Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis

DOCTORATE PROGRAM IN INDUSTRIAL ENGINEERING DOCTORAL THESIS: DETECTION, CLASSIFICATION AND CHARACTERIZATION OF DEFECTS IN PHOTOVOLTAIC MODULES THROUGH THE USE OF THERMOGRAPHY, ELECTROLUMINESCENCE, I-V CURVES AND VISUAL ANALYSIS Presented by Sara Gallardo Saavedra to qualify for the degree of Doctor from the University of Valladolid Directed by: Luis Hernández Callejo Óscar Duque Pérez PROGRAMA DE DOCTORADO EN INGENIERÍA INDUSTRIAL TESIS DOCTORAL: DETECCIÓN, CLASIFICACIÓN Y CARACTERIZACIÓN DE DEFECTOS EN MÓDULOS FOTOVOLTAICOS MEDIANTE LA UTILIZACIÓN DE TERMOGRAFÍA, ELECTROLUMINISCENCIA, CURVAS I-V Y ANÁLISIS VISUALES Presentada por Sara Gallardo Saavedra para optar al grado de Doctora por la Universidad de Valladolid Dirigida por: Luis Hernández Callejo Óscar Duque Pérez A Mario ACKNOWLEDGMENTS / AGRADECIMIENTOS Me gustaría aprovechar esta página para expresar mi gratitud a todas las personas que han hecho posible la realización de esta tesis doctoral. En primer lugar, quisiera dar las gracias a mis dos directores de tesis. Al Dr. Luis Hernández Callejo por su disponibilidad y dedicación, por todo lo que me ha enseñado y aconsejado, por confiar en mí y por su motivación e interés en la investigación, permitiendo a tantos estudiantes crecer, desarrollarse y completar su formación. Al Dr. Óscar Duque Pérez, por ayudarme siempre que lo he necesitado, por todos los conocimientos que me ha transmitido, por su dedicación, su criterio, su disponibilidad y su paciencia. Gracias a los dos por haberos esforzado por ayudarme a llegar al punto en el que hoy me encuentro. Me gustaría agradecer a la Escuela de Ingeniería de la Industria Forestal, Agronómica y de la Bioenergía (EIFAB) por haberme dado la oportunidad de desarrollar mi tesis en el departamento de Ingeniería Agrícola y Forestal y por haberme facilitado los medios necesarios, así como a la Universidad de Valladolid, que ha hecho posible la realización de esta tesis doctoral gracias a su apoyo de financiación predoctoral. Además, quiero expresar mi más profundo agradecimiento a todos los compañeros de la EIFAB, de GDS OPTRONLAB, del CIEMAT, de SOLARIG y al resto de profesionales que han colaborado conmigo durante estos cuatro años, por toda la sabiduría que me han transmitido, por los buenos momentos compartidos, por sus consejos, por estar a mi disposición siempre y ante todo, por su gran valor como personas. Por último, quiero dar las gracias a los incondicionales, a mis pilares. Me gustaría agradecer a mis amigos por sus largas horas de compañía, consejos y alegría, a Mario su cariño durante todos estos años y de una forma muy especial a mis padres, César y Milagros, y a mi hermana, Elena, su incondicional apoyo y el ejemplo de esfuerzo y honestidad que siempre han sido para mí. Me siento afortunada porque sé que esta tesis doctoral existe gracias a las personas que me rodean, así que a todas ellas mi más sincero agradecimiento. A todos ellos gracias. ‘If I have seen further it is by standing on the shoulders of Giants.’ Isaac Newton. VII VIII PROLOGUE / PREÁMBULO De acuerdo con la normativa vigente de presentación y defensa de la tesis doctoral (RESOLUCIÓN de 8 de junio de 2016, del Rectorado de la Universidad de Valladolid, por la que se ordena la publicación del Acuerdo del Consejo de Gobierno de 3 de junio de 2016, por el que se aprueba la normativa para la presentación y defensa de la tesis doctoral en la Universidad de Valladolid), esta Tesis Doctoral se presenta como compendio de publicaciones. A continuación se indican los artículos publicados incluidos dentro del compendio de publicaciones que ha dado lugar a la tesis, los cuales se adjuntan en el CAP. 2 Artículos publicados. Los artículos se han organizado siguiendo el orden conceptual que da forma y sentido a la tesis doctoral. Artículo 1: Aceptado y publicado. Título: Quantitative failure rates and modes analysis in photovoltaic plants. Revista: Energy, volumen 183, pp. 825-836 (2019). DOI: https://doi.org/10.1016/j.energy.2019.06.185. Autores: Sara Gallardo Saavedra, Luis Hernández Callejo y Óscar Duque Pérez. Web of Science (WoS): Journal Citation Reports (JCR) Factor de impacto: 5.537 (2018) Categoría y Ranking: Energy & Fuels, Q1 (15/103) Scimago Journal & Country Rank (SJR) Factor de impacto: 2.05 (2018) Categoría y Ranking: Energy, Q1 (5/136) Artículo 2: Aceptado y publicado. Título: Failure rate determination and Failure Mode, Effect and Criticality Analysis (FMECA) based on historical data for photovoltaic plants. Proceedings congreso: ISES Conference Proceedings of the ISES Solar World Conference 2017 and the IEA SHC Solar Heating and Cooling Conference for Buildings and Industry 2017, pp. 1232-1239. DOI: https://doi.org/10.18086%2Fswc.2017.20.04 Autores: Sara Gallardo Saavedra, Javier Pérez Moreno, Luis Hernández Callejo y Óscar Duque Pérez. IX XVI RESUMEN El reciente crecimiento que han experimentado las energías renovables ha sido liderado por la energía solar fotovoltaica (FV), siendo FV más de la mitad de la energía renovable instalada en 2018. Poder detectar, identificar y cuantificar la gravedad de los defectos en módulos es esencial para constituir sistemas fiables, eficientes y seguros, evitando pérdidas de energía, desajustes y problemas de seguridad. Por lo tanto, la idea de Operación y Mantenimiento avanzado (O&M) es muy atractiva para el sector de explotación FV, ya que garantizará ciertos niveles de eficiencia en sus plantas. Tradicionalmente se han utilizado diferentes técnicas para detectar anomalías en módulos FV, como inspecciones visuales, termografía manual o pruebas eléctricas (curvas I-V). Los avances tecnológicos permiten el desarrollo de técnicas innovadoras de bajo costo, mejorando los laboriosos métodos manuales tradicionales. Esta tesis doctoral presenta un análisis de la detección, caracterización y clasificación de defectos en módulos FV mediante el uso de termografía infrarroja (IRT), EL, curvas I-V y análisis visual, enfocándose en tres líneas principales. En primer lugar, se analiza el estado del arte de los sistemas fotovoltaicos, incluyendo la investigación sobre las tasas y modos de fallo en plantas FV y el diseño, operación y mantenimiento de las plantas. En segundo lugar, se realiza la detección, caracterización y clasificación de defectos en módulos FV, mediante los siguientes enfoques: el análisis y comparación de los resultados experimentales obtenidos con las diferentes técnicas de inspección, la generación de imágenes fusionadas, el desarrollo de circuitos de simulación, el estudio del efecto de las pruebas de electroluminiscencia (EL) en la vida útil de los módulos, el análisis cuantitativo y caracterización de defectos de fabricación, soldadura y roturas en células solares y la clasificación de defectos en función de sus patrones de fallo. En tercer lugar, en referencia al análisis de la termografía aérea como nueva técnica de inspección, las principales contribuciones del trabajo son tres: revisar el equipamiento disponible, enfocándose en las características deseables para su aplicación en inspecciones termográficas aéreas de plantas FV, la revisión de los aspectos más importantes que deben tenerse en cuenta en las pruebas para obtener resultados válidos y el estudio de la influencia de la resolución de las imágenes IRT. Es importante señalar que la investigación realizada y las contribuciones presentadas han surgido de necesidades reales propuestas por operadores, fabricantes y propietarios de instalaciones FV, no solo de España sino también de diversos países, lo cual enfatiza la importancia de la aplicación práctica del estudio desarrollado. XVII XVIII INDEX ACKNOWLEDGMENTS / AGRADECIMIENTOS ........................................................ VII PROLOGUE / PREÁMBULO ........................................................................................ IX ABSTRACT ................................................................................................................. XV RESUMEN ................................................................................................................. XVII CHAP. 1 INTRODUCTION ........................................................................................ 23 1.1. Conceptual framework ................................................................................. 24 1.2. Justification .................................................................................................. 28 1.3. Hypothesis and objectives .......................................................................... 30 1.4. Collaborations with other research groups ............................................... 31 1.5. Project phases and methodology ............................................................... 32 1.5.1. Project inception and theoretical grounding phase .............................. 33 1.5.2. Analytical-dimensioning phase ............................................................ 33 1.5.3. Empirical research phase .................................................................... 36 1.5.4. Analytical-comparative phase .............................................................. 46 1.5.5. Results dissemination phase ............................................................... 50 1.6. Results and discussion ............................................................................... 51 1.6.1. Study of the state of the art (Obj. 1) .................................................... 52 1.6.2. Detection, characterization and classification of defects in PV modules through the use of different inspection techniques (Obj. 2) ................................. 56 1.6.2.1. Detection of PV defects ................................................................... 56 1.6.2.2. Characterization of PV defects ........................................................ 61 1.6.2.3. Classification of PV defects ............................................................. 71 1.6.3. Analysis of aerial IRT as a novel inspection technique (Obj. 3) .......... 75 1.7. References .................................................................................................... 80 1.8. Abbreviations ............................................................................................... 91 1.9. Index of figures ............................................................................................. 92 XIX Index 1.10. Index of tables ........................................................................................... 95 CHAP. 2 PUBLISHED PAPERS ............................................................................... 97 2.1. Study of the state of the art (Obj. 1) ........................................................... 98 2.1.1. Quantitative failure rates and modes analysis in PV plants ................. 99 2.1.2. Failure rate determination and Failure Mode, Effect and Criticality Analysis (FMECA) based on historical data for photovoltaic plants .................. 101 2.1.3. A review of photovoltaic systems: design, operation and maintenance .. ........................................................................................................... 103 2.2. Detection, characterization and classification of defects in PV modules through the use of different inspection techniques (Obj. 2) ............................ 105 2.2.1. Simulation, validation and analysis of shading effects on a PV system .. ........................................................................................................... 106 2.2.2. Influence of large periods of DC current injection in c-Si photovoltaic panels ........................................................................................................... 108 2.2.3. Failure diagnosis on photovoltaic modules using visual inspection, thermography, electroluminescence and I-V techniques ................................... 110 2.2.4. Merged images for fault detection in photovoltaic panels .................. 112 2.2.5. Nondestructive characterization of solar PV cells defects by means of electroluminescence, infrared thermography, I-V curves and visual tests: experimental study and comparison .................................................................. 114 2.2.6. Analysis and characterization of PV module defects by thermographic inspection. ......................................................................................................... 116 2.3. Analysis of aerial thermography as a novel inspection technique (Obj. 3) ..................................................................................................................... 118 2.3.1. Technological review of the instrumentation used in aerial thermographic inspection of photovoltaic plants ....................................................................... 119 2.3.2. Aerial thermographic inspection of photovoltaic plants: analysis and selection of the equipment ................................................................................ 121 2.3.3. Low-cost infrared thermography in aid of photovoltaic panels degradation research ........................................................................................ 123 XX Index 2.3.4. Image Resolution Influence in Aerial Thermographic Inspections of Photovoltaic Plants. ........................................................................................... 125 CHAP. 3 CONCLUSIONS AND FUTURE WORK .................................................. 127 3.1. Conclusions and PhD contributions ........................................................ 128 3.2. Future work ................................................................................................. 132 ANNEX A ADDITIONAL PUBLICATIONS ................................................................ 135 XXI XXII Chapter 1 CAP. 1 INTRODUCTION ‘Nothing in life is to be feared, it is only to be understood. Now is the time to understand more, so that we may fear less.’ Marie Curie. - 23 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis This doctoral thesis begins with an introduction that justifies the thematic relationship of the publications presented in the “compendium of publications” and the relevance of their joint contribution. To do this, the Introduction chapter starts explaining the conceptual framework, followed by the justification of the thesis. Thirdly, the global and partial objectives pursued by the investigation are described. The collaborations carried out with other research groups are detailed below, in the fourth section. This gives way to the phases of the project and the explanation of the methodology used, in section five, for ending in section six with the summary of the main results obtained and their discussion. 1.1. Conceptual framework Renewable energy is now a fully mainstream element in the global electricity mix. Alongside energy efficiency, renewables are playing a critical role in reducing emissions in the energy sector and in end-use sectors. In many locations, new renewable energy is now the lowest-cost way to provide electricity services and can be brought online the fastest. Around the world, renewable electricity has spread thanks to both transferable and reliable technologies and effective policy frameworks. The recent growth in renewable power capacity has been led by solar photovoltaics (PV), with 100 GW of new solar PV capacity installed in 2018 of the more than 180 GW of renewable power installed this year, reaching a total installed PV solar capacity of 505 GW [1,2]. Low-power PV solar plants are proliferating in recent decades, sometimes due to the lack of electricity supply at one site, sometimes due to their geographical conditions (for instance in islands) and others due to the simple fact of reducing energy dependence on fossils fuels and approaching quasi sustainability [3–5]. But in the same way that lowpower PV solar plants are being installed, it is more than likely that large-scale ones are installed much more [6–9], since the need for large-scale power generation is a need, together with the interest of maximizing the benefit of the installation [10,11]. High-power PV solar plants have a common feature, which is that they are formed by a large number of PV solar modules [12,13], and it is expected that in the future they will increase in size (in power and number of modules) [14,15]. It is well-known that the cost of purchasing PV solar modules has drastically fallen in recent decades, and especially in recent years [16–18], which makes the idea of installing PV solar plants even more attractive. Almost all the efforts of the PV exploitation - 24 of 137 - Chapter 1: Introduction are focusing on carrying out inspections of their plants (maintenance) that allow them to maintain their efficiency (operation), with the clear objective of obtaining high productivity levels. Therefore, the idea of advanced Operation and Maintenance (O&M) is very attractive for the PV exploitation sector, since it will guarantee certain levels of efficiency in their PV plants [19,20]. PV cells are basically made up of a PN junction. When sunlight hits the cell, the photons are absorbed by the semiconductor atoms, freeing electrons from the negative layer. This free electron finds its path through an external circuit toward the positive layer resulting in an electric current from the positive layer to the negative one [21]. Crystalline silicon (c-Si) solar cells have dominated the PV market since the very beginning in the 1950’s. Silicon is non-toxic and abundantly available in the earth crust, silicon PV modules have shown their long-term stability over decades in practice [22,23]. Three major families of PV cells are monocrystalline technology, polycrystalline technology and thin-film technologies. For this reason, this research has been performed using monocrystalline and polycrystalline PV cells, although in the market exist other crystalline semiconductor based PV cells in a minority, as Germanium (Ge), Gallium Arsenide (GaAs), Cadmium Telluride (CdTe), or even more