Accuracy investigation in the modeling of partially shaded photovoltaic systems

Abstract

Software for simulation of photovoltaic (PV) systems is widely used for dimensioning and forecasting electrical production. A factor of losses in PV installations is the partial shading caused by surrounding elements, and these software allow the user to estimate this effect. However, the accuracy of these simulated results for shaded PV systems is not widely studied. The purpose of this article is to investigate the accuracy and quantify the differences between simulated and measured data of partially shaded PV systems, obtained with the widely used tools SAM and PVSyst. Measured data from a PV installation were compared to results from simulations performed using the different shading calculation options available in both tools. The simulated outputs were both underestimated and overestimated in the shading situations. This variation was related to the use of an hourly fraction of shading and, in the case of SAM, due to the limitations of the 3D tools available for representation. Another source of differences between simulated and measured values was the use of uniform shading factors for diffuse and albedo. In addition, the simplification of the 3D model had a significant impact on the predicted energy, mainly on cloudy days. Both software overestimated the electricity production for the entire measurement period, reaching differences between the predicted and the measured energy varying from 9% to 24%. Shaded PV systems must be carefully analyzed, and the simulated results may differ from the measured values, which may even influence the decision on the feasibility of an installation.