Shading loss changes in near shading irradiance loss and electrical loss as versions evolve

[Debbie]

Hi PVsyst team, I am noticing some changes in shading losses as I go through version updates.  These are some general observations.  Sometimes running the shading the first time generates different results than the second time.  Sometimes the values within "irradiance loss" seem to reduce, while the values within "electrical loss" increase.  In version 7.2.16, when comparing unlimited sheds (fixed tilt), vs. the same set of inputs in near shadings, the losses are more significantly different than they used to be in earlier versions.  I understand you might need specific files to check what is happening.  But can you comment generally on whether the shading calculation methods and algorithms are getting updated intentionally with version updates?  We now have the capabilities to implement more and more complex shading scenes.  Can you comment on how robust the calculations are with complex scenes?  Do you have any recommendations for us as modelers to ensure that shading calculations remain accurate even when running complex scenes?  Thanks in advance.

[Michele Oliosi]

Hi thanks for the info. Another post was already mentioning this kind of recomputing issues.
 

We are investigating the matter but we haven't been able to pinpoint the exact issue yet.

Regarding differences between unlimited sheds and 3D scene, yes, an example would be valuable. If you have such a project with an earlier calculation and a newer one, you can send it at support@pvsyst.com. It could be very useful for us in terms of tracking possible bugs.

In terms of shadings there mostly have been bug corrections (as far as recent patches are concerned), hence the robustness has mostly improved. But the calculation itself has not changed.

For complex tracking scenes, we have added the possibility to change the 40 tracker cap, above which the diffuse shadings calculation was simplified. This simplified calculation sometimes leads to incorrect results, whenever there are patches of trackers instead of a single homogeneous patch. If you are in such a situation, we recommend you check out this advanced parameter:

 

[Debbie]

Thank you, I will try the 40 tracker cap modification.  What value do you recommend raising it to, for example if I'm modeling  a system with 5000 trackers?

[Michele Oliosi]

You should raise it above the number of trackers, here e.g. 6000, the exact number being unimportant. Else the simplification is going to be applied anyway.

[Debbie]

Does this threshold for partial scene apply differently depending if your model uses backtracking or no backtracking?  I am running a First Solar Array with no backtracking.  The complex layout, which is very spread out in multiple small sub-arrays, results in reduced shading losses compared to a simple rectangular layout, even when setting this threshold greater than the tracker count.  Does a tracking array with no backtracking still make use of a reference tracker for calculations?

 

 

[Debbie]

Ah, I had to also redo the shading table calculation after changing the threshold, then after that it triggered the "computing the integrals for diffuse" while running the model.  So it does seem to work the same with a non-backtracking array.

[Michele Oliosi]

Yes it is the same issue with or without backtracking.