Pmpp Temperature Coefficient Problem

[kjs55]

I am reporting a Pmpp temperature coefficient problem with PVsyst.

In V6.26 and later, PVsyst changed the way that the temperature dependence of PV modules is being modeled.

The brief release notes state that the derivative at 45 C is now taken into account in addition to the derivative at 25 C.

PAN files in the V6 database will lead to different results depending on which version of PVsyst V6 is being run. Namely, the same PAN file used in V6.25 and earlier versions will lead to different results compared to if it is used in V6.26 and later versions of PVsyst.

Also, any PAN file in the database with optimized parameters for fitting the temperature dependence in V6.25 and earlier versions may no longer be optimized for V6.26 and later versions. Likewise, any PAN file that is optimized for V6.26 and later versions may not be optimized for V6.25 and earlier versions. As an example, in PVsyst V6.25 or earlier versions, a temperature coefficient of -0.42 %/C may have been chosen as the input for µPmpp to yield an effective temperature coefficient of -0.43 %/C over the full range of temperature values (e.g., 25-65 C). As of PVsyst V6.26 and later, the inputted temperature coefficient of -0.42 %/C may yield an effective temperature coefficient of -0.42 %/C over this same temperature range.

To demonstrate all of this, I simulated the exact same project in Versions 6.25 and 6.26 (500 kW in Phoenix, Arizona with a 310 W multi-Si module type). The results are as follows:

PVsyst V6.25

961.64 MWh/yr

PV loss due to temperature: -10.0%

PVsyst V6.26

963.74 MWh/yr

PV loss due to temperature: -9.8%

Is it acceptable for module manufacturers to manage this bug by providing two sets of optimized .PAN files for V6; namely, PAN files that are optimized for V6.25 and earlier versions and PAN files for V6.26 and later versions?

Edited December 16, 2019 by kjs55

[André Mermoud]

This is an interpretation of the muPmpp value, as specified on the datasheets (and usually on test reports) and representing a linear behavior.

This new interpretation better matches the supposed linear behavior of the Pmpp.

I doubt that many manufacturers have optimized this parameter in order to better match the simulations in PVsyst.

Now PVsyst is an evolving program. We try to improve the simulation when we discover some weakness.

If the results should be always quite stable, there were no more room for improving the simulation and the program would be frozen.

NB: If you want to avoid this modification, in the Hidden parameters you can set the parameter "PVModules" / "Upper reference temperature for muPmpp default" = 25°C.