Wind Power in Ontario Its Contribution to the Electricity Grid
Bulletin of Science, Technology & Society http://bst.sagepub.com
Wind Power in Ontario: Its Contribution to the Electricity Grid Ian H. Rowlands and Carey Jernigan Bulletin of Science Technology Society 2008; 28; 436 originally published online Jun 6, 2008; DOI: 10.1177/0270467608315942 The online version of this article can be found at: http://bst.sagepub.com/cgi/content/abstract/28/6/436
Published by: http://www.sagepublications.com
On behalf of: National Association for Science, Technology & Society
Additional services and information for Bulletin of Science, Technology & Society can be found at: Email Alerts: http://bst.sagepub.com/cgi/alerts Subscriptions: http://bst.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://bst.sagepub.com/cgi/content/refs/28/6/436
Downloaded from http://bst.sagepub.com by on June 29, 2010
Bulletin of Science, Technology & Society Volume 28 Number 6 December 2008 436-453 © 2008 Sage Publications 10.1177/0270467608315942 http://bsts.sagepub.com hosted at http://online.sagepub.com
Wind Power in Ontario Its Contribution to the Electricity Grid Ian H. Rowlands University of Waterloo, Ontario, Canada
Carey Jernigan NSCAD University, Halifax, Nova Scotia, Canada The purpose of this article is to investigate wind turbine production, the variability of that production, and the relationship between output and system-wide demand. A review of the literature reveals that a variety of measures (and methods) to explore the variability of wind power production exist. Attention then turns to the province of Ontario (Canada), and the performances of four wind farms are examined for 2006 and 2007. Key conclusions include that the wind farms’ capacity factors vary from 27.6% to 35.6%, with higher values in winter as compared to summer; wind power performs better than the seasonal average during peak periods; wind is a better “partner” for the Ontario electricity system in the winter as opposed to the summer; and the increased geographic distribution of wind farms decreases their collective variability. Keywords:
I
Canada; electricity; energy policy; grid performance; markets; Ontario; wind energy
nterest in renewable energy sources has been growing in recent years. This has been motivated by a variety of environmental concerns, not least of all the air quality impacts (smog and global climate change) generated by conventional, fossil fuel–burning power plants (Goldemberg & Johansson, 2004, pp. 40-42). “Energy security” is also a concern on the rise (Yergin, 2006), catalyzed by rising oil prices (Organization of the Petroleum Exporting Countries, 2007), fear of terrorism (Asmus, 2001), the war in Iraq (Yankelovich, 2006), and diminishing reserves of conventional fuels (Deffeyes, 2005). In this context, clean, local, and renewable energy sources are becoming more appealing and thus attracting more attention. The expansion of wind power—one of the world’s leading renewable energy sources—has proceeded accordingly: In 2006, installed wind energy capacity increased 32% worldwide (Global Wind Energy Council, 2006). But with this increased attention has also come increased scrutiny. Although wind power’s attributes are lauded by its supporters, critics maintain that wind’s variability—that is, the fact that it is available not “on demand” but only when the wind is blowing
(what these critics often call “intermittency”)—means that its contribution is ultimately limited: To maintain electricity system security, they continue, its contribution to the grid must be capped. Indeed, in a study on wind power’s role in Ontario (Canada), Adams (2006) worried that “a clean and promising generating technology is being burdened with unrealistic forecasts of reliable production” (p. 1). In this article, we assess the contribution of wind farms to the electricity grid in the province of Ontario (Canada) in 2006 and 2007. Specifically, we explore wind turbine production, the variability of that production, and the relationship between output and system-wide demand; we then briefly discuss strategies for the future.
Electricity in Ontario The article investigates the province of Ontario, Canada’s most populous province and the country’s second largest generator of electricity (of 10 Canadian provinces). In 2005, total Ontario electricity demand reached 157 TWh,1 with nuclear generating stations
Authors’ Note: Support for part of this research was provided by a grant from the Social Sciences and Humanities Research Council of Canada (through its support of a project titled Business and Green Power in Electricity Transformation: Markets and Policies, in the Research Developments Initiative program; see http://www.fes.uwaterloo.ca/research/greenpower). The authors are grateful for this support. The authors would also like to thank Glen Estill and Paul Gipe for their comments on earlier drafts of the article. The authors, however, remain fully responsible for all of the contents of the article.
436
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 437
making the largest contribution to the province’s electricity supply (54.1%), whereas hydroelectric (22.3%), coal (16.0%), and natural gas (6.4%) stations also were major contributors. Other supply sources, including all renewables, accounted for the remaining 1.2% (Government of Ontario, 2007). Peak system demand, meanwhile, reached 26,160 MW on Wednesday, July 13, 2005, at 5 p.m. (eastern daylight time [EDT]). That was surpassed the following year, when Ontario’s demand reached 27,005 MW on Tuesday, August 1, 2006, at 5 p.m. (EDT). But although Ontario is a “summer peaking” system, total electricity demand is higher in the winter than the summer.2 Turning more specifically to wind power, there are presently four wind farms in Ontario (i.e., locations at which there are at least 10 wind turbines in close proximity to each other), with an installed capacity of, collectively, 396.1 MW. A number of individual grid-connected wind turbines also exist in the province—together, these 15 turbines (ranging from 0.6 MW to 1.8 MW) account for 20.1 MW of installed capacity (Canadian Wind Energy Association [CanWEA], 2007).3 A number of additional wind projects are currently in development—both wind farms and individual turbines. (Both larger- and smaller-scale wind projects have been encouraged by the province of Ontario’s two-track strategy with respect to renewable electricity: Larger projects have been catalyzed by a tendering process, whereas smaller ones have been encouraged by the introduction of “standard offer contracts” [Rowlands, 2007].) Ontario’s electricity system operator—that is, the “Independent Electricity System Operator” (IESO)—anticipates that by the end of 2008, five additional wind farms will be fully operational: the Ripley Wind Power Project (a 76 MW project located on the shores of Lake Huron), the Kruger Energy Port Alma Wind Power Project (a 101 MW project located on the shores of Lake Erie), the Malanchton II wind project (a 132 MW project located in central Ontario, near the existing Amaranth Wind Farm), the Wolfe Island Wind Project (a 198 MW project located on Wolfe Island in Lake Ontario, Wolfe Island, near Kingston, Ontario), and the Enbridge Ontario Wind Power Project (a 200 MW project located near Kincardine, Ontario) (IESO, 2007a, p. 13). Moreover, in its “Integrated Power System Plan,” the Ontario Power Authority (OPA) anticipates that the installed capacity of wind power will steadily grow during the coming 20 years, reaching 4,685 MW by 2025 (OPA, 2007b, p. 10). Wind proponents say that this level could be safely much higher, for they maintain that desirable off-shore resources were not given sufficient attention in the OPA’s analyses (e.g., Hornung, 2006).
