In-field monitoring of eight photovoltaic plants: degradation rate over seven years of continuous operation ACTA IMEKO ISSN: 2221-870X December 2018, Volume 7, Number 4, 75 - 81 ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 75 In-field monitoring of eight photovoltaic plants: degradation rate over seven years of continuous operation Alessio Carullo1, Antonella Castellana1, Alberto Vallan1, Alessandro Ciocia2, Filippo Spertino2 1 Politecnico di Torino, Dipartimento di Elettronica e Telecomunicazioni, Corso Duca degli Abruzzi n. 24, 10129 Torino, Italy 2 Politecnico di Torino, Dipartimento Energia, Corso Duca degli Abruzzi n. 24, 10129 Torino, Italy Section: RESEARCH PAPER Keywords: Photovoltaic power systems; degradation; electric variables measurement; data acquisition; uncertainty Citation: Alessio Carullo, Antonella Castellana, Alberto Vallan, Alessandro Ciocia, Filippo Spertino, In-field monitoring of eight photovoltaic plants: degradation rate over seven years of continuous operation, Acta IMEKO, vol. 7, no. 4, article 13, December 2018, identifier: IMEKO-ACTA-07 (2018)-04-13 Editor: Alexandru Salceanu, "Gheorghe Asachi" Technical University of Iasi, Romania Received April 24, 2018; In final form September 21, 2018; Published December 2018 Copyright: © 2018 IMEKO. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited Corresponding author: Alessio Carullo, e-mail: alessio.carullo@polito.it 1. INTRODUCTION The continuous worldwide increase in installations of photovoltaic (PV) plants [1] demands a reliable estimation of their long-term performance. The main parameters of interest are the payback time (PBT) and the energy payback Time (EPBT) [2-6]. Such parameters are largely affected by the actual degradation to which the PV modules are subject, which reduces efficiency during their lifetime. As an example, for a small-size 3 kW plant whose expected energy production is 4500 kWh/year, in the conditions defined in the web tool [7], a payback time of nine years and eight months is estimated using a PV module decay of 0.70 %/year, as suggested in the IEA document [8] for mature module technologies. However, if the PV modules used exhibit a degradation rate of 2.40 % per year, the estimated payback time becomes 10 years and 4 months. With the aim of estimating the degradation rate of the commercially available PV technologies, the authors monitored eight different outdoor PV plants between 2010 and 2017. Electrical and environmental quantities were continuously monitored by means of a specifically conceived data acquisition system, which is subject to a metrological confirmation program [9]. This allows for measurement traceability to be ensured and uncertainty to be stated for each of the measured parameters. Preliminary results are reported in [10-12]. The results reported here refer specifically to the degradation rate of the monitored PV plants, estimated from October 2010 to December 2017. Furthermore, three of the plants based on thin-film technologies, which exhibited the worst degradation rate, have been characterized at module level in order to verify the average degradation of all the involved PV modules and to highlight the presence of possible “outliers.” This characterization was undertaken in March 2018. In Section 2, the test facility is described, providing information about the monitored PV plants, the experimental setup, and the estimated parameters. In Section 3, the results obtained in terms of the degradation rate at plant level are provided for each of the monitored PV plants, while Section 4 reports the results in terms of the degradation rate at module ABSTRACT The results of more than seven years (October 2010-December 2017) of continuous monitoring are presented in this paper. Eight outdoor photovoltaic (PV) plants were monitored. The monitored plants use different technologies: mono-crystalline silicon (m-Si), poli- crystalline silicon (p-Si), string ribbon silicon, copper indium gallium selenide (CIGS), thin film, and cadmium telluride (CdTe) thin film. The thin-film and m-Si modules are used both in fixed installations and on x-y tracking systems. The results are expressed in terms of the degradation rate of the efficiency of each PV plant, which is estimated using the measurements provided by a multi-channel data acquisition system that senses both electrical and environmental quantities. A comparison with the electrical characterization of each plant obtained by means of the transient charge of a capacitive load is also made. In addition, three of the monitored plants are characterized at module level, and the estimated degradation rates are compared to the values obtained with the monitoring system. The main outcome of this work can be summarized as the higher degradation rate of thin-film based PV modules with respect to silicon- based PV modules. mailto:alessio.carullo@polito.it ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 76 level for three of the monitored plants. Finally, in Section 5, the main outcomes related to the obtained results are summarized. 