Quantitative assessment of PV modules affected by Type-C cracks using electroluminescence imaging

Abstract

Cracks in photovoltaic modules are among the defects that can lead to the highest power losses. While the qualitative analysis of such failures has been widely studied, quantitative assessments remain relatively scarce. Accurately estimating the power loss associated with these defects is therefore essential to assess module performance, guide decision-making in maintenance activities and ensure long-term photovoltaic system performance. This paper presents the application of a methodology to quantify Type-C crack severity across the cells of a sample of 100 photovoltaic modules from two different manufacturers that have been in operation for 11 years. The methodology combines automatic crack segmentation of electroluminescence images with a quantitative assessment of the power losses associated with Type-C cracks. The statistical analysis of the sample indicates that the best probability distribution fitting the power loss estimation results is the Generalized Extreme Value distribution, with a mean value of 12.77%. This suggests a slightly higher average power loss than expected according to the acceptance/rejection criteria established, reflecting that the sample exhibits slight degradation while remaining operational. Furthermore, the shape of the resulting distribution suggests that it is unlikely to observe modules exceeding the 20% of power loss. The approach used to estimate power loss highlights the importance of detecting the distribution of cracks across the photovoltaic module, as concentrated Type-C cracks in a cell can lead to significant power losses for the module. The proposed approach enables proactive maintenance by identifying severely cracked modules for replacement, improving plant performance.