United States
Environmental Protection
Agency
The Impact of EPA's
Green Power Purchases
-------
EPA/600/R-07/019
March 2007
The Impact of EPA's Green Power Purchases
Joseph F. DeCarolis
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
Research Triangle Park, NC
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
Notice
The information in this document has been subjected to the Agency's peer and
administrative review and has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability to natural systems to support and nurture life. To meet this
mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future.
The Natural Risk Management Research Laboratory (NRMRL) is the Agency's center for
investigation of technological and management approaches for preventing and reducing risks
from pollution that threaten human health and the environment. The focus of the Laboratory's
research program is on methods and their cost-effectiveness for prevention and control of
pollution to air, land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites, sediments and ground water; prevention and control
of indoor air pollution; and restoration of ecosystems. NRMRI collaborates with both public and
private sector partners to foster technologies that reduce the cost of compliance and to anticipate
emerging problems. NRMRL's research provides solutions to environmental problems by
developing and promoting technologies that protect and improve the environment; advancing
scientific and engineering information to support regulatory and policy decisions; and providing
the technical support and information transfer to ensure implementation of environmental
regulations and strategies at the national, state, and community levels.
This publication has been produced as part of the Laboratory's strategic long-term research plan.
It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
Sally C. Gutierrez, Director
National Risk Management Research
Laboratory
-------
Abstract
All federal agencies, including EPA, are required under Executive Order (EO) 13123 to reduce
life-cycle greenhouse gas emissions attributed to facility energy use by 30% below 1990 levels by
2010. A key approach to reducing facility greenhouse gas emissions, employed by EPA's Office
of Administration and Resources Management (OARM), involves the purchase of "green power".
Green power generally includes renewables (wind, solar, biomass) and other clean energy
technologies (municipal solid waste and landfill gas) that generate electricity. Green tags, which
represent the positive environmental attributes associated with electricity production from green
power sources, are sold through markets to electricity consumers.
The analysis presented in this report meets the following three objectives: (1) establish the 1990
EPA emissions baseline in order to assess progress towards fulfillment of EO 13123, (2) examine
the impact of EPA's green power purchasing on facility-related greenhouse gas emissions and air
pollution, and (3) develop a strategy for future green power purchases. In order to achieve these
objectives, this report describes a new method to estimate net emissions of CO2, SO2, NOX, and
Hg. The estimation of net facility emissions is complicated by the purchase of green tags because
it requires detailed knowledge of which conventional power plants are being offset by purchased
green power. Different offset scenarios are analyzed in order to quantify the uncertainty inherent
in estimating emissions offsets without hour-by-hour system dispatch data.
Keywords: green power, green tags, renewables, greenhouse gas, air quality
-------
Contents
Notice 2
Foreword 3
Abstract 4
Contents 5
List of Tables 6
List of Figures 6
Executive Summary 9
1. Introduction 11
2. EPA Facilities and Emissions Analyzed 12
3. Emissions Estimation from Fuel Consumption 13
4. Emissions Estimation from Electricity Consumption 14
4.1 Facility Emissions from Electricity Use in 1990 14
4.2 Facility Emissions from Electricity Use in 2005 16
4.2.1 Steps 1-2: Calculating Emissions Assuming No Green Power Purchased 17
4.2.2 Steps 3-4: Estimating Offsets from Purchased Green Power 17
4.2.3 Step 5: Estimating Direct Emissions from Green Power 18
4.2.4 Step 6: Calculating Net Emissions from Electricity 19
5. Analysis 20
5.1 Facility Energy Consumption in 1990 and 2005 20
5.2 Estimated CC>2 Emissions from Facility Electricity Use in 2005 21
5.3 Total Emissions from Fuel and Electricity Consumption by Facility in 1990 and
2005 23
5.4 Total Emissions from All Facilities, 1990 and 2005 26
6. Looking Ahead: Maximizing the Environmental Benefits of Green Power 28
6.1 Emissions from Green Power Sources 28
6.2 Best Regions in Which to Buy Green Power 31
7. Conclusions 34
8. Future Work 35
References 36
Appendix 1: Derivation of Biodiesel Emissions Rates 38
Appendix 2: Estimation of Power Plant Thermal Efficiencies By Fuel
Type in 1990 40
Appendix 3: Data for Landfill Gas Emissions Rate Estimates 41
Appendix 4: Description of Spreadsheets Containing Data Analysis 42
Appendix 5: Data Quality Disclaimer 43
-------
-------
List of Tables
Table 1. Emissions factors associated with fuel combustion 13
Table 2. Emissions factors drawn from EPA (2006c) to approximate
average power plant emissions in 1990 15
Table 3. Emissions factors per unit of electricity consumed 16
Table 4. Operating emissions rates from renewable sources 18
Table 6. EPA facilities ordered by the amount of green power purchased 22
Table 7. Operating and life-cycle emissions estimates for
renewable and conventional power sources 30
Table 8. EPA eGRID emissions by NERC subregion,
ranked by the level of CO2 emissions 32
Table 9. R2 values resulting from a regression of coal usage (MWh) versus tons of
emissions 33
List of Figures
Figure 1. Map of NERC subregions 14
Figure 2. Relative mix of power plants in two adjacent
northwestern NERC subregions 15
Figure 3. Semi-log plot of fuel (A) and electricity (B) consumption at select EPA
Facilities in 1990 and 2005 20
Figure 4. Short tons of CO2 emissions resulting from EPA facility electricity
consumption in 2005 22
Figure 5. Total normalized CO26 emissions by EPA facility, 1990 and 2005 23
Figure 6. Total normalized SO2, NOx and Hg emissions by EPA facility,
1990 and 2005 25
Figure 7. Total normalized CO2e, SO2, NOX and Hg emissions, 1990 and 2005 26
Figure 8. Total CO26, SO2, NOx and Hg emissions per gross ft2 in 2005
normalized by 1990 emissions per gross ft2 27
-------
-------
Executive Summary
All federal agencies, including EPA, are required under Executive Order (EO) 13123 to reduce
life-cycle greenhouse gas emissions attributed to facility energy use by 30 percent below 1990
levels by 2010. A key approach to reducing facility greenhouse gas emissions, employed by
EPA's Office of Administration and Resources Management (OARM), involves the purchase of
"green power." Green power generally includes renewables (wind, solar, biomass) and other
clean energy technologies (municipal solid waste and landfill gas) that generate electricity. Green
tags, which represent the positive environmental attributes associated with electricity production
from green power sources, are sold through markets to electricity consumers. By paying a small
premium for green tagged electricity, consumers are given exclusive ownership of the
environmental benefits of the purchased green power, including the greenhouse gas and air
pollutant emissions reductions associated with their use.
The analysis presented in this report meets the following three objectives: (1) establish the 1990
EPA emissions baseline in order to assess progress towards fulfillment of EO 13123, (2) examine
the impact of EPA's green power purchasing on facility-related greenhouse gas emissions and air
pollution, and (3) develop a strategy for future green power purchases. In order to achieve these
objectives, this report describes a new method to estimate net emissions of CO2, SO2, NOX, and
Hg. The estimation of net facility emissions is complicated by the purchase of green tags because
it requires detailed knowledge of which conventional power plants are being offset by purchased
green power. Different offset scenarios are analyzed in order to quantify the uncertainty inherent
in estimating emissions offsets without hour-by-hour system dispatch data.
In 2005, among the OARM-managed EPA facilities existing prior to 1990, one-third have already
met the 30 percent greenhouse reduction specified in EO13123, one-third have reduced
greenhouse gas emissions less than 30 percent below 1990 levels, and one-third have increased
greenhouse gas emissions above 1990 levels. As a whole, OARM-managed EPA facilities have
likely reduced greenhouse gas emissions below 1990 levels, but have not reached the 30 percent
reduction target. Results for SO2, NOX, and Hg emissions are harder to categorize, owing to the
larger ranges of uncertainty in emissions offsets.
To maximize the emissions benefits of future purchases, two key criteria should be used: life-
cycle emissions and location of the green power source. Green power should be purchased from
sources that have minimal emissions, such as wind power and municipal solid waste combustion.
In addition, green power sources should be targeted in regions with the highest levels of coal use
in order to maximize the air quality benefit.
-------
10
-------
1. Introduction
All federal agencies, including EPA, are required under Executive Order (EO) 13123 to reduce
life-cycle greenhouse gas emissions attributed to facility energy use by 30 percent below 1990
levels by 2010. This target is more stringent than the U.S. commitment to the Kyoto Protocol,
which had it been ratified by Congress, would have required a 7 percent reduction below 1990
levels to be achieved on average during the commitment period of 2008 - 2012. Three
noteworthy points drawn from the EO shaped this analysis:
Part 2, Sec.201. "Through life-cycle cost-effective energy measures, each agency shall
reduce its greenhouse gas emissions attributed to facility energy use by 30 percent by
2010 compared to such emissions levels in 1990..."
Part 2, Sec. 204. "Each agency shall strive to expand the use of renewable energy within
its facilities and in its activities by implementing renewable energy projects and by
purchasing electricity from renewable energy sources..."
Part 2, Sec.206. "The Federal Government shall strive to reduce total energy use and
associated greenhouse gas and other air emissions as measured at the source. To that end,
agencies shall undertake life-cycle cost-effective projects in which source energy
decreases, even if site energy use increases..."
