Debating Wind Power Cost Estimates – 5

[Series header: On the Morning of October 15 the Institute for Energy Research in Washington DC released a report I’d written about the federal government’s wind power cost estimates. (Links available here.) Later that day Michael Goggin of the American Wind Energy Association, the lobbying organization in Washington DC that represents the wind energy industry, posted a response on the AWEA website: “Fact check: Fossil-funded think tank strikes out on cost of wind.” I’m considering points made by the AWEA response in a series of posts.]

In the final section of Goggin’s detailed criticisms of my report he takes on my claims with respect to various additional costs associated with the addition of wind power to the grid, including grid integration costs, indirect pollution effects, transmission expenses, and negative prices. He writes:

After starting with a baseline wind cost that is 100% too high, IER compounds the error by claiming that the actual costs of wind are even higher based on obsolete data and a flawed understanding of how the power system works.

IER incorrectly alleges that wind energy imposes large “integration costs” on the power system. In reality, it is far more costly to integrate the unexpected and instantaneous failures of large fossil and nuclear power plants than to accommodate the gradual and predictable changes in wind energy output.

I’m note sure just where the report “alleges that wind energy imposes large ‘integration costs’ on the power system.” All that my report does is (1) observe that grid integration costs are not included in NREL levelized cost of energy estimates so a fuller consideration of costs much include them, (2) summarize the discussion of the topic in the Lawrence Berkeley National Lab’s 2012 Wind Technologies Market Report [WTMR], and (3) highlight factors that tend to increase or decrease those costs.

Here is the core of my wind integration cost claim: “The [WTMR] reported a range of cost estimates from wind power integration studies, with all studies but one falling below $12 per MWh and some studies below $5 per MWh.” From this remark somehow Goggin claims I allege the costs are large.

Goggin then cherry-picks a few examples of low wind integration cost estimates. But each of these examples is included in the far more comprehensive WTMR study produced by the Berkeley Lab. You can find his 3 examples, and 22 others, in figure 37 of the 2012 WTMR, p. 63. The American Wind Energy Association may not like the answers, but again it seems that Goggin’s complaint is with the Berkeley Lab research and not my report.

Goggin again:

IER’s report falsely alleges that wind energy’s pollution reductions are significantly reduced because of this incremental need to operate other power plants more flexibly. IER picked a bad time to once again try to push that myth, as last month a comprehensive report used real-world emissions data from every power plant in the Western U.S. to confirm that wind energy produces the expected pollution reductions. … IER’s claim to the contrary is based on a single report that has been thoroughly debunked for getting the wrong answer because its authors failed to understand how the power system works. [Link in source.]

Goggin discusses issues raised in section 3.3 of my report on additional cycling of baseload units and section 3.4 of my report on environmental costs. I cite a handful of references in these two sections and Goggin doesn’t include links. As best as I can tell by “single report” Goggin is referencing the Katzenstein and Apt article published in the journal Environmental Science & Technology, “Air Emissions Due to Wind and Solar Power,” and the “thorough[] debunking” is the comment on that piece by Mills, Wiser, Milligan and O’Malley.

On this point, while Goggin exaggerates his point in cartoonish fashion, he raises a good point. The nub of the issue is that the Katzenstein and Apt article use a very simplified case to examine the relationship between renewable power intermittency and emissions from dispatchable generators, and the simplified case yields a much higher reduction in emission benefits than renewable power intermittency actually yields when connected to large scale power grids. That is to say, as Katzenstein and Apt acknowledge and Mills et al. emphasize in their comment, the Katzenstein and Apt result is essentially an estimate of the possible upper bound of the effect. Mills et al. make clear that in actual power grids the reduction in emission benefits, while still present, is likely much smaller in practice. In short, the study I emphasize was not the best choice to show the indirect emission effects of renewable energy intermittency in large scale power grids. (Having met Jay Apt once or twice, I’d be very reluctant to accuse him, as Goggin does, of failing to understand how the power grid works.)

Goggin objects to my referencing transmission costs as another factor to be considered as associated with wind power, since “upgrades to the nation’s obsolete and congested electric grid are needed anyway regardless of the addition of wind energy” and transmission upgrades will more than pay for themselves by broadening access to low cost generation. I’m sure Goggin understands enough about how the power grid works to understand that “upgrades to the … grid … needed anyway regardless of the addition of wind energy” will be somewhat different from “upgrades to the … grid … needed” because of the addition of wind energy. Perhaps amusingly, the Western Wind and Solar Integration Study Goggin cites against me on the emissions point tends to support my point on transmission costs: in Phase 1 of the study they assume significant enhancement of the grid in the Western U.S. to accommodate assumed addition of large amounts of wind and solar power.

If modelling assumptions don’t convince Goggin, then surely he has heard of the $6.8 billion CREZ grid upgrades in Texas that were designed accommodate existing and projected wind power production. Most of the CREZ upgrades would not have been useful in the absence of wind power and certainly the ERCOT grid would not have been expanded to overlap the Southwest Power Pool grid footprint in the Texas Panhandle and South Plains area in the absence of a high-quality wind power resources in the region. Transmission upgrades can enhance competition, like Goggin points out, but had ERCOT wanted transmission upgrades primarily to enhance competition then the money would have been spent much differently. The grid upgrade plans and the associated expenses were largely driven by the fact that high-quality wind power resources are location dependent, and those locations are distant from the primary areas of power demand in the state.

Obviously the selection of power plant location is important for any kind of generator, and good locations will always be constrained (usual main factors: access to fuel, access to water for cooling, access to consumers, and cost of land). But for coal, nuclear, and natural gas it is possible to deliver the energy resource to locations nearer ultimate consumers. In the cases of wind, hydropower, and geothermal energy the resource locations are determined primarily by nature (and not with net-system-cost minimization in mind).

Next: Two issues remain, both concerning the effects of wind power on regional power market prices. I’ll look at the price suppression effects of adding wind to the grid in my next post in this series and then I’ll take another look at negative power market prices.