In any case, the debate about wind’s present and future roles in the province of Ontario continues. On the one hand, many are calling for a more prominent position for wind, arguing that its potential contribution is often misunderstood. For example, the national wind energy industry association advances the following position: Everyone knows that the wind is variable. Sometimes it blows, other times it doesn’t. So how can wind power be a reliable source of energy? . . . The fact is, the wind will never stop blowing everywhere at once. . . . With Canada’s large and varied wind resource, there’s no doubt that the wind can power us well into the future. (CanWEA, 2006)
Others, by contrast, challenge any proposed largescale role that wind might play. The president and chief executive officer of one of Ontario’s largest nuclear power plants, for instance, maintains that renewables such as wind and solar power are certainly environmentally friendly. Yet the truth is they cannot be relied upon to keep our factories humming and homes warm because of their intermittent availability and limited scale. Renewables have their place in our energy mix, but we can’t be naive enough to believe they are the large-scale answer. (Hawthorne, 2004, p. 6)
It is against this background that we investigate, in this article, the contribution of wind to the Ontario electricity system.
Wind Power’s Performance in an Electricity Grid The degree of variability of the output of a wind turbine has significant consequences for the potential contribution that it can make to the wider electricity system. If it varies significantly, then the other generators will have to respond accordingly. In addition, the extent to which changes in wind energy output positively correlate with fluctuations in system-wide electricity demand also has important implications for the integration of wind power into the electricity grid. If the two vary in tandem, then increased wind power would reduce both “load-following” requirements (how often generators must alter their power output to match demand, therefore operating less efficiently than with continuous output) and “operating reserve” requirements (stand-by capacity that is kept online in case the power system suffers a supply interruption). If, on the other hand, wind
Downloaded from http://bst.sagepub.com by on June 29, 2010
438 Bulletin of Science, Technology & Society
output randomly fluctuates or negatively correlates with system-wide operating demand, then load-following requirements and the need for larger operating reserves would increase. Thus, the pattern of wind power production (both in an isolated sense and compared to the demand patterns in the broader electricity system) will determine the “value” of its contribution to the electricity grid (e.g., DeCarolis & Keith, 2005; Gipe, 2006b; Holttinen et al., 2007). As such, a better understanding of wind power’s contribution to a power grid as societies consider increased use of renewable resources is critical. Hence, can we identify patterns in wind power production and its relationship to demand? Are they consistent over time? Are they beneficial? Do they differ from one location to the next? How quickly do changes occur? In this section, we review the focus and findings of previous studies on wind power performance, with specific emphasis on patterns of production.
Wind Power Over Time When investigating wind power’s patterns of production, time scale is important. DeCarolis and Keith (2006) argued that three timescales “concern system operators on a day-to-day basis: minute-to-minute, intrahour, and hour- to day-ahead scheduling” (p. 397). For his part, Milborrow (2007) suggested that power changes within an hour should be the focus, for “these tend to have the strongest influence on the ‘costs of variability’” (p. 38). Although most studies have relied on hourly data for ease of access (e.g., Holttinen, 2003; Parsons, Wan, & Kirby, 2001), others have used finer-resolution records to better reflect specific aspects of grid operation. Still others have examined a range: A study carried out by General Electric International Inc. (GE, 2006) in Ontario, for example, examined daily, monthly, and yearly trends as well as 3-hour sustained ramping, 1hour generation scheduling, 10-minute operating reserve requirement, 5-minute load-following requirement, and 1-minute regulation requirement time frames. The study found that variability tends to increase over time; Parsons et al. (2001) found the same pattern in a wind farm in Minnesota. Although the rate of change in wind output has consequences at all time scales, the nature of the broader electricity grid (and the associated peak demand periods generated by its users) means that particular time periods are particularly significant. For many, the day-night distinction is important to make. The Danish Wind Industry Association (DWIA), for example, reported that more wind is typically available during the day than at night, so that “wind electricity generally fits well into
the electricity consumption pattern” (DWIA, 2003d; see similar examinations in, e.g., Li & Li, 2005). Moving from a consideration of changes within a day to changes within a year, many have found wind power performance to vary according to the season. Several wind industry associations (e.g., CanWEA, 2006; DWIA, 2003c), for example, have highlighted the fact that wind output is typically high in the winter, when electrical heating demand rises in high-latitude locations (for additional reflections on seasonal variations, see Belanger & Gagnon, 2002; Milborrow, 2007; Sinden, 2007a). Rather than focusing on times of the day or seasons of the year, still others have highlighted the importance of focusing on high-demand periods, determined by different means—for example, “the top 20 per cent of demand hours” (Sinden, 2007b, p. 65). Holttinen (2003) suggested that wind output varies with temperature more importantly than with season. Moreover, this relationship may not be linear. In Germany, E.ON Netz (2005) found that “cold wintry periods” and hot summer days, when heating and air conditioning costs are highest, respectively, are often symptomatic of high-pressure weather systems when winds are typically low. In other words, wind output may be highest in winter on average but low on cold, still days when heating demand peaks; similarly, wind output in summer may not be high when cooling demand peaks. In the United Kingdom, meanwhile, it has been suggested that large anticyclones with little wind are associated with low temperatures in the winter (when, as a result, “heating and lighting loads can also be at maximum”) and with clear skies and high temperatures in the summer (when “increased demand for cooling and air conditioning result”; Laughton, 2007, pp. 8, 9, 14; also see Adams, 2007). Finally, a consideration of wind power over time can move from examination of performance within a year to investigation of output across years. Estimating output from wind speeds taken over 30 years in the United Kingdom, Sinden (2007a) found that average annual production ranged from 24.1% to 35.7% of installed capacity. Meanwhile, the DWIA (2003c) predicted a 9% to 10% variation in wind output, year to year.
Wind Power Over Space In addition to time, geography plays an important role in wind power’s contribution. First, the conditions of the site at which the turbines are located—the roughness of the terrain and the topography more generally, for example—affect wind power output. Offshore wind turbines therefore produce, on average, more electricity
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 439
than do nearby onshore sites, for grasses, trees, hills, and/or buildings can increase turbulence in the case of the latter (DWIA, 2003b). Second, weather systems are of limited geographic extent and will often pass through different sites within a region at different times; as a consequence, many argue that increasing the geographic dispersion of wind generators can increase the overall reliability of wind power (as a whole). E.ON Netz (2005), Holttinen (2003), Milborrow (2007, p. 36), and NBSO (2005) are among those who have argued that greater geographic dispersion of wind farms improves overall reliability and can reduce forecasting error so that changes in wind output can be better prepared for. (And, indeed, many have suggested that the further apart, the better—that is, the less correlated [e.g., Sinden, 2007a, 117-118; also see Gross et al., 2006].) In a study of Minnesota and the eastern parts of North and South Dakota, it was found that the geographic distribution of wind production “provides for substantial ‘smoothing’ of wind generation variations . . . [and] the number of hours at either very high or very low production are reduced, allowing the aggregate wind generation to behave as a more stable supply of electric energy” (EnerNex Corporation, 2006, p. xxi). Similarly, Holttinen et al. (2007) argued: Because of spatial variations of wind from turbine to turbine in a wind power plant—and to a greater degree from wind power plant to wind power plant— a sudden loss of all wind power on a system simultaneously due to a loss of wind is not a credible event. (p. 12)
More recently, The Economist (“Where the Wind Blows,” 2007) observed, “The question of where the wind is blowing [in Europe] would no longer matter because it is almost always blowing somewhere” (p. 81; also see Drake & Hubacek, 2007).
