Questions on Diffuse Component Decomposition in PVsyst for Client Comparative Study

[Zaki BOULAOUCHE]

Hi PVsyst Team, 

I’m a PV Performance Engineer at TotalEnergies, We’re currently performing a detailed performance comparison using PVsyst, PVlib and SAM, for one of our large-scale PV clients, using the same high-resolution meteorological data derived from on-site field measurements.

We’ve noticed areas where PVsyst’s POA outputs differ, and we’d love your insight so we can explain the differences:

• Negative circumsolar term values Around sunrise/sunset and significant differences in Isotropic diffuse too. Most Perez (1990) implementations cap that at zero (in PVlib, SAM). whether this is intentional, and if so, why PVsyst retains negative values rather than forcing them to zero.
• Is PVsyst’s Horizon brightness implemented as a band?  numbers diverge from SAM and PVlib’s continuous-integration approach. Any guidance on how that band or weighting is defined would help us reconcile the three software outputs.
• do you use other air‐mass equation when computing? because in PVlib defaults is to the Kasten & Young (1989) air mass formula, which can shift the sky brightness attenuation and thus POA irradiance calculations.

Understanding these points will allow us to prepare a clear, defensible comparison report for our client highlighting not just “which software” but why small differences arise at low sun angles or under certain horizon conditions.

Thank you in advance for any details or references you can provide. 
Best regards,
Zaki B.

[Auriane Canesse]

Dear Zaki

• We do not cap negative values in the Perrez model when computing the circumsolar brightness. They arise when implementing the model and we keep them to avoid biasing the result towards positive values. This term is added to the isotropic component and the irradiance itself is never negative.
• The horizon is not implemented as a band. The horizon component is calculated using the Perez/Ineichen's 1990 model, but it is directly added to the diffuse irradiance. Horizion shadings are applied  as a fraction on the diffuse irradiance but not to the horizon irradiance component separately.
• For the air mass, we use the the Kasten model described in Ineichen's thesis  "Mesures d'ensoleillement à Genève", which is corrected for altitude as well (exp(-0.00013*altitude))
  https://unige.swisscovery.slsp.ch/discovery/fulldisplay?vid=41SLSP_UGE%3AVU1&search_scope=MyInst_and_CI&tab=41SLSP_UGE_MyInst_CI&docid=alma991006463789705502&lang=fr&context=L&adaptor=Local Search Engine&query=sub%2Cexact%2CKANTON GENF (SCHWEIZ)%2CAND&mode=advanced&offset=0
  

I hope this helps you in your study