I'd like to point out that none of this is fundamentally insurmountable, it's more a disagreement about what the real cost is going to be. And the level of difficulty.
We judge that much - perhaps all - of our disagreement stems from
differing assumptions, rather than dispute over the factual content such
as the cost and performance of wind turbines or the cost of
long-distance transmission. With this letter, we aim to make our
assumptions explicit and then respond to Jacobson and Masters'
critique of our letter to Science.
We assume the following: 1. Wind energy could realistically effect
deep reductions in the environmental damages (air pollution, CO2)
imposed by fossil-based electric power systems. 2. In response to the
CO2-climate problem, we expect that it will be necessary to make
deep reductions (over 50%) in electric sector emissions. We are
interested in estimating the cost of wind if it were to supply a
substantial fraction, on the order of one-fourth, of U.S. demand. 3. If
wind is to be exploited at very large scales (hundreds of gigawatts of
output), we anticipate that environmental, aesthetic, and economic
considerations will dictate that the bulk of the wind capacity be located
in the windy regions of the Great Plains.
Below we address the critiques you raised regarding our letter.
1. Hirst’s analysis and intermittency. We were impressed by Hirst’s
analysis, “Interactions of Wind Farms With Bulk-Power Operations
and Markets,”(1). The paper analyzes import of wind energy from the
Lake Benton site in southwestern Minnesota to the PJM grid. The
analysis is, however, not pertinent to our disagreement about the cost
of intermittency because it treats a case where the wind power supply
is too small to significantly influence the power market. The Benton
array has a small capacity (~100 megawatts) and is being imported
into a massive grid capable of supporting a peak load of 52 gigawatts.
Hirst addresses this issue by adding a wind multiplier parameter, but
his analysis still only extends to wind serving less than 10 percent of
generation. Hirst’s general conclusion only supports our intuition: “as
the size of the wind farm increases relative to the control area, the
average price it receives for its output declines.”
2. The economics of backup when wind is baseload. There is an
additional complication not presented in the Hirst paper that is only
relevant when wind is treated as baseload capacity. Although
geographically dispersing turbine arrays can decrease the variance in
wind power output, there will still be times when turbine output is
minimal. Therefore, there must be a significant amount of backup
capacity or storage. But because many of these generating or storage
units will be used infrequently only when the wind doesn’t blow, there
use will be small and the amortized cost will be spread over fewer
kilowatt hours of production, making the incremental cost of backup
very expensive. Given points 1 and 2, we think your suggestion that
5 of 19 1/29/2002 4:58 PM
Science -- Published dEbate responses for DeCarolis et al., 294 (5544) 1000-1003 wysiwyg://270/
http://www.sciencemag.org/cgi/eletters/294/5544/1000the cost of intermittency is of order 0.05¢/kWh is implausible. We
think our disagreement here is completely driven by differing
assumptions about wind’s fraction of electric capacity.
3. Correspondence between wind resources and the existing grid. We
do not dispute your statement that several hundred gigawatts of wind
resources exist within 10 miles of existing transmission infrastructure
(2). However, we think that this may not be relevant for three reasons
detailed below.
(a). Economic considerations. Exploiting wind resources close to
existing transmission grids is not necessarily the most cost-effective
solution. Because wind turbine output exhibits a cubic dependence on
wind speed, wind power output is very sensitive to location. For
instance, it may be true that installing 10 gigawatts of turbine capacity
in the Pembina Escarpment of North Dakota, a wind class 5 area, and
transporting the electricity to the PJM grid via HVDC lines is roughly
equivalent in cost to simply installing the wind turbines in southwestern
Pennsylvania, in wind classes 3-4 and neglecting transmission costs.
For the same reason, we do not believe it is coincidental that Hirst
chose to look at wheeling wind power from Lake Benton, a wind class
6 site, to the PJM grid.
(b). Transmission considerations. In addition to considering the
location of wind turbines with respect to the existing grid, a
comprehensive assessment of existing transmission and distribution line
capacity of the local grid must be performed, as your reference clearly
indicates (2). We would wager that the existing grid located near the
Pembina Escarpment would not support the hypothesized 10 gigawatts
of additional electric power from new turbine arrays. As such, we still
believe that long-distance HVDC transmission lines would be a critical
component of large-scale wind. Jacobson and Masters say that the
cost of 1.5 ¢/kWh “is not supported by the actual cost of transmission
lines,” but they provide no reference to other estimates of HVDC
costs. We can cite many studies that show amortized HVDC costs to
lie in the 1-2 ¢/kWh for these distances.
(c). Aesthetic considerations. Although there are substantial wind
resources near population centers (and the grid), we are skeptical that
these would be developed at large scales. For example, where we live
in western Pennsylvania, there are substantial wind resources located
on the mountain ridges, and in principle these could supply power to
the PJM grid. However, to supply substantial power a developer
would need to use almost all the ridge tops, which we believe would
be unacceptable to local residents. We judge that aesthetic and
environmental concerns would push large-scale wind into the Great
Plains.