Air Pollution: Meeting Future Electricity Demand Will Increase	 
Emission of Some Harmful Substances (30-OCT-02, GAO-03-49).	 
                                                                 
Electric power plants burn fuels that can produce harmful	 
emissions, such as carbon dioxide, mercury, nitrogen oxides, and 
sulfur dioxide, which can pose human health and environmental	 
risks. To assess the potential risks of meeting future		 
electricity demand, congressional committees asked GAO to (1)	 
report on the Energy Information Administration's (EIA's)	 
national and regional projections of such emissions by 2020, and 
(2) determine how the projections would change using alternative 
assumptions about future economic growth and other factors that  
advisers in these fields recommended. GAO also assessed the	 
potential effects of future electricity demand on water demand	 
and supply.							 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-03-49						        
    ACCNO:   A05230						        
  TITLE:     Air Pollution: Meeting Future Electricity Demand Will    
Increase Emission of Some Harmful Substances			 
     DATE:   10/30/2002 
  SUBJECT:   Air pollution					 
	     Air pollution control				 
	     Electric energy					 
	     Environmental monitoring				 
	     Environmental policies				 
	     Electric power generation				 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-03-49

Report to Congressional Requesters

United States General Accounting Office

GAO

October 2002 AIR POLLUTION Meeting Future Electricity Demand Will Increase
Emissions of Some Harmful Substances

GAO- 03- 49

EIA forecasts that as electricity generation increases 42 percent by 2020,
power plants* annual carbon dioxide and mercury emissions will rise
nationwide by about 800 million tons and 4 tons, respectively. At the same
time, EIA expects plants* annual emissions of nitrogen oxides and sulfur
dioxide to decrease nationwide by about 100 thousand tons and about 2
million tons, respectively. Regionally, EIA forecasts that emissions of
nitrogen oxides and sulfur dioxide will increase in some areas of the
country; mercury will also increase in some areas, while carbon dioxide
will increase in all areas.

EIA also estimated emissions from three additional scenarios, using
different assumptions based on recommendations from advisers GAO
consulted. Like EIA*s original forecast, the scenarios showed an increase
nationwide in power plants* annual carbon dioxide and mercury emissions
and a decrease in emissions of nitrogen oxides and sulfur dioxide between
2000 and 2020, although at different rates than EIA*s projections.
However, the scenarios also showed that, regionally, emissions of nitrogen
oxides and sulfur dioxide could rise in some areas. Separately, GAO found
that EIA had not used the most current data on certain emissions limits in
its model, although this had a limited impact on the forecasts.

GAO estimates that power plants will use between 3 percent less and 17
percent more water by 2020, although they will use less water for each
unit of electricity produced than they currently do, primarily because of
new technologies that require less water. The total increase in water use
is not likely to create shortages, but it could affect companies*
decisions about where to locate new plants and what type to build.

EIA*s Projected Changes in Harmful Air Emissions from Power Plants by 2020

EIA forecasts nationwide increases in power plants' carbon dioxide and
mercury emissions and decreases in their emissions of nitrogen oxides and
sulfur dioxide by 2020. AIR POLLUTION

Meeting Future Electricity Demand Will Increase Emissions of Some Harmful
Substances

www. gao. gov/ cgi- bin/ getrpt? GAO- 03- 49 To view the full report,
including the scope and methodology, click on the link above. For more
information, contact John B. Stephenson at (202) 512- 6225 or stephensonj@
gao. gov. Highlights of GAO- 03- 49, a report to the

Chairmen of the Senate Committee on Environment and Public Works and its
Subcommittee on Clean Air, Wetlands, and Climate Change

October 2002

Electric power plants burn fuels that can produce harmful emissions, such
as carbon dioxide, mercury, nitrogen oxides, and sulfur dioxide, which can
pose human health and environmental risks. To assess the potential risks
of meeting future electricity demand, congressional committees asked GAO
to (1) report on the Energy Information Administration*s (EIA*s) national
and regional projections of such emissions by 2020, and (2) determine how
the projections would change using alternative assumptions about future
economic growth and other factors that advisers in these fields
recommended. GAO also assessed the potential effects of future electricity
demand on water demand and supply.

GAO recommends that the Administrator, EIA, work with EPA and states to
ensure that EIA incorporates into its modeling of electricity generation
and emissions the most current information on regulatory limits for
certain emissions, such as nitrogen oxides.

Page i GAO- 03- 49 Air Pollution Letter 1

Results in Brief 3 Background 5 Power Plants* Carbon Dioxide and Mercury
Emissions Will

Increase by 2020 8 Alternative Assumptions Lead to Wide- Ranging Estimates
of

Future Carbon Dioxide and Mercury Emissions 16 Power Plants* Demand for
Water to Meet Future Electricity Needs

Should Not Create Shortages, but Could Influence the Location of New
Plants 22 Conclusions 27 Recommendations for Executive Action 27 Agency
Comments 27

Appendix I Scope and Methodology 29

Appendix II Projected Emissions through 2020 under Four Alternative Cases
34

Appendix III Comments from the Energy Information Administration 39

Appendix IV GAO Contacts and Staff Acknowledgments 43

Tables

Table 1: Expert Advisers Who Assisted in Our Review 30 Table 2: Carbon
Dioxide Emissions Projections under Four

Cases, 2000- 2020 35 Table 3: Mercury Emissions Projections under Four
Cases,

2000- 2020 36 Table 4: Projections of Emissions of Nitrogen Oxides under
Four

Cases, 2000- 2020 37 Table 5: Sulfur Dioxide Emissions Projections under
Four Cases,

2000- 2020 38 Contents

Page ii GAO- 03- 49 Air Pollution Figures

Figure 1: Electricity Supply Regions Included in EIA*s Forecasts 6 Figure
2: Projected Changes in Total U. S. Electricity Generation, by

Fuel Source, between 2000 and 2020 9 Figure 3: Electricity Generators*
Projected Annual Emissions of

Nitrogen Oxides in 2000 and 2020, by Region 11 Figure 4: Electricity
Generators* Projected Annual Sulfur Dioxide

Emissions in 2000 and 2020, by Region 13 Figure 5: Electricity Generators*
Projected Annual Mercury

Emissions in 2000 and 2020, by Region 14 Figure 6: Electricity Generators*
Projected Annual Carbon Dioxide

Emissions in 2000 and 2020, by Region 15 Figure 7: Percent Changes in
Emissions under Three Scenarios,

2000- 2020 20 Figure 8: Activities That Draw Fresh Water from Rivers,
Lakes, and

Other Sources 23 Figure 9: Diagram of Electricity Generation by a Steam
Turbine 24 Figure 10: Water Drawn Into Power Plants for Cooling Per Unit
of

Electricity Produced, 1950 to 2020 25

Abbreviations

EIA Energy Information Administration EPA Environmental Protection Agency
USGS U. S. Geological Survey SIP State Implementation Plan

Page 1 GAO- 03- 49 Air Pollution

October 30, 2002 The Honorable James M. Jeffords Chairman, Committee on
Environment

and Public Works United States Senate

The Honorable Joseph I. Lieberman Chairman, Subcommittee on Clean Air,

Wetlands, and Climate Change Committee on Environment and Public Works
United States Senate

Electricity is critical to the nation*s economy. To generate electricity,
power plants use a variety of fuels, including fossil fuels* coal, natural
gas, and oil, which account for about two- thirds of the electricity* and
nuclear fuels as well as other sources. As a result of the combustion
process, power plants emit an array of harmful substances, such as carbon
dioxide, mercury, nitrogen oxides, and sulfur dioxide, which pose human
health and environmental risks, especially if the plants do not take steps
to reduce their emissions. For example, carbon dioxide emissions have been
linked to global climate change, among other effects, and exposure to
mercury can lead to nervous system disorders and birth defects. Although
regulations have been implemented to restrict the emission of nitrogen
oxides and sulfur dioxide, and some power plants have installed equipment
to reduce emissions of these substances, these emissions still contribute
to public health problems, including respiratory illnesses and premature
death as well as environmental problems such as acid rain and smog.

Environmental Protection Agency (EPA) data show that, in 1999, power
plants were the single greatest industrial source of all four substances,
emitting 35 percent of the nation*s carbon dioxide, 37 percent of its
mercury, 23 percent of its nitrogen oxides, and 67 percent of its sulfur
dioxide. As demand for electricity grows, companies not only will build
new power plants, but will also continue operating existing plants. Some
of the older existing plants do not have to meet the emission standards
that new plants must meet. As a result, older plants generally emit more
pollution per unit of electricity generated than newer plants.