In 2004, the federal government consumed roughly 1,200 trillion BTU, which represents slightly
more than 1 percent of total U.S. energy consumption (EIA, 2004). Though government energy
consumption is small on an absolute scale, compliance with EO 13123 presents a significant
opportunity to develop strategies for decision-makers to maximize the benefit of renewable
energy purchases. Insight generated here can be utilized by other federal agencies as well as local
and regional decision makers.
A key approach to reducing facility greenhouse gas emissions, employed by several federal
agencies, involves the purchase of green power. "Green power" generally includes renewables
(wind, solar, biomass) and other clean, non-fossil energy technologies (municipal solid waste and
landfill gas) that generate electricity. Green tags, which represent the positive environmental
attributes associated with electricity production from green sources, are sold through markets to
electricity consumers. By paying a small premium for green tagged electricity, consumers are
given exclusive ownership of the environmental benefits of the purchased green power. As a
result, many private and commercial customers buy green tags to offset the greenhouse gas
and air pollutant emissions associated with their energy use.
EPA's Office of Administration and Resources Management (OARM) has successfully purchased
green power, but as of yet has not been able to estimate total greenhouse gas reductions or
evaluate their progress towards fulfilling the mandate. This report has four objectives: (1)
establish the 1990 emissions baseline in order to assess progress towards fulfillment of EO
13123, (2) examine the impact of EPA's green power purchasing on facility-related greenhouse
gas emissions and air pollution, (3) develop a strategy for future green power purchases, and (4)
propose future work to enhance the analysis presented here.
11
-------
2. EPA Facilities and Emissions Analyzed
Data on 1990 fuel and electricity consumption from the following facilities were available:
Narragansett, Edison, Athens, Gulf Breeze, RTF, Ann Arbor, Duluth / Grosse lie, Cincinnati,
Ada, Las Vegas, Manchester, and Corvallis. In addition to these facilities, the 2005 analysis
included Chelmsford, Fort Meade, Houston, Kansas City STC, Golden, Richmond, and Newport.
The data were provided by the Sustainable Facilities Practices Branch of EPA's Office of
Administration and Resource Management, which is responsible for the operation and
maintenance of select EPA facilities (EPA, 2006a; EPA, 2006b).
Given the significant growth in EPA building space over the last 15 years, the net emissions in
1990 and 2005 are characterized in three ways: facility-by-facility, total emissions across all
facilities, and total emissions across all facilities normalized by building space. Because EO13123
requires absolute emissions reductions, the 2005 emissions estimates were not controlled for
growth in floor space, number of personnel, or variations in the heating and cooling degree days.
Facility emissions came from two sources: fuel combustion and electricity consumption. With
respect to greenhouse gases, CO2 emissions from both electricity and fuel consumption were
tracked. In addition, CFL, and N2O, which are potent greenhouse gases, were estimated for fuel
use and electricity generation in 19901. The CH4 and N2O emissions were multiplied by their
global warming potential (21 and 310, respectively) and added to the CO2 emissions to obtain
CO2 equivalent (CO2e) emissions. Recognizing Part 2, Sec. 206 of the EO (see Section 1 above),
air pollutants (SO2, NOX, and Hg) were also tracked along with greenhouse gas emissions.
1 EPA (2003), which was used to calculate emissions in 2005, provides only CO2 estimates. Examination of
the 1990 electricity data indicates that CH4 and N2O emissions only increase the 1990 CO2e estimates by
approximately 0.2 percent, indicating that their likely impact on the 2005 estimates is negligible.
12
-------
3. Emissions Estimation from Fuel Consumption
The procedure to estimate emissions from facility fuel use in both 1990 and 2005 was identical.
The four fuels consumed by EPA facilities were natural gas, fuel oil, propane, and biodiesel.
Emissions factors, obtained from the EPA (2006c) and Krishna (2004), are shown below in Table
1.
S02
NOX
Hg
C02
CH4
N20
Natural Gas
(lbs/106scf)
0.6
190
0
120,000
2.3
0.64
Fuel Oil
(lbs/103gal)
157
47
1.13X10'4
25,000
0.28
0.11
Propane
(lbs/103gal)
0.1
14
0
12,500
0.2
0.9
Biodiesel
(Ibs / MBTU)
0.328
0.0775
2.4xlO'6
159.3
NA
NA
Table 1. Emissions factors associated with fuel combustion, drawn from EPA (2006c). Emissions from natural gas
assume uncontrolled emissions from a large wall-fired boiler (>100 MBTU/hr) subject to New Source Performance
Standards (NSPS). Emissions from fuel oil are based on a large No. 6 oil boiler (>100 MBTU/hr) with normal firing.
Biodiesel represents a blend of 80 percent No. 2 fuel oil and 20 percent biodiesel - see Appendix 1 for assumptions
regarding the calculation of emissions factors.
The emissions factors in Table 1 were used to calculate facility emissions from fuel use:
roil,p
1x10
3 XQ,oil
propane,p
IxlO3
-------
4. Emissions Estimation from Electricity Consumption
Estimating emissions from facility electricity use is more complicated than estimating fuel
emissions, owing to the fact that electricity is drawn from a large regional grid of power plants.
System operators direct power plant operation to ensure that electricity supply meets demand in
real-time. Control areas are the fundamental entities responsible for performing system dispatch
while maintaining bulk-power reliability, and there are roughly 150 in the U.S. Adjacent control
areas often trade electricity with one another to minimize the amount of required electric
generation capacity. The North American Electric Reliability Council (NERC) has pooled the
control areas into 20 subregions that closely coordinate trading and reliability planning, and those
subregions are also grouped into 10 NERC regions that coordinate activities across the broader
regions. Therefore, each NERC region/subregion is often treated as an inclusive system in
analyses of electric power systems. In this analysis, emissions from facility electricity use and the
emissions offsets obtained through the purchase of renewable electricity are based on the mix of
generators in the relevant NERC subregion. See Figure 1.
hioft
,V"0'
^%HIMS
V*
Figure 1. Map of NERC subregions. The mix of generators in each subregion was used to calculate EPA facility
emissions. Map reproduced from EPA's eGRID.
4. 1 Facility Emissions from Electricity Use in 1990
Comprehensive plant-level data on U.S. power plants emissions were virtually non-existent
before the mid-1990s. The Energy Information Administration (EIA) tracks power plant
emissions in Form-767, but the earliest dataset available is 1996. Likewise, EPA eGRID data
exist from years 1996 to 2000, but not earlier. Boiler level emissions data are available from
EPA's Clean Air Market Division (CAMD) in 1990, but only the boilers subject to the Acid Rain
Program were tracked. As a result, average emissions by NERC subregion can not be estimated
from CAMD data alone.
Lacking direct plant-level emissions data for 1990, emissions by NERC subregion were estimated
directly. The NERC Electricity Supply & Demand (ES&D) database was used to sort plants by
subregion and plant/fuel type2 (NERC, 2004). Six plant types were tracked in each subregion:
2 Although the NERC database contains all plants built through 2004, any plants built after 1990 were filtered
out.
14
-------
coal, oil, natural gas, hydro, nuclear, and wind. Other types present in the database made a
negligible contribution to total regional capacity and were therefore not included. The filtered
data were used to calculate the relative mix of units, expressed as a fraction of total subregional
capacity. Two examples are shown in Figure 2.
NWPN
NWGB
17%
66%
80%
Figure 2. Relative mix of power plants in two adjacent northwestern NERC subregions: Pacific Northwest (NWPN)
and Great Basin (NWGB). The dramatic difference in composition between these regions indicates that emissions will
be much higher in NWGB than NWPN. Note that EPA Las Vegas facility is located in NWGB while EPA's Corvallis
and Manchester facilities are located in NWPN.
To estimate facility emissions, it is also necessary to estimate the emissions rate for the three
fossil-based plant types. The emissions rates were drawn from the EPA (2006c). See Table 2.
S02
NOx
Hg
CO2
CH4
N2O
Coal (Ibs / ton)
38
12
83xlO'5
6040
0.04
0.03
Oil (Ibs /103 gal)
157
47
1.13X10'4
25,000
0.28
0.11
Natural Gas (Ibs / MBTU)
0.0034
0.32
0
110
0.0086
0.003
Table 2. Emissions factors drawn from EPA (2006c) to approximate average power plant emissions in 1990. Coal
emissions factors assume PC-fired, dry bottom, wall-fired boilers burning medium volatility bituminous coal under
New Source Performance Standards. Oil emission factors assume No. 6 oil with 1 percent sulfur content is combusted
in a boiler with normal firing. Natural gas emission rates assume combustion in a stationary turbine.
Using conversion factors, the emissions rates given in Table 2 are converted to units of
Ibs/MBTU3. Because these emissions rates are defined per unit of primary energy and purchased
electricity represents end-use energy, one final transformation is required to account for the
thermal efficiency of electricity production. The average thermal efficiencies for pulverized coal,
oil, and natural gas-fired turbines were assumed to be 33 percent, 32 percent, and 33 percent,
respectively (EIA, 2004). (See Appendix 2 for more details on the thermal efficiency
calculations.) The resulting emissions rates per unit of purchased electricity are given below in
Table 3.
