E-Letter responses to:

letters: Joseph F. DeCarolis, David W. Keith, Mark Z. Jacobson, and Gilbert M. Masters The Real Cost of Wind Energy Science 2001; 294: 1000-1003 [Full text]

E-Letters: Submit a response to this article

Published E-Letter responses:

Response to letter by Howard Gruenspecht of November 21, 2001

Mark Z. Jacobson and Gilbert M. Masters   (28 November 2001)

Response to DeCarolis and Keith response of November 21, 2001

Mark Z. Jacobson and Gilbert M. Masters   (28 November 2001)

Response to Alfred Cavallo letter of November 21, 2001

Mark Z. Jacobson and Gilbert M. Masters   (28 November 2001)

Re: Re: Wind vs. Coal

Howard Gruenspecht   (21 November 2001)

Response to Jacobson and Masters

Joseph F. DeCarolis and David W. Keith   (21 November 2001)

Re: Re: Wind vs. Coal

Alfred Cavallo   (21 November 2001)

Re: Wind vs. Coal

Mark Jacobson and Gilbert Masters   (14 November 2001)

Wind vs. Coal

Dan S. Golomb   (14 November 2001)

Don't Dismiss the Midwest's Power Needs

Dr. Josh Kurutz   (2 November 2001)

Response to letter by Howard Gruenspecht of November 21, 2001 28 November 2001
Mark Z. Jacobson and Gilbert M. Masters, Associate Professor (MZJ); Professor (GMM) Department of Civil and Environmental Engineering, Stanford University Respond to this E-Letter: Re: Response to letter by Howard Gruenspecht of November 21, 2001	We thank Gruenspecht for his comments. We will address them, below. First, Gruenspecht makes the same point as DeCarolis and Keith, that when wind is a large fraction of the energy production, more backup energy may be needed. We addressed the issue in detail in our Response 2 to DeCarolis and Keith's response of 21 November 2001, so we do not repeat it here. Second, Gruenspecht says that 10,000 km provides only 30 km of transmission per gigawatt of wind capacity. This statement is incorrect. One transmission line can transmit 2 GW of actual power = installed GW / capacity factor. For a CF of 0.36, this translates to an installed power of 5.5 GW. Thus, for 225,000 1.5-MW turbines, about 61 lines are needed, and 10,000 km / 61 lines = 163 km/line. As such, for transmission to cost <1% of direct wind cost, the average transmission line for 225,000 turbines can be up to 163 km. This number may be reduced to around 100 km/line to account for overlapping transmission during times when wind is peaking and for other factors. The 100 km/line is consistent with a number derived independently in our response to Cavallo's 21 November 2001 letter, which is based on an earlier analysis by Cavallo, who accounted for overlapping. Even if 50,000 km of transmission lines are required (500 km/line) the net cost is still <5% the direct cost of wind energy (3 to 4 cents/kWh) and <1.5% the price a typical consumer pays for electricity, which is 11 cents/kWh. In addition, we disagree with the implication that the Siemens et al. bid of $5 billion for transmission lines out of a $15 billion project means that transmission is one-third the cost of wind energy. The transmission lines will have lifetimes of 40 to 60 years; the wind turbines, 20 years. The $5 billion investment in transmission lines will survive two to three generations of turbines. In addition, this example applies only if the transmission lines are long, which is not always necessary. Third, Gruenspecht statement that even old coal-fired plants produce electricity at 1 to 1.5 cents per kWh direct cost appears low. Their costs are often cited as 2 to 3 cents/kWh. Regardless, the direct cost is not the cost to the U.S. citizen. The cost to the U.S. citizen is the total cost: the direct + health/environmental + subsidy cost. The health/environmental costs of old coal power plants exceed those of new coal power plants, and both far exceed the total cost of wind. Thus, our conclusion that the direct+health/environmental cost of wind is less than that of coal (whether old or new) still stands, and this is the only cost comparison that matters from a public policy point of view. Fourth, Gruenspecht correctly states that most new capacity is natural gas. The direct cost of a new natural gas power plant is 3.3 to 3.6 cents/kWh (Office of Fossil Energy, 2001). Natural gas emits carbon dioxide, methane, carbon monoxide, nitrogen oxides, particulate matter, reactive organic gases, ammonia, and other pollutants that exacerbate global warming, urban smog, particulate health problems, acid deposition, and visibility degradation. We calculate a tentative, conservative global warming cost of natural gas as 0.7 to 1.1 cents/kWh and other health/environmental cost as 0.5 to 1.1 cents/kWh, which gives the direct+health/environmental cost of natural gas as 4.5 to 5.8 cents/kWh, more expensive to society than wind (3 to 4 cents/kWh) but less expensive than coal (5.5 to 8.3 cents/kWh). As such, we believe our conclusions apply to both coal and natural gas. Fifth, Gruenspecht states that most black lung cases reflect past, not current, mining practices, and the number of black-lung deaths would not be appreciably impacted by the prospective reduction in coal use. This argument glosses over a real problem and misrepresents our point. First, the federal government still pays hundreds of millions of dollars per year in black lung subsidies, and miners in the U.S. still contract black lung disease by working in coal mines today. Miners, globally, contract