A Method for Improving the Accuracy of Numerical Simulations of a Photovoltaic Panel

Abstract Numerical simulations of photovoltaic solar panels are performed using temperature-dependent layer properties. The results are compared with experimental data recorded from a 50 W mono-crystalline panel and a 50 W poly-crystalline panel. The comparison shows that, for both panels, introducing temperature dependencies in the layer properties can significantly improve the accuracy of numerical simulations. On a sample day in August 2019, the mean absolute error in power prediction is found to decrease from 9.13 to 4.32% for the mono-crystalline panel and from 9.49 to 5.55% for the poly-crystalline panel, representing accuracy improvements of 52.7% and 41.5%, respectively. On an annual basis, the accuracy of estimating the power generated by the mono- and poly-crystalline panels improves by 52.8% and 41.4%, respectively. Finally, it is found that as the standard deviation of the temperature distribution on the panel increases, so does the effect of the temperature-dependent layer properties. This study highlights the need to account for the temperature dependencies of the different layer properties when numerically simulating photovoltaic panels.