode radiometers. Part 1: Impacts on global horizontal irradiance.” Solar Energy. v.153, p.755-765, 2017. [16] R. Rüther, L. R. D. Nascimento and R. A. Campos, “Performance assessment issues in utility-scale photovoltaics in warm and sunny climates,” Renew. Energy Environ. Sustain, v.2, p.35, 2017. [17] J. Polo, S. Wilbert, J. A. Ruiz-Arias, R. Meyer, C. Gueymard, M. Súri, L. Martín, T. Mieslinger, P. Blanc, I. Grant, J. Boland, P. Ineichen, J. Remund, R. Escobar, A. Troccoli, M. Sengupta, K. P. Nielsen, D. Renne, N. Geuder and T. Cebecauer, “Preliminary survey on site-adaptation techniques for satellite-derived and reanalysis solar radiation datasets.” Solar Energy, v.132, p.25-37, 2016. [18] C. S. Ruschel and G. P. D. Ponte, “Metodologias de ajuste de dados solarimétricos visando a estimativa de produção de energia de longo prazo.” in VII Congresso Brasileiro de Energia Solar, 2018. Anais. Gramado - SC, 2018.v.p. 978-1-5386-8529-7/18/$31.00 ©2018 IEEE 0108
Compartilhar