MEAN DEGRADATION RATES IN PV SYSTEMS FOR

32nd European Photovoltaic Solar Energy Conference and Exhibition

MEAN DEGRADATION RATES IN PV SYSTEMS FOR VARIOUS KINDS OF PV MODULE FAILURES Marc Köntges1, Sascha Altmann1, Tobias Heimberg1, Ulrike Jahn2, Karl A. Berger3 Institut für Solarenergieforschung Hameln (ISFH), Am Ohrberg 1, D-31860 Emmerthal, Germany 2 TÜV Rheinland Energy GmbH, D-51105 Köln, Germany 3 Austrian Institute of Technology GmbH, Energy Department, 1210 Vienna, Austria

1

ABSTRACT: We analyse the impact of various photovoltaic (PV) module failure modes in PV systems in the field to rank the impact of the failure types on the power generation. Therefore we designed a PV system failure survey based on a spread sheet programme. The survey is distributed to international PV experts who fill in the data. The failure data is collected since Oct. 2015 and the data collection is still in progress. The module failure types “cell cracks”, “potential induced degradation by shunting (PIDs)”, “defect bypass diodes” and “discolouring of pottant“ have the highest fraction in the database. The PIDs effect has a mean annual degradation rate of about 16%/a and affects about 3/5 of a system in moderate climate. This degradation rate is calculated for the PV modules PV systems which show degradation. Cell cracks show a mean degradation rate of around 5%/a for systems in the moderate climate zone and a higher degradation rate of about 8%/a in cold and snow climate. Cell cracks also often affect only some modules in a string showing a relevant power loss. Defect bypass diodes may show in some cases a very high impact on the module and system power degradation. They occur in the first 10 years of system operation. The wear out failure discolouring of pottant occurs during the whole system life but accumulates for very old systems. Its degradation rate is below 1%/a for all climatic zones. Keywords: degradation rate, statistic, PV module failure

1

INTRODUCTION 2.1. Target group of the survey

Plenty of data on photovoltaic (PV) system and PV module reliability has been collected in the past [1][2][3][4][5][6][7][8][9][10]. Some studies focus on the PV module failure in great detail [10], other focus on the power losses of PV systems, but loose the origin of the power loss [11][12]. Jordan showed mean degradation rates for x-Si in the 0.8–0.9%/year range, ~1% for HIT and microcrystalline silicon and for the sum of thin film technologies 1.38%/year. There are strong variations by thin film technologies indicated. For example there are some studies on Cu(In,Ga)Se with low degradation rates like x-Si technology. To collect and evaluate data on a specific failure one has to define the most important failures in advance. As basis for the failure definitions in the survey we used the failure description given in the TASK13 report “Review of Failures of Photovoltaic Modules” [13]. Besides the PV module failures also methods like visual inspection, IV curve analysis, electroluminescence, thermography, UV fluorescence and other techniques are described to detect and identify the failures. A clear definition is a prerequisite to make the failure classification more reliable.

2

The survey is designed to be distributed to international PV experts who fill in the data. However, the primary origin of the data may be from other sources such as system owners, installers, other experts, manufacturers, scientific publications and internet surveys. Each of the provided datasets is checked by a second expert for plausibility at the ISFH. The survey input parameters are reduced to a minimum so that it is still possible to evaluate the data for power loss and degradation rates for a specific failure. We developed the survey to ensure a maximum of anonymity of the PV system. In this work we only evaluate PV module data. Tab. 1 shows publications used in survey data as data source. Tab. 1: List of survey data extracted of publications. Some authors submit additional information (marked with *). Power loss dominating failure Ref. Discolouring of pottant, Delamination, [1] Disconnected cell or string interconnect ribbon Delamination [16] Defect backsheet, Discolouring of pottant* [17] Cell cracks* [18] Dust soiling [19]

EXPERIMENTAL

We designed a survey format to collect failure data of PV systems for various climate zones with the focus on both, the origin of failure and the power loss. A full list of failure types used in the survey is given in Tab.1 in the Appendix. The long-term goal of the survey is to evaluate the possible different impacts of the failures for various climate zones. The survey is implemented in a Microsoft Excel table worksheet. The file and the corresponding help file can be accessed via internet at the homepage of the Photovoltaic Power Systems Programme of the International Energy Agency [14],[15]. The format of the survey is shown in Fig. 1.

