From a pointer from “Bernie” in comments at Bishop Hill’s web site, I took a look at the annual load factor for wind generation in Denmark based on data published at Wikipedia. I chose to do the analysis starting in 1990 as it appeared from the data that the first decade of wind power in Denmark was low–probably caused by the learning factor effect.
The probability distribution is shown here, along with the raw data from Wikipedia. The mean over this 20 year period is about 21%. The trend is generally upward, but there is variation represented in the probability distribution.