complex formulas, like Indium- Gallium-Arsenide (InGaAs) [13]. Defects at the PV solar cells level are really important, since these are the ones that make up the defects at the PV modules, being responsible for the reduction of efficiency, as well as their durability and reliability. There has been extensive research on the types of failures [24–29], as it would be interesting to have them classified and even being able to estimate them. The failures in the plants are numerous and very heterogeneous, so their research is a challenge of interest to the scientific world [30–32]. Faults can appear in manufacturing, transportation, installation and operation. Many different failure modes can appear at the module level. In [33] four different groups of failures found in PV modules are defined. In the first group, it is described the common failures to all PV modules divided in the following failure modes: delamination, back sheet adhesion loss, junction box failure and frame breakage. The second group includes the following failures in silicon wafer-based PV modules: EVA discoloration, cell cracks, snail tracks, burn marks, Potential Induced Degradation (PID), disconnected cell and string interconnect ribbons, defect bypass diode. The thin-film failures described in the third group are: micro arcs at glued connectors and shunt hot spots. And finally, the fourth group describes the following failures found in CdTe thin-film modules: front glass breakage and back contact degradation. In reference [34] different aging mechanisms - 25 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Additionally, for the analysis of the equipment used in aerial thermographic inspections, the first year of the thesis there has also been a collaboration with the School of Electrical and Electronic Engineering of the Valle University in Colombia. From the second year of the doctoral period, the PhD labor has been focused as part of the working group for the research projects ENE2017-89561-C4-R-3 and RTC- 2017-6712-3 of the Spanish Ministry of Science, Innovation and Universities, in collaboration with GdS-Optronlab group. During the third year, it has been cooperation with the National Polytechnic Institute of Mexico for the processing of EL and thermographic merged images and with the Department of Electrical Engineering of the Faculty of Engineering of the University of Cuenca, in Ecuador, for the low-cost IRT application. Throughout the fourth year of the doctoral thesis, a three-month stay in the Photovoltaic Solar Energy Unit at the Energy department of the Center for Energy, Environmental and Technological Research (CIEMAT) in Madrid has been accomplished. There, all the specialized PV equipment of the center has been accessible for the defects characterization and for the inspection techniques comparison. The renewable energies investigation group at the EIFAB has a research collaboration agreement with Solarig, a worldwide operating company with more than 4,000 MW distributed in twenty countries in four continents, that has contributed during the entire doctoral thesis with access to some sites to perform on-site measurements, documentation and experimental data of the operation of its PV plants. 1.5. Project phases and methodology This section explains the methodology followed in the development of the doctoral thesis divided into the five different phases of the project. This section starts with the first phase, the project inception and theoretical grounding, followed by the analytical-dimensioning phase. These two phases allow the understanding of the level of development of PV sites and of the use of different techniques at present, determining the key points to be analyzed and highlighting the importance of the research. The third phase is the empirical research, which comprises the empirical objective and field work. In this phase, the design, development and validation of the different campaigns of measures necessary to carry out the research project are carried out. The fourth phase, - 32 of 137 - Chapter 1: Introduction the analytical comparative phase, is focused on extracting results and drawing conclusions from the empirical data. Finally, the last phase details the dissemination of the results of the research performed. 1.5.1. Project inception and theoretical grounding phase The research work has a first theoretical-analytical phase. The collection of theoretical information is based primarily on peer-reviewed articles published in scientific journals, prioritizing those with high impact factor, as well as books and other sources of information such as conference or congress articles. The library from the UVa (both physically and through the website) has made most of these means accessible. Additionally, it is also remarkable the countless equipment manufacturers, resellers and professional operators who have contributed to the state-of-the-art review, showing the latest trends and key points for the specific cases studied. In this theoretical phase, it has been reviewed, analyzed and synthetized the following information (in parentheses it is indicated the thesis objective related with each theoretical study): • Failure rates and modes in PV systems (Obj. 1). • Design, operation and maintenance of PV plants (Obj. 1). • Typology and relevance of defects at the module level (Obj. 1 and Obj. 2). • State of the art of PV module inspection techniques (Obj. 2 and Obj. 3). • Modelling of solar cells (Obj. 2). • Aerial thermographic inspection of PV plant related work (Obj. 3). • Instrumentation requirements to carry out aerial inspections, including the study of the most suitable characteristics for this application (Obj. 3). • Analysis of the best flight conditions and influencing factors (Obj. 3). This phase has been mainly completed in the Department of Agricultural and Forestry Engineering of the EIFAB. 1.5.2. Analytical-dimensioning phase Within the study of the state of the art (Obj. 1), it is included the analysis of the failures rates and modes in PV plants for the dimensioning of the problem studied. To complete this phase, the alarms in a PV portfolio have been analyzed, corresponding to the ones generated during five years (2012-2016) in sixty-three different PV sites distributed along Italy and Spain, from the smallest of 200 kW to the highest of 10,000 kW. The data has been obtained through the PV plants monitoring systems and the - 33 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis feedback of the field technicians, which feed the Computerized Maintenance Management Systems (CMMS) of the operator. With this information, it has been performed a quantitative analysis of failure rates and modes in PV plants and a Failure Rate, Effect and Criticality Analysis (FMECA) based on PV plants historical data. For the quantitative analysis of failure rates and modes, the PV portfolio has been divided into five different groups depending on their Power (p): p≤750 kW, 750<p≤995 kW, 995<p≤1,300 kW, 1,300<p≤2,300 kW and 1,300<p≤10,000 kW. The main information that can be obtained from the CMMS of each of the alarms is listed right after: state, ID alarm number, type, name, PV plant, country, element affected, element affected code, activation time, repercussion, priority, deactivation time, technician, cause, description, actions performed and report state. In order to process the data, the PV system has been simplified considering the main groups of elements that compose a PV site: PV generator, inverter, Medium Voltage (MV) transformer station, metering elements, security system, communication system, monitoring system, grid and civil works. Along the paper, different statistical indicators are used, as the mean, the typical deviation and the Mean Time Between Failures (MTBF). The latter can be defined as the mean exposure time between consecutive failures of a component. It can be estimated by dividing the exposure time by the number of failures in that period, provided that enough failures have occurred in that period [72]. It has been calculated using the following (Eq. 1). Further information about this study methodology can be found in the paper presented in subsection 2.1.1. MTBF= Power [kW] Number of failures [#failures] Period [year] (Eq. 1) Additionally, for the FMECA, the failure rates, the MTBF and the subjective evaluations of solar experts with more than a decade experience are used to define three ranking criterions: Severity, Detection and Occurrence. The Occurrence based on the failure rates calculated is classified following this criterion: 1 for unlikely failures, 2 for remote probability, 3 for occasional probability, 4 for moderate probability and 5 for high probability. Each ranking follows a scale from 1 to 5, in which 1 denotes the best situation while 5 denotes the worst. As a result, the Risk Priority Number (RPN) goes from 1 to 125, where a highest value of RPN indicates a most risky situation. Finally, thermographic tests have been applied at a module level to complete the FMECA of the whole PV Plant. Further information about this study methodology can be found in the paper presented in subsection 2.1.2. - 34 of 137 - Chapter 1: Introduction Table 1: Severity ranking intervals calculated from the fixing price and the loss of profit that a failure produces Loss of Profit (LOP) 57.000 € 13.300 € 950 € 190 € - € Fixing Price (Spares and Workforce) 60.000 € 117,000 € 73,300 € 60,950 € 60,190 € 60,000 € 15.000 € 72,000 € 28,300 € 15,950 € 15,190 € 15,000 € 1.000 € 58,000 € 14,300 € 1,950 € 1,190 € 1,000 € 500 € 57,500 € 13,800 € 1,450 € 690 € 500 € - € 57,000 € 13,300 € 950 € 190 € 0 € Table 2: Severity ranking criteria Rank X (Fixing Cost + LOP) Description 1 X <690 € Minor failure with almost no influence in the performance of the plant and insignificant parts deterioration 2 1,950 €>X>690 € Failure with low influence in the performance of the plant and parts deterioration 3 28,300 €>X>1,950 € Failure with quite important influence in the performance of the plant and parts deterioration 4 60,950 €>X> 28,300 € Failure with important influence in the performance of the plant and parts deterioration 5 X >60,950 € Major failure with extreme influence in the performance of the plant and parts deterioration Table 3: Detection ranking criteria following [73]. Rank Description 1 Almost certain that the problem will be detected (chance 81–100%) 2 High probability that the problem will be detected (chance 61–80%) 3 Moderate probability that the problem will be detected (chance 41–60%) 4 Low probability that the problem will be detected (chance 21–40%) 5 None/minimal probability that the problem will be detected (chance 0–20%) This phase has also been mainly completed in the Department of Agricultural and Forestry Engineering of the EIFAB in collaboration with Solarig. - 35 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 1.5.3. Empirical research phase Subsequent stages comprise the empirical objective, including the fieldwork and the analysis and treatment of the data and images obtained. This subsection expounds the methodology used in this phase, and it has been divided considering the facilities, instrumentation and experimental techniques and the modules tested. • Facilities, instrumentation and experimental techniques To facilitate the understanding of different failure mechanisms and their effects on the performance of the PV module, simulation, validation and analysis of the electrical model of faulty and shaded modules have been carried out. This research has been performed in collaboration with the Department of Building, Energy and Environmental Engineering of the Faculty of Engineering and Sustainable Development at the University of Gävle, in Sweden. There, two different kinds of tests have been done to validate the simulation results: to a string of six modules and to an individual module at Högskolan i Gävle (HIG) laboratory. In each group, seven different configurations have been studied. These cases start from the reference case, in which any cell is defective or any shading has been induced in the module. After that, an opaque tape is added to generate the shading or simulate the cells defects. The string of modules can be seen in the upper left part of Figure 9 and one of the seven shading configurations studied is presented in Figure 10. For all the shading configurations tested, the benefits of installing optimizers in each module will be considered and detailed in the discussion. The DC-DC optimizers installed at HIG laboratory are dual maximizers model MM-2ES 50 of Tigo Energy manufacturer. Further information about this study methodology can be found in the paper presented in subsection 2.2.1. An important part of the experimental investigations for failure diagnosis on PV modules using visual inspection, IRT, EL and I-V techniques have been performed in the EIFAB facilities, at the Campus Duques de Soria of the Uva. There, indoor and outdoor tests have been conducted. Between the indoor tests, EL has been performed in controlled ambient conditions simultaneously than IRT in the fourth quadrant. For these tests, the University has a temperature and humidity controlled chamber, which is shown in Figure 1. In this chamber, each module is continuously fed with its short circuit current, using a laboratory source, during 72 hours. EL images are captured with a PCO 1300 camera each 30 minutes with an exposure of 5,000 ms. Thermal images have been captured with a Flir C2 system and a Workswell Wiris Pro camera. This capturing system - 36 of 137 - Chapter 1: Introduction is presented in Figure 2, in which can be seen the EL camera, the IR camera and the PC which control the acquisition. Figure 1: Temperature and humidity controlled chamber at the EIFAB in Soria, Spain Figure 2: EL and IR imaging capturing system used in the temperature and humidity controlled chamber Outdoor tests include I-V curves, IRT and Red, Green Blue (RGB) images in the PV field of the Campus Duques de Soria, which can be seen in Figure 3. The thermal camera used outdoor is a Workswell Wiris Pro camera, with a 640x512 pixels resolution, a thermal sensitivity of 0.05⁰C and an accuracy of ±2% o ±2°C. Additionally, the camera has a frame rate of 30Hz, is calibrated to be used with two different lenses, 32⁰ y 69⁰, and includes a Full HD RGB sensor with a resolution of 1920x1080 pixels and x10 zoom. The I-V curves have been traced using an IV HT 1500V Solar I-V tracer. Figure 3: PV field of the Campus Duques de Soria, University of Valladolid. The results obtained throughout these empirical assessments have been used to feed the studies performed on the influence of large periods of Direct Current (DC) injection in c-Si PV panels (subsection 2.2.2), failure diagnosis on PV modules using - 37 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis visual inspection, IRT, EL and I-V techniques (subsection 2.2.3) and merging EL and thermographic images (subsection 2.2.4). The nondestructive characterization of solar PV cells defects by means of EL, IRT, I-V curves and visual tests has been performed in the Photovoltaic Solar Energy Unit at the Energy department of CIEMAT, in Madrid. Firstly, the indoor measurements have been performed in the Solar Simulator Pasan SunSim 3CM, class A, flash type. Starting with the I-V curves acquisition at two different irradiance levels, 1,000 W/m2 and 100 W/m2, with temperature controlled conditions of 25±0.5⁰C. The second irradiation level has been chosen to characterize the defects behavior in low irradiation conditions. With the objective of better characterizing the defects behavior, EL tests to individual cells have been performed to identify the light emission caused by radiative recombination of carriers in each case. During this EL test, the temperature of the different cells has been measured with IRT. In this case, the module is fed with a power source and the EL and IR images are captured with a PCO 1300 and a FLIR SC 640 camera, respectively. EL tests to individual defective cells have been performed at the Isc injected current, 8.9 A, a focal distance of 4 cm and exposure time of 40 s. EL tests to the whole module have been performed at two injected current levels, 8.9 A and 0.89 A, with a focal distance of 4 cm and exposure time 50 s and 15 min, respectively. This image acquisition is presented in Figure 4. RGB images of the module have been taken with a Nikon D80 camera and images in detail of defects in the PV laboratory at the CIEMAT in Madrid with a Nikon SMZ 800 microscope. Finally, IRT has also been performed outdoor using the same IR camera, as presented in Figure 5. Figure 4: Set up of the EL and IRT cameras for the indoor tests at the Photovoltaic Solar Energy Unit of the CIEMAT, in Madrid. Figure 5: Outdoor IRT testing bench at the Photovoltaic Solar Energy Unit of the CIEMAT, in Madrid. - 38 of 137 - Chapter 1: Introduction These images have been taken in the lower current step of the I-V curve at 7.3A- 24 V and in the middle current step of the I-V curve at 7.8 A-15 V at 933 W/m2. Further information about this study methodology can be found in the paper presented in subsection 2.2.5. Within the objectives, once the different techniques have been used and analyzed, it is proposed to give a more thorough diagnosis of the causes of failure and their severity. For this, a broad database based on experimental thermographic measurements at field level on modules with hot spots has been developed, in order to identify the different failure modes that the modules can develop and the patterns of each of the modes of failure. This research has been performed in a 3 MW capacity PV site (17,142 modules), located in Spain, in Castilla y León region, in collaboration with Solarig. The thermographic inspection has been performed using the traditional manual IRT method, walking all around the PV site inspecting each module with the thermographic camera. The manual camera used was a Testo 870-2. The manual camera used captures visual RGB images simultaneously to thermographic images, allowing