2.2 Design of the survey For the database the climatic zone, system data, the component types, and start and analysis date are collected. However, the categorization of each item is as rough as possible in order to be as anonymous as possible.

1435


Furthermore, the failure x effects only a part zi,x of the system. The partitions yi and zi,x are given in percentage of the installed power. These values are the basis of the following degradation rate calculations.

Fig. 1: Input page of the PV system failure survey. The section “Failure specification for” is available five times to specify system parts with different failures. Underlined blue text leads to linked help files. E.g. the climate zone is linked to the Wikipedia classification of the KöppenGeiger climate system. The characterization of the PV failures allows specifying only two failures for one system part. This is for the reason that if there are more than two failures in one part of the system it is very difficult to specify the power loss for each type of failure. Therefore, it is only allowed to specify two failures which are most important for the power loss of the system. If there are various failures for different parts of the system it is possible to specify the various failures for at maximum five system parts. Sometimes it is not possible to find out all reasons for a power loss. In this case one should only state the power loss of the well analysed failures in the system.

Fig. 2: Definition of various parts of a PV system. The nominal system power Pi is given in kWp. All other parts are given in % of Pi. 3.1 Degradation rate The power loss ΔPi,x of a specific module failure x is documented in percent of the affected nominal module power sum. The equation of the degradation rate di,x of a specific module failure type x of dataset i is given by:

In the appendix the selectable failure types in the survey are visualized. The light induced degradation is not added to the choice list of possible failures. This early power degradation in the modules life must be incorporated in the nominal power of the name plate. Even if the manufacturer does not incorporate this degradation in the nominal power or it is higher than assumed it is very difficult to classify this failure. There is no easy way to detect this failure type. Therefore, we do not account for light induced degradations.

𝑑𝑖,𝑥 =

.

(1)

The parameter 𝜏b,i is the date of failure documentation of dataset i and 𝜏a,i is the date of system start of dataset i. The equation for the degradation rate of the whole system is given by:

𝛿𝑖,𝑥 = 𝑑𝑖,𝑥

In addition to the climate zones it is possible to specify special stress loads at the location of the system. Besides the standard item here are two possible items to be chosen: “Island, coastal region (10 km)” and “Agricultural environment”. 3

𝛥𝛥𝑖,𝑥

𝜏𝑏,𝑖 −𝜏𝑎,𝑖

𝑧𝑖,𝑥 𝑦𝑖

,

(2)

where 𝑧𝑖 is the percentage of the system being affected and 𝑦𝑖 is the investigated system part in percent from the total nominal system power Pi. For the further evaluations it is assumed that the investigated system part is chosen large enough that the investigation result is representative for the whole system. In this case 𝛿𝑖,𝑥 shows how much the total system is affected by the failure type x.

EVALUATION OF THE SURVEY

Due to the fact that it is very difficult to obtain power loss data for specific failure types, the power loss choices in the survey are roughly subdivided into the following items and intervals “Unknown”, „No detectable loss“, ]0%-3%], ]3%-10%], ]10%-20%], ]20%-30%] and so on until ]90%-100%]. The intervals are defined in the bracket notation of the International standard ISO 31-11. Because of the rough interval definition we do not differentiate between measured initial power of the PV system or modules and nominal power of the PV system or modules. Fig. 2 illustrates the definition of various partitions of a PV system being important for the definition of degradation rates of power losses. For each survey dataset i the total installed power Pi is collected in the database. For the fault analysis of a failure x very often only a part yi of the total system is investigated.

We define two mean degradation rates for the PV modules: 1. Unweighted mean degradation rate 𝑑𝑥 of PV module failure x of all datasets 𝑛𝑥 with the module failure x: 𝑑𝑥 =

∑ 𝑑i,𝑥 𝑛𝑥

.

