A new approach to comparing photovoltaic simulation software

In this work, a new approach to comparing photovoltaic simulation tools is proposed. It combines so-called cross-validation analysis along with a comparison to measured data from the real PV system. PV*SOL and PVsyst, as two professional tools widely used among PV professionals, are chosen for comparison. Cross-validation was applied to 83 European locations to account for geographical and climate dependence of simulation accuracy. The locations are divided into five groups according to Koppen-Geiger climate classification. Statistics of variation between simulation results are presented. Both seasonal and variations between locations with different climates are studied. The results showed that PVsyst tends to estimate higher plane-of-array radiation, but the difference in estimated energy is smaller. This difference is more pronounced for locations with hot and dry climates. Overall, the difference in monthly values between simulation tools for most locations remains within +/- 10%. Comparison with measurement was presented for a 5-year operation period of 31 kWp grid-connected PV system installed on the Faculty of Electrical Engineering, Mechanical Engineering, and Naval Architecture in Split, Croatia. Booth tools accurately predicted annual and monthly yield, with nRMSE less than 3.2% for the complete period under study.