inconsistency between PVlib and PVsyst when calculating rotated plane azimuth

[mayudong]

When I was comparing the stimulative results of PVlib with PVsyst, I noticed that for PV panels with same configurations, the calculated rotated plane azimuth computed by pvlib.tracking.singleaxis(..)
is different from that of PVsyst.

I did find out where the problem is, and it is related to the coordinator transformation.

Can I create a pull request to submit my fix?

[wholmgren]

@mayudong yes, that would be great.

[wholmgren]

@tafadzwalgurupira, I believe that @mayudong is discussing a potential issue with the single axis tracker code, whereas you are discussing an issue with some part of the plane of array calculations. So, I'd prefer that you make a new issue with more details to discuss the discrepancy between the tmy_to_power tutorial and your pvsyst results. This helps keep discussions focused and on topic. Thanks.

[cwhanse]

@mayudong could you post a summary of your fix? If there's a code error, it's likely also present in the Matlab toolbox and I'd like to fix it there as well. Thanks.

[mayudong]

Sure. I have summarized the issue and potential fix in the attached file.
Please double check and confirm the issue.
Best
NEX-PVSystemModelComparison-110416-1221-2.pdf

[wholmgren]

I tried to simply this part of the matlab code and it's possible that I introduced an error. I'll look at it more carefully.

[mayudong]

Actually the results calculated using Matlab are different from PVsyst too.
I will create a pull request sometime this week

[wholmgren]

I looked at the surface azimuth results more carefully and I agree that they are not correct in pvlib python for non-trivial cases. Example 3 of the python single axis tutorial and the matlab version's documentation simulates a tracker with 20 deg surface tilt. I remade the matlab figure to include the surface azimuth...

I'll look at the proposed fix in #145 next.

[wholmgren]

Closed by #145