Decomposition and transposition models

[dina.christensen.martinsen]

I am currently working on an ongoing research project that involves analyzing decomposition and transposition models for Nordic climates.

Our results show that the default decomposition and transposition models in PVsyst yield higher uncertainties compared to alternative models for several of the analyzed locations. Specifically, the current options in PVsyst (Erbs and Perez) yield significantly higher uncertainties (rRMSD of 12,9% and rMBD of 4,3%) compared to other options with almost no bias in terms of rMBD and rRMSD below 10%.
Would it be possible to provide an option to perform the decomposition and transposition step separately and import a custom file containing not only GHI, but also DNI, DHI, and POA to PVsyst?

Alternatively, could you add more decomposition models and transposition models to choose from in PVsyst?

[Michele Oliosi]

Hello,

This sounds like a great research. Are you going to publish it? If so, we'd gladly go through it once it's out there.

In PVsyst you can always (and it is recommended to do so) import DHI or DNI in addition to GHI. This means that we do not need to rely on DirInt or Erbs for diffuse estimation.

Currently, it is also possible to import POA data as part of the custom file import (https://www.pvsyst.com/help/meteo_convdialog.htm). What this does is that it uses Hay for reverse transposition to get both GHI and DHI, and then uses these values for the simulation (as well as the Hay transposition). This means that the POA used in the simulation will be the imported one, however the decomposition in different irradiance components will be according to the Hay model.

We could add more decomposition / transposition models if necessary. We will establish this type of developments by first having a thorough look at the literature. I think your research would help us as a first step to assess the need to implement other transposition / decomposition models.