(3)

2. Degradation rate 𝑑𝑥𝑥 of failure x weighted by the part 𝑧𝑖,𝑥 of the nominal power 𝑃𝑖 of the PV system affected by the failure x: ∑𝑛 𝑑 ∙𝑃 ∙𝑧 𝑑̅𝑥𝑥 = 𝑖=1𝑛 𝑖,𝑥 𝑖 𝑖,𝑥 . (4) ∑𝑖=1 𝑃𝑖 ∙𝑧𝑖,𝑥

The first degradation rate 𝑑x is unbiased of the size of the systems in the dataset. In this case small systems

1436


have the same impact on the mean module degradation rate for a failure type than large systems. The second degradation rate 𝑑xw accounts for the system size of the failure affected part. Furthermore, we define two mean degradation rates for the PV system affected by a failure x: 1. Unweighted mean degradation rate 𝛿𝑥̅ of the investigated PV system part: 𝛿𝑥 =

∑ 𝛿i,𝑥

(5) ̅ 2. Degradation rate 𝛿𝑥𝑥 of the investigated PV system part weighted by the investigated part 𝑦𝑖,𝑥 of the nominal power 𝑃𝑖,𝑥 of the PV system: 𝑛 ̅ = ∑𝑖=1𝑛 𝛿𝑖,𝑥 ∙𝑃𝑖 ∙𝑦𝑖 . 𝛿𝑥𝑥 (6) 𝑛𝑥

.

∑𝑖=1 𝑃𝑖 ∙𝑦𝑖

The first system degradation rate 𝛿x is again unbiased of the size of the systems in the dataset. The second system degradation rate �𝛿xw accounts for the system size of the failure affected part. These degradation rates allows to assess how a failure x affects the whole system power. Our survey data does not allow checking if the degradation rates of the found failures are constant over time. However the degradation rates calculated with Eq. 1-6 allow comparing and averaging slowly developing power loss effects. 3.2 Mean power loss of sudden events For sudden power losses caused by a storm, a hail storm or a lightning stroke the calculation of a degradation rate makes no sense. Therefore we evaluate how much percent of the investigated system power is affected by any power loss after the sudden event: 𝑝𝑥𝑥 =

∑𝑛 𝑖=1 𝑃𝑖 ∙𝑧𝑖,𝑥 ∑𝑛 𝑖=1 𝑃𝑖 ∙𝑦𝑖

,

(7)

and how much is the mean power loss relative to the investigated system power: 𝑝𝑥𝑥 =

∑𝑛 𝑖=1 𝛥𝛥𝑖,𝑥 ∙𝑃𝑖 ∙𝑧𝑖,𝑥 ∑𝑛 𝑖=1 𝑃𝑖 ∙𝑦𝑖

.

(8)

Fig. 4: Composition of the input data sorted by climate zones (top), by contributor profile (middle) and by PV module technology (bottom).

4 Fig. 3: Survey data distribution over the countries colour coded in percentage of the total number of contributions. Contributions from European countries are summarized in one bubble.

COMPOSITION OF SURVEY DATA

Fig. 3 shows the survey data distribution over the countries. In total 144 failure-survey-data sets from 18 countries have been evaluated. Fig. 4 shows the composition of the survey data. The moderate climate zone dominates the database with 45 % of data sets and

1437


unfortunately we have only 10% data from hot and humid climate. Most of the data is generated by experts from plant investigations or taken from scientific reports (72%). The distribution into mono- with 28%, multi- with 62% and thin film technology with 8% approximately equals the capacity share of these technologies in years 2011-2015 with ~24% for mono, ~66% for multi and ~10% for thin film [20].

5

detected failure. The lower graph shows the occurrence of detected failure which causes a measurable power loss. For the current status of collection of failure reports only some mean degradation rates can be analysed. Results based on less than four reports are not discussed. Therefore in all figures presenting statistical data the basic population of PV systems is shown as number above the mean value. The weighted degradation rates ̅ are currently too much influenced by large 𝑑̅xw and 𝛿xw single PV systems. Therefore, we do not show the results of the weighted degradations here.