3 According to EPA AP 42, there are 26 MBTU / ton coal and 50 MBTU / 103 gal oil.
15
-------
Coal (Ibs/MWh) Petroleum (Ibs/MWh) Natural Gas (Ibs/MWh)
S02
NOX
Hg
CO2
CH4
N2O
15.1
4.77
3.3xlO'5
2400
0.02
0.01
11.2
3.34
8.03 xlO'6
1780
0.02
0.01
0.04
3.31
0
1140
0.09
0.03
Table 3. Emissions factors per unit of electricity consumed. Data in Table 2 were transformed to Ibs/MBTU and then
divided by appropriate thermal efficiencies to account for heat lost during production.
Using this information, electricity-related facility emissions can be calculated:
Mp,f =Ef X Ifacoal >< ^Q,coal)+ta,oil X^RQoil)+('p,gas * ^RQ,gas)J , (Equation 2)
where Mpf is the mass of pollutant emission attributed to electricity use at a particular facility, Ef
is amount of electricity consumed by the facility, rp is the emissions rate for a given
pollutant and fuel, and jVH?Cf) is the fraction of generating capacity using a given fuel in a
particular region. Note that the NERC fractions only sum to 1 if coal, oil and natural gas are the
only fuels used to produce electricity. The other main power sources - nuclear and hydro - do not
explicitly appear in Equation 2 because they are assumed to have zero operating emissions. Note
that Equation 2 assumes that power plants were utilized in rough proportion to their capacity,
which given the paucity of data for 1990, is a necessary simplifying assumption.
4.2 Facility Emissions from Electricity Use in 2005
The latest available (year 2000) EPA eGRID data were used to estimate the facility emissions in
2005. Two developments over the last five years suggest that the eGRID data are likely
overestimate emissions: the rapid construction of clean-burning natural gas turbines, and further
compliance with federal air quality standards. Given that the 5-year gap between 2000 and 2005
is small compared to the average lifetime of power plants, major structural changes that would
affect the conclusions of the analysis were not observed4.
Estimating the emissions from facility electricity use in 2005 is complicated by the purchase of
renewable energy (mostly green tags), which offset total emissions. It is important to note that
green tags are simply an accounting mechanism to support the development of renewables and
are not physically associated with the purchasing entity. The renewable electricity associated with
the green tags can be produced anywhere in the U.S. as long as the energy production is credited
to the purchaser. In most cases, purchased green tags did not come from the NERC subregion in
which the EPA facility was located. In these cases, green tags are improving air quality in a
region of the U.S. apart from the EPA facilities' local area. Unlike air pollutants, greenhouse
gases are globally well-mixed, so the location where the greenhouse gas emissions offsets take
place do not have regional implications for climate change.
Several steps were required to estimate the net emissions from facility electricity use in 2005,
which includes the effect of purchased green power. The steps are enumerated below:
4 To verify this assumption, electricity generation by state and fuel type was compared for the years 2000
and 2005 using EIA (2005). In most cases, there were only single-digit percentage changes in the amount
of electricity coming from coal, petroleum or natural gas. Nationwide, there was roughly a 2 percent
decrease in coal generation and a 3 percent increase in natural gas generation.
16
-------
1. Identify the NERC subregion in which each EPA facility is located.
2. Calculate facility emissions assuming no green power was purchased.
3. Identify the NERC subregion in which green power purchases are made.
4. Calculate emissions offsets in NERC subregion where the green power is physically
located.
5. Calculate emissions associated with the operation of the green power sources.
6. Calculate net emissions, taking into account conventional electricity consumption,
green power offsets, and direct emissions from green power sources.
4.2.1 Steps 1-2: Calculating Emissions Assuming No Green Power Purchased
Step 1 is accomplished by associating the location of each facility with the regions shown in
Figure 1. Estimated facility emissions in Step 2 are the product of total facility electricity
consumption (conventional + green MWh) and the eGRID output emissions rates (Ibs/MWh) for
the NERC subregion in which the facility is located.
4.2.2 Steps 3-4: Estimating Offsets from Purchased Green Power
Step 3 requires identification of the NERC subregions in which the green power is physically
generated. The physical location of EPA's purchased green power was drawn from ERG (2006).5
Step 4 is difficult because there is no precise method to determine the emissions reductions from
the purchase of electricity generated from renewable sources. Assuming that renewables are built
to supplant existing generators rather than meet growing demand, it is necessary to make
assumptions about which generators are displaced by the purchased renewable energy. Because
renewables represent a small fraction of overall electricity supply and often produce electricity
intermittently, they tend to displace load-following units, which can ramp output quickly to
compensate for changes in supply or demand. Accurate identification of the conventional
generators displaced by renewable energy sources requires substantial data regarding the hourly
dispatch of all generators within a NERC subregion. For example, Connors et al. (2003) estimate
the emissions offsets from solar photovoltaics by performing statistical analysis on hourly time
series data of PV, fossil generation, and electricity demand. Their analysis identifies the load-
following units by identifying the units that adjust their output hour-to-hour in the same direction
as total system demand. Then they calculate the hourly emissions offsets from PV assuming the
load-following units are being displaced. While such hour-by-hour analysis is beyond the scope
of this study, the average annual emissions rate for load-following units by NERC subregion-as
calculated by Connors et al. (2003)-is used to estimate the emissions offsets from EPA facilities.
In addition to the emissions offset estimates using Connors et al. (2003) data, three alternative
scenarios are developed to form a wide but plausible range of emissions offsets. The advantage of
this approach is that it uses highly aggregated data from EPA's eGRID, which greatly simplifies
the analysis while providing a characterization of the uncertainty inherent in such an approach. In
the first scenario, purchased renewables displace only the fossil-based generation. Renewables
displacing fossil fuels exclusively can be considered the "environmental option" because it
minimizes air pollution and greenhouse gas emissions. This scenario forms the upper bound on
the emissions offset from renewables, since the plants with highest emissions are being displaced.
In the second scenario, purchased renewables displace only natural gas turbine capacity. All
5 Eastern Research Group (ERG) was contracted by EPA OARM to perform analysis of EPA's green power
purchases.
17
-------
NERC subregions have natural gas turbine capacity, which is often used to meet peak and
shoulder load. Because gas turbines operate with high marginal costs, it would be cost-effective
in most cases for the system operators to ramp down their output when renewable output ramps
up. Renewables displacing natural gas can be considered the "economic option" and forms the
lower bound on the emissions offset estimates since natural gas is clean-burning. In the third
scenario, purchased renewables displace generators in proportion to the total amount of grid
electricity they produce. This scenario forms a middle estimate between the economic and
environmental options and can be considered the "proportional option" because it assumes the
proportional displacement of all generators. This scenario approach has the advantage of
quantifying the uncertainty inherent in estimating emissions offsets without detailed knowledge
of system dispatch6. For comparison, an additional offset estimate was generated by using
estimates from Connors et al. (2003).
A more detailed accounting of system dispatch is likely to result in offset estimates that fall
between the proportional and economic scenarios. In general, low levels of green power are most
likely to displace conventional generating units with high marginal costs that can ramp output
quickly to follow load, which in many regions is dominated by gas turbine capacity. The Connors
et al. estimates-which are based on higher resolution hourly power plant data-confirm this
reasoning. The environmental scenario, in which only fossil-burning plants are displaced by green
power, should be interpreted as a hypothetical scenario meant to minimize emissions. In reality,
many boiler-based fossil plants such as coal do not have the ability to ramp output on the intra- or
inter-hour timescale in order to compensate time-varying green power sources such as wind or
solar. As such, the lower bound net emissions estimates based on the environmental scenario
should be given less weight in the following section.
4.2.3 Step 5: Estimating Direct Emissions from Green Power
Step 5 requires the estimation of operating emissions from the renewables directly7. In 2005,
there were only three green power sources purchased by EPA: wind, biomass, and land-fill gas
(LFG). Wind was assumed to have zero emissions during operation and non-CO2 emissions
factors for LFG and biomass were drawn from the EPA (2006c). CO2 emissions from landfill gas
and pulp and paper mills were estimated at zero, since the ancillary production of electricity at
such facilities does not result in additional CO2 emissions beyond what would be produced by
normal facility operation. See Table 4.
SO2 (Ibs/MWh) NOX (Ibs/MWh) Hg (Ibs/MWh) CO2 (Ibs/MWh)
Wind 0 000
Landfill Gas 0.20 1.8 3.6xlO"6 0
Biomass 0.86 2.4 0 0
(Pulp and paper)
Table 4. Operating emissions rates from renewable sources drawn from the EPA AP 42 (EPA, 2006c). Landfill gas emissions
estimates are based on Burklin (2003). Because 37 percent of LFG electricity is generated by engines and 63 percent by turbines, a
weighted average emissions rate was calculated based on estimates in Appendix 1. The emission rates for biomass pulp and paper are
drawn from EPA (2003) and represent the specific estimates associated with the plant from which green tags were purchased for the
EPA RTF facility.
6 Emissions rate data for all three offset scenarios are available in eGRID (EPA, 2003). The emissions rates for
the "proportional" scenario were obtained with the eGRID software by clicking on the appropriate subregion,
clicking the "Emissions Profile" tab, and using the values in the middle column ("Output Rate (Ibs/MWh)").
Likewise, the emissions rates for the "economic" and "environmental" scenarios were obtained by clicking the
"Display emissions rates for fossil, coal/oil/gas" button on the "Emissions Profile" tab.
7 Operating emissions were used because life-cycle emissions estimates for landfill gas and pulp and paper
were unavailable.
