The team has tested over 70,000 solar modules at its calibration laboratory since 2012, analyzing 1,034 of the collected performance measurements taken from monocrystalline silicon modules under standardized conditions.
Analysis between 2012 and 2016 found measurement deviations between promised power and performance were always less than 1% on average, with some positive deviations also measured. Results from 2016 showed the difference between the manufacturer's power specifications and the measured power in the institute's laboratory was 0.6% on average.
Since 2017, a negative discrepancy between the specifications and results in the institute’s laboratory has emerged. The negative trend was found to be particularly pronounced in the years 2020 to 2023, culminating in an average power reduction of around 1.3% in 2023, with a positive deviation hardly ever observed.
Daniel Phillip, head of the Department Module Characterization and Reliability at Fraunhofer ISE, says a slight trend reversal was observed in 2024, but the average result still showed a power reduction of 1.2% against the manufacturers' specifications.
“If we assume that our data is representative of the German installation market, an average underperformance of 1.2% with an additional 16.2 GW in 2024 corresponds to a total output of around 195 MW,” Phillip added, which would equate to the nominal output of one of the largest solar parks in Germany.
Fraunhofer says the drop between 2023 and 2024 suggests manufacturing companies may have recognized the tendency towards optimistic power ratings as a problem. “The findings also make it clear how important a reliable, continuous and independent infrastructure is for the quality control of PV modules,” added Professor Andreas Bett, Institute Director at Fraunhofer ISE.
Fraunhofer says the data included in the study was filtered according to suitable criteria for the evaluation, with inconsistent data removed and further filters applied to exclude statistical distortions.
The analysis features solar modules from only the top ten manufacturers in each respective year and includes data only from projects where the client and the manufacturer were different, to achieve a result representative of module purchases.