RESULTS

Before we evaluate the degradation rates we show the occurrence of failures over the years of operation in Fig. 5. The upper diagram shows the occurrence of all reported failures and the lower graph shows only failures causing a power loss. Each diagram is split into the occurrence of degrading failures and sudden occurring failures. The occurrence of both failure types accumulates at the first 7 years. When we focus on special types of failures we see that cell crack failures are mostly reported in the very early stage of PV system operation from year 1 to year 2. Systems with PIDs failure are mainly reported during year 3 and year 4. Disconnected cells or strings in the module are reported after year 4 spread over the whole operation time. Discolouring of pottant is spread over the years but power relevant discolouring starts after year 3 with a high accumulation after 18 years of system operation. Defect bypass diodes are spread over the first 10 years of operation. The total occurrence of the other failures is too seldom for further discussion.

Fig. 6: Mean unweighted � 𝐱 of PV degradation rates 𝒅 module failures sorted by climatic zones. The numbers show the quantity of data per failure in the database. The whiskers indicate the upper and lower quartile of the degradation rates. Failures with no data or 0%/a degradation rate are not shown. Right: legend of figure. Fig. 6 shows the mean unweighted degradation rates for the affected part of the PV systems. The highest impact on the performance of PV modules have defective bypass diodes in the hot and dry climate with 11%/a, and in moderate climate with 25%/a. The degradation rate 𝑑x caused by cell cracks in the cold and snow climate is about 3%/a higher than in the moderate climate and 6%/a higher than in hot and dry climate. The PIDs effect shows a mean degradation rate of about 15% per year. In the moderate climate it is the most often found failure together with a high degradation rate. Unfortunately there are not enough PIDs events documented from other climate zones. The discolouring of pottant failure is found in the hot and humid, hot and dry and in the moderate climate as well. In the three climate zones this degradation mechanism is in the mean below 1%/a. Therefore, this effect is most often not the cause for warranty claims.

Fig. 5: Occurrence distribution of failures over the years of PV system operation. The failure occurrence is split into degrading failure and sudden occurring failures. The upper graph shows the total failures occurrence of all

How much a failure dominates the total system power can be seen by the unweighted degradation rate of the investigated system part shown in Fig. 7. For the PIDs effect in the moderate climate the degradation of the

1438


investigated system part is reduced by 3/5 compared to the degradation of the effected system part. Therefore, PIDs affects in average about 3/5 of a system in the moderate climate zone. For cell cracks the same evaluation shows that cell cracks affect about 3/5 of the system in the moderate zone if they cause a power loss.

loss on less than 10% of the modules in the system. The failure “snow load” affects about 20% of the modules in the system. The power loss of a system caused by a sudden event is calculated by Eq. (8) and is shown in Fig. 9. Besides soiling the snow load event has the highest impact on the system power loss. The other sudden events only affect less than 1% of the system power. Except for soiling the calculations in Fig. 8 and Fig. 9 are mostly based on single to three entries in the database and are therewith only show cases.

Fig. 7: Mean unweighted �𝒙 of the degradation rates 𝜹 investigated PV system part of PV module failure sorted by climatic zones. The numbers show the quantity of data per failure in the data-base. The whiskers indicate the upper and lower quartile of the degradation rates. Failures with no data or 0%/a degradation rate are not shown. Right: legend of figure.

Fig. 9: Mean system power 𝒑𝒙𝒙 lost from the investigated system part after a sudden event. The numbers show the quantity of data per failure in the database. The whiskers indicate the upper and lower quartile of the lost system power.

Fig. 10: Box-Whisker-Plot of degradation rate of the investigated part of the system for the PIDs effect sorted by the special stress coastal region and other. The numbers show the quantity of data per failure in the database. The mean value is indicated by a thick line. The whiskers indicate the upper and lower quartile of the degradation rates.

Fig. 8: Mean system part 𝒑𝒙𝒙 affected after a sudden event. The numbers show the quantity of data per failure in the database. The whiskers indicate the upper and lower quartile of the affected system part. Fig. 8 shows how much of a system is affected by a sudden event calculated by Eq. (7). As expected, soiling affects almost the whole systems in nearly all cases. There is one event of “Animal -> bite/corrosion/dirt” where animals also soil all of the modules. The events “lighting stroke”, “storm” and “hail” only cause a power

72% of all reported failures with the special stress loads “Island, coastal region (10 km)” show a PIDs failure. From the rest data only 4.6% shows PIDs. This indicates much higher PIDs risk for island and coastal regions as also found by Berghold [21]. Therefore, one

1439


would expect a much higher PIDs degradation rate for “coastal regions/island” systems. But this is not the case. The total degradation rate of the coastal regions/island systems is shown in Fig. 10 in comparison to all other systems. The mean PIDs degradation rate is much higher for other regions than for “coastal regions/island” systems. However this is caused by some systems in other regions with an extreme high degradation rate.