18
-------
4.2.4 Step 6: Calculating Net Emissions from Electricity
Step 6 incorporates all of the calculations in the previous steps to estimate the net emissions:
MP,f= (£f,c+£f,g)x'"p)C - £f,gx>p,o + £f,gxrP,g (Equations)
emissions assuming no green power emissions offset from green power emissions from green power
Where, as in Equation 2, M represents pollutant mass, E represents electricity consumption, and r
represents the emissions rate. The subscripts are defined as follows: 'f denotes facility, 'c'
denotes conventional electricity, 'g' denotes purchased green electricity, and 'o' denotes
offset (of which there are three scenarios - economic, environmental, and proportional).
The first term in Equation 3 is an estimate of emissions assuming no green power is
purchased, the second term is an estimate of the emissions from conventional capacity
that are being offset by the purchased green power, and the third term is an estimate of
direct emissions from the green power source. To see the relative importance of various
factors in determining net emissions, it is illustrative to rearrange Equation 3 by factoring
out the amount of green power purchased (£f,g):
1 J T^ I ~^f,C
MP,f=£fgx TT-
£fg
(Equation 4)
In addition to the amount of green power purchased (Ef,g), Equation 4 demonstrates that a key
determinant of net emissions is the relative magnitude of the emissions rates associated with the
conventional electricity used by the EPA facility (rp c), the offset of grid electricity by green
power generation (rp 0), and the green power technology itself (rpg).
19
-------
5. Analysis
In this section, the method described above will be utilized to estimate EPA's emissions of
greenhouse gases and air pollutants. These estimates will indicate EPA's progress towards
fulfillment of EO 13123. Net emissions, including green power purchases, will be provided on a
facility-by-facility basis, total across all facilities, and total across all facilities normalized by
building space.
5.1 Facility Energy Consumption in 1990 and 2005
A sharp increase of 85 percent in building space between 1990 and 2005 led to a significant rise
in energy consumption. See Table 5, where fuel and electricity consumption are presented
separately for simplicity8. During this period, electricity consumption increased by 166 percent
and fuel consumption by 6 percent. Interestingly, electricity consumption grew faster than
building space while fuel consumption remained nearly constant.
Building Space (106 Gross ft2)
Electricity (GWh)
Fuel(106BTU)
1990
2.0
106
375
2005
3.7
282
396
Table 5. Total building space and electricity and fuel consumption at select EPA facilities, 1990 and 2005.
For facilities that existed pre-1990, it is instructive to examine the facility-level changes in energy
consumption that took place between 1990 and 2005. See Figure 3.
100.000 -
£ 10.000 -
c
.2
S. 1 .000 -
a 0.100 -
0.010 -
(A)
I-
h
Q 990
2005
1
2. 100
i
(B)
Figure 3. Semi-log plot of fuel (A) and electricity (B) consumption at select EPA facilities in 1990 and 2005. From
1990 to 2005, six facilities reduced fuel consumption while only two facilities decreased electricity use. Note that no
heating fuel is consumed at the Gulf Breeze facility, and Ada dramatically reduced its fuel consumption by installing a
ground source heat pump.
It is important to note that some EPA facilities, such as RTP, switched buildings between 1990
and 2005. Nonetheless, Figure 3 indicates that over the last 15 years, seven facilities reduced fuel
consumption while only two facilities decreased electricity use. The two most likely reasons for
the drop in fuel consumption is a milder winter in 2005 than 1990 and improved energy
8 Estimated energy consumption due to fuel use represents primary energy since it is the heat content of the fuel
consumed. Electricity is shown in end-use units of MWh.
20
-------
efficiency measures9. However, only two facilities (Ann Arbor and Corvallis) decreased their
electricity use. As is the case nationally, the increased use of electrical devices and air
conditioning is likely responsible for the growing electricity consumption at EPA facilities.
5.2 Estimated CO 2 Emissions from Facility Electricity Use in 2005
The key question is how the energy consumption shown in Figure 3 translated into emissions of
greenhouse gases and air pollutants. Estimated facility emissions are complicated by the purchase
of renewable energy because there is no simple way to determine which conventional generators
are displaced by the purchased renewable energy. Accurate estimates require detailed knowledge
of system dispatch10, which is beyond the scope of this analysis. In this analysis, the lack of
system dispatch data creates uncertainty in the emissions offsets. This uncertainty is quantified by
three scenarios, outlined in Section 4.2, which assume that renewables displace different types of
power plants: fossil capacity (environmental scenario), gas turbines with high marginal cost
(economic scenario), and all conventional units (proportional scenario).
Since EPA only began purchasing renewable energy after 1990, this uncertainty in emissions only
applies to the year 2005 emissions. Before proceeding with the 1990/2005 comparative analysis,
it is useful to quantify the uncertainty associated with the purchase of renewable energy credits in
2005. Figure 4 presents facility CO2 emissions from electricity consumption. The three different
offset scenarios result in three distinct emissions estimates for each facility, which are represented
in Figure 4 by the filled circles and error bars.
In six different NERC subregions (NEWE, CALI, NWGB, MAAC, MAINS, SPNO), the
proportional scenario produced higher CO2 emissions than the economic scenario. Equation (3)
implies that this can only happen if the average emissions rate from natural gas plants is higher
than the total average emissions rate, which factors in all power plants. In five of these six
subregions (not SPNO11), the power system was composed of at least 40 percent carbon-free
power sources, mostly nuclear and hydro. As a result, the total average CO2 emissions rate in
these subregions was lower than the CO2 emissions rate from natural gas alone.
The influence of winter weather on fuel consumption can be tested by performing a degree day analysis on a
regional basis for 1990 and 2005.
10 System dispatch is the process by which power plants are called upon by the system operator to produce
electricity; the objective is to minimize generation cost subject to operational limits.
11 According to eGRID, the average natural gas rate in SPNO is very high. As a result, the CO2 emissions rate
from natural gas is higher than the total average rate. It could be due in part to the predominance of low
efficiency natural gas steam boilers and simple-cycle turbines instead of the more efficient combined-cycle
turbines.
21
-------
CL
LU
Corvallis -
Newport, OR -
Manchester, WA -
Richmond, CA -
Las Vegas, NV -
Golden, CO -
Kansas City STC -
Houston, TX -
Ada, OK -
Cincinnati -
Duluth + Grosse lie -
Ann Arbor, Ml -
RTP -
Montgomery -
Gulf Breeze -
Athens -
Fort Meade -
Edison, NJ -
Narragansett, Rl -
Chelmsford, MA -
M
H
K
H
H
-20
-10
10
20
30
CO, Emissions (10 short tons)
Figure 4. Short tons of CO2 emissions resulting from EPA facility electricity consumption in 2005. The three scenarios
that characterize the uncertainty associated with the emissions offsets are represented by the error bars and filled
circles. No green tags were purchased for Newport, Ann Arbor, Montgomery, and Gulf Breeze.
The width of the error bars is explained by the amount of purchased green power. Table 6 ranks
EPA facilities by the amount of green electricity purchased. Note the correspondence between the
facility ranking in Table 6 and the width of the error bars in Figure 4.
Facility
Green Electricity (MWh)
RTP, NC
Cincinnati, OH
Edison, NJ
Las Vegas, NV
Athens, GA
Kansas City STC, MO
Houston, TX
Manchester, WA
Duluth, MN + Grosse He, MI
Golden, CO
Narragansett, RI
Richmond, CA
Ada, OK
Fort Meade, MD
Chelmsford, MA
Corvallis, OR
100,000
15,560
5,027
4,650
4,150
3,529
3,394
3,333
3,074
2,100
1,791
1,434
1,250
800
375
360
Table 6. EPA facilities ordered by the amount of green power purchased. Data drawn from EPA (2006b).
22
-------
5.3 Total Emissions from Fuel and Electricity Consumption by Facility in 1990 and 2005
Adding together CO2e emissions resulting from both fuel and electricity consumption in 1990 and
2005, it is possible to assess EPA's progress towards fulfilling Executive Order 1312312. In this
section, emissions and progress towards the emissions target are presented on a facility-by-
facility basis. Figure 5 shows total CO2e emissions by facility, normalized by each facility's 1990
emissions level. The solid line represents the normalized 1990 CO2e emissions level by facility
and the dotted line represents the normalized emissions target under EO 13123 for 2010, which is
70 percent of 1990 emissions level. The emissions range for 2005 results from the variation in
emissions among the three offset scenarios; the filled circles represent the emissions from the
median scenario. The open circles represent the facility emissions that would have resulted had
the green power purchases not been made. It is clear from inspection of Figure 5 that the
purchased green tags resulted in substantial offsets of CC^e emissions. Facilities can be
grouped into three categories. First, facilities that are likely compliant with EO 13123:
Manchester, Las Vegas, Ann Arbor, RTF, and Edison. Second, facilities that have
reduced CC^e emissions below 1990 levels but not the 30 percent reduction required by
EO 13123: Corvallis, Cincinnati, Duluth+Grosse He, and Narragansett. Third, facilities
that have increased CO26 emissions since 1990: Ada, Gulf Breeze, and Athens.
