An interesting discussion, especially as industry often accepts a longer string length than PVsyst allows and it is difficult to enter this value into PVsyst.
I agree that the string length calculation should factor in the very worst case cell temperature (by using the lowest historical site ambient temperature), however, the formula does not correct Voc with a realistic worst case plane of irradiance value for the start of the day. 1000W/m2 @ STC is overkill as the irradiance in the morning won't be anywhere near this when the central inverters self check prior to closing DC feeders in the morning.
The problem is you will need years of historical meteo data to justify this lower 'worst case' irradiance values in the morning are well below 1000W/m2. What do you think of this approach PVsyst admin?
