How to simulate Tandem data of top and bottom cells

[Shurouq]

Hello, I have the tandem cell details, the bottom and the top. How can I simulate it as a 4T connection? can I do it separately? 

[Luca Antognini]

Hello,

We are starting to think about the question of simulating TANDEM solar cells in PVsyst, but unfortunately for the moment it is not possible.

The behaviour of the two absorbers with temperature and solar spectrum is not trivial and we need to implement a few models to describe it.

Though, I believe the case of 4T (both cells connected in parallel) can be more easily approximated than the 2T case (both cells connected in series), as voltage mismatches create usually less power losses than current mismatches.

I will give here an attempt to set up PVsyst to approximate a 4T Tandem behaviour in the present version of the software (8.0). But bearin mind that this is a very preliminary approach and require more testing.

1) Create two .Pan files, one to represent the top cell, one to represent the bottom cell, as if they were measured independently. This way you set up the one diode parameters for each cell. Use the same number of cells, modules dimensions etc...

2) Create a system with two subarrays, one with each .Pan files, with same number of panels connected in series and number of strings in both. Connect them to a same inverter and use the power sharing feature. You will not be able to connect the two subarrays to a single MPPT here, so at this stage we are neglecting the voltage mismatch (see next step).

3) Advanced losses /  Module Quality LID Mismatch / Strings Voltage Mismatch: You can add here a general percentage to represent the loss of power due to the mismatch. To evaluate this one, you can use PVsyst specific tools > Electrical Behaviour of PV Array > Heterogeneous array. Here you can select your two .PAN files and open the I/V graphs. It will show you the percentage loss due to the mismatch between both modules for a given time of the day and date. Vary a bit those one and you will notice that this percentage is almost constant, as voltage mismatch should behave. Report this number in the Advanced losses.

4) Thermal model: We also need to tweak PVsyst regular parameters. With our setup, PVsyst will see two independent panels with low efficiency (~<15%), and therefore calculate that they are dissipating a lot of energy into heat. Which is not correct since in reality we have a single device with high efficiency (>30%), so more electricity, less heat. We need therefore to adapt the coefficients of the thermal model (https://www.pvsyst.com/help/project-design/array-and-system-losses/array-thermal-losses/index.html#thermal-model)

According to the equation of the thermal model 

Tcell=Tamb+(Alpha⋅Ginc⋅(1−Effic))/U

we could correct the model by using a different U value to compensate for the wrong (1-Effic) term.

In my example (assuming single cells efficiency at ~15% and Tandem at about ~30%), the corrected value would become U_corr = U * (1-0.15) / (1-0.3) = U * 1.21. So in the case of open rack, instead of using the values of Uc = 29 W/m2/K, use something slightly larger Uc_corr = 35 W/m2/K.

This will lower the losses of 1-2%.

5) Spectral correction: Unfortunately, I haven't reflected yet on how to adapt the spectral correction for the Tandem case. I think this cannot be guessed and of course the behaviour will be different than crystalline silicon. But you should keep in mind that it might be an import factor that we ignored too here!

Please do not hesitate to comment on the approach.

And as said above, we are reflecting on what model to implement in the future for dealing with 2T and 4T tandem devices, so any input would be useful for us : )