Method and Data The focus and findings of research on wind power performance not only vary among studies but also reveal significant differences with respect to the methods deployed. Although some researchers analyze production data taken from the operations of wind turbines, others use simulated output. These latter values are calculated based on a location’s wind speed (ideally at turbine hub height and often described using the Weibull distribution or other models [Li & Li, 2005]), perhaps accompanied by temperature (which affects air density
and thus the power of wind), a measure of surface roughness (DWIA, 2003b), and an estimate of availability losses resulting from blade soiling, icing, or other maintenance concerns (GE, 2006). A projected power output curve for a given turbine is then developed and used to simulate wind power performance across time. It is generally accepted that because wind speed measurements are often taken at one location on a site (rather than for each turbine) and roughness and availability losses are not entirely predictable, actual output should be used wherever possible (Holttinen, 2003). It remains, however, that a number of wind performance investigations are taken in advance of wind turbine construction (in order to assist with planning), and therefore simulated output continues to be widely used. Once output data are collected or simulated, various methods are used to study wind power performance. Milligan and Porter (2006) suggested that effective load carrying capacity (ELCC) techniques are most useful for electricity grid planning because they consider wind power’s role within a larger system (also see the discussion in Holttinen et al., 2007, p. 77). Calculations can accommodate hourly load requirements, hourly output data for renewable generators (e.g., wind) and hourly output data for conventional generators (e.g., rated capacity, forced outage rates, and maintenance schedules for coal, natural gas, nuclear, or other power plants). The technique then compares the loss of load expectation (the probability that power generation will fail to meet demand) for the system including wind electricity versus the same system without wind electricity. ELCC can also be used to determine what combination of renewable energy generators are required to achieve the same risk level as conventional technologies, which conventional facilities can be retired once renewable energy generators come online, or how much demand can be accommodated by different supply scenarios (Milligan & Porter, 2006). If, instead, we look only at the relationship between wind output and demand, we ignore interaction with other elements in the system. For example, controlled hydropower dispatched during peak demand hours, power purchased from outside the local area at these times, or routine maintenance that shuts down large generating facilities during low peak seasons (e.g., spring and fall) can shift “high-risk” periods considerably (Milligan & Porter, 2006). However, there are significant data requirements associated with the use of ELCC techniques. Where data describing all system components are not available, or where research specifically focuses on the relationship between wind electricity generation and
Downloaded from http://bst.sagepub.com by on June 29, 2010
440 Bulletin of Science, Technology & Society
system-wide demand, simpler “time-period based methods” are often used instead. Average and/or peak levels of wind power output are plotted against time, and patterns, rates of change, or relationships to system-wide demand are discussed. This approach has the benefit of simplicity and ease of explanation, but data must be appropriately interpreted (Milligan & Porter, 2006). A disadvantage, however, is that these methods are not capable of assessing and finding times that the system may be at risk even though loads are not especially high. If a significant fraction of the generating capacity is on maintenance during the shoulder seasons, this can cause a potentially large increase in LOLP and can result in potentially much higher risk than peak periods. (Holttinen et al., 2007, p. 83)
It is striking that statistical pattern analysis and confidence intervals are largely absent in the literature. This is perhaps indicative of small sample sizes, where data are often only available for a short period of time (often, given wind’s recent growth, less than 1 year), or perhaps reflects the fact that conditions rapidly change with time (as new wind projects are installed or system-wide demand patterns change) and across space so that results from one study rarely reflect those of others. Also, rather than tracking total production (kWh), studies often use ratios (Holttinen, 2003) to facilitate comparison between local conditions and future or distant sites. Perhaps the most commonly used metric in the investigation of wind power performance is “capacity factor” (CF). CF is defined as the ratio of actual electricity production to total installed capacity—the theoretical maximum output or “nameplate capacity” of the generator in question. Variations on CF include “capacity credit” (Milligan & Porter, 2006) and “capacity value” (GE, 2006), which focus on wind output during periods of high demand (e.g., the top 10% of peak load hours or between 3 p.m. and 7 p.m. in the summer). These metrics (often defined differently in different studies) are intended to reflect the “relevance” of variable power sources to overall system planning—the amount of installed conventional power that can be replaced by renewable supply (Voorspools & D’haeseleer, 2006), for example, or the “firm” generating capacity that can be counted on during periods of peak demand (GE, 2006). Although CF is commonly used in the literature, its appropriateness has been called into question by some (e.g., Gipe, 2006a; Wiser & Bolinger, 2007). DWIA
(2003a) referred to a Capacity Factor Paradox: Unlike with other generation technologies, a high CF for wind does not necessarily reflect greater overall production. Wind generators vary in their “cut-in” and “cut-out” wind speeds (the wind speed at which they start or stop producing electricity) and also in the speeds at which they operate most efficiently. Depending on wind conditions at a given site, a turbine with a relatively large generator (relative to rotor diameter) may operate at a low average CF most of the time but be able to take advantage of high winds when they occur. A relatively small generator might operate close to its design limit most of the time (and so have a high average CF) but will “cut out” during high wind events. As wind power production varies with the cube of wind speed, overall production may drop significantly. Wiser and Bolinger (2007) and Gipe (2006a) suggested that, instead of CF, electricity generation per m2 of swept rotor area (kWh/m2) is a better measure of wind power. Gipe referred to this metric as “specific yield” and argued that the area “swept” by a turbine’s blades determines the amount of the available wind resource that is actually intercepted by a turbine and so can potentially be converted to electricity. There is no generally accepted means by which a wind turbine’s performance—particularly with respect to its contribution to the broader grid—is measured. Instead, a variety of approaches, each emphasizing particular elements, continue to be deployed (also see Gross et al., 2006). Keeping many of these methods on hand, we now turn to the particular case of Ontario in order to investigate wind power in that province.
Data: Wind Power Production in Ontario In this article, we investigate the wind power production from the four existing wind farms in Ontario. (Please see Figure 1 for a map.) They are as follows: 1.
2.