2. TEST FACILITY 2.1. Monitored PV plants The eight PV plants that were monitored are located in Piemonte (Italy) at a latitude of about 45 °N, in the Cwc Köppen- Geiger climate zone [13]. Table 1 summarizes the main characteristics of the monitored PV plants, which include silicon- based modules (plants A, B, C, and Ats) and thin-film based modules (D, E, Dts, and Ets). The subscript “ts” denotes the plants whose modules are installed on a x-y tracking system, while the others are oriented towards South (azimuth angle of about 0°) and mounted in a fixed position with a tilt angle of 35°. In Table 1, APV, Pnom, and nom refer to the nameplate area (m2), power (kW), and efficiency (%), respectively. Immediately after the installation of the PV plants, their I-V electrical characteristics were measured in natural sunlight by means of the acquisition of the transient charge of a capacitive load [14], and the actual power Pact and efficiency act in the standard test conditions (STCs) were then estimated and are also reported in Table 1. The relative expanded uncertainty (coverage factor k = 2) for both parameters is about 3.5 %. 2.2. Monitoring system The monitoring system is able to measure direct voltage Vdc and direct current Idc (to upstream the inverter) and temperature tm of a PV module for each of the eight PV plants. Solar irradiance Gm was also measured on the plane of the PV modules by means of two secondary-standard pyranometers: The first one was mounted with the same orientation as the fixed plants, while the second one was mounted on the x-y tracking system of the plant Ats. A measurement of each quantity was carried out every 10 s and stored in a daily file. The architecture of the monitoring system is described in [10], and its metrological confirmation process is described in [9]. The calibration is performed with a periodicity of one year and includes the initial verification of each measuring chain, the adjustment that allows offset and gain drifts to be compensated, and the final verification. The maximum admitted errors used during the verifications are: • (0.5 % reading + 0.2 V) for voltage Vdc in the range of 100 V to 450 V; • (0.4 % reading + 5 mA) for current Idc in the range of 0.5 A to 7 A; • (2.0 % reading + 5 W/m2) for irradiance Gm in the range of 500 W/m2 to 1200 W/m2; and • (0.6 % reading + 0.55 °C) for temperature tm in the range of 10 °C to 80 °C. 2.3. Data processing The efficiency  of each plant at STC is obtained by: STCPV STCmax, GA P   (1) where GSTC = 1000 W/m2 is the reference irradiance value, APV is the area of the PV modules as reported in Table 1, and Pmax,STC is the maximum measured power reported at STC through the simplified model:      tGactSC,actS,tdc tGactSC,dcSTCmax, CCIRCV CCIIP     (2) where ISC,act and RS,act are the short-circuit current and the series resistance of each plant obtained during the preliminary I-V characterization;  (A/°C) and  (V/°C) are the absolute current and voltage temperature coefficients, respectively; and the correction coefficients CG and Ct are obtained by: mSTCt STC m G ;1 ttC G G C  (3) Starting from the available data, a clear day was selected in each month in the period from January 2014 to December 2017, and Equations (1)-(3) were calculated for each plant using the measured parameters that correspond to the values of irradiance Gm that are greater than 800 W/m2. During the initial monitoring period (from September 2010 to December 2013), the same procedure was implemented, but with different sampling, since a clear day was selected for each couple of months. The standard uncertainty u(Pmax,STC) of each PV plant was estimated according to the procedure described in [11], which takes into account both the repeatability contributions and the uncertainty of each measuring chain. The last contribution corresponds to the maximum admitted errors that are periodically verified. One should note that the uncertainty of the efficiency  obtained by means of Equation (1) mainly depends on the uncertainty of the parameter Pmax,STC, since GSTC is a conventional reference value, while APV is known with a negligible uncertainty. 