How are managed the new "twin modules" with half-cells ?

[André Mermoud]

The new technology of "Twin half-cut cells" is more and more frequent on the market.

These modules involve half-cells strings, mounted in the following way:

 

Twin-module with half cells architecture

 

Electrically, these modules are made of 2 strings of half-cells, mounted in parallel.

In the PV modules definition, this should be defined as "Twin half-cells" layout, 60 (or 72) cells in series and 2 strings in parallel.

One of the advantages waited from this configuration is an improvement in the mismatch under partial shadings.

How can this new module configuration be taken into account in the PVsyst shading calculation ?

First of all, we will only consider "regular" shading situations, when the shades affect all lower submodules at a time, i.e. mutual row shadings in sheds arrangement (portrait disposition, modules in portrait of one string on a same row).

In this case as soon as the lower half-cell is shaded the lower submodule becomes completely inactive (for beam component), and the string current will be half the "normal" current (upper sub-module). This is the key point of the benefit of this configuration.

This should be taken differently into account in the different ways PVsyst evaluates the electrical shadings:

- In the "Unlimited sheds" 2D calculation, where the electrical shading loss is considered complete as soon as the first bottom cell is shaded, we can simply consider loosing half the module instead of the full module width.

This option asks for the number of strings in the width of the row. We can simply double this value for twin modules (i.e. consider 2 rows for one module in portrait). This doesn't require any program modification and the calculation is correct.

- In the calculation "according to module strings", the rectangle-string becomes inactive when the rectangle is hitten by a shade.

In our case the halfmodule lower row behaves as a string: as soon as a part of it is shaded the half-modules doesn't produce anything more for the beam component. Therefore you can define the "rectangle-strings" size as half the size of the module. This is only true for one row of modules: in you have several rows (for example U-wiring), this is no more true and the situation becomes much more complex (with reduced gain on the losses).

In the present time with normal modules, there is a correction for sheds arrangement: when the bottom cell is partially shaded, the remaining current should be proportional the the enlighted part of this bottom cell. In practice, PVsyst accounts for electrical shadings only when half the bottom cell is shaded (correct on an average over numerous situations). This correction becomes a little bit optimistic as it is applied on a cell which is half the size.

- With the "Module layout" calculation, the calculation of the I/V curves becomes more complex. It has been fully implemented in the Version 7. Here there will be a tool for deeply understanding the real electrical effect of the partial shades.

In the old versions (V 6.xx) the Twin modules are defined in the same way as usual modules in the ModuleLayout part (3 sub-modules in length). Therefore if you mount them as portrait, they will have slightly higher mismatch losses than the landscape.

Cells temperature

Now some people are wondering about the cell's temperature, and if it should be lowered in the model due to half the current in each sub-module.

I can't imagine why, physically, the cell temperature should be lower than in normal PV modules. The current is half, but in half a cell. Therefore the current per cell area is the same. Now if you have scientific papers with reliable cell-temperature measurements (differential between normal and half-cut modules), please send them to me.

Moreover I don't know the exact relation between the currents in the cells and the module temperature. This should be very low as to my understanding, the temperature elevation would be due to the power loss Rs * Ioper². Now Rs is around 0.3 ohm for a usual PV module, therefore the power loss is 24 W for Imp = 9A (and 6 W when operating at half-irradiance). If we assume an irradiance of 1000 W/m2, this module of 1.6 m2 will receive 1600 W, the current loss is 24/1600 = 1.5% of the irradiance, i.e. roughly a contribution of 1.5% to the U-value (and 0.75% under 500 W/m2). PVsyst desn't pretend to provide default U-values with this kind of accuracy.

By the way, this Rs loss is already accounted in the thermal balance equation, as this equation accounts for the efficiency.

Now if this is really the case, the benefit will probably be exactly the same in hot and cold climates, as the temperature correction wiil be added to the module temperature whatever the original temperature, and therefore lead to similar power differences.