Bifacial Model in PVlib vs PVsyst

[roshni agrawal]

Hi all,

I’m using the infinite shed model in pvlib for bifacial PV system modeling, and I’ve noticed something interesting when comparing results with PVsyst.

While front-side irradiance results are fairly consistent between the two tools, the back-side irradiance from pvlib is consistently higher than what PVsyst reports.

For context, I'm using the average orientation parameters exported from PVsyst as input to the infinite shed model in pvlib.

Has anyone else observed this? Are there known differences in how back-side irradiance is calculated between the two models.

Thanks & Regards,

Roshni

[kevina...@gmail.com]

I'd guess there are several differences, but I will highlight one in particular.  pvlib's infinite sheds model makes the simplifying assumption that light incident on the ground surface is not necessarily reflected from where it hits the ground, but rather uniformly across the entire ground surface.  This means that direct irradiance incident on the ground between rows is effectively shifted to the ground underneath rows, which tends to increase the irradiance on the back side of the modules.  At least this is the case currently (as of pvlib v0.12.0).  Hopefully in a future version, there will be an option to simulate this reflected light more realistically.

I cannot say for sure how PVsyst does it, but my loose understanding is that it makes the more realistic assumption that ground irradiance is incident and reflected from the same point. 

Kevin

[roshni agrawal]

Thank you Kevin for the detailed explanation — that makes a lot of sense. It's helpful to know that this is a known simplification and might be improved in future versions.

If others have tried workarounds (adjusting the ground reflectivity or modifying the scene setup) to minimize this discrepancy, I’d be keen to hear about those too.

Thanks again!

Best regards,
Roshni