5

almost all cells are affected [21]. Therewith the different affected area partitions influence the degradation rate. Berghold suggests that the “Frame PID” with the low degradation rate is found in hot and dry locations. But in our database only two island/coast region system are from hot and dry locations. Therefore, this “Frame PID” effect cannot explain the relative low degradation rate for island/coast region systems compared to other regions. But there may be some systems among the other region systems that show “Surface PID” leading to very high degradation rates that influence the statistic.

DISCUSSION

The occurrence distribution of failures over the operational years deviates from a typical bath tube curve. A reason for the difference may be that the number of installed systems changes exponentially with the year of installation. We have much more young systems in the world than old ones as shown in Fig. 11. Therefore the failure occurrence is highly biased by not equally distributed annual system installations. This may be the reason for the high failure level at the first seven years in Fig. 5. Especially the sudden occurring failures may be highly biased by this effect.

Maybe in coastal region already systems are affected by PIDs if the PIDs effect would not occur for the same systems in the other regions. And in the other regions we find almost heavily degrading PIDs affected systems. The degradation rates 𝑑x of failure x affected modules are important for module manufacturers. It shows how fast modules degrades caused y failure x. The degradation rate 𝛿x̅ of the investigated PV system part is important for the bankability of PV systems. Under the assumption that the people who analysed the failures in the system choose a representative part of the PV system for the failure analysis the degradation rates 𝛿x̅ are representative for such systems. 7

SUMMARY AND CONCLUSIONS

The current status of the data collection of PV system failures is preliminary. For some cases results can be drawn from the data. The degradation rate due to cell cracks is highest in cold and snow climates. Cell cracks dominate the early failures in year one and two after installation. However the system degradation rate stays below 3%/a for all climatic zones, while the degradation rates of the affected part is highest with about 8%/a in the cold and snow climate. The next dominant failure is the potential induced shunts effect in year three and four with mean degradation rates of the system of 9%/a and 16%/a for the affected parts in the moderate zone. The PIDs effect occurs 15 times more often in coastal/island regions than in other regions, but is less severe than in other region. Defect bypass diodes heavily affect the module and the system power. The failure occurrence is spread over the first 10 years of system operation. The discolouring of the pottant is a quite often seen wear out failure. But for all climate zones this is effect is in the mean well below 1%/a for the system and the affected part. In our future work we will expand the analysis also for PV system components.

Fig. 11: Evolution of global annual PV system name plate installations, composed from numbers in [22]. The failure type soiling does not fit into degrading and sudden failures category, because the power of soiled module degrades over time but can fully be recovered. We see a dependence of dust soiling on the climate zones in Fig. 9. However the high mean power loss due to dust in the moderate zone (6%) does not coincident with the expectation to find higher dust soiling losses the hot and dry climate (4%). Probably dust soiling is strongly influenced by local conditions. Herrmann developed a model on dust soiling based on weather data and other influencing factors [23]. He showed a correlation between desert and no desert regions. To prove this correlation we would need a much more detailed climate zone resolution as performed in this study.

8 ACKNOWLEDGMENTS Funding was provided by the Federal Ministry for Economic Affairs and Energy (BMWi) under contract no. FKZ 0325786C/A Task13 Projekt, the State of Lower Saxony, and by the Austrian Climate and Energy Fund (KliEn) within the 1st e!MISSION energy research call 2014, project Infinity, under contract no. 850.414. We like to thank Iris Kunze and Arnaud Morlier for supporting the data collection. Furthermore we thank all TASK13 members and experts who contributed survey data for the database.

