Over two weeks ago Migeru alerted us to a paper by authors C. le Pair, F. Udo and K. de Groot with the title "Wind turbines as yet unsuitable as electricity providers", which, unsurprisingly, does a hack job on wind power with an appearance of seriousness, collating earlier think-tank work and mixing in every anti-wind evergreen from intermittency to lowered real estate prices. I now remembered and looked a bit at its central claim: that less efficient operation of fossil fuel plants results in as much extra energy use and emissions as what wind power is supposed to displace. This is nothing complete so I post it only as a comment.

First of all, IMHO their claim is as much about giving a compelling rhetorical framing as it is about giving the foundation for a shoddy calculation, so let's first attack that compelling framing.

• The argument is compelling as long as people think that fossil fuel plants have to react to wind intermittency – forgetting that they have to react to demand variation, as well as planned and unplanned shutdowns of big power plants; and now solar power changes the picture, too.
  
• Moreover, the efficiency calculation shouldn't be done without some plants staying in cold shutdown: its not like all load-following plants follow the same power curve, rather, some go down while others can continue operating at a higher power (and thus more efficiently).
  
• The study makers do actually have the gall to add construction-related emissions for wind, but not for new conventional plants.

The Pair-Udo-de Groot study heavily references an earlier anti-wind screed, a study by energy consultant Bentek (which is connected to the gas industry). I don't have to pick that one apart as others have done so, AWEA has a good summary (see the pdf opening from the linked summary page). I'm quoting one part, a list of methodological errors, some of which echo or complement what I wrote above:

• Wrongly assuming that all variability on the power system is caused by wind energy, even though variability in electric demand has a far larger impact.
  
• Cherry-picking outlier events for wind variability and then extrapolating them as normal behavior.
  
• Cherry-picking an isolated instance of the pollution control technology at a coal plant failing, blaming wind energy for the emissions increase, and then extrapolating that as the normal impact of adding wind.
  
• Understating the emissions reduction benefits of wind energy in California by excluding the fact that wind is actually displacing the state's large imports of coal power.
  
• Greatly understating the emissions benefits of wind in the Pacific Northwest and California by ignoring the fact that displaced hydropower is stored behind the dam and used later to displace fossil fuel use and emissions.
  
• Only looking at hourly snapshots of the power system and therefore excluding the vast majority of the emissions savings when wind energy causes emitting power plants to turn off for an extended period of time.
  
• Failing to model the process by which grid operators actually choose which power plants to operate. That type of analysis is the only way to accurately assess the impact of adding wind, or any other resource, to the grid.

I suspect that the fourth and the before-last points are particularly potent tools of mis-estimation in a European setting, too.

The AWEA paper also references actual studies (not simulations) on the effect of wind power on electricity sector emissions, which found even greater reductions than direct replacement (that is, the net second-order effects are positive rather than negative), explained by a shift from coal to gas. On this front, it's worth to mention the situation in Ontario, where electricity sector greenhouse emissions reduced by more than 70% in eight years, all the while comsumption increased and Ontario turned from net importer to net exporter: coal's share dropped from 21% to 3% (and that's not counting coal's share in 2003 imports).

Another source they use is the thesis of B. Ummels. They don't mention the conclusions Ummels himself drew, which have been much more positive: he says large amounts of wind power can be integrated, and international exchange alone is a sufficient enabler for the Netherlands. They don't mention Ummels's disclaimer on the so-called load duration curves, either, which makes them unfit for the purpose Pair-Udo-de Groot use them:

...There are however important limitations to the application of duration curves for generation despatch, since duration curves incorporate neither chronology nor consider actual generation unavailabilities [106]. Since load levels are re-organised in a decremental order, all chronological aspects (minimum uptimes and downtimes, ramp rates etc.) are lost. This makes an analysis of power system operation difficult.

_{*Lunatic*, n.
One whose delusions are out of fashion.}