Correct selection of simulation variables for vertical bifacial PV

[Luis Zimmermann]

Hello together,
I want to analyze the deviation between measured pyranometer data and the simulation output data. 
The investigated project consists of vertical bifacial PV modules on a agricultural land. On the site there are three pyranometers measuring GHI, front irradiance and back irradiance. To compare the measured irradiance and the PVsyst data I am unsure which simulation variable to use. 

My current idea:
A) Front irradiance = GlobInc + ReflFrt
I use GlobInc instead of GlobEff, because the front-pyranometer is not affected the same of the optical losses (shadings, IAM, soiling) like the modules.
Question 1: Is that assumption right? 
Question 2: Is ReflFrt already included into GlobInc?

B) Rear irradiance = GlobBak + ReflBck
Question 3: Is ReflBck already included into GlobBak?

Please answer the three questions and give me general feedback on what to consider for the comparison between measured operating data and simulation output data. 
Many thanks in advance!
 

[Michele Oliosi]

Hello !

A) ReflFrt is not already included in GlobInc. It is okay to add it. Instead of GlobInc i would suggest using GlobHrz, if you have a horizon line. Moreover, depending on the positioning of the pyranometer, you may want to deduct the diffuse shading losses ShdDLss.

B) ReflBck should not be added like you suggest. This is because it is already included in GlobBak. Instead, you could add BackShd (which was deduced to get to GlobBak), as a way to reintroduce the structural shadings which probably do not affect your pyranometer.

[Luis Zimmermann]

Thanks for the response!
For B) I followed your advice. In result I get now around 50% more irradiance on the rear side from the measured operation data compared to the simulation within the winter months (Jan, Feb). By adding ReflBck I got a deviation of around 10% in the same months. The power plant is located in a snowy region, thats why I use an albedo factor of 0.7 for those two months. 
Do you have any idea where there could be a mistake?

Edited December 25, 2024 by Luis Zimmermann

[Michele Oliosi]

Dear Luis,

Regarding the albedo, did you enter it in the bifacial model? For the purpose of backside irradiance calculation (and also for the light reflected from ground to front) the albedo should be entered in system > Bifacial system > first tab.

Other than that, I wonder about your pyranometer placement. Do you have some images/details on that ? There could be some differences depending on the height at which they are placed (and also the pitch between your rows).

Finally, could you share a screenshot of your loss diagram ?

[Luis Zimmermann]

1) Yes, I did enter the albedo in the bifacial model. 
2) I attached a picture of the pyranometer placement, it is installed as seen in the first row of the power plant at a height of 3.10m. The power plant has only four rows with a pitch of 8-10m.  
3) Attached you can find the loss diagram out of the report. 

For the front side GlobInc = BeamInc + DiffInc + AlbInc (+ReflFrt), where AlbInc is the reflected irradiance from the far ground.
How or where is AlbInc included in the irradiance of the backside, because for vertical solar the calculation for back and front side should be quit similar, compared to modules with a tilt of f.e. 30 degree (there the front and back side irradiance calculation is definitely more different).

[Michele Oliosi]

2) I suspect that this pyranometer placement is biased towards higher irradiance than what is actually incident on the backside on average. Indeed, because of mutual shadings, the irradiance is not homogeneous on the backside. The lower end receives less light than the top.

For the front side, PVsyst outputs GlobInc, which is the irradiance in the plane of array, without yet applying mutual shadings. For the backside, this is not the case, mutual shadings are already accounted for in the different irradiance contributions.

For the sake of the comparison with the measurements, I would suggest defining a different variant in PVsyst that exchanges the front and backside. In this way, you can compare with BeamInc + CircInc + DifSInc + ReflFrt +( - HrzBLss - HrzCLss - HrzDLss), the latter parenthesis if a horizon exists, of this variant with the real backside measurements. Note that because the backside pyranometer sees other rows, the albedo contribution should be removed.

For the front side, since it is in the front row, there is a far albedo contribution: BeamInc + CircInc + DifSInc + AlbInc+ ReflFrt +( - HrzBLss - HrzCLss - HrzDLss - HrzALss).

3) I am a little surprised by the low backside to front side irradiance ratio. Is the azimuth not 90°?

[Luis Zimmermann]

2) Thank you for the recommendation, to add another simulation variant. In my understanding the Pyranometer for the rear side sees other rows, but the albedo contribution should not be removed completely. Attached you can find a drawing which explains my understanding. I would propose to calculate as followed:  BeamInc + CircInc + DifSInc + ReflFrt + AlbInc/2 + ( - HrzBLss - HrzCLss - HrzDLss - HrzALss/2)
Would you agree with that?

3)The azimuth is -47° for the frontside. That should explain your comment.

[Michele Oliosi]

2) Ah yes, this makes sense. I agree with your suggestion.

[Luis Zimmermann]

Thanks for all your answers! 
Could you explain again why you propose to deduct the the diffuse shading losses in your first answer of this discussion? (Screenshot attached)

[Michele Oliosi]

Originally I thought your pyranometer may not have been at the top of the row. This means that it would have “seen” other PV rows in its zenithal hemisphere, even in hours without direct shadings (which you could filter out). This is accounted for in the diffuse shadings.

But in your case, the diffuse contribution to the measured irradiance is unshaded by mutual shadings.

[sagar b]

On 1/10/2025 at 10:39 AM, Luis Zimmermann said:

Thanks for all your answers! 
Could you explain again why you propose to deduct the the diffuse shading losses in your first answer of this discussion? (Screenshot attached)

Hi Luis,

I would like to know more about your vertical bifacial project.

Would you share me your project location, DC, AC, mounting height with nos. of module in Y and X, and real time project result?

I am planning to consider vertical bifacial in heavy snowfall locations.

Waiting for your response.

Regards,

[Michele Oliosi]

UPDATE and correction to the previous answer.

The "far" albedo (AlbInc, AlbHrz, AlbShd) and the "near" albedo (between PV rows, i.e. ReflFrt) should not be included in the same quantity to compare to POA measurements. The exception to this rule is the case of POA measurements that are made at the end of a row (such as in your case Luis). The problem with including both is that there is a risk of overcounting. For example in the case of a single row, AlbShd and ReflFrt do somwehat overlap.

 

What this means is the following: if you measure POA away from the PV rows, or on the front row of the system, you can compare your measurements with GlobInc, or GlobHrz (which is like GlobInc but using the horizon profile as a shading). Between both, I'd rather use GlobHrz. Do not add ReflFrt.

 

Instead, if you are measuring between the rows, the "far" albedo contribution should not be counted, and you should use ReflFrt instead. This means that you can compare to, e.g., GlobHrz - AlbHrz + ReflFrt. Note that if the POA measurement between rows is well below the top of the PV rows, then you should at least shade diffuse light, i.e., BeamHrz (horizon shaded beam) + CircHrz (horizon shaded circumsolar) + DiffShd (fully shaded diffuse) + ReflFrt (reflected irradiance between the rows).

 

@Luis Zimmermann Sorry for the imprecise previous answers. This interaction between models is always a bit tricky. For your example, instead of 
BeamInc + CircInc + DifSInc + ReflFrt + AlbInc/2 + ( - HrzBLss - HrzCLss - HrzDLss - HrzALss/2)
I would suggest now
BeamInc + CircInc + DifSInc + ReflFrt/2 + AlbInc/2 + ( - HrzBLss - HrzCLss - HrzDLss - HrzALss/2)