Forecasts of future electricity supply and demand and associated air
emissions are used to develop national energy and environmental policies,

United States General Accounting Office Washington, DC 20548

Page 2 GAO- 03- 49 Air Pollution

among other things. The Energy Information Administration (EIA) within the
Department of Energy uses the National Energy Modeling System, a computer-
based modeling system to forecast annually future energy supply, demand,
and prices over a 20- year period. EIA*s forecasts depend on the specific
assumptions used in the model, such as economic growth and world oil
prices. EIA develops a *reference case* forecast, which uses assumptions
and data on known technology, demographic and other trends and current
laws and regulations, including those that limit emissions. It also
develops several alternative forecasts, which it bases on assumptions of
both higher and lower oil prices and economic growth, among other factors.

Because the Congress is considering various proposals for a future energy
policy, you asked us to (1) report on EIA*s overall and region- specific
projections of emissions of carbon dioxide, mercury, nitrogen oxides, and
sulfur dioxide from electricity- generating facilities in 2020, and (2)
determine how the emissions projections in EIA*s reference case would
change using alternative assumptions about key variables, such as economic
growth and fuel prices, based on suggestions by expert advisers we
contacted. In addition, because power plants use large amounts of water
during the process of generating electricity, you asked us to determine
how future electricity demand might affect future water demand and supply.

To address these issues, we analyzed EIA*s reference case forecast of
future electricity demand and associated air emissions contained in its

Annual Energy Outlook 2002 and supporting studies. Our analysis focused on
EIA*s estimates of electricity and emissions from electric utility power
plants and excludes industrial and other facilities that produce
electricity as a by- product of their operations, which they then sell to
utilities for distribution. To advise us in our analysis, we contacted a
wide range of individuals with experience in modeling electricity
generation and its environmental effects, including staff of EPA and the
Department of Energy, and representatives of environmental organizations,
consulting firms, research and academic institutions, and the electric
generation industry (see app. I). We asked these individuals, who were
most frequently identified by their peers as knowledgeable in a particular
field, to review EIA*s model and supporting documentation and suggest any
alternative assumptions, such as the rate of future economic growth or
fuel price increases, that they thought were more likely than those
included in EIA*s reference case, based on their expertise in energy
modeling and related topics. We selected EIA*s reference case because,
according to EIA, it presents a *business- as- usual* forecast, based on

Page 3 GAO- 03- 49 Air Pollution

known technology, demographic and other trends, and current laws and
regulations. We then asked EIA to rerun its model substituting alternative
assumptions based on the advisers* suggestions and analyzed the results.
We did not attempt to evaluate EIA*s model or determine which set of
assumptions was the most likely to occur. Finally, we obtained and
analyzed data on power plants* water use and developed estimates of future
water demand based on EIA*s forecasts of electricity production. Our
review was conducted from October 2001 through October 2002 in accordance
with generally accepted government auditing standards. Appendix I contains
additional information regarding our methodology.

EIA*s reference case shows that electricity generation will increase
overall by 42 percent from 2000 through 2020 and that power plants will
emit 800 million tons (or 35 percent) more carbon dioxide and 4 tons (or 9
percent) more mercury per year by 2020 than they did in 2000. This
anticipated increase in emissions would result from power plants*
increased use of fossil fuels to meet anticipated demand and the general
absence of federal or state regulations establishing emissions standards
for carbon dioxide and mercury from power plants. The projected mercury
emissions could decrease, however, once EPA proposes mercury limits, which
are required by 2004. In contrast, EIA forecasts that by 2020 power
plants* total emissions of nitrogen oxides and sulfur dioxide will
decrease nationwide by about 100 thousand tons (2 percent) and about 2
million tons (19 percent), respectively. This expected decline in
emissions results from the anticipated need for power plants to meet
projected increases in electricity demand while complying with clean air
regulations. This will necessitate building new plants that emit
relatively lower levels of these pollutants and installing emissions
controls at some existing plants. Such practices would coincidentally
reduce mercury emissions, explaining in part why EIA*s model projects a
smaller increase in emissions of mercury than carbon dioxide. Despite
these overall declines, EIA forecasts that emissions of nitrogen oxides
and sulfur dioxide will increase in some regions of the country. Such
regional increases may complicate efforts to improve air quality and curb
acid rain in the areas where pollutants are emitted as well as in adjacent
areas where they may spread via wind currents. EIA forecasts that mercury
emissions may increase in some areas and decrease in others, depending on
the amount of coal used, while carbon dioxide emissions will increase
nationwide.

EIA modeled three additional cases using alternative assumptions that
adjusted the model*s values for electricity demand and natural gas prices
to address uncertainties identified by our advisers. Like the reference
case, Results in Brief

Page 4 GAO- 03- 49 Air Pollution

these alternatives showed that, from 2000 through 2020, annual carbon
dioxide and mercury emissions from power plants would rise in all cases,
although at different rates than EIA*s reference case. EIA*s modeling also
showed that emissions of nitrogen oxides and sulfur dioxide would decrease
under all alternatives. Specifically, the modeling showed that

 carbon dioxide emissions could increase between 659 million tons (28
percent) and 1,129 million tons (48 percent);  mercury emissions could
increase between 5,700 pounds (7 percent) and

17,000 pounds (21 percent);  emissions of nitrogen oxides could decrease
between 41 thousand tons

(1 percent) and 204 thousand tons (5 percent); and  sulfur dioxide
emissions could decrease about 2.1 million tons

(19 percent) under all three alternatives, because federal emissions
limits apply under all alternatives.

The modeling also showed that emissions of nitrogen oxides and sulfur
dioxide could increase in some areas, despite the projected decreases
nationwide. Separately, in working with EIA*s model we found that the
agency had not used the most current data on certain emissions limits,
although this had a limited impact on the emissions forecasts. We are
making a recommendation to the Administrator of EIA to update these data
in the model.

Depending on the type of technology installed, as power plants increase
production to meet EIA*s forecast electricity demand, we estimate that
they will use a total of between 3 percent less and 17 percent more water
per year by 2020. However, we also estimate that they will use less water
for each unit of electricity produced than they currently do primarily
because some will install new technologies that require less water. The
future water use is not likely to pose shortages for most areas because
state and local authorities must ensure that communities will have an
adequate water supply before approving new power plants. Nevertheless,
future water use could have some impact on companies* decisions about
where to locate new plants and what type to build. For example, when
deciding whether it is economically feasible to build a new plant in a
particular location, developers must consider, among other things, the
cost of obtaining the needed water or using alternative technologies that
require little water. Such alternatives could increase construction costs
and consume 2 percent to 10 percent of the power generated by the plant.
Finally, while future water use may not affect a locality*s water supply,
it may affect the ecosystems that depend on that water. For example, if
water discharged from a power plant to a body of water has an elevated

Page 5 GAO- 03- 49 Air Pollution

temperature, it could potentially harm aquatic organisms and habitats
downstream. EPA has developed regulations to address some of these
potential effects.

EIA generally agreed with the findings, conclusions, and recommendations
of the report, but suggested a number of technical changes, which we have
incorporated as appropriate.

The Department of Energy Organization Act of 1977 requires EIA to prepare
annual reports including information on trends and projections for energy
use and supply. EIA develops the annual forecasts using the National
Energy Modeling System, a computer- based model, and publishes the results
in the Annual Energy Outlook. Using the model, EIA projects energy supply
and demand and air emissions, among other things, over a 20- year period.
EIA develops 30 cases with alternative assumptions about economic growth,
world oil prices, and electricity demand growth, among other factors, to
address the uncertainties inherent in mid- to long- term forecasting.
EIA*s 2002 projections are based on federal, state, and local laws and
regulations in effect on September 1, 2001, and on data current as of July
31, 2001.

EIA forecasts electricity generation and emissions levels for 13
electricity supply regions (see fig. 1). 1

1 EIA*s electricity supply regions are based on the North American
Electricity Reliability Council*s (NERC) regional divisions. NERC is a
not- for- profit corporation, consisting of members from all segments of
the electric industry, including investor- owned utilities; federal power
agencies; rural electric cooperatives; state, municipal, and provincial
utilities; independent power producers; power marketers; and end- use
customers. Background

Page 6 GAO- 03- 49 Air Pollution

Figure 1: Electricity Supply Regions Included in EIA*s Forecasts

1. East Central (East Central Area Reliability Coordination Agreement) 2.
Texas (Electric Reliability Council of Texas) 3. Mid- Atlantic (Mid-
Atlantic Area Council) 4. Western Great Lakes (Mid- America Interconnected
Network) 5. Upper Midwest (Mid- Continent Area Power Pool) 6. New York
(Northeast Power Coordinating Council/ New York) 7. New England (Northeast
Power Coordinating Council/ New England) 8. Florida (Southeastern Electric
Reliability Council/ Florida) 9. Southeast (Southeastern Electric
Reliability Council /excluding Florida) 10. Lower Midwest (Southwest Power
Pool) 11. Northwest (Western Systems Coordinating Council/ Northwest Power
Pool Area) 12. Southwest (Western Systems Coordinating Council/ Rocky
Mountain Power Area) 13. California (Western Systems Coordinating Council/
California- Southern Nevada Power) Source: GAO characterization of
information from the Energy Information Administration*s Office of
Integrated Analysis and Forecasting.

