guest ::
| Home > Publications database > Performance and stability of solar cells and modules: From laboratory characterization to field data analysis |
| Home > Publications database > Performance and stability of solar cells and modules: From laboratory characterization to field data analysis |
| Book/Dissertation / PhD Thesis | FZJ-2025-04777 |
2025
Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag
Jülich
ISBN: 978-3-95806-871-1
This record in other databases:
Please use a persistent id in citations: urn:nbn:de:0001-2602031059449.295647120559 doi:10.34734/FZJ-2025-04777
Abstract: The constant growth of the world population and the associated growing global energy demand represents a major challenge. The need to abstain nuclear and coal-fired power plants requires the rapid expansion of renewable energies and thus the expansion of photovoltaics (PV). In addition to increasing production capacity, improving the efficiency and sustainability of PV modules leads to higher yields. The resulting shorter payback times and the lower space required leads to increased economic interest and thus to a faster expansion of the PV industry. This thesis deals with several topics in the area of efficiency and sustainability of PV modules. Research to improve the efficiency of solar cells and solar modules mainly focuses on the cell efficiency. For the purpose of comparability, this is verified conventionally at a fixed temperature and irradiation. The defined standard test conditions (STC) ensure the comparability of efficiency measurements in different laboratories around the world, but are not representative for the expected efficiency or expected yield of a module in operation consisting of several cells. The first topic addressed in this thesis is the extrapolation of the expected yield of a PV module from the laboratory-measured efficiency of a single CIGS solar cell. On the root of the extrapolation model stand temperatureand irradiation-dependent IV measurements. Based on realistic assumptions, a corresponding module characteristic of a 100Wp module is extrapolated to each measured cell IV characteristic. To determine the yield of this module in realistic operating conditions, standard climatic reference profiles are used, which are defined in the IEC61853 standard. In addition, the extrapolation model can be used to determine how much yield potential there is in individual improvements to the cell performance. In addition to the temperature and irradiation dependent efficiency of solar modules, efficiency stability is crucial for the yield of a solar module that is in operation for several years. Long-term experiments of PV modules in operation cover all possible influences affecting the performance. However, the superposition of these many influences also means that individual influences are difficult to separate from one another. In order to investigate the effect of individual influences, laboratory tests are required in which the operating conditions can be adjusted and controlled. One influence on CIGS modules in operation is the illumination with (sun) light. To investigate the separate influence of illumination in more detail, an accelerated degradation experiment was set up in which a total of 24 flexible CIGS solar cells were illuminated under different operating conditions for approximately 1170 h. The operating temperature was varied between 25◦C and 70◦C and the irradiation intensity was varied between 0Wm−2 and 1000Wm−2. In addition, the influence of the operating bias was examined, with 12 cells kept in open circuit and 12 cells in short circuit. The results of the experiment show a strong dependency of the sustainability on the operating bias, with cells in short circuit showing high degradation rates. The process that leads to degradation is temperaturedependent (higher effect at higher temperatures) and mostly independent of the irradiation intensity. In addition, the analysis of the IV measurement datawith the one-diode model indicates an increased probability of a recombination process with a high ideality factor. A possible recombination process that would explain this development is Shockley-Read-Hall (SRH) recombination in the quasi-neutral region of the p-n junction. While laboratory experiments can lead to valuable insights on the effect of accelerated conditions, long-term experiments under realistic operating conditions are ultimately the best way to depict all influences affecting a PV modules efficiency in outdoor operation. In addition to measuring the module performance (often IV characteristics), this also requires the collection of eteorological data (e.g. module temperature and irradiation). Since the environmental influences are not controlled in these long-term experiments, the data structures recorded in this way are prone to errors. To this end, an analysis of the data requires a classification of the data quality and an appropriate filter before the analysis. In this work, two possible filter methods are developed and discussed. The first approach evaluates on the plausibility, which is classified using physical models describing the relationships between the different dimensions of the data recorded. The correlation of several occurring deviations is weighted using the Mahalanobis distance. The second filter method specifically aims to address a systematic error in the data often present in outdoor data, which occurs due to partial shading of the module and/or the irradiation sensor. Irradiation and operating temperature dependencies of the short circuit current of a PV module are described using several Gaussian process regressions (GPRs, a statistical method for interpolating data) and outliers from measurement and expectation are iteratively filtered out. After the data quality has been assessed and an appropriate filter has been applied, the data can be further analysed. The two biggest challenges are the dimensionality and size of the data structures as well as the irradiation and temperature dependencies, which needs to be taken into account rating the performance. To reduce the dimensionality of the data, I use the socalled extended solar cell parameters (ESPs), an extension of the standard solar cell parameters ISC, VOC, IMPP and VMPP on a total of 10 parameters, which describe the complete form of an IV characteristic. Furthermore, a principal component analysis (PCA) is used, in which a new coordinate system is chosen for the 10 ESPs and linear correlations between the ESPs are thus separated. The 10 principal components (PCs) determined in this way are linearly uncorrelated and are described with several Gaussian processes that use the module temperature, irradiation and the time as input. In order to be able to process the amount of data, the data set for each individual PC is divided into monthly subsets and individual Gaussian process regressions are trained. The model can thus reproduce the 10 PCs, consequently the 10 ESPs and the complete IV characteristic for any input of temperature, irradiation and time. In addition, the Gaussian process regressions provide information about the uncertainty of the output, which arises from the measurement uncertainty and the distance (in time, irradiation and temperature) to the actual measurements. The versatility of possible applications of the model is underlined with 3 examples (a simple representation of the time series to visualize seasonality, long-term degradation and acclimatization, the comparison to a classical performance ratioanalysis and an application of the one-diode model based on the output of the GPRs). In addition, an analysis of the model’s uncertainty shows high accuracy and good agreement between real and predicted error distributions.
|
The record appears in these collections: |
PDF
Fulltext by OpenAccess repository