Comparing actual versus modelled POA irradiance data

[auspv]

Hello,

We are reviewing a utility scale solar farm (single-axis tracking) that has been operating for a couple of years. We have taken a complete record of the measured POA irradiance data and compared it with the PVsyst calculated POA irradiance (using measured GHI and other relevant actual weather data) for each comparable and valid hour. The ratio below is POA (PVsyst) divided by POA (measured) and then averaged for each hour and month. 

 

You can see quite good correlation for hours 10-13 but this correlation drops off more and more as we get to either end of the day. 

The conclusion we reach from this is either the model doesn't align with what has been installed or the trackers are not operating.

The alternate viewpoint could be that PVsyst doesn't produce good POA results across all hours which we do not believe is correct.

Keen to hear your thoughts on this?

Thanks.