A study on the effect of thin object shading on the performance of photovoltaic modules

Abstract

This study investigates a specific and underexplored aspect of partial shading of photovoltaic (PV) systems that is caused by thin horizontal objects. Following a review of both fundamental principles and current state-of-the-art, the primary aims were to examine differences in shadow formation and associated power loss between laboratory and outdoor environments, quantify shadow formation characteristics, and to assess the scientific validity of commonly used simplifications in representing penumbra shadows. This study presents a comparative analysis of thin object shadows under laboratory environment conditions using a solar simulator, and under outdoor environmental conditions. Using a 156.0 × 156.0 mm solar cell, 56 thin object shading cases were compared using thin objects of thicknesses of ⌀2.0 mm - ⌀20.0 mm at distances between 10 cm – 40 cm in 5 cm increments. The solar cell placed in outdoor conditions produced higher power losses in the range of 0.34 W - 0.67 W, equivalent to 8.18% to 16.13% power loss respectively, when converted to standard test conditions (STC). This observation was validated by producing a quantitative factor which incorporated both the size and intensity of the overall shadow at equal weight where clearly the outdoor environment produced stronger shadow formation. Moreover, the image capture of all 56 cases found evidence of the laboratory environment producing fringed shadow patterns caused by Fresnel diffraction on the thin objects, most evidently in the shadows of the smallest thin object (⌀2 mm). This thesis also implemented a novel methodology to quantify the shadow formation and the power loss of 104 thin object shading cases, using custom made image analysis tool named ThinShadePV. This tool utilises image analysis and Random Forest machine learning to determine the size and intensity of umbra and penumbra shadow regions from captured shadow images and predict the associated power losses. Validation through 15 real-world thin shadow cases demonstrated strong agreement between predicted and actual power loss values, with a mean deviation of just 0.48% and an accuracy range of –3.68% to +2.71%. Moreover, Spearman correlation and sensitivity analyses revealed that penumbra intensity is the most influential factor for smaller objects (⌀2.8–⌀8.0 mm thick), while the size of the umbra shadow dominates in larger objects (⌀10–⌀16.0 mm) thick. The study also assessed the validity of approximating varying-intensity penumbra shadows with constant-intensity penumbra representations using neutral density (ND) filters. While constant-intensity shadows closely replicated the relationship profile, they consistently overestimated power loss, despite matching the size and average varying intensity shadows. A large effect size (Cohen’s d = –0.893) confirmed a significant discrepancy, indicating that varying-intensity penumbra shadows should not be simplified to constant-intensity models in analytical or simulation contexts. One of the key contributions of this study lies in its significant advancement of current knowledge by empirically quantifying the effects of horizontal thin object shading on PV performance and establishing key relationships between shading parameters and power loss. The development of ThinShadePV, a purpose-built image analysis tool, enables quantitative measurement of shadow characteristics and power loss prediction. Moreover, ThinShadePV holds strong potential for integration into commercial PV simulation software, enhancing the accuracy of performance forecasts under real-world shading conditions.