Large number of renewable energy based distributed energy resources(DERs) are integrated into the conventional power grid using power electronic interfaces. This causes increased need for efficient power conversion, advanced control, and DER situational awareness. In case of photovoltaic(PV) grid integration, power is processed in two stages, namely DC-DC and DC-AC. In this work, two novel soft-switching schemes for quadratic boost DC-DC converters are proposed for PV microinverter application. Both the schemes allow the converter to operate at higher switching frequency, reducing the converter size while still maintaining high power conversion efficiency. Further, to analyze the impact of high penetration DERs on the power system a real-time simulation platform has been developed in this work. A real, large distribution feeder with more than 8000 buses is considered for investigation. The practical challenges in the implementation of a real-time simulation (such as number of buses, simulation time step, and computational burden) and the corresponding solutions are discussed. The feeder under study has a large number of DERs leading to more than 200% instantaneous PV penetration. Opal-RT ePHASORSIM model of the distribution feeder and different types of DER models are discussed in detailed in this work. A novel DER-Edge-Cloud based three-level architecture is proposed for achieving solar situational awareness for the system operators and for real-time control of DERs. This is accomplished using a network of customized edge-intelligent-devices(EIDs) and end-to-end solar energy optimization platform(eSEOP). The proposed architecture attains superior data resolution, data transfer rate and low latency for the end-to-end communication. An advanced PV string inverter with control and communication capabilities exceeding those of state-of-the-art, commercial inverters has been developed to demonstrate the proposed real-time control. A power-hardware-in-loop(PHIL) and EID-in-loop(EIL) testbeds are developed to verify the impact of large number of controllable DERs on the distribution system under different operational modes such as volt-VAr, constant reactive power and constant power factor. Edge level data analytics and intelligent controls such as autonomous reactive power allocation strategy are implemented using EIL testbed for real-time monitoring and control. Finally, virtual oscillator control(VOC) for grid forming inverters and its operation under different X/R conditions are explored.

Operational efficiency of solar energy farms requires detailed analytics and information on each panel regarding voltage, current, temperature, and irradiance. Monitoring utility-scale solar arrays was shown to minimize the cost of maintenance and help optimize the performance of photovoltaic (PV) arrays under various conditions. This dissertation describes a project that focuses on the development of machine learning and neural network algorithms. It also describes an 18kW solar array testbed for the purpose of PV monitoring and control. The use of the 18kW Sensor Signal and Information Processing (SenSIP) PV testbed which consists of 104 modules fitted with smart monitoring devices (SMDs) is described in detail. Each of the SMDs has embedded, a wireless transceiver, and relays that enable continuous monitoring, fault detection, and real-time connection topology changes. Data is obtained in real time using the SenSIP PV testbed. Machine learning and neural network algorithms for PV fault classification is are studied in depth. More specifically, the development of a series of customized neural networks for detection and classification of solar array faults that include soiling, shading, degradation, short circuits and standard test conditions is considered. The evaluation of fault detection and classification methods using metrics such as accuracy, confusion matrices, and the Risk Priority Number (RPN) is performed. The examination and assessment the classification performance of customized neural networks with dropout regularizers is presented in detail. The development and evaluation of neural network pruning strategies and illustration of the trade-off between fault classification model accuracy and algorithm complexity is studied. This study includes data from the National Renewable Energy Laboratory (NREL) database and also real-time data collected from the SenSIP testbed at MTW under various loading and shading conditions. The overall approach for detection and classification promises to elevate the performance and robustness of PV arrays.

In the current photovoltaic (PV) industry, the O&M (operations and maintenance) personnel in the field primarily utilize three approaches to identify the underperforming or defective modules in a string: i) EL (electroluminescence) imaging of all the modules in the string; ii) IR (infrared) thermal imaging of all the modules in the string; and, iii) current-voltage (I-V) curve tracing of all the modules in the string. In the first and second approaches, the EL images are used to detect the modules with broken cells, and the IR images are used to detect the modules with hotspot cells, respectively. These two methods may identify the modules with defective cells only semi-qualitatively, but not accurately and quantitatively. The third method, I-V curve tracing, is a quantitative method to identify the underperforming modules in a string, but it is an extremely time consuming, labor-intensive, and highly ambient conditions dependent method. Since the I-V curves of individual modules in a string are obtained by disconnecting them individually at different irradiance levels, module operating temperatures, angle of incidences (AOI) and air-masses/spectra, all these measured curves are required to be translated to a single reporting condition (SRC) of a single irradiance, single temperature, single AOI and single spectrum. These translations are not only time consuming but are also prone to inaccuracy due to inherent issues in the translation models. Therefore, the current challenges in using the traditional I-V tracers are related to: i) obtaining I-V curves simultaneously of all the modules and substrings in a string at a single irradiance, operating temperature, irradiance spectrum and angle of incidence due to changing weather parameters and sun positions during the measurements, ii) safety of field personnel when disconnecting and reconnecting of cables in high voltage systems (especially field aged connectors), and iii) enormous time and hardship for the test personnel in harsh outdoor climatic conditions. In this thesis work, a non-contact I-V (NCIV) curve tracing tool has been integrated and implemented to address the above mentioned three challenges of the traditional I-V tracers.

This work compares I-V curves obtained using a traditional I-V curve tracer with the I-V curves obtained using a NCIV curve tracer for the string, substring and individual modules of crystalline silicon (c-Si) and cadmium telluride (CdTe) technologies. The NCIV curve tracer equipment used in this study was integrated using three commercially available components: non-contact voltmeters (NCV) with voltage probes to measure the voltages of substrings/modules in a string, a hall sensor to measure the string current and a DAS (data acquisition system) for simultaneous collection of the voltage data obtained from the NCVs and the current data obtained from the hall sensor. This study demonstrates the concept and accuracy of the NCIV curve tracer by comparing the I-V curves obtained using a traditional capacitor-based tracer and the NCIV curve tracer in a three-module string of c-Si modules and of CdTe modules under natural sunlight with uniform light conditions on all the modules in the string and with partially shading one or more of the modules in the string to simulate and quantitatively detect the underperforming module(s) in a string.

The goal of any solar photovoltaic (PV) system is to generate maximum energy throughout its lifetime. The parameters that can affect PV module power output include: solar irradiance, temperature, soil accumulation, shading, encapsulant browning, encapsulant delamination, series resistance increase due to solder bond degradation and corrosion and shunt resistance decrease due to potential induced degradation, etc. Several PV modules together in series makes up a string, and in a power plant there are a number of these strings in parallel which can be referred to as an array. Ideally, PV modules in a string should be identically matched to attain maximum power output from the entire string. Any underperforming module or mismatch among modules within a string can reduce the power output. The goal of this project is to quickly identify and quantitatively determine the underperforming module(s) in an operating string without the use of an I-V curve tracer, irradiance sensor or temperature sensor. This goal was achieved by utilizing Radiovoltmeters (RVM). In this project, it is demonstrated that the voltages at maximum power point (Vmax) of all the individual modules in a string can be simultaneously and quantitatively obtained using RVMs at a single irradiance, single module operating temperature, single spectrum and single angle of incidence. By combining these individual module voltages (Vmax) with the string current (Imax) using a Hall sensor, the power output of individual modules can be obtained, quickly and quantitatively.

Global photovoltaic (PV) module installation in 2018 is estimated to exceed 100 GW, and crystalline Si (c-Si) solar cell-based modules have a share more than 90% of the global PV market. To reduce the social cost of PV electricity, further developments in reliability of solar panels are expected. These will lead to realize longer module lifetime and reduced levelized cost of energy. As many as 86 failure modes are observed in PV modules [1] and series resistance increase is one of the major durability issues of all. Series resistance constitutes emitter sheet resistance, metal-semiconductor contact resistance, and resistance across the metal-solder ribbon. Solder bond degradation at the cell interconnect is one of the primary causes for increase in series resistance, which is also considered to be an invisible defect [1]. Combination of intermetallic compounds (IMC) formation during soldering and their growth due to solid state diffusion over its lifetime result in formation of weak interfaces between the solar cell and the interconnect. Thermal cycling under regular operating conditions induce thermo-mechanical fatigue over these weak interfaces resulting in contact reduction or loss. Contact reduction or loss leads to increase in series resistance which further manifests into power and fill factor loss. The degree of intermixing of metallic interfaces and contact loss depends on climatic conditions as temperature and humidity (moisture ingression into the PV module laminate) play a vital role in reaction kinetics of these layers. Modules from Arizona and Florida served as a good sample set to analyze the effects of hot and humid climatic conditions respectively. The results obtained in the current thesis quantifies the thickness of IMC formation from SEM-EDS profiles, where similar modules obtained from different climatic conditions were compared. The results indicate the thickness of the IMC and detachment degree to be growing with age and operating temperatures of the module. This can be seen in CuxSny IMC which is thicker in the case of Arizona module. The results obtained from FL

ii

aged modules also show that humidity accelerates the formation of IMC as they showed thicker AgxSny layer and weak interconnect-contact interfaces as compared to Arizona modules. It is also shown that climatic conditions have different effects on rate at which CuxSny and AgxSny intermetallic compounds are formed.

The volume of end-of-life photovoltaic (PV) modules is increasing as the global PV market increases, and the global PV waste streams are expected to reach 250,000 metric tons by the end of 2020. If the recycling processes are not in place, there would be 60 million tons of end-of-life PV modules lying in the landfills by 2050, that may not become a not-so-sustainable way of sourcing energy since all PV modules could contain certain amount of toxic substances. Currently in the United States, PV modules are categorized as general waste and can be disposed in landfills. However, potential leaching of toxic chemicals and materials, if any, from broken end-of-life modules may pose health or environmental risks. There is no standard procedure to remove samples from PV modules for chemical toxicity testing in the Toxicity Characteristic Leaching Procedure (TCLP) laboratories as per EPA 1311 standard. The main objective of this thesis is to develop an unbiased sampling approach for the TCLP testing of PV modules. The TCLP testing was concentrated only for the laminate part of the modules, as they are already existing recycling technologies for the frame and junction box components of PV modules. Four different sample removal methods have been applied to the laminates of five different module manufacturers: coring approach, cell-cut approach, strip-cut approach, and hybrid approach. These removed samples were sent to two different TCLP laboratories, and TCLP results were tested for repeatability within a lab and reproducibility between the labs. The pros and cons of each sample removal method have been explored and the influence of sample removal methods on the variability of TCLP results has been discussed. To reduce the variability of TCLP results to an acceptable level, additional improvements in the coring approach, the best of the four tested options, are still needed.

Encapsulant is a key packaging component of photovoltaic (PV) modules, which protects the solar cell from physical, environmental and electrical damages. Ethylene-vinyl acetate (EVA) is one of the major encapsulant materials used in the PV industry. This work focuses on indoor accelerated ultraviolet (UV) stress testing and characterization to investigate the EVA discoloration and delamination in PV modules by using various non-destructive characterization techniques, including current-voltage (IV) measurements, UV fluorescence (UVf) and colorimetry measurements. Mini-modules with glass/EVA/cell/EVA/backsheet construction were fabricated in the laboratory with two types of EVA, UV-cut EVA (UVC) and UV-pass EVA (UVP).

The accelerated UV testing was performed in a UV chamber equipped with UV lights at an ambient temperature of 50°C, little or no humidity and total UV dosage of 400 kWh/m2. The mini-modules were maintained at three different temperatures through UV light heating by placing different thickness of thermal insulation sheets over the backsheet. Also, prior to thermal insulation sheet placement, the backsheet and laminate edges were fully covered with aluminum tape to prevent oxygen diffusion into the module and hence the photobleaching reaction.

The characterization results showed that mini-modules with UV-cut EVA suffered from discoloration while the modules with UV-pass EVA suffered from delamination. UVf imaging technique has the capability to identify the discoloration region in the UVC modules in the very early stage when the discoloration is not visible to the naked eyes, whereas Isc measurement is unable to measure the performance loss until the color becomes visibly darker. YI also provides the direct evidence of yellowing in the encapsulant. As expected, the extent of degradation due to discoloration increases with the increase in module temperature. The Isc loss is dictated by both the regions – discolored area at the center and non-discolored area at the cell edges, whereas the YI is only determined at the discolored region due to low probe area. This led to the limited correlation between Isc and YI in UVC modules.

In case of UVP modules, UV radiation has caused an adverse impact on the interfacial adhesion between the EVA and solar cell, which was detected from UVf images and severe Isc loss. No change in YI confirms that the reason for Isc loss is not due to yellowing but the delamination.