Amaranth: Located near Shelburne, in Melanchton Township, approximately 100 km northwest of Toronto, forty-five 1.5 MW GE turbines (each with three 37 m blades) went into operation on March 4, 2006. Owned by Canadian Hydro Developers, the wind farm has a total installed capacity of 67.5 MW (Canadian Hydro Developers, 2006, p. 2; Canadian Hydro Developers, 2007). Kingsbridge: Located on the shore of Lake Huron (near Goderich), 22 turbines—each a 1.8 MW Vestas turbine—began generating power in March 2006. The blades on each have a length of 39 m. Owned by EPCOR Utilities Inc., the total installed capacity is 39.6 MW (EPCOR Ontario, 2007).
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 441
Figure 1 Location of Four Wind Farms in Ontario
Table 1 Total Energy Production, Ontario Wind Farms, March 2006 to November 2007
ONTARIO Prince Amaranth
Total Production (MWh)
Average Daily Production (MWh)
Period Beginning (Ending November 30, 2007)
284,598 182,677 342,230 544,552
444.7 285.4 579.1 1,148.8
March 1, 2006 March 1, 2006 April 19, 2006a August 14, 2006a
Amaranth Kingsbridge Port Burwell Prince
a. For Port Burwell and Prince farms, “starting dates” are those days on which a reading of at least 2 MWh for 1 hour’s production is noted on the Independent Electricity System Operator Web site.
Kingsbridge Port Burwell
Source: Original map data provided by The Atlas of Canada http://atlas.gc.ca/ © 2008. Data reproduced with permission of Natural Resources Canada.
Figure 2 Average Daily Energy Production (MWh), Ontario Wind Farms, April 2006 to November 2007
by Brookfield Power. There are 126 GE turbines (again, with a capacity of 1.5 MW each, and again with diameter of 77 m), and they were deployed in two stages, with full operation achieved in November 2006. Total installed capacity is thus 189 MW (Brookfield Power, 2007; Hatch Energy, 2006).4
Results: Wind Power Performance in Ontario Our investigation of wind power’s contribution in Ontario proceeds in two parts. First, we consider the operation of the wind farms individually—that is, without any reference to the broader (Ontario) system in which they are located. And second, we investigate the performance of the wind farms set against each other as well as the system-wide (provincial) demand.
Individual Wind Farms’ Performance
3.
4.
Port Burwell: Also known as the Erie Shores Wind Farm, these 66 GE turbines (1.5 MW each) are located on the northern shoreline of Lake Erie. The project “was commissioned in May 2006 and began commercial operation in June 2006.” With a total capacity of 99 MW (and with each turbine having a 77 m blade diameter), the wind farm is owned by AIM PowerGen (Stapletonprice, 2007). Prince: Located in the Sault Ste. Marie District (northern Ontario), this project has been developed
Figure 2 provides information regarding the energy production of Ontario’s wind farms, between April 2006 and November 2007.5 From this, it is clear that production is higher during the winter months as compared to the summer months. In terms of total production for each wind farm, meanwhile, Table 1 provides some details; note that, in this table, we present data for a longer time period than in Figure 2. CF, as defined in the third section of this article, is an additional way of investigating the performance of wind farms. Given the production figures presented in Figure 2, along with the stated nameplate capacities of the wind farms, the CFs can be calculated. These results are presented in Figure 3. For all four wind farms, the highest CFs occur in either February or March, the lowest in either July, August, or September. For the entire periods
Downloaded from http://bst.sagepub.com by on June 29, 2010
442 Bulletin of Science, Technology & Society
Table 2 Annual Capacity Value Factors for Ontario Wind Farms, Selected 12-Month Periods, April 2006 to November 2007 Amaranth (%)
Kingsbridge (%)
Port Burwell (%)
Prince (%)
28.9 29.6 29.4 29.7 29.4 29.5 29.7 29.3 30.2 29.5
32.0 33.2 33.5 33.7 33.6 33.7 34.1 34.2 35.6 33.7
—a — — 29.0 28.9 29.0 27.8 27.6 28.9 28.5
— — — — — — — — 28.7 28.7
April 2006 to March 2007 May 2006 to April 2007 June 2006 to May 2007 July 2006 to June 2007 August 2006 to July 2007 September 2006 to August 2007 October 2006 to September 2007 November 2006 to October 2007 December 2006 to November 2007 Average a. Not applicable.
Figure 3 Monthly Capacity Factor Values for Ontario Wind Farms, April 2006 to November 2007
Table 3 Production and Specific Yield, Ontario Wind Farms, December 2006 to November 2007 Port Amaranth Kingsbridge Burwell Prince
60 50
Production (MWh) 178,679 Swept area per 4,657 turbine (m2) Total swept area (m2) 209,565 Specific yield 85.3 (kWh/m2)
40 30 20
123,508 5,026
250,854 474,905 4,657 4,657
110,572 111.7
307,362 586,782 81.6 80.9
10 0
Amaranth
Kingsbridge
Port Burwell
Prince
under investigation, the CFs are, highest to lowest, 30.5% (Kingsbridge), 28.7% (Prince), 27.6% (Amaranth), and 26.8% (Port Burwell). Recognize, however, that these annual CFs include, for three of the four wind farms, a higher proportion of summer months (times at which production figures are lower). Therefore, to consider annual CFs for Amaranth, Kingsbridge, and Port Burwell, one can consider any of a variety of 12-month periods. From the results in Table 2, it is clear that the CF, in every case, rises, as compared with the CF for the entire study period. Specific yields can also be calculated. Given the lengths of the individual turbines’ rotor blades noted
above,6 Table 3 provides calculations for the wind farms’ specific yields. Wanting not to have a particular season influence the calculation, we consider only the contiguous 1-year period of December 2006 to November 2007 in each case. (Indeed, from this point forward, we focus on this 1-year period for this very reason.) Although Kingsbridge continues to be the top performer on the specific yield measure (as it was in terms of CF), Port Burwell and Prince swap places in terms of their respective rankings. Basic data can also be presented on an hourly basis. In Figure 4, data for the four wind farms are presented, across the “24-hour clock” (with, it must be noted, all times presented in eastern standard time [EST]). The morning period is generally a lower time for all wind farms, with three of the four experiencing their lowest hourly value (indicated by a dashed circle) at either 9 a.m. or 10 a.m. The respective peaks (noted by the solid circle) are more dispersed across the day (3 a.m., 2 p.m., 3 p.m., and 9 p.m.). In addition, note that the difference between the lowest and highest values was modest,
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 443
Figure 4 Average hourly energy production (MWh), Ontario wind farms, December 2006 to November 2007
Figure 5 Percentage of Time During Which Wind Farms’ Capacity Values Are Above, Within, or Below Particular Threshold Values, Ontario, December 2006 to November 2007
70
100% 60
90%
50
80%
40
70%
25.4
24.1
23.9
55.9
58.2
20.0
17.8
Port Burwell
Prince
32.8
60% 30
50% 20
40%
10
30%
56.4 53.5
20%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hour (EST) Amaranth
Kingsbridge
10%
18.3
13.8
0% Port Burwell
Prince
Amaranth >50% capacity
Kingsbridge
2%
Wind Power in Ontario: Its Contribution to the Electricity Grid Ian H. Rowlands and Carey Jernigan Bulletin of Science Technology Society 2008; 28; 436 originally published online Jun 6, 2008; DOI: 10.1177/0270467608315942 The online version of this article can be found at: http://bst.sagepub.com/cgi/content/abstract/28/6/436
