This paper examines Vermont Public Interest Research Group’s (VPIRG) assertion that by 2015 industrial wind turbines on 8.8% (or 46 miles) of Vermont’s ridgelines above 2500 feet could provide 20% of Vermont’s electricity needs. (1) The examination compares VPIRG’s proposal- which is predicated on Vermont’s average electricity consumption- with the utility industry’s standard for measuring wind energy’s contribution to system reliability and peak demand. i.e. its capacity credit. This measurement concludes that for wind energy to provide the reliable generating capacity to meet 20% of Vermont’s peak demand industrial wind turbines would require 44% - 88% (or 226-451 miles) of Vermont’s ridgeline above 2500’.
This conclusion is derived from available Searsburg data and wind energy’s apparent capacity credit in California, Texas, Ontario and Germany.
To provide perspective, a brief review of how electricity grids work precedes this comparative analysis.
This paper concludes with an overview of three related issues- emissions, transmission and governmental incentives. It argues that (1) emissions may or may not be reduced depending on wind energy’s grid specific impact, (2) wind energy poses significant problems to transmission related operations and costs when installed wind capacity reaches 10%-20% of peak demand capacity and (3) incentives promoting renewable energy alternatives should, in the interest of optimizing investment returns and minimizing ratepayers’ costs, reflect their respective verifiable contributions to on-peak/on-season energy generation, system stability and reliability as well as net emissions reduction.
Electricity Grids: How They Work
An electric power grid’s primary responsibility is to provide electricity on demand 24/7/365. This means the grid must match aggregate production and consumption instantaneously, continuously and reliably.
Following policies developed and enforced by the North American Electric Reliability Council (NERC) (3), each grid must have sufficient generating sources that will:
(1) ensure ‘adequacy’ - the ability of the electric system to supply the aggregate electrical demand and energy requirements of the end-use customers at all times, taking into account scheduled and reasonably expected unscheduled outages and
(2) accommodate ‘contingencies’ - the unexpected failure or outage of a system component such as a generator, transmission line, circuit breaker, switch or other electrical element.
This means, in addition to having ‘contingency reserves’, the grid must routinely have generating sources that not only meet base load demand (the minimum amount of electric power delivered or required over a given period at a constant rate) but also meet peak demand (the highest hourly load within a given period-day, month, season or year) as well as the operating flexibility to respond to everything in between.
Since cost is a critical component in source selection, large reliable power plants with low fuel costs are generally dedicated to providing base load power, e.g. coal, nuclear power and (at least seasonally) conventional hydro power (the variable/fuel cost of hydro and nuclear power is essentially zero). Additional capacity is brought on-line on a daily/seasonal basis to meet routine increases in demand and peak demand. Routine, fairly predictable increases in demand can be accommodated by coal, biomass, and hydro (if available). Natural gas and oil are also used but, in recent years, they have become less attractive cost alternatives. To meet short duration peak demand, the key requirements are fast startup and low investment cost, i.e. generally fossil fuel generating units, particularly natural gas units.
Conceptually, the grid’s need to match reliably aggregate production and consumption instantaneously and continuously can be met by controlling generation, demand, or both. In practice, grid operators historically have preferred to control generation almost exclusively.
Consistent with their primary responsibility to provide electricity on demand 24/7/365, grid managers focus intently on each generating source’s capacity credit which is a measure tied to peak demand and reflects a generating source’s contribution to system reliability.
Comparative Analysis: Average Capacity vs. Capacity Credit
As detailed in Exhibit A, VPIRG contends that by 2015 20% of Vermont’s annual electricity needs could be provided by industrial wind energy (1) operating, on average, at 32% of installed capacity (450MW), (2) consisting of 274, 1.65MW turbines and (3) occupying 46 miles of ridgeline or 8.8% of Vermont’s 517 miles of ridgeline above 2500 feet.
VPIRG’s proposal is predicated on projected Vermont annual electricity consumption of 6,200,000MWh which equates to an average daily and hourly consumption of 16,986MWh and 708MW respectively.
The standard measurement, however, employed by the utility industry to determine a generating unit’s contribution to meeting electricity needs is capacity credit, i.e. its reliable contribution to peak demand (Vermont’s = 1,100MW). This measurement, applied to VPIRG’s scenario, concludes that for wind energy to provide the reliable generating capacity to meet 20% of Vermont’s peak demand industrial wind turbines would require 44% - 88% (or 226-451 miles) of Vermont’s ridgeline above 2500’. VIPRG’s 46 miles of ridgeline would contribute 2.1%-4.1% of the electricity generating capacity required to meet Vermont’s peak demand.
A generating unit’s capacity credit is derived from its reliable contribution over time to meeting peak demand in the hour in which annual peak demand occurs. Although Green Mountain Power (GMP) does not provide Searsburg’s contribution to peak demand during the hour it occurred, its data does show Searsburg’s capacity during the month in which peak demand occurred—a range of 8% (August 2001, 2002 & 2003) to 16% (December 2004 and July 2005). (Exhibit B) Of note is that Vermont and New England appear to have set a new peak demand record last August when winds were light. (4)
As capacity credit is a measure of a generating unit’s reliability in meeting peak demand, it can only be established with confidence based on actual experience over time. In the absence of empirical evidence , this comparative analysis assigns an capacity credit to Vermont wind energy of between 5% and 10% based on the aforementioned GMP data and, as addressed below, wind energy’s apparent capacity credit in California, Texas, Ontario and Germany.