certifying the detected failures during the post-processing steps and avoiding false positives. However, the presence of false positives is less significant in case of manual inspection, as specialists performing inspections on site can check the presence of shadows or dirt in modules during the inspection. Every single failure detected during the inspection regardless of its temperature, was registered, identified and reported. The time needed to complete this inspection has been 34 working days; and to post-process and to analyze the results, 26 working days. More information about this analysis and characterization of PV module defects by thermographic inspection is presented in subsection 2.2.6. The experimental tests for the analysis of the image resolution influence in aerial thermographic inspections of PV plants have been performed in the same 3 MW PV site than the previous analysis. Three different tests have been performed on the entire site to study the influence of the resolution on aerial thermographic inspections. − Traditional manual IRT method, walking all around the PV site inspecting each module with the thermographic camera (it has been used the information from the inspection detailed above for the thermographic failures characterization and classification). − The second and third thermographic inspections were carried out on the 4th of August of 2017 using aerial thermographic inspection. This technique consists of a UAV transporting a thermographic camera specially designed to - 39 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis be integrated into aerial vehicles. These aerial inspections were performed at 30 m and 80 m from the ground level respectively. These heights have been selected with the objective of obtaining results that clearly show the impact of Spatial Resolution. With this aim, and considering that the available manual images already have a high resolution, it has been selected 30 m of elevation that provides a pixel size of 3.9 cm x 3.9 cm, which is within the acceptable range for this application (considering an acceptable pixel of 3 cm to 5 cm) and 80 m, that results in a pixel of 10.4 cm x 10.4 cm, out of this acceptable range. Both inspections were performed in 3 hours, considering all the time needed on-site to prepare and mount the equipment, prepare the inspection based on the previously prepared flight plan, taking off and landing, changing the batteries and downloading the images. The UAV used to deliver the FLIR TAU 2 thermographic sensor and the black edition go pro +3 RGB camera during the inspections is presented in Figure 6. It is a hexacopter model DJI S900, prepared to fly with the RGB and the thermographic cameras at the same time, in addition to other telemetry sensors. A hexacopter was chosen since they are more powerful than quadcopters, allowing the transportation of heavier payloads and being safer. Figure 6: Equipment used in the aerial inspections to the 3 MW PV site, composed by a thermographic camera FLIR TAU 2 640 on board of a hexacopter DJI S900. At the back the STREAM monocrystalline modules with two portrait distribution. Additionally, a fourth test was accomplished to six faulty modules, which were selected in detail from the results obtained in the previous tests, in order to clarify the results obtained. It consists on capturing manual images of six different failures at twelve different distances with the Testo 870-2 thermographic camera. More than 250 images were taken in this fourth test in order to study the distribution of temperatures within a PV cell hotspot and to compare the temperature obtained for different hotspots the same - 40 of 137 - Chapter 1: Introduction day, with the same environmental conditions at twelve different distances, revealing how the resolution affects the results. Further information about this study methodology can be found in the paper presented in subsection 2.3.4. The low-cost IRT application has been studied in cooperation with the Department of Electrical Engineering of the Faculty of Engineering of the University of Cuenca, in Ecuador, in which thermographic analysis with different cameras and I-V curves have been captured. University of Cuenca’s solar farm has 35 kWp installed, with a total of 160 modules, and it has been under production since 2016. After manual inspection of all these modules, only two were found with hot spots. These anomalies have been further studied with a Flir One Pro, a Cat S60 and a Flir TG167 cameras, obtaining temperature tables, image reconstruction intensity graphs, X-Y dispersion of the thermal images, three-dimensional mesh, contours of the three-dimensional mesh of the thermal image and standard grayscale static images. Results have been validated using a Solmetric PVA-600 IV tracer [74] and DS18b20 temperature sensors (Figure 7). (a) (b) Figure 7: Temperature sensor matrix validation (a) Setup (b) Sensors deployed. Additionally, the Cat S60 camera has been mounted onboard a DJI Mavic Pro drone and used as proof of concept in order to verify if this camera can show the hot spot on the faulty panel (Figure 8). Further information about this study methodology can be found in the paper presented in subsection 2.3.3. (a) (b) Figure 8: Solar panel IRT with drone (a) Cat S60 thermal camera and DJI Mavic Pro drone (b) captured image. - 41 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Figure 16: Module Windon 265W Mono-crystalline simulated in Ltspice IV with the real single-diode model cells presented in Figure 17. Figure 17: Real single-diode model of a solar cell simulated in Ltspice IV. − Matlab 9.3 [78]: this software has been used for merging EL and thermographic images (subsection 2.2.5). In general terms, image fusion is the process of combining different information from several images (sensors) taken from the same scene to obtain a composite image, which will contain the best information coming from the original images [79]. In that sense, it is expected that the fused image will have better quality (or more information) than any of the original images. In this work, a pixel-level fusion or image signal level is adopted, which - 48 of 137 - Chapter 1: Introduction is the lowest-level type of fusion and involves the combination of raw source images into a single composite image. In order to be able to fuse images at this level, it is essential that the images are adequately aligned prior to being fused. This process is known as image registration, which is the process of overlaying two or more images of the same scene taken at different times, from different viewpoints, and/or by different sensors. In general terms, it geometrically aligns two images, which are usually referred to as the reference image and the sensed image [80]. Image registration helps to correct problems such as image rotation, scale and inclination, which are common when overlaying images. The initial available images in their original format are presented in Figure 18. Note that in EL image, the panel is placed vertically and in the thermographic image horizontally, in addition to being in different formats, EL and thermal image. Hence, image registration of these images is needed. Firstly, for the image alignment process to be performed in a better and automatic way, cropping followed by a rotation of both images was performed manually. Once this is performed, it was observed that the two images to be merged are of different sizes. Therefore, the images were resized to have the same size, in this case the thermal image is resized to the size of the EL image. Also, both images were transformed to grey scale. The resized and transformed images are presented side by side in Figure 19. The registration process was then performed using the Matlab function imregister().Thereafter, nine different fusing image algorithms were performed, as diff, blend, fpde, ifm, gff. Further information about this study methodology can be found in the paper presented in subsection 2.2.4. (a) (b) (c) Figure 18: (a) RGB of the module under analysis (SE-2). Initial available (b) EL and (c) thermographic images in their original format for the fusion. - 49 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Figure 19: EL and thermographic images resized and in grey scale, placed side by side. − Flir Tools [81], Workswell CorePlayer [82] and IRSoft [83]: Cell defects detected in thermographic inspections have been analyzed using the thermographic cameras software. It has been used to import, to edit, to process and to analyze thermographic images, obtaining the relative temperature of the defect, the mean temperature of the healthy area and the difference between them, which indicates the overheat of the fault. Flir Tools has been used for images captured with Flir cameras (FLIR SC 640 camera in subsection 2.2.5, Flir C2 in subsections 2.2.2, 2.2.3 and 2.2.4 and TAU 2 in subsection 2.3.4), CorePlayer for Workswel cameras (Wiris Pro in subsections 2.2.2, 2.2.3 and 2.2.4) and IRSoft for Testo cameras (Testo 870-2 in subsections 2.2.6 and 2.3.4). − Microsoft Office 365 [84]: It has been the program used for most of the storage, analysis, organization and exposure of data, including the use of Excel, Word, OneDrive and PowerPoint. 1.5.5. Results dissemination phase It is also necessary to foster active dissemination of the results of research. It has been done once finished the empirical data acquisition (third phase) and its analysis and comparison for extracting results and drawing conclusions (fourth phase) for each - 50 of 137 - Chapter 1: Introduction research performed throughout the entire duration of the thesis. This phase also includes the writing of the doctoral thesis document. Below is a list of the conferences in which the doctoral student has participated and the scientific journals in which the results obtained have been published. More information about papers published is detailed in the Prologue and in Chapter 2 (Published papers). • Conferences: o EUROSUN 2016: International Conference on Solar Energy for Buildings and Industry (11th Oct – 14th Oct 2016). o ISES SWC 2017: Solar World Congress & IEA Solar Heating and Cooling Conference (29th Oct – 02nd Nov 2017). o ICSC-CITIES 2018: Congreso Iberoamericano de Ciudades Inteligentes (26th Oct – 27th Oct 2018). o PVSEC 2019: 36th European Photovoltaic Solar Energy Conference and Exhibition (09th Oct – 13th Oct 2019). o ICSC-CITIES 2019: Congreso Iberoamericano de Ciudades Inteligentes (07th Oct – 09th Oct 2019). • Scientific journals and books: o Energy, 2019 (Q1). o Solar Energy, 2019 (Q1). o Solar Energy, 2018 (Q1). o Revista Facultad de Ingeniería Universidad de Antioquía, 2019 (Q3). o Energy, 2020 (Q1). o Renewable and Sustainable Energy Reviews, 2018 (Q1). o Revista Facultad de Ingeniería Universidad de Antioquía, 2020 (Q3). o IEEE Transactions on Industrial Informatics, 2018 (Q1). 1.6. Results and discussion This section presents a summary of the most relevant empirical findings of the doctoral thesis and a brief explanation and interpretation of the results. It has been structured in three subsections, coinciding with the three partial objectives of the doctoral thesis. - 51 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 1.6.1. Study of the state of the art (Obj. 1) Like any other technology, PV production is never faultless. It is essential for main players involved in PV plants, as investors, operators and equipment manufacturers, to identify the failure modes and rates that the main equipment experiences to reduce investment risk, to focus their maintenance efforts on preventing those failures and to improve longevity and performance of PV plants. It can be considered as a relatively new technology that has evolved a lot in recent years and there are many manufacturers with very variable qualities. Moreover, from the point of view of the installation, O&M, due to the inherent characteristics of this technology, PV fields are highly atomized, so there is also great diversity in terms of procedures. Besides, PV systems consist of many different vulnerable components and their operation is highly dependent on temperature, power losses, and ambient environments. In addition, the input power is variable and uncontrollable what may cause electrical stress in the panels [15]. The greater challenge that researchers address and indicate while investigating about PV system failures is the lack of accessible reliable real quantitative data, since most operators are private and do not disclose the figures or either they do not have enough capabilities to record this data [85]. Different authors have contributed to the advance of this research with their accessible information [28,86–92]. Through the implementation of advanced PV plant monitoring systems [93], the application of Computerized Maintenance Management Systems (CMMS) and the feedback of the field technicians it is possible to have an experimental database to assess the failure rate analysis of the PV plants [73]. This fact is one of the greater strengths of the research performed, in which the information from the historical data of the sixty-three PV plants portfolio in Italy and Spain has been accessible. Important results have been obtained in this research, which are summarized in this subsection and are further analyzed in the paper presented in subsection 2.1.1. The absolute number of failures per year in the portfolio analyzed is 19,325. Firstly, an analysis of the average number of failures per PV site has been performed, considering defects in all elements, showing that although PV plants are getting older each year, proper maintenance can be responsible for keeping the defects ratio or even reducing it. Additionally, properly performed predictive, preventive and corrective maintenance include improvements in PV plants, which reduces the failure ratio in the long term. An important conclusion from the five years data analyzed in this study is the amount of alarms or failures registered in a PV plant per year, resulting in an - 52 of 137 - Chapter 1: Introduction average ratio of 321 failures/year/plant. Lately, a similar study has been performed, calculating the number of failures per kW instead of per PV site, revealing an average ratio of 0.23 failures/year/kW. After that, the plants have been divided in five groups considering their power. Logically, bigger plants present a higher number of defects per plant, as they have more components and it is just a probabilistic fact, presenting a deviation of -72.6%, +191% from the average previously calculated if the sites capacity is not considered. Therefore, the extension of this study per power unit instead of per site has been done, with the objective of obtaining more accurate failure ratios. It has been seen how there are defects that do not depend on the plant size or the plant number of elements or components; therefore, the failure number per power unit is higher in smaller plants (Figure 20). For instance, regarding the grid, there is usually just a connection point in the plant, regardless of its size and it can generate failures as a consequence of overvoltage, overcurrent or disconnections among other causes. Other example of this fact are the alarms due to failures in the wired ethernet, wireless Wi-Fi or fiber optic of the communication system, which can be independent of the size of the plants in many cases. Three different important annual ratios are obtained; 0.45 failures/year/kW for plants with a capacity under 750 kW, 0.18 failures/year/kW for plants with a capacity beyond 750 kW and 0.23 failures/year/kW without any plant capacity consideration. All these alarms are distributed among the different elements that compose the PV sites as presented in Figure 21. The existence of failures in different components can lead to the detection of failures in the monitoring or communication system, which explains the large amount of failures attributed to them, adding more than 50% of the alarms. For instance, a failure in the auxiliary services transformer can be responsible of an interruption of communications, as it feeds the communication system. In this case, the presence of a failure in the auxiliary services transformer would also generate a communications alarm. Further information about the results of this research can be found in the paper presented in subsection 2.1.1. However, having a higher failure rate does not mean they are the most relevant alarms of the plant, as shown in Figure 22. In a FMECA, the results present the failure modes and the severity of the consequences with relatively high probability, obtaining the RPN that multiplies severity, detection and occurrence based on a ranking criteria. Further information about the results of this research can be found in the paper presented in subsection 2.1.2. - 53 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Figure 20: Average number of failures per kW for the different defined groups (from PV1 to PV5) from 2012 to 2016. Figure 21: Distribution of the registered failures in each of the PV plants monitored elements, considering data from 2012 to 2016. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 2012 2013 2014 2015 2016 Average number of failures per kW (failures/kW) Year PV1 ( p ≤ 750 kW ) PV2 ( 750 < p ≤ 995 kW ) PV3 ( 995 < p ≤ 1,300 kW ) PV4 ( 1,300 < p ≤ 2,300 kW ) PV5 ( 2,300 < p ≤ 10,000 kW) 8.5% 12.5% 3.7% 0.9% 26.3% 30.6% 2.1% 0.7% 11.6% 3.1% Photovoltaic generator Inverter MV Transformer station Metering elements Communication system Monitoring system Security system Civil works Grid Others - 54 of 137 - Chapter 1: Introduction Figure 22: Sum of the RPN for each of the groups of elements under study and the accumulated RPN sum Accordingly, each defect involves each own repairing time to fix a defect, and excluding the elements that usually require the presence of the manufacturer or outsourcing for their reparation and have a higher reparation time, the PV generator is the element with the higher fixing time value, with 8.15 hours/failure. Additionally, as it has been introduced, in utility scale PV plants, typically the monitoring system trails the string series current of multiple modules connected, or even the inverter current, which is the sum of the parallel string current, and does not track the performance of the individual PV modules, which is the key element on a PV system as it converts the incident irradiance into electric power and their operation will affect the overall plant performance. Therefore, PV module alarms are not frequently automatically generated being specially complicated the detection of failures at this level and even more difficult the recognition of the failure cause and mode. PV module’s cost is the highest cost of PV installations, having reached its price up to 50% of the price of the installation in the last decade, according to reference [71] of 2012. All these facts support the need to investigate this type of defects. Further information about PV systems, including design, O&M, module failures and inspection techniques can be found in the paper presented in subsection 2.1.3. As it has been seen, the module is the key element on a PV system, but, even so, it has limited access to local monitoring alarms, which are not frequently automatically 0% 20% 40% 60% 80% 100% 120% 0 100 200 300 400 500 600 700 800 Sum of RPN Accumulated Sum of RPN - 55 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis generated. From the study of the state of the art, it is highlighted the importance of the research on PV modules failures detection, classification and characterization in order to produce reliable, efficient and safety energy. This justifies the research on this investigation line, summarized in the following subsection 1.6.2., and with the complete related published papers in section 2.2. The use of the different inspection techniques has been reviewed, emphasizing aerial IRT as a novel and promising inspection technique in this field, so it has been further studied, presenting the results obtained in subsection 1.6.3 and the related published papers in section 2.3. 1.6.2. Detection, characterization and classification of defects in PV modules through the use of different inspection techniques (Obj. 2) PV modules experience thermo-mechanical loads during manufacturing and subsequent life stages leading to the origination of the appearance of different defects [94], which are responsible for the reduction of their efficiency, as well as their durability and reliability. These defects can result in power loss and module degradation [65]. Consequently, defect finding in PV modules has had substantial attention of researchers during the last years. This subsection summarizes the results of the doctoral thesis on this regard included in section 2.2, and it has been structured in three parts. In the first, it is presented the detection of PV defects using different techniques and the comparison of the information obtained with each of them, the second includes the characterization of different PV defects and the third, the classification of PV defects attending to failure patterns. 1.6.2.1. Detection of PV defects As it has been seen, different techniques can be used to detect and quantify PV modules anomalies, as visual inspections, electrical tests like the I-V curve test, IRT or EL. PV plants operators usually apply only one or two of them within the O&M activities. Additionally, researchers usually studied them separately. However, these methods provide complementary results, glimpsing interesting information about the PV site state. Throughout this doctoral research, all these inspection techniques have been simultaneously applied and analyzed, studying the correlation between them. Regarding the I-V curves, several electrical circuits that model a PV system have been simulated using Ltspice and validated with experimental measurements carried out - 56 of 137 - Chapter 1: Introduction in the laboratory of the University of Gävle (Sweden). The results of this research are summarized in the paper included in subsection 2.2.1. The simulations show remarkable agreement with the experimental data, which means that the designed model supposes a very useful tool that can be used to study the performance of shaded or defective PV systems, allowing the simulation of numerous faults. The novelty introduced in this research is not related to the electrical model of the PV cell, as it has been used a wellknown model to do the simulations, the real single-diode model. This research aims to perform a detailed analysis on how shadowing and other defects, as cell breaks that cause isolation of part of the cell, affect the PV production of the modules in different situations, specifying the way that PV modules work when they are affected by these faults. In relation with EL tests, operators agree that there is a generalized concern about how the current injection needed in EL measurements can affect the PV modules service life, and how these periodical inspections can affect the long term life of the modules. In order to give a practical answer to this problem, a series of tests consisting of long periods of current injection on several monocrystalline silicon modules has been carried out, and it is included in the paper presented in subsection 2.2.2. The modules tested had already fulfilled their useful life and present multiple defects. In order to analyze how the current injection affects the state of the module, images of IRT and EL were acquired during the current injection period (during 96 hours). Figure 23 shows the EL images of the module taken before and after 96 h of Isc current injection. (a) (b) Figure 23: EL images taken before (a) and after 96 h of Isc current injection of a Tynsolar 175W monocrystalline module (b). The mean values of the EL and IRT intensity over the full images were calculated, Figure 24. For a better understanding of the results and to analyze how EL could affect the modules efficiency, the current injection was switched off after 24 h to check the module behavior in a second EL test after completing a first injection cycle. In this way, the module was cooled during 24 h, after which the current injection was reestablished for a total time of 96 h. - 57 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis (a) Cell A2 (b) Cell A5 (c) Cell A7 Figure 30: Indoor IRT images of the low efficiency (a and b) and the short-circuited I defective cells of the first string (string AB) during EL tests, injecting a current of 8.9A. Images taken with FLIR SC 640 IR camera. Their maximum and minimum temperatures in these conditions are: a) Tmax=38.8°C and Tmin=29.1°C. b) Tmax=40.6°C and Tmin=30.8°C and c) Tmax=33.0°C and Tmin=27.6°C. The second characterized failure has been the soldering defects. The soldering of the wires to the solar cells results in generating significant thermomechanical stresses in different layers due to the diversity of thermal expansion of different components, which can lead to small cracks in the solar cells [98,99]. Soldering defective cells present smaller parameter deviations than the rest of the defects studied. Additionally, it has been seen that there are more losses when there is a local soldering defect in just a part of the bus, as the series resistance is considerably increased. Conversely, having all (or some) tabs loose does not interrupt the electrical contact despite not being welded, and its influence is not disastrous. However, this could be altered after some years of operation of the module in which the encapsulation is exposed to thermal cycles. Soldering defects studied are not visible in EL. It could be because the faults are made on the back side of the module. An additional analysis has been performed to check if these defects are visible with IRT performed from the back side of the module in short circuit current injection conditions. The IRT image of the back side of the whole module can be seen in a). In these conditions, the only defect which certainly experiments an overheating is cell C8, corresponding the heating area with the centimeter of the three buses welded. The greatest danger of bad welds is usually the electric arcs, since that bad contact causes the tab to warm up and it can come out to disconnect it. At that moment there is a relevant tension difference between two points very close and micro arcs can be generated. That ends up burning the back sheet behind, or if they are generated in broken cells, the edges of the cell are usually burned and also depending on the - 64 of 137 - Chapter 1: Introduction composition of the paste together with the permeability of the back sheet, the humidity causes the fingers to blacken, being easily identifiable the cell breakages, which is commonly known as snail trails. (a) (b) Figure 31: Indoor IRT images of the back side of the module (a) and of the C8 cell (b) during EL tests, injecting current of 8.9A. Images taken with FLIR SC 640 IR camera. The maximum and minimum temperatures in C8 cell in these conditions are: Tmax=38.2°C and Tmin=29.6°C. Concerning broken cells, it has been analyzed how the shape of the I-V curves of the affected cells is very similar to the one of the nominal equivalent cell but decreasing the current proportionally, as in the case of shading [100,101]. The EL images of defective cells, complemented in this case by their corresponding RGB images that allow seeing the breaking patterns, are shown in Figure 32. It has been seen that if a piece of one cell is placed on another cell it generates a parallel electrical path which is responsible of a small Rsh in the cell, turning the I-V curve into a straight line. (a) Cell E2 (b) Cell E3 (c) Cell E6 (d) Cell E10 (e) Cell F1 - 65 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis (f) Cell F6 (g) Cell F8 (h) Cell F9 (i) Cell F10 Figure 32: EL and RGB images of the broken defective cells of the third string (string EF) with an injected current of 8.9A. EL images taken with PCO 1300 camera, with a focal distance of 4 cm and exposure time 40s. Failures like cell cracks, solder corrosion, and broken cell interconnects have no limits in power loss and the PV module may become unusable [33]. Two distinct types of dark areas are observable in the EL images (Figure 32), irregularly shaped regions and regular rectangular shaped areas. The irregular areas indicate the presence of cracks in the silicon wafers. However, the presence of regular rectangular dark areas can be due to broken front grid fingers, which are broken at the busbar [102,103]. Finally, an optical microscope has been used for imaging a crack in cell F9 solar cell, which displayed a dark irregular area in the crack zone in the EL measurement. It proves how a small crack, not visible for the human eye, can isolate a part of the cell Figure 33. Figure 33: Optical micrographs of a crack in F9 solar cell which displayed a dark irregular area in the crack zone in the EL. Images captured with a Nikon SMZ 800 microscope. - 66 of 137 - Chapter 1: Introduction Regarding the three kinds of defects combination or mix within a module, it has been seen how when the cells are series connected, the cell which delivers the lowest current will limit the current of the string. In this module, there is a total of three bypass diodes, one every 20 cells. As it has been seen, each bypass string contains different defects (of manufacturing, soldering and breaking cells) and each of them is able to produce a different current. Therefore, the electrical characteristic curve of the whole module, which is composed by the three bypass circuits in series, will contain three steps, each at the Isc of each bypass circuit, as seen in Figure 34. It has also been concluded that for soldering defects, in which the series resistance has an important influence, having more buses connecting the cells could increase the whole module performance. The effect of a shunt resistance is particularly severe at low light levels, since there will be less light-generated current. There is a clear tendency to increase the shunt resistance on lower irradiance levels in illumination tests [104], as seen in the following Figure 35, in which this effect is extremely evident in cell E10. Figure 34: I-V Curves of the complete module and separate strings at 1,000 W/m2. The secondary axis presents the P-V curves of the module at this irradiance. 0 50 100 150 200 0 1 2 3 4 5 6 7 8 9 010 20 30 40 Power [W] Current [A] Voltage [V] I-V StringAB I-V StringCD I-V StringEF I-V Module P-V Module - 67 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis (a) (b) Figure 35: EL images of the tested module with an injected current of 8.9 A (a) and 0.89 A (b). EL images taken with PCO 1300 camera, with a focal distance of 4 cm and exposure time 50 s (a) and 15 min (b). In the outdoor IRT, the most restrictive cell is detected as a severe hotspot. It can be seen in Figure 36. (a) (b) Figure 36: Outdoor IRT images of the module in the lower current step of the I-V curve at 7.3 A-24 V (a) and in the middle current step of the I-V curve at 7.8 A-15 V (b) at 933 W/m2. Images taken with FLIR SC 640 IR camera. The maximum temperature values of the different hot spots are indicated in the figures. At the low current step (Figure 36 a), the diodes are not operating. However, in the second current step (Figure 36 b), at 7.8 A, the temperature in F1 increases a lot, as string EF has the most restrictive current in the module and throughout the bypass diode of this string has to circulate the current that the string is not able to do. In the shortcircuited cell (cell A7), all the external energy received from the sun is not used to generate electricity. The absorbed photons generate pairs of electrons in the cell - 68 of 137 - Chapter 1: Introduction material, that are recombined by other mechanisms not useful for the electricity generation and therefore, the material becomes hotter. From this study it has been possible to obtain a concluding remark on each inspection method and defect type, to highlight the information provided by each surveying technique considering the defect type, with the aim of not missing any information during the examination. It has been seen that authors and main players involved in PV systems mainly work with the different inspection techniques separately. The remarks presented in this section will determine the validity of the different alternatives and their complementarity. In relation with manufacturing defects, it has been checked that the different inspection techniques show: • I-V characterization: low-efficiency and medium-efficiency defects are clearly evidenced at STC conditions, as well as the short-circuited cells. Their FF is reduced while increased the irradiance levels and therefore their maximum power output. The efficiency defects present high values of Rs, which is shown as an increment in the curve slope at Voc. Severe defects are also revealed with a slope increment at Isc because of their reduced Rsh. Power loss can almost achieve 50% in low efficiency cells, as seen in the research. • EL and indoor IRT characterization: low efficiency cells can be seen as a tire imprint in the EL images and they experiment a local overheating in the emitting area. Medium efficiency cells do not present any special pattern in EL neither indoor IRT, although it has been observed a tendency to decrease the light emission during EL tests with the efficiency drop. Short-circuited cells do not emit any light during EL and its whole area is seen colder in thermography, which has been denominated as cold spot. • Outdoor IRT characterization: low efficiency cells as well as short-circuited cells are revealed as minor hot spots in normal outdoor operation, with a ∆T of around 5°C. Medium efficiency defects are not evidenced in outdoor IRT examination. Therefore, the most complete inspection technique for efficiency defects is the I- V characterization, in which all the manufacturing defects types analyzed are evidenced. In low efficiency cells, EL complements the I-V curves checking the EL defects patters. Regarding the outdoor IRT, it should be very exhaustive in order to be valid, as these defects present low ∆T. In relation with short-circuited cells, EL and indoor IRT would be - 69 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis the most effective inspections, presenting this kind of defect the clear patterns detailed in these conditions. In relation with soldering defects, it has been checked that the different inspection techniques show: • I-V characterization: local soldering defects, in just a part of the bus, considerably increase the Rs, presenting an I-V curve affected near the Voc, incrementing the curve slope. Additionally, having one, two or all the three tabs not welded reduces the Voc. • EL and indoor IRT characterization: soldering defects analyzed are not visible in EL. Only local soldering defects experiment an overheating during current injection. Therefore, defects with one or more buses not welded in the backside of the cell are apparently neither detectable using indoor IRT. • Outdoor IRT characterization: soldering defects analyzed are not detectable in outdoor IRT. It has been seen that soldering defects performed in the backside of the module present small parameter deviations than the rest of failures studied, being the I-V characterization the most effective to detect them, which can be complemented with indoor IRT in case of local soldering defects. Finally, in relation with broken cells defects, it has been checked that the different inspection techniques show: • I-V characterization: the shape of the I-V curve of the affected cell is very similar to the nominal equivalent cell but decreasing the current proportionally to the disconnected area. The Voc is also slightly decreased. Parallel electrical paths distort the I-V curve, turning it into a straight line due to small Rsh. • EL, visual and indoor IRT characterization: EL clearly evidences disconnected areas and breaks with dark regions. Irregular areas indicate the presence of cracks in the silicon wafers while regular rectangular dark areas can be due to broken front grid fingers. Not the totality of dark regions in EL corresponds with disconnected cell areas, as the case in which a piece of other cell is over placed in the front part of another cell. Broken cells defects analyzed have not been detected in indoor IRT. • Outdoor IRT characterization: Disconnected areas restring the current through the cell, and this can generate a hot spot, which can reach high temperatures. - 70 of 137 - Chapter 1: Introduction The hotspot temperature will depend on the disconnected area. In the broken cell analyzed (F1), the ∆T is 23.6°C in normal operation. Therefore, in the case of breaks in cells, the most interesting information is probably provided by the EL tests, clearly showing the breaks and disconnected areas, while outdoor IRT can complement EL locating the most restrictive cell within a string. It has been seen in the hot spot examined that it present high ∆T, so regarding the outdoor IRT, it should not be such exhaustive as in other defects analyzed in order to be valid. 