Q_
LU
Corvallis -
Las Vegas -
Ada -
Cincinnati -
Duluth + Grosse lie -
PTD -
Gulf Breeze -
Athens -
Edison -
Narragansett -
Targs
1 A A
p
Median Offset
O No Offset
A Connors etal. Offset
1 A A
HA <
a
5tl
H
H*
n
r
H
H o
1990 Level
O
£* O
H O
H»H 0
0
h^*l 0
0
-2
-1
CCXe Emissions Relative to 1990 Baseline
Figure 5. Total normalized CO2e emissions by EPA facility, 1990 and 2005. Values are normalized by each facility's
1990 emissions. The solid line represents the 1990 emissions level and the dotted line represents the emissions target
set by EO 13123. The filled circles and error bars are the estimated net emissions under the three different offset
scenarios. The gray triangles represent the emissions offsets using estimates from Connors et al. (2003). The open
circles represent 2005 CO2e emissions if green tags not been purchased. Because no green tags were purchased in 1990,
there is no uncertainty associated with emissions offsets.
12 Note that the number of facilities being analyzed is less than in the previous section. Facilities analyzed in this
section had available data for both 1990 and 2005.
23
-------
Although the intent of EO 13123 is to reduce greenhouse gas emissions, it is also worthwhile to
examine the change in air pollutant emissions, in particular SO2, NOX, and Hg (mercury). The
purpose of examining these air pollutants in addition to greenhouse gases is to examine whether
and to what degree EPA decreased air pollutant emissions through the purchase of green tags.
The following analysis can also be used to help develop a strategy that maximizes the emissions
offsets of both greenhouse gases and air pollutants.
Air pollutant emissions can be estimated with the same procedure used to calculate CO2e
emissions. The results are shown below in Figure 6. Although EO 13123 does not mandate
reductions in air pollution, the same target of 30 percent below 1990 levels applied to CO2e is
provided for reference. Note that the change in air pollutant emissions from 1990 to 2005 is much
more mixed than for CO2e emissions. Because gas turbines produce negligible SO2 and Hg
emissions, assuming that purchased green power displaces natural gas capacity (economic
scenario) results in little or no offset of those emissions. As a result, the theoretical case where no
green power is purchased often results in the same SO2 and Hg emissions as in the economic
offset scenario. In most cases, there are moderate reductions in NOX emissions, in part because
even displacing clean natural gas capacity in the economic scenario results in less NOX emissions.
24
-------
Corvallis -
Manchester -
Las Vegas -
Ada-
Cincinnati -
Duluth + Grosse lie -
Ann Arbor -
RTP -
Gulf Breeze -
Athens -
Edison -
Narragansett -
Corvallis -
Manchester -
Las Vegas -
Ada -
Cincinnati -
Duluth + Grosse lie -
Ann Arbor -
RTP -
Gulf Breeze -
Athens -
Edison -
Narragansett -
Corvallis -
Manchester -
Las Vegas -
Ada-
Cincinnati -
Duluth + Grosse lie -
Ann Arbor -
RTP -
Gulf Breeze -
Athens -
Edison -
Narragansett -
Target |
H
wq
HO-
O
o
1
1990 Level
-^-» 0
Median Offset
»O O No Offset
, . _ A Connors et al. Offset
r" \J
OH
01234
SO, Emissions Relative to 1990 Baseline
Target |
f*3
H
6
HO9
1990 Level
._ Median Offset
O No Offset
£ *O i Connors et al. Offset
O
^*^°
-2-101 234
NO Emissions Relative to 1990 Baseline
Tare
Median Offset
O No Offset
jet
h- L
C
1990 Level
h«D
0
0
O
c
-4-202468
Hg Emissions Relative to 1990 Baseline
Figure 6. Total normalized SO2, NOX and Hg emissions by EPA facility, 1990 and 2005. Values are normalized by
each facility's 1990 emissions. The solid line represents the 1990 emissions level. Although EO 13123 does not target a
reduction in air pollution emissions, the dotted line representing 30 percent below 1990 levels is provided for reference.
The filled circles and error bars are the estimated emissions under the three different offset scenarios. The gray
triangles represent net emissions using offset estimates from Connors et al. (2003). (Note that Connors et al. did not
analyze Hg emissions.) The open circles represent 2005 emissions if green tags not been purchased. Note the mixed
results compared with the reductions in CO2e emissions.
25
-------
5.4 Total Emissions from All Facilities, 1990 and 2005
The previous section presented net emissions on a facility-by-facility basis. To determine overall
progress towards fulfillment of EO 13123, net emissions from all facilities were summed
together. The central challenge of EO 13123 is achieving the mandated greenhouse gas reductions
despite large increases in building space. The CO2e target based on the 1990 emissions baseline is
roughly 7.4xl04 short tons of CO2e emissions from 12 EPA facilities, which account for roughly
2xl06 gross ft2. By 2005, EPA had added eight new facilities, which increased the total building
space by 85 percent to 3.7 xlO6 gross ft2. Without the purchase of green power, CO2e emissions
would have been nearly 2.2x 105 short tons: an increase of 190 percent over the EO target and 100
percent over 1990 emissions levels. Figure 7 demonstrates how EPA's purchase of green power
has affected net emissions of CO2e, SO2, NOX, and Hg between 1990 and 2005.
o
Q_
MHv
l-ln -
ng
Target
1 A i
i
1990 Level 9 Median Offset
O No Offset
A Connors Offset
I A A e~\
/-N
-1
1 2 3
Emissions Relative to 1990 Baseline
Figure 7. Total normalized CO2e, SO2, NOX and Hg emissions, 1990 and 2005. The solid line represents the 1990
emissions level and the dotted line represents the EO target. The filled circles and error bars are the estimated emissions
under the three different offset scenarios. The gray triangles represent net emissions using offset estimates from
Connors et al. (2003). The open circles represent 2005 emissions if green tags not been purchased. Note that with
uncertainty in emissions offsets, net CO2e emissions in 2005 most likely fall somewhere between the 1990 level and
the EO target.
Figure 7 indicates that EPA has made substantial progress towards EO 13123. CO2e emissions
have roughly remained at the 1990 levels, despite an 85 percent increase in building space over
the last 15 years. CO2e emissions would have been 190 percent higher without any purchase of
green power. Net emissions of SO2 and NOX show potentially modest benefits while uncertainty
in Hg offsets is so large that no increase or decrease can be discerned.
Suppose that EO 13123 had been focused on decreasing CO2e / ft2 rather than absolute CO2e.
Such a standard would relax the target by factoring out the emissions associated with increased
building space. Total emissions (shown in Figure 7) normalized by building space yields Figure
8. In this case, EPA has clearly met the CO2e target already. More appreciable reductions in SO2
26
-------
and NOX are also evident, although the reductions fall within the uncertainty associated with the
emissions offsets. The results in Figures 7 and 8 indicate that EPA is close to meeting the EO
target, but that further reductions are required to counter the rapid increase in building space.
CO2 -
9O9 -
2
~o
Wn -
ng -
Target
1 A 1
1 A
1 A
1990 Level . Median Offset
O No Offset
A Connors Offset
o
^ *~t
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
Emissions Per Gross ft , Relative to 1990 Baseline
Figure 8. Total CO2e, SO2, NOX and Hg emissions per gross ft2 in 2005 normalized by 1990 emissions per gross ft2.
Note that when CO2e emissions are normalized by building space, EPA has reduced emissions well below the target of
70 percent of 1990 levels.
27
-------
6. Looking Ahead: Maximizing the Environmental Benefits of Green Power
In the future, EPA can maximize the emissions offsets provided by the purchase of renewable
energy by carefully considering two factors: the direct emissions from different renewable
sources and the total emissions level in each NERC subregion where green power is located. The
following two subsections address these issues in more detail.
6.1 Emissions from Green Power Sources
As shown in Equation 4, the emissions rate from the green power source (rpig) is a key
determinant of net emissions. Estimated emissions rates for green power sources should factor
into purchasing decisions. In considering emissions from power plants, there is an important
distinction between operating emissions and life-cycle emissions. Operating emissions only
account for the emissions released during the production of electricity, while life-cycle emissions
account for all emissions related to the construction, operation, and disposal of a technology as
well as the emissions associated with extraction, transportation, and combustion of fuel. For
reference, Table 7 below presents operating and life-cycle emissions rates for both renewable
energy sources and conventional sources13. Table 7 is not meant to be comprehensive, but rather
to provide a range of the estimated emissions produced by different electric generating
technologies.
Among renewable sources, wind appears to be the most attractive from an emissions accounting
perspective. It has zero operating emissions, and the life-cycle emissions associated with building
and disposing of wind turbines is low. Solar photovoltaics (PV) also produce zero operating
emissions, but the process to fabricate the crystalline silicon panels is energy intensive, resulting
in higher life-cycle emissions than is the case for wind. However, thin film PV (in which
photovoltaic material is applied to a low cost substrate) has the potential to reduce cost and life-
cycle emissions in the future. Like solar PV, biomass also appears to be a relatively low
emissions source of electricity, but there is currently limited availability in existing green power
markets owing to higher generation costs.
The purchase of both municipal solid waste (MSW) and landfill gas (LFG)-widely available
green power alternatives-has interesting implications for CO2e emissions. The natural
decomposition of trash in a landfill produces methane, a potent greenhouse gas. The direct
combustion of MSW reduces greenhouse gas emissions by preventing the formation of methane.
LFG combustion, on the other hand, allows methane to form in the landfill. The methane is then
collected and combusted in either an engine or a turbine. However, the reported collection
efficiencies of methane from landfills range from 60 to 85 percent, where an average efficiency of
75 percent is often assumed (EPA, 2006c). In a life-cycle analysis of different waste management
options, Thorneloe et al. (2006) find that electricity from MSW produces considerably less CO2e
emissions than LFG. This result is due in large part to the high methane leakage rates associated
with LFG. In addition, MSW combustion avoids the issue of leachate leakage from landfills,
which can be difficult to track and can lead to groundwater contamination.
