The irradiance loss doesn't match after simulation

I have the following losses (there's no shading losses):
Soiling: -5 %
Reflection: -1.693 % 
TOTAL: -6.693 %

The total irradiance (year 1) for:
a =: POA front-side irradiance total nominal: 57723.83 kWh
b =: POA irradiance total after shading and soiling: 53909.12 kWh
Loss = 100 - (b/a) = 6,61%

Where's the 0.08%? Am I doing this right? Does the precision is affected by the simulation?

Hi Leandro,

I was only able to recover one of the images you inserted in your post. Please attach image files to your post (you can then insert it if you want, but you must first attach the file).

The small difference you see is probably due to rounding and to calculating losses from monthly totals rather than hourly (or subhourly) time steps.

Best regards,
Paul.

Here is the Sankey diagram, and a table with my interpretation of the losses (thank you for your patient)