Analysis of discrepancies between simulated and actual energy production for a photovoltaic system in Norway

The purpose of this study was to investigate and analyze parameters that influence the discrep- ancy between simulated and actual energy production for a photovoltaic (PV) system in south- eastern Norway for the year 2015. The aim was to derive solutions for how simulation mod- els should be designed and how to treat loss parameters, by analyzing and discussing possible causes and solutions for deviations. The simulation software PVsyst was used to simulate differ- ent scenarios in order to investigate how the different parameters influenced predicted energy production. To increase the coherency between simulated and produced energy is important in the planning phase of a PV system, and the establishment of a reliable PV system that will meet the required energy demand. Meteorological (meteo) data from weather stations and satellite data was used as input meteo in PVsyst to investigate any potential differences. Satellite data sources were Meteonorm and Meteocontrol, while weather station data came from Ås and ASKO (Vesby). The PV system was carefully designed in PVsyst according to system configuration and layout. The potential loss parameters were estimated, based upon theory and analysis of climate data from the different meteo sources. Different simulation scenarios were conducted in PVsyst to analyze the accu- racy of the different meteo data, and the influence of adjusting loss parameters. Results show that the accuracy of Global Horizontal Irradiation (GHI) data is vital for the coher- ence between simulated and produced energy. The sources with the least deviation in irradia- tion data from the reference value, resulted in the closest estimate to produced energy. Satellite collected data underestimated GHI about 3-4 %, while meteo data from ASKO underestimated 0,6 % and Ås overestimated 1,8 %. Monthly Root Mean Square Deviation (RMSD) was almost 100 % greater for satellite data compared to data from ASKO, with a RMSD of 1647,18 kW h. Individual adjustment of loss parameters improved the accuracy of the simulations. Reducing soiling levels for summer and increasing levels for winter improved the coherence the most. Including thermal loss according to module temperature increased heat loss from the modules, and resulted in monthly increase of predicted energy. A rough estimate of ohmic resistance reduced ohmic loss from 1,5 % at standard test conditions to 0,3 %. The impact of albedo on the simulation result was negligible. Combining the loss parameters improved the accuracy of the simulation and resulted in a consistent monthly overestimate, suggesting possible losses due to light induced degradation or higher soiling loss during summer