Figure 1: Schematic overview of the elements of the PVSAT-2 procedure.

To determine the expected energy yield of a PV system, satellite data and ground based irradiance measurements give information about the solar resource at the site of the PV system. These data are used because local measurements by pyranometers are costly for operators and need periodic maintenance.

First, surface irradiance is derived from METEOSAT-7 images (later METEOSAT-8) with the Heliosat method (Hammer, 1999). Satellite images have a very good spatial and temporal resolution what makes them very suitable for PV applications. To achieve a higher accuracy the satellite measurements are combined with ground measurements by the geostatistical method kriging-of-the-differences (Betcke and Beyer, 2004). These irradiance values are input for a PV simulation (Beyer et al, 2004). According to the submitted plant description, in the second step a PV simulation will determine the expected energy yield of the day.

Here, the central decision support tool carries out the daily performance check. It compares the expected and recorded energy yields on the daily and hourly basis individually for each PV system. It decides on the occurrence of a failure. The detectability of a malfunction depends on the analysis of hourly values (e. g. shading) or is detectable by evaluation of daily values over a longer period of time. In case an error occurred, the
footprint algorithm, part of the error detection routine, determines its cause. This whole procedure of error detection is hardly influenced by the accuracy of the input data.

If the malfunction is found, the operator of the PV system will be informed automatically as part of the automated procedure.