1.6.2.3. Classification of PV defects Once the detection and characterization of different defects have been studied, a classification of thermographic defects has been performed to complete objective 2. An onsite study of 17,142 monocrystalline modules has been carried out with the aim of detecting every single existing defect, classifying them in different groups, studying the variance of the same kind of defect in different modules and the patterns of each group of thermographic defects. This analysis can be reviewed in the paper presented in subsection 2.2.6. In this case, the thermographic analysis for the identification of defects in this research is performed manually, as the spatial resolution of the thermographic images is higher than using UAVs. However, this study is complemented with the results of the aerial thermographic inspections explained in subsection 1.6.3. A 3 MW PV plant, commissioned in 2008, has been chosen with the aim of gathering larger amount of data and different cases in comparison with the information which would be available in a small installation in the rooftop of a building. However, the results are perfectly extended for their subsequent application in aerial thermographic inspections and in small scale installations in roofs or façades of buildings, as same defects have to be identified in all PV inspections. From the 17,142 modules thermographically inspected, the number of detected modules with some failure has been 1,140, which corresponds to 6.65%. According to some recent research, it is predicted that 2% of the PV modules do not meet the manufacturer’s warranty after 11-12 years of operation [105]. The percentage of failures detected is over this rate because every single anomaly has been reported in this study, independently of the temperature difference between the overheated area and the healthy part. However, according to [64], due to normal tolerances in cell sorting and module production, thermal abnormalities of less than 10 % of the recorded modules do not indicate a special quality issue regarding the used modules. - 71 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Attending the results obtained, all these faults detected have been classified in five different thermographic defects modes: hotspot in a cell or in a group of cells, bypass circuit overheated, hotspot in the junction box, hotspot in the connection of the busbar to the junction box and whole module overheated. The distribution of defects among these five groups can be observed in Table 4. As it can be seen, more than three quarters of the affected modules correspond to cell hotspots, presenting one or more cells overheated. This prevailing number of hotspot failures in cells with respect to the rest of failures types is not an isolated case. Statistically, there are a greater amount of cells, 72 cells per module in this case, than of the rest of components, three bypass circuits, one junction box, four bus ribbons and one module. Additionally, there are a large number of causes responsible for the occurrence of cell hotspots, as cell cracks, snail trails, potential induced degradation (PID) or delamination [105]. Although some of the defects are slightly visible to the human eye, as snail trails, most of them are undetectable without the use of a thermographic camera [64]. Additionally, other defects within the PV module can be the result of a cell failure. For instance, a diode fault can be caused by a hotspot in a string of cells that forces the continuous operation of the diode, overheating it. Therefore, a correct characterization of this kind of failures is essential for a proper automatic detection of cell hotspot supported by software, as it represents more than three quarters of the defects. Table 5: Thermographic defects detected classified by the module affected component Affected component Number of defects detected Percentage Hotspot 859 75.35% Bypass circuit 123 10.79% Junction box 79 6.93% Connection 78 6.84% Module 1 0.09% Commonly, the severity of defects is given, based on the difference of temperature between overheated and healthy areas, ΔT, within a module. The absolute temperature of the defect is not used to determine the severity as it is strongly dependent on different weather conditions, as the ambient temperature, the irradiance or the wind speed. This ΔT varies significantly between different kinds of defects, as it can be observed in Figure 37, which presents a box chart showing the difference of temperature for each of the 1,140 defective modules detected, grouped by defect type. - 72 of 137 - Chapter 1: Introduction Figure 37: Box chart showing the delta temperature between overheated and healthy areas, ΔT, grouped by defect type for each of the 1,140 defective modules detected Three different common ranges of temperature for all defects types have been defined attending the severity of the anomalies for manual thermal inspection proposed in [61]. Differences higher than 30°C are considered as severe failures, from 10°C to 30°C are considered as medium severity failures, and lower than 10°C as minor failures. According to [64], temperature gradients lower than 10°C are normally considered as unproblematic and do not have to be listed separately. Temperature gradients from 10°C to 20°C are unproblematic in the current stage [64], although they have to be reported and observed in subsequent regular thermal inspections as they can increase during the operation of the PV power plant. Differences in temperature above 20°C are expected to cause degradations of panel output and eventually, the material compound may even degrade, resulting in a safety issue [64]. Table 5 summarizes the severity of all defects detected, grouped by defect type or affected component. Table 6: All defects detected classified by the affected component and in three temperature groups Affected component ΔT≥30°C 10≤ΔT<30°C ΔT<10°C Total Hotspot 89 10.36% 262 30.50% 508 59.14% 859 Bypass circuit 0 0% 4 3.25% 119 96.75% 123 Junction box 0 0% 4 5.06% 75 94.94% 79 Connection 2 2.56% 32 41.03% 44 56.41% 78 Module 0 0% 1 100% 0 0% 1 - 73 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis It has been proved that the same failure will be registered as a different temperature depending on the distance from the thermographic camera to the faulty cell. Therefore, it is necessary to determine the severity range criteria to classify the failures detected as a function of the flight height, as it cannot be the same as the criteria used previously in the manual inspections. For the inspections performed, this criteria should change from the manual (1) to the aerial inspections at 30 m (2) and 80 m (3) as follows: severe failures, temperature difference higher than 30°C (1), 17.8°C (2) and 9.0°C (3); medium severity failures, from 10°C to 30°C (1), from 4.4°C to 17.8°C (1), from 0°C to 9°C (1) ; and minor failures, lower than 10°C (1), lower than 4.4°C (2) and not detectable (3). 1.7. References [1] REN21. REN21 - 2019 Global Status Report. 2019. [2] Sampaio PGV, González MOA. Photovoltaic solar energy: Conceptual framework. Renew Sustain Energy Rev 2017;74:590–601. https://doi.org/10.1016/j.rser.2017.02.081. [3] Herold D, Horstmann V, Neskakis A, Plettner-Marliani J, Piernavieja G, Calero R. Small scale photovoltaic desalination for rural water supply - demonstration plant in Gran Canaria. Renew Energy 1998;14:293–8. https://doi.org/10.1016/S0960- 1481(98)00080-9. [4] Bakos GC. Distributed power generation: A case study of small scale PV power plant in Greece. Appl Energy 2009;86:1757–66. https://doi.org/10.1016/J.APENERGY.2008.12.021. [5] Bouzerdoum M, Mellit A, Massi Pavan A. A hybrid model (SARIMA–SVM) for short-term power forecasting of a small-scale grid-connected photovoltaic plant. Sol Energy 2013;98:226–35. https://doi.org/10.1016/J.SOLENER.2013.10.002. [6] Araneo R, Lammens S, Grossi M, Bertone S. EMC Issues in High-Power Grid- Connected Photovoltaic Plants. IEEE Trans Electromagn Compat 2009;51:639– 48. https://doi.org/10.1109/TEMC.2009.2026055. [7] Moradi-Shahrbabak Z, Tabesh A, Yousefi GZ. Economical Design of Utility-Scale Photovoltaic Power Plants With Optimum Availability. IEEE Trans Ind Electron 2014;61:3399–406. https://doi.org/10.1109/TIE.2013.2278525. - 80 of 137 - Chapter 1: Introduction [8] Institution of Engineering and Technology. R, Mithulananthan N, Lee KY, Bansal RC. Wide-area measurement signal-based stabiliser for large-scale photovoltaic plants with high variability and uncertainty. vol. 7. Institution of Engineering and Technology; 2007. [9] Cabrera-Tobar A, Bullich-Massagué E, Aragüés-Peñalba M, Gomis-Bellmunt O. Topologies for large scale photovoltaic power plants. Renew Sustain Energy Rev 2016;59:309–19. https://doi.org/10.1016/J.RSER.2015.12.362. [10] Akinyele DO, Rayudu RK, Nair NKC. Development of photovoltaic power plant for remote residential applications: The socio-technical and economic perspectives. Appl Energy 2015;155:131–49. https://doi.org/10.1016/J.APENERGY.2015.05.091. [11] Oliver M, Jackson T. Energy and economic evaluation of building-integrated photovoltaics. Energy 2001;26:431–9. https://doi.org/10.1016/S0360- 5442(01)00009-3. [12] Simola A, Kosonen A, Ahonen T, Ahola J, Korhonen M, Hannula T. Optimal dimensioning of a solar PV plant with measured electrical load curves in Finland. Sol Energy 2018;170:113–23. https://doi.org/10.1016/J.SOLENER.2018.05.058. [13] Hernández-Callejo L, Gallardo-Saavedra S, Alonso-Gómez V. A review of photovoltaic systems: Design, operation and maintenance. Sol Energy 2019;188:426–40. https://doi.org/10.1016/j.solener.2019.06.017. [14] Braun M, Stetz T, Bründlinger R, Mayr C, Ogimoto K, Hatta H, et al. Is the distribution grid ready to accept large-scale photovoltaic deployment? State of the art, progress, and future prospects. Prog Photovoltaics Res Appl 2012;20:681– 97. https://doi.org/10.1002/pip.1204. [15] Zhang P, Li W, Li S, Wang Y, Xiao W. Reliability assessment of photovoltaic power systems: Review of current status and future perspectives. Appl Energy 2013;104:822–33. https://doi.org/10.1016/j.apenergy.2012.12.010. [16] Harder E, Gibson JM. The costs and benefits of large-scale solar photovoltaic power production in Abu Dhabi, United Arab Emirates. Renew Energy 2011;36:789–96. https://doi.org/10.1016/J.RENENE.2010.08.006. [17] de La Tour A, Glachant M, Ménière Y. Predicting the costs of photovoltaic solar modules in 2020 using experience curve models. Energy 2013;62:341–8. https://doi.org/10.1016/J.ENERGY.2013.09.037. - 81 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis [18] Raugei M, Frankl P. Life cycle impacts and costs of photovoltaic systems: Current state of the art and future outlooks. Energy 2009;34:392–9. https://doi.org/10.1016/J.ENERGY.2009.01.001. [19] Spertino F, Corona F. Monitoring and checking of performance in photovoltaic plants: A tool for design, installation and maintenance of grid-connected systems. Renew Energy 2013;60:722–32. https://doi.org/10.1016/J.RENENE.2013.06.011. [20] Moore LM, Post HN. Five years of operating experience at a large, utility-scale photovoltaic generating plant. Prog Photovoltaics Res Appl 2008;16:249–59. https://doi.org/10.1002/pip.800. [21] Salmi T, Bouzguenda M, Gastli A, Masmoudi A. MATLAB/Simulink Based Modelling of Solar Photovoltaic Cell. Int J Renew Energy Res 2012;2:213–8. [22] Glunz SW, Preu R, Biro D. Crystalline silicon solar cells. State-of-the-art and future developments. Compr. Renew. Energy, vol. 1, 2012, p. 353–87. https://doi.org/10.1016/B978-0-08-087872-0.00117-7. [23] Green MA. Crystalline and thin-film silicon solar cells: State of the art and future potential. Sol Energy 2003;74:181–92. https://doi.org/10.1016/S0038- 092X(03)00187-7. [24] Schneller EJ, Brooker RP, Shiradkar NS, Rodgers MP, Dhere NG, Davis KO, et al. Manufacturing metrology for c-Si module reliability and durability Part III: Module manufacturing. Renew Sustain Energy Rev 2016;59:992–1016. https://doi.org/10.1016/J.RSER.2015.12.215. [25] Boix PP, Garcia-Belmonte G, Muñecas U, Neophytou M, Waldauf C, Pacios R. Determination of gap defect states in organic bulk heterojunction solar cells from capacitance measurements. Appl Phys Lett 2009;95:233302. https://doi.org/10.1063/1.3270105. [26] Chen S, Gong XG, Walsh A, Wei SH. Defect physics of the kesterite thin-film solar cell absorber Cu 2 ZnSnS4. Appl Phys Lett 2010;96:021902. https://doi.org/10.1063/1.3275796. [27] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Pérez O. Quantitative failure rates and modes analysis in photovoltaic plants. Energy 2019;183:825–36. https://doi.org/10.1016/j.energy.2019.06.185. [28] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Perez O. Analysis and - 82 of 137 - Chapter 1: Introduction characterization of PV module defects by thermographic inspection. Rev Fac Ing Univ Antioquia 2019:92–104. https://doi.org/10.17533/udea.redin.20190517. [29] Soultanis NL, Papathanasiou SA, Hatziargyriou ND. A Stability Algorithm for the Dynamic Analysis of Inverter Dominated Unbalanced LV Microgrids. IEEE Trans Power Syst 2007;22:294–304. https://doi.org/10.1109/TPWRS.2006.887961. [30] Ji D, Zhang C, Lv M, Ma Y, Guan N. Photovoltaic Array Fault Detection by Automatic Reconfiguration. Energies 2017;10:699. https://doi.org/10.3390/en10050699. [31] Silvestre S, Silva MAD, Chouder A, Guasch D, Karatepe E. New procedure for fault detection in grid connected PV systems based on the evaluation of current and voltage indicators. Energy Convers Manag 2014;86:241–9. https://doi.org/10.1016/j.enconman.2014.05.008. [32] Chine W, Mellit A, Pavan AM, Kalogirou SA. Fault detection method for gridconnected photovoltaic plants. Renew Energy 2014;66:99–110. https://doi.org/10.1016/j.renene.2013.11.073. [33] Köntges M, Kurtz S, Packard C, Jahn U, Berger KA, Kato K, et al. Review on Failures of Photovoltaic Modules, Report IEA-PVPS T13-01:2013. 2013. [34] Manganiello P, Balato M, Vitelli M. A Survey on Mismatching and Aging of PV Modules : The Closed Loop. IEEE Trans Ind Electron 2015;62:7276–86. [35] Paper C, Energy TA, Co AE. Quantifying the Impact of Hot Spots on the Performance of Photovoltaic Modules by Infrared Thermography and a Simulation Model. 2nd Int. Exergy, Life Cycle Assessment, Sustain. Work. Symp., NISYROS - GREECE: 2011, p. 1–8. https://doi.org/10.13140/2.1.3955.7125. [36] Molenbroek E, Waddington DW, Emery KA. Hot spot susceptibility and testing of PV modules. Conf. Rec. IEEE Photovolt. Spec. Conf., vol. 1, 1992, p. 547–52. https://doi.org/10.1109/pvsc.1991.169273. [37] UNE-EN 61215-1:2017 Terrestrial photovoltaic (PV) modules - Design qualification and type approval, 2017. [38] Dueñas S, Pérez E, Castán H, García H, Bailón L. The role of defects in solar cells: Control and detection defects in solar cells. 2013 Spanish Conf. Electron Devices, Valladolid, Spain: 2013. https://doi.org/10.1109/CDE.2013.6481402. [39] Bühler AJ, Perin Gasparin F, Krenzinger A. Post-processing data of measured I– V curves of photovoltaic devices. Renew Energy 2014;68:602–10. - 83 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis https://doi.org/10.1016/J.RENENE.2014.02.048. [40] Deviations IT. Solar I-V Curves Interpreting I-V 2014. [41] IEC 62446 Grid connected photovoltaic systems -Minimum requirements for system documentation, commissioning tests and inspection, 2019. [42] Quater PB, Grimaccia F, Leva S, Mussetta M, Aghaei M. Light Unmanned Aerial Vehicles (UAVs) for cooperative inspection of PV plants. IEEE J Photovolt 2014;4:1107–13. https://doi.org/10.1109/JPHOTOV.2014.2323714. [43] Grimaccia F, Aghaei M, Mussetta M, Leva S, Quater PB. Planning for PV plant performance monitoring by means of unmanned aerial systems (UAS). Int J Energy Env Eng 2015;6:47–54. https://doi.org/10.1007/s40095-014-0149-6. [44] Denio H. Aerial solar Thermography and condition monitoring of photovoltaic systems. 38th IEEE Photovolt. Spec. Conf., Austin, TX, USA: IEEE; 2012, p. 613– 8. https://doi.org/10.1109/PVSC.2012.6317686. [45] Buerhop C, Weißmann R, Scheuerpflug H, Auer R, Brabec C. Quality Control of PV-Modules in the Field Using a Remote-Controlled Drone with an Infrared Camera. 27th Eur. Photovolt. Sol. Energy Conf. Exhib., Frankfurt, Germany: 2012, p. 3370–3. [46] Aghaei M, Bellezza Quater P, Grimaccia F, Leva S, Ogliari E, Mussetta M. PV Plant Planning and Performance Monitoring by means of Unmanned Aerial Systems (UAS). VIII Congr. Naz. AIGE Reggio Emilia, Reggio Emilia, Italy: 2014, p. 55–60. [47] Aghaei M, Quater PB, Grimaccia F, Leva S, Mussetta M. Unmanned Aerial Vehicles in Photovoltaic Systems Monitoring Applications. 29th Eur. Photovolt. Sol. Energy Conf. Exhib. (EU PVSEC 2014), Amsterdam: 2014, p. 2734–9. https://doi.org/10.1017/CBO9781107415324.004. [48] Kauppinen T, Panouillot P-E, Siikanen S, Athanasakou E, Baltas P, Nikopoulos B. About infrared scanning of photovoltaic solar plant. In: Hsieh, SJ and Zalameda J, editor. THERMOSENSE Therm. INFRARED Appl. XXXVII, vol. 9485, 1000 20TH ST, PO BOX 10, BELLINGHAM, WA 98227-0010 USA: SPIE-INT SOC OPTICAL ENGINEERING; 2015, p. 948517 1-14. https://doi.org/10.1117/12.2180165. [49] Muntwyler U, Schüpbach E, Lanz M. Infrared (IR) Drone for Quick and Cheap PV - 84 of 137 - Chapter 1: Introduction Inspection. 