Page 7 GAO- 03- 49 Air Pollution

EIA*s projections of emissions are influenced by existing laws and
regulations that address air pollution. For example, to help limit
emissions and protect air quality, EPA, under the Clean Air Act, regulates
emissions of nitrogen oxides and sulfur dioxide from a variety of sources,
including power plants that burn fossil- fuels. Under the Clean Air Act,
EPA requires new sources of air pollution within certain industries to
meet federal standards. The standards do not apply to older power plants
built before August 17, 1971, that have not been modified, although some
older plants have taken steps to meet the standards. But, when older
plants make *major modifications* that significantly increase their
emissions, they must install modern pollution controls under the
requirements of a program called New Source Review. EPA currently does not
regulate carbon dioxide or mercury emissions from power plants, although
it plans to issue mercury regulations in 2004.

Power plants must limit their emissions of nitrogen oxides and sulfur
dioxide under the acid rain provisions of the Clean Air Act Amendments of
1990. To achieve reductions in emissions of nitrogen oxides, the
provisions allowed companies with multiple power plants to meet the set
limits by calculating the average of their total emissions across two or
more plants and ensuring that the average did not exceed the limits. This
averaging in effect allows some individual power plants to continue
emitting at levels above the limits.

In contrast, the provisions directed EPA to reduce emissions of sulfur
dioxide from electricity generating units by setting a nationwide limit,
known as a *cap,* on emissions from all power plants, not by setting
limits for individual plants, and establishing an emissions- trading
program. Under this program, each plant receives a number of emissions
*allowances* which each represent the right to emit one ton of sulfur
dioxide. The allowances may be bought, sold, or banked for use in later
years, but power plant owners or operators must own enough allowances at
the end of each year to cover their annual emissions.

Page 8 GAO- 03- 49 Air Pollution

In addition, EPA has established air quality standards for six principal
pollutants including nitrogen dioxide (one of the nitrogen oxides), sulfur
dioxide, and ground- level ozone. 2 These *national ambient air quality
standards* seek to protect public health by limiting the allowable level
of these pollutants in the air. To assist in meeting the ozone standard,
EPA has issued two related regulations that further limit emissions of
nitrogen oxides. In October 1998, EPA issued a final rule requiring
certain states to revise their state implementation plan (SIP) measures to
impose additional controls on emissions of nitrogen oxides to mitigate
ozone transport in the eastern United States. 3 The rule* known as the NOx
SIP call* set stringent caps on emissions of nitrogen oxides in 22
midwestern and eastern states (as well as the District of Columbia) during
the summer. 4 In January 2000, EPA issued another rule* known as the
Section 126 rule* in response to petitions from 8 northeastern states that
the emissions of nitrogen oxides from coal- fired power plants in 12
upwind states and the District of Columbia were being transported by wind
patterns into their states, complicating their efforts to meet national
air quality standards for ground- level ozone. The rule required 392
facilities in the upwind states to reduce annual emissions of nitrogen
oxides and established a cap- andtrade program for emissions within each
of those states. 5

EIA*s reference case forecasts that, overall, as generators increase
electricity production to meet rising demand over the next two decades,
emissions of carbon dioxide and mercury from power plants nationwide will
increase, while their emissions of nitrogen oxides and sulfur dioxide will
decrease. On a regional basis, EIA forecasts that power plants* emissions
of mercury, nitrogen oxides, and sulfur dioxide will increase in some
portions of the country and decrease in others. Carbon dioxide emissions
will increase in all areas. These variations in emissions may

2 The other principal pollutants are carbon monoxide, lead, and
particulate matter. 3 Ozone is a regulated pollutant that forms when
nitrogen oxides react with volatile organic compounds in the presence of
heat and sunlight. 4 63 Fed. Reg. 57356 (Oct. 27, 1998). The states were:
Alabama, Connecticut, Delaware, Georgia, Illinois, Indiana, Kentucky,
Maryland, Massachusetts, Michigan, Missouri, New Jersey, New York, North
Carolina, Ohio, Pennsylvania, Rhode Island, South Carolina, Tennessee,
Virginia, West Virginia, and Wisconsin. The District of Columbia Circuit
Court later vacated the NOx SIP call for Georgia, Missouri, and Wisconsin.
Michigan v. EPA, 213 F. 3d 663 (D. C. Cir. 2000).

5 65 Fed. Reg. 2674 (Jan. 18, 2000). Power Plants* Carbon

Dioxide and Mercury Emissions Will Increase by 2020

Page 9 GAO- 03- 49 Air Pollution

complicate some regions* efforts to control their pollution and reduce the
associated risks.

EIA*s reference case forecasts that, from 2000 through 2020, electricity
generation will increase by 42 percent (see fig. 2), from 3.5 trillion
kilowatt hours in 2000 to almost 5 trillion kilowatt hours in 2020. As
this figure also shows, the largest increase in electricity generation*
240 percent* will come from power plants that burn natural gas.

Figure 2: Projected Changes in Total U. S. Electricity Generation, by Fuel
Source, between 2000 and 2020

Note: Renewable sources include, among others, hydropower, solar, and wind
energy. Source: GAO analysis of EIA data.

In response, EIA expects power plants* annual emissions of carbon dioxide
to increase nationally by about 800 million tons (35 percent), from 2.4
billion tons in 2000 to 3.2 billion tons in 2020. Similarly, EIA forecasts
that plants* annual mercury emissions will increase by about 4 tons (9
percent), from about 40 tons in 2000 to about 44 tons in 2020. EPA plans
to issue regulations limiting mercury emissions from power plants in 2004,
Carbon Dioxide and

Mercury Emissions Are Expected to Increase Overall While Nitrogen Oxides
and Sulfur Dioxide Decrease

Page 10 GAO- 03- 49 Air Pollution

which could reduce emissions below the projected levels. The expected
increase in carbon dioxide and mercury will result primarily from a
projected increase in electricity generation from fossil fuels. For
example, natural gas and coal emit large amounts of carbon dioxide when
burned and coal emits mercury. In addition, these emissions from power
plants will increase because there are no federal or state limits on them,
with the exception of Maine*s mercury emission standard. EIA projects that
88 percent of the 355 gigawatts of new generating capacity needed by 2020
will be fueled by natural gas and another 9 percent by coal. 6 These two
fuels are expected to account for 99 percent of the carbon dioxide
emissions from all electricity production in 2020. Even though mercury
emissions from power plants are not currently federally regulated, they
are not expected to increase substantially in the future in part because
certain measures that generators take to limit emissions of nitrogen
oxides and sulfur dioxide* such as switching to cleaner fuels and
installing emissions control technologies* also coincidentally reduce
power plants* mercury emissions.

In contrast to the growth of carbon dioxide and mercury emissions, EIA
forecasts that, by 2020, power plants* annual emissions of nitrogen oxides
will decline from 2000 levels by about 100 thousand tons (2 percent), from
about 4.3 million tons to about 4.2 million tons, and sulfur dioxide
emissions will decrease by about 2 million tons (19 percent), from 11
million tons to about 9 million tons. Emissions of nitrogen oxides and
sulfur dioxide decline, despite increases in electricity generation,
primarily because federal and state regulations limit power plants*
emissions of these substances. As generators build additional plants and
make major modifications to expand capacity at existing plants to meet
growing electricity demand over the next 20 years, they must also comply
with these limits. To limit emissions from new plants, generators are
expected to build both new natural gas- and coal- burning power plants
that will include emission control technologies. To limit emissions at
some existing plants, generators will continue to switch to cleaner fuels*
such as coal that contains less sulfur* and install technologies to
control these emissions.

6 These percentages exclude electricity that is generated by industrial
and other facilities that is then sold to electric utilities.

Page 11 GAO- 03- 49 Air Pollution

Although EIA forecasts that aggregate annual emissions of nitrogen oxides
and sulfur dioxide will decrease nationally by 2020, it projects that
emissions of both of these pollutants will increase in some regions of the
country and decrease in others. For example, EIA expects emissions of
nitrogen oxides to decrease in 7 of the 13 electricity supply regions (see
fig. 3).

Figure 3: Electricity Generators* Projected Annual Emissions of Nitrogen
Oxides in 2000 and 2020, by Region

Source: GAO analysis of EIA data.

Some Areas of the Country Will Face Increased Emissions

Page 12 GAO- 03- 49 Air Pollution

EIA expects power plants in the Mid- Atlantic area of the United States to
experience the smallest decrease in annual emissions of nitrogen oxides* 6
thousand tons, or 3 percent* and plants in the East Central area to
experience the largest decrease* 182 thousand tons, or 16 percent.
However, EIA projects that emissions of nitrogen oxides will increase in 6
regions

 California (10 thousand tons, or 11 percent);  the Lower Midwest (11
thousand tons, or 4 percent);  the Northwest (32 thousand tons, or 20
percent);  the Southwest (34 thousand tons, or 15 percent);  the Western
Great Lakes (44 thousand tons, or 13 percent); and  the Upper Midwest (60
thousand tons, or 23 percent).