13 It is important to note that different methods for estimating life-cycle analysis can affect emissions estimates.
The estimates presented here were not reviewed in detail for consistency.
28
-------
At first pass, it appears that green power from MSW is preferable to that from LFG. However,
more detailed work is required to quantify the tradeoffs between the two generation alternatives.
Life-cycle emission estimates for both MSW and LFG used in the emissions offset calculations
are open to interpretation, since the estimated net emissions depend on assumptions about how
the waste material would have been handled if not combusted. More accurate quantification of
the life-cycle emissions from LFG and MSW incineration under a variety of scenarios is a high
priority for future work.
Note that among the conventional alternatives, nuclear has very low emissions of both
greenhouse gases and air pollutants, although the issue of nuclear waste disposal presents unique
environmental considerations. Coal clearly has the highest CO2 emissions compared with all
other alternatives.
29
-------
Operating Emissions (Ibs / MWh)
Technology Source QQ
Wind
Landfill Gas a
MSW
Incineration
Biomass (pulp
and paper) °
Biomass
(Varied) d
Solar PV
Denholm
(2005)
IEA(1998)
Bergerson
(2005)
EPAAP
42 (2006)
EPAAP
42 (2006)
EPAAP
42 (2003)
IEA(1998)
Corti et al.
(2004)
Bergerson
(2005)
IEA(1998)
Bergerson
(2005)
Fthenakis
etal.
(2006)
Low
0
High
0
N/A
2130
1900
NA
NA
NA
0
Meier et al.
^^^"
Nuclear e
Natural Gas
(Combined
Cycle)
Coalf'g
^^
Denholm
(2005)
Bergerson
(2005)
Fthenakis
etal.
(2006)
Meier et al
(2005)
Gagnon et
al (2002)
EPAAP
42 (2006)
Denholm
(2005)
ORNL-RFF
(1992)
EPAAP
42 (2006)
Denholm
(2005)
ORNL-
RFF
(1992)
Sundqvist
(2004)
Bergerson
(2005)
CUEcost
(2006)
0
0
0
880
NA
NA
2800
NA
NA
NA
2200
2400
NA
SO2
Low
0
High
0
0.2
0.2
3.7
0.26
NA
NA
NA
0
0
0
0
0.03
NA
NA
18
NA
NA
NA
0.2
1.4
15
NOX
Low
0
High
0
1.8
3.9
2.3
NA
NA
NA
0
0
0
0
2.5
NA
NA
5.5
NA
NA
NA
1.3
2.2
7
Hg
Low
0
High
0
3.6x10~°
0.002
0.006
3.5x10-°
NA
NA
NA
0
0
0
Lifecycle Emissions (Ibs / MWh)
CO2e
Low
11
15
0
High
55
20
60
NA
NA
near zero
40
-1310
-80
220
0
30
60
250
400
370
24
110
SO2
Low
High
<0.04
0.04
0
0.2
0.03
NA
NA
NA
0.032
0.013
0.4
0
0.07
0.35
0.8
0.25
NA
NOX
Low
High
<0.22
0.04
0
0.1
0.05
NA
NA
NA
0.5
0.25
0.4
0
1.1
1
0.7
0.05
NA
90 I NA I NA
0
NA
NA
1.66X10"4
NA
NA
NA
NA
NA
20
7x10~5
30
\ 1
60
26
120
40
30
NA
880
1100
1400
NA
2000
2400
2400
2000
2200
3100
2600
NA
^m
<0.1
1.1x10~5
0.15
NA
NA
0.007
NA
0.2
0.9
0
NA
4
20
4
1.8
0.3
34
1.5
NA
1
^
<0.25
8x10~2
0.11
NA
NA
NA
NA
0.4
1.3
1.1
NA
4
9
7
2
1.8
13
2.6
NA
Table 7. Operating and life-cycle emissions estimates for renewable and conventional power sources. Note that the
zeros in the operating emissions rows for wind, solar PV, and nuclear denote negligible operating emissions. NA = Not
Available.
"Calculations by Clint Burklin of ERG's Morrisville Office, based on EPA's AP 42, See Appendix 3.
bHigh estimate based on uncontrolled emissions, low estimate assumes spray dryer + fabric filter, and an assumed
thermal efficiency of 35 percent.
°Assumes a "wet wood-fired boiler" and a thermal efficiency of 35 percent.
30
-------
dThere is a wide range depending on whether the biomass is gasified before combustion and whether CO2 emissions are
sequestered afterwards (producing net negative emissions). CO2e estimates from Corti et al. (2004) are for biomass
integrated gasification combined cycle, with (low estimate) and without (high estimate) the chemical absorption of
C02.
eThe low end of the Bergerson estimate reflects a once-through fuel cycle, while high end includes fuel recycling.
fCoal emissions estimates from EPA's AP 42 are based on a pulverized coal boiler with the following characteristics:
dry bottom, tangentially-fired, post-NSPS (New Source Performance Standard), and combustion of medium volatile
bituminous coal.
gCoal emissions estimates from Bergerson assume low NOX burners and selective catalytic reduction for NOX control,
and flue gas desulfurization for SO? control.
6.2 Best Regions in Which to Buy Green Power
Equation (4) indicates that an optimal green power purchasing strategy should also take into
account the emissions rates of the NERC subregions in which the green power sources are
located, since the higher the subregion's emissions rates, the higher the offset achieved. With
respect to greenhouse gases, purchasing green power in the NERC subregions with the highest
CO2 emissions rates will maximize the CO2 offset achieved by the purchaser. Because CO2 is
non-reactive and has a long residence time in the atmosphere, the location where the emissions
takes place is irrelevant to global climate.
With criteria pollutants such as SO2 and NOX as well as Hg, the location where emissions take
place is very important because atmospheric transport occurs mainly at local to regional scales. In
this case, searching for NERC subregions with the highest rates of air pollution will provide the
highest offset for the purchaser, but does not necessarily improve air quality the most. To target
areas where green power would have the greatest benefit to air quality, a better metric than
emissions rate is total tonnage of emissions by NERC subregion, which is simply the product of
the emissions rate and total electricity generated in the region. Purchasing green power in regions
with the highest levels of air pollution provides the greatest air quality benefit.
Put simply, green power purchasers should focus on the NERC subregions that have the highest
rates of greenhouse gas emissions and the worst air quality. The process could be complicated by
the tradeoff between greenhouse gas and air pollution emissions as well as tradeoffs among
different air pollutants. Fortunately, targeting specific NERC subregions to achieve maximum
emissions offsets is relatively easy for a simple reason: the energy source that creates the highest
emissions of both greenhouse gases and air pollutants is coal. As a result, regions with the highest
levels of electricity production from coal will also have the highest greenhouse gas and air
pollutant emissions. Using emissions data from EPA (2003), CO2, SO2, NOX, and Hg emissions
by NERC subregion were sorted by CO2 emissions from greatest to least. See Table 8.
31
-------
NERC Subregion
ECAR Ohio Valley
ERCOT
SERC South
SERC
Virginia/Carolina
MAPP
MAIN South
MAAC
FRCC
SPP South
SERC Mississippi
Valley
SERC Tennessee
Valley
WECC Southwest
WECC California
ECAR Michigan
WECC Rockies
SPP North
NPCC New England
MAIN North
WECC Great Basin
WECC Pacific
Northwest
NPCC Upstate NY
NPCC
NYC/Westchester
NPCC Long Island
HICC Oahu
ASCC Alaska Grid
HICC Miscellaneous
ASCC Miscellaneous
Correlation
Coefficient
CO2
(106 tons)
488
221
184
171
165
146
145
126
124
113
111
97
92
77
59
59
52
52
47
43
41
15
12
7
3
2
0
SO2
(103 tons)
3351
464
1085
954
496
612
1013
502
274
276
721
143
64
346
105
172
219
202
81
121
245
8
34
12
4
12
0
0.93
NOX
(103 tons)
1229
351
397
381
356
320
327
297
220
243
288
206
110
159
99
134
87
114
100
80
66
16
19
11
10
16
4
0.98
Hg
(tons)
11.33
4.27
4.38
3.37
3.87
4.49
5.18
1.06
2.07
1.13
1.85
1.84
0.32
1.45
0.34
1.15
0.52
1.09
0.49
0.70
0.65
0.11
0.04
0.05
0.01
0.00
0.00
0.95
Table 8. EPA eGRID emissions by NERC subregion, ranked by the level of CO2 emissions. Note that ECAR - Ohio
Valley has the highest emissions; CO2 and air pollutant levels are at least a factor of two higher than ERCOT, which is
second on the list. The last row provides the correlation (0-1) between the CO2 emissions and SO2, NOX, and Hg
emissions. Data drawn from EPA's eGRID database (EPA, 2003).
Table 8 shows that ECAR Ohio Valley is by far the most polluted NERC subregion. It is also
worth noting that 2-4 orders of magnitude separate the greatest and least polluted regions. In
general, the order of air pollution levels by NERC subregion closely tracks the CO2 ranking. Note
the strong correlation between CO2 and SO2, NOX, and Hg emissions, resulting in correlation
coefficients close to 1.