31st Eur. Photovolt. Sol. Energy Conf. Exhib., Hamburg, Germany: WIP; 2015, p. 1804–6. https://doi.org/10.4229/EUPVSEC20152015-5CO.15.6. [50] Aghaei M, Grimaccia F, Gonano CA, Leva S. Innovative Automated Control System for PV Fields Inspection and Remote Control. IEEE Trans Ind Electron 2015;62:7287–96. https://doi.org/10.1109/TIE.2015.2475235. [51] Aghaei M, Gandelli A, Grimaccia F, Leva S, Zich RE. IR real-time analyses for PV system monitoring by digital image processing techniques. 2015 Int. Conf. Eventbased Control. Commun. Signal Process., IEEE; 2015, p. 1–6. https://doi.org/10.1109/EBCCSP.2015.7300708. [52] Aghaei M, Leva S, Grimaccia F. PV power plant inspection by image mosaicing techniques for IR real-time images. 2016 IEEE 43rd Photovolt. Spec. Conf., IEEE; 2016, p. 3100–5. https://doi.org/10.1109/PVSC.2016.7750236. [53] Dotenco S, Dalsass M, Winkler L, Wurzner T, Brabec C, Maier A, et al. Automatic detection and analysis of photovoltaic modules in aerial infrared imagery. 2016 IEEE Winter Conf. Appl. Comput. Vis., Lake Placid, NY, USA: 2016, p. 1–9. https://doi.org/10.1109/WACV.2016.7477658. [54] Aghaei M, Dolara A, Leva S, Grimaccia F. Image resolution and defects detection in PV inspection by unmanned technologies. 2016 IEEE Power Energy Soc. Gen. Meet., IEEE; 2016, p. 1–5. https://doi.org/10.1109/PESGM.2016.7741605. [55] Grimaccia F, Leva S, Dolara A, Aghaei M. Survey on PV Modules’ Common Faults After an O&M Flight Extensive Campaign Over Different Plants in Italy. IEEE J Photovoltaics 2017;7:810–6. https://doi.org/10.1109/JPHOTOV.2017.2674977. [56] Dalsass M, Scheuerpflug H, Fecher FW, Buerhop-Lutz C, Camus C, Brabec CJ. Correlation between the generated string powers of a photovoltaic power plant and module defects detected by aerial thermography. 2017 IEEE 44th Photovolt Spec Conf PVSC 2017 2018:1–6. https://doi.org/10.1109/PVSC.2017.8366737. [57] Tsanakas JA, Vannier G, Plissonnier A, Ha DL, Barruel F. Fault diagnosis and classification of large-scale photovoltaic plants through aerial orthophoto thermal mapping. 31st Eur. Photovolt. Sol. Energy Conf. Exhib. EU PVSEC 2015, Hamburg, Germany: 2015, p. 1783–8. [58] Addabbo P, Angrisano A, Bernardi ML, Gagliarde G, Mennella A, Nisi M, et al. A UAV infrared measurement approach for defect detection in photovoltaic plants. 2017 IEEE Int. Work. Metrol. Aerosp., IEEE; 2017, p. 345–50. - 85 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis https://doi.org/10.1109/MetroAeroSpace.2017.7999594. [59] Abdin Jaffery Z, Kumar Dubey A, Haque A. Scheme for predictive fault diagnosis in photovoltaic modules using thermal imaging. Infrared Phys Technol 2017;83:182–7. https://doi.org/10.1016/j.infrared.2017.04.015. [60] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Pérez O. Analysis and Characterization of Thermographic Defects at the PV Module Level. Commun. Comput. Inf. Sci. 1st Ibero-American Congr. Smart Cities, ICSC-CITIES 2018; Soria; Spain; 26 Sept. 2018 through 27 Sept. 2018, Soria (Spain): Communications in Computer and Information Science; 2019, p. 80–93. https://doi.org/10.1007/978-3-030-12804-3_7. [61] Gallardo-Saavedra S, Hernandez-Callejo L, Duque-Perez O. Image Resolution Influence in Aerial Thermographic Inspections of Photovoltaic Plants. IEEE Trans Ind Informatics 2018;14:5678–86. https://doi.org/10.1109/TII.2018.2865403. [62] Gallardo-Saavedra S, Alfaro-Mejia E, Hernández-Callejo L, Duque-Pérez Ó, Loaiza-Correa H, Franco-Mejia E. Aerial thermographic inspection of photovoltaic plants: analysis and selection of the equipment n.d. https://doi.org/10.18086/swc.2017.20.03. [63] Hallam B, Hamer P, Kim M, Nampalli N, Gorman N, Chen D, et al. Field Inspection of PV Modules: Quantitative Determination of Performance Loss due to Cell Cracks using EL Images. 2017 IEEE 44th Photovolt. Spec. Conf. PVSC 2017, 2017, p. 1858–62. https://doi.org/10.1109/PVSC.2017.8366560. [64] Jahn U, Herz M, Köntges M, Parlevliet D, Paggi M, Tsanakas I, et al. Review on Infrared and Electroluminescence Imaging for PV Field Applications. 2018. [65] Tsanakas JA, Ha L, Buerhop C. Faults and infrared thermographic diagnosis in operating c-Si photovoltaic modules: A review of research and future challenges. Renew Sustain Energy Rev 2016;62:695–709. https://doi.org/10.1016/j.rser.2016.04.079. [66] Youssef A, El-Telbany M, Zekry A. The role of artificial intelligence in photo-voltaic systems design and control: A review. Renew Sustain Energy Rev 2017;78:72–9. https://doi.org/10.1016/j.rser.2017.04.046. [67] Platon R, Martel J, Woodruff N, Chau TY. Online Fault Detection in PV Systems. IEEE Trans Sustain Energy 2015;6:1200–7. - 86 of 137 - Chapter 1: Introduction https://doi.org/10.1109/TSTE.2015.2421447. [68] Dhimish M, Holmes V, Mehrdadi B, Dales M. Comparing Mamdani Sugeno fuzzy logic and RBF ANN network for PV fault detection. Renew Energy 2018;117:257– 74. https://doi.org/10.1016/j.renene.2017.10.066. [69] Mekki H, Mellit A, Salhi H. Artificial neural network-based modelling and fault detection of partial shaded photovoltaic modules. Simul Model Pract Theory 2016;67:1–13. https://doi.org/10.1016/j.simpat.2016.05.005. [70] Chine W, Mellit A, Lughi V, Malek A, Sulligoi G, Massi Pavan A. A novel fault diagnosis technique for photovoltaic systems based on artificial neural networks. Renew Energy 2016;90:501–12. https://doi.org/10.1016/j.renene.2016.01.036. [71] Agroui K. Indoor and Outdoor Characterizations of Photovoltaic Module Based on Mulicrystalline Solar Cells. Energy Procedia 2012;18:857–66. https://doi.org/10.1016/j.egypro.2012.05.100. [72] Power systems reliability subcommittee of the power systems engineering committee of the, Society IIA. IEEE Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems. IEEE Stand Board 1990;Std 493-19:1–382. [73] Colli A. Failure mode and effect analysis for photovoltaic systems. Renew Sustain Energy Rev 2015;50:804–9. https://doi.org/10.1016/j.rser.2015.05.056. [74] Solmetric. Solmetric PV Analyzer I-V Curve Tracer User’s Guide n.d. http://www.solmetric.net/get/Solmetric PV Analyzer Users Guide_1500_en.pdf (accessed November 18, 2019). [75] Suckow S. 2/3-Diode Fit 2014. [76] National Institutes of Health. ImageJ Image Processing and analysis in Java n.d. https://imagej.nih.gov/ij/. [77] Linear Technology - Design Simulation and Device Models n.d. http://www.linear.com/designtools/software/ (accessed March 9, 2018). [78] Matlab Homepage n.d. https://www.mathworks.com/ (accessed August 25, 2019). [79] Pajares G, de la Cruz JM. A wavelet-based image fusion tutorial. Pattern Recognit 2004;37:1855–72. https://doi.org/10.1016/j.patcog.2004.03.010. [80] Zitová B, Flusser J. Image registration methods: A survey. Image Vis Comput 2003;21:977–1000. https://doi.org/10.1016/S0262-8856(03)00137-9. - 87 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis [81] FLIR. FLIR Tools. THERMAL ANALYSIS AND REPORTING (DESKTOP) n.d. https://www.flir.co.uk/products/flir-tools/ (accessed December 26, 2019). [82] Workswell Infrared cameras and systems. Workswell CorePlayer n.d. https://www.workswell-thermal-camera.com/workswellcoreplayer/ (accessed December 4, 2019). [83] Testo. The thermal imager n.d. https://www.testo.com/en-UK/products/thermalimager (accessed January 24, 2020). [84] Microsoft. Office 365 n.d. https://products.office.com/en-us/home (accessed December 19, 2019). [85] Colli A. Extending performance and evaluating risks of PV systems failure using a fault tree and event tree approach: Analysis of the possible application. 38th IEEE Photovoltaics Spec. Conf., 2012, p. 2922–6. https://doi.org/10.1109/PVSC.2012.6318198. [86] Garoudja E, Harrou F, Sun Y, Kara K, Chouder A, Silvestre S. A statistical-based approach for fault detection and diagnosis in a photovoltaic system. 6th Int. Conf. Syst. Control. ICSC 2017, IEEE; 2017, p. 75–80. https://doi.org/10.1109/ICoSC.2017.7958710. [87] Perdue M, Gottschalg R. Energy yields of small grid connected photovoltaic system: effects of component reliability and maintenance. IET Renew Power Gener 2015;9:432–7. https://doi.org/10.1049/iet-rpg.2014.0389. [88] Zini G, Mangeant C, Merten J. Reliability of large-scale grid-connected photovoltaic systems. Renew Energy 2011;36:2334–40. https://doi.org/10.1016/j.renene.2011.01.036. [89] Collins E, Dvorack M, Mahn J, Mundt M, Quintana M. Reliability and availability analysis of a fielded photovoltaic system. IEEE Photovolt. Spec. Conf., IEEE; 2009, p. 002316–21. https://doi.org/10.1109/PVSC.2009.5411343. [90] Stember LH. Reliability Considerations in the Design of Solar Photo-Voltaic Power-Systems. Sol Cells 1981;3:269–85. https://doi.org/10.1016/0379- 6787(81)90008-9. [91] Golnas A. PV system reliability: An operator’s perspective. IEEE J Photovoltaics 2013;3:416–21. https://doi.org/10.1109/JPHOTOV.2012.2215015. [92] Piantoni G, Araneo R. Reliability and maintenance in high-power grid-connected - 88 of 137 - Chapter 1: Introduction photovoltaic systems: A survey of critical issues and failures. 17th IEEE Int. Conf. Environ. Electr. Eng. 2017 1st IEEE Ind. Commer. Power Syst. Eur. EEEIC / I CPS Eur. 2017, IEEE; 2017, p. 1–4. https://doi.org/10.1109/EEEIC.2017.7977870. [93] Cristaldi L, Faifer M, Lazzaroni M, Khalil MMAF, Catelani M, Ciani L. Diagnostic architecture: A procedure based on the analysis of the failure causes applied to photovoltaic plants. Measurement 2015;67:99–107. https://doi.org/10.1016/j.measurement.2015.02.023. [94] Waqar Akram M, Li G, Jin Y, Chen X, Zhu C, Zhao X, et al. Improved outdoor thermography and processing of infrared images for defect detection in PV modules. Sol Energy 2019;190:549–60. https://doi.org/10.1016/j.solener.2019.08.061. [95] Ebner R, Zamini S, Újvári G. Defect analysis in different photovoltaic modules using electroluminescence (EL) and infrared (IR)-thermography. 25th Eur. Photovolt. Sol. Energy Conf. Exhib. / 5th World Conf. Photovolt. Energy Conversion, 6-10 Sept. 2010, Val. Spain, vol. 66, 2010, p. 37–9. https://doi.org/10.4229/25thEUPVSEC2010-1DV.2.8. [96] Gallardo-Saavedra S, Hernández-Callejo L, Alonso-García M del C, Santos JD, Morales-Aragonés JI, Alonso-Gómez V, et al. Nondestructive characterization of solar PV cells defects by means of electroluminescence, infrared thermography, I-V curves and visual tests: experimental study and comparison. Energy 2020. [97] Berardone I, Lopez Garcia J, Paggi M. Quantitative analysis of electroluminescence and infrared thermal images for aged monocrystalline silicon photovoltaic modules. 2017 IEEE 44th Photovolt. Spec. Conf., IEEE; 2018, p. 402–7. https://doi.org/10.1109/pvsc.2017.8366338. [98] Nasr Esfahani S, Asghari S, Rashid-Nadimi S. A numerical model for soldering process in silicon solar cells. Sol Energy 2017;148:49–56. https://doi.org/10.1016/j.solener.2017.03.065. [99] Li G, Akram MW, Jin Y, Chen X, Zhu C, Ahmad A, et al. Thermo-mechanical behavior assessment of smart wire connected and busbarPV modules during production, transportation, and subsequent field loading stages. Energy 2019;168:931–45. https://doi.org/10.1016/j.energy.2018.12.002. [100] Gallardo-Saavedra S, Karlsson B. Simulation, validation and analysis of shading - 89 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Table 3: Detection ranking criteria following [73]. ........................................................ 35 Table 4: Cells defects types and codes per string. ...................................................... 45 Table 4: Thermographic defects detected classified by the module affected component ...................................................................................................................................... 72 Table 5: All defects detected classified by the affected component and in three temperature groups ...................................................................................................... 73 Table 6: Anomalies identified, classified by their temperature. .................................... 77 - 96 of 137 - Chapter 2 CAP. 2 PUBLISHED PAPERS ‘Science knows no country, because knowledge belongs to humanity, and is the torch which illuminates the world. Science is the highest personification of the nation because that nation will remain the first which carries the furthest the works of thought and intelligence.’ Louis Pasteur. - 97 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis This chapter presents the papers and contributions done throughout this doctoral thesis. It has been structured in three subchapters, corresponding each of them with the publications related to each of the three partial objectives established in section 1.3 Hypothesis and objectives. 2.1. Study of the state of the art (Obj. 1) This section includes the following articles: [2.1.1] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Pérez O. Quantitative failure rates and modes analysis in photovoltaic plants. Energy 2019;183:825– 36. https://doi.org/10.1016/j.energy.2019.06.185. [2.1.2] Gallardo-Saavedra S, Pérez-Moreno J, Hernández-Callejo L, Duque-Pérez Ó. Failure rate determination and Failure Mode, Effect and Criticality Analysis (FMECA) based on historical data for photovoltaic plants ISES Conference Proceedings 2017:1-8. https://doi.org/10.18086/swc.2017.20.04. [2.1.3] Hernández-Callejo L, Gallardo-Saavedra S, Alonso-Gómez V. A review of photovoltaic systems: Design, operation and maintenance. Sol Energy 2019;188:426–40. https://doi.org/https://doi.org/10.1016/j.solener.2019.06.017. - 98 of 137 - Chapter 2: Published papers 2.1.1. Quantitative failure rates and modes analysis in PV plants QUANTITATIVE FAILURE RATES AND MODES ANALYSIS IN PHOTOVOLTAIC PLANTS Sara Gallardo-Saavedra a,b*, Luis Hernández-Callejo a, Oscar Duque-Pérez b a Universidad de Valladolid (UVa), School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB), Department of Agricultural and Forestry Engineering, Campus Duques de Soria, 42004, Soria, Spain b Universidad de Valladolid (UVa), Industrial Engineering School, Department of Electrical Engineering, Paseo del Cauce ,59, 47011, Valladolid, Spain *Corresponding autor Energy, vol. 183, pp. 825-836 (2019). https://doi.org/10.1016/j.energy.2019.06.185. - 99 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Abstract The greater challenge that researchers address and indicate while investigating about photovoltaic (PV) system failures during their Operation and Maintenance (O&M) is the lack of accessible reliable real quantitative data. For this reason, several publications have focused on this problem through a qualitative approach. However, this fact is one of the greater strengths of this paper, in which the quantitative information from the historical data of sixty-three PV plants portfolio in Italy and Spain has been accessible. Results obtained from the research provide essential information for main players involved in PV plants to identify failure modes and rates, in order to reduce investment risk and to focus their maintenance efforts on preventing those failures, improving longevity and performance of PV plants. The paper presents failure rates per PV Site and per kW, considering all portfolio and dividing it regarding five PV plants groups per size, distribution of failures per element, Mean Time Between Failures (MTBF), reparation times per affected element and the main failures modes examining each of the almost 100,000 complete alarms registered during the five years analyzed. - 100 of 137 - Chapter 2: Published papers 2.1.2. Failure rate determination and Failure Mode, Effect and Criticality Analysis (FMECA) based on historical data for photovoltaic plants FAILURE RATE DETERMINATION AND FAILURE MODE, EFFECT AND CRITICALITY ANALYSIS (FMECA) BASED ON HISTORICAL DATA FOR PHOTOVOLTAIC PLANTS Sara Gallardo-Saavedra1,2, Javier Pérez-Moreno2, Luis Hernández-Callejo1 and Óscar Duque-Pérez3 1 Universidad de Valladolid (UVa), School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB), Department of Agricultural and Forestry Engineering, Soria (Spain) 2 Solarig, Soria (Spain) 3 Universidad de Valladolid (UVa), Industrial Engineering School, Department of Electrical Engineering, Valladolid (Spain) Proceedings: ISES Conference Proceedings of the ISES Solar World Conference 2017 and the IEA SHC Solar Heating and Cooling Conference for Buildings and Industry 2017, pp. 1232-1239. https://doi.org/10.18086%2Fswc.2017.20.04 - 101 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Abstract It is essential for photovoltaic plants investors, operators and equipment manufacturers to identify the failure modes and rates of the system in order to reduce investment risk, to focus their maintenance efforts on preventing those failures and to improve longevity and performance of the PV Plant. In this paper, it is assessed the importance of the Failure Modes within a real existing portfolio in Spain and Italy of continuous operation since 2008 and it is identified the module level failure modes, which are imperceptible in the standard monitoring systems, through the application of thermographic inspection. The experimental Mean Time Between Failures (MTBF) and the failure rates are calculated and these ratios are used to define the ranking criterion to perform the Failure Mode, Effect and Criticality Analysis (FMECA) and to focus on the module level analysis. The conclusions highlight the most critical sub-system and failure modes within a photovoltaic PV plant. - 102 of 137 - Chapter 2: Published papers 2.1.3. A review of photovoltaic systems: design, operation and maintenance A REVIEW OF PHOTOVOLTAIC SYSTEMS: DESIGN, OPERATION AND MAINTENANCE Luis Hernández-Callejo1,*, Sara Gallardo-Saavedra1, Víctor Alonso-Gómez2 1 Department of Agricultural Engineering and Forestry, University of Valladolid (UVa), Campus Universitario Duques de Soria, 42004 Soria, Spain. L.H- C.: [email protected]; S.G-S.: [email protected]; 2 Department of Physics, University of Valladolid (UVa), Campus Universitario Duques de Soria, 42004 Soria, Spain. V.A-G.: [email protected] * Correspondence: [email protected] ; Tel.: +34-975-129-418; Fax: +34-975-129-401 Solar Energy, 188, pp. 426-440 (2019). https://doi.org/10.1016/j.solener.2019.06.017 - 103 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Abstract Nowadays renewable energies are becoming more important in the generation of electricity. Fossil resources do not present a sustainable option for the future since they are non-renewable sources of energy that contribute to environmental pollution. Within the sources of renewable generation, photovoltaic energy is the most used, and this is due to a large number of solar resources existing throughout the planet. At present, the greatest advances in photovoltaic systems (regardless of the efficiency of different technologies) are focused on improved designs of photovoltaic systems, as well as optimal operation and maintenance. This work intends to make a review of the photovoltaic systems, where the design, operation and maintenance are the key points of these systems. Within the design, the critical components of the system and their own design are revised. Regarding the operation, it is reviewed the general operation and the operation of hybrid systems, as well as the power quality. Finally, in relation to the maintenance of photovoltaic (PV) systems, it has been studied their performance, thermography and electroluminescence, dirt, risks and failure modes. - 104 of 137 - Chapter 2: Published papers 2.2. Detection, characterization and classification of defects in PV modules through the use of different inspection techniques (Obj. 2) This section includes the following articles: [2.2.1] Gallardo-Saavedra S, Karlsson B. Simulation, validation and analysis of shading effects on a PV system. Sol Energy 2018;170:828–39. https://doi.org/10.1016/j.solener.2018.06.035. [2.2.2] Moretón A, Gallardo S, Jiménez MM, Alonso V, Hernández L, Morales JI, et al. Influence of large periods of dc current injection in c-si photovoltaic panels. 