Published by: http://www.sagepublications.com
On behalf of: National Association for Science, Technology & Society
Additional services and information for Bulletin of Science, Technology & Society can be found at: Email Alerts: http://bst.sagepub.com/cgi/alerts Subscriptions: http://bst.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations http://bst.sagepub.com/cgi/content/refs/28/6/436
Downloaded from http://bst.sagepub.com by on June 29, 2010
Bulletin of Science, Technology & Society Volume 28 Number 6 December 2008 436-453 © 2008 Sage Publications 10.1177/0270467608315942 http://bsts.sagepub.com hosted at http://online.sagepub.com
Wind Power in Ontario Its Contribution to the Electricity Grid Ian H. Rowlands University of Waterloo, Ontario, Canada
Carey Jernigan NSCAD University, Halifax, Nova Scotia, Canada The purpose of this article is to investigate wind turbine production, the variability of that production, and the relationship between output and system-wide demand. A review of the literature reveals that a variety of measures (and methods) to explore the variability of wind power production exist. Attention then turns to the province of Ontario (Canada), and the performances of four wind farms are examined for 2006 and 2007. Key conclusions include that the wind farms’ capacity factors vary from 27.6% to 35.6%, with higher values in winter as compared to summer; wind power performs better than the seasonal average during peak periods; wind is a better “partner” for the Ontario electricity system in the winter as opposed to the summer; and the increased geographic distribution of wind farms decreases their collective variability. Keywords:
I
Canada; electricity; energy policy; grid performance; markets; Ontario; wind energy
nterest in renewable energy sources has been growing in recent years. This has been motivated by a variety of environmental concerns, not least of all the air quality impacts (smog and global climate change) generated by conventional, fossil fuel–burning power plants (Goldemberg & Johansson, 2004, pp. 40-42). “Energy security” is also a concern on the rise (Yergin, 2006), catalyzed by rising oil prices (Organization of the Petroleum Exporting Countries, 2007), fear of terrorism (Asmus, 2001), the war in Iraq (Yankelovich, 2006), and diminishing reserves of conventional fuels (Deffeyes, 2005). In this context, clean, local, and renewable energy sources are becoming more appealing and thus attracting more attention. The expansion of wind power—one of the world’s leading renewable energy sources—has proceeded accordingly: In 2006, installed wind energy capacity increased 32% worldwide (Global Wind Energy Council, 2006). But with this increased attention has also come increased scrutiny. Although wind power’s attributes are lauded by its supporters, critics maintain that wind’s variability—that is, the fact that it is available not “on demand” but only when the wind is blowing
(what these critics often call “intermittency”)—means that its contribution is ultimately limited: To maintain electricity system security, they continue, its contribution to the grid must be capped. Indeed, in a study on wind power’s role in Ontario (Canada), Adams (2006) worried that “a clean and promising generating technology is being burdened with unrealistic forecasts of reliable production” (p. 1). In this article, we assess the contribution of wind farms to the electricity grid in the province of Ontario (Canada) in 2006 and 2007. Specifically, we explore wind turbine production, the variability of that production, and the relationship between output and system-wide demand; we then briefly discuss strategies for the future.
Electricity in Ontario The article investigates the province of Ontario, Canada’s most populous province and the country’s second largest generator of electricity (of 10 Canadian provinces). In 2005, total Ontario electricity demand reached 157 TWh,1 with nuclear generating stations
Authors’ Note: Support for part of this research was provided by a grant from the Social Sciences and Humanities Research Council of Canada (through its support of a project titled Business and Green Power in Electricity Transformation: Markets and Policies, in the Research Developments Initiative program; see http://www.fes.uwaterloo.ca/research/greenpower). The authors are grateful for this support. The authors would also like to thank Glen Estill and Paul Gipe for their comments on earlier drafts of the article. The authors, however, remain fully responsible for all of the contents of the article.
436
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 437
making the largest contribution to the province’s electricity supply (54.1%), whereas hydroelectric (22.3%), coal (16.0%), and natural gas (6.4%) stations also were major contributors. Other supply sources, including all renewables, accounted for the remaining 1.2% (Government of Ontario, 2007). Peak system demand, meanwhile, reached 26,160 MW on Wednesday, July 13, 2005, at 5 p.m. (eastern daylight time [EDT]). That was surpassed the following year, when Ontario’s demand reached 27,005 MW on Tuesday, August 1, 2006, at 5 p.m. (EDT). But although Ontario is a “summer peaking” system, total electricity demand is higher in the winter than the summer.2 Turning more specifically to wind power, there are presently four wind farms in Ontario (i.e., locations at which there are at least 10 wind turbines in close proximity to each other), with an installed capacity of, collectively, 396.1 MW. A number of individual grid-connected wind turbines also exist in the province—together, these 15 turbines (ranging from 0.6 MW to 1.8 MW) account for 20.1 MW of installed capacity (Canadian Wind Energy Association [CanWEA], 2007).3 A number of additional wind projects are currently in development—both wind farms and individual turbines. (Both larger- and smaller-scale wind projects have been encouraged by the province of Ontario’s two-track strategy with respect to renewable electricity: Larger projects have been catalyzed by a tendering process, whereas smaller ones have been encouraged by the introduction of “standard offer contracts” [Rowlands, 2007].) Ontario’s electricity system operator—that is, the “Independent Electricity System Operator” (IESO)—anticipates that by the end of 2008, five additional wind farms will be fully operational: the Ripley Wind Power Project (a 76 MW project located on the shores of Lake Huron), the Kruger Energy Port Alma Wind Power Project (a 101 MW project located on the shores of Lake Erie), the Malanchton II wind project (a 132 MW project located in central Ontario, near the existing Amaranth Wind Farm), the Wolfe Island Wind Project (a 198 MW project located on Wolfe Island in Lake Ontario, Wolfe Island, near Kingston, Ontario), and the Enbridge Ontario Wind Power Project (a 200 MW project located near Kincardine, Ontario) (IESO, 2007a, p. 13). Moreover, in its “Integrated Power System Plan,” the Ontario Power Authority (OPA) anticipates that the installed capacity of wind power will steadily grow during the coming 20 years, reaching 4,685 MW by 2025 (OPA, 2007b, p. 10). Wind proponents say that this level could be safely much higher, for they maintain that desirable off-shore resources were not given sufficient attention in the OPA’s analyses (e.g., Hornung, 2006).
