One of our readers noticed a discrepancy between the figures given at the EEX Transparency website and on Agora’s visualization. Today, we explain how the difference comes about.
Paul Neau is the founder of French renewable energy and environmental agency Abies. Recently, he sent me the following image, showing that some 28.2 GW of wind power was being generated on 23 December 2014 at 11 AM according to the EEX Transparency website.
23 December 2014
That figure is for the Phelix grid zone, however, which includes Austria. When you factor out Austria, the EEX shows 27.0 GW of wind power.
EEX 23 December 2014
Simultaneously, over at Agora we see 31.6 GW for the same hour.
Agora 23 December 2014
The discrepancy is tremendous – 4.6 GW, or roughly a sixth of the total. How does this huge discrepancy come about?
My first answer to Neau was that the EEX is a website that only publishes the minimum amount of data required by law. The data come from grid operators, and as the website itself indicates conventional systems are only counted if larger than 100 MW. Not all wind turbines and solar arrays are counted either – but then, Agora gets its data from the EEX.
My colleague Thomas Gerke knew the explanation. In a background document (the PDF seems to be only available in German), Agora explains that the EEX data only covered around 85.2 percent of total wind power production in the first half of 2014 based on the official AGEB numbers (which are not produced live), so Agora applies a correction factor of 1.17 – essentially increasing the EEX figures by a sixth. The assumption is only a rough approximation, however, and it varies from year to year. In 2013, for instance, the correction factor was 1.13. (Interestingly, no scaling is needed for solar because the EEX data covered 98.4 percent of total generation in the first half of 2014.)
Apparently, practically no live data are available for biomass, but the experts assume that such systems run basically at base load level and do not respond to price signals from the exchange because they receive feed-in tariffs. The live production levels are thus simply back-calculated from the installed capacity.
The coverage of lignite and nuclear in the EEX figures is quite high because such systems are generally large. In the first half of 2014, it was 96.2 percent for lignite (there are a few small cogeneration units fired with lignite) and 99.2 percent for nuclear, which seems to be a rounding error. The situation is much different for hard coal, however, only 67 percent of which was covered in the first half of 2014 – a clear sign of the numerous relatively small power and heat plants fired with hard coal in Germany.
Coverage of power from natural gas is dismal on the Transparency website at only 17.3 percent, and the Agora background document also mentions the obstacle of distinguishing between cogeneration and pure power generation – but that discussion would take up an entire article.
Other websites also extrapolate live data, such as SMA’s visualization of solar power production; the inverter manufacturer tallies production data from its numerous systems across Germany and extrapolates the estimate for the country as a whole based on its market share.
And then there is Fraunhofer ISE’s wonderful visualization. The explanations include the following statement:
"The data of the European Energy Exchange EEX does not cover the production of all power stations. Therefore the hourly power values are multiplied with monthly correction factors. The corrected values are displayed."
The individual scaling factors are not spelled out, but it is likely that Fraunhofer uses slightly different numbers for scaling than Agora does. That, at least, would help explain the variations between those two websites that I have commented on before.
In the end, we will have to live with the fact that these figures for live power production are educated guesses.
Incidentally, the EEX has also now added data for the Netherlands and Belgium. I’m not sure what the numbers represent, but they seem wildly unrealistic at the moment.