From Trust, yet verify

In previous post, I was looking for two Dunkelflautes (periods of low wind and not much sunshine) in the German grid data. I looked specifically into the first half of November 2024 and around half December 2024, this because it was reported that Dunkelflautes occurred in those two periods.

A commonly used definition of a Dunkelflaute is a period with a daily production of less than 10% of installed capacity. Unfortunately, I didn’t have the exact capacity information for November and December 2024. A LinkedIn post that analyzed Dunkelflautes in Germany used the installed capacity of the beginning of the year and found 12 consecutive days in November and 3 consecutive days in December. I redid that analysis but with the installed capacity at the end of 2024 (which should be more appropriate for the end of the year) and found 14 consecutive days in November and 4 consecutive days in December.

The actual number of Dunkelflaute days will depend on how much capacity was available at the end of 2024. When all the new capacity came available in the second half of December, then that would be 12 days in November and 3 days in December, as found in the LinkedIn post. When that capacity was already available at the beginning of November, then that would be 14 days in November and 4 in December. When some of that capacity was coming available in November or December, then that would be something in between, depending on how much capacity still had to come available at the end of the year.

I limited myself in previous post to just those two date ranges, but the winter period in general is known for having Dunkelflaute events, so that made me wonder how frequent these occur. So, let’s just extend the same analysis from November 2024 until now.

The Agora data from 1 November 2024 until 28 March 2025 (when I finalize this post) gives this:

I used the installed capacity at the end of 2024 for the entire period (November 2024 → March 2025) in my calculation. Also here, the actual number of Dunkelfaute days in the first three months of 2025 will depend on how much capacity came available. When there was already new capacity available in that period, then there might be more days having less production than the 10% threshold than what I calculated.

The longest Dunkelflaute is still the one of the first half of November 2024. There were however more periods than I expected. I count 14 periods with a production of less than 10% of installed capacity, of which 5 are single day and 9 spanned consecutive days:

	Start date	End date	Duration (days)
1	02 Nov 2024	15 Nov 2024	14
2	29 Nov 2024	29 Nov 2024	1
3	04 Dec 2024	04 Dec 2024	1
4	10 Dec 2024	13 Dec 2024	4
5	24 Dec 2024	29 Dec 2024	6
6	04 Jan 2025	04 Jan 2025	1
7	12 Jan 2025	13 Jan 2025	2
8	15 Jan 2025	22 Jan 2025	8
9	01 Feb 2025	06 Feb 2025	6
10	13 Feb 2025	17 Feb 2025	5
11	25 Feb 2025	01 Mar 2025	5
12	10 Mar 2025	14 Mar 2025	5
13	24 Mar 2025	24 Mar 2025	1
14	26 Mar 2025	26 Mar 2025	1

I expected some periods of Dunkelflaute, but not that many! From the 148 days between 1 November 2024 and 28 March 2025, there were 60 days with a daily production of less than 10% of the installed capacity. That is 40% of the days in the period. That is much more than I expected. The average capacity factor was only 11.8% over the entire period. This in a period of the year when demand is at its highest.