Further, the activation energy of encapsulant discoloration was estimated by using Arrhenius model on two types of data, %Isc drop and ΔYI. The Ea determined from the change in YI data for the EVA encapsulant discoloration reaction without the influence of oxygen and humidity is 0.61 eV. Based on the activation energy determined in this work and hourly weather data of any site, the degradation rate for the encaspulant browning mode can be estimated.

Bifacial photovoltaic modules are a relatively new development in the photovoltaic industry which allows for the collection and conversion of light on both sides of photovoltaic modules to usable electricity. Additional energy yield from bifacial photovoltaic modules, despite a slight increase in cost due to manufacturing processes of the bifacial cells, has the potential to significantly decrease the LCOE of photovoltaic installation. The performance of bifacial modules is dependent on three major factors: incident irradiation on the front side of the module, reflected irradiation on the back side of the module, and the module's bifaciality. Bifaciality is an inherent property of the photovoltaic cells and is determined by the performance of the front and rear side of the module when tested at STC. The reflected light on the back side of the module, however, is determined by several different factors including the incident ground irradiance, shading from the modules and racking system, height of the module installation, and ground albedo. Typical ground surfaces have a low albedo, which means that the magnitude of reflected light is a low percentage of the incident irradiance. Non-uniformity of back-side irradiance can also reduce the power generation due to cell-to-cell mismatch losses. This study investigates the use of controlled back-side reflectors to improve the irradiance on the back side of loosely packed 48-cell bifacial modules and compares this performance to the performance of 48 and 60-cell bifacial modules which rely on the uncontrolled reflection off nearby ground surfaces. Different construction geometries and reflective coating materials were tested to determine optimal construction to improve the reflectivity and uniformity of reflection. Results of this study show a significant improvement of 10-14% total energy production from modules with reflectors when compared to the 48-cell module with an uncontrolled ground reflection.

Performance of photovoltaic (PV) modules decrease as the operating temperatures increase. In hot climatic conditions, the operating temperature can reach as high as 85°C for the rooftop modules. Considering a typical power drop of 0.5%/°C for crystalline silicon modules, a performance decrease of approximately 30% could be expected during peak summer seasons due to the difference between module rated temperature of 25°C and operating temperature of 85°C. Therefore, it is critical to accurately predict the temperature of the modules so the performance can be accurately predicted. The module operating temperature is based not only on the ambient and irradiance conditions but is also based on the thermal properties of module packaging materials. One of the key packaging materials that would influence the module operating temperature is the substrate, polymer backsheet or glass. In this study, the thermal influence of three different polymer backsheet substrates and one glass substrate has been investigated through five tasks:

1. Determination and modeling of substrate or module temperature of coupons using four different substrates (three backsheet materials and one glass material).

2. Determination and modeling of cell temperature of coupons using four different substrates (three backsheet materials and one glass material)

3. Determination of temperature difference between cell and individual substrates for coupons of all four substrates

4. Determination of NOCT (nominal operating cell temperature) of coupons using all four substrate materials

5. Comparison of operating temperature difference between backsheet substrate coupons.

All these five tasks have been executed using the specially constructed one-cell coupons with identical cells but with four different substrates. For redundancy, two coupons per substrate were constructed and investigated. This study has attempted to model the effect of thermal conductivity of backsheet material on the cell and backsheet temperatures.

Solar photovoltaic (PV) deployment has grown at unprecedented rates since the early 2000s. As the global PV market increases, so will the volume of decommissioned PV panels. Growing PV panel waste presents a new environmental challenge, but also unprecedented opportunities to create value and pursue new economic avenues. Currently, in the United States, there are no regulations for governing the recycling of solar panels and the recycling process varies by the manufacturer. To bring in PV specific recycling regulations, whether the PV panels are toxic to the landfills, is to be determined. Per existing EPA regulations, PV panels are categorized as general waste and are subjected to a toxicity characterization leaching procedure (TCLP) to determine if it contains any toxic metals that can possibly leach into the landfill. In this thesis, a standardized procedure is developed for extracting samples from an end of life PV module. A literature review of the existing regulations in Europe and other countries is done. The sample extraction procedure is tested on a crystalline Si module to validate the method. The extracted samples are sent to an independent TCLP testing lab and the results are obtained. Image processing technique developed at ASU PRL is used to detect the particle size in a broken module and the size of samples sent is confirmed to follow the regulation.