Update 7.2.0 creates huge discrepancies for Singel Axis Trackers with backtracking

[ckremin]

Dear PVsyst Team,

I have just installed the new update 7.2.0 and loaded a project with single axis trackers (backtracking enabled) which I simulated with 7.1.8 yesterday. The new results with 7.2.0 show a huge negative delta to 7.1.8!

I deactivated backtracking in the shading scene and did another run. The differences in yield were marginally. Backtracking had close to no effect when it should.

Finally I redid the whole shading scene as I hoped that having made the shading scene in 7.1.8 may cause the issue. To my great surprise the new results with the remodelled shading scenes were even worse and now backtracking was even lower than without. The shading factor table showed an impact even with backtracking enabled, which was never the case before. Please find the results below.

 

 

I am sure there must be an error in the calculation and setup. This cannot be correct.

Kind regards,

Dr Christoph Kremin

Director Systems Engineering

CERO Generation

[M.Herseth]

I am also facing the same problem. The production and yield values of my existing Single-Axis tracker Bifacial Model have gone down when I ran the simulation with the new PVsyst version.

The Bifacial model orientation is getting changed in the report (The tracker spacing and width in the report are not what I had provided in the model ; please see the screenshot)

 

 

It would be really helpful if PVsyst Team could provide a solution,

Thank you,

Michael Herseth

[Bruno Wittmer]

Indeed there seems to be a problem with the bifacial tracking systems. The parameters of pitch and width are not correctly applied.

@ckremin : Can you confirm that the system where you observe the discrepancy is bifacial?

I also did checks on monofacial systems, and found consistent results between PVsyst V7.1.8 and V7.2.0.

We will fix this issue of bifacial trackers as soon as possible and publish an update.

We are sorry for the inconvenience, and thank you for reporting the issue.

[ckremin]

Indeed there seems to be a problem with the bifacial tracking systems. The parameters of pitch and width are not correctly applied.

@ckremin : Can you confirm that the system where you observe the discrepancy is bifacial?

I also did checks on monofacial systems, and found consistent results between PVsyst V7.1.8 and V7.2.0.

We will fix this issue of bifacial trackers as soon as possible and publish an update.

We are sorry for the inconvenience, and thank you for reporting the issue.

 

Yes, the used modules are bi-facial types.

You are welcome and glad to help.

[smeredith]

I'd like to add some information to this topic.

I'm a performance engineer responsible for reviewing models produced by other engineers. I recently came across a tracker project that was run in 7.2.3 and had some significant errors with regards to calculating POA and shading. These errors were corrected when the project was re-run in 7.2.4.

Some background:

-Project located in Maine, USA

-The project is single-axis tracker with one module high in portrait. Backtracking is enabled.

-The project is modeled with a shade scene consisting of tracker tables and tree-like objects (polygons that resemble trees, but not actual tree objects). The shade scene includes a ground object representing the topography of the site. The tracker tables conform to the topography with an average axis tilt of about -0.8 degrees.

-Average row-to-row spacing is about 6.09 m.

What happened:

-Initially, a simulation was run in 7.2.3 (Variant 1). There was a problem with the string partitioning (specifically each table was partitioned one string wide and one string high, when it should have been two strings wide and 2-3 strings high) that required the model to be run again.

-The partitioning was corrected and the simulation was run again in 7.2.3 (Variant 2). When it was re-run I compared the loss diagram from the two simulations and found that the POA and near shading losses were different, despite the fact that only the partitioning changed. As read from the loss diagram, POA gain increased from 24.4% to 26.8% and near shading loss increased from -4.18% to -4.89%. Other downstream losses were affected as well by these two changes.

-I opened the raw variant files for both of these variants and found that for Variant 1 only, within the PVObject_Ombrage=pvShading section, the pitch was listed as 6.6 m. Variant 2 showed the correct pitch of 6.09 m, which leads me to believe that Variant 2 is the correct one.

-I updated my PVsyst version to 7.2.4 and re-ran the original simulation with the incorrect partitioning (Variant 3). I found that the POA gain was 26.8% and near shading loss was -4.87%, nearly matching Variant 2.

So it seems to me there's a problem in 7.2.3 that is causing POA and row-to-row shading to be incorrectly calculated. And string partitioning may have something to do with it?