32
-------
Coal is the fuel that ties all of these emissions together. To prove this assertion, a simple linear
regression of electricity production from coal (MWh) versus the level of emissions (by NERC
subregion) was performed. The results are shown in Table 9.
CO2 SO2 NOX Hg
R2 (Coal MWh) 0.92 0.93 0.98 0.93
Table 9. R values resulting from a regression of coal usage (MWh) versus tons of emissions. The R value estimates
what fraction of the variation in emissions can be explained by coal usage.
Table 9 demonstrates that coal consumption explains more than 90 percent of the variation in
emissions by NERC subregion. As a result, buying green power in areas with the heaviest coal
usage for electricity production produces the greatest greenhouse gas and air pollution benefits.
33
-------
7. Conclusions
The conclusions can be summarized as follows:
Despite growing energy consumption at EPA facilities, the purchase of green power has
offset much of the overall growth in emissions. In many cases, on a facility-by-facility
basis, the purchased green power in 2005 (assuming extension of current contracts) has
already met EPA's mandate under Executive Order 13123. Overall; however, total
reductions of CO2e fall between the 1990 levels and the EO target.
There is uncertainty in determining the emissions offsets from green power because it
depends on which generating units are being displaced by the green power. The scenario
approach used in this analysis quantified the uncertainty in emissions offsets.
Comparison of both operating and life-cycle emissions from various green power sources
revealed that wind and MSW combustion appear to be the cleanest sources that also are
widely available in existing markets.
After comparing NERC subregion emissions of CO2, SO2, NOX, and Hg, it became clear
the best regions to make green power purchases were the ones with the highest levels of
coal use. The three regions with highest emissions were ECAR - Ohio Valley, ERCOT,
and SERC South.
Bottom line: search for the cleanest green power sources in the dirtiest NERC subregions.
A good starting point would be to search for wind power generators selling green tags in
ECAR-Ohio Valley.
34
-------
8. Future Work
There are three goals for further analysis. First, the spreadsheets developed for this analysis could
be automated for use as a decision support tool. A decision-maker would be able to use the
spreadsheet tool to estimate the emissions benefits of different purchasing options. The user
would be able to select and test four key variables: the location of the facility for which green tags
will be purchased, the green power technology, the amount of green energy to be purchased, and
the location of the green power source. Using EPA eGRID data and emissions estimates for green
power sources, the spreadsheet would estimate the net emissions as described in this report.
Second, a dispatch model could be used to identify with greater precision which generating units
are being displaced by purchased green power, which would shrink the size of the error bars
associated with net emissions (see Figures 4-6). The project could build upon work performed by
Stephen Connors et al. (2003), which estimated the emissions offsets from photovoltaics. Their
work utilized plant-level data from EPA's eGRID, hourly plant data from EPA's Clean Air
Markets division, and hourly electricity demand data from the FERC. The data and methodology
could be expanded to create an important decision support resource for green power purchasers.
The resource would allow users to accurately estimate the emissions offsets in each NERC
subregion based on the purchase of electricity from a particular type of generation technology.
Examination of green power purchasing decisions using a tool based on a dispatch model to
estimate emissions offsets would represent a significant improvement for EPA green power
purchasing decisions.
Third, more work needs to be done to better characterize the life-cycle emissions from landfill gas
and MSW combustion. In particular, it is critical to estimate the emissions from these sources
with and without power generation. This analysis can be done by EPA-ORD using the Municipal
Solid Waste Decision Support Tool developed by EPA and RTI (Research Triangle Institute).
This work has already been discussed with the lead developers, and further analysis is planned.
35
-------
References
Bergerson, J. (2005). Future Electricity Generation: An Economic and Environmental Life Cycle
Perspective on Near-, Mid- and Long-Term Technology Options and Policy Implications. PhD
thesis, Department of Engineering and Public Policy, Carnegie Mellon University.
Burklin, C. (2003). Estimated emissions factors for LFG. ERG, Morrisville Office, (see Appendix
1 for assumptions.)
Connors, S., Kern, E., Adams, M., Martin, K., and Asiamah-Adjei, B. (2003). National
Assessment of Emissions Reduction of Photovoltaic (PV) Power Systems. Prepared by Analysis
Group for Regional Electricity Alternatives (AGREA) and Laboratory for Energy and the
Environment (LFEE), Massachusetts Institute of Technology.
Corti, A. and Lombardi, L. (2004). "Biomass integrated gasification combined cycle with reduced
CO2 emissions: Performance analysis and life cycle assessment (LCA)." Energy, 29(2004):
2109-2124.
Denholm, P., Kulcinski, G.L., and Holloway, T. (2005). "Emissions and Energy Efficiency
Assessment of Baeload Wind Energy Systems." Environmental Science & Technology, 39(6):
1903- 1911.
EIA (U.S. Energy Information Administration), Office of Integrated Analysis and Forecasting,
U.S. Department of Energy, (2002). Annual Energy Outlook 2002. DOE/EIA-0384(2002).
EIA (U.S. Energy Information Administration), Office of Energy Markets and End Use, U.S.
Department of Energy, (2004). Annual Energy Review 2003. DOE/EIA-03 84(2003).
EIA (U.S. Energy Information Administration), Office of Coal, Nuclear, Electric and Alternate
Fuels, U.S. Department of Energy, (2005). Electric Power Annual 2005. DOE/EIA-0348(2005).
EPA (2000). Coal Utility Environmental Cost (CUECost) Model Version 1.3. In the Office of
Research and Development, U.S. Environmental Protection Agency.
EPA (2003). Emissions & Generation Resource Integrated Database (EGRID 2002) for Data
Years 1996-2000 (Version 2.01). Prepared by E.H. Pechan and Associates, Inc. (May 2003).
Accessed 2 May 2005 from http://www.epa.gov/cleanenergy/egrid/index.htm.
EPA (2006a). "FY1990_Baseline Data.xls." Electronic spreadsheet provided by the Sustainable
Facilities Practices Branch, Office of Administration and Resource Management, U.S. EPA.
EPA (2006b). "FY05 EPA Lab Energy & Emissions.xls." Electronic spreadsheet provided by the
Sustainable Facilities Practices Branch, Office of Administration and Resource Management,
U.S. EPA.
EPA (2006c).AP 42: Compilation of Air Pollutant Emission Factors, Stationary Point and Area
Sources. Volume 1, Fifth Edition. Accessed 10 April 2005 from
http: //www .epa.gov/ttn/chief/ap42/
36
-------
ERG (2006). "GREEN POWER CHART (just labs) - JAN 19 2005.xls" Electronic spreadsheet
provided by Eliot Metzger of Eastern Research Group (ERG).
Fthenakis, V.M. and Kim, H.C. (in press). "Greenhouse-gas emissions from solar electric and
nuclear power: A life-cycle study." Energy Policy, accepted June 27, 2006.
Gagnon, L., Belanger, C., Uchiyama, Y. (2002). "Life-cycle assessment of electricity generation
options: The status of research in year 2001." Energy Policy, 30: 1267 - 1278.
International Energy Agency. (1998). Benign Energy? The Environmental Implications of
Renewables. OECD/IEA.
Krishna, C.R. (2004). Biodiesel Blends in Space Heating Equipment. National Renewable Energy
Laboratory, Office of Energy Efficiency and Renewable Energy, U.S. Department of Energy.
NREL/SR-510-33579.
Meier, P.J., Wilson, P.H. Kulcinski, G.L., Deholm, P.L. (2005). "U.S. electric industry response
to a carbon constraint: a life-cycle assessment of supply-side alternatives." Energy Policy, 33(9):
1099- 1108.
NERC (North American Electric Reliability Council) (2004). Electricity Supply and Demand
Database and Software, Version 1.1.
Oak Ridge National Laboratory and Resources for the Future. (1992-1996). Estimating the
Externalities of Fuel Cycles. Utility Data Institute and the Integrated Resource Planning Report,
USA.
Sundqvist, T. (2004). "What causes the disparity of electricity externality estimates?" Energy
Policy, 32: 1753 - 1760.
Thorneloe, S.A., Weitz, K., and Jambeck, J. (in print). "Application of the U.S. Decision Support
Tool for Materials and Waste Management." Journal of the Air and Waste Management
Association.
37
-------
Appendix 1: Derivation of Biodiesel Emissions Rates
Two EPA facilities, Narragansett, RI and Manchester, WA, use a biodiesel blend in their boilers.
At both facilities, 80 percent No. 2 fuel oil is blended with 20 percent biodiesel, referred to as
B20. The emissions rates for B20 are estimated from EPA (2006c) and Krishna (2004). Details on
the estimated emissions of CO2, SO2, Hg, and NOX are provided below.
COi Emissions Rate
On a lifecycle basis, the emissions of CO2 from the pure biodiesel are asummed to be negligible.
Based on Table 1.3-12 of the EPA AP 42, the CO2 emissions rate for No. 2 fuel oil is 22,300 Ibs /
103 gal. Based on Krishna (2004), the energy content of B20 is roughly 136,600 BTU per gallon.
Since No. 2 fuel oil represents 80 percent of the B20 mixture,
B20 C02 emissions rate = 0.8 x ( 22'300 lbs CO2 x 1218?! - j = 130.6 ibs/MBTU
^ 103gal 136.6 MBTUj
SOi Emissions Rate
According to Krishna (2004), the sulfur content of the B20 blend used for testing was 0.293
percent. Also, according to Table 1.3-1 of EPA (2006c), the SO2 emissions rate for a No. 2 oil-
fired boiler (>100 MBTU/hr) is 151 S lbs/103 gal, where S is the sulfur content percentage of the
fuel:
B20 SO2 emissions rate = 157 x 0.293 = 461bs x - 12_i?! - = 0 337 ibs/MBTU.