36th Eur. Photovolt. Sol. Energy Conf. Exhib., 2019, p. 1079-108. https://doi.org/10.4229/eupvsec20192019-4av.1.38. [2.2.3] Gallardo S, Moretón A, Jiménez MM, Alonso V, Hernández L, Morales JI, et al. Failure diagnosis on photovoltaic modules using thermography, electroluminescence, RGB and I-V techniques. 36th Eur. Photovolt. Sol. Energy Conf. Exhib., 2019, p. 1171–5. https://doi.org/10.4229/eupvsec20192019- 4av.2.20. [2.2.4] Gallardo-Saavedra S, Hernández-Callejo L, Escamilla-Ambrosio PJ, Alonso- Gómez V. Merged images for fault detection in photovoltaic panels. II Ibero- American Congr. Smart Cities (ICSC-CITIES 2019), 2019, p. 167-178, ISBN: 978-958-5583-78-8. [2.2.5] Gallardo-Saavedra S, Hernández-Callejo L, Alonso-García M del C, Domingo- Santos J, Morales-Aragonés JI, Alonso-Gómez V, et al. Nondestructive characterization of solar PV cells defects by means of electroluminescence, infrared thermography, I-V curves and visual tests: experimental study and comparison. Energy 2020. [2.2.6] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Perez O. Analysis and characterization of PV module defects by thermographic inspection. Rev Fac Ing Univ Antioquia 2019:92–104. https://doi.org/10.17533/udea.redin.20190517. - 105 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 2.2.4. Merged images for fault detection in photovoltaic panels MERGED IMAGES FOR FAULT DETECTION IN PHOTOVOLTAIC PANELS Sara Gallardo-Saavedra1[0000-0002-2834-5591], Luis Hernández-Callejo1[0000-0002-8822-2948], Ponciano Jorge Escamilla-Ambrosio 2[0000-0003-3772-3651] and Víctor Alonso-Gómez1[0000- 0001-5107-4892]. 1 University of Valladolid, Campus Duques de Soria, 42004 Soria, España 2 Instituto Politécnico Nacional: Mexico, Distrito Federal, MX Proceedings of the II Ibero-American Congress of Smart Cities (ICSC-CITIES 2019), p. 167-178. ISBN: 978-958-5583-78-8. - 112 of 137 - Chapter 2: Published papers Abstract Photovoltaic systems are of great interest throughout the world. Its installation, operation and maintenance are crucial for achieving energy sustainability worldwide. In recent years, the inspection of defects in photovoltaic modules using images (thermography, electroluminescence, visible, etc.) is increasing considerably. Therefore, this work presents the experience of merging images to detect faults in photovoltaic modules. - 113 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 2.2.5. Nondestructive characterization of solar PV cells defects by means of electroluminescence, infrared thermography, I-V curves and visual tests: experimental study and comparison NONDESTRUCTIVE CHARACTERIZATION OF SOLAR PV CELLS DEFECTS BY MEANS OF ELECTROLUMINESCENCE, INFRARED THERMOGRAPHY, I-V CURVES AND VISUAL TESTS: EXPERIMENTAL STUDY AND COMPARISON Sara Gallardo-Saavedra a *, Luis Hernández-Callejo a, María del Carmen Alonso- García b, José Domingo Santos b, José Ignacio Morales-Aragonés a, Víctor Alonso-Gómez a, Ángel Moretón-Fernández c, Miguel Ángel González-Rebollo c and Oscar Martínez-Sacristán c. a Universidad de Valladolid (UVa), School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB), Department of Agricultural and Forestry Engineering, Campus Duques de Soria, 42004, Soria, Spain. b Photovoltaic Solar Energy Unit (Energy Department, CIEMAT), Avda. Complutense 40, 28040 Madrid, Spain. c Universidad de Valladolid (UVa), Department of Condensed Matter Physics, GdS Optronlab Group, Paseo de Belen 19, 47011, Valladolid, Spain. *Corresponding author. Energy. (2020). - 114 of 137 - Chapter 2: Published papers Abstract Photovoltaic (PV) modules are the core of every PV system, representing the power generation and their operation will affect the overall plant performance. It is one of the elements within a PV site with the higher failure appearance, being essential their proper operation to produce reliable, efficient and safety energy. Quantitative analysis and characterization of manufacturing, soldering and breaking PV defects is performed by a combination of electroluminescence (EL), infrared thermography (IRT), electrical current voltage (I-V) curves and visual inspection. Equivalent-circuit model characterization and microscope inspection are also performed as additional techniques when they contribute to the defects characterization. A 60-cells polycrystalline module has been ad hoc manufactured for this research, with different defective and nondefective cells. All cells are accessible from the backside of the module and the module includes similar kinds of defects in the same bypass string. This paper characterizes different defects of PV modules to control, mitigate or eliminate their influence and being able to do a quality assessment of a whole PV module, relating the individual cells performance with the combination of defective and non-defective cells within the module strings, with the objective of determining their interaction and mismatch effects, apart from their discrete performance. - 115 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 2.2.6. Analysis and characterization of PV module defects by thermographic inspection. ANALYSIS AND CHARACTERIZATION OF PV MODULE DEFECTS BY THERMOGRAPHIC INSPECTION. Sara Gallardo-Saavedra1,2 (0000-0002-2834-5591), Luis Hernádez-Callejo1 (0000- 0002-8822-2948) and Óscar Duque-Pérez2 (0000-0003-2994-2520) 1Departamento de Ingeniería Agropecuaria y Forestal, Facultad de Ingeniería Forestal, Ingeniería Agronómica y de la Industria de la Bioenergía (EIFAB), Campus Duques de Soria, Universidad de Valladolid (UVa). Calle Universidad, C.P. 42004, Soria, España. 2 Departamento de Ingeniería Eléctrica, Escuela de Ingeniería Industrial, Universidad de Valladolid (UVa). Paseo del Cauce, 59, C.P. 47011, Valladolid, España. Facultad de Ingeniería Universidad de Antioquía, no.93, pp. 92-104, Oct-Dec 2019. https://doi.org/10.17533/udea.redin.20190517 - 116 of 137 - Chapter 2: Published papers Abstract Being able to detect, to identify and to quantify the severity of defects that appear within photovoltaic modules is essential to constitute a reliable, efficient and safety system, avoiding energy losses, mismatches and safety issues. The main objective of this paper is to perform an in-depth, onsite study of 17,142 monocrystalline modules to detect every single existing defect, classifying them in different groups, studying the variance of the same kind of defect in different modules and the patterns of each group of thermal defects. Results can be useful in a subsequent development of a software to automatically detect if a module has an anomaly and its classification. Focusing on the results obtained, all faults detected have been classified in five different thermographic defects modes: hotspot in a cell, bypass circuit overheated, hotspot in the junction box, hotspot in the connection of the busbar to the junction box and whole module overheated. An analysis of patterns of the different defects is included, studiyng location within the module, size and temperature statistical results, as average temperature, standard deviation, maximum temperature, median and first and third quartile. Resumen Ser capaz de detectar, identificar y cuantificar la gravedad de los defectos que aparecen en los módulos fotovoltaicos es esencial para constituir un sistema fiable, eficiente y seguro, evitando pérdidas de energía, desajustes y problemas de seguridad. El objetivo principal de esta investigación es realizar un estudio de 17.142 módulos monocristalinos para detectar cada defecto existente, clasificándolos en diferentes grupos, estudiando la varianza del mismo tipo de defecto en diferentes módulos y los patrones de cada grupo de defectos térmicos. Los resultados obtenidos pueden ser útiles en el desarrollo posterior de un software de detección automática de anomalías en módulos y su clasificación. Atendiendo a los resultados obtenidos, los defectos detectados se han clasificado en cinco modos de fallo termográficos: sobrecalentamiento en celdas, en circuito bypass, en la caja de conexiones, en la conexión entre la barra colectora (busbar) y la caja de conexiones y en el módulo completo. Se incluye un análisis de patrones de los diferentes defectos, estudiando su ubicación dentro del módulo, tamaño y resultados estadísticos de temperatura, como temperatura promedio, desviación estándar, temperatura máxima, mediana y primer y tercer cuartil. - 117 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis 2.3. Analysis of aerial thermography as a novel inspection technique (Obj. 3) This section includes the following articles: [2.3.1] Gallardo-Saavedra S, Hernández-Callejo L, Duque-Perez O. Technological review of the instrumentation used in aerial thermographic inspection of photovoltaic plants. Renew Sustain Energy Rev 2018;93:566–79. https://doi.org/10.1016/J.RSER.2018.05.027. [2.3.2] Gallardo-Saavedra S, Alfaro-Mejia E, Hernández-Callejo L, Duque-Pérez Ó, Loaiza-Correa H, Franco-Mejia E. Aerial thermographic inspection of photovoltaic plants: analysis and selection of the equipment. ISES Conference Proceedings 2017:1-9 https://doi.org/10.18086/swc.2017.20.03. [2.3.3] Dávila-Sacoto M, Hernández-Callejo L, Alonso-Gómez V, Gallardo-Saavedra S, González-Morales L. Low-cost infrared thermography in aid of photovoltaic panels degradation research. Fac Ing Univ Antioquía 2020. [2.3.4] Gallardo-Saavedra S, Hernandez-Callejo L, Duque-Perez O. Image Resolution Influence in Aerial Thermographic Inspections of Photovoltaic Plants. IEEE Trans Ind Informatics 2018;14:5678–86. https://doi.org/10.1109/TII.2018.2865403. - 118 of 137 - Chapter 2: Published papers 2.3.1. Technological review of the instrumentation used in aerial thermographic inspection of photovoltaic plants TECHNOLOGICAL REVIEW OF THE INSTRUMENTATION USED IN AERIAL THERMOGRAPHIC INSPECTION OF PHOTOVOLTAIC PLANTS Sara Gallardo-Saavedra a,b,*, Luis Hernández-Callejo a, Oscar Duque-Pérez b a,* Universidad de Valladolid (UVa), School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB), Department of Agricultural and Forestry Engineering, Campus Duques de Soria, 42004, Soria, Spain b Universidad de Valladolid (UVa), Industrial Engineering School, Department of Electrical Engineering, Paseo del Cauce ,59, 47011, Valladolid, Spain Renewable and sustainable energy reviews, 93, pp. 566-579 (2018). https://doi.org/10.1016%2Fj.rser.2018.05.027 - 119 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Abstract Photogrammetric studies performed with Unmanned Aerial Vehicles (UAV) have recently become more popular as they present an interesting low-cost alternative. A novel application of thermography with UAVs has been validated during the last years: aerial thermography for inspection of photovoltaic plants as a useful diagnostic technique to assess performance of photovoltaic modules, superseding time-consuming traditional manual methods. This paper describes the current state of thermographic cameras and UAV technology, with the aim of examining the general principles of aerial thermographic measurement and required instrumentation, detailing the most important system aspects, discussing new developments and future trends in aerial thermography sensors and instrumentation, and evaluating them for specific application of aerial thermographic inspection of photovoltaic plants. - 120 of 137 - Chapter 2: Published papers 2.3.2. Aerial thermographic inspection of photovoltaic plants: analysis and selection of the equipment AERIAL THERMOGRAPHIC INSPECTION OF PHOTOVOLTAIC PLANTS: ANALYSIS AND SELECTION OF THE EQUIPMENT Sara Gallardo-Saavedra1, Estefanía Alfaro-Mejia2 Luis Hernández-Callejo1, Óscar Duque-Pérez3, Humberto Loaiza-Correa2 and Edinson Franco-Mejia2 1 Universidad de Valladolid (UVa), School of Forestry, Agronomic and Bioenergy Industry Engineering (EIFAB), Department of Agricultural and Forestry Engineering, Soria (Spain) 2 Universidad del Valle, Escuela de Ingeniería Eléctrica y Electrónica, Santiago de Cali (Colombia) 3 Universidad de Valladolid (UVa), Industrial Engineering School, Department of Electrical Engineering, Valladolid (Spain) Proceedings: ISES Conference Proceedings of the ISES Solar World Conference 2017 and the IEA SHC Solar Heating and Cooling Conference for Buildings and Industry 2017, pp. 1223-1231. https://doi.org/10.18086%2Fswc.2017.20.03 - 121 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis This chapter presents the conclusions of the research described in the thesis and the main contributions. The aim and objectives of the research, outlined in the Introduction, are reviewed and their achievement addressed. Finally, proposals for future work are suggested. 3.1. Conclusions and PhD contributions In this dissertation, it has been presented an analysis of the detection, characterization and classification of defects in PV modules through the use of IRT, EL, I-V curves and visual analysis, which is the aim of the research. To achieve this main objective, the research has been focused on three main threads, coinciding with the three partial objectives of the thesis: the study of the state of the art, the detection, characterization and classification of defects in PV modules through the use of different inspection techniques and the analysis of aerial IRT as a novel inspection technique. The main contributions of the thesis on these three areas are presented in this section. Firstly, regarding the study of the state of the art, the failure rates and modes in PV plants and the design, operation and maintenance of these sites have been analyzed. It has been proved how PV production is never faultless, like any other technology, quantifying the failure rates corresponding to each of the elements in a PV plant. It has been seen how although PV plants are getting older each year, a proper maintenance can be responsible of keeping the defects ratio or even reducing it. Accurately performed predictive, preventive and corrective maintenance include improvements in PV plants, which reduces the failure ratio in the long term. It has been notice how there are defects that do not depend on the plant size or the plant number of elements or components; therefore, the failure number per power unit is higher in smaller plants. At the front of the list of elements with more failures are the monitoring and communication systems, followed by the MV elements, the inverter and the PV generator (including the PV modules). Each defect involves each own repairing time to fix a defect, and excluding the elements that usually require the presence of the manufacturer or outsourcing for their reparation and have a higher reparation time, the PV generator is the element with the higher fixing value. Studying the failure ratios occurrence combined with their severity and detection probability, it is seen how inverter, MV, auxiliary services and PV modules are the most relevant systems. It has been notice that the module is - 128 of 137 - Chapter 3: Conclusions and future work the key element on a PV system, but, even so, it has limited access to local monitoring alarms, which are not frequently automatically generated, being specially complicated the detection of failures at this level and even more difficult the recognition of the failure cause and mode. From the study of the state of the art, it is highlighted the importance of the research on PV modules failures detection, classification and characterization in order to produce reliable, efficient and safety energy. Secondly, concerning the detection, characterization and classification of defects in PV modules through the use of different inspection techniques, it has been addressed three different aspects. The first deals with the detection of PV defects using different techniques and the comparison of the information obtained with each of them. From an experimental point of view, our contribution lies in the simultaneous application of these inspection techniques to several PV modules