Page 13 GAO- 03- 49 Air Pollution

Similarly, EIA projects that annual sulfur dioxide emissions from power
plants will decline in 10 of the 13 regions by 2020 (see fig. 4).

Figure 4: Electricity Generators* Projected Annual Sulfur Dioxide
Emissions in 2000 and 2020, by Region

Source: GAO analysis of EIA data.

The smallest decrease* 1 thousand tons (under 1 percent)* is expected to
occur in the Southwest and the largest decrease* 950 thousand tons (29
percent)* is expected in the East Central area of the country. However,
power plants* annual emissions of sulfur dioxide are expected to increase
in three regions

 Texas (28 thousand tons, or 8 percent);  the Lower Midwest (33 thousand
tons, or 8 percent); and  the Upper Midwest (38 thousand tons, or 8
percent).

Page 14 GAO- 03- 49 Air Pollution

According to EIA, decreases in mercury emissions will range from about 335
pounds (20 percent) in the New York State area to about 821 pounds (17
percent) in Texas (see fig. 5).

Figure 5: Electricity Generators* Projected Annual Mercury Emissions in
2000 and 2020, by Region

Source: GAO analysis of EIA data.

Furthermore, EIA expects mercury emissions to increase in all but 3
regions, with the smallest increases* about 30 pounds (3 percent)*
occurring in New England and California, and the largest increase* about
2,600 pounds (13 percent)* in the Southeast.

Page 15 GAO- 03- 49 Air Pollution

In contrast, EIA forecasts that carbon dioxide emissions will increase
both nationwide and in all regions (see fig. 6).

Figure 6: Electricity Generators* Projected Annual Carbon Dioxide
Emissions in 2000 and 2020, by Region

Source: GAO analysis of EIA data.

EIA projects that increases in carbon dioxide emissions are likely to
range from a low of 6 million tons (15 percent) in the New York state area
to as much as 200 million tons (40 percent) in the Southeast.

Regional variations in emissions of these four substances result primarily
from (1) differences in electricity demand, which largely determines where
new generating capacity will be added, and (2) power plants* interactions
across regions within the emissions- trading program. As electricity
demand increases in one region, generators will expand capacity at

Page 16 GAO- 03- 49 Air Pollution

existing plants and build additional plants to meet that demand.
Consequently, emissions are likely to increase in those regions where
capacity is expanded. To comply with federal and state limits on emissions
of nitrogen oxides and sulfur dioxide, generators in those regions will,
among other things, purchase emissions credits from other plants, some of
which may be in other regions, to offset the increases. This typically
requires that the plant selling emissions credits either add emissions
controls or switch to cleaner fuels, thereby reducing emissions in that
region. Accordingly, emissions of nitrogen oxides and sulfur dioxide may
increase in the region with the new plants and decrease in the region
where emissions credits were purchased. In contrast, because there are no
federal or, in most cases, state limits on carbon dioxide or mercury
emissions from power plants, these emissions are generally expected to
increase, both nationally and regionally, with the expansion of generating
capacity. However, plants adding pollution controls to reduce sulfur
dioxide and nitrogen oxides in some regions would also remove some mercury
as a side benefit, thereby decreasing mercury emissions in those regions.

In addition to the alternative cases that EIA runs each year as part of
its forecasts, we asked EIA to model three other cases using different
values for electricity demand and fuel prices. These cases showed that,
between 2000 and 2020, annual carbon dioxide and mercury emissions from
power plants would rise under all alternatives, although mercury emissions
would decrease in some regions. The modeling showed overall decreases in
nitrogen oxides and sulfur dioxide under all alternatives, although these
emissions will increase in some regions. Separately, we found that EIA had
not used the most current data on certain emissions limits, which would
have only a modest impact on estimates.

The majority of our advisers described EIA*s modeling methodology as sound
and suitable for forecasting future electricity generation and emissions,
but they did not always agree with EIA*s values for two of the key drivers
of emissions forecasts* electricity demand and fuel prices. They also said
that forecasting is imprecise and that it is difficult to know which
modeling assumptions are most appropriate. Some of the advisers provided
alternative assumptions, which varied widely, causing most estimates of
future emissions to also vary.

EIA*s reference case forecasts that electricity demand will increase by an
average of 1. 8 percent a year between 2000 and 2020. Advisers*
alternatives Alternative

Assumptions Lead to Wide- Ranging Estimates of Future Carbon Dioxide and
Mercury Emissions

Most of the Advisers Agreed with EIA*s Modeling Methodology but Questioned
Electricity Demand and Fuel Price Assumptions

Page 17 GAO- 03- 49 Air Pollution

ranged from an annual increase of 1.25 percent (about 31 percent lower
than EIA*s estimate) to 2.1 percent (about 17 percent higher than EIA). 7
Of the six advisers who provided alternatives, three said that EIA*s
electricity demand estimates were too high; one agreed with EIA; one said
that demand would be equal to or greater than EIA*s estimate; and one said
that EIA*s estimate was too low. Demand could be higher, according to one
adviser, if new technologies that use electricity, such as
electricitybased transportation, are widely adopted. Another adviser
predicted lower increases in demand than EIA and said the actual numbers
will depend primarily on energy efficiency policies and economic growth.
EIA*s analysis of the accuracy of its last 10 annual forecasts found that
it underestimated electricity demand 96 percent of the time, with an
average error of about 4 percent. Because some air emissions increase with
rising electricity demand, underestimating demand can lead EIA*s model to
underestimate emissions also.

Several advisers raised questions as well about EIA*s forecasts of natural
gas prices. Two of the advisers said that EIA*s methodology overstated the
future price of natural gas. Another said prices were too low and that
EIA*s methodology did not capture the likely volatility in gas prices and
future supply constraints that could occur as more gas is used to generate
electricity. This expert suggested that EIA perform additional sensitivity
analyses to address gas price uncertainties.

EIA*s analysis of the accuracy of its past forecasts also indicates that,
of all its fuel price forecasts, those for natural gas have been the least
accurate, deviating from actual prices by an average of 19 percent in the
last ten forecasts, with a tendency to overestimate (58 percent of the
time) rather than underestimate (42 percent of the time) prices. According
to an EIA official, higher gas prices would make new natural gas plants
less economical and could likely lead to the construction of more new coal
plants in the future to meet demand for additional electricity generating
capacity. This in turn would lead to higher emissions, particularly for
carbon dioxide. Therefore, overestimating gas prices could also lead to
overestimating emissions.

7 EIA has developed an alternative case based on an annual electricity
demand growth rate of 2.5 percent.

Page 18 GAO- 03- 49 Air Pollution

Our review of EIA*s modeling found that it included outdated information
on regulations limiting emissions of nitrogen oxides. EIA used preliminary
data on limits for emissions of nitrogen oxides that will take effect in
19 states and the District of Columbia beginning in 2004. As a result, EIA
used a 488,000 ton overall limit in its forecast instead of the 473,000
ton final limit, which was published in the Federal Register. An EIA
official responsible for the model*s emissions data said that while they
met with industry experts* including EPA staff* in developing their
analyses, these final changes were not brought to their attention.
According to this official, because of the relatively small change* a 3
percent decline* updating the information would slightly reduce the
model*s projected emissions of nitrogen oxides. He said EIA would update
the information in the model for future forecasts.

In addition, our review of EIA*s model found that it included data on the
costs and performance characteristics of equipment that power plants use
to control nitrogen oxides that were from a 1996 EPA study. An EPA
contractor updated this information in 2001 based on information provided
by control equipment suppliers and experience gained through more than 200
installations of the equipment. The cost data fell within the ranges
reported in similar studies. EIA found that industry groups and experts in
the Department of Energy believed that the cost of installing equipment
with the performance characteristics described in the study would be
higher than reported. As a result, EIA continued using the earlier cost
and performance estimates. The EIA official responsible for the model*s
emissions data said that the agency agrees with EPA*s updated performance
data but would use higher cost data for future analyses.

EIA prepared for us three alternative emissions projections to its
reference case by running its model with updated information on pollution
control costs and emissions limits as well as revised assumptions to
address the electricity demand and fuel price uncertainties identified by
the advisers. The first alternative* the *revised reference case**
substituted updated information on limits and costs for controlling
emissions of nitrogen oxides. The second alternative* the *high emissions
case** also substituted assumptions about economic growth and
technological change that, in turn, increased electricity demand and the
price of natural gas. The third alternative* the *low emissions case**
substituted assumptions that lowered electricity demand and natural gas
prices.

