Simulation Losses Incompatible with Array Voltage

[Alexandre V Cavalheiro]

Hi!

I work on a photovoltaic plant with 6030 545W Trina TSM-545DEG19C.20 bifacial solar modules arranged in 201 strings with 30 modules each. We use Huawei SUN2000-215KTL-H0 inverters, and the string voltage shown in the PVsyst simulation for this configuration is 856V.

However, after months of generating less energy than predicted by the PVsyst simulation, we were informed by the inverter manufacturer that the string voltage was below the minimum voltage for nominal power. We reconfigured the arrangement so that 25% of the plant remained with strings of 30 modules (totaling 51 strings), and 75% now had strings of 36 modules (totaling 125 strings). The average voltage of the 36-module strings is 1027V (within the inverter's maximum power range: 930V to 1300V), and the others remained at 856V. We maintained the plant's initial ratio of 1.22.

As a result of this reconfiguration, the plant began to generate considerably more energy (since it was no longer operating below the Vmppt range), and operate within the expected generation:

	PVsyst predicted generation:	Actual Plant Generation:	Measured/Expected:
Jan/24 - May/24 - 30 modules (856V Array Voltage):	2731 MWh	2205 MWh	80,70%
Jun/24 - Sep24 - 36 modules (1027V Array Voltage):	2194 MWh	2179 MWh	99,30%

We updated the PVsyst simulation prior to the construction of the plant to include this changes in the as-built documentation. However, when generating a simulation based on the installed reconfiguration, the results indicated lower generation for arrays of 36 modules, compared to arrays of 30 modules. And this situation clearly differs from reality. It is worth mentioning that the simulation indicated unrealistic losses due to clipping with arrays of 36 modules, as if the inverters were operating outside the optimal voltage range.

Has anyone experienced a similar situation? How did you solve it?

Thank you in advance for your help!

[André Mermoud]

Yes this is normal: with a Vmpp lower than 880V, you have losses due to an excessive current at the MPPT input.

This loss is not accounted in your simulation, because in the parameters of your inverter, you have not defined the value "Minimum Voltage for PNom", which is equivalent to a current limitation.  Without this parameter PVsyst cannot "know" that the current will be limited in your real system. 

According to the curve PNom f(Vin) above,  this parameter should be set at a value of 880 V.

This is fully explained in the help  https://www.pvsyst.com/help/physical-models-used/grid-inverter/inverter-pnom-as-f-voltage.html

Now in this particuar case, the PNom f(Vin) curve is not quite compatible with the hypothesis of the model described in the help (i.e. the violet curve). 

You have:

At  100%, PNom = 215 kVA. With a voltage = 880V, this means a current = 244A  (full inverter)

At 40%, PNom = 86 kVA. With a voltage of 500V, this means a current = 172 A.

Between these points the maximum current varies on the specified curve, but is constant in the simulation (violet dotted curve in the help).  I don't know the "accuracy" of the real behaviour of the PNom limit within the inverter. Perhaps you will have to slightly decrease the value of 880 V for getting your exact waited E_Grid.