Detailed Losses > tracking imperfection ?

[julmou]

A thought occurred to me as I was reading this part of the documentation: Project Design > Shadings > Backtracking Strategy > Backtracking Parameter Management. See screenshot below. 

Quote

During the simulation, the backtracking angle calculation will be the same for all trackers, based on this reference.

So basically, on the typical 50, or 100 or 200 MW solar plants here in Australia, there are thousands of trackers. And the way it's simulated, is that ALL trackers move exactly at the same time and do the same exact movement with the same angle. And I assume that the simulation is making it so that it is back-tracking only as much as it needs to, based on the sun's position and the incidence angle, so this is assuming that every single tracker on the field also has the utmost precision estimating or measuring the sun's position and deducing the right angle to adopt.

My point is, I think the way it's simulated is too perfect, and I doubt that:

1.  all trackers move exactly at the same time, hence there is some kind of mismatch loss, due to the trackers being slightly not all in phase with each other (in practice, they are indeed independent, and wouldn't all move exactly the same way like in the simulation),
2.  and I also doubt that they all have the correct perfect angle, there's maybe some margin to make absolutely sure that there isn't any mutual shading, and I also believe that whatever the way they estimate the proper angle to have at any time, there is some uncertainty in the process

For these two reasons, I was thinking that it would probably be adequate to incorporate in PVsyst a type of "tracking losses" section, with for instance, a mismatch loss, and an "imprecision" or "inaccuracy" or "miscalculation" loss. What do you think?

Besides, maybe it's unrelated, but all tracking solar farms I have re-created so far in PVsyst (with backtracking) perform much less in reality that in the simulation, with the difference being quite notably bigger than what I get with sheds solar farms.

[Michele Oliosi]

Hi,

You are probably right. Your observation that trackers with backtracking underperform is related to these inaccuracies, as well as the topography, see https://www.pvsyst.com/help/backtracking_onhill.htm

Even small amounts of irregularities in the topography, shadings would be generated, hence leading to electrical shading effects.

One option would be to add an optional extra loss as you mention, to be estimated by the user. Defining a default value for that would need an extensive study and is probably not realistic.

More generally I would consider adding a set of generic "custom losses", that the user can define and add at some level in the simulation. I will add this to our possible tickets.

In any case the discussion of these effects is a great and complex topic. Backtracking gathers a lot of expectations but it seems it will not always live to them. Moreover its implementation in PVsyst was complex and leads / has led to many discussions.

[julmou]

Thank you for the answer.

Have you also considered listing the tracker manufacturers in the Database?

It gets really tricky to find what the offer on the market is, especially with tilted or vertical trackers or dual-axis.

Besides, it would be a wonderful comparison tool, being able to see that such and such system have stroke limits of 60°, while this one has 80°, and perhaps this one has backtracking and the other not, and perhaps the same manufacturer offers a range of options with tilted single-axis trackers, etc.

[julmou]

Do you think having a database for tracker systems could be relevant for PVsyst?

[Michele Oliosi]

It would be quite nice, although it would require quite some work, especially keeping the database updated (the electrical components already generate quite some work). It would be a good fit in the orientation window, at least as an informative list. I can imagine that a tracker database as a webapp would already be wonderful, having it directly in PVsyst even better.

In the short term we will however probably prioritize covering more generic cases and offering a more detailed/precise simulation. Especially since more and more tracker manufacturers have special backtracking algorithms, covering these cases would probably be a good start already. But we're always discussing all sorts of avenues; we won't disregard your suggestion.