Michael Giberson
A pair of posts at Master Resource (part I, part II) explore the degree to which variable wind power leads to lower efficiency and increased air emissions when natural gas generators are used to provide energy balancing and back-up reserves (and except when and where sufficient hydropower is available, natural gas generation usually is the low cost provider of these services). Depending on assumptions, Kent Hawkins finds that adding wind power could result in no reduction of fossil fuel use and perhaps even an increase in related emissions.
The results are more dramatic than found by Warren Katzenstein and Jay Apt and published in Environmental Science & Technology, but the difference is primarily in the degree of the effect and not the nature of the issue. (Katzenstein and Apt found under certain circumstances that wind and solar power added to a natural gas-based power system acheived about 80 percent of expected CO2 reductions but no more than half of the expected NOx reductions.)
NOTE: Environmental Science & Technology has published a comment on the Katzenstein and Apt article by Andrew Mills and co-authors and then a reply by Katzenstein and Apt. The Mills et al. comment asserts that the methodology used overstates the need for backup power supplies and so at best indicates a possible upper bound on indirect emissions. The comment suggests that geographic diversity of widespread wind power facilities helps to smooth out some of the variability from individual sites, so studies based on a few units exaggerate the actual effects in larger power systems. In addition, unit commitment and dispatch practices by power system operators can accommodate some variability without contributing to added emissions.
In the reply, Katzenstein and Apt say the central issues are, “how are the fill-in generators to be dispatched?” and “what are the emissions from those generators in that dispatch method?” They note that Mills et al. adopt a different approach than the original article, but with either approach their are incremental emissions associated with the dispatch of the required fill-in generators.
UPDATE: The November/December issue of IEEE’s Power and Energy Magazine is devoted to wind power issues. Included is “Wind Power Myths Debunked,” which claims among other things, “the notion that wind’s variations would actually increase system fuel consumption does not withstand scrutiny.” Unfortunately, subscription required so the link only goes to an abstract. The article also reports another analysis saying that adding up to 20 percent wind may extract an efficiency penalty of no more than 7 percent (i.e., emission reductions may be 7 percent lower than expected).
One thing clear from these discussions is that answers to questions about the effect of wind power on emissions will depend very much on what else is going on on the power system to which the wind power is added.
UPDATE 2009/12/04: Kent Hawkins responds to some of these issues in another post up today at Master Resource.
I am the author of the posts at MasterResource. I respect the Katzenstein and Apt work, and my approach is broadly based on theirs. It is important to note the following quotes from their paper.
“As discussed in the Supporting Information, the emission and heat rate data we obtained for the gas turbines did not cover all combinations of power and ramp rate. We therefore further constrain the model to operate only in regions of the power-ramp rate space for which we have data. Here we focus on estimating the additional emissions caused by variability, and caution that we have made no attempt to ensure the stability of an electrical grid. Grid dynamic response may somewhat change our results.”
and,
“The fraction of expected emissions reduction, ?, is calculated assuming that the emissions predicted to be displaced originate from the same generator type that provides fill-in power: Figure 2a and b assume a LM6000 is displaced and a LM6000 is providing compensating power; Figure 2c and d assume 501FDs. Realistically, displaced generators will differ from the generators providing fill-in power and would produce different results.”
The LM6000 is an OCGT and the 501FD is a CCGT.
I cite their paper in the bibliography, because I presume to extend their work to include more considerations, albeit in a much less sophisticated manner, due in part to limited available data. With respect to the second quote, I find the impact of comparing CCGT alone (displaced generators) to OCGT/CCGT combinations (generators providing fill-in power) very dramatic. I produce similar results to theirs in scenario 2 of my Base Case.
Kent, Thanks for commenting. I meant to mention that your work was based on and cited the Katzenstein and Apt (K&A) analysis, but overlooked that point when writing.
I’m just now adding a bit about the comment in Environmental Science and Technology on K&A’s article and the K&A reply. I think the overall conclusion remains that specific answers will depend on characteristics of the power system to which the wind power is added (and the actual variability of net load in practice, etc.). Nonetheless, with existing power systems clearly one should not expect 100 percent displacement of fossil fuel emissions when adding variable renewable technologies to the generation mix.
This is a useful area for research as it can help see the indirect consequences of adding wind or solar to a power system.
“Leveling effects” of a geographically diverse band of wind generators dictates three types of grid upgrades that would otherwise be unnecessary – congestion relief, continental backbone, and the obvious feeders from remote wind sites to the backbone. All that statistical buffering doesn’t just materialize with the birth of the concept. It’s undertaking might spur “economic development,” but so would moving a pile of rocks to Halifax and back. In other words, if it isn’t broken, don’t pay to fix it.
As well, the statistical odds for rapid sags and spikes among all intermittent wind generation (let’s leave poor solar alone for now) may be reduced, but is never eliminated, and the amplitude prospects go up when more wind is connected, not down. As far as utilizing spinning reserve or standby simple cycle units, I can only say that the mechanisms in place today were built for a reason – to cover existing variability risks over very short to short time periods. The capacity factors are low, but if they really approached zero, the most recent peaker projects would never have been built. If the risk of variability rises during summer peak, then either we need more backup or we tolerate higher risk. if we choose more backup, take your pick – high ramping duty gas or prohibitively expensive storage.
But that’s engineering for you. A small, high hoop to jump through and they all begin to salivate. Just keep the accountants away.
Tom Stacy’s comments here are right on target. The grid is so sophisticated that it could integrate a cumquat if its political bosses insisted. But to what end? Adding the instability of wind flux to the existing instability of demand flux is little short of criminal for systems desiring reliability, security, and affordability.
The Katzenstein/Apt “study” may, as Kent Hawkins states, provide a template for further inquiry and as such prove valuable, since it stimulated his own work. However, in its present state it leaves much to be desired. For there are assumptions in it that are problematic, which Hawkins will no doubt discuss in greater detail–assumptions that are, at best, incomplete and, at worst, wrong.
Even so, Apt does recognize problems. He states, for example:
“That means that when a wind farm is built, some other power source of the same size must be built to provide power during those calm hours. Our research shows that natural gas turbines, that are often used to provide this fill-in power, produce more CO2 and much more nitrous oxide (as they quickly spin up and then slow down to counter the variability of wind than) than they do when they are run steadily.” He continues:
“Another problem is that wind and solar generation are variable. Wind speed changes from moment to moment and clouds block the sun, even in the desert. This variable power challenges the grid to provide reliable, high quality power when wind and solar are contributing more than a few percent of total generation.”
Beyond these caveats, which should, when investigated properly, yield much different results than the conclusion that balancing wind instability will only slightly reduce the amount of CO2 offsets wind energy induces–as Hawkins calculator provisionally shows.
And as Stacy implies, in addition to the problems of the companion generation necessary to mirror/shadow the wind flux, there is also the implication for what increasing amounts of wind volatility does for systems reserves.
It is becoming increasingly clear that when the study of wind behavior is contaminated by reality–and not projections that produce the same level of verisimilitude as college football polls–it will show that wind technology can no little to offset carbon emissions in the production of electricity. Engineers may well be able to ride those wild wind horses, but they will do so by increasing costs–both to consumers in the form of higher rates and to the atmosphere in the form of more carbon dioxide.