Tracking mode electrical losses. Bactracking algorithm

[tecnun]

Hello,

We've been encountering some severe electrical losses in several 1V tracking mode Utility Scale PV farms, where pitch is a little tight (less than 5 m).

For instance, at 5,5 m no electrical shading losses are founded, but at 4,9 m we found a huge drop in PR due to >5% electrical shadings. Does it make sense?

Could it be due to an incorrect backtracking algorithm? What we would expect is an angle shorten during the morning and during the evening like the simple drawing we are attaching.

If we simulate it in 2H instead of 1V there is no such electrical shadings, we attribute it to the bypass diode effect, what we find it cool, but in tracking mode we should not expect it. If the backtracking applies there will be a gaining drop but no so accused as it is reflected now with such a severe electrical drop.

Is there some explanation to this effect?

Best. Jose

[dtarin]

Backtracking fixing.pngBacktracking fixing.pngBacktracking fixing.pngHello,

We've been encountering some severe electrical losses in several 1V tracking mode Utility Scale PV farms, where pitch is a little tight (less than 5 m).

For instance, at 5,5 m no electrical shading losses are founded, but at 4,9 m we found a huge drop in PR due to >5% electrical shadings. Does it make sense?

Could it be due to an incorrect backtracking algorithm? What we would expect is an angle shorten during the morning and during the evening like the simple drawing we are attaching.

 

Are you comparing production data from two different plants? They should be commissioned with different backtracking parameters to account for the difference in pitch and shading. Have you compared the tracking angles reported from the trackers?

[tecnun]

Thanks for your response Dtarin,

There are 2 issues: One theorethical (PVsyst) and the other practical (actual 50 MW PV farm data)

The one i would like to double check is the theorethical point of view. The plant we are simulating through PVsyst algorithms has a 5,5 m pitch in 1/3th of the plant, and a 4,9 m pitch in 2/3th. I am attaching the report, in which we can observe a severe shadowing electrical drop (5,91%), affecting the global PR.

So we have created an imaginary configuration of the plant: The whole plot with 5,5 m of pitch. And the results are absolutely greater as no electrical shadowing effects are displayed (they fade away). It is attached here as well.

We acknowledge the problematic of partial shadowing in 1P (portrait) configurations, but we would like to double check this PVsyst penalization is accurate. It seems like backtracking algorithms is enabling partial shadowings and we would like to understand if they are optimized. For instance, is PVsyst software capable of deciding whether it is better (1) to lose proper angle of sun captation, or (2) to let certain partial shadowing, and if landscape of portrait configurations are taken into the equation in these backtracking internal PVsyst algorithms.

Hope I could now let me explain deeper my question

All comments are very welcome

BR Tecnun

[dtarin]

The one i would like to double check is the theorethical point of view. The plant we are simulating through PVsyst algorithms has a 5,5 m pitch in 1/3th of the plant, and a 4,9 m pitch in 2/3th. I am attaching the report, in which we can observe a severe shadowing electrical drop (5,91%), affecting the global PR.

 

Are you modeling it this way in PVsyst (in the shade scene), where 1/3 plant has 5.5m pitch and 2/3 has 4.9m pitch?

[tecnun]

Yes of course (and according to strings in order to enable the lack of effect of bypass diodes as it is portrait config)

 

 

When we also disable the half-cell effect, the electrical losses are even higher.

We have also considered an equivalent configuration but in 3H and the losses fade away even with 4,9 m of pitch

BR Jose

[dtarin]

That is the issue I think. Have you used backtracking management to set the tracker pair which determines the backtracking algorithm? You should model these two blocks separately so each has the appropriate backtracking algorithm, as they will be commissioned this way (or at least they should be). If you insist on modeling together, try using the 4.9m pitch block as the reference for the backtracking algorithm under backtracking management and see what your results are.

[tecnun]

Thanks a lot Dtarin,

It was 4,9 m pitch backtracking managed. If it was 5,5 the losses would be higher as you said.

Anyway i will have to combine 2 partial simulations (like it was 2 independant PV farms) and see how it behaves, let me show you the results when i got round to.

Thanks again for your help :) very useful!

[dtarin]

If you are using 4.9m pitch as the backtracking, I dont think there should be any shading losses. You will be underpredicting your irradiance gain for the 5.5m block, but you should at least not have any electrical effect losses. I am modeling in 7.2.6 and tested a site in North America to confirm. See attached. You can likely do the same check on your end to see if your results are the same. Set the backtracking pair for each block and run the shading analysis in the shade scene to check if there are any electrical losses.