In any case, the debate about wind’s present and future roles in the province of Ontario continues. On the one hand, many are calling for a more prominent position for wind, arguing that its potential contribution is often misunderstood. For example, the national wind energy industry association advances the following position: Everyone knows that the wind is variable. Sometimes it blows, other times it doesn’t. So how can wind power be a reliable source of energy? . . . The fact is, the wind will never stop blowing everywhere at once. . . . With Canada’s large and varied wind resource, there’s no doubt that the wind can power us well into the future. (CanWEA, 2006)
Others, by contrast, challenge any proposed largescale role that wind might play. The president and chief executive officer of one of Ontario’s largest nuclear power plants, for instance, maintains that renewables such as wind and solar power are certainly environmentally friendly. Yet the truth is they cannot be relied upon to keep our factories humming and homes warm because of their intermittent availability and limited scale. Renewables have their place in our energy mix, but we can’t be naive enough to believe they are the large-scale answer. (Hawthorne, 2004, p. 6)
It is against this background that we investigate, in this article, the contribution of wind to the Ontario electricity system.
Wind Power’s Performance in an Electricity Grid The degree of variability of the output of a wind turbine has significant consequences for the potential contribution that it can make to the wider electricity system. If it varies significantly, then the other generators will have to respond accordingly. In addition, the extent to which changes in wind energy output positively correlate with fluctuations in system-wide electricity demand also has important implications for the integration of wind power into the electricity grid. If the two vary in tandem, then increased wind power would reduce both “load-following” requirements (how often generators must alter their power output to match demand, therefore operating less efficiently than with continuous output) and “operating reserve” requirements (stand-by capacity that is kept online in case the power system suffers a supply interruption). If, on the other hand, wind
Downloaded from http://bst.sagepub.com by on June 29, 2010
438 Bulletin of Science, Technology & Society
output randomly fluctuates or negatively correlates with system-wide operating demand, then load-following requirements and the need for larger operating reserves would increase. Thus, the pattern of wind power production (both in an isolated sense and compared to the demand patterns in the broader electricity system) will determine the “value” of its contribution to the electricity grid (e.g., DeCarolis & Keith, 2005; Gipe, 2006b; Holttinen et al., 2007). As such, a better understanding of wind power’s contribution to a power grid as societies consider increased use of renewable resources is critical. Hence, can we identify patterns in wind power production and its relationship to demand? Are they consistent over time? Are they beneficial? Do they differ from one location to the next? How quickly do changes occur? In this section, we review the focus and findings of previous studies on wind power performance, with specific emphasis on patterns of production.
Wind Power Over Time When investigating wind power’s patterns of production, time scale is important. DeCarolis and Keith (2006) argued that three timescales “concern system operators on a day-to-day basis: minute-to-minute, intrahour, and hour- to day-ahead scheduling” (p. 397). For his part, Milborrow (2007) suggested that power changes within an hour should be the focus, for “these tend to have the strongest influence on the ‘costs of variability’” (p. 38). Although most studies have relied on hourly data for ease of access (e.g., Holttinen, 2003; Parsons, Wan, & Kirby, 2001), others have used finer-resolution records to better reflect specific aspects of grid operation. Still others have examined a range: A study carried out by General Electric International Inc. (GE, 2006) in Ontario, for example, examined daily, monthly, and yearly trends as well as 3-hour sustained ramping, 1hour generation scheduling, 10-minute operating reserve requirement, 5-minute load-following requirement, and 1-minute regulation requirement time frames. The study found that variability tends to increase over time; Parsons et al. (2001) found the same pattern in a wind farm in Minnesota. Although the rate of change in wind output has consequences at all time scales, the nature of the broader electricity grid (and the associated peak demand periods generated by its users) means that particular time periods are particularly significant. For many, the day-night distinction is important to make. The Danish Wind Industry Association (DWIA), for example, reported that more wind is typically available during the day than at night, so that “wind electricity generally fits well into
the electricity consumption pattern” (DWIA, 2003d; see similar examinations in, e.g., Li & Li, 2005). Moving from a consideration of changes within a day to changes within a year, many have found wind power performance to vary according to the season. Several wind industry associations (e.g., CanWEA, 2006; DWIA, 2003c), for example, have highlighted the fact that wind output is typically high in the winter, when electrical heating demand rises in high-latitude locations (for additional reflections on seasonal variations, see Belanger & Gagnon, 2002; Milborrow, 2007; Sinden, 2007a). Rather than focusing on times of the day or seasons of the year, still others have highlighted the importance of focusing on high-demand periods, determined by different means—for example, “the top 20 per cent of demand hours” (Sinden, 2007b, p. 65). Holttinen (2003) suggested that wind output varies with temperature more importantly than with season. Moreover, this relationship may not be linear. In Germany, E.ON Netz (2005) found that “cold wintry periods” and hot summer days, when heating and air conditioning costs are highest, respectively, are often symptomatic of high-pressure weather systems when winds are typically low. In other words, wind output may be highest in winter on average but low on cold, still days when heating demand peaks; similarly, wind output in summer may not be high when cooling demand peaks. In the United Kingdom, meanwhile, it has been suggested that large anticyclones with little wind are associated with low temperatures in the winter (when, as a result, “heating and lighting loads can also be at maximum”) and with clear skies and high temperatures in the summer (when “increased demand for cooling and air conditioning result”; Laughton, 2007, pp. 8, 9, 14; also see Adams, 2007). Finally, a consideration of wind power over time can move from examination of performance within a year to investigation of output across years. Estimating output from wind speeds taken over 30 years in the United Kingdom, Sinden (2007a) found that average annual production ranged from 24.1% to 35.7% of installed capacity. Meanwhile, the DWIA (2003c) predicted a 9% to 10% variation in wind output, year to year.
Wind Power Over Space In addition to time, geography plays an important role in wind power’s contribution. First, the conditions of the site at which the turbines are located—the roughness of the terrain and the topography more generally, for example—affect wind power output. Offshore wind turbines therefore produce, on average, more electricity
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 439
than do nearby onshore sites, for grasses, trees, hills, and/or buildings can increase turbulence in the case of the latter (DWIA, 2003b). Second, weather systems are of limited geographic extent and will often pass through different sites within a region at different times; as a consequence, many argue that increasing the geographic dispersion of wind generators can increase the overall reliability of wind power (as a whole). E.ON Netz (2005), Holttinen (2003), Milborrow (2007, p. 36), and NBSO (2005) are among those who have argued that greater geographic dispersion of wind farms improves overall reliability and can reduce forecasting error so that changes in wind output can be better prepared for. (And, indeed, many have suggested that the further apart, the better—that is, the less correlated [e.g., Sinden, 2007a, 117-118; also see Gross et al., 2006].) In a study of Minnesota and the eastern parts of North and South Dakota, it was found that the geographic distribution of wind production “provides for substantial ‘smoothing’ of wind generation variations . . . [and] the number of hours at either very high or very low production are reduced, allowing the aggregate wind generation to behave as a more stable supply of electric energy” (EnerNex Corporation, 2006, p. xxi). Similarly, Holttinen et al. (2007) argued: Because of spatial variations of wind from turbine to turbine in a wind power plant—and to a greater degree from wind power plant to wind power plant— a sudden loss of all wind power on a system simultaneously due to a loss of wind is not a credible event. (p. 12)
More recently, The Economist (“Where the Wind Blows,” 2007) observed, “The question of where the wind is blowing [in Europe] would no longer matter because it is almost always blowing somewhere” (p. 81; also see Drake & Hubacek, 2007).