3. PLANT DEGRADATION RATE An example of the obtained results that refer to the selected day, 17 January, 2016, is reported in Table 2, where uA(P) represents the standard deviation of the set of values Pmax,STC,i, corresponding to the Gmi values that are greater than 800 W/m2. Meanwhile, uBx(P) is the uncertainty contribution related to the measured quantity x. Among the different contributions, the one that mainly accounts for the uncertainty of the parameter Pmax,STC and, in turn, of the estimated efficiency  is that related to the correction coefficient CG, which depends on the measured quantity Gm. Since this quantity is sensed through a secondary standard pyranometer, this contribution cannot be significantly reduced for monitored outdoor PV plants. In the same table, the estimated efficiency  at STC is reported for each plant with the corresponding standard uncertainty u(). Similar uncertainty values for the parameters Pmax,STC and were obtained during the whole monitored period. 3.1. m-Si based plants The estimated efficiency  of plant A (fixed installation) based on m-Si PV modules during the monitored period is reported in Figure 1, while Figure 2 refers to the plant Ats (x-y tracking system). The red circles represent the estimated  values, the black squares are the values obtained after the adjustment of the Table 1. Main characteristics of the monitored PV plants. Plant PV technology APV (m2) Pnom (kW) Pact (kW) nom (%) act (%) A m-Si 11.2 2.03 1.93 18.1 17.2 B p-Si 13.8 1.85 1.80 13.4 13.1 C String ribbon Si 17.9 2.28 2.16 12.7 12.1 D CIGS 17.5 1.70 1.68 10.3 9.6 E CdTe 17.3 1.74 1.61 10.1 9.3 Ats m-Si 11.2 2.03 1.93 18.1 17.2 Dts CIGS 17.5 1.70 1.68 10.3 9.6 Ets CdTe 17.3 1.74 1.61 10.1 9.3 ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 77 monitoring system, and the black stars represent the value obtained after the PV modules were washed. The former intervention does not show significant effects on the estimated efficiency, thus highlighting that the drift of the measuring chain characteristics during the calibration period is acceptable. In Figure 1 and Figure 2, the efficiency values obtained with the capacitive load technique in months 1 and 78 are also reported (black diamonds), and they demonstrate behavior that is in agreement with the observed season variability. The black triangles identify invalid results, due to a fault in the monitoring system in the month-interval (80-85) and a fault in the plant Ats during months 46, 47, and 48. The continuous straight line is obtained by fitting only the valid  values through a least square algorithm. One should note that for plant Ats, the straight line was fitted from month 40, since previous efficiency estimations were affected by a fault in the board that connects the positive pole of the PV modules to ground. Due to this improper grounding of the PV modules, the two poles of the plant were floating, thus exposing the modules to potential induced degradation (PID) [15], one of the causes of efficiency loss. After the grounding board was replaced (month 40), plant Ats exhibited efficiency values very similar to plant A. The fitted straight line can be represented as:     tStt   0 (4) where t is the time (months), while S is the slope of the fitted line (%/months). The parameter S and its standard uncertainty u(S) were obtained according to the procedure described in [11], taking into account the standard uncertainty u(i) corresponding to each monthly estimation. Starting from the obtained values, the yearly percentage degradation rate DR (%/year) was estimated following:  S DR   12 100 (5) For the m-Si based plant A, a degradation rate DRA equal to -0.03 %/year, expanded uncertainty (95 % confidence level) U(DRA) = 0.35 %/year, was obtained. These values are summarized in Table 3, where the degradation rates and the corresponding uncertainties of the PV plants under investigation are reported for different monitoring periods [11-12]. Taking the uncertainty into account, the estimated degradation rates of plant A are compatible during the different monitored periods, thus showing excellent behavior of the plant based on m-Si PV modules in fixed installations. For plant Ats, the obtained value is DRAts = -0.78 %/year, and the expanded uncertainty is U(DRAts) = 0.40 %/year. This updated value is significantly different from the previous one (+0.20 %/year at the month 72, as indicated in Table 3), thus highlighting an important degradation of the m-Si modules installed on the x-y tracking system. The obtained results are in good agreement with the efficiency values estimated for month 78 with an independent measuring system (black diamonds in Figure 1 and Figure 2). 