grass - r.sun delievering unrealistic Values

I need your help. I'm trying to calculate the global irridiation for a specific area. I'm interested in the results for rooftops in special. I have a DEM, calculated slope, aspect and horizons and started r.sun for every day with a monthly changing Linke Factor. All calculated maps (slope, aspect, horizons) seem to have believable values. My Problem is: On the yearly sum (addition of all 365 days) I get values for total global irridiation of about 1900 KW/m². Which is 1,9times more then expected. Especially when the solar constant is only 1368 kW/m².... Could you imagine what went wrong? I'm lost.