Method and Data The focus and findings of research on wind power performance not only vary among studies but also reveal significant differences with respect to the methods deployed. Although some researchers analyze production data taken from the operations of wind turbines, others use simulated output. These latter values are calculated based on a location’s wind speed (ideally at turbine hub height and often described using the Weibull distribution or other models [Li & Li, 2005]), perhaps accompanied by temperature (which affects air density
and thus the power of wind), a measure of surface roughness (DWIA, 2003b), and an estimate of availability losses resulting from blade soiling, icing, or other maintenance concerns (GE, 2006). A projected power output curve for a given turbine is then developed and used to simulate wind power performance across time. It is generally accepted that because wind speed measurements are often taken at one location on a site (rather than for each turbine) and roughness and availability losses are not entirely predictable, actual output should be used wherever possible (Holttinen, 2003). It remains, however, that a number of wind performance investigations are taken in advance of wind turbine construction (in order to assist with planning), and therefore simulated output continues to be widely used. Once output data are collected or simulated, various methods are used to study wind power performance. Milligan and Porter (2006) suggested that effective load carrying capacity (ELCC) techniques are most useful for electricity grid planning because they consider wind power’s role within a larger system (also see the discussion in Holttinen et al., 2007, p. 77). Calculations can accommodate hourly load requirements, hourly output data for renewable generators (e.g., wind) and hourly output data for conventional generators (e.g., rated capacity, forced outage rates, and maintenance schedules for coal, natural gas, nuclear, or other power plants). The technique then compares the loss of load expectation (the probability that power generation will fail to meet demand) for the system including wind electricity versus the same system without wind electricity. ELCC can also be used to determine what combination of renewable energy generators are required to achieve the same risk level as conventional technologies, which conventional facilities can be retired once renewable energy generators come online, or how much demand can be accommodated by different supply scenarios (Milligan & Porter, 2006). If, instead, we look only at the relationship between wind output and demand, we ignore interaction with other elements in the system. For example, controlled hydropower dispatched during peak demand hours, power purchased from outside the local area at these times, or routine maintenance that shuts down large generating facilities during low peak seasons (e.g., spring and fall) can shift “high-risk” periods considerably (Milligan & Porter, 2006). However, there are significant data requirements associated with the use of ELCC techniques. Where data describing all system components are not available, or where research specifically focuses on the relationship between wind electricity generation and
Downloaded from http://bst.sagepub.com by on June 29, 2010
440 Bulletin of Science, Technology & Society
system-wide demand, simpler “time-period based methods” are often used instead. Average and/or peak levels of wind power output are plotted against time, and patterns, rates of change, or relationships to system-wide demand are discussed. This approach has the benefit of simplicity and ease of explanation, but data must be appropriately interpreted (Milligan & Porter, 2006). A disadvantage, however, is that these methods are not capable of assessing and finding times that the system may be at risk even though loads are not especially high. If a significant fraction of the generating capacity is on maintenance during the shoulder seasons, this can cause a potentially large increase in LOLP and can result in potentially much higher risk than peak periods. (Holttinen et al., 2007, p. 83)
It is striking that statistical pattern analysis and confidence intervals are largely absent in the literature. This is perhaps indicative of small sample sizes, where data are often only available for a short period of time (often, given wind’s recent growth, less than 1 year), or perhaps reflects the fact that conditions rapidly change with time (as new wind projects are installed or system-wide demand patterns change) and across space so that results from one study rarely reflect those of others. Also, rather than tracking total production (kWh), studies often use ratios (Holttinen, 2003) to facilitate comparison between local conditions and future or distant sites. Perhaps the most commonly used metric in the investigation of wind power performance is “capacity factor” (CF). CF is defined as the ratio of actual electricity production to total installed capacity—the theoretical maximum output or “nameplate capacity” of the generator in question. Variations on CF include “capacity credit” (Milligan & Porter, 2006) and “capacity value” (GE, 2006), which focus on wind output during periods of high demand (e.g., the top 10% of peak load hours or between 3 p.m. and 7 p.m. in the summer). These metrics (often defined differently in different studies) are intended to reflect the “relevance” of variable power sources to overall system planning—the amount of installed conventional power that can be replaced by renewable supply (Voorspools & D’haeseleer, 2006), for example, or the “firm” generating capacity that can be counted on during periods of peak demand (GE, 2006). Although CF is commonly used in the literature, its appropriateness has been called into question by some (e.g., Gipe, 2006a; Wiser & Bolinger, 2007). DWIA
(2003a) referred to a Capacity Factor Paradox: Unlike with other generation technologies, a high CF for wind does not necessarily reflect greater overall production. Wind generators vary in their “cut-in” and “cut-out” wind speeds (the wind speed at which they start or stop producing electricity) and also in the speeds at which they operate most efficiently. Depending on wind conditions at a given site, a turbine with a relatively large generator (relative to rotor diameter) may operate at a low average CF most of the time but be able to take advantage of high winds when they occur. A relatively small generator might operate close to its design limit most of the time (and so have a high average CF) but will “cut out” during high wind events. As wind power production varies with the cube of wind speed, overall production may drop significantly. Wiser and Bolinger (2007) and Gipe (2006a) suggested that, instead of CF, electricity generation per m2 of swept rotor area (kWh/m2) is a better measure of wind power. Gipe referred to this metric as “specific yield” and argued that the area “swept” by a turbine’s blades determines the amount of the available wind resource that is actually intercepted by a turbine and so can potentially be converted to electricity. There is no generally accepted means by which a wind turbine’s performance—particularly with respect to its contribution to the broader grid—is measured. Instead, a variety of approaches, each emphasizing particular elements, continue to be deployed (also see Gross et al., 2006). Keeping many of these methods on hand, we now turn to the particular case of Ontario in order to investigate wind power in that province.
Data: Wind Power Production in Ontario In this article, we investigate the wind power production from the four existing wind farms in Ontario. (Please see Figure 1 for a map.) They are as follows: 1.
2.
Amaranth: Located near Shelburne, in Melanchton Township, approximately 100 km northwest of Toronto, forty-five 1.5 MW GE turbines (each with three 37 m blades) went into operation on March 4, 2006. Owned by Canadian Hydro Developers, the wind farm has a total installed capacity of 67.5 MW (Canadian Hydro Developers, 2006, p. 2; Canadian Hydro Developers, 2007). Kingsbridge: Located on the shore of Lake Huron (near Goderich), 22 turbines—each a 1.8 MW Vestas turbine—began generating power in March 2006. The blades on each have a length of 39 m. Owned by EPCOR Utilities Inc., the total installed capacity is 39.6 MW (EPCOR Ontario, 2007).