black lung disease at higher rates. Second, the cumulative federal black-lung payments, brought to present value, are around $70 billion. The cumulative subsidy has allowed coal to gain an advantage in pricing and lobbying, and this advantage has, in turn, resulted in greater pollution output. For wind to obtain a level playing field, we argue that the federal government should spend the same $70 billion by purchasing wind turbines (or that coal should pay back the $70 billion). This amount alone would allow the replacement of 10% of coal with wind. Unlike with the black lung subsidy, the government could recoup its entire investment if it purchased turbines and sold the electricity. References and Notes Office of Fossil Energy, Department of Energy; see http://fossil.energy.gov/coal_power/special_rpts/market _systems/market_sys.html, 2001.
Response to DeCarolis and Keith response of November 21, 2001 28 November 2001
Mark Z. Jacobson and Gilbert M. Masters, Associate Professor (MZJ); Professor (GMM) Department of Civil and Environmental Engineering, Stanford University Respond to this E-Letter: Re: Response to DeCarolis and Keith response of November 21, 2001	We thank DeCarolis and Keith for their important comments and for furthering this debate. We believe that we have reached convergence on several issues and that some issues will not be completely resolved at this time. On other issues, though, we still respectfully disagree, and we believe that the conclusions of our original paper still hold. Below are responses to DeCarolis and Keith's points in their response of 21 November 2001, in the order they are given. (Point 1) Hirst's (2001) analysis addressed one aspect of intermittency, the cost of regulation ancillary service. He found the cost to be small (0.005 to 0.03 cents/kWh, less than 1% of the direct cost of wind energy). A similar study was performed by Hudson et al. (2001) who also found that the cost of regulation ancillary service to be small when wind is integrated into a grid (0.006 cents/kWh). (Point 2) As DeCarolis and Keith correctly point out, there is a second issue related to intermittency, and this is the potential cost of supplying backup energy when wind becomes a large fraction (e.g., 30%) of energy supply and wind's output is low for a given hour. The real question here, though, is not what is the cost to wind, if any, in this case, but what is the difference between the cost to wind and the cost to coal or natural gas (since we did not account for this potential cost with respect to either coal, natural gas, or wind in our paper). We suggest that the cost to wind due to backup reserves could be less than or more than that to natural gas and coal and the net cost, either way, is uncertain. As such, we believe it is incorrect to presume a cost or a benefit as DeCarolis and Keith have done. Before the work of Hirst and Hudson et al., it was commonly presumed that wind had a high cost of regulation ancillary service. This assumption turned out to be incorrect. Similarly, it should not be presumed that expansion of wind energy will result in a higher cost of contingency reserves than the current cost. The main reasons we believe the net difference in contingency reserve costs could be either negative or positive are given as follows. First, backup sources of power are already in place and are used when natural gas or coal power plants fail, supplies tighten, or energy demand increases beyond expectations suddenly. The forced outage rate for all fossil fuel power plants is, on average, 8% (North American Electric Reliability Council, 2000). This compares with a failure rate for modern wind turbines of 2% (Danish Windpower Manufacturers Association, 2001). As such, if wind displaces coal, the reliability rate of replaced energy, in terms of energy source failure alone, will improve immediately by 6%. DeCarolis and Keith use in their example a peak load of 52 GW. If 30% of 52 GW is supplied by replacing coal with wind, backup requirements for failure will be reduced by 1 GW. Replacing natural gas with wind will result in a proportional reduction. Second, whereas wind is an intermittent energy source, natural gas is also an intermittent energy source. This is evidenced by the variations in natural gas prices of 50 to 100% from month to month and year to year (e.g., McFeat, 2001). This price variation is caused in large part by a variation in natural gas supply. The variability of natural gas supplies and prices suggests that, if wind replaces natural gas, backup requirements may not change much. Third, peaker plants are used commonly today when energy demand exceeds expectations. Thus, a certain amount of backup is already necessary, regardless of the energy source. Fourth, there are several ways to provide backup energy. Some include increasing hydroelectric output, transmitting from outside the grid, using peaker plants (usually fossil fuel), and storage. (i) Hydroelectric power supplies 10% of energy in the United States (only 4% outside of California, Oregon, and Washington). When hydroelectric output is increased as a backup, there is little additional cost. (ii) If wind becomes 30% of the energy supply, wind farms will be distributed over greater areas, and grid interconnections will expand, enabling easier transmission of excess wind, solar, hydroelectric, and fossil energy from outside the