The anti-correlation between mean degradation rate and occurrence of PIDs in coastal regions/island systems is unexpected. Berghold presented a classification of PIDs modules into “Frame PID” where only the cells close to the frame are affected and “Surface PID” where

1440


8

REFERENCES

978–989, Jul. 2016. [13] M. Köntges, S. Kurtz, C. Packard, U. Jahn, K. A. . Berger, K. Kato, T. Friesen, H. Liu, M. Van Iseghem, J. Wohlgemuth, D. Miller, M. Kempe, P. Hacke, F. Reil, N. Bogdanski, W. Herrmann, C. Buerhop-Lutz, G. Razongles, and G. Friesen, Review of Failures of Photovoltaic Modules. IEA, 2014. [14] M. Köntges, “PV-failure_survey, 27th Oct. 2015,” 151027__PV-failure_survey.xlsm, 2015. [Online]. Available: http://ieapvps.org/index.php?id=344&eID=dam_frontend_push&d ocID=2918. [Accessed: 29-Jan-2016]. [15] M. Köntges and S. Altmann, “Explanation of the PV System Survey sheet,” 151027_Explanation_PVSystem_Survey.pdf. [Online]. Available: http://ieapvps.org/index.php?id=344&eID=dam_frontend_push&d ocID=2919. [Accessed: 26-Jan-2016]. [16] P. Sánchez-Friera, M. Piliougine, J. Peláez, J. Carretero, and M. Sidrach de Cardona, “Analysis of degradation mechanisms of crystalline silicon PV modules after 12 years of operation in Southern Europe,” Prog. Photovoltaics Res. Appl., vol. 19, no. 6, pp. 658– 666, Sep. 2011. [17] T. Friesen, “PV Modules Failures Modes Observed in Real Use and Long Term Exposure,” in PVModule Reliability, Berlin, April 5-6, 2011. [18] F. Spertino, A. Ciocia, P. Di Leo, R. Tommasini, I. Berardone, M. Corrado, A. Infuso, and M. Paggi, “A power and energy procedure in operating photovoltaic systems to quantify the losses according to the causes,” Sol. Energy, vol. 118, pp. 313–326, Aug. 2015. [19] H. Häberlin and P. Schärf, “Langzeitverhalten von PV-Anlagen über mehr als 15 Jahre Prof.,” in 25. Symposyum Photovoltaische Solarenergie, 2010, p. 0043_8050_3. [20] B. Burger, K. Kiefer, C. Kost, S. Nold, S. Philipps, R. Preu, J. Rentsch, T. Schlegl, G. Stryi-Hipp, G. Willeke, H. Wirth, I. Brucker, A. Häberle, and W. Warmuth, “Photovoltaics Report,” Fraunhofer Institute for Solar Energy Systems, ISE, Freiburg, 2015. [21] J. Berghold, S. Koch, S. Pingel, S. Janke, A. Ukar, P. Grunow, and T. Shioda, “PID: from material properties to outdoor performance and quality control counter measures,” in SOPHIA Workshop, p. 95630A. [22] BP, “BP Statistical Review of World Energy June 2016,” Excel workbook, 2016. [Online]. Available: http://www.bp.com/content/dam/bp/excel/energyeconomics/statistical-review-2016/bp-statistical-reviewof-world-energy-2016-workbook.xlsx. [23] J. Herrmann, K. Slamova, R. Glaser, and M. Köhl, “Modeling the Soiling of Glazing Materials in Arid Regions with Geographic Information Systems (GIS),” Energy Procedia, vol. 48, pp. 715–720, 2014.