Nationally, these analyses show that increases in carbon dioxide and
mercury emissions could vary widely in the future, depending on the EIA*s
Model Had Outdated

Information on Certain Emissions Limits, Which Had Little Effect on
Emissions Projections

Modeling with Revised Electricity Demand and Natural Gas Price Data Leads
to Wide- Ranging Estimates of Carbon Dioxide Increases

Page 19 GAO- 03- 49 Air Pollution

assumptions used, while decreases in emissions of nitrogen oxides and
sulfur dioxide would be unlikely to vary significantly from EIA*s
reference case because of regulations that limit these emissions.
Specifically, the modeling showed that between 2000 and 2020:

 Carbon dioxide emissions could increase by between 659 million tons (28
percent) in the low emissions case and 1,129 million tons (48 percent) in
the high emissions case, compared to 827 million tons (35 percent) in
EIA*s reference case. The variation from the reference case results from
differences in the demand for electricity in each case and the amount of
electricity generated from each fossil fuel. Natural gas has about 40
percent less carbon dioxide per unit of energy than coal, so carbon
dioxide emissions from natural gas combustion are proportionately lower. 
Mercury emissions could increase by between 5,700 pounds (7 percent) in

the low emissions case and 17,000 pounds (21 percent) in the high
emissions case, compared to about 7,200 pounds (9 percent) in EIA*s
reference case. Mercury emissions increase in relation to the amount of
electricity generated by coal plants.  Emissions of nitrogen oxides could
decrease by between 41,000 tons

(1 percent) in the high emissions case and 204,000 tons (5 percent) in the
low emissions case, compared to 104,000 tons (2 percent) in EIA*s
reference case. The nitrogen oxides estimates do not vary significantly
from the reference case due to existing control programs and the fact that
new plants are expected to be very clean.  Sulfur dioxide emissions would
decrease by about 2.1 million tons

(19 percent) in all cases because the Clean Air Act Amendments of 1990
call for reductions in annual sulfur dioxide emissions from electricity
generators. 8

Figure 7 compares the national results of the low and high emissions cases
with EIA*s reference case.

8 The modeling results listed here focus on the low and high emissions
cases because the results of the revised reference case did not vary
substantially from EIA*s reference case.

Page 20 GAO- 03- 49 Air Pollution

Figure 7: Percent Changes in Emissions under Three Scenarios, 2000- 2020

Source: GAO Analysis of EIA data.

Under all three alternatives, carbon dioxide would increase in all regions
but the magnitude of the increases would vary widely. (App. II contains a
summary of the regional emissions projections for the reference case and
the three alternative cases). For example, annual emissions in the
Southeast would increase from about 153 million (30 percent) in the low
emissions case to 300 million tons (59 percent) in the high emissions
case, while those in New England would increase from about 4 million (9
percent) in the low emissions case to 17 million tons (36 percent) in the
high emissions case. For all three alternatives, the Southeast and East
Central regions would have the largest emissions increases because these
areas are projected to have the largest increases in fossil- fuel
generation, while New England and New York would have the smallest
emissions increases.

Mercury emissions would increase in 10 of the 13 regions in the revised
reference case and the low emissions case, and in 12 of the 13 regions in
the high emissions case. The Southeast region has the largest expected
increases in emissions and coal- fired generation. Only Texas would have

Page 21 GAO- 03- 49 Air Pollution

emissions decreases across all three alternatives, ranging from about
1,008 pounds (20 percent) to 1,199 pounds (24 percent).

Even though nationally, emissions of nitrogen oxides would decrease,
regionally they would increase in six areas under all three alternatives*
California, Lower Midwest, Northwest, Southwest, Upper Midwest, and
Western Great Lakes* with the magnitude of the increases varying by region
and alternative. The Upper Midwest region would have the largest increase*
ranging from about 56,000 tons (22 percent) to 66,000 (25 percent). The
alternative modeling showed increased generation from fossil fuels in each
of these regions, which may explain the projected emissions increases.

Similarly, while sulfur dioxide emissions are expected to decrease
nationally, they would increase in the Lower Midwest and Upper Midwest
regions despite the federal limits. Emissions would decrease in 8 of the
13 regions under all 3 alternatives. These trends likely hinge on the
national trading program for sulfur dioxide emissions, whereby plants in
some regions would control their emissions and sell excess emissions
credits to plants in other regions.

EIA has modeled additional cases that project far lower emissions than
those presented in its reference case or the three cases EIA prepared for
us. For example, for an October 2001 report, EIA modeled a case based on
assumptions of policies and programs that would promote clean energy
technologies and further reductions in emissions of carbon dioxide,
mercury, nitrogen oxides, and sulfur dioxide. This case showed that, by
2020, emissions of carbon dioxide would be 48 percent lower, mercury 90
percent lower, nitrogen oxides 61 percent lower, and sulfur dioxide 76
percent lower than in EIA*s reference case.

Page 22 GAO- 03- 49 Air Pollution

Electricity generation requires more fresh water than all other sectors of
the economy except agriculture, according to data from the U. S.
Geological Survey (USGS). Power plants* water requirements will likely
rise as demand for electricity grows over the next two decades. However,
the amount of water needed to generate each unit of electricity would
likely decrease because companies are expected to install new technologies
that require less water. The total increase in water use is not likely to
have an impact on most communities* supplies because state and local
authorities protect certain uses, such as for drinking water, when
approving the construction of new power plants in their areas.
Nevertheless, the increase could influence companies* decisions regarding
the locations and types of new plants and may affect aquatic ecosystems.

Power plants draw the second largest amount of fresh water from rivers,
lakes, and other sources each year* 48.2 trillion gallons* according to
1995 USGS data. Only agricultural activities draw more fresh water (see
fig. 8). Power Plants*

Demand for Water to Meet Future Electricity Needs Should Not Create
Shortages, but Could Influence the Location of New Plants

Overall, Power Plants May Need More Water to Operate in 2020 than 2000,
but They Are Expected to Use Less Water per Unit of Electricity Produced

Page 23 GAO- 03- 49 Air Pollution

Figure 8: Activities That Draw Fresh Water from Rivers, Lakes, and Other
Sources

Note: *Other* activities include such uses as domestic, commercial, and
mining activities. Percentages are national averages.

Source: USGS, 1995 data are the most current available.

Nationwide, power plants also use five times more fresh water than
households use for purposes such as drinking, preparing food, and bathing.

Page 24 GAO- 03- 49 Air Pollution

Power plants consume only about 3 percent of the water they draw from a
particular source during the process of generating electricity; in
contrast, agriculture consumes 61 percent. To generate electricity, most
power plants burn a fuel to heat water and create steam (see fig. 9).

Figure 9: Diagram of Electricity Generation by a Steam Turbine

Source: Environmental Protection Agency.

The steam flows through a turbine connected to a generator, which turns
the blades and produces electricity. The steam leaving the turbine is
carried through pipes, which pass through circulating water. The steam
then condenses back into a liquid as the heat is transferred to the water.
This water, in turn, flows to a cooling tower, where the heat dissipates
through contact with the air, and then recirculates to condense the steam
again. This type of system is known as a *wet- cooling* system. Some
cooling systems pass the cooling water over the steam pipes once, and then
discharge it back to its source or the community*s local sewer system,
where the water can be treated and used for other purposes. Such systems
draw in 98 percent more water than a recirculating system.

Given these cooling processes, we estimate that power plants will need
between 94 billion gallons less water (a reduction of 3 percent) per year
by 2020 and 576 billion gallons more water (an increase of 17 percent) to
meet EIA*s reference case projections of future electricity demand. The
lower estimate assumes all the additional demand is met with a cooling
technology that uses significantly less water, and the higher number
assumes it is met with recirculating wet- cooling systems. Plants will
likely

Page 25 GAO- 03- 49 Air Pollution

use a combination of the two systems. Regardless, newer technologies will
allow plants to consume less water per unit of electricity produced than
they currently do. Such reductions in water use would follow a trend that
has been underway since the 1950s (see fig. 10).

Figure 10: Water Drawn Into Power Plants for Cooling Per Unit of
Electricity Produced, 1950 to 2020

Note: Amounts for 2000 and 2020 are GAO estimates. Source: GAO analysis of
USGS data.

The overall increase in the volume of water used by power plants is
unlikely to cause supply shortages for most communities. Companies
generally must obtain state and local approval to withdraw water and a
permit to discharge it back to the water source or a local sewer system
before they can begin construction. In granting the approvals and permits,
the water authority usually must ensure that the plants* water use will
not adversely diminish regional or local water supplies. To help make this
determination, some states are establishing water *budgets* that allocate
water resources to ensure supplies for drinking water and other critical
activities will remain adequate in the future.

On the other hand, future water use could affect decisions about where to
build new plants. When making these decisions, companies must consider
whether sufficient water is available at a particular location and whether
the cost of withdrawing and discharging the water is prohibitive.
Companies consider these costs, along with other important factors* such
The Potential Need for

More Water Should Not Threaten Local Supplies, but Could Affect Where New
Plants Are Built

Page 26 GAO- 03- 49 Air Pollution

as the anticipated demand for electricity, the proximity to fuel and
transmission lines, and the expected selling price for the electricity* to
determine whether building a plant in a particular location would be
profitable.

