Hello everyone,
I was quickly testing the newly-released components to simulate photovoltaic surfaces with LBT and I have found out some strange behaviors when considering extra shadows/context that can influence electricity production. I compared the electricity production obtained by using the LB Incident Radiation component “total” output (multiplied by the same loss factors of the photovoltaic components), with the “total _DC” output from the new HB Generation Loads component. Despite the two results agree pretty well when no obstacles in present in the model (see Fig.1), if an obstacle/context is introduced, completely different results are obtained (see Fig. 2) like if new photovoltaic components are not fully accounting for the shading obstacles.
In the presented case I tried with an extreme example by completely enclosing the PV surfaces into a “box”, however discrepancies are visible with smaller obstacles as well. Am I doing some mistakes in setting up the model or is it some sort of bug?
Skimming that, it seems like there’s a few effects at play like anti reflective coatings, radiation treated as isotropic an extreme angles, detailed thermal and conversion loss calculations that would account for the differences in the exposed condition results.
For the shaded condition, I’d guess it comes down to the solar and shading algorithms. I think when I’ve tested shading objects in IES VE even if a surface is fully shaded the solar algorithm still simulates it as fully exposed to diffuse solar - I’m guessing that E+ is doing something similar (at least with the settings you’re using), the closest I could find about it was below.
But I couldn’t find anything saying what information E+ passes to PVWatts, ie whether it direct or diffuse shading factors and then what PVWatts does with them.
Thank you @charlie.brooker for investing time in looking into it!
The non-negligible energy generation results in the shaded case are however quite weird to get as an output.
However, if this is a limitation of the algorithm and not a bug, I assume we have to leave with it and just being careful not to always take the results for granted.