[1] R. Dubey, S. Chattopadhyay, V. Kuthanazhi, J. J. John, B. M. Arora, A. Kottantharayil, K. L. Narasimhan, C. S. Solanki, V. Kuber, V. Vasi, A. Kumar, and O. S. Sastry, “All-India India Survey of Photovoltaic Module Degradation : 2013,” 2013. [Online]. Available: www. ncpre. iitb. ac. in/uploads/All_India_Survey_of_Photovoltaic_Module_ Degradation_2013. pdf. [2] A. Jacobson, D. Kammen, R. Duke, and M. Hankins, “Field performance measurements of amorphous silicon photovoltaic modules in Kenya,” in The Solar Conference, 2000, pp. 95–100. [3] A. B. Maish, C. Atcitty, S. Hester, D. Greenberg, D. Osborn, D. Collier, and M. Brine, “Photovoltaic system reliability,” in Conference Record of the Twenty Sixth IEEE Photovoltaic Specialists Conference - 1997, 1997, pp. 1049–1054. [4] K. Stokes and J. Bigger, “Reliability, cost, and performance of PV-powered water pumping systems: a survey for electric utilities,” IEEE Trans. Energy Convers., vol. 8, no. 3, pp. 506–512, 1993. [5] K. Kato, “PVRessQ!: a research activity on reliability of PV systems from an user’s viewpoint in Japan,” in SPIE 8112, Reliability of Photovoltaic Cells, Modules, Components, and Systems IV, 81120K (21 September 2011), 2011, p. 81120K–81120K–9. [6] A. Skoczek, T. Sample, and E. D. Dunlop, “The results of performance measurements of field-aged crystalline silicon photovoltaic modules,” Prog. Photovoltaics Res. Appl., vol. 17, no. 4, pp. 227–240, Jun. 2009. [7] R. J. van der Plas and M. Hankins, “Solar electricity in Africa: a reality,” Energy Policy, vol. 26, no. 4, pp. 295–305, Mar. 1998. [8] S. Kumar, S. C. Bhattacharya, and M. A. Leon, “A survey on PV systems and accessories in Asia,” in In: Sayigh AAM, editor. World Renewable Energy Congress VI. Oxford: Pergamon, 2000, pp. 860–863. [9] S. Djordjevic, D. Parlevliet, and P. Jennings, “Detectable faults on recently installed solar modules in Western Australia,” Renew. Energy, vol. 67, pp. 215– 221, Jul. 2014. [10] E. D. Dunlop, D. Halton, and H. A. Ossenbrink, “20 years of life and more: where is the end of life of a PV module?,” in Conference Record of the Thirty-first IEEE Photovoltaic Specialists Conference, 2005., pp. 1593–1596. [11] D. C. Jordan and S. R. Kurtz, “Photovoltaic Degradation Rates-an Analytical Review,” Prog. Photovoltaics Res. Appl., vol. 21, no. 1, pp. 12–29, Jan. 2013. [12] D. C. Jordan, S. R. Kurtz, K. VanSant, and J. Newmiller, “Compendium of photovoltaic degradation rates,” Prog. Photovoltaics Res. Appl., vol. 24, no. 7, pp.

1441


APPENDIX Table 1: List of PV module failures being selectable in the survey. Type Image Type Delamination Burn marks Photo: c-Si solar cells delaminte from the cell interconnect ribbon.

Photo: Burn mark along cell edge on module front side

Defect junction box

Potential induced shunts (often named PID)

Photo: lid of junction box fell of and pottant discolor

Junction box detached Photo: junction box is not fixed to the module any more

Photo: some spots indicate the shunt positions

Potential induced corrosion (often with thin film modules) Photo: thin film module, corrosion near edge below clamp

Discolouring pottant

of

Photo: back sheet of module discolored

Cell cracks Photo: diagonal cell cracks visible in browned EVA on top of a cell

Disconnected cell or string interconnect ribbon Photo: disconnected cell interconnect ribbon

Defective bypass diode/wrong dimensioned Photo: molton plastic around the bypass diodes

1442

Image


Type Defect backsheet

Image

Type Hail -> glass breakage/cell breakage

Potential induced shunts (often named PID)

Image

Snow load -> deformed frame/glass- /cellbreakage

Electroluminescence image.

Corrosion/abrasion of AR coating

Direct lightning stroke -> defect glass/frame and defect bypass diodes

Glass breakage Excluding other glass breakage reasons also listed here.

Biofilm soiling

Isolation failure

CdTe: back contact degradation

Frame breakage/bent/defect

Source: Sulaiman bin Shaari

Any part of the module shows an isolation failure incl. the cables and connectors of the module.

Dust soiling

Temperature and voltage induce degradation of the CdTe back contact. Can be identified by fill factor loss and “roll over” of the IV curve. Any damage on the frame, exept caused by special evens also listed here.

Animal -> bite/corrosion/dirt

Any damage, corrosion or dirt caused the system to loos power.

Storm -> deformed frame/glass-/cellbreakage

Any damage caused by a storm.

Source: Uwe Hupach

1443