local grid, thereby reducing and potentially eliminating the need for peaker plants for backup. In other words, the expansion of wind energy may reduce the cost of backup energy by enlarging the size of a grid and by facilitating transmission of excess wind and other types of energy from outside the grid when needed. (iii) A large future energy requirement may be to generate hydrogen for fuel cells. In such a case, intermittency is no longer an issue, and only total energy output over a year is. Wind is reliable for producing an aggregate amount of energy over a period of a month to a year. (Point 3)(i) Whereas we agree that the Great Plains contains the largest concentration of ideal wind sites, there are plenty of land and water sites with equal or greater wind power that have still not been exploited. The fastest wind sites in the country (>9.4 m/s on average) are all along the coast, such as of North Carolina, South Carolina, and and Louisiana (Archer and Jacobson, 2001, first figure - please note that the figure gives speeds at the measurement locations only). Sites exist in numerous states that are very fast. (ii) For a distance of 2000 km, we estimate the cost of new HVDC lines as about 0.7 cents/kWh (see footnote), which is about half the average value given by DeCarolis and Keith (although the uncertainty is within the margin of error of their estimate). However, we disagree with their assumption that most lines need to be 2000 km. A total of 840,000 MW of wind power lie within 20 km of existing transmission lines. Our proposal requires the generation of only 128,000 MW of power (225,000 1.5 MW turbines in the presence of 7 to 7.5 m/s winds, giving capacity factors of 0.35 to 0.4). A reasonable portion of our required power can be obtained from turbines close to existing transmission lines. If such lines are already saturated with power, the cost of additional lines is not a cost of wind exclusively but a shared cost among all energy sources using the lines, because coal and natural gas generally do not own such lines, and therefore, do not have an exclusive right to them. (iii) Siting almost all turbines in the Great Plains would at first glance make sense, but the larger the area that turbines are distributed over, the higher the minimum power output summed over all turbines (please see Archer and Jacobson, 2001) and the lesser the contingent backup energy required to account for a worst-case scenario for wind. In other words, if all turbines are placed in one region, a high pressure system could cause slow winds in that region. If turbines are placed in many areas of the country, the chances are slim that all regions will have slow winds at the same time. In addition, our speculation is that most development will ultimately occur offshore, since offshore area is essentially unlimited, most people live near the coast, winds are generally faster over water than land, winds are very regular and predictable near the coast, turbines can be placed far enough out that people don't see them, and new turbines cause minimal environmental damage (the large, slow-moving turbines do not cause bird loss any more). (Point 4) DeCarolis and Keith say, "...but to assert that the cost of wind energy is low less that of coal is not accurate. If it were, we would expect to see wind dominate virtually all new capacity installations (given the 1.5 cents/kWh tax incentives), rather than simply having the fastest relative growth rate..." First, our paper compared the direct cost of new coal with that of new wind, not the direct cost of old coal with new wind. We concluded that the direct costs of new coal and new wind are comparable, in the 3 to 4 cents/kWh range, and this conclusion is supported by Bolinger and Wiser (2001, p. 3), who calculated the 25-year real costs of 17 wind farm proposals in California in 2001 as 3.2 to 3.7 cents/kWh, with a weighted average value of 3.6 cents/kWh. Their analysis also stated that the numbers were based on proposal information that presumably contained worst-case estimates for wind. Second, why should wind dominate coal or gas in the marketplace when the direct costs are similar in both cases. Whereas wind receives a Production Tax Credit, coal receives a percentage depletion allowance for mining operations, deductions for mining exploration and development costs, special capital gains treatment for coal and iron ore, a special deduction for mine reclamation and closing, research subsidies, and black-lung benefits paid for by the federal government. Oil and gas (mined together) receive a percentage depletion allowance, a 15% credit for enhanced oil recovery, a deduction for intangible drilling and development costs, a "passive loss" tax shelter for investors in oil and gas, a nonconventional fuel production credit, and research subsidies. Third, the Production Tax Credit for wind can be fully realized only if the price of wind energy exceeds the cost of wind energy by 2.5 to 3.75 cents/kWh. As such, either wind producers benefit only partially from the credit or the Credit itself drives up the price of wind. This is easily proven: In order to fully realize the credit, the price of wind over the cost of wind must be the credit (1.5 cents/kWh) divided by the marginal tax federal plus state tax rate (40 to 60%), which gives 2.5 to 3.75 cents/kWh. Thus, if the cost of wind is 3.5 cents/kWh, the credit