High Capacity Factor Wind Energy Systems

Wind generated electricity can be fundamentally transformed from an intermittent resource to a baseload power supply. For the case of long distance transmission of wind electricity, this change can be achieved at a negligible increase or even a decrease in the per unit cost of electricity. The economic and technical feasibility of this process can be illustrated by studying the example of a wind farm located in central Kansas and a 2000 km, 2000 megawatt transmission line to southern California. Such a system can have a capacity factor of 60 percent, with no economic penalty and without storage. With compressed air energy storage (CAES) (and with a negligible economic penalty), capacity factors of 70-95 percent can be achieved. This strategy has important implications for the development of wind energy throughout the world since good wind resources are usually located far from major demand centers. INTRODUCTION At present, most wind energy development has occurred in regions with excellent wind resources that are close to load centers, where transmission costs are low and transmission capacity is adequate. In the future, wind farms will be located far from load centers, and transmission cost and availability may constrain development. Also, as a consequence of the passage of the National Energy Policy Act of 1992, utilities are being required to separate transmission from generation and distribution charges. These factors indicate that it is important to consider wind farms and transmission lines as a system rather than as separate entities, and to minimize the cost of delivered electricity, including transmission cost. Minimizing the cost of delivered electricity will entail increasing the system capacity factor. This has the added benefit of weakening an important objection often raised by utilities to renewable energy resources such as photovoltaic and wind systems. These are intermittent: that is they have a low capacity factor and a high forced outage rate. Increasing the capacity factor effectively reduces the intermittent characteristic of the resource. In addition, for a given transmission capacity, wind developers will be able to sell much more energy at no increase in the delivered per unit cost, increasing revenues and profits. Both utilities and wind farm developers will benefit from this approach. Since a utility is accustomed to control, or dispatch, its sources of energy to meet demand at a given time, coping with intermittent generating technologies presents conceptual difficulties and operational challenges. These challenges certainly exist: the theoretical result that at low (<10 percent) system penetration an intermittent supply can be regarded as a negative load and effectively integrated, while completely correct (Haslett and Diesendorf, 1981; Grubb, 1991), does not give any indication of these problems (Friis and Mogens, 1993; Harrison, 1993). In order to understand how it is possible to construct, with a minimum economic penalty, a high capacity factor system or a wind energy baseload system from an intermittent resource, we shall first examine some of the characteristics of wind that influence the wind turbine capacity factor, and then some aspects of transmission line technology. Next, the concept will be illustrated by examining the economic and technical characteristics of a wind farm in western Kansas coupled to a 2000 km transmission line. Finally, the economic and technical attributes of a hybrid system consisting of a wind farm with compressed air energy storage (CAES) using the same transmission line will be examined. This type of system could, for example, replace the nuclear power plants at Diablo Canyon, CA, (2x1100 MWe, average capacity factor 76 percent) around the year 2010, at which time they would have been in operation for 25 years. WIND AND WIND TURBINE CHARACTERISTICS The amount of power generated by a wind turbine is a result of both the design characteristics of the turbine and the properties of the wind resource (the wind speed probability density as a function of wind velocity, f(v)). It has been found that the wind frequency can best be described by a Weibull probability distribution; f(v) can be written as (Johnson, 1985): Here c and k are the scale and shape factors, respectively. The parameter c has dimensions of velocity and is about 1.1 times the average wind velocity, while k largely determines the shape of f(v). A k value close to 1 indicates a highly variable wind regime, while a k greater than 3 indicates more regular, steadier winds. Since detailed information on the wind frequency is often lacking, a k factor of 2 is often assumed in evaluating a wind resource. As will be shown, this can lead to a significant error in the estimate of the capacity factor and the cost of electricity. The power output (Pout) of a wind turbine as a function of wind velocity is written as: The average power output () of a wind turbine can be written as: This is just the power output of the wind turbine at a given velocity times the frequency at which that velocity occurs, summed over all possible velocities. The integral in Eq. 3 is the ratio of the average