3.2. CIGS-based plants Table 2. An example of the uncertainty budget, selected day: January 17, 2016. Plant uA(P) (W) uBIdc(P) (W) uBVdc(P) (W) uBCG(P) (W) uBtm(P) (W) u(Pmax,STC) (W) Pmax,STC (W)  (%) u() (%) A 8.2 5.0 5.7 26 3.2 28 1940 17.33 0.25 B 8.4 4.4 5.0 40 4.1 42 1725 12.50 0.21 C 11.7 5.4 6.2 27 4.9 31 2110 11.77 0.16 D 10.9 3.4 4.1 14 3.2 19 1375 7.95 0.16 E 4.9 3.0 4.3 8.5 3.1 12 1415 8.07 0.16 Ats 8.0 5.9 5.7 24 3.2 26 1990 17.78 0.25 Dts 3.7 4.0 3.8 10.8 3.5 13 1350 7.70 0.16 Ets 3.9 3.9 3.9 11.0 2.6 13 1360 7.84 0.16 Figure 1. Estimated efficiency of plant A based on m-Si PV modules in a fixed installation (red circles); black squares: values after adjustment of the monitoring system; black stars: values after module washing; black diamonds: values obtained by means of the capacitive load technique; black triangles: invalid data. Table 3. Summary of the degradation rates of the plants under investigation estimated during different monitored periods ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 78 The results in terms of efficiency  of plant D (fixed installation) based on CIGS thin-film PV modules are reported in Figure 3. The meaning of the symbols is the same as in Figure 1, and the invalid results are the same as those of plant A. Furthermore, there is good agreement with respect to the values obtained using the capacitive load technique (black diamonds for months 1 and 78), and the results obtained after the adjustment of the monitoring system (black squares) do not significantly differ from the others. The estimated degradation rate DRD is equal to -1.83 %/year (expanded uncertainty U(DRD) = 0.20 %/year), which is compatible with the values obtained in the previous monitored interval (see Table 3). Unfortunately, the results obtained after month 65 for plant Dts that was installed on a x-y tracking system are unreliable. This is due to a misalignment between the tracking systems of plants Ats and Dts in the month-interval (65-72) [12] and to a fault in the electrical measuring chain of this plant in the month-interval (73 - 87) that the plant owner has not fixed. For this reason, the estimated degradation rate for plant Dts is the same as the value stated in [12], i.e., DRDts = -3.34 %/year, U(DRDts) = 0.26 %/year. The obtained uncertainty is low enough for us to confidently state that CIGS modules are subject to an important degradation and that the modules installed on the x-y tracking system exhibit a larger degradation than the same modules in a fixed position. 3.3. CdTe-based plants Figure 4 shows the efficiency  of the plants based on CdTe thin-film PV modules. The left chart refers to plant E (fixed position) and the right chart to plant Ets (x-y tracking system). The same symbols used in the previous figures are used here, and the reasons for the invalid results (black triangles) are the same as for the other plants. The efficiency values obtained by means of the capacitive load technique for months 1 and 78 (black diamonds) are in good agreement with the observed season variability. The estimated degradation rates are DRE = -2.36 %/year, U(DRE) = 0.30 %/year, and DREts = -2.41 %/year, U(DREts) = 0.20 %/year. As summarized in Table 3, these values are in agreement with the degradation rates estimated during the first 72 months of operation, while they seem better than the values estimated after 34 months of operation. However, the large uncertainty of the first estimated values does not allow us to distinguish the degradation rates estimated during the different monitored periods, since they are compatible. Furthermore, in this case, we can conclude that the degradation of the CdTe thin-film modules is higher than the degradation exhibited by the m-Si modules. However, due to the uncertainty, it is not possible to clearly distinguish between the behavior of the modules in a fixed position with respect to the modules installed on the x-y tracking system. 3.4. p-Si and string ribbon Si based plants The results of the last two monitored plants are shown in Figure 5: the left chart refers to the plant based on p-Si PV modules, while the right chart refers to the plant that uses string- ribbon Si modules. The same symbols used in the previous Figure 2. Estimated efficiency of plant Ats based on m-Si PV modules on an x-y tracking system (red circles); black squares: values after adjustment of the monitoring system; black stars: values after module washing; black diamonds: values obtained by means of the capacitive load technique; black triangles: invalid data. Figure 3. Estimated efficiency of plant D based on CIGS thin-film PV modules in a fixed installation (red circles); black squares: values after adjustment of the monitoring system; black stars: values after module washing; black diamonds: values obtained by means of the capacitive load technique; black triangles: invalid data. ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 79 figures are used here, and the reasons for the invalid results (black triangles) are the same as for the other plants in fixed positions. One should note that the adjustment of the monitoring system does not significantly change the efficiency estimations. The fitted straight lines of the two plants are very similar, which is confirmed by the numerical results: for the p-Si based plant, the estimated degradation rate DRB is -0.87 %/year, U(DRB) = 0.35 %/year, while for the string-ribbon Si-based plant, the obtained value is DRC = -0.68 %/year, U(DRC) = 0.25 %/year. Both degradation rates are very similar to the values previously estimated (see Table 3), thus confirming a medium degradation for these plants. 4. MODULE DEGRADATION RATE With the aim of confirming the estimated degradation rates for the plants that exhibited the worst behavior, an electrical characterization at module level was performed for plants D and Dts, which are based on CIGS thin-film modules, and for plant Ets, which is based on CdTe thin-film modules. The I-V characteristic of each of the 24 modules that made up each plant was obtained by means of an experimental setup that is based on a programmable electronic load connected to the PV modules under test. Each I-V characteristic is obtained at the measurement conditions (air temperature in the range of 15.3 °C to 18.0 °C and irradiance in the range of 800 W/m2 to 1030 W/m2) and then reported to STC in order to obtain the parameter Pmax,STC,i of each PV module. The experimental setup and the correction algorithm are the same as those described in [16]. Figure 4. Estimated efficiency of the plants based on CdTe PV modules: E in fixed installation, Ets on an x-y tracking system (red circles); black squares: values after the adjustment of the monitoring system; black stars: values after module washing; black diamonds: values obtained by means of the capacitive load technique; black triangles: invalid data. Figure 5. Estimated efficiency of the plants based on p-Si modules (left chart) and string ribbon Si modules (right chart) (red circles); black squares: values after adjustment of the monitoring system; black stars: values after module washing; black diamonds: values obtained by means of the capacitive load technique; black triangles: not valid data. ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 80 This operation was performed on March 23, 2018, approximatively 90 months after the installation of the PV modules. Once the parameter Pmax,STC,i was obtained, the actual degradation rate of each PV module was obtained as: )%/year( 90 12 100 inom, inom,iSTC,max, i    P PP DR , (6) where Pnom,i is the nameplate power of the investigated PV modules, which is 70 W for CIGS thin-film modules and 72.5 W for CdTe thin-film modules. The obtained results are summarized in Figure 6 in the form of a histogram of occurrences of the parameters DRi, which were obtained according to (6). The top histogram, which refers to the PV modules of the plant D (CIGS thin-film in fixed position), shows the lowest standard deviation (about 0.7 %/year) and an average value of -2.43 %/year. This value is very similar to the degradation rate estimated at plant level, which is -2.35 %/year, thus confirming the obtained outcomes. The middle histogram refers to the PV modules of the plant Dts (CIGS thin-film on x-y tracking system) and exhibits a standard deviation of about 0.8 %/year and an average value of -2.90 %/year. In this case, the average value is lower than the degradation rate estimated at plant level (- 3.34 %/year), but the two values are not comparable, since the monitoring of the plant Dts only refers to the first 65 months of operation, for the reasons described in subsection 3.2. The bottom histogram, which refers to the PV modules of the plant Ets (CdTe thin-film on x-y tracking system), shows the highest standard deviation (about 0.9 %/year) and an average value of -2.72 %/year, which is comparable with the degradation rate at plant level (-2.41 %/year). The high standard deviation is mainly due to a PV module that shows a degradation rate of about -5.0 %/year. However, this is a biased value, since this PV module, which is positioned at the lower-left corner of the plant, presents a visible break, as shown in Figure 7. The average value of the degradation rate recalculated after the broken module has been excluded does not change significantly. 