If companies anticipate water supply problems, they may consider using
alternative supplies or installing technologies that use less water. For
example, 0.5 percent of existing power plants use recycled wastewater,
typically in areas where sufficient water supplies are not available.
California has begun requiring companies to evaluate the feasibility of
using recycled wastewater before approving other water sources. While a
viable alternative, there must be sufficiently large quantities of
wastewater available to meet the power plant*s needs. In addition, plants
must treat the wastewater to remove nutrients and minerals that can foul
equipment and decrease cooling efficiencies, and these treatment costs add
to a plant*s operating costs.

Nearly another 2 percent of existing plants have adopted a technology
known as *dry* cooling, which uses outside air, rather than water, to cool
the steam produced in the plant. Dry- cooling systems can reduce water use
by 90 percent to 95 percent compared to wet- cooling systems that use the
water only once. However, they can cost 2 to 3 times more to construct
than wet recirculating systems. They can also cost significantly more to
operate because the fans and other necessary equipment can themselves
consume from 2 percent to 10 percent of the electricity generated by the
plant. These additional costs can make a dry- cooling system economically
infeasible in some locations.

Although plants* future water use may not affect local water supply, it
can have ecological effects on the original water sources. For example,
pulling water into a plant can kill fish, and discharging water with
elevated temperatures back to its source can damage aquatic organisms and
habitats. However, EPA has developed regulations for new plants and is
developing regulations for existing plants that specify the maximum rates
that plants can take water into the cooling system, among other
requirements. EPA has also proposed that existing plants upgrade their
cooling systems when economically feasible. For example, EPA has proposed
that a plant in Massachusetts reduce the amount of heated water discharged
by almost 96 percent, or approximately 1 billion gallons per day, in order
to lessen the effects on marine life.

Page 27 GAO- 03- 49 Air Pollution

EIA*s forecasts of the future electricity supply and demand as well as
associated air emissions are important for developing national energy and
environmental policies. Both the administration and the Congress have
often relied on EIA*s expertise in modeling and forecasting to assist them
in making decisions about such key policies. Most of the advisers whom we
consulted agreed that EIA*s modeling methodology is sound and suitable for
forecasting future electricity generation and emissions. And while the
advisers disagreed with some of EIA*s values for future electricity demand
and fuel price trends, they and EIA recognize that forecasting is
imprecise and that it is difficult to determine which modeling assumptions
are most appropriate. Nevertheless, regardless of which set of
alternatives becomes reality, the modeling shows that the country will
face elevated levels of carbon dioxide emissions and potentially mercury
emissions. In addition, certain regions of the country will be exposed to
higher levels of emissions of nitrogen oxides and sulfur dioxide, even
though on a national basis, the levels will decrease. Finally, as EIA
continues to assess its modeling accuracy and refine its methodology
accordingly, it is important that the agency use the most current data
available. This includes data on any federal and state regulations that
set limits on emissions, helping to ensure more accurate future estimates.

To ensure that future forecasts of electricity generation and related
environmental effects are as accurate and useful as possible, we recommend
that the Administrator, EIA, work with EPA and states to ensure that the
agency incorporates the most current information on regulatory limits for
certain emissions, such as nitrogen oxides, into the modeling of its
electricity and emissions projections.

EIA provided written comments on a draft of this report. These comments
are reprinted in appendix III. EIA generally agreed with the findings,
conclusions, and recommendations of the report, but believed that there
were areas of the draft report that readers might misunderstand without
additional information. In this regard, EIA suggested a number of
technical changes and clarifications, which we have incorporated as
appropriate. Despite general agreement with the report, EIA disagreed with
what it characterized as the report*s assertion that EIA*s projections
were based on outdated information on the costs of equipment used to
control emissions of nitrogen oxides. However, we believe the report had
already appropriately acknowledged the basis for EIA's decision to
continue to use 1995 data on control costs rather than EPA*s more recent
2001 data. Conclusions

Recommendations for Executive Action

Agency Comments

Page 28 GAO- 03- 49 Air Pollution

EIA made this decision primarily because certain industry representatives
and EIA advisers thought EPA's data underestimated these costs. However,
because the 1995 data used in the model underestimated these costs to an
even greater degree, we asked EIA to incorporate the more recent data for
the three alternative emissions projections the agency prepared for us. In
addition, the Department of Energy*s Office of Energy Efficiency and
Renewable Energy suggested a number of technical changes, which we have
included as appropriate.

As agreed with your offices, unless you publicly announce the contents of
this report earlier, we plan no further distribution until 10 days from
the report date. At that time, we will send copies to the Chairman and
Ranking Minority Member of the House Committee on Energy and Commerce and
its Subcommittee on Energy and Air Quality; the House Committee on
Government Reform and its Subcommittee on Energy Policy, Natural
Resources, and Regulatory Affairs; the Ranking Minority Member of the
Senate Committee on Environment and Public Works, and its Subcommittee on
Clean Air, Wetlands, and Climate Change; other interested members of
Congress; the Administrator, EIA; the Administrator, EPA; the Secretary of
Energy; the Director of the Office of Management and Budget; and other
interested parties. We will also make copies available to others upon
request. In addition, the report will be available at no charge on GAO*s
Web site at http:// www. gao. gov.

If you have any questions about this report, please contact me at (202)
512- 3841. Key contributors to this report are listed in appendix IV.

John B. Stephenson Director, Natural Resources and Environment

Appendix I: Scope and Methodology Page 29 GAO- 03- 49 Air Pollution

To address the first objective, we analyzed EIA*s reference case
projections of future electricity demand and associated air emissions of
carbon dioxide, mercury, nitrogen oxides, and sulfur dioxide, on a
national and regional basis. To obtain this information, we met with EIA
officials responsible for the forecasting model and related emissions
projections and reviewed relevant EIA analyses. We focused on EIA*s
reference case, which accounts for the construction of additional power
plants to meet anticipated electricity demand between 2000 and 2020, as
well as the retirement of those plants that it projects will become
economically unviable.

To address the second objective, we relied on expert advisers who
identified alternative assumptions for EIA*s model. We identified the
advisers using an iterative process (referred to as the *snowball*
technique). We first contacted EIA officials responsible for the agency*s
National Energy Modeling System and its emissions projections to identify
individuals within the government, electricity industry, environmental
organizations, academia, consulting firms, and other organizations who
they believed to be most familiar with EIA*s model and electricity
forecasting. We also spoke with senior officials within EPA, organizations
that perform energy and emissions modeling similar to EIA*s, such as
Platts/ RDI Consulting, and energy and environmental policy analysts from
the electricity industry, such as the Electric Power Research Institute,
and think tanks, such as Resources For the Future. We spoke with these
parties because literature reviews and agency contacts suggested that they
would be best positioned to help us identify individuals with the greatest
knowledge of energy modeling and related issues. We asked them to identify
individuals who are nationally recognized in the fields of energy
modeling, electricity demand and fuel price forecasting, emissions control
technologies, and related areas. We then contacted these individuals and
asked them to identify additional experts in their field. At the
conclusion of this process, we had identified 30 individuals and/ or
organizations.

To select individual advisers from this pool, we applied predetermined
criteria that included (1) area of expertise* to provide adequate coverage
of representatives with detailed knowledge of relevant disciplines,
including electricity modeling and emission control technologies; (2) the
number of times an individual was recommended by others in the same field;
and (3) professional affiliation* to ensure adequate coverage of key
stakeholder groups, including federal agencies, academic institutions,
private consulting firms, and nongovernmental organizations. This process
resulted in the selection of seven advisers who included a cross section
of the various stakeholder groups and specialties. The process was
intended Appendix I: Scope and Methodology

Appendix I: Scope and Methodology Page 30 GAO- 03- 49 Air Pollution

to ensure the selection of advisers who represent a broad range of
opinions and perspectives. Table 1 includes the names and professional
affiliations of the individuals selected.

Table 1: Expert Advisers Who Assisted in Our Review Adviser Affiliation
Type of Organization

Joel Bluestein Energy and Environmental Analysis Consulting Steve Clemmer
Union of Concerned Scientists Environmental Gordon Hester Electric Power
Research Institute Industry Eliot Lieberman U. S. Environmental Protection
Agency Federal government Walter Short National Renewable Energy
Laboratory Federal government Joseph Sutton Westpower Management Team
Consulting Frances Wood OnLocation, Inc. Consulting

In addition to the advisers, we retained the services of Arnold Leitner,
Ph. D.* a nationally recognized energy forecasting expert with Platts/ RDI
Consulting* to analyze the assumptions in EIA*s model and perform
independent modeling.

