225et al. (2019), the curtailment level was 12.6 TWh in 2014 (8%), 33.9 TWh in 2015 (15%), and 49.7 TWh in 2016 (19%). As seen in Fig. 5, both the absolute level of curtailment and curtailment as a percentage of total production increased between 2010 and 2016. There have been indications in nonacademic reports that curtailment rates have fallen sharply since 2018, although trustworthy sources are lacking. It is plausible that the curtailment rate has fallen. Curtailments were growing and in the long run should be noticed and remedied. Grid development and connection in some years lagged installed capacity by over 30%, making wind power generation exceed the acceptance of the grid and creating abandonment and grid instability (Zeng et al., 2015; Zhang et al., 2017).

Curtailment levels exceeded those of other countries. For example, Chinese wind turbines produced 1787 full load hours on average in 2007, which was substantially below turbines in the United Kingdom (2628 h), Australia (2500 h), and the United States (2300 h) (Sahu, 2018). On average, Chinese wind power plants were shut down up to 15% of the time, whereas, in Germany, employees are laid off just under J. Grafström

226250

20

200

15

Percent

TWh

150

10

100

5

50

0

0

2010

2011

2012

2013

2014

2015

2016

Total Wind

Curtailment level

Percent

Fig. 5 Wind curtailment as share of production. Source: Own calculations based on calculations Luo et al. (2016), Zhu et al. (2019), and IRENA (2021)

0.5% of the time (Karltorp et al., 2017). If the average of 1787 full load hours in 2007 had remained, there would have been no economic surplus in 2013. An example from Zhu et al. (2019) indicates that to break even in Guazhou county of Gansu, windfarms needed to produce 1800 h, which they did in 2013 (1859 h), but in 2016