103gal 136.6 MBTU
Hg Emissions Rate
The biodiesel component of B2O is assumed to have zero Hg content. The Hg content of No. 2
fuel oil was drawn from Table 1.3-10 of EPA (2006c):
B20 Hg emissions rate = 0.8 x 131bs x 10 BTU = 2.4 x 10~6 Ibs/MBTU.
12 IMBTU
NOx Emissions Rate
According to Figure 1 8 of Krishna (2004), the concentration of NOX in the commercial boiler is
roughly 42 ppm at 8 percent O2. First, the NOX concentration must be adjusted to a 3 percent O2
level, according to a formula provided by Preferred Instruments (see:
http://www.preferredinstruments.com/engineering data.html#emissions):
NOX ppm (at 3% O2) = " x ppm(actual) .
Iv21-O2(actual)y)
Plugging in the B20 data,
NOX ppm (at 3% O2) = x 42 ppm = 58.1 ppm NOX .
v 21 -8 I
38
-------
Then, according to the Preferred Instruments reference above, the NOX concentration is converted
to Ibs/MBTU through a simple linear transformation:
B20 N(X emissions rate = 58'lppm = 0.0775 Ibs/MBTU .
750
39
-------
Appendix 2: Estimation of Power Plant Thermal Efficiencies By Fuel
Type in 1990
Source: (EIA/DOE, 2004). Annual Energy Review 2003
Electricity Generation by Fuel Type, 1990
Table 8.2a: Electricity Net Generation: Total (All Sectors), 1949-2002
Coal = 1594 billion kWh
Petroleum = 126.6 billion kWh
Natural Gas = 3 72.8 billion kWh
Fuel Consumed for Electricity Generation, 1990
Table 8.3c: Consumption of Combustible Fuels for Electricity Generation: Total (All Sectors),
1949-2002
Coal = 792,457x 103 short tons
Petroleum = 218,997xl03 barrels
Natural Gas = 3,692xlO9 ft3
Conversion Factors, 1990
Table A5: Approximate Heat Content of Coal and Coal Coke, 1949-2002
Electric Power Sector Coal = 20.779 MBTU / short ton
Table A3: Approximate Heat Content of Petroleum Product Weighted Averages, 1949-2002
Electric Power Sector Petroleum = 6.244 MBTU / barrel
Table A4: Approximate Heat Content of Natural Gas, 1949-2002
Electric Power Sector Natural Gas = 1,027 BTU / ft3
Thermal Efficiencies
First, convert primary fuel consumption to MBTU using the conversion factors above. Next,
convert billion kWh of electricity production to MBTU (3.415xl03 MBTU/billion kWh). Divide
MBTUs of energy use electricity by MBTUs of primary energy in the fuel.
5.444xl09
Coal = = 33%
1.647 xlO10
4 323xlO8
Petroleum = = 32%
1.367xl09
AT+ 1.273 xlO9 _-0/
Natural Gas = = 33%
3.792xl09
40
-------
Appendix 3: Data for Landfill Gas Emissions Rate Estimates
Table 4.1998 AP-42 Data
Electricity Generation Technology
LFG 1C Engines
LFG Turbines from MSW LF Chapter (2.4)
LFG Turbines from Gas Turbine Chapter (3.1)"
AP-42 Emission Rates (Ibs/MWh, except LFG flares)
NOX
2.98
1.21
1.98
SO2a
0.186
0.218
0.636
vocb
(0.035 - 0.265)
(0.082-0.311)
0.184
PM
0.572
0.307
0.325
Mercury'
3.62E-06
4.24E-06
N/A
C02
N/A
N/A
N/A
'Assumes all sulfur compounds in LFG are converted to SO2 during combustion. Uses the AP-42 default sulfur content for raw LFG rather than
using landfill-specific data.
b Lower bounds of ranges were calculated by multiplying the default VOC concentration (as hexane) for LFG by typical NMOC control
efficiencies for each technology. Upper bounds of ranges were determined based on the fact that some flares, engines, and turbines are known
to comply with the NSPS limit of 20 ppmv NMOC, which converts to 0.022 Ibs/MMBTU, 0.244 Ibs/MWh, and 0.286 Ibs/MWh VOC, respectively.
Per AP-42, VOC is 39% by weight of NMOC for landfills with no or unknown co-disposal of hazardous waste.
c Uses the default concentration for total mercury in LFG (Table 2.4-1) to estimate mercury released from 1C engines & turbines.
d This value of NOX was not used since it was based on less landfill data.
Constants:
DefaultLFG 1C engine heat rate (BTU/kWh) = 11,100
Default LFG turbine heat rate (BTU/kWh) = 13,000
parasitic loss = 0.92
Methane heating value (BTU/scf) = 1,012
Percent methane in LFG = 50%
Molecular weight of sulfur dioxide (Ib/lb-mol) = 64.1
Molecular weight of hexane (Ib/lb-mol) = 86.2
Molecular weight of mercury (Ib/lb-mol) = 200.6
Scf LFG at standard conditions (scf/lb-mol) = 385.5
% of electricity generated by engines 37
(HHV)
(HHV)
(HHV)
41
-------
Appendix 4: Description of Spreadsheets Containing Data Analysis
These spreadsheets contain the data analysis presented in this report and are available upon
request from the author: decarolis.joseph@epa.gov.
"EPA FY1990 only.xls" contains the following worksheets:
"OARM data" contains original energy data for 1990 obtained from Justin Spenillo.
"NERC pre-1990 generators" provides a comprehensive list of all U.S. power plants
installed in the U.S. before 1990 and still in operation. Data obtained from NERC (2004).
"NERC_frac" provides a summary of power plant capacity by plant type and NERC
subregion. Obtained by filtering data in previous sheet.
"1990 EPA ELC Emissions" performs calculations to estimate EPA facility emissions
from electricity use in 1990.
"1990 EPA Fuel Emissions" performs calculations to estimate EPA facility emissions
from fuel consumption in 1990 using emissions factors from EPA's AP 42.
"Total Emissions" sums the emissions from electricity and fuel consumption.
"EPA FY1990 and FY2005.xls" contains the following worksheets:
"FY05 Totals" contains original energy data for 2005 obtained from Justin Spenillo.
"ELC Emissions 2005" calculates the net emissions by facility, taking into account
emissions offsets from the purchase of green power. For each green power contract, the
net emissions are estimated under 3 different offset scenarios. Tables beginning with cell
B52 summarize the net emissions (absolute and relative) under each scenario.
"Fuel Emissions 2005" performs calculations to estimate EPA facility emissions from
fuel consumption in 2005 using emissions factors from EPA's AP 42.
"Total Energy 1990 - 2005" summarizes both energy associated with electricity and fuel
consumption at EPA facilities in both 1990 and 2005.
"Total Emissions 1990 - 2005" summarizes total CO2e, SO2, NOX and Hg emissions by
EPA facility for 1990 and 2005. The number of facilities used in 1990 to 2005
comparisons is only a subset of the 2005 dataset since less facility data was available for
1990.
42
-------
Appendix 5: Data Quality Disclaimer
This study utilizes several data sources, which can be categorized as follows:
Federal organizations and laboratories
Academic studies
Journal articles
For cases where metadata exist that describe precision, accuracy, completeness, or other
uncertainty measures with respect to the data, the data collector can use this information to assess
data quality. In cases where no quality assurance descriptions are available, the data collector
must accept the data on an as-is basis. The primary sources of the information are ultimately
responsible for the quality of their data. However, the author recognizes that each source
organization may have different levels of resources available to accomplish their mission, or may
have differing commitments to quality assurance within their organization, and consequently data
quality may vary from place to place in ways that cannot be quantified.
Widely cited data from authoritative sources was used wherever possible. The core 1990/2005
comparative analysis was based on several key data sources: EPA (2003, 2006c) for emissions
data, NERC (2004) for 1990 electric generating capacity by NERC subregion, and EIA (2004) for
data used to calculate power plant thermal efficiencies by fuel type in Appendix 2. EPA facility
energy consumption data was drawn from EPA (2006a, 2006b) and details on purchased green
power were drawn from ERG (2006).
Emissions data came from the EPA, which has already been subjected to rigorous internal review.
Electric generating capacity in 1990 came from the North American Electric Reliability Council
(NERC), which is an organization that pools information from all U.S. electric utilities, and is
therefore considered to be a primary source of high quality power system data. Thermal power
plant efficiencies in 1990 were based on estimates from EIA (2004), which is an authoritative
source for U.S. energy data. Finally, the energy consumption data by EPA facility was provided
by EPA's Office of Administration and Resource Management (OARM), which is responsible for
the operation and maintenance of EPA facilities. EPA-OARM also contracted Eastern Research
Group (ERG) to perform analysis of green power purchases, and this report utilizes the reported
physical location of purchased green power from the ERG spreadsheet (ERG, 2006).
Other data from reputable government and academic sources were used to characterize operating
and life-cycle emissions from alternative energy sources.
43
-------
v°/EPA
United States
Environmental Protection
Agency
Office of Research and Development
National Risk Management
Research Laboratory
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/600/R-07/019
March 2007
www.epa.gov
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |