Search

Degradation and partial shading impact the long-term reliability and power production of photovoltaic (PV) modules and power plants. Time-series power ( P mp ) and current–voltage (I-V) curve datastreams from PV modules enable a remote diagnostic approach to quantify active degradation mechanisms and identify partial shading. We study three to nine years of these datastreams, including 3.6 million I-V curves and 36 million P mp values, from eight PV modules, four each of double-glass and glass-backsheet module architectures, located in three distinctly different Koppen-Geiger climate zones, to determine the module’s performance loss rates (PLR), identify active degradation mechanisms and power loss modes, along with partial shading by local objects. Considering both module architectures, PLR results indicate that the BSh climate zone is the most aggressive for module degradation, while the Alpine ET zone is the mildest climate. PLR of double-glass modules located in BWh and BSh climate zones are different due to the significantly greater uniform current loss (ΔPIsc) for double-glass modules in BSh, at a 5% significance level. Power loss for four out of five modules located in the BWh and BSh climates are dominated by uniform current degradation. Statistical analysis of multistep I-V curves detects partial shading experienced by three studied modules with details of the shading profile, the shading Poynting vector diagram for the obstacle’s relative position, shading scenarios, and duration. This work demonstrates how remote monitoring and diagnosis of P mp & I-V time-series of modules can provide quantitative operations and maintenance insights into system performance, degradation mechanisms, and shading.

Weibull analysis and weakest link theory are employed to resolve the probability of crystalline silicon PV cell fracture when measured as bare cells and when stressed in reduced- and full-sized modules. Experimental results indicated that the characteristic cell strength is reduced by ∼20% once packaged into the laminate of a one-cell module and loaded in four-point flexure (4PF). This experimental observation was shown consistent with a weakest link theory prediction that the strength limiting flaws reside on the surface of the cell's edge. The analysis is ultimately extended to present the equivalent loading of four-cell modules by uniform pressure and 4PF and a uniformly loaded full-sized module and demonstrates that smaller, representative, modules must be loaded to a much higher level than their parent full-sized modules to achieve an equivalent probability for cell fracture.

The width tapered beam method for measuring fracture toughness has been contemplated for use in measuring the adhesion of photovoltaic materials, is promoted by it being a viable method for use dir...

Utilizing large-scale outdoor $I$ – $V$ data streams, the authors demonstrated a feasible, repeatable method to construct outdoor time-series $I_{\text{sc}}$ – $V_{\text{oc}}$ curves. Resulting Analytic $I_{\text{sc}}$ – $V_{\text{oc}}$ curves are validated with laboratory $Suns$ - $V_{\text{oc}}$ measurements. Comparing $I_{\text{sc}}$ – $V_{\text{oc}}$ with real $I$ – $V$ curves, power loss modes including uniform current loss, recombination loss, series resistance loss, and current mismatch loss can be quantified in units of watts or percent nominal power, making them directly comparable. Above power loss modes are calculated for long-term $I$ – $V$ data from fielded modules in different climate zones, and dominant power loss modes were determined from the resulting time series. This method has been published as an open source package in R programming language.

Electroluminescence (EL) imaging of photovoltiac (PV) modules offers high-speed, high-resolution information about device performance, affording opportunities for greater insight and efficiency in module characterization across manufacturing, research and development, and power plant operations and management. Predicting module electrical properties from EL image features is a critical step toward these applications. In this article, we demonstrate quantification of both generalized and performance mechanism-specific EL image features, using pixel intensity-based and machine learning classification algorithms. From EL image features, we build predictive models for PV module power and series resistance, using time-series current–voltage ( I–V ) and EL data obtained stepwise on five brands of modules spanning three Si cell types through two accelerated exposures: damp heat (DH) (85 $^\circ$ C/85% RH) and thermal cycling (TC) (IEC 61215). In total, 195 pairs of EL images and I–V characteristics were analyzed, yielding 11 700 individual PV cell images. A convolutional neural network was built to classify cells by the severity of busbar corrosion with high accuracy (95%). Generalized power predictive models estimated the maximum power of PV modules from EL images with high confidence and an adjusted- $R^2$ of 0.88, across all module brands and cell types in extended DH and TC exposures. Mechanistic degradation prediction was demonstrated by quantification of busbar corrosion in EL images of three module brands in DH, and subsequent modeling of series resistance using these mechanism-specific EL image features. For modules exhibiting busbar corrosion, we demonstrated series resistance predictive models with adjusted- $R^2$ of up to 0.73.

Backsheet cracking has been a major issue observed in the field; however, standardized qualification tests, such as IEC61215, are inadequate to reliably identify such failures of PV modules due to the lack of the critical weathering factors applied sequentially or in combination, such as those found in the service environments. To address this problem, we investigated the effects of various environmental variables on the degradation and failure behaviors of the polyamide-based backsheet in PV modules retrieved from five different locations, encompassing a variety of climates, including humid subtropical, hot-summer Mediterranean, tropical savanna climate and hot arid. The correlations between the degradation indicators and the weathering variables were further demonstrated by principle components analysis (PCA). We found strong relationships between: carbonyl formation and reflected solar radiation; hydroxyl formation and module temperature; yellowness and NO2 concentration, while no simple correlation could be found between a specific weathering factor and cracking. By introducing additional stress factors to the aged polyamide-based backsheet films with the novel “fragmentation test”, we successfully reproduced the field cracking behaviour. This study has demonstrated that different degradation modes of PV components respond differently to the environmental stresses encountered in service. Thereby, any accelerated laboratory test based on a single set condition or lacking key environmental variables would be inadequate to assess the long-term performance of PV modules and components. A new reliability-based methodology is proposed to quantitatively link laboratory testing with field results for the service life prediction of PV materials.

Rigorous statistical analysis of whole building, 15-minute interval, time series electricity data enables remote insights into buildings’ operational characteristics. We developed select building m...

Data science has advanced significantly in recent years and allows scientists to harness large-scale data analysis techniques using open source coding frameworks. Data science is a tool that should be taught to science and engineering students in addition to their chosen domain knowledge. An applied data science minor allows students to understand data and data handling as well as statistics and model development. This move will improve reproducibility and openness of research as well as allow for greater interdisciplinarity and more analyses focusing on critical scientific challenges.

In research on photovoltaic (PV) device degradation, current–voltage ( $I$ – $V$ ) datasets carry a large amount of information in addition to the maximum power point. Performance parameters such as short-circuit current, open-circuit voltage, shunt resistance, series resistance, and fill factor are essential for diagnosing the performance and degradation of solar cells and modules. To enable the scaling of $I$ – $V$ studies to millions of $I$ – $V$ curves, we have developed a data-driven method to extract $I$ – $V$ curve parameters and distributed this method as an open-source package in R. In contrast with the traditional practice of fitting the diode equation to $I$ – $V$ curves individually, which requires solving a transcendental equation, this data-driven method can be applied to large volumes of $I$ – $V$ data in a short time. Our data-driven feature extraction technique is tested on $I$ – $V$ curves generated with the single-diode model and applied to $I$ – $V$ curves with different data point densities collected from three different sources. This method has a high repeatability for extracting $I$ – $V$ features, without requiring knowledge of the device or expected parameters to be input by the researcher. We also demonstrate how this method can be applied to large datasets and accommodates nonstandard $I$ – $V$ curves including those showing artifacts of connection problems or shading where bypass diode activation produces multiple “steps.” These features together make the data-driven $I$ – $V$ feature extraction method ideal for evaluating time-series $I$ – $V$ data and analyzing power degradation mechanisms in PV modules through cross comparisons of the extracted parameters.

An automated data analysis pipeline is developed to preprocess electroluminescence (EL) module images, and parse the images into individual cells to be used as an input for machine learning algorithms. The dataset used in the study includes EL images of three 60 cell modules from each of five commercial brands at six steps of damp heat exposure, from 500 to 3000 h. Preprocessing of the original raw EL images includes lens distortion correction, filtering, thresholding, convex hull, regression fitting, and perspective transformation to produce planar indexed module and single cell images. Parsing of PV cells from each of the preprocessed 90 EL module images gives us 5400 cell images, which are function of module brand and damp heat exposure step. From the dataset, two unique degradation categories (“cracked” and “corroded”) were observed, while cells that did not degrade were classified as “good.” For supervised machine learning modeling, cell images were sorted into these three classes yielding 3550 images. A training and testing framework with 80:20 sampling ratio was generated using stratified sampling. Three machine learning algorithms (support vector machine, Random Forest, and convolutional neural network) were trained and tuned independently on the training set and then given the test set to predict the scores for each of the three models. Five-fold cross validation was done on training set to tune hyper-parameters of the models. Model prediction scores showed that convolutional neural network outperforms support vector machine and Random Forest for supervised PV cell classification.

Photovoltaic (PV) module backsheets degrade at different rates because of the specific polymeric materials and the local exposure environments of the installations. Studies of real-world backsheet degradation provide valuable information to understand backsheet degradation and failure. Field surveys of PV module backsheets were conducted on 1310 modules in four commercial PV power plant sites with different exposure times. The backsheet's local exposure environment is determined by its location along the rack length and depth, the modules’ elevation above the ground, and the ground cover albedo. Backsheets that are installed at the ends of the module rack length exhibit larger degradation rates (with yellowness index difference of 1.14 $\pm$ 0.45 and 7.80 $\pm$ 1.3 for backsheets of poly (ethylene terephthalate) and poly (ethylene naphthalate) as the air-side layer, respectively) than backsheets in the center section of the rack. A generalized spatio-temporal model was developed to predict the large-scale backsheet degradation of different backsheet polymers across time/age and location in the PV power plant rack. The model utilizes a cubic-spline relationship between backsheet degradation and rack length, and a quadratic relationship between backsheet discoloration and rack depth. This generalized spatio-temporal model predicts the outdoor backsheet degradation with an adjusted- $R^2$ range between 0.31 to 0.89. The similarity between this model and the spatial variation of the rear-side irradiance, indicates that the irradiance plays a significant role in outdoor backsheet degradation. The generalized spatio-temporal model can be used to evaluate large scale PV backsheet performance, as well as a guidance for PV site designers and operations and maintenance crews.

Heritage data for the class of 9 to 12 wt pct Cr steels are studied using data science to quantify the statistically significant relationships among multiple processing/microstructure and performance variables. The effort is undertaken to find new martensitic steels for creep life of $$10^5{{\text { hours}}}$$ or greater at 650 °C and 100 MPa using machine learning. Linear regression and lasso regression were utilized to identify alloying elements that contribute towards better creep strength. Visualization techniques such as t-distributed stochastic neighbor embedding and pair-wire element specific comparisons were utilized to explore information gaps that exist within the data and are in conflict with existing domain knowledge. Combining all results suggest that the next alloy design to be explored should be 9 wt pct Cr with high W (2 to 3 wt pct) and high Co (2 to 3 wt pct) for creep life of $$10^5\,\,{\text { hours}}$$ or greater at 650 °C, 100 MPa.

Many research show that UV degrades the solar cell and module power. In this work, we study the UV stability of mc-Si bifacial PERC solar modules with different industrial silicon nitride (SiNx) passivation. We find that with exposure to UV at 60 °C, both VOC and JSC decrease then stabilize. The quantum efficiency analysis shows decreased response in both short (blue loss) and long (base collection loss) wavelength range. It infers that deteriorations were not only at passivation layers, but also in the silicon bulk, caused by UV-induced degradation and possible LeTID. No significant difference in degradation has been found between front and rear passivation, or between different PECVD, respectively. While thermal oxidation process employed in mass production effectively improved UV resistance.

The long-term reliability of photovoltaic (PV) modules is essential to decrease the levelized cost of electricity and is dependent on module packaging choices. In this paper, we study the degradation of double glass (DG) and glass-backsheet (GB) PV modules with ethylene-vinyl acetate (EVA) and polyolefin elastomer (POE) encapsulants using multicrystalline PERC cells under accelerated exposures including modified damp heat (mDH) and mDH with full-spectrum light (FSL). The results showed that the modules with opaque rear encapsulant have greater power loss on average than those with UV-cutoff rear encapsulant for each module type. The dominant degradation mechanism was series resistance (R s ) increase indicating interconnect corrosion for most module types. In addition to the increased R s , GB modules with UV-cutoff rear encapsulant experienced power loss by transmission loss, and the POE_GB type under mDH+FSL also had more cell shunting. For modules with opaque rear encapsulant, the POE_DG type under mDH+FSL had power loss dominated by transmission loss.

Assessing photovoltaic module backsheet durability is critical to increasing module lifetime. Lab based accelerating testing has recently failed to predict large scale failures of widely adopted polymeric materials. Field surveyed data is critical to assess the performance of component lifetime. Using a documented field survey protocol, 13 field surveys where conducted. Each measurement is encoded with it’s spatial location in respect to the other modules. By combining field survey data on degradation predictors with real time satellite weather data, data-driven predictive models of backsheet degradation were trained. LOESS models were constructed to investigate the spatial dependence of measurements. It was found that micro-climatic effects like tree-lines, ground surface changes, and elevation changes effected the magnitude and variance of the measurements. A GAM model was created to predict the value of degradation based on measured predictors. The model includes variables on the climate of the system and the location of each measurement in the PV mounting structure. The model performed well with an adj.R2 of 0.95 for yellowness index prediction. The model was cross-validated using k-folds.

To assess the reliability of PERC cells compared to Al-BSF in a commercial setting minimodules with cell and encapsulant combinations are compared in accelerated exposure. In both modified damp heat and modified damp heat with full spectrum light exposures, white EVA samples showed a higher susceptibility for metallization corrosion degradation than all other encapsulants. Al-BSF cells in particular showed higher power loss than PERC cells with white EVA. It was observed that the degree of degradation had a strong significance on the manufacturer of the white EVA encapsulant. In both exposures the encapsulant was a much stronger predictor of degradation than cell type. For modules with the same encapsulant, PERC cells showed the higher performance or were comparable to Al-BSF cells for all but one case.

Network structural equation modeling has been used for degradation modeling of glass/backsheet (GB) and double glass (DG) PERC PV minimodules, made by CSI and CWRU. The encapsulants used were ethylene vinyl acetate (EVA) and polyolefin elastomer (POE). The exposures included modified damp heat (80°C and 85% relative humidity), with and without full spectrum light. Each exposure cycle consists of 2520 hours, 5 steps of 504 hours each. The data from I-V and Suns-V oc was used in the analysis. We observe that most DG minimodules exhibit stability in power with exposure time and GB minimodules by CWRU showed a power loss of 5-6% on average due to corrosion.

In this work, we explore the influence of module size on the rate of interconnect solder bond thermomechanical fatigue (TMF) damage and the probability of cell fracture. For the solder bond TMF damage evaluation, structural mechanics models of crystalline silicon PV models are created to solve with the Finite Element Method. For the probability of cell fracture evaluation, Weibull analysis and weakest link theory are employed to resolve the probability of crystalline silicon PV cell fracture when measured as bare cells and when stressed in reduced- and full-sized modules. Results conclusively demonstrate that the rate of solder bond TMF damage is independent of module size, interconnect location across the cell and cell location across the module and that smaller, representative, modules must be loaded to a much higher level than their parent full-sized modules to achieve an equivalent driving force for cell fracture.

We present the application of FAIR principles to photovoltaic time series data to increase their reusability within the photovoltaic research community. The main requirements for a "FAIRified" dataset is to have a clearly defined data format, and to make accessible all metadata for this dataset to humans and machines. To achieve FAIRification, we implement a data model that separates the photovoltaic data and its metadata. The metadata and their descriptions are registered on a data repository in a human and machine readable format, using JSON-LD. Also, secure APIs are developed to access photovoltaic data. This approach has long term scalability and maintainability.

Cell cracking in PV modules can lead to a variety of changes in module operation, with vastly different performance degradation based on the type and severity of the cracks. In this work, we demonstrate automated measurement of cell crack properties from electroluminescence images, and correlate these properties with current-voltage curve features on 35 four-cell Al-BSF and PERC mini-modules showing a range of crack types and severity. Power loss in PERC modules was associated with more total crack length, resulting in electrical isolation of cell areas and mild shunting and recombination. Many of the Al-BSF modules suffered catastrophic power loss due to crack-related shunts. Mild power loss in Al-BSF modules was not as strongly correlated with total crack length; instead crack angles and branching were better indicators of module performance for this cell type.

The open circuit voltage (VOC) is a critical and common indicator of solar cell performance as well as degradation, for panel down to lab-scale photovoltaics. Detecting VOC at the nanoscale is much more challenging, however, due to experimental limitations on spatial resolution, voltage resolution, and/or measurement times. Accordingly, an approach based on Conductive Atomic Force Microscopy is implemented to directly detect the local VOC, notably for monocrystalline Passivated Emitter Rear Contact (PERC) cells which are the most common industrial-scale solar panel technology in production worldwide. This is demonstrated with cross-sectioned monocrystalline PERC cells around the entire circumference of a poly-aluminum-silicide via through the rear emitter. The VOC maps reveal a local back surface field extending ~ 2 μm into the underlying p-type Si absorber due to Al in-diffusion as designed. Such high spatial resolution methods for photovoltaic performance mapping are especially promising for directly visualizing the effects of processing parameters, as well as identifying signatures of degradation for silicon and other solar cell technologies.