To collect information and views from the advisers, we sent them
questionnaires which (1) summarized the key variables 1 that EIA officials
identified as most directly affecting EIA*s emissions projections, and (2)
asked specific questions regarding whether they agreed with or would
suggest alternatives to EIA*s assumptions and findings. We also provided
them with a list of EIA*s key assumptions and relevant supporting
documentation. We asked respondents to provide us with specific
alternatives in cases where they disagreed with EIA*s assumptions.

After we received and analyzed the advisers* responses, we determined
that, while they generally agreed with EIA*s overall modeling methodology,
they disagreed with many specific assumptions and modeling outputs and
suggested a wide range of alternatives. We interpreted this as evidence of
the uncertainty associated with long- term energy forecasting.
Accordingly, we asked EIA to rerun its model to address the uncertainties.
Specifically, we asked EIA to run several scenarios that would identify
the lower- and upper- bound of possible

1 These include electricity demand, new plant costs, the fuel mix for
electricity production, expected fuel prices, pollution control equipment
costs, and retirements of older plants.

Appendix I: Scope and Methodology Page 31 GAO- 03- 49 Air Pollution

future air emissions based on alternative assumptions identified by the
advisers.

We then met with EIA officials to determine how best to conduct the
additional modeling. Because we wanted EIA to run cases to reflect our
advisers* assumptions that would lead to both higher and lower estimated
emissions, the officials suggested rerunning the model using alternative
values for electricity demand and fuel prices* the two modeling elements
they said had the greatest influence on the model*s emissions projections.
Instead of using the exact values for electricity demand and natural gas
prices the advisers suggested, EIA used values from cases it had already
run that most closely matched the advisers* alternatives. For example, EIA
used electricity demand values from its high and low macroeconomic cases
and natural gas prices from its slow and fast oil and gas technological
progress cases. We did not attempt to assess the relative likelihood of
the alternatives provided by the advisers or the values EIA used for the
additional modeling versus EIA*s reference case. Because EIA*s model is
integrated* it is composed of separate modules, which produce results
that, in turn, are used as data or assumptions driving other modules* EIA
could not easily substitute some of the other information provided by the
advisers.

The specific cases we asked EIA to run included:  A *revised reference
case,* using all of the assumptions from EIA*s Annual

Energy Outlook 2002 reference case, but including updated EPA data on the
costs of controlling nitrogen oxides and revised state emissions *caps*
for the power sector, as published in the Federal Register on March 2,
2000. The updated costs for controlling emissions of nitrogen oxides were
about 9 percent higher than those used in the reference case. Correcting
the data on state caps for emissions of nitrogen oxides resulted in using
473,000 tons as the overall limit, rather than the 488,000 tons used in
EIA*s reference case (a difference of about 3 percent). The net result of
these corrected assumptions was a 1 percent decrease in emissions of
nitrogen oxides.  A *low emissions case,* using all assumptions as in the
revised reference

case above, except substituting assumptions about economic growth and
technological change that resulted in an electricity demand growth rate
that was 4 percent lower than EIA*s reference case, and natural gas prices
that were 21 percent lower than the reference case.  A *high emissions
case,* using all assumptions from the revised reference

case, except substituting assumptions about economic growth and
technological change that resulted in an electricity demand growth rate

Appendix I: Scope and Methodology Page 32 GAO- 03- 49 Air Pollution

that was 4.5 percent higher than EIA*s reference case and natural gas
prices that were 30 percent higher than the reference case.

For each of these scenarios we received projections of emissions of carbon
dioxide, mercury, nitrogen oxides, and sulfur dioxide on a national and
regional basis through 2020.

The alternative electricity demand values and natural gas prices used in
the low and high emissions cases did not vary equally from the values used
in the reference case. For example, electricity demand was 4 percent lower
than the reference case in the low emissions case but 4.5 percent higher
in the high emissions case. As a result, the difference in emissions
levels between the high emissions case and the reference case tends to be
larger than that between the low emissions case and the reference case.
The unequal variation from the reference case in each of the additional
cases is a function of the alternatives provided by the advisers and EIA*s
decision to use values for electricity demand and natural gas prices from
cases it had already run. While EIA*s model is sensitive to these changes,
as demonstrated by the wide- ranging results for carbon dioxide and
mercury, the results should not be considered a sensitivity analysis. It
is also important to note that the high emissions case involved both high
gas prices and high electricity demand, which led to large amounts of
generation from coal and high carbon dioxide and mercury emissions. It is
possible that an alternative case could have similarly high demand but
lower emissions of these substances due to lower gas prices. Similarly,
the low emissions case had low demand and low gas prices, which led to
relatively high levels of generation from natural gas. It is possible that
an alternative case could have equally low demand but higher emissions if
gas prices were higher.

To respond to the third objective, we obtained and analyzed information
from EIA, and a report issued by the California Energy Commission (CEC)
that relied on data from the Electric Power Research Institute and the
Public Interest Energy Research Program. We used EIA*s reference case
projections to determine the amount of electricity that EIA expects each
type of plant (e. g., steam, turbine, nuclear) to produce in the future.
Next, we obtained data on water consumption rates (expressed in gallons
per megawatt hour of power produced) for different types of power plants
from the CEC report. We then multiplied the projected annual generation
produced by each plant type by the typical water use rates. We did not try
to incorporate projections of improvements in generation or cooling
technologies, which in the past have reduced the amount of water used by
power plants. In order to reflect the uncertainty associated with
projecting

Appendix I: Scope and Methodology Page 33 GAO- 03- 49 Air Pollution

water use by power plants in 2020, we calculated estimates using two
extreme assumptions about the methods power plants used for cooling. We
first assumed that all power plants adopted *dry cooling** a method that
uses much less water than the current average. Then we assumed that all
power plants use *wet- cooling* technology available in 2000. This
provided a range of possible water use rates. Assuming all plants adopt
wet cooling likely overstates the actual water needs for 2020 because it
ignores (1) the likely use by some generators of dry- cooling methods and
(2) possible innovations in either generating or cooling technology that
would reduce water use.

Appendix II: Projected Emissions through 2020 under Four Alternative Cases

Page 34 GAO- 03- 49 Air Pollution

The following tables present information on the projected emissions of
carbon dioxide, mercury, nitrogen oxides, and sulfur dioxide through 2020
under EIA*s reference case and the three alternative cases EIA prepared
for us: the low emissions case, the revised reference case, and the high
emissions case. The information in each table includes, by region, the
projected volume and percentage changes of the emissions from 2000 to 2020
under the four cases. Appendix II: Projected Emissions through

2020 under Four Alternative Cases

Appendix II: Projected Emissions through 2020 under Four Alternative Cases

Page 35 GAO- 03- 49 Air Pollution

Table 2: Carbon Dioxide Emissions Projections under Four Cases, 2000- 2020
EIA reference case Low emissions case Revised reference case High
emissions case

Region Change in

emissions 2000- 2020

(million tons)

Percent change 2000- 2020

Change in emissions

2000- 2020 (million

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (million

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (million

tons) Percent

change 2000- 2020

1 154 28 130 24 145 26 148 27 2 31 17 23 12 30 16 45 24 3 53 44 49 40 52
42 68 55 4 59 35 55 32 59 34 67 40 5 44 32 39 28 43 31 67 49 6 6 15 6 14 6
15 12 29 7 716 49 7 15 17 36 8 47 43 36 33 48 44 80 73 9 201 40 153 30 198
39 300 59 10 35 20 31 18 34 20 52 30 11 77 76 39 39 77 75 92 91 12 74 53
59 42 74 52 82 58 13 38 45 39 47 36 43 98 119

Total U. S. 827 35 659 28 808 34 1,129 48

Note: The regions included in the table are 1. East Central (East Central
Area Reliability Coordination Agreement), 2. Texas (Electric Reliability
Council of Texas), 3. Mid- Atlantic (Mid- Atlantic Area Council), 4.
Western Great Lakes (Mid- America Interconnected Network), 5. Upper
Midwest (Mid- Continent Area Power Pool), 6. New York (Northeast Power
Coordinating Council/ New York), 7. New England (Northeast Power
Coordinating Council/ New England), 8. Florida (Southeastern Electric
Reliability Council/ Florida), 9. Southeast (Southeastern Electric
Reliability Council /excluding Florida), 10. Lower Midwest (Southwest
Power Pool), 11. Northwest (Western Systems Coordinating Council/
Northwest Power Pool Area), 12. Southwest (Western Systems Coordinating
Council/ Rocky Mountain Power Area), and 13. California (Western Systems
Coordinating Council/ California- Southern Nevada Power).

Source: EIA.

Appendix II: Projected Emissions through 2020 under Four Alternative Cases

Page 36 GAO- 03- 49 Air Pollution

Table 3: Mercury Emissions Projections under Four Cases, 2000- 2020 EIA
reference case Low emissions case Revised reference case High emissions
case

Region Change in

emissions 2000- 2020

(lbs.) Percent

change 2000- 2020

Change in emissions

2000- 2020 (lbs.)

