Abstract
Many tools implementing Radiative Transfer Equation solvers are used nowadays for the calculation of spectral irradiance. Although many studies proved their accuracy, few of them focused on the uncertainty of these models. This study evaluates how the uncertainty of the input parameters of Radiative Transfer solver SDISORT propagates into the generated spectra and how this in turn affects the uncertainty of two photovoltaic (PV) device calibration parameters: short circuit current (Isc) and spectral mismatch factor (MM). In particular, both the propagation of single input parameter uncertainty and of their combination are evaluated, as well as the definition of possible maximum and minimum uncertainty levels referring to input combinations that differ from a defined reference set.
The analysis considers the spectral range 280 nm - 2500 nm and the spectral response of seven different photovoltaic technologies. Monte Carlo technique is used as a statistical method for evaluating uncertainty propagation, by assuming a rectangular probability density function of SDISORT input parameters as well as defined uncertainty levels. Results show that the lowest uncertainty variability occurs for mono-crystalline silicon technology, the highest for organic technology.