PVplr is an R package designed with the intent to offer a variety of options and side-by-side comparisons when modeling outdoor PV time-series data for performance loss rate (PLR) analysis. Built as a part of the IEA-PVPS task 13 study on the determination and uncertainty of PLR calculation, the package is designed to not to run a single method but to compare multiple commonly used methods on the same systems to determine system stability and identify biases in the analysis. The workflow is designed as a pipeline with steps of data cleaning, weather correction, time-series processing, and PLR determination, with multiple options at each step replicating commonly used methods or specific process created by other groups. Non-linear PLR evaluation capabilities have been added using piecewise linear modeling. Implementation of this pipeline in prior work has shown when 40 unique assumed linear PLR values are calculated for individual systems results can show significant variance depending on the steps chosen in data processing and modeling, highlighting the biases that can be induced by differences in analysis processes. Future updates to the package are planned that add more options to each step of the pipeline and allow python and R integration so python packages such as RdTools and PVlib-python can be used in conjunction with PVplr.

Cell cracking in PV modules can lead to a variety of changes in the modules operation, with vastly different performance degradation based on the type and severity of crack. In this work, we correlate cell crack metrics in images with current-voltage (I-V) curve features on a sample set of 38 four-cell Al-BSF and PERC mini-modules showing a range of fracture and electrical properties. Impacts of cracking on electrical performance demonstrated in this work include cell shunting and electrical isolation of cell regions, shown with electroluminescence (EL) images and I-V tracing. Cracks and other EL image properties are quantified with algorithmic computer vision techniques, and correlated with I-V properties at the cell and module levels.

In this paper, we study the degradation behavior of glass-backsheet (GB) and double glass (DG) multicrystalline silicon monofacial PERC minimodules, fabricated using polyolefin elastomer (POE) and ethylene vinyl acetate (EVA) encapsulants, undergoing exposures of modified damp-heat (80°C / 85% RH), with and without full spectrum light. The completed exposure time, at this time, is 1512 hours with each step of 504 hours. Stepwise measurements are conducted throughout exposures including current-voltage (I-V) curves, $Suns-V_{oc}$ , electroluminescence (EL) images, and four-point proof loading with EL for tracking changes in mechanical properties in encapsulants by cell fracture probability. The results show that GB minimodules of both encapsulants, compared to DG minimodules, experience more changes in features of I-V and $Suns-V_{oc}$ measurements. There is no development of PV cell cracking during proof loading as seen from EL images and the load versus load line displacement plots currently.

In this work we demonstrate how fracture statistics may be employed to track cell reliability through module fabrication steps. We were able to detect how cell cutting equipment could reduce cell fracture strain by 25 % as well as increase its variability. A smaller reduction was found to result from the ribbon soldering operation, and all cells measured were weaker when strained in an orientation perpendicular to the gridlines. Cell fracture measurements were found to be consistent with the number of cracked cells experienced in full sized modules exposed to mechanical loading. A structural mechanics model was created to elucidate how the requisite fracture strain is developed in a cell encapsulated within a PV module and demonstrates that this strain decreases more than 50 % from its center to edge when placed within the inner loading span of four-point flexure. The associated implications on the apparent cell strength is examined with an effective volume treatment.

Photovoltaic (PV) power is a critical resource for meeting future energy demands. As integration of photovoltaic systems increases it is important to categorize climate parameters by conditions that are known to affect the viability of PV. A widely used classification is Koppen Geiger Climate zone which divides the world into five main climates and subcategories based on humidity and average temperature. Despite Koppen Geiger climate classifications widespread popularity it was not designed with parameters that determine the performance or degradation of modules. This paper presents an evaluation of the Koppen Geiger climate classification using real-world weather data from 357 PV sites. Two types of clustering algorithms were applied. The clustering algorithms had difficulty distinguishing between temperature categories. Furthermore, machine learning algorithms were used to predict the climate zone classification. The randomforest model had the highest prediction accuracy of 76%. Both clustering and machine learning failed to capture the seasonality of data resulting in poor performance.