will be fully realized only if the price of wind is 6 to 7.25 cents/kWh (becuase the 3.5 cents/kWh cost is deductible). Wind producers are likely to optimize by raising their bid prices sufficiently to take advantage of the credit but not too high so that their projects are priced out of the selection. Ironically, then, the credit serves as a disincentive to reduce the price of wind (which is not the same as the cost of wind). A direct subsidy would be better because it would not provide incentive to maximize the difference between the price and cost of wind. Fourth, it is commonly known that it is much easier for large producers of any product to offer a lower price, thereby having a smaller profit margin (but a larger net profit summed over all sales) than it is for a small producer, who must offer a higher price at a lower sales volume. Finally, the market for new power plants is not a free market. In California, for example, separate bids are requested for renewable energy sources versus fossil power sources. One reason for the separate treatment is the misperception (as shown by Hirst and Hudson et al), that on a small scale, the intermittency of wind triggers an extra cost. Footnote: The cost of HVDC lines is calculated as follows: Cavallo (1995) estimates the cost of transmitted energy through a 2000-km HVDC transmission line as 2.75 cents/kWh. This translates to 0.00138 cents/kWh/km. Cavallo assumed a capital charge rate of 0.107, which translates to an interest rate of 9% over 20 years. However, transmission lines can last 40 to 60 years. Further, commercial interest rates today are lower than 9%. These combined factors alone would reduce Cavallo's estimate by a factor of 2. Cavallo (1995) also acknowledges that the transmission line cost used was conservative and "could be about one-half what we have assumed." Changes in assumptions about interest rate, transmission line lifetime, and direct costs would change Cavallo's transmission cost estimate to 0.000345 cents/kWh/km. This estimate could be reduced further by piggybacking new lines on existing transmission powers. Nevertheless, the 0.000345 cents/kWh/km cost reaches 1 percent our estimated direct cost of wind energy (3 to 4 cents/kWh) when the average transmission line is 88 to 116 km long. Even if the average transmission line is 500 km long, the cost is still less than 5% the direct cost of wind. In the worst case (2000 km line), the cost is around 20% the direct cost of wind (0.7 cents/kWh). References and Notes Archer, C. L., and M. Z. Jacobson, The regularity and spatial distribution of U.S. windpower, http://fluid.stanford.edu/~lozej/winds/winds.html, 2001. Bolinger, M., and R. Wiser, Summary of Power Authority Letters of Intent for Renewable Energy, Memorandum, Lawrence Berkeley National Laboratory, October 30, 2001. Cavallo, A. J., High-capacity factor wind energy systems, JSEE 117, 137-143, 1995. Danish Windturbine Manufacturers Association, "21 Frequently Asked Questions About Wind Energy," (updated 16 April 2001) http://www.windpower.dk/faqs.htm, 2001. Hirst, E., Interactions of wind farms with bulk-power operations and markets, http://www.EHirst.com/PDF/WindIntegration.pdf, 2001. Hudson, R., b. Kirby, and Y.-H. Wan, The impact of wind generation on system regulation requirements, AWEA Wind Power 2001 Conference, 2001. McFeat, T., The unnatural price of natural gas, CBC News Online, http://www.cbc.ca/news/indepth/background/gas_hikes. html, 2001. North American Electric Reliability Council, Generating Unit Statistical Brochure, 1995-1999, Princeton, NJ, October, 2000.
Response to Alfred Cavallo letter of November 21, 2001 28 November 2001
Mark Z. Jacobson and Gilbert M. Masters, Associate Professor (MZJ); Professor (GMM) Department of Civil and Environmental Engineering, Stanford University Respond to this E-Letter: Re: Response to Alfred Cavallo letter of November 21, 2001	We thank Cavallo for his contribution to this debate. However, we believe his letter misstates the comparison made in our paper. Second, his implication that wind power is not cheap relative to natural gas or coal is contradicted by a third-party analysis of 17 California wind proposals in 2001 that supports our conclusion that the direct cost of wind is 3 to 4 cents/kWh. Third, his implication that the Production Tax Credit reflects the attributes of wind is contradicted by the stated purpose of the credit. Finally, we believe his comments about transmission lines are misplaced and his cost estimates of transmission lines too high. First, Cavallo implicitly assumes that we compared the direct cost of new wind with the direct cost of all (old + new) coal. Instead, our paper compared (i) the direct cost of new coal with that of new wind, and (ii) the direct + health/environmental costs of new or old coal with those of new wind (since the health/environmental costs of old coal are higher than those of new coal, the total cost is likely to be similar in both cases). Nowhere did we discuss the direct cost of old coal with that of new wind, nor do we believe this matters from a public policy point of view. From such a point of view, the only relevant issue is the total (direct+health/environmental+subsidy) cost of wind versus that of coal because whether wind replaces coal is a political, not a marketplace, decision (regrettably, we did not discuss coal or wind subsidies in our