Generation hours

Greetings!
I have a problem in the simulation: generation starts at 4-5 am in some months, which is not true.
Coordinates: 49.925 36.376
For the simulation were used the NSRDB and ECMWF/ERA resource libraries from PVGIS ( re.jrc.ec.europa.eu/pvg_tools/en/tools.html#TMY ).

I attach the simulation file:

Hi Maxim,

Please double-check that the time zone information in the weather file matches the time stamps. Usually, this kind of offset is caused by the time stamps being in UTC instead of local standard time.

You might also consider using data from the NREL NSRDB, which you can download directly from SAM's Location and Resource page. Recent versions of SAM can access NSRDB for most parts of the world. I was able to use SAM 2024.12.12 to download a TMY file from the NSRDB for the location in your file.

If you continue to have trouble and would like me to help investigate, please attach an example of a weather file that causes the problem.

Best regards,
Paul.

Got your reply Paul, thanks!
It is indeed the wrong time measuring, although the GMT2 zone is correct, also I have updated the SAM version, but the problem still persists.
Probably the problem is in the weather data file, I attach it:

Hi Maxim,

This location is at a relatively high latitude near 50 degrees North, so early morning and late evening irradiance makes sense for the summer months. I downloaded a weather file for the same location from the NREL NSRDB and see the same pattern, so believe this reflects the actual available irradiance.

Best regards,
Paul.

Thanks for your reply, Paul! The error is probably the clock forward one hour in the simulation region.
One more question on the topic, I have encountered a bug that with a certain panel orientation: the generation in morning hours can exceed the generation at noon, but this cannot be because of solar radiation diffusion. Can you please tell me what this could be related to? Or is this the correct result? This is observed with all possible models of Sky Diffuse Model in SAM. We have identified the mentioned errors in the "Parametrics" section.

I attach a file of the simulation and the weather data I used for it:

Hi Maxim,

It looks like on some days in January, the DNI is higher in the morning than for the rest of the day. This could be a data anomaly, or an indication of a sunny morning and cloudy rest of the day.

Given that the array is facing east (azimuth = 90), I would expect that there might be some days with sunny mornings and cloudy mid-days where the system would generate more power in the morning than in mid-day.

Best regards,
Paul.