Percent change 2000- 2020

Change in emissions

2000- 2020 (lbs.)

Percent change 2000- 2020

Change in emissions

2000- 2020 (lbs.)

Percent change 2000- 2020

1 148 1 345 2 368 2 677 3 2 -821 -17 -1,008 -20 -1,199 -24 -989 -21 3 -352
-5 -741 -11 -542 -8 674 10 4 1, 173 21 978 17 1,178 20 1,800 32 5 1, 350
26 1,351 26 1,347 26 2,170 43 6 -335 -20 -530 -31 -137 -8 68 4 7 32 332332
3464 8 659 32 463 22 657 32 1,852 89 9 2, 634 13 2,456 12 3,599 18 6,111
31 10 375 7 377 7 377 7 802 15 11 1,058 46 666 29 1,055 47 1,467 65 12
1,286 40 1,090 34 1,283 40 1,308 41 13 32 3 222 29 221 29 1,013 134

Total U. S. 7,240 9 5, 700 7 8, 240 10 17,000 21

Note: The regions included in the table are 1. East Central (East Central
Area Reliability Coordination Agreement), 2. Texas (Electric Reliability
Council of Texas), 3. Mid- Atlantic (Mid- Atlantic Area Council), 4.
Western Great Lakes (Mid- America Interconnected Network), 5. Upper
Midwest (Mid- Continent Area Power Pool), 6. New York (Northeast Power
Coordinating Council/ New York), 7. New England (Northeast Power
Coordinating Council/ New England), 8. Florida (Southeastern Electric
Reliability Council/ Florida), 9. Southeast (Southeastern Electric
Reliability Council /excluding Florida), 10. Lower Midwest (Southwest
Power Pool), 11. Northwest (Western Systems Coordinating Council/
Northwest Power Pool Area), 12. Southwest (Western Systems Coordinating
Council/ Rocky Mountain Power Area), and 13. California (Western Systems
Coordinating Council/ California- Southern Nevada Power).

Source: EIA.

Appendix II: Projected Emissions through 2020 under Four Alternative Cases

Page 37 GAO- 03- 49 Air Pollution

Table 4: Projections of Emissions of Nitrogen Oxides under Four Cases,
2000- 2020 EIA reference case Low emissions case Revised reference case
High emissions case

Region Change in

emissions 2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

1 -182 -16 -228 -20 -193 -17 -207 -18 2 -31 -11 -30 -11 -31 -11 -30 -11 3
-6 -3 -3 -1 00 10 4 44 13 33 10 34 11 299 5 60 23 56 22 59 23 66 25 6 -10
-15 -9 -15 -9 -14 -8 -12 7 -12 -18 -14 -21 -12 -19 -3 -4 8 -24 -11 -34 -16
-23 -11 -5 -2 9 -30 -3 -36 -4 -28 -3 -22 -2 10 11 4 13 5 11 4 16 6 11 32
20 10 6 32 20 40 25 12 34 15 26 12 34 15 38 17 13 10 11 9 11 8 10 45 54

Total U. S. -104 -2 -204 -5 -118 -3 -41 -1

Note: The regions included in the table are 1. East Central (East Central
Area Reliability Coordination Agreement), 2. Texas (Electric Reliability
Council of Texas), 3. Mid- Atlantic (Mid- Atlantic Area Council), 4.
Western Great Lakes (Mid- America Interconnected Network), 5. Upper
Midwest (Mid- Continent Area Power Pool), 6. New York (Northeast Power
Coordinating Council/ New York), 7. New England (Northeast Power
Coordinating Council/ New England), 8. Florida (Southeastern Electric
Reliability Council/ Florida), 9. Southeast (Southeastern Electric
Reliability Council /excluding Florida), 10. Lower Midwest (Southwest
Power Pool), 11. Northwest (Western Systems Coordinating Council/
Northwest Power Pool Area), 12. Southwest (Western Systems Coordinating
Council/ Rocky Mountain Power Area), and 13. California (Western Systems
Coordinating Council/ California- Southern Nevada Power).

Source: EIA.

Appendix II: Projected Emissions through 2020 under Four Alternative Cases

Page 38 GAO- 03- 49 Air Pollution

Table 5: Sulfur Dioxide Emissions Projections under Four Cases, 2000- 2020
EIA reference case Low emissions case Revised reference case High
emissions case

Region Change in

emissions 2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

Change in emissions

2000- 2020 (thousand

tons) Percent

change 2000- 2020

1 -950 -29 -888 -27 -982 -29 -987 -31 2 28 8 29 8 -21 -5 80 23 3 -298 -29
-428 -42 -472 -47 -485 -47 4 -207 -21 -252 -25 -128 -14 -202 -21 5 388286
5111 378 6 -96 -35 -118 -42 -89 -33 -80 -28 7 -76 -32 -98 -40 -79 -32 -16
-6 8 -102 -27 -158 -41 -93 -26 -47 -12 9 -376 -12 -304 -9 -248 -8 -380 -12
10 33 8 33 8 38 9 38 9 11 -59 -36 -67 -41 -59 -36 -56 -33 12 -10 -5-3 -2-1
-10 13 -21 -23 -5 -8 -5 -9 12 20

Total U. S. -2,088 -19 -2,088 -19 -2,088 -19 -2,088 -19

Note: The regions included in the table are 1. East Central (East Central
Area Reliability Coordination Agreement), 2. Texas (Electric Reliability
Council of Texas), 3. Mid- Atlantic (Mid- Atlantic Area Council), 4.
Western Great Lakes (Mid- America Interconnected Network), 5. Upper
Midwest (Mid- Continent Area Power Pool), 6. New York (Northeast Power
Coordinating Council/ New York), 7. New England (Northeast Power
Coordinating Council/ New England), 8. Florida (Southeastern Electric
Reliability Council/ Florida), 9. Southeast (Southeastern Electric
Reliability Council /excluding Florida), 10. Lower Midwest (Southwest
Power Pool), 11. Northwest (Western Systems Coordinating Council/
Northwest Power Pool Area), 12. Southwest (Western Systems Coordinating
Council/ Rocky Mountain Power Area), and 13. California (Western Systems
Coordinating Council/ California- Southern Nevada Power).

Source: EIA.

Appendix III: Comments from the Energy Information Administration

Page 39 GAO- 03- 49 Air Pollution

Appendix III: Comments from the Energy Information Administration

Appendix III: Comments from the Energy Information Administration

Page 40 GAO- 03- 49 Air Pollution

Appendix III: Comments from the Energy Information Administration

Page 41 GAO- 03- 49 Air Pollution

Appendix III: Comments from the Energy Information Administration

Page 42 GAO- 03- 49 Air Pollution

Appendix IV: GAO Contacts and Staff Acknowledgments

Page 43 GAO- 03- 49 Air Pollution

John B. Stephenson (202) 512- 3841 Eileen R. Larence (202) 512- 6510

In addition to the individuals named above, Michael Hix, Vincent Price,
and Laura Yannayon made key contributions to this report. Important
contributions were also made by Frank Rusco and Amy Webbink. Appendix IV:
GAO Contacts and Staff

Acknowledgments GAO Contacts Acknowledgments

(360256)

The General Accounting Office, the investigative arm of Congress, exists
to support Congress in meeting its constitutional responsibilities and to
help improve the performance and accountability of the federal government
for the American people. GAO examines the use of public funds; evaluates
federal programs and policies; and provides analyses, recommendations, and
other assistance to help Congress make informed oversight, policy, and
funding decisions. GAO*s commitment to good government is reflected in its
core values of accountability, integrity, and reliability.

The fastest and easiest way to obtain copies of GAO documents at no cost
is through the Internet. GAO*s Web site (www. gao. gov) contains abstracts
and fulltext files of current reports and testimony and an expanding
archive of older products. The Web site features a search engine to help
you locate documents using key words and phrases. You can print these
documents in their entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as *Today*s Reports,* on its
Web site daily. The list contains links to the full- text document files.
To have GAO e- mail this list to you every afternoon, go to www. gao. gov
and select *Subscribe to daily E- mail alert for newly released products*
under the GAO Reports heading.

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U. S. General Accounting Office 441 G Street NW, Room LM Washington, D. C.
20548

To order by Phone: Voice: (202) 512- 6000 TDD: (202) 512- 2537 Fax: (202)
512- 6061

Contact: Web site: www. gao. gov/ fraudnet/ fraudnet. htm E- mail:
fraudnet@ gao. gov Automated answering system: (800) 424- 5454 or (202)
512- 7470

Jeff Nelligan, Managing Director, nelliganj@ gao. gov (202) 512- 4800 U.
S. General Accounting Office, 441 G Street NW, Room 7149 Washington, D. C.
20548 GAO*s Mission

Obtaining Copies of GAO Reports and Testimony

Order by Mail or Phone To Report Fraud, Waste, and Abuse in Federal
Programs

Public Affairs
*** End of document. ***