original paper). It is not a marketplace decision because, even when wind and old coal prices are exactly the same, there is no incentive for coal producers merely to fold up. Coal producers will fold up only when government decides to (i) require old and new coal to eliminate emissions, (ii) require old and new coal to pay for the health and environmental damage caused by remaining emissions and mining, and (iii) reduce the subsidies given to coal in excess of those given to wind. At the same time, government itself can promote wind to ensure that new fossil energy does not take a larger foothold. Second, we believe our estimated direct cost of energy from wind (3 to 4 cents/kWh) is correct for the conditions assumed and is beginning to be reflected in wind proposals. For example, Bolinger and Wiser (1, p. 3) calculated the 25-year real costs of 17 wind farm proposals in California in 2001 as 3.2 to 3.7 cents/kWh, with a weighted average value of 3.6 cents/kWh. Their analysis also stated that the numbers were based on proposal information that presumably contained worst-case estimates for wind. Third, Cavallo says that "wind power receives premium payments to reflect its attributes...and indeed the U.S. Production Tax Credit program does just this." This is incorrect. The purpose of the tax credit is to level the playing field in terms of past and current subsidies that have favored the development of coal and natural gas technologies and that have kept the price of coal and gas low. Specifically, the House Ways and Means Committee stated (H. Rpt. 102-474, Part 6, p. 42), "The Credit is intended to enhance the development of technology to utilize the specified renewable energy sources and to promote competition between renewable energy sources and conventional energy sources." The credit does not address the attribute of renewable energy, namely, its health, environmental, and climate benefits over natural gas and coal. It addresses inequities in past and present subsidies. Even with the tax credit, tax and current direct government subsidies to coal and natural gas far exceed those of wind. Current tax and other subsidies for coal and natural gas are in the billions of dollars per year (http://www.foe.org/DLS for starters), whereas subsidies under the Production Tax Credit are on the order of $100 million per year. Fourth, we believe Cavallo's comments about transmission lines are out of context. He states that "transmission lines are not only costly, but quite difficult to site." We agree with these general statements but do not believe that translates into a high cost per kilowatt hour for wind energy. (i) Transmission access pathways already crisscross the United States, and many already pass through the Great Plains. If new, long transmission lines are needed for wind plants, most of these lines can piggyback on existing transmission towers, and smaller transmission lines on existing towers can be upgraded. Adding new lines to existing towers or replacing existing lines is less expensive than creating new towers. Only local connections to the nearest long-distance transmission pathway require siting of new transmission pathways. (ii) Whether the high cost of transmission lines per unit distance translates into a high cost per unit energy (kWh) depends on the length of the transmission line, so it is incorrect to label transmission costs as "high" without specifying the length of the line. Cavallo (2) estimated the cost of transmitted energy through a 2000-km HVDC transmission line as 2.75 cents/kWh. This translates to 0.00138 cents/kWh/km. Cavallo assumed a capital charge rate of 0.107, which translates to an interest rate of nearly 9% over 20 years. However, transmission lines can last 40 to 60 years. Further, commercial interest rates today are lower than 9%. These combined factors alone would reduce Cavallo's estimate by a factor of 2. Cavallo also acknowledged that the transmission line cost used was conservative and "could be about one-half what we have assumed" (2). Changes in assumptions about interest rate, transmission line lifetime, and direct costs would change Cavallo's transmission cost estimate to 0.000345 cents/kWh/km. This estimate could be reduced further by piggybacking new lines on existing transmission powers. Nevertheless, the 0.000345 cents/kWh/km cost is 1% of our estimated direct cost of wind energy (3 to 4 cents/kWh) when the average transmission line is 88 to 116 km long. Even if the average transmission line is 500 km long, the cost is still less than 5% of the direct cost of wind (and <1.5% the price a typical consumer pays for electricity, which is 11 cents/kWh). In the worst case (2000 km line), the cost is around 20% of the direct cost of wind (and <6% the price a consumer today pays for electricity). We have cited before that 840,000 MW of wind power lie within 20 km of existing transmission lines. Our proposal requires the generation of only 128,000 MW of power (225,000 1.5-MW turbines in the presence of 7 to 7.5 m/s winds, giving capacity factors of 0.35 to 0.4). Clearly, a reasonable portion of our required power can be obtained from turbines close to existing transmission lines. If such lines are already saturated with power, the cost of additional lines is not a cost of wind exclusively but a shared cost among all energy sources using the lines, because coal and natural gas