turbine output to the maximum turbine output and is defined as the wind turbine capacity factor (WTCF). In Figure 1 the capacity factor of a Vestas V27 wind turbine is plotted as a function of k for a constant wind power density of 440 W/m2, which is typical of that found over large areas of the Great Plains. The published characteristics of the Vestas V27-225 wind turbine (Vestas, 1993), an efficient 225 kW pitch regulated machine with high and low speed generators, and Eq. 1, were used in Eq. 3 to calculate the capacity factor. This shows clearly the importance of a detailed understanding of the wind resource. Typical values of k obtained from data taken in the Department of Energy (DOE) Candidate Wind Turbine Test Site program (Cavallo,1994) are 2.4 to 3 at 50 m elevation over the Great Plains. If k is equal to 3, the capacity factor is 20 percent greater than at the usually assumed value of k=2, implying a correspondingly larger output per machine, and correspondingly lower costs per unit of output. The wind resources of the U.S. have been evaluated using data from a wide variety of sources (Elliott, 1987). Using the results of this survey, the wind electric potential of the U.S. has been estimated (Elliott, 1991) at 1200 GWavg; more than 90 percent of this potential is located in the Great Plains, far from electricity demand centers. If these resources are to be utilized on a significant scale, long distance transmission lines will certainly be an integral part of the development. We have chosen western Kansas for our wind farm location because, based on DOE data from this area, the Weibull K factor at 50 m is about 3 and the yearly average as well as the summer average wind power density is about 440 W/m2; the former indicates an economically viable resource, while the latter indicates that system output will be high in the summer, when utility demand is highest. (Other Great Plains regions experience at least a 20 percent decrease in wind power density during the summer season.) Therefore, the wind regime assumed for these calculations is one with Pw=440 W/m2 and k=3, and is constant over the year; this represents a realistic best case scenario. It is also assumed that the wind turbine hub height is 50 m. TRANSMISSION LINE TECHNOLOGY For this case study, we have chosen high voltage direct current (HVDC) technology for the transmission line. This has been shown to be the lowest cost option for point to point power transfers over distances greater than about 800 km (Wu, 1990). There would initially be a significant difference between transmission line capacity and the output of the wind farm; in order to illustrate the general principles involved, overnight construction of all system components is assumed. The cost estimates used here are based on those incurred in the construction of a 2000 MW, 450 kV, 2222 A, 1500 km HVDC transmission line between the La Grande hydroelectric complex at James Bay, Québec, and Boston, MA (Reason, 1990) as well as on information given by Long (1987). The agreement to build the HQ (Hydro Québec) line was signed in 1983, and construction proceeded in two phases; the line was completed at the end of 1990. Long development times are typical for such projects: Watkins (1991) estimated an 8-12 year development time for the 3000 MW, 1100 km HVAC Pacific Northwest line. Although construction time was projected to be only 2.5 years, preparation of applications and the environmental impact statement, and hearings before various state agencies and commissions lengthened the total project time considerably. This indicates that such projects will require strong utility and governmental leadership. The HQ transmission line was built over an existing right of way in the US, while in Québec the right of way had to be acquired and extensive road construction was necessary. According to project engineers, the transmission line cost about $0.62 million per km ($1 million per mile) both in the U.S. and in Québec. The cost of the converter stations (345 kV AC to 450 kV DC), filters and circuit breakers in the US was $320 million. Converter losses are 0.6% per station; line resistance is 12 Ω per pole (the line has two conductors or poles, operating at +450 kV and -450 kV) so that total ohmic and conversion loss at full power is 140 MW, or 7 percent of the transmitted power. Operation and maintenance costs for the line are negligible. The cost of the HQ transmission line ($682/kV-km) is substantially greater than the HVDC line only costs ($198-229/kV-km) cited by Long (1987). However, the HVAC transmission line cost of $560/kV-km quoted by Watkins (1991) agrees well with that given by Long ($576-607/kV-km for HVAC single circuit transmission lines). It may be that construction costs in the Quebec wilderness and New England are substantially greater than what would be encountered in the Great Plains, and therefore the HQ figures should be considered quite pessimistic. HVDC converter costs of about $110/kW are also quoted by Long (1987), and are significantly less than the $160/kW for the HQ system. The latter, however, includes substantial AC and DC filter and shunt capacitor bank costs, which could account for the