[tecnun]

Dtarin, it was at 4,9 m (default) but unticked, maybe thats the reason (i guess i m sure ?), i thought you did not have to. Tomorrow i will simulate ticking it, and with 2 separated sims to see the differences of the latter. Thanks for your agile and accurate help

Best.

[tecnun]

Thanks Dtarin for your unvalued help, now we are ok, results look quite coherent

Congrats for your software improvements

BR Jose

[tecnun]

By the way...I'm seeing your backtracking algorithm is set to "zero shadowing". So, in the morning, trackers do not follow astronomic perpendicularity to the ray until the is NO shadow at all on the modules captation area, and in the evening trackers begin backtracking when certain shadow is spotted.

This algorithm is the optimal when modules are portrait. But if not...wouldn't be a turning point different to this NO shadow which could harvests more yield (maximizes it)? And if so...are you guys intending to perform this optimization in future versions?

Thanks BR Jose

[tecnun]

Sorry for the insistance, this issue is quite important for us for sharing it with our Clients and Technological Colleagues, in order to be comfortable with the results.

Given that backtracking algorithm is set to avoid every shadow, we cannot understand why in the report, there is a percentage of shadowing identified.

Let me explain with this example:

 

 

As we can see it exists 1,57% of irradiance losses due to near shadings. But...let see now these table & graphs:

 

 

It seems like backtracking has cancelled any shadow during the entire year of simulation. Where are these 1,57% irradiance losses imputed? Gain? it should not i think, as gain is already attributed hereinabove (30,4%). Also when we simulte shading animations in the shading scene, no shadow is spotted.

Is there any explanation? Only reason i see is that 1,57% is the gaining loss due to backtracking movement, and 30,4% is "plain" gain without considering any backtracking. Is it that?

BR

[dtarin]

Diffuse light, partially or in total.

Edit: probably coming from diffuse and albedo light

https://forum.pvsyst.com/viewtopic.php?f=30&t=694

https://forum.pvsyst.com/viewtopic.php?f=30&t=2522

[tecnun]

Yes, but it sais "attenuation for diffuse / albedo = 0.00000", what do you guys mean with "attenuation"?

Diffuse or albedo are not considered isotropic?

BR

[tecnun]

Ok now haha, thanks!

viewtopic.php?f=30&t=2522

With this topic we are seeing they vary with the angle. Anyway this attenuation = 0 should be accounted non equal to zero right?

BR and thanks for the agile response

[dtarin]

Ok now haha, thanks!

viewtopic.php?f=30&t=2522

With this topic we are seeing they vary with the angle. Anyway this attenuation = 0 should be accounted non equal to zero right?

BR and thanks for the agile response

 

Edit: I think the attenuation factor is something separate from the diffuse shading factor, which is "..independent of the sun's position, and therefore constant over the year...". From how the manual reads, the attenuation factor is something the software calculates when there are shading objects in the shade scene. So if you have trees or buildings (or basically a non-zero beam loss), you will get a non-zero attenuation factor. If you do not have objects, this factor remains zero.

"The diffuse attenuation factor should be calculated, by integrating simultaneously the shading factor due to horizon, the near shadings factor according to the table, and the IAM attenuation factors over the visible part of the sky hemisphere. " - We dont know the integrals being performed, but perhaps if near shading (beam) is zero, the result is zero.

[tecnun]

Thank you Dtarin,

Just for double checking: The shading losses are composed by these items:

1) NEAR SHADINGS:

I) Irradiance losses

a) Mutual due to adyacent structures

- Direct irradiation (zero in the case of actual PVsyst backtracking)

- Diffuse irradiation (coming from vaulted ceiling)

b) Other near objects (trees or buildings for example)

II) Electrical losses (due to string configuration)

2) FAR SHADINGS (for example mountains, horizon shadings in general)

1-I, 1-II and 2 are accounted separately on the final losses "tree", in the report.

Attenuation is zero just when direct irradiation + other near objects are equal to zero

BR Jose

[tecnun]

Sorry, what is the role of ALBEDO in this case?

I forgot to mention before...

BR

[dtarin]

Sorry, what is the role of ALBEDO in this case?

I forgot to mention before...

BR

From https://forum.pvsyst.com/viewtopic.php?f=30&t=2522,

 

First, we observe that the Albedo contribution is important.

The transposition model involves an albedo contribution proportional to (1 - cos(tilt))/2, i.e. low at usual tilts (e.g. 4.7% for tilt=25°), but significant at high tilts (25% at 60°).

