When asked to taking into account daylight saving time to perform illuminance analysis I understand to keep fix clock time (9AM) , and then moving back 1h to take proper sola data (8th hour of the day) if we are in summer, in order to eprform calculation at 9AM LST, when people are at work and occupy the building .

This is obviously more severe (radiation is lower) , but with the rational exposed above . Moreover IES LM83-12 section 2.2.1 fixes the period of analysis explicitly referring to local clock time.

Example 1, Option 2 of the Ref. Guide v4 gives for NY City TMY3, September 17 ,  9AM, 618 W/m2 for GloHorRad, and 98 W/m2 for DifHorRad resulting in 520 DirHorRad. However when downloaded it from http://climate.onebuilding.org/   USA_NY_New.York-Central.Park.725033_TMY3  these data are recorded for 10AM !!!??

If you choose (properly) data at 8AM line , it results in 293-53=240 W/m2 DirHorRad , more then an half of reduction !!

I want to share this because on some projects I'm working on it results really challenging to reach good performance taking into account DST in Option 2 (as stated by the Ref. Guide), because of the 9AM conditions. But if I can shift 1h forward (like in the Ref. Guide example) rather than back "everything works better" ...

I would like to know what are you used to do, thanks

(Moreover, March 21, 9AM, in the same example has same  DIfHorRad 136 W/m2 than TMY£, but different GloHorRad 155 exemple vs 566 TMY3, and again on 10AM row ...)