Wind Energy’s Capacity Credit
As an intermittent, uncontrollable and largely unpredictable (except in the very short term) generating source, wind is inherently unreliable. When the wind does blow within required parameters, wind turbines do generate energy. However, because wind is often ‘off-peak and off-season’, its ‘effective capacity’, and thus its capacity value for meeting peak demand, is limited.
A recent study determined that it is difficult to generalize the capacity credit for wind as ‘it is a highly site-specific quantity determined by the correlation between wind resource and load”….with values ranging “from 26 % to 0% of rated capacity”. (5)
This conclusion is based, in part, on a 2003 study by the California Energy Commission that estimated that three wind farm aggregates- Altamont, San Gorgonio and Tehachpi, which collectively represented 75% of California’s deployed wind capacity- had relative capacity credits of 26.0%, 23.9% and 22.0% respectively. While we do not know how these three facilities performed during California’s summer ’06 energy crunch, we do know how California’s aggregate wind power portfolio performed. As has been widely publicized in the press, California wind power produced at 254.6 MW (10.2% of wind’s rated capacity of 2,500MW) at the time of peak demand (on July 24th, 2006) and over the preceding seven days (July 17-23) the facilities produced at 4% of rated capacity. (6)
Addressing wind energy’s performance in California during the July 06 ‘heat storm’, American Wind Energy Association (AWEA) spokesperson, Christine Real de Azua, explained- “You really don’t count on wind energy as capacity. It is different from other technologies because it can’t be dispatched.” (emphasis added) (7)
The following excerpt from the Electricity Reliability Council of Texas’ (ERCOT) 2005 study suggests a more conservative assessment of wind energy's capacity credit. (8)
In addition to meeting the state’s energy needs (MWh), the electric system must also meet expected peak demand (MW). Generation resources other than wind will be needed to meet most of the projected growth in peak demand, as maximum output from wind resources does not correspond to system peak demand. (emphasis added) ERCOT currently assigns 10% of the installed capacity of wind turbines to its calculation of the ERCOT peak capacity reserve margin. Based on a review of historical data of actual wind turbine generation during ERCOT system peaks (from 4 p.m. to 6 p.m. in July and August), the average output for wind turbines was 16.8% of capacity. However, the data also showed that for any hour during these months, the output of the wind turbines could range from 0% of installed capacity to 49% of installed capacity. Stakeholders comprising the ERCOT Generation Adequacy Task Group have expressed concern that use of an average number (i.e., 16.8%) was too optimistic because it fails to adequately recognize the intermittency of wind generation. Accordingly, the group is working to assign a peak capacity value for wind using an appropriate “confidence factor.” While the group has not yet formally made a recommendation to the ERCOT Technical Advisory Committee, it is currently considering recommending a wind capacity value of 2%. In summary, in order to reliably meet system peak demand, dispatchable resources (such as gas, coal, biomass) would be required to replace the wind resources when wind is not blowing." (emphasis added)
ERCOT’s more conservative assessment is shared by Energy Probe’s recent study of Ontario wind plants: (9)
In Ontario, the IESO assumes that 10% of the installed capacity should be considered as firm capacity for meeting peak demands. A Pembina Institute study has commented on this assumption, “Given that the capacity factor for modern land-based wind turbines is accepted to range from 25%–40%, and that wind generating capacity in Ontario will be relatively geographically distributed, this may be an excessively conservative assumption.”
Both the GE Study conclusion and the IESO’s forecast about firm summer peak reliability are inconsistent with Ontario’s actual experience. (emphasis added)
During July and August 2006, the actual average frequency of hours when there was little or no wind output in Ontario – output less than 2% – was 18.6%. These very low production hours were about as likely to occur during the daily peak period as any other time during the day. Ontario’s experience in 2006 shows that the conclusion of the GE Study that wind can reliably supply power in summer equal to 17% of its rated capacity significantly over-estimated the actual results. The actual results for the summer of 2006 also suggest that the IESO should review its forecast that even 10% of the installed wind capacity should be considered as firm capacity for meeting peak demands. During the summer of 2006, wind power provided no firm generation capacity during the peak months. (emphasis added)
The German grid operator, Eon Netz, - one of the world’s largest managers of wind energy addresses wind’s capacity credit in its 2005 Annual Report as follows: (10)
Wind energy is only able to replace traditional power stations to a limited extent. (emphasis added) Their dependence on the prevailing wind conditions means that wind power has a limited load factor even when technically available. It is not possible to guarantee its use for the continual cover of electricity consumption. Consequently, traditional power stations with capacities equal to 90% of the installed wind power capacity must be permanently online in order to guarantee power supply at all times. [emphasis added]
While capacity factors for U.S. industrial wind plants are known, (11) capacity credits for these facilities do not appear to be available. Even if available, it would be inappropriate, as noted above, to extrapolate a given wind plant’s capacity credit to other sites as it is highly site-specific. That said, the above experiences suggest that, in the absence of hard evidence to the contrary, caution should prevail when projecting wind energy’s capacity credit. Experience belied the presumed capacity credit for wind energy in Texas and Ontario. In California, the previously referenced reports (6) described wind energy’s performance during summer 2006 as disappointing. (12)
A. CO2 Emissions: (13)
A primary reason for considering wind as ‘part of our energy mix’ is that it is emissions free; and that every MWh produced by wind energy displaces one MWh produced by other electricity generating sources. While theoretically true for electricity generation- given the grid’s matching of production and consumption instantaneously and continuously- it is not true for emissions because wind is intermittent and often ‘off-peak’.