5. CONCLUSION The results of the monitoring of eight outdoor PV plants based on different technologies along a period longer than seven years have been presented in this paper. The parameters that have been taken into account in order to assess the performance of the PV plants under investigation are the maximum power and the PV efficiency at STC. The latter parameter allowed the degradation rate of each plant to be obtained, which provided interesting information related to the behavior of the different types of plants. The plants that use silicon-based PV modules are subject to a degradation rate that is significantly lower than the plants based on thin-film PV modules. Taking the estimated uncertainty into account, the PV plants that use m-Si, p-Si and string-ribbon Si PV modules exhibited a degradation rate that is in agreement with the indication given in [8] for mature module technologies. On the contrary, the plants based on thin-film PV modules exhibited degradation rates that exceed the typical value for mature technologies. The estimated uncertainty does not allow significant conclusions to be made about the different behavior of the same PV modules installed in fixed positions and on tracking systems. An estimation of the degradation rate at module level was also performed for three of the eight monitored plants, obtaining results that are in good agreement with the degradation rate at plant level. REFERENCES [1] International Energy Agency, Photovoltaic Power Systems Programme, Annual Report, 2016. [2] L. Francke, M. S. Armand, et al., “GHG emissions and energy payback time of AC electricity generated by the SunPower® Oasis® photovoltaic power plant”, Proc. of the 42nd IEEE Figure 6. Degradation rate (%/year) of the 24 PV modules of plants D, Dts, and Ets expressed in the form of a histogram of occurrences. Figure 7. Detail of the broken module in plant Ets (CdTe thin-film on an x-y tracking system). ACTA IMEKO | www.imeko.org December 2018 | Volume 7 | Number 4 | 81 Photovoltaic Specialist Conference (PVSC), 2015, New Orleans, L.A., U.S.A. [3] M. Held, R. Ilg, Update of environmental indicators and energy payback time of CdTe PV systems in Europe, Progress in Photovoltaics 19(5) (2011) pp. 614-626. [4] A. F. Sherwania, J. A. Usmanib, et al., Life cycle assessment of solar PV based electricity generation systems: A review, Renewable and Sustainable Energy Reviews 14(1) (2010) pp. 540- 544. [5] L. Lu, H. X. Yang, Environmental payback time analysis of a roof- mounted building-integrated photovoltaic (BIPV) system in Hong Kong, Applied Energy 87(12) (2010) pp. 3625-3631. [6] A. Nishimuraa, Y. Hayashia, et al., Life cycle assessment and evaluation of energy payback time on high-concentration photovoltaic power generation system, Applied Energy 87(9) (2010) pp. 2797-2807. [7] SunEarthToll.com, Photovoltaic payback [Online] Available: https://www.sunearthtools.com/solar/payback- photovoltaic.php#top [8] International Energy Agency, Methodology Guidelines on Life Cycle Assessment of Photovoltaic Electricity, Report IEA-PVPS T12-03:2011. [9] A. Carullo, S. Corbellini, et al., In situ calibration of heterogeneous acquisition systems: The monitoring system of a photovoltaic plant, IEEE Trans. on Instrum. and Meas. 59(5) (2010) pp. 1098- 1103. [10] A. Carullo, A. Vallan, Outdoor experimental laboratory for long- term estimation of photovoltaic-plant performance, IEEE Trans. on Instrum. and Meas. 61(5) (2012) pp. 1307-1314. [11] A. Carullo, A. Vallan, et al., Uncertainty analysis of degradation parameters estimated in long-term monitoring of photovoltaic plants, Measurement 55 (2014) pp. 641-649. [12] A. Carullo, A. Castellana, et al., “Degradation rate of eight photovoltaic plants: results during six years of continuous monitoring”, Proc. of the 22nd IMEKO TC4 International Symposium, Sept. 14-15, 2017, Iasi, Romania. [13] F. Rubel, M. Kottek, Observed and projected climate shifts 1901- 2100 depicted by world maps of the Kãüppen-Geiger climate classification, Meteorologische Zeitschrift 19(2) (2010) pp. 135- 141. [14] F. Spertino, J. Ahmad, et al., Capacitor charging method for I–V curve tracer and MPPT in photovoltaic systems, Solar Energy 119 (2015) pp. 461-473. [15] M. Martin, R. Krause, et al., “Investigation of potential induced degradation for various module manufacturers and technologies”, Proc. of the 27th European Photovoltaic Solar Energy Conference and Exhibition, 2012, Frankfurt, Germany, pp. 3394-3398. [16] A. Carullo, A. Castellana, et al., “Uncertainty issues in the experimental assessment of degradation rate of power ratings in photovoltaic modules”, Measurement 111 (2017), pp. 432-440. https://www.sunearthtools.com/solar/payback-photovoltaic.php#top https://www.sunearthtools.com/solar/payback-photovoltaic.php#top