generally do not own such lines, and therefore, do not have an exclusive right to them. In sum, we suggest that higher cost of long transmission lines is compensated for by lower cost of shorter transmission lines. If the average transmission line is less than 100 to 500 km long, the resulting cost of energy related to transmission lines is less than 1 to 5% of the price of wind energy. (c) In his letter, Cavallo says that transmission costs are high for wind but omits the fact that, when comparing coal and wind, it is necessary to compare the transmission costs of both, not merely to state that wind has a high transmission cost. There are thousands of coal plants in the United States and tens of thousands of miles of transmission lines needed to transmit coal energy. Not only do current transmission costs exist for coal, but when coal transmission lines do wear out (and most are fairly old now), they need replacing. References and Notes 1. M. Bolinger, R. Wiser, Summary of Power Authority Letters of Intent for Renewable Energy, Memorandum, Lawrence Berkeley National Laboratory, 30 October 2001. 2. A. J. Cavallo, High-capacity factor wind energy systems, JSEE 117, 137 (1995).
Re: Re: Wind vs. Coal 21 November 2001
Howard Gruenspecht, Resident Scholar Resources for the Future Respond to this E-Letter: Re: Re: Re: Wind vs. Coal	Letters from DeCarolis and Keith (Science, Nov. 2) and Golomb (dEbates, Nov 14) argue that the cost of contingency reserves to back up intermittent wind power omitted from the Jacobson and Masters proposal(Science, Aug. 24, p. 1438) to substitute wind for coal on a massive scale is significant. In response, Jacobson and Masters (Nov. 2, Nov 14) reference a recent paper by Hirst (1), which they say shows that these costs are trivial. However, the cited paper explicitly states that its ancillary service cost estimates for integrating wind do not include contingency reserves. Hirst's rationale for excluding contingency reserves, that wind farms (typically a fraction of 1 gigawatt (GW) capacity) do not contribute to the need for reserves required to meet the largest system contingency (typically in the range of 1 GW), clearly does not apply to the Jacobson and Masters proposal to install 321 to 354 GW of wind in the Dakotas. Where wind variation is the largest system contingency, as it would be under the Jacobson and Masters proposal, conventional reliability criteria would require reserves sufficient to meet load under the calmest 1-day-in-10-year wind conditions. With the rapid drop in generation as wind speed falls below its mean level (according to references cited by Jacobson and Masters, generation drops to zero at roughly 3 m/s), required contingency reserves equal to a significant fraction of the wind capacity envisioned under the Jacobson and Masters proposal would be needed. Significantly, Golomb notes that average windspeed in Bismarck is less than 3 m/s 40 percent of the time - a sobering consideration, given the likelihood of significant correlation in wind conditions across individual windfarm sites in North Dakota. DeCarolis and Keith also say that there are likely to be significant costs of moving power to load centers. In response, Jacobson and Masters outline a calculation that costs 10,000 km of transmission lines at $3.1 billion, less than 1% of wind turbine costs. However, 10,000 km provides only 30 km of transmission per gigawatt of wind capacity added under their proposal. This could perhaps meet local interconnection and grid enhancement needs, but not the need for long-distance transmission to load centers. Existing project proposals provide a firmer basis for estimating the latter cost. For example, Siemens and Black, and Veatch, experienced power system engineers and vendors, have recently analyzed a plan to add 8 GW of capacity in North Dakota and connect it to load centers in the Chicago and Los Angeles area by HVDC transmission.(2) Subtracting generating capacity costs from their $15 billion total project cost estimate suggests a transmission component cost of roughly $5 billion. A system capable of carrying 8 GW from North Dakota to only one load center would probably cost $3 billion, since two of the three sets of AC/DC converters would still be needed. Eight gigawatts is only 2 to 2.5 percent of the power that would be moved under the Jacobsen and Masters proposal, validating the DeCarolis argument that long-distance transmission costs are more than noise in the overall cost evaluation of that proposal. In addition to the foregoing comments related to previous exchanges, Jacobson and Masters' focus in theiroriginal Policy Forum on comparisons between the levelized costs of new wind and coal plants is something of a red herring because their actual proposal involves replacing generation from existing coal-fired plants, which costs 1 to 1.5 cents per kilowatt-hour, with new wind power. Furthermore, in cases where new capacity is needed now, the overwhelming choice in today's markets is for gas-fired units, which are both cheaper and cleaner than coal, but not even mentioned. Information they provide regarding black-lung deaths is also misleading in the context of this article. Most black-lung cases reflect past, not current, mining practices, and the number of black-lung deaths would not be appreciably impacted by the prospective reduction in coal use under their