The impact of PV (photovoltaics) on the power system becomes increasingly important to study as the penetration of PV has increased rapidly over the last decade. A physical model for aggregated PV generation has been developed for the Swedish spot market areas. Information about PV systems within the Swedish electricity certificate system and irradiance data from the meteorological model STRANG were used as inputs. The model was trained and validated against production data reported to the Swedish transmission system operator. Our model shows high correlation (0.95-0.99) to reported historical production data. However, it overestimates maximum 1 h ramp rates, which are -20% and 22% for down and up-ramps respectively, compared to -13% and 14% for the reported data. Furthermore a weighting function was developed, which takes demography, available solar irradiance and today's PV deployment into account, to model likely deployment in a Swedish high penetration scenario, where PV covers 6% of the total annual power demand. The difference in large 1 and 4 h step changes before and after introducing PV is small. The model could thus be used with confidence to model the impact on the power system for future scenarios of high PV penetration. (C) 2016 Elsevier Ltd. All rights reserved.