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 441
Figure 1 Location of Four Wind Farms in Ontario
Table 1 Total Energy Production, Ontario Wind Farms, March 2006 to November 2007
ONTARIO Prince Amaranth
Total Production (MWh)
Average Daily Production (MWh)
Period Beginning (Ending November 30, 2007)
284,598 182,677 342,230 544,552
444.7 285.4 579.1 1,148.8
March 1, 2006 March 1, 2006 April 19, 2006a August 14, 2006a
Amaranth Kingsbridge Port Burwell Prince
a. For Port Burwell and Prince farms, “starting dates” are those days on which a reading of at least 2 MWh for 1 hour’s production is noted on the Independent Electricity System Operator Web site.
Kingsbridge Port Burwell
Source: Original map data provided by The Atlas of Canada http://atlas.gc.ca/ © 2008. Data reproduced with permission of Natural Resources Canada.
Figure 2 Average Daily Energy Production (MWh), Ontario Wind Farms, April 2006 to November 2007
by Brookfield Power. There are 126 GE turbines (again, with a capacity of 1.5 MW each, and again with diameter of 77 m), and they were deployed in two stages, with full operation achieved in November 2006. Total installed capacity is thus 189 MW (Brookfield Power, 2007; Hatch Energy, 2006).4
Results: Wind Power Performance in Ontario Our investigation of wind power’s contribution in Ontario proceeds in two parts. First, we consider the operation of the wind farms individually—that is, without any reference to the broader (Ontario) system in which they are located. And second, we investigate the performance of the wind farms set against each other as well as the system-wide (provincial) demand.
Individual Wind Farms’ Performance
3.
4.
Port Burwell: Also known as the Erie Shores Wind Farm, these 66 GE turbines (1.5 MW each) are located on the northern shoreline of Lake Erie. The project “was commissioned in May 2006 and began commercial operation in June 2006.” With a total capacity of 99 MW (and with each turbine having a 77 m blade diameter), the wind farm is owned by AIM PowerGen (Stapletonprice, 2007). Prince: Located in the Sault Ste. Marie District (northern Ontario), this project has been developed
Figure 2 provides information regarding the energy production of Ontario’s wind farms, between April 2006 and November 2007.5 From this, it is clear that production is higher during the winter months as compared to the summer months. In terms of total production for each wind farm, meanwhile, Table 1 provides some details; note that, in this table, we present data for a longer time period than in Figure 2. CF, as defined in the third section of this article, is an additional way of investigating the performance of wind farms. Given the production figures presented in Figure 2, along with the stated nameplate capacities of the wind farms, the CFs can be calculated. These results are presented in Figure 3. For all four wind farms, the highest CFs occur in either February or March, the lowest in either July, August, or September. For the entire periods
Downloaded from http://bst.sagepub.com by on June 29, 2010
442 Bulletin of Science, Technology & Society
Table 2 Annual Capacity Value Factors for Ontario Wind Farms, Selected 12-Month Periods, April 2006 to November 2007 Amaranth (%)
Kingsbridge (%)
Port Burwell (%)
Prince (%)
28.9 29.6 29.4 29.7 29.4 29.5 29.7 29.3 30.2 29.5
32.0 33.2 33.5 33.7 33.6 33.7 34.1 34.2 35.6 33.7
—a — — 29.0 28.9 29.0 27.8 27.6 28.9 28.5
— — — — — — — — 28.7 28.7
April 2006 to March 2007 May 2006 to April 2007 June 2006 to May 2007 July 2006 to June 2007 August 2006 to July 2007 September 2006 to August 2007 October 2006 to September 2007 November 2006 to October 2007 December 2006 to November 2007 Average a. Not applicable.
Figure 3 Monthly Capacity Factor Values for Ontario Wind Farms, April 2006 to November 2007
Table 3 Production and Specific Yield, Ontario Wind Farms, December 2006 to November 2007 Port Amaranth Kingsbridge Burwell Prince
60 50
Production (MWh) 178,679 Swept area per 4,657 turbine (m2) Total swept area (m2) 209,565 Specific yield 85.3 (kWh/m2)
40 30 20
123,508 5,026
250,854 474,905 4,657 4,657
110,572 111.7
307,362 586,782 81.6 80.9
10 0
Amaranth
Kingsbridge
Port Burwell
Prince
under investigation, the CFs are, highest to lowest, 30.5% (Kingsbridge), 28.7% (Prince), 27.6% (Amaranth), and 26.8% (Port Burwell). Recognize, however, that these annual CFs include, for three of the four wind farms, a higher proportion of summer months (times at which production figures are lower). Therefore, to consider annual CFs for Amaranth, Kingsbridge, and Port Burwell, one can consider any of a variety of 12-month periods. From the results in Table 2, it is clear that the CF, in every case, rises, as compared with the CF for the entire study period. Specific yields can also be calculated. Given the lengths of the individual turbines’ rotor blades noted
above,6 Table 3 provides calculations for the wind farms’ specific yields. Wanting not to have a particular season influence the calculation, we consider only the contiguous 1-year period of December 2006 to November 2007 in each case. (Indeed, from this point forward, we focus on this 1-year period for this very reason.) Although Kingsbridge continues to be the top performer on the specific yield measure (as it was in terms of CF), Port Burwell and Prince swap places in terms of their respective rankings. Basic data can also be presented on an hourly basis. In Figure 4, data for the four wind farms are presented, across the “24-hour clock” (with, it must be noted, all times presented in eastern standard time [EST]). The morning period is generally a lower time for all wind farms, with three of the four experiencing their lowest hourly value (indicated by a dashed circle) at either 9 a.m. or 10 a.m. The respective peaks (noted by the solid circle) are more dispersed across the day (3 a.m., 2 p.m., 3 p.m., and 9 p.m.). In addition, note that the difference between the lowest and highest values was modest,
Downloaded from http://bst.sagepub.com by on June 29, 2010
Rowlands, Jernigan / Wind Power in Ontario 443
Figure 4 Average hourly energy production (MWh), Ontario wind farms, December 2006 to November 2007
Figure 5 Percentage of Time During Which Wind Farms’ Capacity Values Are Above, Within, or Below Particular Threshold Values, Ontario, December 2006 to November 2007
70
100% 60
90%
50
80%
40
70%
25.4
24.1
23.9
55.9
58.2
20.0
17.8
Port Burwell
Prince
32.8
60% 30
50% 20
40%
10
30%
56.4 53.5
20%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hour (EST) Amaranth
Kingsbridge
10%
18.3
13.8
0% Port Burwell
Prince
Amaranth >50% capacity
Kingsbridge
2%