proposal. Competition between coal-fired and wind-powered generation will likely grow increasingly important over time. Each will have to overcome fuel-specific hurdles. For wind, these include the costs of contingency reserves and the need to overcome public objections to siting new transmission lines and turbines. For coal, these include the costs of increasingly stringent controls on conventional pollutants and the likely future requirement to capture and sequester carbon emissions. Notwithstanding the shortcomings of the Jacobson and Masters analysis, wind appears well-positioned to provide an important share of generation capacity additions over the coming decades. References and Notes (1) Hirst. Same reference #1 in Jacobson and Masters, Nov 2 (2) Engineering News Record, June 11, 2001 @ http://www.bv.com/bv/news/articles/grid_sols.htm
Response to Jacobson and Masters 21 November 2001
Joseph F. DeCarolis and David W. Keith Engineering and Public Policy, Carnegie Mellon University Respond to this E-Letter: Re: Response to Jacobson and Masters	We have enjoyed our dialog about the cost of large-scale wind energy. 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 the 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. 4. Wind versus coal. We realize that electricity production from coal results in significant environmental externalities, which must be addressed. Rather than speculating on the costs of coal externalities, we assume that coal with carbon capture may present a comparable solution to wind by minimizing power plant emissions. Such costs will likely raise the price of coal to the 5-7 ˘/kWh range (3, 4). This is the price wind will need to compete against. As for the cost of wind, we simply used your claim of 3-4 ˘/kWh for the amortized capital cost of wind turbines, and are therefore confused by your statement that “the authors use wind cost statistics from past experience.” We do not seriously dispute your estimate of the average cost of wind generation at a given site. We believe that wind may present an economically viable alternative to coal with carbon capture, but to assert that, “the cost of wind energy is now less that of coal” is not accurate. If it were, we would expect to see wind dominate virtually all new capacity installations (given the 1.5 ˘/kWh tax incentive), rather than simply having the fastest relative growth rate – not an overly impressive statistic for an energy technology that is cheaper than coal. We welcome any feedback and would like to continue this dialog. References and Notes 1. E. Hirst, Interactions of wind farms with bulk-power operations and markets, http://www.EHirst.com/PDF/WindIntegration.pdf, 2001. 2. M. Shaheen, Wind Energy Issue Brief 9a (October 1997) (http://www.nationalwind.org/pubs/wes/ibrief09a.htm). 3. E.A. Parson and D.W. Keith Science 1998 November 6; 282: 1053-1054. 4. H. Herzog, The Economics of CO2 Separation and Capture, Technology, 7, pp. 13-23 (2000).
Re: Re: Wind vs. Coal 21 November 2001
Alfred Cavallo, Physicist U.S. Department of Energy Respond to this E-Letter: Re: Re: Re: Wind vs. Coal	I would agree with the letter writers that it is an exaggeration to say that wind is competitive with coal. I have examined the issues of distance from demand centers and intermittency. My most recent paper was published in the November issue of JSEE in which I compared storage costs. Intermittent wind generated electricity can be transformed to a constantly available power supply economically by using compressed air energy storage (CAES) systems. Costs, including transmission and storage costs, are computed for a realistic system in (1). Transmission lines are not only costly, but quite difficult to site. Nobody wants one in their neighborhood. This can be overcome, but it takes great diplomacy and political will to accomplish. I do believe that wind energy could supply all of the electricity needed by the United States at a reasonable price, but it will cost more than market priced coal, which does not take into account any environmental damage from coal mining, acid rain, or global warming. People should be prepared to pay a premium for wind energy, or they should be prepared to penalize dirty power to reflect its real cost. There is indeed a boom in wind energy in Europe, but it is not caused by cheap (relative to natural gas or even imported coal) wind power. Wind power receives premium payments to reflect its attributes. This same approach should be used in the United States, and indeed the U.S. Production Tax Credit program does just this. The program is passed for only a few years at a time, then allowed to lapse, guaranteeing turmoil in the U.S. industry. Europe has made a policy decision to support clean power at a reasonable cost, even if it is more costly than fossil fuel generated electricity. The United States should do the same. References and Notes 1. High Capacity Factor Wind Energy Systems, JSEE 117, 137 (May 1995).
Re: Wind vs. Coal 14 November 2001