with different failures. These experiments were performed for studying the correlation between them. Results show that EL and IRT under current injection on modules are closely correlated, while IRT under normal operation (sun exposure) reveals complementary information not detected in EL but existing in the visible spectrum. It has been concluded, it is advisable using as different techniques as possible to characterize the actual state of the module and to explain its I-V curve. It has been studied the possibility to use different algorithms to generate merged images of frames captured by means of EL and IRT in the fourth quadrant, with the objective of exploring how using image processing techniques to align multiple scenes in a single integrated image simplifies the analysis and reporting of results of inspections of PV plants, resulting to be a very interesting and promising field. Regarding the I-V curves, several electrical circuits which model a PV system have been simulated using LTspice and validated with experimental measurements. We have obtained successful results, both on simulation and on real experimentation, showing that the designed model is a very useful tool that can be used to study the performance of shaded or defective PV systems, allowing the simulation of numerous faults. Regarding the analysis of the EL current injection in modules, the main focus of this thesis was on answering to the generalized concern about how this current injection can affect the PV modules service life. In the research it is not appreciated that this current has any negative effect on the performance of damaged modules. In the second investigation, a quantitative analysis and characterization of manufacturing, soldering and breaking PV defects by a combination of the inspection techniques previously analyzed was performed for the characterization of PV cells and modules as an important part of the search process. Equivalent-circuit model characterization and microscope inspection have also been performed as additional techniques when they contribute to - 129 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis the defects characterization. Our contribution in this part was to study in detail and characterize different defects and their combination within a module. In particular, regarding the manufacturing defects, it has been seen that there is a tendency to decrease the light emission during EL tests with the efficiency drop. Short-circuited cell does not emit any light during EL and it is seen colder than the rest in IRT, which authors propose to denominate as cold spot, defining it as cell or group of cells at abnormal low temperature in a PV system. Low-efficiency defects are seen as a tire imprint in the cell EL images and can be originated due to inhomogeneity during the firing process or in the transportation belt in the cell production and they can experiment almost a 50% power loss. Soldering defective cells present smaller parameter deviations than the rest of defects studied. It has been proved that there are more loses when there is a local soldering defect in just a part of the bus, as the series resistance is considerably increased. Conversely, having all (or some) tabs loose does not interrupt the electrical contact despite not being welded, and its influence is not disastrous. However, this could be altered after some years of operation of the module in which the encapsulation is exposed to thermal cycles. It has also been seen that the greatest danger of bad welds are usually the electric arcs, since that bad contact causes the tab to warm up and it can come out to disconnect it. A relevant tension difference between two points very close can generate micro arcs. That ends up burning the back sheet behind, or if they are generated in broken cells, the edges of the cell are usually burned and also depending on the composition of the paste together with the permeability of the back sheet, the humidity causes the fingers to blacken, being easily identifiable the cell breakages, which is commonly known as snail trails. Concerning broken cells, it has been analyzed how the shape of the I-V curves of the affected cells is very similar to the one of the nominal equivalent cell but decreasing the current proportionally. Alternatively, if a piece of other cell over placed in the front part of a cell generates a parallel electrical path which is responsible of a small Rsh in the cell, turning the I-V curve into a straight line. The third point presents the classification of PV defects attending to failure patterns. A discussion on the different thermographic defects have been presented. Our contribution here is the classification of defects in five different modes base on an experimental database generated with numerous on-site measurements and the study of their patterns: hotspot in a cell or in a group of cells, bypass circuit overheated, hotspot in the junction box, hotspot in the connection of the busbar to the junction box and whole module overheated, with more than three quarters of the affected modules correspond to cell hotspots and being this the defect type that shows higher temperature differences, achieving up to - 130 of 137 - Chapter 3: Conclusions and future work 77.4ºC between the overheated hotspot and the healthy area in one of the tested modules. This study results can be useful as a base to develop the patterns of the different kind of defects in a software to automatically detect if a module has an anomaly and its classification. Thirdly, aerial IRT has been proposed to address the drawbacks of traditional inspection techniques by using thermal cameras onboard of UAVs. Up-to-date traditional techniques have been applied to diagnose and assess the performance of PV modules resulting in costly and time-consuming inspections. The use of thermographic cameras and UAVs has been analyzed in detail. Regarding the analysis of aerial IRT as a novel inspection technique the main contributions of the work are three: to review the available equipment focusing on their desirable characteristic for their application on aerial thermographic inspection of PV sites, to review the most important aspects in the tests that should be taken into account in order to obtain valid results and to study the influence of the resolution of IRT images. Although there are several researchers developing this novel technique, there were no clear guidelines for selection of involved instrumentation meeting the needs of PV inspections. The research presented aims to present a clear review and discussion of available equipment and its main characteristics applied to PV plants, as well as to review the equipment already used in this application as reported in the related bibliography. The most important aspects on this regard are reviewed, such as resolution, thermal sensitivity, accuracy, lens and the corresponding FOV, radiometric functionality, visual or RGB images, frame rate or temperature range for the thermographic camera, as well as stability of the system, maximum altitude, flight duration and maximum payload, full compatibility between instruments and duration of batteries for UAVs. The second contribution relies in the review of the most important aspects in the tests that should be taken into account in order to make them valid, as the presence of steady state conditions, minimum irradiance value of at least 600 W/m2, maximum wind speed of 28 km/h, maximum UAV moving speed of 3 m/s, clean modules, accurate angle of view to avoid reflections, setting the correct emissivity, reflected temperature, relative humidity, temperature and distance between the sensor and the object values in the camera, correctly select the temperature range and properly calibrating the camera. Finally, it has been proved that the images resolution will strongly influence the results by means of the particular case of one manual and two aerial inspections at 30 and 80 m, respectively, of a 3 MW site. Important anomalies can be hidden or disdained if the inspection is not well planned. Therefore, it has been seen that the aerial IRT should be arranged as a function of the thermographic camera and lens that will be used, selecting a flight height that satisfies the Spatial and the Measurement - 131 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis Resolutions, understanding how the resolution affects the results and how to interpret them and considering the important factors previously mentioned. Thus, it is considered that the main objective of the doctoral thesis, which is the analysis of the detection, characterization and classification of defects in PV modules through the use of IRT, EL, I-V curves and visual analysis have been satisfactorily achieved by means of the suitably completion of the three partial objectives established. 3.2. Future work Although it is considered that the goal of the thesis has been accomplished, there are plenty of advances that could be done in order to achieve supplementary results. In this section are detailed some of the open issues that could deserve further research for some of the aspects related to this thesis (some of which we are already working on). • Aerial inspection with EL sensors It has been seen how the detection accuracy of IRT is limited by many aspects, as the weather conditions, and to faults causing a sufficient increase in temperature. Another alternative fast and accurate automatic UAV-based inspection system for PV plants can be based on EL sensors. This technique could be a significant leap forward in PV plant inspection technologies, combining the speed of UAV-based inspections with the in-depth analysis of EL, detecting a wider range of PV panel failures. This automation of the EL imaging process can be done through a bidirectional PV inverter, which will require full coordination with the UAV planning. • Bidirectional inverters for PV maintenance With respect to EL, as mentioned, it is a limitation to have to disconnect the PV solar modules and connect them to an external power supply. In line with the development of aerial inspections using EL sensors, it is essential the research on bidirectional inverters. This kind of inverters can be used in aerial EL inspections to feed electricity to the strings of modules that are going to be inspected without the necessity of disconnecting the strings from the inverter neither having an external DC source that would slow down the inspection process as well as raise the price. Actually, in utility scale PV plants this kind of inverters can be already used to regulate and monitor the flow of power between a DC bus and an AC grid and to restrict the voltage expanse at - 132 of 137 - Chapter 3: Conclusions and future work the former to only a certain permissible range of voltages. A bidirectional inverter is the one that not only performs the DC to AC conversion, but also performs the conversion of AC power to DC, with the major advance of giving the flexibility to the user to decide when to buy power from an electrical grid and when to sell so as to make the maximum profit based on the price of electricity at a particular point in time. • Advanced measures in PV plants It has been seen that many different failure modes can appear at the module level. However, typically the monitoring system trails the string series current of multiple modules connected, or even the inverter current, which is the sum of the parallel string current, and does not track the performance of the individual PV modules. Therefore, PV module alarms are not frequently automatically generated being specially complicated the detection of failures at this level and even more difficult the recognition of the failure cause and mode. The need for advanced and intelligent devices to apply to PV solar modules is a necessity to turn them into plants of the future. Up to now, I-V curves is a process totally dependent on the human operator, who needs to interrupt a PV solar string to measure a single module, with the partial loss of production due to the disconnection. An element of intelligence associated with a PV solar plant of the future, could consist on a centralized / decentralized measurement of I-V curves measurement of all PV solar modules in a string, and in a coordinated and simultaneous way, minimizing the energy losses during the measures. It would allow the simultaneous electrical parameters acquisition of all modules in a string, making easier the PV plant assessment. • Characterization of soldering defects in the front of the module It is proposed the analysis of the same soldering defects performed on the front site of the cells instead of on the back site as a future work, in order to characterize them and compare the effect of the soldering defect location within the cell. It could be interesting seeing if they are detectable with EL and if the lack of this metal sheet in the front site generates heat differences in the front soldering defects, as well as comparing the power production changes in case the defects are in the front or in the back of cells. Additionally, as it has been seen that having all tabs loose may not interrupt the current in a new manufactured module, it would also be interesting to study its influence after some years of operation or after several thermal cycles. - 133 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis • Characterization of combination of PV defects by simulations The defects characterization performed within this thesis in combination with the electrical circuits developed, which model a PV system using LTspice, could be used to simulate the module behavior depending on the defects combination and location within a module. With this research, it could be possible predicting the PV modules behavior according to their defects without the necessity of physically having them, and to determine the impact of different defects combinations or patterns. • Image processing automation and PV defects detection and classification Something that seems to fit in the advances in O&M is the use of techniques based on AI, to be able to guarantee the early detection of failures, or the increase of the energy efficiency of PV solar plants. Once a discussion on the different thermographic defects has been provided within the thesis objectives, these results can be useful as a base to set the patterns of the different kind of defects in a software to automatically detect if a module has an anomaly and its classification. - 134 of 137 - Annex A ANEXO A ADDITIONAL PUBLICATIONS ‘A winner is a dreamer who never gives up.’ Nelson Mandela. - 135 of 137- Detection, classification and characterization of defects in photovoltaic modules through the use of thermography, electroluminescence, I-V curves and visual analysis This annex includes other publications in which the author of this thesis has also contributed, although they have not been included in the “compendium of articles” of the doctoral thesis. • Hernández-Callejo L, Gallardo-Saavedra S, Diez-Cercadillo A, Alonso-Gómez V. Analysis of the influence of DC optimizers on photovoltaic production. Rev Fac Ing 2020:43–55. https://doi.org/10.17533/udea.redin.20190521. • Hernández-Callejo L, Gallardo-Saavedra S, Diez-Cercadillo A, Alonso-Gómez V. Study of the Influence of DC-DC Optimizers on PV-Energy Generation. Smart Cities. Commun. Comput. Inf. Sci., 2019, p. 1–17. https://doi.org/https://doi.org/10.1007/978-3-030-12804-3_1. • Hernández-Callejo L, Gallardo-Saavedra S, Diez-Cercadillo A, Alonso-Gómez V. Study of the Influence of DC-DC Optimizers on PV-Energy Generation. I Ibero- American Congr. Smart Cities (ICSC-CITIES 2018), Soria (Spain): 2018, p. 616– 33. • Gallardo-Saavedra S, Hernández-Callejo L, Duque-Pérez Ó. Analysis and Characterization of Thermographic Defects at the PV Module Level. In: Nesmachnow S, Hernández Callejo L, editors. Smart Cities. Commun. Comput. Inf. Sci., vol. 978, Cham: Springer International Publishing; 2019, p. 80–93. https://doi.org/https://doi.org/10.1007/978-3-030-12804-3_7. • Gallardo-Saavedra S, Hernádez-Callejo L, Duque-Pérez Ó. Analysis and characterization of thermographic defects at the PV module level. I Ibero- American Congr. Smart Cities (ICSC-CITIES 2018), Soria (Spain): 2018, p. 487– 502. • Dávila-Sacoto M, Hernández-Callejo L, Alonso-Gómez V, Gallardo-Saavedra S, González LG. Detecting Hot Spots in Photovoltaic Panels Using Low-Cost Thermal Cameras. Smart Cities. Commun. Comput. Inf. Sci., vol. 1152 CCIS, Springer; 2020, p. 38–53. https://doi.org/10.1007/978-3-030-38889-8_4. • Dávila-Sacoto M, Hernández-Callejo L, Alonso-Gómez V, Gallardo-Saavedra S, González LG. Detecting Hot Spots in Photovoltaic Panels Using Low-Cost Thermal Cameras. II Ibero-American Congr. Smart Cities (ICSC-CITIES 2019), 2019, p. 151–166. • Hernández-Martínez B, Gallardo-Saavedra S, Hernández-Callejo L, Alonso- Gómez V, Morales-Aragonés JI. General Purpose I-V Tester Developed to Measure a Wide Range of Photovoltaic Systems. Smart Cities. Commun. Comput. Inf. Sci., vol. 1152 CCIS, Springer; 2020, p. 135–45. https://doi.org/10.1007/978-3-030-38889-8_11. - 136 of 137 - Annex A: Additional publications • Hernández-Martínez B, Gallardo-Saavedra S, Hernández-Callejo L, Alonso- Gómez V, Morales-Aragonés JI. General purpose I-V tester developed to measure a wide range of photovoltaic systems. II Ibero-American Congr. Smart Cities (ICSC-CITIES 2019), 2019, p. 33–44. • Hernández-Martínez B, Hernández-Callejo L, Gallardo-Saavedra S, Alonso- Gómez V, Morales-Aragones JI. Low-cost illumination system for photovoltaic devices validation at the control and constant irradiance. II Ibero-American Congr. Smart Cities (ICSC-CITIES 2019), 2019, p. 842–851. • Ballestín-Fuertes J, Muñoz-Cruzado-Alba J, Sanz-Osorio JF, Hernández-Callejo L, Alonso-Gómez V, Morales-Aragones I, Gallardo-Saavedra S, et al. Novel Utility-Scale Photovoltaic Plant Electroluminescence Maintenance Technique by Means of Bidirectional Power Inverter Controller. Appl Sci 2020: 3084. https://doi.org/10.3390/app10093084. - 137 of 137-