Average time shift on clear days

[ckoessler]

Hy

When importing hourly data sets ( i.e. long term correlations based on weather data from Meteo France)) I observed that average time shift on clear days is quite similar to the difference between TSV/UTC or the Time Equation in general. The following figure is an example:

https://drive.google.com/open?id=0B_jtkQGMQ2rqdXZ5RHRXTFRnTlE&authuser=0

Data from Meteo France has UTC format.The clear sky model applied by PVsyst is based on astrological algorithms. Consequently, there might be a difference between your model and observations made by Meteo France.

Therefor I'd like to know if this observation can have an impact on simulation results?

If I am not mistaken, differences between your clear sky model and hourly data for global irradiation provoke a difference in the diffuse fraction calculation.

Is there a way to minimize errors or can we estimate the uncertainty of results due to this problem?

[André Mermoud]

The Clear day model shown in PVsyst gives a reference of the time at which the solar geometry is computed within the software.

Imported data should match the internal time of PVsyst. This is usually a difficult question, explained in the help "Meteorological data > Hourly Meteorological Data > Time shift".

When importing hourly ASCII files, PVsyst allows to specify a correction for the real measured data time with respect to the internal PVsyst time.

But sorry, a correction of data written in UT is not yet available in this tool.

[ckoessler]

Would it be possible for you to detail briefly the internal PVsyst time?

N.B. Here below,I attached the image I was talking about in my first post.

[André Mermoud]

These data will be difficult to accurately manage within PVsyst, as the time shift is highly varying along the year.

This indicates that your datalogger's time reference has shifted, and the data are difficult to accurately recover (not possible in PVsyst).

In this case the best choice would be to specify a time shift of around -15 min, so that the time variation of the data with respect to the average will be +/- 15 min.

Now such a variation could be explaind by the fact that you imported the data using the "Solar time". Please make sure that you have specified "Legal time" in the second page of the import format.

[ckoessler]

The indicated time format for the import is "legal time", but there might be a problem with the time format for some years of our weather station data.

Is is known that there is no impact on cumulative daily global radiation by the choice of TSV or UT. But there is actually an impact on cumulative hourly global radiation due to the time-shift.

Can this have an impact on diffuse part calculation in PVsyst?

Are you using the daily or hourly clearness index for calculating the diffuse part of radiation?

[André Mermoud]

The relevant time is the time for the calculation of the solar geometry (should be in the middle of the accumulation interval).

Yes indeed, this will have an impact on the diffuse evaluation, and much more on the transposition model.