Statistical and Domain Analytics Applied to PV Module Lifetime and Degradation Science

Abstract

A better understanding of the degradation modes and rates for photovoltaic (PV) modules is necessary to optimize and extend the lifetime of these modules. Lifetime and degradation science (L&DS) is used to understand degradation modes, mechanisms and rates of materials, components and systems to predict lifetime of PV modules. A PV module lifetime and degradation science (PVM L&DS) model is an essential component to predict lifetime and mitigate degradation of PV modules using reproducible open data science. Previously published accelerated testing data from Underwriter Laboratories on PV modules with fluorinated polyester backsheets, which included eight modules that were exposed up to 4000 hrs of damp heat (85% relative humidity at 85 • C) and eight exposed up to 4000 hrs of ultraviolet light (80 W/m 2 of 280-400 nm wavelengths at 60 • C) (UV preconditioning) were used to determine statistically significant relationships between the applied stresses and measured responses. There were 15 different variables tracking aspects of system performance, degradation mechanisms, component metrics and time. Modules were analyzed for three system performance metrics (fill factor, peak power, and wet insulation). The results were statistically analyzed to identify variable transformations, statistically significant relationships (SSRs) and to develop the PVM L&DS model informed by a generalization of structural equation modeling techniques. The SSRs and significant model coefficients, combined with domain analytics, incorporating materials science, chemistry, and physics expertise, produced a pathway diagram ranking the variables' impact on the system performance, which were iteratively examined using sound statistical analysis and diagnostics. The SSRs determined from the damp heat exposure for the system response of Pmax corresponded to the degradation pathway of polyester terephthalate (PET) and ethylene vinyl acetate (EVA) hydrolysis. A linear change point for the damp heat exposure with the system response of Pmax was determined to be 1890 hrs. The UV preconditioning exposure did not induce sufficient degradation shown by the quality of the R 2 values for many of the best fitting models. This exemplifies the development of a methodology to determine rank ordered lifetime and degradation pathways present in modules and their effects on module performance over lifetime.

Degradation pathway models for photovoltaics module lifetime performance

“…5. The statistical pathway model generated using Principle 1 for the variables T ime, Hac, IR2, IRBS1, and P max was found to contain the most information content and mapped well to domain knowledge [26].…”

“…Of the 32 possible combinations of the chosen variables that were modeled with each of the two previously described principles of variable selection, the Principle 1 modeling technique applied to the variables T ime, Hac, IR2, IRBS1, and P max was found to possess the most information content and mapped well to domain knowledge [26]. A full prognostic model was not developed due to some limitations in the data set, though valuable observations about PV module response to damp heat testing conditions were evident from the statistical analysis.…”

“…The second principle examines the collective additive influence of variables upon one another utilizing Akaike Information Criterion (AIC) values to perform stepwise variable selections iteratively to indicate networks of relationships between the variables [26].…”

Statistical and domain analytics for informed study protocols

“…The first principle examines only univariate relationships based on a selection of known functional forms directly from one variable to another, and performs iterative variable selection steps on the basis of a 0.2 cutoff of adjusted-r-squared value. The second principle examines the collective additive influence of variables upon one another utilizing Akaike Information Criterion (AIC) values to perform stepwise variable selections iteratively to indicate networks of relationships between the variables [27].…”

Insights into metastability of photovoltaic materials at the mesoscale through massiveI–Vanalytics

“…Considerable research has already been published on observed degradation modes or mechanisms, but it is not clear how these mechanisms interact or how the network of mechanistic degradation pathways, acting in parallel or series, actually produce the degradation in performance. 39 Reliability research efforts conducted on complete modules are typically conducted in the laboratory under accelerated aging conditions, exposing the module to high doses of heat/ humidity and/or light well in excess of the extremes of the real world. Relative humidity of 85 C and 85% is a typical exposure condition for accelerated aging.…”