Because wind is intermittent, grids must have ‘fast responsive’ backup generating sources available at all times to ensure grid stability and reliability.
The ‘fast responsive’ generating units of choice for most grids to backup wind’s intermittency are conventional hydro power (if available) and natural gas. Hydro is emissions free while natural gas is relatively emissions benign. (14)
Because wind is often ‘off-peak’, e.g. it blows more at night when demand is low and less during the day when demand is high, grids aren’t always able to ‘ramp down’ base-load generating units that are economically and operationally designed to generate electricity continuously.
The base-load generating units of choice for most grids are coal, nuclear power and/or conventional hydropower (if available). Nuclear power and hydropower are emissions free. Large ‘slow responsive’ coal generating units designed to provide base-load capacity are not suited- economically or operationally- for providing ‘standby’ generating capacity. Where wind generation displaces coal generation the emissions emanating from ‘standby’ coal units- operating inefficiently- may offset assumed emissions savings. (15)
In sum, the emissions displaced - if any- by wind energy is grid specific, i.e. emissions displacement is a function of which and how specific generating units are impacted by wind generation. Perhaps more importantly, given wind energy’s limited capacity credit, future emissions levels will be determined largely by the reliable generating sources we choose to meet growing electricity demand.
Vermont CO2 emissions are generated by transportation (56.4%), followed by residential (24.5%), commercial (10.7%), industrial (7.9%) and electric power (0.47%). (16)
Vermont’s share of NEPOOL CO2 emissions from electricity generation is statistically insignificant- in 2005, Vermont generated 13,582 metric tons of CO2 or 0.02% of total NEPOOL CO2 emissions. NEPOOL, in turn, accounts for 2.24% of total US CO2 emissions from electricity generation. (17) This performance is attributable to Vermont’s emissions benign electric generation portfolio. (18)
These figures suggest that Vermont environmental initiatives should focus on reducing automobile/truck emissions and on expanding current programs designed to improve residential efficiency and conservation.
Recent studies (19) suggest that the impact of wind’s intermittency on a grid’s operations and associated costs can become significant when installed wind capacity reaches 10%-20% of peak demand based upon capacity. Wind energy’s impact on a specific grid will depend on its ‘robustness’, specifically whether transmission constraints exist or not between the location of wind generation and demand areas. At Vermont’s current peak demand of 1,100MW and assuming Vermont’s transmission capacity is adequately robust, this suggests that Vermont could accommodate 110MW- 220MW of installed wind capacity before encountering significant operating and cost issues. This range of installed capacity would mean 67-133 turbines on 11-22 miles of ridgeline providing between 0.5% and 2.0% of Vermont’s peak demand. (20)
Legislation promoting renewable energy (e.g. Production tax credits (21), Renewable Portfolio Standards, Renewable Energy Credits, etc) should reflect the value of their respective contributions to (1) ‘on-peak and on-season’ energy generation, (2) system stability and reliability and (3) net emissions reduction as well as require each and every facility to provide verifiable and transparent evidence of these benefits. For example, current incentives generally do not differentiate between the value of biomass, a source of base-load generation, and wind energy which contributes little to meeting peak demand and does not obviate the need to build additional reliable capacity as peak demand grows.
Differentiation among the various alternatives available based on their value is consistent with optimizing investment returns and minimizing ratepayers’ costs which, after ensuring system stability and reliability, is a grid’s primary objective. For example, the cost of Renewable Energy Credits (RECs) is passed on to all ratepayers and is, as such, equivalent to a regressive energy tax. Differentiation would minimize the impact of this energy tax on consumers by favoring energy sources that provide more value towards meeting the capacity required to meet peak demand.
In sum, and from a somewhat broader perspective, we should be encouraging energy sources that (1) are reliable and add capacity to meet peak demand, (2) meet or exceed relevant environmental laws, and (3) are low cost and built close to demand areas to minimize transmission costs. The comparative cost analysis should include all costs, i.e. fixed costs, variable costs and the costs of incentive programs. ERCOT’s comparative analysis of the costs, characteristics, benefits and drawbacks by alternative electricity generating sources provides a helpful framework though this analysis does not include conventional hydropower. (Exhibit C)