Then, as only the first tracker "sees" the albedo of the ground in front of the system, the albedo is affected by a shading factor of (n-1)/n, where n = number of sheds. This means that the albedo contribution is almost completely lost in big systems

[tecnun]

Ok that's really clear. I guess this plot of angles in fixed structures considers the same pitch, bcs if we decrease it the albedo shading would remain...

Thanxs again Dtarin

[tecnun]

Ok now haha, thanks!

viewtopic.php?f=30&t=2522

With this topic we are seeing they vary with the angle. Anyway this attenuation = 0 should be accounted non equal to zero right?

BR and thanks for the agile response

 

Edit: I think the attenuation factor is something separate from the diffuse shading factor, which is "..independent of the sun's position, and therefore constant over the year...". From how the manual reads, the attenuation factor is something the software calculates when there are shading objects in the shade scene. So if you have trees or buildings (or basically a non-zero beam loss), you will get a non-zero attenuation factor. If you do not have objects, this factor remains zero.

"The diffuse attenuation factor should be calculated, by integrating simultaneously the shading factor due to horizon, the near shadings factor according to the table, and the IAM attenuation factors over the visible part of the sky hemisphere. " - We dont know the integrals being performed, but perhaps if near shading (beam) is zero, the result is zero.

 

In the case of backtracking mode there is no attenuation factor, all the shadings come from diffuse+albedo, this is absolutely clear thanks for last response.

But, looking deeper to the hourly shading factors, we observe that, apart from diffuse and albedo, there exists global factors below 1, which to my understanding they should be equal to diffuse (global = direct + diffuse). Am I right? Could it be a bug?

BR

[tecnun]

Ok now haha, thanks!

viewtopic.php?f=30&t=2522

With this topic we are seeing they vary with the angle. Anyway this attenuation = 0 should be accounted non equal to zero right?

BR and thanks for the agile response

 

Edit: I think the attenuation factor is something separate from the diffuse shading factor, which is "..independent of the sun's position, and therefore constant over the year...". From how the manual reads, the attenuation factor is something the software calculates when there are shading objects in the shade scene. So if you have trees or buildings (or basically a non-zero beam loss), you will get a non-zero attenuation factor. If you do not have objects, this factor remains zero.

"The diffuse attenuation factor should be calculated, by integrating simultaneously the shading factor due to horizon, the near shadings factor according to the table, and the IAM attenuation factors over the visible part of the sky hemisphere. " - We dont know the integrals being performed, but perhaps if near shading (beam) is zero, the result is zero.

 

In the case of backtracking mode there is no attenuation factor, all the shadings come from diffuse+albedo, this is absolutely clear thanks for last response.

But, looking deeper to the hourly shading factors, we observe that, apart from diffuse and albedo, there exists global factors below 1, which to my understanding they should be equal to diffuse (global = direct + diffuse). Am I right? Could it be a bug?

BR

[dtarin]

What this is telling us is that the shading losses are contributed by the diffuse and albedo components. If you refer to the link below , it states that the albedo component is affected by both the size of the plant and by the tilt, and effectively for large plants the albedo is near-zero. I suspect there is a mathematical operation on the albedo shading factor we see in the 8760 before computing the global shade factor.

https://forum.pvsyst.com/viewtopic.php?f=30&t=2522

[tecnun]

Thanks Dtarin,

Sorry but I still have not understood these exported figures:

 

 

What do they stand for?

I ve highlighted one specific hour where FShdGl, FShdDif and FShAlb are not zero,

FShdGl = Global ?? Global = Direct + Diffuse, there is no direct shading so it should be equal to FShdDif as far as I know...

Anyway i still dont understand why there is no direct shadings, they should be the most important ones, shouldnt they??

Please need some help :idea:

[dtarin]

These are shading loss factors, so 1-FShdBm = 0% beam loss. 1-FShdGl = 1-0.9846 = 1.54% global shading loss. The individual shading ratios are not as informative when we dont know the calculation steps PVsyst is taking, but the global shading ratio can be calculated, see screenshot. When there is backtracking and no shading objects or terrain consideration, the beam loss is zero, which is why it is showing a value of 1 for the ratio. So, a value of 1 = 0% loss, a value less than 1 means there is a shading loss contribution for that irradiance component.

*There is not a horizon shading loss in this example, but if there were, then it would not be GlobInc in the denominator, but instead GlobHrz (global irradiance after horizon loss)