Mark Jacobson and Gilbert Masters Terman Engineering Center, Stanford University Respond to this E-Letter: Re: Re: Wind vs. Coal	We disagree with Golomb's premise and believe that our conclusions still stand. If wind energy replaces 59% of coal energy, then wind will supply about 30% of U.S. electricity, whereas 70% of electricity will still be supplied by other sources. As such, there is still plenty of backup electricity even if wind energy for a day hypothetically went to zero, which is not even a remote possibility, given the consistency of daily U.S.-averaged winds. The issue then is, what is the intermittency cost (the cost of regulation ancillary service) of wind. A study on this issue has been performed, and it shows that such costs are about 0.005 to 0.03 cents per kilowatt hour (kWh), which is less than 1% of the price of wind energy (1). The cost can be reduced further simply by using an hour-by-hour persistence forecast at the given location (1). In addition, the greater the number of turbines at a given wind farm and the greater the number of wind farms, the more intermittency of individual turbines cancel each other out. One can imagine a scenario where winds are slow one day at one wind farm. These slow wind speeds can be made up for by power generated at one of several other farms, where wind speeds are faster. It should also be noted that winds near the coast and offshore are regular and predictable and subject to less intermittency than winds away from the coast. Based on the above, we believe it is incorrect to state that wind cannot replace conventional power generators. Further, our paper discussed replacement of coal with wind, but we also believe that new wind should replace new natural gas, whose emissions enhance acid deposition, urban smog, human health and mortality, visibility degradation, and global warming, all of which have real costs. In sum, we believe our conclusions stand. References and Notes 1. E. Hirst, Interactions of wind farms with bulk-power operations and markets, http://www.EHirst.com/PDF/WindIntegration.pdf, 2001. 2. Danish Windturbine Manufacters Assoc. (2001). http://www.windpower.dk/faqs.htm
Wind vs. Coal 14 November 2001
Dan S. Golomb, Professor University of Massachusetts, Lowell Respond to this E-Letter: Re: Wind vs. Coal	Jacobson and Masters (24 Aug., p. 1438) make the case that wind-derived electricity could replace a significant fraction of coal derived electricity, thereby reducing coal carbon dioxide emissions by up to 59%. The cost of wind-derived electricity is comparable with that of coal-derived electricity. There is no doubt that a wind turbine does not emit any carbon dioxide (except that emitted by fossil fuels used to fabricate and construct the turbine), and does not emit any of the other harmful air pollutants associated with mining, transport, and combustion of coal. But in balancing the cost-benefit equation, we should be more judicious. Because wind is intermittent, back-up power generators must be available. Even in North Dakota, arguably the windiest state in the United States, winds do not blow all the time. For example, in Bismarck, North Dakota, winds are calm 5% of the time, and blow less than 3 m/s 40% of the time (1). (The efficiency of wind turbines declines precipitously when winds blow less than 3 m/s.) Thus, wind power cannot replace conventional power generators, but only displace the fuel that conventional generators would use when the wind generators are in operation. Many coal fired power plants supply the base load, because they cannot follow the fluctuating demand during peak hours. Peak power is mainly supplied by gas or diesel fired generators. An efficient combined cycle gas fired power plant might emit only half as much carbon dioxide per kilowatt hour as a coal fired generator, so the savings in carbon dioxide emissions by wind generators might be much less than Jacobson and Masters calculated. Typically, the fraction of fuel cost to total production cost of pulverized coal fired electricity generators is in the 24 to 30% range, and of natural gas combined cycle generators is in the 48 to 58% range (2). Thus, it is not correct to compare the total cost of coal versus wind generating costs, and total carbon dioxide emissions of coal versus wind, but only the fuel cost and carbon dioxide emissions that wind power displaces when the wind generators operate. This is not to say that wind-derived electricity is not worthwhile. The savings in carbon dioxide and other pollutant emissions are real, as well as the savings in fuel cost. But the cost-benefit equation must be properly balanced. References and Notes 1. Data supplied by J. Enz, State Climatologist, North Dakota State University. 2. Same as reference 1. in M. Z. Jacobson and G. M. Masters, Science 293, 1438 (2001).
Don't Dismiss the Midwest's Power Needs 2 November 2001
Dr. Josh Kurutz, Postdoctoral Fellow University of Chicago, Chemistry Department Respond to this E-Letter: Re: Don't Dismiss the Midwest's Power Needs	In their argument against wind power, DeCarolis and Keith dismiss this method of power generation in part because the best generation would come from the Great Plains, "far from demand centers concentrated on the coasts." Even if transmission costs prohibited transcontinental power distribution, Denver, Chicago, Minneapolis, Milwaukee, St. Louis, Indianapolis, Detroit, Des Moines, Topeka, Kansas City, Winnipeg, Calgary, Saskatoon, Edmonton, and, possibly, Dallas and Houston would benefit greatly from plains-derived wind power. Even if it cost the same or slightly more, wind power would allow more polluting resources to be made available to the coasts. Just because a good energy solution might not benefit America's coastal cities does not mean it should be ignored.